Files
ANSLibs/swig-4.3.0/Doc/Manual/SWIGDocumentation.html

66329 lines
2.4 MiB

<!DOCTYPE html>
<HTML>
<HEAD>
<TITLE>SWIG-4.3 Documentation</TITLE>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=iso-8859-1">
<STYLE TYPE="text/css"><!--
div.sectiontoc {
border-style: dotted;
border-width: 2px;
padding: 2pt;
}
h2 {
padding: 3px;
color: #000000;
border-bottom: 2px
solid #dddddd;
}
h3, h4 {
margin-left: 1em;
}
p,li,table,dl {
margin-left: 2em;
margin-right: 2em;
}
div.indent {
margin-left: 4em;
margin-right: 4em;
}
div.code {
border-style: solid;
border-width: 1px;
padding: 2pt;
margin-left: 4em;
margin-right: 4em;
background-color: #F0FFFF;
font-family: "Courier New", Courier, "Courier 10 Pitch", monospace;
}
div.targetlang {
border-style: solid;
border-width: 1px;
padding: 2pt;
margin-left: 4em;
margin-right: 4em;
background-color: #d7f6bb;
font-family: "Courier New", Courier, "Courier 10 Pitch", monospace;
}
div.shell {
border-style: solid;
border-width: 1px;
padding: 2pt;
margin-left: 4em;
margin-right: 4em;
background-color: #DCDCDC;
font-family: "Courier New", Courier, "Courier 10 Pitch", monospace;
}
div.diagram {
border-style: solid;
border-width: 1px;
padding: 2pt;
margin-left: 4em;
margin-right: 4em;
background-color: #FFEBCD;
font-family: "Courier New", Courier, "Courier 10 Pitch", monospace;
}
div.diagram li {
margin-left: 0;
}
ul li p {
margin-left: 0;
margin-right: 0;
}
ol li p {
margin-left: 0;
margin-right: 0;
}
dl dd p {
margin-left: 0;
margin-right: 0;
}
div.indent p {
margin-left: 0;
margin-right: 0;
}
pre, code, tt {
font-family: "Courier New", Courier, "Courier 10 Pitch", monospace;
}
body { font-family: serif; }
--></STYLE>
</HEAD>
<BODY BGCOLOR="#ffffff" LINK="#0000ff" VLINK="#0000ff" ALINK="#0000ff">
<CENTER><A HREF="#CONTENTS"><IMG SRC="swig16.png" BORDER="0" WIDTH="167" HEIGHT="85" ALT="SWIG-4.3 Documentation"><BR>
<H1>SWIG-4.3 Documentation</H1></A><BR>
</CENTER>
<HR NOSHADE>
<h1 ALIGN="CENTER"><a NAME="CONTENTS">Table of Contents</a></h1>
<br>
<br><b><a HREF="#Sections">SWIG-4.3 Documentation</a></b>
<ul>
<li><a HREF="#Sections_Sections">Sections</a>
<ul>
<li><a HREF="#Sections_core_docs">SWIG Core Documentation</a></li>
<li><a HREF="#Sections_language_modules">Supported Language Modules
Documentation</a></li>
<li><a HREF="#Sections_experimental_language_modules">Experimental
Language Modules Documentation</a></li>
<li><a HREF="#Sections_deprecated_language_modules">Deprecated Language
Modules Documentation</a></li>
<li><a HREF="#Sections_developers_docs">Developer Documentation</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Preface">1 Preface</a></b>
<ul>
<li><a HREF="#Preface_nn2">1.1 Introduction</a></li>
<li><a HREF="#Preface_nn4">1.2 SWIG Versions</a></li>
<li><a HREF="#Preface_license">1.3 SWIG License</a></li>
<li><a HREF="#Preface_nn5">1.4 SWIG resources</a></li>
<li><a HREF="#Preface_nn6">1.5 Prerequisites</a></li>
<li><a HREF="#Preface_nn7">1.6 Organization of this manual</a></li>
<li><a HREF="#Preface_nn8">1.7 How to avoid reading the manual</a></li>
<li><a HREF="#Preface_nn9">1.8 Backwards compatibility</a></li>
<li><a HREF="#Preface_release_notes">1.9 Release notes</a></li>
<li><a HREF="#Preface_nn10">1.10 Credits</a></li>
<li><a HREF="#Preface_nn11">1.11 Bug reports</a></li>
<li><a HREF="#Preface_installation">1.12 Installation</a>
<ul>
<li><a HREF="#Preface_windows_installation">1.12.1 Windows installation</a>
</li>
<li><a HREF="#Preface_unix_installation">1.12.2 Unix installation</a></li>
<li><a HREF="#Preface_osx_installation">1.12.3 Macintosh OS X
installation</a></li>
<li><a HREF="#Preface_testing">1.12.4 Testing</a></li>
<li><a HREF="#Preface_examples">1.12.5 Examples</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Introduction">2 Introduction</a></b>
<ul>
<li><a HREF="#Introduction_nn2">2.1 What is SWIG?</a></li>
<li><a HREF="#Introduction_nn3">2.2 Why use SWIG?</a></li>
<li><a HREF="#Introduction_target_languages">2.3 Target languages</a>
<ul>
<li><a HREF="#Introduction_supported_status">2.3.1 Supported status</a></li>
<li><a HREF="#Introduction_experimental_status">2.3.2 Experimental
status</a></li>
<li><a HREF="#Introduction_deprecated_status">2.3.3 Deprecated status</a>
</li>
</ul>
</li>
<li><a HREF="#Introduction_nn4">2.4 A SWIG example</a>
<ul>
<li><a HREF="#Introduction_nn5">2.4.1 SWIG interface file</a></li>
<li><a HREF="#Introduction_nn6">2.4.2 The swig command</a></li>
<li><a HREF="#Introduction_nn7">2.4.3 Building a Perl5 module</a></li>
<li><a HREF="#Introduction_nn8">2.4.4 Building a Python module</a></li>
<li><a HREF="#Introduction_nn9">2.4.5 Shortcuts</a></li>
</ul>
</li>
<li><a HREF="#Introduction_nn10">2.5 Supported C/C++ language features</a>
</li>
<li><a HREF="#Introduction_nn11">2.6 Non-intrusive interface building</a>
</li>
<li><a HREF="#Introduction_build_system">2.7 Incorporating SWIG into a
build system</a></li>
<li><a HREF="#Introduction_nn12">2.8 Hands off code generation</a></li>
<li><a HREF="#Introduction_nn13">2.9 SWIG and freedom</a></li>
</ul>
<b><a HREF="#Windows">3 Getting started on Windows</a></b>
<ul>
<li><a HREF="#Windows_installation">3.1 Installation on Windows</a>
<ul>
<li><a HREF="#Windows_executable">3.1.1 Windows Executable</a></li>
</ul>
</li>
<li><a HREF="#Windows_examples">3.2 SWIG Windows Examples</a>
<ul>
<li><a HREF="#Windows_visual_studio">3.2.1 Instructions for using the
Examples with Visual Studio</a>
<ul>
<li><a HREF="#Windows_csharp">3.2.1.1 C#</a></li>
<li><a HREF="#Windows_java">3.2.1.2 Java</a></li>
<li><a HREF="#Windows_python">3.2.1.3 Python</a></li>
<li><a HREF="#Windows_tcl">3.2.1.4 TCL</a></li>
</ul>
</li>
<li><a HREF="#Windows_other_compilers">3.2.2 Instructions for using the
Examples with other compilers</a></li>
</ul>
</li>
<li><a HREF="#Windows_swig_exe">3.3 Building swig.exe on Windows</a>
<ul>
<li><a HREF="#Windows_cmake">3.3.1 Building swig.exe using CMake</a></li>
<li><a HREF="#Windows_msys2">3.3.2 Building swig.exe using MSYS2 and
MinGW-w64</a></li>
<li><a HREF="#Windows_mingw_msys">3.3.3 Building swig.exe using MinGW
and MSYS</a></li>
<li><a HREF="#Windows_cygwin">3.3.4 Building swig.exe using Cygwin</a>
<ul>
<li><a HREF="#Windows_examples_cygwin">3.3.4.1 Running the examples on
Windows using Cygwin</a></li>
</ul>
</li>
</ul>
</li>
<li><a HREF="#Windows_interface_file">3.4 Microsoft extensions and other
Windows quirks</a>
<ul>
<li><a HREF="#Windows_msvc_cpp_standards">3.4.1 Visual C++ standards
compliance</a></li>
<li><a HREF="#Windows_calling_conventions">3.4.2 Calling conventions</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Scripting">4 Scripting Languages</a></b>
<ul>
<li><a HREF="#Scripting_nn2">4.1 The two language view of the world</a></li>
<li><a HREF="#Scripting_nn3">4.2 How does a scripting language talk to
C?</a>
<ul>
<li><a HREF="#Scripting_nn4">4.2.1 Wrapper functions</a></li>
<li><a HREF="#Scripting_nn5">4.2.2 Variable linking</a></li>
<li><a HREF="#Scripting_nn6">4.2.3 Constants</a></li>
<li><a HREF="#Scripting_nn7">4.2.4 Structures and classes</a></li>
<li><a HREF="#Scripting_nn8">4.2.5 Proxy classes</a></li>
</ul>
</li>
<li><a HREF="#Scripting_nn9">4.3 Building scripting language extensions</a>
<ul>
<li><a HREF="#Scripting_nn10">4.3.1 Shared libraries and dynamic loading</a>
</li>
<li><a HREF="#Scripting_nn11">4.3.2 Linking with shared libraries</a></li>
<li><a HREF="#Scripting_nn12">4.3.3 Static linking</a></li>
</ul>
</li>
</ul>
<b><a HREF="#SWIG">5 SWIG Basics</a></b>
<ul>
<li><a HREF="#SWIG_nn2">5.1 Running SWIG</a>
<ul>
<li><a HREF="#SWIG_nn3">5.1.1 Input format</a></li>
<li><a HREF="#SWIG_output">5.1.2 SWIG output</a></li>
<li><a HREF="#SWIG_nn5">5.1.3 Comments</a></li>
<li><a HREF="#SWIG_nn6">5.1.4 C Preprocessor</a></li>
<li><a HREF="#SWIG_nn7">5.1.5 SWIG directives</a></li>
<li><a HREF="#SWIG_nn8">5.1.6 Parser limitations</a></li>
<li><a HREF="#SWIG_parse_tree">5.1.7 Parse tree</a></li>
</ul>
</li>
<li><a HREF="#SWIG_nn9">5.2 Wrapping simple C declarations</a>
<ul>
<li><a HREF="#SWIG_nn10">5.2.1 Basic type handling</a></li>
<li><a HREF="#SWIG_nn11">5.2.2 Global variables</a></li>
<li><a HREF="#SWIG_nn12">5.2.3 Constants</a></li>
<li><a HREF="#SWIG_nn13">5.2.4 A brief word about const</a></li>
<li><a HREF="#SWIG_nn14">5.2.5 A cautionary tale of char *</a></li>
</ul>
</li>
<li><a HREF="#SWIG_nn15">5.3 Pointers and complex objects</a>
<ul>
<li><a HREF="#SWIG_nn16">5.3.1 Simple pointers</a></li>
<li><a HREF="#SWIG_nn17">5.3.2 Run time pointer type checking</a></li>
<li><a HREF="#SWIG_nn18">5.3.3 Derived types, structs, and classes</a></li>
<li><a HREF="#SWIG_nn19">5.3.4 Undefined datatypes</a></li>
<li><a HREF="#SWIG_nn20">5.3.5 Typedef</a></li>
</ul>
</li>
<li><a HREF="#SWIG_nn21">5.4 Other Practicalities</a>
<ul>
<li><a HREF="#SWIG_nn22">5.4.1 Passing structures by value</a></li>
<li><a HREF="#SWIG_nn23">5.4.2 Return by value</a></li>
<li><a HREF="#SWIG_nn24">5.4.3 Linking to structure variables</a></li>
<li><a HREF="#SWIG_nn25">5.4.4 Linking to char *</a></li>
<li><a HREF="#SWIG_nn26">5.4.5 Arrays</a></li>
<li><a HREF="#SWIG_readonly_variables">5.4.6 Creating read-only
variables</a></li>
<li><a HREF="#SWIG_rename_ignore">5.4.7 Renaming and ignoring
declarations</a>
<ul>
<li><a HREF="#SWIG_nn29">5.4.7.1 Simple renaming of specific identifiers</a>
</li>
<li><a HREF="#SWIG_ignore">5.4.7.2 Ignoring identifiers</a></li>
<li><a HREF="#SWIG_advanced_renaming">5.4.7.3 Advanced renaming support</a>
</li>
<li><a HREF="#SWIG_limiting_renaming">5.4.7.4 Limiting global renaming
rules</a></li>
<li><a HREF="#SWIG_chosen_unignore">5.4.7.5 Ignoring everything then
wrapping a few selected symbols</a></li>
</ul>
</li>
<li><a HREF="#SWIG_default_args">5.4.8 Default/optional arguments</a></li>
<li><a HREF="#SWIG_nn30">5.4.9 Pointers to functions and callbacks</a></li>
</ul>
</li>
<li><a HREF="#SWIG_nn31">5.5 Structures and unions</a>
<ul>
<li><a HREF="#SWIG_nn32">5.5.1 Typedef and structures</a></li>
<li><a HREF="#SWIG_nn33">5.5.2 Character strings and structures</a></li>
<li><a HREF="#SWIG_nn34">5.5.3 Array members</a></li>
<li><a HREF="#SWIG_structure_data_members">5.5.4 Structure data members</a>
</li>
<li><a HREF="#SWIG_nn36">5.5.5 C constructors and destructors</a></li>
<li><a HREF="#SWIG_adding_member_functions">5.5.6 Adding member
functions to C structures</a></li>
<li><a HREF="#SWIG_nested_structs">5.5.7 Nested structures</a></li>
<li><a HREF="#SWIG_nn39">5.5.8 Other things to note about structure
wrapping</a></li>
</ul>
</li>
<li><a HREF="#SWIG_nn40">5.6 Code Insertion</a>
<ul>
<li><a HREF="#SWIG_nn41">5.6.1 The output of SWIG</a></li>
<li><a HREF="#SWIG_nn42">5.6.2 Code insertion blocks</a></li>
<li><a HREF="#SWIG_nn43">5.6.3 Inlined code blocks</a></li>
<li><a HREF="#SWIG_nn44">5.6.4 Initialization blocks</a></li>
</ul>
</li>
<li><a HREF="#SWIG_nn45">5.7 An Interface Building Strategy</a>
<ul>
<li><a HREF="#SWIG_nn46">5.7.1 Preparing a C program for SWIG</a></li>
<li><a HREF="#SWIG_nn47">5.7.2 The SWIG interface file</a></li>
<li><a HREF="#SWIG_nn48">5.7.3 Why use separate interface files?</a></li>
<li><a HREF="#SWIG_nn49">5.7.4 Getting the right header files</a></li>
<li><a HREF="#SWIG_nn50">5.7.5 What to do with main()</a></li>
</ul>
</li>
</ul>
<b><a HREF="#SWIGPlus">6 SWIG and C++</a></b>
<ul>
<li><a HREF="#SWIGPlus_nn2">6.1 Comments on C++ Wrapping</a></li>
<li><a HREF="#SWIGPlus_nn3">6.2 Approach</a></li>
<li><a HREF="#SWIGPlus_nn4">6.3 Supported C++ features</a></li>
<li><a HREF="#SWIGPlus_nn5">6.4 Command line options and compilation</a></li>
<li><a HREF="#SWIGPlus_nn38">6.5 Proxy classes</a>
<ul>
<li><a HREF="#SWIGPlus_nn39">6.5.1 Construction of proxy classes</a></li>
<li><a HREF="#SWIGPlus_nn40">6.5.2 Resource management in proxies</a></li>
<li><a HREF="#SWIGPlus_nn41">6.5.3 Language specific details</a></li>
</ul>
</li>
<li><a HREF="#SWIGPlus_nn6">6.6 Simple C++ wrapping</a>
<ul>
<li><a HREF="#SWIGPlus_nn7">6.6.1 Constructors and destructors</a></li>
<li><a HREF="#SWIGPlus_nn8">6.6.2 Default constructors, copy
constructors and implicit destructors</a></li>
<li><a HREF="#SWIGPlus_nn9">6.6.3 When constructor wrappers aren't
created</a></li>
<li><a HREF="#SWIGPlus_nn10">6.6.4 Copy constructors</a></li>
<li><a HREF="#SWIGPlus_nn11">6.6.5 Member functions</a></li>
<li><a HREF="#SWIGPlus_nn12">6.6.6 Static members</a></li>
<li><a HREF="#SWIGPlus_member_data">6.6.7 Member data</a></li>
</ul>
</li>
<li><a HREF="#SWIGPlus_nn15">6.7 Protection</a></li>
<li><a HREF="#SWIGPlus_nn16">6.8 Enums and constants</a></li>
<li><a HREF="#SWIGPlus_nn17">6.9 Friends</a>
<ul>
<li><a HREF="#SWIGPlus_friend_classes">6.9.1 Friend classes</a></li>
<li><a HREF="#SWIGPlus_friend_function_definitions">6.9.2 Friend
function definitions</a></li>
<li><a HREF="#SWIGPlus_friend_function_declarations">6.9.3 Friend
function declarations</a></li>
<li><a HREF="#SWIGPlus_friends_unqualified">6.9.4 Unqualified friend
functions</a></li>
</ul>
</li>
<li><a HREF="#SWIGPlus_nn18">6.10 References and pointers</a></li>
<li><a HREF="#SWIGPlus_nn19">6.11 Pass and return by value</a></li>
<li><a HREF="#SWIGPlus_nn20">6.12 Inheritance</a></li>
<li><a HREF="#SWIGPlus_nn21">6.13 A brief discussion of multiple
inheritance, pointers, and type checking</a></li>
<li><a HREF="#SWIGPlus_default_args">6.14 Default arguments</a></li>
<li><a HREF="#SWIGPlus_overloaded_methods">6.15 Overloaded functions and
methods</a>
<ul>
<li><a HREF="#SWIGPlus_nn24">6.15.1 Dispatch function generation</a></li>
<li><a HREF="#SWIGPlus_nn25">6.15.2 Ambiguity in overloading</a></li>
<li><a HREF="#SWIGPlus_ambiguity_resolution_renaming">6.15.3 Renaming
and ambiguity resolution</a></li>
<li><a HREF="#SWIGPlus_nn27">6.15.4 Comments on overloading</a></li>
</ul>
</li>
<li><a HREF="#SWIGPlus_nn28">6.16 Overloaded operators</a></li>
<li><a HREF="#SWIGPlus_class_extension">6.17 Class extension</a>
<ul>
<li><a HREF="#SWIGPlus_replacing_methods">6.17.1 Replacing class methods</a>
</li>
</ul>
</li>
<li><a HREF="#SWIGPlus_nn30">6.18 Templates</a>
<ul>
<li><a HREF="#SWIGPlus_template_directive">6.18.1 The %template
directive</a></li>
<li><a HREF="#SWIGPlus_template_functions">6.18.2 Function templates</a></li>
<li><a HREF="#SWIGPlus_template_classes">6.18.3 Default template
arguments</a></li>
<li><a HREF="#SWIGPlus_template_class_inheritance">6.18.4 Template base
classes</a></li>
<li><a HREF="#SWIGPlus_template_empty">6.18.5 Empty template
instantiation</a></li>
<li><a HREF="#SWIGPlus_template_specialization">6.18.6 Template
specialization</a></li>
<li><a HREF="#SWIGPlus_template_member">6.18.7 Member templates</a></li>
<li><a HREF="#SWIGPlus_template_scoping">6.18.8 Scoping and templates</a>
</li>
<li><a HREF="#SWIGPlus_template_more">6.18.9 More on templates</a></li>
</ul>
</li>
<li><a HREF="#SWIGPlus_namespaces">6.19 Namespaces</a>
<ul>
<li><a HREF="#SWIGPlus_nspace">6.19.1 The nspace feature for namespaces</a>
<ul>
<li><a HREF="#SWIGPlus_nspace_feature_flag">6.19.1.1 %nspace for
mirroring namespace hierarchies</a></li>
<li><a HREF="#SWIGPlus_nspacemove">6.19.1.2 %nspacemove for modifying
namespace hierarchies</a></li>
<li><a HREF="#SWIGPlus_nspace_more">6.19.1.3 More about the nspace
feature</a></li>
</ul>
</li>
</ul>
</li>
<li><a HREF="#SWIGPlus_renaming_templated_types_namespaces">6.20
Renaming templated types in namespaces</a></li>
<li><a HREF="#SWIGPlus_exception_specifications">6.21 Exception
specifications</a></li>
<li><a HREF="#SWIGPlus_catches">6.22 Exception handling with %catches</a>
</li>
<li><a HREF="#SWIGPlus_nn33">6.23 Pointers to Members</a></li>
<li><a HREF="#SWIGPlus_smart_pointers">6.24 Smart pointers and
operator-&gt;()</a></li>
<li><a HREF="#SWIGPlus_ref_unref">6.25 C++ reference counted objects -
ref/unref feature</a></li>
<li><a HREF="#SWIGPlus_nn35">6.26 Using declarations and inheritance</a></li>
<li><a HREF="#SWIGPlus_nested_classes">6.27 Nested classes</a></li>
<li><a HREF="#SWIGPlus_const">6.28 A brief rant about const-correctness</a>
</li>
<li><a HREF="#SWIGPlus_target_language_callbacks">6.29 Callbacks to the
target language</a>
<ul>
<li><a HREF="#SWIGPlus_director_classes_introduction">6.29.1
Introduction to director classes</a></li>
<li><a HREF="#SWIGPlus_directors_for_function_pointers">6.29.2 Using
directors and target language callbacks</a></li>
</ul>
</li>
<li><a HREF="#SWIGPlus_nn42">6.30 Where to go for more information</a></li>
</ul>
<b><a HREF="#CPlusPlus11">7 SWIG and C++11</a></b>
<ul>
<li><a HREF="#CPlusPlus11_introduction">7.1 Introduction</a></li>
<li><a HREF="#CPlusPlus11_core_language_changes">7.2 Core language
changes</a>
<ul>
<li><a HREF="#CPlusPlus11_rvalue_reference_and_move_semantics">7.2.1
Rvalue reference and move semantics</a>
<ul>
<li><a HREF="#CPlusPlus11_rvalue_reference_inputs">7.2.1.1 Rvalue
reference inputs</a></li>
<li><a HREF="#CPlusPlus11_rvalue_reference_outputs">7.2.1.2 Rvalue
reference outputs</a></li>
<li><a HREF="#CPlusPlus11_move_only">7.2.1.3 Movable and move-only types
by value</a></li>
</ul>
</li>
<li><a HREF="#CPlusPlus11_generalized_constant_expressions">7.2.2
Generalized constant expressions</a></li>
<li><a HREF="#CPlusPlus11_extern_template">7.2.3 Extern template</a></li>
<li><a HREF="#CPlusPlus11_initializer_lists">7.2.4 Initializer lists</a></li>
<li><a HREF="#CPlusPlus11_uniform_initialization">7.2.5 Uniform
initialization</a></li>
<li><a HREF="#CPlusPlus11_type_inference">7.2.6 Type inference</a></li>
<li><a HREF="#CPlusPlus11_range_based_for_loop">7.2.7 Range-based
for-loop</a></li>
<li><a HREF="#CPlusPlus11_lambda_functions_and_expressions">7.2.8 Lambda
functions and expressions</a></li>
<li><a HREF="#CPlusPlus11_alternate_function_syntax">7.2.9 Alternate
function syntax</a></li>
<li><a HREF="#CPlusPlus11_object_construction_improvement">7.2.10 Object
construction improvement</a></li>
<li><a HREF="#CPlusPlus11_explicit_overrides_final">7.2.11 Explicit
overrides and final</a></li>
<li><a HREF="#CPlusPlus11_null_pointer_constant">7.2.12 Null pointer
constant</a></li>
<li><a HREF="#CPlusPlus11_strongly_typed_enumerations">7.2.13 Strongly
typed enumerations</a></li>
<li><a HREF="#CPlusPlus11_double_angle_brackets">7.2.14 Double angle
brackets</a></li>
<li><a HREF="#CPlusPlus11_explicit_conversion_operators">7.2.15 Explicit
conversion operators</a></li>
<li><a HREF="#CPlusPlus11_alias_templates">7.2.16 Type alias and alias
templates</a></li>
<li><a HREF="#CPlusPlus11_unrestricted_unions">7.2.17 Unrestricted
unions</a></li>
<li><a HREF="#CPlusPlus11_variadic_templates">7.2.18 Variadic templates</a>
</li>
<li><a HREF="#CPlusPlus11_new_char_literals">7.2.19 New character
literals</a></li>
<li><a HREF="#CPlusPlus11_new_string_literals">7.2.20 New string
literals</a></li>
<li><a HREF="#CPlusPlus11_user_defined_literals">7.2.21 User-defined
literals</a></li>
<li><a HREF="#CPlusPlus11_thread_local_storage">7.2.22 Thread-local
storage</a></li>
<li><a HREF="#CPlusPlus11_defaulted_deleted">7.2.23 Explicitly defaulted
functions and deleted functions</a></li>
<li><a HREF="#CPlusPlus11_type_long_long_int">7.2.24 Type long long int</a>
</li>
<li><a HREF="#CPlusPlus11_static_assertions">7.2.25 Static assertions</a>
</li>
<li><a HREF="#CPlusPlus11_sizeof">7.2.26 Allow sizeof to work on members
of classes without an explicit object</a></li>
<li><a HREF="#CPlusPlus11_noexcept">7.2.27 Exception specifications and
noexcept</a></li>
<li><a HREF="#CPlusPlus11_alignment">7.2.28 Control and query object
alignment</a></li>
<li><a HREF="#CPlusPlus11_attributes">7.2.29 Attributes</a></li>
<li><a HREF="#CPlusPlus11_ref_qualifiers">7.2.30 Methods with
ref-qualifiers</a></li>
</ul>
</li>
<li><a HREF="#CPlusPlus11_standard_library_changes">7.3 Standard library
changes</a>
<ul>
<li><a HREF="#CPlusPlus11_threading_facilities">7.3.1 Threading
facilities</a></li>
<li><a HREF="#CPlusPlus11_tuple_types">7.3.2 Tuple types</a></li>
<li><a HREF="#CPlusPlus11_hash_tables">7.3.3 Hash tables</a></li>
<li><a HREF="#CPlusPlus11_regular_expressions">7.3.4 Regular expressions</a>
</li>
<li><a HREF="#CPlusPlus11_general_purpose_smart_pointers">7.3.5
General-purpose smart pointers</a></li>
<li><a HREF="#CPlusPlus11_extensible_random_number_facility">7.3.6
Extensible random number facility</a></li>
<li><a HREF="#CPlusPlus11_wrapper_reference">7.3.7 Wrapper reference</a></li>
<li><a HREF="#CPlusPlus11_polymorphous_wrappers_for_function_objects">
7.3.8 Polymorphic wrappers for function objects</a></li>
<li><a HREF="#CPlusPlus11_type_traits_for_metaprogramming">7.3.9 Type
traits for metaprogramming</a></li>
<li><a HREF="#CPlusPlus11_uniform_method_for_computing_return_type_of_function_objects">
7.3.10 Uniform method for computing return type of function objects</a></li>
</ul>
</li>
</ul>
<b><a HREF="#CPlusPlus14">8 SWIG and C++14</a></b>
<ul>
<li><a HREF="#CPlusPlus14_introduction">8.1 Introduction</a></li>
<li><a HREF="#CPlusPlus14_core_language_changes">8.2 Core language
changes</a>
<ul>
<li><a HREF="#CPlusPlus14_binary_literals">8.2.1 Binary integer literals</a>
</li>
<li><a HREF="#CPlusPlus14_return_type_deduction">8.2.2 Return type
deduction</a></li>
</ul>
</li>
<li><a HREF="#CPlusPlus14_standard_library_changes">8.3 Standard library
changes</a></li>
</ul>
<b><a HREF="#CPlusPlus17">9 SWIG and C++17</a></b>
<ul>
<li><a HREF="#CPlusPlus17_introduction">9.1 Introduction</a></li>
<li><a HREF="#CPlusPlus17_core_language_changes">9.2 Core language
changes</a>
<ul>
<li><a HREF="#CPlusPlus17_nested_namespaces">9.2.1 Nested namespace
definitions</a></li>
<li><a HREF="#CPlusPlus17_u8_char_literals">9.2.2 UTF-8 character
literals</a></li>
<li><a HREF="#CPlusPlus17_hexadecimal_floating_literals">9.2.3
Hexadecimal floating literals</a></li>
<li><a HREF="#CPlusPlus17_fold_expressions">9.2.4 Fold expressions</a></li>
</ul>
</li>
<li><a HREF="#CPlusPlus17_standard_library_changes">9.3 Standard library
changes</a></li>
</ul>
<b><a HREF="#CPlusPlus20">10 SWIG and C++20</a></b>
<ul>
<li><a HREF="#CPlusPlus20_introduction">10.1 Introduction</a></li>
<li><a HREF="#CPlusPlus20_core_language_changes">10.2 Core language
changes</a>
<ul>
<li><a HREF="#CPlusPlus20_spaceship_operator">10.2.1 Spaceship operator</a>
</li>
<li><a HREF="#CPlusPlus20_lambda_templates">10.2.2 Lambda templates</a></li>
<li><a HREF="#CPlusPlus20_constexpr_destructors">10.2.3 Constexpr
destructors</a></li>
</ul>
</li>
<li><a HREF="#CPlusPlus20_preprocessor_changes">10.3 Preprocessor
changes</a>
<ul>
<li><a HREF="#CPlusPlus20_va_opt">10.3.1 __VA_OPT__()</a></li>
</ul>
</li>
<li><a HREF="#CPlusPlus20_standard_library_changes">10.4 Standard
library changes</a></li>
</ul>
<b><a HREF="#Preprocessor">11 Preprocessing</a></b>
<ul>
<li><a HREF="#Preprocessor_nn2">11.1 File inclusion</a></li>
<li><a HREF="#Preprocessor_nn3">11.2 File imports</a></li>
<li><a HREF="#Preprocessor_condition_compilation">11.3 Conditional
Compilation</a></li>
<li><a HREF="#Preprocessor_nn5">11.4 Macro Expansion</a></li>
<li><a HREF="#Preprocessor_nn6">11.5 SWIG Macros</a></li>
<li><a HREF="#Preprocessor_nn7">11.6 Variadic Macros</a></li>
<li><a HREF="#Preprocessor_delimiters">11.7 Preprocessing and delimiters</a>
<ul>
<li><a HREF="#Preprocessor_nn8">11.7.1 Preprocessing and %{ ... %} &amp; &quot;
... &quot; delimiters</a></li>
<li><a HREF="#Preprocessor_nn9">11.7.2 Preprocessing and { ... }
delimiters</a></li>
</ul>
</li>
<li><a HREF="#Preprocessor_typemap_delimiters">11.8 Preprocessor and
Typemaps</a></li>
<li><a HREF="#Preprocessor_nn10">11.9 Viewing preprocessor output</a></li>
<li><a HREF="#Preprocessor_warning_error">11.10 The #error and #warning
directives</a></li>
<li><a HREF="#Preprocessor_trigraphs">11.11 Trigraphs</a></li>
<li><a HREF="#Preprocessor_digraphs">11.12 Digraphs</a></li>
</ul>
<b><a HREF="#Library">12 SWIG library</a></b>
<ul>
<li><a HREF="#Library_nn2">12.1 The %include directive and library
search path</a></li>
<li><a HREF="#Library_nn3">12.2 C arrays and pointers</a>
<ul>
<li><a HREF="#Library_argcargv">12.2.1 argcargv.i</a></li>
<li><a HREF="#Library_nn4">12.2.2 cpointer.i</a></li>
<li><a HREF="#Library_carrays">12.2.3 carrays.i</a></li>
<li><a HREF="#Library_nn6">12.2.4 cmalloc.i</a></li>
<li><a HREF="#Library_nn7">12.2.5 cdata.i</a></li>
</ul>
</li>
<li><a HREF="#Library_nn8">12.3 C string handling</a>
<ul>
<li><a HREF="#Library_nn9">12.3.1 Default string handling</a></li>
<li><a HREF="#Library_nn10">12.3.2 Passing a string with length</a></li>
<li><a HREF="#Library_nn11">12.3.3 Using %newobject to release memory</a>
</li>
<li><a HREF="#Library_nn12">12.3.4 cstring.i</a></li>
</ul>
</li>
<li><a HREF="#Library_c_standard_library">12.4 C standard library</a>
<ul>
<li><a HREF="#Library_complex">12.4.1 Complex floating types</a></li>
</ul>
</li>
<li><a HREF="#Library_stl_cpp_library">12.5 STL/C++ library</a>
<ul>
<li><a HREF="#Library_std_string">12.5.1 std::string</a></li>
<li><a HREF="#Library_std_string_view">12.5.2 std::string_view</a></li>
<li><a HREF="#Library_std_vector">12.5.3 std::vector</a></li>
<li><a HREF="#Library_stl_exceptions">12.5.4 STL exceptions</a></li>
<li><a HREF="#Library_std_shared_ptr">12.5.5 shared_ptr smart pointer</a>
<ul>
<li><a HREF="#Library_shared_ptr_basics">12.5.5.1 shared_ptr basics</a></li>
<li><a HREF="#Library_shared_ptr_inheritance">12.5.5.2 shared_ptr and
inheritance</a></li>
<li><a HREF="#Library_shared_ptr_overloading">12.5.5.3 shared_ptr and
method overloading</a></li>
<li><a HREF="#Library_shared_ptr_templates">12.5.5.4 shared_ptr and
templates</a></li>
<li><a HREF="#Library_shared_ptr_directors">12.5.5.5 shared_ptr and
directors</a></li>
</ul>
</li>
<li><a HREF="#Library_std_unique_ptr">12.5.6 unique_ptr smart pointer</a>
<ul>
<li><a HREF="#Library_std_unique_ptr_by_value">12.5.6.1 unique_ptr
passed by value</a></li>
<li><a HREF="#Library_std_unique_ptr_by_ref">12.5.6.2 unique_ptr passed
by reference</a></li>
</ul>
</li>
<li><a HREF="#Library_std_auto_ptr">12.5.7 auto_ptr smart pointer</a></li>
</ul>
</li>
<li><a HREF="#Library_nn16">12.6 Utility Libraries</a>
<ul>
<li><a HREF="#Library_nn17">12.6.1 exception.i</a></li>
<li><a HREF="#Library_attributes">12.6.2 attribute.i</a>
<ul>
<li><a HREF="#Library_attribute_templates">12.6.2.1 %attribute and C++
templates</a></li>
</ul>
</li>
</ul>
</li>
</ul>
<b><a HREF="#Arguments">13 Argument Handling</a></b>
<ul>
<li><a HREF="#Arguments_nn2">13.1 The typemaps.i library</a>
<ul>
<li><a HREF="#Arguments_nn3">13.1.1 Introduction</a></li>
<li><a HREF="#Arguments_nn4">13.1.2 Input parameters</a></li>
<li><a HREF="#Arguments_nn5">13.1.3 Output parameters</a></li>
<li><a HREF="#Arguments_nn6">13.1.4 Input/Output parameters</a></li>
<li><a HREF="#Arguments_nn7">13.1.5 Using different names</a></li>
</ul>
</li>
<li><a HREF="#Arguments_nn8">13.2 Applying constraints to input values</a>
<ul>
<li><a HREF="#Arguments_nn9">13.2.1 Simple constraint example</a></li>
<li><a HREF="#Arguments_nn10">13.2.2 Constraint methods</a></li>
<li><a HREF="#Arguments_nn11">13.2.3 Applying constraints to new
datatypes</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Typemaps">14 Typemaps</a></b>
<ul>
<li><a HREF="#Typemaps_nn2">14.1 Introduction</a>
<ul>
<li><a HREF="#Typemaps_nn3">14.1.1 Type conversion</a></li>
<li><a HREF="#Typemaps_nn4">14.1.2 Typemaps</a></li>
<li><a HREF="#Typemaps_nn5">14.1.3 Pattern matching</a></li>
<li><a HREF="#Typemaps_nn6">14.1.4 Reusing typemaps</a></li>
<li><a HREF="#Typemaps_nn7">14.1.5 What can be done with typemaps?</a></li>
<li><a HREF="#Typemaps_nn8">14.1.6 What can't be done with typemaps?</a></li>
<li><a HREF="#Typemaps_aspects">14.1.7 Similarities to Aspect Oriented
Programming</a></li>
<li><a HREF="#Typemaps_nn9">14.1.8 The rest of this chapter</a></li>
</ul>
</li>
<li><a HREF="#Typemaps_nn10">14.2 Typemap specifications</a>
<ul>
<li><a HREF="#Typemaps_defining">14.2.1 Defining a typemap</a></li>
<li><a HREF="#Typemaps_nn12">14.2.2 Typemap scope</a></li>
<li><a HREF="#Typemaps_nn13">14.2.3 Copying a typemap</a></li>
<li><a HREF="#Typemaps_nn14">14.2.4 Deleting a typemap</a></li>
<li><a HREF="#Typemaps_nn15">14.2.5 Placement of typemaps</a></li>
</ul>
</li>
<li><a HREF="#Typemaps_pattern_matching">14.3 Pattern matching rules</a>
<ul>
<li><a HREF="#Typemaps_nn17">14.3.1 Basic matching rules</a></li>
<li><a HREF="#Typemaps_typedef_reductions">14.3.2 Typedef reductions
matching</a></li>
<li><a HREF="#Typemaps_nn19">14.3.3 Default typemap matching rules</a></li>
<li><a HREF="#Typemaps_multi_argument_typemaps_patterns">14.3.4
Multi-arguments typemaps</a></li>
<li><a HREF="#Typemaps_matching_template_comparison">14.3.5 Matching
rules compared to C++ templates</a></li>
<li><a HREF="#Typemaps_debugging_search">14.3.6 Debugging typemap
pattern matching</a></li>
</ul>
</li>
<li><a HREF="#Typemaps_nn21">14.4 Code generation rules</a>
<ul>
<li><a HREF="#Typemaps_nn22">14.4.1 Scope</a></li>
<li><a HREF="#Typemaps_nn23">14.4.2 Declaring new local variables</a></li>
<li><a HREF="#Typemaps_special_variables">14.4.3 Special variables</a>
<ul>
<li><a HREF="#Typemaps_type_special_variables">14.4.3.1 Type related
special variables</a></li>
<li><a HREF="#Typemaps_non_type_special_variables">14.4.3.2 Non-type
related special variables</a></li>
</ul>
</li>
<li><a HREF="#Typemaps_special_variable_macros">14.4.4 Special variable
macros</a>
<ul>
<li><a HREF="#Typemaps_special_macro_descriptor">14.4.4.1
$descriptor(type)</a></li>
<li><a HREF="#Typemaps_special_macro_typemap">14.4.4.2 $typemap(method,
typepattern)</a></li>
<li><a HREF="#Typemaps_special_macro_typemap_attribute">14.4.4.3
$typemap(method:attribute, typepattern)</a></li>
</ul>
</li>
<li><a HREF="#Typemaps_special_variable_attributes">14.4.5 Special
variables and typemap attributes</a></li>
<li><a HREF="#Typemaps_special_variables_and_macros">14.4.6 Special
variables combined with special variable macros</a></li>
</ul>
</li>
<li><a HREF="#Typemaps_nn25">14.5 Common typemap methods</a>
<ul>
<li><a HREF="#Typemaps_nn26">14.5.1 &quot;in&quot; typemap</a></li>
<li><a HREF="#Typemaps_nn27">14.5.2 &quot;typecheck&quot; typemap</a></li>
<li><a HREF="#Typemaps_nn28">14.5.3 &quot;out&quot; typemap</a></li>
<li><a HREF="#Typemaps_nn29">14.5.4 &quot;arginit&quot; typemap</a></li>
<li><a HREF="#Typemaps_nn30">14.5.5 &quot;default&quot; typemap</a></li>
<li><a HREF="#Typemaps_nn31">14.5.6 &quot;check&quot; typemap</a></li>
<li><a HREF="#Typemaps_nn32">14.5.7 &quot;argout&quot; typemap</a></li>
<li><a HREF="#Typemaps_nn33">14.5.8 &quot;freearg&quot; typemap</a></li>
<li><a HREF="#Typemaps_nn34">14.5.9 &quot;newfree&quot; typemap</a></li>
<li><a HREF="#Typemaps_ret">14.5.10 &quot;ret&quot; typemap</a></li>
<li><a HREF="#Typemaps_nn35">14.5.11 &quot;memberin&quot; typemap</a></li>
<li><a HREF="#Typemaps_nn36">14.5.12 &quot;varin&quot; typemap</a></li>
<li><a HREF="#Typemaps_nn37">14.5.13 &quot;varout&quot; typemap</a></li>
<li><a HREF="#Typemaps_throws_typemap">14.5.14 &quot;throws&quot; typemap</a></li>
</ul>
</li>
<li><a HREF="#Typemaps_nn39">14.6 Some typemap examples</a>
<ul>
<li><a HREF="#Typemaps_nn40">14.6.1 Typemaps for arrays</a></li>
<li><a HREF="#Typemaps_nn41">14.6.2 Implementing constraints with
typemaps</a></li>
</ul>
</li>
<li><a HREF="#Typemaps_nn43">14.7 Typemaps for multiple target languages</a>
</li>
<li><a HREF="#Typemaps_optimal">14.8 Optimal code generation when
returning by value</a></li>
<li><a HREF="#Typemaps_multi_argument_typemaps">14.9 Multi-argument
typemaps</a></li>
<li><a HREF="#Typemaps_warnings">14.10 Typemap warnings</a></li>
<li><a HREF="#Typemaps_fragments">14.11 Typemap fragments</a>
<ul>
<li><a HREF="#Typemaps_fragment_type_specialization">14.11.1 Fragment
type specialization</a></li>
<li><a HREF="#Typemaps_automatic_specialization">14.11.2 Fragments and
automatic typemap specialization</a></li>
</ul>
</li>
<li><a HREF="#Typemaps_runtime_type_checker">14.12 The run-time type
checker</a>
<ul>
<li><a HREF="#Typemaps_nn45">14.12.1 Implementation</a></li>
<li><a HREF="#Typemaps_runtime_type_checker_usage">14.12.2 Usage</a></li>
</ul>
</li>
<li><a HREF="#Typemaps_overloading">14.13 Typemaps and overloading</a>
<ul>
<li><a HREF="#Typemaps_typecheck_pointer">14.13.1 SWIG_TYPECHECK_POINTER
precedence level and the typecheck typemap</a></li>
</ul>
</li>
<li><a HREF="#Typemaps_nn48">14.14 More about %apply and %clear</a></li>
<li><a HREF="#Typemaps_nn47">14.15 Passing data between typemaps</a></li>
<li><a HREF="#Typemaps_nn52">14.16 C++ &quot;this&quot; pointer</a></li>
<li><a HREF="#Typemaps_nn51">14.17 Where to go for more information?</a></li>
</ul>
<b><a HREF="#Customization">15 Customization Features</a></b>
<ul>
<li><a HREF="#Customization_exception">15.1 Exception handling with
%exception</a>
<ul>
<li><a HREF="#Customization_nn3">15.1.1 Handling exceptions in C code</a>
</li>
<li><a HREF="#Customization_nn4">15.1.2 Exception handling with
longjmp()</a></li>
<li><a HREF="#Customization_nn5">15.1.3 Handling C++ exceptions</a></li>
<li><a HREF="#Customization_allowexcept">15.1.4 Exception handlers for
variables</a></li>
<li><a HREF="#Customization_nn6">15.1.5 Defining different exception
handlers</a></li>
<li><a HREF="#Customization_exception_special_variables">15.1.6 Special
variables for %exception</a></li>
<li><a HREF="#Customization_nn7">15.1.7 Using The SWIG exception library</a>
</li>
</ul>
</li>
<li><a HREF="#Customization_ownership">15.2 Object ownership and
%newobject</a></li>
<li><a HREF="#Customization_features">15.3 Features and the %feature
directive</a>
<ul>
<li><a HREF="#Customization_feature_attributes">15.3.1 Feature
attributes</a></li>
<li><a HREF="#Customization_feature_flags">15.3.2 Feature flags</a></li>
<li><a HREF="#Customization_clearing_features">15.3.3 Clearing features</a>
</li>
<li><a HREF="#Customization_features_default_args">15.3.4 Features and
default arguments</a></li>
<li><a HREF="#Customization_features_example">15.3.5 Feature example</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Contract">16 Contracts</a></b>
<ul>
<li><a HREF="#Contract_nn2">16.1 The %contract directive</a></li>
<li><a HREF="#Contract_nn3">16.2 %contract and classes</a></li>
<li><a HREF="#Contract_nn4">16.3 Constant aggregation and
%aggregate_check</a></li>
<li><a HREF="#Contract_nn5">16.4 Notes</a></li>
</ul>
<b><a HREF="#Varargs">17 Variable Length Arguments</a></b>
<ul>
<li><a HREF="#Varargs_nn2">17.1 Introduction</a></li>
<li><a HREF="#Varargs_nn3">17.2 The Problem</a></li>
<li><a HREF="#Varargs_nn4">17.3 Default varargs support</a></li>
<li><a HREF="#Varargs_nn5">17.4 Argument replacement using %varargs</a></li>
<li><a HREF="#Varargs_nn6">17.5 Varargs and typemaps</a></li>
<li><a HREF="#Varargs_nn7">17.6 Varargs wrapping with libffi</a></li>
<li><a HREF="#Varargs_nn8">17.7 Wrapping of va_list</a></li>
<li><a HREF="#Varargs_nn9">17.8 C++ Issues</a></li>
<li><a HREF="#Varargs_nn10">17.9 Discussion</a></li>
</ul>
<b><a HREF="#Doxygen">18 SWIG and Doxygen Translation</a></b>
<ul>
<li><a HREF="#Doxygen_translation_overview">18.1 Doxygen translation
overview</a></li>
<li><a HREF="#Doxygen_file_preparation">18.2 Preparations</a>
<ul>
<li><a HREF="#Doxygen_running_swig">18.2.1 Enabling Doxygen translation</a>
</li>
<li><a HREF="#Doxygen_features">18.2.2 Doxygen-specific %feature
directives</a>
<ul>
<li><a HREF="#Doxygen_notranslate">18.2.2.1 doxygen:notranslate</a></li>
<li><a HREF="#Doxygen_alias">18.2.2.2 doxygen:alias:&lt;command-name&gt;</a></li>
<li><a HREF="#Doxygen_ignore">18.2.2.3 doxygen:ignore:&lt;command-name&gt;</a></li>
<li><a HREF="#Doxygen_nolinktranslate">18.2.2.4 doxygen:nolinktranslate</a>
</li>
<li><a HREF="#Doxygen_nostripparams">18.2.2.5 doxygen:nostripparams</a></li>
</ul>
</li>
<li><a HREF="#Doxygen_additional_options">18.2.3 Additional command line
options</a></li>
</ul>
</li>
<li><a HREF="#Doxygen_javadoc">18.3 Doxygen to Javadoc</a>
<ul>
<li><a HREF="#Doxygen_javadoc_basic_example">18.3.1 Basic Javadoc
example</a></li>
<li><a HREF="#Doxygen_javadoc_tags">18.3.2 Javadoc tags</a></li>
<li><a HREF="#Doxygen_javadoc_unsupported_tags">18.3.3 Unsupported tags
for Javadoc</a></li>
</ul>
</li>
<li><a HREF="#Doxygen_pydoc">18.4 Doxygen to Pydoc</a>
<ul>
<li><a HREF="#Doxygen_pydoc_basic_example">18.4.1 Basic Pydoc example</a>
</li>
<li><a HREF="#Doxygen_pydoc_tags">18.4.2 Pydoc translator</a></li>
<li><a HREF="#Doxygen_pydoc_unsupported_tags">18.4.3 Unsupported tags
for Pydoc</a></li>
</ul>
</li>
<li><a HREF="#Doxygen_csharp">18.5 Doxygen to XML C# documentation</a>
<ul>
<li><a HREF="#Doxygen_csharp_basic_example">18.5.1 Basic C# example</a></li>
<li><a HREF="#Doxygen_csharp_tags">18.5.2 C# tags</a></li>
<li><a HREF="#Doxygen_csharp_unsupported_tags">18.5.3 Unsupported tags
for C#</a></li>
</ul>
</li>
<li><a HREF="#Doxygen_troubleshooting">18.6 Troubleshooting</a>
<ul>
<li><a HREF="#troubleshooting_ifndef">18.6.1 Problem with conditional
compilation</a></li>
</ul>
</li>
<li><a HREF="#Doxygen_developer_details">18.7 Developer information</a>
<ul>
<li><a HREF="#Doxygen_translator_design">18.7.1 Doxygen translator
design</a></li>
<li><a HREF="#Doxygen_debugging_commands">18.7.2 Debugging the Doxygen
parser and translator</a></li>
<li><a HREF="#Doxygen_tests">18.7.3 Tests</a></li>
</ul>
</li>
<li><a HREF="#Doxygen_language_extension">18.8 Extending to other
languages</a></li>
</ul>
<b><a HREF="#Warnings">19 Warning Messages</a></b>
<ul>
<li><a HREF="#Warnings_nn2">19.1 Introduction</a></li>
<li><a HREF="#Warnings_suppression">19.2 Warning message suppression</a></li>
<li><a HREF="#Warnings_nn4">19.3 Enabling extra warnings</a></li>
<li><a HREF="#Warnings_nn5">19.4 Issuing a warning message</a></li>
<li><a HREF="#Warnings_symbolic_symbols">19.5 Symbolic symbols</a></li>
<li><a HREF="#Warnings_nn6">19.6 Commentary</a></li>
<li><a HREF="#Warnings_nn7">19.7 Warnings as errors</a></li>
<li><a HREF="#Warnings_nn8">19.8 Message output format</a></li>
<li><a HREF="#Warnings_nn9">19.9 Warning number reference</a>
<ul>
<li><a HREF="#Warnings_nn10">19.9.1 Deprecated features (100-199)</a></li>
<li><a HREF="#Warnings_nn11">19.9.2 Preprocessor (200-299)</a></li>
<li><a HREF="#Warnings_nn12">19.9.3 C/C++ Parser (300-399)</a></li>
<li><a HREF="#Warnings_nn13">19.9.4 Types and typemaps (400-499)</a></li>
<li><a HREF="#Warnings_nn14">19.9.5 Code generation (500-559)</a></li>
<li><a HREF="#Warnings_doxygen">19.9.6 Doxygen comments (560-599)</a></li>
<li><a HREF="#Warnings_nn15">19.9.7 Language module specific (700-899)</a>
</li>
<li><a HREF="#Warnings_nn16">19.9.8 User defined (900-999)</a></li>
</ul>
</li>
<li><a HREF="#Warnings_nn17">19.10 History</a></li>
</ul>
<b><a HREF="#Modules">20 Working with Modules</a></b>
<ul>
<li><a HREF="#Modules_introduction">20.1 Modules Introduction</a></li>
<li><a HREF="#Modules_nn1">20.2 Basics</a></li>
<li><a HREF="#Modules_nn2">20.3 The SWIG runtime code</a></li>
<li><a HREF="#Modules_external_run_time">20.4 External access to the
runtime</a></li>
<li><a HREF="#Modules_nn4">20.5 A word of caution about static libraries</a>
</li>
<li><a HREF="#Modules_nn5">20.6 References</a></li>
<li><a HREF="#Modules_nn6">20.7 Reducing the wrapper file size</a></li>
</ul>
<b><a HREF="#CCache">21 Using SWIG with ccache - ccache-swig(1) manpage</a>
</b>
<ul>
<li><a HREF="#CCache_nn2">21.1 NAME</a></li>
<li><a HREF="#CCache_nn3">21.2 SYNOPSIS</a></li>
<li><a HREF="#CCache_nn4">21.3 DESCRIPTION</a></li>
<li><a HREF="#CCache_nn5">21.4 OPTIONS SUMMARY</a></li>
<li><a HREF="#CCache_nn6">21.5 OPTIONS</a></li>
<li><a HREF="#CCache_nn7">21.6 INSTALLATION</a></li>
<li><a HREF="#CCache_nn8">21.7 EXTRA OPTIONS</a></li>
<li><a HREF="#CCache_nn9">21.8 ENVIRONMENT VARIABLES</a></li>
<li><a HREF="#CCache_nn10">21.9 CACHE SIZE MANAGEMENT</a></li>
<li><a HREF="#CCache_nn11">21.10 CACHE COMPRESSION</a></li>
<li><a HREF="#CCache_nn12">21.11 HOW IT WORKS</a></li>
<li><a HREF="#CCache_nn13">21.12 USING CCACHE WITH DISTCC</a></li>
<li><a HREF="#CCache_nn14">21.13 SHARING A CACHE</a></li>
<li><a HREF="#CCache_nn15">21.14 HISTORY</a></li>
<li><a HREF="#CCache_nn16">21.15 DIFFERENCES FROM COMPILERCACHE</a></li>
<li><a HREF="#CCache_nn17">21.16 CREDITS</a></li>
<li><a HREF="#CCache_nn18">21.17 AUTHOR</a></li>
</ul>
<b><a HREF="#Android">22 SWIG and Android</a></b>
<ul>
<li><a HREF="#Android_overview">22.1 Overview</a></li>
<li><a HREF="#Android_examples">22.2 Android examples</a>
<ul>
<li><a HREF="#Android_examples_intro">22.2.1 Examples introduction</a></li>
<li><a HREF="#Android_example_simple">22.2.2 Simple C example</a></li>
<li><a HREF="#Android_example_class">22.2.3 C++ class example</a></li>
<li><a HREF="#Android_examples_other">22.2.4 Other examples</a></li>
</ul>
</li>
<li><a HREF="#Android_stl">22.3 C++ STL</a></li>
</ul>
<b><a HREF="#CSharp">23 SWIG and C#</a></b>
<ul>
<li><a HREF="#CSharp_introduction">23.1 Introduction</a>
<ul>
<li><a HREF="#CSharp_introduction_swig2_compatibility">23.1.1 SWIG 2
Compatibility</a></li>
<li><a HREF="#CSharp_commandline">23.1.2 Additional command line options</a>
</li>
</ul>
</li>
<li><a HREF="#CSharp_differences_java">23.2 Differences to the Java
module</a></li>
<li><a HREF="#CSharp_type_mapping">23.3 Type mapping</a>
<ul>
<li><a HREF="#CSharp_primitive_types">23.3.1 Primitive types</a></li>
<li><a HREF="#CSharp_other_type_mappings">23.3.2 Other types</a></li>
<li><a HREF="#CSharp_void_pointers">23.3.3 Void pointers</a></li>
</ul>
</li>
<li><a HREF="#CSharp_arrays">23.4 C# Arrays</a>
<ul>
<li><a HREF="#CSharp_arrays_swig_library">23.4.1 The SWIG C arrays
library</a></li>
<li><a HREF="#CSharp_arrays_pinvoke_default_array_marshalling">23.4.2
Managed arrays using P/Invoke default array marshalling</a></li>
<li><a HREF="#CSharp_arrays_pinning">23.4.3 Managed arrays using pinning</a>
</li>
</ul>
</li>
<li><a HREF="#CSharp_exceptions">23.5 C# Exceptions</a>
<ul>
<li><a HREF="#CSharp_exception_example_check_typemap">23.5.1 C#
exception example using &quot;check&quot; typemap</a></li>
<li><a HREF="#CSharp_exception_example_percent_exception">23.5.2 C#
exception example using %exception</a></li>
<li><a HREF="#CSharp_exception_example_exception_specifications">23.5.3
C# exception example using exception specifications</a></li>
<li><a HREF="#CSharp_custom_application_exception">23.5.4 Custom C#
ApplicationException example</a></li>
</ul>
</li>
<li><a HREF="#CSharp_directors">23.6 C# Directors</a>
<ul>
<li><a HREF="#CSharp_directors_example">23.6.1 Directors example</a></li>
<li><a HREF="#CSharp_directors_implementation">23.6.2 Directors
implementation</a></li>
<li><a HREF="#CSharp_director_caveats">23.6.3 Director caveats</a></li>
</ul>
</li>
<li><a HREF="#CSharp_multiple_modules">23.7 Multiple modules</a></li>
<li><a HREF="#CSharp_named_arguments">23.8 C# named and optional
arguments</a></li>
<li><a HREF="#CSharp_typemap_examples">23.9 C# Typemap examples</a>
<ul>
<li><a HREF="#CSharp_memory_management_member_variables">23.9.1 Memory
management when returning references to member variables</a></li>
<li><a HREF="#CSharp_memory_management_objects">23.9.2 Memory management
for objects passed to the C++ layer</a></li>
<li><a HREF="#CSharp_date_marshalling">23.9.3 Date marshalling using the
csin typemap and associated attributes</a></li>
<li><a HREF="#CSharp_date_properties">23.9.4 A date example
demonstrating marshalling of C# properties</a></li>
<li><a HREF="#CSharp_date_pre_post_directors">23.9.5 Date example
demonstrating the 'pre' and 'post' typemap attributes for directors</a></li>
<li><a HREF="#CSharp_partial_classes">23.9.6 Turning proxy classes into
partial classes</a></li>
<li><a HREF="#CSharp_sealed_proxy_class">23.9.7 Turning proxy classes
into sealed classes</a></li>
<li><a HREF="#CSharp_extending_proxy_class">23.9.8 Extending proxy
classes with additional C# code</a></li>
<li><a HREF="#CSharp_enum_underlying_type">23.9.9 Underlying type for
enums</a></li>
</ul>
</li>
</ul>
<b><a HREF="#D">24 SWIG and D</a></b>
<ul>
<li><a HREF="#D_introduction">24.1 Introduction</a></li>
<li><a HREF="#D_command_line_invocation">24.2 Command line invocation</a>
</li>
<li><a HREF="#D_typemaps">24.3 Typemaps</a>
<ul>
<li><a HREF="#D_typemap_name_comparison">24.3.1 C# &lt;-&gt; D name comparison</a>
</li>
<li><a HREF="#D_ctype_imtype_dtype">24.3.2 ctype, imtype, dtype</a></li>
<li><a HREF="#D_in_out_directorin_direcetorout">24.3.3 in, out,
directorin, directorout</a></li>
<li><a HREF="#D_din_dout_ddirectorin_ddirectorout">24.3.4 din, dout,
ddirectorin, ddirectorout</a></li>
<li><a HREF="#D_typecheck_typemaps">24.3.5 typecheck typemaps</a></li>
<li><a HREF="#D_code_injection_typemaps">24.3.6 Code injection typemaps</a>
</li>
<li><a HREF="#D_special_variables">24.3.7 Special variable macros</a></li>
</ul>
</li>
<li><a HREF="#D_other_code_control">24.4 Other D code control features</a>
<ul>
<li><a HREF="#D_module">24.4.1 D begin</a></li>
<li><a HREF="#D_features">24.4.2 D and %feature</a></li>
</ul>
</li>
<li><a HREF="#D_pragmas">24.5 Pragmas</a></li>
<li><a HREF="#D_exceptions">24.6 D Exceptions</a></li>
<li><a HREF="#D_directors">24.7 D Directors</a></li>
<li><a HREF="#D_other_features">24.8 Other features</a>
<ul>
<li><a HREF="#D_nspace">24.8.1 Extended namespace support (nspace)</a></li>
<li><a HREF="#D_native_pointer_support">24.8.2 Native pointer support</a>
</li>
<li><a HREF="#D_operator_overloading">24.8.3 Operator overloading</a></li>
<li><a HREF="#D_test_suite">24.8.4 Running the test-suite</a></li>
</ul>
</li>
<li><a HREF="#D_typemap_examples">24.9 D Typemap examples</a></li>
<li><a HREF="#D_planned_features">24.10 Work in progress and planned
features</a></li>
</ul>
<b><a HREF="#Go">25 SWIG and Go</a></b>
<ul>
<li><a HREF="#Go_overview">25.1 Overview</a></li>
<li><a HREF="#Go_examples">25.2 Examples</a></li>
<li><a HREF="#Go_running_swig">25.3 Running SWIG with Go</a>
<ul>
<li><a HREF="#Go_commandline">25.3.1 Go-specific Commandline Options</a></li>
<li><a HREF="#Go_outputs">25.3.2 Generated Wrapper Files</a></li>
</ul>
</li>
<li><a HREF="#Go_basic_tour">25.4 A tour of basic C/C++ wrapping</a>
<ul>
<li><a HREF="#Go_package">25.4.1 Go Package Name</a></li>
<li><a HREF="#Go_names">25.4.2 Go Names</a></li>
<li><a HREF="#Go_constants">25.4.3 Go Constants</a></li>
<li><a HREF="#Go_enumerations">25.4.4 Go Enumerations</a></li>
<li><a HREF="#Go_classes">25.4.5 Go Classes</a>
<ul>
<li><a HREF="#Go_class_memory">25.4.5.1 Go Class Memory Management</a></li>
<li><a HREF="#Go_class_inheritance">25.4.5.2 Go Class Inheritance</a></li>
</ul>
</li>
<li><a HREF="#Go_templates">25.4.6 Go Templates</a></li>
<li><a HREF="#Go_threads">25.4.7 Go and C/C++ Threads</a></li>
<li><a HREF="#Go_exceptions">25.4.8 Go and C++ Exceptions</a></li>
<li><a HREF="#Go_director_classes">25.4.9 Go Director Classes</a>
<ul>
<li><a HREF="#Go_director_example_cpp_code">25.4.9.1 Example C++ code</a>
</li>
<li><a HREF="#Go_director_enable">25.4.9.2 Enable director feature</a></li>
<li><a HREF="#Go_director_ctor_dtor">25.4.9.3 Constructor and destructor</a>
</li>
<li><a HREF="#Go_director_overriding">25.4.9.4 Override virtual methods</a>
</li>
<li><a HREF="#Go_director_base_methods">25.4.9.5 Call base methods</a></li>
<li><a HREF="#Go_director_subclass">25.4.9.6 Subclass via embedding</a></li>
<li><a HREF="#Go_director_finalizer">25.4.9.7 Memory management with
runtime.SetFinalizer</a></li>
<li><a HREF="#Go_director_foobargo_class">25.4.9.8 Complete FooBarGo
example class</a></li>
</ul>
</li>
<li><a HREF="#Go_primitive_type_mappings">25.4.10 Default Go primitive
type mappings</a></li>
<li><a HREF="#Go_output_arguments">25.4.11 Output arguments</a></li>
<li><a HREF="#Go_adding_additional_code">25.4.12 Adding additional go
code</a></li>
<li><a HREF="#Go_typemaps">25.4.13 Go typemaps</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Guile">26 SWIG and Guile</a></b>
<ul>
<li><a HREF="#Guile_nn1">26.1 Supported Guile Versions</a></li>
<li><a HREF="#Guile_nn2">26.2 Meaning of &quot;Module&quot;</a></li>
<li><a HREF="#Guile_nn3">26.3 Old GH Guile API</a></li>
<li><a HREF="#Guile_nn4">26.4 Linkage</a>
<ul>
<li><a HREF="#Guile_nn5">26.4.1 Simple Linkage</a></li>
<li><a HREF="#Guile_nn6">26.4.2 Passive Linkage</a></li>
<li><a HREF="#Guile_nn7">26.4.3 Native Guile Module Linkage</a></li>
<li><a HREF="#Guile_nn8">26.4.4 Old Auto-Loading Guile Module Linkage</a>
</li>
<li><a HREF="#Guile_nn9">26.4.5 Hobbit4D Linkage</a></li>
</ul>
</li>
<li><a HREF="#Guile_nn10">26.5 Underscore Folding</a></li>
<li><a HREF="#Guile_nn11">26.6 Typemaps</a></li>
<li><a HREF="#Guile_nn12">26.7 Representation of pointers as smobs</a>
<ul>
<li><a HREF="#Guile_nn14">26.7.1 Smobs</a></li>
<li><a HREF="#Guile_nn15">26.7.2 Garbage Collection</a></li>
</ul>
</li>
<li><a HREF="#Guile_nn16">26.8 Native Guile pointers</a></li>
<li><a HREF="#Guile_nn17">26.9 Exception Handling</a></li>
<li><a HREF="#Guile_nn18">26.10 Procedure documentation</a></li>
<li><a HREF="#Guile_nn19">26.11 Procedures with setters</a></li>
<li><a HREF="#Guile_nn20">26.12 GOOPS Proxy Classes</a>
<ul>
<li><a HREF="#Guile_nn21">26.12.1 Naming Issues</a></li>
<li><a HREF="#Guile_nn22">26.12.2 Linking</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Java">27 SWIG and Java</a></b>
<ul>
<li><a HREF="#Java_overview">27.1 Overview</a></li>
<li><a HREF="#Java_preliminaries">27.2 Preliminaries</a>
<ul>
<li><a HREF="#Java_running_swig">27.2.1 Running SWIG</a></li>
<li><a HREF="#Java_commandline">27.2.2 Additional Commandline Options</a>
</li>
<li><a HREF="#Java_getting_right_headers">27.2.3 Getting the right
header files</a></li>
<li><a HREF="#Java_compiling_dynamic">27.2.4 Compiling a dynamic module</a>
</li>
<li><a HREF="#Java_using_module">27.2.5 Using your module</a></li>
<li><a HREF="#Java_dynamic_linking_problems">27.2.6 Dynamic linking
problems</a></li>
<li><a HREF="#Java_compilation_problems_cpp">27.2.7 Compilation problems
and compiling with C++</a></li>
<li><a HREF="#Java_building_windows">27.2.8 Building on Windows</a>
<ul>
<li><a HREF="#Java_visual_studio">27.2.8.1 Running SWIG from Visual
Studio</a></li>
<li><a HREF="#Java_nmake">27.2.8.2 Using NMAKE</a></li>
</ul>
</li>
</ul>
</li>
<li><a HREF="#Java_basic_tour">27.3 A tour of basic C/C++ wrapping</a>
<ul>
<li><a HREF="#Java_module_packages_classes">27.3.1 Modules, packages and
generated Java classes</a></li>
<li><a HREF="#Java_functions">27.3.2 Functions</a></li>
<li><a HREF="#Java_global_variables">27.3.3 Global variables</a></li>
<li><a HREF="#Java_constants">27.3.4 Constants</a></li>
<li><a HREF="#Java_enumerations">27.3.5 Enumerations</a>
<ul>
<li><a HREF="#Java_anonymous_enums">27.3.5.1 Anonymous enums</a></li>
<li><a HREF="#Java_typesafe_enums">27.3.5.2 Typesafe enums</a></li>
<li><a HREF="#Java_proper_enums">27.3.5.3 Proper Java enums</a></li>
<li><a HREF="#Java_typeunsafe_enums">27.3.5.4 Type unsafe enums</a></li>
<li><a HREF="#Java_simple_enums">27.3.5.5 Simple enums</a></li>
</ul>
</li>
<li><a HREF="#Java_pointers">27.3.6 Pointers</a></li>
<li><a HREF="#Java_structures">27.3.7 Structures</a></li>
<li><a HREF="#Java_classes">27.3.8 C++ classes</a></li>
<li><a HREF="#Java_inheritance">27.3.9 C++ inheritance</a></li>
<li><a HREF="#Java_pointers_refs_arrays">27.3.10 Pointers, references,
arrays and pass by value</a>
<ul>
<li><a HREF="#Java_null_pointers">27.3.10.1 Null pointers</a></li>
</ul>
</li>
<li><a HREF="#Java_overloaded_functions">27.3.11 C++ overloaded
functions</a></li>
<li><a HREF="#Java_default_arguments">27.3.12 C++ default arguments</a></li>
<li><a HREF="#Java_namespaces">27.3.13 C++ namespaces</a></li>
<li><a HREF="#Java_templates">27.3.14 C++ templates</a></li>
<li><a HREF="#Java_smart_pointers">27.3.15 C++ Smart Pointers</a>
<ul>
<li><a HREF="#Java_smart_pointers_shared_ptr">27.3.15.1 The shared_ptr
Smart Pointer</a></li>
<li><a HREF="#Java_smart_pointers_generic">27.3.15.2 Generic Smart
Pointers</a></li>
</ul>
</li>
</ul>
</li>
<li><a HREF="#Java_further_details">27.4 Further details on the
generated Java classes</a>
<ul>
<li><a HREF="#Java_imclass">27.4.1 The intermediary JNI class</a>
<ul>
<li><a HREF="#Java_imclass_pragmas">27.4.1.1 The intermediary JNI class
pragmas</a></li>
</ul>
</li>
<li><a HREF="#Java_module_class">27.4.2 The Java module class</a>
<ul>
<li><a HREF="#Java_module_class_pragmas">27.4.2.1 The Java module class
pragmas</a></li>
</ul>
</li>
<li><a HREF="#Java_constants_interface">27.4.3 The Java constants
interface</a>
<ul>
<li><a HREF="#Java_constants_interface_pragmas">27.4.3.1 The Java
constants interface pragmas</a></li>
</ul>
</li>
<li><a HREF="#Java_proxy_classes">27.4.4 Java proxy classes</a>
<ul>
<li><a HREF="#Java_memory_management">27.4.4.1 Memory management</a></li>
<li><a HREF="#Java_inheritance_mirroring">27.4.4.2 Inheritance</a></li>
<li><a HREF="#Java_proxy_classes_gc">27.4.4.3 Proxy classes and garbage
collection</a></li>
<li><a HREF="#Java_pgcpp">27.4.4.4 The premature garbage collection
prevention parameter for proxy class marshalling</a></li>
<li><a HREF="#Java_multithread_libraries">27.4.4.5 Single threaded
applications and thread safety</a></li>
</ul>
</li>
<li><a HREF="#Java_type_wrapper_classes">27.4.5 Type wrapper classes</a></li>
<li><a HREF="#Java_enum_classes">27.4.6 Enum classes</a>
<ul>
<li><a HREF="#Java_typesafe_enums_classes">27.4.6.1 Typesafe enum
classes</a></li>
<li><a HREF="#Java_proper_enums_classes">27.4.6.2 Proper Java enum
classes</a></li>
<li><a HREF="#Java_typeunsafe_enums_classes">27.4.6.3 Type unsafe enum
classes</a></li>
</ul>
</li>
<li><a HREF="#Java_interfaces">27.4.7 Interfaces</a></li>
</ul>
</li>
<li><a HREF="#Java_directors">27.5 Cross language polymorphism using
directors</a>
<ul>
<li><a HREF="#Java_enabling_directors">27.5.1 Enabling directors</a></li>
<li><a HREF="#Java_directors_classes">27.5.2 Director classes</a></li>
<li><a HREF="#Java_directors_overhead">27.5.3 Overhead and code bloat</a>
</li>
<li><a HREF="#Java_directors_example">27.5.4 Simple directors example</a>
</li>
<li><a HREF="#Java_directors_threading">27.5.5 Director threading issues</a>
</li>
<li><a HREF="#Java_directors_performance">27.5.6 Director performance
tuning</a></li>
<li><a HREF="#Java_exceptions_from_directors">27.5.7 Java exceptions
from directors</a>
<ul>
<li><a HREF="#Java_customizing_director_exceptions">27.5.7.1 Customizing
director exceptions</a></li>
</ul>
</li>
<li><a HREF="#Java_protected_virtual_methods">27.5.8 Accessing virtual
protected methods</a></li>
<li><a HREF="#Java_allprotected">27.5.9 Accessing non-virtual protected
members</a></li>
</ul>
</li>
<li><a HREF="#Java_common_customization">27.6 Common customization
features</a>
<ul>
<li><a HREF="#Java_helper_functions">27.6.1 C/C++ helper functions</a></li>
<li><a HREF="#Java_class_extension">27.6.2 Class extension with %extend</a>
</li>
<li><a HREF="#Java_proxycode">27.6.3 Class extension with %proxycode</a></li>
<li><a HREF="#Java_exception_handling">27.6.4 Exception handling with
%exception and %javaexception</a></li>
<li><a HREF="#Java_method_access">27.6.5 Method access with
%javamethodmodifiers</a></li>
<li><a HREF="#Java_begin">27.6.6 Java begin</a></li>
</ul>
</li>
<li><a HREF="#Java_tips_techniques">27.7 Tips and techniques</a>
<ul>
<li><a HREF="#Java_input_output_parameters">27.7.1 Input and output
parameters using primitive pointers and references</a></li>
<li><a HREF="#Java_simple_pointers">27.7.2 Simple pointers</a></li>
<li><a HREF="#Java_c_arrays">27.7.3 Wrapping C arrays with Java arrays</a>
</li>
<li><a HREF="#Java_unbounded_c_arrays">27.7.4 Unbounded C Arrays</a></li>
<li><a HREF="#Java_string_length">27.7.5 Passing a string with length</a>
</li>
<li><a HREF="#Java_heap_allocations">27.7.6 Overriding new and delete to
allocate from Java heap</a></li>
</ul>
</li>
<li><a HREF="#Java_typemaps">27.8 Java typemaps</a>
<ul>
<li><a HREF="#Java_default_primitive_type_mappings">27.8.1 Default
primitive type mappings</a></li>
<li><a HREF="#Java_default_non_primitive_typemaps">27.8.2 Default
typemaps for non-primitive types</a></li>
<li><a HREF="#Java_jvm64">27.8.3 Sixty four bit JVMs</a></li>
<li><a HREF="#Java_what_is_typemap">27.8.4 What is a typemap?</a></li>
<li><a HREF="#Java_typemaps_c_to_java_types">27.8.5 Typemaps for mapping
C/C++ types to Java types</a></li>
<li><a HREF="#Java_typemap_attributes">27.8.6 Java typemap attributes</a>
</li>
<li><a HREF="#Java_special_variables">27.8.7 Java special variables</a></li>
<li><a HREF="#Java_typemaps_for_c_and_cpp">27.8.8 Typemaps for both C
and C++ compilation</a></li>
<li><a HREF="#Java_code_typemaps">27.8.9 Java code typemaps</a></li>
<li><a HREF="#Java_directors_typemaps">27.8.10 Director specific
typemaps</a></li>
</ul>
</li>
<li><a HREF="#Java_typemap_examples">27.9 Typemap Examples</a>
<ul>
<li><a HREF="#Java_simpler_enum_classes">27.9.1 Simpler Java enums for
enums without initializers</a></li>
<li><a HREF="#Java_exception_typemap">27.9.2 Handling C++ exception
specifications as Java exceptions</a></li>
<li><a HREF="#Java_nan_exception_typemap">27.9.3 NaN Exception -
exception handling for a particular type</a></li>
<li><a HREF="#Java_converting_java_string_arrays">27.9.4 Converting Java
String arrays to char **</a></li>
<li><a HREF="#Java_expanding_java_object">27.9.5 Expanding a Java object
to multiple arguments</a></li>
<li><a HREF="#Java_using_typemaps_return_arguments">27.9.6 Using
typemaps to return arguments</a></li>
<li><a HREF="#Java_adding_downcasts">27.9.7 Adding Java downcasts to
polymorphic return types</a></li>
<li><a HREF="#Java_adding_equals_method">27.9.8 Adding an equals method
to the Java classes</a></li>
<li><a HREF="#Java_void_pointers">27.9.9 Void pointers and a common Java
base class</a></li>
<li><a HREF="#Java_struct_pointer_pointer">27.9.10 Struct pointer to
pointer</a></li>
<li><a HREF="#Java_memory_management_member_variables">27.9.11 Memory
management when returning references to member variables</a></li>
<li><a HREF="#Java_memory_management_objects">27.9.12 Memory management
for objects passed to the C++ layer</a></li>
<li><a HREF="#Java_date_marshalling">27.9.13 Date marshalling using the
javain typemap and associated attributes</a></li>
</ul>
</li>
<li><a HREF="#Java_directors_faq">27.10 Living with Java Directors</a></li>
<li><a HREF="#Java_odds_ends">27.11 Odds and ends</a>
<ul>
<li><a HREF="#Java_javadoc_comments">27.11.1 JavaDoc comments</a></li>
<li><a HREF="#Java_functional_interface">27.11.2 Functional interface
without proxy classes</a></li>
<li><a HREF="#Java_using_own_jni_functions">27.11.3 Using your own JNI
functions</a></li>
<li><a HREF="#Java_performance">27.11.4 Performance concerns and hints</a>
</li>
<li><a HREF="#Java_debugging">27.11.5 Debugging</a></li>
</ul>
</li>
<li><a HREF="#Java_examples">27.12 Java Examples</a></li>
</ul>
<b><a HREF="#Javascript">28 SWIG and Javascript</a></b>
<ul>
<li><a HREF="#Javascript_overview">28.1 Overview</a></li>
<li><a HREF="#Javascript_preliminaries">28.2 Preliminaries</a>
<ul>
<li><a HREF="#Javascript_running_swig">28.2.1 Running SWIG</a></li>
<li><a HREF="#Javascript_running_tests_examples">28.2.2 Running Tests
and Examples</a></li>
<li><a HREF="#Javascript_known_issues">28.2.3 Known Issues</a></li>
</ul>
</li>
<li><a HREF="#Javascript_integration">28.3 Integration</a>
<ul>
<li><a HREF="#Javascript_node_extensions">28.3.1 Creating node.js
Extensions</a>
<ul>
<li><a HREF="#Javascript_using_yeoman">28.3.1.1 Using yeoman to generate
a Node-API skeleton</a></li>
<li><a HREF="#Javascript_troubleshooting">28.3.1.2 Troubleshooting</a></li>
</ul>
</li>
<li><a HREF="#Javascript_embedded_webkit">28.3.2 Embedded Webkit</a>
<ul>
<li><a HREF="#Javascript_osx">28.3.2.1 Mac OS X</a></li>
<li><a HREF="#Javascript_gtk">28.3.2.2 GTK</a></li>
</ul>
</li>
<li><a HREF="#Javascript_applications_webkit">28.3.3 Creating
Applications with node-webkit</a></li>
</ul>
</li>
<li><a HREF="#Javascript_examples">28.4 Examples</a>
<ul>
<li><a HREF="#Javascript_simple_example">28.4.1 Simple</a></li>
<li><a HREF="#Javascript_class_example">28.4.2 Class</a></li>
</ul>
</li>
<li><a HREF="#Javascript_implementation">28.5 Implementation</a>
<ul>
<li><a HREF="#Javascript_source_code">28.5.1 Source Code</a></li>
<li><a HREF="#Javascript_code_templates">28.5.2 Code Templates</a></li>
<li><a HREF="#Javascript_emitter">28.5.3 Emitter</a></li>
<li><a HREF="#Javascript_emitter_states">28.5.4 Emitter states</a></li>
<li><a HREF="#Javascript_jsc_exceptions">28.5.5 Handling Exceptions in
JavascriptCore</a></li>
<li><a HREF="#Javascript_napi_exceptions">28.5.6 Handling Exceptions in
Node-API</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Lua">29 SWIG and Lua</a></b>
<ul>
<li><a HREF="#Lua_nn2">29.1 Preliminaries</a></li>
<li><a HREF="#Lua_nn3">29.2 Running SWIG</a>
<ul>
<li><a HREF="#Lua_commandline">29.2.1 Additional command line options</a>
</li>
<li><a HREF="#Lua_nn4">29.2.2 Compiling and Linking and Interpreter</a></li>
<li><a HREF="#Lua_nn5">29.2.3 Compiling a dynamic module</a></li>
<li><a HREF="#Lua_nn6">29.2.4 Using your module</a></li>
</ul>
</li>
<li><a HREF="#Lua_nn7">29.3 A tour of basic C/C++ wrapping</a>
<ul>
<li><a HREF="#Lua_nn8">29.3.1 Modules</a></li>
<li><a HREF="#Lua_nn9">29.3.2 Functions</a></li>
<li><a HREF="#Lua_nn10">29.3.3 Global variables</a></li>
<li><a HREF="#Lua_nn11">29.3.4 Constants and enums</a>
<ul>
<li><a HREF="#Lua_nn13">29.3.4.1 Constants/enums and classes/structures</a>
</li>
</ul>
</li>
<li><a HREF="#Lua_nn12">29.3.5 Pointers</a></li>
<li><a HREF="#Lua_structures">29.3.6 Structures</a></li>
<li><a HREF="#Lua_nn14">29.3.7 C++ classes</a></li>
<li><a HREF="#Lua_nn15">29.3.8 C++ inheritance</a></li>
<li><a HREF="#Lua_nn16">29.3.9 Pointers, references, values, and arrays</a>
</li>
<li><a HREF="#Lua_nn17">29.3.10 C++ overloaded functions</a></li>
<li><a HREF="#Lua_nn18">29.3.11 C++ operators</a></li>
<li><a HREF="#Lua_nn19">29.3.12 Class extension with %extend</a></li>
<li><a HREF="#Lua_nn20">29.3.13 Using %newobject to release memory</a></li>
<li><a HREF="#Lua_nn21">29.3.14 C++ templates</a></li>
<li><a HREF="#Lua_nn22">29.3.15 C++ Smart Pointers</a></li>
<li><a HREF="#Lua_nn23">29.3.16 C++ Exceptions</a></li>
<li><a HREF="#Lua_namespaces">29.3.17 Namespaces</a>
<ul>
<li><a HREF="#Lua_nn27">29.3.17.1 Compatibility Note</a></li>
<li><a HREF="#Lua_nn29">29.3.17.2 Names</a></li>
<li><a HREF="#Lua_nn30">29.3.17.3 Inheritance</a></li>
</ul>
</li>
</ul>
</li>
<li><a HREF="#Lua_nn24">29.4 Typemaps</a>
<ul>
<li><a HREF="#Lua_nn25">29.4.1 What is a typemap?</a></li>
<li><a HREF="#Lua_nn26">29.4.2 Using typemaps</a></li>
<li><a HREF="#Lua_typemap_arrays">29.4.3 Typemaps and arrays</a></li>
<li><a HREF="#Lua_typemaps_ptr_ptr_functions">29.4.4 Typemaps and
pointer-pointer functions</a></li>
</ul>
</li>
<li><a HREF="#Lua_writing_typemaps">29.5 Writing typemaps</a>
<ul>
<li><a HREF="#Lua_typemaps_write">29.5.1 Typemaps you can write</a></li>
<li><a HREF="#Lua_nn31">29.5.2 SWIG's Lua-C API</a></li>
</ul>
</li>
<li><a HREF="#Lua_nn32">29.6 Customization of your Bindings</a>
<ul>
<li><a HREF="#Lua_nn33">29.6.1 Writing your own custom wrappers</a></li>
<li><a HREF="#Lua_nn34">29.6.2 Adding additional Lua code</a></li>
</ul>
</li>
<li><a HREF="#Lua_nn35">29.7 Details on the Lua binding</a>
<ul>
<li><a HREF="#Lua_nn36">29.7.1 Binding global data into the module.</a></li>
<li><a HREF="#Lua_nn37">29.7.2 Userdata and Metatables</a></li>
<li><a HREF="#Lua_nn38">29.7.3 Memory management</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Octave">30 SWIG and Octave</a></b>
<ul>
<li><a HREF="#Octave_nn2">30.1 Preliminaries</a></li>
<li><a HREF="#Octave_nn3">30.2 Running SWIG</a>
<ul>
<li><a HREF="#Octave_nn4">30.2.1 Command-line options</a></li>
<li><a HREF="#Octave_nn5">30.2.2 Compiling a dynamic module</a></li>
<li><a HREF="#Octave_nn6">30.2.3 Using your module</a></li>
</ul>
</li>
<li><a HREF="#Octave_nn7">30.3 A tour of basic C/C++ wrapping</a>
<ul>
<li><a HREF="#Octave_nn8">30.3.1 Modules</a></li>
<li><a HREF="#Octave_nn9">30.3.2 Functions</a></li>
<li><a HREF="#Octave_nn10">30.3.3 Global variables</a></li>
<li><a HREF="#Octave_nn11">30.3.4 Constants and enums</a></li>
<li><a HREF="#Octave_nn12">30.3.5 Pointers</a></li>
<li><a HREF="#Octave_nn13">30.3.6 Structures and C++ classes</a></li>
<li><a HREF="#Octave_nn15">30.3.7 C++ inheritance</a></li>
<li><a HREF="#Octave_nn17">30.3.8 C++ overloaded functions</a></li>
<li><a HREF="#Octave_nn18">30.3.9 C++ operators</a></li>
<li><a HREF="#Octave_nn19">30.3.10 Class extension with %extend</a></li>
<li><a HREF="#Octave_nn20">30.3.11 C++ templates</a></li>
<li><a HREF="#Octave_nn21">30.3.12 C++ Smart Pointers</a>
<ul>
<li><a HREF="#Octave_smart_pointers_shared_ptr">30.3.12.1 The shared_ptr
Smart Pointer</a></li>
<li><a HREF="#Octave_smart_pointers_generic">30.3.12.2 Generic Smart
Pointers</a></li>
</ul>
</li>
<li><a HREF="#Octave_nn22">30.3.13 Directors (calling Octave from C++
code)</a></li>
<li><a HREF="#Octave_nn23">30.3.14 Threads</a></li>
<li><a HREF="#Octave_nn24">30.3.15 Memory management</a></li>
<li><a HREF="#Octave_nn25">30.3.16 STL support</a></li>
<li><a HREF="#Octave_nn26">30.3.17 Matrix typemaps</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Perl5">31 SWIG and Perl5</a></b>
<ul>
<li><a HREF="#Perl5_nn2">31.1 Overview</a></li>
<li><a HREF="#Perl5_nn3">31.2 Preliminaries</a>
<ul>
<li><a HREF="#Perl5_nn4">31.2.1 Getting the right header files</a></li>
<li><a HREF="#Perl5_nn5">31.2.2 Compiling a dynamic module</a></li>
<li><a HREF="#Perl5_nn6">31.2.3 Building a dynamic module with MakeMaker</a>
</li>
<li><a HREF="#Perl5_nn7">31.2.4 Building a static version of Perl</a></li>
<li><a HREF="#Perl5_nn8">31.2.5 Using the module</a></li>
<li><a HREF="#Perl5_nn9">31.2.6 Compilation problems and compiling with
C++</a></li>
<li><a HREF="#Perl5_nn10">31.2.7 Compiling for 64-bit platforms</a></li>
</ul>
</li>
<li><a HREF="#Perl5_nn11">31.3 Building Perl Extensions under Windows</a>
<ul>
<li><a HREF="#Perl5_nn12">31.3.1 Running SWIG from Developer Studio</a></li>
<li><a HREF="#Perl5_nn13">31.3.2 Using other compilers</a></li>
</ul>
</li>
<li><a HREF="#Perl5_nn14">31.4 The low-level interface</a>
<ul>
<li><a HREF="#Perl5_nn15">31.4.1 Functions</a></li>
<li><a HREF="#Perl5_nn16">31.4.2 Global variables</a></li>
<li><a HREF="#Perl5_nn17">31.4.3 Constants</a></li>
<li><a HREF="#Perl5_nn18">31.4.4 Pointers</a></li>
<li><a HREF="#Perl5_nn19">31.4.5 Structures</a></li>
<li><a HREF="#Perl5_nn20">31.4.6 C++ classes</a></li>
<li><a HREF="#Perl5_nn21">31.4.7 C++ classes and type-checking</a></li>
<li><a HREF="#Perl5_nn22">31.4.8 C++ overloaded functions</a></li>
<li><a HREF="#Perl5_nn23">31.4.9 Operators</a></li>
<li><a HREF="#Perl5_nn24">31.4.10 Modules and packages</a></li>
</ul>
</li>
<li><a HREF="#Perl5_nn25">31.5 Input and output parameters</a></li>
<li><a HREF="#Perl5_nn26">31.6 Exception handling</a></li>
<li><a HREF="#Perl5_nn27">31.7 Remapping datatypes with typemaps</a>
<ul>
<li><a HREF="#Perl5_nn28">31.7.1 A simple typemap example</a></li>
<li><a HREF="#Perl5_nn29">31.7.2 Perl5 typemaps</a></li>
<li><a HREF="#Perl5_nn30">31.7.3 Typemap variables</a></li>
<li><a HREF="#Perl5_nn31">31.7.4 Useful functions</a></li>
</ul>
</li>
<li><a HREF="#Perl5_nn32">31.8 Typemap Examples</a>
<ul>
<li><a HREF="#Perl5_nn33">31.8.1 Converting a Perl5 array to a char **</a>
</li>
<li><a HREF="#Perl5_nn34">31.8.2 Return values</a></li>
<li><a HREF="#Perl5_nn35">31.8.3 Returning values from arguments</a></li>
<li><a HREF="#Perl5_nn36">31.8.4 Accessing array structure members</a></li>
<li><a HREF="#Perl5_nn37">31.8.5 Turning Perl references into C pointers</a>
</li>
<li><a HREF="#Perl5_nn38">31.8.6 Pointer handling</a></li>
</ul>
</li>
<li><a HREF="#Perl5_nn39">31.9 Proxy classes</a>
<ul>
<li><a HREF="#Perl5_nn40">31.9.1 Preliminaries</a></li>
<li><a HREF="#Perl5_nn41">31.9.2 Structure and class wrappers</a></li>
<li><a HREF="#Perl5_nn42">31.9.3 Object Ownership</a></li>
<li><a HREF="#Perl5_nn43">31.9.4 Nested Objects</a></li>
<li><a HREF="#Perl5_nn44">31.9.5 Proxy Functions</a></li>
<li><a HREF="#Perl5_nn45">31.9.6 Inheritance</a></li>
<li><a HREF="#Perl5_nn46">31.9.7 Modifying the proxy methods</a></li>
</ul>
</li>
<li><a HREF="#Perl5_nn47">31.10 Adding additional Perl code</a></li>
<li><a HREF="#Perl5_directors">31.11 Cross language polymorphism</a>
<ul>
<li><a HREF="#Perl5_nn48">31.11.1 Enabling directors</a></li>
<li><a HREF="#Perl5_nn49">31.11.2 Director classes</a></li>
<li><a HREF="#Perl5_nn50">31.11.3 Ownership and object destruction</a></li>
<li><a HREF="#Perl5_nn51">31.11.4 Exception unrolling</a></li>
<li><a HREF="#Perl5_nn52">31.11.5 Overhead and code bloat</a></li>
<li><a HREF="#Perl5_nn53">31.11.6 Typemaps</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Php">32 SWIG and PHP</a></b>
<ul>
<li><a HREF="#Php_nn1">32.1 Generating PHP Extensions</a>
<ul>
<li><a HREF="#Php_nn1_1">32.1.1 Building a loadable extension</a></li>
<li><a HREF="#Php_nn1_3">32.1.2 Using PHP Extensions</a></li>
</ul>
</li>
<li><a HREF="#Php_nn2">32.2 Basic PHP interface</a>
<ul>
<li><a HREF="#Php_nn2_1">32.2.1 Constants</a></li>
<li><a HREF="#Php_nn2_2">32.2.2 Global Variables</a></li>
<li><a HREF="#Php_nn2_3">32.2.3 Functions</a></li>
<li><a HREF="#Php_nn2_4">32.2.4 Overloading</a></li>
<li><a HREF="#Php_nn2_5">32.2.5 Pointers and References</a></li>
<li><a HREF="#Php_nn2_6">32.2.6 Structures and C++ classes</a>
<ul>
<li><a HREF="#Php_nn2_6_1">32.2.6.1 Using -noproxy</a></li>
<li><a HREF="#Php_nn2_6_2">32.2.6.2 Constructors and Destructors</a></li>
<li><a HREF="#Php_nn2_6_3">32.2.6.3 Static Member Variables</a></li>
<li><a HREF="#Php_nn2_6_4">32.2.6.4 Static Member Functions</a></li>
<li><a HREF="#Php_nn2_6_5">32.2.6.5 Specifying Implemented Interfaces</a>
</li>
<li><a HREF="#Php_nn2_6_6">32.2.6.6 Dynamic Properties</a></li>
</ul>
</li>
<li><a HREF="#Php_nn2_7">32.2.7 PHP Pragmas, Startup and Shutdown code</a>
</li>
</ul>
</li>
<li><a HREF="#Php_nn3">32.3 Cross language polymorphism</a>
<ul>
<li><a HREF="#Php_nn3_1">32.3.1 Enabling directors</a></li>
<li><a HREF="#Php_nn3_2">32.3.2 Director classes</a></li>
<li><a HREF="#Php_nn3_3">32.3.3 Ownership and object destruction</a></li>
<li><a HREF="#Php_nn3_4">32.3.4 Exception unrolling</a></li>
<li><a HREF="#Php_nn3_5">32.3.5 Overhead and code bloat</a></li>
<li><a HREF="#Php_nn3_6">32.3.6 Typemaps</a></li>
<li><a HREF="#Php_nn3_7">32.3.7 Miscellaneous</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Python">33 SWIG and Python</a></b>
<ul>
<li><a HREF="#Python_nn2">33.1 Overview</a></li>
<li><a HREF="#Python_nn3">33.2 Preliminaries</a>
<ul>
<li><a HREF="#Python_nn4">33.2.1 Running SWIG</a></li>
<li><a HREF="#Python_nn6">33.2.2 Using distutils</a></li>
<li><a HREF="#Python_nn7">33.2.3 Hand compiling a dynamic module</a></li>
<li><a HREF="#Python_nn8">33.2.4 Static linking</a></li>
<li><a HREF="#Python_nn9">33.2.5 Using your module</a></li>
<li><a HREF="#Python_nn10">33.2.6 Compilation of C++ extensions</a></li>
<li><a HREF="#Python_nn11">33.2.7 Compiling for 64-bit platforms</a></li>
<li><a HREF="#Python_nn12">33.2.8 Building Python extensions under
Windows</a></li>
<li><a HREF="#Python_commandline">33.2.9 Additional Python commandline
options</a></li>
</ul>
</li>
<li><a HREF="#Python_nn13">33.3 A tour of basic C/C++ wrapping</a>
<ul>
<li><a HREF="#Python_nn14">33.3.1 Modules</a></li>
<li><a HREF="#Python_nn15">33.3.2 Functions</a></li>
<li><a HREF="#Python_nn16">33.3.3 Global variables</a></li>
<li><a HREF="#Python_nn17">33.3.4 Constants and enums</a></li>
<li><a HREF="#Python_nn18">33.3.5 Pointers</a></li>
<li><a HREF="#Python_nn19">33.3.6 Structures</a></li>
<li><a HREF="#Python_nn20">33.3.7 C++ classes</a></li>
<li><a HREF="#Python_nn21">33.3.8 C++ inheritance</a></li>
<li><a HREF="#Python_nn22">33.3.9 Pointers, references, values, and
arrays</a></li>
<li><a HREF="#Python_nn23">33.3.10 C++ overloaded functions</a></li>
<li><a HREF="#Python_nn24">33.3.11 C++ operators</a></li>
<li><a HREF="#Python_nn25">33.3.12 C++ namespaces</a></li>
<li><a HREF="#Python_nn26">33.3.13 C++ templates</a></li>
<li><a HREF="#Python_nn27">33.3.14 C++ Smart Pointers</a>
<ul>
<li><a HREF="#Python_smart_pointers_shared_ptr">33.3.14.1 The shared_ptr
Smart Pointer</a></li>
<li><a HREF="#Python_smart_pointers_generic">33.3.14.2 Generic Smart
Pointers</a></li>
</ul>
</li>
<li><a HREF="#Python_nn27a">33.3.15 C++ reference counted objects</a></li>
</ul>
</li>
<li><a HREF="#Python_nn28">33.4 Further details on the Python class
interface</a>
<ul>
<li><a HREF="#Python_nn29">33.4.1 Proxy classes</a></li>
<li><a HREF="#Python_builtin_types">33.4.2 Built-in Types</a>
<ul>
<li><a HREF="#Python_builtin_limitations">33.4.2.1 Limitations</a></li>
<li><a HREF="#Python_builtin_overloads">33.4.2.2 Operator overloads and
slots -- use them!</a></li>
</ul>
</li>
<li><a HREF="#Python_nn30">33.4.3 Memory management</a></li>
</ul>
</li>
<li><a HREF="#Python_directors">33.5 Cross language polymorphism</a>
<ul>
<li><a HREF="#Python_nn33">33.5.1 Enabling directors</a></li>
<li><a HREF="#Python_nn34">33.5.2 Director classes</a></li>
<li><a HREF="#Python_nn35">33.5.3 Ownership and object destruction</a></li>
<li><a HREF="#Python_nn36">33.5.4 Exception unrolling</a></li>
<li><a HREF="#Python_nn37">33.5.5 Overhead and code bloat</a></li>
<li><a HREF="#Python_nn38">33.5.6 Typemaps</a></li>
<li><a HREF="#Python_director_thread_safety">33.5.7 Thread safety</a></li>
<li><a HREF="#Python_nn39">33.5.8 Miscellaneous</a></li>
</ul>
</li>
<li><a HREF="#Python_nn40">33.6 Common customization features</a>
<ul>
<li><a HREF="#Python_nn41">33.6.1 C/C++ helper functions</a></li>
<li><a HREF="#Python_nn42">33.6.2 Adding additional Python code</a></li>
<li><a HREF="#Python_nn43">33.6.3 Class extension with %extend</a></li>
<li><a HREF="#Python_nn44">33.6.4 Exception handling with %exception</a></li>
<li><a HREF="#Python_optimization">33.6.5 Optimization options</a>
<ul>
<li><a HREF="#Python_fastproxy">33.6.5.1 -fastproxy</a></li>
</ul>
</li>
<li><a HREF="#Python_stable_abi">33.6.6 Stable ABI</a></li>
</ul>
</li>
<li><a HREF="#Python_nn45">33.7 Tips and techniques</a>
<ul>
<li><a HREF="#Python_nn46">33.7.1 Input and output parameters</a></li>
<li><a HREF="#Python_nn47">33.7.2 Simple pointers</a></li>
<li><a HREF="#Python_nn48">33.7.3 Unbounded C Arrays</a></li>
<li><a HREF="#Python_nn49">33.7.4 String handling</a></li>
<li><a HREF="#Python_default_args">33.7.5 Default arguments</a></li>
</ul>
</li>
<li><a HREF="#Python_nn53">33.8 Typemaps</a>
<ul>
<li><a HREF="#Python_nn54">33.8.1 What is a typemap?</a></li>
<li><a HREF="#Python_nn55">33.8.2 Python typemaps</a></li>
<li><a HREF="#Python_nn56">33.8.3 Typemap variables</a></li>
<li><a HREF="#Python_nn57">33.8.4 Useful Python Functions</a></li>
</ul>
</li>
<li><a HREF="#Python_nn58">33.9 Typemap Examples</a>
<ul>
<li><a HREF="#Python_nn59">33.9.1 Converting a Python list to a char **</a>
</li>
<li><a HREF="#Python_nn60">33.9.2 Expanding a Python object into
multiple arguments</a></li>
<li><a HREF="#Python_nn61">33.9.3 Using typemaps to return arguments</a></li>
<li><a HREF="#Python_nn62">33.9.4 Mapping Python tuples into small
arrays</a></li>
<li><a HREF="#Python_nn63">33.9.5 Mapping sequences to C arrays</a></li>
<li><a HREF="#Python_nn64">33.9.6 Pointer handling</a></li>
<li><a HREF="#Python_memory_management_member_variables">33.9.7 Memory
management when returning references to member variables</a></li>
</ul>
</li>
<li><a HREF="#Python_nn65">33.10 Docstring Features</a>
<ul>
<li><a HREF="#Python_nn66">33.10.1 Module docstring</a></li>
<li><a HREF="#Python_nn67">33.10.2 %feature(&quot;autodoc&quot;)</a>
<ul>
<li><a HREF="#Python_nn68">33.10.2.1 %feature(&quot;autodoc&quot;, &quot;0&quot;)</a></li>
<li><a HREF="#Python_nn69">33.10.2.2 %feature(&quot;autodoc&quot;, &quot;1&quot;)</a></li>
<li><a HREF="#Python_autodoc2">33.10.2.3 %feature(&quot;autodoc&quot;, &quot;2&quot;)</a></li>
<li><a HREF="#Python_autodoc3">33.10.2.4 %feature(&quot;autodoc&quot;, &quot;3&quot;)</a></li>
<li><a HREF="#Python_nn70">33.10.2.5 %feature(&quot;autodoc&quot;, &quot;docstring&quot;)</a>
</li>
</ul>
</li>
<li><a HREF="#Python_nn71">33.10.3 %feature(&quot;docstring&quot;)</a></li>
<li><a HREF="#Python_doxygen_docstrings">33.10.4 Doxygen comments</a></li>
</ul>
</li>
<li><a HREF="#Python_nn72">33.11 Python Packages</a>
<ul>
<li><a HREF="#Python_modulepackage">33.11.1 Setting the Python package</a>
</li>
<li><a HREF="#Python_absrelimports">33.11.2 Absolute and relative
imports</a></li>
<li><a HREF="#Python_absimport">33.11.3 Enforcing absolute import
semantics</a></li>
<li><a HREF="#Python_importfrominit">33.11.4 Importing from __init__.py</a>
</li>
<li><a HREF="#Python_implicit_namespace_packages">33.11.5 Implicit
namespace packages</a></li>
<li><a HREF="#Python_package_search">33.11.6 Location of modules</a>
<ul>
<li><a HREF="#Python_package_search_both_package_modules">33.11.6.1 Both
modules in the same package</a></li>
<li><a HREF="#Python_package_search_both_global_modules">33.11.6.2 Both
modules are global</a></li>
<li><a HREF="#Python_package_search_wrapper_split">33.11.6.3 Split
modules custom configuration</a></li>
<li><a HREF="#Python_custom_module_import">33.11.6.4 More on customizing
the module import code</a></li>
<li><a HREF="#Python_package_search_static">33.11.6.5 Statically linked
C modules</a></li>
</ul>
</li>
</ul>
</li>
<li><a HREF="#Python_python3support">33.12 Python 3 Support</a>
<ul>
<li><a HREF="#Python_annotations">33.12.1 Python function annotations
and variable annotations</a>
<ul>
<li><a HREF="#Python_annotations_c">33.12.1.1 C/C++ annotation types</a></li>
</ul>
</li>
<li><a HREF="#Python_nn75">33.12.2 Buffer interface</a></li>
<li><a HREF="#Python_nn76">33.12.3 Abstract base classes</a></li>
<li><a HREF="#Python_nn77">33.12.4 Byte string output conversion</a></li>
<li><a HREF="#Python_2_unicode">33.12.5 Python 2 Unicode</a></li>
</ul>
</li>
<li><a HREF="#Python_multithreaded">33.13 Support for Multithreaded
Applications</a>
<ul>
<li><a HREF="#Python_thread_UI">33.13.1 UI for Enabling Multithreading
Support</a></li>
<li><a HREF="#Python_thread_performance">33.13.2 Multithread Performance</a>
</li>
</ul>
</li>
</ul>
<b><a HREF="#R">34 SWIG and R</a></b>
<ul>
<li><a HREF="#R_nn2">34.1 Bugs</a></li>
<li><a HREF="#R_nn3">34.2 Using R and SWIG</a></li>
<li><a HREF="#R_nn4">34.3 Precompiling large R files</a></li>
<li><a HREF="#R_nn5">34.4 General policy</a></li>
<li><a HREF="#R_language_conventions">34.5 Language conventions</a></li>
<li><a HREF="#R_nn6">34.6 C++ classes</a>
<ul>
<li><a HREF="#R_class_examples">34.6.1 Examples</a></li>
</ul>
</li>
<li><a HREF="#R_nn7">34.7 Enumerations</a></li>
</ul>
<b><a HREF="#Ruby">35 SWIG and Ruby</a></b>
<ul>
<li><a HREF="#Ruby_nn2">35.1 Preliminaries</a>
<ul>
<li><a HREF="#Ruby_nn3">35.1.1 Running SWIG</a></li>
<li><a HREF="#Ruby_nn4">35.1.2 Getting the right header files</a></li>
<li><a HREF="#Ruby_nn5">35.1.3 Compiling a dynamic module</a></li>
<li><a HREF="#Ruby_nn6">35.1.4 Using your module</a></li>
<li><a HREF="#Ruby_nn7">35.1.5 Static linking</a></li>
<li><a HREF="#Ruby_nn8">35.1.6 Compilation of C++ extensions</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn9">35.2 Building Ruby Extensions under Windows
95/NT</a>
<ul>
<li><a HREF="#Ruby_nn10">35.2.1 Running SWIG from Developer Studio</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn11">35.3 The Ruby-to-C/C++ Mapping</a>
<ul>
<li><a HREF="#Ruby_nn12">35.3.1 Modules</a></li>
<li><a HREF="#Ruby_nn13">35.3.2 Functions</a></li>
<li><a HREF="#Ruby_nn14">35.3.3 Variable Linking</a></li>
<li><a HREF="#Ruby_nn15">35.3.4 Constants</a></li>
<li><a HREF="#Ruby_nn16">35.3.5 Pointers</a></li>
<li><a HREF="#Ruby_nn17">35.3.6 Structures</a></li>
<li><a HREF="#Ruby_nn18">35.3.7 C++ classes</a></li>
<li><a HREF="#Ruby_nn19">35.3.8 C++ Inheritance</a></li>
<li><a HREF="#Ruby_nn20">35.3.9 C++ Overloaded Functions</a></li>
<li><a HREF="#Ruby_nn21">35.3.10 C++ Operators</a></li>
<li><a HREF="#Ruby_nn22">35.3.11 C++ namespaces</a></li>
<li><a HREF="#Ruby_nn23">35.3.12 C++ templates</a></li>
<li><a HREF="#Ruby_nn23_1">35.3.13 C++ Standard Template Library (STL)</a>
</li>
<li><a HREF="#Ruby_C_STL_Functors">35.3.14 C++ STL Functors</a></li>
<li><a HREF="#Ruby_C_Iterators">35.3.15 C++ STL Iterators</a></li>
<li><a HREF="#Ruby_nn24">35.3.16 C++ Smart Pointers</a>
<ul>
<li><a HREF="#Ruby_smart_pointers_shared_ptr">35.3.16.1 The shared_ptr
Smart Pointer</a></li>
<li><a HREF="#Ruby_smart_pointers_generic">35.3.16.2 Generic Smart
Pointers</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn25">35.3.17 Cross-Language Polymorphism</a>
<ul>
<li><a HREF="#Ruby_nn26">35.3.17.1 Exception Unrolling</a></li>
</ul>
</li>
</ul>
</li>
<li><a HREF="#Ruby_nn27">35.4 Naming</a>
<ul>
<li><a HREF="#Ruby_nn28">35.4.1 Defining Aliases</a></li>
<li><a HREF="#Ruby_nn29">35.4.2 Predicate Methods</a></li>
<li><a HREF="#Ruby_nn30">35.4.3 Bang Methods</a></li>
<li><a HREF="#Ruby_nn31">35.4.4 Getters and Setters</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn32">35.5 Input and output parameters</a></li>
<li><a HREF="#Ruby_nn33">35.6 Exception handling</a>
<ul>
<li><a HREF="#Ruby_nn34">35.6.1 Using the %exception directive</a></li>
<li><a HREF="#Ruby_nn34_2">35.6.2 Handling Ruby Blocks</a></li>
<li><a HREF="#Ruby_nn35">35.6.3 Raising exceptions</a></li>
<li><a HREF="#Ruby_nn36">35.6.4 Exception classes</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn37">35.7 Typemaps</a>
<ul>
<li><a HREF="#Ruby_nn38">35.7.1 What is a typemap?</a></li>
<li><a HREF="#Ruby_Typemap_scope">35.7.2 Typemap scope</a></li>
<li><a HREF="#Ruby_Copying_a_typemap">35.7.3 Copying a typemap</a></li>
<li><a HREF="#Ruby_Deleting_a_typemap">35.7.4 Deleting a typemap</a></li>
<li><a HREF="#Ruby_Placement_of_typemaps">35.7.5 Placement of typemaps</a>
</li>
<li><a HREF="#Ruby_nn39">35.7.6 Ruby typemaps</a>
<ul>
<li><a HREF="#Ruby_in_typemap">35.7.6.1 &quot;in&quot; typemap</a></li>
<li><a HREF="#Ruby_typecheck_typemap">35.7.6.2 &quot;typecheck&quot; typemap</a></li>
<li><a HREF="#Ruby_out_typemap">35.7.6.3 &quot;out&quot; typemap</a></li>
<li><a HREF="#Ruby_arginit_typemap">35.7.6.4 &quot;arginit&quot; typemap</a></li>
<li><a HREF="#Ruby_default_typemap">35.7.6.5 &quot;default&quot; typemap</a></li>
<li><a HREF="#Ruby_check_typemap">35.7.6.6 &quot;check&quot; typemap</a></li>
<li><a HREF="#Ruby_argout_typemap_">35.7.6.7 &quot;argout&quot; typemap</a></li>
<li><a HREF="#Ruby_freearg_typemap_">35.7.6.8 &quot;freearg&quot; typemap</a></li>
<li><a HREF="#Ruby_newfree_typemap">35.7.6.9 &quot;newfree&quot; typemap</a></li>
<li><a HREF="#Ruby_memberin_typemap">35.7.6.10 &quot;memberin&quot; typemap</a></li>
<li><a HREF="#Ruby_varin_typemap">35.7.6.11 &quot;varin&quot; typemap</a></li>
<li><a HREF="#Ruby_varout_typemap_">35.7.6.12 &quot;varout&quot; typemap</a></li>
<li><a HREF="#Ruby_throws_typemap">35.7.6.13 &quot;throws&quot; typemap</a></li>
<li><a HREF="#Ruby_directorin_typemap">35.7.6.14 directorin typemap</a></li>
<li><a HREF="#Ruby_directorout_typemap">35.7.6.15 directorout typemap</a>
</li>
<li><a HREF="#Ruby_directorargout_typemap">35.7.6.16 directorargout
typemap</a></li>
<li><a HREF="#Ruby_ret_typemap">35.7.6.17 ret typemap</a></li>
<li><a HREF="#Ruby_globalin_typemap">35.7.6.18 globalin typemap</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn40">35.7.7 Typemap variables</a></li>
<li><a HREF="#Ruby_nn41">35.7.8 Useful Functions</a>
<ul>
<li><a HREF="#Ruby_nn42">35.7.8.1 C Datatypes to Ruby Objects</a></li>
<li><a HREF="#Ruby_nn43">35.7.8.2 Ruby Objects to C Datatypes</a></li>
<li><a HREF="#Ruby_nn44">35.7.8.3 Macros for VALUE</a></li>
<li><a HREF="#Ruby_nn45">35.7.8.4 Exceptions</a></li>
<li><a HREF="#Ruby_nn46">35.7.8.5 Iterators</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn47">35.7.9 Typemap Examples</a></li>
<li><a HREF="#Ruby_nn48">35.7.10 Converting a Ruby array to a char **</a>
</li>
<li><a HREF="#Ruby_nn49">35.7.11 Collecting arguments in a hash</a></li>
<li><a HREF="#Ruby_nn50">35.7.12 Pointer handling</a>
<ul>
<li><a HREF="#Ruby_nn51">35.7.12.1 Ruby Datatype Wrapping</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn52">35.7.13 Example: STL Vector to Ruby Array</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn65">35.8 Docstring Features</a>
<ul>
<li><a HREF="#Ruby_nn66">35.8.1 Module docstring</a></li>
<li><a HREF="#Ruby_nn67">35.8.2 %feature(&quot;autodoc&quot;)</a>
<ul>
<li><a HREF="#Ruby_nn68">35.8.2.1 %feature(&quot;autodoc&quot;, &quot;0&quot;)</a></li>
<li><a HREF="#Ruby_autodoc1">35.8.2.2 %feature(&quot;autodoc&quot;, &quot;1&quot;)</a></li>
<li><a HREF="#Ruby_autodoc2">35.8.2.3 %feature(&quot;autodoc&quot;, &quot;2&quot;)</a></li>
<li><a HREF="#Ruby_feature_autodoc3">35.8.2.4 %feature(&quot;autodoc&quot;, &quot;3&quot;)</a>
</li>
<li><a HREF="#Ruby_nn70">35.8.2.5 %feature(&quot;autodoc&quot;, &quot;docstring&quot;)</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn71">35.8.3 %feature(&quot;docstring&quot;)</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn53">35.9 Advanced Topics</a>
<ul>
<li><a HREF="#Ruby_operator_overloading">35.9.1 Operator overloading</a></li>
<li><a HREF="#Ruby_nn55">35.9.2 Creating Multi-Module Packages</a></li>
<li><a HREF="#Ruby_nn56">35.9.3 Specifying Mixin Modules</a></li>
</ul>
</li>
<li><a HREF="#Ruby_nn57">35.10 Memory Management</a>
<ul>
<li><a HREF="#Ruby_nn58">35.10.1 Mark and Sweep Garbage Collector</a></li>
<li><a HREF="#Ruby_nn59">35.10.2 Object Ownership</a></li>
<li><a HREF="#Ruby_nn60">35.10.3 Object Tracking</a></li>
<li><a HREF="#Ruby_nn61">35.10.4 Mark Functions</a></li>
<li><a HREF="#Ruby_nn62">35.10.5 Free Functions</a></li>
<li><a HREF="#Ruby_nn63">35.10.6 Embedded Ruby and the C++ Stack</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Scilab">36 SWIG and Scilab</a></b>
<ul>
<li><a HREF="#Scilab_preliminaries">36.1 Preliminaries</a></li>
<li><a HREF="#Scilab_running_swig">36.2 Running SWIG</a>
<ul>
<li><a HREF="#Scilab_running_swig_generating_module">36.2.1 Generating
the module</a></li>
<li><a HREF="#Scilab_running_swig_building_module">36.2.2 Building the
module</a></li>
<li><a HREF="#Scilab_running_swig_loading_module">36.2.3 Loading the
module</a></li>
<li><a HREF="#Scilab_running_swig_using_module">36.2.4 Using the module</a>
</li>
<li><a HREF="#Scilab_running_swig_options">36.2.5 Scilab command line
options</a></li>
</ul>
</li>
<li><a HREF="#Scilab_wrapping">36.3 A basic tour of C/C++ wrapping</a>
<ul>
<li><a HREF="#Scilab_wrapping_overview">36.3.1 Overview</a></li>
<li><a HREF="#Scilab_wrapping_identifiers">36.3.2 Identifiers</a></li>
<li><a HREF="#Scilab_wrapping_functions">36.3.3 Functions</a>
<ul>
<li><a HREF="#Scilab_nn13">36.3.3.1 Argument passing</a></li>
<li><a HREF="#Scilab_nn14">36.3.3.2 Multiple output arguments</a></li>
</ul>
</li>
<li><a HREF="#Scilab_wrapping_global_variables">36.3.4 Global variables</a>
</li>
<li><a HREF="#Scilab_wrapping_constants_and_enums">36.3.5 Constants and
enumerations</a>
<ul>
<li><a HREF="#Scilab_wrapping_constants">36.3.5.1 Constants</a></li>
<li><a HREF="#Scilab_wrapping_enums">36.3.5.2 Enumerations</a></li>
</ul>
</li>
<li><a HREF="#Scilab_wrapping_pointers">36.3.6 Pointers</a>
<ul>
<li><a HREF="#Scilab_wrapping_pointers_utility_functions">36.3.6.1
Utility functions</a></li>
<li><a HREF="#Scilab_wrapping_pointers_null_pointers">36.3.6.2 Null
pointers:</a></li>
</ul>
</li>
<li><a HREF="#Scilab_wrapping_structs">36.3.7 Structures</a></li>
<li><a HREF="#Scilab_wrapping_cpp_classes">36.3.8 C++ classes</a></li>
<li><a HREF="#Scilab_wrapping_cpp_inheritance">36.3.9 C++ inheritance</a>
</li>
<li><a HREF="#Scilab_wrapping_cpp_overloading">36.3.10 C++ overloading</a>
</li>
<li><a HREF="#Scilab_wrapping_pointers_references_values_arrays">36.3.11
Pointers, references, values, and arrays</a></li>
<li><a HREF="#Scilab_wrapping_cpp_templates">36.3.12 C++ templates</a></li>
<li><a HREF="#Scilab_wrapping_cpp_operators">36.3.13 C++ operators</a></li>
<li><a HREF="#Scilab_wrapping_cpp_namespaces">36.3.14 C++ namespaces</a></li>
<li><a HREF="#Scilab_wrapping_cpp_exceptions">36.3.15 C++ exceptions</a></li>
<li><a HREF="#Scilab_wrapping_cpp_stl">36.3.16 C++ STL</a></li>
</ul>
</li>
<li><a HREF="#Scilab_typemaps">36.4 Type mappings and libraries</a>
<ul>
<li><a HREF="#Scilab_typemaps_primitive_types">36.4.1 Default primitive
type mappings</a></li>
<li><a HREF="#Scilab_typemaps_arrays">36.4.2 Arrays</a></li>
<li><a HREF="#Scilab_typemaps_pointer-to-pointers">36.4.3
Pointer-to-pointers</a></li>
<li><a HREF="#Scilab_typemaps_matrices">36.4.4 Matrices</a></li>
<li><a HREF="#Scilab_typemaps_stl">36.4.5 STL</a></li>
</ul>
</li>
<li><a HREF="#Scilab_module_initialization">36.5 Module initialization</a>
</li>
<li><a HREF="#Scilab_building_modes">36.6 Building modes</a>
<ul>
<li><a HREF="#Scilab_building_modes_nobuilder_mode">36.6.1 No-builder
mode</a></li>
<li><a HREF="#Scilab_building_modes_builder_mode">36.6.2 Builder mode</a>
</li>
</ul>
</li>
<li><a HREF="#Scilab_generated_scripts">36.7 Generated scripts</a>
<ul>
<li><a HREF="#Scilab_generated_scripts_builder_script">36.7.1 Builder
script</a></li>
<li><a HREF="#Scilab_generated_scripts_loader_script">36.7.2 Loader
script</a></li>
<li><a HREF="#Scilab_generated_scripts_gateway">36.7.3 Gateway XML files</a>
</li>
</ul>
</li>
<li><a HREF="#Scilab_other_resources">36.8 Other resources</a></li>
</ul>
<b><a HREF="#Tcl">37 SWIG and Tcl</a></b>
<ul>
<li><a HREF="#Tcl_nn2">37.1 Preliminaries</a>
<ul>
<li><a HREF="#Tcl_nn3">37.1.1 Getting the right header files</a></li>
<li><a HREF="#Tcl_nn4">37.1.2 Compiling a dynamic module</a></li>
<li><a HREF="#Tcl_nn5">37.1.3 Static linking</a></li>
<li><a HREF="#Tcl_nn6">37.1.4 Using your module</a></li>
<li><a HREF="#Tcl_nn7">37.1.5 Compilation of C++ extensions</a></li>
<li><a HREF="#Tcl_nn8">37.1.6 Compiling for 64-bit platforms</a></li>
<li><a HREF="#Tcl_nn9">37.1.7 Setting a package prefix</a></li>
<li><a HREF="#Tcl_nn10">37.1.8 Using namespaces</a></li>
</ul>
</li>
<li><a HREF="#Tcl_nn11">37.2 Building Tcl/Tk Extensions under Windows
95/NT</a>
<ul>
<li><a HREF="#Tcl_nn12">37.2.1 Running SWIG from Developer Studio</a></li>
<li><a HREF="#Tcl_nn13">37.2.2 Using NMAKE</a></li>
</ul>
</li>
<li><a HREF="#Tcl_nn14">37.3 A tour of basic C/C++ wrapping</a>
<ul>
<li><a HREF="#Tcl_nn15">37.3.1 Modules</a></li>
<li><a HREF="#Tcl_nn16">37.3.2 Functions</a></li>
<li><a HREF="#Tcl_nn17">37.3.3 Global variables</a></li>
<li><a HREF="#Tcl_nn18">37.3.4 Constants and enums</a></li>
<li><a HREF="#Tcl_nn19">37.3.5 Pointers</a></li>
<li><a HREF="#Tcl_nn20">37.3.6 Structures</a></li>
<li><a HREF="#Tcl_nn21">37.3.7 C++ classes</a></li>
<li><a HREF="#Tcl_nn22">37.3.8 C++ inheritance</a></li>
<li><a HREF="#Tcl_nn23">37.3.9 Pointers, references, values, and arrays</a>
</li>
<li><a HREF="#Tcl_nn24">37.3.10 C++ overloaded functions</a></li>
<li><a HREF="#Tcl_nn25">37.3.11 C++ operators</a></li>
<li><a HREF="#Tcl_nn26">37.3.12 C++ namespaces</a></li>
<li><a HREF="#Tcl_nn27">37.3.13 C++ templates</a></li>
<li><a HREF="#Tcl_nn28">37.3.14 C++ Smart Pointers</a></li>
</ul>
</li>
<li><a HREF="#Tcl_nn29">37.4 Further details on the Tcl class interface</a>
<ul>
<li><a HREF="#Tcl_nn30">37.4.1 Proxy classes</a></li>
<li><a HREF="#Tcl_nn31">37.4.2 Memory management</a></li>
</ul>
</li>
<li><a HREF="#Tcl_nn32">37.5 Input and output parameters</a></li>
<li><a HREF="#Tcl_nn33">37.6 Exception handling</a></li>
<li><a HREF="#Tcl_nn34">37.7 Typemaps</a>
<ul>
<li><a HREF="#Tcl_nn35">37.7.1 What is a typemap?</a></li>
<li><a HREF="#Tcl_nn36">37.7.2 Tcl typemaps</a></li>
<li><a HREF="#Tcl_nn37">37.7.3 Typemap variables</a></li>
<li><a HREF="#Tcl_nn38">37.7.4 Converting a Tcl list to a char **</a></li>
<li><a HREF="#Tcl_nn39">37.7.5 Returning values in arguments</a></li>
<li><a HREF="#Tcl_nn40">37.7.6 Useful functions</a></li>
<li><a HREF="#Tcl_nn41">37.7.7 Standard typemaps</a></li>
<li><a HREF="#Tcl_nn42">37.7.8 Pointer handling</a></li>
</ul>
</li>
<li><a HREF="#Tcl_nn43">37.8 Turning a SWIG module into a Tcl Package.</a>
</li>
<li><a HREF="#Tcl_nn44">37.9 Building new kinds of Tcl interfaces (in
Tcl)</a>
<ul>
<li><a HREF="#Tcl_nn45">37.9.1 Proxy classes</a></li>
</ul>
</li>
<li><a HREF="#Tcl_nn46">37.10 Tcl/Tk Stubs</a></li>
</ul>
<b><a HREF="#C">38 SWIG and C as the target language</a></b>
<ul>
<li><a HREF="#C_overview">38.1 Overview</a>
<ul>
<li><a HREF="#C_shortcomings">38.1.1 Known C++ Shortcomings in Generated
C API</a></li>
</ul>
</li>
<li><a HREF="#C_preliminaries">38.2 Preliminaries</a>
<ul>
<li><a HREF="#C_running_swig">38.2.1 Running SWIG</a></li>
<li><a HREF="#C_commandline">38.2.2 Command line options</a></li>
<li><a HREF="#C_dynamic">38.2.3 Compiling a dynamic module</a></li>
<li><a HREF="#C_using_module">38.2.4 Using the generated module</a></li>
</ul>
</li>
<li><a HREF="#C_basic_c_wrapping">38.3 Basic C wrapping</a>
<ul>
<li><a HREF="#C_functions">38.3.1 Functions</a></li>
<li><a HREF="#C_variables">38.3.2 Variables</a></li>
</ul>
</li>
<li><a HREF="#C_basic_cpp_wrapping">38.4 Basic C++ wrapping</a>
<ul>
<li><a HREF="#C_enums">38.4.1 Enums</a></li>
<li><a HREF="#C_classes">38.4.2 Classes</a></li>
</ul>
</li>
<li><a HREF="#C_developer">38.5 Backend Developer Documentation</a></li>
<li><a HREF="#C_typemaps">38.6 Typemaps</a>
<ul>
<li><a HREF="#C_typemaps_walkthrough">38.6.1 C Typemaps, a Code
Generation Walkthrough</a>
<ul>
<li><a HREF="#C_walkthrough_interface">38.6.1.1 The Interface</a></li>
<li><a HREF="#C_walkthrough_wrapper">38.6.1.2 The Wrapper</a></li>
<li><a HREF="#C_walkthrough_proxy">38.6.1.3 The Proxy</a></li>
</ul>
</li>
</ul>
</li>
<li><a HREF="#C_exceptions">38.7 Exception handling</a></li>
<li><a HREF="#C_cxx_wrappers">38.8 C++ Wrappers</a>
<ul>
<li><a HREF="#C_additional_possibilities">38.8.1 Additional
customization possibilities</a></li>
<li><a HREF="#C_exception_handling">38.8.2 Exception handling</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Mzscheme">39 SWIG and MzScheme/Racket</a></b>
<ul>
<li><a HREF="#MzScheme_nn2">39.1 Creating native structures</a></li>
<li><a HREF="#MzScheme_simple">39.2 Simple example</a></li>
<li><a HREF="#MzScheme_external_docs">39.3 External documentation</a></li>
</ul>
<b><a HREF="#Ocaml">40 SWIG and OCaml</a></b>
<ul>
<li><a HREF="#Ocaml_nn2">40.1 Preliminaries</a>
<ul>
<li><a HREF="#Ocaml_nn3">40.1.1 Running SWIG</a></li>
<li><a HREF="#Ocaml_nn4">40.1.2 Compiling the code</a></li>
<li><a HREF="#Ocaml_nn5">40.1.3 The camlp4 module</a></li>
<li><a HREF="#Ocaml_nn6">40.1.4 Using your module</a></li>
<li><a HREF="#Ocaml_nn7">40.1.5 Compilation problems and compiling with
C++</a></li>
</ul>
</li>
<li><a HREF="#Ocaml_nn8">40.2 The low-level Ocaml/C interface</a>
<ul>
<li><a HREF="#Ocaml_nn9">40.2.1 The generated module</a></li>
<li><a HREF="#Ocaml_nn10">40.2.2 Enums</a>
<ul>
<li><a HREF="#Ocaml_nn11">40.2.2.1 Enum typing in Ocaml</a></li>
</ul>
</li>
<li><a HREF="#Ocaml_nn12">40.2.3 Arrays</a>
<ul>
<li><a HREF="#Ocaml_nn13">40.2.3.1 Simple types of bounded arrays</a></li>
<li><a HREF="#Ocaml_nn14">40.2.3.2 Complex and unbounded arrays</a></li>
<li><a HREF="#Ocaml_nn15">40.2.3.3 Using an object</a></li>
<li><a HREF="#Ocaml_nn16">40.2.3.4 Example typemap for a function taking
float * and int</a></li>
</ul>
</li>
<li><a HREF="#Ocaml_nn17">40.2.4 C++ Classes</a>
<ul>
<li><a HREF="#Ocaml_nn18">40.2.4.1 STL vector and string Example</a></li>
<li><a HREF="#Ocaml_nn19">40.2.4.2 C++ Class Example</a></li>
<li><a HREF="#Ocaml_nn20">40.2.4.3 Compiling the example</a></li>
<li><a HREF="#Ocaml_nn21">40.2.4.4 Sample Session</a></li>
</ul>
</li>
<li><a HREF="#Ocaml_nn22">40.2.5 Director Classes</a>
<ul>
<li><a HREF="#Ocaml_nn23">40.2.5.1 Director Introduction</a></li>
<li><a HREF="#Ocaml_nn24">40.2.5.2 Overriding Methods in Ocaml</a></li>
<li><a HREF="#Ocaml_nn25">40.2.5.3 Director Usage Example</a></li>
<li><a HREF="#Ocaml_nn26">40.2.5.4 Creating director objects</a></li>
<li><a HREF="#Ocaml_nn27">40.2.5.5 Typemaps for directors, directorin,
directorout, directorargout</a></li>
<li><a HREF="#Ocaml_nn28">40.2.5.6 directorin typemap</a></li>
<li><a HREF="#Ocaml_nn29">40.2.5.7 directorout typemap</a></li>
<li><a HREF="#Ocaml_nn30">40.2.5.8 directorargout typemap</a></li>
</ul>
</li>
<li><a HREF="#Ocaml_nn31">40.2.6 Exceptions</a></li>
</ul>
</li>
<li><a HREF="#Ocaml_nn32">40.3 Documentation Features</a>
<ul>
<li><a HREF="#Ocaml_nn33">40.3.1 Module docstring</a></li>
</ul>
</li>
</ul>
<b><a HREF="#Extending">41 Extending SWIG to support new languages</a></b>
<ul>
<li><a HREF="#Extending_nn2">41.1 Introduction</a></li>
<li><a HREF="#Extending_nn3">41.2 Prerequisites</a></li>
<li><a HREF="#Extending_nn4">41.3 The Big Picture</a></li>
<li><a HREF="#Extending_nn5">41.4 Execution Model</a>
<ul>
<li><a HREF="#Extending_nn6">41.4.1 Preprocessing</a></li>
<li><a HREF="#Extending_nn7">41.4.2 Parsing</a></li>
<li><a HREF="#Extending_nn8">41.4.3 Parse Trees</a></li>
<li><a HREF="#Extending_nn9">41.4.4 Attribute namespaces</a></li>
<li><a HREF="#Extending_nn10">41.4.5 Symbol Tables</a></li>
<li><a HREF="#Extending_nn11">41.4.6 The %feature directive</a></li>
<li><a HREF="#Extending_nn12">41.4.7 Code Generation</a></li>
<li><a HREF="#Extending_nn13">41.4.8 SWIG and XML</a></li>
</ul>
</li>
<li><a HREF="#Extending_nn14">41.5 Primitive Data Structures</a>
<ul>
<li><a HREF="#Extending_nn15">41.5.1 Strings</a></li>
<li><a HREF="#Extending_nn16">41.5.2 Hashes</a></li>
<li><a HREF="#Extending_nn17">41.5.3 Lists</a></li>
<li><a HREF="#Extending_nn18">41.5.4 Common operations</a></li>
<li><a HREF="#Extending_nn19">41.5.5 Iterating over Lists and Hashes</a></li>
<li><a HREF="#Extending_nn20">41.5.6 I/O</a></li>
</ul>
</li>
<li><a HREF="#Extending_nn21">41.6 Navigating and manipulating parse
trees</a></li>
<li><a HREF="#Extending_nn22">41.7 Working with attributes</a></li>
<li><a HREF="#Extending_nn23">41.8 Type system</a>
<ul>
<li><a HREF="#Extending_nn24">41.8.1 String encoding of types</a></li>
<li><a HREF="#Extending_nn25">41.8.2 Type construction</a></li>
<li><a HREF="#Extending_nn26">41.8.3 Type tests</a></li>
<li><a HREF="#Extending_nn27">41.8.4 Typedef and inheritance</a></li>
<li><a HREF="#Extending_nn28">41.8.5 Lvalues</a></li>
<li><a HREF="#Extending_nn29">41.8.6 Output functions</a></li>
</ul>
</li>
<li><a HREF="#Extending_nn30">41.9 Parameters</a></li>
<li><a HREF="#Extending_nn31">41.10 Writing a Language Module</a>
<ul>
<li><a HREF="#Extending_nn32">41.10.1 Execution model</a></li>
<li><a HREF="#Extending_starting_out">41.10.2 Starting out</a></li>
<li><a HREF="#Extending_nn34">41.10.3 Command line options</a></li>
<li><a HREF="#Extending_nn35">41.10.4 Configuration and preprocessing</a>
</li>
<li><a HREF="#Extending_nn36">41.10.5 Entry point to code generation</a></li>
<li><a HREF="#Extending_nn37">41.10.6 Module I/O and wrapper skeleton</a>
</li>
<li><a HREF="#Extending_nn38">41.10.7 Low-level code generators</a></li>
<li><a HREF="#Extending_configuration_files">41.10.8 Configuration files</a>
</li>
<li><a HREF="#Extending_nn40">41.10.9 Runtime support</a></li>
<li><a HREF="#Extending_nn41">41.10.10 Standard library files</a></li>
<li><a HREF="#Extending_nn42">41.10.11 User examples</a></li>
<li><a HREF="#Extending_test_suite">41.10.12 Test driven development and
the test-suite</a>
<ul>
<li><a HREF="#Extending_running_test_suite">41.10.12.1 Running the
test-suite</a></li>
</ul>
</li>
<li><a HREF="#Extending_nn43">41.10.13 Documentation</a></li>
<li><a HREF="#Extending_coding_style_guidelines">41.10.14 Coding style
guidelines</a></li>
<li><a HREF="#Extending_language_status">41.10.15 Target language status</a>
<ul>
<li><a HREF="#Extending_supported_status">41.10.15.1 Supported status</a>
</li>
<li><a HREF="#Extending_experimental_status">41.10.15.2 Experimental
status</a></li>
<li><a HREF="#Extending_deprecated_status">41.10.15.3 Deprecated status</a>
</li>
</ul>
</li>
<li><a HREF="#Extending_prerequisites">41.10.16 Prerequisites for adding
a new language module to the SWIG distribution</a></li>
</ul>
</li>
<li><a HREF="#Extending_debugging_options">41.11 Debugging Options</a></li>
<li><a HREF="#Extending_nn46">41.12 Guide to parse tree nodes</a></li>
<li><a HREF="#Extending_further_info">41.13 Further Development
Information</a></li>
</ul>
<HR NOSHADE>
<h1><a name="Sections">SWIG-4.3 Documentation</a></h1>
<p> Last update : SWIG-4.3.0 (20 Oct 2024)</p>
<h2><a name="Sections_Sections">Sections</a></h2>
<h3><a name="Sections_core_docs">SWIG Core Documentation</a></h3>
<ul>
<li><a href="#Preface">Preface</a></li>
<li><a href="#Introduction">Introduction</a></li>
<li><a href="#Windows">Getting started on Windows</a></li>
<li><a href="#Scripting">Scripting</a></li>
<li><a href="#SWIG">SWIG Basics</a> (Read this!)</li>
<li><a href="#SWIGPlus">SWIG and C++</a></li>
<li><a href="#CPlusPlus11">SWIG and C++11</a></li>
<li><a href="#CPlusPlus14">SWIG and C++14</a></li>
<li><a href="#CPlusPlus17">SWIG and C++17</a></li>
<li><a href="#CPlusPlus20">SWIG and C++20</a></li>
<li><a href="#Preprocessor">The SWIG preprocessor</a></li>
<li><a href="#Library">The SWIG library</a></li>
<li><a href="#Arguments">Argument handling</a></li>
<li><a href="#Typemaps">Typemaps</a></li>
<li><a href="#Customization">Customization features</a></li>
<li><a href="#Contract">Contracts</a></li>
<li><a href="#Varargs">Variable length arguments</a></li>
<li><a href="#Doxygen">Doxygen documentation comments</a></li>
<li><a href="#Warnings">Warning messages</a></li>
<li><a href="#Modules">Working with Modules</a></li>
<li><a href="#CCache">Using SWIG with ccache</a></li>
</ul>
<h3><a name="Sections_language_modules">Supported Language Modules
Documentation</a></h3>
<ul>
<li><a href="#Android">Android support</a></li>
<li><a href="#CSharp">C# support</a></li>
<li><a href="#D">D support</a></li>
<li><a href="#Go">Go support</a></li>
<li><a href="#Guile">Guile support</a></li>
<li><a href="#Java">Java support</a></li>
<li><a href="#Javascript">Javascript support</a></li>
<li><a href="#Lua">Lua support</a></li>
<li><a href="#Octave">Octave support</a></li>
<li><a href="#Perl5">Perl5 support</a></li>
<li><a href="#Php">PHP support</a></li>
<li><a href="#Python">Python support</a></li>
<li><a href="#R">R support</a></li>
<li><a href="#Ruby">Ruby support</a></li>
<li><a href="#Scilab">Scilab support</a></li>
<li><a href="#Tcl">Tcl support</a></li>
</ul>
<h3><a name="Sections_experimental_language_modules">Experimental
Language Modules Documentation</a></h3>
<ul>
<li><a href="#C">C target language support</a></li>
<li><a href="#Ocaml">OCaml support</a></li>
</ul>
<h3><a name="Sections_deprecated_language_modules">Deprecated Language
Modules Documentation</a></h3>
<ul>
<li><a href="#Mzscheme">MzScheme/Racket support</a></li>
</ul>
<h3><a name="Sections_developers_docs">Developer Documentation</a></h3>
<ul>
<li><a href="#Extending">Extending SWIG</a></li>
</ul>
<HR NOSHADE>
<h1><a name="Preface">1 Preface</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Preface_nn2">Introduction</a></li>
<li><a href="#Preface_nn4">SWIG Versions</a></li>
<li><a href="#Preface_license">SWIG License</a></li>
<li><a href="#Preface_nn5">SWIG resources</a></li>
<li><a href="#Preface_nn6">Prerequisites</a></li>
<li><a href="#Preface_nn7">Organization of this manual</a></li>
<li><a href="#Preface_nn8">How to avoid reading the manual</a></li>
<li><a href="#Preface_nn9">Backwards compatibility</a></li>
<li><a href="#Preface_release_notes">Release notes</a></li>
<li><a href="#Preface_nn10">Credits</a></li>
<li><a href="#Preface_nn11">Bug reports</a></li>
<li><a href="#Preface_installation">Installation</a>
<ul>
<li><a href="#Preface_windows_installation">Windows installation</a></li>
<li><a href="#Preface_unix_installation">Unix installation</a></li>
<li><a href="#Preface_osx_installation">Macintosh OS X installation</a></li>
<li><a href="#Preface_testing">Testing</a></li>
<li><a href="#Preface_examples">Examples</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="Preface_nn2">1.1 Introduction</a></h2>
<p> SWIG (Simplified Wrapper and Interface Generator) is a software
development tool for building scripting language interfaces to C and
C++ programs. Originally developed in 1995, SWIG was first used by
scientists in the Theoretical Physics Division at Los Alamos National
Laboratory for building user interfaces to simulation codes running on
the Connection Machine 5 supercomputer. In this environment, scientists
needed to work with huge amounts of simulation data, complex hardware,
and a constantly changing code base. The use of a scripting language
interface provided a simple yet highly flexible foundation for solving
these types of problems. SWIG simplifies development by largely
automating the task of scripting language integration--allowing
developers and users to focus on more important problems.</p>
<p> Although SWIG was originally developed for scientific applications,
it has since evolved into a general purpose tool that is used in a wide
variety of applications--in fact almost anything where C/C++
programming is involved.</p>
<h2><a name="Preface_nn4">1.2 SWIG Versions</a></h2>
<p> In the late 1990's, the most stable version of SWIG was release
1.1p5. Versions 1.3.x were officially development versions and these
were released over a period of 10 years starting from the year 2000.
The final version in the 1.3.x series was 1.3.40, but in truth the
1.3.x series had been stable for many years. An official stable version
was released along with the decision to make SWIG license changes and
this gave rise to version 2.0.0 in 2010. Version 3.0.0 was released in
2014 focusing on adding C++11 support and C++ nested classes. Version
4.0.0 was released in 2019 to add in Doxygen support. Some target
languages were disabled as part of a clean up and others were given a
new status of either 'Supported' or 'Experimental'.</p>
<h2><a name="Preface_license">1.3 SWIG License</a></h2>
<p> The LICENSE file shipped with SWIG in the top level directory
contains the SWIG license. For further insight into the license
including the license of SWIG's output code, please visit the SWIG
legal page - <a href="https://www.swig.org/legal.html">
https://www.swig.org/legal.html</a>.</p>
<p> The license was clarified in version 2.0.0 so that the code that
SWIG generated could be distributed under license terms of the user's
choice/requirements and at the same time the SWIG source was placed
under the GNU General Public License version 3.</p>
<h2><a name="Preface_nn5">1.4 SWIG resources</a></h2>
<p> The official location of SWIG related material is</p>
<div class="shell">
<pre>
<a href="https://www.swig.org">https://www.swig.org</a>
</pre>
</div>
<p> This site contains the latest version of the software, users guide,
and information regarding bugs, installation problems, and
implementation tricks.</p>
<p> You can also subscribe to the swig-user mailing list by visiting the
page</p>
<div class="shell">
<pre>
<a href="https://www.swig.org/mail.html">https://www.swig.org/mail.html</a>
</pre>
</div>
<p> The mailing list often discusses some of the more technical aspects
of SWIG along with information about beta releases and future work.</p>
<p> Git and Subversion access to the latest version of SWIG is also
available. More information about this can be obtained at:</p>
<div class="shell">
<pre>
<a href="https://www.swig.org/svn.html">SWIG Bleeding Edge</a>
</pre>
</div>
<h2><a name="Preface_nn6">1.5 Prerequisites</a></h2>
<p> This manual assumes that you know how to write C/C++ programs and
that you have at least heard of scripting languages such as Tcl,
Python, and Perl. A detailed knowledge of these scripting languages is
not required although some familiarity won't hurt. No prior experience
with building C extensions to these languages is required---after all,
this is what SWIG does automatically. However, you should be reasonably
familiar with the use of compilers, linkers, and makefiles since making
scripting language extensions is somewhat more complicated than writing
a normal C program.</p>
<p> Over time SWIG releases have become significantly more capable in
their C++ handling--especially support for advanced features like
namespaces, overloaded operators, and templates. Whenever possible,
this manual tries to cover the technicalities of this interface.
However, this isn't meant to be a tutorial on C++ programming. For many
of the gory details, you will almost certainly want to consult a good
C++ reference. If you don't program in C++, you may just want to skip
those parts of the manual.</p>
<h2><a name="Preface_nn7">1.6 Organization of this manual</a></h2>
<p> The first few chapters of this manual describe SWIG in general and
provide an overview of its capabilities. The remaining chapters are
devoted to specific SWIG language modules and are self contained. Thus,
if you are using SWIG to build Python interfaces, you can probably skip
to that chapter and find almost everything you need to know.</p>
<h2><a name="Preface_nn8">1.7 How to avoid reading the manual</a></h2>
<p> If you hate reading manuals, glance at the &quot;Introduction&quot; which
contains a few simple examples. These examples contain about 95% of
everything you need to know to use SWIG. After that, simply use the
language-specific chapters as a reference. The SWIG distribution also
comes with a large directory of examples that illustrate different
topics.</p>
<h2><a name="Preface_nn9">1.8 Backwards compatibility</a></h2>
<p> If you are a previous user of SWIG, don't expect SWIG to provide
complete backwards compatibility. Although the developers strive to the
utmost to keep backwards compatibility, this isn't always possible as
the primary goal over time is to make SWIG better---a process that
would simply be impossible if the developers are constantly bogged down
with backwards compatibility issues. Potential incompatibilities are
clearly marked in the detailed <a href="#Preface_release_notes">release
notes</a>.</p>
<p> If you need to work with different versions of SWIG and backwards
compatibility is an issue, you can use the SWIG_VERSION preprocessor
symbol which holds the version of SWIG being executed. SWIG_VERSION is
a hexadecimal integer such as 0x010311 (corresponding to SWIG-1.3.11).
This can be used in an interface file to define different typemaps,
take advantage of different features etc:</p>
<div class="code">
<pre>
#if SWIG_VERSION &gt;= 0x010311
/* Use some fancy new SWIG feature */
#endif
</pre>
</div>
<p> Note: The SWIG preprocessor has defined SWIG_VERSION since
SWIG-1.3.11.</p>
<p> The SWIG_VERSION macro is also generated into the SWIG wrapper file
for use by the C preprocessor in the generated code since SWIG-4.1.0.</p>
<h2><a name="Preface_release_notes">1.9 Release notes</a></h2>
<p> The CHANGES.current, CHANGES and RELEASENOTES files shipped with
SWIG in the top level directory contain, respectively, detailed release
notes for the current version, detailed release notes for previous
releases and summary release notes from SWIG-1.3.22 onwards.</p>
<h2><a name="Preface_nn10">1.10 Credits</a></h2>
<p> SWIG is an unfunded project that would not be possible without the
contributions of many people working in their spare time. If you have
benefitted from using SWIG, please consider <a href="https://www.swig.org/donate.html">
Donating to SWIG</a> to keep development going. There have been a large
varied number of people who have made contributions at all levels over
time. Contributors are mentioned either in the COPYRIGHT file or
CHANGES files shipped with SWIG or in submitted bugs.</p>
<h2><a name="Preface_nn11">1.11 Bug reports</a></h2>
<p> Although every attempt has been made to make SWIG bug-free, we are
also trying to make feature improvements that may introduce bugs. To
report a bug, either send mail to the SWIG developer list at the <a href="https://www.swig.org/mail.html">
swig-devel mailing list</a> or report a bug at the <a href="https://www.swig.org/bugs.html">
SWIG bug tracker</a>. In your report, be as specific as possible,
including (if applicable), error messages, tracebacks (if a core dump
occurred), corresponding portions of the SWIG interface file used, and
any important pieces of the SWIG generated wrapper code. We can only
fix bugs if we know about them.</p>
<h2><a name="Preface_installation">1.12 Installation</a></h2>
<h3><a name="Preface_windows_installation">1.12.1 Windows installation</a>
</h3>
<p> Please see the dedicated <a href="#Windows">Windows chapter</a> for
instructions on installing SWIG on Windows and running the examples.
The Windows distribution is called swigwin and includes a prebuilt SWIG
executable, swig.exe, included in the top level directory. Otherwise it
is exactly the same as the main SWIG distribution. There is no need to
download anything else.</p>
<h3><a name="Preface_unix_installation">1.12.2 Unix installation</a></h3>
<p> These installation instructions are for using the distributed
tarball, for example, <tt>swig-3.0.8.tar.gz</tt>. If you wish to build
and install from source on Github, extra steps are required. Please see
the <a href="https://swig.org/svn.html">Bleeding Edge</a> page on the
SWIG website.</p>
<p> You must use <a href="https://www.gnu.org/software/make/">GNU make</a>
to build and install SWIG.</p>
<p> <a href="https://www.pcre.org/">PCRE2</a> needs to be installed on
your system to build SWIG, in particular pcre2-config must be
available. If you have PCRE2 headers and libraries but not pcre2-config
itself or, alternatively, wish to override the compiler or linker flags
returned by pcre2-config, you may set PCRE2_LIBS and PCRE2_CFLAGS
variables to be used instead. And if you don't have PCRE2 at all, the
configure script will provide instructions for obtaining it.</p>
<p> To build and install SWIG, simply type the following:</p>
<div class="shell">
<pre>
$ ./configure
$ make
$ make install
</pre>
</div>
<p> By default SWIG installs itself in /usr/local. If you need to
install SWIG in a different location or in your home directory, use the
<tt>--prefix</tt> option to <tt>./configure</tt>. For example:</p>
<div class="shell">
<pre>
$ ./configure --prefix=/home/yourname/projects
$ make
$ make install
</pre>
</div>
<p> Note: the directory given to <tt>--prefix</tt> must be an absolute
pathname. Do<b> not</b> use the ~ shell-escape to refer to your home
directory. SWIG won't work properly if you do this.</p>
<p> The INSTALL file shipped in the top level directory details more
about using configure. Also try</p>
<div class="shell">
<pre>
$ ./configure --help.
</pre>
</div>
<p> The configure script will attempt to locate various packages on your
machine including Tcl, Perl5, Python and all the other target languages
that SWIG supports. Don't panic if you get 'not found' messages -- SWIG
does not need these packages to compile or run. The configure script is
actually looking for these packages so that you can try out the SWIG
examples contained in the 'Examples' directory without having to hack
Makefiles. Note that the <tt>--without-xxx</tt> options, where xxx is a
target language, have minimal effect. All they do is reduce the amount
of testing done with 'make check'. The SWIG executable and library
files installed cannot currently be configured with a subset of target
languages.</p>
<p> SWIG used to include a set of runtime libraries for some languages
for working with multiple modules. These are no longer built during the
installation stage. However, users can build them just like any wrapper
module as described in the <a href="#Modules">Modules chapter</a>. The
CHANGES file shipped with SWIG in the top level directory also lists
some examples which build the runtime library.</p>
<p> Note:</p>
<ul>
<li> If you checked the code out via Git, you will have to run <tt>
./autogen.sh</tt> before <tt>./configure</tt>. In addition, a full build
of SWIG requires a number of packages to be installed. Full
instructions at <a href="https://www.swig.org/svn.html">SWIG bleeding
edge</a>.</li>
</ul>
<h3><a name="Preface_osx_installation">1.12.3 Macintosh OS X
installation</a></h3>
<p> SWIG is known to work on various flavors of OS X. Follow the Unix
installation instructions above. However, as of this writing, there is
still great deal of inconsistency with how shared libraries are handled
by various scripting languages on OS X.</p>
<p> Users of OS X should be aware that Darwin handles shared libraries
and linking in a radically different way than most Unix systems. In
order to test SWIG and run the examples, SWIG configures itself to use
flat namespaces and to allow undefined symbols (<tt>-flat_namespace
-undefined suppress</tt>). This mostly closely follows the Unix model
and makes it more likely that the SWIG examples will work with whatever
installation of software you might have. However, this is generally not
the recommended technique for building larger extension modules.
Instead, you should utilize Darwin's two-level namespaces. Some details
about this can be found here <a href="https://developer.apple.com/library/mac/documentation/Porting/Conceptual/PortingUnix/compiling/compiling.html#//apple_ref/doc/uid/TP40002850-BCIHJBBF">
Understanding Two-Level Namespaces</a>.</p>
<p> Needless to say, you might have to experiment a bit to get things
working at first.</p>
<h3><a name="Preface_testing">1.12.4 Testing</a></h3>
<p> If you want to test SWIG after building it, a check can be performed
on Unix operating systems. Type the following:</p>
<div class="shell">
<pre>
$ make -k check
</pre>
</div>
<p> This step can be performed either before or after installation. The
check requires at least one of the target languages to be installed. If
it fails, it may mean that you have an uninstalled language module or
that the file 'Examples/Makefile' has been incorrectly configured. It
may also fail due to compiler issues such as a broken C++ compiler.
Even if the check fails, there is a pretty good chance SWIG still works
correctly --- you will just have to mess around with one of the
examples and some makefiles to get it to work. Some tests may also fail
due to missing dependency packages, eg PCRE or Boost, but this will
require careful analysis of the configure output done during
configuration.</p>
<p> The test suite executed by the check is designed to stress-test many
parts of the implementation including obscure corner cases. If some of
these tests fail or generate warning messages, there is no reason for
alarm --- the test may be related to some new SWIG feature or a
difficult bug that we're trying to resolve. Chances are that SWIG will
work just fine for you. Note that if you have more than one CPU/core,
then you can use parallel make to speed up the check as it does take
quite some time to run, for example:</p>
<div class="shell">
<pre>
$ make -j2 -k check
</pre>
</div>
<p> Also, SWIG's support for C++ is sufficiently advanced that certain
tests may fail on older C++ compilers (for instance if your compiler
does not support member templates). These errors are harmless if you
don't intend to use these features in your own programs.</p>
<p> Note: The test-suite currently contains over 800 tests. If you have
many different target languages installed and a slow machine, it might
take more than an hour to run the test-suite.</p>
<h3><a name="Preface_examples">1.12.5 Examples</a></h3>
<p> The Examples directory contains a variety of examples of using SWIG
and it has some browsable documentation. Simply point your browser to
the file &quot;Example/index.html&quot;.</p>
<p> The Examples directory also includes Visual C++ project 6 (.dsp)
files for building some of the examples on Windows. Later versions of
Visual Studio will convert these old style project files into a current
solution file.</p>
<HR NOSHADE>
<h1><a name="Introduction">2 Introduction</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Introduction_nn2">What is SWIG?</a></li>
<li><a href="#Introduction_nn3">Why use SWIG?</a></li>
<li><a href="#Introduction_target_languages">Target languages</a>
<ul>
<li><a href="#Introduction_supported_status">Supported status</a></li>
<li><a href="#Introduction_experimental_status">Experimental status</a></li>
<li><a href="#Introduction_deprecated_status">Deprecated status</a></li>
</ul>
</li>
<li><a href="#Introduction_nn4">A SWIG example</a>
<ul>
<li><a href="#Introduction_nn5">SWIG interface file</a></li>
<li><a href="#Introduction_nn6">The swig command</a></li>
<li><a href="#Introduction_nn7">Building a Perl5 module</a></li>
<li><a href="#Introduction_nn8">Building a Python module</a></li>
<li><a href="#Introduction_nn9">Shortcuts</a></li>
</ul>
</li>
<li><a href="#Introduction_nn10">Supported C/C++ language features</a></li>
<li><a href="#Introduction_nn11">Non-intrusive interface building</a></li>
<li><a href="#Introduction_build_system">Incorporating SWIG into a build
system</a></li>
<li><a href="#Introduction_nn12">Hands off code generation</a></li>
<li><a href="#Introduction_nn13">SWIG and freedom</a></li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="Introduction_nn2">2.1 What is SWIG?</a></h2>
<p> SWIG is a software development tool that simplifies the task of
interfacing different languages to C and C++ programs. In a nutshell,
SWIG is a compiler that takes C/C++ declarations and creates the
wrappers needed to access those declarations from other languages
including Perl, Python, Tcl, Ruby, Guile, and Java. SWIG normally
requires no modifications to existing code and can often be used to
build a usable interface in only a few minutes. Possible applications
of SWIG include:</p>
<ul>
<li>Building interpreted interfaces to existing C programs.</li>
<li>Rapid prototyping and application development.</li>
<li>Interactive debugging.</li>
<li>Reengineering or refactoring of legacy software into scripting
language components.</li>
<li>Making a graphical user interface (using Tk for example).</li>
<li>Testing of C libraries and programs (using scripts).</li>
<li>Building high performance C modules for scripting languages.</li>
<li>Making C programming more enjoyable (or tolerable depending on your
point of view).</li>
<li>Impressing your friends.</li>
<li>Obtaining vast sums of research funding (although obviously not
applicable to the author).</li>
</ul>
<p> SWIG was originally designed to make it extremely easy for
scientists and engineers to build extensible scientific software
without having to get a degree in software engineering. Because of
this, the use of SWIG tends to be somewhat informal and ad-hoc (e.g.,
SWIG does not require users to provide formal interface specifications
as you would find in a dedicated IDL compiler). Although this style of
development isn't appropriate for every project, it is particularly
well suited to software development in the small; especially the
research and development work that is commonly found in scientific and
engineering projects. However, nowadays SWIG is known to be used in
many large open source and commercial projects.</p>
<h2><a name="Introduction_nn3">2.2 Why use SWIG?</a></h2>
<p> As stated in the previous section, the primary purpose of SWIG is to
simplify the task of integrating C/C++ with other programming
languages. However, why would anyone want to do that? To answer that
question, it is useful to list a few strengths of C/C++ programming:</p>
<ul>
<li>Excellent support for writing programming libraries.</li>
<li>High performance (number crunching, data processing, graphics,
etc.).</li>
<li>Systems programming and systems integration.</li>
<li>Large user community and software base.</li>
</ul>
<p> Next, let's list a few problems with C/C++ programming</p>
<ul>
<li>Writing a user interface is rather painful (i.e., consider
programming with MFC, X11, GTK, or any number of other libraries).</li>
<li>Testing is time consuming (the compile/debug cycle).</li>
<li>Not easy to reconfigure or customize without recompilation.</li>
<li>Modularization can be tricky.</li>
<li>Security concerns (buffer overflows for instance).</li>
</ul>
<p> To address these limitations, many programmers have arrived at the
conclusion that it is much easier to use different programming
languages for different tasks. For instance, writing a graphical user
interface may be significantly easier in a scripting language like
Python or Tcl (consider the reasons why millions of programmers have
used languages like Visual Basic if you need more proof). An
interactive interpreter might also serve as a useful debugging and
testing tool. Other languages like Java might greatly simplify the task
of writing distributed computing software. The key point is that
different programming languages offer different strengths and
weaknesses. Moreover, it is extremely unlikely that any programming is
ever going to be perfect. Therefore, by combining languages together,
you can utilize the best features of each language and greatly simplify
certain aspects of software development.</p>
<p> From the standpoint of C/C++, a lot of people use SWIG because they
want to break out of the traditional monolithic C programming model
which usually results in programs that resemble this:</p>
<ul>
<li>A collection of functions and variables that do something useful.</li>
<li>A <tt>main()</tt> program that starts everything.</li>
<li>A horrible collection of hacks that form some kind of user interface
(but which no-one really wants to touch).</li>
</ul>
<p> Instead of going down that route, incorporating C/C++ into a higher
level language often results in a more modular design, less code,
better flexibility, and increased programmer productivity.</p>
<p> SWIG tries to make the problem of C/C++ integration as painless as
possible. This allows you to focus on the underlying C program and
using the high-level language interface, but not the tedious and
complex chore of making the two languages talk to each other. At the
same time, SWIG recognizes that all applications are different.
Therefore, it provides a wide variety of customization features that
let you change almost every aspect of the language bindings. This is
the main reason why SWIG has such a large user manual ;-).</p>
<h2><a name="Introduction_target_languages">2.3 Target languages</a></h2>
<p> SWIG in essence is a tool to generate code for making C/C++ code
available to various other programming languages. These higher level
programming languages are the target languages for the SWIG code
generator and C or C++ are the input languages. A single target
language must be specified when SWIG is run. This results in generating
code for C/C++ and the specified target language to interface with each
other. SWIG can be invoked multiple times, but with a different target
language specified on each invocation. This ability to interface C/C++
to many different target languages is one of SWIG's core strengths and
features.</p>
<p> SWIG is very broadly composed of two components. A core component
creates a parse tree from the input ISO C/C++ and SWIG directives
(extensions to the C/C++ standards). The parse tree is then passed to a
second component, one of the target language modules for generating
code specific to a higher level language. SWIG supports many different
target languages. These target languages are given a status of either
Supported, Experimental or Deprecated. This status is provided to
indicate the level of maturity to expect when using a particular target
language as not all target languages are fully developed or being kept
up to date.</p>
<p> The second part of the SWIG documentation contains a chapter for
each target level language. The target language chapters are under one
of the sections indicating the status (Supported, Experimental or
Deprecated) for that language.</p>
<h3><a name="Introduction_supported_status">2.3.1 Supported status</a></h3>
<p> A target language is given the 'Supported' status when</p>
<ul>
<li>It is in a mature, well functioning state.</li>
<li>It has its own comprehensive chapter in the documentation.</li>
<li>It passes all of the main SWIG test-suite and has a range of working
examples.</li>
<li>It supports the vast majority of SWIG features.</li>
<li>It provides strong backwards compatibility between releases.</li>
</ul>
<p> The above is a short summary and further details are outlined in the
<a href="#Extending_supported_status">Supported status</a> section in
the Extending chapter. The good news is that all the well-known and
most popular languages have this status.</p>
<h3><a name="Introduction_experimental_status">2.3.2 Experimental status</a>
</h3>
<p> A target language is given the 'Experimental' status when</p>
<ul>
<li>It is of sub-standard quality, failing to meet the above 'Supported'
status.</li>
<li>It is somewhere between the mid to mature stage of development.</li>
<li>It does not guarantee any backwards compatibility between releases.</li>
<li>It is in need of help to finish development.</li>
</ul>
<p> Anyone using an experimental target language is strongly urged to
assist with development of the target language module if they wish to
use it.</p>
<p> SWIG displays a warning when an experimental target language is used
in order to set expectations and emphasize the experimental status of
the target language. The usual <a href="#Warnings_suppression">warning
suppression</a> techniques can be used if required.</p>
<p> The above is a short summary and further details are outlined in the
<a href="#Extending_experimental_status">Experimental status</a>
section in the Extending chapter.</p>
<h3><a name="Introduction_deprecated_status">2.3.3 Deprecated status</a></h3>
<p> A target language that was once 'Supported' or 'Experimental' is
changed to the 'Deprecated' status when it has been neglected over time
and become non-functional. It requires interested community member's
contributions to restore it to its former status.</p>
<p> Please see the <a href="#Extending_deprecated_status">Deprecated
status</a> section in the Extending chapter for more details.</p>
<p> SWIG displays a deprecated warning notice whenever a 'Deprecated'
language is used. The usual <a href="#Warnings_suppression">warning
suppression</a> techniques can be used if required.</p>
<h2><a name="Introduction_nn4">2.4 A SWIG example</a></h2>
<p> The best way to illustrate SWIG is with a simple example. Consider
the following C code:</p>
<div class="code">
<pre>
/* File : example.c */
double My_variable = 3.0;
/* Compute factorial of n */
int fact(int n) {
if (n &lt;= 1)
return 1;
else
return n*fact(n-1);
}
/* Compute n mod m */
int my_mod(int n, int m) {
return(n % m);
}
</pre>
</div>
<p> Suppose that you wanted to access these functions and the global
variable <tt>My_variable</tt> from Tcl. You start by making a SWIG
interface file as shown below (by convention, these files carry a .i
suffix) :</p>
<h3><a name="Introduction_nn5">2.4.1 SWIG interface file</a></h3>
<div class="code">
<pre>
/* File : example.i */
%module example
%{
/* Put headers and other declarations here */
extern double My_variable;
extern int fact(int);
extern int my_mod(int n, int m);
%}
extern double My_variable;
extern int fact(int);
extern int my_mod(int n, int m);
</pre>
</div>
<p> The interface file contains ISO C function prototypes and variable
declarations. The <tt>%module</tt> directive defines the name of the
module that will be created by SWIG. The <tt>%{ %}</tt> block provides
a location for inserting additional code, such as C header files or
additional C declarations, into the generated C wrapper code.</p>
<h3><a name="Introduction_nn6">2.4.2 The swig command</a></h3>
<p> SWIG is invoked using the <tt>swig</tt> command. We can use this to
build a Tcl module (under Linux) as follows :</p>
<div class="shell">
<pre>
$ <b>swig -tcl example.i</b>
$ <b>gcc -c -fPIC example.c example_wrap.c -I/usr/include/tcl8.7</b>
$ <b>gcc -shared example.o example_wrap.o -o example.so</b>
$ <b>tclsh</b>
% <b>load ./example.so</b>
% <b>fact 4</b>
24
% <b>my_mod 23 7</b>
2
% <b>expr $My_variable + 4.5</b>
7.5
%
</pre>
</div>
<p> The <tt>swig</tt> command produced a new file called <tt>
example_wrap.c</tt> that should be compiled along with the <tt>example.c</tt>
file. Most operating systems and scripting languages now support
dynamic loading of modules. In our example, our Tcl module has been
compiled into a shared library that can be loaded into Tcl. When
loaded, Tcl can now access the functions and variables declared in the
SWIG interface. A look at the file <tt>example_wrap.c</tt> reveals a
hideous mess. However, you almost never need to worry about it.</p>
<h3><a name="Introduction_nn7">2.4.3 Building a Perl5 module</a></h3>
<p> Now, let's turn these functions into a Perl5 module. Without making
any changes type the following (shown for Solaris):</p>
<div class="shell">
<pre>
unix &gt; <b>swig -perl5 example.i</b>
unix &gt; <b>gcc -c example.c example_wrap.c \
-I/usr/local/lib/perl5/sun4-solaris/5.003/CORE</b>
unix &gt; <b>ld -G example.o example_wrap.o -o example.so</b> # This is for Solaris
unix &gt; <b>perl5.003
use example;
print example::fact(4), &quot;\n&quot;;
print example::my_mod(23, 7), &quot;\n&quot;;
print $example::My_variable + 4.5, &quot;\n&quot;;
&lt;ctrl-d&gt;</b>
24
2
7.5
unix &gt;
</pre>
</div>
<h3><a name="Introduction_nn8">2.4.4 Building a Python module</a></h3>
<p> Finally, let's build a module for Python (shown for Linux).</p>
<div class="shell">
<pre>
$ <b>swig -python example.i</b>
$ <b>gcc -c -fPIC example.c example_wrap.c -I/usr/include/python3.12</b>
$ <b>gcc -shared example.o example_wrap.o -o _example.so</b>
$ <b>python3</b>
Python 3.12.4 (main, Jun 12 2024, 19:06:53) [GCC 13.2.0] on linux
Type &quot;help&quot;, &quot;copyright&quot;, &quot;credits&quot; or &quot;license&quot; for more information.
&gt;&gt;&gt; <b>import example</b>
&gt;&gt;&gt; <b>example.fact(4)</b>
24
&gt;&gt;&gt; <b>example.my_mod(23, 7)</b>
2
&gt;&gt;&gt; <b>example.cvar.My_variable + 4.5</b>
7.5
</pre>
</div>
<h3><a name="Introduction_nn9">2.4.5 Shortcuts</a></h3>
<p> To the truly lazy programmer, one may wonder why we needed the extra
interface file at all. As it turns out, you can often do without it.
For example, you could also build a Perl5 module by just running SWIG
on the C header file and specifying a module name as follows</p>
<div class="shell">
<pre>
unix &gt; <b>swig -perl5 -module example example.h</b>
unix &gt; <b>gcc -c example.c example_wrap.c \
-I/usr/local/lib/perl5/sun4-solaris/5.003/CORE</b>
unix &gt; <b>ld -G example.o example_wrap.o -o example.so</b>
unix &gt; <b>perl5.003
use example;
print example::fact(4), &quot;\n&quot;;
print example::my_mod(23, 7), &quot;\n&quot;;
print $example::My_variable + 4.5, &quot;\n&quot;;
&lt;ctrl-d&gt;</b>
24
2
7.5
</pre>
</div>
<h2><a name="Introduction_nn10">2.5 Supported C/C++ language features</a>
</h2>
<p> A primary goal of the SWIG project is to make the language binding
process extremely easy. Although a few simple examples have been shown,
SWIG is quite capable in supporting most of C++. Some of the major
features include:</p>
<ul>
<li>Full C99 preprocessing.</li>
<li>All ISO C and C++ datatypes.</li>
<li>Functions, variables, and constants.</li>
<li>Classes.</li>
<li>Single and multiple inheritance.</li>
<li>Overloaded functions and methods.</li>
<li>Overloaded operators.</li>
<li>C++ templates (including member templates, specialization, and
partial specialization).</li>
<li>Namespaces.</li>
<li>Variable length arguments.</li>
<li>C++ smart pointers.</li>
</ul>
<p> Most of C++11 is also supported. Details are in the <a href="#CPlusPlus11">
C++11</a> chapter. C++14 support is covered in the <a href="#CPlusPlus14">
C++14</a> chapter. C++17 support is covered in the <a href="#CPlusPlus17">
C++17</a> chapter. C++20 support is covered in the <a href="#CPlusPlus20">
C++20</a> chapter.</p>
<p> It is important to stress that SWIG is not a simplistic C++ lexing
tool like several apparently similar wrapper generation tools. SWIG not
only parses C++, it implements the full C++ type system and it is able
to understand C++ semantics. SWIG generates its wrappers with full
knowledge of this information. As a result, you will find SWIG to be
just as capable of dealing with nasty corner cases as it is in wrapping
simple C++ code. In fact, SWIG is able to handle C++ code that stresses
the very limits of many C++ compilers.</p>
<h2><a name="Introduction_nn11">2.6 Non-intrusive interface building</a></h2>
<p> When used as intended, SWIG requires minimal (if any) modification
to existing C or C++ code. This makes SWIG extremely easy to use with
existing packages and promotes software reuse and modularity. By making
the C/C++ code independent of the high level interface, you can change
the interface and reuse the code in other applications. It is also
possible to support different types of interfaces depending on the
application.</p>
<h2><a name="Introduction_build_system">2.7 Incorporating SWIG into a
build system</a></h2>
<p> SWIG is a command line tool and as such can be incorporated into any
build system that supports invoking external tools/compilers. SWIG is
most commonly invoked from within a Makefile, but is also known to be
invoked from popular IDEs such as Microsoft Visual Studio.</p>
<p> If you are using the GNU Autotools (<a href="https://www.gnu.org/software/autoconf/">
Autoconf</a>/ <a href="https://www.gnu.org/software/automake/">Automake</a>
/ <a href="https://www.gnu.org/software/libtool/">Libtool</a>) to
configure SWIG use in your project, the SWIG Autoconf macros can be
used. The primary macro is <tt>ax_pkg_swig</tt>, see <a href="https://www.gnu.org/software/autoconf-archive/ax_pkg_swig.html#ax_pkg_swig">
http://www.gnu.org/software/autoconf-archive/ax_pkg_swig.html#ax_pkg_swig
</a>. The <tt>ax_python_devel</tt> macro is also helpful for generating
Python extensions. See the <a href="https://www.gnu.org/software/autoconf-archive/">
Autoconf Archive</a> for further information on this and other Autoconf
macros.</p>
<p> There is growing support for SWIG in some build tools, for example <a
href="https://cmake.org">CMake</a> is a cross-platform, open-source
build manager with built in support for SWIG. CMake can detect the SWIG
executable and many of the target language libraries for linking
against. CMake knows how to build shared libraries and loadable modules
on many different operating systems. This allows easy cross platform
SWIG development. It can also generate the custom commands necessary
for driving SWIG from IDEs and makefiles. All of this can be done from
a single cross platform input file. The following example is a CMake
input file for creating a Python wrapper for the SWIG interface file,
example.i:</p>
<div class="code">
<pre>
# This is a CMake example for Python
FIND_PACKAGE(SWIG REQUIRED)
INCLUDE(${SWIG_USE_FILE})
FIND_PACKAGE(PythonLibs)
INCLUDE_DIRECTORIES(${PYTHON_INCLUDE_PATH})
INCLUDE_DIRECTORIES(${CMAKE_CURRENT_SOURCE_DIR})
SET(CMAKE_SWIG_FLAGS &quot;&quot;)
SET_SOURCE_FILES_PROPERTIES(example.i PROPERTIES CPLUSPLUS ON)
SET_SOURCE_FILES_PROPERTIES(example.i PROPERTIES SWIG_FLAGS &quot;-includeall&quot;)
SWIG_ADD_MODULE(example python example.i example.cxx)
SWIG_LINK_LIBRARIES(example ${PYTHON_LIBRARIES})
</pre>
</div>
<p> The above example will generate native build files such as
makefiles, nmake files and Visual Studio projects which will invoke
SWIG and compile the generated C++ files into _example.so (UNIX) or
_example.pyd (Windows). For other target languages on Windows a dll,
instead of a .pyd file, is usually generated.</p>
<h2><a name="Introduction_nn12">2.8 Hands off code generation</a></h2>
<p> SWIG is designed to produce working code that needs no
hand-modification (in fact, if you look at the output, you probably
won't want to modify it). You should think of your target language
interface being defined entirely by the input to SWIG, not the
resulting output file. While this approach may limit flexibility for
hard-core hackers, it allows others to forget about the low-level
implementation details.</p>
<h2><a name="Introduction_nn13">2.9 SWIG and freedom</a></h2>
<p> No, this isn't a special section on the sorry state of world
politics. However, it may be useful to know that SWIG was written with
a certain &quot;philosophy&quot; about programming---namely that programmers are
smart and that tools should just stay out of their way. Because of
that, you will find that SWIG is extremely permissive in what it lets
you get away with. In fact, you can use SWIG to go well beyond
&quot;shooting yourself in the foot&quot; if dangerous programming is your goal.
On the other hand, this kind of freedom may be exactly what is needed
to work with complicated and unusual C/C++ applications.</p>
<p> Ironically, the freedom that SWIG provides is countered by an
extremely conservative approach to code generation. At its core, SWIG
tries to distill even the most advanced C++ code down to a small
well-defined set of interface building techniques based on ISO C
programming. Because of this, you will find that SWIG interfaces can be
easily compiled by virtually every C/C++ compiler and that they can be
used on any platform. Again, this is an important part of staying out
of the programmer's way----the last thing any developer wants to do is
to spend their time debugging the output of a tool that relies on
non-portable or unreliable programming features. Dependencies are often
a source of incompatibilities and problems and so additional third
party libraries are not used in the generated code. SWIG will also
generally avoid generating code that introduces a dependency on the C++
Standard Template Library (STL). SWIG will generate code that depends
on the C libraries though.</p>
<HR NOSHADE>
<h1><a name="Windows">3 Getting started on Windows</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Windows_installation">Installation on Windows</a>
<ul>
<li><a href="#Windows_executable">Windows Executable</a></li>
</ul>
</li>
<li><a href="#Windows_examples">SWIG Windows Examples</a>
<ul>
<li><a href="#Windows_visual_studio">Instructions for using the Examples
with Visual Studio</a>
<ul>
<li><a href="#Windows_csharp">C#</a></li>
<li><a href="#Windows_java">Java</a></li>
<li><a href="#Windows_python">Python</a></li>
<li><a href="#Windows_tcl">TCL</a></li>
</ul>
</li>
<li><a href="#Windows_other_compilers">Instructions for using the
Examples with other compilers</a></li>
</ul>
</li>
<li><a href="#Windows_swig_exe">Building swig.exe on Windows</a>
<ul>
<li><a href="#Windows_cmake">Building swig.exe using CMake</a></li>
<li><a href="#Windows_msys2">Building swig.exe using MSYS2 and MinGW-w64</a>
</li>
<li><a href="#Windows_mingw_msys">Building swig.exe using MinGW and MSYS</a>
</li>
<li><a href="#Windows_cygwin">Building swig.exe using Cygwin</a>
<ul>
<li><a href="#Windows_examples_cygwin">Running the examples on Windows
using Cygwin</a></li>
</ul>
</li>
</ul>
</li>
<li><a href="#Windows_interface_file">Microsoft extensions and other
Windows quirks</a>
<ul>
<li><a href="#Windows_msvc_cpp_standards">Visual C++ standards
compliance</a></li>
<li><a href="#Windows_calling_conventions">Calling conventions</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> This chapter describes SWIG usage on Microsoft Windows. Installing
SWIG and running the examples is covered as well as building the SWIG
executable. Usage within the Unix like environments MinGW and Cygwin is
also detailed.</p>
<h2><a name="Windows_installation">3.1 Installation on Windows</a></h2>
<p> SWIG does not come with the usual Windows type installation program,
however it is quite easy to get started. The main steps are:</p>
<ul>
<li>Download the swigwin zip package from the <a href="https://www.swig.org">
SWIG website</a> and unzip into a directory. This is all that needs
downloading for the Windows platform.</li>
<li>Set environment variables as described in the <a href="#Windows_examples">
SWIG Windows Examples</a> section in order to run examples using Visual
C++.</li>
</ul>
<h3><a name="Windows_executable">3.1.1 Windows Executable</a></h3>
<p> The swigwin distribution contains the SWIG Windows 64-bit
executable, swig.exe, which will only run on 64-bit versions of
Windows. If you want to build your own swig.exe have a look at <a href="#Windows_swig_exe">
Building swig.exe on Windows</a>.</p>
<h2><a name="Windows_examples">3.2 SWIG Windows Examples</a></h2>
<p> Microsoft Visual C++ is used for compiling and linking SWIG's output
on Windows for some languages. However, MinGW and gcc is often the only
supported toolchain for a number of target languages. The Examples
directory has a few Visual C++ project files (.vcxproj files) where the
target languages are known to work with Visual C++. These were produced
by Visual Studio 2019. Newer versions of Visual Studio, such as Visual
Studio 2022, are able to open and convert these project files if
necessary. Each C# example comes with a Visual Studio 2019 solution in
addition to the .vcxproj file the C# (.csproj) project file. The
project files have been set up to execute SWIG in a custom build rule
for the SWIG interface (.i) file. Alternatively run the <a href="#Windows_examples_cygwin">
examples using Cygwin</a>.</p>
<p> More information on each of the examples is available with the
examples distributed with SWIG (Examples/index.html).</p>
<h3><a name="Windows_visual_studio">3.2.1 Instructions for using the
Examples with Visual Studio</a></h3>
<p> Ensure the SWIG executable is as supplied in the SWIG root directory
in order for the examples to work. Most languages require some
environment variables to be set<b> before</b> running Visual C++. Note
that Visual Studio must be re-started to pick up any changes in
environment variables. Open up an example .vcxproj file or .sln file,
Visual Studio will prompt you to upgrade the project if necessary.
Ensure the Release build is selected then do a Rebuild Solution from
the Build menu. The required environment variables are displayed with
their current values during the build.</p>
<p> The list of required environment variables for each module language
is also listed below. They are usually set from the Control Panel and
System properties, but this depends on which flavour of Windows you are
running. If you don't want to use environment variables then change all
occurrences of the environment variables in the .dsp files with hard
coded values. If you are interested in how the project files are set up
there is explanatory information in some of the language module's
documentation.</p>
<h4><a name="Windows_csharp">3.2.1.1 C#</a></h4>
<p> The C# examples do not require any environment variables to be set
as a C# project file is included. Just open up the .sln solution file
in Visual Studio 2019 or later, select Release Build, and do a Rebuild
Solution from the Build menu. The accompanying C# and C++ project files
are automatically used by the solution file.</p>
<h4><a name="Windows_java">3.2.1.2 Java</a></h4>
<p><b> <tt>JAVA_INCLUDE</tt></b> : Set this to the directory containing
jni.h
<br><b> <tt>JAVA_BIN</tt></b> : Set this to the bin directory containing
javac.exe</p>
<p> Example using the openjdk package installed in a Conda environment:
<br> <tt>JAVA_INCLUDE: C:\miniconda3\envs\java\Library\lib\jvm\include
<br> JAVA_BIN: C:\miniconda3\envs\java\Library\lib\jvm\bin
<br></tt></p>
<h4><a name="Windows_python">3.2.1.3 Python</a></h4>
<p><b> <tt>PYTHON_INCLUDE</tt></b> : Set this to the directory that
contains Python.h
<br><b> <tt>PYTHON_LIB</tt></b> : Set this to the Python library
including path for linking</p>
<p> Example using Python 3.13 installed in a Conda environment:
<br> <tt>PYTHON_INCLUDE: C:\miniconda3\envs\python\include
<br> PYTHON_LIB: C:\miniconda3\envs\python\libs\python313.lib
<br></tt></p>
<h4><a name="Windows_tcl">3.2.1.4 TCL</a></h4>
<p><b> <tt>TCL_INCLUDE</tt></b> : Set this to the directory containing
tcl.h
<br><b> <tt>TCL_LIB</tt></b> : Set this to the TCL library including
path for linking</p>
<p> Example using ActiveTcl 8.6
<br> <tt>TCL_INCLUDE: C:\ActiveTcl\include
<br> TCL_LIB: C:\ActiveTcl\lib\tcl86t.lib
<br></tt></p>
<h3><a name="Windows_other_compilers">3.2.2 Instructions for using the
Examples with other compilers</a></h3>
<p> If you do not have access to Visual C++ you will have to set up
project files / Makefiles for your chosen compiler. There is a section
in each of the language modules detailing what needs setting up using
Visual C++ which may be of some guidance. Alternatively you may want to
use Cygwin as described in the following section.</p>
<h2><a name="Windows_swig_exe">3.3 Building swig.exe on Windows</a></h2>
<p> The SWIG distribution provides a pre-built swig.exe and so it is not
necessary for users to build the SWIG executable. However, this section
is provided for those that want to modify the SWIG source code in a
Windows environment. Normally this is not needed, so most people will
want to ignore this section.</p>
<p> There are various ways to build the SWIG executable including <a href="https://cmake.org/">
CMake</a> which is able to generate project files for building with <a href="https://visualstudio.microsoft.com/">
Visual Studio</a> <a href="https://docs.microsoft.com/cpp/">MSVC</a>.
<br> SWIG can also be compiled and run using <a href="https://www.msys2.org/">
MSYS2</a> with <a href="https://www.mingw-w64.org/">MinGW-w64</a>, <a href="https://www.cygwin.com">
Cygwin</a> or <a href="https://osdn.net/projects/mingw/">MinGW</a>, all
of which provide a Unix like front end to Windows and comes free with
the <a href="https://gcc.gnu.org/">GCC</a> C/C++ compiler.
<br> SWIG can also be compiled with MSYS2 using MSVC and SWIG <a href="https://github.com/swig/cccl">
MSVC wrapper</a>.
<br></p>
<h3><a name="Windows_cmake">3.3.1 Building swig.exe using CMake</a></h3>
<p> SWIG can be built using <a href="https://cmake.org/">CMake</a> and
Visual Studio rather than autotools. As with the other approaches to
building SWIG the dependencies need to be installed. The steps below
are one of a number of ways of installing the dependencies without
requiring Cygwin or MinGW. For fully working build steps always check
the Continuous Integration (CI) setups currently detailed in the <a href="https://github.com/swig/swig/tree/master/.github/workflows/nuget.yml">
GitHub Actions YAML file</a>.</p>
<ol>
<li> Install Nuget from <a href="https://www.nuget.org/downloads">
https://www.nuget.org/downloads</a> (v6.0.0 is used in this example, and
installed to <tt>C:\Tools</tt>). Nuget is the package manager for .NET,
but allows us to easily install <a href="https://cmake.org/">CMake</a>
and other dependencies required by SWIG.</li>
<li> Install <a href="https://www.nuget.org/packages/CMake-win64/">
CMake-win64 Nuget package</a> using the following command:
<pre>C:\Tools\nuget install CMake-win64 -Version 3.15.5 -OutputDirectory C:\Tools\CMake</pre>
Using PowerShell the equivalent syntax is:
<pre>&amp; &quot;C:\Tools\nuget&quot; install CMake-win64 -Version 3.15.5 -OutputDirectory C:\Tools\CMake</pre>
Alternatively you can download CMake from <a href="https://cmake.org/download/">
https://cmake.org/download/</a>.</li>
<li> Install the <a href="https://www.nuget.org/packages/bison/">Bison
Nuget package</a> using the following command:
<pre>C:\Tools\nuget install Bison -Version 3.7.4 -OutputDirectory C:\Tools\bison</pre>
Alternatively download Bison from <a href="https://sourceforge.net/projects/winflexbison/files/">
https://sourceforge.net/projects/winflexbison/files/</a> (Bison 3.7.4 is
used in this example) and save to a folder e.g. <tt>C:\Tools\Bison</tt></li>
<li> Install the <a href="https://www.nuget.org/packages/pcre2/">PCRE2
Nuget package</a> using the following command:
<pre>C:\Tools\nuget install PCRE2 -Version 10.39 -OutputDirectory C:\Tools\pcre2</pre>
Note this is a x64 build, if this is not suitable PCRE2 can be built
from source using <a href="https://github.com/PhilipHazel/pcre2/">
https://github.com/PhilipHazel/pcre2/</a>. Alternatively, set <tt>
WITH_PCRE=OFF</tt> to disable PCRE2 support if you are sure you do not
require it.</li>
<li> We will also need the SWIG source code. Either download a zipped
archive from GitHub, or if git is installed clone the latest codebase
using:
<pre>git clone https://github.com/swig/swig.git</pre>
In this example we are assuming the source code is available at <tt>
C:\swig</tt></li>
<li>
<p> Now we have all the required dependencies we can build SWIG using
PowerShell and the commands below. We are assuming Visual Studio 2019
is installed. For other versions of Visual Studio change <tt>&quot;Visual
Studio 16 2019 -A x64&quot;</tt> to the relevant <a href="https://cmake.org/cmake/help/latest/manual/cmake-generators.7.html#visual-studio-generators">
Visual Studio Generator</a> and architecture. We add the required build
tools to the system PATH, and then build a Release version of SWIG. If
all runs successfully a new swig.exe should be generated in the <tt>
C:/swig/install2/bin</tt> folder.</p>
</li>
</ol>
<div class="shell">
<pre>
cd C:\swig
$env:PATH=&quot;C:\Tools\CMake\CMake-win64.3.15.5\bin;C:\Tools\bison\Bison.3.7.4\bin;&quot; + $env:PATH
$PCRE_ROOT=&quot;C:\Tools\pcre2\PCRE2.10.39.0&quot;
cmake -G &quot;Visual Studio 16 2019&quot; -A &quot;x64&quot; `
-DCMAKE_INSTALL_PREFIX=&quot;C:/swig/install2&quot; `
-DCMAKE_C_FLAGS=&quot;/DPCRE2_STATIC&quot; `
-DCMAKE_CXX_FLAGS=&quot;/DPCRE2_STATIC&quot; `
-DPCRE2_INCLUDE_DIR=&quot;$PCRE_ROOT/include&quot; `
-DPCRE2_LIBRARY=&quot;$PCRE_ROOT/lib/pcre2-8-static.lib&quot; `
-S . -B build
cmake --build build --config Release
cmake --install build --config Release
# to test the exe built correctly
cd install2/bin
./swig.exe -version
./swig.exe -help
</pre>
</div>
<p> In addition to Release builds you can create a Debug build using:</p>
<div class="shell">
<pre>cmake --build build --config Debug</pre>
</div>
<p> A Visual Studio solution file should be generated named swig.sln.
This can be opened and debugged by running the swig project and setting
<tt>Properties &gt; Debugging &gt; Command Arguments</tt>. For example to
debug one of the test-suite .i files included with the SWIG source use
the following:</p>
<div class="shell">
<pre>-python -c++ -o C:\Temp\doxygen_parsing.cpp C:\swig\Examples\test-suite\doxygen_parsing.i</pre>
</div>
<h3><a name="Windows_msys2">3.3.2 Building swig.exe using MSYS2 and
MinGW-w64</a></h3>
<p> Download and install MSYS2 from <a href="https://www.msys2.org/">
www.msys2.org</a> (tested with version msys2-x86_64-20201109). Launch
the MSYS2 shell.</p>
<p> Install the packages needed to build swig:
<br></p>
<div class="shell">
<pre>
pacman -S git autoconf automake bison gcc make pcre2-devel
</pre>
</div>
<p> Clone the repository to /usr/src/:</p>
<div class="shell">
<pre>
mkdir /usr/src/
cd /usr/src/
git clone https://github.com/swig/swig.git
</pre>
</div>
<p> Configure and build:</p>
<div class="shell">
<pre>
cd /usr/src/swig
./autogen.sh
./configure
make
</pre>
</div>
<p> Finally you may also want to install SWIG:</p>
<div class="shell">
<pre>
make install
</pre>
</div>
<h3><a name="Windows_mingw_msys">3.3.3 Building swig.exe using MinGW and
MSYS</a></h3>
<p>Warning: These instructions were added in 2006 and have barely
changed since so are unlikely to work exactly as written.</p>
<p> The short abbreviated instructions follow...</p>
<ul>
<li>Install MinGW and MSYS from the <a href="https://osdn.net/projects/mingw/">
MinGW</a> site. This provides a Unix environment on Windows.</li>
<li>Follow the usual Unix instructions in the README file in the SWIG
root directory to build swig.exe from the MinGW command prompt.</li>
</ul>
<p> The step by step instructions to download and install MinGW and
MSYS, then download and build the latest version of SWIG from Github
follow... Note that the instructions for obtaining SWIG from Github are
also online at <a href="https://www.swig.org/svn.html">SWIG Bleeding
Edge</a>.</p>
<p><b> Pitfall note:</b> Execute the steps in the order shown and don't
use spaces in path names. In fact it is best to use the default
installation directories.</p>
<ol>
<li> Download the following packages from the <a href="https://osdn.net/projects/mingw/releases/">
MinGW download page</a>. Note that at the time of writing, the majority
of these are in the Current release list and some are in the Snapshot
or Previous release list.
<ul>
<li>MinGW-3.1.0-1.exe</li>
<li>MSYS-1.0.11-2004.04.30-1.exe</li>
<li>msysDTK-1.0.1.exe</li>
<li>bison-2.0-MSYS.tar.gz</li>
<li>msys-autoconf-2.59.tar.bz2</li>
<li>msys-automake-1.8.2.tar.bz2</li>
</ul>
</li>
<li> Install MinGW-3.1.0-1.exe (C:\MinGW is default location.)</li>
<li> Install MSYS-1.0.11-2004.04.30-1.exe. Make sure you install it on
the same windows drive letter as MinGW (C:\msys\1.0 is default). In the
post install script,
<ul>
<li>Answer y to the &quot;do you wish to continue with the post install?&quot;</li>
<li>Answer y to the &quot;do you have MinGW installed?&quot;</li>
<li>Type in the folder in which you installed MinGW (C:/MinGW is
default)</li>
</ul>
</li>
<li> Install msysDTK-1.0.1.exe to the same folder that you installed
MSYS (C:\msys\1.0 is default).</li>
<li> Copy the following to the MSYS install folder (C:\msys\1.0 is
default):
<ul>
<li>msys-automake-1.8.2.tar.bz2</li>
<li>msys-autoconf-2.59.tar.bz2</li>
<li>bison-2.0-MSYS.tar.gz</li>
</ul>
</li>
<li> Start the MSYS command prompt and execute:<div class="shell">
<pre>
cd /
tar -jxf msys-automake-1.8.2.tar.bz2
tar -jxf msys-autoconf-2.59.tar.bz2
tar -zxf bison-2.0-MSYS.tar.gz
</pre>
</div></li>
<li> The very latest development version of SWIG is available from <a href="https://github.com/swig/swig">
SWIG on Github</a> and can be downloaded as a zip file or if you have
Git installed, via Git. Either download the latest <a href="https://github.com/swig/swig/archive/master.zip">
Zip file</a> snapshot and unzip and rename the top level folder to
/usr/src/swig. Otherwise if using Git, type in the following:<div class="shell">
<pre>
mkdir /usr/src
cd /usr/src
git clone https://github.com/swig/swig.git
</pre>
</div><b> Pitfall note:</b> If you want to place SWIG in a different
folder to the proposed /usr/src/swig, do not use MSYS emulated windows
drive letters, because the autotools will fail miserably on those.</li>
<li> The PCRE2 third party library needs to be built next. Download the
latest PCRE2 source tarball, such as <tt>pcre2-10.39.tar.bz2</tt>, from
<a href="https://www.pcre.org">www.pcre.org</a> and place in the <tt>
/usr/src/swig</tt> directory. Build PCRE2 as a static library using the
Tools/pcre-build.sh script as follows:<div class="shell">
<pre>
cd /usr/src/swig
Tools/pcre-build.sh
</pre>
</div></li>
<li> You are now ready to build SWIG. Execute the following commands to
build swig.exe:<div class="shell">
<pre>
cd /usr/src/swig
./autogen.sh
./configure
make
</pre>
</div></li>
</ol>
<h3><a name="Windows_cygwin">3.3.4 Building swig.exe using Cygwin</a></h3>
<p> Note that SWIG can also be built using Cygwin. However, SWIG will
then require the Cygwin DLL when executing. Follow the Unix
instructions in the README file in the SWIG root directory. Note that
the Cygwin environment will also allow one to regenerate the autotool
generated files which are supplied with the release distribution. These
files are generated using the <tt>autogen.sh</tt> script and will only
need regenerating in circumstances such as changing the build system.</p>
<h4><a name="Windows_examples_cygwin">3.3.4.1 Running the examples on
Windows using Cygwin</a></h4>
<p> The examples and test-suite work as successfully on Cygwin as on any
other Unix operating system. The modules which are known to work are
Python, Tcl, Perl, Ruby, Java and C#. Follow the Unix instructions in
the README file in the SWIG root directory to build the examples.</p>
<h2><a name="Windows_interface_file">3.4 Microsoft extensions and other
Windows quirks</a></h2>
<h3><a name="Windows_msvc_cpp_standards">3.4.1 Visual C++ standards
compliance</a></h3>
<p> The Visual C++ compiler (MSVC) has a long history of not being
standards compliant, but this has been getting better over the years.</p>
<p> SWIG's <a href="#SWIGPlus_nn3">approach for C++</a> is to generate
standards compliant C++98 code along with enhancements for later
standards if the C++ compiler supports a later standard. Unfortunately
by default, in 2024, Visual C++ still does not set the <tt>__cplusplus</tt>
macro correctly to pick up these later C++ standard features. MSVC
users are urged to ensure this macro is defined correctly by consulting
the latest Microsoft Visual C++ documentation, in particular, <a href="https://learn.microsoft.com/en-us/cpp/build/reference/std-specify-language-standard-version">
/std</a> and <a href="https://learn.microsoft.com/en-us/cpp/build/reference/zc-cplusplus">
/Zc:__cplusplus</a>.</p>
<h3><a name="Windows_calling_conventions">3.4.2 Calling conventions</a></h3>
<p> A common problem when using SWIG on Windows are the Microsoft
function calling conventions which are not in the C++ standard. SWIG
parses ISO C/C++ so cannot deal with proprietary conventions such as <tt>
__declspec(dllimport)</tt>, <tt>__stdcall</tt> etc. There is a Windows
interface file, <tt>windows.i</tt>, to deal with these calling
conventions though. The file also contains typemaps for handling
commonly used Windows specific types such as <tt>__int64</tt>, <tt>BOOL</tt>
, <tt>DWORD</tt> etc. Include it like you would any other interface
file, for example:</p>
<div class="code">
<pre>
%include &lt;windows.i&gt;
__declspec(dllexport) ULONG __stdcall foo(DWORD, __int32);
</pre>
</div>
<p>Note that if you follow Microsoft's recommendation of wrapping the <tt>
__declspec</tt> calls in a preprocessor definition, you will need to
make sure that the definition is included by SWIG as well, by either
defining it manually or via a header. For example, if you have
specified the preprocessor definition in a header named <tt>
export_lib.h</tt> and include other headers which depend on it, you
should use the <tt>%include</tt> directive to include the definition
explicitly. For example, if you had a header file, <tt>bar.h</tt>,
which depended on <tt>export_lib.h</tt>, your SWIG definition file
might look like:</p>
<div class="code">
<pre>
// bar.i
%module bar
%include &lt;windows.i&gt;
%include &quot;export_lib.h&quot;
%include &quot;bar.h&quot;
</pre>
</div>
<p> where export_lib.h may contain:</p>
<div class="code">
<pre>
// export_lib.h
#define BAR_API __declspec(dllexport)
</pre>
</div>
<p> and bar.h may look like:</p>
<div class="code">
<pre>
// bar.h
#include &quot;export_lib.h&quot;
BAR_API void bar_function(int, double);
</pre>
</div>
<p> Using the preprocessor to remove BAR_API is a popular simpler
solution:</p>
<div class="code">
<pre>
// bar.i
%module bar
#define BAR_API
%include &quot;bar.h&quot;
</pre>
</div><HR NOSHADE>
<h1><a name="Scripting">4 Scripting Languages</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Scripting_nn2">The two language view of the world</a></li>
<li><a href="#Scripting_nn3">How does a scripting language talk to C?</a>
<ul>
<li><a href="#Scripting_nn4">Wrapper functions</a></li>
<li><a href="#Scripting_nn5">Variable linking</a></li>
<li><a href="#Scripting_nn6">Constants</a></li>
<li><a href="#Scripting_nn7">Structures and classes</a></li>
<li><a href="#Scripting_nn8">Proxy classes</a></li>
</ul>
</li>
<li><a href="#Scripting_nn9">Building scripting language extensions</a>
<ul>
<li><a href="#Scripting_nn10">Shared libraries and dynamic loading</a></li>
<li><a href="#Scripting_nn11">Linking with shared libraries</a></li>
<li><a href="#Scripting_nn12">Static linking</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> This chapter provides a brief overview of scripting language
extension programming and the mechanisms by which scripting language
interpreters access C and C++ code.</p>
<h2><a name="Scripting_nn2">4.1 The two language view of the world</a></h2>
<p> When a scripting language is used to control a C program, the
resulting system tends to look as follows:</p>
<center><img alt="Scripting language input - C/C++ functions output" HEIGHT="149"
src="ch2.1.png" WIDTH="377"></center>
<p> In this programming model, the scripting language interpreter is
used for high level control whereas the underlying functionality of the
C/C++ program is accessed through special scripting language
&quot;commands.&quot; If you have ever tried to write your own simple command
interpreter, you might view the scripting language approach to be a
highly advanced implementation of that. Likewise, If you have ever used
a package such as MATLAB or IDL, it is a very similar model--the
interpreter executes user commands and scripts. However, most of the
underlying functionality is written in a low-level language like C or
Fortran.</p>
<p> The two-language model of computing is extremely powerful because it
exploits the strengths of each language. C/C++ can be used for maximal
performance and complicated systems programming tasks. Scripting
languages can be used for rapid prototyping, interactive debugging,
scripting, and access to high-level data structures such associative
arrays.</p>
<h2><a name="Scripting_nn3">4.2 How does a scripting language talk to C?</a>
</h2>
<p> Scripting languages are built around a parser that knows how to
execute commands and scripts. Within this parser, there is a mechanism
for executing commands and accessing variables. Normally, this is used
to implement the builtin features of the language. However, by
extending the interpreter, it is usually possible to add new commands
and variables. To do this, most languages define a special API for
adding new commands. Furthermore, a special foreign function interface
defines how these new commands are supposed to hook into the
interpreter.</p>
<p> Typically, when you add a new command to a scripting interpreter you
need to do two things; first you need to write a special &quot;wrapper&quot;
function that serves as the glue between the interpreter and the
underlying C function. Then you need to give the interpreter
information about the wrapper by providing details about the name of
the function, arguments, and so forth. The next few sections illustrate
the process.</p>
<h3><a name="Scripting_nn4">4.2.1 Wrapper functions</a></h3>
<p> Suppose you have an ordinary C function like this :</p>
<div class="code">
<pre>
int fact(int n) {
if (n &lt;= 1)
return 1;
else
return n*fact(n-1);
}
</pre>
</div>
<p> In order to access this function from a scripting language, it is
necessary to write a special &quot;wrapper&quot; function that serves as the glue
between the scripting language and the underlying C function. A wrapper
function must do three things :</p>
<ul>
<li>Gather function arguments and make sure they are valid.</li>
<li>Call the C function.</li>
<li>Convert the return value into a form recognized by the scripting
language.</li>
</ul>
<p> As an example, the Tcl wrapper function for the <tt>fact()</tt>
function above example might look like the following :</p>
<div class="code">
<pre>
int wrap_fact(ClientData clientData, Tcl_Interp *interp, int argc, char *argv[]) {
int result;
int arg0;
if (argc != 2) {
interp-&gt;result = &quot;wrong # args&quot;;
return TCL_ERROR;
}
arg0 = atoi(argv[1]);
result = fact(arg0);
sprintf(interp-&gt;result, &quot;%d&quot;, result);
return TCL_OK;
}
</pre>
</div>
<p> Once you have created a wrapper function, the final step is to tell
the scripting language about the new function. This is usually done in
an initialization function called by the language when the module is
loaded. For example, adding the above function to the Tcl interpreter
requires code like the following :</p>
<div class="code">
<pre>
int Wrap_Init(Tcl_Interp *interp) {
Tcl_CreateCommand(interp, &quot;fact&quot;, wrap_fact, (ClientData) NULL,
(Tcl_CmdDeleteProc *) NULL);
return TCL_OK;
}
</pre>
</div>
<p> When executed, Tcl will now have a new command called &quot;<tt>fact</tt>
&quot; that you can use like any other Tcl command.</p>
<p> Although the process of adding a new function to Tcl has been
illustrated, the procedure is almost identical for Perl and Python.
Both require special wrappers to be written and both need additional
initialization code. Only the specific details are different.</p>
<h3><a name="Scripting_nn5">4.2.2 Variable linking</a></h3>
<p> Variable linking refers to the problem of mapping a C/C++ global
variable to a variable in the scripting language interpreter. For
example, suppose you had the following variable:</p>
<div class="code">
<pre>
double Foo = 3.5;
</pre>
</div>
<p> It might be nice to access it from a script as follows (shown for
Perl):</p>
<div class="targetlang">
<pre>
$a = $Foo * 2.3; # Evaluation
$Foo = $a + 2.0; # Assignment
</pre>
</div>
<p> To provide such access, variables are commonly manipulated using a
pair of get/set functions. For example, whenever the value of a
variable is read, a &quot;get&quot; function is invoked. Similarly, whenever the
value of a variable is changed, a &quot;set&quot; function is called.</p>
<p> In many languages, calls to the get/set functions can be attached to
evaluation and assignment operators. Therefore, evaluating a variable
such as <tt>$Foo</tt> might implicitly call the get function.
Similarly, typing <tt>$Foo = 4</tt> would call the underlying set
function to change the value.</p>
<h3><a name="Scripting_nn6">4.2.3 Constants</a></h3>
<p> In many cases, a C program or library may define a large collection
of constants. For example:</p>
<div class="code">
<pre>
#define RED 0xff0000
#define BLUE 0x0000ff
#define GREEN 0x00ff00
</pre>
</div>
<p> To make constants available, their values can be stored in scripting
language variables such as <tt>$RED</tt>, <tt>$BLUE</tt>, and <tt>
$GREEN</tt>. Virtually all scripting languages provide C functions for
creating variables so installing constants is usually a trivial
exercise.</p>
<h3><a name="Scripting_nn7">4.2.4 Structures and classes</a></h3>
<p> Although scripting languages have no trouble accessing simple
functions and variables, accessing C/C++ structures and classes present
a different problem. This is because the implementation of structures
is largely related to the problem of data representation and layout.
Furthermore, certain language features are difficult to map to an
interpreter. For instance, what does C++ inheritance mean in a Perl
interface?</p>
<p> The most straightforward technique for handling structures is to
implement a collection of accessor functions that hide the underlying
representation of a structure. For example,</p>
<div class="code">
<pre>
struct Vector {
Vector();
~Vector();
double x, y, z;
};
</pre>
</div>
<p> can be transformed into the following set of functions :</p>
<div class="code">
<pre>
Vector *new_Vector();
void delete_Vector(Vector *v);
double Vector_x_get(Vector *v);
double Vector_y_get(Vector *v);
double Vector_z_get(Vector *v);
void Vector_x_set(Vector *v, double x);
void Vector_y_set(Vector *v, double y);
void Vector_z_set(Vector *v, double z);
</pre>
</div>
<p> Now, from an interpreter these function might be used as follows:</p>
<div class="targetlang">
<pre>
% set v [new_Vector]
% Vector_x_set $v 3.5
% Vector_y_get $v
% delete_Vector $v
% ...
</pre>
</div>
<p> Since accessor functions provide a mechanism for accessing the
internals of an object, the interpreter does not need to know anything
about the actual representation of a <tt>Vector</tt>.</p>
<h3><a name="Scripting_nn8">4.2.5 Proxy classes</a></h3>
<p> In certain cases, it is possible to use the low-level accessor
functions to create a proxy class, also known as a shadow class. A
proxy class is a special kind of object that gets created in a
scripting language to access a C/C++ class (or struct) in a way that
looks like the original structure (that is, it proxies the real C++
class). For example, if you have the following C++ definition :</p>
<div class="code">
<pre>
class Vector {
public:
Vector();
~Vector();
double x, y, z;
};
</pre>
</div>
<p> A proxy classing mechanism would allow you to access the structure
in a more natural manner from the interpreter. For example, in Python,
you might want to do this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; v = Vector()
&gt;&gt;&gt; v.x = 3
&gt;&gt;&gt; v.y = 4
&gt;&gt;&gt; v.z = -13
&gt;&gt;&gt; ...
&gt;&gt;&gt; del v
</pre>
</div>
<p> Similarly, in Perl5 you may want the interface to work like this:</p>
<div class="targetlang">
<pre>
$v = new Vector;
$v-&gt;{x} = 3;
$v-&gt;{y} = 4;
$v-&gt;{z} = -13;
</pre>
</div>
<p> Finally, in Tcl :</p>
<div class="targetlang">
<pre>
Vector v
v configure -x 3 -y 4 -z -13
</pre>
</div>
<p> When proxy classes are used, two objects are really at work--one in
the scripting language, and an underlying C/C++ object. Operations
affect both objects equally and for all practical purposes, it appears
as if you are simply manipulating a C/C++ object.</p>
<h2><a name="Scripting_nn9">4.3 Building scripting language extensions</a>
</h2>
<p> The final step in using a scripting language with your C/C++
application is adding your extensions to the scripting language itself.
There are two primary approaches for doing this. The preferred
technique is to build a dynamically loadable extension in the form of a
shared library. Alternatively, you can recompile the scripting language
interpreter with your extensions added to it.</p>
<h3><a name="Scripting_nn10">4.3.1 Shared libraries and dynamic loading</a>
</h3>
<p> To create a shared library or DLL, you often need to look at the
manual pages for your compiler and linker. However, the procedure for a
few common platforms is shown below:</p>
<div class="shell">
<pre>
# Build a shared library for Solaris
gcc -fPIC -c example.c example_wrap.c -I/usr/local/include
ld -G example.o example_wrap.o -o example.so
# Build a shared library for Linux
gcc -fPIC -c example.c example_wrap.c -I/usr/local/include
gcc -shared example.o example_wrap.o -o example.so
</pre>
</div>
<p> To use your shared library, you simply use the corresponding command
in the scripting language (load, import, use, etc...). This will import
your module and allow you to start using it. For example:</p>
<div class="targetlang">
<pre>
% load ./example.so
% fact 4
24
%
</pre>
</div>
<p> When working with C++ codes, the process of building shared
libraries may be more complicated--primarily due to the fact that C++
modules may need additional code in order to operate correctly. On many
machines, you can build a shared C++ module by following the above
procedures, but changing the link line to the following :</p>
<div class="shell">
<pre>
c++ -shared example.o example_wrap.o -o example.so
</pre>
</div>
<h3><a name="Scripting_nn11">4.3.2 Linking with shared libraries</a></h3>
<p> When building extensions as shared libraries, it is not uncommon for
your extension to rely upon other shared libraries on your machine. In
order for the extension to work, it needs to be able to find all of
these libraries at run-time. Otherwise, you may get an error such as
the following :</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import graph
Traceback (innermost last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
File &quot;/home/sci/data1/beazley/graph/graph.py&quot;, line 2, in ?
import graphc
ImportError: 1101:/home/sci/data1/beazley/bin/python: rld: Fatal Error: cannot
successfully map soname 'libgraph.so' under any of the filenames /usr/lib/libgraph.so:/
lib/libgraph.so:/lib/cmplrs/cc/libgraph.so:/usr/lib/cmplrs/cc/libgraph.so:
&gt;&gt;&gt;
</pre>
</div>
<p> What this error means is that the extension module created by SWIG
depends upon a shared library called &quot;<tt>libgraph.so</tt>&quot; that the
system was unable to locate. To fix this problem, there are a few
approaches you can take.</p>
<ul>
<li>Link your extension and explicitly tell the linker where the
required libraries are located. Often times, this can be done with a
special linker flag such as <tt>-R</tt>, <tt>-rpath</tt>, etc. This is
not implemented in a standard manner so read the man pages for your
linker to find out more about how to set the search path for shared
libraries.</li>
<li>Put shared libraries in the same directory as the executable. This
technique is sometimes required for correct operation on non-Unix
platforms.</li>
<li>Set the UNIX environment variable <tt>LD_LIBRARY_PATH</tt> to the
directory where shared libraries are located before running Python.
Although this is an easy solution, it is not recommended. Consider
setting the path using linker options instead.</li>
</ul>
<h3><a name="Scripting_nn12">4.3.3 Static linking</a></h3>
<p> With static linking, you rebuild the scripting language interpreter
with extensions. The process usually involves compiling a short main
program that adds your customized commands to the language and starts
the interpreter. You then link your program with a library to produce a
new scripting language executable.</p>
<p> Although static linking is supported on all platforms, this is not
the preferred technique for building scripting language extensions. In
fact, there are very few practical reasons for doing this--consider
using shared libraries instead.</p>
<HR NOSHADE>
<h1><a name="SWIG">5 SWIG Basics</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#SWIG_nn2">Running SWIG</a>
<ul>
<li><a href="#SWIG_nn3">Input format</a></li>
<li><a href="#SWIG_output">SWIG output</a></li>
<li><a href="#SWIG_nn5">Comments</a></li>
<li><a href="#SWIG_nn6">C Preprocessor</a></li>
<li><a href="#SWIG_nn7">SWIG directives</a></li>
<li><a href="#SWIG_nn8">Parser limitations</a></li>
<li><a href="#SWIG_parse_tree">Parse tree</a></li>
</ul>
</li>
<li><a href="#SWIG_nn9">Wrapping simple C declarations</a>
<ul>
<li><a href="#SWIG_nn10">Basic type handling</a></li>
<li><a href="#SWIG_nn11">Global variables</a></li>
<li><a href="#SWIG_nn12">Constants</a></li>
<li><a href="#SWIG_nn13">A brief word about const</a></li>
<li><a href="#SWIG_nn14">A cautionary tale of char *</a></li>
</ul>
</li>
<li><a href="#SWIG_nn15">Pointers and complex objects</a>
<ul>
<li><a href="#SWIG_nn16">Simple pointers</a></li>
<li><a href="#SWIG_nn17">Run time pointer type checking</a></li>
<li><a href="#SWIG_nn18">Derived types, structs, and classes</a></li>
<li><a href="#SWIG_nn19">Undefined datatypes</a></li>
<li><a href="#SWIG_nn20">Typedef</a></li>
</ul>
</li>
<li><a href="#SWIG_nn21">Other Practicalities</a>
<ul>
<li><a href="#SWIG_nn22">Passing structures by value</a></li>
<li><a href="#SWIG_nn23">Return by value</a></li>
<li><a href="#SWIG_nn24">Linking to structure variables</a></li>
<li><a href="#SWIG_nn25">Linking to char *</a></li>
<li><a href="#SWIG_nn26">Arrays</a></li>
<li><a href="#SWIG_readonly_variables">Creating read-only variables</a></li>
<li><a href="#SWIG_rename_ignore">Renaming and ignoring declarations</a>
<ul>
<li><a href="#SWIG_nn29">Simple renaming of specific identifiers</a></li>
<li><a href="#SWIG_ignore">Ignoring identifiers</a></li>
<li><a href="#SWIG_advanced_renaming">Advanced renaming support</a></li>
<li><a href="#SWIG_limiting_renaming">Limiting global renaming rules</a></li>
<li><a href="#SWIG_chosen_unignore">Ignoring everything then wrapping a
few selected symbols</a></li>
</ul>
</li>
<li><a href="#SWIG_default_args">Default/optional arguments</a></li>
<li><a href="#SWIG_nn30">Pointers to functions and callbacks</a></li>
</ul>
</li>
<li><a href="#SWIG_nn31">Structures and unions</a>
<ul>
<li><a href="#SWIG_nn32">Typedef and structures</a></li>
<li><a href="#SWIG_nn33">Character strings and structures</a></li>
<li><a href="#SWIG_nn34">Array members</a></li>
<li><a href="#SWIG_structure_data_members">Structure data members</a></li>
<li><a href="#SWIG_nn36">C constructors and destructors</a></li>
<li><a href="#SWIG_adding_member_functions">Adding member functions to C
structures</a></li>
<li><a href="#SWIG_nested_structs">Nested structures</a></li>
<li><a href="#SWIG_nn39">Other things to note about structure wrapping</a>
</li>
</ul>
</li>
<li><a href="#SWIG_nn40">Code Insertion</a>
<ul>
<li><a href="#SWIG_nn41">The output of SWIG</a></li>
<li><a href="#SWIG_nn42">Code insertion blocks</a></li>
<li><a href="#SWIG_nn43">Inlined code blocks</a></li>
<li><a href="#SWIG_nn44">Initialization blocks</a></li>
</ul>
</li>
<li><a href="#SWIG_nn45">An Interface Building Strategy</a>
<ul>
<li><a href="#SWIG_nn46">Preparing a C program for SWIG</a></li>
<li><a href="#SWIG_nn47">The SWIG interface file</a></li>
<li><a href="#SWIG_nn48">Why use separate interface files?</a></li>
<li><a href="#SWIG_nn49">Getting the right header files</a></li>
<li><a href="#SWIG_nn50">What to do with main()</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> This chapter describes the basic operation of SWIG, the structure of
its input files, and how it handles standard ISO C declarations. C++
support is described in the next chapter. However, C++ programmers
should still read this chapter to understand the basics. Specific
details about each target language are described in later chapters.</p>
<h2><a name="SWIG_nn2">5.1 Running SWIG</a></h2>
<p> To run SWIG, use the <tt>swig</tt> command with options and a
filename like this:</p>
<div class="shell">
<pre>
swig [ options ] filename
</pre>
</div>
<p> where <tt>filename</tt> is a SWIG interface file or a C/C++ header
file. Full help can be seen by running <tt>swig -help</tt>. Below is
the common set of options that can be used. Additional options are also
defined for each target language. A full list can be obtained by
running <tt>swig<em> -&lt;lang&gt;</em> -help</tt> for language<em> &lt;lang&gt;</em>
specific options, for example, <tt>swig -ruby -help</tt> for Ruby.</p>
<div class="shell">
<pre>
Supported Target Language Options
-csharp - Generate C# wrappers
-d - Generate D wrappers
-go - Generate Go wrappers
-guile - Generate Guile wrappers
-java - Generate Java wrappers
-javascript - Generate Javascript wrappers
-lua - Generate Lua wrappers
-octave - Generate Octave wrappers
-perl5 - Generate Perl 5 wrappers
-php7 - Generate PHP 8 or later wrappers
-python - Generate Python wrappers
-r - Generate R (aka GNU S) wrappers
-ruby - Generate Ruby wrappers
-scilab - Generate Scilab wrappers
-tcl8 - Generate Tcl 8 wrappers
-xml - Generate XML wrappers
Experimental Target Language Options
-c - Generate C wrappers
-ocaml - Generate OCaml wrappers
Deprecated Target Language Options
-mzscheme - Generate MzScheme/Racket wrappers
General Options
-addextern - Add extra extern declarations
-c++ - Enable C++ processing
-co &lt;file&gt; - Check &lt;file&gt; out of the SWIG library
-copyctor - Automatically generate copy constructors wherever possible
-cpperraswarn - Treat the preprocessor #error statement as #warning (default)
-cppext &lt;ext&gt; - Change file extension of generated C++ files to &lt;ext&gt;
(default is cxx)
-copyright - Display copyright notices
-debug-classes - Display information about the classes found in the interface
-debug-module &lt;n&gt;- Display module parse tree at stages 1-4, &lt;n&gt; is a csv list of stages
-debug-symtabs - Display symbol tables information
-debug-symbols - Display target language symbols in the symbol tables
-debug-csymbols - Display C symbols in the symbol tables
-debug-lsymbols - Display target language layer symbols
-debug-quiet - Display less parse tree node debug info when using other -debug options
-debug-tags - Display information about the tags found in the interface
-debug-template - Display information for debugging templates
-debug-top &lt;n&gt; - Display entire parse tree at stages 1-4, &lt;n&gt; is a csv list of stages
-debug-typedef - Display information about the types and typedefs in the interface
-debug-typemap - Display typemap debugging information
-debug-tmsearch - Display typemap search debugging information
-debug-tmused - Display typemaps used debugging information
-directors - Turn on director mode for all the classes, mainly for testing
-dirprot - Turn on wrapping of protected members for director classes (default)
-D&lt;symbol&gt;[=&lt;value&gt;] - Define symbol &lt;symbol&gt; (for conditional compilation)
-E - Preprocess only, does not generate wrapper code
-external-runtime [file] - Export the SWIG runtime stack
-fakeversion &lt;v&gt;- Make SWIG fake the program version number to &lt;v&gt;
-fcompact - Compile in compact mode
-features &lt;list&gt;- Set global features, where &lt;list&gt; is a comma separated list of
features, eg -features directors,autodoc=1
If no explicit value is given to the feature, a default of 1 is used
-fastdispatch - Enable fast dispatch mode to produce faster overload dispatcher code
-Fmicrosoft - Display error/warning messages in Microsoft format
-Fstandard - Display error/warning messages in commonly used format
-fvirtual - Compile in virtual elimination mode
-help - Display help
-I- - Don't search the current directory
-I&lt;dir&gt; - Look for SWIG files in directory &lt;dir&gt;
-ignoremissing - Ignore missing include files
-importall - Follow all #include statements as imports
-includeall - Follow all #include statements
-l&lt;ifile&gt; - Include SWIG library file &lt;ifile&gt;
-macroerrors - Report errors inside macros
-M - List all dependencies
-MD - Is equivalent to `-M -MF &lt;file&gt;', except `-E' is not implied
-MF &lt;file&gt; - Generate dependencies into &lt;file&gt; and continue generating wrappers
-MM - List dependencies, but omit files in SWIG library
-MMD - Like `-MD', but omit files in SWIG library
-module &lt;name&gt; - Set module name to &lt;name&gt;
-MP - Generate phony targets for all dependencies
-MT &lt;target&gt; - Set the target of the rule emitted by dependency generation
-nocontract - Turn off contract checking
-nocpperraswarn - Do not treat the preprocessor #error statement as #warning
-nodefaultctor - Do not generate implicit default constructors
-nodefaultdtor - Do not generate implicit default destructors
-nodirprot - Do not wrap director protected members
-noexcept - Do not wrap exception specifiers
-nofastdispatch - Disable fast dispatch mode (default)
-nopreprocess - Skip the preprocessor step
-notemplatereduce - Disable reduction of the typedefs in templates
-O - Enable the optimization options:
-fastdispatch -fvirtual
-o &lt;outfile&gt; - Set name of C/C++ output file to &lt;outfile&gt;
-oh &lt;headfile&gt; - Set name of C++ output header file for directors to &lt;headfile&gt;
-outcurrentdir - Set default output dir to current dir instead of input file's path
-outdir &lt;dir&gt; - Set language specific files output directory to &lt;dir&gt;
-pcreversion - Display PCRE2 version information
-small - Compile in virtual elimination and compact mode
-std=&lt;standard&gt; - Set the C or C++ language &lt;standard&gt; for inputs
-swiglib - Report location of SWIG library and exit
-templatereduce - Reduce all the typedefs in templates
-U&lt;symbol&gt; - Undefine symbol &lt;symbol&gt;
-v - Run in verbose mode
-version - Display SWIG version number
-Wall - Remove all warning suppression, also implies -Wextra
-Wallkw - Enable keyword warnings for all the supported languages
-Werror - Treat warnings as errors
-Wextra - Adds the following additional warnings: 309,403,405,512,321,322
-w&lt;list&gt; - Suppress/add warning messages, eg -w401,+321 - see Warnings.html
-xmlout &lt;file&gt; - Write XML version of the parse tree to &lt;file&gt; after normal processing
</pre>
</div>
<p> Arguments may also be passed in a command-line options file (also
known as a response file) which is useful if they exceed the system
command line length limit. To do this, put the arguments in a file,
then provide the file name prefixed with <tt>@</tt> like so:</p>
<div class="shell">
<pre>
swig @<em>file</em>
</pre>
</div>
<p> The options read from the file are inserted in place of the file
option. If the file does not exist, or cannot be read, then the option
will be treated literally and not removed.</p>
<p> Options in the file are separated by whitespace. A whitespace
character may be included in an option by surrounding the entire option
in either single or double quotes. Any character (including a
backslash) may be included by prefixing the character to be included
with a backslash. The file may itself contain additional <tt>@file</tt>
options; any such options will be processed recursively.</p>
<h3><a name="SWIG_nn3">5.1.1 Input format</a></h3>
<p> As input, SWIG expects a file containing ISO C/C++ declarations and
special SWIG directives. More often than not, this is a special SWIG
interface file which is usually denoted with a special <tt>.i</tt> or <tt>
.swg</tt> suffix. In certain cases, SWIG can be used directly on raw
header files or source files. However, this is not the most typical
case and there are several reasons why you might not want to do this
(described later).</p>
<p> The most common format of a SWIG interface is as follows:</p>
<div class="code">
<pre>
%module mymodule
%{
#include &quot;myheader.h&quot;
%}
// Now list ISO C/C++ declarations
int foo;
int bar(int x);
...
</pre>
</div>
<p> The module name is supplied using the special <tt>%module</tt>
directive. Modules are described further in the <a href="#Modules_introduction">
Modules Introduction</a> section.</p>
<p> Everything in the <tt>%{ ... %}</tt> block is simply copied verbatim
to the resulting wrapper file created by SWIG. This section is almost
always used to include header files and other declarations that are
required to make the generated wrapper code compile. It is important to
emphasize that just because you include a declaration in a SWIG input
file, that declaration does<em> not</em> automatically appear in the
generated wrapper code---therefore you need to make sure you include
the proper header files in the <tt>%{ ... %}</tt> section. It should be
noted that the text enclosed in <tt>%{ ... %}</tt> is not parsed or
interpreted by SWIG. The <tt>%{...%}</tt> syntax and semantics in SWIG
is analogous to that of the declarations section used in input files to
parser generation tools such as yacc or bison.</p>
<h3><a name="SWIG_output">5.1.2 SWIG output</a></h3>
<p> The output of SWIG is a C/C++ file that contains all of the wrapper
code needed to build an extension module. SWIG may generate some
additional files depending on the target language. By default, an input
file with the name <tt>file.i</tt> is transformed into a file <tt>
file_wrap.c</tt> or <tt>file_wrap.cxx</tt> (depending on whether or not
the <tt>-c++</tt> option has been used). The name of the output C/C++
file can be changed using the <tt>-o</tt> option. In certain cases,
file suffixes are used by the compiler to determine the source language
(C, C++, etc.). Therefore, you have to use the <tt>-o</tt> option to
change the suffix of the SWIG-generated wrapper file if you want
something different than the default. For example:</p>
<div class="shell">
<pre>
$ swig -c++ -python -o example_wrap.cpp example.i
</pre>
</div>
<p> The C/C++ output file created by SWIG often contains everything that
is needed to construct an extension module for the target scripting
language. SWIG is not a stub compiler nor is it usually necessary to
edit the output file (and if you look at the output, you probably won't
want to). To build the final extension module, the SWIG output file is
compiled and linked with the rest of your C/C++ program to create a
shared library.</p>
<p> For many target languages SWIG will also generate proxy class files
in the target language. The default output directory for these language
specific files is the same directory as the generated C/C++ file. This
can be modified using the <tt>-outdir</tt> option. For example:</p>
<div class="shell">
<pre>
$ swig -c++ -python -outdir pyfiles -o cppfiles/example_wrap.cpp example.i
</pre>
</div>
<p> If the directories <tt>cppfiles</tt> and <tt>pyfiles</tt> exist, the
following will be generated:</p>
<div class="shell">
<pre>
cppfiles/example_wrap.cpp
pyfiles/example.py
</pre>
</div>
<p> If the <tt>-outcurrentdir</tt> option is used (without <tt>-o</tt>)
then SWIG behaves like a typical C/C++ compiler and the default output
directory is then the current directory. Without this option the
default output directory is the path to the input file. If <tt>-o</tt>
and <tt>-outcurrentdir</tt> are used together, <tt>-outcurrentdir</tt>
is effectively ignored as the output directory for the language files
is the same directory as the generated C/C++ file if not overridden
with <tt>-outdir</tt>.</p>
<h3><a name="SWIG_nn5">5.1.3 Comments</a></h3>
<p> C and C++ style comments may appear anywhere in interface files. In
previous versions of SWIG, comments were used to generate documentation
files. Doxygen comments can now be used to generate target language
specific documentation, see the <a href="Doxygen.html">Doxygen</a>
chapter.</p>
<h3><a name="SWIG_nn6">5.1.4 C Preprocessor</a></h3>
<p> Like C, SWIG preprocesses all input files through an enhanced
version of the C preprocessor. All standard preprocessor features are
supported including file inclusion, conditional compilation and macros.
However, <tt>#include</tt> statements are ignored unless the <tt>
-includeall</tt> command line option has been supplied. The reason for
disabling includes is that SWIG is sometimes used to process raw C
header files. In this case, you usually only want the extension module
to include functions in the supplied header file rather than everything
that might be included by that header file (i.e., system headers, C
library functions, etc.).</p>
<p> It should also be noted that the SWIG preprocessor skips all text
enclosed inside a <tt>%{...%}</tt> block. In addition, the preprocessor
includes a number of macro handling enhancements that make it more
powerful than the normal C preprocessor. These extensions are described
in the &quot;<a href="#Preprocessor">Preprocessor</a>&quot; chapter.</p>
<h3><a name="SWIG_nn7">5.1.5 SWIG directives</a></h3>
<p> Most of SWIG's operation is controlled by special directives that
are always preceded by a &quot;<tt>%</tt>&quot; to distinguish them from normal C
declarations. These directives are used to give SWIG hints or to alter
SWIG's parsing behavior in some manner.</p>
<p> Since SWIG directives are not legal C syntax, it is generally not
possible to include them in header files. However, SWIG directives can
be included in C header files using conditional compilation like this:</p>
<div class="code">
<pre>
/* header.h --- Some header file */
/* SWIG directives -- only seen if SWIG is running */
#ifdef SWIG
%module foo
#endif
</pre>
</div>
<p> <tt>SWIG</tt> is a special preprocessing symbol defined by SWIG when
it is parsing an input file.</p>
<h3><a name="SWIG_nn8">5.1.6 Parser limitations</a></h3>
<p> Although SWIG can parse most C/C++ declarations, it does not provide
a complete C/C++ parser implementation. Most of these limitations
pertain to very complicated type declarations and certain advanced C++
features. Specifically, the following features are not currently
supported:</p>
<ul>
<li>
<p> Non-conventional type declarations. For example, SWIG does not
support declarations such as the following (even though this is legal
C):</p>
<div class="code">
<pre>
/* Non-conventional placement of storage specifier (extern) */
const int extern Number;
/* Extra declarator grouping */
Matrix (foo); // A global variable
/* Extra declarator grouping in parameters */
void bar(Spam (Grok)(Doh));
</pre>
</div>
<p> In practice, few (if any) C programmers actually write code like
this since this style is never featured in programming books. However,
if you're feeling particularly obfuscated, you can certainly break SWIG
(although why would you want to?).</p>
</li>
<li>
<p> Running SWIG on C++ source files (the code in a .C, .cpp or .cxx
file) is not recommended. The usual approach is to feed SWIG header
files for parsing C++ definitions and declarations. The main reason is
if SWIG parses a scoped definition or declaration (as is normal for C++
source files), it is ignored, unless a declaration for the symbol was
parsed earlier. For example</p>
<div class="code">
<pre>
/* bar not wrapped unless foo has been defined and
the declaration of bar within foo has already been parsed */
int foo::bar(int) {
... whatever ...
}
</pre>
</div></li>
<li>
<p> Certain advanced features of C++ such as nested classes are not yet
fully supported. Please see the C++ <a href="#SWIGPlus_nested_classes">
Nested classes</a> section for more information.</p>
</li>
</ul>
<p> In the event of a parsing error, conditional compilation can be used
to skip offending code. For example:</p>
<div class="code">
<pre>
#ifndef SWIG
... some bad declarations ...
#endif
</pre>
</div>
<p> Alternatively, you can just delete the offending code from the
interface file.</p>
<p> One of the reasons why SWIG does not provide a full C++ parser
implementation is that it has been designed to work with incomplete
specifications and to be very permissive in its handling of C/C++
datatypes (e.g., SWIG can generate interfaces even when there are
missing class declarations or opaque datatypes). Unfortunately, this
approach makes it extremely difficult to implement certain parts of a
C/C++ parser as most compilers use type information to assist in the
parsing of more complex declarations (for the truly curious, the
primary complication in the implementation is that the SWIG parser does
not utilize a separate<em> typedef-name</em> terminal symbol as
described on p. 234 of K&amp;R).</p>
<h3><a name="SWIG_parse_tree">5.1.7 Parse tree</a></h3>
<p> SWIG can dump its parse tree in either a simple text or XML format.
This can be useful for advanced development or debugging purposes.
Details are covered in the developer documentation <a href="#Extending">
Extending SWIG</a> chapter.</p>
<h2><a name="SWIG_nn9">5.2 Wrapping simple C declarations</a></h2>
<p> SWIG wraps simple C declarations by creating an interface that
closely matches the way in which the declarations would be used in a C
program. For example, consider the following interface file:</p>
<div class="code">
<pre>
%module example
%inline %{
extern double sin(double x);
extern int strcmp(const char *, const char *);
extern int Foo;
%}
#define STATUS 50
#define VERSION &quot;1.1&quot;
</pre>
</div>
<p> In this file, there are two functions <tt>sin()</tt> and <tt>
strcmp()</tt>, a global variable <tt>Foo</tt>, and two constants <tt>
STATUS</tt> and <tt>VERSION</tt>. When SWIG creates an extension module,
these declarations are accessible as scripting language functions,
variables, and constants respectively. For example, in Tcl:</p>
<div class="targetlang">
<pre>
% sin 3
5.2335956
% strcmp Dave Mike
-1
% puts $Foo
42
% puts $STATUS
50
% puts $VERSION
1.1
</pre>
</div>
<p> Or in Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; example.sin(3)
5.2335956
&gt;&gt;&gt; example.strcmp('Dave', 'Mike')
-1
&gt;&gt;&gt; print example.cvar.Foo
42
&gt;&gt;&gt; print example.STATUS
50
&gt;&gt;&gt; print example.VERSION
1.1
</pre>
</div>
<p> Whenever possible, SWIG creates an interface that closely matches
the underlying C/C++ code. However, due to subtle differences between
languages, run-time environments, and semantics, it is not always
possible to do so. The next few sections describe various aspects of
this mapping.</p>
<h3><a name="SWIG_nn10">5.2.1 Basic type handling</a></h3>
<p> In order to build an interface, SWIG has to convert C/C++ datatypes
to equivalent types in the target language. Generally, scripting
languages provide a more limited set of primitive types than C.
Therefore, this conversion process involves a certain amount of type
coercion.</p>
<p> Most scripting languages provide a single integer type that is
implemented using the <tt>int</tt> or <tt>long</tt> datatype in C. The
following list shows all of the C datatypes that SWIG will convert to
and from integers in the target language:</p>
<div class="code">
<pre>
int
short
long
unsigned
signed
unsigned short
unsigned long
unsigned char
signed char
bool
</pre>
</div>
<p> When an integral value is converted from C, a cast is used to
convert it to the representation in the target language. Thus, a 16 bit
short in C may be promoted to a 32 bit integer. When integers are
converted in the other direction, the value is cast back into the
original C type. If the value is too large to fit, it is silently
truncated.
<!-- Dave: Maybe we should fix this -->
</p>
<p> <tt>unsigned char</tt> and <tt>signed char</tt> are special cases
that are handled as small 8-bit integers. Normally, the <tt>char</tt>
datatype is mapped as a one-character ASCII string.</p>
<p> The <tt>bool</tt> datatype is cast to and from an integer value of 0
and 1 unless the target language provides a special boolean type.</p>
<p> Some care is required when working with large integer values. Most
scripting languages use 32-bit integers so mapping a 64-bit long
integer may lead to truncation errors. Similar problems may arise with
32 bit unsigned integers (which may appear as large negative numbers).
As a rule of thumb, the <tt>int</tt> datatype and all variations of <tt>
char</tt> and <tt>short</tt> datatypes are safe to use. For <tt>unsigned
int</tt> and <tt>long</tt> datatypes, you will need to carefully check
the correct operation of your program after it has been wrapped with
SWIG.</p>
<p> Although the SWIG parser supports the <tt>long long</tt> datatype,
not all language modules support it. This is because <tt>long long</tt>
usually exceeds the integer precision available in the target language.
In certain modules such as Tcl and Perl5, <tt>long long</tt> integers
are encoded as strings. This allows the full range of these numbers to
be represented. However, it does not allow <tt>long long</tt> values to
be used in arithmetic expressions. It should also be noted that
although <tt>long long</tt> is part of the ISO C99 standard, it is not
universally supported by all C compilers. Make sure you are using a
compiler that supports <tt>long long</tt> before trying to use this
type with SWIG.</p>
<p> SWIG recognizes the following floating point types :</p>
<div class="code">
<pre>
float
double
</pre>
</div>
<p> Floating point numbers are mapped to and from the natural
representation of floats in the target language. This is almost always
a C <tt>double</tt>. The rarely used datatype of <tt>long double</tt>
is not supported by SWIG.</p>
<p> The <tt>char</tt> datatype is mapped into a NULL terminated ASCII
string with a single character. When used in a scripting language it
shows up as a tiny string containing the character value. When
converting the value back into C, SWIG takes a character string from
the scripting language and strips off the first character as the char
value. Thus if the value &quot;foo&quot; is assigned to a <tt>char</tt> datatype,
it gets the value `f'.</p>
<p> The <tt>char *</tt> datatype is handled as a NULL-terminated ASCII
string. SWIG maps this into a 8-bit character string in the target
scripting language. SWIG converts character strings in the target
language to NULL terminated strings before passing them into C/C++. The
default handling of these strings does not allow them to have embedded
NULL bytes. Therefore, the <tt>char *</tt> datatype is not generally
suitable for passing binary data. However, it is possible to change
this behavior by defining a SWIG typemap. See the chapter on <a href="#Typemaps">
Typemaps</a> for details about this.</p>
<p> At this time, SWIG provides limited support for Unicode and
wide-character strings (the C <tt>wchar_t</tt> type). Some languages
provide typemaps for wchar_t, but bear in mind these might not be
portable across different operating systems. This is a delicate topic
that is poorly understood by many programmers and not implemented in a
consistent manner across languages. For those scripting languages that
provide Unicode support, Unicode strings are often available in an
8-bit representation such as UTF-8 that can be mapped to the <tt>char *</tt>
type (in which case the SWIG interface will probably work). If the
program you are wrapping uses Unicode, there is no guarantee that
Unicode characters in the target language will use the same internal
representation (e.g., UCS-2 vs. UCS-4). You may need to write some
special conversion functions.</p>
<h3><a name="SWIG_nn11">5.2.2 Global variables</a></h3>
<p> Whenever possible, SWIG maps C/C++ global variables into scripting
language variables. For example,</p>
<div class="code">
<pre>
%module example
double foo;
</pre>
</div>
<p> results in a scripting language variable like this:</p>
<div class="code">
<pre>
# Tcl
set foo [3.5] ;# Set foo to 3.5
puts $foo ;# Print the value of foo
# Python
cvar.foo = 3.5 # Set foo to 3.5
print(cvar.foo) # Print value of foo
# Perl
$foo = 3.5; # Set foo to 3.5
print $foo, &quot;\n&quot;; # Print value of foo
# Ruby
Module.foo = 3.5 # Set foo to 3.5
print Module.foo, &quot;\n&quot; # Print value of foo
</pre>
</div>
<p> Whenever the scripting language variable is used, the underlying C
global variable is accessed. Although SWIG makes every attempt to make
global variables work like scripting language variables, it is not
always possible to do so. For instance, in Python, all global variables
must be accessed through a special variable object known as <tt>cvar</tt>
(shown above). In Ruby, variables are accessed as attributes of the
module. Other languages may convert variables to a pair of accessor
functions. For example, the Java module generates a pair of functions <tt>
double get_foo()</tt> and <tt>set_foo(double val)</tt> that are used to
manipulate the value.</p>
<p> Finally, if a global variable has been declared as <tt>const</tt>,
it only supports read-only access. Note: this behavior is new to
SWIG-1.3. Earlier versions of SWIG incorrectly handled <tt>const</tt>
and created constants instead.</p>
<h3><a name="SWIG_nn12">5.2.3 Constants</a></h3>
<p> Constants can be created using <tt>#define</tt>, enumerations, or a
special <tt>%constant</tt> directive. The following interface file
shows a few valid constant declarations :</p>
<div class="code">
<pre>
#define I_CONST 5 // An integer constant
#define PI 3.14159 // A Floating point constant
#define S_CONST &quot;hello world&quot; // A string constant
#define NEWLINE '\n' // Character constant
enum boolean {NO=0, YES=1};
enum months {JAN, FEB, MAR, APR, MAY, JUN, JUL, AUG,
SEP, OCT, NOV, DEC};
%constant double BLAH = 42.37;
#define PI_4 PI/4
#define FLAGS 0x04 | 0x08 | 0x40
</pre>
</div>
<p> In <tt>#define</tt> declarations, the type of a constant is inferred
by syntax. For example, a number with a decimal point is assumed to be
floating point. In addition, SWIG must be able to fully resolve all of
the symbols used in a <tt>#define</tt> in order for a constant to
actually be created. This restriction is necessary because <tt>#define</tt>
is also used to define preprocessor macros that are definitely not
meant to be part of the scripting language interface. For example:</p>
<div class="code">
<pre>
#define EXTERN extern
EXTERN void foo();
</pre>
</div>
<p> In this case, you probably don't want to create a constant called <tt>
EXTERN</tt> (what would the value be?). In general, SWIG will not create
constants for macros unless the value can be completely determined by
the preprocessor. For instance, in the above example, the declaration</p>
<div class="code">
<pre>
#define PI_4 PI/4
</pre>
</div>
<p> defines a constant because <tt>PI</tt> was already defined as a
constant and the value is known. However, for the same conservative
reasons even a constant with a simple cast will be ignored, such as</p>
<div class="code">
<pre>
#define F_CONST (double) 5 // A floating point constant with cast
</pre>
</div>
<p> This logic can lead to false attempts at converting <tt>#define</tt>
into <tt>%constant</tt> though. For example the following case does not
have any undefined symbols within the macro:</p>
<div class="code">
<pre>
// For indicating pure virtual functions such as: virtual void f() PURE;
#define PURE = 0
</pre>
</div>
<p> A warning is issued:</p>
<div class="shell">
<pre>
pure.h:1: Warning 305: Bad constant value (ignored).
</pre>
</div>
<p> In such cases simply ignore the warning or suppress it using the
normal warning suppression techniques.</p>
<p> The use of constant expressions is allowed, but SWIG does not
evaluate them. Rather, it passes them through to the output file and
lets the C compiler perform the final evaluation (SWIG does perform a
limited form of type-checking however).</p>
<p> For enumerations, it is critical that the original enum definition
be included somewhere in the interface file (either in a header file or
in the <tt>%{ %}</tt> block). SWIG only translates the enumeration into
code needed to add the constants to a scripting language. It needs the
original enumeration declaration in order to get the correct enum
values as assigned by the C compiler.</p>
<p> The <tt>%constant</tt> directive is used to more precisely create
constants corresponding to different C datatypes. Although it is not
usually needed for simple values, it is more useful when working with
pointers and other more complex datatypes. Typically, <tt>%constant</tt>
is only used when you want to add constants to the scripting language
interface that are not defined in the original header file.</p>
<h3><a name="SWIG_nn13">5.2.4 A brief word about const</a></h3>
<p> A common confusion with C programming is the semantic meaning of the
<tt>const</tt> qualifier in declarations--especially when it is mixed
with pointers and other type modifiers. In fact, previous versions of
SWIG handled <tt>const</tt> incorrectly--a situation that SWIG-1.3.7
and newer releases have fixed.</p>
<p> Starting with SWIG-1.3, all variable declarations, regardless of any
use of <tt>const</tt>, are wrapped as global variables. If a
declaration happens to be declared as <tt>const</tt>, it is wrapped as
a read-only variable. To tell if a variable is <tt>const</tt> or not,
you need to look at the right-most occurrence of the <tt>const</tt>
qualifier (that appears before the variable name). If the right-most <tt>
const</tt> occurs after all other type modifiers (such as pointers),
then the variable is <tt>const</tt>. Otherwise, it is not.</p>
<p> Here are some examples of <tt>const</tt> declarations.</p>
<div class="code">
<pre>
const char a; // A constant character
char const b; // A constant character (the same)
char *const c; // A constant pointer to a character
const char *const d; // A constant pointer to a constant character
</pre>
</div>
<p> Here is an example of a declaration that is not <tt>const</tt>:</p>
<div class="code">
<pre>
const char *e; // A pointer to a constant character. The pointer
// may be modified.
</pre>
</div>
<p> In this case, the pointer <tt>e</tt> can change---it's only the
value being pointed to that is read-only.</p>
<p> Please note that for const parameters or return types used in a
function, SWIG pretty much ignores the fact that these are const, see
the section on <a href="#SWIGPlus_const">const-correctness</a> for more
information.</p>
<p><b> Compatibility Note:</b> One reason for changing SWIG to handle <tt>
const</tt> declarations as read-only variables is that there are many
situations where the value of a <tt>const</tt> variable might change.
For example, a library might export a symbol as <tt>const</tt> in its
public API to discourage modification, but still allow the value to
change through some other kind of internal mechanism. Furthermore,
programmers often overlook the fact that with a constant declaration
like <tt>char *const</tt>, the underlying data being pointed to can be
modified--it's only the pointer itself that is constant. In an embedded
system, a <tt>const</tt> declaration might refer to a read-only memory
address such as the location of a memory-mapped I/O device port (where
the value changes, but writing to the port is not supported by the
hardware). Rather than trying to build a bunch of special cases into
the <tt>const</tt> qualifier, the new interpretation of <tt>const</tt>
as &quot;read-only&quot; is simple and exactly matches the actual semantics of <tt>
const</tt> in C/C++. If you really want to create a constant as in older
versions of SWIG, use the <tt>%constant</tt> directive instead. For
example:</p>
<div class="code">
<pre>
%constant double PI = 3.14159;
</pre>
</div>
<p> or</p>
<div class="code">
<pre>
#ifdef SWIG
#define const %constant
#endif
const double foo = 3.4;
const double bar = 23.4;
const int spam = 42;
#ifdef SWIG
#undef const
#endif
...
</pre>
</div>
<h3><a name="SWIG_nn14">5.2.5 A cautionary tale of char *</a></h3>
<p> Before going any further, there is one bit of caution involving <tt>
char *</tt> that must now be mentioned. When strings are passed from a
scripting language to a C <tt>char *</tt>, the pointer usually points
to string data stored inside the interpreter. It is almost always a
really bad idea to modify this data. Furthermore, some languages may
explicitly disallow it. For instance, in Python, strings are supposed
to be immutable. If you violate this, you will probably receive a vast
amount of wrath when you unleash your module on the world.</p>
<p> The primary source of problems are functions that might modify
string data in place. A classic example would be a function like this:</p>
<div class="code">
<pre>
char *strcat(char *s, const char *t)
</pre>
</div>
<p> Although SWIG will certainly generate a wrapper for this, its
behavior will be undefined. In fact, it will probably cause your
application to crash with a segmentation fault or other memory related
problem. This is because <tt>s</tt> refers to some internal data in the
target language---data that you shouldn't be touching.</p>
<p> The bottom line: don't rely on <tt>char *</tt> for anything other
than read-only input values. However, it must be noted that you could
change the behavior of SWIG using <a href="#Typemaps">typemaps</a>.</p>
<h2><a name="SWIG_nn15">5.3 Pointers and complex objects</a></h2>
<p> Most C programs manipulate arrays, structures, and other types of
objects. This section discusses the handling of these datatypes.</p>
<h3><a name="SWIG_nn16">5.3.1 Simple pointers</a></h3>
<p> Pointers to primitive C datatypes such as</p>
<div class="code">
<pre>
int *
double ***
char **
</pre>
</div>
<p> are fully supported by SWIG. Rather than trying to convert the data
being pointed to into a scripting representation, SWIG simply encodes
the pointer itself into a representation that contains the actual value
of the pointer and a type-tag. Thus, the SWIG representation of the
above pointers (in Tcl), might look like this:</p>
<div class="targetlang">
<pre>
_10081012_p_int
_1008e124_ppp_double
_f8ac_pp_char
</pre>
</div>
<p> A NULL pointer is represented by the string &quot;NULL&quot; or the value 0
encoded with type information.</p>
<p> All pointers are treated as opaque objects by SWIG. Thus, a pointer
may be returned by a function and passed around to other C functions as
needed. For all practical purposes, the scripting language interface
works in exactly the same way as you would use the pointer in a C
program. The only difference is that there is no mechanism for
dereferencing the pointer since this would require the target language
to understand the memory layout of the underlying object.</p>
<p> The scripting language representation of a pointer value should
never be manipulated directly. Even though the values shown look like
hexadecimal addresses, the numbers used may differ from the actual
machine address (e.g., on little-endian machines, the digits may appear
in reverse order). Furthermore, SWIG does not normally map pointers
into high-level objects such as associative arrays or lists (for
example, converting an <tt>int *</tt> into an list of integers). There
are several reasons why SWIG does not do this:</p>
<ul>
<li>There is not enough information in a C declaration to properly map
pointers into higher level constructs. For example, an <tt>int *</tt>
may indeed be an array of integers, but if it contains ten million
elements, converting it into a list object is probably a bad idea.</li>
<li>The underlying semantics associated with a pointer is not known by
SWIG. For instance, an <tt>int *</tt> might not be an array at
all--perhaps it is an output value!</li>
<li>By handling all pointers in a consistent manner, the implementation
of SWIG is greatly simplified and less prone to error.</li>
</ul>
<h3><a name="SWIG_nn17">5.3.2 Run time pointer type checking</a></h3>
<p> By allowing pointers to be manipulated from a scripting language,
extension modules effectively bypass compile-time type checking in the
C/C++ compiler. To prevent errors, a type signature is encoded into all
pointer values and is used to perform run-time type checking. This
type-checking process is an integral part of SWIG and can not be
disabled or modified without using typemaps (described in later
chapters).</p>
<p> Like C, <tt>void *</tt> matches any kind of pointer. Furthermore, <tt>
NULL</tt> pointers can be passed to any function that expects to receive
a pointer. Although this has the potential to cause a crash, <tt>NULL</tt>
pointers are also sometimes used as sentinel values or to denote a
missing/empty value. Therefore, SWIG leaves NULL pointer checking up to
the application.</p>
<h3><a name="SWIG_nn18">5.3.3 Derived types, structs, and classes</a></h3>
<p> For everything else (structs, classes, arrays, etc...) SWIG applies
a very simple rule :</p>
<center><b> Everything else is a pointer</b></center>
<p> In other words, SWIG manipulates everything else by reference. This
model makes sense because most C/C++ programs make heavy use of
pointers and SWIG can use the type-checked pointer mechanism already
present for handling pointers to basic datatypes.</p>
<p> Although this probably sounds complicated, it's really quite simple.
Suppose you have an interface file like this :</p>
<div class="code">
<pre>
%module fileio
FILE *fopen(char *, char *);
int fclose(FILE *);
unsigned fread(void *ptr, unsigned size, unsigned nobj, FILE *);
unsigned fwrite(void *ptr, unsigned size, unsigned nobj, FILE *);
void *malloc(int nbytes);
void free(void *);
</pre>
</div>
<p> In this file, SWIG doesn't know what a <tt>FILE</tt> is, but since
it's used as a pointer, so it doesn't really matter what it is. If you
wrapped this module into Python, you can use the functions just like
you expect :</p>
<div class="targetlang">
<pre>
# Copy a file
def filecopy(source, target):
f1 = fopen(source, &quot;r&quot;)
f2 = fopen(target, &quot;w&quot;)
buffer = malloc(8192)
nbytes = fread(buffer, 8192, 1, f1)
while (nbytes &gt; 0):
fwrite(buffer, 8192, 1, f2)
nbytes = fread(buffer, 8192, 1, f1)
free(buffer)
</pre>
</div>
<p> In this case <tt>f1</tt>, <tt>f2</tt>, and <tt>buffer</tt> are all
opaque objects containing C pointers. It doesn't matter what value they
contain--our program works just fine without this knowledge.</p>
<h3><a name="SWIG_nn19">5.3.4 Undefined datatypes</a></h3>
<p> When SWIG encounters an undeclared datatype, it automatically
assumes that it is a structure or class. For example, suppose the
following function appeared in a SWIG input file:</p>
<div class="code">
<pre>
void matrix_multiply(Matrix *a, Matrix *b, Matrix *c);
</pre>
</div>
<p> SWIG has no idea what a &quot;<tt>Matrix</tt>&quot; is. However, it is
obviously a pointer to something so SWIG generates a wrapper using its
generic pointer handling code.</p>
<p> Unlike C or C++, SWIG does not actually care whether <tt>Matrix</tt>
has been previously defined in the interface file or not. This allows
SWIG to generate interfaces from only partial or limited information.
In some cases, you may not care what a <tt>Matrix</tt> really is as
long as you can pass an opaque reference to one around in the scripting
language interface.</p>
<p> An important detail to mention is that SWIG will gladly generate
wrappers for an interface when there are unspecified type names.
However,<b> all unspecified types are internally handled as pointers to
structures or classes!</b> For example, consider the following
declaration:</p>
<div class="code">
<pre>
void foo(size_t num);
</pre>
</div>
<p> If <tt>size_t</tt> is undeclared, SWIG generates wrappers that
expect to receive a type of <tt>size_t *</tt> (this mapping is
described shortly). As a result, the scripting interface might behave
strangely. For example:</p>
<div class="code">
<pre>
foo(40);
TypeError: expected a _p_size_t.
</pre>
</div>
<p> The only way to fix this problem is to make sure you properly
declare type names using <tt>typedef</tt>.</p>
<!-- We might want to add an error reporting flag to swig -->
<h3><a name="SWIG_nn20">5.3.5 Typedef</a></h3>
<p> Like C, <tt>typedef</tt> can be used to define new type names in
SWIG. For example:</p>
<div class="code">
<pre>
typedef unsigned int size_t;
</pre>
</div>
<p> <tt>typedef</tt> definitions appearing in a SWIG interface are not
propagated to the generated wrapper code. Therefore, they either need
to be defined in an included header file or placed in the declarations
section like this:</p>
<div class="code">
<pre>
%{
/* Include in the generated wrapper file */
typedef unsigned int size_t;
%}
/* Tell SWIG about it */
typedef unsigned int size_t;
</pre>
</div>
<p> or</p>
<div class="code">
<pre>
%inline %{
typedef unsigned int size_t;
%}
</pre>
</div>
<p> In certain cases, you might be able to include other header files to
collect type information. For example:</p>
<div class="code">
<pre>
%module example
%import &quot;sys/types.h&quot;
</pre>
</div>
<p> In this case, you might run SWIG as follows:</p>
<div class="shell">
<pre>
$ swig -I/usr/include -includeall example.i
</pre>
</div>
<p> It should be noted that your mileage will vary greatly here. System
headers are notoriously complicated and may rely upon a variety of
non-standard C coding extensions (e.g., such as special directives to
GCC). Unless you exactly specify the right include directories and
preprocessor symbols, this may not work correctly (you will have to
experiment).</p>
<p> SWIG tracks <tt>typedef</tt> declarations and uses this information
for run-time type checking. For instance, if you use the above <tt>
typedef</tt> and had the following function declaration:</p>
<div class="code">
<pre>
void foo(unsigned int *ptr);
</pre>
</div>
<p> The corresponding wrapper function will accept arguments of type <tt>
unsigned int *</tt> or <tt>size_t *</tt>.</p>
<h2><a name="SWIG_nn21">5.4 Other Practicalities</a></h2>
<p> So far, this chapter has presented almost everything you need to
know to use SWIG for simple interfaces. However, some C programs use
idioms that are somewhat more difficult to map to a scripting language
interface. This section describes some of these issues.</p>
<h3><a name="SWIG_nn22">5.4.1 Passing structures by value</a></h3>
<p> Sometimes a C function takes structure parameters that are passed by
value. For example, consider the following function:</p>
<div class="code">
<pre>
double dot_product(Vector a, Vector b);
</pre>
</div>
<p> To deal with this, SWIG transforms the function to use pointers by
creating a wrapper equivalent to the following:</p>
<div class="code">
<pre>
double wrap_dot_product(Vector *a, Vector *b) {
Vector x = *a;
Vector y = *b;
return dot_product(x, y);
}
</pre>
</div>
<p> In the target language, the <tt>dot_product()</tt> function now
accepts pointers to Vectors instead of Vectors. For the most part, this
transformation is transparent so you might not notice.</p>
<h3><a name="SWIG_nn23">5.4.2 Return by value</a></h3>
<p> C functions that return structures or classes datatypes by value are
more difficult to handle. Consider the following function:</p>
<div class="code">
<pre>
Vector cross_product(Vector v1, Vector v2);
</pre>
</div>
<p> This function wants to return <tt>Vector</tt>, but SWIG only really
supports pointers. As a result, SWIG creates a wrapper like this:</p>
<div class="code">
<pre>
Vector *wrap_cross_product(Vector *v1, Vector *v2) {
Vector x = *v1;
Vector y = *v2;
Vector *result;
result = (Vector *) malloc(sizeof(Vector));
*(result) = cross_product(x, y);
return result;
}
</pre>
</div>
<p> or if SWIG was run with the <tt>-c++</tt> option:</p>
<div class="code">
<pre>
Vector *wrap_cross_product(Vector *v1, Vector *v2) {
Vector x = *v1;
Vector y = *v2;
Vector *result = new Vector(cross_product(x, y)); // Uses default copy or move constructor
return result;
}
</pre>
</div>
<p> In both cases, SWIG allocates a new object and returns a reference
to it. It is up to the user to delete the returned object when it is no
longer in use. Clearly, this will leak memory if you are unaware of the
implicit memory allocation and don't take steps to free the result.
That said, it should be noted that some language modules can now
automatically track newly created objects and reclaim memory for you.
Consult the documentation for each language module for more details.</p>
<p> It should also be noted that the handling of pass/return by value in
C++ has some special cases. For example, the above code fragments don't
work correctly if <tt>Vector</tt> doesn't define a default constructor.
The section on SWIG and C++ has more information about this case.</p>
<h3><a name="SWIG_nn24">5.4.3 Linking to structure variables</a></h3>
<p> When global variables or class members involving structures are
encountered, SWIG handles them as pointers. For example, a global
variable like this</p>
<div class="code">
<pre>
Vector unit_i;
</pre>
</div>
<p> gets mapped to an underlying pair of set/get functions like this :</p>
<div class="code">
<pre>
Vector *unit_i_get() {
return &amp;unit_i;
}
void unit_i_set(Vector *value) {
unit_i = *value;
}
</pre>
</div>
<p> Again some caution is in order. A global variable created in this
manner will show up as a pointer in the target scripting language. It
would be an extremely bad idea to free or destroy such a pointer. Also,
C++ classes must supply a properly defined copy constructor in order
for assignment to work correctly.</p>
<h3><a name="SWIG_nn25">5.4.4 Linking to char *</a></h3>
<p> When a global variable of type <tt>char *</tt> appears, SWIG uses <tt>
malloc()</tt> or <tt>new</tt> to allocate memory for the new value.
Specifically, if you have a variable like this</p>
<div class="code">
<pre>
char *foo;
</pre>
</div>
<p> SWIG generates the following code:</p>
<div class="code">
<pre>
/* C mode */
void foo_set(char *value) {
free(foo);
foo = (char *) malloc(strlen(value)+1);
strcpy(foo, value);
}
/* C++ mode. When -c++ option is used */
void foo_set(char *value) {
delete [] foo;
foo = new char[strlen(value)+1];
strcpy(foo, value);
}
</pre>
</div>
<p> If this is not the behavior that you want, consider making the
variable read-only using the <tt>%immutable</tt> directive.
Alternatively, you might write a short assist-function to set the value
exactly like you want. For example:</p>
<div class="code">
<pre>
%inline %{
void set_foo(char *value) {
strncpy(foo, value, 50);
}
%}
</pre>
</div>
<p> Note: If you write an assist function like this, you will have to
call it as a function from the target scripting language (it does not
work like a variable). For example, in Python you will have to write:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; set_foo(&quot;Hello World&quot;)
</pre>
</div>
<p> A common mistake with <tt>char *</tt> variables is to link to a
variable declared like this:</p>
<div class="code">
<pre>
char *VERSION = &quot;1.0&quot;;
</pre>
</div>
<p> In this case, the variable will be readable, but any attempt to
change the value results in a segmentation or general protection fault.
This is due to the fact that SWIG is trying to release the old value
using <tt>free</tt> or <tt>delete</tt> when the string literal value
currently assigned to the variable wasn't allocated using <tt>malloc()</tt>
or <tt>new</tt>. To fix this behavior, you can either mark the variable
as read-only, write a typemap (as described in Chapter 6), or write a
special set function as shown. Another alternative is to declare the
variable as an array:</p>
<div class="code">
<pre>
char VERSION[64] = &quot;1.0&quot;;
</pre>
</div>
<p> When variables of type <tt>const char *</tt> are declared, SWIG
still generates functions for setting and getting the value. However,
the default behavior does<em> not</em> release the previous contents
(resulting in a possible memory leak). In fact, you may get a warning
message such as this when wrapping such a variable:</p>
<div class="shell">
<pre>
example.i:20. Typemap warning. Setting const char * variable may leak memory
</pre>
</div>
<p> The reason for this behavior is that <tt>const char *</tt> variables
are often used to point to string literals. For example:</p>
<div class="code">
<pre>
const char *foo = &quot;Hello World\n&quot;;
</pre>
</div>
<p> Therefore, it's a really bad idea to call <tt>free()</tt> on such a
pointer. On the other hand, it<em> is</em> legal to change the pointer
to point to some other value. When setting a variable of this type,
SWIG allocates a new string (using malloc or new) and changes the
pointer to point to the new value. However, repeated modifications of
the value will result in a memory leak since the old value is not
released.</p>
<h3><a name="SWIG_nn26">5.4.5 Arrays</a></h3>
<p> Arrays are fully supported by SWIG, but they are always handled as
pointers instead of mapping them to a special array object or list in
the target language. Thus, the following declarations :</p>
<div class="code">
<pre>
int foobar(int a[40]);
void grok(char *argv[]);
void transpose(double a[20][20]);
</pre>
</div>
<p> are processed as if they were really declared like this:</p>
<div class="code">
<pre>
int foobar(int *a);
void grok(char **argv);
void transpose(double (*a)[20]);
</pre>
</div>
<p> Like C, SWIG does not perform array bounds checking. It is up to the
user to make sure the pointer points to a suitably allocated region of
memory.</p>
<p> Multi-dimensional arrays are transformed into a pointer to an array
of one less dimension. For example:</p>
<div class="code">
<pre>
int [10]; // Maps to int *
int [10][20]; // Maps to int (*)[20]
int [10][20][30]; // Maps to int (*)[20][30]
</pre>
</div>
<p> It is important to note that in the C type system, a
multidimensional array <tt>a[][]</tt> is<b> NOT</b> equivalent to a
single pointer <tt>*a</tt> or a double pointer such as <tt>**a</tt>.
Instead, a pointer to an array is used (as shown above) where the
actual value of the pointer is the starting memory location of the
array. The reader is strongly advised to dust off their C book and
re-read the section on arrays before using them with SWIG.</p>
<p> Array variables are supported, but are read-only by default. For
example:</p>
<div class="code">
<pre>
int a[100][200];
</pre>
</div>
<p> In this case, reading the variable 'a' returns a pointer of type <tt>
int (*)[200]</tt> that points to the first element of the array <tt>
&amp;a[0][0]</tt>. Trying to modify 'a' results in an error. This is because
SWIG does not know how to copy data from the target language into the
array. To work around this limitation, you may want to write a few
simple assist functions like this:</p>
<div class="code">
<pre>
%inline %{
void a_set(int i, int j, int val) {
a[i][j] = val;
}
int a_get(int i, int j) {
return a[i][j];
}
%}
</pre>
</div>
<p> To dynamically create arrays of various sizes and shapes, it may be
useful to write some helper functions in your interface. For example:</p>
<div class="code">
<pre>
// Some array helpers
%inline %{
/* Create any sort of [size] array */
int *int_array(int size) {
return (int *) malloc(size*sizeof(int));
}
/* Create a two-dimension array [size][10] */
int (*int_array_10(int size))[10] {
return (int (*)[10]) malloc(size*10*sizeof(int));
}
%}
</pre>
</div>
<p> Arrays of <tt>char</tt> are handled as a special case by SWIG. In
this case, strings in the target language can be stored in the array.
For example, if you have a declaration like this,</p>
<div class="code">
<pre>
char pathname[256];
</pre>
</div>
<p> SWIG generates functions for both getting and setting the value that
are equivalent to the following code:</p>
<div class="code">
<pre>
char *pathname_get() {
return pathname;
}
void pathname_set(char *value) {
strncpy(pathname, value, 256);
}
</pre>
</div>
<p> In the target language, the value can be set like a normal variable.</p>
<h3><a name="SWIG_readonly_variables">5.4.6 Creating read-only variables</a>
</h3>
<p> A read-only variable can be created by using the <tt>%immutable</tt>
directive as shown :</p>
<div class="code">
<pre>
// File : interface.i
int a; // Can read/write
%immutable;
int b, c, d; // Read only variables
%mutable;
double x, y; // read/write
</pre>
</div>
<p> The <tt>%immutable</tt> directive enables read-only mode until it is
explicitly disabled using the <tt>%mutable</tt> directive. As an
alternative to turning read-only mode off and on like this, individual
declarations can also be tagged as immutable. For example:</p>
<div class="code">
<pre>
%immutable x; // Make x read-only
...
double x; // Read-only (from earlier %immutable directive)
double y; // Read-write
...
</pre>
</div>
<p> The <tt>%mutable</tt> and <tt>%immutable</tt> directives are
actually <a href="#Customization_features">%feature directives</a>
defined like this:</p>
<div class="code">
<pre>
#define %immutable %feature(&quot;immutable&quot;)
#define %mutable %feature(&quot;immutable&quot;, &quot;&quot;)
</pre>
</div>
<p> If you wanted to make all wrapped variables read-only, barring one
or two, it might be easier to take this approach:</p>
<div class="code">
<pre>
%immutable; // Make all variables read-only
%feature(&quot;immutable&quot;, &quot;0&quot;) x; // except, make x read/write
...
double x;
double y;
double z;
...
</pre>
</div>
<p> Read-only variables are also created when declarations are declared
as <tt>const</tt>. For example:</p>
<div class="code">
<pre>
const int foo; /* Read only variable */
char * const version=&quot;1.0&quot;; /* Read only variable */
</pre>
</div>
<p> SWIG will also create read-only variables for non-assignable types,
more details in the <a href="#SWIGPlus_member_data">Member data</a>
section in the C++ chapter.</p>
<h3><a name="SWIG_rename_ignore">5.4.7 Renaming and ignoring
declarations</a></h3>
<h4><a name="SWIG_nn29">5.4.7.1 Simple renaming of specific identifiers</a>
</h4>
<p> Normally, the name of a C declaration is used when that declaration
is wrapped into the target language. However, this may generate a
conflict with a keyword or already existing function in the scripting
language. To resolve a name conflict, you can use the <tt>%rename</tt>
directive as shown :</p>
<div class="code">
<pre>
// interface.i
%rename(my_print) print;
extern void print(const char *);
%rename(foo) a_really_long_and_annoying_name;
extern int a_really_long_and_annoying_name;
</pre>
</div>
<p> SWIG still calls the correct C function, but in this case the
function <tt>print()</tt> will really be called &quot;<tt>my_print()</tt>&quot;
in the target language.</p>
<p> The placement of the <tt>%rename</tt> directive is arbitrary as long
as it appears before the declarations to be renamed. A common technique
is to write code for wrapping a header file like this:</p>
<div class="code">
<pre>
// interface.i
%rename(my_print) print;
%rename(foo) a_really_long_and_annoying_name;
%include &quot;header.h&quot;
</pre>
</div>
<p> <tt>%rename</tt> applies a renaming operation to all future
occurrences of a name. The renaming applies to functions, variables,
class and structure names, member functions, and member data. For
example, if you had two-dozen C++ classes, all with a member function
named `print' (which is a keyword in Python), you could rename them all
to `output' by specifying :</p>
<div class="code">
<pre>
%rename(output) print; // Rename all `print' functions to `output'
</pre>
</div>
<p> A new <tt>%rename</tt> for the same name will replace the current <tt>
%rename</tt> for all uses after it in the file, and setting the new name
to &quot;&quot; will remove the rename. So, for instance, if you wanted to rename
some things in one file and not in another, you could do:</p>
<div class="code">
<pre>
%rename(print1) print;
%include &quot;header1.h&quot; //Anything &quot;print&quot; in here will become &quot;print1&quot;
%rename(print2) print;
%include &quot;header2.h&quot; //Anything &quot;print&quot; in here will become &quot;print2&quot;
%rename(&quot;&quot;) print;
%include &quot;header3.h&quot; //Anything &quot;print&quot; in here will remain &quot;print&quot;
</pre>
</div>
<p> SWIG does not normally perform any checks to see if the functions it
wraps are already defined in the target scripting language. However, if
you are careful about namespaces and your use of modules, you can
usually avoid these problems.</p>
<p> When wrapping C code, simple use of identifiers/symbols with <tt>
%rename</tt> usually suffices. When wrapping C++ code, simple use of
simple identifiers/symbols with <tt>%rename</tt> might be too limiting
when using C++ features such as function overloading, default
arguments, namespaces, template specialization etc. If you are using
the <tt>%rename</tt> directive and C++, make sure you read the <a href="#SWIGPlus">
SWIG and C++</a> chapter and in particular the section on <a href="#SWIGPlus_ambiguity_resolution_renaming">
Renaming and ambiguity resolution</a> for method overloading and default
arguments.</p>
<h4><a name="SWIG_ignore">5.4.7.2 Ignoring identifiers</a></h4>
<p> Closely related to <tt>%rename</tt> is the <tt>%ignore</tt>
directive. <tt>%ignore</tt> instructs SWIG to ignore declarations that
match a given identifier. For example:</p>
<div class="code">
<pre>
%ignore print; // Ignore all declarations named print
%ignore MYMACRO; // Ignore a macro
...
#define MYMACRO 123
void print(const char *);
...
</pre>
</div>
<p> Any function, variable etc which matches <tt>%ignore</tt> will not
be wrapped and therefore will not be available from the target
language. A common usage of <tt>%ignore</tt> is to selectively remove
certain declarations from a header file without having to add
conditional compilation to the header. However, it should be stressed
that this only works for simple declarations. If you need to remove a
whole section of problematic code, the SWIG preprocessor should be used
instead.</p>
<h4><a name="SWIG_advanced_renaming">5.4.7.3 Advanced renaming support</a>
</h4>
<p> While writing <tt>%rename</tt> for specific declarations is simple
enough, sometimes the same renaming rule needs to be applied to many,
maybe all, identifiers in the SWIG input. For example, it may be
necessary to apply some transformation to all the names in the target
language to better follow its naming conventions, like adding a
specific prefix to all wrapped functions. Doing it individually for
each function is impractical so SWIG supports applying a renaming rule
to all declarations if the name of the identifier to be renamed is not
specified:</p>
<div class="code">
<pre>
%rename(&quot;myprefix_%s&quot;) &quot;&quot;; // print&nbsp;-&gt;&nbsp;myprefix_print
</pre>
</div>
<p> This also shows that the argument of <tt>%rename</tt> doesn't have
to be a literal string but can be a <tt>printf()</tt>-like format
string. In the simplest form, <tt>&quot;%s&quot;</tt> is replaced with the name
of the original declaration, as shown above. However this is not always
enough and SWIG provides extensions to the usual format string syntax
to allow applying a (SWIG-defined) function to the argument. For
example, to wrap all C functions <tt>do_something_long()</tt> as more
Java-like <tt>doSomethingLong()</tt> you can use the <tt>
&quot;lowercamelcase&quot;</tt> extended format specifier like this:</p>
<div class="code">
<pre>
%rename(&quot;%(lowercamelcase)s&quot;) &quot;&quot;; // foo_bar -&gt; fooBar; FooBar -&gt; fooBar
</pre>
</div>
<p> Some functions can be parametrized, for example the <tt>&quot;strip&quot;</tt>
one strips the provided prefix from its argument. The prefix is
specified as part of the format string, following a colon after the
function name:</p>
<div class="code">
<pre>
%rename(&quot;%(strip:[wx])s&quot;) &quot;&quot;; // wxHello -&gt; Hello; FooBar -&gt; FooBar
</pre>
</div>
<p> Below is the table summarizing all currently defined functions with
an example of applying each one. Note that some of them have two names,
a shorter one and a more descriptive one, but the two functions are
otherwise equivalent:</p>
<table border="1" cellpadding="5" summary="Format string functions">
<tr><th>Function</th><th>Returns</th><th colspan="2">Example (in/out)</th>
</tr>
<tr><td><tt>uppercase</tt> or <tt>upper</tt></td><td>Upper case version
of the string.</td><td><tt>Print</tt></td><td><tt>PRINT</tt></td></tr>
<tr><td><tt>lowercase</tt> or <tt>lower</tt></td><td>Lower case version
of the string.</td><td><tt>Print</tt></td><td><tt>print</tt></td></tr>
<tr><td><tt>title</tt></td><td>String with first letter capitalized and
the rest in lower case.</td><td><tt>print</tt></td><td><tt>Print</tt></td>
</tr>
<tr><td><tt>firstuppercase</tt></td><td>String with the first letter
capitalized and the rest unchanged.</td><td><tt>printIt</tt></td><td><tt>
PrintIt</tt></td></tr>
<tr><td><tt>firstlowercase</tt></td><td>String with the first letter in
lower case and the rest unchanged.</td><td><tt>PrintIt</tt></td><td><tt>
printIt</tt></td></tr>
<tr><td><tt>camelcase</tt> or <tt>ctitle</tt></td><td>String with
capitalized first letter and any letter following an underscore (which
are removed in the process) and rest in lower case.</td><td><tt>
print_it</tt></td><td><tt>PrintIt</tt></td></tr>
<tr><td><tt>lowercamelcase</tt> or <tt>lctitle</tt></td><td>String with
every letter following an underscore (which is removed in the process)
capitalized and rest, including the first letter, in lower case.</td><td>
<tt>print_it</tt></td><td><tt>printIt</tt></td></tr>
<tr><td><tt>undercase</tt> or <tt>utitle</tt></td><td>Lower case string
with underscores inserted before every upper case letter in the
original string and any number not at the end of string. Logically,
this is the reverse of <tt>camelcase</tt>.</td><td><tt>PrintIt</tt></td><td>
<tt>print_it</tt></td></tr>
<tr><td><tt>schemify</tt></td><td>String with all underscores replaced
with dashes, resulting in more Lispers/Schemers-pleasing name.</td><td><tt>
print_it</tt></td><td><tt>print-it</tt></td></tr>
<tr><td><tt>strip:[prefix]</tt></td><td>String without the given prefix
or the original string if it doesn't start with this prefix. Note that
square brackets should be used literally, e.g. <tt>
%rename(&quot;strip:[wx]&quot;)</tt></td><td><tt>wxPrint</tt></td><td><tt>Print</tt>
</td></tr>
<tr><td><tt>rstrip:[suffix]</tt></td><td>String without the given suffix
or the original string if it doesn't end with this suffix. Note that
square brackets should be used literally, e.g. <tt>
%rename(&quot;rstrip:[Cls]&quot;)</tt></td><td><tt>PrintCls</tt></td><td><tt>Print</tt>
</td></tr>
<tr><td><span style="white-space: nowrap;"><tt>regex:/pattern/subst/</tt>
</span></td><td>String after (Perl-like) regex substitution operation.
This function allows applying arbitrary regular expressions to the
identifier names. The<i> pattern</i> part is a regular expression in
Perl syntax (as supported by the <a href="https://www.pcre.org/">Perl
Compatible Regular Expressions</a>) (PCRE2 library) and the<i> subst</i>
string can contain back-references of the form <tt>\N</tt> where <tt>N</tt>
is a digit from 0 to 9, or one of the following escape sequences: <tt>
\l</tt>, <tt>\L</tt>, <tt>\u</tt>, <tt>\U</tt> or <tt>\E</tt>. The
back-references are replaced with the contents of the corresponding
capture group while the escape sequences perform the case conversion in
the substitution string: <tt>\l</tt> and <tt>\L</tt> convert to the
lower case, while <tt>\u</tt> and <tt>\U</tt> convert to the upper
case. The difference between the elements of each pair is that <tt>\l</tt>
and <tt>\u</tt> change the case of the next character only, while <tt>
\L</tt> and <tt>\U</tt> do it for all the remaining characters or until <tt>
\E</tt> is encountered. Finally please notice that backslashes need to
be escaped in C strings, so in practice <tt>&quot;\\&quot;</tt> must be used in
all these escape sequences. For example, to remove any alphabetic
prefix before an underscore and capitalize the remaining part you could
use the following directive: <tt>%rename(&quot;regex:/(\\w+)_(.*)/\\u\\2/&quot;)</tt>
</td><td><tt>prefix_print</tt></td><td><tt>Print</tt></td></tr>
</table>
<p> The most general function of all of the above ones is the <tt>regex</tt>
one. Here are some more examples of its use:</p>
<div class="code">
<pre>
// Strip the wx prefix from all identifiers except those starting with wxEVT
%rename(&quot;%(regex:/wx(?!EVT)(.*)/\\1/)s&quot;) &quot;&quot;; // wxSomeWidget -&gt; SomeWidget
// wxEVT_PAINT -&gt; wxEVT_PAINT
// Apply a rule for renaming the enum elements to avoid the common prefixes
// which are redundant in C#/Java
%rename(&quot;%(regex:/^([A-Z][a-z]+)+_(.*)/\\2/)s&quot;, %$isenumitem) &quot;&quot;; // Colour_Red -&gt; Red
// Remove all &quot;Set/Get&quot; prefixes.
%rename(&quot;%(regex:/^(Set|Get)(.*)/\\2/)s&quot;) &quot;&quot;; // SetValue -&gt; Value
// GetValue -&gt; Value
</pre>
</div>
<p> As before, everything that was said above about <tt>%rename</tt>
also applies to <tt>%ignore</tt>. In fact, the latter is just a special
case of the former and ignoring an identifier is the same as renaming
it to the special <tt>&quot;$ignore&quot;</tt> value. So the following snippets</p>
<div class="code">
<pre>
%ignore print;
</pre>
</div>
<p> and</p>
<div class="code">
<pre>
%rename(&quot;$ignore&quot;) print;
</pre>
</div>
<p> are exactly equivalent and <tt>%rename</tt> can be used to
selectively ignore multiple declarations using the previously described
matching possibilities.</p>
<h4><a name="SWIG_limiting_renaming">5.4.7.4 Limiting global renaming
rules</a></h4>
<p> As explained in the previous sections, it is possible to either
rename individual declarations or apply a rename rule to all of them at
once. In practice, the latter is however rarely appropriate as there
are always some exceptions to the general rules. To deal with them, the
scope of an unnamed <tt>%rename</tt> can be limited using subsequent <tt>
match</tt> parameters. They can be applied to any of the attributes
associated by SWIG with the declarations appearing in its input. For
example:</p>
<div class="code">
<pre>
%rename(&quot;foo&quot;, match$name=&quot;bar&quot;) &quot;&quot;;
</pre>
</div>
<p> can be used to achieve the same effect as the simpler</p>
<div class="code">
<pre>
%rename(&quot;foo&quot;) bar;
</pre>
</div>
<p> and so is not very interesting on its own. However <tt>match</tt>
can also be applied to the declaration type, for example <tt>
match=&quot;class&quot;</tt> restricts the match to class declarations only (in
C++) and <tt>match=&quot;enumitem&quot;</tt> restricts it to the enum elements.
SWIG also provides convenience macros for such match expressions, for
example</p>
<div class="code">
<pre>
%rename(&quot;%(title)s&quot;, %$isenumitem) &quot;&quot;;
// same as:
%rename(&quot;%(title)s&quot;, match=&quot;enumitem&quot;) &quot;&quot;;
</pre>
</div>
<p> will capitalize the names of all the enum elements but not change
the case of the other declarations. Similarly, <tt>%$isclass</tt>, <tt>
%$isfunction</tt>, <tt>%$isconstructor</tt>, <tt>%$isunion</tt>, <tt>
%$istemplate</tt>, and <tt>%$isvariable</tt> can be used. Many other
checks are possible and this documentation is not exhaustive, see the
&quot;%rename predicates&quot; section in <tt>swig.swg</tt> for the full list of
supported match expressions.</p>
<p> A logical not is also possible by using <tt>notmatch</tt>. For
example, <tt>notmatch=&quot;enumitem&quot;</tt> will restrict the match to all
items that are not enum elements. There is also a <tt>%$not</tt> macro
which simply expands to &quot;not&quot;. Be careful using this as some of the
other macros in <tt>swig.swg</tt> are complex expressions and so it
will only &quot;notmatch&quot; the first part of the expression.</p>
<div class="code">
<pre>
%rename(&quot;%(title)s&quot;, %$not %$isenumitem) &quot;&quot;;
// same as:
%rename(&quot;%(title)s&quot;, notmatch=&quot;enumitem&quot;) &quot;&quot;;
</pre>
</div>
<p> For a comprehensive understanding of how the matching works, the
internal <a href="#Extending_nn8">parse tree</a> needs to be examined
using the command line option: <tt>-debug-module 1 -debug-quiet</tt>. A
snippet of the resulting output might be:</p>
<div class="shell">
<pre>
+++ destructor ----------------------------------------
| access - &quot;public&quot;
| decl - &quot;f().&quot;
| ismember - &quot;1&quot;
| name - &quot;~Shape&quot;
| storage - &quot;virtual&quot;
| sym:name - &quot;~Shape&quot;
</pre>
</div>
<p> Here the node type is a &quot;destructor&quot; and in order to match all
destructor nodes, use <tt>match=&quot;destructor&quot;</tt>. To match one of the
listed attributes in the node, such as when the storage is virtual, use
<tt>match$storage=&quot;virtual&quot;</tt>. This will match all nodes that have a
storage attribute set to &quot;virtual&quot;. To match only virtual destructors,
combine them and use <tt>match=&quot;destructor&quot;, match$storage=&quot;virtual&quot;</tt>
.</p>
<p> While the vast majority of these internal parse tree nodes are
unlikely to change from one version of SWIG to the next,<b> use these
matching rules at your own risk</b> as there are no guarantees that
they will not change.</p>
<p> In addition to literally matching some string with <tt>match</tt>
you can also use <tt>regexmatch</tt> or <tt>notregexmatch</tt> to match
a string against a regular expression. For example, to ignore all
functions having &quot;Old&quot; as a suffix you could use</p>
<div class="code">
<pre>
%rename(&quot;$ignore&quot;, regexmatch$name=&quot;Old$&quot;) &quot;&quot;;
</pre>
</div>
<p> For simple cases like this, specifying the regular expression for
the declaration name directly can be preferable and can also be done
using <tt>regextarget</tt>:</p>
<div class="code">
<pre>
%rename(&quot;$ignore&quot;, regextarget=1) &quot;Old$&quot;;
</pre>
</div>
<p> Notice that the check is done only against the name of the
declaration itself, if you need to match the full name of a C++
declaration you must use <tt>fullname</tt> attribute:</p>
<div class="code">
<pre>
%rename(&quot;$ignore&quot;, regextarget=1, fullname=1) &quot;NameSpace::ClassName::.*Old$&quot;;
</pre>
</div>
<p> As for <tt>notregexmatch</tt>, it restricts the match only to the
strings not matching the specified regular expression. So to rename all
declarations to lower case except those consisting of capital letters
only:</p>
<div class="code">
<pre>
%rename(&quot;$(lower)s&quot;, notregexmatch$name=&quot;^[A-Z]+$&quot;) &quot;&quot;;
</pre>
</div>
<p> Finally, variants of <tt>%rename</tt> and <tt>%ignore</tt>
directives can be used to help wrap C++ overloaded functions and
methods or C++ methods which use default arguments. This is described
in the <a href="#SWIGPlus_ambiguity_resolution_renaming">Renaming and
ambiguity resolution</a> section in the C++ chapter.</p>
<h4><a name="SWIG_chosen_unignore">5.4.7.5 Ignoring everything then
wrapping a few selected symbols</a></h4>
<p> Using the techniques described above it is possible to ignore
everything in a header and then selectively wrap a few chosen methods
or classes. For example, consider a header, <tt>myheader.h</tt> which
has many classes in it and just the one class called <tt>Star</tt> is
wanted within this header, the following approach could be taken:</p>
<div class="code">
<pre>
%ignore &quot;&quot;; // Ignore everything
// Unignore chosen class 'Star'
%rename(&quot;%s&quot;) Star;
// As the ignore everything will include the constructor, destructor, methods etc
// in the class, these have to be explicitly unignored too:
%rename(&quot;%s&quot;) Star::Star;
%rename(&quot;%s&quot;) Star::~Star;
%rename(&quot;%s&quot;) Star::shine; // named method
%include &quot;myheader.h&quot;
%rename(&quot;%s&quot;) &quot;&quot;; // Undo the %ignore
</pre>
</div>
<p> If <tt>Star</tt> was in the <tt>Galaxy</tt> namespace, you would
need to unignore the namespace, too, and add the namespace to all the
renames:</p>
<div class="code">
<pre>
%rename(&quot;%s&quot;) Galaxy;
%rename(&quot;%s&quot;) Galaxy::Star;
%rename(&quot;%s&quot;) Galaxy::Star::Star;
...
</pre>
</div>
<p> Another approach which might be more suitable as it does not require
naming all the methods in the chosen class is to begin by ignoring just
the classes. This does not add an explicit ignore to any members of the
class, so when the chosen class is unignored, all of its methods will
be wrapped.</p>
<div class="code">
<pre>
%rename($ignore, %$isclass) &quot;&quot;; // Only ignore all classes
%rename(&quot;%s&quot;) Star; // Unignore 'Star'
%include &quot;myheader.h&quot;
%rename(&quot;%s&quot;, %$isclass) &quot;&quot;; // Stop ignoring all classes
</pre>
</div>
<h3><a name="SWIG_default_args">5.4.8 Default/optional arguments</a></h3>
<p> SWIG supports default arguments in both C and C++ code. For example:</p>
<div class="code">
<pre>
int plot(double x, double y, int color=WHITE);
</pre>
</div>
<p> In this case, SWIG generates wrapper code where the default
arguments are optional in the target language. For example, this
function could be used in Tcl as follows :</p>
<div class="targetlang">
<pre>
% plot -3.4 7.5 # Use default value
% plot -3.4 7.5 10 # set color to 10 instead
</pre>
</div>
<p> Although the ISO C standard does not allow default arguments,
default arguments specified in a SWIG interface work with both C and
C++.</p>
<p><b> Note:</b> There is a subtle semantic issue concerning the use of
default arguments and the SWIG generated wrapper code. When default
arguments are used in C code, the default values are emitted into the
wrappers and the function is invoked with a full set of arguments. This
is different to when wrapping C++ where an overloaded wrapper method is
generated for each defaulted argument. Please refer to the section on <a
href="#SWIGPlus_default_args">default arguments</a> in the C++ chapter
for further details.</p>
<h3><a name="SWIG_nn30">5.4.9 Pointers to functions and callbacks</a></h3>
<p> Occasionally, a C library may include functions that expect to
receive pointers to functions--possibly to serve as callbacks. SWIG
provides full support for function pointers when the callback functions
are defined in C and not in the target language. For example, consider
a function like this:</p>
<div class="code">
<pre>
int binary_op(int a, int b, int (*op)(int, int));
</pre>
</div>
<p> When you first wrap something like this into an extension module,
you may find the function to be impossible to use. For instance, in
Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; def add(x, y):
... return x+y
...
&gt;&gt;&gt; binary_op(3, 4, add)
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
TypeError: Type error. Expected _p_f_int_int__int
&gt;&gt;&gt;
</pre>
</div>
<p> The reason for this error is that SWIG doesn't know how to map a
scripting language function into a C callback. However, existing C
functions can be used as arguments provided you install them as
constants. One way to do this is to use the <tt>%constant</tt>
directive like this:</p>
<div class="code">
<pre>
/* Function with a callback */
int binary_op(int a, int b, int (*op)(int, int));
/* Some callback functions */
%constant int add(int, int);
%constant int sub(int, int);
%constant int mul(int, int);
</pre>
</div>
<p> In this case, <tt>add</tt>, <tt>sub</tt>, and <tt>mul</tt> become
function pointer constants in the target scripting language. This
allows you to use them as follows:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; binary_op(3, 4, add)
7
&gt;&gt;&gt; binary_op(3, 4, mul)
12
&gt;&gt;&gt;
</pre>
</div>
<p> Unfortunately, by declaring the callback functions as constants,
they are no longer accessible as functions. For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; add(3, 4)
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
TypeError: object is not callable: '_ff020efc_p_f_int_int__int'
&gt;&gt;&gt;
</pre>
</div>
<p> If you want to make a function available as both a callback function
and a function, you can use the <tt>%callback</tt> and <tt>%nocallback</tt>
directives like this:</p>
<div class="code">
<pre>
/* Function with a callback */
int binary_op(int a, int b, int (*op)(int, int));
/* Some callback functions */
%callback(&quot;%s_cb&quot;);
int add(int, int);
int sub(int, int);
int mul(int, int);
%nocallback;
</pre>
</div>
<p> The argument to <tt>%callback</tt> is a printf-style format string
that specifies the naming convention for the callback constants (<tt>%s</tt>
gets replaced by the function name). The callback mode remains in
effect until it is explicitly disabled using <tt>%nocallback</tt>. When
you do this, the interface now works as follows:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; binary_op(3, 4, add_cb)
7
&gt;&gt;&gt; binary_op(3, 4, mul_cb)
12
&gt;&gt;&gt; add(3, 4)
7
&gt;&gt;&gt; mul(3, 4)
12
</pre>
</div>
<p> Notice that when the function is used as a callback, special names
such as <tt>add_cb</tt> are used instead. To call the function
normally, just use the original function name such as <tt>add()</tt>.</p>
<p> SWIG provides a number of extensions to standard C printf formatting
that may be useful in this context. For instance, the following
variation installs the callbacks as all upper case constants such as <tt>
ADD</tt>, <tt>SUB</tt>, and <tt>MUL</tt>:</p>
<div class="code">
<pre>
/* Some callback functions */
%callback(&quot;%(uppercase)s&quot;);
int add(int, int);
int sub(int, int);
int mul(int, int);
%nocallback;
</pre>
</div>
<p> A format string of <tt>&quot;%(lowercase)s&quot;</tt> converts all characters
to lower case. A string of <tt>&quot;%(title)s&quot;</tt> capitalizes the first
character and converts the rest to lower case.</p>
<p> And now, a final note about function pointer support. Although SWIG
does not normally allow callback functions to be written in the target
language, this can be accomplished with the use of typemaps and other
advanced SWIG features. See the <a href="#Typemaps">Typemaps chapter</a>
for more about typemaps and individual target language chapters for
more on callbacks. The 'director' feature can be used to make callbacks
from C/C++ into the target language, see <a href="#SWIGPlus_target_language_callbacks">
Callbacks to the target language</a>.</p>
<h2><a name="SWIG_nn31">5.5 Structures and unions</a></h2>
<p> This section describes the behavior of SWIG when processing ISO C
structures and union declarations. Extensions to handle C++ are
described in the next section.</p>
<p> ISO C has a separate tag name space in which the names of
structures, unions and enumerated types are put, which is separate from
the name space for ordinary identifiers (function names, object names,
typedef names, enumeration constants). For example, this is valid ISO C
because <tt>Foo</tt> the struct tag and <tt>Foo</tt> the function name
are in different name spaces:</p>
<div class="code">
<pre>
struct Foo {
int bar;
};
int Foo(void) { return 42; }
</pre>
</div>
<p> SWIG doesn't currently implement this separate tag name space and
for the above example you'll get:</p>
<div class="code">
<pre>
foo.i:5: Warning 302: Redefinition of identifier 'Foo' as Foo(void) ignored,
foo.i:1: Warning 302: previous definition of 'Foo'.
</pre>
</div>
<p> In practice this rarely actually causes problems, particular because
SWIG has special handling for <tt>typedef</tt> so cases such as this
work:</p>
<div class="code">
<pre>
typedef struct Foo {
int bar;
} Foo;
</pre>
</div>
<p> If SWIG encounters the definition of a structure or union, it
creates a set of accessor functions. Although SWIG does not need
structure definitions to build an interface, providing definitions
makes it possible to access structure members. The accessor functions
generated by SWIG simply take a pointer to an object and allow access
to an individual member. For example, the declaration :</p>
<div class="code">
<pre>
struct Vector {
double x, y, z;
};
</pre>
</div>
<p> gets transformed into the following set of accessor functions :</p>
<div class="code">
<pre>
double Vector_x_get(struct Vector *obj) {
return obj-&gt;x;
}
double Vector_y_get(struct Vector *obj) {
return obj-&gt;y;
}
double Vector_z_get(struct Vector *obj) {
return obj-&gt;z;
}
void Vector_x_set(struct Vector *obj, double value) {
obj-&gt;x = value;
}
void Vector_y_set(struct Vector *obj, double value) {
obj-&gt;y = value;
}
void Vector_z_set(struct Vector *obj, double value) {
obj-&gt;z = value;
}
</pre>
</div>
<p> In addition, SWIG creates default constructor and destructor
functions if none are defined in the interface. For example:</p>
<div class="code">
<pre>
struct Vector *new_Vector() {
return (Vector *) calloc(1, sizeof(struct Vector));
}
void delete_Vector(struct Vector *obj) {
free(obj);
}
</pre>
</div>
<p> Using these low-level accessor functions, an object can be minimally
manipulated from the target language using code like this:</p>
<div class="code">
<pre>
v = new_Vector()
Vector_x_set(v, 2)
Vector_y_set(v, 10)
Vector_z_set(v, -5)
...
delete_Vector(v)
</pre>
</div>
<p> However, most of SWIG's language modules also provide a high-level
interface that is more convenient. Keep reading.</p>
<h3><a name="SWIG_nn32">5.5.1 Typedef and structures</a></h3>
<p> SWIG supports the following construct which is quite common in C
programs :</p>
<div class="code">
<pre>
typedef struct {
double x, y, z;
} Vector;
</pre>
</div>
<p> When encountered, SWIG assumes that the name of the object is
`Vector' and creates accessor functions like before. The only
difference is that the use of <tt>typedef</tt> allows SWIG to drop the <tt>
struct</tt> keyword on its generated code. For example:</p>
<div class="code">
<pre>
double Vector_x_get(Vector *obj) {
return obj-&gt;x;
}
</pre>
</div>
<p> If two different names are used like this :</p>
<div class="code">
<pre>
typedef struct vector_struct {
double x, y, z;
} Vector;
</pre>
</div>
<p> the name <tt>Vector</tt> is used instead of <tt>vector_struct</tt>
since this is more typical C programming style. If declarations defined
later in the interface use the type <tt>struct vector_struct</tt>, SWIG
knows that this is the same as <tt>Vector</tt> and it generates the
appropriate type-checking code.</p>
<h3><a name="SWIG_nn33">5.5.2 Character strings and structures</a></h3>
<p> Structures involving character strings require some care. SWIG
assumes that all members of type <tt>char *</tt> have been dynamically
allocated using <tt>malloc()</tt> and that they are NULL-terminated
ASCII strings. When such a member is modified, the previous contents
will be released, and the new contents allocated. For example :</p>
<div class="code">
<pre>
%module mymodule
...
struct Foo {
char *name;
...
}
</pre>
</div>
<p> This results in the following accessor functions :</p>
<div class="code">
<pre>
char *Foo_name_get(Foo *obj) {
return Foo-&gt;name;
}
char *Foo_name_set(Foo *obj, char *c) {
free(obj-&gt;name);
obj-&gt;name = (char *) malloc(strlen(c)+1);
strcpy(obj-&gt;name, c);
return obj-&gt;name;
}
</pre>
</div>
<p> If this behavior differs from what you need in your applications,
the SWIG &quot;memberin&quot; typemap can be used to change it. See the typemaps
chapter for further details.</p>
<p> Note: If the <tt>-c++</tt> option is used, <tt>new</tt> and <tt>
delete</tt> are used to perform memory allocation.</p>
<h3><a name="SWIG_nn34">5.5.3 Array members</a></h3>
<p> Arrays may appear as the members of structures, but they will be
read-only. SWIG will write an accessor function that returns the
pointer to the first element of the array, but will not write a
function to change the contents of the array itself. When this
situation is detected, SWIG may generate a warning message such as the
following :</p>
<div class="shell">
<pre>
interface.i:116. Warning. Array member will be read-only
</pre>
</div>
<p> To eliminate the warning message, typemaps can be used, but this is
discussed in a later chapter. In many cases, the warning message is
harmless.</p>
<h3><a name="SWIG_structure_data_members">5.5.4 Structure data members</a>
</h3>
<p> Occasionally, a structure will contain data members that are
themselves structures. For example:</p>
<div class="code">
<pre>
typedef struct Foo {
int x;
} Foo;
typedef struct Bar {
int y;
Foo f; /* struct member */
} Bar;
</pre>
</div>
<p> When a structure member is wrapped, it is handled as a pointer,
unless the <tt>%naturalvar</tt> directive is used where it is handled
more like a C++ reference (see <a href="#SWIGPlus_member_data">C++
Member data</a>). The accessors to the member variable as a pointer are
effectively wrapped as follows:</p>
<div class="code">
<pre>
Foo *Bar_f_get(Bar *b) {
return &amp;b-&gt;f;
}
void Bar_f_set(Bar *b, Foo *value) {
b-&gt;f = *value;
}
</pre>
</div>
<p> The reasons for this are somewhat subtle but have to do with the
problem of modifying and accessing data inside the data member. For
example, suppose you wanted to modify the value of <tt>f.x</tt> of a <tt>
Bar</tt> object like this:</p>
<div class="code">
<pre>
Bar *b;
b-&gt;f.x = 37;
</pre>
</div>
<p> Translating this assignment to function calls (as would be used
inside the scripting language interface) results in the following code:</p>
<div class="code">
<pre>
Bar *b;
Foo_x_set(Bar_f_get(b), 37);
</pre>
</div>
<p> In this code, if the <tt>Bar_f_get()</tt> function were to return a <tt>
Foo</tt> instead of a <tt>Foo *</tt>, then the resulting modification
would be applied to a<em> copy</em> of <tt>f</tt> and not the data
member <tt>f</tt> itself. Clearly that's not what you want!</p>
<p> It should be noted that this transformation to pointers only occurs
if SWIG knows that a data member is a structure or class. For instance,
if you had a structure like this,</p>
<div class="code">
<pre>
struct Foo {
WORD w;
};
</pre>
</div>
<p> and nothing was known about <tt>WORD</tt>, then SWIG will generate
more normal accessor functions like this:</p>
<div class="code">
<pre>
WORD Foo_w_get(Foo *f) {
return f-&gt;w;
}
void Foo_w_set(FOO *f, WORD value) {
f-&gt;w = value;
}
</pre>
</div>
<p> If you have accessor methods that you want to use as attributes in
the target language, you can make them appear as data members using <a href="#Library_attributes">
attributes.i</a>.</p>
<p><b> Compatibility Note:</b> SWIG-1.3.11 and earlier releases
transformed all non-primitive member datatypes to pointers. Starting in
SWIG-1.3.12, this transformation<em> only</em> occurs if a datatype is
known to be a structure, class, or union. This is unlikely to break
existing code. However, if you need to tell SWIG that an undeclared
datatype is really a struct, simply use a forward struct declaration
such as <tt>&quot;struct Foo;&quot;</tt>.</p>
<h3><a name="SWIG_nn36">5.5.5 C constructors and destructors</a></h3>
<p> When wrapping structures, it is generally useful to have a mechanism
for creating and destroying objects. If you don't do anything, SWIG
will automatically generate functions for creating and destroying
objects using <tt>malloc()</tt> and <tt>free()</tt>. Note: the use of <tt>
malloc()</tt> only applies when SWIG is used on C code (i.e., when the <tt>
-c++</tt> option is<em> not</em> supplied on the command line). C++ is
handled differently.</p>
<p> If you don't want SWIG to generate default constructors for your
interfaces, you can use the <tt>%nodefaultctor</tt> directive or the <tt>
-nodefaultctor</tt> command line option. For example:</p>
<div class="shell">
<pre>
swig -nodefaultctor example.i
</pre>
</div>
<p> or</p>
<div class="code">
<pre>
%module foo
...
%nodefaultctor; // Don't create default constructors
... declarations ...
%clearnodefaultctor; // Re-enable default constructors
</pre>
</div>
<p> If you need more precise control, <tt>%nodefaultctor</tt> can
selectively target individual structure definitions. For example:</p>
<div class="code">
<pre>
%nodefaultctor Foo; // No default constructor for Foo
...
struct Foo { // No default constructor generated.
};
struct Bar { // Default constructor generated.
};
</pre>
</div>
<p> Since ignoring the implicit or default destructors most of the time
produces memory leaks, SWIG will always try to generate them. If
needed, however, you can selectively disable the generation of the
default/implicit destructor by using <tt>%nodefaultdtor</tt></p>
<div class="code">
<pre>
%nodefaultdtor Foo; // No default/implicit destructor for Foo
...
struct Foo { // No default destructor is generated.
};
struct Bar { // Default destructor generated.
};
</pre>
</div>
<p><b> Compatibility note:</b> Prior to SWIG-1.3.7, SWIG did not
generate default constructors or destructors unless you explicitly
turned them on. However, it appears that most users want to have
constructor and destructor functions so it has now been enabled as the
default behavior.</p>
<p><b> Note:</b> There is also the <tt>%nodefault</tt> directive, which
disables both the default or implicit destructor generation. This could
lead to memory leaks across the target languages, and it is highly
recommended you don't use them.</p>
<h3><a name="SWIG_adding_member_functions">5.5.6 Adding member functions
to C structures</a></h3>
<p> Most languages provide a mechanism for creating classes and
supporting object oriented programming. From a C standpoint, object
oriented programming really just boils down to the process of attaching
functions to structures. These functions normally operate on an
instance of the structure (or object). Although there is a natural
mapping of C++ to such a scheme, there is no direct mechanism for
utilizing it with C code. However, SWIG provides a special <tt>%extend</tt>
directive that makes it possible to attach methods to C structures for
purposes of building an object oriented interface. Suppose you have a C
header file with the following declaration :</p>
<div class="code">
<pre>
/* file : vector.h */
...
typedef struct Vector {
double x, y, z;
} Vector;
</pre>
</div>
<p> You can make a <tt>Vector</tt> look a lot like a class by writing a
SWIG interface like this:</p>
<div class="code">
<pre>
// file : vector.i
%module mymodule
%{
#include &quot;vector.h&quot;
%}
%include &quot;vector.h&quot; // Just grab original C header file
%extend Vector { // Attach these functions to struct Vector
Vector(double x, double y, double z) {
Vector *v;
v = (Vector *) malloc(sizeof(Vector));
v-&gt;x = x;
v-&gt;y = y;
v-&gt;z = z;
return v;
}
~Vector() {
free($self);
}
double magnitude() {
return sqrt($self-&gt;x*$self-&gt;x+$self-&gt;y*$self-&gt;y+$self-&gt;z*$self-&gt;z);
}
void print() {
printf(&quot;Vector [%g, %g, %g]\n&quot;, $self-&gt;x, $self-&gt;y, $self-&gt;z);
}
};
</pre>
</div>
<p> Note the usage of the <tt>$self</tt> special variable. Its usage is
identical to a C++ 'this' pointer and should be used whenever access to
the struct instance is required. Also note that C++ constructor and
destructor syntax has been used to simulate a constructor and
destructor, even for C code. There is one subtle difference to a normal
C++ constructor implementation though and that is although the
constructor declaration is as per a normal C++ constructor, the newly
constructed object must be returned<b> as if</b> the constructor
declaration had a return value, a <tt>Vector *</tt> in this case.</p>
<p> Now, when used with proxy classes in Python, you can do things like
this :</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; v = Vector(3, 4, 0) # Create a new vector
&gt;&gt;&gt; print(v.magnitude()) # Print magnitude
5.0
&gt;&gt;&gt; v.print() # Print it out
[ 3, 4, 0 ]
&gt;&gt;&gt; del v # Destroy it
</pre>
</div>
<p> The <tt>%extend</tt> directive can also be used inside the
definition of the Vector structure. For example:</p>
<div class="code">
<pre>
// file : vector.i
%module mymodule
%{
#include &quot;vector.h&quot;
%}
typedef struct Vector {
double x, y, z;
%extend {
Vector(double x, double y, double z) { ... }
~Vector() { ... }
...
}
} Vector;
</pre>
</div>
<p> Note that <tt>%extend</tt> can be used to access externally written
functions provided they follow the naming convention used in this
example :</p>
<div class="code">
<pre>
/* File : vector.c */
/* Vector methods */
#include &quot;vector.h&quot;
Vector *new_Vector(double x, double y, double z) {
Vector *v;
v = (Vector *) malloc(sizeof(Vector));
v-&gt;x = x;
v-&gt;y = y;
v-&gt;z = z;
return v;
}
void delete_Vector(Vector *v) {
free(v);
}
double Vector_magnitude(Vector *v) {
return sqrt(v-&gt;x*v-&gt;x+v-&gt;y*v-&gt;y+v-&gt;z*v-&gt;z);
}
// File : vector.i
// Interface file
%module mymodule
%{
#include &quot;vector.h&quot;
%}
typedef struct Vector {
double x, y, z;
%extend {
Vector(int, int, int); // This calls new_Vector()
~Vector(); // This calls delete_Vector()
double magnitude(); // This will call Vector_magnitude()
...
}
} Vector;
</pre>
</div>
<p> You'll also need to use these names if you want to directly call
methods added using <tt>%extend</tt> from other C/C++ code.</p>
<p> The name used for %extend should be the name of the struct and not
the name of any typedef to the struct. For example:</p>
<div class="code">
<pre>
typedef struct Integer {
int value;
} Int;
%extend Integer { ... } /* Correct name */
%extend Int { ... } /* Incorrect name */
struct Float {
float value;
};
typedef struct Float FloatValue;
%extend Float { ... } /* Correct name */
%extend FloatValue { ... } /* Incorrect name */
</pre>
</div>
<p> There is one exception to this rule and that is when the struct is
anonymously named such as:</p>
<div class="code">
<pre>
typedef struct {
double value;
} Double;
%extend Double { ... } /* Okay */
</pre>
</div>
<p> A little known feature of the <tt>%extend</tt> directive is that it
can also be used to add synthesized attributes or to modify the
behavior of existing data attributes. For example, suppose you wanted
to make <tt>magnitude</tt> a read-only attribute of <tt>Vector</tt>
instead of a method. To do this, you might write some code like this:</p>
<div class="code">
<pre>
// Add a new attribute to Vector
%extend Vector {
const double magnitude;
}
// Now supply the implementation of the Vector_magnitude_get function
%{
const double Vector_magnitude_get(Vector *v) {
return (const double) sqrt(v-&gt;x*v-&gt;x+v-&gt;y*v-&gt;y+v-&gt;z*v-&gt;z);
}
%}
</pre>
</div>
<p> Now, for all practical purposes, <tt>magnitude</tt> will appear like
an attribute of the object.</p>
<p> A similar technique can also be used to work with data members that
you want to process. For example, consider this interface:</p>
<div class="code">
<pre>
typedef struct Person {
char name[50];
...
} Person;
</pre>
</div>
<p> Say you wanted to ensure <tt>name</tt> was always upper case, you
can rewrite the interface as follows to ensure this occurs whenever a
name is read or written to:</p>
<div class="code">
<pre>
typedef struct Person {
%extend {
char name[50];
}
...
} Person;
%{
#include &lt;string.h&gt;
#include &lt;ctype.h&gt;
void make_upper(char *name) {
char *c;
for (c = name; *c; ++c)
*c = (char)toupper((int)*c);
}
/* Specific implementation of set/get functions forcing capitalization */
char *Person_name_get(Person *p) {
make_upper(p-&gt;name);
return p-&gt;name;
}
void Person_name_set(Person *p, char *val) {
strncpy(p-&gt;name, val, 50);
make_upper(p-&gt;name);
}
%}
</pre>
</div>
<p> Finally, it should be stressed that even though <tt>%extend</tt> can
be used to add new data members, these new members can not require the
allocation of additional storage in the object (e.g., their values must
be entirely synthesized from existing attributes of the structure or
obtained elsewhere).</p>
<h3><a name="SWIG_nested_structs">5.5.7 Nested structures</a></h3>
<p> Occasionally, a C program will involve structures like this :</p>
<div class="code">
<pre>
typedef struct Object {
int objtype;
union {
int ivalue;
double dvalue;
char *strvalue;
void *ptrvalue;
} intRep;
} Object;
</pre>
</div>
<p> When SWIG encounters this, it performs a structure splitting
operation that transforms the declaration into the equivalent of the
following:</p>
<div class="code">
<pre>
typedef union {
int ivalue;
double dvalue;
char *strvalue;
void *ptrvalue;
} Object_intRep;
typedef struct Object {
int objType;
Object_intRep intRep;
} Object;
</pre>
</div>
<p> SWIG will then create an <tt>Object_intRep</tt> structure for use
inside the interface file. Accessor functions will be created for both
structures. In this case, functions like this would be created :</p>
<div class="code">
<pre>
Object_intRep *Object_intRep_get(Object *o) {
return (Object_intRep *) &amp;o-&gt;intRep;
}
int Object_intRep_ivalue_get(Object_intRep *o) {
return o-&gt;ivalue;
}
int Object_intRep_ivalue_set(Object_intRep *o, int value) {
return (o-&gt;ivalue = value);
}
double Object_intRep_dvalue_get(Object_intRep *o) {
return o-&gt;dvalue;
}
... etc ...
</pre>
</div>
<p> Although this process is a little hairy, it works like you would
expect in the target scripting language--especially when proxy classes
are used. For instance, in Perl:</p>
<div class="targetlang">
<pre>
# Perl5 script for accessing nested member
$o = CreateObject(); # Create an object somehow
$o-&gt;{intRep}-&gt;{ivalue} = 7 # Change value of o.intRep.ivalue
</pre>
</div>
<p> If you have a lot of nested structure declarations, it is advisable
to double-check them after running SWIG. Although, there is a good
chance that they will work, you may have to modify the interface file
in certain cases.</p>
<p> Finally, note that nesting is handled differently in C++ mode, see <a
href="#SWIGPlus_nested_classes">Nested classes</a>.</p>
<h3><a name="SWIG_nn39">5.5.8 Other things to note about structure
wrapping</a></h3>
<p> SWIG doesn't care if the declaration of a structure in a <tt>.i</tt>
file exactly matches that used in the underlying C code (except in the
case of nested structures). For this reason, there are no problems
omitting problematic members or simply omitting the structure
definition altogether. If you are happy passing pointers around, this
can be done without ever giving SWIG a structure definition.</p>
<p> Starting with SWIG1.3, a number of improvements have been made to
SWIG's code generator. Specifically, even though structure access has
been described in terms of high-level accessor functions such as this,</p>
<div class="code">
<pre>
double Vector_x_get(Vector *v) {
return v-&gt;x;
}
</pre>
</div>
<p> most of the generated code is actually inlined directly into wrapper
functions. Therefore, no function <tt>Vector_x_get()</tt> actually
exists in the generated wrapper file. For example, when creating a Tcl
module, the following function is generated instead:</p>
<div class="code">
<pre>
static int
_wrap_Vector_x_get(ClientData clientData, Tcl_Interp *interp,
int objc, Tcl_Obj *const objv[]) {
struct Vector *arg1 ;
double result ;
if (SWIG_GetArgs(interp, objc, objv, &quot;p:Vector_x_get self &quot;, &amp;arg0,
SWIGTYPE_p_Vector) == TCL_ERROR)
return TCL_ERROR;
result = (double ) (arg1-&gt;x);
Tcl_SetObjResult(interp, Tcl_NewDoubleObj((double) result));
return TCL_OK;
}
</pre>
</div>
<p> The only exception to this rule are methods defined with <tt>%extend</tt>
. In this case, the added code is contained in a separate function.</p>
<p> Finally, it is important to note that most language modules may
choose to build a more advanced interface. Although you may never use
the low-level interface described here, most of SWIG's language modules
use it in some way or another.</p>
<h2><a name="SWIG_nn40">5.6 Code Insertion</a></h2>
<p> Sometimes it is necessary to insert special code into the resulting
wrapper file generated by SWIG. For example, you may want to include
additional C code to perform initialization or other operations. There
are four common ways to insert code, but it's useful to know how the
output of SWIG is structured first.</p>
<h3><a name="SWIG_nn41">5.6.1 The output of SWIG</a></h3>
<p> When SWIG creates its output C/C++ file, it is broken up into five
sections corresponding to runtime code, headers, wrapper functions, and
module initialization code (in that order).</p>
<ul>
<li><b>Begin section</b>.
<br> A placeholder for users to put code at the beginning of the C/C++
wrapper file. This is most often used to define preprocessor macros
that are used in later sections.</li>
<li><b>Runtime code</b>.
<br> This code is internal to SWIG and is used to include type-checking
and other support functions that are used by the rest of the module.</li>
<li><b>Header section</b>.
<br> This is user-defined support code that has been included by the <tt>
%{ ... %}</tt> directive. Usually this consists of header files and
other helper functions.</li>
<li><b>Wrapper code</b>.
<br> These are the wrappers generated automatically by SWIG.</li>
<li><b>Module initialization</b>.
<br> The function generated by SWIG to initialize the module upon
loading.</li>
</ul>
<h3><a name="SWIG_nn42">5.6.2 Code insertion blocks</a></h3>
<p> The <tt>%insert</tt> directive enables inserting blocks of code into
a given section of the generated code. It can be used in one of two
ways:</p>
<div class="code">
<pre>
%insert(&quot;section&quot;) &quot;filename&quot;
%insert(&quot;section&quot;) %{ ... %}
</pre>
</div>
<p> The first will dump the contents of the file in the given <tt>
filename</tt> into the named <tt>section</tt>. The second inserts the
code between the braces into the named <tt>section</tt>. For example,
the following adds code into the runtime section:</p>
<div class="code">
<pre>
%insert(&quot;runtime&quot;) %{
... code in runtime section ...
%}
</pre>
</div>
<p> There are the 5 sections, however, some target languages add in
additional sections and some of these result in code being generated
into a target language file instead of the C/C++ wrapper file. These
are documented when available in the target language chapters. Macros
named after the code sections are available as additional directives
and these macro directives are normally used instead of <tt>%insert</tt>
. For example, <tt>%runtime</tt> is used instead of <tt>
%insert(&quot;runtime&quot;)</tt>. The valid sections and order of the sections in
the generated C/C++ wrapper file is as shown:</p>
<div class="code">
<pre>
%begin %{
... code in begin section ...
%}
%runtime %{
... code in runtime section ...
%}
%header %{
... code in header section ...
%}
%wrapper %{
... code in wrapper section ...
%}
%init %{
... code in init section ...
%}
</pre>
</div>
<p> The bare <tt>%{ ... %}</tt> directive is a shortcut that is the same
as <tt>%header %{ ... %}</tt>.</p>
<p> The <tt>%begin</tt> section is effectively empty as it just contains
the SWIG banner by default. This section is provided as a way for users
to insert code at the top of the wrapper file before any other code is
generated. Everything in a code insertion block is copied verbatim into
the output file and is not parsed by SWIG. Most SWIG input files have
at least one such block to include header files and support C code.
Additional code blocks may be placed anywhere in a SWIG file as needed.</p>
<div class="code">
<pre>
%module mymodule
%{
#include &quot;my_header.h&quot;
%}
... Declare functions here
%{
void some_extra_function() {
...
}
%}
</pre>
</div>
<p> A common use for code blocks is to write &quot;helper&quot; functions. These
are functions that are used specifically for the purpose of building an
interface, but which are generally not visible to the normal C program.
For example :</p>
<div class="code">
<pre>
%{
/* Create a new vector */
static Vector *new_Vector() {
return (Vector *) malloc(sizeof(Vector));
}
%}
// Now wrap it
Vector *new_Vector();
</pre>
</div>
<h3><a name="SWIG_nn43">5.6.3 Inlined code blocks</a></h3>
<p> Since the process of writing helper functions is fairly common,
there is a special inlined form of code block that is used as follows :</p>
<div class="code">
<pre>
%inline %{
/* Create a new vector */
Vector *new_Vector() {
return (Vector *) malloc(sizeof(Vector));
}
%}
</pre>
</div>
<p> This is the same as writing:</p>
<div class="code">
<pre>
%{
/* Create a new vector */
Vector *new_Vector() {
return (Vector *) malloc(sizeof(Vector));
}
%}
/* Create a new vector */
Vector *new_Vector() {
return (Vector *) malloc(sizeof(Vector));
}
</pre>
</div>
<p> In other words, the <tt>%inline</tt> directive inserts all of the
code that follows verbatim into the header portion of an interface
file. The code is then parsed by both the SWIG preprocessor and parser.
Thus, the above example creates a new command <tt>new_Vector</tt> using
only one declaration. Since the code inside an <tt>%inline %{ ... %}</tt>
block is given to both the C compiler and SWIG, it is illegal to
include any SWIG directives inside a <tt>%{ ... %}</tt> block.</p>
<p><b> Note:</b> The usual SWIG C preprocessor rules apply to code in <tt>
%apply</tt> blocks when SWIG parses this code. For example, as mentioned
earlier, <a href="#SWIG_nn6">SWIG's C Preprocessor</a> does not follow <tt>
#include</tt> directives by default.</p>
<h3><a name="SWIG_nn44">5.6.4 Initialization blocks</a></h3>
<p> When code is included in the <tt>%init</tt> section, it is copied
directly into the module initialization function. For example, if you
needed to perform some extra initialization on module loading, you
could write this:</p>
<div class="code">
<pre>
%init %{
init_variables();
%}
</pre>
</div>
<p> Please note that some language backends (e.g. C# or Java) don't have
any initialization function, hence you should define a global object
performing the necessary initialization for them instead:</p>
<div class="code">
<pre>
%init %{
static struct MyInit { MyInit() { init_variables(); } } myInit;
%}
</pre>
</div>
<h2><a name="SWIG_nn45">5.7 An Interface Building Strategy</a></h2>
<p> This section describes the general approach for building interfaces
with SWIG. The specifics related to a particular scripting language are
found in later chapters.</p>
<h3><a name="SWIG_nn46">5.7.1 Preparing a C program for SWIG</a></h3>
<p> SWIG doesn't require modifications to your C code, but if you feed
it a collection of raw C header files or source code, the results might
not be what you expect---in fact, they might be awful. Here's a series
of steps you can follow to make an interface for a C program :</p>
<ul>
<li>Identify the functions that you want to wrap. It's probably not
necessary to access every single function of a C program--thus, a
little forethought can dramatically simplify the resulting scripting
language interface. C header files are a particularly good source for
finding things to wrap.</li>
<li>Create a new interface file to describe the scripting language
interface to your program.</li>
<li>Copy the appropriate declarations into the interface file or use
SWIG's <tt>%include</tt> directive to process an entire C source/header
file.</li>
<li>Make sure everything in the interface file uses ISO C/C++ syntax.</li>
<li>Make sure all necessary `<tt>typedef</tt>' declarations and
type-information is available in the interface file. In particular,
ensure that the type information is specified in the correct order as
required by a C/C++ compiler. Most importantly, define a type before it
is used! A C compiler will tell you if the full type information is not
available if it is needed, whereas SWIG will usually not warn or error
out as it is designed to work without full type information. However,
if type information is not specified correctly, the wrappers can be
sub-optimal and even result in uncompilable C/C++ code.</li>
<li>If your program has a main() function, you may need to rename it
(read on).</li>
<li>Run SWIG and compile.</li>
</ul>
<p> Although this may sound complicated, the process turns out to be
fairly easy once you get the hang of it.</p>
<p> In the process of building an interface, SWIG may encounter syntax
errors or other problems. The best way to deal with this is to simply
copy the offending code into a separate interface file and edit it.
However, the SWIG developers have worked very hard to improve the SWIG
parser--you should report parsing errors to the <a href="https://www.swig.org/mail.html">
swig-devel mailing list</a> or to the <a href="https://www.swig.org/bugs.html">
SWIG bug tracker</a>.</p>
<h3><a name="SWIG_nn47">5.7.2 The SWIG interface file</a></h3>
<p> The preferred method of using SWIG is to generate a separate
interface file. Suppose you have the following C header file :</p>
<div class="code">
<pre>
/* File : header.h */
#include &lt;stdio.h&gt;
#include &lt;math.h&gt;
extern int foo(double);
extern double bar(int, int);
extern void dump(FILE *f);
</pre>
</div>
<p> A typical SWIG interface file for this header file would look like
the following :</p>
<div class="code">
<pre>
/* File : interface.i */
%module mymodule
%{
#include &quot;header.h&quot;
%}
extern int foo(double);
extern double bar(int, int);
extern void dump(FILE *f);
</pre>
</div>
<p> Of course, in this case, our header file is pretty simple so we
could use a simpler approach and use an interface file like this:</p>
<div class="code">
<pre>
/* File : interface.i */
%module mymodule
%{
#include &quot;header.h&quot;
%}
%include &quot;header.h&quot;
</pre>
</div>
<p> The main advantage of this approach is minimal maintenance of an
interface file for when the header file changes in the future. In more
complex projects, an interface file containing numerous <tt>%include</tt>
and <tt>#include</tt> statements like this is one of the most common
approaches to interface file design due to lower maintenance overhead.</p>
<h3><a name="SWIG_nn48">5.7.3 Why use separate interface files?</a></h3>
<p> Although SWIG can parse many header files, it is more common to
write a special <tt>.i</tt> file defining the interface to a package.
There are several reasons why you might want to do this:</p>
<ul>
<li>It is rarely necessary to access every single function in a large
package. Many C functions might have little or no use in a scripted
environment. Therefore, why wrap them?</li>
<li>Separate interface files provide an opportunity to provide more
precise rules about how an interface is to be constructed.</li>
<li>Interface files can provide more structure and organization.</li>
<li>SWIG can't parse certain definitions that appear in header files.
Having a separate file allows you to eliminate or work around these
problems.</li>
<li>Interface files provide a more precise definition of what the
interface is. Users wanting to extend the system can go to the
interface file and immediately see what is available without having to
dig it out of header files.</li>
</ul>
<h3><a name="SWIG_nn49">5.7.4 Getting the right header files</a></h3>
<p> Sometimes, it is necessary to use certain header files in order for
the code generated by SWIG to compile properly. Make sure you include
certain header files by using a <tt>%{ %}</tt> block like this:</p>
<div class="code">
<pre>
%module graphics
%{
#include &lt;GL/gl.h&gt;
#include &lt;GL/glu.h&gt;
%}
// Put the rest of the declarations here
...
</pre>
</div>
<h3><a name="SWIG_nn50">5.7.5 What to do with main()</a></h3>
<p> If your program defines a <tt>main()</tt> function, you may need to
get rid of it or rename it in order to use a scripting language. Most
scripting languages define their own <tt>main()</tt> procedure that is
called instead. <tt>main()</tt> also makes no sense when working with
dynamic loading. There are a few approaches to solving the <tt>main()</tt>
conflict :</p>
<ul>
<li>Get rid of <tt>main()</tt> entirely.</li>
<li>Rename <tt>main()</tt> to something else. You can do this by
compiling your C program with an option like <tt>-Dmain=oldmain</tt>.</li>
<li>Use conditional compilation to only include <tt>main()</tt> when not
using a scripting language.</li>
</ul>
<p> Getting rid of <tt>main()</tt> may cause potential initialization
problems of a program. To handle this problem, you may consider writing
a special function called <tt>program_init()</tt> that initializes your
program upon startup. This function could then be called either from
the scripting language as the first operation, or when the SWIG
generated module is loaded.</p>
<p> As a general note, many C programs only use the <tt>main()</tt>
function to parse command line options and to set parameters. However,
by using a scripting language, you are probably trying to create a
program that is more interactive. In many cases, the old <tt>main()</tt>
program can be completely replaced by a Perl, Python, or Tcl script.</p>
<p><b> Note:</b> In some cases, you might be inclined to create a
scripting language wrapper for <tt>main()</tt>. If you do this, the
compilation will probably work and your module might even load
correctly. The only trouble is that when you call your <tt>main()</tt>
wrapper, you will find that it actually invokes the <tt>main()</tt> of
the scripting language interpreter itself! This behavior is a side
effect of the symbol binding mechanism used in the dynamic linker. The
bottom line: don't do this.</p>
<HR NOSHADE>
<h1><a name="SWIGPlus">6 SWIG and C++</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#SWIGPlus_nn2">Comments on C++ Wrapping</a></li>
<li><a href="#SWIGPlus_nn3">Approach</a></li>
<li><a href="#SWIGPlus_nn4">Supported C++ features</a></li>
<li><a href="#SWIGPlus_nn5">Command line options and compilation</a></li>
<li><a href="#SWIGPlus_nn38">Proxy classes</a>
<ul>
<li><a href="#SWIGPlus_nn39">Construction of proxy classes</a></li>
<li><a href="#SWIGPlus_nn40">Resource management in proxies</a></li>
<li><a href="#SWIGPlus_nn41">Language specific details</a></li>
</ul>
</li>
<li><a href="#SWIGPlus_nn6">Simple C++ wrapping</a>
<ul>
<li><a href="#SWIGPlus_nn7">Constructors and destructors</a></li>
<li><a href="#SWIGPlus_nn8">Default constructors, copy constructors and
implicit destructors</a></li>
<li><a href="#SWIGPlus_nn9">When constructor wrappers aren't created</a></li>
<li><a href="#SWIGPlus_nn10">Copy constructors</a></li>
<li><a href="#SWIGPlus_nn11">Member functions</a></li>
<li><a href="#SWIGPlus_nn12">Static members</a></li>
<li><a href="#SWIGPlus_member_data">Member data</a></li>
</ul>
</li>
<li><a href="#SWIGPlus_nn15">Protection</a></li>
<li><a href="#SWIGPlus_nn16">Enums and constants</a></li>
<li><a href="#SWIGPlus_nn17">Friends</a>
<ul>
<li><a href="#SWIGPlus_friend_classes">Friend classes</a></li>
<li><a href="#SWIGPlus_friend_function_definitions">Friend function
definitions</a></li>
<li><a href="#SWIGPlus_friend_function_declarations">Friend function
declarations</a></li>
<li><a href="#SWIGPlus_friends_unqualified">Unqualified friend functions</a>
</li>
</ul>
</li>
<li><a href="#SWIGPlus_nn18">References and pointers</a></li>
<li><a href="#SWIGPlus_nn19">Pass and return by value</a></li>
<li><a href="#SWIGPlus_nn20">Inheritance</a></li>
<li><a href="#SWIGPlus_nn21">A brief discussion of multiple inheritance,
pointers, and type checking</a></li>
<li><a href="#SWIGPlus_default_args">Default arguments</a></li>
<li><a href="#SWIGPlus_overloaded_methods">Overloaded functions and
methods</a>
<ul>
<li><a href="#SWIGPlus_nn24">Dispatch function generation</a></li>
<li><a href="#SWIGPlus_nn25">Ambiguity in overloading</a></li>
<li><a href="#SWIGPlus_ambiguity_resolution_renaming">Renaming and
ambiguity resolution</a></li>
<li><a href="#SWIGPlus_nn27">Comments on overloading</a></li>
</ul>
</li>
<li><a href="#SWIGPlus_nn28">Overloaded operators</a></li>
<li><a href="#SWIGPlus_class_extension">Class extension</a>
<ul>
<li><a href="#SWIGPlus_replacing_methods">Replacing class methods</a></li>
</ul>
</li>
<li><a href="#SWIGPlus_nn30">Templates</a>
<ul>
<li><a href="#SWIGPlus_template_directive">The %template directive</a></li>
<li><a href="#SWIGPlus_template_functions">Function templates</a></li>
<li><a href="#SWIGPlus_template_classes">Default template arguments</a></li>
<li><a href="#SWIGPlus_template_class_inheritance">Template base classes</a>
</li>
<li><a href="#SWIGPlus_template_empty">Empty template instantiation</a></li>
<li><a href="#SWIGPlus_template_specialization">Template specialization</a>
</li>
<li><a href="#SWIGPlus_template_member">Member templates</a></li>
<li><a href="#SWIGPlus_template_scoping">Scoping and templates</a></li>
<li><a href="#SWIGPlus_template_more">More on templates</a></li>
</ul>
</li>
<li><a href="#SWIGPlus_namespaces">Namespaces</a>
<ul>
<li><a href="#SWIGPlus_nspace">The nspace feature for namespaces</a>
<ul>
<li><a href="#SWIGPlus_nspace_feature_flag">%nspace for mirroring
namespace hierarchies</a></li>
<li><a href="#SWIGPlus_nspacemove">%nspacemove for modifying namespace
hierarchies</a></li>
<li><a href="#SWIGPlus_nspace_more">More about the nspace feature</a></li>
</ul>
</li>
</ul>
</li>
<li><a href="#SWIGPlus_renaming_templated_types_namespaces">Renaming
templated types in namespaces</a></li>
<li><a href="#SWIGPlus_exception_specifications">Exception
specifications</a></li>
<li><a href="#SWIGPlus_catches">Exception handling with %catches</a></li>
<li><a href="#SWIGPlus_nn33">Pointers to Members</a></li>
<li><a href="#SWIGPlus_smart_pointers">Smart pointers and operator-&gt;()</a>
</li>
<li><a href="#SWIGPlus_ref_unref">C++ reference counted objects -
ref/unref feature</a></li>
<li><a href="#SWIGPlus_nn35">Using declarations and inheritance</a></li>
<li><a href="#SWIGPlus_nested_classes">Nested classes</a></li>
<li><a href="#SWIGPlus_const">A brief rant about const-correctness</a></li>
<li><a href="#SWIGPlus_target_language_callbacks">Callbacks to the
target language</a>
<ul>
<li><a href="#SWIGPlus_director_classes_introduction">Introduction to
director classes</a></li>
<li><a href="#SWIGPlus_directors_for_function_pointers">Using directors
and target language callbacks</a></li>
</ul>
</li>
<li><a href="#SWIGPlus_nn42">Where to go for more information</a></li>
</ul>
</div>
<!-- INDEX -->
<p> This chapter describes SWIG's support for wrapping C++. It is mostly
concerned about C++ as defined by the C++ 98 and 03 standards. For
additions to the original C++ standard, please read the <a href="#CPlusPlus11">
SWIG and C++11</a>, <a href="#CPlusPlus14">SWIG and C++14</a>, <a href="#CPlusPlus17">
SWIG and C++17</a> and <a href="#CPlusPlus20">SWIG and C++20</a>
chapters. As a prerequisite, you should first read the chapter <a href="#SWIG">
SWIG Basics</a> to see how SWIG wraps ISO C. Support for C++ builds upon
ISO C wrapping and that material will be useful in understanding this
chapter.</p>
<h2><a name="SWIGPlus_nn2">6.1 Comments on C++ Wrapping</a></h2>
<p> Because of its complexity and the fact that C++ can be difficult to
integrate with itself let alone other languages, SWIG only provides
support for a subset of C++ features. Fortunately, this is now a rather
large subset.</p>
<p> In part, the problem with C++ wrapping is that there is no
semantically obvious (or automatic ) way to map many of its advanced
features into other languages. As a simple example, consider the
problem of wrapping C++ multiple inheritance to a target language with
no such support. Similarly, the use of overloaded operators and
overloaded functions can be problematic when no such capability exists
in a target language.</p>
<p> A more subtle issue with C++ has to do with the way that some C++
programmers think about programming libraries. In the world of SWIG,
you are really trying to create binary-level software components for
use in other languages. In order for this to work, a &quot;component&quot; has to
contain real executable instructions and there has to be some kind of
binary linking mechanism for accessing its functionality. In contrast,
C++ has increasingly relied upon generic programming and templates for
much of its functionality. Although templates are a powerful feature,
they are largely orthogonal to the whole notion of binary components
and libraries. For example, an STL <tt>vector</tt> does not define any
kind of binary object for which SWIG can just create a wrapper. To
further complicate matters, these libraries often utilize a lot of
behind the scenes magic in which the semantics of seemingly basic
operations (e.g., pointer dereferencing, procedure call, etc.) can be
changed in dramatic and sometimes non-obvious ways. Although this
&quot;magic&quot; may present few problems in a C++-only universe, it greatly
complicates the problem of crossing language boundaries and provides
many opportunities to shoot yourself in the foot. You will just have to
be careful.</p>
<h2><a name="SWIGPlus_nn3">6.2 Approach</a></h2>
<p> To wrap C++, SWIG uses a layered approach to code generation. At the
lowest level, SWIG generates a collection of procedural ISO C style
wrappers. These wrappers take care of basic type conversion, type
checking, error handling, and other low-level details of the C++
binding. These wrappers are also sufficient to bind C++ into any target
language that supports built-in procedures. In some sense, you might
view this layer of wrapping as providing a C library interface to C++.
On top of the low-level procedural (flattened) interface, SWIG
generates proxy classes that provide a natural object-oriented (OO)
interface to the underlying code. The proxy classes are typically
written in the target language itself. For instance, in Python, a real
Python class is used to provide a wrapper around the underlying C++
object.</p>
<p> It is important to emphasize that SWIG takes a deliberately
conservative and non-intrusive approach to C++ wrapping. SWIG does not
encapsulate C++ classes inside a special C++ adaptor, it does not rely
upon templates, nor does it add in additional C++ inheritance when
generating wrappers. The last thing that most C++ programs need is even
more compiler magic. Therefore, SWIG tries to maintain a very strict
and clean separation between the implementation of your C++ application
and the resulting wrapper code. You might say that SWIG has been
written to follow the principle of least surprise--it does not play
sneaky tricks with the C++ type system, it doesn't mess with your class
hierarchies, and it doesn't introduce new semantics. Although this
approach might not provide the most seamless integration with C++, it
is safe, simple, portable, and debuggable.</p>
<p> This approach results in generated code that uses just the most
basic of C and C++ features, no matter how complex the input C++ is.
SWIG will not force a newer standard of C++ to be required as the
generated code complies with ISO standard C++98. However, there are a
few features in later C++ standards that can be advantageous to use.
For example, C++11 move semantics are available to provide performance
improvements. So in some instances, SWIG will use features from newer
C++ standards provided they are available. This is made possible via
the <tt>__cplusplus</tt> macro, which is set by the C++ compiler to an
appropriate standards defined value. This means that if your compiler
is configured to support say the C++11 standard, then there may be some
C++11 benefits available in the generated code, such as C++11 move
semantics.</p>
<p> Some of this chapter focuses on the low-level procedural interface
to C++ that is used as the foundation for all language modules. Keep in
mind that the target languages also provide the high-level OO interface
via proxy classes. More detailed coverage can be found in the
documentation for each target language.</p>
<h2><a name="SWIGPlus_nn4">6.3 Supported C++ features</a></h2>
<p> SWIG currently supports most C++ features including the following:</p>
<ul>
<li>Classes</li>
<li>Constructors and destructors</li>
<li>Virtual functions</li>
<li>Public inheritance (including multiple inheritance)</li>
<li>Static functions</li>
<li>Function and method overloading</li>
<li>Operator overloading for many standard operators</li>
<li>References</li>
<li>Templates (including specialization and member templates)</li>
<li>Pointers to members</li>
<li>Namespaces</li>
<li>Default parameters</li>
<li>Smart pointers</li>
</ul>
<p> The following C++ features are not currently supported:</p>
<ul>
<li>Overloaded versions of certain operators (new, delete, etc.)</li>
</ul>
<p> As a rule of thumb, SWIG should not be used on raw C++ source files,
use header files only.</p>
<p> SWIG's C++ support is an ongoing project so some of these
limitations may be lifted in future releases. However, we make no
promises. Also, submitting a bug report is a very good way to get
problems fixed (wink).</p>
<h2><a name="SWIGPlus_nn5">6.4 Command line options and compilation</a></h2>
<p> When wrapping C++ code, it is critical that SWIG be called with the
`<tt>-c++</tt>' option. This changes the way a number of critical
features such as memory management are handled. It also enables the
recognition of C++ keywords. Without the <tt>-c++</tt> flag, SWIG will
either issue a warning or a large number of syntax errors if it
encounters C++ code in an interface file.</p>
<p> When compiling and linking the resulting wrapper file, it is normal
to use the C++ compiler. For example:</p>
<div class="shell">
<pre>
$ swig -c++ -tcl example.i
$ c++ -fPIC -c example_wrap.cxx
$ c++ example_wrap.o $(OBJS) -o example.so
</pre>
</div>
<p> Unfortunately, the process varies slightly on each platform. Make
sure you refer to the documentation on each target language for further
details. The SWIG Wiki also has further details.</p>
<b> Compatibility Note:</b> Early versions of SWIG generated just a
flattened low-level C style API to C++ classes by default. The <tt>
-noproxy</tt> commandline option is recognised by some target languages
and will generate just this interface as in earlier versions.
<h2><a name="SWIGPlus_nn38">6.5 Proxy classes</a></h2>
<p> In order to provide a natural mapping from C++ classes to the target
language classes, SWIG's target languages mostly wrap C++ classes with
special proxy classes. These proxy classes are typically implemented in
the target language itself. For example, if you're building a Python
module, each C++ class is wrapped by a Python proxy class. Or if you're
building a Java module, each C++ class is wrapped by a Java proxy
class.</p>
<h3><a name="SWIGPlus_nn39">6.5.1 Construction of proxy classes</a></h3>
<p> Proxy classes are always constructed as an extra layer of wrapping
that uses low-level accessor functions. To illustrate, suppose you had
a C++ class like this:</p>
<div class="code">
<pre>
class Foo {
public:
Foo();
~Foo();
int bar(int x);
int x;
};
</pre>
</div>
<p> Using C++ as pseudocode, a proxy class looks something like this:</p>
<div class="code">
<pre>
class FooProxy {
private:
Foo *self;
public:
FooProxy() {
self = new_Foo();
}
~FooProxy() {
delete_Foo(self);
}
int bar(int x) {
return Foo_bar(self, x);
}
int x_get() {
return Foo_x_get(self);
}
void x_set(int x) {
Foo_x_set(self, x);
}
};
</pre>
</div>
<p> Of course, always keep in mind that the real proxy class is written
in the target language. For example, in Python, the proxy might look
roughly like this:</p>
<div class="targetlang">
<pre>
class Foo:
def __init__(self):
self.this = new_Foo()
def __del__(self):
delete_Foo(self.this)
def bar(self, x):
return Foo_bar(self.this, x)
def __getattr__(self, name):
if name == 'x':
return Foo_x_get(self.this)
...
def __setattr__(self, name, value):
if name == 'x':
Foo_x_set(self.this, value)
...
</pre>
</div>
<p> Again, it's important to emphasize that the low-level accessor
functions are always used by the proxy classes. Whenever possible,
proxies try to take advantage of language features that are similar to
C++. This might include operator overloading, exception handling, and
other features.</p>
<h3><a name="SWIGPlus_nn40">6.5.2 Resource management in proxies</a></h3>
<p> A major issue with proxies concerns the memory management of wrapped
objects. Consider the following C++ code:</p>
<div class="code">
<pre>
class Foo {
public:
Foo();
~Foo();
int bar(int x);
int x;
};
class Spam {
public:
Foo *value;
...
};
</pre>
</div>
<p> Consider some script code that uses these classes:</p>
<div class="targetlang">
<pre>
f = Foo() # Creates a new Foo
s = Spam() # Creates a new Spam
s.value = f # Stores a reference to f inside s
g = s.value # Returns stored reference
g = 4 # Reassign g to some other value
del f # Destroy f
</pre>
</div>
<p> Now, ponder the resulting memory management issues. When objects are
created in the script, the objects are wrapped by newly created proxy
classes. That is, there is both a new proxy class instance and a new
instance of the underlying C++ class. In this example, both <tt>f</tt>
and <tt>s</tt> are created in this way. However, the statement <tt>
s.value</tt> is rather curious---when executed, a pointer to <tt>f</tt>
is stored inside another object. This means that the scripting proxy
class<em> AND</em> another C++ class share a reference to the same
object. To make matters even more interesting, consider the statement <tt>
g = s.value</tt>. When executed, this creates a new proxy class <tt>g</tt>
that provides a wrapper around the C++ object stored in <tt>s.value</tt>
. In general, there is no way to know where this object came from---it
could have been created by the script, but it could also have been
generated internally. In this particular example, the assignment of <tt>
g</tt> results in a second proxy class for <tt>f</tt>. In other words, a
reference to <tt>f</tt> is now shared by two proxy classes<em> and</em>
a C++ class.</p>
<p> Finally, consider what happens when objects are destroyed. In the
statement, <tt>g=4</tt>, the variable <tt>g</tt> is reassigned. In many
languages, this makes the old value of <tt>g</tt> available for garbage
collection. Therefore, this causes one of the proxy classes to be
destroyed. Later on, the statement <tt>del f</tt> destroys the other
proxy class. Of course, there is still a reference to the original
object stored inside another C++ object. What happens to it? Is the
object still valid?</p>
<p> To deal with memory management problems, proxy classes provide an
API for controlling ownership. In C++ pseudocode, ownership control
might look roughly like this:</p>
<div class="code">
<pre>
class FooProxy {
public:
Foo *self;
int thisown;
FooProxy() {
self = new_Foo();
thisown = 1; // Newly created object
}
~FooProxy() {
if (thisown) delete_Foo(self);
}
...
// Ownership control API
void disown() {
thisown = 0;
}
void acquire() {
thisown = 1;
}
};
class FooPtrProxy: public FooProxy {
public:
FooPtrProxy(Foo *s) {
self = s;
thisown = 0;
}
};
class SpamProxy {
...
FooProxy *value_get() {
return FooPtrProxy(Spam_value_get(self));
}
void value_set(FooProxy *v) {
Spam_value_set(self, v-&gt;self);
v-&gt;disown();
}
...
};
</pre>
</div>
<p> Looking at this code, there are a few central features:</p>
<ul>
<li>Each proxy class keeps an extra flag to indicate ownership. C++
objects are only destroyed if the ownership flag is set.</li>
<li>When new objects are created in the target language, the ownership
flag is set.</li>
<li>When a reference to an internal C++ object is returned, it is
wrapped by a proxy class, but the proxy class does not have ownership.</li>
<li>In certain cases, ownership is adjusted. For instance, when a value
is assigned to the member of a class, ownership is lost.</li>
<li>Manual ownership control is provided by special <tt>disown()</tt>
and <tt>acquire()</tt> methods.</li>
</ul>
<p> Given the tricky nature of C++ memory management, it is impossible
for proxy classes to automatically handle every possible memory
management problem. However, proxies do provide a mechanism for manual
control that can be used (if necessary) to address some of the more
tricky memory management problems.</p>
<h3><a name="SWIGPlus_nn41">6.5.3 Language specific details</a></h3>
<p> Language specific details on proxy classes are contained in the
chapters describing each target language. This chapter has merely
introduced the topic in a very general way.</p>
<h2><a name="SWIGPlus_nn6">6.6 Simple C++ wrapping</a></h2>
<p> The following code shows a SWIG interface file for a simple C++
class.</p>
<div class="code">
<pre>
%module list
%{
#include &quot;list.h&quot;
%}
// Very simple C++ example for linked list
class List {
public:
List();
~List();
int search(char *value);
void insert(char *);
void remove(char *);
char *get(int n);
int length;
static void print(List *l);
};
</pre>
</div>
<p> To generate wrappers for this class, SWIG first reduces the class to
a collection of low-level C-style accessor functions which are then
used by the proxy classes.</p>
<h3><a name="SWIGPlus_nn7">6.6.1 Constructors and destructors</a></h3>
<p> C++ constructors and destructors are translated into accessor
functions such as the following :</p>
<div class="code">
<pre>
List * new_List(void) {
return new List;
}
void delete_List(List *l) {
delete l;
}
</pre>
</div>
<h3><a name="SWIGPlus_nn8">6.6.2 Default constructors, copy constructors
and implicit destructors</a></h3>
<p> Following the C++ rules for implicit constructor and destructors,
SWIG will automatically assume there is one even when they are not
explicitly declared in the class interface.</p>
<p> In general then:</p>
<ul>
<li> If a C++ class does not declare any explicit constructor, SWIG will
automatically generate a wrapper for one.</li>
<li> If a C++ class does not declare an explicit copy constructor, SWIG
will automatically generate a wrapper for one if <tt>%copyctor</tt> is
used.</li>
<li> If a C++ class does not declare an explicit destructor, SWIG will
automatically generate a wrapper for one.</li>
</ul>
<p> And as in C++, a few rules that alters the previous behavior:</p>
<ul>
<li>A default constructor is not created if a class already defines a
constructor with arguments.</li>
<li>Default constructors are not generated for classes with pure virtual
methods or for classes that inherit from an abstract class, but don't
provide definitions for all of the pure methods.</li>
<li>A default constructor is not created unless all base classes support
a default constructor.</li>
<li>Default constructors and implicit destructors are not created if a
class defines them in a <tt>private</tt> or <tt>protected</tt> section.</li>
<li>Default constructors and implicit destructors are not created if any
base class defines a non-public default constructor or destructor.</li>
</ul>
<p> SWIG should never generate a default constructor, copy constructor
or default destructor wrapper for a class in which it is illegal to do
so. In some cases, however, it could be necessary (if the complete
class declaration is not visible from SWIG, and one of the above rules
is violated) or desired (to reduce the size of the final interface) by
manually disabling the implicit constructor/destructor generation.</p>
<p> To manually disable these, the <tt>%nodefaultctor</tt> and <tt>
%nodefaultdtor</tt> <a href="#Customization_feature_flags">feature flag</a>
directives can be used. Note that these directives only affects the
implicit generation, and they have no effect if the default/copy
constructors or destructor are explicitly declared in the class
interface.</p>
<p> For example:</p>
<div class="code">
<pre>
%nodefaultctor Foo; // Disable the default constructor for class Foo.
class Foo { // No default constructor is generated, unless one is declared
...
};
class Bar { // A default constructor is generated, if possible
...
};
</pre>
</div>
<p> The directive <tt>%nodefaultctor</tt> can also be applied
&quot;globally&quot;, as in:</p>
<div class="code">
<pre>
%nodefaultctor; // Disable creation of default constructors
class Foo { // No default constructor is generated, unless one is declared
...
};
class Bar {
public:
Bar(); // The default constructor is generated, since one is declared
};
%clearnodefaultctor; // Enable the creation of default constructors again
</pre>
</div>
<p> The corresponding <tt>%nodefaultdtor</tt> directive can be used to
disable the generation of the default or implicit destructor, if
needed. Be aware, however, that this could lead to memory leaks in the
target language. Hence, it is recommended to use this directive only in
well known cases. For example:</p>
<div class="code">
<pre>
%nodefaultdtor Foo; // Disable the implicit/default destructor for class Foo.
class Foo { // No destructor is generated, unless one is declared
...
};
</pre>
</div>
<p><b> Compatibility Note:</b> The generation of default
constructors/implicit destructors was made the default behavior in SWIG
1.3.7. This may break certain older modules, but the old behavior can
be easily restored using <tt>%nodefault</tt>. Furthermore, in order for
SWIG to properly generate (or not generate) default constructors, it
must be able to gather information from both the <tt>private</tt> and <tt>
protected</tt> sections (specifically, it needs to know if a private or
protected constructor/destructor is defined). In older versions of
SWIG, it was fairly common to simply remove or comment out the private
and protected sections of a class due to parser limitations. However,
this removal may now cause SWIG to erroneously generate constructors
for classes that define a constructor in those sections. Consider
restoring those sections in the interface or using <tt>%nodefault</tt>
to fix the problem.</p>
<p><b> Note:</b> The <tt>%nodefault</tt> directive described above,
which disables both the default constructor and the implicit
destructors, could lead to memory leaks, and so it is strongly
recommended to not use it.</p>
<h3><a name="SWIGPlus_nn9">6.6.3 When constructor wrappers aren't
created</a></h3>
<p> If a class defines a constructor, SWIG normally tries to generate a
wrapper for it. However, SWIG will not generate a constructor wrapper
if it thinks that it will result in illegal wrapper code. There are
really two cases where this might show up.</p>
<p> First, SWIG won't generate wrappers for protected or private
constructors. For example:</p>
<div class="code">
<pre>
class Foo {
protected:
Foo(); // Not wrapped.
public:
...
};
</pre>
</div>
<p> Next, SWIG won't generate wrappers for a class if it appears to be
abstract--that is, it has undefined pure virtual methods. Here are some
examples:</p>
<div class="code">
<pre>
class Bar {
public:
Bar(); // Not wrapped. Bar is abstract.
virtual void spam(void) = 0;
};
class Grok : public Bar {
public:
Grok(); // Not wrapped. No implementation of abstract spam().
};
</pre>
</div>
<p> Some users are surprised (or confused) to find missing constructor
wrappers in their interfaces. In almost all cases, this is caused when
classes are determined to be abstract. To see if this is the case, run
SWIG with all of its warnings turned on:</p>
<div class="shell">
<pre>
% swig -Wall -python module.i
</pre>
</div>
<p> In this mode, SWIG will issue a warning for all abstract classes. It
is possible to force a class to be non-abstract using this:</p>
<div class="code">
<pre>
%feature(&quot;notabstract&quot;) Foo;
class Foo : public Bar {
public:
Foo(); // Generated no matter what---not abstract.
...
};
</pre>
</div>
<p> More information about <tt>%feature</tt> can be found in the <a href="#Customization">
Customization features</a> chapter.</p>
<h3><a name="SWIGPlus_nn10">6.6.4 Copy constructors</a></h3>
<p> If a class defines more than one constructor, its behavior depends
on the capabilities of the target language. If overloading is
supported, the copy constructor is accessible using the normal
constructor function. For example, if you have this:</p>
<div class="code">
<pre>
class List {
public:
List();
List(const List &amp;); // Copy constructor
...
};
</pre>
</div>
<p> then the copy constructor can be used as follows:</p>
<div class="targetlang">
<pre>
x = List() # Create a list
y = List(x) # Copy list x
</pre>
</div>
<p> If the target language does not support overloading, then the copy
constructor is available through a special function like this:</p>
<div class="code">
<pre>
List *copy_List(List *f) {
return new List(*f);
}
</pre>
</div>
<p><b> Note:</b> For a class <tt>X</tt>, SWIG only treats a constructor
as a copy constructor if it can be applied to an object of type <tt>X</tt>
or <tt>X *</tt>. If more than one copy constructor is defined, only the
first definition that appears is used as the copy constructor--other
definitions will result in a name-clash. Constructors such as <tt>
X(const X &amp;)</tt>, <tt>X(X &amp;)</tt>, and <tt>X(X *)</tt> are handled as
copy constructors in SWIG.</p>
<p><b> Note:</b> SWIG does<em> not</em> generate a copy constructor
wrapper unless one is explicitly declared in the class. This differs
from the treatment of default constructors and destructors. However,
copy constructor wrappers can be generated if using the <tt>copyctor</tt>
<a href="#Customization_feature_flags">feature flag</a>. For example:</p>
<div class="code">
<pre>
%copyctor List;
class List {
public:
List();
};
</pre>
</div>
<p> Will generate a copy constructor wrapper for <tt>List</tt>.</p>
<p><b> Compatibility note:</b> Special support for copy constructors was
not added until SWIG-1.3.12. In previous versions, copy constructors
could be wrapped, but they had to be renamed. For example:</p>
<div class="code">
<pre>
%rename(CopyFoo) Foo::Foo(const Foo &amp;);
class Foo {
public:
Foo();
Foo(const Foo &amp;);
...
};
</pre>
</div>
<p> For backwards compatibility, SWIG does not perform any special
copy-constructor handling if the constructor has been manually renamed.
For instance, in the above example, the name of the constructor is set
to <tt>new_CopyFoo()</tt>. This is the same as in older versions.</p>
<h3><a name="SWIGPlus_nn11">6.6.5 Member functions</a></h3>
<p> All member functions are roughly translated into accessor functions
like this :</p>
<div class="code">
<pre>
int List_search(List *obj, char *value) {
return obj-&gt;search(value);
}
</pre>
</div>
<p> This translation is the same even if the member function has been
declared as <tt>virtual</tt>.</p>
<p> It should be noted that SWIG does not<em> actually</em> create a C
accessor function in the code it generates. Instead, member access such
as <tt>obj-&gt;search(value)</tt> is directly inlined into the generated
wrapper functions. However, the name and calling convention of the
low-level procedural wrappers match the accessor function prototype
described above.</p>
<h3><a name="SWIGPlus_nn12">6.6.6 Static members</a></h3>
<p> Static member functions are called directly without making any
special transformations. For example, the static member function <tt>
print(List *l)</tt> directly invokes <tt>List::print(List *l)</tt> in
the generated wrapper code.</p>
<h3><a name="SWIGPlus_member_data">6.6.7 Member data</a></h3>
<p> Member data is handled in exactly the same manner as for C
structures. A pair of accessor functions are effectively created. For
example :</p>
<div class="code">
<pre>
int List_length_get(List *obj) {
return obj-&gt;length;
}
int List_length_set(List *obj, int value) {
obj-&gt;length = value;
return value;
}
</pre>
</div>
<p> A read-only member can be created using the <tt>%immutable</tt> and <tt>
%mutable</tt> <a href="#Customization_feature_flags">feature flag</a>
directive. For example, we probably wouldn't want the user to change
the length of a list so we could do the following to make the value
available, but read-only.</p>
<div class="code">
<pre>
class List {
public:
...
%immutable;
int length;
%mutable;
...
};
</pre>
</div>
<p> Alternatively, you can specify an immutable member in advance like
this:</p>
<div class="code">
<pre>
%immutable List::length;
...
class List {
...
int length; // Immutable by above directive
...
};
</pre>
</div>
<p> Similarly, all data attributes declared as <tt>const</tt> are
wrapped as read-only members.</p>
<p> By default, SWIG uses the const reference typemaps for members that
are primitive types. There are some subtle issues when wrapping data
members that are not primitive types, such as classes. For instance, if
you had another class like this,</p>
<div class="code">
<pre>
class Foo {
public:
List items;
...
</pre>
</div>
<p> then the low-level accessor to the <tt>items</tt> member actually
uses pointers. For example:</p>
<div class="code">
<pre>
List *Foo_items_get(Foo *self) {
return &amp;self-&gt;items;
}
void Foo_items_set(Foo *self, List *value) {
self-&gt;items = *value;
}
</pre>
</div>
<p> More information about this can be found in the SWIG Basics chapter
in the <a href="#SWIG_structure_data_members">Structure data members</a>
and <a href="#SWIG_readonly_variables">Creating read-only variables</a>
sections.</p>
<p> Additionally, any C++ type that is not assignable is also wrapped as
a read-only member. Consider the class below which is non-assignable
due to the private assignment operator:</p>
<div class="code">
<pre>
class NonAssignable {
private:
NonAssignable &amp; operator=(const NonAssignable &amp;);
public:
NonAssignable();
};
struct ImmutableVars {
NonAssignable non_assignable;
};
</pre>
</div>
<p> The <tt>non_assignable</tt> member variable is immutable by default
as SWIG detects that <tt>NonAssignable</tt> is not assignable. SWIG
must of course see the full type information in order to get this
correct, otherwise you may have to use <tt>%immutable</tt> to make the
variable read-only.</p>
<p> The wrapper code to generate the accessors for classes comes from
the pointer typemaps. This can be somewhat unnatural for some types.
For example, a user would expect the STL std::string class member
variables to be wrapped as a string in the target language, rather than
a pointer to this class. The const reference typemaps offer this type
of marshalling, so there is a feature to tell SWIG to use the const
reference typemaps rather than the pointer typemaps. It is the
naturalvar feature and can be used to effectively change the way
accessors are generated to the following:</p>
<div class="code">
<pre>
const List &amp;Foo_items_get(Foo *self) {
return self-&gt;items;
}
void Foo_items_set(Foo *self, const List &amp;value) {
self-&gt;items = value;
}
</pre>
</div>
<p> The <tt>%naturalvar</tt> directive is a macro for, and hence
equivalent to, <tt>%feature(&quot;naturalvar&quot;)</tt>. It can be used as
follows:</p>
<div class="code">
<pre>
// All List variables will use const List&amp; typemaps
%naturalvar List;
// Only Foo::myList will use const List&amp; typemaps
%naturalvar Foo::myList;
struct Foo {
List myList;
};
// All non-primitive types will use const reference typemaps
%naturalvar;
</pre>
</div>
<p> The observant reader will notice that <tt>%naturalvar</tt> works
like any other <a href="#Customization_feature_flags">feature flag</a>
directive but with some extra flexibility. The first of the example
usages above shows <tt>%naturalvar</tt> attaching to the <tt>myList</tt>
's variable type, that is the <tt>List</tt> class. The second usage
shows <tt>%naturalvar</tt> attaching to the variable name. Hence the
naturalvar feature can be used on either the variable's name or type.
Note that using the naturalvar feature on a variable's name overrides
any naturalvar feature attached to the variable's type.</p>
<p> It is generally a good idea to use this feature globally as the
reference typemaps have extra NULL checking compared to the pointer
typemaps. A pointer can be NULL, whereas a reference cannot, so the
extra checking ensures that the target language user does not pass in a
value that translates to a NULL pointer and thereby preventing any
potential NULL pointer dereferences. The <tt>%naturalvar</tt> feature
will apply to global variables in addition to member variables in some
language modules, eg C# and Java.</p>
<p> The naturalvar behavior can also be turned on as a global setting
via the <tt>-naturalvar</tt> commandline option or the module mode
option, <tt>%module(naturalvar=1)</tt>. However, any use of <tt>
%feature(&quot;naturalvar&quot;)</tt> will override the global setting.</p>
<p><b> Compatibility note:</b> The <tt>%naturalvar</tt> feature was
introduced in SWIG-1.3.28, prior to which it was necessary to manually
apply the const reference typemaps, eg <tt>%apply const std::string &amp; {
std::string * }</tt>, but this example would also apply the typemaps to
methods taking a <tt>std::string</tt> pointer.</p>
<p><b> Compatibility note:</b> Prior to SWIG-1.3.12, all members of
unknown type were wrapped into accessor functions using pointers. For
example, if you had a structure like this</p>
<div class="code">
<pre>
struct Foo {
size_t len;
};
</pre>
</div>
<p> and nothing was known about <tt>size_t</tt>, then accessors would be
written to work with <tt>size_t *</tt>. Starting in SWIG-1.3.12, this
behavior has been modified. Specifically, pointers will<em> only</em>
be used if SWIG knows that a datatype corresponds to a structure or
class. Therefore, the above code would be wrapped into accessors
involving <tt>size_t</tt>. This change is subtle, but it smooths over a
few problems related to structure wrapping and some of SWIG's
customization features.</p>
<h2><a name="SWIGPlus_nn15">6.7 Protection</a></h2>
<p> SWIG wraps class members that are public following the C++
conventions, i.e., by explicit public declaration or by the use of <tt>
using</tt> declarations. In general, anything specified in a private or
protected section will be ignored, although the internal code generator
sometimes looks at the contents of the private and protected sections
so that it can properly generate code for default constructors and
destructors. Directors could also modify the way non-public virtual
protected members are treated.</p>
<p> By default, members of a class definition are assumed to be private
until you explicitly give a `<tt>public:</tt>' declaration (This is the
same convention used by C++).</p>
<h2><a name="SWIGPlus_nn16">6.8 Enums and constants</a></h2>
<p> Enumerations and constants are handled differently by the different
language modules and are described in detail in the appropriate
language chapter. However, many languages map enums and constants in a
class definition into constants with the classname as a prefix. For
example :</p>
<div class="code">
<pre>
class Swig {
public:
enum {ALE, LAGER, PORTER, STOUT};
};
</pre>
</div>
<p> Generates the following set of constants in the target scripting
language :</p>
<div class="targetlang">
<pre>
Swig_ALE = Swig::ALE
Swig_LAGER = Swig::LAGER
Swig_PORTER = Swig::PORTER
Swig_STOUT = Swig::STOUT
</pre>
</div>
<p> Members declared as <tt>const</tt> are wrapped as read-only members
and do not create constants.</p>
<h2><a name="SWIGPlus_nn17">6.9 Friends</a></h2>
<h3><a name="SWIGPlus_friend_classes">6.9.1 Friend classes</a></h3>
<p> Friend classes are a C++ feature that do not affect SWIG wrappers.
SWIG simply parses the friend class declarations, but they are
effectively ignored as they have no meaningful effect for wrappers. An
example of friend classes:</p>
<div class="code">
<pre>
class X;
class Y;
class C {
// Friend classes have no effect on generated wrappers
friend class X;
friend Y;
...
};
</pre>
</div>
<h3><a name="SWIGPlus_friend_function_definitions">6.9.2 Friend function
definitions</a></h3>
<p> A friend function definition in a C++ class defines a non-member
function of the class and simultaneously makes it a friend of the
class. For example, if you have this code:</p>
<div class="code">
<pre>
class Buddy {
int val;
friend int blah(Buddy *b) { return b-&gt;val; }
public:
...
};
</pre>
</div>
<p> then the <tt>friend</tt> function definition results in wrapper code
equivalent to one generated for the following:</p>
<div class="code">
<pre>
class Buddy {
int val;
friend int blah(Buddy *b);
public:
...
};
int blah(Buddy *b) { return b-&gt;val; }
</pre>
</div>
<p> Access from target languages is thus as if <tt>blah</tt> was wrapped
as a non-member function. The function is available and wrapped even if
the friend is defined with private or protected access.</p>
<p> A friend definition, as in C++, is understood to be in the same
scope that the class is defined in, hence the scoping required for SWIG
directives, such as <tt>%ignore</tt>, is as follows:</p>
<div class="code">
<pre>
%ignore bar::blah(Buddy *b);
// Not: %ignore bar::Buddy::blah(Buddy *b);
namespace bar {
class Buddy {
int val;
friend int blah(Buddy *b) { return b-&gt;val; }
public:
...
};
}
</pre>
</div>
<p> and a wrapper for <tt>blah</tt> will not be generated.</p>
<h3><a name="SWIGPlus_friend_function_declarations">6.9.3 Friend
function declarations</a></h3>
<p> A C++ class can specify friends via friend function declarations.
These functions are allowed access to the private and protected members
of a class. This is pure C++ functionality and these friend function
declarations are hence quietly ignored by SWIG and do not result in any
wrappers. Well, not always! The C++ rules for friends that SWIG needs
to follow are not that simple. Technically, only qualified function
declarations are silently ignored by SWIG. Below are some examples of
qualified friend declarations in <tt>A</tt> that are quietly ignored:</p>
<div class="code">
<pre>
struct B {
int f();
B();
~B();
...
};
int g();
class A {
public:
// The following friend function-declarations are silently ignored (including constructor and destructor friends)
friend B::B();
friend B::~B();
friend int B::f();
friend int ::g();
...
};
</pre>
</div>
<p> In the example above, if SWIG parses the struct <tt>B</tt> and
global function <tt>g()</tt>, then they are of course wrapped as
normal.</p>
<h3><a name="SWIGPlus_friends_unqualified">6.9.4 Unqualified friend
functions</a></h3>
<p> Further clarification is required regarding both friend function
definitions and declarations. In C++, friend function definitions can
only be unqualified, whereas, friend function declarations can be
either unqualified or qualified. Qualified friend function declarations
are silently ignored by SWIG as covered in the previous section. SWIG
does generate wrappers for any unqualified friend functions that it
parses. This section goes through some of the complexities of wrapping
unqualified friend functions.</p>
<p> Consider an unqualified friend function definition:</p>
<div class="code">
<pre>
class Chum {
int val;
friend int blah() { Chum c; c.private_function(); return c.val; }
void private_function();
public:
...
};
</pre>
</div>
<p> The <tt>Chum::blah()</tt> friend is very similar to the <tt>
Buddy::blah(Buddy *)</tt> friend presented earlier. However, the
generated code to call <tt>blah()</tt> may not compile unlike the code
to call <tt>blah(Buddy *)</tt>. The compiler error will be something
like:<div class="shell">
<pre>
error: 'blah' was not declared in this scope
</pre>
</div></p>
<p> The reason one works and the other doesn't is due to the rules
around unqualified friend definitions/declarations. Broadly, friends
are not visible for lookup except via argument dependent lookup that
considers the class that the friend is defined/declared in, unless
there is a<i> matching declaration</i> at namespace scope. This will
probably only make sense if you are conversant with this C++ concept,
which is covered quite well at <a href="https://en.cppreference.com/w/cpp/language/adl">
Argument-dependent lookup</a>. In our examples, <tt>blah(Buddy *)</tt>
is visible via argument dependent lookup, but <tt>blah()</tt> is not.
The solution is thus to provide a<i> matching declaration</i> in order
to make the function visible to the compiler. Simply add:</p>
<div class="code">
<pre>
int blah();
</pre>
</div>
<p> SWIG does<b> not</b> have to parse it. In all likelihood, your code
already has the<i> matching declaration</i> as it is required in order
for the friend function definition to be usable from pure C++ code.</p>
<p> The same potential problem applies to unqualified friend function
declarations, such as:</p>
<div class="code">
<pre>
class Mate {
int val;
friend int blah(); // Unqualified friend function declaration
void private_function();
public:
...
};
</pre>
</div>
<p> Again, the actual function declaration needs to be visible to the
compiler. Or just the actual function definition as shown below. This
must be defined in the same scope as <tt>Mate</tt>. Of course the
function definition is necessary in order to avoid linking issues too.</p>
<div class="code">
<pre>
int blah() { Mate m; m.private_function(); return m.val; }
</pre>
</div>
<h2><a name="SWIGPlus_nn18">6.10 References and pointers</a></h2>
<p> C++ references are supported, but SWIG transforms them back into
pointers. For example, a declaration like this :</p>
<div class="code">
<pre>
class Foo {
public:
double bar(double &amp;a);
}
</pre>
</div>
<p> has a low-level accessor</p>
<div class="code">
<pre>
double Foo_bar(Foo *obj, double *a) {
obj-&gt;bar(*a);
}
</pre>
</div>
<p> As a special case, most language modules pass <tt>const</tt>
references to primitive datatypes (<tt>int</tt>, <tt>short</tt>, <tt>
float</tt>, etc.) by value instead of pointers. For example, if you have
a function like this,</p>
<div class="code">
<pre>
void foo(const int &amp;x);
</pre>
</div>
<p> it is called from a script as follows:</p>
<div class="targetlang">
<pre>
foo(3) # Notice pass by value
</pre>
</div>
<p> Functions that return a reference are remapped to return a pointer
instead. For example:</p>
<div class="code">
<pre>
class Bar {
public:
Foo &amp;spam();
};
</pre>
</div>
<p> Generates an accessor like this:</p>
<div class="code">
<pre>
Foo *Bar_spam(Bar *obj) {
Foo &amp;result = obj-&gt;spam();
return &amp;result;
}
</pre>
</div>
<p> However, functions that return <tt>const</tt> references to
primitive datatypes (<tt>int</tt>, <tt>short</tt>, etc.) normally
return the result as a value rather than a pointer. For example, a
function like this,</p>
<div class="code">
<pre>
const int &amp;bar();
</pre>
</div>
<p> will return integers such as 37 or 42 in the target scripting
language rather than a pointer to an integer.</p>
<p> Don't return references to objects allocated as local variables on
the stack. SWIG doesn't make a copy of the objects so this will
probably cause your program to crash.</p>
<p><b> Note:</b> The special treatment for references to primitive
datatypes is necessary to provide more seamless integration with more
advanced C++ wrapping applications---especially related to templates
and the STL. This was first added in SWIG-1.3.12.</p>
<h2><a name="SWIGPlus_nn19">6.11 Pass and return by value</a></h2>
<p> Occasionally, a C++ program will pass and return class objects by
value. For example, a function like this might appear:</p>
<div class="code">
<pre>
Vector cross_product(Vector a, Vector b);
</pre>
</div>
<p> If no information is supplied about <tt>Vector</tt>, SWIG creates a
wrapper function similar to the following:</p>
<div class="code">
<pre>
Vector *wrap_cross_product(Vector *a, Vector *b) {
Vector x;
Vector y;
Vector r;
x = *a;
y = *b;
r = cross_product(x, y);
return new Vector(r);
}</pre>
</div>
<p> In order for the wrapper code to compile, <tt>Vector</tt> must
define a default constructor and a copy assignment operator (and/or a
move assignment operator for C++11 and later). The <a href="#CPlusPlus11_move_only">
Movable and move-only types</a> section should be read regarding C++11
move semantics and return by value.</p>
<p> If <tt>Vector</tt> is defined as a class in the interface, but it
does not support a default constructor and an assignment operator, SWIG
changes the wrapper code by encapsulating the arguments inside a
special C++ template wrapper class, through a process called the
&quot;Fulton Transform&quot;. This produces a wrapper that looks like this:</p>
<div class="code">
<pre>
Vector cross_product(Vector *a, Vector *b) {
SwigValueWrapper&lt;Vector&gt; x;
SwigValueWrapper&lt;Vector&gt; y;
SwigValueWrapper&lt;Vector&gt; r;
x = *a;
y = *b;
r = cross_product(x, y);
return new Vector(r);
}
</pre>
</div>
<p> This transformation is a little sneaky, but it provides support for
pass-by-value even when a class is not default constructible nor
assignable and it makes it possible to properly support a number of
SWIG's customization options. The definition of <tt>SwigValueWrapper</tt>
can be found by reading the SWIG wrapper code. This class is really
nothing more than a thin wrapper around a pointer.</p>
<p> Although SWIG usually detects the classes to which the Fulton
Transform should be applied, in some situations it's necessary to
override it. That's done with <tt>%feature(&quot;valuewrapper&quot;)</tt> to
ensure it is used and <tt>%feature(&quot;novaluewrapper&quot;)</tt> to ensure it
is not used:</p>
<div class="code">
<pre>
%feature(&quot;novaluewrapper&quot;) A;
class A;
%feature(&quot;valuewrapper&quot;) B;
struct B {
B();
// ....
};
</pre>
</div>
<p> It is well worth considering turning this feature on for classes
that do have a default constructor. It will remove a redundant
constructor call at the point of the variable declaration in the
wrapper, so will generate notably better performance for large objects
or for classes with expensive construction. Alternatively consider
returning a reference or a pointer.</p>
<p><b> Note:</b> this transformation has no effect on typemaps or any
other part of SWIG---it should be transparent except that you may see
this code when reading the SWIG output file.</p>
<p><b> Note:</b> This template transformation is new in SWIG-1.3.11 and
may be refined in future SWIG releases. In practice, it is only
absolutely necessary to do this for classes that don't define a default
constructor.</p>
<p><b> Note:</b> The use of this template only occurs when objects are
passed or returned by value. It is not used for C++ pointers or
references.</p>
<h2><a name="SWIGPlus_nn20">6.12 Inheritance</a></h2>
<p> SWIG supports C++ inheritance of classes and allows both single and
multiple inheritance, as limited or allowed by the target language. The
SWIG type-checker knows about the relationship between base and derived
classes and allows pointers to any object of a derived class to be used
in functions of a base class. The type-checker properly casts pointer
values and is safe to use with multiple inheritance.</p>
<p> SWIG treats private or protected inheritance as close to the C++
spirit, and target language capabilities, as possible. In most cases,
this means that SWIG will parse the non-public inheritance
declarations, but that will have no effect in the generated code,
besides the implicit policies derived for constructors and destructors.</p>
<p> The following example shows how SWIG handles inheritance. For
clarity, the full C++ code has been omitted.</p>
<div class="code">
<pre>
// shapes.i
%module shapes
%{
#include &quot;shapes.h&quot;
%}
class Shape {
public:
double x, y;
virtual double area() = 0;
virtual double perimeter() = 0;
void set_location(double x, double y);
};
class Circle : public Shape {
public:
Circle(double radius);
~Circle();
double area();
double perimeter();
};
class Square : public Shape {
public:
Square(double size);
~Square();
double area();
double perimeter();
}
</pre>
</div>
<p> When wrapped into Python, we can perform the following operations
(shown using the low level Python accessors):</p>
<div class="targetlang">
<pre>
$ python
&gt;&gt;&gt; import shapes
&gt;&gt;&gt; circle = shapes.new_Circle(7)
&gt;&gt;&gt; square = shapes.new_Square(10)
&gt;&gt;&gt; print(shapes.Circle_area(circle))
153.93804004599999757
&gt;&gt;&gt; print(shapes.Shape_area(circle))
153.93804004599999757
&gt;&gt;&gt; print(shapes.Shape_area(square))
100.00000000000000000
&gt;&gt;&gt; shapes.Shape_set_location(square, 2, -3)
&gt;&gt;&gt; print(shapes.Shape_perimeter(square))
40.00000000000000000
&gt;&gt;&gt;
</pre>
</div>
<p> In this example, Circle and Square objects have been created. Member
functions can be invoked on each object by making calls to <tt>
Circle_area</tt>, <tt>Square_area</tt>, and so on. However, the same
results can be accomplished by simply using the <tt>Shape_area</tt>
function on either object.</p>
<p> One important point concerning inheritance is that the low-level
accessor functions are only generated for classes in which they are
actually declared. For instance, in the above example, the method <tt>
set_location()</tt> is only accessible as <tt>Shape_set_location()</tt>
and not as <tt>Circle_set_location()</tt> or <tt>Square_set_location()</tt>
. Of course, the <tt>Shape_set_location()</tt> function will accept any
kind of object derived from Shape. Similarly, accessor functions for
the attributes <tt>x</tt> and <tt>y</tt> are generated as <tt>
Shape_x_get()</tt>, <tt>Shape_x_set()</tt>, <tt>Shape_y_get()</tt>, and <tt>
Shape_y_set()</tt>. Functions such as <tt>Circle_x_get()</tt> are not
available--instead you should use <tt>Shape_x_get()</tt>.</p>
<p> Note that there is a one to one correlation between the low-level
accessor functions and the proxy methods and therefore there is also a
one to one correlation between the C++ class methods and the generated
proxy class methods.</p>
<p><b> Note:</b> For the best results, SWIG requires all base classes to
be defined in an interface. Otherwise, you may get a warning message
like this:</p>
<div class="shell">
<pre>
example.i:18: Warning 401: Nothing known about base class 'Foo'. Ignored.
</pre>
</div>
<p> If any base class is undefined, SWIG still generates correct type
relationships. For instance, a function accepting a <tt>Foo *</tt> will
accept any object derived from <tt>Foo</tt> regardless of whether or
not SWIG actually wrapped the <tt>Foo</tt> class. If you really don't
want to generate wrappers for the base class, but you want to silence
the warning, you might consider using the <tt>%import</tt> directive to
include the file that defines <tt>Foo</tt>. <tt>%import</tt> simply
gathers type information, but doesn't generate wrappers. Alternatively,
you could just define <tt>Foo</tt> as an empty class in the SWIG
interface or use <a href="#Warnings_suppression">warning suppression</a>
.</p>
<p><b> Note:</b> <tt>typedef</tt>-names<em> can</em> be used as base
classes. For example:</p>
<div class="code">
<pre>
class Foo {
...
};
typedef Foo FooObj;
class Bar : public FooObj { // Ok. Base class is Foo
...
};
</pre>
</div>
<p> Similarly, <tt>typedef</tt> allows unnamed structures to be used as
base classes. For example:</p>
<div class="code">
<pre>
typedef struct {
...
} Foo;
class Bar : public Foo { // Ok.
...
};
</pre>
</div>
<p><b> Compatibility Note:</b> Starting in version 1.3.7, SWIG only
generates low-level accessor wrappers for the declarations that are
actually defined in each class. This differs from SWIG1.1 which used to
inherit all of the declarations defined in base classes and regenerate
specialized accessor functions such as <tt>Circle_x_get()</tt>, <tt>
Square_x_get()</tt>, <tt>Circle_set_location()</tt>, and <tt>
Square_set_location()</tt>. This behavior resulted in huge amounts of
replicated code for large class hierarchies and made it awkward to
build applications spread across multiple modules (since accessor
functions are duplicated in every single module). It is also
unnecessary to have such wrappers when advanced features like proxy
classes are used.<b> Note:</b> Further optimizations are enabled when
using the <tt>-fvirtual</tt> option, which avoids the regenerating of
wrapper functions for virtual members that are already defined in a
base class.</p>
<h2><a name="SWIGPlus_nn21">6.13 A brief discussion of multiple
inheritance, pointers, and type checking</a></h2>
<p> When a target scripting language refers to a C++ object, it normally
uses a tagged pointer object that contains both the value of the
pointer and a type string. For example, in Tcl, a C++ pointer might be
encoded as a string like this:</p>
<div class="diagram">
<pre>
_808fea88_p_Circle
</pre>
</div>
<p> A somewhat common question is whether or not the type-tag could be
safely removed from the pointer. For instance, to get better
performance, could you strip all type tags and just use simple integers
instead?</p>
<p> In general, the answer to this question is no. In the wrappers, all
pointers are converted into a common data representation in the target
language. Typically this is the equivalent of casting a pointer to <tt>
void *</tt>. This means that any C++ type information associated with
the pointer is lost in the conversion.</p>
<p> The problem with losing type information is that it is needed to
properly support many advanced C++ features--especially multiple
inheritance. For example, suppose you had code like this:</p>
<div class="code">
<pre>
class A {
public:
int x;
};
class B {
public:
int y;
};
class C : public A, public B {
};
int A_function(A *a) {
return a-&gt;x;
}
int B_function(B *b) {
return b-&gt;y;
}
</pre>
</div>
<p> Now, consider the following code that uses <tt>void *</tt>.</p>
<div class="code">
<pre>
C *c = new C();
void *p = (void *) c;
...
int x = A_function((A *) p);
int y = B_function((B *) p);
</pre>
</div>
<p> In this code, both <tt>A_function()</tt> and <tt>B_function()</tt>
may legally accept an object of type <tt>C *</tt> (via inheritance).
However, one of the functions will always return the wrong result when
used as shown. The reason for this is that even though <tt>p</tt>
points to an object of type <tt>C</tt>, the casting operation doesn't
work like you would expect. Internally, this has to do with the data
representation of <tt>C</tt>. With multiple inheritance, the data from
each base class is stacked together. For example:</p>
<div class="diagram">
<pre>
------------ &lt;--- (C *), (A *)
| A |
|------------| &lt;--- (B *)
| B |
------------
</pre>
</div>
<p> Because of this stacking, a pointer of type <tt>C *</tt> may change
value when it is converted to a <tt>A *</tt> or <tt>B *</tt>. However,
this adjustment does<em> not</em> occur if you are converting from a <tt>
void *</tt>.</p>
<p> The use of type tags marks all pointers with the real type of the
underlying object. This extra information is then used by SWIG
generated wrappers to correctly cast pointer values under inheritance
(avoiding the above problem).</p>
<p> Some of the language modules are able to solve the problem by
storing multiple instances of the pointer, for example, <tt>A *</tt>,
in the A proxy class as well as <tt>C *</tt> in the C proxy class. The
correct cast can then be made by choosing the correct <tt>void *</tt>
pointer to use and is guaranteed to work as the cast to a void pointer
and back to the same type does not lose any type information:</p>
<div class="code">
<pre>
C *c = new C();
void *p = (void *) c;
void *pA = (void *) c;
void *pB = (void *) c;
...
int x = A_function((A *) pA);
int y = B_function((B *) pB);
</pre>
</div>
<p> In practice, the pointer is held as an integral number in the target
language proxy class.</p>
<h2><a name="SWIGPlus_default_args">6.14 Default arguments</a></h2>
<p> SWIG will wrap all types of functions that have default arguments.
For example member functions:</p>
<div class="code">
<pre>
class Foo {
public:
void bar(int x, int y = 3, int z = 4);
};
</pre>
</div>
<p> SWIG handles default arguments by generating an extra overloaded
method for each defaulted argument. SWIG is effectively handling
methods with default arguments as if it was wrapping the equivalent
overloaded methods. Thus for the example above, it is as if we had
instead given the following to SWIG:</p>
<div class="code">
<pre>
class Foo {
public:
void bar(int x, int y, int z);
void bar(int x, int y);
void bar(int x);
};
</pre>
</div>
<p> The wrappers produced are exactly the same as if the above code was
instead fed into SWIG. Details of this are covered in the next section <a
href="#SWIGPlus_overloaded_methods">Overloaded functions and methods</a>
. This approach allows SWIG to wrap all possible default arguments, but
can be verbose. For example if a method has ten default arguments, then
eleven wrapper methods are generated.</p>
<p> Please see the <a href="#Customization_features_default_args">
Features and default arguments</a> section for more information on using
<tt>%feature</tt> with functions with default arguments. The <a href="#SWIGPlus_ambiguity_resolution_renaming">
Renaming and ambiguity resolution</a> section also deals with using <tt>
%rename</tt> and <tt>%ignore</tt> on methods with default arguments. If
you are writing your own typemaps for types used in methods with
default arguments, you may also need to write a <tt>typecheck</tt>
typemap. See the <a href="#Typemaps_overloading">Typemaps and
overloading</a> section for details or otherwise use the <tt>
compactdefaultargs</tt> feature flag as mentioned below.</p>
<p> For C# please see the <a href="#CSharp_named_arguments">C# named and
optional arguments</a> section for information on special handling of
default arguments available specifically for C#.</p>
<p><b> Compatibility note:</b> Versions of SWIG prior to SWIG-1.3.23
wrapped default arguments slightly differently. Instead a single
wrapper method was generated and the default values were copied into
the C++ wrappers so that the method being wrapped was then called with
all the arguments specified. If the size of the wrappers are a concern
then this approach to wrapping methods with default arguments can be
re-activated by using the <tt>compactdefaultargs</tt> <a href="#Customization_feature_flags">
feature flag</a>.</p>
<div class="code">
<pre>
%feature(&quot;compactdefaultargs&quot;) Foo::bar;
class Foo {
public:
void bar(int x, int y = 3, int z = 4);
};
</pre>
</div>
<p> This is great for reducing the size of the wrappers, but the caveat
is it does not work for the statically typed languages, such as C# and
Java, which don't have optional arguments in the language, Another
restriction of this feature is that it cannot handle default arguments
that are not public. The following example illustrates this:</p>
<div class="code">
<pre>
class Foo {
private:
static const int spam;
public:
void bar(int x, int y = spam); // Won't work with %feature(&quot;compactdefaultargs&quot;) -
// private default value
};
</pre>
</div>
<p> This produces uncompilable wrapper code because default values in
C++ are evaluated in the same scope as the member function whereas SWIG
evaluates them in the scope of a wrapper function (meaning that the
values have to be public).</p>
<p> The <tt>compactdefaultargs</tt> feature is automatically turned on
when wrapping <a href="#SWIG_default_args">C code with default
arguments</a>. Some target languages will also automatically turn on
this feature if the keyword arguments feature (kwargs) is specified for
either C or C++ functions, and the target language supports kwargs, the
<tt>compactdefaultargs</tt> feature is also automatically turned on.
Keyword arguments are a language feature of some scripting languages,
for example Ruby and Python. SWIG is unable to support kwargs when
wrapping overloaded methods, so the default approach cannot be used.</p>
<h2><a name="SWIGPlus_overloaded_methods">6.15 Overloaded functions and
methods</a></h2>
<p> In many language modules, SWIG provides partial support for
overloaded functions, methods, and constructors. For example, if you
supply SWIG with overloaded functions like this:</p>
<div class="code">
<pre>
void foo(int x) {
printf(&quot;x is %d\n&quot;, x);
}
void foo(char *x) {
printf(&quot;x is '%s'\n&quot;, x);
}
</pre>
</div>
<p> The function is used in a completely natural way. For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; foo(3)
x is 3
&gt;&gt;&gt; foo(&quot;hello&quot;)
x is 'hello'
&gt;&gt;&gt;
</pre>
</div>
<p> Overloading works in a similar manner for methods and constructors.
For example if you have this code,</p>
<div class="code">
<pre>
class Foo {
public:
Foo();
Foo(const Foo &amp;); // Copy constructor
void bar(int x);
void bar(char *s, int y);
};
</pre>
</div>
<p> it might be used like this</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; f = Foo() # Create a Foo
&gt;&gt;&gt; f.bar(3)
&gt;&gt;&gt; g = Foo(f) # Copy Foo
&gt;&gt;&gt; f.bar(&quot;hello&quot;, 2)
</pre>
</div>
<h3><a name="SWIGPlus_nn24">6.15.1 Dispatch function generation</a></h3>
<p> The implementation of overloaded functions and methods is somewhat
complicated due to the dynamic nature of scripting languages. Unlike
C++, which binds overloaded methods at compile time, SWIG must
determine the proper function as a runtime check for scripting language
targets. This check is further complicated by the typeless nature of
certain scripting languages. For instance, in Tcl, all types are simply
strings. Therefore, if you have two overloaded functions like this,</p>
<div class="code">
<pre>
void foo(char *x);
void foo(int x);
</pre>
</div>
<p> the order in which the arguments are checked plays a rather critical
role.</p>
<p> For statically typed languages, SWIG uses the language's method
overloading mechanism. To implement overloading for the scripting
languages, SWIG generates a dispatch function that checks the number of
passed arguments and their types. To create this function, SWIG first
examines all of the overloaded methods and ranks them according to the
following rules:</p>
<ol>
<li><b>Number of required arguments.</b> Methods are sorted by
increasing number of required arguments.</li>
<li>
<p><b>Argument type precedence.</b> All C++ datatypes are assigned a
numeric type precedence value (which is determined by the language
module).</p>
<div class="diagram">
<pre>
Type Precedence
---------------- ----------
TYPE * 0 (High)
void * 20
Integers 40
Floating point 60
char 80
Strings 100 (Low)
</pre>
</div>
<p> Using these precedence values, overloaded methods with the same
number of required arguments are sorted in increased order of
precedence values.</p>
</li>
</ol>
<p> This may sound very confusing, but an example will help. Consider
the following collection of overloaded methods:</p>
<div class="code">
<pre>
void foo(double);
void foo(int);
void foo(Bar *);
void foo();
void foo(int x, int y, int z, int w);
void foo(int x, int y, int z = 3);
void foo(double x, double y);
void foo(double x, Bar *z);
</pre>
</div>
<p> The first rule simply ranks the functions by required argument
count. This would produce the following list:</p>
<div class="diagram">
<pre>
rank
-----
[0] foo()
[1] foo(double);
[2] foo(int);
[3] foo(Bar *);
[4] foo(int x, int y, int z = 3);
[5] foo(double x, double y)
[6] foo(double x, Bar *z)
[7] foo(int x, int y, int z, int w);
</pre>
</div>
<p> The second rule, simply refines the ranking by looking at argument
type precedence values.</p>
<div class="diagram">
<pre>
rank
-----
[0] foo()
[1] foo(Bar *);
[2] foo(int);
[3] foo(double);
[4] foo(int x, int y, int z = 3);
[5] foo(double x, Bar *z)
[6] foo(double x, double y)
[7] foo(int x, int y, int z, int w);
</pre>
</div>
<p> Finally, to generate the dispatch function, the arguments passed to
an overloaded method are simply checked in the same order as they
appear in this ranking.</p>
<p> If you're still confused, don't worry about it---SWIG is probably
doing the right thing.</p>
<h3><a name="SWIGPlus_nn25">6.15.2 Ambiguity in overloading</a></h3>
<p> Regrettably, SWIG is not able to support every possible use of valid
C++ overloading. Consider the following example:</p>
<div class="code">
<pre>
void foo(int x);
void foo(long x);
</pre>
</div>
<p> In C++, this is perfectly legal. However, in a scripting language,
there is generally only one kind of integer object. Therefore, which
one of these functions do you pick? Clearly, there is no way to truly
make a distinction just by looking at the value of the integer itself (<tt>
int</tt> and <tt>long</tt> may even be the same precision). Therefore,
when SWIG encounters this situation, it may generate a warning message
like this for scripting languages:</p>
<div class="shell">
<pre>
example.i:4: Warning 509: Overloaded method foo(long) effectively ignored,
example.i:3: Warning 509: as it is shadowed by foo(int).
</pre>
</div>
<p> or for statically typed languages like Java:</p>
<div class="shell">
<pre>
example.i:4: Warning 516: Overloaded method foo(long) ignored,
example.i:3: Warning 516: using foo(int) instead.
</pre>
</div>
<p> This means that the second overloaded function will be inaccessible
from a scripting interface or the method won't be wrapped at all. This
is done as SWIG does not know how to disambiguate it from an earlier
method.</p>
<p> Ambiguity problems are known to arise in the following situations:</p>
<ul>
<li>Integer conversions. Datatypes such as <tt>int</tt>, <tt>long</tt>,
and <tt>short</tt> cannot be disambiguated in some languages. Shown
above.</li>
<li>Floating point conversion. <tt>float</tt> and <tt>double</tt> can
not be disambiguated in some languages.</li>
<li>Pointers and references. For example, <tt>Foo *</tt> and <tt>Foo &amp;</tt>
.</li>
<li>Pointers and arrays. For example, <tt>Foo *</tt> and <tt>Foo [4]</tt>
.</li>
<li>Pointers and instances. For example, <tt>Foo</tt> and <tt>Foo *</tt>
. Note: SWIG converts all instances to pointers.</li>
<li>Qualifiers. For example, <tt>const Foo *</tt> and <tt>Foo *</tt>.</li>
<li>Default vs. non default arguments. For example, <tt>foo(int a, int
b)</tt> and <tt>foo(int a, int b = 3)</tt>.</li>
</ul>
<p> When an ambiguity arises, methods are checked in the same order as
they appear in the interface file. Therefore, earlier methods will
shadow methods that appear later.</p>
<p> When wrapping an overloaded function, there is a chance that you
will get a warning message like this:</p>
<div class="shell">
<pre>
example.i:3: Warning 467: Overloaded foo(int) not supported (incomplete type checking rule -
no precedence level in typecheck typemap for 'int').
</pre>
</div>
<p> This error means that the target language module supports
overloading, but for some reason there is no type-checking rule that
can be used to generate a working dispatch function. The resulting
behavior is then undefined. You should report this as a bug to the <a href="https://www.swig.org/bugs.html">
SWIG bug tracking database</a> if this is due to one of the typemaps
supplied with SWIG.</p>
<p> If you get an error message such as the following,</p>
<div class="shell">
<pre>
foo.i:6. Overloaded declaration ignored. Spam::foo(double )
foo.i:5. Previous declaration is Spam::foo(int )
foo.i:7. Overloaded declaration ignored. Spam::foo(Bar *, Spam *, int )
foo.i:5. Previous declaration is Spam::foo(int )
</pre>
</div>
<p> it means that the target language module has not yet implemented
support for overloaded functions and methods. The only way to fix the
problem is to read the next section.</p>
<h3><a name="SWIGPlus_ambiguity_resolution_renaming">6.15.3 Renaming and
ambiguity resolution</a></h3>
<p> If an ambiguity in overload resolution occurs or if a module doesn't
allow overloading, there are a few strategies for dealing with the
problem. First, you can tell SWIG to ignore one of the methods. This is
easy---simply use the <tt>%ignore</tt> directive. For example:</p>
<div class="code">
<pre>
%ignore foo(long);
void foo(int);
void foo(long); // Ignored. Oh well.
</pre>
</div>
<p> The other alternative is to rename one of the methods. This can be
done using <tt>%rename</tt>. For example:</p>
<div class="code">
<pre>
%rename(&quot;foo_short&quot;) foo(short);
%rename(foo_long) foo(long);
void foo(int);
void foo(short); // Accessed as foo_short()
void foo(long); // Accessed as foo_long()
</pre>
</div>
<p> Note that the quotes around the new name are optional, however,
should the new name be a C/C++ keyword they would be essential in order
to avoid a parsing error. The <tt>%ignore</tt> and <tt>%rename</tt>
directives are both rather powerful in their ability to match
declarations. When used in their simple form, they apply to both global
functions and methods. For example:</p>
<div class="code">
<pre>
/* Forward renaming declarations */
%rename(foo_i) foo(int);
%rename(foo_d) foo(double);
...
void foo(int); // Becomes 'foo_i'
void foo(char *c); // Stays 'foo' (not renamed)
class Spam {
public:
void foo(int); // Becomes 'foo_i'
void foo(double); // Becomes 'foo_d'
...
};
</pre>
</div>
<p> If you only want the renaming to apply to a certain scope, the C++
scope resolution operator (::) can be used. For example:</p>
<div class="code">
<pre>
%rename(foo_i) ::foo(int); // Only rename foo(int) in the global scope.
// (will not rename class members)
%rename(foo_i) Spam::foo(int); // Only rename foo(int) in class Spam
</pre>
</div>
<p> When a renaming operator is applied to a class as in <tt>
Spam::foo(int)</tt>, it is applied to that class and all derived
classes. This can be used to apply a consistent renaming across an
entire class hierarchy with only a few declarations. For example:</p>
<div class="code">
<pre>
%rename(foo_i) Spam::foo(int);
%rename(foo_d) Spam::foo(double);
class Spam {
public:
virtual void foo(int); // Renamed to foo_i
virtual void foo(double); // Renamed to foo_d
...
};
class Bar : public Spam {
public:
virtual void foo(int); // Renamed to foo_i
virtual void foo(double); // Renamed to foo_d
...
};
class Grok : public Bar {
public:
virtual void foo(int); // Renamed to foo_i
virtual void foo(double); // Renamed to foo_d
...
};
</pre>
</div>
<p> It is also possible to include <tt>%rename</tt> specifications in
the class definition itself. For example:</p>
<div class="code">
<pre>
class Spam {
%rename(foo_i) foo(int);
%rename(foo_d) foo(double);
public:
virtual void foo(int); // Renamed to foo_i
virtual void foo(double); // Renamed to foo_d
...
};
class Bar : public Spam {
public:
virtual void foo(int); // Renamed to foo_i
virtual void foo(double); // Renamed to foo_d
...
};
</pre>
</div>
<p> In this case, the <tt>%rename</tt> directives still get applied
across the entire inheritance hierarchy, but it's no longer necessary
to explicitly specify the class prefix <tt>Spam::</tt>.</p>
<p> A special form of <tt>%rename</tt> can be used to apply a renaming
just to class members (of all classes):</p>
<div class="code">
<pre>
%rename(foo_i) *::foo(int); // Only rename foo(int) if it appears in a class.
</pre>
</div>
<p> Note: the <tt>*::</tt> syntax is non-standard C++, but the '*' is
meant to be a wildcard that matches any class name (we couldn't think
of a better alternative so if you have a better idea, send email to the
<a href="https://www.swig.org/mail.html">swig-devel mailing list</a>.</p>
<p> Although this discussion has primarily focused on <tt>%rename</tt>
all of the same rules also apply to <tt>%ignore</tt>. For example:</p>
<div class="code">
<pre>
%ignore foo(double); // Ignore all foo(double)
%ignore Spam::foo; // Ignore foo in class Spam (and foo in any derived classes)
%ignore Spam::foo(double); // Ignore foo(double) in class Spam (and foo in any derived classes)
%ignore *::foo(double); // Ignore foo(double) in all classes
</pre>
</div>
<p> When applied to a base class, <tt>%ignore</tt> forces all
definitions in derived classes to disappear. For example, <tt>%ignore
Spam::foo(double)</tt> will eliminate <tt>foo(double)</tt> in <tt>Spam</tt>
and all classes derived from <tt>Spam</tt>.</p>
<p><b> Notes on %rename and %ignore:</b></p>
<ul>
<li>
<p>Since, the <tt>%rename</tt> declaration is used to declare a renaming
in advance, it can be placed at the start of an interface file. This
makes it possible to apply a consistent name resolution without having
to modify header files. For example:</p>
<div class="code">
<pre>
%module foo
/* Rename these overloaded functions */
%rename(foo_i) foo(int);
%rename(foo_d) foo(double);
%include &quot;header.h&quot;
</pre>
</div></li>
<li>
<p>The scope qualifier (::) can also be used on simple names. For
example:</p>
<div class="code">
<pre>
%rename(bar) ::foo; // Rename foo to bar in global scope only
%rename(bar) Spam::foo; // Rename foo to bar in class Spam only
%rename(bar) *::foo; // Rename foo in classes only
</pre>
</div></li>
<li>
<p>Name matching tries to find the most specific match that is defined.
A qualified name such as <tt>Spam::foo</tt> always has higher
precedence than an unqualified name <tt>foo</tt>. <tt>Spam::foo</tt>
has higher precedence than <tt>*::foo</tt> and <tt>*::foo</tt> has
higher precedence than <tt>foo</tt>. A parameterized name has higher
precedence than an unparameterized name within the same scope level.
However, an unparameterized name with a scope qualifier has higher
precedence than a parameterized name in global scope (e.g., a renaming
of <tt>Spam::foo</tt> takes precedence over a renaming of <tt>foo(int)</tt>
).</p>
</li>
<li>
<p> Renaming a class member, using an unparameterized but qualified
name, such as <tt>Spam::foo</tt>, also applies to members in all
derived classes that have members with the same name. This can be used
to simply rename a method, across an entire class hierarchy for all
overloaded and non-overloaded methods. This also applies to methods
introduced via <tt>using</tt> declarations, see <a href="#SWIGPlus_nn35">
Using declarations and inheritance</a>. For example:</p>
<div class="code">
<pre>
%rename(foo_new) Spam::foo;
class Spam {
public:
virtual void foo(int); // Renamed to foo_new
};
class Bar : public Spam {
public:
virtual void foo(int); // Renamed to foo_new
void foo(bool, short, int); // Renamed to foo_new
};
class Grok : public Bar {
public:
virtual void foo(int); // Renamed to foo_new
void foo(bool, int); // Renamed to foo_new
void foo(const char *); // Renamed to foo_new
void foo(Bar *); // Renamed to foo_new
};
class Spok : public Grok {
public:
void foo(); // Renamed to foo_new
};
class Knock : public Spok {
public:
using Grok::foo; // Introduced methods renamed to foo_new
};
</pre>
</div></li>
<li>
<p> The order in which <tt>%rename</tt> directives are defined does not
matter as long as they appear before the declarations to be renamed.
Thus, there is no difference between saying:</p>
<div class="code">
<pre>
%rename(bar) foo;
%rename(foo_i) Spam::foo(int);
%rename(Foo) Spam::foo;
</pre>
</div>
<p> and this</p>
<div class="code">
<pre>
%rename(Foo) Spam::foo;
%rename(bar) foo;
%rename(foo_i) Spam::foo(int);
</pre>
</div>
<p> (the declarations are not stored in a linked list and order has no
importance). Of course, a repeated <tt>%rename</tt> directive will
change the setting for a previous <tt>%rename</tt> directive if exactly
the same name, scope, and parameters are supplied.</p>
</li>
<li>For multiple inheritance where renaming rules are defined for
multiple base classes, the first renaming rule found on a depth-first
traversal of the class hierarchy is used.</li>
<li>
<p>The name matching rules strictly follow member qualifier rules. For
example, if you have a class and member with a member that is const
qualified like this:</p>
<div class="code">
<pre>
class Spam {
public:
...
void bar() const;
...
};
</pre>
</div>
<p> the declaration</p>
<div class="code">
<pre>
%rename(name) Spam::bar();
</pre>
</div>
<p> will not apply as there is no unqualified member <tt>bar()</tt>. The
following will apply the rename as the qualifier matches correctly:</p>
<div class="code">
<pre>
%rename(name) Spam::bar() const;
</pre>
</div>
<p> Similarly for combinations of cv-qualifiers and ref-qualifiers, all
the qualifiers must be specified to match correctly:</p>
<div class="code">
<pre>
%rename(name) Jam::bar(); // will not match
%rename(name) Jam::bar() &amp;; // will not match
%rename(name) Jam::bar() const; // will not match
%rename(name) Jam::bar() const &amp;; // ok, will match
class Jam {
public:
...
void bar() const &amp;;
...
};
</pre>
</div>
<p> An often overlooked C++ feature is that classes can define two
different overloaded members that differ only in their qualifiers, like
this:</p>
<div class="code">
<pre>
class Spam {
public:
...
void bar(); // Unqualified member
void bar() const; // Qualified member
...
};
</pre>
</div>
<p> %rename can then be used to target each of the overloaded methods
individually. For example we can give them separate names in the target
language:</p>
<div class="code">
<pre>
%rename(name1) Spam::bar();
%rename(name2) Spam::bar() const;
</pre>
</div>
<p> Similarly, if you merely wanted to ignore one of the declarations,
use <tt>%ignore</tt> with the full qualifier. For example, the
following directive would tell SWIG to ignore the <tt>const</tt>
version of <tt>bar()</tt> above:</p>
<div class="code">
<pre>
%ignore Spam::bar() const; // Ignore bar() const, but leave other bar() alone
</pre>
</div></li>
<li>
<p> Currently no resolution is performed in order to match function
parameters. This means function parameter types must match exactly. For
example, namespace qualifiers and typedefs will not work. The following
usage of typedefs demonstrates this:</p>
<div class="code">
<pre>
typedef int Integer;
%rename(foo_i) foo(int);
class Spam {
public:
void foo(Integer); // Stays 'foo' (not renamed)
};
class Ham {
public:
void foo(int); // Renamed to foo_i
};
</pre>
</div></li>
<li>
<p> The name matching rules also use default arguments for finer control
when wrapping methods that have default arguments. Recall that methods
with default arguments are wrapped as if the equivalent overloaded
methods had been parsed (<a href="#SWIGPlus_default_args">Default
arguments</a> section). Let's consider the following example class:</p>
<div class="code">
<pre>
class Spam {
public:
...
void bar(int i=-1, double d=0.0);
...
};
</pre>
</div>
<p> The following <tt>%rename</tt> will match exactly and apply to all
the target language overloaded methods because the declaration with the
default arguments exactly matches the wrapped method:</p>
<div class="code">
<pre>
%rename(newbar) Spam::bar(int i=-1, double d=0.0);
</pre>
</div>
<p> The C++ method can then be called from the target language with the
new name no matter how many arguments are specified, for example: <tt>
newbar(2, 2.0)</tt>, <tt>newbar(2)</tt> or <tt>newbar()</tt>. However,
if the <tt>%rename</tt> does not contain the default arguments:</p>
<div class="code">
<pre>
%rename(newbar) Spam::bar(int i, double d);
</pre>
</div>
<p> then only one of the three equivalent overloaded methods will be
renamed and wrapped as if SWIG parsed:</p>
<div class="code">
<pre>
void Spam::newbar(int i, double d);
void Spam::bar(int i);
void Spam::bar();
</pre>
</div>
<p> The C++ method must then be called from the target language with the
new name <tt>newbar(2, 2.0)</tt> when both arguments are supplied or
with the original name as <tt>bar(2)</tt> (one argument) or <tt>bar()</tt>
(no arguments).</p>
<p> In fact it is possible to use <tt>%rename</tt> on the equivalent
overloaded methods, to rename all the equivalent overloaded methods:</p>
<div class="code">
<pre>
%rename(bar_2args) Spam::bar(int i, double d);
%rename(bar_1arg) Spam::bar(int i);
%rename(bar_default) Spam::bar();
</pre>
</div>
<p> Similarly, the extra overloaded methods can be selectively ignored
using <tt>%ignore</tt>.</p>
<p><b> Compatibility note:</b> The <tt>%rename</tt> directive introduced
the default argument matching rules in SWIG-1.3.23 at the same time as
the changes to wrapping methods with default arguments was introduced.</p>
</li>
</ul>
<h3><a name="SWIGPlus_nn27">6.15.4 Comments on overloading</a></h3>
<p> Support for overloaded methods was first added in SWIG-1.3.14. The
implementation is somewhat unusual when compared to similar tools. For
instance, the order in which declarations appear is largely irrelevant
in SWIG. Furthermore, SWIG does not rely upon trial execution or
exception handling to figure out which method to invoke.</p>
<p> Internally, the overloading mechanism is completely configurable by
the target language module. Therefore, the degree of overloading
support may vary from language to language. As a general rule,
statically typed languages like Java are able to provide more support
than dynamically typed languages like Perl, Python, Ruby, and Tcl.</p>
<h2><a name="SWIGPlus_nn28">6.16 Overloaded operators</a></h2>
<p> C++ overloaded operator declarations can be wrapped. For example,
consider a class like this:</p>
<div class="code">
<pre>
class Complex {
private:
double rpart, ipart;
public:
Complex(double r = 0, double i = 0) : rpart(r), ipart(i) { }
Complex(const Complex &amp;c) : rpart(c.rpart), ipart(c.ipart) { }
Complex &amp;operator=(const Complex &amp;c) {
rpart = c.rpart;
ipart = c.ipart;
return *this;
}
Complex operator+(const Complex &amp;c) const {
return Complex(rpart+c.rpart, ipart+c.ipart);
}
Complex operator-(const Complex &amp;c) const {
return Complex(rpart-c.rpart, ipart-c.ipart);
}
Complex operator*(const Complex &amp;c) const {
return Complex(rpart*c.rpart - ipart*c.ipart,
rpart*c.ipart + c.rpart*ipart);
}
Complex operator-() const {
return Complex(-rpart, -ipart);
}
double re() const { return rpart; }
double im() const { return ipart; }
};
</pre>
</div>
<p> When operator declarations appear, they are handled in<em> exactly</em>
the same manner as regular methods. However, the names of these methods
are set to strings like &quot;<tt>operator +</tt>&quot; or &quot;<tt>operator -</tt>&quot;.
The problem with these names is that they are illegal identifiers in
most scripting languages. For instance, you can't just create a method
called &quot;<tt>operator +</tt>&quot; in Python--there won't be any way to call
it.</p>
<p> Some language modules already know how to automatically handle
certain operators (mapping them into operators in the target language).
However, the underlying implementation of this is really managed in a
very general way using the <tt>%rename</tt> directive. For example, in
Python a declaration similar to this is used:</p>
<div class="code">
<pre>
%rename(__add__) Complex::operator+;
</pre>
</div>
<p> This binds the + operator to a method called <tt>__add__</tt> (which
is conveniently the same name used to implement the Python + operator).
Internally, the generated wrapper code for a wrapped operator will look
something like this pseudocode:</p>
<div class="code">
<pre>
_wrap_Complex___add__(args) {
... get args ...
obj-&gt;operator+(args);
...
}
</pre>
</div>
<p> When used in the target language, it may now be possible to use the
overloaded operator normally. For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = Complex(3, 4)
&gt;&gt;&gt; b = Complex(5, 2)
&gt;&gt;&gt; c = a + b # Invokes __add__ method
</pre>
</div>
<p> It is important to realize that there is nothing magical happening
here. The <tt>%rename</tt> directive really only picks a valid method
name. If you wrote this:</p>
<div class="code">
<pre>
%rename(add) operator+;
</pre>
</div>
<p> The resulting scripting interface might work like this:</p>
<div class="targetlang">
<pre>
a = Complex(3, 4)
b = Complex(5, 2)
c = a.add(b) # Call a.operator+(b)
</pre>
</div>
<p> All of the techniques described to deal with overloaded functions
also apply to operators. For example:</p>
<div class="code">
<pre>
%ignore Complex::operator=; // Ignore = in class Complex
%ignore *::operator=; // Ignore = in all classes
%ignore operator=; // Ignore = everywhere.
%rename(__sub__) Complex::operator-;
%rename(__neg__) Complex::operator-(); // Unary -
</pre>
</div>
<p> The last part of this example illustrates how multiple definitions
of the <tt>operator-</tt> method might be handled.</p>
<p> Handling operators in this manner is mostly straightforward.
However, there are a few subtle issues to keep in mind:</p>
<ul>
<li>
<p>In C++, it is fairly common to define different versions of the
operators to account for different types. For example, a class might
also include a friend function like this:</p>
<div class="code">
<pre>
class Complex {
public:
friend Complex operator+(Complex &amp;, double);
};
Complex operator+(Complex &amp;, double);
</pre>
</div>
<p> SWIG simply ignores all <tt>friend</tt> declarations. Furthermore,
it doesn't know how to associate the associated <tt>operator+</tt> with
the class (because it's not a member of the class).</p>
<p> It's still possible to make a wrapper for this operator, but you'll
have to handle it like a normal function. For example:</p>
<div class="code">
<pre>
%rename(add_complex_double) operator+(Complex &amp;, double);
</pre>
</div></li>
<li>
<p>Certain operators are ignored by default. For instance, <tt>new</tt>
and <tt>delete</tt> operators are ignored as well as conversion and
index operators. A warning such as the one below is shown:</p>
<div class="shell">
<pre>
example.i:12: Warning 503: Can't wrap 'operator []' unless renamed to a valid identifier.
</pre>
</div></li>
<li>
<p>The index operator, <tt>operator[]</tt>, is particularly difficult to
overload due to differences in C++ implementations. Specifically, the
get and set operators in other languages typically are separated into
two methods such that additional logic can be packed into the
operations; C# uses <tt>this[type key] { get { ... } set { ... }}</tt>,
Python uses <tt>__getitem__</tt> and <tt>__setitem__</tt>, etc. In C++
if the return type of <tt>operator[]</tt> is a reference and the method
is const, it is often indicative of the<i> setter</i>, and the<i>
getter</i> is usually a const function return an object by value. In
the absence of any hard and fast rules and the fact that there may be
multiple index operators, it is up to the user to choose the getter and
setter to use by using %rename as shown earlier.</p>
</li>
<li>The semantics of certain C++ operators may not match those in the
target language.</li>
</ul>
<h2><a name="SWIGPlus_class_extension">6.17 Class extension</a></h2>
<p> New methods can be added to a class using the <tt>%extend</tt>
directive. This directive is primarily used in conjunction with proxy
classes to add additional functionality to an existing class. For
example :</p>
<div class="code">
<pre>
%module vector
%{
#include &quot;vector.h&quot;
%}
class Vector {
public:
double x, y, z;
Vector();
~Vector();
... bunch of C++ methods ...
%extend {
char *__str__() {
static char temp[256];
sprintf(temp, &quot;[ %g, %g, %g ]&quot;, $self-&gt;x, $self-&gt;y, $self-&gt;z);
return &amp;temp[0];
}
}
};
</pre>
</div>
<p> This code adds a <tt>__str__</tt> method to our class for producing
a string representation of the object. In Python, such a method would
allow us to print the value of an object using the <tt>print</tt>
command.</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt;
&gt;&gt;&gt; v = Vector();
&gt;&gt;&gt; v.x = 3
&gt;&gt;&gt; v.y = 4
&gt;&gt;&gt; v.z = 0
&gt;&gt;&gt; print(v)
[ 3.0, 4.0, 0.0 ]
&gt;&gt;&gt;
</pre>
</div>
<p> The C++ 'this' pointer is often needed to access member variables,
methods etc. The <tt>$self</tt> special variable should be used
wherever you could use 'this'. The example above demonstrates this for
accessing member variables. Note that the members dereferenced by <tt>
$self</tt> must be public members as the code is ultimately generated
into a global function and so will not have any access to non-public
members. The implicit 'this' pointer that is present in C++ methods is
not present in <tt>%extend</tt> methods. In order to access anything in
the extended class or its base class, an explicit 'this' is required.
The following example shows how one could access base class members:</p>
<div class="code">
<pre>
struct Base {
virtual void method(int v) {
...
}
int value;
};
struct Derived : Base {
};
%extend Derived {
virtual void method(int v) {
$self-&gt;Base::method(v); // akin to this-&gt;Base::method(v);
$self-&gt;value = v; // akin to this-&gt;value = v;
...
}
}
</pre>
</div>
<p> The following special variables are expanded if used within a
%extend block: $name, $symname, $overname, $decl, $fulldecl,
$parentclassname and $parentclasssymname. The <a href="#Customization_exception_special_variables">
Special variables</a> section provides more information each of these
special variables.</p>
<p> The <tt>%extend</tt> directive follows all of the same conventions
as its use with C structures. Please refer to the <a href="#SWIG_adding_member_functions">
Adding member functions to C structures</a> section for further details.</p>
<h3><a name="SWIGPlus_replacing_methods">6.17.1 Replacing class methods</a>
</h3>
<p> Suppose there is a method in a class that you need to replace and
keep the method name the same. This can be achieved combining the <tt>
%extend</tt> and <tt>%ignore</tt> directives covered earlier. Here is an
example to replace the <tt>MyClass::mymethod()</tt>:<div class="code">
<pre>
%extend MyClass {
void mymethod() {
std::cout &lt;&lt; &quot;swig mymethod&quot; &lt;&lt; std::endl;
}
}
%ignore MyClass::mymethod;
%inline %{
class MyClass {
public:
void mymethod() {
std::cout &lt;&lt; &quot;class mymethod&quot; &lt;&lt; std::endl;
}
};
%}
</pre>
</div></p>
<p> Or if your code organization makes more sense to put the <tt>%extend</tt>
after the class definition, you would need the following:</p>
<div class="code">
<pre>
%rename(&quot;&quot;) MyClass::mymethod; // unignores the method
</pre>
</div>
<p> before the <tt>%extend</tt> or SWIG will continue to ignore <tt>
mymethod()</tt>, even in an <tt>%extend</tt>.</p>
<p> Note that you can call the class method from the method in <tt>
%extend</tt>, just use <tt>self-&gt;mymethod()</tt> and it will call the
class method, not the one in <tt>%extend</tt>.</p>
<h2><a name="SWIGPlus_nn30">6.18 Templates</a></h2>
<p> Template type names may appear anywhere a type is expected in an
interface file. For example:</p>
<div class="code">
<pre>
void foo(vector&lt;int&gt; *a, int n);
void bar(std::array&lt;int, 100&gt; *x);
</pre>
</div>
<p> There are some restrictions on the use of non-type arguments. Simple
literals are supported, and so are most constant expressions. However,
there are some limitations on the use of '&lt;' and '&gt;' in constant
expressions (but note that '&lt;=' and '&gt;=' are fully supported). For
example:</p>
<div class="code">
<pre>
void bar(std::array&lt;int, 100&gt; *x); // OK
void bar(std::array&lt;int, 2*50&gt; *x); // OK
void bar(std::array&lt;int, (1&lt;2 ? 100 : 50)&gt; *x) // OK
void bar(std::array&lt;int, 1&lt;2 ? 100 : 50&gt; *x) // Not supported
void bar(std::array&lt;int, (2&gt;1 ? 100 : 50)&gt; *x) // Not supported
</pre>
</div>
<p> The type system is smart enough to figure out clever games you might
try to play with <tt>typedef</tt>. For instance, consider this code:</p>
<div class="code">
<pre>
typedef int Integer;
void foo(vector&lt;int&gt; *x, vector&lt;Integer&gt; *y);
</pre>
</div>
<p> In this case, <tt>vector&lt;Integer&gt;</tt> is exactly the same type as <tt>
vector&lt;int&gt;</tt>. The wrapper for <tt>foo()</tt> will accept either
variant.</p>
<h3><a name="SWIGPlus_template_directive">6.18.1 The %template directive</a>
</h3>
<p> There are a couple of important points about template wrapping.
First, a bare C++ template does not define any sort of runnable
object-code for which SWIG can normally create a wrapper. Therefore, in
order to wrap a template, you need to give SWIG information about a
particular template instantiation (e.g., <tt>vector&lt;int&gt;</tt>, <tt>
array&lt;double&gt;</tt>, etc.). Second, an instantiation name such as <tt>
vector&lt;int&gt;</tt> is generally not a valid identifier name in most target
languages. Thus, you will need to give the template instantiation a
more suitable name such as <tt>intvector</tt>.</p>
<p> To illustrate, consider the following class template definition:</p>
<div class="code">
<pre>
template&lt;class T&gt; class List {
private:
T *data;
int nitems;
int maxitems;
public:
List(int max) {
data = new T [max];
nitems = 0;
maxitems = max;
}
~List() {
delete [] data;
};
void append(T obj) {
if (nitems &lt; maxitems) {
data[nitems++] = obj;
}
}
int length() {
return nitems;
}
T get(int n) {
return data[n];
}
};
</pre>
</div>
<p> By itself, this class template is useless--SWIG simply ignores it
because it doesn't know how to generate any code unless a definition of
<tt>T</tt> is provided. The <tt>%template</tt> directive is required to
instantiate the template for use in a target language. The directive
requires an identifier name for use in the target language plus the
template for instantiation. The example below instantiates <tt>
List&lt;int&gt;</tt> for use as a class named <tt>intList</tt>:</p>
<div class="code">
<pre>
%template(intList) List&lt;int&gt;;
</pre>
</div>
<p> The instantiation expands the template code as a C++ compiler would
do and then makes it available under the given identifier name.
Essentially it is the same as wrapping the following concept code where
the class template definition has <tt>T</tt> expanded to <tt>int</tt>
(note that this is not entirely valid syntax):</p>
<div class="code">
<pre>
%rename(intList) List&lt;int&gt;; // Rename to a suitable identifier
class List&lt;int&gt; {
private:
int *data;
int nitems;
int maxitems;
public:
List(int max);
~List();
void append(int obj);
int length();
int get(int n);
};
</pre>
</div>
<p> The argument to <tt>%template()</tt> is the name of the
instantiation in the target language. The name you choose should not
conflict with any other declarations in the interface file with one
exception---it is okay for the template name to match that of a typedef
declaration. For example:</p>
<div class="code">
<pre>
%template(intList) List&lt;int&gt;;
...
typedef List&lt;int&gt; intList; // OK
</pre>
</div>
<p> The <tt>%template</tt> directive must always appear<em> after</em>
the definition of the template to be expanded, so the following will
work:</p>
<div class="code">
<pre>
template&lt;class T&gt; class List { ... };
%template(intList) List&lt;int&gt;;
</pre>
</div>
<p> but if %template is used before the template definition, such as:</p>
<div class="code">
<pre>
%template(intList) List&lt;int&gt;;
template&lt;class T&gt; class List { ... };
</pre>
</div>
<p> SWIG will generate an error:</p>
<div class="shell">
<pre>
example.i:3: Error: Template 'List' undefined.
</pre>
</div>
<p> Since the type system knows how to handle <tt>typedef</tt>, it is
generally not necessary to instantiate different versions of a template
for typenames that are equivalent. For instance, consider this code:</p>
<div class="code">
<pre>
%template(intList) List&lt;int&gt;;
typedef int Integer;
...
void foo(List&lt;Integer&gt; *x);
</pre>
</div>
<p> In this case, <tt>List&lt;Integer&gt;</tt> is exactly the same type as <tt>
List&lt;int&gt;</tt>. Any use of <tt>List&lt;Integer&gt;</tt> is mapped back to the
instantiation of <tt>List&lt;int&gt;</tt> created earlier. Therefore, it is
not correct to instantiate a new class for the type <tt>Integer</tt>.
An attempt to do so such as:</p>
<div class="code">
<pre>
%template(intList) List&lt;int&gt;;
%template(IntegerList) List&lt;Integer&gt;; // Ignored
</pre>
</div>
<p> will result in the duplicate instantiation being ignored with a
warning:</p>
<div class="shell">
<pre>
example.i:48: Warning 404: Duplicate template instantiation of 'List&lt; Integer &gt;' with name 'IntegerList' ignored,
example.i:47: Warning 404: previous instantiation of 'List&lt; int &gt;' with name 'intList'.
</pre>
</div>
<p> The template provided to <tt>%template</tt> for instantiation must
be the actual template and not a typedef to a template.</p>
<div class="code">
<pre>
typedef List&lt;int&gt; ListOfInt;
%template(intList) List&lt;int&gt;; // ok
%template(intList) ListOfInt; // illegal - Syntax error
</pre>
</div>
<h3><a name="SWIGPlus_template_functions">6.18.2 Function templates</a></h3>
<p> SWIG can also generate wrappers for function templates using a
similar technique to that shown above for class templates. For example:</p>
<div class="code">
<pre>
// Function template
template&lt;class T&gt; T max(T a, T b) { return a &gt; b ? a : b; }
// Make some different versions of this function
%template(maxint) max&lt;int&gt;;
%template(maxdouble) max&lt;double&gt;;
</pre>
</div>
<p> In this case, <tt>maxint</tt> and <tt>maxdouble</tt> become unique
names for specific instantiations of the function.</p>
<p> SWIG even supports overloaded templated functions. As usual the <tt>
%template</tt> directive is used to wrap templated functions. For
example:</p>
<div class="code">
<pre>
template&lt;class T&gt; void foo(T x) { };
template&lt;class T&gt; void foo(T x, T y) { };
%template(foo) foo&lt;int&gt;;
</pre>
</div>
<p> This will generate two overloaded wrapper methods, the first will
take a single integer as an argument and the second will take two
integer arguments.</p>
<h3><a name="SWIGPlus_template_classes">6.18.3 Default template
arguments</a></h3>
<p> The number of arguments supplied to <tt>%template</tt> should match
that in the original template definition. Template default arguments
are supported. For example:</p>
<div class="code">
<pre>
template &lt;typename T, int max=100&gt; class vector {
...
};
%template(intvec) vector&lt;int&gt;; // OK
%template(vec1000) vector&lt;int, 1000&gt;; // OK
</pre>
</div>
<p> The <tt>%template</tt> directive should not be used to wrap the same
template instantiation more than once. This also applies to default
parameters where a template parameter specified in the instantiation is
the same as the default parameter. For example:</p>
<div class="code">
<pre>
%template(vec) vector&lt;double&gt;; // OK
%template(vec100) vector&lt;double, 100&gt;; // Ignored
</pre>
</div>
<p> will warn:</p>
<div class="shell">
<pre>
example.i:59: Warning 404: Duplicate template instantiation of 'vector&lt; double,100 &gt;' with name 'vec100' ignored,
example.i:58: Warning 404: previous instantiation of 'vector&lt; double &gt;' with name 'vec'.
</pre>
</div>
<p> If this was not ignored, the template expansion would result in two
identical classes. An identical instantiation is only wrapped once in
order to reduce code bloat.</p>
<p><b> Compatibility Note</b>: Versions prior to SWIG-4.2.0 would
sometimes not detect and prevent duplicate instantiations, such as when
the wrapped name was different.</p>
<h3><a name="SWIGPlus_template_class_inheritance">6.18.4 Template base
classes</a></h3>
<p> When a template is instantiated using <tt>%template</tt>,
information about that class is saved by SWIG and used elsewhere in the
program. For example, if you wrote code like this,</p>
<div class="code">
<pre>
...
%template(intList) List&lt;int&gt;;
...
class UltraList : public List&lt;int&gt; {
...
};
</pre>
</div>
<p> then SWIG knows that <tt>List&lt;int&gt;</tt> was already wrapped as a
class called <tt>intList</tt> and arranges to handle the inheritance
correctly. If, on the other hand, nothing is known about <tt>List&lt;int&gt;</tt>
, you will get a warning message similar to this:</p>
<div class="shell">
<pre>
example.h:42: Warning 401. Nothing known about class 'List&lt; int &gt;'. Ignored.
example.h:42: Warning 401. Maybe you forgot to instantiate 'List&lt; int &gt;' using %template.
</pre>
</div>
<p> If a class template inherits from another class template, you need
to make sure that base classes are instantiated before derived classes.
For example:</p>
<div class="code">
<pre>
template&lt;class T&gt; class Foo {
...
};
template&lt;class T&gt; class Bar : public Foo&lt;T&gt; {
...
};
// Instantiate base classes first
%template(intFoo) Foo&lt;int&gt;;
%template(doubleFoo) Foo&lt;double&gt;;
// Now instantiate derived classes
%template(intBar) Bar&lt;int&gt;;
%template(doubleBar) Bar&lt;double&gt;;
</pre>
</div>
<p> The order is important since SWIG uses the instantiation names to
properly set up the inheritance hierarchy in the resulting wrapper code
(and base classes need to be wrapped before derived classes). Don't
worry--if you get the order wrong, SWIG should generate a warning
message.</p>
<p> If you have to instantiate a lot of different classes for many
different types, you might consider writing a SWIG macro. For example:</p>
<div class="code">
<pre>
%define TEMPLATE_WRAP(prefix, T...)
%template(prefix ## Foo) Foo&lt;T &gt;;
%template(prefix ## Bar) Bar&lt;T &gt;;
...
%enddef
TEMPLATE_WRAP(int, int)
TEMPLATE_WRAP(double, double)
TEMPLATE_WRAP(String, char *)
TEMPLATE_WRAP(PairStringInt, std::pair&lt;string, int&gt;)
...
</pre>
</div>
<p> Note the use of a vararg macro for the type T. If this wasn't used,
the comma in the templated type in the last example would not be
possible.</p>
<h3><a name="SWIGPlus_template_empty">6.18.5 Empty template
instantiation</a></h3>
<p> Occasionally, you may need to tell SWIG about classes that are
defined by templates, but which aren't supposed to be wrapped. Since
SWIG is not able to automatically instantiate templates for this
purpose, you must do it manually. To do this, simply use <tt>
%template()</tt>, that is the empty template instantiation that omits
providing a name. For example:</p>
<div class="code">
<pre>
template&lt;typename T&gt; struct Traits {
typedef T type;
};
%}
%template() Traits&lt;int&gt;; // instantiate Traits&lt;int&gt;, but don't wrap it
void traitor(Traits&lt;int&gt;::type val);
</pre>
</div>
<p> Without a template instantiation, SWIG does not know that the first
parameter to the <tt>traitor</tt> function is type int and passing an
integer to this function from any target language won't work. The empty
template instantiation adds the appropriate type information into
SWIG's type system, without forcing one to wrap the <tt>Traits</tt>
class.</p>
<p> Duplicate template instantiation are not allowed, as described in
the <a href="#SWIGPlus_template_classes">Default template arguments</a>
section above. There is one exception where a named template
instantiation can be followed by an empty template instantiation.
Duplicate empty template instantiations are silently ignored, unlike
duplicate named template instantiations.</p>
<p> Unlike template class instantiations, template function
instantiations must have a name. Consider the following:</p>
<div class="code">
<pre>
template&lt;class T&gt; T tfunc(T x) { };
%template() tfunc&lt;double&gt;;
</pre>
</div>
<p> The empty template instantiation will be ignored with:</p>
<div class="shell">
<pre>
example.i:9: Warning 519: %template() contains no name. Template method ignored: tfunc&lt; double &gt;(double)
</pre>
</div>
<h3><a name="SWIGPlus_template_specialization">6.18.6 Template
specialization</a></h3>
<p> The SWIG template mechanism<em> does</em> support specialization.
For instance, if you define a class like this,</p>
<div class="code">
<pre>
template&lt;&gt; class List&lt;int&gt; {
private:
int *data;
int nitems;
int maxitems;
public:
List(int max);
~List();
void append(int obj);
int length();
int get(int n);
};
</pre>
</div>
<p> then SWIG will use this code whenever the user expands <tt>List&lt;int&gt;</tt>
. In practice, this may have very little effect on the underlying
wrapper code since specialization is often used to provide slightly
modified method bodies (which are ignored by SWIG). However, special
SWIG directives such as <tt>%typemap</tt>, <tt>%extend</tt>, and so
forth can be attached to a specialization to provide customization for
specific types.</p>
<p> Partial template specialization is partially supported by SWIG. For
example, this code defines a template that is applied when the template
argument is a pointer.</p>
<div class="code">
<pre>
template&lt;class T&gt; class List&lt;T*&gt; {
private:
T *data;
int nitems;
int maxitems;
public:
List(int max);
~List();
void append(T obj);
int length();
T get(int n);
};
</pre>
</div>
<p> SWIG supports both template explicit specialization and partial
specialization. Consider:</p>
<div class="code">
<pre>
template&lt;class T1, class T2&gt; class Foo { }; // (1) primary template
template&lt;&gt; class Foo&lt;double *, int *&gt; { }; // (2) explicit specialization
template&lt;class T1, class T2&gt; class Foo&lt;T1, T2 *&gt; { }; // (3) partial specialization
</pre>
</div>
<p> SWIG is able to properly match explicit instantiations:</p>
<div class="code">
<pre>
<tt>Foo&lt;double *, int *&gt;</tt> // explicit specialization matching (2)
</pre>
</div>
<p> SWIG implements template argument deduction so that the following
partial specialization examples work just like they would with a C++
compiler:</p>
<div class="code">
<pre>
<tt>Foo&lt;int *, int *&gt;</tt> // partial specialization matching (3)
<tt>Foo&lt;int *, const int *&gt;</tt> // partial specialization matching (3)
<tt>Foo&lt;int *, int **&gt;</tt> // partial specialization matching (3)
</pre>
</div>
<h3><a name="SWIGPlus_template_member">6.18.7 Member templates</a></h3>
<p> Member templates are supported. The underlying principle is the same
as for normal templates--SWIG can't create a wrapper unless you provide
more information about types. For example, a class with a member
function template might look like this:</p>
<div class="code">
<pre>
class Foo {
public:
template&lt;class T&gt; void bar(T x, T y) { ... };
...
};
</pre>
</div>
<p> To expand the template, simply use <tt>%template</tt> inside the
class.</p>
<div class="code">
<pre>
class Foo {
public:
template&lt;class T&gt; void bar(T x, T y) { ... };
...
%template(barint) bar&lt;int&gt;;
%template(bardouble) bar&lt;double&gt;;
};
</pre>
</div>
<p> Or, if you want to leave the original class definition alone, just
do this:</p>
<div class="code">
<pre>
class Foo {
public:
template&lt;class T&gt; void bar(T x, T y) { ... };
...
};
...
%extend Foo {
%template(barint) bar&lt;int&gt;;
%template(bardouble) bar&lt;double&gt;;
};
</pre>
</div>
<p> or simply</p>
<div class="code">
<pre>
class Foo {
public:
template&lt;class T&gt; void bar(T x, T y) { ... };
...
};
...
%template(bari) Foo::bar&lt;int&gt;;
%template(bard) Foo::bar&lt;double&gt;;
</pre>
</div>
<p> In this case, the <tt>%extend</tt> directive is not needed, and <tt>
%template</tt> does exactly the same job, i.e., it adds two new methods
to the Foo class.</p>
<p> Now, if your target language supports overloading, you can even try</p>
<div class="code">
<pre>
%template(bar) Foo::bar&lt;int&gt;;
%template(bar) Foo::bar&lt;double&gt;;
</pre>
</div>
<p> and since the two new wrapped methods have the same name 'bar', they
will be overloaded, and when called, the correct method will be
dispatched depending on the argument type.</p>
<p> When used with members, the <tt>%template</tt> directive may be
placed in another class template. Here is a slightly perverse example:</p>
<div class="code">
<pre>
// A template
template&lt;class T&gt; class Foo {
public:
// A member template
template&lt;class S&gt; T bar(S x, S y) { ... };
...
};
// Expand a few member templates
%extend Foo {
%template(bari) bar&lt;int&gt;;
%template(bard) bar&lt;double&gt;;
}
// Create some wrappers for the template
%template(Fooi) Foo&lt;int&gt;;
%template(Food) Foo&lt;double&gt;;
</pre>
</div>
<p> Miraculously, you will find that each expansion of <tt>Foo</tt> has
member functions <tt>bari()</tt> and <tt>bard()</tt> added.</p>
<p> A common use of member templates is to define constructors for
copies and conversions. For example:</p>
<div class="code">
<pre>
template&lt;class T1, class T2&gt; struct pair {
T1 first;
T2 second;
pair() : first(T1()), second(T2()) { }
pair(const T1 &amp;x, const T2 &amp;y) : first(x), second(y) { }
template&lt;class U1, class U2&gt; pair(const pair&lt;U1, U2&gt; &amp;x)
: first(x.first), second(x.second) { }
};
</pre>
</div>
<p> This declaration is perfectly acceptable to SWIG, but the
constructor template will be ignored unless you explicitly expand it.
To do that, you could expand a few versions of the constructor in the
class template itself. For example:</p>
<div class="code">
<pre>
%extend pair {
%template(pair) pair&lt;T1, T2&gt;; // Generate default copy constructor
};
</pre>
</div>
<p> When using <tt>%extend</tt> in this manner, notice how you can still
use the template parameters in the original template definition.</p>
<p> Alternatively, you could expand the constructor template in selected
instantiations. For example:</p>
<div class="code">
<pre>
// Instantiate a few versions
%template(pairii) pair&lt;int, int&gt;;
%template(pairdd) pair&lt;double, double&gt;;
// Create a default constructor only
%extend pair&lt;int, int&gt; {
%template(paird) pair&lt;int, int&gt;; // Default constructor
};
// Create default and conversion constructors
%extend pair&lt;double, double&gt; {
%template(paird) pair&lt;double, double&gt;; // Default constructor
%template(pairc) pair&lt;int, int&gt;; // Conversion constructor
};
</pre>
</div>
<p>And if your target language supports overloading, then you can try
instead:</p>
<div class="code">
<pre>
// Create default and conversion constructors
%extend pair&lt;double, double&gt; {
%template(pair) pair&lt;double, double&gt;; // Default constructor
%template(pair) pair&lt;int, int&gt;; // Conversion constructor
};
</pre>
</div>
<p> In this case, the default and conversion constructors have the same
name. Hence, SWIG will overload them and define an unique visible
constructor, that will dispatch the proper call depending on the
argument type.</p>
<h3><a name="SWIGPlus_template_scoping">6.18.8 Scoping and templates</a></h3>
<p> The <tt>%template</tt> directive for a class template is the
equivalent to an explicit instantiation of a C++ class template. The
scope for a valid <tt>%template</tt> instantiation is the same as the
scope required for a valid explicit instantiation of a C++ template. A
definition of the template for the explicit instantiation must be in
scope where the instantiation is declared and must not be enclosed
within a different namespace.</p>
<p> For example, a few <tt>%template</tt> instantiations and C++
explicit instantiations are shown below:</p>
<div class="code">
<pre>
namespace N {
template&lt;typename T&gt; class C {};
}
// valid
%template(cin) N::C&lt;int&gt;;
template class N::C&lt;int&gt;;
// valid
namespace N {
%template(cin) C&lt;int&gt;;
template class C&lt;int&gt;;
}
// valid
using namespace N;
%template(cin) C&lt;int&gt;;
template class C&lt;int&gt;;
// valid
using N::C;
%template(cin) C&lt;int&gt;;
template class C&lt;int&gt;;
// ill-formed
namespace unrelated {
using N::C;
%template(cin) C&lt;int&gt;;
template class C&lt;int&gt;;
}
// ill-formed
namespace unrelated {
using namespace N;
%template(cin) C&lt;int&gt;;
template class C&lt;int&gt;;
}
// ill-formed
namespace unrelated {
namespace N {
%template(cin) C&lt;int&gt;;
template class C&lt;int&gt;;
}
}
// ill-formed
namespace unrelated {
%template(cin) N::C&lt;int&gt;;
template class N::C&lt;int&gt;;
}
</pre>
</div>
<p> When the scope is incorrect, such as for the ill-formed examples
above, an error occurs:</p>
<div class="shell">
<pre>
cpp_template_scope.i:34: Error: 'C' resolves to 'N::C' and was incorrectly instantiated
in scope 'unrelated' instead of within scope 'N'.
</pre>
</div>
<p> A note for the C++ standard geeks out there; a valid instantiation
is one which conforms to the C++03 standard as C++11 made a change to
disallow using declarations and using directives to find a template.</p>
<div class="code">
<pre>
// valid C++03, ill-formed C++11
using N::C;
template class C&lt;int&gt;;
</pre>
</div>
<p><b> Compatibility Note</b>: Versions prior to SWIG-4.0.0 did not
error out with incorrectly scoped <tt>%template</tt> declarations, but
this led to numerous subtle template scope problems.</p>
<h3><a name="SWIGPlus_template_more">6.18.9 More on templates</a></h3>
<p> If all of this isn't quite enough and you really want to make
someone's head explode, SWIG directives such as <tt>%rename</tt>, <tt>
%extend</tt>, and <tt>%typemap</tt> can be included directly in template
definitions. For example:</p>
<div class="code">
<pre>
// File : list.h
template&lt;class T&gt; class List {
...
public:
%rename(__getitem__) get(int);
List(int max);
~List();
...
T get(int index);
%extend {
char *__str__() {
/* Make a string representation */
...
}
}
};
</pre>
</div>
<p> In this example, the extra SWIG directives are propagated to<em>
every</em> template instantiation.</p>
<p> It is also possible to separate these declarations from the class
template. For example:</p>
<div class="code">
<pre>
%rename(__getitem__) List::get;
%extend List {
char *__str__() {
/* Make a string representation */
...
}
/* Make a copy */
T *__copy__() {
return new List&lt;T&gt;(*$self);
}
};
...
template&lt;class T&gt; class List {
...
public:
List() { }
T get(int index);
...
};
</pre>
</div>
<p> When <tt>%extend</tt> is decoupled from the class definition, it is
legal to use the same template parameters as provided in the class
definition. These are replaced when the template is expanded. In
addition, the <tt>%extend</tt> directive can be used to add additional
methods to a specific instantiation. For example:</p>
<div class="code">
<pre>
%template(intList) List&lt;int&gt;;
%extend List&lt;int&gt; {
void blah() {
printf(&quot;Hey, I'm an List&lt;int&gt;!\n&quot;);
}
};
</pre>
</div>
<p> It is even possible to extend a class via <tt>%extend</tt> with
template methods, for example:</p>
<div class="code">
<pre>
%include &lt;std_string.i&gt;
%inline %{
class ExtendMe {
public:
template &lt;typename T&gt;
T do_stuff_impl(int a, T b, double d) {
return b;
}
};
%}
%extend ExtendMe {
template&lt;typename T&gt;
T do_overloaded_stuff(T b) {
return $self-&gt;do_stuff_impl(0, b, 4.0);
}
}
%template(do_overloaded_stuff) ExtendMe::do_overloaded_stuff&lt;std::string&gt;;
%template(do_overloaded_stuff) ExtendMe::do_overloaded_stuff&lt;double&gt;;
</pre>
</div>
<p> The wrapped <tt>ExtendMe</tt> class will then have two (overloaded)
methods called <tt>do_overloaded_stuff</tt>.</p>
<p><b> Compatibility Note</b>: Extending a class with template methods
was added in version 3.0.12</p>
<p> Needless to say, SWIG's template support provides plenty of
opportunities to break the universe. That said, an important final
point is that<b> SWIG does not perform extensive error checking of
templates!</b> Specifically, SWIG does not perform type checking nor
does it check to see if the actual contents of the template declaration
make any sense. Since the C++ compiler checks this when it compiles the
resulting wrapper file, there is no practical reason for SWIG to
duplicate this functionality.</p>
<a name="SWIGPlus_template_nested_class_example"></a>
<p> As SWIG's template support does not perform type checking <tt>
%template</tt> can be used as early as after a template declaration. You
can, and rarely have to, use <tt>%template</tt> before the template
parameters have been declared. For example:</p>
<div class="code">
<pre>
template &lt;class T&gt; class OuterTemplateClass {};
// The nested class OuterClass::InnerClass inherits from the class template
// OuterTemplateClass&lt;OuterClass::InnerStruct&gt; and thus the template needs
// to be expanded with %template before the OuterClass declaration.
%template(OuterTemplateClass_OuterClass__InnerStruct)
OuterTemplateClass&lt;OuterClass::InnerStruct&gt;
// Don't forget to use %feature(&quot;flatnested&quot;) for OuterClass::InnerStruct and
// OuterClass::InnerClass if the target language doesn't support nested classes.
class OuterClass {
public:
// Forward declarations:
struct InnerStruct;
class InnerClass;
};
struct OuterClass::InnerStruct {};
// Expanding the template at this point with %template is too late as the
// OuterClass::InnerClass declaration is processed inside OuterClass.
class OuterClass::InnerClass : public OuterTemplateClass&lt;InnerStruct&gt; {};
</pre>
</div>
<p><b> Compatibility Note</b>: The first implementation of template
support relied heavily on macro expansion in the preprocessor.
Templates have been more tightly integrated into the parser and type
system in SWIG-1.3.12 and the preprocessor is no longer used. Code that
relied on preprocessing features in template expansion will no longer
work. However, SWIG still allows the # operator to be used to generate
a string from a template argument.</p>
<p><b> Compatibility Note</b>: In earlier versions of SWIG, the <tt>
%template</tt> directive introduced a new class name. This name could
then be used with other directives. For example:</p>
<div class="code">
<pre>
%template(vectori) vector&lt;int&gt;;
%extend vectori {
void somemethod() { }
};
</pre>
</div>
<p> This behavior is no longer supported. Instead, you should use the
original template name as the class name. For example:</p>
<div class="code">
<pre>
%template(vectori) vector&lt;int&gt;;
%extend vector&lt;int&gt; {
void somemethod() { }
};
</pre>
</div>
<p> Similar changes apply to typemaps and other customization features.</p>
<h2><a name="SWIGPlus_namespaces">6.19 Namespaces</a></h2>
<p> Support for C++ namespaces is comprehensive, but by default simple,
however, some target languages can turn on more advanced namespace
support via the <a href="#SWIGPlus_nspace">nspace feature</a>,
described later. Code within unnamed namespaces is ignored as there is
no external access to symbols declared within the unnamed namespace.
Before detailing the default implementation for named namespaces, it is
worth noting that the semantics of C++ namespaces is extremely
non-trivial--especially with regard to the C++ type system and class
machinery. At a most basic level, namespaces are sometimes used to
encapsulate common functionality. For example:</p>
<div class="code">
<pre>
namespace math {
double sin(double);
double cos(double);
class Complex {
double im, re;
public:
...
};
...
};
</pre>
</div>
<p> Members of the namespace are accessed in C++ by prepending the
namespace prefix to names. For example:</p>
<div class="code">
<pre>
double x = math::sin(1.0);
double magnitude(math::Complex *c);
math::Complex c;
...
</pre>
</div>
<p> At this level, namespaces are relatively easy to manage. However,
things start to get very ugly when you throw in the other ways a
namespace can be used. For example, selective symbols can be exported
from a namespace with a <tt>using</tt> declaration:</p>
<div class="code">
<pre>
using math::Complex; // Using declaration
double magnitude(Complex *c); // Namespace prefix stripped
</pre>
</div>
<p> Similarly, the contents of an entire namespace can be made available
via a <tt>using</tt> directive:</p>
<div class="code">
<pre>
using namespace math; // Using directive
double x = sin(1.0);
double magnitude(Complex *c);
</pre>
</div>
<p> Alternatively, a namespace can be aliased:</p>
<div class="code">
<pre>
namespace M = math;
double x = M::sin(1.0);
double magnitude(M::Complex *c);
</pre>
</div>
<p> Using combinations of these features, it is possible to write
head-exploding code like this:</p>
<div class="code">
<pre>
namespace A {
class Foo {
};
}
namespace B {
namespace C {
using namespace A;
}
typedef C::Foo FooClass;
}
namespace BIGB = B;
namespace D {
using BIGB::FooClass;
class Bar : public FooClass {
}
};
class Spam : public D::Bar {
};
void evil(A::Foo *a, B::FooClass *b, B::C::Foo *c, BIGB::FooClass *d,
BIGB::C::Foo *e, D::FooClass *f);
</pre>
</div>
<p> Given the possibility for such perversion, it's hard to imagine how
every C++ programmer might want such code wrapped into the target
language. Clearly this code defines three different classes. However,
one of those classes is accessible under at least six different names!</p>
<p> SWIG fully supports C++ namespaces in its internal type system and
class handling code. If you feed SWIG the above code, it will be parsed
correctly, it will generate compilable wrapper code, and it will
produce a working scripting language module. However, the default
wrapping behavior is to flatten namespaces in the target language. This
means that the contents of all namespaces are merged together in the
resulting scripting language module. For example, if you have code like
this,</p>
<div class="code">
<pre>
%module foo
namespace foo {
void bar(int);
void spam();
}
namespace bar {
void blah();
}
</pre>
</div>
<p> then SWIG simply creates three wrapper functions <tt>bar()</tt>, <tt>
spam()</tt>, and <tt>blah()</tt> in the target language. SWIG does not
prepend the names with a namespace prefix nor are the functions
packaged in any kind of nested scope. Note that the default handling of
flattening all the namespace scopes in the target language can be
changed via the <a href="#SWIGPlus_nspace">nspace feature</a>.</p>
<p> There is some rationale for taking this approach. Since C++
namespaces are often used to define modules in C++, there is a natural
correlation between the likely contents of a SWIG module and the
contents of a namespace. For instance, it would not be unreasonable to
assume that a programmer might make a separate extension module for
each C++ namespace. In this case, it would be redundant to prepend
everything with an additional namespace prefix when the module itself
already serves as a namespace in the target language. Or put another
way, if you want SWIG to keep namespaces separate, simply wrap each
namespace with its own SWIG interface.</p>
<p> Because namespaces are flattened, it is possible for symbols defined
in different namespaces to generate a name conflict in the target
language. For example:</p>
<div class="code">
<pre>
namespace A {
void foo(int);
}
namespace B {
void foo(double);
}
</pre>
</div>
<p> When this conflict occurs, you will get an error message that
resembles this:</p>
<div class="shell">
<pre>
example.i:26: Error: 'foo' is multiply defined in the generated target language module.
example.i:23: Error: Previous declaration of 'foo'
</pre>
</div>
<p> To resolve this error, simply use <tt>%rename</tt> to disambiguate
the declarations. For example:</p>
<div class="code">
<pre>
%rename(B_foo) B::foo;
...
namespace A {
void foo(int);
}
namespace B {
void foo(double); // Gets renamed to B_foo
}
</pre>
</div>
<p> Similarly, <tt>%ignore</tt> can be used to ignore declarations.</p>
<p> C++ <tt>using</tt> directives and <tt>using</tt> declarations do not
add any code to the generated wrapper code. However, there is an
exception in one context, see <a href="#SWIGPlus_nn35">Using
declarations and inheritance</a> for introducing members of a base
class into a derived class definition. C++ <tt>using</tt> declarations
and directives<em> are</em> used by the internal type system to track
type-names. Therefore, if you have code like this:</p>
<div class="code">
<pre>
namespace A {
typedef int Integer;
}
using namespace A;
void foo(Integer x);
</pre>
</div>
<p> SWIG knows that <tt>Integer</tt> is the same as <tt>A::Integer</tt>
which is the same as <tt>int</tt>.</p>
<p> Namespaces may be combined with templates. If necessary, the <tt>
%template</tt> directive can be used to expand a template defined in a
different namespace. For example:</p>
<div class="code">
<pre>
namespace foo {
template&lt;typename T&gt; T max(T a, T b) { return a &gt; b ? a : b; }
}
using foo::max;
%template(maxint) max&lt;int&gt;; // Okay.
%template(maxfloat) foo::max&lt;float&gt;; // Okay (qualified name).
namespace bar {
using namespace foo;
%template(maxdouble) max&lt;double&gt;; // Okay.
}
</pre>
</div>
<p> The combination of namespaces and other SWIG directives may
introduce subtle scope-related problems. The key thing to keep in mind
is that all SWIG generated wrappers are produced in the<em> global</em>
namespace. Symbols from other namespaces are always accessed using
fully qualified names---names are never imported into the global space
unless the interface happens to do so with a <tt>using</tt>
declaration. In almost all cases, SWIG adjusts typenames and symbols to
be fully qualified. However, this is not done in code fragments such as
function bodies, typemaps, exception handlers, and so forth. For
example, consider the following:</p>
<div class="code">
<pre>
namespace foo {
typedef int Integer;
class bar {
public:
...
};
}
%extend foo::bar {
Integer add(Integer x, Integer y) {
Integer r = x + y; // Error. Integer not defined in this scope
return r;
}
};
</pre>
</div>
<p> In this case, SWIG correctly resolves the added method parameters
and return type to <tt>foo::Integer</tt>. However, since function
bodies aren't parsed and such code is emitted in the global namespace,
this code produces a compiler error about <tt>Integer</tt>. To fix the
problem, make sure you use fully qualified names. For example:</p>
<div class="code">
<pre>
%extend foo::bar {
Integer add(Integer x, Integer y) {
foo::Integer r = x + y; // Ok.
return r;
}
};
</pre>
</div>
<p><b> Note:</b> SWIG does<em> not</em> propagate <tt>using</tt>
declarations to the resulting wrapper code. If these declarations
appear in an interface, they should<em> also</em> appear in any header
files that might have been included in a <tt>%{ ... %}</tt> section. In
other words, don't insert extra <tt>using</tt> declarations into a SWIG
interface unless they also appear in the underlying C++ code.</p>
<p><b> Note:</b> Code inclusion directives such as <tt>%{ ... %}</tt> or
<tt>%inline %{ ... %}</tt> should not be placed inside a namespace
declaration. The code emitted by these directives will not be enclosed
in a namespace and you may get very strange results. If you need to use
namespaces with these directives, consider the following:</p>
<div class="code">
<pre>
// Good version
%inline %{
namespace foo {
void bar(int) { ... }
...
}
%}
// Bad version. Emitted code not placed in namespace.
namespace foo {
%inline %{
void bar(int) { ... } /* I'm bad */
...
%}
}
</pre>
</div>
<p><b> Note:</b> When the <tt>%extend</tt> directive is used inside a
namespace, the namespace name is included in the generated functions.
For example, if you have code like this,</p>
<div class="code">
<pre>
namespace foo {
class bar {
public:
%extend {
int blah(int x);
};
};
}
</pre>
</div>
<p> the added method <tt>blah()</tt> is mapped to a function <tt>int
foo_bar_blah(foo::bar *self, int x)</tt>. This function resides in the
global namespace.</p>
<p><b> Note:</b> Although namespaces are flattened in the target
language, the SWIG generated wrapper code observes the same namespace
conventions as used in the input file. Thus, if there are no symbol
conflicts in the input, there will be no conflicts in the generated
code.</p>
<p><b> Note:</b> In the same way that no resolution is performed on
parameters, a conversion operator name must match exactly to how it is
defined. Do not change the qualification of the operator. For example,
suppose you had an interface like this:</p>
<div class="code">
<pre>
namespace foo {
class bar;
class spam {
public:
...
operator bar(); // Conversion of spam -&gt; bar
...
};
}
</pre>
</div>
<p> The following is how the feature is expected to be written for a
successful match:</p>
<div class="code">
<pre>
%rename(tofoo) foo::spam::operator bar();
</pre>
</div>
<p> The following does not work as no namespace resolution is performed
in the matching of conversion operator names:</p>
<div class="code">
<pre>
%rename(tofoo) foo::spam::operator <b>foo::</b>bar();
</pre>
</div>
<p> Note, however, that if the operator is defined using a qualifier in
its name, then the feature must use it too...</p>
<div class="code">
<pre>
%rename(tofoo) foo::spam::operator bar(); // will not match
%rename(tofoo) foo::spam::operator foo::bar(); // will match
namespace foo {
class bar;
class spam {
public:
...
operator foo::bar();
...
};
}
</pre>
</div>
<p><b> Compatibility Note:</b> Versions of SWIG prior to 1.3.32 were
inconsistent in this approach. A fully qualified name was usually
required, but would not work in some situations.</p>
<p><b> Note:</b> The flattening of namespaces is only intended to serve
as a basic namespace implementation. More advanced handling of
namespaces is discussed next.</p>
<h3><a name="SWIGPlus_nspace">6.19.1 The nspace feature for namespaces</a>
</h3>
<p> The <tt>nspace</tt> feature operates in two modes. Firstly, in a
simple enable/disable mode to mirror the C++ namespaces into the target
language specific concept of a C++ namespace. Secondly, a complex mode
for modifying, renaming or moving the hierarchies of the language
specific concept of a C++ namespace.</p>
<p> The types of symbols that the feature can be applied to are any
class, struct, union or enum declared within a named namespace. This
also includes templates of the aforementioned types. The feature wraps
the type within the target language specific concept of a namespace,
such as a Java package or C# namespace. Only some target languages
provide support for the <tt>nspace</tt> feature. Please see the target
language specific sections to see if the language you are interested in
supports the nspace feature.</p>
<h4><a name="SWIGPlus_nspace_feature_flag">6.19.1.1 %nspace for
mirroring namespace hierarchies</a></h4>
<p> In this simple mode the <tt>nspace</tt> feature works as a <a href="#Customization_feature_flags">
feature flag</a> to enable or disable the feature for a given C++
symbol. As described earlier, all namespace are flattened by default,
hence the <tt>nspace</tt> feature is disabled by default. When the
feature is enabled, the C++ namespace hierarchies are mirrored into the
target language concept of a namespace.</p>
<p> The feature flag is demonstrated below using the following example:</p>
<div class="code">
<pre>
%feature(&quot;nspace&quot;) MyWorld::Material::Color;
%nspace MyWorld::Wrapping::Color; // %nspace is a macro for %feature(&quot;nspace&quot;)
namespace MyWorld {
namespace Material {
class Color {
...
};
}
namespace Wrapping {
class Color {
...
};
}
}
</pre>
</div>
<p> By default, without the above <tt>nspace</tt> feature flags (or an
appropriate <tt>%rename</tt>), SWIG outputs the following warning
resulting in just one of the <tt>Color</tt> classes being available for
use from the target language:</p>
<div class="shell">
<pre>
example.i:9: Error: 'Color' is multiply defined in the generated target language module.
example.i:5: Error: Previous declaration of 'Color'
</pre>
</div>
<p> Let's consider C# as the target language. With the two <tt>nspace</tt>
feature flags, the two C# <tt>Color</tt> proxy classes are generated
into the equivalent C# namespaces that mirror the C++ namespace
hierarchies. A fully qualified constructor call of each these two types
in C# is then:</p>
<div class="targetlang">
<pre>
var materialColor = new MyWorld.Material.Color();
var wrappingColor = new MyWorld.Wrapping.Color();
</pre>
</div>
<p> Without the <tt>nspace</tt> feature flag, no namespaces are
available in C# and the fully qualified constructor call of the type in
C# would simply be:</p>
<div class="targetlang">
<pre>
var materialColor = new Color();
</pre>
</div>
<p> Note that the <tt>nspace</tt> feature does not apply to variables
and functions simply declared in a namespace. For example, the
following symbols cannot co-exist in the target language without
renaming. This may change in a future version.</p>
<div class="code">
<pre>
namespace MyWorld {
namespace Material {
int quantity;
void dispatch();
}
namespace Wrapping {
int quantity;
void dispatch();
}
}
</pre>
</div>
<p><b> Compatibility Note:</b> The simple <tt>%nspace</tt> feature flag
was first introduced in SWIG-2.0.0.</p>
<h4><a name="SWIGPlus_nspacemove">6.19.1.2 %nspacemove for modifying
namespace hierarchies</a></h4>
<p> The more complex mode for <tt>nspace</tt> provides the ability to
move a type into a differently named target language equivalent of a
namespace. This allows a fully flexible approach to mapping C++
namespaces into a target language equivalent of a namespace, such as:</p>
<ul>
<li> Renaming a namespace.</li>
<li> Renaming any number of sub-namespaces of a namespace.</li>
<li> Moving a type to a completely different namespace.</li>
<li> Splitting many types in one namespace into multiple different
namespaces.</li>
<li> Adding nested sub-namespaces to a namespace.</li>
<li> Removing nested sub-namespaces of namespace.</li>
<li> Combinations of the above.</li>
</ul>
<p> The <tt>nspace</tt> feature attaches to a C++ symbol and provides a
target language namespace for that symbol to move to. The <tt>
%nspacemove</tt> directive is a macro for the <tt>nspace</tt> feature as
follows:</p>
<div class="code">
<pre>
#define %nspacemove(NAMESPACE) %feature(&quot;nspace&quot;, #NAMESPACE)
</pre>
</div>
<p> where the target namespace is provided in <tt>NAMESPACE</tt> and is
specified in valid C++ syntax, such as <tt>A::B::C</tt>. Internally,
SWIG converts the C++ namespace into a target language equivalent, so
this might for example, result in a C# namespace called <tt>A.B.C</tt>
or a Java package named <tt>A.B.C</tt>. Note that the target namespace
does not have to actually exist in any of the C++ code that SWIG
parses; it could be entirely made-up.</p>
<p> Either the <tt>%nspacemove</tt> macro or the more verbose <tt>
%feature</tt> syntax can be used. Please see the <a href="#Customization_features">
Features and the %feature directive</a> section for more details of SWIG
features.</p>
<p> An example follows:</p>
<div class="code">
<pre>
// Enable the nspace feature flag for all symbols
%nspace;
// Override the nspace feature for a subset of symbols by moving them into different namespaces
%nspacemove(A) A::B::C::Struct1;
%feature(&quot;nspace&quot;, &quot;A::B::X&quot;) A::B::C::Struct2;
%nspacemove(A::B::C::D) A::B::C::Struct4;
%nspacemove(Somewhere::Else) A::B::C::Struct5;
%inline %{
namespace A {
namespace B {
namespace C {
struct Struct1 {
// ...
};
struct Struct2 {
// ...
};
struct Struct3 {
// ...
};
struct Struct4 {
// ...
};
struct Struct5 {
// ...
};
}
}
}
%}
</pre>
</div>
<p> The C# code below constructs each of the above classes. It shows the
namespaces that the C# proxy classes have been moved into, noting
though that <tt>Struct4</tt> merely exactly mirrors the C++ namespace
hierarchy as it has the <tt>nspace</tt> feature flag attached to it.</p>
<div class="targetlang">
<pre>
var s1 = new A.Struct1();
var s2 = new A.B.X.Struct2();
var s3 = new A.B.C.Struct3();
var s4 = new A.B.C.D.Struct4();
var s5 = new Somewhere.Else.Struct5();
</pre>
</div>
<h4><a name="SWIGPlus_nspace_more">6.19.1.3 More about the nspace
feature</a></h4>
<p> When the <tt>nspace</tt> feature is attached to a class or enum, all
contained symbols (members) are also automatically moved into the
target language namespace. Contained symbols include all enum values,
static and non-static class members as well as nested classes. There is
no need for additional <tt>nspace</tt> features to be specified for all
the contained symbols. Below is an example showing this. It also shows
the nspace feature working for templates.</p>
<div class="code">
<pre>
// Easy way to move all the Structure template instantiations into a different namespace
%nspacemove(A::Different::Space) Space::Structure;
%inline %{
namespace Space {
template&lt;typename T&gt;
struct Structure {
static int Count;
static void StaticMethod(T t) {
// ...
}
struct NestedStruct {
// ...
};
};
template&lt;typename T&gt;
int Structure&lt;T&gt;::Count = 0;
}
%}
%template(StructureInt) Space::Structure&lt;int&gt;;
%template(StructureString) Space::Structure&lt;const char *&gt;;
</pre>
</div>
<p> The C# code below shows the full C# namespace <tt>A.Different.Space</tt>
being used as one would expect for all the contained symbols within a
C# class.</p>
<div class="targetlang">
<pre>
var s = new A.Different.Space.StructureInt();
int count = A.Different.Space.StructureInt.Count;
var n = new A.Different.Space.StructureInt.NestedStruct();
A.Different.Space.StructureInt.StaticMethod(99);
A.Different.Space.StructureString.StaticMethod(&quot;hi&quot;);
</pre>
</div>
<p> Any attempt to give a different namespace value to a nested class or
enum will issue a warning. For example, adding the following to the
above in an attempt to move one of the instantiated nested classes into
another namespace like this:</p>
<div class="code">
<pre>
%nspacemove(Bad::Space) Space::Structure&lt;int&gt;::NestedStruct;
... rest of example above ...
</pre>
</div>
<p> will result in the following warning:</p>
<div class="shell">
<pre>
Warning 406: Ignoring nspace setting (Bad::Space) for 'Space::Structure&lt; int &gt;::NestedStruct',
Warning 406: as it conflicts with the nspace setting (A::Different::Space) for outer class 'Space::Structure&lt; int &gt;'.
</pre>
</div>
<p><b> Compatibility Note:</b> Modifying namespace hierarchies via <tt>
%nspacemove</tt> was first introduced in SWIG-4.3.0.</p>
<h2><a name="SWIGPlus_renaming_templated_types_namespaces">6.20 Renaming
templated types in namespaces</a></h2>
<p> As has been mentioned, when %rename includes parameters, the
parameter types must match exactly (no typedef or namespace resolution
is performed). SWIG treats templated types slightly differently and has
an additional matching rule so unlike non-templated types, an exact
match is not always required. If the fully qualified templated type is
specified, it will have a higher precedence over the generic template
type. In the example below, the generic template type is used to rename
to <tt>bbb</tt> and the fully qualified type is used to rename to <tt>
ccc</tt>.</p>
<div class="code">
<pre>
%rename(bbb) Space::ABC::aaa(T t); // will match but with lower precedence than ccc
%rename(ccc) Space::ABC&lt;Space::XYZ&gt;::aaa(Space::XYZ t);// will match but with higher precedence
// than bbb
namespace Space {
class XYZ {};
template&lt;typename T&gt; struct ABC {
void aaa(T t) {}
};
}
%template(ABCXYZ) Space::ABC&lt;Space::XYZ&gt;;
</pre>
</div>
<p> It should now be apparent that there are many ways to achieve a
renaming with %rename. This is demonstrated by the following two
examples, which are effectively the same as the above example. Below
shows how %rename can be placed inside a namespace.</p>
<div class="code">
<pre>
namespace Space {
%rename(bbb) ABC::aaa(T t); // will match but with lower precedence than ccc
%rename(ccc) ABC&lt;Space::XYZ&gt;::aaa(Space::XYZ t);// will match but with higher precedence than bbb
%rename(ddd) ABC&lt;Space::XYZ&gt;::aaa(XYZ t); // will not match
}
namespace Space {
class XYZ {};
template&lt;typename T&gt; struct ABC {
void aaa(T t) {}
};
}
%template(ABCXYZ) Space::ABC&lt;Space::XYZ&gt;;
</pre>
</div>
<p> Note that <tt>ddd</tt> does not match as there is no namespace
resolution for parameter types and the fully qualified type must be
specified for template type expansion. The following example shows how
%rename can be placed within %extend.</p>
<div class="code">
<pre>
namespace Space {
%extend ABC {
%rename(bbb) aaa(T t); // will match but with lower precedence than ccc
}
%extend ABC&lt;Space::XYZ&gt; {
%rename(ccc) aaa(Space::XYZ t);// will match but with higher precedence than bbb
%rename(ddd) aaa(XYZ t); // will not match
}
}
namespace Space {
class XYZ {};
template&lt;typename T&gt; struct ABC {
void aaa(T t) {}
};
}
%template(ABCXYZ) Space::ABC&lt;Space::XYZ&gt;;
</pre>
</div>
<h2><a name="SWIGPlus_exception_specifications">6.21 Exception
specifications</a></h2>
<p> When C++ programs utilize exceptions, exceptional behavior is
sometimes specified as part of a function or method declaration. For
example:</p>
<div class="code">
<pre>
class Error { };
class Foo {
public:
...
void blah() throw(Error);
...
};
</pre>
</div>
<p> If an exception specification is used, SWIG automatically generates
wrapper code for catching the indicated exception and, when possible,
rethrowing it into the target language, or converting it into an error
in the target language otherwise. For example, in Python, you can write
code like this:</p>
<div class="targetlang">
<pre>
f = Foo()
try:
f.blah()
except Error, e:
# e is a wrapped instance of &quot;Error&quot;
</pre>
</div>
<p> Details of how to tailor code for handling the caught C++ exception
and converting it into the target language's exception/error handling
mechanism is outlined in the <a href="#Typemaps_throws_typemap">
&quot;throws&quot; typemap</a> section.</p>
<p> Since exception specifications are sometimes only used sparingly,
this alone may not be enough to properly handle C++ exceptions. To do
that, a different set of special SWIG directives are used. Consult the
&quot;<a href="#Customization_exception">Exception handling with %exception</a>
&quot; section for details. The next section details a way of simulating an
exception specification or replacing an existing one.</p>
<h2><a name="SWIGPlus_catches">6.22 Exception handling with %catches</a></h2>
<p> Exceptions are automatically handled for methods with an exception
specification. Similar handling can be achieved for methods without
exception specifications through the <tt>%catches</tt> feature. It is
also possible to replace any declared exception specification using the
<tt>%catches</tt> feature. In fact, <tt>%catches</tt> uses the same <a href="#Typemaps_throws_typemap">
&quot;throws&quot; typemaps</a> that SWIG uses for exception specifications in
handling exceptions. The <tt>%catches</tt> feature must contain a list
of possible types that can be thrown. For each type that is in the
list, SWIG will generate a catch handler, in the same way that it would
for types declared in the exception specification. Note that the list
can also include the catch all specification &quot;...&quot;. For example,</p>
<div class="code">
<pre>
struct EBase { virtual ~EBase(); };
struct Error1 : EBase { };
struct Error2 : EBase { };
struct Error3 : EBase { };
struct Error4 : EBase { };
%catches(Error1, Error2, ...) Foo::bar();
%catches(EBase) Foo::blah();
class Foo {
public:
...
void bar();
void blah() throw(Error1, Error2, Error3, Error4);
...
};
</pre>
</div>
<p> For the <tt>Foo::bar()</tt> method, which can throw anything, SWIG
will generate catch handlers for <tt>Error1</tt>, <tt>Error2</tt> as
well as a catch all handler (...). Each catch handler will convert the
caught exception and convert it into a target language error/exception.
The catch all handler will convert the caught exception into an unknown
error/exception.</p>
<p> Without the <tt>%catches</tt> feature being attached to <tt>
Foo::blah()</tt>, SWIG will generate catch handlers for all of the types
in the exception specification, that is, <tt>Error1, Error2, Error3,
Error4</tt>. However, with the <tt>%catches</tt> feature above, just a
single catch handler for the base class, <tt>EBase</tt> will be
generated to convert the C++ exception into a target language
error/exception.</p>
<h2><a name="SWIGPlus_nn33">6.23 Pointers to Members</a></h2>
<p> Starting with SWIG-1.3.7, there is limited parsing support for
pointers to C++ class members. For example:</p>
<div class="code">
<pre>
double do_op(Object *o, double (Object::*callback)(double, double));
extern double (Object::*fooptr)(double, double);
%constant double (Object::*FOO)(double, double) = &amp;Object::foo;
</pre>
</div>
<p> Although these kinds of pointers can be parsed and represented by
the SWIG type system, few language modules know how to handle them due
to implementation differences from standard C pointers. Readers are<em>
strongly</em> advised to consult an advanced text such as the &quot;The
Annotated C++ Manual&quot; for specific details.</p>
<p> When pointers to members are supported, the pointer value might
appear as a special string like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; print(example.FOO)
_ff0d54a800000000_m_Object__f_double_double__double
&gt;&gt;&gt;
</pre>
</div>
<p> In this case, the hexadecimal digits represent the entire value of
the pointer which is usually the contents of a small C++ structure on
most machines.</p>
<p> SWIG's type-checking mechanism is also more limited when working
with member pointers. Normally SWIG tries to keep track of inheritance
when checking types. However, no such support is currently provided for
member pointers.</p>
<h2><a name="SWIGPlus_smart_pointers">6.24 Smart pointers and
operator-&gt;()</a></h2>
<p> In some C++ programs, objects are often encapsulated by
smart-pointers or proxy classes. This is sometimes done to implement
automatic memory management (reference counting) or persistence.
Typically a smart-pointer is defined by a class template where the <tt>
-&gt;</tt> operator has been overloaded. This class is then wrapped around
some other class. For example:</p>
<div class="code">
<pre>
// Smart-pointer class
template&lt;class T&gt; class SmartPtr {
T *pointee;
public:
SmartPtr(T *p) : pointee(p) { ... }
T *operator-&gt;() {
return pointee;
}
...
};
// Ordinary class
class Foo_Impl {
public:
int x;
virtual void bar();
...
};
// Smart-pointer wrapper
typedef SmartPtr&lt;Foo_Impl&gt; Foo;
// Create smart pointer Foo
Foo make_Foo() {
return SmartPtr&lt;Foo_Impl&gt;(new Foo_Impl());
}
// Do something with smart pointer Foo
void do_something(Foo f) {
printf(&quot;x = %d\n&quot;, f-&gt;x);
f-&gt;bar();
}
// Call the wrapped smart pointer proxy class in the target language 'Foo'
%template(Foo) SmartPtr&lt;Foo_Impl&gt;;
</pre>
</div>
<p> A key feature of this approach is that by defining <tt>operator-&gt;</tt>
the methods and attributes of the object wrapped by a smart pointer are
transparently accessible. For example, expressions such as these (from
the previous example),</p>
<div class="code">
<pre>
f-&gt;x
f-&gt;bar()
</pre>
</div>
<p> are transparently mapped to the following</p>
<div class="code">
<pre>
(f.operator-&gt;())-&gt;x;
(f.operator-&gt;())-&gt;bar();
</pre>
</div>
<p> When generating wrappers, SWIG tries to emulate this functionality
to the extent that it is possible. To do this, whenever <tt>
operator-&gt;()</tt> is encountered in a class, SWIG looks at its returned
type and uses it to generate wrappers for accessing attributes of the
underlying object. For example, wrapping the above code produces
wrappers like this:</p>
<div class="code">
<pre>
int Foo_x_get(Foo *f) {
return (*f)-&gt;x;
}
void Foo_x_set(Foo *f, int value) {
(*f)-&gt;x = value;
}
void Foo_bar(Foo *f) {
(*f)-&gt;bar();
}
</pre>
</div>
<p> These wrappers take a smart-pointer instance as an argument, but
dereference it in a way to gain access to the object returned by <tt>
operator-&gt;()</tt>. You should carefully compare these wrappers to those
in the first part of this chapter (they are slightly different).</p>
<p> The end result is that access looks very similar to C++. For
example, you could do this in Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; f = make_Foo()
&gt;&gt;&gt; print(f.x)
0
&gt;&gt;&gt; f.bar()
&gt;&gt;&gt;
</pre>
</div>
<p> When generating wrappers through a smart-pointer, SWIG tries to
generate wrappers for all methods and attributes that might be
accessible through <tt>operator-&gt;()</tt>. This includes any methods
that might be accessible through inheritance. However, there are a
number of restrictions:</p>
<ul>
<li>Member variables and methods are wrapped through a smart pointer.
Enumerations, constructors, and destructors are not wrapped.</li>
<li>
<p>If the smart-pointer class and the underlying object both define a
method or variable of the same name, then the smart-pointer version has
precedence. For example, if you have this code</p>
<div class="code">
<pre>
class Foo {
public:
int x;
};
class Bar {
public:
int x;
Foo *operator-&gt;();
};
</pre>
</div>
<p> then the wrapper for <tt>Bar::x</tt> accesses the <tt>x</tt> defined
in <tt>Bar</tt>, and not the <tt>x</tt> defined in <tt>Foo</tt>.</p>
</li>
</ul>
<p> If your intent is to only expose the smart-pointer class in the
interface, it is not necessary to wrap both the smart-pointer class and
the class for the underlying object. However, you must still tell SWIG
about both classes if you want the technique described in this section
to work. To only generate wrappers for the smart-pointer class, you can
use the %ignore directive. For example:</p>
<div class="code">
<pre>
%ignore Foo;
class Foo { // Ignored
};
class Bar {
public:
Foo *operator-&gt;();
...
};
</pre>
</div>
<p> Alternatively, you can import the definition of <tt>Foo</tt> from a
separate file using <tt>%import</tt>.</p>
<p><b> Note:</b> When a class defines <tt>operator-&gt;()</tt>, the
operator itself is wrapped as a method <tt>__deref__()</tt>. For
example:</p>
<div class="targetlang">
<pre>
f = Foo() # Smart-pointer
p = f.__deref__() # Raw pointer from operator-&gt;
</pre>
</div>
<p><b> Note:</b> To disable the smart-pointer behavior, use <tt>%ignore</tt>
to ignore <tt>operator-&gt;()</tt>. For example:</p>
<div class="code">
<pre>
%ignore Bar::operator-&gt;;
</pre>
</div>
<p><b> Note:</b> Smart pointer support was first added in SWIG-1.3.14.</p>
<h2><a name="SWIGPlus_ref_unref">6.25 C++ reference counted objects -
ref/unref feature</a></h2>
<p> Another similar idiom in C++ is the use of reference counted
objects. Consider for example:</p>
<div class="code">
<pre>
class RCObj {
// implement the ref counting mechanism
int add_ref();
int del_ref();
int ref_count();
public:
virtual ~RCObj() = 0;
int ref() const {
return add_ref();
}
int unref() const {
if (ref_count() == 0 || del_ref() == 0 ) {
delete this;
return 0;
}
return ref_count();
}
};
class A : RCObj {
public:
A();
int foo();
};
class B {
A *_a;
public:
B(A *a) : _a(a) {
a-&gt;ref();
}
~B() {
a-&gt;unref();
}
};
int main() {
A *a = new A(); // (count: 0)
a-&gt;ref(); // 'a' ref here (count: 1)
B *b1 = new B(a); // 'a' ref here (count: 2)
if (1 + 1 == 2) {
B *b2 = new B(a); // 'a' ref here (count: 3)
delete b2; // 'a' unref, but not deleted (count: 2)
}
delete b1; // 'a' unref, but not deleted (count: 1)
a-&gt;unref(); // 'a' unref and deleted (count: 0)
}
</pre>
</div>
<p> In the example above, the 'A' class instance 'a' is a reference
counted object, which can't be deleted arbitrarily since it is shared
between the objects 'b1' and 'b2'. 'A' is derived from a<i> Reference
Counted Object</i> 'RCObj', which implements the ref/unref idiom.</p>
<p> To tell SWIG that 'RCObj' and all its derived classes are reference
counted objects, use the &quot;ref&quot; and &quot;unref&quot; <a href="#Customization_features">
features</a>. These are also available as <tt>%refobject</tt> and <tt>
%unrefobject</tt>, respectively. For example:</p>
<div class="code">
<pre>
%module example
...
%feature(&quot;ref&quot;) RCObj &quot;$this-&gt;ref();&quot;
%feature(&quot;unref&quot;) RCObj &quot;$this-&gt;unref();&quot;
%include &quot;rcobj.h&quot;
%include &quot;A.h&quot;
...
</pre>
</div>
<p> where the code passed to the &quot;ref&quot; and &quot;unref&quot; features will be
executed as needed whenever a new object is passed to Python, or when
Python tries to release the proxy object instance, respectively.</p>
<p> On the Python side, the use of a reference counted object is no
different to any other regular instance:</p>
<div class="targetlang">
<pre>
def create_A():
a = A() # SWIG ref 'a' - new object is passed to Python (count: 1)
b1 = B(a) # C++ ref 'a (count: 2)
if 1 + 1 == 2:
b2 = B(a) # C++ ref 'a' (count: 3)
return a # 'b1' and 'b2' are released and deleted, C++ unref 'a' twice (count: 1)
a = create_A() # (count: 1)
exit # 'a' is released, SWIG unref 'a' called in the destructor wrapper (count: 0)
</pre>
</div>
<p> Note that the user doesn't explicitly need to call 'a-&gt;ref()' nor
'a-&gt;unref()' (and neither 'delete a'). Instead, SWIG takes cares of
executing the &quot;ref&quot; and &quot;unref&quot; calls as needed. If the user doesn't
specify the &quot;ref/unref&quot; feature for a type, SWIG will produce code
equivalent to defining these features:</p>
<div class="code">
<pre>
%feature(&quot;ref&quot;) &quot;&quot;
%feature(&quot;unref&quot;) &quot;delete $this;&quot;
</pre>
</div>
<p> In other words, SWIG will not do anything special when a new object
is passed to Python, and it will always 'delete' the underlying object
when Python releases the proxy instance.</p>
<p> The <a href="#Customization_ownership">%newobject feature</a> is
designed to indicate to the target language that it should take
ownership of the returned object. When used in conjunction with a type
that has the &quot;ref&quot; feature associated with it, it additionally emits
the code in the &quot;ref&quot; feature into the C++ wrapper. Consider wrapping
the following factory function in addition to the above:</p>
<div class="code">
<pre>
%newobject AFactory;
A *AFactory() {
return new A();
}
</pre>
</div>
<p> The <tt>AFactory</tt> function now acts much like a call to the <tt>
A</tt> constructor with respect to memory handling:</p>
<div class="targetlang">
<pre>
a = AFactory() # SWIG ref 'a' due to %newobject (count: 1)
exit # 'a' is released, SWIG unref 'a' called in the destructor wrapper (count: 0)
</pre>
</div>
<h2><a name="SWIGPlus_nn35">6.26 Using declarations and inheritance</a></h2>
<p> C++ <tt>using</tt> declarations are sometimes used to introduce
members of base classes. For example:</p>
<div class="code">
<pre>
class Foo {
public:
int blah(int x);
};
class Bar {
public:
double blah(double x);
};
class FooBar : public Foo, public Bar {
public:
using Foo::blah;
using Bar::blah;
char *blah(const char *x);
};
</pre>
</div>
<p> In this example, the <tt>using</tt> declarations make different
versions of the overloaded <tt>blah()</tt> method accessible from the
derived class. For example:</p>
<div class="code">
<pre>
FooBar *f;
f-&gt;blah(3); // Ok. Invokes Foo::blah(int)
f-&gt;blah(3.5); // Ok. Invokes Bar::blah(double)
f-&gt;blah(&quot;hello&quot;); // Ok. Invokes FooBar::blah(const char *);
</pre>
</div>
<p> SWIG emulates the same functionality when creating wrappers. For
example, if you wrap this code in Python, the module works just like
you would expect:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
&gt;&gt;&gt; f = example.FooBar()
&gt;&gt;&gt; f.blah(3)
&gt;&gt;&gt; f.blah(3.5)
&gt;&gt;&gt; f.blah(&quot;hello&quot;)
</pre>
</div>
<p> The C++11 standard supports using declarations for inheriting
constructors and this is covered in <a href="CPlusPlus11_object_construction_improvement">
Object construction improvement</a>.</p>
<p> C++ <tt>using</tt> declarations can also be used to change access
when applicable. For example, protected methods in a base class can be
made public in a derived class:</p>
<div class="code">
<pre>
class Foo {
protected:
int x;
int blah(int x);
};
class Bar : public Foo {
public:
using Foo::x; // Make x public
using Foo::blah; // Make blah public
};
</pre>
</div>
<p> This also works in SWIG---the exposed declarations will be wrapped
normally.</p>
<p> When <tt>using</tt> declarations are used as shown in these
examples, declarations from the base classes are copied into the
derived class and wrapped normally. When copied, the declarations
retain any properties that might have been attached using <tt>%rename</tt>
, <tt>%ignore</tt>, or <tt>%feature</tt>. Thus, if a method is ignored
in a base class, it will also be ignored by a <tt>using</tt>
declaration.</p>
<p> Because a <tt>using</tt> declaration does not provide fine-grained
control over the declarations that get imported, because a single <tt>
using</tt> declaration may introduce multiple methods, it may be
difficult to manage such declarations in applications that make heavy
use of SWIG customization features. If you can't get <tt>using</tt> to
work correctly, you can always modify the C++ code to handle SWIG
differently such as:</p>
<div class="code">
<pre>
class FooBar : public Foo, public Bar {
public:
#ifndef SWIG
using Foo::blah;
using Bar::blah;
#else
int blah(int x); // explicitly tell SWIG about other declarations
double blah(double x);
#endif
char *blah(const char *x);
};
</pre>
</div>
<p> If the C++ code being wrapped cannot be changed, make judicious
usage of <tt>%extend</tt> and <tt>%rename</tt> to ignore and unignore
declarations. The example below is effectively the same as above:</p>
<div class="code">
<pre>
%extend FooBar {
int blah(int x) { return $self-&gt;Foo::blah(x); }
double blah(double x) { return $self-&gt;Bar::blah(x); }
}
%ignore FooBar::blah; // ignore all FooBar::blah below
%rename(&quot;&quot;) FooBar::blah(const char *x); // parameterized unignore
class FooBar : public Foo, public Bar {
public:
using Foo::blah;
using Bar::blah;
char *blah(const char *x);
};
</pre>
</div>
<p><b> Notes:</b></p>
<ul>
<li>
<p>If a derived class introduces a method defined in a base class via a <tt>
using</tt> declaration, there won't be a conflict due to incorrect
additional methods. For example:</p>
<div class="code">
<pre>
class Foo {
public:
int blah(int );
double blah(double);
};
class Bar : public Foo {
public:
using Foo::blah; // Only introduces blah(double);
int blah(int);
};
</pre>
</div></li>
<li>
<p>Renaming methods may prevent methods from being introduced into the
derived class via <tt>using</tt> declarations. For example:</p>
<div class="code">
<pre>
%rename(blah_long) Foo::blah(long);
class Foo {
public:
int blah(int);
long blah(long); // Renamed to blah_long
};
class Bar : public Foo {
public:
using Foo::blah; // Only introduces blah(int)
double blah(double x);
};
</pre>
</div>
<p> The problem here is <tt>Foo::blah</tt> is renamed to <tt>blah_long</tt>
in the target language, but the <tt>using</tt> declaration in Bar is
not renamed in the target language and thinks all introduced methods
should simply be called <tt>blah</tt>. It is not clear what target
language names should be used in Bar and so the conflicting names are
effectively ignored as they are not introduced into the derived class
for the target language wrappers. In such situations SWIG will emit a
warning:</p>
<div class="shell">
<pre>
example.i:15: Warning 526: Using declaration Foo::blah, with name 'blah', is not actually using
example.i:10: Warning 526: the method from Foo::blah(long), with name 'blah_long', as the names are different.
</pre>
</div>
<p><b> Compatibility Note:</b> This warning message was introduced in
SWIG-4.1.0. Prior versions also effectively ignored the using
declaration for the same reasons, but were silent about it.</p>
<p> If methods really need different names, please use of combinations
of <tt>%rename</tt>, <tt>%ignore</tt> and <tt>%extend</tt> to achieve
the desired outcome.</p>
</li>
</ul>
<h2><a name="SWIGPlus_nested_classes">6.27 Nested classes</a></h2>
<p> If the target language supports the nested classes concept (like
Java), the nested C++ classes are wrapped as nested target language
proxy classes. (In case of Java - &quot;static&quot; nested classes.) Only public
nested classes are wrapped. Otherwise there is little difference
between nested and normal classes.</p>
<p> If the target language doesn't support nested classes directly, or
the support is not implemented in the language module (like for Python
currently), then the visible nested classes are moved to the same name
space as the containing class (nesting hierarchy is &quot;flattened&quot;). The
same behaviour may be turned on for C# and Java by the %feature
(&quot;flatnested&quot;); If there is a class with the same name in the outer
namespace the inner class (or the global one) may be renamed or
ignored:</p>
<div class="code">
<pre>
%rename (Bar_Foo) Bar::Foo;
class Foo {};
class Bar {
public:
class Foo {};
};
</pre>
</div>
<p> If a nested class, within an outer class, has to be used as a
template parameter within the outer class, then the template will have
to be instantiated with <tt>%template</tt> before the beginning of the
outer class. An example can be found in the <a href="#SWIGPlus_template_nested_class_example">
Templates</a> section.</p>
<p><b> Compatibility Note:</b> Prior to SWIG-3.0.0, there was limited
nested class support. Nested classes were treated as opaque pointers.
However, there was a workaround for nested class support in these older
versions requiring the user to replicate the nested class in the global
scope, adding in a typedef for the nested class in the global scope and
using the &quot;nestedworkaround&quot; feature on the nested class. This resulted
in approximately the same behaviour as the &quot;flatnested&quot; feature. With
proper nested class support now available in SWIG-3.0.0, this feature
was deprecated and has now been removed. If you see the following
error:</p>
<div class="shell">
<pre>
example.i:8: Error: Unknown directive '%nestedworkaround'
</pre>
</div>
<p> consider using the &quot;flatnested&quot; feature discussed above which
generates a non-nested proxy class, like the &quot;nestedworkaround&quot; feature
did. Alternatively, use the default nested class code generation, which
may generate an equivalent to a nested proxy class in the target
language, depending on the target language support.</p>
<p> SWIG-1.3.40 and earlier versions did not have the <tt>
nestedworkaround</tt> feature and the generated code resulting from
parsing nested classes did not always compile. Nested class warnings
could also not be suppressed using %warnfilter.</p>
<h2><a name="SWIGPlus_const">6.28 A brief rant about const-correctness</a>
</h2>
<p> A common issue when working with C++ programs is dealing with all
possible ways in which the <tt>const</tt> qualifier (or lack thereof)
will break your program, all programs linked against your program, and
all programs linked against those programs.</p>
<p> Although SWIG knows how to correctly deal with <tt>const</tt> in its
internal type system and it knows how to generate wrappers that are
free of const-related warnings, SWIG does not make any attempt to
preserve const-correctness in the target language. Thus, it is possible
to pass <tt>const</tt> qualified objects to non-const methods and
functions. For example, consider the following code in C++:</p>
<div class="code">
<pre>
const Object * foo();
void bar(Object *);
...
// C++ code
void blah() {
bar(foo()); // Error: bar discards const
};
</pre>
</div>
<p> Now, consider the behavior when wrapped into a Python module:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; bar(foo()) # Okay
&gt;&gt;&gt;
</pre>
</div>
<p> Although this is clearly a violation of the C++ type-system, fixing
the problem doesn't seem to be worth the added implementation
complexity that would be required to support it in the SWIG run-time
type system. There are no plans to change this in future releases
(although we'll never rule anything out entirely).</p>
<p> The bottom line is that this particular issue does not appear to be
a problem for most SWIG projects. Of course, you might want to consider
using another tool if maintaining constness is the most important part
of your project.</p>
<h2><a name="SWIGPlus_target_language_callbacks">6.29 Callbacks to the
target language</a></h2>
<p> C/C++ function pointers are often used for callbacks and this is
discussed in the <a href="#SWIG_nn30">Pointers to functions and
callbacks</a> section. The callback techniques described therein
provide a way to control callbacks to a C/C++ function but not
callbacks into the target language. The techniques described below show
how the director feature can be used to support callbacks from C/C++ to
the target language.</p>
<h3><a name="SWIGPlus_director_classes_introduction">6.29.1 Introduction
to director classes</a></h3>
<p> The director feature enables the ability for a target language class
to derive from a wrapped C++ class. The target language can override
virtual methods of a wrapped C++ class, thereby supporting
cross-language polymorphism. Code can 'call up' from C++ into the
target language by simply calling a virtual method overridden in a
derived class in the target language. The wrapped C++ classes that have
this ability are termed 'director' classes. The director feature is
documented individually in each target language and the reader should
locate and read this to obtain a full understanding of directors.</p>
<h3><a name="SWIGPlus_directors_for_function_pointers">6.29.2 Using
directors and target language callbacks</a></h3>
<p> SWIG's primary goal is to make it possible to call C/C++ code from a
target language, however, the director feature enables the reverse.
While there isn't simple direct support for calling target language
code from C, the director feature makes this possible. It does require
some work and additional wrapper code to be provided by the user. The
additional code required must be C++ and not C code and hence may
introduce a small dependency on C++ if using a pure C project. In a
nutshell, the user must create a C++ base class and turn it into a
director class. A virtual method in the director base class is
required. SWIG generates the code to call up into the target language
when wrapping the director virtual method.</p>
<p> Let's look at some details next. Consider the same function pointer
for a callback called <tt>binary_op</tt> from the <a href="#SWIG_nn30">
Pointers to functions and callbacks</a> section. For completeness, the
code required for the module and director feature is also shown:</p>
<div class="code">
<pre>
%module(directors=&quot;1&quot;) example
%{
int binary_op(int a, int b, int (*op)(int, int)) {
return op(a, b);
}
%}
</pre>
</div>
<p> The goal is to have a target language function that gets called by <tt>
binary_op</tt>. The target language function should have the equivalent
signature as the C/C++ function pointer <tt>int (*op)(int, int)</tt>.
As we are using directors, we need a C++ virtual method with this
signature, so let's define the C++ class and pure virtual method first
and make it a director class via the director feature:</p>
<div class="code">
<pre>
%feature(&quot;director&quot;) BinaryOp;
%inline %{
struct BinaryOp {
virtual int handle(int a, int b) = 0;
virtual ~BinaryOp() {}
};
%}
</pre>
</div>
<p> The following <tt>handler_helper</tt> function and <tt>
binary_op_wrapper</tt> function completes the code needed in the
C++/SWIG layer. The <tt>binary_op_wrapper</tt> function is wrapped by
SWIG and is very similar to the <tt>binary_op</tt> function, however,
it takes a pointer to the director base class <tt>BinaryOp</tt> instead
of a C/C++ function pointer.</p>
<div class="code">
<pre>
%{
static BinaryOp *handler_ptr = NULL;
static int handler_helper(int a, int b) {
// Make the call up to the target language when handler_ptr
// is an instance of a target language director class
return handler_ptr-&gt;handle(a, b);
}
// If desired, handler_ptr above could be changed to a thread-local variable in order to make thread-safe
%}
%inline %{
int binary_op_wrapper(int a, int b, BinaryOp *handler) {
handler_ptr = handler;
int result = binary_op(a, b, &amp;handler_helper);
handler = NULL;
return result;
}
%}
</pre>
</div>
<p> On the target language side, we need to derive a class from <tt>
BinaryOp</tt> and override the <tt>handle</tt> method. In Python this
could be as simple as:</p>
<div class="targetlang">
<pre>
import example
# PythonBinaryOp class is defined and derived from C++ class BinaryOp
class PythonBinaryOp(example.BinaryOp):
# Define Python class 'constructor'
def __init__(self):
# Call C++ base class constructor
example.BinaryOp.__init__(self)
# Override C++ method: virtual int handle(int a, int b) = 0;
def handle(self, a, b):
# Return the product
return a * b
</pre>
</div>
<p> For this to work from Python, an instance of the <tt>PythonBinaryOp</tt>
class is created and then passed to <tt>binary_op_wrapper</tt>. The net
result is the <tt>binary_op</tt> function will in turn be called which
will call <tt>handler_helper</tt> which will call the virtual <tt>
handle</tt> method, that is, the Python method <tt>handle</tt> in the
PythonBinaryOp class. The result will be the product of 10 and 20 and
make its way back to Python and hence 200 will be printed with the
following code:</p>
<div class="targetlang">
<pre>
handler = PythonBinaryOp()
result = example.binary_op_wrapper(10, 20, handler)
print(result)
</pre>
</div>
<p> This has thus demonstrated a C/C++ function pointer calling back
into a target language function. The code could be made a little more
user friendly by using <tt>%rename</tt> to provide the original <tt>
binary_op</tt> name from the target language instead of <tt>
binary_op_wrapper</tt>. A C++ functor base class and Python functor
class could also be used instead, but these are left as exercises for
the reader.</p>
<h2><a name="SWIGPlus_nn42">6.30 Where to go for more information</a></h2>
<p> If you're wrapping serious C++ code, you might want to pick up a
copy of &quot;The Annotated C++ Reference Manual&quot; by Ellis and Stroustrup.
This is the reference document we use to guide a lot of SWIG's C++
support.</p>
<!-- LocalWords: destructors Enums Namespaces const SWIG's STL OO adaptor tcl
-->
<!-- LocalWords: debuggable cxx OBJS Wiki accessor nodefault makedefault
-->
<!-- LocalWords: notabstract CopyFoo
-->
<HR NOSHADE>
<h1><a name="CPlusPlus11">7 SWIG and C++11</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#CPlusPlus11_introduction">Introduction</a></li>
<li><a href="#CPlusPlus11_core_language_changes">Core language changes</a>
<ul>
<li><a href="#CPlusPlus11_rvalue_reference_and_move_semantics">Rvalue
reference and move semantics</a>
<ul>
<li><a href="#CPlusPlus11_rvalue_reference_inputs">Rvalue reference
inputs</a></li>
<li><a href="#CPlusPlus11_rvalue_reference_outputs">Rvalue reference
outputs</a></li>
<li><a href="#CPlusPlus11_move_only">Movable and move-only types by
value</a></li>
</ul>
</li>
<li><a href="#CPlusPlus11_generalized_constant_expressions">Generalized
constant expressions</a></li>
<li><a href="#CPlusPlus11_extern_template">Extern template</a></li>
<li><a href="#CPlusPlus11_initializer_lists">Initializer lists</a></li>
<li><a href="#CPlusPlus11_uniform_initialization">Uniform initialization</a>
</li>
<li><a href="#CPlusPlus11_type_inference">Type inference</a></li>
<li><a href="#CPlusPlus11_range_based_for_loop">Range-based for-loop</a></li>
<li><a href="#CPlusPlus11_lambda_functions_and_expressions">Lambda
functions and expressions</a></li>
<li><a href="#CPlusPlus11_alternate_function_syntax">Alternate function
syntax</a></li>
<li><a href="#CPlusPlus11_object_construction_improvement">Object
construction improvement</a></li>
<li><a href="#CPlusPlus11_explicit_overrides_final">Explicit overrides
and final</a></li>
<li><a href="#CPlusPlus11_null_pointer_constant">Null pointer constant</a>
</li>
<li><a href="#CPlusPlus11_strongly_typed_enumerations">Strongly typed
enumerations</a></li>
<li><a href="#CPlusPlus11_double_angle_brackets">Double angle brackets</a>
</li>
<li><a href="#CPlusPlus11_explicit_conversion_operators">Explicit
conversion operators</a></li>
<li><a href="#CPlusPlus11_alias_templates">Type alias and alias
templates</a></li>
<li><a href="#CPlusPlus11_unrestricted_unions">Unrestricted unions</a></li>
<li><a href="#CPlusPlus11_variadic_templates">Variadic templates</a></li>
<li><a href="#CPlusPlus11_new_char_literals">New character literals</a></li>
<li><a href="#CPlusPlus11_new_string_literals">New string literals</a></li>
<li><a href="#CPlusPlus11_user_defined_literals">User-defined literals</a>
</li>
<li><a href="#CPlusPlus11_thread_local_storage">Thread-local storage</a></li>
<li><a href="#CPlusPlus11_defaulted_deleted">Explicitly defaulted
functions and deleted functions</a></li>
<li><a href="#CPlusPlus11_type_long_long_int">Type long long int</a></li>
<li><a href="#CPlusPlus11_static_assertions">Static assertions</a></li>
<li><a href="#CPlusPlus11_sizeof">Allow sizeof to work on members of
classes without an explicit object</a></li>
<li><a href="#CPlusPlus11_noexcept">Exception specifications and
noexcept</a></li>
<li><a href="#CPlusPlus11_alignment">Control and query object alignment</a>
</li>
<li><a href="#CPlusPlus11_attributes">Attributes</a></li>
<li><a href="#CPlusPlus11_ref_qualifiers">Methods with ref-qualifiers</a>
</li>
</ul>
</li>
<li><a href="#CPlusPlus11_standard_library_changes">Standard library
changes</a>
<ul>
<li><a href="#CPlusPlus11_threading_facilities">Threading facilities</a></li>
<li><a href="#CPlusPlus11_tuple_types">Tuple types</a></li>
<li><a href="#CPlusPlus11_hash_tables">Hash tables</a></li>
<li><a href="#CPlusPlus11_regular_expressions">Regular expressions</a></li>
<li><a href="#CPlusPlus11_general_purpose_smart_pointers">
General-purpose smart pointers</a></li>
<li><a href="#CPlusPlus11_extensible_random_number_facility">Extensible
random number facility</a></li>
<li><a href="#CPlusPlus11_wrapper_reference">Wrapper reference</a></li>
<li><a href="#CPlusPlus11_polymorphous_wrappers_for_function_objects">
Polymorphic wrappers for function objects</a></li>
<li><a href="#CPlusPlus11_type_traits_for_metaprogramming">Type traits
for metaprogramming</a></li>
<li><a href="#CPlusPlus11_uniform_method_for_computing_return_type_of_function_objects">
Uniform method for computing return type of function objects</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="CPlusPlus11_introduction">7.1 Introduction</a></h2>
<p>This chapter gives you a brief overview about the SWIG implementation
of the C++11 standard.</p>
<p>SWIG supports the new C++ syntax changes with some minor limitations
in some areas such as decltype expressions and variadic templates.
Wrappers for the new STL types (unordered_ containers, result_of,
tuples) are incomplete. The wrappers for the new containers would work
much like the C++03 containers and users are welcome to help by
adapting the existing container interface files and submitting them as
a patch for inclusion in future versions of SWIG.</p>
<h2><a name="CPlusPlus11_core_language_changes">7.2 Core language
changes</a></h2>
<h3><a name="CPlusPlus11_rvalue_reference_and_move_semantics">7.2.1
Rvalue reference and move semantics</a></h3>
<p> SWIG correctly parses the rvalue reference syntax '&amp;&amp;', for example
the typical usage of it in the move constructor and move assignment
operator below:</p>
<div class="code">
<pre>
class MyClass {
...
std::vector&lt;int&gt; numbers;
public:
MyClass() : numbers() {}
MyClass(MyClass &amp;&amp;other) : numbers(std::move(other.numbers)) {}
MyClass &amp; operator=(MyClass &amp;&amp;other) {
numbers = std::move(other.numbers);
return *this;
}
};
</pre>
</div>
<p> Rvalue references are designed for C++ temporaries and are not
particularly useful when used from non-C++ target languages. One option
is to just ignore them via <tt>%ignore</tt>. For example, ignore the
move constructor:</p>
<div class="code">
<pre>
%ignore MyClass::MyClass(MyClass &amp;&amp;);
</pre>
</div>
<h4><a name="CPlusPlus11_rvalue_reference_inputs">7.2.1.1 Rvalue
reference inputs</a></h4>
<p> Rvalue reference parameters are useful as input parameters in C++
for implementing move semantics, such as, in the move constructor and
move assignment operator. This type of usage can be useful from target
languages too to avoid copying large objects.</p>
<p> If you do wrap a function/constructor with an rvalue reference
parameter and pass a proxy class to it, SWIG will assume that after the
call, the rvalue reference parameter object will have been 'moved'. The
proxy class passed as the rvalue reference, will own the underlying C++
object up until it is used as an rvalue reference parameter.
Afterwards, the proxy class will have the underlying C++ pointer set to
the nullptr so that the proxy class instance cannot be used again and
the underlying (moved from) C++ object will be deleted after the
function/constructor call has returned.</p>
<p> In this way, the SWIG proxy class works much like an exclusively
owned smart pointer (think of <tt>std::unique_ptr</tt>), passing
ownership to the called C++ function/constructor. Let's consider an
example in Java using the wrapped proxy class from above:</p>
<div class="targetlang">
<pre>
MyClass mc = new MyClass();
MyClass mc1 = new MyClass(mc); // move constructor
MyClass mc2 = new MyClass(mc); // move constructor fails
</pre>
</div>
<p> The second call to the move constructor will fail as the <tt>mc</tt>
proxy instance has been moved. Each target language handles the moved
proxy class slightly differently when attempting to move it again, but
typically you'll get an exception such as in Java:</p>
<div class="shell">
<pre>
Exception in thread &quot;main&quot; java.lang.RuntimeException: Cannot release ownership as memory is not owned
at MyClass.swigRelease(MyClass.java:27)
at MyClass.&lt;init&gt;(MyClass.java:55)
at runme.main(runme.java:18)
</pre>
</div>
<p> Note that both normal copy assignment operators as well as move
assignment operators are ignored by default in the target languages
with the following warning:</p>
<div class="shell">
<pre>
example.i:18: Warning 503: Can't wrap 'operator =' unless renamed to a valid identifier.
</pre>
</div>
<p> Using a <tt>%rename</tt> will remove the warning and also makes the
move assignment operator available from the target language:</p>
<div class="code">
<pre>
%rename(MoveAssign) MyClass::operator=(MyClass &amp;&amp;);
</pre>
</div>
<p> You can then use it, but like the move constructor example above,
you cannot use a proxy class once it has already been moved:</p>
<div class="targetlang">
<pre>
MyClass mc = new MyClass();
MyClass mc2 = mc.MoveAssign(mc);
MyClass mc3 = mc.MoveAssign(mc); // Use of mc again will fail
</pre>
</div>
<p> It is of course perfectly possible in C++ for a function/constructor
to not move an object passed to it in an rvalue reference parameter.
The assumption that SWIG makes would then not hold and customisation of
the appropriate input typemaps would be required. For scripting
languages, this would be for the 'in' typemap and for the non-scripting
languages additional typemaps such as the 'javain' typemap, which is
used to set the memory ownership of the underlying C++ object for Java,
would also need copying and modifying appropriately.</p>
<p><b> Compatibility note:</b> SWIG-4.1.0 changed the way that rvalue
reference parameters were handled and implemented typemaps assuming
that the proxy class owns the underlying C++ object and transfers
ownership of the object when a function/constructor with an rvalue
reference parameter is called.</p>
<h4><a name="CPlusPlus11_rvalue_reference_outputs">7.2.1.2 Rvalue
reference outputs</a></h4>
<p> While rvalue reference parameter inputs are not uncommon in C++ and
can be usefully utilised from target languages, this cannot be said for
rvalue reference outputs. Firstly, it is quite unusual in C++ to have
functions that return an rvalue reference. Secondly, these cases are
nigh on impossible to use from a target language. The main problem is
these references are for C++ compiler temporaries used on the stack and
the target languages use objects on the heap and the concept of
compiler temporary objects doesn't make sense from another language.</p>
<p> Using <tt>MyClass</tt> from earlier and this C++ code:</p>
<div class="code">
<pre>
void use(MyClass &amp;&amp;mc);
MyClass &amp;&amp; get1();
MyClass &amp; get2();
</pre>
</div>
<p> SWIG wraps the <tt>get1</tt> and <tt>get2</tt> functions more or
less identically. The returned references are converted into pointers
that are not owned by the target language. It means that the following
perfectly valid C++ has no equivalent in any of the target languages:</p>
<div class="code">
<pre>
use(get1());
use(std::move(get2()));
</pre>
</div>
<p> An attempt to call the equivalent <tt>use(get1())</tt> from one of
the target languages will result in the ownership failure mentioned in
the previous section as the object being passed to the <tt>use</tt>
function is not owned by the proxy class. In order to own the object,
it would need to be cloned for the object to move from the stack to the
heap, for which an appropriate clone function would be required, but
may not even be available. Note that a move constructor or copy
constructor may slice the object when inheritance is involved.
Alternatively, customising the input rvalue reference typemap, as
mentioned in the previous section, could remove the ownership
requirement. Another alternative would be to modify the output rvalue
reference typemap to always clone the rvalue reference object.
Fortunately you're highly unlikely to have to solve any of these
issues!</p>
<h4><a name="CPlusPlus11_move_only">7.2.1.3 Movable and move-only types
by value</a></h4>
<p> SWIG has traditionally relied on wrapped C++ types to be copy
constructible or copy assignable, either via an explicit or implicit
copy constructor and copy assignment operator. Prior to C++11, a
function could not return nor take a type by value that was not
copyable. In C++11 this is no longer the case. A type can also be
movable if it has has a move constructor and a move assignment
operator. A move-only type is movable but not copyable; it has both the
copy constructor and copy assignment operator deleted. Movable types
can appear in function signatures for passing 'by value' and in C++11
the object can then be moved rather than copied.</p>
<p> SWIG has support for both copyable and/or movable types. Support for
move semantics is quite seamless when returning by value from a
function. Support for move semantics is less so and may require some
customisation when passing by value to a function. First let's consider
returning by value from a function.</p>
<p> The support for function return values is generically implemented in
the &quot;out&quot; <tt>SWIGTYPE</tt> typemap which supports any type, including
copyable, movable and move-only types. The typemap code is very simple
and written so that the compiler will call the move constructor if
possible, otherwise the copy constructor:</p>
<div class="code">
<pre>
%typemap(out) SWIGTYPE %{
$result = new $1_ltype($1);
%}
</pre>
</div>
<p> The above typemap is for C# and when used to wrap a move-only type
such as:</p>
<div class="code">
<pre>
struct MoveOnly {
int val;
MoveOnly(): val(0) {}
MoveOnly(const MoveOnly &amp;) = delete;
MoveOnly(MoveOnly &amp;&amp;) = default;
MoveOnly &amp; operator=(const MoveOnly &amp;) = delete;
MoveOnly &amp; operator=(MoveOnly &amp;&amp;) = default;
static MoveOnly create() { return MoveOnly(); }
static void take(MoveOnly mo);
};
</pre>
</div>
<p> will generate wrapper code for the <tt>create</tt> factory method:</p>
<div class="code">
<pre>
SWIGEXPORT void * SWIGSTDCALL CSharp_MoveOnly_create() {
void * jresult ;
SwigValueWrapper&lt; MoveOnly &gt; result;
result = MoveOnly::create();
jresult = new MoveOnly(result);
return jresult;
}
</pre>
</div>
<p> <tt>SwigValueWrapper</tt> is covered in <a href="#SWIGPlus_nn19">
Pass and return by value</a> and is automatically generated in this case
as <tt>MoveOnly</tt> is not assignable. Note that the generated code
could be optimised further using the <a href="#Typemaps_optimal">
&quot;optimal&quot; attribute</a> in the &quot;out&quot; typemap, so if the above typemap is
customised as follows (note that this is C# specific):</p>
<div class="code">
<pre>
%typemap(out, optimal=&quot;1&quot;) MoveOnly %{
$result = new $1_ltype($1);
%}
</pre>
</div>
<p> then the generated code will result in the object being optimally
moved:</p>
<div class="code">
<pre>
SWIGEXPORT void * SWIGSTDCALL CSharp_MoveOnly_create() {
void * jresult ;
jresult = new MoveOnly(MoveOnly::create());
return jresult;
}
</pre>
</div>
<p> Now let's consider passing by value. We'll consider three cases;
namely types that are:</p>
<ol>
<li> Copyable and not movable - <tt>CopyOnly</tt>.</li>
<li> Copyable and movable - <tt>MovableCopyable</tt>.</li>
<li> Movable and not copyable - <tt>MoveOnly</tt>.</li>
</ol>
<p> and for clarification, define these two additional types as follows:</p>
<div class="code">
<pre>
struct CopyOnly {
int val;
CopyOnly(): val(0) {}
CopyOnly(const CopyOnly &amp;) = default;
CopyOnly &amp; operator=(const CopyOnly &amp;) = default;
static CopyOnly create() { return CopyOnly(); }
static void take(CopyOnly co);
};
struct MovableCopyable {
int val;
MovableCopyable(): val(0) {}
MovableCopyable(const MovableCopyable &amp;) = default;
MovableCopyable(MovableCopyable &amp;&amp;) = default;
MovableCopyable &amp; operator=(const MovableCopyable &amp;) = default;
MovableCopyable &amp; operator=(MovableCopyable &amp;&amp;) = default;
static MovableCopyable create() { return MovableCopyable(); }
static void take(MovableCopyable mc);
};
</pre>
</div>
<p> The generated code is shown below for <tt>CopyOnly::take</tt> (with
additional comments for when constructors and assignment operators are
called). While the code shown is C# specific, the generated constructor
and/or assignment operator calls are ultimately the same for all target
languages.</p>
<div class="code">
<pre>
SWIGEXPORT void SWIGSTDCALL CSharp_CopyOnly_take(void * jarg1) {
CopyOnly arg1 ; // (a) Default constructor
CopyOnly *argp1 ;
argp1 = (CopyOnly *)jarg1;
if (!argp1) {
SWIG_CSharpSetPendingExceptionArgument(SWIG_CSharpArgumentNullException, &quot;Attempt to dereference null CopyOnly&quot;, 0);
return ;
}
arg1 = *argp1; // (b) Copy assignment
CopyOnly::take(SWIG_STD_MOVE(arg1)); // (c) Copy constructor
}
</pre>
</div>
<p> Note that <tt>SWIG_STD_MOVE</tt> is a macro defined as shown below
to use <tt>std::move</tt> which is only available from C++11 onwards:</p>
<div class="code">
<pre>
#if __cplusplus &gt;=201103L
# define SWIG_STD_MOVE(OBJ) std::move(OBJ)
#else
# define SWIG_STD_MOVE(OBJ) OBJ
#endif
</pre>
</div>
<p> Also note:<i> (c) Copy constructor</i>. Yes, when passing by value
the copy constructor is called for all versions of C++, even C++11 and
later even though std::move is specified. It's a C++ language feature
for types that don't have move semantics!</p>
<p> The generated code for <tt>MovableCopyable::take</tt> is the same as
for <tt>CopyOnly::take</tt>, however, the C++ compiler will choose the
move constructor this time where commented<i> (c) Move constructor</i>:</p>
<div class="code">
<pre>
SWIGEXPORT void SWIGSTDCALL CSharp_MovableCopyable_take(void * jarg1) {
MovableCopyable arg1 ; // (a) Default constructor
MovableCopyable *argp1 ;
argp1 = (MovableCopyable *)jarg1;
if (!argp1) {
SWIG_CSharpSetPendingExceptionArgument(SWIG_CSharpArgumentNullException, &quot;Attempt to dereference null MovableCopyable&quot;, 0);
return ;
}
arg1 = *argp1; // (b) Copy assignment
MovableCopyable::take(SWIG_STD_MOVE(arg1)); // (c) Move constructor
}
</pre>
</div>
<p> There are two optimisation opportunities available.</p>
<ol>
<li> Remove the default constructor call with the <tt>
%feature(&quot;valuewrapper&quot;)</tt> covered in <a href="#SWIGPlus_nn19">Pass
and return by value</a> and replace it with <tt>SwigValueWrapper</tt>.</li>
<li> Apply the SWIGTYPE MOVE typemaps which are designed specifically to
implement full move semantics when passing parameters by value. They
replace the copy assignment with a call to <tt>SwigValueWrapper::reset</tt>
, which works much like <tt>std::unique_ptr::reset</tt>. These typemaps
could alternatively have replaced the copy assignment with a move
assignment, but this is not maximally optimal.</li>
</ol>
<p> Simply add the following before the <tt>MovableCopyable::take</tt>
method is parsed:</p>
<div class="code">
<pre>
%valuewrapper MovableCopyable;
%include &lt;swigmove.i&gt;
%apply SWIGTYPE MOVE { MovableCopyable }
</pre>
</div>
<p> will result in this optimal code where just one move constructor is
invoked:</p>
<div class="code">
<pre>
SWIGEXPORT void SWIGSTDCALL CSharp_MovableCopyable_take(void * jarg1) {
SwigValueWrapper&lt; MovableCopyable &gt; arg1 ; // (a) No constructors invoked
MovableCopyable *argp1 ;
argp1 = (MovableCopyable *)jarg1;
if (!argp1) {
SWIG_CSharpSetPendingExceptionArgument(SWIG_CSharpArgumentNullException, &quot;Attempt to dereference null MovableCopyable&quot;, 0);
return ;
}
SwigValueWrapper&lt; MovableCopyable &gt;::reset(arg1, argp1); // (b) No constructor or assignment operator invoked
MovableCopyable::take(SWIG_STD_MOVE(arg1)); // (c) Move constructor
}
</pre>
</div>
<p> Note that <tt>SwigValueWrapper</tt> will call the destructor for the
pointer passed to it in the <tt>reset</tt> function. This pointer is
the underlying C++ object that the proxy class owns. The details aren't
shown, but the 'csin' typemap also generates C# code to ensure that the
proxy class releases ownership of the object. Please see the 'SWIGTYPE
MOVE' typemaps in the swigmove.i file provided for each target
language. Therefore full move semantics are implemented; ownership is
moved from the proxy class into the C++ layer and the net effect is the
same as using an <a href="#CPlusPlus11_rvalue_reference_inputs">rvalue
reference parameter</a> discussed earlier.</p>
<p> Lastly, let's consider the <tt>MoveOnly::take</tt> function defined
earlier. By default the generated code fails to compile as <tt>MoveOnly</tt>
does not have a copy assignment operator. SWIG is not designed to
select a different typemap automatically for move-only types and the
user must apply the SWIGTYPE MOVE typemaps to ensure that only
move-only semantics are used. However, SWIG is able to automatically
use <tt>%feature(&quot;valuewrapper&quot;)</tt> for move-only types so it is not
necessary to explicitly use this feature. So in this move-only case,
simply add the following before <tt>MoveOnly::take</tt> is parsed,
which results in the same optimal code shown above for <tt>
MovableCopyable</tt>:</p>
<div class="code">
<pre>
%include &lt;swigmove.i&gt;
%apply SWIGTYPE MOVE { MoveOnly }
</pre>
</div>
<p><b> Compatibility note:</b> SWIG-4.1.0 introduced support for taking
advantage of types with move semantics and making it possible to easily
use move only types.</p>
<h3><a name="CPlusPlus11_generalized_constant_expressions">7.2.2
Generalized constant expressions</a></h3>
<p>SWIG parses and identifies the keyword <tt>constexpr</tt>, but cannot
fully utilise it. These C++ compile time constants are usable as
runtime constants from the target languages. Below shows example usage
for assigning a C++ compile time constant from a compile time constant
function:</p>
<div class="code">
<pre>
constexpr int XXX() { return 10; }
constexpr int YYY = XXX() + 100;
</pre>
</div>
<p> When either of these is used from a target language, a runtime call
is made to obtain the underlying constant.</p>
<h3><a name="CPlusPlus11_extern_template">7.2.3 Extern template</a></h3>
<p>SWIG correctly parses <tt>extern template</tt> explicit instantiation
declarations. However, this template instantiation suppression in a
translation unit has no relevance outside of the C++ compiler and so is
not used by SWIG. SWIG only uses <tt>%template</tt> for instantiating
and wrapping templates. Consider the class template below:</p>
<div class="code">
<pre>
// Class template
template class std::vector&lt;int&gt;; // C++03 template explicit instantiation definition in C++
extern template class std::vector&lt;int&gt;; // C++11 template explicit instantiation declaration (extern template)
%template(VectorInt) std::vector&lt;int&gt;; // SWIG template instantiation
</pre>
</div>
<p> The above result in warnings:</p>
<div class="shell">
<pre>
example.i:2: Warning 320: Explicit template instantiation ignored.
example.i:3: Warning 327: Extern template ignored.
</pre>
</div>
<p> Similarly for the function template below:</p>
<div class="code">
<pre>
// Function template
template void Func&lt;int&gt;(); // C++03 template explicit instantiation definition in C++
extern template void Func&lt;int&gt;(); // C++11 template explicit instantiation declaration (extern template)
%template(FuncInt) Func&lt;int&gt;; // SWIG template instantiation
</pre>
</div>
<h3><a name="CPlusPlus11_initializer_lists">7.2.4 Initializer lists</a></h3>
<p> Initializer lists are very much a C++ compiler construct and are not
very accessible from wrappers as they are intended for compile time
initialization of classes using the special <tt>std::initializer_list</tt>
type. SWIG detects usage of initializer lists and will emit a special
informative warning each time one is used:</p>
<div class="shell">
<pre>
example.i:33: Warning 476: Initialization using std::initializer_list.
</pre>
</div>
<p> Initializer lists usually appear in constructors but can appear in
any function or method. They often appear in constructors which are
overloaded with alternative approaches to initializing a class, such as
the std container's push_back method for adding elements to a
container. The recommended approach then is to simply ignore the
initializer-list constructor, for example:</p>
<div class="code">
<pre>
%ignore Container::Container(std::initializer_list&lt;int&gt;);
class Container {
public:
Container(std::initializer_list&lt;int&gt;); // initializer-list constructor
Container();
void push_back(const int &amp;);
...
};
</pre>
</div>
<p>Alternatively you could modify the class and add another constructor
for initialization by some other means, for example by a <tt>
std::vector</tt>:</p>
<div class="code">
<pre>
%include &lt;std_vector.i&gt;
class Container {
public:
Container(const std::vector&lt;int&gt; &amp;);
Container(std::initializer_list&lt;int&gt;); // initializer-list constructor
Container();
void push_back(const int &amp;);
...
};
</pre>
</div>
<p>And then call this constructor from your target language, for
example, in Python, the following will call the constructor taking the <tt>
std::vector</tt>:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; c = Container( [1, 2, 3, 4] )
</pre>
</div>
<p> If you are unable to modify the class being wrapped, consider
ignoring the initializer-list constructor and using %extend to add in
an alternative constructor:</p>
<div class="code">
<pre>
%include &lt;std_vector.i&gt;
%extend Container {
Container(const std::vector&lt;int&gt; &amp;elements) {
Container *c = new Container();
for (int element : elements)
c-&gt;push_back(element);
return c;
}
}
%ignore Container::Container(std::initializer_list&lt;int&gt;);
class Container {
public:
Container(std::initializer_list&lt;int&gt;); // initializer-list constructor
Container();
void push_back(const int &amp;);
...
};
</pre>
</div>
<p> The above makes the wrappers look is as if the class had been
declared as follows:</p>
<div class="code">
<pre>
%include &lt;std_vector.i&gt;
class Container {
public:
Container(const std::vector&lt;int&gt; &amp;);
// Container(std::initializer_list&lt;int&gt;); // initializer-list constructor (ignored)
Container();
void push_back(const int &amp;);
...
};
</pre>
</div>
<p> <tt>std::initializer_list</tt> is simply a container that can only
be initialized at compile time. As it is just a C++ type, it is
possible to write typemaps for a target language container to map onto <tt>
std::initializer_list</tt>. However, this can only be done for a fixed
number of elements as initializer lists are not designed to be
constructed with a variable number of arguments at runtime. The example
below is a very simple approach which ignores any parameters passed in
and merely initializes with a fixed list of fixed integer values chosen
at compile time:</p>
<div class="code">
<pre>
%typemap(in) std::initializer_list&lt;int&gt; {
$1 = {10, 20, 30, 40, 50};
}
class Container {
public:
Container(std::initializer_list&lt;int&gt;); // initializer-list constructor
Container();
void push_back(const int &amp;);
...
};
</pre>
</div>
<p> Any attempt at passing in values from the target language will be
ignored and be replaced by <tt>{10, 20, 30, 40, 50}</tt>. Needless to
say, this approach is very limited, but could be improved upon, but
only slightly. A typemap could be written to map a fixed number of
elements on to the <tt>std::initializer_list</tt>, but with values
decided at runtime. The typemaps would be target language specific.</p>
<p> Note that the default typemap for <tt>std::initializer_list</tt>
does nothing but issue the warning and hence any user supplied typemaps
will override it and suppress the warning.</p>
<h3><a name="CPlusPlus11_uniform_initialization">7.2.5 Uniform
initialization</a></h3>
<p>The curly brackets {} for member initialization are fully supported
by SWIG:</p>
<div class="code">
<pre>
struct BasicStruct {
int x;
double y;
};
struct AltStruct {
AltStruct(int x, double y) : x_{x}, y_{y} {}
int x_;
double y_;
};
BasicStruct var1{5, 3.2}; // only fills the struct components
AltStruct var2{2, 4.3}; // calls the constructor
</pre>
</div>
<p>Uniform initialization does not affect usage from the target
language, for example in Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = AltStruct(10, 142.15)
&gt;&gt;&gt; a.x_
10
&gt;&gt;&gt; a.y_
142.15
</pre>
</div>
<h3><a name="CPlusPlus11_type_inference">7.2.6 Type inference</a></h3>
<p><tt>decltype()</tt> is supported with a few limitations. SWIG can
parse all uses, but can't deduce the type in every situation where a
C++ compiler can. The cases SWIG can deduce have expanded with time and
hopefully will continue to. For example, for the code</p>
<div class="code">
<pre>
int i;
decltype(i) j;
decltype(i+j) k;
</pre>
</div>
<p>SWIG is able to deduce that the variable <tt>i</tt> and the
expression <tt>i+j</tt> both have type <tt>int</tt>.</p>
<p> Using an expression for the decltype which SWIG can't handle results
in a warning:</p>
<div class="code">
<pre>
int foo(int);
decltype(foo(0)) k; // Warning 344: Unable to deduce decltype for 'foo(0)'.
</pre>
</div>
<p> This warning should be viewed as a prompt to add in a manual ignore
of the variable/function as in most cases the generated code will not
compile. For the example above, ignore the symbol that is declared
using <tt>decltype</tt> and perhaps additionally suppress the warning
as follows:</p>
<div class="code">
<pre>
#pragma SWIG nowarn=SWIGWARN_CPP11_DECLTYPE
%ignore k;
</pre>
</div>
<p> If an ignore is not acceptable, a workaround is to redefine the
symbol with the actual type, for example:</p>
<div class="code">
<pre>
int k; // define k with the actual type
%ignore k; // ignore the real definition of k
</pre>
</div>
<p>You would typically put one of these workarounds in your interface
file before using <tt>%include</tt> to get SWIG to parse the header
which defines <tt>k</tt>.</p>
<p>SWIG supports <tt>auto</tt> as a type specifier for variables (with
the same limitations for actually deducing the type as for <tt>
decltype()</tt>), and for specifying the return type of <a href="#CPlusPlus11_lambda_functions_and_expressions">
lambdas</a> and <a href="#CPlusPlus11_alternate_function_syntax">
functions</a>.</p>
<h3><a name="CPlusPlus11_range_based_for_loop">7.2.7 Range-based
for-loop</a></h3>
<p>This feature is part of the implementation block only. SWIG ignores
it.</p>
<h3><a name="CPlusPlus11_lambda_functions_and_expressions">7.2.8 Lambda
functions and expressions</a></h3>
<p>SWIG correctly parses most of the Lambda functions syntax. For
example:</p>
<div class="code">
<pre>
auto val = [] { return something; };
auto sum = [](int x, int y) { return x+y; };
auto sum = [](int x, int y) -&gt; int { return x+y; };
</pre>
</div>
<p>The lambda functions are removed from the wrappers for now, because
of the lack of support for closures (scope of the lambda functions) in
the target languages.</p>
<p> Lambda functions used to create variables can also be parsed, but
due to limited support of <tt>auto</tt> when the type is deduced from
the expression, the variables are simply ignored.</p>
<div class="code">
<pre>
auto six = [](int x, int y) { return x+y; }(4, 2);
</pre>
</div>
<p> Better support should be available in a later release.</p>
<h3><a name="CPlusPlus11_alternate_function_syntax">7.2.9 Alternate
function syntax</a></h3>
<p>SWIG fully supports the new definition of functions. For example:</p>
<div class="code">
<pre>
struct SomeStruct {
int FuncName(int x, int y);
};
</pre>
</div>
<p>can now be written as in C++11:</p>
<div class="code">
<pre>
struct SomeStruct {
auto FuncName(int x, int y) -&gt; int;
};
auto SomeStruct::FuncName(int x, int y) -&gt; int {
return x + y;
}
</pre>
</div>
<p>The usage in the target languages remains the same, for example in
Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = SomeStruct()
&gt;&gt;&gt; a.FuncName(10, 5)
15
</pre>
</div>
<p>SWIG will also deal with type inference for the return type, as per
the limitations described earlier. For example:</p>
<div class="code">
<pre>
auto square(float a, float b) -&gt; decltype(a);
</pre>
</div>
<h3><a name="CPlusPlus11_object_construction_improvement">7.2.10 Object
construction improvement</a></h3>
<p> There are three parts to object construction improvement. The first
improvement is constructor delegation such as the following:</p>
<div class="code">
<pre>
class A {
public:
int a;
int b;
int c;
A() : A(10) {}
A(int aa) : A(aa, 20) {}
A(int aa, int bb) : A(aa, bb, 30) {}
A(int aa, int bb, int cc) { a=aa; b=bb; c=cc; }
};
</pre>
</div>
<p> where peer constructors can be called. SWIG handles this without any
issue.</p>
<p> The second improvement is constructor inheritance via a <tt>using</tt>
declaration. The extra constructors provided by the <tt>using</tt>
declaration will add the appropriate constructors into the target
language proxy derived classes. In the example below a wrapper for the <tt>
DerivedClass(int)</tt> constructor is added to <tt>DerivedClass</tt>:</p>
<div class="code">
<pre>
class BaseClass {
public:
BaseClass(int iValue);
};
class DerivedClass: public BaseClass {
public:
using BaseClass::BaseClass; // Adds DerivedClass(int) constructor
};
</pre>
</div>
<p><b> Compatibility note:</b> SWIG-4.2.0 was the first version to
generate wrappers for constructors inherited via <tt>using</tt>
declarations.</p>
<p> The final part is member initialization at the site of the
declaration. This kind of initialization is handled by SWIG.</p>
<div class="code">
<pre>
class SomeClass {
public:
SomeClass() {}
explicit SomeClass(int new_value) : value(new_value) {}
int value = 5;
};
</pre>
</div>
<h3><a name="CPlusPlus11_explicit_overrides_final">7.2.11 Explicit
overrides and final</a></h3>
<p> The special identifiers <tt>final</tt> and <tt>override</tt> can be
used on methods and destructors, such as in the following example:</p>
<div class="code">
<pre>
struct BaseStruct {
virtual void ab() const = 0;
virtual void cd();
virtual void ef();
virtual ~BaseStruct();
};
struct DerivedStruct : BaseStruct {
virtual void ab() const override;
virtual void cd() final;
virtual void ef() final override;
virtual ~DerivedStruct() override;
};
</pre>
</div>
<p> Classes can also be marked as final, such as</p>
<div class="code">
<pre>
struct FinalDerivedStruct final : BaseStruct {
virtual void ab() const override;
};
</pre>
</div>
<p><b> Compatibility note:</b> Final methods were supported much earlier
than final classes. SWIG-4.1.0 was the first version to support classes
marked as final.</p>
<h3><a name="CPlusPlus11_null_pointer_constant">7.2.12 Null pointer
constant</a></h3>
<p>The <tt>nullptr</tt> constant is mostly unimportant in wrappers. In
the few places it has an effect, it is treated like <tt>NULL</tt>.</p>
<h3><a name="CPlusPlus11_strongly_typed_enumerations">7.2.13 Strongly
typed enumerations</a></h3>
<p>SWIG supports strongly typed enumerations and parses the new <tt>enum
class</tt> syntax and forward declarator for the enums, such as:</p>
<div class="code">
<pre>
enum class MyEnum : unsigned int;
</pre>
</div>
<p> Strongly typed enums are often used to avoid name clashes such as
the following:</p>
<div class="code">
<pre>
struct Color {
enum class RainbowColors : unsigned int {
Red, Orange, Yellow, Green, Blue, Indigo, Violet
};
enum class WarmColors {
Yellow, Orange, Red
};
// Note normal enum
enum PrimeColors {
Red=100, Green, Blue
};
};
</pre>
</div>
<p> There are various ways that the target languages handle enums, so it
is not possible to precisely state how they are handled in this
section. However, generally, most scripting languages mangle in the
strongly typed enumeration's class name, but do not use any additional
mangling for normal enumerations. For example, in Python, the following
code</p>
<div class="targetlang">
<pre>
print(Color.RainbowColors_Red, Color.WarmColors_Red, Color.Red)
</pre>
</div>
<p> results in</p>
<div class="shell">
<pre>
0 2 100
</pre>
</div>
<p> The strongly typed languages often wrap normal enums into an enum
class and so treat normal enums and strongly typed enums the same. The
equivalent in Java is:</p>
<div class="targetlang">
<pre>
System.out.println(Color.RainbowColors.Red.swigValue() + &quot; &quot; + Color.WarmColors.Red.swigValue() + &quot; &quot; + Color.PrimeColors.Red.swigValue());
</pre>
</div>
<p> The C++11 enum base type, such as <tt>unsigned int</tt>, in the
example above, is used by some language modules and is missing support
in others. For example, in C#, the enum base type in the example above
is used and converted into a C# <tt>uint</tt> to specify the underlying
C# enumeration type as follows:</p>
<div class="targetlang">
<pre>
public enum RainbowColors : uint {
Red,
Orange,
Yellow,
Green,
Blue,
Indigo,
Violet
}
</pre>
</div>
<h3><a name="CPlusPlus11_double_angle_brackets">7.2.14 Double angle
brackets</a></h3>
<p>SWIG correctly parses the symbols &gt;&gt; as closing the template block,
if found inside it at the top level, or as the right shift operator &gt;&gt;
otherwise.</p>
<div class="code">
<pre>
std::vector&lt;std::vector&lt;int&gt;&gt; myIntTable;
</pre>
</div>
<h3><a name="CPlusPlus11_explicit_conversion_operators">7.2.15 Explicit
conversion operators</a></h3>
<p>SWIG correctly parses the keyword <tt>explicit</tt> for operators in
addition to constructors now. For example:</p>
<div class="code">
<pre>
class U {
public:
int u;
};
class V {
public:
int v;
};
class TestClass {
public:
//implicit converting constructor
TestClass(U const &amp;val) { t=val.u; }
// explicit constructor
explicit TestClass(V const &amp;val) { t=val.v; }
int t;
};
struct Testable {
// explicit conversion operator
explicit operator bool() const {
return false;
}
};
</pre>
</div>
<p> The effect of explicit constructors and operators has little
relevance for the proxy classes as target languages don't have the same
concepts of implicit conversions as C++. Conversion operators either
with or without <tt>explicit</tt> need renaming to a valid identifier
name in order to make them available as a normal proxy method.</p>
<h3><a name="CPlusPlus11_alias_templates">7.2.16 Type alias and alias
templates</a></h3>
<p> A type alias is a statement of the form:</p>
<div class="code">
<pre>
using PFD = void (*)(double); // New introduced syntax
</pre>
</div>
<p> which is equivalent to the old style typedef:</p>
<div class="code">
<pre>
typedef void (*PFD)(double); // The old style
</pre>
</div>
<p> The following is an example of an alias template:<div class="code">
<pre>
template&lt; typename T1, typename T2, int N &gt;
class SomeType {
public:
T1 a;
T2 b;
};
template&lt; typename T2 &gt;
using TypedefName = SomeType&lt;char*, T2, 5&gt;;
</pre>
</div></p>
<p> SWIG supports both type aliasing and alias templates. However, in
order to use an alias template, two <tt>%template</tt> directives must
be used:</p>
<div class="code">
<pre>
%template(SomeTypeBool) SomeType&lt;char*, bool, 5&gt;;
%template() TypedefName&lt;bool&gt;;
</pre>
</div>
<p>Firstly, the actual template is instantiated with a name to be used
by the target language, as per any template being wrapped. Secondly,
the empty template instantiation, <tt>%template()</tt>, is required for
the alias template. This second requirement is necessary to add the
appropriate instantiated template type into the type system as SWIG
does not automatically instantiate templates. See the <a href="#SWIGPlus_nn30">
Templates</a> section for more general information on wrapping
templates.</p>
<h3><a name="CPlusPlus11_unrestricted_unions">7.2.17 Unrestricted unions</a>
</h3>
<p>SWIG fully supports any type inside a union even if it does not
define a trivial constructor. For example, the wrapper for the
following code correctly provides access to all members in the union:</p>
<div class="code">
<pre>
struct point {
point() {}
point(int x, int y) : x_(x), y_(y) {}
int x_, y_;
};
#include &lt;new&gt; // For placement 'new' in the constructor below
union P {
int z;
double w;
point p; // Illegal in C++03; legal in C++11.
// Due to the point member, a constructor definition is required.
P() {
new(&amp;p) point();
}
} p1;
</pre>
</div>
<h3><a name="CPlusPlus11_variadic_templates">7.2.18 Variadic templates</a>
</h3>
<p>SWIG supports the variadic templates including the &lt;&gt; variadic class
inheritance, variadic methods, variadic constructors and initializers.
Example:</p>
<div class="code">
<pre>
template &lt;typename... BaseClasses&gt; class ClassName : public BaseClasses... {
public:
ClassName(BaseClasses &amp;&amp;... baseClasses) : BaseClasses(baseClasses)... {}
void InstanceMethod(const BaseClasses&amp;... baseClasses) {}
};
</pre>
</div>
<p> The <tt>%template</tt> directive works as expected for variable
template parameters.</p>
<div class="code">
<pre>
struct A {
virtual void amethod();
virtual ~A();
};
struct B {
virtual void bmethod();
virtual ~B();
};
%template(ClassName0) ClassName&lt;&gt;
%template(ClassName1) ClassName&lt;A&gt;
%template(ClassName2) ClassName&lt;A, B&gt;
</pre>
</div>
<p> Example usage from say Python:</p>
<div class="targetlang">
<pre>
cn0 = ClassName0()
cn0.InstanceMethod()
a = A()
cn1 = ClassName1(a)
cn1.amethod()
cn1.InstanceMethod(a)
b = B()
cn2 = ClassName2(a, b)
cn2.InstanceMethod(a, b)
cn2.amethod()
cn2.bmethod()
</pre>
</div>
<p>Support for the variadic <tt>sizeof()</tt> function also works:</p>
<div class="code">
<pre>
const int SIZE = sizeof...(ClassName&lt;A, B&gt;);
</pre>
</div>
<p> In the above example <tt>SIZE</tt> is of course wrapped as a
constant.</p>
<p><b> Compatibility note:</b> SWIG-4.2.0 was the first version to fully
support variadic templates. SWIG-3.0.0 provided initial support and was
limited to only one variadic parameter.</p>
<h3><a name="CPlusPlus11_new_char_literals">7.2.19 New character
literals</a></h3>
<p> C++11 adds support for UCS-2 and UCS-4 character literals. These
character literals are preceded by either 'u' or 'U'.</p>
<div class="code">
<pre>
char16_t a = u'a';
char32_t b = U'b';
</pre>
</div>
<p><b> Compatibility note:</b> SWIG-4.0.0 was the first version to
support these Universal Coded Character Set (UCS) character literals.</p>
<h3><a name="CPlusPlus11_new_string_literals">7.2.20 New string literals</a>
</h3>
<p>SWIG supports wide string and Unicode string constants and raw string
literals.</p>
<div class="code">
<pre>
// New string literals
wstring aa = L&quot;Wide string&quot;;
const char *bb = u8&quot;UTF-8 string&quot;;
const char16_t *cc = u&quot;UTF-16 string&quot;;
const char32_t *dd = U&quot;UTF-32 string&quot;;
// Raw string literals
const char *xx = &quot;)I'm an \&quot;ascii\&quot; \\ string.&quot;;
const char *ee = R&quot;XXX()I'm an &quot;ascii&quot; \ string.)XXX&quot;; // same as xx
wstring ff = LR&quot;XXX(I'm a &quot;raw wide&quot; \ string.)XXX&quot;;
const char *gg = u8R&quot;XXX(I'm a &quot;raw UTF-8&quot; \ string.)XXX&quot;;
const char16_t *hh = uR&quot;XXX(I'm a &quot;raw UTF-16&quot; \ string.)XXX&quot;;
const char32_t *ii = UR&quot;XXX(I'm a &quot;raw UTF-32&quot; \ string.)XXX&quot;;
</pre>
</div>
<p> Non-ASCII string support varies quite a bit among the various target
languages though.</p>
<p> Note: There is a bug currently where SWIG's preprocessor incorrectly
parses an odd number of double quotes inside raw string literals.</p>
<h3><a name="CPlusPlus11_user_defined_literals">7.2.21 User-defined
literals</a></h3>
<p> SWIG parses the declaration of user-defined literals, that is, the <tt>
operator &quot;&quot; _mysuffix()</tt> function syntax.</p>
<p> Some examples are the raw literal:</p>
<div class="code">
<pre>
OutputType operator &quot;&quot; _myRawLiteral(const char * value);
</pre>
</div>
<p> numeric cooked literals:</p>
<div class="code">
<pre>
OutputType operator &quot;&quot; _mySuffixIntegral(unsigned long long);
OutputType operator &quot;&quot; _mySuffixFloat(long double);
</pre>
</div>
<p> and cooked string literals:</p>
<div class="code">
<pre>
OutputType operator &quot;&quot; _mySuffix(const char * string_values, size_t num_chars);
OutputType operator &quot;&quot; _mySuffix(const wchar_t * string_values, size_t num_chars);
OutputType operator &quot;&quot; _mySuffix(const char16_t * string_values, size_t num_chars);
OutputType operator &quot;&quot; _mySuffix(const char32_t * string_values, size_t num_chars);
</pre>
</div>
<p> Like other operators that SWIG parses, a warning is given about
renaming the operator in order for it to be wrapped:</p>
<div class="shell">
<pre>
example.i:27: Warning 503: Can't wrap 'operator &quot;&quot; _myRawLiteral' unless renamed to a valid identifier.
</pre>
</div>
<p> If %rename is used, then it can be called like any other wrapped
method. Currently you need to specify the full declaration including
parameters for %rename:</p>
<div class="code">
<pre>
%rename(MyRawLiteral) operator&quot;&quot; _myRawLiteral(const char * value);
</pre>
</div>
<p> Or if you just wish to ignore it altogether:</p>
<div class="code">
<pre>
%ignore operator &quot;&quot; _myRawLiteral(const char * value);
</pre>
</div>
<p> Note that use of user-defined literals such as the following still
give a syntax error:</p>
<div class="code">
<pre>
OutputType var1 = &quot;1234&quot;_suffix;
OutputType var2 = 1234_suffix;
OutputType var3 = 3.1416_suffix;
</pre>
</div>
<h3><a name="CPlusPlus11_thread_local_storage">7.2.22 Thread-local
storage</a></h3>
<p>SWIG correctly parses the <tt>thread_local</tt> keyword. For example,
variables reachable by the current thread can be defined as:</p>
<div class="code">
<pre>
struct A {
static thread_local int val;
};
thread_local int global_val;
</pre>
</div>
<p> The use of the <tt>thread_local</tt> storage specifier does not
affect the wrapping process; it does not modify the wrapper code
compared to when it is not specified. A variable will be thread local
if accessed from different threads from the target language in the same
way that it will be thread local if accessed from C++ code.</p>
<h3><a name="CPlusPlus11_defaulted_deleted">7.2.23 Explicitly defaulted
functions and deleted functions</a></h3>
<p>SWIG handles explicitly defaulted functions, that is, <tt>= default</tt>
added to a function declaration. Deleted definitions, which are also
called deleted functions, have <tt>= delete</tt> added to the function
declaration. For example:</p>
<div class="code">
<pre>
struct NonCopyable {
NonCopyable &amp; operator=(const NonCopyable &amp;) = delete; /* Removes operator= */
NonCopyable(const NonCopyable &amp;) = delete; /* Removes copy constructor */
NonCopyable() = default; /* Explicitly allows the empty constructor */
};
</pre>
</div>
<p> Wrappers for deleted functions will not be available in the target
language. Wrappers for defaulted functions will of course be available
in the target language. Explicitly defaulted functions have no direct
effect for SWIG wrapping as the declaration is handled much like any
other method declaration parsed by SWIG.</p>
<p> Deleted functions are also designed to prevent implicit conversions
when calling the function. For example, the C++ compiler will not
compile any code which attempts to use an int as the type of the
parameter passed to <tt>f</tt> below:</p>
<div class="code">
<pre>
struct NoInt {
void f(double i);
void f(int) = delete;
};
</pre>
</div>
<p> This is a C++ compile time check and SWIG does not make any attempt
to detect if the target language is using an int instead of a double
though, so in this case it is entirely possible to pass an int instead
of a double to <tt>f</tt> from Java, Python etc.</p>
<h3><a name="CPlusPlus11_type_long_long_int">7.2.24 Type long long int</a>
</h3>
<p>SWIG correctly parses and uses the new <tt>long long</tt> type
already introduced in C99 some time ago.</p>
<h3><a name="CPlusPlus11_static_assertions">7.2.25 Static assertions</a></h3>
<p> SWIG correctly parses the new <tt>static_assert</tt> declarations
(though 3.0.12 and earlier had a bug which meant this wasn't accepted
at file scope). This is a C++ compile time directive so there isn't
anything useful that SWIG can do with it.</p>
<div class="code">
<pre>
template &lt;typename T&gt;
struct Check {
static_assert(sizeof(int) &lt;= sizeof(T), &quot;not big enough&quot;);
};
</pre>
</div>
<h3><a name="CPlusPlus11_sizeof">7.2.26 Allow sizeof to work on members
of classes without an explicit object</a></h3>
<p> SWIG can parse the new sizeof() on types as well as on objects. For
example:</p>
<div class="code">
<pre>
struct A {
int member;
};
const int SIZE = sizeof(A::member); // does not work with C++03. Okay with C++11
</pre>
</div>
<p>In Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; SIZE
8
</pre>
</div>
<h3><a name="CPlusPlus11_noexcept">7.2.27 Exception specifications and
noexcept</a></h3>
<p> C++11 added in the noexcept specification to exception
specifications to indicate that a function simply may or may not throw
an exception, without actually naming any exception. SWIG understands
these, although there isn't any useful way that this information can be
taken advantage of by target languages, so it is as good as ignored
during the wrapping process. Below are some examples of noexcept in
function declarations:</p>
<div class="code">
<pre>
static void noex1() noexcept;
int noex2(int) noexcept(true);
int noex3(int, bool) noexcept(false);
</pre>
</div>
<h3><a name="CPlusPlus11_alignment">7.2.28 Control and query object
alignment</a></h3>
<p> An <tt>alignof</tt> operator is used mostly within C++ to return
alignment in number of bytes, but could be used to initialize a
variable as shown below. The variable's value will be available for
access by the target language as any other variable's compile time
initialised value.<div class="code">
<pre>
const int align1 = alignof(A::member);
</pre>
</div></p>
<p> The <tt>alignas</tt> specifier for variable alignment is not yet
supported. Example usage:</p>
<div class="code">
<pre>
struct alignas(16) S {
int num;
};
alignas(double) unsigned char c[sizeof(double)];
</pre>
</div>
<p> Use the preprocessor to work around this for now:</p>
<div class="code">
<pre>
#define alignas(T)
</pre>
</div>
<h3><a name="CPlusPlus11_attributes">7.2.29 Attributes</a></h3>
<p> Attributes such as those shown below, are supported since SWIG 4.1.0
but are currently crudely ignored by the parser's tokeniser so they
have no effect on SWIG's code generation.</p>
<div class="code">
<pre>
int [[attr1]] i [[attr2, attr3]];
[[noreturn, nothrow]] void f [[noreturn]] ();
</pre>
</div>
<h3><a name="CPlusPlus11_ref_qualifiers">7.2.30 Methods with
ref-qualifiers</a></h3>
<p> C++11 non-static member functions can be declared with
ref-qualifiers. Member functions declared with a <tt>&amp;</tt> lvalue
ref-qualifiers are wrapped like any other function without
ref-qualifiers. Member functions declared with a <tt>&amp;&amp;</tt> rvalue
ref-qualifiers are ignored by default as they are unlikely to be
required from non-C++ languages where the concept of<i> rvalue-ness</i>
for the implied *this pointer does not apply. The warning is hidden by
default, but can be displayed as described in the section on <a href="#Warnings_nn4">
Enabling extra warnings</a>.</p>
<p> Consider:</p>
<div class="code">
<pre>
struct RQ {
void m1(int x) &amp;;
void m2(int x) &amp;&amp;;
};
</pre>
</div>
<p> The only wrapped method will be the lvalue ref-qualified method <tt>
m1</tt> and if SWIG is run with the <tt>-Wextra</tt> command-line
option, the following warning will be issued indicating <tt>m2</tt> is
not wrapped:</p>
<div class="shell">
<pre>
example.i:7: Warning 405: Method with rvalue ref-qualifier m2(int) &amp;&amp; ignored.
</pre>
</div>
<p> If you unignore the method as follows, wrappers for <tt>m2</tt> will
be generated:</p>
<div class="code">
<pre>
%feature(&quot;ignore&quot;, &quot;0&quot;) RQ::m2(int x) &amp;&amp;;
struct RQ {
void m1(int x) &amp;;
void m2(int x) &amp;&amp;;
};
</pre>
</div>
<p> Inspection of the generated C++ code, will show that <tt>std::move</tt>
is used on the instance of the <tt>RQ *</tt> class:</p>
<div class="code">
<pre>
RQ *arg1 = (RQ *) 0 ;
int arg2 ;
arg1 = ...marshalled from target language...
arg2 = ...marshalled from target language...
std::move(*arg1).m2(arg2);
</pre>
</div>
<p> This will compile but when run, the move effects may not be what you
want. As stated earlier, rvalue ref-qualifiers aren't really applicable
outside the world of C++. However, if you really know what you are
doing, full control over the call to the method is possible via the
low-level &quot;action&quot; feature. This feature completely replaces the call
to the underlying function, that is, the last line in the snippet of
code above.</p>
<div class="code">
<pre>
%feature(&quot;ignore&quot;, &quot;0&quot;) RQ::m2(int x) &amp;&amp;;
%feature(&quot;action&quot;) RQ::m2(int x) &amp;&amp; %{
RQ().m2(arg2);
%}
struct RQ {
void m1(int x) &amp;;
void m2(int x) &amp;&amp;;
};
</pre>
</div>
<p> resulting in:</p>
<div class="code">
<pre>
RQ *arg1 = (RQ *) 0 ;
int arg2 ;
arg1 = ...marshalled from target language...
arg2 = ...marshalled from target language...
RQ().m2(arg2);
</pre>
</div>
<p><b> Compatibility note:</b> SWIG-4.0.0 was the first version to
support ref-qualifiers.</p>
<h2><a name="CPlusPlus11_standard_library_changes">7.3 Standard library
changes</a></h2>
<h3><a name="CPlusPlus11_threading_facilities">7.3.1 Threading
facilities</a></h3>
<p>SWIG does not currently wrap or use any of the new threading classes
introduced (thread, mutex, locks, condition variables, task). The main
reason is that SWIG target languages offer their own threading
facilities so there is limited use for them.</p>
<h3><a name="CPlusPlus11_tuple_types">7.3.2 Tuple types</a></h3>
<p> SWIG does not provide library files for the new tuple types yet.
Variadic template support requires further work to provide substantial
tuple wrappers.</p>
<h3><a name="CPlusPlus11_hash_tables">7.3.3 Hash tables</a></h3>
<p> The new hash tables in the STL are <tt>unordered_set</tt>, <tt>
unordered_multiset</tt>, <tt>unordered_map</tt>, <tt>unordered_multimap</tt>
. These are not available in all target languages. Any missing support
can in principle be easily implemented by adapting the current STL
containers.</p>
<h3><a name="CPlusPlus11_regular_expressions">7.3.4 Regular expressions</a>
</h3>
<p> While SWIG could provide wrappers for the new C++11 regular
expressions classes, there is little need as the target languages have
their own regular expression facilities.</p>
<h3><a name="CPlusPlus11_general_purpose_smart_pointers">7.3.5
General-purpose smart pointers</a></h3>
<p> SWIG provides special smart pointer handling for <tt>std::shared_ptr</tt>
in the same way it has support for <tt>boost::shared_ptr</tt>. Please
see the <a href="#Library_std_shared_ptr">shared_ptr smart pointer</a>
and <a href="#Library_std_unique_ptr">unique_ptr smart pointer</a>
library sections. There is no special smart pointer handling available
for <tt>std::weak_ptr</tt>.</p>
<h3><a name="CPlusPlus11_extensible_random_number_facility">7.3.6
Extensible random number facility</a></h3>
<p>This feature extends and standardizes the standard library only and
does not affect the C++ language nor SWIG.</p>
<h3><a name="CPlusPlus11_wrapper_reference">7.3.7 Wrapper reference</a></h3>
<p> Wrapper references are similar to normal C++ references but are
copy-constructible and copy-assignable. They could conceivably be used
in public APIs. There is no special support for <tt>
std::reference_wrapper</tt> in SWIG though. Users would need to write
their own typemaps if wrapper references are being used and these would
be similar to the plain C++ reference typemaps.</p>
<h3><a name="CPlusPlus11_polymorphous_wrappers_for_function_objects">
7.3.8 Polymorphic wrappers for function objects</a></h3>
<p> SWIG supports functor classes in a few languages in a very natural
way. However nothing is provided yet for the new <tt>std::function</tt>
template. SWIG will parse usage of the template like any other
template.</p>
<div class="code">
<pre>
%rename(__call__) Test::operator(); // Default renaming used for Python
struct Test {
bool operator()(int x, int y); // function object
};
#include &lt;functional&gt;
std::function&lt;void (int, int)&gt; pF = Test; // function template wrapper
</pre>
</div>
<p> Example of supported usage of the plain functor from Python is shown
below. It does not involve <tt>std::function</tt>.</p>
<div class="targetlang">
<pre>
t = Test()
b = t(1, 2) # invoke C++ function object
</pre>
</div>
<h3><a name="CPlusPlus11_type_traits_for_metaprogramming">7.3.9 Type
traits for metaprogramming</a></h3>
<p>The type_traits functions to support C++ metaprogramming is useful at
compile time and is aimed specifically at C++ development:</p>
<div class="code">
<pre>
#include &lt;type_traits&gt;
// First way of operating.
template&lt; bool B &gt; struct algorithm {
template&lt; class T1, class T2 &gt; static int do_it(T1 &amp;, T2 &amp;) { /*...*/ return 1; }
};
// Second way of operating.
template&lt;&gt; struct algorithm&lt;true&gt; {
template&lt; class T1, class T2 &gt; static int do_it(T1, T2) { /*...*/ return 2; }
};
// Instantiating 'elaborate' will automatically instantiate the correct way to operate, depending on the types used.
template&lt; class T1, class T2 &gt; int elaborate(T1 A, T2 B) {
// Use the second way only if 'T1' is an integer and if 'T2' is a floating point,
// otherwise use the first way.
return algorithm&lt; std::is_integral&lt;T1&gt;::value &amp;&amp; std::is_floating_point&lt;T2&gt;::value &gt;::do_it(A, B);
}
</pre>
</div>
<p> SWIG correctly parses the template specialization, template types
etc. However, metaprogramming and the additional support in the
type_traits header is really for compile time and is not much use at
runtime for the target languages. For example, as SWIG requires
explicit instantiation of templates via <tt>%template</tt>, there isn't
much that <tt>std::is_integral&lt;int&gt;</tt> is going to provide by itself.
However, template functions using such metaprogramming techniques might
be useful to wrap. For example, the following instantiations could be
made:</p>
<div class="code">
<pre>
%template(Elaborate) elaborate&lt;int, int&gt;;
%template(Elaborate) elaborate&lt;int, double&gt;;
</pre>
</div>
<p> Then the appropriate algorithm can be called for the subset of types
given by the above <tt>%template</tt> instantiations from a target
language, such as Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; Elaborate(0, 0)
1
&gt;&gt;&gt; Elaborate(0, 0.0)
2
</pre>
</div>
<h3><a name="CPlusPlus11_uniform_method_for_computing_return_type_of_function_objects">
7.3.10 Uniform method for computing return type of function objects</a></h3>
<p> The new <tt>std::result_of</tt> class introduced in the &lt;functional&gt;
header provides a generic way to obtain the return type of a function
type via <tt>std::result_of::type</tt>. There isn't any library
interface file to support this type. With a bit of work, SWIG will
deduce the return type of functions when used in <tt>std::result_of</tt>
using the approach shown below. The technique basically forward
declares the <tt>std::result_of</tt> template class, then partially
specializes it for the function types of interest. SWIG will use the
partial specialization and hence correctly use the <tt>
std::result_of::type</tt> provided in the partial specialization.</p>
<div class="code">
<pre>
%inline %{
#include &lt;functional&gt;
typedef double(*fn_ptr)(double);
%}
namespace std {
// Forward declaration of result_of
template&lt;typename Func&gt; struct result_of;
// Add in a partial specialization of result_of
template&lt;&gt; struct result_of&lt; fn_ptr(double) &gt; {
typedef double type;
};
}
%template() std::result_of&lt; fn_ptr(double) &gt;;
%inline %{
double square(double x) {
return (x * x);
}
template&lt;class Fun, class Arg&gt;
typename std::result_of&lt;Fun(Arg)&gt;::type test_result_impl(Fun fun, Arg arg) {
return fun(arg);
}
%}
%template(test_result) test_result_impl&lt; fn_ptr, double &gt;;
%constant double (*SQUARE)(double) = square;
</pre>
</div>
<p> Note the first use of <tt>%template</tt> which SWIG requires to
instantiate the template. The empty template instantiation suffices as
no proxy class is required for <tt>std::result_of&lt;Fun(Arg)&gt;::type</tt>
as this type is really just a <tt>double</tt>. The second <tt>%template</tt>
instantiates the template function which is being wrapped for use as a
callback. The <tt>%constant</tt> can then be used for any callback
function as described in <a href="#SWIG_nn30">Pointers to functions and
callbacks</a>.</p>
<p> Example usage from Python should give the not too surprising result:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; test_result(SQUARE, 5.0)
25.0
</pre>
</div>
<p> Phew, that is a lot of hard work to get a callback working. You
could just go with the more attractive option of just using <tt>double</tt>
as the return type in the function declaration instead of <tt>result_of</tt>
!</p>
<HR NOSHADE>
<h1><a name="CPlusPlus14">8 SWIG and C++14</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#CPlusPlus14_introduction">Introduction</a></li>
<li><a href="#CPlusPlus14_core_language_changes">Core language changes</a>
<ul>
<li><a href="#CPlusPlus14_binary_literals">Binary integer literals</a></li>
<li><a href="#CPlusPlus14_return_type_deduction">Return type deduction</a>
</li>
</ul>
</li>
<li><a href="#CPlusPlus14_standard_library_changes">Standard library
changes</a></li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="CPlusPlus14_introduction">8.1 Introduction</a></h2>
<p>This chapter gives you a brief overview about the SWIG implementation
of the C++14 standard. There isn't much in C++14 that affects SWIG,
however, work has only just begun on adding C++14 support.</p>
<p><b> Compatibility note:</b> SWIG-4.0.0 is the first version to
support any C++14 features.</p>
<h2><a name="CPlusPlus14_core_language_changes">8.2 Core language
changes</a></h2>
<h3><a name="CPlusPlus14_binary_literals">8.2.1 Binary integer literals</a>
</h3>
<p> C++14 added binary integer literals and SWIG supports these.
Example:</p>
<div class="code">
<pre>
int b = 0b101011;
</pre>
</div>
<h3><a name="CPlusPlus14_return_type_deduction">8.2.2 Return type
deduction</a></h3>
<p> C++14 added the ability to specify <tt>auto</tt> for the return type
of a function and have the compiler deduce it from the body of the
function (in C++11 you had to explicitly specify a trailing return type
if you used <tt>auto</tt> for the return type).</p>
<p> SWIG parses these types of functions, but with one significant
limitation: SWIG can't actually deduce the return type! If you want to
wrap such a function you will need to tell SWIG the return type
explicitly.</p>
<p> The trick for specifying the return type is to use <tt>%ignore</tt>
to tell SWIG to ignore the function with the deduced return type, but
first provide SWIG with an alternative declaration of the function with
an explicit return type. The generated wrapper will wrap this
alternative declaration, and the call in the wrapper to the function
will call the actual declaration. Here is an actual example:</p>
<div class="code">
<pre>
std::tuple&lt;int, int&gt; va_static_cast();
%ignore va_static_cast();
#pragma SWIG nowarn=SWIGWARN_CPP14_AUTO
%inline %{
#include &lt;tuple&gt;
auto va_static_cast() {
return std::make_tuple(0, 0);
}
%}
</pre>
</div>
<p> For member methods the trick is to use <tt>%extend</tt> to redeclare
the method and call it as follows:</p>
<div class="code">
<pre>
%extend X {
const char * a() const { return $self-&gt;a(); }
}
%inline %{
struct X {
auto a() const {
return &quot;a string&quot;;
}
};
%}
</pre>
</div>
<p><b> Compatibility note:</b> SWIG-4.2.0 first introduced support for
functions declared with an auto return without a trailing return type.</p>
<h2><a name="CPlusPlus14_standard_library_changes">8.3 Standard library
changes</a></h2>
<HR NOSHADE>
<h1><a name="CPlusPlus17">9 SWIG and C++17</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#CPlusPlus17_introduction">Introduction</a></li>
<li><a href="#CPlusPlus17_core_language_changes">Core language changes</a>
<ul>
<li><a href="#CPlusPlus17_nested_namespaces">Nested namespace
definitions</a></li>
<li><a href="#CPlusPlus17_u8_char_literals">UTF-8 character literals</a></li>
<li><a href="#CPlusPlus17_hexadecimal_floating_literals">Hexadecimal
floating literals</a></li>
<li><a href="#CPlusPlus17_fold_expressions">Fold expressions</a></li>
</ul>
</li>
<li><a href="#CPlusPlus17_standard_library_changes">Standard library
changes</a></li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="CPlusPlus17_introduction">9.1 Introduction</a></h2>
<p>This chapter gives you a brief overview about the SWIG implementation
of the C++17 standard. There isn't much in C++17 that affects SWIG,
however, work has only just begun on adding C++17 support.</p>
<p><b> Compatibility note:</b> SWIG-4.0.0 is the first version to
support any C++17 features.</p>
<h2><a name="CPlusPlus17_core_language_changes">9.2 Core language
changes</a></h2>
<h3><a name="CPlusPlus17_nested_namespaces">9.2.1 Nested namespace
definitions</a></h3>
<p> C++17 offers a more concise syntax for defining namespaces. SWIG has
support for nested namespace definitions such as:</p>
<div class="code">
<pre>
namespace A::B::C {
...
}
</pre>
</div>
<p> This is the equivalent to the C++98 namespace definitions:</p>
<div class="code">
<pre>
namespace A {
namespace B {
namespace C {
...
}
}
}
</pre>
</div>
<h3><a name="CPlusPlus17_u8_char_literals">9.2.2 UTF-8 character
literals</a></h3>
<p> C++17 added UTF-8 (u8) character literals. These are of type char.
Example:</p>
<div class="code">
<pre>
char a = u8'a';
</pre>
</div>
<h3><a name="CPlusPlus17_hexadecimal_floating_literals">9.2.3
Hexadecimal floating literals</a></h3>
<p> C++17 added hexadecimal floating literals. For example:</p>
<div class="code">
<pre>
double f = 0xF.68p2;
</pre>
</div>
<h3><a name="CPlusPlus17_fold_expressions">9.2.4 Fold expressions</a></h3>
<p> C++17 added template fold expressions. SWIG 4.3.0 and later support
parsing these with a few restrictions. Unary left fold expressions are
not supported currently. Also the same restrictions that apply to other
expressions apply here too.</p>
<h2><a name="CPlusPlus17_standard_library_changes">9.3 Standard library
changes</a></h2>
<HR NOSHADE>
<h1><a name="CPlusPlus20">10 SWIG and C++20</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#CPlusPlus20_introduction">Introduction</a></li>
<li><a href="#CPlusPlus20_core_language_changes">Core language changes</a>
<ul>
<li><a href="#CPlusPlus20_spaceship_operator">Spaceship operator</a></li>
<li><a href="#CPlusPlus20_lambda_templates">Lambda templates</a></li>
<li><a href="#CPlusPlus20_constexpr_destructors">Constexpr destructors</a>
</li>
</ul>
</li>
<li><a href="#CPlusPlus20_preprocessor_changes">Preprocessor changes</a>
<ul>
<li><a href="#CPlusPlus20_va_opt">__VA_OPT__()</a></li>
</ul>
</li>
<li><a href="#CPlusPlus20_standard_library_changes">Standard library
changes</a></li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="CPlusPlus20_introduction">10.1 Introduction</a></h2>
<p>This chapter gives you a brief overview about the SWIG implementation
of the C++20 standard. Work has only just begun on adding C++20
support.</p>
<p><b> Compatibility note:</b> SWIG-4.1.0 is the first version to
support any C++20 features.</p>
<h2><a name="CPlusPlus20_core_language_changes">10.2 Core language
changes</a></h2>
<h3><a name="CPlusPlus20_spaceship_operator">10.2.1 Spaceship operator</a>
</h3>
<p> SWIG supports the spaceship operator <tt>&lt;=&gt;</tt> in constant
expressions. To simplify handling of the return value type, it is
currently treated as an integer rather than <tt>std::strong_ordering</tt>
, etc. In practice we think that should do the right thing in most
cases.</p>
<p> SWIG also recognises <tt>operator&lt;=&gt;</tt> which can be wrapped if
renamed. There is not currently any default renaming for the operator
or any attempt to automatically map it to a three-way comparison
operator in any of the target languages.</p>
<h3><a name="CPlusPlus20_lambda_templates">10.2.2 Lambda templates</a></h3>
<p> SWIG should parse lambda templates, but like <a href="#CPlusPlus11_lambda_functions_and_expressions">
non-templated lambdas</a> they aren't currently wrapped.</p>
<h3><a name="CPlusPlus20_constexpr_destructors">10.2.3 Constexpr
destructors</a></h3>
<p> Destructors that are declared <tt>constexpr</tt> are parsed and
handled like any other constructor. For example:</p>
<div class="code">
<pre>
class DtorA {
public:
constexpr ~DtorA() {}
};
</pre>
</div>
<h2><a name="CPlusPlus20_preprocessor_changes">10.3 Preprocessor changes</a>
</h2>
<h3><a name="CPlusPlus20_va_opt">10.3.1 __VA_OPT__()</a></h3>
<p> Support for <tt>__VA_OPT__()</tt> was added in SWIG 4.3.0.</p>
<h2><a name="CPlusPlus20_standard_library_changes">10.4 Standard library
changes</a></h2>
<HR NOSHADE>
<h1><a name="Preprocessor">11 Preprocessing</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Preprocessor_nn2">File inclusion</a></li>
<li><a href="#Preprocessor_nn3">File imports</a></li>
<li><a href="#Preprocessor_condition_compilation">Conditional
Compilation</a></li>
<li><a href="#Preprocessor_nn5">Macro Expansion</a></li>
<li><a href="#Preprocessor_nn6">SWIG Macros</a></li>
<li><a href="#Preprocessor_nn7">Variadic Macros</a></li>
<li><a href="#Preprocessor_delimiters">Preprocessing and delimiters</a>
<ul>
<li><a href="#Preprocessor_nn8">Preprocessing and %{ ... %} &amp; &quot; ... &quot;
delimiters</a></li>
<li><a href="#Preprocessor_nn9">Preprocessing and { ... } delimiters</a></li>
</ul>
</li>
<li><a href="#Preprocessor_typemap_delimiters">Preprocessor and Typemaps</a>
</li>
<li><a href="#Preprocessor_nn10">Viewing preprocessor output</a></li>
<li><a href="#Preprocessor_warning_error">The #error and #warning
directives</a></li>
<li><a href="#Preprocessor_trigraphs">Trigraphs</a></li>
<li><a href="#Preprocessor_digraphs">Digraphs</a></li>
</ul>
</div>
<!-- INDEX -->
<p> SWIG includes its own enhanced version of the C preprocessor. The
preprocessor supports the standard preprocessor directives and macro
expansion rules. However, a number of modifications and enhancements
have been made. This chapter describes some of these modifications.</p>
<h2><a name="Preprocessor_nn2">11.1 File inclusion</a></h2>
<p> To include another file into a SWIG interface, use the <tt>%include</tt>
directive like this:</p>
<div class="code">
<pre>
%include &quot;cpointer.i&quot;
</pre>
</div>
<p> Unlike, <tt>#include</tt>, <tt>%include</tt> includes each file once
(and will not reload the file on subsequent <tt>%include</tt>
declarations). Therefore, it is not necessary to use include-guards in
SWIG interfaces.</p>
<p> By default, the <tt>#include</tt> is ignored unless you run SWIG
with the <tt>-includeall</tt> option. The reason for ignoring
traditional includes is that you often don't want SWIG to try and wrap
everything included in standard header system headers and auxiliary
files.</p>
<h2><a name="Preprocessor_nn3">11.2 File imports</a></h2>
<p> SWIG provides another file inclusion directive with the <tt>%import</tt>
directive. For example:</p>
<div class="code">
<pre>
%import &quot;foo.i&quot;
</pre>
</div>
<p> The purpose of <tt>%import</tt> is to collect certain information
from another SWIG interface file or a header file without actually
generating any wrapper code. Such information generally includes type
declarations (e.g., <tt>typedef</tt>) as well as C++ classes that might
be used as base-classes for class declarations in the interface. The
use of <tt>%import</tt> is also important when SWIG is used to generate
extensions as a collection of related modules. This is an advanced
topic and is described in later in the <a href="#Modules">Working with
Modules</a> chapter.</p>
<p> The <tt>-importall</tt> directive tells SWIG to follow all <tt>
#include</tt> statements as imports. This might be useful if you want to
extract type definitions from system header files without generating
any wrappers.</p>
<h2><a name="Preprocessor_condition_compilation">11.3 Conditional
Compilation</a></h2>
<p> SWIG fully supports the use of <tt>#if</tt>, <tt>#ifdef</tt>, <tt>
#ifndef</tt>, <tt>#else</tt>, <tt>#endif</tt> to conditionally include
parts of an interface.</p>
<p> SWIG's preprocessor conditionals support the standard C/C++
preprocessor integer expressions. As a SWIG-specific extension, string
equality and inequality tests are also supported, for example:</p>
<div class="code">
<pre>
#if defined __cplusplus &amp;&amp; (#__VA_ARGS__ != &quot;&quot; || #TYPE == &quot;void&quot;)
</pre>
</div>
<p> The following symbols are predefined by SWIG when it is parsing the
interface:</p>
<div class="code">
<pre>
SWIG Always defined when SWIG is processing a file
SWIGIMPORTED Defined when SWIG is importing a file with <tt>%import</tt>
SWIG_VERSION Hexadecimal (binary-coded decimal) number containing SWIG version,
such as 0x010311 (corresponding to SWIG-1.3.11).
SWIGCSHARP Defined when using C#
SWIGD Defined when using D
SWIGGO Defined when using Go
SWIGGUILE Defined when using Guile
SWIGJAVA Defined when using Java
SWIGJAVASCRIPT Defined when using Javascript
SWIG_JAVASCRIPT_JSC Defined when using Javascript with -jsc
SWIG_JAVASCRIPT_V8 Defined when using Javascript with -v8 or -node
SWIG_JAVASCRIPT_NAPI Defined when using Javascript with -napi
SWIGLUA Defined when using Lua
SWIGMZSCHEME Defined when using Mzscheme
SWIGOCAML Defined when using OCaml
SWIGOCTAVE Defined when using Octave
SWIGPERL Defined when using Perl
SWIGPHP Defined when using PHP (any version)
SWIGPHP7 Defined when using PHP 7 or later (with a compatible C API)
SWIGPYTHON Defined when using Python
SWIGR Defined when using R
SWIGRUBY Defined when using Ruby
SWIGSCILAB Defined when using Scilab
SWIGTCL Defined when using Tcl
SWIGXML Defined when using XML
</pre>
</div>
<p> SWIG also defines <tt>SWIG_VERSION</tt> and a target language macro
in the generated wrapper file (since SWIG 4.1.0 - in older versions
these were defined for some target languages but this wasn't
consistent). Best practice is to use SWIG-time conditional checks
because that results in smaller generated wrapper sources.</p>
<p> In addition, SWIG defines the following set of standard C/C++
macros:</p>
<div class="code">
<pre>
__LINE__ Current line number
__FILE__ Current file name
__STDC__ Defined to indicate ISO C/C++
__cplusplus Defined when -c++ option used, value controlled by <tt>-std=c++NN</tt>
__STDC_VERSION__ May be defined when -c++ option is not used, value controlled by <tt>-std=cNN</tt>
</pre>
</div>
<p> Since SWIG 4.2.0, <tt>__STDC__</tt> is defined to <tt>1</tt> to
match the behaviour of ISO C/C++ compilers. Before this SWIG defined it
to have an empty value.</p>
<p> Since SWIG 4.2.0, <tt>__cplusplus</tt> is defined to <tt>199711L</tt>
(the value for C++98) by default. Before this SWIG always defined it to
have the<b> value</b> <tt>__cplusplus</tt>.</p>
<p> Since SWIG 4.2.0, SWIG supports command line options <tt>-std=cNN</tt>
and <tt>-std=c++NN</tt> to specify the C/C++ standards version. The
only effect of these options is to set appropriate values for <tt>
__STDC_VERSION__</tt> and <tt>__cplusplus</tt> respectively, which is
useful if you're wrapping headers which have preprocessor checks based
on their values.</p>
<p> If your code requires these macros to be set to a version of the
standard that is not a final official version, or one that SWIG is not
yet aware of, you can simply redefine the appropriate macro to an
alternative value at the top of your interface file, for example:</p>
<div class="code">
<pre>
#undef __cplusplus
#define __cplusplus 202211L
</pre>
</div>
<p> The following are language specific symbols that might be defined:</p>
<div class="code">
<pre>
SWIG_D_VERSION Unsigned integer target version when using D
SWIGGO_CGO Defined when using Go for cgo
SWIGGO_GCCGO Defined when using Go for gccgo
SWIGGO_INTGO_SIZE Size of the Go type int when using Go (32 or 64)
SWIGPYTHON_BUILTIN Defined when using Python with -builtin
SWIG_RUBY_AUTORENAME Defined when using Ruby with -autorename
</pre>
</div>
<p> Interface files can look at these symbols as necessary to change the
way in which an interface is generated or to mix SWIG directives with C
code.</p>
<h2><a name="Preprocessor_nn5">11.4 Macro Expansion</a></h2>
<p> Traditional preprocessor macros can be used in SWIG interfaces. Be
aware that the <tt>#define</tt> statement is also used to try and
detect constants. Therefore, if you have something like this in your
file,</p>
<div class="code">
<pre>
#ifndef FOO_H 1
#define FOO_H 1
...
#endif
</pre>
</div>
<p> you may get some extra constants such as <tt>FOO_H</tt> showing up
in the scripting interface.</p>
<p> More complex macros can be defined in the standard way. For example:</p>
<div class="code">
<pre>
#define EXTERN extern
#ifdef __STDC__
#define ISOC_(args) (args)
#else
#define ISOC_(args) ()
#endif
</pre>
</div>
<p> The following operators can appear in macro definitions:</p>
<ul>
<li><tt>#x</tt>
<br> Converts macro argument <tt>x</tt> to a string surrounded by double
quotes (&quot;x&quot;).</li>
<li><tt>x ## y</tt>
<br> Concatenates x and y together to form <tt>xy</tt>.</li>
<li><tt>`x`</tt>
<br> If <tt>x</tt> is a string surrounded by double quotes, do nothing.
Otherwise, turn into a string like <tt>#x</tt>. This is a non-standard
SWIG extension.</li>
</ul>
<h2><a name="Preprocessor_nn6">11.5 SWIG Macros</a></h2>
<p> SWIG provides an enhanced macro capability with the <tt>%define</tt>
and <tt>%enddef</tt> directives. For example:</p>
<div class="code">
<pre>
%define ARRAYHELPER(type, name)
%inline %{
type *new_ ## name (int nitems) {
return (type *) malloc(sizeof(type)*nitems);
}
void delete_ ## name(type *t) {
free(t);
}
type name ## _get(type *t, int index) {
return t[index];
}
void name ## _set(type *t, int index, type val) {
t[index] = val;
}
%}
%enddef
ARRAYHELPER(int, IntArray)
ARRAYHELPER(double, DoubleArray)
</pre>
</div>
<p> The primary purpose of <tt>%define</tt> is to define large macros of
code. Unlike normal C preprocessor macros, it is not necessary to
terminate each line with a continuation character (\)--the macro
definition extends to the first occurrence of <tt>%enddef</tt>.
Furthermore, when such macros are expanded, they are reparsed through
the C preprocessor. Thus, SWIG macros can contain all other
preprocessor directives except for nested <tt>%define</tt> statements.</p>
<p> The SWIG macro capability is a very quick and easy way to generate
large amounts of code. In fact, many of SWIG's advanced features and
libraries are built using this mechanism (such as C++ template
support).</p>
<h2><a name="Preprocessor_nn7">11.6 Variadic Macros</a></h2>
<p> SWIG-1.3.12 and newer releases support variadic preprocessor macros
which were standardised by C99 and C++11. For example:</p>
<div class="code">
<pre>
#define DEBUGF(fmt, ...) fprintf(stderr, fmt, __VA_ARGS__)
</pre>
</div>
<p> When used, any extra arguments to <tt>...</tt> are placed into the
special variable <tt>__VA_ARGS__</tt>. This also works with special
SWIG macros defined using <tt>%define</tt>.</p>
<p> The variable arguments can be empty. However, this often results in
an extra comma (<tt>,</tt>) and syntax error in the resulting
expansion. For example:</p>
<div class="code">
<pre>
DEBUGF(&quot;hello&quot;); --&gt; fprintf(stderr, &quot;hello&quot;, );
</pre>
</div>
<p> C++20 and C23 added <tt>__VA_OPT__()</tt> as a solution to this,
which SWIG 4.3.0 added support for. <tt>__VA_OPT__()</tt> expands to
its argument if the variable arguments contain any tokens, and to
nothing otherwise. It can be used to solve the problem above like so:</p>
<div class="code">
<pre>
#define DEBUGF(fmt, ...) fprintf(stderr, fmt __VA_OPT__(,) __VA_ARGS__)
</pre>
</div>
<p> An early non-standardised solution to this problem which gave a
special meaning to the token sequence <tt>, ## __VA_ARGS__</tt> is
supported by several C and C++ compilers, and also by SWIG 4.3.0 and
later (it was documented as supported by earlier SWIG versions, but
didn't actually work in at least SWIG 2.x and 3.x). Using this feature
you can get rid of the extra comma like this:</p>
<div class="code">
<pre>
#define DEBUGF(fmt, ...) fprintf(stderr, fmt, ##__VA_ARGS__)
</pre>
</div>
<p> SWIG also supports GNU-style variadic macros, which specify a name
for the variable arguments instead of using <tt>__VA_ARGS__</tt>. For
example:</p>
<div class="code">
<pre>
#define DEBUGF(fmt, args...) fprintf(stdout, fmt, args)
</pre>
</div>
<p> SWIG supports <tt>__VA_OPT__()</tt> in combination with GNU-style
variadic macros (following the lead of GCC and clang which also support
this, albeit with a warning by default).</p>
<h2><a name="Preprocessor_delimiters">11.7 Preprocessing and delimiters</a>
</h2>
<p> The preprocessor handles { }, &quot; &quot; and %{ %} delimiters differently.</p>
<h3><a name="Preprocessor_nn8">11.7.1 Preprocessing and %{ ... %} &amp; &quot;
... &quot; delimiters</a></h3>
<p> The SWIG preprocessor does not process any text enclosed in a code
block %{ ... %}. Therefore, if you write code like this,</p>
<div class="code">
<pre>
%{
#ifdef NEED_BLAH
int blah() {
...
}
#endif
%}
</pre>
</div>
<p> the contents of the <tt>%{ ... %}</tt> block are copied without
modification to the output (including all preprocessor directives).</p>
<h3><a name="Preprocessor_nn9">11.7.2 Preprocessing and { ... }
delimiters</a></h3>
<p> SWIG always runs the preprocessor on text appearing inside <tt>{ ...
}</tt>. However, sometimes it is desirable to make a preprocessor
directive pass through to the output file. For example:</p>
<div class="code">
<pre>
%extend Foo {
void bar() {
#ifdef DEBUG
printf(&quot;I'm in bar\n&quot;);
#endif
}
}
</pre>
</div>
<p> By default, SWIG will interpret the <tt>#ifdef DEBUG</tt> statement.
However, if you really wanted that code to actually go into the wrapper
file, prefix the preprocessor directives with <tt>%</tt> like this:</p>
<div class="code">
<pre>
%extend Foo {
void bar() {
%#ifdef DEBUG
printf(&quot;I'm in bar\n&quot;);
%#endif
}
}
</pre>
</div>
<p> SWIG will strip the extra <tt>%</tt> and leave the preprocessor
directive in the code.</p>
<h2><a name="Preprocessor_typemap_delimiters">11.8 Preprocessor and
Typemaps</a></h2>
<p> <a href="#Typemaps">Typemaps</a> support a special attribute called <tt>
noblock</tt> where the { ... } delimiters can be used, but the
delimiters are not actually generated into the code. The effect is then
similar to using &quot;&quot; or %{ %} delimiters but the code<b> is</b> run
through the preprocessor. For example:</p>
<div class="code">
<pre>
#define SWIG_macro(CAST) (CAST)$input
%typemap(in) Int {$1= SWIG_macro(int);}
</pre>
</div>
<p> might generate</p>
<div class="code">
<pre>
{
arg1=(int)jarg1;
}
</pre>
</div>
<p> whereas</p>
<div class="code">
<pre>
#define SWIG_macro(CAST) (CAST)$input
%typemap(in, noblock=1) Int {$1= SWIG_macro(int);}
</pre>
</div>
<p> might generate</p>
<div class="code">
<pre>
arg1=(int)jarg1;
</pre>
</div>
<p> and</p>
<div class="code">
<pre>
#define SWIG_macro(CAST) (CAST)$input
%typemap(in) Int %{$1=SWIG_macro(int);%}
</pre>
</div>
<p> would generate</p>
<div class="code">
<pre>
arg1=SWIG_macro(int);
</pre>
</div>
<h2><a name="Preprocessor_nn10">11.9 Viewing preprocessor output</a></h2>
<p> Like many compilers, SWIG supports a <tt>-E</tt> command line option
to display the output from the preprocessor. When the <tt>-E</tt>
option is used, SWIG will not generate any wrappers. Instead the
results after the preprocessor has run are displayed. This might be
useful as an aid to debugging and viewing the results of macro
expansions.</p>
<h2><a name="Preprocessor_warning_error">11.10 The #error and #warning
directives</a></h2>
<p> SWIG supports the standard <tt>#warning</tt> and <tt>#error</tt>
preprocessor directives. The <tt>#warning</tt> directive will cause
SWIG to issue a warning then continue processing. It was standardised
in C++23 and C23, and has been widely supported as an extension by most
C and C++ compilers for a long time. The <tt>#error</tt> directive will
cause SWIG to exit with a fatal error. Example usage:</p>
<div class="code">
<pre>
#error &quot;This is a fatal error message&quot;
#warning &quot;This is a warning message&quot;
</pre>
</div>
<p> The <tt>#error</tt> behaviour can be made to work like <tt>#warning</tt>
if the <tt>-cpperraswarn</tt> commandline option is used.
Alternatively, the <tt>#pragma</tt> directive can be used to the same
effect, for example:</p>
<div class="code">
<pre>
/* Modified behaviour: #error does not cause SWIG to exit with error */
#pragma SWIG cpperraswarn=1
/* Normal behaviour: #error does cause SWIG to exit with error */
#pragma SWIG cpperraswarn=0
</pre>
</div>
<h2><a name="Preprocessor_trigraphs">11.11 Trigraphs</a></h2>
<p> SWIG's preprocessor does not implement trigraphs (such as <tt>??!</tt>
being mapped to <tt>|</tt>). They are very rarely used deliberately but
these character sequences sometimes occur in code where they aren't
intended as trigraphs. Compilers typically don't enable trigraph
support by default, and they've been removed in C++17 and C23.</p>
<h2><a name="Preprocessor_digraphs">11.12 Digraphs</a></h2>
<p> SWIG's preprocessor does not currently implement digraphs (such as <tt>
&lt;%</tt> being an alternative way to write the token <tt>{</tt>). These
are standard in C++ and C95, but they're intended to support working
with code on systems with very restricted character sets which are
really rare these days so digraphs just don't seem to be used in
practice.</p>
<HR NOSHADE>
<h1><a name="Library">12 SWIG library</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Library_nn2">The %include directive and library search
path</a></li>
<li><a href="#Library_nn3">C arrays and pointers</a>
<ul>
<li><a href="#Library_argcargv">argcargv.i</a></li>
<li><a href="#Library_nn4">cpointer.i</a></li>
<li><a href="#Library_carrays">carrays.i</a></li>
<li><a href="#Library_nn6">cmalloc.i</a></li>
<li><a href="#Library_nn7">cdata.i</a></li>
</ul>
</li>
<li><a href="#Library_nn8">C string handling</a>
<ul>
<li><a href="#Library_nn9">Default string handling</a></li>
<li><a href="#Library_nn10">Passing a string with length</a></li>
<li><a href="#Library_nn11">Using %newobject to release memory</a></li>
<li><a href="#Library_nn12">cstring.i</a></li>
</ul>
</li>
<li><a href="#Library_c_standard_library">C standard library</a>
<ul>
<li><a href="#Library_complex">Complex floating types</a></li>
</ul>
</li>
<li><a href="#Library_stl_cpp_library">STL/C++ library</a>
<ul>
<li><a href="#Library_std_string">std::string</a></li>
<li><a href="#Library_std_string_view">std::string_view</a></li>
<li><a href="#Library_std_vector">std::vector</a></li>
<li><a href="#Library_stl_exceptions">STL exceptions</a></li>
<li><a href="#Library_std_shared_ptr">shared_ptr smart pointer</a>
<ul>
<li><a href="#Library_shared_ptr_basics">shared_ptr basics</a></li>
<li><a href="#Library_shared_ptr_inheritance">shared_ptr and inheritance</a>
</li>
<li><a href="#Library_shared_ptr_overloading">shared_ptr and method
overloading</a></li>
<li><a href="#Library_shared_ptr_templates">shared_ptr and templates</a></li>
<li><a href="#Library_shared_ptr_directors">shared_ptr and directors</a></li>
</ul>
</li>
<li><a href="#Library_std_unique_ptr">unique_ptr smart pointer</a>
<ul>
<li><a href="#Library_std_unique_ptr_by_value">unique_ptr passed by
value</a></li>
<li><a href="#Library_std_unique_ptr_by_ref">unique_ptr passed by
reference</a></li>
</ul>
</li>
<li><a href="#Library_std_auto_ptr">auto_ptr smart pointer</a></li>
</ul>
</li>
<li><a href="#Library_nn16">Utility Libraries</a>
<ul>
<li><a href="#Library_nn17">exception.i</a></li>
<li><a href="#Library_attributes">attribute.i</a>
<ul>
<li><a href="#Library_attribute_templates">%attribute and C++ templates</a>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> To help build extension modules, SWIG is packaged with a library of
support files that you can include in your own interfaces. These files
often define new SWIG directives or provide utility functions that can
be used to access parts of the standard C and C++ libraries. This
chapter provides a reference to the current set of supported library
files.</p>
<p><b> Compatibility note:</b> Older versions of SWIG included a number
of library files for manipulating pointers, arrays, and other
structures. Most these files are now deprecated and have been removed
from the distribution. Alternative libraries provide similar
functionality. Please read this chapter carefully if you used the old
libraries.</p>
<h2><a name="Library_nn2">12.1 The %include directive and library search
path</a></h2>
<p> Library files are included using the <tt>%include</tt> directive.
When searching for files, directories are searched in the following
order:</p>
<ol>
<li>The current directory</li>
<li>Directories specified with the <tt>-I</tt> command line option</li>
<li>.<tt>/swig_lib</tt></li>
<li>SWIG library install location as reported by <tt>swig -swiglib</tt>,
for example <tt>/usr/local/share/swig/1.3.30</tt></li>
<li>On Windows, a directory <tt>Lib</tt> relative to the location of <tt>
swig.exe</tt> is also searched.</li>
</ol>
<p> Within directories mentioned in points 3-5, SWIG first looks for a
subdirectory corresponding to a target language (e.g., <tt>python</tt>,
<tt>tcl</tt>, etc.). If found, SWIG will search the language specific
directory first. This allows for language-specific implementations of
library files.</p>
<p> You can ignore the installed SWIG library by setting the <tt>
SWIG_LIB</tt> environment variable. Set the environment variable to hold
an alternative library directory.</p>
<p> The directories that are searched are displayed when using <tt>
-verbose</tt> commandline option.</p>
<h2><a name="Library_nn3">12.2 C arrays and pointers</a></h2>
<p> This section describes library modules for manipulating low-level C
arrays and pointers. The primary use of these modules is in supporting
C declarations that manipulate bare pointers such as <tt>int *</tt>, <tt>
double *</tt>, or <tt>void *</tt>. The modules can be used to allocate
memory, manufacture pointers, dereference memory, and wrap pointers as
class-like objects. Since these functions provide direct access to
memory, their use is potentially unsafe and you should exercise
caution.</p>
<h3><a name="Library_argcargv">12.2.1 argcargv.i</a></h3>
<p> The argcargv.i library is a simple library providing multi-argument
typemaps for handling C argc argv command line argument C string
arrays. The <tt>argc</tt> parameter contains the argument count and <tt>
argv</tt> contains the argument vector array.</p>
<p> This library provides the following multi-argument typemap:</p>
<p><b> <tt>(int ARGC, char **ARGV)</tt></b></p>
<p> Apply this multi-argument typemap to your use case, for example:</p>
<div class="code">
<pre>
%apply (int ARGC, char **ARGV) { (size_t argc, const char **argv) }
int mainApp(size_t argc, const char **argv);
</pre>
</div>
<p> then from Ruby:</p>
<div class="targetlang">
<pre>
$args = [&quot;myarg1&quot;, &quot;myarg2&quot;]
mainApp(args);
</pre>
</div>
<h3><a name="Library_nn4">12.2.2 cpointer.i</a></h3>
<p> The <tt>cpointer.i</tt> module defines macros that can be used to
generate wrappers around simple C pointers. The primary use of this
module is in generating pointers to primitive datatypes such as <tt>int</tt>
and <tt>double</tt>.</p>
<p><b> <tt>%pointer_functions(type, name)</tt></b></p>
<div class="indent">
<p>Generates a collection of four functions for manipulating a pointer <tt>
type *</tt>:</p>
<p> <tt>type *new_name()</tt></p>
<div class="indent">
<p> Creates a new object of type <tt>type</tt> and returns a pointer to
it. In C, the object is created using <tt>calloc()</tt>. In C++, <tt>
new</tt> is used.</p>
</div>
<p> <tt>type *copy_name(type value)</tt></p>
<div class="indent">
<p> Creates a new object of type <tt>type</tt> and returns a pointer to
it. An initial value is set by copying it from <tt>value</tt>. In C,
the object is created using <tt>calloc()</tt>. In C++, <tt>new</tt> is
used.</p>
</div>
<p> <tt>type *delete_name(type *obj)</tt></p>
<div class="indent">
<p> Deletes an object type <tt>type</tt>.</p>
</div>
<p> <tt>void name_assign(type *obj, type value)</tt></p>
<div class="indent">
<p> Assigns <tt>*obj = value</tt>.</p>
</div>
<p> <tt>type name_value(type *obj)</tt></p>
<div class="indent">
<p> Returns the value of <tt>*obj</tt>.</p>
</div>
<p> When using this macro, <tt>type</tt> may be any type and <tt>name</tt>
must be a legal identifier in the target language. <tt>name</tt> should
not correspond to any other name used in the interface file.</p>
<p> Here is a simple example of using <tt>%pointer_functions()</tt>:</p>
<div class="code">
<pre>
%module example
%include &quot;cpointer.i&quot;
/* Create some functions for working with &quot;int *&quot; */
%pointer_functions(int, intp);
/* A function that uses an &quot;int *&quot; */
void add(int x, int y, int *result);
</pre>
</div>
<p> Now, in Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
&gt;&gt;&gt; c = example.new_intp() # Create an &quot;int&quot; for storing result
&gt;&gt;&gt; example.add(3, 4, c) # Call function
&gt;&gt;&gt; example.intp_value(c) # Dereference
7
&gt;&gt;&gt; example.delete_intp(c) # Delete
</pre>
</div></div>
<p><b> <tt>%pointer_class(type, name)</tt></b></p>
<div class="indent">
<p> Wraps a pointer of <tt>type *</tt> inside a class-based interface.
This interface is as follows:</p>
<div class="code">
<pre>
struct name {
name(); // Create pointer object
~name(); // Delete pointer object
void assign(type value); // Assign value
type value(); // Get value
type *cast(); // Cast the pointer to original type
static name *frompointer(type *); // Create class wrapper from existing
// pointer
};
</pre>
</div>
<p> When using this macro, <tt>type</tt> is restricted to a simple type
name like <tt>int</tt>, <tt>float</tt>, or <tt>Foo</tt>. Pointers and
other complicated types are not allowed. <tt>name</tt> must be a valid
identifier not already in use. When a pointer is wrapped as a class,
the &quot;class&quot; may be transparently passed to any function that expects
the pointer.</p>
<p> If the target language does not support proxy classes, the use of
this macro will produce the example same functions as <tt>
%pointer_functions()</tt> macro.</p>
<p> It should be noted that the class interface does introduce a new
object or wrap a pointer inside a special structure. Instead, the raw
pointer is used directly.</p>
<p> Here is the same example using a class instead:</p>
<div class="code">
<pre>
%module example
%include &quot;cpointer.i&quot;
/* Wrap a class interface around an &quot;int *&quot; */
%pointer_class(int, intp);
/* A function that uses an &quot;int *&quot; */
void add(int x, int y, int *result);
</pre>
</div>
<p> Now, in Python (using proxy classes)</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
&gt;&gt;&gt; c = example.intp() # Create an &quot;int&quot; for storing result
&gt;&gt;&gt; example.add(3, 4, c) # Call function
&gt;&gt;&gt; c.value() # Dereference
7
</pre>
</div>
<p> Of the two macros, <tt>%pointer_class</tt> is probably the most
convenient when working with simple pointers. This is because the
pointers are access like objects and they can be easily garbage
collected (destruction of the pointer object destroys the underlying
object).</p>
</div>
<p><b> <tt>%pointer_cast(type1, type2, name)</tt></b></p>
<div class="indent">
<p> Creates a casting function that converts <tt>type1</tt> to <tt>type2</tt>
. The name of the function is <tt>name</tt>. For example:</p>
<div class="code">
<pre>
%pointer_cast(int *, unsigned int *, int_to_uint);
</pre>
</div>
<p> In this example, the function <tt>int_to_uint()</tt> would be used
to cast types in the target language.</p>
</div>
<p><b> Note:</b> None of these macros can be used to safely work with
strings (<tt>char *</tt> or <tt>char **</tt>).</p>
<p><b> Note:</b> When working with simple pointers, typemaps can often
be used to provide more seamless operation.</p>
<h3><a name="Library_carrays">12.2.3 carrays.i</a></h3>
<p> This module defines macros that assist in wrapping ordinary C
pointers as arrays. The module does not provide any safety or an extra
layer of wrapping--it merely provides functionality for creating,
destroying, and modifying the contents of raw C array data.</p>
<p><b> <tt>%array_functions(type, name)</tt></b></p>
<div class="indent">
<p>Creates four functions.</p>
<p> <tt>type *new_name(size_t nelements)</tt></p>
<div class="indent">
<p> Creates a new array of objects of type <tt>type</tt>. In C, the
array is allocated using <tt>calloc()</tt>. In C++, <tt>new []</tt> is
used.</p>
</div>
<p> <tt>type *delete_name(type *ary)</tt></p>
<div class="indent">
<p> Deletes an array. In C, <tt>free()</tt> is used. In C++, <tt>delete
[]</tt> is used.</p>
</div>
<p> <tt>type name_getitem(type *ary, size_t index)</tt></p>
<div class="indent">
<p> Returns the value <tt>ary[index]</tt>.</p>
</div>
<p> <tt>void name_setitem(type *ary, size_t index, type value)</tt></p>
<div class="indent">
<p> Assigns <tt>ary[index] = value</tt>.</p>
</div>
<p> When using this macro, <tt>type</tt> may be any type and <tt>name</tt>
must be a legal identifier in the target language. <tt>name</tt> should
not correspond to any other name used in the interface file.</p>
<p> Here is an example of <tt>%array_functions()</tt>. Suppose you had a
function like this:</p>
<div class="code">
<pre>
void print_array(double x[10]) {
int i;
for (i = 0; i &lt; 10; i++) {
printf(&quot;[%d] = %g\n&quot;, i, x[i]);
}
}
</pre>
</div>
<p> To wrap it, you might write this:</p>
<div class="code">
<pre>
%module example
%include &quot;carrays.i&quot;
%array_functions(double, doubleArray);
void print_array(double x[10]);
</pre>
</div>
<p> Now, in a scripting language, you might write this:</p>
<div class="targetlang">
<pre>
a = new_doubleArray(10) # Create an array
for i in range(0, 10):
doubleArray_setitem(a, i, 2 * i) # Set a value
print_array(a) # Pass to C
delete_doubleArray(a) # Destroy array
</pre>
</div></div>
<p><b> <tt>%array_class(type, name)</tt></b></p>
<div class="indent">
<p> Wraps a pointer of <tt>type *</tt> inside a class-based interface.
This interface is as follows:</p>
<div class="code">
<pre>
struct name {
name(size_t nelements); // Create an array
~name(); // Delete array
type getitem(size_t index); // Return item
void setitem(size_t index, type value); // Set item
type *cast(); // Cast to original type
static name *frompointer(type *); // Create class wrapper from
// existing pointer
};
</pre>
</div>
<p> When using this macro, <tt>type</tt> is restricted to a simple type
name like <tt>int</tt> or <tt>float</tt>. Pointers and other
complicated types are not allowed. <tt>name</tt> must be a valid
identifier not already in use. When a pointer is wrapped as a class, it
can be transparently passed to any function that expects the pointer.</p>
<p> When combined with proxy classes, the <tt>%array_class()</tt> macro
can be especially useful. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;carrays.i&quot;
%array_class(double, doubleArray);
void print_array(double x[10]);
</pre>
</div>
<p> Allows you to do this:</p>
<div class="targetlang">
<pre>
import example
c = example.doubleArray(10) # Create double[10]
for i in range(0, 10):
c[i] = 2 * i # Assign values
example.print_array(c) # Pass to C
</pre>
</div></div>
<p><b> Note:</b> These macros do not encapsulate C arrays inside a
special data structure or proxy. There is no bounds checking or safety
of any kind. If you want this, you should consider using a special
array object rather than a bare pointer.</p>
<p><b> Note:</b> <tt>%array_functions()</tt> and <tt>%array_class()</tt>
should not be used with types of <tt>char</tt> or <tt>char *</tt>.
SWIG's default handling of these types is to handle them as character
strings and the two macros do not do enough to change this.</p>
<h3><a name="Library_nn6">12.2.4 cmalloc.i</a></h3>
<p> This module defines macros for wrapping the low-level C memory
allocation functions <tt>malloc()</tt>, <tt>calloc()</tt>, <tt>
realloc()</tt>, and <tt>free()</tt>.</p>
<p><b> <tt>%malloc(type [, name=type])</tt></b></p>
<div class="indent">
<p> Creates a wrapper around <tt>malloc()</tt> with the following
prototype:</p>
<div class="code">
<pre>
<em>type</em> *malloc_<em>name</em>(int nbytes = sizeof(<em>type</em>));
</pre>
</div>
<p> If <tt>type</tt> is <tt>void</tt>, then the size parameter <tt>
nbytes</tt> is required. The <tt>name</tt> parameter only needs to be
specified when wrapping a type that is not a valid identifier (e.g., &quot;<tt>
int *</tt>&quot;, &quot;<tt>double **</tt>&quot;, etc.).</p>
</div>
<p><b> <tt>%calloc(type [, name=type])</tt></b></p>
<div class="indent">
<p> Creates a wrapper around <tt>calloc()</tt> with the following
prototype:</p>
<div class="code">
<pre>
<em>type</em> *calloc_<em>name</em>(int nobj =1, int sz = sizeof(<em>type</em>));
</pre>
</div>
<p> If <tt>type</tt> is <tt>void</tt>, then the size parameter <tt>sz</tt>
is required.</p>
</div>
<p><b> <tt>%realloc(type [, name=type])</tt></b></p>
<div class="indent">
<p> Creates a wrapper around <tt>realloc()</tt> with the following
prototype:</p>
<div class="code">
<pre>
<em>type</em> *realloc_<em>name</em>(<em>type</em> *ptr, int nitems);
</pre>
</div>
<p> Note: unlike the C <tt>realloc()</tt>, the wrapper generated by this
macro implicitly includes the size of the corresponding type. For
example, <tt>realloc_int(p, 100)</tt> reallocates <tt>p</tt> so that it
holds 100 integers.</p>
</div>
<p><b> <tt>%free(type [, name=type])</tt></b></p>
<div class="indent">
<p> Creates a wrapper around <tt>free()</tt> with the following
prototype:</p>
<div class="code">
<pre>
void free_<em>name</em>(<em>type</em> *ptr);
</pre>
</div></div>
<p><b> <tt>%sizeof(type [, name=type])</tt></b></p>
<div class="indent">
<p> Creates the constant:</p>
<div class="code">
<pre>
%constant int sizeof_<em>name</em> = sizeof(<em>type</em>);
</pre>
</div></div>
<p><b> <tt>%allocators(type [, name=type])</tt></b></p>
<div class="indent">
<p> Generates wrappers for all five of the above operations.</p>
</div>
<p> Here is a simple example that illustrates the use of these macros:</p>
<div class="code">
<pre>
// SWIG interface
%module example
%include &quot;cmalloc.i&quot;
%malloc(int);
%free(int);
%malloc(int *, intp);
%free(int *, intp);
%allocators(double);
</pre>
</div>
<p> Now, in a script:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; from example import *
&gt;&gt;&gt; a = malloc_int()
&gt;&gt;&gt; a
'_000efa70_p_int'
&gt;&gt;&gt; free_int(a)
&gt;&gt;&gt; b = malloc_intp()
&gt;&gt;&gt; b
'_000efb20_p_p_int'
&gt;&gt;&gt; free_intp(b)
&gt;&gt;&gt; c = calloc_double(50)
&gt;&gt;&gt; c
'_000fab98_p_double'
&gt;&gt;&gt; c = realloc_double(100000)
&gt;&gt;&gt; free_double(c)
&gt;&gt;&gt; print sizeof_double
8
&gt;&gt;&gt;
</pre>
</div>
<h3><a name="Library_nn7">12.2.5 cdata.i</a></h3>
<p> The <tt>cdata.i</tt> module defines functions for converting raw C
data to and from a target language type. The target language type is
either:</p>
<ul>
<li> A string for the scripting languages. The scripting language must
support strings with embedded binary data in order for this to work.</li>
<li> A byte array for the statically typed languages; namely, <tt>byte[]</tt>
for C# and Java, <tt>ubyte[]</tt> for D, <tt>[]byte</tt> for Go.</li>
</ul>
<p> The primary applications of this module would be packing/unpacking
of binary data structures---for instance, if you needed to extract data
from a buffer.</p>
<p> The available APIs are:</p>
<p><b> <tt>const char *cdata(void *ptr, size_t nbytes)</tt></b></p>
<div class="indent">
<p> Converts <tt>nbytes</tt> of data at <tt>ptr</tt> into the target
language type. <tt>ptr</tt> can be any pointer.</p>
</div>
<p><b> <tt>void memmove(void *ptr, const char *s)</tt></b></p>
<div class="indent">
<p> This is actually a wrapper of the standard C library <tt>memmove</tt>
function,<b> <tt>void *memmove(void *ptr, const void *src, size_t n)</tt>
</b>, which copies <tt>n</tt> bytes of data from <tt>src</tt> into the
destination pointed to by <tt>ptr</tt>. Multi-argument typemaps are
used so that the last two parameters, <tt>src</tt> and <tt>n</tt> are
replaced by <tt>s</tt>, a string or byte array target language type,
mentioned earlier. The string/byte array may contain embedded NULL
bytes. Unlike the C library function nothing is returned by the wrapper
function.</p>
</div>
<p> One use of these functions is packing and unpacking data from
memory. Here is a short example:</p>
<div class="code">
<pre>
// SWIG interface
%module example
%include &quot;carrays.i&quot;
%include &quot;cdata.i&quot;
%array_class(int, intArray);
</pre>
</div>
<p> Python example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = intArray(10)
&gt;&gt;&gt; for i in range(0, 10):
... a[i] = i
&gt;&gt;&gt; b = cdata(a, 40)
&gt;&gt;&gt; b
'\x00\x00\x00\x00\x00\x00\x00\x01\x00\x00\x00\x02\x00\x00\x00\x03\x00\x00\x00\x04
\x00\x00\x00\x05\x00\x00\x00\x06\x00\x00\x00\x07\x00\x00\x00\x08\x00\x00\x00\t'
&gt;&gt;&gt; c = intArray(10)
&gt;&gt;&gt; memmove(c, b)
&gt;&gt;&gt; print c[4]
4
&gt;&gt;&gt;
</pre>
</div>
<p> Since the size of data is not always known, the following macro is
also defined:</p>
<p><b> <tt>%cdata(type [, name=type])</tt></b></p>
<div class="indent">
<p> Generates the following function for extracting C data for a given
type.</p>
<div class="code">
<pre>
char *cdata_<em>name</em>(type* ptr, int nitems)
</pre>
</div>
<p> <tt>nitems</tt> is the number of items of the given type to extract.</p>
</div>
<p><b> Note:</b> These functions provide direct access to memory and can
be used to overwrite data. Clearly they are unsafe.</p>
<h2><a name="Library_nn8">12.3 C string handling</a></h2>
<p> A common problem when working with C programs is dealing with
functions that manipulate raw character data using <tt>char *</tt>. In
part, problems arise because there are different interpretations of <tt>
char *</tt>---it could be a NULL-terminated string or it could point to
binary data. Moreover, functions that manipulate raw strings may mutate
data, perform implicit memory allocations, or utilize fixed-sized
buffers.</p>
<p> The problems (and perils) of using <tt>char *</tt> are well-known.
However, SWIG is not in the business of enforcing morality. The modules
in this section provide basic functionality for manipulating raw C
strings.</p>
<h3><a name="Library_nn9">12.3.1 Default string handling</a></h3>
<p> Suppose you have a C function with this prototype:</p>
<div class="code">
<pre>
char *foo(char *s);
</pre>
</div>
<p> The default wrapping behavior for this function is to set <tt>s</tt>
to a raw <tt>char *</tt> that refers to the internal string data in the
target language. In other words, if you were using a language like Tcl,
and you wrote this,</p>
<div class="targetlang">
<pre>
% foo Hello
</pre>
</div>
<p> then <tt>s</tt> would point to the representation of &quot;Hello&quot; inside
the Tcl interpreter. When returning a <tt>char *</tt>, SWIG assumes
that it is a NULL-terminated string and makes a copy of it. This gives
the target language its own copy of the result.</p>
<p> There are obvious problems with the default behavior. First, since a
<tt>char *</tt> argument points to data inside the target language, it
is<b> NOT</b> safe for a function to modify this data (doing so may
corrupt the interpreter and lead to a crash). Furthermore, the default
behavior does not work well with binary data. Instead, strings are
assumed to be NULL-terminated.</p>
<h3><a name="Library_nn10">12.3.2 Passing a string with length</a></h3>
<p> If you have a function that expects string with a length,</p>
<div class="code">
<pre>
size_t parity(char *str, size_t len, size_t initial);
</pre>
</div>
<p> you can wrap the parameters <tt>(char *str, size_t len)</tt> as a
single argument using a typemap. Just do this:</p>
<div class="code">
<pre>
%apply (char *STRING, size_t LENGTH) { (char *str, size_t len) };
...
size_t parity(char *str, size_t len, size_t initial);
</pre>
</div>
<p> Now, in the target language, you can use the string like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; s = &quot;H\x00\x15eg\x09\x20&quot;
&gt;&gt;&gt; parity(s, 0)
</pre>
</div>
<p> In the wrapper function, the passed string will be expanded to a
pointer and length parameter. The <tt>(char *STRING, int LENGTH)</tt>
multi-argument typemap is also available in addition to <tt>(char
*STRING, size_t LENGTH)</tt>.</p>
<h3><a name="Library_nn11">12.3.3 Using %newobject to release memory</a></h3>
<p> If you have a function that allocates memory like this,</p>
<div class="code">
<pre>
char *foo() {
char *result = (char *) malloc(...);
...
return result;
}
</pre>
</div>
<p> then the SWIG generated wrappers will have a memory leak--the
returned data will be copied into a string object and the old contents
ignored.</p>
<p> To fix the memory leak, use the <tt>%newobject</tt> directive.</p>
<div class="code">
<pre>
%newobject foo;
...
char *foo();
</pre>
</div>
<p> This will release the result if the appropriate target language
support is available. SWIG provides the appropriate &quot;newfree&quot; typemap
for <tt>char *</tt> so that the memory is released, however, you may
need to provide your own &quot;newfree&quot; typemap for other types. See <a href="#Customization_ownership">
Object ownership and %newobject</a> for more details.</p>
<h3><a name="Library_nn12">12.3.4 cstring.i</a></h3>
<p> The <tt>cstring.i</tt> library file provides a collection of macros
for dealing with functions that either mutate string arguments or which
try to output string data through their arguments. An example of such a
function might be this rather questionable implementation:</p>
<div class="code">
<pre>
void get_path(char *s) {
// Potential buffer overflow---uh, oh.
sprintf(s, &quot;%s/%s&quot;, base_directory, sub_directory);
}
...
// Somewhere else in the C program
{
char path[1024];
...
get_path(path);
...
}
</pre>
</div>
<p> (Off topic rant: If your program really has functions like this, you
would be well-advised to replace them with safer alternatives involving
bounds checking).</p>
<p> The macros defined in this module all expand to various combinations
of typemaps. Therefore, the same pattern matching rules and ideas
apply.</p>
<p><b> %cstring_bounded_output(parm, maxsize)</b></p>
<div class="indent">
<p> Turns parameter <tt><em>parm</em></tt> into an output value. The
output string is assumed to be NULL-terminated and smaller than <tt><em>
maxsize</em></tt> characters. Here is an example:</p>
<div class="code">
<pre>
%cstring_bounded_output(char *path, 1024);
...
void get_path(char *path);
</pre>
</div>
<p> In the target language:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; get_path()
/home/beazley/packages/Foo/Bar
&gt;&gt;&gt;
</pre>
</div>
<p> Internally, the wrapper function allocates a small buffer (on the
stack) of the requested size and passes it as the pointer value. Data
stored in the buffer is then returned as a function return value. If
the function already returns a value, then the return value and the
output string are returned together (multiple return values).<b> If
more than <tt><em>maxsize</em></tt> bytes are written, your program
will crash with a buffer overflow!</b></p>
</div>
<p><b> %cstring_chunk_output(parm, chunksize)</b></p>
<div class="indent">
<p> Turns parameter <tt><em>parm</em></tt> into an output value. The
output string is always <tt><em>chunksize</em></tt> and may contain
binary data. Here is an example:</p>
<div class="code">
<pre>
%cstring_chunk_output(char *packet, PACKETSIZE);
...
void get_packet(char *packet);
</pre>
</div>
<p> In the target language:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; get_packet()
'\xa9Y:\xf6\xd7\xe1\x87\xdbH;y\x97\x7f\xd3\x99\x14V\xec\x06\xea\xa2\x88'
&gt;&gt;&gt;
</pre>
</div>
<p> This macro is essentially identical to <tt>%cstring_bounded_output</tt>
. The only difference is that the result is always <tt><em>chunksize</em>
</tt> characters. Furthermore, the result can contain binary data.<b> If
more than <tt><em>maxsize</em></tt> bytes are written, your program
will crash with a buffer overflow!</b></p>
</div>
<p><b> %cstring_bounded_mutable(parm, maxsize)</b></p>
<div class="indent">
<p> Turns parameter <tt><em>parm</em></tt> into a mutable string
argument. The input string is assumed to be NULL-terminated and smaller
than <tt><em>maxsize</em></tt> characters. The output string is also
assumed to be NULL-terminated and less than <tt><em>maxsize</em></tt>
characters.</p>
<div class="code">
<pre>
%cstring_bounded_mutable(char *ustr, 1024);
...
void make_upper(char *ustr);
</pre>
</div>
<p> In the target language:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; make_upper(&quot;hello world&quot;)
'HELLO WORLD'
&gt;&gt;&gt;
</pre>
</div>
<p> Internally, this macro is almost exactly the same as <tt>
%cstring_bounded_output</tt>. The only difference is that the parameter
accepts an input value that is used to initialize the internal buffer.
It is important to emphasize that this function does not mutate the
string value passed---instead it makes a copy of the input value,
mutates it, and returns it as a result.<b> If more than <tt><em>maxsize</em>
</tt> bytes are written, your program will crash with a buffer overflow!</b>
</p>
</div>
<p><b> %cstring_mutable(parm [, expansion])</b></p>
<div class="indent">
<p> Turns parameter <tt><em>parm</em></tt> into a mutable string
argument. The input string is assumed to be NULL-terminated. An
optional parameter <tt><em>expansion</em></tt> specifies the number of
extra characters by which the string might grow when it is modified.
The output string is assumed to be NULL-terminated and less than the
size of the input string plus any expansion characters.</p>
<div class="code">
<pre>
%cstring_mutable(char *ustr);
...
void make_upper(char *ustr);
%cstring_mutable(char *hstr, HEADER_SIZE);
...
void attach_header(char *hstr);
</pre>
</div>
<p> In the target language:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; make_upper(&quot;hello world&quot;)
'HELLO WORLD'
&gt;&gt;&gt; attach_header(&quot;Hello world&quot;)
'header: Hello world'
&gt;&gt;&gt;
</pre>
</div>
<p> This macro differs from <tt>%cstring_bounded_mutable()</tt> in that
a buffer is dynamically allocated (on the heap using <tt>malloc/new</tt>
). This buffer is always large enough to store a copy of the input value
plus any expansion bytes that might have been requested. It is
important to emphasize that this function does not directly mutate the
string value passed---instead it makes a copy of the input value,
mutates it, and returns it as a result.<b> If the function expands the
result by more than <tt><em>expansion</em></tt> extra bytes, then the
program will crash with a buffer overflow!</b></p>
</div>
<p><b> %cstring_output_maxsize(parm, maxparm)</b></p>
<div class="indent">
<p> This macro is used to handle bounded character output functions
where both a <tt>char *</tt> and a maximum length parameter are
provided. As input, a user simply supplies the maximum length. The
return value is assumed to be a NULL-terminated string.</p>
<div class="code">
<pre>
%cstring_output_maxsize(char *path, int maxpath);
...
void get_path(char *path, int maxpath);
</pre>
</div>
<p> In the target language:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; get_path(1024)
'/home/beazley/Packages/Foo/Bar'
&gt;&gt;&gt;
</pre>
</div>
<p> This macro provides a safer alternative for functions that need to
write string data into a buffer. User supplied buffer size is used to
dynamically allocate memory on heap. Results are placed into that
buffer and returned as a string object.</p>
</div>
<p><b> %cstring_output_withsize(parm, maxparm)</b></p>
<div class="indent">
<p> This macro is used to handle bounded character output functions
where both a <tt>char *</tt> and a pointer <tt>int *</tt> are passed.
Initially, the <tt>int *</tt> parameter points to a value containing
the maximum size. On return, this value is assumed to contain the
actual number of bytes. As input, a user simply supplies the maximum
length. The output value is a string that may contain binary data.</p>
<div class="code">
<pre>
%cstring_output_withsize(char *data, int *maxdata);
...
void get_data(char *data, int *maxdata);
</pre>
</div>
<p> In the target language:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; get_data(1024)
'x627388912'
&gt;&gt;&gt; get_data(1024)
'xyzzy'
&gt;&gt;&gt;
</pre>
</div>
<p> This macro is a somewhat more powerful version of <tt>
%cstring_output_chunk()</tt>. Memory is dynamically allocated and can be
arbitrary large. Furthermore, a function can control how much data is
actually returned by changing the value of the <tt>maxparm</tt>
argument.</p>
</div>
<p><b> %cstring_output_allocate(parm, release)</b></p>
<div class="indent">
<p> This macro is used to return strings that are allocated within the
program and returned in a parameter of type <tt>char **</tt>. For
example:</p>
<div class="code">
<pre>
void foo(char **s) {
*s = (char *) malloc(64);
sprintf(*s, &quot;Hello world\n&quot;);
}
</pre>
</div>
<p> The returned string is assumed to be NULL-terminated. <tt><em>
release</em></tt> specifies how the allocated memory is to be released
(if applicable). Here is an example:</p>
<div class="code">
<pre>
%cstring_output_allocate(char **s, free(*$1));
...
void foo(char **s);
</pre>
</div>
<p> In the target language:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; foo()
'Hello world\n'
&gt;&gt;&gt;
</pre>
</div></div>
<p><b> %cstring_output_allocate_size(parm, szparm, release)</b></p>
<div class="indent">
<p> This macro is used to return strings that are allocated within the
program and returned in two parameters of type <tt>char **</tt> and <tt>
int *</tt>. For example:</p>
<div class="code">
<pre>
void foo(char **s, int *sz) {
*s = (char *) malloc(64);
*sz = 64;
// Write some binary data
...
}
</pre>
</div>
<p> The returned string may contain binary data. <tt><em>release</em></tt>
specifies how the allocated memory is to be released (if applicable).
Here is an example:</p>
<div class="code">
<pre>
%cstring_output_allocate_size(char **s, int *slen, free(*$1));
...
void foo(char **s, int *slen);
</pre>
</div>
<p> In the target language:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; foo()
'\xa9Y:\xf6\xd7\xe1\x87\xdbH;y\x97\x7f\xd3\x99\x14V\xec\x06\xea\xa2\x88'
&gt;&gt;&gt;
</pre>
</div>
<p> This is the safest and most reliable way to return binary string
data in SWIG. If you have functions that conform to another prototype,
you might consider wrapping them with a helper function. For example,
if you had this:</p>
<div class="code">
<pre>
char *get_data(int *len);
</pre>
</div>
<p> You could wrap it with a function like this:</p>
<div class="code">
<pre>
void my_get_data(char **result, int *len) {
*result = get_data(len);
}
</pre>
</div></div>
<p><b> Comments:</b></p>
<ul>
<li>Support for the <tt>cstring.i</tt> module depends on the target
language. Not all SWIG modules currently support this library.</li>
<li>Reliable handling of raw C strings is a delicate topic. There are
many ways to accomplish this in SWIG. This library provides support for
a few common techniques.</li>
<li>If used in C++, this library uses <tt>new</tt> and <tt>delete []</tt>
for memory allocation. If using C, the library uses <tt>malloc()</tt>
and <tt>free()</tt>.</li>
<li>Rather than manipulating <tt>char *</tt> directly, you might
consider using a special string structure or class instead.</li>
</ul>
<h2><a name="Library_c_standard_library">12.4 C standard library</a></h2>
<h3><a name="Library_complex">12.4.1 Complex floating types</a></h3>
<p> SWIG has some support for complex floating types. By default the
keyword <tt>_Complex</tt> is understood by the parser but <tt>complex</tt>
is not treated as a keyword because it may be used as an identifier.</p>
<p> SWIG is only able to fully wrap complex floating types for some
target languages. If you are using such a target language for complex
floating types, you should include <tt>complex.i</tt> to enable this
support:</p>
<div class="code">
<pre>
%include &quot;complex.i&quot;
</pre>
</div>
<p> If SWIG is in C++ code this will actually include <tt>std_complex.i</tt>
for you; otherwise it will include <tt>ccomplex.i</tt>.</p>
<p> For target languages without support this will fail to find a file
to include. In this case, if you are wrapping headers which use <tt>
complex</tt> then you'll need to add a define to your SWIG interface
file to get SWIG to parse the headers:</p>
<div class="code">
<pre>
#define complex _Complex
</pre>
</div>
<p> Then if wrapping as opaque types is not useful to you, you can use <a
href="#SWIG_ignore"><tt>%ignore</tt></a> to tell SWIG not to wrap the
functions and/or variables which use complex floating types.</p>
<h2><a name="Library_stl_cpp_library">12.5 STL/C++ library</a></h2>
<p> The library modules in this section provide access to parts of the
standard C++ library including the STL. SWIG support for the STL is an
ongoing effort. Support is quite comprehensive for some language
modules but some of the lesser used modules do not have quite as much
library code written.</p>
<p> The following table shows which C++ classes are supported and the
equivalent SWIG interface library file for the C++ library.</p>
<table BORDER summary="SWIG C++ library files">
<tr VALIGN="TOP"><td><b>C++ class</b></td><td><b>C++ Library file</b></td><td>
<b>SWIG Interface library file</b></td></tr>
<tr><td>std::array (C++11)</td><td>array</td><td>std_array.i</td></tr>
<tr><td>std::auto_ptr</td><td>memory</td><td>std_auto_ptr.i</td></tr>
<tr><td>std::complex</td><td>complex</td><td>std_complex.i</td></tr>
<tr><td>std::deque</td><td>deque</td><td>std_deque.i</td></tr>
<tr><td>std::list</td><td>list</td><td>std_list.i</td></tr>
<tr><td>std::map</td><td>map</td><td>std_map.i</td></tr>
<tr><td>std::multimap (C++11)</td><td>multimap</td><td>std_multimap.i</td>
</tr>
<tr><td>std::multiset (C++11)</td><td>multiset</td><td>std_multiset.i</td>
</tr>
<tr><td>std::pair</td><td>utility</td><td>std_pair.i</td></tr>
<tr><td>std::set</td><td>set</td><td>std_set.i</td></tr>
<tr><td>std::shared_ptr (C++11)</td><td>shared_ptr</td><td>
std_shared_ptr.i</td></tr>
<tr><td>std::string</td><td>string</td><td>std_string.i</td></tr>
<tr><td>std::string_view (C++17)</td><td>string_view</td><td>
std_string_view.i</td></tr>
<tr><td>std::unordered_map (C++11)</td><td>unordered_map</td><td>
std_unordered_map.i</td></tr>
<tr><td>std::unordered_multimap (C++11)</td><td>unordered_multimap</td><td>
std_unordered_multimap.i</td></tr>
<tr><td>std::unordered_multiset (C++11)</td><td>unordered_multiset</td><td>
std_unordered_multiset.i</td></tr>
<tr><td>std::unordered_set (C++11)</td><td>unordered_set</td><td>
std_unordered_set.i</td></tr>
<tr><td>std::vector</td><td>vector</td><td>std_vector.i</td></tr>
<tr><td>std::wstring</td><td>wstring</td><td>std_wstring.i</td></tr>
</table>
<p> The list is by no means complete; some language modules support a
subset of the above and some support additional STL classes. Please
look for the library files in the appropriate language library
directory.</p>
<h3><a name="Library_std_string">12.5.1 std::string</a></h3>
<p> The <tt>std_string.i</tt> library provides typemaps for converting
C++ <tt>std::string</tt> objects to and from strings in the target
scripting language. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;std_string.i&quot;
std::string foo();
void bar(const std::string &amp;x);
</pre>
</div>
<p> In the target language:</p>
<div class="targetlang">
<pre>
x = foo(); # Returns a string object
bar(&quot;Hello World&quot;); # Pass string as std::string
</pre>
</div>
<p> A common problem that people encounter is that of classes/structures
containing a <tt>std::string</tt>. This can be overcome by defining a
typemap. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;std_string.i&quot;
%apply const std::string&amp; {std::string* foo};
struct my_struct
{
std::string foo;
};
</pre>
</div>
<p> In the target language:</p>
<div class="targetlang">
<pre>
x = my_struct();
x.foo = &quot;Hello World&quot;; # assign with string
print x.foo; # print as string
</pre>
</div>
<p> This module only supports types <tt>std::string</tt> and <tt>const
std::string &amp;</tt>. Pointers and non-const references are left
unmodified and returned as SWIG pointers.</p>
<p> This library file is fully aware of C++ namespaces. If you export <tt>
std::string</tt> or rename it with a typedef, make sure you include
those declarations in your interface. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;std_string.i&quot;
using namespace std;
typedef std::string String;
...
void foo(string s, const String &amp;t); // std_string typemaps still applied
</pre>
</div>
<h3><a name="Library_std_string_view">12.5.2 std::string_view</a></h3>
<p> The <tt>std_string_view.i</tt> library provides typemaps for
converting C++17 <tt>std::string_view</tt> objects to and from strings
in the target scripting language. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;std_string_view.i&quot;
std::string_view foo();
void bar(std::string_view x);
</pre>
</div>
<p> In the target language:</p>
<div class="targetlang">
<pre>
x = foo(); # Returns a string object
bar(&quot;Hello World&quot;); # Pass string as std::string_view
</pre>
</div>
<p> For target languages for which SWIG supports directors, <tt>
directorout</tt> typemaps are provided for <tt>std::string_view</tt>,
but these require extra care to use safely. The issue is that returning
<tt>std::string_view</tt> effectively returns a pointer to string data
but doesn't own the pointed to data. For target languages where there
isn't a native narrow string representation (e.g. C#, Java) a <tt>
static std::string</tt> is used to cache the data, which works but isn't
thread/reentrant safe. For target languages where there is a native
narrow string representation SWIG will return a <tt>std::string_view</tt>
pointing to that data, so you need to store the string to return
somewhere which will persist for the lifetime the caller needs (e.g.
put it in a member variable) - you can't return a temporary target
language string. In both cases SWIG will issue a warning by default.</p>
<h3><a name="Library_std_vector">12.5.3 std::vector</a></h3>
<p> The <tt>std_vector.i</tt> library provides support for the C++ <tt>
std::vector</tt> class in the STL. Using this library involves the use
of the <tt>%template</tt> directive. All you need to do is to
instantiate different versions of <tt>vector</tt> for the types that
you want to use. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;std_vector.i&quot;
namespace std {
%template(vectori) vector&lt;int&gt;;
%template(vectord) vector&lt;double&gt;;
};
</pre>
</div>
<p> When a template <tt>vector&lt;X&gt;</tt> is instantiated a number of
things happen:</p>
<ul>
<li>A class that exposes the C++ API is created in the target language .
This can be used to create objects, invoke methods, etc. This class is
currently a subset of the real STL vector class.</li>
<li>Input typemaps are defined for <tt>vector&lt;X&gt;</tt>, <tt>const
vector&lt;X&gt; &amp;</tt>, and <tt>const vector&lt;X&gt; *</tt>. For each of these, a
pointer <tt>vector&lt;X&gt; *</tt> may be passed or a native list object in
the target language.</li>
<li>An output typemap is defined for <tt>vector&lt;X&gt;</tt>. In this case,
the values in the vector are expanded into a list object in the target
language.</li>
<li>For all other variations of the type, the wrappers expect to receive
a <tt>vector&lt;X&gt; *</tt> object in the usual manner.</li>
<li>An exception handler for <tt>std::out_of_range</tt> is defined.</li>
<li>Optionally, special methods for indexing, item retrieval, slicing,
and element assignment may be defined. This depends on the target
language.</li>
</ul>
<p> To illustrate the use of this library, consider the following
functions:</p>
<div class="code">
<pre>
/* File : example.h */
#include &lt;vector&gt;
#include &lt;algorithm&gt;
#include &lt;functional&gt;
#include &lt;numeric&gt;
double average(std::vector&lt;int&gt; v) {
return std::accumulate(v.begin(), v.end(), 0.0)/v.size();
}
std::vector&lt;double&gt; half(const std::vector&lt;double&gt;&amp; v) {
std::vector&lt;double&gt; w(v);
for (unsigned int i=0; i&lt;w.size(); i++)
w[i] /= 2.0;
return w;
}
void halve_in_place(std::vector&lt;double&gt;&amp; v) {
for (std::vector&lt;double&gt;::iterator it = v.begin(); it != v.end(); ++it)
*it /= 2.0;
}
</pre>
</div>
<p> To wrap with SWIG, you might write the following:</p>
<div class="code">
<pre>
%module example
%{
#include &quot;example.h&quot;
%}
%include &quot;std_vector.i&quot;
// Instantiate templates used by example
namespace std {
%template(IntVector) vector&lt;int&gt;;
%template(DoubleVector) vector&lt;double&gt;;
}
// Include the header file with above prototypes
%include &quot;example.h&quot;
</pre>
</div>
<p> Now, to illustrate the behavior in the scripting interpreter,
consider this Python example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; from example import *
&gt;&gt;&gt; iv = IntVector(4) # Create an vector&lt;int&gt;
&gt;&gt;&gt; for i in range(0, 4):
... iv[i] = i
&gt;&gt;&gt; average(iv) # Call method
1.5
&gt;&gt;&gt; average([0, 1, 2, 3]) # Call with list
1.5
&gt;&gt;&gt; half([1, 2, 3]) # Half a list
(0.5, 1.0, 1.5)
&gt;&gt;&gt; halve_in_place([1, 2, 3]) # Oops
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
TypeError: Type error. Expected _p_std__vectorTdouble_t
&gt;&gt;&gt; dv = DoubleVector(4)
&gt;&gt;&gt; for i in range(0, 4):
... dv[i] = i
&gt;&gt;&gt; halve_in_place(dv) # Ok
&gt;&gt;&gt; for i in dv:
... print i
...
0.0
0.5
1.0
1.5
&gt;&gt;&gt; dv[20] = 4.5
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
File &quot;example.py&quot;, line 81, in __setitem__
def __setitem__(*args): return apply(examplec.DoubleVector___setitem__, args)
IndexError: vector index out of range
&gt;&gt;&gt;
</pre>
</div>
<p> This library module is fully aware of C++ namespaces. If you use
vectors with other names, make sure you include the appropriate <tt>
using</tt> or typedef directives. For example:</p>
<div class="code">
<pre>
%include &quot;std_vector.i&quot;
namespace std {
%template(IntVector) vector&lt;int&gt;;
}
using namespace std;
typedef std::vector Vector;
void foo(vector&lt;int&gt; *x, const Vector &amp;x);
</pre>
</div>
<p><b> Note:</b> This module makes use of several advanced SWIG features
including templatized typemaps and template partial specialization. If
you are trying to wrap other C++ code with templates, you might look at
the code contained in <tt>std_vector.i</tt>. Alternatively, you can
show them the code if you want to make their head explode.</p>
<p><b> Note:</b> This module is defined for all SWIG target languages.
However argument conversion details and the public API exposed to the
interpreter vary.</p>
<h3><a name="Library_stl_exceptions">12.5.4 STL exceptions</a></h3>
<p> Many of the STL wrapper functions add parameter checking and will
throw a language dependent error/exception should the values not be
valid. The classic example is array bounds checking. The library
wrappers are written to throw a C++ exception in the case of error. The
C++ exception in turn gets converted into an appropriate
error/exception for the target language. By and large this handling
should not need customising, however, customisation can easily be
achieved by supplying appropriate &quot;throws&quot; typemaps. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;std_vector.i&quot;
%typemap(throws) std::out_of_range {
// custom exception handler
}
%template(VectInt) std::vector&lt;int&gt;;
</pre>
</div>
<p> The custom exception handler might, for example, log the exception
then convert it into a specific error/exception for the target
language.</p>
<p> When using the STL it is advisable to add in an exception handler to
catch all STL exceptions. The <tt>%exception</tt> directive can be used
by placing the following code before any other methods or libraries to
be wrapped:</p>
<div class="code">
<pre>
%include &quot;exception.i&quot;
%exception {
try {
$action
} catch (const std::exception&amp; e) {
SWIG_exception(SWIG_RuntimeError, e.what());
}
}
</pre>
</div>
<p> Any thrown STL exceptions will then be gracefully handled instead of
causing a crash.</p>
<h3><a name="Library_std_shared_ptr">12.5.5 shared_ptr smart pointer</a></h3>
<h4><a name="Library_shared_ptr_basics">12.5.5.1 shared_ptr basics</a></h4>
<p> Some target languages have support for handling the shared_ptr
reference counted smart pointer. This smart pointer is available in the
standard C++11 library as <tt>std::shared_ptr</tt>. It was also in TR1
as <tt>std::tr1::shared_ptr</tt> before it was fully standardized.
Support for the widely used <tt>boost::shared_ptr</tt> is also
available.</p>
<p> In order to use <tt>std::shared_ptr</tt>, the <tt>std_shared_ptr.i</tt>
library file should be included:</p>
<div class="code">
<pre>
%include &lt;std_shared_ptr.i&gt;
</pre>
</div>
<p> The pre-standard <tt>std::tr1::shared_ptr</tt> can be used by
including the following macro before including the <tt>std_shared_ptr.i</tt>
library file:</p>
<div class="code">
<pre>
#define SWIG_SHARED_PTR_SUBNAMESPACE tr1
%include &lt;std_shared_ptr.i&gt;
</pre>
</div>
<p> In order to use <tt>boost::shared_ptr</tt>, the <tt>
boost_shared_ptr.i</tt> library file should be included:</p>
<div class="code">
<pre>
%include &lt;boost_shared_ptr.i&gt;
</pre>
</div>
<p> You can only use one of these variants of shared_ptr in your
interface file at a time (also SWIG doesn't currently support using
both <tt>%shared_ptr(T)</tt> and <tt>%unique_ptr&lt;T&gt;</tt> on the same
type <tt>T</tt>). All three variants must be used in conjunction with
the <tt>%shared_ptr(T)</tt> macro, where <tt>T</tt> is the underlying
pointer type equating to usage <tt>shared_ptr&lt;T&gt;</tt>. The type <tt>T</tt>
must be non-primitive. A simple example demonstrates usage:</p>
<div class="code">
<pre>
%module example
%include &lt;boost_shared_ptr.i&gt;
%shared_ptr(IntValue)
%inline %{
#include &lt;boost/shared_ptr.hpp&gt;
struct IntValue {
int value;
IntValue(int v) : value(v) {}
};
static int extractValue(const IntValue &amp;t) {
return t.value;
}
static int extractValueSmart(boost::shared_ptr&lt;IntValue&gt; t) {
return t-&gt;value;
}
%}
</pre>
</div>
<p> Note that the <tt>%shared_ptr(IntValue)</tt> declaration occurs
after the inclusion of the <tt>boost_shared_ptr.i</tt> library which
provides the macro and, very importantly, before any usage or
declaration of the type, <tt>IntValue</tt>. The <tt>%shared_ptr</tt>
macro provides, a few things for handling this smart pointer, but
mostly a number of typemaps. These typemaps override the default
typemaps so that the underlying proxy class is stored and passed around
as a pointer to a <tt>shared_ptr</tt> instead of a plain pointer to the
underlying type. This approach means that any instantiation of the type
can be passed to methods taking the type by value, reference, pointer
or as a smart pointer. The interested reader might want to look at the
generated code, however, usage is simple and no different handling is
required from the target language. For example, a simple use case of
the above code from Java would be:</p>
<div class="targetlang">
<pre>
IntValue iv = new IntValue(1234);
int val1 = example.extractValue(iv);
int val2 = example.extractValueSmart(iv);
System.out.println(val1 + &quot; &quot; + val2);
</pre>
</div>
<h4><a name="Library_shared_ptr_inheritance">12.5.5.2 shared_ptr and
inheritance</a></h4>
<p> The shared_ptr library works quite differently to SWIG's normal, but
somewhat limited, <a href="#SWIGPlus_smart_pointers">smart pointer
handling</a>. The shared_ptr library does not generate extra wrappers,
just for smart pointer handling, in addition to the proxy class. The
normal proxy class including inheritance relationships is generated as
usual. The only real change introduced by the <tt>%shared_ptr</tt>
macro is that the proxy class stores a pointer to the shared_ptr
instance instead of a raw pointer to the instance. A proxy class
derived from a base which is being wrapped with shared_ptr can and<b>
must</b> be wrapped as a shared_ptr too. In other words all classes in
an inheritance hierarchy must all be used with the <tt>%shared_ptr</tt>
macro. For example the following code can be used with the base class
shown earlier:</p>
<div class="code">
<pre>
%shared_ptr(DerivedIntValue)
%inline %{
struct DerivedIntValue : IntValue {
DerivedIntValue(int value) : IntValue(value) {}
...
};
%}
</pre>
</div>
<p> A shared_ptr of the derived class can now be passed to a method
where the base is expected in the target language, just as it can in
C++:</p>
<div class="targetlang">
<pre>
DerivedIntValue div = new DerivedIntValue(5678);
int val3 = example.extractValue(div);
int val4 = example.extractValueSmart(div);
</pre>
</div>
<p> If the <tt>%shared_ptr</tt> macro is omitted for any class in the
inheritance hierarchy, SWIG will warn about this and the generated code
may or may not result in a C++ compilation error. For example, the
following input:</p>
<div class="code">
<pre>
%include &quot;boost_shared_ptr.i&quot;
%shared_ptr(Parent);
%inline %{
#include &lt;boost/shared_ptr.hpp&gt;
struct GrandParent {
virtual ~GrandParent() {}
};
struct Parent : GrandParent {
virtual ~Parent() {}
};
struct Child : Parent {
virtual ~Child() {}
};
%}
</pre>
</div>
<p> warns about the missing smart pointer information:</p>
<div class="shell">
<pre>
example.i:12: Warning 520: Base class 'GrandParent' of 'Parent' is not similarly marked as a smart pointer.
example.i:16: Warning 520: Derived class 'Child' of 'Parent' is not similarly marked as a smart pointer.
</pre>
</div>
<p> Adding the missing <tt>%shared_ptr</tt> macros will fix this:</p>
<div class="code">
<pre>
%include &lt;boost_shared_ptr.i&gt;
%shared_ptr(GrandParent);
%shared_ptr(Parent);
%shared_ptr(Child);
... as before ...
</pre>
</div>
<h4><a name="Library_shared_ptr_overloading">12.5.5.3 shared_ptr and
method overloading</a></h4>
<p> A C++ compiler can disambiguate a method overloaded by a shared_ptr
and one using the raw underlying type. For example, either one of these
methods can be called in C++:</p>
<div class="code">
<pre>
int age(std::shared_ptr&lt;GrandParent&gt; num);
int age(GrandParent&amp; num);
</pre>
</div>
<p> When wrapped by SWIG, disambiguation is not possible using the
overloaded names as there is just one equivalent type (<tt>GrandParent</tt>
) in the target language. SWIG will choose to wrap just the first method
by default. <a href="#SWIGPlus_nn25">Ambiguity in overloading</a>
discusses ways to control which method(s) gets wrapped using <tt>
%ignore</tt> or <tt>%rename</tt>. For the interested reader, SWIG
detects that they are equivalent types via the <a href="#Typemaps_typecheck_pointer">
typecheck typemaps</a> in the shared_ptr library.</p>
<h4><a name="Library_shared_ptr_templates">12.5.5.4 shared_ptr and
templates</a></h4>
<p> The <tt>%shared_ptr</tt> macro should be used for all the required
instantiations of the template before each of the <tt>%template</tt>
instantiations. For example, consider <tt>number.h</tt> containing the
following illustrative template:</p>
<div class="code">
<pre>
#include &lt;memory&gt;
template&lt;int N&gt; struct Number {
int num;
Number() : num(N) {}
static std::shared_ptr&lt;Number&lt;N&gt;&gt; make() { return std::make_shared&lt;Number&lt;N&gt;&gt;(); }
};
</pre>
</div>
<p> The SWIG code below shows the required ordering:</p>
<div class="code">
<pre>
%include &lt;std_shared_ptr.i&gt;
%shared_ptr(Number&lt;10&gt;);
%shared_ptr(Number&lt;42&gt;);
%{
#include &quot;number.h&quot;
%}
%include &quot;number.h&quot;
%template(Number10) Number&lt;10&gt;;
%template(Number42) Number&lt;42&gt;;
</pre>
</div>
<h4><a name="Library_shared_ptr_directors">12.5.5.5 shared_ptr and
directors</a></h4>
<p> The languages that support shared_ptr also have support for using
shared_ptr with directors.</p>
<h3><a name="Library_std_unique_ptr">12.5.6 unique_ptr smart pointer</a></h3>
<p> The <tt>std_unique_ptr.i</tt> library file provides SWIG's
unique_ptr support. It provides move semantics for the smart pointer's
underlying object, both from C++ to the target language and vice versa.</p>
<p> The library defines typemaps and a macro, <tt>%unique_ptr(T)</tt>,
to use for handling <tt>std::unique_ptr&lt;T&gt;</tt> for a type <tt>T</tt>.
The type <tt>T</tt> must be non-primitive. This macro should be used
before any code declaring or using type <tt>T</tt>. Ordering
requirements for using this smart pointer macro are the same as the
equivalent <tt>%shared_ptr(T)</tt> macro covered in the previous
section. The ownership and move semantics described here can of course
be modified if not suitable by copying and customising the typemaps in
the appropriate <tt>std_unique_ptr.i</tt> library file.</p>
<p> Note that SWIG doesn't currently support using both <tt>
%shared_ptr(T)</tt> and <tt>%unique_ptr&lt;T&gt;</tt> on the same type <tt>T</tt>
.</p>
<h4><a name="Library_std_unique_ptr_by_value">12.5.6.1 unique_ptr passed
by value</a></h4>
<p> Example usage of a <tt>std::unique_ptr</tt> being returned from a
function by value is shown below.</p>
<div class="code">
<pre>
%include &lt;std_unique_ptr.i&gt;
%unique_ptr(Klass)
%inline %{
#include &lt;memory&gt;
class Klass {
public:
// Factory function creating objects of this class:
static std::unique_ptr&lt;Klass&gt; Create(int value) {
return std::unique_ptr&lt;Klass&gt;(new Klass(value));
}
int getValue() const { return m_value; }
private:
Klass(int value) : m_value(value) {}
int m_value;
};
%}
</pre>
</div>
<p> The returned objects can be used naturally from the target language,
e.g. from C#:</p>
<div class="targetlang">
<pre>
Klass k = Klass.Create(17);
int value = k.getValue();
</pre>
</div>
<p> The implementation simply calls <tt>std::unique_ptr::release()</tt>
to obtain the underlying raw pointer. The pointer is then used to
create a target language proxy class in the same way that SWIG handles
a C++ function returning a class by value. The target language proxy
class then owns the memory pointed to by the raw pointer and memory
handling is identical to normal SWIG proxy class handling of the
underlying C++ memory. Note that an object returned by value is first
copied/moved from the stack onto the heap in order to obtain a raw
pointer on the heap, whereas the underlying raw pointer in <tt>
std::unique_ptr</tt> already points to an object the heap.</p>
<p> Note that the implementation is quite different to the <tt>
std::shared_ptr</tt> smart pointer, where the proxy class manages the
underlying C++ memory as a pointer to a shared_ptr instead of a plain
raw pointer.</p>
<p> A possibly less common usage of this smart pointer is as a parameter
to a function. When used like this it indicates that memory usage of
the object pointed to by the underlying pointer is transferred to the
function being called. The code that SWIG generates assumes this
happens. First, it is assumed that a proxy class already owns the
underlying C++ object and is used to pass the object to the C++
function being called. Second, the ownership is transferred from the
proxy class to the C++ function being called and lifetime is then
controlled by the function. Finally, it is assumed the lifetime of the
object may not last beyond returning from the C++ function and hence
the proxy class can no longer be used.</p>
<p> Consider expanding the example above with a function that takes a <tt>
std::unique_ptr</tt> as follows:</p>
<div class="code">
<pre>
void take(std::unique_ptr&lt;Klass&gt;);
</pre>
</div>
<p> and use from C#:</p>
<div class="targetlang">
<pre>
Klass k = Klass.Create(17); // create an instance of Klass any way you like
int value = k.getValue(); // ok
example.take(k); // memory ownership passes from C# layer to C++ layer
int v = k.getValue(); // don't do this - invalid use of k
</pre>
</div>
<p> Attempts to use <tt>k</tt> after the ownership has been passed into
the <tt>take</tt> function should not be attempted. The implementation
sets the proxy class to an invalid state by setting the class's
underlying C++ pointer to null after the return from the <tt>take</tt>
function. Subsequent use of an invalid proxy class instance is very
much dependent on the implementation in the target language and ranges
from a segfault to giving a nice error. Consider implementing
additional checks via the 'check' typemap.</p>
<p> Attempts to pass ownership from a proxy class to a <tt>std::unique</tt>
parameter more than once will result in a &quot;Cannot release ownership as
memory is not owned&quot; exception. For example, if <tt>example.take(k)</tt>
in the example above is called twice.</p>
<h4><a name="Library_std_unique_ptr_by_ref">12.5.6.2 unique_ptr passed
by reference</a></h4>
<p> The effect of passing a <tt>std::unique_ptr</tt> by rvalue reference
into a function is identical to passing it by value. The ownership of
the memory of the object being pointed to by the underyling pointer is
transferred from the proxy class to the C++ function being called.
Example:</p>
<div class="code">
<pre>
void grab(std::unique_ptr&lt;Klass&gt; &amp;&amp;);
</pre>
</div>
<p> Passing non-const lvalue references into a function is a bit quirky
and not perfect due to ambiguities as to what the function may do. The
approach taken is the ownership is transferred out of the target
language from the proxy class into C++ space and the proxy class can
then no longer be used after the wrapped function returns. In summary
it works much like passing a <tt>std::unique_ptr</tt> by value into a
function. The assumption is the function will not modify the <tt>
std::unique_ptr</tt>. If this is not true and the underlying pointer is
changed, such as calling the member functions, <tt>swap</tt>, <tt>reset</tt>
or <tt>release</tt>, then the modified <tt>std::unique_ptr</tt> will
effectively be ignored. It is destroyed when the function exits C++
space on return to the target language. Example:</p>
<div class="code">
<pre>
void process(std::unique_ptr&lt;Klass&gt; &amp;);
</pre>
</div>
<p> Passing const lvalue references into a function works much like
passing any wrapped class. The proxy class owning the underling C++
object continues to own the underying C++ object after calling the
function, the function cannot modify the <tt>std::unique_ptr</tt> or
take ownership. Example:</p>
<div class="code">
<pre>
void use(const std::unique_ptr&lt;Klass&gt; &amp;);
</pre>
</div>
<p> Move semantics are not provided when wrapping a C++ function that
returns a <tt>std::unique_ptr</tt> by reference. The target language
proxy class wrapper that is returned does not own the underlying C++
object. This applies to all reference types, such as:</p>
<div class="code">
<pre>
std::unique_ptr&lt;Klass&gt; &amp; LvalueRefReturn();
std::unique_ptr&lt;Klass&gt; &amp;&amp; RvalueRefReturn();
</pre>
</div>
<p><b> Compatibility note:</b> Support for <tt>std::unique_ptr</tt> was
first added in SWIG-4.1.0. This initial support contained the move
semantics when passing a <tt>std::unique_ptr</tt> around by value.
Support for passing a <tt>std::unique_ptr</tt> around by reference was
added in SWIG-4.3.0.</p>
<h3><a name="Library_std_auto_ptr">12.5.7 auto_ptr smart pointer</a></h3>
<p> While <tt>std::auto_ptr</tt> is deprecated in C++11, some existing
code may still be using it. SWIG provides support for this class which
is nearly identical to <tt>std::unique_ptr</tt>.</p>
<p> The <tt>std_auto_ptr.i</tt> library file provides SWIG's auto_ptr
support. It defines typemaps and a macro, <tt>%auto_ptr(T)</tt>, to use
for handling <tt>std::auto_ptr&lt;T&gt;</tt> for a type <tt>T</tt>. The type <tt>
T</tt> must be non-primitive. This macro should be used before any code
declaring or using type <tt>T</tt>. Ordering requirements for using
this smart pointer macro are the same as the equivalent <tt>
%shared_ptr(T)</tt> and <tt>%unique_ptr</tt> macros covered in the
previous two sections.</p>
<p> Example usage of a <tt>std::auto_ptr</tt> being returned from a
function is shown below.</p>
<div class="code">
<pre>
%include &lt;std_auto_ptr.i&gt;
%auto_ptr(Klass)
%inline %{
#include &lt;memory&gt;
class Klass {
public:
// Factory function creating objects of this class:
static std::auto_ptr&lt;Klass&gt; Create(int value) {
return std::auto_ptr&lt;Klass&gt;(new Klass(value));
}
int getValue() const { return m_value; }
private:
Klass(int value) : m_value(value) {}
int m_value;
};
%}
</pre>
</div>
<p> The returned objects can be used naturally from the target language,
e.g. from C#:</p>
<div class="targetlang">
<pre>
Klass k = Klass.Create(17);
int value = k.getValue();
</pre>
</div>
<p> The implementation simply calls <tt>std::auto_ptr::release()</tt> to
obtain the underlying raw pointer. That is, it works the same way
covered in the previous section for <tt>std::unique_ptr</tt>.</p>
<p> Input parameters also work the same way as <tt>std::unique_ptr</tt>
covered in the previous section.</p>
<h2><a name="Library_nn16">12.6 Utility Libraries</a></h2>
<h3><a name="Library_nn17">12.6.1 exception.i</a></h3>
<p> The <tt>exception.i</tt> library provides a language-independent
function for raising a run-time exception in the target language. This
library is largely used by the SWIG library writers. If possible, use
the error handling scheme available to your target language as there is
greater flexibility in what errors/exceptions can be thrown.</p>
<p><b> <tt>SWIG_exception(int code, const char *message)</tt></b></p>
<div class="indent">
<p> Raises an exception in the target language. <tt>code</tt> is one of
the following symbolic constants:</p>
<div class="code">
<pre>
SWIG_MemoryError
SWIG_IOError
SWIG_RuntimeError
SWIG_IndexError
SWIG_TypeError
SWIG_DivisionByZero
SWIG_OverflowError
SWIG_SyntaxError
SWIG_ValueError
SWIG_SystemError
SWIG_NullReferenceError
</pre>
</div>
<p> <tt>message</tt> is a string indicating more information about the
problem.</p>
</div>
<p> The primary use of this module is in writing language-independent
exception handlers. For example:</p>
<div class="code">
<pre>
%include &quot;exception.i&quot;
%exception std::vector::getitem {
try {
$action
} catch (std::out_of_range&amp; e) {
SWIG_exception(SWIG_IndexError, const_cast&lt;char*&gt;(e.what()));
}
}
</pre>
</div>
<h3><a name="Library_attributes">12.6.2 attribute.i</a></h3>
<p> The attribute library contains a set of macros to convert a pair of
set/get methods into a &quot;native&quot; attribute/property.</p>
<p> Use <tt>%attribute</tt> when you have a pair of get/set methods to a
primitive type like:</p>
<div class="code">
<pre>
%include &quot;attribute.i&quot;
%attribute(A, int, a, get_a, set_a);
struct A {
int get_a() const;
void set_a(int aa);
};
</pre>
</div>
<p> and you want to provide that variable as an attribute in the target
language. This example only works for primitive types, not derived
types. Now you can use the attributes like so (in Python):</p>
<div class="targetlang">
<pre>
x = A()
x.a = 3 # calls A::set_a(3)
print(x.a) # calls A::get_a() const
</pre>
</div>
<p> If you don't provide a 'set' method, a 'read-only' attribute is
generated, ie, like:</p>
<div class="code">
<pre>
%attribute(A, int, c, get_c);
</pre>
</div>
<p> Use <tt>%attributeref</tt> when you have const/non-const reference
access methods for primitive types or class/structs, like:</p>
<div class="code">
<pre>
%attributeref(A, int, b);
struct A {
const int &amp; b() const;
int &amp; b();
};
%attributeref(B, int, c);
struct B {
int &amp; c();
};
</pre>
</div>
<p> Use the attributes like so (in Python):</p>
<div class="targetlang">
<pre>
x = A()
x.b = 3 # calls A::b()
print(x.b) # calls A::b() const
</pre>
</div>
<p> You can also use</p>
<div class="code">
<pre>
%attributeref(Class, AttributeType, AttributeName, AccessorMethod)
</pre>
</div>
<p> if the internal C++ reference methods have a different name from the
attribute you want, so</p>
<div class="code">
<pre>
%attributeref(B, int, d, c);
</pre>
</div>
<p> is the same as the last example, but instead of the attribute 'c'
being called 'c', it is called 'd'.</p>
<p> Use <tt>%attribute2</tt> instead of <tt>%attribute</tt> to indicate
that reference-pointer translation is required. Use <tt>%attribute2</tt>
instead of <tt>%attribute</tt> in cases like this:</p>
<div class="code">
<pre>
%attribute2(MyClass, MyFoo, Foo, GetFoo, SetFoo);
%inline %{
struct MyFoo {
int x;
};
class MyClass {
MyFoo foo;
public:
MyFoo &amp; GetFoo() { return foo; }
void SetFoo(const MyFoo &amp;other) { foo = other; }
};
%}
</pre>
</div>
<p> Here, the data type of the property is a wrapped type <tt>MyFoo</tt>
and on the C++ side it is passed by reference. The problem is that the
SWIG wrapper will pass around a pointer (MyFoo *) which is not
compatible with the reference type of the accessors (MyFoo &amp;).
Therefore, if you use <tt>%attribute</tt>, you'll get an error from
your C/C++ compiler. <tt>%attribute2</tt> translates between a pointer
and a reference to eliminate the error. In case you're confused, let's
make it simple: just use <tt>%attribute</tt> at first, but if the C/C++
compiler gives an error while compiling the wrapper, try <tt>
%attribute2</tt> instead.</p>
<p> NOTE: remember that if the type contains commas, such as <tt>
std::pair&lt;int, int&gt;</tt>, you need to use the macro like:</p>
<div class="code">
<pre>
%attributeref(A, %arg(std::pair&lt;int, int&gt;), pval);
</pre>
</div>
<p> where <tt>%arg()</tt> 'normalizes' the type to be understood as a
single argument, otherwise the macro will get confused by the comma.</p>
<p> The <tt>%attributeval</tt> is the same as <tt>%attribute</tt>, but
should be used when the type is a class/struct (ie a non-primitive
type) and when the get and set methods return/pass by value. The
following is very similar to the above example, but note that the
access is by value rather than reference.</p>
<div class="code">
<pre>
%attributeval(MyClassVal, MyFoo, ReadWriteFoo, GetFoo, SetFoo);
%attributeval(MyClassVal, MyFoo, ReadOnlyFoo, GetFoo);
%inline %{
class MyClassVal {
MyFoo foo;
public:
MyFoo GetFoo() { return foo; }
void SetFoo(MyFoo other) { foo = other; }
};
%}
</pre>
</div>
<p> The <tt>%attributestring</tt> is the same as <tt>%attributeval</tt>,
but should be used for string class types, which are unusual as they
are a class on the C++ side, but normally an immutable/primitive type
in the target language. Example usage for <tt>std::string</tt>:</p>
<div class="code">
<pre>
%include &lt;std_string.i&gt;
%attributestring(MyStringyClass, std::string, ReadWriteString, GetString, SetString);
%attributestring(MyStringyClass, std::string, ReadOnlyString, GetString);
%inline %{
class MyStringyClass {
std::string str;
public:
MyStringyClass(const std::string &amp;val) : str(val) {}
std::string GetString() { return str; }
void SetString(std::string other) { str = other; }
};
%}
</pre>
</div>
<p> The <tt>%attributestring</tt> also works for class types that have <tt>
%naturalvar</tt> turned on and so is also useful for shared_ptr which
has <tt>%naturalvar</tt> turned on in <tt>%shared_ptr</tt>.</p>
<h4><a name="Library_attribute_templates">12.6.2.1 %attribute and C++
templates</a></h4>
<p> <tt>%attribute</tt> and friends have to be used on fully specified
classes. For example</p>
<div class="code">
<pre>
%attributeref(A&lt;int&gt;, int, a);
%inline %{
template &lt;class T&gt; struct A {
T a() const;
void a(T &amp;);
};
%}
</pre>
</div>
<p> Note the use of a template-id (i.e., <tt>A&lt;int&gt;</tt> not <tt>A&lt;T&gt;</tt>
or just <tt>A</tt>). This means that <tt>%attribute</tt> statements
have to be repeated for any template-id that you want to use with <tt>
%template</tt>.</p>
<HR NOSHADE>
<h1><a name="Arguments">13 Argument Handling</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Arguments_nn2">The typemaps.i library</a>
<ul>
<li><a href="#Arguments_nn3">Introduction</a></li>
<li><a href="#Arguments_nn4">Input parameters</a></li>
<li><a href="#Arguments_nn5">Output parameters</a></li>
<li><a href="#Arguments_nn6">Input/Output parameters</a></li>
<li><a href="#Arguments_nn7">Using different names</a></li>
</ul>
</li>
<li><a href="#Arguments_nn8">Applying constraints to input values</a>
<ul>
<li><a href="#Arguments_nn9">Simple constraint example</a></li>
<li><a href="#Arguments_nn10">Constraint methods</a></li>
<li><a href="#Arguments_nn11">Applying constraints to new datatypes</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> In Chapter 5, SWIG's treatment of basic datatypes and pointers was
described. In particular, primitive types such as <tt>int</tt> and <tt>
double</tt> are mapped to corresponding types in the target language.
For everything else, pointers are used to refer to structures, classes,
arrays, and other user-defined datatypes. However, in certain
applications it is desirable to change SWIG's handling of a specific
datatype. For example, you might want to return multiple values through
the arguments of a function. This chapter describes some of the
techniques for doing this.</p>
<h2><a name="Arguments_nn2">13.1 The typemaps.i library</a></h2>
<p> This section describes the <tt>typemaps.i</tt> library
file--commonly used to change certain properties of argument
conversion.</p>
<h3><a name="Arguments_nn3">13.1.1 Introduction</a></h3>
<p> Suppose you had a C function like this:</p>
<div class="code">
<pre>
void add(double a, double b, double *result) {
*result = a + b;
}
</pre>
</div>
<p> From reading the source code, it is clear that the function is
storing a value in the <tt>double *result</tt> parameter. However,
since SWIG does not examine function bodies, it has no way to know that
this is the underlying behavior.</p>
<p> One way to deal with this is to use the <tt>typemaps.i</tt> library
file and write interface code like this:</p>
<div class="code">
<pre>
// Simple example using typemaps
%module example
%include &quot;typemaps.i&quot;
%apply double *OUTPUT { double *result };
%inline %{
extern void add(double a, double b, double *result);
%}
</pre>
</div>
<p> The <tt>%apply</tt> directive tells SWIG that you are going to apply
a special type handling rule to a type. The &quot;<tt>double *OUTPUT</tt>&quot;
specification is the name of a rule that defines how to return an
output value from an argument of type <tt>double *</tt>. This rule gets
applied to all of the datatypes listed in curly braces-- in this case &quot;<tt>
double *result</tt>&quot;.</p>
<p> When the resulting module is created, you can now use the function
like this (shown for Python):</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = add(3, 4)
&gt;&gt;&gt; print a
7
&gt;&gt;&gt;
</pre>
</div>
<p> In this case, you can see how the output value normally returned in
the third argument has magically been transformed into a function
return value. Clearly this makes the function much easier to use since
it is no longer necessary to manufacture a special <tt>double *</tt>
object and pass it to the function somehow.</p>
<p> Once a typemap has been applied to a type, it stays in effect for
all future occurrences of the type and name. For example, you could
write the following:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
%apply double *OUTPUT { double *result };
%inline %{
extern void add(double a, double b, double *result);
extern void sub(double a, double b, double *result);
extern void mul(double a, double b, double *result);
extern void div(double a, double b, double *result);
%}
...
</pre>
</div>
<p> In this case, the <tt>double *OUTPUT</tt> rule is applied to all of
the functions that follow.</p>
<p> Typemap transformations can even be extended to multiple return
values. For example, consider this code:</p>
<div class="code">
<pre>
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *width, int *height };
// Returns a pair (width, height)
void getwinsize(int winid, int *width, int *height);
</pre>
</div>
<p> In this case, the function returns multiple values, allowing it to
be used like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; w, h = genwinsize(wid)
&gt;&gt;&gt; print w
400
&gt;&gt;&gt; print h
300
&gt;&gt;&gt;
</pre>
</div>
<p> It should also be noted that although the <tt>%apply</tt> directive
is used to associate typemap rules to datatypes, you can also use the
rule names directly in arguments. For example, you could write this:</p>
<div class="code">
<pre>
// Simple example using typemaps
%module example
%include &quot;typemaps.i&quot;
%{
extern void add(double a, double b, double *OUTPUT);
%}
extern void add(double a, double b, double *OUTPUT);
</pre>
</div>
<p> Typemaps stay in effect until they are explicitly deleted or
redefined to something else. To clear a typemap, the <tt>%clear</tt>
directive should be used. For example:</p>
<div class="code">
<pre>
%clear double *result; // Remove all typemaps for double *result
</pre>
</div>
<h3><a name="Arguments_nn4">13.1.2 Input parameters</a></h3>
<p> The following typemaps instruct SWIG that a pointer really only
holds a single input value:</p>
<div class="code">
<pre>
int *INPUT
short *INPUT
long *INPUT
unsigned int *INPUT
unsigned short *INPUT
unsigned long *INPUT
double *INPUT
float *INPUT
</pre>
</div>
<p> When used, it allows values to be passed instead of pointers. For
example, consider this function:</p>
<div class="code">
<pre>
double add(double *a, double *b) {
return *a+*b;
}
</pre>
</div>
<p> Now, consider this SWIG interface:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
...
%{
extern double add(double *, double *);
%}
extern double add(double *INPUT, double *INPUT);
</pre>
</div>
<p> When the function is used in the scripting language interpreter, it
will work like this:</p>
<div class="targetlang">
<pre>
result = add(3, 4)
</pre>
</div>
<h3><a name="Arguments_nn5">13.1.3 Output parameters</a></h3>
<p> The following typemap rules tell SWIG that pointer is the output
value of a function. When used, you do not need to supply the argument
when calling the function. Instead, one or more output values are
returned.</p>
<div class="code">
<pre>
int *OUTPUT
short *OUTPUT
long *OUTPUT
unsigned int *OUTPUT
unsigned short *OUTPUT
unsigned long *OUTPUT
double *OUTPUT
float *OUTPUT
</pre>
</div>
<p> These methods can be used as shown in an earlier example. For
example, if you have this C function :</p>
<div class="code">
<pre>
void add(double a, double b, double *c) {
*c = a+b;
}
</pre>
</div>
<p> A SWIG interface file might look like this :</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
...
%inline %{
extern void add(double a, double b, double *OUTPUT);
%}
</pre>
</div>
<p> In this case, only a single output value is returned, but this is
not a restriction. An arbitrary number of output values can be returned
by applying the output rules to more than one argument (as shown
previously).</p>
<p> If the function also returns a value, it is returned along with the
argument. For example, if you had this:</p>
<div class="code">
<pre>
extern int foo(double a, double b, double *OUTPUT);
</pre>
</div>
<p> The function will return two values like this:</p>
<div class="targetlang">
<pre>
iresult, dresult = foo(3.5, 2)
</pre>
</div>
<h3><a name="Arguments_nn6">13.1.4 Input/Output parameters</a></h3>
<p> When a pointer serves as both an input and output value you can use
the following typemaps :</p>
<div class="code">
<pre>
int *INOUT
short *INOUT
long *INOUT
unsigned int *INOUT
unsigned short *INOUT
unsigned long *INOUT
double *INOUT
float *INOUT
</pre>
</div>
<p> A C function that uses this might be something like this:</p>
<div class="code">
<pre>
void negate(double *x) {
*x = -(*x);
}
</pre>
</div>
<p> To make x function as both and input and output value, declare the
function like this in an interface file :</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
...
%{
extern void negate(double *);
%}
extern void negate(double *INOUT);
</pre>
</div>
<p> Now within a script, you can simply call the function normally :</p>
<div class="targetlang">
<pre>
a = negate(3); # a = -3 after calling this
</pre>
</div>
<p> One subtle point of the <tt>INOUT</tt> rule is that many scripting
languages enforce mutability constraints on primitive objects (meaning
that simple objects like integers and strings aren't supposed to
change). Because of this, you can't just modify the object's value in
place as the underlying C function does in this example. Therefore, the
<tt>INOUT</tt> rule returns the modified value as a new object rather
than directly overwriting the value of the original input object.</p>
<h3><a name="Arguments_nn7">13.1.5 Using different names</a></h3>
<p> As previously shown, the <tt>%apply</tt> directive can be used to
apply the <tt>INPUT</tt>, <tt>OUTPUT</tt>, and <tt>INOUT</tt> typemaps
to different argument names. For example:</p>
<div class="code">
<pre>
// Make double *result an output value
%apply double *OUTPUT { double *result };
// Make Int32 *in an input value
%apply int *INPUT { Int32 *in };
// Make long *x inout
%apply long *INOUT {long *x};
</pre>
</div>
<p> To clear a rule, the <tt>%clear</tt> directive is used:</p>
<div class="code">
<pre>
%clear double *result;
%clear Int32 *in, long *x;
</pre>
</div>
<p> Typemap declarations are lexically scoped so a typemap takes effect
from the point of definition to the end of the file or a matching <tt>
%clear</tt> declaration.</p>
<h2><a name="Arguments_nn8">13.2 Applying constraints to input values</a>
</h2>
<p> In addition to changing the handling of various input values, it is
also possible to use typemaps to apply constraints. For example, maybe
you want to insure that a value is positive, or that a pointer is
non-NULL. This can be accomplished including the <tt>constraints.i</tt>
library file.</p>
<h3><a name="Arguments_nn9">13.2.1 Simple constraint example</a></h3>
<p> The constraints library is best illustrated by the following
interface file :</p>
<div class="code">
<pre>
// Interface file with constraints
%module example
%include &quot;constraints.i&quot;
double exp(double x);
double log(double POSITIVE); // Allow only positive values
double sqrt(double NONNEGATIVE); // Non-negative values only
double inv(double NONZERO); // Non-zero values
int fclose(FILE *NONNULL); // Non-NULL pointers only
</pre>
</div>
<p> The behavior of this file is exactly as you would expect. If any of
the arguments violate the constraint condition, a scripting language
exception will be raised. As a result, it is possible to catch bad
values, prevent mysterious program crashes and so on.</p>
<h3><a name="Arguments_nn10">13.2.2 Constraint methods</a></h3>
<p> The following constraints are currently available</p>
<div class="code">
<pre>
POSITIVE Any number &gt; 0 (not zero)
NEGATIVE Any number &lt; 0 (not zero)
NONNEGATIVE Any number &gt;= 0
NONPOSITIVE Any number &lt;= 0
NONZERO Nonzero number
NONNULL Non-NULL pointer (pointers only).
</pre>
</div>
<h3><a name="Arguments_nn11">13.2.3 Applying constraints to new
datatypes</a></h3>
<p> The constraints library only supports the primitive C datatypes, but
it is easy to apply it to new datatypes using <tt>%apply</tt>. For
example :</p>
<div class="code">
<pre>
// Apply a constraint to a Real variable
%apply Number POSITIVE { Real in };
// Apply a constraint to a pointer type
%apply Pointer NONNULL { Vector * };
</pre>
</div>
<p> The special types of &quot;Number&quot; and &quot;Pointer&quot; can be applied to any
numeric and pointer variable type respectively. To later remove a
constraint, the <tt>%clear</tt> directive can be used :</p>
<div class="code">
<pre>
%clear Real in;
%clear Vector *;
</pre>
</div><HR NOSHADE>
<h1><a name="Typemaps">14 Typemaps</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Typemaps_nn2">Introduction</a>
<ul>
<li><a href="#Typemaps_nn3">Type conversion</a></li>
<li><a href="#Typemaps_nn4">Typemaps</a></li>
<li><a href="#Typemaps_nn5">Pattern matching</a></li>
<li><a href="#Typemaps_nn6">Reusing typemaps</a></li>
<li><a href="#Typemaps_nn7">What can be done with typemaps?</a></li>
<li><a href="#Typemaps_nn8">What can't be done with typemaps?</a></li>
<li><a href="#Typemaps_aspects">Similarities to Aspect Oriented
Programming</a></li>
<li><a href="#Typemaps_nn9">The rest of this chapter</a></li>
</ul>
</li>
<li><a href="#Typemaps_nn10">Typemap specifications</a>
<ul>
<li><a href="#Typemaps_defining">Defining a typemap</a></li>
<li><a href="#Typemaps_nn12">Typemap scope</a></li>
<li><a href="#Typemaps_nn13">Copying a typemap</a></li>
<li><a href="#Typemaps_nn14">Deleting a typemap</a></li>
<li><a href="#Typemaps_nn15">Placement of typemaps</a></li>
</ul>
</li>
<li><a href="#Typemaps_pattern_matching">Pattern matching rules</a>
<ul>
<li><a href="#Typemaps_nn17">Basic matching rules</a></li>
<li><a href="#Typemaps_typedef_reductions">Typedef reductions matching</a>
</li>
<li><a href="#Typemaps_nn19">Default typemap matching rules</a></li>
<li><a href="#Typemaps_multi_argument_typemaps_patterns">Multi-arguments
typemaps</a></li>
<li><a href="#Typemaps_matching_template_comparison">Matching rules
compared to C++ templates</a></li>
<li><a href="#Typemaps_debugging_search">Debugging typemap pattern
matching</a></li>
</ul>
</li>
<li><a href="#Typemaps_nn21">Code generation rules</a>
<ul>
<li><a href="#Typemaps_nn22">Scope</a></li>
<li><a href="#Typemaps_nn23">Declaring new local variables</a></li>
<li><a href="#Typemaps_special_variables">Special variables</a>
<ul>
<li><a href="#Typemaps_type_special_variables">Type related special
variables</a></li>
<li><a href="#Typemaps_non_type_special_variables">Non-type related
special variables</a></li>
</ul>
</li>
<li><a href="#Typemaps_special_variable_macros">Special variable macros</a>
<ul>
<li><a href="#Typemaps_special_macro_descriptor">$descriptor(type)</a></li>
<li><a href="#Typemaps_special_macro_typemap">$typemap(method,
typepattern)</a></li>
<li><a href="#Typemaps_special_macro_typemap_attribute">
$typemap(method:attribute, typepattern)</a></li>
</ul>
</li>
<li><a href="#Typemaps_special_variable_attributes">Special variables
and typemap attributes</a></li>
<li><a href="#Typemaps_special_variables_and_macros">Special variables
combined with special variable macros</a></li>
</ul>
</li>
<li><a href="#Typemaps_nn25">Common typemap methods</a>
<ul>
<li><a href="#Typemaps_nn26">&quot;in&quot; typemap</a></li>
<li><a href="#Typemaps_nn27">&quot;typecheck&quot; typemap</a></li>
<li><a href="#Typemaps_nn28">&quot;out&quot; typemap</a></li>
<li><a href="#Typemaps_nn29">&quot;arginit&quot; typemap</a></li>
<li><a href="#Typemaps_nn30">&quot;default&quot; typemap</a></li>
<li><a href="#Typemaps_nn31">&quot;check&quot; typemap</a></li>
<li><a href="#Typemaps_nn32">&quot;argout&quot; typemap</a></li>
<li><a href="#Typemaps_nn33">&quot;freearg&quot; typemap</a></li>
<li><a href="#Typemaps_nn34">&quot;newfree&quot; typemap</a></li>
<li><a href="#Typemaps_ret">&quot;ret&quot; typemap</a></li>
<li><a href="#Typemaps_nn35">&quot;memberin&quot; typemap</a></li>
<li><a href="#Typemaps_nn36">&quot;varin&quot; typemap</a></li>
<li><a href="#Typemaps_nn37">&quot;varout&quot; typemap</a></li>
<li><a href="#Typemaps_throws_typemap">&quot;throws&quot; typemap</a></li>
</ul>
</li>
<li><a href="#Typemaps_nn39">Some typemap examples</a>
<ul>
<li><a href="#Typemaps_nn40">Typemaps for arrays</a></li>
<li><a href="#Typemaps_nn41">Implementing constraints with typemaps</a></li>
</ul>
</li>
<li><a href="#Typemaps_nn43">Typemaps for multiple target languages</a></li>
<li><a href="#Typemaps_optimal">Optimal code generation when returning
by value</a></li>
<li><a href="#Typemaps_multi_argument_typemaps">Multi-argument typemaps</a>
</li>
<li><a href="#Typemaps_warnings">Typemap warnings</a></li>
<li><a href="#Typemaps_fragments">Typemap fragments</a>
<ul>
<li><a href="#Typemaps_fragment_type_specialization">Fragment type
specialization</a></li>
<li><a href="#Typemaps_automatic_specialization">Fragments and automatic
typemap specialization</a></li>
</ul>
</li>
<li><a href="#Typemaps_runtime_type_checker">The run-time type checker</a>
<ul>
<li><a href="#Typemaps_nn45">Implementation</a></li>
<li><a href="#Typemaps_runtime_type_checker_usage">Usage</a></li>
</ul>
</li>
<li><a href="#Typemaps_overloading">Typemaps and overloading</a>
<ul>
<li><a href="#Typemaps_typecheck_pointer">SWIG_TYPECHECK_POINTER
precedence level and the typecheck typemap</a></li>
</ul>
</li>
<li><a href="#Typemaps_nn48">More about %apply and %clear</a></li>
<li><a href="#Typemaps_nn47">Passing data between typemaps</a></li>
<li><a href="#Typemaps_nn52">C++ &quot;this&quot; pointer</a></li>
<li><a href="#Typemaps_nn51">Where to go for more information?</a></li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="Typemaps_nn2">14.1 Introduction</a></h2>
<p> Chances are, you are reading this chapter for one of two reasons;
you either want to customize SWIG's behavior or you overheard someone
mumbling some incomprehensible drivel about &quot;typemaps&quot; and you asked
yourself &quot;typemaps, what are those?&quot; That said, let's start with a
short disclaimer that &quot;typemaps&quot; are an advanced customization feature
that provide direct access to SWIG's low-level code generator. Not only
that, they are an integral part of the SWIG C++ type system (a
non-trivial topic of its own). Typemaps are generally<em> not</em> a
required part of using SWIG. Therefore, you might want to re-read the
earlier chapters if you have found your way to this chapter with only a
vague idea of what SWIG already does by default.</p>
<h3><a name="Typemaps_nn3">14.1.1 Type conversion</a></h3>
<p> One of the most important problems in wrapper code generation is the
conversion or marshalling of datatypes between programming languages.
Specifically, for every C/C++ declaration, SWIG must somehow generate
wrapper code that allows values to be passed back and forth between
languages. Since every programming language represents data
differently, this is not a simple of matter of simply linking code
together with the C linker. Instead, SWIG has to know something about
how data is represented in each language and how it can be manipulated.</p>
<p> To illustrate, suppose you had a simple C function like this:</p>
<div class="code">
<pre>
int factorial(int n);
</pre>
</div>
<p> To access this function from Python, a pair of Python API functions
are used to convert integer values. For example:</p>
<div class="code">
<pre>
long PyInt_AsLong(PyObject *obj); /* Python --&gt; C */
PyObject *PyInt_FromLong(long x); /* C --&gt; Python */
</pre>
</div>
<p> The first function is used to convert the input argument from a
Python integer object to C <tt>long</tt>. The second function is used
to convert a value from C back into a Python integer object.</p>
<p> Inside the wrapper function, you might see these functions used like
this:</p>
<div class="code">
<pre>
PyObject *wrap_factorial(PyObject *self, PyObject *args) {
int arg1;
int result;
PyObject *obj1;
PyObject *resultobj;
if (!PyArg_ParseTuple(&quot;O:factorial&quot;, &amp;obj1)) return NULL;
<b>arg1 = PyInt_AsLong(obj1);</b>
result = factorial(arg1);
<b>resultobj = PyInt_FromLong(result);</b>
return resultobj;
}
</pre>
</div>
<p> Every target language supported by SWIG has functions that work in a
similar manner. For example, in Perl, the following functions are used:</p>
<div class="code">
<pre>
IV SvIV(SV *sv); /* Perl --&gt; C */
void sv_setiv(SV *sv, IV val); /* C --&gt; Perl */
</pre>
</div>
<p> In Tcl:</p>
<div class="code">
<pre>
int Tcl_GetLongFromObj(Tcl_Interp *interp, Tcl_Obj *obj, long *value);
Tcl_Obj *Tcl_NewIntObj(long value);
</pre>
</div>
<p> The precise details are not so important. What is important is that
all of the underlying type conversion is handled by collections of
utility functions and short bits of C code like this---you simply have
to read the extension documentation for your favorite language to know
how it works (an exercise left to the reader).</p>
<h3><a name="Typemaps_nn4">14.1.2 Typemaps</a></h3>
<p> Since type handling is so central to wrapper code generation, SWIG
allows it to be completely defined (or redefined) by the user. To do
this, a special <tt>%typemap</tt> directive is used. For example:</p>
<div class="code">
<pre>
/* Convert from Python --&gt; C */
%typemap(in) int {
$1 = PyInt_AsLong($input);
}
/* Convert from C --&gt; Python */
%typemap(out) int {
$result = PyInt_FromLong($1);
}
</pre>
</div>
<p> At first glance, this code will look a little confusing. However,
there is really not much to it. The first typemap (the &quot;in&quot; typemap) is
used to convert a value from the target language to C. The second
typemap (the &quot;out&quot; typemap) is used to convert in the other direction.
The content of each typemap is a small fragment of code that is
inserted directly into the SWIG generated wrapper functions. The code
is usually C or C++ code which will be generated into the C/C++ wrapper
functions. Note that this isn't always the case as some target language
modules allow target language code within the typemaps which gets
generated into target language specific files. Within this code, a
number of special variables prefixed with a <tt>$</tt> are expanded.
These are really just placeholders for C/C++ variables that are
generated in the course of creating the wrapper function. In this case,
<tt>$input</tt> refers to an input object that needs to be converted to
C/C++ and <tt>$result</tt> refers to an object that is going to be
returned by a wrapper function. <tt>$1</tt> refers to a C/C++ variable
that has the same type as specified in the typemap declaration (an <tt>
int</tt> in this example).</p>
<p> A short example might make this a little more clear. If you were
wrapping a function like this:</p>
<div class="code">
<pre>
int gcd(int x, int y);
</pre>
</div>
<p> A wrapper function would look approximately like this:</p>
<div class="code">
<pre>
PyObject *wrap_gcd(PyObject *self, PyObject *args) {
int arg1;
int arg2;
int result;
PyObject *obj1;
PyObject *obj2;
PyObject *resultobj;
if (!PyArg_ParseTuple(&quot;OO:gcd&quot;, &amp;obj1, &amp;obj2)) return NULL;
/* &quot;in&quot; typemap, argument 1 */<b>
{
arg1 = PyInt_AsLong(obj1);
}
</b>
/* &quot;in&quot; typemap, argument 2 */<b>
{
arg2 = PyInt_AsLong(obj2);
}
</b>
result = gcd(arg1, arg2);
/* &quot;out&quot; typemap, return value */<b>
{
resultobj = PyInt_FromLong(result);
}
</b>
return resultobj;
}
</pre>
</div>
<p> In this code, you can see how the typemap code has been inserted
into the function. You can also see how the special $ variables have
been expanded to match certain variable names inside the wrapper
function. This is really the whole idea behind typemaps--they simply
let you insert arbitrary code into different parts of the generated
wrapper functions. Because arbitrary code can be inserted, it possible
to completely change the way in which values are converted.</p>
<h3><a name="Typemaps_nn5">14.1.3 Pattern matching</a></h3>
<p> As the name implies, the purpose of a typemap is to &quot;map&quot; C
datatypes to types in the target language. Once a typemap is defined
for a C datatype, it is applied to all future occurrences of that type
in the input file. For example:</p>
<div class="code">
<pre>
/* Convert from Perl --&gt; C */
%typemap(in) <b>int</b> {
$1 = SvIV($input);
}
...
int factorial(<b>int</b> n);
int gcd(<b>int</b> x, <b>int</b> y);
int count(char *s, char *t, <b>int</b> max);
</pre>
</div>
<p> The matching of typemaps to C datatypes is more than a simple
textual match. In fact, typemaps are fully built into the underlying
type system. Therefore, typemaps are unaffected by <tt>typedef</tt>,
namespaces, and other declarations that might hide the underlying type.
For example, you could have code like this:</p>
<div class="code">
<pre>
/* Convert from Ruby--&gt; C */
%typemap(in) <b>int</b> {
$1 = NUM2INT($input);
}
...
typedef int Integer;
namespace foo {
typedef Integer Number;
};
int foo(<b>int</b> x);
int bar(<b>Integer</b> y);
int spam(<b>foo::Number</b> a, <b>foo::Number</b> b);
</pre>
</div>
<p> In this case, the typemap is still applied to the proper arguments
even though typenames don't always match the text &quot;int&quot;. This ability
to track types is a critical part of SWIG--in fact, all of the target
language modules work merely define a family of typemaps for the basic
types. Yet, it is never necessary to write new typemaps for typenames
introduced by <tt>typedef</tt>.</p>
<p> In addition to tracking typenames, typemaps may also be specialized
to match against a specific argument name. For example, you could write
a typemap like this:</p>
<div class="code">
<pre>
%typemap(in) <b>double nonnegative</b> {
$1 = PyFloat_AsDouble($input);
if ($1 &lt; 0) {
PyErr_SetString(PyExc_ValueError, &quot;argument must be nonnegative.&quot;);
SWIG_fail;
}
}
...
double sin(double x);
double cos(double x);
double sqrt(<b>double nonnegative</b>);
typedef double Real;
double log(<b>Real nonnegative</b>);
...
</pre>
</div>
<p> For certain tasks such as input argument conversion, typemaps can be
defined for sequences of consecutive arguments. For example:</p>
<div class="code">
<pre>
%typemap(in) (<b>char *str, int len</b>) {
$1 = PyString_AsString($input); /* char *str */
$2 = PyString_Size($input); /* int len */
}
...
int count(<b>char *str, int len</b>, char c);
</pre>
</div>
<p> In this case, a single input object is expanded into a pair of C
arguments. This example also provides a hint to the unusual variable
naming scheme involving <tt>$1</tt>, <tt>$2</tt>, and so forth.</p>
<h3><a name="Typemaps_nn6">14.1.4 Reusing typemaps</a></h3>
<p> Typemaps are normally defined for specific type and argument name
patterns. However, typemaps can also be copied and reused. One way to
do this is to use assignment like this:</p>
<div class="code">
<pre>
%typemap(in) Integer = int;
%typemap(in) (char *buffer, int size) = (char *str, int len);
</pre>
</div>
<p> There is a more powerful way to copy a family of typemaps though.
Consider the following family of two typemap methods, &quot;in&quot; and &quot;out&quot;
for type <tt>int</tt>:</p>
<div class="code">
<pre>
%typemap(in) int {
/* Convert an integer argument */
...
}
%typemap(out) int {
/* Return an integer value */
...
}
</pre>
</div>
<p> Each of the two typemap methods could be copied individually for
type <tt>size_t</tt> as follows:</p>
<div class="code">
<pre>
/* Apply all of the int typemaps to size_t */
%typemap(in) size_t = int;
%typemap(out) size_t = int;
</pre>
</div>
<p> A more powerful form of copying is available from the <tt>%apply</tt>
directive. The code below is identical to the above:</p>
<div class="code">
<pre>
/* Apply all of the int typemaps to size_t */
%apply int { size_t };
</pre>
</div>
<p> <tt>%apply</tt> merely takes<em> all</em> of the typemaps that are
defined for one type and applies them to other types. Note: you can
include a comma separated set of types in the <tt>{ ... }</tt> part of <tt>
%apply</tt>.</p>
<p> It should be noted that it is not necessary to copy typemaps for
types that are related by <tt>typedef</tt>. For example, if you have
this,</p>
<div class="code">
<pre>
typedef int size_t;
</pre>
</div>
<p> then SWIG already knows that the <tt>int</tt> typemaps apply. You
don't have to do anything.</p>
<h3><a name="Typemaps_nn7">14.1.5 What can be done with typemaps?</a></h3>
<p> The primary use of typemaps is for defining wrapper generation
behavior at the level of individual C/C++ datatypes. There are
currently six general categories of problems that typemaps address:</p>
<p><b> Argument handling</b></p>
<div class="code">
<pre>
int foo(<b>int x, double y, char *s</b>);
</pre>
</div>
<ul>
<li>Input argument conversion (&quot;in&quot; typemap).</li>
<li>Input argument type checking for types used in overloaded methods
(&quot;typecheck&quot; typemap).</li>
<li>Output argument handling (&quot;argout&quot; typemap).</li>
<li>Input argument value checking (&quot;check&quot; typemap).</li>
<li>Input argument initialization (&quot;arginit&quot; typemap).</li>
<li>Default arguments (&quot;default&quot; typemap).</li>
<li>Input argument resource management (&quot;freearg&quot; typemap).</li>
</ul>
<p><b> Return value handling</b></p>
<div class="code">
<pre>
<b>int</b> foo(int x, double y, char *s);
</pre>
</div>
<ul>
<li>Function return value conversion (&quot;out&quot; typemap).</li>
<li>Return value resource management (&quot;ret&quot; typemap).</li>
<li>Resource management for newly allocated objects (&quot;newfree&quot; typemap).</li>
</ul>
<p><b> Exception handling</b></p>
<div class="code">
<pre>
<b>int</b> foo(int x, double y, char *s) throw(<b>MemoryError, IndexError</b>);
</pre>
</div>
<ul>
<li>Handling of C++ exception specifications. (&quot;throw&quot; typemap).</li>
</ul>
<p><b> Global variables</b></p>
<div class="code">
<pre>
<b>int foo;</b>
</pre>
</div>
<ul>
<li>Assignment of a global variable. (&quot;varin&quot; typemap).</li>
<li>Reading a global variable. (&quot;varout&quot; typemap).</li>
</ul>
<p><b> Member variables</b></p>
<div class="code">
<pre>
struct Foo {
<b>int x[20]</b>;
};
</pre>
</div>
<ul>
<li>Assignment of data to a class/structure member. (&quot;memberin&quot;
typemap).</li>
</ul>
<p><b> Constant creation</b></p>
<div class="code">
<pre>
#define FOO 3
%constant int BAR = 42;
enum { ALE, LAGER, STOUT };
</pre>
</div>
<ul>
<li>Creation of constant values. (&quot;consttab&quot; or &quot;constcode&quot; typemap).</li>
</ul>
<p> Details of each of these typemaps will be covered shortly. Also,
certain language modules may define additional typemaps that expand
upon this list. For example, the Java module defines a variety of
typemaps for controlling additional aspects of the Java bindings.
Consult language specific documentation for further details.</p>
<h3><a name="Typemaps_nn8">14.1.6 What can't be done with typemaps?</a></h3>
<p> Typemaps can't be used to define properties that apply to C/C++
declarations as a whole. For example, suppose you had a declaration
like this,</p>
<div class="code">
<pre>
Foo *make_Foo(int n);
</pre>
</div>
<p> and you wanted to tell SWIG that <tt>make_Foo(int n)</tt> returned a
newly allocated object (for the purposes of providing better memory
management). Clearly, this property of <tt>make_Foo(int n)</tt> is<em>
not</em> a property that would be associated with the datatype <tt>Foo
*</tt> by itself. Therefore, a completely different SWIG customization
mechanism (<tt>%feature</tt>) is used for this purpose. Consult the <a href="#Customization">
Customization Features</a> chapter for more information about that.</p>
<p> Typemaps also can't be used to rearrange or transform the order of
arguments. For example, if you had a function like this:</p>
<div class="code">
<pre>
void foo(int, char *);
</pre>
</div>
<p> you can't use typemaps to interchange the arguments, allowing you to
call the function like this:</p>
<div class="targetlang">
<pre>
foo(&quot;hello&quot;, 3) # Reversed arguments
</pre>
</div>
<p> If you want to change the calling conventions of a function, write a
helper function instead. For example:</p>
<div class="code">
<pre>
%rename(foo) wrap_foo;
%inline %{
void wrap_foo(char *s, int x) {
foo(x, s);
}
%}
</pre>
</div>
<h3><a name="Typemaps_aspects">14.1.7 Similarities to Aspect Oriented
Programming</a></h3>
<p> SWIG has parallels to <a href="https://en.wikipedia.org/wiki/Aspect-oriented_programming">
Aspect Oriented Software Development (AOP)</a>. The <a href="https://en.wikipedia.org/wiki/Aspect-oriented_programming#Terminology">
AOP terminology</a> with respect to SWIG typemaps can be viewed as
follows:</p>
<ul>
<li><b> Cross-cutting concerns</b>: The cross-cutting concerns are the
modularization of the functionality that the typemaps implement, which
is primarily marshalling of types from/to the target language and
C/C++.</li>
<li><b> Advice</b>: The typemap body contains code which is executed
whenever the marshalling is required.</li>
<li><b> Pointcut</b>: The pointcuts are the positions in the wrapper
code that the typemap code is generated into.</li>
<li><b> Aspect</b>: Aspects are the combination of the pointcut and the
advice, hence each typemap is an aspect.</li>
</ul>
<p> SWIG can also be viewed as has having a second set of aspects based
around <a href="#Customization">%feature</a>. Features such as <tt>
%exception</tt> are also cross-cutting concerns as they encapsulate code
that can be used to add logging or exception handling to any function.</p>
<h3><a name="Typemaps_nn9">14.1.8 The rest of this chapter</a></h3>
<p> The rest of this chapter provides detailed information for people
who want to write new typemaps. This information is of particular
importance to anyone who intends to write a new SWIG target language
module. Power users can also use this information to write application
specific type conversion rules.</p>
<p> Since typemaps are strongly tied to the underlying C++ type system,
subsequent sections assume that you are reasonably familiar with the
basic details of values, pointers, references, arrays, type qualifiers
(e.g., <tt>const</tt>), structures, namespaces, templates, and memory
management in C/C++. If not, you would be well-advised to consult a
copy of &quot;The C Programming Language&quot; by Kernighan and Ritchie or &quot;The
C++ Programming Language&quot; by Stroustrup before going any further.</p>
<h2><a name="Typemaps_nn10">14.2 Typemap specifications</a></h2>
<p> This section describes the behavior of the <tt>%typemap</tt>
directive itself.</p>
<h3><a name="Typemaps_defining">14.2.1 Defining a typemap</a></h3>
<p> New typemaps are defined using the <tt>%typemap</tt> declaration.
The general form of this declaration is as follows (parts enclosed in [
... ] are optional):</p>
<div class="code">
<pre>
%typemap(<em>method</em> [, <em>modifiers</em>]) <em>typelist</em> <em>code</em> ;
</pre>
</div>
<p><em> method</em> is a simply a name that specifies what kind of
typemap is being defined. It is usually a name like <tt>&quot;in&quot;</tt>, <tt>
&quot;out&quot;</tt>, or <tt>&quot;argout&quot;</tt>. The purpose of these methods is
described later.</p>
<p><em> modifiers</em> is an optional comma separated list of <tt>
name=&quot;value&quot;</tt> values. These are sometimes to attach extra
information to a typemap and is often target-language dependent. They
are also known as typemap attributes.</p>
<p><em> typelist</em> is a list of the C++ type patterns that the
typemap will match. The general form of this list is as follows:</p>
<div class="diagram">
<pre>
typelist : typepattern [, typepattern, typepattern, ... ] ;
typepattern : type [ (parms) ]
| type name [ (parms) ]
| ( typelist ) [ (parms) ]
</pre>
</div>
<p> Each type pattern is either a simple type, a simple type and
argument name, or a list of types in the case of multi-argument
typemaps. In addition, each type pattern can be parameterized with a
list of temporary variables (parms). The purpose of these variables
will be explained shortly.</p>
<p><em>code</em> specifies the code used in the typemap. Usually this is
C/C++ code, but in the statically typed target languages, such as Java
and C#, this can contain target language code for certain typemaps. It
can take any one of the following forms:</p>
<div class="diagram">
<pre>
code : { ... }
| &quot; ... &quot;
| %{ ... %}
</pre>
</div>
<p> Note that the preprocessor will expand code within the {}
delimiters, but not in the last two styles of delimiters, see <a href="#Preprocessor_delimiters">
Preprocessor and Typemaps</a>. Here are some examples of valid typemap
specifications:</p>
<div class="code">
<pre>
/* Simple typemap declarations */
%typemap(in) int {
$1 = PyInt_AsLong($input);
}
%typemap(in) int &quot;$1 = PyInt_AsLong($input);&quot;
%typemap(in) int %{
$1 = PyInt_AsLong($input);
%}
/* Typemap with extra argument name */
%typemap(in) int nonnegative {
...
}
/* Multiple types in one typemap */
%typemap(in) int, short, long {
$1 = SvIV($input);
}
/* Typemap with modifiers */
%typemap(in, doc=&quot;integer&quot;) int &quot;$1 = scm_to_int($input);&quot;
/* Typemap applied to patterns of multiple arguments */
%typemap(in) (char *str, int len),
(char *buffer, int size)
{
$1 = PyString_AsString($input);
$2 = PyString_Size($input);
}
/* Typemap with extra pattern parameters */
%typemap(in, numinputs=0) int *output (int temp),
long *output (long temp)
{
$1 = &amp;temp;
}
</pre>
</div>
<p> Admittedly, it's not the most readable syntax at first glance.
However, the purpose of the individual pieces will become clear.</p>
<h3><a name="Typemaps_nn12">14.2.2 Typemap scope</a></h3>
<p> Once defined, a typemap remains in effect for all of the
declarations that follow. A typemap may be redefined for different
sections of an input file. For example:</p>
<div class="code">
<pre>
// typemap1
%typemap(in) int {
...
}
int fact(int); // typemap1
int gcd(int x, int y); // typemap1
// typemap2
%typemap(in) int {
...
}
int isprime(int); // typemap2
</pre>
</div>
<p> One exception to the typemap scoping rules pertains to the <tt>
%extend</tt> declaration. <tt>%extend</tt> is used to attach new
declarations to a class or structure definition. Because of this, all
of the declarations in an <tt>%extend</tt> block are subject to the
typemap rules that are in effect at the point where the class itself is
defined. For example:</p>
<div class="code">
<pre>
class Foo {
...
};
%typemap(in) int {
...
}
%extend Foo {
int blah(int x); // typemap has no effect. Declaration is attached to Foo which
// appears before the %typemap declaration.
};
</pre>
</div>
<h3><a name="Typemaps_nn13">14.2.3 Copying a typemap</a></h3>
<p> A typemap is copied by using assignment. For example:</p>
<div class="code">
<pre>
%typemap(in) Integer = int;
</pre>
</div>
<p> or this:</p>
<div class="code">
<pre>
%typemap(in) Integer, Number, int32_t = int;
</pre>
</div>
<p> Types are often managed by a collection of different typemaps. For
example:</p>
<div class="code">
<pre>
%typemap(in) int { ... }
%typemap(out) int { ... }
%typemap(varin) int { ... }
%typemap(varout) int { ... }
</pre>
</div>
<p> To copy all of these typemaps to a new type, use <tt>%apply</tt>.
For example:</p>
<div class="code">
<pre>
%apply int { Integer }; // Copy all int typemaps to Integer
%apply int { Integer, Number }; // Copy all int typemaps to both Integer and Number
</pre>
</div>
<p> The patterns for <tt>%apply</tt> follow the same rules as for <tt>
%typemap</tt>. For example:</p>
<div class="code">
<pre>
%apply int *output { Integer *output }; // Typemap with name
%apply (char *buf, int len) { (char *buffer, int size) }; // Multiple arguments
</pre>
</div>
<h3><a name="Typemaps_nn14">14.2.4 Deleting a typemap</a></h3>
<p> A particular typemap can be deleted / cleared by simply defining no
code. For example:</p>
<div class="code">
<pre>
%typemap(in) int; // Clears the &quot;in&quot; typemap for int
%typemap(in) int, long, short; // Clears the &quot;in&quot; typemap for int, long, short
%typemap(in) int *output;
</pre>
</div>
<p> The above syntax deletes a typemap for just one typemap method - the
&quot;in&quot; method in each of the examples above. The <tt>%clear</tt>
directive is more powerful and will delete / clear a family of
typemaps, that is, all the typemap methods for a given type. For
example:</p>
<div class="code">
<pre>
%clear int; // Delete all typemaps (&quot;in&quot;, &quot;out&quot;, &quot;varin&quot;, ...) for int
%clear int *output, long *output;
</pre>
</div>
<p><b> Note:</b> Since SWIG's default behavior is defined by typemaps,
clearing a fundamental type like <tt>int</tt> will make that type
unusable unless you also define a new family of typemaps immediately
after the clear operation.</p>
<h3><a name="Typemaps_nn15">14.2.5 Placement of typemaps</a></h3>
<p> Typemap declarations can be declared in the global scope, within a
C++ namespace, and within a C++ class. For example:</p>
<div class="code">
<pre>
%typemap(in) int {
...
}
namespace std {
class string;
%typemap(in) string {
...
}
}
class Bar {
public:
typedef const int &amp; const_reference;
%typemap(out) const_reference {
...
}
};
</pre>
</div>
<p> When a typemap appears inside a namespace or class, it stays in
effect until the end of the SWIG input (just like before). However, the
typemap takes the local scope into account. Therefore, this code</p>
<div class="code">
<pre>
namespace std {
class string;
%typemap(in) string {
...
}
}
</pre>
</div>
<p> is really defining a typemap for the type <tt>std::string</tt>. You
could have code like this:</p>
<div class="code">
<pre>
namespace std {
class string;
%typemap(in) string { /* std::string */
...
}
}
namespace Foo {
class string;
%typemap(in) string { /* Foo::string */
...
}
}
</pre>
</div>
<p> In this case, there are two completely distinct typemaps that apply
to two completely different types (<tt>std::string</tt> and <tt>
Foo::string</tt>).</p>
<p> It should be noted that for scoping to work, SWIG has to know that <tt>
string</tt> is a typename defined within a particular namespace. In this
example, this is done using the forward class declaration <tt>class
string</tt>.</p>
<h2><a name="Typemaps_pattern_matching">14.3 Pattern matching rules</a></h2>
<p> The section describes the pattern matching rules by which C/C++
datatypes are associated with typemaps. The matching rules can be
observed in practice by using the debugging options also described.</p>
<h3><a name="Typemaps_nn17">14.3.1 Basic matching rules</a></h3>
<p> Typemaps are matched using both a type and a name (typically the
name of an argument, but in the case of <tt>out</tt> typemaps, the name
of a function, qualified by the class name if it's a class method). For
a given <tt>TYPE NAME</tt> pair, the following rules are applied, in
order, to find a match. The first typemap found is used.</p>
<ul>
<li>Typemaps that exactly match <tt>TYPE</tt> and <tt>NAME</tt>.</li>
<li>Typemaps that exactly match <tt>TYPE</tt> only.</li>
<li>If <tt>TYPE</tt> is a C++ template of type <tt>T&lt; TPARMS &gt;</tt>,
where <tt>TPARMS</tt> are the template parameters, the type is stripped
of the template parameters and the following checks are then made:
<ul>
<li>Typemaps that exactly match <tt>T</tt> and <tt>NAME</tt>.</li>
<li>Typemaps that exactly match <tt>T</tt> only.</li>
</ul>
</li>
</ul>
<p> If <tt>TYPE</tt> includes qualifiers (const, volatile, etc.), each
qualifier is stripped one at a time to form a new stripped type and the
matching rules above are repeated on the stripped type. The left-most
qualifier is stripped first, resulting in the right-most (or top-level)
qualifier being stripped last. For example <tt>int const*const</tt> is
first stripped to <tt>int *const</tt> then <tt>int *</tt>.</p>
<p> If <tt>TYPE</tt> is an array. The following transformation is made:</p>
<ul>
<li>Replace all dimensions to <tt>[ANY]</tt> and look for a generic
array typemap.</li>
</ul>
<p> To illustrate, suppose that you had a function like this:</p>
<div class="code">
<pre>
int foo(const char *s);
</pre>
</div>
<p> To find a typemap for the argument <tt>const char *s</tt>, SWIG will
search for the following typemaps:</p>
<div class="diagram">
<pre>
const char *s Exact type and name match
const char * Exact type match
char *s Type and name match (qualifier stripped)
char * Type match (qualifier stripped)
</pre>
</div>
<p> When more than one typemap rule might be defined, only the first
match found is actually used. Here is an example that shows how some of
the basic rules are applied:</p>
<div class="code">
<pre>
%typemap(in) int *x {
... typemap 1
}
%typemap(in) int * {
... typemap 2
}
%typemap(in) const int *z {
... typemap 3
}
%typemap(in) int [4] {
... typemap 4
}
%typemap(in) int [ANY] {
... typemap 5
}
void A(int *x); // int *x rule (typemap 1)
void B(int *y); // int * rule (typemap 2)
void C(const int *x); // int *x rule (typemap 1)
void D(const int *z); // const int *z rule (typemap 3)
void E(int x[4]); // int [4] rule (typemap 4)
void F(int x[1000]); // int [ANY] rule (typemap 5)
</pre>
</div>
<p><b> Compatibility note:</b> SWIG-2.0.0 introduced stripping the
qualifiers one step at a time. Prior versions stripped all qualifiers
in one step.</p>
<h3><a name="Typemaps_typedef_reductions">14.3.2 Typedef reductions
matching</a></h3>
<p> If no match is found using the rules in the previous section, SWIG
applies a typedef reduction to the type and repeats the typemap search
for the reduced type. To illustrate, suppose you had code like this:</p>
<div class="code">
<pre>
%typemap(in) int {
... typemap 1
}
typedef int Integer;
void blah(Integer x);
</pre>
</div>
<p> To find the typemap for <tt>Integer x</tt>, SWIG will first search
for the following typemaps:</p>
<div class="diagram">
<pre>
Integer x
Integer
</pre>
</div>
<p> Finding no match, it then applies a reduction <tt>Integer -&gt; int</tt>
to the type and repeats the search.</p>
<div class="diagram">
<pre>
int x
int --&gt; match: typemap 1
</pre>
</div>
<p> Even though two types might be the same via typedef, SWIG allows
typemaps to be defined for each typename independently. This allows for
interesting customization possibilities based solely on the typename
itself. For example, you could write code like this:</p>
<div class="code">
<pre>
typedef double pdouble; // Positive double
// typemap 1
%typemap(in) double {
... get a double ...
}
// typemap 2
%typemap(in) pdouble {
... get a positive double ...
}
double sin(double x); // typemap 1
pdouble sqrt(pdouble x); // typemap 2
</pre>
</div>
<p> When reducing the type, only one typedef reduction is applied at a
time. The search process continues to apply reductions until a match is
found or until no more reductions can be made.</p>
<p> For complicated types, the reduction process can generate a long
list of patterns. Consider the following:</p>
<div class="code">
<pre>
typedef int Integer;
typedef Integer Row4[4];
void foo(Row4 rows[10]);
</pre>
</div>
<p> To find a match for the <tt>Row4 rows[10]</tt> argument, SWIG would
check the following patterns, stopping only when it found a match:</p>
<div class="code">
<pre>
Row4 rows[10]
Row4 [10]
Row4 rows[ANY]
Row4 [ANY]
# Reduce Row4 --&gt; Integer[4]
Integer rows[10][4]
Integer [10][4]
Integer rows[ANY][ANY]
Integer [ANY][ANY]
# Reduce Integer --&gt; int
int rows[10][4]
int [10][4]
int rows[ANY][ANY]
int [ANY][ANY]
</pre>
</div>
<p> For parameterized types like templates, the situation is even more
complicated. Suppose you had some declarations like this:</p>
<div class="code">
<pre>
typedef int Integer;
typedef foo&lt;Integer, Integer&gt; fooii;
void blah(fooii *x);
</pre>
</div>
<p> In this case, the following typemap patterns are searched for the
argument <tt>fooii *x</tt>:</p>
<div class="code">
<pre>
fooii *x
fooii *
# Reduce fooii --&gt; foo&lt;Integer, Integer&gt;
foo&lt;Integer, Integer&gt; *x
foo&lt;Integer, Integer&gt; *
# Reduce Integer -&gt; int
foo&lt;int, Integer&gt; *x
foo&lt;int, Integer&gt; *
# Reduce Integer -&gt; int
foo&lt;int, int&gt; *x
foo&lt;int, int&gt; *
</pre>
</div>
<p> Typemap reductions are always applied to the left-most type that
appears. Only when no reductions can be made to the left-most type are
reductions made to other parts of the type. This behavior means that
you could define a typemap for <tt>foo&lt;int, Integer&gt;</tt>, but a
typemap for <tt>foo&lt;Integer, int&gt;</tt> would never be matched.
Admittedly, this is rather esoteric--there's little practical reason to
write a typemap quite like that. Of course, you could rely on this to
confuse your coworkers even more.</p>
<p> As a point of clarification, it is worth emphasizing that typedef
matching is a typedef<b> reduction</b> process only, that is, SWIG does
not search for every single possible typedef. Given a type in a
declaration, it will only reduce the type, it won't build it up looking
for typedefs. For example, given the type <tt>Struct</tt>, the typemap
below will not be used for the <tt>aStruct</tt> parameter, because <tt>
Struct</tt> is fully reduced:</p>
<div class="code">
<pre>
struct Struct {...};
typedef Struct StructTypedef;
%typemap(in) StructTypedef {
...
}
void go(Struct aStruct);
</pre>
</div>
<h3><a name="Typemaps_nn19">14.3.3 Default typemap matching rules</a></h3>
<p> If the basic pattern matching rules result in no match being made,
even after typedef reductions, the default typemap matching rules are
used to look for a suitable typemap match. These rules match a generic
typemap based on the reserved <tt>SWIGTYPE</tt> base type. For example
pointers will use <tt>SWIGTYPE *</tt> and references will use <tt>
SWIGTYPE &amp;</tt>. More precisely, the rules are based on the C++ class
template partial specialization matching rules used by C++ compilers
when looking for an appropriate partial template specialization. This
means that a match is chosen from the most specialized set of generic
typemap types available. For example, when looking for a match to <tt>
int const *</tt>, the rules will prefer to match <tt>SWIGTYPE const *</tt>
if available before matching <tt>SWIGTYPE *</tt>, before matching <tt>
SWIGTYPE</tt>.</p>
<p> Most SWIG language modules use typemaps to define the default
behavior of the C primitive types. This is entirely straightforward.
For example, a set of typemaps for primitives marshalled by value or
const reference are written like this:</p>
<div class="code">
<pre>
%typemap(in) int &quot;... convert to int ...&quot;
%typemap(in) short &quot;... convert to short ...&quot;
%typemap(in) float &quot;... convert to float ...&quot;
...
%typemap(in) const int &amp; &quot;... convert ...&quot;
%typemap(in) const short &amp; &quot;... convert ...&quot;
%typemap(in) const float &amp; &quot;... convert ...&quot;
...
</pre>
</div>
<p> Since typemap matching follows all <tt>typedef</tt> declarations,
any sort of type that is mapped to a primitive type by value or const
reference through <tt>typedef</tt> will be picked up by one of these
primitive typemaps. Most language modules also define typemaps for char
pointers and char arrays to handle strings, so these non-default types
will also be used in preference as the basic typemap matching rules
provide a better match than the default typemap matching rules.</p>
<p> Below is a list of the typical default types supplied by language
modules, showing what the &quot;in&quot; typemap would look like:</p>
<div class="code">
<pre>
%typemap(in) SWIGTYPE &amp; { ... default reference handling ... }
%typemap(in) SWIGTYPE * { ... default pointer handling ... }
%typemap(in) SWIGTYPE *const { ... default pointer const handling ... }
%typemap(in) SWIGTYPE *const&amp; { ... default pointer const reference handling ... }
%typemap(in) SWIGTYPE[ANY] { ... 1D fixed size arrays handling ... }
%typemap(in) SWIGTYPE [] { ... unknown sized array handling ... }
%typemap(in) enum SWIGTYPE { ... default handling for enum values ... }
%typemap(in) const enum SWIGTYPE &amp; { ... default handling for const enum reference values ... }
%typemap(in) SWIGTYPE (CLASS::*) { ... default pointer member handling ... }
%typemap(in) SWIGTYPE { ... simple default handling ... }
</pre>
</div>
<p> If you wanted to change SWIG's default handling for simple pointers,
you would simply redefine the rule for <tt>SWIGTYPE *</tt>. Note, the
simple default typemap rule is used to match against simple types that
don't match any other rules:</p>
<div class="code">
<pre>
%typemap(in) SWIGTYPE { ... simple default handling ... }
</pre>
</div>
<p> This typemap is important because it is the rule that gets triggered
when call or return by value is used. For instance, if you have a
declaration like this:</p>
<div class="code">
<pre>
double dot_product(Vector a, Vector b);
</pre>
</div>
<p> The <tt>Vector</tt> type will usually just get matched against <tt>
SWIGTYPE</tt>. The default implementation of <tt>SWIGTYPE</tt> is to
convert the value into pointers (<a href="#SWIG_nn22">as described in
this earlier section</a>).</p>
<p> By redefining <tt>SWIGTYPE</tt> it may be possible to implement
other behavior. For example, if you cleared all typemaps for <tt>
SWIGTYPE</tt>, SWIG simply won't wrap any unknown datatype (which might
be useful for debugging). Alternatively, you might modify SWIGTYPE to
marshal objects into strings instead of converting them to pointers.</p>
<p> Let's consider an example where the following typemaps are defined
and SWIG is looking for the best match for the enum shown below:</p>
<div class="code">
<pre>
%typemap(in) const Hello &amp; { ... }
%typemap(in) const enum SWIGTYPE &amp; { ... }
%typemap(in) enum SWIGTYPE &amp; { ... }
%typemap(in) SWIGTYPE &amp; { ... }
%typemap(in) SWIGTYPE { ... }
enum Hello {};
const Hello &amp;hi;
</pre>
</div>
<p> The typemap at the top of the list will be chosen, not because it is
defined first, but because it is the closest match for the type being
wrapped. If any of the typemaps in the above list were not defined,
then the next one on the list would have precedence.</p>
<p> The best way to explore the default typemaps is to look at the ones
already defined for a particular language module. Typemap definitions
are usually found in the SWIG library in a file such as <tt>java.swg</tt>
, <tt>csharp.swg</tt> etc. However, for many of the target languages the
typemaps are hidden behind complicated macros, so the best way to view
the default typemaps, or any typemaps for that matter, is to look at
the preprocessed output by running <tt>swig -E</tt> on any interface
file. Finally the best way to view the typemap matching rules in action
is via the <a href="#Typemaps_debugging_search">debugging typemap
pattern matching</a> options covered later on.</p>
<p><b> Compatibility note:</b> The default typemap matching rules were
modified in SWIG-2.0.0 from a slightly simpler scheme to match the
current C++ class template partial specialization matching rules.</p>
<h3><a name="Typemaps_multi_argument_typemaps_patterns">14.3.4
Multi-arguments typemaps</a></h3>
<p> When multi-argument typemaps are specified, they take precedence
over any typemaps specified for a single type. For example:</p>
<div class="code">
<pre>
%typemap(in) (char *buffer, int len) {
// typemap 1
}
%typemap(in) char *buffer {
// typemap 2
}
void foo(char *buffer, int len, int count); // (char *buffer, int len)
void bar(char *buffer, int blah); // char *buffer
</pre>
</div>
<p> Multi-argument typemaps are also more restrictive in the way that
they are matched. Currently, the first argument follows the matching
rules described in the previous section, but all subsequent arguments
must match exactly.</p>
<h3><a name="Typemaps_matching_template_comparison">14.3.5 Matching
rules compared to C++ templates</a></h3>
<p> For those intimately familiar with C++ templates, a comparison of
the typemap matching rules and template type deduction is interesting.
The two areas considered are firstly the default typemaps and their
similarities to partial template specialization and secondly,
non-default typemaps and their similarities to full template
specialization.</p>
<p> For default (SWIGTYPE) typemaps the rules are inspired by C++ class
template partial specialization. For example, given partial
specialization for <tt>T const&amp;</tt> :</p>
<div class="code">
<pre>
template &lt;typename T&gt; struct X { void a(); };
template &lt;typename T&gt; struct X&lt; T const&amp; &gt; { void b(); };
</pre>
</div>
<p> The full (unspecialized) template is matched with most types, such
as:</p>
<div class="code">
<pre>
X&lt; int &amp; &gt; x1; x1.a();
</pre>
</div>
<p> and the following all match the <tt>T const&amp;</tt> partial
specialization:</p>
<div class="code">
<pre>
X&lt; int *const&amp; &gt; x2; x2.b();
X&lt; int const*const&amp; &gt; x3; x3.b();
X&lt; int const&amp; &gt; x4; x4.b();
</pre>
</div>
<p> Now, given just these two default typemaps, where T is analogous to
SWIGTYPE:</p>
<div class="code">
<pre>
%typemap(...) SWIGTYPE { ... }
%typemap(...) SWIGTYPE const&amp; { ... }
</pre>
</div>
<p> The generic default typemap <tt>SWIGTYPE</tt> is used with most
types, such as</p>
<div class="code">
<pre>
int &amp;
</pre>
</div>
<p> and the following all match the <tt>SWIGTYPE const&amp;</tt> typemap,
just like the partial template matching:</p>
<div class="code">
<pre>
int *const&amp;
int const*const&amp;
int const&amp;
</pre>
</div>
<p> Note that the template and typemap matching rules are not identical
for all default typemaps though, for example, with arrays.</p>
<p> For non-default typemaps, one might expect SWIG to follow the fully
specialized template rules. This is nearly the case, but not quite.
Consider a very similar example to the earlier partially specialized
template but this time there is a fully specialized template:</p>
<div class="code">
<pre>
template &lt;typename T&gt; struct Y { void a(); };
template &lt;&gt; struct Y&lt; int const &amp; &gt; { void b(); };
</pre>
</div>
<p> Only the one type matches the specialized template exactly:</p>
<div class="code">
<pre>
Y&lt; int &amp; &gt; y1; y1.a();
Y&lt; int *const&amp; &gt; y2; y2.a();
Y&lt; int const *const&amp; &gt; y3; y3.a();
Y&lt; int const&amp; &gt; y4; y4.b(); // fully specialized match
</pre>
</div>
<p> Given typemaps with the same types used for the template declared
above, where T is again analogous to SWIGTYPE:</p>
<div class="code">
<pre>
%typemap(...) SWIGTYPE { ... }
%typemap(...) int const&amp; { ... }
</pre>
</div>
<p> The comparison between non-default typemaps and fully specialized
single parameter templates turns out to be the same, as just the one
type will match the non-default typemap:</p>
<div class="code">
<pre>
int &amp;
int *const&amp;
int const*const&amp;
int const&amp; // matches non-default typemap int const&amp;
</pre>
</div>
<p> However, if a non-const type is used instead:</p>
<div class="code">
<pre>
%typemap(...) SWIGTYPE { ... }
%typemap(...) int &amp; { ... }
</pre>
</div>
<p> then there is a clear difference to template matching as both the
const and non-const types match the typemap:</p>
<div class="code">
<pre>
int &amp; // matches non-default typemap int &amp;
int *const&amp;
int const*const&amp;
int const&amp; // matches non-default typemap int &amp;
</pre>
</div>
<p> There are other subtle differences such as typedef handling, but at
least it should be clear that the typemap matching rules are similar to
those for specialized template handling.</p>
<h3><a name="Typemaps_debugging_search">14.3.6 Debugging typemap pattern
matching</a></h3>
<p> There are two useful debug command line options available for
debugging typemaps, <tt>-debug-tmsearch</tt> and <tt>-debug-tmused</tt>
.</p>
<p> The <tt>-debug-tmsearch</tt> option is a verbose option for
debugging typemap searches. This can be very useful for watching the
pattern matching process in action and for debugging which typemaps are
used. The option displays all the typemaps and types that are looked
for until a successful pattern match is made. As the display includes
searches for each and every type needed for wrapping, the amount of
information displayed can be large. Normally you would manually search
through the displayed information for the particular type that you are
interested in.</p>
<p> For example, consider some of the code used in the <a href="#Typemaps_typedef_reductions">
Typedef reductions</a> section already covered:</p>
<div class="code">
<pre>
typedef int Integer;
typedef Integer Row4[4];
void foo(Row4 rows[10]);
</pre>
</div>
<p> A sample of the debugging output is shown below for the &quot;in&quot;
typemap:</p>
<div class="shell">
<pre>
swig -perl -debug-tmsearch example.i
...
example.h:3: Searching for a suitable 'in' typemap for: Row4 rows[10]
Looking for: Row4 rows[10]
Looking for: Row4 [10]
Looking for: Row4 rows[ANY]
Looking for: Row4 [ANY]
Looking for: Integer rows[10][4]
Looking for: Integer [10][4]
Looking for: Integer rows[ANY][ANY]
Looking for: Integer [ANY][ANY]
Looking for: int rows[10][4]
Looking for: int [10][4]
Looking for: int rows[ANY][ANY]
Looking for: int [ANY][ANY]
Looking for: SWIGTYPE rows[ANY][ANY]
Looking for: SWIGTYPE [ANY][ANY]
Looking for: SWIGTYPE rows[ANY][]
Looking for: SWIGTYPE [ANY][]
Looking for: SWIGTYPE *rows[ANY]
Looking for: SWIGTYPE *[ANY]
Looking for: SWIGTYPE rows[ANY]
Looking for: SWIGTYPE [ANY]
Looking for: SWIGTYPE rows[]
Looking for: SWIGTYPE []
Using: %typemap(in) SWIGTYPE []
...
</pre>
</div>
<p> showing that the best default match supplied by SWIG is the <tt>
SWIGTYPE []</tt> typemap. As the example shows, the successful match
displays the used typemap source including typemap method, type and
optional name in one of these simplified formats:</p>
<ul>
<li> <tt>Using: %typemap(method) type name</tt></li>
<li> <tt>Using: %typemap(method) type name = type2 name2</tt></li>
<li> <tt>Using: %apply type2 name2 { type name }</tt></li>
</ul>
<p> This information might meet your debugging needs, however, you might
want to analyze further. If you next invoke SWIG with the <tt>-E</tt>
option to display the preprocessed output, and search for the
particular typemap used, you'll find the full typemap contents (example
shown below for Python):</p>
<div class="code">
<pre>
%typemap(in, noblock=1) SWIGTYPE [] (void *argp = 0, int res = 0) {
res = SWIG_ConvertPtr($input, &amp;argp, $descriptor, $disown | 0 );
if (!SWIG_IsOK(res)) {
SWIG_exception_fail(SWIG_ArgError(res), &quot;in method '&quot; &quot;$symname&quot; &quot;', argument &quot;
&quot;$argnum&quot;&quot; of type '&quot; &quot;$type&quot;&quot;'&quot;);
}
$1 = ($ltype)(argp);
}
</pre>
</div>
<p> The generated code for the <tt>foo</tt> wrapper will then contain
the snippets of the typemap with the special variables expanded. The
rest of this chapter will need reading though to fully understand all
of this, however, the relevant parts of the generated code for the
above typemap can be seen below:</p>
<div class="code">
<pre>
SWIGINTERN PyObject *_wrap_foo(PyObject *SWIGUNUSEDPARM(self), PyObject *args) {
...
void *argp1 = 0 ;
int res1 = 0 ;
...
res1 = SWIG_ConvertPtr(obj0, &amp;argp1, SWIGTYPE_p_a_4__int, 0 | 0 );
if (!SWIG_IsOK(res1)) {
SWIG_exception_fail(SWIG_ArgError(res1), &quot;in method '&quot; &quot;foo&quot; &quot;', argument &quot;
&quot;1&quot;&quot; of type '&quot; &quot;int [10][4]&quot;&quot;'&quot;);
}
arg1 = (int (*)[4])(argp1);
...
}
</pre>
</div>
<p> Searches for multi-argument typemaps are not mentioned unless a
matching multi-argument typemap does actually exist. For example, the
output for the code in the <a href="#Typemaps_multi_argument_typemaps_patterns">
earlier multi-arguments section</a> is as follows:</p>
<div class="shell">
<pre>
...
example.h:39: Searching for a suitable 'in' typemap for: char *buffer
Looking for: char *buffer
Multi-argument typemap found...
Using: %typemap(in) (char *buffer, int len)
...
</pre>
</div>
<p> The second option for debugging is <tt>-debug-tmused</tt> and this
displays the typemaps used. This option is a less verbose version of
the <tt>-debug-tmsearch</tt> option as it only displays each
successfully found typemap on a separate single line. The output
displays the type, and name if present, the typemap method in brackets
and then the actual typemap used in the same simplified format output
by the <tt>-debug-tmsearch</tt> option. Below is the output for the
example code at the start of this section on debugging.</p>
<div class="shell">
<pre>
$ swig -perl -debug-tmused example.i
example.h:3: Typemap for Row4 rows[10] (in) : %typemap(in) SWIGTYPE []
example.h:3: Typemap for Row4 rows[10] (typecheck) : %typemap(typecheck) SWIGTYPE *
example.h:3: Typemap for Row4 rows[10] (freearg) : %typemap(freearg) SWIGTYPE []
example.h:3: Typemap for void foo (out) : %typemap(out) void
</pre>
</div>
<p> Now, consider the following interface file:</p>
<div class="code">
<pre>
%module example
%{
void set_value(const char* val) {}
%}
%typemap(check) char *NON_NULL {
if (!$1) {
/* ... error handling ... */
}
}
// use default pointer handling instead of strings
%apply SWIGTYPE * { const char* val, const char* another_value }
%typemap(check) const char* val = char* NON_NULL;
%typemap(arginit, noblock=1) const char* val {
$1 = &quot;&quot;;
}
void set_value(const char* val);
</pre>
</div>
<p> and the output debug:</p>
<div class="shell">
<pre>
swig -perl5 -debug-tmused example.i
example.i:21: Typemap for char const *val (arginit) : %typemap(arginit) char const *val
example.i:21: Typemap for char const *val (in) : %apply SWIGTYPE * { char const *val }
example.i:21: Typemap for char const *val (typecheck) : %apply SWIGTYPE * { char const *val }
example.i:21: Typemap for char const *val (check) : %typemap(check) char const *val = char *NON_NULL
example.i:21: Typemap for char const *val (freearg) : %apply SWIGTYPE * { char const *val }
example.i:21: Typemap for void set_value (out) : %typemap(out) void
</pre>
</div>
<p> The following observations about what is displayed can be noted (the
same applies for <tt>-debug-tmsearch</tt>):</p>
<ul>
<li> The relevant typemap is shown, but for typemap copying, the
appropriate <tt>%typemap</tt> or <tt>%apply</tt> is displayed, for
example, the &quot;check&quot; and &quot;in&quot; typemaps.</li>
<li> The typemap modifiers are not shown, eg the <tt>noblock=1</tt>
modifier in the &quot;arginit&quot; typemap.</li>
<li> The exact <tt>%apply</tt> statement might look different to what is
in the actual code. For example, the <tt>const char* another_value</tt>
is not shown as it is not relevant here. Also the types may be
displayed slightly differently - <tt>char const *</tt> and not <tt>
const char*</tt>.</li>
</ul>
<h2><a name="Typemaps_nn21">14.4 Code generation rules</a></h2>
<p> This section describes rules by which typemap code is inserted into
the generated wrapper code.</p>
<h3><a name="Typemaps_nn22">14.4.1 Scope</a></h3>
<p> When a typemap is defined like this:</p>
<div class="code">
<pre>
%typemap(in) int {
$1 = PyInt_AsLong($input);
}
</pre>
</div>
<p> the typemap code is inserted into the wrapper function using a new
block scope. In other words, the wrapper code will look like this:</p>
<div class="code">
<pre>
wrap_whatever() {
...
// Typemap code
{
arg1 = PyInt_AsLong(obj1);
}
...
}
</pre>
</div>
<p> Because the typemap code is enclosed in its own block, it is legal
to declare temporary variables for use during typemap execution. For
example:</p>
<div class="code">
<pre>
%typemap(in) short {
long temp; /* Temporary value */
if (Tcl_GetLongFromObj(interp, $input, &amp;temp) != TCL_OK) {
return TCL_ERROR;
}
$1 = (short) temp;
}
</pre>
</div>
<p> Of course, any variables that you declare inside a typemap are
destroyed as soon as the typemap code has executed (they are not
visible to other parts of the wrapper function or other typemaps that
might use the same variable names).</p>
<p> Occasionally, typemap code will be specified using a few alternative
forms. For example:</p>
<div class="code">
<pre>
%typemap(in) int &quot;$1 = PyInt_AsLong($input);&quot;
%typemap(in) int %{
$1 = PyInt_AsLong($input);
%}
%typemap(in, noblock=1) int {
$1 = PyInt_AsLong($input);
}
</pre>
</div>
<p> These three forms are mainly used for cosmetics--the specified code
is not enclosed inside a block scope when it is emitted. This sometimes
results in a less complicated looking wrapper function. Note that only
the third of the three typemaps have the typemap code passed through
the SWIG preprocessor.</p>
<h3><a name="Typemaps_nn23">14.4.2 Declaring new local variables</a></h3>
<p> Sometimes it is useful to declare a new local variable that exists
within the scope of the entire wrapper function. A good example of this
might be an application in which you wanted to marshal strings. Suppose
you had a C++ function like this</p>
<div class="code">
<pre>
int foo(std::string *s);
</pre>
</div>
<p> and you wanted to pass a native string in the target language as an
argument. For instance, in Perl, you wanted the function to work like
this:</p>
<div class="targetlang">
<pre>
$x = foo(&quot;Hello World&quot;);
</pre>
</div>
<p> To do this, you can't just pass a raw Perl string as the <tt>
std::string *</tt> argument. Instead, you have to create a temporary <tt>
std::string</tt> object, copy the Perl string data into it, and then
pass a pointer to the object. To do this, simply specify the typemap
with an extra parameter like this:</p>
<div class="code">
<pre>
%typemap(in) std::string * <b>(std::string temp)</b> {
unsigned int len;
char *s;
s = SvPV($input, len); /* Extract string data */
temp.assign(s, len); /* Assign to temp */
$1 = &amp;temp; /* Set argument to point to temp */
}
</pre>
</div>
<p> In this case, <tt>temp</tt> becomes a local variable in the scope of
the entire wrapper function. For example:</p>
<div class="code">
<pre>
wrap_foo() {
std::string temp; &lt;--- Declaration of temp goes here
...
/* Typemap code */
{
...
temp.assign(s, len);
...
}
...
}
</pre>
</div>
<p> When you set <tt>temp</tt> to a value, it persists for the duration
of the wrapper function and gets cleaned up automatically on exit.</p>
<p> It is perfectly safe to use more than one typemap involving local
variables in the same declaration. For example, you could declare a
function as :</p>
<div class="code">
<pre>
void foo(std::string *x, std::string *y, std::string *z);
</pre>
</div>
<p> This is safely handled because SWIG actually renames all local
variable references by appending an argument number suffix. Therefore,
the generated code would actually look like this:</p>
<div class="code">
<pre>
wrap_foo() {
int *arg1; /* Actual arguments */
int *arg2;
int *arg3;
std::string temp1; /* Locals declared in the typemap */
std::string temp2;
std::string temp3;
...
{
char *s;
unsigned int len;
...
temp1.assign(s, len);
arg1 = *temp1;
}
{
char *s;
unsigned int len;
...
temp2.assign(s, len);
arg2 = &amp;temp2;
}
{
char *s;
unsigned int len;
...
temp3.assign(s, len);
arg3 = &amp;temp3;
}
...
}
</pre>
</div>
<p>There is an exception: if the variable name starts with the <tt>
_global_</tt> prefix, the argument number is not appended. Such
variables can be used throughout the generated wrapper function. For
example, the above typemap could be rewritten to use <tt>_global_temp</tt>
instead of <tt>temp</tt> and the generated code would then contain a
single <tt>_global_temp</tt> variable instead of <tt>temp1</tt>, <tt>
temp2</tt> and <tt>temp3</tt>:</p>
<div class="code">
<pre>
%typemap(in) std::string * <b>(std::string _global_temp)</b> {
... as above ...
}
</pre>
</div>
<p> Some typemaps do not recognize local variables (or they may simply
not apply). At this time, only typemaps that apply to argument
conversion support this (input typemaps such as the &quot;in&quot; typemap).</p>
<p><b> Note:</b></p>
<p> When declaring a typemap for multiple types, each type must have its
own local variable declaration.</p>
<div class="code">
<pre>
%typemap(in) const std::string *, std::string * (std::string temp) // NO!
// only std::string * has a local variable
// const std::string * does not (oops)
....
%typemap(in) const std::string * (std::string temp), std::string * (std::string temp) // Correct
....
</pre>
</div>
<h3><a name="Typemaps_special_variables">14.4.3 Special variables</a></h3>
<p> Special variables are prefixed with a <tt>$</tt> and are expanded by
SWIG as part of the code generation process.</p>
<h4><a name="Typemaps_type_special_variables">14.4.3.1 Type related
special variables</a></h4>
<p> Within all typemaps, the following special variables are expanded.
These are related in some way to a typemap's type. This is by no means
a complete list as some target languages have additional special
variables which are documented in the language specific chapters.</p>
<center>
<table border="1" summary="Typemap type related special variables">
<tr><th>Variable</th><th>Meaning</th></tr>
<tr><td>$<em>n</em></td><td> A C local variable corresponding to type<em>
n</em> in the typemap pattern.</td></tr>
<tr><td>$argnum</td><td>Argument number. Only available in typemaps
related to argument conversion</td></tr>
<tr><td>$<em>n</em>_name</td><td>Argument name</td></tr>
<tr><td>$<em>n</em>_type</td><td>Real C datatype of type<em> n</em>.</td>
</tr>
<tr><td>$<em>n</em>_ltype</td><td>ltype of type<em> n</em></td></tr>
<tr><td>$<em>n</em>_mangle</td><td>Mangled form of type<em> n</em>. For
example <tt>_p_Foo</tt></td></tr>
<tr><td>$<em>n</em>_descriptor</td><td>Type descriptor structure for
type<em> n</em>. For example <tt>SWIGTYPE_p_Foo</tt>. This is primarily
used when interacting with the run-time type checker (described later).</td>
</tr>
<tr><td>$*<em>n</em>_type</td><td>Real C datatype of type<em> n</em>
with one pointer removed.</td></tr>
<tr><td>$*<em>n</em>_ltype</td><td>ltype of type<em> n</em> with one
pointer removed.</td></tr>
<tr><td>$*<em>n</em>_mangle</td><td>Mangled form of type<em> n</em> with
one pointer removed.</td></tr>
<tr><td>$*<em>n</em>_descriptor</td><td>Type descriptor structure for
type<em> n</em> with one pointer removed.</td></tr>
<tr><td>$&amp;<em>n</em>_type</td><td>Real C datatype of type<em> n</em>
with one pointer added.</td></tr>
<tr><td>$&amp;<em>n</em>_ltype</td><td>ltype of type<em> n</em> with one
pointer added.</td></tr>
<tr><td>$&amp;<em>n</em>_mangle</td><td>Mangled form of type<em> n</em> with
one pointer added.</td></tr>
<tr><td>$&amp;<em>n</em>_descriptor</td><td>Type descriptor structure for
type<em> n</em> with one pointer added.</td></tr>
<tr><td>$<em>n</em>_basetype</td><td>Base typename with all pointers and
qualifiers stripped.</td></tr>
</table>
</center>
<p> Within the table, $<em>n</em> refers to a specific type within the
typemap specification. For example, if you write this</p>
<div class="code">
<pre>
%typemap(in) int *INPUT {
}
</pre>
</div>
<p> then $1 refers to <tt>int *INPUT</tt>. If you have a typemap like
this,</p>
<div class="code">
<pre>
%typemap(in) (int argc, char *argv[]) {
...
}
</pre>
</div>
<p> then $1 refers to <tt>int argc</tt> and $2 refers to <tt>char
*argv[]</tt>.</p>
<p> Substitutions related to types and names always fill in values from
the actual code that was matched. This is useful when a typemap might
match multiple C datatype. For example:</p>
<div class="code">
<pre>
%typemap(in) int, short, long {
$1 = ($1_ltype) PyInt_AsLong($input);
}
</pre>
</div>
<p> In this case, <tt>$1_ltype</tt> is replaced with the datatype that
is actually matched.</p>
<p> When typemap code is emitted, the C/C++ datatype of the special
variables <tt>$1</tt> and <tt>$2</tt> is always an &quot;ltype.&quot; An &quot;ltype&quot;
is simply a type that can legally appear on the left-hand side of a C
assignment operation. Here are a few examples of types and ltypes:</p>
<div class="diagram">
<pre>
type ltype
------ ----------------
int int
const int int
const int * int *
int [4] int *
int [4][5] int (*)[5]
</pre>
</div>
<p> In most cases a ltype is simply the C datatype with qualifiers
stripped off. In addition, arrays are converted into pointers.</p>
<p> Variables such as <tt>$&amp;1_type</tt> and <tt>$*1_type</tt> are used
to safely modify the type by removing or adding pointers. Although not
needed in most typemaps, these substitutions are sometimes needed to
properly work with typemaps that convert values between pointers and
values.</p>
<p> If necessary, type related substitutions can also be used when
declaring locals. For example:</p>
<div class="code">
<pre>
%typemap(in) int * ($*1_type temp) {
temp = PyInt_AsLong($input);
$1 = &amp;temp;
}
</pre>
</div>
<p> There is one word of caution about declaring local variables in this
manner. If you declare a local variable using a type substitution such
as <tt>$1_ltype temp</tt>, it won't work like you expect for arrays and
certain kinds of pointers. For example, if you wrote this,</p>
<div class="code">
<pre>
%typemap(in) int [10][20] {
$1_ltype temp;
}
</pre>
</div>
<p> then the declaration of <tt>temp</tt> will be expanded as</p>
<div class="code">
<pre>
int (*)[20] temp;
</pre>
</div>
<p> This is illegal C syntax and won't compile. There is currently no
straightforward way to work around this problem in SWIG due to the way
that typemap code is expanded and processed. However, one possible
workaround is to simply pick an alternative type such as <tt>void *</tt>
and use casts to get the correct type when needed. For example:</p>
<div class="code">
<pre>
%typemap(in) int [10][20] {
void *temp;
...
(($1_ltype) temp)[i][j] = x; /* set a value */
...
}
</pre>
</div>
<p> Another approach, which only works for arrays is to use the <tt>
$1_basetype</tt> substitution. For example:</p>
<div class="code">
<pre>
%typemap(in) int [10][20] {
$1_basetype temp[10][20];
...
temp[i][j] = x; /* set a value */
...
}
</pre>
</div>
<h4><a name="Typemaps_non_type_special_variables">14.4.3.2 Non-type
related special variables</a></h4>
<p> The following are additional special variables. These are not
specifically related to a typemap's type.</p>
<center>
<table border="1" summary="Non-type related special variables">
<tr><th>Variable</th><th>Meaning</th></tr>
<tr><td>$<em>isvoid</em></td><td> Expands to 1 if the wrapped function
has a void return, otherwise expands to 0. Note that it expands to 1 in
destructors and 0 in constructors.</td></tr>
<tr><td>$<em>symname</em></td><td> Name of function/method being
wrapped.</td></tr>
</table>
</center>
<h3><a name="Typemaps_special_variable_macros">14.4.4 Special variable
macros</a></h3>
<p> Special variable macros are like macro functions in that they take
one or more input arguments which are used for the macro expansion.
They look like macro/function calls but use the special variable <tt>$</tt>
prefix to the macro name. Note that unlike normal macros, the expansion
is not done by the preprocessor, it is done during the SWIG
parsing/compilation stages. The following special variable macros are
available across all language modules.</p>
<h4><a name="Typemaps_special_macro_descriptor">14.4.4.1
$descriptor(type)</a></h4>
<p> This macro expands into the type descriptor structure for any C/C++
type specified in <tt>type</tt>. It behaves like the <tt>$1_descriptor</tt>
special variable described above except that the type to expand is
taken from the macro argument rather than inferred from the typemap
type. For example, <tt>$descriptor(std::vector&lt;int&gt; *)</tt> will expand
into <tt>SWIGTYPE_p_std__vectorT_int_t</tt>. This macro is mostly used
in the scripting target languages and is demonstrated later in the <a href="#Typemaps_runtime_type_checker_usage">
Run-time type checker usage</a> section.</p>
<h4><a name="Typemaps_special_macro_typemap">14.4.4.2 $typemap(method,
typepattern)</a></h4>
<p> This macro uses the <a href="#Typemaps_pattern_matching">pattern
matching rules</a> described earlier to lookup and then substitute the
special variable macro with the code in the matched typemap. The
typemap to search for is specified by the arguments, where <tt>method</tt>
is the typemap method name and <tt>typepattern</tt> is a type pattern
as per the <tt>%typemap</tt> specification in the <a href="#Typemaps_defining">
Defining a typemap</a> section.</p>
<p> The special variables within the matched typemap are expanded into
those for the matched typemap type, not the typemap within which the
macro is called. In practice, there is little use for this macro in the
scripting target languages. It is mostly used in the target languages
that are statically typed as a way to obtain the target language type
given the C/C++ type and more commonly only when the C++ type is a
template parameter.</p>
<p> The example below is for C# only and uses some typemap method names
documented in the C# chapter, but it shows some of the possible syntax
variations.</p>
<div class="code">
<pre>
%typemap(cstype) unsigned long &quot;uint&quot;
%typemap(cstype) unsigned long bb &quot;bool&quot;
%typemap(cscode) BarClass %{
void foo($typemap(cstype, unsigned long aa) var1,
$typemap(cstype, unsigned long bb) var2,
$typemap(cstype, (unsigned long bb)) var3,
$typemap(cstype, unsigned long) var4)
{
// do something
}
%}
</pre>
</div>
<p> The result is the following expansion</p>
<div class="code">
<pre>
%typemap(cstype) unsigned long &quot;uint&quot;
%typemap(cstype) unsigned long bb &quot;bool&quot;
%typemap(cscode) BarClass %{
void foo(uint var1,
bool var2,
bool var3,
uint var4)
{
// do something
}
%}
</pre>
</div>
<h4><a name="Typemaps_special_macro_typemap_attribute">14.4.4.3
$typemap(method:attribute, typepattern)</a></h4>
<p> An enhanced version of <tt>$typemap</tt> provides access to typemap
attributes by appending a colon and the attribute name after the method
name. In the example below, &quot;cstype&quot; is the typemap method and &quot;out&quot; is
the typemap attribute.</p>
<div class="code">
<pre>
%typemap(cstype, out=&quot;object&quot;) XClass &quot;XClass&quot;
%typemap(cscode) BarClass %{
$typemap(cstype:out, XClass) bar()
{
return null;
}
</pre>
</div>
<p> which expands to</p>
<div class="code">
<pre>
object bar()
{
return null;
}
</pre>
</div>
<p><b> Compatibility note:</b> Support for typemap attributes in <tt>
$typemap</tt> was introduced in SWIG-4.1.0.</p>
<h3><a name="Typemaps_special_variable_attributes">14.4.5 Special
variables and typemap attributes</a></h3>
<p> As of SWIG-3.0.7 typemap attributes will also expand special
variables and special variable macros.</p>
<p> Example usage showing the expansion in the 'out' attribute (C#
specific) as well as the main typemap body:</p>
<div class="code">
<pre>
%typemap(ctype, out=&quot;$*1_ltype&quot;) unsigned int&amp; &quot;$*1_ltype&quot;
</pre>
</div>
<p> is equivalent to the following as <tt>$*1_ltype</tt> expands to <tt>
unsigned int</tt>:</p>
<div class="code">
<pre>
%typemap(ctype, out=&quot;unsigned int&quot;) unsigned int&amp; &quot;unsigned int&quot;
</pre>
</div>
<h3><a name="Typemaps_special_variables_and_macros">14.4.6 Special
variables combined with special variable macros</a></h3>
<p> Special variables can also be used within special variable macros.
The special variables are expanded before they are used in the special
variable macros.</p>
<p> Consider the following C# typemaps:</p>
<div class="code">
<pre>
%typemap(cstype) unsigned int &quot;uint&quot;
%typemap(cstype, out=&quot;$typemap(cstype, $*1_ltype)&quot;) unsigned int&amp; &quot;$typemap(cstype, $*1_ltype)&quot;
</pre>
</div>
<p> Special variables are expanded first and hence the above is
equivalent to:</p>
<div class="code">
<pre>
%typemap(cstype) unsigned int &quot;uint&quot;
%typemap(cstype, out=&quot;$typemap(cstype, unsigned int)&quot;) unsigned int&amp; &quot;$typemap(cstype, unsigned int)&quot;
</pre>
</div>
<p> which then expands to:</p>
<div class="code">
<pre>
%typemap(cstype) unsigned int &quot;uint&quot;
%typemap(cstype, out=&quot;uint&quot;) unsigned int&amp; &quot;uint&quot;
</pre>
</div>
<h2><a name="Typemaps_nn25">14.5 Common typemap methods</a></h2>
<p> The family of typemaps recognized by a language module may vary.
However, the following typemap methods are nearly universal:</p>
<h3><a name="Typemaps_nn26">14.5.1 &quot;in&quot; typemap</a></h3>
<p> The &quot;in&quot; typemap is used to convert function arguments from the
target language to C. For example:</p>
<div class="code">
<pre>
%typemap(in) int {
$1 = PyInt_AsLong($input);
}
</pre>
</div>
<p> The following additional special variables are available:</p>
<div class="code">
<pre>
$input - Input object holding value to be converted.
</pre>
</div>
<p> This is probably the most commonly redefined typemap because it can
be used to implement customized conversions.</p>
<p> In addition, the &quot;in&quot; typemap allows the number of converted
arguments to be specified. The <tt>numinputs</tt> attributes
facilitates this. For example:</p>
<div class="code">
<pre>
// Ignored argument.
%typemap(in, numinputs=0) int *out (int temp) {
$1 = &amp;temp;
}
</pre>
</div>
<p> At this time, only zero or one arguments may be converted. When <tt>
numinputs</tt> is set to 0, the argument is effectively ignored and
cannot be supplied from the target language. The argument is still
required when making the C/C++ call and the above typemap shows the
value used is instead obtained from a locally declared variable called <tt>
temp</tt>. Usually <tt>numinputs</tt> is not specified, whereupon the
default value is 1, that is, there is a one to one mapping of the
number of arguments when used from the target language to the C/C++
call. <a href="#Typemaps_multi_argument_typemaps">Multi-argument
typemaps</a> provide a similar concept where the number of arguments
mapped from the target language to C/C++ can be changed for multiple
adjacent C/C++ arguments.</p>
<h3><a name="Typemaps_nn27">14.5.2 &quot;typecheck&quot; typemap</a></h3>
<p> The &quot;typecheck&quot; typemap is used to support overloaded functions and
methods. It merely checks an argument to see whether or not it matches
a specific type. For example:</p>
<div class="code">
<pre>
%typemap(typecheck, precedence=SWIG_TYPECHECK_INTEGER) int {
$1 = PyInt_Check($input) ? 1 : 0;
}
</pre>
</div>
<p> For typechecking, the $1 variable is always a simple integer that is
set to 1 or 0 depending on whether or not the input argument is the
correct type. Set to 1 if the input argument is the correct type
otherwise set to 0.</p>
<p> If you define new &quot;in&quot; typemaps<em> and</em> your program uses
overloaded methods, you should also define a collection of &quot;typecheck&quot;
typemaps. More details about this follow in the <a href="#Typemaps_overloading">
Typemaps and overloading</a> section.</p>
<h3><a name="Typemaps_nn28">14.5.3 &quot;out&quot; typemap</a></h3>
<p> The &quot;out&quot; typemap is used to convert function/method return values
from C into the target language. For example:</p>
<div class="code">
<pre>
%typemap(out) int {
$result = PyInt_FromLong($1);
}
</pre>
</div>
<p> The following additional special variables are available.</p>
<div class="code">
<pre>
$result - Result object returned to target language.
</pre>
</div>
<p> The &quot;out&quot; typemap supports an optional attribute flag called
&quot;optimal&quot;. This is for code optimisation and is detailed in the <a href="#Typemaps_optimal">
Optimal code generation when returning by value</a> section.</p>
<h3><a name="Typemaps_nn29">14.5.4 &quot;arginit&quot; typemap</a></h3>
<p> The &quot;arginit&quot; typemap is used to set the initial value of a function
argument--before any conversion has occurred. This is not normally
necessary, but might be useful in highly specialized applications. For
example:</p>
<div class="code">
<pre>
// Set argument to NULL before any conversion occurs
%typemap(arginit) int *data {
$1 = NULL;
}
</pre>
</div>
<h3><a name="Typemaps_nn30">14.5.5 &quot;default&quot; typemap</a></h3>
<p> The &quot;default&quot; typemap is used to turn an argument into a default
argument. For example:</p>
<div class="code">
<pre>
%typemap(default) int flags {
$1 = DEFAULT_FLAGS;
}
...
int foo(int x, int y, int flags);
</pre>
</div>
<p> The primary use of this typemap is to either change the wrapping of
default arguments or specify a default argument in a language where
they aren't supported (like C). Target languages that do not support
optional arguments, such as Java and C#, effectively ignore the value
specified by this typemap as all arguments must be given.</p>
<p> Once a default typemap has been applied to an argument, all
arguments that follow must have default values. See the <a href="#SWIG_default_args">
Default/optional arguments</a> section for further information on
default argument wrapping.</p>
<h3><a name="Typemaps_nn31">14.5.6 &quot;check&quot; typemap</a></h3>
<p> The &quot;check&quot; typemap is used to supply value checking code during
argument conversion. The typemap is applied<em> after</em> arguments
have been converted. For example:</p>
<div class="code">
<pre>
%typemap(check) int positive {
if ($1 &lt;= 0) {
SWIG_exception(SWIG_ValueError, &quot;Expected positive value.&quot;);
}
}
</pre>
</div>
<h3><a name="Typemaps_nn32">14.5.7 &quot;argout&quot; typemap</a></h3>
<p> The &quot;argout&quot; typemap is used to return values from arguments. This
is most commonly used to write wrappers for C/C++ functions that need
to return multiple values. The &quot;argout&quot; typemap is almost always
combined with an &quot;in&quot; typemap---possibly to ignore the input value. For
example:</p>
<div class="code">
<pre>
/* Set the input argument to point to a temporary variable */
%typemap(in, numinputs=0) int *out (int temp) {
$1 = &amp;temp;
}
%typemap(argout) int *out {
// Append output value $1 to $result
...
}
</pre>
</div>
<p> The following additional special variables are available.</p>
<div class="diagram">
<pre>
$result - Result object returned to target language.
$input - The original input object passed.
</pre>
</div>
<p> The code supplied to the &quot;argout&quot; typemap is always placed after the
&quot;out&quot; typemap. If multiple return values are used, the extra return
values are often appended to return value of the function.</p>
<p> See the <tt>typemaps.i</tt> library file for examples.</p>
<h3><a name="Typemaps_nn33">14.5.8 &quot;freearg&quot; typemap</a></h3>
<p> The &quot;freearg&quot; typemap is used to cleanup argument data. It is only
used when an argument might have allocated resources that need to be
cleaned up when the wrapper function exits. The &quot;freearg&quot; typemap
usually cleans up argument resources allocated by the &quot;in&quot; typemap. For
example:</p>
<div class="code">
<pre>
// Get a list of integers
%typemap(in) int *items {
int nitems = Length($input);
$1 = (int *) malloc(sizeof(int)*nitems);
}
// Free the list
%typemap(freearg) int *items {
free($1);
}
</pre>
</div>
<p> The &quot;freearg&quot; typemap inserted at the end of the wrapper function,
just before control is returned back to the target language. This code
is also placed into a special variable <tt>$cleanup</tt> that may be
used in other typemaps whenever a wrapper function needs to abort
prematurely.</p>
<h3><a name="Typemaps_nn34">14.5.9 &quot;newfree&quot; typemap</a></h3>
<p> The &quot;newfree&quot; typemap is used in conjunction with the <tt>%newobject</tt>
directive and is used to deallocate memory used by the return result of
a function. For example:</p>
<div class="code">
<pre>
%typemap(newfree) string * {
delete $1;
}
%typemap(out) string * {
$result = PyString_FromString($1-&gt;c_str());
}
...
%newobject foo;
...
string *foo();
</pre>
</div>
<p> See <a href="#Customization_ownership">Object ownership and
%newobject</a> for further details.</p>
<h3><a name="Typemaps_ret">14.5.10 &quot;ret&quot; typemap</a></h3>
<p> The &quot;ret&quot; typemap is not used very often, but can be useful for
anything associated with the return type, such as resource management,
return value error checking, etc. Usually this can all be done in the
&quot;out&quot; typemap, but sometimes it is handy to use the &quot;out&quot; typemap code
untouched and add to the generated code using the code in the &quot;ret&quot;
typemap. One such case is memory clean up. For example, a <tt>
stringheap_t</tt> type is defined indicating that the returned memory
must be deleted and a <tt>string_t</tt> type is defined indicating that
the returned memory must not be deleted.</p>
<div class="code">
<pre>
%typemap(ret) stringheap_t %{
free($1);
%}
typedef char * string_t;
typedef char * stringheap_t;
string_t MakeString1();
stringheap_t MakeString2();
</pre>
</div>
<p> The &quot;ret&quot; typemap above will only be used for <tt>MakeString2</tt>,
but both functions will use the default &quot;out&quot; typemap for <tt>char *</tt>
provided by SWIG. The code above would ensure the appropriate memory is
freed in all target languages as the need to provide custom &quot;out&quot;
typemaps (which involve target language specific code) is not
necessary.</p>
<p> This approach is an alternative to using the &quot;newfree&quot; typemap and <tt>
%newobject</tt> as there is no need to list all the functions that
require the memory cleanup, it is purely done on types.</p>
<h3><a name="Typemaps_nn35">14.5.11 &quot;memberin&quot; typemap</a></h3>
<p> The &quot;memberin&quot; typemap is used to copy data from<em> an already
converted input value</em> into a structure member. It is typically
used to handle array members and other special cases. For example:</p>
<div class="code">
<pre>
%typemap(memberin) int [4] {
memmove($1, $input, 4*sizeof(int));
}
</pre>
</div>
<p> It is rarely necessary to write &quot;memberin&quot; typemaps---SWIG already
provides a default implementation for arrays, strings, and other
objects.</p>
<h3><a name="Typemaps_nn36">14.5.12 &quot;varin&quot; typemap</a></h3>
<p> The &quot;varin&quot; typemap is used to convert objects in the target
language to C for the purposes of assigning to a C/C++ global variable.
This is implementation specific.</p>
<h3><a name="Typemaps_nn37">14.5.13 &quot;varout&quot; typemap</a></h3>
<p> The &quot;varout&quot; typemap is used to convert a C/C++ object to an object
in the target language when reading a C/C++ global variable. This is
implementation specific.</p>
<h3><a name="Typemaps_throws_typemap">14.5.14 &quot;throws&quot; typemap</a></h3>
<p> The &quot;throws&quot; typemap is only used when SWIG parses a C++ method with
an exception specification or has the <tt>%catches</tt> feature
attached to the method (see <a href="#SWIGPlus_catches">Exception
handling with %catches</a>). It provides a default mechanism for
handling C++ methods that have declared the exceptions they will throw.
The purpose of this typemap is to convert a C++ exception into an error
or exception in the target language. It is slightly different to the
other typemaps as it is based around the exception type rather than the
type of a parameter or variable. For example:</p>
<div class="code">
<pre>
%typemap(throws) const char * %{
PyErr_SetString(PyExc_RuntimeError, $1);
SWIG_fail;
%}
// Either an exception specification on the method
void bar() throw (const char *);
// Or a %catches feature attached to the method
%catches(const char *) bar();
void bar();
</pre>
</div>
<p> As can be seen from the resulting generated code below, SWIG
generates an exception handler when wrapping the <tt>bar</tt> function
with the catch block comprising the &quot;throws&quot; typemap content.</p>
<div class="code">
<pre>
...
try {
bar();
} catch(char const *_e) {
PyErr_SetString(PyExc_RuntimeError, _e);
SWIG_fail;
}
...
</pre>
</div>
<p> Note that if your methods do not have an exception specification but
they do throw exceptions and you are not using <tt>%catches</tt>, SWIG
cannot know how to deal with them. Please also see the <a href="#Customization_exception">
Exception handling with %exception</a> section for another way to handle
exceptions.</p>
<h2><a name="Typemaps_nn39">14.6 Some typemap examples</a></h2>
<p> This section contains a few examples. Consult language module
documentation for more examples.</p>
<h3><a name="Typemaps_nn40">14.6.1 Typemaps for arrays</a></h3>
<p> A common use of typemaps is to provide support for C arrays
appearing both as arguments to functions and as structure members.</p>
<p> For example, suppose you had a function like this:</p>
<div class="code">
<pre>
void set_vector(int type, float value[4]);
</pre>
</div>
<p> If you wanted to handle <tt>float value[4]</tt> as a list of floats,
you might write a typemap similar to this:</p>
<div class="code">
<pre>
%typemap(in) float value[4] (float temp[4]) {
int i;
if (!PySequence_Check($input)) {
PyErr_SetString(PyExc_ValueError, &quot;Expected a sequence&quot;);
SWIG_fail;
}
if (PySequence_Length($input) != 4) {
PyErr_SetString(PyExc_ValueError, &quot;Size mismatch. Expected 4 elements&quot;);
SWIG_fail;
}
for (i = 0; i &lt; 4; i++) {
PyObject *o = PySequence_GetItem($input, i);
if (PyNumber_Check(o)) {
temp[i] = (float) PyFloat_AsDouble(o);
} else {
PyErr_SetString(PyExc_ValueError, &quot;Sequence elements must be numbers&quot;);
SWIG_fail;
}
}
$1 = temp;
}
</pre>
</div>
<p> In this example, the variable <tt>temp</tt> allocates a small array
on the C stack. The typemap then populates this array and passes it to
the underlying C function.</p>
<p> When used from Python, the typemap allows the following type of
function call:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; set_vector(type, [ 1, 2.5, 5, 20 ])
</pre>
</div>
<p> If you wanted to generalize the typemap to apply to arrays of all
dimensions you might write this:</p>
<div class="code">
<pre>
%typemap(in) float value[ANY] (float temp[$1_dim0]) {
int i;
if (!PySequence_Check($input)) {
PyErr_SetString(PyExc_ValueError, &quot;Expected a sequence&quot;);
SWIG_fail;
}
if (PySequence_Length($input) != $1_dim0) {
PyErr_SetString(PyExc_ValueError, &quot;Size mismatch. Expected $1_dim0 elements&quot;);
SWIG_fail;
}
for (i = 0; i &lt; $1_dim0; i++) {
PyObject *o = PySequence_GetItem($input, i);
if (PyNumber_Check(o)) {
temp[i] = (float) PyFloat_AsDouble(o);
} else {
PyErr_SetString(PyExc_ValueError, &quot;Sequence elements must be numbers&quot;);
SWIG_fail;
}
}
$1 = temp;
}
</pre>
</div>
<p> In this example, the special variable <tt>$1_dim0</tt> is expanded
with the actual array dimensions. Multidimensional arrays can be
matched in a similar manner. For example:</p>
<div class="code">
<pre>
%typemap(in) float matrix[ANY][ANY] (float temp[$1_dim0][$1_dim1]) {
... convert a 2d array ...
}
</pre>
</div>
<p> For large arrays, it may be impractical to allocate storage on the
stack using a temporary variable as shown. To work with heap allocated
data, the following technique can be used.</p>
<div class="code">
<pre>
%typemap(in) float value[ANY] {
int i;
if (!PySequence_Check($input)) {
PyErr_SetString(PyExc_ValueError, &quot;Expected a sequence&quot;);
SWIG_fail;
}
if (PySequence_Length($input) != $1_dim0) {
PyErr_SetString(PyExc_ValueError, &quot;Size mismatch. Expected $1_dim0 elements&quot;);
SWIG_fail;
}
$1 = (float *) malloc($1_dim0*sizeof(float));
for (i = 0; i &lt; $1_dim0; i++) {
PyObject *o = PySequence_GetItem($input, i);
if (PyNumber_Check(o)) {
$1[i] = (float) PyFloat_AsDouble(o);
} else {
free($1);
PyErr_SetString(PyExc_ValueError, &quot;Sequence elements must be numbers&quot;);
SWIG_fail;
}
}
}
%typemap(freearg) float value[ANY] {
free($1);
}
</pre>
</div>
<p> In this case, an array is allocated using <tt>malloc</tt>. The <tt>
freearg</tt> typemap is then used to release the argument after the
function has been called.</p>
<p> Another common use of array typemaps is to provide support for array
structure members. Due to subtle differences between pointers and
arrays in C, you can't just &quot;assign&quot; to a array structure member.
Instead, you have to explicitly copy elements into the array. For
example, suppose you had a structure like this:</p>
<div class="code">
<pre>
struct SomeObject {
float value[4];
...
};
</pre>
</div>
<p> When SWIG runs, it won't produce any code to set the <tt>vec</tt>
member. You may even get a warning message like this:</p>
<div class="shell">
<pre>
$ swig -python example.i
example.i:10: Warning 462: Unable to set variable of type float [4].
</pre>
</div>
<p> These warning messages indicate that SWIG does not know how you want
to set the <tt>vec</tt> field.</p>
<p> To fix this, you can supply a special &quot;memberin&quot; typemap like this:</p>
<div class="code">
<pre>
%typemap(memberin) float [ANY] {
int i;
for (i = 0; i &lt; $1_dim0; i++) {
$1[i] = $input[i];
}
}
</pre>
</div>
<p> The memberin typemap is used to set a structure member from data
that has already been converted from the target language to C. In this
case, <tt>$input</tt> is the local variable in which converted input
data is stored. This typemap then copies this data into the structure.</p>
<p> When combined with the earlier typemaps for arrays, the combination
of the &quot;in&quot; and &quot;memberin&quot; typemap allows the following usage:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; s = SomeObject()
&gt;&gt;&gt; s.x = [1, 2.5, 5, 10]
</pre>
</div>
<p> Related to structure member input, it may be desirable to return
structure members as a new kind of object. For example, in this
example, you will get very odd program behavior where the structure
member can be set nicely, but reading the member simply returns a
pointer:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; s = SomeObject()
&gt;&gt;&gt; s.x = [1, 2.5, 5, 10]
&gt;&gt;&gt; print s.x
_1008fea8_p_float
&gt;&gt;&gt;
</pre>
</div>
<p> To fix this, you can write an &quot;out&quot; typemap. For example:</p>
<div class="code">
<pre>
%typemap(out) float [ANY] {
int i;
$result = PyList_New($1_dim0);
for (i = 0; i &lt; $1_dim0; i++) {
PyObject *o = PyFloat_FromDouble((double) $1[i]);
PyList_SetItem($result, i, o);
}
}
</pre>
</div>
<p> Now, you will find that member access is quite nice:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; s = SomeObject()
&gt;&gt;&gt; s.x = [1, 2.5, 5, 10]
&gt;&gt;&gt; print s.x
[ 1, 2.5, 5, 10]
</pre>
</div>
<p><b> Compatibility Note:</b> SWIG1.1 used to provide a special
&quot;memberout&quot; typemap. However, it was mostly useless and has since been
eliminated. To return structure members, simply use the &quot;out&quot; typemap.</p>
<h3><a name="Typemaps_nn41">14.6.2 Implementing constraints with
typemaps</a></h3>
<p> One particularly interesting application of typemaps is the
implementation of argument constraints. This can be done with the
&quot;check&quot; typemap. When used, this allows you to provide code for
checking the values of function arguments. For example:</p>
<div class="code">
<pre>
%module math
%typemap(check) double posdouble {
if ($1 &lt; 0) {
croak(&quot;Expecting a positive number&quot;);
}
}
...
double sqrt(double posdouble);
</pre>
</div>
<p> This provides a sanity check to your wrapper function. If a negative
number is passed to this function, a Perl exception will be raised and
your program terminated with an error message.</p>
<p> This kind of checking can be particularly useful when working with
pointers. For example:</p>
<div class="code">
<pre>
%typemap(check) Vector * {
if ($1 == 0) {
PyErr_SetString(PyExc_TypeError, &quot;NULL Pointer not allowed&quot;);
SWIG_fail;
}
}
</pre>
</div>
<p> will prevent any function involving a <tt>Vector *</tt> from
accepting a NULL pointer. As a result, SWIG can often prevent a
potential segmentation faults or other run-time problems by raising an
exception rather than blindly passing values to the underlying C/C++
program.</p>
<h2><a name="Typemaps_nn43">14.7 Typemaps for multiple target languages</a>
</h2>
<p> The code within typemaps is usually language dependent, however,
many target languages support the same typemaps. In order to
distinguish typemaps across different languages, the preprocessor
should be used. For example, the &quot;in&quot; typemap for Perl and Ruby could
be written as:</p>
<div class="code">
<pre>
#if defined(SWIGPERL)
%typemap(in) int &quot;$1 = ($1_ltype) SvIV($input);&quot;
#elif defined(SWIGRUBY)
%typemap(in) int &quot;$1 = NUM2INT($input);&quot;
#else
#warning no &quot;in&quot; typemap defined
#endif
</pre>
</div>
<p> The full set of language specific macros is defined in the <a href="#Preprocessor_condition_compilation">
Conditional Compilation</a> section. The example above also shows a
common approach of issuing a warning for an as yet unsupported
language.</p>
<p><b> Compatibility note:</b> In SWIG-1.1 different languages could be
distinguished with the language name being put within the <tt>%typemap</tt>
directive, but this was deprecated in SWIG 1.3.28 and support finally
dropped completely in SWIG 4.1.0 so you'll need to update any remaining
uses to use the approach above. For example,
<br> <tt>%typemap(ruby, in) int &quot;$1 = NUM2INT($input);&quot;</tt>.</p>
<h2><a name="Typemaps_optimal">14.8 Optimal code generation when
returning by value</a></h2>
<p> The &quot;out&quot; typemap is the main typemap for return types. This typemap
supports an optional attribute flag called &quot;optimal&quot;, which is for
reducing the number of temporary variables and the amount of generated
code, thereby giving the compiler the opportunity to use<i> return
value optimization</i> for generating faster executing code. It only
really makes a difference when returning objects by value and has some
limitations on usage, as explained later on.</p>
<p> When a function returns an object by value, SWIG generates code that
instantiates the default type on the stack then assigns the value
returned by the function call to it. A copy of this object is then made
on the heap and this is what is ultimately stored and used from the
target language. This will be clearer considering an example. Consider
running the following code through SWIG:</p>
<div class="code">
<pre>
%typemap(out) SWIGTYPE %{
$result = new $1_ltype($1);
%}
%inline %{
#include &lt;iostream&gt;
using namespace std;
struct XX {
XX() { cout &lt;&lt; &quot;XX()&quot; &lt;&lt; endl; }
XX(int i) { cout &lt;&lt; &quot;XX(&quot; &lt;&lt; i &lt;&lt; &quot;)&quot; &lt;&lt; endl; }
XX(const XX &amp;other) { cout &lt;&lt; &quot;XX(const XX &amp;)&quot; &lt;&lt; endl; }
XX &amp; operator =(const XX &amp;other) { cout &lt;&lt; &quot;operator=(const XX &amp;)&quot; &lt;&lt; endl; return *this; }
~XX() { cout &lt;&lt; &quot;~XX()&quot; &lt;&lt; endl; }
static XX create() {
return XX(0);
}
};
%}
</pre>
</div>
<p> The &quot;out&quot; typemap shown is the default typemap for C# when returning
objects by value. When making a call to <tt>XX::create()</tt> from C#,
the output is as follows:</p>
<div class="targetlang">
<pre>
XX()
XX(0)
operator=(const XX &amp;)
~XX()
XX(const XX &amp;)
~XX()
~XX()
</pre>
</div>
<p> Note that three objects are being created as well as an assignment.
Wouldn't it be great if the <tt>XX::create()</tt> method was the only
time a constructor was called? As the method returns by value, this is
asking a lot and the code that SWIG generates by default makes it
impossible for the compiler to use<i> return value optimisation (RVO)</i>
. However, this is where the &quot;optimal&quot; attribute in the &quot;out&quot; typemap
can help out. If the typemap code is kept the same and just the
&quot;optimal&quot; attribute specified like this:</p>
<div class="code">
<pre>
%typemap(out, optimal=&quot;1&quot;) SWIGTYPE %{
$result = new $1_ltype($1);
%}
</pre>
</div>
<p> then when the code is run again, the output is simply:</p>
<div class="targetlang">
<pre>
XX(0)
~XX()
</pre>
</div>
<p> How the &quot;optimal&quot; attribute works is best explained using the
generated code. Without &quot;optimal&quot;, the generated code is:</p>
<div class="code">
<pre>
SWIGEXPORT void * SWIGSTDCALL CSharp_XX_create() {
void * jresult ;
XX result;
result = XX::create();
jresult = new XX(result);
return jresult;
}
</pre>
</div>
<p> With the &quot;optimal&quot; attribute, the code is:</p>
<div class="code">
<pre>
SWIGEXPORT void * SWIGSTDCALL CSharp_XX_create() {
void * jresult ;
jresult = new XX(XX::create());
return jresult;
}
</pre>
</div>
<p> The major difference is the <tt>result</tt> temporary variable
holding the value returned from <tt>XX::create()</tt> is no longer
generated and instead the copy constructor call is made directly from
the value returned by <tt>XX::create()</tt>. With modern compilers
implementing RVO, the copy is not actually done, in fact the object is
never created on the stack in <tt>XX::create()</tt> at all, it is
simply created directly on the heap. In the first instance, the <tt>$1</tt>
special variable in the typemap is expanded into <tt>result</tt>. In
the second instance, <tt>$1</tt> is expanded into <tt>XX::create()</tt>
and this is essentially what the &quot;optimal&quot; attribute is telling SWIG to
do.</p>
<p> The &quot;optimal&quot; attribute optimisation is not turned on by default as
it has a number of restrictions. Firstly, some code cannot be condensed
into a simple call for passing into the copy constructor. One common
occurrence is when <a href="#Customization_exception">%exception</a> is
used. Consider adding the following <tt>%exception</tt> to the example:</p>
<div class="code">
<pre>
%exception XX::create() %{
try {
$action
} catch(const std::exception &amp;e) {
cout &lt;&lt; e.what() &lt;&lt; endl;
}
%}
</pre>
</div>
<p> SWIG can detect when the &quot;optimal&quot; attribute cannot be used and will
ignore it and in this case will issue the following warning:</p>
<div class="targetlang">
<pre>
example.i:28: Warning 474: Method XX::create() usage of the optimal attribute ignored
example.i:14: Warning 474: in the out typemap as the following cannot be used to generate
optimal code:
try {
result = XX::create();
} catch(const std::exception &amp;e) {
cout &lt;&lt; e.what() &lt;&lt; endl;
}
</pre>
</div>
<p> It should be clear that the above code cannot be used as the
argument to the copy constructor call, that is, for the <tt>$1</tt>
substitution.</p>
<p> Secondly, if the typemap uses <tt>$1</tt> more than once, then
multiple calls to the wrapped function will be made. Obviously that is
not very optimal. In fact SWIG attempts to detect this and will issue a
warning something like:</p>
<div class="targetlang">
<pre>
example.i:21: Warning 475: Multiple calls to XX::create() might be generated due to
example.i:7: Warning 475: optimal attribute usage in the out typemap.
</pre>
</div>
<p> However, it doesn't always get it right, for example when <tt>$1</tt>
is within some commented out code.</p>
<h2><a name="Typemaps_multi_argument_typemaps">14.9 Multi-argument
typemaps</a></h2>
<p> So far, the typemaps presented have focused on the problem of
dealing with single values. For example, converting a single input
object to a single argument in a function call. However, certain
conversion problems are difficult to handle in this manner. As an
example, consider the example at the very beginning of this chapter:</p>
<div class="code">
<pre>
int foo(int argc, char *argv[]);
</pre>
</div>
<p> Suppose that you wanted to wrap this function so that it accepted a
single list of strings like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; foo([&quot;ale&quot;, &quot;lager&quot;, &quot;stout&quot;])
</pre>
</div>
<p> To do this, you not only need to map a list of strings to <tt>char
*argv[]</tt>, but the value of <tt>int argc</tt> is implicitly
determined by the length of the list. Using only simple typemaps, this
type of conversion is possible, but extremely painful. Multi-argument
typemaps help in this situation.</p>
<p> A multi-argument typemap is a conversion rule that specifies how to
convert a<em> single</em> object in the target language to a set of
consecutive function arguments in C/C++. For example, the following
multi-argument maps perform the conversion described for the above
example:</p>
<div class="code">
<pre>
%typemap(in) (int argc, char *argv[]) {
int i;
if (!PyList_Check($input)) {
PyErr_SetString(PyExc_ValueError, &quot;Expecting a list&quot;);
SWIG_fail;
}
$1 = PyList_Size($input);
$2 = (char **) malloc(($1+1)*sizeof(char *));
for (i = 0; i &lt; $1; i++) {
PyObject *s = PyList_GetItem($input, i);
if (!PyString_Check(s)) {
free($2);
PyErr_SetString(PyExc_ValueError, &quot;List items must be strings&quot;);
SWIG_fail;
}
$2[i] = PyString_AsString(s);
}
$2[i] = 0;
}
%typemap(freearg) (int argc, char *argv[]) {
free($2);
}
/* Required for C++ method overloading */
%typecheck(SWIG_TYPECHECK_STRING_ARRAY) (int argc, char *argv[]) {
$1 = PyList_Check($input) ? 1 : 0;
}
</pre>
</div>
<p> A multi-argument map is always specified by surrounding the
arguments with parentheses as shown. For example:</p>
<div class="code">
<pre>
%typemap(in) (int argc, char *argv[]) { ... }
</pre>
</div>
<p> Within the typemap code, the variables <tt>$1</tt>, <tt>$2</tt>, and
so forth refer to each type in the map. All of the usual substitutions
apply--just use the appropriate <tt>$1</tt> or <tt>$2</tt> prefix on
the variable name (e.g., <tt>$2_type</tt>, <tt>$1_ltype</tt>, etc.)</p>
<p> Multi-argument typemaps always have precedence over simple typemaps
and SWIG always performs longest-match searching. Therefore, you will
get the following behavior:</p>
<div class="code">
<pre>
%typemap(in) int argc { ... typemap 1 ... }
%typemap(in) (int argc, char *argv[]) { ... typemap 2 ... }
%typemap(in) (int argc, char *argv[], char *env[]) { ... typemap 3 ... }
int foo(int argc, char *argv[]); // Uses typemap 2
int bar(int argc, int x); // Uses typemap 1
int spam(int argc, char *argv[], char *env[]); // Uses typemap 3
</pre>
</div>
<p> It should be stressed that multi-argument typemaps can appear
anywhere in a function declaration and can appear more than once. For
example, you could write this:</p>
<div class="code">
<pre>
%typemap(in) (int scount, char *swords[]) { ... }
%typemap(in) (int wcount, char *words[]) { ... }
void search_words(int scount, char *swords[], int wcount, char *words[], int maxcount);
</pre>
</div>
<p> Other directives such as <tt>%apply</tt> and <tt>%clear</tt> also
work with multi-argument maps. For example:</p>
<div class="code">
<pre>
%apply (int argc, char *argv[]) {
(int scount, char *swords[]),
(int wcount, char *words[])
};
...
%clear (int scount, char *swords[]), (int wcount, char *words[]);
...
</pre>
</div>
<p> Don't forget to also provide a suitable <a href="#Typemaps_overloading">
typemap for overloaded functions</a>, such as <tt>%typecheck</tt> shown
for foo above. This is only required if the function is overloaded in
C++.</p>
<p> Although multi-argument typemaps may seem like an exotic, little
used feature, there are several situations where they make sense.
First, suppose you wanted to wrap functions similar to the low-level <tt>
read()</tt> and <tt>write()</tt> system calls. For example:</p>
<div class="code">
<pre>
typedef unsigned int size_t;
int read(int fd, void *rbuffer, size_t len);
int write(int fd, void *wbuffer, size_t len);
</pre>
</div>
<p> As is, the only way to use the functions would be to allocate memory
and pass some kind of pointer as the second argument---a process that
might require the use of a helper function. However, using
multi-argument maps, the functions can be transformed into something
more natural. For example, you might write typemaps like this:</p>
<div class="code">
<pre>
// typemap for an outgoing buffer
%typemap(in) (void *wbuffer, size_t len) {
if (!PyString_Check($input)) {
PyErr_SetString(PyExc_ValueError, &quot;Expecting a string&quot;);
SWIG_fail;
}
$1 = (void *) PyString_AsString($input);
$2 = PyString_Size($input);
}
// typemap for an incoming buffer
%typemap(in) (void *rbuffer, size_t len) {
if (!PyInt_Check($input)) {
PyErr_SetString(PyExc_ValueError, &quot;Expecting an integer&quot;);
SWIG_fail;
}
$2 = PyInt_AsLong($input);
if ($2 &lt; 0) {
PyErr_SetString(PyExc_ValueError, &quot;Positive integer expected&quot;);
SWIG_fail;
}
$1 = (void *) malloc($2);
}
// Return the buffer. Discarding any previous return result
%typemap(argout) (void *rbuffer, size_t len) {
Py_DecRef($result); /* Blow away any previous result */
if (result &lt; 0) { /* Check for I/O error */
free($1);
PyErr_SetFromErrno(PyExc_IOError);
return NULL;
}
$result = PyString_FromStringAndSize($1, result);
free($1);
}
</pre>
</div>
<p> (note: In the above example, <tt>$result</tt> and <tt>result</tt>
are two different variables. <tt>result</tt> is the real C datatype
that was returned by the function. <tt>$result</tt> is the scripting
language object being returned to the interpreter.).</p>
<p> Now, in a script, you can write code that simply passes buffers as
strings like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; f = example.open(&quot;Makefile&quot;)
&gt;&gt;&gt; example.read(f, 40)
'TOP = ../..\nSWIG = $(TOP)/.'
&gt;&gt;&gt; example.read(f, 40)
'./swig\nSRCS = example.c\nTARGET '
&gt;&gt;&gt; example.close(f)
0
&gt;&gt;&gt; g = example.open(&quot;foo&quot;, example.O_WRONLY | example.O_CREAT, 0644)
&gt;&gt;&gt; example.write(g, &quot;Hello world\n&quot;)
12
&gt;&gt;&gt; example.write(g, &quot;This is a test\n&quot;)
15
&gt;&gt;&gt; example.close(g)
0
&gt;&gt;&gt;
</pre>
</div>
<p> A number of multi-argument typemap problems also arise in libraries
that perform matrix-calculations--especially if they are mapped onto
low-level Fortran or C code. For example, you might have a function
like this:</p>
<div class="code">
<pre>
int is_symmetric(double *mat, int rows, int columns);
</pre>
</div>
<p> In this case, you might want to pass some kind of higher-level
object as an matrix. To do this, you could write a multi-argument
typemap like this:</p>
<div class="code">
<pre>
%typemap(in) (double *mat, int rows, int columns) {
MatrixObject *a;
a = GetMatrixFromObject($input); /* Get matrix somehow */
/* Get matrix properties */
$1 = GetPointer(a);
$2 = GetRows(a);
$3 = GetColumns(a);
}
</pre>
</div>
<p> This kind of technique can be used to hook into scripting-language
matrix packages such as Numeric Python. However, it should also be
stressed that some care is in order. For example, when crossing
languages you may need to worry about issues such as row-major vs.
column-major ordering (and perform conversions if needed). Note that
multi-argument typemaps cannot deal with non-consecutive C/C++
arguments; a workaround such as a helper function re-ordering the
arguments to make them consecutive will need to be written.</p>
<h2><a name="Typemaps_warnings">14.10 Typemap warnings</a></h2>
<p> Warnings can be added to typemaps so that SWIG generates a warning
message whenever the typemap is used. See the information in the <a href="#Warnings_nn5">
issuing warnings</a> section.</p>
<h2><a name="Typemaps_fragments">14.11 Typemap fragments</a></h2>
<p> The primary purpose of fragments is to reduce code bloat that
repeated use of typemap code can lead to. Fragments are snippets of
code that can be thought of as code dependencies of a typemap. If a
fragment is used by more than one typemap, then the snippet of code
within the fragment is only generated once. Code bloat is typically
reduced by moving typemap code into a support function and then placing
the support function into a fragment.</p>
<p> For example, if you have a very long typemap</p>
<div class="code">
<pre>
%typemap(in) MyClass * {
MyClass *value = 0;
... many lines of marshalling code ...
$result = value;
}
</pre>
</div>
<p> the same marshalling code is often repeated in several typemaps,
such as &quot;in&quot;, &quot;varin&quot;, &quot;directorout&quot;, etc. SWIG copies the code for
each argument that requires the typemap code, easily leading to code
bloat in the generated code. To eliminate this, define a fragment that
includes the common marshalling code:</p>
<div class="code">
<pre>
%fragment(&quot;AsMyClass&quot;, &quot;header&quot;) {
MyClass *AsMyClass(PyObject *obj) {
MyClass *value = 0;
... many lines of marshalling code ...
return value;
}
}
%typemap(in, fragment=&quot;AsMyClass&quot;) MyClass * {
$result = AsMyClass($input);
}
%typemap(varin, fragment=&quot;AsMyClass&quot;) MyClass * {
$result = AsMyClass($input);
}
</pre>
</div>
<p> When the &quot;in&quot; or &quot;varin&quot; typemaps for MyClass are required, the
contents of the fragment called &quot;AsMyClass&quot; are added to the &quot;header&quot;
section within the generated code, and then the typemap code is
emitted. Hence, the method <tt>AsMyClass</tt> will be generated into
the wrapper code before any typemap code that calls it.</p>
<p> To define a fragment you need a fragment name, a section name for
generating the fragment code into, and the code itself. See <a href="#SWIG_nn42">
Code insertion blocks</a> for a full list of section names. Usually the
section name used is &quot;header&quot;. Different delimiters can be used:</p>
<div class="code">
<pre>
%fragment(&quot;my_name&quot;, &quot;header&quot;) %{ ... %}
%fragment(&quot;my_name&quot;, &quot;header&quot;) { ... }
%fragment(&quot;my_name&quot;, &quot;header&quot;) &quot; ... &quot;
</pre>
</div>
<p> and these follow the usual preprocessing rules mentioned in the <a href="#Preprocessor_delimiters">
Preprocessing delimiters</a> section. The following are some rules and
guidelines for using fragments:</p>
<ol>
<li>
<p> A fragment is added to the wrapping code only once. When using the <tt>
MyClass *</tt> typemaps above and wrapping the method:</p>
<div class="code">
<pre>
void foo(MyClass *a, MyClass *b);
</pre>
</div>
<p> the generated code will look something like:</p>
<div class="code">
<pre>
MyClass *AsMyClass(PyObject *obj) {
...
}
void _wrap_foo(...) {
....
arg1 = AsMyClass(obj1);
arg2 = AsMyClass(obj2);
...
foo(arg1, arg2);
}
</pre>
</div>
<p> even as there is duplicated typemap code to process both <tt>a</tt>
and <tt>b</tt>, the <tt>AsMyClass</tt> method will be defined only
once.</p>
</li>
<li>
<p> A fragment should only be defined once. If there is more than one
definition, the first definition is the one used. All other definitions
are silently ignored. For example, if you have</p>
<div class="code">
<pre>
%fragment(&quot;AsMyClass&quot;, &quot;header&quot;) { ...definition 1... }
....
%fragment(&quot;AsMyClass&quot;, &quot;header&quot;) { ...definition 2... }
</pre>
</div>
<p> only the first definition is used. In this way you can override the
default fragments in a SWIG library by defining your fragment before
the library <tt>%include</tt>. Note that this behavior is the opposite
to typemaps, where the last typemap defined/applied prevails. Fragments
follow the first-in-first-out convention since they are intended to be
global, while typemaps are intended to be locally specialized.</p>
</li>
<li>
<p> Fragment names cannot contain commas.</p>
</li>
<li>
<p> A fragment can use one or more additional fragments, for example:</p>
<div class="code">
<pre>
%fragment(&quot;&lt;limits.h&gt;&quot;, &quot;header&quot;) %{
#include &lt;limits.h&gt;
%}
%fragment(&quot;AsMyClass&quot;, &quot;header&quot;, fragment=&quot;&lt;limits.h&gt;&quot;) {
MyClass *AsMyClass(PyObject *obj) {
MyClass *value = 0;
... some marshalling code ...
if (ival &lt; CHAR_MIN /*defined in &lt;limits.h&gt;*/) {
...
} else {
...
}
...
return value;
}
}
</pre>
</div>
<p> in this case, when the &quot;AsMyClass&quot; fragment is emitted, it also
triggers the inclusion of the &quot;&lt;limits.h&gt;&quot; fragment.</p>
</li>
<li>
<p> A fragment can have dependencies on a number of other fragments, for
example:</p>
<div class="code">
<pre>
%fragment(&quot;bigfragment&quot;, &quot;header&quot;, fragment=&quot;frag1&quot;, fragment=&quot;frag2&quot;, fragment=&quot;frag3&quot;) &quot;&quot;;
</pre>
</div>
<p> When the &quot;bigfragment&quot; is used, the three dependent fragments
&quot;frag1&quot;, &quot;frag2&quot; and &quot;frag3&quot; are also pulled in. Note that as
&quot;bigframent&quot; is empty (the empty string - &quot;&quot;), it does not add any code
itself, but merely triggers the inclusion of the other fragments.</p>
</li>
<li>
<p> A typemap can also use more than one fragment:</p>
<div class="code">
<pre>
%typemap(&quot;in&quot;, fragment=&quot;frag1&quot;, fragment=&quot;frag2&quot;, fragment=&quot;frag3&quot;) {...}
</pre>
</div>
<p><b> Compatibility note:</b> The ability to use multiple <tt>fragment</tt>
keys as shown above was introduced in SWIG-4.1.0.</p>
<p> Multiple fragments can alternatively be specified as a comma
separated list value in a single <tt>fragment</tt> key. Note that no
whitespace is allowed within this comma separated list. The following
is the equivalent to the above:</p>
<div class="code">
<pre>
%typemap(in, fragment=&quot;frag1,frag2,frag3&quot;) {...}
</pre>
</div>
<p> which in turn is functionally equivalent to:</p>
<div class="code">
<pre>
%typemap(in, fragment=&quot;bigfragment&quot;) {...}
</pre>
</div>
<p> when used with the &quot;bigfragment&quot; defined above.</p>
</li>
<li>
<p> Finally, you can force the inclusion of a fragment at any point in
the generated code as follows:</p>
<div class="code">
<pre>
%fragment(&quot;bigfragment&quot;);
</pre>
</div>
<p> which, for example, is very useful inside a template class. Another
useful case is when using <tt>%extend</tt> inside a class where the
additional code in the <tt>%extend</tt> block depends on the contents
of the fragment.</p>
<div class="code">
<pre>
%fragment(&quot;&lt;limits.h&gt;&quot;, &quot;header&quot;) %{
#include &lt;limits.h&gt;
%}
struct X {
...
%extend {
%fragment(&quot;&lt;limits.h&gt;&quot;);
bool check(short val) {
if (val &lt; SHRT_MIN /*defined in &lt;limits.h&gt;*/) {
return true;
} else {
return false;
}
}
}
};
</pre>
</div>
<p> Forced inclusion of fragments can be used as a replacement for <a href="#SWIG_nn42">
code insertion block</a>, ensuring the code block is only generated
once. Consider the contents of FileA.i below which first uses a code
insertion block and then a forced fragment inclusion to generate code:</p>
<div class="code">
<pre>
// FileA.i
%{
#include &lt;stdio.h&gt;
%}
%fragment(&quot;&lt;limits.h&gt;&quot;);
</pre>
</div>
<p> and another file including the above:</p>
<div class="code">
<pre>
// FileB.i
%include &quot;FileA.i&quot;
</pre>
</div>
<p> The resulting code in the wrappers for FileB.i is:</p>
<div class="code">
<pre>
#include &lt;stdio.h&gt;
#include &lt;limits.h&gt;
</pre>
</div>
<p> A note of caution must be mentioned when using <tt>%fragment</tt>
forced inclusion or code insertion blocks with <tt>%import</tt>. If <tt>
%import</tt> is used instead:</p>
<div class="code">
<pre>
// FileC.i
%import &quot;FileA.i&quot;
</pre>
</div>
<p> then nothing is generated in the resulting code in the wrappers for
FileC.i. This is because <tt>%import</tt> is for collecting type
information and does not result in any code being generated, see <a href="#Preprocessor_nn3">
File Imports</a>.</p>
</li>
</ol>
<p> Most readers will probably want to skip the next two sub-sections on
advanced fragment usage unless a desire to really get to grips with
some powerful but tricky macro and fragment usage that is used in parts
of the SWIG typemap library.</p>
<h3><a name="Typemaps_fragment_type_specialization">14.11.1 Fragment
type specialization</a></h3>
<p> Fragments can be<i> type specialized</i>. The syntax is as follows:</p>
<div class="code">
<pre>
%fragment(&quot;name&quot;, &quot;header&quot;) { ...a type independent fragment... }
%fragment(&quot;name&quot;{type}, &quot;header&quot;) { ...a type dependent fragment... }
</pre>
</div>
<p> where <tt>type</tt> is a C/C++ type. Like typemaps, fragments can
also be used inside templates, for example:</p>
<div class="code">
<pre>
template &lt;class T&gt;
struct A {
%fragment(&quot;incode&quot;{A&lt;T&gt;}, &quot;header&quot;) {
... 'incode' specialized fragment ...
}
%typemap(in, fragment=&quot;incode&quot;{A&lt;T&gt;}) {
... here we use the 'type specialized' fragment &quot;incode&quot;{A&lt;T&gt;} ...
}
};
</pre>
</div>
<h3><a name="Typemaps_automatic_specialization">14.11.2 Fragments and
automatic typemap specialization</a></h3>
<p> Since fragments can be type specialized, they can be elegantly used
to specialize typemaps. For example, if you have something like:</p>
<div class="code">
<pre>
%fragment(&quot;incode&quot;{float}, &quot;header&quot;) {
float in_method_float(PyObject *obj) {
...
}
}
%fragment(&quot;incode&quot;{long}, &quot;header&quot;) {
float in_method_long(PyObject *obj) {
...
}
}
// %my_typemaps macro definition
%define %my_typemaps(Type)
%typemap(in, fragment=&quot;incode&quot;{Type}) Type {
value = in_method_##Type(obj);
}
%enddef
%my_typemaps(float);
%my_typemaps(long);
</pre>
</div>
<p> then the proper <tt>&quot;incode&quot;{float}</tt> or <tt>&quot;incode&quot;{long}</tt>
fragment will be used, and the <tt>in_method_float</tt> and <tt>
in_method_long</tt> methods will be called whenever the <tt>float</tt>
or <tt>long</tt> types are used as input parameters.</p>
<p> This feature is used a lot in the typemaps shipped in the SWIG
library for some scripting languages. The interested (or very brave)
reader can take a look at the fragments.swg file shipped with SWIG to
see this in action.</p>
<h2><a name="Typemaps_runtime_type_checker">14.12 The run-time type
checker</a></h2>
<p> Most scripting languages need type information at run-time. This
type information can include how to construct types, how to garbage
collect types, and the inheritance relationships between types. If the
language interface does not provide its own type information storage,
the generated SWIG code needs to provide it.</p>
<p> Requirements for the type system:</p>
<ul>
<li>Store inheritance and type equivalence information and be able to
correctly re-create the type pointer.</li>
<li>Share type information between modules.</li>
<li>Modules can be loaded in any order, regardless of actual type
dependency.</li>
<li>Avoid the use of dynamically allocated memory, and library/system
calls in general.</li>
<li>Provide a reasonably fast implementation, minimizing the lookup time
for all language modules.</li>
<li>Custom, language specific information can be attached to types.</li>
<li>Modules can be unloaded from the type system.</li>
</ul>
<h3><a name="Typemaps_nn45">14.12.1 Implementation</a></h3>
<p> The run-time type checker is used by many, but not all, of SWIG's
supported target languages. The run-time type checker features are not
required and are thus not used for statically typed languages such as
Java and C#. The scripting and scheme based languages rely on it and it
forms a critical part of SWIG's operation for these languages.</p>
<p> When pointers, arrays, and objects are wrapped by SWIG, they are
normally converted into typed pointer objects. For example, an instance
of <tt>Foo *</tt> might be a string encoded like this:</p>
<div class="diagram">
<pre>
_108e688_p_Foo
</pre>
</div>
<p> At a basic level, the type checker simply restores some type-safety
to extension modules. However, the type checker is also responsible for
making sure that wrapped C++ classes are handled correctly---especially
when inheritance is used. This is especially important when an
extension module makes use of multiple inheritance. For example:</p>
<div class="code">
<pre>
class Foo {
public:
int x;
};
class Bar {
public:
int y;
};
class FooBar : public Foo, public Bar {
public:
int z;
};
</pre>
</div>
<p> When the class <tt>FooBar</tt> is organized in memory, it contains
the contents of the classes <tt>Foo</tt> and <tt>Bar</tt> as well as
its own data members. For example:</p>
<div class="diagram">
<pre>
FooBar --&gt; | -----------| &lt;-- Foo
| int x |
|------------| &lt;-- Bar
| int y |
|------------|
| int z |
|------------|
</pre>
</div>
<p> Because of the way that base class data is stacked together, the
casting of a <tt>Foobar *</tt> to either of the base classes may change
the actual value of the pointer. This means that it is generally not
safe to represent pointers using a simple integer or a bare <tt>void *</tt>
---type tags are needed to implement correct handling of pointer values
(and to make adjustments when needed).</p>
<p> In the wrapper code generated for each language, pointers are
handled through the use of special type descriptors and conversion
functions. For example, if you look at the wrapper code for Python, you
will see code similar to the following (simplified for brevity):</p>
<div class="code">
<pre>
if (!SWIG_IsOK(SWIG_ConvertPtr(obj0, (void **) &amp;arg1, SWIGTYPE_p_Foo, 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method 'GrabVal', expecting type Foo&quot;);
}
</pre>
</div>
<p> In this code, <tt>SWIGTYPE_p_Foo</tt> is the type descriptor that
describes <tt>Foo *</tt>. The type descriptor is actually a pointer to
a structure that contains information about the type name to use in the
target language, a list of equivalent typenames (via typedef or
inheritance), and pointer value handling information (if applicable).
The <tt>SWIG_ConvertPtr()</tt> function is simply a utility function
that takes a pointer object in the target language and a
type-descriptor object and uses this information to generate a C++
pointer. The <tt>SWIG_IsOK</tt> macro checks the return value for
errors and <tt>SWIG_exception_fail</tt> can be called to raise an
exception in the target language. However, the exact name and calling
conventions of the conversion function depends on the target language
(see language specific chapters for details).</p>
<p> The actual type code is in swigrun.swg, and gets inserted near the
top of the generated swig wrapper file. The phrase &quot;a type X that can
cast into a type Y&quot; means that given a type X, it can be converted into
a type Y. In other words, X is a derived class of Y or X is a typedef
of Y. The structure to store type information looks like this:</p>
<div class="code">
<pre>
/* Structure to store information on one type */
typedef struct swig_type_info {
const char *name; /* mangled name of this type */
const char *str; /* human readable name for this type */
swig_dycast_func dcast; /* dynamic cast function down a hierarchy */
struct swig_cast_info *cast; /* Linked list of types that can cast into this type */
void *clientdata; /* Language specific type data */
} swig_type_info;
/* Structure to store a type and conversion function used for casting */
typedef struct swig_cast_info {
swig_type_info *type; /* pointer to type that is equivalent to this type */
swig_converter_func converter; /* function to cast the void pointers */
struct swig_cast_info *next; /* pointer to next cast in linked list */
struct swig_cast_info *prev; /* pointer to the previous cast */
} swig_cast_info;
</pre>
</div>
<p> Each <tt>swig_type_info</tt> stores a linked list of types that it
is equivalent to. Each entry in this doubly linked list stores a
pointer back to another swig_type_info structure, along with a pointer
to a conversion function. This conversion function is used to solve the
above problem of the FooBar class, correctly returning a pointer to the
type we want.</p>
<p> The basic problem we need to solve is verifying and building
arguments passed to functions. So going back to the <tt>
SWIG_ConvertPtr()</tt> function example from above, we are expecting a <tt>
Foo *</tt> and need to check if <tt>obj0</tt> is in fact a <tt>Foo *</tt>
. From before, <tt>SWIGTYPE_p_Foo</tt> is just a pointer to the <tt>
swig_type_info</tt> structure describing <tt>Foo *</tt>. So we loop
through the linked list of <tt>swig_cast_info</tt> structures attached
to <tt>SWIGTYPE_p_Foo</tt>. If we see that the type of <tt>obj0</tt> is
in the linked list, we pass the object through the associated
conversion function and then return a positive. If we reach the end of
the linked list without a match, then <tt>obj0</tt> can not be
converted to a <tt>Foo *</tt> and an error is generated.</p>
<p> Another issue needing to be addressed is sharing type information
between multiple modules. More explicitly, we need to have ONE <tt>
swig_type_info</tt> for each type. If two modules both use the type, the
second module loaded must lookup and use the swig_type_info structure
from the module already loaded. Because no dynamic memory is used and
the circular dependencies of the casting information, loading the type
information is somewhat tricky, and not explained here. A complete
description is in the <tt>Lib/swiginit.swg</tt> file (and near the top
of any generated file).</p>
<p> Each module has one swig_module_info structure which looks like
this:</p>
<div class="code">
<pre>
/* Structure used to store module information
* Each module generates one structure like this, and the runtime collects
* all of these structures and stores them in a circularly linked list.*/
typedef struct swig_module_info {
swig_type_info **types; /* Array of pointers to swig_type_info structs in this module */
int size; /* Number of types in this module */
struct swig_module_info *next; /* Pointer to next element in circularly linked list */
swig_type_info **type_initial; /* Array of initially generated type structures */
swig_cast_info **cast_initial; /* Array of initially generated casting structures */
void *clientdata; /* Language specific module data */
} swig_module_info;
</pre>
</div>
<p> Each module stores an array of pointers to <tt>swig_type_info</tt>
structures and the number of types in this module. So when a second
module is loaded, it finds the <tt>swig_module_info</tt> structure for
the first module and searches the array of types. If any of its own
types are in the first module and have already been loaded, it uses
those <tt>swig_type_info</tt> structures rather than creating new ones.
These <tt>swig_module_info</tt> structures are chained together in a
circularly linked list.</p>
<h3><a name="Typemaps_runtime_type_checker_usage">14.12.2 Usage</a></h3>
<p>This section covers how to use these functions from typemaps. To
learn how to call these functions from external files (not the
generated _wrap.c file), see the <a href="#Modules_external_run_time">
External access to the run-time system</a> section.</p>
<p>When pointers are converted in a typemap, the typemap code often
looks similar to this:</p>
<div class="code">
<pre>
%typemap(in) Foo * {
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;$1, $1_descriptor, 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method '$symname', expecting type Foo&quot;);
}
}
</pre>
</div>
<p> The most critical part is the typemap is the use of the <tt>
$1_descriptor</tt> special variable. When placed in a typemap, this is
expanded into the <tt>SWIGTYPE_*</tt> type descriptor object above. As
a general rule, you should always use <tt>$1_descriptor</tt> instead of
trying to hard-code the type descriptor name directly.</p>
<p> There is another reason why you should always use the <tt>
$1_descriptor</tt> variable. When this special variable is expanded,
SWIG marks the corresponding type as &quot;in use.&quot; When type-tables and
type information is emitted in the wrapper file, descriptor information
is only generated for those datatypes that were actually used in the
interface. This greatly reduces the size of the type tables and
improves efficiency.</p>
<p> Occasionally, you might need to write a typemap that needs to
convert pointers of other types. To handle this, the special variable
macro <tt>$descriptor(type)</tt> covered earlier can be used to
generate the SWIG type descriptor name for any C datatype. For example:</p>
<div class="code">
<pre>
%typemap(in) Foo * {
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;$1, $1_descriptor, 0))) {
Bar *temp;
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;temp, $descriptor(Bar *), 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method '$symname', expecting type Foo or Bar&quot;);
}
$1 = (Foo *)temp;
}
}
</pre>
</div>
<p> The primary use of <tt>$descriptor(type)</tt> is when writing
typemaps for container objects and other complex data structures. There
are some restrictions on the argument---namely it must be a fully
defined C datatype. It can not be any of the special typemap variables.</p>
<p> In certain cases, SWIG may not generate type-descriptors like you
expect. For example, if you are converting pointers in some
non-standard way or working with an unusual combination of interface
files and modules, you may find that SWIG omits information for a
specific type descriptor. To fix this, you may need to use the <tt>
%types</tt> directive. For example:</p>
<div class="code">
<pre>
%types(int *, short *, long *, float *, double *);
</pre>
</div>
<p> When <tt>%types</tt> is used, SWIG generates type-descriptor
information even if those datatypes never appear elsewhere in the
interface file.</p>
<p> Further details about the run-time type checking can be found in the
documentation for individual language modules. Reading the source code
may also help. The file <tt>Lib/swigrun.swg</tt> in the SWIG library
contains all of the source of the generated code for type-checking.
This code is also included in every generated wrapped file so you
probably just look at the output of SWIG to get a better sense for how
types are managed.</p>
<h2><a name="Typemaps_overloading">14.13 Typemaps and overloading</a></h2>
<p> This section does not apply to the statically typed languages like
Java and C#, where overloading of the types is handled much like C++ by
generating overloaded methods in the target language. In many of the
other target languages, SWIG still fully supports C++ overloaded
methods and functions. For example, if you have a collection of
functions like this:</p>
<div class="code">
<pre>
int foo(int x);
int foo(double x);
int foo(char *s, int y);
</pre>
</div>
<p> You can access the functions in a normal way from the scripting
interpreter:</p>
<div class="targetlang">
<pre>
# Python
foo(3) # foo(int)
foo(3.5) # foo(double)
foo(&quot;hello&quot;, 5) # foo(char *, int)
# Tcl
foo 3 # foo(int)
foo 3.5 # foo(double)
foo hello 5 # foo(char *, int)
</pre>
</div>
<p> To implement overloading, SWIG generates a separate wrapper function
for each overloaded method. For example, the above functions would
produce something roughly like this:</p>
<div class="code">
<pre>
// wrapper pseudocode
_wrap_foo_0(argc, args[]) { // foo(int)
int arg1;
int result;
...
arg1 = FromInteger(args[0]);
result = foo(arg1);
return ToInteger(result);
}
_wrap_foo_1(argc, args[]) { // foo(double)
double arg1;
int result;
...
arg1 = FromDouble(args[0]);
result = foo(arg1);
return ToInteger(result);
}
_wrap_foo_2(argc, args[]) { // foo(char *, int)
char *arg1;
int arg2;
int result;
...
arg1 = FromString(args[0]);
arg2 = FromInteger(args[1]);
result = foo(arg1, arg2);
return ToInteger(result);
}
</pre>
</div>
<p> Next, a dynamic dispatch function is generated:</p>
<div class="code">
<pre>
_wrap_foo(argc, args[]) {
if (argc == 1) {
if (IsInteger(args[0])) {
return _wrap_foo_0(argc, args);
}
if (IsDouble(args[0])) {
return _wrap_foo_1(argc, args);
}
}
if (argc == 2) {
if (IsString(args[0]) &amp;&amp; IsInteger(args[1])) {
return _wrap_foo_2(argc, args);
}
}
error(&quot;No matching function!\n&quot;);
}
</pre>
</div>
<p> The purpose of the dynamic dispatch function is to select the
appropriate C++ function based on argument types---a task that must be
performed at runtime in most of SWIG's target languages.</p>
<p> The generation of the dynamic dispatch function is a relatively
tricky affair. Not only must input typemaps be taken into account
(these typemaps can radically change the types of arguments accepted),
but overloaded methods must also be sorted and checked in a very
specific order to resolve potential ambiguity. A high-level overview of
this ranking process is found in the &quot;<a href="#SWIGPlus">SWIG and C++</a>
&quot; chapter. What isn't mentioned in that chapter is the mechanism by
which it is implemented---as a collection of typemaps.</p>
<p> To support dynamic dispatch, SWIG first defines a general purpose
type hierarchy as follows:</p>
<div class="diagram">
<pre>
Symbolic Name Precedence Value
------------------------------ ------------------
SWIG_TYPECHECK_POINTER 0
SWIG_TYPECHECK_ITERATOR 5
SWIG_TYPECHECK_VOIDPTR 10
SWIG_TYPECHECK_BOOL 15
SWIG_TYPECHECK_UINT8 20
SWIG_TYPECHECK_INT8 25
SWIG_TYPECHECK_UINT16 30
SWIG_TYPECHECK_INT16 35
SWIG_TYPECHECK_UINT32 40
SWIG_TYPECHECK_INT32 45
SWIG_TYPECHECK_SIZE 47
SWIG_TYPECHECK_PTRDIFF 48
SWIG_TYPECHECK_UINT64 50
SWIG_TYPECHECK_INT64 55
SWIG_TYPECHECK_UINT128 60
SWIG_TYPECHECK_INT128 65
SWIG_TYPECHECK_INTEGER 70
SWIG_TYPECHECK_FLOAT 80
SWIG_TYPECHECK_DOUBLE 90
SWIG_TYPECHECK_CPLXFLT 95
SWIG_TYPECHECK_CPLXDBL 100
SWIG_TYPECHECK_COMPLEX 105
SWIG_TYPECHECK_UNICHAR 110
SWIG_TYPECHECK_STDUNISTRING 115
SWIG_TYPECHECK_UNISTRING 120
SWIG_TYPECHECK_CHAR 130
SWIG_TYPECHECK_STDSTRING 135
SWIG_TYPECHECK_STRING 140
SWIG_TYPECHECK_PAIR 150
SWIG_TYPECHECK_STDARRAY 155
SWIG_TYPECHECK_VECTOR 160
SWIG_TYPECHECK_DEQUE 170
SWIG_TYPECHECK_LIST 180
SWIG_TYPECHECK_SET 190
SWIG_TYPECHECK_MULTISET 200
SWIG_TYPECHECK_MAP 210
SWIG_TYPECHECK_MULTIMAP 220
SWIG_TYPECHECK_STACK 230
SWIG_TYPECHECK_QUEUE 240
SWIG_TYPECHECK_BOOL_ARRAY 1015
SWIG_TYPECHECK_INT8_ARRAY 1025
SWIG_TYPECHECK_INT16_ARRAY 1035
SWIG_TYPECHECK_INT32_ARRAY 1045
SWIG_TYPECHECK_INT64_ARRAY 1055
SWIG_TYPECHECK_INT128_ARRAY 1065
SWIG_TYPECHECK_FLOAT_ARRAY 1080
SWIG_TYPECHECK_DOUBLE_ARRAY 1090
SWIG_TYPECHECK_CHAR_ARRAY 1130
SWIG_TYPECHECK_STRING_ARRAY 1140
SWIG_TYPECHECK_OBJECT_ARRAY 1150
SWIG_TYPECHECK_BOOL_PTR 2015
SWIG_TYPECHECK_UINT8_PTR 2020
SWIG_TYPECHECK_INT8_PTR 2025
SWIG_TYPECHECK_UINT16_PTR 2030
SWIG_TYPECHECK_INT16_PTR 2035
SWIG_TYPECHECK_UINT32_PTR 2040
SWIG_TYPECHECK_INT32_PTR 2045
SWIG_TYPECHECK_UINT64_PTR 2050
SWIG_TYPECHECK_INT64_PTR 2055
SWIG_TYPECHECK_FLOAT_PTR 2080
SWIG_TYPECHECK_DOUBLE_PTR 2090
SWIG_TYPECHECK_CHAR_PTR 2130
SWIG_TYPECHECK_SWIGOBJECT 5000
</pre>
</div>
<p> (These precedence levels are defined in <tt>swig.swg</tt>, a library
file that's included by all target language modules.)</p>
<p> In this table, the precedence-level determines the order in which
types are going to be checked. Low values are always checked before
higher values. For example, integers are checked before floats, single
values are checked before arrays, and so forth.</p>
<p> Using the above table as a guide, each target language defines a
collection of &quot;typecheck&quot; typemaps. The following excerpt from the
Python module illustrates this:</p>
<div class="code">
<pre>
/* Python type checking rules */
/* Note: %typecheck(X) is a macro for %typemap(typecheck, precedence=X) */
%typecheck(SWIG_TYPECHECK_INTEGER)
int, short, long,
unsigned int, unsigned short, unsigned long,
signed char, unsigned char,
long long, unsigned long long,
const int &amp;, const short &amp;, const long &amp;,
const unsigned int &amp;, const unsigned short &amp;, const unsigned long &amp;,
const long long &amp;, const unsigned long long &amp;,
enum SWIGTYPE,
bool, const bool &amp;
{
$1 = (PyInt_Check($input) || PyLong_Check($input)) ? 1 : 0;
}
%typecheck(SWIG_TYPECHECK_DOUBLE)
float, double,
const float &amp;, const double &amp;
{
$1 = (PyFloat_Check($input) || PyInt_Check($input) || PyLong_Check($input)) ? 1 : 0;
}
%typecheck(SWIG_TYPECHECK_CHAR) char {
$1 = (PyString_Check($input) &amp;&amp; (PyString_Size($input) == 1)) ? 1 : 0;
}
%typecheck(SWIG_TYPECHECK_STRING) char * {
$1 = PyString_Check($input) ? 1 : 0;
}
%typemap(typecheck, precedence=SWIG_TYPECHECK_POINTER, noblock=1) SWIGTYPE * {
void *vptr = 0;
int res = SWIG_ConvertPtr($input, &amp;vptr, $1_descriptor, 0);
$1 = SWIG_IsOK(res) ? 1 : 0;
}
%typecheck(SWIG_TYPECHECK_POINTER) PyObject * {
$1 = ($input != 0);
}
</pre>
</div>
<p> It might take a bit of contemplation, but this code has merely
organized all of the basic C++ types, provided some simple
type-checking code, and assigned each type a precedence value.</p>
<p> Finally, to generate the dynamic dispatch function, SWIG uses the
following algorithm:</p>
<ul>
<li>Overloaded methods are first sorted by the number of required
arguments.</li>
<li>Methods with the same number of arguments are then sorted by
precedence values of argument types.</li>
<li>Typecheck typemaps are then emitted to produce a dispatch function
that checks arguments in the correct order.</li>
</ul>
<p> If you haven't written any typemaps of your own, it is unnecessary
to worry about the typechecking rules. However, if you have written new
input typemaps, you might have to supply a typechecking rule as well.
An easy way to do this is to simply copy one of the existing
typechecking rules. Here is an example,</p>
<div class="code">
<pre>
// Typemap for a C++ string
%typemap(in) std::string {
if (PyString_Check($input)) {
$1 = std::string(PyString_AsString($input));
} else {
SWIG_exception(SWIG_TypeError, &quot;string expected&quot;);
}
}
// Copy the typecheck code for &quot;char *&quot;.
%typemap(typecheck) std::string = char *;
</pre>
</div>
<p> The bottom line: If you are writing new typemaps and you are using
overloaded methods, you will probably have to write new typecheck code
or copy and modify existing typecheck code.</p>
<p> If you write a typecheck typemap and omit the precedence level, for
example commenting it out as shown below:</p>
<div class="code">
<pre>
%typemap(typecheck /*, precedence=SWIG_TYPECHECK_INTEGER*/) int {
$1 = PyInt_Check($input) ? 1 : 0;
}
</pre>
</div>
<p> then the type is given a precedence higher than any other known
precedence level and a <a href="#Warnings">warning</a> is issued:</p>
<div class="shell">
<pre>
example.i:18: Warning 467: Overloaded method foo(int) not supported (incomplete type
checking rule - no precedence level in typecheck typemap for 'int').
</pre>
</div>
<p><b> Notes:</b></p>
<ul>
<li>Typecheck typemaps are not used for non-overloaded methods. Because
of this, it is still always necessary to check types in any &quot;in&quot;
typemaps.</li>
<li>The dynamic dispatch process is only meant to be a heuristic. There
are many corner cases where SWIG simply can't disambiguate types to the
same degree as C++. The only way to resolve this ambiguity is to use
the %rename directive to rename one of the overloaded methods
(effectively eliminating overloading).</li>
<li> Typechecking may be partial. For example, if working with arrays,
the typecheck code might simply check the type of the first array
element and use that to dispatch to the correct function. Subsequent
&quot;in&quot; typemaps would then perform more extensive type-checking.</li>
<li>Make sure you read the section on <a href="#SWIGPlus_overloaded_methods">
overloading</a> in the SWIG and C++ chapter.</li>
</ul>
<h3><a name="Typemaps_typecheck_pointer">14.13.1 SWIG_TYPECHECK_POINTER
precedence level and the typecheck typemap</a></h3>
<p> When it comes to overloading of a particular type passed by value,
pointer or reference (const and non-const), a C++ compiler can
disambiguate which overloaded function to call. However, SWIG
effectively treats these as pointers in the target language and thus as
equivalent types. For example, consider:</p>
<div class="code">
<pre>
class X { ... };
void m(X const &amp;c); // equivalent: void m(X *c);
void m(X &amp;r); // equivalent: void m(X *r);
void m(X *p); // equivalent: void m(X *p);
</pre>
</div>
<p> These cannot be disambiguated in the target languages and so SWIG
will choose the first method and ignore the subsequent two methods. The
scripting languages do this by using the overload dispatch mechanism
described earlier and warnings indicate this:</p>
<div class="shell">
<pre>
example.i:6: Warning 509: Overloaded method m(X &amp;) effectively ignored,
example.i:5: Warning 509: as it is shadowed by m(X const &amp;).
example.i:7: Warning 509: Overloaded method m(X *) effectively ignored,
example.i:5: Warning 509: as it is shadowed by m(X const &amp;).
</pre>
</div>
<p> The statically typed languages like Java and C# automatically ignore
all but the first equivalent overloaded methods with warnings:</p>
<div class="shell">
<pre>
example.i:6: Warning 516: Overloaded method m(X &amp;) ignored,
example.i:5: Warning 516: using m(X const &amp;) instead.
example.i:7: Warning 516: Overloaded method m(X *) ignored,
example.i:5: Warning 516: using m(X const &amp;) instead.
</pre>
</div>
<p> You can select the overloaded method you would like to wrap by
ignoring the other two with <tt>%ignore</tt> or rename two of them with
<tt>%rename</tt> and this will of course remove the warnings too. The
problem of ambiguity is also discussed in the C++ chapter on <a href="#SWIGPlus_overloaded_methods">
overloading</a>.</p>
<p> So how does this work with respect to typemaps? The typemaps SWIG
provides to handle overloading for these three methods are from the
SWIGTYPE family. As discussed earlier, in <a href="#Typemaps_nn19">
Default typemap matching rules</a>, the <tt>SWIGTYPE &amp;</tt> typemaps are
used for references and <tt>SWIGTYPE *</tt> typemaps are used for
pointers. SWIG uses the special <tt>SWIG_TYPECHECK_POINTER</tt> (0)
precedence level to handle these types in the &quot;typecheck&quot; typemap:</p>
<div class="code">
<pre>
%typemap(typecheck, precedence=SWIG_TYPECHECK_POINTER) SWIGTYPE &amp; &quot;...&quot;
%typemap(typecheck, precedence=SWIG_TYPECHECK_POINTER) SWIGTYPE * &quot;...&quot;
</pre>
</div>
<p> When the SWIGTYPE &quot;typecheck&quot; typemaps use the <tt>
SWIG_TYPECHECK_POINTER</tt> precedence level, SWIG converts the type to
a pointer equivalent type and then uses the equivalent type to detect
if it can be disambiguated in an overloaded method in the target
language. In our example above, the equivalent types for <tt>X const &amp;</tt>
, <tt>X &amp;</tt> and <tt>X *</tt> are all <tt>X *</tt>. As they are the
same, they cannot be disambiguated and so just the first overloaded
method is chosen.</p>
<p> The automatic conversion to equivalent types and subsequent type
comparison is triggered via the use of the special <tt>
SWIG_TYPECHECK_POINTER</tt> precedence level and works for types passed
by value, pointer and reference. Alas, there are more ways to overload
a method that also need handling. C++ smart pointers are such a type
which can be disambiguated by a C++ compiler but not automatically by
SWIG. SWIG does not automatically know that a smart pointer has an
equivalent type, but it can be told manually. Just specify the
'equivalent' attribute in the &quot;typecheck&quot; typemap with a pointer to the
underlying type.</p>
<div class="code">
<pre>
%typemap(typecheck, precedence=SWIG_TYPECHECK_POINTER, equivalent=&quot;X *&quot;) MySmartPtr&lt;X&gt; &quot; ... &quot;
void m(X &amp;r); // equivalent: void m(X *r);
void m(MySmartPtr&lt;X&gt; s); // equivalent: void m(X *s);
</pre>
</div>
<p> Now SWIG will detect the two types are equivalent and generate valid
code by wrapping just the first overloaded method. You can of course
choose which method to wrap by ignoring one of them with <tt>%ignore</tt>
. Otherwise both can be wrapped by removing the overloading name
ambiguity by renaming one of them with <tt>%rename</tt>.</p>
<p> The 'equivalent' attribute is used in the implementation for the <a href="#Library_std_shared_ptr">
shared_ptr smart pointer</a> library.</p>
<h2><a name="Typemaps_nn48">14.14 More about %apply and %clear</a></h2>
<p> In order to implement certain kinds of program behavior, it is
sometimes necessary to write a family of typemap methods. For example,
to support output arguments, one often writes a family of typemaps like
this:</p>
<div class="code">
<pre>
%typemap(in, numinputs=0) int *OUTPUT (int temp) {
$1 = &amp;temp;
}
%typemap(argout) int *OUTPUT {
// return value somehow
}
</pre>
</div>
<p> To make it easier to apply the typemap to different argument types
and names, the <tt>%apply</tt> directive performs a copy of all
typemaps from a source type to one or more set of target types. For
example, if you specify this,</p>
<div class="code">
<pre>
%apply int *OUTPUT { int *retvalue, int32 *output };
</pre>
</div>
<p> then all of the <tt>int *OUTPUT</tt> (source) typemap methods are
copied to <tt>int *retvalue</tt> and <tt>int32 *output</tt> (the
targets).</p>
<p> However, there is a subtle aspect of <tt>%apply</tt> that needs
clarification. Namely, if a target contains a typemap method that the
source does not, the target typemap method remains in place and
unchanged. This behavior allows you to do two things:</p>
<ul>
<li>You can specialize parts of a complex typemap rule by first defining
a few typemaps and then using <tt>%apply</tt> to incorporate the
remaining pieces.</li>
<li>Different typemaps can be applied to the same datatype using
repeated <tt>%apply</tt> directives.</li>
</ul>
<p> For example:</p>
<div class="code">
<pre>
%typemap(in) int *INPUT (int temp) {
temp = ... get value from $input ...;
$1 = &amp;temp;
}
%typemap(check) int *POSITIVE {
if (*$1 &lt;= 0) {
SWIG_exception(SWIG_ValueError, &quot;Expected a positive number!\n&quot;);
return NULL;
}
}
%typemap(arginit) int *invalue %{
$1 = NULL;
%}
...
%apply int *INPUT { int *invalue };
%apply int *POSITIVE { int *invalue };
</pre>
</div>
<p> In this example, neither of the two <tt>%apply</tt> directives will
overwrite / delete the &quot;arginit&quot; typemap as neither has an &quot;arginit&quot;
typemap. The result is a family of three relevant typemaps for <tt>int
*invalue</tt>. Since <tt>%apply</tt> does not overwrite / delete any
existing rules, the only way to reset behavior is to delete them, such
as with the <tt>%clear</tt> directive. For example:</p>
<div class="code">
<pre>
%clear int *invalue;
</pre>
</div>
<p> will delete the typemaps for all the typemap methods; namely &quot;in&quot;,
&quot;check&quot; and &quot;arginit&quot;. Alternatively delete each one individually:</p>
<div class="code">
<pre>
%typemap(in) int *invalue;
%typemap(check) int *invalue;
%typemap(arginit) int *invalue;
</pre>
</div>
<h2><a name="Typemaps_nn47">14.15 Passing data between typemaps</a></h2>
<p> It is also important to note that the primary use of local variables
is to create stack-allocated objects for temporary use inside a wrapper
function (this is faster and less-prone to error than allocating data
on the heap). In general, the variables are not intended to pass
information between different types of typemaps. However, this can be
done if you realize that local names have the argument number appended
to them. For example, you could do this:</p>
<div class="code">
<pre>
%typemap(in) int *(int temp) {
temp = (int) PyInt_AsLong($input);
$1 = &amp;temp;
}
%typemap(argout) int * {
PyObject *o = PyInt_FromLong(temp$argnum);
...
}
</pre>
</div>
<p> In this case, the <tt>$argnum</tt> variable is expanded into the
argument number. Therefore, the code will reference the appropriate
local such as <tt>temp1</tt> and <tt>temp2</tt>. It should be noted
that there are plenty of opportunities to break the universe here and
that accessing locals in this manner should probably be avoided. At the
very least, you should make sure that the typemaps sharing information
have exactly the same types and names.</p>
<h2><a name="Typemaps_nn52">14.16 C++ &quot;this&quot; pointer</a></h2>
<p> All the rules discussed for typemaps apply to C++ as well as C.
However in addition C++ passes an extra parameter into every non-static
class method -- the <tt>this</tt> pointer. Occasionally it can be
useful to apply a typemap to this pointer (for example to check and
make sure <tt>this</tt> is non-null before deferencing). Actually, C
also has an the equivalent of the <tt>this</tt> pointer which is used
when accessing variables in a C struct.</p>
<p> In order to customise the <tt>this</tt> pointer handling, target a
variable named <tt>self</tt> in your typemaps. <tt>self</tt> is the
name SWIG uses to refer to the extra parameter in wrapped functions.</p>
<p> For example, if wrapping for Java generation:</p>
<div class="code">
<pre>
%typemap(check) SWIGTYPE *self %{
if (!$1) {
SWIG_JavaThrowException(jenv, SWIG_JavaNullPointerException,
&quot;invalid native object; delete() likely already called&quot;);
return $null;
}
%}
</pre>
</div>
<p> In the above case, the <tt>$1</tt> variable is expanded into the
argument name that SWIG is using as the <tt>this</tt> pointer. SWIG
will then insert the check code before the actual C++ class method is
called, and will raise an exception rather than crash the Java virtual
machine. The generated code will look something like:</p>
<div class="code">
<pre>
if (!arg1) {
SWIG_JavaThrowException(jenv, SWIG_JavaNullPointerException,
&quot;invalid native object; delete() likely already called&quot;);
return ;
}
(arg1)-&gt;wrappedFunction(...);
</pre>
</div>
<p> Note that if you have a parameter named <tt>self</tt> then it will
also match the typemap. One work around is to create an interface file
that wraps the method, but gives the argument a name other than <tt>
self</tt>.</p>
<h2><a name="Typemaps_nn51">14.17 Where to go for more information?</a></h2>
<p> The best place to find out more information about writing typemaps
is to look in the SWIG library. Most language modules define all of
their default behavior using typemaps. These are found in files such as
<tt>python.swg</tt>, <tt>perl5.swg</tt>, <tt>tcl8.swg</tt> and so
forth. The <tt>typemaps.i</tt> file in the library also contains
numerous examples. You should look at these files to get a feel for how
to define typemaps of your own. Some of the language modules support
additional typemaps and further information is available in the
individual chapters for each target language. There you may also find
more hands-on practical examples.</p>
<HR NOSHADE>
<h1><a name="Customization">15 Customization Features</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Customization_exception">Exception handling with
%exception</a>
<ul>
<li><a href="#Customization_nn3">Handling exceptions in C code</a></li>
<li><a href="#Customization_nn4">Exception handling with longjmp()</a></li>
<li><a href="#Customization_nn5">Handling C++ exceptions</a></li>
<li><a href="#Customization_allowexcept">Exception handlers for
variables</a></li>
<li><a href="#Customization_nn6">Defining different exception handlers</a>
</li>
<li><a href="#Customization_exception_special_variables">Special
variables for %exception</a></li>
<li><a href="#Customization_nn7">Using The SWIG exception library</a></li>
</ul>
</li>
<li><a href="#Customization_ownership">Object ownership and %newobject</a>
</li>
<li><a href="#Customization_features">Features and the %feature
directive</a>
<ul>
<li><a href="#Customization_feature_attributes">Feature attributes</a></li>
<li><a href="#Customization_feature_flags">Feature flags</a></li>
<li><a href="#Customization_clearing_features">Clearing features</a></li>
<li><a href="#Customization_features_default_args">Features and default
arguments</a></li>
<li><a href="#Customization_features_example">Feature example</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> In many cases, it is desirable to change the default wrapping of
particular declarations in an interface. For example, you might want to
provide hooks for catching C++ exceptions, add assertions, or provide
hints to the underlying code generator. This chapter describes some of
these customization techniques. First, a discussion of exception
handling is presented. Then, a more general-purpose customization
mechanism known as &quot;features&quot; is described.</p>
<h2><a name="Customization_exception">15.1 Exception handling with
%exception</a></h2>
<p> The <tt>%exception</tt> directive allows you to define a general
purpose exception handler. For example, you can specify the following:</p>
<div class="code">
<pre>
%exception {
try {
$action
}
catch (RangeError) {
... handle error ...
}
}
</pre>
</div>
<p> How the exception is handled depends on the target language, for
example, Python:</p>
<div class="code">
<pre>
%exception {
try {
$action
}
catch (RangeError) {
PyErr_SetString(PyExc_IndexError, &quot;index out-of-bounds&quot;);
SWIG_fail;
}
}
</pre>
</div>
<p> When defined, the code enclosed in braces is inserted directly into
the low-level wrapper functions. The special variable <tt>$action</tt>
is one of a few <a href="#Customization_exception_special_variables">
%exception special variables</a> supported and gets replaced with the
actual operation to be performed (a function call, method invocation,
attribute access, etc.). An exception handler remains in effect until
it is explicitly deleted. This is done by using either <tt>%exception</tt>
or <tt>%noexception</tt> with no code. For example:</p>
<div class="code">
<pre>
%exception; // Deletes any previously defined handler
</pre>
</div>
<h3><a name="Customization_nn3">15.1.1 Handling exceptions in C code</a></h3>
<p> C has no formal exception handling mechanism so there are several
approaches that might be used. A somewhat common technique is to simply
set a special error code. For example:</p>
<div class="code">
<pre>
/* File : except.c */
static char error_message[256];
static int error_status = 0;
void throw_exception(char *msg) {
strncpy(error_message, msg, 256);
error_status = 1;
}
void clear_exception() {
error_status = 0;
}
char *check_exception() {
if (error_status)
return error_message;
else
return NULL;
}
</pre>
</div>
<p> To use these functions, functions simply call <tt>throw_exception()</tt>
to indicate an error occurred. For example :</p>
<div class="code">
<pre>
double inv(double x) {
if (x != 0)
return 1.0/x;
else {
throw_exception(&quot;Division by zero&quot;);
return 0;
}
}
</pre>
</div>
<p> To catch the exception, you can write a simple exception handler
such as the following (shown for Perl5) :</p>
<div class="code">
<pre>
%exception {
char *err;
clear_exception();
$action
if ((err = check_exception())) {
croak(err);
}
}
</pre>
</div>
<p> In this case, when an error occurs, it is translated into a Perl
error. Each target language has its own approach to creating a runtime
error/exception in and for Perl it is the <tt>croak</tt> method shown
above.</p>
<h3><a name="Customization_nn4">15.1.2 Exception handling with longjmp()</a>
</h3>
<p> Exception handling can also be added to C code using the <tt>
&lt;setjmp.h&gt;</tt> library. Here is a minimalistic implementation that
relies on the C preprocessor :</p>
<div class="code">
<pre>
/* File : except.c
Just the declaration of a few global variables we're going to use */
#include &lt;setjmp.h&gt;
jmp_buf exception_buffer;
int exception_status;
/* File : except.h */
#include &lt;setjmp.h&gt;
extern jmp_buf exception_buffer;
extern int exception_status;
#define try if ((exception_status = setjmp(exception_buffer)) == 0)
#define catch(val) else if (exception_status == val)
#define throw(val) longjmp(exception_buffer, val)
#define finally else
/* Exception codes */
#define RangeError 1
#define DivisionByZero 2
#define OutOfMemory 3
</pre>
</div>
<p> Now, within a C program, you can do the following :</p>
<div class="code">
<pre>
double inv(double x) {
if (x)
return 1.0/x;
else
throw(DivisionByZero);
}
</pre>
</div>
<p> Finally, to create a SWIG exception handler, write the following :</p>
<div class="code">
<pre>
%{
#include &quot;except.h&quot;
%}
%exception {
try {
$action
} catch(RangeError) {
croak(&quot;Range Error&quot;);
} catch(DivisionByZero) {
croak(&quot;Division by zero&quot;);
} catch(OutOfMemory) {
croak(&quot;Out of memory&quot;);
} finally {
croak(&quot;Unknown exception&quot;);
}
}
</pre>
</div>
<p> Note: This implementation is only intended to illustrate the general
idea. To make it work better, you'll need to modify it to handle nested
<tt>try</tt> declarations.</p>
<h3><a name="Customization_nn5">15.1.3 Handling C++ exceptions</a></h3>
<p> Handling C++ exceptions is also straightforward. For example:</p>
<div class="code">
<pre>
%exception {
try {
$action
} catch(RangeError) {
croak(&quot;Range Error&quot;);
} catch(DivisionByZero) {
croak(&quot;Division by zero&quot;);
} catch(OutOfMemory) {
croak(&quot;Out of memory&quot;);
} catch(...) {
croak(&quot;Unknown exception&quot;);
}
}
</pre>
</div>
<p> The exception types need to be declared as classes elsewhere,
possibly in a header file :</p>
<div class="code">
<pre>
class RangeError {};
class DivisionByZero {};
class OutOfMemory {};
</pre>
</div>
<h3><a name="Customization_allowexcept">15.1.4 Exception handlers for
variables</a></h3>
<p> By default all variables will ignore <tt>%exception</tt>, so it is
effectively turned off for all variables wrappers. This applies to
global variables, member variables and static member variables. The
approach is certainly a logical one when wrapping variables in C.
However, in C++, it is quite possible for an exception to be thrown
while the variable is being assigned. To ensure <tt>%exception</tt> is
used when wrapping variables, it needs to be 'turned on' using the <tt>
%allowexception</tt> feature. Note that <tt>%allowexception</tt> is just
a macro for <tt>%feature(&quot;allowexcept&quot;)</tt>, that is, it is a feature
called &quot;allowexcept&quot;. Any variable which has this feature attached to
it, will then use the <tt>%exception</tt> feature, but of course, only
if there is a <tt>%exception</tt> attached to the variable in the first
place. The <tt>%allowexception</tt> feature works like any other
feature and so can be used globally or for selective variables.</p>
<div class="code">
<pre>
%allowexception; // turn on globally
%allowexception Klass::MyVar; // turn on for a specific variable
%noallowexception Klass::MyVar; // turn off for a specific variable
%noallowexception; // turn off globally
</pre>
</div>
<h3><a name="Customization_nn6">15.1.5 Defining different exception
handlers</a></h3>
<p> By default, the <tt>%exception</tt> directive creates an exception
handler that is used for all wrapper functions that follow it. Unless
there is a well-defined (and simple) error handling mechanism in place,
defining one universal exception handler may be unwieldy and result in
excessive code bloat since the handler is inlined into each wrapper
function.</p>
<p> To fix this, you can be more selective about how you use the <tt>
%exception</tt> directive. One approach is to only place it around
critical pieces of code. For example:</p>
<div class="code">
<pre>
%exception {
... your exception handler ...
}
/* Define critical operations that can throw exceptions here */
%exception;
/* Define non-critical operations that don't throw exceptions */
</pre>
</div>
<p> More precise control over exception handling can be obtained by
attaching an exception handler to specific declaration name. For
example:</p>
<div class="code">
<pre>
%exception allocate {
try {
$action
}
catch (MemoryError) {
croak(&quot;Out of memory&quot;);
}
}
</pre>
</div>
<p> In this case, the exception handler is only attached to declarations
named &quot;allocate&quot;. This would include both global and member functions.
The names supplied to <tt>%exception</tt> follow the same rules as for <tt>
%rename</tt> described in the section on <a href="#SWIGPlus_ambiguity_resolution_renaming">
Renaming and ambiguity resolution</a>. For example, if you wanted to
define an exception handler for a specific class, you might write this:</p>
<div class="code">
<pre>
%exception Object::allocate {
try {
$action
}
catch (MemoryError) {
croak(&quot;Out of memory&quot;);
}
}
</pre>
</div>
<p> When a class prefix is supplied, the exception handler is applied to
the corresponding declaration in the specified class as well as for
identically named functions appearing in derived classes.</p>
<p> <tt>%exception</tt> can even be used to pinpoint a precise
declaration when overloading is used. For example:</p>
<div class="code">
<pre>
%exception Object::allocate(int) {
try {
$action
}
catch (MemoryError) {
croak(&quot;Out of memory&quot;);
}
}
</pre>
</div>
<p> Attaching exceptions to specific declarations is a good way to
reduce code bloat. It can also be a useful way to attach exceptions to
specific parts of a header file. For example:</p>
<div class="code">
<pre>
%module example
%{
#include &quot;someheader.h&quot;
%}
// Define a few exception handlers for specific declarations
%exception Object::allocate(int) {
try {
$action
}
catch (MemoryError) {
croak(&quot;Out of memory&quot;);
}
}
%exception Object::getitem {
try {
$action
}
catch (RangeError) {
croak(&quot;Index out of range&quot;);
}
}
...
// Read a raw header file
%include &quot;someheader.h&quot;
</pre>
</div>
<h3><a name="Customization_exception_special_variables">15.1.6 Special
variables for %exception</a></h3>
<p> The %exception directive supports a few special variables which are
placeholders for code substitution. The following table shows the
available special variables and details what the special variables are
replaced with.</p>
<table summary="Special variables for %exception">
<tr><td>$action</td><td>The actual operation to be performed (a function
call, method invocation, variable access, etc.)</td></tr>
<tr><td>$decl</td><td>The fully qualified C/C++ declaration of the
method being wrapped without the return type.</td></tr>
<tr><td>$fulldecl</td><td>The fully qualified C/C++ declaration of the
method being wrapped including the return type.</td></tr>
<tr><td>$isvoid</td><td>Expands to 1 if the wrapped function has a void
return, otherwise expands to 0. Note that it expands to 1 in
destructors and 0 in constructors.</td></tr>
<tr><td>$name</td><td>The C/C++ symbol name for the function.</td></tr>
<tr><td>$overname</td><td>The extra mangling used in the symbol name for
overloaded method. Expands to nothing if the wrapped method is not
overloaded.</td></tr>
<tr><td>$parentclassname</td><td>The parent class name (if any) for a
method.</td></tr>
<tr><td>$parentclasssymname</td><td>The target language parent class
name (if any) for a method.</td></tr>
<tr><td>$symname</td><td>The symbol name used internally by SWIG.</td></tr>
<tr><td>$wrapname</td><td>The language specific wrapper name (usually a
C function name exported from the shared object/dll).</td></tr>
</table>
<p> The special variables are often used in situations where method
calls are logged. Exactly which form of the method call needs logging
is up to individual requirements, but the example code below shows all
the possible expansions, plus how an exception message could be
tailored to show the C++ method declaration:</p>
<div class="code">
<pre>
%exception Special::something {
log(&quot;symname: $symname&quot;);
log(&quot;overname: $overname&quot;);
log(&quot;wrapname: $wrapname&quot;);
log(&quot;decl: $decl&quot;);
log(&quot;fulldecl: $fulldecl&quot;);
try {
$action
}
catch (MemoryError) {
croak(&quot;Out of memory in $decl&quot;);
}
}
void log(const char *message);
struct Special {
void something(const char *c);
void something(int i);
};
</pre>
</div>
<p> Below shows the expansions for the 1st of the overloaded <tt>
something</tt> wrapper methods for Perl:</p>
<div class="code">
<pre>
log(&quot;symname: Special_something&quot;);
log(&quot;overname: __SWIG_0&quot;);
log(&quot;wrapname: _wrap_Special_something__SWIG_0&quot;);
log(&quot;decl: Special::something(char const *)&quot;);
log(&quot;fulldecl: void Special::something(char const *)&quot;);
try {
(arg1)-&gt;something((char const *)arg2);
}
catch (MemoryError) {
croak(&quot;Out of memory in Special::something(char const *)&quot;);
}
</pre>
</div>
<h3><a name="Customization_nn7">15.1.7 Using The SWIG exception library</a>
</h3>
<p> The <tt>exception.i</tt> library file provides support for creating
language independent exceptions in your interfaces. To use it, simply
put an &quot;<tt>%include exception.i</tt>&quot; in your interface file. This
provides a function <tt>SWIG_exception()</tt> that can be used to raise
common scripting language exceptions in a portable manner. For example
:</p>
<div class="code">
<pre>
// Language independent exception handler
%include exception.i
%exception {
try {
$action
} catch(RangeError) {
SWIG_exception(SWIG_ValueError, &quot;Range Error&quot;);
} catch(DivisionByZero) {
SWIG_exception(SWIG_DivisionByZero, &quot;Division by zero&quot;);
} catch(OutOfMemory) {
SWIG_exception(SWIG_MemoryError, &quot;Out of memory&quot;);
} catch(...) {
SWIG_exception(SWIG_RuntimeError, &quot;Unknown exception&quot;);
}
}
</pre>
</div>
<p> As arguments, <tt>SWIG_exception()</tt> takes an error type code (an
integer) and an error message string. The currently supported error
types are :</p>
<div class="diagram">
<pre>
SWIG_UnknownError
SWIG_IOError
SWIG_RuntimeError
SWIG_IndexError
SWIG_TypeError
SWIG_DivisionByZero
SWIG_OverflowError
SWIG_SyntaxError
SWIG_ValueError
SWIG_SystemError
SWIG_AttributeError
SWIG_MemoryError
SWIG_NullReferenceError
</pre>
</div>
<p> The <tt>SWIG_exception()</tt> function can also be used in typemaps.</p>
<h2><a name="Customization_ownership">15.2 Object ownership and
%newobject</a></h2>
<p> A common problem in some applications is managing proper ownership
of objects. For example, consider a function like this:</p>
<div class="code">
<pre>
Foo *blah() {
Foo *f = new Foo();
return f;
}
</pre>
</div>
<p> If you wrap the function <tt>blah()</tt>, SWIG has no idea that the
return value is a newly allocated object. As a result, the resulting
extension module may produce a memory leak (SWIG is conservative and
will never delete objects unless it knows for certain that the returned
object was newly created).</p>
<p> To fix this, you can provide an extra hint to the code generator
using the <tt>%newobject</tt> directive. For example:</p>
<div class="code">
<pre>
%newobject blah;
Foo *blah();
</pre>
</div>
<p> <tt>%newobject</tt> works exactly like <tt>%rename</tt> and <tt>
%exception</tt>. In other words, you can attach it to class members and
parameterized declarations as before. For example:</p>
<div class="code">
<pre>
%newobject ::blah(); // Only applies to global blah
%newobject Object::blah(int, double); // Only blah(int, double) in Object
%newobject *::copy; // Copy method in all classes
...
</pre>
</div>
<p> When <tt>%newobject</tt> is supplied, many language modules will
arrange to take ownership of the return value. This allows the value to
be automatically garbage-collected when it is no longer in use.
However, this depends entirely on the target language (a language
module may also choose to ignore the <tt>%newobject</tt> directive).</p>
<p> Closely related to <tt>%newobject</tt> is a special typemap. The
&quot;newfree&quot; typemap can be used to deallocate a newly allocated return
value. It is only available on methods for which <tt>%newobject</tt>
has been applied and is commonly used to clean-up string results. For
example:</p>
<div class="code">
<pre>
%typemap(newfree) char * &quot;free($1);&quot;
...
%newobject strdup;
...
char *strdup(const char *s);
</pre>
</div>
<p> In this case, the result of the function is a string in the target
language. Since this string is a copy of the original result, the data
returned by <tt>strdup()</tt> is no longer needed. The &quot;newfree&quot;
typemap in the example simply releases this memory.</p>
<p> As a complement to the <tt>%newobject</tt>, from SWIG 1.3.28, you
can use the <tt>%delobject</tt> directive. For example, if you have two
methods, one to create objects and one to destroy them, you can use:</p>
<div class="code">
<pre>
%newobject create_foo;
%delobject destroy_foo;
...
Foo *create_foo();
void destroy_foo(Foo *foo);
</pre>
</div>
<p> or in a member method as:</p>
<div class="code">
<pre>
%delobject Foo::destroy;
class Foo {
public:
void destroy() { delete this;}
private:
~Foo();
};
</pre>
</div>
<p> <tt>%delobject</tt> instructs SWIG that the first argument passed to
the method will be destroyed, and therefore, the target language should
not attempt to deallocate it twice. This is similar to use the DISOWN
typemap in the first method argument, and in fact, it also depends on
the target language on implementing the 'disown' mechanism properly.</p>
<p> The use of <tt>%newobject</tt> is also integrated with reference
counting and is covered in the <a href="#SWIGPlus_ref_unref">C++
reference counted objects</a> section.</p>
<h2><a name="Customization_features">15.3 Features and the %feature
directive</a></h2>
<p> Both <tt>%exception</tt> and <tt>%newobject</tt> are examples of a
more general purpose customization mechanism known as &quot;features.&quot; A
feature is simply a user-definable property that is attached to
specific declarations. Features are attached using the <tt>%feature</tt>
directive. For example:</p>
<div class="code">
<pre>
%feature(&quot;except&quot;) Object::allocate {
try {
$action
}
catch (MemoryError) {
croak(&quot;Out of memory&quot;);
}
}
%feature(&quot;new&quot;, &quot;1&quot;) *::copy;
</pre>
</div>
<p> In fact, the <tt>%exception</tt> and <tt>%newobject</tt> directives
are really nothing more than macros involving <tt>%feature</tt>:</p>
<div class="code">
<pre>
#define %exception %feature(&quot;except&quot;)
#define %newobject %feature(&quot;new&quot;, &quot;1&quot;)
</pre>
</div>
<p> The name matching rules outlined in the <a href="#SWIGPlus_ambiguity_resolution_renaming">
Renaming and ambiguity resolution</a> section applies to all <tt>
%feature</tt> directives. In fact the <tt>%rename</tt> directive is just
a special form of <tt>%feature</tt>. The matching rules mean that
features are very flexible and can be applied with pinpoint accuracy to
specific declarations if needed. Additionally, if no declaration name
is given, a global feature is said to be defined. This feature is then
attached to<em> every</em> declaration that follows. This is how global
exception handlers are defined. For example:</p>
<div class="code">
<pre>
/* Define a global exception handler */
%feature(&quot;except&quot;) {
try {
$action
}
...
}
... bunch of declarations ...
</pre>
</div>
<p> The <tt>%feature</tt> directive can be used with different syntax.
The following are all equivalent:</p>
<div class="code">
<pre>
%feature(&quot;except&quot;) Object::method { $action };
%feature(&quot;except&quot;) Object::method %{ $action %};
%feature(&quot;except&quot;) Object::method &quot; $action &quot;;
%feature(&quot;except&quot;, &quot;$action&quot;) Object::method;
</pre>
</div>
<p> The syntax in the first variation will generate the <tt>{ }</tt>
delimiters used whereas the other variations will not.</p>
<h3><a name="Customization_feature_attributes">15.3.1 Feature attributes</a>
</h3>
<p> The <tt>%feature</tt> directive also accepts XML style attributes in
the same way that typemaps do. Any number of attributes can be
specified. The following is the generic syntax for features:</p>
<div class="code">
<pre>
%feature(&quot;name&quot;, &quot;value&quot;, attribute1=&quot;AttributeValue1&quot;) symbol;
%feature(&quot;name&quot;, attribute1=&quot;AttributeValue1&quot;) symbol {value};
%feature(&quot;name&quot;, attribute1=&quot;AttributeValue1&quot;) symbol %{value%};
%feature(&quot;name&quot;, attribute1=&quot;AttributeValue1&quot;) symbol &quot;value&quot;;
</pre>
</div>
<p> More than one attribute can be specified using a comma separated
list. The Java module is an example that uses attributes in <tt>
%feature(&quot;except&quot;)</tt>. The <tt>throws</tt> attribute specifies the
name of a Java class to add to a proxy method's throws clause. In the
following example, <tt>MyExceptionClass</tt> is the name of the Java
class for adding to the throws clause.</p>
<div class="code">
<pre>
%feature(&quot;except&quot;, throws=&quot;MyExceptionClass&quot;) Object::method {
try {
$action
} catch (...) {
... code to throw a MyExceptionClass Java exception ...
}
};
</pre>
</div>
<p> Further details can be obtained from the <a href="#Java_exception_handling">
Java exception handling</a> section.</p>
<h3><a name="Customization_feature_flags">15.3.2 Feature flags</a></h3>
<p> Feature flags are used to enable or disable a particular feature.
Feature flags are a common but simple usage of <tt>%feature</tt> and
the feature value should be either <tt>1</tt> to enable or <tt>0</tt>
to disable the feature.</p>
<div class="code">
<pre>
%feature(&quot;featurename&quot;) // enables feature
%feature(&quot;featurename&quot;, &quot;1&quot;) // enables feature
%feature(&quot;featurename&quot;, &quot;x&quot;) // enables feature
%feature(&quot;featurename&quot;, &quot;0&quot;) // disables feature
%feature(&quot;featurename&quot;, &quot;&quot;) // clears feature
</pre>
</div>
<p> Actually any value other than zero will enable the feature. Note
that if the value is omitted completely, the default value becomes <tt>
1</tt>, thereby enabling the feature. A feature is cleared by specifying
no value, see <a href="#Customization_clearing_features">Clearing
features</a>. The <tt>%immutable</tt> directive described in the <a href="#SWIG_readonly_variables">
Creating read-only variables</a> section, is just a macro for <tt>
%feature(&quot;immutable&quot;)</tt>, and can be used to demonstrates feature
flags:</p>
<div class="code">
<pre>
// features are disabled by default
int red; // mutable
%feature(&quot;immutable&quot;); // global enable
int orange; // immutable
%feature(&quot;immutable&quot;, &quot;0&quot;); // global disable
int yellow; // mutable
%feature(&quot;immutable&quot;, &quot;1&quot;); // another form of global enable
int green; // immutable
%feature(&quot;immutable&quot;, &quot;&quot;); // clears the global feature
int blue; // mutable
</pre>
</div>
<p> Note that features are disabled by default and must be explicitly
enabled either globally or by specifying a targeted declaration. The
above intersperses SWIG directives with C code. Of course you can
target features explicitly, so the above could also be rewritten as:</p>
<div class="code">
<pre>
%feature(&quot;immutable&quot;, &quot;1&quot;) orange;
%feature(&quot;immutable&quot;, &quot;1&quot;) green;
int red; // mutable
int orange; // immutable
int yellow; // mutable
int green; // immutable
int blue; // mutable
</pre>
</div>
<p> The above approach allows for the C declarations to be separated
from the SWIG directives for when the C declarations are parsed from a
C header file. The logic above can of course be inverted and rewritten
as:</p>
<div class="code">
<pre>
%feature(&quot;immutable&quot;, &quot;1&quot;);
%feature(&quot;immutable&quot;, &quot;0&quot;) red;
%feature(&quot;immutable&quot;, &quot;0&quot;) yellow;
%feature(&quot;immutable&quot;, &quot;0&quot;) blue;
int red; // mutable
int orange; // immutable
int yellow; // mutable
int green; // immutable
int blue; // mutable
</pre>
</div>
<p> As hinted above for <tt>%immutable</tt>, most feature flags can also
be specified via alternative syntax. The alternative syntax is just a
macro in the <tt>swig.swg</tt> Library file. The following shows the
alternative syntax for the imaginary <tt>featurename</tt> feature:</p>
<div class="code">
<pre>
%featurename // equivalent to %feature(&quot;featurename&quot;, &quot;1&quot;) ie enables feature
%nofeaturename // equivalent to %feature(&quot;featurename&quot;, &quot;0&quot;) ie disables feature
%clearfeaturename // equivalent to %feature(&quot;featurename&quot;, &quot;&quot;) ie clears feature
</pre>
</div>
<p> The concept of clearing features is discussed next.</p>
<h3><a name="Customization_clearing_features">15.3.3 Clearing features</a>
</h3>
<p> A feature stays in effect until it is explicitly cleared. A feature
is cleared by supplying a <tt>%feature</tt> directive with no value.
For example <tt>%feature(&quot;name&quot;, &quot;&quot;)</tt>. A cleared feature means that
any feature exactly matching any previously defined feature is no
longer used in the name matching rules. So if a feature is cleared, it
might mean that another name matching rule will apply. To clarify,
let's consider the <tt>except</tt> feature again (<tt>%exception</tt>):</p>
<div class="code">
<pre>
// Define global exception handler
%feature(&quot;except&quot;) {
try {
$action
} catch (...) {
croak(&quot;Unknown C++ exception&quot;);
}
}
// Define exception handler for all clone methods to log the method calls
%feature(&quot;except&quot;) *::clone() {
try {
logger.info(&quot;$action&quot;);
$action
} catch (...) {
croak(&quot;Unknown C++ exception&quot;);
}
}
... initial set of class declarations with clone methods ...
// clear the previously defined feature
%feature(&quot;except&quot;, &quot;&quot;) *::clone();
... final set of class declarations with clone methods ...
</pre>
</div>
<p> In the above scenario, the initial set of clone methods will log all
method invocations from the target language. This specific feature is
cleared for the final set of clone methods. However, these clone
methods will still have an exception handler (without logging) as the
next best feature match for them is the global exception handler.</p>
<p> Note that clearing a feature is not always the same as disabling it.
Clearing the feature above with <tt>%feature(&quot;except&quot;, &quot;&quot;) *::clone()</tt>
is not the same as specifying <tt>%feature(&quot;except&quot;, &quot;0&quot;) *::clone()</tt>
. The former will disable the feature for clone methods - the feature is
still a better match than the global feature. If on the other hand, no
global exception handler had been defined at all, then clearing the
feature would be the same as disabling it as no other feature would
have matched.</p>
<p> Note that the feature must match exactly for it to be cleared by any
previously defined feature. For example the following attempt to clear
the initial feature will not work:</p>
<div class="code">
<pre>
%feature(&quot;except&quot;) clone() { logger.info(&quot;$action&quot;); $action }
%feature(&quot;except&quot;, &quot;&quot;) *::clone();
</pre>
</div>
<p> but this will:</p>
<div class="code">
<pre>
%feature(&quot;except&quot;) clone() { logger.info(&quot;$action&quot;); $action }
%feature(&quot;except&quot;, &quot;&quot;) clone();
</pre>
</div>
<p> SWIG provides macros for disabling and clearing features. Many of
these can be found in the <tt>swig.swg</tt> library file. The typical
pattern is to define three macros; one to define the feature itself,
one to disable the feature and one to clear the feature. The three
macros below show this for the &quot;except&quot; feature:</p>
<div class="code">
<pre>
#define %exception %feature(&quot;except&quot;)
#define %noexception %feature(&quot;except&quot;, &quot;0&quot;)
#define %clearexception %feature(&quot;except&quot;, &quot;&quot;)
</pre>
</div>
<h3><a name="Customization_features_default_args">15.3.4 Features and
default arguments</a></h3>
<p> SWIG treats methods with default arguments as separate overloaded
methods as detailed in the <a href="#SWIGPlus_default_args">default
arguments</a> section. Any <tt>%feature</tt> targeting a method with
default arguments will apply to all the extra overloaded methods that
SWIG generates if the default arguments are specified in the feature.
If the default arguments are not specified in the feature, then the
feature will match that exact wrapper method only and not the extra
overloaded methods that SWIG generates. For example:</p>
<div class="code">
<pre>
%feature(&quot;except&quot;) hello(int i=0, double d=0.0) { ... }
void hello(int i=0, double d=0.0);
</pre>
</div>
<p> will apply the feature to all three wrapper methods, that is:</p>
<div class="code">
<pre>
void hello(int i, double d);
void hello(int i);
void hello();
</pre>
</div>
<p> If the default arguments are not specified in the feature:</p>
<div class="code">
<pre>
%feature(&quot;except&quot;) hello(int i, double d) { ... }
void hello(int i=0, double d=0.0);
</pre>
</div>
<p> then the feature will only apply to this wrapper method:</p>
<div class="code">
<pre>
void hello(int i, double d);
</pre>
</div>
<p> and not these wrapper methods:</p>
<div class="code">
<pre>
void hello(int i);
void hello();
</pre>
</div>
<p> If <a href="#SWIGPlus_default_args">compactdefaultargs</a> are being
used, then the difference between specifying or not specifying default
arguments in a feature is not applicable as just one wrapper is
generated.</p>
<p><b> Compatibility note:</b> The different behaviour of features
specified with or without default arguments was introduced in
SWIG-1.3.23 when the approach to wrapping methods with default
arguments was changed.</p>
<h3><a name="Customization_features_example">15.3.5 Feature example</a></h3>
<p> As has been shown earlier, the intended use for the <tt>%feature</tt>
directive is as a highly flexible customization mechanism that can be
used to annotate declarations with additional information for use by
specific target language modules. Another example is in the Python
module. You might use <tt>%feature</tt> to rewrite proxy/shadow class
code as follows:</p>
<div class="code">
<pre>
%module example
%rename(bar_id) bar(int, double);
// Rewrite bar() to allow some nice overloading
%feature(&quot;shadow&quot;) Foo::bar(int) %{
def bar(*args):
if len(args) == 3:
return apply(examplec.Foo_bar_id, args)
return apply(examplec.Foo_bar, args)
%}
class Foo {
public:
int bar(int x);
int bar(int x, double y);
}
</pre>
</div>
<p> Further details of <tt>%feature</tt> usage is described in the
documentation for specific language modules.</p>
<HR NOSHADE>
<h1><a name="Contract">16 Contracts</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Contract_nn2">The %contract directive</a></li>
<li><a href="#Contract_nn3">%contract and classes</a></li>
<li><a href="#Contract_nn4">Constant aggregation and %aggregate_check</a>
</li>
<li><a href="#Contract_nn5">Notes</a></li>
</ul>
</div>
<!-- INDEX -->
<p> A common problem that arises when wrapping C libraries is that of
maintaining reliability and checking for errors. The fact of the matter
is that many C programs are notorious for not providing error checks.
Not only that, when you expose the internals of an application as a
library, it often becomes possible to crash it simply by providing bad
inputs or using it in a way that wasn't intended.</p>
<p> This chapter describes SWIG's support for software contracts. In the
context of SWIG, a contract can be viewed as a runtime constraint that
is attached to a declaration. For example, you can easily attach
argument checking rules, check the output values of a function and
more. When one of the rules is violated by a script, a runtime
exception is generated rather than having the program continue to
execute.</p>
<h2><a name="Contract_nn2">16.1 The %contract directive</a></h2>
<p> Contracts are added to a declaration using the %contract directive.
Here is a simple example:</p>
<div class="code">
<pre>
%contract sqrt(double x) {
require:
x &gt;= 0;
ensure:
sqrt &gt;= 0;
}
...
double sqrt(double);
</pre>
</div>
<p> In this case, a contract is being added to the <tt>sqrt()</tt>
function. The <tt>%contract</tt> directive must always appear before
the declaration in question. Within the contract there are two
sections, both of which are optional. The <tt>require:</tt> section
specifies conditions that must hold before the function is called.
Typically, this is used to check argument values. The <tt>ensure:</tt>
section specifies conditions that must hold after the function is
called. This is often used to check return values or the state of the
program. In both cases, the conditions that must hold must be specified
as boolean expressions.</p>
<p> In the above example, we're simply making sure that sqrt() returns a
non-negative number (if it didn't, then it would be broken in some
way).</p>
<p> Once a contract has been specified, it modifies the behavior of the
resulting module. For example:</p>
<div class="shell">
<pre>
&gt;&gt;&gt; example.sqrt(2)
1.4142135623730951
&gt;&gt;&gt; example.sqrt(-2)
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
RuntimeError: Contract violation: require: (arg1&gt;=0)
&gt;&gt;&gt;
</pre>
</div>
<h2><a name="Contract_nn3">16.2 %contract and classes</a></h2>
<p> The <tt>%contract</tt> directive can also be applied to class
methods and constructors. For example:</p>
<div class="code">
<pre>
%contract Foo::bar(int x, int y) {
require:
x &gt; 0;
ensure:
bar &gt; 0;
}
%contract Foo::Foo(int a) {
require:
a &gt; 0;
}
class Foo {
public:
Foo(int);
int bar(int, int);
};
</pre>
</div>
<p> The way in which <tt>%contract</tt> is applied is exactly the same
as the <tt>%feature</tt> directive. Thus, any contract that you
specified for a base class will also be attached to inherited methods.
For example:</p>
<div class="code">
<pre>
class Spam : public Foo {
public:
int bar(int, int); // Gets contract defined for Foo::bar(int, int)
};
</pre>
</div>
<p> In addition to this, separate contracts can be applied to both the
base class and a derived class. For example:</p>
<div class="code">
<pre>
%contract Foo::bar(int x, int) {
require:
x &gt; 0;
}
%contract Spam::bar(int, int y) {
require:
y &gt; 0;
}
class Foo {
public:
int bar(int, int); // Gets Foo::bar contract.
};
class Spam : public Foo {
public:
int bar(int, int); // Gets Foo::bar and Spam::bar contract
};
</pre>
</div>
<p> When more than one contract is applied, the conditions specified in
a &quot;require:&quot; section are combined together using a logical-AND
operation. In other words conditions specified for the base class and
conditions specified for the derived class all must hold. In the above
example, this means that both the arguments to <tt>Spam::bar</tt> must
be positive.</p>
<h2><a name="Contract_nn4">16.3 Constant aggregation and
%aggregate_check</a></h2>
<p> Consider an interface file that contains the following code:</p>
<div class="code">
<pre>
#define UP 1
#define DOWN 2
#define RIGHT 3
#define LEFT 4
void move(SomeObject *, int direction, int distance);
</pre>
</div>
<p> One thing you might want to do is impose a constraint on the
direction parameter to make sure it's one of a few accepted values. To
do that, SWIG provides an easy to use macro %aggregate_check() that
works like this:</p>
<div class="code">
<pre>
%aggregate_check(int, check_direction, UP, DOWN, LEFT, RIGHT);
</pre>
</div>
<p> This merely defines a utility function of the form</p>
<div class="code">
<pre>
int check_direction(int x);
</pre>
</div>
<p> That checks the argument x to see if it is one of the values listed.
This utility function can be used in contracts. For example:</p>
<div class="code">
<pre>
%aggregate_check(int, check_direction, UP, DOWN, RIGHT, LEFT);
%contract move(SomeObject *, int direction, in) {
require:
check_direction(direction);
}
#define UP 1
#define DOWN 2
#define RIGHT 3
#define LEFT 4
void move(SomeObject *, int direction, int distance);
</pre>
</div>
<p> Alternatively, it can be used in typemaps and other directives. For
example:</p>
<div class="code">
<pre>
%aggregate_check(int, check_direction, UP, DOWN, RIGHT, LEFT);
%typemap(check) int direction {
if (!check_direction($1)) SWIG_exception(SWIG_ValueError, &quot;Bad direction&quot;);
}
#define UP 1
#define DOWN 2
#define RIGHT 3
#define LEFT 4
void move(SomeObject *, int direction, int distance);
</pre>
</div>
<p> Regrettably, there is no automatic way to perform similar checks
with enums values. Maybe in a future release.</p>
<h2><a name="Contract_nn5">16.4 Notes</a></h2>
<p> Contract support was implemented by Songyan (Tiger) Feng and first
appeared in SWIG-1.3.20.</p>
<HR NOSHADE>
<h1><a name="Varargs">17 Variable Length Arguments</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Varargs_nn2">Introduction</a></li>
<li><a href="#Varargs_nn3">The Problem</a></li>
<li><a href="#Varargs_nn4">Default varargs support</a></li>
<li><a href="#Varargs_nn5">Argument replacement using %varargs</a></li>
<li><a href="#Varargs_nn6">Varargs and typemaps</a></li>
<li><a href="#Varargs_nn7">Varargs wrapping with libffi</a></li>
<li><a href="#Varargs_nn8">Wrapping of va_list</a></li>
<li><a href="#Varargs_nn9">C++ Issues</a></li>
<li><a href="#Varargs_nn10">Discussion</a></li>
</ul>
</div>
<!-- INDEX -->
<p><b> (a.k.a, &quot;The horror. The horror.&quot;)</b></p>
<p> This chapter describes the problem of wrapping functions that take a
variable number of arguments. For instance, generating wrappers for the
C <tt>printf()</tt> family of functions.</p>
<p> This topic is sufficiently advanced to merit its own chapter. In
fact, support for varargs is an often requested feature that was first
added in SWIG-1.3.12. Most other wrapper generation tools have wisely
chosen to avoid this issue.</p>
<h2><a name="Varargs_nn2">17.1 Introduction</a></h2>
<p> Some C and C++ programs may include functions that accept a variable
number of arguments. For example, most programmers are familiar with
functions from the C library such as the following:</p>
<div class="code">
<pre>
int printf(const char *fmt, ...)
int fprintf(FILE *, const char *fmt, ...);
int sprintf(char *s, const char *fmt, ...);
</pre>
</div>
<p> Although there is probably little practical purpose in wrapping
these specific C library functions in a scripting language (what would
be the point?), a library may include its own set of special functions
based on a similar API. For example:</p>
<div class="code">
<pre>
int traceprintf(const char *fmt, ...);
</pre>
</div>
<p> In this case, you may want to have some kind of access from the
target language.</p>
<p> Before describing the SWIG implementation, it is important to
discuss the common uses of varargs that you are likely to encounter in
real programs. Obviously, there are the <tt>printf()</tt> style output
functions as shown. Closely related to this would be <tt>scanf()</tt>
style input functions that accept a format string and a list of
pointers into which return values are placed. However, variable length
arguments are also sometimes used to write functions that accept a
NULL-terminated list of pointers. A good example of this would be a
function like this:</p>
<div class="code">
<pre>
int execlp(const char *path, const char *arg1, ...);
...
/* Example */
execlp(&quot;ls&quot;, &quot;ls&quot;, &quot;-l&quot;, NULL);
</pre>
</div>
<p> In addition, varargs is sometimes used to fake default arguments in
older C libraries. For instance, the low level <tt>open()</tt> system
call is often declared as a varargs function so that it will accept two
or three arguments:</p>
<div class="code">
<pre>
int open(const char *path, int oflag, ...);
...
/* Examples */
f = open(&quot;foo&quot;, O_RDONLY);
g = open(&quot;bar&quot;, O_WRONLY | O_CREAT, 0644);
</pre>
</div>
<p> Finally, to implement a varargs function, recall that you have to
use the C library functions defined in <tt>&lt;stdarg.h&gt;</tt>. For
example:</p>
<div class="code">
<pre>
List make_list(const char *s, ...) {
va_list ap;
List x;
...
va_start(ap, s);
while (s) {
x.append(s);
s = va_arg(ap, const char *);
}
va_end(ap);
return x;
}
</pre>
</div>
<h2><a name="Varargs_nn3">17.2 The Problem</a></h2>
<p> Generating wrappers for a variable length argument function presents
a number of special challenges. Although C provides support for
implementing functions that receive variable length arguments, there
are no functions that can go in the other direction. Specifically, you
can't write a function that dynamically creates a list of arguments and
which invokes a varargs function on your behalf.</p>
<p> Although it is possible to write functions that accept the special
type <tt>va_list</tt>, this is something entirely different. You can't
take a <tt>va_list</tt> structure and pass it in place of the variable
length arguments to another varargs function. It just doesn't work.</p>
<p> The reason this doesn't work has to do with the way that function
calls get compiled. For example, suppose that your program has a
function call like this:</p>
<div class="code">
<pre>
printf(&quot;Hello %s. Your number is %d\n&quot;, name, num);
</pre>
</div>
<p> When the compiler looks at this, it knows that you are calling <tt>
printf()</tt> with exactly three arguments. Furthermore, it knows that
the number of arguments as well are their types and sizes is<em> never</em>
going to change during program execution. Therefore, this gets turned
to machine code that sets up a three-argument stack frame followed by a
call to <tt>printf()</tt>.</p>
<p> In contrast, suppose you attempted to make some kind of wrapper
around <tt>printf()</tt> using code like this:</p>
<div class="code">
<pre>
int wrap_printf(const char *fmt, ...) {
va_list ap;
va_start(ap, fmt);
...
printf(fmt, ap);
...
va_end(ap);
};
</pre>
</div>
<p> Although this code might compile, it won't do what you expect. This
is because the call to <tt>printf()</tt> is compiled as a procedure
call involving only two arguments. However, clearly a two-argument
configuration of the call stack is completely wrong if your intent is
to pass an arbitrary number of arguments to the real <tt>printf()</tt>.
Needless to say, it won't work.</p>
<p> Unfortunately, the situation just described is exactly the problem
faced by wrapper generation tools. In general, the number of passed
arguments will not be known until run-time. To make matters even worse,
you won't know the types and sizes of arguments until run-time as well.
Needless to say, there is no obvious way to make the C compiler
generate code for a function call involving an unknown number of
arguments of unknown types.</p>
<p> In theory, it<em> is</em> possible to write a wrapper that does the
right thing. However, this involves knowing the underlying ABI for the
target platform and language as well as writing special purpose code
that manually constructed the call stack before making a procedure
call. Unfortunately, both of these tasks require the use of inline
assembly code. Clearly, that's the kind of solution you would much
rather avoid.</p>
<p> With this nastiness in mind, SWIG provides a number of solutions to
the varargs wrapping problem. Most of these solutions are compromises
that provide limited varargs support without having to resort to
assembly language. However, SWIG can also support real varargs wrapping
(with stack-frame manipulation) if you are willing to get hands dirty.
Keep reading.</p>
<h2><a name="Varargs_nn4">17.3 Default varargs support</a></h2>
<p> When variable length arguments appear in an interface, the default
behavior is to drop the variable argument list entirely, replacing them
with a single NULL pointer. For example, if you had this function,</p>
<div class="code">
<pre>
void traceprintf(const char *fmt, ...);
</pre>
</div>
<p> it would be wrapped as if it had been declared as follows:</p>
<div class="code">
<pre>
void traceprintf(const char *fmt);
</pre>
</div>
<p> When the function is called inside the wrappers, it is called as
follows:</p>
<div class="code">
<pre>
traceprintf(arg1, NULL);
</pre>
</div>
<p> Arguably, this approach seems to defeat the whole point of variable
length arguments. However, this actually provides enough support for
many simple kinds of varargs functions to still be useful, however it
does come with a caveat. For instance, you could make function calls
like this (in Python):</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; traceprintf(&quot;Hello World&quot;)
&gt;&gt;&gt; traceprintf(&quot;Hello %s. Your number is %d\n&quot; % (name, num))
&gt;&gt;&gt; traceprintf(&quot;Your result is 90%%.&quot;)
</pre>
</div>
<p> Notice how string formatting is being done in Python instead of C.
The caveat is the strings passed must be safe to use in C though. For
example if name was to contain a &quot;%&quot; it should be double escaped in
order to avoid unpredictable behaviour:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; traceprintf(&quot;Your result is 90%.\n&quot;) # unpredictable behaviour
&gt;&gt;&gt; traceprintf(&quot;Your result is 90%%.\n&quot;) # good
</pre>
</div>
<p> Read on for further solutions.</p>
<h2><a name="Varargs_nn5">17.4 Argument replacement using %varargs</a></h2>
<p> Instead of dropping the variable length arguments, an alternative
approach is to replace <tt>(...)</tt> with a set of suitable arguments.
SWIG provides a special <tt>%varargs</tt> directive that can be used to
do this. For example,</p>
<div class="code">
<pre>
%varargs(int mode = 0) open;
...
int open(const char *path, int oflags, ...);
</pre>
</div>
<p> is equivalent to this:</p>
<div class="code">
<pre>
int open(const char *path, int oflags, int mode = 0);
</pre>
</div>
<p> In this case, <tt>%varargs</tt> is simply providing more specific
information about the extra arguments that might be passed to a
function. If the arguments to a varargs function are of uniform type, <tt>
%varargs</tt> can also accept a numerical argument count as follows:</p>
<div class="code">
<pre>
%varargs(3, char *str = NULL) execlp;
...
int execlp(const char *path, const char *arg, ...);
</pre>
</div>
<p> and is effectively seen as:</p>
<div class="code">
<pre>
int execlp(const char *path, const char *arg,
char *str1 = NULL,
char *str2 = NULL,
char *str3 = NULL);
</pre>
</div>
<p> This would wrap <tt>execlp()</tt> as a function that accepted up to
3 optional arguments. Depending on the application, this may be more
than enough for practical purposes.</p>
<p> The handling of <a href="#SWIGPlus_default_args">default arguments</a>
can be changed via the <tt>compactdefaultargs</tt> feature. If this
feature is used, for example</p>
<div class="code">
<pre>
%feature(&quot;compactdefaultargs&quot;) execlp;
%varargs(3, char *str = NULL) execlp;
...
int execlp(const char *path, const char *arg, ...);
</pre>
</div>
<p> a call from the target language which does not provide the maximum
number of arguments, such as, <tt>execlp(&quot;a&quot;, &quot;b&quot;, &quot;c&quot;)</tt> will
generate C code which includes the missing default values, that is, <tt>
execlp(&quot;a&quot;, &quot;b&quot;, &quot;c&quot;, NULL, NULL)</tt>. If <tt>compactdefaultargs</tt>
is not used, then the generated code will be <tt>execlp(&quot;a&quot;, &quot;b&quot;, &quot;c&quot;)</tt>
. The former is useful for helping providing a sentinel to terminate the
argument list. However, this is not guaranteed, for example when a user
passes a non-NULL value for all the parameters. When using <tt>
compactdefaultargs</tt> it is possible to guarantee the NULL sentinel is
passed through the, <tt>numinputs=0</tt> <a href="#Typemaps_nn26">'in'
typemap attribute</a>, naming the<b> last parameter</b>. For example,</p>
<div class="code">
<pre>
%feature(&quot;compactdefaultargs&quot;) execlp;
%varargs(3, char *str = NULL) execlp;
%typemap(in, numinputs=0) char *str3 &quot;&quot;
...
int execlp(const char *path, const char *arg, ...);
</pre>
</div>
<p> Note that <tt>str3</tt> is the name of the last argument, as we have
used <tt>%varargs</tt> with 3. Now <tt>execlp(&quot;a&quot;, &quot;b&quot;, &quot;c&quot;, &quot;d&quot;, &quot;e&quot;)</tt>
will result in an error as one too many arguments has been passed, as
now only 2 additional 'str' arguments can be passed with the 3rd one
always using the specified default <tt>NULL</tt>.</p>
<p> Argument replacement is most appropriate in cases where the types of
the extra arguments are uniform and the maximum number of arguments are
known. Argument replacement is not as useful when working with
functions that accept mixed argument types such as <tt>printf()</tt>.
Providing general purpose wrappers to such functions presents special
problems (covered shortly).</p>
<h2><a name="Varargs_nn6">17.5 Varargs and typemaps</a></h2>
<p> Variable length arguments may be used in typemap specifications. For
example:</p>
<div class="code">
<pre>
%typemap(in) (...) {
// Get variable length arguments (somehow)
...
}
%typemap(in) (const char *fmt, ...) {
// Multi-argument typemap
}
</pre>
</div>
<p> However, this immediately raises the question of what &quot;type&quot; is
actually used to represent <tt>(...)</tt>. For lack of a better
alternative, the type of <tt>(...)</tt> is set to <tt>void *</tt>.
Since there is no way to dynamically pass arguments to a varargs
function (as previously described), the <tt>void *</tt> argument value
is intended to serve as a place holder for storing some kind of
information about the extra arguments (if any). In addition, the
default behavior of SWIG is to pass the <tt>void *</tt> value as an
argument to the function. Therefore, you could use the pointer to hold
a valid argument value if you wanted.</p>
<p> To illustrate, here is a safer version of wrapping <tt>printf()</tt>
in Python:</p>
<div class="code">
<pre>
%typemap(in) (const char *fmt, ...) {
$1 = &quot;%s&quot;; /* Fix format string to %s */
$2 = (void *) PyString_AsString($input); /* Get string argument */
};
...
int printf(const char *fmt, ...);
</pre>
</div>
<p> In this example, the format string is implicitly set to <tt>&quot;%s&quot;</tt>
. This prevents a program from passing a bogus format string to the
extension. Then, the passed input object is decoded and placed in the <tt>
void *</tt> argument defined for the <tt>(...)</tt> argument. When the
actual function call is made, the underlying wrapper code will look
roughly like this:</p>
<div class="code">
<pre>
wrap_printf() {
char *arg1;
void *arg2;
int result;
arg1 = &quot;%s&quot;;
arg2 = (void *) PyString_AsString(arg2obj);
...
result = printf(arg1, arg2);
...
}
</pre>
</div>
<p> Notice how both arguments are passed to the function and it does
what you would expect.</p>
<p> The next example illustrates a more advanced kind of varargs
typemap. Disclaimer: this requires special support in the target
language module and is not guaranteed to work with all SWIG modules at
this time. It also starts to illustrate some of the more fundamental
problems with supporting varargs in more generality.</p>
<p> If a typemap is defined for any form of <tt>(...)</tt>, many SWIG
modules will generate wrappers that accept a variable number of
arguments as input and will make these arguments available in some
form. The precise details of this depends on the language module being
used (consult the appropriate chapter for more details). However,
suppose that you wanted to create a Python wrapper for the <tt>execlp()</tt>
function shown earlier. To do this using a typemap instead of using <tt>
%varargs</tt>, you might first write a typemap like this:</p>
<div class="code">
<pre>
%typemap(in) (...)(char *vargs[10]) {
int i;
Py_ssize_t argc;
for (i = 0; i &lt; 10; i++) vargs[i] = 0;
argc = PyTuple_Size(varargs);
if (argc &gt; 10) {
PyErr_SetString(PyExc_ValueError, &quot;Too many arguments&quot;);
SWIG_fail;
}
for (i = 0; i &lt; argc; i++) {
PyObject *pyobj = PyTuple_GetItem(varargs, i);
char *str = 0;
%#if PY_VERSION_HEX&gt;=0x03000000
const char *strtmp = 0;
PyObject *pystr;
if (!PyUnicode_Check(pyobj)) {
PyErr_SetString(PyExc_ValueError, &quot;Expected a string&quot;);
SWIG_fail;
}
pystr = PyUnicode_AsUTF8String(pyobj);
if (!pystr) {
SWIG_fail;
}
strtmp = PyBytes_AsString(pystr);
str = (char *)malloc(strlen(strtmp) + 1);
if (str)
strcpy(str, strtmp);
Py_DecRef(pystr);
%#else
if (!PyString_Check(pyobj)) {
PyErr_SetString(PyExc_ValueError, &quot;Expected a string&quot;);
SWIG_fail;
}
str = PyString_AsString(pyobj);
%#endif
vargs[i] = str;
}
$1 = (void *)vargs;
}
%typemap(freearg) (...) {
%#if PY_VERSION_HEX&gt;=0x03000000
int i;
for (i = 0; i &lt; 10; i++) {
free(vargs$argnum[i]);
}
%#endif
}
</pre>
</div>
<p> In the 'in' typemap, the special variable <tt>varargs</tt> is a
tuple holding all of the extra arguments passed (this is specific to
the Python module). The typemap then pulls this apart and sticks the
values into the array of strings <tt>args</tt>. Then, the array is
assigned to <tt>$1</tt> (recall that this is the <tt>void *</tt>
variable corresponding to <tt>(...)</tt>). However, this assignment is
only half of the picture----clearly this alone is not enough to make
the function work. The 'freearg' typemap cleans up memory allocated in
the 'in' typemap; this code is generated to be called after the <tt>
execlp</tt> function is called. To patch everything up, you have to
rewrite the underlying action code using the <tt>%feature</tt>
directive like this:</p>
<div class="code">
<pre>
%feature(&quot;action&quot;) execlp {
char **vargs = (char **) arg3;
result = execlp(arg1, arg2, vargs[0], vargs[1], vargs[2], vargs[3], vargs[4],
vargs[5], vargs[6], vargs[7], vargs[8], vargs[9], NULL);
}
int execlp(const char *path, const char *arg, ...);
</pre>
</div>
<p> This patches everything up and creates a function that more or less
works. However, don't try explaining this to your coworkers unless you
know for certain that they've had several cups of coffee. If you really
want to elevate your guru status and increase your job security,
continue to the next section.</p>
<h2><a name="Varargs_nn7">17.6 Varargs wrapping with libffi</a></h2>
<p> All of the previous examples have relied on features of SWIG that
are portable and which don't rely upon any low-level machine-level
details. In many ways, they have all dodged the real issue of variable
length arguments by recasting a varargs function into some weaker
variation with a fixed number of arguments of known types. In many
cases, this works perfectly fine. However, if you want more generality
than this, you need to bring out some bigger guns.</p>
<p> One way to do this is to use a special purpose library such as
libffi (<a href="https://www.sourceware.org/libffi/">
https://www.sourceware.org/libffi/</a>). libffi is a library that allows
you to dynamically construct call-stacks and invoke procedures in a
relatively platform independent manner. Details about the library can
be found in the libffi distribution and are not repeated here.</p>
<p> To illustrate the use of libffi, suppose that you<em> really</em>
wanted to create a wrapper for <tt>execlp()</tt> that accepted<em> any</em>
number of arguments. To do this, you might make a few adjustments to
the previous example. For example:</p>
<div class="code">
<pre>
/* Take an arbitrary number of extra arguments and place into an array
of strings */
%typemap(in) (...) {
char **argv;
int argc;
int i;
argc = PyTuple_Size(varargs);
argv = (char **) malloc(sizeof(char *)*(argc+1));
for (i = 0; i &lt; argc; i++) {
PyObject *o = PyTuple_GetItem(varargs, i);
if (!PyString_Check(o)) {
free(argv);
PyErr_SetString(PyExc_ValueError, &quot;Expected a string&quot;);
SWIG_fail;
}
argv[i] = PyString_AsString(o);
}
argv[i] = NULL;
$1 = (void *) argv;
}
/* Rewrite the function call, using libffi */
%feature(&quot;action&quot;) execlp {
int i, vc;
ffi_cif cif;
ffi_type **types;
void **values;
char **args;
vc = PyTuple_Size(varargs);
types = (ffi_type **) malloc((vc+3)*sizeof(ffi_type *));
values = (void **) malloc((vc+3)*sizeof(void *));
args = (char **) arg3;
/* Set up path parameter */
types[0] = &amp;ffi_type_pointer;
values[0] = &amp;arg1;
/* Set up first argument */
types[1] = &amp;ffi_type_pointer;
values[1] = &amp;arg2;
/* Set up rest of parameters */
for (i = 0; i &lt;= vc; i++) {
types[2+i] = &amp;ffi_type_pointer;
values[2+i] = &amp;args[i];
}
if (ffi_prep_cif(&amp;cif, FFI_DEFAULT_ABI, vc+3,
&amp;ffi_type_uint, types) == FFI_OK) {
ffi_call(&amp;cif, (void (*)()) execlp, &amp;result, values);
} else {
free(types);
free(values);
free(arg3);
PyErr_SetString(PyExc_RuntimeError, &quot;Whoa!!!!!&quot;);
SWIG_fail;
}
free(types);
free(values);
free(arg3);
}
/* Declare the function. Whew! */
int execlp(const char *path, const char *arg1, ...);
</pre>
</div>
<p> Looking at this example, you may start to wonder if SWIG is making
life any easier. Given the amount of code involved, you might also
wonder why you didn't just write a hand-crafted wrapper! Either that or
you're wondering &quot;why in the hell am I trying to wrap this varargs
function in the first place?!?&quot; Obviously, those are questions you'll
have to answer for yourself.</p>
<p> As a more extreme example of libffi, here is some code that attempts
to wrap <tt>printf()</tt>,</p>
<div class="code">
<pre>
/* A wrapper for printf() using libffi */
%{
/* Structure for holding passed arguments after conversion */
typedef struct {
int type;
union {
int ivalue;
double dvalue;
void *pvalue;
} val;
} vtype;
enum { VT_INT, VT_DOUBLE, VT_POINTER };
%}
%typemap(in) (const char *fmt, ...) {
vtype *argv;
int argc;
int i;
/* Format string */
$1 = PyString_AsString($input);
/* Variable length arguments */
argc = PyTuple_Size(varargs);
argv = (vtype *) malloc(argc*sizeof(vtype));
for (i = 0; i &lt; argc; i++) {
PyObject *o = PyTuple_GetItem(varargs, i);
if (PyInt_Check(o)) {
argv[i].type = VT_INT;
argv[i].val.ivalue = PyInt_AsLong(o);
} else if (PyFloat_Check(o)) {
argv[i].type = VT_DOUBLE;
argv[i].val.dvalue = PyFloat_AsDouble(o);
} else if (PyString_Check(o)) {
argv[i].type = VT_POINTER;
argv[i].val.pvalue = (void *) PyString_AsString(o);
} else {
free(argv);
PyErr_SetString(PyExc_ValueError, &quot;Unsupported argument type&quot;);
return NULL;
}
}
$2 = (void *) argv;
}
/* Rewrite the function call using libffi */
%feature(&quot;action&quot;) printf {
int i, vc;
ffi_cif cif;
ffi_type **types;
void **values;
vtype *args;
vc = PyTuple_Size(varargs);
types = (ffi_type **) malloc((vc+1)*sizeof(ffi_type *));
values = (void **) malloc((vc+1)*sizeof(void *));
args = (vtype *) arg2;
/* Set up fmt parameter */
types[0] = &amp;ffi_type_pointer;
values[0] = &amp;arg1;
/* Set up rest of parameters */
for (i = 0; i &lt; vc; i++) {
switch(args[i].type) {
case VT_INT:
types[1+i] = &amp;ffi_type_uint;
values[1+i] = &amp;args[i].val.ivalue;
break;
case VT_DOUBLE:
types[1+i] = &amp;ffi_type_double;
values[1+i] = &amp;args[i].val.dvalue;
break;
case VT_POINTER:
types[1+i] = &amp;ffi_type_pointer;
values[1+i] = &amp;args[i].val.pvalue;
break;
default:
abort(); /* Whoa! We're seriously hosed */
break;
}
}
if (ffi_prep_cif(&amp;cif, FFI_DEFAULT_ABI, vc+1,
&amp;ffi_type_uint, types) == FFI_OK) {
ffi_call(&amp;cif, (void (*)()) printf, &amp;result, values);
} else {
free(types);
free(values);
free(args);
PyErr_SetString(PyExc_RuntimeError, &quot;Whoa!!!!!&quot;);
SWIG_fail;
}
free(types);
free(values);
free(args);
}
/* The function */
int printf(const char *fmt, ...);
</pre>
</div>
<p> Much to your amazement, it even seems to work if you try it:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
&gt;&gt;&gt; example.printf(&quot;Grade: %s %d/60 = %0.2f%%\n&quot;, &quot;Dave&quot;, 47, 47.0*100/60)
Grade: Dave 47/60 = 78.33%
&gt;&gt;&gt;
</pre>
</div>
<p> Of course, there are still some limitations to consider:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; example.printf(&quot;la de da de da %s&quot;, 42)
Segmentation fault (core dumped)
</pre>
</div>
<p> And, on this note, we leave further exploration of libffi to the
reader as an exercise. Although Python has been used as an example,
most of the techniques in this section can be extrapolated to other
language modules with a bit of work. The only details you need to know
is how the extra arguments are accessed in each target language. For
example, in the Python module, we used the special <tt>varargs</tt>
variable to get these arguments. Modules such as Tcl8 and Perl5 simply
provide an argument number for the first extra argument. This can be
used to index into an array of passed arguments to get values. Please
consult the chapter on each language module for more details.</p>
<h2><a name="Varargs_nn8">17.7 Wrapping of va_list</a></h2>
<p> Closely related to variable length argument wrapping, you may
encounter functions that accept a parameter of type <tt>va_list</tt>.
For example:</p>
<div class="code">
<pre>
int vprintf(const char *fmt, va_list ap);
</pre>
</div>
<p> As far as we know, there is no obvious way to wrap these functions
with SWIG. This is because there is no documented way to assemble the
proper va_list structure (there are no C library functions to do it and
the contents of va_list are opaque). Not only that, the contents of a <tt>
va_list</tt> structure are closely tied to the underlying call-stack.
It's not clear that exporting a <tt>va_list</tt> would have any use or
that it would work at all.</p>
<p> A workaround can be implemented by writing a simple varargs C
wrapper and then using the techniques discussed earlier in this chapter
for varargs. Below is a simple wrapper for <tt>vprintf</tt> renamed so
that it can still be called as <tt>vprintf</tt> from your target
language. The <tt>%varargs</tt> used in the example restricts the
function to taking one string argument.</p>
<div class="code">
<pre>
%{
int vprintf(const char *fmt, va_list ap);
%}
%varargs(const char *) my_vprintf;
%rename(vprintf) my_vprintf;
%inline %{
int my_vprintf(const char *fmt, ...) {
va_list ap;
int result;
va_start(ap, fmt);
result = vprintf(fmt, ap);
va_end(ap);
return result;
}
%}
</pre>
</div>
<h2><a name="Varargs_nn9">17.8 C++ Issues</a></h2>
<p> Wrapping of C++ member functions that accept a variable number of
arguments presents a number of challenges. By far, the easiest way to
handle this is to use the <tt>%varargs</tt> directive. This is portable
and it fully supports classes much like the <tt>%rename</tt> directive.
For example:</p>
<div class="code">
<pre>
%varargs (10, char * = NULL) Foo::bar;
class Foo {
public:
virtual void bar(char *arg, ...); // gets varargs above
};
class Spam: public Foo {
public:
virtual void bar(char *arg, ...); // gets varargs above
};
</pre>
</div>
<p> <tt>%varargs</tt> also works with constructors, operators, and any
other C++ programming construct that accepts variable arguments.</p>
<p> Doing anything more advanced than this is likely to involve a
serious world of pain. In order to use a library like libffi, you will
need to know the underlying calling conventions and details of the C++
ABI. For instance, the details of how <tt>this</tt> is passed to member
functions as well as any hidden arguments that might be used to pass
additional information. These details are implementation specific and
may differ between compilers and even different versions of the same
compiler. Also, be aware that invoking a member function is further
complicated if it is a virtual method. In this case, invocation might
require a table lookup to obtain the proper function address (although
you might be able to obtain an address by casting a bound pointer to a
pointer to function as described in the C++ ARM section 18.3.4).</p>
<p> If you do decide to change the underlying action code, be aware that
SWIG always places the <tt>this</tt> pointer in <tt>arg1</tt>. Other
arguments are placed in <tt>arg2</tt>, <tt>arg3</tt>, and so forth. For
example:</p>
<div class="code">
<pre>
%feature(&quot;action&quot;) Foo::bar {
...
result = arg1-&gt;bar(arg2, arg3, etc.);
...
}
</pre>
</div>
<p> Given the potential to shoot yourself in the foot, it is probably
easier to reconsider your design or to provide an alternative interface
using a helper function than it is to create a fully general wrapper to
a varargs C++ member function.</p>
<h2><a name="Varargs_nn10">17.9 Discussion</a></h2>
<p> This chapter has provided a number of techniques that can be used to
address the problem of variable length argument wrapping. If you care
about portability and ease of use, the <tt>%varargs</tt> directive is
probably the easiest way to tackle the problem. However, using
typemaps, it is possible to do some very advanced kinds of wrapping.</p>
<p> One point of discussion concerns the structure of the libffi
examples in the previous section. Looking at that code, it is not at
all clear that this is the easiest way to solve the problem. However,
there are a number of subtle aspects of the solution to
consider--mostly concerning the way in which the problem has been
decomposed. First, the example is structured in a way that tries to
maintain separation between wrapper-specific information and the
declaration of the function itself. The idea here is that you might
structure your interface like this:</p>
<div class="code">
<pre>
%typemap(const char *fmt, ...) {
...
}
%feature(&quot;action&quot;) traceprintf {
...
}
/* Include some header file with traceprintf in it */
%include &quot;someheader.h&quot;
</pre>
</div>
<p> Second, careful scrutiny will reveal that the typemaps involving <tt>
(...)</tt> have nothing whatsoever to do with the libffi library. In
fact, they are generic with respect to the way in which the function is
actually called. This decoupling means that it will be much easier to
consider other library alternatives for making the function call. For
instance, if libffi wasn't supported on a certain platform, you might
be able to use something else instead. You could use conditional
compilation to control this:</p>
<div class="code">
<pre>
#ifdef USE_LIBFFI
%feature(&quot;action&quot;) printf {
...
}
#endif
#ifdef USE_OTHERFFI
%feature(&quot;action&quot;) printf {
...
}
#endif
</pre>
</div>
<p> Finally, even though you might be inclined to just write a
hand-written wrapper for varargs functions, the techniques used in the
previous section have the advantage of being compatible with all other
features of SWIG such as exception handling.</p>
<p> As a final word, some C programmers seem to have the assumption that
the wrapping of variable length argument functions is an easily solved
problem. However, this section has hopefully dispelled some of these
myths. All things being equal, you are better off avoiding variable
length arguments if you can. If you can't avoid them, please consider
some of the simple solutions first. If you can't live with a simple
solution, proceed with caution. At the very least, make sure you
carefully read the section &quot;A7.3.2 Function Calls&quot; in Kernighan and
Ritchie and make sure you fully understand the parameter passing
conventions used for varargs. Also, be aware of the platform
dependencies and reliability issues that this will introduce. Good
luck.</p>
<HR NOSHADE>
<h1><a name="Doxygen">18 SWIG and Doxygen Translation</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Doxygen_translation_overview">Doxygen translation overview</a>
</li>
<li><a href="#Doxygen_file_preparation">Preparations</a>
<ul>
<li><a href="#Doxygen_running_swig">Enabling Doxygen translation</a></li>
<li><a href="#Doxygen_features">Doxygen-specific %feature directives</a>
<ul>
<li><a href="#Doxygen_notranslate">doxygen:notranslate</a></li>
<li><a href="#Doxygen_alias">doxygen:alias:&lt;command-name&gt;</a></li>
<li><a href="#Doxygen_ignore">doxygen:ignore:&lt;command-name&gt;</a></li>
<li><a href="#Doxygen_nolinktranslate">doxygen:nolinktranslate</a></li>
<li><a href="#Doxygen_nostripparams">doxygen:nostripparams</a></li>
</ul>
</li>
<li><a href="#Doxygen_additional_options">Additional command line
options</a></li>
</ul>
</li>
<li><a href="#Doxygen_javadoc">Doxygen to Javadoc</a>
<ul>
<li><a href="#Doxygen_javadoc_basic_example">Basic Javadoc example</a></li>
<li><a href="#Doxygen_javadoc_tags">Javadoc tags</a></li>
<li><a href="#Doxygen_javadoc_unsupported_tags">Unsupported tags for
Javadoc</a></li>
</ul>
</li>
<li><a href="#Doxygen_pydoc">Doxygen to Pydoc</a>
<ul>
<li><a href="#Doxygen_pydoc_basic_example">Basic Pydoc example</a></li>
<li><a href="#Doxygen_pydoc_tags">Pydoc translator</a></li>
<li><a href="#Doxygen_pydoc_unsupported_tags">Unsupported tags for Pydoc</a>
</li>
</ul>
</li>
<li><a href="#Doxygen_csharp">Doxygen to XML C# documentation</a>
<ul>
<li><a href="#Doxygen_csharp_basic_example">Basic C# example</a></li>
<li><a href="#Doxygen_csharp_tags">C# tags</a></li>
<li><a href="#Doxygen_csharp_unsupported_tags">Unsupported tags for C#</a>
</li>
</ul>
</li>
<li><a href="#Doxygen_troubleshooting">Troubleshooting</a>
<ul>
<li><a href="#troubleshooting_ifndef">Problem with conditional
compilation</a></li>
</ul>
</li>
<li><a href="#Doxygen_developer_details">Developer information</a>
<ul>
<li><a href="#Doxygen_translator_design">Doxygen translator design</a></li>
<li><a href="#Doxygen_debugging_commands">Debugging the Doxygen parser
and translator</a></li>
<li><a href="#Doxygen_tests">Tests</a></li>
</ul>
</li>
<li><a href="#Doxygen_language_extension">Extending to other languages</a>
</li>
</ul>
</div>
<!-- INDEX -->
<p> This chapter describes SWIG's support for translating Doxygen
comments found in interface and header files into a target language's
normal documentation language. Currently only Javadoc and Pydoc is
supported.</p>
<h2><a name="Doxygen_translation_overview">18.1 Doxygen translation
overview</a></h2>
<p> The Doxygen Translation module of SWIG adds an extra layer of
functionality to SWIG, allowing automated translation of <a href="https://www.doxygen.nl/manual/">
Doxygen</a> formatted comments from input files into a documentation
language more suited for the target language. Currently this module
only translates into Javadoc and Pydoc for the SWIG Java and Python
modules. Other extensions could be added at a later date. The Doxygen
Translation module originally started as a <a href="https://developers.google.com/open-source/gsoc/2008/">
Google Summer of Code</a> proposal from Summer 2008.</p>
<h2><a name="Doxygen_file_preparation">18.2 Preparations</a></h2>
<p> To make use of the comment translation system, your documentation
comments must be in properly formatted <a href="https://www.doxygen.nl/manual/">
Doxygen.</a> Doxygen comments can be present in your main SWIG interface
file or any header file that it imports. You are advised to be validate
that your comments compile properly with Doxygen before you try to
translate them. Doxygen itself is a more comprehensive tool and can
provide you better feedback for correcting any syntax errors that may
be present. Please look at Doxygen's <a href="https://www.doxygen.nl/manual/docblocks.html">
Documenting the code</a> for the full comment format specifications.
However, SWIG's Doxygen parser will still report many errors and
warnings found in comments (like unterminated strings or missing ending
tags).</p>
<p> Currently, the whole subset of Doxygen comment styles is supported
(See <a href="https://www.doxygen.nl/manual/docblocks.html">
Documenting the code</a>). Here they are:</p>
<div class="code">
<pre>
/**
* Javadoc style comment, multiline
*/
/*!
* QT-style comment, multiline
*/
/**
Any of the above, but without intermediate *'s
*/
/// Single-line comment
//! Another single-line comment
</pre>
</div>
<p> Also any of the above with '<tt>&lt;</tt>' added after comment-starting
symbol, like <tt>/**&lt;, /*!&lt;, ///&lt;,</tt> or <tt>//!&lt;</tt> will be
treated as a post-comment and will be assigned to the code before the
comment. Any number of '<tt>*</tt>' or '<tt>/</tt>' within a Doxygen
comment is considered to be a separator and is not included in the
final comment, so you may safely use comments like <tt>/*********/</tt>
or <tt>//////////</tt>.</p>
<p> Please note, as SWIG parses the input file by itself with strict
grammar, there is only a limited support for various cases of comment
placement in the file.</p>
<p> Comments can be placed before C/C++ expressions on separate lines:</p>
<div class="code">
<pre>
/**
* Some comment
*/
void someOtherFunction();
/**
* Some comment
*/
void someFunction();
class Shape {
/*
* Calculate the area in cm^2
*/
int getArea();
}
</pre>
</div>
<p> After C/C++ expressions at the end of the line:</p>
<div class="code">
<pre>
int someVariable = 9; ///&lt; This is a var holding magic number 9
void doNothing(); ///&lt; This does nothing, nop
</pre>
</div>
<p> and in some special cases, like function parameter comments:</p>
<div class="code">
<pre>
void someFunction(
int a ///&lt; Some parameter
);
</pre>
</div>
<p> or enum element comments:</p>
<div class="code">
<pre>
enum E_NUMBERS
{
EN_ZERO, ///&lt; The first enum item, gets zero as its value
EN_ONE, ///&lt; The second, EN_ONE=1
EN_THREE
};
</pre>
</div>
<p> Currently only comments directly before or after the code items are
supported. Doxygen also supports comments containing structural
commands, where the comments for a code item are not put directly
before or after the code item. These structural commands are stripped
out by SWIG and are not assigned to anything.</p>
<h3><a name="Doxygen_running_swig">18.2.1 Enabling Doxygen translation</a>
</h3>
<p> Doxygen comments translation is disabled by default and needs to be
explicitly enabled using the command line <tt>-doxygen</tt> option for
the languages that do support it (currently Java and Python).</p>
<h3><a name="Doxygen_features">18.2.2 Doxygen-specific %feature
directives</a></h3>
<p> Translation of Doxygen comments is influenced by the following <a href="#Customization_features">
%feature directives</a>:</p>
<h4><a name="Doxygen_notranslate">18.2.2.1 doxygen:notranslate</a></h4>
<p> Turns off translation of Doxygen comments to the target language
syntax: the original comment will be copied to the output unchanged.
This is useful if you want to use Doxygen itself to generate
documentation for the target language instead of the corresponding
language tool (<tt>javadoc</tt>, <tt>sphinx</tt>, ...).</p>
<h4><a name="Doxygen_alias">18.2.2.2 doxygen:alias:&lt;command-name&gt;</a></h4>
<p> Specify an alias for a Doxygen command with the given name. This can
be useful for custom Doxygen commands which can be defined using <tt>
ALIASES</tt> option for Doxygen itself but which are unknown to SWIG. <tt>
&quot;command-name&quot;</tt> is the name of the command in the Doxyfile, e.g. if
it contains</p>
<div class="code">
<pre>
ALIASES = &quot;sideeffect=\par Side Effects:\n&quot;
</pre>
</div>
<p> Then you could also specify the same expansion for SWIG with:</p>
<div class="code">
<pre>
%feature(&quot;doxygen:alias:sideeffect&quot;) &quot;\par Side Effects:\n&quot;
</pre>
</div>
<p> Please note that command arguments are not currently supported with
this feature.</p>
<p> Notice that it is perfectly possible and potentially useful to
define the alias expansion differently depending on the target
language, e.g. with</p>
<div class="code">
<pre>
#ifdef SWIGJAVA
%feature(&quot;doxygen:alias:not_for_java&quot;) &quot;This functionality is not available for Java&quot;
#else
%feature(&quot;doxygen:alias:not_for_java&quot;) &quot;&quot;
#endif
</pre>
</div>
<p> you could use <tt>@not_for_java</tt> in the documentation comments
of all functions which can't, for whatever reason, be currently exposed
in Java wrappers of the C++ API.</p>
<h4><a name="Doxygen_ignore">18.2.2.3 doxygen:ignore:&lt;command-name&gt;</a></h4>
<p> This feature makes it possible to just ignore an unknown Doxygen
command, instead of replacing it with the predefined text that <tt>
doxygen:alias</tt> does. For example, you could use</p>
<div class="code">
<pre>
%feature(&quot;doxygen:ignore:transferfull&quot;) Fantastic();
/**
A fantastic function.
@transferfull Command ignored, but anything here is still included.
*/
int * Fantastic();
</pre>
</div>
<p> if you use a custom Doxygen <tt>transferfull</tt> command to
indicate that the return value ownership is transferred to the caller,
as this information doesn't make much sense for the other languages
without explicit ownership management.</p>
<p> Doxygen syntax is rather rich and, in addition to simple commands
such as <tt>@transferfull</tt>, it is also possible to define commands
with arguments. As explained in <a href="https://www.doxygen.nl/manual/commands.html">
Doxygen documentation</a>, the arguments can have a range of a single
word, everything until the end of line or everything until the end of
the next paragraph. Currently, only the &quot;end of line&quot; case is supported
using the <tt>range=&quot;line&quot;</tt> argument of the feature directive:</p>
<div class="code">
<pre>
// Ignore occurrences of
//
// @compileroptions Some special C++ compiler options.
//
// in Doxygen comments as C++ options are not interesting for the target language
// developers.
%feature(&quot;doxygen:ignore:compileroptions&quot;, range=&quot;line&quot;) Amazing();
/**
An amazing function.
@compileroptions This function must be compiled with /EHa when using MSVC.
*/
void Amazing();
</pre>
</div>
<p> In addition, it is also possible to have custom pairs of begin/end
tags, similarly to the standard Doxygen <tt>@code/@endcode</tt>, for
example. Such tags can also be ignored using the special value of <tt>
range</tt> starting with <tt>end</tt> to indicate that the range is an
interval, for example:</p>
<div class="code">
<pre>
%feature(&quot;doxygen:ignore:forcpponly&quot;, range=&quot;end&quot;); // same as &quot;end:endforcpponly&quot;
/**
An incredible function.
@forcpponly
This is C++-specific.
@endforcpponly
*/
void Incredible();
</pre>
</div>
<p> would ignore everything between <tt>@forcpponly</tt> and <tt>
@endforcpponly</tt> commands in Doxygen comments. By default, the name
of the end command is the same as of the start one with &quot;end&quot; prefix,
following Doxygen conventions, but this can be overridden by providing
the end command name after the colon.</p>
<p> This example shows how custom tags can be used to bracket anything
specific to C++ and prevent it from appearing in the target language
documentation. Conversely, another pair of custom tags could be used to
put target language specific information in the C++ comments. In this
case, only the custom tags themselves should be ignored, but their
contents should be parsed as usual and <tt>contents=&quot;parse&quot;</tt> can be
used for this:</p>
<div class="code">
<pre>
%feature(&quot;doxygen:ignore:beginPythonOnly&quot;, range=&quot;end:endPythonOnly&quot;, contents=&quot;parse&quot;);
/**
A splendid function.
@beginPythonOnly
This is specific to @b Python.
@endPythonOnly
*/
void Splendid();
</pre>
</div>
<p> Putting everything together, if these directives are in effect:</p>
<div class="code">
<pre>
%feature(&quot;doxygen:ignore:transferfull&quot;);
%feature(&quot;doxygen:ignore:compileroptions&quot;, range=&quot;line&quot;);
%feature(&quot;doxygen:ignore:forcpponly&quot;, range=&quot;end&quot;);
%feature(&quot;doxygen:ignore:beginPythonOnly&quot;, range=&quot;end:endPythonOnly&quot;, contents=&quot;parse&quot;);
</pre>
</div>
<p> then the following C++ Doxygen comment:</p>
<div class="code">
<pre>
/**
A contrived example of ignoring too many commands in one comment.
@forcpponly
This is C++-specific.
@endforcpponly
@beginPythonOnly
This is specific to @b Python.
@endPythonOnly
@transferfull Command ignored, but anything here is still included.
@compileroptions This function must be compiled with /EHa when using MSVC.
*/
int * Contrived();
</pre>
</div>
<p> would be translated to this comment in Python:</p>
<div class="code">
<pre>
def func():
r&quot;&quot;&quot;
A contrived example of ignoring too many commands in one comment.
This is specific to **Python**.
Command ignored, but anything here is still included.
&quot;&quot;&quot;
...
</pre>
</div>
<h4><a name="Doxygen_nolinktranslate">18.2.2.4 doxygen:nolinktranslate</a>
</h4>
<p> Turn off automatic link-objects translation. This is only applicable
to Java at the moment.</p>
<h4><a name="Doxygen_nostripparams">18.2.2.5 doxygen:nostripparams</a></h4>
<p> Turn off stripping of <tt>@param</tt> and <tt>@tparam</tt> Doxygen
commands if the parameter is not found in the function signature. This
is only applicable to Java at the moment.</p>
<h3><a name="Doxygen_additional_options">18.2.3 Additional command line
options</a></h3>
<p> ALSO TO BE ADDED (Javadoc auto brief?)</p>
<h2><a name="Doxygen_javadoc">18.3 Doxygen to Javadoc</a></h2>
<p> If translation is enabled, Javadoc formatted comments should be
automatically placed in the correct locations in the resulting module
and proxy files.</p>
<h3><a name="Doxygen_javadoc_basic_example">18.3.1 Basic Javadoc example</a>
</h3>
<p> Here is an example segment from an included header file</p>
<div class="code">
<pre>
/*! This is describing class Shape
\author Bob
*/
class Shape {
public:
Shape() {
nshapes++;
}
virtual ~Shape() {
nshapes--;
};
double x, y; /*!&lt; Important Variables */
void move(double dx, double dy); /*!&lt; Moves the Shape */
virtual double area(void) = 0; /*!&lt; \return the area */
virtual double perimeter(void) = 0; /*!&lt; \return the perimeter */
static int nshapes;
};
</pre>
</div>
<p> Simply running SWIG should result in the following code being
present in Shapes.java</p>
<div class="targetlang">
<pre>
/**
* This is describing class Shape
* @author Bob
*
*/
public class Shape {
...
/**
* Important Variables
*/
public void setX(double value) {
ShapesJNI.Shape_x_set(swigCPtr, this, value);
}
/**
* Important Variables
*/
public double getX() {
return ShapesJNI.Shape_x_get(swigCPtr, this);
}
/**
* Moves the Shape
*/
public void move(double dx, double dy) {
ShapesJNI.Shape_move(swigCPtr, this, dx, dy);
}
/**
* @return the area
*/
public double area() {
return ShapesJNI.Shape_area(swigCPtr, this);
}
/**
* @return the perimeter
*/
public double perimeter() {
return ShapesJNI.Shape_perimeter(swigCPtr, this);
}
}
</pre>
</div>
<p> The code Java-wise should be identical to what would have been
generated without the doxygen functionality enabled. When the Doxygen
Translator module encounters a comment that contains nothing useful or
a doxygen comment that it cannot parse, it will not affect the
functionality of the SWIG generated code.</p>
<p> The Javadoc translator will handle most of the tags conversions (see
the table below). It will also automatically translate link-objects
params, in <tt>\see</tt> and <tt>\link...\endlink</tt> commands. For
example, <tt>someFunction(std::string)</tt> will be converted to <tt>
someFunction(String)</tt>. If you don't want such behaviour, you could
turn this off by using the <tt>doxygen:nolinktranslate</tt> feature.
Also all <tt>\param</tt> and <tt>\tparam</tt> commands are stripped
out, if the specified parameter is not present in the function. Use <tt>
doxygen:nostripparams</tt> to avoid.</p>
<p> Javadoc translator features summary (see <a href="#Customization_features">
%feature directives</a>):</p>
<h3><a name="Doxygen_javadoc_tags">18.3.2 Javadoc tags</a></h3>
<p> Here is the list of all Doxygen tags and the description of how they
are translated to Javadoc</p>
<div class="diagram">
<table border="0" summary="Java Doxygen tags">
<tr><th align="left">Doxygen tags</th></tr>
<tr><td>\a</td><td>wrapped with &lt;i&gt; html tag</td></tr>
<tr><td>\arg</td><td>wrapped with &lt;li&gt; html tag</td></tr>
<tr><td>\author</td><td>translated to @author</td></tr>
<tr><td>\authors</td><td>translated to @author</td></tr>
<tr><td>\b</td><td>wrapped with &lt;b&gt; html tag</td></tr>
<tr><td>\c</td><td>wrapped with &lt;code&gt; html tag</td></tr>
<tr><td>\cite</td><td>wrapped with &lt;i&gt; html tag</td></tr>
<tr><td>\code</td><td>translated to {@code ...}</td></tr>
<tr><td>\code{&lt;ext&gt;}</td><td>translated to {@code ...}; code language
extension is ignored</td></tr>
<tr><td>\cond</td><td>translated to 'Conditional comment: &lt;condition&gt;'</td>
</tr>
<tr><td>\copyright</td><td>replaced with 'Copyright:'</td></tr>
<tr><td>\deprecated</td><td>translated to @deprecated</td></tr>
<tr><td>\e</td><td>wrapped with &lt;i&gt; html tag</td></tr>
<tr><td>\else</td><td>replaced with '}Else:{'</td></tr>
<tr><td>\elseif</td><td>replaced with '}Else if: &lt;condition&gt;{'</td></tr>
<tr><td>\em</td><td>wrapped with &lt;i&gt; html tag</td></tr>
<tr><td>\endcode</td><td>see note for \code</td></tr>
<tr><td>\endcond</td><td>replaced with 'End of conditional comment.'</td>
</tr>
<tr><td>\endif</td><td>replaced with '}'</td></tr>
<tr><td>\endlink</td><td>see note for \link</td></tr>
<tr><td>\endverbatim</td><td>see note for \verbatim</td></tr>
<tr><td>\exception</td><td>translated to @exception</td></tr>
<tr><td>\f$, \f[, \f], \f{, \f}</td><td>LateX formulas are left
unchanged</td></tr>
<tr><td>\if</td><td>replaced with 'If: &lt;condition&gt; {'</td></tr>
<tr><td>\ifnot</td><td>replaced with 'If not: &lt;condition&gt; {'</td></tr>
<tr><td>\image</td><td>translated to &lt;img/&gt; html tag only if target=HTML</td>
</tr>
<tr><td>\li</td><td>wrapped with &lt;li&gt; html tag</td></tr>
<tr><td>\link</td><td>translated to {@link ...}</td></tr>
<tr><td>\n</td><td>replaced with newline char</td></tr>
<tr><td>\note</td><td>replaced with 'Note:'</td></tr>
<tr><td>\overload</td><td>prints 'This is an overloaded ...' according
to Doxygen docs</td></tr>
<tr><td>\p</td><td>wrapped with &lt;code&gt; html tag</td></tr>
<tr><td>\par</td><td>replaced with &lt;p alt='title'&gt;...&lt;/p&gt;</td></tr>
<tr><td>\param</td><td>translated to @param</td></tr>
<tr><td>\param[&lt;dir&gt;]</td><td>translated to @param; parameter direction
('in'; 'out'; or 'in,out') is ignored</td></tr>
<tr><td>\remark</td><td>replaced with 'Remarks:'</td></tr>
<tr><td>\remarks</td><td>replaced with 'Remarks:'</td></tr>
<tr><td>\result</td><td>translated to @return</td></tr>
<tr><td>\return</td><td>translated to @return</td></tr>
<tr><td>\returns</td><td>translated to @return</td></tr>
<tr><td>\sa</td><td>translated to @see</td></tr>
<tr><td>\see</td><td>translated to @see</td></tr>
<tr><td>\since</td><td>translated to @since</td></tr>
<tr><td>\throw</td><td>translated to @throws</td></tr>
<tr><td>\throws</td><td>translated to @throws</td></tr>
<tr><td>\todo</td><td>replaced with 'TODO:'</td></tr>
<tr><td>\tparam</td><td>translated to @param</td></tr>
<tr><td>\verbatim</td><td>translated to {@literal ...}</td></tr>
<tr><td>\version</td><td>translated to @version</td></tr>
<tr><td>\warning</td><td>translated to 'Warning:'</td></tr>
<tr><td>\$</td><td>prints $ char</td></tr>
<tr><td>\@</td><td>prints @ char</td></tr>
<tr><td>\\</td><td>prints \ char</td></tr>
<tr><td>\&amp;</td><td>prints &amp; char</td></tr>
<tr><td>\~</td><td>prints ~ char</td></tr>
<tr><td>\&lt;</td><td>prints &lt; char</td></tr>
<tr><td>\&gt;</td><td>prints &gt; char</td></tr>
<tr><td>\#</td><td>prints # char</td></tr>
<tr><td>\%</td><td>prints % char</td></tr>
<tr><td>\&quot;</td><td>prints &quot; char</td></tr>
<tr><td>\.</td><td>prints . char</td></tr>
<tr><td>\::</td><td>prints ::</td></tr>
</table>
</div>
<h3><a name="Doxygen_javadoc_unsupported_tags">18.3.3 Unsupported tags
for Javadoc</a></h3>
<p> Doxygen has a wealth of tags such as <tt>@latexonly</tt> that have
no equivalent in Javadoc (all supported tags are listed in <a href="https://docs.oracle.com/javase/7/docs/technotes/tools/windows/javadoc.html">
Javadoc documentation</a>). As a result several tags have no translation
or particular use, such as some linking and section tags. These are
suppressed with their content just printed out (if the tag has any
sense, typically text content). Here is the list of these tags:</p>
<div class="diagram"><b> Unsupported Doxygen tags</b>
<ul style="list-style-type:none;column-count:4;">
<li>\addindex</li>
<li>\addtogroup</li>
<li>\anchor</li>
<li>\attention</li>
<li>\brief</li>
<li>\bug</li>
<li>\callergraph</li>
<li>\callgraph</li>
<li>\category</li>
<li>\class</li>
<li>\copybrief</li>
<li>\copydetails</li>
<li>\copydoc</li>
<li>\date</li>
<li>\def</li>
<li>\defgroup</li>
<li>\details</li>
<li>\dir</li>
<li>\dontinclude</li>
<li>\dot</li>
<li>\dotfile</li>
<li>\enddot</li>
<li>\endhtmlonly</li>
<li>\endinternal</li>
<li>\endlatexonly</li>
<li>\endmanonly</li>
<li>\endmsc</li>
<li>\endrtfonly</li>
<li>\endxmlonly</li>
<li>\enum</li>
<li>\example</li>
<li>\extends</li>
<li>\file</li>
<li>\fn</li>
<li>\headerfile</li>
<li>\hideinitializer</li>
<li>\htmlinclude</li>
<li>\htmlonly</li>
<li>\implements</li>
<li>\include</li>
<li>\includelineno</li>
<li>\ingroup</li>
<li>\interface</li>
<li>\internal</li>
<li>\invariant</li>
<li>\latexonly</li>
<li>\line</li>
<li>\mainpage</li>
<li>\manonly</li>
<li>\memberof</li>
<li>\msc</li>
<li>\mscfile</li>
<li>\name</li>
<li>\namespace</li>
<li>\nosubgrouping</li>
<li>\package</li>
<li>\page</li>
<li>\paragraph</li>
<li>\post</li>
<li>\pre</li>
<li>\private</li>
<li>\privatesection</li>
<li>\property</li>
<li>\protected</li>
<li>\protectedsection</li>
<li>\protocol</li>
<li>\public</li>
<li>\publicsection</li>
<li>\ref</li>
<li>\related</li>
<li>\relatedalso</li>
<li>\relates</li>
<li>\relatesalso</li>
<li>\retval</li>
<li>\rtfonly</li>
<li>\section</li>
<li>\short</li>
<li>\showinitializer</li>
<li>\skip</li>
<li>\skipline</li>
<li>\snippet</li>
<li>\struct</li>
<li>\subpage</li>
<li>\subsection</li>
<li>\subsubsection</li>
<li>\tableofcontents</li>
<li>\test</li>
<li>\typedef</li>
<li>\union</li>
<li>\until</li>
<li>\var</li>
<li>\verbinclude</li>
<li>\weakgroup</li>
<li>\xmlonly</li>
<li>\xrefitem</li>
</ul>
</div>
<p> If one of the following Doxygen tags appears as the first tag in a
comment, the whole comment block is ignored:
<!-- see parser.y, function isStructuralDoxygen() -->
</p>
<div class="diagram"><b> Ignored Doxygen tags</b>
<ul style="list-style-type:none;column-count:4;">
<li>\addtogroup</li>
<li>\callergraph</li>
<li>\callgraph</li>
<li>\category</li>
<li>\class</li>
<li>\def</li>
<li>\defgroup</li>
<li>\dir</li>
<li>\enum</li>
<li>\example</li>
<li>\file</li>
<li>\fn</li>
<li>\headerfile</li>
<li>\hideinitializer</li>
<li>\interface</li>
<li>\internal</li>
<li>\mainpage</li>
<li>\name</li>
<li>\namespace</li>
<li>\nosubgrouping</li>
<li>\overload</li>
<li>\package</li>
<li>\page</li>
<li>\property</li>
<li>\protocol</li>
<li>\relates</li>
<li>\relatesalso</li>
<li>\showinitializer</li>
<li>\struct</li>
<li>\typedef</li>
<li>\union</li>
<li>\var</li>
<li>\weakgroup</li>
</ul>
</div>
<h2><a name="Doxygen_pydoc">18.4 Doxygen to Pydoc</a></h2>
<p> If translation is enabled, Pydoc formatted comments should be
automatically placed in the correct locations in the resulting module
and proxy files. The problem is that Pydoc has no tag mechanism like
Doxygen or Javadoc, so most of Doxygen commands are translated by
merely copying the appropriate command text.</p>
<h3><a name="Doxygen_pydoc_basic_example">18.4.1 Basic Pydoc example</a></h3>
<p> Here is an example segment from an included header file</p>
<div class="code">
<pre>
/*! This is describing class Shape
\author Bob
*/
class Shape {
public:
Shape() {
nshapes++;
}
virtual ~Shape() {
nshapes--;
};
double x, y; /*!&lt; Important Variables */
void move(double dx, double dy); /*!&lt; Moves the Shape */
virtual double area(void) = 0; /*!&lt; \return the area */
virtual double perimeter(void) = 0; /*!&lt; \return the perimeter */
static int nshapes;
};
</pre>
</div>
<p> Simply running SWIG should result in the following code being
present in Shapes.py</p>
<div class="targetlang">
<pre>
...
class Shape(_object):
&quot;&quot;&quot;
This is describing class Shape
Authors:
Bob
&quot;&quot;&quot;
...
def move(self, *args):
&quot;&quot;&quot;
Moves the Shape
&quot;&quot;&quot;
return _Shapes.Shape_move(self, *args)
def area(self):
&quot;&quot;&quot;
Return:
the area
&quot;&quot;&quot;
return _Shapes.Shape_area(self)
def perimeter(self):
&quot;&quot;&quot;
Return:
the perimeter
&quot;&quot;&quot;
return _Shapes.Shape_perimeter(self)
</pre>
</div>
<p> If any parameters of a function or a method are documented in the
Doxygen comment, their description is copied into the generated output
using <a href="https://www.sphinx-doc.org/">Sphinx</a> documentation
conventions. For example</p>
<div class="code">
<pre>
/**
Set a breakpoint at the given location.
@param filename The full path to the file.
@param line_number The line number in the file.
*/
bool SetBreakpoint(const char* filename, int line_number);
</pre>
</div>
<p> would be translated to</p>
<div class="targetlang">
<pre>
def SetBreakpoint(filename, line_number):
r&quot;&quot;&quot;
Set a breakpoint at the given location.
:type filename: string
:param filename: The full path to the file.
:type line_number: int
:param line_number: The line number in the file.
&quot;&quot;&quot;
</pre>
</div>
<p> The types used for the parameter documentation come from the
&quot;doctype&quot; typemap which is defined for all the primitive types and a
few others (e.g. <tt>std::string</tt> and <tt>shared_ptr&lt;T&gt;</tt>) but
for non-primitive types is taken to be just the C++ name of the type
with namespace scope delimiters (<tt>::</tt>) replaced with a dot. To
change this, you can define your own typemaps for the custom types,
e.g:</p>
<div class="code">
<pre>
%typemap(doctype) MyDate &quot;datetime.date&quot;
</pre>
</div>
<p> Currently Doxygen comments assigned to global variables and static
member variables are not present in generated code, so they have no
comment translated for them.</p>
<p><b> Whitespace and tables</b> Whitespace is preserved when
translating comments, so it makes sense to have Doxygen comments
formatted in a readable way. This includes tables, where tags &lt;th&gt;,
&lt;td&gt; and &lt;/tr&gt;are translated to '|'. The line after line with &lt;th&gt; tags
contains dashes. If we take care about whitespace, comments in Python
are much more readable. Example:<div class="code">
<pre>
/**
* &lt;table border = '1'&gt;
* &lt;caption&gt;Animals&lt;/caption&gt;
* &lt;tr&gt;&lt;th&gt; Column 1 &lt;/th&gt;&lt;th&gt; Column 2 &lt;/th&gt;&lt;/tr&gt;
* &lt;tr&gt;&lt;td&gt; cow &lt;/td&gt;&lt;td&gt; dog &lt;/td&gt;&lt;/tr&gt;
* &lt;tr&gt;&lt;td&gt; cat &lt;/td&gt;&lt;td&gt; mouse &lt;/td&gt;&lt;/tr&gt;
* &lt;tr&gt;&lt;td&gt; horse &lt;/td&gt;&lt;td&gt; parrot &lt;/td&gt;&lt;/tr&gt;
* &lt;/table&gt;
*/
</pre>
</div></p>
<p> translates to Python as:</p>
<div class="diagram">
<pre>
Animals
| Column 1 | Column 2 |
-----------------------
| cow | dog |
| cat | mouse |
| horse | parrot |
</pre>
</div>
<p><b> Overloaded functions</b> Since all the overloaded functions in
c++ are wrapped into one Python function, Pydoc translator will combine
every comment of every overloaded function and put it into the comment
for the one wrapper function.</p>
<p> If you intend to use resulting generated Python file with the
Doxygen docs generator, rather than Pydoc, you may want to turn off
translation completely (doxygen:notranslate feature). Then SWIG will
just copy the comments to the proxy file and reformat them if needed,
but all the comment content will be left as is. As Doxygen doesn't
support special commands in Python comments (see <a href="https://www.doxygen.nl/manual/docblocks.html#pythonblocks">
Doxygen docs</a>), you may want to use some tool like doxypy (<a href="https://pypi.org/project/doxypy/">
doxypy</a>) to do the work.</p>
<h3><a name="Doxygen_pydoc_tags">18.4.2 Pydoc translator</a></h3>
<p> Here is the list of all Doxygen tags and the description of how they
are translated to Pydoc</p>
<div class="diagram">
<table border="0" summary="Python Doxygen tags">
<tr><th align="left">Doxygen tags</th></tr>
<tr><td>\a</td><td>wrapped with '*'</td></tr>
<tr><td>\arg</td><td>prepended with '* '</td></tr>
<tr><td>\author</td><td>prints 'Author:'</td></tr>
<tr><td>\authors</td><td>prints 'Authors:'</td></tr>
<tr><td>\b</td><td>wrapped with '**'</td></tr>
<tr><td>\c</td><td>wrapped with '``'</td></tr>
<tr><td>\cite</td><td>wrapped with single quotes</td></tr>
<tr><td>\code</td><td>replaced with '.. code-block:: c++'</td></tr>
<tr><td>\code{&lt;ext&gt;}</td><td>replaced with '.. code-block:: &lt;lang&gt;',
where the following doxygen code languages are recognized: .c -&gt; C, .py
-&gt; python, .java &gt; java</td></tr>
<tr><td>\cond</td><td>translated to 'Conditional comment: &lt;condition&gt;'</td>
</tr>
<tr><td>\copyright</td><td>prints 'Copyright:'</td></tr>
<tr><td>\deprecated</td><td>prints 'Deprecated:'</td></tr>
<tr><td>\e</td><td>wrapped with '*'</td></tr>
<tr><td>\else</td><td>replaced with '}Else:{'</td></tr>
<tr><td>\elseif</td><td>replaced with '}Else if: &lt;condition&gt;{'</td></tr>
<tr><td>\em</td><td>wrapped with '*'</td></tr>
<tr><td>\endcond</td><td>replaced with 'End of conditional comment.'</td>
</tr>
<tr><td>\endif</td><td>replaced with '}'</td></tr>
<tr><td>\example</td><td>replaced with 'Example:'</td></tr>
<tr><td>\exception</td><td>replaced with ':raises:'</td></tr>
<tr><td>\f$</td><td>rendered using ':math:``'</td></tr>
<tr><td>\f[</td><td>rendered using '.. math::'</td></tr>
<tr><td>\f{</td><td>rendered using '.. math::'</td></tr>
<tr><td>\if</td><td>replaced with 'If: &lt;condition&gt; {'</td></tr>
<tr><td>\ifnot</td><td>replaced with 'If not: &lt;condition&gt; {'</td></tr>
<tr><td>\li</td><td>prepended with '* '</td></tr>
<tr><td>\n</td><td>replaced with newline char</td></tr>
<tr><td>\note</td><td>replaced with 'Note:'</td></tr>
<tr><td>\overload</td><td>prints 'This is an overloaded ...' according
to Doxygen docs</td></tr>
<tr><td>\p</td><td>wrapped with '``'</td></tr>
<tr><td>\par</td><td>replaced with 'Title: ...'</td></tr>
<tr><td>\param</td><td>add ':type:' and ':param:' directives</td></tr>
<tr><td>\param[&lt;dir&gt;]</td><td>same as \param, but direction ('in';
'out'; 'in,out') is included in ':type:' directive</td></tr>
<tr><td>\remark</td><td>replaced with 'Remarks:'</td></tr>
<tr><td>\remarks</td><td>replaced with 'Remarks:'</td></tr>
<tr><td>\result</td><td>add ':rtype:' and ':return:' directives</td></tr>
<tr><td>\return</td><td>add ':rtype:' and ':return:' directives</td></tr>
<tr><td>\returns</td><td>add ':rtype:' and ':return:' directives</td></tr>
<tr><td>\sa</td><td>replaced with 'See also:'</td></tr>
<tr><td>\see</td><td>replaced with 'See also:'</td></tr>
<tr><td>\since</td><td>replaced with 'Since:'</td></tr>
<tr><td>\throw</td><td>replaced with ':raises:'</td></tr>
<tr><td>\throws</td><td>replaced with ':raises:'</td></tr>
<tr><td>\todo</td><td>replaced with 'TODO:'</td></tr>
<tr><td>\tparam</td><td>add ':type:' and ':param:' directives</td></tr>
<tr><td>\verbatim</td><td>content copied verbatim</td></tr>
<tr><td>\version</td><td>replaced with 'Version:'</td></tr>
<tr><td>\warning</td><td>translated to 'Warning:'</td></tr>
<tr><td>\$</td><td>prints $ char</td></tr>
<tr><td>\@</td><td>prints @ char</td></tr>
<tr><td>\\</td><td>prints \ char</td></tr>
<tr><td>\&amp;</td><td>prints &amp; char</td></tr>
<tr><td>\~</td><td>prints ~ char</td></tr>
<tr><td>\&lt;</td><td>prints &lt; char</td></tr>
<tr><td>\&gt;</td><td>prints &gt; char</td></tr>
<tr><td>\#</td><td>prints # char</td></tr>
<tr><td>\%</td><td>prints % char</td></tr>
<tr><td>\&quot;</td><td>prints &quot; char</td></tr>
<tr><td>\.</td><td>prints . character</td></tr>
<tr><td>\::</td><td>prints ::</td></tr>
</table>
</div>
<h3><a name="Doxygen_pydoc_unsupported_tags">18.4.3 Unsupported tags for
Pydoc</a></h3>
<p> Doxygen has a wealth of tags such as <tt>@latexonly</tt> that have
no equivalent in Pydoc. As a result several tags that have no
translation (or particular use, such as some linking and section tags)
are suppressed with their content just printed out (if it has any
sense, typically text content). Here is the list of these tags:</p>
<div class="diagram"><b> Unsupported Python Doxygen tags</b>
<ul style="list-style-type:none;column-count:4;">
<li>\addindex</li>
<li>\addtogroup</li>
<li>\anchor</li>
<li>\attention</li>
<li>\brief</li>
<li>\bug</li>
<li>\callergraph</li>
<li>\callgraph</li>
<li>\category</li>
<li>\class</li>
<li>\copybrief</li>
<li>\copydetails</li>
<li>\copydoc</li>
<li>\date</li>
<li>\def</li>
<li>\defgroup</li>
<li>\details</li>
<li>\dir</li>
<li>\dontinclude</li>
<li>\dot</li>
<li>\dotfile</li>
<li>\enddot</li>
<li>\endhtmlonly</li>
<li>\endinternal</li>
<li>\endlatexonly</li>
<li>\endlink</li>
<li>\endmanonly</li>
<li>\endmsc</li>
<li>\endrtfonly</li>
<li>\endxmlonly</li>
<li>\enum</li>
<li>\extends</li>
<li>\file</li>
<li>\fn</li>
<li>\headerfile</li>
<li>\hideinitializer</li>
<li>\htmlinclude</li>
<li>\htmlonly</li>
<li>\image</li>
<li>\implements</li>
<li>\include</li>
<li>\includelineno</li>
<li>\ingroup</li>
<li>\interface</li>
<li>\internal</li>
<li>\invariant</li>
<li>\latexonly</li>
<li>\line</li>
<li>\link</li>
<li>\mainpage</li>
<li>\manonly</li>
<li>\memberof</li>
<li>\msc</li>
<li>\mscfile</li>
<li>\name</li>
<li>\namespace</li>
<li>\nosubgrouping</li>
<li>\package</li>
<li>\page</li>
<li>\paragraph</li>
<li>\post</li>
<li>\pre</li>
<li>\private</li>
<li>\privatesection</li>
<li>\property</li>
<li>\protected</li>
<li>\protectedsection</li>
<li>\protocol</li>
<li>\public</li>
<li>\publicsection</li>
<li>\ref</li>
<li>\related</li>
<li>\relatedalso</li>
<li>\relates</li>
<li>\relatesalso</li>
<li>\retval</li>
<li>\rtfonly</li>
<li>\section</li>
<li>\short</li>
<li>\showinitializer</li>
<li>\skip</li>
<li>\skipline</li>
<li>\snippet</li>
<li>\struct</li>
<li>\subpage</li>
<li>\subsection</li>
<li>\subsubsection</li>
<li>\tableofcontents</li>
<li>\test</li>
<li>\typedef</li>
<li>\union</li>
<li>\until</li>
<li>\var</li>
<li>\verbinclude</li>
<li>\weakgroup</li>
<li>\xmlonly</li>
<li>\xrefitem</li>
</ul>
</div>
<h2><a name="Doxygen_csharp">18.5 Doxygen to XML C# documentation</a></h2>
<p> If translation is enabled, XML formatted comments should be
automatically placed in the correct locations in the resulting module
and proxy files.</p>
<h3><a name="Doxygen_csharp_basic_example">18.5.1 Basic C# example</a></h3>
<p> Here is an example segment from an included header file</p>
<div class="code">
<pre>
/*! This is describing class Shape
\author Bob
*/
class Shape {
public:
Shape() {
nshapes++;
}
virtual ~Shape() {
nshapes--;
};
/** Important Variables x*/
double x;
/** Important Variables y*/
double y;
/** Moves the Shape
* @param dx delta on x
* @param dy delta on y */
void move(double dx, double dy);
/** \return the area */
virtual double area(void) = 0;
/** \return the perimeter */
virtual double perimeter(void) = 0;
static int nshapes;
};
</pre>
</div>
<p> Simply running SWIG should result in the following code being
present in Shapes.cs</p>
<div class="targetlang">
<pre>
/// &lt;summary&gt;This is describing class Shape
/// Author: Bob
/// &lt;/summary&gt;
public class Shape {
...
/// Important Variables x
public double x {
set {
ShapesCsPINVOKE.Shape_x_set(swigCPtr, value);
}
get {
double ret = ShapesCsPINVOKE.Shape_x_get(swigCPtr);
return ret;
}
}
/// Important Variables y
public double y {
set {
ShapesCsPINVOKE.Shape_y_set(swigCPtr, value);
}
get {
double ret = ShapesCsPINVOKE.Shape_y_get(swigCPtr);
return ret;
}
}
/// Moves the Shape
/// &lt;param name=&quot;dx&quot;&gt; delta on x&lt;/param&gt;
/// &lt;param name=&quot;dy&quot;&gt; delta on y&lt;/param&gt;
public void move(double dx, double dy) {
ShapesCsPINVOKE.Shape_move(swigCPtr, dx, dy);
}
/// &lt;returns&gt;the area&lt;/returns&gt;
public virtual double area() {
double ret = ShapesCsPINVOKE.Shape_area(swigCPtr);
return ret;
}
/// &lt;returns&gt;the perimeter&lt;/returns&gt;
public virtual double perimeter() {
double ret = ShapesCsPINVOKE.Shape_perimeter(swigCPtr);
return ret;
}
public static int nshapes {
set {
ShapesCsPINVOKE.Shape_nshapes_set(value);
}
get {
int ret = ShapesCsPINVOKE.Shape_nshapes_get();
return ret;
}
}
}
</pre>
</div>
<p> Otherwise, the code should be identical to what would have been
generated without the Doxygen functionality enabled. When the Doxygen
Translator module encounters a comment that contains nothing useful or
a Doxygen comment that it cannot parse, it will not affect the
functionality of the SWIG generated code.</p>
<p> The translator will handle most of the tags conversions (see the
table below). It will also automatically translate link-objects params,
in <tt>&lt;seealso&gt;</tt> tags . Also all <tt>\param</tt> and <tt>\tparam</tt>
commands are stripped out, if the specified parameter is not present in
the function. Use <tt>doxygen:nostripparams</tt> to avoid.</p>
<p> C# translator features summary (see <a href="#Customization_features">
%feature directives</a>):</p>
<h3><a name="Doxygen_csharp_tags">18.5.2 C# tags</a></h3>
<p> Here is the list of all Doxygen tags and the description of how they
are translated to C# XML</p>
<div class="diagram">
<table border="0" summary="CSharp XML Doxygen tags">
<tr><th align="left">Doxygen tags</th></tr>
<tr><td>\a</td><td>wrapped with '*' markdown character</td></tr>
<tr><td>\arg</td><td>prefixed with '*' markdown character</td></tr>
<tr><td>\author</td><td>Made simple text with prefix 'Author:'</td></tr>
<tr><td>\authors</td><td>Made simple text with prefix 'Author:'</td></tr>
<tr><td>\b</td><td>wrapped with '**' markdown character</td></tr>
<tr><td>\c</td><td>wrapped with `` markdown character</td></tr>
<tr><td>\cite</td><td>started with ' markdown character</td></tr>
<tr><td>\code</td><td>wrapped in &lt;code&gt; tag</td></tr>
<tr><td>\code{&lt;ext&gt;}</td><td>wrapped in &lt;code&gt; tag. code language
extension is ignored</td></tr>
<tr><td>\cond</td><td>Ignored</td></tr>
<tr><td>\copyright</td><td>wrapped in &lt;remark&gt; tag</td></tr>
<tr><td>\deprecated</td><td>Made simple text with prefix 'Deprecated:'</td>
</tr>
<tr><td>\e</td><td>prefixed with '*' markdown character</td></tr>
<tr><td>\else</td><td>Ignored</td></tr>
<tr><td>\elseif</td><td>Ignored</td></tr>
<tr><td>\em</td><td>prefixed with '*' markdown character</td></tr>
<tr><td>\endcode</td><td>wrapped in &lt;code&gt; tag. code language extension
is ignored</td></tr>
<tr><td>\endcond</td><td>Ignored</td></tr>
<tr><td>\endif</td><td>Ignored</td></tr>
<tr><td>\endlink</td><td>Ignored</td></tr>
<tr><td>\endverbatim</td><td>see note for \verbatim</td></tr>
<tr><td>\exception</td><td>translated to &lt;exception&gt; section</td></tr>
<tr><td>\f$, \f[, \f], \f{, \f}</td><td>Ignored</td></tr>
<tr><td>\if</td><td>Ignored</td></tr>
<tr><td>\ifnot</td><td>Ignored</td></tr>
<tr><td>\image</td><td>Ignored</td></tr>
<tr><td>\li</td><td>Ignored</td></tr>
<tr><td>\link</td><td>Ignored</td></tr>
<tr><td>\n</td><td>replaced with newline char</td></tr>
<tr><td>\note</td><td>Content copied</td></tr>
<tr><td>\overload</td><td>Ignored</td></tr>
<tr><td>\p</td><td>wrapped with `` markdown characters</td></tr>
<tr><td>\par</td><td>Made simple text with prefix &quot;Title:&quot;</td></tr>
<tr><td>\param</td><td>translated to &lt;param&gt; xml section</td></tr>
<tr><td>\param[&lt;dir&gt;]</td><td>translated to &lt;param&gt; xml section;
parameter direction ('in'; 'out'; or 'in,out') is ignored</td></tr>
<tr><td>\remark</td><td>Made simple text with prefix &quot;remarks:&quot;</td></tr>
<tr><td>\remarks</td><td>Made simple text with prefix &quot;remarks:&quot;</td></tr>
<tr><td>\result</td><td>translated to &lt;return&gt; xml section</td></tr>
<tr><td>\return</td><td>translated to &lt;return&gt; xml section</td></tr>
<tr><td>\returns</td><td>translated to &lt;return&gt; xml section</td></tr>
<tr><td>\sa</td><td>translated to &lt;seealso&gt; xml section</td></tr>
<tr><td>\see</td><td>translated to &lt;seealso&gt; xml section</td></tr>
<tr><td>\since</td><td>Content copied</td></tr>
<tr><td>\throw</td><td>translated to &lt;exception&gt;</td></tr>
<tr><td>\throws</td><td>translated to &lt;exception&gt;</td></tr>
<tr><td>\todo</td><td>Prefixed with 'TODO:'</td></tr>
<tr><td>\tparam</td><td>translated to &lt;exception&gt; xml section</td></tr>
<tr><td>\verbatim</td><td>copied without any processing</td></tr>
<tr><td>\version</td><td>Made simple text with prefix 'Version:&quot;</td></tr>
<tr><td>\warning</td><td>Made simple text with prefix 'remarks:&quot;</td></tr>
<tr><td>\$</td><td>prints $ char</td></tr>
<tr><td>\@</td><td>prints @ char</td></tr>
<tr><td>\\</td><td>prints \ char</td></tr>
<tr><td>\&amp;</td><td>prints &amp; char</td></tr>
<tr><td>\~</td><td>prints ~ char</td></tr>
<tr><td>\&lt;</td><td>prints &lt; char</td></tr>
<tr><td>\&gt;</td><td>prints &gt; char</td></tr>
<tr><td>\#</td><td>prints # char</td></tr>
<tr><td>\%</td><td>prints % char</td></tr>
<tr><td>\&quot;</td><td>prints &quot; char</td></tr>
<tr><td>\.</td><td>prints . char</td></tr>
<tr><td>\::</td><td>prints ::</td></tr>
</table>
</div>
<h3><a name="Doxygen_csharp_unsupported_tags">18.5.3 Unsupported tags
for C#</a></h3>
<p> Doxygen has a wealth of tags such as <tt>@latexonly</tt> that have
no equivalent in C# XML (all supported tags are listed in <a href="https://www.doxygen.nl/manual/xmlcmds.html">
XML documentation</a>). As a result several tags have no translation or
particular use, such as some linking and section tags. These are
suppressed with their content just printed out (if the tag has any
sense, typically text content). Here is the list of these tags:</p>
<div class="diagram"><b> Unsupported Doxygen tags</b>
<ul style="list-style-type:none;column-count:4;">
<li>\addindex</li>
<li>\addtogroup</li>
<li>\anchor</li>
<li>\attention</li>
<li>\brief</li>
<li>\bug</li>
<li>\callergraph</li>
<li>\callgraph</li>
<li>\category</li>
<li>\class</li>
<li>\copybrief</li>
<li>\copydetails</li>
<li>\copydoc</li>
<li>\date</li>
<li>\def</li>
<li>\defgroup</li>
<li>\details</li>
<li>\dir</li>
<li>\dontinclude</li>
<li>\dot</li>
<li>\dotfile</li>
<li>\enddot</li>
<li>\endhtmlonly</li>
<li>\endinternal</li>
<li>\endlatexonly</li>
<li>\endmanonly</li>
<li>\endmsc</li>
<li>\endrtfonly</li>
<li>\endxmlonly</li>
<li>\enum</li>
<li>\example</li>
<li>\extends</li>
<li>\file</li>
<li>\fn</li>
<li>\headerfile</li>
<li>\hideinitializer</li>
<li>\htmlinclude</li>
<li>\htmlonly</li>
<li>\implements</li>
<li>\include</li>
<li>\includelineno</li>
<li>\ingroup</li>
<li>\interface</li>
<li>\internal</li>
<li>\invariant</li>
<li>\latexonly</li>
<li>\line</li>
<li>\mainpage</li>
<li>\manonly</li>
<li>\memberof</li>
<li>\msc</li>
<li>\mscfile</li>
<li>\name</li>
<li>\namespace</li>
<li>\nosubgrouping</li>
<li>\package</li>
<li>\page</li>
<li>\paragraph</li>
<li>\post</li>
<li>\pre</li>
<li>\private</li>
<li>\privatesection</li>
<li>\property</li>
<li>\protected</li>
<li>\protectedsection</li>
<li>\protocol</li>
<li>\public</li>
<li>\publicsection</li>
<li>\ref</li>
<li>\related</li>
<li>\relatedalso</li>
<li>\relates</li>
<li>\relatesalso</li>
<li>\retval</li>
<li>\rtfonly</li>
<li>\section</li>
<li>\short</li>
<li>\showinitializer</li>
<li>\skip</li>
<li>\skipline</li>
<li>\snippet</li>
<li>\struct</li>
<li>\subpage</li>
<li>\subsection</li>
<li>\subsubsection</li>
<li>\tableofcontents</li>
<li>\test</li>
<li>\typedef</li>
<li>\union</li>
<li>\until</li>
<li>\var</li>
<li>\verbinclude</li>
<li>\weakgroup</li>
<li>\xmlonly</li>
<li>\xrefitem</li>
</ul>
</div>
<p> If one of the following Doxygen tags appears as the first tag in a
comment, the whole comment block is ignored:
<!-- see parser.y, function isStructuralDoxygen() -->
</p>
<div class="diagram"><b> Ignored Doxygen tags</b>
<ul style="list-style-type:none;column-count:4;">
<li>\addtogroup</li>
<li>\callergraph</li>
<li>\callgraph</li>
<li>\category</li>
<li>\class</li>
<li>\def</li>
<li>\defgroup</li>
<li>\dir</li>
<li>\enum</li>
<li>\example</li>
<li>\file</li>
<li>\fn</li>
<li>\headerfile</li>
<li>\hideinitializer</li>
<li>\interface</li>
<li>\internal</li>
<li>\mainpage</li>
<li>\name</li>
<li>\namespace</li>
<li>\nosubgrouping</li>
<li>\overload</li>
<li>\package</li>
<li>\page</li>
<li>\property</li>
<li>\protocol</li>
<li>\relates</li>
<li>\relatesalso</li>
<li>\showinitializer</li>
<li>\struct</li>
<li>\typedef</li>
<li>\union</li>
<li>\var</li>
<li>\weakgroup</li>
</ul>
</div>
<h2><a name="Doxygen_troubleshooting">18.6 Troubleshooting</a></h2>
<p> When running SWIG with command line option <tt>-doxygen</tt>, it may
happen that SWIG will fail to parse the code, which is valid C++ code
and is parsed without problems without the option. The problem is, that
Doxygen comments are not tokens (the C/C++ compiler actually never sees
them) and that they can appear anywhere in the code. That's why it is
practically impossible to handle all corner cases with the parser.
However, these problems can usually be avoided by minor changes in the
code or comment. Known problems and solutions are shown in this
section.</p>
<p> Recommended approach is to first run SWIG without command line
option <tt>-doxygen</tt>. When it successfully processes the code,
include the option and fix problems with Doxygen comments.</p>
<h3><a name="troubleshooting_ifndef">18.6.1 Problem with conditional
compilation</a></h3>
<p> Inserting a conditional compilation preprocessor directive between a
Doxygen comment and a commented item may break parsing:</p>
<div class="code">
<pre>
class A {
/**
* Some func.
*/
<font color="#ff0000">#ifndef SWIG</font>
void myfunc()
{
}
#endif
};
</pre>
</div>
<p> The solution is to move the directive above the comment:</p>
<div class="code">
<pre>
class A {
<font color="#00d000">#ifndef SWIG</font>
/**
* Some func.
*/
void myfunc()
{
}
#endif
};
</pre>
</div>
<h2><a name="Doxygen_developer_details">18.7 Developer information</a></h2>
<p> This section contains information for developers enhancing the
Doxygen translator.</p>
<h3><a name="Doxygen_translator_design">18.7.1 Doxygen translator design</a>
</h3>
<p> If this functionality is turned on, SWIG places all comments found
into the SWIG parse tree. Nodes contain an additional attribute called <tt>
doxygen</tt> when a comment is present. Individual nodes containing
Doxygen with Structural Indicators, such as <tt>@file</tt>, as their
first command, are also present in the parse tree. These individual
&quot;blobs&quot; of Doxygen such as :</p>
<div class="code">
<pre>
/*! This is describing function Foo
\param x some random variable
\author Bob
\return Foo
*/
</pre>
</div>
<p> are passed on individually to the Doxygen Translator module. This
module builds its own private parse tree and hands it to a separate
class for translation into the target documentation language. For
example, <tt>JavaDocConverter</tt> is the Javadoc module class.</p>
<h3><a name="Doxygen_debugging_commands">18.7.2 Debugging the Doxygen
parser and translator</a></h3>
<p> There are two handy command line options, that enable lots of
detailed debug information printing.</p>
<div class="shell">
<pre>
-debug-doxygen-parser - Display Doxygen parser module debugging information
-debug-doxygen-translator - Display Doxygen translator module debugging information
</pre>
</div>
<h3><a name="Doxygen_tests">18.7.3 Tests</a></h3>
<p> Doxygen tests have been added to the regular SWIG test-suite. There
are a number of tests beginning <tt>doxygen_</tt> in the
Examples/test-suite sub-directory.</p>
<p> Like any other SWIG test case, the tests are included in
Examples/test-suite/common.mk and can be tested with commands like <tt>
make check-test-suite</tt> or <tt>make check-python-test-suite</tt>. To
run them individually, type <tt>make -s &lt;testname&gt;.cpptest</tt> in the
language-specific sub-directory in <tt>Examples/test-suite</tt>
directory. For example:</p>
<div class="shell">
<pre>
Examples/test-suite/java $ make -s doxygen_parsing.cpptest
</pre>
</div>
<p> If the test fails, both expected and translated comments are printed
to std out, but also written to files<i> expected.txt</i> and<i>
got.txt</i>. Since it is often difficult to find a single character
difference in several lines of text, we can use some diff tool, for
example:</p>
<div class="shell">
<pre>
Examples/test-suite/java $ kdiff3 expected.txt got.txt
</pre>
</div>
<p> Runtime tests in Java are implemented using Javadoc doclets. To make
that work, you should have tools.jar from the JDK in your classpath. Or
you should have JAVA_HOME environment variable defined and pointing to
the JDK location.</p>
<p> The Java's comment parsing code (the testing part) is located in
commentParser.java. It checks the generated code. It is possible to run
this file as a stand-alone program, with <tt>java commentParser &lt;some
java package&gt;</tt>, and it will print the list of comments found in the
specified directory (in the format it has used in the runtime tests).
So, when you want to create a new Doxygen test case, just copy an
existing one and replace the actual comment content (section of entries
in form 'wantedComments.put(...)' with the output of the above command.</p>
<p> Runtime tests in Python are just plain string comparisons of the
__doc__ properties.</p>
<h2><a name="Doxygen_language_extension">18.8 Extending to other
languages</a></h2>
<p> In general, an extension to another language requires a fairly deep
understanding of the target language module, such as Modules/python.cxx
for Python. Searching for &quot;doxygen&quot; in the java.cxx module can give you
a good idea of the process for placing documentation comments into the
correct areas. The basic gist is that anywhere a comment may reside on
a node, there needs to be a catch for it in front of where that
function, class, or other object is written out to a target language
file. The other half of extension is building a target documentation
language comment generator that handles one blob at a time. However,
this is relatively simple and nowhere near as complex as the wrapper
generating modules in SWIG. See Source/Doxygen/javadoc.cxx for a good
example. The target language module passes the Doxygen Translator the
blob to translate, and receives back the translated text.</p>
<p><b> What is given to the Doxygen Translator</b></p>
<div class="code">
<pre>
/*! This is describing function Foo
\param x some random variable
\author Bob
\return Foo
*/
</pre>
</div>
<p><b> What is received back by java.cxx</b></p>
<div class="targetlang">
<pre>
/** This is describing function Foo
*
* @param x some random variable
* @author Bob
* @return Foo
*/
</pre>
</div>
<p> Development of the comment translator itself is simplified by the
fact that the Doxygen Translator module can easily include a <tt>main</tt>
function and thus be developed, compiled, and tested independently of
SWIG.</p>
<HR NOSHADE>
<h1><a name="Warnings">19 Warning Messages</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Warnings_nn2">Introduction</a></li>
<li><a href="#Warnings_suppression">Warning message suppression</a></li>
<li><a href="#Warnings_nn4">Enabling extra warnings</a></li>
<li><a href="#Warnings_nn5">Issuing a warning message</a></li>
<li><a href="#Warnings_symbolic_symbols">Symbolic symbols</a></li>
<li><a href="#Warnings_nn6">Commentary</a></li>
<li><a href="#Warnings_nn7">Warnings as errors</a></li>
<li><a href="#Warnings_nn8">Message output format</a></li>
<li><a href="#Warnings_nn9">Warning number reference</a>
<ul>
<li><a href="#Warnings_nn10">Deprecated features (100-199)</a></li>
<li><a href="#Warnings_nn11">Preprocessor (200-299)</a></li>
<li><a href="#Warnings_nn12">C/C++ Parser (300-399)</a></li>
<li><a href="#Warnings_nn13">Types and typemaps (400-499)</a></li>
<li><a href="#Warnings_nn14">Code generation (500-559)</a></li>
<li><a href="#Warnings_doxygen">Doxygen comments (560-599)</a></li>
<li><a href="#Warnings_nn15">Language module specific (700-899)</a></li>
<li><a href="#Warnings_nn16">User defined (900-999)</a></li>
</ul>
</li>
<li><a href="#Warnings_nn17">History</a></li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="Warnings_nn2">19.1 Introduction</a></h2>
<p> During compilation, SWIG may generate a variety of warning messages.
For example:</p>
<div class="shell">
<pre>
example.i:16: Warning 501: Overloaded declaration ignored. bar(double)
example.i:15: Warning 501: Previous declaration is bar(int)
</pre>
</div>
<p> Typically, warning messages indicate non-fatal problems with the
input where the generated wrapper code will probably compile, but it
may not work like you expect.</p>
<h2><a name="Warnings_suppression">19.2 Warning message suppression</a></h2>
<p> All warning messages have a numeric code that is shown in the
warning message itself. To suppress the printing of a warning message,
a number of techniques can be used. First, you can run SWIG with the <tt>
-w</tt> command line option. For example:</p>
<div class="shell">
<pre>
% swig -python -w501 example.i
% swig -python -w501,505,401 example.i
</pre>
</div>
<p> Alternatively, warnings can be suppressed by inserting a special
preprocessor pragma into the input file:</p>
<div class="code">
<pre>
%module example
#pragma SWIG nowarn=501
#pragma SWIG nowarn=501,505,401
</pre>
</div>
<p> Finally, code-generation warnings can be disabled on a declaration
by declaration basis using the <tt>%warnfilter</tt> directive. For
example:</p>
<div class="code">
<pre>
%module example
%warnfilter(501) foo;
...
int foo(int);
int foo(double); // Silently ignored.
</pre>
</div>
<p> The <tt>%warnfilter</tt> directive has the same semantics as other
declaration modifiers like <tt>%rename</tt>, <tt>%ignore</tt> and <tt>
%feature</tt>, see the <a href="#Customization_features">%feature
directive</a> section. For example, if you wanted to suppress a warning
for a method in a class hierarchy, you could do this:</p>
<div class="code">
<pre>
%warnfilter(501) Object::foo;
class Object {
public:
int foo(int);
int foo(double); // Silently ignored
...
};
class Derived : public Object {
public:
int foo(int);
int foo(double); // Silently ignored
...
};
</pre>
</div>
<p> Warnings can be suppressed for an entire class by supplying a class
name. For example:</p>
<div class="code">
<pre>
%warnfilter(501) Object;
class Object {
public:
... // All 501 warnings ignored in class
};
</pre>
</div>
<p> There is no option to suppress all SWIG warning messages. The
warning messages are there for a reason---to tell you that something
may be<em> broken</em> in your interface. Ignore the warning messages
at your own peril.</p>
<h2><a name="Warnings_nn4">19.3 Enabling extra warnings</a></h2>
<p> Some warning messages are disabled by default and are generated only
to provide additional diagnostics. These warnings can be turned on
using the <tt>-Wextra</tt> option. For example:</p>
<div class="shell">
<pre>
% swig -Wextra -python example.i
</pre>
</div>
<p> Preprocessor warning 202 (&quot;Could not evaluate expression<em> expr</em>
.&quot;) was formally off by default and enabled by <tt>-Wextra</tt>, but
since SWIG 4.1.0 this warning is on by default because suppressing it
tends to hide genuine problems. If you really don't want to see it, you
can suppress it with <tt>-w202</tt> or using <tt>%warnfilter</tt> as
described below. Both will work with older versions of SWIG too.</p>
<p> To selectively turn on extra warning messages, you can use the
directives and options in the previous section--simply add a &quot;+&quot; to all
warning numbers. For example:</p>
<div class="shell">
<pre>
% swig -w+309,+452 example.i
</pre>
</div>
<p> or in your interface file use either</p>
<div class="code">
<pre>
#pragma SWIG nowarn=+309,+452
</pre>
</div>
<p> or</p>
<div class="code">
<pre>
%warnfilter(+309,+452) foo;
</pre>
</div>
<p> Note: selective enabling of warnings with <tt>%warnfilter</tt>
overrides any global settings you might have made using <tt>-w</tt> or <tt>
#pragma</tt>.</p>
<p> You can of course also enable all warnings and suppress a select
few, for example:</p>
<div class="shell">
<pre>
% swig -Wextra -w309,452 example.i
</pre>
</div>
<p> The warnings on the right take precedence over the warnings on the
left, so in the above example <tt>-Wextra</tt> adds numerous warnings
including 452, but then <tt>-w309,452</tt> overrides this and so 452 is
suppressed.</p>
<p> If you would like all warnings to appear, regardless of the warning
filters used, then use the <tt>-Wall</tt> option. The <tt>-Wall</tt>
option also turns on the extra warnings that <tt>-Wextra</tt> adds,
however, it is subtely different. When <tt>-Wall</tt> is used, it also
disables all other warning filters, that is, any warnings suppressed or
added in <tt>%warnfilter</tt>, <tt>#pragma SWIG nowarn</tt> or the <tt>
-w</tt> option.</p>
<h2><a name="Warnings_nn5">19.4 Issuing a warning message</a></h2>
<p> Warning messages can be issued from an interface file using a number
of directives. The <tt>%warn</tt> directive is the most simple:</p>
<div class="code">
<pre>
%warn &quot;900:This is your last warning!&quot;
</pre>
</div>
<p> All warning messages are optionally prefixed by the warning number
to use. If you are generating your own warnings, make sure you don't
use numbers defined in the table at the end of this section.</p>
<p> The <tt>%ignorewarn</tt> directive is the same as <tt>%ignore</tt>
except that it issues a warning message whenever a matching declaration
is found. For example:</p>
<div class="code">
<pre>
%ignorewarn(&quot;362:operator= ignored&quot;) operator=;
</pre>
</div>
<p> Warning messages can be associated with typemaps using the <tt>
warning</tt> attribute of a typemap declaration. For example:</p>
<div class="code">
<pre>
%typemap(in, warning=&quot;901:You are really going to regret this usage of $1_type $1_name&quot;) blah * {
...
}
</pre>
</div>
<p> In this case, the warning message will be printed whenever the
typemap is actually used and the <a href="#Typemaps_special_variables">
special variables</a> will be expanded as appropriate, for example:</p>
<div class="shell">
<pre>
example.i:23: Warning 901: You are really going to regret this usage of blah * self
example.i:24: Warning 901: You are really going to regret this usage of blah * stuff
</pre>
</div>
<h2><a name="Warnings_symbolic_symbols">19.5 Symbolic symbols</a></h2>
<p> The <tt>swigwarn.swg</tt> file that is installed with SWIG contains
symbol constants that could also be used in <tt>%warnfilter</tt> and <tt>
#pragma SWIG nowarn</tt>. For example this file contains the following
line:</p>
<div class="code">
<pre>
%define SWIGWARN_TYPE_UNDEFINED_CLASS 401 %enddef
</pre>
</div>
<p> so <tt>SWIGWARN_TYPE_UNDEFINED_CLASS</tt> could be used instead of
401, for example:</p>
<div class="code">
<pre>
#pragma SWIG nowarn=SWIGWARN_TYPE_UNDEFINED_CLASS
</pre>
</div>
<p> or</p>
<div class="code">
<pre>
%warnfilter(SWIGWARN_TYPE_UNDEFINED_CLASS) Foo;
</pre>
</div>
<h2><a name="Warnings_nn6">19.6 Commentary</a></h2>
<p> The ability to suppress warning messages is really only provided for
advanced users and is not recommended in normal use. You are advised to
modify your interface to fix the problems highlighted by the warnings
wherever possible instead of suppressing warnings.</p>
<p> Certain types of SWIG problems are errors. These usually arise due
to parsing errors (bad syntax) or semantic problems for which there is
no obvious recovery. There is no mechanism for suppressing error
messages.</p>
<h2><a name="Warnings_nn7">19.7 Warnings as errors</a></h2>
<p> Warnings can be handled as errors by using the <tt>-Werror</tt>
command line option. This will cause SWIG to exit with a non successful
exit code if a warning is encountered.</p>
<h2><a name="Warnings_nn8">19.8 Message output format</a></h2>
<p> The output format for both warnings and errors can be selected for
integration with your favourite IDE/editor. Editors and IDEs can
usually parse error messages and if in the appropriate format will
easily take you directly to the source of the error. The standard
format is used by default except on Windows where the Microsoft format
is used by default. These can be overridden using command line options,
for example:</p>
<div class="shell">
<pre>
$ swig -python -Fstandard example.i
example.i:4: Syntax error in input(1).
$ swig -python -Fmicrosoft example.i
example.i(4) : Syntax error in input(1).
</pre>
</div>
<h2><a name="Warnings_nn9">19.9 Warning number reference</a></h2>
<h3><a name="Warnings_nn10">19.9.1 Deprecated features (100-199)</a></h3>
<p>None currently.</p>
<h3><a name="Warnings_nn11">19.9.2 Preprocessor (200-299)</a></h3>
<ul>
<li>201. Unable to find<em> filename</em>.</li>
<li>202. Could not evaluate expression<em> expr</em>.</li>
<li>203. Both includeall and importall are defined: using includeall.</li>
<li>204. CPP #warning, &quot;<em>warning</em>&quot;.</li>
<li>205. CPP #error, &quot;<em>error</em>&quot;.</li>
<li>206. Unexpected tokens after #<em>directive</em> directive.</li>
</ul>
<h3><a name="Warnings_nn12">19.9.3 C/C++ Parser (300-399)</a></h3>
<ul>
<li>301. <tt>class</tt> keyword used, but not in C++ mode.</li>
<li>302. Redefinition of identifier '<em>name</em>' as<em> decl</em>
ignored.</li>
<li>303. <tt>%extend</tt> defined for an undeclared class '<em>name</em>
'.</li>
<li>304. Unsupported constant value (ignored).</li>
<li>305. Bad constant value (ignored).</li>
<li>308. Namespace alias '<em>name</em>' not allowed here. Assuming '<em>
name</em>'</li>
<li>309. [private | protected] inheritance ignored.</li>
<li>312. Unnamed nested class not currently supported (ignored).</li>
<li>313. Unrecognized extern type &quot;<em>name</em>&quot; (ignored).</li>
<li>314. '<em>identifier</em>' is a<em> lang</em> keyword.</li>
<li>315. Nothing known about '<em>identifier</em>'.</li>
<li>317. Specialization of non-template '<em>name</em>'.</li>
<li>318. Instantiation of template '<em>name</em>' is ambiguous,
instantiation<em> templ</em> used, instantiation<em> templ</em>
ignored.</li>
<li>319. No access specifier given for base class<em> name</em>
(ignored).</li>
<li>320. Explicit template instantiation ignored.</li>
<li>321.<em> identifier</em> conflicts with a built-in name.</li>
<li>322. Redundant redeclaration of identifier '<em>name</em>' as<em>
decl</em> ignored.</li>
<li>323. Recursive scope inheritance of '<em>name</em>'.</li>
<li>324. Named nested template instantiations not supported. Processing
as if no name was given to %template().</li>
<li>325. Nested<em> kind</em> not currently supported (<em>name</em>
ignored).</li>
<li>326. Deprecated %extend name used - the<em> kind</em> name '<em>name</em>
' should be used instead of the typedef name '<em>name</em>'.</li>
<li>327. Extern template ignored.</li>
<li>328. Value assigned to<em> name</em> not used due to limited parsing
implementation.</li>
<li>329. Using declaration '<em>name</em>' for inheriting constructors
uses base '<em>name</em>' which is not an immediate base of '<em>name</em>
'.</li>
<li>330. Template forward class '<em>name</em>' cannot be used to
instantiate a full template class with name '<em>name</em>'.</li>
<li>331. Unsupported template nested class '<em>name</em>' cannot be
used to instantiate a full template class with name '<em>name</em>'.</li>
<li>340. Lambda expressions and closures are not fully supported yet.</li>
<li>344. Unable to deduce decltype for '<em>expr</em>'.</li>
<li>345. Unable to deduce auto return type for '<em>name</em>'
(ignored).</li>
<li>346. Unable to deduce auto type for variable '<em>name</em>'
(ignored).</li>
<li>350. operator new ignored.</li>
<li>351. operator delete ignored.</li>
<li>352. operator+ ignored.</li>
<li>353. operator- ignored.</li>
<li>354. operator* ignored.</li>
<li>355. operator/ ignored.</li>
<li>356. operator% ignored.</li>
<li>357. operator^ ignored.</li>
<li>358. operator&amp; ignored.</li>
<li>359. operator| ignored.</li>
<li>360. operator~ ignored.</li>
<li>361. operator! ignored.</li>
<li>362. operator= ignored.</li>
<li>363. operator&lt; ignored.</li>
<li>364. operator&gt; ignored.</li>
<li>365. operator+= ignored.</li>
<li>366. operator-= ignored.</li>
<li>367. operator*= ignored.</li>
<li>368. operator/= ignored.</li>
<li>369. operator%= ignored.</li>
<li>370. operator^= ignored.</li>
<li>371. operator&amp;= ignored.</li>
<li>372. operator|= ignored.</li>
<li>373. operator&lt;&lt; ignored.</li>
<li>374. operator&gt;&gt;ignored.</li>
<li>375. operator&lt;&lt;= ignored.</li>
<li>376. operator&gt;&gt;= ignored.</li>
<li>377. operator== ignored.</li>
<li>378. operator!= ignored.</li>
<li>379. operator&lt;= ignored.</li>
<li>380. operator&gt;= ignored.</li>
<li>381. operator&amp;&amp; ignored.</li>
<li>382. operator|| ignored.</li>
<li>383. operator++ ignored.</li>
<li>384. operator-- ignored.</li>
<li>385. operator, ignored.</li>
<li>386. operator-&lt;* ignored.</li>
<li>387. operator-&lt; ignored.</li>
<li>388. operator() ignored.</li>
<li>389. operator[] ignored.</li>
<li>390. operator+ ignored (unary).</li>
<li>391. operator- ignored (unary).</li>
<li>392. operator* ignored (unary).</li>
<li>393. operator&amp; ignored (unary).</li>
<li>394. operator new[] ignored.</li>
<li>395. operator delete[] ignored.</li>
<li>396. operator*() ignored.</li>
<li>397. operator&lt;=&gt; delete[] ignored.</li>
</ul>
<h3><a name="Warnings_nn13">19.9.4 Types and typemaps (400-499)</a></h3>
<ul>
<li>401. Nothing known about class 'name'. Ignored.</li>
<li>402. Base class 'name' is incomplete.</li>
<li>403. Class 'name' might be abstract.</li>
<li>404. Duplicate template instantiation of '<em>type</em>' with name '<em>
name</em>' ignored, previous instantiation of '<em>type</em>' with name
'<em>name</em>'.</li>
<li>405. Method with rvalue ref-qualifier<em> name</em> ignored.</li>
<li>406. Ignoring nspace setting (<em>setting</em>) for '<em>type</em>',
as it conflicts with the nspace setting (<em>setting</em>) for outer
class '<em>type</em>'</li>
<li>450. Reserved</li>
<li>451. Setting const char * variable may leak memory.</li>
<li>452. Reserved</li>
<li>453. Can't apply (pattern). No typemaps are defined.</li>
<li>454. Setting a pointer/reference variable may leak memory</li>
<li>455. Setting a const wchar_t * variable may leak memory.</li>
<li>460. Unable to use type<em> type</em> as a function argument.</li>
<li>461. Unable to use return type<em> type</em> in function<em> name</em>
.</li>
<li>462. Unable to set variable of type<em> type</em>.</li>
<li>463. Unable to read variable of type<em> type</em>.</li>
<li>464. Unsupported constant value.</li>
<li>465. Unable to handle type<em> type</em>.</li>
<li>466. Unsupported variable type<em> type</em>.</li>
<li>467. Overloaded<em> declaration</em> not supported (incomplete type
checking rule - no precedence level in typecheck typemap for '<em>type</em>
')</li>
<li>468. No 'throw' typemap defined for exception type<em> type</em></li>
<li>469. No or improper directorin typemap defined for<em> type</em></li>
<li>470. Thread/reentrant unsafe wrapping, consider returning by value
instead.</li>
<li>471. Unable to use return type<em> type</em> in director method</li>
<li>472. Overloaded method<em> method</em> with no explicit typecheck
typemap for arg<em> number</em> of type '<em>type</em>'. Dispatching
calls to this method may not work correctly, see the 'Typemaps and
Overloading' section in the Typemaps chapter of the SWIG documentation.</li>
<li>473. Returning a reference, pointer or pointer wrapper in a director
method is not recommended.</li>
<li>474. Method<em> method</em> usage of the optimal attribute ignored
in the out typemap as the following cannot be used to generate optimal
code:<em> code</em></li>
<li>475. Multiple calls to<em> method</em> might be generated due to
optimal attribute usage in the out typemap.</li>
<li>476. Initialization using std::initializer_list.</li>
<li>477. No directorthrows typemap defined for<em> type</em></li>
<li>490. Fragment '<em>name</em>' not found.</li>
</ul>
<h3><a name="Warnings_nn14">19.9.5 Code generation (500-559)</a></h3>
<ul>
<li>501. Overloaded declaration ignored.<em> decl</em>. Previous
declaration is<em> decl</em>.</li>
<li>502. Overloaded constructor ignored.<em> decl</em>. Previous
declaration is<em> decl</em>.</li>
<li>503. Can't wrap '<em>identifier</em>' unless renamed to a valid
identifier.</li>
<li>504. Function<em> name</em> must have a return type. Ignored.</li>
<li>505. Variable length arguments discarded.</li>
<li>506. Can't wrap varargs with keyword arguments enabled.</li>
<li>507. Adding native function<em> name</em> not supported (ignored).</li>
<li>508. Declaration of '<em>name</em>' shadows declaration accessible
via operator-&gt;(), previous declaration of'<em>declaration</em>'.</li>
<li>509. Overloaded method<em> declaration</em> effectively ignored, as
it is shadowed by<em> declaration</em>.</li>
<li>510. Friend function '<em>name</em>' ignored.</li>
<li>511. Can't use keyword arguments with overloaded functions.</li>
<li>512. Overloaded method<em> declaration</em> ignored, using non-const
method<em> declaration</em> instead.</li>
<li>513. Can't generate wrappers for unnamed struct/class.</li>
<li>514.</li>
<li>515.</li>
<li>516. Overloaded method<em> declaration</em> ignored, using<em>
declaration</em> instead.</li>
<li>517.</li>
<li>518. Portability warning: File<em> file1</em> will be overwritten by<em>
file2</em> on case insensitive filesystems such as Windows' FAT32 and
NTFS unless the class/module name is renamed.</li>
<li>519. %template() contains no name. Template method ignored:<em>
declaration</em></li>
<li>520.<em> Base/Derived</em> class '<em>classname1</em>' of '<em>
classname2</em>' is not similarly marked as a smart pointer.</li>
<li>521. Illegal destructor name<em> name</em>. Ignored.</li>
<li>522. Use of an illegal constructor name '<em>name</em>' in %extend
is deprecated, the constructor name should be '<em>name</em>'.</li>
<li>523. Use of an illegal destructor name '<em>name</em>' in %extend is
deprecated, the destructor name should be '<em>name</em>'.</li>
<li>524. Experimental target language. Target language<em> language</em>
specified by<em> lang</em> is an experimental language. See the 'Target
Languages' section in the Introduction chapter of the SWIG
documentation.</li>
<li>525. Destructor<em> declaration</em> is final,<em> name</em> cannot
be a director class.</li>
<li>526. Using declaration<em> declaration</em>, with name '<em>name</em>
', is not actually using the method from<em> declaration</em>, with name
'<em>name</em>', as the names are different.</li>
<li>527. Deprecated target language. Target language<em> language</em>
specified by<em> lang</em> is a deprecated language. It will be removed
in the next release of SWIG unless a new maintainer steps forward to
bring it up to at least experimental status. See the 'Target Languages'
section in the Introduction chapter of the SWIG documentation.</li>
</ul>
<h3><a name="Warnings_doxygen">19.9.6 Doxygen comments (560-599)</a></h3>
<ul>
<li>560: Unknown Doxygen command:<em> command</em>.</li>
<li>561: Unexpected end of Doxygen comment encountered.</li>
<li>562: Expected Doxygen command:<em> command</em></li>
<li>563: Doxygen HTML error for tag<em> tag</em>:<em> error text</em>.</li>
<li>564: Error parsing Doxygen command<em> command</em>:<em> error text</em>
. Command ignored.&quot;</li>
</ul>
<h3><a name="Warnings_nn15">19.9.7 Language module specific (700-899)</a>
</h3>
<ul>
<li>801. Wrong name (corrected to '<em>name</em>'). (Ruby).</li>
</ul>
<ul>
<li>810. No jni typemap defined for<em> type</em> (Java).</li>
<li>811. No jtype typemap defined for<em> type</em> (Java).</li>
<li>812. No jstype typemap defined for<em> type</em> (Java).</li>
<li>813. Warning for<em> classname</em>, base<em> baseclass</em>
ignored. Multiple inheritance is not supported in Java. (Java).</li>
<li>814.</li>
<li>815. No javafinalize typemap defined for<em> type</em> (Java).</li>
<li>816. No javabody typemap defined for<em> type</em> (Java).</li>
<li>817. No javaout typemap defined for<em> type</em> (Java).</li>
<li>818. No javain typemap defined for<em> type</em> (Java).</li>
<li>819. No javadirectorin typemap defined for<em> type</em> (Java).</li>
<li>820. No javadirectorout typemap defined for<em> type</em> (Java).</li>
<li>821.</li>
<li>822. Covariant return types not supported in Java. Proxy method will
return<em> basetype</em> (Java).</li>
<li>823. No javaconstruct typemap defined for<em> type</em> (Java).</li>
<li>824. Missing JNI descriptor in directorin typemap defined for<em>
type</em> (Java).</li>
<li>825. &quot;directorconnect&quot; attribute missing in<em> type</em>
&quot;javaconstruct&quot; typemap. (Java).</li>
<li>826. The nspace feature is used on '<em>type</em>' without -package.
The generated code may not compile as Java does not support types
declared in a named package accessing types declared in an unnamed
package. (Java).</li>
</ul>
<ul>
<li>830. No ctype typemap defined for<em> type</em> (C#).</li>
<li>831. No cstype typemap defined for<em> type</em> (C#).</li>
<li>832. No cswtype typemap defined for<em> type</em> (C#).</li>
<li>833. Warning for<em> classname</em>, base<em> baseclass</em>
ignored. Multiple inheritance is not supported in C#. (C#).</li>
<li>834.</li>
<li>835. No csfinalize typemap defined for<em> type</em> (C#).</li>
<li>836. No csbody typemap defined for<em> type</em> (C#).</li>
<li>837. No csout typemap defined for<em> type</em> (C#).</li>
<li>838. No csin typemap defined for<em> type</em> (C#).</li>
<li>839.</li>
<li>840.</li>
<li>841.</li>
<li>842. Covariant return types not supported in C#. Proxy method will
return<em> basetype</em> (C#).</li>
<li>843. No csconstruct typemap defined for<em> type</em> (C#).</li>
<li>844. C# exception may not be thrown - no $excode or excode attribute
in<em> typemap</em> typemap. (C#).</li>
<li>845. Unmanaged code contains a call to a
SWIG_CSharpSetPendingException method and C# code does not handle
pending exceptions via the canthrow attribute. (C#).</li>
</ul>
<ul>
<li>870. Warning for<em> classname</em>: Base<em> baseclass</em>
ignored. Multiple inheritance is not supported in PHP. (Php).</li>
<li>871. Unrecognized pragma<em> pragma</em>. (Php).</li>
</ul>
<h3><a name="Warnings_nn16">19.9.8 User defined (900-999)</a></h3>
<p> These numbers can be used by your own application.</p>
<h2><a name="Warnings_nn17">19.10 History</a></h2>
<p> The ability to control warning messages was first added to
SWIG-1.3.12.</p>
<HR NOSHADE>
<h1><a name="Modules">20 Working with Modules</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Modules_introduction">Modules Introduction</a></li>
<li><a href="#Modules_nn1">Basics</a></li>
<li><a href="#Modules_nn2">The SWIG runtime code</a></li>
<li><a href="#Modules_external_run_time">External access to the runtime</a>
</li>
<li><a href="#Modules_nn4">A word of caution about static libraries</a></li>
<li><a href="#Modules_nn5">References</a></li>
<li><a href="#Modules_nn6">Reducing the wrapper file size</a></li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="Modules_introduction">20.1 Modules Introduction</a></h2>
<p> Each invocation of SWIG requires a module name to be specified. The
module name is used to name the resulting target language extension
module. Exactly what this means and what the name is used for depends
on the target language, for example the name can define a target
language namespace or merely be a useful name for naming files or
helper classes. Essentially, a module comprises target language
wrappers for a chosen collection of global variables/functions,
structs/classes and other C/C++ types.</p>
<p> The module name can be supplied in one of two ways. The first is to
specify it with the special <tt>%module</tt> directive. This directive
must appear at the beginning of the interface file. The general form of
this directive is:</p>
<div class="code">
<pre>
<tt>%module(option1=&quot;value1&quot;, option2=&quot;value2&quot;, ...) modulename</tt>
</pre>
</div>
<p> where the modulename is mandatory and the options add one or more
optional additional features. Typically no options are specified, for
example:</p>
<div class="code">
<pre>
<tt>%module mymodule</tt>
</pre>
</div>
<p> The second way to specify the module name is with the <tt>-module</tt>
command line option, for example <tt>-module mymodule</tt>. If the
module name is supplied on the command line, it overrides the name
specified by the <tt>%module</tt> directive.</p>
<p> When first working with SWIG, users commonly start by creating a
single module. That is, you might define a single SWIG interface that
wraps some set of C/C++ code. You then compile all of the generated
wrapper code together and use it. For large applications, however, this
approach is problematic---the size of the generated wrapper code can be
rather large. Moreover, it is probably easier to manage the target
language interface when it is broken up into smaller pieces.</p>
<p> This chapter describes the problem of using SWIG in programs where
you want to create a collection of modules. Each module in the
collection is created via separate invocations of SWIG.</p>
<h2><a name="Modules_nn1">20.2 Basics</a></h2>
<p> The basic usage case with multiple modules is when modules do not
have cross-references (ie. when wrapping multiple independent C APIs).
In that case, swig input files should just work out of the box - you
simply create multiple wrapper .cxx files, link them into your
application, and insert/load each in the scripting language runtime as
you would do for the single module case.</p>
<p> A bit more complex is the case in which modules need to share
information. For example, when one module extends the class of another
by deriving from it:</p>
<div class="code">
<pre>
// File: base.h
class base {
public:
int foo();
};
</pre>
</div> &nbsp;<div class="code">
<pre>
// File: base_module.i
%module base_module
%{
#include &quot;base.h&quot;
%}
%include &quot;base.h&quot;
</pre>
</div> &nbsp;<div class="code">
<pre>
// File: derived_module.i
%module derived_module
%{
#include &quot;base.h&quot;
%}
%import &quot;base_module.i&quot;
%inline %{
class derived : public base {
public:
int bar();
};
%}
</pre>
</div>
<p>To create the wrapper properly, module <tt>derived_module</tt> needs
to know about the <tt>base</tt> class and that its interface is covered
in another module. The line <tt>%import &quot;base_module.i&quot;</tt> lets SWIG
know exactly that. Often the <tt>.h</tt> file is passed to <tt>%import</tt>
instead of the <tt>.i</tt>, which unfortunately doesn't work for all
language modules. For example, Python requires the name of module that
the base class exists in so that the proxy classes can fully inherit
the base class's methods. Typically you will get a warning when the
module name is missing, eg:</p>
<div class="shell">
<pre>
derived_module.i:8: Warning 401: Base class 'base' ignored - unknown module name for base. Either
import
the appropriate module interface file or specify the name of the module in the %import directive.
</pre>
</div>
<p> It is sometimes desirable to import the header file rather than the
interface file and overcome the above warning. For example in the case
of the imported interface being quite large, it may be desirable to
simplify matters and just import a small header file of dependent
types. This can be done by specifying the optional <tt>module</tt>
attribute in the <tt>%import</tt> directive. The <tt>derived_module.i</tt>
file shown above could be replaced with the following:<div class="code">
<pre>
// File: derived_module.i
%module derived_module
%{
#include &quot;base.h&quot;
%}
%import(module=&quot;base_module&quot;) &quot;base.h&quot;
%inline %{
class derived : public base {
public:
int bar();
};
</pre>
</div></p>
<p> Note that &quot;base_module&quot; is the module name and is the same as that
specified in <tt>%module</tt> in <tt>base_module.i</tt> as well as the <tt>
%import</tt> in <tt>derived_module.i</tt>.</p>
<p> Another issue to beware of is that multiple dependent wrappers
should not be linked/loaded in parallel from multiple threads as SWIG
provides no locking - for more on that issue, read on.</p>
<h2><a name="Modules_nn2">20.3 The SWIG runtime code</a></h2>
<p> Many of SWIG's target languages generate a set of functions commonly
known as the &quot;SWIG runtime.&quot; These functions are primarily related to
the runtime type system which checks pointer types and performs other
tasks such as proper casting of pointer values in C++. As a general
rule, the statically typed target languages, such as Java, use the
language's built in static type checking and have no need for a SWIG
runtime. All the dynamically typed / interpreted languages rely on the
SWIG runtime.</p>
<p> The runtime functions are private to each SWIG-generated module.
That is, the runtime functions are declared with &quot;static&quot; linkage and
are visible only to the wrapper functions defined in that module. The
only problem with this approach is that when more than one SWIG module
is used in the same application, those modules often need to share type
information. This is especially true for C++ programs where SWIG must
collect and share information about inheritance relationships that
cross module boundaries.</p>
<p> To solve the problem of sharing information across modules, a
pointer to the type information is stored in a global variable in the
target language namespace. During module initialization, type
information is loaded into the global data structure of type
information from all modules.</p>
<p> There are a few trade offs with this approach. This type information
is global across all SWIG modules loaded, and can cause type conflicts
between modules that were not designed to work together. To solve this
approach, the SWIG runtime code uses a define SWIG_TYPE_TABLE to
provide a unique type table. This behavior can be enabled when
compiling the generated _wrap.cxx or _wrap.c file by adding
-DSWIG_TYPE_TABLE=myprojectname to the command line argument.</p>
<p> Then, only modules compiled with SWIG_TYPE_TABLE set to
myprojectname will share type information. So if your project has three
modules, all three should be compiled with
-DSWIG_TYPE_TABLE=myprojectname, and then these three modules will
share type information. But any other project's types will not
interfere or clash with the types in your module.</p>
<p> Another issue relating to the global type table is thread safety. If
two modules try and load at the same time, the type information can
become corrupt. SWIG currently does not provide any locking, and if you
use threads, you must make sure that modules are loaded serially. Be
careful if you use threads and the automatic module loading that some
scripting languages provide. One solution is to load all modules before
spawning any threads, or use SWIG_TYPE_TABLE to separate type tables so
they do not clash with each other.</p>
<p> Lastly, SWIG uses a #define SWIG_RUNTIME_VERSION, located in
Lib/swigrun.swg and near the top of every generated module. This number
gets incremented when the data structures change, so that SWIG modules
generated with different versions can peacefully coexist. So the type
structures are separated by the (SWIG_TYPE_TABLE, SWIG_RUNTIME_VERSION)
pair, where by default SWIG_TYPE_TABLE is empty. Only modules compiled
with the same pair will share type information.</p>
<h2><a name="Modules_external_run_time">20.4 External access to the
runtime</a></h2>
<p>As described in <a href="#Typemaps_runtime_type_checker">The run-time
type checker</a>, the functions <tt>SWIG_TypeQuery</tt>, <tt>
SWIG_NewPointerObj</tt>, and others sometimes need to be called. Calling
these functions from a typemap is supported, since the typemap code is
embedded into the <tt>_wrap.c</tt> file, which has those declarations
available. If you need to call the SWIG run-time functions from another
C file, there is one header you need to include. To generate the header
that needs to be included, SWIG can be run in a different mode via <tt>
-external-runtime</tt> to generate the run-time instead of the normal
mode of processing an input interface file. For example:<div class="shell">
<pre>
$ swig -python -external-runtime &lt;filename&gt;
</pre>
</div></p>
<p>The filename argument is optional and if it is not passed, then the
default filename will be something like <tt>swigpyrun.h</tt>, depending
on the language. This header file should be treated like any of the
other _wrap.c output files, and should be regenerated when the _wrap
files are. After including this header, your code will be able to call <tt>
SWIG_TypeQuery</tt>, <tt>SWIG_NewPointerObj</tt>, <tt>SWIG_ConvertPtr</tt>
and others. The exact argument parameters for these functions might
differ between language modules; please check the language module
chapters for more information.</p>
<p>Inside this header the functions are declared static and are included
inline into the file, and thus the file does not need to be linked
against any SWIG libraries or code (you might still need to link
against the language libraries like libpython-2.3). Data is shared
between this file and the _wrap.c files through a global variable in
the scripting language. It is also possible to copy this header file
along with the generated wrapper files into your own package, so that
you can distribute a package that can be compiled without SWIG
installed (this works because the header file is self-contained, and
does not need to link with anything).</p>
<p> This header will also use the -DSWIG_TYPE_TABLE described above, so
when compiling any code which includes the generated header file should
define the SWIG_TYPE_TABLE to be the same as the module whose types you
are trying to access.</p>
<h2><a name="Modules_nn4">20.5 A word of caution about static libraries</a>
</h2>
<p> When working with multiple SWIG modules, you should take care not to
use static libraries. For example, if you have a static library <tt>
libfoo.a</tt> and you link a collection of SWIG modules with that
library, each module will get its own private copy of the library code
inserted into it. This is very often<b> NOT</b> what you want and it
can lead to unexpected or bizarre program behavior. When working with
dynamically loadable modules, you should try to work exclusively with
shared libraries.</p>
<h2><a name="Modules_nn5">20.6 References</a></h2>
<p> Due to the complexity of working with shared libraries and multiple
modules, it might be a good idea to consult an outside reference. John
Levine's &quot;Linkers and Loaders&quot; is highly recommended.</p>
<h2><a name="Modules_nn6">20.7 Reducing the wrapper file size</a></h2>
<p> Using multiple modules with the <tt>%import</tt> directive is the
most common approach to modularising large projects. In this way a
number of different wrapper files can be generated, thereby avoiding
the generation of a single large wrapper file. There are a couple of
alternative solutions for reducing the size of a wrapper file through
the use of command line options and features.</p>
<p><b> -fcompact</b>
<br> This command line option will compact the size of the wrapper file
without changing the code generated into the wrapper file. It simply
removes blank lines and joins lines of code together. This is useful
for compilers that have a maximum file size that can be handled.</p>
<p><b> -fvirtual</b>
<br> This command line option will remove the generation of superfluous
virtual method wrappers. Consider the following inheritance hierarchy:</p>
<div class="code">
<pre>
struct Base {
virtual void method();
...
};
struct Derived : Base {
virtual void method();
...
};
</pre>
</div>
<p> Normally wrappers are generated for both methods, whereas this
command line option will suppress the generation of a wrapper for <tt>
Derived::method</tt>. Normal polymorphic behaviour remains as <tt>
Derived::method</tt> will still be called should you have a <tt>Derived</tt>
instance and call the wrapper for <tt>Base::method</tt>.</p>
<p><b> %feature(&quot;compactdefaultargs&quot;)</b>
<br> This feature can reduce the number of wrapper methods when wrapping
methods with default arguments. The section on <a href="#SWIGPlus_default_args">
default arguments</a> discusses the feature and its limitations.</p>
<HR NOSHADE>
<h1><a name="CCache">21 Using SWIG with ccache - ccache-swig(1) manpage</a>
</h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#CCache_nn2">NAME</a></li>
<li><a href="#CCache_nn3">SYNOPSIS</a></li>
<li><a href="#CCache_nn4">DESCRIPTION</a></li>
<li><a href="#CCache_nn5">OPTIONS SUMMARY</a></li>
<li><a href="#CCache_nn6">OPTIONS</a></li>
<li><a href="#CCache_nn7">INSTALLATION</a></li>
<li><a href="#CCache_nn8">EXTRA OPTIONS</a></li>
<li><a href="#CCache_nn9">ENVIRONMENT VARIABLES</a></li>
<li><a href="#CCache_nn10">CACHE SIZE MANAGEMENT</a></li>
<li><a href="#CCache_nn11">CACHE COMPRESSION</a></li>
<li><a href="#CCache_nn12">HOW IT WORKS</a></li>
<li><a href="#CCache_nn13">USING CCACHE WITH DISTCC</a></li>
<li><a href="#CCache_nn14">SHARING A CACHE</a></li>
<li><a href="#CCache_nn15">HISTORY</a></li>
<li><a href="#CCache_nn16">DIFFERENCES FROM COMPILERCACHE</a></li>
<li><a href="#CCache_nn17">CREDITS</a></li>
<li><a href="#CCache_nn18">AUTHOR</a></li>
</ul>
</div>
<!-- INDEX -->
<p></p>
<h2><a name="CCache_nn2">21.1 NAME</a></h2>
<p> ccache-swig - a fast compiler cache</p>
<p></p>
<h2><a name="CCache_nn3">21.2 SYNOPSIS</a></h2>
<p> ccache-swig [OPTION]</p>
<p> ccache-swig &lt;compiler&gt; [COMPILER OPTIONS]</p>
<p> &lt;compiler&gt; [COMPILER OPTIONS]</p>
<p></p>
<h2><a name="CCache_nn4">21.3 DESCRIPTION</a></h2>
<p> ccache-swig is a compiler cache. It speeds up re-compilation of
C/C++/SWIG code by caching previous compiles and detecting when the
same compile is being done again. ccache-swig is ccache plus support
for SWIG. ccache and ccache-swig are used interchangeably in this
document.</p>
<p></p>
<h2><a name="CCache_nn5">21.4 OPTIONS SUMMARY</a></h2>
<p> Here is a summary of the options to ccache-swig.</p>
<p></p>
<pre>
-s show statistics summary
-z zero statistics
-c run a cache cleanup
-C clear the cache completely
-F &lt;n&gt; set maximum files in cache
-M &lt;n&gt; set maximum size of cache (use G, M or K)
-h this help page
-V print version number
</pre>
<p></p>
<h2><a name="CCache_nn6">21.5 OPTIONS</a></h2>
<p> These options only apply when you invoke ccache as &quot;ccache-swig&quot;.
When invoked as a compiler none of these options apply. In that case
your normal compiler options apply and you should refer to your
compilers documentation.</p>
<p></p>
<dl>
<p>
<dt><strong><strong>-h</strong></strong></dt>
<dd> Print a options summary page
<p></p>
<p></p>
</dd>
<dt><strong><strong>-s</strong></strong></dt>
<dd> Print the current statistics summary for the cache. The statistics
are stored spread across the subdirectories of the cache. Using
&quot;ccache-swig -s&quot; adds up the statistics across all subdirectories and
prints the totals.
<p></p>
<p></p>
</dd>
<dt><strong><strong>-z</strong></strong></dt>
<dd> Zero the cache statistics.
<p></p>
<p></p>
</dd>
<dt><strong><strong>-V</strong></strong></dt>
<dd> Print the ccache version number
<p></p>
<p></p>
</dd>
<dt><strong><strong>-c</strong></strong></dt>
<dd> Clean the cache and re-calculate the cache file count and size
totals. Normally the -c option should not be necessary as ccache keeps
the cache below the specified limits at runtime and keeps statistics up
to date on each compile. This option is mostly useful if you manually
modify the cache contents or believe that the cache size statistics may
be inaccurate.
<p></p>
<p></p>
</dd>
<dt><strong><strong>-C</strong></strong></dt>
<dd> Clear the entire cache, removing all cached files.
<p></p>
<p></p>
</dd>
<dt><strong><strong>-F &lt;maxfiles&gt;</strong></strong></dt>
<dd> This sets the maximum number of files allowed in the cache. The
value is stored inside the cache directory and applies to all future
compiles. Due to the way the value is stored the actual value used is
always rounded down to the nearest multiple of 16.
<p></p>
<p></p>
</dd>
<dt><strong><strong>-M &lt;maxsize&gt;</strong></strong></dt>
<dd> This sets the maximum cache size. You can specify a value in
gigabytes, megabytes or kilobytes by appending a G, M or K to the
value. The default is gigabytes. The actual value stored is rounded
down to the nearest multiple of 16 kilobytes.
<p></p>
</dd>
</p>
</dl>
<p></p>
<h2><a name="CCache_nn7">21.6 INSTALLATION</a></h2>
<p> There are two ways to use ccache. You can either prefix your compile
commands with &quot;ccache-swig&quot; or you can create a symbolic link between
ccache-swig and the names of your compilers. The first method is most
convenient if you just want to try out ccache or wish to use it for
some specific projects. The second method is most useful for when you
wish to use ccache for all your compiles.</p>
<p> To install for usage by the first method just copy ccache-swig to
somewhere in your path.</p>
<p> To install for the second method do something like this:</p>
<pre>
cp ccache-swig /usr/local/bin/
ln -s /usr/local/bin/ccache-swig /usr/local/bin/gcc
ln -s /usr/local/bin/ccache-swig /usr/local/bin/g++
ln -s /usr/local/bin/ccache-swig /usr/local/bin/cc
ln -s /usr/local/bin/ccache-swig /usr/local/bin/swig
</pre>
This will work as long as /usr/local/bin comes before the path to gcc
(which is usually in /usr/bin). After installing you may wish to run
&quot;which gcc&quot; to make sure that the correct link is being used.
<p> Note! Do not use a hard link, use a symbolic link. A hardlink will
cause &quot;interesting&quot; problems.</p>
<p></p>
<h2><a name="CCache_nn8">21.7 EXTRA OPTIONS</a></h2>
<p> When run as a compiler front end ccache usually just takes the same
command line options as the compiler you are using. The only exception
to this is the option '--ccache-skip'. That option can be used to tell
ccache that the next option is definitely not a input filename, and
should be passed along to the compiler as-is.</p>
<p> The reason this can be important is that ccache does need to parse
the command line and determine what is an input filename and what is a
compiler option, as it needs the input filename to determine the name
of the resulting object file (among other things). The heuristic ccache
uses in this parse is that any string on the command line that exists
as a file is treated as an input file name (usually a C file). By using
--ccache-skip you can force an option to not be treated as an input
file name and instead be passed along to the compiler as a command line
option.</p>
<p></p>
<h2><a name="CCache_nn9">21.8 ENVIRONMENT VARIABLES</a></h2>
<p> ccache uses a number of environment variables to control operation.
In most cases you won't need any of these as the defaults will be fine.</p>
<p></p>
<dl>
<p></p>
<p>
<dt><strong><strong>CCACHE_DIR</strong></strong></dt>
<dd> the CCACHE_DIR environment variable specifies where ccache will
keep its cached compiler output. The default is &quot;$HOME/.ccache&quot;.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_TEMPDIR</strong></strong></dt>
<dd> the CCACHE_TEMPDIR environment variable specifies where ccache will
put temporary files. The default is the same as CCACHE_DIR. Note that
the CCACHE_TEMPDIR path must be on the same filesystem as the
CCACHE_DIR path, so that renames of files between the two directories
can work.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_LOGFILE</strong></strong></dt>
<dd> If you set the CCACHE_LOGFILE environment variable then ccache will
write some log information on cache hits and misses in that file. This
is useful for tracking down problems.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_VERBOSE</strong></strong></dt>
<dd> If you set the CCACHE_VERBOSE environment variable then ccache will
display on stdout all the compiler invocations that it makes. This can
useful for debugging unexpected problems.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_PATH</strong></strong></dt>
<dd> You can optionally set CCACHE_PATH to a colon separated path where
ccache will look for the real compilers. If you don't do this then
ccache will look for the first executable matching the compiler name in
the normal PATH that isn't a symbolic link to ccache itself.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_CC</strong></strong></dt>
<dd> You can optionally set CCACHE_CC to force the name of the compiler
to use. If you don't do this then ccache works it out from the command
line.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_PREFIX</strong></strong></dt>
<dd> This option adds a prefix to the command line that ccache runs when
invoking the compiler. Also see the section below on using ccache with
distcc.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_DISABLE</strong></strong></dt>
<dd> If you set the environment variable CCACHE_DISABLE then ccache will
just call the real compiler, bypassing the cache completely.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_READONLY</strong></strong></dt>
<dd> the CCACHE_READONLY environment variable tells ccache to attempt to
use existing cached object files, but not to try to add anything new to
the cache. If you are using this because your CCACHE_DIR is read-only,
then you may find that you also need to set CCACHE_TEMPDIR as otherwise
ccache will fail to create the temporary files.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_CPP2</strong></strong></dt>
<dd> If you set the environment variable CCACHE_CPP2 then ccache will
not use the optimisation of avoiding the 2nd call to the pre-processor
by compiling the pre-processed output that was used for finding the
hash in the case of a cache miss. This is primarily a debugging option,
although it is possible that some unusual compilers will have problems
with the intermediate filename extensions used in this optimisation, in
which case this option could allow ccache to be used.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_NOCOMPRESS</strong></strong></dt>
<dd> If you set the environment variable CCACHE_NOCOMPRESS then there is
no compression used on files that go into the cache. However, this
setting has no effect on how files are retrieved from the cache,
compressed results will still be usable.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_NOSTATS</strong></strong></dt>
<dd> If you set the environment variable CCACHE_NOSTATS then ccache will
not update the statistics files on each compile.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_NLEVELS</strong></strong></dt>
<dd> The environment variable CCACHE_NLEVELS allows you to choose the
number of levels of hash in the cache directory. The default is 2. The
minimum is 1 and the maximum is 8.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_HARDLINK</strong></strong></dt>
<dd> If you set the environment variable CCACHE_HARDLINK then ccache
will attempt to use hard links from the cache directory when creating
the compiler output rather than using a file copy. Using hard links is
faster, but can confuse programs like 'make' that rely on modification
times. Hard links are never made for compressed cache files.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_RECACHE</strong></strong></dt>
<dd> This forces ccache to not use any cached results, even if it finds
them. New results are still cached, but existing cache entries are
ignored.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_UMASK</strong></strong></dt>
<dd> This sets the umask for ccache and all child processes (such as the
compiler). This is mostly useful when you wish to share your cache with
other users. Note that this also affects the file permissions set on
the object files created from your compilations.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_HASHDIR</strong></strong></dt>
<dd> This tells ccache to hash the current working directory when
calculating the hash that is used to distinguish two compiles. This
prevents a problem with the storage of the current working directory in
the debug info of a object file, which can lead ccache to give a cached
object file that has the working directory in the debug info set
incorrectly. This option is off by default as the incorrect setting of
this debug info rarely causes problems. If you strike problems with gdb
not using the correct directory then enable this option.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_UNIFY</strong></strong></dt>
<dd> If you set the environment variable CCACHE_UNIFY then ccache will
use the C/C++ unifier when hashing the pre-processor output if -g is
not used in the compile. The unifier is slower than a normal hash, so
setting this environment variable loses a little bit of speed, but it
means that ccache can take advantage of not recompiling when the
changes to the source code consist of reformatting only. Note that
using CCACHE_UNIFY changes the hash, so cached compiles with
CCACHE_UNIFY set cannot be used when CCACHE_UNIFY is not set and vice
versa. The reason the unifier is off by default is that it can give
incorrect line number information in compiler warning messages.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_EXTENSION</strong></strong></dt>
<dd> Normally ccache tries to automatically determine the extension to
use for intermediate C pre-processor files based on the type of file
being compiled. Unfortunately this sometimes doesn't work, for example
when using the aCC compiler on HP-UX. On systems like this you can use
the CCACHE_EXTENSION option to override the default. On HP-UX set this
environment variable to &quot;i&quot; if you use the aCC compiler.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_STRIPC</strong></strong></dt>
<dd> If you set the environment variable CCACHE_STRIPC then ccache will
strip the -c option when invoking the preprocessor. This option is
primarily for the Sun Workshop C++ compiler as without this option an
unwarranted warning is displayed: CC: Warning: &quot;-E&quot; redefines product
from &quot;object&quot; to &quot;source (stdout)&quot; when -E and -c is used together.
<p></p>
<p></p>
</dd>
<dt><strong><strong>CCACHE_SWIG</strong></strong></dt>
<dd> When using SWIG as the compiler and it does not have 'swig' in the
executable name, then the CCACHE_SWIG environment variable needs to be
set in order for ccache to work correctly with SWIG. The use of
CCACHE_CPP2 is also recommended for SWIG due to some preprocessor
quirks, however, use of CCACHE_CPP2 can often be skipped -- check your
generated code with and without this option set. Known problems are
using preprocessor directives within %inline blocks and the use of
'#pragma SWIG'.
<p></p>
</dd>
</p>
</dl>
<p></p>
<h2><a name="CCache_nn10">21.9 CACHE SIZE MANAGEMENT</a></h2>
<p> By default ccache has a one gigabyte limit on the cache size and no
maximum number of files. You can set a different limit using the
&quot;ccache -M&quot; and &quot;ccache -F&quot; options, which set the size and number of
files limits.</p>
<p> When these limits are reached ccache will reduce the cache to 20%
below the numbers you specified in order to avoid doing the cache clean
operation too often.</p>
<p></p>
<h2><a name="CCache_nn11">21.10 CACHE COMPRESSION</a></h2>
<p> By default on most platforms ccache will compress all files it puts
into the cache using the zlib compression. While this involves a
negligible performance slowdown, it significantly increases the number
of files that fit in the cache. You can turn off compression setting
the CCACHE_NOCOMPRESS environment variable.</p>
<p></p>
<h2><a name="CCache_nn12">21.11 HOW IT WORKS</a></h2>
<p> The basic idea is to detect when you are compiling exactly the same
code a 2nd time and use the previously compiled output. You detect that
it is the same code by forming a hash of:</p>
<p></p>
<ul>
<li> the pre-processor output from running the compiler with -E</li>
<li> the command line options</li>
<li> the real compilers size and modification time</li>
<li> any stderr output generated by the compiler</li>
</ul>
<p> These are hashed using md4 (a strong hash) and a cache file is
formed based on that hash result. When the same compilation is done a
second time ccache is able to supply the correct compiler output
(including all warnings etc) from the cache.</p>
<p> ccache has been carefully written to always produce exactly the same
compiler output that you would get without the cache. If you ever
discover a case where ccache changes the output of your compiler then
please let me know.</p>
<p></p>
<h2><a name="CCache_nn13">21.12 USING CCACHE WITH DISTCC</a></h2>
<p> distcc is a very useful program for distributing compilation across
a range of compiler servers. It is often useful to combine distcc with
ccache, so that compiles that are done are sped up by distcc, but that
ccache avoids the compile completely where possible.</p>
<p> To use distcc with ccache I recommend using the CCACHE_PREFIX
option. You just need to set the environment variable CCACHE_PREFIX to
'distcc' and ccache will prefix the command line used with the compiler
with the command 'distcc'.</p>
<p></p>
<h2><a name="CCache_nn14">21.13 SHARING A CACHE</a></h2>
<p> A group of developers can increase the cache hit rate by sharing a
cache directory. The hard links however cause unwanted side effects, as
all links to a cached file share the file's modification timestamp.
This results in false dependencies to be triggered by timestamp-based
build systems whenever another user links to an existing file.
Typically, users will see that their libraries and binaries are
relinked without reason. To share a cache without side effects, the
following conditions need to be met:</p>
<p></p>
<ul>
<li> Use the same<strong> CCACHE_DIR</strong> environment variable
setting</li>
<li> Unset the<strong> CCACHE_HARDLINK</strong> environment variable</li>
<li> Make sure everyone sets the CCACHE_UMASK environment variable to
002, this ensures that cached files are accessible to everyone in the
group.</li>
<li> Make sure that all users have write permission in the entire cache
directory (and that you trust all users of the shared cache).</li>
<li> Make sure that the setgid bit is set on all directories in the
cache. This tells the filesystem to inherit group ownership for new
directories. The command &quot;chmod g+s `find $CCACHE_DIR -type d`&quot; might
be useful for this.</li>
<li> Set<strong> CCACHE_NOCOMPRESS</strong> for all users, if there are
users with versions of ccache that do not support compression.</li>
</ul>
<p></p>
<h2><a name="CCache_nn15">21.14 HISTORY</a></h2>
<p> ccache was inspired by the compilercache shell script written by
Erik Thiele and I would like to thank him for an excellent piece of
work. See <a href="http://www.erikyyy.de/compilercache/">
http://www.erikyyy.de/compilercache/</a> for the Erik's scripts.
ccache-swig is a port of the original ccache with support added for use
with SWIG.</p>
<p> I wrote ccache because I wanted to get a bit more speed out of a
compiler cache and I wanted to remove some of the limitations of the
shell-script version.</p>
<p></p>
<h2><a name="CCache_nn16">21.15 DIFFERENCES FROM COMPILERCACHE</a></h2>
<p> The biggest differences between Erik's compilercache script and
ccache are:</p>
<ul>
<li> ccache is written in C, which makes it a bit faster (calling out to
external programs is mostly what slowed down the scripts).</li>
<li> ccache can automatically find the real compiler</li>
<li> ccache keeps statistics on hits/misses</li>
<li> ccache can do automatic cache management</li>
<li> ccache can cache compiler output that includes warnings. In many
cases this gives ccache a much higher cache hit rate.</li>
<li> ccache can handle a much wider ranger of compiler options</li>
<li> ccache avoids a double call to cpp on a cache miss</li>
</ul>
<p></p>
<h2><a name="CCache_nn17">21.16 CREDITS</a></h2>
<p> Thanks to the following people for their contributions to ccache</p>
<ul>
<li> Erik Thiele for the original compilercache script</li>
<li> Luciano Rocha for the idea of compiling the pre-processor output to
avoid a 2nd cpp pass</li>
<li> Paul Russell for many suggestions and the debian packaging</li>
</ul>
<p></p>
<h2><a name="CCache_nn18">21.17 AUTHOR</a></h2>
<p> ccache was written by Andrew Tridgell <a href="https://www.samba.org/~tridge/">
https://www.samba.org/~tridge/</a>. ccache was adapted to create
ccache-swig for use with SWIG by William Fulton.</p>
<p> If you wish to report a problem or make a suggestion then please
email the SWIG developers on the swig-devel mailing list, see <a href="https://www.swig.org/mail.html">
https://www.swig.org/mail.html</a></p>
<p> ccache is released under the GNU General Public License version 2 or
later. Please see the file COPYING for license details.</p>
<p></p>
<HR NOSHADE>
<h1><a name="Android">22 SWIG and Android</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Android_overview">Overview</a></li>
<li><a href="#Android_examples">Android examples</a>
<ul>
<li><a href="#Android_examples_intro">Examples introduction</a></li>
<li><a href="#Android_example_simple">Simple C example</a></li>
<li><a href="#Android_example_class">C++ class example</a></li>
<li><a href="#Android_examples_other">Other examples</a></li>
</ul>
</li>
<li><a href="#Android_stl">C++ STL</a></li>
</ul>
</div>
<!-- INDEX -->
<p> This chapter describes SWIG's support of Android.</p>
<h2><a name="Android_overview">22.1 Overview</a></h2>
<p> The Android chapter is fairly short as support for Android is the
same as for Java, where the Java Native Interface (JNI) is used to call
from Android Java into C or C++ compiled code. Everything in the <a href="#Java">
Java chapter</a> applies to generating code for access from Android Java
code. This chapter contains a few Android specific notes and examples.</p>
<h2><a name="Android_examples">22.2 Android examples</a></h2>
<h3><a name="Android_examples_intro">22.2.1 Examples introduction</a></h3>
<p> The examples require the <a href="https://developer.android.com/sdk/">
Android SDK</a> and <a href="https://developer.android.com/ndk/">Android
NDK</a> which can be installed as per instructions in the links. The
Eclipse version is not required for these examples as just the command
line tools are used (shown for Linux as the host, but Windows will be
very similar, if not identical in most places). Add the SDK tools and
NDK tools to your path and create a directory somewhere for your
Android projects (adjust PATH as necessary to where you installed the
tools):</p>
<div class="shell">
<pre>
$ export PATH=$HOME/android/android-sdk-linux_x86/tools:$HOME/android/android-sdk-linux_x86/platform-tools:$HOME/android/android-ndk-r6b:$PATH
$ mkdir AndroidApps
$ cd AndroidApps
</pre>
</div>
<p> The examples use a target id of 1. This might need changing
depending on your setup. After installation of the Android SDK, the
available target ids can be viewed by running the command below. Please
adjust the id to suit your target device.</p>
<div class="shell">
<pre>
$ android list targets
</pre>
</div>
<p> The following examples are shipped with SWIG under the
Examples/android directory and include a Makefile to build and install
each example.</p>
<h3><a name="Android_example_simple">22.2.2 Simple C example</a></h3>
<p> This simple C example shows how to call a C function as well as read
and modify a global variable. First we'll create and build a pure Java
Android app. Afterwards the JNI code will be generated by SWIG and
built into the app. First create and build an app called <tt>SwigSimple</tt>
in a subdirectory called <tt>simple</tt> using the commands below.
Adjust the <tt>--target</tt> id as mentioned earlier in the <a href="#Android_examples_intro">
Examples introduction</a>. <a href="http://developer.android.com/guide/developing/projects/projects-cmdline.html">
Managing Projects from the Command Line</a> on the Android developer's
site is a useful reference for these steps.</p>
<div class="shell">
<pre>
$ android create project --target 1 --name SwigSimple --path ./simple --activity SwigSimple --package org.swig.simple
$ cd simple
$ ant debug
</pre>
</div>
<p> Modify <tt>src/org/swig/simple/SwigSimple.java</tt> from the default
to:</p>
<div class="code">
<pre>
package org.swig.simple;
import android.app.Activity;
import android.os.Bundle;
import android.view.View;
import android.widget.Button;
import android.widget.TextView;
import android.widget.ScrollView;
import android.text.method.ScrollingMovementMethod;
public class SwigSimple extends Activity
{
TextView outputText = null;
ScrollView scroller = null;
/** Called when the activity is first created. */
@Override
public void onCreate(Bundle savedInstanceState)
{
super.onCreate(savedInstanceState);
setContentView(R.layout.main);
outputText = (TextView)findViewById(R.id.OutputText);
outputText.setText(&quot;Press 'Run' to start...\n&quot;);
outputText.setMovementMethod(new ScrollingMovementMethod());
scroller = (ScrollView)findViewById(R.id.Scroller);
}
public void onRunButtonClick(View view)
{
outputText.append(&quot;Started...\n&quot;);
nativeCall();
outputText.append(&quot;Finished!\n&quot;);
// Ensure scroll to end of text
scroller.post(new Runnable() {
public void run() {
scroller.fullScroll(ScrollView.FOCUS_DOWN);
}
});
}
/** Calls into C/C++ code */
public void nativeCall()
{
// TODO
}
}
</pre>
</div>
<p> The above simply adds a<i> Run</i> button and scrollable text view
as the GUI aspects of the program. The associated resources need to be
created, modify <tt>res/layout/main.xml</tt> as follows:</p>
<div class="code">
<pre>
&lt;?xml version=&quot;1.0&quot; encoding=&quot;utf-8&quot;?&gt;
&lt;LinearLayout xmlns:android=&quot;http://schemas.android.com/apk/res/android&quot;
android:orientation=&quot;vertical&quot;
android:layout_width=&quot;fill_parent&quot;
android:layout_height=&quot;fill_parent&quot;
&gt;
&lt;Button
android:id=&quot;@+id/RunButton&quot;
android:layout_width=&quot;wrap_content&quot;
android:layout_height=&quot;wrap_content&quot;
android:text=&quot;Run...&quot;
android:onClick=&quot;onRunButtonClick&quot;
/&gt;
&lt;ScrollView
android:id=&quot;@+id/Scroller&quot;
android:layout_width=&quot;fill_parent&quot;
android:layout_height=&quot;fill_parent&quot;
&gt;
&lt;TextView
android:id=&quot;@+id/OutputText&quot;
android:layout_width=&quot;wrap_content&quot;
android:layout_height=&quot;wrap_content&quot;
/&gt;
&lt;/ScrollView&gt;
&lt;/LinearLayout&gt;
</pre>
</div>
<p> Rebuild the project with your changes:</p>
<div class="shell">
<pre>
$ ant debug
</pre>
</div>
<p> Although there are no native function calls in the code, yet, you
may want to check that this simple pure Java app runs before adding in
the native calls. First, set up your Android device for hardware
debugging, see <a href="http://developer.android.com/guide/developing/device.html">
Using hardware devices</a> on the Android developer's site. When
complete your device should be listed in those attached, something
like:</p>
<div class="shell">
<pre>
$ adb devices
List of devices attached
A32-6DBE0001-9FF80000-015D62C3-02018028 device
</pre>
</div>
<p> This means you are now ready to install the application...</p>
<div class="shell">
<pre>
$ adb install bin/SwigSimple-debug.apk
95 KB/s (4834 bytes in 0.049s)
pkg: /data/local/tmp/SwigSimple-debug.apk
Success
</pre>
</div>
<p> The newly installed 'SwigSimple' app will be amongst all your other
applications on the home screen. Run the app and it will show a<i> Run</i>
button text box below it. Press the<i> Run</i> button to see the simple
text output.</p>
<p> The application can be uninstalled like any other application and in
fact must be uninstalled before installing an updated version.
Uninstalling is quite easy too from your host computer:</p>
<div class="shell">
<pre>
$ adb uninstall org.swig.simple
Success
</pre>
</div>
<p> Now that you have a pure Java Android app working, let's add some
JNI code generated from SWIG.</p>
<p> First create a <tt>jni</tt> subdirectory and then create some C
source code in <tt>jni/example.c</tt>:</p>
<div class="code">
<pre>
/* File : example.c */
/* A global variable */
double Foo = 3.0;
/* Compute the greatest common divisor of positive integers */
int gcd(int x, int y) {
int g;
g = y;
while (x &gt; 0) {
g = x;
x = y % x;
y = g;
}
return g;
}
</pre>
</div>
<p> Create a SWIG interface file for this C code, <tt>jni/example.i</tt>
:</p>
<div class="code">
<pre>
/* File : example.i */
%module example
%inline %{
extern int gcd(int x, int y);
extern double Foo;
%}
</pre>
</div>
<p> Invoke SWIG as follows:</p>
<div class="shell">
<pre>
$ swig -java -package org.swig.simple -outdir src/org/swig/simple -o jni/example_wrap.c jni/example.i
</pre>
</div>
<p> SWIG generates the following files:</p>
<ul>
<li><tt>src/org/swig/simple/exampleJNI.java</tt></li>
<li><tt>src/org/swig/simple/example.java</tt></li>
<li><tt>jni/example_wrap.c</tt></li>
</ul>
<p> Next we need to create a standard Android NDK build system file <tt>
jni/Android.mk</tt>:</p>
<div class="code">
<pre>
# File: Android.mk
LOCAL_PATH := $(call my-dir)
include $(CLEAR_VARS)
LOCAL_MODULE := example
LOCAL_SRC_FILES := example_wrap.c example.c
include $(BUILD_SHARED_LIBRARY)
</pre>
</div>
<p> See the <a href="https://developer.android.com/ndk/">Android NDK
documentation</a> for more on the NDK build system and getting started
with the NDK. A simple invocation of ndk-build will compile the .c
files and generate a shared object/system library. Output will be
similar to:</p>
<div class="shell">
<pre>
$ ndk-build
Compile thumb : example &lt;= example_wrap.c
Compile thumb : example &lt;= example.c
SharedLibrary : libexample.so
Install : libexample.so =&gt; libs/armeabi/libexample.so
</pre>
</div>
<p> Now that the C JNI layer has been built, we can write Java code to
call into the this layer. Modify the <tt>nativeCall</tt> method in <tt>
src/org/swig/simple/SwigSimple.java</tt> to call the JNI code as follows
and add the static constructor to load the system library containing
the compiled JNI C code:</p>
<div class="code">
<pre>
/** Calls into C/C++ code */
public void nativeCall()
{
// Call our gcd() function
int x = 42;
int y = 105;
int g = example.gcd(x, y);
outputText.append(&quot;The greatest common divisor of &quot; + x + &quot; and &quot; + y + &quot; is &quot; + g + &quot;\n&quot;);
// Manipulate the Foo global variable
// Output its current value
double foo = example.getFoo();
outputText.append(&quot;Foo = &quot; + foo + &quot;\n&quot;);
// Change its value
example.setFoo(3.1415926);
// See if the change took effect
outputText.append(&quot;Foo = &quot; + example.getFoo() + &quot;\n&quot;);
// Restore value
example.setFoo(foo);
}
/** static constructor */
static {
System.loadLibrary(&quot;example&quot;);
}
</pre>
</div>
<p> Compile the Java code as usual, uninstall the old version of the app
if still installed and re-install the new app:</p>
<div class="shell">
<pre>
$ ant debug
$ adb uninstall org.swig.simple
$ adb install bin/SwigSimple-debug.apk
</pre>
</div>
<p> Run the app again and this time you will see the output pictured
below, showing the result of calls into the C code:</p>
<center><img alt="Android screenshot of SwigSimple example" HEIGHT="400" src="android-simple.png"
WIDTH="256"></center>
<h3><a name="Android_example_class">22.2.3 C++ class example</a></h3>
<p> The steps for calling C++ code are almost identical to those in the
previous C code example. All the steps required to compile and use a
simple hierarchy of classes for shapes are shown in this example.</p>
<p> First create an Android project called <tt>SwigClass</tt> in a
subdirectory called <tt>class</tt>. The steps below create and build
the JNI C++ app. Adjust the <tt>--target</tt> id as mentioned earlier
in the <a href="#Android_examples_intro">Examples introduction</a>.</p>
<div class="shell">
<pre>
$ android create project --target 1 --name SwigClass --path ./class --activity SwigClass --package org.swig.classexample
$ cd class
</pre>
</div>
<p> Now create a <tt>jni</tt> subdirectory and then create a C++ header
file <tt>jni/example.h</tt> which defines our hierarchy of shape
classes:</p>
<div class="code">
<pre>
/* File : example.h */
class Shape {
public:
Shape() {
nshapes++;
}
virtual ~Shape() {
nshapes--;
}
double x, y;
void move(double dx, double dy);
virtual double area() = 0;
virtual double perimeter() = 0;
static int nshapes;
};
class Circle : public Shape {
private:
double radius;
public:
Circle(double r) : radius(r) { }
virtual double area();
virtual double perimeter();
};
class Square : public Shape {
private:
double width;
public:
Square(double w) : width(w) { }
virtual double area();
virtual double perimeter();
};
</pre>
</div>
<p> and create the implementation in the <tt>jni/example.cpp</tt> file:</p>
<div class="code">
<pre>
/* File : example.cpp */
#include &quot;example.h&quot;
#define M_PI 3.14159265358979323846
/* Move the shape to a new location */
void Shape::move(double dx, double dy) {
x += dx;
y += dy;
}
int Shape::nshapes = 0;
double Circle::area() {
return M_PI*radius*radius;
}
double Circle::perimeter() {
return 2*M_PI*radius;
}
double Square::area() {
return width*width;
}
double Square::perimeter() {
return 4*width;
}
</pre>
</div>
<p> Create a SWIG interface file for this C++ code in <tt>jni/example.i</tt>
:</p>
<div class="code">
<pre>
/* File : example.i */
%module example
%{
#include &quot;example.h&quot;
%}
/* Let's just grab the original header file here */
%include &quot;example.h&quot;
</pre>
</div>
<p> Invoke SWIG as follows, note that the -c++ option is required for
C++ code:</p>
<div class="shell">
<pre>
$ swig -c++ -java -package org.swig.classexample -outdir src/org/swig/classexample -o jni/example_wrap.cpp jni/example.i
</pre>
</div>
<p> SWIG generates the following files:</p>
<ul>
<li><tt>src/org/swig/classexample/Square.java</tt></li>
<li><tt>src/org/swig/classexample/exampleJNI.java</tt></li>
<li><tt>src/org/swig/classexample/example.java</tt></li>
<li><tt>src/org/swig/classexample/Circle.java</tt></li>
<li><tt>src/org/swig/classexample/Shape.java</tt></li>
<li><tt>jni/example_wrap.cpp</tt></li>
</ul>
<p> Next we need to create an Android NDK build system file for
compiling the C++ code <tt>jni/Android.mk</tt>. The <tt>-frtti</tt>
compiler flag isn't strictly needed for this example, but is needed for
any code that uses C++ RTTI:</p>
<div class="code">
<pre>
# File: Android.mk
LOCAL_PATH := $(call my-dir)
include $(CLEAR_VARS)
LOCAL_MODULE := example
LOCAL_SRC_FILES := example_wrap.cpp example.cpp
LOCAL_CFLAGS := -frtti
include $(BUILD_SHARED_LIBRARY)
</pre>
</div>
<p> A simple invocation of ndk-build will compile the .cpp files and
generate a shared object/system library. Output will be similar to:</p>
<div class="shell">
<pre>
$ ndk-build
Compile++ thumb : example &lt;= example_wrap.cpp
Compile++ thumb : example &lt;= example.cpp
StaticLibrary : libstdc++.a
SharedLibrary : libexample.so
Install : libexample.so =&gt; libs/armeabi/libexample.so
</pre>
</div>
<p> Now that the C JNI layer has been built, we can write Java code to
call into this layer. Modify <tt>
src/org/swig/classexample/SwigClass.java</tt> from the default to:</p>
<div class="code">
<pre>
package org.swig.classexample;
import android.app.Activity;
import android.os.Bundle;
import android.view.View;
import android.widget.Button;
import android.widget.TextView;
import android.widget.ScrollView;
import android.text.method.ScrollingMovementMethod;
public class SwigClass extends Activity
{
TextView outputText = null;
ScrollView scroller = null;
/** Called when the activity is first created. */
@Override
public void onCreate(Bundle savedInstanceState)
{
super.onCreate(savedInstanceState);
setContentView(R.layout.main);
outputText = (TextView)findViewById(R.id.OutputText);
outputText.setText(&quot;Press 'Run' to start...\n&quot;);
outputText.setMovementMethod(new ScrollingMovementMethod());
scroller = (ScrollView)findViewById(R.id.Scroller);
}
public void onRunButtonClick(View view)
{
outputText.append(&quot;Started...\n&quot;);
nativeCall();
outputText.append(&quot;Finished!\n&quot;);
// Ensure scroll to end of text
scroller.post(new Runnable() {
public void run() {
scroller.fullScroll(ScrollView.FOCUS_DOWN);
}
});
}
/** Calls into C/C++ code */
public void nativeCall()
{
// ----- Object creation -----
outputText.append( &quot;Creating some objects:\n&quot; );
Circle c = new Circle(10);
outputText.append( &quot; Created circle &quot; + c + &quot;\n&quot;);
Square s = new Square(10);
outputText.append( &quot; Created square &quot; + s + &quot;\n&quot;);
// ----- Access a static member -----
outputText.append( &quot;\nA total of &quot; + Shape.getNshapes() + &quot; shapes were created\n&quot; );
// ----- Member data access -----
// Notice how we can do this using functions specific to
// the 'Circle' class.
c.setX(20);
c.setY(30);
// Now use the same functions in the base class
Shape shape = s;
shape.setX(-10);
shape.setY(5);
outputText.append( &quot;\nHere is their current position:\n&quot; );
outputText.append( &quot; Circle = (&quot; + c.getX() + &quot; &quot; + c.getY() + &quot;)\n&quot; );
outputText.append( &quot; Square = (&quot; + s.getX() + &quot; &quot; + s.getY() + &quot;)\n&quot; );
// ----- Call some methods -----
outputText.append( &quot;\nHere are some properties of the shapes:\n&quot; );
Shape[] shapes = {c, s};
for (int i=0; i&lt;shapes.length; i++)
{
outputText.append( &quot; &quot; + shapes[i].toString() + &quot;\n&quot; );
outputText.append( &quot; area = &quot; + shapes[i].area() + &quot;\n&quot; );
outputText.append( &quot; perimeter = &quot; + shapes[i].perimeter() + &quot;\n&quot; );
}
// Notice how the area() and perimeter() functions really
// invoke the appropriate virtual method on each object.
// ----- Delete everything -----
outputText.append( &quot;\nGuess I'll clean up now\n&quot; );
// Note: this invokes the virtual destructor
// You could leave this to the garbage collector
c.delete();
s.delete();
outputText.append( Shape.getNshapes() + &quot; shapes remain\n&quot; );
outputText.append( &quot;Goodbye\n&quot; );
}
/** static constructor */
static {
System.loadLibrary(&quot;example&quot;);
}
}
</pre>
</div>
<p> Note the static constructor and the interesting JNI code is in the <tt>
nativeCall</tt> method. The remaining code deals with the GUI aspects
which are identical to the previous C simple example. Modify <tt>
res/layout/main.xml</tt> to contain the xml for the 'Run' button and
scrollable text view:</p>
<div class="code">
<pre>
&lt;?xml version=&quot;1.0&quot; encoding=&quot;utf-8&quot;?&gt;
&lt;LinearLayout xmlns:android=&quot;http://schemas.android.com/apk/res/android&quot;
android:orientation=&quot;vertical&quot;
android:layout_width=&quot;fill_parent&quot;
android:layout_height=&quot;fill_parent&quot;
&gt;
&lt;Button
android:id=&quot;@+id/RunButton&quot;
android:layout_width=&quot;wrap_content&quot;
android:layout_height=&quot;wrap_content&quot;
android:text=&quot;Run...&quot;
android:onClick=&quot;onRunButtonClick&quot;
/&gt;
&lt;ScrollView
android:id=&quot;@+id/Scroller&quot;
android:layout_width=&quot;fill_parent&quot;
android:layout_height=&quot;fill_parent&quot;
&gt;
&lt;TextView
android:id=&quot;@+id/OutputText&quot;
android:layout_width=&quot;wrap_content&quot;
android:layout_height=&quot;wrap_content&quot;
/&gt;
&lt;/ScrollView&gt;
&lt;/LinearLayout&gt;
</pre>
</div>
<p> Compile the Java code as usual, uninstall the old version of the app
if installed and re-install the new app:</p>
<div class="shell">
<pre>
$ ant debug
$ adb uninstall org.swig.classexample
$ adb install bin/SwigClass-debug.apk
</pre>
</div>
<p> Run the app to see the result of calling the C++ code from Java:</p>
<center><img alt="Android screenshot of SwigClass example" HEIGHT="400" src="android-class.png"
WIDTH="256"></center>
<h3><a name="Android_examples_other">22.2.4 Other examples</a></h3>
<p> The Examples/android directory contains further examples which can
be run and installed in a similar manner to the previous two examples.</p>
<p> Note that the 'extend' example is demonstrates the directors
feature. Normally C++ exception handling and the STL is not available
by default in the version of g++ shipped with Android, but this example
turns these features on as described in the next section.</p>
<h2><a name="Android_stl">22.3 C++ STL</a></h2>
<p> Should the C++ Standard Template Library (STL) be required, an <tt>
Application.mk</tt> file needs to be created in the same directory as
the <tt>Android.mk</tt> directory containing information about the STL
to use. See the NDK documentation in the $NDKROOT/docs folder
especially CPLUSPLUS-SUPPORT.html. Below is an example of the <tt>
Application.mk</tt> file to make the STLport static library available
for use:</p>
<div class="code">
<pre>
# File: Application.mk
APP_STL := gnustl_static
</pre>
</div><HR NOSHADE>
<h1><a name="CSharp">23 SWIG and C#</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#CSharp_introduction">Introduction</a>
<ul>
<li><a href="#CSharp_introduction_swig2_compatibility">SWIG 2
Compatibility</a></li>
<li><a href="#CSharp_commandline">Additional command line options</a></li>
</ul>
</li>
<li><a href="#CSharp_differences_java">Differences to the Java module</a>
</li>
<li><a href="#CSharp_type_mapping">Type mapping</a>
<ul>
<li><a href="#CSharp_primitive_types">Primitive types</a></li>
<li><a href="#CSharp_other_type_mappings">Other types</a></li>
<li><a href="#CSharp_void_pointers">Void pointers</a></li>
</ul>
</li>
<li><a href="#CSharp_arrays">C# Arrays</a>
<ul>
<li><a href="#CSharp_arrays_swig_library">The SWIG C arrays library</a></li>
<li><a href="#CSharp_arrays_pinvoke_default_array_marshalling">Managed
arrays using P/Invoke default array marshalling</a></li>
<li><a href="#CSharp_arrays_pinning">Managed arrays using pinning</a></li>
</ul>
</li>
<li><a href="#CSharp_exceptions">C# Exceptions</a>
<ul>
<li><a href="#CSharp_exception_example_check_typemap">C# exception
example using &quot;check&quot; typemap</a></li>
<li><a href="#CSharp_exception_example_percent_exception">C# exception
example using %exception</a></li>
<li><a href="#CSharp_exception_example_exception_specifications">C#
exception example using exception specifications</a></li>
<li><a href="#CSharp_custom_application_exception">Custom C#
ApplicationException example</a></li>
</ul>
</li>
<li><a href="#CSharp_directors">C# Directors</a>
<ul>
<li><a href="#CSharp_directors_example">Directors example</a></li>
<li><a href="#CSharp_directors_implementation">Directors implementation</a>
</li>
<li><a href="#CSharp_director_caveats">Director caveats</a></li>
</ul>
</li>
<li><a href="#CSharp_multiple_modules">Multiple modules</a></li>
<li><a href="#CSharp_named_arguments">C# named and optional arguments</a>
</li>
<li><a href="#CSharp_typemap_examples">C# Typemap examples</a>
<ul>
<li><a href="#CSharp_memory_management_member_variables">Memory
management when returning references to member variables</a></li>
<li><a href="#CSharp_memory_management_objects">Memory management for
objects passed to the C++ layer</a></li>
<li><a href="#CSharp_date_marshalling">Date marshalling using the csin
typemap and associated attributes</a></li>
<li><a href="#CSharp_date_properties">A date example demonstrating
marshalling of C# properties</a></li>
<li><a href="#CSharp_date_pre_post_directors">Date example demonstrating
the 'pre' and 'post' typemap attributes for directors</a></li>
<li><a href="#CSharp_partial_classes">Turning proxy classes into partial
classes</a></li>
<li><a href="#CSharp_sealed_proxy_class">Turning proxy classes into
sealed classes</a></li>
<li><a href="#CSharp_extending_proxy_class">Extending proxy classes with
additional C# code</a></li>
<li><a href="#CSharp_enum_underlying_type">Underlying type for enums</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="CSharp_introduction">23.1 Introduction</a></h2>
<p> The purpose of the C# module is to offer an automated way of
accessing existing C/C++ code from .NET languages. The wrapper code
implementation uses C# and the Platform Invoke (PInvoke) interface to
access natively compiled C/C++ code. The PInvoke interface has been
chosen over Microsoft's Managed C++ interface as it is portable to both
Microsoft Windows and non-Microsoft platforms. PInvoke is part of the
ECMA/ISO C# specification. It is also better suited for robust
production environments due to the Managed C++ flaw called the <a href="https://msdn.microsoft.com/en-us/ie/aa290048(v=vs.94)">
Mixed DLL Loading Problem</a>. SWIG C# works equally well on
non-Microsoft operating systems such as Linux, Solaris and Apple Mac
using <a href="https://www.mono-project.com/Main_Page/">Mono</a>.</p>
<p> SWIG 3 and later requires .NET 2.0 at a minimum. There are some
minor exceptions, where the minimum required is .NET 4.0. This is when
using the <tt>std::complex</tt> and <tt>std::list</tt> STL containers.</p>
<p> To get the most out of this chapter an understanding of interop is
required. The <a href="https://msdn.microsoft.com">Microsoft Developer
Network (MSDN)</a> has a good reference guide in a section titled
&quot;Interop Marshaling&quot;. Monodoc, available from the Mono project, has a
very useful section titled <a href="https://www.mono-project.com/docs/advanced/pinvoke/">
Interop with native libraries</a>.</p>
<h3><a name="CSharp_introduction_swig2_compatibility">23.1.1 SWIG 2
Compatibility</a></h3>
<p> In order to minimize name collisions between names generated based
on input to SWIG and names used in the generated code from the .NET
framework, SWIG 3 fully qualifies the use of all .NET types.
Furthermore, SWIG 3 avoids <tt>using</tt> directives in generated code.
This breaks backwards compatibility with typemaps, pragmas, etc written
for use with SWIG 2 that assume the presence of <tt>using System;</tt>
or <tt>using System.Runtime.InteropServices;</tt> directives in the
intermediate class imports, module imports, or proxy imports. SWIG 3
supports backwards compatibility though the use of the <tt>SWIG2_CSHARP</tt>
macro. If <tt>SWIG2_CSHARP</tt> is defined, SWIG 3 generates <tt>using</tt>
directives in the intermediate class, module class, and proxy class
code similar to those generated by SWIG 2. This can be done without
modifying any of the input code by passing the <tt>-DSWIG2_CSHARP</tt>
commandline parameter when executing <tt>swig</tt>.</p>
<h3><a name="CSharp_commandline">23.1.2 Additional command line options</a>
</h3>
<p> The following table lists the additional commandline options
available for the C# module. They can also be seen by using:</p>
<div class="code">
<pre>
swig -csharp -help
</pre>
</div>
<table summary="C# specific options">
<tr><th>C# specific options</th></tr>
<tr><td>-dllimport &lt;dl&gt;</td><td>Override DllImport attribute name to
&lt;dl&gt;</td></tr>
<tr><td>-namespace &lt;nm&gt;</td><td>Generate wrappers into C# namespace &lt;nm&gt;</td>
</tr>
<tr><td>-noproxy</td><td>Generate the low-level functional interface
instead of proxy classes</td></tr>
<tr><td>-oldvarnames</td><td>Old intermediary method names for variable
wrappers</td></tr>
<tr><td>-outfile &lt;file&gt;</td><td>Write all C# into a single &lt;file&gt;
located in the output directory</td></tr>
<tr><td>-doxygen</td><td> Convert C++ doxygen comments to CSharp
comments</td></tr>
</table>
<p> The -outfile option combines all the generated C# code into a single
output file instead of creating multiple C# files. The default, when
this option is not provided, is to generate separate .cs files for the
module class, intermediary class and each of the generated proxy and
type wrapper classes. Note that the file extension (.cs) will not be
automatically added and needs to be provided. Due to possible compiler
limits it is not advisable to use <tt>-outfile</tt> for large projects.</p>
<h2><a name="CSharp_differences_java">23.2 Differences to the Java
module</a></h2>
<p> The C# module is very similar to the Java module, so until some more
complete documentation has been written, please use the <a href="#Java">
Java documentation</a> as a guide to using SWIG with C#. The C# module
has the same major SWIG features as the Java module. The rest of this
section should be read in conjunction with the Java documentation as it
lists the main differences. The most notable differences to Java are
the following:</p>
<ul>
<li> When invoking SWIG use the <tt>-csharp</tt> command line option
instead of <tt>-java</tt>.</li>
<li> The <tt>-nopgcpp</tt> command line option does not exist.</li>
<li> The <tt>-package</tt> command line option does not exist.</li>
<li> The <tt>-namespace &lt;name&gt;</tt> commandline option will generate all
code into the namespace specified by <tt>&lt;name&gt;</tt>. C# supports
nested namespaces that are not lexically nested, so nested namespaces
will of course also work. For example: <tt>-namespace com.bloggs.widget</tt>
, will generate code into C# namespaces:<div class="code">
<pre>
namespace com.bloggs.widget {
...
}
</pre>
</div> Note that by default, the generated C# classes have no namespace
and the module name is unrelated to namespaces. The module name is just
like in Java and is merely used to name some of the generated classes.</li>
<li> The <a href="#SWIGPlus_nspace">nspace feature</a> is also supported
as described in this general section with a C# example. Unlike Java
which requires the use of the -package option when using the <tt>nspace</tt>
feature, the -namespace option is not mandatory for C#.</li>
<li> The <tt>-dllimport &lt;name&gt;</tt> commandline option specifies the
name of the DLL for the <tt>DllImport</tt> attribute for every PInvoke
method. If this commandline option is not given, the <tt>DllImport</tt>
DLL name is the same as the module name. This option is useful for when
one wants to invoke SWIG multiple times on different modules, yet
compile all the resulting code into a single DLL.</li>
<li> C/C++ variables are wrapped with C# properties and not JavaBean
style getters and setters.</li>
<li> Global constants are generated into the module class. There is no
constants interface.</li>
<li> There is no implementation for type unsafe enums - not deemed
necessary.</li>
<li> The default enum wrapping approach is proper C# enums, not typesafe
enums.
<br> Note that %csconst(0) will be ignored when wrapping C/C++ enums
with proper C# enums. This is because C# enum items must be initialised
from a compile time constant. If an enum item has an initialiser and
the initialiser doesn't compile as C# code, then the %csconstvalue
directive must be used as %csconst(0) will have no effect. If it was
used, it would generate an illegal runtime initialisation via a PInvoke
call.</li>
<li> C# doesn't support the notion of throws clauses. Therefore there is
no 'throws' typemap attribute support for adding exception classes to a
throws clause. Likewise there is no need for an equivalent to <tt>
%javaexception</tt>. In fact, throwing C# exceptions works quite
differently, see <a href="#CSharp_exceptions">C# Exceptions</a> below.</li>
<li> The majority of the typemaps are in csharp.swg, not java.swg.</li>
<li>
<p>Typemap equivalent names:</p>
<div class="code">
<pre>
jni -&gt; ctype
jtype -&gt; imtype
jstype -&gt; cstype
javain -&gt; csin
javaout -&gt; csout
javadirectorin -&gt; csdirectorin
javadirectorout -&gt; csdirectorout
javainterfaces -&gt; csinterfaces and csinterfaces_derived
javabase -&gt; csbase
javaclassmodifiers -&gt; csclassmodifiers
javacode -&gt; cscode
javaimports -&gt; csimports
javabody -&gt; csbody
javafinalize -&gt; csfinalize
javadestruct -&gt; csdisposing and csdispose
javadestruct_derived -&gt; csdisposing_derived and csdispose_derived
javainterfacemodifiers -&gt; csinterfacemodifiers
javainterfacecode -&gt; csinterfacecode
</pre>
</div></li>
<li>
<p>Typemap macros:</p>
<div class="code">
<pre>
SWIG_JAVABODY_PROXY -&gt; SWIG_CSBODY_PROXY
SWIG_JAVABODY_TYPEWRAPPER -&gt; SWIG_CSBODY_TYPEWRAPPER
</pre>
</div></li>
<li>
<p>Additional typemaps:</p>
<div class="code">
<pre>
csvarin C# code property set typemap
csvarout C# code property get typemap
csattributes C# attributes for attaching to proxy classes/enums
</pre>
</div></li>
<li>
<p>Additional typemap attributes:</p>
<p> The &quot;null&quot; attribute in the &quot;out&quot; typemap can be specified to
provide a value for <tt>$null</tt> to expand into for wrapped functions
that return non-void. Normally the default value of <tt>0</tt> is used.
For example this is needed if you change the return type to void:</p>
<div class="code">
<pre>
%typemap(ctype) Status &quot;void&quot;
%typemap(out, null=&quot;&quot;) Status { ... }
</pre>
</div></li>
<li>
<p>Feature equivalent names:</p>
<div class="code">
<pre>
%javaconst -&gt; %csconst
%javaconstvalue -&gt; %csconstvalue
%javamethodmodifiers -&gt; %csmethodmodifiers
</pre>
</div></li>
<li>
<p>Pragma equivalent names:</p>
<div class="code">
<pre>
%pragma(java) -&gt; %pragma(csharp)
jniclassbase -&gt; imclassbase
jniclassclassmodifiers -&gt; imclassclassmodifiers
jniclasscode -&gt; imclasscode
jniclassimports -&gt; imclassimports
jniclassinterfaces -&gt; imclassinterfaces
</pre>
</div></li>
<li>
<p>Special variable equivalent names:</p>
<div class="code">
<pre>
$javaclassname -&gt; $csclassname
$&amp;javaclassname -&gt; $&amp;csclassname
$*javaclassname -&gt; $*csclassname
$javaclazzname -&gt; $csclazzname
$javainput -&gt; $csinput
$jnicall -&gt; $imcall
$javainterfacename -&gt; $csinterfacename
$&amp;javainterfacename -&gt; $&amp;csinterfacename
$*javainterfacename -&gt; $*csinterfacename
</pre>
</div></li>
<li>
<p> Unlike the &quot;javain&quot; typemap, the &quot;csin&quot; typemap does not support the
'pgcpp' attribute as the C# module does not have a premature garbage
collection prevention parameter. The &quot;csin&quot; typemap supports additional
optional attributes called 'cshin' and 'terminator'. The &quot;csdirectorin&quot;
typemap supports additional optional attributes called 'terminator'.
The 'cshin' attribute should contain the parameter type and name
whenever a <a href="#Java_constructor_helper_function">constructor
helper function</a> is generated due to the 'pre' or 'post' attributes.
The 'terminator' attribute normally just contains a closing brace for
when the 'pre' attribute contains an opening brace, such as when a C# <tt>
using</tt> or <tt>fixed</tt> block is started. Note that 'pre', 'post',
'terminator' and 'cshin' attributes are not used for marshalling the
property set. Please see the <a href="#CSharp_date_marshalling">Date
marshalling example</a> and <a href="#CSharp_date_properties">Date
marshalling of properties example</a> for further understanding of
these &quot;csin&quot; applicable attributes. Please see the <a href="#CSharp_date_pre_post_directors">
Date marshalling director example</a> for further understanding of the
&quot;csdirectorin&quot; attributes.</p>
</li>
<li>
<p> Support for asymmetric type marshalling. The 'ctype', 'imtype' and
'cstype' typemaps support an optional <tt>out</tt> attribute which is
used for output types. If this typemap attribute is specified, then the
type specified in the attribute is used for output types and the type
specified in the typemap itself is used for the input type. If this
typemap attribute is not specified, then the type used for both input
and output is the type specified in the typemap. An example shows that <tt>
char *</tt> could be marshalled in different ways,</p>
<div class="code">
<pre>
%typemap(imtype, out=&quot;global::System.IntPtr&quot;) char * &quot;string&quot;
char * function(char *);
</pre>
</div>
<p> The output type is thus IntPtr and the input type is string. The
resulting intermediary C# code is:</p>
<div class="code">
<pre>
public static extern global::System.IntPtr function(string jarg1);
</pre>
</div></li>
<li>
<p> Support for type attributes. The 'imtype' and 'cstype' typemaps can
have an optional <tt>inattributes</tt> and <tt>outattributes</tt>
typemap attribute. The 'imtype' typemap can also have an optional <tt>
directorinattributes</tt> and <tt>directoroutattributes</tt> typemap
attribute which attaches to director delegates, an implementation
detail of directors, see <a href="#CSharp_directors_implementation">
directors implementation</a>. Note that there are C# attributes and
typemap attributes, don't get confused between the two!! The C#
attributes specified in these typemap attributes are generated wherever
the type is used in the C# wrappers. These can be used to specify any
C# attribute associated with a C/C++ type, but are more typically used
for the C# <tt>MarshalAs</tt> attribute. For example:</p>
<div class="code">
<pre>
%typemap(imtype,
inattributes=&quot;[global::System.Runtime.InteropServices.MarshalAs(UnmanagedType.LPStr)]&quot;,
outattributes=&quot;[return: global::System.Runtime.InteropServices.MarshalAs(UnmanagedType.LPStr)]&quot;) const char * &quot;String&quot;
const char * GetMsg() {}
void SetMsg(const char *msg) {}
</pre>
</div>
<p> The intermediary class will then have the marshalling as specified
by everything in the 'imtype' typemap:</p>
<div class="code">
<pre>
class examplePINVOKE {
...
[global::System.Runtime.InteropServices.DllImport(&quot;example&quot;, EntryPoint=&quot;CSharp_GetMsg&quot;)]
[return: global::System.Runtime.InteropServices.MarshalAs(UnmanagedType.LPStr)]
public static extern String GetMsg();
[global::System.Runtime.InteropServices.DllImport(&quot;example&quot;, EntryPoint=&quot;CSharp_SetMsg&quot;)]
public static extern void SetMsg([global::System.Runtime.InteropServices.MarshalAs(UnmanagedType.LPStr)]String jarg1);
}
</pre>
</div>
<p> Note that the <tt>DllImport</tt> attribute is always generated,
irrespective of any additional attributes specified.</p>
<p> These attributes are associated with the C/C++ parameter type or
return type, which is subtly different to the attribute features and
typemaps covered next. Note that all these different C# attributes can
be combined so that a method has more than one attribute.</p>
<p> The <tt>directorinattributes</tt> and <tt>directoroutattributes</tt>
typemap attribute are attached to the delegates in the director class,
for example, the SwigDelegateBase_0</p>
</li>
<li>
<p> Support for attaching C# attributes to wrapped methods, variables
and enum values. This is done using the <tt>%csattributes</tt> feature,
see <a href="#Customization_features">%feature directives</a>. Note
that C# attributes are attached to proxy classes and enums using the <tt>
csattributes</tt> typemap. For example, imagine we have a custom
attribute class, <tt>ThreadSafeAttribute</tt>, for labelling thread
safety. The following SWIG code shows how to attach this C# attribute
to some methods and the class declaration itself:</p>
<div class="code">
<pre>
%typemap(csattributes) AClass &quot;[ThreadSafe]&quot;
%csattributes AClass::AClass(double d) &quot;[ThreadSafe(false)]&quot;
%csattributes AClass::AMethod() &quot;[ThreadSafe(true)]&quot;
%inline %{
class AClass {
public:
AClass(double a) {}
void AMethod() {}
};
%}
</pre>
</div>
<p> will generate a C# proxy class:</p>
<div class="code">
<pre>
[ThreadSafe]
public class AClass : global::System.IDisposable {
...
[ThreadSafe(false)]
public AClass(double a) ...
[ThreadSafe(true)]
public void AMethod() ...
}
</pre>
</div>
<p> If C# attributes need adding to the <tt>set</tt> or <tt>get</tt>
part of C# properties, when wrapping C/C++ variables, they can be added
using the 'csvarin' and 'csvarout' typemaps respectively. Note that the
type used for the property is specified in the 'cstype' typemap. If the
'out' attribute exists in this typemap, then the type used is from the
'out' attribute.</p>
<p> An example for attaching attributes to the enum and enum values is
shown below.</p>
<div class="code">
<pre>
%typemap(csattributes) Couleur &quot;[global::System.ComponentModel.Description(\&quot;Colours\&quot;)]&quot;
%csattributes Rouge &quot;[global::System.ComponentModel.Description(\&quot;Red\&quot;)]&quot;
%csattributes Vert &quot;[global::System.ComponentModel.Description(\&quot;Green\&quot;)]&quot;
%inline %{
enum Couleur { Rouge, Orange, Vert };
%}
</pre>
</div>
<p> which will result in the following C# enum:</p>
<div class="code">
<pre>
[global::System.ComponentModel.Description(&quot;Colours&quot;)]
public enum Couleur {
[global::System.ComponentModel.Description(&quot;Red&quot;)]
Rouge,
Orange,
[global::System.ComponentModel.Description(&quot;Green&quot;)]
Vert
}
</pre>
</div></li>
<li>
<p> The intermediary classname has <tt>PINVOKE</tt> appended after the
module name instead of <tt>JNI</tt>, for example <tt>modulenamePINVOKE</tt>
.</p>
</li>
<li>
<p> The <tt>%csmethodmodifiers</tt> feature can also be applied to
variables as well as methods. In addition to the default <tt>public</tt>
modifier that SWIG generates when <tt>%csmethodmodifiers</tt> is not
specified, the feature will also replace the <tt>virtual</tt>/<tt>new</tt>
/<tt>override</tt> modifiers that SWIG thinks is appropriate. This
feature is useful for some obscure cases where SWIG might get the <tt>
virtual</tt>/<tt>new</tt>/<tt>override</tt> modifiers incorrect, for
example with multiple inheritance.</p>
</li>
<li> <a name="CSharp_module_directive"></a>
<p> The name of the intermediary class can be changed from its default,
that is, the module name with PINVOKE appended after it. The module
directive attribute <tt>imclassname</tt> is used to achieve this:</p>
<div class="code">
<pre>
%module(imclassname=&quot;name&quot;) modulename
</pre>
</div>
<p> If <tt>name</tt> is the same as <tt>modulename</tt> then the module
class name gets changed from <tt>modulename</tt> to <tt>
modulenameModule</tt>.</p>
</li>
<li> <a name="CSharp_begin"></a>
<p> The <tt>%module</tt> directive supports the <tt>csbegin</tt> option
for adding code to the start of every generated C# file. This is useful
for adding common comments, using statements and/or preprocessor
statements into all generated .cs files. For example, C# 8 nullable
reference types can be enabled via a C# preprocessor directive by
adding <tt>#nullable enable</tt> into C# files as follows:</p>
<div class="code">
<pre>
%module(csbegin=&quot;#nullable enable\n&quot;) mymodule
</pre>
</div>
<p> It might be easier to use a macro for multiple lines of code, for
example:</p>
<div class="code">
<pre>
%define CSBEGIN_CODE
&quot;
/* Copyright statement */
using System.Text;
#nullable enable
&quot;
%enddef
%module(csbegin=CSBEGIN_CODE) mymodule
</pre>
</div></li>
<li> There is no additional 'premature garbage collection prevention
parameter' as the marshalling of the <tt>HandleRef</tt> object takes
care of ensuring a reference to the proxy class is held until the
unmanaged call completed.</li>
</ul>
<p><b> <tt>$dllimport</tt></b>
<br> This is a C# only special variable that can be used in typemaps,
pragmas, features etc. The special variable will get translated into
the value specified by the <tt>-dllimport</tt> commandline option if
specified, otherwise it is equivalent to the<b> $module</b> special
variable.</p>
<p><b> <tt>$imclassname</tt></b>
<br> This special variable expands to the intermediary class name. For
C# this is usually the same as '$modulePINVOKE' ('$moduleJNI' for
Java), unless the imclassname attribute is specified in the <a href="#CSharp_module_directive">
%module directive</a>.</p>
<p><b> <tt>$imfuncname</tt></b>
<br> This special variable expands to the name of the function in the
intermediary class that will be used in $imcall. Like, $imcall, this
special variable is only expanded in the &quot;csout&quot;, &quot;csvarin&quot; and
&quot;csvarout&quot; typemaps.</p>
<p> The directory <tt>Examples/csharp</tt> has a number of simple
examples. Visual Studio .NET 2003 solution and project files are
available for compiling with the Microsoft .NET C# compiler on Windows.
This also works with newer versions of Visual Studio if you allow it to
convert the solution to the latest version. If your SWIG installation
went well on a Unix environment and your C# compiler was detected, you
should be able to type <tt>make</tt> in each example directory. After
SWIG has run and both the C# and C/C++ compilers have finished
building, the examples will be run, by either running <tt>runme.exe</tt>
or by running <tt>mono runme.exe</tt> (Mono C# compiler). Windows users
can also get the examples working using a <a href="http://www.cygwin.com">
Cygwin</a> or <a href="https://osdn.net/projects/mingw/">MinGW</a>
environment for automatic configuration of the example makefiles. Any
one of the C# compilers (Mono or Microsoft) can be detected from within
a Cygwin or MinGW environment if installed in your path.</p>
<h2><a name="CSharp_type_mapping">23.3 Type mapping</a></h2>
<p> The marshalling of the types and typemaps used for marshalling
across the managed/unmanaged layers are discussed in this section. The
interested reader will find the implementation in the csharp.swg file.</p>
<h3><a name="CSharp_primitive_types">23.3.1 Primitive types</a></h3>
<p> Primitive types are marshalled between the unmanaged and managed
layers as blittable types.</p>
<ul>
<li> The first column in the table below shows various C/C++ types that
might be parsed by SWIG.</li>
<li> The second column contains the default type provided by the 'ctype'
typemap, that is, the type used for marshalling on the C side.</li>
<li> The third column shows the default type provided by both the
'imtype' and the 'cstype' typemaps, that is, the equivalent type on the
C# side.</li>
<li> The fourth column shows the size or number of bytes of the
'imtype'/'cstype', which may or may not match the size of the C/C++
type and is discussed next.</li>
</ul>
<table BORDER summary="Default primitive type mappings">
<tr><td><b>C/C++ type</b></td><td><b>ctype</b></td><td><b>imtype/cstype</b>
</td><td><b>Size</b></td></tr>
<tr><td>bool
<br> const bool &amp;</td><td>unsigned int</td><td>bool</td><td>1</td></tr>
<tr><td>char
<br>const char &amp;</td><td>char</td><td>char</td><td>1</td></tr>
<tr><td>signed char
<br>const signed char &amp;</td><td>signed char</td><td>sbyte</td><td>1</td></tr>
<tr><td>unsigned char
<br>const unsigned char &amp;</td><td>unsigned char</td><td>byte</td><td>1</td>
</tr>
<tr><td>short
<br>const short &amp;</td><td>short</td><td>short</td><td>2</td></tr>
<tr><td>unsigned short
<br> const unsigned short &amp;</td><td>unsigned short</td><td>ushort</td><td>
2</td></tr>
<tr><td>int
<br> const int &amp;</td><td>int</td><td>int</td><td>4</td></tr>
<tr><td>unsigned int
<br> const unsigned int &amp;</td><td>unsigned int</td><td>uint</td><td>4</td>
</tr>
<tr><td>long
<br>const long &amp;</td><td>int</td><td>int</td><td>4</td></tr>
<tr><td>unsigned long
<br>const unsigned long &amp;</td><td>unsigned int</td><td>uint</td><td>4</td>
</tr>
<tr><td>long long
<br> const long long &amp;</td><td>long long</td><td>long</td><td>8</td></tr>
<tr><td>unsigned long long
<br>const unsigned long long &amp;</td><td>unsigned long long</td><td>ulong</td><td>
8</td></tr>
<tr><td>float
<br>const float &amp;</td><td>float</td><td>float</td><td>4</td></tr>
<tr><td>double
<br> const double &amp;</td><td>double</td><td>double</td><td>8</td></tr>
<tr><td>size_t
<br> const size_t &amp;</td><td>unsigned int</td><td>uint</td><td>4</td></tr>
</table>
<p> The size in bytes of the C type, 'ctype', should match the C# type,
'imtype' for blitting across the managed/unmanaged layers. They do
match across the common 32-bit and 64-bit operating systems, Unix,
Windows and MacOS, except for the C long/unsigned long and size_t
types. From the table above the default is to handle C long and size_t
as a 32-bit (4 byte) type, so large numbers could be truncated on some
64-bit operating systems. If <tt>SWIGWORDSIZE64</tt> is defined the C
long type is instead handled as a 64-bit (8 byte) type, as per the
table below.</p>
<table BORDER summary="SWIGWORDSIZE64 primitive type mappings">
<tr><td><b>C/C++ type</b></td><td><b>ctype</b></td><td><b>imtype/cstype</b>
</td><td><b>Size</b></td></tr>
<tr><td>long
<br>const long &amp;</td><td>long long</td><td>long</td><td>8</td></tr>
<tr><td>unsigned long
<br>const unsigned long &amp;</td><td>unsigned long long</td><td>ulong</td><td>
8</td></tr>
</table>
<p> However, truncation may then occur when the C long type is actually
32-bits (4 bytes). It's best to avoid using C long for portability
across different operating systems!</p>
<p> If you need to support long on a range of systems where the size of
long varies, then steps must be taken before invoking SWIG to determine
whether or not to define <tt>SWIGWORDSIZE64</tt> when invoking SWIG.</p>
<p> In order to treat the C size_t type as a 64-bit (8 byte) type, apply
the 64-bit typemaps as follows:</p>
<div class="code">
<pre>
%apply unsigned long long { size_t };
%apply const unsigned long long &amp; { const size_t &amp; };
</pre>
</div>
<p> The net effect then changes from the default shown earlier to:</p>
<table BORDER summary="size_t long long mappings">
<tr><td><b>C/C++ type</b></td><td><b>ctype</b></td><td><b>imtype/cstype</b>
</td><td><b>Size</b></td></tr>
<tr><td>size_t
<br>const size_t &amp;</td><td>unsigned long long</td><td>ulong</td><td>8</td>
</tr>
</table>
<p> If you need to support size_t on a range of systems where the size
of size_t varies, then steps must be taken before invoking SWIG to
determine whether or not to apply the typemaps. Conditionally applying
the typemaps using a macro is easily done. For example define <tt>
MY_SIZET_WORDSIZE64</tt> to generate 64-bit (8 byte) handling using the
following:</p>
<div class="code">
<pre>
#if defined(MY_SIZET_WORDSIZE64)
%apply unsigned long long { size_t };
%apply const unsigned long long &amp; { const size_t &amp; };
#endif
</pre>
</div>
<h3><a name="CSharp_other_type_mappings">23.3.2 Other types</a></h3>
<p> The table below shows the equivalent mappings for pointers and
strings. Classes and structs are marshalled using a pointer to the
instance of the object. Note the types in the 'imtype' and 'cstype'
typemaps can be different.</p>
<ul>
<li> The 'imtype' type is used for marshalling across the managed and
unmanaged boundary.</li>
<li> The 'cstype' is the final C# proxy or intermediary class type.</li>
<li> In the table below, the 'imtype' type is used for marshalling from
the managed to the unmanaged layer.</li>
<li> The 'imtype out' type is used for marshalling from the unmanaged to
the managed layer.</li>
</ul>
<table BORDER summary="Other type mappings">
<tr><td><b>C/C++ type</b></td><td><b>ctype</b></td><td><b>imtype</b></td><td>
<b>imtype out</b></td><td><b>cstype</b></td></tr>
<tr><td>char *
<br>char []</td><td>char *</td><td>string</td><td>string</td><td>string</td>
</tr>
<tr><td>void *</td><td>void *</td><td>
System.Runtime.InteropServices.HandleRef</td><td>System.IntPtr</td><td>
SWIGTYPE_p_void</td></tr>
</table>
<h3><a name="CSharp_void_pointers">23.3.3 Void pointers</a></h3>
<p> By default SWIG treats <tt>void *</tt> as any other pointer and
hence marshalls it as a type wrapper class called <tt>SWIGTYPE_p_void</tt>
, as shown in the table in the previous section. If you want to marshall
with the .NET <tt>System.IntPtr</tt> type instead, there is a simple
set of named typemaps called <tt>void *VOID_INT_PTR</tt> that can be
used. The net effect is then:</p>
<table BORDER summary="VOID_INT_PTR type mappings">
<tr><td><b>C/C++ type</b></td><td><b>ctype</b></td><td><b>imtype</b></td><td>
<b>imtype out</b></td><td><b>cstype</b></td></tr>
<tr><td>void *VOID_INT_PTR</td><td>void *</td><td>System.IntPtr</td><td>
System.IntPtr</td><td>System.IntPtr</td></tr>
</table>
<p> This is achieved by applying them like any other named typemaps:</p>
<div class="code">
<pre>
%apply void *VOID_INT_PTR { void * }
void * f(void *v);
</pre>
</div>
<h2><a name="CSharp_arrays">23.4 C# Arrays</a></h2>
<p> There are various ways to pass arrays from C# to C/C++. The default
wrapping treats arrays as pointers and as such simple type wrapper
classes are generated, eg <tt>SWIGTYPE_p_int</tt> when wrapping the C
type <tt>int []</tt> or <tt>int *</tt>. This gives a rather restricted
use of the underlying unmanaged code and the most practical way to use
arrays is to enhance or customise with one of the following three
approaches; namely the SWIG C arrays library, P/Invoke default array
marshalling or pinned arrays.</p>
<h3><a name="CSharp_arrays_swig_library">23.4.1 The SWIG C arrays
library</a></h3>
<p> The C arrays library keeps all the array memory in the unmanaged
layer. The library is available to all language modules and is
documented in the <a href="#Library_carrays">carrays.i library</a>
section. Please refer to this section for details, but for convenience,
the C# usage for the two examples outlined there is shown below.</p>
<p> For the <tt>%array_functions</tt> example, the equivalent usage
would be:</p>
<div class="code">
<pre>
SWIGTYPE_p_double a = example.new_doubleArray(10); // Create an array
for (int i=0; i&lt;10; i++)
example.doubleArray_setitem(a, i, 2*i); // Set a value
example.print_array(a); // Pass to C
example.delete_doubleArray(a); // Destroy array
</pre>
</div>
<p> and for the <tt>%array_class</tt> example, the equivalent usage
would be:</p>
<div class="code">
<pre>
doubleArray c = new doubleArray(10); // Create double[10]
for (int i=0; i&lt;10; i++)
c.setitem(i, 2*i); // Assign values
example.print_array(c.cast()); // Pass to C
</pre>
</div>
<h3><a name="CSharp_arrays_pinvoke_default_array_marshalling">23.4.2
Managed arrays using P/Invoke default array marshalling</a></h3>
<p> In the P/Invoke default marshalling scheme, one needs to designate
whether the invoked function will treat a managed array parameter as
input, output, or both. When the function is invoked, the CLR allocates
a separate chunk of memory as big as the given managed array, which is
automatically released at the end of the function call. If the array
parameter is marked as being input, the content of the managed array is
copied into this buffer when the call is made. Correspondingly, if the
array parameter is marked as being output, the contents of the reserved
buffer are copied back into the managed array after the call returns. A
pointer to this buffer is passed to the native function.</p>
<p> The reason for allocating a separate buffer is to leave the CLR free
to relocate the managed array object during garbage collection. If the
overhead caused by the copying is causing a significant performance
penalty, consider pinning the managed array and passing a direct
reference as described in the next section.</p>
<p> For more information on the subject, see the <a href="https://docs.microsoft.com/en-us/dotnet/framework/interop/default-marshaling-for-arrays">
Default Marshaling for Arrays</a> article on MSDN.</p>
<p> The P/Invoke default marshalling is supported by the <tt>
arrays_csharp.i</tt> library via the INPUT, OUTPUT and INOUT typemaps.
Let's look at some example usage. Consider the following C function:</p>
<div class="code">
<pre>
void myArrayCopy(int *sourceArray, int *targetArray, int nitems);
</pre>
</div>
<p> We can now instruct SWIG to use the default marshalling typemaps by</p>
<div class="code">
<pre>
%include &quot;arrays_csharp.i&quot;
%apply int INPUT[] {int *sourceArray}
%apply int OUTPUT[] {int *targetArray}
</pre>
</div>
<p> As a result, we get the following method in the module class:</p>
<div class="code">
<pre>
public static void myArrayCopy(int[] sourceArray, int[] targetArray, int nitems) {
examplePINVOKE.myArrayCopy(sourceArray, targetArray, nitems);
}
</pre>
</div>
<p> If we look beneath the surface at the corresponding intermediary
class code, we see that SWIG has generated code that uses attributes
(from the System.Runtime.InteropServices namespace) to tell the CLR to
use default marshalling for the arrays:</p>
<div class="code">
<pre>
[global::System.Runtime.InteropServices.DllImport(&quot;example&quot;, EntryPoint=&quot;CSharp_myArrayCopy&quot;)]
public static extern void myArrayCopy([global::System.Runtime.InteropServices.In, global::System.Runtime.InteropServices.MarshalAs(UnmanagedType.LPArray)]int[] jarg1,
[global::System.Runtime.InteropServices.Out, global::System.Runtime.InteropServices.MarshalAs(UnmanagedType.LPArray)]int[] jarg2,
int jarg3);
</pre>
</div>
<p> As an example of passing an inout array (i.e. the target function
will both read from and write to the array), consider this C function
that swaps a given number of elements in the given arrays:</p>
<div class="code">
<pre>
void myArraySwap(int *array1, int *array2, int nitems);
</pre>
</div>
<p> Now, we can instruct SWIG to wrap this by</p>
<div class="code">
<pre>
%include &quot;arrays_csharp.i&quot;
%apply int INOUT[] {int *array1}
%apply int INOUT[] {int *array2}
</pre>
</div>
<p> This results in the module class method</p>
<div class="code">
<pre>
public static void myArraySwap(int[] array1, int[] array2, int nitems) {
examplePINVOKE.myArraySwap(array1, array2, nitems);
}
</pre>
</div>
<p> and intermediary class method</p>
<div class="code">
<pre>
[global::System.Runtime.InteropServices.DllImport(&quot;example&quot;, EntryPoint=&quot;CSharp_myArraySwap&quot;)]
public static extern void myArraySwap([global::System.Runtime.InteropServices.In, global::System.Runtime.InteropServices.Out, global::System.Runtime.InteropServices.MarshalAs(UnmanagedType.LPArray)]int[] jarg1,
[global::System.Runtime.InteropServices.In, global::System.Runtime.InteropServices.Out, global::System.Runtime.InteropServices.MarshalAs(UnmanagedType.LPArray)]int[] jarg2,
int jarg3);
</pre>
</div>
<h3><a name="CSharp_arrays_pinning">23.4.3 Managed arrays using pinning</a>
</h3>
<p> It is also possible to pin a given array in memory (i.e. fix its
location in memory), obtain a direct pointer to it, and then pass this
pointer to the wrapped C/C++ function. This approach involves no
copying, but it makes the work of the garbage collector harder as the
managed array object can not be relocated before the fix on the array
is released. You should avoid fixing arrays in memory in cases where
the control may re-enter the managed side via a callback and/or another
thread may produce enough garbage to trigger garbage collection.</p>
<p> For more information, see the <a href="https://docs.microsoft.com/en-us/dotnet/csharp/language-reference/keywords/fixed-statement">
fixed statement</a> in the C# language reference.</p>
<p> Now let's look at an example using pinning, thus avoiding the CLR
making copies of the arrays passed as parameters. The <tt>
arrays_csharp.i</tt> library file again provides the required support
via the <tt>FIXED</tt> typemaps. Let's use the same function from the
previous section:</p>
<div class="code">
<pre>
void myArrayCopy(int *sourceArray, int *targetArray, int nitems);
</pre>
</div>
<p> We now need to declare the module class method unsafe, as we are
using pointers:</p>
<div class="code">
<pre>
%csmethodmodifiers myArrayCopy &quot;public unsafe&quot;;
</pre>
</div>
<p> Apply the appropriate typemaps to the array parameters:</p>
<div class="code">
<pre>
%include &quot;arrays_csharp.i&quot;
%apply int FIXED[] {int *sourceArray}
%apply int FIXED[] {int *targetArray}
</pre>
</div>
<p> Notice that there is no need for separate in, out or inout typemaps
as is the case when using P/Invoke default marshalling.</p>
<p> As a result, we get the following method in the module class:</p>
<div class="code">
<pre>
public unsafe static void myArrayCopy(int[] sourceArray, int[] targetArray, int nitems) {
fixed ( int *swig_ptrTo_sourceArray = sourceArray ) {
fixed ( int *swig_ptrTo_targetArray = targetArray ) {
{
examplePINVOKE.myArrayCopy((global::System.IntPtr)swig_ptrTo_sourceArray, (global::System.IntPtr)swig_ptrTo_targetArray,
nitems);
}
}
}
}
</pre>
</div>
<p> On the method signature level the only difference to the version
using P/Invoke default marshalling is the &quot;unsafe&quot; quantifier, which is
required because we are handling pointers.</p>
<p> Also the intermediary class method looks a little different from the
default marshalling example - the method is expecting an IntPtr as the
parameter type.</p>
<div class="code">
<pre>
[global::System.Runtime.InteropServices.DllImport(&quot;example&quot;, EntryPoint=&quot;CSharp_myArrayCopy&quot;)]
public static extern void myArrayCopy(global::System.IntPtr jarg1, global::System.IntPtr jarg2, int jarg3);
</pre>
</div>
<h2><a name="CSharp_exceptions">23.5 C# Exceptions</a></h2>
<p> It is possible to throw a C# Exception from C/C++ code. SWIG already
provides the framework for throwing C# exceptions if it is able to
detect that a C++ exception could be thrown. Automatically detecting
that a C++ exception could be thrown is only possible when a C++
exception specification is used, see <a href="#SWIGPlus_exception_specifications">
Exception specifications</a>. The <a href="#Customization_exception">
Exception handling with %exception</a> section details the <tt>
%exception</tt> feature. Customised code for handling exceptions with or
without a C++ exception specification is possible and the details
follow. However anyone wishing to do this should be familiar with the
contents of the sections referred to above.</p>
<p> Unfortunately a C# exception cannot simply be thrown from unmanaged
code for a variety of reasons. Most notably being that throwing a C#
exception results in exceptions being thrown across the C PInvoke
interface and C does not understand exceptions. The design revolves
around a C# exception being constructed and stored as a pending
exception, to be thrown only when the unmanaged code has completed.
Implementing this is a tad involved and there are thus some unusual
typemap constructs. Some practical examples follow and they should be
read in conjunction with the rest of this section.</p>
<p> First some details about the design that must be followed. Each
typemap or feature that generates<b> unmanaged code</b> supports an
attribute called <tt>canthrow</tt>. This is simply a flag which when
set indicates that the code in the typemap/feature has code which might
want to throw a C# exception. The code in the typemap/feature can then
raise a C# exception by calling one of the C functions, <tt>
SWIG_CSharpSetPendingException()</tt> or <tt>
SWIG_CSharpSetPendingExceptionArgument()</tt>. When called, the function
makes a callback into the managed world via a delegate. The callback
creates and stores an exception ready for throwing when the unmanaged
code has finished. The typemap/feature unmanaged code is then expected
to force an immediate return from the unmanaged wrapper function, so
that the pending managed exception can then be thrown. The support code
has been carefully designed to be efficient as well as thread-safe.
However to achieve the goal of efficiency requires some optional code
generation in the<b> managed code</b> typemaps. Code to check for
pending exceptions is generated if and only if the unmanaged code has
code to set a pending exception, that is if the <tt>canthrow</tt>
attribute is set. The optional managed code is generated using the <tt>
excode</tt> typemap attribute and <tt>$excode</tt> special variable in
the relevant managed code typemaps. Simply, if any relevant unmanaged
code has the <tt>canthrow</tt> attribute set, then any occurrences of <tt>
$excode</tt> is replaced with the code in the <tt>excode</tt> attribute.
If the <tt>canthrow</tt> attribute is not set, then any occurrences of <tt>
$excode</tt> are replaced with nothing.</p>
<p> The prototypes for the <tt>SWIG_CSharpSetPendingException()</tt> and
<tt>SWIG_CSharpSetPendingExceptionArgument()</tt> functions are</p>
<div class="code">
<pre>
static void SWIG_CSharpSetPendingException(SWIG_CSharpExceptionCodes code,
const char *msg);
static void SWIG_CSharpSetPendingExceptionArgument(SWIG_CSharpExceptionArgumentCodes code,
const char *msg,
const char *param_name);
</pre>
</div>
<p> The first parameter defines which .NET exceptions can be thrown:</p>
<div class="code">
<pre>
typedef enum {
SWIG_CSharpApplicationException,
SWIG_CSharpArithmeticException,
SWIG_CSharpDivideByZeroException,
SWIG_CSharpIndexOutOfRangeException,
SWIG_CSharpInvalidCastException,
SWIG_CSharpInvalidOperationException,
SWIG_CSharpIOException,
SWIG_CSharpNullReferenceException,
SWIG_CSharpOutOfMemoryException,
SWIG_CSharpOverflowException,
SWIG_CSharpSystemException
} SWIG_CSharpExceptionCodes;
typedef enum {
SWIG_CSharpArgumentException,
SWIG_CSharpArgumentNullException,
SWIG_CSharpArgumentOutOfRangeException,
} SWIG_CSharpExceptionArgumentCodes;
</pre>
</div>
<p> where, for example, <tt>SWIG_CSharpApplicationException</tt>
corresponds to the .NET exception, <tt>ApplicationException</tt>. The <tt>
msg</tt> and <tt>param_name</tt> parameters contain the C# exception
message and parameter name associated with the exception.</p>
<p> The <tt>%exception</tt> feature in C# has the <tt>canthrow</tt>
attribute set. The <tt>%csnothrowexception</tt> feature is like <tt>
%exception</tt>, but it does not have the <tt>canthrow</tt> attribute
set so should only be used when a C# exception is not created.</p>
<h3><a name="CSharp_exception_example_check_typemap">23.5.1 C# exception
example using &quot;check&quot; typemap</a></h3>
<p> Let's say we have the following simple C++ method:</p>
<div class="code">
<pre>
void positivesonly(int number);
</pre>
</div>
<p> and we want to check that the input <tt>number</tt> is always
positive and if not throw a C# <tt>ArgumentOutOfRangeException</tt>.
The &quot;check&quot; typemap is designed for checking input parameters. Below
you will see the <tt>canthrow</tt> attribute is set because the code
contains a call to <tt>SWIG_CSharpSetPendingExceptionArgument()</tt>.
The full example follows:</p>
<div class="code">
<pre>
%module example
%typemap(check, canthrow=1) int number %{
if ($1 &lt; 0) {
SWIG_CSharpSetPendingExceptionArgument(SWIG_CSharpArgumentOutOfRangeException,
&quot;only positive numbers accepted&quot;, &quot;number&quot;);
return $null;
}
// SWIGEXCODE is a macro used by many other csout typemaps
%define SWIGEXCODE
&quot;\n if ($modulePINVOKE.SWIGPendingException.Pending)&quot;
&quot;\n throw $modulePINVOKE.SWIGPendingException.Retrieve();&quot;
%enddef
%typemap(csout, excode=SWIGEXCODE) void {
$imcall;$excode
}
%}
%inline %{
void positivesonly(int number) {
}
%}
</pre>
</div>
<p> When the following C# code is executed:</p>
<div class="code">
<pre>
public class runme {
static void Main() {
example.positivesonly(-1);
}
}
</pre>
</div>
<p> The exception is thrown:</p>
<div class="code">
<pre>
Unhandled Exception: System.ArgumentOutOfRangeException: only positive numbers accepted
Parameter name: number
in &lt;0x00034&gt; example:positivesonly (int)
in &lt;0x0000c&gt; runme:Main ()
</pre>
</div>
<p> Now let's analyse the generated code to gain a fuller understanding
of the typemaps. The generated unmanaged C++ code is:</p>
<div class="code">
<pre>
SWIGEXPORT void SWIGSTDCALL CSharp_positivesonly(int jarg1) {
int arg1 ;
arg1 = (int)jarg1;
if (arg1 &lt; 0) {
SWIG_CSharpSetPendingExceptionArgument(SWIG_CSharpArgumentOutOfRangeException,
&quot;only positive numbers accepted&quot;, &quot;number&quot;);
return ;
}
positivesonly(arg1);
}
</pre>
</div>
<p> This largely comes from the &quot;check&quot; typemap. The managed code in the
module class is:</p>
<div class="code">
<pre>
public class example {
public static void positivesonly(int number) {
examplePINVOKE.positivesonly(number);
if (examplePINVOKE.SWIGPendingException.Pending)
throw examplePINVOKE.SWIGPendingException.Retrieve();
}
}
</pre>
</div>
<p> This comes largely from the &quot;csout&quot; typemap.</p>
<p> The &quot;csout&quot; typemap is the same as the default void &quot;csout&quot; typemap
so is not strictly necessary for the example. However, it is shown to
demonstrate what managed output code typemaps should contain, that is,
a <tt>$excode</tt> special variable and an <tt>excode</tt> attribute.
Also note that <tt>$excode</tt> is expanded into the code held in the <tt>
excode</tt> attribute. The <tt>$imcall</tt> as always expands into <tt>
examplePINVOKE.positivesonly(number)</tt>. The exception support code in
the intermediary class, <tt>examplePINVOKE</tt>, is not shown, but is
contained within the inner classes, <tt>SWIGPendingException</tt> and <tt>
SWIGExceptionHelper</tt> and is always generated. These classes can be
seen in any of the generated wrappers. However, all that is required of
a user is as demonstrated in the &quot;csin&quot; typemap above. That is, is to
check <tt>SWIGPendingException.Pending</tt> and to throw the exception
returned by <tt>SWIGPendingException.Retrieve()</tt>.</p>
<p> If the &quot;check&quot; typemap did not exist, then the following module
class would instead be generated:</p>
<div class="code">
<pre>
public class example {
public static void positivesonly(int number) {
examplePINVOKE.positivesonly(number);
}
}
</pre>
</div>
<p> Here we see the pending exception checking code is omitted. In fact,
the code above would be generated if the <tt>canthrow</tt> attribute
was not in the &quot;check&quot; typemap, such as:</p>
<div class="code">
<pre>
%typemap(check) int number %{
if ($1 &lt; 0) {
SWIG_CSharpSetPendingExceptionArgument(SWIG_CSharpArgumentOutOfRangeException,
&quot;only positive numbers accepted&quot;, &quot;number&quot;);
return $null;
}
%}
</pre>
</div>
<p> Note that if SWIG detects you have used <tt>
SWIG_CSharpSetPendingException()</tt> or <tt>
SWIG_CSharpSetPendingExceptionArgument()</tt> without setting the <tt>
canthrow</tt> attribute you will get a warning message similar to</p>
<div class="code">
<pre>
example.i:21: Warning 845: Unmanaged code contains a call to a SWIG_CSharpSetPendingException
method and C# code does not handle pending exceptions via the canthrow attribute.
</pre>
</div>
<p> Actually it will issue this warning for any function beginning with <tt>
SWIG_CSharpSetPendingException</tt>.</p>
<h3><a name="CSharp_exception_example_percent_exception">23.5.2 C#
exception example using %exception</a></h3>
<p> Let's consider a similar, but more common example that throws a C++
exception from within a wrapped function. We can use <tt>%exception</tt>
as mentioned in <a href="#Customization_exception">Exception handling
with %exception</a>.</p>
<div class="code">
<pre>
%exception negativesonly(int value) %{
try {
$action
} catch (std::out_of_range e) {
SWIG_CSharpSetPendingException(SWIG_CSharpApplicationException, e.what());
return $null;
}
%}
%inline %{
#include &lt;stdexcept&gt;
void negativesonly(int value) {
if (value &gt;= 0)
throw std::out_of_range(&quot;number should be negative&quot;);
}
%}
</pre>
</div>
<p> The generated unmanaged code this time catches the C++ exception and
converts it into a C# <tt>ApplicationException</tt>.</p>
<div class="code">
<pre>
SWIGEXPORT void SWIGSTDCALL CSharp_negativesonly(int jarg1) {
int arg1 ;
arg1 = (int)jarg1;
try {
negativesonly(arg1);
} catch (std::out_of_range e) {
SWIG_CSharpSetPendingException(SWIG_CSharpApplicationException, e.what());
return ;
}
}
</pre>
</div>
<p> The managed code generated does check for the pending exception as
mentioned earlier as the C# version of <tt>%exception</tt> has the <tt>
canthrow</tt> attribute set by default:</p>
<div class="code">
<pre>
public static void negativesonly(int value) {
examplePINVOKE.negativesonly(value);
if (examplePINVOKE.SWIGPendingException.Pending)
throw examplePINVOKE.SWIGPendingException.Retrieve();
}
</pre>
</div>
<h3><a name="CSharp_exception_example_exception_specifications">23.5.3
C# exception example using exception specifications</a></h3>
<p> When C++ exception specifications are used, SWIG is able to detect
that the method might throw an exception. By default SWIG will
automatically generate code to catch the exception and convert it into
a managed <tt>ApplicationException</tt>, as defined by the default
&quot;throws&quot; typemaps. The following example has a user supplied &quot;throws&quot;
typemap which is used whenever an exception specification contains a <tt>
std::out_of_range</tt>, such as the <tt>evensonly</tt> method below.</p>
<div class="code">
<pre>
%typemap(throws, canthrow=1) std::out_of_range {
SWIG_CSharpSetPendingExceptionArgument(SWIG_CSharpArgumentException, $1.what(), NULL);
return $null;
}
%inline %{
#include &lt;stdexcept&gt;
void evensonly(int input) throw (std::out_of_range) {
if (input%2 != 0)
throw std::out_of_range(&quot;number is not even&quot;);
}
%}
</pre>
</div>
<p> Note that the type for the throws typemap is the type in the
exception specification. SWIG generates a try catch block with the
throws typemap code in the catch handler.</p>
<div class="code">
<pre>
SWIGEXPORT void SWIGSTDCALL CSharp_evensonly(int jarg1) {
int arg1 ;
arg1 = (int)jarg1;
try {
evensonly(arg1);
}
catch(std::out_of_range &amp;_e) {
{
SWIG_CSharpSetPendingExceptionArgument(SWIG_CSharpArgumentException, (&amp;_e)-&gt;what(), NULL);
return ;
}
}
}
</pre>
</div>
<p> Multiple catch handlers are generated should there be more than one
exception specifications declared.</p>
<h3><a name="CSharp_custom_application_exception">23.5.4 Custom C#
ApplicationException example</a></h3>
<p> This example involves a user defined exception. The conventional
.NET exception handling approach is to create a custom <tt>
ApplicationException</tt> and throw it in your application. The goal in
this example is to convert the STL <tt>std::out_of_range</tt> exception
into one of these custom .NET exceptions.</p>
<p> The default exception handling is quite easy to use as the <tt>
SWIG_CSharpSetPendingException()</tt> and <tt>
SWIG_CSharpSetPendingExceptionArgument()</tt> methods are provided by
SWIG. However, for a custom C# exception, the boiler plate code that
supports these functions needs replicating. In essence this consists of
some C/C++ code and C# code. The C/C++ code can be generated into the
wrapper file using the <tt>%insert(runtime)</tt> directive and the C#
code can be generated into the intermediary class using the <tt>
imclasscode</tt> pragma as follows:</p>
<div class="code">
<pre>
%insert(runtime) %{
// Code to handle throwing of C# CustomApplicationException from C/C++ code.
// The equivalent delegate to the callback, CSharpExceptionCallback_t, is CustomExceptionDelegate
// and the equivalent customExceptionCallback instance is customDelegate
typedef void (SWIGSTDCALL* CSharpExceptionCallback_t)(const char *);
CSharpExceptionCallback_t customExceptionCallback = NULL;
extern &quot;C&quot; SWIGEXPORT
void SWIGSTDCALL CustomExceptionRegisterCallback(CSharpExceptionCallback_t customCallback) {
customExceptionCallback = customCallback;
}
// Note that SWIG detects any method calls named starting with
// SWIG_CSharpSetPendingException for warning 845
static void SWIG_CSharpSetPendingExceptionCustom(const char *msg) {
customExceptionCallback(msg);
}
%}
%pragma(csharp) imclasscode=%{
class CustomExceptionHelper {
// C# delegate for the C/C++ customExceptionCallback
public delegate void CustomExceptionDelegate(string message);
static CustomExceptionDelegate customDelegate =
new CustomExceptionDelegate(SetPendingCustomException);
[global::System.Runtime.InteropServices.DllImport(&quot;$dllimport&quot;, EntryPoint=&quot;CustomExceptionRegisterCallback&quot;)]
public static extern
void CustomExceptionRegisterCallback(CustomExceptionDelegate customCallback);
static void SetPendingCustomException(string message) {
SWIGPendingException.Set(new CustomApplicationException(message));
}
static CustomExceptionHelper() {
CustomExceptionRegisterCallback(customDelegate);
}
}
static CustomExceptionHelper exceptionHelper = new CustomExceptionHelper();
%}
</pre>
</div>
<p> The method stored in the C# delegate instance, <tt>customDelegate</tt>
is what gets called by the C/C++ callback. However, the equivalent to
the C# delegate, that is the C/C++ callback, needs to be assigned
before any unmanaged code is executed. This is achieved by putting the
initialisation code in the intermediary class. Recall that the
intermediary class contains all the PInvoke methods, so the static
variables in the intermediary class will be initialised before any of
the PInvoke methods in this class are called. The <tt>exceptionHelper</tt>
static variable ensures the C/C++ callback is initialised with the
value in <tt>customDelegate</tt> by calling the <tt>
CustomExceptionRegisterCallback</tt> method in the <tt>
CustomExceptionHelper</tt> static constructor. Once this has been done,
unmanaged code can make callbacks into the managed world as <tt>
customExceptionCallback</tt> will be initialised with a valid
callback/delegate. Any calls to <tt>
SWIG_CSharpSetPendingExceptionCustom()</tt> will make the callback to
create the pending exception in the same way that <tt>
SWIG_CSharpSetPendingException()</tt> and <tt>
SWIG_CSharpSetPendingExceptionArgument()</tt> does. In fact the method
has been similarly named so that SWIG can issue the warning about
missing <tt>canthrow</tt> attributes as discussed earlier. It is an
invaluable warning as it is easy to forget the <tt>canthrow</tt>
attribute when writing typemaps/features.</p>
<p> The <tt>SWIGPendingException</tt> helper class is not shown, but is
generated as an inner class into the intermediary class. It stores the
pending exception in Thread Local Storage so that the exception
handling mechanism is thread safe.</p>
<p> The boiler plate code above must be used in addition to a
handcrafted <tt>CustomApplicationException</tt>:</p>
<div class="code">
<pre>
// Custom C# Exception
class CustomApplicationException : global::System.ApplicationException {
public CustomApplicationException(string message)
: base(message) {
}
}
</pre>
</div>
<p> and the SWIG interface code:</p>
<div class="code">
<pre>
%typemap(throws, canthrow=1) std::out_of_range {
SWIG_CSharpSetPendingExceptionCustom($1.what());
return $null;
}
%inline %{
void oddsonly(int input) throw (std::out_of_range) {
if (input%2 != 1)
throw std::out_of_range(&quot;number is not odd&quot;);
}
%}
</pre>
</div>
<p> The &quot;throws&quot; typemap now simply calls our new <tt>
SWIG_CSharpSetPendingExceptionCustom()</tt> function so that the
exception can be caught, as such:</p>
<div class="code">
<pre>
try {
example.oddsonly(2);
} catch (CustomApplicationException e) {
...
}
</pre>
</div>
<h2><a name="CSharp_directors">23.6 C# Directors</a></h2>
<p> The SWIG directors feature adds extra code to the generated C# proxy
classes that enable these classes to be used in cross-language
polymorphism. Essentially, it enables unmanaged C++ code to call back
into managed code for virtual methods so that a C# class can derive
from a wrapped C++ class.</p>
<p> The following sections provide information on the C# director
implementation and contain most of the information required to use the
C# directors. However, the <a href="#Java_directors">Java directors</a>
section should also be read in order to gain more insight into
directors.</p>
<h3><a name="CSharp_directors_example">23.6.1 Directors example</a></h3>
<p> Imagine we are wrapping a C++ base class, <tt>Base</tt>, from which
we would like to inherit in C#. Such a class is shown below as well as
another class, <tt>Caller</tt>, which calls the virtual method <tt>
UIntMethod</tt> from pure unmanaged C++ code.</p>
<div class="code">
<pre>
// file: example.h
class Base {
public:
virtual ~Base() {}
virtual unsigned int UIntMethod(unsigned int x) {
std::cout &lt;&lt; &quot;Base - UIntMethod(&quot; &lt;&lt; x &lt;&lt; &quot;)&quot; &lt;&lt; std::endl;
return x;
}
virtual void BaseBoolMethod(const Base &amp;b, bool flag) {}
};
class Caller {
public:
Caller(): m_base(0) {}
~Caller() { delBase(); }
void set(Base *b) { delBase(); m_base = b; }
void reset() { m_base = 0; }
unsigned int UIntMethodCall(unsigned int x) { return m_base-&gt;UIntMethod(x); }
private:
Base *m_base;
void delBase() { delete m_base; m_base = 0; }
};
</pre>
</div>
<p> The director feature is turned off by default and the following
simple interface file shows how directors are enabled for the class <tt>
Base</tt>.</p>
<div class="code">
<pre>
/* File : example.i */
%module(directors=&quot;1&quot;) example
%{
#include &quot;example.h&quot;
%}
%feature(&quot;director&quot;) Base;
%include &quot;example.h&quot;
</pre>
</div>
<p> The following is a C# class inheriting from <tt>Base</tt>:</p>
<div class="code">
<pre>
public class CSharpDerived : Base
{
public override uint UIntMethod(uint x)
{
global::System.Console.WriteLine(&quot;CSharpDerived - UIntMethod({0})&quot;, x);
return x;
}
}
</pre>
</div>
<p> The <tt>Caller</tt> class can demonstrate the <tt>UIntMethod</tt>
method being called from unmanaged code using the following C# code:</p>
<div class="targetlang">
<pre>
public class runme
{
static void Main()
{
Caller myCaller = new Caller();
// Test pure C++ class
using (Base myBase = new Base())
{
makeCalls(myCaller, myBase);
}
// Test director / C# derived class
using (Base myBase = new CSharpDerived())
{
makeCalls(myCaller, myBase);
}
}
static void makeCalls(Caller myCaller, Base myBase)
{
myCaller.set(myBase);
myCaller.UIntMethodCall(123);
myCaller.reset();
}
}
</pre>
</div>
<p> If the above is run, the output is then:</p>
<div class="shell">
<pre>
Base - UIntMethod(123)
CSharpDerived - UIntMethod(123)
</pre>
</div>
<h3><a name="CSharp_directors_implementation">23.6.2 Directors
implementation</a></h3>
<p> The previous section demonstrated a simple example where the virtual
<tt>UIntMethod</tt> method was called from C++ code, even when the
overridden method is implemented in C#. The intention of this section
is to gain an insight into how the director feature works. It shows the
generated code for the two virtual methods, <tt>UIntMethod</tt> and <tt>
BaseBoolMethod</tt>, when the director feature is enabled for the <tt>
Base</tt> class.</p>
<p> Below is the generated C# <tt>Base</tt> director class.</p>
<div class="code">
<pre>
public class Base : global::System.IDisposable {
private global::System.Runtime.InteropServices.HandleRef swigCPtr;
protected bool swigCMemOwn;
internal Base(global::System.IntPtr cPtr, bool cMemoryOwn) {
swigCMemOwn = cMemoryOwn;
swigCPtr = new global::System.Runtime.InteropServices.HandleRef(this, cPtr);
}
internal static global::System.Runtime.InteropServices.HandleRef getCPtr(Base obj) {
return (obj == null) ? new global::System.Runtime.InteropServices.HandleRef(null, global::System.IntPtr.Zero) : obj.swigCPtr;
}
~Base() {
Dispose();
}
public virtual void Dispose() {
lock(this) {
if(swigCPtr.Handle != global::System.IntPtr.Zero &amp;&amp; swigCMemOwn) {
swigCMemOwn = false;
examplePINVOKE.delete_Base(swigCPtr);
}
swigCPtr = new global::System.Runtime.InteropServices.HandleRef(null, global::System.IntPtr.Zero);
global::System.GC.SuppressFinalize(this);
}
}
public virtual uint UIntMethod(uint x) {
uint ret = examplePINVOKE.Base_UIntMethod(swigCPtr, x);
return ret;
}
public virtual void BaseBoolMethod(Base b, bool flag) {
examplePINVOKE.Base_BaseBoolMethod(swigCPtr, Base.getCPtr(b), flag);
if (examplePINVOKE.SWIGPendingException.Pending)
throw examplePINVOKE.SWIGPendingException.Retrieve();
}
public Base() : this(examplePINVOKE.new_Base(), true) {
SwigDirectorConnect();
}
private void SwigDirectorConnect() {
if (SwigDerivedClassHasMethod(&quot;UIntMethod&quot;, swigMethodTypes0))
swigDelegate0 = new SwigDelegateBase_0(SwigDirectorMethodUIntMethod);
if (SwigDerivedClassHasMethod(&quot;BaseBoolMethod&quot;, swigMethodTypes1))
swigDelegate1 = new SwigDelegateBase_1(SwigDirectorMethodBaseBoolMethod);
examplePINVOKE.Base_director_connect(swigCPtr, swigDelegate0, swigDelegate1);
}
private bool SwigDerivedClassHasMethod(string methodName, global::System.global::System.Type[] methodTypes) {
System.Reflection.MethodInfo methodInfo = this.GetType().GetMethod(methodName, methodTypes);
bool hasDerivedMethod = methodInfo.DeclaringType.IsSubclassOf(typeof(Base));
return hasDerivedMethod;
}
private uint SwigDirectorMethodUIntMethod(uint x) {
return UIntMethod(x);
}
private void SwigDirectorMethodBaseBoolMethod(global::System.IntPtr b, bool flag) {
BaseBoolMethod(new Base(b, false), flag);
}
public delegate uint SwigDelegateBase_0(uint x);
public delegate void SwigDelegateBase_1(global::System.IntPtr b, bool flag);
private SwigDelegateBase_0 swigDelegate0;
private SwigDelegateBase_1 swigDelegate1;
private static global::System.Type[] swigMethodTypes0 = new global::System.Type[] { typeof(uint) };
private static global::System.Type[] swigMethodTypes1 = new global::System.Type[] { typeof(Base), typeof(bool) };
}
</pre>
</div>
<p> Everything from the <tt>SwigDirectorConnect()</tt> method and below
is code that is only generated when directors are enabled. The design
comprises a C# delegate being initialised for each virtual method on
construction of the class. Let's examine the <tt>BaseBoolMethod</tt>.</p>
<p> In the <tt>Base</tt> constructor a call is made to <tt>
SwigDirectorConnect()</tt> which contains the initialisation code for
all the virtual methods. It uses a support method, <tt>
SwigDerivedClassHasMethod()</tt>, which simply uses reflection to
determine if the named method, BaseBoolMethod, with the list of
required parameter types, exists in a subclass. If it does not exist,
the delegate is not initialised as there is no need for unmanaged code
to call back into managed C# code. However, if there is an overridden
method in any subclass, the delegate is required. It is then
initialised to the <tt>SwigDirectorMethodBaseBoolMethod</tt> which in
turn will call <tt>BaseBoolMethod</tt> if invoked. The delegate is not
initialised to the <tt>BaseBoolMethod</tt> directly as quite often
types will need marshalling from the unmanaged type to the managed type
in which case an intermediary method (<tt>
SwigDirectorMethodBaseBoolMethod</tt>) is required for the marshalling.
In this case, the C# <tt>Base</tt> class needs to be created from the
unmanaged <tt>IntPtr</tt> type.</p>
<p> The last thing that <tt>SwigDirectorConnect()</tt> does is to pass
the delegates to the unmanaged code. It calls the intermediary method <tt>
Base_director_connect()</tt> which is really a call to the C function <tt>
CSharp_Base_director_connect()</tt>. This method simply maps each C#
delegate onto a C function pointer.</p>
<div class="code">
<pre>
SWIGEXPORT void SWIGSTDCALL CSharp_Base_director_connect(void *objarg,
SwigDirector_Base::SWIG_Callback0_t callback0,
SwigDirector_Base::SWIG_Callback1_t callback1) {
Base *obj = (Base *)objarg;
SwigDirector_Base *director = dynamic_cast&lt;SwigDirector_Base *&gt;(obj);
if (director) {
director-&gt;swig_connect_director(callback0, callback1);
}
}
class SwigDirector_Base : public Base, public Swig::Director {
public:
SwigDirector_Base();
virtual unsigned int UIntMethod(unsigned int x);
virtual ~SwigDirector_Base();
virtual void BaseBoolMethod(Base const &amp;b, bool flag);
typedef unsigned int (SWIGSTDCALL* SWIG_Callback0_t)(unsigned int);
typedef void (SWIGSTDCALL* SWIG_Callback1_t)(void *, unsigned int);
void swig_connect_director(SWIG_Callback0_t callbackUIntMethod,
SWIG_Callback1_t callbackBaseBoolMethod);
private:
SWIG_Callback0_t swig_callbackUIntMethod;
SWIG_Callback1_t swig_callbackBaseBoolMethod;
void swig_init_callbacks();
};
void SwigDirector_Base::swig_connect_director(SWIG_Callback0_t callbackUIntMethod,
SWIG_Callback1_t callbackBaseBoolMethod) {
swig_callbackUIntMethod = callbackUIntMethod;
swig_callbackBaseBoolMethod = callbackBaseBoolMethod;
}
</pre>
</div>
<p> Note that for each director class SWIG creates an unmanaged director
class for making the callbacks. For example <tt>Base</tt> has <tt>
SwigDirector_Base</tt> and <tt>SwigDirector_Base</tt> is derived from <tt>
Base</tt>. Should a C# class be derived from <tt>Base</tt>, the
underlying C++ <tt>SwigDirector_Base</tt> is created rather than <tt>
Base</tt>. The <tt>SwigDirector_Base</tt> class then implements all the
virtual methods, redirecting calls up to managed code if the
callback/delegate is non-zero. The implementation of <tt>
SwigDirector_Base::BaseBoolMethod</tt> shows this - the callback is made
by invoking the <tt>swig_callbackBaseBoolMethod</tt> function pointer:</p>
<div class="code">
<pre>
void SwigDirector_Base::BaseBoolMethod(Base const &amp;b, bool flag) {
void * jb = 0 ;
unsigned int jflag ;
if (!swig_callbackBaseBoolMethod) {
Base::BaseBoolMethod(b, flag);
return;
} else {
jb = (Base *) &amp;b;
jflag = flag;
swig_callbackBaseBoolMethod(jb, jflag);
}
}
</pre>
</div>
<p> The delegates from the above example are <tt>public</tt> by default:</p>
<div class="code">
<pre>
public delegate uint SwigDelegateBase_0(uint x);
public delegate void SwigDelegateBase_1(global::System.IntPtr b, bool flag);
</pre>
</div>
<p> These can be changed if desired via the <tt>
csdirectordelegatemodifiers</tt> <a href="#Customization_features">
%feature directive</a>. For example, using <tt>
%feature(&quot;csdirectordelegatemodifiers&quot;) &quot;internal&quot;</tt> before SWIG
parses the Base class will change all the delegates to <tt>internal</tt>
:</p>
<div class="code">
<pre>
internal delegate uint SwigDelegateBase_0(uint x);
internal delegate void SwigDelegateBase_1(global::System.IntPtr b, bool flag);
</pre>
</div>
<h3><a name="CSharp_director_caveats">23.6.3 Director caveats</a></h3>
<p> There is a subtle gotcha with directors. If default parameters are
used, it is recommended to follow a pattern of always calling a single
method in any C# derived class. An example will clarify this and the
reasoning behind the recommendation. Consider the following C++ class
wrapped as a director class:</p>
<div class="code">
<pre>
class Defaults {
public:
virtual ~Defaults();
virtual void DefaultMethod(int a=-100);
};
</pre>
</div>
<p> Recall that C++ methods with default parameters generate overloaded
methods for each defaulted parameter, so a C# derived class can be
created with two <tt>DefaultMethod</tt> override methods:</p>
<div class="code">
<pre>
public class CSharpDefaults : Defaults
{
public override void DefaultMethod()
{
DefaultMethod(-100); // note C++ default value used
}
public override void DefaultMethod(int x)
{
}
}
</pre>
</div>
<p> It may not be clear at first, but should a user intend to call <tt>
CSharpDefaults.DefaultMethod()</tt> from C++, a call is actually made to
<tt>CSharpDefaults.DefaultMethod(int)</tt>. This is because the initial
call is made in C++ and therefore the <tt>DefaultMethod(int)</tt>
method will be called as is expected with C++ calls to methods with
defaults, with the default being set to -100. The callback/delegate
matching this method is of course the overloaded method <tt>
DefaultMethod(int)</tt>. However, a call from C# to <tt>
CSharpDefaults.DefaultMethod()</tt> will of course call this exact
method and in order for behaviour to be consistent with calls from C++,
the implementation should pass the call on to <tt>
CSharpDefaults.DefaultMethod(int)</tt>using the C++ default value, as
shown above.</p>
<h2><a name="CSharp_multiple_modules">23.7 Multiple modules</a></h2>
<p> When using <a href="#Modules">multiple modules</a> it is possible to
compile each SWIG generated wrapper into a different assembly. However,
by default the generated code may not compile if generated classes in
one assembly use generated classes in another assembly. The visibility
of the <tt>getCPtr()</tt> and pointer constructor generated from the <tt>
csbody</tt> typemaps needs changing. The default visibility is <tt>
internal</tt> but it needs to be <tt>public</tt> for access from a
different assembly. Just changing 'internal' to 'public' in the typemap
achieves this. Two macros are available in <tt>csharp.swg</tt> to make
this easier and using them is the preferred approach over simply
copying the typemaps and modifying as this is forward compatible with
any changes in the <tt>csbody</tt> typemap in future versions of SWIG.
The macros are for the proxy and typewrapper classes and can
respectively be used to to make the method and constructor public:</p>
<div class="code">
<pre>
SWIG_CSBODY_PROXY(public, public, SWIGTYPE)
SWIG_CSBODY_TYPEWRAPPER(public, public, public, SWIGTYPE)
</pre>
</div>
<p> Alternatively, instead of exposing these as public, consider using
the <tt>[assembly:InternalsVisibleTo(&quot;Name&quot;)]</tt> attribute available
in the .NET framework when you know which assemblies these can be
exposed to.</p>
<p> Another approach would be to make these public, but also to hide
them from intellisense by using the <tt>
[System.ComponentModel.EditorBrowsable(System.ComponentModel.EditorBrowsableState.Never)]
</tt> attribute if you don't want users to easily stumble upon these so
called 'internal workings' of the wrappers. You can do this like so:</p>
<div class="code">
<pre>
#define PUBLIC_BUT_HIDDEN [System.ComponentModel.EditorBrowsable(System.ComponentModel.EditorBrowsableState.Never)] public
SWIG_CSBODY_PROXY(PUBLIC_BUT_HIDDEN, PUBLIC_BUT_HIDDEN, SWIGTYPE)
SWIG_CSBODY_TYPEWRAPPER(PUBLIC_BUT_HIDDEN, PUBLIC_BUT_HIDDEN, PUBLIC_BUT_HIDDEN, SWIGTYPE)
</pre>
</div>
<h2><a name="CSharp_named_arguments">23.8 C# named and optional
arguments</a></h2>
<p> In C++ you can specify default arguments for functions, methods, and
constructors. C# offers named arguments. This feature, specific to C#,
lets you bind default argument functions with default arguments in C#.
SWIG also allows you to add default arguments to C# functions which
don't have default arguments in C++.</p>
<p> The <tt>cs:defaultargs</tt> feature enables C# named arguments with
C# default values. Using this feature will turn off SWIG's default
handling for default arguments, which would create an override for each
defaulted argument.</p>
<p>For this feature, you first specify the function/method/constructor
you want it to impact. Inside the feature call you specify each
argument you want to override. If you specify none, it will take the
literal text from c++ for each argument and apply that in C#. That
often works fine, but when it doesn't you can supply a string literal
with a C# expression to be used as an override. Or you can supply an
int literal, or a float literal. If you want to give a literal string
you need to include an escaped quote at the start and end of the
literal such as <tt>&quot;\&quot;a string\&quot;&quot;</tt>.</p>
<p> Let's consider an example:</p>
<div class="code">
<pre>
%feature(&quot;cs:defaultargs&quot;) Foo::Foo;
%feature(&quot;cs:defaultargs&quot;, x=0, z=4) Foo::bar;
%feature(&quot;cs:defaultargs&quot;, x=&quot;\&quot;five\&quot;&quot;) Foo::zoo;
%inline %{
class Foo {
public:
Foo(int a, int b=1, int c=2)
{
}
int bar(int x, int y=2, int z=3)
{
return x+y+z;
}
int bat(int x=1, int y=2, int z=3)
{
return x+y+z;
}
int zoo(std::string x=&quot;four&quot;)
{
return (int)x.size();
}
};
%}
</pre>
</div>
<p> The generated C# proxy class contains:</p>
<div class="code">
<pre>
public class Foo : global::System.IDisposable {
...
public Foo(int a, int b=1, int c=2) ...
public int bar(int x=0, int y=2, int z=4) ...
public int bat(int x, int y, int z) ...
public int bat(int x, int y) ...
public int bat(int x) ...
public int bat() ...
public int zoo(string x=&quot;five&quot;) ...
}
</pre>
</div>
<p> Note that:</p>
<ol>
<li> The constructor uses the default arguments exactly as taken from
C++.</li>
<li> The <tt>bar</tt> method uses the default arguments exactly as taken
from C++ apart from <tt>z</tt> whose default value is changed to 4, and
<tt>x</tt> did not have a default value in C++, but does in the
generated C#.</li>
<li> The <tt>bat</tt> method does not use the <tt>cs:defaultargs</tt>
feature and so the default handling of one overloaded C# method per
defaulted C++ argument is generated.</li>
<li> The <tt>zoo</tt> method's string value is overridden by the new
string value from the feature.</li>
</ol>
<p><b> Compatibility Note:</b> SWIG-4.2.0 added support for the <tt>
cs:defaultargs</tt> feature.</p>
<h2><a name="CSharp_typemap_examples">23.9 C# Typemap examples</a></h2>
This section includes a few examples of typemaps. For more examples,
you might look at the files &quot;<tt>csharp.swg</tt>&quot; and &quot;<tt>typemaps.i</tt>
&quot; in the SWIG library.
<h3><a name="CSharp_memory_management_member_variables">23.9.1 Memory
management when returning references to member variables</a></h3>
<p> This example shows how to prevent premature garbage collection of
objects when the underlying C++ class returns a pointer or reference to
a member variable. The example is a direct equivalent to this <a href="#Java_memory_management_objects">
Java equivalent</a>.</p>
<p> Consider the following C++ code:</p>
<div class="code">
<pre>
struct Wheel {
int size;
Wheel(int sz = 0) : size(sz) {}
};
class Bike {
Wheel wheel;
public:
Bike(int val) : wheel(val) {}
Wheel&amp; getWheel() { return wheel; }
};
</pre>
</div>
<p> and the following usage from C# after running the code through SWIG:</p>
<div class="code">
<pre>
Wheel wheel = new Bike(10).getWheel();
global::System.Console.WriteLine(&quot;wheel size: &quot; + wheel.size);
// Simulate a garbage collection
global::System.GC.Collect();
global::System.GC.WaitForPendingFinalizers();
global::System.Console.WriteLine(&quot;wheel size: &quot; + wheel.size);
</pre>
</div>
<p> Don't be surprised that if the resulting output gives strange
results such as...</p>
<div class="shell">
<pre>
wheel size: 10
wheel size: 135019664
</pre>
</div>
<p> What has happened here is the garbage collector has collected the <tt>
Bike</tt> instance as it doesn't think it is needed any more. The proxy
instance, <tt>wheel</tt>, contains a reference to memory that was
deleted when the <tt>Bike</tt> instance was collected. In order to
prevent the garbage collector from collecting the <tt>Bike</tt>
instance a reference to the <tt>Bike</tt> must be added to the <tt>
wheel</tt> instance. You can do this by adding the reference when the <tt>
getWheel()</tt> method is called using the following typemaps.</p>
<div class="code">
<pre>
%typemap(cscode) Wheel %{
// Ensure that the GC doesn't collect any Bike instance set from C#
private Bike bikeReference;
internal void addReference(Bike bike) {
bikeReference = bike;
}
%}
// Add a C# reference to prevent premature garbage collection and resulting use
// of dangling C++ pointer. Intended for methods that return pointers or
// references to a member variable.
%typemap(csout, excode=SWIGEXCODE) Wheel&amp; getWheel {
global::System.IntPtr cPtr = $imcall;$excode
$csclassname ret = null;
if (cPtr != global::System.IntPtr.Zero) {
ret = new $csclassname(cPtr, $owner);
ret.addReference(this);
}
return ret;
}
</pre>
</div>
<p> The code in the first typemap gets added to the <tt>Wheel</tt> proxy
class. The code in the second typemap constitutes the bulk of the code
in the generated <tt>getWheel()</tt> function:</p>
<div class="code">
<pre>
public class Wheel : global::System.IDisposable {
...
// Ensure that the GC doesn't collect any Bike instance set from C#
private Bike bikeReference;
internal void addReference(Bike bike) {
bikeReference = bike;
}
}
public class Bike : global::System.IDisposable {
...
public Wheel getWheel() {
global::System.IntPtr cPtr = examplePINVOKE.Bike_getWheel(swigCPtr);
Wheel ret = null;
if (cPtr != global::System.IntPtr.Zero) {
ret = new Wheel(cPtr, false);
ret.addReference(this);
}
return ret;
}
}
</pre>
</div>
<p> Note the <tt>addReference</tt> call.</p>
<h3><a name="CSharp_memory_management_objects">23.9.2 Memory management
for objects passed to the C++ layer</a></h3>
<p> The example is a direct equivalent to this <a href="#Java_memory_management_objects">
Java equivalent</a>. Managing memory can be tricky when using C++ and C#
proxy classes. The previous example shows one such case and this
example looks at memory management for a class passed to a C++ method
which expects the object to remain in scope after the function has
returned. Consider the following two C++ classes:</p>
<div class="code">
<pre>
struct Element {
int value;
Element(int val) : value(val) {}
};
class Container {
Element* element;
public:
Container() : element(0) {}
void setElement(Element* e) { element = e; }
Element* getElement() { return element; }
};
</pre>
</div>
<p> and usage from C++</p>
<div class="code">
<pre>
Container container;
Element element(20);
container.setElement(&amp;element);
cout &lt;&lt; &quot;element.value: &quot; &lt;&lt; container.getElement()-&gt;value &lt;&lt; endl;
</pre>
</div>
<p> and more or less equivalent usage from C#</p>
<div class="code">
<pre>
Container container = new Container();
Element element = new Element(20);
container.setElement(element);
</pre>
</div>
<p> The C++ code will always print out 20, but the value printed out may
not be this in the C# equivalent code. In order to understand why,
consider a garbage collection occurring...</p>
<div class="code">
<pre>
Container container = new Container();
Element element = new Element(20);
container.setElement(element);
global::System.Console.WriteLine(&quot;element.value: &quot; + container.getElement().value);
// Simulate a garbage collection
global::System.GC.Collect();
global::System.GC.WaitForPendingFinalizers();
global::System.Console.WriteLine(&quot;element.value: &quot; + container.getElement().value);
</pre>
</div>
<p> The temporary element created with <tt>new Element(20)</tt> could
get garbage collected which ultimately means the <tt>container</tt>
variable is holding a dangling pointer, thereby printing out any old
random value instead of the expected value of 20. One solution is to
add in the appropriate references in the C# layer...</p>
<div class="code">
<pre>
public class Container : global::System.IDisposable {
...
// Ensure that the GC doesn't collect any Element set from C#
// as the underlying C++ class stores a shallow copy
private Element elementReference;
public void setElement(Element e) {
examplePINVOKE.Container_setElement(swigCPtr, Element.getCPtr(e));
elementReference = e;
}
}
</pre>
</div>
<p> The following typemaps can be used to generate this code:</p>
<div class="code">
<pre>
%typemap(cscode) Container %{
// Ensure that the GC doesn't collect any Element set from C#
// as the underlying C++ class stores a shallow copy
private Element elementReference;
%}
%typemap(csin,
post=&quot; elementReference = $csinput;&quot;
) Element *e &quot;Element.getCPtr($csinput)&quot;
</pre>
</div>
<p> The 'cscode' typemap simply adds in the specified code into the C#
proxy class. The 'csin' typemap matches the input parameter type and
name for the <tt>setElement</tt> method and the 'post' typemap
attribute allows adding code after the PInvoke call. The 'post' code is
generated into a finally block after the PInvoke call so the resulting
code isn't quite as mentioned earlier, <tt>setElement</tt> is actually:</p>
<div class="code">
<pre>
public void setElement(Element e) {
try {
examplePINVOKE.Container_setElement(swigCPtr, Element.getCPtr(e));
} finally {
elementReference = e;
}
}
</pre>
</div>
<h3><a name="CSharp_date_marshalling">23.9.3 Date marshalling using the
csin typemap and associated attributes</a></h3>
<p> The <a href="#Java_nan_exception_typemap">NaN Exception example</a>
is a simple example of the &quot;javain&quot; typemap and its 'pre' attribute.
This example demonstrates how a C++ date class, say <tt>CDate</tt>, can
be mapped onto the standard .NET date class, <tt>System.DateTime</tt>
by using the 'pre', 'post' and 'pgcppname' attributes of the &quot;csin&quot;
typemap (the C# equivalent to the &quot;javain&quot; typemap). The example is an
equivalent to the <a href="#Java_date_marshalling">Java Date
marshalling example</a>. The idea is that the <tt>System.DateTime</tt>
is used wherever the C++ API uses a <tt>CDate</tt>. Let's assume the
code being wrapped is as follows:</p>
<div class="code">
<pre>
class CDate {
public:
CDate();
CDate(int year, int month, int day);
int getYear();
int getMonth();
int getDay();
...
};
struct Action {
static int doSomething(const CDate &amp;dateIn, CDate &amp;dateOut);
Action(const CDate &amp;date, CDate &amp;dateOut);
};
</pre>
</div>
<p> Note that <tt>dateIn</tt> is const and therefore read only and <tt>
dateOut</tt> is a non-const output type.</p>
<p> First let's look at the code that is generated by default, where the
C# proxy class <tt>CDate</tt> is used in the proxy interface:</p>
<div class="code">
<pre>
public class Action : global::System.IDisposable {
...
public Action(CDate dateIn, CDate dateOut)
: this(examplePINVOKE.new_Action(CDate.getCPtr(dateIn), CDate.getCPtr(dateOut)), true) {
if (examplePINVOKE.SWIGPendingException.Pending)
throw examplePINVOKE.SWIGPendingException.Retrieve();
}
public int doSomething(CDate dateIn, CDate dateOut) {
int ret = examplePINVOKE.Action_doSomething(swigCPtr,
CDate.getCPtr(dateIn),
CDate.getCPtr(dateOut));
if (examplePINVOKE.SWIGPendingException.Pending)
throw examplePINVOKE.SWIGPendingException.Retrieve();
return ret;
}
}
</pre>
</div>
<p> The <tt>CDate &amp;</tt> and <tt>const CDate &amp;</tt> C# code is generated
from the following two default typemaps:</p>
<div class="code">
<pre>
%typemap(cstype) SWIGTYPE &amp; &quot;$csclassname&quot;
%typemap(csin) SWIGTYPE &amp; &quot;$csclassname.getCPtr($csinput)&quot;
</pre>
</div>
<p> where '$csclassname' is translated into the proxy class name, <tt>
CDate</tt> and '$csinput' is translated into the name of the parameter,
eg <tt>dateIn</tt>. From C#, the intention is then to call into a
modified API with something like:</p>
<div class="code">
<pre>
System.DateTime dateIn = new System.DateTime(2011, 4, 13);
System.DateTime dateOut = new System.DateTime();
// Note in calls below, dateIn remains unchanged and dateOut
// is set to a new value by the C++ call
Action action = new Action(dateIn, out dateOut);
dateIn = new System.DateTime(2012, 7, 14);
</pre>
</div>
<p> To achieve this mapping, we need to alter the default code
generation slightly so that at the C# layer, a <tt>System.DateTime</tt>
is converted into a <tt>CDate</tt>. The intermediary layer will still
take a pointer to the underlying <tt>CDate</tt> class. The typemaps to
achieve this are shown below.</p>
<div class="code">
<pre>
%typemap(cstype) const CDate &amp; &quot;System.DateTime&quot;
%typemap(csin,
pre=&quot; CDate temp$csinput = new CDate($csinput.Year, $csinput.Month, $csinput.Day);&quot;
) const CDate &amp;
&quot;$csclassname.getCPtr(temp$csinput)&quot;
%typemap(cstype) CDate &amp; &quot;out System.DateTime&quot;
%typemap(csin,
pre=&quot; CDate temp$csinput = new CDate();&quot;,
post=&quot; $csinput = new System.DateTime(temp$csinput.getYear(),&quot;
&quot; temp$csinput.getMonth(), temp$csinput.getDay(), 0, 0, 0);&quot;,
cshin=&quot;out $csinput&quot;
) CDate &amp;
&quot;$csclassname.getCPtr(temp$csinput)&quot;
</pre>
</div>
<p> The resulting generated proxy code in the <tt>Action</tt> class
follows:</p>
<div class="code">
<pre>
public class Action : global::System.IDisposable {
...
public int doSomething(System.DateTime dateIn, out System.DateTime dateOut) {
CDate tempdateIn = new CDate(dateIn.Year, dateIn.Month, dateIn.Day);
CDate tempdateOut = new CDate();
try {
int ret = examplePINVOKE.Action_doSomething(swigCPtr,
CDate.getCPtr(tempdateIn),
CDate.getCPtr(tempdateOut));
if (examplePINVOKE.SWIGPendingException.Pending)
throw examplePINVOKE.SWIGPendingException.Retrieve();
return ret;
} finally {
dateOut = new System.DateTime(tempdateOut.getYear(),
tempdateOut.getMonth(), tempdateOut.getDay(), 0, 0, 0);
}
}
static private global::System.IntPtr SwigConstructAction(System.DateTime dateIn, out System.DateTime dateOut) {
CDate tempdateIn = new CDate(dateIn.Year, dateIn.Month, dateIn.Day);
CDate tempdateOut = new CDate();
try {
return examplePINVOKE.new_Action(CDate.getCPtr(tempdateIn), CDate.getCPtr(tempdateOut));
} finally {
dateOut = new System.DateTime(tempdateOut.getYear(),
tempdateOut.getMonth(), tempdateOut.getDay(), 0, 0, 0);
}
}
public Action(System.DateTime dateIn, out System.DateTime dateOut)
: this(Action.SwigConstructAction(dateIn, out dateOut), true) {
if (examplePINVOKE.SWIGPendingException.Pending)
throw examplePINVOKE.SWIGPendingException.Retrieve();
}
}
</pre>
</div>
<p> A few things to note:</p>
<ul>
<li> The &quot;cstype&quot; typemap has changed the parameter type to <tt>
System.DateTime</tt> instead of the default generated <tt>CDate</tt>
proxy.</li>
<li> The non-const <tt>CDate &amp;</tt> type is marshalled as a reference
parameter in C# as the date cannot be explicitly set once the object
has been created, so a new object is created instead.</li>
<li> The code in the 'pre' attribute appears before the intermediary
call (<tt>examplePINVOKE.new_Action</tt> / <tt>
examplePINVOKE.Action_doSomething</tt>).</li>
<li> The code in the 'post' attribute appears after the intermediary
call.</li>
<li> A try .. finally block is generated with the intermediary call in
the try block and 'post' code in the finally block. The alternative of
just using a temporary variable for the return value from the
intermediary call and the 'post' code being inserted before the return
statement is not possible given that the intermediary call and method
return comes from a single source (the &quot;csout&quot; typemap).</li>
<li> The temporary variables in the &quot;csin&quot; typemaps are called <tt>
temp$csin</tt>, where &quot;$csin&quot; is replaced with the parameter name.
&quot;$csin&quot; is used to mangle the variable name so that more than one <tt>
CDate &amp;</tt> type can be used as a parameter in a method, otherwise two
or more local variables with the same name would be generated.</li>
<li> The use of the &quot;csin&quot; typemap causes a constructor helper function
(<tt>SwigConstructAction</tt>) to be generated. This allows C# code to
be called before the intermediary call made in the constructor
initialization list.</li>
<li> The 'cshin' attribute is required for the <tt>SwigConstructAction</tt>
constructor helper function so that the 2nd parameter is declared as <tt>
out dateOut</tt> instead of just <tt>dateOut</tt>.</li>
</ul>
<p> So far we have considered the date as an input only and an output
only type. Now let's consider <tt>CDate *</tt> used as an input/output
type. Consider the following C++ function which modifies the date
passed in:</p>
<div class="code">
<pre>
void addYears(CDate *pDate, int years) {
*pDate = CDate(pDate-&gt;getYear() + years, pDate-&gt;getMonth(), pDate-&gt;getDay());
}
</pre>
</div>
<p> If usage of <tt>CDate *</tt> commonly follows this input/output
pattern, usage from C# like the following</p>
<div class="code">
<pre>
System.DateTime christmasEve = new System.DateTime(2000, 12, 24);
example.addYears(ref christmasEve, 10); // christmasEve now contains 2010-12-24
</pre>
</div>
<p> will be possible with the following <tt>CDate *</tt> typemaps</p>
<div class="code">
<pre>
%typemap(cstype, out=&quot;System.DateTime&quot;) CDate * &quot;ref System.DateTime&quot;
%typemap(csin,
pre=&quot; CDate temp$csinput = new CDate($csinput.Year, $csinput.Month, $csinput.Day);&quot;,
post=&quot; $csinput = new System.DateTime(temp$csinput.getYear(),&quot;
&quot; temp$csinput.getMonth(), temp$csinput.getDay(), 0, 0, 0);&quot;,
cshin=&quot;ref $csinput&quot;
) CDate *
&quot;$csclassname.getCPtr(temp$csinput)&quot;
</pre>
</div>
<p> Globals are wrapped by the module class and for a module called
example, the typemaps result in the following code:</p>
<div class="code">
<pre>
public class example {
public static void addYears(ref System.DateTime pDate, int years) {
CDate temppDate = new CDate(pDate.Year, pDate.Month, pDate.Day);
try {
examplePINVOKE.addYears(CDate.getCPtr(temppDate), years);
} finally {
pDate = new System.DateTime(temppDate.getYear(), temppDate.getMonth(), temppDate.getDay(),
0, 0, 0);
}
}
...
}
</pre>
</div>
<p> The following typemap is the same as the previous but demonstrates
how a using block can be used for the temporary variable. The only
change to the previous typemap is the introduction of the 'terminator'
attribute to terminate the <tt>using</tt> block. The <tt>subtractYears</tt>
method is nearly identical to the above <tt>addYears</tt> method.</p>
<div class="code">
<pre>
%typemap(csin,
pre=&quot; using (CDate temp$csinput = new CDate($csinput.Year, $csinput.Month, $csinput.Day)) {&quot;,
post=&quot; $csinput = new System.DateTime(temp$csinput.getYear(),&quot;
&quot; temp$csinput.getMonth(), temp$csinput.getDay(), 0, 0, 0);&quot;,
terminator=&quot; } // terminate temp$csinput using block&quot;,
cshin=&quot;ref $csinput&quot;
) CDate *
&quot;$csclassname.getCPtr(temp$csinput)&quot;
void subtractYears(CDate *pDate, int years) {
*pDate = CDate(pDate-&gt;getYear() - years, pDate-&gt;getMonth(), pDate-&gt;getDay());
}
</pre>
</div>
<p> The resulting generated code shows the termination of the <tt>using</tt>
block:</p>
<div class="code">
<pre>
public class example {
public static void subtractYears(ref System.DateTime pDate, int years) {
using (CDate temppDate = new CDate(pDate.Year, pDate.Month, pDate.Day)) {
try {
examplePINVOKE.subtractYears(CDate.getCPtr(temppDate), years);
} finally {
pDate = new System.DateTime(temppDate.getYear(), temppDate.getMonth(), temppDate.getDay(),
0, 0, 0);
}
} // terminate temppDate using block
}
...
}
</pre>
</div>
<h3><a name="CSharp_date_properties">23.9.4 A date example demonstrating
marshalling of C# properties</a></h3>
<p> The previous section looked at converting a C++ date class to <tt>
System.DateTime</tt> for parameters. This section extends this idea so
that the correct marshalling is obtained when wrapping C++ variables.
Consider the same <tt>CDate</tt> class from the previous section and a
global variable:</p>
<div class="code">
<pre>
CDate ImportantDate = CDate(1999, 12, 31);
</pre>
</div>
<p> The aim is to use <tt>System.DateTime</tt> from C# when accessing
this date as shown in the following usage where the module name is
'example':</p>
<div class="code">
<pre>
example.ImportantDate = new System.DateTime(2000, 11, 22);
System.DateTime importantDate = example.ImportantDate;
Console.WriteLine(&quot;Important date: &quot; + importantDate);
</pre>
</div>
<p> When SWIG wraps a variable that is a class/struct/union, it is
wrapped using a pointer to the type for the reasons given in <a href="#SWIG_structure_data_members">
Structure data members</a>. The typemap type required is thus <tt>CDate
*</tt>. Given that the previous section already designed <tt>CDate *</tt>
typemaps, we'll use those same typemaps plus the 'csvarin' and
'csvarout' typemaps.<div class="code">
<pre>
%typemap(cstype, out=&quot;System.DateTime&quot;) CDate * &quot;ref System.DateTime&quot;
%typemap(csin,
pre=&quot; CDate temp$csinput = new CDate($csinput.Year, $csinput.Month, $csinput.Day);&quot;,
post=&quot; $csinput = new System.DateTime(temp$csinput.getYear(),&quot;
&quot; temp$csinput.getMonth(), temp$csinput.getDay(), 0, 0, 0);&quot;,
cshin=&quot;ref $csinput&quot;
) CDate *
&quot;$csclassname.getCPtr(temp$csinput)&quot;
%typemap(csvarin, excode=SWIGEXCODE2) CDate * %{
/* csvarin typemap code */
set {
CDate temp$csinput = new CDate($csinput.Year, $csinput.Month, $csinput.Day);
$imcall;$excode
} %}
%typemap(csvarout, excode=SWIGEXCODE2) CDate * %{
/* csvarout typemap code */
get {
global::System.IntPtr cPtr = $imcall;
CDate tempDate = (cPtr == global::System.IntPtr.Zero) ? null : new CDate(cPtr, $owner);$excode
return new System.DateTime(tempDate.getYear(), tempDate.getMonth(), tempDate.getDay(),
0, 0, 0);
} %}
</pre>
</div></p>
<p> For a module called example, the typemaps result in the following
code:</p>
<div class="code">
<pre>
public class example {
public static System.DateTime ImportantDate {
/* csvarin typemap code */
set {
CDate tempvalue = new CDate(value.Year, value.Month, value.Day);
examplePINVOKE.ImportantDate_set(CDate.getCPtr(tempvalue));
}
/* csvarout typemap code */
get {
global::System.IntPtr cPtr = examplePINVOKE.ImportantDate_get();
CDate tempDate = (cPtr == global::System.IntPtr.Zero) ? null : new CDate(cPtr, false);
return new System.DateTime(tempDate.getYear(), tempDate.getMonth(), tempDate.getDay(),
0, 0, 0);
}
}
...
}
</pre>
</div>
<p> Some points to note:</p>
<ul>
<li>The property set comes from the 'csvarin' typemap and the property
get comes from the 'csvarout' typemap.</li>
<li>The type used for the property comes from the 'cstype' typemap. This
particular example has the 'out' attribute set in the typemap and as it
is specified, it is used in preference to the type in the typemap body.
This is because the type in the 'out' attribute can never include
modifiers such as 'ref', thereby avoiding code such as <tt>public
static ref System.DateTime ImportantDate { ...</tt>, which would of
course not compile.</li>
<li>The <tt>$excode</tt> special variable expands to nothing as there
are no exception handlers specified in any of the unmanaged code
typemaps (in fact the marshalling was done using the default unmanaged
code typemaps.)</li>
<li>The <tt>$imcall</tt> typemap expands to the appropriate intermediary
method call in the <tt>examplePINVOKE</tt> class.</li>
<li>The <tt>$csinput</tt> special variable in the 'csin' typemap always
expands to <tt>value</tt> for properties. In this case <tt>
$csclassname.getCPtr(temp$csinput)</tt> expands to <tt>
CDate.getCPtr(tempvalue)</tt>.</li>
<li>The 'csin' typemap has 'pre', 'post' and 'cshin' attributes, and
these are all ignored in the property set. The code in these attributes
must instead be replicated within the 'csvarin' typemap. The line
creating the <tt>temp$csinput</tt> variable is such an example; it is
identical to what is in the 'pre' attribute.</li>
</ul>
<h3><a name="CSharp_date_pre_post_directors">23.9.5 Date example
demonstrating the 'pre' and 'post' typemap attributes for directors</a></h3>
<p> The 'pre' and 'post' attributes in the &quot;csdirectorin&quot; typemap act
like the attributes of the same name in the &quot;csin&quot; typemap. For example
if we modify the <a href="#CSharp_date_marshalling">Date marshalling
example</a> like this:</p>
<div class="code">
<pre>
class CDate {
...
void setYear(int);
void setMonth(int);
void setDay(int);
};
struct Action {
virtual void someCallback(CDate &amp;date);
virtual ~Action();
...
};
</pre>
</div>
<p> and declare <tt>%feature (&quot;director&quot;)</tt> for the <tt>Action</tt>
class, we would have to define additional marshalling rules for <tt>
CDate &amp;</tt> parameter. The typemap may look like this:</p>
<div class="code">
<pre>
%typemap(csdirectorin,
pre=&quot;System.DateTime temp$iminput = new System.DateTime();&quot;,
post=&quot;CDate temp2$iminput = new CDate($iminput, false);\n&quot;
&quot;temp2$iminput.setYear(tempdate.Year);\n&quot;
&quot;temp2$iminput.setMonth(tempdate.Month);\n&quot;
&quot;temp2$iminput.setDay(tempdate.Day);&quot;
) CDate &amp;date &quot;out temp$iminput&quot;
</pre>
</div>
<p> The generated proxy class code will then contain the following
wrapper for calling user-overloaded <tt>someCallback()</tt>:</p>
<div class="code">
<pre>
...
private void SwigDirectorMethodsomeCallback(global::System.IntPtr date) {
System.DateTime tempdate = new System.DateTime();
try {
someCallback(out tempdate);
} finally {
// we create a managed wrapper around the existing C reference, just for convenience
CDate temp2date = new CDate(date, false);
temp2date.setYear(tempdate.Year);
temp2date.setMonth(tempdate.Month);
temp2date.setDay(tempdate.Day);
}
}
...
</pre>
</div>
<p> Pay special attention to the memory management issues, using these
attributes.</p>
<h3><a name="CSharp_partial_classes">23.9.6 Turning proxy classes into
partial classes</a></h3>
<p> C# supports the notion of partial classes whereby a class definition
can be split into more than one file. It is possible to turn the
wrapped C++ class into a partial C# class using the <tt>
csclassmodifiers</tt> typemap. Consider a C++ class called <tt>ExtendMe</tt>
:</p>
<div class="code">
<pre>
class ExtendMe {
public:
int Part1() { return 1; }
};
</pre>
</div>
<p> The default C# proxy class generated is:</p>
<div class="code">
<pre>
public class ExtendMe : global::System.IDisposable {
...
public int Part1() {
...
}
}
</pre>
</div>
<p> The default csclassmodifiers typemap shipped with SWIG is</p>
<div class="code">
<pre>
%typemap(csclassmodifiers) SWIGTYPE &quot;public class&quot;
</pre>
</div>
<p> Note that the type used is the special catch all type <tt>SWIGTYPE</tt>
. If instead we use the following typemap to override this for just the <tt>
ExtendMe</tt> class:</p>
<div class="code">
<pre>
%typemap(csclassmodifiers) ExtendMe &quot;public partial class&quot;
</pre>
</div>
<p> The C# proxy class becomes a partial class:</p>
<div class="code">
<pre>
public partial class ExtendMe : global::System.IDisposable {
...
public int Part1() {
...
}
}
</pre>
</div>
<p> You can then of course declare another part of the partial class
elsewhere, for example:</p>
<div class="code">
<pre>
public partial class ExtendMe : global::System.IDisposable {
public int Part2() {
return 2;
}
}
</pre>
</div>
<p> and compile the following code:</p>
<div class="code">
<pre>
ExtendMe em = new ExtendMe();
Console.WriteLine(&quot;part1: {0}&quot;, em.Part1());
Console.WriteLine(&quot;part2: {0}&quot;, em.Part2());
</pre>
</div>
<p> demonstrating that the class contains methods calling both unmanaged
code - <tt>Part1()</tt> and managed code - <tt>Part2()</tt>. The
following example is an alternative approach to adding managed code to
the generated proxy class.</p>
<h3><a name="CSharp_sealed_proxy_class">23.9.7 Turning proxy classes
into sealed classes</a></h3>
<p> The technique in the previous section can be used to make the proxy
class a sealed class. Consider a C++ class <tt>NotABaseClass</tt> that
you don't want to be derived from in C#:</p>
<div class="code">
<pre>
struct NotABaseClass {
NotABaseClass();
~NotABaseClass();
};
</pre>
</div>
<p> The default C# proxy class method generated with Dispose method is:</p>
<div class="code">
<pre>
public class NotABaseClass : global::System.IDisposable {
...
public virtual void Dispose() {
...
}
}
</pre>
</div>
<p> The <tt>csclassmodifiers</tt> typemap can be used to modify the
class modifiers and the <tt>csmethodmodifiers</tt> feature can be used
on the destructor to modify the proxy's <tt>Dispose</tt> method:</p>
<div class="code">
<pre>
%typemap(csclassmodifiers) NotABaseClass &quot;public sealed class&quot;
%csmethodmodifiers NotABaseClass::~NotABaseClass &quot;public /*virtual*/&quot;;
</pre>
</div>
<p> The relevant generated code is thus:</p>
<div class="code">
<pre>
public sealed class NotABaseClass : global::System.IDisposable {
...
public /*virtual*/ void Dispose() {
...
}
}
</pre>
</div>
<p> Any attempt to derive from the <tt>NotABaseClass</tt> in C# will
result in a C# compiler error, for example:</p>
<div class="code">
<pre>
public class Derived : NotABaseClass {
};
</pre>
</div><div class="shell">
<pre>
runme.cs(6,14): error CS0509: `Derived': cannot derive from sealed type `NotABaseClass'
</pre>
</div>
<p> Finally, if you get a warning about use of 'protected' in the
generated base class:</p>
<div class="shell">
<pre>
NotABaseClass.cs(14,18): warning CS0628: `NotABaseClass.swigCMemOwn': new protected member declared in sealed class
</pre>
</div>
<p> Either suppress the warning or modify the generated code by copying
and tweaking the default 'csbody' typemap code in csharp.swg by
modifying swigCMemOwn to not be protected.</p>
<h3><a name="CSharp_extending_proxy_class">23.9.8 Extending proxy
classes with additional C# code</a></h3>
<p> The previous example showed how to use partial classes to add
functionality to a generated C# proxy class. It is also possible to
extend a wrapped struct/class with C/C++ code by using the <a href="#SWIGPlus_class_extension">
%extend directive</a>. A third approach is to add some C# methods into
the generated proxy class with the <tt>cscode</tt> typemap. If we
declare the following typemap before SWIG parses the <tt>ExtendMe</tt>
class used in the previous example</p>
<div class="code">
<pre>
%typemap(cscode) ExtendMe %{
public int Part3() {
return 3;
}
%}
</pre>
</div>
<p> The generated C# proxy class will instead be:</p>
<div class="code">
<pre>
public class ExtendMe : global::System.IDisposable {
...
public int Part3() {
return 3;
}
public int Part1() {
...
}
}
</pre>
</div>
<h3><a name="CSharp_enum_underlying_type">23.9.9 Underlying type for
enums</a></h3>
<p> C# enums use int as the underlying type for each enum item, unless
there is a C++11 enum base specifying the underlying C++ enum type. If
there is a C++ base enum then this is automatically converted to the
equivalent C# integral type. If you wish to change the underlying type
to something else, then use the <tt>csbase</tt> typemap. For example
when your C++ code uses a value larger than int, this is necessary as
the C# compiler will not compile values which are too large to fit into
an int. Here is an example:</p>
<div class="code">
<pre>
%typemap(csbase) BigNumbers &quot;uint&quot;
%inline %{
enum BigNumbers { big=0x80000000, bigger };
%}
</pre>
</div>
<p> The generated enum will then use the given underlying type and
compile correctly:</p>
<div class="targetlang">
<pre>
public enum BigNumbers : uint {
big = 0x80000000,
bigger
}
</pre>
</div>
<p> If a C++11 enum base is specified, such as <tt>unsigned short</tt>
in the following:</p>
<div class="code">
<pre>
%inline %{
enum SmallNumbers : unsigned short { tiny, small=1 };
%}
</pre>
</div>
<p> The underlying type is automatically converted to the C# equivalent,
<tt>ushort</tt>:</p>
<div class="targetlang">
<pre>
public enum SmallNumbers : ushort {
tiny,
small = 1
}
</pre>
</div>
<p> The underlying C# type can still be changed to something else using
the csbase typemap but the <tt>replace</tt> attribute must be set to
avoid an ignored warnings as there are effectively two specified bases,
which of course is not possible. For example:</p>
<div class="code">
<pre>
%typemap(csbase, replace=&quot;1&quot;) SmallNumbers &quot;byte&quot;
%inline %{
enum SmallNumbers : unsigned short { tiny, small=1 };
%}
</pre>
</div>
<p> which generates the desired underlying enum type:</p>
<div class="targetlang">
<pre>
public enum SmallNumbers : byte {
tiny,
small = 1
}
</pre>
</div>
<p> If SWIG cannot find appropriate type information for the C++ enum
base, then a type wrapper class (something like SWIGTYPE_p_xxxx) is
generated for the C# enum base. This will not compile in C#, as a C#
integral base is required. To fix this, make sure the C++ enum base
type information is parsed by SWIG; this is usually a typedef to an
integral type. Note that SWIG uses the cstype typemap to lookup the C#
type for any given C++ type. Using the csbase typemap, as described
above, overrides the default cstype for looking up the C# integral base
type.</p>
<HR NOSHADE>
<h1><a name="D">24 SWIG and D</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#D_introduction">Introduction</a></li>
<li><a href="#D_command_line_invocation">Command line invocation</a></li>
<li><a href="#D_typemaps">Typemaps</a>
<ul>
<li><a href="#D_typemap_name_comparison">C# &lt;-&gt; D name comparison</a></li>
<li><a href="#D_ctype_imtype_dtype">ctype, imtype, dtype</a></li>
<li><a href="#D_in_out_directorin_direcetorout">in, out, directorin,
directorout</a></li>
<li><a href="#D_din_dout_ddirectorin_ddirectorout">din, dout,
ddirectorin, ddirectorout</a></li>
<li><a href="#D_typecheck_typemaps">typecheck typemaps</a></li>
<li><a href="#D_code_injection_typemaps">Code injection typemaps</a></li>
<li><a href="#D_special_variables">Special variable macros</a></li>
</ul>
</li>
<li><a href="#D_other_code_control">Other D code control features</a>
<ul>
<li><a href="#D_module">D begin</a></li>
<li><a href="#D_features">D and %feature</a></li>
</ul>
</li>
<li><a href="#D_pragmas">Pragmas</a></li>
<li><a href="#D_exceptions">D Exceptions</a></li>
<li><a href="#D_directors">D Directors</a></li>
<li><a href="#D_other_features">Other features</a>
<ul>
<li><a href="#D_nspace">Extended namespace support (nspace)</a></li>
<li><a href="#D_native_pointer_support">Native pointer support</a></li>
<li><a href="#D_operator_overloading">Operator overloading</a></li>
<li><a href="#D_test_suite">Running the test-suite</a></li>
</ul>
</li>
<li><a href="#D_typemap_examples">D Typemap examples</a></li>
<li><a href="#D_planned_features">Work in progress and planned features</a>
</li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="D_introduction">24.1 Introduction</a></h2>
<p>From the <a href="https://www.digitalmars.com/d/">D Programming
Language</a> web site:<em> D is a systems programming language. Its
focus is on combining the power and high performance of C and C++ with
the programmer productivity of modern languages like Ruby and Python.
[...] The D language is statically typed and compiles directly to
machine code.</em> As such, it is not very surprising that D is able to
directly <a href="https://www.digitalmars.com/d/1.0/interfaceToC.html">
interface with C libraries</a>. Why would a SWIG module for D be needed
then in the first place?</p>
<p>Well, besides the obvious downside that the C header files have to be
manually converted to D modules for this to work, there is one major
inconvenience with this approach: D code usually is on a higher
abstraction level than C, and many of the features that make D
interesting are simply not available when dealing with C libraries,
requiring you e.g. to manually convert strings between pointers to <tt>
\0</tt>-terminated char arrays and D char arrays, making the algorithms
from the D2 standard library unusable with C arrays and data
structures, and so on.</p>
<p>While these issues can be worked around relatively easy by
hand-coding a thin wrapper layer around the C library in question,
there is another issue where writing wrapper code per hand is not
feasible: C++ libraries. Support for C++ was added in D2 via <tt>
extern(C++)</tt> but this support is still very limited, and a custom
wrapper layer is still required in many cases.</p>
<p>To help addressing these issues, the SWIG C# module has been forked
to support D. Is has evolved quite a lot since then, but there are
still many similarities, so if you do not find what you are looking for
on this page, it might be worth having a look at the chapter on <a href="#CSharp">
C#</a> (and also on <a href="#Java">Java</a>, since the C# module was in
turn forked from it).</p>
<h2><a name="D_command_line_invocation">24.2 Command line invocation</a></h2>
<p>To activate the D module, pass the <tt>-d</tt> option to SWIG at the
command line. The same standard command line options as with any other
language module are available, plus the following D specific ones:</p>
<dl>
<dt><tt>-d2</tt></dt>
<dd>
<p>Prior to SWIG 4.2.0, SWIG generated wrappers for D1/Tango by default
and <tt>-d2</tt> could be used to generate D2/Phobos wrappers instead.
SWIG 4.2.0 dropped support for D1, and D2 wrappers are now produced by
default. This option is still recognised to allow existing build
systems calling SWIG to work, but is now a no-op.</p>
</dd>
<dt><a name="D_splitproxy"></a><tt>-splitproxy</tt></dt>
<dd>
<p>By default, SWIG generates two D modules: the<em> proxy</em> module,
named like the source module (either specified via the <tt>%module</tt>
directive or via the <tt>module</tt> command line option), which
contains all the proxy classes, functions, enums, etc., and the<em>
intermediary</em> module (named like the proxy module, but suffixed
with <tt>_im</tt>), which contains all the <tt>extern(C)</tt> function
declarations and other private parts only used internally by the proxy
module.</p>
<p>If the split proxy mode is enabled by passing this option at the
command line, all proxy classes and enums are emitted to their own D
module instead. The main proxy module only contains free functions and
constants in this case.</p>
</dd>
<dt><tt>-package &lt;pkg&gt;</tt></dt>
<dd>
<p>By default, the proxy D modules and the intermediary D module are
written to the root package. Using this option, you can specify another
target package instead.</p>
</dd>
<dt><tt>-wrapperlibrary &lt;wl&gt;</tt></dt>
<dd>
<p>The code SWIG generates to dynamically load the C/C++ wrapper layer
looks for a library called <tt>$module_wrap</tt> by default. With this
option, you can override the name of the file the wrapper code loads at
runtime (the <tt>lib</tt> prefix and the suffix for shared libraries
are appended automatically, depending on the OS).</p>
<p>This might especially be useful if you want to invoke SWIG several
times on separate modules, but compile the resulting code into a single
shared library.</p>
</dd>
</dl>
<h2><a name="D_typemaps">24.3 Typemaps</a></h2>
<h3><a name="D_typemap_name_comparison">24.3.1 C# &lt;-&gt; D name comparison</a>
</h3>
<p>If you already know the SWIG C# module, you might find the following
name comparison table useful:</p>
<div class="diagram">
<pre>
ctype &lt;-&gt; ctype
imtype &lt;-&gt; imtype
cstype &lt;-&gt; dtype
csin &lt;-&gt; din
csout &lt;-&gt; dout
csdirectorin &lt;-&gt; ddirectorin
csdirectorout &lt;-&gt; ddirectorout
csinterfaces &lt;-&gt; dinterfaces
csinterfaces_derived &lt;-&gt; dinterfaces_derived
csbase &lt;-&gt; dbase
csclassmodifiers &lt;-&gt; dclassmodifiers
cscode &lt;-&gt; dcode
csimports &lt;-&gt; dimports
csbody &lt;-&gt; dbody
csfinalize &lt;-&gt; ddestructor
csdisposing &lt;-&gt; ddispose
csdisposing_derived &lt;-&gt; ddispose_derived
</pre>
</div>
<h3><a name="D_ctype_imtype_dtype">24.3.2 ctype, imtype, dtype</a></h3>
<p>Mapping of types between the C/C++ library, the C/C++ library wrapper
exposing the C functions, the D wrapper module importing these
functions and the D proxy code.</p>
<p>The <tt>ctype</tt> typemap is used to determine the types to use in
the C wrapper functions. The types from the <tt>imtype</tt> typemap are
used in the extern(C) declarations of these functions in the
intermediary D module. The <tt>dtype</tt> typemap contains the D types
used in the D proxy module/class.</p>
<h3><a name="D_in_out_directorin_direcetorout">24.3.3 in, out,
directorin, directorout</a></h3>
<p>Used for converting between the types for C/C++ and D when generating
the code for the wrapper functions (on the C++ side).</p>
<p>The code from the <tt>in</tt> typemap is used to convert arguments to
the C wrapper function to the type used in the wrapped code (<tt>ctype</tt>
-&gt;original C++ type), the <tt>out</tt> typemap is utilized to convert
values from the wrapped code to wrapper function return types (original
C++ type-&gt;<tt>ctype</tt>).</p>
<p>The <tt>directorin</tt> typemap is used to convert parameters to the
type used in the D director callback function, its return value is
processed by <tt>directorout</tt> (see below).</p>
<h3><a name="D_din_dout_ddirectorin_ddirectorout">24.3.4 din, dout,
ddirectorin, ddirectorout</a></h3>
<p>Typemaps for code generation in D proxy and type wrapper classes.</p>
<p><a name="D_din"></a>The <tt>din</tt> typemap is used for converting
function parameter types from the type used in the proxy module or
class to the type used in the intermediary D module (the <a href="#D_dinput">
$dinput</a> macro is replaced). To inject further parameter processing
code before or after the call to the intermediary layer, the <tt>pre</tt>
, <tt>post</tt> and <tt>terminator</tt> attributes can be used (please
refer to the <a href="#CSharp_date_marshalling">C# date marshalling
example</a> for more information on these).</p>
<p><a name="D_dout"></a>The <tt>dout</tt> typemap is used for converting
function return values from the return type used in the intermediary D
module to the type returned by the proxy function. The <tt>$excode</tt>
special variable in <tt>dout</tt> typemaps is replaced by the <tt>
excode</tt> typemap attribute code if the method can throw any
exceptions from unmanaged code, otherwise by nothing (the <a href="#D_imcall">
<tt>$imcall</tt> and <tt>$owner</tt></a> macros are replaced).</p>
<p><a name="D_ddirectorinout"></a>The code from the <tt>ddirectorin</tt>
and <tt>ddirectorout</tt> typemaps is used for conversion in director
callback functions. Arguments are converted to the type used in the
proxy class method they are calling by using the code from <tt>
ddirectorin</tt>, the proxy class method return value is converted to
the type the C++ code expects via the <tt>ddirectorout</tt> typemap
(the <a href="#D_dpcall"><tt>$dcall</tt> and <tt>$winput</tt></a>
macros are replaced).</p>
<p>The full chain of type conversions when a director callback is
invoked looks like this:</p>
<div class="diagram">
<pre>
type CPPClass::method(type a)
&uarr; &darr;
&lt;directorout&gt; &lt;directorin&gt;
&uarr; &darr;
ctype methodCallback(ctype a) C++
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
imtype methodCallback(imtype a) D
&uarr; &darr;
&lt;ddirectorout&gt; &lt;ddirectorin&gt;
&uarr; &darr;
dtype DClass.method(dtype a)</pre>
</div>
<h3><a name="D_typecheck_typemaps">24.3.5 typecheck typemaps</a></h3>
<p>Because, unlike many scripting languages supported by SWIG, D does
not need any dynamic dispatch helper to access an overloaded function,
the purpose of these is merely to issue a warning for overloaded C++
functions that cannot be overloaded in D (as more than one C++ type
maps to a single D type).</p>
<h3><a name="D_code_injection_typemaps">24.3.6 Code injection typemaps</a>
</h3>
<p>These typemaps are used for generating the skeleton of proxy classes
for C++ types.</p>
<p>By overriding <tt>dbase</tt>, <tt>dinterfaces</tt> or <tt>
dinterfaces_derived</tt>, the inheritance chain of the generated proxy
class for a type can be modified. <tt>dclassmodifiers</tt> allows you
to add any custom modifiers around the class keyword.</p>
<p>Using <tt>dcode</tt> and <tt>dimports</tt>, you can specify
additional D code which will be emitted into the class body
respectively the imports section of the D module the class is written
to.</p>
<p><a name="D_class_code_typemaps"></a><tt>dconstructor</tt>, <tt>
ddestructor</tt>, <tt>ddispose</tt> and <tt>ddispose_derived</tt> are
used to generate the class constructor, destructor and <tt>dispose()</tt>
method, respectively. The auxiliary code for handling the pointer to
the C++ object is stored in <tt>dbody</tt> and <tt>dbody_derived</tt>.
You can override them for specific types.</p>
<p> Code can also be injected into the D proxy class using <tt>
%proxycode</tt>.</p>
<h3><a name="D_special_variables">24.3.7 Special variable macros</a></h3>
<p>The standard SWIG special variables are available for use within
typemaps as described in the <a href="#Typemaps">Typemaps documentation</a>
, for example <tt>$1</tt>, <tt>$input</tt>, <tt>$result</tt> etc.</p>
<p>When generating D wrappers, a few additional macros are available:</p>
<dl>
<dt><tt>$dclassname</tt> (C#: <tt>$csclassname</tt>)</dt>
<dd>
<p>This special variable works similar to <a href="#Typemaps_special_variables">
<tt>$n_type</tt></a> in that it returns the name of a type - it expands
to the D proxy class name of the type being wrapped. If the type does
not have an associated proxy class, it expands to the type wrapper
class name, for example, <tt>SWIGTYPE_p_p_SomeCppClass</tt> is
generated when wrapping <tt>SomeCppClass **</tt>.</p>
<p>There are two other variants available, <tt>$&amp;dclassname</tt> and <tt>
$*dclassname</tt>. The former adds a level of indirection, while the
latter removes one. For instance, when wrapping <tt>Foo **</tt>, <tt>
$*dclassname</tt> would be replaced by the proxy class name
corresponding to <tt>Foo *</tt>.</p>
</dd>
<dt><tt>$dclazzname</tt> (C#: <tt>$csclazzname</tt>)</dt>
<dd>
<p>This special variable expands the fully qualified C++ class into the
package name, if used by the <a href="#SWIGPlus_nspace"><tt>nspace</tt>
feature</a>, and the proxy class name, mangled for use as a function
name. For example, <tt>Namespace1::Namespace2::Klass</tt> is expanded
into <tt>Namespace1_Namespace2_Klass_</tt>.</p>
<p>This special variable might be useful for calling certain functions
in the wrapper layer (e.g. upcast wrappers) which are mangled like
this.</p>
</dd>
<dt><tt>$null</tt></dt>
<dd>
<p>In code inserted into the generated C/C++ wrapper functions, this
variable is replaced by either <tt>0</tt> or nothing at all, depending
on whether the function has a return value or not. It can be used to
bail out early e.g. in case of errors (<tt>return $null;</tt>).</p>
</dd>
<dt><a name="D_dinput"></a><tt>$dinput</tt> (C#: <tt>$csinput</tt>)</dt>
<dd>
<p>This variable is used in <tt><a href="#D_din">din</a></tt> typemaps
and is replaced by the expression which is to be passed to C/C++.</p>
<p>For example, this input</p>
<div class="code">
<pre>
%typemap(din) SomeClass * &quot;SomeClass.getCPointer($dinput)&quot;
%inline %{
class SomeClass {};
void foo(SomeClass *arg);
%}</pre>
</div>
<p>leads to the following D proxy code being generated:</p>
<div class="targetlang">
<pre>
void foo(SomeClass arg) {
example_im.foo(SomeClass.getCPointer(arg));
}</pre>
</div></dd>
<dt><a name="D_imcall"></a><tt>$imcall</tt> and <tt>$owner</tt> (C#: <tt>
$imcall</tt>)</dt>
<dd>
<p>These variables are used in <tt><a href="#D_dout">dout</a></tt>
typemaps. <tt>$imcall</tt> contains the call to the intermediary module
which provides the value to be used, and <tt>$owner</tt> signals if the
caller is responsible for managing the object lifetime (that is, if the
called method is a constructor or has been marked via <tt>%newobject</tt>
).</p>
<p>Consider the following example:</p>
<div class="code">
<pre>
%typemap(dout) SomeClass * {
return new SomeClass($imcall, $owner);
}
%inline %{
class SomeClass;
SomeClass *foo();
%newobject bar();
SomeClass *bar();
%}
</pre>
</div>
<p>The code generated for <tt>foo()</tt> and <tt>bar()</tt> looks like
this:</p>
<div class="targetlang">
<pre>
SomeClass foo() {
return new SomeClass(example_im.foo(), false);
}
SomeClass bar() {
return new SomeClass(example_im.bar(), true);
}
</pre>
</div></dd>
<dt><tt>$dcall</tt> and <tt>$winput</tt> (C#: <tt>$cscall</tt>, <tt>
$iminput</tt>)</dt>
<dd><a name="D_dpcall"></a>
<p>These variables are used in the director-specific typemaps <a href="#D_ddirectorinout">
<tt>ddirectorin</tt></a> and <a href="#D_ddirectorinout"><tt>
ddirectorout</tt></a>. They are more or less the reverse of the <tt>
$imcall</tt> and <tt>$dinput</tt> macros: <tt>$dcall</tt> contains the
invocation of the D proxy method of which the return value is to be
passed back to C++, <tt>$winput</tt> contains the parameter value from
C++.</p>
</dd>
<dt><tt>$excode</tt></dt>
<dd>
<p>This variable is used in <tt>dout</tt> and <tt>dconstructor</tt>
typemaps and is filled with the contents of the <tt>excode</tt> typemap
attribute if an exception could be thrown from the C++ side. See the <a href="#CSharp_exceptions">
C# documentation</a> for details.</p>
</dd>
<dt><tt>$dbaseclass</tt></dt>
<dd>
<p>Currently for internal use only, it contains the D name of the C++
base class (if any) inside proxy classes.</p>
</dd>
<dt><tt>$directorconnect</tt></dt>
<dd>
<p>This macro is only valid inside the <tt><a href="#D_class_code_typemaps">
dconstructor</a></tt> typemap and contains the value of the <tt>
dconstructor</tt> typemap attribute if the currently wrapped class has
directors enabled.</p>
<p>This is how the default <tt>dconstructor</tt> typemap looks like (you
usually do not want to specify a custom one):</p>
<div class="code">
<pre>
%typemap(dconstructor, excode=SWIGEXCODE,
directorconnect=&quot;\n swigDirectorConnect();&quot;) SWIGTYPE {
this($imcall, true);$excode$directorconnect
}
</pre>
</div></dd>
<dt><tt>$imfuncname</tt></dt>
<dd>
<p> This special variable expands to the name of the function in the
intermediary class that will be used in $imcall. Like, $imcall, this
special variable is only expanded in the &quot;dout&quot; typemap.</p>
</dd>
<dt><a name="D_importtype"></a><tt>$importtype(SomeDType)</tt></dt>
<dd>
<p>This macro is used in the <tt>dimports</tt> typemap if a dependency
on another D type generated by SWIG is added by a custom typemap.</p>
<p>Consider the following code snippet:</p>
<div class="code">
<pre>
%typemap(dinterfaces) SomeClass &quot;AnInterface, AnotherInterface&quot;
</pre>
</div>
<p>This causes SWIG to add <tt>AnInterface</tt> and <tt>AnotherInterface</tt>
to the base class list of <tt>SomeClass</tt>:</p>
<div class="targetlang">
<pre>
class SomeClass : AnInterface, AnotherInterface {
...
}
</pre>
</div>
<p>For this to work, <tt>AnInterface</tt> and <tt>AnotherInterface</tt>
have to be in scope. If SWIG is not in split proxy mode, this is
already the case, but if it is, they have to be added to the import
list via the <tt>dimports</tt> typemap. Additionally, the import
statement depends on the package SWIG is configured to emit the modules
to.</p>
<p>The <tt>$importtype</tt> macro helps you to elegantly solve this
problem:</p>
<div class="code">
<pre>
%typemap(dimports) RemoteMpe %{
$importtype(AnInterface)
$importtype(AnotherInterface)
%}
</pre>
</div>
<p>If SWIG is in split proxy mode, it expands to an <tt>import</tt>
statement for the specified type, to nothing if not.</p>
</dd>
<dt><tt>$module</tt></dt>
<dd>
<p>Expands to the name of the main proxy D module.</p>
</dd>
<dt><tt>$imdmodule</tt></dt>
<dd>
<p>Contains the fully qualified name of the intermediary D module.</p>
</dd>
</dl>
<h2><a name="D_other_code_control">24.4 Other D code control features</a>
</h2>
<h3><a name="D_module">24.4.1 D begin</a></h3>
<p> It is possible to add a common comment at the start of every
generated D file. The <tt>%module</tt> directive supports the <tt>
dbegin</tt> option for this. The provided text is generated at the very
beginning of each generated D file. As it is generated before the D
module statement, is only really useful for adding in a common comment
into all generated D files. For example, copyright text for each file:</p>
<div class="code">
<pre>
%module(dbegin=&quot;/* Common comment. Copyright (C) 2000 Mr Nobody. */\n&quot;) nobodymodule
</pre>
</div>
<h3><a name="D_features">24.4.2 D and %feature</a></h3>
<p>The D module defines a number of directives which modify the <a href="#Customization_features">
SWIG features</a> set globally or for a specific declaration:</p>
<dl>
<dt><tt>%dmanifestconst</tt> and <tt>%dconstvalue(value)</tt></dt>
<dd>
<p>Out of the box, SWIG generates accessor methods for C <tt>#defines</tt>
and C++ constants. The <tt>%dmanifestconst</tt> directive enables
wrapping these constants as D manifest constants (<tt>enum</tt> in D2).</p>
<p>For this to work, the C/C++ code for the constant value must directly
compile as D code, though. If this is not the case, you can manually
override the expression written to the D proxy module using the <tt>
%dconstvalue</tt> directive, passing the new value as parameter.</p>
<p>For <tt>enum</tt>s, again <tt>%dconstvalue</tt> can be used to
override the value of an enum item if the initializer should not
compile in D.</p>
</dd>
<dt><tt>%dmethodmodifiers</tt></dt>
<dd>
<p>This directive can be used to override the modifiers for a proxy
function. For instance, you could make a <tt>public</tt> C++ member
function <tt>private</tt> in D like this:</p>
<div class="code">
<pre>
%dmethodmodifiers A::foo &quot;private&quot;;
%inline %{
struct A {
void foo();
};
%}
</pre>
</div></dd>
</dl>
<h2><a name="D_pragmas">24.5 Pragmas</a></h2>
<p>There are a few SWIG pragmas specific to the D module, which you can
use to influence the D code SWIG generates:</p>
<dl>
<dt><tt>%pragma(d) imdmodulecode</tt></dt>
<dd>
<p>The passed text (D code) is copied verbatim to the intermediary D
module. For example, it can be (and is, internally) used to emit
additional private helper code for the use by proxy typemaps.</p>
</dd>
<dt><tt>%pragma(d) imdmoduleimports</tt></dt>
<dd>
<p>Additional code to be emitted to the imports section of the
intermediary D module (the <a href="#D_importtype">$importtype</a>
macro can be used here). You probably want to use this in conjunction
with the <tt>imdmodulecode</tt> pragma.</p>
</dd>
<dt><tt>%pragma(d) proxydmodulecode</tt></dt>
<dd>
<p>Just like <tt>proxydmodulecode</tt>, the argument is copied to the
proxy D module (if SWIG is in <a href="#D_splitproxy">split proxy mode</a>
and/or the <tt>nspace</tt> feature is used, it is emitted to the main
proxy D module only).</p>
</dd>
<dt><tt>%pragma(d) globalproxyimports</tt></dt>
<dd>
<p>The D module currently does not support specifying dependencies on
external modules (e.g. from the standard library) for the D typemaps.
To add the import statements to the proxy modules (resp. to<em> all</em>
proxy modules if in split proxy mode), you can use the <tt>
globalproxyimports</tt> directive.</p>
<p>For example:</p>
<div class="code">
<pre>
%typemap(din) char[] &quot;($dinput ? tango.stdc.stringz.toStringz($dinput) : null)&quot;
%pragma(d) globalproxyimports = &quot;static import tango.stdc.stringz;&quot;;
</pre>
</div></dd>
<dt><tt>%pragma(d) wrapperloadercode</tt></dt>
<dd>
<p>The D code for loading the wrapper library (it is copied to the
intermediary D module). The <tt>$wrapperloaderbindcode</tt> variable is
replaced by the list of commands for binding the functions from the
wrapper library to the symbols in the intermediary D module.</p>
<p>Each time this pragma is specified, the previous value is
overwritten.</p>
</dd>
<dt><tt>%pragma(d) wrapperloaderbindcommand</tt></dt>
<dd>
<p>The D command to use for binding the wrapper functions from the C/C++
library to the symbols in the intermediary D module. The <tt>$function</tt>
variable contains the name of the D function in the wrap module, the <tt>
$symbol</tt> variable is replaced by the name of the symbol in the
library.</p>
<p>Each time this pragma is specified, the previous value is
overwritten.</p>
</dd>
</dl>
<h2><a name="D_exceptions">24.6 D Exceptions</a></h2>
<p>Out of the box, C++ exceptions are fundamentally incompatible to
their equivalent in the D world and cannot simply be propagated to a
calling D method. There is, however, an easy way to solve this problem:
Just catch the exception in the C/C++ wrapper layer, pass the contents
to D, and make the wrapper code rethrow the exception in the D world.</p>
<p>The implementation details of this are a bit crude, but the SWIG D
module automatically takes care of this, as long as it is able to
detect that an exception could potentially be thrown (e.g. because the
C++ method has a <tt>throw(...)</tt> exception specification).</p>
<p>As this feature is implemented in exactly the same way it is for C#,
please see the <a href="#CSharp_exceptions">C# documentation</a> for a
more detailed explanation.</p>
<h2><a name="D_directors">24.7 D Directors</a></h2>
<p>When the directors feature is activated, SWIG generates extra code on
both the C++ and the D side to enable cross-language polymorphism.
Essentially, this means that if you subclass a proxy class in D, C++
code can access any overridden virtual methods just as if you created a
derived class in C++.</p>
<p>There is no D specific documentation yet, but the way the feature is
implemented is very similar to how it is done in <a href="#Java_directors">
Java</a> and <a href="#CSharp_directors">C#</a>.</p>
<h2><a name="D_other_features">24.8 Other features</a></h2>
<h3><a name="D_nspace">24.8.1 Extended namespace support (nspace)</a></h3>
<p>By default, SWIG flattens all C++ namespaces into a flattened D
module hierarchy, but as for Java and C#, the <a href="#SWIGPlus_nspace">
<tt>nspace</tt></a> feature is supported for D. If the feature is
active, C++ namespaces are mapped to D packages/modules. This includes
the <tt>nspace</tt> feature flag for mirroring the C++ namespaces into
D namespaces as a scoped D module/package name. It also includes <tt>
%nspacemove</tt> for transforming a C++ namespace into a completely
different scoped D module/package name. Note, however, that like for
the other languages,<em> free</em> variables and functions are not
supported yet; currently, they are all available in the main proxy D
module.</p>
<h3><a name="D_native_pointer_support">24.8.2 Native pointer support</a></h3>
<p>Contrary to many of the scripting languages supported by SWIG, D
fully supports C-style pointers. The D module thus includes a custom
mechanism to wrap C pointers directly as D pointers where applicable,
that is, if the type that is pointed to is represented the same in C
and D (on the bit-level), dubbed a<em> primitive type</em> below.</p>
<p>Central to this custom pointer handling scheme are two typemap
attributes: the <tt>cprimitive</tt> attribute on the <tt>dtype</tt>
typemap and the <tt>nativepointer</tt> attribute on all the typemaps
which influence the D side of the code (<tt>dtype</tt>, <tt>din</tt>, <tt>
dout</tt>, ...). When a D typemap is looked up, the following happens
behind the scenes:</p>
<p>First, the matching typemap is determined by the usual typemap lookup
rules. Then, it is checked if the result has the <tt>nativepointer</tt>
attribute set. If it is present, it means that its value should replace
the typemap value<em> if and only if</em> the actual type the typemap
is looked up for is a primitive type, a pointer to a primitive type
(through an arbitrary level of indirections), or a function pointer
with only primitive types in its signature.</p>
<p>To determine if a type should be considered primitive, the <tt>
cprimitive</tt> attribute on its <tt>dtype</tt> attribute is used. For
example, the <tt>dtype</tt> typemap for <tt>float</tt> has <tt>
cprimitive=&quot;1&quot;</tt>, so the code from the <tt>nativepointer</tt>
attribute is taken into account e.g. for <tt>float **</tt> or the
function pointer <tt>float (*)(float *)</tt>.</p>
<h3><a name="D_operator_overloading">24.8.3 Operator overloading</a></h3>
<p>The D module comes with basic operator overloading support. There
are, however, a few limitations arising from conceptual differences
between C++ and D:</p>
<p>The first key difference is that C++ supports free functions as
operators (along with argument-dependent lookup), while D requires
operators to be member functions of the class they are operating on.
SWIG can only automatically generate wrapping code for member function
operators; if you want to use operators defined as free functions in D,
you need to handle them manually.</p>
<p>Another set of differences between C++ and D concerns individual
operators. For example, there are quite a few operators which are
overloadable in C++, but not in D, for example <tt>&amp;&amp;</tt> and <tt>||</tt>
, but also <tt>!</tt>, and postfix increment/decrement operators (see
the <a href="https://www.digitalmars.com/d/2.0/operatoroverloading.html">
D documentation</a> for details).</p>
<p>There are also some cases where the operators can be translated to D,
but the differences in the implementation details are big enough that a
rather involved scheme would be required for automatic wrapping them,
which has not been implemented yet. This affects, for example, the
array subscript operator, <tt>[]</tt>, in combination with assignments
- while <tt>operator []</tt> in C++ simply returns a reference which is
then written to, D resorts to a separate <tt>opIndexAssign</tt> method
-, or implicit casting (which was introduced in D2 via <tt>alias this</tt>
). Despite the lack of automatic support, manually handling these cases
should be perfectly possible.</p>
<h3><a name="D_test_suite">24.8.4 Running the test-suite</a></h3>
<p>As with any other language, the SWIG test-suite can be built for D
using the <tt>*-d-test-suite</tt> targets of the top-level Makefile.</p>
<p>Note: If you want to use GDC on Linux or another platform which
requires you to link <tt>libdl</tt> for dynamically loading the shared
library, you might have to add <tt>-ldl</tt> manually to the <tt>
d_compile</tt> target in <tt>Examples/Makefile</tt>, because GDC does
not currently honor the <tt>pragma(lib, ...)</tt> statement.</p>
<h2><a name="D_typemap_examples">24.9 D Typemap examples</a></h2>
<p>There are no D-specific typemap examples yet. However, with the above
<a href="#D_typemap_name_comparison">name comparison table</a>, you
should be able to get an idea what can be done by looking at the <a href="#CSharp_typemap_examples">
corresponding C# section</a>.</p>
<h2><a name="D_planned_features">24.10 Work in progress and planned
features</a></h2>
<p>There are a couple of features which are not implemented yet, but
would be very useful and might be added in the near future:</p>
<ul>
<li><em>Static linking:</em> Currently, the C wrapper code is compiled
into a dynamic library, out of which the symbol addresses are looked up
at runtime by the D part. If statically linking the different languages
into one binary was supported, a tool-chain capable of performing IPO
at link time could inline the wrapping code, effectively reducing the
overhead for simple calls to zero.</li>
<li><em>C array handling:</em> Many data structures in some C/C++
libraries contain array containing of a pointer to the first element
and the element count. Currently, one must manually writing wrapper
code to be able to access these from D. It should be possible to add a
set of SWIG macros to semi-automatically generate conversion code.</li>
</ul>
<p>Some generated code might also be a bit rough around the edges,
particularly in the following areas:</p>
<ul>
<li><em>Memory management:</em> Although the currently generated wrapper
code works fine with regard to the GC for the test-suite, there might
be issues coming up in real-world multi-threaded usage.</li>
<li><em>D2 support</em>: Originally, the module has been developed for
the use with D1, D2/Phobos support has been added in later. The basic
features should work equally well for both, but there<em> could</em> be
issues concerning const-correctness etc.</li>
</ul>
<HR NOSHADE>
<h1><a name="Go">25 SWIG and Go</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Go_overview">Overview</a></li>
<li><a href="#Go_examples">Examples</a></li>
<li><a href="#Go_running_swig">Running SWIG with Go</a>
<ul>
<li><a href="#Go_commandline">Go-specific Commandline Options</a></li>
<li><a href="#Go_outputs">Generated Wrapper Files</a></li>
</ul>
</li>
<li><a href="#Go_basic_tour">A tour of basic C/C++ wrapping</a>
<ul>
<li><a href="#Go_package">Go Package Name</a></li>
<li><a href="#Go_names">Go Names</a></li>
<li><a href="#Go_constants">Go Constants</a></li>
<li><a href="#Go_enumerations">Go Enumerations</a></li>
<li><a href="#Go_classes">Go Classes</a>
<ul>
<li><a href="#Go_class_memory">Go Class Memory Management</a></li>
<li><a href="#Go_class_inheritance">Go Class Inheritance</a></li>
</ul>
</li>
<li><a href="#Go_templates">Go Templates</a></li>
<li><a href="#Go_threads">Go and C/C++ Threads</a></li>
<li><a href="#Go_exceptions">Go and C++ Exceptions</a></li>
<li><a href="#Go_director_classes">Go Director Classes</a>
<ul>
<li><a href="#Go_director_example_cpp_code">Example C++ code</a></li>
<li><a href="#Go_director_enable">Enable director feature</a></li>
<li><a href="#Go_director_ctor_dtor">Constructor and destructor</a></li>
<li><a href="#Go_director_overriding">Override virtual methods</a></li>
<li><a href="#Go_director_base_methods">Call base methods</a></li>
<li><a href="#Go_director_subclass">Subclass via embedding</a></li>
<li><a href="#Go_director_finalizer">Memory management with
runtime.SetFinalizer</a></li>
<li><a href="#Go_director_foobargo_class">Complete FooBarGo example
class</a></li>
</ul>
</li>
<li><a href="#Go_primitive_type_mappings">Default Go primitive type
mappings</a></li>
<li><a href="#Go_output_arguments">Output arguments</a></li>
<li><a href="#Go_adding_additional_code">Adding additional go code</a></li>
<li><a href="#Go_typemaps">Go typemaps</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> This chapter describes SWIG's support of Go. For more information on
the Go programming language see <a href="https://golang.org/">
golang.org</a>.</p>
<h2><a name="Go_overview">25.1 Overview</a></h2>
<p> Go does not support direct calling of functions written in C/C++.
The <a href="https://golang.org/cmd/cgo/">cgo program</a> may be used
to generate wrappers to call C code from Go, but there is no convenient
way to call C++ code. SWIG fills this gap.</p>
<p> There are (at least) two different Go compilers. The first is the gc
compiler of the <a href="https://golang.org/doc/install">Go
distribution</a>, normally invoked via the <a href="https://golang.org/cmd/go/">
go tool</a>. SWIG supports the gc compiler version 1.2 or later. The
second Go compiler is the <a href="https://golang.org/doc/install/gccgo">
gccgo compiler</a>, which is a frontend to the GCC compiler suite. The
interface to C/C++ code is completely different for the two Go
compilers. SWIG supports both Go compilers, selected by the <tt>-gccgo</tt>
command line option.</p>
<p> Go is a type-safe compiled language and the wrapper code generated
by SWIG is type-safe as well. In case of type issues the build will
fail and hence SWIG's <a href="#Modules_nn2">runtime library</a> and <a href="#Typemaps_runtime_type_checker">
runtime type checking</a> are not used.</p>
<h2><a name="Go_examples">25.2 Examples</a></h2>
<p> Working examples can be found in the <a href="https://github.com/swig/swig/tree/master/Examples/go">
SWIG source tree</a> .</p>
<p> Please note that the examples in the SWIG source tree use makefiles
with the .i SWIG interface file extension for backwards compatibility
with Go 1.</p>
<h2><a name="Go_running_swig">25.3 Running SWIG with Go</a></h2>
<p> Most Go programs are built using the <a href="https://golang.org/cmd/go/">
go tool</a>. Since Go 1.1 the go tool has support for SWIG. To use it,
give your SWIG interface file the extension .swig (for C code) or
.swigcxx (for C++ code). Put that file in a GOPATH/src directory as
usual for Go sources. Put other C/C++ code in the same directory with
extensions of .c and .cxx. The <tt>go build</tt> and <tt>go install</tt>
commands will automatically run SWIG for you and compile the generated
wrapper code. To check the SWIG command line options the go tool uses
run <tt>go build -x</tt>. To access the automatically generated files
run <tt>go build -work</tt>. You'll find the files under the temporary
WORK directory.</p>
<p> To manually generate and compile C/C++ wrapper code for Go, use the <tt>
-go</tt> option with SWIG. By default SWIG will generate code for the Go
compiler of the Go distribution. To generate code for gccgo, you should
also use the <tt>-gccgo</tt> option.</p>
<p> By default SWIG will generate files that can be used directly by <tt>
go build</tt>. This requires Go 1.2 or later. Put your SWIG interface
file in a directory under GOPATH/src, and give it a name that does<b>
not</b> end in the .swig or .swigcxx extension. Typically the SWIG
interface file extension is .i in this case.</p>
<div class="code">
<pre>
% swig -go example.i
% go install
</pre>
</div>
<p> You will now have a Go package that you can import from other Go
packages as usual.</p>
<h3><a name="Go_commandline">25.3.1 Go-specific Commandline Options</a></h3>
<p> These are the command line options for SWIG's Go module. They can
also be seen by using:</p>
<div class="code">
<pre>
swig -go -help
</pre>
</div>
<table summary="Go-specific options">
<tr><th>Go-specific options</th></tr>
<tr><td>-cgo</td><td>Generate files to be used as input for the Go cgo
tool. This is the default.</td></tr>
<tr><td>-no-cgo</td><td>This option is no longer supported.</td></tr>
<tr><td>-intgosize &lt;s&gt;</td><td>Set the size for the Go type <tt>int</tt>
. This controls the size that the C/C++ code expects to see. The &lt;s&gt;
argument should be 32 or 64. This option was required during the
transition from Go 1.0 to Go 1.1, as the size of <tt>int</tt> on 64-bit
x86 systems changed between those releases (from 32 bits to 64 bits).
It was made optional in SWIG 4.1.0 and if not specified SWIG will
assume that the size of <tt>int</tt> is the size of a C pointer.</td></tr>
<tr><td>-gccgo</td><td>Generate code for gccgo. The default is to
generate code for the Go compiler of the Go distribution.</td></tr>
<tr><td>-package &lt;name&gt;</td><td>Set the name of the Go package to
&lt;name&gt;. The default package name is the SWIG module name.</td></tr>
<tr><td>-use-shlib</td><td>Tell SWIG to emit code that uses a shared
library. This is only meaningful for the Go compiler of the Go
distribution, which needs to know at compile time whether a shared
library will be used.</td></tr>
<tr><td>-soname &lt;name&gt;</td><td>Set the runtime name of the shared
library that the dynamic linker should include at runtime. The default
is the package name with &quot;.so&quot; appended. This is only used when
generating code for the Go compiler of the Go distribution; when using
gccgo, the equivalent name will be taken from the <code>-soname</code>
option passed to the linker. Using this option implies the -use-shlib
option.</td></tr>
<tr><td>-go-pkgpath &lt;pkgpath&gt;</td><td>When generating code for gccgo,
set the pkgpath to use. This corresponds to the <tt>-fgo-pkgpath</tt>
option to gccgo.</td></tr>
<tr><td>-go-prefix &lt;prefix&gt;</td><td>When generating code for gccgo, set
the prefix to use. This corresponds to the <tt>-fgo-prefix</tt> option
to gccgo. If <tt>-go-pkgpath</tt> is used, <tt>-go-prefix</tt> will be
ignored.</td></tr>
<tr><td>-import-prefix &lt;prefix&gt;</td><td>A prefix to add when turning a
%import prefix in the SWIG interface file into an import statement in
the Go file. For example, with <code>-import-prefix mymodule</code>, a
SWIG interface file <code>%import mypackage</code> will become a Go
import statement <code>import &quot;mymodule/mypackage&quot;</code>.</td></tr>
</table>
<h3><a name="Go_outputs">25.3.2 Generated Wrapper Files</a></h3>
<p> SWIG will generate the following files when generating wrapper code:</p>
<ul>
<li> MODULE.go will contain the Go functions that your Go code will
call. These functions will be wrappers for the C++ functions defined by
your module. This file should, of course, be compiled with the Go
compiler.</li>
<li> MODULE_wrap.c or MODULE_wrap.cxx will contain C/C++ functions will
be invoked by the Go wrapper code. This file should be compiled with
the usual C or C++ compiler.</li>
<li> MODULE_wrap.h will be generated if you use the directors feature.
It provides a definition of the generated C++ director classes. It is
generally not necessary to use this file, but in some special cases it
may be helpful to include it in your code, compiled with the usual C or
C++ compiler.</li>
</ul>
<h2><a name="Go_basic_tour">25.4 A tour of basic C/C++ wrapping</a></h2>
<p> By default, SWIG attempts to build a natural Go interface to your
C/C++ code. However, the languages are somewhat different, so some
modifications have to occur. This section briefly covers the essential
aspects of this wrapping.</p>
<h3><a name="Go_package">25.4.1 Go Package Name</a></h3>
<p> All Go source code lives in a package. The name of this package will
default to the name of the module from SWIG's <tt>%module</tt>
directive. You may override this by using SWIG's <tt>-package</tt>
command line option.</p>
<h3><a name="Go_names">25.4.2 Go Names</a></h3>
<p> In Go, a function is only visible outside the current package if the
first letter of the name is uppercase. This is quite different from
C/C++. Because of this, C/C++ names are modified when generating the Go
interface: the first letter is forced to be uppercase if it is not
already. This affects the names of functions, methods, variables,
constants, enums, and classes.</p>
<p> C/C++ variables are wrapped with setter and getter functions in Go.
First the first letter of the variable name will be forced to
uppercase, and then <tt>Get</tt> or <tt>Set</tt> will be prepended. For
example, if the C/C++ variable is called <tt>var</tt>, then SWIG will
define the functions <tt>GetVar</tt> and <tt>SetVar</tt>. If a variable
is declared as <tt>const</tt>, or if SWIG's <a href="#SWIG_readonly_variables">
<tt>%immutable</tt> directive</a> is used for the variable, then only
the getter will be defined.</p>
<p> C++ classes will be discussed further below. Here we'll note that
the first letter of the class name will be forced to uppercase to give
the name of a type in Go. A constructor will be named <tt>New</tt>
followed by that name, and the destructor will be named <tt>Delete</tt>
followed by that name.</p>
<h3><a name="Go_constants">25.4.3 Go Constants</a></h3>
<p> C/C++ constants created via <tt>#define</tt> or the <tt>%constant</tt>
directive become Go constants, declared with a <tt>const</tt>
declaration.</p>
<h3><a name="Go_enumerations">25.4.4 Go Enumerations</a></h3>
<p> C/C++ enumeration types will cause SWIG to define an integer type
with the name of the enumeration (with first letter forced to uppercase
as usual). The values of the enumeration will become variables in Go;
code should avoid modifying those variables.</p>
<h3><a name="Go_classes">25.4.5 Go Classes</a></h3>
<p> Go has interfaces, methods and inheritance, but it does not have
classes in the same sense as C++. This sections describes how SWIG
represents C++ classes represented in Go.</p>
<p> For a C++ class <tt>ClassName</tt>, SWIG will define two types in
Go: an underlying type, which will just hold a pointer to the C++ type,
and an interface type. The interface type will be named <tt>ClassName</tt>
. SWIG will define a function <tt>NewClassName</tt> which will take any
constructor arguments and return a value of the interface type <tt>
ClassName</tt>. SWIG will also define a destructor <tt>DeleteClassName</tt>
.</p>
<p> SWIG will represent any methods of the C++ class as methods on the
underlying type, and also as methods of the interface type. Thus C++
methods may be invoked directly using the usual <tt>val.MethodName</tt>
syntax. Public members of the C++ class will be given getter and setter
functions defined as methods of the class.</p>
<p> SWIG will represent static methods of C++ classes as ordinary Go
functions. SWIG will use names like <tt>ClassNameMethodName</tt>. SWIG
will give static members getter and setter functions with names like <tt>
GetClassNameVarName</tt>.</p>
<p> Given a value of the interface type, Go code can retrieve the
pointer to the C++ type by calling the <tt>Swigcptr</tt> method. This
will return a value of type <tt>SwigcptrClassName</tt>, which is just a
name for <tt>uintptr</tt>. A Go type conversion can be used to convert
this value to a different C++ type, but note that this conversion will
not be type checked and is essentially equivalent to <tt>
reinterpret_cast</tt>. This should only be used for very special cases,
such as where C++ would use a <tt>dynamic_cast</tt>.</p>
<p>Note that C++ pointers to compound objects are represented in go as
objects themselves, not as go pointers. So, for example, if you wrap
the following function:</p>
<div class="code">
<pre>
class MyClass {
int MyMethod();
static MyClass *MyFactoryFunction();
};
</pre>
</div>
<p>You will get go code that looks like this:</p>
<div class="code">
<pre>
type MyClass interface {
Swigcptr() uintptr
SwigIsMyClass()
MyMethod() int
}
func MyClassMyFactoryFunction() MyClass {
// swig magic here
}
</pre>
</div>
<p>Note that the factory function does not return a go pointer; it
actually returns a go interface. If the returned pointer can be null,
you can check for this by calling the Swigcptr() method.</p>
<h4><a name="Go_class_memory">25.4.5.1 Go Class Memory Management</a></h4>
<p> Calling <tt>NewClassName</tt> for a C++ class <tt>ClassName</tt>
will allocate memory using the C++ memory allocator. This memory will
not be automatically freed by Go's garbage collector as the object
ownership is not tracked. When you are done with the C++ object you
must free it using <tt>DeleteClassName</tt>.
<br>
<br> The most Go idiomatic way to manage the memory for some C++ class
is to call <tt>NewClassName</tt> followed by a <tt><a href="https://golang.org/doc/effective_go.html#defer">
defer</a></tt> of the <tt>DeleteClassName</tt> call. Using <tt>defer</tt>
ensures that the memory of the C++ object is freed as soon as the
function containing the <tt>defer</tt> statement returns. Furthermore <tt>
defer</tt> works great for short-lived objects and fits nicely C++'s
RAII idiom. Example:</p>
<div class="code">
<pre>
func UseClassName(...) ... {
o := NewClassName(...)
defer DeleteClassName(o)
// Use the ClassName object
return ...
}
</pre>
</div>
<p> With increasing complexity, especially complex C++ object
hierarchies, the correct placement of <tt>defer</tt> statements becomes
harder and harder as C++ objects need to be freed in the correct order.
This problem can be eased by keeping a C++ object function local so
that it is only available to the function that creates a C++ object and
functions called by this function. Example:</p>
<div class="code">
<pre>
func WithClassName(constructor args, f func(ClassName, ...interface{}) error, data ...interface{}) error {
o := NewClassName(constructor args)
defer DeleteClassName(o)
return f(o, data...)
}
func UseClassName(o ClassName, data ...interface{}) (err error) {
// Use the ClassName object and additional data and return error.
}
func main() {
WithClassName(constructor args, UseClassName, additional data)
}
</pre>
</div>
<p> Using <tt>defer</tt> has limitations though, especially when it
comes to long-lived C++ objects whose lifetimes are hard to predict.
For such C++ objects a common technique is to store the C++ object into
a Go object, and to use the Go function <tt>runtime.SetFinalizer</tt>
to add a finalizer which frees the C++ object when the Go object is
freed. It is strongly recommended to read the <a href="https://golang.org/pkg/runtime/#SetFinalizer">
runtime.SetFinalizer</a> documentation before using this technique to
understand the <tt>runtime.SetFinalizer</tt> limitations.
<br></p>
<p> Common pitfalls with <tt>runtime.SetFinalizer</tt> are:</p>
<ul>
<li> If a hierarchy of C++ objects will be automatically freed by Go
finalizers then the Go objects that store the C++ objects need to
replicate the hierarchy of the C++ objects to prevent that C++ objects
are freed prematurely while other C++ objects still rely on them.</li>
<li> The usage of Go finalizers is problematic with C++'s RAII idiom as
it isn't predictable when the finalizer will run and this might require
a Close or Delete method to be added the Go object that stores a C++
object to mitigate.</li>
</ul>
<p> <tt>runtime.SetFinalizer</tt> Example:</p>
<div class="code">
<pre>
import (
&quot;runtime&quot;
&quot;wrap&quot; // SWIG generated wrapper code
)
type GoClassName struct {
wcn wrap.ClassName
}
func NewGoClassName() *GoClassName {
o := &amp;GoClassName{wcn: wrap.NewClassName()}
runtime.SetFinalizer(o, deleteGoClassName)
return o
}
func deleteGoClassName(o *GoClassName) {
// Runs typically in a different OS thread!
wrap.DeleteClassName(o.wcn)
o.wcn = nil
}
func (o *GoClassName) Close() {
// If the C++ object has a Close method.
o.wcn.Close()
// If the GoClassName object is no longer in an usable state.
runtime.SetFinalizer(o, nil) // Remove finalizer.
deleteGoClassName() // Free the C++ object.
}
</pre>
</div>
<h4><a name="Go_class_inheritance">25.4.5.2 Go Class Inheritance</a></h4>
<p> C++ class inheritance is automatically represented in Go due to its
use of interfaces. The interface for a child class will be a superset
of the interface of its parent class. Thus a value of the child class
type in Go may be passed to a function which expects the parent class.
Doing the reverse will require an explicit type assertion, which will
be checked dynamically.</p>
<h3><a name="Go_templates">25.4.6 Go Templates</a></h3>
<p> In order to use C++ templates in Go, you must tell SWIG to create
wrappers for a particular template instantiation. To do this, use the <tt>
%template</tt> directive.</p>
<h3><a name="Go_threads">25.4.7 Go and C/C++ Threads</a></h3>
<p> C and C++ code can use operating system threads and thread local
storage. Go code uses goroutines, which are multiplexed onto operating
system threads. This multiplexing means that Go code can change to run
on a different thread at any time. C/C++ code, on the other hand, may
assume that it runs on a single thread; this is true in particular if
the C/C++ code uses thread local storage.</p>
<p> In order to use Go code with C/C++ code that expects to run on a
single thread, the Go code must call the <a href="https://pkg.go.dev/runtime#LockOSThread">
<code>runtime.LockOSThread</code></a> function to lock the goroutine
onto a single thread.</p>
<h3><a name="Go_exceptions">25.4.8 Go and C++ Exceptions</a></h3>
<p> C++ exceptions do not interoperate with Go code. Attempts to throw
C++ exceptions through a Go caller are unreliable: in many cases the
C++ exception handler will be unable to unwind the stack, and the
program will crash. The only safe way to handle C++ exceptions is to
catch them in C++ before returning to Go.</p>
<h3><a name="Go_director_classes">25.4.9 Go Director Classes</a></h3>
<p> SWIG's director feature permits a Go type to act as the subclass of
a C++ class. This is complicated by the fact that C++ and Go define
inheritance differently. SWIG normally represents the C++ class
inheritance automatically in Go via interfaces but with a Go type
representing a subclass of a C++ class some manual work is necessary.</p>
<p> This subchapter gives a step by step guide how to properly subclass
a C++ class with a Go type. In general it is strongly recommended to
follow this guide completely to avoid common pitfalls with directors in
Go.</p>
<h4><a name="Go_director_example_cpp_code">25.4.9.1 Example C++ code</a></h4>
<p> The step by step guide is based on two example C++ classes.
FooBarAbstract is an abstract C++ class and the FooBarCpp class
inherits from it. This guide explains how to implement a FooBarGo class
similar to the FooBarCpp class.</p>
<p> <tt>FooBarAbstract</tt> abstract C++ class:</p>
<div class="code">
<pre>
class FooBarAbstract
{
public:
FooBarAbstract() {};
virtual ~FooBarAbstract() {};
std::string FooBar() {
return this-&gt;Foo() + &quot;, &quot; + this-&gt;Bar();
};
protected:
virtual std::string Foo() {
return &quot;Foo&quot;;
};
virtual std::string Bar() = 0;
};
</pre>
</div>
<p> <tt>FooBarCpp</tt> C++ class:</p>
<div class="code">
<pre>
class FooBarCpp : public FooBarAbstract
{
protected:
virtual std::string Foo() {
return &quot;C++ &quot; + FooBarAbstract::Foo();
}
virtual std::string Bar() {
return &quot;C++ Bar&quot;;
}
};
</pre>
</div>
<p> Returned string by the <tt>FooBarCpp::FooBar</tt> method is:</p>
<div class="code">
<pre>
C++ Foo, C++ Bar
</pre>
</div>
<p> The complete example, including the <tt>FooBarGoo</tt> class
implementation, can be found in <a href="#Go_director_foobargo_class">
the end of the guide</a>.</p>
<h4><a name="Go_director_enable">25.4.9.2 Enable director feature</a></h4>
<p> The director feature is disabled by default. To use directors you
must make two changes to the interface file. First, add the &quot;directors&quot;
option to the %module directive, like this:</p>
<div class="code">
<pre>
%module(directors=&quot;1&quot;) modulename
</pre>
</div>
<p> Second, you must use the %feature(&quot;director&quot;) directive to tell SWIG
which classes should get directors. In the example the FooBarAbstract
class needs the director feature enabled so that the FooBarGo class can
inherit from it, like this:</p>
<div class="code">
<pre>
%feature(&quot;director&quot;) FooBarAbstract;
</pre>
</div>
<p> For a more detailed documentation of the director feature and how to
enable or disable it for specific classes and virtual methods see
SWIG's Java documentation on directors.</p>
<h4><a name="Go_director_ctor_dtor">25.4.9.3 Constructor and destructor</a>
</h4>
<p> SWIG creates an additional set of constructor and destructor
functions once the director feature has been enabled for a C++ class. <tt>
NewDirectorClassName</tt> allows overriding virtual methods on the new
object instance and <tt>DeleteDirectorClassName</tt> needs to be used
to free a director object instance created with <tt>
NewDirectorClassName</tt>. More on overriding virtual methods follows
later in this guide under <a href="#Go_director_overriding">overriding
virtual methods</a>.</p>
<p> The default constructor and destructor functions <tt>NewClassName</tt>
and <tt>DeleteClassName</tt> can still be used as before so that
existing code doesn't break just because the director feature has been
enabled for a C++ class. The behavior is undefined if the default and
director constructor and destructor functions get mixed and so great
care needs to be taken that only one of the constructor and destructor
function pairs is used for any object instance. Both constructor
functions, the default and the director one, return the same interface
type. This makes it potentially hard to know which destructor function,
the default or the director one, needs to be called to delete an object
instance.</p>
<p> In<b> theory</b> the <tt>DirectorInterface</tt> method could be used
to determine if an object instance was created via <tt>
NewDirectorClassName</tt>:</p>
<div class="code">
<pre>
if o.DirectorInterface() != nil {
DeleteDirectorClassName(o)
} else {
DeleteClassName(o)
}
</pre>
</div>
<p> In<b> practice</b> it is strongly recommended to embed a director
object instance in a Go struct so that a director object instance will
be represented as a distinct Go type that subclasses a C++ class. For
this Go type custom constructor and destructor functions take care of
the director constructor and destructor function calls and the
resulting Go class will appear to the user as any other SWIG wrapped
C++ class. More on properly subclassing a C++ class follows later in
this guide under <a href="#Go_director_subclass">subclass via embedding</a>
.</p>
<h4><a name="Go_director_overriding">25.4.9.4 Override virtual methods</a>
</h4>
<p> In order to override virtual methods on a C++ class with Go methods
the <tt>NewDirectorClassName</tt> constructor functions receives a <tt>
DirectorInterface</tt> argument. The methods in the <tt>
DirectorInterface</tt> are a subset of the public and protected virtual
methods of the C++ class. Virtual methods that have a final specifier
are unsurprisingly excluded. If the <tt>DirectorInterface</tt> contains
a method with a matching signature to a virtual method of the C++ class
then the virtual C++ method will be overwritten with the Go method. As
Go doesn't support protected methods all overridden protected virtual
C++ methods will be public in Go.</p>
<p> As an example see part of the <tt>FooBarGo</tt> class:</p>
<div class="code">
<pre>
type overwrittenMethodsOnFooBarAbstract struct {
fb FooBarAbstract
}
func (om *overwrittenMethodsOnFooBarAbstract) Foo() string {
...
}
func (om *overwrittenMethodsOnFooBarAbstract) Bar() string {
...
}
func NewFooBarGo() FooBarGo {
om := &amp;overwrittenMethodsOnFooBarAbstract{}
fb := NewDirectorFooBarAbstract(om)
om.fb = fb
...
}
</pre>
</div>
<p> The complete example, including the <tt>FooBarGoo</tt> class
implementation, can be found in <a href="#Go_director_foobargo_class">
the end of the guide</a>. In this part of the example the virtual
methods <tt>FooBarAbstract::Foo</tt> and <tt>FooBarAbstract::Bar</tt>
have been overwritten with Go methods similarly to how the <tt>
FooBarAbstract</tt> virtual methods are overwritten by the <tt>FooBarCpp</tt>
class.</p>
<p> The <tt>DirectorInterface</tt> in the example is implemented by the <tt>
overwrittenMethodsOnFooBarAbstract</tt> Go struct type. A pointer to a <tt>
overwrittenMethodsOnFooBarAbstract</tt> struct instance will be given to
the <tt>NewDirectorFooBarAbstract</tt> constructor function. The
constructor return value implements the <tt>FooBarAbstract</tt>
interface. <tt>overwrittenMethodsOnFooBarAbstract</tt> could in theory
be any Go type but in practice a struct is used as it typically
contains at least a value of the C++ class interface so that the
overwritten methods can use the rest of the C++ class. If the <tt>
FooBarGo</tt> class would receive additional constructor arguments then
these would also typically be stored in the <tt>
overwrittenMethodsOnFooBarAbstract</tt> struct so that they can be used
by the Go methods.</p>
<h4><a name="Go_director_base_methods">25.4.9.5 Call base methods</a></h4>
<p> Often a virtual method will be overwritten to extend the original
behavior of the method in the base class. This is also the case for the
<tt>FooBarCpp::Foo</tt> method of the example code:</p>
<div class="code">
<pre>
virtual std::string Foo() {
return &quot;C++ &quot; + FooBarAbstract::Foo();
}
</pre>
</div>
<p> To use base methods the <tt>DirectorClassNameMethodName</tt> wrapper
functions are automatically generated by SWIG for public and protected
virtual methods. The <tt>FooBarGo.Foo</tt> implementation in the
example looks like this:</p>
<div class="code">
<pre>
func (om *overwrittenMethodsOnFooBarAbstract) Foo() string {
return &quot;Go &quot; + DirectorFooBarAbstractFoo(om.fb)
}
</pre>
</div>
<p> The complete example, including the <tt>FooBarGoo</tt> class
implementation, can be found in <a href="#Go_director_foobargo_class">
the end of the guide</a>.</p>
<h4><a name="Go_director_subclass">25.4.9.6 Subclass via embedding</a></h4>
<p> <a href="#Go_director_ctor_dtor">As previously mentioned in this
guide</a> the default and director constructor functions return the
same interface type. To properly subclass a C++ class with a Go type
the director object instance returned by the <tt>NewDirectorClassName</tt>
constructor function should be embedded into a Go struct so that it
represents a distinct but compatible type in Go's type system. This Go
struct should be private and the constructor and destructor functions
should instead work with a public interface type so that the Go class
that subclasses a C++ class can be used as a compatible drop in.</p>
<p> The subclassing part of the <tt>FooBarGo</tt> class for an example
looks like this:</p>
<div class="code">
<pre>
type FooBarGo interface {
FooBarAbstract
deleteFooBarAbstract()
IsFooBarGo()
}
type fooBarGo struct {
FooBarAbstract
}
func (fbgs *fooBarGo) deleteFooBarAbstract() {
DeleteDirectorFooBarAbstract(fbgs.FooBarAbstract)
}
func (fbgs *fooBarGo) IsFooBarGo() {}
func NewFooBarGo() FooBarGo {
om := &amp;overwrittenMethodsOnFooBarAbstract{}
fb := NewDirectorFooBarAbstract(om)
om.fb = fb
return &amp;fooBarGo{FooBarAbstract: fb}
}
func DeleteFooBarGo(fbg FooBarGo) {
fbg.deleteFooBarAbstract()
}
</pre>
</div>
<p> The complete example, including the <tt>FooBarGoo</tt> class
implementation, can be found in <a href="#Go_director_foobargo_class">
the end of the guide</a>. In this part of the example the private <tt>
fooBarGo</tt> struct embeds <tt>FooBarAbstract</tt> which lets the <tt>
fooBarGo</tt> Go type &quot;inherit&quot; all the methods of the <tt>
FooBarAbstract</tt> C++ class by means of embedding. The public <tt>
FooBarGo</tt> interface type includes the <tt>FooBarAbstract</tt>
interface and hence <tt>FooBarGo</tt> can be used as a drop in
replacement for <tt>FooBarAbstract</tt> while the reverse isn't
possible and would raise a compile time error. Furthermore the
constructor and destructor functions <tt>NewFooBarGo</tt> and <tt>
DeleteFooBarGo</tt> take care of all the director specifics and to the
user the class appears as any other SWIG wrapped C++ class.</p>
<h4><a name="Go_director_finalizer">25.4.9.7 Memory management with
runtime.SetFinalizer</a></h4>
<p> In general all guidelines for <a href="#Go_class_memory">C++ class
memory management</a> apply as well to director classes. One often
overlooked limitation with <tt>runtime.SetFinalizer</tt> is that a
finalizer doesn't run in case of a cycle and director classes typically
have a cycle. The cycle in the <tt>FooBarGo</tt> class is here:</p>
<div class="code">
<pre>
type overwrittenMethodsOnFooBarAbstract struct {
fb FooBarAbstract
}
func NewFooBarGo() FooBarGo {
om := &amp;overwrittenMethodsOnFooBarAbstract{}
fb := NewDirectorFooBarAbstract(om) // fb.v = om
om.fb = fb // Backlink causes cycle as fb.v = om!
...
}
</pre>
</div>
<p> In order to be able to use <tt>runtime.SetFinalizer</tt>
nevertheless the finalizer needs to be set on something that isn't in a
cycle and that references the director object instance. In the <tt>
FooBarGo</tt> class example the <tt>FooBarAbstract</tt> director
instance can be automatically deleted by setting the finalizer on <tt>
fooBarGo</tt>:</p>
<div class="code">
<pre>
type fooBarGo struct {
FooBarAbstract
}
type overwrittenMethodsOnFooBarAbstract struct {
fb FooBarAbstract
}
func NewFooBarGo() FooBarGo {
om := &amp;overwrittenMethodsOnFooBarAbstract{}
fb := NewDirectorFooBarAbstract(om)
om.fb = fb // Backlink causes cycle as fb.v = om!
fbgs := &amp;fooBarGo{FooBarAbstract: fb}
runtime.SetFinalizer(fbgs, FooBarGo.deleteFooBarAbstract)
return fbgs
}
</pre>
</div>
<p> Furthermore if <tt>runtime.SetFinalizer</tt> is in use either the <tt>
DeleteClassName</tt> destructor function needs to be removed or the <tt>
fooBarGo</tt> struct needs additional data to prevent double deletion.
Please read the <a href="#Go_class_memory">C++ class memory management</a>
subchapter before using <tt>runtime.SetFinalizer</tt> to know all of
its gotchas.</p>
<h4><a name="Go_director_foobargo_class">25.4.9.8 Complete FooBarGo
example class</a></h4>
<p> The complete and annotated <tt>FooBarGo</tt> class looks like this:</p>
<div class="code">
<pre>
// FooBarGo is a superset of FooBarAbstract and hence FooBarGo can be used as a
// drop in replacement for FooBarAbstract but the reverse causes a compile time
// error.
type FooBarGo interface {
FooBarAbstract
deleteFooBarAbstract()
IsFooBarGo()
}
// Via embedding fooBarGo &quot;inherits&quot; all methods of FooBarAbstract.
type fooBarGo struct {
FooBarAbstract
}
func (fbgs *fooBarGo) deleteFooBarAbstract() {
DeleteDirectorFooBarAbstract(fbgs.FooBarAbstract)
}
// The IsFooBarGo method ensures that FooBarGo is a superset of FooBarAbstract.
// This is also how the class hierarchy gets represented by the SWIG generated
// wrapper code. For an instance FooBarCpp has the IsFooBarAbstract and
// IsFooBarCpp methods.
func (fbgs *fooBarGo) IsFooBarGo() {}
// Go type that defines the DirectorInterface. It contains the Foo and Bar
// methods that overwrite the respective virtual C++ methods on FooBarAbstract.
type overwrittenMethodsOnFooBarAbstract struct {
// Backlink to FooBarAbstract so that the rest of the class can be used by
// the overridden methods.
fb FooBarAbstract
// If additional constructor arguments have been given they are typically
// stored here so that the overridden methods can use them.
}
func (om *overwrittenMethodsOnFooBarAbstract) Foo() string {
// DirectorFooBarAbstractFoo calls the base method FooBarAbstract::Foo.
return &quot;Go &quot; + DirectorFooBarAbstractFoo(om.fb)
}
func (om *overwrittenMethodsOnFooBarAbstract) Bar() string {
return &quot;Go Bar&quot;
}
func NewFooBarGo() FooBarGo {
// Instantiate FooBarAbstract with selected methods overridden. The methods
// that will be overwritten are defined on
// overwrittenMethodsOnFooBarAbstract and have a compatible signature to the
// respective virtual C++ methods. Furthermore additional constructor
// arguments will be typically stored in the
// overwrittenMethodsOnFooBarAbstract struct.
om := &amp;overwrittenMethodsOnFooBarAbstract{}
fb := NewDirectorFooBarAbstract(om)
om.fb = fb // Backlink causes cycle as fb.v = om!
fbgs := &amp;fooBarGo{FooBarAbstract: fb}
// The memory of the FooBarAbstract director object instance can be
// automatically freed once the FooBarGo instance is garbage collected by
// uncommenting the following line. Please make sure to understand the
// runtime.SetFinalizer specific gotchas before doing this. Furthermore
// DeleteFooBarGo should be deleted if a finalizer is in use or the fooBarGo
// struct needs additional data to prevent double deletion.
// runtime.SetFinalizer(fbgs, FooBarGo.deleteFooBarAbstract)
return fbgs
}
// Recommended to be removed if runtime.SetFinalizer is in use.
func DeleteFooBarGo(fbg FooBarGo) {
fbg.deleteFooBarAbstract()
}
</pre>
</div>
<p> Returned string by the <tt>FooBarGo.FooBar</tt> method is:</p>
<div class="code">
<pre>
Go Foo, Go Bar
</pre>
</div>
<p> For comparison the <tt>FooBarCpp</tt> class looks like this:</p>
<div class="code">
<pre>
class FooBarCpp : public FooBarAbstract
{
protected:
virtual std::string Foo() {
return &quot;C++ &quot; + FooBarAbstract::Foo();
}
virtual std::string Bar() {
return &quot;C++ Bar&quot;;
}
};
</pre>
</div>
<p> For comparison the returned string by the <tt>FooBarCpp::FooBar</tt>
method is:</p>
<div class="code">
<pre>
C++ Foo, C++ Bar
</pre>
</div>
<p> The complete source of this example can be found under <a href="https://github.com/swig/swig/tree/master/Examples/go/director">
SWIG/Examples/go/director/</a>.</p>
<h3><a name="Go_primitive_type_mappings">25.4.10 Default Go primitive
type mappings</a></h3>
<p> The following table lists the default type mapping from C/C++ to Go.
This table will tell you which Go type to expect for a function which
uses a given C/C++ type.</p>
<table BORDER summary="Go primitive type mappings">
<tr><td><b>C/C++ type</b></td><td><b>Go type</b></td></tr>
<tr><td>bool</td><td>bool</td></tr>
<tr><td>char</td><td>byte</td></tr>
<tr><td>signed char</td><td>int8</td></tr>
<tr><td>unsigned char</td><td>byte</td></tr>
<tr><td>short</td><td>int16</td></tr>
<tr><td>unsigned short</td><td>uint16</td></tr>
<tr><td>int</td><td>int</td></tr>
<tr><td>unsigned int</td><td>uint</td></tr>
<tr><td>long</td><td>int64</td></tr>
<tr><td>unsigned long</td><td>uint64</td></tr>
<tr><td>long long</td><td>int64</td></tr>
<tr><td>unsigned long long</td><td>uint64</td></tr>
<tr><td>float</td><td>float32</td></tr>
<tr><td>double</td><td>float64</td></tr>
<tr><td>char *
<br>char []</td><td>string</td></tr>
</table>
<p> Note that SWIG wraps the C <tt>char</tt> type as a character.
Pointers and arrays of this type are wrapped as strings. The <tt>signed
char</tt> type can be used if you want to treat <tt>char</tt> as a
signed number rather than a character. Also note that all const
references to primitive types are treated as if they are passed by
value.</p>
<p> These type mappings are defined by the &quot;gotype&quot; typemap. You may
change that typemap, or add new values, to control how C/C++ types are
mapped into Go types.</p>
<h3><a name="Go_output_arguments">25.4.11 Output arguments</a></h3>
<p>Because of limitations in the way output arguments are processed in
swig, a function with output arguments will not have multiple return
values. Instead, you must pass a pointer into the C++ function to tell
it where to store the output value. In go, you supply a slice in the
place of the output argument.</p>
<p>For example, suppose you were trying to wrap the modf() function in
the C math library which splits x into integral and fractional parts
(and returns the integer part in one of its parameters):</p>
<div class="code">
<pre>
double modf(double x, double *ip);
</pre>
</div>
<p>You could wrap it with SWIG as follows:</p>
<div class="code">
<pre>
%include &lt;typemaps.i&gt;
double modf(double x, double *OUTPUT);
</pre>
</div>
<p>or you can use the <code>%apply</code> directive:</p>
<div class="code">
<pre>
%include &lt;typemaps.i&gt;
%apply double *OUTPUT { double *ip };
double modf(double x, double *ip);
</pre>
</div>
<p>In Go you would use it like this:</p>
<div class="code">
<pre>
ptr := []float64{0.0}
fraction := modulename.Modf(5.0, ptr)
</pre>
</div>
<p>Since this is ugly, you may want to wrap the swig-generated API with
some <a href="#Go_adding_additional_code">additional functions written
in go</a> that hide the ugly details.</p>
<p>There are no <code>char&nbsp;*OUTPUT</code> typemaps. However you can
apply the <code>signed&nbsp;char&nbsp;*</code> typemaps instead:</p>
<div class="code">
<pre>
%include &lt;typemaps.i&gt;
%apply signed char *OUTPUT {char *output};
void f(char *output);
</pre>
</div>
<h3><a name="Go_adding_additional_code">25.4.12 Adding additional go
code</a></h3>
<p>Often the APIs generated by swig are not very natural in go,
especially if there are output arguments. You can insert additional go
wrapping code to add new APIs with <code>%insert(go_wrapper)</code>,
like this:</p>
<div class="code">
<pre>
%include &lt;typemaps.i&gt;
// Change name of what swig generates to Wrapped_modf. This function will
// have the following signature in go:
// func Wrapped_modf(float64, []float64) float64
%rename(wrapped_modf) modf(double x, double *ip);
%apply double *OUTPUT { double *ip };
double modf(double x, double *ip);
%insert(go_wrapper) %{
// The improved go interface to this function, which has two return values,
// in the more natural go idiom:
func Modf(x float64) (fracPart float64, intPart float64) {
ip := []float64{0.0}
fracPart = Wrapped_modf(x, ip)
intPart = ip[0]
return
}
%}
</pre>
</div>
<p>For classes, since swig generates an interface, you can add
additional methods by defining another interface that includes the
swig-generated interface. For example,</p>
<div class="code">
<pre>
%rename(Wrapped_MyClass) MyClass;
%rename(Wrapped_GetAValue) MyClass::GetAValue(int *x);
%apply int *OUTPUT { int *x };
class MyClass {
public:
MyClass();
int AFineMethod(const char *arg); // Swig's wrapping is fine for this one.
bool GetAValue(int *x);
};
%insert(go_wrapper) %{
type MyClass interface {
Wrapped_MyClass
GetAValue() (int, bool)
}
func (arg SwigcptrWrapped_MyClass) GetAValue() (int, bool) {
ip := []int{0}
ok := arg.Wrapped_GetAValue(ip)
return ip[0], ok
}
%}
</pre>
</div>
<p>Of course, if you have to rewrite most of the methods, instead of
just a few, then you might as well define your own struct that includes
the swig-wrapped object, instead of adding methods to the
swig-generated object.</p>
<p>If you need to import other go packages, you can do this with <code>
%go_import</code>. For example,</p>
<div class="code">
<pre>
%go_import(&quot;fmt&quot;, _ &quot;unusedPackage&quot;, rp &quot;renamed/package&quot;)
%insert(go_wrapper) %{
func foo() {
fmt.Println(&quot;Some string:&quot;, rp.GetString())
}
// Importing the same package twice is permitted,
// Go code will be generated with only the first instance of the import.
%go_import(&quot;fmt&quot;)
%insert(go_wrapper) %{
func bar() {
fmt.Println(&quot;Hello world!&quot;)
}
%}
</pre>
</div>
<h3><a name="Go_typemaps">25.4.13 Go typemaps</a></h3>
<p> You can use the <tt>%typemap</tt> directive to modify SWIG's default
wrapping behavior for specific C/C++ types. You need to be familiar
with the material in the general &quot;<a href="#Typemaps">Typemaps</a>&quot;
chapter. That chapter explains how to define a typemap. This section
describes some specific typemaps used for Go.</p>
<p> In general type conversion code may be written either in C/C++ or in
Go. The choice to make normally depends on where memory should be
allocated. To allocate memory controlled by the Go garbage collector,
write Go code. To allocate memory in the C/C++ heap, write C code.</p>
<table BORDER summary="Go Typemaps">
<tr><td><b>Typemap</b></td><td><b>Description</b></td></tr>
<tr><td>gotype</td><td> The Go type to use for a C++ type. This type
will appear in the generated Go wrapper function. If this is not
defined SWIG will use a default as <a href="#Go_primitive_type_mappings">
described above</a>.</td></tr>
<tr><td>imtype</td><td> An intermediate Go type used by the &quot;goin&quot;,
&quot;goout&quot;, &quot;godirectorin&quot;, and &quot;godirectorout&quot; typemaps. If this typemap
is not defined for a C/C++ type, the gotype typemap will be used. This
is useful when gotype is best converted to C/C++ using Go code.</td></tr>
<tr><td>goin</td><td> Go code to convert from gotype to imtype when
calling a C/C++ function. SWIG will then internally convert imtype to a
C/C++ type and pass it down. If this is not defined, or is the empty
string, no conversion is done.</td></tr>
<tr><td>in</td><td> C/C++ code to convert the internally generated C/C++
type, based on imtype, into the C/C++ type that a function call
expects. If this is not defined the value will simply be cast to the
desired type.</td></tr>
<tr><td>out</td><td> C/C++ code to convert the C/C++ type that a
function call returns into the internally generated C/C++ type, based
on imtype, that will be returned to Go. If this is not defined the
value will simply be cast to the desired type.</td></tr>
<tr><td>goout</td><td> Go code to convert a value returned from a C/C++
function from imtype to gotype. If this is not defined, or is the empty
string, no conversion is done.</td></tr>
<tr><td>argout</td><td> C/C++ code to adjust an argument value when
returning from a function. This is called after the real C/C++ function
has run. This uses the internally generated C/C++ type, based on
imtype. This is only useful for a pointer type of some sort. If this is
not defined nothing will be done.</td></tr>
<tr><td>goargout</td><td> Go code to adjust an argument value when
returning from a function. This is called after the real C/C++ function
has run. The value will be in imtype. This is only useful for a pointer
type of some sort. If this is not defined, or is the empty string,
nothing will be done.</td></tr>
<tr><td>directorin</td><td> C/C++ code to convert the C/C++ type used to
call a director method into the internally generated C/C++ type, based
on imtype, that will be passed to Go. If this is not defined the value
will simply be cast to the desired type.</td></tr>
<tr><td>godirectorin</td><td> Go code to convert a value used to call a
director method from imtype to gotype. If this is not defined, or is
the empty string, no conversion is done.</td></tr>
<tr><td>godirectorout</td><td> Go code to convert a value returned from
a director method from gotype to imtype. If this is not defined, or is
the empty string, no conversion is done.</td></tr>
<tr><td>directorout</td><td> C/C++ code to convert a value returned from
a director method from the internally generated C/C++ type, based on
imtype, into the type that the method should return If this is not
defined the value will simply be cast to the desired type.</td></tr>
</table>
<HR NOSHADE>
<h1><a name="Guile">26 SWIG and Guile</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Guile_nn1">Supported Guile Versions</a></li>
<li><a href="#Guile_nn2">Meaning of &quot;Module&quot;</a></li>
<li><a href="#Guile_nn3">Old GH Guile API</a></li>
<li><a href="#Guile_nn4">Linkage</a>
<ul>
<li><a href="#Guile_nn5">Simple Linkage</a></li>
<li><a href="#Guile_nn6">Passive Linkage</a></li>
<li><a href="#Guile_nn7">Native Guile Module Linkage</a></li>
<li><a href="#Guile_nn8">Old Auto-Loading Guile Module Linkage</a></li>
<li><a href="#Guile_nn9">Hobbit4D Linkage</a></li>
</ul>
</li>
<li><a href="#Guile_nn10">Underscore Folding</a></li>
<li><a href="#Guile_nn11">Typemaps</a></li>
<li><a href="#Guile_nn12">Representation of pointers as smobs</a>
<ul>
<li><a href="#Guile_nn14">Smobs</a></li>
<li><a href="#Guile_nn15">Garbage Collection</a></li>
</ul>
</li>
<li><a href="#Guile_nn16">Native Guile pointers</a></li>
<li><a href="#Guile_nn17">Exception Handling</a></li>
<li><a href="#Guile_nn18">Procedure documentation</a></li>
<li><a href="#Guile_nn19">Procedures with setters</a></li>
<li><a href="#Guile_nn20">GOOPS Proxy Classes</a>
<ul>
<li><a href="#Guile_nn21">Naming Issues</a></li>
<li><a href="#Guile_nn22">Linking</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> This section details guile-specific support in SWIG.</p>
<h2><a name="Guile_nn1">26.1 Supported Guile Versions</a></h2>
<p> SWIG is known to work with Guile versions 2.0.x, 2.2.x and 3.0.x
(these are all tested via CI). SWIG probably still works with Guile
1.8.x but we're no longer able to regularly test this either in CI or
by hand. Support for Guile 1.6.x has been dropped (SWIG 2.0.9 was the
last version of SWIG to support it).</p>
<p> Note that starting with guile 2.0, the guile sources can be compiled
for improved performance. This is currently not tested with swig so
your mileage may vary. To be safe set environment variable <tt>
GUILE_AUTO_COMPILE</tt> to 0 when using swig generated guile code.</p>
<h2><a name="Guile_nn2">26.2 Meaning of &quot;Module&quot;</a></h2>
<p> There are three different concepts of &quot;module&quot; involved, defined
separately for SWIG, Guile, and Libtool. To avoid horrible confusion,
we explicitly prefix the context, e.g., &quot;guile-module&quot;.</p>
<h2><a name="Guile_nn3">26.3 Old GH Guile API</a></h2>
<p>Guile 1.8 and older could be interfaced using two different api's,
the SCM or the GH API. The GH interface to guile is deprecated. Read
more about why in the <a href="https://www.gnu.org/software/guile/docs/docs-1.6/guile-ref/GH.html#GH">
Guile manual</a>.</p>
<p>Support for the guile GH wrapper code generation has been dropped
from SWIG. The last version of SWIG that can still generate guile GH
wrapper code is 2.0.9. Please use that version if you really need the
GH wrapper code.</p>
<h2><a name="Guile_nn4">26.4 Linkage</a></h2>
<p> Guile support is complicated by a lack of user community
cohesiveness, which manifests in multiple shared-library usage
conventions. A set of policies implementing a usage convention is
called a<b> linkage</b>.</p>
<h3><a name="Guile_nn5">26.4.1 Simple Linkage</a></h3>
<p> The default linkage is the simplest; nothing special is done. In
this case the function <code>SWIG_init()</code> is exported. Simple
linkage can be used in several ways:</p>
<ul>
<li><b>Embedded Guile, no modules.</b> You want to embed a Guile
interpreter into your program; all bindings made by SWIG shall show up
in the root module. Then call <code>SWIG_init()</code> in the <code>
inner_main()</code> function. See the &quot;simple&quot; and &quot;matrix&quot; examples
under <code>Examples/guile</code>.</li>
<li>
<p><b>Dynamic module mix-in.</b> You want to create a Guile module using
<code>define-module</code>, containing both Scheme code and bindings
made by SWIG; you want to load the SWIG modules as shared libraries
into Guile.</p>
<div class="targetlang">
<pre>
(define-module (my module))
(define my-so (dynamic-link &quot;./libexample.so&quot;))
(dynamic-call &quot;SWIG_init&quot; my-so) ; make SWIG bindings
;; Scheme definitions can go here
</pre>
</div>
<p> Newer Guile versions provide a shorthand for <code>dynamic-link</code>
and <code>dynamic-call</code>:</p>
<div class="targetlang">
<pre>
(load-extension &quot;./libexample.so&quot; &quot;SWIG_init&quot;)
</pre>
</div>
<p> A more portable approach would be to drop the shared library
extension:</p>
<div class="targetlang">
<pre>
(load-extension &quot;./libexample&quot; &quot;SWIG_init&quot;)
</pre>
</div>
<p> You need to explicitly export those bindings made by SWIG that you
want to import into other modules:</p>
<div class="targetlang">
<pre>
(export foo bar)
</pre>
</div>
<p> In this example, the procedures <code>foo</code> and <code>bar</code>
would be exported. Alternatively, you can export all bindings with the
following module-system hack:</p>
<div class="targetlang">
<pre>
(module-map (lambda (sym var)
(module-export! (current-module) (list sym)))
(current-module))
</pre>
</div>
<p>SWIG can also generate this Scheme stub (from <code>define-module</code>
up to <code>export</code>) semi-automagically if you pass it the
command-line argument <code>-scmstub</code>. The code will be exported
in a file called <code><i>module</i>.scm</code> in the directory
specified by <code>-outdir</code> or the current directory if <code>
-outdir</code> is not specified. Since SWIG doesn't know how to load
your extension module (with <code>dynamic-link</code> or <code>
load-extension</code>), you need to supply this information by including
a directive like this in the interface file:</p>
<div class="code">
<pre>
%scheme %{ (load-extension &quot;./libexample.so&quot; &quot;SWIG_init&quot;) %}
</pre>
</div>
<p> (The <code>%scheme</code> directive allows inserting arbitrary
Scheme code into the generated file <code><var>module.scm</var></code>;
it is placed between the <code>define-module</code> form and the <code>
export</code> form.)</p>
</li>
</ul>
<p>If you want to include several SWIG modules, you would need to rename
<code>SWIG_init</code> via a preprocessor define to avoid symbol
clashes. For this case, however, passive linkage is available.</p>
<h3><a name="Guile_nn6">26.4.2 Passive Linkage</a></h3>
<p>Passive linkage is just like simple linkage, but it generates an
initialization function whose name is derived from the module and
package name (see below).</p>
<p>You should use passive linkage rather than simple linkage when you
are using multiple modules.</p>
<h3><a name="Guile_nn7">26.4.3 Native Guile Module Linkage</a></h3>
<p>SWIG can also generate wrapper code that does all the Guile module
declarations on its own if you pass it the <code>-Linkage module</code>
command-line option.</p>
<p>The module name is set with the <code>-package</code> and <code>
-module</code> command-line options. Suppose you want to define a module
with name <code>(my lib foo)</code>; then you would have to pass the
options <code>-package<var> my</var>/<var>lib</var> -module<var> foo</var>
</code>. Note that the last part of the name can also be set via the
SWIG directive <code>%module</code>.</p>
<p>You can use this linkage in several ways:</p>
<ul>
<li><b>Embedded Guile with SWIG modules.</b> You want to embed a Guile
interpreter into your program; the SWIG bindings shall be put into
different modules. Simply call the function <code>scm_init_<var>my</var>
_<var>modules</var>_<var>foo</var>_module</code> in the <code>
inner_main()</code> function.</li>
<li><b>Dynamic Guile modules.</b> You want to load the SWIG modules as
shared libraries into Guile; all bindings are automatically put in
newly created Guile modules.<div class="targetlang">
<pre>
(define my-so (dynamic-link &quot;./libfoo&quot;))
;; create new module and put bindings there:
(dynamic-call &quot;scm_init_my_modules_foo_module&quot; my-so)
</pre>
</div> Newer Guile versions have a shorthand procedure for this:<div class="targetlang">
<pre>
(load-extension &quot;./libfoo.so&quot; &quot;scm_init_my_modules_foo_module&quot;)
</pre>
</div></li>
</ul>
<h3><a name="Guile_nn8">26.4.4 Old Auto-Loading Guile Module Linkage</a></h3>
<p>Guile used to support an autoloading facility for object-code
modules, but this support was deprecated and removed in Guile version
1.4.1. SWIG supported this via option <code>-Linkage ltdlmod</code>,
but this support is no longer useful and was removed in SWIG 4.2.0.</p>
<h3><a name="Guile_nn9">26.4.5 Hobbit4D Linkage</a></h3>
<p> The only other linkage supported at this time creates shared object
libraries suitable for use by hobbit's <code>(hobbit4d link)</code>
guile module. This is called the &quot;hobbit&quot; linkage, and requires also
using the &quot;-package&quot; command line option to set the part of the module
name before the last symbol. For example, both command lines:</p>
<div class="shell">
<pre>
swig -guile -package my/lib foo.i
swig -guile -package my/lib -module foo foo.i
</pre>
</div>
<p> would create module <code>(my lib foo)</code> (assuming in the first
case foo.i declares the module to be &quot;foo&quot;). The installed files are
my/lib/libfoo.so.X.Y.Z and friends. This scheme is still very
experimental; the (hobbit4d link) conventions are not well understood.</p>
<h2><a name="Guile_nn10">26.5 Underscore Folding</a></h2>
<p> Underscores are converted to dashes in identifiers. Guile support
may grow an option to inhibit this folding in the future, but no one
has complained so far.</p>
<p>You can use the <a href="#SWIG_rename_ignore">SWIG directive <code>
%rename</code></a> to specify the Guile names of the wrapped functions
and variables.</p>
<h2><a name="Guile_nn11">26.6 Typemaps</a></h2>
<p> The Guile module handles all types via typemaps. This information is
read from <code>Lib/guile/typemaps.i</code>. Some non-standard typemap
substitutions are supported:</p>
<ul>
<li><code>$descriptor</code> expands to a type descriptor for use with
the <code>SWIG_NewPointerObj()</code> and <code>SWIG_ConvertPtr</code>
functions.</li>
<li>For pointer types, <code>$*descriptor</code> expands to a descriptor
for the direct base type (i.e., one pointer is stripped), whereas <code>
$basedescriptor</code> expands to a descriptor for the base type (i.e.,
all pointers are stripped).</li>
</ul>
<p>A function returning <code>void</code> (more precisely, a function
whose <code>out</code> typemap returns <code>SCM_UNSPECIFIED</code>) is
treated as returning no values. In <code>argout</code> typemaps, one
can use the macro <code>GUILE_APPEND_RESULT</code> in order to append a
value to the list of function return values.</p>
<p>Multiple values can be passed up to Scheme in one of three ways:</p>
<ul>
<li>
<p><em>Multiple values as lists.</em> By default, if more than one value
is to be returned, a list of the values is created and returned; to
switch back to this behavior, use</p>
<div class="code">
<pre>
%values_as_list;</pre>
</div></li>
<li>
<p><em>Multiple values as vectors.</em> By issuing</p>
<div class="code">
<pre>
%values_as_vector;</pre>
</div>
<p> vectors instead of lists will be used.</p>
</li>
<li>
<p><em>Multiple values for multiple-value continuations.</em><strong>
This is the most elegant way.</strong> By issuing</p>
<div class="code">
<pre>
%multiple_values;</pre>
</div>
<p> multiple values are passed to the multiple-value continuation, as
created by <code>call-with-values</code> or the convenience macro <code>
receive</code>. The latter is available if you issue <code>(use-modules
(srfi srfi-8))</code>. Assuming that your <code>divide</code> function
wants to return two values, a quotient and a remainder, you can write:</p>
<div class="targetlang">
<pre>
(receive (quotient remainder)
(divide 35 17)
<var>body</var>...)
</pre>
</div>
<p> In <code><var>body</var></code>, the first result of <code>divide</code>
will be bound to the variable <code>quotient</code>, and the second
result to <code>remainder</code>.</p>
</li>
</ul>
<p> See also the &quot;multivalue&quot; example.</p>
<p>Constants are exported as a function that returns the value. The
%feature(&quot;constasvar&quot;) can be applied to any constant, immutable
variable, or enum. Instead of exporting the constant as a function that
must be called, the constant will appear as a scheme variable. See <a href="#Customization_features">
Features and the %feature directive</a> for info on how to apply the
%feature.</p>
<h2><a name="Guile_nn12">26.7 Representation of pointers as smobs</a></h2>
<p> For pointer types, SWIG uses Guile smobs. SWIG smobs print like
this: <code>#&lt;swig struct xyzzy * 0x1234affe&gt;</code> Two of them are <code>
equal?</code> if and only if they have the same type and value.</p>
<p> To construct a Scheme object from a C pointer, the wrapper code
calls the function <code>SWIG_NewPointerObj()</code>, passing a pointer
to a struct representing the pointer type. The type index to store in
the upper half of the CAR is read from this struct. To get the pointer
represented by a smob, the wrapper code calls the function <code>
SWIG_ConvertPtr()</code>, passing a pointer to a struct representing the
expected pointer type. See also <a href="#Typemaps_runtime_type_checker">
The run-time type checker</a>. If the Scheme object passed was not a
SWIG smob representing a compatible pointer, a <code>wrong-type-arg</code>
exception is raised.</p>
<h3><a name="Guile_nn14">26.7.1 Smobs</a></h3>
<p> In earlier versions of SWIG, C pointers were represented as Scheme
strings containing a hexadecimal rendering of the pointer value and a
mangled type name. As Guile allows registering user types, so-called
&quot;smobs&quot; (small objects), a much cleaner representation has been
implemented now. The details will be discussed in the following.</p>
<p>The whole type system, when it is first initialized, creates two
smobs named &quot;swig&quot; and &quot;collected_swig&quot;. The swig smob is used for
non-garbage collected smobs, while the collected_swig smob is used as
described below. Each smob has the same format, which is a double cell
created by SCM_NEWSMOB2() The first word of data is the pointer to the
object and the second word of data is the swig_type_info * structure
describing this type. If a generated GOOPS module has been loaded,
smobs will be wrapped by the corresponding GOOPS class.</p>
<h3><a name="Guile_nn15">26.7.2 Garbage Collection</a></h3>
<p>Garbage collection is a feature of Guile since version 1.6. As SWIG
now requires Guile &gt; 1.8, it is automatically included. Garbage
collection works like this. Every swig_type_info structure stores in
its clientdata field a pointer to the destructor for this type. The
destructor is the generated wrapper around the delete function. So swig
still exports a wrapper for the destructor, it just does not call
scm_c_define_gsubr() for the wrapped delete function. So the only way
to delete an object is from the garbage collector, since the delete
function is not available to scripts. How swig determines if a type
should be garbage collected is exactly like described in <a href="#Customization_ownership">
Object ownership and %newobject</a> in the SWIG manual. All typemaps
use an $owner var, and the guile module replaces $owner with 0 or 1
depending on feature:new.</p>
<h2><a name="Guile_nn16">26.8 Native Guile pointers</a></h2>
<p> In addition to SWIG smob pointers, <a href="https://www.gnu.org/software/guile/manual/html_node/Foreign-Pointers.html">
Guile's native pointer type</a> are accepted as arguments to wrapped
SWIG functions. This can be useful for passing <a href="https://www.gnu.org/software/guile/manual/html_node/Void-Pointers-and-Byte-Access.html#">
pointers to bytevector data</a> to wrapped functions.</p>
<h2><a name="Guile_nn17">26.9 Exception Handling</a></h2>
<p> SWIG code calls <code>scm_error</code> on exception, using the
following mapping:<div class="code">
<pre>
MAP(SWIG_MemoryError, &quot;swig-memory-error&quot;);
MAP(SWIG_IOError, &quot;swig-io-error&quot;);
MAP(SWIG_RuntimeError, &quot;swig-runtime-error&quot;);
MAP(SWIG_IndexError, &quot;swig-index-error&quot;);
MAP(SWIG_TypeError, &quot;swig-type-error&quot;);
MAP(SWIG_DivisionByZero, &quot;swig-division-by-zero&quot;);
MAP(SWIG_OverflowError, &quot;swig-overflow-error&quot;);
MAP(SWIG_SyntaxError, &quot;swig-syntax-error&quot;);
MAP(SWIG_ValueError, &quot;swig-value-error&quot;);
MAP(SWIG_SystemError, &quot;swig-system-error&quot;);
MAP(SWIG_NullReferenceError, &quot;swig-null-reference-error&quot;);
</pre>
</div></p>
<p> The default when not specified here is to use &quot;swig-error&quot;. See
Lib/exception.i for details.</p>
<h2><a name="Guile_nn18">26.10 Procedure documentation</a></h2>
<p>If invoked with the command-line option <code>-procdoc<var> file</var>
</code>, SWIG creates documentation strings for the generated wrapper
functions, describing the procedure signature and return value, and
writes them to<var> file</var>.</p>
<p>SWIG can generate documentation strings in three formats, which are
selected via the command-line option <code>-procdocformat<var> format</var>
</code>:</p>
<ul>
<li><code>guile-1.4</code> (default): Generates a format suitable for
Guile 1.4.</li>
<li><code>plain</code>: Generates a format suitable for Guile 1.4.1 and
later.</li>
<li><code>texinfo</code>: Generates texinfo source, which must be run
through texinfo in order to get a format suitable for Guile 1.4.1 and
later.</li>
</ul>
<p>You need to register the generated documentation file with Guile like
this:<div class="targetlang">
<pre>
(use-modules (ice-9 documentation))
(set! documentation-files
(cons &quot;<var>file</var>&quot; documentation-files))
</pre>
</div></p>
<p>Documentation strings can be configured using the Guile-specific
typemap argument <code>doc</code>. See <code>Lib/guile/typemaps.i</code>
for details.</p>
<h2><a name="Guile_nn19">26.11 Procedures with setters</a></h2>
<p>For global variables, SWIG creates a single wrapper procedure <code>(<var>
variable</var> :optional value)</code>, which is used for both getting
and setting the value. For struct members, SWIG creates two wrapper
procedures <code>(<var>struct</var>-<var>member</var>-get pointer)</code>
and <code>(<var>struct-member</var>-set pointer value)</code>.</p>
<p>If invoked with the command-line option <code>-emit-setters</code> (<em>
recommended</em>), SWIG will additionally create procedures with
setters. For global variables, the procedure-with-setter <code><var>
variable</var></code> is created, so you can use <code>(<var>variable</var>
)</code> to get the value and <code>(set! (<var>variable</var>)<var>
value</var>)</code> to set it. For struct members, the
procedure-with-setter <code><var>struct</var>-<var>member</var></code>
is created, so you can use <code>(<var>struct</var>-<var>member</var><var>
pointer</var>)</code> to get the value and <code>(set! (<var>struct</var>
-<var>member</var><var> pointer</var>)<var> value</var>)</code> to set
it.</p>
<p>If invoked with the command-line option <code>-only-setters</code>,
SWIG will<em> only</em> create procedures with setters, i.e., for
struct members, the procedures <code>(<var>struct</var>-<var>member</var>
-get pointer)</code> and <code>(<var>struct-member</var>-set pointer
value)</code> are<em> not</em> generated.</p>
<h2><a name="Guile_nn20">26.12 GOOPS Proxy Classes</a></h2>
<p>SWIG can also generate classes and generic functions for use with
Guile's Object-Oriented Programming System (GOOPS). GOOPS is a
sophisticated object system in the spirit of the Common Lisp Object
System (CLOS).</p>
<p>To enable GOOPS support, pass the <code>-proxy</code> argument to
swig. This will export the GOOPS wrapper definitions into the <code><i>
module</i>.scm</code> file in the directory specified by -outdir or the
current directory. GOOPS support requires either passive or module
linkage.</p>
<p>The generated file will contain definitions of GOOPS classes
mimicking the C++ class hierarchy.</p>
<p>Enabling GOOPS support implies <code>-emit-setters</code>.</p>
<p>If <code>-emit-slot-accessors</code> is also passed as an argument,
then the generated file will contain accessor methods for all the slots
in the classes and for global variables. The input class</p>
<div class="code">
<pre>
class Foo {
public:
Foo(int i) : a(i) {}
int a;
int getMultBy(int i) { return a * i; }
Foo getFooMultBy(int i) { return Foo(a * i); }
};
Foo getFooPlus(int i) { return Foo(a + i); }
</pre>
</div>
<p> will produce (if <code>-emit-slot-accessors</code> is not passed as
a parameter)</p>
<div class="targetlang">
<pre>
(define-class &lt;Foo&gt; (&lt;swig&gt;)
(a #:allocation #:swig-virtual
#:slot-ref primitive:Foo-a-get
#:slot-set! primitive:Foo-a-set)
#:metaclass &lt;swig-metaclass&gt;
#:new-function primitive:new-Foo
)
(define-method (getMultBy (swig_smob &lt;Foo&gt;) i)
(primitive:Foo-getMultBy (slot-ref swig_smob 'smob) i))
(define-method (getFooMultBy (swig_smob &lt;Foo&gt;) i)
(make &lt;Foo&gt; #:init-smob (primitive:Foo-getFooMultBy (slot-ref swig_smob 'smob) i)))
(define-method (getFooPlus i)
(make &lt;Foo&gt; #:init-smob (primitive:getFooPlus i)))
(export &lt;Foo&gt; getMultBy getFooMultBy getFooPlus )
</pre>
</div>
<p> and will produce (if <code>-emit-slot-accessors</code> is passed as
a parameter)</p>
<div class="targetlang">
<pre>
(define-class &lt;Foo&gt; (&lt;swig&gt;)
(a #:allocation #:swig-virtual
#:slot-ref primitive:Foo-a-get
#:slot-set! primitive:Foo-a-set
<b>#:accessor a</b>)
#:metaclass &lt;swig-metaclass&gt;
#:new-function primitive:new-Foo
)
(define-method (getMultBy (swig_smob &lt;Foo&gt;) i)
(primitive:Foo-getMultBy (slot-ref swig_smob 'smob) i))
(define-method (getFooMultBy (swig_smob &lt;Foo&gt;) i)
(make &lt;Foo&gt; #:init-smob (primitive:Foo-getFooMultBy (slot-ref swig_smob 'smob) i)))
(define-method (getFooPlus i)
(make &lt;Foo&gt; #:init-smob (primitive:getFooPlus i)))
(export &lt;Foo&gt; <b>a</b> getMultBy getFooMultBy getFooPlus )
</pre>
</div>
<p> which can then be used by this code</p>
<div class="targetlang">
<pre>
;; not using getters and setters
(define foo (make &lt;Foo&gt; #:args '(45)))
(slot-ref foo 'a)
(slot-set! foo 'a 3)
(getMultBy foo 4)
(define foo2 (getFooMultBy foo 7))
(slot-ref foo 'a)
(slot-ref (getFooPlus foo 4) 'a)
;; using getters and setters
(define foo (make &lt;Foo&gt; #:args '(45)))
(a foo)
(set! (a foo) 5)
(getMultBy foo 4)
(a (getFooMultBy foo 7))
</pre>
</div>
<p>Notice that constructor arguments are passed as a list after the <code>
#:args</code> keyword. Hopefully in the future the following will be
valid <code>(make &lt;Foo&gt; #:a 5 #:b 4)</code></p>
<p>Also note that the order the declarations occur in the .i file make a
difference. For example,</p>
<div class="code">
<pre>
%module test
%{ #include &quot;foo.h&quot; %}
%inline %{
int someFunc(Foo &amp;a) {
...
}
%}
%include &quot;foo.h&quot;
</pre>
</div>
<p> This is a valid SWIG file it will work as you think it will for
primitive support, but the generated GOOPS file will be broken. Since
the <code>someFunc</code> definition is parsed by SWIG before all the
declarations in foo.h, the generated GOOPS file will contain the
definition of <code>someFunc()</code> before the definition of &lt;Foo&gt;.
The generated GOOPS file would look like</p>
<div class="targetlang">
<pre>
;;...
(define-method (someFunc (swig_smob &lt;Foo&gt;))
(primitive:someFunc (slot-ref swig_smob 'smob)))
;;...
(define-class &lt;Foo&gt; (&lt;swig&gt;)
;;...
)
;;...
</pre>
</div>
<p> Notice that &lt;Foo&gt; is used before it is defined. The fix is to just
put the <code>%import &quot;foo.h&quot;</code> before the <code>%inline</code>
block.</p>
<h3><a name="Guile_nn21">26.12.1 Naming Issues</a></h3>
<p>As you can see in the example above, there are potential naming
conflicts. The default exported accessor for the <code>Foo::a</code>
variable is named <code>a</code>. The name of the wrapper global
function is <code>getFooPlus</code>. If the <code>-useclassprefix</code>
option is passed to swig, the name of all accessors and member
functions will be prepended with the class name. So the accessor will
be called <code>Foo-a</code> and the member functions will be called <code>
Foo-getMultBy</code>. Also, if the <code>-goopsprefix goops:</code>
argument is passed to swig, every identifier will be prefixed by <code>
goops:</code></p>
<p>Two guile-modules are created by SWIG. The first module contains the
primitive definitions of all the wrapped functions and variables, and
is located either in the _wrap.cxx file (with <code>-Linkage module</code>
) or in the scmstub file (if <code>-Linkage passive -scmstub</code>).
The name of this guile-module is the swig-module name (given on the
command line with the -module argument or with the %module directive)
concatenated with the string &quot;-primitive&quot;. For example, if <code>
%module Test</code> is set in the swig interface file, the name of the
guile-module in the scmstub or <code>-Linkage module</code> will be <code>
Test-primitive</code>. Also, the scmstub file will be named <code>
Test-primitive.scm</code>. The string &quot;primitive&quot; can be changed by the <code>
-primsuffix</code> swig argument. So the same interface, with the <code>
-primsuffix base</code> will produce a module called <code>Test-base</code>
. The second generated guile-module contains all the GOOPS class
definitions and is located in a file named<i> module</i>.scm in the
directory specified with -outdir or the current directory. The name of
this guile-module is the name of the swig-module (given on the command
line or with the <code>%module</code> directive). In the previous
example, the GOOPS definitions will be in a file named Test.scm.</p>
<p>Because of the naming conflicts, you can't in general use both the <code>
-primitive</code> and the GOOPS guile-modules at the same time. To do
this, you need to rename the exported symbols from one or both
guile-modules. For example,</p>
<div class="targetlang">
<pre>
(use-modules ((Test-primitive) #:renamer (symbol-prefix-proc 'primitive:)))
(use-modules ((Test) #:renamer (symbol-prefix-proc 'goops:)))
</pre>
</div>
<h3><a name="Guile_nn22">26.12.2 Linking</a></h3>
<p>The guile-modules generated above all need to be linked together.
GOOPS support requires either passive or module linkage. The exported
GOOPS guile-module will be the name of the swig-module and should be
located in a file called<i> Module</i>.scm. This should be installed on
the autoload path for guile, so that <code>(use-modules (<i>Package
Module</i>))</code> will load everything needed. Thus, the top of the
GOOPS guile-module will contain code to load everything needed by the
interface (the shared library, the scmstub module, etc.). The <code>
%goops</code> directive inserts arbitrary code into the generated GOOPS
guile-module, and should be used to load the dependent libraries.</p>
<p>This breaks up into three cases</p>
<ul>
<li><b>Passive Linkage without -scmstub</b>: Note that this linkage
style has the potential for naming conflicts, since the primitive
exported function and variable names are not wrapped in a guile-module
and might conflict with names from the GOOPS guile-module (see above).
Pass the -goopsprefix argument to solve this problem. If the <code>
-exportprimitive</code> option is passed to SWIG the <code>(export ...)</code>
code that would be exported into the scmstub file is exported at the
bottom of the generated GOOPS guile-module. The <code>%goops</code>
directive should contain code to load the shared library.<div class="code">
<pre>
%goops %{ (load-extension &quot;./libfoo.so&quot; &quot;scm_init_my_modules_foo_module&quot;) %}
</pre>
</div>
<p> Produces the following code at the top of the generated GOOPS
guile-module (with the <code>-package my/modules -module foo</code>
command line arguments)</p>
<div class="targetlang">
<pre>
(define-module (my modules foo))
;; %goops directive goes here
(load-extension &quot;./libfoo.so&quot; &quot;scm_init_my_modules_foo_module&quot;)
(use-modules (oop goops) (Swig common))
</pre>
</div></li>
<li>
<p><b>Passive Linkage with -scmstub</b>: Here, the name of the scmstub
file should be <code>Module-primitive.scm</code> (with<i> primitive</i>
replaced with whatever is given with the <code>-primsuffix</code>
argument. The code to load the shared library should be located in the <code>
%scheme</code> directive, which will then be added to the scmstub file.
SWIG will automatically generate the line <code>(use-modules (<i>
Package</i><i> Module-primitive</i>))</code> into the GOOPS
guile-module. So if<i> Module-primitive.scm</i> is on the autoload path
for guile, the <code>%goops</code> directive can be empty. Otherwise,
the <code>%goops</code> directive should contain whatever code is
needed to load the<i> Module-primitive.scm</i> file into guile.</p>
<div class="targetlang">
<pre>
%scheme %{ (load-extension &quot;./libfoo.so&quot; &quot;scm_init_my_modules_foo_module&quot;) %}
// only include the following definition if (my modules foo) cannot
// be loaded automatically
%goops %{
(primitive-load &quot;/path/to/foo-primitive.scm&quot;)
(primitive-load &quot;/path/to/Swig/common.scm&quot;)
%}
</pre>
</div>
<p> Produces the following code at the top of the generated GOOPS
guile-module</p>
<div class="targetlang">
<pre>
(define-module (my modules foo))
;; %goops directive goes here (if any)
(primitive-load &quot;/path/to/foo-primitive.scm&quot;)
(primitive-load &quot;/path/to/Swig/common.scm&quot;)
(use-modules (oop goops) (Swig common))
(use-modules ((my modules foo-primitive) :renamer (symbol-prefix-proc
'primitive:)))
</pre>
</div></li>
<li>
<p><b>Module Linkage</b>: This is very similar to passive linkage with a
scmstub file. SWIG will also automatically generate the line <code>
(use-modules (<i>Package</i><i> Module-primitive</i>))</code> into the
GOOPS guile-module. Again the <code>%goops</code> directive should
contain whatever code is needed to get that module loaded into guile.</p>
<div class="code">
<pre>
%goops %{ (load-extension &quot;./libfoo.so&quot; &quot;scm_init_my_modules_foo_module&quot;) %}
</pre>
</div>
<p> Produces the following code at the top of the generated GOOPS
guile-module</p>
<div class="targetlang">
<pre>
(define-module (my modules foo))
;; %goops directive goes here (if any)
(load-extension &quot;./libfoo.so&quot; &quot;scm_init_my_modules_foo_module&quot;)
(use-modules (oop goops) (Swig common))
(use-modules ((my modules foo-primitive) :renamer (symbol-prefix-proc
'primitive:)))
</pre>
</div></li>
</ul>
<p><b>(Swig common)</b>: The generated GOOPS guile-module also imports
definitions from the (Swig common) guile-module. This module is
included with SWIG and should be installed by SWIG into the autoload
path for guile (based on the configure script and whatever arguments
are passed). If it is not, then the <code>%goops</code> directive also
needs to contain code to load the <code>common.scm</code> file into
guile. Also note that if you are trying to install the generated
wrappers on a computer without SWIG installed, you will need to include
the common.swg file along with the install.</p>
<p><b>Multiple Modules</b>: Type dependencies between modules is
supported. For example, if <code>mod1</code> includes definitions of
some classes, and <code>mod2</code> includes some classes derived from
classes in <code>mod1</code>, the generated GOOPS file for <code>mod2</code>
will declare the correct superclasses. The only problem is that since <code>
mod2</code> uses symbols from <code>mod1</code>, the <code>mod2</code>
GOOPS file must include a <code>(use-modules (mod2))</code>. Currently,
SWIG does not automatically export this line; it must be included in
the <code>%goops</code> directive of <code>mod2</code>. Maybe in the
future SWIG can detect dependencies and export this line. (how do other
language modules handle this problem?)</p>
<HR NOSHADE>
<h1><a name="Java">27 SWIG and Java</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Java_overview">Overview</a></li>
<li><a href="#Java_preliminaries">Preliminaries</a>
<ul>
<li><a href="#Java_running_swig">Running SWIG</a></li>
<li><a href="#Java_commandline">Additional Commandline Options</a></li>
<li><a href="#Java_getting_right_headers">Getting the right header files</a>
</li>
<li><a href="#Java_compiling_dynamic">Compiling a dynamic module</a></li>
<li><a href="#Java_using_module">Using your module</a></li>
<li><a href="#Java_dynamic_linking_problems">Dynamic linking problems</a>
</li>
<li><a href="#Java_compilation_problems_cpp">Compilation problems and
compiling with C++</a></li>
<li><a href="#Java_building_windows">Building on Windows</a>
<ul>
<li><a href="#Java_visual_studio">Running SWIG from Visual Studio</a></li>
<li><a href="#Java_nmake">Using NMAKE</a></li>
</ul>
</li>
</ul>
</li>
<li><a href="#Java_basic_tour">A tour of basic C/C++ wrapping</a>
<ul>
<li><a href="#Java_module_packages_classes">Modules, packages and
generated Java classes</a></li>
<li><a href="#Java_functions">Functions</a></li>
<li><a href="#Java_global_variables">Global variables</a></li>
<li><a href="#Java_constants">Constants</a></li>
<li><a href="#Java_enumerations">Enumerations</a>
<ul>
<li><a href="#Java_anonymous_enums">Anonymous enums</a></li>
<li><a href="#Java_typesafe_enums">Typesafe enums</a></li>
<li><a href="#Java_proper_enums">Proper Java enums</a></li>
<li><a href="#Java_typeunsafe_enums">Type unsafe enums</a></li>
<li><a href="#Java_simple_enums">Simple enums</a></li>
</ul>
</li>
<li><a href="#Java_pointers">Pointers</a></li>
<li><a href="#Java_structures">Structures</a></li>
<li><a href="#Java_classes">C++ classes</a></li>
<li><a href="#Java_inheritance">C++ inheritance</a></li>
<li><a href="#Java_pointers_refs_arrays">Pointers, references, arrays
and pass by value</a>
<ul>
<li><a href="#Java_null_pointers">Null pointers</a></li>
</ul>
</li>
<li><a href="#Java_overloaded_functions">C++ overloaded functions</a></li>
<li><a href="#Java_default_arguments">C++ default arguments</a></li>
<li><a href="#Java_namespaces">C++ namespaces</a></li>
<li><a href="#Java_templates">C++ templates</a></li>
<li><a href="#Java_smart_pointers">C++ Smart Pointers</a>
<ul>
<li><a href="#Java_smart_pointers_shared_ptr">The shared_ptr Smart
Pointer</a></li>
<li><a href="#Java_smart_pointers_generic">Generic Smart Pointers</a></li>
</ul>
</li>
</ul>
</li>
<li><a href="#Java_further_details">Further details on the generated
Java classes</a>
<ul>
<li><a href="#Java_imclass">The intermediary JNI class</a>
<ul>
<li><a href="#Java_imclass_pragmas">The intermediary JNI class pragmas</a>
</li>
</ul>
</li>
<li><a href="#Java_module_class">The Java module class</a>
<ul>
<li><a href="#Java_module_class_pragmas">The Java module class pragmas</a>
</li>
</ul>
</li>
<li><a href="#Java_constants_interface">The Java constants interface</a>
<ul>
<li><a href="#Java_constants_interface_pragmas">The Java constants
interface pragmas</a></li>
</ul>
</li>
<li><a href="#Java_proxy_classes">Java proxy classes</a>
<ul>
<li><a href="#Java_memory_management">Memory management</a></li>
<li><a href="#Java_inheritance_mirroring">Inheritance</a></li>
<li><a href="#Java_proxy_classes_gc">Proxy classes and garbage
collection</a></li>
<li><a href="#Java_pgcpp">The premature garbage collection prevention
parameter for proxy class marshalling</a></li>
<li><a href="#Java_multithread_libraries">Single threaded applications
and thread safety</a></li>
</ul>
</li>
<li><a href="#Java_type_wrapper_classes">Type wrapper classes</a></li>
<li><a href="#Java_enum_classes">Enum classes</a>
<ul>
<li><a href="#Java_typesafe_enums_classes">Typesafe enum classes</a></li>
<li><a href="#Java_proper_enums_classes">Proper Java enum classes</a></li>
<li><a href="#Java_typeunsafe_enums_classes">Type unsafe enum classes</a>
</li>
</ul>
</li>
<li><a href="#Java_interfaces">Interfaces</a></li>
</ul>
</li>
<li><a href="#Java_directors">Cross language polymorphism using
directors</a>
<ul>
<li><a href="#Java_enabling_directors">Enabling directors</a></li>
<li><a href="#Java_directors_classes">Director classes</a></li>
<li><a href="#Java_directors_overhead">Overhead and code bloat</a></li>
<li><a href="#Java_directors_example">Simple directors example</a></li>
<li><a href="#Java_directors_threading">Director threading issues</a></li>
<li><a href="#Java_directors_performance">Director performance tuning</a>
</li>
<li><a href="#Java_exceptions_from_directors">Java exceptions from
directors</a>
<ul>
<li><a href="#Java_customizing_director_exceptions">Customizing director
exceptions</a></li>
</ul>
</li>
<li><a href="#Java_protected_virtual_methods">Accessing virtual
protected methods</a></li>
<li><a href="#Java_allprotected">Accessing non-virtual protected members</a>
</li>
</ul>
</li>
<li><a href="#Java_common_customization">Common customization features</a>
<ul>
<li><a href="#Java_helper_functions">C/C++ helper functions</a></li>
<li><a href="#Java_class_extension">Class extension with %extend</a></li>
<li><a href="#Java_proxycode">Class extension with %proxycode</a></li>
<li><a href="#Java_exception_handling">Exception handling with
%exception and %javaexception</a></li>
<li><a href="#Java_method_access">Method access with
%javamethodmodifiers</a></li>
<li><a href="#Java_begin">Java begin</a></li>
</ul>
</li>
<li><a href="#Java_tips_techniques">Tips and techniques</a>
<ul>
<li><a href="#Java_input_output_parameters">Input and output parameters
using primitive pointers and references</a></li>
<li><a href="#Java_simple_pointers">Simple pointers</a></li>
<li><a href="#Java_c_arrays">Wrapping C arrays with Java arrays</a></li>
<li><a href="#Java_unbounded_c_arrays">Unbounded C Arrays</a></li>
<li><a href="#Java_string_length">Passing a string with length</a></li>
<li><a href="#Java_heap_allocations">Overriding new and delete to
allocate from Java heap</a></li>
</ul>
</li>
<li><a href="#Java_typemaps">Java typemaps</a>
<ul>
<li><a href="#Java_default_primitive_type_mappings">Default primitive
type mappings</a></li>
<li><a href="#Java_default_non_primitive_typemaps">Default typemaps for
non-primitive types</a></li>
<li><a href="#Java_jvm64">Sixty four bit JVMs</a></li>
<li><a href="#Java_what_is_typemap">What is a typemap?</a></li>
<li><a href="#Java_typemaps_c_to_java_types">Typemaps for mapping C/C++
types to Java types</a></li>
<li><a href="#Java_typemap_attributes">Java typemap attributes</a></li>
<li><a href="#Java_special_variables">Java special variables</a></li>
<li><a href="#Java_typemaps_for_c_and_cpp">Typemaps for both C and C++
compilation</a></li>
<li><a href="#Java_code_typemaps">Java code typemaps</a></li>
<li><a href="#Java_directors_typemaps">Director specific typemaps</a></li>
</ul>
</li>
<li><a href="#Java_typemap_examples">Typemap Examples</a>
<ul>
<li><a href="#Java_simpler_enum_classes">Simpler Java enums for enums
without initializers</a></li>
<li><a href="#Java_exception_typemap">Handling C++ exception
specifications as Java exceptions</a></li>
<li><a href="#Java_nan_exception_typemap">NaN Exception - exception
handling for a particular type</a></li>
<li><a href="#Java_converting_java_string_arrays">Converting Java String
arrays to char **</a></li>
<li><a href="#Java_expanding_java_object">Expanding a Java object to
multiple arguments</a></li>
<li><a href="#Java_using_typemaps_return_arguments">Using typemaps to
return arguments</a></li>
<li><a href="#Java_adding_downcasts">Adding Java downcasts to
polymorphic return types</a></li>
<li><a href="#Java_adding_equals_method">Adding an equals method to the
Java classes</a></li>
<li><a href="#Java_void_pointers">Void pointers and a common Java base
class</a></li>
<li><a href="#Java_struct_pointer_pointer">Struct pointer to pointer</a></li>
<li><a href="#Java_memory_management_member_variables">Memory management
when returning references to member variables</a></li>
<li><a href="#Java_memory_management_objects">Memory management for
objects passed to the C++ layer</a></li>
<li><a href="#Java_date_marshalling">Date marshalling using the javain
typemap and associated attributes</a></li>
</ul>
</li>
<li><a href="#Java_directors_faq">Living with Java Directors</a></li>
<li><a href="#Java_odds_ends">Odds and ends</a>
<ul>
<li><a href="#Java_javadoc_comments">JavaDoc comments</a></li>
<li><a href="#Java_functional_interface">Functional interface without
proxy classes</a></li>
<li><a href="#Java_using_own_jni_functions">Using your own JNI functions</a>
</li>
<li><a href="#Java_performance">Performance concerns and hints</a></li>
<li><a href="#Java_debugging">Debugging</a></li>
</ul>
</li>
<li><a href="#Java_examples">Java Examples</a></li>
</ul>
</div>
<!-- INDEX -->
<p> This chapter describes SWIG's support of Java. It covers most SWIG
features, but certain low-level details are covered in less depth than
in earlier chapters.</p>
<h2><a name="Java_overview">27.1 Overview</a></h2>
<p> The 100% Pure Java effort is a commendable concept, however in the
real world programmers often either need to re-use their existing code
or in some situations want to take advantage of Java but are forced
into using some native (C/C++) code. The Java extension to SWIG makes
it very easy to plumb in existing C/C++ code for access from Java, as
SWIG writes the Java Native Interface (JNI) code for you. It is
different to using the 'javah' tool as SWIG will wrap existing C/C++
code, whereas javah takes 'native' Java function declarations and
creates C/C++ function prototypes. SWIG wraps C/C++ code using Java
proxy classes and is very useful if you want to have access to large
amounts of C/C++ code from Java. If only one or two JNI functions are
needed then using SWIG may be overkill. SWIG enables a Java program to
easily call into C/C++ code from Java. Historically, SWIG was not able
to generate any code to call into Java code from C++. However, SWIG now
supports full cross language polymorphism and code is generated to call
up from C++ to Java when wrapping C++ virtual methods via the director
feature.</p>
<p> Java is one of the few non-scripting language modules in SWIG. As
SWIG utilizes the type safety that the Java language offers, it takes a
somewhat different approach to that used for scripting languages. In
particular runtime type checking and the runtime library are not used
by Java. This should be borne in mind when reading the rest of the SWIG
documentation. This chapter on Java is relatively self contained and
will provide you with nearly everything you need for using SWIG and
Java. However, the &quot;<a href="#SWIG">SWIG Basics</a>&quot; chapter will be a
useful read in conjunction with this one.</p>
<p> This chapter starts with a few practicalities on running SWIG and
compiling the generated code. If you are looking for the minimum amount
to read, have a look at the sections up to and including the <a href="#Java_basic_tour">
tour of basic C/C++ wrapping</a> section which explains how to call the
various C/C++ code constructs from Java. Following this section are
details of the C/C++ code and Java classes that SWIG generates. Due to
the complexities of C and C++ there are different ways in which C/C++
code could be wrapped and called from Java. SWIG is a powerful tool and
the rest of the chapter details how the default code wrapping can be
tailored. Various customisation tips and techniques using SWIG
directives are covered. The latter sections cover the advanced
techniques of using typemaps for complete control of the wrapping
process.</p>
<h2><a name="Java_preliminaries">27.2 Preliminaries</a></h2>
<p> SWIG 1.1 works with JDKs from JDK 1.1 to JDK1.4 (Java 2 SDK1.4) and
should also work with any later versions. Given the choice, you should
probably use the latest version of Sun's JDK. The SWIG Java module is
known to work using Sun's JVM on Solaris, Linux and the various
flavours of Microsoft Windows including Cygwin. The Kaffe JVM is known
to give a few problems and at the time of writing was not a fully
fledged JVM with full JNI support. The generated code is also known to
work on vxWorks using WindRiver's PJava 3.1. The best way to determine
whether your combination of operating system and JDK will work is to
test the examples and test-suite that comes with SWIG. Run <tt>make -k
check</tt> from the SWIG root directory after installing SWIG on Unix
systems.</p>
<p> The Java module requires your system to support shared libraries and
dynamic loading. This is the commonly used method to load JNI code so
your system will more than likely support this.</p>
<p> Android uses Java JNI and also works with SWIG. Please read the <a href="#Android">
Android chapter</a> in conjunction with this one if you are targeting
Android.</p>
<h3><a name="Java_running_swig">27.2.1 Running SWIG</a></h3>
<p> Suppose that you defined a SWIG module such as the following:</p>
<div class="code">
<pre>
/* File: example.i */
%module test
%{
#include &quot;stuff.h&quot;
%}
int fact(int n);
</pre>
</div>
<p> To build a Java module, run SWIG using the <tt>-java</tt> option :</p>
<div class="code">
<pre>
%swig -java example.i
</pre>
</div>
<p> If building C++, add the <tt>-c++</tt> option:</p>
<div class="code">
<pre>
$ swig -c++ -java example.i
</pre>
</div>
<p> This creates two different files; a C/C++ source file <tt>
example_wrap.c</tt> or <tt>example_wrap.cxx</tt> and numerous Java
files. The generated C/C++ source file contains the JNI wrapper code
that needs to be compiled and linked with the rest of your C/C++
application.</p>
<p> The name of the wrapper file is derived from the name of the input
file. For example, if the input file is <tt>example.i</tt>, the name of
the wrapper file is <tt>example_wrap.c</tt>. To change this, you can
use the <tt>-o</tt> option. It is also possible to change the <a href="#SWIG_output">
output directory</a> that the Java files are generated into using <tt>
-outdir</tt>.</p>
<p> The module name, specified with <tt>%module</tt>, determines the
name of various generated classes as discussed <a href="#Java_module_packages_classes">
later</a>. Note that the module name does not define a Java package and
by default, the generated Java classes do not have a Java package. The <tt>
-package</tt> option described below can specify a Java package name to
use.</p>
<p> The following sections have further practical examples and details
on how you might go about compiling and using the generated files.</p>
<h3><a name="Java_commandline">27.2.2 Additional Commandline Options</a></h3>
<p> The following table lists the additional commandline options
available for the Java module. They can also be seen by using:</p>
<div class="code">
<pre>
swig -java -help
</pre>
</div>
<table summary="Java specific options">
<tr><th>Java specific options</th></tr>
<tr><td>-nopgcpp</td><td>suppress the premature garbage collection
prevention parameter</td></tr>
<tr><td>-noproxy</td><td>generate the low-level functional interface
instead of proxy classes</td></tr>
<tr><td>-package &lt;name&gt;</td><td>set name of the Java package to &lt;name&gt;</td>
</tr>
</table>
<p> Their use will become clearer by the time you have finished reading
this section on SWIG and Java.</p>
<h3><a name="Java_getting_right_headers">27.2.3 Getting the right header
files</a></h3>
<p> In order to compile the C/C++ wrappers, the compiler needs the <tt>
jni.h</tt> and <tt>jni_md.h</tt> header files which are part of the JDK.
They are usually in directories like this:</p>
<div class="code">
<pre>
/usr/java/include
/usr/java/include/&lt;operating_system&gt;
</pre>
</div>
<p> The exact location may vary on your machine, but the above locations
are typical.</p>
<h3><a name="Java_compiling_dynamic">27.2.4 Compiling a dynamic module</a>
</h3>
<p> The JNI code exists in a dynamic module or shared library (DLL on
Windows) and gets loaded by the JVM. Assuming you have code you need to
link to in a file called <tt>example.c</tt>, in order to build a shared
library file, you need to compile your module in a manner similar to
the following (shown for Solaris):</p>
<div class="code">
<pre>
$ swig -java example.i
$ gcc -fPIC -c example_wrap.c -I/usr/java/include -I/usr/java/include/solaris
$ gcc -fPIC -c example.c
$ ld -G example_wrap.o example.o -o libexample.so
</pre>
</div>
<p> The exact commands for doing this vary from platform to platform.
However, SWIG tries to guess the right options when it is installed.
Therefore, you may want to start with one of the examples in the <tt>
Examples/java</tt> directory. If that doesn't work, you will need to
read the man-pages for your compiler and linker to get the right set of
options. You might also check the <a href="https://github.com/swig/swig/wiki">
SWIG Wiki</a> for additional information.</p>
<p><b> Important</b>
<br> If you are going to use optimisations turned on with gcc (for
example -O2), ensure you also compile with -fno-strict-aliasing. The
GCC optimisations have become more aggressive from gcc-4.0 onwards and
will result in code that fails with strict aliasing optimisations
turned on. See the <a href="#Java_typemaps_c_to_java_types">C/C++ to
Java typemaps</a> section for more details.</p>
<p> The name of the shared library output file is important. If the name
of your SWIG module is &quot;<tt>example</tt>&quot;, the name of the
corresponding shared library file should be &quot;<tt>libexample.so</tt>&quot;
(or equivalent depending on your machine, see <a href="#Java_dynamic_linking_problems">
Dynamic linking problems</a> for more information). The name of the
module is specified using the <tt>%module</tt> directive or <tt>-module</tt>
command line option.</p>
<h3><a name="Java_using_module">27.2.5 Using your module</a></h3>
<p> To load your shared native library module in Java, simply use Java's
<tt>System.loadLibrary</tt> method in a Java class:</p>
<div class="code">
<pre>
// runme.java
public class runme {
static {
System.loadLibrary(&quot;example&quot;);
}
public static void main(String argv[]) {
System.out.println(example.fact(4));
}
}
</pre>
</div>
<p> Compile all the Java files and run:</p>
<div class="code">
<pre>
$ javac *.java
$ java runme
24
$
</pre>
</div>
<p> If it doesn't work have a look at the following section which
discusses problems loading the shared library.</p>
<h3><a name="Java_dynamic_linking_problems">27.2.6 Dynamic linking
problems</a></h3>
<p> As shown in the previous section the code to load a native library
(shared library) is <tt>System.loadLibrary(&quot;name&quot;)</tt>. This can fail
with an UnsatisfiedLinkError exception and can be due to a number of
reasons.</p>
<p> You may get an exception similar to this:</p>
<div class="code">
<pre>
$ java runme
Exception in thread &quot;main&quot; java.lang.UnsatisfiedLinkError: no example in java.library.path
at java.lang.ClassLoader.loadLibrary(ClassLoader.java:1312)
at java.lang.Runtime.loadLibrary0(Runtime.java:749)
at java.lang.System.loadLibrary(System.java:820)
at runme.&lt;clinit&gt;(runme.java:5)
</pre>
</div>
<p> The most common cause for this is an incorrect naming of the native
library for the name passed to the <tt>loadLibrary</tt> function. The
string passed to the <tt>loadLibrary</tt> function must not include the
file extension name in the string, that is<i> .dll</i> or<i> .so</i>.
The string must be<i> name</i> and not<i> libname</i> for all
platforms. On Windows the native library must then be called<i>
name.dll</i> and on most Unix systems it must be called<i> libname.so</i>
.</p>
<p> Another common reason for the native library not loading is because
it is not in your path. On Windows make sure the<i> path</i>
environment variable contains the path to the native library. On Unix
make sure that your<i> LD_LIBRARY_PATH</i> contains the path to the
native library. Adding paths to<i> LD_LIBRARY_PATH</i> can slow down
other programs on your system so you may want to consider alternative
approaches. For example you could recompile your native library with
extra path information using <tt>-rpath</tt> if you're using GNU, see
the GNU linker documentation (<tt>ld</tt> man page). You could use a
command such as <tt>ldconfig</tt> (Linux) or <tt>crle</tt> (Solaris) to
add additional search paths to the default system configuration (this
requires root access and you will need to read the man pages).</p>
<p> The native library will also not load if there are any unresolved
symbols in the compiled C/C++ code. The following exception is
indicative of this:</p>
<div class="code">
<pre>
$ java runme
Exception in thread &quot;main&quot; java.lang.UnsatisfiedLinkError: libexample.so: undefined
symbol: fact
at java.lang.ClassLoader$NativeLibrary.load(Native Method)
at java.lang.ClassLoader.loadLibrary0(ClassLoader.java, Compiled Code)
at java.lang.ClassLoader.loadLibrary(ClassLoader.java, Compiled Code)
at java.lang.Runtime.loadLibrary0(Runtime.java, Compiled Code)
at java.lang.System.loadLibrary(System.java, Compiled Code)
at runme.&lt;clinit&gt;(runme.java:5)
$
</pre>
</div>
<p> This error usually indicates that you forgot to include some object
files or libraries in the linking of the native library file. Make sure
you compile both the SWIG wrapper file and the code you are wrapping
into the native library file. If you forget to compile and link in the
SWIG wrapper file into your native library file, you will get a message
similar to the following:</p>
<div class="code">
<pre>
$ java runme
Exception in thread &quot;main&quot; java.lang.UnsatisfiedLinkError: exampleJNI.gcd(II)I
at exampleJNI.gcd(Native Method)
at example.gcd(example.java:12)
at runme.main(runme.java:18)
</pre>
</div>
<p> where <tt>gcd</tt> is the missing JNI function that SWIG generated
into the wrapper file. Also make sure you pass all of the required
libraries to the linker. The <tt>java -verbose:jni</tt> commandline
option is also a great way to get more information on unresolved
symbols. One last piece of advice is to beware of the common faux pas
of having more than one native library version in your path.</p>
<p> In summary, ensure that you are using the correct C/C++ compiler and
linker combination and options for successful native library loading.
If you are using the examples that ship with SWIG, then the
Examples/Makefile must have these set up correctly for your system. The
SWIG installation package makes a best attempt at getting these correct
but does not get it right 100% of the time. The <a href="https://github.com/swig/swig/wiki">
SWIG Wiki</a> also has some settings for commonly used compiler and
operating system combinations. The following section also contains some
C++ specific linking problems and solutions.</p>
<h3><a name="Java_compilation_problems_cpp">27.2.7 Compilation problems
and compiling with C++</a></h3>
<p> On most machines, shared library files should be linked using the
C++ compiler. For example:</p>
<div class="code">
<pre>
% swig -c++ -java example.i
% g++ -c -fPIC example.cxx
% g++ -c -fPIC example_wrap.cxx -I/usr/java/j2sdk1.4.1/include -I/usr/java/j2sdk1.4.1/include/linux
% g++ -shared example.o example_wrap.o -o libexample.so
</pre>
</div>
<p> In addition to this, you may need to include additional library
files to make it work. For example, if you are using the Sun C++
compiler on Solaris, you often need to add an extra library <tt>-lCrun</tt>
like this:</p>
<div class="code">
<pre>
% swig -c++ -java example.i
% CC -c example.cxx
% CC -c example_wrap.cxx -I/usr/java/include -I/usr/java/include/solaris
% CC -G example.o example_wrap.o -L/opt/SUNWspro/lib -o libexample.so -lCrun
</pre>
</div>
<p> If you aren't entirely sure about the linking for C++, you might
look at an existing C++ program. On many Unix machines, the <tt>ldd</tt>
command will list library dependencies. This should give you some clues
about what you might have to include when you link your shared library.
For example:</p>
<div class="code">
<pre>
$ ldd swig
libstdc++-libc6.1-1.so.2 =&gt; /usr/lib/libstdc++-libc6.1-1.so.2 (0x40019000)
libm.so.6 =&gt; /lib/libm.so.6 (0x4005b000)
libc.so.6 =&gt; /lib/libc.so.6 (0x40077000)
/lib/ld-linux.so.2 =&gt; /lib/ld-linux.so.2 (0x40000000)
$
</pre>
</div>
<p> Finally make sure the version of JDK header files matches the
version of Java that you are running as incompatibilities could lead to
compilation problems or unpredictable behaviour.</p>
<h3><a name="Java_building_windows">27.2.8 Building on Windows</a></h3>
<p> Building on Windows is roughly similar to the process used with
Unix. You will want to produce a DLL that can be loaded by the Java
Virtual Machine. This section covers the process of using SWIG with
Microsoft Visual C++ 6 although the procedure may be similar with other
compilers. In order for everything to work, you will need to have a JDK
installed on your machine in order to read the JNI header files.</p>
<h4><a name="Java_visual_studio">27.2.8.1 Running SWIG from Visual
Studio</a></h4>
<p> If you are developing your application within Microsoft Visual
studio, SWIG can be invoked as a custom build option. The Examples\java
directory has a few <a href="#Windows_examples">Windows Examples</a>
containing Visual Studio project (.dsp) files. The process to re-create
the project files for a C project are roughly:</p>
<ul>
<li>Open up a new workspace and use the AppWizard to select a DLL
project.</li>
<li>Add both the SWIG interface file (the .i file), any supporting C
files, and the name of the wrapper file that will be created by SWIG
(ie. <tt>example_wrap.c</tt>). Don't worry if the wrapper file doesn't
exist yet--Visual Studio will keep a reference to it.</li>
<li>Select the SWIG interface file and go to the settings menu. Under
settings, select the &quot;Custom Build&quot; option.</li>
<li>Enter &quot;SWIG&quot; in the description field.</li>
<li>Enter &quot;<tt>swig -java -o $(ProjDir)\$(InputName)_wrap.c $(InputPath)</tt>
&quot; in the &quot;Build command(s) field&quot;</li>
<li>Enter &quot;<tt>$(ProjDir)\$(InputName)_wrap.c</tt>&quot; in the &quot;Output
files(s) field&quot;.</li>
<li>Next, select the settings for the entire project and go to C/C++ tab
and select the Preprocessor category. Add the include directories to
the JNI header files under &quot;Additional include directories&quot;, eg
&quot;C:\jdk1.3\include, C:\jdk1.3\include\win32&quot;.</li>
<li>Next, select the settings for the entire project and go to Link tab
and select the General category. Set the name of the output file to
match the name of your Java module (ie. example.dll).</li>
<li>Next, select the example.c and example_wrap.c files and go to the
C/C++ tab and select the Precompiled Headers tab in the project
settings. Disabling precompiled headers for these files will overcome
any precompiled header errors while building.</li>
<li>Finally, add the java compilation as a post build rule in the
Post-build step tab in project settings, eg, &quot;c:\jdk1.3\bin\javac
*.java&quot;</li>
<li>Build your project.</li>
</ul>
<p> Note: If using C++, choose a C++ suffix for the wrapper file, for
example <tt>example_wrap.cxx</tt>. Use <tt>_wrap.cxx</tt> instead of <tt>
_wrap.c</tt> in the instructions above and add -c++ when invoking swig.</p>
<p> Now, assuming all went well, SWIG will be automatically invoked when
you build your project. When doing a build, any changes made to the
interface file will result in SWIG being automatically invoked to
produce a new version of the wrapper file.</p>
<p> The Java classes that SWIG output should also be compiled into
.class files. To run the native code in the DLL (example.dll), make
sure that it is in your path then run your Java program which uses it,
as described in the previous section. If the library fails to load have
a look at <a href="#Java_dynamic_linking_problems">Dynamic linking
problems</a>.</p>
<h4><a name="Java_nmake">27.2.8.2 Using NMAKE</a></h4>
<p> Alternatively, a Makefile for use by NMAKE can be written. Make sure
the environment variables for MSVC++ are available and the MSVC++ tools
are in your path. Now, just write a short Makefile like this :</p>
<div class="code">
<pre>
# Makefile for using SWIG and Java for C code
SRCS = example.c
IFILE = example
INTERFACE = $(IFILE).i
WRAPFILE = $(IFILE)_wrap.c
# Location of the Visual C++ tools (32 bit assumed)
TOOLS = c:\msdev
TARGET = example.dll
CC = $(TOOLS)\bin\cl.exe
LINK = $(TOOLS)\bin\link.exe
INCLUDE32 = -I$(TOOLS)\include
MACHINE = IX86
# C Library needed to build a DLL
DLLIBC = msvcrt.lib oldnames.lib
# Windows libraries that are apparently needed
WINLIB = kernel32.lib advapi32.lib user32.lib gdi32.lib comdlg32.lib winspool.lib
# Libraries common to all DLLs
LIBS = $(DLLIBC) $(WINLIB)
# Linker options
LOPT = -debug:full -debugtype:cv /NODEFAULTLIB /RELEASE /NOLOGO \
/MACHINE:$(MACHINE) -entry:_DllMainCRTStartup@12 -dll
# C compiler flags
CFLAGS = /Z7 /Od /c /nologo
JAVA_INCLUDE = -ID:\jdk1.3\include -ID:\jdk1.3\include\win32
java::
swig -java -o $(WRAPFILE) $(INTERFACE)
$(CC) $(CFLAGS) $(JAVA_INCLUDE) $(SRCS) $(WRAPFILE)
set LIB=$(TOOLS)\lib
$(LINK) $(LOPT) -out:example.dll $(LIBS) example.obj example_wrap.obj
javac *.java
</pre>
</div>
<p> To build the DLL and compile the java code, run NMAKE (you may need
to run <tt>vcvars32</tt> first). This is a pretty simplistic Makefile,
but hopefully it's enough to get you started. Of course you may want to
make changes for it to work for C++ by adding in the -c++ command line
option for swig and replacing .c with .cxx.</p>
<h2><a name="Java_basic_tour">27.3 A tour of basic C/C++ wrapping</a></h2>
<p> By default, SWIG attempts to build a natural Java interface to your
C/C++ code. Functions are wrapped as functions, classes are wrapped as
classes, variables are wrapped with JavaBean type getters and setters
and so forth. This section briefly covers the essential aspects of this
wrapping.</p>
<h3><a name="Java_module_packages_classes">27.3.1 Modules, packages and
generated Java classes</a></h3>
<p> The SWIG <tt>%module</tt> directive specifies the name of the Java
module. When you specify `<tt>%module example</tt>', the<i> module name</i>
determines the name of some of the generated files in the module. The
generated code consists of a<i> module class</i> file <tt>example.java</tt>
, an<i> intermediary JNI class</i> file, <tt>exampleJNI.java</tt> as
well as numerous other Java<i> proxy class</i> files. Each proxy class
is named after the structs, unions and classes you are wrapping. You
may also get a<i> constants interface</i> file if you are wrapping any
unnamed enumerations or constants, for example <tt>
exampleConstants.java</tt>. When choosing a module name, make sure you
don't use the same name as one of the generated proxy class files nor a
Java keyword. Sometimes a C/C++ type cannot be wrapped by a proxy
class, for example a pointer to a primitive type. In these situations a<i>
type wrapper class</i> is generated. Wrapping an enum generates an<i>
enum class</i>, either a proper Java enum or a Java class that
simulates the enums pattern. Details of all these generated classes
will unfold as you read this section.</p>
<p> The JNI (C/C++) code is generated into a file which also contains
the module name, for example <tt>example_wrap.cxx</tt> or <tt>
example_wrap.c</tt>. These C or C++ files complete the contents of the
module.</p>
<p> The generated Java classes can be placed into a Java package by
using the <tt>-package</tt> commandline option. This is often combined
with the <tt>-outdir</tt> to specify a package directory for generating
the Java files.</p>
<div class="code">
<pre>
swig -java -package com.bloggs.swig -outdir com/bloggs/swig example.i
</pre>
</div>
<p> SWIG won't create the directory, so make sure it exists beforehand.</p>
<h3><a name="Java_functions">27.3.2 Functions</a></h3>
<p> There is no such thing as a global Java function so global C
functions are wrapped as static methods in the module class. For
example,</p>
<div class="code">
<pre>
%module example
int fact(int n);
</pre>
</div>
<p> creates a static function that works exactly like you think it
might:</p>
<div class="code">
<pre>
public class example {
public static int fact(int n) {
// makes call using JNI to the C function
}
}
</pre>
</div>
<p> The Java class <tt>example</tt> is the<i> module class</i>. The
function can be used as follows from Java:</p>
<div class="code">
<pre>
System.out.println(example.fact(4));
</pre>
</div>
<h3><a name="Java_global_variables">27.3.3 Global variables</a></h3>
<p> C/C++ global variables are fully supported by SWIG. Java does not
allow the overriding of the dot operator so all variables are accessed
through getters and setters. Again because there is no such thing as a
Java global variable, access to C/C++ global variables is done through
static getter and setter functions in the module class.</p>
<div class="code">
<pre>
// SWIG interface file with global variables
%module example
...
%inline %{
extern int My_variable;
extern double density;
%}
...
</pre>
</div>
<p> Now in Java :</p>
<div class="code">
<pre>
// Print out value of a C global variable
System.out.println(&quot;My_variable = &quot; + example.getMy_variable());
// Set the value of a C global variable
example.setDensity(0.8442);
</pre>
</div>
<p> The value returned by the getter will always be up to date even if
the value is changed in C. Note that the getters and setters produced
follow the JavaBean property design pattern. That is the first letter
of the variable name is capitalized and preceded with set or get. If
you have the misfortune of wrapping two variables that differ only in
the capitalization of their first letters, use %rename to change one of
the variable names. For example:</p>
<div class="code">
<pre>
%rename Clash RenamedClash;
float Clash;
int clash;
</pre>
</div>
<p> If a variable is declared as <tt>const</tt>, it is wrapped as a
read-only variable. That is only a getter is produced.</p>
<p> To make ordinary variables read-only, you can use the <tt>%immutable</tt>
directive. For example:</p>
<div class="code">
<pre>
%{
extern char *path;
%}
%immutable;
extern char *path;
%mutable;
</pre>
</div>
<p> The <tt>%immutable</tt> directive stays in effect until it is
explicitly disabled or cleared using <tt>%mutable</tt>. See the <a href="#SWIG_readonly_variables">
Creating read-only variables</a> section for further details.</p>
<p> If you just want to make a specific variable immutable, supply a
declaration name. For example:</p>
<div class="code">
<pre>
%{
extern char *path;
%}
%immutable path;
...
extern char *path; // Read-only (due to %immutable)
</pre>
</div>
<h3><a name="Java_constants">27.3.4 Constants</a></h3>
<p> C/C++ constants are wrapped as Java static final variables. To
create a constant, use <tt>#define</tt> or the <tt>%constant</tt>
directive. For example:</p>
<div class="code">
<pre>
#define PI 3.14159
#define VERSION &quot;1.0&quot;
%constant int FOO = 42;
%constant const char *path = &quot;/usr/local&quot;;
</pre>
</div>
<p> By default the generated static final variables are initialized by
making a JNI call to get their value. The constants are generated into
the constants interface and look like this:</p>
<div class="code">
<pre>
public interface exampleConstants {
public final static double PI = exampleJNI.PI_get();
public final static String VERSION = exampleJNI.VERSION_get();
public final static int FOO = exampleJNI.FOO_get();
public final static String path = exampleJNI.path_get();
}
</pre>
</div>
<p> Note that SWIG has inferred the C type and used an appropriate Java
type that will fit the range of all possible values for the C type. By
default SWIG generates<b> runtime constants</b>. They are not<b>
compiler constants</b> that can, for example, be used in a switch
statement. This can be changed by using the <tt>%javaconst(flag)</tt>
directive. It works like all the other <a href="#Customization_features">
%feature directives</a>. The default is <tt>%javaconst(0)</tt>. It is
possible to initialize all wrapped constants from pure Java code by
placing a <tt>%javaconst(1)</tt><b> before</b> SWIG parses the
constants. Putting it at the top of your interface file would ensure
this. Here is an example:</p>
<div class="code">
<pre>
%javaconst(1);
%javaconst(0) BIG;
%javaconst(0) LARGE;
#define EXPRESSION (0x100+5)
#define BIG 1000LL
#define LARGE 2000ULL
</pre>
</div>
<p> generates:</p>
<div class="code">
<pre>
public interface exampleConstants {
public final static int EXPRESSION = (0x100+5);
public final static long BIG = exampleJNI.BIG_get();
public final static java.math.BigInteger LARGE = exampleJNI.LARGE_get();
}
</pre>
</div>
<p> Note that SWIG has inferred the C <tt>long long</tt> type from <tt>
BIG</tt> and used an appropriate Java type (<tt>long</tt>) as a Java <tt>
long</tt> is the smallest sized Java type that will take all possible
values for a C <tt>long long</tt>. Similarly for <tt>LARGE</tt>.</p>
<p> Be careful using the <tt>%javaconst(1)</tt> directive as not all C
code will compile as Java code. For example neither the <tt>1000LL</tt>
value for <tt>BIG</tt> nor <tt>2000ULL</tt> for <tt>LARGE</tt> above
would generate valid Java code. The example demonstrates how you can
target particular constants (<tt>BIG</tt> and <tt>LARGE</tt>) with <tt>
%javaconst</tt>. SWIG doesn't use <tt>%javaconst(1)</tt> as the default
as it tries to generate code that will always compile. However, using a
<tt>%javaconst(1)</tt> at the top of your interface file is strongly
recommended as the preferred compile time constants will be generated
and most C constants will compile as Java code and in any case the odd
constant that doesn't can be fixed using <tt>%javaconst(0)</tt>.</p>
<p> There is an alternative directive which can be used for these rare
constant values that won't compile as Java code. This is the <tt>
%javaconstvalue(value)</tt> directive, where <tt>value</tt> is a Java
code replacement for the C constant and can be either a string or a
number. This is useful if you do not want to use either the parsed C
value nor a JNI call, such as when the C parsed value will not compile
as Java code and a compile time constant is required. The same example
demonstrates this:</p>
<div class="code">
<pre>
%javaconst(1);
%javaconstvalue(&quot;new java.math.BigInteger(\&quot;2000\&quot;)&quot;) LARGE;
%javaconstvalue(1000) BIG;
#define EXPRESSION (0x100+5)
#define BIG 1000LL
#define LARGE 2000ULL
</pre>
</div>
<p> Note the string quotes for <tt>&quot;2000&quot;</tt> are escaped. The
following is then generated:</p>
<div class="code">
<pre>
public interface exampleConstants {
public final static int EXPRESSION = (0x100+5);
public final static long BIG = 1000;
public final static java.math.BigInteger LARGE = new java.math.BigInteger(&quot;2000&quot;);
}
</pre>
</div>
<p> Note: declarations declared as <tt>const</tt> are wrapped as
read-only variables and will be accessed using a getter as described in
the previous section. They are not wrapped as constants. The exception
to this rule are static const integral values defined within a
class/struct, where they are wrapped as constants, eg:.</p>
<div class="code">
<pre>
struct Maths {
static const int FIVE = 5;
};
</pre>
</div>
<p><b> Compatibility Note:</b> In SWIG-1.3.19 and earlier releases, the
constants were generated into the module class and the constants
interface didn't exist. Backwards compatibility is maintained as the
module class implements the constants interface (even though some
consider this type of interface implementation to be bad practice):</p>
<div class="code">
<pre>
public class example implements exampleConstants {
}
</pre>
</div>
<p> You thus have the choice of accessing these constants from either
the module class or the constants interface, for example, <tt>
example.EXPRESSION</tt> or <tt>exampleConstants.EXPRESSION</tt>. Or if
you decide this practice isn't so bad and your own class implements <tt>
exampleConstants</tt>, you can of course just use <tt>EXPRESSION</tt>.</p>
<h3><a name="Java_enumerations">27.3.5 Enumerations</a></h3>
<p> SWIG handles both named and unnamed (anonymous) enumerations. There
is a choice of approaches to wrapping named C/C++ enums. This is due to
historical reasons as SWIG's initial support for enums was limited and
Java did not originally have support for enums. Each approach has
advantages and disadvantages and it is important for the user to decide
which is the most appropriate solution. There are four approaches of
which the first is the default approach based on the so called Java
typesafe enum pattern. The second generates proper Java enums. The
final two approaches use simple integers for each enum item. Before
looking at the various approaches for wrapping named C/C++ enums,
anonymous enums are considered.</p>
<h4><a name="Java_anonymous_enums">27.3.5.1 Anonymous enums</a></h4>
<p> There is no name for anonymous enums and so they are handled like
constants. For example:</p>
<div class="code">
<pre>
enum { ALE, LAGER=10, STOUT, PILSNER, PILZ=PILSNER };
</pre>
</div>
<p> is wrapped into the constants interface, in a similar manner as
constants (see previous section):</p>
<div class="code">
<pre>
public interface exampleConstants {
public final static int ALE = exampleJNI.ALE_get();
public final static int LAGER = exampleJNI.LAGER_get();
public final static int STOUT = exampleJNI.STOUT_get();
public final static int PILSNER = exampleJNI.PILSNER_get();
public final static int PILZ = exampleJNI.PILZ_get();
}
</pre>
</div>
<p> The <tt>%javaconst(flag)</tt> and <tt>%javaconstvalue(value)</tt>
directive introduced in the previous section on constants can also be
used with enums. As is the case for constants, the default is <tt>
%javaconst(0)</tt> as not all C values will compile as Java code.
However, it is strongly recommended to add in a <tt>%javaconst(1)</tt>
directive at the top of your interface file as it is only on very rare
occasions that this will produce code that won't compile under Java.
Using <tt>%javaconst(1)</tt> will ensure compile time constants are
generated, thereby allowing the enum values to be used in Java switch
statements. Example usage:</p>
<div class="code">
<pre>
%javaconst(1);
%javaconst(0) PILSNER;
enum { ALE, LAGER=10, STOUT, PILSNER, PILZ=PILSNER };
</pre>
</div>
<p> generates:</p>
<div class="code">
<pre>
public interface exampleConstants {
public final static int ALE = 0;
public final static int LAGER = 10;
public final static int STOUT = LAGER + 1;
public final static int PILSNER = exampleJNI.PILSNER_get();
public final static int PILZ = PILSNER;
}
</pre>
</div>
<p> As in the case of constants, you can access them through either the
module class or the constants interface, for example, <tt>example.ALE</tt>
or <tt>exampleConstants.ALE</tt>.</p>
<h4><a name="Java_typesafe_enums">27.3.5.2 Typesafe enums</a></h4>
<p> This is the default approach to wrapping named enums. The typesafe
enum pattern is a relatively well known construct to work around the
lack of enums in versions of Java prior to JDK 1.5. It basically
defines a class for the enumeration and permits a limited number of
final static instances of the class. Each instance equates to an enum
item within the enumeration. The implementation is in the
&quot;enumtypesafe.swg&quot; file. Let's look at an example:</p>
<div class="code">
<pre>
%include &quot;enumtypesafe.swg&quot; // optional as typesafe enums are the default
enum Beverage { ALE, LAGER=10, STOUT, PILSNER, PILZ=PILSNER };
</pre>
</div>
<p>will generate:</p>
<div class="code">
<pre>
public final class Beverage {
public final static Beverage ALE = new Beverage(&quot;ALE&quot;);
public final static Beverage LAGER = new Beverage(&quot;LAGER&quot;, exampleJNI.LAGER_get());
public final static Beverage STOUT = new Beverage(&quot;STOUT&quot;);
public final static Beverage PILSNER = new Beverage(&quot;PILSNER&quot;);
public final static Beverage PILZ = new Beverage(&quot;PILZ&quot;, exampleJNI.PILZ_get());
[... additional support methods omitted for brevity ...]
}
</pre>
</div>
<p> See <a href="#Java_typesafe_enums_classes">Typesafe enum classes</a>
to see the omitted support methods. Note that the enum item with an
initializer (LAGER) is initialized with the enum value obtained via a
JNI call. However, as with anonymous enums and constants, use of the <tt>
%javaconst</tt> directive is strongly recommended to change this
behaviour:</p>
<div class="code">
<pre>
%include &quot;enumtypesafe.swg&quot; // optional as typesafe enums are the default
%javaconst(1);
enum Beverage { ALE, LAGER=10, STOUT, PILSNER, PILZ=PILSNER };
</pre>
</div>
<p> will generate:</p>
<div class="code">
<pre>
public final class Beverage {
public final static Beverage ALE = new Beverage(&quot;ALE&quot;);
public final static Beverage LAGER = new Beverage(&quot;LAGER&quot;, 10);
public final static Beverage STOUT = new Beverage(&quot;STOUT&quot;);
public final static Beverage PILSNER = new Beverage(&quot;PILSNER&quot;);
public final static Beverage PILZ = new Beverage(&quot;PILZ&quot;, PILSNER);
[... additional support methods omitted for brevity ...]
}
</pre>
</div>
<p> The generated code is easier to read and more efficient as a true
constant is used instead of a JNI call. As is the case for constants,
the default is <tt>%javaconst(0)</tt> as not all C values will compile
as Java code. However, it is recommended to add in a <tt>%javaconst(1)</tt>
directive at the top of your interface file as it is only on very rare
occasions that this will produce code that won't compile under Java.
The <tt>%javaconstvalue(value)</tt> directive can also be used for
typesafe enums. Note that global enums are generated into a Java class
within whatever package you are using. C++ enums defined within a C++
class are generated into a static final inner Java class within the
Java proxy class.</p>
<p> Typesafe enums have their advantages over using plain integers in
that they can be used in a typesafe manner. However, there are
limitations. For example, they cannot be used in switch statements and
serialization is an issue. Please look at the following references for
further information: <a href="http://java.sun.com/developer/Books/shiftintojava/page1.html#replaceenums">
Replace Enums with Classes</a> in<i> Effective Java Programming</i> on
the Sun website, <a href="https://www.javaworld.com/article/2076970/create-enumerated-constants-in-java.html">
Create enumerated constants in Java</a> JavaWorld article, <a href="https://www.javaworld.com/article/2077499/java-tip-133--more-on-typesafe-enums.html">
Java Tip 133: More on typesafe enums</a> and <a href="https://www.javaworld.com/article/2077487/java-tip-122--beware-of-java-typesafe-enumerations.html">
Java Tip 122: Beware of Java typesafe enumerations</a> JavaWorld tips.</p>
<p> Note that the syntax required for using typesafe enums is the same
as that for proper Java enums. This is useful during the period that a
project has to support legacy versions of Java. When upgrading to JDK
1.5 or later, proper Java enums could be used instead, without users
having to change their code. The following section details proper Java
enum generation.</p>
<h4><a name="Java_proper_enums">27.3.5.3 Proper Java enums</a></h4>
<p> Proper Java enums were only introduced in JDK 1.5 so this approach
is only compatible with more recent versions of Java. Java enums have
been designed to overcome all the limitations of both typesafe and type
unsafe enums and should be the choice solution, provided older versions
of Java do not have to be supported. In this approach, each named C/C++
enum is wrapped by a Java enum. Java enums, by default, do not support
enums with initializers. Java enums are in many respects similar to
Java classes in that they can be customised with additional methods.
SWIG takes advantage of this feature to facilitate wrapping C/C++ enums
that have initializers. In order to wrap all possible C/C++ enums using
proper Java enums, the &quot;enums.swg&quot; file must be used. Let's take a look
at an example.</p>
<div class="code">
<pre>
%include &quot;enums.swg&quot;
%javaconst(1);
enum Beverage { ALE, LAGER=10, STOUT, PILSNER, PILZ=PILSNER };
</pre>
</div>
<p> will generate:</p>
<div class="code">
<pre>
public enum Beverage {
ALE,
LAGER(10),
STOUT,
PILSNER,
PILZ(PILSNER);
[... additional support methods omitted for brevity ...]
}
</pre>
</div>
<p> See <a href="#Java_proper_enums_classes">Proper Java enum classes</a>
to see the omitted support methods. The generated Java enum has
numerous additional methods to support enums with initializers, such as
<tt>LAGER</tt> above. Note that as with the typesafe enum pattern, enum
items with initializers are by default initialized with the enum value
obtained via a JNI call. However, this is not the case above as we have
used the recommended <tt>%javaconst(1)</tt> to avoid the JNI call. The <tt>
%javaconstvalue(value)</tt> directive covered in the <a href="#Java_constants">
Constants</a> section can also be used for proper Java enums.</p>
<p> The additional support methods need not be generated if none of the
enum items have initializers and this is covered later in the <a href="#Java_simpler_enum_classes">
Simpler Java enums for enums without initializers</a> section.</p>
<h4><a name="Java_typeunsafe_enums">27.3.5.4 Type unsafe enums</a></h4>
<p> In this approach each enum item in a named enumeration is wrapped as
a static final integer in a class named after the C/C++ enum name. This
is a commonly used pattern in Java to simulate C/C++ enums, but it is
not typesafe. However, the main advantage over the typesafe enum
pattern is enum items can be used in switch statements. In order to use
this approach, the &quot;enumtypeunsafe.swg&quot; file must be used. Let's take a
look at an example.</p>
<div class="code">
<pre>
%include &quot;enumtypeunsafe.swg&quot;
%javaconst(1);
enum Beverage { ALE, LAGER=10, STOUT, PILSNER, PILZ=PILSNER };
</pre>
</div>
<p> will generate:</p>
<div class="code">
<pre>
public final class Beverage {
public final static int ALE = 0;
public final static int LAGER = 10;
public final static int STOUT = LAGER + 1;
public final static int PILSNER = STOUT + 1;
public final static int PILZ = PILSNER;
}
</pre>
</div>
<p> As is the case previously, the default is <tt>%javaconst(0)</tt> as
not all C/C++ values will compile as Java code. However, again it is
recommended to add in a <tt>%javaconst(1)</tt> directive. and the <tt>
%javaconstvalue(value)</tt> directive covered in the <a href="#Java_constants">
Constants</a> section can also be used for type unsafe enums. Note that
global enums are generated into a Java class within whatever package
you are using. C++ enums defined within a C++ class are generated into
a static final inner Java class within the Java proxy class.</p>
<p> Note that unlike typesafe enums, this approach requires users to
mostly use different syntax compared with proper Java enums. Thus the
upgrade path to proper enums provided in JDK 1.5 is more painful.</p>
<h4><a name="Java_simple_enums">27.3.5.5 Simple enums</a></h4>
<p> This approach is similar to the type unsafe approach. Each enum item
is also wrapped as a static final integer. However, these integers are
not generated into a class named after the C/C++ enum. Instead, global
enums are generated into the constants interface. Also, enums defined
in a C++ class have their enum items generated directly into the Java
proxy class rather than an inner class within the Java proxy class. In
fact, this approach is effectively wrapping the enums as if they were
anonymous enums and the resulting code is as per <a href="#Java_anonymous_enums">
anonymous enums</a>. The implementation is in the &quot;enumsimple.swg&quot; file.</p>
<p><b> Compatibility Note:</b> SWIG-1.3.21 and earlier versions wrapped
all enums using this approach. The type unsafe approach is preferable
to this one and this simple approach is only included for backwards
compatibility with these earlier versions of SWIG.</p>
<h3><a name="Java_pointers">27.3.6 Pointers</a></h3>
<p> C/C++ pointers are fully supported by SWIG. Furthermore, SWIG has no
problem working with incomplete type information. Here is a rather
simple interface:</p>
<div class="code">
<pre>
%module example
FILE *fopen(const char *filename, const char *mode);
int fputs(const char *, FILE *);
int fclose(FILE *);
</pre>
</div>
<p> When wrapped, you will be able to use the functions in a natural way
from Java. For example:</p>
<div class="code">
<pre>
SWIGTYPE_p_FILE f = example.fopen(&quot;junk&quot;, &quot;w&quot;);
example.fputs(&quot;Hello World\n&quot;, f);
example.fclose(f);
</pre>
</div>
<p> C pointers in the Java module are stored in a Java <tt>long</tt> and
cross the JNI boundary held within this 64 bit number. Many other SWIG
language modules use an encoding of the pointer in a string. These
scripting languages use the SWIG runtime type checker for dynamic type
checking as they do not support static type checking by a compiler. In
order to implement static type checking of pointers within Java, they
are wrapped by a simple Java class. In the example above the <tt>FILE *</tt>
pointer is wrapped with a<i> type wrapper class</i> called <tt>
SWIGTYPE_p_FILE</tt>.</p>
<p> Once obtained, a type wrapper object can be freely passed around to
different C functions that expect to receive an object of that type.
The only thing you can't do is dereference the pointer from Java. Of
course, that isn't much of a concern in this example.</p>
<p> As much as you might be inclined to modify a pointer value directly
from Java, don't. The value is not necessarily the same as the logical
memory address of the underlying object. The value will vary depending
on the native byte-ordering of the platform (i.e., big-endian vs.
little-endian). Most JVMs are 32 bit applications so any JNI code must
also be compiled as 32 bit. The net result is pointers in JNI code are
also 32 bits and are stored in the high order 4 bytes on big-endian
machines and in the low order 4 bytes on little-endian machines. By
design it is also not possible to manually cast a pointer to a new type
by using Java casts as it is particularly dangerous especially when
casting C++ objects. If you need to cast a pointer or change its value,
consider writing some helper functions instead. For example:</p>
<div class="code">
<pre>
%inline %{
/* C-style cast */
Bar *FooToBar(Foo *f) {
return (Bar *) f;
}
/* C++-style cast */
Foo *BarToFoo(Bar *b) {
return dynamic_cast&lt;Foo*&gt;(b);
}
Foo *IncrFoo(Foo *f, int i) {
return f+i;
}
%}
</pre>
</div>
<p> Also, if working with C++, you should always try to use the new C++
style casts. For example, in the above code, the C-style cast may
return a bogus result whereas as the C++-style cast will return a NULL
pointer if the conversion can't be performed.</p>
<h3><a name="Java_structures">27.3.7 Structures</a></h3>
<p> If you wrap a C structure, it is wrapped by a Java class with
getters and setters for access to the member variables. For example,</p>
<div class="code">
<pre>
struct Vector {
double x, y, z;
};
</pre>
</div>
<p> is used as follows:</p>
<div class="code">
<pre>
Vector v = new Vector();
v.setX(3.5);
v.setY(7.2);
double x = v.getX();
double y = v.getY();
</pre>
</div>
<p> The variable setters and getters are also based on the JavaBean
design pattern already covered under the Global variables section.
Similar access is provided for unions and the public data members of
C++ classes.</p>
<p> This object is actually an instance of a Java class that has been
wrapped around a pointer to the C structure. This instance doesn't
actually do anything--it just serves as a proxy. The pointer to the C
object is held in the Java proxy class in much the same way as pointers
are held by type wrapper classes. Further details about Java proxy
classes are covered a little later.</p>
<p> <tt>const</tt> members of a structure are read-only. Data members
can also be forced to be read-only using the <tt>%immutable</tt>
directive. For example:</p>
<div class="code">
<pre>
struct Foo {
...
%immutable;
int x; /* Read-only members */
char *name;
%mutable;
...
};
</pre>
</div>
<p> When <tt>char *</tt> members of a structure are wrapped, the
contents are assumed to be dynamically allocated using <tt>malloc</tt>
or <tt>new</tt> (depending on whether or not SWIG is run with the -c++
option). When the structure member is set, the old contents will be
released and a new value created. If this is not the behavior you want,
you will have to use a typemap (described later).</p>
<p> If a structure contains arrays, access to those arrays is managed
through pointers. For example, consider this:</p>
<div class="code">
<pre>
struct Bar {
int x[16];
};
</pre>
</div>
<p> If accessed in Java, you will see behavior like this:</p>
<div class="code">
<pre>
Bar b = new Bar();
SWIGTYPE_p_int x = b.getX();
</pre>
</div>
<p> This pointer can be passed around to functions that expect to
receive an <tt>int *</tt> (just like C). You can also set the value of
an array member using another pointer. For example:</p>
<div class="code">
<pre>
Bar b = new Bar();
SWIGTYPE_p_int x = b.getX();
Bar c = new Bar();
c.setX(x); // Copy contents of b.x to c.x
</pre>
</div>
<p> For array assignment (setters not getters), SWIG copies the entire
contents of the array starting with the data pointed to by <tt>b.x</tt>
. In this example, 16 integers would be copied. Like C, SWIG makes no
assumptions about bounds checking---if you pass a bad pointer, you may
get a segmentation fault or access violation. The default wrapping
makes it hard to set or get just one element of the array and so array
access from Java is somewhat limited. This can be changed easily though
by using the approach outlined later in the <a href="#Java_c_arrays">
Wrapping C arrays with Java arrays</a> and <a href="#Java_unbounded_c_arrays">
Unbounded C Arrays</a> sections.</p>
<p> When a member of a structure is itself a structure, it is handled as
a pointer. For example, suppose you have two structures like this:</p>
<div class="code">
<pre>
struct Foo {
int a;
};
struct Bar {
Foo f;
};
</pre>
</div>
<p> Now, suppose that you access the <tt>f</tt> member of <tt>Bar</tt>
like this:</p>
<div class="code">
<pre>
Bar b = new Bar();
Foo x = b.getF();
</pre>
</div>
<p> In this case, <tt>x</tt> is a pointer that points to the <tt>Foo</tt>
that is inside <tt>b</tt>. This is the same value as generated by this
C code:</p>
<div class="code">
<pre>
Bar b;
Foo *x = &amp;b-&gt;f; /* Points inside b */
</pre>
</div>
<p> Because the pointer points inside the structure, you can modify the
contents and everything works just like you would expect. For example:</p>
<div class="code">
<pre>
Bar b = new Bar();
b.getF().setA(3); // Modify b.f.a
Foo x = b.getF();
x.setA(3); // Modify x.a - this is the same as b.f.a
</pre>
</div>
<h3><a name="Java_classes">27.3.8 C++ classes</a></h3>
<p> C++ classes are wrapped by Java classes as well. For example, if you
have this class,</p>
<div class="code">
<pre>
class List {
public:
List();
~List();
int search(char *item);
void insert(char *item);
void remove(char *item);
char *get(int n);
int length;
};
</pre>
</div>
<p> you can use it in Java like this:</p>
<div class="code">
<pre>
List l = new List();
l.insert(&quot;Ale&quot;);
l.insert(&quot;Stout&quot;);
l.insert(&quot;Lager&quot;);
String item = l.get(2);
int length = l.getLength();
</pre>
</div>
<p> Class data members are accessed in the same manner as C structures.</p>
<p> Static class members are unsurprisingly wrapped as static members of
the Java class:</p>
<div class="code">
<pre>
class Spam {
public:
static void foo();
static int bar;
};
</pre>
</div>
<p> The static members work like any other Java static member:</p>
<div class="code">
<pre>
Spam.foo();
int bar = Spam.getBar();
</pre>
</div>
<h3><a name="Java_inheritance">27.3.9 C++ inheritance</a></h3>
<p> SWIG is fully aware of issues related to C++ inheritance. Therefore,
if you have classes like this</p>
<div class="code">
<pre>
class Foo {
...
};
class Bar : public Foo {
...
};
</pre>
</div>
<p> those classes are wrapped into a hierarchy of Java classes that
reflect the same inheritance structure:</p>
<div class="code">
<pre>
Bar b = new Bar();
Class c = b.getClass();
System.out.println(c.getSuperclass().getName());
</pre>
</div>
<p> will of course display:</p>
<div class="code">
<pre>
Foo
</pre>
</div>
<p> Furthermore, if you have functions like this</p>
<div class="code">
<pre>
void spam(Foo *f);
</pre>
</div>
<p> then the Java function <tt>spam()</tt> accepts instances of <tt>Foo</tt>
or instances of any other proxy classes derived from <tt>Foo</tt>.</p>
<p> Note that Java does not support multiple inheritance so any multiple
inheritance in the C++ code is not going to work. A warning is given
when multiple inheritance is detected and only the first base class is
used.</p>
<p> Private and protected methods are not wrapped as they are
inaccessible outside of the class. Protected methods can be made
accessible though via the directors feature, see the <a href="#Java_protected_virtual_methods">
Accessing virtual protected methods</a> in the directors section.</p>
<h3><a name="Java_pointers_refs_arrays">27.3.10 Pointers, references,
arrays and pass by value</a></h3>
<p> In C++, there are many different ways a function might receive and
manipulate objects. For example:</p>
<div class="code">
<pre>
void spam1(Foo *x); // Pass by pointer
void spam2(Foo &amp;x); // Pass by reference
void spam3(Foo x); // Pass by value
void spam4(Foo x[]); // Array of objects
</pre>
</div>
<p> In Java, there is no detailed distinction like this--specifically,
there are only instances of classes. There are no pointers nor
references. Because of this, SWIG unifies all of these types together
in the wrapper code. For instance, if you actually had the above
functions, it is perfectly legal to do this from Java:</p>
<div class="code">
<pre>
Foo f = new Foo(); // Create a Foo
example.spam1(f); // Ok. Pointer
example.spam2(f); // Ok. Reference
example.spam3(f); // Ok. Value.
example.spam4(f); // Ok. Array (1 element)
</pre>
</div>
<p> Similar behavior occurs for return values. For example, if you had
functions like this,</p>
<div class="code">
<pre>
Foo *spam5();
Foo &amp;spam6();
Foo spam7();
</pre>
</div>
<p> then all three functions will return a pointer to some <tt>Foo</tt>
object. Since the third function (spam7) returns a value, newly
allocated memory is used to hold the result and a pointer is returned
(Java will release this memory when the returned object's finalizer is
run by the garbage collector).</p>
<h4><a name="Java_null_pointers">27.3.10.1 Null pointers</a></h4>
<p> Working with null pointers is easy. A Java <tt>null</tt> can be used
whenever a method expects a proxy class or typewrapper class. However,
it is not possible to pass null to C/C++ functions that take parameters
by value or by reference. If you try you will get a
NullPointerException.</p>
<div class="code">
<pre>
example.spam1(null); // Pointer - ok
example.spam2(null); // Reference - NullPointerException
example.spam3(null); // Value - NullPointerException
example.spam4(null); // Array - ok
</pre>
</div>
<p> For <tt>spam1</tt> and <tt>spam4</tt> above the Java <tt>null</tt>
gets translated into a NULL pointer for passing to the C/C++ function.
The converse also occurs, that is, NULL pointers are translated into <tt>
null</tt> Java objects when returned from a C/C++ function.</p>
<h3><a name="Java_overloaded_functions">27.3.11 C++ overloaded functions</a>
</h3>
<p> C++ overloaded functions, methods, and constructors are mostly
supported by SWIG. For example, if you have two functions like this:</p>
<div class="code">
<pre>
%module example
void foo(int);
void foo(char *c);
</pre>
</div>
<p> You can use them in Java in a straightforward manner:</p>
<div class="code">
<pre>
example.foo(3); // foo(int)
example.foo(&quot;Hello&quot;); // foo(char *c)
</pre>
</div>
<p> Similarly, if you have a class like this,</p>
<div class="code">
<pre>
class Foo {
public:
Foo();
Foo(const Foo &amp;);
...
};
</pre>
</div>
<p> you can write Java code like this:</p>
<div class="code">
<pre>
Foo f = new Foo(); // Create a Foo
Foo g = new Foo(f); // Copy f
</pre>
</div>
<p> Overloading support is not quite as flexible as in C++. Sometimes
there are methods that SWIG cannot disambiguate as there can be more
than one C++ type mapping onto a single Java type. For example:</p>
<div class="code">
<pre>
void spam(int);
void spam(unsigned short);
</pre>
</div>
<p> Here both int and unsigned short map onto a Java int. Here is
another example:</p>
<div class="code">
<pre>
void foo(Bar *b);
void foo(Bar &amp;b);
</pre>
</div>
<p> If declarations such as these appear, you will get a warning message
like this:</p>
<div class="code">
<pre>
example.i:12: Warning 515: Overloaded method spam(unsigned short) ignored.
Method spam(int) at example.i:11 used.
</pre>
</div>
<p> To fix this, you either need to either <a href="#SWIG_rename_ignore">
rename or ignore</a> one of the methods. For example:</p>
<div class="code">
<pre>
%rename(spam_ushort) spam(unsigned short);
...
void spam(int);
void spam(unsigned short); // Now renamed to spam_ushort
</pre>
</div>
<p> or</p>
<div class="code">
<pre>
%ignore spam(unsigned short);
...
void spam(int);
void spam(unsigned short); // Ignored
</pre>
</div>
<h3><a name="Java_default_arguments">27.3.12 C++ default arguments</a></h3>
<p> Any function with a default argument is wrapped by generating an
additional function for each argument that is defaulted. For example,
if we have the following C++:</p>
<div class="code">
<pre>
%module example
void defaults(double d=10.0, int i=0);
</pre>
</div>
<p> The following methods are generated in the Java module class:</p>
<div class="code">
<pre>
public class example {
public static void defaults(double d, int i) { ... }
public static void defaults(double d) { ... }
public static void defaults() { ... }
}
</pre>
</div>
<p> It is as if SWIG had parsed three separate overloaded methods. The
same approach is taken for static methods, constructors and member
methods.</p>
<p><b> Compatibility note:</b> Versions of SWIG prior to SWIG-1.3.23
wrapped these with a single wrapper method and so the default values
could not be taken advantage of from Java. Further details on default
arguments and how to restore this approach are given in the more
general <a href="#SWIGPlus_default_args">Default arguments</a> section.</p>
<h3><a name="Java_namespaces">27.3.13 C++ namespaces</a></h3>
<p> SWIG is aware of named C++ namespaces and they can be mapped to Java
packages, however, the default wrapping flattens the namespaces,
effectively ignoring them. So by default, the namespace names do not
appear in the module nor do namespaces result in a module that is
broken up into submodules or packages. For example, if you have a file
like this,</p>
<div class="code">
<pre>
%module example
namespace foo {
int fact(int n);
struct Vector {
double x, y, z;
};
};
</pre>
</div>
<p> it works in Java as follows:</p>
<div class="code">
<pre>
int f = example.fact(3);
Vector v = new Vector();
v.setX(3.4);
double y = v.getY();
</pre>
</div>
<p> If your program has more than one namespace, name conflicts (if any)
can be resolved using <tt>%rename</tt> For example:</p>
<div class="code">
<pre>
%rename(Bar_spam) Bar::spam;
namespace Foo {
int spam();
}
namespace Bar {
int spam();
}
</pre>
</div>
<p> If you have more than one namespace and you want to keep their
symbols separate, consider wrapping them as separate SWIG modules. Each
SWIG module can be placed into a separate package.</p>
<p> The default behaviour described above can be improved via the <a href="#SWIGPlus_nspace">
nspace feature</a>. Note that it only works for classes, structs, unions
and enums declared within a named C++ namespace. When the nspace
feature is used (either the <tt>nspace</tt> feature flag or <tt>
%nspacemove</tt>), the C++ namespaces are converted into Java packages
of the same name. Proxy classes are thus declared within a package and
this proxy makes numerous calls to the JNI intermediary class which is
declared in the unnamed package by default. As Java does not support
types declared in a named package accessing types declared in an
unnamed package, the <tt>-package</tt> commandline option described
earlier generally should be used to provide a parent package. So if
SWIG is run using the <tt>-package com.myco</tt> option, a wrapped
class, <tt>MyWorld::Material::Color</tt>, can then be accessed as <tt>
com.myco.MyWorld.Material.Color</tt>. If you don't specify a package,
you will get the following warning:</p>
<div class="shell">
<pre>
example.i:16: Warning 826: The nspace feature is used on 'MyWorld::Material::Color' without -package. The generated code
may not compile as Java does not support types declared in a named package accessing types declared in an unnamed package.
</pre>
</div>
<p> If it is undesirable to have a single top level package, the nspace
feature may be used without the <tt>-package</tt> commandline option
(and the resulting warning ignored) if all of the types exposed using
SWIG are placed in a package using the nspace feature and the
'jniclasspackage' pragma is used to specify a package for the JNI
intermediary class.</p>
<p> If the resulting use of the nspace feature and hence packages
results in a proxy class in one package deriving or using a proxy class
from another package, you will need to open up the visibility for the
pointer constructor and <tt>getCPtr</tt> method from the default
'protected' to 'public' with the <tt>SWIG_JAVABODY_PROXY</tt> macro.
See <a href="#Java_code_typemaps">Java code typemaps</a>.</p>
<h3><a name="Java_templates">27.3.14 C++ templates</a></h3>
<p> C++ templates don't present a huge problem for SWIG. However, in
order to create wrappers, you have to tell SWIG to create wrappers for
a particular template instantiation. To do this, you use the <tt>
%template</tt> directive. For example:</p>
<div class="code">
<pre>
%module example
%{
#include &lt;utility&gt;
%}
template&lt;class T1, class T2&gt;
struct pair {
typedef T1 first_type;
typedef T2 second_type;
T1 first;
T2 second;
pair();
pair(const T1&amp;, const T2&amp;);
~pair();
};
%template(pairii) pair&lt;int, int&gt;;
</pre>
</div>
<p> In Java:</p>
<div class="code">
<pre>
pairii p = new pairii(3, 4);
int first = p.getFirst();
int second = p.getSecond();
</pre>
</div>
<p> Obviously, there is more to template wrapping than shown in this
example. More details can be found in the <a href="#SWIGPlus">SWIG and
C++</a> chapter.</p>
<h3><a name="Java_smart_pointers">27.3.15 C++ Smart Pointers</a></h3>
<h4><a name="Java_smart_pointers_shared_ptr">27.3.15.1 The shared_ptr
Smart Pointer</a></h4>
<p> The C++11 standard provides <tt>std::shared_ptr</tt> which was
derived from the Boost implementation, <tt>boost::shared_ptr</tt>. Both
of these are available for Java in the SWIG library and usage is
outlined in the <a href="#Library_std_shared_ptr">shared_ptr smart
pointer</a> library section.</p>
<h4><a name="Java_smart_pointers_generic">27.3.15.2 Generic Smart
Pointers</a></h4>
<p> In certain C++ programs, it is common to use classes that have been
wrapped by so-called &quot;smart pointers.&quot; Generally, this involves the use
of a template class that implements <tt>operator-&gt;()</tt> like this:</p>
<div class="code">
<pre>
template&lt;class T&gt; class SmartPtr {
...
T *operator-&gt;();
...
}
</pre>
</div>
<p> Then, if you have a class like this,</p>
<div class="code">
<pre>
class Foo {
public:
int x;
int bar();
};
</pre>
</div>
<p> A smart pointer would be used in C++ as follows:</p>
<div class="code">
<pre>
SmartPtr&lt;Foo&gt; p = CreateFoo(); // Created somehow (not shown)
...
p-&gt;x = 3; // Foo::x
int y = p-&gt;bar(); // Foo::bar
</pre>
</div>
<p> To wrap this in Java, simply tell SWIG about the <tt>SmartPtr</tt>
class and the low-level <tt>Foo</tt> object. Make sure you instantiate <tt>
SmartPtr</tt> using <tt>%template</tt> if necessary. For example:</p>
<div class="code">
<pre>
%module example
...
%template(SmartPtrFoo) SmartPtr&lt;Foo&gt;;
...
</pre>
</div>
<p> Now, in Java, everything should just &quot;work&quot;:</p>
<div class="code">
<pre>
SmartPtrFoo p = example.CreateFoo(); // Create a smart-pointer somehow
p.setX(3); // Foo::x
int y = p.bar(); // Foo::bar
</pre>
</div>
<p> If you ever need to access the underlying pointer returned by <tt>
operator-&gt;()</tt> itself, simply use the <tt>__deref__()</tt> method.
For example:</p>
<div class="code">
<pre>
Foo f = p.__deref__(); // Returns underlying Foo *
</pre>
</div>
<h2><a name="Java_further_details">27.4 Further details on the generated
Java classes</a></h2>
<p> In the previous section, a high-level view of Java wrapping was
presented. A key component of this wrapping is that structures and
classes are wrapped by Java proxy classes and type wrapper classes are
used in situations where no proxies are generated. This provides a very
natural, type safe Java interface to the C/C++ code and fits in with
the Java programming paradigm. However, a number of low-level details
were omitted. This section provides a brief overview of how the proxy
classes work and then covers the type wrapper classes. Finally enum
classes are covered. First, the crucial intermediary JNI class is
considered.</p>
<h3><a name="Java_imclass">27.4.1 The intermediary JNI class</a></h3>
<p> In the <a href="#SWIG">&quot;SWIG basics&quot;</a> and <a href="#SWIGPlus">
&quot;SWIG and C++&quot;</a> chapters, details of low-level structure and class
wrapping are described. To summarize those chapters, if you have a
global function and class like this</p>
<div class="code">
<pre>
class Foo {
public:
int x;
int spam(int num, Foo* foo);
};
void egg(Foo* chips);
</pre>
</div>
<p> then SWIG transforms the class into a set of low-level procedural
wrappers. These procedural wrappers essentially perform the equivalent
of this C++ code:</p>
<div class="code">
<pre>
Foo *new_Foo() {
return new Foo();
}
void delete_Foo(Foo *f) {
delete f;
}
int Foo_x_get(Foo *f) {
return f-&gt;x;
}
void Foo_x_set(Foo *f, int value) {
f-&gt;x = value;
}
int Foo_spam(Foo *f, int num, Foo* foo) {
return f-&gt;spam(num, foo);
}
</pre>
</div>
<p> These procedural function names don't actually exist, but their
functionality appears inside the generated JNI functions. The JNI
functions have to follow a particular naming convention so the function
names are actually:</p>
<div class="code">
<pre>
SWIGEXPORT jlong JNICALL Java_exampleJNI_new_1Foo(JNIEnv *jenv, jclass jcls);
SWIGEXPORT void JNICALL Java_exampleJNI_delete_1Foo(JNIEnv *jenv, jclass jcls,
jlong jarg1);
SWIGEXPORT void JNICALL Java_exampleJNI_Foo_1x_1set(JNIEnv *jenv, jclass jcls,
jlong jarg1, jobject jarg1_, jint jarg2);
SWIGEXPORT jint JNICALL Java_exampleJNI_Foo_1x_1get(JNIEnv *jenv, jclass jcls,
jlong jarg1, jobject jarg1_);
SWIGEXPORT jint JNICALL Java_exampleJNI_Foo_1spam(JNIEnv *jenv, jclass jcls,
jlong jarg1, jobject jarg1_, jint jarg2,
jlong jarg3, jobject jarg3_);
SWIGEXPORT void JNICALL Java_exampleJNI_egg(JNIEnv *jenv, jclass jcls,
jlong jarg1, jobject jarg1_);
</pre>
</div>
<p> For every JNI C function there has to be a static native Java
function. These appear in the intermediary JNI class:</p>
<div class="code">
<pre>
class exampleJNI {
public final static native long new_Foo();
public final static native void delete_Foo(long jarg1);
public final static native void Foo_x_set(long jarg1, Foo jarg1_, int jarg2);
public final static native int Foo_x_get(long jarg1, Foo jarg1_);
public final static native int Foo_spam(long jarg1, Foo jarg1_, int jarg2,
long jarg3, Foo jarg3_);
public final static native void egg(long jarg1, Foo jarg1_);
}
</pre>
</div>
<p> This class contains the complete Java - C/C++ interface so all
function calls go via this class. As this class acts as a go-between
for all JNI calls to C/C++ code from the Java <a href="#Java_proxy_classes">
proxy classes</a>, <a href="#Java_type_wrapper_classes">type wrapper
classes</a> and <a href="#Java_module_class">module class</a>, it is
known as the intermediary JNI class.</p>
<p> You may notice that SWIG uses a Java long wherever a pointer or
class object needs to be marshalled across the Java-C/C++ boundary.
This approach leads to minimal JNI code which makes for better
performance as JNI code involves a lot of string manipulation. SWIG
favours generating Java code over JNI code as Java code is compiled
into byte code and avoids the costly string operations needed in JNI
code. This approach has a downside though as the proxy class might get
collected before the native method has completed. You might notice
above that there is an additional parameters with a underscore postfix,
eg <tt>jarg1_</tt>. These are added in order to prevent <a href="#Java_pgcpp">
premature garbage collection when marshalling proxy classes</a>.</p>
<p> The functions in the intermediary JNI class cannot be accessed
outside of its package. Access to them is gained through the module
class for globals otherwise the appropriate proxy class.</p>
<a name="Java_module_directive"></a>
<p> The name of the intermediary JNI class can be changed from its
default, that is, the module name with JNI appended after it. The
module directive attribute <tt>jniclassname</tt> is used to achieve
this:</p>
<div class="code">
<pre>
%module(jniclassname=&quot;name&quot;) modulename
</pre>
</div>
<p> If <tt>name</tt> is the same as <tt>modulename</tt> then the module
class name gets changed from <tt>modulename</tt> to <tt>
modulenameModule</tt>.</p>
<h4><a name="Java_imclass_pragmas">27.4.1.1 The intermediary JNI class
pragmas</a></h4>
<p> The intermediary JNI class can be tailored through the use of
pragmas, but is not commonly done. The pragmas for this class are:</p>
<table BORDER summary="Intermediary JNI class pragmas">
<tr VALIGN="TOP"><td><b>Pragma</b></td><td><b>Description</b></td></tr>
<tr><td>jniclassbase</td><td>Base class for the intermediary JNI class</td>
</tr>
<tr><td>jniclasspackage</td><td>Package in which to place the
intermediary JNI class</td></tr>
<tr><td>jniclassclassmodifiers</td><td>Class modifiers and class type
for the intermediary JNI class</td></tr>
<tr><td>jniclasscode</td><td>Java code is copied verbatim into the
intermediary JNI class</td></tr>
<tr><td>jniclassimports</td><td>Java code, usually one or more import
statements, placed before the intermediary JNI class definition</td></tr>
<tr><td>jniclassinterfaces</td><td>Comma separated interface classes for
the intermediary JNI class</td></tr>
</table>
<p> The pragma code appears in the generated intermediary JNI class
where you would expect:</p>
<div class="code">
<pre>
[ jniclassimports pragma ]
[ jniclassclassmodifiers pragma ] jniclassname extends [ jniclassbase pragma ]
implements [ jniclassinterfaces pragma ] {
[ jniclasscode pragma ]
... SWIG generated native methods ...
}
</pre>
</div>
<p> The <tt>jniclasscode</tt> pragma is quite useful for adding in a
static block for loading the shared library / dynamic link library and
demonstrates how pragmas work:</p>
<div class="code">
<pre>
%pragma(java) jniclasscode=%{
static {
try {
System.loadLibrary(&quot;example&quot;);
} catch (UnsatisfiedLinkError e) {
System.err.println(&quot;Native code library failed to load. \n&quot; + e);
System.exit(1);
}
}
%}
</pre>
</div>
<p> Pragmas will take either <tt>&quot;&quot;</tt> or <tt>%{ %}</tt> as
delimiters. For example, let's change the intermediary JNI class access
to just the default package-private access.</p>
<div class="code">
<pre>
%pragma(java) jniclassclassmodifiers=&quot;class&quot;
</pre>
</div>
<p> All the methods in the intermediary JNI class will then not be
callable outside of the package as the method modifiers have been
changed from public access to default access. This is useful if you
want to prevent users calling these low level functions.</p>
<h3><a name="Java_module_class">27.4.2 The Java module class</a></h3>
<p> All global functions and variable getters/setters appear in the
module class. For our example, there is just one function:</p>
<div class="code">
<pre>
public class example {
public static void egg(Foo chips) {
exampleJNI.egg(Foo.getCPtr(chips), chips);
}
}
</pre>
</div>
<p> The module class is necessary as there is no such thing as a global
in Java so all the C globals are put into this class. They are
generated as static functions and so must be accessed as such by using
the module name in the static function call:</p>
<div class="code">
<pre>
example.egg(new Foo());
</pre>
</div>
<p> The primary reason for having the module class wrapping the calls in
the intermediary JNI class is to implement static type checking. In
this case only a <tt>Foo</tt> can be passed to the <tt>egg</tt>
function, whereas any <tt>long</tt> can be passed to the <tt>egg</tt>
function in the intermediary JNI class.</p>
<h4><a name="Java_module_class_pragmas">27.4.2.1 The Java module class
pragmas</a></h4>
<p> The module class can be tailored through the use of pragmas, in the
same manner as the intermediary JNI class. The pragmas are similarly
named and are used in the same way. The complete list follows:</p>
<table BORDER summary="Java module class pragmas">
<tr VALIGN="TOP"><td><b>Pragma</b></td><td><b>Description</b></td></tr>
<tr><td>modulebase</td><td>Base class for the module class</td></tr>
<tr><td>moduleclassmodifiers</td><td>Class modifiers and class type for
the module class</td></tr>
<tr><td>modulecode</td><td>Java code is copied verbatim into the module
class</td></tr>
<tr><td>moduleimports</td><td>Java code, usually one or more import
statements, placed before the module class definition</td></tr>
<tr><td>moduleinterfaces</td><td>Comma separated interface classes for
the module class</td></tr>
</table>
<p> The pragma code appears in the generated module class like this:</p>
<div class="code">
<pre>
[ moduleimports pragma ]
[ modulemodifiers pragma ] modulename extends [ modulebase pragma ]
implements [ moduleinterfaces pragma ] {
[ modulecode pragma ]
... SWIG generated wrapper functions ...
}
</pre>
</div>
<p> See <a href="#Java_imclass_pragmas">The intermediary JNI class
pragmas</a> section for further details on using pragmas.</p>
<h3><a name="Java_constants_interface">27.4.3 The Java constants
interface</a></h3>
<p> C/C++ constants are generated as final static members in a constants
interface as mentioned in the <a href="#Java_constants">Constants</a>
section, such as the example in this section:</p>
<div class="code">
<pre>
public interface exampleConstants {
public final static int EXPRESSION = (0x100+5);
public final static long BIG = exampleJNI.BIG_get();
public final static java.math.BigInteger LARGE = exampleJNI.LARGE_get();
}
</pre>
</div>
<p> C/C++ enums can also be generated into the same constants interface
as described in the <a href="#Java_enumerations">Enumerations</a>
section.</p>
<h4><a name="Java_constants_interface_pragmas">27.4.3.1 The Java
constants interface pragmas</a></h4>
<p> Scope for tailoring the generated interface is limited to one
pragma, in the same manner as the intermediary JNI class pragmas and
module class pragmas. The pragma details are:</p>
<table BORDER summary="Java constants interface pragmas">
<tr VALIGN="TOP"><td><b>Pragma</b></td><td><b>Description</b></td></tr>
<tr><td>constantsmodifiers</td><td>Class modifiers and class type for
the constants interface, defaults to <tt>public interface</tt></td></tr>
</table>
<p> The pragma code appears in the generated constants interface like
this:</p>
<div class="code">
<pre>
[ constantsmodifiers pragma ] constantsname {
... SWIG generated final static variables ...
}
</pre>
</div>
<p> where <tt>constantsname</tt> is a name created by concatenating the
module name and <tt>Constants</tt>, for example, <tt>exampleConstants</tt>
for a module named <tt>example</tt>.</p>
<p> The only real use for this pragma is to change the visibility from
public to default with:</p>
<div class="code">
<pre>
%pragma(java) constantsmodifiers=&quot;interface&quot;
</pre>
</div>
<h3><a name="Java_proxy_classes">27.4.4 Java proxy classes</a></h3>
<p> A Java proxy class is generated for each structure, union or C++
class that is wrapped. Proxy classes have also been called <a href="http://java.sun.com/docs/books/jni/html/stubs.html">
peer classes</a>. The default proxy class for our previous example looks
like this:</p>
<div class="code">
<pre>
public class Foo {
private transient long swigCPtr;
protected transient boolean swigCMemOwn;
protected Foo(long cPtr, boolean cMemoryOwn) {
swigCMemOwn = cMemoryOwn;
swigCPtr = cPtr;
}
protected static long getCPtr(Foo obj) {
return (obj == null) ? 0 : obj.swigCPtr;
}
protected void finalize() {
delete();
}
public synchronized void delete() {
if(swigCPtr != 0 &amp;&amp; swigCMemOwn) {
swigCMemOwn = false;
exampleJNI.delete_Foo(swigCPtr);
}
swigCPtr = 0;
}
public void setX(int value) {
exampleJNI.Foo_x_set(swigCPtr, this, value);
}
public int getX() {
return exampleJNI.Foo_x_get(swigCPtr, this);
}
public int spam(int num, Foo foo) {
return exampleJNI.Foo_spam(swigCPtr, this, num, Foo.getCPtr(foo), foo);
}
public Foo() {
this(exampleJNI.new_Foo(), true);
}
}
</pre>
</div>
<p> This class merely holds a pointer to the underlying C++ object (<tt>
swigCPtr</tt>). It also contains all the methods in the C++ class it is
proxying plus getters and setters for public member variables. These
functions call the native methods in the intermediary JNI class. The
advantage of having this extra layer is the type safety that the proxy
class functions offer. It adds static type checking which leads to
fewer surprises at runtime. For example, you can see that if you
attempt to use the <tt>spam()</tt> function it will only compile when
the parameters passed are an <tt>int</tt> and a <tt>Foo</tt>. From a
user's point of view, it makes the class work as if it were a Java
class:</p>
<div class="code">
<pre>
Foo f = new Foo();
f.setX(3);
int y = f.spam(5, new Foo());
</pre>
</div>
<h4><a name="Java_memory_management">27.4.4.1 Memory management</a></h4>
<p> Each proxy class has an ownership flag <tt>swigCMemOwn</tt>. The
value of this flag determines who is responsible for deleting the
underlying C++ object. If set to <tt>true</tt>, the proxy class's
finalizer will destroy the C++ object when the proxy class is garbage
collected. If set to false, then the destruction of the proxy class has
no effect on the C++ object.</p>
<p> When an object is created by a constructor or returned by value,
Java automatically takes ownership of the result. On the other hand,
when pointers or references are returned to Java, there is often no way
to know where they came from. Therefore, the ownership is set to false.
For example:</p>
<div class="code">
<pre>
class Foo {
public:
Foo();
Foo bar1();
Foo &amp;bar2();
Foo *bar2();
};
</pre>
</div>
<p> In Java:</p>
<div class="code">
<pre>
Foo f = new Foo(); // f.swigCMemOwn = true
Foo f1 = f.bar1(); // f1.swigCMemOwn = true
Foo f2 = f.bar2(); // f2.swigCMemOwn = false
Foo f3 = f.bar3(); // f3.swigCMemOwn = false
</pre>
</div>
<p> This behavior for pointers and references is especially important
for classes that act as containers. For example, if a method returns a
pointer to an object that is contained inside another object, you
definitely don't want Java to assume ownership and destroy it!</p>
<p> For the most part, memory management issues remain hidden. However,
there are situations where you might have to manually change the
ownership of an object. For instance, consider code like this:</p>
<div class="code">
<pre>
class Obj {};
class Node {
Obj *value;
public:
void set_value(Obj *v) { value = v; }
};
</pre>
</div>
<p> Now, consider the following Java code:</p>
<div class="code">
<pre>
Node n = new Node(); // Create a node
{
Obj o = new Obj(); // Create an object
n.set_value(o); // Set value
} // o goes out of scope
</pre>
</div>
<p> In this case, the Node <tt>n</tt> is holding a reference to <tt>o</tt>
internally. However, SWIG has no way to know that this has occurred.
The Java proxy class still thinks that it has ownership of <tt>o</tt>.
As <tt>o</tt> has gone out of scope, it could be garbage collected in
which case the C++ destructor will be invoked and <tt>n</tt> will then
be holding a stale-pointer to <tt>o</tt>. If you're lucky, you will
only get a segmentation fault.</p>
<p> To work around this, the ownership flag of <tt>o</tt> needs changing
to <tt>false</tt>. The ownership flag is a private member variable of
the proxy class so this is not possible without some customization of
the proxy class. This can be achieved by using a typemap to customise
the proxy class with pure Java code as detailed later in the section on
<a href="#Java_typemaps">Java typemaps</a>.</p>
<p> Sometimes a function will create memory and return a pointer to a
newly allocated object. SWIG has no way of knowing this so by default
the proxy class does not manage the returned object. However, you can
tell the proxy class to manage the memory if you specify the <tt>
%newobject</tt> directive. Consider:</p>
<div class="code">
<pre>
class Obj {...};
class Factory {
public:
static Obj *createObj() { return new Obj(); }
};
</pre>
</div>
<p> If we call the factory function, then we have to manually delete the
memory:</p>
<div class="code">
<pre>
Obj obj = Factory.createObj(); // obj.swigCMemOwn = false
...
obj.delete();
</pre>
</div>
<p> Now add in the %newobject directive:</p>
<div class="code">
<pre>
%newobject Factory::createObj();
class Obj {...};
class Factory {
public:
static Obj *createObj() { return new Obj(); }
};
</pre>
</div>
<p> A call to <tt>delete()</tt> is no longer necessary as the garbage
collector will make the C++ destructor call because <tt>swigCMemOwn</tt>
is now true.</p>
<div class="code">
<pre>
Obj obj = Factory.createObj(); // obj.swigCMemOwn = true;
...
</pre>
</div>
<p> Some memory management issues are quite tricky to fix and may only
be noticeable after using for a long time. One such issue is premature
garbage collection of an object created from Java and resultant usage
from C++ code. The section on typemap examples cover two such
scenarios, <a href="#Java_memory_management_objects">Memory management
for objects passed to the C++ layer</a> and <a href="#Java_memory_management_member_variables">
Memory management when returning references to member variables</a></p>
<h4><a name="Java_inheritance_mirroring">27.4.4.2 Inheritance</a></h4>
<p> Java proxy classes will mirror C++ inheritance chains. For example,
given the base class <tt>Base</tt> and its derived class <tt>Derived</tt>
:</p>
<div class="code">
<pre>
class Base {
public:
virtual double foo();
};
class Derived : public Base {
public:
virtual double foo();
};
</pre>
</div>
<p> The base class is generated much like any other proxy class seen so
far:</p>
<div class="code">
<pre>
public class Base {
private transient long swigCPtr;
protected transient boolean swigCMemOwn;
protected Base(long cPtr, boolean cMemoryOwn) {
swigCMemOwn = cMemoryOwn;
swigCPtr = cPtr;
}
protected static long getCPtr(Base obj) {
return (obj == null) ? 0 : obj.swigCPtr;
}
protected void finalize() {
delete();
}
public synchronized void delete() {
if(swigCPtr != 0 &amp;&amp; swigCMemOwn) {
swigCMemOwn = false;
exampleJNI.delete_Base(swigCPtr);
}
swigCPtr = 0;
}
public double foo() {
return exampleJNI.Base_foo(swigCPtr, this);
}
public Base() {
this(exampleJNI.new_Base(), true);
}
}
</pre>
</div>
<p> The <tt>Derived</tt> class extends <tt>Base</tt> mirroring the C++
class inheritance hierarchy.</p>
<div class="code">
<pre>
public class Derived extends Base {
private transient long swigCPtr;
protected Derived(long cPtr, boolean cMemoryOwn) {
super(exampleJNI.SWIGDerivedUpcast(cPtr), cMemoryOwn);
swigCPtr = cPtr;
}
protected static long getCPtr(Derived obj) {
return (obj == null) ? 0 : obj.swigCPtr;
}
protected void finalize() {
delete();
}
public synchronized void delete() {
if(swigCPtr != 0 &amp;&amp; swigCMemOwn) {
swigCMemOwn = false;
exampleJNI.delete_Derived(swigCPtr);
}
swigCPtr = 0;
super.delete();
}
public double foo() {
return exampleJNI.Derived_foo(swigCPtr, this);
}
public Derived() {
this(exampleJNI.new_Derived(), true);
}
}
</pre>
</div>
<p> Note the memory ownership is controlled by the base class. However
each class in the inheritance hierarchy has its own pointer value which
is obtained during construction. The <tt>SWIGDerivedUpcast()</tt> call
converts the pointer from a <tt>Derived *</tt> to a <tt>Base *</tt>.
This is a necessity as C++ compilers are free to implement pointers in
the inheritance hierarchy with different values.</p>
<p> It is of course possible to extend <tt>Base</tt> using your own Java
classes. If <tt>Derived</tt> is provided by the C++ code, you could for
example add in a pure Java class <tt>Extended</tt> derived from <tt>
Base</tt>. There is a caveat and that is any C++ code will not know
about your pure Java class <tt>Extended</tt> so this type of derivation
is restricted. However, true cross language polymorphism can be
achieved using the <a href="#Java_directors">directors</a> feature.</p>
<h4><a name="Java_proxy_classes_gc">27.4.4.3 Proxy classes and garbage
collection</a></h4>
<p> By default each proxy class has a <tt>delete()</tt> and a <tt>
finalize()</tt> method. The <tt>finalize()</tt> method calls <tt>
delete()</tt> which frees any malloc'd memory for wrapped C structs or
calls the C++ class destructors. The idea is for <tt>delete()</tt> to
be called when you have finished with the C/C++ object. Ideally you
need not call <tt>delete()</tt>, but rather leave it to the garbage
collector to call it from the finalizer. When a program exits, the
garbage collector does not guarantee to call all finalizers. An insight
into the reasoning behind this can be obtained from <a href="https://www.hpl.hp.com/techreports/2002/HPL-2002-335.html">
Hans Boehm's Destructors, Finalizers, and Synchronization</a> paper.
Depending on what the finalizers do and which operating system you use,
this may or may not be a problem.</p>
<p> If the <tt>delete()</tt> call into JNI code is just for memory
handling, there is not a problem when run on most operating systems,
for example Windows and Unix. Say your JNI code creates memory on the
heap which your finalizers should clean up, the finalizers may or may
not be called before the program exits. In Windows and Unix all memory
that a process uses is returned to the system on exit, so this isn't a
problem. This is not the case in some operating systems like vxWorks.
If however, your finalizer calls into JNI code invoking the C++
destructor which in turn releases a TCP/IP socket for example, there is
no guarantee that it will be released. Note that with long running
programs the garbage collector will eventually run, thereby calling any
unreferenced object's finalizers.</p>
<p> Some not so ideal solutions are:</p>
<ol>
<li>
<p> Call the <tt>System.runFinalizersOnExit(true)</tt> or <tt>
Runtime.getRuntime().runFinalizersOnExit(true)</tt> to ensure the
finalizers are called before the program exits. The catch is that this
is a deprecated function call as the documentation says:</p>
<div class="code"><i> This method is inherently unsafe. It may result in
finalizers being called on live objects while other threads are
concurrently manipulating those objects, resulting in erratic behavior
or deadlock.</i></div>
<p>In many cases you will be lucky and find that it works, but it is not
to be advocated. Have a look at <a href="https://www.oracle.com/technetwork/java/index.html">
Java web site</a> and search for <tt>runFinalizersOnExit</tt>.</p>
</li>
<li>
<p> From jdk1.3 onwards a new function, <tt>addShutdownHook()</tt>, was
introduced which is guaranteed to be called when your program exits.
You can encourage the garbage collector to call the finalizers, for
example, add this static block to the class that has the <tt>main()</tt>
function:</p>
<div class="code">
<pre>
static {
Runtime.getRuntime().addShutdownHook(
new Thread() {
public void run() { System.gc(); System.runFinalization(); }
}
);
}
</pre>
</div>
<p>Although this usually works, the documentation doesn't guarantee that
<tt>runFinalization()</tt> will actually call the finalizers. As the
shutdown hook is guaranteed you could also make a JNI call to clean up
any resources that are being tracked by the C/C++ code.</p>
</li>
<li>
<p>Call the <tt>delete()</tt> function manually which will immediately
invoke the C++ destructor. As a suggestion it may be a good idea to set
the object to null so that should the object be inadvertently used
again a Java null pointer exception is thrown, the alternative would
crash the JVM by using a null C pointer. For example given a SWIG
generated class A:</p>
<div class="code">
<pre>
A myA = new A();
// use myA ...
myA.delete();
// any use of myA here would crash the JVM
myA=null;
// any use of myA here would cause a Java null pointer exception to be thrown
</pre>
</div>
<p> The SWIG generated code ensures that the memory is not deleted
twice, in the event the finalizers get called in addition to the manual
<tt>delete()</tt> call.</p>
</li>
<li>
<p> Write your own object manager in Java. You could derive all SWIG
classes from a single base class which could track which objects have
had their finalizers run, then call the rest of them on program
termination. The section on <a href="#Java_typemaps">Java typemaps</a>
details how to specify a pure Java base class.</p>
</li>
</ol>
<p> See the <a href="http://www.devx.com/Java/Article/30192">How to
Handle Java Finalization's Memory-Retention Issues</a> article for
alternative approaches to managing memory by avoiding finalizers
altogether.</p>
<h4><a name="Java_pgcpp">27.4.4.4 The premature garbage collection
prevention parameter for proxy class marshalling</a></h4>
<p> As covered earlier, the C/C++ struct/class pointer is stored in the
proxy class as a Java long and when needed is passed into the native
method where it is cast into the appropriate type. This approach
provides very fast marshalling but could be susceptible to premature
garbage collection. Consider the following C++ code:</p>
<div class="code">
<pre>
class Wibble {
};
void wobble(Wibble &amp;w);
</pre>
</div>
<p> The module class contains the Java wrapper for the global <tt>wobble</tt>
method:</p>
<div class="code">
<pre>
public class example {
...
public static void wobble(Wibble w) {
exampleJNI.wobble(Wibble.getCPtr(w), w);
}
}
</pre>
</div>
<p> where <tt>example</tt> is the name of the module. All native methods
go through the intermediary class which has the native method declared
as such:</p>
<div class="code">
<pre>
public class exampleJNI {
...
public final static native void wobble(long jarg1, Wibble jarg1_);
}
</pre>
</div>
<p> The second parameter, <tt>jarg1_</tt>, is the premature garbage
collection prevention parameter and is added to the native method
parameter list whenever a C/C++ struct or class is marshalled as a Java
long. In order to understand why, consider the alternative where the
intermediary class method is declared without the additional parameter:</p>
<div class="code">
<pre>
public class exampleJNI {
...
public final static native void wobble(long jarg1);
}
</pre>
</div>
<p> and the following simple call to <tt>wobble</tt>:</p>
<div class="code">
<pre>
{
Wibble w = new Wibble();
example.wobble(w);
}
</pre>
</div>
<p> The hotspot compiler effectively sees something like:</p>
<div class="code">
<pre>
{
Wibble w = new Wibble();
long w_ptr = Wibble.getCPtr(w);
// w is no longer reachable
exampleJNI.wobble(w_ptr);
}
</pre>
</div>
<p> The <tt>Wibble</tt> object is no longer reachable after the point
shown as in this bit of code, the <tt>Wibble</tt> object is not
referenced again after this point. This means that it is a candidate
for garbage collection. Should <tt>wobble</tt> be a long running
method, it is quite likely that the finalizer for the <tt>Wibble</tt>
instance will be called. This in turn will call its underlying C++
destructor which is obviously disastrous while the method <tt>wobble</tt>
is running using this object. Even if <tt>wobble</tt> is not a long
running method, it is possible for the <tt>Wibble</tt> instance to be
finalized. By passing the <tt>Wibble</tt> instance into the native
method, it will not be finalized as the JVM guarantees not to finalize
any objects until the native method returns. Effectively, the code then
becomes</p>
<div class="code">
<pre>
{
Wibble w = new Wibble();
long w_ptr = Wibble.getCPtr(w);
exampleJNI.wobble(w_ptr, w);
// w is no longer reachable
}
</pre>
</div>
<p> and therefore there is no possibility of premature garbage
collection. In practice, this premature garbage collection was only
ever observed in Sun's server JVM from jdk-1.3 onwards and in Sun's
client JVM from jdk-1.6 onwards.</p>
<p> The premature garbage collection prevention parameter for proxy
classes is generated by default whenever proxy classes are passed by
value, reference or with a pointer. The implementation for this extra
parameter generation requires the &quot;jtype&quot; typemap to contain <tt>long</tt>
and the &quot;jstype&quot; typemap to contain the name of a proxy class.</p>
<p> The additional parameter does impose a slight performance overhead
and the parameter generation can be suppressed globally with the <tt>
-nopgcpp</tt> commandline option. More selective suppression is possible
with the 'nopgcpp' attribute in the &quot;jtype&quot; <a href="#Java_typemaps">
Java typemap</a>. The attribute is a flag and so should be set to &quot;1&quot; to
enable the suppression, or it can be omitted or set to &quot;0&quot; to disable.
For example:</p>
<div class="code">
<pre>
%typemap(jtype, nopgcpp=&quot;1&quot;) Wibble &amp; &quot;long&quot;
</pre>
</div>
<p><b> Compatibility note:</b> The generation of this additional
parameter did not occur in versions prior to SWIG-1.3.30.</p>
<h4><a name="Java_multithread_libraries">27.4.4.5 Single threaded
applications and thread safety</a></h4>
<p> Single threaded Java applications using JNI need to consider thread
safety. The same applies for the C# module where the .NET wrappers use
PInvoke. Consider the C++ class:</p>
<div class="code">
<pre>
class Test {
string str;
public:
Test() : str(&quot;initial&quot;) {}
};
</pre>
</div>
<p> and the Java proxy class generated by SWIG:</p>
<div class="code">
<pre>
public class Test {
private transient long swigCPtr;
protected transient boolean swigCMemOwn;
protected Test(long cPtr, boolean cMemoryOwn) {
swigCMemOwn = cMemoryOwn;
swigCPtr = cPtr;
}
protected static long getCPtr(Test obj) {
return (obj == null) ? 0 : obj.swigCPtr;
}
protected void finalize() {
delete();
}
// Call C++ destructor
public synchronized void delete() {
if(swigCPtr != 0 &amp;&amp; swigCMemOwn) {
swigCMemOwn = false;
exampleJNI.delete_Test(swigCPtr);
}
swigCPtr = 0;
}
// Call C++ constructor
public Test() {
this(exampleJNI.new_Test(), true);
}
}
</pre>
</div>
<p> It has two methods that call JNI methods, namely, <tt>
exampleJNI.new_Test()</tt> for the C++ constructor and <tt>
exampleJNI.delete_Test()</tt> for the C++ destructor. If the garbage
collector collects an instance of this class, ie <tt>delete()</tt> is
not explicitly called, then the C++ destructor will be run in a
different thread to the main thread. This is because when an object is
marked for garbage collection, any objects with finalizers are added to
a finalization queue and the objects in the finalization queue have
their <tt>finalize()</tt> methods run in a separate finalization
thread. Therefore, if the C memory allocator is not thread safe, then
the heap will get corrupted sooner or later, when a concurrent C++
delete and new are executed. It is thus essential, even in single
threaded usage, to link to the C multi-thread runtime libraries, for
example, use the /MD option for Visual C++ on Windows. Alternatively,
lock all access to C++ functions that have heap
allocation/deallocation.</p>
<p> Note that some of the STL in Visual C++ 6 is not thread safe, so
although code might be linked to the multithread runtime libraries,
undefined behaviour might still occur in a single threaded Java
program. Similarly some older versions of Sun Studio have bugs in the
multi-threaded implementation of the std::string class and so will lead
to undefined behaviour in these supposedly single threaded Java
applications.</p>
<p> The following innocuous Java usage of Test is an example that will
crash very quickly on a multiprocessor machine if the JNI compiled code
is linked against the single thread C runtime libraries.</p>
<div class="code">
<pre>
for (int i=0; i&lt;100000; i++) {
System.out.println(&quot;Iteration &quot; + i);
for (int k=0; k&lt;10; k++) {
Test test = new Test();
}
System.gc();
}
</pre>
</div>
<h3><a name="Java_type_wrapper_classes">27.4.5 Type wrapper classes</a></h3>
<p> The generated type wrapper class, for say an <tt>int *</tt>, looks
like this:</p>
<div class="code">
<pre>
public class SWIGTYPE_p_int {
private transient long swigCPtr;
protected SWIGTYPE_p_int(long cPtr, boolean bFutureUse) {
swigCPtr = cPtr;
}
protected SWIGTYPE_p_int() {
swigCPtr = 0;
}
protected static long getCPtr(SWIGTYPE_p_int obj) {
return obj.swigCPtr;
}
}
</pre>
</div>
<p> The methods do not have public access, so by default it is
impossible to do anything with objects of this class other than pass
them around. The methods in the class are part of the inner workings of
SWIG. If you need to mess around with pointers you will have to use
some typemaps specific to the Java module to achieve this. The section
on <a href="#Java_typemaps">Java typemaps</a> details how to modify the
generated code.</p>
<p> Note that if you use a pointer or reference to a proxy class in a
function then no type wrapper class is generated because the proxy
class can be used as the function parameter. If however, you need
anything more complicated like a pointer to a pointer to a proxy class
then a typewrapper class is generated for your use.</p>
<p> Note that SWIG generates a type wrapper class and not a proxy class
when it has not parsed the definition of a type that gets used. For
example, say SWIG has not parsed the definition of <tt>class Snazzy</tt>
because it is in a header file that you may have forgotten to use the <tt>
%include</tt> directive on. Should SWIG parse <tt>Snazzy *</tt> being
used in a function parameter, it will then generates a type wrapper
class around a <tt>Snazzy</tt> pointer. Also recall from earlier that
SWIG will use a pointer when a class is passed by value or by
reference:</p>
<div class="code">
<pre>
void spam(Snazzy *x, Snazzy &amp;y, Snazzy z);
</pre>
</div>
<p> Should SWIG not know anything about <tt>Snazzy</tt> then a <tt>
SWIGTYPE_p_Snazzy</tt> must be used for all 3 parameters in the <tt>spam</tt>
function. The Java function generated is:</p>
<div class="code">
<pre>
public static void spam(SWIGTYPE_p_Snazzy x, SWIGTYPE_p_Snazzy y, SWIGTYPE_p_Snazzy z) {
...
}
</pre>
</div>
<p> Note that typedefs are tracked by SWIG and the typedef name is used
to construct the type wrapper class name. For example, consider the
case where <tt>Snazzy</tt> is a typedef to an <tt>int</tt> which SWIG
does parse:</p>
<div class="code">
<pre>
typedef int Snazzy;
void spam(Snazzy *x, Snazzy &amp;y, Snazzy z);
</pre>
</div>
<p> Because the typedefs have been tracked the Java function generated
is:</p>
<div class="code">
<pre>
public static void spam(SWIGTYPE_p_int x, SWIGTYPE_p_int y, int z) { ... }
</pre>
</div>
<h3><a name="Java_enum_classes">27.4.6 Enum classes</a></h3>
<p> SWIG can generate three types of enum classes. The <a href="#Java_enumerations">
Enumerations</a> section discussed these but omitted all the details.
The following sub-sections detail the various types of enum classes
that can be generated.</p>
<h4><a name="Java_typesafe_enums_classes">27.4.6.1 Typesafe enum classes</a>
</h4>
<p> The following example demonstrates the typesafe enum classes which
SWIG generates:</p>
<div class="code">
<pre>
%include &quot;enumtypesafe.swg&quot;
%javaconst(1);
enum Beverage { ALE, LAGER=10, STOUT, PILSNER, PILZ=PILSNER };
</pre>
</div>
<p> The following is the code that SWIG generates:</p>
<div class="code">
<pre>
public final class Beverage {
public final static Beverage ALE = new Beverage(&quot;ALE&quot;);
public final static Beverage LAGER = new Beverage(&quot;LAGER&quot;, 10);
public final static Beverage STOUT = new Beverage(&quot;STOUT&quot;);
public final static Beverage PILSNER = new Beverage(&quot;PILSNER&quot;);
public final static Beverage PILZ = new Beverage(&quot;PILZ&quot;, PILSNER);
public final int swigValue() {
return swigValue;
}
public String toString() {
return swigName;
}
public static Beverage swigToEnum(int swigValue) {
if (swigValue &lt; swigValues.length &amp;&amp; swigValue &gt;= 0 &amp;&amp;
swigValues[swigValue].swigValue == swigValue)
return swigValues[swigValue];
for (int i = 0; i &lt; swigValues.length; i++)
if (swigValues[i].swigValue == swigValue)
return swigValues[i];
throw new IllegalArgumentException(&quot;No enum &quot; + Beverage.class + &quot; with value &quot; +
swigValue);
}
private Beverage(String swigName) {
this.swigName = swigName;
this.swigValue = swigNext++;
}
private Beverage(String swigName, int swigValue) {
this.swigName = swigName;
this.swigValue = swigValue;
swigNext = swigValue+1;
}
private Beverage(String swigName, Beverage swigEnum) {
this.swigName = swigName;
this.swigValue = swigEnum.swigValue;
swigNext = this.swigValue+1;
}
private static Beverage[] swigValues = { ALE, LAGER, STOUT, PILSNER, PILZ };
private static int swigNext = 0;
private final int swigValue;
private final String swigName;
}
</pre>
</div>
<p> As can be seen, there are a fair number of support methods for the
typesafe enum pattern. The typesafe enum pattern involves creating a
fixed number of static instances of the enum class. The constructors
are private to enforce this. Three constructors are available - two for
C/C++ enums with an initializer and one for those without an
initializer. Note that the two enums with initializers, <tt>LAGER</tt>
and <tt>PILZ</tt>, each call one the two different initializer
constructors. In order to use one of these typesafe enums, the <tt>
swigToEnum</tt> static method must be called to return a reference to
one of the static instances. The JNI layer returns the enum value from
the C/C++ world as an integer and this method is used to find the
appropriate Java enum static instance. The <tt>swigValue</tt> method is
used for marshalling in the other direction. The <tt>toString</tt>
method is overridden so that the enum name is available.</p>
<h4><a name="Java_proper_enums_classes">27.4.6.2 Proper Java enum
classes</a></h4>
<p> The following example demonstrates the Java enums approach:</p>
<div class="code">
<pre>
%include &quot;enums.swg&quot;
%javaconst(1);
enum Beverage { ALE, LAGER=10, STOUT, PILSNER, PILZ=PILSNER };
</pre>
</div>
<p> SWIG will generate the following Java enum:</p>
<div class="code">
<pre>
public enum Beverage {
ALE,
LAGER(10),
STOUT,
PILSNER,
PILZ(PILSNER);
public final int swigValue() {
return swigValue;
}
public static Beverage swigToEnum(int swigValue) {
Beverage[] swigValues = Beverage.class.getEnumConstants();
if (swigValue &lt; swigValues.length &amp;&amp; swigValue &gt;= 0 &amp;&amp;
swigValues[swigValue].swigValue == swigValue)
return swigValues[swigValue];
for (Beverage swigEnum : swigValues)
if (swigEnum.swigValue == swigValue)
return swigEnum;
throw new IllegalArgumentException(&quot;No enum &quot; + Beverage.class +
&quot; with value &quot; + swigValue);
}
private Beverage() {
this.swigValue = SwigNext.next++;
}
private Beverage(int swigValue) {
this.swigValue = swigValue;
SwigNext.next = swigValue+1;
}
private Beverage(Beverage swigEnum) {
this.swigValue = swigEnum.swigValue;
SwigNext.next = this.swigValue+1;
}
private final int swigValue;
private static class SwigNext {
private static int next = 0;
}
}
</pre>
</div>
<p> The enum items appear first. Like the typesafe enum pattern, the
constructors are private. The constructors are required to handle C/C++
enums with initializers. The <tt>next</tt> variable is in the <tt>
SwigNext</tt> inner class rather than in the enum class as static
primitive variables cannot be modified from within enum constructors.
Marshalling between Java enums and the C/C++ enum integer value is
handled via the <tt>swigToEnum</tt> and <tt>swigValue</tt> methods. All
the constructors and methods in the Java enum are required just to
handle C/C++ enums with initializers. These needn't be generated if the
enum being wrapped does not have any initializers and the <a href="#Java_simpler_enum_classes">
Simpler Java enums for enums without initializers</a> section describes
how typemaps can be used to achieve this.</p>
<h4><a name="Java_typeunsafe_enums_classes">27.4.6.3 Type unsafe enum
classes</a></h4>
<p> The following example demonstrates type unsafe enums:</p>
<div class="code">
<pre>
%include &quot;enumtypeunsafe.swg&quot;
%javaconst(1);
enum Beverage { ALE, LAGER=10, STOUT, PILSNER, PILZ=PILSNER };
</pre>
</div>
<p> SWIG will generate the following simple class:</p>
<div class="code">
<pre>
public final class Beverage {
public final static int ALE = 0;
public final static int LAGER = 10;
public final static int STOUT = LAGER + 1;
public final static int PILSNER = STOUT + 1;
public final static int PILZ = PILSNER;
}
</pre>
</div>
<h3><a name="Java_interfaces">27.4.7 Interfaces</a></h3>
<p> By default SWIG wraps all C++ classes as Java classes. As Java only
supports derivation from a single base class, SWIG has to ignore all
bases except the first when a C++ class inherits from more than one
base class. However, there is a family of SWIG macros that change the
default wrapping and allows a C++ class to be wrapped as a Java
interface instead of a Java class. These macros provide a way to
support some sort of multiple inheritance as there is no limit to the
number of interfaces that a Java class can inherit from.</p>
<p> When a C++ class is wrapped as a Java interface, a Java proxy class
is still needed. The <tt>swiginterface.i</tt> library file provides
three macros for marking a C++ class to be wrapped as a Java interface.
There is more than one macro in order to provide a choice for choosing
the Java interface and Java proxy names.</p>
<table BORDER summary="Java interface macros">
<tr VALIGN="TOP"><td><b>Interface Macro Name</b></td><td><b>Description</b>
</td></tr>
<tr><td><tt>%interface(CTYPE)</tt></td><td>For C++ class <tt>CTYPE</tt>,
proxy class name is unchanged without any suffix added, interface name
has <tt>SwigInterface</tt> added as a suffix.</td></tr>
<tr><td><tt>%interface_impl(CTYPE)</tt></td><td>For C++ class <tt>CTYPE</tt>
, proxy class name has <tt>SwigImpl</tt> added as a suffix, interface
name has no added suffix.</td></tr>
<tr><td><tt>%interface_custom(&quot;PROXY&quot;, &quot;INTERFACE&quot;, CTYPE)</tt></td><td>
For C++ class <tt>CTYPE</tt>, proxy class name is given by the string <tt>
PROXY</tt>, interface name is given by the string <tt>INTERFACE</tt>.
The <tt>PROXY</tt> and <tt>INTERFACE</tt> names can use the <a href="#SWIG_advanced_renaming">
string formatting functions</a> used in <tt>%rename</tt>.</td></tr>
<tr><td><tt>%interface_additional(&quot;PROXY&quot;, &quot;INTERFACE&quot;, &quot;ADDITIONAL&quot;,
CTYPE)</tt></td><td>For C++ class <tt>CTYPE</tt>, proxy class name is
given by the string <tt>PROXY</tt>, interface name is given by the
string <tt>INTERFACE</tt>. The <tt>PROXY</tt> and <tt>INTERFACE</tt>
names can use the <a href="#SWIG_advanced_renaming">string formatting
functions</a> used in <tt>%rename</tt>. <tt>ADDITIONAL</tt> should
contain a comma separated list of interfaces for the interface class to
additionally extend. This is for adding interfaces not parsed by SWIG
and is useful for adding in pure Java interfaces</td></tr>
</table>
<p> The table below has a few examples showing the resulting proxy and
interface names for a C++ class called <tt>Base</tt>.</p>
<table BORDER summary="Java interface macro examples">
<tr VALIGN="TOP"><td><b>Example Usage</b></td><td><b>Proxy Class Name</b>
</td><td><b>Interface Class Name</b></td></tr>
<tr><td><tt>%interface(Base)</tt></td><td><tt>Base</tt></td><td><tt>
BaseSwigInterface</tt></td></tr>
<tr><td><tt>%interface_impl(Base)</tt></td><td><tt>BaseSwigImpl</tt></td><td>
<tt>Base</tt></td></tr>
<tr><td><tt>%interface_custom(&quot;BaseProxy&quot;, &quot;IBase&quot;, Base)</tt></td><td><tt>
BaseProxy</tt></td><td><tt>IBase</tt></td></tr>
<tr><td><tt>%interface_custom(&quot;%sProxy&quot;, &quot;IBase&quot;, Base)</tt></td><td><tt>
BaseProxy</tt></td><td><tt>IBase</tt></td></tr>
<tr><td><tt>%interface_custom(&quot;%sProxy&quot;, &quot;%sInterface&quot;, Base)</tt></td><td>
<tt>BaseProxy</tt></td><td><tt>BaseProxyInterface</tt></td></tr>
<tr><td><tt>%interface_custom(&quot;%sProxy&quot;, &quot;%(rstrip:[Proxy])sInterface&quot;,
Base)</tt></td><td><tt>BaseProxy</tt></td><td><tt>BaseInterface</tt></td>
</tr>
</table>
<p> The 2nd last example shows the names used in the string formatting
functions. The input for <tt>PROXY</tt> that <tt>&quot;%s&quot;</tt> expands to
is the proxy name, that is, Base. The input for <tt>INTERFACE</tt> that
<tt>&quot;%s&quot;</tt> expands to is the proxy name, that is, <tt>BaseProxy</tt>
.</p>
<p> The last example shows <tt>rstrip</tt> and in this case strips the <tt>
Proxy</tt> suffix and then adds on <tt>Interface</tt>.</p>
<p> Consider the following C++ code:</p>
<div class="code">
<pre>
namespace Space {
struct Base1 {
virtual void Method1();
virtual Base1();
};
struct Base2 {
virtual void Method2();
virtual Base2();
};
struct Derived : Base1, Base2 {
};
void UseBases(const Base1 &amp;b1, const Base2 &amp;b2);
}
</pre>
</div>
<p> By default all classes are wrapped and are available in Java, but, <tt>
Derived</tt> has all bases ignored except the first. SWIG generates a
warning for the above code:</p>
<div class="shell">
<pre>
example.i:12: Warning 813: Warning for Derived, base Base2 ignored.
Multiple inheritance is not supported in Java.
</pre>
</div>
<p> If we decide to wrap the two base classes as interfaces and add the
following before SWIG parses the above example code:</p>
<div class="code">
<pre>
%include &lt;swiginterface.i&gt;
%interface_impl(Space::Base1);
%interface_impl(Space::Base2);
</pre>
</div>
<p> then two interface files are generated, Base1.java and Base2.java in
addition to proxy class files, Base1SwigImpl.java and
Base2SwigImpl.java. The contents of interface file Base1.java for <tt>
Base1</tt> is shown below:</p>
<div class="code">
<pre>
public interface Base1 {
long Base1_GetInterfaceCPtr();
void Method1();
}
</pre>
</div>
<p> The proxy class in Base1SwigImpl.java for Base1 is as it would have
been if <tt>%interface</tt> was not used, except the name has changed
to <tt>Base1SwigImpl</tt> and it implements the appropriate base:</p>
<div class="code">
<pre>
public class Base1SwigImpl implements Base1 {
...
public long Base1_GetInterfaceCPtr() {
return exampleJNI.Base1SwigImpl_Base1_GetInterfaceCPtr(swigCPtr);
}
public void Method1() {
exampleJNI.Base1SwigImpl_Method1(swigCPtr, this);
}
...
}
</pre>
</div>
<p> In fact any class using <tt>Base</tt> as an immediate base class
will now implement the interface instead of deriving from it (or
ignoring the base in the case of multiple base classes). Hence the <tt>
Derived</tt> proxy class will now implement both bases:</p>
<div class="code">
<pre>
public class Derived implements Base1, Base2 {
...
public long Base1_GetInterfaceCPtr() {
return exampleJNI.Derived_Base1_GetInterfaceCPtr(swigCPtr);
}
public long Base2_GetInterfaceCPtr() {
return exampleJNI.Derived_Base2_GetInterfaceCPtr(swigCPtr);
}
public void Method1() {
exampleJNI.Derived_Method1(swigCPtr, this);
}
public void Method2() {
exampleJNI.Derived_Method2(swigCPtr, this);
}
...
}
</pre>
</div>
<p> The proxy class has methods added to it, from the implemented bases,
so that the underlying C++ implementation can be called. In the example
above, <tt>Method1</tt> and <tt>Method2</tt> have been added from the
implemented bases. If a method is ignored in the base, such as via <tt>
%ignore</tt>, then that method will be excluded from the interface and
there will not be an additional method added to the proxy class
implementing that interface.</p>
<p> The Java interface only ever contains virtual and non-virtual
instance methods from the wrapped C++ class. Any static methods, enums
or variables in the wrapped C++ class are not supported and are not
added to the interface. They are of course still available in the Java
proxy class.</p>
<p> Wherever a class marked as an interface is used, such as the <tt>
UseBases</tt> method in the example, the interface name is used as the
type in the Java layer:</p>
<div class="code">
<pre>
public static void UseBases(Base1 b1, Base2 b2) {
exampleJNI.UseBases(b1.Base1_GetInterfaceCPtr(), b1, b2.Base2_GetInterfaceCPtr(), b2);
}
</pre>
</div>
<p> Note that each Java interface has a method added to obtain the
correct C++ pointer for passing to the native function - <tt>
Base1_GetInterfaceCPtr</tt> for <tt>Base1</tt>. This method is similar
to the <tt>getCPtr</tt> method in the proxy classes. In fact, as shown
above in the <tt>Derived</tt> class, the proxy classes implement this
generated interface by calling a native method (<tt>
Derived_Base1_GetInterfaceCPtr</tt>) which calls an appropriate C++ cast
of the pointer up the inheritance chain.</p>
<p> The <tt>%interface_additional</tt> macro is useful for adding
additional interfaces to the Java interface class. Consider a simple
class hierarchy, where <tt>Whizz</tt> inherits from <tt>Bang</tt> and
we want these two classes to be wrapped as Java interfaces and the <tt>
Whizz</tt> interface to additionally extend from the pure Java interface
<tt>java.util.EventListener</tt>.</p>
<div class="code">
<pre>
%interface_custom(&quot;Bang&quot;, &quot;IBang&quot;, Space::Bang);
%interface_additional(&quot;Whizz&quot;, &quot;IWhizz&quot;, &quot;java.util.EventListener&quot;, Space::Whizz)
namespace Space {
struct Bang {
virtual void bang() = 0;
};
struct Whizz : Bang {
virtual void whizz() = 0;
};
}
</pre>
</div>
<p> The classes and interfaces generated are shown below, noting that <tt>
IWhizz</tt> extends the additional interface <tt>java.util.EventListener</tt>
:</p>
<div class="code">
<pre>
public interface IBang { ... }
public interface IWhizz extends java.util.EventListener, IBang { ... }
public class Bang implements IBang { ... }
public class Whizz implements IWhizz, IBang { ... }
</pre>
</div>
<p> Note the subtle difference to the <tt>javainterfaces</tt> typemap
discussed elsewhere. The typemap applies to the generated Java proxy
class only, not the Java interface. A slight change to our example
above as follows:</p>
<div class="code">
<pre>
%interface_custom(&quot;Bang&quot;, &quot;IBang&quot;, Space::Bang);
%interface_custom(&quot;Whizz&quot;, &quot;IWhizz&quot;, Space::Whizz)
%typemap(javainterfaces) Space::Whizz &quot;java.util.EventListener&quot;
namespace Space {
... as shown above ...
}
</pre>
</div>
<p> will then the generate different code for the proxy class <tt>Whizz</tt>
and the interface <tt>IWhizz</tt> as follows:</p>
<div class="code">
<pre>
public interface IBang { ... }
public interface IWhizz extends IBang { ... }
public class Bang implements IBang { ... }
public class Whizz implements IWhizz, IBang, java.util.EventListener { ... }
</pre>
</div>
<p> Finally, a note on the implementation of the interface macros. The
interface macros are implemented using the <tt>interface</tt> feature
and typemaps. A couple of examples:</p>
<div class="code">
<pre>
%define %interface(CTYPE...)
%feature(&quot;interface&quot;, name=&quot;%sSwigInterface&quot;) CTYPE;
INTERFACE_TYPEMAPS(CTYPE)
%enddef
%define %interface_additional(PROXY, INTERFACE, ADDITIONAL, CTYPE...)
%rename(PROXY) CTYPE;
%feature(&quot;interface&quot;, name=INTERFACE, additional=ADDITIONAL) CTYPE;
INTERFACE_TYPEMAPS(CTYPE)
%enddef
</pre>
</div>
<p> The feature accepts two attributes named <tt>name</tt> and <tt>
additional</tt>. The <tt>name</tt> attribute should contain the name of
the Java interface. The <tt>additional</tt> attribute should contain
additional interfaces for the interface class, not parsed by SWIG. The <tt>
INTERFACE_TYPEMAPS</tt> macro implements the typemaps and can be viewed
in the <tt>swiginterface.i</tt> file and contains the usual Java
typemaps for generating code plus the <tt>javainterfacecode</tt>
typemap which is only used when a class is marked with the <tt>
interface</tt> feature. See <a href="#Java_code_typemaps">Java code
typemaps</a> for details.</p>
<p><b> Compatibility note:</b> The <tt>additional</tt> attribute and <tt>
%interface_additional</tt> macro was added in SWIG-4.3.0.</p>
<h2><a name="Java_directors">27.5 Cross language polymorphism using
directors</a></h2>
<p> Proxy classes provide a natural, object-oriented way to wrap C++
classes. as described earlier, each proxy instance has an associated
C++ instance, and method calls from Java to the proxy are passed to the
C++ instance transparently via C wrapper functions.</p>
<p> This arrangement is asymmetric in the sense that no corresponding
mechanism exists to pass method calls down the inheritance chain from
C++ to Java. In particular, if a C++ class has been extended in Java
(by deriving from the proxy class), these classes will not be visible
from C++ code. Virtual method calls from C++ are thus not able to
access the lowest implementation in the inheritance chain.</p>
<p> SWIG can address this problem and make the relationship between C++
classes and proxy classes more symmetric. To achieve this goal, new
classes called directors are introduced at the bottom of the C++
inheritance chain. The job of the directors is to route method calls
correctly, either to C++ implementations higher in the inheritance
chain or to Java implementations lower in the inheritance chain. The
upshot is that C++ classes can be extended in Java and from C++ these
extensions look exactly like native C++ classes. Neither C++ code nor
Java code needs to know where a particular method is implemented: the
combination of proxy classes, director classes, and C wrapper functions
transparently takes care of all the cross-language method routing.</p>
<h3><a name="Java_enabling_directors">27.5.1 Enabling directors</a></h3>
<p> The director feature is disabled by default. To use directors you
must make two changes to the interface file. First, add the &quot;directors&quot;
option to the %module directive, like this:</p>
<div class="code">
<pre>
%module(directors=&quot;1&quot;) modulename
</pre>
</div>
<p> Without this option no director code will be generated. Second, you
must use the %feature(&quot;director&quot;) directive to tell SWIG which classes
and methods should get directors. The %feature directive can be applied
globally, to specific classes, and to specific methods, like this:</p>
<div class="code">
<pre>
// generate directors for all classes that have virtual methods
%feature(&quot;director&quot;);
// generate directors for the virtual methods in class Foo
%feature(&quot;director&quot;) Foo;
</pre>
</div>
<p> You can use the %feature(&quot;nodirector&quot;) directive to turn off
directors for specific classes or methods. So for example,</p>
<div class="code">
<pre>
%feature(&quot;director&quot;) Foo;
%feature(&quot;nodirector&quot;) Foo::bar;
</pre>
</div>
<p> will generate directors for the virtual methods of class Foo except
bar().</p>
<p> Directors can also be generated implicitly through inheritance. In
the following, class Bar will get a director class that handles the
methods one() and two() (but not three()):</p>
<div class="code">
<pre>
%feature(&quot;director&quot;) Foo;
class Foo {
public:
virtual void one();
virtual void two();
};
class Bar: public Foo {
public:
virtual void three();
};
</pre>
</div>
<h3><a name="Java_directors_classes">27.5.2 Director classes</a></h3>
<p> For each class that has directors enabled, SWIG generates a new
class that derives from both the class in question and a special <tt>
Swig::Director</tt> class. These new classes, referred to as director
classes, can be loosely thought of as the C++ equivalent of the Java
proxy classes. The director classes store a pointer to their underlying
Java proxy classes.</p>
<p> For simplicity let's ignore the <tt>Swig::Director</tt> class and
refer to the original C++ class as the director's base class. By
default, a director class extends all virtual methods in the
inheritance chain of its base class (see the preceding section for how
to modify this behavior). Virtual methods that have a final specifier
are unsurprisingly excluded. Thus the virtual method calls, whether
they originate in C++ or in Java via proxy classes, eventually end up
in at the implementation in the director class. The job of the director
methods is to route these method calls to the appropriate place in the
inheritance chain. By &quot;appropriate place&quot; we mean the method that would
have been called if the C++ base class and its Java derived classes
were seamlessly integrated. That seamless integration is exactly what
the director classes provide, transparently skipping over all the messy
JNI glue code that binds the two languages together.</p>
<p> In reality, the &quot;appropriate place&quot; is one of only two
possibilities: C++ or Java. Once this decision is made, the rest is
fairly easy. If the correct implementation is in C++, then the lowest
implementation of the method in the C++ inheritance chain is called
explicitly. If the correct implementation is in Java, the Java API is
used to call the method of the underlying Java object (after which the
usual virtual method resolution in Java automatically finds the right
implementation).</p>
<h3><a name="Java_directors_overhead">27.5.3 Overhead and code bloat</a></h3>
<p> Enabling directors for a class will generate a new director method
for every virtual method in the class' inheritance chain. This alone
can generate a lot of code bloat for large hierarchies. Method
arguments that require complex conversions to and from Java types can
result in large director methods. For this reason it is recommended
that directors are selectively enabled only for specific classes that
are likely to be extended in Java and used in C++.</p>
<p> Although directors make it natural to mix native C++ objects with
Java objects (as director objects), one should be aware of the obvious
fact that method calls to Java objects from C++ will be much slower
than calls to C++ objects. Additionally, compared to classes that do
not use directors, the call routing in the director methods adds a
small overhead. This situation can be optimized by selectively enabling
director methods (using the %feature directive) for only those methods
that are likely to be extended in Java.</p>
<h3><a name="Java_directors_example">27.5.4 Simple directors example</a></h3>
<p> Consider the following SWIG interface file:</p>
<div class="code">
<pre>
%module(directors=&quot;1&quot;) example;
%feature(&quot;director&quot;) DirectorBase;
class DirectorBase {
public:
virtual ~DirectorBase() {}
virtual void upcall_method() {}
};
void callup(DirectorBase *director) {
director-&gt;upcall_method();
}
</pre>
</div>
<p> The following <code>DirectorDerived</code> Java class is derived
from the Java proxy class <code>DirectorBase</code> and overrides <code>
upcall_method()</code>. When C++ code invokes <code>upcall_method()</code>
, the SWIG-generated C++ code redirects the call via JNI to the Java <code>
DirectorDerived</code> subclass. Naturally, the SWIG generated C++ code
and the generated Java intermediary class marshal and convert arguments
between C++ and Java when needed.</p>
<div class="code">
<pre>
class DirectorDerived extends DirectorBase {
@Override
public void upcall_method() {
System.out.println(&quot;DirectorDerived.upcall_method() invoked.&quot;);
}
}
</pre>
</div>
<p> Running the following Java code</p>
<div class="code">
<pre>
DirectorDerived director = new DirectorDerived();
example.callup(director);
</pre>
</div>
<p> will result in the following being output:</p>
<div class="code">
<pre>
DirectorDerived.upcall_method() invoked.
</pre>
</div>
<h3><a name="Java_directors_threading">27.5.5 Director threading issues</a>
</h3>
<p> Depending on your operating system and version of Java and how you
are using threads, you might find the JVM hangs on exit. There are a
couple of solutions to try out. The preferred solution requires jdk-1.4
and later and uses <tt>AttachCurrentThreadAsDaemon</tt> instead of <tt>
AttachCurrentThread</tt> whenever a call into the JVM is required. This
can be enabled by defining the
SWIG_JAVA_ATTACH_CURRENT_THREAD_AS_DAEMON macro when compiling the C++
wrapper code. For older JVMs define SWIG_JAVA_NO_DETACH_CURRENT_THREAD
instead, to avoid the <tt>DetachCurrentThread</tt> call but this will
result in a memory leak instead. For further details inspect the source
code in the java/director.swg library file.</p>
<p> Macros can be defined on the commandline when compiling your C++
code, or alternatively added to the C++ wrapper file as shown below:</p>
<div class="code">
<pre>
%insert(&quot;runtime&quot;) %{
#define SWIG_JAVA_NO_DETACH_CURRENT_THREAD
%}
</pre>
</div>
<h3><a name="Java_directors_performance">27.5.6 Director performance
tuning</a></h3>
<p> When a new instance of a director (or subclass) is created in Java,
the C++ side of the director performs a runtime check per director
method to determine if that particular method is overridden in Java or
if it should invoke the C++ base implementation directly. Although this
makes initialization slightly more expensive, it is generally a good
overall tradeoff.</p>
<p> However, if all director methods are expected to usually be
overridden by Java subclasses, then initialization can be made faster
by avoiding these checks via the <tt>assumeoverride</tt> attribute. For
example:</p>
<div class="code">
<pre>
%feature(&quot;director&quot;, assumeoverride=1) Foo;
</pre>
</div>
<p> The disadvantage is that invocation of director methods from C++
when Java doesn't actually override the method will require an
additional call up into Java and back to C++. As such, this option is
only useful when overrides are extremely common and instantiation is
frequent enough that its performance is critical.</p>
<h3><a name="Java_exceptions_from_directors">27.5.7 Java exceptions from
directors</a></h3>
<p> With directors routing method calls to Java, and proxies routing
them to C++, the handling of exceptions is an important concern. The
default behavior for Java exceptions thrown in a director method
overridden in Java is to store the thrown Java exception into a SWIG
defined <code>Swig::DirectorException</code> C++ class exception in the
C++ layer and then throw this C++ exception.</p>
<p> Of course, should this exception be thrown, your C++ code must catch
it and handle it before returning back to Java. The default generated
code<b> does not</b> attempt to handle the C++ exception, but there is
a simple way to make this all work by catching the C++ exception and
extracting the original Java exception by using <tt>%catches</tt> for <tt>
Swig::DirectorException</tt>. Consider the example shown earlier with a
modification to the <tt>upcall_method</tt> Java method to throw a Java
exception:</p>
<div class="code">
<pre>
class DirectorDerived extends DirectorBase {
@Override
public void upcall_method() {
System.out.println(&quot;DirectorDerived.upcall_method() invoked.&quot;);
throw new RuntimeException(&quot;There was a problem!&quot;);
}
}
</pre>
</div>
<p> Now, by default, the JVM will abort when <tt>
example.callup(director)</tt> is called as the C++ <tt>
Swig::DirectorException</tt> (storing the Java exception) is thrown and
not handled by the <tt>callup</tt> method. Needless to say this is not
very user friendly and so the recommendation is to add the following
simple <tt>%catches</tt> directive before SWIG parses the <tt>callup</tt>
function:</p>
<div class="code">
<pre>
%catches(Swig::DirectorException) callup;
</pre>
</div>
<p> Or target all wrapped methods using:</p>
<div class="code">
<pre>
%catches(Swig::DirectorException);
</pre>
</div>
<p> This tells SWIG to generate a C++ catch handler using some code from
the <a href="#Typemaps_throws_typemap">throws typemap</a> for <tt>
Swig::DirectorException</tt> that SWIG supplies by default, see <a href="#SWIGPlus_catches">
Exception handling with %catches</a>. This typemap code is written to
simply catch the C++ <tt>Swig::DirectorException</tt> class and
immediately return to Java throwing the original Java exception that it
has stored. The net result is a stack trace containing the original
Java exception including the location that the exception was thrown
from.</p>
<div class="shell">
<pre>
DirectorDerived.upcall_method() invoked.
Exception in thread &quot;main&quot; java.lang.RuntimeException: There was a problem!
at DirectorDerived.upcall_method(runme.java:4)
at exampleJNI.SwigDirector_DirectorBase_upcall_method(exampleJNI.java:20)
at exampleJNI.callup(Native Method)
at example.callup(example.java:12)
at runme.main(runme.java:21)
</pre>
</div>
<p> More on the <tt>Swig::DirectorException</tt> class can be found in
the next section which details how to customize the handling of
director exceptions.</p>
<h4><a name="Java_customizing_director_exceptions">27.5.7.1 Customizing
director exceptions</a></h4>
<p> This section is for advanced customization of director exceptions.
The recommendation for most users is to use the simple <tt>%catches</tt>
directive described above as it should be sufficient for most users
needs.</p>
<p> The conversion of Java exceptions into C++ exceptions can be
customized in two different ways using the <code>director:except</code>
<a href="#Customization_features">feature</a>. In the first approach, a
code block is attached to each director method to handle the mapping of
Java exceptions into C++ exceptions. The code block is generated just
after the call up from the C++ director method into the overloaded
method in Java. Its primary function is to check if a Java exception
has been thrown and then handle it in C++. The example below converts a
<tt>java.lang.IndexOutOfBoundsException</tt> into a C++ <tt>
std::out_of_range</tt> exception and converts a user's Java <tt>
MyJavaException</tt> into a C++ <tt>MyCppException</tt> exception. If
the Java exception doesn't match either of these, a fallback <tt>
std::runtime_error</tt> C++ exception is thrown.</p>
<div class="code">
<pre>
%feature(&quot;director:except&quot;) MyClass::dirmethod(int x) {
jthrowable $error = jenv-&gt;ExceptionOccurred();
if ($error) {
if (Swig::ExceptionMatches(jenv, $error, &quot;java/lang/IndexOutOfBoundsException&quot;))
throw std::out_of_range(Swig::JavaExceptionMessage(jenv, $error).message());
if (Swig::ExceptionMatches(jenv, $error, &quot;$packagepath/MyJavaException&quot;))
throw MyCppException(Swig::JavaExceptionMessage(jenv, $error).message());
throw std::runtime_error(&quot;Unexpected exception thrown in MyClass::dirmethod&quot;);
}
}
class MyClass {
public:
/** Throws either a std::out_of_range or MyCppException on error */
virtual void dirmethod(int x);
virtual ~MyClass();
};
</pre>
</div>
<p> A few special variables are expanded within the <tt>director:except</tt>
feature.</p>
<ul>
<li> The special variable <tt>$error</tt> is expanded into a unique
variable name (swigerror) and should be used for the assignment of the
jthrowable exception that occurred.</li>
<li> The special variable <tt>$packagepath</tt> is replaced by the outer
package provided for SWIG generation by the -package option.</li>
<li> The special variable <tt>$directorthrowshandlers</tt> is not shown
above, but is replaced by applicable &quot;directorthrows&quot; typemap contents
(covered later in this section).</li>
<li> The special variable <tt>$null</tt> is not shown above, but is
replaced by a suitable default constructed object for returning from
the director method (or nothing if the director method has a void
return).</li>
</ul>
<p> Utility functions/classes in director.swg are provided to aid the
exception conversion as follows:</p>
<div class="code">
<pre>
namespace Swig {
// Helper method to determine if a Java throwable matches a particular Java class type
// Note side effect of clearing any pending exceptions
bool ExceptionMatches(JNIEnv *jenv, jthrowable throwable, const char *classname);
// Helper class to extract the exception message from a Java throwable
class JavaExceptionMessage {
public:
JavaExceptionMessage(JNIEnv *jenv, jthrowable throwable);
// Return a C string of the exception message in the jthrowable passed in the constructor
// If no message is available, null_string is return instead
const char *message(const char *null_string =
&quot;Could not get exception message in JavaExceptionMessage&quot;) const;
};
// C++ Exception class for handling Java exceptions thrown during a director method Java upcall
class DirectorException : public std::exception {
public:
// Construct exception from a Java throwable
DirectorException(JNIEnv *jenv, jthrowable throwable);
// More general constructor for handling as a java.lang.RuntimeException
DirectorException(const char *msg);
// Return exception message extracted from the Java throwable
const char *what() const throw();
// Reconstruct and raise/throw the Java Exception that caused the DirectorException
// Note that any error in the JNI exception handling results in a Java RuntimeException
void throwException(JNIEnv *jenv) const;
// Create and throw the DirectorException
static void raise(JNIEnv *jenv, jthrowable throwable) {
throw DirectorException(jenv, throwable);
}
};
}
</pre>
</div>
<p> The utility function <code>Swig::ExceptionMatches</code> and class <code>
Swig::JavaExceptionMessage</code> are provided to simplify writing code
for wrappers that use the <code>director:except</code> feature. The
function <code>Swig::ExceptionMatches</code> matches the type of the <code>
jthrowable</code> thrown against a<b> fully qualified</b> JNI style
class name, such as <code>&quot;java/lang/IOError&quot;</code>. If the throwable
class is the same type, or derives from the given type, <code>
Swig::ExceptionMatches</code> will return true. Care must be taken to
provide the correct fully qualified name, since for wrapped exceptions
the generated proxy class will have an additional package
qualification, depending on the '-package' argument and use of the <a href="#Java_namespaces">
nspace feature</a>. The utility class <code>Swig::JavaExceptionMessage</code>
is a holder providing access to the message from the thrown Java
exception. The <code>message()</code> method returns the exception
message as a <code>const char *</code>, which is only valid during the
lifetime of the holder. Any code using this message needs to copy it,
for example into a std::string or a newly constructed C++ exception.</p>
<p> Using the first approach above to write handlers for a large number
of methods will require repetitive duplication of the <code>
director:except</code> feature code for each director method. To
mitigate this, a second approach is provided via typemaps in a fashion
analogous to the <a href="#Typemaps_throws_typemap">&quot;throws&quot; typemap</a>
. The &quot;throws&quot; typemap provides a way to map all the C++ exceptions
listed in a method's defined exceptions (either from a C++<em>
exception specification</em> or a <code>%catches</code> feature) into
Java exceptions. The &quot;directorthrows&quot; typemap provides the inverse
mapping and should contain code to convert a suitably matching Java
exception into a C++ exception. Only use this typemap if you wish to
write custom conversions of Java exceptions into C++ exceptions and
apply them to many different methods. The default handling which uses
the <code>Swig::DirectorException</code> class should otherwise meet
your needs.</p>
<p> The example below converts a Java <code>
java.lang.IndexOutOfBoundsException</code> exception to the typemap's
type, that is a <code>std::out_of_range</code> C++ exception:<div class="code">
<pre>
%typemap(directorthrows) std::out_of_range %{
if (Swig::ExceptionMatches(jenv, $error, &quot;java/lang/IndexOutOfBoundsException&quot;)) {
throw std::out_of_range(Swig::JavaExceptionMessage(jenv, $error).message());
}
%}
</pre>
</div></p>
<p> The &quot;directorthrows&quot; typemap is then used in conjunction with the <code>
director:except</code> feature if the <code>$directorthrowshandlers</code>
special variable is used in the code block. Consider the following,
which also happens to be the default:</p>
<div class="code">
<pre>
%feature(&quot;director:except&quot;) %{
jthrowable $error = jenv-&gt;ExceptionOccurred();
if ($error) {
$directorthrowshandlers
Swig::DirectorException::raise(jenv, $error);
}
%}
</pre>
</div>
<p> where <tt>Swig::DirectorException::raise</tt> is the helper method
to throw a C++ <tt>Swig::DirectorException</tt>, see above. The code
generated from the <code>director:except</code> feature has the <code>
$directorthrowshandlers</code> special variable replaced with the code
in the relevant &quot;directorthrows&quot; typemaps, for each and every exception
defined for the method. The relevant exceptions can be defined either
with a C++ exception specification or <code>%catches</code> as
described for the <a href="#Typemaps_throws_typemap">&quot;throws&quot; typemap</a>
.</p>
<p> Let's try and put all this together by considering the following
director method:</p>
<div class="code">
<pre>
struct X {
virtual void doSomething(int index) throw (std::out_of_range);
...
};
OR
%catches(std::out_of_range) X::doSomething;
struct X {
virtual void doSomething(int index);
...
};
</pre>
</div>
<p> When combined with the default <code>director:except</code> feature
and the &quot;directorthrows&quot; typemap above, the resulting code generated in
the director method after calling up to Java will be:</p>
<div class="code">
<pre>
jthrowable swigerror = jenv-&gt;ExceptionOccurred();
if (swigerror) {
if (Swig::ExceptionMatches(jenv, swigerror, &quot;java/lang/IndexOutOfBoundsException&quot;)) {
throw std::out_of_range(Swig::JavaExceptionMessage(jenv, swigerror).message());
}
Swig::DirectorException::raise(jenv, swigerror);
}
</pre>
</div>
<p><em> Note: Beware of using exception specifications as the SWIG
director methods will be generated with the same exception
specifications and if the director method throws an exception that is
not specified in the exception specifications list it is likely to
terminate your program. See the C++ standard for more details. Using
the %catches feature instead to define the handled exceptions does not
suffer this potential fate.</em></p>
<p>Because the default code generation maps any unhandled Java
exceptions to <code>Swig::DirectorException</code>, any director
methods that have exception specifications may cause program
termination as this exception class won't be in the exception
specifications list. You can avoid throwing <tt>Swig::DirectorException</tt>
by changing the default handling for all methods by adding a <tt>
director:except</tt> feature without any method name. For example, you
can just ignore them:</p>
<div class="code">
<pre>
%feature(&quot;director:except&quot;) %{
jthrowable $error = jenv-&gt;ExceptionOccurred();
if ($error) {
$directorthrowshandlers
jenv-&gt;ExceptionClear();
return $null; // exception is ignored
}
%}
</pre>
</div>
<p>Alternatively an exception compatible with the existing director
method exception specifications can be thrown. Assuming that all
methods allow std::runtime_error to be thrown, the <code>return&nbsp;$null</code>
line above could be changed to:</p>
<div class="code">
<pre>
throw std::runtime_error(Swig::JavaExceptionMessage(jenv, $error).message());
</pre>
</div>
<p>In more complex situations, a separate <code>director:except</code>
feature may need to be attached to specific methods by providing a
method name to the <tt>director:except</tt> feature.</p>
<p>This is all no doubt quite hard to follow without seeing a full
example and some code. Below is a complete example demonstrating the
use of most of the exception customizations one can use, that is,
&quot;directorthrows&quot; and &quot;throws&quot; typemaps, %exception and %catches. See
the <a href="#Java_exception_handling">Exception handling with
%exception and %javaexception</a> section for more on converting C++
exceptions to Java exceptions. The example also has a user defined C++
exception class called <tt>MyNS::MyException</tt> and this is wrapped
as a Java exception. The director class being wrapped is <tt>MyClass</tt>
and the director method is called <tt>MyClass::dirmethod</tt>. A number
of <tt>std::cout</tt> calls have been added to help understand code
flow. You can copy the code below into an interface file and run SWIG
on it and examine the generated code.<div class="code">
<pre>
%module(directors=&quot;1&quot;) example
%{
#include &lt;stdexcept&gt;
#include &lt;iostream&gt;
%}
// Generic catch handler for all wrapped methods
%exception %{
try {
$action
} catch (const std::exception &amp;e) {
std::cout &lt;&lt; &quot;Generic std::exception catch handler&quot; &lt;&lt; std::endl;
jclass clazz = jenv-&gt;FindClass(&quot;java/lang/RuntimeException&quot;);
jenv-&gt;ThrowNew(clazz, e.what());
return $null;
}
%}
// Expose C++ exception as a Java Exception by changing the Java base class and providing a getMessage()
%typemap(javabase) MyNS::MyException &quot;java.lang.RuntimeException&quot;
%rename(getMessage) MyNS::MyException::whatsup;
%inline %{
namespace MyNS {
class MyException {
std::string msg;
public:
MyException(const char *msg) : msg(msg) {}
const char * whatsup() const { return msg.c_str(); }
};
}
%}
%typemap(directorthrows) MyNS::MyException %{
if (Swig::ExceptionMatches(jenv, $error, &quot;$packagepath/MyException&quot;)) {
std::cout &lt;&lt; &quot;$1_type exception matched (directorthrows typemap)&quot; &lt;&lt; std::endl;
throw $1_type(Swig::JavaExceptionMessage(jenv, $error).message());
}
%}
%typemap(throws) MyNS::MyException %{
std::cout &lt;&lt; &quot;$1_type caught (throws typemap)&quot; &lt;&lt; std::endl;
jclass excep = jenv-&gt;FindClass(&quot;MyException&quot;);
if (excep) {
std::cout &lt;&lt; &quot;$1_type class found (throws typemap)&quot; &lt;&lt; std::endl;
jenv-&gt;ThrowNew(excep, $1.whatsup());
}
return $null;
%}
// These are the exceptions that the director method MyClass::dirmethod will have catch handlers for.
// Note that this is also a virtual method / director method and the C++ exceptions listed can be
// thrown after converting them from Java exceptions.
%catches(MyNS::MyException, Swig::DirectorException) MyClass::dirmethod;
// These are the exceptions that call_dirmethod C++ wrapper will have catch handlers for.
// Note that this is not a virtual method, hence not a director method.
%catches(MyNS::MyException, Swig::DirectorException) call_dirmethod;
%feature(&quot;director&quot;) MyClass;
%feature(&quot;director:except&quot;) MyClass::dirmethod(int x) {
jthrowable $error = jenv-&gt;ExceptionOccurred();
if ($error) {
std::cout &lt;&lt; &quot;Upcall finished, an exception was thrown in Java&quot; &lt;&lt; std::endl;
$directorthrowshandlers
std::cout &lt;&lt; &quot;Upcall finished, no exception conversion, throwing DirectorException&quot; &lt;&lt; std::endl;
Swig::DirectorException::raise(jenv, $error);
}
}
%inline %{
class MyClass {
public:
/** Throws either a std::out_of_range or MyException on error */
virtual void dirmethod(int x) {
if (x &lt;= 0)
throw std::out_of_range(&quot;MyClass::dirmethod index is out of range&quot;);
else if (x == 1)
throw MyNS::MyException(&quot;MyClass::dirmethod some problem!&quot;);
}
virtual ~MyClass() {}
static void call_dirmethod(MyClass&amp; c, int x) {
return c.dirmethod(x);
}
};
%}
</pre>
</div></p>
<p> The generated code for the <tt>call_dirmethod</tt> wrapper contains
the various exception handlers. The outer exception handler is from the
<tt>%exception</tt> directive and the others are from the &quot;throws&quot;
typemaps.</p>
<div class="code">
<pre>
SWIGEXPORT void JNICALL Java_exampleJNI_MyClass_1call_1dirmethod(JNIEnv *jenv, jclass jcls, jlong jarg1, jobject jarg1_, jint jarg2) {
...
try {
try {
MyClass::call_dirmethod(*arg1,arg2);
} catch(MyNS::MyException &amp;_e) {
std::cout &lt;&lt; &quot;MyNS::MyException caught (throws typemap)&quot; &lt;&lt; std::endl;
jclass excep = jenv-&gt;FindClass(&quot;MyException&quot;);
if (excep) {
std::cout &lt;&lt; &quot;MyNS::MyException class found (throws typemap)&quot; &lt;&lt; std::endl;
jenv-&gt;ThrowNew(excep, (&amp;_e)-&gt;whatsup());
}
return ;
} catch(Swig::DirectorException &amp;_e) {
(&amp;_e)-&gt;throwException(jenv);
return ;
}
} catch (const std::exception &amp;e) {
std::cout &lt;&lt; &quot;Generic std::exception catch handler&quot; &lt;&lt; std::endl;
jclass clazz = jenv-&gt;FindClass(&quot;java/lang/RuntimeException&quot;);
jenv-&gt;ThrowNew(clazz, e.what());
return ;
}
</pre>
</div>
<p> The director method calling up to Java contains the exception
handling code from the &quot;directorthrows&quot; typemaps and <tt>
director:except</tt> feature.</p>
<div class="code">
<pre>
void SwigDirector_MyClass::dirmethod(int x) {
... [call up to Java using CallStaticVoidMethod]
jthrowable swigerror = jenv-&gt;ExceptionOccurred();
if (swigerror) {
std::cout &lt;&lt; &quot;Upcall finished, an exception was thrown in Java&quot; &lt;&lt; std::endl;
if (Swig::ExceptionMatches(jenv, swigerror, &quot;MyException&quot;)) {
std::cout &lt;&lt; &quot;MyNS::MyException exception matched (directorthrows typemap)&quot; &lt;&lt; std::endl;
throw MyNS::MyException(Swig::JavaExceptionMessage(jenv, swigerror).message());
}
std::cout &lt;&lt; &quot;Upcall finished, no exception conversion, throwing DirectorException&quot; &lt;&lt; std::endl;
Swig::DirectorException::raise(jenv, swigerror);
}
</pre>
</div>
<p> Let's use the following Java class to override the director method.</p>
<div class="code">
<pre>
class DerivedClass extends MyClass {
@Override
public void dirmethod(int x) {
if (x &lt; 0)
throw new IndexOutOfBoundsException(&quot;Index is negative&quot;);
else if (x == 0)
throw new MyException(&quot;MyException: bad dirmethod&quot;);
}
}
public class runme {
public static void main(String argv[]) {
System.loadLibrary(&quot;example&quot;);
... code snippets shown below ...
}
}
</pre>
</div>
<p> Consider the output using the Java code in the four slightly
different scenarios below.</p>
<p> 1. Non-director C++ class is used, thus, no upcall to a Java
director method is made. A <tt>std::out_of_range</tt> exception is
thrown, which is derived from <tt>std::exception</tt>, and hence caught
by the generic exception handler in the <tt>call_dirmethod</tt>
wrapper. The Java code snippet and resulting output is:</p>
<div class="code">
<pre>
MyClass.call_dirmethod(new MyClass(), 0);
</pre>
</div><div class="shell">
<pre>
Generic std::exception catch handler
Exception in thread &quot;main&quot; java.lang.RuntimeException: MyClass::dirmethod index is out of range
at exampleJNI.MyClass_call_dirmethod(Native Method)
at MyClass.call_dirmethod(MyClass.java:57)
at runme.main(runme.java:14)
</pre>
</div>
<p> 2. Non-director C++ class again but this time the <tt>
MyNS::MyException</tt> class is thrown and caught:</p>
<div class="code">
<pre>
MyClass.call_dirmethod(new MyClass(), 1);
</pre>
</div><div class="shell">
<pre>
MyNS::MyException caught (throws typemap)
MyNS::MyException class found (throws typemap)
Exception in thread &quot;main&quot; MyException: MyClass::dirmethod some problem!
at exampleJNI.MyClass_call_dirmethod(Native Method)
at MyClass.call_dirmethod(MyClass.java:57)
at runme.main(runme.java:15)
</pre>
</div>
<p> 3. The <tt>DerivedClass</tt> director class is used so the upcall to
Java occurs, but it throws a Java <tt>MyException</tt>, which gets
converted into a C++ <tt>MyNS::MyException</tt>, then caught and
converted back into a Java <tt>MyException</tt>:</p>
<div class="code">
<pre>
MyClass.call_dirmethod(new DerivedClass(), 0);
</pre>
</div><div class="shell">
<pre>
Upcall finished, an exception was thrown in Java
MyNS::MyException exception matched (directorthrows typemap)
MyNS::MyException caught (throws typemap)
MyNS::MyException class found (throws typemap)
Exception in thread &quot;main&quot; MyException: MyException: bad dirmethod
at exampleJNI.MyClass_call_dirmethod(Native Method)
at MyClass.call_dirmethod(MyClass.java:57)
at runme.main(runme.java:16)
</pre>
</div>
<p> 4. The director class is used again, but this time the director
method throws a Java <tt>IndexOutOfBoundsException</tt> exception which
is converted into a C++ <tt>Swig::DirectorException</tt>, thrown and
caught again. This time the original Java exception is extracted from
the <tt>Swig::DirectorException</tt> and rethrown. Note that this
approach keeps the stack trace information of the original exception,
so it has the exact location of where the <tt>IndexOutOfBoundsException</tt>
exception was thrown. This is arguably an improvement over the approach
above that converts from a Java excepton to C++ exception and then back
to a new Java exception, losing the location of the original exception.</p>
<div class="code">
<pre>
MyClass.call_dirmethod(new DerivedClass(), -1);
</pre>
</div><div class="shell">
<pre>
Upcall finished, an exception was thrown in Java
Upcall finished, no exception conversion, throwing DirectorException
Exception in thread &quot;main&quot; java.lang.IndexOutOfBoundsException: Index is negative
at DerivedClass.dirmethod(runme.java:5)
at exampleJNI.SwigDirector_MyClass_dirmethod(exampleJNI.java:23)
at exampleJNI.MyClass_call_dirmethod(Native Method)
at MyClass.call_dirmethod(MyClass.java:57)
at runme.main(runme.java:17)
</pre>
</div>
<h3><a name="Java_protected_virtual_methods">27.5.8 Accessing virtual
protected methods</a></h3>
<p> By default, without enabling the director feature, protected methods
are not wrapped and so cannot be accessed from Java. When using
directors, the protected virtual methods are wrapped. These methods are
wrapped with a protected Java proxy method, so the only way that Java
code can access these is from within a Java class derived from the
director class.</p>
<p> Unfortunately the rules for classes in an inheritance chain when
changing the access specifiers in C++ are different to Java access
modifiers. They are similar, but, unlike in C+++, one cannot assign
weaker access modifiers in Java. Consider the following inheritance
chain and using directors to turn on wrapping protected methods:</p>
<div class="code">
<pre>
%feature(&quot;director&quot;) BaseClass;
%feature(&quot;director&quot;) DerivedClass;
%inline %{
class BaseClass {
public:
virtual ~BaseClass();
virtual void baseMethod1();
virtual void baseMethod2();
protected:
virtual void baseMethod3();
virtual void baseMethod4();
};
class DerivedClass : public BaseClass {
public:
virtual ~DerivedClass();
virtual void baseMethod1();
protected:
virtual void baseMethod2();
virtual void baseMethod3();
public:
virtual void baseMethod4();
};
%}
</pre>
</div>
<p> SWIG will generate Java access modifiers that are a one-to-one
mapping of the C++ access specifiers. So if a method is <tt>public</tt>
in C++ it will also be <tt>public</tt> in Java, and similarly <tt>
protected</tt> in C++ becomes <tt>protected</tt> in Java. This is fine
for all of the methods in the above example, except for the <tt>
DerivedClass.baseMethod2</tt>. The Java compiler issues an error:</p>
<div class="shell">
<pre>
DerivedClass.java:69: error: baseMethod2() in DerivedClass cannot override baseMethod2() in BaseClass
protected void baseMethod2() {
^
attempting to assign weaker access privileges; was public
</pre>
</div>
<p> This requires fixing with one of two solutions:</p>
<ol>
<li>
<p>Change the access modifier to public for the derived class by adding
the following before SWIG parses the <tt>DerivedClass:</tt></p>
<div class="code">
<pre>
%feature(&quot;java:methodmodifiers&quot;) DerivedClass::baseMethod2 &quot;public&quot;
</pre>
</div></li>
<li>
<p>Simply ignore the method in the derived class altogether as it cannot
be used as a <tt>protected</tt> method in Java as the C++ design
intended:</p>
<div class="code">
<pre>
%ignore DerivedClass::baseMethod2;
</pre>
</div></li>
</ol>
<h3><a name="Java_allprotected">27.5.9 Accessing non-virtual protected
members</a></h3>
<p> Members which are protected and non-virtual can also be accessed
when using the 'allprotected' mode. The allprotected mode requires
directors and is turned on by setting the <tt>allprotected</tt> option
in addition to the <tt>directors</tt> option in the %module directive,
like this:</p>
<div class="code">
<pre>
%module(directors=&quot;1&quot;, allprotected=&quot;1&quot;) modulename
</pre>
</div>
<p> Protected member variables and methods (both static and non-static)
will then be wrapped with protected access in the Java proxy class.</p>
<p><b> Note:</b> Neither the directors option nor the allprotected mode
support types defined with protected scope. This includes any enums or
typedefs declared in the protected section of the C++ class.</p>
<p> The following simple example is a class with numerous protected
members, including the constructor and destructor:</p>
<div class="code">
<pre>
%module(directors=&quot;1&quot;, allprotected=&quot;1&quot;) example
%feature(&quot;director&quot;) ProtectedBase;
// Ignore use of unsupported types (those defined in the protected section)
%ignore ProtectedBase::typedefs;
%inline %{
class ProtectedBase {
protected:
ProtectedBase() {}
virtual ~ProtectedBase() {}
virtual void virtualMethod() const {}
void nonStaticMethod(double d) const {}
static void staticMethod(int i) {}
int instanceMemberVariable;
static int staticMemberVariable;
// unsupported: types defined with protected access and the methods/variables which use them
typedef int IntegerType;
IntegerType typedefs(IntegerType it) { return it; }
};
int ProtectedBase::staticMemberVariable = 10;
%}
</pre>
</div>
<p> Note that the <tt>IntegerType</tt> has protected scope and the
members which use this type must be ignored as they cannot be wrapped.</p>
<p> The proxy methods are protected, so the only way the protected
members can be accessed is within a class that derives from the
director class, such as the following:</p>
<div class="code">
<pre>
class MyProtectedBase extends ProtectedBase
{
public MyProtectedBase() {
}
public void accessProtected() {
virtualMethod();
nonStaticMethod(1.2);
staticMethod(99);
setInstanceMemberVariable(5);
int i = getInstanceMemberVariable();
setStaticMemberVariable(10);
i = getStaticMemberVariable();
}
}
</pre>
</div>
<h2><a name="Java_common_customization">27.6 Common customization
features</a></h2>
<p> An earlier section presented the absolute basics of C/C++ wrapping.
If you do nothing but feed SWIG a header file, you will get an
interface that mimics the behavior described. However, sometimes this
isn't enough to produce a nice module. Certain types of functionality
might be missing or the interface to certain functions might be
awkward. This section describes some common SWIG features that are used
to improve the interface to existing C/C++ code.</p>
<h3><a name="Java_helper_functions">27.6.1 C/C++ helper functions</a></h3>
<p> Sometimes when you create a module, it is missing certain bits of
functionality. For example, if you had a function like this</p>
<div class="code">
<pre>
typedef struct Image {...};
void set_transform(Image *im, double m[4][4]);
</pre>
</div>
<p> it would be accessible from Java, but there may be no easy way to
call it. The problem here is that a type wrapper class is generated for
the two dimensional array parameter so there is no easy way to
construct and manipulate a suitable <tt>double [4][4]</tt> value. To
fix this, you can write some extra C helper functions. Just use the <tt>
%inline</tt> directive. For example:</p>
<div class="code">
<pre>
%inline %{
/* Note: double[4][4] is equivalent to a pointer to an array double (*)[4] */
double (*new_mat44())[4] {
return (double (*)[4]) malloc(16*sizeof(double));
}
void free_mat44(double (*x)[4]) {
free(x);
}
void mat44_set(double x[4][4], int i, int j, double v) {
x[i][j] = v;
}
double mat44_get(double x[4][4], int i, int j) {
return x[i][j];
}
%}
</pre>
</div>
<p> From Java, you could then write code like this:</p>
<div class="code">
<pre>
Image im = new Image();
SWIGTYPE_p_a_4__double a = example.new_mat44();
example.mat44_set(a, 0, 0, 1.0);
example.mat44_set(a, 1, 1, 1.0);
example.mat44_set(a, 2, 2, 1.0);
...
example.set_transform(im, a);
example.free_mat44(a);
</pre>
</div>
<p> Admittedly, this is not the most elegant looking approach. However,
it works and it wasn't too hard to implement. It is possible to improve
on this using Java code, typemaps, and other customization features as
covered in later sections, but sometimes helper functions are a quick
and easy solution to difficult cases.</p>
<h3><a name="Java_class_extension">27.6.2 Class extension with %extend</a>
</h3>
<p> One of the more interesting features of SWIG is that it can extend
structures and classes with new methods or constructors. Here is a
simple example:</p>
<div class="code">
<pre>
%module example
%{
#include &quot;someheader.h&quot;
%}
struct Vector {
double x, y, z;
};
%extend Vector {
char *toString() {
static char tmp[1024];
sprintf(tmp, &quot;Vector(%g, %g, %g)&quot;, $self-&gt;x, $self-&gt;y, $self-&gt;z);
return tmp;
}
Vector(double x, double y, double z) {
Vector *v = (Vector *) malloc(sizeof(Vector));
v-&gt;x = x;
v-&gt;y = y;
v-&gt;z = z;
return v;
}
};
</pre>
</div>
<p> Now, in Java</p>
<div class="code">
<pre>
Vector v = new Vector(2, 3, 4);
System.out.println(v);
</pre>
</div>
<p> will display</p>
<div class="code">
<pre>
Vector(2, 3, 4)
</pre>
</div>
<p> <tt>%extend</tt> works with both C and C++ code. It does not modify
the underlying object in any way---the extensions only show up in the
Java interface.</p>
<h3><a name="Java_proxycode">27.6.3 Class extension with %proxycode</a></h3>
<p> The previous section described how to extend a wrapped class with C
or C++ code. This section describes how to extend a wrapped class with
Java code instead of C/C++ code. The <tt>%proxycode</tt> directive is
used and is just a macro for <tt>%insert(&quot;proxycode&quot;)</tt>. The <a href="#SWIG_nn42">
Code insertion block</a> section describes the <tt>%insert</tt>
directive. The section of code for insertion is &quot;proxycode&quot;, that is,
the Java proxy class. This directive must hence only be used within the
scope of a class, otherwise it is silently ignored. There are two
common ways to get the scope correct.</p>
<p> The first is to use <tt>%proxycode</tt> inside a class that SWIG
parses, for example a <tt>toString()</tt> method can be added to a C++
class using pure Java code. A C++ header file can mix C++ and Java code
inside the C++ class as follows:</p>
<div class="code">
<pre>
// flag.h header file
class Flag {
bool flag;
public:
Flag(bool flag) : flag(flag) {}
bool FetchFlag() { return flag; }
#if defined(SWIG)
%proxycode %{
public String toString() {
boolean flag = FetchFlag();
return Boolean.toString(flag);
}
%}
#endif
};
</pre>
</div>
<p> and wrapped using:</p>
<div class="code">
<pre>
%{
#include &quot;flag.h&quot;
%}
%include &quot;flag.h&quot;
</pre>
</div>
<p> The second is to use <tt>%proxycode</tt> within <tt>%extend</tt> as
everything within a <tt>%extend</tt> block is effectively within the
scope of the class, for example:</p>
<div class="code">
<pre>
// flag.h header file
class Flag {
bool flag;
public:
Flag(bool flag) : flag(flag) {}
bool FetchFlag() { return flag; }
};
</pre>
</div>
<p> and wrapped using:</p>
<div class="code">
<pre>
%{
#include &quot;flag.h&quot;
%}
%include &quot;flag.h&quot;
%extend Flag {
#if defined(SWIG)
%proxycode %{
public String toString() {
boolean flag = FetchFlag();
return Boolean.toString(flag);
}
%}
#endif
}
</pre>
</div>
<p> There is some very limited support of typemaps within a <tt>
%proxycode</tt> block. A useful trick is to obtain the Java type for a
given C/C++ type using the <a href="#Typemaps_special_macro_typemap">
$typemap</a> special macro. The following C++ template demonstrates
this:</p>
<div class="code">
<pre>
%inline %{
template&lt;typename T&gt; struct Value {
T value;
Value(const T&amp; val) : value(val) {}
};
%}
%extend Value {
%proxycode %{
public String toString() {
// Note template type expansion is supported, so T is expanded to 'unsigned int' in this example
// and $typemap(jstype, unsigned int) in turn is expanded to 'long'
$typemap(jstype, T) val = getValue();
return &quot;$javaclassname value: &quot; + val + &quot; Java type: $typemap(jstype, T) JNI type: $typemap(jni, T)&quot;;
}
%}
}
%template(ValueUnsignedInt) Value&lt;unsigned int&gt;;
</pre>
</div>
<p> The generated Java contains the expanded special variable and macro
resulting in Java proxy code:</p>
<div class="code">
<pre>
public class ValueUnsignedInt {
...
public String toString() {
long val = getValue();
return &quot;ValueUnsignedInt value: &quot; + val + &quot; Java type: long JNI type: jlong&quot;;
}
}
</pre>
</div>
<h3><a name="Java_exception_handling">27.6.4 Exception handling with
%exception and %javaexception</a></h3>
<p> If a C or C++ function throws an error, you may want to convert that
error into a Java exception. To do this, you can use the <tt>%exception</tt>
directive. The <tt>%exception</tt> directive simply lets you rewrite
part of the generated wrapper code to include an error check. It is
detailed in full in the <a href="#Customization_exception">Exception
handling with %exception</a> section.</p>
<p> In C, a function often indicates an error by returning a status code
(a negative number or a NULL pointer perhaps). Here is a simple example
of how you might handle that:</p>
<div class="code">
<pre>
%exception malloc {
$action
if (!result) {
jclass clazz = (*jenv)-&gt;FindClass(jenv, &quot;java/lang/OutOfMemoryError&quot;);
(*jenv)-&gt;ThrowNew(jenv, clazz, &quot;Not enough memory&quot;);
return $null;
}
}
void *malloc(size_t nbytes);
</pre>
</div>
<p> In Java,</p>
<div class="code">
<pre>
SWIGTYPE_p_void a = example.malloc(2000000000);
</pre>
</div>
<p> will produce a familiar looking Java exception:</p>
<div class="code">
<pre>
Exception in thread &quot;main&quot; java.lang.OutOfMemoryError: Not enough memory
at exampleJNI.malloc(Native Method)
at example.malloc(example.java:16)
at runme.main(runme.java:112)
</pre>
</div>
<p> If a library provides some kind of general error handling framework,
you can also use that. For example:</p>
<div class="code">
<pre>
%exception malloc {
$action
if (err_occurred()) {
jclass clazz = (*jenv)-&gt;FindClass(jenv, &quot;java/lang/OutOfMemoryError&quot;);
(*jenv)-&gt;ThrowNew(jenv, clazz, &quot;Not enough memory&quot;);
return $null;
}
}
void *malloc(size_t nbytes);
</pre>
</div>
<p> If no declaration name is given to <tt>%exception</tt>, it is
applied to all wrapper functions. The <tt>$action</tt> is a SWIG
special variable and is replaced by the C/C++ function call being
wrapped. The <tt>return $null;</tt> handles all native method return
types, namely those that have a void return and those that do not. This
is useful for typemaps that will be used in native method returning all
return types. See the section on <a href="#Java_special_variables">Java
special variables</a> for further explanation.</p>
<p> C++ exceptions are also easy to handle. We can catch the C++
exception and rethrow it as a Java exception like this:</p>
<div class="code">
<pre>
%exception getitem {
try {
$action
} catch (std::out_of_range &amp;e) {
jclass clazz = jenv-&gt;FindClass(&quot;java/lang/Exception&quot;);
jenv-&gt;ThrowNew(clazz, &quot;Range error&quot;);
return $null;
}
}
class FooClass {
public:
FooClass *getitem(int index); // Might throw std::out_of_range exception
...
};
</pre>
</div>
<p> In the example above, <tt>java.lang.Exception</tt> is a checked
exception class and so ought to be declared in the throws clause of <tt>
getitem</tt>. Classes can be specified for adding to the throws clause
using <tt>%javaexception(classes)</tt> instead of <tt>%exception</tt>,
where <tt>classes</tt> is a string containing one or more comma
separated Java classes. The <tt>%clearjavaexception</tt> feature is the
equivalent to <tt>%clearexception</tt> and clears previously declared
exception handlers. The <tt>%nojavaexception</tt> feature is the
equivalent to <tt>%noexception</tt> and disables the exception handler.
See <a href="#Customization_clearing_features">Clearing features</a>
for the difference on disabling and clearing features.</p>
<div class="code">
<pre>
%javaexception(&quot;java.lang.Exception&quot;) getitem {
try {
$action
} catch (std::out_of_range &amp;e) {
jclass clazz = jenv-&gt;FindClass(&quot;java/lang/Exception&quot;);
jenv-&gt;ThrowNew(clazz, &quot;Range error&quot;);
return $null;
}
}
class FooClass {
public:
FooClass *getitem(int index); // Might throw std::out_of_range exception
...
};
</pre>
</div>
<p> The generated proxy method now generates a throws clause containing <tt>
java.lang.Exception</tt>:</p>
<div class="code">
<pre>
public class FooClass {
...
public FooClass getitem(int index) throws java.lang.Exception { ... }
...
}
</pre>
</div>
<p> The examples above first use the C JNI calling syntax then the C++
JNI calling syntax. The C++ calling syntax will not compile as C and
also vice versa. It is however possible to write JNI calls which will
compile under both C and C++ and is covered in the <a href="#Java_typemaps_for_c_and_cpp">
Typemaps for both C and C++ compilation</a> section.</p>
<p> The language-independent <tt>exception.i</tt> library file can also
be used to raise exceptions. See the <a href="#Library">SWIG Library</a>
chapter. The typemap example <a href="#Java_exception_typemap">Handling
C++ exception specifications as Java exceptions</a> provides further
exception handling capabilities.</p>
<h3><a name="Java_method_access">27.6.5 Method access with
%javamethodmodifiers</a></h3>
<p> A Java feature called <tt>%javamethodmodifiers</tt> can be used to
change the method modifiers from the default <tt>public</tt>. It
applies to both module class methods and proxy class methods. For
example:</p>
<div class="code">
<pre>
%javamethodmodifiers protect_me() &quot;protected&quot;;
void protect_me();
</pre>
</div>
<p> Will produce the method in the module class with protected access.</p>
<div class="code">
<pre>
protected static void protect_me() {
exampleJNI.protect_me();
}
</pre>
</div>
<h3><a name="Java_begin">27.6.6 Java begin</a></h3>
<p> It is possible to add a common comment at the start of every
generated Java file. The <tt>%module</tt> directive supports the <tt>
javabegin</tt> option for this. The provided text is generated at the
very beginning of each generated .java file. As it is generated before
the package statement, is only really useful for adding in a common
comment into all generated .java files. For example, copyright text for
each file:</p>
<div class="code">
<pre>
%module(javabegin=&quot;/* Common comment. Copyright (C) 2000 Mr Nobody. */\n&quot;) nobodymodule
</pre>
</div>
<h2><a name="Java_tips_techniques">27.7 Tips and techniques</a></h2>
<p> Although SWIG is largely automatic, there are certain types of
wrapping problems that require additional user input. Examples include
dealing with output parameters, strings and arrays. This chapter
discusses the common techniques for solving these problems.</p>
<h3><a name="Java_input_output_parameters">27.7.1 Input and output
parameters using primitive pointers and references</a></h3>
<p> A common problem in some C programs is handling parameters passed as
simple pointers or references. For example:</p>
<div class="code">
<pre>
void add(int x, int y, int *result) {
*result = x + y;
}
</pre>
</div>
<p> or perhaps</p>
<div class="code">
<pre>
int sub(int *x, int *y) {
return *x-*y;
}
</pre>
</div>
<p> The <tt>typemaps.i</tt> library file will help in these situations.
For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
void add(int, int, int *OUTPUT);
int sub(int *INPUT, int *INPUT);
</pre>
</div>
<p> In Java, this allows you to pass simple values. For example:</p>
<div class="code">
<pre>
int result = example.sub(7, 4);
System.out.println(&quot;7 - 4 = &quot; + result);
int[] sum = {0};
example.add(3, 4, sum);
System.out.println(&quot;3 + 4 = &quot; + sum[0]);
</pre>
</div>
<p> Which will display:</p>
<div class="code">
<pre>
7 - 4 = 3
3 + 4 = 7
</pre>
</div>
<p> Notice how the <tt>INPUT</tt> parameters allow integer values to be
passed instead of pointers and how the <tt>OUTPUT</tt> parameter will
return the result in the first element of the integer array.</p>
<p> If you don't want to use the names <tt>INPUT</tt> or <tt>OUTPUT</tt>
, use the <tt>%apply</tt> directive. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *result };
%apply int *INPUT { int *x, int *y};
void add(int x, int y, int *result);
int sub(int *x, int *y);
</pre>
</div>
<p> If a function mutates one of its parameters like this,</p>
<div class="code">
<pre>
void negate(int *x) {
*x = -(*x);
}
</pre>
</div>
<p> you can use <tt>INOUT</tt> like this:</p>
<div class="code">
<pre>
%include &quot;typemaps.i&quot;
...
void negate(int *INOUT);
</pre>
</div>
<p> In Java, the input parameter is the first element in a 1 element
array and is replaced by the output of the function. For example:</p>
<div class="code">
<pre>
int[] neg = {3};
example.negate(neg);
System.out.println(&quot;Negative of 3 = &quot; + neg[0]);
</pre>
</div>
<p> And no prizes for guessing the output:</p>
<div class="code">
<pre>
Negative of 3 = -3
</pre>
</div>
<p> These typemaps can also be applied to C++ references. The above
examples would work the same if they had been defined using references
instead of pointers. For example, the Java code to use the <tt>negate</tt>
function would be the same if it were defined either as it is above:</p>
<div class="code">
<pre>
void negate(int *INOUT);
</pre>
</div>
<p> or using a reference:</p>
<div class="code">
<pre>
void negate(int &amp;INOUT);
</pre>
</div>
<p> Note: Since most Java primitive types are immutable and are passed
by value, it is not possible to perform in-place modification of a type
passed as a parameter.</p>
<p> Be aware that the primary purpose of the <tt>typemaps.i</tt> file is
to support primitive datatypes. Writing a function like this</p>
<div class="code">
<pre>
void foo(Bar *OUTPUT);
</pre>
</div>
<p> will not have the intended effect since <tt>typemaps.i</tt> does not
define an OUTPUT rule for <tt>Bar</tt>.</p>
<h3><a name="Java_simple_pointers">27.7.2 Simple pointers</a></h3>
<p> If you must work with simple pointers such as <tt>int *</tt> or <tt>
double *</tt> another approach to using <tt>typemaps.i</tt> is to use
the <tt>cpointer.i</tt> pointer library file. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;cpointer.i&quot;
%inline %{
extern void add(int x, int y, int *result);
%}
%pointer_functions(int, intp);
</pre>
</div>
<p> The <tt>%pointer_functions(type, name)</tt> macro generates five
helper functions that can be used to create, destroy, copy, assign, and
dereference a pointer. In this case, the functions are as follows:</p>
<div class="code">
<pre>
int *new_intp();
int *copy_intp(int *x);
void delete_intp(int *x);
void intp_assign(int *x, int value);
int intp_value(int *x);
</pre>
</div>
<p> In Java, you would use the functions like this:</p>
<div class="code">
<pre>
SWIGTYPE_p_int intPtr = example.new_intp();
example.add(3, 4, intPtr);
int result = example.intp_value(intPtr);
System.out.println(&quot;3 + 4 = &quot; + result);
</pre>
</div>
<p> If you replace <tt>%pointer_functions(int, intp)</tt> by <tt>
%pointer_class(int, intp)</tt>, the interface is more class-like.</p>
<div class="code">
<pre>
intp intPtr = new intp();
example.add(3, 4, intPtr.cast());
int result = intPtr.value();
System.out.println(&quot;3 + 4 = &quot; + result);
</pre>
</div>
<p> See the <a href="#Library">SWIG Library</a> chapter for further
details.</p>
<h3><a name="Java_c_arrays">27.7.3 Wrapping C arrays with Java arrays</a>
</h3>
<p> SWIG can wrap arrays in a more natural Java manner than the default
by using the <tt>arrays_java.i</tt> library file. Let's consider an
example:</p>
<div class="code">
<pre>
%include &quot;arrays_java.i&quot;;
int array[4];
void populate(int x[]) {
int i;
for (i=0; i&lt;4; i++)
x[i] = 100 + i;
}
</pre>
</div>
<p> These one dimensional arrays can then be used as if they were Java
arrays:</p>
<div class="code">
<pre>
int[] array = new int[4];
example.populate(array);
System.out.print(&quot;array: &quot;);
for (int i=0; i&lt;array.length; i++)
System.out.print(array[i] + &quot; &quot;);
example.setArray(array);
int[] global_array = example.getArray();
System.out.print(&quot;\nglobal_array: &quot;);
for (int i=0; i&lt;array.length; i++)
System.out.print(global_array[i] + &quot; &quot;);
</pre>
</div>
<p> Java arrays are always passed by reference, so any changes a
function makes to the array will be seen by the calling function. Here
is the output after running this code:</p>
<div class="code">
<pre>
array: 100 101 102 103
global_array: 100 101 102 103
</pre>
</div>
<p> Note that for assigning array variables the length of the C variable
is used, so it is possible to use a Java array that is bigger than the
C code will cope with. Only the number of elements in the C array will
be used. However, if the Java array is not large enough then you are
likely to get a segmentation fault or access violation, just like you
would in C. When arrays are used in functions like <tt>populate</tt>,
the size of the C array passed to the function is determined by the
size of the Java array.</p>
<p> Please be aware that the typemaps in this library are not efficient
as all the elements are copied from the Java array to a C array
whenever the array is passed to and from JNI code. There is an
alternative approach using the SWIG array library and this is covered
in the next section.</p>
<h3><a name="Java_unbounded_c_arrays">27.7.4 Unbounded C Arrays</a></h3>
<p> Sometimes a C function expects an array to be passed as a pointer.
For example,</p>
<div class="code">
<pre>
int sumitems(int *first, int nitems) {
int i, sum = 0;
for (i = 0; i &lt; nitems; i++) {
sum += first[i];
}
return sum;
}
</pre>
</div>
<p> One of the ways to wrap this is to apply the Java array typemaps
that come in the <tt>arrays_java.i</tt> library file:</p>
<div class="code">
<pre>
%include &quot;arrays_java.i&quot;
%apply int[] {int *};
</pre>
</div>
<p> The <tt>ANY</tt> size will ensure the typemap is applied to arrays
of all sizes. You could narrow the typemap matching rules by specifying
a particular array size. Now you can use a pure Java array and pass it
to the C code:</p>
<div class="code">
<pre>
int[] array = new int[10000000]; // Array of 10-million integers
for (int i=0; i&lt;array.length; i++) { // Set some values
array[i] = i;
}
int sum = example.sumitems(array, 10000);
System.out.println(&quot;Sum = &quot; + sum);
</pre>
</div>
<p> and the sum would be displayed:</p>
<div class="code">
<pre>
Sum = 49995000
</pre>
</div>
<p> This approach is probably the most natural way to use arrays.
However, it suffers from performance problems when using large arrays
as a lot of copying of the elements occurs in transferring the array
from the Java world to the C++ world. An alternative approach to using
Java arrays for C arrays is to use an alternative SWIG library file <tt>
carrays.i</tt>. This approach can be more efficient for large arrays as
the array is accessed one element at a time. For example:</p>
<div class="code">
<pre>
%include &quot;carrays.i&quot;
%array_functions(int, intArray);
</pre>
</div>
<p> The <tt>%array_functions(type, name)</tt> macro generates four
helper functions that can be used to create and destroy arrays and
operate on elements. In this case, the functions are as follows:</p>
<div class="code">
<pre>
int *new_intArray(size_t nelements);
void delete_intArray(int *x);
int intArray_getitem(int *x, size_t index);
void intArray_setitem(int *x, size_t index, int value);
</pre>
</div>
<p> In Java, you would use the functions like this:</p>
<div class="code">
<pre>
SWIGTYPE_p_int array = example.new_intArray(10000000); // Array of 10-million integers
for (int i=0; i&lt;10000; i++) { // Set some values
example.intArray_setitem(array, i, i);
}
int sum = example.sumitems(array, 10000);
System.out.println(&quot;Sum = &quot; + sum);
</pre>
</div>
<p> If you replace <tt>%array_functions(int, intp)</tt> by <tt>
%array_class(int, intp)</tt>, the interface is more class-like and a
couple more helper functions are available for casting between the
array and the type wrapper class.</p>
<div class="code">
<pre>
%include &quot;carrays.i&quot;
%array_class(int, intArray);
</pre>
</div>
<p> The <tt>%array_class(type, name)</tt> macro creates wrappers for an
unbounded array object that can be passed around as a simple pointer
like <tt>int *</tt> or <tt>double *</tt>. For instance, you will be
able to do this in Java:</p>
<div class="code">
<pre>
intArray array = new intArray(10000000); // Array of 10-million integers
for (int i=0; i&lt;10000; i++) { // Set some values
array.setitem(i, i);
}
int sum = example.sumitems(array.cast(), 10000);
System.out.println(&quot;Sum = &quot; + sum);
</pre>
</div>
<p> The array &quot;object&quot; created by <tt>%array_class()</tt> does not
encapsulate pointers inside a special array object. In fact, there is
no bounds checking or safety of any kind (just like in C). Because of
this, the arrays created by this library are extremely low-level
indeed. You can't iterate over them nor can you even query their
length. In fact, any valid memory address can be accessed if you want
(negative indices, indices beyond the end of the array, etc.). Needless
to say, this approach is not going to suit all applications. On the
other hand, this low-level approach is extremely efficient and well
suited for applications in which you need to create buffers, package
binary data, etc.</p>
<h3><a name="Java_string_length">27.7.5 Passing a string with length</a></h3>
<p> SWIG provides multi-argument typemap available as mentioned in <a href="#Library_nn10">
Passing a string with length</a>. The following simple example
demonstrates passing a string to the wrapped function.</p>
<div class="code">
<pre>
%apply (char *STRING, size_t LENGTH) { (const char data[], size_t len) }
%inline %{
void binaryChar1(const char data[], size_t len) {
printf(&quot;len: %d data: &quot;, len);
for (size_t i=0; i&lt;len; ++i)
printf(&quot;%x &quot;, data[i]);
printf(&quot;\n&quot;);
}
%}
</pre>
</div>
<p> Calling from Java requires a string to be passed in as the
multi-argument typemap being applied reduces the number of arguments in
the target language to one, from the original two:</p>
<div class="code">
<pre>
String str = &quot;hi\0jk&quot;;
example.binaryChar1(str);
</pre>
</div>
<p> resulting in the output</p>
<div class="code">
<pre>
$ java runme
len: 5 data: 68 69 0 6a 6b
</pre>
</div>
<p> The typemap uses Java <tt>String::getBytes()</tt> to convert the
string to the default system encoding. If you wish to use another
encoding, you can modify the typemap. For example to convert the string
to UTF-8 use:</p>
<div class="code">
<pre>
%typemap(javain, throws=&quot;java.io.IllegalCharsetNameException&quot;)
<br>
(const char *STRING, size_t LENGTH) %{($javainput == null) ? null : $javainput.getBytes(&quot;UTF-8&quot;)%}
</pre>
</div>
<h3><a name="Java_heap_allocations">27.7.6 Overriding new and delete to
allocate from Java heap</a></h3>
<p> Unlike some languages supported by SWIG, Java has a true garbage
collection subsystem. Other languages will free SWIG wrapped objects
when their reference count reaches zero. Java only schedules these
objects for finalization, which may not occur for some time. Because
SWIG objects are allocated on the C heap, Java users may find the JVM
memory use quickly exceeds the assigned limits, as memory fills with
unfinalized proxy objects. Forcing garbage collection is clearly an
undesirable solution.</p>
<p> An elegant fix for C++ users is to override new and delete using the
following code (here shown included in a SWIG interface file)</p>
<div class="code">
<pre>
/* File: java_heap.i */
%module test
%{
#include &lt;stdexcept&gt;
#include &quot;jni.h&quot;
/**
* A stash area embedded in each allocation to hold java handles
*/
struct Jalloc {
jbyteArray jba;
jobject ref;
};
static JavaVM *cached_jvm = 0;
JNIEXPORT jint JNICALL JNI_OnLoad(JavaVM *jvm, void *reserved) {
cached_jvm = jvm;
return JNI_VERSION_1_2;
}
static JNIEnv * JNU_GetEnv() {
JNIEnv *env;
jint rc = cached_jvm-&gt;GetEnv((void **)&amp;env, JNI_VERSION_1_2);
if (rc == JNI_EDETACHED)
throw std::runtime_error(&quot;current thread not attached&quot;);
if (rc == JNI_EVERSION)
throw std::runtime_error(&quot;jni version not supported&quot;);
return env;
}
void * operator new(size_t t) {
if (cached_jvm != 0) {
JNIEnv *env = JNU_GetEnv();
jbyteArray jba = env-&gt;NewByteArray((int) t + sizeof(Jalloc));
if (env-&gt;ExceptionOccurred())
throw bad_alloc();
void *jbuffer = static_cast&lt;void *&gt;(env-&gt;GetByteArrayElements(jba, 0));
if (env-&gt;ExceptionOccurred())
throw bad_alloc();
Jalloc *pJalloc = static_cast&lt;Jalloc *&gt;(jbuffer);
pJalloc-&gt;jba = jba;
/* Assign a global reference so byte array will persist until delete'ed */
pJalloc-&gt;ref = env-&gt;NewGlobalRef(jba);
if (env-&gt;ExceptionOccurred())
throw bad_alloc();
return static_cast&lt;void *&gt;(static_cast&lt;char *&gt;(jbuffer) + sizeof(Jalloc));
}
else { /* JNI_OnLoad not called, use malloc and mark as special */
Jalloc *pJalloc = static_cast&lt;Jalloc *&gt;(malloc((int) t + sizeof(Jalloc)));
if (!pJalloc)
throw bad_alloc();
pJalloc-&gt;ref = 0;
return static_cast&lt;void *&gt;(
static_cast&lt;char *&gt;(static_cast&lt;void *&gt;(pJalloc)) + sizeof(Jalloc));
}
}
void operator delete(void *v) {
if (v != 0) {
void *buffer = static_cast&lt;void *&gt;( static_cast&lt;char *&gt;(v) - sizeof(Jalloc));
Jalloc *pJalloc = static_cast&lt;Jalloc *&gt;(buffer);
if (pJalloc-&gt;ref) {
JNIEnv *env = JNU_GetEnv();
env-&gt;DeleteGlobalRef(pJalloc-&gt;ref);
env-&gt;ReleaseByteArrayElements(pJalloc-&gt;jba, static_cast&lt;jbyte *&gt;(buffer), 0);
}
else {
free(buffer);
}
}
}
%}
...
</pre>
</div>
<p> This code caches the Java environment during initialization, and
when new is called, a Java ByteArray is allocated to provide the SWIG
objects with space in the Java heap. This has the combined effect of
re-asserting the Java virtual machine's limit on memory allocation, and
puts additional pressure on the garbage collection system to run more
frequently. This code is made slightly more complicated because
allowances must be made if new is called before the JNI_OnLoad is
executed. This can happen during static class initialization, for
example.</p>
<p> Unfortunately, because most Java implementations call malloc and
free, this solution will not work for C wrapped structures. However,
you are free to make functions that allocate and free memory from the
Java heap using this model and use these functions in place of malloc
and free in your own code.</p>
<h2><a name="Java_typemaps">27.8 Java typemaps</a></h2>
<p> This section describes how you can modify SWIG's default wrapping
behavior for various C/C++ datatypes using the <tt>%typemap</tt>
directive. You are advised to be familiar with the material in the &quot;<a href="#Typemaps">
Typemaps</a>&quot; chapter. While not absolutely essential knowledge, this
section assumes some familiarity with the Java Native Interface (JNI).
JNI documentation can be consulted either online at <a href="https://www.oracle.com/technetwork/java/index.html">
the Java web site</a> or from a good JNI book. The following two books
are recommended:</p>
<ul>
<li> Title: 'Essential JNI: Java Native Interface.' Author: Rob Gordon.
Publisher: Prentice Hall. ISBN: 0-13-679895-0.</li>
<li> Title: 'The Java Native Interface: Programmer's Guide and
Specification.' Author: Sheng Liang. Publisher: Addison-Wesley. ISBN:
0-201-32577-2. Also available <a href="http://java.sun.com/docs/books/jni">
online</a> at the Sun Developer Network.</li>
</ul>
<p> Before proceeding, it should be stressed that typemaps are not a
required part of using SWIG---the default wrapping behavior is enough
in most cases. Typemaps are only used if you want to change some aspect
of the generated code.</p>
<h3><a name="Java_default_primitive_type_mappings">27.8.1 Default
primitive type mappings</a></h3>
<p> The following table lists the default type mapping from Java to
C/C++.</p>
<table BORDER summary="Default primitive type mappings">
<tr><td><b>C/C++ type</b></td><td><b>Java type</b></td><td><b>JNI type</b>
</td></tr>
<tr><td>bool
<br> const bool &amp;</td><td>boolean</td><td>jboolean</td></tr>
<tr><td>char
<br>const char &amp;</td><td>char</td><td>jchar</td></tr>
<tr><td>signed char
<br>const signed char &amp;</td><td>byte</td><td>jbyte</td></tr>
<tr><td>unsigned char
<br>const unsigned char &amp;</td><td>short</td><td>jshort</td></tr>
<tr><td>short
<br>const short &amp;</td><td>short</td><td>jshort</td></tr>
<tr><td>unsigned short
<br> const unsigned short &amp;</td><td>int</td><td>jint</td></tr>
<tr><td>int
<br> const int &amp;</td><td>int</td><td>jint</td></tr>
<tr><td>unsigned int
<br> const unsigned int &amp;</td><td>long</td><td>jlong</td></tr>
<tr><td>long
<br>const long &amp;</td><td>int</td><td>jint</td></tr>
<tr><td>unsigned long
<br>const unsigned long &amp;</td><td>long</td><td>jlong</td></tr>
<tr><td>long long
<br> const long long &amp;</td><td>long</td><td>jlong</td></tr>
<tr><td>unsigned long long
<br>const unsigned long long &amp;</td><td>java.math.BigInteger</td><td>
jobject</td></tr>
<tr><td>float
<br>const float &amp;</td><td>float</td><td>jfloat</td></tr>
<tr><td>double
<br> const double &amp;</td><td>double</td><td>jdouble</td></tr>
<tr><td>size_t
<br> const size_t &amp;</td><td>long</td><td>jlong</td></tr>
<tr><td>char *
<br>char []</td><td>String</td><td>jstring</td></tr>
</table>
<p> Note that default mappings for the C long type is more suitable for
32-bit systems. If long is 64-bit, the full range can be obtained by
defining <tt>SWIGWORDSIZE64</tt> when invoking SWIG. The long type will
instead be mapped as follows:</p>
<table BORDER summary="SWIGWORDSIZE64 primitive type mappings">
<tr><td><b>C/C++ type</b></td><td><b>Java type</b></td><td><b>JNI type</b>
</td></tr>
<tr><td>long
<br> const long &amp;</td><td>long</td><td>jlong</td></tr>
<tr><td>unsigned long
<br>const unsigned long &amp;</td><td>java.math.BigInteger</td><td>jobject</td>
</tr>
</table>
<p> Note that when size_t is 64-bit in C, the full unsigned range is not
available. This can be fixed by applying the 64-bit unsigned long long
typemaps as follows:</p>
<div class="code">
<pre>
%apply unsigned long long { size_t };
%apply const unsigned long long &amp; { const size_t &amp; };
</pre>
</div>
<p> The net effect then changes from the default shown earlier to:</p>
<table BORDER summary="64-bit size_t primitive type mappings">
<tr><td><b>C/C++ type</b></td><td><b>Java type</b></td><td><b>JNI type</b>
</td></tr>
<tr><td>size_t
<br> const size_t &amp;</td><td>java.math.BigInteger</td><td>jobject</td></tr>
</table>
<p> Note that SWIG wraps the C <tt>char</tt> type as a character.
Pointers and arrays of this type are wrapped as strings. The <tt>signed
char</tt> type can be used if you want to treat <tt>char</tt> as a
signed number rather than a character. Also note that all const
references to primitive types are treated as if they are passed by
value.</p>
<p> Given the following C function:</p>
<div class="code">
<pre>
void func(unsigned short a, char *b, const long &amp;c, unsigned long long d);
</pre>
</div>
<p> The module class method would be:</p>
<div class="code">
<pre>
public static void func(int a, String b, int c, java.math.BigInteger d) {...}
</pre>
</div>
<p> The intermediary JNI class would use the same types:</p>
<div class="code">
<pre>
public final static native void func(int jarg1, String jarg2, int jarg3,
java.math.BigInteger jarg4);
</pre>
</div>
<p> and the JNI function would look like this:</p>
<div class="code">
<pre>
SWIGEXPORT void JNICALL Java_exampleJNI_func(JNIEnv *jenv, jclass jcls,
jint jarg1, jstring jarg2, jint jarg3, jobject jarg4) {...}
</pre>
</div>
<p> The mappings for C <tt>int</tt> and C <tt>long</tt> are appropriate
for 32 bit applications which are used in the 32 bit JVMs. There is no
perfect mapping between Java and C as Java doesn't support all the
unsigned C data types. However, the mappings allow the full range of
values for each C type from Java.</p>
<h3><a name="Java_default_non_primitive_typemaps">27.8.2 Default
typemaps for non-primitive types</a></h3>
<p> The previous section covered the primitive type mappings.
Non-primitive types such as classes and structs are mapped using
pointers on the C/C++ side and storing the pointer into a Java <tt>long</tt>
variable which is held by the proxy class or type wrapper class. This
applies whether the type is marshalled as a pointer, by reference or by
value. It also applies for any unknown/incomplete types which use type
wrapper classes.</p>
<p> So in summary, the C/C++ pointer to non-primitive types is cast into
the 64 bit Java <tt>long</tt> type and therefore the JNI type is a <tt>
jlong</tt>. The Java type is either the proxy class or type wrapper
class.</p>
<h3><a name="Java_jvm64">27.8.3 Sixty four bit JVMs</a></h3>
<p> If you are using a 64 bit JVM you may have to override the C long,
but probably not C int default mappings. Mappings will be system
dependent, for example long will need remapping on Unix LP64 systems
(long, pointer 64 bits, int 32 bits), but not on Microsoft 64 bit
Windows which will be using a P64 IL32 (pointer 64 bits and int, long
32 bits) model. This may be automated in a future version of SWIG. Note
that the Java write once run anywhere philosophy holds true for all
pure Java code when moving to a 64 bit JVM. Unfortunately it won't of
course hold true for JNI code.</p>
<h3><a name="Java_what_is_typemap">27.8.4 What is a typemap?</a></h3>
<p> A typemap is nothing more than a code generation rule that is
attached to a specific C datatype. For example, to convert integers
from Java to C, you might define a typemap like this:</p>
<div class="code">
<pre>
%module example
%typemap(in) int {
$1 = $input;
printf(&quot;Received an integer : %d\n&quot;, $1);
}
%inline %{
extern int fact(int nonnegative);
%}
</pre>
</div>
<p> Typemaps are always associated with some specific aspect of code
generation. In this case, the &quot;in&quot; method refers to the conversion of
input arguments to C/C++. The datatype <tt>int</tt> is the datatype to
which the typemap will be applied. The supplied C code is used to
convert values. In this code a number of special variables prefaced by
a <tt>$</tt> are used. The <tt>$1</tt> variable is a placeholder for a
local variable of type <tt>int</tt>. The <tt>$input</tt> variable
contains the Java data, the JNI <tt>jint</tt> in this case.</p>
<p> When this example is compiled into a Java module, it can be used as
follows:</p>
<div class="code">
<pre>
System.out.println(example.fact(6));
</pre>
</div>
<p> and the output will be:</p>
<div class="code">
<pre>
Received an integer : 6
720
</pre>
</div>
<p> In this example, the typemap is applied to all occurrences of the <tt>
int</tt> datatype. You can refine this by supplying an optional
parameter name. For example:</p>
<div class="code">
<pre>
%module example
%typemap(in) int nonnegative {
$1 = $input;
printf(&quot;Received an integer : %d\n&quot;, $1);
}
%inline %{
extern int fact(int nonnegative);
%}
</pre>
</div>
<p> In this case, the typemap code is only attached to arguments that
exactly match <tt>int nonnegative</tt>.</p>
<p> The application of a typemap to specific datatypes and argument
names involves more than simple text-matching--typemaps are fully
integrated into the SWIG C++ type-system. When you define a typemap for
<tt>int</tt>, that typemap applies to <tt>int</tt> and qualified
variations such as <tt>const int</tt>. In addition, the typemap system
follows <tt>typedef</tt> declarations. For example:</p>
<div class="code">
<pre>
%typemap(in) int nonnegative {
$1 = $input;
printf(&quot;Received an integer : %d\n&quot;, $1);
}
%inline %{
typedef int Integer;
extern int fact(Integer nonnegative); // Above typemap is applied
%}
</pre>
</div>
<p> However, the matching of <tt>typedef</tt> only occurs in one
direction. If you defined a typemap for <tt>Integer</tt>, it is not
applied to arguments of type <tt>int</tt>.</p>
<p> Typemaps can also be defined for groups of consecutive arguments.
For example:</p>
<div class="code">
<pre>
%typemap(in) (char *str, int len) {
...
};
int count(char c, char *str, int len);
</pre>
</div>
<p> When a multi-argument typemap is defined, the arguments are always
handled as a single Java parameter. This allows the function to be used
like this (notice how the length parameter is omitted):</p>
<div class="code">
<pre>
int c = example.count('e', &quot;Hello World&quot;);
</pre>
</div>
<h3><a name="Java_typemaps_c_to_java_types">27.8.5 Typemaps for mapping
C/C++ types to Java types</a></h3>
<p> The typemaps available to the Java module include the common
typemaps listed in the main typemaps section. There are a number of
additional typemaps which are necessary for using SWIG with Java. The
most important of these implement the mapping of C/C++ types to Java
types:</p>
<br> &nbsp;
<table BORDER summary="Typemap mappings for C/C++ types to Java types">
<tr><td><b>Typemap</b></td><td><b>Description</b></td></tr>
<tr><td>jni</td><td>JNI C types. These provide the default mapping of
types from C/C++ to JNI for use in the JNI (C/C++) code.</td></tr>
<tr><td>jtype</td><td>Java intermediary types. These provide the default
mapping of types from C/C++ to Java for use in the native functions in
the intermediary JNI class. The type must be the equivalent Java type
for the JNI C type specified in the &quot;jni&quot; typemap.</td></tr>
<tr><td>jstype</td><td>Java types. These provide the default mapping of
types from C/C++ to Java for use in the Java module class, proxy
classes and type wrapper classes.</td></tr>
<tr><td>javain</td><td>Conversion from jstype to jtype. These are Java
code typemaps which transform the type used in the Java module class,
proxy classes and type wrapper classes (as specified in the &quot;jstype&quot;
typemap) to the type used in the Java intermediary JNI class (as
specified in the &quot;jtype&quot; typemap). In other words the typemap provides
the conversion to the native method call parameter types.</td></tr>
<tr><td>javaout</td><td>Conversion from jtype to jstype. These are Java
code typemaps which transform the type used in the Java intermediary
JNI class (as specified in the &quot;jtype&quot; typemap) to the Java type used
in the Java module class, proxy classes and type wrapper classes (as
specified in the &quot;jstype&quot; typemap). In other words the typemap provides
the conversion from the native method call return type.</td></tr>
<tr><td>jboxtype</td><td>Java boxed type. These are Java code typemaps
to provide the Java boxed type, such as, <tt>Integer</tt> for C type <tt>
int</tt>. As autoboxing is only relevant to the Java primitive types,
these are only provided for the C types that map to Java primitive
types. This typemap is usually only used by C++ STL container wrappers
that are wrapped by Java generic types as the boxed type must be used
instead of the unboxed/primitive type when declaring a Java generic
type.</td></tr>
<tr><td>javadirectorin</td><td>Conversion from jtype to jstype for
director methods. These are Java code typemaps which transform the type
used in the Java intermediary JNI class (as specified in the &quot;jtype&quot;
typemap) to the Java type used in the Java module class, proxy classes
and type wrapper classes (as specified in the &quot;jstype&quot; typemap). This
typemap provides the conversion for the parameters in the director
methods when calling up from C++ to Java. See <a href="#Java_directors_typemaps">
Director typemaps</a>.</td></tr>
<tr><td>javadirectorout</td><td>Conversion from jstype to jtype for
director methods. These are Java code typemaps which transform the type
used in the Java module class, proxy classes and type wrapper classes
(as specified in the &quot;jstype&quot; typemap) to the type used in the Java
intermediary JNI class (as specified in the &quot;jtype&quot; typemap). This
typemap provides the conversion for the return type in the director
methods when returning from the C++ to Java upcall. See <a href="#Java_directors_typemaps">
Director typemaps</a>.</td></tr>
<tr><td>directorin</td><td>Conversion from C++ type to jni type for
director methods. These are C++ typemaps which convert the parameters
used in the C++ director method to the appropriate JNI intermediary
type. The conversion is done in JNI code prior to calling the Java
function from the JNI code. See <a href="#Java_directors_typemaps">
Director typemaps</a>.</td></tr>
<tr><td>directorout</td><td>Conversion from jni type to C++ type for
director methods. These are C++ typemaps which convert the JNI return
type used in the C++ director method to the appropriate C++ return
type. The conversion is done in JNI code after calling the Java
function from the JNI code. See <a href="#Java_directors_typemaps">
Director typemaps</a>.</td></tr>
</table>
<p> If you are writing your own typemaps to handle a particular type,
you will normally have to write a collection of them. The default
typemaps are in &quot;<tt>java.swg</tt>&quot; and so might be a good place for
finding typemaps to base any new ones on.</p>
<p> The &quot;jni&quot;, &quot;jtype&quot; and &quot;jstype&quot; typemaps are usually defined
together to handle the Java to C/C++ type mapping. An &quot;in&quot; typemap
should be accompanied by a &quot;javain&quot; typemap and likewise an &quot;out&quot;
typemap by a &quot;javaout&quot; typemap. If an &quot;in&quot; typemap is written, a
&quot;freearg&quot; and &quot;argout&quot; typemap may also need to be written as some
types have a default &quot;freearg&quot; and/or &quot;argout&quot; typemap which may need
overriding. The &quot;freearg&quot; typemap sometimes releases memory allocated
by the &quot;in&quot; typemap. The &quot;argout&quot; typemap sometimes sets values in
function parameters which are passed by reference in Java.</p>
<p> Note that the &quot;in&quot; typemap marshals the JNI type held in the &quot;jni&quot;
typemap to the real C/C++ type and for the opposite direction, the
&quot;out&quot; typemap marshals the real C/C++ type to the JNI type held in the
&quot;jni&quot; typemap. For <a href="#Java_default_non_primitive_typemaps">
non-primitive types</a> the &quot;in&quot; and &quot;out&quot; typemaps are responsible for
casting between the C/C++ pointer and the 64 bit <tt>jlong</tt> type.
There is no portable way to cast a pointer into a 64 bit integer type
and the approach taken by SWIG is mostly portable, but breaks C/C++
aliasing rules. In summary, these rules state that a pointer to any
type must never be dereferenced by a pointer to any other incompatible
type. The following code snippet might aid in understand aliasing rules
better:</p>
<div class="code">
<pre>
short a;
short* pa = 0;
int i = 0x1234;
a = (short)i; /* okay */
a = *(short*)&amp;i; /* breaks aliasing rules */
</pre>
</div>
<p> An email posting, <a href="http://mail-index.netbsd.org/tech-kern/2003/08/11/0001.html">
Aliasing, pointer casts and gcc 3.3</a> elaborates further on the
subject. In SWIG, the &quot;in&quot; and &quot;out&quot; typemaps for pointers are
typically</p>
<div class="code">
<pre>
%typemap(in) struct Foo * %{
$1 = *(struct Foo **)&amp;$input; /* cast jlong into C ptr */
%}
%typemap(out) struct Bar * %{
*(struct Bar **)&amp;$result = $1; /* cast C ptr into jlong */
%}
struct Bar {...};
struct Foo {...};
struct Bar * FooBar(struct Foo *f);
</pre>
</div>
<p> resulting in the following code which breaks the aliasing rules:</p>
<div class="code">
<pre>
SWIGEXPORT jlong JNICALL Java_exampleJNI_FooBar(JNIEnv *jenv, jclass jcls,
jlong jarg1, jobject jarg1_) {
jlong jresult = 0 ;
struct Foo *arg1 = (struct Foo *) 0 ;
struct Bar *result = 0 ;
(void)jenv;
(void)jcls;
(void)jarg1_;
arg1 = *(struct Foo **)&amp;jarg1;
result = (struct Bar *)FooBar(arg1);
*(struct Bar **)&amp;jresult = result;
return jresult;
}
</pre>
</div>
<p> If you are using gcc as your C compiler, you might get a
&quot;dereferencing type-punned pointer will break strict-aliasing rules&quot;
warning about this. Please see <a href="#Java_compiling_dynamic">
Compiling a dynamic module</a> to avoid runtime problems with these
strict aliasing rules.</p>
<p> The default code generated by SWIG for the Java module comes from
the typemaps in the &quot;<tt>java.swg</tt>&quot; library file which implements
the <a href="#Java_default_primitive_type_mappings">Default primitive
type mappings</a> and <a href="#Java_default_non_primitive_typemaps">
Default typemaps for non-primitive types</a> covered earlier. There are
other type mapping typemaps in the Java library. These are listed
below:</p>
<br> &nbsp;
<table BORDER summary="Java library typemap mappings">
<tr VALIGN="TOP"><td><b>C Type</b></td><td><b>Typemap</b></td><td><b>
File</b></td><td><b>Kind</b></td><td><b>Java Type</b></td><td><b>
Function</b></td></tr>
<tr><td>primitive pointers and references</td><td>INPUT</td><td>
typemaps.i</td><td>input</td><td>Java basic types</td><td>Allows values
to be used for C functions taking pointers for data input.
<tr><td>primitive pointers and references</td><td>OUTPUT</td><td>
typemaps.i</td><td>output</td><td>Java basic type arrays</td><td>Allows
values held within an array to be used for C functions taking pointers
for data output.
<tr><td>primitive pointers and references</td><td>INOUT</td><td>
typemaps.i</td><td>input
<br>output</td><td>Java basic type arrays</td><td>Allows values held
within an array to be used for C functions taking pointers for data
input and output.
<tr><td>string
<br> wstring</td><td>[unnamed]</td><td>std_string.i</td><td>input
<br> output</td><td>String</td><td>Use for std::string mapping to Java
String.</td></tr>
<tr><td>arrays of primitive types</td><td>[unnamed]</td><td>
arrays_java.i</td><td>input
<br> output</td><td>arrays of primitive Java types</td><td>Use for
mapping C arrays to Java arrays.</td></tr>
<tr><td>arrays of classes/structs/unions</td><td>JAVA_ARRAYSOFCLASSES
macro</td><td>arrays_java.i</td><td>input
<br> output</td><td>arrays of proxy classes</td><td>Use for mapping C
arrays to Java arrays.</td></tr>
<tr><td>arrays of enums</td><td>ARRAYSOFENUMS</td><td>arrays_java.i</td><td>
input
<br> output</td><td>int[]</td><td>Use for mapping C arrays to Java
arrays (typeunsafe and simple enum wrapping approaches only).</td></tr>
<tr VALIGN="TOP"><td>char *</td><td>BYTE</td><td>various.i</td><td>input</td><td>
byte[]</td><td VALIGN="TOP">Java byte array is converted to char array</td>
</tr>
<tr><td>char **</td><td>STRING_ARRAY</td><td>various.i</td><td>input
<br> output</td><td>String[]</td><td>Use for mapping NULL terminated
arrays of C strings to Java String arrays</td></tr>
<tr><td>unsigned char *</td><td>NIOBUFFER</td><td>various.i</td><td>
input
<br> output</td><td>java.nio.Buffer</td><td>Use for mapping directly
allocated buffers to c/c++. useful with directors and long lived memory
objects</td></tr>
</td></tr>
</td></tr>
</td></tr>
</table>
<h3><a name="Java_typemap_attributes">27.8.6 Java typemap attributes</a></h3>
<p> There are a few additional typemap attributes that the Java module
supports.</p>
<p> The first of these is the 'throws' attribute. The throws attribute
is optional and specified after the typemap name and contains one or
more comma separated classes for adding to the throws clause for any
methods that use that typemap. It is analogous to the <a href="#Java_exception_handling">
%javaexception</a> feature's throws attribute.</p>
<div class="code">
<pre>
%typemap(typemapname, throws=&quot;ExceptionClass1, ExceptionClass2&quot;) type { ... }
</pre>
</div>
<p> The attribute is necessary for supporting Java checked exceptions
and can be added to just about any typemap. The list of typemaps
include all the C/C++ (JNI) typemaps in the &quot;<a href="#Typemaps">
Typemaps</a>&quot; chapter and the Java specific typemaps listed in <a href="#Java_typemaps_c_to_java_types">
the previous section</a>, barring the &quot;jni&quot;, &quot;jtype&quot; and &quot;jstype&quot;
typemaps as they could never contain code to throw an exception.</p>
<p> The throws clause is generated for the proxy method as well as the
JNI method in the JNI intermediary class. If a method uses more than
one typemap and each of those typemaps have classes specified in the
throws clause, the union of the exception classes is added to the
throws clause ensuring there are no duplicate classes. See the <a href="#Java_nan_exception_typemap">
NaN exception example</a> for further usage.</p>
<p> The &quot;jtype&quot; typemap has the optional 'nopgcpp' attribute which can
be used to suppress the generation of the <a href="#Java_pgcpp">
premature garbage collection prevention parameter</a>.</p>
<p> The &quot;javain&quot; typemap has the optional 'pre', 'post' and 'pgcppname'
attributes. These are used for generating code before and after the JNI
call in the proxy class or module class. The 'pre' attribute contains
code that is generated before the JNI call and the 'post' attribute
contains code generated after the JNI call. The 'pgcppname' attribute
is used to change the <a href="#Java_pgcpp">premature garbage
collection prevention parameter</a> name passed to the JNI function.
This is sometimes needed when the 'pre' typemap creates a temporary
variable which is then passed to the JNI function.</p>
<p> <a name="Java_constructor_helper_function"></a> Note that when the
'pre' or 'post' attributes are specified and the associated type is
used in a constructor, a constructor helper function is generated. This
is necessary as the Java proxy constructor wrapper makes a call to a
support constructor using a<i> this</i> call. In Java the<i> this</i>
call must be the first statement in the constructor body. The
constructor body thus calls the helper function and the helper function
instead makes the JNI call, ensuring the 'pre' code is called before
the JNI call is made. There is a <a href="#Java_date_marshalling">Date
marshalling</a> example showing 'pre', 'post' and 'pgcppname'
attributes in action.</p>
<h3><a name="Java_special_variables">27.8.7 Java special variables</a></h3>
<p> The standard SWIG special variables are available for use within
typemaps as described in the <a href="#Typemaps">Typemaps documentation</a>
, for example <tt>$1</tt>, <tt>$input</tt>, <tt>$result</tt> etc.</p>
<p> The Java module uses a few additional special variables:</p>
<p><b> <tt>$javaclassname</tt></b>
<br> This special variable works like the other <a href="#Typemaps_special_variables">
special variables</a> and <tt>$javaclassname</tt> is similar to <tt>
$1_type</tt>. It expands to the class name for use in Java given a
pointer. SWIG wraps unions, structs and classes using pointers and in
this case it expands to the Java proxy class name. For example, <tt>
$javaclassname</tt> is replaced by the proxy classname <tt>Foo</tt> when
wrapping a <tt>Foo *</tt> and <tt>$&amp;javaclassname</tt> expands to the
proxy classname when wrapping the C/C++ type <tt>Foo</tt> and <tt>
$*javaclassname</tt> expands to the proxy classname when wrapping <tt>
Foo *&amp;</tt>. If the type does not have an associated proxy class, it
expands to the type wrapper class name, for example, <tt>
SWIGTYPE_p_unsigned_short</tt> is generated when wrapping <tt>unsigned
short *</tt>. The class name is fully qualified with the package name
when using the <a href="#SWIGPlus_nspace">nspace feature</a>.</p>
<p><b> <tt>$javaclazzname</tt></b>
<br> This special variable works like <tt>$javaclassname</tt>, but
expands the fully qualified C++ class into the package name, if used by
the <a href="#SWIGPlus_nspace">nspace feature</a>, and the proxy class
name, mangled for use as a function name. For example, <tt>
Namespace1::Namespace2::Klass</tt> is expanded into <tt>
Namespace1_Namespace2_Klass_</tt>. This special variable is usually used
for making calls to a function in the intermediary JNI class, as they
are mangled with this prefix.</p>
<p><b> <tt>$null</tt></b>
<br> Used in input typemaps to return early from JNI functions that have
either void or a non-void return type. Example:</p>
<div class="code">
<pre>
%typemap(check) int * %{
if (error) {
SWIG_JavaThrowException(jenv, SWIG_JavaIndexOutOfBoundsException, &quot;Array element error&quot;);
return $null;
}
%}
</pre>
</div>
<p> If the typemap gets put into a function with void as return, $null
will expand to nothing:</p>
<div class="code">
<pre>
SWIGEXPORT void JNICALL Java_jnifn(...) {
if (error) {
SWIG_JavaThrowException(jenv, SWIG_JavaIndexOutOfBoundsException, &quot;Array element error&quot;);
return ;
}
...
}
</pre>
</div>
<p> otherwise $null expands to<i> NULL</i></p>
<div class="code">
<pre>
SWIGEXPORT jobject JNICALL Java_jnifn(...) {
if (error) {
SWIG_JavaThrowException(jenv, SWIG_JavaIndexOutOfBoundsException, &quot;Array element error&quot;);
return NULL;
}
...
}
</pre>
</div>
<p><b> <tt>$javainput, $jnicall and $owner</tt></b>
<br> The $javainput special variable is used in &quot;javain&quot; typemaps and
$jnicall and $owner are used in &quot;javaout&quot; typemaps. $jnicall is
analogous to $action in %exception. It is replaced by the call to the
native method in the intermediary JNI class. $owner is replaced by
either <tt>true</tt> if %newobject has been used, otherwise <tt>false</tt>
. $javainput is analogous to the $input special variable. It is replaced
by the parameter name.</p>
<p> Here is an example:</p>
<div class="code">
<pre>
%typemap(javain) Class &quot;Class.getCPtr($javainput)&quot;
%typemap(javain) unsigned short &quot;$javainput&quot;
%typemap(javaout) Class * {
return new Class($jnicall, $owner);
}
%inline %{
class Class {...};
Class * bar(Class cls, unsigned short ush) { return new Class(); };
%}
</pre>
</div>
<p> The generated proxy code is then:</p>
<div class="code">
<pre>
public static Class bar(Class cls, int ush) {
return new Class(exampleJNI.bar(Class.getCPtr(cls), cls, ush), false);
}
</pre>
</div>
<p> Here $javainput has been replaced by <tt>cls</tt> and <tt>ush</tt>.
$jnicall has been replaced by the native method call, <tt>
exampleJNI.bar(...)</tt> and $owner has been replaced by <tt>false</tt>.
If %newobject is used by adding the following at the beginning of our
example:</p>
<div class="code">
<pre>
%newobject bar(Class cls, unsigned short ush);
</pre>
</div>
<p> The generated code constructs the return type using <tt>true</tt>
indicating the proxy class <tt>Class</tt> is responsible for destroying
the C++ memory allocated for it in <tt>bar</tt>:</p>
<div class="code">
<pre>
public static Class bar(Class cls, int ush) {
return new Class(exampleJNI.bar(Class.getCPtr(cls), cls, ush), true);
}
</pre>
</div>
<p><b> <tt>$static</tt></b>
<br> This special variable expands to either<i> static</i> or nothing
depending on whether the class is an inner Java class or not. It is
used in the &quot;javaclassmodifiers&quot; typemap so that global classes can be
wrapped as Java proxy classes and nested C++ classes/enums can be
wrapped with the Java equivalent, that is, static inner proxy classes.</p>
<p><b> <tt>$error, $jniinput, $javacall and $packagepath</tt></b>
<br> These special variables are used in the directors typemaps. See <a href="#Java_directors_typemaps">
Director specific typemaps</a> for details.</p>
<p><b> <tt>$module</tt></b>
<br> This special variable expands to the module name, as specified by <tt>
%module</tt> or the <tt>-module</tt> commandline option.</p>
<p><b> <tt>$imclassname</tt></b>
<br> This special variable expands to the intermediary class name.
Usually this is the same as '$moduleJNI', unless the jniclassname
attribute is specified in the <a href="#Java_module_directive">%module
directive</a>.</p>
<p><b> <tt>$imfuncname</tt></b>
<br> This special variable expands to the name of the function in the
intermediary class that will be used in $jnicall. Like, $jnicall, this
special variable is only expanded in the &quot;javaout&quot; typemap.</p>
<p><b> <tt>$javainterfacename</tt></b>
<br> This special variable is only expanded when the <tt>interface</tt>
feature is applied to a class. It works much like <tt>$javaclassname</tt>
, but instead of expanding to the proxy classname, it expands to the
value in the <tt>name</tt> attribute in the <tt>interface</tt> feature.
For example:</p>
<div class="code">
<pre>
%feature(&quot;interface&quot;, name=&quot;MyInterface&quot;) MyClass;
%typemap(jstype) MyClass &quot;$&amp;javainterfacename&quot;
%typemap(jstype) MyClass * &quot;$javainterfacename&quot;
</pre>
</div>
<p> will result in the <tt>jstype</tt> typemap expanding to <tt>
MyInterface</tt> for both <tt>MyClass</tt> and <tt>MyClass *</tt>. The
interface name is fully qualified with the package name when using the <a
href="#SWIGPlus_nspace">nspace feature</a>.</p>
<p><b> <tt>$interfacename</tt></b>
<br> This special variable is only expanded when the <tt>interface</tt>
feature is applied to a class. It expands to just the interface name
and is thus different to <tt>$javainterfacename</tt> in that it is not
fully qualified with the package name when using the <a href="#SWIGPlus_nspace">
nspace feature</a>.</p>
<h3><a name="Java_typemaps_for_c_and_cpp">27.8.8 Typemaps for both C and
C++ compilation</a></h3>
<p> JNI calls must be written differently depending on whether the code
is being compiled as C or C++. For example C compilation requires the
pointer to a function pointer struct member syntax like</p>
<div class="code">
<pre>
const jclass clazz = (*jenv)-&gt;FindClass(jenv, &quot;java/lang/String&quot;);
</pre>
</div>
<p> whereas C++ code compilation of the same function call is a member
function call using a class pointer like</p>
<div class="code">
<pre>
const jclass clazz = jenv-&gt;FindClass(&quot;java/lang/String&quot;);
</pre>
</div>
<p> To enable typemaps to be used for either C or C++ compilation, a set
of JCALLx macros have been defined in Lib/java/javahead.swg, where x is
the number of arguments in the C++ version of the JNI call. The above
JNI calls would be written in a typemap like this</p>
<div class="code">
<pre>
const jclass clazz = JCALL1(FindClass, jenv, &quot;java/lang/String&quot;);
</pre>
</div>
<p> Note that the SWIG preprocessor expands these into the appropriate C
or C++ JNI calling convention. The C calling convention is emitted by
default and the C++ calling convention is emitted when using the -c++
SWIG commandline option. If you do not intend your code to be targeting
both C and C++ then your typemaps can use the appropriate JNI calling
convention and need not use the JCALLx macros.</p>
<h3><a name="Java_code_typemaps">27.8.9 Java code typemaps</a></h3>
<p> Most of SWIG's typemaps are used for the generation of C/C++ code.
The typemaps in this section are used solely for the generation of Java
code. Elements of proxy classes and type wrapper classes come from the
following typemaps (the defaults).</p>
<p><tt>%typemap(javabase)</tt></p>
<div class="indent"> base (extends) for Java class: empty default
<br> Note that this typemap accepts a <tt>replace</tt> attribute as an
optional flag. When set to &quot;1&quot;, it will replace/override any C++ base
classes that might have been parsed. If this flag is not specified and
there are C++ base classes, then a multiple inheritance warning is
issued and the code in the typemap is ignored. The typemap also accepts
a <tt>notderived</tt> attribute as an optional flag. When set to &quot;1&quot;,
it will not apply to classes that are derived from a C++ base. When
used with the SWIGTYPE type, it is useful for giving a common base for
all proxy classes, that is, providing a base class that sits in between
all proxy classes and the Java base class <tt>Object</tt> for example: <tt>
%typemap(javabase, notderived=&quot;1&quot;) SWIGTYPE &quot;CommonBase&quot;</tt>.</div>
<p><tt>%typemap(javabody)</tt></p>
<div class="indent"> the essential support body for proxy classes (proxy
base classes only), typewrapper classes and enum classes. Default
contains extra constructors, memory ownership control member variables
(<tt>swigCMemOwn</tt>, <tt>swigCPtr</tt>), the <tt>getCPtr</tt> method
etc.</div>
<p><tt>%typemap(javabody_derived)</tt></p>
<div class="indent"> the essential support body for proxy classes
(derived classes only). Same as &quot;javabody&quot; typemap, but only used for
proxy derived classes.</div>
<p><tt>%typemap(javaclassmodifiers)</tt></p>
<div class="indent"> class modifiers for the Java class: default is
&quot;public class&quot;</div>
<p><tt>%typemap(javacode)</tt></p>
<div class="indent"> Java code is copied verbatim to the Java class:
empty default As there can only be one &quot;javacode&quot; typemap per class,
also consider using the <a href="#Java_proxycode">%proxycode</a>
directive which can be used multiple times per class and offers nearly
identical functionality.</div>
<p><tt>%typemap(javadestruct, methodname=&quot;delete&quot;,
methodmodifiers=&quot;public synchronized&quot;, parameters=&quot;&quot;)</tt>
<br></p>
<div class="indent"> destructor wrapper - the <tt>delete()</tt> method
(proxy classes only), used for all proxy classes except those which
have a base class : default calls C++ destructor (or frees C memory)
and resets <tt>swigCPtr</tt> and <tt>swigCMemOwn</tt> flags
<br>
<br> Note that the <tt>delete()</tt> method name is configurable and is
specified by the <tt>methodname</tt> attribute. The method modifiers
are also configurable via the <tt>methodmodifiers</tt> attribute. If a <tt>
%javamethodmodifiers</tt> is attached to the class' destructor, it will
be used in preference to the <tt>methodmodifiers</tt> typemap attribute
for the class. The <tt>delete</tt> method's parameters declaration can
be provided in the optional <tt>parameters</tt> typemap attribute.</div>
<p><tt>%typemap(javadestruct_derived, methodname=&quot;delete&quot;,
methodmodifiers=&quot;public synchronized&quot;, parameters=&quot;&quot;)</tt></p>
<div class="indent"> destructor wrapper - the <tt>delete()</tt> method
(proxy classes only), same as &quot;javadestruct&quot; but only used for derived
proxy classes : default calls C++ destructor (or frees C memory) and
resets <tt>swigCPtr</tt> and <tt>swigCMemOwn</tt> flags
<br>
<br> Note that the <tt>delete()</tt> method name is configurable and is
specified by the <tt>methodname</tt> attribute. The method modifiers
are also configurable via the <tt>methodmodifiers</tt> attribute. If a <tt>
%javamethodmodifiers</tt> is attached to the class' destructor, it will
be used in preference to the <tt>methodmodifiers</tt> typemap attribute
for the class. The <tt>delete</tt> method's parameters declaration can
be provided in the optional <tt>parameters</tt> typemap attribute.</div>
<p><tt>%typemap(javaimports)</tt></p>
<div class="indent"> import statements for Java class: empty default</div>
<p><tt>%typemap(javainterfaces)</tt></p>
<div class="indent"> interfaces (implements) for Java class: empty
default</div>
<p><tt>%typemap(javafinalize)</tt></p>
<div class="indent"> the <tt>finalize()</tt> method (proxy classes
only): default calls the <tt>delete()</tt> method
<p> Note that the default javafinalize typemap must contain the full
implementation of the finalize method. Any customization to this
typemap must still declare a java finalize method with the correct
signature. Note also that the name of the generated &quot;delete&quot; method may
be affected by <tt>javadestruct</tt> and <tt>javadestruct_derived</tt>
typemaps. Below shows an example modifying the finalizer, assuming the <tt>
delete</tt> method has been renamed to <tt>swig_delete</tt>.</p>
<div class="code">
<pre>
%typemap(javafinalize) SWIGTYPE %{
protected void finalize() {
swig_delete(); // renamed to prevent conflict with existing delete method
}
]%
</pre>
</div></div>
<p><tt>%typemap(javainterfacemodifiers)</tt></p>
<div class="indent"> Interface modifiers for the Java interface
generated when using the <tt>interface</tt> feature, see <a href="#Java_interfaces">
Java interfaces</a> section. The default is &quot;public interface&quot;.
<p><b> Compatibility note:</b> This typemap was added in SWIG-4.1.0.</p>
</div>
<p><tt>%typemap(javainterfacecode, declaration=&quot;...&quot;, cptrmethod=&quot;...&quot;)</tt>
</p>
<div class="indent">
<p> The code in this typemap is added to the body of a Java proxy class
but only when a class is marked with the <tt>interface</tt> feature.
The typemap is used in the proxy class marked with the interface
feature as well as all proxy classes derived from the marked C++ class,
as they are all generated as implementing the Java interface. The
default typemap used in the <tt>%interface</tt> family of macros
mentioned in the <a href="#Java_interfaces">Java interfaces</a>
section, where <tt>CTYPE</tt> is the C++ class macro argument, is as
follows:</p>
<div class="code">
<pre>
%typemap(javainterfacecode,
declaration=&quot; long $interfacename_GetInterfaceCPtr();\n&quot;,
cptrmethod=&quot;$interfacename_GetInterfaceCPtr&quot;) CTYPE %{
public long $interfacename_GetInterfaceCPtr() {
return $imclassname.$javaclazzname$interfacename_GetInterfaceCPtr(swigCPtr);
}
%}
</pre>
</div></div>
<p> The special variable <tt>$interfacename</tt> is expanded into the
name specified in the <tt>interface</tt> feature.</p>
<p><b> Compatibility Note:</b> In SWIG-1.3.21 and earlier releases,
typemaps called &quot;javagetcptr&quot; and &quot;javaptrconstructormodifiers&quot; were
available. These are deprecated and the &quot;javabody&quot; typemap can be used
instead. The <tt>javainterfacecode</tt> typemap and interface feature
was introduced in SWIG-3.0.9.</p>
<p> In summary the contents of the typemaps make up a proxy class like
this:</p>
<div class="code">
<pre>
[ javaimports typemap ]
[ javaclassmodifiers typemap ] javaclassname extends [ javabase typemap ]
implements [ javainterfaces typemap ] {
[ javabody or javabody_derived typemap ]
[ javafinalize typemap ]
public synchronized void <i>delete</i>() [ javadestruct OR javadestruct_derived typemap ]
[ javacode typemap ]
[ javainterfacecode typemap]
... proxy functions ...
}
</pre>
</div>
<p> Note the <tt><i>delete</i>()</tt> methodname and method modifiers
are configurable, see &quot;javadestruct&quot; and &quot;javadestruct_derived&quot;
typemaps above.</p>
<p> The <tt>javainterfacecode</tt> typemap is only used when bases are
marked by the <tt>interface</tt> feature and the <tt>implements</tt>
list will also then be expanded to include these Java interfaces.</p>
<p> The type wrapper class is similar in construction:</p>
<div class="code">
<pre>
[ javaimports typemap ]
[ javaclassmodifiers typemap ] javaclassname extends [ javabase typemap ]
implements [ javainterfaces typemap ] {
[ javabody typemap ]
[ javacode typemap ]
}
</pre>
</div>
<p>The enum class is also similar in construction:</p>
<div class="code">
<pre>
[ javaimports typemap ]
[ javaclassmodifiers typemap ] javaclassname extends [ javabase typemap ]
implements [ javainterfaces typemap ] {
... Enum values ...
[ javabody typemap ]
[ javacode typemap ]
}
</pre>
</div>
<p> The &quot;javaimports&quot; typemap is ignored if the enum class is wrapped by
an inner Java class, that is when wrapping an enum declared within a
C++ class.</p>
<p>The Java interface turned on by the <tt>interface</tt> feature is
fairly simple:</p>
<div class="code">
<pre>
[ javaimports typemap ]
[ javainterfacemodifiers typemap ] [ name ] [extends additional [, ...] ] {
[ javainterfacecode:cptrmethod typemap attribute ]
... interface declarations ...
}
</pre>
</div>
<p> where <tt>name</tt> is the <tt>name</tt> attribute and <tt>
additional</tt> is the <tt>additional</tt> attribute in the <a href="#Java_interfaces">
interface feature</a>.</p>
<p> The defaults can be overridden to tailor the generated classes. Here
is an example which will change the <tt>getCPtr</tt> method and
constructor from the default public access to protected access. If the
classes in one package are not using the classes in another package,
then these methods need not be public and removing access to these low
level implementation details, is a good thing. If you are invoking SWIG
more than once and generating the wrapped classes into different
packages in each invocation, then you cannot do this as you will then
have different packages.</p>
<div class="code">
<pre>
%typemap(javabody) SWIGTYPE %{
private transient long swigCPtr;
protected transient boolean swigCMemOwn;
protected $javaclassname(long cPtr, boolean cMemoryOwn) {
swigCMemOwn = cMemoryOwn;
swigCPtr = cPtr;
}
protected static long getCPtr($javaclassname obj) {
return (obj == null) ? 0 : obj.swigCPtr;
}
%}
</pre>
</div>
<p> The typemap code is the same that is in &quot;<tt>java.swg</tt>&quot;, barring
the last two method modifiers. Note that <tt>SWIGTYPE</tt> will target
all proxy classes, but not the type wrapper classes. Also the above
typemap is only used for proxy classes that are potential base classes.
To target proxy classes that are derived from a wrapped class as well,
the &quot;javabody_derived&quot; typemap should also be overridden.</p>
<p> For the typemap to be used in all type wrapper classes, all the
different types that type wrapper classes could be used for should be
targeted:</p>
<div class="code">
<pre>
%typemap(javabody) SWIGTYPE *, SWIGTYPE &amp;, SWIGTYPE [], SWIGTYPE (CLASS::*) %{
private transient long swigCPtr;
protected $javaclassname(long cPtr, boolean bFutureUse) {
swigCPtr = cPtr;
}
protected $javaclassname() {
swigCPtr = 0;
}
protected static long getCPtr($javaclassname obj) {
return (obj == null) ? 0 : obj.swigCPtr;
}
%}
</pre>
</div>
<p> Again this is the same that is in &quot;<tt>java.swg</tt>&quot;, barring the
method modifier for <tt>getCPtr</tt>.</p>
<p> When using <a href="#Modules">multiple modules</a> or the <a href="#Java_namespaces">
nspace feature</a> it is common to invoke SWIG with a different <tt>
-package</tt> command line option for each module. However, by default
the generated code may not compile if generated classes in one package
use generated classes in another package. The visibility of the <tt>
getCPtr()</tt> and pointer constructor generated from the <tt>javabody</tt>
typemaps needs changing. The default visibility is <tt>protected</tt>
but it needs to be <tt>public</tt> for access from a different package.
Just changing 'protected' to 'public' in the typemap achieves this. Two
macros are available in <tt>java.swg</tt> to make this easier and using
them is the preferred approach over simply copying the typemaps and
modifying as this is forward compatible with any changes in the <tt>
javabody</tt> typemap in future versions of SWIG. The macros are for the
proxy and typewrapper classes and can respectively be used to to make
the method and constructor public:</p>
<div class="code">
<pre>
SWIG_JAVABODY_PROXY(public, public, SWIGTYPE)
SWIG_JAVABODY_TYPEWRAPPER(public, public, public, SWIGTYPE)
</pre>
</div>
<h3><a name="Java_directors_typemaps">27.8.10 Director specific typemaps</a>
</h3>
<p> The Java directors feature requires the &quot;javadirectorin&quot;,
&quot;javadirectorout&quot;, &quot;directorin&quot; and the &quot;directorout&quot; typemaps in order
to work properly. The &quot;javapackage&quot; typemap is an optional typemap used
to identify the Java package path for individual SWIG generated proxy
classes used in director methods.</p>
<p><tt>%typemap(directorin)</tt></p>
<div class="indent">
<p> The &quot;directorin&quot; typemap is used for converting arguments in the C++
director class to the appropriate JNI type before the upcall to Java.
This typemap also specifies the JNI field descriptor for the type in
the &quot;descriptor&quot; attribute. For example, integers are converted as
follows:</p>
<div class="code">
<pre>
%typemap(directorin, descriptor=&quot;I&quot;) int &quot;$input = (jint) $1;&quot;
</pre>
</div>
<p> <code>$input</code> is the SWIG name of the JNI temporary variable
passed to Java in the upcall. The <code>descriptor=&quot;I&quot;</code> will put
an <code>I</code> into the JNI field descriptor that identifies the
Java method that will be called from C++. For more about JNI field
descriptors and their importance, refer to the <a href="#Java_typemaps">
JNI documentation mentioned earlier</a>. A typemap for C character
strings is:</p>
<div class="code">
<pre>
%typemap(directorin, descriptor=&quot;Ljava/lang/String;&quot;, noblock=1) char * {
$input = 0;
if ($1) {
$input = JCALL1(NewStringUTF, jenv, (const char *)$1);
if (!$input) return $null;
}
Swig::LocalRefGuard $1_refguard(jenv, $input);
}
</pre>
</div>
<p> The <tt>Swig::LocalRefGuard</tt> class should be used in directorin
typemaps for newly allocated objects. It is used to control local
reference counts ensuring the count is decremented after the call up
into Java has completed. Its destructor simply calls <tt>
jenv-&gt;DeleteLocalRef(obj)</tt> on the <tt>obj</tt> passed in during
construction.</p>
<p> User-defined types have the default &quot;descriptor&quot; attribute &quot;<code>
L$packagepath/$javaclassname;</code>&quot; where <code>$packagepath</code> is
the package name passed from the SWIG command line and <code>
$javaclassname</code> is the Java proxy class' name. If the <tt>-package</tt>
commandline option is not used to specify the package, then
'$packagepath/' will be removed from the resulting output JNI field
descriptor.<b> Do not forget the terminating ';' for JNI field
descriptors starting with 'L'.</b> If the ';' is left out, Java will
generate a &quot;method not found&quot; runtime error. Note that the <code>
$packagepath</code> substitution always uses the path separator '/' when
expanded. The <code>$javaclassname</code> expansion can be confusing as
it is normally expanded using the '.' separator. However, <code>
$javaclassname</code> is expanded using the path separator '/' in
typemap's &quot;descriptor&quot; attribute as well as in the &quot;directorthrows&quot;
typemap.</p>
</div>
<p><tt>%typemap(directorout)</tt></p>
<div class="indent">
<p> The &quot;directorout&quot; typemap is used for converting the JNI return type
in the C++ director class to the appropriate C++ type after the upcall
to Java. For example, integers are converted as follows:</p>
<div class="code">
<pre>
%typemap(directorout) int %{ $result = (int)$input; %}
</pre>
</div>
<p> <code>$input</code> is the SWIG name of the JNI temporary variable
returned from Java after the upcall. <code>$result</code> is the
resulting output. A typemap for C character strings is:</p>
<div class="code">
<pre>
%typemap(directorout) char * {
$1 = 0;
if ($input) {
$result = (char *)jenv-&gt;GetStringUTFChars($input, 0);
if (!$1) return $null;
}
}
</pre>
</div></div>
<p><tt>%typemap(javadirectorin)</tt></p>
<div class="indent">
<p> Conversion from jtype to jstype for director methods. These are Java
code typemaps which transform the type used in the Java intermediary
JNI class (as specified in the &quot;jtype&quot; typemap) to the Java type used
in the Java module class, proxy classes and type wrapper classes (as
specified in the &quot;jstype&quot; typemap). This typemap provides the
conversion for the parameters in the director methods when calling up
from C++ to Java.</p>
<p> For primitive types, this typemap is usually specified as:</p>
<div class="code">
<pre>
%typemap(javadirectorin) int &quot;$jniinput&quot;
</pre>
</div>
<p> The <code>$jniinput</code> special variable is analogous to <code>
$javainput</code> special variable. It is replaced by the input
parameter name.</p>
</div>
<p><tt>%typemap(javadirectorout)</tt></p>
<div class="indent">
<p> Conversion from jstype to jtype for director methods. These are Java
code typemaps which transform the type used in the Java module class,
proxy classes and type wrapper classes (as specified in the &quot;jstype&quot;
typemap) to the type used in the Java intermediary JNI class (as
specified in the &quot;jtype&quot; typemap). This typemap provides the conversion
for the return type in the director methods when returning from the C++
to Java upcall.</p>
<p> For primitive types, this typemap is usually specified as:</p>
<div class="code">
<pre>
%typemap(javadirectorout) int &quot;$javacall&quot;
</pre>
</div>
<p> The <code>$javacall</code> special variable is analogous to the <code>
$jnicall</code> special variable. It is replaced by the call to the
target Java method. The target method is the method in the Java proxy
class which overrides the virtual C++ method in the C++ base class.</p>
</div>
<p><tt>%typemap(directorthrows)</tt></p>
<div class="indent">
<p> Conversion of Java exceptions to C++ exceptions in director method's
exception handling. This typemap is expected to test the <tt>$error</tt>
special variable for a matching Java exception and if successful
convert and throw it into a C++ exception given by the typemap's type.
The <code>$error</code> special variable is of type <code>jthrowable</code>
and is substituted with a unique variable name in the generated code.</p>
<p> The example below converts a Java <code>
java.lang.IndexOutOfBoundsException</code> exception to the typemap's
type, that is <code>std::out_of_range</code>:</p>
<div class="code">
<pre>
%typemap(directorthrows) std::out_of_range %{
if (Swig::ExceptionMatches(jenv, $error, &quot;java/lang/IndexOutOfBoundsException&quot;)) {
throw std::out_of_range(Swig::JavaExceptionMessage(jenv, $error).message());
}
%}
</pre>
</div>
<p> The utility function <code>Swig::ExceptionMatches</code> and class <code>
Swig::JavaExceptionMessage</code> are helpers available when using
directors and are described in the <a href="#Java_exceptions_from_directors">
Java Exceptions from Directors</a> section.</p>
</div>
<p><tt>%typemap(javapackage)</tt></p>
<div class="indent">
<p> The &quot;javapackage&quot; typemap is optional; it serves to identify a
class's Java package. This typemap should be used in conjunction with
classes that are defined outside of the current SWIG interface file.
The typemap is only used if the type is used in a director method, that
is, in a virtual method in a director class. For example:</p>
<div class="code">
<pre>
// class Foo is handled in a different interface file:
%import &quot;Foo.i&quot;
%feature(&quot;director&quot;) Example;
%inline {
class Bar { };
class Example {
public:
virtual ~Example();
virtual void ping(Foo *arg1, Bar *arg2);
};
}
</pre>
</div>
<p> Assume that the Foo class is part of the Java package<i>
com.wombat.foo</i> but the above interface file is part of the Java
package<i> com.wombat.example</i>. Without the &quot;javapackage&quot; typemap,
SWIG will assume that the Foo class belongs to<i> com.wombat.example</i>
class. The corrected interface file looks like:</p>
<div class="code">
<pre>
// class Foo is handled in a different interface file:
%import &quot;Foo.i&quot;
%typemap(&quot;javapackage&quot;) Foo, Foo *, Foo &amp; &quot;com.wombat.foo&quot;
%feature(&quot;director&quot;) Example;
%inline {
class Bar { };
class Example {
public:
virtual ~Example();
virtual void ping(Foo *arg1, Bar *arg2);
};
}
</pre>
</div>
<p> SWIG looks up the package based on the<b> actual</b> type (plain
Foo, Foo pointer and Foo reference), so it is important to associate
all three types with the desired package. Practically speaking, you
should create a separate SWIG interface file, which is %import-ed into
each SWIG interface file, when you have multiple Java packages. Note
the helper macros below, <code>OTHER_PACKAGE_SPEC</code> and <code>
ANOTHER_PACKAGE_SPEC</code>, which reduce the amount of extra typing. &quot;<code>
TYPE...</code>&quot; is useful when passing templated types to the macro,
since multiargument template types appear to the SWIG preprocessor as
multiple macro arguments.</p>
<div class="code">
<pre>
%typemap(&quot;javapackage&quot;) SWIGTYPE, SWIGTYPE *, SWIGTYPE &amp;
&quot;package.for.most.classes&quot;;
%define OTHER_PACKAGE_SPEC(TYPE...)
%typemap(&quot;javapackage&quot;) TYPE, TYPE *, TYPE &amp; &quot;package.for.other.classes&quot;
%enddef
%define ANOTHER_PACKAGE_SPEC(TYPE...)
%typemap(&quot;javapackage&quot;) TYPE, TYPE *, TYPE &amp; &quot;package.for.another.set&quot;
%enddef
OTHER_PACKAGE_SPEC(Package_2_class_one)
ANOTHER_PACKAGE_SPEC(Package_3_class_two)
/* etc */
</pre>
</div>
<p> The basic strategy here is to provide a default package typemap for
the majority of the classes, only providing &quot;javapackage&quot; typemaps for
the exceptions.</p>
</div>
<h2><a name="Java_typemap_examples">27.9 Typemap Examples</a></h2>
<p> This section includes a few examples of typemaps. For more examples,
you might look at the files &quot;<tt>java.swg</tt>&quot; and &quot;<tt>typemaps.i</tt>
&quot; in the SWIG library.</p>
<h3><a name="Java_simpler_enum_classes">27.9.1 Simpler Java enums for
enums without initializers</a></h3>
<p> The default <a href="#Java_proper_enums_classes">Proper Java enums</a>
approach to wrapping enums is somewhat verbose. This is to handle all
possible C/C++ enums, in particular enums with initializers. The
generated code can be simplified if the enum being wrapped does not
have any initializers.</p>
<p> The following shows how to remove the support methods that are
generated by default and instead use the methods in the Java enum base
class <tt>java.lang.Enum</tt> and <tt>java.lang.Class</tt> for
marshalling enums between C/C++ and Java. The type used for the
typemaps below is <tt>enum SWIGTYPE</tt> which is the default type used
for all enums. The &quot;enums.swg&quot; file should be examined in order to see
the original overridden versions of the typemaps.</p>
<div class="code">
<pre>
%include &quot;enums.swg&quot;
%typemap(javain) enum SWIGTYPE &quot;$javainput.ordinal()&quot;
%typemap(javaout) enum SWIGTYPE {
return $javaclassname.class.getEnumConstants()[$jnicall];
}
%typemap(javabody) enum SWIGTYPE &quot;&quot;
%inline %{
enum HairType { blonde, ginger, brunette };
void setHair(HairType h);
HairType getHair();
%}
</pre>
</div>
<p> SWIG will generate the following Java enum, which is somewhat
simpler than the default:</p>
<div class="code">
<pre>
public enum HairType {
blonde,
ginger,
brunette;
}
</pre>
</div>
<p> and the two Java proxy methods will be:</p>
<div class="code">
<pre>
public static void setHair(HairType h) {
exampleJNI.setHair(h.ordinal());
}
public static HairType getHair() {
return HairType.class.getEnumConstants()[exampleJNI.getHair()];
}
</pre>
</div>
<p> For marshalling Java enums to C/C++ enums, the <tt>ordinal</tt>
method is used to convert the Java enum into an integer value for
passing to the JNI layer, see the &quot;javain&quot; typemap. For marshalling
C/C++ enums to Java enums, the C/C++ enum value is cast to an integer
in the C/C++ typemaps (not shown). This integer value is then used to
index into the array of enum constants that the Java language provides.
See the <tt>getEnumConstants</tt> method in the &quot;javaout&quot; typemap.</p>
<p> These typemaps can often be used as the default for wrapping enums
as in many cases there won't be any enum initializers. In fact a good
strategy is to always use these typemaps and to specifically handle
enums with initializers using %apply. This would be done by using the
original versions of these typemaps in &quot;enums.swg&quot; under another
typemap name for applying using %apply.</p>
<h3><a name="Java_exception_typemap">27.9.2 Handling C++ exception
specifications as Java exceptions</a></h3>
<p> This example demonstrates various ways in which C++ exceptions can
be tailored and converted into Java exceptions. Let's consider a simple
file class <tt>SimpleFile</tt> and an exception class <tt>FileException</tt>
which it may throw on error:</p>
<div class="code">
<pre>
%include &quot;std_string.i&quot; // for std::string typemaps
#include &lt;string&gt;
class FileException {
std::string message;
public:
FileException(const std::string&amp; msg) : message(msg) {}
std::string what() {
return message;
}
};
class SimpleFile {
std::string filename;
public:
SimpleFile(const std::string&amp; filename) : filename(filename) {}
void open() throw(FileException) {
...
}
};
</pre>
</div>
<p> As the <tt>open</tt> method has a C++ exception specification, SWIG
will parse this and know that the method can throw an exception. The <a href="Typemaps.html#throws_typemap">
&quot;throws&quot; typemap</a> is then used when SWIG encounters an exception
specification. The default generic &quot;throws&quot; typemap looks like this:</p>
<div class="code">
<pre>
%typemap(throws) SWIGTYPE, SWIGTYPE &amp;, SWIGTYPE *, SWIGTYPE [ANY] %{
SWIG_JavaThrowException(jenv, SWIG_JavaRuntimeException,
&quot;C++ $1_type exception thrown&quot;);
return $null;
%}
</pre>
</div>
<p> Basically SWIG will generate a C++ try catch block and the body of
the &quot;throws&quot; typemap constitutes the catch block. The above typemap
calls a SWIG supplied method which throws a <tt>
java.lang.RuntimeException</tt>. This exception class is a runtime
exception and therefore not a checked exception. If, however, we wanted
to throw a checked exception, say <tt>java.io.IOException</tt>, then we
could use the following typemap:</p>
<div class="code">
<pre>
%typemap(throws, throws=&quot;java.io.IOException&quot;) FileException {
jclass excep = jenv-&gt;FindClass(&quot;java/io/IOException&quot;);
if (excep)
jenv-&gt;ThrowNew(excep, $1.what().c_str());
return $null;
}
</pre>
</div>
<p> Note that this typemap uses the 'throws' <a href="#Java_typemap_attributes">
typemap attribute</a> to ensure a throws clause is generated. The
generated proxy method then specifies the checked exception by
containing <tt>java.io.IOException</tt> in the throws clause:</p>
<div class="code">
<pre>
public class SimpleFile {
...
public void open() throws java.io.IOException { ... }
}
</pre>
</div>
<p> Lastly, if you don't want to map your C++ exception into one of the
standard Java exceptions, the C++ class can be wrapped and turned into
a custom Java exception class. If we go back to our example, the first
thing we must do is get SWIG to wrap <tt>FileException</tt> and ensure
that it derives from <tt>java.lang.Exception</tt>. Additionally, we
might want to override the <tt>java.lang.Exception.getMessage()</tt>
method. The typemaps to use then are as follows:</p>
<div class="code">
<pre>
%typemap(javabase) FileException &quot;java.lang.Exception&quot;
%typemap(javacode) FileException %{
public String getMessage() {
return what();
}
%}
</pre>
</div>
<p> This generates:</p>
<div class="code">
<pre>
public class FileException extends java.lang.Exception {
...
public String getMessage() {
return what();
}
public FileException(String msg) { ... }
public String what() {
return exampleJNI.FileException_what(swigCPtr, this);
}
}
</pre>
</div>
<p> We could alternatively have used <tt>%rename</tt> to rename <tt>
what()</tt> into <tt>getMessage()</tt>.</p>
<h3><a name="Java_nan_exception_typemap">27.9.3 NaN Exception -
exception handling for a particular type</a></h3>
<p> A Java exception can be thrown from any Java or JNI code. Therefore,
as most typemaps contain either Java or JNI code, just about any
typemap could throw an exception. The following example demonstrates
exception handling on a type by type basis by checking for 'Not a
number' (NaN) whenever a parameter of type <tt>float</tt> is wrapped.</p>
<p> Consider the following C++ code:</p>
<div class="code">
<pre>
bool calculate(float first, float second);
</pre>
</div>
<p> To validate every <tt>float</tt> being passed to C++, we could
precede the code being wrapped by the following typemap which throws a
runtime exception whenever the <tt>float</tt> is 'Not a Number':</p>
<div class="code">
<pre>
%module example
%typemap(javain) float &quot;$module.CheckForNaN($javainput)&quot;
%pragma(java) modulecode=%{
/** Simply returns the input value unless it is not a number,
whereupon an exception is thrown. */
static protected float CheckForNaN(float num) {
if (Float.isNaN(num))
throw new RuntimeException(&quot;Not a number&quot;);
return num;
}
%}
</pre>
</div>
<p> Note that the <tt>CheckForNaN</tt> support method has been added to
the module class using the <tt>modulecode</tt> pragma. The following
shows the generated code of interest:</p>
<div class="code">
<pre>
public class example {
...
/** Simply returns the input value unless it is not a number,
whereupon an exception is thrown. */
static protected float CheckForNaN(float num) {
if (Float.isNaN(num))
throw new RuntimeException(&quot;Not a number&quot;);
return num;
}
public static boolean calculate(float first, float second) {
return exampleJNI.calculate(example.CheckForNaN(first), example.CheckForNaN(second));
}
}
</pre>
</div>
<p> Note that the &quot;javain&quot; typemap is used for every occurrence of a <tt>
float</tt> being used as an input. Of course, we could have targeted the
typemap at a particular parameter by using <tt>float first</tt>, say,
instead of just <tt>float</tt>.</p>
<p> The exception checking could alternatively have been placed into the
'pre' attribute that the &quot;javain&quot; typemap supports. The &quot;javain&quot;
typemap above could be replaced with the following:</p>
<div class="code">
<pre>
%typemap(javain, pre=&quot; $module.CheckForNaN($javainput);&quot;) float &quot;$javainput&quot;
</pre>
</div>
<p> which would modify the <tt>calculate</tt> function to instead be
generated as:</p>
<div class="code">
<pre>
public class example {
...
public static boolean calculate(float first, float second) {
example.CheckForNaN(first);
example.CheckForNaN(second);
{
return exampleJNI.calculate(first, second);
}
}
}
</pre>
</div>
<p> See the <a href="#Java_date_marshalling">Date marshalling example</a>
for an example using further &quot;javain&quot; typemap attributes.</p>
<p> If we decide that what we actually want is a checked exception
instead of a runtime exception, we can change this easily enough. The
proxy method that uses <tt>float</tt> as an input, must then add the
exception class to the throws clause. SWIG can handle this as it
supports the 'throws' <a href="#Java_typemap_attributes">typemap
attribute</a> for specifying classes for the throws clause. Thus we can
modify the pragma and the typemap for the throws clause:</p>
<div class="code">
<pre>
%typemap(javain, throws=&quot;java.lang.Exception&quot;) float &quot;$module.CheckForNaN($javainput)&quot;
%pragma(java) modulecode=%{
/** Simply returns the input value unless it is not a number,
whereupon an exception is thrown. */
static protected float CheckForNaN(float num) throws java.lang.Exception {
if (Float.isNaN(num))
throw new RuntimeException(&quot;Not a number&quot;);
return num;
}
%}
</pre>
</div>
<p> The <tt>calculate</tt> method now has a throws clause and even
though the typemap is used twice for both <tt>float first</tt> and <tt>
float second</tt>, the throws clause contains a single instance of <tt>
java.lang.Exception</tt>:</p>
<div class="code">
<pre>
public class example {
...
/** Simply returns the input value unless it is not a number,
whereupon an exception is thrown. */
static protected float CheckForNaN(float num) throws java.lang.Exception {
if (Float.isNaN(num))
throw new RuntimeException(&quot;Not a number&quot;);
return num;
}
public static boolean calculate(float first, float second) throws java.lang.Exception {
return exampleJNI.calculate(example.CheckForNaN(first), example.CheckForNaN(second));
}
}
</pre>
</div>
<p> If we were a martyr to the JNI cause, we could replace the succinct
code within the &quot;javain&quot; typemap with a few pages of JNI code. If we
had, we would have put it in the &quot;in&quot; typemap which, like all JNI and
Java typemaps, also supports the 'throws' attribute.</p>
<h3><a name="Java_converting_java_string_arrays">27.9.4 Converting Java
String arrays to char **</a></h3>
<p> A common problem in many C programs is the processing of command
line arguments, which are usually passed in an array of NULL terminated
strings. The following SWIG interface file allows a Java String array
to be used as a <tt>char **</tt> object.</p>
<div class="code">
<pre>
%module example
/* This tells SWIG to treat char ** as a special case when used as a parameter
in a function call */
%typemap(in) char ** (jint size) {
int i = 0;
size = (*jenv)-&gt;GetArrayLength(jenv, $input);
$1 = (char **) malloc((size+1)*sizeof(char *));
/* make a copy of each string */
for (i = 0; i&lt;size; i++) {
jstring j_string = (jstring)(*jenv)-&gt;GetObjectArrayElement(jenv, $input, i);
const char * c_string = (*jenv)-&gt;GetStringUTFChars(jenv, j_string, 0);
$1[i] = malloc((strlen(c_string)+1)*sizeof(char));
strcpy($1[i], c_string);
(*jenv)-&gt;ReleaseStringUTFChars(jenv, j_string, c_string);
(*jenv)-&gt;DeleteLocalRef(jenv, j_string);
}
$1[i] = 0;
}
/* This cleans up the memory we malloc'd before the function call */
%typemap(freearg) char ** {
int i;
for (i=0; i&lt;size$argnum-1; i++)
free($1[i]);
free($1);
}
/* This allows a C function to return a char ** as a Java String array */
%typemap(out) char ** {
int i;
int len=0;
jstring temp_string;
const jclass clazz = (*jenv)-&gt;FindClass(jenv, &quot;java/lang/String&quot;);
while ($1[len]) len++;
jresult = (*jenv)-&gt;NewObjectArray(jenv, len, clazz, NULL);
/* exception checking omitted */
for (i=0; i&lt;len; i++) {
temp_string = (*jenv)-&gt;NewStringUTF(jenv, *result++);
(*jenv)-&gt;SetObjectArrayElement(jenv, jresult, i, temp_string);
(*jenv)-&gt;DeleteLocalRef(jenv, temp_string);
}
}
/* These 3 typemaps tell SWIG what JNI and Java types to use */
%typemap(jni) char ** &quot;jobjectArray&quot;
%typemap(jtype) char ** &quot;String[]&quot;
%typemap(jstype) char ** &quot;String[]&quot;
/* These 2 typemaps handle the conversion of the jtype to jstype typemap type
and vice versa */
%typemap(javain) char ** &quot;$javainput&quot;
%typemap(javaout) char ** {
return $jnicall;
}
/* Now a few test functions */
%inline %{
int print_args(char **argv) {
int i = 0;
while (argv[i]) {
printf(&quot;argv[%d] = %s\n&quot;, i, argv[i]);
i++;
}
return i;
}
char **get_args() {
static char *values[] = { &quot;Dave&quot;, &quot;Mike&quot;, &quot;Susan&quot;, &quot;John&quot;, &quot;Michelle&quot;, 0};
return &amp;values[0];
}
%}
</pre>
</div>
<p> Note that the 'C' JNI calling convention is used. Checking for any
thrown exceptions after JNI function calls has been omitted. When this
module is compiled, our wrapped C functions can be used by the
following Java program:</p>
<div class="code">
<pre>
// File runme.java
public class runme {
static {
try {
System.loadLibrary(&quot;example&quot;);
} catch (UnsatisfiedLinkError e) {
System.err.println(&quot;Native code library failed to load. &quot; + e);
System.exit(1);
}
}
public static void main(String argv[]) {
String animals[] = {&quot;Cat&quot;, &quot;Dog&quot;, &quot;Cow&quot;, &quot;Goat&quot;};
example.print_args(animals);
String args[] = example.get_args();
for (int i=0; i&lt;args.length; i++)
System.out.println(i + &quot;:&quot; + args[i]);
}
}
</pre>
</div>
<p> When compiled and run we get:</p>
<div class="code">
<pre>
$ java runme
argv[0] = Cat
argv[1] = Dog
argv[2] = Cow
argv[3] = Goat
0:Dave
1:Mike
2:Susan
3:John
4:Michelle
</pre>
</div>
<p> In the example, a few different typemaps are used. The &quot;in&quot; typemap
is used to receive an input argument and convert it to a C array. Since
dynamic memory allocation is used to allocate memory for the array, the
&quot;freearg&quot; typemap is used to later release this memory after the
execution of the C function. The &quot;out&quot; typemap is used for function
return values. Lastly the &quot;jni&quot;, &quot;jtype&quot; and &quot;jstype&quot; typemaps are also
required to specify what Java types to use.</p>
<h3><a name="Java_expanding_java_object">27.9.5 Expanding a Java object
to multiple arguments</a></h3>
<p> Suppose that you had a collection of C functions with arguments such
as the following:</p>
<div class="code">
<pre>
int foo(int argc, char **argv);
</pre>
</div>
<p> In the previous example, a typemap was written to pass a Java String
array as the <tt>char **argv</tt>. This allows the function to be used
from Java as follows:</p>
<div class="code">
<pre>
example.foo(4, new String[]{&quot;red&quot;, &quot;green&quot;, &quot;blue&quot;, &quot;white&quot;});
</pre>
</div>
<p> Although this works, it's a little awkward to specify the argument
count. To fix this, a multi-argument typemap can be defined. This is
not very difficult--you only have to make slight modifications to the
previous example's typemaps:</p>
<div class="code">
<pre>
%typemap(in) (int argc, char **argv) {
int i = 0;
$1 = (*jenv)-&gt;GetArrayLength(jenv, $input);
$2 = (char **) malloc(($1+1)*sizeof(char *));
/* make a copy of each string */
for (i = 0; i&lt;$1; i++) {
jstring j_string = (jstring)(*jenv)-&gt;GetObjectArrayElement(jenv, $input, i);
const char * c_string = (*jenv)-&gt;GetStringUTFChars(jenv, j_string, 0);
$2[i] = malloc((strlen(c_string)+1)*sizeof(char));
strcpy($2[i], c_string);
(*jenv)-&gt;ReleaseStringUTFChars(jenv, j_string, c_string);
(*jenv)-&gt;DeleteLocalRef(jenv, j_string);
}
$2[i] = 0;
}
%typemap(freearg) (int argc, char **argv) {
int i;
for (i=0; i&lt;$1-1; i++)
free($2[i]);
free($2);
}
%typemap(jni) (int argc, char **argv) &quot;jobjectArray&quot;
%typemap(jtype) (int argc, char **argv) &quot;String[]&quot;
%typemap(jstype) (int argc, char **argv) &quot;String[]&quot;
%typemap(javain) (int argc, char **argv) &quot;$javainput&quot;
</pre>
</div>
<p> When writing a multiple-argument typemap, each of the types is
referenced by a variable such as <tt>$1</tt> or <tt>$2</tt>. The
typemap code simply fills in the appropriate values from the supplied
Java parameter.</p>
<p> With the above typemap in place, you will find it no longer
necessary to supply the argument count. This is automatically set by
the typemap code. For example:</p>
<div class="code">
<pre>
example.foo(new String[]{&quot;red&quot;, &quot;green&quot;, &quot;blue&quot;, &quot;white&quot;});
</pre>
</div>
<h3><a name="Java_using_typemaps_return_arguments">27.9.6 Using typemaps
to return arguments</a></h3>
<p> A common problem in some C programs is that values may be returned
in function parameters rather than in the return value of a function.
The <tt>typemaps.i</tt> file defines INPUT, OUTPUT and INOUT typemaps
which can be used to solve some instances of this problem. This library
file uses an array as a means of moving data to and from Java when
wrapping a C function that takes non const pointers or non const
references as parameters.</p>
<p> Now we are going to outline an alternative approach to using arrays
for C pointers. The INOUT typemap uses a <tt>double[]</tt> array for
receiving and returning the <tt>double*</tt> parameters. In this
approach we are able to use a Java class <tt>myDouble</tt> instead of <tt>
double[]</tt> arrays where the C pointer <tt>double*</tt> is required.</p>
<p> Here is our example function:</p>
<div class="code">
<pre>
/* Returns a status value and two values in out1 and out2 */
int spam(double a, double b, double *out1, double *out2);
</pre>
</div>
<p> If we define a structure <tt>MyDouble</tt> containing a <tt>double</tt>
member variable and use some typemaps we can solve this problem. For
example we could put the following through SWIG:</p>
<div class="code">
<pre>
%module example
/* Define a new structure to use instead of double * */
%inline %{
typedef struct {
double value;
} MyDouble;
%}
%{
/* Returns a status value and two values in out1 and out2 */
int spam(double a, double b, double *out1, double *out2) {
int status = 1;
*out1 = a*10.0;
*out2 = b*100.0;
return status;
}
%}
/*
This typemap will make any double * function parameters with name <tt>OUTVALUE</tt> take an
argument of MyDouble instead of double *. This will
allow the calling function to read the double * value after returning from the function.
*/
%typemap(in) double *OUTVALUE {
jclass clazz = jenv-&gt;FindClass(&quot;MyDouble&quot;);
jfieldID fid = jenv-&gt;GetFieldID(clazz, &quot;swigCPtr&quot;, &quot;J&quot;);
jlong cPtr = jenv-&gt;GetLongField($input, fid);
MyDouble *pMyDouble = NULL;
*(MyDouble **)&amp;pMyDouble = *(MyDouble **)&amp;cPtr;
$1 = &amp;pMyDouble-&gt;value;
}
%typemap(jtype) double *OUTVALUE &quot;MyDouble&quot;
%typemap(jstype) double *OUTVALUE &quot;MyDouble&quot;
%typemap(jni) double *OUTVALUE &quot;jobject&quot;
%typemap(javain) double *OUTVALUE &quot;$javainput&quot;
/* Now we apply the typemap to the named variables */
%apply double *OUTVALUE { double *out1, double *out2 };
int spam(double a, double b, double *out1, double *out2);
</pre>
</div>
<p> Note that the C++ JNI calling convention has been used this time and
so must be compiled as C++ and the -c++ commandline must be passed to
SWIG. JNI error checking has been omitted for clarity.</p>
<p> What the typemaps do are make the named <tt>double*</tt> function
parameters use our new <tt>MyDouble</tt> wrapper structure. The &quot;in&quot;
typemap takes this structure, gets the C++ pointer to it, takes the <tt>
double value</tt> member variable and passes it to the C++ <tt>spam</tt>
function. In Java, when the function returns, we use the SWIG created <tt>
getValue()</tt> function to get the output value. The following Java
program demonstrates this:</p>
<div class="code">
<pre>
// File: runme.java
public class runme {
static {
try {
System.loadLibrary(&quot;example&quot;);
} catch (UnsatisfiedLinkError e) {
System.err.println(&quot;Native code library failed to load. &quot; + e);
System.exit(1);
}
}
public static void main(String argv[]) {
MyDouble out1 = new MyDouble();
MyDouble out2 = new MyDouble();
int ret = example.spam(1.2, 3.4, out1, out2);
System.out.println(ret + &quot; &quot; + out1.getValue() + &quot; &quot; + out2.getValue());
}
}
</pre>
</div>
<p> When compiled and run we get:</p>
<div class="code">
<pre>
$ java runme
1 12.0 340.0
</pre>
</div>
<h3><a name="Java_adding_downcasts">27.9.7 Adding Java downcasts to
polymorphic return types</a></h3>
<p> SWIG support for polymorphism works in that the appropriate virtual
function is called. However, the default generated code does not allow
for downcasting. Let's examine this with the following code:</p>
<div class="code">
<pre>
%include &quot;std_string.i&quot;
#include &lt;iostream&gt;
using namespace std;
class Vehicle {
public:
virtual void start() = 0;
...
};
class Ambulance : public Vehicle {
string vol;
public:
Ambulance(string volume) : vol(volume) {}
virtual void start() {
cout &lt;&lt; &quot;Ambulance started&quot; &lt;&lt; endl;
}
void sound_siren() {
cout &lt;&lt; vol &lt;&lt; &quot; siren sounded!&quot; &lt;&lt; endl;
}
...
};
Vehicle *vehicle_factory() {
return new Ambulance(&quot;Very loud&quot;);
}
</pre>
</div>
<p> If we execute the following Java code:</p>
<div class="code">
<pre>
Vehicle vehicle = example.vehicle_factory();
vehicle.start();
Ambulance ambulance = (Ambulance)vehicle;
ambulance.sound_siren();
</pre>
</div>
<p> We get:</p>
<div class="code">
<pre>
Ambulance started
java.lang.ClassCastException
at runme.main(runme.java:16)
</pre>
</div>
<p> Even though we know from examination of the C++ code that <tt>
vehicle_factory</tt> returns an object of type <tt>Ambulance</tt>, we
are not able to use this knowledge to perform the downcast in Java.
This occurs because the runtime type information is not completely
passed from C++ to Java when returning the type from <tt>
vehicle_factory()</tt>. Usually this is not a problem as virtual
functions do work by default, such as in the case of <tt>start()</tt>.
There are a few solutions to getting downcasts to work.</p>
<p> The first is not to use a Java cast but a call to C++ to make the
cast. Add this to your code:</p>
<div class="code">
<pre>
%exception Ambulance::dynamic_cast(Vehicle *vehicle) {
$action
if (!result) {
jclass excep = jenv-&gt;FindClass(&quot;java/lang/ClassCastException&quot;);
if (excep) {
jenv-&gt;ThrowNew(excep, &quot;dynamic_cast exception&quot;);
}
}
}
%extend Ambulance {
static Ambulance *dynamic_cast(Vehicle *vehicle) {
return dynamic_cast&lt;Ambulance *&gt;(vehicle);
}
};
</pre>
</div>
<p> It would then be used from Java like this</p>
<div class="code">
<pre>
Ambulance ambulance = Ambulance.dynamic_cast(vehicle);
ambulance.sound_siren();
</pre>
</div>
<p> Should <tt>vehicle</tt> not be of type <tt>ambulance</tt> then a
Java <tt>ClassCastException</tt> is thrown. The next solution is a
purer solution in that Java downcasts can be performed on the types.
Add the following before the definition of <tt>vehicle_factory</tt>:</p>
<div class="code">
<pre>
%typemap(out) Vehicle * {
Ambulance *downcast = dynamic_cast&lt;Ambulance *&gt;($1);
*(Ambulance **)&amp;$result = downcast;
}
%typemap(javaout) Vehicle * {
return new Ambulance($jnicall, $owner);
}
</pre>
</div>
<p> Here we are using our knowledge that <tt>vehicle_factory</tt> always
returns type <tt>Ambulance</tt> so that the Java proxy is created as a
type <tt>Ambulance</tt>. If <tt>vehicle_factory</tt> can manufacture
any type of <tt>Vehicle</tt> and we want to be able to downcast using
Java casts for any of these types, then a different approach is needed.
Consider expanding our example with a new Vehicle type and a more
flexible factory function:</p>
<div class="code">
<pre>
class FireEngine : public Vehicle {
public:
FireEngine() {}
virtual void start() {
cout &lt;&lt; &quot;FireEngine started&quot; &lt;&lt; endl;
}
void roll_out_hose() {
cout &lt;&lt; &quot;Hose rolled out&quot; &lt;&lt; endl;
}
...
};
Vehicle *vehicle_factory(int vehicle_number) {
if (vehicle_number == 0)
return new Ambulance(&quot;Very loud&quot;);
else
return new FireEngine();
}
</pre>
</div>
<p> To be able to downcast with this sort of Java code:</p>
<div class="code">
<pre>
FireEngine fireengine = (FireEngine)example.vehicle_factory(1);
fireengine.roll_out_hose();
Ambulance ambulance = (Ambulance)example.vehicle_factory(0);
ambulance.sound_siren();
</pre>
</div>
<p> the following typemaps targeted at the <tt>vehicle_factory</tt>
function will achieve this. Note that in this case, the Java class is
constructed using JNI code rather than passing a pointer across the JNI
boundary in a Java long for construction in Java code.</p>
<div class="code">
<pre>
%typemap(jni) Vehicle *vehicle_factory &quot;jobject&quot;
%typemap(jtype) Vehicle *vehicle_factory &quot;Vehicle&quot;
%typemap(jstype) Vehicle *vehicle_factory &quot;Vehicle&quot;
%typemap(javaout) Vehicle *vehicle_factory {
return $jnicall;
}
%typemap(out) Vehicle *vehicle_factory {
Ambulance *ambulance = dynamic_cast&lt;Ambulance *&gt;($1);
FireEngine *fireengine = dynamic_cast&lt;FireEngine *&gt;($1);
if (ambulance) {
// call the Ambulance(long cPtr, boolean cMemoryOwn) constructor
jclass clazz = jenv-&gt;FindClass(&quot;Ambulance&quot;);
if (clazz) {
jmethodID mid = jenv-&gt;GetMethodID(clazz, &quot;&lt;init&gt;&quot;, &quot;(JZ)V&quot;);
if (mid) {
jlong cptr = 0;
*(Ambulance **)&amp;cptr = ambulance;
$result = jenv-&gt;NewObject(clazz, mid, cptr, false);
}
}
} else if (fireengine) {
// call the FireEngine(long cPtr, boolean cMemoryOwn) constructor
jclass clazz = jenv-&gt;FindClass(&quot;FireEngine&quot;);
if (clazz) {
jmethodID mid = jenv-&gt;GetMethodID(clazz, &quot;&lt;init&gt;&quot;, &quot;(JZ)V&quot;);
if (mid) {
jlong cptr = 0;
*(FireEngine **)&amp;cptr = fireengine;
$result = jenv-&gt;NewObject(clazz, mid, cptr, false);
}
}
}
else {
cout &lt;&lt; &quot;Unexpected type &quot; &lt;&lt; endl;
}
if (!$result)
cout &lt;&lt; &quot;Failed to create new java object&quot; &lt;&lt; endl;
}
</pre>
</div>
<p> Better error handling would need to be added into this code. There
are other solutions to this problem, but this last example demonstrates
some more involved JNI code. SWIG usually generates code which
constructs the proxy classes using Java code as it is easier to handle
error conditions and is faster. Note that the JNI code above uses a
number of string lookups to call a constructor, whereas this would not
occur using byte compiled Java code.</p>
<h3><a name="Java_adding_equals_method">27.9.8 Adding an equals method
to the Java classes</a></h3>
<p> When a pointer is returned from a JNI function, it is wrapped using
a new Java proxy class or type wrapper class. Even when the pointers
are the same, it will not be possible to know that the two Java classes
containing those pointers are actually the same object. It is common in
Java to use the <tt>equals()</tt> method to check whether two objects
are equivalent. The <tt>equals()</tt> method is usually accompanied by
a <tt>hashCode()</tt> method in order to fulfill the requirement that
the hash code is equal for equal objects. Pure Java code methods like
these can be easily added:</p>
<div class="code">
<pre>
%typemap(javacode) SWIGTYPE %{
public boolean equals(Object obj) {
boolean equal = false;
if (obj instanceof $javaclassname)
equal = ((($javaclassname)obj).swigCPtr == this.swigCPtr);
return equal;
}
public int hashCode() {
return (int)getPointer();
}
%}
class Foo { };
Foo* returnFoo(Foo *foo) { return foo; }
</pre>
</div>
<p> The following would display <tt>false</tt> without the <tt>javacode</tt>
typemap above. With the typemap defining the <tt>equals</tt> method the
result is <tt>true</tt>.</p>
<div class="code">
<pre>
Foo foo1 = new Foo();
Foo foo2 = example.returnFoo(foo1);
System.out.println(&quot;foo1? &quot; + foo1.equals(foo2));
</pre>
</div>
<h3><a name="Java_void_pointers">27.9.9 Void pointers and a common Java
base class</a></h3>
<p> One might wonder why the common code that SWIG emits for the proxy
and type wrapper classes is not pushed into a base class. The reason is
that although <tt>swigCPtr</tt> could be put into a common base class
for all classes wrapping C structures, it would not work for C++
classes involved in an inheritance chain. Each class derived from a
base needs a separate <tt>swigCPtr</tt> because C++ compilers sometimes
use a different pointer value when casting a derived class to a base.
Additionally as Java only supports single inheritance, it would not be
possible to derive wrapped classes from your own pure Java classes if
the base class has been 'used up' by SWIG. However, you may want to
move some of the common code into a base class. Here is an example
which uses a common base class for all proxy classes and type wrapper
classes:</p>
<div class="code">
<pre>
%typemap(javabase) SWIGTYPE, SWIGTYPE *, SWIGTYPE &amp;, SWIGTYPE [],
SWIGTYPE (CLASS::*) &quot;SWIG&quot;
%typemap(javacode) SWIGTYPE, SWIGTYPE *, SWIGTYPE &amp;, SWIGTYPE [],
SWIGTYPE (CLASS::*) %{
protected long getPointer() {
return swigCPtr;
}
%}
</pre>
</div>
<p> Define new base class called SWIG:</p>
<div class="code">
<pre>
public abstract class SWIG {
protected abstract long getPointer();
public boolean equals(Object obj) {
boolean equal = false;
if (obj instanceof SWIG)
equal = (((SWIG)obj).getPointer() == this.getPointer());
return equal;
}
SWIGTYPE_p_void getVoidPointer() {
return new SWIGTYPE_p_void(getPointer(), false);
}
}
</pre>
</div>
<p> This example contains some useful functionality which you may want
in your code.</p>
<ul>
<li> It has an <tt>equals()</tt> method. Unlike the previous example,
the method code isn't replicated in all classes.</li>
<li> It also has a function which effectively implements a cast from the
type of the proxy/type wrapper class to a void pointer. This is
necessary for passing a proxy class or a type wrapper class to a
function that takes a void pointer.</li>
</ul>
<h3><a name="Java_struct_pointer_pointer">27.9.10 Struct pointer to
pointer</a></h3>
<p> Pointers to pointers are often used as output parameters in C
factory type functions. These are a bit more tricky to handle. Consider
the following situation where a <tt>Butler</tt> can be hired and fired:</p>
<div class="code">
<pre>
typedef struct {
int hoursAvailable;
char *greeting;
} Butler;
// Note: HireButler will allocate the memory
// The caller must free the memory by calling FireButler()!!
extern int HireButler(Butler **ppButler);
extern void FireButler(Butler *pButler);
</pre>
</div>
<p> C code implementation:</p>
<div class="code">
<pre>
int HireButler(Butler **ppButler) {
Butler *pButler = (Butler *)malloc(sizeof(Butler));
pButler-&gt;hoursAvailable = 24;
pButler-&gt;greeting = (char *)malloc(32);
strcpy(pButler-&gt;greeting, &quot;At your service Sir&quot;);
*ppButler = pButler;
return 1;
}
void FireButler(Butler *pButler) {
free(pButler-&gt;greeting);
free(pButler);
}
</pre>
</div>
<p> Let's take two approaches to wrapping this code. The first is to
provide a functional interface, much like the original C interface. The
following Java code shows how we intend the code to be used:</p>
<div class="code">
<pre>
Butler jeeves = new Butler();
example.HireButler(jeeves);
System.out.println(&quot;Greeting: &quot; + jeeves.getGreeting());
System.out.println(&quot;Availability: &quot; + jeeves.getHoursAvailable() + &quot; hours per day&quot;);
example.FireButler(jeeves);
</pre>
</div>
<p> Resulting in the following output when run:</p>
<div class="shell">
<pre>
Greeting: At your service Sir
Availability: 24 hours per day
</pre>
</div>
<p> Note the usage is very much like it would be used if we were writing
C code, that is, explicit memory management is needed. No C memory is
allocated in the construction of the <tt>Butler</tt> proxy class and
the proxy class will not destroy the underlying C memory when it is
collected. A number of typemaps and features are needed to implement
this approach. The following interface file code should be placed
before SWIG parses the above C code.</p>
<div class="code">
<pre>
%module example
// Do not generate the default proxy constructor or destructor
%nodefaultctor Butler;
%nodefaultdtor Butler;
// Add in pure Java code proxy constructor
%typemap(javacode) Butler %{
/** This constructor creates the proxy which initially does not create nor own any C memory */
public Butler() {
this(0, false);
}
%}
// Type typemaps for marshalling Butler **
%typemap(jni) Butler ** &quot;jobject&quot;
%typemap(jtype) Butler ** &quot;Butler&quot;
%typemap(jstype) Butler ** &quot;Butler&quot;
// Typemaps for Butler ** as a parameter output type
%typemap(in) Butler ** (Butler *ppButler = 0) %{
$1 = &amp;ppButler;
%}
%typemap(argout) Butler ** {
// Give Java proxy the C pointer (of newly created object)
jclass clazz = (*jenv)-&gt;FindClass(jenv, &quot;Butler&quot;);
jfieldID fid = (*jenv)-&gt;GetFieldID(jenv, clazz, &quot;swigCPtr&quot;, &quot;J&quot;);
jlong cPtr = 0;
*(Butler **)&amp;cPtr = *$1;
(*jenv)-&gt;SetLongField(jenv, $input, fid, cPtr);
}
%typemap(javain) Butler ** &quot;$javainput&quot;
</pre>
</div>
<p> Note that the JNI code sets the proxy's <tt>swigCPtr</tt> member
variable to point to the newly created object. The <tt>swigCMemOwn</tt>
remains unchanged (at false), so that the proxy does not own the
memory.</p>
<p> Note: The old %nodefault directive disabled the default constructor
and destructor at the same time. This is unsafe in most of the cases,
and you can use the explicit %nodefaultctor and %nodefaultdtor
directives to achieve the same result if needed.</p>
<p> The second approach offers a more object oriented interface to the
Java user. We do this by making the Java proxy class's constructor call
the <tt>HireButler()</tt> method to create the underlying C object.
Additionally we get the proxy to take ownership of the memory so that
the finalizer will call the <tt>FireButler()</tt> function. The proxy
class will thus take ownership of the memory and clean it up when no
longer needed. We will also prevent the user from being able to
explicitly call the <tt>HireButler()</tt> and <tt>FireButler()</tt>
functions. Usage from Java will simply be:</p>
<div class="code">
<pre>
Butler jeeves = new Butler();
System.out.println(&quot;Greeting: &quot; + jeeves.getGreeting());
System.out.println(&quot;Availability: &quot; + jeeves.getHoursAvailable() + &quot; hours per day&quot;);
</pre>
</div>
<p> Note that the Butler class is used just like any other Java class
and no extra coding by the user needs to be written to clear up the
underlying C memory as the finalizer will be called by the garbage
collector which in turn will call the <tt>FireButler()</tt> function.
To implement this, we use the above interface file code but remove the <tt>
javacode</tt> typemap and add the following:</p>
<div class="code">
<pre>
// Don't expose the memory allocation/de-allocation functions
%ignore FireButler(Butler *pButler);
%ignore HireButler(Butler **ppButler);
// Add in a custom proxy constructor and destructor
%extend Butler {
Butler() {
Butler *pButler = 0;
HireButler(&amp;pButler);
return pButler;
}
~Butler() {
FireButler($self);
}
}
</pre>
</div>
<p> Note that the code in <tt>%extend</tt> is using a C++ type
constructor and destructor, yet the generated code will still compile
as C code, see <a href="#SWIG_adding_member_functions">Adding member
functions to C structures</a>. The C functional interface has been
completely morphed into an object-oriented interface and the Butler
class would behave much like any pure Java class and feel more natural
to Java users.</p>
<h3><a name="Java_memory_management_member_variables">27.9.11 Memory
management when returning references to member variables</a></h3>
<p> This example shows how to prevent premature garbage collection of
objects when the underlying C++ class returns a pointer or reference to
a member variable.</p>
<p> Consider the following C++ code:</p>
<div class="code">
<pre>
struct Wheel {
int size;
Wheel(int sz = 0) : size(sz) {}
};
class Bike {
Wheel wheel;
public:
Bike(int val) : wheel(val) {}
Wheel&amp; getWheel() { return wheel; }
};
</pre>
</div>
<p> and the following usage from Java after running the code through
SWIG:</p>
<div class="code">
<pre>
Wheel wheel = new Bike(10).getWheel();
System.out.println(&quot;wheel size: &quot; + wheel.getSize());
// Simulate a garbage collection
System.gc();
System.runFinalization();
System.out.println(&quot;wheel size: &quot; + wheel.getSize());
</pre>
</div>
<p> Don't be surprised that if the resulting output gives strange
results such as...</p>
<div class="shell">
<pre>
wheel size: 10
wheel size: 135019664
</pre>
</div>
<p> What has happened here is the garbage collector has collected the <tt>
Bike</tt> instance as it doesn't think it is needed any more. The proxy
instance, <tt>wheel</tt>, contains a reference to memory that was
deleted when the <tt>Bike</tt> instance was collected. In order to
prevent the garbage collector from collecting the <tt>Bike</tt>
instance a reference to the <tt>Bike</tt> must be added to the <tt>
wheel</tt> instance. You can do this by adding the reference when the <tt>
getWheel()</tt> method is called using the following typemaps.</p>
<div class="code">
<pre>
%typemap(javacode) Wheel %{
// Ensure that the GC doesn't collect any Bike instance set from Java
private Bike bikeReference;
protected void addReference(Bike bike) {
bikeReference = bike;
}
%}
// Add a Java reference to prevent premature garbage collection and resulting use
// of dangling C++ pointer. Intended for methods that return pointers or
// references to a member variable.
%typemap(javaout) Wheel&amp; getWheel {
long cPtr = $jnicall;
$javaclassname ret = null;
if (cPtr != 0) {
ret = new $javaclassname(cPtr, $owner);
ret.addReference(this);
}
return ret;
}
</pre>
</div>
<p> The code in the first typemap gets added to the <tt>Wheel</tt> proxy
class. The code in the second typemap constitutes the bulk of the code
in the generated <tt>getWheel()</tt> function:</p>
<div class="code">
<pre>
public class Wheel {
...
// Ensure that the GC doesn't collect any bike set from Java
private Bike bikeReference;
protected void addReference(Bike bike) {
bikeReference = bike;
}
}
public class Bike {
...
public Wheel getWheel() {
long cPtr = exampleJNI.Bike_getWheel(swigCPtr, this);
Wheel ret = null;
if (cPtr != 0) {
ret = new Wheel(cPtr, false);
ret.addReference(this);
}
return ret;
}
}
</pre>
</div>
<p> Note the <tt>addReference</tt> call.</p>
<h3><a name="Java_memory_management_objects">27.9.12 Memory management
for objects passed to the C++ layer</a></h3>
<p> Managing memory can be tricky when using C++ and Java proxy classes.
The previous example shows one such case and this example looks at
memory management for a class passed to a C++ method which expects the
object to remain in scope after the function has returned. Consider the
following two C++ classes:</p>
<div class="code">
<pre>
struct Element {
int value;
Element(int val) : value(val) {}
};
class Container {
Element* element;
public:
Container() : element(0) {}
void setElement(Element* e) { element = e; }
Element* getElement() { return element; }
};
</pre>
</div>
<p> and usage from C++</p>
<div class="code">
<pre>
Container container;
Element element(20);
container.setElement(&amp;element);
cout &lt;&lt; &quot;element.value: &quot; &lt;&lt; container.getElement()-&gt;value &lt;&lt; endl;
</pre>
</div>
<p> and more or less equivalent usage from Java</p>
<div class="code">
<pre>
Container container = new Container();
container.setElement(new Element(20));
System.out.println(&quot;element value: &quot; + container.getElement().getValue());
</pre>
</div>
<p> The C++ code will always print out 20, but the value printed out may
not be this in the Java equivalent code. In order to understand why,
consider a garbage collection occurring...</p>
<div class="code">
<pre>
Container container = new Container();
container.setElement(new Element(20));
// Simulate a garbage collection
System.gc();
System.runFinalization();
System.out.println(&quot;element value: &quot; + container.getElement().getValue());
</pre>
</div>
<p> The temporary element created with <tt>new Element(20)</tt> could
get garbage collected which ultimately means the <tt>container</tt>
variable is holding a dangling pointer, thereby printing out any old
random value instead of the expected value of 20. One solution is to
add in the appropriate references in the Java layer...</p>
<div class="code">
<pre>
public class Container {
...
// Ensure that the GC doesn't collect any Element set from Java
// as the underlying C++ class stores a shallow copy
private Element elementReference;
public void setElement(Element e) {
exampleJNI.Container_setElement(swigCPtr, this, Element.getCPtr(e), e);
elementReference = e;
}
}
</pre>
</div>
<p> The following typemaps can be used to generate this code:</p>
<div class="code">
<pre>
%typemap(javacode) Container %{
// Ensure that the GC doesn't collect any element set from Java
// as the underlying C++ class stores a shallow copy
private Element elementReference;
%}
%typemap(javain,
post=&quot; elementReference = $javainput;&quot;
) Element *e &quot;Element.getCPtr($javainput)&quot;
</pre>
</div>
<p> The 'javacode' typemap simply adds in the specified code into the
Java proxy class. The 'javain' typemap matches the input parameter type
and name for the <tt>setElement</tt> method and the 'post' typemap
attribute allows adding code after the JNI call. The 'post' code is
generated into a finally block after the JNI call so the resulting code
isn't quite as mentioned earlier, <tt>setElement</tt> is actually:</p>
<div class="code">
<pre>
public void setElement(Element e) {
try {
exampleJNI.Container_setElement(swigCPtr, this, Element.getCPtr(e), e);
} finally {
elementReference = e;
}
}
</pre>
</div>
<h3><a name="Java_date_marshalling">27.9.13 Date marshalling using the
javain typemap and associated attributes</a></h3>
<p> The <a href="#Java_nan_exception_typemap">NaN Exception example</a>
is a simple example of the &quot;javain&quot; typemap and its 'pre' attribute.
This example demonstrates how a C++ date class, say <tt>CDate</tt>, can
be mapped onto the standard Java date class, <tt>
java.util.GregorianCalendar</tt> by using the 'pre', 'post' and
'pgcppname' attributes of the &quot;javain&quot; typemap. The idea is that the <tt>
GregorianCalendar</tt> is used wherever the C++ API uses a <tt>CDate</tt>
. Let's assume the code being wrapped is as follows:</p>
<div class="code">
<pre>
class CDate {
public:
CDate(int year, int month, int day);
int getYear();
int getMonth();
int getDay();
...
};
struct Action {
static int doSomething(const CDate &amp;dateIn, CDate &amp;dateOut);
Action(const CDate &amp;date, CDate &amp;dateOut);
};
</pre>
</div>
<p> Note that <tt>dateIn</tt> is const and therefore read only and <tt>
dateOut</tt> is a non-const output type.</p>
<p> First let's look at the code that is generated by default, where the
Java proxy class <tt>CDate</tt> is used in the proxy interface:</p>
<div class="code">
<pre>
public class Action {
...
public static int doSomething(CDate dateIn, CDate dateOut) {
return exampleJNI.Action_doSomething(CDate.getCPtr(dateIn), dateIn,
CDate.getCPtr(dateOut), dateOut);
}
public Action(CDate date, CDate dateOut) {
this(exampleJNI.new_Action(CDate.getCPtr(date), date,
CDate.getCPtr(dateOut), dateOut), true);
}
}
</pre>
</div>
<p> The <tt>CDate &amp;</tt> and <tt>const CDate &amp;</tt> Java code is
generated from the following two default typemaps:</p>
<div class="code">
<pre>
%typemap(jstype) SWIGTYPE &amp; &quot;$javaclassname&quot;
%typemap(javain) SWIGTYPE &amp; &quot;$javaclassname.getCPtr($javainput)&quot;
</pre>
</div>
<p> where '$javaclassname' is translated into the proxy class name, <tt>
CDate</tt> and '$javainput' is translated into the name of the
parameter, eg <tt>dateIn</tt>. From Java, the intention is then to call
into a modified API with something like:</p>
<div class="code">
<pre>
java.util.GregorianCalendar calendarIn =
new java.util.GregorianCalendar(2011, java.util.Calendar.APRIL, 13, 0, 0, 0);
java.util.GregorianCalendar calendarOut = new java.util.GregorianCalendar();
// Note in calls below, calendarIn remains unchanged and calendarOut
// is set to a new value by the C++ call
Action.doSomething(calendarIn, calendarOut);
Action action = new Action(calendarIn, calendarOut);
</pre>
</div>
<p> To achieve this mapping, we need to alter the default code
generation slightly so that at the Java layer, a <tt>GregorianCalendar</tt>
is converted into a <tt>CDate</tt>. The JNI intermediary layer will
still take a pointer to the underlying <tt>CDate</tt> class. The
typemaps to achieve this are shown below.</p>
<div class="code">
<pre>
%typemap(jstype) const CDate&amp; &quot;java.util.GregorianCalendar&quot;
%typemap(javain,
pre=&quot; CDate temp$javainput = new CDate($javainput.get(java.util.Calendar.YEAR), &quot;
&quot;$javainput.get(java.util.Calendar.MONTH), $javainput.get(java.util.Calendar.DATE));&quot;,
pgcppname=&quot;temp$javainput&quot;) const CDate &amp;
&quot;$javaclassname.getCPtr(temp$javainput)&quot;
%typemap(jstype) CDate&amp; &quot;java.util.Calendar&quot;
%typemap(javain,
pre=&quot; CDate temp$javainput = new CDate($javainput.get(java.util.Calendar.YEAR), &quot;
&quot;$javainput.get(java.util.Calendar.MONTH), $javainput.get(java.util.Calendar.DATE));&quot;,
post=&quot; $javainput.set(temp$javainput.getYear(), temp$javainput.getMonth(), &quot;
&quot;temp$javainput.getDay(), 0, 0, 0);&quot;,
pgcppname=&quot;temp$javainput&quot;) CDate &amp;
&quot;$javaclassname.getCPtr(temp$javainput)&quot;
</pre>
</div>
<p> The resulting generated proxy code in the <tt>Action</tt> class
follows:</p>
<div class="code">
<pre>
public class Action {
...
public static int doSomething(java.util.GregorianCalendar dateIn,
java.util.Calendar dateOut) {
CDate tempdateIn = new CDate(dateIn.get(java.util.Calendar.YEAR),
dateIn.get(java.util.Calendar.MONTH),
dateIn.get(java.util.Calendar.DATE));
CDate tempdateOut = new CDate(dateOut.get(java.util.Calendar.YEAR),
dateOut.get(java.util.Calendar.MONTH),
dateOut.get(java.util.Calendar.DATE));
try {
return exampleJNI.Action_doSomething(CDate.getCPtr(tempdateIn), tempdateIn,
CDate.getCPtr(tempdateOut), tempdateOut);
} finally {
dateOut.set(tempdateOut.getYear(), tempdateOut.getMonth(), tempdateOut.getDay(), 0, 0, 0);
}
}
static private long SwigConstructAction(java.util.GregorianCalendar date,
java.util.Calendar dateOut) {
CDate tempdate = new CDate(date.get(java.util.Calendar.YEAR),
date.get(java.util.Calendar.MONTH),
date.get(java.util.Calendar.DATE));
CDate tempdateOut = new CDate(dateOut.get(java.util.Calendar.YEAR),
dateOut.get(java.util.Calendar.MONTH),
dateOut.get(java.util.Calendar.DATE));
try {
return exampleJNI.new_Action(CDate.getCPtr(tempdate), tempdate,
CDate.getCPtr(tempdateOut), tempdateOut);
} finally {
dateOut.set(tempdateOut.getYear(), tempdateOut.getMonth(), tempdateOut.getDay(), 0, 0, 0);
}
}
public Action(java.util.GregorianCalendar date, java.util.Calendar dateOut) {
this(Action.SwigConstructAction(date, dateOut), true);
}
}
</pre>
</div>
<p> A few things to note:</p>
<ul>
<li> The &quot;javatype&quot; typemap has changed the parameter type to <tt>
java.util.GregorianCalendar</tt> or <tt>java.util.Calendar</tt> instead
of the default generated <tt>CDate</tt> proxy.</li>
<li> The code in the 'pre' attribute appears before the JNI call (<tt>
exampleJNI.new_Action</tt> / <tt>exampleJNI.Action_doSomething</tt>).</li>
<li> The code in the 'post' attribute appears after the JNI call.</li>
<li> A try .. finally block is generated with the JNI call in the try
block and 'post' code in the finally block. The alternative of just
using a temporary variable for the return value from the JNI call and
the 'post' code being generated before the return statement is not
possible given that the JNI call is in one line and comes from the
&quot;javaout&quot; typemap.</li>
<li> The temporary variables in the &quot;javain&quot; typemaps are called <tt>
temp$javain</tt>, where &quot;$javain&quot; is replaced with the parameter name.
&quot;$javain&quot; is used to mangle the variable name so that more than one <tt>
CDate &amp;</tt> type can be used as a parameter in a method, otherwise two
or more local variables with the same name would be generated.</li>
<li> The use of the &quot;javain&quot; typemap causes a constructor helper
function (<tt>SwigConstructAction</tt>) to be generated. This allows
Java code to be called before the JNI call and is required as the Java
compiler won't compile code inserted before the 'this' call.</li>
<li> The 'pgcppname' attribute is used to modify the object being passed
as the <a href="#Java_pgcpp">premature garbage collection prevention
parameter</a> (the 2nd and 4th parameters in the JNI calls).</li>
</ul>
<h2><a name="Java_directors_faq">27.10 Living with Java Directors</a></h2>
<p> This section is intended to address frequently asked questions and
frequently encountered problems when using Java directors.</p>
<ol>
<li><i>When my program starts up, it complains that</i> method_foo<i>
cannot be found in a Java method called</i> swig_module_init<i>. How do
I fix this?</i>
<p> Open up the C++ wrapper source code file and look for <code>
&quot;method_foo&quot;</code> (include the double quotes, they are important!)
Look at the JNI field descriptor and make sure that each class that
occurs in the descriptor has the correct package name in front of it.
If the package name is incorrect, put a &quot;javapackage&quot; typemap in your
SWIG interface file.</p>
</li>
<li><i>I'm compiling my code and I'm using templates. I provided a
javapackage typemap, but SWIG doesn't generate the right JNI field
descriptor.</i>
<p> Use the template's renamed name as the argument to the &quot;javapackage&quot;
typemap:</p>
<div class="code">
<pre>
%typemap(javapackage) std::vector&lt;int&gt; &quot;your.package.here&quot;
%template(VectorOfInt) std::vector&lt;int&gt;;
</pre>
</div></li>
<li>
<p><i>When I pass class pointers or references through a C++ upcall and
I try to type cast them, Java complains with a ClassCastException. What
am I doing wrong?</i></p>
<p> Normally, a non-director generated Java proxy class creates
temporary Java objects as follows:</p>
<div class="code">
<pre>
public static void MyClass_method_upcall(MyClass self, long jarg1)
{
Foo darg1 = new Foo(jarg1, false);
self.method_upcall(darg1);
}
</pre>
</div>
<p>Unfortunately, this loses the Java type information that is part of
the underlying Foo director proxy class's Java object pointer causing
the type cast to fail. The SWIG Java module's director code attempts to
correct the problem,<b> but only for director-enabled classes</b>,
since the director class retains a global reference to its Java object.
Thus, for director-enabled classes<b> and only for director-enabled
classes</b>, the generated proxy Java code looks something like:</p>
<div class="code">
<pre>
public static void MyClass_method_upcall(MyClass self, long jarg1,
Foo jarg1_object)
{
Foo darg1 = (jarg1_object != null ? jarg1_object : new Foo(jarg1, false));
self.method_upcall(darg1);
}
</pre>
</div>
<p> When you import a SWIG interface file containing class definitions,
the classes you want to be director-enabled must be have the <code>
feature(&quot;director&quot;)</code> enabled for type symmetry to work. This
applies even when the class being wrapped isn't a director-enabled
class but takes parameters that are director-enabled classes.</p>
<p> The current &quot;type symmetry&quot; design will work for simple C++
inheritance, but will most likely fail for anything more complicated
such as tree or diamond C++ inheritance hierarchies. Those who are
interested in challenging problems are more than welcome to hack the <code>
Java::Java_director_declaration</code> method in <code>
Source/Modules/java.cxx</code>.</p>
<p> If all else fails, you can use the downcastXXXXX() method to attempt
to recover the director class's Java object pointer. For the Java Foo
proxy class, the Foo director class's java object pointer can be
accessed through the javaObjectFoo() method. The generated method's
signature is:</p>
<div class="code">
<pre>
public static Foo javaObjectFoo(Foo obj);
</pre>
</div>
<p> From your code, this method is invoked as follows:</p>
<div class="code">
<pre>
public class MyClassDerived {
public void method_upcall(Foo foo_object)
{
FooDerived derived = (foo_object != null ?
(FooDerived) Foo.downcastFoo(foo_object) : null);
/* rest of your code here */
}
}
</pre>
</div>
<p> An good approach for managing downcasting is placing a static method
in each derived class that performs the downcast from the superclass,
e.g.,</p>
<div class="code">
<pre>
public class FooDerived extends Foo {
/* ... */
public static FooDerived downcastFooDerived(Foo foo_object)
{
try {
return foo_object != null ? (FooDerived) Foo.downcastFoo(foo_object);
}
catch (ClassCastException exc) {
// Wasn't a FooDerived object, some other subclass of Foo
return null;
}
}
}
</pre>
</div>
<p> Then change the code in MyClassDerived as follows:</p>
<div class="code">
<pre>
public class MyClassDerived extends MyClass {
/* ... */
public void method_upcall(Foo foo_object)
{
FooDerived derived = FooDerived.downcastFooDerived(foo_object);
/* rest of your code here */
}
}
</pre>
</div></li>
<li>
<p><i>Why isn't the proxy class declared abstract? Why aren't the
director upcall methods in the proxy class declared abstract?</i></p>
<p> Declaring the proxy class and its methods abstract would break the
JNI argument marshalling and SWIG's downcall functionality (going from
Java to C++.) Create an abstract Java subclass that inherits from the
director-enabled class instead. Using the previous Foo class example:</p>
<div class="code">
<pre>
public abstract class UserVisibleFoo extends Foo {
/** Make sure user overrides this method, it's where the upcall
* happens.
*/
public abstract void method_upcall(Foo foo_object);
/// Downcast from Foo to UserVisibleFoo
public static UserVisibleFoo downcastUserVisibleFoo(Foo foo_object)
{
try {
return foo_object != null ? (FooDerived) Foo.downcastFoo(foo_object) : null;
} catch (ClassCastException exc) {
// Wasn't a FooDerived object, some other subclass of Foo
return null;
}
}
}
</pre>
</div>
<p>This doesn't prevent the user from creating subclasses derived from
Foo, however, UserVisibleFoo provides the safety net that reminds the
user to override the <code>method_upcall()</code> method.</p>
</li>
</ol>
<h2><a name="Java_odds_ends">27.11 Odds and ends</a></h2>
<h3><a name="Java_javadoc_comments">27.11.1 JavaDoc comments</a></h3>
<p> SWIG can translate <a href="https://doxygen.org/">Doxygen</a>
comments in the C/C++ headers being wrapped to JavaDoc. For details of
this, see the <a href="Doxygen.html#Doxygen_to_javadoc">Doxygen to
Javadoc</a> section.</p>
<p> If you don't have Doxygen comments to translate then it's possible
to add JavaDoc comments from your interface file. The <tt>
%javamethodmodifiers</tt> feature can be used for adding proxy class
method comments and module class method comments. The &quot;javaimports&quot;
typemap can be hijacked for adding in proxy class JavaDoc comments. The
<tt>jniclassimports</tt> or <tt>jniclassclassmodifiers</tt> pragmas can
also be used for adding intermediary JNI class comments and likewise
the <tt>moduleimports</tt> or <tt>moduleclassmodifiers</tt> pragmas for
the module class. Here is an example adding in a proxy class and method
comment:</p>
<div class="code">
<pre>
%javamethodmodifiers Barmy::lose_marbles() &quot;
/**
* Calling this method will make you mad.
* Use with &lt;b&gt;utmost&lt;/b&gt; caution.
*/
public&quot;;
%typemap(javaimports) Barmy &quot;
/** The crazy class. Use as a last resort. */&quot;
class Barmy {
public:
void lose_marbles() {}
};
</pre>
</div>
<p> Note the &quot;public&quot; added at the end of the <tt>%javamethodmodifiers</tt>
as this is the default for this feature. The generated proxy class with
JavaDoc comments is then as follows:</p>
<div class="code">
<pre>
/** The crazy class. Use as a last resort. */
public class Barmy {
...
/**
* Calling this method will make you mad.
* Use with &lt;b&gt;utmost&lt;/b&gt; caution.
*/
public void lose_marbles() {
...
}
...
}
</pre>
</div>
<h3><a name="Java_functional_interface">27.11.2 Functional interface
without proxy classes</a></h3>
<p> It is possible to run SWIG in a mode that does not produce proxy
classes by using the -noproxy commandline option. The interface is
rather primitive when wrapping structures or classes and is accessed
through function calls to the module class. All the functions in the
module class are wrapped by functions with identical names as those in
the intermediary JNI class.</p>
<p> Consider the example we looked at when examining proxy classes:</p>
<div class="code">
<pre>
class Foo {
public:
int x;
int spam(int num, Foo* foo);
};
</pre>
</div>
<p> When using <tt>-noproxy</tt>, type wrapper classes are generated
instead of proxy classes. Access to all the functions and variables is
through a C like set of functions where the first parameter passed is
the pointer to the class, that is an instance of a type wrapper class.
Here is what the module class looks like:</p>
<div class="code">
<pre>
public class example {
public static void Foo_x_get(SWIGTYPE_p_Foo self, int x) {...}
public static int Foo_x_get(SWIGTYPE_p_Foo self) {...}
public static int Foo_spam(SWIGTYPE_p_Foo self, int num, SWIGTYPE_p_Foo foo) {...}
public static SWIGTYPE_p_Foo new_Foo() {...}
public static void delete_Foo(SWIGTYPE_p_Foo self) {...}
}
</pre>
</div>
<p> This approach is not nearly as natural as using proxy classes as the
functions need to be used like this:</p>
<div class="code">
<pre>
SWIGTYPE_p_Foo foo = example.new_Foo();
example.Foo_x_set(foo, 10);
int var = example.Foo_x_get(foo);
example.Foo_spam(foo, 20, foo);
example.delete_Foo(foo);
</pre>
</div>
<p> Unlike proxy classes, there is no attempt at tracking memory. All
destructors have to be called manually for example the <tt>
delete_Foo(foo)</tt> call above.</p>
<h3><a name="Java_using_own_jni_functions">27.11.3 Using your own JNI
functions</a></h3>
<p> You may have some hand written JNI functions that you want to use in
addition to the SWIG generated JNI functions. Adding these to your SWIG
generated package is possible using the <tt>%native</tt> directive. If
you don't want SWIG to wrap your JNI function then of course you can
simply use the <tt>%ignore</tt> directive. However, if you want SWIG to
generate just the Java code for a JNI function then use the <tt>%native</tt>
directive. The C types for the parameters and return type must be
specified in place of the JNI types and the function name must be the
native method name. For example:</p>
<div class="code">
<pre>
%native (HandRolled) void HandRolled(int, char *);
%{
JNIEXPORT void JNICALL Java_packageName_moduleName_HandRolled(JNIEnv *, jclass,
jlong, jstring);
%}
</pre>
</div>
<p> No C JNI function will be generated and the <tt>
Java_packageName_moduleName_HandRolled</tt> function will be accessible
using the SWIG generated Java native method call in the intermediary
JNI class which will look like this:</p>
<div class="code">
<pre>
public final static native void HandRolled(int jarg1, String jarg2);
</pre>
</div>
<p> and as usual this function is wrapped by another which for a global
C function would appear in the module class:</p>
<div class="code">
<pre>
public static void HandRolled(int arg0, String arg1) {
exampleJNI.HandRolled(arg0, arg1);
}
</pre>
</div>
<p> The <tt>packageName</tt> and <tt>moduleName</tt> must of course be
correct else you will get linker errors when the JVM dynamically loads
the JNI function. You may have to add in some &quot;jtype&quot;, &quot;jstype&quot;,
&quot;javain&quot; and &quot;javaout&quot; typemaps when wrapping some JNI types. Here the
default typemaps work for <tt>int</tt> and <tt>char *</tt>.</p>
<p> Note that if you're wanting to effectively<b> replace</b> the JNI
code generated for a C/C++ function then you'll need to use <tt>%ignore</tt>
as well to tell SWIG not to automatically generate a JNI wrapper for
it.</p>
<p> In summary the <tt>%native</tt> directive is telling SWIG to
generate the Java code to access the JNI C code, but not the JNI C
function itself. This directive is only really useful if you want to
mix your own hand crafted JNI code and the SWIG generated code into one
Java class or package.</p>
<h3><a name="Java_performance">27.11.4 Performance concerns and hints</a>
</h3>
<p> If you're directly manipulating huge arrays of complex objects from
Java, performance may suffer greatly when using the array functions in <tt>
arrays_java.i</tt>. Try and minimise the expensive JNI calls to C/C++
functions, perhaps by using temporary Java variables instead of
accessing the information directly from the C/C++ object.</p>
<p> Java classes without any finalizers generally speed up code
execution as there is less for the garbage collector to do. Finalizer
generation can be stopped by using an empty <tt>javafinalize</tt>
typemap:</p>
<div class="code">
<pre>
%typemap(javafinalize) SWIGTYPE &quot;&quot;
</pre>
</div>
<p> However, you will have to be careful about memory management and
make sure that you code in a call to the <tt>delete()</tt> member
function. This method normally calls the C++ destructor or <tt>free()</tt>
for C code.</p>
<h3><a name="Java_debugging">27.11.5 Debugging</a></h3>
<p> The generated code can be debugged using both a Java debugger and a
C++ debugger using the usual debugging techniques. Breakpoints can be
set in either Java or C++ code and so both can be debugged
simultaneously. Most debuggers do not understand both Java and C++,
with one notable exception of Sun Studio, where it is possible to step
from Java code into a JNI method within one environment.</p>
<p> Alternatively, debugging can involve placing debug printout
statements in the JNI layer using the <tt>%exception</tt> directive.
See the <a href="#Customization_exception_special_variables">special
variables for %exception</a> section. Many of the default typemaps can
also be overridden and modified for adding in extra logging/debug
display information.</p>
<p> The <tt>-Xcheck:jni</tt> and <tt>-Xcheck:nabounds</tt> Java
executable options are useful for debugging to make sure the JNI code
is behaving. The -verbose:jni and -verbose:gc are also useful options
for monitoring code behaviour.</p>
<h2><a name="Java_examples">27.12 Java Examples</a></h2>
<p> The directory Examples/java has a number of further examples. Take a
look at these if you want to see some of the techniques described in
action. The Examples/index.html file in the parent directory contains
the SWIG Examples Documentation and is a useful starting point. If your
SWIG installation went well Unix users should be able to type <tt>make</tt>
in each example directory, then <tt>java main</tt> to see them running.
For the benefit of Windows users, there are also Visual C++ project
files in a couple of the <a href="#Windows_examples">Windows Examples</a>
. There are also many regression tests in the Examples/test-suite
directory. Many of these have runtime tests in the java subdirectory.</p>
<HR NOSHADE>
<h1><a name="Javascript">28 SWIG and Javascript</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Javascript_overview">Overview</a></li>
<li><a href="#Javascript_preliminaries">Preliminaries</a>
<ul>
<li><a href="#Javascript_running_swig">Running SWIG</a></li>
<li><a href="#Javascript_running_tests_examples">Running Tests and
Examples</a></li>
<li><a href="#Javascript_known_issues">Known Issues</a></li>
</ul>
</li>
<li><a href="#Javascript_integration">Integration</a>
<ul>
<li><a href="#Javascript_node_extensions">Creating node.js Extensions</a>
<ul>
<li><a href="#Javascript_using_yeoman">Using <code>yeoman</code> to
generate a Node-API skeleton</a></li>
<li><a href="#Javascript_troubleshooting">Troubleshooting</a></li>
</ul>
</li>
<li><a href="#Javascript_embedded_webkit">Embedded Webkit</a>
<ul>
<li><a href="#Javascript_osx">Mac OS X</a></li>
<li><a href="#Javascript_gtk">GTK</a></li>
</ul>
</li>
<li><a href="#Javascript_applications_webkit">Creating Applications with
node-webkit</a></li>
</ul>
</li>
<li><a href="#Javascript_examples">Examples</a>
<ul>
<li><a href="#Javascript_simple_example">Simple</a></li>
<li><a href="#Javascript_class_example">Class</a></li>
</ul>
</li>
<li><a href="#Javascript_implementation">Implementation</a>
<ul>
<li><a href="#Javascript_source_code">Source Code</a></li>
<li><a href="#Javascript_code_templates">Code Templates</a></li>
<li><a href="#Javascript_emitter">Emitter</a></li>
<li><a href="#Javascript_emitter_states">Emitter states</a></li>
<li><a href="#Javascript_jsc_exceptions">Handling Exceptions in
JavascriptCore</a></li>
<li><a href="#Javascript_napi_exceptions">Handling Exceptions in
Node-API</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p>This chapter describes SWIG's support of Javascript. It does not
cover SWIG basics, but only information that is specific to this
module.</p>
<h2><a name="Javascript_overview">28.1 Overview</a></h2>
<p>Javascript is a prototype-based scripting language that is dynamic,
weakly typed and has first-class functions. Its arguably the most
popular language for web development. Javascript has gone beyond being
a browser-based scripting language and can be used as a backend
development language with <a href="https://nodejs.org">node.js</a>.</p>
<p>Native Javascript extensions can be used for applications that embed
a web-browser view or that embed a Javascript engine (such as<em>
node.js</em>). Extending a general purpose web-browser is not possible
as this would be a severe security issue.</p>
<p>SWIG/Javascript currently supports:</p>
<ul>
<li><strong>Node.js</strong> 12 or later. SWIG can optionally generate
Node-API code, which needs Node.js 12.17 or later.</li>
<li><strong>JavascriptCore</strong> 4.0 or later. JavascriptCore is used
by Safari/Webkit</li>
<li><a href="https://v8.dev/"><strong>v8</strong></a> 7.4 or later (we
currently CI test with 7.8 and 10.2). v8 is used by Chromium and
Node.js.</li>
</ul>
<p><a href="https://webkit.org/">WebKit</a> is a modern browser
implementation available as open-source which can be embedded into an
application. <a href="https://github.com/nwjs/nw.js">NW.jst</a>
provides an app runtime which uses Google's Chromium as Web-Browser
widget and node.js for javascript extensions.</p>
<h2><a name="Javascript_preliminaries">28.2 Preliminaries</a></h2>
<h3><a name="Javascript_running_swig">28.2.1 Running SWIG</a></h3>
<p>Suppose that you defined a SWIG module such as the following:</p>
<div class="code">
<pre>
%module example
%{
#include &quot;example.h&quot;
%}
int gcd(int x, int y);
extern double Foo;</pre>
</div>
<p>To build a Javascript module, run SWIG using the <code>-javascript</code>
option and a desired target engine <code>-jsc</code>, <code>-v8</code>,
<code>-node</code> or <code>-napi</code>. <code>-v8</code> allows for
interfacing with a raw embedded version of V8. In this case, it is up
to the user to implement a binary module loading protocol. There are
two generators supporting Node.js. The older generator for <code>node</code>
is essentially delegating to the <code>v8</code> generator and adds
some necessary preprocessor definitions. The more recent <code>-napi</code>
generator produces <code>node-addon-api</code> that interfaces to
Node.js through Node-API. The V8 generator is more mature, while the
Node-API generator offers a number of advantages such as binary stable
ABI allowing for publishing of universal binary modules on npm,
Electron support and automatic multi-threading.</p>
<div class="shell">
<pre>
$ swig -javascript -jsc example.i</pre>
</div>
<p>If building a C++ extension, add the -c++ option:</p>
<div class="shell">
<pre>
$ swig -c++ -javascript -jsc example.i</pre>
</div>
<p>To generate code for V8, you would run swig like so:</p>
<div class="shell">
<pre>
$ swig -c++ -javascript -v8 example.i</pre>
</div>
<p>This creates a C/C++ source file <code>example_wrap.c</code> or <code>
example_wrap.cxx</code>. The generated C source file contains the
low-level wrappers that need to be compiled and linked with the rest of
your C/C++ application to create an extension module.</p>
<p>The name of the wrapper file is derived from the name of the input
file. For example, if the input file is <code>example.i</code>, the
name of the wrapper file is <code>example_wrap.c</code>. To change
this, you can use the -o option. The wrapped module will export one
function which must be called to register the module with the
Javascript interpreter. For example, if your module is named <code>
example</code> the corresponding initializer for JavascriptCore would be</p>
<div class="code">
<pre>
bool example_initialize(JSGlobalContextRef context, JSObjectRef *exports)</pre>
</div>
<p>and for v8:</p>
<div class="code">
<pre>
void example_initialize(v8::Handle&lt;v8::Object&gt; exports)</pre>
</div>
<p><b> Note</b>: be aware that <code>v8</code> has a C++ API, and thus,
the generated modules must be compiled as C++.</p>
<h3><a name="Javascript_running_tests_examples">28.2.2 Running Tests and
Examples</a></h3>
<p>The configuration for tests and examples currently supports Linux and
Mac only and not MinGW (Windows) yet.</p>
<p>The default interpreter is <code>node.js</code> as it is available on
all platforms and convenient to use.</p>
<p>Running the examples with JavascriptCore requires <code>
libjavascriptcoregtk-1.0</code> to be installed, e.g., under Ubuntu with</p>
<div class="shell">
<pre>
$ sudo apt-get install libjavascriptcoregtk-1.0-dev</pre>
</div>
<p>Running with <code>V8</code> requires <code>libv8</code>:</p>
<div class="shell">
<pre>
$ sudo apt-get install libnode-dev</pre>
</div>
<p>Running with Node-API requires <code>node-addon-api</code>:</p>
<div class="shell">
<pre>
$ sudo npm install -g node-addon-api</pre>
</div>
<p>Examples can be run using</p>
<div class="shell">
<pre>
$ make check-javascript-examples ENGINE=jsc</pre>
</div>
<p><code>ENGINE</code> can be <code>node</code>, <code>jsc</code>, <code>
v8</code>, or <code>napi</code>.</p>
<p>The test-suite can be run using</p>
<div class="shell">
<pre>
$ make check-javascript-test-suite ENGINE=jsc</pre>
</div>
<h3><a name="Javascript_known_issues">28.2.3 Known Issues</a></h3>
<p>At the moment, the Javascript generators pass all tests
syntactically, i.e., the generated source code compiles. However, there
are still remaining runtime issues.</p>
<ul>
<li>
<p>Default optional arguments do not work for all targeted interpreters
except Node-API</p>
</li>
<li>
<p>Multiple output arguments do not work for JSC</p>
</li>
<li>
<p>C89 incompatibility: the JSC generator might still generate C89
violating code</p>
</li>
<li>
<p><code>long long</code> is not supported except with Node-API</p>
</li>
<li>
<p>Javascript callbacks are not supported</p>
</li>
<li>
<p><code>instanceOf</code> does not work under JSC</p>
</li>
</ul>
<p>The primary development environment has been Linux (Ubuntu 22.04).
Windows and Mac OS X have been tested sporadically. Therefore, the
generators might have more issues on those platforms. Please report
back any problem you observe to help us improving this module quickly.</p>
<h2><a name="Javascript_integration">28.3 Integration</a></h2>
<p>This chapter gives a short introduction how to use a native
Javascript extension: as a <code>node.js</code> module, and as an
extension for an embedded Webkit.</p>
<h3><a name="Javascript_node_extensions">28.3.1 Creating node.js
Extensions</a></h3>
<p>To install <code>node.js</code> you can download an installer from
their <a href="https://launchpad.net/~chris-lea/+archive/node.js">
web-site</a> for Mac OS X and Windows. For Linux you can either build
the source yourself and run <code>sudo checkinstall</code> or keep to
the (probably stone-age) packaged version. For Ubuntu there is a <a href="https://launchpad.net/~chris-lea/+archive/ubuntu/node.js/">
PPA</a> available.</p>
<div class="shell">
<pre>
$ sudo add-apt-repository ppa:chris-lea/node.js
$ sudo apt-get update
$ sudo apt-get install nodejs</pre>
</div>
<p>As <code>v8</code> is written in C++ and comes as a C++ library it is
crucial to compile your module using the same compiler flags as used
for building v8. To make things easier, <code>node.js</code> provides a
build tool called <code>node-gyp</code>.</p>
<p>You have to install it using <code>npm</code>:</p>
<div class="shell">
<pre>
$ sudo npm install -g node-gyp</pre>
</div>
<p><code>node-gyp</code> expects a configuration file named <code>
binding.gyp</code> which is basically in JSON format and conforms to the
same format that is used with Google's build-tool <code>gyp</code>.</p>
<p><code>binding.gyp</code>:</p>
<div class="code">
<pre>
{
&quot;targets&quot;: [
{
&quot;target_name&quot;: &quot;example&quot;,
&quot;sources&quot;: [ &quot;example.cxx&quot;, &quot;example_wrap.cxx&quot; ]
}
]
}</pre>
</div>
<p>First create the wrapper using SWIG:</p>
<div class="shell">
<pre>
$ swig -javascript -node -c++ example.i</pre>
</div>
<p>Then run <code>node-gyp build</code> to actually create the module:</p>
<div class="shell">
<pre>
$ node-gyp build</pre>
</div>
<p>This will create a <code>build</code> folder containing the native
module. To use the extension you need to 'require' it in your
Javascript source file:</p>
<div class="code">
<pre>
require(&quot;./build/Release/example&quot;)</pre>
</div>
<p>A more detailed explanation is given in the <a href="#Javascript_examples">
Examples</a> section.</p>
<h4><a name="Javascript_using_yeoman">28.3.1.1 Using <code>yeoman</code>
to generate a Node-API skeleton</a></h4>
<p>If targeting Node-API, the easiest way to bootstrap a project is by
using the <code>yeoman</code> generator:</p>
<div class="shell">
<pre>
$ sudo npm install -g yo
$ sudo npm install -g generator-napi-module
$ mkdir example
$ cd example
$ yo napi-module # the choice of template is irrelevant, SWIG will replace the C++ code
$ npm install node-addon-api@latest # the yeoman version is outdated
$ swig -javascript -napi -c++ -o src/example.cc example.i
$ node-gyp configure
$ node-gyp build
</pre>
</div>
<p>There is also the <a href="https://github.com/mmomtchev/node-magickwand">
<code>node-magickwand</code></a> project that can be used as a tutorial
for building and publishing a complex C++ library to npm as a
ready-to-use real-world binary module.</p>
<h4><a name="Javascript_troubleshooting">28.3.1.2 Troubleshooting</a></h4>
<ul>
<li><em>'module' object has no attribute 'script_main'</em></li>
</ul>
<p>This error happens when <code>gyp</code> is installed as a
distribution package. It seems to be outdated. Removing it resolves the
problem.</p>
<div class="shell">
<pre>
$ sudo apt-get remove gyp</pre>
</div>
<h3><a name="Javascript_embedded_webkit">28.3.2 Embedded Webkit</a></h3>
<p>Webkit is pre-installed on Mac OS X and available as a library for
GTK.</p>
<h4><a name="Javascript_osx">28.3.2.1 Mac OS X</a></h4>
<p>There is general information about programming with WebKit on <a href="https://developer.apple.com/library/mac/documentation/cocoa/conceptual/DisplayWebContent/DisplayWebContent.html">
Apple Developer Documentation</a>. Details about <code>Cocoa</code>
programming are not covered here.</p>
<p>An integration of a native extension 'example' would look like this:</p>
<div class="code">
<pre>
#import &quot;appDelegate.h&quot;
extern bool example_initialize(JSGlobalContextRef context, JSObjectRef* exports);
@implementation ExampleAppDelegate
@synthesize webView;
- (void)addGlobalObject:(JSContextRef) context:(NSString *)objectName:(JSObjectRef) theObject {
JSObjectRef global = JSContextGetGlobalObject(context);
JSStringRef objectJSName = JSStringCreateWithCFString( (CFStringRef) objectName )
if ( objectJSName != NULL ) {
JSObjectSetProperty(context, global, objectJSName, theObject, kJSPropertyAttributeReadOnly, NULL);
JSStringRelease( objectJSName );
}
}
- (void)applicationDidFinishLaunching:(NSNotification *)aNotification {
// Start a webview with the bundled index.html file
NSString *path = [[NSBundle mainBundle] bundlePath];
NSString *url = [NSString stringWithFormat: @&quot;file://%@/Contents/Assets/index.html&quot;, path];
WebFrame *webframe = [webView mainFrame];
JSGlobalContextRef context = [webframe globalContext];
JSObjectRef example;
example_initialize(context, &amp;example);
[self addGlobalObject:context:@&quot;example&quot;:example]
JSObjectSetProperty(context, global, JSStringRef propertyName, example, JSPropertyAttributes attributes, NULL);
[ [webView mainFrame] loadRequest:
[NSURLRequest requestWithURL: [NSURL URLWithString:url] ]
];
}
@end</pre>
</div>
<h4><a name="Javascript_gtk">28.3.2.2 GTK</a></h4>
<p>There is general information about programming GTK at <a href="https://developer.gnome.org/gtk2/">
GTK documentation</a> and in the <a href="https://developer.gnome.org/gtk-tutorial/">
GTK tutorial</a>, and for Webkit there is a <a href="https://webkitgtk.org/reference/webkitgtk/stable/index.html">
Webkit GTK+ API Reference</a>.</p>
<p>An integration of a native extension 'example' would look like this:</p>
<div class="code">
<pre>
#include &lt;gtk/gtk.h&gt;
#include &lt;webkit/webkit.h&gt;
extern bool example_initialize(JSGlobalContextRef context);
int main(int argc, char* argv[])
{
// Initialize GTK+
gtk_init(&amp;argc, &amp;argv);
...
// Create a browser instance
WebKitWebView *webView = WEBKIT_WEB_VIEW(webkit_web_view_new());
WebFrame *webframe = webkit_web_view_get_main_frame(webView);
JSGlobalContextRef context = webkit_web_frame_get_global_context(webFrame);
JSObjectRef global = JSContextGetGlobalObject(context);
JSObjectRef exampleModule;
example_initialize(context, &amp;exampleModule);
JSStringRef jsName = JSStringCreateWithUTF8CString(&quot;example&quot;);
JSObjectSetProperty(context, global, jsName, exampleModule, kJSPropertyAttributeReadOnly, NULL);
JSStringRelease(jsName);
...
// Load a web page into the browser instance
webkit_web_view_load_uri(webView, &quot;https://www.webkitgtk.org/&quot;);
...
// Run the main GTK+ event loop
gtk_main();
return 0;
}</pre>
</div>
<h3><a name="Javascript_applications_webkit">28.3.3 Creating
Applications with node-webkit</a></h3>
<p>To get started with <code>node-webkit</code> there is a very
informative set of <a href="https://github.com/rogerwang/node-webkit/wiki">
wiki pages</a>.</p>
<p>Similar to <code>node.js</code>, <code>node-webkit</code> is started
from command line within a <code>node.js</code> project directory.
Native extensions are created in the very same way as for <code>node.js</code>
, except that a customized <code>gyp</code> derivate has to be used: <a href="https://github.com/rogerwang/nw-gyp">
nw-gyp</a>.</p>
<p> A simple example would have the following structure:</p>
<div class="code">
<pre>
- package.json
- app.html
- app.js
- node_modules
/ example
... (as known from node.js)
</pre>
</div>
<p> The configuration file essentially conforms to <code>node.js</code>
syntax. It has some extras to configure <code>node-webkit</code>. See
the <a href="https://github.com/rogerwang/node-webkit/wiki/Manifest-format">
Manifest</a> specification for more details.</p>
<p> <code>package.json</code>:</p>
<div class="code">
<pre>
{
&quot;name&quot;: &quot;example&quot;,
&quot;main&quot;: &quot;app.html&quot;,
&quot;window&quot;: {
&quot;show&quot;: true,
&quot;width&quot;: 800,
&quot;height&quot;: 600
}
}</pre>
</div>
<p> The <code>'main'</code> property of <code>package.json</code>
specifies a web-page to be rendered in the main window.</p>
<p> <code>app.html</code>:</p>
<div class="code">
<pre>
&lt;html&gt;
&lt;head&gt;
&lt;script src=&quot;app.js&quot;&gt;&lt;/script&gt;
&lt;/head&gt;
&lt;body&gt;
&lt;div&gt;
The greatest common divisor of
&lt;span id=&quot;x&quot;&gt;&lt;/span&gt; and
&lt;span id=&quot;y&quot;&gt;&lt;/span&gt; is
&lt;span id=&quot;z&quot;&gt;&lt;/span&gt;.
&lt;/div&gt;
&lt;/body&gt;
&lt;/html&gt;</pre>
</div>
<p> As known from <code>node.js</code> one can use <code>require</code>
to load javascript modules. Additionally, <code>node-webkit</code>
provides an API that allows manipulating the window's menu, open new
windows, and many more things.</p>
<p> <code>app.js</code>:</p>
<div class="code">
<pre>window.onload = function() {
var example = require(&quot;example&quot;);
var x = 18;
var y = 24;
var z = example.gcd(x, y);
document.querySelector('#x').innerHTML = x;
document.querySelector('#y').innerHTML = y;
document.querySelector('#z').innerHTML = z;
};</pre>
</div>
<h2><a name="Javascript_examples">28.4 Examples</a></h2>
<p>Some basic examples are shown here in more detail.</p>
<h3><a name="Javascript_simple_example">28.4.1 Simple</a></h3>
<p>The common example <code>simple</code> looks like this:</p>
<div class="code">
<pre>
/* File : example.i */
%module example
%inline %{
extern int gcd(int x, int y);
extern double Foo;
%}</pre>
</div>
<p>To make this available as a node extension a <code>binding.gyp</code>
has to be created:</p>
<div class="code">
<pre>
{
&quot;targets&quot;: [
{
&quot;target_name&quot;: &quot;example&quot;,
&quot;sources&quot;: [ &quot;example.cxx&quot;, &quot;example_wrap.cxx&quot; ]
}
]
}</pre>
</div>
<p>Then <code>node-gyp</code> is used to build the extension:</p>
<div class="shell">
<pre>
$ node-gyp configure build</pre>
</div>
<p>From a 'nodejs` application the extension would be used like this:</p>
<div class="code">
<pre>
// import the extension via require
var example = require(&quot;./build/Release/example&quot;);
// calling the global method
var x = 42;
var y = 105;
var g = example.gcd(x, y);
// Accessing the global variable
var f = example.Foo;
example.Foo = 3.1415926;</pre>
</div>
<p>First the module <code>example</code> is loaded from the previously
built extension. Global methods and variables are available in the
scope of the module.</p>
<p><b>Note</b>: ECMAScript 5, the currently implemented Javascript
standard, does not have modules. <code>node.js</code> and other
implementations provide this mechanism defined by the <a href="https://wiki.commonjs.org/wiki/CommonJS">
CommonJS</a> group. For browsers this is provided by <a href="https://browserify.org">
Browserify</a>, for instance.</p>
<h3><a name="Javascript_class_example">28.4.2 Class</a></h3>
<p>The common example <code>class</code> defines three classes, <code>
Shape</code>, <code>Circle</code>, and <code>Square</code>:</p>
<div class="code">
<pre>
class Shape {
public:
Shape() {
nshapes++;
}
virtual ~Shape() {
nshapes--;
}
double x, y;
void move(double dx, double dy);
virtual double area(void) = 0;
virtual double perimeter(void) = 0;
static int nshapes;
};
class Circle : public Shape {
private:
double radius;
public:
Circle(double r) : radius(r) { }
virtual double area(void);
virtual double perimeter(void);
};
class Square : public Shape {
private:
double width;
public:
Square(double w) : width(w) { }
virtual double area(void);
virtual double perimeter(void);
};</pre>
</div>
<p><code>Circle</code> and <code>Square</code> inherit from <code>Shape</code>
. <code>Shape</code> has a static variable <code>nshapes</code>, a
function <code>move</code> that can't be overridden (non-virtual), and
two abstract functions <code>area</code> and <code>perimeter</code>
(pure virtual) that must be overridden by the sub-classes.</p>
<p>A <code>nodejs</code> extension is built the same way as for the <code>
simple</code> example.</p>
<p>In Javascript it can be used as follows:</p>
<div class="code">
<pre>
var example = require(&quot;./build/Release/example&quot;);
// local aliases for convenience
var Shape = example.Shape;
var Circle = example.Circle;
var Square = example.Square;
// creating new instances using the 'new' operator
var c = new Circle(10);
var s = new Square(10);
// accessing a static member
Shape.nshapes;
// accessing member variables
c.x = 20;
c.y = 30;
s.x = -10;
s.y = 5;
// calling some methods
c.area();
c.perimeter();
s.area();
s.perimeter();
// instantiation of Shape is not permitted
new Shape();</pre>
</div>
<p>Running these commands in an interactive node shell results in the
following output:</p>
<div class="shell">
<pre>
$ node -i
&amp; var example = require(&quot;./build/Release/example&quot;);
undefined
&amp; var Shape = example.Shape;
undefined
&amp; var Circle = example.Circle;
undefined
&amp; var Square = example.Square;
undefined
&amp; var c = new Circle(10);
undefined
&amp; var s = new Square(10);
undefined
&amp; Shape.nshapes;
2
&amp; c.x = 20;
20
&amp; c.y = 30;
30
&amp; s.x = -10;
-10
&amp; s.y = 5;
5
&amp; c.area();
314.1592653589793
&amp; c.perimeter();
62.83185307179586
&amp; s.area();
100
&amp; s.perimeter();
40
&amp; c.move(40, 40)
undefined
&amp; c.x
60
&amp; c.y
70
&amp; new Shape()
Error: Class Shape can not be instantiated
at repl:1:2
at REPLServer.self.eval (repl.js:110:21)
at Interface.&lt;anonymous&gt; (repl.js:239:12)
at Interface.EventEmitter.emit (events.js:95:17)
at Interface._onLine (readline.js:202:10)
at Interface._line (readline.js:531:8)
at Interface._ttyWrite (readline.js:760:14)
at ReadStream.onkeypress (readline.js:99:10)
at ReadStream.EventEmitter.emit (events.js:98:17)
at emitKey (readline.js:1095:12)</pre>
</div>
<p><b> Note</b>: In ECMAScript 5 there is no concept for classes.
Instead each function can be used as a constructor function which is
executed by the 'new' operator. Furthermore, during construction the
key property <code>prototype</code> of the constructor function is used
to attach a prototype instance to the created object. A prototype is
essentially an object itself that is the first-class delegate of a
class used whenever the access to a property of an object fails. The
very same prototype instance is shared among all instances of one type.
Prototypal inheritance is explained in more detail on in <a href="https://developer.mozilla.org/en-US/docs/Web/JavaScript/Inheritance_and_the_prototype_chain">
Inheritance and the prototype chain</a>, for instance.</p>
<h2><a name="Javascript_implementation">28.5 Implementation</a></h2>
<p>The Javascript Module implementation has taken a very different
approach compared to other language modules in order to support
different Javascript interpreters.</p>
<h3><a name="Javascript_source_code">28.5.1 Source Code</a></h3>
<p>The Javascript module is implemented in <code>
Source/Modules/javascript.cxx</code>. It dispatches the code generation
to a <code>JSEmitter</code> instance, <code>V8Emitter</code>, <code>
JSCEmitter</code> or <code>NAPIEmitter</code>. Additionally there are
some helpers: <code>Template</code>, for templated code generation, and
<code>JSEmitterState</code>, which is used to manage state information
during AST traversal. This rough map shall make it easier to find a way
through this huge source file:</p>
<div class="code">
<pre>
// module wide defines
#define NAME &quot;name&quot;
...
// ###############################
// # Helper class declarations
class JSEmitterState { ... };
class Template { ... };
// ###############################
// # JSEmitter declaration
class JSEmitter { ... };
// Emitter factory declarations
JSEmitter *swig_javascript_create_JSCEmitter();
JSEmitter *swig_javascript_create_V8Emitter();
JSEmitter *swig_javascript_create_NAPIEmitter();
// ###############################
// # Javascript module
// Javascript module declaration
class JAVASCRIPT:public Language { ... };
// Javascript module implementation
int JAVASCRIPT::functionWrapper(Node *n) { ... }
...
// Module factory implementation
static Language *new_swig_javascript() { ... }
extern &quot;C&quot; Language *swig_javascript(void) { ... }
// ###############################
// # JSEmitter base implementation
JSEmitter::JSEmitter() { ... }
Template JSEmitter::getTemplate(const String *name) { ... }
...
// ###############################
// # JSCEmitter
// JSCEmitter declaration
class JSCEmitter: public JSEmitter { ... };
// JSCEmitter implementation
JSCEmitter::JSCEmitter() { ... }
void JSCEmitter::marshalInputArgs(Node *n, ParmList *parms, Wrapper *wrapper, MarshallingMode mode, bool is_member, bool is_static) { ... }
...
// JSCEmitter factory
JSEmitter *swig_javascript_create_JSCEmitter() { ... }
// ###############################
// # V8Emitter
// V8Emitter declaration
class V8Emitter: public JSEmitter { ... };
// V8Emitter implementation
V8Emitter::V8Emitter() { ... }
int V8Emitter::initialize(Node *n) { ... }
// V8Emitter factory
JSEmitter *swig_javascript_create_V8Emitter() { ... }
// ###############################
// # Helper implementation (JSEmitterState, Template)
JSEmitterState::JSEmitterState() { ... }
...
Template::Template(const String *code_) { ... }
...</pre>
</div>
<h3><a name="Javascript_code_templates">28.5.2 Code Templates</a></h3>
<p>All generated code is created on the basis of code templates. The
templates for<em> JavascriptCore</em> can be found in <code>
Lib/javascript/jsc/javascriptcode.swg</code>, for<em> v8</em> in <code>
Lib/javascript/v8/javascriptcode.swg</code> and for<em> Node-API</em> in
<code>Lib/javascript/napi/javascriptcode.swg</code>.</p>
<p>To track the originating code template for generated code you can run</p>
<div class="shell">
<pre>
$ swig -javascript -jsc -debug-codetemplates</pre>
</div>
<p>which wraps generated code with a descriptive comment</p>
<div class="code">
<pre>
/* begin fragment(&quot;template_name&quot;) */
...generated code ...
/* end fragment(&quot;template_name&quot;) */</pre>
</div>
<p>The Template class is used like this:</p>
<div class="code">
<pre>
Template t_register = getTemplate(&quot;jsv8_register_static_variable&quot;);
t_register.replace(&quot;$jsparent&quot;, state.clazz(NAME_MANGLED))
.replace(&quot;$jsname&quot;, state.variable(NAME))
.replace(&quot;$jsgetter&quot;, state.variable(GETTER))
.replace(&quot;$jssetter&quot;, state.variable(SETTER))
.trim().
print(f_init_static_wrappers);</pre>
</div>
<p>A code template is registered with the<em> JSEmitter</em> via <code>
fragment(name, &quot;template&quot;)</code>, e.g.,</p>
<div class="code">
<pre>
%fragment (&quot;jsc_variable_declaration&quot;, &quot;templates&quot;)
%{
{&quot;$jsname&quot;, $jsgetter, $jssetter, kJSPropertyAttributeNone},
%}</pre>
</div>
<p><code>Template</code> creates a copy of that string and <code>
Template::replace</code> uses Swig's <code>Replaceall</code> to replace
variables in the template. <code>Template::trim</code> can be used to
eliminate leading and trailing whitespaces. <code>Template::print</code>
is used to write the final template string to a Swig <code>DOH</code>
(based on <code>Printv</code>). All methods allow chaining.</p>
<h3><a name="Javascript_emitter">28.5.3 Emitter</a></h3>
<p>The Javascript module delegates code generation to a <code>JSEmitter</code>
instance. The following extract shows the essential interface:</p>
<div class="code">
<pre>
class JSEmitter {
...
/**
* Opens output files and temporary output DOHs.
*/
virtual int initialize(Node *n);
/**
* Writes all collected code into the output file(s).
*/
virtual int dump(Node *n) = 0;
/**
* Cleans up all open output DOHs.
*/
virtual int close() = 0;
...
/**
* Invoked at the beginning of the classHandler.
*/
virtual int enterClass(Node *);
/**
* Invoked at the end of the classHandler.
*/
virtual int exitClass(Node *) {
return SWIG_OK;
}
/**
* Invoked at the beginning of the variableHandler.
*/
virtual int enterVariable(Node *);
/**
* Invoked at the end of the variableHandler.
*/
virtual int exitVariable(Node *) {
return SWIG_OK;
}
/**
* Invoked at the beginning of the functionHandler.
*/
virtual int enterFunction(Node *);
/**
* Invoked at the end of the functionHandler.
*/
virtual int exitFunction(Node *) {
return SWIG_OK;
}
/**
* Invoked by functionWrapper callback after call to Language::functionWrapper.
*/
virtual int emitWrapperFunction(Node *n);
/**
* Invoked from constantWrapper after call to Language::constantWrapper.
**/
virtual int emitConstant(Node *n);
/**
* Registers a given code snippet for a given key name.
*
* This method is called by the fragmentDirective handler
* of the JAVASCRIPT language module.
**/
int registerTemplate(const String *name, const String *code);
/**
* Retrieve the code template registered for a given name.
*/
Template getTemplate(const String *name);
State &amp;getState();
...
}</pre>
</div>
<p>The module calls <code>initialize</code>, <code>dump</code>, and <code>
close</code> from within the <code>top</code> method:</p>
<div class="code">
<pre>
int JAVASCRIPT::top(Node *n) {
emitter-&gt;initialize(n);
Language::top(n);
emitter-&gt;dump(n);
emitter-&gt;close();
return SWIG_OK;
}</pre>
</div>
<p>The methods <code>enterClass</code> and <code>exitClass</code> are
called from within the <code>classHandler</code> method:</p>
<div class="code">
<pre>
int JAVASCRIPT::classHandler(Node *n) {
emitter-&gt;enterClass(n);
Language::classHandler(n);
emitter-&gt;exitClass(n);
return SWIG_OK;
}</pre>
</div>
<p>In <code>enterClass</code> the emitter stores state information that
is necessary when processing class members. In <code>exitClass</code>
the wrapper code for the whole class is generated.</p>
<h3><a name="Javascript_emitter_states">28.5.4 Emitter states</a></h3>
<p>For storing information during the AST traversal the emitter provides
a <code>JSEmitterState</code> with different slots to store data
representing the scopes global, class, function, and variable.</p>
<div class="code">
<pre>
class JSEmitterState {
public:
JSEmitterState();
~JSEmitterState();
DOH *global();
DOH *global(const char* key, DOH *initial = 0);
DOH *clazz(bool reset = false);
DOH *clazz(const char* key, DOH *initial = 0);
DOH *function(bool reset = false);
DOH *function(const char* key, DOH *initial = 0);
DOH *variable(bool reset = false);
DOH *variable(const char* key, DOH *initial = 0);
static int IsSet(DOH *val);
...
};</pre>
</div>
<p>When entering a scope, such as in <code>enterClass</code>, the
corresponding state is reset and new data is stored:</p>
<div class="code">
<pre>
state.clazz(RESET);
state.clazz(NAME, Getattr(n, &quot;sym:name&quot;));</pre>
</div>
<p>State information can be retrieved using <code>state.clazz(NAME)</code>
or with <code>Getattr</code> on <code>state.clazz()</code> which
actually returns a <code>Hash</code> instance.</p>
<h3><a name="Javascript_jsc_exceptions">28.5.5 Handling Exceptions in
JavascriptCore</a></h3>
<p>Applications with an embedded JavascriptCore should be able to
present detailed exception messages that occur in the Javascript
engine. Below is an example derived from code provided by Brian Barnes
on how these exception details can be extracted.</p>
<div class="code">
<pre>
void script_exception_to_string(JSContextRef js_context, JSValueRef exception_value_ref, char* return_error_string, int return_error_string_max_length)
{
JSObjectRef exception_object;
JSValueRef value_ref;
JSStringRef jsstring_property_name = NULL;
JSValueRef temporary_exception = NULL;
JSStringRef js_return_string = NULL;
size_t bytes_needed;
char* c_result_string = NULL;
exception_object = JSValueToObject(js_context, exception_value_ref, NULL);
/* source url */
strcpy(return_error_string, &quot;[&quot;);
jsstring_property_name = JSStringCreateWithUTF8CString(&quot;sourceURL&quot;);
value_ref = JSObjectGetProperty(js_context, exception_object, jsstring_property_name, &amp;temporary_exception);
JSStringRelease(jsstring_property_name);
js_return_string = JSValueToStringCopy(js_context, value_ref, NULL);
bytes_needed = JSStringGetMaximumUTF8CStringSize(js_return_string);
c_result_string = (char*)calloc(bytes_needed, sizeof(char));
JSStringGetUTF8CString(js_return_string, c_result_string, bytes_needed);
JSStringRelease(js_return_string);
strncat(return_error_string, c_result_string, return_error_string_max_length-1);
free(c_result_string);
strncat(return_error_string, &quot;:&quot;, return_error_string_max_length-1);
/* line number */
jsstring_property_name = JSStringCreateWithUTF8CString(&quot;line&quot;);
value_ref = JSObjectGetProperty(js_context, exception_object, jsstring_property_name, &amp;temporary_exception);
JSStringRelease(jsstring_property_name);
js_return_string = JSValueToStringCopy(js_context, value_ref, NULL);
bytes_needed = JSStringGetMaximumUTF8CStringSize(js_return_string);
c_result_string = (char*)calloc(bytes_needed, sizeof(char));
JSStringGetUTF8CString(js_return_string, c_result_string, bytes_needed);
JSStringRelease(js_return_string);
strncat(return_error_string, c_result_string, return_error_string_max_length-1);
free(c_result_string);
strncat(return_error_string, &quot;]&quot;, return_error_string_max_length-1);
/* error message */
jsstring_property_name = JSStringCreateWithUTF8CString(&quot;message&quot;);
value_ref = JSObjectGetProperty(js_context, exception_object, jsstring_property_name, &amp;temporary_exception);
JSStringRelease(jsstring_property_name);
if(NULL == value_ref)
{
strncat(return_error_string, &quot;Unknown Error&quot;, return_error_string_max_length-1);
}
else
{
js_return_string = JSValueToStringCopy(js_context, value_ref, NULL);
bytes_needed = JSStringGetMaximumUTF8CStringSize(js_return_string);
c_result_string = (char*)calloc(bytes_needed, sizeof(char));
JSStringGetUTF8CString(js_return_string, c_result_string, bytes_needed);
JSStringRelease(js_return_string);
strncat(return_error_string, c_result_string, return_error_string_max_length-1);
free(c_result_string);
}
}</pre>
</div>
<p>It would be used in the following way:</p>
<div class="code">
<pre>
if(js_exception)
{
char return_error_string[256];
script_exception_to_string(js_context, js_exception, return_error_string, 256);
printf(&quot;Compile error is %s&quot;, return_error_string);
}</pre>
</div>
<h3><a name="Javascript_napi_exceptions">28.5.6 Handling Exceptions in
Node-API</a></h3>
<p>Node-API is the only generator that provides fully automatic
conversion of C++ exceptions to JavaScript exceptions when building
with C++ exceptions enabled in `binding.gyp`:</p>
<div class="code">
<pre>
'conditions': [
['OS==&quot;mac&quot;',
{
'xcode_settings': {
'GCC_ENABLE_CPP_RTTI': 'YES',
'GCC_ENABLE_CPP_EXCEPTIONS' : 'YES'
}
}
],
['OS==&quot;linux&quot; or OS==&quot;freebsd&quot; or OS==&quot;openbsd&quot; or OS==&quot;solaris&quot;',
{
'cflags!': [ '-fno-exceptions' ],
'cflags_cc!': [ '-fno-exceptions', '-fno-rtti' ]
}
]
]
</pre>
</div>
<p>In this case, nothing else is needed for the C++ exceptions to be
passed to JavaScript.</p>
<HR NOSHADE>
<h1><a name="Lua">29 SWIG and Lua</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Lua_nn2">Preliminaries</a></li>
<li><a href="#Lua_nn3">Running SWIG</a>
<ul>
<li><a href="#Lua_commandline">Additional command line options</a></li>
<li><a href="#Lua_nn4">Compiling and Linking and Interpreter</a></li>
<li><a href="#Lua_nn5">Compiling a dynamic module</a></li>
<li><a href="#Lua_nn6">Using your module</a></li>
</ul>
</li>
<li><a href="#Lua_nn7">A tour of basic C/C++ wrapping</a>
<ul>
<li><a href="#Lua_nn8">Modules</a></li>
<li><a href="#Lua_nn9">Functions</a></li>
<li><a href="#Lua_nn10">Global variables</a></li>
<li><a href="#Lua_nn11">Constants and enums</a>
<ul>
<li><a href="#Lua_nn13">Constants/enums and classes/structures</a></li>
</ul>
</li>
<li><a href="#Lua_nn12">Pointers</a></li>
<li><a href="#Lua_structures">Structures</a></li>
<li><a href="#Lua_nn14">C++ classes</a></li>
<li><a href="#Lua_nn15">C++ inheritance</a></li>
<li><a href="#Lua_nn16">Pointers, references, values, and arrays</a></li>
<li><a href="#Lua_nn17">C++ overloaded functions</a></li>
<li><a href="#Lua_nn18">C++ operators</a></li>
<li><a href="#Lua_nn19">Class extension with %extend</a></li>
<li><a href="#Lua_nn20">Using %newobject to release memory</a></li>
<li><a href="#Lua_nn21">C++ templates</a></li>
<li><a href="#Lua_nn22">C++ Smart Pointers</a></li>
<li><a href="#Lua_nn23">C++ Exceptions</a></li>
<li><a href="#Lua_namespaces">Namespaces</a>
<ul>
<li><a href="#Lua_nn27">Compatibility Note</a></li>
<li><a href="#Lua_nn29">Names</a></li>
<li><a href="#Lua_nn30">Inheritance</a></li>
</ul>
</li>
</ul>
</li>
<li><a href="#Lua_nn24">Typemaps</a>
<ul>
<li><a href="#Lua_nn25">What is a typemap?</a></li>
<li><a href="#Lua_nn26">Using typemaps</a></li>
<li><a href="#Lua_typemap_arrays">Typemaps and arrays</a></li>
<li><a href="#Lua_typemaps_ptr_ptr_functions">Typemaps and
pointer-pointer functions</a></li>
</ul>
</li>
<li><a href="#Lua_writing_typemaps">Writing typemaps</a>
<ul>
<li><a href="#Lua_typemaps_write">Typemaps you can write</a></li>
<li><a href="#Lua_nn31">SWIG's Lua-C API</a></li>
</ul>
</li>
<li><a href="#Lua_nn32">Customization of your Bindings</a>
<ul>
<li><a href="#Lua_nn33">Writing your own custom wrappers</a></li>
<li><a href="#Lua_nn34">Adding additional Lua code</a></li>
</ul>
</li>
<li><a href="#Lua_nn35">Details on the Lua binding</a>
<ul>
<li><a href="#Lua_nn36">Binding global data into the module.</a></li>
<li><a href="#Lua_nn37">Userdata and Metatables</a></li>
<li><a href="#Lua_nn38">Memory management</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> Lua is an extension programming language designed to support general
procedural programming with data description facilities. It also offers
good support for object-oriented programming, functional programming,
and data-driven programming. Lua is intended to be used as a powerful,
light-weight configuration language for any program that needs one. Lua
is implemented as a library, written in clean C (that is, in the common
subset of ISO C and C++). It's also a<em> really</em> tiny language,
less than 6000 lines of code, which compiles to &lt;100 kilobytes of
binary code. It can be found at <a href="https://www.lua.org">
https://www.lua.org</a></p>
<p> eLua stands for Embedded Lua (can be thought of as a flavor of Lua)
and offers the full implementation of the Lua programming language to
the embedded world, extending it with specific features for efficient
and portable software embedded development. eLua runs on smaller
devices like microcontrollers and provides the full features of the
regular Lua desktop version. More information on eLua can be found
here: <a href="http://www.eluaproject.net">http://www.eluaproject.net</a>
</p>
<h2><a name="Lua_nn2">29.1 Preliminaries</a></h2>
<p> The current SWIG implementation is designed to work with Lua 5.0.x,
5.1.x and 5.2.x. It should work with later versions of Lua, but
certainly not with Lua 4.0 due to substantial API changes. It is
possible to either static link or dynamic link a Lua module into the
interpreter (normally Lua static links its libraries, as dynamic
linking is not available on all platforms). SWIG also has support for
eLua starting from eLua 0.8. Due to substantial changes between SWIG
2.x and SWIG 3.0 and unavailability of testing platform, eLua status
was downgraded to 'experimental'.</p>
<h2><a name="Lua_nn3">29.2 Running SWIG</a></h2>
<p> Suppose that you defined a SWIG module such as the following:</p>
<div class="code">
<pre>
%module example
%{
#include &quot;example.h&quot;
%}
int gcd(int x, int y);
extern double Foo;
</pre>
</div>
<p> To build a Lua module, run SWIG using the <tt>-lua</tt> option.</p>
<div class="shell">
<pre>
$ swig -lua example.i
</pre>
</div>
<p> If building a C++ extension, add the <tt>-c++</tt> option:</p>
<div class="shell">
<pre>
$ swig -c++ -lua example.i
</pre>
</div>
<p> This creates a C/C++ source file <tt>example_wrap.c</tt> or <tt>
example_wrap.cxx</tt>. The generated C source file contains the
low-level wrappers that need to be compiled and linked with the rest of
your C/C++ application to create an extension module.</p>
<p> The name of the wrapper file is derived from the name of the input
file. For example, if the input file is <tt>example.i</tt>, the name of
the wrapper file is <tt>example_wrap.c</tt>. To change this, you can
use the -o option. The wrapped module will export one function <tt>&quot;int
luaopen_example(lua_State* L)&quot;</tt> which must be called to register
the module with the Lua interpreter. The name &quot;luaopen_example&quot; depends
upon the name of the module.</p>
<p> To build an eLua module, run SWIG using <tt>-lua</tt> and add either
<tt>-elua</tt> or <tt>-eluac</tt>.</p>
<div class="shell">
<pre>
$ swig -lua -elua example.i
</pre>
</div>
<p> or</p>
<div class="shell">
<pre>
$ swig -lua -eluac example.i
</pre>
</div>
<p> The <tt>-elua</tt> option puts all the C function wrappers and
variable get/set wrappers in rotables. It also generates a metatable
which will control the access to these variables from eLua. It also
offers a significant amount of module size compression. On the other
hand, the <tt>-eluac</tt> option puts all the wrappers in a single
rotable. With this option, no matter how huge the module, it will
consume no additional microcontroller SRAM (crass compression). There
is a catch though: Metatables are not generated with <tt>-eluac</tt>.
To access any value from eLua, one must directly call the wrapper
function associated with that value.</p>
<h3><a name="Lua_commandline">29.2.1 Additional command line options</a></h3>
<p> The following table list the additional commandline options
available for the Lua module. They can also be seen by using:</p>
<div class="shell">
<pre>
swig -lua -help
</pre>
</div>
<table summary="Lua specific options">
<tr><th>Lua specific options</th></tr>
<tr><td>-elua</td><td>Generates LTR compatible wrappers for smaller
devices running elua.</td></tr>
<tr><td>-eluac</td><td>LTR compatible wrappers in &quot;crass compress&quot; mode
for elua.</td></tr>
<tr><td>-nomoduleglobal</td><td>Do not register the module name as a
global variable but return the module table from calls to require.</td></tr>
<tr><td>-no-old-metatable-bindings</td><td>Disable backward
compatibility: old-style binding names generations and a few other
things. Explanations are included in appropriate later sections. This
option is considered deprecated and will be removed in the near future.</td>
</tr>
<tr><td>-squash-bases</td><td>Squashes symbols from all inheritance tree
of a given class into itself. Emulates pre-SWIG3.0 inheritance.
Insignificantly speeds things up, but increases memory consumption.</td>
</tr>
</table>
<h3><a name="Lua_nn4">29.2.2 Compiling and Linking and Interpreter</a></h3>
<p> Normally Lua is embedded into another program and will be statically
linked. An extremely simple stand-alone interpreter (<tt>min.c</tt>) is
given below:</p>
<div class="code">
<pre>
#include &lt;stdio.h&gt;
#include &quot;lua.h&quot;
#include &quot;lualib.h&quot;
#include &quot;lauxlib.h&quot;
extern int luaopen_example(lua_State* L); // declare the wrapped module
int main(int argc, char* argv[])
{
lua_State *L;
if (argc&lt;2)
{
printf(&quot;%s: &lt;filename.lua&gt;\n&quot;, argv[0]);
return 0;
}
L=lua_open();
luaopen_base(L); // load basic libs (eg. print)
luaopen_example(L); // load the wrapped module
if (luaL_loadfile(L, argv[1])==0) // load and run the file
lua_pcall(L, 0, 0, 0);
else
printf(&quot;unable to load %s\n&quot;, argv[1]);
lua_close(L);
return 0;
}
</pre>
</div>
<p> A much improved set of code can be found in the Lua distribution <tt>
src/lua/lua.c</tt>. Include your module, just add the external
declaration &amp; add a <tt>#define LUA_EXTRALIBS {&quot;example&quot;,
luaopen_example}</tt>, at the relevant place.</p>
<p> The exact commands for compiling and linking vary from platform to
platform. Here is a possible set of commands of doing this:</p>
<div class="shell">
<pre>
$ swig -lua example.i -o example_wrap.c
$ gcc -I/usr/include/lua -c min.c -o min.o
$ gcc -I/usr/include/lua -c example_wrap.c -o example_wrap.o
$ gcc -c example.c -o example.o
$ gcc -I/usr/include/lua -L/usr/lib/lua min.o example_wrap.o example.o -o my_lua
</pre>
</div>
<p> For eLua, the source must be built along with the wrappers generated
by SWIG. Make sure the eLua source files <tt>platform_conf.h</tt> and <tt>
auxmods.h</tt> are updated with the entries of your new module. Please
note: <tt>&quot;mod&quot;</tt> is the module name.</p>
<div class="code">
<pre>
/* Sample platform_conf.h */
#define LUA_PLATFORM_LIBS_ROM\
_ROM( AUXLIB_PIO, luaopen_pio, pio_map )\
_ROM( AUXLIB_TMR, luaopen_tmr, tmr_map )\
_ROM( AUXLIB_MOD, luaopen_mod, mod_map )\
....
</pre>
</div><div class="code">
<pre>
/* Sample auxmods.h */
#define AUXLIB_PIO &quot;pio&quot;
LUALIB_API int ( luaopen_pio )(lua_State *L );
#define AUXLIB_MOD &quot;mod&quot;
LUALIB_API int ( luaopen_mod )(lua_State *L );
....
</pre>
</div>
<p> More information on building and configuring eLua can be found here:
<a href="http://www.eluaproject.net/doc/v0.8/en_building.html">
http://www.eluaproject.net/doc/v0.8/en_building.html</a></p>
<h3><a name="Lua_nn5">29.2.3 Compiling a dynamic module</a></h3>
<p> Most, but not all platforms support the dynamic loading of modules
(Windows &amp; Linux do). Refer to the Lua manual to determine if your
platform supports it. For compiling a dynamically loaded module the
same wrapper can be used. Assuming you have code you need to link to in
a file called <tt>example.c</tt>, the commands will be something like
this:</p>
<div class="shell">
<pre>
$ swig -lua example.i -o example_wrap.c
$ gcc -fPIC -I/usr/include/lua -c example_wrap.c -o example_wrap.o
$ gcc -fPIC -c example.c -o example.o
$ gcc -shared -I/usr/include/lua -L/usr/lib/lua example_wrap.o example.o -o example.so
</pre>
</div>
<p> The wrappers produced by SWIG can be compiled and linked with Lua
5.1.x and later. The loading is extremely simple.</p>
<div class="targetlang">
<pre>
require(&quot;example&quot;)
</pre>
</div>
<p> For those using Lua 5.0.x, you will also need an interpreter with
the loadlib function (such as the default interpreter compiled with
Lua). In order to dynamically load a module you must call the loadlib
function with two parameters: the filename of the shared library, and
the function exported by SWIG. Calling loadlib should return the
function, which you then call to initialise the module</p>
<div class="targetlang">
<pre>
my_init=loadlib(&quot;example.so&quot;, &quot;luaopen_example&quot;) -- for Unix/Linux
--my_init=loadlib(&quot;example.dll&quot;, &quot;luaopen_example&quot;) -- for Windows
assert(my_init) -- make sure it's not nil
my_init() -- call the init fn of the lib
</pre>
</div>
<p> Or can be done in a single line of Lua code</p>
<div class="targetlang">
<pre>
assert(loadlib(&quot;example.so&quot;, &quot;luaopen_example&quot;))()
</pre>
</div>
<p> If the code didn't work, don't panic. The best thing to do is to
copy the module and your interpreter into a single directory and then
execute the interpreter and try to manually load the module (take care,
all this code is case sensitive).</p>
<div class="targetlang">
<pre>
a, b, c=package.loadlib(&quot;example.so&quot;, &quot;luaopen_example&quot;) -- for Unix/Linux
--a, b, c=package.loadlib(&quot;example.dll&quot;, &quot;luaopen_example&quot;) -- for Windows
print(a, b, c)
</pre>
</div>
<p> Note: for Lua 5.0:
<br> The loadlib() function is in the global namespace, not in a
package. So it's just loadlib().</p>
<p> if 'a' is a function, this is all working fine, all you need to do
is call it</p>
<div class="targetlang">
<pre>
a()
</pre>
</div>
<p> to load your library which will add a table 'example' with all the
functions added.</p>
<p> If it doesn't work, look at the error messages, in particular
message 'b'
<br> <tt> The specified module could not be found.</tt>
<br> Means that is cannot find the module, check your the location and
spelling of the module.
<br> <tt> The specified procedure could not be found.</tt>
<br> Means that it loaded the module, but cannot find the named
function. Again check the spelling, and if possible check to make sure
the functions were exported correctly.
<br> <tt> 'loadlib' not installed/supported</tt>
<br> Is quite obvious (Go back and consult the Lua documents on how to
enable loadlib for your platform).</p>
<h3><a name="Lua_nn6">29.2.4 Using your module</a></h3>
<p> Assuming all goes well, you will be able to this:</p>
<div class="targetlang">
<pre>
$ ./my_lua
&gt; print(example.gcd(4, 6))
2
&gt; print(example.Foo)
3
&gt; example.Foo=4
&gt; print(example.Foo)
4
&gt;
</pre>
</div>
<h2><a name="Lua_nn7">29.3 A tour of basic C/C++ wrapping</a></h2>
<p> By default, SWIG tries to build a very natural Lua interface to your
C/C++ code. This section briefly covers the essential aspects of this
wrapping.</p>
<h3><a name="Lua_nn8">29.3.1 Modules</a></h3>
<p> The SWIG module directive specifies the name of the Lua module. If
you specify `module example', then everything is wrapped into a Lua
table 'example' containing all the functions and variables. When
choosing a module name, make sure you don't use the same name as a
built-in Lua command or standard module name.</p>
<h3><a name="Lua_nn9">29.3.2 Functions</a></h3>
<p> Global functions are wrapped as new Lua built-in functions. For
example,</p>
<div class="code">
<pre>
%module example
int fact(int n);</pre>
</div>
<p> creates a built-in function <tt>example.fact(n)</tt> that works
exactly like you think it does:</p>
<div class="targetlang">
<pre>
&gt; print example.fact(4)
24
&gt;
</pre>
</div>
<p> To avoid name collisions, SWIG create a Lua table which keeps all
the functions, constants, classes and global variables in. It is
possible to copy the functions, constants and classes (but not
variables) out of this and into the global environment with the
following code. This can easily overwrite existing functions, so this
must be used with care.</p>
<div class="targetlang">
<pre>
&gt; for k, v in pairs(example) do _G[k]=v end
&gt; print(fact(4))
24
&gt;
</pre>
</div>
<p> It is also possible to rename the module with an assignment.</p>
<div class="targetlang">
<pre>
&gt; e=example
&gt; print(e.fact(4))
24
&gt; print(example.fact(4))
24
</pre>
</div>
<h3><a name="Lua_nn10">29.3.3 Global variables</a></h3>
<p> Global variables (which are linked to C code) are supported, and
appear to be just another variable in Lua. However the actual mechanism
is more complex. Given a global variable:</p>
<div class="code">
<pre>%module example
extern double Foo;
</pre>
</div>
<p> SWIG will effectively generate two functions <tt>example.Foo_set()</tt>
and <tt>example.Foo_get()</tt>. It then adds a metatable to the table
'example' to call these functions at the correct time (when you attempt
to set or get examples.Foo). Therefore if you were to attempt to assign
the global to another variable, you will get a local copy within the
interpreter, which is no longer linked to the C code.</p>
<div class="targetlang">
<pre>
&gt; print(example.Foo)
3
&gt; c=example.Foo -- c is a COPY of example.Foo, not the same thing
&gt; example.Foo=4
&gt; print(c)
3
&gt; c=5 -- this will not affect the original example.Foo
&gt; print(example.Foo, c)
4 5
</pre>
</div>
<p> It is therefore not possible to 'move' the global variable into the
global namespace as it is with functions. It is however, possible to
rename the module with an assignment, to make it more convenient.</p>
<div class="targetlang">
<pre>
&gt; e=example
&gt; -- e and example are the same table
&gt; -- so e.Foo and example.Foo are the same thing
&gt; example.Foo=4
&gt; print(e.Foo)
4
</pre>
</div>
<p> If a variable is marked with the %immutable directive then any
attempts to set this variable will cause a Lua error. Given a global
variable:</p>
<div class="code">
<pre>%module example
%immutable;
extern double Foo;
%mutable;
</pre>
</div>
<p> SWIG will allow the reading of <tt>Foo</tt> but when a set attempt
is made, an error function will be called.</p>
<div class="targetlang">
<pre>
&gt; print(e.Foo) -- reading works ok
4
&gt; example.Foo=40 -- but writing does not
This variable is immutable
stack traceback:
[C]: ?
[C]: ?
stdin:1: in main chunk
[C]: ?
</pre>
</div>
<p> For those people who would rather that SWIG silently ignore the
setting of immutables (as previous versions of the Lua bindings did),
adding a <tt>-DSWIGLUA_IGNORE_SET_IMMUTABLE</tt> compile option will
remove this.</p>
<p> Unlike earlier versions of the binding, it is now possible to add
new functions or variables to the module, just as if it were a normal
table. This also allows the user to rename/remove existing functions
and constants (but not linked variables, mutable or immutable).
Therefore users are recommended to be careful when doing so.</p>
<div class="targetlang">
<pre>
&gt; -- example.PI does not exist
&gt; print(example.PI)
nil
&gt; example.PI=3.142 -- new value added
&gt; print(example.PI)
3.142
</pre>
</div>
<p> If you have used the <tt>-eluac</tt> option for your eLua module,
you will have to follow a different approach while manipulating global
variables. (This is not applicable for wrappers generated with <tt>
-elua</tt>)</p>
<div class="targetlang">
<pre>
&gt; -- Applicable only with -eluac. (num is defined)
&gt; print(example.num_get())
20
&gt; example.num_set(50) -- new value added
&gt; print(example.num_get())
50
</pre>
</div>
<p> In general, functions of the form <tt>&quot;variable_get()&quot;</tt> and <tt>
&quot;variable_set()&quot;</tt> are automatically generated by SWIG for use with <tt>
-eluac</tt>.</p>
<h3><a name="Lua_nn11">29.3.4 Constants and enums</a></h3>
<p> Because Lua doesn't really have the concept of constants, C/C++
constants are not really constant in Lua. They are actually just a copy
of the value into the Lua interpreter. Therefore they can be changed
just as any other value. For example given some constants:</p>
<div class="code">
<pre>%module example
%constant int ICONST=42;
#define SCONST &quot;Hello World&quot;
enum Days{SUNDAY, MONDAY, TUESDAY, WEDNESDAY, THURSDAY, FRIDAY, SATURDAY};
</pre>
</div>
<p> This is 'effectively' converted into the following Lua code:</p>
<div class="targetlang">
<pre>
example.ICONST=42
example.SCONST=&quot;Hello World&quot;
example.SUNDAY=0
....
</pre>
</div>
<p> Constants are not guaranteed to remain constant in Lua. The name of
the constant could be accidentally reassigned to refer to some other
object. Unfortunately, there is no easy way for SWIG to generate code
that prevents this. You will just have to be careful.</p>
<p> If you're using eLua and have used <tt>-elua</tt> or <tt>-eluac</tt>
to generate your wrapper, macro constants and enums should be accessed
through a rotable called <tt>&quot;const&quot;</tt>. In eLua, macro constants and
enums are guaranteed to remain constants since they are all contained
within a rotable. A regular C constant is accessed from eLua just as if
it were a regular global variable, just that the property of value
immutability is demonstrated if an attempt at modifying a C constant is
made.</p>
<div class="targetlang">
<pre>
&gt; print(example.ICONST)
10
&gt; print(example.const.SUNDAY)
0
&gt; print(example.const.SCONST)
Hello World
</pre>
</div>
<h4><a name="Lua_nn13">29.3.4.1 Constants/enums and classes/structures</a>
</h4>
<p> Enums are exported into a class table. For example, given some
enums:</p>
<div class="code">
<pre>%module example
enum Days { SUNDAY = 0, MONDAY, TUESDAY, WEDNESDAY, THURSDAY, FRIDAY, SATURDAY };
struct Test {
enum { TEST1 = 10, TEST2 = 20 };
#ifdef __cplusplus // There are no static members in C
static const int ICONST = 12;
#endif
};
</pre>
</div>
<p> There is a slight difference in behaviour wrapping C and C++ code
due to the different scoping rules of C and C++. The wrapped C++ code
is used as follows from Lua code:</p>
<div class="targetlang">
<pre>
&gt; print(example.SUNDAY)
0
&gt; print(example.Test.TEST1)
10
&gt; print(example.Test.ICONST)
12
</pre>
</div>
<p>Enums within a C struct are in the global namespace and are used as
follows from Lua</p>
<div class="targetlang">
<pre>
&gt; print(example.SUNDAY)
0
&gt; -- See the difference here
&gt; print(example.TEST1)
10
</pre>
</div>
<p><b> Compatibility Note:</b> Versions of SWIG prior to SWIG-3.0.0 did
not generate the class table members above. There is no change in the C
wrappers, but the following code was the only way to access these
constants/enums when wrapping C++ member constants:</p>
<div class="targetlang">
<pre>
&gt; print(example.Test_TEST1)
10
&gt; print(example.Test_ICONST)
12
</pre>
</div>
<p> The old-style bindings are still generated in addition to the new
ones. If the <tt>-no-old-metatable-bindings</tt> option is used, then
these old-style bindings are not generated.</p>
<p> It is worth mentioning, that <tt>example.Test.TEST1</tt> and <tt>
example.Test_TEST1</tt> are different entities and changing one does not
change the other. Given the fact that these are constants and they are
not supposed to be changed, it is up to you to avoid such issues.</p>
<h3><a name="Lua_nn12">29.3.5 Pointers</a></h3>
<p> C/C++ pointers are fully supported by SWIG. Furthermore, SWIG has no
problem working with incomplete type information. Given a wrapping of
the &lt;file.h&gt; interface:</p>
<div class="code">
<pre>%module example
FILE *fopen(const char *filename, const char *mode);
int fputs(const char *, FILE *);
int fclose(FILE *);
</pre>
</div>
<p> When wrapped, you will be able to use the functions in a natural way
from Lua. For example:</p>
<div class="targetlang">
<pre>
&gt; f=example.fopen(&quot;junk&quot;, &quot;w&quot;)
&gt; example.fputs(&quot;Hello World&quot;, f)
&gt; example.fclose(f)
</pre>
</div>
<p> Unlike many scripting languages, Lua has had support for pointers to
C/C++ object built in for a long time. They are called 'userdata'.
Unlike many other SWIG versions which use some kind of encoded
character string, all objects will be represented as a userdata. The
SWIG-Lua bindings provides a special function <tt>swig_type()</tt>,
which if given a userdata object will return the type of object pointed
to as a string (assuming it was a SWIG wrapped object).</p>
<div class="targetlang">
<pre>
&gt; print(f)
userdata: 003FDA80
&gt; print(swig_type(f))
FILE * -- it's a FILE*
</pre>
</div>
<p> Lua enforces the integrity of its userdata, so it is virtually
impossible to corrupt the data. But as the user of the pointer, you are
responsible for freeing it, or closing any resources associated with it
(just as you would in a C program). This does not apply so strictly to
classes &amp; structs (see below). One final note: if a function returns a
NULL pointer, this is not encoded as a userdata, but as a Lua nil.</p>
<div class="targetlang">
<pre>
&gt; f=example.fopen(&quot;not there&quot;, &quot;r&quot;) -- this will return a NULL in C
&gt; print(f)
nil
</pre>
</div>
<h3><a name="Lua_structures">29.3.6 Structures</a></h3>
<p> If you wrap a C structure, it is also mapped to a Lua userdata. By
adding a metatable to the userdata, this provides a very natural
interface. For example,</p>
<div class="code">
<pre>struct Point{
int x, y;
};
</pre>
</div>
<p> is used as follows:</p>
<div class="targetlang">
<pre>
&gt; p=example.new_Point()
&gt; p.x=3
&gt; p.y=5
&gt; print(p.x, p.y)
3 5
&gt;
</pre>
</div>
<p> Similar access is provided for unions and the data members of C++
classes.
<br> C structures can be created using a function <tt>new_Point()</tt>,
and both C structures and C++ classes can be created using just the
name <tt>Point()</tt>.</p>
<p> If you print out the value of p in the above example, you will see
something like this:</p>
<div class="targetlang">
<pre>
&gt; print(p)
userdata: 003FA320
</pre>
</div>
<p> Like the pointer in the previous section, this is held as a
userdata. However, additional features have been added to make this
more usable. SWIG effectively creates some accessor/mutator functions
to get and set the data. These functions will be added to the
userdata's metatable. This provides the natural access to the member
variables that were shown above (see end of the document for full
details).</p>
<p> <tt>const</tt> members of a structure are read-only. Data members
can also be forced to be read-only using the immutable directive. As
with other immutables, setting attempts will be cause an error. For
example:</p>
<div class="code">
<pre>struct Foo {
...
%immutable;
int x; // Read-only members
char *name;
%mutable;
...
};
</pre>
</div>
<p> The mechanism for managing char* members as well as array members is
similar to other languages. It is somewhat cumbersome and should
probably be better handled by defining of typemaps (described later).</p>
<p> When a member of a structure is itself a structure, it is handled as
a pointer. For example, suppose you have two structures like this:</p>
<div class="code">
<pre>struct Foo {
int a;
};
struct Bar {
Foo f;
};
</pre>
</div>
<p> Now, suppose that you access the f attribute of Bar like this:</p>
<div class="targetlang">
<pre>
&gt; b = Bar()
&gt; x = b.f
</pre>
</div>
<p> In this case, x is a pointer that points to the Foo that is inside
b. This is the same value as generated by this C code:</p>
<div class="code">
<pre>
Bar b;
Foo *x = &amp;b-&gt;f; // Points inside b
</pre>
</div>
<p> Because the pointer points inside the structure, you can modify the
contents and everything works just like you would expect. For example:</p>
<div class="targetlang">
<pre>
&gt; b = Bar()
&gt; b.f.a = 3 -- Modify attribute of structure member
&gt; x = b.f
&gt; x.a = 3 -- Modifies the same structure
</pre>
</div>
<p> For eLua with the <tt>-eluac</tt> option, structure manipulation has
to be performed with specific structure functions generated by SWIG.
Let's say you have the following structure definition:</p>
<div class="code">
<pre>struct data {
int x, y;
double z;
};
&gt; --From eLua
&gt; a = example.new_data()
&gt; example.data_x_set(a, 10)
&gt; example.data_y_set(a, 20)
&gt; print(example.data_x_get(a), example.data_y_get(a))
10 20
</pre>
</div>
<p> In general, functions of the form <tt>&quot;new_struct()&quot;</tt>, <tt>
&quot;struct_member_get()&quot;</tt>, <tt>&quot;struct_member_set()&quot;</tt> and <tt>
&quot;free_struct()&quot;</tt> are automatically generated by SWIG for each
structure defined in C. (Please note: This doesn't apply for modules
generated with the <tt>-elua</tt> option)</p>
<h3><a name="Lua_nn14">29.3.7 C++ classes</a></h3>
<p> C++ classes are wrapped by a Lua userdata as well. For example, if
you have this class,</p>
<div class="code">
<pre>class List {
public:
List();
~List();
int search(char *item);
void insert(char *item);
void remove(char *item);
char *get(int n);
int length;
};
</pre>
</div>
<p> you can use it in Lua like this:</p>
<div class="targetlang">
<pre>
&gt; l = example.List()
&gt; l:insert(&quot;Ale&quot;)
&gt; l:insert(&quot;Stout&quot;)
&gt; l:insert(&quot;Lager&quot;)
&gt; print(l:get(1))
Stout
&gt; print(l:length)
3
&gt;
</pre>
</div>
<p> (Note: for calling methods of a class, you use <tt>
class:method(args)</tt>, not <tt>class.method(args)</tt>, it's an easy
mistake to make. However for data attributes it is <tt>class.attribute</tt>
)</p>
<p> Class data members are accessed in the same manner as C structures.
Static class members present a special problem for Lua, as Lua doesn't
have support for such features. Therefore, SWIG generates wrappers that
try to work around some of these issues. To illustrate, suppose you
have a class like this:</p>
<div class="code">
<pre>
class Spam {
public:
static void foo();
static int bar;
};
</pre>
</div>
<p> In Lua, C++ static members can be accessed as follows:</p>
<div class="targetlang">
<pre>
&gt; example.Spam.foo() -- calling Spam::foo()
&gt; a=example.Spam.bar -- reading Spam::bar
&gt; example.Spam.bar=b -- writing to Spam::bar
</pre>
</div>
<p> It is not (currently) possible to access static members of an
instance:</p>
<div class="targetlang">
<pre>
&gt; s=example.Spam() -- s is a Spam instance
&gt; s.foo() -- Spam::foo() via an instance
-- does NOT work
</pre>
</div>
<p><b> Compatibility Note:</b> In versions prior to SWIG-3.0.0 only the
following names would work:</p>
<div class="targetlang">
<pre>
&gt; example.Spam_foo() -- calling Spam::foo()
&gt; a=example.Spam_bar -- reading Spam::bar
&gt; example.Spam_bar=b -- writing to Spam::bar
</pre>
</div>
<p> Both style names are generated by default now. However, if the <tt>
-no-old-metatable-bindings</tt> option is used, then the backward
compatible names are not generated in addition to ordinary ones.</p>
<h3><a name="Lua_nn15">29.3.8 C++ inheritance</a></h3>
<p> SWIG is fully aware of issues related to C++ inheritance. Therefore,
if you have classes like this</p>
<div class="code">
<pre>class Foo {
...
};
class Bar : public Foo {
...
};
</pre>
</div>
<p> And if you have functions like this</p>
<div class="code">
<pre>void spam(Foo *f);
</pre>
</div>
<p> then the function <tt>spam()</tt> accepts a Foo pointer or a pointer
to any class derived from Foo.</p>
<p> It is safe to use multiple inheritance with SWIG.</p>
<h3><a name="Lua_nn16">29.3.9 Pointers, references, values, and arrays</a>
</h3>
<p> In C++, there are many different ways a function might receive and
manipulate objects. For example:</p>
<div class="code">
<pre>void spam1(Foo *x); // Pass by pointer
void spam2(Foo &amp;x); // Pass by reference
void spam3(Foo x); // Pass by value
void spam4(Foo x[]); // Array of objects
</pre>
</div>
<p> In SWIG, there is no detailed distinction like this--specifically,
there are only &quot;objects&quot;. There are no pointers, references, arrays,
and so forth. Because of this, SWIG unifies all of these types together
in the wrapper code. For instance, if you actually had the above
functions, it is perfectly legal to do this:</p>
<div class="targetlang">
<pre>
&gt; f = Foo() -- Create a Foo
&gt; spam1(f) -- Ok. Pointer
&gt; spam2(f) -- Ok. Reference
&gt; spam3(f) -- Ok. Value.
&gt; spam4(f) -- Ok. Array (1 element)
</pre>
</div>
<p> Similar behaviour occurs for return values. For example, if you had
functions like this,</p>
<div class="code">
<pre>Foo *spam5();
Foo &amp;spam6();
Foo spam7();
</pre>
</div>
<p> then all three functions will return a pointer to some Foo object.
Since the third function (spam7) returns a value, newly allocated
memory is used to hold the result and a pointer is returned (Lua will
release this memory when the return value is garbage collected). The
other two are pointers which are assumed to be managed by the C code
and so will not be garbage collected.</p>
<h3><a name="Lua_nn17">29.3.10 C++ overloaded functions</a></h3>
<p> C++ overloaded functions, methods, and constructors are mostly
supported by SWIG. For example, if you have two functions like this:</p>
<div class="code">
<pre>void foo(int);
void foo(char *c);
</pre>
</div>
<p> You can use them in Lua in a straightforward manner:</p>
<div class="targetlang">
<pre>
&gt; foo(3) -- foo(int)
&gt; foo(&quot;Hello&quot;) -- foo(char *c)
</pre>
</div>
<p> However due to Lua's coercion mechanism is can sometimes do strange
things.</p>
<div class="targetlang">
<pre>
&gt; foo(&quot;3&quot;) -- &quot;3&quot; can be coerced into an int, so it calls foo(int)!
</pre>
</div>
<p> As this coercion mechanism is an integral part of Lua, there is no
easy way to get around this other than renaming of functions (see
below).</p>
<p> Similarly, if you have a class like this,</p>
<div class="code">
<pre>class Foo {
public:
Foo();
Foo(const Foo &amp;);
...
};
</pre>
</div>
<p> you can write Lua code like this:</p>
<div class="targetlang">
<pre>
&gt; f = Foo() -- Create a Foo
&gt; g = Foo(f) -- Copy f
</pre>
</div>
<p> Overloading support is not quite as flexible as in C++. Sometimes
there are methods that SWIG can't disambiguate. For example:</p>
<div class="code">
<pre>void spam(int);
void spam(short);
</pre>
</div>
<p> or</p>
<div class="code">
<pre>void foo(Bar *b);
void foo(Bar &amp;b);
</pre>
</div>
<p> If declarations such as these appear, you will get a warning message
like this:</p>
<div class="shell">
<pre>
example.i:12: Warning 509: Overloaded method spam(short) effectively ignored,
example.i:11: Warning 509: as it is shadowed by spam(int).
</pre>
</div>
<p> To fix this, you either need to ignore or rename one of the methods.
For example:</p>
<div class="code">
<pre>%rename(spam_short) spam(short);
...
void spam(int);
void spam(short); // Accessed as spam_short
</pre>
</div>
<p> or</p>
<div class="code">
<pre>%ignore spam(short);
...
void spam(int);
void spam(short); // Ignored
</pre>
</div>
<p> SWIG resolves overloaded functions and methods using a
disambiguation scheme that ranks and sorts declarations according to a
set of type-precedence rules. The order in which declarations appear in
the input does not matter except in situations where ambiguity
arises--in this case, the first declaration takes precedence.</p>
<p> Please refer to the &quot;SWIG and C++&quot; chapter for more information
about overloading.</p>
<p> Dealing with the Lua coercion mechanism, the priority is roughly
(integers, floats, strings, userdata). But it is better to rename the
functions rather than rely upon the ordering.</p>
<h3><a name="Lua_nn18">29.3.11 C++ operators</a></h3>
<p> Certain C++ overloaded operators can be handled automatically by
SWIG. For example, consider a class like this:</p>
<div class="code">
<pre>class Complex {
private:
double rpart, ipart;
public:
Complex(double r = 0, double i = 0) : rpart(r), ipart(i) { }
Complex(const Complex &amp;c) : rpart(c.rpart), ipart(c.ipart) { }
Complex &amp;operator=(const Complex &amp;c);
Complex operator+(const Complex &amp;c) const;
Complex operator-(const Complex &amp;c) const;
Complex operator*(const Complex &amp;c) const;
Complex operator-() const;
double re() const { return rpart; }
double im() const { return ipart; }
};
</pre>
</div>
<p> When wrapped, it works like you expect:</p>
<div class="targetlang">
<pre>
&gt; c = Complex(3, 4)
&gt; d = Complex(7, 8)
&gt; e = c + d
&gt; e:re()
10.0
&gt; e:im()
12.0
</pre>
</div>
<p> One restriction with operator overloading support is that SWIG is
not able to fully handle operators that aren't defined as part of the
class. For example, if you had code like this</p>
<div class="code">
<pre>
class Complex {
...
friend Complex operator+(double, const Complex &amp;c);
...
};
</pre>
</div>
<p> then SWIG doesn't know what to do with the friend function--in fact,
it simply ignores it and issues a warning. You can still wrap the
operator, but you may have to encapsulate it in a special function. For
example:</p>
<div class="code">
<pre>
%rename(Complex_add_dc) operator+(double, const Complex &amp;);
...
Complex operator+(double, const Complex &amp;c);
</pre>
</div>
<p> There are ways to make this operator appear as part of the class
using the <tt>%extend</tt> directive. Keep reading.</p>
<p> Also, be aware that certain operators don't map cleanly to Lua, and
some Lua operators don't map cleanly to C++ operators. For instance,
overloaded assignment operators don't map to Lua semantics and will be
ignored, and C++ doesn't support Lua's concatenation operator (<tt>..</tt>
).</p>
<p> In order to keep maximum compatibility within the different
languages in SWIG, the Lua bindings uses the same set of operator names
as Python. Although internally it renames the functions to something
else (on order to work with Lua).</p>
<p> The current list of operators which can be overloaded (and the
alternative function names) are:</p>
<ul>
<li><tt>__add__</tt> operator+</li>
<li><tt>__sub__</tt> operator-</li>
<li><tt>__mul__</tt> operator *</li>
<li><tt>__div__</tt> operator/</li>
<li><tt>__unm__</tt> unary minus</li>
<li><tt>__call__</tt> operator<tt>()</tt> (often used in functor
classes)</li>
<li><tt>__pow__</tt> the exponential fn (no C++ equivalent, Lua uses <tt>
^</tt>)</li>
<li><tt>__concat__</tt> the concatenation operator (Lua's <tt>..</tt>)</li>
<li><tt>__eq__</tt> operator<tt>==</tt></li>
<li><tt>__lt__</tt> operator<tt>&lt;</tt></li>
<li><tt>__le__</tt> operator<tt>&lt;=</tt></li>
</ul>
<p> Note: in Lua, only the equals, less than, and less than equals
operators are defined. The other operators (!=, &gt;, &gt;=) are achieved by
using a logical not applied to the results of other operators.</p>
<p> The following operators cannot be overloaded (mainly because they
are not supported in Lua)</p>
<ul>
<li>++ and --</li>
<li>+=, -=, *= etc</li>
<li>% operator (you have to use math.mod)</li>
<li>assignment operator</li>
<li>all bitwise/logical operations</li>
</ul>
<p> SWIG also accepts the <tt>__str__()</tt> member function which
converts an object to a string. This function should return a const
char*, preferably to static memory. This will be used for the <tt>
print()</tt> and <tt>tostring()</tt> functions in Lua. Assuming the
complex class has a function</p>
<div class="code">
<pre>const char* __str__() {
static char buffer[255];
sprintf(buffer, &quot;Complex(%g, %g)&quot;, this-&gt;re(), this-&gt;im());
return buffer;
}
</pre>
</div>
<p> Then this will support the following code in Lua</p>
<div class="targetlang">
<pre>
&gt; c = Complex(3, 4)
&gt; d = Complex(7, 8)
&gt; e = c + d
&gt; print(e)
Complex(10, 12)
&gt; s=tostring(e) -- s is the number in string form
&gt; print(s)
Complex(10, 12)
</pre>
</div>
<p> It is also possible to overload the operator<tt>[]</tt>, but
currently this cannot be automatically performed. To overload the
operator<tt>[]</tt> you need to provide two functions, <tt>
__getitem__()</tt> and <tt>__setitem__()</tt></p>
<div class="code">
<pre>class Complex {
//....
double __getitem__(int i)const; // i is the index, returns the data
void __setitem__(int i, double d); // i is the index, d is the data
};
</pre>
</div>
<p> C++ operators are mapped to Lua predefined metafunctions. Class
inherits from its bases the following list of metafunctions ( thus
inheriting the folloging operators and pseudo-operators):</p>
<ul>
<li><tt>__add__</tt></li>
<li><tt>__sub__</tt></li>
<li><tt>__mul__</tt></li>
<li><tt>__div__</tt></li>
<li><tt>__unm__</tt></li>
<li><tt>__mod__</tt></li>
<li><tt>__call__</tt></li>
<li><tt>__pow__</tt></li>
<li><tt>__concat__</tt></li>
<li><tt>__eq__</tt></li>
<li><tt>__lt__</tt></li>
<li><tt>__le__</tt></li>
<li><tt>__len__</tt></li>
<li><tt>__getitem__</tt></li>
<li><tt>__setitem__</tt></li>
<li><tt>__tostring</tt> used internally by Lua for tostring() function.
__str__ is mapped to this function</li>
</ul>
<p>No other lua metafunction is inherited. For example, __gc is not
inherited and must be redefined in every class. <tt>__tostring</tt> is
subject to a special handling. If absent in class and in class bases, a
default one will be provided by SWIG.</p>
<h3><a name="Lua_nn19">29.3.12 Class extension with %extend</a></h3>
<p> One of the more interesting features of SWIG is that it can extend
structures and classes with new methods. In the previous section, the
Complex class would have benefited greatly from an __str__() method as
well as some repairs to the operator overloading. It can also be used
to add additional functions to the class if they are needed.</p>
<p> Take the original Complex class</p>
<div class="code">
<pre>class Complex {
private:
double rpart, ipart;
public:
Complex(double r = 0, double i = 0) : rpart(r), ipart(i) { }
Complex(const Complex &amp;c) : rpart(c.rpart), ipart(c.ipart) { }
Complex &amp;operator=(const Complex &amp;c);
Complex operator+(const Complex &amp;c) const;
Complex operator-(const Complex &amp;c) const;
Complex operator*(const Complex &amp;c) const;
Complex operator-() const;
double re() const { return rpart; }
double im() const { return ipart; }
};
</pre>
</div>
<p> Now we extend it with some new code</p>
<div class="code">
<pre>%extend Complex {
const char *__str__() {
static char tmp[1024];
sprintf(tmp, &quot;Complex(%g, %g)&quot;, $self-&gt;re(), $self-&gt;im());
return tmp;
}
bool operator==(const Complex&amp; c) {
return ($self-&gt;re()==c.re() &amp;&amp; $self-&gt;im()==c.im());
}
};
</pre>
</div>
<p> Now, in Lua</p>
<div class="targetlang">
<pre>
&gt; c = Complex(3, 4)
&gt; d = Complex(7, 8)
&gt; e = c + d
&gt; print(e) -- print uses __str__ to get the string form to print
Complex(10, 12)
&gt; print(e==Complex(10, 12)) -- testing the == operator
true
&gt; print(e!=Complex(12, 12)) -- the != uses the == operator
true
</pre>
</div>
<p> Extend works with both C and C++ code, on classes and structs. It
does not modify the underlying object in any way---the extensions only
show up in the Lua interface. The only item to take note of is the code
has to use the '$self' instead of 'this', and that you cannot access
protected/private members of the code (as you are not officially part
of the class).</p>
<h3><a name="Lua_nn20">29.3.13 Using %newobject to release memory</a></h3>
<p> If you have a function that allocates memory like this,</p>
<div class="code">
<pre>char *foo() {
char *result = (char *) malloc(...);
...
return result;
}
</pre>
</div>
<p> then the SWIG generated wrappers will have a memory leak--the
returned data will be copied into a string object and the old contents
ignored.</p>
<p> To fix the memory leak, use the <a href="#Customization_ownership">
%newobject directive</a>.</p>
<div class="code">
<pre>%newobject foo;
...
char *foo();
</pre>
</div>
<p> This will release the allocated memory.</p>
<h3><a name="Lua_nn21">29.3.14 C++ templates</a></h3>
<p> C++ templates don't present a huge problem for SWIG. However, in
order to create wrappers, you have to tell SWIG to create wrappers for
a particular template instantiation. To do this, you use the template
directive. For example:</p>
<div class="code">
<pre>%module example
%{
#include &quot;pair.h&quot;
%}
template&lt;class T1, class T2&gt;
struct pair {
typedef T1 first_type;
typedef T2 second_type;
T1 first;
T2 second;
pair();
pair(const T1&amp;, const T2&amp;);
~pair();
};
%template(pairii) pair&lt;int, int&gt;;
</pre>
</div>
<p> In Lua:</p>
<div class="targetlang">
<pre>
&gt; p = example.pairii(3, 4)
&gt; print(p.first, p.second)
3 4
</pre>
</div>
<p> Obviously, there is more to template wrapping than shown in this
example. More details can be found in the SWIG and C++ chapter. Some
more complicated examples will appear later.</p>
<h3><a name="Lua_nn22">29.3.15 C++ Smart Pointers</a></h3>
<p> In certain C++ programs, it is common to use classes that have been
wrapped by so-called &quot;smart pointers.&quot; Generally, this involves the use
of a template class that implements operator-&gt;() like this:</p>
<div class="code">
<pre>template&lt;class T&gt; class SmartPtr {
...
T *operator-&gt;();
...
}
</pre>
</div>
<p> Then, if you have a class like this,</p>
<div class="code">
<pre>class Foo {
public:
int x;
int bar();
};
</pre>
</div>
<p> A smart pointer would be used in C++ as follows:</p>
<div class="code">
<pre>SmartPtr&lt;Foo&gt; p = CreateFoo(); // Created somehow (not shown)
...
p-&gt;x = 3; // Foo::x
int y = p-&gt;bar(); // Foo::bar
</pre>
</div>
<p> To wrap this, simply tell SWIG about the SmartPtr class and the
low-level Foo object. Make sure you instantiate SmartPtr using template
if necessary. For example:</p>
<div class="code">
<pre>%module example
...
%template(SmartPtrFoo) SmartPtr&lt;Foo&gt;;
...
</pre>
</div>
<p> Now, in Lua, everything should just &quot;work&quot;:</p>
<div class="targetlang">
<pre>
&gt; p = example.CreateFoo() -- Create a smart-pointer somehow
&gt; p.x = 3 -- Foo::x
&gt; print(p:bar()) -- Foo::bar
</pre>
</div>
<p> If you ever need to access the underlying pointer returned by <tt>
operator-&gt;()</tt> itself, simply use the <tt>__deref__()</tt> method.
For example:</p>
<div class="targetlang">
<pre>
&gt; f = p:__deref__() -- Returns underlying Foo *
</pre>
</div>
<h3><a name="Lua_nn23">29.3.16 C++ Exceptions</a></h3>
<p> Lua does not natively support exceptions, but it has errors which
are similar. When a Lua function terminates with an error it returns
one value back to the caller. SWIG automatically maps any basic type
which is thrown into a Lua error. Therefore for a function:</p>
<div class="code">
<pre>
int message() throw(const char *) {
throw(&quot;I died.&quot;);
return 1;
}
</pre>
</div>
<p> SWIG will automatically convert this to a Lua error.</p>
<div class="targetlang">
<pre>
&gt; message()
I died.
stack traceback:
[C]: in function 'message'
stdin:1: in main chunk
[C]: ?
&gt;
</pre>
</div>
<p> If you want to catch an exception, you must use either pcall() or
xpcall(), which are documented in the Lua manual. Using xpcall will
allow you to obtain additional debug information (such as a
stacktrace).</p>
<div class="targetlang">
<pre>
&gt; function a() b() end -- function a() calls function b()
&gt; function b() message() end -- function b() calls C++ function message(), which throws
&gt; ok, res=pcall(a) -- call the function
&gt; print(ok, res)
false I died.
&gt; ok, res=xpcall(a, debug.traceback) -- call the function
&gt; print(ok, res)
false I died.
stack traceback:
[C]: in function 'message'
runme.lua:70: in function 'b'
runme.lua:67: in function &lt;runme.lua:66&gt;
[C]: in function 'xpcall'
runme.lua:95: in main chunk
[C]: ?
</pre>
</div>
<p> SWIG is able to throw numeric types, enums, chars, char*'s and
std::string's without problem. It has also written typemaps for
std::exception and its derived classes, which convert the exception
into an error string.</p>
<p> However it's not so simple to throw other types of objects. Thrown
objects are not valid outside the 'catch' block. Therefore they cannot
be returned to the interpreter. The obvious ways to overcome this would
be to either return a copy of the object, or to convert the object to a
string and return that. Though it seems obvious to perform the former,
in some cases this is not possible, most notably when SWIG has no
information about the object, or the object is not copyable/creatable.</p>
<p> Therefore by default SWIG converts all thrown object into strings
and returns them. So given a function:</p>
<div class="code">
<pre>
void throw_A() throw(A*) {
throw new A();
}
</pre>
</div>
<p> SWIG will just convert it (poorly) to a string and use that as its
error. (This is not that useful, but it always works).</p>
<div class="targetlang">
<pre>
&gt; throw_A()
object exception:A *
stack traceback:
[C]: in function 'unknown'
stdin:1: in main chunk
[C]: ?
&gt;
</pre>
</div>
<p> To get a more useful behaviour out of SWIG you must either: provide
a way to convert your exceptions into strings, or throw objects which
can be copied.</p>
<p> If you have your own class which you want output as a string you
will need to add a typemap something like this:</p>
<div class="code">
<pre>
%typemap(throws) my_except
%{
lua_pushstring(L, $1.what()); // assuming my_except::what() returns a const char* message
SWIG_fail; // trigger the error handler
%}
</pre>
</div>
<p> If you wish your exception to be returned to the interpreter, it
must firstly be copyable. Then you must have an additional <tt>%apply</tt>
statement, to tell SWIG to return a copy of this object to the
interpreter. For example:</p>
<div class="code">
<pre>
%apply SWIGTYPE EXCEPTION_BY_VAL {Exc}; // tell SWIG to return Exc by value to interpreter
class Exc {
public:
Exc(int c, const char *m) {
code = c;
strncpy(msg, m, 256);
}
int code;
char msg[256];
};
void throw_exc() throw(Exc) {
throw(Exc(42, &quot;Hosed&quot;));
}
</pre>
</div>
<p> Then the following code can be used (note: we use pcall to catch the
error so we can process the exception).</p>
<div class="targetlang">
<pre>
&gt; ok, res=pcall(throw_exc)
&gt; print(ok)
false
&gt; print(res)
userdata: 0003D880
&gt; print(res.code, res.msg)
42 Hosed
&gt;
</pre>
</div>
<p> Note: it is also possible (though tedious) to have a function throw
several different kinds of exceptions. To process this will require a
pcall, followed by a set of if statements checking the type of the
error.</p>
<p> All of this code assumes that your C++ code uses exception
specification (which a lot doesn't). If it doesn't consult the &quot;<a href="#SWIGPlus_catches">
Exception handling with %catches</a>&quot; section and the &quot;<a href="#Customization_exception">
Exception handling with %exception</a>&quot; section, for more details on how
to add exception specification to functions or globally (respectively).</p>
<h3><a name="Lua_namespaces">29.3.17 Namespaces</a></h3>
<p> C++ namespaces are supported via the %nspace feature.</p>
<p> Namespaces are mapped into Lua tables. Each of those tables contains
names that were defined within an appropriate namespace. Namespace
hierarchies (a.k.a nested namespaces) are preserved. Consider the
following C++ code:</p>
<div class="code">
<pre>
%module example
%nspace MyWorld::World;
%nspace MyWorld::Nested::Dweller;
int module_function() { return 7; }
int module_variable = 9;
namespace MyWorld {
class World {
public:
World() : world_max_count(9) {}
int create_world() { return 17; }
const int world_max_count; // = 9
};
namespace Nested {
class Dweller {
public:
enum Gender { MALE = 0, FEMALE = 1 };
static int count() { return 19; }
};
}
}
</pre>
</div>
<p> Now, from Lua usage is as follows:</p>
<div class="targetlang">
<pre>
&gt; print(example.module_function())
7.0
&gt; print(example.module_variable)
9.0
&gt; print(example.MyWorld.World():create_world())
17.0
&gt; print(example.MyWorld.World().world_max_count)
9.0
&gt; print(example.MyWorld.Nested.Dweller.MALE)
0
&gt; print(example.MyWorld.Nested.Dweller.count())
19.0
</pre>
</div>
<p> The hierarchies in Lua are the same as the C++ hierarchies when
using <tt>%nspace</tt>, however, we could instead completely change
these hierarchies with <tt>%nspacemove</tt>. Consider the code above
with the two <tt>%nspace</tt> lines of code removed and replaced with
the following:</p>
<div class="code">
<pre>
%nspacemove(DifferentWorld::SubSpace1::SubSpace2) MyWorld::World;
%nspacemove(DifferentWorld) MyWorld::Nested::Dweller;
</pre>
</div>
<p> The Lua code uses the completely modified hierarchies:</p>
<div class="targetlang">
<pre>
&gt; print(example.module_function())
7.0
&gt; print(example.module_variable)
9.0
&gt; print(example.DifferentWorld.SubSpace1.SubSpace2.World():create_world())
17.0
&gt; print(example.DifferentWorld.SubSpace1.SubSpace2.World().world_max_count)
9.0
&gt; print(example.DifferentWorld.Dweller.MALE)
0
&gt; print(example.DifferentWorld.Dweller.count())
19.0
</pre>
</div>
<h4><a name="Lua_nn27">29.3.17.1 Compatibility Note</a></h4>
<p> The nspace feature was first supported by the Lua module in
SWIG-3.0.0</p>
<p> If SWIG is running in a backwards compatible way, i.e. without the <tt>
-no-old-metatable-bindings</tt> option, then additional old-style names
are generated (notice the underscore):</p>
<div class="targetlang">
<pre>
&gt; print(example.MyWorld.Nested.Dweller_MALE)
0
&gt; print(example.MyWorld.Nested.Dweller_count())
19.0
</pre>
</div>
<p> The ability to move symbols into different namespaces via <tt>
%nspacemove</tt> was introduced in SWIG-4.3.0.</p>
<h4><a name="Lua_nn29">29.3.17.2 Names</a></h4>
<p> If SWIG is launched without <tt>-no-old-metatable-bindings</tt>
option, then it enters backward-compatible mode. While in this mode, it
tries to generate additional names for static functions, class static
constants and class enums. Those names are in a form <tt>
$classname_$symbolname</tt> and are added to the scope surrounding the
class. If %nspace is enabled, then class namespace is taken as scope.
If there is no namespace, or %nspace is disabled, then module is
considered a class namespace.</p>
<p> Consider the following C++ code</p>
<div class="code">
<pre>%module example
%nspace MyWorld::Test;
namespace MyWorld {
class Test {
public:
enum { TEST1 = 10, TEST2 }
static const int ICONST = 12;
};
class Test2 {
public:
enum { TEST3 = 20, TEST4 }
static const int ICONST2 = 23;
}
</pre>
</div>
<p> When in backward compatible mode, in addition to the usual names,
the following ones will be generated (notice the underscore):</p>
<div class="targetlang">
<pre>
9
&gt; print(example.MyWorld.Test_TEST1) -- Test has %nspace enabled
10
&gt; print(example.MyWorld.Test_ICONST) -- Test has %nspace enabled
12
&gt; print(example.Test2_TEST3) -- Test2 doesn't have %nspace enabled
20
&gt; print(example.Test2_ICONST2) -- Test2 doesn't have %nspace enabled
23
&gt;
</pre>
</div>
<p> There is a slight difference with enums when in C mode. As per C
standard, enums from C structures are exported to surrounding scope
without any prefixing. Pretending that Test2 is a struct, not class,
that would be:</p>
<div class="targetlang">
<pre>
&gt; print(example.TEST3) -- NOT Test2_TEST3
20
&gt;
</pre>
</div>
<h4><a name="Lua_nn30">29.3.17.3 Inheritance</a></h4>
<p> The internal organization of inheritance has changed. Consider the
following C++ code:</p>
<div class="code">
<pre>%module example
class Base {
public:
int base_func()
};
class Derived : public Base {
public:
int derived_func()
}
</pre>
</div>
<p>Lets assume for a moment that class member functions are stored in <tt>
.fn</tt> table. Previously, when classes were exported to Lua during
module initialization, for every derived class all service tables <tt>
ST(i.e. &quot;.fn&quot;)</tt> were squashed and added to corresponding derived
class <tt>ST</tt>: Everything from <tt>.fn</tt> table of class Base was
copied to <tt>.fn</tt> table of class Derived and so on. This was a
recursive procedure, so in the end the whole inheritance tree of
derived class was squashed into derived class.</p>
<p> That means that any changes done to class Base after module
initialization wouldn't affect class Derived:</p>
<div class="targetlang">
<pre>
base = example.Base()
der = example.Derived()
&gt; print(base.base_func)
function: 0x1367940
&gt; getmetatable(base)[&quot;.fn&quot;].new_func = function (x) return x -- Adding new function to class Base (to class, not to an instance!)
&gt; print(base.new_func) -- Checking this function
function
&gt; print(der.new_func) -- Wouldn't work. Derived doesn't check Base any more.
nil
&gt;
</pre>
</div>
<p> This behaviour was changed. Now unless -squash-bases option is
provided, Derived stores a list of its bases and if some symbol is not
found in its own service tables then its bases are searched for it.
Option -squash-bases will effectively return old behaviour.</p>
<div class="targetlang">
<pre>
&gt; print(der.new_func) -- Now it works
function
&gt;
</pre>
</div>
<h2><a name="Lua_nn24">29.4 Typemaps</a></h2>
<p>This section explains what typemaps are and how to use them. The
default wrapping behaviour of SWIG is enough in most cases. However
sometimes SWIG may need a little additional assistance to know which
typemap to apply to provide the best wrapping. This section will be
explaining how to use typemaps to best effect</p>
<h3><a name="Lua_nn25">29.4.1 What is a typemap?</a></h3>
<p>A typemap is nothing more than a code generation rule that is
attached to a specific C datatype. For example, to convert integers
from Lua to C, you might define a typemap like this:</p>
<div class="code">
<pre>%module example
%typemap(in) int {
$1 = (int) lua_tonumber(L, $input);
printf(&quot;Received an integer : %d\n&quot;, $1);
}
%inline %{
extern int fact(int n);
%}
</pre>
</div>
<p><i>Note: you shouldn't use this typemap, as SWIG already has a
typemap for this task. This is purely for example.</i></p>
<p>Typemaps are always associated with some specific aspect of code
generation. In this case, the &quot;in&quot; method refers to the conversion of
input arguments to C/C++. The datatype int is the datatype to which the
typemap will be applied. The supplied C code is used to convert values.
In this code a number of special variable prefaced by a $ are used. The
$1 variable is placeholder for a local variable of type int. The $input
is the index on the Lua stack for the value to be used.</p>
<p>When this example is compiled into a Lua module, it operates as
follows:</p>
<div class="targetlang">
<pre>&gt; require &quot;example&quot;
&gt; print(example.fact(6))
Received an integer : 6
720
</pre>
</div>
<h3><a name="Lua_nn26">29.4.2 Using typemaps</a></h3>
<p>There are many ready written typemaps built into SWIG for all common
types (int, float, short, long, char*, enum and more), which SWIG uses
automatically, with no effort required on your part.</p>
<p>However for more complex functions which use input/output parameters
or arrays, you will need to make use of &lt;typemaps.i&gt;, which contains
typemaps for these situations. For example, consider these functions:</p>
<div class="code">
<pre>void add(int x, int y, int *result) {
*result = x + y;
}
int sub(int *x1, int *y1) {
return *x1-*y1;
}
void swap(int *sx, int *sy) {
int t=*sx;
*sx=*sy;
*sy=t;
}
</pre>
</div>
<p>It is clear to the programmer, that 'result' is an output parameter,
'x1' and 'y1' are input parameters and 'sx' and 'sy' are input/output
parameters. However is not apparent to SWIG, so SWIG must to informed
about which kind they are, so it can wrapper accordingly.</p>
<p>One means would be to rename the argument name to help SWIG, eg <tt>
void add(int x, int y, int *OUTPUT)</tt>, however it is easier to use
the <tt>%apply</tt> to achieve the same result, as shown below.</p>
<div class="code">
<pre>%include &lt;typemaps.i&gt;
%apply int* OUTPUT {int *result}; // int *result is output
%apply int* INPUT {int *x1, int *y1}; // int *x1 and int *y1 are input
%apply int* INOUT {int *sx, int *sy}; // int *sx and int *sy are input and output
void add(int x, int y, int *result);
int sub(int *x1, int *y1);
void swap(int *sx, int *sy);
</pre>
</div>
<p>When wrapped, it gives the following results:</p>
<div class="targetlang">
<pre>&gt; require &quot;example&quot;
&gt; print(example.add(1, 2))
3
&gt; print(demo.sub(1, 2))
-1
&gt; a, b=1, 2
&gt; c, d=demo.swap(a, b)
&gt; print(a, b, c, d)
1 2 2 1
</pre>
</div>
<p>Notice, that 'result' is not required in the arguments to call the
function, as it an output parameter only. For 'sx' and 'sy' they must
be passed in (as they are input), but the original value is not
modified (Lua does not have a pass by reference feature). The modified
results are then returned as two return values. All INPUT/OUTPUT/INOUT
arguments will behave in a similar manner.</p>
<p>Note: C++ references must be handled exactly the same way. However
SWIG will automatically wrap a <tt>const int&amp;</tt> as an input
parameter (since that it obviously input).</p>
<h3><a name="Lua_typemap_arrays">29.4.3 Typemaps and arrays</a></h3>
<p>Arrays present a challenge for SWIG, because like pointers SWIG does
not know whether these are input or output values, nor does SWIG have
any indication of how large an array should be. However with the proper
guidance SWIG can easily wrapper arrays for convenient usage.</p>
<p>Given the functions:</p>
<div class="code">
<pre>extern void sort_int(int* arr, size_t len);
extern void sort_double(double* arr, size_t len);
</pre>
</div>
<p>There are basically two ways that SWIG can deal with this. The first
way, uses the <tt>&lt;carrays.i&gt;</tt> library to create an array in C/C++
then this can be filled within Lua and passed into the function. It
works, but it's a bit tedious. More details can be found in the <a href="#Library_carrays">
carrays.i</a> documentation.</p>
<p>The second and more intuitive way, would be to pass a Lua table
directly into the function, and have SWIG automatically convert between
Lua-table and C-array. Within the <tt>&lt;typemaps.i&gt;</tt> file there are
typemaps ready written to perform this task. To use them is again a
matter of using %apply in the correct manner.</p>
<p>The wrapper file below, shows both the use of carrays as well as the
use of the typemap to wrap arrays.</p>
<div class="code">
<pre>// using the C-array
%include &lt;carrays.i&gt;
// this declares a batch of functions for manipulating C integer arrays
%array_functions(int, int)
extern void sort_int(int* arr, size_t len); // the function to wrap
// using typemaps
%include &lt;typemaps.i&gt;
%apply (double *INOUT, int) {(double* arr, size_t len)};
extern void sort_double(double* arr, size_t len); // the function to wrap
</pre>
</div>
<p>Once wrapped, the functions can both be called, though with different
ease of use:</p>
<div class="targetlang">
<pre>require &quot;example&quot;
ARRAY_SIZE=10
-- passing a C array to the sort_int()
arr=example.new_int(ARRAY_SIZE) -- create the array
for i=0, ARRAY_SIZE-1 do -- index 0..9 (just like C)
example.int_setitem(arr, i, math.random(1000))
end
example.sort_int(arr, ARRAY_SIZE) -- call the function
example.delete_int(arr) -- must delete the allocated memory
-- use a typemap to call with a Lua-table
-- one item of note: the typemap creates a copy, rather than edit-in-place
t={} -- a Lua table
for i=1, ARRAY_SIZE do -- index 1..10 (Lua style)
t[i]=math.random(1000)/10
end
t=example.sort_double(t) -- replace t with the result
</pre>
</div>
<p>Obviously the first version could be made less tedious by writing a
Lua function to perform the conversion from a table to a C-array. The <tt>
%luacode</tt> directive is good for this. See SWIG\Examples\lua\arrays
for an example of this.</p>
<p><b>Warning:</b> in C indexes start at ZERO, in Lua indexes start at
ONE. SWIG expects C-arrays to be filled for 0..N-1 and Lua tables to be
1..N, (the indexing follows the norm for the language). In the typemap
when it converts the table to an array it quietly changes the indexing
accordingly. Take note of this behaviour if you have a C function which
returns indexes.</p>
<p>Note: SWIG also can support arrays of pointers in a similar manner.</p>
<h3><a name="Lua_typemaps_ptr_ptr_functions">29.4.4 Typemaps and
pointer-pointer functions</a></h3>
<p>Several C++ libraries use a pointer-pointer functions to create its
objects. These functions require a pointer to a pointer which is then
filled with the pointer to the new object. Microsoft's COM and DirectX
as well as many other libraries have this kind of function. An example
is given below:</p>
<div class="code">
<pre>struct iMath; // some structure
int Create_Math(iMath** pptr); // its creator (assume it mallocs)
</pre>
</div>
<p>Which would be used with the following C code:</p>
<div class="code">
<pre>iMath* ptr;
int ok;
ok=Create_Math(&amp;ptr);
// do things with ptr
//...
free(ptr); // dispose of iMath
</pre>
</div>
<p>SWIG has a ready written typemap to deal with such a kind of function
in &lt;typemaps.i&gt;. It provides the correct wrapping as well as setting
the flag to inform Lua that the object in question should be garbage
collected. Therefore the code is simply:</p>
<div class="code">
<pre>%include &lt;typemaps.i&gt;
%apply SWIGTYPE** OUTPUT{iMath **pptr }; // tell SWIG it's an output
struct iMath; // some structure
int Create_Math(iMath** pptr); // its creator (assume it mallocs)
</pre>
</div>
<p>The usage is as follows:</p>
<div class="targetlang">
<pre>ok, ptr=Create_Math() -- ptr is an iMath* which is returned with the int (ok)
ptr=nil -- the iMath* will be GC'ed as normal
</pre>
</div>
<h2><a name="Lua_writing_typemaps">29.5 Writing typemaps</a></h2>
<p>This section describes how you can modify SWIG's default wrapping
behavior for various C/C++ datatypes using the <tt>%typemap</tt>
directive. This is an advanced topic that assumes familiarity with the
Lua C API as well as the material in the &quot;<a href="#Typemaps">Typemaps</a>
&quot; chapter.</p>
<p>Before proceeding, it should be stressed that writing typemaps is
rarely needed unless you want to change some aspect of the wrapping, or
to achieve an effect which in not available with the default bindings.</p>
<p>Before proceeding, you should read the previous section on using
typemaps, and look at the existing typemaps found in luatypemaps.swg
and typemaps.i. These are both well documented and fairly easy to read.
You should not attempt to write your own typemaps until you have read
and can understand both of these files (they may well also give you an
idea to base your work on).</p>
<h3><a name="Lua_typemaps_write">29.5.1 Typemaps you can write</a></h3>
<p>There are many different types of typemap that can be written, the
full list can be found in the &quot;<a href="#Typemaps">Typemaps</a>&quot;
chapter. However the following are the most commonly used ones.</p>
<ul>
<li><tt>in</tt> this is for input arguments to functions</li>
<li><tt>out</tt> this is for return types from functions</li>
<li><tt>argout</tt> this is for a function argument which is actually
returning something</li>
<li><tt>typecheck</tt> this is used to determine which overloaded
function should be called (the syntax for the typecheck is different
from the typemap, see typemaps for details).</li>
</ul>
<h3><a name="Lua_nn31">29.5.2 SWIG's Lua-C API</a></h3>
<p>This section explains the SWIG specific Lua-C API. It does not cover
the main Lua-C api, as this is well documented and not worth covering.</p>
<p><tt>int SWIG_ConvertPtr(lua_State* L, int index, void** ptr,
swig_type_info *type, int flags);</tt></p>
<div class="indent"> This is the standard function used for converting a
Lua userdata to a void*. It takes the value at the given index in the
Lua state and converts it to a userdata. It will then provide the
necessary type checks, confirming that the pointer is compatible with
the type given in 'type'. Then finally setting '*ptr' to the pointer.
If flags is set to SWIG_POINTER_DISOWN, this is will clear any
ownership flag set on the object.
<br> This returns a value which can be checked with the macro
SWIG_IsOK()</div>
<p><tt>void SWIG_NewPointerObj(lua_State* L, void* ptr, swig_type_info
*type, int own);</tt></p>
<div class="indent"> This is the opposite of SWIG_ConvertPtr, as it
pushes a new userdata which wrappers the pointer 'ptr' of type 'type'.
The parameter 'own' specifies if the object is owned be Lua and if it
is 1 then Lua will GC the object when the userdata is disposed of.</div>
<p><tt>void* SWIG_MustGetPtr(lua_State* L, int index, swig_type_info
*type, int flags, int argnum, const char* func_name);</tt></p>
<div class="indent"> This function is a version of SWIG_ConvertPtr(),
except that it will either work, or it will trigger a lua_error() with
a text error message. This function is rarely used, and may be
deprecated in the future.</div>
<p><tt>SWIG_fail</tt></p>
<div class="indent"> This macro, when called within the context of a
SWIG wrapped function, will jump to the error handler code. This will
call any cleanup code (freeing any temp variables) and then triggers a
lua_error.
<br> A common use for this code is:
<br>
<pre>
if (!SWIG_IsOK(SWIG_ConvertPtr( .....)){
lua_pushstring(L, &quot;something bad happened&quot;);
SWIG_fail;
}</pre>
</div>
<p><tt>SWIG_fail_arg(char* func_name, int argnum, char* type)</tt></p>
<div class="indent"> This macro, when called within the context of a
SWIG wrapped function, will display the error message and jump to the
error handler code. The error message is of the form
<pre>
&quot;Error in <i>func_name</i> (arg <i>argnum</i>), expected '<i>type</i>' got '<i>whatever the type was</i>'&quot;
</pre>
</div>
<p><tt>SWIG_fail_ptr(const char* fn_name, int argnum, swig_type_info*
type);</tt></p>
<div class="indent"> Similar to SWIG_fail_arg, except that it will
display the swig_type_info information instead.</div>
<h2><a name="Lua_nn32">29.6 Customization of your Bindings</a></h2>
<p> This section covers adding of some small extra bits to your module
to add the last finishing touches.</p>
<h3><a name="Lua_nn33">29.6.1 Writing your own custom wrappers</a></h3>
<p> Sometimes, it may be necessary to add your own special functions,
which bypass the normal SWIG wrapper method, and just use the native
Lua API calls. These 'native' functions allow direct adding of your own
code into the module. This is performed with the <tt>%native</tt>
directive as follows:</p>
<div class="code">
<pre>%native(my_func) int native_function(lua_State*L); // registers native_function() with SWIG
...
%{
int native_function(lua_State*L) // my native code
{
...
}
%}
</pre>
</div>
<p> The <tt>%native</tt> directive in the above example, tells SWIG that
there is a function <tt>int native_function(lua_State*L);</tt> which is
to be added into the module under the name '<tt>my_func</tt>'. SWIG
will not add any wrapper for this function, beyond adding it into the
function table. How you write your code is entirely up to you.</p>
<h3><a name="Lua_nn34">29.6.2 Adding additional Lua code</a></h3>
<p> As well as adding additional C/C++ code, it's also possible to add
your own Lua code to the module as well. This code is executed once all
other initialisation, including the %init code has been called.</p>
<p> The directive <tt>%luacode</tt> adds code into the module which is
executed upon loading. Normally you would use this to add your own
functions to the module. Though you could easily perform other tasks.</p>
<div class="code">
<pre>%module example;
%luacode {
function example.greet()
print &quot;hello world&quot;
end
print &quot;Module loaded ok&quot;
}
...
%}
</pre>
</div>
<p> Notice that the code is not part of the module table. Therefore any
references to the module must have the module name added.</p>
<p> Should there be an error in the Lua code, this will<em> not</em>
stop loading of the module. The default behaviour of SWIG is to print
an error message to stderr and then continue. It is possible to change
this behaviour by using a <tt>#define SWIG_DOSTRING_FAIL(STR)</tt> to
define a different behaviour should the code fail.</p>
<p> Good uses for this feature is adding of new code, or writing helper
functions to simplify some of the code. See Examples/lua/arrays for an
example of this code.</p>
<h2><a name="Lua_nn35">29.7 Details on the Lua binding</a></h2>
<p> In the previous section, a high-level view of Lua wrapping was
presented. Obviously a lot of stuff happens behind the scenes to make
this happen. This section will explain some of the low-level details on
how this is achieved.</p>
<p><i> If you just want to use SWIG and don't care how it works, then
stop reading here. This is going into the guts of the code and how it
works. It's mainly for people who need to know what's going on within
the code.</i></p>
<h3><a name="Lua_nn36">29.7.1 Binding global data into the module.</a></h3>
<p> Assuming that you had some global data that you wanted to share
between C and Lua. How does SWIG do it?</p>
<div class="code">
<pre>%module example;
extern double Foo;
</pre>
</div>
<p> SWIG will effectively generate the pair of functions</p>
<div class="code">
<pre>void Foo_set(double);
double Foo_get();
</pre>
</div>
<p> At initialisation time, it will then add to the interpreter a table
called 'example', which represents the module. It will then add all its
functions to the module. (Note: older versions of SWIG actually added
the Foo_set() and Foo_get() functions, current implementation does not
add these functions any more.) But it also adds a metatable to this
table, which has two functions (<tt>__index</tt> and <tt>__newindex</tt>
) as well as two tables (<tt>.get</tt> and <tt>.set</tt>) The following
Lua code will show these hidden features.</p>
<div class="targetlang">
<pre>
&gt; print(example)
table: 003F8F90
&gt; m=getmetatable(example)
&gt; table.foreach(m, print)
.set table: 003F9088
.get table: 003F9038
__index function: 003F8FE0
__newindex function: 003F8FF8
&gt; g=m['.get']
&gt; table.foreach(g, print)
Foo function: 003FAFD8
&gt;
</pre>
</div>
<p> The .get and .set tables are lookups connecting the variable name
'Foo' to the accessor/mutator functions (Foo_set, Foo_get)</p>
<p> The Lua equivalent of the code for the <tt>__index</tt> and <tt>
__newindex</tt> looks a bit like this</p>
<div class="targetlang">
<pre>
function __index(mod, name)
local g=getmetatable(mod)['.get'] -- gets the table
if not g then return nil end
local f=g[name] -- looks for the function
-- calls it &amp; returns the value
if type(f)==&quot;function&quot; then return f() end
return nil
end
function __newindex(mod, name, value)
local s=getmetatable(mod)['.set'] -- gets the table
if not s then return end
local f=s[name] -- looks for the function
-- calls it to set the value
if type(f)==&quot;function&quot; then f(value)
else rawset(mod, name, value) end
end
</pre>
</div>
<p> That way when you call '<tt>a=example.Foo</tt>', the interpreter
looks at the table 'example' sees that there is no field 'Foo' and
calls __index. This will in turn check in '.get' table and find the
existence of 'Foo' and then return the value of the C function call
'Foo_get()'. Similarly for the code '<tt>example.Foo=10</tt>', the
interpreter will check the table, then call the __newindex which will
then check the '.set' table and call the C function 'Foo_set(10)'.</p>
<h3><a name="Lua_nn37">29.7.2 Userdata and Metatables</a></h3>
<p> As mentioned earlier, classes and structures, are all held as
pointer, using the Lua 'userdata' structure. This structure is actually
a pointer to a C structure 'swig_lua_userdata', which contains the
pointer to the data, a pointer to the swig_type_info (an internal SWIG
struct) and a flag which marks if the object is to be disposed of when
the interpreter no longer needs it. The actual accessing of the object
is done via the metatable attached to this userdata.</p>
<p> The metatable is a Lua 5.0 feature (which is also why SWIG cannot
wrap Lua 4.0). It's a table which holds a list of functions, operators
and attributes. This is what gives the userdata the feeling that it is
a real object and not just a hunk of memory.</p>
<p> Given a class</p>
<div class="code">
<pre>%module excpp;
class Point
{
public:
int x, y;
Point(){x=y=0;}
~Point(){}
virtual void Print(){printf(&quot;Point @%p (%d, %d)\n&quot;, this, x, y);}
};
</pre>
</div>
<p> SWIG will create a module excpp, with all the various functions
inside. However to allow the intuitive use of the userdata, SWIG also
creates up a set of metatables. As seen in the above section on global
variables, use of the metatables allows for wrappers to be used
intuitively. To save effort, the code creates one metatable per class
and stores it inside Lua's registry. Then when a new object is
instantiated, the metatable is found in the registry and the userdata
associated with the metatable. Currently, derived classes make a
complete copy of the base class' table and then add on their own
additional functions.</p>
<p> Some of the internals can be seen by looking at the metatable of a
class:</p>
<div class="targetlang">
<pre>
&gt; p=excpp.Point()
&gt; print(p)
userdata: 003FDB28
&gt; m=getmetatable(p)
&gt; table.foreach(m, print)
.type Point
__gc function: 003FB6C8
__newindex function: 003FB6B0
__index function: 003FB698
.get table: 003FB4D8
.set table: 003FB500
.fn table: 003FB528
</pre>
</div>
<p> The '.type' attribute is the name of the class. The '.get' and
'.set' tables work in a similar manner to the modules, the main
difference is the '.fn' table which also holds all the member
functions. (The '__gc' function is the class' destructor function)</p>
<p> The Lua equivalent of the code for enabling functions looks a little
like this</p>
<div class="targetlang">
<pre>
function __index(obj, name)
local m=getmetatable(obj) -- gets the metatable
if not m then return nil end
local g=m['.get'] -- gets the attribute table
if not g then return nil end
local f=g[name] -- looks for the get_attribute function
-- calls it &amp; returns the value
if type(f)==&quot;function&quot; then return f() end
-- ok, so it not an attribute, maybe it's a function
local fn=m['.fn'] -- gets the function table
if not fn then return nil end
local f=fn[name] -- looks for the function
-- if found the fn then return the function
-- so the interpreter can call it
if type(f)==&quot;function&quot; then return f end
return nil
end
</pre>
</div>
<p> So when 'p:Print()' is called, the __index looks on the object
metatable for a 'Print' attribute, then looks for a 'Print' function.
When it finds the function, it returns the function, and then
interpreter can call 'Point_Print(p)'</p>
<p> In theory, you can play with this usertable &amp; add new features, but
remember that it is a shared table between all instances of one class,
and you could very easily corrupt the functions in all the instances.</p>
<p> Note: Both the opaque structures (like the FILE*) and normal wrapped
classes/structs use the same 'swig_lua_userdata' structure. Though the
opaque structures do not have a metatable attached, or any information
on how to dispose of them when the interpreter has finished with them.</p>
<p> Note: Operator overloads are basically done in the same way, by
adding functions such as '__add' &amp; '__call' to the class' metatable.
The current implementation is a bit rough as it will add any member
function beginning with '__' into the metatable too, assuming it's an
operator overload.</p>
<h3><a name="Lua_nn38">29.7.3 Memory management</a></h3>
<p> Lua is very helpful with the memory management. The
'swig_lua_userdata' is fully managed by the interpreter itself. This
means that neither the C code nor the Lua code can damage it. Once a
piece of userdata has no references to it, it is not instantly
collected, but will be collected when Lua deems is necessary. (You can
force collection by calling the Lua function <tt>collectgarbage()</tt>
). Once the userdata is about to be free'ed, the interpreter will check
the userdata for a metatable and for a function '__gc'. If this exists
this is called. For all complete types (ie normal wrapped classes &amp;
structs) this should exist. The '__gc' function will check the
'swig_lua_userdata' to check for the 'own' field and if this is true
(which is will be for all owned data) it will then call the destructor
on the pointer.</p>
<p> It is currently not recommended to edit this field or add some user
code, to change the behaviour. Though for those who wish to try, here
is where to look.</p>
<p> It is also currently not possible to change the ownership flag on
the data (unlike most other scripting languages, Lua does not permit
access to the data from within the interpreter).</p>
<HR NOSHADE>
<h1><a name="Octave">30 SWIG and Octave</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Octave_nn2">Preliminaries</a></li>
<li><a href="#Octave_nn3">Running SWIG</a>
<ul>
<li><a href="#Octave_nn4">Command-line options</a></li>
<li><a href="#Octave_nn5">Compiling a dynamic module</a></li>
<li><a href="#Octave_nn6">Using your module</a></li>
</ul>
</li>
<li><a href="#Octave_nn7">A tour of basic C/C++ wrapping</a>
<ul>
<li><a href="#Octave_nn8">Modules</a></li>
<li><a href="#Octave_nn9">Functions</a></li>
<li><a href="#Octave_nn10">Global variables</a></li>
<li><a href="#Octave_nn11">Constants and enums</a></li>
<li><a href="#Octave_nn12">Pointers</a></li>
<li><a href="#Octave_nn13">Structures and C++ classes</a></li>
<li><a href="#Octave_nn15">C++ inheritance</a></li>
<li><a href="#Octave_nn17">C++ overloaded functions</a></li>
<li><a href="#Octave_nn18">C++ operators</a></li>
<li><a href="#Octave_nn19">Class extension with %extend</a></li>
<li><a href="#Octave_nn20">C++ templates</a></li>
<li><a href="#Octave_nn21">C++ Smart Pointers</a>
<ul>
<li><a href="#Octave_smart_pointers_shared_ptr">The shared_ptr Smart
Pointer</a></li>
<li><a href="#Octave_smart_pointers_generic">Generic Smart Pointers</a></li>
</ul>
</li>
<li><a href="#Octave_nn22">Directors (calling Octave from C++ code)</a></li>
<li><a href="#Octave_nn23">Threads</a></li>
<li><a href="#Octave_nn24">Memory management</a></li>
<li><a href="#Octave_nn25">STL support</a></li>
<li><a href="#Octave_nn26">Matrix typemaps</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> Octave is a high-level language intended for numerical programming
that is mostly compatible with MATLAB. More information can be found at
<a href="https://octave.org/">Octave web site</a>.</p>
<p> This chapter is intended to give an introduction to using the
module. You should also read the SWIG documentation that is not
specific to Octave. Also, there are a dozen or so examples in the
Examples/octave directory, and hundreds in the test suite
(Examples/test-suite and Examples/test-suite/octave).</p>
<h2><a name="Octave_nn2">30.1 Preliminaries</a></h2>
<p> SWIG is regularly tested against the following versions of Octave:
5.2, 6.4. SWIG 4.2.1 has also been manually tested with 7.3, 8.4 and
9.0.</p>
<p> Every effort is made to maintain backward compatibility with older
versions of Octave. This cannot be guaranteed however, as in recent
times new Octave releases have required nontrivial updates to SWIG,
which may break backward compatibility for older Octave versions
against which SWIG is not regularly tested.</p>
<p> The SWIG runtime exports the function <tt>swig_octave_prereq()</tt>
for checking the version of Octave.</p>
<h2><a name="Octave_nn3">30.2 Running SWIG</a></h2>
<p> Let's start with a very simple SWIG interface file, example.i:</p>
<div class="code">
<pre>
%module swigexample
%{
#include &quot;example.h&quot;
%}
int gcd(int x, int y);
extern double Foo; </pre>
</div>
<p> To build an Octave module when wrapping C code, run SWIG using the <tt>
-octave</tt> option:</p>
<div class="shell">
<pre>$ swig -octave -o example_wrap.cpp example.i </pre>
</div>
<p> The <tt>-c++</tt> option is also required when wrapping C++ code:</p>
<div class="shell">
<pre>$ swig -octave -c++ -o example_wrap.cpp example.i </pre>
</div>
<h3><a name="Octave_nn4">30.2.1 Command-line options</a></h3>
<p> The swig command line has a number of options you can use, like to
redirect its output. Use <tt>swig -help</tt> to learn about these.
Options specific to the Octave module are:</p>
<div class="shell">
<pre>$ swig -octave -help
...
Octave Options (available with -octave)
-globals <em>name</em> - Set <em>name</em> used to access C global variables [default: 'cvar']
Use '.' to load C global variables into module namespace
-opprefix <em>str</em> - Prefix <em>str</em> for global operator functions [default: 'op_']
</pre>
</div>
<p> The<em> -globals</em> option sets the name of the variable which is
the namespace for C global variables exported by the module. The
special name &quot;.&quot; loads C global variables into the module namespace,
i.e. alongside C functions and structs exported by the module. The<em>
-opprefix</em> options sets the prefix of the names of global/friend <a href="#Octave_nn18">
operator</a> functions.</p>
<h3><a name="Octave_nn5">30.2.2 Compiling a dynamic module</a></h3>
<p> Octave modules are DLLs/shared objects having the &quot;.oct&quot; suffix.
Building an oct file is usually done with the mkoctfile command (either
within Octave itself, or from the shell). For example,</p>
<div class="shell">
<pre>
$ swig -octave -c++ -o example_wrap.cpp example.i
$ mkoctfile example_wrap.cpp example.c
</pre>
</div>
<p> where &quot;example.c&quot; is the file containing the gcd() implementation.</p>
<p> mkoctfile can also be used to extract the build parameters required
to invoke the compiler and linker yourself. See the Octave manual and
mkoctfile man page.</p>
<p> mkoctfile will produce &quot;swigexample.oct&quot;, which contains the
compiled extension module. Loading it into Octave is then a matter of
invoking</p>
<div class="targetlang">
<pre>octave:1&gt; swigexample</pre>
</div>
<h3><a name="Octave_nn6">30.2.3 Using your module</a></h3>
<p> Assuming all goes well, you will be able to do this:
<br></p>
<div class="targetlang">
<pre>$ octave -q
octave:1&gt; swigexample
octave:2&gt; swigexample.gcd(4, 6)
ans = 2
octave:3&gt; swigexample.cvar.Foo
ans = 3
octave:4&gt; swigexample.cvar.Foo=4;
octave:5&gt; swigexample.cvar.Foo
ans = 4 </pre>
</div>
<h2><a name="Octave_nn7">30.3 A tour of basic C/C++ wrapping</a></h2>
<h3><a name="Octave_nn8">30.3.1 Modules</a></h3>
<p> The SWIG module directive specifies the name of the Octave module.
If you specify &quot;module swigexample&quot;, then in Octave everything in the
module will be accessible under &quot;swigexample&quot;, as in the above example.
When choosing a module name, make sure you don't use the same name as a
built-in Octave command or standard module name.</p>
<p> When Octave is asked to invoke <tt>swigexample</tt>, it will try to
find the &quot;.m&quot; or &quot;.oct&quot; file that defines the function &quot;swigexample&quot;.
You therefore need to make sure that &quot;swigexample.oct&quot; is in Octave's
search path, which can be specified with the environment variable
&quot;OCTAVE_PATH&quot;.</p>
<p> To load an Octave module, simply type its name:</p>
<div class="targetlang">
<pre>
octave:1&gt; swigexample;
octave:2&gt; gcd(4, 6)
ans = 2
octave:3&gt; cvar.Foo
ans = 3
octave:4&gt; cvar.Foo=4;
octave:5&gt; cvar.Foo
ans = 4
</pre>
</div>
<p> Modules can also be loaded from within functions, even before being
loaded in the base context. If the module is also used in the base
context, however, it must first be loaded again:</p>
<div class="targetlang">
<pre>
octave:1&gt; function l = my_lcm(a, b)
&gt; swigexample
&gt; l = abs(a*b)/swigexample.gcd(a, b);
&gt; endfunction
octave:2&gt; my_lcm(4, 6)
ans = 12
octave:3&gt; swigexample.gcd(4, 6)
error: can't perform indexing operations for &lt;unknown type&gt; type
octave:3&gt; swigexample;
octave:4&gt; swigexample.gcd(4, 6)
ans = 2
</pre>
</div>
<h3><a name="Octave_nn9">30.3.2 Functions</a></h3>
<p> Global functions are wrapped as new Octave built-in functions. For
example,</p>
<div class="code">
<pre>%module swigexample
int fact(int n); </pre>
</div>
<p> creates a built-in function <tt>swigexample.fact(n)</tt> that works
exactly like you think it does:</p>
<div class="targetlang">
<pre>octave:1&gt; swigexample.fact(4)
24 </pre>
</div>
<h3><a name="Octave_nn10">30.3.3 Global variables</a></h3>
<p> Global variables are a little special in Octave. Given a global
variable:</p>
<div class="code">
<pre>%module swigexample
extern double Foo;
</pre>
</div>
<p> To expose variables, SWIG actually generates two functions, to get
and set the value. In this case, Foo_set and Foo_set would be
generated. SWIG then automatically calls these functions when you get
and set the variable-- in the former case creating a local copy in the
interpreter of the C variables, and in the latter case copying an
interpreter variables onto the C variable.</p>
<div class="targetlang">
<pre>octave:1&gt; swigexample;
octave:2&gt; c=swigexample.cvar.Foo
c = 3
octave:3&gt; swigexample.cvar.Foo=4;
octave:4&gt; c
c = 3
octave:5&gt; swigexample.cvar.Foo
ans = 4</pre>
</div>
<p> If a variable is marked with the %immutable directive then any
attempts to set this variable will cause an Octave error. Given a
global variable:</p>
<div class="code">
<pre>%module swigexample
%immutable;
extern double Foo;
%mutable;
</pre>
</div>
<p> SWIG will allow the reading of <tt>Foo</tt> but when a set attempt
is made, an error function will be called.</p>
<div class="targetlang">
<pre>octave:1&gt; swigexample
octave:2&gt; swigexample.Foo=4
error: attempt to set immutable member variable
error: assignment failed, or no method for `swig_type = scalar'
error: evaluating assignment expression near line 2, column 12 </pre>
</div>
<p> It is possible to add new functions or variables to the module. This
also allows the user to rename/remove existing functions and constants
(but not linked variables, mutable or immutable). Therefore users are
recommended to be careful when doing so.</p>
<div class="targetlang">
<pre>octave:1&gt; swigexample;
octave:2&gt; swigexample.PI=3.142;
octave:3&gt; swigexample.PI
ans = 3.1420 </pre>
</div>
<h3><a name="Octave_nn11">30.3.4 Constants and enums</a></h3>
<p> Because Octave doesn't really have the concept of constants, C/C++
constants are not really constant in Octave. They are actually just a
copy of the value into the Octave interpreter. Therefore they can be
changed just as any other value. For example given some constants:</p>
<div class="code">
<pre>%module swigexample
%constant int ICONST=42;
#define SCONST &quot;Hello World&quot;
enum Days{SUNDAY, MONDAY, TUESDAY, WEDNESDAY, THURSDAY, FRIDAY, SATURDAY};
</pre>
</div>
<p> This is 'effectively' converted into the following Octave code:</p>
<div class="targetlang">
<pre>swigexample.ICONST=42
swigexample.SCONST=&quot;Hello World&quot;
swigexample.SUNDAY=0
.... </pre>
</div>
<h3><a name="Octave_nn12">30.3.5 Pointers</a></h3>
<p> C/C++ pointers are fully supported by SWIG. Furthermore, SWIG has no
problem working with incomplete type information. Given a wrapping of
the &lt;file.h&gt; interface: C/C++ pointers are fully supported by SWIG.
Furthermore, SWIG has no problem working with incomplete type
information. Given a wrapping of the &lt;file.h&gt; interface:</p>
<div class="code">
<pre>%module swigexample
FILE *fopen(const char *filename, const char *mode);
int fputs(const char *, FILE *);
int fclose(FILE *);
</pre>
</div>
<p> When wrapped, you will be able to use the functions in a natural way
from Octave. For example:</p>
<div class="targetlang">
<pre>
octave:1&gt; swigexample;
octave:2&gt; f=swigexample.fopen(&quot;w&quot;, &quot;junk&quot;);
octave:3&gt; swigexample.fputs(&quot;Hello world&quot;, f);
octave:4&gt; swigexample.fclose(f);
</pre>
</div>
<p> Simply printing the value of a wrapped C++ type will print its
typename. E.g.,</p>
<div class="targetlang">
<pre>octave:1&gt; swigexample;
octave:2&gt; f=swigexample.fopen(&quot;junk&quot;, &quot;w&quot;);
octave:3&gt; f
f =
{
_p_FILE, ptr = 0x9b0cd00
} </pre>
</div>
<p> As the user of the pointer, you are responsible for freeing it, or
closing any resources associated with it (just as you would in a C
program). This does not apply so strictly to classes and structs (see
below).</p>
<div class="targetlang">
<pre>octave:1&gt; swigexample;
octave:2&gt; f=swigexample.fopen(&quot;not there&quot;, &quot;r&quot;);
error: value on right hand side of assignment is undefined
error: evaluating assignment expression near line 2, column 2 </pre>
</div>
<p> NULL C/C++ pointers are represented by the Octave null matrix, <tt>
[]</tt>.</p>
<h3><a name="Octave_nn13">30.3.6 Structures and C++ classes</a></h3>
<p> SWIG wraps C structures and C++ classes by using a special Octave
type called a <tt>swig_ref</tt>. A <tt>swig_ref</tt> contains a
reference to one or more instances of C/C++ objects, or just the type
information for an object. For each wrapped structure and class, a <tt>
swig_ref</tt> will be exposed that has the name of the type. When
invoked as a function, it creates a new object of its type and returns
a <tt>swig_ref</tt> that points to that instance. This provides a very
natural interface. For example,</p>
<div class="code">
<pre>struct Point{
int x, y;
};
</pre>
</div>
<p> is used as follows:</p>
<div class="targetlang">
<pre>octave:1&gt; swigexample;
octave:2&gt; p=swigexample.Point();
octave:3&gt; p.x=3;
octave:4&gt; p.y=5;
octave:5&gt; p.x, p.y
ans = 3
ans = 5
</pre>
</div>
<p> In C++, invoking the type object in this way calls the object's
constructor. <tt>swig_ref</tt> objects can also be acquired by having a
wrapped function return a pointer, reference, or value of a
non-primitive type.</p>
<p> The swig_ref type handles indexing operations such that usage maps
closely to what you would have in C/C++. Structure members are accessed
as in the above example, by calling set and get methods for C++
variables. Methods also work as expected. For example, code wrapped in
the following way</p>
<div class="code">
<pre>class Point{
public:
int x, y;
Point(int _x, int _y) : x(_x), y(_y) {}
double distance(const Point&amp; rhs) {
return sqrt(pow(x-rhs.x, 2)+pow(y-rhs.y, 2));
}
void set(int _x, int _y) {
x=_x; y=_y;
}
};
</pre>
</div>
<p> can be used from Octave like this</p>
<div class="targetlang">
<pre>octave:1&gt; swigexample;
octave:2&gt; p1=swigexample.Point(3, 5);
octave:3&gt; p2=swigexample.Point(1, 2);
octave:4&gt; p1.distance(p2)
ans = 3.6056
</pre>
</div>
<p> By using the <tt>swig_this()</tt> and <tt>swig_type()</tt>
functions, one can discover the pointers to and types of the underlying
C/C++ object.</p>
<div class="targetlang">
<pre>
octave:5&gt; swig_this(p1)
ans = 162504808
octave:6&gt; swig_type(p1)
ans = Point
</pre>
</div>
<p> Note that <tt>swig_ref</tt> is a reference-counted pointer to a
C/C++ object/type, and as such has pass-by-reference semantics. For
example if one has a allocated a single object but has two <tt>swig_ref</tt>
's pointing to it, modifying the object through either of them will
change the single allocated object. This differs from the usual
pass-by-value (copy-on-write) semantics that Octave maintains for
built-in types. For example, in the following snippet, modifying <tt>b</tt>
does not modify <tt>a</tt>,</p>
<div class="targetlang">
<pre>
octave:7&gt; a=struct('x', 4)
a =
{
x = 4
}
octave:8&gt; b=a
b =
{
x = 4
}
octave:9&gt; b.y=4
b =
{
x = 4
y = 4
}
octave:10&gt; a
a =
{
x = 4
}
</pre>
</div>
<p> However, when dealing with wrapped objects, one gets the behavior</p>
<div class="targetlang">
<pre>
octave:2&gt; a=Point(3, 5)
a =
{
Point, ptr = 0x9afbbb0
}
octave:3&gt; b=a
b =
{
Point, ptr = 0x9afbbb0
}
octave:4&gt; b.set(2, 1);
octave:5&gt; b.x, b.y
ans = 2
ans = 1
octave:6&gt; a.x, a.y
ans = 2
ans = 1
</pre>
</div>
<p> Depending on the ownership setting of a <tt>swig_ref</tt>, it may
call C++ destructors when its reference count goes to zero. See the
section on memory management below for details.</p>
<h3><a name="Octave_nn15">30.3.7 C++ inheritance</a></h3>
<p> Single and multiple inheritance are fully supported. The <tt>
swig_ref</tt> type carries type information along with any C++ object
pointer it holds. This information contains the full class hierarchy.
When an indexing operation (such as a method invocation) occurs, the
tree is walked to find a match in the current class as well as any of
its bases. The lookup is then cached in the <tt>swig_ref</tt>.</p>
<h3><a name="Octave_nn17">30.3.8 C++ overloaded functions</a></h3>
<p> Overloaded functions are supported, and handled as in other modules.
That is, each overload is wrapped separately (under internal names),
and a dispatch function is also emitted under the external/visible
name. The dispatch function selects which overload to call (if any)
based on the passed arguments. <tt>typecheck</tt> typemaps are used to
analyze each argument, as well as assign precedence. See the chapter on
typemaps for details.</p>
<h3><a name="Octave_nn18">30.3.9 C++ operators</a></h3>
<p> C++ operator overloading is supported, in a way similar to other
modules. The <tt>swig_ref</tt> type supports all unary and binary
operators between itself and all other types that exist in the system
at module load time. When an operator is used (where one of the
operands is a <tt>swig_ref</tt>), the runtime routes the call to either
a member function of the given object, or to a global function whose
named is derived from the types of the operands (either both or just
the lhs or rhs).</p>
<p> For example, if <tt>a</tt> and <tt>b</tt> are SWIG variables in
Octave, <tt>a+b</tt> becomes <tt>a.__add__(b)</tt>. The wrapper is then
free to implement __add__ to do whatever it wants. A wrapper may define
the <tt>__add__</tt> function manually, %rename some other function to
it, or %rename a C++ operator to it.</p>
<p> By default the C++ operators are renamed to their corresponding
Octave operators. So without doing any work, the following interface</p>
<div class="code">
<pre>
%inline {
struct A {
int value;
A(int _value) : value(_value) {}
A operator+ (const A&amp; x) {
return A(value+x.value);
}
};
}
</pre>
</div>
<p> is usable from Octave like this:</p>
<div class="targetlang">
<pre>
a=A(2), b=A(3), c=a+b
assert(c.value==5);
</pre>
</div>
<p> Octave operators are mapped in the following way:</p>
<div class="code">
<pre>
__brace__ a{args}
__brace_asgn__ a{args} = rhs
__paren__ a(args)
__paren_asgn__ a(args) = rhs
__str__ generates string rep
__not__ !a
__uplus__ +a
__uminus__ -a
__transpose__ a.'
__hermitian__ a'
__incr__ a++
__decr__ a--
__add__ a + b
__sub__ a - b
__mul__ a * b
__div__ a / b
__pow__ a ^ b
__ldiv__ a \ b
__lshift__ a &lt;&lt; b
__rshift__ a &gt;&gt; b
__lt__ a &lt; b
__le__ a &lt;= b
__eq__ a == b
__ge__ a &gt;= b
__gt__ a &gt; b
__ne__ a != b
__el_mul__ a .* b
__el_div__ a ./ b
__el_pow__ a .^ b
__el_ldiv__ a .\ b
__el_and__ a &amp; b
__el_or__ a | b
</pre>
</div>
<p> On the C++ side, the default mappings are as follows:</p>
<div class="code">
<pre>
%rename(__add__) *::operator+;
%rename(__add__) *::operator+();
%rename(__add__) *::operator+() const;
%rename(__sub__) *::operator-;
%rename(__uminus__) *::operator-();
%rename(__uminus__) *::operator-() const;
%rename(__mul__) *::operator*;
%rename(__div__) *::operator/;
%rename(__mod__) *::operator%;
%rename(__lshift__) *::operator&lt;&lt;;
%rename(__rshift__) *::operator&gt;&gt;;
%rename(__el_and__) *::operator&amp;&amp;;
%rename(__el_or__) *::operator||;
%rename(__xor__) *::operator^;
%rename(__invert__) *::operator~;
%rename(__lt__) *::operator&lt;;
%rename(__le__) *::operator&lt;=;
%rename(__gt__) *::operator&gt;;
%rename(__ge__) *::operator&gt;=;
%rename(__eq__) *::operator==;
%rename(__ne__) *::operator!=;
%rename(__not__) *::operator!;
%rename(__incr__) *::operator++;
%rename(__decr__) *::operator--;
%rename(__paren__) *::operator();
%rename(__brace__) *::operator[];
</pre>
</div>
<p> Octave can also utilise friend (i.e. non-member) operators with a
simple %rename: see the example in the Examples/octave/operator
directory.</p>
<p> Octave has several operators for which no corresponding C++
operators exist. For example, the Octave code</p>
<div class="targetlang">
<pre>
x=[a,b,c];
</pre>
</div>
<p> calls the Octave operator <tt>horzcat</tt> of the class of <tt>a</tt>
. Hence, if <tt>a</tt> is of type <tt>swig_ref</tt> you can write an
overload for this operator for your wrapped C++ class by placing a file
<tt>@swig_ref/horzcat.m</tt> in the Octave load path (like for every
Octave class, see <a href="https://docs.octave.org/latest/Creating-a-Class.html">
Creating a Class</a>). This Octave function file is then called whenever
the above Octave code is executed for a variable of type <tt>swig_ref</tt>
.</p>
<h3><a name="Octave_nn19">30.3.10 Class extension with %extend</a></h3>
<p> The %extend directive works the same as in other modules.</p>
<p> You can use it to define special behavior, like for example defining
Octave operators not mapped to C++ operators, or defining certain
Octave mechanisms such as how an object prints. For example, the <tt>
octave_value::{is_string, string_value, print}</tt> functions are routed
to a special method <tt>__str__</tt> that can be defined inside an
%extend.</p>
<div class="code">
<pre>
%extend A {
string __str__() {
stringstream sout;
sout&lt;&lt;$self-&gt;value;
return sout.str();
}
}
</pre>
</div>
<p> Then in Octave one gets,</p>
<div class="targetlang">
<pre>
octave:1&gt; a=A(4);
octave:2&gt; a
a = 4
octave:3&gt; printf(&quot;%s\n&quot;, a);
4
octave:4&gt; a.__str__()
4
</pre>
</div>
<p> Similarly, Octave can use the <tt>__float__</tt> method to convert
an object to a numeric value.</p>
<p> Octave 3.8.0 and later versions will also map unary functions X() to
the corresponding <tt>__X__</tt> method, where X includes: abs(),
acos(), acosh(), angle(), arg(), asin(), asinh(), atan(), atanh(),
cbrt(), ceil(), conj(), cos(), cosh(), dawson(), erf(), erfc(),
erfcinv(), erfcx(), erfi(), erfinv(), exp(), expm1(), finite(), fix(),
floor(), gamma(), imag(), isalnum(), isalpha(), isascii(), iscntrl(),
isdigit(), isgraph(), isinf(), islower(), isna(), isnan(), isprint(),
ispunct(), isspace(), isupper(), isxdigit(), lgamma(), log(), log10(),
log1p(), log2(), real(), round(), roundb(), signbit(), signum(), sin(),
sinh(), sqrt(), tan(), tanh(), toascii(), tolower(), toupper()</p>
<h3><a name="Octave_nn20">30.3.11 C++ templates</a></h3>
<p> C++ class and function templates are fully supported as in other
modules, in that the %template directive may used to create explicit
instantiations of templated types. For example, function templates can
be instantiated as follows:</p>
<div class="code">
<pre>%module swigexample
%inline {
template&lt;class __scalar&gt;
__scalar mul(__scalar a, __scalar b) {
return a*b;
}
}
%include &lt;std_complex.i&gt;
%template(mul) mul&lt;std::complex&lt;double&gt; &gt;
%template(mul) mul&lt;double&gt;
</pre>
</div>
<p> and then used from Octave</p>
<div class="targetlang">
<pre>
octave:1&gt; mul(4, 3)
ans = 12
octave:2&gt; mul(4.2, 3.6)
ans = 15.120
octave:3&gt; mul(3+4i, 10+2i)
ans = 22 + 46i
</pre>
</div>
<p> Similarly, class templates can be instantiated as in the following
example,</p>
<div class="code">
<pre>%module swigexample
%include &lt;std_complex.i&gt;
%include &lt;std_string.i&gt;
%inline {
#include &lt;sstream&gt;
template&lt;class __scalar&gt; class sum {
__scalar s;
public:
sum(__scalar _s=0) : s(_s) {}
sum&amp; add(__scalar _s) {
s+=_s;
return *this;
}
std::string __str__() const {
std::stringstream sout;
sout&lt;&lt;s;
return sout.str();
}
};
}
%template(sum_complex) sum&lt;std::complex&lt;double&gt; &gt;;
%template(sum_double) sum&lt;double&gt;;
</pre>
</div>
<p> and then used from Octave</p>
<div class="targetlang">
<pre>
octave:2&gt; a=sum_complex(2+3i);
octave:3&gt; a.add(2)
ans =
(4, 3)
octave:4&gt; a.add(3+i)
ans =
(7, 4)
</pre>
</div>
<h3><a name="Octave_nn21">30.3.12 C++ Smart Pointers</a></h3>
<h4><a name="Octave_smart_pointers_shared_ptr">30.3.12.1 The shared_ptr
Smart Pointer</a></h4>
<p> The C++11 standard provides <tt>std::shared_ptr</tt> which was
derived from the Boost implementation, <tt>boost::shared_ptr</tt>. Both
of these are available for Octave in the SWIG library and usage is
outlined in the <a href="#Library_std_shared_ptr">shared_ptr smart
pointer</a> library section.</p>
<h4><a name="Octave_smart_pointers_generic">30.3.12.2 Generic Smart
Pointers</a></h4>
<p> C++ smart pointers are fully supported as in other modules.</p>
<h3><a name="Octave_nn22">30.3.13 Directors (calling Octave from C++
code)</a></h3>
<p> There is full support for SWIG Directors, which permits Octave code
to subclass C++ classes, and implement their virtual methods.</p>
<p> Octave has no direct support for object oriented programming,
however the <tt>swig_ref</tt> type provides some of this support. You
can manufacture a <tt>swig_ref</tt> using the <tt>subclass</tt>
function (provided by the SWIG/Octave runtime).</p>
<p> For example,</p>
<div class="targetlang">
<pre>
octave:1&gt; a=subclass();
octave:2&gt; a.my_var = 4;
octave:3&gt; a.my_method = @(self) printf(&quot;my_var = &quot;, self.my_var);
octave:4&gt; a.my_method();
my_var = 4
</pre>
</div>
<p> <tt>subclass()</tt> can also be used to subclass one or more C++
types. Suppose you have an interface defined by</p>
<div class="code">
<pre>
%inline {
class A {
public:
virtual my_method() {
printf(&quot;c-side routine called\n&quot;);
}
};
void call_your_method(A&amp; a) {
a.my_method();
}
}
</pre>
</div>
<p> Then from Octave you can say:</p>
<div class="targetlang">
<pre>
octave:1&gt; B=@() subclass(A(), @my_method);
octave:2&gt; function my_method(self)
octave:3&gt; printf(&quot;octave-side routine called\n&quot;);
octave:4&gt; end
octave:5&gt; call_your_method(B());
octave-side routine called
</pre>
</div>
<p> or more concisely,</p>
<div class="targetlang">
<pre>
octave:1&gt; B=@() subclass(A(), 'my_method', @(self) printf(&quot;octave-side routine called\n&quot;));
octave:2&gt; call_your_method(B());
octave-side routine called
</pre>
</div>
<p> Note that you have to enable directors via the %feature directive
(see other modules for this).</p>
<p> <tt>subclass()</tt> will accept any number of C++ bases or other <tt>
subclass()</tt>'ed objects, <tt>(string, octave_value)</tt> pairs, and <tt>
function_handles</tt>. In the first case, these are taken as base
classes; in the second case, as named members (either variables or
functions, depending on whether the given value is a function handle);
in the third case, as member functions whose name is taken from the
given function handle. E.g.,</p>
<div class="targetlang">
<pre>
octave:1&gt; B=@(some_var=2) subclass(A(), 'some_var', some_var, @some_func, 'another_func',
@(self) do_stuff())
</pre>
</div>
<p> You can also assign non-C++ member variables and functions after
construct time. There is no support for non-C++ static members.</p>
<p> There is limited support for explicitly referencing C++ bases. So,
in the example above, we could have</p>
<div class="targetlang">
<pre>
octave:1&gt; B=@() subclass(A(), @my_method);
octave:2&gt; function my_method(self)
octave:3&gt; self.A.my_method();
octave:4&gt; printf(&quot;octave-side routine called\n&quot;);
octave:5&gt; end
octave:6&gt; call_your_method(B());
c-side routine called
octave-side routine called
</pre>
</div>
<h3><a name="Octave_nn23">30.3.14 Threads</a></h3>
<p> The use of threads in wrapped Director code is not supported; i.e.,
an Octave-side implementation of a C++ class must be called from the
Octave interpreter's thread. Anything fancier (apartment/queue model,
whatever) is left to the user. Without anything fancier, this amounts
to the limitation that Octave must drive the module... like, for
example, an optimization package that calls Octave to evaluate an
objective function.</p>
<h3><a name="Octave_nn24">30.3.15 Memory management</a></h3>
<p> As noted above, <tt>swig_ref</tt> represents a reference counted
pointer to a C/C++-side object. It also contains a flag indicating
whether Octave or the C/C++ code owns the object. If Octave owns it,
any destructors will be called when the reference count reaches zero.
If the C/C++ side owns the object, then destructors will not be called
when the reference count goes to zero. As noted above, <tt>swig_ref</tt>
represents a reference counted pointer to a C/C++-side object. It also
contains a flag indicating whether Octave or the C/C++ code owns the
object. If Octave owns it, any destructors will be called when the
reference count reaches zero. If the C/C++ side owns the object, then
destructors will not be called when the reference count goes to zero.</p>
<p> For example,<div class="code">
<pre>
%inline {
class A {
public:
A() { printf(&quot;A constructing\n&quot;); }
~A() { printf(&quot;A destructing\n&quot;); }
};
}
</pre>
</div></p>
<p> Would produce this behavior in Octave:</p>
<div class="targetlang">
<pre>
octave:1&gt; a=A();
A constructing
octave:2&gt; b=a;
octave:3&gt; clear a;
octave:4&gt; b=4;
A destructing
</pre>
</div>
<p> The %newobject directive may be used to control this behavior for
pointers returned from functions.</p>
<p> In the case where one wishes for the C++ side to own an object that
was created in Octave (especially a Director object), one can use the
__disown() method to invert this logic. Then letting the Octave
reference count go to zero will not destroy the object, but destroying
the object will invalidate the Octave-side object if it still exists
(and call destructors of other C++ bases in the case of multiple
inheritance/<tt>subclass()</tt>'ing).</p>
<h3><a name="Octave_nn25">30.3.16 STL support</a></h3>
<p> Various STL library files are provided for wrapping STL containers.</p>
<h3><a name="Octave_nn26">30.3.17 Matrix typemaps</a></h3>
<p> Octave provides a rich set of classes for dealing with matrices.
Currently there are no built-in typemaps to deal with those. However,
these are relatively straight forward for users to add themselves (see
the docs on typemaps). Without much work (a single typemap decl-- say,
5 lines of code in the interface file), it would be possible to have a
function</p>
<div class="code">
<pre>
double my_det(const double* mat, int m, int n);
</pre>
</div>
<p> that is accessed from Octave as,</p>
<div class="targetlang">
<pre>
octave:1&gt; my_det(rand(4));
ans = -0.18388
</pre>
</div> <tt>
<br></tt><HR NOSHADE>
<h1><a name="Perl5">31 SWIG and Perl5</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Perl5_nn2">Overview</a></li>
<li><a href="#Perl5_nn3">Preliminaries</a>
<ul>
<li><a href="#Perl5_nn4">Getting the right header files</a></li>
<li><a href="#Perl5_nn5">Compiling a dynamic module</a></li>
<li><a href="#Perl5_nn6">Building a dynamic module with MakeMaker</a></li>
<li><a href="#Perl5_nn7">Building a static version of Perl</a></li>
<li><a href="#Perl5_nn8">Using the module</a></li>
<li><a href="#Perl5_nn9">Compilation problems and compiling with C++</a></li>
<li><a href="#Perl5_nn10">Compiling for 64-bit platforms</a></li>
</ul>
</li>
<li><a href="#Perl5_nn11">Building Perl Extensions under Windows</a>
<ul>
<li><a href="#Perl5_nn12">Running SWIG from Developer Studio</a></li>
<li><a href="#Perl5_nn13">Using other compilers</a></li>
</ul>
</li>
<li><a href="#Perl5_nn14">The low-level interface</a>
<ul>
<li><a href="#Perl5_nn15">Functions</a></li>
<li><a href="#Perl5_nn16">Global variables</a></li>
<li><a href="#Perl5_nn17">Constants</a></li>
<li><a href="#Perl5_nn18">Pointers</a></li>
<li><a href="#Perl5_nn19">Structures</a></li>
<li><a href="#Perl5_nn20">C++ classes</a></li>
<li><a href="#Perl5_nn21">C++ classes and type-checking</a></li>
<li><a href="#Perl5_nn22">C++ overloaded functions</a></li>
<li><a href="#Perl5_nn23">Operators</a></li>
<li><a href="#Perl5_nn24">Modules and packages</a></li>
</ul>
</li>
<li><a href="#Perl5_nn25">Input and output parameters</a></li>
<li><a href="#Perl5_nn26">Exception handling</a></li>
<li><a href="#Perl5_nn27">Remapping datatypes with typemaps</a>
<ul>
<li><a href="#Perl5_nn28">A simple typemap example</a></li>
<li><a href="#Perl5_nn29">Perl5 typemaps</a></li>
<li><a href="#Perl5_nn30">Typemap variables</a></li>
<li><a href="#Perl5_nn31">Useful functions</a></li>
</ul>
</li>
<li><a href="#Perl5_nn32">Typemap Examples</a>
<ul>
<li><a href="#Perl5_nn33">Converting a Perl5 array to a char **</a></li>
<li><a href="#Perl5_nn34">Return values</a></li>
<li><a href="#Perl5_nn35">Returning values from arguments</a></li>
<li><a href="#Perl5_nn36">Accessing array structure members</a></li>
<li><a href="#Perl5_nn37">Turning Perl references into C pointers</a></li>
<li><a href="#Perl5_nn38">Pointer handling</a></li>
</ul>
</li>
<li><a href="#Perl5_nn39">Proxy classes</a>
<ul>
<li><a href="#Perl5_nn40">Preliminaries</a></li>
<li><a href="#Perl5_nn41">Structure and class wrappers</a></li>
<li><a href="#Perl5_nn42">Object Ownership</a></li>
<li><a href="#Perl5_nn43">Nested Objects</a></li>
<li><a href="#Perl5_nn44">Proxy Functions</a></li>
<li><a href="#Perl5_nn45">Inheritance</a></li>
<li><a href="#Perl5_nn46">Modifying the proxy methods</a></li>
</ul>
</li>
<li><a href="#Perl5_nn47">Adding additional Perl code</a></li>
<li><a href="#Perl5_directors">Cross language polymorphism</a>
<ul>
<li><a href="#Perl5_nn48">Enabling directors</a></li>
<li><a href="#Perl5_nn49">Director classes</a></li>
<li><a href="#Perl5_nn50">Ownership and object destruction</a></li>
<li><a href="#Perl5_nn51">Exception unrolling</a></li>
<li><a href="#Perl5_nn52">Overhead and code bloat</a></li>
<li><a href="#Perl5_nn53">Typemaps</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p><b> Caution: This chapter is under repair!</b></p>
<p> This chapter describes SWIG's support of Perl5. Although the Perl5
module is one of the earliest SWIG modules, it has continued to evolve
and has been improved greatly with the help of SWIG users. As of SWIG
4.1.0, the minimum version of Perl we aim to support is Perl 5.8.0. We
can no longer easily test with older versions, and they no longer seem
to be in active use.</p>
<h2><a name="Perl5_nn2">31.1 Overview</a></h2>
<p> To build Perl extension modules, SWIG uses a layered approach. At
the lowest level, simple procedural wrappers are generated for
functions, classes, methods, and other declarations in the input file.
Then, for structures and classes, an optional collection of Perl proxy
classes can be generated in order to provide a more natural object
oriented Perl interface. These proxy classes simply build upon the
low-level interface.</p>
<p> In describing the Perl interface, this chapter begins by covering
the essentials. First, the problem of configuration, compiling, and
installing Perl modules is discussed. Next, the low-level procedural
interface is presented. Finally, proxy classes are described. Advanced
customization features, typemaps, and other options are found near the
end of the chapter.</p>
<h2><a name="Perl5_nn3">31.2 Preliminaries</a></h2>
<p> To build a Perl5 module, run SWIG using the <tt>-perl</tt> or <tt>
-perl5</tt> option as follows:</p>
<div class="code">
<pre>
swig -perl example.i
</pre>
</div>
<p> This produces two files. The first file, <tt>example_wrap.c</tt>
contains all of the C code needed to build a Perl5 module. The second
file, <tt>example.pm</tt> contains supporting Perl code needed to
properly load the module.</p>
<p> To build the module, you will need to compile the file <tt>
example_wrap.c</tt> and link it with the rest of your program.</p>
<h3><a name="Perl5_nn4">31.2.1 Getting the right header files</a></h3>
<p> In order to compile, SWIG extensions need the following Perl5 header
files:</p>
<div class="code">
<pre>
#include &quot;Extern.h&quot;
#include &quot;perl.h&quot;
#include &quot;XSUB.h&quot;
</pre>
</div>
<p> These are typically located in a directory like this</p>
<div class="code">
<pre>
/usr/lib/perl/5.14/CORE
</pre>
</div>
<p> The SWIG configuration script automatically tries to locate this
directory so that it can compile examples. However, if you need to find
out where the directory is located, an easy way to find out is to ask
Perl itself:</p>
<div class="code">
<pre>
$ perl -e 'use Config; print &quot;$Config{archlib}\n&quot;;'
/usr/lib/perl/5.14
</pre>
</div>
<h3><a name="Perl5_nn5">31.2.2 Compiling a dynamic module</a></h3>
<p> The preferred approach to building an extension module is to compile
it into a shared object file or DLL. Assuming you have code you need to
link to in a file called <tt>example.c</tt>, you will need to compile
your program using commands like this (shown for Linux):</p>
<div class="code">
<pre>
$ swig -perl example.i
$ gcc -fPIC example.c
$ gcc -fPIC -c example_wrap.c -I/usr/lib/perl/5.14/CORE -Dbool=char
$ gcc -shared example.o example_wrap.o -o example.so
</pre>
</div>
<p> The exact compiler options vary from platform to platform. SWIG
tries to guess the right options when it is installed. Therefore, you
may want to start with one of the examples in the <tt>
SWIG/Examples/perl5</tt> directory. If that doesn't work, you will need
to read the man-pages for your compiler and linker to get the right set
of options. You might also check the <a href="https://github.com/swig/swig/wiki">
SWIG Wiki</a> for additional information.</p>
<p> When linking the module, the name of the shared object file must
match the module name used in the SWIG interface file. If you used `<tt>
%module example</tt>', then the target should be named `<tt>example.so</tt>
', `<tt>example.sl</tt>', or the appropriate dynamic module name on your
system.</p>
<h3><a name="Perl5_nn6">31.2.3 Building a dynamic module with MakeMaker</a>
</h3>
<p> It is also possible to use Perl to build dynamically loadable
modules for you using the MakeMaker utility. To do this, write a Perl
script such as the following:</p>
<div class="targetlang">
<pre>
# File : Makefile.PL
use ExtUtils::MakeMaker;
WriteMakefile(
`NAME' =&gt; `example', # Name of package
`LIBS' =&gt; [`-lm'], # Name of custom libraries
`OBJECT' =&gt; `example.o example_wrap.o' # Object files
);
</pre>
</div>
<p> Now, to build a module, simply follow these steps:</p>
<div class="code">
<pre>
$ perl Makefile.PL
$ make
$ make install
</pre>
</div>
<p> If you are planning to distribute a SWIG-generated module, this is
the preferred approach to compilation. More information about MakeMaker
can be found in &quot;Programming Perl, 2nd ed.&quot; by Larry Wall, Tom
Christiansen, and Randal Schwartz.</p>
<h3><a name="Perl5_nn7">31.2.4 Building a static version of Perl</a></h3>
<p> If you machine does not support dynamic loading or if you've tried
to use it without success, you can build a new version of the Perl
interpreter with your SWIG extensions added to it. To build a static
extension, you first need to invoke SWIG as follows:</p>
<div class="code">
<pre>
$ swig -perl -static example.i
</pre>
</div>
<p> By default SWIG includes code for dynamic loading, but the <tt>
-static</tt> option takes it out.</p>
<p> Next, you will need to supply a <tt>main()</tt> function that
initializes your extension and starts the Perl interpreter. While, this
may sound daunting, SWIG can do this for you automatically as follows:</p>
<div class="targetlang">
<pre>
%module example
%inline %{
extern double My_variable;
extern int fact(int);
%}
// Include code for rebuilding Perl
%include &lt;perlmain.i&gt;
</pre>
</div>
<p> The same thing can be accomplished by running SWIG as follows:</p>
<div class="code">
<pre>
$ swig -perl -static -lperlmain.i example.i
</pre>
</div>
<p> The <tt>perlmain.i</tt> file inserts Perl's <tt>main()</tt> function
into the wrapper code and automatically initializes the SWIG generated
module. If you just want to make a quick a dirty module, this may be
the easiest way. By default, the <tt>perlmain.i</tt> code does not
initialize any other Perl extensions. If you need to use other
packages, you will need to modify it appropriately. You can do this by
just copying <tt>perlmain.i</tt> out of the SWIG library, placing it in
your own directory, and modifying it to suit your purposes.</p>
<p> To build your new Perl executable, follow the exact same procedure
as for a dynamic module, but change the link line to something like
this:</p>
<div class="code">
<pre>
$ gcc example.o example_wrap.o -L/usr/lib/perl/5.14/CORE \
-lperl -lsocket -lnsl -lm -o myperl
</pre>
</div>
<p> This will produce a new version of Perl called <tt>myperl</tt>. It
should be functionality identical to Perl with your C/C++ extension
added to it. Depending on your machine, you may need to link with
additional libraries such as <tt>-lsocket, -lnsl, -ldl</tt>, etc.</p>
<h3><a name="Perl5_nn8">31.2.5 Using the module</a></h3>
<p> To use the module, simply use the Perl <tt>use</tt> statement. If
all goes well, you will be able to do this:</p>
<div class="targetlang">
<pre>
$ perl
use example;
print example::fact(4), &quot;\n&quot;;
24
</pre>
</div>
<p> A common error received by first-time users is the following:</p>
<div class="targetlang">
<pre>
use example;
Can't locate example.pm in @INC (@INC contains: /etc/perl /usr/local/lib/perl/5.14.2 /usr/local/share/perl/5.14.2 /usr/lib/perl5 /usr/share/perl5 /usr/lib/perl/5.14 /usr/share/perl/5.14 /usr/local/lib/site_perl .) at - line 1.
BEGIN failed--compilation aborted at - line 1.
</pre>
</div>
<p> This error is almost caused when the name of the shared object file
you created doesn't match the module name you specified with the <tt>
%module</tt> directive.</p>
<p> A somewhat related, but slightly different error is this:</p>
<div class="targetlang">
<pre>
use example;
Can't find 'boot_example' symbol in ./example.so
at - line 1
BEGIN failed--compilation aborted at - line 1.
</pre>
</div>
<p> This error is generated because Perl can't locate the module
bootstrap function in the SWIG extension module. This could be caused
by a mismatch between the module name and the shared library name.
However, another possible cause is forgetting to link the
SWIG-generated wrapper code with the rest of your application when you
linked the extension module.</p>
<p> Another common error is the following:</p>
<div class="targetlang">
<pre>
use example;
Can't load './example.so' for module example: ./example.so:
undefined symbol: Foo at /usr/lib/perl/5.14/i386-linux/DynaLoader.pm line 169.
at - line 1
BEGIN failed--compilation aborted at - line 1.
</pre>
</div>
<p> This error usually indicates that you forgot to include some object
files or libraries in the linking of the shared library file. Make sure
you compile both the SWIG wrapper file and your original program into a
shared library file. Make sure you pass all of the required libraries
to the linker.</p>
<p> Sometimes unresolved symbols occur because a wrapper has been
created for a function that doesn't actually exist in a library. This
usually occurs when a header file includes a declaration for a function
that was never actually implemented or it was removed from a library
without updating the header file. To fix this, you can either edit the
SWIG input file to remove the offending declaration or you can use the <tt>
%ignore</tt> directive to ignore the declaration. Better yet, update the
header file so that it doesn't have an undefined declaration.</p>
<p> Finally, suppose that your extension module is linked with another
library like this:</p>
<div class="code">
<pre>
$ gcc -shared example.o example_wrap.o -L/home/beazley/projects/lib -lfoo \
-o example.so
</pre>
</div>
<p> If the <tt>foo</tt> library is compiled as a shared library, you
might get the following error when you try to use your module:</p>
<div class="targetlang">
<pre>
use example;
Can't load './example.so' for module example: libfoo.so: cannot open shared object file:
No such file or directory at /usr/lib/perl/5.14/i386-linux/DynaLoader.pm line 169.
at - line 1
BEGIN failed--compilation aborted at - line 1.
&gt;&gt;&gt;
</pre>
</div>
<p> This error is generated because the dynamic linker can't locate the <tt>
libfoo.so</tt> library. When shared libraries are loaded, the system
normally only checks a few standard locations such as <tt>/usr/lib</tt>
and <tt>/usr/local/lib</tt>. To get the loader to look in other
locations, there are several things you can do. First, you can
recompile your extension module with extra path information. For
example, on Linux you can do this:</p>
<div class="code">
<pre>
$ gcc -shared example.o example_wrap.o -L/home/beazley/projects/lib -lfoo \
<b>-Xlinker -rpath /home/beazley/projects/lib \</b>
-o example.so
</pre>
</div>
<p> Alternatively, you can set the <tt>LD_LIBRARY_PATH</tt> environment
variable to include the directory with your shared libraries. If
setting <tt>LD_LIBRARY_PATH</tt>, be aware that setting this variable
can introduce a noticeable performance impact on all other applications
that you run. To set it only for Perl, you might want to do this
instead:</p>
<div class="code">
<pre>
$ env LD_LIBRARY_PATH=/home/beazley/projects/lib perl
</pre>
</div>
<p> Finally, you can use a command such as <tt>ldconfig</tt> (Linux) or <tt>
crle</tt> (Solaris) to add additional search paths to the default system
configuration (this requires root access and you will need to read the
man pages).</p>
<h3><a name="Perl5_nn9">31.2.6 Compilation problems and compiling with
C++</a></h3>
<p> Compilation of C++ extensions has traditionally been a tricky
problem. Since the Perl interpreter is written in C, you need to take
steps to make sure C++ is properly initialized and that modules are
compiled correctly.</p>
<p> On most machines, C++ extension modules should be linked using the
C++ compiler. For example:</p>
<div class="code">
<pre>
$ swig -c++ -perl example.i
$ g++ -fPIC -c example.cxx
$ g++ -fPIC -c example_wrap.cxx -I/usr/lib/perl/5.14/i386-linux/CORE
$ <b>g++ -shared example.o example_wrap.o -o example.so</b>
</pre>
</div>
<p> In addition to this, you may need to include additional library
files to make it work. For example, if you are using the Sun C++
compiler on Solaris, you often need to add an extra library <tt>-lCrun</tt>
like this:</p>
<div class="code">
<pre>
$ swig -c++ -perl example.i
$ CC -Kpic -c example.cxx
$ CC -Kpic -c example_wrap.cxx -I/usr/lib/perl/5.14/i386-linux/CORE
$ CC -shared example.o example_wrap.o -o example.so <b>-lCrun</b>
</pre>
</div>
<p> Of course, the names of the extra libraries are completely
non-portable---you will probably need to do some experimentation.</p>
<p> Another possible compile problem comes from recent versions of Perl
(5.8.0) and the GNU tools. If you see errors having to do with
_crypt_struct, that means _GNU_SOURCE is not defined and it needs to
be. So you should compile the wrapper like:</p>
<div class="code">
<pre>
$ g++ -fPIC -c example_wrap.cxx -I/usr/lib/perl/5.8.0/CORE -D_GNU_SOURCE
</pre>
</div>
<p> -D_GNU_SOURCE is also included in the Perl ccflags, which can be
found by running</p>
<div class="code">
<pre>
$ perl -e 'use Config; print &quot;$Config{ccflags}\n&quot;;'
</pre>
</div>
<p> So you could also compile the wrapper like</p>
<div class="code">
<pre>
$ g++ -fPIC -c example_wrap.cxx -I/usr/lib/perl/5.8.0/CORE \
`perl -MConfig -e 'print $Config{ccflags}'`
</pre>
</div>
<p> Sometimes people have suggested that it is necessary to relink the
Perl interpreter using the C++ compiler to make C++ extension modules
work. In the experience of this author, this has never actually
appeared to be necessary on most platforms. Relinking the interpreter
with C++ really only includes the special run-time libraries described
above---as long as you link your extension modules with these
libraries, it should not be necessary to rebuild Perl.</p>
<p> If you aren't entirely sure about the linking of a C++ extension,
you might look at an existing C++ program. On many Unix machines, the <tt>
ldd</tt> command will list library dependencies. This should give you
some clues about what you might have to include when you link your
extension module. For example, notice the first line of output here:</p>
<div class="code">
<pre>
$ ldd swig
<b>libstdc++-libc6.1-1.so.2 =&gt; /usr/lib/libstdc++-libc6.1-1.so.2 (0x40019000)</b>
libm.so.6 =&gt; /lib/libm.so.6 (0x4005b000)
libc.so.6 =&gt; /lib/libc.so.6 (0x40077000)
/lib/ld-linux.so.2 =&gt; /lib/ld-linux.so.2 (0x40000000)
$
</pre>
</div>
<p> If linking wasn't enough of a problem, another major complication of
C++ is that it does not define any sort of standard for binary linking
of libraries. This means that C++ code compiled by different compilers
will not link together properly as libraries nor is the memory layout
of classes and data structures implemented in any kind of portable
manner. In a monolithic C++ program, this problem may be unnoticed.
However, in Perl, it is possible for different extension modules to be
compiled with different C++ compilers. As long as these modules are
self-contained, this probably won't matter. However, if these modules
start sharing data, you will need to take steps to avoid segmentation
faults and other erratic program behavior. Also, be aware that certain
C++ features, especially RTTI, can behave strangely when working with
multiple modules.</p>
<p> It should be noted that you may get a lot of error messages about
the '<tt>bool</tt>' datatype when compiling a C++ Perl module. If you
experience this problem, you can try the following:</p>
<ul>
<li>Use <tt>-DHAS_BOOL</tt> when compiling the SWIG wrapper code</li>
<li>Or use <tt>-Dbool=char</tt> when compiling.</li>
</ul>
<p> Finally, recent versions of Perl (5.8.0) have namespace conflict
problems. Perl defines a bunch of short macros to make the Perl API
function names shorter. For example, in
/usr/lib/perl/5.8.0/CORE/embed.h there is a line:</p>
<div class="code">
<pre>
#define do_open Perl_do_open
</pre>
</div>
<p> The problem is, in the &lt;iostream&gt; header from GNU libstdc++v3 there
is a private function named do_open. If &lt;iostream&gt; is included after
the perl headers, then the Perl macro causes the iostream do_open to be
renamed, which causes compile errors. Hopefully in the future Perl will
support a PERL_NO_SHORT_NAMES flag, but for now the only solution is to
undef the macros that conflict. Lib/perl5/noembed.h in the SWIG source
has a list of macros that are known to conflict with either standard
headers or other headers. But if you get macro type conflicts from
other macros not included in Lib/perl5/noembed.h while compiling the
wrapper, you will have to find the macro that conflicts and add an
#undef into the .i file. Please report any conflicting macros you find
to <a href="https://www.swig.org/mail.html">swig-user mailing list</a>.</p>
<h3><a name="Perl5_nn10">31.2.7 Compiling for 64-bit platforms</a></h3>
<p> On platforms that support 64-bit applications (Solaris, Irix, etc.),
special care is required when building extension modules. On these
machines, 64-bit applications are compiled and linked using a different
set of compiler/linker options. In addition, it is not generally
possible to mix 32-bit and 64-bit code together in the same
application.</p>
<p> To utilize 64-bits, the Perl executable will need to be recompiled
as a 64-bit application. In addition, all libraries, wrapper code, and
every other part of your application will need to be compiled for
64-bits. If you plan to use other third-party extension modules, they
will also have to be recompiled as 64-bit extensions.</p>
<p> If you are wrapping commercial software for which you have no source
code, you will be forced to use the same linking standard as used by
that software. This may prevent the use of 64-bit extensions. It may
also introduce problems on platforms that support more than one linking
standard (e.g., -o32 and -n32 on Irix).</p>
<h2><a name="Perl5_nn11">31.3 Building Perl Extensions under Windows</a></h2>
<p> Building a SWIG extension to Perl under Windows is roughly similar
to the process used with Unix. Normally, you will want to produce a DLL
that can be loaded into the Perl interpreter. This section assumes you
are using SWIG with Microsoft Visual C++ although the procedure may be
similar with other compilers.</p>
<h3><a name="Perl5_nn12">31.3.1 Running SWIG from Developer Studio</a></h3>
<p> If you are developing your application within Microsoft developer
studio, SWIG can be invoked as a custom build option. The process
roughly requires these steps:</p>
<ul>
<li>Open up a new workspace and use the AppWizard to select a DLL
project.</li>
<li>Add both the SWIG interface file (the .i file), any supporting C
files, and the name of the wrapper file that will be created by SWIG
(ie. <tt>example_wrap.c</tt>). Note: If using C++, choose a different
suffix for the wrapper file such as <tt>example_wrap.cxx</tt>. Don't
worry if the wrapper file doesn't exist yet--Developer studio will keep
a reference to it around.</li>
<li>Select the SWIG interface file and go to the settings menu. Under
settings, select the &quot;Custom Build&quot; option.</li>
<li>Enter &quot;SWIG&quot; in the description field.</li>
<li>Enter &quot;<tt>swig -perl5 -o $(ProjDir)\$(InputName)_wrap.cxx
$(InputPath)</tt>&quot; in the &quot;Build command(s) field&quot;</li>
<li>Enter &quot;<tt>$(ProjDir)\$(InputName)_wrap.c</tt>xx&quot; in the &quot;Output
files(s) field&quot;.</li>
<li>Next, select the settings for the entire project and go to
&quot;C++:Preprocessor&quot;. Add the include directories for your Perl 5
installation under &quot;Additional include directories&quot;.</li>
<li>Define the symbols WIN32 and MSWIN32 under preprocessor options.
Note that all extensions to the ActiveWare port must be compiled with
the C++ compiler since Perl has been encapsulated in a C++ class.</li>
<li>Finally, select the settings for the entire project and go to &quot;Link
Options&quot;. Add the Perl library file to your link libraries. For example
&quot;perl.lib&quot;. Also, set the name of the output file to match the name of
your Perl module (ie. example.dll).</li>
<li>Build your project.</li>
</ul>
<p> Now, assuming you made it this far, SWIG will be automatically
invoked when you build your project. Any changes made to the interface
file will result in SWIG being automatically invoked to produce a new
version of the wrapper file. To run your new Perl extension, simply run
Perl and use the use command as normal. For example:</p>
<div class="targetlang">
<pre>
DOS &gt; perl
use example;
$a = example::fact(4);
print &quot;$a\n&quot;;
</pre>
</div>
<h3><a name="Perl5_nn13">31.3.2 Using other compilers</a></h3>
<p> SWIG is known to work with Cygwin and may work with other compilers
on Windows. For general hints and suggestions refer to the <a href="#Windows">
Windows</a> chapter.</p>
<h2><a name="Perl5_nn14">31.4 The low-level interface</a></h2>
<p> At its core, the Perl module uses a simple low-level interface to C
function, variables, constants, and classes. This low-level interface
can be used to control your application. However, it is also used to
construct more user-friendly proxy classes as described in the next
section.</p>
<h3><a name="Perl5_nn15">31.4.1 Functions</a></h3>
<p> C functions are converted into new Perl built-in commands (or
subroutines). For example:</p>
<div class="targetlang">
<pre>
%module example
int fact(int a);
...
</pre>
</div>
<p> Now, in Perl:</p>
<div class="targetlang">
<pre>
use example;
$a = &amp;example::fact(2);
</pre>
</div>
<h3><a name="Perl5_nn16">31.4.2 Global variables</a></h3>
<p> Global variables are handled using Perl's magic variable mechanism.
SWIG generates a pair of functions that intercept read/write operations
and attaches them to a Perl variable with the same name as the C global
variable. Thus, an interface like this</p>
<div class="targetlang">
<pre>
%module example;
...
double Spam;
...
</pre>
</div>
<p> is accessed as follows:</p>
<div class="targetlang">
<pre>
use example;
print $example::Spam, &quot;\n&quot;;
$example::Spam = $example::Spam + 4
# ... etc ...
</pre>
</div>
<p> If a variable is declared as <tt>const</tt>, it is wrapped as a
read-only variable. Attempts to modify its value will result in an
error.</p>
<p> To make ordinary variables read-only, you can also use the <tt>
%immutable</tt> directive. For example:</p>
<div class="code">
<pre>
%{
extern char *path;
%}
%immutable;
extern char *path;
%mutable;
</pre>
</div>
<p> The <tt>%immutable</tt> directive stays in effect until it is
explicitly disabled or cleared using <tt>%mutable</tt>. See the <a href="#SWIG_readonly_variables">
Creating read-only variables</a> section for further details.</p>
<p> It is also possible to tag a specific variable as read-only like
this:</p>
<div class="code">
<pre>
%{
extern char *path;
%}
%immutable path;
...
...
extern char *path; // Declared later in the input
</pre>
</div>
<h3><a name="Perl5_nn17">31.4.3 Constants</a></h3>
<p> By default, constants are wrapped as read-only Perl variables. For
example:</p>
<div class="code">
<pre>
%module example
#define FOO 42
</pre>
</div>
<p> In Perl:</p>
<div class="targetlang">
<pre>
use example;
print $example::FOO, &quot;\n&quot;; # OK
$example::FOO = 2; # Error
</pre>
</div>
<p> Alternatively, if you use swig's <tt>-const</tt> option, constants
are wrapped such that the leading $ isn't required (by using a constant
subroutine), which usually gives a more natural Perl interface, for
example:</p>
<div class="targetlang">
<pre>
use example;
print example::FOO, &quot;\n&quot;;
</pre>
</div>
<h3><a name="Perl5_nn18">31.4.4 Pointers</a></h3>
<p> SWIG represents pointers as blessed references. A blessed reference
is the same as a Perl reference except that it has additional
information attached to it indicating what kind of reference it is.
That is, if you have a C declaration like this:</p>
<div class="code">
<pre>
Matrix *new_Matrix(int n, int m);
</pre>
</div>
<p> The module returns a value generated as follows:</p>
<div class="targetlang">
<pre>
$ptr = new_Matrix(int n, int m); # Save pointer return result
bless $ptr, &quot;p_Matrix&quot;; # Bless it as a pointer to Matrix
</pre>
</div>
<p> SWIG uses the &quot;blessing&quot; to check the datatype of various pointers.
In the event of a mismatch, an error or warning message is generated.</p>
<p> To check to see if a value is the NULL pointer, use the <tt>
defined()</tt> command:</p>
<div class="targetlang">
<pre>
if (defined($ptr)) {
print &quot;Not a NULL pointer.&quot;;
} else {
print &quot;Is a NULL pointer.&quot;;
}
</pre>
</div>
<p> To create a NULL pointer, you should pass the <tt>undef</tt> value
to a function.</p>
<p> The &quot;value&quot; of a Perl reference is not the same as the underlying C
pointer that SWIG wrapper functions return. Suppose that <tt>$a</tt>
and <tt>$b</tt> are two references that point to the same C object. In
general, <tt>$a</tt> and <tt>$b</tt> will be different--since they are
different references. Thus, it is a mistake to check the equality of <tt>
$a</tt> and <tt>$b</tt> to check the equality of two C pointers. The
correct method to check equality of C pointers is to dereference them
as follows:</p>
<div class="targetlang">
<pre>
if ($$a == $$b) {
print &quot;a and b point to the same thing in C&quot;;
} else {
print &quot;a and b point to different objects.&quot;;
}
</pre>
</div>
<p> As much as you might be inclined to modify a pointer value directly
from Perl, don't. Manipulating pointer values is architecture dependent
and could cause your program to crash. Similarly, don't try to manually
cast a pointer to a new type by reblessing a pointer. This may not work
like you expect and it is particularly dangerous when casting C++
objects. If you need to cast a pointer or change its value, consider
writing some helper functions instead. For example:</p>
<div class="code">
<pre>
%inline %{
/* C-style cast */
Bar *FooToBar(Foo *f) {
return (Bar *) f;
}
/* C++-style cast */
Foo *BarToFoo(Bar *b) {
return dynamic_cast&lt;Foo*&gt;(b);
}
Foo *IncrFoo(Foo *f, int i) {
return f+i;
}
%}
</pre>
</div>
<p> Also, if working with C++, you should always try to use the new C++
style casts. For example, in the above code, the C-style cast may
return a bogus result whereas as the C++-style cast will return <tt>
NULL</tt> if the conversion can't be performed.</p>
<p><b> Compatibility Note:</b> In earlier versions, SWIG tried to
preserve the same pointer naming conventions as XS and <tt>xsubpp</tt>.
Given the advancement of the SWIG typesystem and the growing
differences between SWIG and XS, this is no longer supported.</p>
<h3><a name="Perl5_nn19">31.4.5 Structures</a></h3>
<p> Access to the contents of a structure are provided through a set of
low-level accessor functions as described in the &quot;SWIG Basics&quot; chapter.
For example,</p>
<div class="code">
<pre>
struct Vector {
double x, y, z;
};
</pre>
</div>
<p> gets mapped into the following collection of accessor functions:</p>
<div class="code">
<pre>
struct Vector *new_Vector();
void delete_Vector(Vector *v);
double Vector_x_get(Vector *obj)
void Vector_x_set(Vector *obj, double x)
double Vector_y_get(Vector *obj)
void Vector_y_set(Vector *obj, double y)
double Vector_z_get(Vector *obj)
void Vector_z_set(Vector *obj, double z)
</pre>
</div>
<p> These functions are then used to access structure data from Perl as
follows:</p>
<div class="targetlang">
<pre>
$v = example::new_Vector();
print example::Vector_x_get($v), &quot;\n&quot;; # Get x component
example::Vector_x_set($v, 7.8); # Change x component
</pre>
</div>
<p> Similar access is provided for unions and the data members of C++
classes.</p>
<p> <tt>const</tt> members of a structure are read-only. Data members
can also be forced to be read-only using the <tt>%immutable</tt>
directive. For example:</p>
<div class="code">
<pre>
struct Foo {
...
%immutable;
int x; /* Read-only members */
char *name;
%mutable;
...
};
</pre>
</div>
<p> When <tt>char *</tt> members of a structure are wrapped, the
contents are assumed to be dynamically allocated using <tt>malloc</tt>
or <tt>new</tt> (depending on whether or not SWIG is run with the -c++
option). When the structure member is set, the old contents will be
released and a new value created. If this is not the behavior you want,
you will have to use a typemap (described later).</p>
<p> Array members are normally wrapped as read-only. For example,</p>
<div class="code">
<pre>
struct Foo {
int x[50];
};
</pre>
</div>
<p> produces a single accessor function like this:</p>
<div class="code">
<pre>
int *Foo_x_get(Foo *self) {
return self-&gt;x;
};
</pre>
</div>
<p> If you want to set an array member, you will need to supply a
&quot;memberin&quot; typemap described later in this chapter. As a special case,
SWIG does generate code to set array members of type <tt>char</tt>
(allowing you to store a Perl string in the structure).</p>
<p> When structure members are wrapped, they are handled as pointers.
For example,</p>
<div class="code">
<pre>
struct Foo {
...
};
struct Bar {
Foo f;
};
</pre>
</div>
<p> generates accessor functions such as this:</p>
<div class="code">
<pre>
Foo *Bar_f_get(Bar *b) {
return &amp;b-&gt;f;
}
void Bar_f_set(Bar *b, Foo *val) {
b-&gt;f = *val;
}
</pre>
</div>
<h3><a name="Perl5_nn20">31.4.6 C++ classes</a></h3>
<p> C++ classes are wrapped by building a set of low level accessor
functions. Consider the following class:</p>
<div class="code">
<pre>
class List {
public:
List();
~List();
int search(char *item);
void insert(char *item);
void remove(char *item);
char *get(int n);
int length;
static void print(List *l);
};
</pre>
</div>
<p> When wrapped by SWIG, the following functions are created:</p>
<div class="code">
<pre>
List *new_List();
void delete_List(List *l);
int List_search(List *l, char *item);
void List_insert(List *l, char *item);
void List_remove(List *l, char *item);
char *List_get(List *l, int n);
int List_length_get(List *l);
void List_length_set(List *l, int n);
void List_print(List *l);
</pre>
</div>
<p> In Perl, these functions are used in a straightforward manner:</p>
<div class="targetlang">
<pre>
use example;
$l = example::new_List();
example::List_insert($l, &quot;Ale&quot;);
example::List_insert($l, &quot;Stout&quot;);
example::List_insert($l, &quot;Lager&quot;)
example::List_print($l)
Lager
Stout
Ale
print example::List_length_get($l), &quot;\n&quot;;
3
</pre>
</div>
<p> At this low level, C++ objects are really just typed pointers.
Member functions are accessed by calling a C-like wrapper with an
instance pointer as the first argument. Although this interface is
fairly primitive, it provides direct access to C++ objects. A higher
level interface using Perl proxy classes can be built using these
low-level accessors. This is described shortly.</p>
<h3><a name="Perl5_nn21">31.4.7 C++ classes and type-checking</a></h3>
<p> The SWIG type-checker is fully aware of C++ inheritance. Therefore,
if you have classes like this</p>
<div class="code">
<pre>
class Foo {
...
};
class Bar : public Foo {
...
};
</pre>
</div>
<p> and a function</p>
<div class="code">
<pre>
void spam(Foo *f);
</pre>
</div>
<p> then the function <tt>spam()</tt> accepts <tt>Foo *</tt> or a
pointer to any class derived from <tt>Foo</tt>. If necessary, the
type-checker also adjusts the value of the pointer (as is necessary
when multiple inheritance is used).</p>
<h3><a name="Perl5_nn22">31.4.8 C++ overloaded functions</a></h3>
<p> If you have a C++ program with overloaded functions or methods, you
will need to disambiguate those methods using <tt>%rename</tt>. For
example:</p>
<div class="code">
<pre>
/* Forward renaming declarations */
%rename(foo_i) foo(int);
%rename(foo_d) foo(double);
...
void foo(int); // Becomes 'foo_i'
void foo(char *c); // Stays 'foo' (not renamed)
class Spam {
public:
void foo(int); // Becomes 'foo_i'
void foo(double); // Becomes 'foo_d'
...
};
</pre>
</div>
<p> Now, in Perl, the methods are accessed as follows:</p>
<div class="targetlang">
<pre>
use example;
example::foo_i(3);
$s = example::new_Spam();
example::Spam_foo_i($s, 3);
example::Spam_foo_d($s, 3.14);
</pre>
</div>
<p> Please refer to the &quot;SWIG Basics&quot; chapter for more information.</p>
<h3><a name="Perl5_nn23">31.4.9 Operators</a></h3>
<p> As of version 1.3.27 SWIG automatically renames the most common C++
operators, and maps them into the perl module with the proper 'use
overload ...' so you don't need to do any work.</p>
<p> The following C++ operators are currently supported by the Perl
module:</p>
<ul>
<li>operator++</li>
<li>operator--</li>
<li>operator+</li>
<li>operator-</li>
<li>operator*</li>
<li>operator/</li>
<li>operator==</li>
<li>operator!=</li>
<li>operator%</li>
<li>operator&gt;</li>
<li>operator&lt;</li>
<li>operator and</li>
<li>operator or</li>
</ul>
<h3><a name="Perl5_nn24">31.4.10 Modules and packages</a></h3>
<p> When you create a SWIG extension, everything gets placed into a
single Perl module. The name of the module is determined by the <tt>
%module</tt> directive. To use the module, do the following:</p>
<div class="targetlang">
<pre>
$ perl5
use example; # load the example module
print example::fact(4), &quot;\n&quot; # Call a function in it
24
</pre>
</div>
<p> Usually, a module consists of a collection of code that is contained
within a single file. A package, on the other hand, is the Perl
equivalent of a namespace. A package is a lot like a module, except
that it is independent of files. Any number of files may be part of the
same package--or a package may be broken up into a collection of
modules if you prefer to think about it in this way.</p>
<p> SWIG installs its functions into a package with the same name as the
module.</p>
<p><b> Incompatible Change:</b> previous versions of SWIG enabled you to
change the name of the package by using the -package option, this
feature has been removed in order to properly support modules that used
nested namespaces, e.g. Foo::Bar::Baz. To give your module a nested
namespace simply provide the fully qualified name in your %module
directive:</p>
<div class="code">
<pre>
%module &quot;Foo::Bar::Baz&quot;
</pre>
</div>
<p><b> NOTE:</b> the double quotes are necessary.</p>
<p> Using the <tt>package</tt> option of the <tt>%module</tt> directive
allows you to specify what Perl namespace that the module will be
living in when installed. This is useful in the situation where a
module maintainer wants to split a large module into smaller pieces to
make maintenance easier, but doesn't want to have that affect the
module name used by applications. So for example, if I wanted to split <tt>
XML::Xerces</tt> into <tt>XML::Xerces::SAX</tt>, etc. , but I wanted all
the applications to be able to access the classes using the <tt>
XML::Xerces</tt> namespace I could use:</p>
<div class="code">
<pre>
%module(package=&quot;XML::Xerces&quot;) &quot;XML::Xerces::SAX
</pre>
</div>
<p> And now all the applications could use the class <tt>
XML::Xerces::SAXParser</tt>. Without the <tt>package</tt> directive
splitting the module would force applications to use the class <tt>
XML::Xerces::SAX::SAXParser</tt>. This could break compatibility for
existing applications that are already using the class under the name <tt>
XML::Xerces::SAXParser</tt>.</p>
<!--
&lt;p&gt;
This can be changed by giving SWIG the -package
option:
&lt;/p&gt;
&lt;div class=&quot;code&quot;&gt;&lt;pre&gt;
$ swig -perl -package Foo example.i
&lt;/pre&gt;&lt;/div&gt;
&lt;p&gt;
In this case, you still create a module called `&lt;tt&gt;example&lt;/tt&gt;' exactly as before, but
all of the functions in that module will be installed into the package
`&lt;tt&gt;Foo&lt;/tt&gt;.' For example:
&lt;/p&gt;
&lt;div class=&quot;targetlang&quot;&gt;&lt;pre&gt;
use example; # Load the module like before
print Foo::fact(4), &quot;\n&quot;; # Call a function in package FooBar
&lt;/pre&gt;&lt;/div&gt;
-->
<h2><a name="Perl5_nn25">31.5 Input and output parameters</a></h2>
<p> A common problem in some C programs is handling parameters passed as
simple pointers. For example:</p>
<div class="code">
<pre>
void add(int x, int y, int *result) {
*result = x + y;
}
</pre>
</div>
<p> or perhaps</p>
<div class="code">
<pre>
int sub(int *x, int *y) {
return *x+*y;
}
</pre>
</div>
<p> The easiest way to handle these situations is to use the <tt>
typemaps.i</tt> file. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
void add(int, int, int *OUTPUT);
int sub(int *INPUT, int *INPUT);
</pre>
</div>
<p> In Perl, this allows you to pass simple values. For example:</p>
<div class="targetlang">
<pre>
$a = example::add(3, 4);
print &quot;$a\n&quot;;
7
$b = example::sub(7, 4);
print &quot;$b\n&quot;;
3
</pre>
</div>
<p> Notice how the <tt>INPUT</tt> parameters allow integer values to be
passed instead of pointers and how the <tt>OUTPUT</tt> parameter
creates a return result.</p>
<p> If you don't want to use the names <tt>INPUT</tt> or <tt>OUTPUT</tt>
, use the <tt>%apply</tt> directive. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *result };
%apply int *INPUT { int *x, int *y};
void add(int x, int y, int *result);
int sub(int *x, int *y);
</pre>
</div>
<p> If a function mutates one of its parameters like this,</p>
<div class="code">
<pre>
void negate(int *x) {
*x = -(*x);
}
</pre>
</div>
<p> you can use <tt>INOUT</tt> like this:</p>
<div class="code">
<pre>
%include &quot;typemaps.i&quot;
...
void negate(int *INOUT);
</pre>
</div>
<p> In Perl, a mutated parameter shows up as a return value. For
example:</p>
<div class="targetlang">
<pre>
$a = example::negate(3);
print &quot;$a\n&quot;;
-3
</pre>
</div>
<p> The most common use of these special typemap rules is to handle
functions that return more than one value. For example, sometimes a
function returns a result as well as a special error code:</p>
<div class="code">
<pre>
/* send message, return number of bytes sent, along with success code */
int send_message(char *text, int *success);
</pre>
</div>
<p> To wrap such a function, simply use the <tt>OUTPUT</tt> rule above.
For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *success };
...
int send_message(char *text, int *success);
</pre>
</div>
<p> When used in Perl, the function will return multiple values.</p>
<div class="targetlang">
<pre>
($bytes, $success) = example::send_message(&quot;Hello World&quot;);
</pre>
</div>
<p> Another common use of multiple return values are in query functions.
For example:</p>
<div class="code">
<pre>
void get_dimensions(Matrix *m, int *rows, int *columns);
</pre>
</div>
<p> To wrap this, you might use the following:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *rows, int *columns };
...
void get_dimensions(Matrix *m, int *rows, *columns);
</pre>
</div>
<p> Now, in Perl:</p>
<div class="targetlang">
<pre>
($r, $c) = example::get_dimensions($m);
</pre>
</div>
<p> In certain cases, it is possible to treat Perl references as C
pointers. To do this, use the <tt>REFERENCE</tt> typemap. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
void add(int x, int y, int *REFERENCE);
</pre>
</div>
<p> In Perl:</p>
<div class="targetlang">
<pre>
use example;
$c = 0.0;
example::add(3, 4, \$c);
print &quot;$c\n&quot;;
7
</pre>
</div>
<p><b> Note:</b> The <tt>REFERENCE</tt> feature is only currently
supported for numeric types (integers and floating point).</p>
<h2><a name="Perl5_nn26">31.6 Exception handling</a></h2>
<p> The SWIG <tt>%exception</tt> directive can be used to create a
user-definable exception handler for converting exceptions in your
C/C++ program into Perl exceptions. The chapter on customization
features contains more details, but suppose you have a C++ class like
the following:</p>
<div class="code">
<pre>
class RangeError {}; // Used for an exception
class DoubleArray {
private:
int n;
double *ptr;
public:
// Create a new array of fixed size
DoubleArray(int size) {
ptr = new double[size];
n = size;
}
// Destroy an array
~DoubleArray() {
delete ptr;
}
// Return the length of the array
int length() {
return n;
}
// Get an item from the array and perform bounds checking.
double getitem(int i) {
if ((i &gt;= 0) &amp;&amp; (i &lt; n))
return ptr[i];
else
throw RangeError();
}
// Set an item in the array and perform bounds checking.
void setitem(int i, double val) {
if ((i &gt;= 0) &amp;&amp; (i &lt; n))
ptr[i] = val;
else
throw RangeError();
}
};
</pre>
</div>
<p> Since several methods in this class can throw an exception for an
out-of-bounds access, you might want to catch this in the Perl
extension by writing the following in an interface file:</p>
<div class="code">
<pre>
%exception {
try {
$action
}
catch (RangeError) {
croak(&quot;Array index out-of-bounds&quot;);
}
}
class DoubleArray {
...
};
</pre>
</div>
<p> The exception handling code is inserted directly into generated
wrapper functions. The <tt>$action</tt> variable is replaced with the
C/C++ code being executed by the wrapper. When an exception handler is
defined, errors can be caught and used to gracefully generate a Perl
error instead of forcing the entire program to terminate with an
uncaught error.</p>
<p> As shown, the exception handling code will be added to every wrapper
function. Since this is somewhat inefficient. You might consider
refining the exception handler to only apply to specific methods like
this:</p>
<div class="code">
<pre>
%exception getitem {
try {
$action
}
catch (RangeError) {
croak(&quot;Array index out-of-bounds&quot;);
}
}
%exception setitem {
try {
$action
}
catch (RangeError) {
croak(&quot;Array index out-of-bounds&quot;);
}
}
</pre>
</div>
<p> In this case, the exception handler is only attached to methods and
functions named <tt>getitem</tt> and <tt>setitem</tt>.</p>
<p> If you had a lot of different methods, you can avoid extra typing by
using a macro. For example:</p>
<div class="code">
<pre>
%define RANGE_ERROR
{
try {
$action
}
catch (RangeError) {
croak(&quot;Array index out-of-bounds&quot;);
}
}
%enddef
%exception getitem RANGE_ERROR;
%exception setitem RANGE_ERROR;
</pre>
</div>
<p> Since SWIG's exception handling is user-definable, you are not
limited to C++ exception handling. See the chapter on &quot;<a href="#Customization">
Customization features</a>&quot; for more examples.</p>
<h2><a name="Perl5_nn27">31.7 Remapping datatypes with typemaps</a></h2>
<p> This section describes how you can modify SWIG's default wrapping
behavior for various C/C++ datatypes using the <tt>%typemap</tt>
directive. This is an advanced topic that assumes familiarity with the
Perl C API as well as the material in the &quot;<a href="#Typemaps">Typemaps</a>
&quot; chapter.</p>
<p> Before proceeding, it should be stressed that typemaps are<em> not</em>
a required part of using SWIG---the default wrapping behavior is enough
in most cases. Typemaps are only used if you want to change some aspect
of the primitive C-Perl interface.</p>
<h3><a name="Perl5_nn28">31.7.1 A simple typemap example</a></h3>
<p> A typemap is nothing more than a code generation rule that is
attached to a specific C datatype. For example, to convert integers
from Perl to C, you might define a typemap like this:</p>
<div class="code">
<pre>
%module example
%typemap(in) int {
$1 = (int) SvIV($input);
printf(&quot;Received an integer : %d\n&quot;, $1);
}
...
%inline %{
extern int fact(int n);
%}
</pre>
</div>
<p> Typemaps are always associated with some specific aspect of code
generation. In this case, the &quot;in&quot; method refers to the conversion of
input arguments to C/C++. The datatype <tt>int</tt> is the datatype to
which the typemap will be applied. The supplied C code is used to
convert values. In this code a number of special variable prefaced by a
<tt>$</tt> are used. The <tt>$1</tt> variable is placeholder for a
local variable of type <tt>int</tt>. The <tt>$input</tt> variable is
the input object (usually a <tt>SV *</tt>).</p>
<p> When this example is used in Perl5, it will operate as follows:</p>
<div class="targetlang">
<pre>
use example;
$n = example::fact(6);
print &quot;$n\n&quot;;
...
Output:
Received an integer : 6
720
</pre>
</div>
<p> The application of a typemap to specific datatypes and argument
names involves more than simple text-matching--typemaps are fully
integrated into the SWIG type-system. When you define a typemap for <tt>
int</tt>, that typemap applies to <tt>int</tt> and qualified variations
such as <tt>const int</tt>. In addition, the typemap system follows <tt>
typedef</tt> declarations. For example:</p>
<div class="targetlang">
<pre>
%typemap(in) int n {
$1 = (int) SvIV($input);
printf(&quot;n = %d\n&quot;, $1);
}
%inline %{
typedef int Integer;
extern int fact(Integer n); // Above typemap is applied
%}
</pre>
</div>
<p> It should be noted that the matching of <tt>typedef</tt> only occurs
in one direction. If you defined a typemap for <tt>Integer</tt>, it is
not applied to arguments of type <tt>int</tt>.</p>
<p> Typemaps can also be defined for groups of consecutive arguments.
For example:</p>
<div class="targetlang">
<pre>
%typemap(in) (char *str, unsigned len) {
$1 = SvPV($input, $2);
};
int count(char c, char *str, unsigned len);
</pre>
</div>
<p> When a multi-argument typemap is defined, the arguments are always
handled as a single Perl object. This allows the function to be used
like this (notice how the length parameter is omitted):</p>
<div class="targetlang">
<pre>
example::count(&quot;e&quot;, &quot;Hello World&quot;);
1
&gt;&gt;&gt;
</pre>
</div>
<h3><a name="Perl5_nn29">31.7.2 Perl5 typemaps</a></h3>
<p> The previous section illustrated an &quot;in&quot; typemap for converting Perl
objects to C. A variety of different typemap methods are defined by the
Perl module. For example, to convert a C integer back into a Perl
object, you might define an &quot;out&quot; typemap like this:</p>
<div class="targetlang">
<pre>
%typemap(out) int {
$result = sv_newmortal();
sv_setiv($result, (IV) $1);
argvi++;
}
</pre>
</div>
<p> The following typemap methods are available:</p>
<p> <tt>%typemap(in)</tt></p>
<div class="indent"> Converts Perl5 object to input function arguments.</div>
<p> <tt>%typemap(out)</tt></p>
<div class="indent"> Converts function return value to a Perl5 value.</div>
<p> <tt>%typemap(varin)</tt></p>
<div class="indent"> Converts a Perl5 object to a global variable.</div>
<p> <tt>%typemap(varout)</tt></p>
<div class="indent"> Converts a global variable to a Perl5 object.</div>
<p> <tt>%typemap(freearg)</tt></p>
<div class="indent"> Cleans up a function argument after a function call</div>
<p> <tt>%typemap(argout)</tt></p>
<div class="indent"> Output argument handling</div>
<p> <tt>%typemap(ret)</tt></p>
<div class="indent"> Clean up return value from a function.</div>
<p> <tt>%typemap(memberin)</tt></p>
<div class="indent"> Setting of C++ member data (all languages).</div>
<p> <tt>%typemap(check)</tt></p>
<div class="indent"> Check value of input parameter.</div>
<h3><a name="Perl5_nn30">31.7.3 Typemap variables</a></h3>
<p> Within typemap code, a number of special variables prefaced with a <tt>
$</tt> may appear. A full list of variables can be found in the &quot;<a href="#Typemaps">
Typemaps</a>&quot; chapter. This is a list of the most common variables:</p>
<p> <tt>$1</tt></p>
<div class="indent"> A C local variable corresponding to the actual type
specified in the <tt>%typemap</tt> directive. For input values, this is
a C local variable that's supposed to hold an argument value. For
output values, this is the raw result that's supposed to be returned to
Perl.</div>
<p> <tt>$input</tt></p>
<div class="indent"> A Perl object holding the value of an argument of
variable value.</div>
<p> <tt>$result</tt></p>
<div class="indent"> A Perl object that holds the result to be returned
to Perl.</div>
<p> <tt>$1_name</tt></p>
<div class="indent"> The parameter name that was matched.</div>
<p> <tt>$1_type</tt></p>
<div class="indent"> The actual C datatype matched by the typemap.</div>
<p> <tt>$1_ltype</tt></p>
<div class="indent"> An assignable version of the datatype matched by
the typemap (a type that can appear on the left-hand-side of a C
assignment operation). This type is stripped of qualifiers and may be
an altered version of <tt>$1_type</tt>. All arguments and local
variables in wrapper functions are declared using this type so that
their values can be properly assigned.</div>
<p> <tt>$symname</tt></p>
<div class="indent"> The Perl name of the wrapper function being
created.</div>
<h3><a name="Perl5_nn31">31.7.4 Useful functions</a></h3>
<p> When writing typemaps, it is necessary to work directly with Perl5
objects. This, unfortunately, can be a daunting task. Consult the
&quot;perlguts&quot; man-page for all of the really ugly details. A short summary
of commonly used functions is provided here for reference. It should be
stressed that SWIG can be used quite effectively without knowing any of
these details--especially now that there are typemap libraries that can
already been written.</p>
<p><b> Perl Integer Functions</b></p>
<div class="code">
<pre>
int SvIV(SV *);
void sv_setiv(SV *sv, IV value);
SV *newSViv(IV value);
int SvIOK(SV *);
</pre>
</div>
<p><b> Perl Floating Point Functions</b></p>
<div class="code">
<pre>
double SvNV(SV *);
void sv_setnv(SV *, double value);
SV *newSVnv(double value);
int SvNOK(SV *);
</pre>
</div>
<p><b> Perl String Functions</b></p>
<div class="code">
<pre>
char *SvPV(SV *, STRLEN len);
void sv_setpv(SV *, char *val);
void sv_setpvn(SV *, char *val, STRLEN len);
SV *newSVpv(char *value, STRLEN len);
int SvPOK(SV *);
void sv_catpv(SV *, char *);
void sv_catpvn(SV *, char *, STRLEN);
</pre>
</div>
<p><b> Perl References</b></p>
<div class="code">
<pre>
void sv_setref_pv(SV *, char *, void *ptr);
int sv_isobject(SV *);
SV *SvRV(SV *);
int sv_isa(SV *, char *0;
</pre>
</div>
<h2><a name="Perl5_nn32">31.8 Typemap Examples</a></h2>
<p> This section includes a few examples of typemaps. For more examples,
you might look at the files &quot;<tt>perl5.swg</tt>&quot; and &quot;<tt>typemaps.i</tt>
&quot; in the SWIG library.</p>
<h3><a name="Perl5_nn33">31.8.1 Converting a Perl5 array to a char **</a>
</h3>
<p> A common problem in many C programs is the processing of command
line arguments, which are usually passed in an array of NULL terminated
strings. The following SWIG interface file allows a Perl5 array
reference to be used as a char ** datatype.</p>
<div class="code">
<pre>
%module argv
// This tells SWIG to treat char ** as a special case
%typemap(in) char ** {
AV *tempav;
I32 len;
int i;
SV **tv;
if (!SvROK($input))
croak(&quot;Argument $argnum is not a reference.&quot;);
if (SvTYPE(SvRV($input)) != SVt_PVAV)
croak(&quot;Argument $argnum is not an array.&quot;);
tempav = (AV*)SvRV($input);
len = av_len(tempav);
$1 = (char **) malloc((len+2)*sizeof(char *));
for (i = 0; i &lt;= len; i++) {
tv = av_fetch(tempav, i, 0);
$1[i] = (char *) SvPV(*tv, PL_na);
}
$1[i] = NULL;
};
// This cleans up the char ** array after the function call
%typemap(freearg) char ** {
free($1);
}
// Creates a new Perl array and places a NULL-terminated char ** into it
%typemap(out) char ** {
AV *myav;
SV **svs;
int i = 0, len = 0;
/* Figure out how many elements we have */
while ($1[len])
len++;
svs = (SV **) malloc(len*sizeof(SV *));
for (i = 0; i &lt; len ; i++) {
svs[i] = sv_newmortal();
sv_setpv((SV*)svs[i], $1[i]);
};
myav = av_make(len, svs);
free(svs);
$result = newRV_noinc((SV*)myav);
sv_2mortal($result);
argvi++;
}
// Now a few test functions
%inline %{
int print_args(char **argv) {
int i = 0;
while (argv[i]) {
printf(&quot;argv[%d] = %s\n&quot;, i, argv[i]);
i++;
}
return i;
}
// Returns a char ** list
char **get_args() {
static char *values[] = { &quot;Dave&quot;, &quot;Mike&quot;, &quot;Susan&quot;, &quot;John&quot;, &quot;Michelle&quot;, 0};
return &amp;values[0];
}
%}
</pre>
</div>
<p> When this module is compiled, the wrapped C functions can be used in
a Perl script as follows:</p>
<div class="targetlang">
<pre>
use argv;
@a = (&quot;Dave&quot;, &quot;Mike&quot;, &quot;John&quot;, &quot;Mary&quot;); # Create an array of strings
argv::print_args(\@a); # Pass it to our C function
$b = argv::get_args(); # Get array of strings from C
print @$b, &quot;\n&quot;; # Print it out
</pre>
</div>
<h3><a name="Perl5_nn34">31.8.2 Return values</a></h3>
<p> Return values are placed on the argument stack of each wrapper
function. The current value of the argument stack pointer is contained
in a variable <tt>argvi</tt>. Whenever a new output value is added, it
is critical that this value be incremented. For multiple output values,
the final value of <tt>argvi</tt> should be the total number of output
values.</p>
<p> The total number of return values should not exceed the number of
input values unless you explicitly extend the argument stack. This can
be done using the <tt>EXTEND()</tt> macro as in:</p>
<div class="code">
<pre>
%typemap(argout) int *OUTPUT {
if (argvi &gt;= items) {
EXTEND(sp, 1); /* Extend the stack by 1 object */
}
$result = sv_newmortal();
sv_setiv($result, (IV) *($1));
argvi++;
}
</pre>
</div>
<h3><a name="Perl5_nn35">31.8.3 Returning values from arguments</a></h3>
<p> Sometimes it is desirable for a function to return a value in one of
its arguments. This example describes the implementation of the <tt>
OUTPUT</tt> typemap.</p>
<div class="code">
<pre>
%module return
// This tells SWIG to treat an double * argument with name 'OutDouble' as
// an output value.
%typemap(argout) double *OUTPUT {
$result = sv_newmortal();
sv_setnv($result, *$input);
argvi++; /* Increment return count -- important! */
}
// We don't care what the input value is. Ignore, but set to a temporary variable
%typemap(in, numinputs=0) double *OUTPUT(double junk) {
$1 = &amp;junk;
}
// Now a function to test it
%{
/* Returns the first two input arguments */
int multout(double a, double b, double *out1, double *out2) {
*out1 = a;
*out2 = b;
return 0;
};
%}
// If we name both parameters OutDouble both will be output
int multout(double a, double b, double *OUTPUT, double *OUTPUT);
...
</pre>
</div>
<p> When this function is called, the output arguments are appended to
the stack used to return results. This shows up an array in Perl. For
example:</p>
<div class="targetlang">
<pre>
@r = multout(7, 13);
print &quot;multout(7, 13) = @r\n&quot;;
($x, $y) = multout(7, 13);
</pre>
</div>
<h3><a name="Perl5_nn36">31.8.4 Accessing array structure members</a></h3>
<p> Consider the following data structure:</p>
<div class="code">
<pre>
#define SIZE 8
typedef struct {
int values[SIZE];
...
} Foo;
</pre>
</div>
<p> By default, SWIG doesn't know how to the handle the values structure
member because it's an array, not a pointer. In this case, SWIG makes
the array member read-only. Reading will simply return a pointer to the
first item in the array. To make the member writable, a &quot;memberin&quot;
typemap can be used.</p>
<div class="code">
<pre>
%typemap(memberin) int [SIZE] {
int i;
for (i = 0; i &lt; SIZE; i++) {
$1[i] = $input[i];
}
}
</pre>
</div>
<p> Whenever a <tt>int [SIZE]</tt> member is encountered in a structure
or class, this typemap provides a safe mechanism for setting its value.</p>
<p> As in the previous example, the typemap can be generalized for any
dimension. For example:</p>
<div class="code">
<pre>
%typemap(memberin) int [ANY] {
int i;
for (i = 0; i &lt; $1_dim0; i++) {
$1[i] = $input[i];
}
}
</pre>
</div>
<p> When setting structure members, the input object is always assumed
to be a C array of values that have already been converted from the
target language. Because of this, the <tt>memberin</tt> typemap is
almost always combined with the use of an &quot;in&quot; typemap. For example,
the &quot;in&quot; typemap in the previous section would be used to convert an <tt>
int[]</tt> array to C whereas the &quot;memberin&quot; typemap would be used to
copy the converted array into a C data structure.</p>
<h3><a name="Perl5_nn37">31.8.5 Turning Perl references into C pointers</a>
</h3>
<p> A frequent confusion on the SWIG mailing list is errors caused by
the mixing of Perl references and C pointers. For example, suppose you
have a C function that modifies its arguments like this:</p>
<div class="code">
<pre>
void add(double a, double b, double *c) {
*c = a + b;
}
</pre>
</div>
<p> A common misinterpretation of this function is the following Perl
script:</p>
<div class="targetlang">
<pre>
# Perl script
$a = 3.5;
$b = 7.5;
$c = 0.0; # Output value
add($a, $b, \$c); # Place result in c (Except that it doesn't work)
</pre>
</div>
<p> To make this work with a reference, you can use a typemap such as
this:</p>
<div class="code">
<pre>
%typemap(in) double * (double dvalue) {
SV* tempsv;
if (!SvROK($input)) {
croak(&quot;expected a reference\n&quot;);
}
tempsv = SvRV($input);
if ((!SvNOK(tempsv)) &amp;&amp; (!SvIOK(tempsv))) {
croak(&quot;expected a double reference\n&quot;);
}
dvalue = SvNV(tempsv);
$1 = &amp;dvalue;
}
%typemap(argout) double * {
SV *tempsv;
tempsv = SvRV($input);
sv_setnv(tempsv, *$1);
}
</pre>
</div>
<p> Now, if you place this before the add function, you can do this:</p>
<div class="targetlang">
<pre>
$a = 3.5;
$b = 7.5;
$c = 0.0;
add($a, $b, \$c); # Now it works!
print &quot;$c\n&quot;;
</pre>
</div>
<h3><a name="Perl5_nn38">31.8.6 Pointer handling</a></h3>
<p> Occasionally, it might be necessary to convert pointer values that
have been stored using the SWIG typed-pointer representation. To
convert a pointer from Perl to C, the following function is used:</p>
<p> <tt>int SWIG_ConvertPtr(SV *obj, void **ptr, swig_type_info *ty, int
flags)</tt></p>
<div class="indent"> Converts a Perl object <tt>obj</tt> to a C pointer.
The result of the conversion is placed into the pointer located at <tt>
ptr</tt>. <tt>ty</tt> is a SWIG type descriptor structure. <tt>flags</tt>
is used to handle error checking and other aspects of conversion. <tt>
flags</tt> is currently undefined and reserved for future expansion.
Returns 0 on success and -1 on error.</div>
<p> <tt>void *SWIG_MakePtr(SV *obj, void *ptr, swig_type_info *ty, int
flags)</tt></p>
<div class="indent"> Creates a new Perl pointer object. <tt>obj</tt> is
a Perl SV that has been initialized to hold the result, <tt>ptr</tt> is
the pointer to convert, <tt>ty</tt> is the SWIG type descriptor
structure that describes the type, and <tt>flags</tt> is a flag that
controls properties of the conversion. <tt>flags</tt> is currently
undefined and reserved.</div>
<p> Both of these functions require the use of a special SWIG
type-descriptor structure. This structure contains information about
the mangled name of the datatype, type-equivalence information, as well
as information about converting pointer values under C++ inheritance.
For a type of <tt>Foo *</tt>, the type descriptor structure is usually
accessed as follows:</p>
<div class="code">
<pre>
Foo *f;
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;f, SWIGTYPE_p_Foo, 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method '$symname', expecting type Foo&quot;);
}
SV *sv = sv_newmortal();
SWIG_MakePtr(sv, f, SWIGTYPE_p_Foo, 0);
</pre>
</div>
<p> In a typemap, the type descriptor should always be accessed using
the special typemap variable <tt>$1_descriptor</tt>. For example:</p>
<div class="code">
<pre>
%typemap(in) Foo * {
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;$1, $1_descriptor, 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method '$symname', expecting type Foo&quot;);
}
}
</pre>
</div>
<p> If necessary, the descriptor for any type can be obtained using the <tt>
$descriptor()</tt> macro in a typemap. For example:</p>
<div class="code">
<pre>
%typemap(in) Foo * {
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;$1, $descriptor(Foo *), 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method '$symname', expecting type Foo&quot;);
}
}
</pre>
</div>
<h2><a name="Perl5_nn39">31.9 Proxy classes</a></h2>
<p><b> Out of date. Needs update.</b></p>
<p> Using the low-level procedural interface, SWIG can also construct a
high-level object oriented interface to C structures and C++ classes.
This is done by constructing a Perl proxy class (also known as a shadow
class) that provides an OO wrapper to the underlying code. This section
describes the implementation details of the proxy interface.</p>
<h3><a name="Perl5_nn40">31.9.1 Preliminaries</a></h3>
<p> Proxy classes, are generated by default. If you want to turn them
off, use the <tt>-noproxy</tt> command line option. For example:</p>
<div class="code">
<pre>
$ swig -c++ -perl -noproxy example.i
</pre>
</div>
<p> When proxy classes are used, SWIG moves all of the low-level
procedural wrappers to another package name. By default, this package
is named 'modulec' where 'module' is the name of the module you
provided with the <tt>%module</tt> directive. Then, in place of the
original module, SWIG creates a collection of high-level Perl wrappers.
In your scripts, you will use these high level wrappers. The wrappers,
in turn, interact with the low-level procedural module.</p>
<h3><a name="Perl5_nn41">31.9.2 Structure and class wrappers</a></h3>
<p> Suppose you have the following SWIG interface file:</p>
<div class="code">
<pre>
%module example
struct Vector {
Vector(double x, double y, double z);
~Vector();
double x, y, z;
};
</pre>
</div>
<p> When wrapped, SWIG creates the following set of low-level accessor
functions as described in previous sections.</p>
<div class="code">
<pre>
Vector *new_Vector(double x, double y, double z);
void delete_Vector(Vector *v);
double Vector_x_get(Vector *v);
double Vector_x_set(Vector *v, double value);
double Vector_y_get(Vector *v);
double Vector_y_set(Vector *v, double value);
double Vector_z_get(Vector *v);
double Vector_z_set(Vector *v, double value);
</pre>
</div>
<p> However, when proxy classes are enabled, these accessor functions
are wrapped inside a Perl class like this:</p>
<div class="targetlang">
<pre>
package example::Vector;
@ISA = qw( example );
%OWNER = ();
%BLESSEDMEMBERS = ();
sub new () {
my $self = shift;
my @args = @_;
$self = vectorc::new_Vector(@args);
return undef if (!defined($self));
bless $self, &quot;example::Vector&quot;;
$OWNER{$self} = 1;
my %retval;
tie %retval, &quot;example::Vector&quot;, $self;
return bless \%retval, &quot;Vector&quot;;
}
sub DESTROY {
return unless $_[0]-&gt;isa('HASH');
my $self = tied(%{$_[0]});
delete $ITERATORS{$self};
if (exists $OWNER{$self}) {
examplec::delete_Vector($self));
delete $OWNER{$self};
}
}
sub FETCH {
my ($self, $field) = @_;
my $member_func = &quot;vectorc::Vector_${field}_get&quot;;
my $val = &amp;$member_func($self);
if (exists $BLESSEDMEMBERS{$field}) {
return undef if (!defined($val));
my %retval;
tie %retval, $BLESSEDMEMBERS{$field}, $val;
return bless \%retval, $BLESSEDMEMBERS{$field};
}
return $val;
}
sub STORE {
my ($self, $field, $newval) = @_;
my $member_func = &quot;vectorc::Vector_${field}_set&quot;;
if (exists $BLESSEDMEMBERS{$field}) {
&amp;$member_func($self, tied(%{$newval}));
} else {
&amp;$member_func($self, $newval);
}
}
</pre>
</div>
<p> Each structure or class is mapped into a Perl package of the same
name. The C++ constructors and destructors are mapped into constructors
and destructors for the package and are always named &quot;new&quot; and
&quot;DESTROY&quot;. The constructor always returns a tied hash table. This hash
table is used to access the member variables of a structure in addition
to being able to invoke member functions. The <tt>%OWNER</tt> and <tt>
%BLESSEDMEMBERS</tt> hash tables are implementation details used
internally and described shortly.</p>
<p> To use our new proxy class we can simply do the following:</p>
<div class="targetlang">
<pre>
# Perl code using Vector class
$v = new Vector(2, 3, 4);
$w = Vector-&gt;new(-1, -2, -3);
# Assignment of a single member
$v-&gt;{x} = 7.5;
# Assignment of all members
%$v = ( x=&gt;3,
y=&gt;9,
z=&gt;-2);
# Reading members
$x = $v-&gt;{x};
# Destruction
$v-&gt;DESTROY();
</pre>
</div>
<h3><a name="Perl5_nn42">31.9.3 Object Ownership</a></h3>
<p> In order for proxy classes to work properly, it is necessary for
Perl to manage some mechanism of object ownership. Here's the crux of
the problem---suppose you had a function like this:</p>
<div class="code">
<pre>
Vector *Vector_get(Vector *v, int index) {
return &amp;v[i];
}
</pre>
</div>
<p> This function takes a Vector pointer and returns a pointer to
another Vector. Such a function might be used to manage arrays or lists
of vectors (in C). Now contrast this function with the constructor for
a Vector object:</p>
<div class="code">
<pre>
Vector *new_Vector(double x, double y, double z) {
Vector *v;
v = new Vector(x, y, z); // Call C++ constructor
return v;
}
</pre>
</div>
<p> Both functions return a Vector, but the constructor is returning a
brand-new Vector while the other function is returning a Vector that
was already created (hopefully). In Perl, both vectors will be
indistinguishable---clearly a problem considering that we would
probably like the newly created Vector to be destroyed when we are done
with it.</p>
<p> To manage these problems, each class contains two methods that
access an internal hash table called <tt>%OWNER</tt>. This hash keeps a
list of all of the objects that Perl knows that it has created. This
happens in two cases: (1) when the constructor has been called, and (2)
when a function implicitly creates a new object (as is done when SWIG
needs to return a complex datatype by value). When the destructor is
invoked, the Perl proxy class module checks the <tt>%OWNER</tt> hash to
see if Perl created the object. If so, the C/C++ destructor is invoked.
If not, we simply destroy the Perl object and leave the underlying C
object alone (under the assumption that someone else must have created
it).</p>
<p> This scheme works remarkably well in practice but it isn't
foolproof. In fact, it will fail if you create a new C object in Perl,
pass it on to a C function that remembers the object, and then destroy
the corresponding Perl object (this situation turns out to come up
frequently when constructing objects like linked lists and trees). When
C takes possession of an object, you can change Perl's ownership by
calling the <tt>DISOWN</tt> method (which will delete the object from
the internal <tt>%OWNER</tt> hash).</p>
<p> The <tt>%OWNER</tt> hash is an implementation detail, discussed here
only to help clarify the operation of <tt>ACQUIRE</tt> and <tt>DISOWN</tt>
. You should not access <tt>%OWNER</tt> directly - the details of how it
works (and possibly even its existence) may change in future SWIG
versions.</p>
<div class="targetlang">
<pre>
# Perl code to change ownership of an object
$v = new Vector(x, y, z);
$v-&gt;DISOWN();
</pre>
</div>
<p> To acquire ownership of an object, the <tt>ACQUIRE</tt> method can
be used.</p>
<div class="targetlang">
<pre>
# Given Perl ownership of a file
$u = Vector_get($v);
$u-&gt;ACQUIRE();
</pre>
</div>
<p> As always, a little care is in order. SWIG does not provide
reference counting, garbage collection, or advanced features one might
find in sophisticated languages.</p>
<h3><a name="Perl5_nn43">31.9.4 Nested Objects</a></h3>
<p> Suppose that we have a new object that looks like this:</p>
<div class="code">
<pre>
struct Particle {
Vector r;
Vector v;
Vector f;
int type;
}
</pre>
</div>
<p> In this case, the members of the structure are complex objects that
have already been encapsulated in a Perl proxy class. To handle these
correctly, we use the <tt>%BLESSEDMEMBERS</tt> hash which would look
like this (along with some supporting code):</p>
<div class="targetlang">
<pre>
package Particle;
...
%BLESSEDMEMBERS = (
r =&gt; `Vector',
v =&gt; `Vector',
f =&gt; `Vector',
);
</pre>
</div>
<p> When fetching members from the structure, <tt>%BLESSEDMEMBERS</tt>
is checked. If the requested field is present, we create a tied-hash
table and return it. If not, we just return the corresponding member
unmodified.</p>
<p> This implementation allows us to operate on nested structures as
follows:</p>
<div class="targetlang">
<pre>
# Perl access of nested structure
$p = new Particle();
$p-&gt;{f}-&gt;{x} = 0.0;
%${$p-&gt;{v}} = ( x=&gt;0, y=&gt;0, z=&gt;0);
</pre>
</div>
<h3><a name="Perl5_nn44">31.9.5 Proxy Functions</a></h3>
<p> When functions take arguments involving a complex object, it is
sometimes necessary to write a proxy function. For example:</p>
<div class="code">
<pre>
double dot_product(Vector *v1, Vector *v2);
</pre>
</div>
<p> Since Vector is an object already wrapped into a proxy class, we
need to modify this function to accept arguments that are given in the
form of tied hash tables. This is done by creating a Perl function like
this:</p>
<div class="targetlang">
<pre>
sub dot_product {
my @args = @_;
$args[0] = tied(%{$args[0]}); # Get the real pointer values
$args[1] = tied(%{$args[1]});
my $result = vectorc::dot_product(@args);
return $result;
}
</pre>
</div>
<p> This function replaces the original function, but operates in an
identical manner.</p>
<h3><a name="Perl5_nn45">31.9.6 Inheritance</a></h3>
<p> Simple C++ inheritance is handled using the Perl <tt>@ISA</tt> array
in each class package. For example, if you have the following interface
file:</p>
<div class="code">
<pre>
// shapes.i
// SWIG interface file for shapes class
%module shapes
%{
#include &quot;shapes.h&quot;
%}
class Shape {
public:
virtual double area() = 0;
virtual double perimeter() = 0;
void set_location(double x, double y);
};
class Circle : public Shape {
public:
Circle(double radius);
~Circle();
double area();
double perimeter();
};
class Square : public Shape {
public:
Square(double size);
~Square();
double area();
double perimeter();
}
</pre>
</div>
<p> The resulting, Perl wrapper class will create the following code:</p>
<div class="targetlang">
<pre>
Package Shape;
@ISA = (shapes);
...
Package Circle;
@ISA = (shapes Shape);
...
Package Square;
@ISA = (shapes Shape);
</pre>
</div>
<p> The <tt>@ISA</tt> array determines where to look for methods of a
particular class. In this case, both the <tt>Circle</tt> and <tt>Square</tt>
classes inherit functions from <tt>Shape</tt> so we'll want to look in
the <tt>Shape</tt> base class for them. All classes also inherit from
the top-level module <tt>shapes</tt>. This is because certain common
operations needed to implement proxy classes are implemented only once
and reused in the wrapper code for various classes and structures.</p>
<p> Since SWIG proxy classes are implemented in Perl, it is easy to
subclass from any SWIG generated class. To do this, simply put the name
of a SWIG class in the <tt>@ISA</tt> array for your new class. However,
be forewarned that this is not a trivial problem. In particular,
inheritance of data members is extremely tricky (and I'm not even sure
if it really works).</p>
<h3><a name="Perl5_nn46">31.9.7 Modifying the proxy methods</a></h3>
<p> It is possible to override the SWIG generated proxy/shadow methods,
using <tt>%feature(&quot;shadow&quot;)</tt>. It works like all the other <a href="#Customization_features">
%feature directives</a>. Here is a simple example showing how to add
some Perl debug code to the constructor:</p>
<div class="targetlang">
<pre>
/* Let's make the constructor of the class Square more verbose */
%feature(&quot;shadow&quot;) Square(double w)
%{
sub new {
my $pkg = shift;
my $self = examplec::new_Square(@_);
print STDERR &quot;Constructed an @{[ref($self)]}\n&quot;;
bless $self, $pkg if defined($self);
}
%}
class Square {
public:
Square(double w);
...
};
</pre>
</div>
<h2><a name="Perl5_nn47">31.10 Adding additional Perl code</a></h2>
<p> If writing support code in C isn't enough, it is also possible to
write code in Perl. This code gets inserted in to the <tt>.pm</tt> file
created by SWIG. One use of Perl code might be to supply a high-level
interface to certain functions. For example:</p>
<div class="code">
<pre>
void set_transform(Image *im, double x[4][4]);
...
/* Rewrite the high level interface to set_transform */
%perlcode %{
sub set_transform
{
my ($im, $x) = @_;
my $a = new_mat44();
for (my $i = 0; $i &lt; 4, $i++)
{
for (my $j = 0; $j &lt; 4, $j++)
{
mat44_set($a, $i, $j, $x-&gt;[i][j])
}
}
example.set_transform($im, $a);
free_mat44($a);
}
%}
</pre>
</div>
<p> In this example, <tt>set_transform()</tt> provides a high-level Perl
interface built on top of low-level helper functions. For example, this
code now seems to work:</p>
<div class="targetlang">
<pre>
my $a =
[[1, 0, 0, 0],
[0, 1, 0, 0],
[0, 0, 1, 0],
[0, 0, 0, 1]];
set_transform($im, $a);
</pre>
</div>
<h2><a name="Perl5_directors">31.11 Cross language polymorphism</a></h2>
<p> Proxy classes provide a more natural, object-oriented way to access
extension classes. As described above, each proxy instance has an
associated C++ instance, and method calls to the proxy are passed to
the C++ instance transparently via C wrapper functions.</p>
<p> This arrangement is asymmetric in the sense that no corresponding
mechanism exists to pass method calls down the inheritance chain from
C++ to Perl. In particular, if a C++ class has been extended in Perl
(by extending the proxy class), these extensions will not be visible
from C++ code. Virtual method calls from C++ are thus not able access
the lowest implementation in the inheritance chain.</p>
<p> Changes have been made to SWIG to address this problem and make the
relationship between C++ classes and proxy classes more symmetric. To
achieve this goal, new classes called directors are introduced at the
bottom of the C++ inheritance chain. The job of the directors is to
route method calls correctly, either to C++ implementations higher in
the inheritance chain or to Perl implementations lower in the
inheritance chain. The upshot is that C++ classes can be extended in
Perl and from C++ these extensions look exactly like native C++
classes. Neither C++ code nor Perl code needs to know where a
particular method is implemented: the combination of proxy classes,
director classes, and C wrapper functions takes care of all the
cross-language method routing transparently.</p>
<h3><a name="Perl5_nn48">31.11.1 Enabling directors</a></h3>
<p> The director feature is disabled by default. To use directors you
must make two changes to the interface file. First, add the &quot;directors&quot;
option to the %module directive, like this:</p>
<div class="code">
<pre>
%module(directors=&quot;1&quot;) modulename
</pre>
</div>
<p> Without this option no director code will be generated. Second, you
must use the %feature(&quot;director&quot;) directive to tell SWIG which classes
and methods should get directors. The %feature directive can be applied
globally, to specific classes, and to specific methods, like this:</p>
<div class="code">
<pre>
// generate directors for all classes that have virtual methods
%feature(&quot;director&quot;);
// generate directors for the virtual methods in class Foo
%feature(&quot;director&quot;) Foo;
</pre>
</div>
<p> You can use the %feature(&quot;nodirector&quot;) directive to turn off
directors for specific classes or methods. So for example,</p>
<div class="code">
<pre>
%feature(&quot;director&quot;) Foo;
%feature(&quot;nodirector&quot;) Foo::bar;
</pre>
</div>
<p> will generate directors for the virtual methods of class Foo except
bar().</p>
<p> Directors can also be generated implicitly through inheritance. In
the following, class Bar will get a director class that handles the
methods one() and two() (but not three()):</p>
<div class="code">
<pre>
%feature(&quot;director&quot;) Foo;
class Foo {
public:
Foo(int foo);
virtual void one();
virtual void two();
};
class Bar: public Foo {
public:
virtual void three();
};
</pre>
</div>
<p> then at the Perl side you can define</p>
<div class="targetlang">
<pre>
use mymodule;
package MyFoo;
use base 'mymodule::Foo';
sub one {
print &quot;one from Perl\n&quot;;
}
</pre>
</div>
<h3><a name="Perl5_nn49">31.11.2 Director classes</a></h3>
<p> For each class that has directors enabled, SWIG generates a new
class that derives from both the class in question and a special <tt>
Swig::Director</tt> class. These new classes, referred to as director
classes, can be loosely thought of as the C++ equivalent of the Perl
proxy classes. The director classes store a pointer to their underlying
Perl object and handle various issues related to object ownership.</p>
<p> For simplicity let's ignore the <tt>Swig::Director</tt> class and
refer to the original C++ class as the director's base class. By
default, a director class extends all virtual methods in the
inheritance chain of its base class (see the preceding section for how
to modify this behavior). Virtual methods that have a final specifier
are unsurprisingly excluded. Thus the virtual method calls, whether
they originate in C++ or in Perl via proxy classes, eventually end up
in at the implementation in the director class. The job of the director
methods is to route these method calls to the appropriate place in the
inheritance chain. By &quot;appropriate place&quot; we mean the method that would
have been called if the C++ base class and its extensions in Perl were
seamlessly integrated. That seamless integration is exactly what the
director classes provide, transparently skipping over all the messy
extension API glue that binds the two languages together.</p>
<p> In reality, the &quot;appropriate place&quot; is one of only two
possibilities: C++ or Perl. Once this decision is made, the rest is
fairly easy. If the correct implementation is in C++, then the lowest
implementation of the method in the C++ inheritance chain is called
explicitly. If the correct implementation is in Perl, the Perl API is
used to call the method of the underlying Perl object (after which the
usual virtual method resolution in Perl automatically finds the right
implementation).</p>
<p> Now how does the director decide which language should handle the
method call? The basic rule is to handle the method in Perl, unless
there's a good reason not to. The reason for this is simple: Perl has
the most &quot;extended&quot; implementation of the method. This assertion is
guaranteed, since at a minimum the Perl proxy class implements the
method. If the method in question has been extended by a class derived
from the proxy class, that extended implementation will execute exactly
as it should. If not, the proxy class will route the method call into a
C wrapper function, expecting that the method will be resolved in C++.
The wrapper will call the virtual method of the C++ instance, and since
the director extends this the call will end up right back in the
director method. Now comes the &quot;good reason not to&quot; part. If the
director method were to blindly call the Perl method again, it would
get stuck in an infinite loop. We avoid this situation by adding
special code to the C wrapper function that tells the director method
to not do this. The C wrapper function compares the pointer to the Perl
object that called the wrapper function to the pointer stored by the
director. If these are the same, then the C wrapper function tells the
director to resolve the method by calling up the C++ inheritance chain,
preventing an infinite loop.</p>
<p> One more point needs to be made about the relationship between
director classes and proxy classes. When a proxy class instance is
created in Perl, SWIG creates an instance of the original C++ class.
This is exactly what happens without directors and is true even if
directors are enabled for the particular class in question. When a
class<i> derived</i> from a proxy class is created, however, SWIG then
creates an instance of the corresponding C++ director class. The reason
for this difference is that user-defined subclasses may override or
extend methods of the original class, so the director class is needed
to route calls to these methods correctly. For unmodified proxy
classes, all methods are ultimately implemented in C++ so there is no
need for the extra overhead involved with routing the calls through
Perl.</p>
<h3><a name="Perl5_nn50">31.11.3 Ownership and object destruction</a></h3>
<p> Memory management issues are slightly more complicated with
directors than for proxy classes alone. Perl instances hold a pointer
to the associated C++ director object, and the director in turn holds a
pointer back to a Perl object. By default, proxy classes own their C++
director object and take care of deleting it when they are garbage
collected.</p>
<p> This relationship can be reversed by calling the special <tt>
DISOWN()</tt> method of the proxy class. After calling this method the
director class increments the reference count of the Perl object. When
the director class is deleted it decrements the reference count.
Assuming no outstanding references to the Perl object remain, the Perl
object will be destroyed at the same time. This is a good thing, since
directors and proxies refer to each other and so must be created and
destroyed together. Destroying one without destroying the other will
likely cause your program to segfault.</p>
<p> Also note that due to the proxy implementation, the <tt>DESTROY()</tt>
method on directors can be called for several reasons, many of which
have little to do with the teardown of an object instance. To help
disambiguate this, a second argument is added to the <tt>DESTROY()</tt>
call when a C++ director object is being released. So, to avoid running
your clean-up code when an object is not really going away, or after it
has already been reclaimed, it is suggested that custom destructors in
Perl subclasses looks something like:</p>
<div class="targetlang">
<pre>
sub DESTROY {
my($self, $final) = @_;
if($final) {
# real teardown code
}
shift-&gt;SUPER::DESTROY(@_);
}
</pre>
</div>
<h3><a name="Perl5_nn51">31.11.4 Exception unrolling</a></h3>
<p> With directors routing method calls to Perl, and proxies routing
them to C++, the handling of exceptions is an important concern. By
default, the directors ignore exceptions that occur during method calls
that are resolved in Perl. To handle such exceptions correctly, it is
necessary to temporarily translate them into C++ exceptions. This can
be done with the %feature(&quot;director:except&quot;) directive. The following
code should suffice in most cases:</p>
<div class="code">
<pre>
%feature(&quot;director:except&quot;) {
if ($error != NULL) {
throw Swig::DirectorMethodException();
}
}
</pre>
</div>
<p> This code will check the Perl error state after each method call
from a director into Perl, and throw a C++ exception if an error
occurred. This exception can be caught in C++ to implement an error
handler.</p>
<p> It may be the case that a method call originates in Perl, travels up
to C++ through a proxy class, and then back into Perl via a director
method. If an exception occurs in Perl at this point, it would be nice
for that exception to find its way back to the original caller. This
can be done by combining a normal %exception directive with the <tt>
director:except</tt> handler shown above. Here is an example of a
suitable exception handler:</p>
<div class="code">
<pre>
%exception {
try { $action }
catch (Swig::DirectorException &amp;e) { SWIG_fail; }
}
</pre>
</div>
<p> The class Swig::DirectorException used in this example is actually a
base class of Swig::DirectorMethodException, so it will trap this
exception. Because the Perl error state is still set when
Swig::DirectorMethodException is thrown, Perl will register the
exception as soon as the C wrapper function returns.</p>
<h3><a name="Perl5_nn52">31.11.5 Overhead and code bloat</a></h3>
<p> Enabling directors for a class will generate a new director method
for every virtual method in the class' inheritance chain. This alone
can generate a lot of code bloat for large hierarchies. Method
arguments that require complex conversions to and from target language
types can result in large director methods. For this reason it is
recommended that you selectively enable directors only for specific
classes that are likely to be extended in Perl and used in C++.</p>
<p> Compared to classes that do not use directors, the call routing in
the director methods does add some overhead. In particular, at least
one dynamic cast and one extra function call occurs per method call
from Perl. Relative to the speed of Perl execution this is probably
completely negligible. For worst case routing, a method call that
ultimately resolves in C++ may take one extra detour through Perl in
order to ensure that the method does not have an extended Perl
implementation. This could result in a noticeable overhead in some
cases.</p>
<p> Although directors make it natural to mix native C++ objects with
Perl objects (as director objects) via a common base class pointer, one
should be aware of the obvious fact that method calls to Perl objects
will be much slower than calls to C++ objects. This situation can be
optimized by selectively enabling director methods (using the %feature
directive) for only those methods that are likely to be extended in
Perl.</p>
<h3><a name="Perl5_nn53">31.11.6 Typemaps</a></h3>
<p> Typemaps for input and output of most of the basic types from
director classes have been written. These are roughly the reverse of
the usual input and output typemaps used by the wrapper code. The
typemap operation names are 'directorin', 'directorout', and
'directorargout'. The director code does not currently use any of the
other kinds of typemaps. It is not clear at this point which kinds are
appropriate and need to be supported.</p>
<HR NOSHADE>
<h1><a name="Php">32 SWIG and PHP</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Php_nn1">Generating PHP Extensions</a>
<ul>
<li><a href="#Php_nn1_1">Building a loadable extension</a></li>
<li><a href="#Php_nn1_3">Using PHP Extensions</a></li>
</ul>
</li>
<li><a href="#Php_nn2">Basic PHP interface</a>
<ul>
<li><a href="#Php_nn2_1">Constants</a></li>
<li><a href="#Php_nn2_2">Global Variables</a></li>
<li><a href="#Php_nn2_3">Functions</a></li>
<li><a href="#Php_nn2_4">Overloading</a></li>
<li><a href="#Php_nn2_5">Pointers and References</a></li>
<li><a href="#Php_nn2_6">Structures and C++ classes</a>
<ul>
<li><a href="#Php_nn2_6_1">Using -noproxy</a></li>
<li><a href="#Php_nn2_6_2">Constructors and Destructors</a></li>
<li><a href="#Php_nn2_6_3">Static Member Variables</a></li>
<li><a href="#Php_nn2_6_4">Static Member Functions</a></li>
<li><a href="#Php_nn2_6_5">Specifying Implemented Interfaces</a></li>
<li><a href="#Php_nn2_6_6">Dynamic Properties</a></li>
</ul>
</li>
<li><a href="#Php_nn2_7">PHP Pragmas, Startup and Shutdown code</a></li>
</ul>
</li>
<li><a href="#Php_nn3">Cross language polymorphism</a>
<ul>
<li><a href="#Php_nn3_1">Enabling directors</a></li>
<li><a href="#Php_nn3_2">Director classes</a></li>
<li><a href="#Php_nn3_3">Ownership and object destruction</a></li>
<li><a href="#Php_nn3_4">Exception unrolling</a></li>
<li><a href="#Php_nn3_5">Overhead and code bloat</a></li>
<li><a href="#Php_nn3_6">Typemaps</a></li>
<li><a href="#Php_nn3_7">Miscellaneous</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> In this chapter, we discuss SWIG's support of PHP. Currently any
PHP8 release should work.</p>
<p> Support for PHP8 was added in SWIG 4.1.0. Support for PHP7 was added
in SWIG 3.0.11 and removed in 4.2.0. Support for PHP5 was removed in
SWIG 4.0.0 and support for PHP4 was removed in SWIG 1.3.37. There never
was a PHP6 release.</p>
<p> In order to use this module, you will need to have a copy of the PHP
include files to compile the SWIG generated C/C++ sources. If you
installed PHP from a binary package, you may need to install a
&quot;php-dev&quot; or &quot;php-devel&quot; package for these to be installed. You can
find out where these files are by running <tt>php-config --includes</tt>
. To use the built PHP module you will need either the php binary or the
Apache php module. If you want to build your extension into php
directly, you will need the complete PHP source tree available.</p>
<h2><a name="Php_nn1">32.1 Generating PHP Extensions</a></h2>
<p> To build a PHP extension, run swig using the <tt>-php</tt> option
(you can also use <tt>-php7</tt> - PHP7 and PHP8 have a largely
compatible C extension API, hence the same option name has been used
for both). For example:</p>
<div class="code">
<pre>
swig -php example.i
</pre>
</div>
<p> This will produce 2 files: example_wrap.c and php_example.h. The
first file, <tt>example_wrap.c</tt> contains all of the C code needed
to build a PHP extension. The second file, <tt>php_example.h</tt>
contains the header information needed if you wish to statically link
the extension into the php interpreter.</p>
<p> If the interface file uses <tt>%pragma(php) include=</tt>... or <tt>
%pragma(php) code=</tt>... then SWIG will also generate a third file, <tt>
example.php</tt> to contain what these specify. In SWIG &lt; 4.1.0, this
third file was always generated as it defined the PHP classes, etc (but
this is now done via C code in <tt>example_wrap.c</tt>) and also
contained code to dynamically load the extension (but this used the PHP
<tt>dl()</tt> function, which isn't recommended nowadays).</p>
<p> SWIG can generate PHP extensions from C++ libraries as well when
given the <tt>-c++</tt> option. The support for C++ is discussed in
more detail in <a href="#Php_nn2_6">section 27.2.6</a>. The generated
C++ wrapper will be called example_wrap.cxx. You can specify a
different extension for the C++ wrapper using <tt>-cppext</tt> - e.g.
if you want example_wrap.cc use <tt>-cppext cc</tt>.</p>
<p> The usual (and recommended) way is to build the extension as a
separate dynamically loaded module (which is supported by all modern
operating systems).</p>
<p> It is also possible to rebuild PHP from source so that your module
is statically linked into the php executable/library. This is a lot
more work, and also requires a full rebuild of PHP to update your
module, and it doesn't play nicely with package system. We don't
recommend this approach, or provide explicit support for it.</p>
<h3><a name="Php_nn1_1">32.1.1 Building a loadable extension</a></h3>
<p> To build your module as a dynamically loadable extension, use
compilation commands like these (if you aren't using GCC, the commands
will be different, and there may be some variation between platforms -
these commands should at least work for Linux though):</p>
<div class="shell">
<pre>
$ gcc `php-config --includes` -fPIC -c example_wrap.c example.c
$ gcc -shared example_wrap.o example.o -o example.so
</pre>
</div>
<h3><a name="Php_nn1_3">32.1.2 Using PHP Extensions</a></h3>
<p> To test the extension from a PHP script, you first need to tell PHP
to load it. The recommended (and simplest!) way to do this is to copy
it to PHP's default extension directory and add a line like this to the
<tt>[PHP]</tt> section of <tt>php.ini</tt>:</p>
<div class="code">
<pre>
extension=modulename
</pre>
</div>
<p> If the module is not in PHP's default extension directory, you also
need to specify the path, in which case you'll also need to deal with
platform-specific naming - for example, on Linux:</p>
<div class="code">
<pre>
extension=/path/to/modulename.so
</pre>
</div>
<p> but on Microsoft Windows you'd need to use:</p>
<div class="code">
<pre>
extension=/path/to/php_modulename.dll
</pre>
</div>
<p> If you're using the PHP CLI SAPI it's possible (but not recommended)
to use the <a href="https://www.php.net/manual/en/function.dl.php">dl()
function</a> to load an extension at run time, by adding a line like
this to the start of each PHP script which uses your extension:</p>
<div class="code">
<pre>
dl(&quot;modulename&quot;); // Load the module
</pre>
</div>
<p> Again, if the module isn't in PHP's default extension directory
you'll also need to specify the path and deal with that varying by
platform.</p>
<p> For security reasons PHP no longer supports <tt>dl()</tt> when
running PHP through a webserver, so this isn't an option there.</p>
<h2><a name="Php_nn2">32.2 Basic PHP interface</a></h2>
<p> It is important to understand that PHP uses a single global
namespace into which all symbols from extension modules are loaded. It
is quite possible for names of symbols in one extension module to clash
with other symbols unless care is taken to <tt>%rename</tt> them. At
present SWIG doesn't have support for generating wrappers which make
use of PHP's namespace feature.</p>
<h3><a name="Php_nn2_1">32.2.1 Constants</a></h3>
<p> These work in much the same way as in C/C++. Constants can be
defined by using either the normal C pre-processor declarations, or the
<tt>%constant</tt> SWIG directive. These will then be available from
your PHP script as a PHP constant, (i.e. no dollar sign is needed to
access them.) For example, with a swig interface file like this,</p>
<div class="code">
<pre>
%module example
#define PI 3.14159
%constant int E = 2.71828
</pre>
</div>
<p> you can access the constants in your PHP script like this,</p>
<div class="code">
<pre>
echo &quot;PI = &quot; . PI . &quot;\n&quot;;
echo &quot;E = &quot; . E . &quot;\n&quot;;
</pre>
</div>
<h3><a name="Php_nn2_2">32.2.2 Global Variables</a></h3>
<p> Because PHP does not provide a mechanism to intercept access and
assignment of global variables, global variables are supported through
the use of automatically generated accessor functions.</p>
<div class="code">
<pre>
%module example;
%inline %{
double seki = 2;
void print_seki() {
zend_printf(&quot;seki is now %f\n&quot;, seki);
}
%}
</pre>
</div>
<p> is accessed as follows:</p>
<div class="code">
<pre>
print seki_get();
seki_set( seki_get() * 2); # The C variable is now 4.
print seki_get();
</pre>
</div>
<p> SWIG supports global variables of all C datatypes including pointers
and complex objects. To support additional types, you just need to
supply the standard <tt>in</tt> and <tt>out</tt> typemaps, which get
used because of the wrapping as <tt>_get()</tt> and <tt>_set()</tt>
functions.</p>
<p> SWIG honors the <tt>%immutable</tt> modifier by not generating a <tt>
_set</tt> method (so attempting to call it will give a PHP fatal error).
A <tt>_get</tt> method is still generated so this provides read-only
access to the variable from the PHP script.</p>
<p> At this time SWIG does not support custom accessor methods.</p>
<h3><a name="Php_nn2_3">32.2.3 Functions</a></h3>
<p> C functions are converted into PHP functions. Default/optional
arguments are also allowed. An interface file like this :</p>
<div class="code">
<pre>
%module example
int foo(int a);
double bar(double, double b = 3.0);
...
</pre>
</div>
<p> Will be accessed in PHP like this :</p>
<div class="code">
<pre>
$a = foo(2);
$b = bar(3.5, -1.5);
$c = bar(3.5); # Use default argument for 2nd parameter
</pre>
</div>
<p> SWIG generates PHP type declarations for function parameters and
return types for PHP 8 and later.</p>
<p> You can control the generation of PHP type declarations using the
&quot;php:type&quot; %feature. This has three settings:</p>
<ul>
<li>
<p>If unset or set to &quot;0&quot; then no type declarations are generated, e.g.:
<tt>%feature(&quot;php:type&quot;, &quot;0&quot;);</tt></p>
</li>
<li>
<p>If set to &quot;1&quot; then type declarations are generated for both
parameters and return types, e.g.: <tt>%feature(&quot;php:type&quot;, &quot;1&quot;);</tt></p>
</li>
<li>
<p>The default setting is &quot;compat&quot;, which is the same as &quot;1&quot; except no
return type declarations are generated for virtual methods for which
directors are enabled. This provides better compatibility for PHP
subclasses of wrapped virtual methods in existing SWIG-generated
bindings, e.g.: <tt>%feature(&quot;php:type&quot;, &quot;compat&quot;);</tt></p>
</li>
</ul>
<p> If you have an existing PHP interface and are upgrading to SWIG &gt;=
4.1.0 then the default &quot;compat&quot; setting should work well.</p>
<p> If you're writing a new set of bindings and<b> only targeting PHP8
or newer</b> then enabling type declarations everywhere probably makes
sense. It will only actually make a difference if you enable directors
and are wrapping C++ classes with virtual methods, but doing it anyway
means you won't forget to if the code you are wrapping later evolves to
have such classes and methods.</p>
<p> The type declaration information will make the generated source code
and compiler extension module larger, so you might want to turn off
type declarations if keeping these small is important to you. If you
find you need to turn off type declarations to fix a problem, please
let us know via our github issue tracker.</p>
<p> Note that being a SWIG feature this can be specified globally (like
above) or per class, per method, etc. See the <a href="#Customization_features">
%feature directives</a> section for full details of how to control at a
fine-grained level.</p>
<p> The PHP type information is specified via a &quot;phptype&quot; attribute on
&quot;in&quot; and &quot;out&quot; typemaps, and these have been added for all the typemaps
we supply for PHP. We don't currently support this for &quot;argout&quot;
templates, but probably will in a future version.</p>
<p> If you have written custom SWIG typemaps for PHP and want to add PHP
type declarations, then the syntax is very like how you'd specify the
type in PHP code, e.g. <tt>%typemap(in, phptype=&quot;int|string|Foo&quot;)</tt>
means the typemap accepts a PHP int or string or an object of class
Foo, <tt>%typemap(in, phptype=&quot;?int&quot;)</tt> means a PHP int or NULL,
etc. As well as the standard PHP type declaration types, SWIG also
understands the special type &quot;SWIGTYPE&quot; as an entry in phptype, which
means the PHP type corresponding to the type that this typemap matched
on - for a object this will give you the PHP class for the object, and
for a pointer to a non-class type it will give you the name of the PHP
class SWIG created for that pointer type.</p>
<!-- This isn't correct for 1.3.30 and needs rewriting to reflect reality
&lt;p&gt;
Because PHP is a dynamically typed language, the default typemaps
used for simple types will attempt to coerce the arguments into the appropriate type. That is the following invocations are equivalent:
&lt;/p&gt;
&lt;div class=&quot;code&quot;&gt;&lt;pre&gt;
$a = foo(2);
$a = foo(&quot;2&quot;);
$a = foo(2.0);
&lt;/pre&gt;&lt;/div&gt;
&lt;p&gt;
Functions are invoked using pass by value semantics like all of PHP.
This means the conversion which automatically takes place when
invoking a swig wrapped method does not change the native type of the
argument variable.
&lt;/p&gt;
&lt;div class=&quot;code&quot;&gt;&lt;pre&gt;
$s = &quot;2 A string representing two&quot;;
$a = foo($s); # invokes 'foo(2)';
print $s; # The value of $s was not changed.
&lt;/pre&gt;&lt;/div&gt;
-->
<h3><a name="Php_nn2_4">32.2.4 Overloading</a></h3>
<p> Although PHP does not support overloading functions natively, swig
will generate dispatch functions which will use <tt>%typecheck</tt>
typemaps to allow overloading. This dispatch function's operation and
precedence is described in <a href="#SWIGPlus_overloaded_methods">
Overloaded functions and methods</a>.</p>
<!-- This isn't correct for 1.3.30 and needs rewriting to reflect reality
&lt;p&gt;
Because PHP is a dynamically typed language, simple values can be
silently converted from one type to another. For example, integers,
doubles and strings silently convert to each other depending on
context. This situation make overloading slightly problematic because
given the following function:
&lt;/p&gt;
&lt;div class=&quot;code&quot;&gt;&lt;pre&gt;
void doit( int i );
void doit( double i );
&lt;/pre&gt;&lt;/div&gt;
&lt;p&gt;
it is questionable which to invoke when &lt;tt&gt;doit(&quot;2&quot;);&lt;/tt&gt; is used in
PHP. The string &lt;tt&gt;&quot;2&quot;&lt;/tt&gt; simultaneously represents the integer
&lt;tt&gt;2&lt;/tt&gt; and the double &lt;tt&gt;2.0&lt;/tt&gt;.
&lt;/p&gt;
&lt;p&gt;
In order to provide the most natural experience to PHP programmers,
the default &lt;tt&gt;%typecheck&lt;/tt&gt; implemented in &lt;tt&gt;php.swg&lt;/tt&gt;
allows any simple type (integer, double, string) in PHP to be used for
any simple C type (int, double, char *). The function selected then
depends only on the argument type precedence defined by SWIG.
&lt;/p&gt;
&lt;p&gt;
It should be noted that &lt;tt&gt;SWIGTYPE&lt;/tt&gt; references and pointers will
not be silently converted. So these two functions:
&lt;/p&gt;
&lt;div class=&quot;code&quot;&gt;&lt;pre&gt;
void doit( const Vector &amp; );
void doit( int i );
&lt;/pre&gt;&lt;/div&gt;
&lt;p&gt;
Cause less confusion and &lt;tt&gt;doit(&quot;2&quot;);&lt;/tt&gt; will invoke the function
taking the integer argument.
&lt;/p&gt;
-->
<h3><a name="Php_nn2_5">32.2.5 Pointers and References</a></h3>
<p> Since SWIG 4.1.0, SWIG wraps C/C++ classes directly with PHP
objects. Pointers to other types are also wrapped as PHP objects -
mostly this is an implementation detail, but it's visible from PHP via <tt>
is_object()</tt> and similar. In earlier SWIG versions, PHP resources
were used to wrap both classes and pointers to other types.</p>
<p> There are multiple ways to wrap pointers to simple types. Given the
following C method:</p>
<div class="code">
<pre>
void add( int *in1, int *in2, int *result);
</pre>
</div>
<p> One can include<b> cpointer.i</b> to generate PHP wrappers to <tt>
int *</tt>.</p>
<div class="code">
<pre>
%module example
%include &quot;cpointer.i&quot;
%pointer_functions(int, intp)
void add( int *in1, int *in2, int *result);
</pre>
</div>
<p> This will result in the following usage in PHP:</p>
<div class="code">
<pre>
&lt;?php
$in1=copy_intp(3);
$in2=copy_intp(5);
$result=new_intp();
add( $in1, $in2, $result );
echo &quot;The sum &quot; . intp_value($in1) . &quot; + &quot; . intp_value($in2) . &quot; = &quot; . intp_value( $result) . &quot;\n&quot;;
</pre>
</div>
<p> An alternative would be to use the include<b> typemaps.i</b> which
defines named typemaps for INPUT, OUTPUT and INOUT variables. One needs
to either <tt>%apply</tt> the appropriate typemap or adjust the
parameter names as appropriate.</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
void add( int *INPUT, int *INPUT, int *OUTPUT);
</pre>
</div>
<p> This will result in the following usage in PHP:</p>
<div class="code">
<pre>
&lt;?php
$in1 = 3;
$in2 = 5;
$result= add($in1, $in2); # Note using variables for the input is unnecessary.
echo &quot;The sum $in1 + $in2 = $result\n&quot;;
</pre>
</div>
<p> Because PHP has a native concept of reference, it may seem more
natural to the PHP developer to use references to pass pointers. To
enable this, one needs to include<b> phppointers.i</b> which defines
the named typemap REF.</p>
<p> In case you write your own typemaps, SWIG supports an attribute
called <tt>byref</tt>: if you set that, then SWIG will make sure that
the generated wrapper function will want the input parameter as a
reference.</p>
<div class="code">
<pre>
%module example
%include &quot;phppointers.i&quot;
void add( int *REF, int *REF, int *REF);
</pre>
</div>
<p> This will result in the following usage in PHP:</p>
<div class="code">
<pre>
&lt;?php
$in1 = 3;
$in2 = 5;
$result = 0;
add($in1, $in2, $result);
echo &quot;The sum $in1 + $in2 = $result\n&quot;;
</pre>
</div>
<p> It is important to note that a php variable which is NULL when
passed by reference would end up passing a NULL pointer into the
function. In PHP, an unassigned variable (i.e. where the first
reference to the variable is not an assignment) is NULL. In the above
example, if any of the three variables had not been assigned, a NULL
pointer would have been passed into <tt>add</tt>. Depending on the
implementation of the function, this may or may not be a good thing.</p>
<p> We chose to allow passing NULL pointers into functions because that
is sometimes required in C libraries. A NULL pointer can be created in
PHP in a number of ways: by using <tt>unset</tt> on an existing
variable, or assigning <tt>NULL</tt> to a variable.</p>
<h3><a name="Php_nn2_6">32.2.6 Structures and C++ classes</a></h3>
<p> SWIG wraps C++ structs and classes with PHP classes. Since SWIG
4.1.0, this is done entirely via PHP's C API - earlier SWIG versions
generated a PHP wrapper script which defined proxy classes which called
a set of flat functions which actually wrapped the C++ class.</p>
<p> This interface file</p>
<div class="code">
<pre>
%module vector
class Vector {
public:
double x, y, z;
Vector();
~Vector();
double magnitude();
};
struct Complex {
double re, im;
};
</pre>
</div>
<p> Would be used in the following way from PHP:</p>
<div class="code">
<pre>
&lt;?php
$v = new Vector();
$v-&gt;x = 3;
$v-&gt;y = 4;
$v-&gt;z = 5;
echo &quot;Magnitude of ($v-&gt;x, $v-&gt;y, $v-&gt;z) = &quot; . $v-&gt;magnitude() . &quot;\n&quot;;
$v = NULL; # destructor called.
$c = new Complex();
$c-&gt;re = 0;
$c-&gt;im = 0;
# $c destructor called when $c goes out of scope.
</pre>
</div>
<p> Member variables and methods are accessed using the <tt>-&gt;</tt>
operator.</p>
<h4><a name="Php_nn2_6_1">32.2.6.1 Using -noproxy</a></h4>
<p> SWIG/PHP used to support a <tt>-noproxy</tt> option to flatten the
class structure and generate collections of named flat functions. This
is no longer supported as of SWIG 4.1.0.</p>
<h4><a name="Php_nn2_6_2">32.2.6.2 Constructors and Destructors</a></h4>
<p> The constructor is called when <tt>new Object()</tt> is used to
create an instance of the object. If multiple constructors are defined
for an object, function overloading will be used to determine which
constructor to execute.</p>
<p> Because PHP uses reference counting, simple assignment of one
variable to another such as:</p>
<div class="code">
<pre>
$ref = $v;
</pre>
</div>
<p> causes the symbol <tt>$ref</tt> to refer to the same underlying
object as <tt>$v</tt>. This does not result in a call to the C++ copy
constructor or copy assignment operator.</p>
<p> One can force execution of the copy constructor by using:</p>
<div class="code">
<pre>
$o_copy = new Object($o);
</pre>
</div>
<p> Destructors are automatically called when all variables referencing
the instance are reassigned or go out of scope. The destructor is not
available to be called manually. To force a destructor to be called the
programmer can either reassign the variable or call <tt>unset($v)</tt></p>
<h4><a name="Php_nn2_6_3">32.2.6.3 Static Member Variables</a></h4>
<p> Static member variables in C++ are not wrapped as such in PHP as it
does not appear to be possible to intercept accesses to such variables.
Therefore, static member variables are wrapped using a class function
with the same name, which returns the current value of the class
variable. For example</p>
<div class="code">
<pre>
%module example
class Ko {
static int threats;
};
</pre>
</div>
<p> would be accessed in PHP as,</p>
<div class="code">
<pre>
echo &quot;There have now been &quot; . Ko::threats() . &quot; threats\n&quot;;
</pre>
</div>
<p> To set the static member variable, pass the value as the argument to
the class function, e.g.</p>
<div class="code">
<pre>
Ko::threats(10);
echo &quot;There have now been &quot; . Ko::threats() . &quot; threats\n&quot;;
</pre>
</div>
<h4><a name="Php_nn2_6_4">32.2.6.4 Static Member Functions</a></h4>
<p> Static member functions are supported in PHP using the <tt>
class::function()</tt> syntax. For example</p>
<div class="code">
<pre>
%module example
class Ko {
static void threats();
};
</pre>
</div>
<p> would be executed in PHP as</p>
<div class="code">
<pre>
Ko::threats();
</pre>
</div>
<h4><a name="Php_nn2_6_5">32.2.6.5 Specifying Implemented Interfaces</a></h4>
<p> PHP supports the concept of abstract interfaces which a class can
implement. Since SWIG 3.0.3, you can tell SWIG that a wrapped class
(for example <code>MyIterator</code>) implements the <code>Iterator</code>
interface like so:</p>
<div class="code">
<pre>
%typemap(&quot;phpinterfaces&quot;) MyIterator &quot;Iterator&quot;
</pre>
</div>
<p> If there are multiple interfaces, just list them separated by
commas.</p>
<h4><a name="Php_nn2_6_6">32.2.6.6 Dynamic Properties</a></h4>
<p> Historically PHP has supported dynamic class properties and SWIG has
implemented them too (because we implement the magic <tt>__get()</tt>, <tt>
__set()</tt> and <tt>__isset()</tt> methods we need to include explicit
handling).</p>
<p> PHP 8.2 <a href="https://wiki.php.net/rfc/deprecate_dynamic_properties">
deprecates dynamic class properties</a> - initially they'll warn, and
apparently they'll not work by default in PHP 9.0.</p>
<p> In PHP code dynamic properties can be enabled for a class by marking
that class with the attribute <tt>#[AllowDynamicProperties]</tt>.</p>
<p> To follow this PHP change, as of SWIG 4.1.0 you now need enable
dynamic properties for any classes you want to support them. To enable
for class <tt>Foo</tt>:</p>
<div class="code">
<pre>
%feature(&quot;php:allowdynamicproperties&quot;, 1) Foo;
</pre>
</div>
<p> or to enable them for all wrapped classes:</p>
<div class="code">
<pre>
%feature(&quot;php:allowdynamicproperties&quot;, 1);
</pre>
</div>
<p> Note that unknown features are ignored, so you can add use these
unconditionally in your interface file and it'll work with older SWIG
too.</p>
<p> There was a bug in SWIG 4.1.0 where setting this feature to any
value enabled it - SWIG 4.2.0 fixed this and you can now set it to 0 to
turn it off (for example, you might want to enabled it for everything
and then selectively turn it off for specific classes).</p>
<h3><a name="Php_nn2_7">32.2.7 PHP Pragmas, Startup and Shutdown code</a>
</h3>
<p> You can get SWIG to generate an &quot;example.php&quot; file by specifying the
code to put in it using the<b> code</b> pragma.</p>
<div class="code">
<pre>
%module example
%pragma(php) code=&quot;
# This code is inserted into example.php
echo \&quot;example.php execution\\n\&quot;;
&quot;
</pre>
</div>
<p> Results in the following in &quot;example.php&quot;</p>
<div class="code">
<pre>
# This code is inserted into example.php
echo &quot;example.php execution\n&quot;;
</pre>
</div>
<p> The<b> version</b> pragma can be used to add version to generated
PHP extension module. The version is inserted in the zend_module_entry
block.</p>
<div class="code">
<pre>
%module example
%pragma(php) version=&quot;1.5&quot;
</pre>
</div>
<p> The<b> include</b> pragma is a short cut to add include statements
to the example.php file.</p>
<div class="code">
<pre>
%module example
%pragma(php) code=&quot;
include \&quot;include.php\&quot;;
&quot;
%pragma(php) include=&quot;include.php&quot; // equivalent.
</pre>
</div>
<p> The<b> phpinfo</b> pragma inserts code in the <tt>PHP_MINFO_FUNCTION</tt>
which is called from PHP's phpinfo() function.</p>
<div class="code">
<pre>
%module example;
%pragma(php) phpinfo=&quot;
zend_printf(&quot;An example of PHP support through SWIG\n&quot;);
php_info_print_table_start();
php_info_print_table_header(2, \&quot;Directive\&quot;, \&quot;Value\&quot;);
php_info_print_table_row(2, \&quot;Example support\&quot;, \&quot;enabled\&quot;);
php_info_print_table_end();
&quot;
</pre>
</div>
<p> To insert code into the <tt>PHP_MINIT_FUNCTION</tt>, one can use
either <tt>%init</tt> or <tt>%minit</tt>.</p>
<div class="code">
<pre>
%module example;
%init {
zend_printf(&quot;Inserted into PHP_MINIT_FUNCTION\n&quot;);
}
%minit {
zend_printf(&quot;Inserted into PHP_MINIT_FUNCTION\n&quot;);
}
</pre>
</div>
<p> To insert code into the <tt>PHP_MSHUTDOWN_FUNCTION</tt>, one can use
either <tt>%shutdown</tt> or <tt>%mshutdown</tt>.</p>
<div class="code">
<pre>
%module example;
%mshutdown {
zend_printf(&quot;Inserted into PHP_MSHUTDOWN_FUNCTION\n&quot;);
}
</pre>
</div>
<p> The <tt>%rinit</tt> and <tt>%rshutdown</tt> statements are very
similar but insert code into the request init (PHP_RINIT_FUNCTION) and
request shutdown (PHP_RSHUTDOWN_FUNCTION) code respectively.</p>
<h2><a name="Php_nn3">32.3 Cross language polymorphism</a></h2>
<p> Proxy classes provide a more natural, object-oriented way to access
extension classes. As described above, each proxy instance has an
associated C++ instance, and method calls to the proxy are passed to
the C++ instance transparently.</p>
<p> This arrangement is asymmetric in the sense that no corresponding
mechanism exists to pass method calls down the inheritance chain from
C++ to PHP. In particular, if a C++ class has been extended in PHP (by
extending the proxy class), these extensions will not be visible from
C++ code. Virtual method calls from C++ are thus not able access the
lowest implementation in the inheritance chain.</p>
<p> Changes have been made to SWIG 1.3.18 to address this problem and
make the relationship between C++ classes and proxy classes more
symmetric. To achieve this goal, new classes called directors are
introduced at the bottom of the C++ inheritance chain. Support for
generating PHP classes has been added in SWIG 1.3.40. The job of the
directors is to route method calls correctly, either to C++
implementations higher in the inheritance chain or to PHP
implementations lower in the inheritance chain. The upshot is that C++
classes can be extended in PHP and from C++ these extensions look
exactly like native C++ classes. Neither C++ code nor PHP code needs to
know where a particular method is implemented: the combination of proxy
classes, director classes, and C wrapper functions takes care of all
the cross-language method routing transparently.</p>
<h3><a name="Php_nn3_1">32.3.1 Enabling directors</a></h3>
<p> The director feature is disabled by default. To use directors you
must make two changes to the interface file. First, add the &quot;directors&quot;
option to the %module directive, like this:</p>
<div class="code">
<pre>
%module(directors=&quot;1&quot;) modulename
</pre>
</div>
<p> Without this option no director code will be generated. Second, you
must use the %feature(&quot;director&quot;) directive to tell SWIG which classes
and methods should get directors. The %feature directive can be applied
globally, to specific classes, and to specific methods, like this:</p>
<div class="code">
<pre>
// generate directors for all classes that have virtual methods
%feature(&quot;director&quot;);
// generate directors for the virtual methods in class Foo
%feature(&quot;director&quot;) Foo;
</pre>
</div>
<p> You can use the %feature(&quot;nodirector&quot;) directive to turn off
directors for specific classes or methods. So for example,</p>
<div class="code">
<pre>
%feature(&quot;director&quot;) Foo;
%feature(&quot;nodirector&quot;) Foo::bar;
</pre>
</div>
<p> will generate directors for the virtual methods of class Foo except
bar().</p>
<p> Directors can also be generated implicitly through inheritance. In
the following, class Bar will get a director class that handles the
methods one() and two() (but not three()):</p>
<div class="code">
<pre>
%feature(&quot;director&quot;) Foo;
class Foo {
public:
Foo(int foo);
virtual void one();
virtual void two();
};
class Bar: public Foo {
public:
virtual void three();
};
</pre>
</div>
<p> then at the PHP side you can define</p>
<div class="targetlang">
<pre>
class MyFoo extends Foo {
function one() {
print &quot;one from php\n&quot;;
}
}
</pre>
</div>
<h3><a name="Php_nn3_2">32.3.2 Director classes</a></h3>
<p> For each class that has directors enabled, SWIG generates a new
class that derives from both the class in question and a special <tt>
Swig::Director</tt> class. These new classes, referred to as director
classes, can be loosely thought of as the C++ equivalent of the PHP
proxy classes. The director classes store a pointer to their underlying
PHP object. Indeed, this is quite similar to <tt>struct
swig_object_wrapper</tt> which is used to implement the PHP proxy
classes.</p>
<p> For simplicity let's ignore the <tt>Swig::Director</tt> class and
refer to the original C++ class as the director's base class. By
default, a director class extends all virtual methods in the
inheritance chain of its base class (see the preceding section for how
to modify this behavior). Virtual methods that have a final specifier
are unsurprisingly excluded. Thus the virtual method calls, whether
they originate in C++ or in PHP via proxy classes, eventually end up in
at the implementation in the director class. The job of the director
methods is to route these method calls to the appropriate place in the
inheritance chain. By &quot;appropriate place&quot; we mean the method that would
have been called if the C++ base class and its extensions in PHP were
seamlessly integrated. That seamless integration is exactly what the
director classes provide, transparently skipping over all the messy
extension API glue that binds the two languages together.</p>
<p> In reality, the &quot;appropriate place&quot; is one of only two
possibilities: C++ or PHP. Once this decision is made, the rest is
fairly easy. If the correct implementation is in C++, then the lowest
implementation of the method in the C++ inheritance chain is called
explicitly. If the correct implementation is in PHP, the Zend API is
used to call the method of the underlying PHP object (after which the
usual virtual method resolution in PHP automatically finds the right
implementation).</p>
<p> Now how does the director decide which language should handle the
method call? The basic rule is to handle the method in PHP, unless
there's a good reason not to. The reason for this is simple: PHP has
the most &quot;extended&quot; implementation of the method. This assertion is
guaranteed, since at a minimum the PHP proxy class implements the
method. If the method in question has been extended by a class derived
from the proxy class, that extended implementation will execute exactly
as it should. If not, the proxy class will route the method call into a
C wrapper function, expecting that the method will be resolved in C++.
The wrapper will call the virtual method of the C++ instance, and since
the director extends this the call will end up right back in the
director method. Now comes the &quot;good reason not to&quot; part. If the
director method were to blindly call the PHP method again, it would get
stuck in an infinite loop. We avoid this situation by adding special
code to the C wrapper function that tells the director method to not do
this. The C wrapper function compares the called and the declaring
class name of the given method. If these are not the same, then the C
wrapper function tells the director to resolve the method by calling up
the C++ inheritance chain, preventing an infinite loop.</p>
<p> One more point needs to be made about the relationship between
director classes and proxy classes. When a proxy class instance is
created in PHP, SWIG creates an instance of the original C++ class and
stores it in the <tt>struct swig_object_wrapper</tt>. This is true
whether or not directors are enabled for the particular class in
question. However when a class<i> derived</i> from a proxy class is
created, SWIG instead creates an instance of the corresponding C++
director class. The reason for this difference is that user-defined
subclasses may override or extend methods of the original class, so the
director class is needed to route calls to these methods correctly. For
unmodified proxy classes, all methods are ultimately implemented in C++
so there is no need for the extra overhead involved with routing the
calls through PHP.</p>
<h3><a name="Php_nn3_3">32.3.3 Ownership and object destruction</a></h3>
<p> Memory management issues are slightly more complicated with
directors than for proxy classes alone. PHP instances hold a pointer to
the associated C++ director object, and the director in turn holds a
pointer back to the PHP object. By default, proxy classes own their C++
director object and take care of deleting it when they are garbage
collected.</p>
<p> This relationship can be reversed by calling the special <tt>
-&gt;thisown</tt> property of the proxy class. After setting this property
to <tt>0</tt>, the director class no longer destroys the PHP object.
Assuming no outstanding references to the PHP object remain, the PHP
object will be destroyed at the same time. This is a good thing, since
directors and proxies refer to each other and so must be created and
destroyed together. Destroying one without destroying the other will
likely cause your program to segfault.</p>
<p> Here is an example:</p>
<div class="code">
<pre>
class Foo {
public:
...
};
class FooContainer {
public:
void addFoo(Foo *);
...
};
</pre>
</div>
<br><div class="targetlang">
<pre>
$c = new FooContainer();
$a = new Foo();
$a-&gt;thisown = 0;
$c-&gt;addFoo($a);
</pre>
</div>
<p> In this example, we are assuming that FooContainer will take care of
deleting all the Foo pointers it contains at some point.</p>
<h3><a name="Php_nn3_4">32.3.4 Exception unrolling</a></h3>
<p> With directors routing method calls to PHP, and proxies routing them
to C++, the handling of exceptions is an important concern. By default,
an exception thrown in PHP code called from C++ causes the PHP
interpreter to flag that an exception is thrown, then return passes to
C++ as if the PHP function had returned <code>Null</code>. Assuming the
directorout typemaps handle this (those SWIG defines by default should)
then once control returns to PHP code again, the PHP exception will
actually propagate.</p>
<p> Sometimes this control flow is problematic, and you want to skip any
handling in the C++ code. To achieve this, it is necessary to
temporarily translate the PHP exception into a C++ exception. This can
be achieved using the %feature(&quot;director:except&quot;) directive. The
following code should suffice in most cases:</p>
<div class="code">
<pre>
%feature(&quot;director:except&quot;) {
#if SWIG_VERSION &gt;= 0x040100
if ($error != NULL)
#else
if ($error == FAILURE)
#endif
{
throw Swig::DirectorMethodException();
}
}
</pre>
</div>
<p> If you only need to support SWIG &gt;= 4.1.0, you can just use the <tt>
($error != NULL)</tt> condition.</p>
<p> In SWIG 4.1.0, <tt>$error</tt> was changed in the SWIG/PHP director
implementation to make it work more like how it does for other
languages. Previously, <tt>$error</tt> didn't actually indicate an
exception, but instead was only set to <tt>FAILURE</tt> if there was a
problem calling the PHP method. Now <tt>$error</tt> indicates if the
PHP method threw a PHP exception, and directorout typemaps for PHP no
longer need to be gated by <tt>if (EG(exception))</tt>.</p>
<p> This code will check the PHP error state after each method call from
a director into PHP, and throw a C++ exception if an error occurred.
This exception can be caught in C++ to implement an error handler.
Currently no information about the PHP error is stored in the
Swig::DirectorMethodException object, but this will likely change in
the future.</p>
<p> It may be the case that a method call originates in PHP, travels up
to C++ through a proxy class, and then back into PHP via a director
method. If an exception occurs in PHP at this point, it would be nice
for that exception to find its way back to the original caller. This
can be done by combining a normal %exception directive with the <tt>
director:except</tt> handler shown above. Here is an example of a
suitable exception handler:</p>
<div class="code">
<pre>
%exception {
try { $action }
catch (Swig::DirectorException &amp;e) { SWIG_fail; }
}
</pre>
</div>
<p> The class Swig::DirectorException used in this example is actually a
base class of Swig::DirectorMethodException, so it will trap this
exception. Because the PHP error state is still set when
Swig::DirectorMethodException is thrown, PHP will register the
exception as soon as the C wrapper function returns.</p>
<h3><a name="Php_nn3_5">32.3.5 Overhead and code bloat</a></h3>
<p> Enabling directors for a class will generate a new director method
for every virtual method in the class' inheritance chain. This alone
can generate a lot of code bloat for large hierarchies. Method
arguments that require complex conversions to and from target language
types can result in large director methods. For this reason it is
recommended that you selectively enable directors only for specific
classes that are likely to be extended in PHP and used in C++.</p>
<p> Compared to classes that do not use directors, the call routing in
the director methods does add some overhead. In particular, at least
one dynamic cast and one extra function call occurs per method call
from PHP. Relative to the speed of PHP execution this is probably
completely negligible. For worst case routing, a method call that
ultimately resolves in C++ may take one extra detour through PHP in
order to ensure that the method does not have an extended PHP
implementation. This could result in a noticeable overhead in some
cases.</p>
<p> Although directors make it natural to mix native C++ objects with
PHP objects (as director objects) via a common base class pointer, one
should be aware of the obvious fact that method calls to PHP objects
will be much slower than calls to C++ objects. This situation can be
optimized by selectively enabling director methods (using the %feature
directive) for only those methods that are likely to be extended in
PHP.</p>
<h3><a name="Php_nn3_6">32.3.6 Typemaps</a></h3>
<p> Typemaps for input and output of most of the basic types from
director classes have been written. These are roughly the reverse of
the usual input and output typemaps used by the wrapper code. The
typemap operation names are 'directorin', 'directorout', and
'directorargout'. The director code does not currently use any of the
other kinds of typemaps. It is not clear at this point which kinds are
appropriate and need to be supported.</p>
<h3><a name="Php_nn3_7">32.3.7 Miscellaneous</a></h3>
<p> Director typemaps for STL classes are mostly in place, and hence you
should be able to use std::string, etc., as you would any other type.</p>
<HR NOSHADE>
<h1><a name="Python">33 SWIG and Python</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Python_nn2">Overview</a></li>
<li><a href="#Python_nn3">Preliminaries</a>
<ul>
<li><a href="#Python_nn4">Running SWIG</a></li>
<li><a href="#Python_nn6">Using distutils</a></li>
<li><a href="#Python_nn7">Hand compiling a dynamic module</a></li>
<li><a href="#Python_nn8">Static linking</a></li>
<li><a href="#Python_nn9">Using your module</a></li>
<li><a href="#Python_nn10">Compilation of C++ extensions</a></li>
<li><a href="#Python_nn11">Compiling for 64-bit platforms</a></li>
<li><a href="#Python_nn12">Building Python extensions under Windows</a></li>
<li><a href="#Python_commandline">Additional Python commandline options</a>
</li>
</ul>
</li>
<li><a href="#Python_nn13">A tour of basic C/C++ wrapping</a>
<ul>
<li><a href="#Python_nn14">Modules</a></li>
<li><a href="#Python_nn15">Functions</a></li>
<li><a href="#Python_nn16">Global variables</a></li>
<li><a href="#Python_nn17">Constants and enums</a></li>
<li><a href="#Python_nn18">Pointers</a></li>
<li><a href="#Python_nn19">Structures</a></li>
<li><a href="#Python_nn20">C++ classes</a></li>
<li><a href="#Python_nn21">C++ inheritance</a></li>
<li><a href="#Python_nn22">Pointers, references, values, and arrays</a></li>
<li><a href="#Python_nn23">C++ overloaded functions</a></li>
<li><a href="#Python_nn24">C++ operators</a></li>
<li><a href="#Python_nn25">C++ namespaces</a></li>
<li><a href="#Python_nn26">C++ templates</a></li>
<li><a href="#Python_nn27">C++ Smart Pointers</a>
<ul>
<li><a href="#Python_smart_pointers_shared_ptr">The shared_ptr Smart
Pointer</a></li>
<li><a href="#Python_smart_pointers_generic">Generic Smart Pointers</a></li>
</ul>
</li>
<li><a href="#Python_nn27a">C++ reference counted objects</a></li>
</ul>
</li>
<li><a href="#Python_nn28">Further details on the Python class interface</a>
<ul>
<li><a href="#Python_nn29">Proxy classes</a></li>
<li><a href="#Python_builtin_types">Built-in Types</a>
<ul>
<li><a href="#Python_builtin_limitations">Limitations</a></li>
<li><a href="#Python_builtin_overloads">Operator overloads and slots --
use them!</a></li>
</ul>
</li>
<li><a href="#Python_nn30">Memory management</a></li>
</ul>
</li>
<li><a href="#Python_directors">Cross language polymorphism</a>
<ul>
<li><a href="#Python_nn33">Enabling directors</a></li>
<li><a href="#Python_nn34">Director classes</a></li>
<li><a href="#Python_nn35">Ownership and object destruction</a></li>
<li><a href="#Python_nn36">Exception unrolling</a></li>
<li><a href="#Python_nn37">Overhead and code bloat</a></li>
<li><a href="#Python_nn38">Typemaps</a></li>
<li><a href="#Python_director_thread_safety">Thread safety</a></li>
<li><a href="#Python_nn39">Miscellaneous</a></li>
</ul>
</li>
<li><a href="#Python_nn40">Common customization features</a>
<ul>
<li><a href="#Python_nn41">C/C++ helper functions</a></li>
<li><a href="#Python_nn42">Adding additional Python code</a></li>
<li><a href="#Python_nn43">Class extension with %extend</a></li>
<li><a href="#Python_nn44">Exception handling with %exception</a></li>
<li><a href="#Python_optimization">Optimization options</a>
<ul>
<li><a href="#Python_fastproxy">-fastproxy</a></li>
</ul>
</li>
<li><a href="#Python_stable_abi">Stable ABI</a></li>
</ul>
</li>
<li><a href="#Python_nn45">Tips and techniques</a>
<ul>
<li><a href="#Python_nn46">Input and output parameters</a></li>
<li><a href="#Python_nn47">Simple pointers</a></li>
<li><a href="#Python_nn48">Unbounded C Arrays</a></li>
<li><a href="#Python_nn49">String handling</a></li>
<li><a href="#Python_default_args">Default arguments</a></li>
</ul>
</li>
<li><a href="#Python_nn53">Typemaps</a>
<ul>
<li><a href="#Python_nn54">What is a typemap?</a></li>
<li><a href="#Python_nn55">Python typemaps</a></li>
<li><a href="#Python_nn56">Typemap variables</a></li>
<li><a href="#Python_nn57">Useful Python Functions</a></li>
</ul>
</li>
<li><a href="#Python_nn58">Typemap Examples</a>
<ul>
<li><a href="#Python_nn59">Converting a Python list to a char **</a></li>
<li><a href="#Python_nn60">Expanding a Python object into multiple
arguments</a></li>
<li><a href="#Python_nn61">Using typemaps to return arguments</a></li>
<li><a href="#Python_nn62">Mapping Python tuples into small arrays</a></li>
<li><a href="#Python_nn63">Mapping sequences to C arrays</a></li>
<li><a href="#Python_nn64">Pointer handling</a></li>
<li><a href="#Python_memory_management_member_variables">Memory
management when returning references to member variables</a></li>
</ul>
</li>
<li><a href="#Python_nn65">Docstring Features</a>
<ul>
<li><a href="#Python_nn66">Module docstring</a></li>
<li><a href="#Python_nn67">%feature(&quot;autodoc&quot;)</a>
<ul>
<li><a href="#Python_nn68">%feature(&quot;autodoc&quot;, &quot;0&quot;)</a></li>
<li><a href="#Python_nn69">%feature(&quot;autodoc&quot;, &quot;1&quot;)</a></li>
<li><a href="#Python_autodoc2">%feature(&quot;autodoc&quot;, &quot;2&quot;)</a></li>
<li><a href="#Python_autodoc3">%feature(&quot;autodoc&quot;, &quot;3&quot;)</a></li>
<li><a href="#Python_nn70">%feature(&quot;autodoc&quot;, &quot;docstring&quot;)</a></li>
</ul>
</li>
<li><a href="#Python_nn71">%feature(&quot;docstring&quot;)</a></li>
<li><a href="#Python_doxygen_docstrings">Doxygen comments</a></li>
</ul>
</li>
<li><a href="#Python_nn72">Python Packages</a>
<ul>
<li><a href="#Python_modulepackage">Setting the Python package</a></li>
<li><a href="#Python_absrelimports">Absolute and relative imports</a></li>
<li><a href="#Python_absimport">Enforcing absolute import semantics</a></li>
<li><a href="#Python_importfrominit">Importing from __init__.py</a></li>
<li><a href="#Python_implicit_namespace_packages">Implicit namespace
packages</a></li>
<li><a href="#Python_package_search">Location of modules</a>
<ul>
<li><a href="#Python_package_search_both_package_modules">Both modules
in the same package</a></li>
<li><a href="#Python_package_search_both_global_modules">Both modules
are global</a></li>
<li><a href="#Python_package_search_wrapper_split">Split modules custom
configuration</a></li>
<li><a href="#Python_custom_module_import">More on customizing the
module import code</a></li>
<li><a href="#Python_package_search_static">Statically linked C modules</a>
</li>
</ul>
</li>
</ul>
</li>
<li><a href="#Python_python3support">Python 3 Support</a>
<ul>
<li><a href="#Python_annotations">Python function annotations and
variable annotations</a>
<ul>
<li><a href="#Python_annotations_c">C/C++ annotation types</a></li>
</ul>
</li>
<li><a href="#Python_nn75">Buffer interface</a></li>
<li><a href="#Python_nn76">Abstract base classes</a></li>
<li><a href="#Python_nn77">Byte string output conversion</a></li>
<li><a href="#Python_2_unicode">Python 2 Unicode</a></li>
</ul>
</li>
<li><a href="#Python_multithreaded">Support for Multithreaded
Applications</a>
<ul>
<li><a href="#Python_thread_UI">UI for Enabling Multithreading Support</a>
</li>
<li><a href="#Python_thread_performance">Multithread Performance</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> This chapter describes SWIG's support of Python. SWIG is compatible
with all recent Python versions (Python 2.7 and Python &gt;= 3.5). Python
3.3 and 3.4 may still work but we are no longer regularly testing with
them. SWIG 4.0.x supported Python 3.2. SWIG 3.0.x supported older
Python 2.x and 3.x.</p>
<p> This chapter covers most SWIG features, but certain low-level
details are covered in less depth than in earlier chapters. At the very
least, make sure you read the &quot;<a href="#SWIG">SWIG Basics</a>&quot;
chapter.</p>
<h2><a name="Python_nn2">33.1 Overview</a></h2>
<p> To build Python extension modules, SWIG uses a layered approach in
which parts of the extension module are defined in C and other parts
are defined in Python. The C layer contains low-level wrappers whereas
Python code is used to define high-level features.</p>
<p> This layered approach recognizes the fact that certain aspects of
extension building are better accomplished in each language (instead of
trying to do everything in C or C++). Furthermore, by generating code
in both languages, you get a lot more flexibility since you can enhance
the extension module with support code in either language.</p>
<p> In describing the Python interface, this chapter starts by covering
the basics of configuration, compiling, and installing Python modules.
Next, the Python interface to common C and C++ programming features is
described. Advanced customization features such as typemaps are then
described followed by a discussion of low-level implementation details.</p>
<h2><a name="Python_nn3">33.2 Preliminaries</a></h2>
<h3><a name="Python_nn4">33.2.1 Running SWIG</a></h3>
<p> Suppose that you defined a SWIG module such as the following:</p>
<div class="code">
<pre>
/* File: example.i */
%module example
%{
#include &quot;example.h&quot;
%}
int fact(int n);
</pre>
</div>
<p> This <tt>.i</tt> file wraps the following simple C file:</p>
<div class="code">
<pre>
/* File: example.c */
#include &quot;example.h&quot;
int fact(int n) {
if (n &lt; 0) { /* This should probably return an error, but this is simpler */
return 0;
}
if (n == 0) {
return 1;
} else {
/* testing for overflow would be a good idea here */
return n * fact(n-1);
}
}
</pre>
</div>
<p> With the header file:</p>
<div class="code">
<pre>
/* File: example.h */
int fact(int n);
</pre>
</div>
<p> To build a Python module, run SWIG using the <tt>-python</tt>
option:</p>
<div class="shell">
<pre>
$ swig -python example.i
</pre>
</div>
<p> If building a C++ extension, add the <tt>-c++</tt> option:</p>
<div class="shell">
<pre>
$ swig -c++ -python example.i
</pre>
</div>
<p> This creates two different files; a C/C++ source file <tt>
example_wrap.c</tt> or <tt>example_wrap.cxx</tt> and a Python source
file <tt>example.py</tt>. The generated C source file contains the
low-level wrappers that need to be compiled and linked with the rest of
your C/C++ application to create an extension module. The Python source
file contains high-level support code. This is the file that you will
import to use the module.</p>
<p> The name of the wrapper file is derived from the name of the input
file. For example, if the input file is <tt>example.i</tt>, the name of
the wrapper file is <tt>example_wrap.c</tt>. To change this, you can
use the <tt>-o</tt> option. The name of the Python file is derived from
the module name specified with <tt>%module</tt>. If the module name is <tt>
example</tt>, then a file <tt>example.py</tt> is created.</p>
<p> The following sections have further practical examples and details
on how you might go about compiling and using the generated files.</p>
<h3><a name="Python_nn6">33.2.2 Using distutils</a></h3>
<p> The preferred approach to building an extension module for Python is
to compile it with distutils, which comes with all recent versions of
Python (<a href="https://docs.python.org/3/library/distutils.html">
Distutils Docs</a>).</p>
<p> Distutils takes care of making sure that your extension is built
with all the correct flags, headers, etc. for the version of Python it
is run with. Distutils will compile your extension into a shared object
file or DLL (<tt>.so</tt> on Linux, <tt>.pyd</tt> on Windows, etc). In
addition, distutils can handle installing your package into
site-packages, if that is desired. A configuration file (conventionally
called: <tt>setup.py</tt>) describes the extension (and related Python
modules). The distutils will then generate all the right compiler
directives to build it for you.</p>
<p> Here is a sample <tt>setup.py</tt> file for the above example:</p>
<div class="code">
<pre>
#!/usr/bin/env python3
&quot;&quot;&quot;
setup.py file for SWIG example
&quot;&quot;&quot;
from distutils.core import setup, Extension
example_module = Extension('_example',
sources=['example_wrap.c', 'example.c'],
)
setup (name = 'example',
version = '0.1',
author = &quot;SWIG Docs&quot;,
description = &quot;&quot;&quot;Simple swig example from docs&quot;&quot;&quot;,
ext_modules = [example_module],
py_modules = [&quot;example&quot;],
)
</pre>
</div>
<p> In this example, the line: <tt>example_module = Extension(....)</tt>
creates an Extension module object, defining the name as <tt>_example</tt>
, and using the source code files: <tt>example_wrap.c</tt>, generated by
swig, and <tt>example.c</tt>, your original c source. The swig (and
other Python extension modules) tradition is for the compiled extension
to have the name of the Python portion, prefixed by an underscore. If
the name of your Python module is &quot;<tt>example.py</tt>&quot;, then the name
of the corresponding object file will be&quot;<tt>_example.so</tt>&quot;</p>
<p> The <tt>setup</tt> call then sets up distutils to build your
package, defining some meta data, and passing in your extension module
object. Once this is saved as <tt>setup.py</tt>, you can build your
extension with these commands:</p>
<div class="shell">
<pre>
$ swig -python example.i
$ python setup.py build_ext --inplace
</pre>
</div>
<p> And a .so, or .pyd or... will be created for you. It will build a
version that matches the Python that you run the command with. Taking
apart the command line:</p>
<ul>
<li> <tt>python</tt> -- the version of Python you want to build for</li>
<li> <tt>setup.py</tt> -- the name of your setup script (it can be
called anything, but setup.py is the tradition)</li>
<li> <tt>build_ext</tt> -- telling distutils to build extensions</li>
<li> <tt>--inplace</tt> -- this tells distutils to put the extension lib
in the current dir. Otherwise, it will put it inside a build hierarchy,
and you'd have to move it to use it.</li>
</ul>
<p> The distutils have many other features, consult the Python distutils
docs for details.</p>
<p> This same approach works on all platforms if the appropriate
compiler is installed. (it can even build extensions to the standard
Windows Python using MingGW)</p>
<h3><a name="Python_nn7">33.2.3 Hand compiling a dynamic module</a></h3>
<p> While the preferred approach to building an extension module is to
use the distutils, some people like to integrate building extensions
with a larger build system, and thus may wish to compile their modules
without the distutils. To do this, you need to compile your program
using commands like this (shown for Linux):</p>
<div class="shell">
<pre>
$ swig -python example.i
$ gcc -O2 -fPIC -c example.c
$ gcc -O2 -fPIC -c example_wrap.c -I/usr/local/include/python2.5
$ gcc -shared example.o example_wrap.o -o _example.so
</pre>
</div>
<p> The exact commands for doing this vary from platform to platform.
However, SWIG tries to guess the right options when it is installed.
Therefore, you may want to start with one of the examples in the <tt>
SWIG/Examples/python</tt> directory. If that doesn't work, you will need
to read the man-pages for your compiler and linker to get the right set
of options. You might also check the <a href="https://github.com/swig/swig/wiki">
SWIG Wiki</a> for additional information.</p>
<p> When linking the module,<b> the name of the output file has to match
the name of the module prefixed by an underscore</b>. If the name of
your module is &quot;<tt>example</tt>&quot;, then the name of the corresponding
object file should be &quot;<tt>_example.so</tt>&quot; or &quot;<tt>_examplemodule.so</tt>
&quot;. The name of the module is specified using the <tt>%module</tt>
directive or the <tt>-module</tt> command line option.</p>
<p><b> Compatibility Note:</b> In SWIG-1.3.13 and earlier releases,
module names did not include the leading underscore. This is because
modules were normally created as C-only extensions without the extra
Python support file (instead, creating Python code was supported as an
optional feature). This has been changed in SWIG-1.3.14 and is
consistent with other Python extension modules. For example, the <tt>
socket</tt> module actually consists of two files; <tt>socket.py</tt>
and <tt>_socket.so</tt>. Many other built-in Python modules follow a
similar convention.</p>
<h3><a name="Python_nn8">33.2.4 Static linking</a></h3>
<p> An alternative approach to dynamic linking is to rebuild the Python
interpreter with your extension module added to it. In the past, this
approach was sometimes necessary due to limitations in dynamic loading
support on certain machines. However, the situation has improved
greatly over the last few years and you should not consider this
approach unless there is really no other option.</p>
<p> The usual procedure for adding a new module to Python involves
finding the Python source, adding an entry to the <tt>Modules/Setup</tt>
file, and rebuilding the interpreter using the Python Makefile.
However, newer Python versions have changed the build process. You may
need to edit the 'setup.py' file in the Python distribution instead.</p>
<p> In earlier versions of SWIG, the <tt>embed.i</tt> library file could
be used to rebuild the interpreter. For example:</p>
<div class="code">
<pre>
%module example
%inline %{
extern int fact(int);
extern int mod(int, int);
extern double My_variable;
%}
%include &quot;embed.i&quot; // Include code for a static version of Python
</pre>
</div>
<p> The <tt>embed.i</tt> library file includes supporting code that
contains everything needed to rebuild Python. To rebuild the
interpreter, you simply do something like this:</p>
<div class="shell">
<pre>
$ swig -python -lembed.i example.i
$ gcc example.c example_wrap.c \
-Xlinker -export-dynamic \
-DHAVE_CONFIG_H -I/usr/include/python2.7 \
-I/usr/lib/python2.7/config-x86_64-linux-gnu \
-I/usr/lib/python2.7/config \
-L/usr/lib/python2.7/config -lpython2.7 -lm -ldl \
-o mypython
</pre>
</div>
<p> You will need to supply the same libraries that were used to build
Python the first time. This may include system libraries such as <tt>
-lsocket</tt>, <tt>-lnsl</tt>, and <tt>-lpthread</tt>. Assuming this
actually works, the new version of Python should be identical to the
default version except that your extension module will be a built-in
part of the interpreter.</p>
<p><b> Comment:</b> In practice, you should probably try to avoid static
linking if possible. Some programmers may be inclined to use static
linking in the interest of getting better performance. However, the
performance gained by static linking tends to be rather minimal in most
situations (and quite frankly not worth the extra hassle in the opinion
of this author).</p>
<p><b> Compatibility note:</b> The <tt>embed.i</tt> library file is
deprecated and has not been actively maintained for many years. Even
though it appears to &quot;work&quot; with Python 2.7, no future support is
guaranteed. If using static linking, you might want to rely on a
different approach (perhaps using distutils).</p>
<h3><a name="Python_nn9">33.2.5 Using your module</a></h3>
<p> To use your module, simply use the Python <tt>import</tt> statement.
If all goes well, you will be able to run this:</p>
<div class="targetlang">
<pre>
$ python
&gt;&gt;&gt; import example
&gt;&gt;&gt; example.fact(4)
24
&gt;&gt;&gt;
</pre>
</div>
<p> A common error received by first-time users is the following:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
File &quot;example.py&quot;, line 2, in ?
import _example
ImportError: No module named _example
</pre>
</div>
<p> If you get this message, it means that you either forgot to compile
the wrapper code into an extension module or you didn't give the
extension module the right name. Make sure that you compiled the
wrappers into a module called <tt>_example.so</tt>. And don't forget
the leading underscore (_).</p>
<p> Another possible error is the following:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
ImportError: dynamic module does not define init function (init_example)
&gt;&gt;&gt;
</pre>
</div>
<p> This error is almost always caused when a bad name is given to the
shared object file. For example, if you created a file <tt>example.so</tt>
instead of <tt>_example.so</tt> you would get this error.
Alternatively, this error could arise if the name of the module is
inconsistent with the module name supplied with the <tt>%module</tt>
directive. Double-check the interface to make sure the module name and
the shared object filename match. Another possible cause of this error
is forgetting to link the SWIG-generated wrapper code with the rest of
your application when creating the extension module.</p>
<p> Another common error is something similar to the following:</p>
<div class="targetlang">
<pre>
Traceback (most recent call last):
File &quot;example.py&quot;, line 3, in ?
import example
ImportError: ./_example.so: undefined symbol: fact
</pre>
</div>
<p> This error usually indicates that you forgot to include some object
files or libraries in the linking of the shared library file. Make sure
you compile both the SWIG wrapper file and your original program into a
shared library file. Make sure you pass all of the required libraries
to the linker.</p>
<p> Sometimes unresolved symbols occur because a wrapper has been
created for a function that doesn't actually exist in a library. This
usually occurs when a header file includes a declaration for a function
that was never actually implemented or it was removed from a library
without updating the header file. To fix this, you can either edit the
SWIG input file to remove the offending declaration or you can use the <tt>
%ignore</tt> directive to ignore the declaration.</p>
<p> Finally, suppose that your extension module is linked with another
library like this:</p>
<div class="shell">
<pre>
$ gcc -shared example.o example_wrap.o -L/home/beazley/projects/lib <b>-lfoo</b> \
-o _example.so
</pre>
</div>
<p> If the <tt>foo</tt> library is compiled as a shared library, you
might encounter the following problem when you try to use your module:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
ImportError: libfoo.so: cannot open shared object file: No such file or directory
&gt;&gt;&gt;
</pre>
</div>
<p> This error is generated because the dynamic linker can't locate the <tt>
libfoo.so</tt> library. When shared libraries are loaded, the system
normally only checks a few standard locations such as <tt>/usr/lib</tt>
and <tt>/usr/local/lib</tt>. To fix this problem, there are several
things you can do. First, you can recompile your extension module with
extra path information. For example, on Linux you can do this:</p>
<div class="shell">
<pre>
$ gcc -shared example.o example_wrap.o -L/home/beazley/projects/lib -lfoo \
<b>-Xlinker -rpath /home/beazley/projects/lib </b> \
-o _example.so
</pre>
</div>
<p> Alternatively, you can set the <tt>LD_LIBRARY_PATH</tt> environment
variable to include the directory with your shared libraries. If
setting <tt>LD_LIBRARY_PATH</tt>, be aware that setting this variable
can introduce a noticeable performance impact on all other applications
that you run. To set it only for Python, you might want to do this
instead:</p>
<div class="shell">
<pre>
$ env LD_LIBRARY_PATH=/home/beazley/projects/lib python
</pre>
</div>
<p> Finally, you can use a command such as <tt>ldconfig</tt> (Linux) or <tt>
crle</tt> (Solaris) to add additional search paths to the default system
configuration (this requires root access and you will need to read the
man pages).</p>
<h3><a name="Python_nn10">33.2.6 Compilation of C++ extensions</a></h3>
<p> Compilation of C++ extensions has traditionally been a tricky
problem. Since the Python interpreter is written in C, you need to take
steps to make sure C++ is properly initialized and that modules are
compiled correctly. This should be a non-issue if you're using
distutils, as it takes care of all that for you. The following is
included for historical reasons, and in case you need to compile on
your own.</p>
<p> On most machines, C++ extension modules should be linked using the
C++ compiler. For example:</p>
<div class="shell">
<pre>
$ swig -c++ -python example.i
$ g++ -O2 -fPIC -c example.cxx
$ g++ -O2 -fPIC -c example_wrap.cxx -I/usr/local/include/python2.5
$ g++ -shared example.o example_wrap.o -o _example.so
</pre>
</div>
<p> The -fPIC option tells GCC to generate position-independent code
(PIC) which is required for most architectures (it's not vital on x86,
but still a good idea as it allows code pages from the library to be
shared between processes). Other compilers may need a different option
specified instead of -fPIC.</p>
<p> In addition to this, you may need to include additional library
files to make it work. For example, if you are using the Sun C++
compiler on Solaris, you often need to add an extra library <tt>-lCrun</tt>
like this:</p>
<div class="shell">
<pre>
$ swig -c++ -python example.i
$ CC -c example.cxx
$ CC -c example_wrap.cxx -I/usr/local/include/python2.5
$ CC -G example.o example_wrap.o -L/opt/SUNWspro/lib -o _example.so -lCrun
</pre>
</div>
<p> Of course, the extra libraries to use are completely
non-portable---you will probably need to do some experimentation.</p>
<p> Sometimes people have suggested that it is necessary to relink the
Python interpreter using the C++ compiler to make C++ extension modules
work. In the experience of this author, this has never actually
appeared to be necessary. Relinking the interpreter with C++ really
only includes the special run-time libraries described above---as long
as you link your extension modules with these libraries, it should not
be necessary to rebuild Python.</p>
<p> If you aren't entirely sure about the linking of a C++ extension,
you might look at an existing C++ program. On many Unix machines, the <tt>
ldd</tt> command will list library dependencies. This should give you
some clues about what you might have to include when you link your
extension module. For example:</p>
<div class="shell">
<pre>
$ ldd swig
libstdc++-libc6.1-1.so.2 =&gt; /usr/lib/libstdc++-libc6.1-1.so.2 (0x40019000)
libm.so.6 =&gt; /lib/libm.so.6 (0x4005b000)
libc.so.6 =&gt; /lib/libc.so.6 (0x40077000)
/lib/ld-linux.so.2 =&gt; /lib/ld-linux.so.2 (0x40000000)
</pre>
</div>
<p> As a final complication, a major weakness of C++ is that it does not
define any sort of standard for binary linking of libraries. This means
that C++ code compiled by different compilers will not link together
properly as libraries nor is the memory layout of classes and data
structures implemented in any kind of portable manner. In a monolithic
C++ program, this problem may be unnoticed. However, in Python, it is
possible for different extension modules to be compiled with different
C++ compilers. As long as these modules are self-contained, this
probably won't matter. However, if these modules start sharing data,
you will need to take steps to avoid segmentation faults and other
erratic program behavior. If working with lots of software components,
you might want to investigate using a more formal standard such as COM.</p>
<h3><a name="Python_nn11">33.2.7 Compiling for 64-bit platforms</a></h3>
<p> On platforms that support 64-bit applications (Solaris, Irix, etc.),
special care is required when building extension modules. On these
machines, 64-bit applications are compiled and linked using a different
set of compiler/linker options. In addition, it is not generally
possible to mix 32-bit and 64-bit code together in the same
application.</p>
<p> To utilize 64-bits, the Python executable will need to be recompiled
as a 64-bit application. In addition, all libraries, wrapper code, and
every other part of your application will need to be compiled for
64-bits. If you plan to use other third-party extension modules, they
will also have to be recompiled as 64-bit extensions.</p>
<p> If you are wrapping commercial software for which you have no source
code, you will be forced to use the same linking standard as used by
that software. This may prevent the use of 64-bit extensions. It may
also introduce problems on platforms that support more than one linking
standard (e.g., -o32 and -n32 on Irix).</p>
<p> On the Linux x86_64 platform (Opteron or EM64T), besides of the
required compiler option -fPIC discussed above, you will need to be
careful about the libraries you link with or the library path you use.
In general, a Linux distribution will have two set of libraries, one
for native x86_64 programs (under /usr/lib64), and another for 32 bits
compatibility (under /usr/lib). Also, the compiler options -m32 and
-m64 allow you to choose the desired binary format for your Python
extension.</p>
<h3><a name="Python_nn12">33.2.8 Building Python extensions under
Windows</a></h3>
<p> Building a SWIG extension to Python under Windows is roughly similar
to the process used with Unix. Using the distutils, it is essentially
identical. If you have the same version of the MS compiler that Python
was built with (the python2.4 and python2.5 distributed by python.org
are built with Visual Studio 2003), the standard <tt>python setup.py
build</tt> should just work.</p>
<p> As of python2.5, the distutils support building extensions with
MingGW out of the box. Following the instruction here: <a href="http://boodebr.org/main/python/build-windows-extensions">
Building Python extensions for Windows with only free tools</a> should
get you started.</p>
<p> If you need to build it on your own, the following notes are
provided:</p>
<p> You will need to create a DLL that can be loaded into the
interpreter. This section briefly describes the use of SWIG with
Microsoft Visual C++. As a starting point, many of SWIG's examples
include project files (.dsp files) for Visual C++ 6. These can be
opened by more recent versions of Visual Studio. You might want to take
a quick look at these examples in addition to reading this section.</p>
<p> In Developer Studio, SWIG should be invoked as a custom build
option. This is usually done as follows:</p>
<ul>
<li>Open up a new workspace and use the AppWizard to select a DLL
project.</li>
<li>Add both the SWIG interface file (the .i file), any supporting C
files, and the name of the wrapper file that will be created by SWIG
(ie. <tt>example_wrap.c</tt>). Note : If using C++, choose a different
suffix for the wrapper file such as <tt>example_wrap.cxx</tt>. Don't
worry if the wrapper file doesn't exist yet--Developer Studio keeps a
reference to it.</li>
<li>Select the SWIG interface file and go to the settings menu. Under
settings, select the &quot;Custom Build&quot; option.</li>
<li>Enter &quot;SWIG&quot; in the description field.</li>
<li>Enter &quot;<tt>swig -python -o $(ProjDir)\$(InputName)_wrap.c
$(InputPath)</tt>&quot; in the &quot;Build command(s) field&quot;</li>
<li>Enter &quot;<tt>$(ProjDir)\$(InputName)_wrap.c</tt>&quot; in the &quot;Output
files(s) field&quot;.</li>
<li>Next, select the settings for the entire project and go to
&quot;C++:Preprocessor&quot;. Add the include directories for your Python
installation under &quot;Additional include directories&quot;.</li>
<li>Define the symbol __WIN32__ under preprocessor options.</li>
<li>Finally, select the settings for the entire project and go to &quot;Link
Options&quot;. Add the Python library file to your link libraries. For
example &quot;python27.lib&quot;. Also, set the name of the output file to match
the name of your Python module, i.e. <tt>_example.pyd</tt></li>
<li>Build your project.</li>
</ul>
<p> If all went well, SWIG will be automatically invoked whenever you
build your project. Any changes made to the interface file will result
in SWIG being automatically executed to produce a new version of the
wrapper file.</p>
<p> To run your new Python extension, simply run Python and use the <tt>
import</tt> command as normal. For example :</p>
<div class="targetlang">
<pre>
$ python
&gt;&gt;&gt; import example
&gt;&gt;&gt; print(example.fact(4))
24
&gt;&gt;&gt;
</pre>
</div>
<p> If you get an <tt>ImportError</tt> exception when importing the
module, you may have forgotten to include additional library files when
you built your module. If you get an access violation or some kind of
general protection fault immediately upon import, you have a more
serious problem. This is often caused by linking your extension module
against the wrong set of Win32 debug or thread libraries. You will have
to fiddle around with the build options of project to try and track
this down.</p>
<p> A 'Debug' build of the wrappers requires a debug build of the Python
interpreter. This normally requires building the Python interpreter
from source, which is not a job for the feint-hearted. Alternatively
you can use the 'Release' build of the Python interpreter with a
'Debug' build of your wrappers by defining the <tt>
SWIG_PYTHON_INTERPRETER_NO_DEBUG</tt> symbol under the preprocessor
options. Or you can ensure this macro is defined at the beginning of
the wrapper code using the following in your interface file, where <tt>
_MSC_VER</tt> ensures it is only used by the Visual Studio compiler:</p>
<div class="code">
<pre>
%begin %{
#ifdef _MSC_VER
#define SWIG_PYTHON_INTERPRETER_NO_DEBUG
#endif
%}
</pre>
</div>
<p> Some users have reported success in building extension modules using
Cygwin and other compilers. However, the problem of building usable
DLLs with these compilers tends to be rather problematic. For the
latest information, you may want to consult the <a href="https://github.com/swig/swig/wiki">
SWIG Wiki</a>.</p>
<h3><a name="Python_commandline">33.2.9 Additional Python commandline
options</a></h3>
<p> The following table lists the additional commandline options
available for the Python module. They can also be seen by using:</p>
<div class="code">
<pre>
swig -python -help
</pre>
</div>
<table summary="Python specific options">
<tr><th>Python specific options</th></tr>
<tr><td>-builtin</td><td>Create Python built-in types rather than proxy
classes, for better performance</td></tr>
<tr><td>-castmode</td><td>Enable the casting mode, which allows implicit
cast between types in Python</td></tr>
<tr><td>-debug-doxygen-parser</td><td>Display doxygen parser module
debugging information</td></tr>
<tr><td>-debug-doxygen-translator</td><td>Display doxygen translator
module debugging information</td></tr>
<tr><td>-dirvtable</td><td>Generate a pseudo virtual table for directors
for faster dispatch (warning: currently the vtable is per-object and
never freed so this option causes memory leaks)</td></tr>
<tr><td>-doxygen</td><td>Convert C++ doxygen comments to pydoc comments
in proxy classes</td></tr>
<tr><td>-extranative</td><td>Return extra native wrappers for C++ std
containers wherever possible</td></tr>
<tr><td>-fastproxy</td><td>Use fast proxy mechanism for member methods</td>
</tr>
<tr><td>-flatstaticmethod</td><td>Generate additional flattened Python
methods for C++ static methods</td></tr>
<tr><td>-globals &lt;name&gt;</td><td>Set &lt;name&gt; used to access C global
variable (default: 'cvar')</td></tr>
<tr><td>-interface &lt;mod&gt;</td><td>Set low-level C/C++ module name to
&lt;mod&gt; (default: module name prefixed by '_')</td></tr>
<tr><td>-keyword</td><td>Use keyword arguments</td></tr>
<tr><td>-nofastunpack</td><td>Use traditional UnpackTuple method to
parse the argument functions</td></tr>
<tr><td>-noh</td><td>Don't generate the output header file</td></tr>
<tr><td>-noproxy</td><td>Don't generate proxy classes</td></tr>
<tr><td>-nortti</td><td>Disable the use of the native C++ RTTI with
directors</td></tr>
<tr><td>-nothreads</td><td>Disable thread support for the entire
interface</td></tr>
<tr><td>-olddefs</td><td>Keep the old method definitions when using
-fastproxy</td></tr>
<tr><td>-relativeimport</td><td>Use relative Python imports</td></tr>
<tr><td>-threads</td><td>Add thread support for all the interface</td></tr>
<tr><td>-O</td><td>Enable the following optimization options:
-fastdispatch -fastproxy -fvirtual</td></tr>
</table>
<p> Many of these options are covered later on and their use should
become clearer by the time you have finished reading this section on
SWIG and Python.</p>
<h2><a name="Python_nn13">33.3 A tour of basic C/C++ wrapping</a></h2>
<p> By default, SWIG tries to build a very natural Python interface to
your C/C++ code. Functions are wrapped as functions, classes are
wrapped as classes, and so forth. This section briefly covers the
essential aspects of this wrapping.</p>
<h3><a name="Python_nn14">33.3.1 Modules</a></h3>
<p> The SWIG <tt>%module</tt> directive specifies the name of the Python
module. If you specify `<tt>%module example</tt>', then everything is
wrapped into a Python '<tt>example</tt>' module. Underneath the covers,
this module consists of a Python source file <tt>example.py</tt> and a
low-level extension module <tt>_example.so</tt>. When choosing a module
name, make sure you don't use the same name as a built-in Python
command or standard module name.</p>
<h3><a name="Python_nn15">33.3.2 Functions</a></h3>
<p> Global functions are wrapped as new Python built-in functions. For
example,</p>
<div class="code">
<pre>
%module example
int fact(int n);
</pre>
</div>
<p> creates a built-in function <tt>example.fact(n)</tt> that works
exactly like you think it does:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
&gt;&gt;&gt; print(example.fact(4))
24
&gt;&gt;&gt;
</pre>
</div>
<h3><a name="Python_nn16">33.3.3 Global variables</a></h3>
<p> C/C++ global variables are fully supported by SWIG. However, the
underlying mechanism is somewhat different than you might expect due to
the way that Python assignment works. When you type the following in
Python</p>
<div class="targetlang">
<pre>
a = 3.4
</pre>
</div>
<p> &quot;a&quot; becomes a name for an object containing the value 3.4. If you
later type</p>
<div class="targetlang">
<pre>
b = a
</pre>
</div>
<p> then &quot;a&quot; and &quot;b&quot; are both names for the object containing the value
3.4. Thus, there is only one object containing 3.4 and &quot;a&quot; and &quot;b&quot; are
both names that refer to it. This is quite different than C where a
variable name refers to a memory location in which a value is stored
(and assignment copies data into that location). Because of this, there
is no direct way to map variable assignment in C to variable assignment
in Python.</p>
<p> To provide access to C global variables, SWIG creates a special
object called `<tt>cvar</tt>' that is added to each SWIG generated
module. Global variables are then accessed as attributes of this
object. For example, consider this interface</p>
<div class="code">
<pre>
// SWIG interface file with global variables
%module example
...
%inline %{
extern int My_variable;
extern double density;
%}
...
</pre>
</div>
<p> Now look at the Python interface:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
&gt;&gt;&gt; # Print out value of a C global variable
&gt;&gt;&gt; print(example.cvar.My_variable)
4
&gt;&gt;&gt; # Set the value of a C global variable
&gt;&gt;&gt; example.cvar.density = 0.8442
&gt;&gt;&gt; # Use in a math operation
&gt;&gt;&gt; example.cvar.density = example.cvar.density*1.10
</pre>
</div>
<p> If you make an error in variable assignment, you will receive an
error message. For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; example.cvar.density = &quot;Hello&quot;
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
TypeError: C variable 'density (double )'
&gt;&gt;&gt;
</pre>
</div>
<p> If a variable is declared as <tt>const</tt>, it is wrapped as a
read-only variable. Attempts to modify its value will result in an
error.</p>
<p> To make ordinary variables read-only, you can use the <tt>%immutable</tt>
directive. For example:</p>
<div class="code">
<pre>
%{
extern char *path;
%}
%immutable;
extern char *path;
%mutable;
</pre>
</div>
<p> The <tt>%immutable</tt> directive stays in effect until it is
explicitly disabled or cleared using <tt>%mutable</tt>. See the <a href="#SWIG_readonly_variables">
Creating read-only variables</a> section for further details.</p>
<p> If you just want to make a specific variable immutable, supply a
declaration name. For example:</p>
<div class="code">
<pre>
%{
extern char *path;
%}
%immutable path;
...
extern char *path; // Read-only (due to %immutable)
</pre>
</div>
<p> If you would like to access variables using a name other than &quot;<tt>
cvar</tt>&quot;, it can be changed using the <tt>-globals</tt> option :</p>
<div class="shell">
<pre>
$ swig -python -globals myvar example.i
</pre>
</div>
<p> Some care is in order when importing multiple SWIG modules. If you
use the &quot;<tt>from &lt;file&gt; import *</tt>&quot; style of importing, you will
get a name clash on the variable `<tt>cvar</tt>' and you will only be
able to access global variables from the last module loaded. To prevent
this, you might consider renaming <tt>cvar</tt> or making it private to
the module by giving it a name that starts with a leading underscore.
SWIG does not create <tt>cvar</tt> if there are no global variables in
a module.</p>
<h3><a name="Python_nn17">33.3.4 Constants and enums</a></h3>
<p> C/C++ constants are installed as Python objects containing the
appropriate value. To create a constant, use <tt>#define</tt>, <tt>enum</tt>
, or the <tt>%constant</tt> directive. For example:</p>
<div class="code">
<pre>
#define PI 3.14159
#define VERSION &quot;1.0&quot;
enum Beverage { ALE, LAGER, STOUT, PILSNER };
%constant int FOO = 42;
%constant const char *path = &quot;/usr/local&quot;;
</pre>
</div>
<p> For enums, make sure that the definition of the enumeration actually
appears in a header file or in the wrapper file somehow---if you just
stick an enum in a SWIG interface without also telling the C compiler
about it, the wrapper code won't compile.</p>
<p> Note: declarations declared as <tt>const</tt> are wrapped as
read-only variables and will be accessed using the <tt>cvar</tt> object
described in the previous section. They are not wrapped as constants.
For further discussion about this, see the <a href="#SWIG">SWIG Basics</a>
chapter.</p>
<p> Constants are not guaranteed to remain constant in Python---the name
of the constant could be accidentally reassigned to refer to some other
object. Unfortunately, there is no easy way for SWIG to generate code
that prevents this. You will just have to be careful.</p>
<h3><a name="Python_nn18">33.3.5 Pointers</a></h3>
<p> C/C++ pointers are fully supported by SWIG. Furthermore, SWIG has no
problem working with incomplete type information. Here is a rather
simple interface:</p>
<div class="code">
<pre>
%module example
FILE *fopen(const char *filename, const char *mode);
int fputs(const char *, FILE *);
int fclose(FILE *);
</pre>
</div>
<p> When wrapped, you will be able to use the functions in a natural way
from Python. For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
&gt;&gt;&gt; f = example.fopen(&quot;junk&quot;, &quot;w&quot;)
&gt;&gt;&gt; example.fputs(&quot;Hello World\n&quot;, f)
&gt;&gt;&gt; example.fclose(f)
</pre>
</div>
<p> If this makes you uneasy, rest assured that there is no deep magic
involved. Underneath the covers, pointers to C/C++ objects are simply
represented as opaque values using an especial Python container object:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; print(f)
&lt;Swig Object of type 'FILE *' at 0xb7d6f470&gt;
</pre>
</div>
<p> This pointer value can be freely passed around to different C
functions that expect to receive an object of type <tt>FILE *</tt>. The
only thing you can't do is dereference the pointer from Python. Of
course, that isn't much of a concern in this example.</p>
<p> In older versions of SWIG (1.3.22 or older), pointers were
represented using a plain string object. If you have an old package
that still requires that representation, or you just feel nostalgic,
you can always retrieve it by casting the pointer object to a string:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; print(str(f))
_c0671108_p_FILE
</pre>
</div>
<p> Also, if you need to pass the raw pointer value to some external
Python library, you can do it by casting the pointer object to an
integer:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; print(int(f))
135833352
</pre>
</div>
<p> However, the inverse operation is not possible, i.e., you can't
build a SWIG pointer object from a raw integer value.</p>
<p> Note also that the '0' or NULL pointer is always represented by <tt>
None</tt>, no matter what type swig is addressing. In the previous
example, you can call:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; example.fclose(None)
</pre>
</div>
<p> and that will be equivalent to the following, but not really useful,
C code:</p>
<div class="code">
<pre>
FILE *f = NULL;
fclose(f);
</pre>
</div>
<p> As much as you might be inclined to modify a pointer value directly
from Python, don't. The hexadecimal encoding is not necessarily the
same as the logical memory address of the underlying object. Instead it
is the raw byte encoding of the pointer value. The encoding will vary
depending on the native byte-ordering of the platform (i.e., big-endian
vs. little-endian). Similarly, don't try to manually cast a pointer to
a new type by simply replacing the type-string. This may not work like
you expect, it is particularly dangerous when casting C++ objects. If
you need to cast a pointer or change its value, consider writing some
helper functions instead. For example:</p>
<div class="code">
<pre>
%inline %{
/* C-style cast */
Bar *FooToBar(Foo *f) {
return (Bar *) f;
}
/* C++-style cast */
Foo *BarToFoo(Bar *b) {
return dynamic_cast&lt;Foo*&gt;(b);
}
Foo *IncrFoo(Foo *f, int i) {
return f+i;
}
%}
</pre>
</div>
<p> Also, if working with C++, you should always try to use the new C++
style casts. For example, in the above code, the C-style cast may
return a bogus result whereas as the C++-style cast will return <tt>
None</tt> if the conversion can't be performed.</p>
<h3><a name="Python_nn19">33.3.6 Structures</a></h3>
<p> If you wrap a C structure, it is wrapped by a Python class. This
provides a very natural interface. For example,</p>
<div class="code">
<pre>
struct Vector {
double x, y, z;
};
</pre>
</div>
<p> is used as follows:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; v = example.Vector()
&gt;&gt;&gt; v.x = 3.5
&gt;&gt;&gt; v.y = 7.2
&gt;&gt;&gt; print(v.x, v.y, v.z)
3.5 7.2 0.0
&gt;&gt;&gt;
</pre>
</div>
<p> Similar access is provided for unions and the data members of C++
classes.</p>
<p> If you print out the value of <tt>v</tt> in the above example, you
will see something like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; print(v)
&lt;C Vector instance at _18e31408_p_Vector&gt;
</pre>
</div>
<p> This object is actually a Python instance that has been wrapped
around a pointer to the low-level C structure. This instance doesn't
actually do anything--it just serves as a proxy. The pointer to the C
object can be found in the <tt>.this</tt> attribute. For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; print(v.this)
_18e31408_p_Vector
&gt;&gt;&gt;
</pre>
</div>
<p> Further details about the Python proxy class are covered a little
later.</p>
<p> <tt>const</tt> members of a structure are read-only. Data members
can also be forced to be read-only using the <tt>%immutable</tt>
directive. For example:</p>
<div class="code">
<pre>
struct Foo {
...
%immutable;
int x; /* Read-only members */
char *name;
%mutable;
...
};
</pre>
</div>
<p> When <tt>char *</tt> members of a structure are wrapped, the
contents are assumed to be dynamically allocated using <tt>malloc</tt>
or <tt>new</tt> (depending on whether or not SWIG is run with the -c++
option). When the structure member is set, the old contents will be
released and a new value created. If this is not the behavior you want,
you will have to use a typemap (described later).</p>
<p> If a structure contains arrays, access to those arrays is managed
through pointers. For example, consider this:</p>
<div class="code">
<pre>
struct Bar {
int x[16];
};
</pre>
</div>
<p> If accessed in Python, you will see behavior like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; b = example.Bar()
&gt;&gt;&gt; print(b.x)
_801861a4_p_int
&gt;&gt;&gt;
</pre>
</div>
<p> This pointer can be passed around to functions that expect to
receive an <tt>int *</tt> (just like C). You can also set the value of
an array member using another pointer. For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; c = example.Bar()
&gt;&gt;&gt; c.x = b.x # Copy contents of b.x to c.x
</pre>
</div>
<p> For array assignment, SWIG copies the entire contents of the array
starting with the data pointed to by <tt>b.x</tt>. In this example, 16
integers would be copied. Like C, SWIG makes no assumptions about
bounds checking---if you pass a bad pointer, you may get a segmentation
fault or access violation.</p>
<p> When a member of a structure is itself a structure, it is handled as
a pointer. For example, suppose you have two structures like this:</p>
<div class="code">
<pre>
struct Foo {
int a;
};
struct Bar {
Foo f;
};
</pre>
</div>
<p> Now, suppose that you access the <tt>f</tt> attribute of <tt>Bar</tt>
like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; b = Bar()
&gt;&gt;&gt; x = b.f
</pre>
</div>
<p> In this case, <tt>x</tt> is a pointer that points to the <tt>Foo</tt>
that is inside <tt>b</tt>. This is the same value as generated by this
C code:</p>
<div class="code">
<pre>
Bar b;
Foo *x = &amp;b-&gt;f; /* Points inside b */
</pre>
</div>
<p> Because the pointer points inside the structure, you can modify the
contents and everything works just like you would expect. For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; b = Bar()
&gt;&gt;&gt; b.f.a = 3 # Modify attribute of structure member
&gt;&gt;&gt; x = b.f
&gt;&gt;&gt; x.a = 3 # Modifies the same structure
</pre>
</div>
<p> Note that there is a limitation with structs within structs that
will cause a problem if the outer struct is not a named variable in
Python. The following will cause a segfault:</p>
<div class="targetlang">
<pre>
Bar().f.a = 3
</pre>
</div>
<p> because the unnamed Python proxy class for <tt>Bar()</tt> has its
reference count decremented by the Python interpreter after <tt>f</tt>
has been obtained from it and before <tt>f</tt> is used to obtain <tt>a</tt>
. This results in the underlying <tt>Bar</tt> instance being deleted,
which of course also deletes <tt>f</tt> inside it. Hence the pointer to
<tt>f</tt> points to deleted memory and use of it results in a segfault
or some sort of other undefined behaviour.</p>
<h3><a name="Python_nn20">33.3.7 C++ classes</a></h3>
<p> C++ classes are wrapped by Python classes as well. For example, if
you have this class,</p>
<div class="code">
<pre>
class List {
public:
List();
~List();
int search(char *item);
void insert(char *item);
void remove(char *item);
char *get(int n);
int length;
};
</pre>
</div>
<p> you can use it in Python like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; l = example.List()
&gt;&gt;&gt; l.insert(&quot;Ale&quot;)
&gt;&gt;&gt; l.insert(&quot;Stout&quot;)
&gt;&gt;&gt; l.insert(&quot;Lager&quot;)
&gt;&gt;&gt; l.get(1)
'Stout'
&gt;&gt;&gt; print(l.length)
3
&gt;&gt;&gt;
</pre>
</div>
<p> Class data members are accessed in the same manner as C structures.</p>
<p> Static class members present a special problem for Python. Prior to
Python-2.2, Python classes had no support for static methods and no
version of Python supports static member variables in a manner that
SWIG can utilize. Therefore, SWIG generates wrappers that try to work
around some of these issues. To illustrate, suppose you have a class
like this:</p>
<div class="code">
<pre>
class Spam {
public:
static void foo();
static int bar;
};
</pre>
</div>
<p> In Python, the static member can be accessed in three different
ways:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; s = example.Spam()
&gt;&gt;&gt; s.foo() # Spam::foo() via an instance
&gt;&gt;&gt; example.Spam.foo() # Spam::foo() using class method
&gt;&gt;&gt; example.Spam_foo() # Spam::foo() &quot;flattened&quot; name
</pre>
</div>
<p> The last technique is only available when using the <tt>
-flatstaticmethod</tt> option. This option is not recommended, it is
only available for backwards compatibility as ancient versions of
Python did not have Python class methods.</p>
<p> Static member variables are currently accessed as global variables.
This means, they are accessed through <tt>cvar</tt> or via an instance
property:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; example.cvar.Spam_bar # Spam::bar
7
&gt;&gt;&gt; s = example.Spam()
&gt;&gt;&gt; s.bar # Spam::bar via an instance property
7
</pre>
</div>
<p> The <tt>-builtin</tt> option uses a metaclass to additionally
provide access as follows:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; example.Spam.bar # Spam::bar using -builtin option only
7
</pre>
</div>
<h3><a name="Python_nn21">33.3.8 C++ inheritance</a></h3>
<p> SWIG is fully aware of issues related to C++ inheritance. Therefore,
if you have classes like this</p>
<div class="code">
<pre>
class Foo {
...
};
class Bar : public Foo {
...
};
</pre>
</div>
<p> those classes are wrapped into a hierarchy of Python classes that
reflect the same inheritance structure. All of the usual Python utility
functions work normally:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; b = Bar()
&gt;&gt;&gt; instance(b, Foo)
1
&gt;&gt;&gt; issubclass(Bar, Foo)
1
&gt;&gt;&gt; issubclass(Foo, Bar)
0
</pre>
</div>
<p> Furthermore, if you have functions like this</p>
<div class="code">
<pre>
void spam(Foo *f);
</pre>
</div>
<p> then the function <tt>spam()</tt> accepts <tt>Foo *</tt> or a
pointer to any class derived from <tt>Foo</tt>.</p>
<p> It is safe to use multiple inheritance with SWIG.</p>
<h3><a name="Python_nn22">33.3.9 Pointers, references, values, and
arrays</a></h3>
<p> In C++, there are many different ways a function might receive and
manipulate objects. For example:</p>
<div class="code">
<pre>
void spam1(Foo *x); // Pass by pointer
void spam2(Foo &amp;x); // Pass by reference
void spam3(const Foo &amp;x);// Pass by const reference
void spam4(Foo x); // Pass by value
void spam5(Foo x[]); // Array of objects
</pre>
</div>
<p> In Python, there is no detailed distinction like this--specifically,
there are only &quot;objects&quot;. There are no pointers, references, arrays,
and so forth. Because of this, SWIG unifies all of these types together
in the wrapper code. For instance, if you actually had the above
functions, it is perfectly legal to do this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; f = Foo() # Create a Foo
&gt;&gt;&gt; spam1(f) # Ok. Pointer
&gt;&gt;&gt; spam2(f) # Ok. Reference
&gt;&gt;&gt; spam3(f) # Ok. Const reference
&gt;&gt;&gt; spam4(f) # Ok. Value.
&gt;&gt;&gt; spam5(f) # Ok. Array (1 element)
</pre>
</div>
<p> Similar behavior occurs for return values. For example, if you had
functions like this,</p>
<div class="code">
<pre>
Foo *spam6();
Foo &amp;spam7();
Foo spam8();
const Foo &amp;spam9();
</pre>
</div>
<p> then all three functions will return a pointer to some <tt>Foo</tt>
object. Since the third function (spam8) returns a value, newly
allocated memory is used to hold the result and a pointer is returned
(Python will release this memory when the return value is garbage
collected). The fourth case (spam9) which returns a const reference, in
most of the cases will be treated as a returning value, and it will
follow the same allocation/deallocation process.</p>
<h3><a name="Python_nn23">33.3.10 C++ overloaded functions</a></h3>
<p> C++ overloaded functions, methods, and constructors are mostly
supported by SWIG. For example, if you have two functions like this:</p>
<div class="code">
<pre>
void foo(int);
void foo(char *c);
</pre>
</div>
<p> You can use them in Python in a straightforward manner:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; foo(3) # foo(int)
&gt;&gt;&gt; foo(&quot;Hello&quot;) # foo(char *c)
</pre>
</div>
<p> Similarly, if you have a class like this,</p>
<div class="code">
<pre>
class Foo {
public:
Foo();
Foo(const Foo &amp;);
...
};
</pre>
</div>
<p> you can write Python code like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; f = Foo() # Create a Foo
&gt;&gt;&gt; g = Foo(f) # Copy f
</pre>
</div>
<p> Overloading support is not quite as flexible as in C++. Sometimes
there are methods that SWIG can't disambiguate. For example:</p>
<div class="code">
<pre>
void spam(int);
void spam(short);
</pre>
</div>
<p> or</p>
<div class="code">
<pre>
void foo(Bar *b);
void foo(Bar &amp;b);
</pre>
</div>
<p> If declarations such as these appear, you will get a warning message
like this:</p>
<div class="shell">
<pre>
example.i:12: Warning 509: Overloaded method spam(short) effectively ignored,
example.i:11: Warning 509: as it is shadowed by spam(int).
</pre>
</div>
<p> To fix this, you either need to ignore or rename one of the methods.
For example:</p>
<div class="code">
<pre>
%rename(spam_short) spam(short);
...
void spam(int);
void spam(short); // Accessed as spam_short
</pre>
</div>
<p> or</p>
<div class="code">
<pre>
%ignore spam(short);
...
void spam(int);
void spam(short); // Ignored
</pre>
</div>
<p> SWIG resolves overloaded functions and methods using a
disambiguation scheme that ranks and sorts declarations according to a
set of type-precedence rules. The order in which declarations appear in
the input does not matter except in situations where ambiguity
arises--in this case, the first declaration takes precedence.</p>
<p> Please refer to the &quot;SWIG and C++&quot; chapter for more information
about overloading.</p>
<h3><a name="Python_nn24">33.3.11 C++ operators</a></h3>
<p> Certain C++ overloaded operators can be handled automatically by
SWIG. For example, consider a class like this:</p>
<div class="code">
<pre>
class Complex {
private:
double rpart, ipart;
public:
Complex(double r = 0, double i = 0) : rpart(r), ipart(i) { }
Complex(const Complex &amp;c) : rpart(c.rpart), ipart(c.ipart) { }
Complex &amp;operator=(const Complex &amp;c);
Complex operator+=(const Complex &amp;c) const;
Complex operator+(const Complex &amp;c) const;
Complex operator-(const Complex &amp;c) const;
Complex operator*(const Complex &amp;c) const;
Complex operator-() const;
double re() const { return rpart; }
double im() const { return ipart; }
};
</pre>
</div>
<p> When wrapped, it works like you expect:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; c = Complex(3, 4)
&gt;&gt;&gt; d = Complex(7, 8)
&gt;&gt;&gt; e = c + d
&gt;&gt;&gt; e.re()
10.0
&gt;&gt;&gt; e.im()
12.0
&gt;&gt;&gt; c += d
&gt;&gt;&gt; c.re()
10.0
&gt;&gt;&gt; c.im()
12.0
</pre>
</div>
<p> One restriction with operator overloading support is that SWIG is
not able to fully handle operators that aren't defined as part of the
class. For example, if you had code like this</p>
<div class="code">
<pre>
class Complex {
...
friend Complex operator+(double, const Complex &amp;c);
...
};
</pre>
</div>
<p> then SWIG ignores it and issues a warning. You can still wrap the
operator, but you may have to encapsulate it in a special function. For
example:</p>
<div class="code">
<pre>
%rename(Complex_add_dc) operator+(double, const Complex &amp;);
</pre>
</div>
<p> There are ways to make this operator appear as part of the class
using the <tt>%extend</tt> directive. Keep reading.</p>
<p> Also, be aware that certain operators don't map cleanly to Python.
For instance, overloaded assignment operators don't map to Python
semantics and will be ignored.</p>
<p> Operator overloading is implemented in the <tt>pyopers.swg</tt>
library file. In particular overloaded operators are marked with the <tt>
python:maybecall</tt> feature, also known as <tt>%pythonmaybecall</tt>.
This feature forces SWIG to generate code that returns an instance of
Python's <tt>NotImplemented</tt> instead of raising the usual <tt>
TypeError</tt> exception when an incorrect type is passed to a SWIG
wrapped method. This follows the guidelines in <a href="https://www.python.org/dev/peps/pep-0207/">
PEP 207 - Rich Comparisons</a> and <a href="https://docs.python.org/3/library/constants.html#NotImplemented">
NotImplemented Python constant</a>.</p>
<h3><a name="Python_nn25">33.3.12 C++ namespaces</a></h3>
<p> SWIG is aware of C++ namespaces, but namespace names do not appear
in the module nor do namespaces result in a module that is broken up
into submodules or packages. For example, if you have a file like this,</p>
<div class="code">
<pre>
%module example
namespace foo {
int fact(int n);
struct Vector {
double x, y, z;
};
};
</pre>
</div>
<p> it works in Python as follows:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
&gt;&gt;&gt; example.fact(3)
6
&gt;&gt;&gt; v = example.Vector()
&gt;&gt;&gt; v.x = 3.4
&gt;&gt;&gt; print(v.y)
0.0
&gt;&gt;&gt;
</pre>
</div>
<p> If your program has more than one namespace, name conflicts (if any)
can be resolved using <tt>%rename</tt> For example:</p>
<div class="code">
<pre>
%rename(Bar_spam) Bar::spam;
namespace Foo {
int spam();
}
namespace Bar {
int spam();
}
</pre>
</div>
<p> If you have more than one namespace and your want to keep their
symbols separate, consider wrapping them as separate SWIG modules. For
example, make the module name the same as the namespace and create
extension modules for each namespace separately. If your program
utilizes thousands of small deeply nested namespaces each with
identical symbol names, well, then you get what you deserve.</p>
<h3><a name="Python_nn26">33.3.13 C++ templates</a></h3>
<p> C++ templates don't present a huge problem for SWIG. However, in
order to create wrappers, you have to tell SWIG to create wrappers for
a particular template instantiation. To do this, you use the <tt>
%template</tt> directive. For example:</p>
<div class="code">
<pre>
%module example
%{
#include &quot;pair.h&quot;
%}
template&lt;class T1, class T2&gt;
struct pair {
typedef T1 first_type;
typedef T2 second_type;
T1 first;
T2 second;
pair();
pair(const T1&amp;, const T2&amp;);
~pair();
};
%template(pairii) pair&lt;int, int&gt;;
</pre>
</div>
<p> In Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
&gt;&gt;&gt; p = example.pairii(3, 4)
&gt;&gt;&gt; p.first
3
&gt;&gt;&gt; p.second
4
</pre>
</div>
<p> Obviously, there is more to template wrapping than shown in this
example. More details can be found in the <a href="#SWIGPlus">SWIG and
C++</a> chapter. Some more complicated examples will appear later.</p>
<h3><a name="Python_nn27">33.3.14 C++ Smart Pointers</a></h3>
<h4><a name="Python_smart_pointers_shared_ptr">33.3.14.1 The shared_ptr
Smart Pointer</a></h4>
<p> The C++11 standard provides <tt>std::shared_ptr</tt> which was
derived from the Boost implementation, <tt>boost::shared_ptr</tt>. Both
of these are available for Python in the SWIG library and usage is
outlined in the <a href="#Library_std_shared_ptr">shared_ptr smart
pointer</a> library section.</p>
<h4><a name="Python_smart_pointers_generic">33.3.14.2 Generic Smart
Pointers</a></h4>
<p> In certain C++ programs, it is common to use classes that have been
wrapped by so-called &quot;smart pointers.&quot; Generally, this involves the use
of a template class that implements <tt>operator-&gt;()</tt> like this:</p>
<div class="code">
<pre>
template&lt;class T&gt; class SmartPtr {
...
T *operator-&gt;();
...
}
</pre>
</div>
<p> Then, if you have a class like this,</p>
<div class="code">
<pre>
class Foo {
public:
int x;
int bar();
};
</pre>
</div>
<p> A smart pointer would be used in C++ as follows:</p>
<div class="code">
<pre>
SmartPtr&lt;Foo&gt; p = CreateFoo(); // Created somehow (not shown)
...
p-&gt;x = 3; // Foo::x
int y = p-&gt;bar(); // Foo::bar
</pre>
</div>
<p> To wrap this in Python, simply tell SWIG about the <tt>SmartPtr</tt>
class and the low-level <tt>Foo</tt> object. Make sure you instantiate <tt>
SmartPtr</tt> using <tt>%template</tt> if necessary. For example:</p>
<div class="code">
<pre>
%module example
...
%template(SmartPtrFoo) SmartPtr&lt;Foo&gt;;
...
</pre>
</div>
<p> Now, in Python, everything should just &quot;work&quot;:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; p = example.CreateFoo() # Create a smart-pointer somehow
&gt;&gt;&gt; p.x = 3 # Foo::x
&gt;&gt;&gt; p.bar() # Foo::bar
</pre>
</div>
<p> If you ever need to access the underlying pointer returned by <tt>
operator-&gt;()</tt> itself, simply use the <tt>__deref__()</tt> method.
For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; f = p.__deref__() # Returns underlying Foo *
</pre>
</div>
<h3><a name="Python_nn27a">33.3.15 C++ reference counted objects</a></h3>
<p> The <a href="#SWIGPlus_ref_unref">C++ reference counted objects</a>
section contains Python examples of memory management using referencing
counting.</p>
<h2><a name="Python_nn28">33.4 Further details on the Python class
interface</a></h2>
<p> In the previous section, a high-level view of Python wrapping was
presented. A key component of this wrapping is that structures and
classes are wrapped by Python proxy classes. This provides a very
natural Python interface and allows SWIG to support a number of
advanced features such as operator overloading. However, a number of
low-level details were omitted. This section provides a brief overview
of how the proxy classes work.</p>
<p><b>New in SWIG version 2.0.4:</b> The use of Python proxy classes has
performance implications that may be unacceptable for a
high-performance library. The new <tt>-builtin</tt> option instructs
SWIG to forego the use of proxy classes, and instead create wrapped
types as new built-in Python types. When this option is used, the
following section (&quot;Proxy classes&quot;) does not apply. Details on the use
of the <tt>-builtin</tt> option are in the <a href="#Python_builtin_types">
Built-in Types</a> section.</p>
<h3><a name="Python_nn29">33.4.1 Proxy classes</a></h3>
<p> In the <a href="#SWIG">&quot;SWIG basics&quot;</a> and <a href="#SWIGPlus">
&quot;SWIG and C++&quot;</a> chapters, details of low-level structure and class
wrapping are described. To summarize those chapters, if you have a
class like this</p>
<div class="code">
<pre>
class Foo {
public:
int x;
int spam(int);
...
</pre>
</div>
<p> then SWIG transforms it into a set of low-level procedural wrappers.
For example:</p>
<div class="code">
<pre>
Foo *new_Foo() {
return new Foo();
}
void delete_Foo(Foo *f) {
delete f;
}
int Foo_x_get(Foo *f) {
return f-&gt;x;
}
void Foo_x_set(Foo *f, int value) {
f-&gt;x = value;
}
int Foo_spam(Foo *f, int arg1) {
return f-&gt;spam(arg1);
}
</pre>
</div>
<p> These wrappers can be found in the low-level extension module (e.g.,
<tt>_example</tt>).</p>
<p> Using these wrappers, SWIG generates a high-level Python proxy class
(also known as a shadow class) like this (shown for Python 2.2):</p>
<div class="targetlang">
<pre>
import _example
class Foo(object):
def __init__(self):
self.this = _example.new_Foo()
self.thisown = True
def __del__(self):
if self.thisown:
_example.delete_Foo(self.this)
def spam(self, arg1):
return _example.Foo_spam(self.this, arg1)
x = property(_example.Foo_x_get, _example.Foo_x_set)
</pre>
</div>
<p> This class merely holds a pointer to the underlying C++ object (<tt>
.this</tt>) and dispatches methods and member variable access to that
object using the low-level accessor functions. From a user's point of
view, it makes the class work normally:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; f = example.Foo()
&gt;&gt;&gt; f.x = 3
&gt;&gt;&gt; y = f.spam(5)
</pre>
</div>
<p> The fact that the class has been wrapped by a real Python class
offers certain advantages. For instance, you can attach new Python
methods to the class and you can even inherit from it (something not
supported by Python built-in types until Python 2.2).</p>
<h3><a name="Python_builtin_types">33.4.2 Built-in Types</a></h3>
<p> The <tt>-builtin</tt> option provides a significant performance
improvement in the wrapped code. To understand the difference between
proxy classes and built-in types, let's take a look at what a wrapped
object looks like under both circumstances.</p>
<p>When proxy classes are used, each wrapped object in Python is an
instance of a pure Python class. As a reminder, here is what the <tt>
__init__</tt> method looks like in a proxy class:</p>
<div class="targetlang">
<pre>
class Foo(object):
def __init__(self):
self.this = _example.new_Foo()
self.thisown = True
</pre>
</div>
<p>When a <tt>Foo</tt> instance is created, the call to <tt>
_example.new_Foo()</tt> creates a new C++ <tt>Foo</tt> instance; wraps
that C++ instance inside an instance of a Python built-in type called <tt>
SwigPyObject</tt>; and stores the <tt>SwigPyObject</tt> instance in the
'this' field of the Python Foo object. Did you get all that? So, the
Python <tt>Foo</tt> object is composed of three parts:</p>
<ul>
<li> The Python <tt>Foo</tt> instance, which contains...</li>
<li> ... an instance of <tt>struct SwigPyObject</tt>, which contains...</li>
<li> ... a C++ <tt>Foo</tt> instance</li>
</ul>
<p>When <tt>-builtin</tt> is used, the pure Python layer is stripped
off. Each wrapped class is turned into a new Python built-in type which
inherits from <tt>SwigPyObject</tt>, and <tt>SwigPyObject</tt>
instances are returned directly from the wrapped methods. For more
information about Python built-in extensions, please refer to the
Python documentation:</p>
<p><a href="https://docs.python.org/3/extending/newtypes.html">
https://docs.python.org/3/extending/newtypes.html</a></p>
<h4><a name="Python_builtin_limitations">33.4.2.1 Limitations</a></h4>
<p>Use of the <tt>-builtin</tt> option implies a couple of limitations:</p>
<ul>
<li>
<p>Some legacy syntax is no longer supported; in particular:</p>
<ul>
<li>The functional interface is no longer exposed. For example, you may
no longer call <tt>Whizzo.new_CrunchyFrog()</tt>. Instead, you must use
<tt>Whizzo.CrunchyFrog()</tt>.</li>
<li>Static member variables are no longer accessed through the 'cvar'
field (e.g., <tt>Dances.cvar.FishSlap</tt>). They are instead accessed
in the idiomatic way (<tt>Dances.FishSlap</tt>).</li>
</ul>
</li>
<li>
<p>Wrapped types may not be raised as Python exceptions. Here's why: the
Python internals expect that all sub-classes of Exception will have
this struct layout:</p>
<div class="code">
<pre>
typedef struct {
PyObject_HEAD
PyObject *dict;
PyObject *args;
PyObject *message;
} PyBaseExceptionObject;
</pre>
</div>
<p>But swig-generated wrappers expect that all swig-wrapped classes will
have this struct layout:</p>
<div class="code">
<pre>
typedef struct {
PyObject_HEAD
void *ptr;
swig_type_info *ty;
int own;
PyObject *next;
PyObject *dict;
} SwigPyObject;
</pre>
</div>
<p>There are workarounds for this. For example, if you wrap this class:<div
class="code">
<pre>
class MyException {
public:
MyException (const char *msg_);
~MyException ();
const char *what () const;
private:
char *msg;
};
</pre>
</div></p>
<p>... you can define this Python class, which may be raised as an
exception:</p>
<div class="targetlang">
<pre>
class MyPyException(Exception):
def __init__(self, msg, *args):
Exception.__init__(self, *args)
self.myexc = MyException(msg)
def what(self):
return self.myexc.what()
</pre>
</div></li>
<li>
<p>Reverse binary operators (e.g., <tt>__radd__</tt>) are not supported.</p>
<p>To illustrate this point, if you have a wrapped class called <tt>
MyString</tt>, and you want to use instances of <tt>MyString</tt>
interchangeably with native Python strings, you can define an <tt>
'operator+ (const char*)'</tt> method :</p>
<div class="code">
<pre>
class MyString {
public:
MyString (const char *init);
MyString operator+ (const char *other) const;
...
};
</pre>
</div>
<p> SWIG will automatically create an operator overload in Python that
will allow this:</p>
<div class="targetlang">
<pre>
from MyModule import MyString
mystr = MyString(&quot;No one expects&quot;)
episode = mystr + &quot; the Spanish Inquisition&quot;
</pre>
</div>
<p> This works because the first operand (<tt>mystr</tt>) defines a way
to add a native string to itself. However, the following will<b> not</b>
work:</p>
<div class="targetlang">
<pre>
from MyModule import MyString
mystr = MyString(&quot;Parrot&quot;)
episode = &quot;Dead &quot; + mystr
</pre>
</div>
<p> The above code fails, because the first operand -- a native Python
string -- doesn't know how to add an instance of <tt>MyString</tt> to
itself.</p>
</li>
<li>
<p>If you have multiple SWIG modules that share type information (<a href="#Modules_nn2">
more info</a>), the <tt>-builtin</tt> option requires a bit of extra
discipline to ensure that base classes are initialized before derived
classes. Specifically:</p>
<ul>
<li>
<p>There must be an unambiguous dependency graph for the modules.</p>
</li>
<li>
<p>Module dependencies must be explicitly stated with <tt>%import</tt>
statements in the SWIG interface file.</p>
</li>
</ul>
<p>As an example, suppose module <tt>A</tt> has this interface in <tt>
A.i</tt> :</p>
<div class="code">
<pre>
%module &quot;A&quot;;
class Base {
...
};
</pre>
</div>
<p>If you want to wrap another module containing a class that inherits
from <tt>A</tt>, this is how it would look :</p>
<div class="code">
<pre>
%module &quot;B&quot;;
%import &quot;A.i&quot;
class Derived : public Base {
...
};
</pre>
</div>
<p>The <tt>import &quot;A.i&quot;</tt> statement is required, because module <tt>B</tt>
depends on module <tt>A</tt>.</p>
<p>As long as you obey these requirements, your Python code may import
the modules in any order :</p>
<div class="targetlang">
<pre>
import B
import A
assert(issubclass(B.Derived, A.Base))
</pre>
</div></li>
<li>
<p><a href="#Python_annotations">Python annotations</a> are not
supported.</p>
</li>
</ul>
<h4><a name="Python_builtin_overloads">33.4.2.2 Operator overloads and
slots -- use them!</a></h4>
<p>The entire justification for the <tt>-builtin</tt> option is improved
performance. To that end, the best way to squeeze maximum performance
out of your wrappers is to<b> use operator overloads.</b> Named method
dispatch is slow in Python, even when compared to other scripting
languages. However, Python built-in types have a large number of
&quot;slots&quot;, analogous to C++ operator overloads, which allow you to
short-circuit named method dispatch for certain common operations.</p>
<p>By default, SWIG will translate most C++ arithmetic operator
overloads into Python slot entries. For example, suppose you have this
class:<div class="code">
<pre>
class Twit {
public:
Twit operator+ (const Twit&amp; twit) const;
// Forward to operator+
Twit add (const Twit&amp; twit) const {
return *this + twit;
}
};
</pre>
</div></p>
<p>SWIG will automatically register <tt>operator+</tt> as a Python slot
operator for addition. You may write Python code like this:</p>
<div class="targetlang">
<pre>
from MyModule import Twit
nigel = Twit()
emily = Twit()
percival = nigel + emily
percival = nigel.add(emily)
</pre>
</div>
<p>The last two lines of the Python code are equivalent, but<b> the line
that uses the '+' operator is much faster</b>.</p>
<p>In-place operators (e.g., <tt>operator+=</tt>) and comparison
operators (<tt>operator==, operator&lt;</tt>, etc.) are also converted to
Python slot operators. For a complete list of C++ operators that are
automatically converted to Python slot operators, refer to the file <tt>
python/pyopers.swg</tt> in the SWIG library.</p>
<p> Read about all of the available Python slots here: <a href="https://docs.python.org/3/c-api/typeobj.html">
https://docs.python.org/3/c-api/typeobj.html</a></p>
<p> There are two ways to define a Python slot function: dispatch to a
statically defined function; or dispatch to a method defined on the
operand.</p>
<p> To dispatch to a statically defined function, use
%feature(&quot;python:&lt;slot&gt;&quot;), where &lt;slot&gt; is the name of a field in a <tt>
PyTypeObject, PyNumberMethods, PyMappingMethods, PySequenceMethods</tt>
or <tt>PyBufferProcs</tt>. You may override (almost) all of these
slots.</p>
<p> Let's consider an example setting the <tt>tp_hash</tt> slot for the <tt>
MyClass</tt> type. This is akin to providing a <tt>__hash__</tt> method
(for non-builtin types) to make a type hashable. The hashable type can
then for example be added to a Python <tt>dict</tt>.</p>
<div class="code">
<pre>
%feature(&quot;python:tp_hash&quot;) MyClass &quot;myHashFunc&quot;;
class MyClass {
public:
long field1;
long field2;
...
};
%{
#if PY_VERSION_HEX &gt;= 0x03020000
static Py_hash_t myHashFunc(PyObject *pyobj)
#else
static long myHashFunc(PyObject *pyobj)
#endif
{
MyClass *cobj;
// Convert pyobj to cobj
return (cobj-&gt;field1 * (cobj-&gt;field2 &lt;&lt; 7));
}
%}
</pre>
</div>
<p> If you examine the generated code, the supplied hash function will
now be the function callback in the tp_hash slot for the builtin type
for <tt>MyClass</tt>:</p>
<div class="code">
<pre>
static PyHeapTypeObject SwigPyBuiltin__MyClass_type = {
...
(hashfunc) myHashFunc, /* tp_hash */
...
</pre>
</div>
<p> NOTE: It is the responsibility of the programmer (that's you!) to
ensure that a statically defined slot function has the correct
signature, the <tt>hashfunc</tt> typedef in this case.</p>
<p> If, instead, you want to dispatch to an instance method, you can use
%feature(&quot;python:slot&quot;). For example:</p>
<div class="code">
<pre>
%feature(&quot;python:slot&quot;, &quot;tp_hash&quot;, functype=&quot;hashfunc&quot;) MyClass::myHashFunc;
#if PY_VERSION_HEX &lt; 0x03020000
#define Py_hash_t long
#endif
class MyClass {
public:
Py_hash_t myHashFunc() const;
...
};
</pre>
</div>
<p> NOTE: Some Python slots use a method signature which does not match
the signature of SWIG-wrapped methods. For those slots, SWIG will
automatically generate a &quot;closure&quot; function to re-marshal the arguments
before dispatching to the wrapped method. Setting the &quot;functype&quot;
attribute of the feature enables SWIG to generate the chosen closure
function.</p>
<p> There is further information on <tt>%feature(&quot;python:slot&quot;)</tt> in
the file <tt>python/pyopers.swg</tt> in the SWIG library.</p>
<h3><a name="Python_nn30">33.4.3 Memory management</a></h3>
<p>NOTE: Although this section refers to proxy objects, everything here
also applies when the <tt>-builtin</tt> option is used.</p>
<p> Associated with proxy object, is an ownership flag <tt>thisown</tt>.
The value of this flag determines who is responsible for deleting the
underlying C++ object. If set to <tt>True</tt>, the Python interpreter
will destroy the C++ object when the proxy class is garbage collected.
If set to <tt>False</tt> (or if the attribute is missing), then the
destruction of the proxy class has no effect on the C++ object.</p>
<p> When an object is created by a constructor or returned by value,
Python automatically takes ownership of the result. For example:</p>
<div class="code">
<pre>
class Foo {
public:
Foo();
~Foo();
};
Foo CreateFoo();
</pre>
</div>
<p> In Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; f = Foo()
&gt;&gt;&gt; f.thisown
True
&gt;&gt;&gt; g = CreateFoo()
&gt;&gt;&gt; g.thisown
True
</pre>
</div>
<p> The type information is a Python class, a proxy of the underlying
C++ type:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; type(f)
&lt;class 'example.Foo'&gt;
&gt;&gt;&gt; f
&lt;example.Foo; proxy of &lt;Swig Object of type 'Foo *' at 0x7f5a86d48600&gt; &gt;
True
</pre>
</div>
<p> Unlike a C or C++ compiler, SWIG can work with incomplete type
information and generate wrapper code even if the definition of <tt>
class Foo</tt> is not provided to SWIG. However, this has negative
consequences in the form of a memory leak.</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; h = CreateFoo()
&gt;&gt;&gt; type(h)
&lt;class 'SwigPyObject'&gt;
&gt;&gt;&gt; h
&lt;Swig Object of type 'Foo *' at 0x7fc341b4d6f0&gt;
&gt;&gt;&gt; quit()
swig/python detected a memory leak of type 'Foo *', no destructor found.
</pre>
</div>
<p> Note that the type is no longer a proxy Python class, but is instead
a generic <tt>SwigPyObject</tt> type. The memory leak warning is shown
as SWIG has no way of knowing how to call the destructor with this
generic type. The recommendation is to fix the leak by providing the
missing type information. However, for those who don't care about
memory leaks, the warning can also be suppressed by by defining <tt>
SWIG_PYTHON_SILENT_MEMLEAK</tt> when compiling the generated C/C++
wrapper code, such as passing <tt>-DSWIG_PYTHON_SILENT_MEMLEAK</tt> to
the compiler or getting SWIG to define this macro at the beginning of
the generated wrapper code by adding the following to the interface
file using:</p>
<div class="code">
<pre>
%begin %{
#define SWIG_PYTHON_SILENT_MEMLEAK
%}
</pre>
</div>
<p> The generic <tt>SwigPyObject</tt> type also has a <tt>disown()</tt>
method which can be called to deactivate the warning.</p>
<p> Next, let's consider returning raw pointers even when the full type
information is available to SWIG. When pointers or references are
returned to Python, there is often no way to know where they came from
and how to manage the memory that they point to. Therefore, the
ownership is set to <tt>False</tt>. For example:</p>
<div class="code">
<pre>
class Foo {
public:
Foo();
~Foo();
};
Foo *PointerCreate();
Foo &amp;ReferenceCreate();
</pre>
</div>
<br><div class="targetlang">
<pre>
&gt;&gt;&gt; p = PointerCreate()
&gt;&gt;&gt; p.thisown
False
&gt;&gt;&gt; r = ReferenceCreate()
&gt;&gt;&gt; r.thisown
False
</pre>
</div>
<p> This behavior is especially important for classes that act as
containers. For example, if a method returns a pointer to an object
that is contained inside another object, you definitely don't want
Python to assume ownership and destroy it!</p>
<p> A good way to indicate that ownership should be set for a returned
pointer is to use the <a href="#Library_nn11">%newobject directive</a>.</p>
<p> Related to containers, ownership issues can arise whenever an object
is assigned to a member or global variable. For example, consider this
interface:</p>
<div class="code">
<pre>
%module example
struct Foo {
int value;
Foo *next;
};
Foo *head = 0;
</pre>
</div>
<p> When wrapped in Python, careful observation will reveal that
ownership changes whenever an object is assigned to a global variable.
For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; f = example.Foo()
&gt;&gt;&gt; f.thisown
True
&gt;&gt;&gt; example.cvar.head = f
&gt;&gt;&gt; f.thisown
False
&gt;&gt;&gt;
</pre>
</div>
<p> In this case, C is now holding a reference to the object---you
probably don't want Python to destroy it. Similarly, this occurs for
members. For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; f = example.Foo()
&gt;&gt;&gt; g = example.Foo()
&gt;&gt;&gt; f.thisown
True
&gt;&gt;&gt; g.thisown
True
&gt;&gt;&gt; f.next = g
&gt;&gt;&gt; g.thisown
False
&gt;&gt;&gt;
</pre>
</div>
<p> For the most part, memory management issues remain hidden. However,
there are occasionally situations where you might have to manually
change the ownership of an object. For instance, consider code like
this:</p>
<div class="code">
<pre>
class Node {
Object *value;
public:
void set_value(Object *v) { value = v; }
...
};
</pre>
</div>
<p> Now, consider the following Python code:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; v = Object() # Create an object
&gt;&gt;&gt; n = Node() # Create a node
&gt;&gt;&gt; n.set_value(v) # Set value
&gt;&gt;&gt; v.thisown
True
&gt;&gt;&gt; del v
</pre>
</div>
<p> In this case, the object <tt>n</tt> is holding a reference to <tt>v</tt>
internally. However, SWIG has no way to know that this has occurred.
Therefore, Python still thinks that it has ownership of the object.
Should the proxy object be destroyed, then the C++ destructor will be
invoked and <tt>n</tt> will be holding a stale-pointer. If you're
lucky, you will only get a segmentation fault.</p>
<p> To work around this, it is always possible to flip the ownership
flag. For example,</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; v.thisown = False
</pre>
</div>
<p> It is also possible to deal with situations like this using
typemaps--an advanced topic discussed later.</p>
<h2><a name="Python_directors">33.5 Cross language polymorphism</a></h2>
<p> Proxy classes provide a more natural, object-oriented way to access
extension classes. As described above, each proxy instance has an
associated C++ instance, and method calls to the proxy are passed to
the C++ instance transparently via C wrapper functions.</p>
<p> This arrangement is asymmetric in the sense that no corresponding
mechanism exists to pass method calls down the inheritance chain from
C++ to Python. In particular, if a C++ class has been extended in
Python (by extending the proxy class), these extensions will not be
visible from C++ code. Virtual method calls from C++ are thus not able
access the lowest implementation in the inheritance chain.</p>
<p> Changes have been made to SWIG 1.3.18 to address this problem and
make the relationship between C++ classes and proxy classes more
symmetric. To achieve this goal, new classes called directors are
introduced at the bottom of the C++ inheritance chain. The job of the
directors is to route method calls correctly, either to C++
implementations higher in the inheritance chain or to Python
implementations lower in the inheritance chain. The upshot is that C++
classes can be extended in Python and from C++ these extensions look
exactly like native C++ classes. Neither C++ code nor Python code needs
to know where a particular method is implemented: the combination of
proxy classes, director classes, and C wrapper functions takes care of
all the cross-language method routing transparently.</p>
<h3><a name="Python_nn33">33.5.1 Enabling directors</a></h3>
<p> The director feature is disabled by default. To use directors you
must make two changes to the interface file. First, add the &quot;directors&quot;
option to the %module directive, like this:</p>
<div class="code">
<pre>
%module(directors=&quot;1&quot;) modulename
</pre>
</div>
<p> Without this option no director code will be generated. Second, you
must use the %feature(&quot;director&quot;) directive to tell SWIG which classes
and methods should get directors. The %feature directive can be applied
globally, to specific classes, and to specific methods, like this:</p>
<div class="code">
<pre>
// generate directors for all classes that have virtual methods
%feature(&quot;director&quot;);
// generate directors for the virtual methods in class Foo
%feature(&quot;director&quot;) Foo;
</pre>
</div>
<p> You can use the %feature(&quot;nodirector&quot;) directive to turn off
directors for specific classes or methods. So for example,</p>
<div class="code">
<pre>
%feature(&quot;director&quot;) Foo;
%feature(&quot;nodirector&quot;) Foo::bar;
</pre>
</div>
<p> will generate directors for the virtual methods of class Foo except
bar().</p>
<p> Directors can also be generated implicitly through inheritance. In
the following, class Bar will get a director class that handles the
methods one() and two() (but not three()):</p>
<div class="code">
<pre>
%feature(&quot;director&quot;) Foo;
class Foo {
public:
Foo(int foo);
virtual ~Foo();
virtual void one();
virtual void two();
};
class Bar: public Foo {
public:
virtual void three();
};
</pre>
</div>
<p> then at the Python side you can define</p>
<div class="targetlang">
<pre>
import mymodule
class MyFoo(mymodule.Foo):
def __init__(self, foo):
mymodule.Foo.__init__(self, foo)
# super().__init__(foo) # Alternative construction for Python3
def one(self):
print(&quot;one from Python&quot;)
</pre>
</div>
<h3><a name="Python_nn34">33.5.2 Director classes</a></h3>
<p> For each class that has directors enabled, SWIG generates a new
class that derives from both the class in question and a special <tt>
Swig::Director</tt> class. These new classes, referred to as director
classes, can be loosely thought of as the C++ equivalent of the Python
proxy classes. The director classes store a pointer to their underlying
Python object and handle various issues related to object ownership.
Indeed, this is quite similar to the &quot;this&quot; and &quot;thisown&quot; members of
the Python proxy classes.</p>
<p> For simplicity let's ignore the <tt>Swig::Director</tt> class and
refer to the original C++ class as the director's base class. By
default, a director class extends all virtual methods in the
inheritance chain of its base class (see the preceding section for how
to modify this behavior). Virtual methods that have a final specifier
are unsurprisingly excluded. Thus the virtual method calls, whether
they originate in C++ or in Python via proxy classes, eventually end up
in at the implementation in the director class. The job of the director
methods is to route these method calls to the appropriate place in the
inheritance chain. By &quot;appropriate place&quot; we mean the method that would
have been called if the C++ base class and its extensions in Python
were seamlessly integrated. That seamless integration is exactly what
the director classes provide, transparently skipping over all the messy
extension API glue that binds the two languages together.</p>
<p> In reality, the &quot;appropriate place&quot; is one of only two
possibilities: C++ or Python. Once this decision is made, the rest is
fairly easy. If the correct implementation is in C++, then the lowest
implementation of the method in the C++ inheritance chain is called
explicitly. If the correct implementation is in Python, the Python API
is used to call the method of the underlying Python object (after which
the usual virtual method resolution in Python automatically finds the
right implementation).</p>
<p> Now how does the director decide which language should handle the
method call? The basic rule is to handle the method in Python, unless
there's a good reason not to. The reason for this is simple: Python has
the most &quot;extended&quot; implementation of the method. This assertion is
guaranteed, since at a minimum the Python proxy class implements the
method. If the method in question has been extended by a class derived
from the proxy class, that extended implementation will execute exactly
as it should. If not, the proxy class will route the method call into a
C wrapper function, expecting that the method will be resolved in C++.
The wrapper will call the virtual method of the C++ instance, and since
the director extends this the call will end up right back in the
director method. Now comes the &quot;good reason not to&quot; part. If the
director method were to blindly call the Python method again, it would
get stuck in an infinite loop. We avoid this situation by adding
special code to the C wrapper function that tells the director method
to not do this. The C wrapper function compares the pointer to the
Python object that called the wrapper function to the pointer stored by
the director. If these are the same, then the C wrapper function tells
the director to resolve the method by calling up the C++ inheritance
chain, preventing an infinite loop.</p>
<p> One more point needs to be made about the relationship between
director classes and proxy classes. When a proxy class instance is
created in Python, SWIG creates an instance of the original C++ class
and assigns it to <tt>.this</tt>. This is exactly what happens without
directors and is true even if directors are enabled for the particular
class in question. When a class<i> derived</i> from a proxy class is
created, however, SWIG then creates an instance of the corresponding
C++ director class. The reason for this difference is that user-defined
subclasses may override or extend methods of the original class, so the
director class is needed to route calls to these methods correctly. For
unmodified proxy classes, all methods are ultimately implemented in C++
so there is no need for the extra overhead involved with routing the
calls through Python.</p>
<h3><a name="Python_nn35">33.5.3 Ownership and object destruction</a></h3>
<p> Memory management issues are slightly more complicated with
directors than for proxy classes alone. Python instances hold a pointer
to the associated C++ director object, and the director in turn holds a
pointer back to the Python object. By default, proxy classes own their
C++ director object and take care of deleting it when they are garbage
collected.</p>
<p> This relationship can be reversed by calling the special <tt>
__disown__()</tt> method of the proxy class. After calling this method,
the <tt>thisown</tt> flag is set to <tt>False</tt>, and the director
class increments the reference count of the Python object. When the
director class is deleted it decrements the reference count. Assuming
no outstanding references to the Python object remain, the Python
object will be destroyed at the same time. This is a good thing, since
directors and proxies refer to each other and so must be created and
destroyed together. Destroying one without destroying the other will
likely cause your program to segfault.</p>
<p> To help ensure that no references to the Python object remain after
calling <tt>__disown__()</tt>, this method returns a weak reference to
the Python object. Here is an example:</p>
<div class="code">
<pre>
class Foo {
public:
...
};
class FooContainer {
public:
void addFoo(Foo *);
...
};
</pre>
</div>
<br><div class="targetlang">
<pre>
&gt;&gt;&gt; c = FooContainer()
&gt;&gt;&gt; a = Foo().__disown__()
&gt;&gt;&gt; c.addFoo(a)
&gt;&gt;&gt; b = Foo()
&gt;&gt;&gt; b = b.__disown__()
&gt;&gt;&gt; c.addFoo(b)
&gt;&gt;&gt; c.addFoo(Foo().__disown__())
</pre>
</div>
<p> In this example, we are assuming that FooContainer will take care of
deleting all the Foo pointers it contains at some point. Note that no
hard references to the Foo objects remain in Python.</p>
<h3><a name="Python_nn36">33.5.4 Exception unrolling</a></h3>
<p> With directors routing method calls to Python, and proxies routing
them to C++, the handling of exceptions is an important concern. By
default, the directors ignore exceptions that occur during method calls
that are resolved in Python. To handle such exceptions correctly, it is
necessary to temporarily translate them into C++ exceptions. This can
be done with the %feature(&quot;director:except&quot;) directive. The following
code should suffice in most cases:</p>
<div class="code">
<pre>
%feature(&quot;director:except&quot;) {
if ($error != NULL) {
throw Swig::DirectorMethodException();
}
}
</pre>
</div>
<p> This code will check the Python error state after each method call
from a director into Python, and throw a C++ exception if an error
occurred. This exception can be caught in C++ to implement an error
handler. Currently no information about the Python error is stored in
the Swig::DirectorMethodException object, but this will likely change
in the future.</p>
<p> It may be the case that a method call originates in Python, travels
up to C++ through a proxy class, and then back into Python via a
director method. If an exception occurs in Python at this point, it
would be nice for that exception to find its way back to the original
caller. This can be done by combining a normal %exception directive
with the <tt>director:except</tt> handler shown above. Here is an
example of a suitable exception handler:</p>
<div class="code">
<pre>
%exception {
try { $action }
catch (Swig::DirectorException &amp;e) { SWIG_fail; }
}
</pre>
</div>
<p> The class Swig::DirectorException used in this example is actually a
base class of Swig::DirectorMethodException, so it will trap this
exception. Because the Python error state is still set when
Swig::DirectorMethodException is thrown, Python will register the
exception as soon as the C wrapper function returns.</p>
<h3><a name="Python_nn37">33.5.5 Overhead and code bloat</a></h3>
<p> Enabling directors for a class will generate a new director method
for every virtual method in the class' inheritance chain. This alone
can generate a lot of code bloat for large hierarchies. Method
arguments that require complex conversions to and from target language
types can result in large director methods. For this reason it is
recommended that you selectively enable directors only for specific
classes that are likely to be extended in Python and used in C++.</p>
<p> Compared to classes that do not use directors, the call routing in
the director methods does add some overhead. In particular, at least
one dynamic cast and one extra function call occurs per method call
from Python. Relative to the speed of Python execution this is probably
completely negligible. For worst case routing, a method call that
ultimately resolves in C++ may take one extra detour through Python in
order to ensure that the method does not have an extended Python
implementation. This could result in a noticeable overhead in some
cases.</p>
<p> Although directors make it natural to mix native C++ objects with
Python objects (as director objects) via a common base class pointer,
one should be aware of the obvious fact that method calls to Python
objects will be much slower than calls to C++ objects. This situation
can be optimized by selectively enabling director methods (using the
%feature directive) for only those methods that are likely to be
extended in Python.</p>
<h3><a name="Python_nn38">33.5.6 Typemaps</a></h3>
<p> Typemaps for input and output of most of the basic types from
director classes have been written. These are roughly the reverse of
the usual input and output typemaps used by the wrapper code. The
typemap operation names are 'directorin', 'directorout', and
'directorargout'. The director code does not currently use any of the
other kinds of typemaps. It is not clear at this point which kinds are
appropriate and need to be supported.</p>
<h3><a name="Python_director_thread_safety">33.5.7 Thread safety</a></h3>
<p> The director implementation uses a <tt>std::map</tt> to manage C++
pointer ownership. This code is not thread-safe by default. See <a href="#Python_multithreaded">
Thread safety</a> for turning on thread-safety. This defines a macro
called <tt>SWIG_THREADS</tt> which forces the C++ code to then use a
mutex for managing the pointer ownership, which in turn makes the
directors implementation thread-safe.</p>
<p> Please see the <tt>director_guard.swg</tt> Library file for details
of the implementation.</p>
<h3><a name="Python_nn39">33.5.8 Miscellaneous</a></h3>
<p> Director typemaps for STL classes are in place, and hence you should
be able to use std::vector, std::string, etc., as you would any other
type.</p>
<p><b> Note:</b> The director typemaps for return types based on const
references, such as<div class="code">
<pre>
class Foo {
&hellip;
virtual const int&amp; bar();
&hellip;
};
</pre>
</div></p>
<p> will work only for simple call scenarios. Usually the resulting code
is neither thread nor reentrant safe. Hence, the user is advised to
avoid returning const references in director methods. For example, the
user could modify the method interface to use return by value semantics
instead.</p>
<div class="code">
<pre>
class Foo {
&hellip;
virtual int bar();
&hellip;
};
</pre>
</div>
<p> If that is not possible, the user should avoid enabling the director
feature for reentrant, recursive or threaded member methods that return
const references.</p>
<h2><a name="Python_nn40">33.6 Common customization features</a></h2>
<p> The last section presented the absolute basics of C/C++ wrapping. If
you do nothing but feed SWIG a header file, you will get an interface
that mimics the behavior described. However, sometimes this isn't
enough to produce a nice module. Certain types of functionality might
be missing or the interface to certain functions might be awkward. This
section describes some common SWIG features that are used to improve
your the interface to an extension module.</p>
<h3><a name="Python_nn41">33.6.1 C/C++ helper functions</a></h3>
<p> Sometimes when you create a module, it is missing certain bits of
functionality. For example, if you had a function like this</p>
<div class="code">
<pre>
void set_transform(Image *im, double m[4][4]);
</pre>
</div>
<p> it would be accessible from Python, but there may be no easy way to
call it. For example, you might get errors like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = [
... [1, 0, 0, 0],
... [0, 1, 0, 0],
... [0, 0, 1, 0],
... [0, 0, 0, 1]]
&gt;&gt;&gt; set_transform(im, a)
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
TypeError: Type error. Expected _p_a_4__double
</pre>
</div>
<p> The problem here is that there is no easy way to construct and
manipulate a suitable <tt>double [4][4]</tt> value to use. To fix this,
you can write some extra C helper functions. Just use the <tt>%inline</tt>
directive. For example:</p>
<div class="code">
<pre>
%inline %{
/* Note: double[4][4] is equivalent to a pointer to an array double (*)[4] */
double (*new_mat44())[4] {
return (double (*)[4]) malloc(16*sizeof(double));
}
void free_mat44(double (*x)[4]) {
free(x);
}
void mat44_set(double x[4][4], int i, int j, double v) {
x[i][j] = v;
}
double mat44_get(double x[4][4], int i, int j) {
return x[i][j];
}
%}
</pre>
</div>
<p> From Python, you could then write code like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = new_mat44()
&gt;&gt;&gt; mat44_set(a, 0, 0, 1.0)
&gt;&gt;&gt; mat44_set(a, 1, 1, 1.0)
&gt;&gt;&gt; mat44_set(a, 2, 2, 1.0)
...
&gt;&gt;&gt; set_transform(im, a)
&gt;&gt;&gt;
</pre>
</div>
<p> Admittedly, this is not the most elegant looking approach. However,
it works and it wasn't too hard to implement. It is possible to clean
this up using Python code, typemaps, and other customization features
as covered in later sections.</p>
<h3><a name="Python_nn42">33.6.2 Adding additional Python code</a></h3>
<p> If writing support code in C isn't enough, it is also possible to
write code in Python. This code gets inserted in to the <tt>.py</tt>
file created by SWIG. One use of Python code might be to supply a
high-level interface to certain functions. For example:</p>
<div class="code">
<pre>
/* Rename the SWIG-generated wrapper. */
%rename _set_transform set_transform;
...
void set_transform(Image *im, double x[4][4]);
...
/* Rewrite the high level interface to set_transform */
%pythoncode %{
def set_transform(im, x):
a = new_mat44()
for i in range(4):
for j in range(4):
mat44_set(a, i, j, x[i][j])
_example._set_transform(im, a)
free_mat44(a)
%}
</pre>
</div>
<p> In this example, <tt>set_transform()</tt> provides a high-level
Python interface built on top of low-level helper functions. For
example, this code now seems to work:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = [
... [1, 0, 0, 0],
... [0, 1, 0, 0],
... [0, 0, 1, 0],
... [0, 0, 0, 1]]
&gt;&gt;&gt; set_transform(im, a)
&gt;&gt;&gt;
</pre>
</div>
<p> Admittedly, this whole scheme for wrapping the two-dimension array
argument is rather ad-hoc. Besides, shouldn't a Python list or a
Numeric Python array just work normally? We'll get to those examples
soon enough. For now, think of this example as an illustration of what
can be done without having to rely on any of the more advanced
customization features.</p>
<p> There is also <tt>%pythonbegin</tt> which is another directive very
similar to <tt>%pythoncode</tt>, but generates the given Python code at
the beginning of the <tt>.py</tt> file. This directive works in the
same way as <tt>%pythoncode</tt>, except the code is copied just after
the SWIG banner (comment) at the top of the file, before any real code.
This provides an opportunity to add your own description in a comment
near the top of the file as well as Python imports that have to appear
at the top of the file, such as &quot;<tt>from __future__ import</tt>&quot;
statements.</p>
<p> The following example for Python 2.x shows how to insert code into
the generated wrapper to enable <tt>print</tt> to be used as a
Python3-compatible function instead of a statement:</p>
<div class="code">
<pre>
%pythonbegin %{
# This module provides wrappers to the Whizz Bang library
%}
%pythonbegin %{
from __future__ import print_function
print(&quot;Loading&quot;, &quot;Whizz&quot;, &quot;Bang&quot;, sep=' ... ')
%}
</pre>
</div>
<p> The insert code can be seen at the start of the generated <tt>.py</tt>
file:</p>
<div class="code">
<pre>
# This file was automatically generated by SWIG (https://www.swig.org).
# Version 4.0.0
#
# Do not make changes to this file unless you know what you are doing--modify
# the SWIG interface file instead.
# This module provides wrappers to the Whizz Bang library
from __future__ import print_function
print(&quot;Loading&quot;, &quot;Whizz&quot;, &quot;Bang&quot;, sep=' ... ')
</pre>
</div>
<p>When using <tt>%pythoncode</tt> and <tt>%pythonbegin</tt> you
generally want to make sure that the block is delimited by <tt>%{</tt>
and <tt>%}</tt>. If you delimit it with <tt>{</tt> and <tt>}</tt> then
any lines with a leading <tt>#</tt> will be handled by SWIG as
preprocessor directives, when you probably meant them as Python
comments. Prior to SWIG 3.0.3, invalid preprocessor directives were
silently ignored, so generally using the wrong delimiters resulted in
such comments not appearing in the generated output (though a comment
starting with a valid preprocessor directive could cause problems, for
example: <tt># error handling</tt>). SWIG 3.0.3 and later report an
error for invalid preprocessor directives, so you may have to update
existing interface files to delimit blocks of Python code correctly.</p>
<p>As an alternative to providing a block containing Python code, you
can include Python code from a file. The code is inserted exactly as in
the file, so this avoids any issues with the SWIG preprocessor. It's a
good approach if you have a non-trivial chunk of Python code to insert.
To use this feature you specify a filename in double quotes, for
example:</p>
<div class="code">
<pre>
%pythoncode &quot;somecode.py&quot;
</pre>
</div>
<p>Sometimes you may want to replace or modify the wrapper function that
SWIG creates in the proxy <tt>.py</tt> file. The Python module in SWIG
provides some features that enable you to do this. First, to entirely
replace a proxy function you can use <tt>%feature(&quot;shadow&quot;)</tt>. For
example:</p>
<div class="code">
<pre>
%module example
// Rewrite bar() Python code
%feature(&quot;shadow&quot;) Foo::bar(int) %{
def bar(*args):
#do something before
$action
#do something after
%}
class Foo {
public:
int bar(int x);
};
</pre>
</div>
<p> where <tt>$action</tt> will be replaced by the call to the C/C++
proper method. Note that this does not include the arguments to the
C/C++ method.</p>
<p> Often the proxy function created by SWIG is fine, but you simply
want to add code to it without touching the rest of the generated
function body. For these cases SWIG provides the <tt>pythonprepend</tt>
and <tt>pythonappend</tt> features which do exactly as their names
suggest. The <tt>pythonprepend</tt> feature will insert its value at
the beginning of the proxy function, and <tt>pythonappend</tt> will
insert code at the end of the proxy, just before the return statement.</p>
<div class="code">
<pre>
%module example
// Add Python code to bar()
%feature(&quot;pythonprepend&quot;) Foo::bar(int) %{
#do something before C++ call
%}
%feature(&quot;pythonappend&quot;) Foo::bar(int) %{
#do something after C++ call
#the 'val' variable holds the return value
%}
class Foo {
public:
int bar(int x);
};
</pre>
</div>
<p> Notes: Usually the <tt>pythonappend</tt> and <tt>pythonprepend</tt>
features are safer to use than the <tt>shadow</tt> feature. Also, from
SWIG version 1.3.28 you can use the directive forms <tt>%pythonappend</tt>
and <tt>%pythonprepend</tt> as follows:</p>
<div class="code">
<pre>
%module example
// Add Python code to bar()
%pythonprepend Foo::bar(int) %{
#do something before C++ call
%}
%pythonappend Foo::bar(int) %{
#do something after C++ call
#the 'val' variable holds the return value
%}
class Foo {
public:
int bar(int x);
};
</pre>
</div>
<p> Note that when the underlying C++ method is overloaded, there is
only one proxy Python method for multiple C++ methods. In this case,
only one of parsed methods is examined for the feature. You are better
off specifying the feature without the argument list to ensure it will
get used, as it will then get attached to all the overloaded C++
methods. For example:</p>
<div class="code">
<pre>
%module example
// Add Python code to bar()
%pythonprepend Foo::bar %{
#do something before C++ call
%}
%pythonappend Foo::bar %{
#do something after C++ call
%}
class Foo {
public:
int bar(int x);
int bar();
};
</pre>
</div>
<p> The same applies for overloaded constructors.</p>
<h3><a name="Python_nn43">33.6.3 Class extension with %extend</a></h3>
<p> One of the more interesting features of SWIG is that it can extend
structures and classes with new methods--at least in the Python
interface. Here is a simple example:</p>
<div class="code">
<pre>
%module example
%{
#include &quot;someheader.h&quot;
%}
struct Vector {
double x, y, z;
};
%extend Vector {
char *__str__() {
static char tmp[1024];
sprintf(tmp, &quot;Vector(%g, %g, %g)&quot;, $self-&gt;x, $self-&gt;y, $self-&gt;z);
return tmp;
}
Vector(double x, double y, double z) {
Vector *v = (Vector *) malloc(sizeof(Vector));
v-&gt;x = x;
v-&gt;y = y;
v-&gt;z = z;
return v;
}
};
</pre>
</div>
<p> Now, in Python</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; v = example.Vector(2, 3, 4)
&gt;&gt;&gt; print(v)
Vector(2, 3, 4)
&gt;&gt;&gt;
</pre>
</div>
<p> <tt>%extend</tt> can be used for many more tasks than this. For
example, if you wanted to overload a Python operator, you might do
this:</p>
<div class="code">
<pre>
%extend Vector {
Vector __add__(Vector *other) {
Vector v;
v.x = $self-&gt;x + other-&gt;x;
v.y = $self-&gt;y + other-&gt;y;
v.z = $self-&gt;z + other-&gt;z;
return v;
}
};
</pre>
</div>
<p> Use it like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; import example
&gt;&gt;&gt; v = example.Vector(2, 3, 4)
&gt;&gt;&gt; w = example.Vector(10, 11, 12)
&gt;&gt;&gt; print(v+w)
Vector(12, 14, 16)
&gt;&gt;&gt;
</pre>
</div>
<p> <tt>%extend</tt> works with both C and C++ code. It does not modify
the underlying object in any way---the extensions only show up in the
Python interface.</p>
<h3><a name="Python_nn44">33.6.4 Exception handling with %exception</a></h3>
<p> If a C or C++ function throws an error, you may want to convert that
error into a Python exception. To do this, you can use the <tt>
%exception</tt> directive. <tt>%exception</tt> simply lets you rewrite
part of the generated wrapper code to include an error check.</p>
<p> In C, a function often indicates an error by returning a status code
(a negative number or a NULL pointer perhaps). Here is a simple example
of how you might handle that:</p>
<div class="code">
<pre>
%exception malloc {
$action
if (!result) {
PyErr_SetString(PyExc_MemoryError, &quot;Not enough memory&quot;);
SWIG_fail;
}
}
void *malloc(size_t nbytes);
</pre>
</div>
<p> In Python,</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = example.malloc(2000000000)
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
MemoryError: Not enough memory
&gt;&gt;&gt;
</pre>
</div>
<p> If a library provides some kind of general error handling framework,
you can also use that. For example:</p>
<div class="code">
<pre>
%exception {
$action
if (err_occurred()) {
PyErr_SetString(PyExc_RuntimeError, err_message());
SWIG_fail;
}
}
</pre>
</div>
<p> No declaration name is given to <tt>%exception</tt>, it is applied
to all wrapper functions.</p>
<p> C++ exceptions are also easy to handle. For example, you can write
code like this:</p>
<div class="code">
<pre>
%exception getitem {
try {
$action
} catch (std::out_of_range &amp;e) {
PyErr_SetString(PyExc_IndexError, const_cast&lt;char*&gt;(e.what()));
SWIG_fail;
}
}
class Base {
public:
Foo *getitem(int index); // Exception handled added
...
};
</pre>
</div>
<p> When raising a Python exception from C, use the <tt>
PyErr_SetString()</tt> function as shown above followed by <tt>SWIG_fail</tt>
. The following exception types can be used as the first argument.</p>
<div class="code">
<pre>
PyExc_ArithmeticError
PyExc_AssertionError
PyExc_AttributeError
PyExc_EnvironmentError
PyExc_EOFError
PyExc_Exception
PyExc_FloatingPointError
PyExc_ImportError
PyExc_IndexError
PyExc_IOError
PyExc_KeyError
PyExc_KeyboardInterrupt
PyExc_LookupError
PyExc_MemoryError
PyExc_NameError
PyExc_NotImplementedError
PyExc_OSError
PyExc_OverflowError
PyExc_RuntimeError
PyExc_StandardError
PyExc_SyntaxError
PyExc_SystemError
PyExc_TypeError
PyExc_UnicodeError
PyExc_ValueError
PyExc_ZeroDivisionError
</pre>
</div>
<p> <tt>SWIG_fail</tt> is a C macro which when called within the context
of SWIG wrapper function, will jump to the error handler code. This
will call any cleanup code (freeing any temp variables) and then return
from the wrapper function so that the Python interpreter can raise the
Python exception. This macro should always be called after setting a
Python error in code snippets, such as typemaps and <tt>%exception</tt>
, that are ultimately generated into the wrapper function.</p>
<p> The language-independent <tt>exception.i</tt> library file can also
be used to raise exceptions. See the <a href="#Library">SWIG Library</a>
chapter.</p>
<h3><a name="Python_optimization">33.6.5 Optimization options</a></h3>
<h4><a name="Python_fastproxy">33.6.5.1 -fastproxy</a></h4>
<p> The <tt>-fastproxy</tt> command line option enables faster method
calling as the call is made directly into the C/C++ layer rather than
going through a method wrapper.</p>
<p> Consider wrapping a C++ class:</p>
<div class="code">
<pre>
struct Go {
void callme0() {}
void callme4(int a, int b, int c, int d) {}
void callme8(double a, double b, double c, double d, double e, double f, double g, double i) {}
};
</pre>
</div>
<p> The default generated proxy class is:</p>
<div class="targetlang">
<pre>
class Go(object):
def callme0(self):
return _example.Go_callme0(self)
def callme4(self, a, b, c, d):
return _example.Go_callme4(self, a, b, c, d)
def callme8(self, a, b, c, d, e, f, g, i):
return _example.Go_callme8(self, a, b, c, d, e, f, g, i)
...
</pre>
</div>
<p> Each method in the Python class contains a Python proxy method which
passes the arguments on to the underlying function in the low-level
C/C++ module (_example in this case). The generated proxy class when
using <tt>-fastproxy</tt> is:</p>
<div class="targetlang">
<pre>
%module example
class Go(object):
callme0 = _swig_new_instance_method(_example.Go_callme0)
callme4 = _swig_new_instance_method(_example.Go_callme4)
callme8 = _swig_new_instance_method(_example.Go_callme8)
...
</pre>
</div>
<p> where <tt>_swig_new_instance_method</tt> adds the method to the
proxy class via C API calls for direct access to the underlying
function in the low-level C/C++ module. Note that for some methods it
is not possible to generate the direct access call and so <tt>
-fastproxy</tt> is ignored. This happens, for example, when adding <a href="#Python_nn42">
additional code</a> to Python proxy methods, such as using <tt>
%pythonprepend</tt>.</p>
<p> The overhead calling into C/C++ from Python is reduced slightly
using <tt>-fastproxy</tt>. Below are some timings in microseconds
calling the 3 functions in the example above. Also included in the
table for comparison is using the <tt>-builtin</tt> option covered in
the <a href="#Python_builtin_types">Built-in Types</a>.</p>
<table summary="Python fastproxy performance">
<tr><th>Method name</th><th>Default</th><th>-fastproxy</th><th>-builtin</th>
</tr>
<tr><td>callme0</td><td>0.15</td><td>0.09</td><td>0.07</td></tr>
<tr><td>callme4</td><td>0.26</td><td>0.16</td><td>0.14</td></tr>
<tr><td>callme8</td><td>0.32</td><td>0.20</td><td>0.17</td></tr>
</table>
<p> Although the <tt>-fastproxy</tt> option results in faster code over
the default, the generated proxy code is not as user-friendly as
docstring/doxygen comments, <a href="#Python_annotations">Python
annotations</a> and functions with default values are not visible in
the generated Python proxy class. The <tt>-olddefs</tt> option can
rectify this.</p>
<p> The generated proxy class for the example above when using <tt>
-fastproxy -olddefs</tt> is:</p>
<div class="targetlang">
<pre>
class Go(object):
def callme0(self):
return _example.Go_callme0(self)
callme0 = _swig_new_instance_method(_example.Go_callme0)
def callme4(self, a, b, c, d):
return _example.Go_callme4(self, a, b, c, d)
callme4 = _swig_new_instance_method(_example.Go_callme4)
def callme8(self, a, b, c, d, e, f, g, i):
return _example.Go_callme8(self, a, b, c, d, e, f, g, i)
callme8 = _swig_new_instance_method(_example.Go_callme8)
...
</pre>
</div>
<p> The class defines each method in two different ways. The first
definition is replaced by the second definition and so the second
definition is the one used when the method is called. While this
possibly provides the best of both worlds, the time to import the
module will be slightly slower when the class is defined due to the
additional method definitions.</p>
<p> The command line options mentioned above also apply to wrapped C/C++
global functions, not just class methods.</p>
<h3><a name="Python_stable_abi">33.6.6 Stable ABI</a></h3>
<p> By default, the version of Python used to compile the wrappers needs
to be the same as that used during runtime. Alternatvely, the <a href="https://docs.python.org/3/c-api/stable.html">
Python Stable ABI</a> enables a single compiled binary to be used by
different versions of Python. This is enabled by defining <tt>
Py_LIMITED_API</tt> during the compilation of the C/C++ wrapper code and
setting this macro to a particular minimum version of Python that one
wants to support.</p>
<p> SWIG supports the stable ABI, but only version 3.4 of Python and
later is supported. There are two recommended approaches for using SWIG
and the stable ABI and both require setting the <tt>Py_LIMITED_API</tt>
macro to be set to <tt>0x03040000</tt> as a minimum value (Python 3.4).
Either set this using <tt>%begin</tt> by adding the following into your
interface file so that this macro appears at the beginning of the
generated C/C++ code:</p>
<div class="code">
<pre>
%begin %{
#define Py_LIMITED_API 0x03040000
%}
</pre>
</div>
<p> or simply define the macro using your C/C++ compiler's <tt>-D</tt>
command line option, for example, <tt>-DPy_LIMITED_API=0x03040000</tt>.</p>
<p> The default SWIG command line options generate code that enable the
limited API/stable ABI. Some options, such as <tt>-builtin</tt>, <tt>
-fast</tt> (used by <tt>-O</tt>) do not use the limited API and hence
when <tt>Py_LIMITED_API</tt> is defined there will be missing symbols
during compilation. Compatibility of user's custom typemaps is of
course dependent on the Python APIs used in the typemaps.</p>
<p> As discussed in the official <a href="https://docs.python.org/3/c-api/stable.html">
Python Stable ABI</a> documentation, there is no guarantee that code
conforms to the Limited API or Stable ABI when defining <tt>
Py_LIMITED_API</tt>. There are a number of recommendations and caveats
detailed there which are worth studying. The Stable ABI is also
evolving and being modified with each Python release. In fact there is
a Python C API working group, <a href="https://github.com/capi-workgroup/">
capi-workgroup</a>, which was setup in the first half of 2023. The <a href="https://github.com/capi-workgroup/problems/issues">
capi-workgroup/problems issues</a> is very enlightening reading for
anyone using the Stable ABI. <a href="https://peps.python.org/pep-0733/">
PEP 733</a> came out of the working group and is an evaluation of
Python's C API.</p>
<p> There is a lot of information to digest in this space, but a simple
approach would be to follow the offical recommendation of testing an
extension with all minor Python versions you want to support, and to
build with the lowest supported version. SWIG supports at least Python
3.4 for the Stable ABI so this would be a good version to compile a
SWIG extension against. There is also a useful tool <a href="https://github.com/trailofbits/abi3audit/">
abi3audit</a> which can be used to check the compiled extension for
Stable ABI conformance. It can be easily installed with pip and then
used to check a SWIG generated shared object or wheel even. Note that
if you are using it to check a shared object, the shared object must be
renamed to have abi3 in the name. For example, below is the output on
Linux for the class example shipped with SWIG, if <tt>Py_LIMITED_API</tt>
is set as described earlier:</p>
<div class="shell">
<pre>
~/swig/Examples/python/class $ mv _example.so _example.abi3.so
~/swig/Examples/python/class $ abi3audit --assume-minimum-abi3 3.4 _example.abi3.so
[22:12:18] WARNING no wheel to infer abi3 version from; assuming (3.4)
[22:12:18] _example.abi3.so: 1 extensions scanned; 0 ABI version mismatches and 0 ABI violations found
</pre>
</div>
<p> If <tt>Py_LIMITED_API</tt> is not set, the check fails, but this
should be expected:</p>
<div class="shell">
<pre>
~/swig/Examples/python/class $ mv _example.so _example.abi3.so
~/swig/Examples/python/class $ abi3audit --report --assume-minimum-abi3 3.4 _example.abi3.so
[22:22:27] WARNING no wheel to infer abi3 version from; assuming (3.4)
[22:22:27] _example.abi3.so: 1 extensions scanned; 1 ABI version mismatches and 0 ABI violations found
{&quot;specs&quot;: {&quot;_example.abi3.so&quot;: {&quot;kind&quot;: &quot;object&quot;, &quot;object&quot;: {&quot;name&quot;: &quot;_example.abi3.so&quot;,
&quot;result&quot;: {&quot;is_abi3&quot;: true, &quot;is_abi3_baseline_compatible&quot;: false, &quot;baseline&quot;: &quot;3.4&quot;, &quot;computed&quot;: &quot;3.10&quot;,
&quot;non_abi3_symbols&quot;: [], &quot;future_abi3_objects&quot;: {&quot;PyUnicode_AsUTF8AndSize&quot;: &quot;3.10&quot;}}}}}}
</pre>
</div>
<p> Note that ABI conformance as measured by <tt>abi3audit</tt> is far
from simple and can depend on function inlining. There are functions in
Python.h, which if inlined, become ABI compliant. Your chances of your
C or C++ compiler inlining functions is greater with higher compiler
optimisation levels. So merely by changing your compiler optimisation
level may switch your Python extension from being identified as ABI
compliant or not.</p>
<p><b> Compatibility Note:</b> Support for the stable ABI was added in
SWIG-4.2.0.</p>
<h2><a name="Python_nn45">33.7 Tips and techniques</a></h2>
<p> Although SWIG is largely automatic, there are certain types of
wrapping problems that require additional user input. Examples include
dealing with output parameters, strings, binary data, and arrays. This
chapter discusses the common techniques for solving these problems.</p>
<h3><a name="Python_nn46">33.7.1 Input and output parameters</a></h3>
<p> A common problem in some C programs is handling parameters passed as
simple pointers. For example:</p>
<div class="code">
<pre>
void add(int x, int y, int *result) {
*result = x + y;
}
</pre>
</div>
<p> or perhaps</p>
<div class="code">
<pre>
int sub(int *x, int *y) {
return *x-*y;
}
</pre>
</div>
<p> The easiest way to handle these situations is to use the <tt>
typemaps.i</tt> file. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
void add(int, int, int *OUTPUT);
int sub(int *INPUT, int *INPUT);
</pre>
</div>
<p> In Python, this allows you to pass simple values. For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = add(3, 4)
&gt;&gt;&gt; print(a)
7
&gt;&gt;&gt; b = sub(7, 4)
&gt;&gt;&gt; print(b)
3
&gt;&gt;&gt;
</pre>
</div>
<p> Notice how the <tt>INPUT</tt> parameters allow integer values to be
passed instead of pointers and how the <tt>OUTPUT</tt> parameter
creates a return result.</p>
<p> If you don't want to use the names <tt>INPUT</tt> or <tt>OUTPUT</tt>
, use the <tt>%apply</tt> directive. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *result };
%apply int *INPUT { int *x, int *y};
void add(int x, int y, int *result);
int sub(int *x, int *y);
</pre>
</div>
<p> If a function mutates one of its parameters like this,</p>
<div class="code">
<pre>
void negate(int *x) {
*x = -(*x);
}
</pre>
</div>
<p> you can use <tt>INOUT</tt> like this:</p>
<div class="code">
<pre>
%include &quot;typemaps.i&quot;
...
void negate(int *INOUT);
</pre>
</div>
<p> In Python, a mutated parameter shows up as a return value. For
example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = negate(3)
&gt;&gt;&gt; print(a)
-3
&gt;&gt;&gt;
</pre>
</div>
<p> Note: Since most primitive Python objects are immutable, it is not
possible to perform in-place modification of a Python object passed as
a parameter.</p>
<p> The most common use of these special typemap rules is to handle
functions that return more than one value. For example, sometimes a
function returns a result as well as a special error code:</p>
<div class="code">
<pre>
/* send message, return number of bytes sent, along with success code */
int send_message(char *text, int *success);
</pre>
</div>
<p> To wrap such a function, simply use the <tt>OUTPUT</tt> rule above.
For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *success };
...
int send_message(char *text, int *success);
</pre>
</div>
<p> When used in Python, the function will return multiple values.</p>
<div class="targetlang">
<pre>
bytes, success = send_message(&quot;Hello World&quot;)
if not success:
print(&quot;Whoa!&quot;)
else:
print(&quot;Sent&quot;, bytes)
</pre>
</div>
<p> Another common use of multiple return values are in query functions.
For example:</p>
<div class="code">
<pre>
void get_dimensions(Matrix *m, int *rows, int *columns);
</pre>
</div>
<p> To wrap this, you might use the following:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *rows, int *columns };
...
void get_dimensions(Matrix *m, int *rows, *columns);
</pre>
</div>
<p> Now, in Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; r, c = get_dimensions(m)
</pre>
</div>
<p> Be aware that the primary purpose of the <tt>typemaps.i</tt> file is
to support primitive datatypes. Writing a function like this</p>
<div class="code">
<pre>
void foo(Bar *OUTPUT);
</pre>
</div>
<p> may not have the intended effect since <tt>typemaps.i</tt> does not
define an OUTPUT rule for <tt>Bar</tt>.</p>
<h3><a name="Python_nn47">33.7.2 Simple pointers</a></h3>
<p> If you must work with simple pointers such as <tt>int *</tt> or <tt>
double *</tt> and you don't want to use <tt>typemaps.i</tt>, consider
using the <tt>cpointer.i</tt> library file. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;cpointer.i&quot;
%inline %{
extern void add(int x, int y, int *result);
%}
%pointer_functions(int, intp);
</pre>
</div>
<p> The <tt>%pointer_functions(type, name)</tt> macro generates five
helper functions that can be used to create, destroy, copy, assign, and
dereference a pointer. In this case, the functions are as follows:</p>
<div class="code">
<pre>
int *new_intp();
int *copy_intp(int *x);
void delete_intp(int *x);
void intp_assign(int *x, int value);
int intp_value(int *x);
</pre>
</div>
<p> In Python, you would use the functions like this:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; result = new_intp()
&gt;&gt;&gt; print(result)
_108fea8_p_int
&gt;&gt;&gt; add(3, 4, result)
&gt;&gt;&gt; print(intp_value(result))
7
&gt;&gt;&gt;
</pre>
</div>
<p> If you replace <tt>%pointer_functions()</tt> by <tt>
%pointer_class(type, name)</tt>, the interface is more class-like.</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; result = intp()
&gt;&gt;&gt; add(3, 4, result)
&gt;&gt;&gt; print(result.value())
7
</pre>
</div>
<p> See the <a href="#Library">SWIG Library</a> chapter for further
details.</p>
<h3><a name="Python_nn48">33.7.3 Unbounded C Arrays</a></h3>
<p> Sometimes a C function expects an array to be passed as a pointer.
For example,</p>
<div class="code">
<pre>
int sumitems(int *first, int nitems) {
int i, sum = 0;
for (i = 0; i &lt; nitems; i++) {
sum += first[i];
}
return sum;
}
</pre>
</div>
<p> To wrap this into Python, you need to pass an array pointer as the
first argument. A simple way to do this is to use the <tt>carrays.i</tt>
library file. For example:</p>
<div class="code">
<pre>
%include &quot;carrays.i&quot;
%array_class(int, intArray);
</pre>
</div>
<p> The <tt>%array_class(type, name)</tt> macro creates wrappers for an
unbounded array object that can be passed around as a simple pointer
like <tt>int *</tt> or <tt>double *</tt>. For instance, you will be
able to do this in Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = intArray(10000000) # Array of 10-million integers
&gt;&gt;&gt; for i in xrange(10000): # Set some values
... a[i] = i
&gt;&gt;&gt; sumitems(a, 10000)
49995000
&gt;&gt;&gt;
</pre>
</div>
<p> The array &quot;object&quot; created by <tt>%array_class()</tt> does not
encapsulate pointers inside a special array object. In fact, there is
no bounds checking or safety of any kind (just like in C). Because of
this, the arrays created by this library are extremely low-level
indeed. You can't iterate over them nor can you even query their
length. In fact, any valid memory address can be accessed if you want
(negative indices, indices beyond the end of the array, etc.). Needless
to say, this approach is not going to suit all applications. On the
other hand, this low-level approach is extremely efficient and well
suited for applications in which you need to create buffers, package
binary data, etc.</p>
<h3><a name="Python_nn49">33.7.4 String handling</a></h3>
<p> If a C function has an argument of <tt>char *</tt>, then a Python
string can be passed as input. For example:</p>
<div class="code">
<pre>
// C
void foo(char *s);
</pre>
</div><div class="targetlang">
<pre>
# Python
&gt;&gt;&gt; foo(&quot;Hello&quot;)
</pre>
</div>
<p> When a Python string is passed as a parameter, the C function
receives a pointer to the raw data contained in the string. Since
Python strings are immutable, it is illegal for your program to change
the value. In fact, doing so will probably crash the Python
interpreter.</p>
<p> If your program modifies the input parameter or uses it to return
data, consider using the <tt>cstring.i</tt> library file described in
the <a href="#Library">SWIG Library</a> chapter.</p>
<p> When functions return a <tt>char *</tt>, it is assumed to be a
NULL-terminated string. Data is copied into a new Python string and
returned.</p>
<p> If your program needs to work with binary data, you can use a
typemap to expand a Python string into a pointer/length argument pair.
As luck would have it, just such a typemap is already defined. Just do
this:</p>
<div class="code">
<pre>
%apply (char *STRING, size_t LENGTH) { (char *data, size_t size) };
...
int parity(char *data, size_t size, int initial);
</pre>
</div>
<p> Now in Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; parity(&quot;e\x09ffss\x00\x00\x01\nx&quot;, 0)
</pre>
</div>
<p> If you need to return binary data, you might use the <tt>cstring.i</tt>
library file. The <tt>cdata.i</tt> library can also be used to extra
binary data from arbitrary pointers.</p>
<h3><a name="Python_default_args">33.7.5 Default arguments</a></h3>
<p> C++ default argument code generation is documented in the main <a href="#SWIGPlus_default_args">
Default arguments</a> section. There is also an optional Python specific
feature that can be used called the <tt>python:cdefaultargs</tt> <a href="#Customization_feature_flags">
feature flag</a>. By default, SWIG attempts to convert C++ default
argument values into Python values and generates code into the Python
layer containing these values. For example:</p>
<div class="code">
<pre>
struct CDA {
int fff(int a = 1, bool b = false);
};
</pre>
</div>
<p> From Python this can be called as follows:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; CDA().fff() # C++ layer receives a=1 and b=false
&gt;&gt;&gt; CDA().fff(2) # C++ layer receives a=2 and b=false
&gt;&gt;&gt; CDA().fff(3, True) # C++ layer receives a=3 and b=true
</pre>
</div>
<p> The default code generation in the Python layer is:</p>
<div class="targetlang">
<pre>
class CDA(object):
...
def fff(self, a=1, b=False):
return _default_args.CDA_fff(self, a, b)
</pre>
</div>
<p> Adding the feature:</p>
<div class="code">
<pre>
%feature(&quot;python:cdefaultargs&quot;) CDA::fff;
struct CDA {
int fff(int a = 1, bool b = false);
</pre>
</div>
<p> results in identical behaviour when called from Python, however, it
results in different code generation:</p>
<div class="targetlang">
<pre>
class CDA(object):
...
def fff(self, *args):
return _default_args.CDA_fff(self, *args)
</pre>
</div>
<p> The default arguments are obtained in the C++ wrapper layer instead
of the Python layer. Some code generation modes are quite different, eg
<tt>-builtin</tt> and <tt>-fastproxy</tt>, and are unaffected by <tt>
python:cdefaultargs</tt> as the default values are always obtained from
the C++ layer.</p>
<p> Note that not all default arguments can be converted into a Python
equivalent. When SWIG does not convert them, it will generate code to
obtain them from the C++ layer as if <tt>python:cdefaultargs</tt> was
specified. This will happen if just one argument cannot be converted
into a Python equivalent. This occurs typically when the argument is
not fully numeric, such as <tt>int(1)</tt>:</p>
<div class="code">
<pre>
struct CDA {
int fff(int a = int(1), bool b = false);
};
</pre>
</div>
<p><b> Compatibility Note:</b> SWIG-3.0.6 introduced the <tt>
python:cdefaultargs</tt> feature. Versions of SWIG prior to this varied
in their ability to convert C++ default values into equivalent Python
default argument values.</p>
<h2><a name="Python_nn53">33.8 Typemaps</a></h2>
<p> This section describes how you can modify SWIG's default wrapping
behavior for various C/C++ datatypes using the <tt>%typemap</tt>
directive. This is an advanced topic that assumes familiarity with the
Python C API as well as the material in the &quot;<a href="#Typemaps">
Typemaps</a>&quot; chapter.</p>
<p> Before proceeding, it should be stressed that typemaps are not a
required part of using SWIG---the default wrapping behavior is enough
in most cases. Typemaps are only used if you want to change some aspect
of the primitive C-Python interface or if you want to elevate your guru
status.</p>
<h3><a name="Python_nn54">33.8.1 What is a typemap?</a></h3>
<p> A typemap is nothing more than a code generation rule that is
attached to a specific C datatype. For example, to convert integers
from Python to C, you might define a typemap like this:</p>
<div class="code">
<pre>
%module example
%typemap(in) int {
$1 = (int) PyLong_AsLong($input);
printf(&quot;Received an integer : %d\n&quot;, $1);
}
%inline %{
extern int fact(int n);
%}
</pre>
</div>
<p> Typemaps are always associated with some specific aspect of code
generation. In this case, the &quot;in&quot; method refers to the conversion of
input arguments to C/C++. The datatype <tt>int</tt> is the datatype to
which the typemap will be applied. The supplied C code is used to
convert values. In this code a number of special variable prefaced by a
<tt>$</tt> are used. The <tt>$1</tt> variable is placeholder for a
local variable of type <tt>int</tt>. The <tt>$input</tt> variable is
the input object of type <tt>PyObject *</tt>.</p>
<p> When this example is compiled into a Python module, it operates as
follows:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; from example import *
&gt;&gt;&gt; fact(6)
Received an integer : 6
720
</pre>
</div>
<p> In this example, the typemap is applied to all occurrences of the <tt>
int</tt> datatype. You can refine this by supplying an optional
parameter name. For example:</p>
<div class="code">
<pre>
%module example
%typemap(in) int nonnegative {
$1 = (int) PyLong_AsLong($input);
if ($1 &lt; 0) {
PyErr_SetString(PyExc_ValueError, &quot;Expected a nonnegative value.&quot;);
SWIG_fail;
}
}
%inline %{
extern int fact(int nonnegative);
%}
</pre>
</div>
<p> In this case, the typemap code is only attached to arguments that
exactly match <tt>int nonnegative</tt>.</p>
<p> The application of a typemap to specific datatypes and argument
names involves more than simple text-matching--typemaps are fully
integrated into the SWIG C++ type-system. When you define a typemap for
<tt>int</tt>, that typemap applies to <tt>int</tt> and qualified
variations such as <tt>const int</tt>. In addition, the typemap system
follows <tt>typedef</tt> declarations. For example:</p>
<div class="code">
<pre>
%typemap(in) int n {
$1 = (int) PyLong_AsLong($input);
printf(&quot;n = %d\n&quot;, $1);
}
%inline %{
typedef int Integer;
extern int fact(Integer n); // Above typemap is applied
%}
</pre>
</div>
<p> Typemaps can also be defined for groups of consecutive arguments.
For example:</p>
<div class="code">
<pre>
%typemap(in) (char *str, int len) {
$1 = PyString_AsString($input);
$2 = PyString_Size($input);
};
int count(char c, char *str, int len);
</pre>
</div>
<p> When a multi-argument typemap is defined, the arguments are always
handled as a single Python object. This allows the function to be used
like this (notice how the length parameter is omitted):</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; example.count('e', 'Hello World')
1
&gt;&gt;&gt;
</pre>
</div>
<h3><a name="Python_nn55">33.8.2 Python typemaps</a></h3>
<p> The previous section illustrated an &quot;in&quot; typemap for converting
Python objects to C. A variety of different typemap methods are defined
by the Python module. For example, to convert a C integer back into a
Python object, you might define an &quot;out&quot; typemap like this:</p>
<div class="code">
<pre>
%typemap(out) int {
$result = PyInt_FromLong((long) $1);
}
</pre>
</div>
<p> A detailed list of available methods can be found in the &quot;<a href="#Typemaps">
Typemaps</a>&quot; chapter.</p>
<p> However, the best source of typemap information (and examples) is
probably the Python module itself. In fact, all of SWIG's default type
handling is defined by typemaps. You can view these typemaps by looking
at the files in the SWIG library. Just take into account that in the
latest versions of swig (1.3.22+), the library files are not very
pristine clear for the casual reader, as they used to be. The extensive
use of macros and other ugly techniques in the latest version produce a
very powerful and consistent Python typemap library, but at the cost of
simplicity and pedagogic value.</p>
<p> To learn how to write a simple or your first typemap, you better
take a look at the SWIG library version 1.3.20 or so.</p>
<h3><a name="Python_nn56">33.8.3 Typemap variables</a></h3>
<p> Within typemap code, a number of special variables prefaced with a <tt>
$</tt> may appear. A full list of variables can be found in the &quot;<a href="#Typemaps">
Typemaps</a>&quot; chapter. This is a list of the most common variables:</p>
<p> <tt>$1</tt></p>
<div class="indent"> A C local variable corresponding to the actual type
specified in the <tt>%typemap</tt> directive. For input values, this is
a C local variable that's supposed to hold an argument value. For
output values, this is the raw result that's supposed to be returned to
Python.</div>
<p> <tt>$input</tt></p>
<div class="indent"> A <tt>PyObject *</tt> holding a raw Python object
with an argument or variable value.</div>
<p> <tt>$result</tt></p>
<div class="indent"> A <tt>PyObject *</tt> that holds the result to be
returned to Python.</div>
<p> <tt>$1_name</tt></p>
<div class="indent"> The parameter name that was matched.</div>
<p> <tt>$1_type</tt></p>
<div class="indent"> The actual C datatype matched by the typemap.</div>
<p> <tt>$1_ltype</tt></p>
<div class="indent"> An assignable version of the datatype matched by
the typemap (a type that can appear on the left-hand-side of a C
assignment operation). This type is stripped of qualifiers and may be
an altered version of <tt>$1_type</tt>. All arguments and local
variables in wrapper functions are declared using this type so that
their values can be properly assigned.</div>
<p> <tt>$symname</tt></p>
<div class="indent"> The Python name of the wrapper function being
created.</div>
<h3><a name="Python_nn57">33.8.4 Useful Python Functions</a></h3>
<p> When you write a typemap, you usually have to work directly with
Python objects. The following functions may prove to be useful.</p>
<p><b> Python Integer Functions</b></p>
<div class="code">
<pre>
PyObject *PyInt_FromLong(long l);
long PyInt_AsLong(PyObject *);
int PyInt_Check(PyObject *);
</pre>
</div>
<p><b> Python Floating Point Functions</b></p>
<div class="code">
<pre>
PyObject *PyFloat_FromDouble(double);
double PyFloat_AsDouble(PyObject *);
int PyFloat_Check(PyObject *);
</pre>
</div>
<p><b> Python String Functions</b></p>
<div class="code">
<pre>
PyObject *PyString_FromString(char *);
PyObject *PyString_FromStringAndSize(char *, lint len);
int PyString_Size(PyObject *);
char *PyString_AsString(PyObject *);
int PyString_Check(PyObject *);
</pre>
</div>
<p><b> Python List Functions</b></p>
<div class="code">
<pre>
PyObject *PyList_New(int size);
int PyList_Size(PyObject *list);
PyObject *PyList_GetItem(PyObject *list, int i);
int PyList_SetItem(PyObject *list, int i, PyObject *item);
int PyList_Insert(PyObject *list, int i, PyObject *item);
int PyList_Append(PyObject *list, PyObject *item);
PyObject *PyList_GetSlice(PyObject *list, int i, int j);
int PyList_SetSlice(PyObject *list, int i, int , PyObject *list2);
int PyList_Sort(PyObject *list);
int PyList_Reverse(PyObject *list);
PyObject *PyList_AsTuple(PyObject *list);
int PyList_Check(PyObject *);
</pre>
</div>
<p><b> Python Tuple Functions</b></p>
<div class="code">
<pre>
PyObject *PyTuple_New(int size);
int PyTuple_Size(PyObject *);
PyObject *PyTuple_GetItem(PyObject *, int i);
int PyTuple_SetItem(PyObject *, int i, PyObject *item);
PyObject *PyTuple_GetSlice(PyObject *t, int i, int j);
int PyTuple_Check(PyObject *);
</pre>
</div>
<p><b> Python Dictionary Functions</b></p>
<div class="code">
<pre>
PyObject *PyDict_New();
int PyDict_Check(PyObject *);
int PyDict_SetItem(PyObject *p, PyObject *key, PyObject *val);
int PyDict_SetItemString(PyObject *p, const char *key, PyObject *val);
int PyDict_DelItem(PyObject *p, PyObject *key);
int PyDict_DelItemString(PyObject *p, char *key);
PyObject* PyDict_Keys(PyObject *p);
PyObject* PyDict_Values(PyObject *p);
PyObject* PyDict_GetItem(PyObject *p, PyObject *key);
PyObject* PyDict_GetItemString(PyObject *p, const char *key);
int PyDict_Next(PyObject *p, Py_ssize_t *ppos, PyObject **pkey, PyObject **pvalue);
Py_ssize_t PyDict_Size(PyObject *p);
int PyDict_Update(PyObject *a, PyObject *b);
int PyDict_Merge(PyObject *a, PyObject *b, int override);
PyObject* PyDict_Items(PyObject *p);
</pre>
</div>
<p><b> Python File Conversion Functions</b></p>
<div class="code">
<pre>
PyObject *PyFile_FromFile(FILE *f);
FILE *PyFile_AsFile(PyObject *);
int PyFile_Check(PyObject *);
</pre>
</div>
<p><b> Abstract Object Interface</b></p>
<div class="code">
<pre>
write me
</pre>
</div>
<h2><a name="Python_nn58">33.9 Typemap Examples</a></h2>
<p> This section includes a few examples of typemaps. For more examples,
you might look at the files &quot;<tt>python.swg</tt>&quot; and &quot;<tt>typemaps.i</tt>
&quot; in the SWIG library.</p>
<h3><a name="Python_nn59">33.9.1 Converting a Python list to a char **</a>
</h3>
<p> A common problem in many C programs is the processing of command
line arguments, which are usually passed in an array of NULL terminated
strings. SWIG provides typemaps which allow passing a Python list or
tuple - see <a href="#Library_argcargv">argcargv.i</a>.</p>
<h3><a name="Python_nn60">33.9.2 Expanding a Python object into multiple
arguments</a></h3>
<p> Suppose that you had a collection of C functions with arguments such
as the following:</p>
<div class="code">
<pre>
int foo(int argc, char **argv);
</pre>
</div>
<p> In the previous example, a typemap was written to pass a Python list
as the <tt>char **argv</tt>. This allows the function to be used from
Python as follows:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; foo(4, [&quot;foo&quot;, &quot;bar&quot;, &quot;spam&quot;, &quot;1&quot;])
</pre>
</div>
<p> Although this works, it's a little awkward to specify the argument
count. To fix this, a multi-argument typemap can be defined. This is
not very difficult--you only have to make slight modifications to the
previous example:</p>
<div class="code">
<pre>
%typemap(in) (int argc, char **argv) {
/* Check if is a list */
if (PyList_Check($input)) {
int i;
$1 = PyList_Size($input);
$2 = (char **) malloc(($1+1)*sizeof(char *));
for (i = 0; i &lt; $1; i++) {
PyObject *o = PyList_GetItem($input, i);
if (PyString_Check(o)) {
$2[i] = PyString_AsString(PyList_GetItem($input, i));
} else {
PyErr_SetString(PyExc_TypeError, &quot;list must contain strings&quot;);
SWIG_fail;
}
}
$2[i] = 0;
} else {
PyErr_SetString(PyExc_TypeError, &quot;not a list&quot;);
SWIG_fail;
}
}
%typemap(freearg) (int argc, char **argv) {
free((char *) $2);
}
</pre>
</div>
<p> When writing a multiple-argument typemap, each of the types is
referenced by a variable such as <tt>$1</tt> or <tt>$2</tt>. The
typemap code simply fills in the appropriate values from the supplied
Python object.</p>
<p> With the above typemap in place, you will find it no longer
necessary to supply the argument count. This is automatically set by
the typemap code. For example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; foo([&quot;foo&quot;, &quot;bar&quot;, &quot;spam&quot;, &quot;1&quot;])
</pre>
</div>
<p> If your function is overloaded in C++, for example:</p>
<div class="code">
<pre>
int foo(int argc, char **argv);
int foo();
</pre>
</div>
<p> don't forget to also provide a suitable <a href="#Typemaps_overloading">
typecheck typemap for overloading</a> such as:</p>
<div class="code">
<pre>
%typecheck(SWIG_TYPECHECK_STRING_ARRAY) (int argc, char **argv) {
$1 = PyList_Check($input) ? 1 : 0;
}
</pre>
</div>
<p> If you don't you'll get an error message along the lines of:</p>
<div class="shell">
<pre>
Traceback (most recent call last):
File &quot;runme.py&quot;, line 3, in &gt;module&lt;
example.foo([&quot;foo&quot;, &quot;bar&quot;, &quot;spam&quot;, &quot;1&quot;])
TypeError: Wrong number or type of arguments for overloaded function 'foo'.
Possible C/C++ prototypes are:
foo(int, char **)
foo()
</pre>
</div>
<h3><a name="Python_nn61">33.9.3 Using typemaps to return arguments</a></h3>
<p> A common problem in some C programs is that values may be returned
in arguments rather than in the return value of a function. For
example:</p>
<div class="code">
<pre>
/* Returns a status value and two values in out1 and out2 */
int spam(double a, double b, double *out1, double *out2) {
... Do a bunch of stuff ...
*out1 = result1;
*out2 = result2;
return status;
}
</pre>
</div>
<p> A typemap can be used to handle this case as follows :</p>
<div class="code">
<pre>
%module outarg
// This tells SWIG to treat an double * argument with name 'OutValue' as
// an output value. We'll append the value to the current result which
// is guaranteed to be a List object by SWIG.
%typemap(argout) double *OutValue {
PyObject *o, *o2, *o3;
o = PyFloat_FromDouble(*$1);
if ((!$result) || ($result == Py_None)) {
$result = o;
} else {
if (!PyTuple_Check($result)) {
PyObject *o2 = $result;
$result = PyTuple_New(1);
PyTuple_SetItem($result, 0, o2);
}
o3 = PyTuple_New(1);
PyTuple_SetItem(o3, 0, o);
o2 = $result;
$result = PySequence_Concat(o2, o3);
Py_DecRef(o2);
Py_DecRef(o3);
}
}
int spam(double a, double b, double *OutValue, double *OutValue);
</pre>
</div>
<p> The typemap works as follows. First, a check is made to see if any
previous result exists. If so, it is turned into a tuple and the new
output value is concatenated to it. Otherwise, the result is returned
normally. For the sample function <tt>spam()</tt>, there are three
output values--meaning that the function will return a 3-tuple of the
results.</p>
<p> As written, the function must accept 4 arguments as input values,
last two being pointers to doubles. If these arguments are only used to
hold output values (and have no meaningful input value), an additional
typemap can be written. For example:</p>
<div class="code">
<pre>
%typemap(in, numinputs=0) double *OutValue(double temp) {
$1 = &amp;temp;
}
</pre>
</div>
<p> By specifying numinputs=0, the input value is ignored. However,
since the argument still has to be set to some meaningful value before
calling C, it is set to point to a local variable <tt>temp</tt>. When
the function stores its output value, it will simply be placed in this
local variable. As a result, the function can now be used as follows:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; a = spam(4, 5)
&gt;&gt;&gt; print(a)
(0, 2.45, 5.0)
&gt;&gt;&gt; x, y, z = spam(4, 5)
&gt;&gt;&gt;
</pre>
</div>
<h3><a name="Python_nn62">33.9.4 Mapping Python tuples into small arrays</a>
</h3>
<p> In some applications, it is sometimes desirable to pass small arrays
of numbers as arguments. For example :</p>
<div class="code">
<pre>
extern void set_direction(double a[4]); // Set direction vector
</pre>
</div>
<p> This too, can be handled used typemaps as follows :</p>
<div class="code">
<pre>
// Grab a 4 element array as a Python 4-tuple
%typemap(in) double[4](double temp[4]) { // temp[4] becomes a local variable
int i;
if (PyTuple_Check($input)) {
if (!PyArg_ParseTuple($input, &quot;dddd&quot;, temp, temp+1, temp+2, temp+3)) {
PyErr_SetString(PyExc_TypeError, &quot;tuple must have 4 elements&quot;);
SWIG_fail;
}
$1 = &amp;temp[0];
} else {
PyErr_SetString(PyExc_TypeError, &quot;expected a tuple.&quot;);
SWIG_fail;
}
}
</pre>
</div>
<p> This allows our <tt>set_direction</tt> function to be called from
Python as follows :</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; set_direction((0.5, 0.0, 1.0, -0.25))
</pre>
</div>
<p> Since our mapping copies the contents of a Python tuple into a C
array, such an approach would not be recommended for huge arrays, but
for small structures, this approach works fine.</p>
<h3><a name="Python_nn63">33.9.5 Mapping sequences to C arrays</a></h3>
<p> Suppose that you wanted to generalize the previous example to handle
C arrays of different sizes. To do this, you might write a typemap as
follows:</p>
<div class="code">
<pre>
// Map a Python sequence into any sized C double array
%typemap(in) double[ANY](double temp[$1_dim0]) {
int i;
if (!PySequence_Check($input)) {
PyErr_SetString(PyExc_TypeError, &quot;Expecting a sequence&quot;);
SWIG_fail;
}
if (PyObject_Length($input) != $1_dim0) {
PyErr_SetString(PyExc_ValueError, &quot;Expecting a sequence with $1_dim0 elements&quot;);
SWIG_fail;
}
for (i =0; i &lt; $1_dim0; i++) {
PyObject *o = PySequence_GetItem($input, i);
if (!PyFloat_Check(o)) {
Py_DecRef(o);
PyErr_SetString(PyExc_ValueError, &quot;Expecting a sequence of floats&quot;);
SWIG_fail;
}
temp[i] = PyFloat_AsDouble(o);
Py_DecRef(o);
}
$1 = &amp;temp[0];
}
</pre>
</div>
<p> In this case, the variable <tt>$1_dim0</tt> is expanded to match the
array dimensions actually used in the C code. This allows the typemap
to be applied to types such as:</p>
<div class="code">
<pre>
void foo(double x[10]);
void bar(double a[4], double b[8]);
</pre>
</div>
<p> Since the above typemap code gets inserted into every wrapper
function where used, it might make sense to use a helper function
instead. This will greatly reduce the amount of wrapper code. For
example:</p>
<div class="code">
<pre>
%{
static int convert_darray(PyObject *input, double *ptr, int size) {
int i;
if (!PySequence_Check(input)) {
PyErr_SetString(PyExc_TypeError, &quot;Expecting a sequence&quot;);
return 0;
}
if (PyObject_Length(input) != size) {
PyErr_SetString(PyExc_ValueError, &quot;Sequence size mismatch&quot;);
return 0;
}
for (i =0; i &lt; size; i++) {
PyObject *o = PySequence_GetItem(input, i);
if (!PyFloat_Check(o)) {
Py_DecRef(o);
PyErr_SetString(PyExc_ValueError, &quot;Expecting a sequence of floats&quot;);
return 0;
}
ptr[i] = PyFloat_AsDouble(o);
Py_DecRef(o);
}
return 1;
}
%}
%typemap(in) double [ANY](double temp[$1_dim0]) {
if (!convert_darray($input, temp, $1_dim0)) {
SWIG_fail;
}
$1 = &amp;temp[0];
}
</pre>
</div>
<h3><a name="Python_nn64">33.9.6 Pointer handling</a></h3>
<p> Occasionally, it might be necessary to convert pointer values that
have been stored using the SWIG typed-pointer representation. Since
there are several ways in which pointers can be represented, the
following two functions are used to safely perform this conversion:</p>
<p> <tt>int SWIG_ConvertPtr(PyObject *obj, void **ptr, swig_type_info
*ty, int flags)</tt></p>
<div class="indent"> Converts a Python object <tt>obj</tt> to a C
pointer. The result of the conversion is placed into the pointer
located at <tt>ptr</tt>. <tt>ty</tt> is a SWIG type descriptor
structure. <tt>flags</tt> is used to handle error checking and other
aspects of conversion. It is the bitwise-or of several flag values
including <tt>SWIG_POINTER_DISOWN</tt> (which steals ownership of the
object) and <tt>SWIG_POINTER_NO_NULL</tt> (which makes the conversion
fail if the C pointer would be <tt>NULL</tt>). Returns 0 on success and
-1 on error.</div>
<p> <tt>PyObject *SWIG_NewPointerObj(void *ptr, swig_type_info *ty, int
own)</tt></p>
<div class="indent"> Creates a new Python pointer object. <tt>ptr</tt>
is the pointer to convert, <tt>ty</tt> is the SWIG type descriptor
structure that describes the type, and <tt>own</tt> is a flag that
indicates whether or not Python should take ownership of the pointer.</div>
<p> Both of these functions require the use of a special SWIG
type-descriptor structure. This structure contains information about
the mangled name of the datatype, type-equivalence information, as well
as information about converting pointer values under C++ inheritance.
For a type of <tt>Foo *</tt>, the type descriptor structure is usually
accessed as follows:</p>
<div class="code">
<pre>
Foo *f;
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;f, SWIGTYPE_p_Foo, 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method '$symname', expecting type Foo&quot;);
}
PyObject *obj;
obj = SWIG_NewPointerObj(f, SWIGTYPE_p_Foo, 0);
</pre>
</div>
<p> In a typemap, the type descriptor should always be accessed using
the special typemap variable <tt>$1_descriptor</tt>. For example:</p>
<div class="code">
<pre>
%typemap(in) Foo * {
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;$1, $1_descriptor, 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method '$symname', expecting type Foo&quot;);
}
}
</pre>
</div>
<p> If necessary, the descriptor for any type can be obtained using the <tt>
$descriptor()</tt> macro in a typemap. For example:</p>
<div class="code">
<pre>
%typemap(in) Foo * {
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;$1, $descriptor(Foo *), 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method '$symname', expecting type Foo&quot;);
}
}
</pre>
</div>
<p> Although the pointer handling functions are primarily intended for
manipulating low-level pointers, both functions are fully aware of
Python proxy classes. Specifically, <tt>SWIG_ConvertPtr()</tt> will
retrieve a pointer from any object that has a <tt>this</tt> attribute.
In addition, <tt>SWIG_NewPointerObj()</tt> can automatically generate a
proxy class object (if applicable).</p>
<h3><a name="Python_memory_management_member_variables">33.9.7 Memory
management when returning references to member variables</a></h3>
<p> This example shows how to prevent premature garbage collection of
objects when the underlying C++ class returns a pointer or reference to
a member variable. The example is a direct equivalent to this <a href="#Java_memory_management_objects">
Java equivalent</a>.</p>
<p> Consider the following C++ code:</p>
<div class="code">
<pre>
#include &lt;iostream&gt;
struct Wheel {
int size;
Wheel(int sz = 0) : size(sz) {}
~Wheel() { std::cout &lt;&lt; &quot;~Wheel&quot; &lt;&lt; std::endl; }
};
class Bike {
Wheel wheel;
public:
Bike(int val) : wheel(val) {}
Wheel&amp; getWheel() { return wheel; }
};
</pre>
</div>
<p> and the following usage from Python after running the code through
SWIG:</p>
<div class="code">
<pre>
bike = Bike(10)
wheel = bike.getWheel()
print(&quot;wheel size: {}&quot;.format(wheel.size))
del bike # Allow bike to be garbage collected
print(&quot;wheel size: {}&quot;.format(wheel.size))
</pre>
</div>
<p> Don't be surprised that if the resulting output gives strange
results such as...</p>
<div class="shell">
<pre>
wheel size: 10
~Wheel
wheel size: 135019664
</pre>
</div>
<p> What has happened here is the garbage collector has collected the <tt>
Bike</tt> instance as it doesn't think it is needed any more. The proxy
instance, <tt>wheel</tt>, contains a reference to memory that was
deleted when the <tt>Bike</tt> instance was collected. In order to
prevent the garbage collector from collecting the <tt>Bike</tt>
instance, a reference to the <tt>Bike</tt> must be added to the <tt>
wheel</tt> instance.</p>
<p> You can do this by adding the reference when the <tt>getWheel()</tt>
method is called using one of three approaches:</p>
<p> The easier, but less optimized, way is to use the <tt>%pythonappend</tt>
directive (see <a href="#Python_nn42">Adding additional Python code</a>
):</p>
<div class="code">
<pre>
%pythonappend getWheel %{
# val is the Wheel proxy, self is the Bike instance
val.__bike_reference = self
%}
</pre>
</div>
<p> The code gets appended to the Python code generated for the <tt>
Bike::getWheel</tt> wrapper function, where we store the <tt>Bike</tt>
proxy instance onto the <tt>Wheel</tt> proxy instance before it is
returned to the caller as follows.</p>
<div class="targetlang">
<pre>
class Bike(object):
...
def getWheel(self):
val = _example.Bike_getWheel(self)
# val is the Wheel proxy, self is the Bike instance
val.__bike_reference = self
return val
</pre>
</div>
<p> The second option, which performs better and is required if you use
the <tt>-builtin</tt> option, is to set the reference in the CPython
implementation:<div class="code">
<pre>
%extend Wheel {
// A reference to the parent class is added to ensure the underlying C++
// object is not deleted while the item is in use
%typemap(ret) Wheel&amp; getWheel {
PyObject *bike_reference_string = SWIG_Python_str_FromChar(&quot;__bike_reference&quot;);
PyObject_SetAttr($result, bike_reference_string, $self);
Py_DecRef(bike_reference_string);
}
}
</pre>
</div></p>
<p> The third approach, shown below, is an optimization of the above
approach and creates the &quot;__bike_reference&quot; Python string object just
once. While this looks more complex, it is just a small variation on
the above typemap plus a support function <tt>bike_reference()</tt> in
a fragment called <tt>bike_reference_function</tt>. The <tt>
bike_reference_init</tt> typemap generates code into the &quot;init&quot; section
for an initial call to <tt>bike_reference()</tt> when the module is
initialized and is done to create the &quot;__bike_reference&quot; Python string
singleton in a thread-safe manner.</p>
<div class="code">
<pre>
%fragment(&quot;bike_reference_init&quot;, &quot;init&quot;) {
// Thread-safe initialization - initialize during Python module initialization
bike_reference();
}
%fragment(&quot;bike_reference_function&quot;, &quot;header&quot;, fragment=&quot;bike_reference_init&quot;) {
static PyObject *bike_reference() {
static PyObject *bike_reference_string = SWIG_Python_str_FromChar(&quot;__bike_reference&quot;);
return bike_reference_string;
}
}
%extend Wheel {
// A reference to the parent class is added to ensure the underlying C++
// object is not deleted while the item is in use
%typemap(ret, fragment=&quot;bike_reference_function&quot;) Wheel&amp; getWheel %{
PyObject_SetAttr($result, bike_reference(), $self);
%}
}
</pre>
</div>
<h2><a name="Python_nn65">33.10 Docstring Features</a></h2>
<p> Using docstrings in Python code is becoming more and more important
and more tools are coming on the scene that take advantage of them,
everything from full-blown documentation generators to class browsers
and popup call-tips in Python-aware IDEs. Given the way that SWIG
generates the proxy code by default, your users will normally get
something like <tt>&quot;function_name(*args)&quot;</tt> in the popup calltip of
their IDE which is next to useless when the real function prototype
might be something like this:</p>
<div class="code">
<pre>
bool function_name(int x, int y, Foo* foo=NULL, Bar* bar=NULL);
</pre>
</div>
<p> The features described in this section make it easy for you to add
docstrings to your modules, functions and methods that can then be used
by the various tools out there to make the programming experience of
your users much simpler.</p>
<h3><a name="Python_nn66">33.10.1 Module docstring</a></h3>
<p> Python allows a docstring at the beginning of the <tt>.py</tt> file
before any other statements, and it is typically used to give a general
description of the entire module. SWIG supports this by setting an
option of the <tt>%module</tt> directive. For example:</p>
<div class="code">
<pre>
%module(docstring=&quot;This is the example module's docstring&quot;) example
</pre>
</div>
<p> When you have more than just a line or so then you can retain the
easy readability of the <tt>%module</tt> directive by using a macro.
For example:</p>
<div class="code">
<pre>
%define DOCSTRING
&quot;The `XmlResource` class allows program resources defining menus,
layout of controls on a panel, etc. to be loaded from an XML file.&quot;
%enddef
%module(docstring=DOCSTRING) xrc
</pre>
</div>
<h3><a name="Python_nn67">33.10.2 %feature(&quot;autodoc&quot;)</a></h3>
<p> As alluded to above SWIG will generate all the function and method
proxy wrappers with just &quot;*args&quot; (or &quot;*args, **kwargs&quot; if the -keyword
option is used) for a parameter list and will then sort out the
individual parameters in the C wrapper code. This is nice and simple
for the wrapper code, but makes it difficult to be programmer and tool
friendly as anyone looking at the <tt>.py</tt> file will not be able to
find out anything about the parameters that the functions accept.</p>
<p>But since SWIG does know everything about the function it is possible
to generate a docstring containing the parameter types, names and
default values. Since many of the docstring tools are adopting a
standard of recognizing if the first thing in the docstring is a
function prototype then using that instead of what they found from
introspection, then life is good once more.</p>
<p>SWIG's Python module provides support for the &quot;autodoc&quot; feature,
which when attached to a node in the parse tree will cause a docstring
to be generated that includes the name of the function, parameter
names, default values if any, and return type if any. There are also
four levels for autodoc controlled by the value given to the feature, <tt>
%feature(&quot;autodoc&quot;, &quot;<i>level</i>&quot;)</tt>. The four values for<i> level</i>
are covered in the following sub-sections.</p>
<h4><a name="Python_nn68">33.10.2.1 %feature(&quot;autodoc&quot;, &quot;0&quot;)</a></h4>
<p> When level &quot;0&quot; is used then the types of the parameters will<em> not</em>
be included in the autodoc string. For example, given this function
prototype:</p>
<div class="code">
<pre>
%feature(&quot;autodoc&quot;, &quot;0&quot;);
bool function_name(int x, int y, Foo* foo=NULL, Bar* bar=NULL);
</pre>
</div>
<p> Then Python code like this will be generated:</p>
<div class="targetlang">
<pre>
def function_name(*args, **kwargs):
&quot;&quot;&quot;function_name(x, y, foo=None, bar=None) -&gt; bool&quot;&quot;&quot;
...
</pre>
</div>
<h4><a name="Python_nn69">33.10.2.2 %feature(&quot;autodoc&quot;, &quot;1&quot;)</a></h4>
<p> When level &quot;1&quot; is used then the parameter types<em> will</em> be
used in the autodoc string. In addition, an attempt is made to simplify
the type name such that it makes more sense to the Python user.
Pointer, reference and const info is removed if the associated type is
has an associated Python type (<tt>%rename</tt>'s are thus shown
correctly). This works most of the time, otherwise a C/C++ type will be
used. See the next section for the &quot;docstring&quot; feature for tweaking the
docstrings to your liking. Given the example above, then turning on the
parameter types with level &quot;1&quot; will result in Python code like this:</p>
<div class="targetlang">
<pre>
def function_name(*args, **kwargs):
&quot;&quot;&quot;function_name(int x, int y, Foo foo=None, Bar bar=None) -&gt; bool&quot;&quot;&quot;
...
</pre>
</div>
<h4><a name="Python_autodoc2">33.10.2.3 %feature(&quot;autodoc&quot;, &quot;2&quot;)</a></h4>
<p> Level &quot;2&quot; results in the function prototype as per level &quot;0&quot;. In
addition, a line of documentation is generated for each parameter using
<a href="https://github.com/numpy/numpy/blob/master/doc/HOWTO_DOCUMENT.rst.txt">
numpydoc</a> style. Using the previous example, the generated code will
be:</p>
<div class="targetlang">
<pre>
def function_name(*args, **kwargs):
&quot;&quot;&quot;
function_name(x, y, foo=None, bar=None) -&gt; bool
Parameters
----------
x: int
y: int
foo: Foo *
bar: Bar *
&quot;&quot;&quot;
...
</pre>
</div>
<p> Note that the documentation for each parameter is sourced from the
&quot;doc&quot; typemap which by default shows the C/C++ type rather than the
simplified Python type name described earlier for level &quot;1&quot;. Typemaps
can of course change the output for any particular type, for example
the <tt>int x</tt> parameter:</p>
<div class="code">
<pre>
%feature(&quot;autodoc&quot;, &quot;2&quot;);
%typemap(&quot;doc&quot;) int x &quot;$1_name (C++ type: $1_type) -- Input $1_name dimension&quot;
bool function_name(int x, int y, Foo* foo=NULL, Bar* bar=NULL);
</pre>
</div>
<p> resulting in</p>
<div class="targetlang">
<pre>
def function_name(*args, **kwargs):
&quot;&quot;&quot;
function_name(x, y, foo=None, bar=None) -&gt; bool
Parameters
----------
x (C++ type: int) -- Input x dimension
y: int
foo: Foo *
bar: Bar *
&quot;&quot;&quot;
</pre>
</div>
<h4><a name="Python_autodoc3">33.10.2.4 %feature(&quot;autodoc&quot;, &quot;3&quot;)</a></h4>
<p> Level &quot;3&quot; results in the function prototype as per level &quot;1&quot; but
also contains the same additional line of documentation for each
parameter as per level &quot;2&quot;. Using our earlier example again, the
generated code will be:</p>
<div class="targetlang">
<pre>
def function_name(*args, **kwargs):
&quot;&quot;&quot;
function_name(int x, int y, Foo foo=None, Bar bar=None) -&gt; bool
Parameters
----------
x: int
y: int
foo: Foo *
bar: Bar *
&quot;&quot;&quot;
...
</pre>
</div>
<h4><a name="Python_nn70">33.10.2.5 %feature(&quot;autodoc&quot;, &quot;docstring&quot;)</a></h4>
<p> Finally, there are times when the automatically generated autodoc
string will make no sense for a Python programmer, particularly when a
typemap is involved. So if you give an explicit value for the autodoc
feature then that string will be used in place of the automatically
generated string. For example:</p>
<div class="code">
<pre>
%feature(&quot;autodoc&quot;, &quot;GetPosition() -&gt; (x, y)&quot;) GetPosition;
void GetPosition(int* OUTPUT, int* OUTPUT);
</pre>
</div>
<h3><a name="Python_nn71">33.10.3 %feature(&quot;docstring&quot;)</a></h3>
<p> In addition to the autodoc strings described above, you can also
attach any arbitrary descriptive text to a node in the parse tree with
the &quot;docstring&quot; feature. When the proxy module is generated then any
docstring associated with classes, function or methods are output. If
an item already has an autodoc string then it is combined with the
docstring and they are output together. If the docstring is all on a
single line then it is output like this::</p>
<div class="targetlang">
<pre>
&quot;&quot;&quot;This is the docstring&quot;&quot;&quot;
</pre>
</div>
<p> Otherwise, to aid readability it is output like this:</p>
<div class="targetlang">
<pre>
&quot;&quot;&quot;
This is a multi-line docstring
with more than one line.
&quot;&quot;&quot;
</pre>
</div>
<h3><a name="Python_doxygen_docstrings">33.10.4 Doxygen comments</a></h3>
<p> Please see the separate <a href="#Doxygen">Doxygen</a> chapter for
information on making use of C++ Doxygen comments and translating them
into Python docstring comments.</p>
<p> Note that when generating docstrings and Doxygen comments have also
been turned on, the <a href="#Python_nn71">docstring feature</a> will
take precedence over a Doxygen comment. If the <a href="#Python_nn67">
autodoc feature</a> is also turned on, then it will be used in
conjunction with the docstring feature. However, if there is no
docstring feature present and there is a Doxygen comment, then the
autodoc docstring will not be generated. The Doxygen comment alone will
be used.</p>
<p> This way, if the autodoc feature is specified globally it will fill
in any missing Doxygen documentation comments. Doxygen comments can be
overridden by using the docstring feature.</p>
<h2><a name="Python_nn72">33.11 Python Packages</a></h2>
<p>Python has concepts of modules and packages. Modules are separate
units of code and may be grouped together to form a package. Packages
may be nested, that is they may contain subpackages. This leads to
tree-like hierarchy, with packages as intermediate nodes and modules as
leaf nodes.</p>
<p>The hierarchy of Python packages/modules follows the hierarchy of <tt>
*.py</tt> files found in a source tree (or, more generally, in the
Python path). Normally, the developer creates new module by placing a <tt>
*.py</tt> file somewhere under Python path; the module is then named
after that <tt>*.py</tt> file. A package is created by placing an <tt>
__init__.py</tt> file within a directory; the package is then named
after that directory. For example, the following source tree:</p>
<div class="diagram">
<pre>
mod1.py
pkg1/__init__.py
pkg1/mod2.py
pkg1/pkg2/__init__.py
pkg1/pkg2/mod3.py
</pre>
</div>
<p> defines the following Python packages and modules:</p>
<div class="diagram">
<pre>
pkg1 # package
pkg1.pkg2 # package
mod1 # module
pkg1.mod2 # module
pkg1.pkg2.mod3 # module
</pre>
</div>
<p> The purpose of an <tt>__init__.py</tt> file is two-fold. First, the
existence of <tt>__init__.py</tt> in a directory informs the Python
interpreter that this directory contains a Python package. Second, the
code in <tt>__init__.py</tt> is loaded/executed automatically when the
package is initialized (when it or its submodule/subpackage gets <tt>
import</tt>'ed). By default, SWIG generates proxy Python code &ndash; one <tt>
*.py</tt> file for each <tt>*.i</tt> interface. The <tt>__init__.py</tt>
files, however, are not generated by SWIG. They should be created by
other means. Both files (module <tt>*.py</tt> and <tt>__init__.py</tt>)
should be installed in appropriate destination directories in order to
obtain a desirable package/module hierarchy.</p>
<p> Python3 adds another option for packages with <a href="https://www.python.org/dev/peps/pep-0420/">
PEP 0420</a> (implicit namespace packages). Implicit namespace packages
no longer use __init__.py files. SWIG generated Python modules support
implicit namespace packages. See <a href="#Python_implicit_namespace_packages">
Implicit namespace packages</a> for more information.</p>
<p> You can place a SWIG generated module into a Python package or keep
as a global module, details are covered a little later in <a href="#Python_package_search">
Location of modules</a>.</p>
<p>The way Python defines its modules and packages impacts SWIG users.
Some users may need to use special features such as the <tt>package</tt>
option in the <tt>%module</tt> directive or import related command line
options. These are explained in the following sections.</p>
<h3><a name="Python_modulepackage">33.11.1 Setting the Python package</a>
</h3>
<p> Using the <tt>package</tt> option in the <tt>%module</tt> directive
allows you to specify a Python package that the module will be in when
installed.</p>
<div class="code">
<pre>
%module(package=&quot;wx&quot;) xrc
</pre>
</div>
<p> This is useful when the <tt>.i</tt> file is <tt>%import</tt>ed by
another <tt>.i</tt> file. By default SWIG will assume that the importer
is able to find the importee with just the module name, but if they
live in separate Python packages then this won't work. However if the
importee specifies what its package is with the <tt>%module</tt> option
then the Python code generated for the importer will use that package
name when importing the other module and in base class declarations,
etc..</p>
<p>SWIG assumes that the <tt>package</tt> option provided to <tt>%module</tt>
together with the <tt>module</tt> name (that is, <tt>wx.xrc</tt> in the
above example) forms a fully qualified (absolute) name of a module (in
Python terms). This is important especially for Python 3, where
absolute imports are used by default. It's up to you to place the
generated module files (<tt>.py</tt>, <tt>.so</tt>) in appropriate
subdirectories. For example, if you have an interface file <tt>foo.i</tt>
with:</p>
<div class="code">
<pre>
%module(package=&quot;pkg1.pkg2&quot;) foo
</pre>
</div>
<p> then the resulting directory layout should be</p>
<div class="diagram">
<pre>
pkg1/
pkg1/__init__.py
pkg1/pkg2/__init__.py
pkg1/pkg2/foo.py # (generated by SWIG)
pkg1/pkg2/_foo.so # (shared library built from C/C++ code generated by SWIG)
</pre>
</div>
<h3><a name="Python_absrelimports">33.11.2 Absolute and relative imports</a>
</h3>
<p>Suppose, we have the following hierarchy of files:</p>
<div class="diagram">
<pre>
pkg1/
pkg1/__init__.py
pkg1/mod2.py
pkg1/pkg2/__init__.py
pkg1/pkg2/mod3.py
</pre>
</div>
<p>Let the contents of <tt>pkg1/pkg2/mod3.py</tt> be</p>
<div class="targetlang">
<pre>
class M3: pass
</pre>
</div>
<p> We edit <tt>pkg1/mod2.py</tt> and want to import module <tt>
pkg1/pkg2/mod3.py</tt> in order to derive from class <tt>M3</tt>. We can
write appropriate Python code in several ways, for example:</p>
<ol>
<li>
<p>Using &quot;<tt>import &lt;&gt;</tt>&quot; syntax with absolute package name:</p>
<div class="targetlang">
<pre>
# pkg1/mod2.py
import pkg1.pkg2.mod3
class M2(pkg1.pkg2.mod3.M3): pass
</pre>
</div></li>
<li>
<p>Using &quot;<tt>import &lt;&gt;</tt>&quot; syntax with package name relative to <tt>
pkg1</tt> (only in Python 2.7 and earlier):</p>
<div class="targetlang">
<pre>
# pkg1/mod2.py
import pkg2.mod3
class M2(pkg2.mod3.M3): pass
</pre>
</div></li>
<li>
<p>Using &quot;<tt>from &lt;&gt; import &lt;&gt;</tt>&quot; syntax (relative import syntax,
only in Python 2.5 and later):</p>
<div class="targetlang">
<pre>
# pkg1/mod2.py
from .pkg2 import mod3
class M2(mod3.M3): pass
</pre>
</div></li>
<li>
<p>Other variants, for example the following construction in order to
have the <tt>pkg2.mod3.M3</tt> symbol available in <tt>mod2</tt> as in
point 2 above (but now under Python 3):</p>
<div class="targetlang">
<pre>
# pkg1/mod2.py
from . import pkg2
from .pkg2 import mod3
class M2(pkg2.mod3.M3): pass
</pre>
</div></li>
</ol>
<p>Now suppose we have <tt>mod2.i</tt> with</p>
<div class="code">
<pre>
// mod2.i
%module (package=&quot;pkg1&quot;) mod2
%import &quot;mod3.i&quot;
// ...
</pre>
</div>
<p>and <tt>mod3.i</tt> with</p>
<div class="code">
<pre>
// mod3.i
%module (package=&quot;pkg1.pkg2&quot;) mod3
// ...
</pre>
</div>
<p>By default, SWIG will generate <tt>mod2.py</tt> proxy file with <tt>
import</tt> directive as in point 1. This can be changed with the <tt>
-relativeimport</tt> command line option. The <tt>-relativeimport</tt>
instructs SWIG to organize imports as in point 4 for Python 2.7.0 and
newer.</p>
<p><b> Compatibility Note:</b> Versions of SWIG prior to SWIG-4.0.0
supported Python &lt; 2.7.0 and would organize the imports as in point 2
if an older version of Python was detected at runtime.</p>
<p> In short, if you have <tt>mod2.i</tt> and <tt>mod3.i</tt> as above,
then without <tt>-relativeimport</tt> SWIG will write</p>
<div class="targetlang">
<pre>
import pkg1.pkg2.mod3
</pre>
</div>
<p>to <tt>mod2.py</tt> proxy file, and with <tt>-relativeimport</tt> it
will write</p>
<div class="targetlang">
<pre>
from . import pkg2
from .pkg2 import mod3
</pre>
</div>
<p>You should avoid using relative imports and use absolute ones
whenever possible. There are some cases, however, when relative imports
may be necessary. The first example is, when some (legacy) Python code
refers entities imported by proxy files generated by SWIG, and it
assumes that the proxy file uses relative imports. Second case is, when
one puts import directives in <tt>__init__.py</tt> to import symbols
from submodules or subpackages and the submodule depends on other
submodules (discussed later).</p>
<h3><a name="Python_absimport">33.11.3 Enforcing absolute import
semantics</a></h3>
<p>As you may know, there is an incompatibility in import semantics (for
the <tt>import &lt;&gt;</tt> syntax) between Python 2 and 3. In Python 2.4
and earlier it is not clear whether</p>
<div class="targetlang">
<pre>
import foo
</pre>
</div>
<p>refers to a top-level module or to another module inside the current
package. In Python 3 it always refers to a top-level module (see <a href="https://www.python.org/dev/peps/pep-0328/">
PEP 328</a>). To instruct Python 2.5 through 2.7 to use new semantics
(that is <tt>import foo</tt> is interpreted as absolute import), one
has to put the following line</p>
<div class="targetlang">
<pre>
from __future__ import absolute_import
</pre>
</div>
<p>at the very beginning of your proxy <tt>*.py</tt> file. In SWIG, it
may be accomplished with <tt>%pythonbegin</tt> directive as follows:</p>
<div class="code">
<pre>
%pythonbegin %{
from __future__ import absolute_import
%}
</pre>
</div>
<h3><a name="Python_importfrominit">33.11.4 Importing from __init__.py</a>
</h3>
<p>Imports in <tt>__init__.py</tt> are handy when you want to populate a
package's namespace with names imported from other modules. In SWIG
based projects this approach may also be used to split large pieces of
code into smaller modules, compile them in parallel and then
re-assemble everything at runtime by importing submodules' contents in <tt>
__init__.py</tt>, for example.</p>
<p>Unfortunately import directives in <tt>__init__.py</tt> may cause
problems, especially if they refer to a package's submodules. This is
caused by the way Python initializes packages. If you spot problems
with imports from <tt>__init__.py</tt> try using <tt>-relativeimport</tt>
option. Below we explain in detail one issue, for which the <tt>
-relativeimport</tt> workaround may be helpful.</p>
<p>Consider the following example (Python 3):</p>
<div class="diagram">
<pre>
pkg1/__init__.py # (empty)
pkg1/pkg2/__init__.py # (imports something from bar.py)
pkg1/pkg2/foo.py
pkg1/pkg2/bar.py # (imports foo.py)
</pre>
</div>
<p>If the file contents are:</p>
<ul>
<li>
<p><tt>pkg1/pkg2/__init__.py:</tt></p>
<div class="targetlang">
<pre>
# pkg1/pkg2/__init__.py
from .bar import Bar
</pre>
</div></li>
<li>
<p><tt>pkg1/pkg2/foo.py:</tt></p>
<div class="targetlang">
<pre>
# pkg1/pkg2/foo.py
class Foo: pass
</pre>
</div></li>
<li>
<p><tt>pkg1/pkg2/bar.py:</tt></p>
<div class="targetlang">
<pre>
# pkg1/pkg2/bar.py
import pkg1.pkg2.foo
class Bar(pkg1.pkg2.foo.Foo): pass
</pre>
</div></li>
</ul>
<p>Now if one simply used <tt>import pkg1.pkg2</tt>, it will usually
fail:</p>
<div class="diagram">
<pre>
&gt;&gt;&gt; import pkg1.pkg2
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in &lt;module&gt;
File &quot;./pkg1/pkg2/__init__.py&quot;, line 2, in &lt;module&gt;
from .bar import Bar
File &quot;./pkg1/pkg2/bar.py&quot;, line 3, in &lt;module&gt;
class Bar(pkg1.pkg2.foo.Foo): pass
AttributeError: 'module' object has no attribute 'pkg2'
</pre>
</div>
<p>Surprisingly, if we execute the <tt>import pkg1.pkg2</tt> directive
for the second time, it succeeds. The reason seems to be following:
when Python spots the <tt>from .bar import Bar</tt> directive in <tt>
pkg1/pkg2/__init__.py</tt> it starts loading <tt>pkg1/pkg2/bar.py</tt>.
This module imports <tt>pkg1.pkg2.foo</tt> in turn and tries to use <tt>
pkg1.pkg2.foo.Foo</tt>, but the package <tt>pkg1</tt> is not fully
initialized yet (the initialization procedure is actually in progress)
and it seems like the effect of the already seen <tt>import
pkg1.pkg2.pkg3.foo</tt> is &quot;delayed&quot; or ignored. Exactly the same may
happen to a proxy module generated by SWIG.</p>
<p>One workaround for this case is to use a relative import in <tt>
pkg1/pkg2/bar.py</tt>. If we change <tt>bar.py</tt> to be:</p>
<div class="targetlang">
<pre>
from .pkg3 import foo
class Bar(foo.Foo): pass
</pre>
</div>
<p>or</p>
<div class="targetlang">
<pre>
from . import pkg3
from .pkg3 import foo
class Bar(pkg3.foo.Foo): pass
</pre>
</div>
<p>then the example works again. With SWIG, you need to enable the <tt>
-relativeimport</tt> option in order to have the above workaround in
effect (note, that the Python 2 case also needs the <tt>-relativeimport</tt>
workaround).</p>
<h3><a name="Python_implicit_namespace_packages">33.11.5 Implicit
namespace packages</a></h3>
<p> Python 3.3 introduced <a href="https://www.python.org/dev/peps/pep-0420/">
PEP 0420</a> which implements implicit namespace packages. In a
nutshell, implicit namespace packages remove the requirement of an
__init__.py file and allow packages to be split across multiple PATH
elements. For example:</p>
<div class="diagram">
<pre>
/fragment1/pkg1/mod1.py
/fragment2/pkg1/mod2.py
/fragment3/pkg1/mod3.py
</pre>
</div>
<p>If PYTHONPATH is set to &quot;/fragment1:/fragment2:/fragment3&quot;, then
mod1, mod2 and mod3 will be part of pkg1. This allows for splitting of
packages into separate pieces. This can be useful for SWIG generated
wrappers in the following way.</p>
<p> Suppose you create a SWIG wrapper for a module called robin. The
SWIG generated code consists of two files robin.py and _robin.so. You
wish to make these modules part of a subpackage (brave.sir). With
implicit namespace packages you can place these files in the following
configurations:</p>
<p>Using PYTHONPATH=&quot;/some/path&quot;</p>
<div class="diagram">
<pre>
/some/path/brave/sir/robin.py
/some/path/brave/sir/_robin.so
</pre>
</div>
<p>Using PYTHONPATH=&quot;/some/path:/some/other/path&quot;<div class="diagram">
<pre>
/some/path/brave/sir/robin.py
/some/other/path/brave/sir/_robin.so
</pre>
</div></p>
<p> Finally suppose that your pure Python code is stored in a .zip file
or some other way (database, web service connection, etc). Python can
load the robin.py module using a custom importer. But the _robin.so
module will need to be located on a file system. Implicit namespace
packages make this possible. For example, using
PYTHONPATH=&quot;/some/path/foo.zip:/some/other/path&quot;</p>
<p> Contents of foo.zip</p>
<div class="diagram">
<pre>
brave/
brave/sir/
brave/sir/robin.py
</pre>
</div>
<p> File system contents</p>
<div class="diagram">
<pre>
/some/other/path/brave/sir/_robin.so
</pre>
</div>
<p>Support for implicit namespace packages was added to python-3.3. The
zipimporter requires python-3.5.1 or newer to work with subpackages.</p>
<p><b> Compatibility Note:</b> Support for implicit namespace packages
was added in SWIG-3.0.9.</p>
<h3><a name="Python_package_search">33.11.6 Location of modules</a></h3>
<p> When SWIG creates wrappers from an interface file, say foo.i, two
Python modules are created. There is a pure Python module (foo.py) and
C/C++ code which is compiled and linked into a dynamically (or
statically) loaded low-level module _foo (see the <a href="#Python_nn3">
Preliminaries section</a> for details). So, the interface file really
defines two Python modules. How these two modules are loaded is covered
next.</p>
<p> The pure Python module needs to load the C/C++ module in order to
call the wrapped C/C++ methods. To do this it must make some
assumptions about the location of the C/C++ module. There are two
configurations that are supported by default.</p>
<ol>
<li>
<p>Both modules in the same package</p>
</li>
<li>
<p>Both modules are global</p>
</li>
</ol>
<p> Additional configurations are supported but require custom import
code.</p>
<p> The following sub-sections look more closely at the two default
configurations as well as some customized configurations. An input
interface file, foo.i, results in the two modules foo.py and _foo.so
for each of the configurations.</p>
<h4><a name="Python_package_search_both_package_modules">33.11.6.1 Both
modules in the same package</a></h4>
<p> In this configuration, the pure Python module, foo.py, tries to load
the C/C++ module, _foo, from the same package foo.py is located in. The
package name is determined from the <tt>__package__</tt> attribute if
available, see <a href="https://www.python.org/dev/peps/pep-0366/">PEP
366</a>. Otherwise it is derived from the <tt>__name__</tt> attribute
given to foo.py by the Python loader that imported foo.py. The
interface file for this configuration would contain:</p>
<div class="code">
<pre>
%module(package=&quot;mypackage&quot;) foo
</pre>
</div>
<p>The location of the files could be as follows:</p>
<div class="diagram">
<pre>
/dir/mypackage/foo.py
/dir/mypackage/__init__.py
/dir/mypackage/_foo.so
</pre>
</div>
<p>Assuming /dir/ is in PYTHONPATH, the module can be imported using</p>
<div class="targetlang">
<pre>
from mypackage import foo
</pre>
</div>
<h4><a name="Python_package_search_both_global_modules">33.11.6.2 Both
modules are global</a></h4>
<p> In this configuration, there are no packages. If foo.py is not in a
package, that is, it is a global module, then _foo is loaded as a
global module. The interface file for this configuration would contain:</p>
<div class="code">
<pre>
%module foo
</pre>
</div>
<p>The location of the files could be as follows:</p>
<div class="diagram">
<pre>
/dir/foo.py
/dir/_foo.so
</pre>
</div>
<p>Assuming /dir/ is in PYTHONPATH, the module can be imported using</p>
<div class="targetlang">
<pre>
import foo
</pre>
</div>
<h4><a name="Python_package_search_wrapper_split">33.11.6.3 Split
modules custom configuration</a></h4>
<p>In this non-standard 'split module' configuration, the pure Python
module is in a package and the low level C/C++ module is global. This
configuration is not generally recommended and is not supported by
default as it needs a custom configuration. The module import code
customization required is via the <tt>moduleimport</tt> attribute in
the <tt>%module</tt> directive. The next sub-section elaborates further
on this. The interface file for this split module configuration would
contain:</p>
<div class="code">
<pre>
%module(package=&quot;mypackage&quot;, moduleimport=&quot;import _foo&quot;) foo
</pre>
</div>
<p> When using <tt>-builtin</tt>, use the following instead (the reasons
are also covered shortly in the next sub-section):</p>
<div class="code">
<pre>
%module(package=&quot;mypackage&quot;, moduleimport=&quot;from _foo import *&quot;) foo
</pre>
</div>
<p>The location of the files could be as follows:</p>
<div class="diagram">
<pre>
/dir/mypackage/foo.py
/dir/mypackage/__init__.py
/dir/_foo.so
</pre>
</div>
<p>Assuming /dir/ is in PYTHONPATH, the module can be imported using</p>
<div class="targetlang">
<pre>
from mypackage import foo
</pre>
</div>
<p><b> Compatibility Note:</b> Versions of SWIG prior to SWIG-4.0.0
supported split modules without the above customization. However, this
had to be removed as the default import code often led to confusion due
to obfuscation of genuine Python <tt>ImportError</tt> problems. Using
one of the two default configurations is the recommended approach now.</p>
<h4><a name="Python_custom_module_import">33.11.6.4 More on customizing
the module import code</a></h4>
<p> The Python code implementing the default import logic is shown
below. It supports the two configurations described earlier, that is,
either both modules are in a package or loading both as global modules.
The code is generated into the pure Python module, foo.py, and merely
imports the low-level _foo module.</p>
<div class="targetlang">
<pre>
if __package__ or '.' in __name__:
from . import _foo
else:
import _foo
</pre>
</div>
<p> This import code implementation is non-trivial but it can be
replaced with custom code providing opportunities to make it simpler
and/or more flexible. This is not normally recommended though unless
you have a good understanding of the intricacies of importing Python
modules. The custom code can be specified by setting the <tt>
moduleimport</tt> option of the <tt>%module</tt> directive with the
appropriate import code. For example:</p>
<div class="code">
<pre>
%module(moduleimport=&quot;import _foo&quot;) foo
</pre>
</div>
<p> This will replace the default import logic above and generate the
following into the pure Python module, foo.py:</p>
<div class="targetlang">
<pre>
import _foo
</pre>
</div>
<p> In fact the above is a simplification customization for the
configuration where both modules are global; it removes the logic for
also handling the modules being in a package.</p>
<p> There is a special variable, <tt>$module</tt>, which is expanded
into the low-level C/C++ module name, <tt>_foo</tt> in the case above.
The identical output would be generated if instead the following had
been used:</p>
<div class="code">
<pre>
%module(moduleimport=&quot;import $module&quot;) foo
</pre>
</div>
<p> When you have many lines you can retain the easy readability of the <tt>
%module</tt> directive by using a macro. For example:</p>
<div class="code">
<pre>
%define MODULEIMPORT
&quot;
print('Loading low-level module $module')
import $module
print('Module has loaded')
&quot;
%enddef
%module(moduleimport=MODULEIMPORT) foo
</pre>
</div>
<p> This will of course generate the following into the pure Python
module:</p>
<div class="targetlang">
<pre>
print('Loading low-level module $module')
import _foo
print('Module has loaded')
</pre>
</div>
<p> When using the <tt>-builtin</tt> option, the link between the pure
Python module and the low-level C/C++ module is slightly different as
all the objects from the low-level module are imported directly into
the pure Python module. The default import loading code is thus
different:</p>
<div class="targetlang">
<pre>
if __package__ or '.' in __name__:
from ._foo import *
else:
from _foo import *
</pre>
</div>
<p> Any customizations must import the code in a similar manner. The
best way to support both with and without <tt>-builtin</tt> is to make
use of the <tt>SWIGPYTHON_BUILTIN</tt> macro which is defined when <tt>
-builtin</tt> is specified. The following will do this for the <a href="#Python_package_search_wrapper_split">
split modules</a> case above.</p>
<div class="code">
<pre>
#if defined(SWIGPYTHON_BUILTIN) /* defined when using -builtin */
%module(package=&quot;mypackage&quot;, moduleimport=&quot;from $module import *&quot;) foo
#else
%module(package=&quot;mypackage&quot;, moduleimport=&quot;import $module&quot;) foo
#endif
</pre>
</div>
<h4><a name="Python_package_search_static">33.11.6.5 Statically linked C
modules</a></h4>
<p>It is strongly recommended to use dynamically linked modules for the
C portion of your pair of Python modules. If for some reason you still
need to link the C module of the pair of Python modules generated by
SWIG into your interpreter, then this section provides some details on
how this impacts the pure Python modules ability to locate the other
part of the pair. Please also see the <a href="#Python_nn8">Static
Linking</a> section.</p>
<p>When Python is extended with C code the Python interpreter needs to
be informed about details of the new C functions that have been linked
into the executable. The code to do this is created by SWIG and is
automatically called in the correct way when the module is dynamically
loaded. However when the code is not dynamically loaded (because it is
statically linked) Then the initialization method for the module
created by SWIG is not called automatically and the Python interpreter
has no idea that the new SWIG C module exists.</p>
<p>Before Python 3, one could simply call the init method created by
SWIG which would have normally been called when the shared object was
dynamically loaded. The specific name of this method is not given here
because statically linked modules are not encouraged with SWIG (<a href="#Python_nn8">
Static Linking</a>). However one can find this init function in the C
file generated by SWIG.</p>
<p>If you are really keen on static linking there are two ways to
initialize the SWIG generated C module with the init method. Which way
you use depends on what version of Python your module is being linked
with. Python 2 and Python 3 treat this init function differently. And
the way they treat it affects how the pure Python module will be able
to locate the C module.</p>
<p>The details concerning this are covered completely in the
documentation for Python itself. Links to the relevant sections follow:</p>
<ul>
<li><a href="https://docs.python.org/2/extending/extending.html#methodtable">
Extending in python2</a></li>
<li><a href="https://docs.python.org/3.6/extending/extending.html#the-module-s-method-table-and-initialization-function">
Extending in python3</a></li>
</ul>
<p>There are two keys things to understand. The first is that in Python
2 the init() function returns void. In Python 3 the init() function
returns a <tt>PyObject *</tt> which points to the new module. Secondly,
when you call the init() method manually, you are the Python importer.
So, you determine which package the C module will be located in.</p>
<p>So, if you are using Python 3 it is important that you follow what is
described in the Python documentation linked above. In particular, you
can't simply call the init() function generated by SWIG and cast the <tt>
PyObject</tt> pointer it returns over the side. If you do then Python 3
will have no idea that your C module exists and the pure Python half of
your wrapper will not be able to find it. You need to register your
module with the Python interpreter as described in the Python docs.</p>
<p>With Python 2 things are somewhat more simple. In this case the
init() function returns void. Calling it will register your new C
module as a<b> global</b> module. The pure Python part of the SWIG
wrapper will be able to find it because it tries both the pure Python
module it is part of and the global module. If you wish not to have the
statically linked module be a global module then you will either need
to refer to the Python documentation on how to do this (remember you
are now the Python importer) or use dynamic linking.</p>
<h2><a name="Python_python3support">33.12 Python 3 Support</a></h2>
<p> SWIG is able to support Python 3.x. The wrapper code generated by
SWIG can be compiled with both Python 2.x or 3.x.</p>
<p> The list of known-to-be-broken features around Python 3 are:</p>
<ul>
<li>No more support for FILE* typemaps, because PyFile_AsFile has been
dropped in Python 3.</li>
<li>The <tt>-apply</tt> command line option is removed and generating
code using apply() is no longer supported.</li>
</ul>
<p> The following are Python 3 new features that are currently supported
by SWIG.</p>
<h3><a name="Python_annotations">33.12.1 Python function annotations and
variable annotations</a></h3>
<p> Python 3 supports function annotations as defined in <a href="https://www.python.org/dev/peps/pep-3107/">
PEP 3107</a>. Python 3.6 and later additionally support variable
annotations as defined in <a href="https://www.python.org/dev/peps/pep-526/">
PEP 526</a>. Note that currently there is no annotations support in SWIG
for the <tt>-builtin</tt> nor the <tt>-fastproxy</tt> option.
Annotations are added via the <tt>python:annotations</tt> <a href="#Customization_features">
%feature directives</a>. SWIG currently supports one type of function
annotation.</p>
<h4><a name="Python_annotations_c">33.12.1.1 C/C++ annotation types</a></h4>
<p> The <tt>%feature(&quot;python:annotations&quot;, &quot;c&quot;)</tt> directive generates
annotations containing C/C++ types. For example:</p>
<div class="code">
<pre>
%feature(&quot;python:annotations&quot;, &quot;c&quot;) global_ints;
int *global_ints(int &amp;ri);
</pre>
</div>
<p> The generated code then contains function annotations containing the
C++ types:</p>
<div class="targetlang">
<pre>
def global_ints(ri: &quot;int &amp;&quot;) -&gt; &quot;int *&quot;:
return _example.global_ints(ri)
</pre>
</div>
<p> There are some limitations with function annotations support, for
example, overloaded functions use <tt>*args</tt> or <tt>**kwargs</tt>
when keyword arguments are enabled. The parameter names and types are
then not shown. For example, with input:</p>
<div class="code">
<pre>
int *global_overloaded(int &amp;ri);
int *global_overloaded();
</pre>
</div>
<p> The generated Python function including annotations is shown below.
Only the return type is annotated.</p>
<div class="targetlang">
<pre>
def global_overloaded(*args) -&gt; &quot;int *&quot;:
return _example.global_overloaded(*args)
</pre>
</div>
<p> Below is an example demonstrating variable annotations.</p>
<div class="code">
<pre>
%feature(&quot;python:annotations&quot;, &quot;c&quot;);
struct V {
float val;
};
</pre>
</div>
<p> The generated code contains a variable annotation containing the C <tt>
float</tt> type:</p>
<div class="targetlang">
<pre>
class V(object):
val: &quot;float&quot; = property(_example.V_val_get, _example.V_val_set)
...
</pre>
</div>
<p> Variable annotations are only supported from Python 3.6. If you need
to support earlier versions of Python, you'll need to turn variable
annotations off via the <tt>python:annotations:novar</tt> feature flag.
It is quite easy to support function annotations but turn off variable
annotations. The next example shows how to do this for all variables.</p>
<div class="code">
<pre>
%feature(&quot;python:annotations&quot;, &quot;c&quot;); // Turn on function annotations and variable annotations globally
%feature(&quot;python:annotations:novar&quot;); // Turn off variable annotations globally
struct V {
float val;
void vv(float *v) const;
};
</pre>
</div>
<p> The resulting code will work with versions older than Python 3.6 as
the variable annotations are turned off:</p>
<div class="targetlang">
<pre>
class V(object):
val = property(_example.V_val_get, _example.V_val_set)
def vv(self, v: &quot;float *&quot;) -&gt; &quot;void&quot;:
return _example.V_vv(self, v)
...
</pre>
</div>
<p><b> Compatibility Note:</b> SWIG-4.1.0 changed the way that function
annotations are generated. Prior versions required the (now removed) <tt>
-py3</tt> option to generate function annotation support containing
C/C++ types instead of supporting <tt>%feature(&quot;python:annotations&quot;,
&quot;c&quot;)</tt>. Variable annotations were also added in SWIG-4.1.0.</p>
<h3><a name="Python_nn75">33.12.2 Buffer interface</a></h3>
<p> SWIG has a series of typemaps to support buffer interfaces. These
typemap macros are defined in <tt>pybuffer.i</tt>, which must be
included in order to use them. By using these typemaps, your wrapped
function will be able to accept any Python object that exposes a
suitable buffer interface.</p>
<p> For example, the <tt>get_path()</tt> function puts the path string
into the memory pointed to by its argument:</p>
<div class="code">
<pre>
void get_path(char *s);
</pre>
</div>
<p> Then you can write a typemap like this: (the following example is
applied to both Python 2 and 3, since the <tt>bytearray</tt> type was
backported to 2.6.</p>
<div class="code">
<pre>
%include &lt;pybuffer.i&gt;
%pybuffer_mutable_string(char *str);
void get_path(char *str);
</pre>
</div>
<p> And then on the Python side the wrapped <tt>get_path</tt> could be
used in this way:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; p = bytearray(10)
&gt;&gt;&gt; get_path(p)
&gt;&gt;&gt; print(p)
bytearray(b'/Foo/Bar/\x00')
</pre>
</div>
<p> The macros defined in <tt>pybuffer.i</tt> are similar to those in <tt>
cstring.i</tt>:</p>
<p><b> %pybuffer_mutable_binary(parm, size_parm)</b></p>
<div class="indent">
<p> The macro can be used to generate a typemap which maps a buffer of
an object to a pointer provided by <tt>parm</tt> and a size argument
provided by <tt>size_parm</tt>. For example:</p>
<div class="code">
<pre>
%pybuffer_mutable_binary(char *str, size_t size);
...
int snprintf(char *str, size_t size, const char *format, ...);
</pre>
</div>
<p> In Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; buf = bytearray(6)
&gt;&gt;&gt; snprintf(buf, &quot;Hello world!&quot;)
&gt;&gt;&gt; print(buf)
bytearray(b'Hello\x00')
&gt;&gt;&gt;
</pre>
</div></div>
<p><b> %pybuffer_mutable_string(parm)</b></p>
<div class="indent">
<p> This typemap macro requires the buffer to be a zero terminated
string, and maps the pointer of the buffer to <tt>parm</tt>. For
example:</p>
<div class="code">
<pre>
%pybuffer_mutable_string(char *str);
...
size_t make_upper(char *str);
</pre>
</div>
<p> In Python:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; buf = bytearray(b'foo\x00')
&gt;&gt;&gt; make_upper(buf)
&gt;&gt;&gt; print(buf)
bytearray(b'FOO\x00')
&gt;&gt;&gt;
</pre>
</div>
<p> Both <tt>%pybuffer_mutable_binary</tt> and <tt>
%pybuffer_mutable_string</tt> require the provided buffer to be mutable,
eg. they can accept a <tt>bytearray</tt> type but can't accept an
immutable <tt>byte</tt> type.</p>
</div>
<p><b> %pybuffer_binary(parm, size_parm)</b></p>
<div class="indent">
<p> This macro maps an object's buffer to a pointer <tt>parm</tt> and a
size <tt>size_parm</tt>. It is similar to <tt>%pybuffer_mutable_binary</tt>
, except the <tt>%pybuffer_binary</tt> can accept both mutable and
immutable buffers. As a result, the wrapped function should not modify
the buffer.</p>
</div>
<p><b> %pybuffer_string(parm)</b></p>
<div class="indent">
<p> This macro maps an object's buffer as a string pointer <tt>parm</tt>
. It is similar to <tt>%pybuffer_mutable_string</tt> but the buffer
could be both mutable and immutable. And your function should not
modify the buffer.</p>
</div>
<h3><a name="Python_nn76">33.12.3 Abstract base classes</a></h3>
<p> By including <tt>pyabc.i</tt> in your interface file, the proxy
classes of the STL containers will automatically gain an appropriate
abstract base class from the <tt>collections.abc</tt> module for Python
3.3 and later, otherwise from the <tt>collections</tt> module. For
example, the following SWIG interface:</p>
<div class="code">
<pre>
%include &lt;pyabc.i&gt;
%include &lt;std_map.i&gt;
%include &lt;std_list.i&gt;
namespace std {
%template(Mapii) map&lt;int, int&gt;;
%template(IntList) list&lt;int&gt;;
}
</pre>
</div>
<p> will generate a Python proxy class <tt>Mapii</tt> inheriting from <tt>
collections.abc.MutableMap</tt> for Python 3.3 and later, or <tt>
collections.MutableMap</tt> for earlier versions and a proxy class <tt>
IntList</tt> inheriting from <tt>collections.abc.MutableSequence</tt>
for Python 3.3 or later, or <tt>collections.MutableSequence</tt> for
earlier versions.</p>
<p> <tt>pyabc.i</tt> also provides a macro <tt>%pythonabc</tt> that
could be used to define an abstract base class for your own C++ class:</p>
<div class="code">
<pre>
%pythonabc(MySet, collections.abc.MutableSet); # Python 3.3 and later
%pythonabc(MySet, collections.MutableSet); # Prior to Python 3.3
%pythonabc(MySet, &quot;collections.abc.MutableSet if _swig_python_version_info &gt;= (3, 3) else collections.MutableSet&quot;); # All Python versions
</pre>
</div>
<p> For details of abstract base class, please see <a href="https://www.python.org/dev/peps/pep-3119/">
PEP 3119</a>.</p>
<p><b> Compatibility Note:</b> SWIG-4.0.0 changed the base classes to
use the <tt>collections.abc</tt> module instead of <tt>collections</tt>
due to the deprecation of the classes in the <tt>collections</tt>
module in Python 3.7. The <tt>collections.abc</tt> module was
introduced in Python 3.3 and hence this feature requires Python 3.3 or
later. SWIG-4.1.0 introduced the flexibility of using either the <tt>
collections.abc</tt> module for Python 3.3 and later or the <tt>
collections</tt> module for earlier Python versions.</p>
<h3><a name="Python_nn77">33.12.4 Byte string output conversion</a></h3>
<p> By default, any byte string (<tt>char*</tt> or <tt>std::string</tt>)
returned from C or C++ code is decoded to text as UTF-8. This decoding
uses the <tt>surrogateescape</tt> error handler under Python 3.1 or
higher -- this error handler decodes invalid byte sequences to high
surrogate characters in the range U+DC80 to U+DCFF. As an example,
consider the following SWIG interface, which exposes a byte string that
cannot be completely decoded as UTF-8:</p>
<div class="code">
<pre>
%module example
%inline %{
const char * non_utf8_c_str(void) {
return &quot;h\xe9llo w\xc3\xb6rld&quot;;
}
void instring(const char *s) {
...
}
%}
</pre>
</div>
<p> Note that &quot;\xe9&quot; is an invalid UTF-8 encoding, but &quot;\xc3\xb6&quot; is
valid. When this method is called from Python 3, the return value is
the following text string:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; s = example.non_utf8_c_str()
&gt;&gt;&gt; s
'h\udce9llo w&ouml;rld'
</pre>
</div>
<p> Since the C string contains bytes that cannot be decoded as UTF-8,
those raw bytes are represented as high surrogate characters that can
be used to obtain the original byte sequence:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; b = s.encode('utf-8', errors='surrogateescape')
&gt;&gt;&gt; b
b'h\xe9llo w\xc3\xb6rld'
</pre>
</div>
<p> One can then attempt a different encoding, if desired (or simply
leave the byte string as a raw sequence of bytes for use in binary
protocols):</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; b.decode('latin-1')
'h&eacute;llo w&Atilde;&para;rld'
</pre>
</div>
<p> Note, however, that text strings containing surrogate characters are
rejected with the default <tt>strict</tt> codec error handler. For
example:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; with open('test', 'w') as f:
... print(s, file=f)
...
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 2, in &lt;module&gt;
UnicodeEncodeError: 'utf-8' codec can't encode character '\udce9' in position 1: surrogates not allowed
</pre>
</div>
<p> This requires the user to check most strings returned by SWIG
bindings, but the alternative is for a non-UTF8 byte string to be
completely inaccessible in Python 3 code.</p>
<p> For more details about the <tt>surrogateescape</tt> error handler,
please see <a href="https://www.python.org/dev/peps/pep-0383/">PEP 383</a>
.</p>
<p> When Python 3 strings are passed to the C/C++ layer, they are
expected to be valid UTF8 Unicode strings too. For example, when the <tt>
instring</tt> method above is wrapped and called, any invalid UTF8
Unicode code strings will result in a TypeError because the attempted
conversion fails:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; example.instring('h\xe9llo')
&gt;&gt;&gt; example.instring('h\udce9llo')
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in &lt;module&gt;
TypeError: in method 'instring', argument 1 of type 'char const *'
</pre>
</div>
<p> In some cases, users may wish to instead handle all byte strings as
bytes objects in Python 3. This can be accomplished by adding <tt>
SWIG_PYTHON_STRICT_BYTE_CHAR</tt> to the generated code:</p>
<div class="code">
<pre>
%module char_to_bytes
%begin %{
#define SWIG_PYTHON_STRICT_BYTE_CHAR
%}
char *charstring(char *s) {
return s;
}
</pre>
</div>
<p> This will modify the behavior so that only Python 3 bytes objects
will be accepted and converted to a C/C++ string, and any string
returned from C/C++ will be converted to a bytes object in Python 3:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; from char_to_bytes import *
&gt;&gt;&gt; charstring(b&quot;hi&quot;) # Byte string
b'hi'
&gt;&gt;&gt; charstring(&quot;hi&quot;) # Unicode string
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
TypeError: in method 'charstring', argument 1 of type 'char *'
</pre>
</div>
<p> Note that in Python 2, defining <tt>SWIG_PYTHON_STRICT_BYTE_CHAR</tt>
has no effect, since strings in Python 2 are equivalent to Python 3
bytes objects. However, there is a similar capability to force
unicode-only handling for wide characters C/C++ strings (<tt>wchar_t *</tt>
or <tt>std::wstring</tt> types) in Python 2. By default, in Python 2
both strings and unicode strings are converted to C/C++ wide strings,
and returned wide strings are converted to a Python unicode string. To
instead only convert unicode strings to wide strings, users can add <tt>
SWIG_PYTHON_STRICT_UNICODE_WCHAR</tt> to the generated code:</p>
<div class="code">
<pre>
%module wchar_to_unicode
%begin %{
#define SWIG_PYTHON_STRICT_UNICODE_WCHAR
%}
wchar_t *wcharstring(wchar_t *s) {
return s;
}
</pre>
</div>
<p> This ensures that only unicode strings are accepted by wcharstring
in both Python 2 and Python 3:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; from wchar_to_unicode import *
&gt;&gt;&gt; wcharstring(u&quot;hi&quot;) # Unicode string
u'hi'
&gt;&gt;&gt; wcharstring(b&quot;hi&quot;) # Byte string
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
TypeError: in method 'charstring', argument 1 of type 'wchar_t *'
</pre>
</div>
<p> By defining both <tt>SWIG_PYTHON_STRICT_BYTE_CHAR</tt> and <tt>
SWIG_PYTHON_STRICT_UNICODE_WCHAR</tt>, Python wrapper code can support
overloads taking both std::string (as Python bytes) and std::wstring
(as Python unicode).</p>
<h3><a name="Python_2_unicode">33.12.5 Python 2 Unicode</a></h3>
<p> A Python 3 string is a Unicode string so by default a Python 3
string that contains Unicode characters passed to C/C++ will be
accepted and converted to a C/C++ string (<tt>char *</tt> or <tt>
std::string</tt> types). A Python 2 string is not a unicode string by
default and should a Unicode string be passed to C/C++ it will fail to
convert to a C/C++ string (<tt>char *</tt> or <tt>std::string</tt>
types). The Python 2 behavior can be made more like Python 3 by
defining <tt>SWIG_PYTHON_2_UNICODE</tt> when compiling the generated
C/C++ code. By default when the following is wrapped:</p>
<div class="code">
<pre>
%module unicode_strings
char *charstring(char *s) {
return s;
}
</pre>
</div>
<p> An error will occur when using Unicode strings in Python 2:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; from unicode_strings import *
&gt;&gt;&gt; charstring(&quot;hi&quot;)
'hi'
&gt;&gt;&gt; charstring(u&quot;hi&quot;)
Traceback (most recent call last):
File &quot;&lt;stdin&gt;&quot;, line 1, in ?
TypeError: in method 'charstring', argument 1 of type 'char *'
</pre>
</div>
<p> When the <tt>SWIG_PYTHON_2_UNICODE</tt> macro is added to the
generated code:</p>
<div class="code">
<pre>
%module unicode_strings
%begin %{
#define SWIG_PYTHON_2_UNICODE
%}
char *charstring(char *s) {
return s;
}
</pre>
</div>
<p> Unicode strings will be successfully accepted and converted from
UTF-8, but note that they are returned as a normal Python 2 string:</p>
<div class="targetlang">
<pre>
&gt;&gt;&gt; from unicode_strings import *
&gt;&gt;&gt; charstring(&quot;hi&quot;)
'hi'
&gt;&gt;&gt; charstring(u&quot;hi&quot;)
'hi'
&gt;&gt;&gt;
</pre>
</div>
<p> Note that defining both <tt>SWIG_PYTHON_2_UNICODE</tt> and <tt>
SWIG_PYTHON_STRICT_BYTE_CHAR</tt> at the same time is not allowed, since
the first is allowing unicode conversion and the second is explicitly
prohibiting it.</p>
<h2><a name="Python_multithreaded">33.13 Support for Multithreaded
Applications</a></h2>
<p>By default, SWIG does not enable support for multithreaded Python
applications. More specifically, the Python wrappers generated by SWIG
will not release the Python's interpreter's Global Interpreter Lock
(GIL) when wrapped C/C++ code is entered. Hence, while any of the
wrapped C/C++ code is executing, the Python interpreter will not be
able to run any other threads, even if the wrapped C/C++ code is
waiting in a blocking call for something like network or disk IO.
Fortunately, SWIG does have the ability to enable multithreaded support
and automatic release of the GIL either for all wrapped code in a
module or on a more selective basis. The user interface for this is
described in the next section.</p>
<h3><a name="Python_thread_UI">33.13.1 UI for Enabling Multithreading
Support</a></h3>
<p>The user interface is as follows:</p>
<ol>
<li>
<p>Module thread support can be enabled in two ways:</p>
<ul>
<li>
<p> The <tt>-threads</tt> SWIG Python option at the command line (or in <tt>
setup.py</tt>):</p>
<div class="shell">
<pre>$ swig -python -threads example.i</pre>
</div></li>
<li>
<p> The <tt>threads</tt> module option in the *.i template file:</p>
<div class="code">
<pre>%module(threads=&quot;1&quot;)</pre>
</div></li>
</ul>
</li>
<li>
<p>You can disable thread support for a given method:</p>
<div class="code">
<pre>%feature(&quot;nothread&quot;) method;</pre>
</div> or<div class="code">
<pre>%nothread method;</pre>
</div></li>
<li>
<p>You can partially disable thread support for a given method:</p>
<ul>
<li>
<p>To disable the C++/Python thread protection:</p>
<div class="code">
<pre>%feature(&quot;nothreadblock&quot;) method;</pre>
</div> or<div class="code">
<pre>%nothreadblock method;</pre>
</div></li>
<li>
<p>To disable the Python/C++ thread protection</p>
<div class="code">
<pre>%feature(&quot;nothreadallow&quot;) method;</pre>
</div> or<div class="code">
<pre>%nothreadallow method;</pre>
</div></li>
</ul>
</li>
</ol>
<h3><a name="Python_thread_performance">33.13.2 Multithread Performance</a>
</h3>
<p> For the curious about performance, here are some numbers for the
profiletest.i test, which is used to check the speed of the wrapped
code:</p>
<table summary="Python multithread performance">
<tr><th>Thread Mode</th><th>Execution Time (sec)</th><th>Comment</th></tr>
<tr><td>Single Threaded</td><td>9.6</td><td>no &quot;-threads&quot; option given</td>
</tr>
<tr><td>Fully Multithreaded</td><td>15.5</td><td>&quot;-threads&quot; option =
'allow' + 'block'</td></tr>
<tr><td>No Thread block</td><td>12.2</td><td>only 'allow'</td></tr>
<tr><td>No Thread Allow</td><td>13.6</td><td>only block'</td></tr>
</table>
<p> Fully threaded code decreases the wrapping performance by around
60%. If that is important to your application, you can tune each method
using the different 'nothread', 'nothreadblock' or 'nothreadallow'
features as needed. Note that for some methods deactivating the 'thread
block' or 'thread allow' code is not an option, so, be careful.</p>
<!-- LocalWords: polymorphism Typemaps STL typemap typemaps Docstring autodoc
-->
<!-- LocalWords: docstring SWIG's cxx py GCC linux DLL gcc fPIC Wiki Xlinker
-->
<!-- LocalWords: examplemodule DHAVE CONFIG lpython lm ldl mypython lsocket
-->
<!-- LocalWords: lnsl lpthread distutils enums namespaces
-->
<HR NOSHADE>
<h1><a name="R">34 SWIG and R</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#R_nn2">Bugs</a></li>
<li><a href="#R_nn3">Using R and SWIG</a></li>
<li><a href="#R_nn4">Precompiling large R files</a></li>
<li><a href="#R_nn5">General policy</a></li>
<li><a href="#R_language_conventions">Language conventions</a></li>
<li><a href="#R_nn6">C++ classes</a>
<ul>
<li><a href="#R_class_examples">Examples</a></li>
</ul>
</li>
<li><a href="#R_nn7">Enumerations</a></li>
</ul>
</div>
<!-- INDEX -->
<p> R is a GPL'ed open source statistical and plotting environment.
Information about R can be found at <a href="https://www.r-project.org/">
www.r-project.org</a>.</p>
<p> The R bindings are under active development. They have been used to
compile and run an R interface to QuantLib running on Mandriva Linux
with gcc. They are also used to create the SimpleITK R package, which
runs on Linux and MacOS. SWIG is used to create all wrapper interfaces
to <a href="https://www.simpleitk.org/">SimpleITK</a>. The R bindings
also work on Microsoft Windows using Visual C++.</p>
<h2><a name="R_nn2">34.1 Bugs</a></h2>
<p> Currently the following features are not implemented or broken:</p>
<ul>
<li>Garbage collection of some created objects. Finalizers are available
for wrapped C++ classes and are called by the garbage collection
system.</li>
<li>C Array wrappings</li>
</ul>
<h2><a name="R_nn3">34.2 Using R and SWIG</a></h2>
<p> To use R and SWIG in C mode, execute the following commands where
example.c is the name of the file with the functions in them</p>
<div class="shell">
<pre>
swig -r example.i
R CMD SHLIB example_wrap.c example.c
</pre>
</div>
<p> The corresponding options for C++ mode are</p>
<div class="shell">
<pre>
swig -c++ -r -o example_wrap.cpp example.i
R CMD SHLIB example_wrap.cpp example.cpp
</pre>
</div>
<p> Note that R is sensitive to the names of the files. The name of the
wrapper file must be the name of the library unless you use the -o
option to R when building the library, for example:</p>
<div class="shell">
<pre>
swig -c++ -r -o example_wrap.cpp example.i
R CMD SHLIB -o example.so example_wrap.cpp example.cpp
</pre>
</div>
<p> R is also sensitive to the name of the file extension in C and C++
mode. In C++ mode, the file extension must be .cpp rather than .cxx for
the R compile command to recognize it. If your C++ code is in a file
using something other than a .cpp extension, then it may still work
using PKG_LIBS:</p>
<div class="shell">
<pre>
swig -c++ -r -o example_wrap.cpp example.i
PKG_LIBS=&quot;example.cxx&quot; R CMD SHLIB -o example example_wrap.cpp
</pre>
</div>
<p> The commands produces two files. A dynamic shared object file called
example.so, or example.dll, and an R wrapper file called example.R. To
load these files, start up R and type in the following commands</p>
<div class="shell">
<pre>
dyn.load(paste(&quot;example&quot;, .Platform$dynlib.ext, sep=&quot;&quot;))
source(&quot;example.R&quot;)
cacheMetaData(1)
</pre>
</div>
<p> The cacheMetaData(1) will cause R to refresh its object tables.
Without it, inheritance of wrapped objects may fail. These two files
can be loaded in any order.</p>
<p> If you are compiling code yourself (not using R itself), there are a
few things to watch out for:</p>
<ul>
<li>The output shared library name (to the left of the file extension)
MUST match the module name, or alternatively, you can also set the
-package NAME command line argument. See swig -r -help for more
information</li>
<li>If you do not set the output file name appropriately, you might see
errors like<div class="shell">
<pre>
&gt; fact(4)
Error in .Call(&quot;R_swig_fact&quot;, s_arg1, as.logical(.copy), PACKAGE = &quot;example&quot;) :
&quot;R_swig_fact&quot; not available for .Call() for package &quot;example&quot;
</pre>
</div></li>
<li>Make sure the architecture of the shared library(x64 for instance),
matches the architecture of the R program you want to load your shared
library into</li>
</ul>
<h2><a name="R_nn4">34.3 Precompiling large R files</a></h2>
<p> In cases where the R file is large, one make save a lot of loading
time by precompiling the R wrapper. This can be done by creating the
file makeRData.R which contains the following</p>
<div class="code">
<pre>
source('BigFile.R')
save(list=ls(all=TRUE), file=&quot;BigFile.RData&quot;, compress=TRUE)
q(save=&quot;no&quot;)
</pre>
</div>
<p> This will generate a compiled R file called BigFile.RData that will
save a large amount of loading time.</p>
<p> There is no need to precompile large R files if the SWIG-generated
code is being included in an R package. The package infrastructure
provides this service during package installation.</p>
<h2><a name="R_nn5">34.4 General policy</a></h2>
<p> The general policy of the module is to treat the C/C++ as a basic
wrapping over the underlying functions and rely on the R type system to
provide R syntax.</p>
<h2><a name="R_language_conventions">34.5 Language conventions</a></h2>
<p> getitem and setitem use C++ conventions (i.e. zero based indices).
[&lt;- and [ are overloaded to allow for R syntax (one based indices and
slices)</p>
<h2><a name="R_nn6">34.6 C++ classes</a></h2>
<p> Wrapping of C++ classes for R works quite well. R has a special
type, known as an external reference, that can be used as a pointer to
arbitrary things, including C++ classes. The proxy layers generated for
other classes are not required.</p>
<p> SWIG currently creates a custom hierarchy of R classes derived from
the external reference type and implements type checking and function
overloading in the R code it generates. In the future we hope to
utilise the built in R6 class structures.</p>
<p> The R interface has the following capabilities:</p>
<ul>
<li> Destructor methods are registered and called automatically by the R
garbage collector.</li>
<li> A range of std::vector types are converted automatically to R
equivalents via the std_vector.i library.</li>
<li> The $ operator is used for method access.</li>
<li> Variable accessors are automatically generated and called via the
$, [, [[, $&lt;-, [&lt;-, [[&lt;- operators.</li>
</ul>
<h3><a name="R_class_examples">34.6.1 Examples</a></h3>
<p> Consider the following simple example:</p>
<div class="code">
<pre>
class Vehicle {
private:
int m_axles;
public:
int Axles() {
return(m_axles);
}
bool Available;
Vehicle() {
Available=false;
m_axles=2;
}
Vehicle(int ax) {
Available=false;
m_axles=ax;
}
};
</pre>
</div>
<p> The following options are available in R:</p>
<div class="code">
<pre>
v1 &lt;- Vehicle()
v2 &lt;- Vehicle(4)
# access members
v1$Axles()
[1] 2
v2$Axles
[1] 4
v1$Available
[1] FALSE
# Set availability
v1$Available &lt;- TRUE
v1$Available
[1] TRUE
</pre>
</div>
<p> A useful trick to determine the methods that are available is to
query the R method definition as follows:</p>
<div class="code">
<pre>
# display the methods for the class
getMethod(&quot;$&quot;, class(v1))
Method Definition:
function (x, name)
{
accessorFuns = list(Axles = Vehicle_Axles, Available = Vehicle_Available_get)
vaccessors = c(&quot;Available&quot;)
idx = pmatch(name, names(accessorFuns))
if (is.na(idx))
return(callNextMethod(x, name))
f = accessorFuns[[idx]]
if (is.na(match(name, vaccessors)))
function(...) {
f(x, ...)
}
else f(x)
}
Signatures:
x
target &quot;_p_Vehicle&quot;
defined &quot;_p_Vehicle&quot;
</pre>
</div>
<p> The names in the <tt>accessorFuns</tt> list correspond to class
methods while names in the <tt>vaccessors</tt> section correspond to
variables that may be modified.</p>
<h2><a name="R_nn7">34.7 Enumerations</a></h2>
<p> R doesn't have a native enumeration type. Enumerations are
represented as character strings in R, with calls to R functions that
convert back and forth between integers.</p>
<p> The details of enumeration names and contents are stored in hidden R
environments, which are named according to the enumeration name - for
example, an enumeration colour:</p>
<div class="code">
<pre>
enum colour { red=-1, blue, green = 10 };
</pre>
</div>
<p> will be initialized by the following call in R:</p>
<div class="code">
<pre>
defineEnumeration(&quot;_colour&quot;,
.values=c(&quot;red&quot; = .Call('R_swig_colour_red_get',FALSE, PACKAGE='enum_thorough'),
&quot;blue&quot; = .Call('R_swig_colour_blue_get',FALSE, PACKAGE='enum_thorough'),
&quot;green&quot; = .Call('R_swig_colour_green_get',FALSE, PACKAGE='enum_thorough')))
</pre>
</div>
<p> which will create an environment named <tt>.__E___colour</tt>. The
enumeration values are initialised via calls to C/C++ code, allowing
complex values for enumerations to be used. Calls to the C/C++ code
require the compiled library to be loaded, so a <tt>delayedAssign</tt>
is employed within <tt>defineEnumeration</tt> in order to allow the
code to be easily used in R packages.</p>
<p> The user typically does not need to access the enumeration lookup
functions or know the name of the enumeration type used by R.
Attributes containing the type information are attached by swig to
functions requiring enumeration arguments or returning enumeration
values, and those attributes are used to identify and access the
appropriate environments and thus translate between characters and
integers.</p>
<p> The relevant functions, for debugging purposes, are <tt>
enumToInteger</tt> and <tt>enumFromInteger</tt>.</p>
<p> Anonymous enumerations are ignored by the binding generation
process, leaving no way of accessing the value of anonymous
enumerations from R code.</p>
<HR NOSHADE>
<h1><a name="Ruby">35 SWIG and Ruby</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Ruby_nn2">Preliminaries</a>
<ul>
<li><a href="#Ruby_nn3">Running SWIG</a></li>
<li><a href="#Ruby_nn4">Getting the right header files</a></li>
<li><a href="#Ruby_nn5">Compiling a dynamic module</a></li>
<li><a href="#Ruby_nn6">Using your module</a></li>
<li><a href="#Ruby_nn7">Static linking</a></li>
<li><a href="#Ruby_nn8">Compilation of C++ extensions</a></li>
</ul>
</li>
<li><a href="#Ruby_nn9">Building Ruby Extensions under Windows 95/NT</a>
<ul>
<li><a href="#Ruby_nn10">Running SWIG from Developer Studio</a></li>
</ul>
</li>
<li><a href="#Ruby_nn11">The Ruby-to-C/C++ Mapping</a>
<ul>
<li><a href="#Ruby_nn12">Modules</a></li>
<li><a href="#Ruby_nn13">Functions</a></li>
<li><a href="#Ruby_nn14">Variable Linking</a></li>
<li><a href="#Ruby_nn15">Constants</a></li>
<li><a href="#Ruby_nn16">Pointers</a></li>
<li><a href="#Ruby_nn17">Structures</a></li>
<li><a href="#Ruby_nn18">C++ classes</a></li>
<li><a href="#Ruby_nn19">C++ Inheritance</a></li>
<li><a href="#Ruby_nn20">C++ Overloaded Functions</a></li>
<li><a href="#Ruby_nn21">C++ Operators</a></li>
<li><a href="#Ruby_nn22">C++ namespaces</a></li>
<li><a href="#Ruby_nn23">C++ templates</a></li>
<li><a href="#Ruby_nn23_1">C++ Standard Template Library (STL)</a></li>
<li><a href="#Ruby_C_STL_Functors">C++ STL Functors</a></li>
<li><a href="#Ruby_C_Iterators">C++ STL Iterators</a></li>
<li><a href="#Ruby_nn24">C++ Smart Pointers</a>
<ul>
<li><a href="#Ruby_smart_pointers_shared_ptr">The shared_ptr Smart
Pointer</a></li>
<li><a href="#Ruby_smart_pointers_generic">Generic Smart Pointers</a></li>
</ul>
</li>
<li><a href="#Ruby_nn25">Cross-Language Polymorphism</a>
<ul>
<li><a href="#Ruby_nn26">Exception Unrolling</a></li>
</ul>
</li>
</ul>
</li>
<li><a href="#Ruby_nn27">Naming</a>
<ul>
<li><a href="#Ruby_nn28">Defining Aliases</a></li>
<li><a href="#Ruby_nn29">Predicate Methods</a></li>
<li><a href="#Ruby_nn30">Bang Methods</a></li>
<li><a href="#Ruby_nn31">Getters and Setters</a></li>
</ul>
</li>
<li><a href="#Ruby_nn32">Input and output parameters</a></li>
<li><a href="#Ruby_nn33">Exception handling</a>
<ul>
<li><a href="#Ruby_nn34">Using the %exception directive</a></li>
<li><a href="#Ruby_nn34_2">Handling Ruby Blocks</a></li>
<li><a href="#Ruby_nn35">Raising exceptions</a></li>
<li><a href="#Ruby_nn36">Exception classes</a></li>
</ul>
</li>
<li><a href="#Ruby_nn37">Typemaps</a>
<ul>
<li><a href="#Ruby_nn38">What is a typemap?</a></li>
<li><a href="#Ruby_Typemap_scope">Typemap scope</a></li>
<li><a href="#Ruby_Copying_a_typemap">Copying a typemap</a></li>
<li><a href="#Ruby_Deleting_a_typemap">Deleting a typemap</a></li>
<li><a href="#Ruby_Placement_of_typemaps">Placement of typemaps</a></li>
<li><a href="#Ruby_nn39">Ruby typemaps</a>
<ul>
<li><a href="#Ruby_in_typemap">&quot;in&quot; typemap</a></li>
<li><a href="#Ruby_typecheck_typemap">&quot;typecheck&quot; typemap</a></li>
<li><a href="#Ruby_out_typemap">&quot;out&quot; typemap</a></li>
<li><a href="#Ruby_arginit_typemap">&quot;arginit&quot; typemap</a></li>
<li><a href="#Ruby_default_typemap">&quot;default&quot; typemap</a></li>
<li><a href="#Ruby_check_typemap">&quot;check&quot; typemap</a></li>
<li><a href="#Ruby_argout_typemap_">&quot;argout&quot; typemap</a></li>
<li><a href="#Ruby_freearg_typemap_">&quot;freearg&quot; typemap</a></li>
<li><a href="#Ruby_newfree_typemap">&quot;newfree&quot; typemap</a></li>
<li><a href="#Ruby_memberin_typemap">&quot;memberin&quot; typemap</a></li>
<li><a href="#Ruby_varin_typemap">&quot;varin&quot; typemap</a></li>
<li><a href="#Ruby_varout_typemap_">&quot;varout&quot; typemap</a></li>
<li><a href="#Ruby_throws_typemap">&quot;throws&quot; typemap</a></li>
<li><a href="#Ruby_directorin_typemap">directorin typemap</a></li>
<li><a href="#Ruby_directorout_typemap">directorout typemap</a></li>
<li><a href="#Ruby_directorargout_typemap">directorargout typemap</a></li>
<li><a href="#Ruby_ret_typemap">ret typemap</a></li>
<li><a href="#Ruby_globalin_typemap">globalin typemap</a></li>
</ul>
</li>
<li><a href="#Ruby_nn40">Typemap variables</a></li>
<li><a href="#Ruby_nn41">Useful Functions</a>
<ul>
<li><a href="#Ruby_nn42">C Datatypes to Ruby Objects</a></li>
<li><a href="#Ruby_nn43">Ruby Objects to C Datatypes</a></li>
<li><a href="#Ruby_nn44">Macros for VALUE</a></li>
<li><a href="#Ruby_nn45">Exceptions</a></li>
<li><a href="#Ruby_nn46">Iterators</a></li>
</ul>
</li>
<li><a href="#Ruby_nn47">Typemap Examples</a></li>
<li><a href="#Ruby_nn48">Converting a Ruby array to a char **</a></li>
<li><a href="#Ruby_nn49">Collecting arguments in a hash</a></li>
<li><a href="#Ruby_nn50">Pointer handling</a>
<ul>
<li><a href="#Ruby_nn51">Ruby Datatype Wrapping</a></li>
</ul>
</li>
<li><a href="#Ruby_nn52">Example: STL Vector to Ruby Array</a></li>
</ul>
</li>
<li><a href="#Ruby_nn65">Docstring Features</a>
<ul>
<li><a href="#Ruby_nn66">Module docstring</a></li>
<li><a href="#Ruby_nn67">%feature(&quot;autodoc&quot;)</a>
<ul>
<li><a href="#Ruby_nn68">%feature(&quot;autodoc&quot;, &quot;0&quot;)</a></li>
<li><a href="#Ruby_autodoc1">%feature(&quot;autodoc&quot;, &quot;1&quot;)</a></li>
<li><a href="#Ruby_autodoc2">%feature(&quot;autodoc&quot;, &quot;2&quot;)</a></li>
<li><a href="#Ruby_feature_autodoc3">%feature(&quot;autodoc&quot;, &quot;3&quot;)</a></li>
<li><a href="#Ruby_nn70">%feature(&quot;autodoc&quot;, &quot;docstring&quot;)</a></li>
</ul>
</li>
<li><a href="#Ruby_nn71">%feature(&quot;docstring&quot;)</a></li>
</ul>
</li>
<li><a href="#Ruby_nn53">Advanced Topics</a>
<ul>
<li><a href="#Ruby_operator_overloading">Operator overloading</a></li>
<li><a href="#Ruby_nn55">Creating Multi-Module Packages</a></li>
<li><a href="#Ruby_nn56">Specifying Mixin Modules</a></li>
</ul>
</li>
<li><a href="#Ruby_nn57">Memory Management</a>
<ul>
<li><a href="#Ruby_nn58">Mark and Sweep Garbage Collector</a></li>
<li><a href="#Ruby_nn59">Object Ownership</a></li>
<li><a href="#Ruby_nn60">Object Tracking</a></li>
<li><a href="#Ruby_nn61">Mark Functions</a></li>
<li><a href="#Ruby_nn62">Free Functions</a></li>
<li><a href="#Ruby_nn63">Embedded Ruby and the C++ Stack</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p>This chapter describes SWIG's support of Ruby.</p>
<h2><a name="Ruby_nn2">35.1 Preliminaries</a></h2>
<p> SWIG 4.2 is known to work with Ruby versions 2.0 and later. Given
the choice, you should use the latest stable version of Ruby. You
should also determine if your system supports shared libraries and
dynamic loading. SWIG will work with or without dynamic loading, but
the compilation process will vary.</p>
<p>This chapter covers most SWIG features, but in less depth than is
found in earlier chapters. At the very least, make sure you also read
the &quot;<a href="#SWIG">SWIG Basics</a>&quot; chapter. It is also assumed that
the reader has a basic understanding of Ruby.</p>
<h3><a name="Ruby_nn3">35.1.1 Running SWIG</a></h3>
<p> To build a Ruby module, run SWIG using the <tt>-ruby</tt> option:</p>
<div class="code shell">
<pre>$ swig -ruby example.i
</pre>
</div>
<p> If building a C++ extension, add the <tt>-c++</tt> option:</p>
<div class="code shell">
<pre>$ swig -c++ -ruby example.i
</pre>
</div>
<p> This creates a file <tt>example_wrap.c</tt> (<tt>example_wrap.cxx</tt>
if compiling a C++ extension) that contains all of the code needed to
build a Ruby extension module. To finish building the module, you need
to compile this file and link it with the rest of your program.</p>
<h3><a name="Ruby_nn4">35.1.2 Getting the right header files</a></h3>
<p> In order to compile the wrapper code, the compiler needs the <tt>
ruby.h</tt> header file and its dependencies, notably <tt>ruby/config.h</tt>
which is found in a different, architecture-dependent, directory. The
best way to find the compiler options needed to compile the code is to
ask Ruby itself:</p>
<div class="code shell">
<pre>$ ruby -rrbconfig -e 'puts &quot;-I#{RbConfig::CONFIG[%q{rubyhdrdir}]} -I#{RbConfig::CONFIG[%q{rubyarchhdrdir}]}&quot;'
-I/usr/include/ruby-2.1.0 -I/usr/include/x86_64-linux-gnu/ruby-2.1.0
</pre>
</div>
<h3><a name="Ruby_nn5">35.1.3 Compiling a dynamic module</a></h3>
<p> Ruby extension modules are typically compiled into shared libraries
that the interpreter loads dynamically at runtime. Since the exact
commands for doing this vary from platform to platform, your best bet
is to follow the steps described in the <tt>README.EXT</tt> file from
the Ruby distribution:</p>
<ol>
<li>
<p>Create a file called <tt>extconf.rb</tt> that looks like the
following:</p>
<div class="code targetlang">
<pre>require 'mkmf'
create_makefile('example')</pre>
</div></li>
<li>
<p>Type the following to build the extension:</p>
<div class="code shell">
<pre>
$ ruby extconf.rb
$ make
$ make install
</pre>
</div></li>
</ol>
<p> Of course, there is the problem that mkmf does not work correctly on
all platforms, e.g, HPUX. If you need to add your own make rules to the
file that <tt>extconf.rb</tt> produces, you can add this:</p>
<div class="code targetlang">
<pre>open(&quot;Makefile&quot;, &quot;a&quot;) { |mf|
puts &lt;&lt;EOM
# Your make rules go here
EOM
}
</pre>
</div>
<p> to the end of the <tt>extconf.rb</tt> file. If for some reason you
don't want to use the standard approach, you'll need to determine the
correct compiler and linker flags for your build platform. For example,
assuming you have code you need to link to in a file called <tt>
example.c</tt>, a typical sequence of commands for the Linux operating
system would look something like this:</p>
<div class="code shell">
<pre>$ swig -ruby example.i
$ gcc -O2 -fPIC -c example.c
$ gcc -O2 -fPIC -c example_wrap.c -I/usr/include/ruby-2.1.0
$ gcc -shared example.o example_wrap.o -o example.so
</pre>
</div>
<p> The -fPIC option tells GCC to generate position-independent code
(PIC) which is required for most architectures (it's not vital on x86,
but still a good idea as it allows code pages from the library to be
shared between processes). Other compilers may need a different option
specified instead of -fPIC.</p>
<p> If in doubt, consult the manual pages for your compiler and linker
to determine the correct set of options. You might also check the <a href="https://github.com/swig/swig/wiki">
SWIG Wiki</a> for additional information.</p>
<h3><a name="Ruby_nn6">35.1.4 Using your module</a></h3>
<p> Ruby<i> module</i> names must be capitalized, but the convention for
Ruby<i> feature</i> names is to use lowercase names. So, for example,
the<b> Etc</b> extension module is imported by requiring the<b> etc</b>
feature:</p>
<div class="code targetlang">
<pre># The feature name begins with a lowercase letter...
require 'etc'
# ... but the module name begins with an uppercase letter
puts &quot;Your login name: #{Etc.getlogin}&quot;
</pre>
</div>
<p> To stay consistent with this practice, you should always specify a<b>
lowercase</b> module name with SWIG's <tt>%module</tt> directive. SWIG
will automatically correct the resulting Ruby module name for your
extension. So for example, a SWIG interface file that begins with:</p>
<div class="code">
<pre>%module example</pre>
</div>
<p> will result in an extension module using the feature name &quot;example&quot;
and Ruby module name &quot;Example&quot;.</p>
<h3><a name="Ruby_nn7">35.1.5 Static linking</a></h3>
<p> An alternative approach to dynamic linking is to rebuild the Ruby
interpreter with your extension module added to it. In the past, this
approach was sometimes necessary due to limitations in dynamic loading
support on certain machines. However, the situation has improved
greatly over the last few years and you should not consider this
approach unless there is really no other option.</p>
<p>The usual procedure for adding a new module to Ruby involves finding
the Ruby source, adding an entry to the <tt>ext/Setup</tt> file, adding
your directory to the list of extensions in the file, and finally
rebuilding Ruby.</p>
<h3><a name="Ruby_nn8">35.1.6 Compilation of C++ extensions</a></h3>
<p> On most machines, C++ extension modules should be linked using the
C++ compiler. For example:</p>
<div class="code shell">
<pre>
$ swig -c++ -ruby example.i
$ g++ -fPIC -c example.cxx
$ g++ -fPIC -c example_wrap.cxx -I/usr/include/ruby-2.1.0
$ g++ -shared example.o example_wrap.o -o example.so
</pre>
</div>
<p> If you've written an <tt>extconf.rb</tt> script to automatically
generate a <tt>Makefile</tt> for your C++ extension module, keep in
mind that (as of this writing) Ruby still uses <tt>gcc</tt> and not <tt>
g++</tt> as its linker. As a result, the required C++ runtime library
support will not be automatically linked into your extension module and
it may fail to load on some platforms. A workaround for this problem is
use the <tt>mkmf</tt> module's <tt>append_library()</tt> method to add
one of the C++ runtime libraries to the list of libraries linked into
your extension, e.g.</p>
<div class="code targetlang">
<pre>require 'mkmf'
$libs = append_library($libs, &quot;supc++&quot;)
create_makefile('example')</pre>
</div>
<h2><a name="Ruby_nn9">35.2 Building Ruby Extensions under Windows 95/NT</a>
</h2>
<p> Building a SWIG extension to Ruby under Windows 95/NT is roughly
similar to the process used with Unix. Normally, you will want to
produce a DLL that can be loaded into the Ruby interpreter. For all
recent versions of Ruby, the procedure described above (i.e. using an <tt>
extconf.rb</tt> script) will work with Windows as well; you should be
able to build your code into a DLL by typing:</p>
<div class="code shell">
<pre>
C:\swigtest&gt; ruby extconf.rb
C:\swigtest&gt; nmake
C:\swigtest&gt; nmake install
</pre>
</div>
<p> The remainder of this section covers the process of compiling
SWIG-generated Ruby extensions with Microsoft Visual C++ 6 (i.e. within
the Developer Studio IDE, instead of using the command line tools). In
order to build extensions, you may need to download the source
distribution to the Ruby package, as you will need the Ruby header
files.</p>
<h3><a name="Ruby_nn10">35.2.1 Running SWIG from Developer Studio</a></h3>
<p> If you are developing your application within Microsoft developer
studio, SWIG can be invoked as a custom build option. The process
roughly follows these steps :</p>
<ul>
<li> Open up a new workspace and use the AppWizard to select a DLL
project.</li>
<li> Add both the SWIG interface file (the .i file), any supporting C
files, and the name of the wrapper file that will be created by SWIG
(i.e. <tt>example_wrap.c</tt>). Note : If using C++, choose a different
suffix for the wrapper file such as <tt>example_wrap.cxx</tt>. Don't
worry if the wrapper file doesn't exist yet--Developer Studio will keep
a reference to it around.</li>
<li> Select the SWIG interface file and go to the settings menu. Under
settings, select the &quot;Custom Build&quot; option.</li>
<li> Enter &quot;SWIG&quot; in the description field.</li>
<li> Enter &quot;<tt>swig -ruby -o $(ProjDir)\$(InputName)_wrap.c
$(InputPath)</tt>&quot; in the &quot;Build command(s) field&quot;. You may have to
include the path to swig.exe.</li>
<li> Enter &quot;<tt>$(ProjDir)\$(InputName)_wrap.c</tt>&quot; in the &quot;Output
files(s) field&quot;.</li>
<li> Next, select the settings for the entire project and go to the
C/C++ tab and select the Preprocessor category. Add NT=1 to the
Preprocessor definitions. This must be set else you will get
compilation errors. Also add IMPORT to the preprocessor definitions,
else you may get runtime errors. Also add the include directories for
your Ruby installation under &quot;Additional include directories&quot;.</li>
<li> Next, select the settings for the entire project and go to the Link
tab and select the General category. Set the name of the output file to
match the name of your Ruby module (i.e.. example.dll). Next add the
Ruby library file to your link libraries under Object/Library modules.
For example &quot;mswin32-ruby16.lib. You also need to add the path to the
library under the Input tab - Additional library path.</li>
<li> Build your project.</li>
</ul>
<p> Now, assuming all went well, SWIG will be automatically invoked when
you build your project. Any changes made to the interface file will
result in SWIG being automatically invoked to produce a new version of
the wrapper file. To run your new Ruby extension, simply run Ruby and
use the <tt>require</tt> command as normal. For example if you have
this ruby file run.rb:</p>
<div class="code targetlang">
<pre># file: run.rb
require 'Example'
# Call a c function
print &quot;Foo = &quot;, Example.Foo, &quot;\n&quot;</pre>
</div>
<p> Ensure the dll just built is in your path or current directory, then
run the Ruby script from the DOS/Command prompt:</p>
<div class="code shell">
<pre>
C:\swigtest&gt; ruby run.rb
Foo = 3.0
</pre>
</div>
<h2><a name="Ruby_nn11">35.3 The Ruby-to-C/C++ Mapping</a></h2>
<p> This section describes the basics of how SWIG maps C or C++
declarations in your SWIG interface files to Ruby constructs.</p>
<h3><a name="Ruby_nn12">35.3.1 Modules</a></h3>
<p> The SWIG <tt>%module</tt> directive specifies the name of the Ruby
module. If you specify:</p>
<div class="code">
<pre>%module example</pre>
</div>
<p> then everything is wrapped into a Ruby module named <tt>Example</tt>
that is nested directly under the global module. You can specify a more
deeply nested module by specifying the fully-qualified module name in
quotes, e.g.</p>
<div class="code">
<pre>%module &quot;foo::bar::spam&quot;</pre>
</div>
<p> An alternate method of specifying a nested module name is to use the
<tt>-prefix</tt> option on the SWIG command line. The prefix that you
specify with this option will be prepended to the module name specified
with the <tt>%module</tt> directive in your SWIG interface file. So for
example, this declaration at the top of your SWIG interface file:</p>
<div class="code">
<pre>%module &quot;foo::bar::spam&quot;</pre>
</div>
<p> will result in a nested module name of <tt>Foo::Bar::Spam</tt>, but
you can achieve the <span style="font-style: italic;">same</span>
effect by specifying:</p>
<div class="code">
<pre>%module spam</pre>
</div>
<p> and then running SWIG with the <tt>-prefix</tt> command line option:</p>
<div class="code shell">
<pre>
$ swig -ruby -prefix &quot;foo::bar::&quot; example.i
</pre>
</div>
<p> Starting with SWIG 1.3.20, you can also choose to wrap everything
into the global module by specifying the <tt>-globalmodule</tt> option
on the SWIG command line, i.e.</p>
<div class="code shell">
<pre>
$ swig -ruby -globalmodule example.i
</pre>
</div>
<p> Note that this does not relieve you of the requirement of specifying
the SWIG module name with the <tt>%module</tt> directive (or the <tt>
-module</tt> command-line option) as described earlier.</p>
<p>When choosing a module name, do not use the same name as a built-in
Ruby command or standard module name, as the results may be
unpredictable. Similarly, if you're using the <tt>-globalmodule</tt>
option to wrap everything into the global module, take care that the
names of your constants, classes and methods don't conflict with any of
Ruby's built-in names.</p>
<h3><a name="Ruby_nn13">35.3.2 Functions</a></h3>
<p> Global functions are wrapped as Ruby module methods. For example,
given the SWIG interface file <tt>example.i</tt>:</p>
<div class="code">
<pre>%module example
int fact(int n);</pre>
</div>
<p> and C source file <tt>example.c</tt>:</p>
<div class="code">
<pre>int fact(int n) {
if (n == 0)
return 1;
return (n * fact(n-1));
}</pre>
</div>
<p> SWIG will generate a method<i> fact</i> in the<i> Example</i> module
that can be used like so:</p>
<div class="code targetlang">
<pre>$ <b>irb</b>
irb(main):001:0&gt; <b>require 'example'</b>
true
irb(main):002:0&gt; <b>Example.fact(4)</b>
24</pre>
</div>
<h3><a name="Ruby_nn14">35.3.3 Variable Linking</a></h3>
<p> C/C++ global variables are wrapped as a pair of singleton methods
for the module: one to get the value of the global variable and one to
set it. For example, the following SWIG interface file declares two
global variables:</p>
<div class="code">
<pre>// SWIG interface file with global variables
%module example
...
%inline %{
extern int variable1;
extern double Variable2;
%}
...</pre>
</div>
<p> Now look at the Ruby interface:</p>
<div class="code targetlang">
<pre>$ <b>irb</b>
irb(main):001:0&gt; <b>require 'Example'</b>
true
irb(main):002:0&gt; <b>Example.variable1 = 2</b>
2
irb(main):003:0&gt; <b>Example.Variable2 = 4 * 10.3</b>
41.2
irb(main):004:0&gt; <b>Example.Variable2</b>
41.2</pre>
</div>
<p> If you make an error in variable assignment, you will receive an
error message. For example:</p>
<div class="code targetlang">
<pre>irb(main):005:0&gt; <b>Example.Variable2 = &quot;hello&quot;</b>
TypeError: no implicit conversion to float from string
from (irb):5:in `Variable2='
from (irb):5</pre>
</div>
<p> If a variable is declared as <tt>const</tt>, it is wrapped as a
read-only variable. Attempts to modify its value will result in an
error.</p>
<p>To make ordinary variables read-only, you can also use the <tt>
%immutable</tt> directive. For example:</p>
<div class="code">
<pre>%immutable;
%inline %{
extern char *path;
%}
%mutable;</pre>
</div>
<p> The <tt>%immutable</tt> directive stays in effect until it is
explicitly disabled using <tt>%mutable</tt>.</p>
<p>Note: When SWIG is invoked with the <tt>-globalmodule</tt> option in
effect, the C/C++ global variables will be translated into Ruby global
variables. Type-checking and the optional read-only characteristic are
available in the same way as described above. However the example would
then have to be modified and executed in the following way:<div class="code targetlang">
<pre>$ <b>irb</b>
irb(main):001:0&gt; <b>require 'Example'</b>
true
irb(main):002:0&gt; <b>$variable1 = 2</b>
2
irb(main):003:0&gt; <b>$Variable2 = 4 * 10.3</b>
41.2
irb(main):004:0&gt; <b>$Variable2</b>
41.2</pre>
</div></p>
<h3><a name="Ruby_nn15">35.3.4 Constants</a></h3>
<p> C/C++ constants are wrapped as module constants initialized to the
appropriate value. To create a constant, use <tt>#define</tt> or the <tt>
%constant</tt> directive. For example:</p>
<div class="code">
<pre>#define PI 3.14159
#define VERSION &quot;1.0&quot;
%constant int FOO = 42;
%constant const char *path = &quot;/usr/local&quot;;
const int BAR = 32;</pre>
</div>
<p> Remember to use the :: operator in Ruby to get at these constant
values, e.g.</p>
<div class="code targetlang">
<pre>$ <b>irb</b>
irb(main):001:0&gt; <b>require 'Example'</b>
true
irb(main):002:0&gt; <b>Example::PI</b>
3.14159</pre>
</div>
<h3><a name="Ruby_nn16">35.3.5 Pointers</a></h3>
<p> &quot;Opaque&quot; pointers to arbitrary C/C++ types (i.e. types that aren't
explicitly declared in your SWIG interface file) are wrapped as data
objects. So, for example, consider a SWIG interface file containing
only the declarations:</p>
<div class="code">
<pre>Foo *get_foo();
void set_foo(Foo *foo);</pre>
</div>
<p> For this case, the<i> get_foo()</i> method returns an instance of an
internally generated Ruby class:</p>
<div class="code targetlang">
<pre>irb(main):001:0&gt; <b>foo = Example::get_foo()</b>
#&lt;SWIG::TYPE_p_Foo:0x402b1654&gt;</pre>
</div>
<p> A <tt>NULL</tt> pointer is always represented by the Ruby <tt>nil</tt>
object.</p>
<h3><a name="Ruby_nn17">35.3.6 Structures</a></h3>
<p> C/C++ structs are wrapped as Ruby classes, with accessor methods
(i.e. &quot;getters&quot; and &quot;setters&quot;) for all of the struct members. For
example, this struct declaration:</p>
<div class="code">
<pre>struct Vector {
double x, y;
};</pre>
</div>
<p> gets wrapped as a <tt>Vector</tt> class, with Ruby instance methods <tt>
x</tt>, <tt>x=</tt>, <tt>y</tt> and <tt>y=</tt>. These methods can be
used to access structure data from Ruby as follows:</p>
<div class="code targetlang">
<pre>$ <b>irb</b>
irb(main):001:0&gt; <b>require 'Example'</b>
true
irb(main):002:0&gt; <b>f = Example::Vector.new</b>
#&lt;Example::Vector:0x4020b268&gt;
irb(main):003:0&gt; <b>f.x = 10</b>
nil
irb(main):004:0&gt; <b>f.x</b>
10.0</pre>
</div>
<p> Similar access is provided for unions and the public data members of
C++ classes.</p>
<p><tt>const</tt> members of a structure are read-only. Data members can
also be forced to be read-only using the <tt>%immutable</tt> directive
(in C++, <tt>private</tt> may also be used). For example:</p>
<div class="code">
<pre>struct Foo {
...
%immutable;
int x; /* Read-only members */
char *name;
%mutable;
...
};</pre>
</div>
<p> When <tt>char *</tt> members of a structure are wrapped, the
contents are assumed to be dynamically allocated using <tt>malloc</tt>
or <tt>new</tt> (depending on whether or not SWIG is run with the <tt>
-c++</tt> option). When the structure member is set, the old contents
will be released and a new value created. If this is not the behavior
you want, you will have to use a typemap (described shortly).</p>
<p>Array members are normally wrapped as read-only. For example, this
code:</p>
<div class="code">
<pre>struct Foo {
int x[50];
};</pre>
</div>
<p> produces a single accessor function like this:</p>
<div class="code">
<pre>int *Foo_x_get(Foo *self) {
return self-&gt;x;
};</pre>
</div>
<p> If you want to set an array member, you will need to supply a
&quot;memberin&quot; typemap described in the <a href="#Ruby_memberin_typemap">
section on typemaps</a>. As a special case, SWIG does generate code to
set array members of type <tt>char</tt> (allowing you to store a Ruby
string in the structure).</p>
<p>When structure members are wrapped, they are handled as pointers. For
example,</p>
<div class="code">
<pre>struct Foo {
...
};
struct Bar {
Foo f;
};</pre>
</div>
<p> generates accessor functions such as this:</p>
<div class="code">
<pre>Foo *Bar_f_get(Bar *b) {
return &amp;b-&gt;f;
}
void Bar_f_set(Bar *b, Foo *val) {
b-&gt;f = *val;
}</pre>
</div>
<h3><a name="Ruby_nn18">35.3.7 C++ classes</a></h3>
<p> Like structs, C++ classes are wrapped by creating a new Ruby class
of the same name with accessor methods for the public class member
data. Additionally, public member functions for the class are wrapped
as Ruby instance methods, and public static member functions are
wrapped as Ruby singleton methods. So, given the C++ class declaration:</p>
<div class="code">
<pre>class List {
public:
List();
~List();
int search(char *item);
void insert(char *item);
void remove(char *item);
char *get(int n);
int length;
static void print(List *l);
};</pre>
</div>
<p> SWIG would create a <tt>List</tt> class with:</p>
<ul>
<li> instance methods<i> search</i>,<i> insert</i>,<i> remove</i>, and<i>
get</i>;</li>
<li> instance methods<i> length</i> and<i> length=</i> (to get and set
the value of the<i> length</i> data member); and,</li>
<li> a<i> print</i> singleton method for the class.</li>
</ul>
<p> In Ruby, these functions are used as follows:</p>
<div class="code targetlang">
<pre>require 'Example'
l = Example::List.new
l.insert(&quot;Ale&quot;)
l.insert(&quot;Stout&quot;)
l.insert(&quot;Lager&quot;)
Example.print(l)
l.length()
----- produces the following output
Lager
Stout
Ale
3</pre>
</div>
<h3><a name="Ruby_nn19">35.3.8 C++ Inheritance</a></h3>
<p> The SWIG type-checker is fully aware of C++ inheritance. Therefore,
if you have classes like this:</p>
<div class="code">
<pre>class Parent {
...
};
class Child : public Parent {
...
};</pre>
</div>
<p> those classes are wrapped into a hierarchy of Ruby classes that
reflect the same inheritance structure. All of the usual Ruby utility
methods work normally:</p>
<div class="code">
<pre>irb(main):001:0&gt; <b>c = Child.new</b>
#&lt;Bar:0x4016efd4&gt;
irb(main):002:0&gt; <b>c.instance_of? Child</b>
true
irb(main):003:0&gt; <b>b.instance_of? Parent</b>
false
irb(main):004:0&gt; <b>b.is_a? Child</b>
true
irb(main):005:0&gt; <b>b.is_a? Parent</b>
true
irb(main):006:0&gt; <b>Child &lt; Parent</b>
true
irb(main):007:0&gt; <b>Child &gt; Parent</b>
false</pre>
</div>
<p> Furthermore, if you have a function like this:</p>
<div class="code">
<pre>void spam(Parent *f);</pre>
</div>
<p> then the function <tt>spam()</tt> accepts <tt>Parent</tt>* or a
pointer to any class derived from <tt>Parent</tt>.</p>
<p>Until recently, the Ruby module for SWIG didn't support multiple
inheritance, and this is still the default behavior. This doesn't mean
that you can't wrap C++ classes which inherit from multiple base
classes; it simply means that only the<b> first</b> base class listed
in the class declaration is considered, and any additional base classes
are ignored. As an example, consider a SWIG interface file with a
declaration like this:</p>
<div class="code">
<pre>class Derived : public Base1, public Base2
{
...
};</pre>
</div>
<p> For this case, the resulting Ruby class (<tt>Derived</tt>) will only
consider <tt>Base1</tt> as its superclass. It won't inherit any of <tt>
Base2</tt>'s member functions or data and it won't recognize <tt>Base2</tt>
as an &quot;ancestor&quot; of <tt>Derived</tt> (i.e. the<em> is_a?</em>
relationship would fail). When SWIG processes this interface file,
you'll see a warning message like:</p>
<div class="code shell">
<pre>example.i:5: Warning 802: Warning for Derived: Base Base2 ignored.
Multiple inheritance is not supported in Ruby.</pre>
</div>
<p> Starting with SWIG 1.3.20, the Ruby module for SWIG provides limited
support for multiple inheritance. Because the approach for dealing with
multiple inheritance introduces some limitations, this is an optional
feature that you can activate with the <tt>-minherit</tt> command-line
option:</p>
<div class="code shell">
<pre>
$ swig -c++ -ruby -minherit example.i
</pre>
</div>
<p> Using our previous example, if your SWIG interface file contains a
declaration like this:</p>
<div class="code">
<pre>class Derived : public Base1, public Base2
{
...
};</pre>
</div>
<p> and you run SWIG with the <tt>-minherit</tt> command-line option,
then you will end up with a Ruby class <tt>Derived</tt> that appears to
&quot;inherit&quot; the member data and functions from both <tt>Base1</tt> and <tt>
Base2</tt>. What actually happens is that three different top-level
classes are created, with Ruby's <tt>Object</tt> class as their
superclass. Each of these classes defines a nested module named <tt>
Impl</tt>, and it's in these nested <tt>Impl</tt> modules that the
actual instance methods for the classes are defined, i.e.</p>
<div class="code targetlang">
<pre>class Base1
module Impl
# Define Base1 methods here
end
include Impl
end
class Base2
module Impl
# Define Base2 methods here
end
include Impl
end
class Derived
module Impl
include Base1::Impl
include Base2::Impl
# Define Derived methods here
end
include Impl
end</pre>
</div>
<p> Observe that after the nested <tt>Impl</tt> module for a class is
defined, it is mixed-in to the class itself. Also observe that the <tt>
Derived::Impl</tt> module first mixes-in its base classes' <tt>Impl</tt>
modules, thus &quot;inheriting&quot; all of their behavior.</p>
<p>The primary drawback is that, unlike the default mode of operation,
neither <tt>Base1</tt> nor <tt>Base2</tt> is a true superclass of <tt>
Derived</tt> anymore:</p>
<div class="code targetlang">
<pre>obj = Derived.new
obj.is_a? Base1 # this will return false...
obj.is_a? Base2 # ... and so will this</pre>
</div>
<p> In most cases, this is not a serious problem since objects of type <tt>
Derived</tt> will otherwise behave as though they inherit from both <tt>
Base1</tt> and <tt>Base2</tt> (i.e. they exhibit <a href="http://c2.com/cgi/wiki?DuckTyping">
&quot;Duck Typing&quot;</a>).</p>
<h3><a name="Ruby_nn20">35.3.9 C++ Overloaded Functions</a></h3>
<p> C++ overloaded functions, methods, and constructors are mostly
supported by SWIG. For example, if you have two functions like this:</p>
<div class="code">
<pre>void foo(int);
void foo(char *c);</pre>
</div>
<p> You can use them in Ruby in a straightforward manner:</p>
<div class="code targetlang">
<pre>irb(main):001:0&gt; <b>foo(3)</b> # foo(int)
irb(main):002:0&gt; <b>foo(&quot;Hello&quot;)</b> # foo(char *c)</pre>
</div>
<p>Similarly, if you have a class like this,</p>
<div class="code">
<pre>class Foo {
public:
Foo();
Foo(const Foo &amp;);
...
};</pre>
</div>
<p>you can write Ruby code like this:</p>
<div class="code targetlang">
<pre>irb(main):001:0&gt; <b>f = Foo.new</b> # Create a Foo
irb(main):002:0&gt; <b>g = Foo.new(f)</b> # Copy f</pre>
</div>
<p> Overloading support is not quite as flexible as in C++. Sometimes
there are methods that SWIG can't disambiguate. For example:</p>
<div class="code">
<pre>void spam(int);
void spam(short);</pre>
</div>
<p>or</p>
<div class="code">
<pre>void foo(Bar *b);
void foo(Bar &amp;b);</pre>
</div>
<p> If declarations such as these appear, you will get a warning message
like this:</p>
<div class="code shell">
<pre>
example.i:12: Warning 509: Overloaded method spam(short) effectively ignored,
example.i:11: Warning 509: as it is shadowed by spam(int).
</pre>
</div>
<p> To fix this, you either need to ignore or rename one of the methods.
For example:</p>
<div class="code">
<pre>%rename(spam_short) spam(short);
...
void spam(int);
void spam(short); // Accessed as spam_short</pre>
</div>
<p>or</p>
<div class="code">
<pre>%ignore spam(short);
...
void spam(int);
void spam(short); // Ignored</pre>
</div>
<p> SWIG resolves overloaded functions and methods using a
disambiguation scheme that ranks and sorts declarations according to a
set of type-precedence rules. The order in which declarations appear in
the input does not matter except in situations where ambiguity
arises--in this case, the first declaration takes precedence.</p>
<p>Please refer to the <a href="#SWIGPlus">&quot;SWIG and C++&quot;</a> chapter
for more information about overloading.</p>
<h3><a name="Ruby_nn21">35.3.10 C++ Operators</a></h3>
<p> For the most part, overloaded operators are handled automatically by
SWIG and do not require any special treatment on your part. So if your
class declares an overloaded addition operator, e.g.</p>
<div class="code">
<pre>class Complex {
...
Complex operator+(Complex &amp;);
...
};</pre>
</div>
<p> the resulting Ruby class will also support the addition (+) method
correctly.</p>
<p>For cases where SWIG's built-in support is not sufficient, C++
operators can be wrapped using the <tt>%rename</tt> directive
(available on SWIG 1.3.10 and later releases). All you need to do is
give the operator the name of a valid Ruby identifier. For example:</p>
<div class="code">
<pre>%rename(add_complex) operator+(Complex &amp;, Complex &amp;);
...
Complex operator+(Complex &amp;, Complex &amp;);</pre>
</div>
<p>Now, in Ruby, you can do this:</p>
<div class="code targetlang">
<pre>a = Example::Complex.new(2, 3)
b = Example::Complex.new(4, -1)
c = Example.add_complex(a, b)</pre>
</div>
<p> More details about wrapping C++ operators into Ruby operators is
discussed in the <a href="#Ruby_operator_overloading">section on
operator overloading</a>.</p>
<h3><a name="Ruby_nn22">35.3.11 C++ namespaces</a></h3>
<p> SWIG is aware of C++ namespaces, but namespace names do not appear
in the module nor do namespaces result in a module that is broken up
into submodules or packages. For example, if you have a file like this,</p>
<div class="code">
<pre>%module example
namespace foo {
int fact(int n);
struct Vector {
double x, y, z;
};
};</pre>
</div>
<p>it works in Ruby as follows:</p>
<div class="code targetlang">
<pre>irb(main):001:0&gt; <b>require 'example'</b>
true
irb(main):002:0&gt; <b>Example.fact(3)</b>
6
irb(main):003:0&gt; <b>v = Example::Vector.new</b>
#&lt;Example::Vector:0x4016f4d4&gt;
irb(main):004:0&gt; <b>v.x = 3.4</b>
3.4
irb(main):004:0&gt; <b>v.y</b>
0.0</pre>
</div>
<p> If your program has more than one namespace, name conflicts (if any)
can be resolved using <tt>%rename</tt> For example:</p>
<div class="code">
<pre>%rename(Bar_spam) Bar::spam;
namespace Foo {
int spam();
}
namespace Bar {
int spam();
}</pre>
</div>
<p> If you have more than one namespace and your want to keep their
symbols separate, consider wrapping them as separate SWIG modules. For
example, make the module name the same as the namespace and create
extension modules for each namespace separately. If your program
utilizes thousands of small deeply nested namespaces each with
identical symbol names, well, then you get what you deserve.</p>
<h3><a name="Ruby_nn23">35.3.12 C++ templates</a></h3>
<p> C++ templates don't present a huge problem for SWIG. However, in
order to create wrappers, you have to tell SWIG to create wrappers for
a particular template instantiation. To do this, you use the <tt>
%template</tt> directive. For example:</p>
<div class="code">
<pre>%module example
%{
#include &quot;pair.h&quot;
%}
template&lt;class T1, class T2&gt;
struct pair {
typedef T1 first_type;
typedef T2 second_type;
T1 first;
T2 second;
pair();
pair(const T1&amp;, const T2&amp;);
~pair();
};
%template(Pairii) pair&lt;int, int&gt;;</pre>
</div>
<p>In Ruby:</p>
<div class="code targetlang">
<pre>irb(main):001:0&gt; <b>require 'example'</b>
true
irb(main):002:0&gt; <b>p = Example::Pairii.new(3, 4)</b>
#&lt;Example:Pairii:0x4016f4df&gt;
irb(main):003:0&gt; <b>p.first</b>
3
irb(main):004:0&gt; <b>p.second</b>
4</pre>
</div>
<h3><a name="Ruby_nn23_1">35.3.13 C++ Standard Template Library (STL)</a>
</h3>
<p> On a related note, the standard SWIG library contains a number of
modules that provide typemaps for standard C++ library classes (such as
<tt>std::pair</tt>, <tt>std::string</tt> and <tt>std::vector</tt>).
These library modules don't provide wrappers around the templates
themselves, but they do make it convenient for users of your extension
module to pass Ruby objects (such as arrays and strings) to wrapped C++
code that expects instances of standard C++ templates. For example,
suppose the C++ library you're wrapping has a function that expects a
vector of floats:</p>
<div class="code">
<pre>%module example
float sum(const std::vector&lt;float&gt;&amp; values);</pre>
</div>
<p> Rather than go through the hassle of writing an &quot;in&quot; typemap to
convert an array of Ruby numbers into a std::vector&lt;float&gt;, you can
just use the <tt>std_vector.i</tt> module from the standard SWIG
library:</p>
<div class="code">
<pre>%module example
%include std_vector.i
float sum(const std::vector&lt;float&gt;&amp; values);</pre>
</div>
<p>Ruby's STL wrappings provide additional methods to make them behave
more similarly to Ruby's native classes.</p>
<p>Thus, you can do, for example:</p>
<div class="targetlang">
<pre>v = IntVector.new
v &lt;&lt; 2
v &lt;&lt; 3
v &lt;&lt; 4
v.each { |x| puts x }
=&gt; 2
3
4
v.delete_if { |x| x == 3 }
=&gt; [2, 4]</pre>
</div>
<p>The SWIG Ruby module provides also the ability for all the STL
containers to carry around Ruby native objects (Fixnum, Classes, etc)
making them act almost like Ruby's own Array, Hash, etc. To do that,
you need to define a container that contains a swig::GC_VALUE, like:</p>
<div class="code">
<pre>
%module nativevector
%{
std::vector&lt; swig::GC_VALUE &gt; NativeVector;
%}
%template(NativeVector) std::vector&lt; swig::GC_VALUE &gt;;
</pre>
</div>
<p>This vector can then contain any Ruby object, making them almost
identical to Ruby's own Array class.</p>
<div class="targetlang">
<pre>require 'nativevector'
include NativeVector
v = NativeVector.new
v &lt;&lt; 1
v &lt;&lt; [1, 2]
v &lt;&lt; 'hello'
class A; end
v &lt;&lt; A.new
puts v
=&gt; [1, [1, 2], 'hello', #&lt;A:0x245325&gt;]
</pre>
</div>
<p>Obviously, there is a lot more to template wrapping than shown in
these examples. More details can be found in the <a href="#SWIGPlus">
SWIG and C++</a> chapter.</p>
<h3><a name="Ruby_C_STL_Functors">35.3.14 C++ STL Functors</a></h3>
<p>Some containers in the STL allow you to modify their default behavior
by using so called functors or function objects. Functors are often
just a very simple struct with <tt>operator()</tt> redefined or an
actual C/C++ function. This allows you, for example, to always keep the
sort order of a STL container to your liking.</p>
<p>The Ruby STL mappings allows you to modify those containers that
support functors using Ruby procs or methods, instead. Currently, this
includes <tt>std::set</tt>, <tt>set::map</tt>, <tt>std::multiset</tt>
and <tt>std::multimap</tt>.</p>
<p>The functors in swig are called <tt>swig::UnaryFunction</tt> and <tt>
swig::BinaryFunction</tt>. For C++ predicates (ie. functors that must
return bool as a result) <tt>swig::UnaryPredicate</tt> and <tt>
swig::BinaryPredicate</tt> are provided.</p>
<p>As an example, if given this swig file:</p>
<div class="code">
<pre>
%module intset;
%include &lt;std_set.i&gt;
%template(IntSet) std::set&lt; int, swig::BinaryPredicate &gt;;
</pre>
</div>
<p>You can then use the set from Ruby with or without a proc object as a
predicate:</p>
<div class="targetlang">
<pre>
require 'intset'
include Intset
# Default sorting behavior defined in C++
a = IntSet.new
a &lt;&lt; 1
a &lt;&lt; 2
a &lt;&lt; 3
a
<b>=&gt; [1, 2, 3]</b>
# Custom sorting behavior defined by a Ruby proc
b = IntSet.new( proc { |a, b| a &gt; b } )
b &lt;&lt; 1
b &lt;&lt; 2
b &lt;&lt; 3
b
<b>=&gt; [3, 2, 1]</b>
</pre>
</div>
<h3><a name="Ruby_C_Iterators">35.3.15 C++ STL Iterators</a></h3>
<p>The STL is well known for the use of iterators. There are a number of
iterators possible with different properties, but in general there are
two main categories: const iterators and non-const iterators. The const
iterators can access and not modify the values they point at, while the
non-const iterators can both read and modify the values.</p>
<p>The Ruby STL wrappings support both type of iterators by using a
proxy class in-between. This proxy class is <tt>swig::Iterator</tt> or <tt>
swig::ConstIterator</tt>. Derived from them are template classes that
need to be initialized with the actual iterator for the container you
are wrapping and often times with the beginning and ending points of
the iteration range.</p>
<p>The SWIG STL library already provides typemaps to all the standard
containers to do this wrapping automatically for you, but if you have
your own STL-like iterator, you will need to write your own typemap for
them. For out typemaps, the special functions <tt>make_const_iterator</tt>
and <tt>make_nonconst_iterator</tt> are provided.</p>
<p>These can be used either like:</p>
<div class="code">
<pre>
make_const_iterator( iterator, rubyclass );
make_const_iterator( iterator, iterator_begin, iterator_end, rubyclass );
</pre>
</div>
<p>The iterators support a <tt>next()</tt> and <tt>previous()</tt>
member function to just change the iterator without returning anything.
<tt>previous()</tt> should obviously only be used for bidirectional
iterators. You can also advance the iterator multiple steps by using
standard math operations like <tt>+=</tt>.</p>
<p>The value the iterator points at can be accessed with <tt>value()</tt>
-- this is equivalent to dereferencing it with <tt>*i</tt>. For
non-const iterators, a <tt>value=()</tt> function is also provided
which allows you to change the value pointed by the iterator. This is
equivalent to the C++ construct of dereferencing and assignment, like <tt>
*i = something</tt>.</p>
<p>Thus, given say a vector class of doubles defined as:</p>
<div class="code">
<pre>
%module doublevector
%include std_vector.i
%template(DoubleVector) std::vector&lt;double&gt;;
</pre>
</div>
<p>Its iterator can then be used from Ruby like:</p>
<div class="targetlang">
<pre>
require 'doublevector'
include Doublevector
v = DoubleVector.new
v &lt;&lt; 1
v &lt;&lt; 2
v &lt;&lt; 3
#
# an elaborate and less efficient way of doing v.map! { |x| x+2 }
#
i = v.begin
e = v.end
while i != e
val = i.value
val += 2
i.value = val
i.next
end
i
<b>&gt;&gt; [3, 4, 5 ]</b>
</pre>
</div>
<p>If you'd rather have STL classes without any iterators, you should
define <tt>-DSWIG_NO_EXPORT_ITERATOR_METHODS</tt> when running swig.</p>
<h3><a name="Ruby_nn24">35.3.16 C++ Smart Pointers</a></h3>
<h4><a name="Ruby_smart_pointers_shared_ptr">35.3.16.1 The shared_ptr
Smart Pointer</a></h4>
<p> The C++11 standard provides <tt>std::shared_ptr</tt> which was
derived from the Boost implementation, <tt>boost::shared_ptr</tt>. Both
of these are available for Ruby in the SWIG library and usage is
outlined in the <a href="#Library_std_shared_ptr">shared_ptr smart
pointer</a> library section.</p>
<h4><a name="Ruby_smart_pointers_generic">35.3.16.2 Generic Smart
Pointers</a></h4>
<p> In certain C++ programs, it is common to use classes that have been
wrapped by so-called &quot;smart pointers.&quot; Generally, this involves the use
of a template class that implements <tt>operator-&gt;()</tt> like this:</p>
<div class="code">
<pre>template&lt;class T&gt; class SmartPtr {
...
T *operator-&gt;();
...
}</pre>
</div>
<p>Then, if you have a class like this,</p>
<div class="code">
<pre>class Foo {
public:
int x;
int bar();
};</pre>
</div>
<p>A smart pointer would be used in C++ as follows:</p>
<div class="code">
<pre>SmartPtr&lt;Foo&gt; p = CreateFoo(); // Created somehow (not shown)
...
p-&gt;x = 3; // Foo::x
int y = p-&gt;bar(); // Foo::bar</pre>
</div>
<p> To wrap this in Ruby, simply tell SWIG about the <tt>SmartPtr</tt>
class and the low-level <tt>Foo</tt> object. Make sure you instantiate <tt>
SmartPtr</tt> using <tt>%template</tt> if necessary. For example:</p>
<div class="code">
<pre>%module example
...
%template(SmartPtrFoo) SmartPtr&lt;Foo&gt;;
...</pre>
</div>
<p>Now, in Ruby, everything should just &quot;work&quot;:</p>
<div class="code targetlang">
<pre>irb(main):001:0&gt; <b>p = Example::CreateFoo()</b> # Create a smart-pointer somehow
#&lt;Example::SmartPtrFoo:0x4016f4df&gt;
irb(main):002:0&gt; <b>p.x = 3</b> # Foo::x
3
irb(main):003:0&gt; <b>p.bar()</b> # Foo::bar</pre>
</div>
<p> If you ever need to access the underlying pointer returned by <tt>
operator-&gt;()</tt> itself, simply use the <tt>__deref__()</tt> method.
For example:</p>
<div class="code targetlang">
<pre>irb(main):004:0&gt; <b>f = p.__deref__()</b> # Returns underlying Foo *</pre>
</div>
<h3><a name="Ruby_nn25">35.3.17 Cross-Language Polymorphism</a></h3>
<p> SWIG's Ruby module supports cross-language polymorphism (a.k.a. the
&quot;directors&quot; feature) similar to that for SWIG's Python module. Rather
than duplicate the information presented in the <a href="#Python">
Python</a> chapter, this section just notes the differences that you
need to be aware of when using this feature with Ruby.</p>
<h4><a name="Ruby_nn26">35.3.17.1 Exception Unrolling</a></h4>
<p> Whenever a C++ director class routes one of its virtual member
function calls to a Ruby instance method, there's always the
possibility that an exception will be raised in the Ruby code. By
default, those exceptions are ignored, which simply means that the
exception will be exposed to the Ruby interpreter. If you would like to
change this behavior, you can use the <tt>%feature(&quot;director:except&quot;)</tt>
directive to indicate what action should be taken when a Ruby exception
is raised. The following code should suffice in most cases:</p>
<div class="code">
<pre>%feature(&quot;director:except&quot;) {
throw Swig::DirectorMethodException($error);
}</pre>
</div>
<p> When this feature is activated, the call to the Ruby instance method
is &quot;wrapped&quot; using the <tt>rb_rescue2()</tt> function from Ruby's C
API. If any Ruby exception is raised, it will be caught here and a C++
exception is raised in its place.</p>
<h2><a name="Ruby_nn27">35.4 Naming</a></h2>
<p>Ruby has several common naming conventions. Constants are generally
in upper case, module and class names are in camel case and methods are
in lower case with underscores. For example:</p>
<div class="code">
<ul>
<li><strong>MATH::PI</strong> is a constant name</li>
<li><strong>MyClass</strong> is a class name</li>
<li><strong>my_method</strong> is a method name</li>
</ul>
</div>
<p>Prior to version 1.3.28, SWIG did not support these Ruby conventions.
The only modifications it made to names was to capitalize the first
letter of constants (which includes module and class names).</p>
<p>SWIG 1.3.28 introduces the new -autorename command line parameter.
When this parameter is specified, SWIG will automatically change
constant, class and method names to conform with the standard Ruby
naming conventions. For example:</p>
<div class="code shell">
<pre>$ swig -ruby -autorename example.i
</pre>
</div>
<p>To disable renaming use the -noautorename command line option.</p>
<p>Since this change significantly changes the wrapper code generated by
SWIG, it is turned off by default in SWIG 1.3.28. However, it is
planned to become the default option in future releases.</p>
<h3><a name="Ruby_nn28">35.4.1 Defining Aliases</a></h3>
<p> It's a fairly common practice in the Ruby built-ins and standard
library to provide aliases for method names. For example,<em>
Array#size</em> is an alias for<em> Array#length</em>. If you would
like to provide an alias for one of your class' instance methods, one
approach is to use SWIG's <tt>%extend</tt> directive to add a new
method of the aliased name that calls the original function. For
example:</p>
<div class="code">
<pre>class MyArray {
public:
// Construct an empty array
MyArray();
// Return the size of this array
size_t length() const;
};
%extend MyArray {
// MyArray#size is an alias for MyArray#length
size_t size() const {
return $self-&gt;length();
}
}
</pre>
</div>
<p> A better solution is to use the <tt>%alias</tt> directive (unique to
SWIG's Ruby module). The previous example could then be rewritten as:</p>
<div class="code">
<pre>// MyArray#size is an alias for MyArray#length
%alias MyArray::length &quot;size&quot;;
class MyArray {
public:
// Construct an empty array
MyArray();
// Return the size of this array
size_t length() const;
};</pre>
</div>
<p> Multiple aliases can be associated with a method by providing a
comma-separated list of aliases to the <tt>%alias</tt> directive, e.g.</p>
<div class="code">
<pre>%alias MyArray::length &quot;amount, quantity, size&quot;;</pre>
</div>
<p> From an end-user's standpoint, there's no functional difference
between these two approaches; i.e. they should get the same result from
calling either<em> MyArray#size</em> or<em> MyArray#length</em>.
However, when the <tt>%alias</tt> directive is used, SWIG doesn't need
to generate all of the wrapper code that's usually associated with
added methods like our<em> MyArray::size()</em> example.</p>
<p>Note that the <tt>%alias</tt> directive is implemented using SWIG's
&quot;features&quot; mechanism and so the same name matching rules used for other
kinds of features apply (see the chapter on <a href="#Customization">
&quot;Customization Features&quot;</a>) for more details).</p>
<h3><a name="Ruby_nn29">35.4.2 Predicate Methods</a></h3>
<p> Ruby methods that return a boolean value and end in a question mark
are known as predicate methods. Examples of predicate methods in
standard Ruby classes include<em> Array#empty?</em> (which returns <tt>
true</tt> for an array containing no elements) and<em>
Object#instance_of?</em> (which returns <tt>true</tt> if the object is
an instance of the specified class). For consistency with Ruby
conventions, methods that return boolean values should be marked as
predicate methods.</p>
<p>One cumbersome solution to this problem is to rename the method
(using SWIG's <tt>%rename</tt> directive) and provide a custom typemap
that converts the function's actual return type to Ruby's <tt>true</tt>
or <tt>false</tt>. For example:</p>
<div class="code">
<pre>%rename(&quot;is_it_safe?&quot;) is_it_safe();
%typemap(out) int is_it_safe &quot;$result = ($1 != 0) ? Qtrue : Qfalse;&quot;
int is_it_safe();</pre>
</div>
<p> A better solution is to use the <tt>%predicate</tt> directive
(unique to SWIG's Ruby module) to designate a method as a predicate
method. For the previous example, this would look like:</p>
<div class="code">
<pre>%predicate is_it_safe();
int is_it_safe();</pre>
</div>
<p>This method would be invoked from Ruby code like this:</p>
<div class="code targetlang">
<pre>irb(main):001:0&gt; <b>Example::is_it_safe?</b>
true</pre>
</div>
<p> The <tt>%predicate</tt> directive is implemented using SWIG's
&quot;features&quot; mechanism and so the same name matching rules used for other
kinds of features apply (see the chapter on <a href="#Customization">
&quot;Customization Features&quot;</a>) for more details).</p>
<h3><a name="Ruby_nn30">35.4.3 Bang Methods</a></h3>
<p> Ruby methods that modify an object in-place and end in an
exclamation mark are known as bang methods. An example of a bang method
is<em> Array#sort!</em> which changes the ordering of items in an
array. Contrast this with<em> Array#sort</em>, which returns a copy of
the array with the items sorted instead of modifying the original
array. For consistency with Ruby conventions, methods that modify
objects in place should be marked as bang methods.</p>
<p>Bang methods can be marked using the <tt>%bang</tt> directive which
is unique to the Ruby module and was introduced in SWIG 1.3.28. For
example:</p>
<div class="code">
<pre>%bang sort(int arr[]);
int sort(int arr[]); </pre>
</div>
<p>This method would be invoked from Ruby code like this:</p>
<div class="code">
<pre>irb(main):001:0&gt; <b>Example::sort!(arr)</b></pre>
</div>
<p> The <tt>%bang</tt> directive is implemented using SWIG's &quot;features&quot;
mechanism and so the same name matching rules used for other kinds of
features apply (see the chapter on <a href="#Customization">
&quot;Customization Features&quot;</a>) for more details).</p>
<h3><a name="Ruby_nn31">35.4.4 Getters and Setters</a></h3>
<p> Often times a C++ library will expose properties through getter and
setter methods. For example:</p>
<div class="code">
<pre>class Foo {
Foo() {}
int getValue() { return value_; }
void setValue(int value) { value_ = value; }
private:
int value_;
};</pre>
</div>
<p>By default, SWIG will expose these methods to Ruby as <tt>get_value</tt>
and <tt>set_value.</tt> However, it more natural for these methods to
be exposed in Ruby as <tt>value</tt> and <tt>value=.</tt> That allows
the methods to be used like this:</p>
<div class="code">
<pre>irb(main):001:0&gt; <b>foo = Foo.new()</b>
irb(main):002:0&gt; <b>foo.value = 5</b>
irb(main):003:0&gt; <b>puts foo.value</b></pre>
</div>
<p> This can be done by using the %rename directive:</p>
<div class="code">
<pre>%rename(&quot;value&quot;) Foo::getValue();
%rename(&quot;value=&quot;) Foo::setValue(int value);</pre>
</div>
<h2><a name="Ruby_nn32">35.5 Input and output parameters</a></h2>
<p> A common problem in some C programs is handling parameters passed as
simple pointers. For example:</p>
<div class="code">
<pre>void add(int x, int y, int *result) {
*result = x + y;
}</pre>
</div>
<p> or</p>
<div class="code">
<pre>
int sub(int *x, int *y) {
return *x-*y;
}</pre>
</div>
<p> The easiest way to handle these situations is to use the <tt>
typemaps.i</tt> file. For example:</p>
<div class="code">
<pre>%module Example
%include &quot;typemaps.i&quot;
void add(int, int, int *OUTPUT);
int sub(int *INPUT, int *INPUT);</pre>
</div>
<p>In Ruby, this allows you to pass simple values. For example:</p>
<div class="code targetlang">
<pre>a = Example.add(3, 4)
puts a
7
b = Example.sub(7, 4)
puts b
3</pre>
</div>
<p> Notice how the <tt>INPUT</tt> parameters allow integer values to be
passed instead of pointers and how the <tt>OUTPUT</tt> parameter
creates a return result.</p>
<p>If you don't want to use the names <tt>INPUT</tt> or <tt>OUTPUT</tt>,
use the <tt>%apply</tt> directive. For example:</p>
<div class="code">
<pre>%module Example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *result };
%apply int *INPUT { int *x, int *y};
void add(int x, int y, int *result);
int sub(int *x, int *y);</pre>
</div>
<p> If a function mutates one of its parameters like this,</p>
<div class="code">
<pre>void negate(int *x) {
*x = -(*x);
}</pre>
</div>
<p>you can use <tt>INOUT</tt> like this:</p>
<div class="code">
<pre>%include &quot;typemaps.i&quot;
...
void negate(int *INOUT);</pre>
</div>
<p>In Ruby, a mutated parameter shows up as a return value. For example:</p>
<div class="code targetlang">
<pre>a = Example.negate(3)
print a
-3</pre>
</div>
<p> The most common use of these special typemap rules is to handle
functions that return more than one value. For example, sometimes a
function returns a result as well as a special error code:</p>
<div class="code">
<pre>/* send message, return number of bytes sent, success code, and error_code */
int send_message(char *text, int *success, int *error_code);</pre>
</div>
<p> To wrap such a function, simply use the <tt>OUTPUT</tt> rule above.
For example:</p>
<div class="code">
<pre>%module example
%include &quot;typemaps.i&quot;
...
int send_message(char *, int *OUTPUT, int *OUTPUT);</pre>
</div>
<p> When used in Ruby, the function will return an array of multiple
values.</p>
<div class="code targetlang">
<pre>bytes, success, error_code = send_message(&quot;Hello World&quot;)
if not success
print &quot;error #{error_code} : in send_message&quot;
else
print &quot;Sent&quot;, bytes
end</pre>
</div>
<p> Another way to access multiple return values is to use the <tt>
%apply</tt> rule. In the following example, the parameters rows and
columns are related to SWIG as <tt>OUTPUT</tt> values through the use
of <tt>%apply</tt></p>
<div class="code">
<pre>%module Example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *rows, int *columns };
...
void get_dimensions(Matrix *m, int *rows, int*columns);</pre>
</div>
<p>In Ruby:</p>
<div class="code targetlang">
<pre>r, c = Example.get_dimensions(m)</pre>
</div>
<h2><a name="Ruby_nn33">35.6 Exception handling</a></h2>
<h3><a name="Ruby_nn34">35.6.1 Using the %exception directive</a></h3>
<p>The SWIG <tt>%exception</tt> directive can be used to define a
user-definable exception handler that can convert C/C++ errors into
Ruby exceptions. The chapter on <a href="#Customization">Customization
Features</a> contains more details, but suppose you have a C++ class
like the following :</p>
<div class="code">
<pre>class DoubleArray {
private:
int n;
double *ptr;
public:
// Create a new array of fixed size
DoubleArray(int size) {
ptr = new double[size];
n = size;
}
// Destroy an array
~DoubleArray() {
delete ptr;
}
// Return the length of the array
int length() {
return n;
}
// Get an array item and perform bounds checking.
double getitem(int i) {
if ((i &gt;= 0) &amp;&amp; (i &lt; n))
return ptr[i];
else
throw RangeError();
}
// Set an array item and perform bounds checking.
void setitem(int i, double val) {
if ((i &gt;= 0) &amp;&amp; (i &lt; n))
ptr[i] = val;
else {
throw RangeError();
}
}
};</pre>
</div>
<p> Since several methods in this class can throw an exception for an
out-of-bounds access, you might want to catch this in the Ruby
extension by writing the following in an interface file:</p>
<div class="code">
<pre>%exception {
try {
$action
}
catch (const RangeError&amp;) {
static VALUE cpperror = rb_define_class(&quot;CPPError&quot;, rb_eStandardError);
rb_raise(cpperror, &quot;Range error.&quot;);
}
}
class DoubleArray {
...
};</pre>
</div>
<p> The exception handling code is inserted directly into generated
wrapper functions. When an exception handler is defined, errors can be
caught and used to gracefully raise a Ruby exception instead of forcing
the entire program to terminate with an uncaught error.</p>
<p>As shown, the exception handling code will be added to every wrapper
function. Because this is somewhat inefficient, you might consider
refining the exception handler to only apply to specific methods like
this:</p>
<div class="code">
<pre>%exception getitem {
try {
$action
} catch (const RangeError&amp;) {
static VALUE cpperror = rb_define_class(&quot;CPPError&quot;, rb_eStandardError);
rb_raise(cpperror, &quot;Range error in getitem.&quot;);
}
}
%exception setitem {
try {
$action
} catch (const RangeError&amp;) {
static VALUE cpperror = rb_define_class(&quot;CPPError&quot;, rb_eStandardError);
rb_raise(cpperror, &quot;Range error in setitem.&quot;);
}
}</pre>
</div>
<p> In this case, the exception handler is only attached to methods and
functions named <tt>getitem</tt> and <tt>setitem</tt>.</p>
<p>Since SWIG's exception handling is user-definable, you are not
limited to C++ exception handling. See the chapter on <a href="#Customization">
Customization Features</a> for more examples.</p>
<h3><a name="Ruby_nn34_2">35.6.2 Handling Ruby Blocks</a></h3>
<p>One of the highlights of Ruby and most of its standard library is the
use of blocks, which allow the easy creation of continuations and other
niceties. Blocks in ruby are also often used to simplify the passing of
many arguments to a class.</p>
<p>In order to make your class constructor support blocks, you can take
advantage of the %exception directive, which will get run after the C++
class' constructor was called.</p>
<p>For example, this yields the class over after its construction:</p>
<div class="code">
<pre>class Window
{
public:
Window(int x, int y, int w, int h);
// .... other methods here ....
};
// Add support for yielding self in the Class' constructor.
%exception Window::Window {
$action
if (rb_block_given_p()) {
rb_yield(self);
}
}</pre>
</div>
<p> Then, in ruby, it can be used like:</p>
<div class="targetlang">
<pre>
Window.new(0, 0, 360, 480) { |w|
w.color = Fltk::RED
w.border = false
}
</pre>
</div>
<p>For other methods, you can usually use a dummy parameter with a
special in typemap, like:</p>
<div class="code">
<pre>
//
// original function was:
//
// void func(int x);
%typemap(in, numinputs=0) int RUBY_YIELD_SELF {
if ( !rb_block_given_p() )
rb_raise(&quot;No block given&quot;);
return rb_yield(self);
}
%extend {
void func(int x, int RUBY_YIELD_SELF );
}
</pre>
</div>
<p>For more information on typemaps, see <a href="#Ruby_nn37">Typemaps</a>
.</p>
<h3><a name="Ruby_nn35">35.6.3 Raising exceptions</a></h3>
<p>There are three ways to raise exceptions from C++ code to Ruby.</p>
<p>The first way is to use <tt>SWIG_exception(int code, const char *msg)</tt>
. The following table shows the mappings from SWIG error codes to Ruby
exceptions:</p>
<div class="diagram">
<table border="1" class="diagram" summary="Mapping between SWIG error codes and Ruby exceptions."
width="80%"><tbody>
<tr><td class="diagram"><div>SWIG_MemoryError</div></td><td><div>
rb_eNoMemError</div></td></tr>
<tr><td class="diagram"><div>SWIG_IOError</div></td><td><div>rb_eIOError</div>
</td></tr>
<tr><td class="diagram"><div>SWIG_RuntimeError</div></td><td><div>
rb_eRuntimeError</div></td></tr>
<tr><td class="diagram"><div>SWIG_IndexError</div></td><td><div>
rb_eIndexError</div></td></tr>
<tr><td class="diagram"><div>SWIG_TypeError</div></td><td><div>
rb_eTypeError</div></td></tr>
<tr><td class="diagram"><div>SWIG_DivisionByZero</div></td><td><div>
rb_eZeroDivError</div></td></tr>
<tr><td class="diagram"><div>SWIG_OverflowError</div></td><td><div>
rb_eRangeError</div></td></tr>
<tr><td class="diagram"><div>SWIG_SyntaxError</div></td><td><div>
rb_eSyntaxError</div></td></tr>
<tr><td class="diagram"><div>SWIG_ValueError</div></td><td><div>
rb_eArgError</div></td></tr>
<tr><td class="diagram"><div>SWIG_SystemError</div></td><td><div>
rb_eFatal</div></td></tr>
<tr><td class="diagram"><div>SWIG_AttributeError</div></td><td><div>
rb_eRuntimeError</div></td></tr>
<tr><td class="diagram"><div>SWIG_NullReferenceError</div></td><td><div>
rb_eNullReferenceError*</div></td></tr>
<tr><td class="diagram"><div>SWIG_ObjectPreviouslyDeletedError</div></td><td>
<div>rb_eObjectPreviouslyDeleted*</div></td></tr>
<tr><td class="diagram"><div>SWIG_UnknownError</div></td><td><div>
rb_eRuntimeError</div></td></tr>
<tr class="diagram"><td colspan="2"><div>* These error classes are
created by SWIG and are not built-in Ruby exception classes</div></td></tr>
</tbody></table>
</div>
<p>The second way to raise errors is to use <tt>SWIG_Raise(obj, type,
desc)</tt>. Obj is a C++ instance of an exception class, type is a
string specifying the type of exception (for example, &quot;MyError&quot;) and
desc is the SWIG description of the exception class. For example:</p>
<div class="code">
<pre>
%raise(SWIG_NewPointerObj(e, SWIGTYPE_p_AssertionFailedException, 0), &quot;:AssertionFailedException&quot;, SWIGTYPE_p_AssertionFailedException);
</pre>
</div>
<p>This is useful when you want to pass the current exception object
directly to Ruby, particularly when the object is an instance of class
marked as an <tt>%exceptionclass</tt> (see the next section for more
information).</p>
<p>Last, you can raise an exception by directly calling Ruby's C api.
This is done by invoking the <tt>rb_raise()</tt> function. The first
argument passed to <tt>rb_raise()</tt> is the exception type. You can
raise a custom exception type or one of the built-in Ruby exception
types.</p>
<h3><a name="Ruby_nn36">35.6.4 Exception classes</a></h3>
<p>Starting with SWIG 1.3.28, the Ruby module supports the <tt>
%exceptionclass</tt> directive, which is used to identify C++ classes
that are used as exceptions. Classes that are marked with the <tt>
%exceptionclass</tt> directive are exposed in Ruby as child classes of <tt>
rb_eRuntimeError</tt>. This allows C++ exceptions to be directly mapped
to Ruby exceptions, providing for a more natural integration between
C++ code and Ruby code.</p>
<div class="code">
<pre>%exceptionclass CustomError;
%inline %{
class CustomError { };
class Foo {
public:
void test() { throw CustomError; }
};
%}</pre>
</div>
<p>From Ruby you can now call this method like this:</p>
<div class="code targetlang">
<pre>foo = Foo.new
begin
foo.test()
rescue CustomError =&gt; e
puts &quot;Caught custom error&quot;
end </pre>
</div>
<p>For another example look at swig/Examples/ruby/exception_class.</p>
<h2><a name="Ruby_nn37">35.7 Typemaps</a></h2>
<p> This section describes how you can modify SWIG's default wrapping
behavior for various C/C++ datatypes using the <tt>%typemap</tt>
directive. This is an advanced topic that assumes familiarity with the
Ruby C API as well as the material in the &quot;<a href="#Typemaps">Typemaps</a>
&quot; chapter.</p>
<p>Before proceeding, it should be stressed that typemaps are not a
required part of using SWIG---the default wrapping behavior is enough
in most cases. Typemaps are only used if you want to change some aspect
of the primitive C-Ruby interface.</p>
<h3><a name="Ruby_nn38">35.7.1 What is a typemap?</a></h3>
<p> A typemap is nothing more than a code generation rule that is
attached to a specific C datatype. The general form of this declaration
is as follows ( parts enclosed in [...] are optional ):</p>
<div class="code">
<pre>
%typemap( method [, modifiers...] ) typelist code;
</pre>
</div>
<p><em> method</em> is a simply a name that specifies what kind of
typemap is being defined. It is usually a name like <tt>&quot;in&quot;</tt>, <tt>
&quot;out&quot;</tt>, or <tt>&quot;argout&quot;</tt> (or its director variations). The
purpose of these methods is described later.</p>
<p><em> modifiers</em> is an optional comma separated list of <tt>
name=&quot;value&quot;</tt> values. These are sometimes to attach extra
information to a typemap and is often target-language dependent.</p>
<p><em> typelist</em> is a list of the C++ type patterns that the
typemap will match. The general form of this list is as follows:</p>
<div class="diagram">
<pre>typelist : typepattern [, typepattern, typepattern, ... ] ;
typepattern : type [ (parms) ]
| type name [ (parms) ]
| ( typelist ) [ (parms) ]</pre>
</div>
<p> Each type pattern is either a simple type, a simple type and
argument name, or a list of types in the case of multi-argument
typemaps. In addition, each type pattern can be parameterized with a
list of temporary variables (parms). The purpose of these variables
will be explained shortly.</p>
<p><em>code</em> specifies the C code used in the typemap. It can take
any one of the following forms:</p>
<div class="diagram">
<pre>code : { ... }
| &quot; ... &quot;
| %{ ... %}</pre>
</div>
<p>For example, to convert integers from Ruby to C, you might define a
typemap like this:</p>
<div class="code">
<pre>%module example
%typemap(in) int {
$1 = (int) NUM2INT($input);
printf(&quot;Received an integer : %d\n&quot;, $1);
}
%inline %{
extern int fact(int n);
%}</pre>
</div>
<p> Typemaps are always associated with some specific aspect of code
generation. In this case, the &quot;in&quot; method refers to the conversion of
input arguments to C/C++. The datatype <tt>int</tt> is the datatype to
which the typemap will be applied. The supplied C code is used to
convert values. In this code a number of special variables prefaced by
a <tt>$</tt> are used. The <tt>$1</tt> variable is placeholder for a
local variable of type <tt>int</tt>. The <tt>$input</tt> variable is
the input Ruby object.</p>
<p>When this example is compiled into a Ruby module, the following
sample code:</p>
<div class="code targetlang">
<pre>require 'example'
puts Example.fact(6)</pre>
</div>
<p>prints the result:</p>
<div class="code shell">
<pre>
Received an integer : 6
720
</pre>
</div>
<p> In this example, the typemap is applied to all occurrences of the <tt>
int</tt> datatype. You can refine this by supplying an optional
parameter name. For example:</p>
<div class="code">
<pre>%module example
%typemap(in) int n {
$1 = (int) NUM2INT($input);
printf(&quot;n = %d\n&quot;, $1);
}
%inline %{
extern int fact(int n);
%}</pre>
</div>
<p> In this case, the typemap code is only attached to arguments that
exactly match &quot;<tt>int n</tt>&quot;.</p>
<p>The application of a typemap to specific datatypes and argument names
involves more than simple text-matching--typemaps are fully integrated
into the SWIG type-system. When you define a typemap for <tt>int</tt>,
that typemap applies to <tt>int</tt> and qualified variations such as <tt>
const int</tt>. In addition, the typemap system follows <tt>typedef</tt>
declarations. For example:</p>
<div class="code">
<pre>%typemap(in) int n {
$1 = (int) NUM2INT($input);
printf(&quot;n = %d\n&quot;, $1);
}
typedef int Integer;
extern int fact(Integer n); // Above typemap is applied</pre>
</div>
<p> However, the matching of <tt>typedef</tt> only occurs in one
direction. If you defined a typemap for <tt>Integer</tt>, it is not
applied to arguments of type <tt>int</tt>.</p>
<p>Typemaps can also be defined for groups of consecutive arguments. For
example:</p>
<div class="code">
<pre>%typemap(in) (char *str, int len) {
$1 = StringValuePtr($input);
$2 = (int) RSTRING($input)-&gt;len;
};
int count(char c, char *str, int len);</pre>
</div>
<p> When a multi-argument typemap is defined, the arguments are always
handled as a single Ruby object. This allows the function <tt>count</tt>
to be used as follows (notice how the length parameter is omitted):</p>
<div class="code targetlang">
<pre>puts Example.count('o', 'Hello World')
2</pre>
</div>
<h3><a name="Ruby_Typemap_scope">35.7.2 Typemap scope</a></h3>
<p> Once defined, a typemap remains in effect for all of the
declarations that follow. A typemap may be redefined for different
sections of an input file. For example:</p>
<div class="code">
<pre>// typemap1
%typemap(in) int {
...
}
int fact(int); // typemap1
int gcd(int x, int y); // typemap1
// typemap2
%typemap(in) int {
...
}
int isprime(int); // typemap2</pre>
</div>
<p> One exception to the typemap scoping rules pertains to the <tt>
%extend</tt> declaration. <tt>%extend</tt> is used to attach new
declarations to a class or structure definition. Because of this, all
of the declarations in an <tt>%extend</tt> block are subject to the
typemap rules that are in effect at the point where the class itself is
defined. For example:</p>
<div class="code">
<pre>class Foo {
...
};
%typemap(in) int {
...
}
%extend Foo {
int blah(int x); // typemap has no effect. Declaration is attached to Foo which
// appears before the %typemap declaration.
};</pre>
</div>
<h3><a name="Ruby_Copying_a_typemap">35.7.3 Copying a typemap</a></h3>
<p> A typemap is copied by using assignment. For example:</p>
<div class="code">
<pre>%typemap(in) Integer = int;</pre>
</div>
<p> or this:</p>
<div class="code">
<pre>%typemap(in) Integer, Number, int32_t = int;</pre>
</div>
<p> Types are often managed by a collection of different typemaps. For
example:</p>
<div class="code">
<pre>%typemap(in) int { ... }
%typemap(out) int { ... }
%typemap(varin) int { ... }
%typemap(varout) int { ... }</pre>
</div>
<p> To copy all of these typemaps to a new type, use <tt>%apply</tt>.
For example:</p>
<div class="code">
<pre>%apply int { Integer }; // Copy all int typemaps to Integer
%apply int { Integer, Number }; // Copy all int typemaps to both Integer and Number</pre>
</div>
<p> The patterns for <tt>%apply</tt> follow the same rules as for <tt>
%typemap</tt>. For example:</p>
<div class="code">
<pre>%apply int *output { Integer *output }; // Typemap with name
%apply (char *buf, int len) { (char *buffer, int size) }; // Multiple arguments</pre>
</div>
<h3><a name="Ruby_Deleting_a_typemap">35.7.4 Deleting a typemap</a></h3>
<p> A typemap can be deleted by simply defining no code. For example:</p>
<div class="code">
<pre>%typemap(in) int; // Clears typemap for int
%typemap(in) int, long, short; // Clears typemap for int, long, short
%typemap(in) int *output; </pre>
</div>
<p> The <tt>%clear</tt> directive clears all typemaps for a given type.
For example:</p>
<div class="code">
<pre>%clear int; // Removes all types for int
%clear int *output, long *output;</pre>
</div>
<p><b> Note:</b> Since SWIG's default behavior is defined by typemaps,
clearing a fundamental type like <tt>int</tt> will make that type
unusable unless you also define a new set of typemaps immediately after
the clear operation.</p>
<h3><a name="Ruby_Placement_of_typemaps">35.7.5 Placement of typemaps</a>
</h3>
<p> Typemap declarations can be declared in the global scope, within a
C++ namespace, and within a C++ class. For example:</p>
<div class="code">
<pre>%typemap(in) int {
...
}
namespace std {
class string;
%typemap(in) string {
...
}
}
class Bar {
public:
typedef const int &amp; const_reference;
%typemap(out) const_reference {
...
}
};</pre>
</div>
<p> When a typemap appears inside a namespace or class, it stays in
effect until the end of the SWIG input (just like before). However, the
typemap takes the local scope into account. Therefore, this code</p>
<div class="code">
<pre>namespace std {
class string;
%typemap(in) string {
...
}
}</pre>
</div>
<p> is really defining a typemap for the type <tt>std::string</tt>. You
could have code like this:</p>
<div class="code">
<pre>namespace std {
class string;
%typemap(in) string { /* std::string */
...
}
}
namespace Foo {
class string;
%typemap(in) string { /* Foo::string */
...
}
}</pre>
</div>
<p> In this case, there are two completely distinct typemaps that apply
to two completely different types (<tt>std::string</tt> and <tt>
Foo::string</tt>).</p>
<p> It should be noted that for scoping to work, SWIG has to know that <tt>
string</tt> is a typename defined within a particular namespace. In this
example, this is done using the class declaration <tt>class string</tt>
.</p>
<h3><a name="Ruby_nn39">35.7.6 Ruby typemaps</a></h3>
<p>The following list details all of the typemap methods that can be
used by the Ruby module:</p>
<h4><a name="Ruby_in_typemap">35.7.6.1 &quot;in&quot; typemap</a></h4>
<p>Converts Ruby objects to input function arguments. For example:</p>
<div class="code">
<pre>%typemap(in) int {
$1 = NUM2INT($input);
}</pre>
</div>
<p> The following special variables are available:</p>
<div class="diagram">
<table border="1" cellpadding="2" cellspacing="2" summary="Special variables - in typemap"
width="100%"><tbody>
<tr><td>$input</td><td> Input object holding value to be converted.</td></tr>
<tr><td>$symname</td><td> Name of function/method being wrapped</td></tr>
<tr><td>$1...n</td><td> Argument being sent to the function</td></tr>
<tr><td>$1_name</td><td> Name of the argument (if provided)</td></tr>
<tr><td>$1_type</td><td> The actual C datatype matched by the typemap.</td>
</tr>
<tr><td>$1_ltype</td><td> The assignable version of the C datatype
matched by the typemap.</td></tr>
</tbody></table>
</div>
<p> This is probably the most commonly redefined typemap because it can
be used to implement customized conversions.</p>
<p> In addition, the &quot;in&quot; typemap allows the number of converted
arguments to be specified. For example:</p>
<div class="code">
<pre>// Ignored argument.
%typemap(in, numinputs=0) int *out (int temp) {
$1 = &amp;temp;
}</pre>
</div>
<p> At this time, only zero or one arguments may be converted.</p>
<h4><a name="Ruby_typecheck_typemap">35.7.6.2 &quot;typecheck&quot; typemap</a></h4>
<p> The &quot;typecheck&quot; typemap is used to support overloaded functions and
methods. It merely checks an argument to see whether or not it matches
a specific type. For example:</p>
<div class="code">
<pre>%typemap(typecheck, precedence=SWIG_TYPECHECK_INTEGER) int {
$1 = FIXNUM_P($input) ? 1 : 0;
}</pre>
</div>
<p> For typechecking, the $1 variable is always a simple integer that is
set to 1 or 0 depending on whether or not the input argument is the
correct type.</p>
<p> If you define new &quot;in&quot; typemaps<em> and</em> your program uses
overloaded methods, you should also define a collection of &quot;typecheck&quot;
typemaps. More details about this follow in a later section on
&quot;Typemaps and Overloading.&quot;</p>
<h4><a name="Ruby_out_typemap">35.7.6.3 &quot;out&quot; typemap</a></h4>
<p>Converts return value of a C function to a Ruby object.</p>
<div class="code">
<pre>%typemap(out) int {
$result = INT2NUM( $1 );
}
</pre>
</div>
<p> The following special variables are available.</p>
<div class="diagram">
<table border="1" cellpadding="2" cellspacing="2" summary="Special variables - out typemap"
width="100%"><tbody>
<tr><td>$result</td><td> Result object returned to target language.</td></tr>
<tr><td>$symname</td><td> Name of function/method being wrapped</td></tr>
<tr><td>$1...n</td><td> Argument being wrapped</td></tr>
<tr><td>$1_name</td><td> Name of the argument (if provided)</td></tr>
<tr><td>$1_type</td><td> The actual C datatype matched by the typemap.</td>
</tr>
<tr><td>$1_ltype</td><td> The assignable version of the C datatype
matched by the typemap.</td></tr>
</tbody></table>
</div>
<h4><a name="Ruby_arginit_typemap">35.7.6.4 &quot;arginit&quot; typemap</a></h4>
<p> The &quot;arginit&quot; typemap is used to set the initial value of a function
argument--before any conversion has occurred. This is not normally
necessary, but might be useful in highly specialized applications. For
example:</p>
<div class="code">
<pre>// Set argument to NULL before any conversion occurs
%typemap(arginit) int *data {
$1 = NULL;
}</pre>
</div>
<h4><a name="Ruby_default_typemap">35.7.6.5 &quot;default&quot; typemap</a></h4>
<p> The &quot;default&quot; typemap is used to turn an argument into a default
argument. For example:</p>
<div class="code">
<pre>%typemap(default) int flags {
$1 = DEFAULT_FLAGS;
}
...
int foo(int x, int y, int flags);</pre>
</div>
<p> The primary use of this typemap is to either change the wrapping of
default arguments or specify a default argument in a language where
they aren't supported (like C). Target languages that do not support
optional arguments, such as Java and C#, effectively ignore the value
specified by this typemap as all arguments must be given.</p>
<p> Once a default typemap has been applied to an argument, all
arguments that follow must have default values. See the <a href="#SWIG_default_args">
Default/optional arguments</a> section for further information on
default argument wrapping.</p>
<h4><a name="Ruby_check_typemap">35.7.6.6 &quot;check&quot; typemap</a></h4>
<p> The &quot;check&quot; typemap is used to supply value checking code during
argument conversion. The typemap is applied<em> after</em> arguments
have been converted. For example:</p>
<div class="code">
<pre>%typemap(check) int positive {
if ($1 &lt;= 0) {
SWIG_exception(SWIG_ValueError, &quot;Expected positive value.&quot;);
}
}</pre>
</div>
<h4><a name="Ruby_argout_typemap_">35.7.6.7 &quot;argout&quot; typemap</a></h4>
<p> The &quot;argout&quot; typemap is used to return values from arguments. This
is most commonly used to write wrappers for C/C++ functions that need
to return multiple values. The &quot;argout&quot; typemap is almost always
combined with an &quot;in&quot; typemap---possibly to ignore the input value. For
example:</p>
<div class="code">
<pre>/* Set the input argument to point to a temporary variable */
%typemap(in, numinputs=0) int *out (int temp) {
$1 = &amp;temp;
}
%typemap(argout) int *out {
// Append output value $1 to $result (assuming a single integer in this case)
$result = SWIG_AppendOutput( $result, INT2NUM(*$1) );
}</pre>
</div>
<p> The following special variables are available.</p>
<div class="diagram">
<table border="1" cellpadding="2" cellspacing="2" summary="Special variables - argout typemap"
width="100%"><tbody>
<tr><td>$result</td><td> Result object returned to target language.</td></tr>
<tr><td>$input</td><td> The original input object passed.</td></tr>
<tr><td>$symname</td><td> Name of function/method being wrapped.</td></tr>
</tbody></table>
</div>
<p> The code supplied to the &quot;argout&quot; typemap is always placed after the
&quot;out&quot; typemap. If multiple return values are used, the extra return
values are often appended to return value of the function.</p>
<p>Output helper is a fragment that usually defines a macro to some
function like SWIG_Ruby_AppendOutput.</p>
<p> See the <tt>typemaps.i</tt> library for examples.</p>
<h4><a name="Ruby_freearg_typemap_">35.7.6.8 &quot;freearg&quot; typemap</a></h4>
<p> The &quot;freearg&quot; typemap is used to cleanup argument data. It is only
used when an argument might have allocated resources that need to be
cleaned up when the wrapper function exits. The &quot;freearg&quot; typemap
usually cleans up argument resources allocated by the &quot;in&quot; typemap. For
example:</p>
<div class="code">
<pre>// Get a list of integers
%typemap(in) int *items {
int nitems = Length($input);
$1 = (int *) malloc(sizeof(int)*nitems);
}
// Free the list
%typemap(freearg) int *items {
free($1);
}</pre>
</div>
<p> The &quot;freearg&quot; typemap inserted at the end of the wrapper function,
just before control is returned back to the target language. This code
is also placed into a special variable <tt>$cleanup</tt> that may be
used in other typemaps whenever a wrapper function needs to abort
prematurely.</p>
<h4><a name="Ruby_newfree_typemap">35.7.6.9 &quot;newfree&quot; typemap</a></h4>
<p> The &quot;newfree&quot; typemap is used in conjunction with the <tt>%newobject</tt>
directive and is used to deallocate memory used by the return result of
a function. For example:</p>
<div class="code">
<pre>%typemap(newfree) string * {
delete $1;
}
%typemap(out) string * {
$result = PyString_FromString($1-&gt;c_str());
}
...
%newobject foo;
...
string *foo();</pre>
</div>
<p> See <a href="#Customization_ownership">Object ownership and
%newobject</a> for further details.</p>
<h4><a name="Ruby_memberin_typemap">35.7.6.10 &quot;memberin&quot; typemap</a></h4>
<p> The &quot;memberin&quot; typemap is used to copy data from<em> an already
converted input value</em> into a structure member. It is typically
used to handle array members and other special cases. For example:</p>
<div class="code">
<pre>%typemap(memberin) int [4] {
memmove($1, $input, 4*sizeof(int));
}</pre>
</div>
<p> It is rarely necessary to write &quot;memberin&quot; typemaps---SWIG already
provides a default implementation for arrays, strings, and other
objects.</p>
<h4><a name="Ruby_varin_typemap">35.7.6.11 &quot;varin&quot; typemap</a></h4>
<p> The &quot;varin&quot; typemap is used to convert objects in the target
language to C for the purposes of assigning to a C/C++ global variable.
This is implementation specific.</p>
<h4><a name="Ruby_varout_typemap_">35.7.6.12 &quot;varout&quot; typemap</a></h4>
<p> The &quot;varout&quot; typemap is used to convert a C/C++ object to an object
in the target language when reading a C/C++ global variable. This is
implementation specific.</p>
<h4><a name="Ruby_throws_typemap">35.7.6.13 &quot;throws&quot; typemap</a></h4>
<p> The &quot;throws&quot; typemap is only used when SWIG parses a C++ method with
an exception specification or has the <tt>%catches</tt> feature
attached to the method. It provides a default mechanism for handling
C++ methods that have declared the exceptions they will throw. The
purpose of this typemap is to convert a C++ exception into an error or
exception in the target language. It is slightly different to the other
typemaps as it is based around the exception type rather than the type
of a parameter or variable. For example:</p>
<div class="code">
<pre>%typemap(throws) const char * %{
rb_raise(rb_eRuntimeError, $1);
SWIG_fail;
%}
void bar() throw (const char *);</pre>
</div>
<p> As can be seen from the generated code below, SWIG generates an
exception handler with the catch block comprising the &quot;throws&quot; typemap
content.</p>
<div class="code">
<pre>...
try {
bar();
}
catch(char const *_e) {
rb_raise(rb_eRuntimeError, _e);
SWIG_fail;
}
...</pre>
</div>
<p> Note that if your methods do not have an exception specification yet
they do throw exceptions, SWIG cannot know how to deal with them. For a
neat way to handle these, see the <a href="#Customization_exception">
Exception handling with %exception</a> section.</p>
<h4><a name="Ruby_directorin_typemap">35.7.6.14 directorin typemap</a></h4>
<p>Converts C++ objects in director member functions to ruby objects. It
is roughly the opposite of the &quot;in&quot; typemap, making its typemap rule
often similar to the &quot;out&quot; typemap.</p>
<div class="code">
<pre>
%typemap(directorin) int {
$result = INT2NUM($1);
}
</pre>
</div>
<p> The following special variables are available.</p>
<div class="diagram">
<table border="1" cellpadding="2" cellspacing="2" summary="Special variables - directorin typemap"
width="100%"><tbody>
<tr><td>$result</td><td> Result object returned to target language.</td></tr>
<tr><td>$symname</td><td> Name of function/method being wrapped</td></tr>
<tr><td>$1...n</td><td> Argument being wrapped</td></tr>
<tr><td>$1_name</td><td> Name of the argument (if provided)</td></tr>
<tr><td>$1_type</td><td> The actual C datatype matched by the typemap.</td>
</tr>
<tr><td>$1_ltype</td><td> The assignable version of the C datatype
matched by the typemap.</td></tr>
<tr><td>this</td><td> C++ this, referring to the class itself.</td></tr>
</tbody></table>
</div>
<h4><a name="Ruby_directorout_typemap">35.7.6.15 directorout typemap</a></h4>
<p>Converts Ruby objects in director member functions to C++ objects. It
is roughly the opposite of the &quot;out&quot; typemap, making its rule often
similar to the &quot;in&quot; typemap.</p>
<div class="code">
<pre>
%typemap(directorout) int {
$result = NUM2INT($1);
}
</pre>
</div>
<p> The following special variables are available:</p>
<div class="diagram">
<table border="1" cellpadding="2" cellspacing="2" summary="Special variables - directorout typemap"
width="100%"><tbody>
<tr><td>$input</td><td>Ruby object being sent to the function</td></tr>
<tr><td>$symname</td><td>Name of function/method being wrapped</td></tr>
<tr><td>$1...n</td><td>Argument being sent to the function</td></tr>
<tr><td>$1_name</td><td> Name of the argument (if provided)</td></tr>
<tr><td>$1_type</td><td> The actual C datatype matched by the typemap.</td>
</tr>
<tr><td>$1_ltype</td><td> The assignable version of the C datatype
matched by the typemap.</td></tr>
<tr><td>this</td><td> C++ this, referring to the class itself.</td></tr>
</tbody></table>
</div>
<p>Currently, the directorout nor the out typemap support the option <tt>
numoutputs</tt>, but the Ruby module provides that functionality through
a %feature directive. Thus, a function can be made to return &quot;nothing&quot;
if you do:</p>
<div class="code">
<pre>
%feature(&quot;numoutputs&quot;, &quot;0&quot;) MyClass::function;
</pre>
</div>
<p>This feature can be useful if a function returns a status code, which
you want to discard but still use the typemap to raise an exception.</p>
<h4><a name="Ruby_directorargout_typemap">35.7.6.16 directorargout
typemap</a></h4>
<p>Output argument processing in director member functions.</p>
<div class="code">
<pre>
%typemap(directorargout) int {
$result = SWIG_AppendOutput( $result, NUM2INT($1) );
}
</pre>
</div>
<p> The following special variables are available:</p>
<div class="diagram">
<table border="1" cellpadding="2" cellspacing="2" style="text-align: left; width: 100%;"
summary="Special variables - directorargout typemap"><tbody>
<tr><td>$result</td><td>Result that the director function returns</td></tr>
<tr><td>$input</td><td>Ruby object being sent to the function</td></tr>
<tr><td>$symname</td><td>name of the function/method being wrapped</td></tr>
<tr><td>$1...n</td><td>Argument being sent to the function</td></tr>
<tr><td>$1_name</td><td>Name of the argument (if provided)</td></tr>
<tr><td>$1_type</td><td>The actual C datatype matched by the typemap</td>
</tr>
<tr><td>$1_ltype</td><td>The assignable version of the C datatype
matched by the typemap</td></tr>
<tr><td>this</td><td>C++ this, referring to the instance of the class
itself</td></tr>
</tbody></table>
</div>
<h4><a name="Ruby_ret_typemap">35.7.6.17 ret typemap</a></h4>
<p>Cleanup of function return values</p>
<h4><a name="Ruby_globalin_typemap">35.7.6.18 globalin typemap</a></h4>
<p>Setting of C global variables</p>
<h3><a name="Ruby_nn40">35.7.7 Typemap variables</a></h3>
<p> Within a typemap, a number of special variables prefaced with a <tt>
$</tt> may appear. A full list of variables can be found in the &quot;<a href="#Typemaps">
Typemaps</a>&quot; chapter. This is a list of the most common variables:</p>
<p><tt>$1</tt></p>
<div class="indent">A C local variable corresponding to the actual type
specified in the <tt>%typemap</tt> directive. For input values, this is
a C local variable that is supposed to hold an argument value. For
output values, this is the raw result that is supposed to be returned
to Ruby.</div>
<p><tt>$input</tt></p>
<div class="indent">A <tt>VALUE</tt> holding a raw Ruby object with an
argument or variable value.</div>
<p><tt>$result</tt></p>
<div class="indent">A <tt>VALUE</tt> that holds the result to be
returned to Ruby.</div>
<p><tt>$1_name</tt></p>
<div class="indent">The parameter name that was matched.</div>
<p><tt>$1_type</tt></p>
<div class="indent">The actual C datatype matched by the typemap.</div>
<p><tt>$1_ltype</tt></p>
<div class="indent">An assignable version of the datatype matched by the
typemap (a type that can appear on the left-hand-side of a C assignment
operation). This type is stripped of qualifiers and may be an altered
version of <tt>$1_type</tt>. All arguments and local variables in
wrapper functions are declared using this type so that their values can
be properly assigned.</div>
<p><tt>$symname</tt></p>
<div class="indent">The Ruby name of the wrapper function being created.</div>
<h3><a name="Ruby_nn41">35.7.8 Useful Functions</a></h3>
<p> When you write a typemap, you usually have to work directly with
Ruby objects. The following functions may prove to be useful. (These
functions plus many more can be found in<em> Programming Ruby</em>
book, by David Thomas and Andrew Hunt.)</p>
<p>In addition, we list equivalent functions that SWIG defines, which
provide a language neutral conversion (these functions are defined for
each swig language supported). If you are trying to create a swig file
that will work under multiple languages, it is recommended you stick to
the swig functions instead of the native Ruby functions. That should
help you avoid having to rewrite a lot of typemaps across multiple
languages.</p>
<h4><a name="Ruby_nn42">35.7.8.1 C Datatypes to Ruby Objects</a></h4>
<div class="diagram">
<table border="1" cellpadding="2" cellspacing="2" style="width: 100%;" summary="Datatypes">
<tbody>
<tr><th><b>RUBY</b></th><th><b>SWIG</b></th><td></td></tr>
<tr><td>INT2NUM(long or int)</td><td>SWIG_From_int(int x)</td><td> int
to Fixnum or Bignum</td></tr>
<tr><td>INT2FIX(long or int)</td><td></td><td> int to Fixnum (faster
than INT2NUM)</td></tr>
<tr><td>CHR2FIX(char)</td><td>SWIG_From_char(char x)</td><td> char to
Fixnum</td></tr>
<tr><td>rb_str_new2(char*)</td><td>SWIG_FromCharPtrAndSize(char*,
size_t)</td><td> char* to String</td></tr>
<tr><td>rb_float_new(double)</td><td>SWIG_From_double(double),
<br> SWIG_From_float(float)</td><td>float/double to Float</td></tr>
</tbody></table>
</div>
<h4><a name="Ruby_nn43">35.7.8.2 Ruby Objects to C Datatypes</a></h4>
<p>Here, while the Ruby versions return the value directly, the SWIG
versions do not, but return a status value to indicate success (<tt>
SWIG_OK</tt>). While more awkward to use, this allows you to write
typemaps that report more helpful error messages, like:</p>
<div class="code">
<pre>
%typemap(in) size_t (int ok)
ok = SWIG_AsVal_size_t($input, &amp;$1);
if (!SWIG_IsOK(ok)) {
SWIG_exception_fail(SWIG_ArgError(ok), Ruby_Format_TypeError( &quot;$1_name&quot;, &quot;$1_type&quot;, &quot;$symname&quot;, $argnum, $input));
}
}
</pre>
</div><div class="diagram">
<table border="1" cellpadding="2" cellspacing="2" summary="Ruby objects" width="100%">
<tbody>
<tr><td>int NUM2INT(Numeric)</td><td>SWIG_AsVal_int(VALUE, int*)</td></tr>
<tr><td>int FIX2INT(Numeric)</td><td>SWIG_AsVal_int(VALUE, int*)</td></tr>
<tr><td>unsigned int NUM2UINT(Numeric)</td><td>
SWIG_AsVal_unsigned_SS_int(VALUE, int*)</td></tr>
<tr><td>unsigned int FIX2UINT(Numeric)</td><td>
SWIG_AsVal_unsigned_SS_int(VALUE, int*)</td></tr>
<tr><td>long NUM2LONG(Numeric)</td><td>SWIG_AsVal_long(VALUE, long*)</td>
</tr>
<tr><td>long FIX2LONG(Numeric)</td><td>SWIG_AsVal_long(VALUE, long*)</td>
</tr>
<tr><td>unsigned long FIX2ULONG(Numeric)</td><td>
SWIG_AsVal_unsigned_SS_long(VALUE, unsigned long*)</td></tr>
<tr><td>char NUM2CHR(Numeric or String)</td><td>SWIG_AsVal_char(VALUE,
int*)</td></tr>
<tr><td>char * StringValuePtr(String)</td><td>
SWIG_AsCharPtrAndSize(VALUE, char**, size_t*, int* alloc)</td></tr>
<tr><td>char * rb_str2cstr(String, int*length)</td><td></td></tr>
<tr><td>double NUM2DBL(Numeric)</td><td>(double) SWIG_AsVal_int(VALUE)
or similar</td></tr>
</tbody></table>
</div>
<h4><a name="Ruby_nn44">35.7.8.3 Macros for VALUE</a></h4>
<p> <tt>RSTRING_LEN(str)</tt></p>
<div class="indent">length of the Ruby string</div>
<p><tt>RSTRING_PTR(str)</tt></p>
<div class="indent">pointer to string storage</div>
<p><tt>RARRAY_LEN(arr)</tt></p>
<div class="indent">length of the Ruby array</div>
<p><tt>RARRAY(arr)-&gt;capa</tt></p>
<div class="indent">capacity of the Ruby array</div>
<p><tt>RARRAY_PTR(arr)</tt></p>
<div class="indent">pointer to array storage</div>
<h4><a name="Ruby_nn45">35.7.8.4 Exceptions</a></h4>
<p> <tt>void rb_raise(VALUE exception, const char *fmt, ...)</tt></p>
<div class="indent"> Raises an exception. The given format string<i> fmt</i>
and remaining arguments are interpreted as with <tt>printf()</tt>.</div>
<p><tt>void rb_fatal(const char *fmt, ...)</tt></p>
<div class="indent"> Raises a fatal exception, terminating the process.
No rescue blocks are called, but ensure blocks will be called. The
given format string<i> fmt</i> and remaining arguments are interpreted
as with <tt>printf()</tt>.</div>
<p><tt>void rb_bug(const char *fmt, ...)</tt></p>
<div class="indent"> Terminates the process immediately -- no handlers
of any sort will be called. The given format string<i> fmt</i> and
remaining arguments are interpreted as with <tt>printf()</tt>. You
should call this function only if a fatal bug has been exposed.</div>
<p><tt>void rb_sys_fail(const char *msg)</tt></p>
<div class="indent"> Raises a platform-specific exception corresponding
to the last known system error, with the given string<i> msg</i>.</div>
<p><tt>VALUE rb_rescue(VALUE (*body)(VALUE), VALUE args,
VALUE(*rescue)(VALUE, VALUE), VALUE rargs)</tt></p>
<div class="indent"> Executes<i> body</i> with the given<i> args</i>. If
a <tt>StandardError</tt> exception is raised, then execute<i> rescue</i>
with the given<i> rargs</i>.</div>
<p><tt>VALUE rb_ensure(VALUE(*body)(VALUE), VALUE args,
VALUE(*ensure)(VALUE), VALUE eargs)</tt></p>
<div class="indent"> Executes<i> body</i> with the given<i> args</i>.
Whether or not an exception is raised, execute<i> ensure</i> with the
given<i> rargs</i> after<i> body</i> has completed.</div>
<p><tt>VALUE rb_protect(VALUE (*body)(VALUE), VALUE args, int *result)</tt>
</p>
<div class="indent"> Executes<i> body</i> with the given<i> args</i> and
returns nonzero in result if any exception was raised.</div>
<p><tt>void rb_notimplement()</tt></p>
<div class="indent"> Raises a <tt>NotImpError</tt> exception to indicate
that the enclosed function is not implemented yet, or not available on
this platform.</div>
<p><tt>void rb_exit(int status)</tt></p>
<div class="indent"> Exits Ruby with the given<i> status</i>. Raises a <tt>
SystemExit</tt> exception and calls registered exit functions and
finalizers.</div>
<p><tt>void rb_warn(const char *fmt, ...)</tt></p>
<div class="indent"> Unconditionally issues a warning message to
standard error. The given format string<i> fmt</i> and remaining
arguments are interpreted as with <tt>printf()</tt>.</div>
<p><tt>void rb_warning(const char *fmt, ...)</tt></p>
<div class="indent"> Conditionally issues a warning message to standard
error if Ruby was invoked with the <tt>-w</tt> flag. The given format
string<i> fmt</i> and remaining arguments are interpreted as with <tt>
printf()</tt>.</div>
<h4><a name="Ruby_nn46">35.7.8.5 Iterators</a></h4>
<p> <tt>void rb_iter_break()</tt></p>
<div class="indent"> Breaks out of the enclosing iterator block.</div>
<p><tt>VALUE rb_each(VALUE obj)</tt></p>
<div class="indent"> Invokes the <tt>each</tt> method of the given<i>
obj</i>.</div>
<p><tt>VALUE rb_yield(VALUE arg)</tt></p>
<div class="indent"> Transfers execution to the iterator block in the
current context, passing<i> arg</i> as an argument. Multiple values may
be passed in an array.</div>
<p><tt>int rb_block_given_p()</tt></p>
<div class="indent"> Returns <tt>true</tt> if <tt>yield</tt> would
execute a block in the current context; that is, if a code block was
passed to the current method and is available to be called.</div>
<p><tt>VALUE rb_iterate(VALUE (*method)(VALUE), VALUE args, VALUE
(*block)(VALUE, VALUE), VALUE arg2)</tt></p>
<div class="indent"> Invokes<i> method</i> with argument<i> args</i> and
block<i> block</i>. A <tt>yield</tt> from that method will invoke<i>
block</i> with the argument given to <tt>yield</tt>, and a second
argument<i> arg2</i>.</div>
<p><tt>VALUE rb_catch(const char *tag, VALUE (*proc)(VALUE, VALUE),
VALUE value)</tt></p>
<div class="indent"> Equivalent to Ruby's <tt>catch</tt>.</div>
<p><tt>void rb_throw(const char *tag, VALUE value)</tt></p>
<div class="indent"> Equivalent to Ruby's <tt>throw</tt>.</div>
<h3><a name="Ruby_nn47">35.7.9 Typemap Examples</a></h3>
<p> This section includes a few examples of typemaps. For more examples,
you might look at the examples in the <tt>Example/ruby</tt> directory.</p>
<h3><a name="Ruby_nn48">35.7.10 Converting a Ruby array to a char **</a></h3>
<p> A common problem in many C programs is the processing of command
line arguments, which are usually passed in an array of <tt>NULL</tt>
terminated strings. The following SWIG interface file allows a Ruby
Array instance to be used as a <tt>char **</tt> object.</p>
<div class="code">
<pre>%module argv
// This tells SWIG to treat char ** as a special case
%typemap(in) char ** {
/* Get the length of the array */
int size = RARRAY($input)-&gt;len;
int i;
$1 = (char **) malloc((size+1)*sizeof(char *));
/* Get the first element in memory */
VALUE *ptr = RARRAY($input)-&gt;ptr;
for (i=0; i &lt; size; i++, ptr++) {
/* Convert Ruby Object String to char* */
$1[i]= StringValuePtr(*ptr);
}
$1[i]=NULL; /* End of list */
}
// This cleans up the char ** array created before
// the function call
%typemap(freearg) char ** {
free((char *) $1);
}
// Now a test function
%inline %{
int print_args(char **argv) {
int i = 0;
while (argv[i]) {
printf(&quot;argv[%d] = %s\n&quot;, i, argv[i]);
i++;
}
return i;
}
%}</pre>
</div>
<p> When this module is compiled, the wrapped C function now operates as
follows :</p>
<div class="code targetlang">
<pre>require 'Argv'
Argv.print_args([&quot;Dave&quot;, &quot;Mike&quot;, &quot;Mary&quot;, &quot;Jane&quot;, &quot;John&quot;])
argv[0] = Dave
argv[1] = Mike
argv[2] = Mary
argv[3] = Jane
argv[4] = John</pre>
</div>
<p> In the example, two different typemaps are used. The &quot;in&quot; typemap is
used to receive an input argument and convert it to a C array. Since
dynamic memory allocation is used to allocate memory for the array, the
&quot;freearg&quot; typemap is used to later release this memory after the
execution of the C function.</p>
<h3><a name="Ruby_nn49">35.7.11 Collecting arguments in a hash</a></h3>
<p> Ruby's solution to the &quot;keyword arguments&quot; capability of some other
languages is to allow the programmer to pass in one or more key-value
pairs as arguments to a function. All of those key-value pairs are
collected in a single <tt>Hash</tt> argument that's presented to the
function. If it makes sense, you might want to provide similar
functionality for your Ruby interface. For example, suppose you'd like
to wrap this C function that collects information about people's vital
statistics:</p>
<div class="code">
<pre>void setVitalStats(const char *person, int nattributes, const char **names, int *values);</pre>
</div>
<p> and you'd like to be able to call it from Ruby by passing in an
arbitrary number of key-value pairs as inputs, e.g.</p>
<div class="code targetlang">
<pre>setVitalStats(&quot;Fred&quot;,
'weight' =&gt; 270,
'age' =&gt; 42
)</pre>
</div>
<p> To make this work, you need to write a typemap that expects a Ruby <tt>
Hash</tt> as its input and somehow extracts the last three arguments (<i>
nattributes</i>,<i> names</i> and<i> values</i>) needed by your C
function. Let's start with the basics:</p>
<div class="code">
<pre>%typemap(in) (int nattributes, const char **names, const int *values)
(VALUE keys_arr, int i, VALUE key, VALUE val) {
}
</pre>
</div>
<p> This <tt>%typemap</tt> directive tells SWIG that we want to match
any function declaration that has the specified types and names of
arguments somewhere in the argument list. The fact that we specified
the argument names (<i>nattributes</i>,<i> names</i> and<i> values</i>)
in our typemap is significant; this ensures that SWIG won't try to
apply this typemap to<i> other</i> functions it sees that happen to
have a similar declaration with different argument names. The arguments
that appear in the second set of parentheses (<i>keys_arr</i>,<i> i</i>
,<i> key</i> and<i> val</i>) define local variables that our typemap
will need.</p>
<p>Since we expect the input argument to be a <tt>Hash</tt>, let's next
add a check for that:</p>
<div class="code">
<pre>%typemap(in) (int nattributes, const char **names, const int *values)
(VALUE keys_arr, int i, VALUE key, VALUE val) {
<b>Check_Type($input, T_HASH);</b>
}</pre>
</div>
<p> <tt>Check_Type()</tt> is just a macro (defined in the Ruby header
files) that confirms that the input argument is of the correct type; if
it isn't, an exception will be raised.</p>
<p>The next task is to determine how many key-value pairs are present in
the hash; we'll assign this number to the first typemap argument (<tt>
$1</tt>). This is a little tricky since the Ruby/C API doesn't provide a
public function for querying the size of a hash, but we can get around
that by calling the hash's<i> size</i> method directly and converting
its result to a C <tt>int</tt> value:</p>
<div class="code">
<pre>%typemap(in) (int nattributes, const char **names, const int *values)
(VALUE keys_arr, int i, VALUE key, VALUE val) {
Check_Type($input, T_HASH);
<b>$1 = NUM2INT(rb_funcall($input, rb_intern(&quot;size&quot;), 0, Qnil));</b>
}</pre>
</div>
<p> So now we know the number of attributes. Next we need to initialize
the second and third typemap arguments (i.e. the two C arrays) to <tt>
NULL</tt> and set the stage for extracting the keys and values from the
hash:</p>
<div class="code">
<pre>%typemap(in) (int nattributes, const char **names, const int *values)
(VALUE keys_arr, int i, VALUE key, VALUE val) {
Check_Type($input, T_HASH);
$1 = NUM2INT(rb_funcall($input, rb_intern(&quot;size&quot;), 0, Qnil));
<b>$2 = NULL;
$3 = NULL;
if ($1 &gt; 0) {
$2 = (char **) malloc($1*sizeof(char *));
$3 = (int *) malloc($1*sizeof(int));
}</b>
}</pre>
</div>
<p> There are a number of ways we could extract the keys and values from
the input hash, but the simplest approach is to first call the hash's<i>
keys</i> method (which returns a Ruby array of the keys) and then start
looping over the elements in that array:</p>
<div class="code">
<pre>%typemap(in) (int nattributes, const char **names, const int *values)
(VALUE keys_arr, int i, VALUE key, VALUE val) {
Check_Type($input, T_HASH);
$1 = NUM2INT(rb_funcall($input, rb_intern(&quot;size&quot;), 0, Qnil));
$2 = NULL;
$3 = NULL;
if ($1 &gt; 0) {
$2 = (char **) malloc($1*sizeof(char *));
$3 = (int *) malloc($1*sizeof(int));
<b>keys_arr = rb_funcall($input, rb_intern(&quot;keys&quot;), 0, Qnil);
for (i = 0; i &lt; $1; i++) {
}</b>
}
}</pre>
</div>
<p> Recall that<i> keys_arr</i> and<i> i</i> are local variables for
this typemap. For each element in the<i> keys_arr</i> array, we want to
get the key itself, as well as the value corresponding to that key in
the hash:</p>
<div class="code">
<pre>%typemap(in) (int nattributes, const char **names, const int *values)
(VALUE keys_arr, int i, VALUE key, VALUE val) {
Check_Type($input, T_HASH);
$1 = NUM2INT(rb_funcall($input, rb_intern(&quot;size&quot;), 0, Qnil));
$2 = NULL;
$3 = NULL;
if ($1 &gt; 0) {
$2 = (char **) malloc($1*sizeof(char *));
$3 = (int *) malloc($1*sizeof(int));
keys_arr = rb_funcall($input, rb_intern(&quot;keys&quot;), 0, Qnil);
for (i = 0; i &lt; $1; i++) {
<b>key = rb_ary_entry(keys_arr, i);
val = rb_hash_aref($input, key);</b>
}
}
}</pre>
</div>
<p> To be safe, we should again use the <tt>Check_Type()</tt> macro to
confirm that the key is a <tt>String</tt> and the value is a <tt>Fixnum</tt>
:</p>
<div class="code">
<pre>%typemap(in) (int nattributes, const char **names, const int *values)
(VALUE keys_arr, int i, VALUE key, VALUE val) {
Check_Type($input, T_HASH);
$1 = NUM2INT(rb_funcall($input, rb_intern(&quot;size&quot;), 0, Qnil));
$2 = NULL;
$3 = NULL;
if ($1 &gt; 0) {
$2 = (char **) malloc($1*sizeof(char *));
$3 = (int *) malloc($1*sizeof(int));
keys_arr = rb_funcall($input, rb_intern(&quot;keys&quot;), 0, Qnil);
for (i = 0; i &lt; $1; i++) {
key = rb_ary_entry(keys_arr, i);
val = rb_hash_aref($input, key);
<b>Check_Type(key, T_STRING);
Check_Type(val, T_FIXNUM);</b>
}
}
}</pre>
</div>
<p> Finally, we can convert these Ruby objects into their C equivalents
and store them in our local C arrays:</p>
<div class="code">
<pre>%typemap(in) (int nattributes, const char **names, const int *values)
(VALUE keys_arr, int i, VALUE key, VALUE val) {
Check_Type($input, T_HASH);
$1 = NUM2INT(rb_funcall($input, rb_intern(&quot;size&quot;), 0, Qnil));
$2 = NULL;
$3 = NULL;
if ($1 &gt; 0) {
$2 = (char **) malloc($1*sizeof(char *));
$3 = (int *) malloc($1*sizeof(int));
keys_arr = rb_funcall($input, rb_intern(&quot;keys&quot;), 0, Qnil);
for (i = 0; i &lt; $1; i++) {
key = rb_ary_entry(keys_arr, i);
val = rb_hash_aref($input, key);
Check_Type(key, T_STRING);
Check_Type(val, T_FIXNUM);
<b>$2[i] = StringValuePtr(key);
$3[i] = NUM2INT(val);</b>
}
}
}</pre>
</div>
<p> We're not done yet. Since we used <tt>malloc()</tt> to dynamically
allocate the memory used for the<i> names</i> and<i> values</i>
arguments, we need to provide a corresponding &quot;freearg&quot; typemap to free
that memory so that there is no memory leak. Fortunately, this typemap
is a lot easier to write:</p>
<div class="code">
<pre>%typemap(freearg) (int nattributes, const char **names, const int *values) {
free((void *) $2);
free((void *) $3);
}</pre>
</div>
<p> All of the code for this example, as well as a sample Ruby program
that uses the extension, can be found in the <tt>Examples/ruby/hashargs</tt>
directory of the SWIG distribution.</p>
<h3><a name="Ruby_nn50">35.7.12 Pointer handling</a></h3>
<p> Occasionally, it might be necessary to convert pointer values that
have been stored using the SWIG typed-pointer representation. Since
there are several ways in which pointers can be represented, the
following two functions are used to safely perform this conversion:</p>
<p><tt>int SWIG_ConvertPtr(VALUE obj, void **ptr, swig_type_info *ty,
int flags)</tt></p>
<div class="indent">Converts a Ruby object<i> obj</i> to a C pointer
whose address is<i> ptr</i> (i.e.<i> ptr</i> is a pointer to a
pointer). The third argument,<i> ty</i>, is a pointer to a SWIG type
descriptor structure. If<i> ty</i> is not <tt>NULL</tt>, that type
information is used to validate type compatibility and other aspects of
the type conversion. If<i> flags</i> is non-zero, any type errors
encountered during this validation result in a Ruby <tt>TypeError</tt>
exception being raised; if<i> flags</i> is zero, such type errors will
cause <tt>SWIG_ConvertPtr()</tt> to return -1 but not raise an
exception. If<i> ty</i> is <tt>NULL</tt>, no type-checking is
performed.</div>
<p> <tt>VALUE SWIG_NewPointerObj(void *ptr, swig_type_info *ty, int own)</tt>
</p>
<div class="indent">Creates a new Ruby pointer object. Here,<i> ptr</i>
is the pointer to convert,<i> ty</i> is the SWIG type descriptor
structure that describes the type, and<i> own</i> is a flag that
indicates whether or not Ruby should take ownership of the pointer
(i.e. whether Ruby should free this data when the corresponding Ruby
instance is garbage-collected).</div>
<p> Both of these functions require the use of a special SWIG
type-descriptor structure. This structure contains information about
the mangled name of the datatype, type-equivalence information, as well
as information about converting pointer values under C++ inheritance.
For a type of <tt>Foo *</tt>, the type descriptor structure is usually
accessed as follows:</p>
<div class="indent code">
<pre>Foo *foo;
SWIG_ConvertPtr($input, (void **) &amp;foo, SWIGTYPE_p_Foo, 1);
VALUE obj;
obj = SWIG_NewPointerObj(f, SWIGTYPE_p_Foo, 0);</pre>
</div>
<p> In a typemap, the type descriptor should always be accessed using
the special typemap variable <tt>$1_descriptor</tt>. For example:</p>
<div class="indent code">
<pre>%typemap(in) Foo * {
SWIG_ConvertPtr($input, (void **) &amp;$1, $1_descriptor, 1);
}</pre>
</div>
<h4><a name="Ruby_nn51">35.7.12.1 Ruby Datatype Wrapping</a></h4>
<p> <tt>VALUE Data_Wrap_Struct(VALUE class, void (*mark)(void *), void
(*free)(void *), void *ptr)</tt></p>
<div class="indent">Given a pointer<i> ptr</i> to some C data, and the
two garbage collection routines for this data (<i>mark</i> and<i> free</i>
), return a <tt>VALUE</tt> for the Ruby object.</div>
<p><tt>VALUE Data_Make_Struct(VALUE class,<i> c-type</i>, void
(*mark)(void *), void (*free)(void *),<i> c-type</i> *ptr)</tt></p>
<div class="indent">Allocates a new instance of a C data type<i> c-type</i>
, assigns it to the pointer<i> ptr</i>, then wraps that pointer with <tt>
Data_Wrap_Struct()</tt> as above.</div>
<p><tt>Data_Get_Struct(VALUE obj,<i> c-type</i>,<i> c-type</i> *ptr)</tt>
</p>
<div class="indent">Retrieves the original C pointer of type<i> c-type</i>
from the data object<i> obj</i> and assigns that pointer to<i> ptr</i>.</div>
<h3><a name="Ruby_nn52">35.7.13 Example: STL Vector to Ruby Array</a></h3>
<p>Another use for macros and type maps is to create a Ruby array from a
STL vector of pointers. In essence, copy of all the pointers in the
vector into a Ruby array. The use of the macro is to make the typemap
so generic that any vector with pointers can use the type map. The
following is an example of how to construct this type of macro/typemap
and should give insight into constructing similar typemaps for other
STL structures:</p>
<div class="code">
<pre>%define PTR_VECTOR_TO_RUBY_ARRAY(vectorclassname, classname)
%typemap(out) vectorclassname &amp;, const vectorclassname &amp; {
VALUE arr = rb_ary_new2($1-&gt;size());
vectorclassname::iterator i = $1-&gt;begin(), iend = $1-&gt;end();
for ( ; i!=iend; i++ )
rb_ary_push(arr, Data_Wrap_Struct(c ## classname.klass, 0, 0, *i));
$result = arr;
}
%typemap(out) vectorclassname, const vectorclassname {
VALUE arr = rb_ary_new2($1.size());
vectorclassname::iterator i = $1.begin(), iend = $1.end();
for ( ; i!=iend; i++ )
rb_ary_push(arr, Data_Wrap_Struct(c ## classname.klass, 0, 0, *i));
$result = arr;
}
%enddef</pre>
</div>
<p> Note, that the &quot;<tt>c ## classname.klass&quot;</tt> is used in the
preprocessor step to determine the actual object from the class name.</p>
<p>To use the macro with a class Foo, the following is used:</p>
<div class="code">
<pre>PTR_VECTOR_TO_RUBY_ARRAY(vector&lt;foo *=&quot;&quot;&gt;, Foo)</pre>
</div>
<p> It is also possible to create a STL vector of Ruby objects:</p>
<div class="code">
<pre>%define RUBY_ARRAY_TO_PTR_VECTOR(vectorclassname, classname)
%typemap(in) vectorclassname &amp;, const vectorclassname &amp; {
Check_Type($input, T_ARRAY);
vectorclassname *vec = new vectorclassname;
int len = RARRAY($input)-&gt;len;
for (int i=0; i!=len; i++) {
VALUE inst = rb_ary_entry($input, i);
//The following _should_ work but doesn't on HPUX
// Check_Type(inst, T_DATA);
classname *element = NULL;
Data_Get_Struct(inst, classname, element);
vec-&gt;push_back(element);
}
$1 = vec;
}
%typemap(freearg) vectorclassname &amp;, const vectorclassname &amp; {
delete $1;
}
%enddef</pre>
</div>
<p> It is also possible to create a Ruby array from a vector of static
data types:</p>
<div class="code">
<pre>%define VECTOR_TO_RUBY_ARRAY(vectorclassname, classname)
%typemap(out) vectorclassname &amp;, const vectorclassname &amp; {
VALUE arr = rb_ary_new2($1-&gt;size());
vectorclassname::iterator i = $1-&gt;begin(), iend = $1-&gt;end();
for ( ; i!=iend; i++ )
rb_ary_push(arr, Data_Wrap_Struct(c ## classname.klass, 0, 0, &amp;(*i)));
$result = arr;
}
%typemap(out) vectorclassname, const vectorclassname {
VALUE arr = rb_ary_new2($1.size());
vectorclassname::iterator i = $1.begin(), iend = $1.end();
for ( ; i!=iend; i++ )
rb_ary_push(arr, Data_Wrap_Struct(c ## classname.klass, 0, 0, &amp;(*i)));
$result = arr;
}
%enddef</pre>
</div> Note that this is mostly an example of typemaps. If you want to
use the STL with ruby, you are advised to use the standard swig STL
library, which does much more than this. Refer to the section called
the<a href="#Ruby_nn23_1"> C++ Standard Template Library</a>.
<h2><a name="Ruby_nn65">35.8 Docstring Features</a></h2>
<p> Using ri and rdoc web pages in Ruby libraries is a common practice.
Given the way that SWIG generates the extensions by default, your users
will normally not get any documentation for it, even if they run 'rdoc'
on the resulting .c or .cxx file.</p>
<p>The features described in this section make it easy for you to add
rdoc strings to your modules, functions and methods that can then be
read by Ruby's rdoc tool to generate html web pages, ri documentation,
Windows chm file and an .xml description.</p>
<p>rdoc can then be run from a console or shell window on a swig
generated file.</p>
<p>For example, to generate html web pages from a C++ file, you'd do:</p>
<div class="code shell">
<pre>
$ rdoc -E cxx=c -f html file_wrap.cxx
</pre>
</div>
<p>To generate ri documentation from a c wrap file, you could do:</p>
<div class="code shell">
<pre>
$ rdoc -r file_wrap.c
</pre>
</div>
<h3><a name="Ruby_nn66">35.8.1 Module docstring</a></h3>
<p> Ruby allows a docstring at the beginning of the file before any
other statements, and it is typically used to give a general
description of the entire module. SWIG supports this by setting an
option of the <tt>%module</tt> directive. For example:</p>
<div class="code">
<pre>%module(docstring=&quot;This is the example module's docstring&quot;) example</pre>
</div>
<p> When you have more than just a line or so then you can retain the
easy readability of the <tt>%module</tt> directive by using a macro.
For example:</p>
<div class="code">
<pre>%define DOCSTRING
&quot;The `XmlResource` class allows program resources defining menus,
layout of controls on a panel, etc. to be loaded from an XML file.&quot;
%enddef
%module(docstring=DOCSTRING) xrc</pre>
</div>
<h3><a name="Ruby_nn67">35.8.2 %feature(&quot;autodoc&quot;)</a></h3>
<p>Since SWIG does know everything about the function it wraps, it is
possible to generate an rdoc containing the parameter types, names and
default values. Since Ruby ships with one of the best documentation
systems of any language, it makes sense to take advantage of it.</p>
<p>SWIG's Ruby module provides support for the &quot;autodoc&quot; feature, which
when attached to a node in the parse tree will cause an rdoc comment to
be generated in the wrapper file that includes the name of the
function, parameter names, default values if any, and return type if
any. There are also several options for autodoc controlled by the value
given to the feature, described below.</p>
<h4><a name="Ruby_nn68">35.8.2.1 %feature(&quot;autodoc&quot;, &quot;0&quot;)</a></h4>
<p> When the &quot;0&quot; option is given then the types of the parameters will<em>
not</em> be included in the autodoc string. For example, given this
function prototype:</p>
<div class="code">
<pre>%feature(&quot;autodoc&quot;, &quot;0&quot;);
bool function_name(int x, int y, Foo* foo=NULL, Bar* bar=NULL);</pre>
</div>
<p> Then Ruby code like this will be generated:</p>
<div class="targetlang">
<pre>function_name(x, y, foo=nil, bar=nil) -&gt; bool
...</pre>
</div>
<h4><a name="Ruby_autodoc1">35.8.2.2 %feature(&quot;autodoc&quot;, &quot;1&quot;)</a></h4>
<p> When the &quot;1&quot; option is used then the parameter types<em> will</em>
be used in the rdoc string. In addition, an attempt is made to simplify
the type name such that it makes more sense to the Ruby user. Pointer,
reference and const info is removed, <tt>%rename</tt>'s are evaluated,
etc. (This is not always successful, but works most of the time. See
the next section for what to do when it doesn't.) Given the example
above, then turning on the parameter types with the &quot;1&quot; option will
result in rdoc code like this:</p>
<div class="targetlang">
<pre>function_name(int x, int y, Foo foo=nil, Bar bar=nil) -&gt; bool
...</pre>
</div>
<h4><a name="Ruby_autodoc2">35.8.2.3 %feature(&quot;autodoc&quot;, &quot;2&quot;)</a></h4>
<p> When the &quot;2&quot; option is used then the parameter types will not be
used in the rdoc string. However, they will be listed in full after the
function. Given the example above, then turning on the parameter types
with the &quot;2&quot; option will result in Ruby code like this:</p>
<h4><a name="Ruby_feature_autodoc3">35.8.2.4 %feature(&quot;autodoc&quot;, &quot;3&quot;)</a>
</h4>
<p> When the &quot;3&quot; option is used then the function will be documented
using a combination of &quot;1&quot; and &quot;2&quot; above. Given the example above, then
turning on the parameter types with the &quot;2&quot; option will result in Ruby
code like this:</p>
<div class="targetlang">
<pre>function_name(int x, int y, Foo foo=nil, Bar bar=nil) -&gt; bool
Parameters:
x - int
y - int
foo - Foo
bar - Bar</pre>
</div>
<h4><a name="Ruby_nn70">35.8.2.5 %feature(&quot;autodoc&quot;, &quot;docstring&quot;)</a></h4>
<p> Finally, there are times when the automatically generated autodoc
string will make no sense for a Ruby programmer, particularly when a
typemap is involved. So if you give an explicit value for the autodoc
feature then that string will be used in place of the automatically
generated string. For example:</p>
<div class="code">
<pre>%feature(&quot;autodoc&quot;, &quot;GetPosition() -&gt; (x, y)&quot;) GetPosition;
void GetPosition(int* OUTPUT, int* OUTPUT);</pre>
</div>
<h3><a name="Ruby_nn71">35.8.3 %feature(&quot;docstring&quot;)</a></h3>
<p> In addition to the autodoc strings described above, you can also
attach any arbitrary descriptive text to a node in the parse tree with
the &quot;docstring&quot; feature. When the proxy module is generated then any
docstring associated with classes, function or methods are output. If
an item already has an autodoc string then it is combined with the
docstring and they are output together.</p>
<h2><a name="Ruby_nn53">35.9 Advanced Topics</a></h2>
<h3><a name="Ruby_operator_overloading">35.9.1 Operator overloading</a></h3>
<p> SWIG allows operator overloading with, by using the <tt>%extend</tt>
or <tt>%rename</tt> commands in SWIG and the following operator names
(derived from Python):</p>
<div class="code diagram">
<table border="1" cellpadding="2" cellspacing="2" style="width: 100%; font-family: monospace;"
summary="operator names"><tbody>
<tr><td><b> General</b></td></tr>
<tr><td>__repr__</td><td> inspect</td></tr>
<tr><td>__str__</td><td> to_s</td></tr>
<tr><td>__cmp__</td><td> &lt;=&gt;</td></tr>
<tr><td>__hash__</td><td> hash</td></tr>
<tr><td>__nonzero__</td><td> nonzero?</td></tr>
<tr><td></td></tr>
<tr><td><b> Callable</b></td></tr>
<tr><td>__call__</td><td> call</td></tr>
<tr><td></td></tr>
<tr><td><b> Collection</b></td></tr>
<tr><td>__len__</td><td> length</td></tr>
<tr><td>__getitem__</td><td> []</td></tr>
<tr><td>__setitem__</td><td> []=</td></tr>
<tr><td></td></tr>
<tr><td><b> Numeric</b></td></tr>
<tr><td>__add__</td><td> +</td></tr>
<tr><td>__sub__</td><td> -</td><td></td></tr>
<tr><td>__mul__</td><td> *</td></tr>
<tr><td>__div__</td><td> /</td></tr>
<tr><td>__mod__</td><td> %</td></tr>
<tr><td>__divmod__</td><td> divmod</td></tr>
<tr><td>__pow__</td><td> **</td></tr>
<tr><td>__lshift__</td><td> &lt;&lt;</td></tr>
<tr><td>__rshift__</td><td> &gt;&gt;</td></tr>
<tr><td>__and__</td><td> &amp;</td></tr>
<tr><td>__xor__</td><td> ^</td></tr>
<tr><td>__or__</td><td> |</td></tr>
<tr><td>__neg__</td><td> -@</td><td></td></tr>
<tr><td>__pos__</td><td> +@</td></tr>
<tr><td>__abs__</td><td> abs</td></tr>
<tr><td>__invert__</td><td> ~</td></tr>
<tr><td>__int__</td><td> to_i</td></tr>
<tr><td>__float__</td><td> to_f</td></tr>
<tr><td>__coerce__</td><td> coerce</td></tr>
<tr><td></td></tr>
<tr><td><b>Additions in 1.3.13</b></td></tr>
<tr><td>__lt__</td><td> &lt;</td></tr>
<tr><td>__le__</td><td> &lt;=</td></tr>
<tr><td>__eq__</td><td> ==</td></tr>
<tr><td>__gt__</td><td> &gt;</td></tr>
<tr><td>__ge__</td><td> &gt;=</td></tr>
</tbody></table>
</div>
<p> Note that although SWIG supports the <tt>__eq__</tt> magic method
name for defining an equivalence operator, there is no separate method
for handling<i> inequality</i> since Ruby parses the expression<i> a !=
b</i> as<i> !(a == b)</i>.</p>
<h3><a name="Ruby_nn55">35.9.2 Creating Multi-Module Packages</a></h3>
<p> The chapter on <a href="#Modules">Working with Modules</a> discusses
the basics of creating multi-module extensions with SWIG, and in
particular the considerations for sharing runtime type information
among the different modules.</p>
<p>As an example, consider one module's interface file (<tt>shape.i</tt>
) that defines our base class:</p>
<div class="code">
<pre>%module shape
%{
#include &quot;Shape.h&quot;
%}
class Shape {
protected:
double xpos;
double ypos;
protected:
Shape(double x, double y);
public:
double getX() const;
double getY() const;
};</pre>
</div>
<p> We also have a separate interface file (<tt>circle.i</tt>) that
defines a derived class:</p>
<div class="code">
<pre>%module circle
%{
#include &quot;Shape.h&quot;
#include &quot;Circle.h&quot;
%}
// Import the base class definition from Shape module
%import shape.i
class Circle : public Shape {
protected:
double radius;
public:
Circle(double x, double y, double r);
double getRadius() const;
};</pre>
</div>
<p> We'll start by building the<b> Shape</b> extension module:</p>
<div class="code shell">
<pre>$ swig -c++ -ruby shape.i
</pre>
</div>
<p> SWIG generates a wrapper file named <tt>shape_wrap.cxx</tt>. To
compile this into a dynamically loadable extension for Ruby, prepare an
<tt>extconf.rb</tt> script using this template:</p>
<div class="code targetlang">
<pre>require 'mkmf'
# Since the SWIG runtime support library for Ruby
# depends on the Ruby library, make sure it's in the list
# of libraries.
$libs = append_library($libs, Config::CONFIG['RUBY_INSTALL_NAME'])
# Create the makefile
create_makefile('shape')</pre>
</div>
<p> Run this script to create a <tt>Makefile</tt> and then type <tt>make</tt>
to build the shared library:</p>
<div class="code targetlang">
<pre>$ <b>ruby extconf.rb</b>
creating Makefile
$ <b>make</b>
g++ -fPIC -g -O2 -I. -I/usr/include/ruby-2.1.0 \
-I. -c shape_wrap.cxx
gcc -shared -L/usr/local/lib -o shape.so shape_wrap.o -L. \
-lruby -lruby -lc</pre>
</div>
<p> Note that depending on your installation, the outputs may be
slightly different; these outputs are those for a Linux-based
development environment. The end result should be a shared library
(here, <tt>shape.so</tt>) containing the extension module code. Now
repeat this process in a separate directory for the<b> Circle</b>
module:</p>
<ol>
<li> Run SWIG to generate the wrapper code (<tt>circle_wrap.cxx</tt>);</li>
<li> Write an <tt>extconf.rb</tt> script that your end-users can use to
create a platform-specific <tt>Makefile</tt> for the extension;</li>
<li> Build the shared library for this extension by typing <tt>make</tt>
.</li>
</ol>
<p> Once you've built both of these extension modules, you can test them
interactively in IRB to confirm that the <tt>Shape</tt> and <tt>Circle</tt>
modules are properly loaded and initialized:</p>
<div class="code targetlang">
<pre>$ <b>irb</b>
irb(main):001:0&gt; <b>require 'shape'</b>
true
irb(main):002:0&gt; <b>require 'circle'</b>
true
irb(main):003:0&gt; <b>c = Circle::Circle.new(5, 5, 20)</b>
#&lt;Circle::Circle:0xa097208&gt;
irb(main):004:0&gt; <b>c.kind_of? Shape::Shape</b>
true
irb(main):005:0&gt; <b>c.getX()</b>
5.0</pre>
</div>
<h3><a name="Ruby_nn56">35.9.3 Specifying Mixin Modules</a></h3>
<p> The Ruby language doesn't support multiple inheritance, but it does
allow you to mix one or more modules into a class using Ruby's <tt>
include</tt> method. For example, if you have a Ruby class that defines
an<em> each</em> instance method, e.g.</p>
<div class="code targetlang">
<pre>class Set
def initialize
@members = []
end
def each
@members.each { |m| yield m }
end
end</pre>
</div>
<p> then you can mix-in Ruby's <tt>Enumerable</tt> module to easily add
a lot of functionality to your class:</p>
<div class="code targetlang">
<pre>class Set
<b>include Enumerable</b>
def initialize
@members = []
end
def each
@members.each { |m| yield m }
end
end</pre>
</div>
<p> To get the same benefit for your SWIG-wrapped classes, you can use
the <tt>%mixin</tt> directive to specify the names of one or more
modules that should be mixed-in to a class. For the above example, the
SWIG interface specification might look like this:</p>
<div class="code">
<pre>%mixin Set &quot;Enumerable&quot;;
class Set {
public:
// Constructor
Set();
// Iterates through set members
void each();
};</pre>
</div>
<p> Multiple modules can be mixed into a class by providing a
comma-separated list of module names to the <tt>%mixin</tt> directive,
e.g.</p>
<div class="code">
<pre>%mixin Set &quot;Fee, Fi, Fo, Fum&quot;;</pre>
</div>
<p> Note that the <tt>%mixin</tt> directive is implemented using SWIG's
&quot;features&quot; mechanism and so the same name matching rules used for other
kinds of features apply (see the chapter on <a href="#Customization">
&quot;Customization Features&quot;</a>) for more details).</p>
<h2><a name="Ruby_nn57">35.10 Memory Management</a></h2>
<p>One of the most common issues in generating SWIG bindings for Ruby is
proper memory management. The key to proper memory management is
clearly defining whether a wrapper Ruby object owns the underlying C
struct or C++ class. There are two possibilities:</p>
<ul>
<li> The Ruby object is responsible for freeing the C struct or C++
object</li>
<li> The Ruby object should not free the C struct or C++ object because
it will be freed by the underlying C or C++ code</li>
</ul>
<p>To complicate matters, object ownership may transfer from Ruby to C++
(or vice versa) depending on what function or methods are invoked.
Clearly, developing a SWIG wrapper requires a thorough understanding of
how the underlying library manages memory.</p>
<h3><a name="Ruby_nn58">35.10.1 Mark and Sweep Garbage Collector</a></h3>
<p>Ruby uses a mark and sweep garbage collector. When the garbage
collector runs, it finds all the &quot;root&quot; objects, including local
variables, global variables, global constants, hardware registers and
the C stack. For each root object, the garbage collector sets its mark
flag to true and calls <tt>rb_gc_mark</tt> on the object. The job of <tt>
rb_gc_mark</tt> is to recursively mark all the objects that a Ruby
object has a reference to (ignoring those objects that have already
been marked). Those objects, in turn, may reference other objects. This
process will continue until all active objects have been &quot;marked.&quot;
After the mark phase comes the sweep phase. In the sweep phase, all
objects that have not been marked will be garbage collected.</p>
<p>The Ruby C/API provides extension developers two hooks into the
garbage collector - a &quot;mark&quot; function and a &quot;sweep&quot; function. By
default these functions are set to NULL.</p>
<p>If a C struct or C++ class references any other Ruby objects, then it
must provide a &quot;mark&quot; function. The &quot;mark&quot; function should identify any
referenced Ruby objects by calling the rb_gc_mark function for each
one. Unsurprisingly, this function will be called by the Ruby garbage
during the &quot;mark&quot; phase.</p>
<p>During the sweep phase, Ruby destroys any unused objects. If any
memory has been allocated in creating the underlying C struct or C++
struct, then a &quot;free&quot; function must be defined that deallocates this
memory.</p>
<h3><a name="Ruby_nn59">35.10.2 Object Ownership</a></h3>
<p>As described above, memory management depends on clearly defining who
is responsible for freeing the underlying C struct or C++ class. If the
Ruby object is responsible for freeing the C++ object, then a &quot;free&quot;
function must be registered for the object. If the Ruby object is not
responsible for freeing the underlying memory, then a &quot;free&quot; function
must not be registered for the object.</p>
<p>For the most part, SWIG takes care of memory management issues. The
rules it uses are:</p>
<ul>
<li> When calling a C++ object's constructor from Ruby, SWIG will assign
a &quot;free&quot; function thereby making the Ruby object responsible for
freeing the C++ object</li>
<li> When calling a C++ member function that returns a pointer, SWIG
will not assign a &quot;free&quot; function thereby making the underlying library
responsible for freeing the object.</li>
</ul>
<p>To make this clearer, let's look at an example. Assume we have a Foo
and a Bar class.</p>
<div class="code">
<pre>/* File &quot;RubyOwernshipExample.h&quot; */
class Foo
{
public:
Foo() {}
~Foo() {}
};
class Bar
{
Foo *foo_;
public:
Bar(): foo_(new Foo) {}
~Bar() { delete foo_; }
Foo* get_foo() { return foo_; }
Foo* get_new_foo() { return new Foo; }
void set_foo(Foo *foo) { delete foo_; foo_ = foo; }
};</pre>
</div>
<p>First, consider this Ruby code:</p>
<div class="code targetlang">
<pre>foo = Foo.new</pre>
</div>
<p>In this case, the Ruby code calls the underlying <tt>Foo</tt> C++
constructor, thus creating a new <tt>foo</tt> object. By default, SWIG
will assign the new Ruby object a &quot;free&quot; function. When the Ruby object
is garbage collected, the &quot;free&quot; function will be called. It in turn
will call <tt>Foo</tt>'s destructor.</p>
<p>Next, consider this code:</p>
<div class="code targetlang">
<pre>bar = Bar.new
foo = bar.get_foo()</pre>
</div>
<p>In this case, the Ruby code calls a C++ member function, <tt>get_foo</tt>
. By default, SWIG will not assign the Ruby object a &quot;free&quot; function.
Thus, when the Ruby object is garbage collected the underlying C++ <tt>
foo</tt> object is not affected.</p>
<p>Unfortunately, the real world is not as simple as the examples above.
For example:</p>
<div class="code targetlang">
<pre>bar = Bar.new
foo = bar.get_new_foo()</pre>
</div>
<p>In this case, the default SWIG behavior for calling member functions
is incorrect. The Ruby object should assume ownership of the returned
object. This can be done by using the %newobject directive. See <a href="#Customization_ownership">
Object ownership and %newobject</a> for more information.</p>
<p>The SWIG default mappings are also incorrect in this case:</p>
<div class="code targetlang">
<pre>foo = Foo.new
bar = Bar.new
bar.set_foo(foo)</pre>
</div>
<p>Without modification, this code will cause a segmentation fault. When
the Ruby <tt>foo</tt> object goes out of scope, it will free the
underlying C++ <tt>foo</tt> object. However, when the Ruby bar object
goes out of scope, it will call the C++ bar destructor which will also
free the C++ <tt>foo</tt> object. The problem is that object ownership
is transferred from the Ruby object to the C++ object when the <tt>
set_foo</tt> method is called. This can be done by using the special
DISOWN type map, which was added to the Ruby bindings in SWIG-1.3.26.</p>
<p>Thus, a correct SWIG interface file correct mapping for these classes
is:</p>
<div class="code">
<pre>/* File RubyOwnershipExample.i */
%module RubyOwnershipExample
%{
#include &quot;RubyOwnershipExample.h&quot;
%}
class Foo
{
public:
Foo();
~Foo();
};
class Bar
{
Foo *foo_;
public:
Bar();
~Bar();
Foo* get_foo();
<b> %newobject get_new_foo;</b>
Foo* get_new_foo();
<b> %apply SWIGTYPE *DISOWN {Foo *foo};</b>
void set_foo(Foo *foo);
<b> %clear Foo *foo;</b>
};
</pre>
</div>
<p> This code can be seen in swig/examples/ruby/tracking.</p>
<h3><a name="Ruby_nn60">35.10.3 Object Tracking</a></h3>
<p>The remaining parts of this section will use the class library shown
below to illustrate different memory management techniques. The class
library models a zoo and the animals it contains.</p>
<div class="code">
<pre>%module zoo
%{
#include &lt;string&gt;
#include &lt;vector&gt;
#include &quot;zoo.h&quot;
%}
class Animal
{
private:
typedef std::vector&lt;Animal*&gt; AnimalsType;
typedef AnimalsType::iterator IterType;
protected:
AnimalsType animals;
protected:
std::string name_;
public:
// Construct an animal with this name
Animal(const char* name) : name_(name) {}
// Return the animal's name
const char* get_name() const { return name.c_str(); }
};
class Zoo
{
protected:
std::vector&lt;Animal *&gt; animals;
public:
// Construct an empty zoo
Zoo() {}
/* Create a new animal. */
static Animal* Zoo::create_animal(const char* name) {
return new Animal(name);
}
// Add a new animal to the zoo
void add_animal(Animal* animal) {
animals.push_back(animal);
}
Animal* remove_animal(size_t i) {
Animal* result = this-&gt;animals[i];
IterType iter = this-&gt;animals.begin();
std::advance(iter, i);
this-&gt;animals.erase(iter);
return result;
}
// Return the number of animals in the zoo
size_t get_num_animals() const {
return animals.size();
}
// Return a pointer to the ith animal
Animal* get_animal(size_t i) const {
return animals[i];
}
};</pre>
</div>
<p>Let's say you SWIG this code and then run IRB:</p>
<div class="code targetlang">
<pre>$ <b>irb</b>
irb(main):001:0&gt; <b>require 'example'</b>
=&gt; true
irb(main):002:0&gt; <b>tiger1 = Example::Animal.new(&quot;tiger1&quot;)</b>
=&gt; #&lt;Example::Animal:0x2be3820&gt;
irb(main):004:0&gt; <b>tiger1.get_name()</b>
=&gt; &quot;tiger1&quot;
irb(main):003:0&gt; <b>zoo = Example::Zoo.new()</b>
=&gt; #&lt;Example::Zoo:0x2be0a60&gt;
irb(main):006:0&gt; <b>zoo.add_animal(tiger)</b>
=&gt; nil
irb(main):007:0&gt; <b>zoo.get_num_animals()</b>
=&gt; 1
irb(main):007:0&gt; <b>tiger2 = zoo.remove_animal(0)</b>
=&gt; #&lt;Example::Animal:0x2bd4a18&gt;
irb(main):008:0&gt; <b>tiger2.get_name()</b>
=&gt; &quot;tiger1&quot;
irb(main):009:0&gt; <b>tiger1.equal?(tiger2)</b>
=&gt; false
</pre>
</div>
<p>Pay particular attention to the code <tt>tiger1.equal?(tiger2)</tt>.
Note that the two Ruby objects are not the same - but they reference
the same underlying C++ object. This can cause problems. For example:</p>
<div class="code targetlang">
<pre>irb(main):010:0&gt; <b>tiger1 = nil</b>
=&gt; nil
irb(main):011:0&gt; <b>GC.start</b>
=&gt; nil
irb(main):012:0&gt; <b>tiger2.get_name()</b>
(irb):12: [BUG] Segmentation fault
</pre>
</div>
<p>After the garbage collector runs, as a result of our call to <tt>
GC.start</tt>, calling<tt>tiger2.get_name()</tt> causes a segmentation
fault. The problem is that when <tt>tiger1</tt> is garbage collected,
it frees the underlying C++ object. Thus, when <tt>tiger2</tt> calls
the <tt>get_name()</tt> method it invokes it on a destroyed object.</p>
<p>This problem can be avoided if SWIG enforces a one-to-one mapping
between Ruby objects and C++ classes. This can be done via the use of
the <tt>%trackobjects</tt> functionality available in SWIG-1.3.26. and
later.</p>
<p>When the <tt>%trackobjects</tt> is turned on, SWIG automatically
keeps track of mappings between C++ objects and Ruby objects. Note that
enabling object tracking causes a slight performance degradation. Test
results show this degradation to be about 3% to 5% when creating and
destroying 100,000 animals in a row.</p>
<p>Since <tt>%trackobjects</tt> is implemented as a <tt>%feature</tt>,
it uses the same name matching rules as other kinds of features (see
the chapter on <a href="#Customization"> &quot;Customization Features&quot;</a>)
. Thus it can be applied on a class-by-class basis if needed. To fix
the example above:</p>
<div class="code">
<pre>%module example
%{
#include &quot;example.h&quot;
%}
<b>/* Tell SWIG that create_animal creates a new object */</b>
<b>%newobject Zoo::create_animal;</b>
<b>/* Tell SWIG to keep track of mappings between C/C++ structs/classes. */</b>
<b>%trackobjects;</b>
%include &quot;example.h&quot;</pre>
</div>
<p>When this code runs we see:</p>
<div class="code targetlang">
<pre>$ <b>irb</b>
irb(main):001:0&gt; <b>require 'example'</b>
=&gt; true
irb(main):002:0&gt; <b>tiger1 = Example::Animal.new(&quot;tiger1&quot;)</b>
=&gt; #&lt;Example::Animal:0x2be37d8&gt;
irb(main):003:0&gt; <b>zoo = Example::Zoo.new()</b>
=&gt; #&lt;Example::Zoo:0x2be0a18&gt;
irb(main):004:0&gt; <b>zoo.add_animal(tiger1)</b>
=&gt; nil
irb(main):006:0&gt; <b>tiger2 = zoo.remove_animal(0)</b>
=&gt; #&lt;Example::Animal:0x2be37d8&gt;
irb(main):007:0&gt; <b>tiger1.equal?(tiger2)</b>
=&gt; true
irb(main):008:0&gt; <b>tiger1 = nil</b>
=&gt; nil
irb(main):009:0&gt; <b>GC.start</b>
=&gt; nil
irb(main):010:0&gt; <b>tiger.get_name()</b>
=&gt; &quot;tiger1&quot;
irb(main):011:0&gt;</pre>
</div>
<p>For those who are interested, object tracking is implemented by
storing Ruby objects in a hash table and keying them on C++ pointers.
The underlying API is:</p>
<div class="code">
<pre>static void SWIG_RubyAddTracking(void* ptr, VALUE object);
static VALUE SWIG_RubyInstanceFor(void* ptr) ;
static void SWIG_RubyRemoveTracking(void* ptr);
static void SWIG_RubyUnlinkObjects(void* ptr);</pre>
</div>
<p>When an object is created, SWIG will automatically call the <tt>
SWIG_RubyAddTracking</tt> method. Similarly, when an object is deleted,
SWIG will call the <tt>SWIG_RubyRemoveTracking</tt>. When an object is
returned to Ruby from C++, SWIG will use the <tt>SWIG_RubyInstanceFor</tt>
method to ensure a one-to-one mapping from Ruby to C++ objects. Last,
the <tt>RubyUnlinkObjects</tt> method unlinks a Ruby object from its
underlying C++ object.</p>
<p>In general, you will only need to use the <tt>SWIG_RubyInstanceFor</tt>
, which is required for implementing mark functions as shown below.
However, if you implement your own free functions (see below) you may
also have to call the <tt>SWIG_RubyRemoveTracking</tt> and <tt>
RubyUnlinkObjects</tt> methods.</p>
<h3><a name="Ruby_nn61">35.10.4 Mark Functions</a></h3>
<p>With a bit more testing, we see that our class library still has
problems. For example:</p>
<div class="targetlang">
<pre>$ <b>irb</b>
irb(main):001:0&gt; <b>require 'example'</b>
=&gt; true
irb(main):002:0&gt; tiger1 = <b>Example::Animal.new(&quot;tiger1&quot;)</b>
=&gt; #&lt;Example::Animal:0x2bea6a8&gt;
irb(main):003:0&gt; zoo = <b>Example::Zoo.new()</b>
=&gt; #&lt;Example::Zoo:0x2be7960&gt;
irb(main):004:0&gt; <b>zoo.add_animal(tiger1)</b>
=&gt; nil
irb(main):007:0&gt; <b>tiger1 = nil</b>
=&gt; nil
irb(main):007:0&gt; <b>GC.start</b>
=&gt; nil
irb(main):005:0&gt; <b>tiger2 = zoo.get_animal(0)</b>
(irb):12: [BUG] Segmentation fault</pre>
</div>
<p>The problem is that Ruby does not know that the <tt>zoo</tt> object
contains a reference to a Ruby object. Thus, when Ruby garbage collects
<tt>tiger1</tt> it frees the underlying C++ object.</p>
<p>This can be fixed by implementing a <tt>mark</tt> function as
described above in the <a href="#Ruby_nn52">Mark and Sweep Garbage
Collector</a> section. You can specify a mark function by using the <tt>
%markfunc</tt> directive. Since the <tt>%markfunc</tt> directive is
implemented using SWIG's' &quot;features&quot; mechanism it uses the same name
matching rules as other kinds of features (see the chapter on <a href="#Customization">
&quot;Customization Features&quot;</a> for more details).</p>
<p>A <tt>mark</tt> function takes a single argument, which is a pointer
to the C++ object being marked; it should, in turn, call <tt>
rb_gc_mark()</tt> for any instances that are reachable from the current
object. The mark function for our <tt>Zoo</tt> class should therefore
loop over all of the C++ animal objects in the zoo object, look up
their Ruby object equivalent, and then call <tt>rb_gc_mark()</tt>. One
possible implementation is:</p>
<div class="code">
<pre>%module example
%{
#include &quot;example.h&quot;
%}
/* Keep track of mappings between C/C++ structs/classes
and Ruby objects so we can implement a mark function. */
<b>%trackobjects;</b>
/* Specify the mark function */
<b>%markfunc Zoo &quot;mark_Zoo&quot;;</b>
%include &quot;example.h&quot;
%header %{
static void mark_Zoo(void* ptr) {
Zoo* zoo = (Zoo*) ptr;
/* Loop over each object and tell the garbage collector
that we are holding a reference to them. */
int count = zoo-&gt;get_num_animals();
for(int i = 0; i &lt; count; ++i) {
Animal* animal = zoo-&gt;get_animal(i);
VALUE object = SWIG_RubyInstanceFor(animal);
if (object != Qnil) {
rb_gc_mark(object);
}
}
}
%}</pre>
</div>
<p> Note the <tt>mark</tt> function is dependent on the <tt>
SWIG_RUBY_InstanceFor</tt> method, and thus requires that <tt>
%trackobjects</tt> is enabled. For more information, please refer to the
ruby_track_objects.i test case in the SWIG test suite.</p>
<p>When this code is compiled we now see:</p>
<div class="targetlang">
<pre>$ <b>irb
</b>irb(main):002:0&gt; <b>tiger1=Example::Animal.new(&quot;tiger1&quot;)</b>
=&gt; #&lt;Example::Animal:0x2be3bf8&gt;
irb(main):003:0&gt; <b>Example::Zoo.new()</b>
=&gt; #&lt;Example::Zoo:0x2be1780&gt;
irb(main):004:0&gt; <b>zoo = Example::Zoo.new()</b>
=&gt; #&lt;Example::Zoo:0x2bde9c0&gt;
irb(main):005:0&gt; <b>zoo.add_animal(tiger1)</b>
=&gt; nil
irb(main):009:0&gt; <b>tiger1 = nil</b>
=&gt; nil
irb(main):010:0&gt; <b>GC.start</b>
=&gt; nil
irb(main):014:0&gt; <b>tiger2 = zoo.get_animal(0)</b>
=&gt; #&lt;Example::Animal:0x2be3bf8&gt;
irb(main):015:0&gt; <b>tiger2.get_name()</b>
=&gt; &quot;tiger1&quot;
irb(main):016:0&gt;</pre>
</div>
<p>This code can be seen in swig/examples/ruby/mark_function.</p>
<h3><a name="Ruby_nn62">35.10.5 Free Functions</a></h3>
<p>By default, SWIG creates a &quot;free&quot; function that is called when a Ruby
object is garbage collected. The free function simply calls the C++
object's destructor.</p>
<p>However, sometimes an appropriate destructor does not exist or
special processing needs to be performed before the destructor is
called. Therefore, SWIG allows you to manually specify a &quot;free&quot;
function via the use of the <tt>%freefunc</tt> directive. The <tt>
%freefunc</tt> directive is implemented using SWIG's' &quot;features&quot;
mechanism and so the same name matching rules used for other kinds of
features apply (see the chapter on <a href="#Customization">
&quot;Customization Features&quot;</a>) for more details).</p>
<p>IMPORTANT ! - If you define your own free function, then you must
ensure that you call the underlying C++ object's destructor. In
addition, if object tracking is activated for the object's class, you
must also call the <tt>SWIG_RubyRemoveTracking</tt> function (of course
call this before you destroy the C++ object). Note that it is harmless
to call this method if object tracking if off so it is advised to
always call it.</p>
<p>Note there is a subtle interaction between object ownership and free
functions. A custom defined free function will only be called if the
Ruby object owns the underlying C++ object. This also to Ruby objects
which are created, but then transfer ownership to C++ objects via the
use of the <tt>disown</tt> typemap described above.</p>
<p>To show how to use the <tt>%freefunc</tt> directive, let's slightly
change our example. Assume that the zoo object is responsible for
freeing any animal that it contains. This means that the <tt>
Zoo::add_animal</tt> function should be marked with a <tt>DISOWN</tt>
typemap and the destructor should be updated as below:</p>
<div class="code">
<pre>
Zoo::~Zoo() {
IterType iter = this-&gt;animals.begin();
IterType end = this-&gt;animals.end();
for(iter; iter != end; ++iter) {
Animal* animal = *iter;
delete animal;
}
}
</pre>
</div>
<p>When we use these objects in IRB we see:</p>
<div class="code targetlang">
<pre class="targetlang"><b>$irb</b>
irb(main):002:0&gt; <b>require 'example'</b>
=&gt; true
irb(main):003:0&gt; <b>zoo = Example::Zoo.new()</b>
=&gt; #&lt;Example::Zoo:0x2be0fe8&gt;
irb(main):005:0&gt; <b>tiger1 = Example::Animal.new(&quot;tiger1&quot;)</b>
=&gt; #&lt;Example::Animal:0x2bda760&gt;
irb(main):006:0&gt; <b>zoo.add_animal(tiger1)</b>
=&gt; nil
irb(main):007:0&gt; <b>zoo = nil</b>
=&gt; nil
irb(main):008:0&gt; <b>GC.start</b>
=&gt; nil
irb(main):009:0&gt; <b>tiger1.get_name()</b>
(irb):12: [BUG] Segmentation fault
</pre>
</div>
<p>The error happens because the C++ <tt>animal</tt> object is freed
when the <tt>zoo</tt> object is freed. Although this error is
unavoidable, we can at least prevent the segmentation fault. To do this
requires enabling object tracking and implementing a custom free
function that calls the <tt>SWIG_RubyUnlinkObjects</tt> function for
each animal object that is destroyed. The <tt>SWIG_RubyUnlinkObjects</tt>
function notifies SWIG that a Ruby object's underlying C++ object is no
longer valid. Once notified, SWIG will intercept any calls from the
existing Ruby object to the destroyed C++ object and raise an
exception.</p>
<div class="code">
<pre>%module example
%{
#include &quot;example.h&quot;
%}
/* Specify that ownership is transferred to the zoo when calling add_animal */
%apply SWIGTYPE *DISOWN { Animal* animal };
/* Track objects */
%trackobjects;
/* Specify the mark function */
%freefunc Zoo &quot;free_Zoo&quot;;
%include &quot;example.h&quot;
%header %{
static void free_Zoo(void* ptr) {
Zoo* zoo = (Zoo*) ptr;
/* Loop over each animal */
int count = zoo-&gt;get_num_animals();
for(int i = 0; i &lt; count; ++i) {
/* Get an animal */
Animal* animal = zoo-&gt;get_animal(i);
/* Unlink the Ruby object from the C++ object */
SWIG_RubyUnlinkObjects(animal);
/* Now remove the tracking for this animal */
SWIG_RubyRemoveTracking(animal);
}
/* Now call SWIG_RubyRemoveTracking for the zoo */
SWIG_RubyRemoveTracking(ptr);
/* Now free the zoo which will free the animals it contains */
delete zoo;
}
%} </pre>
</div>
<p>Now when we use these objects in IRB we see:</p>
<div class="code targetlang">
<pre><b>$irb</b>
irb(main):002:0&gt; <b>require 'example'</b>
=&gt; true
irb(main):003:0&gt; <b>zoo = Example::Zoo.new()</b>
=&gt; #&lt;Example::Zoo:0x2be0fe8&gt;
irb(main):005:0&gt; <b>tiger1 = Example::Animal.new(&quot;tiger1&quot;)</b>
=&gt; #&lt;Example::Animal:0x2bda760&gt;
irb(main):006:0&gt; <b>zoo.add_animal(tiger1)</b>
=&gt; nil
irb(main):007:0&gt; <b>zoo = nil</b>
=&gt; nil
irb(main):008:0&gt; <b>GC.start</b>
=&gt; nil
irb(main):009:0&gt; <b>tiger1.get_name()</b>
RuntimeError: This Animal * already released
from (irb):10:in `get_name'
from (irb):10
irb(main):011:0&gt;</pre>
</div>
<p>Notice that SWIG can now detect the underlying C++ object has been
freed, and thus raises a runtime exception.</p>
<p>This code can be seen in swig/examples/ruby/free_function.</p>
<h3><a name="Ruby_nn63">35.10.6 Embedded Ruby and the C++ Stack</a></h3>
<p>As has been said, the Ruby GC runs and marks objects before its sweep
phase. When the garbage collector is called, it will also try to mark
any Ruby objects (VALUE) it finds in the machine registers and in the
C++ stack.</p>
<p>The stack is basically the history of the functions that have been
called and also contains local variables, such as the ones you define
whenever you do inside a function:</p>
<div class="diagram">VALUE obj;</div>
<p>For ruby to determine where its stack space begins, during
initialization a normal Ruby interpreter will call the ruby_init()
function which in turn will call a function called Init_stack or
similar. This function will store a pointer to the location where the
stack points at that point in time.</p>
<p>ruby_init() is presumed to always be called within the main()
function of your program and whenever the GC is called, ruby will
assume that the memory between the current location in memory and the
pointer that was stored previously represents the stack, which may
contain local (and temporary) VALUE ruby objects. Ruby will then be
careful not to remove any of those objects in that location.</p>
<p>So far so good. For a normal Ruby session, all the above is
completely transparent and magic to the extensions developer.</p>
<p>However, with an embedded Ruby, it may not always be possible to
modify main() to make sure ruby_init() is called there. As such,
ruby_init() will likely end up being called from within some other
function. This can lead Ruby to measure incorrectly where the stack
begins and can result in Ruby incorrectly collecting those temporary
VALUE objects that are created once another function is called. The end
result: random crashes and segmentation faults.</p>
<p>This problem will often be seen in director functions that are used
for callbacks, for example.</p>
<p>To solve the problem, SWIG can now generate code with director
functions containing the optional macros SWIG_INIT_STACK and
SWIG_RELEASE_STACK. These macros will try to force Ruby to reinitialize
the beginning of the stack the first time a director function is
called. This will lead Ruby to measure and not collect any VALUE
objects defined from that point on.</p>
<p>To mark functions to either reset the ruby stack or not, you can use:</p>
<div class="code">
<pre>
%initstack Class::memberfunction; // only re-init the stack in this director method
%ignorestack Class::memberfunction; // do not re-init the stack in this director method
%initstack Class; // init the stack on all the methods of this class
%initstack; // all director functions will re-init the stack
</pre>
</div><HR NOSHADE>
<h1><a name="Scilab">36 SWIG and Scilab</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Scilab_preliminaries">Preliminaries</a></li>
<li><a href="#Scilab_running_swig">Running SWIG</a>
<ul>
<li><a href="#Scilab_running_swig_generating_module">Generating the
module</a></li>
<li><a href="#Scilab_running_swig_building_module">Building the module</a>
</li>
<li><a href="#Scilab_running_swig_loading_module">Loading the module</a></li>
<li><a href="#Scilab_running_swig_using_module">Using the module</a></li>
<li><a href="#Scilab_running_swig_options">Scilab command line options</a>
</li>
</ul>
</li>
<li><a href="#Scilab_wrapping">A basic tour of C/C++ wrapping</a>
<ul>
<li><a href="#Scilab_wrapping_overview">Overview</a></li>
<li><a href="#Scilab_wrapping_identifiers">Identifiers</a></li>
<li><a href="#Scilab_wrapping_functions">Functions</a>
<ul>
<li><a href="#Scilab_nn13">Argument passing</a></li>
<li><a href="#Scilab_nn14">Multiple output arguments</a></li>
</ul>
</li>
<li><a href="#Scilab_wrapping_global_variables">Global variables</a></li>
<li><a href="#Scilab_wrapping_constants_and_enums">Constants and
enumerations</a>
<ul>
<li><a href="#Scilab_wrapping_constants">Constants</a></li>
<li><a href="#Scilab_wrapping_enums">Enumerations</a></li>
</ul>
</li>
<li><a href="#Scilab_wrapping_pointers">Pointers</a>
<ul>
<li><a href="#Scilab_wrapping_pointers_utility_functions">Utility
functions</a></li>
<li><a href="#Scilab_wrapping_pointers_null_pointers">Null pointers:</a></li>
</ul>
</li>
<li><a href="#Scilab_wrapping_structs">Structures</a></li>
<li><a href="#Scilab_wrapping_cpp_classes">C++ classes</a></li>
<li><a href="#Scilab_wrapping_cpp_inheritance">C++ inheritance</a></li>
<li><a href="#Scilab_wrapping_cpp_overloading">C++ overloading</a></li>
<li><a href="#Scilab_wrapping_pointers_references_values_arrays">
Pointers, references, values, and arrays</a></li>
<li><a href="#Scilab_wrapping_cpp_templates">C++ templates</a></li>
<li><a href="#Scilab_wrapping_cpp_operators">C++ operators</a></li>
<li><a href="#Scilab_wrapping_cpp_namespaces">C++ namespaces</a></li>
<li><a href="#Scilab_wrapping_cpp_exceptions">C++ exceptions</a></li>
<li><a href="#Scilab_wrapping_cpp_stl">C++ STL</a></li>
</ul>
</li>
<li><a href="#Scilab_typemaps">Type mappings and libraries</a>
<ul>
<li><a href="#Scilab_typemaps_primitive_types">Default primitive type
mappings</a></li>
<li><a href="#Scilab_typemaps_arrays">Arrays</a></li>
<li><a href="#Scilab_typemaps_pointer-to-pointers">Pointer-to-pointers</a>
</li>
<li><a href="#Scilab_typemaps_matrices">Matrices</a></li>
<li><a href="#Scilab_typemaps_stl">STL</a></li>
</ul>
</li>
<li><a href="#Scilab_module_initialization">Module initialization</a></li>
<li><a href="#Scilab_building_modes">Building modes</a>
<ul>
<li><a href="#Scilab_building_modes_nobuilder_mode">No-builder mode</a></li>
<li><a href="#Scilab_building_modes_builder_mode">Builder mode</a></li>
</ul>
</li>
<li><a href="#Scilab_generated_scripts">Generated scripts</a>
<ul>
<li><a href="#Scilab_generated_scripts_builder_script">Builder script</a>
</li>
<li><a href="#Scilab_generated_scripts_loader_script">Loader script</a></li>
<li><a href="#Scilab_generated_scripts_gateway">Gateway XML files</a></li>
</ul>
</li>
<li><a href="#Scilab_other_resources">Other resources</a></li>
</ul>
</div>
<!-- INDEX -->
<p> Scilab is a scientific software package for numerical computations
providing a powerful open computing environment for engineering and
scientific applications that is mostly compatible with MATLAB. More
information can be found at <a href="https://www.scilab.org">
www.scilab.org</a>.</p>
<p> This chapter explains how to use SWIG for Scilab. After this
introduction, you should be able to generate with SWIG a Scilab
external module from a C/C++ library.</p>
<h2><a name="Scilab_preliminaries">36.1 Preliminaries</a></h2>
<p> SWIG for Scilab supports Linux. Other operating systems haven't been
tested.</p>
<p> Scilab is supported from version 5.5.2 onwards, the SWIG generated
code is supported on both Scilab 5, Scilab 6 and more recent versions.</p>
<p> SWIG for Scilab supports C language. C++ is partially supported. See
<a href="#Scilab_wrapping">A basic tour of C/C++ wrapping</a> for
further details.</p>
<h2><a name="Scilab_running_swig">36.2 Running SWIG</a></h2>
<p> Let's see how to use SWIG for Scilab on a small example.
<br> In this example we bind from C a function and a global variable
into Scilab. The SWIG interface (stored in a file named <tt>example.i</tt>
), is the following:</p>
<div class="code">
<pre>
%module example
%inline %{
double Foo = 3.0;
int fact(int n) {
if (n &lt; 0) {
return 0;
}
else if (n == 0) {
return 1;
}
else {
return n * fact(n-1);
}
}
%}
</pre>
</div>
<p> Note: a code in an <tt>%inline</tt> section is both parsed and
wrapped by SWIG, and inserted as is in the wrapper source file.</p>
<h3><a name="Scilab_running_swig_generating_module">36.2.1 Generating
the module</a></h3>
<p> The module is generated using the <tt>swig</tt> executable and its <tt>
-scilab</tt> option.</p>
<div class="shell">
<pre>
$ swig -scilab example.i
</pre>
</div>
<p> This command generates two files:</p>
<ul>
<li><tt>example_wrap.c</tt>: a C source file containing the wrapping
code and also here the wrapped code (the <tt>fact()</tt> and <tt>Foo</tt>
definitions)</li>
<li><tt>loader.sce</tt>: a Scilab script used to load the module into
Scilab</li>
</ul>
<p> Note: if the following error is returned:</p>
<div class="shell">
<pre>
:1: Error: Unable to find 'swig.swg'
:3: Error: Unable to find 'scilab.swg'
</pre>
</div>
<p> it may be because the SWIG library is not found. Check the <tt>
SWIG_LIB</tt> environment variable or your SWIG installation.</p>
<p> Note: SWIG for Scilab can work in two modes related to the way the
module is built, see the <a href="#Scilab_building_modes">Building
modes</a> section for details. This example uses the <tt>builder</tt>
mode.</p>
<p> The <tt>swig</tt> executable has several other command line options
you can use. See <a href="#Scilab_running_swig_options">Scilab command
line options</a> for further details.</p>
<h3><a name="Scilab_running_swig_building_module">36.2.2 Building the
module</a></h3>
<p> To be loaded in Scilab, the wrapper has to be built into a dynamic
module (or shared library).</p>
<p> The commands to compile and link the wrapper (with <tt>gcc</tt>)
into the shared library <tt>libexample.so</tt> are:</p>
<div class="shell">
<pre>
$ gcc -fPIC -c -I/usr/local/include/scilab example_wrap.c
$ gcc -shared example_wrap.o -o libexample.so
</pre>
</div>
<p> Note: we supposed in this example that the path to the Scilab
include directory is <tt>/usr/local/include/scilab</tt> (which is the
case in a Debian environment), this should be changed for another
environment.</p>
<h3><a name="Scilab_running_swig_loading_module">36.2.3 Loading the
module</a></h3>
<p> Loading a module is done by running the loader script in Scilab:</p>
<div class="targetlang">
<pre>
--&gt; exec loader.sce
</pre>
</div>
<p> Scilab should output the following messages:</p>
<div class="targetlang">
<pre>
Shared archive loaded.
Link done.
</pre>
</div>
<p> which means that Scilab has successfully loaded the shared library.
The module functions and other symbols are now available in Scilab.</p>
<h3><a name="Scilab_running_swig_using_module">36.2.4 Using the module</a>
</h3>
<p> In Scilab, the function <tt>fact()</tt> is simply called as
following:</p>
<div class="targetlang">
<pre>
--&gt; fact(5)
ans =
120.
</pre>
</div>
<p>For the <tt>Foo</tt> global variable, the accessors need to be used:<div
class="targetlang">
<pre>
--&gt; Foo_get
ans =
3.
--&gt; Foo_set(4);
--&gt; Foo_get
ans =
4.
</pre>
</div></p>
<p> Note: for conciseness, we assume in the subsequent Scilab code
examples that the modules have been beforehand built and loaded in
Scilab.</p>
<h3><a name="Scilab_running_swig_options">36.2.5 Scilab command line
options</a></h3>
<p> The following table lists the Scilab specific command line options
in addition to the generic SWIG options:</p>
<table summary="Scilab specific options">
<tr><td><tt>-builder</tt></td><td>Generate the Scilab builder script</td>
</tr>
<tr><td><tt>-buildercflags &lt;cflags&gt;</tt></td><td>Add &lt;cflags&gt; to the
builder compiler flags</td></tr>
<tr><td><tt>-builderldflags &lt;ldflags&gt;</tt></td><td>Add &lt;ldlags&gt; to the
builder linker flags</td></tr>
<tr><td><tt>-buildersources &lt;files&gt;</tt></td><td>Add the (comma
separated) files &lt;files&gt; to the builder sources</td></tr>
<tr><td><tt>-builderverbositylevel &lt;level&gt;</tt></td><td>Set the build
verbosity level to &lt;level&gt; (default 0: off, 2: high)</td></tr>
<tr><td><tt>-builderflagscript &lt;file&gt;</tt></td><td>Use the Scilab script
&lt;file&gt; to configure the compiler and linker flags</td></tr>
<tr><td><tt>-gatewayxml &lt;gateway_id&gt;</tt></td><td>Generate the gateway
XML with the given &lt;gateway_id&gt;</td></tr>
<tr><td><tt>-gatewayxml6</tt></td><td>Generate a gateway XML file
compatible with Scilab 6</td></tr>
</table>
<p> These options can be displayed with:</p>
<div class="shell">
<pre>
$ swig -scilab -help
</pre>
</div>
<h2><a name="Scilab_wrapping">36.3 A basic tour of C/C++ wrapping</a></h2>
<h3><a name="Scilab_wrapping_overview">36.3.1 Overview</a></h3>
<p> SWIG for Scilab provides only a low-level C interface for Scilab
(see <a href="#Scripting">Scripting Languages</a> for the general
approach to wrapping). This means that functions, structs, classes,
variables, etc... are interfaced through C functions. These C functions
are mapped as Scilab functions. There are a few exceptions, such as
constants and enumerations, which can be wrapped directly as Scilab
variables.</p>
<h3><a name="Scilab_wrapping_identifiers">36.3.2 Identifiers</a></h3>
<p> In Scilab 5.x, identifier names are composed of 24 characters
maximum (this limitation disappears from Scilab 6.0 onwards).
<br> By default, variable, member, and function names longer than 24
characters are truncated, and a warning is produced for each
truncation.</p>
<p>This can cause ambiguities, especially when wrapping structs/classes,
for which the wrapped function name is composed of the struct/class
name and field names. In these cases, the <a href="#SWIG_rename_ignore">
%rename directive</a> can be used to choose a different Scilab name.</p>
<h3><a name="Scilab_wrapping_functions">36.3.3 Functions</a></h3>
<p> Functions are wrapped as new Scilab built-in functions. For example:</p>
<div class="code">
<pre>
%module example
%inline %{
int fact(int n) {
if (n &gt; 1)
return n * fact(n - 1);
else
return 1;
}
%}
</pre>
</div>
<p> creates a built-in function <tt>fact(n)</tt> in Scilab:</p>
<div class="targetlang">
<pre>
--&gt; fact(4)
ans =
24.
</pre>
</div>
<h4><a name="Scilab_nn13">36.3.3.1 Argument passing</a></h4>
<p> In the above example, the function parameter is a primitive type and
is marshalled by value. So this function is wrapped without any
additional customization. Argument values are converted between C types
and Scilab types through type mappings. There are several default type
mappings for primitive and complex types, described later in the <a href="#Scilab_typemaps">
Scilab typemaps</a> section.</p>
<p> When a parameter is not passed by value, such as a pointer or
reference, SWIG does not know if it is an input, output (or both)
parameter. The INPUT, OUTPUT, INOUT typemaps defined in the <tt>
typemaps.i</tt> library can be used to specify this.</p>
<p> Let's see this on two simple functions: <tt>sub()</tt> which has an
output parameter, and <tt>inc()</tt>, which as input/output parameter:</p>
<div class="code">
<pre>
%module example
%include &lt;typemaps.i&gt;
extern void sub(int *INPUT, int *INPUT, int *OUTPUT);
extern void inc(int *INOUT, int *INPUT);
%{
void sub(int *x, int *y, int *result) {
*result = *x - *y;
}
void inc(int *x, int *delta) {
*x = *x + *delta;
}
%}
</pre>
</div>
<p> In Scilab, parameters are passed by value. The output (and inout)
parameters are returned as the result of the functions:</p>
<div class="targetlang">
<pre>
--&gt; sub(5, 3)
ans =
2.
--&gt; inc(4, 3)
ans =
7.
</pre>
</div>
<h4><a name="Scilab_nn14">36.3.3.2 Multiple output arguments</a></h4>
<p> A C function can have several output parameters. They can all be
returned as results of the wrapped function as Scilab supports multiple
return values from a function when using the <tt>typemaps.i</tt>
library. If the C function itself returns a result, this is returned
first before the parameter outputs.</p>
<p> The example below shows this for a C function returning 2 values and
a result:</p>
<div class="code">
<pre>
%module example
%include &lt;typemaps.i&gt;
int divide(int n, int d, int *OUTPUT, int *OUTPUT);
%{
int divide(int n, int d, int q*, int *r) {
if (d != 0) {
*q = n / d;
*r = n % d;
return 1;
} else {
return 0;
}
}
%}
</pre>
</div>
<br><div class="targetlang">
<pre>
--&gt; [ret, q, r] = divide(20, 6)
r =
2.
q =
3.
ret =
1.
</pre>
</div>
<h3><a name="Scilab_wrapping_global_variables">36.3.4 Global variables</a>
</h3>
<p> Global variables are manipulated through generated accessor
functions. For example, for a given <tt>Foo</tt> global variable, SWIG
actually generates two functions: <tt>Foo_get()</tt> to get the value
of <tt>Foo</tt>, and <tt>Foo_set()</tt> to set the value. These
functions are used as following:</p>
<div class="targetlang">
<pre>
--&gt; exec loader.sce;
--&gt; c = Foo_get();
--&gt; Foo_set(4);
--&gt; c
c =
3.
--&gt; Foo_get()
ans =
4.
</pre>
</div>
<p> It works for variables of primitive type, but also for non-primitive
types: arrays, and structs/classes which are described later. For now,
an example with two global primitive arrays x and y is shown:</p>
<div class="code">
<pre>
%module example
%inline %{
int x[10];
double y[7];
void initArrays()
{
int i;
for (i = 0; i &lt; 10; i++)
x[i] = 1;
for (i = 0; i &lt; 7; i++)
y[i] = 1.0f;
}
%}
</pre>
</div>
<p> It works the same:</p>
<div class="targetlang">
<pre>
--&gt; exec loader.sce
--&gt; initArrays();
--&gt; x_get()
ans =
1. 1. 1. 1. 1. 1. 1. 1. 1. 1.
--&gt; y_set([0:6] / 10);
--&gt; y_get()
ans =
0. 0.1 0.2 0.3 0.4 0.5 0.6
</pre>
</div>
<h3><a name="Scilab_wrapping_constants_and_enums">36.3.5 Constants and
enumerations</a></h3>
<h4><a name="Scilab_wrapping_constants">36.3.5.1 Constants</a></h4>
<p> There is not any constant in Scilab. By default, C/C++ constants are
wrapped as getter functions. For example, for the following constants:</p>
<div class="code">
<pre>
%module example
#define ICONST 42
#define FCONST 2.1828
#define CCONST 'x'
#define CCONST2 '\n'
#define SCONST &quot;Hello World&quot;
#define SCONST2 &quot;\&quot;Hello World\&quot;&quot;
</pre>
</div>
<p> the following getter functions are generated:</p>
<div class="targetlang">
<pre>
--&gt; exec loader.sce;
--&gt; ICONST_get();
ans =
42.
--&gt; FCONST_get();
ans =
2.1828
--&gt; CCONST_get();
ans =
x
--&gt; CCONST2_get();
ans =
--&gt; SCONST_get();
ans =
Hello World
--&gt; SCONST2_get();
ans =
&quot;Hello World&quot;
--&gt; EXPR_get();
ans =
48.5484
--&gt; iconst_get();
ans =
37.
--&gt; fconst_get();
ans =
3.14
</pre>
</div>
<p> There is another mode in which constants are wrapped as Scilab
variables. The variables are easier to use than functions, but the
drawback is that variables are not constant and so can be modified.</p>
<p> This mode can be enabled/disabled at any time in the interface file
with <tt>%scilabconst()</tt>, which works like all the other <a href="#Customization_features">
%feature directives</a>. Use the argument value &quot;1&quot; to enable and &quot;0&quot; to
disable this mode. For example in this mode the previous constants:</p>
<div class="code">
<pre>
%module example
%scilabconst(1);
#define ICONST 42
#define FCONST 2.1828
#define CCONST 'x'
#define CCONST2 '\n'
#define SCONST &quot;Hello World&quot;
#define SCONST2 &quot;\&quot;Hello World\&quot;&quot;
</pre>
</div>
<p> are mapped to Scilab variables, with the same name:</p>
<div class="targetlang">
<pre>
--&gt; exec loader.sce;
--&gt; ICONST
ans =
42
--&gt; FCONST
ans =
2.1828
--&gt; CCONST
ans =
x
--&gt; CCONST2
ans =
--&gt; SCONST
ans =
Hello World
--&gt; SCONST2
ans =
&quot;Hello World&quot;
--&gt; EXPR
ans =
48.5484
--&gt; iconst
ans =
37
--&gt; fconst
ans =
3.14
</pre>
</div>
<h4><a name="Scilab_wrapping_enums">36.3.5.2 Enumerations</a></h4>
<p> The wrapping of enums is the same as for constants. By default,
enums are wrapped as getter functions. For example, with the following
enumeration:</p>
<div class="code">
<pre>%module example
typedef enum { RED, BLUE, GREEN } color;
</pre>
</div>
<p> a getter function will be generated for each value of the
enumeration:</p>
<div class="targetlang">
<pre>
--&gt; exec loader.sce;
--&gt; RED_get()
ans =
0.
--&gt; BLUE_get()
ans =
1.
--&gt; GREEN_get()
ans =
2.
</pre>
</div>
<p> The <tt>%scilabconst()</tt> feature is also available for
enumerations:</p>
<div class="code">
<pre>%module example
%scilabconst(1) color;
typedef enum { RED, BLUE, GREEN } color;
</pre>
</div>
<br><div class="targetlang">
<pre>
--&gt; exec loader.sce;
--&gt; RED
ans =
0.
--&gt; BLUE
ans =
1.
--&gt; GREEN
ans =
2.
</pre>
</div>
<h3><a name="Scilab_wrapping_pointers">36.3.6 Pointers</a></h3>
<p> Pointers are supported by SWIG. A pointer can be returned from a
wrapped C/C++ function, stored in a Scilab variable, and used in input
argument of another C/C++ function.</p>
<p> Also, thanks to the SWIG runtime which stores information about
types, pointer types are tracked between exchanges Scilab and the
native code. Indeed pointer types are stored alongside the pointer
address. A pointer is mapped to a Scilab structure (<a href="https://help.scilab.org/docs/5.5.2/en_US/tlist.html">
tlist</a>), which contains as fields the pointer address and the pointer
type (in fact a pointer to the type information structure in the SWIG
runtime).
<br> Why a native pointer is not mapped to a Scilab pointer (type name:
&quot;pointer&quot;, type ID: 128) ? The big advantage of mapping to a <tt>tlist</tt>
is that it exposes a new type for the pointer in Scilab, type which can
be accessed in Scilab with the <a href="https://help.scilab.org/docs/5.5.2/en_US/typeof.html">
typeof</a> function, and manipulated using the <a href="https://help.scilab.org/docs/5.5.2/en_US/overloading.html">
overloading</a> mechanism.</p>
<p> Notes:</p>
<ul>
<li>type tracking needs the SWIG runtime to be first initialized with
the appropriate function (see the <a href="#Scilab_module_initialization">
Module initialization</a> section).</li>
<li>for any reason, if a wrapped pointer type is unknown (or if the SWIG
runtime is not initialized), SWIG maps it to a Scilab pointer. Also, a
Scilab pointer is always accepted as a pointer argument of a wrapped
function. The drawback is that pointer type is lost.</li>
</ul>
<p> Following is an example of the wrapping of the C <tt>FILE*</tt>
pointer:</p>
<div class="code">
<pre>
%module example
%{
#include &lt;stdio.h&gt;
%}
FILE *fopen(const char *filename, const char *mode);
int fputs(const char *, FILE *);
int fclose(FILE *);
</pre>
</div>
<p> These functions can be used the same way as in C from Scilab:</p>
<div class="targetlang">
<pre>
--&gt; example_Init();
--&gt; f = fopen(&quot;junk&quot;, &quot;w&quot;);
--&gt; typeof(f)
ans =
_p_FILE
--&gt; fputs(&quot;Hello World&quot;, f);
--&gt; fclose(f);
</pre>
</div>
<p> Note: the type name <tt>_p_FILE</tt> which means &quot;pointer to FILE&quot;.</p>
<p> The user of a pointer is responsible for freeing it or, like in the
example, closing any resources associated with it (just as is required
in a C program).</p>
<h4><a name="Scilab_wrapping_pointers_utility_functions">36.3.6.1
Utility functions</a></h4>
<p> As a scripting language, Scilab does not provide functions to
manipulate pointers. However, in some cases it can be useful, such as
for testing or debugging.</p>
<p> SWIG comes with two pointer utility functions:</p>
<ul>
<li><tt>SWIG_this()</tt>: returns the address value of a pointer</li>
<li><tt>SWIG_ptr()</tt>: creates a pointer from an address value</li>
</ul>
<p> Note: a pointer created by <tt>SWIG_ptr()</tt> does not have any
type and is mapped as a Scilab pointer.</p>
<p>Following we use the utility functions on the previous example:</p>
<div class="targetlang">
<pre>
--&gt; f = fopen(&quot;junk&quot;, &quot;w&quot;);
--&gt; fputs(&quot;Hello&quot;, f);
--&gt; addr = SWIG_this(f)
ans =
8219088.
--&gt; p = SWIG_ptr(addr);
--&gt; typeof(p)
ans =
pointer
--&gt; fputs(&quot; World&quot;, p);
--&gt; fclose(f);
</pre>
</div>
<h4><a name="Scilab_wrapping_pointers_null_pointers">36.3.6.2 Null
pointers:</a></h4>
<p> Using the previous <tt>SWIG_this()</tt> and <tt>SWIG_ptr()</tt>, it
is possible to create and check null pointers:</p>
<div class="targetlang">
<pre>
--&gt; p = SWIG_ptr(0);
--&gt; SWIG_this(p) == 0
ans =
T
</pre>
</div>
<h3><a name="Scilab_wrapping_structs">36.3.7 Structures</a></h3>
<p> Structs exist in Scilab, but C structs are not (at least in this
version of SWIG) mapped to Scilab structs. A C structure is wrapped
through low-level accessor functions, i.e. functions that give access
to the member variables of this structure. In Scilab, a structure is
manipulated through a pointer which is passed as an argument to the
accessor functions.</p>
<p> Let's see it on an example of a struct with two members:</p>
<div class="code">
<pre>
%module example
%inline %{
typedef struct {
int x;
int arr[4];
} Foo;
%}
</pre>
</div>
<p> Several functions are generated:</p>
<ul>
<li>a constructor function <tt>new_Foo()</tt> which returns a pointer to
a newly created struct <tt>Foo</tt>.</li>
<li>two member getter functions <tt>Foo_x_get()</tt>, <tt>Foo_arr_get()</tt>
, to get the values of <tt>x</tt> and <tt>y</tt> for the struct pointer
(provided as the first parameter to these functions)</li>
<li>two member setter functions <tt>Foo_x_set()</tt>, <tt>Foo_arr_set()</tt>
, to set the values of <tt>x</tt> and <tt>y</tt> for the struct pointer
(provided as the first parameter to these functions).</li>
<li>a destructor function <tt>delete_Foo()</tt> to release the struct
pointer.</li>
</ul>
<p> Usage example:</p>
<div class="targetlang">
<pre>
--&gt; f = new_Foo();
--&gt; Foo_x_set(f, 100);
--&gt; Foo_x_get(f)
ans =
100.
--&gt; Foo_arr_set(f, [0:3]);
--&gt; Foo_arr_get(f)
ans =
0. 1. 2. 3.
--&gt; delete_Foo(f);
</pre>
</div>
<p> Members of a structure that are also structures are also accepted
and wrapped as a pointer:</p>
<div class="code">
<pre>
%module example
%inline %{
typedef struct {
int x;
} Bar;
typedef struct {
Bar b;
} Foo;
%}
</pre>
</div>
<br><div class="targetlang">
<pre>
--&gt; b = new_Bar();
--&gt; Bar_x_set(b, 20.);
--&gt; f = new_Foo();
--&gt; Foo_b_set(f, b);
--&gt; b2 = Foo_b_get(f);
--&gt; Bar_x_get(b2);
ans =
20.
</pre>
</div>
<p> Note: the pointer to the struct works as described in <a href="#Scilab_wrapping_pointers">
Pointers</a>. For example, the type of the struct pointer can be get
with <tt>typeof</tt>, as following:</p>
<div class="targetlang">
<pre>
--&gt; example_Init();
--&gt; b = new_Bar();
--&gt; typeof(b)
ans =
_p_Bar
--&gt; delete_Bar(b);
</pre>
</div>
<h3><a name="Scilab_wrapping_cpp_classes">36.3.8 C++ classes</a></h3>
<p> Classes do not exist in Scilab. The classes are wrapped the same way
as structs. Low-level accessor functions are generated for class
members. Also, constructor and destructor functions are generated to
create and destroy an instance of the class.</p>
<p> For example, the following class:</p>
<div class="code">
<pre>
%module example
%inline %{
class Point {
public:
int x, y;
Point(int _x, int _y) : x(_x), y(_y) {}
double distance(const Point&amp; rhs) {
return sqrt(pow(x-rhs.x, 2) + pow(y-rhs.y, 2));
}
void set(int _x, int _y) {
x=_x;
y=_y;
}
};
%}
</pre>
</div>
<p> can be used in Scilab like this:</p>
<div class="targetlang">
<pre>
--&gt; p1 = Point_new(3, 5);
--&gt; p2 = Point_new(1, 2);
--&gt; p1.distance(p2)
ans =
3.6056
--&gt; delete_Point(p1);
--&gt; delete_Point(p2);
</pre>
</div>
<p> Note: like structs, class pointers are mapped as described in <a href="#Scilab_wrapping_pointers">
Pointers</a>. Let's give an example which shows that each class pointer
type is a new type in Scilab that can be used for example (through <a href="https://help.scilab.org/docs/5.5.2/en_US/overloading.html">
overloading</a>) to implement a custom print for the <tt>Point</tt>
class:</p>
<div class="targetlang">
<pre>
--&gt; function %_p_Point_p(p)
--&gt; mprintf('[%d, %d]\n', Point_x_get(p), Point_y_get(p));
--&gt; endfunction
--&gt; example_Init();
--&gt; p = new_Point(1, 2)
p =
[1, 2]
--&gt; delete_Point(p);
</pre>
</div>
<h3><a name="Scilab_wrapping_cpp_inheritance">36.3.9 C++ inheritance</a></h3>
<p> Inheritance is supported. SWIG knows the inheritance relationship
between classes.</p>
<p> A function is only generated for the class in which it is actually
declared. But if one of its parameters is a class, any instance of a
derived class is accepted as the argument.</p>
<p> This mechanism also applies for accessor functions: they are
generated only in the class in which they are defined. But any instance
of a derived class can be used as the argument to these accessor
functions.</p>
<p> For example, let's take a base class <tt>Shape</tt> and two derived
classes <tt>Circle</tt> and <tt>Square</tt>:</p>
<div class="code">
<pre>
%module example
%inline %{
class Shape {
public:
double x, y;
void set_location(double _x, double _y) { x = _x; y = _y; }
virtual double get_perimeter() { return 0; };
};
class Circle : public Shape {
public:
int radius;
Circle(int _radius): radius(_radius) {};
virtual double get_perimeter() { return 6.28 * radius; }
};
class Square : public Shape {
public:
int size;
Square(int _size): size(_size) {};
virtual double get_perimeter() { return 4 * size; }
};
%}
</pre>
</div>
<p> To set the location of the <tt>Circle</tt>, we have to use the
function <tt>set_location()</tt> of the parent <tt>Shape</tt>. But we
can use either use the <tt>get_perimeter()</tt> function of the parent
class or the derived class:</p>
<div class="targetlang">
<pre>
--&gt; c = new_Circle(3);
--&gt; Shape_set_location(c, 2, 3);
--&gt; Shape_x_get(c)
ans =
2.
--&gt; Circle_get_perimeter(c)
ans =
18.84
--&gt; Shape_get_perimeter(c)
ans =
18.84
</pre>
</div>
<h3><a name="Scilab_wrapping_cpp_overloading">36.3.10 C++ overloading</a>
</h3>
<p> As explained in <a href="#SWIGPlus_overloaded_methods">Overloaded
functions and methods</a> SWIG provides support for overloaded
functions and constructors.</p>
<p>As SWIG knows pointer types, the overloading works also with pointer
types, here is an example with a function <tt>magnify</tt> overloaded
for the previous classes <tt>Shape</tt> and <tt>Circle</tt>:</p>
<div class="code">
<pre>
%module example
void magnify(Square *square, double factor) {
square-&gt;size *= factor;
};
void magnify(Circle *circle, double factor) {
square-&gt;radius *= factor;
};
</pre>
</div><div class="targetlang">
<pre>
--&gt; example_Init();
--&gt; c = new_Circle(3);
--&gt; s = new_Square(2);
--&gt; magnify(c, 10);
--&gt; Circle_get_radius(c)
ans =
30;
--&gt; magnify(s, 10);
--&gt; Square_get_size(s)
ans =
20;
</pre>
</div>
<h3><a name="Scilab_wrapping_pointers_references_values_arrays">36.3.11
Pointers, references, values, and arrays</a></h3>
<p> In C++ objects can be passed by value, pointer, reference, or by an
array:</p>
<div class="code">
<pre>
%module example
%{
#include &lt;sciprint.h&gt;
%}
%inline %{
class Foo {
public:
Foo(int _x) : x(_x) {}
int x;
};
void spam1(Foo *f) { sciprint(&quot;%d\n&quot;, f-&gt;x); } // Pass by pointer
void spam2(Foo &amp;f) { sciprint(&quot;%d\n&quot;, f.x); } // Pass by reference
void spam3(Foo f) { sciprint(&quot;%d\n&quot;, f.x); } // Pass by value
void spam4(Foo f[]) { sciprint(&quot;%d\n&quot;, f[0].x); } // Array of objects
%}
</pre>
</div>
<p> In SWIG, there is no real distinction between these. So in Scilab,
it is perfectly legal to do this:</p>
<div class="targetlang">
<pre>
--&gt; f = new_Foo()
--&gt; spam1(f)
3
--&gt; spam2(f)
3
--&gt; spam3(f)
3
--&gt; spam4(f)
3
</pre>
</div>
<p> Similar behaviour occurs for return values. For example, if you had
functions like this:</p>
<div class="code">
<pre>
Foo *spam5();
Foo &amp;spam6();
Foo spam7();
</pre>
</div>
<p> All these functions will return a pointer to an instance of <tt>Foo</tt>
. As the function <tt>spam7</tt> returns a value, new instance of <tt>
Foo</tt> has to be allocated, and a pointer on this instance is
returned.</p>
<h3><a name="Scilab_wrapping_cpp_templates">36.3.12 C++ templates</a></h3>
<p> As in other languages, function and class templates are supported in
SWIG Scilab.</p>
<p> You have to tell SWIG to create wrappers for a particular template
instantiation. The <tt>%template</tt> directive is used for this
purpose. For example:</p>
<div class="code">
<pre>
%module example
template&lt;class T1, class T2, class T3&gt;
struct triplet {
T1 first;
T2 second;
T3 third;
triplet(const T1&amp; a, const T2&amp; b, const T3&amp; c) {
third = a; second = b; third = c;
}
};
%template(IntTriplet) triplet&lt;int, int, int&gt;;
</pre>
</div>
<p> Then in Scilab:</p>
<div class="targetlang">
<pre>
--&gt; t = new_IntTriplet(3, 4, 1);
--&gt; IntTriplet_first_get(t)
ans =
3.
--&gt; IntTriplet_second_get(t)
ans =
4.
--&gt; IntTriplet_third_get(t)
ans =
1.
--&gt; delete_IntTriplet(t);
</pre>
</div>
<p> More details on template support can be found in the <a href="#SWIGPlus_nn30">
templates</a> documentation.</p>
<h3><a name="Scilab_wrapping_cpp_operators">36.3.13 C++ operators</a></h3>
<p> C++ operators are partially supported. Operator overloading exists
in Scilab, but a C++ operator is not (in this version) wrapped by SWIG
as a Scilab operator, but as a function. It is not automatic, you have
to rename each operator (with the instruction <tt>%rename</tt>) with
the suitable wrapper name.</p>
<p> Let's see it with an example of class with two operators <tt>+</tt>
and <tt>double()</tt>:</p>
<div class="code">
<pre>
%module example
%rename(plus) operator +;
%rename(toDouble) operator double();
%inline %{
class Complex {
public:
Complex(double re, double im) : real(re), imag(im) {};
Complex operator+(const Complex&amp; other) {
double result_real = real + other.real;
double result_imaginary = imag + other.imag;
return Complex(result_real, result_imaginary);
}
operator double() { return real; }
private:
double real;
double imag;
};
%}
</pre>
</div>
<br><div class="targetlang">
<pre>
--&gt; c1 = new_Complex(3, 7);
--&gt; c2 = Complex_plus(c, new_Complex(1, 1));
--&gt; Complex_toDouble(c2)
ans =
4.
</pre>
</div>
<h3><a name="Scilab_wrapping_cpp_namespaces">36.3.14 C++ namespaces</a></h3>
<p> SWIG is aware of C++ namespaces, but does not use it for wrappers.
The module is not broken into submodules, nor do namespace appear in
functions names. All the namespaces are all flattened in the module.
For example with one namespace <tt>Foo</tt>:</p>
<div class="code">
<pre>
%module example
%inline %{
namespace foo {
int fact(int n) {
if (n &gt; 1)
return n * fact(n-1);
else
return 1;
}
struct Vector {
double x, y, z;
};
};
%}
</pre>
</div>
<p> In Scilab, there is no need to the specify the <tt>Foo</tt>
namespace:</p>
<div class="targetlang">
<pre>
--&gt; fact(3)
ans =
6.
--&gt; v = new_Vector();
--&gt; Vector_x_set(v, 3.4);
--&gt; Vector_y_get(v)
ans =
0.
</pre>
</div>
<p> If your program has more than one namespace, name conflicts can be
resolved using <tt>%rename</tt>. For example:</p>
<div class="code">
<pre>
%rename(Bar_spam) Bar::spam;
namespace Foo {
int spam();
}
namespace Bar {
int spam();
}
</pre>
</div>
<p> Note: the <a href="#SWIGPlus_nspace">nspace</a> feature is not
supported.</p>
<h3><a name="Scilab_wrapping_cpp_exceptions">36.3.15 C++ exceptions</a></h3>
<p> Scilab does not natively support exceptions, but has errors. When an
exception is thrown, SWIG catches it, and sets a Scilab error. An error
message is displayed in Scilab. For example:</p>
<div class="code">
<pre>
%module example
%inline %{
void throw_exception() throw(char const *) {
throw &quot;Bye world !&quot;;
}
%}
</pre>
</div>
<br><div class="targetlang">
<pre>
--&gt;throw_exception()
!--error 999
SWIG/Scilab: Exception (char const *) occurred: Bye world !
</pre>
</div>
<p> Scilab has a <tt>try-catch</tt> mechanism (and a similar instruction
<tt>execstr()</tt>) to handle exceptions. It can be used with the <tt>
lasterror()</tt> function as following:</p>
<div class="targetlang">
<pre>
--&gt; execstr('throw_exception()', 'errcatch');
ans =
999.
--&gt; lasterror()
ans =
SWIG/Scilab: Exception (char const *) occurred: Bye world !
</pre>
</div>
<p> If the function has a <tt>throw</tt> exception specification, SWIG
can automatically map the exception type and set an appropriate Scilab
error message. It works for a few primitive types, and also for STL
exceptions (the library <tt>std_except.i</tt> has to be included to get
the STL exception support):</p>
<div class="code">
<pre>
%module example
%include &lt;std_except.i&gt;
%inline %{
void throw_int() throw(int) {
throw 12;
}
void throw_stl_invalid_arg(int i) throw(std::invalid_argument) {
if (i &lt; 0)
throw std::invalid_argument(&quot;argument is negative.&quot;);
}
%}
</pre>
</div>
<br><div class="targetlang">
<pre>
--&gt; throw_int();
!--error 999
SWIG/Scilab: Exception (int) occurred: 12
--&gt;throw_stl_invalid_arg(-1);
!--error 999
SWIG/Scilab: ValueError: argument is negative.
</pre>
</div>
<p> More complex or custom exception types require specific exception
typemaps to be implemented in order to specifically handle a thrown
type. See the <a href="#SWIGPlus">SWIG C++ documentation</a> for more
details.</p>
<h3><a name="Scilab_wrapping_cpp_stl">36.3.16 C++ STL</a></h3>
<p> The Standard Template Library (STL) is partially supported. See <a href="#Scilab_typemaps_stl">
STL</a> for more details.</p>
<h2><a name="Scilab_typemaps">36.4 Type mappings and libraries</a></h2>
<h3><a name="Scilab_typemaps_primitive_types">36.4.1 Default primitive
type mappings</a></h3>
<p> The following table provides the equivalent Scilab type for C/C++
primitive types.</p>
<div class="table">
<table border="1" summary="Scilab default primitive type mappings">
<tr><td><b>C/C++ type</b></td><td><b>Scilab type</b></td></tr>
<tr><td>bool</td><td>boolean</td></tr>
<tr><td>char</td><td>string</td></tr>
<tr><td>signed char</td><td>double or int8</td></tr>
<tr><td>unsigned char</td><td>double or uint8</td></tr>
<tr><td>short</td><td>double or int16</td></tr>
<tr><td>unsigned short</td><td>double or uint16</td></tr>
<tr><td>int</td><td>double or int32</td></tr>
<tr><td>unsigned int</td><td>double or uint32</td></tr>
<tr><td>long</td><td>double or int32</td></tr>
<tr><td>unsigned long</td><td>double or uint32</td></tr>
<tr><td>signed long long</td><td>not supported in Scilab 5.x</td></tr>
<tr><td>unsigned long long</td><td>not supported in Scilab 5.x</td></tr>
<tr><td>float</td><td>double</td></tr>
<tr><td>double</td><td>double</td></tr>
<tr><td>char * or char[]</td><td>string</td></tr>
</table>
</div>
<p> Notes:</p>
<ul>
<li>In Scilab the <tt>double</tt> type is far more used than any integer
type. This is why integer values (<tt>int32</tt>, <tt>uint32</tt>, ...)
are automatically converted to Scilab <tt>double</tt> values when
marshalled from C into Scilab. Additionally on input to a C function,
Scilab <tt>double</tt> values are converted into the related integer
type.</li>
<li> When an integer is expected, if the input is a double, the value
must be an integer, i.e. it must not have any decimal part, otherwise a
SWIG value error occurs.</li>
<li> In SWIG for Scilab 5.x, the <tt>long long</tt> type is not
supported, since Scilab 5.x does not have a 64-bit integer type. The
default behaviour is for SWIG to generate code that will give a runtime
error if <tt>long long</tt> type arguments are used from Scilab.</li>
</ul>
<h3><a name="Scilab_typemaps_arrays">36.4.2 Arrays</a></h3>
<p> Typemaps are available by default for arrays. Primitive type arrays
are automatically converted to/from Scilab matrices. Typemaps are also
provided to handle members of a struct or class that are arrays.</p>
<p> In input, the matrix is usually one-dimensional (it can be either a
row or column vector). But it can also be a two-dimensional matrix.
Warning: in Scilab, the values are column-major ordered, unlike in C,
which is row-major ordered.</p>
<p> The type mappings used for arrays is the same for primitive types,
described <a href="#Scilab_typemaps_primitive_types">earlier</a>. This
means that, if needed, a Scilab <tt>double</tt> vector is converted in
input into the related C integer array and this C integer array is
automatically converted on output into a Scilab <tt>double</tt> vector.
Note that unlike scalars, no control is done for arrays when a <tt>
double</tt> is converted into an integer.</p>
<p> The following example illustrates all this:</p>
<div class="code">
<pre>
%module example
%#include &lt;stdio.h&gt;
%inline %{
void printArray(int values[], int len) {
int i = 0;
for (i = 0; i &lt; len; i++) {
printf(&quot;%s %d %s&quot;, i==0?&quot;[&quot;:&quot;&quot;, values[i], i==len-1?&quot;]\n&quot;:&quot;&quot;);
}
}
%}
</pre>
</div>
<br><div class="targetlang">
<pre>
--&gt; printArray([0 1 2 3], 4)
[ 0 1 2 3 ]
--&gt; printArray([0.2; -1.8; 2; 3.7], 4)
[ 0 -1 2 3 ]
--&gt; printArray([0 1; 2 3], 4)
[ 0 2 1 3 ]
--&gt; printArray([0; 1; 2; 3], 4)
[ 0 1 2 3 ]
</pre>
</div>
<h3><a name="Scilab_typemaps_pointer-to-pointers">36.4.3
Pointer-to-pointers</a></h3>
<p> There are no specific typemaps for pointer-to-pointers, they are
mapped as pointers in Scilab.</p>
<p> Pointer-to-pointers are sometimes used to implement matrices in C.
The following is an example of this:</p>
<div class="code">
<pre>
%module example
%inline %{
// Returns the matrix [1 2; 3 4];
double **create_matrix() {
double **M;
int i;
M = (double **) malloc(2 * sizeof(double *));
for (i = 0; i &lt; 2; i++) {
M[i] = (double *) malloc(2 * sizeof(double));
M[i][0] = 2 * i + 1;
M[i][1] = 2 * i + 2;
}
return M;
}
// Gets the item M(i, j) value
double get_matrix(double **M, int i, int j) {
return M[i][j];
}
// Sets the item M(i, j) value to be val
void set_matrix(double **M, int i, int j, double val) {
M[i][j] = val;
}
// Prints a matrix (2, 2) to console
void print_matrix(double **M, int nbRows, int nbCols) {
int i, j;
for (i = 0; i &lt; 2; i++) {
for (j = 0; j &lt; 2; j++) {
printf(&quot;%3g &quot;, M[i][j]);
}
printf(&quot;\n&quot;);
}
}
%}
</pre>
</div>
<p> These functions are used like this in Scilab:</p>
<div class="targetlang">
<pre>
--&gt; m = create_matrix();
--&gt; print_matrix(m);
1. 2.
3. 4.
--&gt; set_matrix(m, 1, 1, 5.);
--&gt; get_matrix(m, 1, 1)
ans =
5.
</pre>
</div>
<h3><a name="Scilab_typemaps_matrices">36.4.4 Matrices</a></h3>
<p> The <tt>matrix.i</tt> library provides a set of typemaps which can
be useful when working with one-dimensional and two-dimensional
matrices.</p>
<p> In order to use this library, just include it in the interface file:</p>
<div class="code">
<pre>
%include &lt;matrix.i&gt;
</pre>
</div>
<p> Several typemaps are available for the common Scilab matrix types:</p>
<ul>
<li><tt>double</tt></li>
<li><tt>int</tt></li>
<li><tt>char *</tt></li>
<li><tt>bool</tt></li>
</ul>
<p> For example: for a matrix of <tt>int</tt>, we have the typemaps, for
input:</p>
<ul>
<li><tt>(int *IN, int IN_ROWCOUNT, int IN_COLCOUNT)</tt></li>
<li><tt>(int IN_ROWCOUNT, int IN_COLCOUNT, int *IN)</tt></li>
<li><tt>(int *IN, int IN_SIZE)</tt></li>
<li><tt>(int IN_SIZE, int *IN)</tt></li>
</ul>
<p> and output:</p>
<ul>
<li><tt>(int **OUT, int *OUT_ROWCOUNT, int *OUT_COLCOUNT)</tt></li>
<li><tt>(int *OUT_ROWCOUNT, int *OUT_COLCOUNT, int **OUT)</tt></li>
<li><tt>(int **OUT, int *OUT_SIZE)</tt></li>
<li><tt>(int *OUT_SIZE, int **OUT)</tt></li>
</ul>
<p> They marshall a Scilab matrix type into the appropriate 2 or 3 C
parameters. The following is an example using the typemaps in this
library:</p>
<div class="code">
<pre>
%module example
%include &lt;matrix.i&gt;
%apply (int *IN, int IN_ROWCOUNT, int IN_COLCOUNT) { (int *matrix, int matrixNbRow, int matrixNbCol) };
%apply (int **OUT, int *OUT_ROWCOUNT, int *OUT_COLCOUNT) { (int **outMatrix, int *outMatrixNbRow, int *outMatrixNbCol) };
%inline %{
void absolute(int *matrix, int matrixNbRow, int matrixNbCol,
int **outMatrix, int *outMatrixNbRow, int *outMatrixNbCol) {
int i, j;
*outMatrixNbRow = matrixNbRow;
*outMatrixNbCol = matrixNbCol;
*outMatrix = malloc(matrixNbRow * matrixNbCol * sizeof(int));
for (i=0; i &lt; matrixNbRow * matrixNbCol; i++) {
(*outMatrix)[i] = matrix[i] &gt; 0 ? matrix[i]:-matrix[i];
}
}
%}
</pre>
</div>
<br><div class="targetlang">
<pre>
--&gt; absolute([-0 1 -2; 3 4 -5])
ans =
0. 1. 2.
3. 4. 5.
</pre>
</div>
<p> The remarks made earlier for arrays also apply here:</p>
<ul>
<li>The values of matrices in Scilab are column-major orderered,</li>
<li>There is no control while converting <tt>double</tt> values to
integers, <tt>double</tt> values are truncated without any checking or
warning.</li>
</ul>
<h3><a name="Scilab_typemaps_stl">36.4.5 STL</a></h3>
<p> The STL library wraps some containers defined in the STL (Standard
Template Library), so that they can be manipulated in Scilab. This
library also provides the appropriate typemaps to use the containers in
functions and variables.</p>
<p> The list of wrapped sequence containers are:</p>
<ul>
<li><tt>std::vector</tt></li>
<li><tt>std::list</tt></li>
<li><tt>std::deque</tt></li>
</ul>
<p> And associative containers are:</p>
<ul>
<li><tt>std::set</tt></li>
<li><tt>std::multiset</tt></li>
</ul>
<p> Typemaps are available for the following container types:</p>
<ul>
<li><tt>double</tt></li>
<li><tt>float</tt></li>
<li><tt>int</tt></li>
<li><tt>string</tt></li>
<li><tt>bool</tt></li>
<li><tt>pointer</tt></li>
</ul>
<p> Containers of other item types are not supported. Using them does
not break compilation, but provokes a runtime error. Containers of enum
are not supported yet.</p>
<p> In order to use the STL, the library must first be included in the
SWIG interface file:</p>
<div class="code">
<pre>
%include &lt;stl.i&gt;
</pre>
</div>
<p>Then for each container used, the appropriate template must be
instantiated, in the <tt>std</tt> namespace:<div class="code">
<pre>
namespace std {
%template(IntVector) vector&lt;int&gt;;
%template(DoubleVector) vector&lt;double&gt;;
}
</pre>
</div></p>
<p> Additionally, the module initialization function has to be executed
first in Scilab, so that all the types are known to Scilab. See the <a href="#Scilab_module_initialization">
Module initialization</a> section for more details.</p>
<p> Because in Scilab matrices exist for basic types only, a sequence
container of pointers is mapped to a Scilab list. For other item types
(double, int, string...) the sequence container is mapped to a Scilab
matrix.</p>
<p> The first example below shows how to create a vector (of <tt>int</tt>
) in Scilab, add some values to the vector and pass it as an argument of
a function. It also shows, thanks to the typemaps, that we can also
pass a Scilab matrix of values directly into the function:</p>
<div class="code">
<pre>
%module example
%include &lt;stl.i&gt;
namespace std {
%template(IntVector) vector&lt;int&gt;;
}
%{
#include &lt;numeric&gt;
%}
%inline %{
double average(std::vector&lt;int&gt; v) {
return std::accumulate(v.begin(), v.end(), 0.0) / v.size();
}
%}
</pre>
</div>
<br><div class="targetlang">
<pre>
--&gt; example_Init();
--&gt; v = new_IntVector();
--&gt; for i = 1:4
--&gt; IntVector_push_back(v, i);
--&gt; end;
--&gt; average(v)
ans =
2.5
--&gt; average([0 1 2 3])
ans =
2.5
--&gt; delete_IntVector();
</pre>
</div>
<p> In the second example, a set of struct (<tt>Person</tt>) is wrapped.
A function performs a search in this set, and returns a subset. As one
can see, the result in Scilab is a list of pointers:</p>
<div class="code">
<pre>
%module example
%include &lt;stl.i&gt;
%{
#include &lt;string&gt;
%}
%inline %{
struct Person {
Person(std::string _name, int _age) : name(_name), age(_age) {};
std::string name;
int age;
};
typedef Person * PersonPtr;
%}
namespace std {
%template(PersonPtrSet) set&lt;PersonPtr&gt;;
}
%inline %{
std::set&lt;PersonPtr&gt; findPersonsByAge(std::set&lt;PersonPtr&gt; persons, int minAge, int maxAge) {
std::set&lt;PersonPtr&gt; foundPersons;
for (std::set&lt;PersonPtr&gt;::iterator it = persons.begin(); it != persons.end(); it++) {
if (((*it)-&gt;age &gt;= minAge) &amp;&amp; ((*it)-&gt;age &lt;= maxAge)) {
foundPersons.insert(*it);
}
}
return foundPersons;
}
%}
</pre>
</div>
<br><div class="targetlang">
<pre>
--&gt; example_Init();
--&gt; joe = new_Person(&quot;Joe&quot;, 25);
--&gt; susan = new_Person(&quot;Susan&quot;, 32);
--&gt; bill = new_Person(&quot;Bill&quot;, 50);
--&gt; p = new_PersonPtrSet();
--&gt; PersonPtrSet_insert(p, susan);
--&gt; PersonPtrSet_insert(p, joe);
--&gt; PersonPtrSet_insert(p, bill);
--&gt; l = findPersonsByAge(p, 20, 40);
--&gt; size(l)
ans =
2.
--&gt; Person_name_get(l(1))
ans =
Susan
--&gt; Person_name_get(l(2))
ans =
Joe
--&gt; delete_PersonPtrSet(p);
</pre>
</div>
<h2><a name="Scilab_module_initialization">36.5 Module initialization</a>
</h2>
<p> The wrapped module contains an initialization function to:</p>
<ul>
<li>initialize the SWIG runtime, needed for pointer type tracking or
when working with the STL</li>
<li>initialize the module constants and enumerations declared with <tt>
%scilabconst()</tt></li>
</ul>
<p> This initialization function should be executed at the start of a
script, before the wrapped library has to be used.</p>
<p> The function has the name of the module suffixed by <tt>_Init</tt>.
For example, to initialize the module <tt>example</tt>:</p>
<div class="targetlang">
<pre>
--&gt; example_Init();
</pre>
</div>
<h2><a name="Scilab_building_modes">36.6 Building modes</a></h2>
<p> The mechanism to load an external module in Scilab is called<i>
Dynamic Link</i> and works with dynamic modules (or shared libraries, <tt>
.so</tt> files).</p>
<p> To produce a dynamic module, when generating the wrapper, there are
two possibilities, or build modes:</p>
<ul>
<li>the <tt>nobuilder</tt> mode, this is the default mode in SWIG. The
user is responsible of the build.</li>
<li>the <tt>builder</tt> mode. In this mode, Scilab is responsible of
building.</li>
</ul>
<h3><a name="Scilab_building_modes_nobuilder_mode">36.6.1 No-builder
mode</a></h3>
<p> In this mode, used by default, SWIG generates the wrapper sources,
which have to be manually compiled and linked. A loader script <tt>
loader.sce</tt> is also produced, this one is executed further in Scilab
to load the module.</p>
<p> This mode is the best option to use when you have to integrate the
module build into a larger build process.</p>
<h3><a name="Scilab_building_modes_builder_mode">36.6.2 Builder mode</a></h3>
<p> In this mode, in addition to the wrapper sources, SWIG produces a
builder Scilab script (<tt>builder.sce</tt>), which is executed in
Scilab to build the module. In a few words, the Scilab <tt>ilib_build()</tt>
command is used, which produces the shared library file, and the loader
script <tt>loader.sce</tt> (and also a cleaner script <tt>cleaner.sce</tt>
).</p>
<p> An advantage of this mode is that it hides all the complexity of the
build and other platform issues. Also it allows the module to conform
to a Scilab external module convention which is that an external module
should be simply built by calling a builder script.</p>
<p> The builder mode is activated with the <tt>-builder</tt> SWIG
option. In this mode, the following SWIG options may be used to setup
the build:</p>
<ul>
<li><tt><b>-buildersources</b></tt>: to add sources to the build
(several files must be separated by a comma)</li>
<li><tt><b>-buildercflags</b></tt>: to add flags to the builder compiler
flags, for example to set library dependencies include paths</li>
<li><tt><b>-builderldflags</b></tt>: to add flags to the linker flags,
for example to set library dependency names and paths</li>
</ul>
<p> Let's give an example how to build a module <tt>example</tt>,
composed of two sources, and using a library dependency:</p>
<ul>
<li>the sources are <tt>baa1.c</tt> and <tt>baa2.c</tt> (and are stored
in the current directory)</li>
<li>the library is <tt>libfoo</tt> in <tt>/opt/foo</tt> (headers stored
in <tt>/opt/foo/include</tt>, and shared library in <tt>/opt/foo/lib</tt>
)</li>
</ul>
<p> The command is:</p>
<div class="shell">
<pre>
$ swig -scilab -builder -buildercflags -I/opt/foo/include -builderldflags &quot;-L/opt/foo/lib -lfoo&quot; -buildersources baa1.cxx, baa2.cxx example.i
</pre>
</div>
<h2><a name="Scilab_generated_scripts">36.7 Generated scripts</a></h2>
<p> In this part we give some details about the generated Scilab
scripts.</p>
<h3><a name="Scilab_generated_scripts_builder_script">36.7.1 Builder
script</a></h3>
<p> <tt>builder.sce</tt> is the name of the builder script generated by
SWIG in <tt>builder</tt> mode. It contains code like this:</p>
<div class="code">
<pre>
ilib_name = &quot;examplelib&quot;;
files = [&quot;example_wrap.c&quot;];
libs = [];
table = [&quot;fact&quot;, &quot;_wrap_fact&quot;;&quot;Foo_set&quot;, &quot;_wrap_Foo_set&quot;;&quot;Foo_get&quot;, &quot;_wrap_Foo_get&quot;;];
ilib_build(ilib_name, table, files, libs);
</pre>
</div>
<p> <tt>ilib_build(lib_name, table, files, libs)</tt> is used to create
shared libraries, and to generate a loader file used to dynamically
load the shared library into Scilab.</p>
<ul>
<li><tt><b>ilib_name</b></tt>: a character string, the generic name of
the library without path and extension.</li>
<li><tt><b>files</b></tt>: string matrix containing objects files needed
for shared library creation.</li>
<li><tt><b>libs</b></tt>: string matrix containing extra libraries
needed for shared library creation.</li>
<li><tt><b>table</b></tt>: two column string matrix containing a table
of pairs of 'scilab function name', 'C function name'.</li>
</ul>
<h3><a name="Scilab_generated_scripts_loader_script">36.7.2 Loader
script</a></h3>
<p> The loader script is used to load in Scilab all the module
functions. When loaded, these functions can be used as other Scilab
functions.</p>
<p> The loader script <tt>loader.sce</tt> contains code similar to:</p>
<div class="code">
<pre>
// ------------------------------------------------------
// generated by builder.sce: Please do not edit this file
// ------------------------------------------------------
libexamplelib_path = get_file_path('loader.sce');
list_functions = [ 'fact';
'Foo_set';
'Foo_get';
];
addinter(libexamplelib_path+'/libexamplelib.so', 'libexamplelib', list_functions);
// remove temp. variables on stack
clear libexamplelib_path;
clear list_functions;
clear get_file_path;
// ------------------------------------------------------
</pre>
</div>
<p> <tt>addinter(files, spname, fcts)</tt> performs dynamic linking of a
compiled C interface function.</p>
<ul>
<li><tt><b>files</b></tt>: a character string or a vector of character
strings defining the object files (containing the C interface
functions) to link with.</li>
<li><tt><b>spname</b></tt>: a character string. Name of interface
routine entry point.</li>
<li><tt><b>fcts</b></tt>: vector of character strings. The name of new
Scilab function.</li>
</ul>
<h3><a name="Scilab_generated_scripts_gateway">36.7.3 Gateway XML files</a>
</h3>
<p>If you need to post-process the entry points, Scilab gateway files
are XML files that can be used to retrieve all SWIG-generated entry
points. With these XML files you can write your own <tt>
builder_swig.sce</tt> file to add custom Scilab for building or linking
the generated code. Documentation stubs can also be generated thanks to
these function listings.</p>
<p>As an example, for a SWIG <a href="Modules.html">module</a> named <tt>
fmuswig</tt> the Scilab code below can be used to store all
SWIG-generated functions in a variable named <tt>funs</tt>.</p>
<div class="code">
<pre>
// src_swig_path is a path to the directory containing the fmuswig.i file
doc = xmlRead(src_swig_path + &quot;/fmuswig_gateway.xml&quot;);
names = xmlAsText(xmlXPath(doc, &quot;//gateway/@name&quot;));
funs = xmlAsText(xmlXPath(doc, &quot;//gateway/@function&quot;));
xmlDelete(doc);
</pre>
</div>
<h2><a name="Scilab_other_resources">36.8 Other resources</a></h2>
<ul>
<li>Example use cases can be found in the <tt>Examples/scilab</tt>
directory.</li>
<li>The test suite in the <tt>Examples/test-suite/scilab</tt> can be
another source of useful use cases.</li>
<li>The <a href="https://help.scilab.org/api_scilab.html">Scilab API</a>
is used in the generated code and is a useful reference when examining
the output.</li>
<li>This <a href="https://wiki.scilab.org/howto/Create%20a%20toolbox">
guide</a> describes the Scilab external modules structure and files, in
particular the files that are generated by SWIG for Scilab.</li>
</ul>
<HR NOSHADE>
<h1><a name="Tcl">37 SWIG and Tcl</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Tcl_nn2">Preliminaries</a>
<ul>
<li><a href="#Tcl_nn3">Getting the right header files</a></li>
<li><a href="#Tcl_nn4">Compiling a dynamic module</a></li>
<li><a href="#Tcl_nn5">Static linking</a></li>
<li><a href="#Tcl_nn6">Using your module</a></li>
<li><a href="#Tcl_nn7">Compilation of C++ extensions</a></li>
<li><a href="#Tcl_nn8">Compiling for 64-bit platforms</a></li>
<li><a href="#Tcl_nn9">Setting a package prefix</a></li>
<li><a href="#Tcl_nn10">Using namespaces</a></li>
</ul>
</li>
<li><a href="#Tcl_nn11">Building Tcl/Tk Extensions under Windows 95/NT</a>
<ul>
<li><a href="#Tcl_nn12">Running SWIG from Developer Studio</a></li>
<li><a href="#Tcl_nn13">Using NMAKE</a></li>
</ul>
</li>
<li><a href="#Tcl_nn14">A tour of basic C/C++ wrapping</a>
<ul>
<li><a href="#Tcl_nn15">Modules</a></li>
<li><a href="#Tcl_nn16">Functions</a></li>
<li><a href="#Tcl_nn17">Global variables</a></li>
<li><a href="#Tcl_nn18">Constants and enums</a></li>
<li><a href="#Tcl_nn19">Pointers</a></li>
<li><a href="#Tcl_nn20">Structures</a></li>
<li><a href="#Tcl_nn21">C++ classes</a></li>
<li><a href="#Tcl_nn22">C++ inheritance</a></li>
<li><a href="#Tcl_nn23">Pointers, references, values, and arrays</a></li>
<li><a href="#Tcl_nn24">C++ overloaded functions</a></li>
<li><a href="#Tcl_nn25">C++ operators</a></li>
<li><a href="#Tcl_nn26">C++ namespaces</a></li>
<li><a href="#Tcl_nn27">C++ templates</a></li>
<li><a href="#Tcl_nn28">C++ Smart Pointers</a></li>
</ul>
</li>
<li><a href="#Tcl_nn29">Further details on the Tcl class interface</a>
<ul>
<li><a href="#Tcl_nn30">Proxy classes</a></li>
<li><a href="#Tcl_nn31">Memory management</a></li>
</ul>
</li>
<li><a href="#Tcl_nn32">Input and output parameters</a></li>
<li><a href="#Tcl_nn33">Exception handling</a></li>
<li><a href="#Tcl_nn34">Typemaps</a>
<ul>
<li><a href="#Tcl_nn35">What is a typemap?</a></li>
<li><a href="#Tcl_nn36">Tcl typemaps</a></li>
<li><a href="#Tcl_nn37">Typemap variables</a></li>
<li><a href="#Tcl_nn38">Converting a Tcl list to a char **</a></li>
<li><a href="#Tcl_nn39">Returning values in arguments</a></li>
<li><a href="#Tcl_nn40">Useful functions</a></li>
<li><a href="#Tcl_nn41">Standard typemaps</a></li>
<li><a href="#Tcl_nn42">Pointer handling</a></li>
</ul>
</li>
<li><a href="#Tcl_nn43">Turning a SWIG module into a Tcl Package.</a></li>
<li><a href="#Tcl_nn44">Building new kinds of Tcl interfaces (in Tcl)</a>
<ul>
<li><a href="#Tcl_nn45">Proxy classes</a></li>
</ul>
</li>
<li><a href="#Tcl_nn46">Tcl/Tk Stubs</a></li>
</ul>
</div>
<!-- INDEX -->
<p><b> Caution: This chapter is under repair!</b></p>
<p> This chapter discusses SWIG's support of Tcl. Since SWIG 4.1.0, Tcl
8.4 or a later release is required. Prior to that earlier Tcl 8.x
releases were also supported. Tcl 9.0 is supported since SWIG 4.2.1.</p>
<h2><a name="Tcl_nn2">37.1 Preliminaries</a></h2>
<p> To build a Tcl module, run SWIG using the <tt>-tcl</tt> or <tt>-tcl8</tt>
option :</p>
<div class="code">
<pre>
$ swig -tcl example.i
</pre>
</div>
<p> If building a C++ extension, add the <tt>-c++</tt> option:</p>
<div class="code">
<pre>
$ swig -c++ -tcl example.i
</pre>
</div>
<p> This creates a file <tt>example_wrap.c</tt> or <tt>example_wrap.cxx</tt>
that contains all of the code needed to build a Tcl extension module.
To finish building the module, you need to compile this file and link
it with the rest of your program.</p>
<h3><a name="Tcl_nn3">37.1.1 Getting the right header files</a></h3>
<p> In order to compile the wrapper code, the compiler needs the <tt>
tcl.h</tt> header file. This file is usually contained in the directory</p>
<div class="code">
<pre>
/usr/local/include
</pre>
</div>
<p> Be aware that some Tcl versions install this header file with a
version number attached to it. If this is the case, you should probably
make a symbolic link so that <tt>tcl.h</tt> points to the correct
header file.</p>
<h3><a name="Tcl_nn4">37.1.2 Compiling a dynamic module</a></h3>
<p> The preferred approach to building an extension module is to compile
it into a shared object file or DLL. Assuming you have code you need to
link to in a file called <tt>example.c</tt>, you will need to compile
your program using commands like this (shown for Linux):</p>
<div class="code">
<pre>
$ swig -tcl example.i
$ gcc -fPIC -c example.c
$ gcc -fPIC -c example_wrap.c -I/usr/local/include
$ gcc -shared example.o example_wrap.o -o example.so
</pre>
</div>
<p> The exact commands for doing this vary from platform to platform.
SWIG tries to guess the right options when it is installed. Therefore,
you may want to start with one of the examples in the <tt>
SWIG/Examples/tcl</tt> directory. If that doesn't work, you will need to
read the man-pages for your compiler and linker to get the right set of
options. You might also check the <a href="https://github.com/swig/swig/wiki">
SWIG Wiki</a> for additional information.</p>
<p> When linking the module, the name of the output file has to match
the name of the module. If the name of your SWIG module is &quot;<tt>example</tt>
&quot;, the name of the corresponding object file should be &quot;<tt>example.so</tt>
&quot;. The name of the module is specified using the <tt>%module</tt>
directive or the <tt>-module</tt> command line option.</p>
<h3><a name="Tcl_nn5">37.1.3 Static linking</a></h3>
<p> An alternative approach to dynamic linking is to rebuild the Tcl
interpreter with your extension module added to it. In the past, this
approach was sometimes necessary due to limitations in dynamic loading
support on certain machines. However, the situation has improved
greatly over the last few years and you should not consider this
approach unless there is really no other option.</p>
<p> The usual procedure for adding a new module to Tcl involves writing
a special function <tt>Tcl_AppInit()</tt> and using it to initialize
the interpreter and your module. With SWIG, the <tt>tclsh.i</tt> and <tt>
wish.i</tt> library files can be used to rebuild the <tt>tclsh</tt> and <tt>
wish</tt> interpreters respectively. For example:</p>
<div class="code">
<pre>
%module example
%inline %{
extern int fact(int);
extern int mod(int, int);
extern double My_variable;
%}
%include &quot;tclsh.i&quot; // Include code for rebuilding tclsh
</pre>
</div>
<p> The <tt>tclsh.i</tt> library file includes supporting code that
contains everything needed to rebuild tclsh. To rebuild the
interpreter, you simply do something like this:</p>
<div class="code">
<pre>
$ swig -tcl example.i
$ gcc example.c example_wrap.c \
-Xlinker -export-dynamic \
-DHAVE_CONFIG_H -I/usr/local/include/ \
-L/usr/local/lib -ltcl -lm -ldl \
-o mytclsh
</pre>
</div>
<p> You will need to supply the same libraries that were used to build
Tcl the first time. This may include system libraries such as <tt>
-lsocket</tt>, <tt>-lnsl</tt>, and <tt>-lpthread</tt>. If this actually
works, the new version of Tcl should be identical to the default
version except that your extension module will be a built-in part of
the interpreter.</p>
<p><b> Comment:</b> In practice, you should probably try to avoid static
linking if possible. Some programmers may be inclined to use static
linking in the interest of getting better performance. However, the
performance gained by static linking tends to be rather minimal in most
situations (and quite frankly not worth the extra hassle in the opinion
of this author).</p>
<h3><a name="Tcl_nn6">37.1.4 Using your module</a></h3>
<p> To use your module, simply use the Tcl <tt>load</tt> command. If all
goes well, you will be able to this:</p>
<div class="code">
<pre>
$ tclsh
% load ./example.so
% fact 4
24
%
</pre>
</div>
<p> A common error received by first-time users is the following:</p>
<div class="code">
<pre>
% load ./example.so
couldn't find procedure Example_Init
%
</pre>
</div>
<p> This error is almost always caused when the name of the shared
object file doesn't match the name of the module supplied using the
SWIG <tt>%module</tt> directive. Double-check the interface to make
sure the module name and the shared object file match. Another possible
cause of this error is forgetting to link the SWIG-generated wrapper
code with the rest of your application when creating the extension
module.</p>
<p> Another common error is something similar to the following:</p>
<div class="code">
<pre>
% load ./example.so
couldn't load file &quot;./example.so&quot;: ./example.so: undefined symbol: fact
%
</pre>
</div>
<p> This error usually indicates that you forgot to include some object
files or libraries in the linking of the shared library file. Make sure
you compile both the SWIG wrapper file and your original program into a
shared library file. Make sure you pass all of the required libraries
to the linker.</p>
<p> Sometimes unresolved symbols occur because a wrapper has been
created for a function that doesn't actually exist in a library. This
usually occurs when a header file includes a declaration for a function
that was never actually implemented or it was removed from a library
without updating the header file. To fix this, you can either edit the
SWIG input file to remove the offending declaration or you can use the <tt>
%ignore</tt> directive to ignore the declaration.</p>
<p> Finally, suppose that your extension module is linked with another
library like this:</p>
<div class="code">
<pre>
$ gcc -shared example.o example_wrap.o -L/home/beazley/projects/lib -lfoo \
-o example.so
</pre>
</div>
<p> If the <tt>foo</tt> library is compiled as a shared library, you
might get the following problem when you try to use your module:</p>
<div class="code">
<pre>
% load ./example.so
couldn't load file &quot;./example.so&quot;: libfoo.so: cannot open shared object file:
No such file or directory
%
</pre>
</div>
<p> This error is generated because the dynamic linker can't locate the <tt>
libfoo.so</tt> library. When shared libraries are loaded, the system
normally only checks a few standard locations such as <tt>/usr/lib</tt>
and <tt>/usr/local/lib</tt>. To fix this problem, there are several
things you can do. First, you can recompile your extension module with
extra path information. For example, on Linux you can do this:</p>
<div class="code">
<pre>
$ gcc -shared example.o example_wrap.o -L/home/beazley/projects/lib -lfoo \
-Xlinker -rpath /home/beazley/projects/lib \
-o example.so
</pre>
</div>
<p> Alternatively, you can set the <tt>LD_LIBRARY_PATH</tt> environment
variable to include the directory with your shared libraries. If
setting <tt>LD_LIBRARY_PATH</tt>, be aware that setting this variable
can introduce a noticeable performance impact on all other applications
that you run. To set it only for Tcl, you might want to do this
instead:</p>
<div class="code">
<pre>
$ env LD_LIBRARY_PATH=/home/beazley/projects/lib tclsh
</pre>
</div>
<p> Finally, you can use a command such as <tt>ldconfig</tt> to add
additional search paths to the default system configuration (this
requires root access and you will need to read the man pages).</p>
<h3><a name="Tcl_nn7">37.1.5 Compilation of C++ extensions</a></h3>
<p> Compilation of C++ extensions has traditionally been a tricky
problem. Since the Tcl interpreter is written in C, you need to take
steps to make sure C++ is properly initialized and that modules are
compiled correctly.</p>
<p> On most machines, C++ extension modules should be linked using the
C++ compiler. For example:</p>
<div class="code">
<pre>
% swig -c++ -tcl example.i
% g++ -fPIC -c example.cxx
% g++ -fPIC -c example_wrap.cxx -I/usr/local/include
% g++ -shared example.o example_wrap.o -o example.so
</pre>
</div>
<p> In addition to this, you may need to include additional library
files to make it work. For example, if you are using the Sun C++
compiler on Solaris, you often need to add an extra library <tt>-lCrun</tt>
like this:</p>
<div class="code">
<pre>
% swig -c++ -tcl example.i
% CC -KPIC -c example.cxx
% CC -KPIC -c example_wrap.cxx -I/usr/local/include
% CC -G example.o example_wrap.o -L/opt/SUNWspro/lib -o example.so -lCrun
</pre>
</div>
<p> Of course, the extra libraries to use are completely
non-portable---you will probably need to do some experimentation.</p>
<p> Sometimes people have suggested that it is necessary to relink the
Tcl interpreter using the C++ compiler to make C++ extension modules
work. In the experience of this author, this has never actually
appeared to be necessary. Relinking the interpreter with C++ really
only includes the special run-time libraries described above---as long
as you link your extension modules with these libraries, it should not
be necessary to rebuild Tcl.</p>
<p> If you aren't entirely sure about the linking of a C++ extension,
you might look at an existing C++ program. On many Unix machines, the <tt>
ldd</tt> command will list library dependencies. This should give you
some clues about what you might have to include when you link your
extension module. For example:</p>
<div class="code">
<pre>
$ ldd swig
libstdc++-libc6.1-1.so.2 =&gt; /usr/lib/libstdc++-libc6.1-1.so.2 (0x40019000)
libm.so.6 =&gt; /lib/libm.so.6 (0x4005b000)
libc.so.6 =&gt; /lib/libc.so.6 (0x40077000)
/lib/ld-linux.so.2 =&gt; /lib/ld-linux.so.2 (0x40000000)
$
</pre>
</div>
<p> As a final complication, a major weakness of C++ is that it does not
define any sort of standard for binary linking of libraries. This means
that C++ code compiled by different compilers will not link together
properly as libraries nor is the memory layout of classes and data
structures implemented in any kind of portable manner. In a monolithic
C++ program, this problem may be unnoticed. However, in Tcl, it is
possible for different extension modules to be compiled with different
C++ compilers. As long as these modules are self-contained, this
probably won't matter. However, if these modules start sharing data,
you will need to take steps to avoid segmentation faults and other
erratic program behavior. If working with lots of software components,
you might want to investigate using a more formal standard such as COM.</p>
<h3><a name="Tcl_nn8">37.1.6 Compiling for 64-bit platforms</a></h3>
<p> On platforms that support 64-bit applications (Solaris, Irix, etc.),
special care is required when building extension modules. On these
machines, 64-bit applications are compiled and linked using a different
set of compiler/linker options. In addition, it is not generally
possible to mix 32-bit and 64-bit code together in the same
application.</p>
<p> To utilize 64-bits, the Tcl executable will need to be recompiled as
a 64-bit application. In addition, all libraries, wrapper code, and
every other part of your application will need to be compiled for
64-bits. If you plan to use other third-party extension modules, they
will also have to be recompiled as 64-bit extensions.</p>
<p> If you are wrapping commercial software for which you have no source
code, you will be forced to use the same linking standard as used by
that software. This may prevent the use of 64-bit extensions. It may
also introduce problems on platforms that support more than one linking
standard (e.g., -o32 and -n32 on Irix).</p>
<h3><a name="Tcl_nn9">37.1.7 Setting a package prefix</a></h3>
<p> To avoid namespace problems, you can instruct SWIG to append a
package prefix to all of your functions and variables. This is done
using the -prefix option as follows :</p>
<div class="code">
<pre>
swig -tcl -prefix Foo example.i
</pre>
</div>
<p> If you have a function &quot;<tt>bar</tt>&quot; in the SWIG file, the prefix
option will append the prefix to the name when creating a command and
call it &quot;<tt>Foo_bar</tt>&quot;.</p>
<h3><a name="Tcl_nn10">37.1.8 Using namespaces</a></h3>
<p> Alternatively, you can have SWIG install your module into a Tcl
namespace by specifying the <tt>-namespace</tt> option :</p>
<div class="code">
<pre>
swig -tcl -namespace example.i
</pre>
</div>
<p> By default, the name of the namespace will be the same as the module
name, but you can override it using the <tt>-prefix</tt> option.</p>
<p> When the <tt>-namespace</tt> option is used, objects in the module
are always accessed with the namespace name such as <tt>Foo::bar</tt>.</p>
<h2><a name="Tcl_nn11">37.2 Building Tcl/Tk Extensions under Windows
95/NT</a></h2>
<p> Building a SWIG extension to Tcl/Tk under Windows 95/NT is roughly
similar to the process used with Unix. Normally, you will want to
produce a DLL that can be loaded into tclsh or wish. This section
covers the process of using SWIG with Microsoft Visual C++. although
the procedure may be similar with other compilers.</p>
<h3><a name="Tcl_nn12">37.2.1 Running SWIG from Developer Studio</a></h3>
<p> If you are developing your application within Microsoft developer
studio, SWIG can be invoked as a custom build option. The process
roughly follows these steps :</p>
<ul>
<li>Open up a new workspace and use the AppWizard to select a DLL
project.</li>
<li>Add both the SWIG interface file (the .i file), any supporting C
files, and the name of the wrapper file that will be created by SWIG
(ie. <tt>example_wrap.c</tt>). Note : If using C++, choose a different
suffix for the wrapper file such as <tt>example_wrap.cxx</tt>. Don't
worry if the wrapper file doesn't exist yet--Developer studio will keep
a reference to it around.</li>
<li>Select the SWIG interface file and go to the settings menu. Under
settings, select the &quot;Custom Build&quot; option.</li>
<li>Enter &quot;SWIG&quot; in the description field.</li>
<li>Enter &quot;<tt>swig -tcl -o $(ProjDir)\$(InputName)_wrap.c $(InputPath)</tt>
&quot; in the &quot;Build command(s) field&quot;</li>
<li>Enter &quot;<tt>$(ProjDir)\$(InputName)_wrap.c</tt>&quot; in the &quot;Output
files(s) field&quot;.</li>
<li>Next, select the settings for the entire project and go to
&quot;C++:Preprocessor&quot;. Add the include directories for your Tcl
installation under &quot;Additional include directories&quot;.</li>
<li>Finally, select the settings for the entire project and go to &quot;Link
Options&quot;. Add the Tcl library file to your link libraries. For example
&quot;<tt>tcl80.lib</tt>&quot;. Also, set the name of the output file to match
the name of your Tcl module (ie. example.dll).</li>
<li>Build your project.</li>
</ul>
<p> Now, assuming all went well, SWIG will be automatically invoked when
you build your project. Any changes made to the interface file will
result in SWIG being automatically invoked to produce a new version of
the wrapper file. To run your new Tcl extension, simply run <tt>tclsh</tt>
or <tt>wish</tt> and use the <tt>load</tt> command. For example :</p>
<div class="code">
<pre>
MSDOS &gt; tclsh80
% load example.dll
% fact 4
24
%
</pre>
</div>
<h3><a name="Tcl_nn13">37.2.2 Using NMAKE</a></h3>
<p> Alternatively, SWIG extensions can be built by writing a Makefile
for NMAKE. To do this, make sure the environment variables for MSVC++
are available and the MSVC++ tools are in your path. Now, just write a
short Makefile like this :</p>
<div class="code">
<pre>
# Makefile for building various SWIG generated extensions
SRCS = example.c
IFILE = example
INTERFACE = $(IFILE).i
WRAPFILE = $(IFILE)_wrap.c
# Location of the Visual C++ tools (32 bit assumed)
TOOLS = c:\msdev
TARGET = example.dll
CC = $(TOOLS)\bin\cl.exe
LINK = $(TOOLS)\bin\link.exe
INCLUDE32 = -I$(TOOLS)\include
MACHINE = IX86
# C Library needed to build a DLL
DLLIBC = msvcrt.lib oldnames.lib
# Windows libraries that are apparently needed
WINLIB = kernel32.lib advapi32.lib user32.lib gdi32.lib comdlg32.lib
winspool.lib
# Libraries common to all DLLs
LIBS = $(DLLIBC) $(WINLIB)
# Linker options
LOPT = -debug:full -debugtype:cv /NODEFAULTLIB /RELEASE /NOLOGO /
MACHINE:$(MACHINE) -entry:_DllMainCRTStartup@12 -dll
# C compiler flags
CFLAGS = /Z7 /Od /c /nologo
TCL_INCLUDES = -Id:\tcl8.0a2\generic -Id:\tcl8.0a2\win
TCLLIB = d:\tcl8.0a2\win\tcl80.lib
tcl:
..\..\swig -tcl -o $(WRAPFILE) $(INTERFACE)
$(CC) $(CFLAGS) $(TCL_INCLUDES) $(SRCS) $(WRAPFILE)
set LIB=$(TOOLS)\lib
$(LINK) $(LOPT) -out:example.dll $(LIBS) $(TCLLIB) example.obj example_wrap.obj
</pre>
</div>
<p> To build the extension, run NMAKE (you may need to run vcvars32
first). This is a pretty minimal Makefile, but hopefully it's enough to
get you started. With a little practice, you'll be making lots of Tcl
extensions.</p>
<h2><a name="Tcl_nn14">37.3 A tour of basic C/C++ wrapping</a></h2>
<p> By default, SWIG tries to build a very natural Tcl interface to your
C/C++ code. Functions are wrapped as functions, classes are wrapped in
an interface that mimics the style of Tk widgets and [incr Tcl]
classes. This section briefly covers the essential aspects of this
wrapping.</p>
<h3><a name="Tcl_nn15">37.3.1 Modules</a></h3>
<p> The SWIG <tt>%module</tt> directive specifies the name of the Tcl
module. If you specify `<tt>%module example</tt>', then everything is
compiled into an extension module <tt>example.so</tt>. When choosing a
module name, make sure you don't use the same name as a built-in Tcl
command.</p>
<p> One pitfall to watch out for is module names involving numbers. If
you specify a module name like <tt>%module md5</tt>, you'll find that
the load command no longer seems to work:</p>
<div class="code">
<pre>
% load ./md5.so
couldn't find procedure Md_Init
</pre>
</div>
<p> To fix this, supply an extra argument to <tt>load</tt> like this:</p>
<div class="code">
<pre>
% load ./md5.so md5
</pre>
</div>
<h3><a name="Tcl_nn16">37.3.2 Functions</a></h3>
<p> Global functions are wrapped as new Tcl built-in commands. For
example,</p>
<div class="code">
<pre>
%module example
int fact(int n);
</pre>
</div>
<p> creates a built-in function <tt>fact</tt> that works exactly like
you think it does:</p>
<div class="code">
<pre>
% load ./example.so
% fact 4
24
% set x [fact 6]
%
</pre>
</div>
<h3><a name="Tcl_nn17">37.3.3 Global variables</a></h3>
<p> C/C++ global variables are wrapped by Tcl global variables. For
example:</p>
<div class="code">
<pre>
// SWIG interface file with global variables
%module example
...
%inline %{
extern double density;
%}
...
</pre>
</div>
<p> Now look at the Tcl interface:</p>
<div class="code">
<pre>
% puts $density # Output value of C global variable
1.0
% set density 0.95 # Change value
</pre>
</div>
<p> If you make an error in variable assignment, you will get an error
message. For example:</p>
<div class="code">
<pre>
% set density &quot;hello&quot;
can't set &quot;density&quot;: Type error. expected a double.
%
</pre>
</div>
<p> If a variable is declared as <tt>const</tt>, it is wrapped as a
read-only variable. Attempts to modify its value will result in an
error.</p>
<p> To make ordinary variables read-only, you can use the <tt>%immutable</tt>
directive. For example:</p>
<div class="code">
<pre>
%{
extern char *path;
%}
%immutable;
extern char *path;
%mutable;
</pre>
</div>
<p> The <tt>%immutable</tt> directive stays in effect until it is
explicitly disabled or cleared using <tt>%mutable</tt>. See the <a href="#SWIG_readonly_variables">
Creating read-only variables</a> section for further details.</p>
<p> If you just want to make a specific variable immutable, supply a
declaration name. For example:</p>
<div class="code">
<pre>
%{
extern char *path;
%}
%immutable path;
...
extern char *path; // Read-only (due to %immutable)
</pre>
</div>
<h3><a name="Tcl_nn18">37.3.4 Constants and enums</a></h3>
<p> C/C++ constants are installed as global Tcl variables containing the
appropriate value. To create a constant, use <tt>#define</tt>, <tt>enum</tt>
, or the <tt>%constant</tt> directive. For example:</p>
<div class="code">
<pre>
#define PI 3.14159
#define VERSION &quot;1.0&quot;
enum Beverage { ALE, LAGER, STOUT, PILSNER };
%constant int FOO = 42;
%constant const char *path = &quot;/usr/local&quot;;
</pre>
</div>
<p> For enums, make sure that the definition of the enumeration actually
appears in a header file or in the wrapper file somehow---if you just
stick an enum in a SWIG interface without also telling the C compiler
about it, the wrapper code won't compile.</p>
<p> Note: declarations declared as <tt>const</tt> are wrapped as
read-only variables and will be accessed using the <tt>cvar</tt> object
described in the previous section. They are not wrapped as constants.
For further discussion about this, see the <a href="#SWIG">SWIG Basics</a>
chapter.</p>
<p> Constants are not guaranteed to remain constant in Tcl---the value
of the constant could be accidentally reassigned. You will just have to
be careful.</p>
<p> A peculiarity of installing constants as variables is that it is
necessary to use the Tcl <tt>global</tt> statement to access constants
in procedure bodies. For example:</p>
<div class="code">
<pre>
proc blah {} {
global FOO
bar $FOO
}
</pre>
</div>
<h3><a name="Tcl_nn19">37.3.5 Pointers</a></h3>
<p> C/C++ pointers are fully supported by SWIG. Furthermore, SWIG has no
problem working with incomplete type information. Here is a rather
simple interface:</p>
<div class="code">
<pre>
%module example
FILE *fopen(const char *filename, const char *mode);
int fputs(const char *, FILE *);
int fclose(FILE *);
</pre>
</div>
<p> When wrapped, you will be able to use the functions in a natural way
from Tcl. For example:</p>
<div class="code">
<pre>
% load ./example.so
% set f [fopen junk w]
% fputs &quot;Hello World\n&quot; $f
% fclose $f
</pre>
</div>
<p> If this makes you uneasy, rest assured that there is no deep magic
involved. Underneath the covers, pointers to C/C++ objects are simply
represented as opaque values--normally an encoded character string like
this:</p>
<div class="code">
<pre>
% puts $f
_c0671108_p_FILE
%
</pre>
</div>
<p> This pointer value can be freely passed around to different C
functions that expect to receive an object of type <tt>FILE *</tt>. The
only thing you can't do is dereference the pointer from Tcl.</p>
<p> The NULL pointer is represented by the string <tt>NULL</tt>.</p>
<p> As much as you might be inclined to modify a pointer value directly
from Tcl, don't. The hexadecimal encoding is not necessarily the same
as the logical memory address of the underlying object. Instead it is
the raw byte encoding of the pointer value. The encoding will vary
depending on the native byte-ordering of the platform (i.e., big-endian
vs. little-endian). Similarly, don't try to manually cast a pointer to
a new type by simply replacing the type-string. This may not work like
you expect and it is particularly dangerous when casting C++ objects.
If you need to cast a pointer or change its value, consider writing
some helper functions instead. For example:</p>
<div class="code">
<pre>
%inline %{
/* C-style cast */
Bar *FooToBar(Foo *f) {
return (Bar *) f;
}
/* C++-style cast */
Foo *BarToFoo(Bar *b) {
return dynamic_cast&lt;Foo*&gt;(b);
}
Foo *IncrFoo(Foo *f, int i) {
return f+i;
}
%}
</pre>
</div>
<p> Also, if working with C++, you should always try to use the new C++
style casts. For example, in the above code, the C-style cast may
return a bogus result whereas as the C++-style cast will return <tt>
None</tt> if the conversion can't be performed.</p>
<h3><a name="Tcl_nn20">37.3.6 Structures</a></h3>
<p> If you wrap a C structure, it is wrapped by a Tcl interface that
somewhat resembles a Tk widget. This provides a very natural interface.
For example,</p>
<div class="code">
<pre>
struct Vector {
double x, y, z;
};
</pre>
</div>
<p> is used as follows:</p>
<div class="code">
<pre>
% Vector v
% v configure -x 3.5 -y 7.2
% puts &quot;[v cget -x] [v cget -y] [v cget -z]&quot;
3.5 7.2 0.0
%
</pre>
</div>
<p> Similar access is provided for unions and the data members of C++
classes.</p>
<p> In the above example, <tt>v</tt> is a name that's used for the
object. However, underneath the covers, there's a pointer to a raw C
structure. This can be obtained by looking at the <tt>-this</tt>
attribute. For example:</p>
<div class="code">
<pre>
% puts [v cget -this]
_88e31408_p_Vector
</pre>
</div>
<p> Further details about the relationship between the Tcl and the
underlying C structure are covered a little later.</p>
<p> <tt>const</tt> members of a structure are read-only. Data members
can also be forced to be read-only using the <tt>%immutable</tt>
directive. For example:</p>
<div class="code">
<pre>
struct Foo {
...
%immutable;
int x; /* Read-only members */
char *name;
%mutable;
...
};
</pre>
</div>
<p> When <tt>char *</tt> members of a structure are wrapped, the
contents are assumed to be dynamically allocated using <tt>malloc</tt>
or <tt>new</tt> (depending on whether or not SWIG is run with the -c++
option). When the structure member is set, the old contents will be
released and a new value created. If this is not the behavior you want,
you will have to use a typemap (described later).</p>
<p> If a structure contains arrays, access to those arrays is managed
through pointers. For example, consider this:</p>
<div class="code">
<pre>
struct Bar {
int x[16];
};
</pre>
</div>
<p> If accessed in Tcl, you will see behavior like this:</p>
<div class="code">
<pre>
% Bar b
% puts [b cget -x]
_801861a4_p_int
%
</pre>
</div>
<p> This pointer can be passed around to functions that expect to
receive an <tt>int *</tt> (just like C). You can also set the value of
an array member using another pointer. For example:</p>
<div class="code">
<pre>
% Bar c
% c configure -x [b cget -x] # Copy contents of b.x to c.x
</pre>
</div>
<p> For array assignment, SWIG copies the entire contents of the array
starting with the data pointed to by <tt>b.x</tt>. In this example, 16
integers would be copied. Like C, SWIG makes no assumptions about
bounds checking---if you pass a bad pointer, you may get a segmentation
fault or access violation.</p>
<p> When a member of a structure is itself a structure, it is handled as
a pointer. For example, suppose you have two structures like this:</p>
<div class="code">
<pre>
struct Foo {
int a;
};
struct Bar {
Foo f;
};
</pre>
</div>
<p> Now, suppose that you access the <tt>f</tt> attribute of <tt>Bar</tt>
like this:</p>
<div class="code">
<pre>
% Bar b
% set x [b cget -f]
</pre>
</div>
<p> In this case, <tt>x</tt> is a pointer that points to the <tt>Foo</tt>
that is inside <tt>b</tt>. This is the same value as generated by this
C code:</p>
<div class="code">
<pre>
Bar b;
Foo *x = &amp;b-&gt;f; /* Points inside b */
</pre>
</div>
<p> However, one peculiarity of accessing a substructure like this is
that the returned value does work quite like you might expect. For
example:</p>
<div class="code">
<pre>
% Bar b
% set x [b cget -f]
% x cget -a
invalid command name &quot;x&quot;
</pre>
</div>
<p> This is because the returned value was not created in a normal way
from the interpreter (x is not a command object). To make it function
normally, just evaluate the variable like this:</p>
<div class="code">
<pre>
% Bar b
% set x [b cget -f]
% $x cget -a
0
%
</pre>
</div>
<p> In this example, <tt>x</tt> points inside the original structure.
This means that modifications work just like you would expect. For
example:</p>
<div class="code">
<pre>
% Bar b
% set x [b cget -f]
% $x configure -a 3 # Modifies contents of f (inside b)
% [b cget -f] -configure -a 3 # Same thing
</pre>
</div>
<p> In many of these structure examples, a simple name like &quot;v&quot; or &quot;b&quot;
has been given to wrapped structures. If necessary, this name can be
passed to functions that expect to receive an object. For example, if
you have a function like this,</p>
<div class="code">
<pre>
void blah(Foo *f);
</pre>
</div>
<p> you can call the function in Tcl as follows:</p>
<div class="code">
<pre>
% Foo x # Create a Foo object
% blah x # Pass the object to a function
</pre>
</div>
<p> It is also possible to call the function using the raw pointer
value. For instance:</p>
<div class="code">
<pre>
% blah [x cget -this] # Pass object to a function
</pre>
</div>
<p> It is also possible to create and use objects using variables. For
example:</p>
<div class="code">
<pre>
% set b [Bar] # Create a Bar
% $b cget -f # Member access
% puts $b
_108fea88_p_Bar
%
</pre>
</div>
<p> Finally, to destroy objects created from Tcl, you can either let the
object name go out of scope or you can explicitly delete the object as
shown below. Objects won't get automatically destroyed when the Tcl
program exits, so if it's important that the C++ destructor is called
for a class you'll need to make sure that you explicitly do this for
objects of that class before program exit.</p>
<p> For example:</p>
<div class="code">
<pre>
% Foo f # Create object f
% rename f &quot;&quot;
</pre>
</div>
<p> or</p>
<div class="code">
<pre>
% Foo f # Create object f
% f -delete
</pre>
</div>
<p> Note: Tcl only destroys the underlying object if it has ownership.
See the memory management section that appears shortly.</p>
<h3><a name="Tcl_nn21">37.3.7 C++ classes</a></h3>
<p> C++ classes are wrapped as an extension of structure wrapping. For
example, if you have this class,</p>
<div class="code">
<pre>
class List {
public:
List();
~List();
int search(char *item);
void insert(char *item);
void remove(char *item);
char *get(int n);
int length;
};
</pre>
</div>
<p> you can use it in Tcl like this:</p>
<div class="code">
<pre>
% List x
% x insert Ale
% x insert Stout
% x insert Lager
% x get 1
Stout
% puts [x cget -length]
3
%
</pre>
</div>
<p> Class data members are accessed in the same manner as C structures.</p>
<p> Static class members are accessed as global functions or variables.
To illustrate, suppose you have a class like this:</p>
<div class="code">
<pre>
class Spam {
public:
static void foo();
static int bar;
};
</pre>
</div>
<p> In Tcl, the static member is accessed as follows:</p>
<div class="code">
<pre>
% Spam_foo # Spam::foo()
% puts $Spam_bar # Spam::bar
</pre>
</div>
<h3><a name="Tcl_nn22">37.3.8 C++ inheritance</a></h3>
<p> SWIG is fully aware of issues related to C++ inheritance. Therefore,
if you have classes like this</p>
<div class="code">
<pre>
class Foo {
...
};
class Bar : public Foo {
...
};
</pre>
</div>
<p> An object of type <tt>Bar</tt> can be used where a <tt>Foo</tt> is
expected. For example, if you have this function:</p>
<div class="code">
<pre>
void spam(Foo *f);
</pre>
</div>
<p> then the function <tt>spam()</tt> accepts a <tt>Foo *</tt> or a
pointer to any class derived from <tt>Foo</tt>. For instance:</p>
<div class="code">
<pre>
% Foo f # Create a Foo
% Bar b # Create a Bar
% spam f # OK
% spam b # OK
</pre>
</div>
<p> It is safe to use multiple inheritance with SWIG.</p>
<h3><a name="Tcl_nn23">37.3.9 Pointers, references, values, and arrays</a>
</h3>
<p> In C++, there are many different ways a function might receive and
manipulate objects. For example:</p>
<div class="code">
<pre>
void spam1(Foo *x); // Pass by pointer
void spam2(Foo &amp;x); // Pass by reference
void spam3(Foo x); // Pass by value
void spam4(Foo x[]); // Array of objects
</pre>
</div>
<p> In Tcl, there is no detailed distinction like this. Because of this,
SWIG unifies all of these types together in the wrapper code. For
instance, if you actually had the above functions, it is perfectly
legal to do this:</p>
<div class="code">
<pre>
% Foo f # Create a Foo
% spam1 f # Ok. Pointer
% spam2 f # Ok. Reference
% spam3 f # Ok. Value.
% spam4 f # Ok. Array (1 element)
</pre>
</div>
<p> Similar behavior occurs for return values. For example, if you had
functions like this,</p>
<div class="code">
<pre>
Foo *spam5();
Foo &amp;spam6();
Foo spam7();
</pre>
</div>
<p> then all three functions will return a pointer to some <tt>Foo</tt>
object. Since the third function (spam7) returns a value, newly
allocated memory is used to hold the result and a pointer is returned
(Tcl will release this memory when the return value is garbage
collected).</p>
<h3><a name="Tcl_nn24">37.3.10 C++ overloaded functions</a></h3>
<p> C++ overloaded functions, methods, and constructors are mostly
supported by SWIG. For example, if you have two functions like this:</p>
<div class="code">
<pre>
void foo(int);
void foo(char *c);
</pre>
</div>
<p> You can use them in Tcl in a straightforward manner:</p>
<div class="code">
<pre>
% foo 3 # foo(int)
% foo Hello # foo(char *c)
</pre>
</div>
<p> Similarly, if you have a class like this,</p>
<div class="code">
<pre>
class Foo {
public:
Foo();
Foo(const Foo &amp;);
...
};
</pre>
</div>
<p> you can write Tcl code like this:</p>
<div class="code">
<pre>
% Foo f # Create a Foo
% Foo g f # Copy f
</pre>
</div>
<p> Overloading support is not quite as flexible as in C++. Sometimes
there are methods that SWIG can't disambiguate. For example:</p>
<div class="code">
<pre>
void spam(int);
void spam(short);
</pre>
</div>
<p> or</p>
<div class="code">
<pre>
void foo(Bar *b);
void foo(Bar &amp;b);
</pre>
</div>
<p> If declarations such as these appear, you will get a warning message
like this:</p>
<div class="code">
<pre>
example.i:12: Warning 509: Overloaded method spam(short) effectively ignored,
example.i:11: Warning 509: as it is shadowed by spam(int).
</pre>
</div>
<p> To fix this, you either need to ignore or rename one of the methods.
For example:</p>
<div class="code">
<pre>
%rename(spam_short) spam(short);
...
void spam(int);
void spam(short); // Accessed as spam_short
</pre>
</div>
<p> or</p>
<div class="code">
<pre>
%ignore spam(short);
...
void spam(int);
void spam(short); // Ignored
</pre>
</div>
<p> SWIG resolves overloaded functions and methods using a
disambiguation scheme that ranks and sorts declarations according to a
set of type-precedence rules. The order in which declarations appear in
the input does not matter except in situations where ambiguity
arises--in this case, the first declaration takes precedence.</p>
<p> Please refer to the &quot;SWIG and C++&quot; chapter for more information
about overloading.</p>
<h3><a name="Tcl_nn25">37.3.11 C++ operators</a></h3>
<p> Certain C++ overloaded operators can be handled automatically by
SWIG. For example, consider a class like this:</p>
<div class="code">
<pre>
class Complex {
private:
double rpart, ipart;
public:
Complex(double r = 0, double i = 0) : rpart(r), ipart(i) { }
Complex(const Complex &amp;c) : rpart(c.rpart), ipart(c.ipart) { }
Complex &amp;operator=(const Complex &amp;c);
Complex operator+(const Complex &amp;c) const;
Complex operator-(const Complex &amp;c) const;
Complex operator*(const Complex &amp;c) const;
Complex operator-() const;
double re() const { return rpart; }
double im() const { return ipart; }
};
</pre>
</div>
<p> When wrapped, it works like this:</p>
<div class="code">
<pre>
% Complex c 3 4
% Complex d 7 8
% set e [c + d]
% $e re
10.0
% $e im
12.0
</pre>
</div>
<p> It should be stressed that operators in SWIG have no relationship to
operators in Tcl. In fact, the only thing that's happening here is that
an operator like <tt>operator +</tt> has been renamed to a method <tt>+</tt>
. Therefore, the statement <tt>[c + d]</tt> is really just invoking the <tt>
+</tt> method on <tt>c</tt>. When more than operator is defined (with
different arguments), the standard method overloading facilities are
used. Here is a rather odd looking example:</p>
<div class="code">
<pre>
% Complex c 3 4
% Complex d 7 8
% set e [c - d] # operator-(const Complex &amp;)
% puts &quot;[$e re] [$e im]&quot;
10.0 12.0
% set f [c -] # operator-()
% puts &quot;[$f re] [$f im]&quot;
-3.0 -4.0
%
</pre>
</div>
<p> One restriction with operator overloading support is that SWIG is
not able to fully handle operators that aren't defined as part of the
class. For example, if you had code like this</p>
<div class="code">
<pre>
class Complex {
...
friend Complex operator+(double, const Complex &amp;c);
...
};
</pre>
</div>
<p> then SWIG doesn't know what to do with the friend function--in fact,
it simply ignores it and issues a warning. You can still wrap the
operator, but you may have to encapsulate it in a special function. For
example:</p>
<div class="code">
<pre>
%rename(Complex_add_dc) operator+(double, const Complex &amp;);
...
Complex operator+(double, const Complex &amp;c);
</pre>
</div>
<p> There are ways to make this operator appear as part of the class
using the <tt>%extend</tt> directive. Keep reading.</p>
<h3><a name="Tcl_nn26">37.3.12 C++ namespaces</a></h3>
<p> SWIG is aware of C++ namespaces, but namespace names do not appear
in the module nor do namespaces result in a module that is broken up
into submodules or packages. For example, if you have a file like this,</p>
<div class="code">
<pre>
%module example
namespace foo {
int fact(int n);
struct Vector {
double x, y, z;
};
};
</pre>
</div>
<p> it works in Tcl as follows:</p>
<div class="code">
<pre>
% load ./example.so
% fact 3
6
% Vector v
% v configure -x 3.4
</pre>
</div>
<p> If your program has more than one namespace, name conflicts (if any)
can be resolved using <tt>%rename</tt> For example:</p>
<div class="code">
<pre>
%rename(Bar_spam) Bar::spam;
namespace Foo {
int spam();
}
namespace Bar {
int spam();
}
</pre>
</div>
<p> If you have more than one namespace and your want to keep their
symbols separate, consider wrapping them as separate SWIG modules. For
example, make the module name the same as the namespace and create
extension modules for each namespace separately. If your program
utilizes thousands of small deeply nested namespaces each with
identical symbol names, well, then you get what you deserve.</p>
<h3><a name="Tcl_nn27">37.3.13 C++ templates</a></h3>
<p> C++ templates don't present a huge problem for SWIG. However, in
order to create wrappers, you have to tell SWIG to create wrappers for
a particular template instantiation. To do this, you use the <tt>
%template</tt> directive. For example:</p>
<div class="code">
<pre>
%module example
%{
#include &quot;pair.h&quot;
%}
template&lt;class T1, class T2&gt;
struct pair {
typedef T1 first_type;
typedef T2 second_type;
T1 first;
T2 second;
pair();
pair(const T1&amp;, const T2&amp;);
~pair();
};
%template(pairii) pair&lt;int, int&gt;;
</pre>
</div>
<p> In Tcl:</p>
<div class="code">
<pre>
% pairii p 3 4
% p cget -first
3
% p cget -second
4
</pre>
</div>
<p> Obviously, there is more to template wrapping than shown in this
example. More details can be found in the <a href="#SWIGPlus">SWIG and
C++</a> chapter. Some more complicated examples will appear later.</p>
<h3><a name="Tcl_nn28">37.3.14 C++ Smart Pointers</a></h3>
<p> In certain C++ programs, it is common to use classes that have been
wrapped by so-called &quot;smart pointers.&quot; Generally, this involves the use
of a template class that implements <tt>operator-&gt;()</tt> like this:</p>
<div class="code">
<pre>
template&lt;class T&gt; class SmartPtr {
...
T *operator-&gt;();
...
}
</pre>
</div>
<p> Then, if you have a class like this,</p>
<div class="code">
<pre>
class Foo {
public:
int x;
int bar();
};
</pre>
</div>
<p> A smart pointer would be used in C++ as follows:</p>
<div class="code">
<pre>
SmartPtr&lt;Foo&gt; p = CreateFoo(); // Created somehow (not shown)
...
p-&gt;x = 3; // Foo::x
int y = p-&gt;bar(); // Foo::bar
</pre>
</div>
<p> To wrap this in Tcl, simply tell SWIG about the <tt>SmartPtr</tt>
class and the low-level <tt>Foo</tt> object. Make sure you instantiate <tt>
SmartPtr</tt> using <tt>%template</tt> if necessary. For example:</p>
<div class="code">
<pre>
%module example
...
%template(SmartPtrFoo) SmartPtr&lt;Foo&gt;;
...
</pre>
</div>
<p> Now, in Tcl, everything should just &quot;work&quot;:</p>
<div class="code">
<pre>
% set p [CreateFoo] # Create a smart-pointer somehow
% $p configure -x 3 # Foo::x
% $p bar # Foo::bar
</pre>
</div>
<p> If you ever need to access the underlying pointer returned by <tt>
operator-&gt;()</tt> itself, simply use the <tt>__deref__()</tt> method.
For example:</p>
<div class="code">
<pre>
% set f [$p __deref__] # Returns underlying Foo *
</pre>
</div>
<h2><a name="Tcl_nn29">37.4 Further details on the Tcl class interface</a>
</h2>
<p> In the previous section, a high-level view of Tcl wrapping was
presented. A key component of this wrapping is that structures and
classes are wrapped by Tcl class-like objects. This provides a very
natural Tcl interface and allows SWIG to support a number of advanced
features such as operator overloading. However, a number of low-level
details were omitted. This section provides a brief overview of how the
proxy classes work.</p>
<h3><a name="Tcl_nn30">37.4.1 Proxy classes</a></h3>
<p> In the <a href="#SWIG">&quot;SWIG basics&quot;</a> and <a href="#SWIGPlus">
&quot;SWIG and C++&quot;</a> chapters, details of low-level structure and class
wrapping are described. To summarize those chapters, if you have a
class like this</p>
<div class="code">
<pre>
class Foo {
public:
int x;
int spam(int);
...
</pre>
</div>
<p> then SWIG transforms it into a set of low-level procedural wrappers.
For example:</p>
<div class="code">
<pre>
Foo *new_Foo() {
return new Foo();
}
void delete_Foo(Foo *f) {
delete f;
}
int Foo_x_get(Foo *f) {
return f-&gt;x;
}
void Foo_x_set(Foo *f, int value) {
f-&gt;x = value;
}
int Foo_spam(Foo *f, int arg1) {
return f-&gt;spam(arg1);
}
</pre>
</div>
<p> These wrappers are actually found in the Tcl extension module. For
example, you can certainly do this:</p>
<div class="code">
<pre>
% load ./example.so
% set f [new_Foo]
% Foo_x_get $f
0
% Foo_spam $f 3
1
%
</pre>
</div>
<p> However, in addition to this, the classname <tt>Foo</tt> is used as
an object constructor function. This allows objects to be encapsulated
objects that look a lot like Tk widgets as shown in the last section.</p>
<h3><a name="Tcl_nn31">37.4.2 Memory management</a></h3>
<p> Associated with each wrapped object, is an ownership flag <tt>
thisown</tt> The value of this flag determines who is responsible for
deleting the underlying C++ object. If set to 1, the Tcl interpreter
destroys the C++ object when the proxy class is garbage collected. If
set to 0 (or if the attribute is missing), then the destruction of the
proxy class has no effect on the C++ object.</p>
<p> When an object is created by a constructor or returned by value, Tcl
automatically takes ownership of the result. For example:</p>
<div class="code">
<pre>
class Foo {
public:
Foo();
Foo bar();
};
</pre>
</div>
<p> In Tcl:</p>
<div class="code">
<pre>
% Foo f
% f cget -thisown
1
% set g [f bar]
% $g cget -thisown
1
</pre>
</div>
<p> On the other hand, when pointers are returned to Tcl, there is often
no way to know where they came from. Therefore, the ownership is set to
zero. For example:</p>
<div class="code">
<pre>
class Foo {
public:
...
Foo *spam();
...
};
</pre>
</div>
<br><div class="code">
<pre>
% Foo f
% set s [f spam]
% $s cget -thisown
0
%
</pre>
</div>
<p> This behavior is especially important for classes that act as
containers. For example, if a method returns a pointer to an object
that is contained inside another object, you definitely don't want Tcl
to assume ownership and destroy it!</p>
<p> Related to containers, ownership issues can arise whenever an object
is assigned to a member or global variable. For example, consider this
interface:</p>
<div class="code">
<pre>
%module example
struct Foo {
int value;
Foo *next;
};
Foo *head = 0;
</pre>
</div>
<p> When wrapped in Tcl, careful observation will reveal that ownership
changes whenever an object is assigned to a global variable. For
example:</p>
<div class="code">
<pre>
% Foo f
% f cget -thisown
1
% set head f
% f cget -thisown
0
</pre>
</div>
<p> In this case, C is now holding a reference to the object---you
probably don't want Tcl to destroy it. Similarly, this occurs for
members. For example:</p>
<div class="code">
<pre>
% Foo f
% Foo g
% f cget -thisown
1
% g cget -thisown
1
% f configure -next g
% g cget -thisown
0
%
</pre>
</div>
<p> For the most part, memory management issues remain hidden. However,
there are occasionally situations where you might have to manually
change the ownership of an object. For instance, consider code like
this:</p>
<div class="code">
<pre>
class Node {
Object *value;
public:
void set_value(Object *v) { value = v; }
...
};
</pre>
</div>
<p> Now, consider the following Tcl code:</p>
<div class="code">
<pre>
% Object v # Create an object
% Node n # Create a node
% n setvalue v # Set value
% v cget -thisown
1
%
</pre>
</div>
<p> In this case, the object <tt>n</tt> is holding a reference to <tt>v</tt>
internally. However, SWIG has no way to know that this has occurred.
Therefore, Tcl still thinks that it has ownership of the object. Should
the proxy object be destroyed, then the C++ destructor will be invoked
and <tt>n</tt> will be holding a stale-pointer. If you're lucky, you
will only get a segmentation fault.</p>
<p> To work around this, it is always possible to flip the ownership
flag. For example,</p>
<div class="code">
<pre>
% v -disown # Give ownership to C/C++
% v -acquire # Acquire ownership
</pre>
</div>
<p> It is also possible to deal with situations like this using
typemaps--an advanced topic discussed later.</p>
<h2><a name="Tcl_nn32">37.5 Input and output parameters</a></h2>
<p> A common problem in some C programs is handling parameters passed as
simple pointers. For example:</p>
<div class="code">
<pre>
void add(int x, int y, int *result) {
*result = x + y;
}
</pre>
</div>
<p> or perhaps</p>
<div class="code">
<pre>
int sub(int *x, int *y) {
return *x+*y;
}
</pre>
</div>
<p> The easiest way to handle these situations is to use the <tt>
typemaps.i</tt> file. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
void add(int, int, int *OUTPUT);
int sub(int *INPUT, int *INPUT);
</pre>
</div>
<p> In Tcl, this allows you to pass simple values instead of pointer.
For example:</p>
<div class="code">
<pre>
set a [add 3 4]
puts $a
7
</pre>
</div>
<p> Notice how the <tt>INPUT</tt> parameters allow integer values to be
passed instead of pointers and how the <tt>OUTPUT</tt> parameter
creates a return result.</p>
<p> If you don't want to use the names <tt>INPUT</tt> or <tt>OUTPUT</tt>
, use the <tt>%apply</tt> directive. For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *result };
%apply int *INPUT { int *x, int *y};
void add(int x, int y, int *result);
int sub(int *x, int *y);
</pre>
</div>
<p> If a function mutates one of its parameters like this,</p>
<div class="code">
<pre>
void negate(int *x) {
*x = -(*x);
}
</pre>
</div>
<p> you can use <tt>INOUT</tt> like this:</p>
<div class="code">
<pre>
%include &quot;typemaps.i&quot;
...
void negate(int *INOUT);
</pre>
</div>
<p> In Tcl, a mutated parameter shows up as a return value. For example:</p>
<div class="code">
<pre>
set a [negate 3]
puts $a
-3
</pre>
</div>
<p> The most common use of these special typemap rules is to handle
functions that return more than one value. For example, sometimes a
function returns a result as well as a special error code:</p>
<div class="code">
<pre>
/* send message, return number of bytes sent, along with success code */
int send_message(char *text, int *success);
</pre>
</div>
<p> To wrap such a function, simply use the <tt>OUTPUT</tt> rule above.
For example:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *success };
...
int send_message(char *text, int *success);
</pre>
</div>
<p> When used in Tcl, the function will return multiple values as a
list.</p>
<div class="code">
<pre>
set r [send_message &quot;Hello World&quot;]
set bytes [lindex $r 0]
set success [lindex $r 1]
</pre>
</div>
<p> Another common use of multiple return values are in query functions.
For example:</p>
<div class="code">
<pre>
void get_dimensions(Matrix *m, int *rows, int *columns);
</pre>
</div>
<p> To wrap this, you might use the following:</p>
<div class="code">
<pre>
%module example
%include &quot;typemaps.i&quot;
%apply int *OUTPUT { int *rows, int *columns };
...
void get_dimensions(Matrix *m, int *rows, *columns);
</pre>
</div>
<p> Now, in Perl:</p>
<div class="code">
<pre>
set dim [get_dimensions $m]
set r [lindex $dim 0]
set c [lindex $dim 1]
</pre>
</div>
<h2><a name="Tcl_nn33">37.6 Exception handling</a></h2>
<p> The <tt>%exception</tt> directive can be used to create a
user-definable exception handler in charge of converting exceptions in
your C/C++ program into Tcl exceptions. The chapter on customization
features contains more details, but suppose you extended the array
example into a C++ class like the following :</p>
<div class="code">
<pre>
class RangeError {}; // Used for an exception
class DoubleArray {
private:
int n;
double *ptr;
public:
// Create a new array of fixed size
DoubleArray(int size) {
ptr = new double[size];
n = size;
}
// Destroy an array
~DoubleArray() {
delete ptr;
}
// Return the length of the array
int length() {
return n;
}
// Get an item from the array and perform bounds checking.
double getitem(int i) {
if ((i &gt;= 0) &amp;&amp; (i &lt; n))
return ptr[i];
else
throw RangeError();
}
// Set an item in the array and perform bounds checking.
void setitem(int i, double val) {
if ((i &gt;= 0) &amp;&amp; (i &lt; n))
ptr[i] = val;
else {
throw RangeError();
}
}
};
</pre>
</div>
<p> The functions associated with this class can throw a C++ range
exception for an out-of-bounds array access. We can catch this in our
Tcl extension by specifying the following in an interface file :</p>
<div class="code">
<pre>
%exception {
try {
$action // Gets substituted by actual function call
}
catch (RangeError) {
Tcl_SetResult(interp, (char *)&quot;Array index out-of-bounds&quot;, TCL_STATIC);
return TCL_ERROR;
}
}
</pre>
</div>
<p> As shown, the exception handling code will be added to every wrapper
function. Since this is somewhat inefficient. You might consider
refining the exception handler to only apply to specific methods like
this:</p>
<div class="code">
<pre>
%exception getitem {
try {
$action
}
catch (RangeError) {
Tcl_SetResult(interp, (char *)&quot;Array index out-of-bounds&quot;, TCL_STATIC);
return TCL_ERROR;
}
}
%exception setitem {
try {
$action
}
catch (RangeError) {
Tcl_SetResult(interp, (char *)&quot;Array index out-of-bounds&quot;, TCL_STATIC);
return TCL_ERROR;
}
}
</pre>
</div>
<p> In this case, the exception handler is only attached to methods and
functions named <tt>getitem</tt> and <tt>setitem</tt>.</p>
<p> If you had a lot of different methods, you can avoid extra typing by
using a macro. For example:</p>
<div class="code">
<pre>
%define RANGE_ERROR
{
try {
$action
}
catch (RangeError) {
Tcl_SetResult(interp, (char *)&quot;Array index out-of-bounds&quot;, TCL_STATIC);
return TCL_ERROR;
}
}
%enddef
%exception getitem RANGE_ERROR;
%exception setitem RANGE_ERROR;
</pre>
</div>
<p> Since SWIG's exception handling is user-definable, you are not
limited to C++ exception handling. See the chapter on &quot;<a href="#Customization">
Customization Features</a>&quot; for more examples.</p>
<h2><a name="Tcl_nn34">37.7 Typemaps</a></h2>
<p> This section describes how you can modify SWIG's default wrapping
behavior for various C/C++ datatypes using the <tt>%typemap</tt>
directive. This is an advanced topic that assumes familiarity with the
Tcl C API as well as the material in the &quot;<a href="#Typemaps">Typemaps</a>
&quot; chapter.</p>
<p> Before proceeding, it should be stressed that typemaps are not a
required part of using SWIG---the default wrapping behavior is enough
in most cases. Typemaps are only used if you want to change some aspect
of the primitive C-Tcl interface.</p>
<h3><a name="Tcl_nn35">37.7.1 What is a typemap?</a></h3>
<p> A typemap is nothing more than a code generation rule that is
attached to a specific C datatype. For example, to convert integers
from Tcl to C, you might define a typemap like this:</p>
<div class="code">
<pre>
%module example
%typemap(in) int {
if (Tcl_GetIntFromObj(interp, $input, &amp;$1) == TCL_ERROR)
return TCL_ERROR;
printf(&quot;Received an integer : %d\n&quot;, $1);
}
%inline %{
extern int fact(int n);
%}
</pre>
</div>
<p> Typemaps are always associated with some specific aspect of code
generation. In this case, the &quot;in&quot; method refers to the conversion of
input arguments to C/C++. The datatype <tt>int</tt> is the datatype to
which the typemap will be applied. The supplied C code is used to
convert values. In this code a number of special variable prefaced by a
<tt>$</tt> are used. The <tt>$1</tt> variable is placeholder for a
local variable of type <tt>int</tt>. The <tt>$input</tt> variable is
the input object of type <tt>Tcl_Obj *</tt>.</p>
<p> When this example is compiled into a Tcl module, it operates as
follows:</p>
<div class="code">
<pre>
% load ./example.so
% fact 6
Received an integer : 6
720
</pre>
</div>
<p> In this example, the typemap is applied to all occurrences of the <tt>
int</tt> datatype. You can refine this by supplying an optional
parameter name. For example:</p>
<div class="code">
<pre>
%module example
%typemap(in) int n {
if (Tcl_GetIntFromObj(interp, $input, &amp;$1) == TCL_ERROR)
return TCL_ERROR;
printf(&quot;n = %d\n&quot;, $1);
}
%inline %{
extern int fact(int n);
%}
</pre>
</div>
<p> In this case, the typemap code is only attached to arguments that
exactly match <tt>int n</tt>.</p>
<p> The application of a typemap to specific datatypes and argument
names involves more than simple text-matching--typemaps are fully
integrated into the SWIG type-system. When you define a typemap for <tt>
int</tt>, that typemap applies to <tt>int</tt> and qualified variations
such as <tt>const int</tt>. In addition, the typemap system follows <tt>
typedef</tt> declarations. For example:</p>
<div class="code">
<pre>
%typemap(in) int n {
if (Tcl_GetIntFromObj(interp, $input, &amp;$1) == TCL_ERROR)
return TCL_ERROR;
printf(&quot;n = %d\n&quot;, $1);
}
%inline %{
typedef int Integer;
extern int fact(Integer n); // Above typemap is applied
%}
</pre>
</div>
<p> However, the matching of <tt>typedef</tt> only occurs in one
direction. If you defined a typemap for <tt>Integer</tt>, it is not
applied to arguments of type <tt>int</tt>.</p>
<p> Typemaps can also be defined for groups of consecutive arguments.
For example:</p>
<div class="code">
<pre>
%typemap(in) (char *str, int len) {
$1 = Tcl_GetStringFromObj($input, &amp;$2);
};
int count(char c, char *str, int len);
</pre>
</div>
<p> When a multi-argument typemap is defined, the arguments are always
handled as a single Tcl object. This allows the function to be used
like this (notice how the length parameter is omitted):</p>
<div class="code">
<pre>
% count e &quot;Hello World&quot;
1
</pre>
</div>
<h3><a name="Tcl_nn36">37.7.2 Tcl typemaps</a></h3>
<p> The previous section illustrated an &quot;in&quot; typemap for converting Tcl
objects to C. A variety of different typemap methods are defined by the
Tcl module. For example, to convert a C integer back into a Tcl object,
you might define an &quot;out&quot; typemap like this:</p>
<div class="code">
<pre>
%typemap(out) int {
Tcl_SetObjResult(interp, Tcl_NewIntObj($1));
}
</pre>
</div>
<p> The following list details all of the typemap methods that can be
used by the Tcl module:</p>
<p> <tt>%typemap(in)</tt></p>
<div class="indent"> Converts Tcl objects to input function arguments</div>
<p> <tt>%typemap(out)</tt></p>
<div class="indent"> Converts return value of a C function to a Tcl
object</div>
<p> <tt>%typemap(varin)</tt></p>
<div class="indent"> Assigns a C global variable from a Tcl object</div>
<p> <tt>%typemap(varout)</tt></p>
<div class="indent"> Returns a C global variable as a Tcl object</div>
<p> <tt>%typemap(freearg)</tt></p>
<div class="indent"> Cleans up a function argument (if necessary)</div>
<p> <tt>%typemap(argout)</tt></p>
<div class="indent"> Output argument processing</div>
<p> <tt>%typemap(ret)</tt></p>
<div class="indent"> Cleanup of function return values</div>
<p> <tt>%typemap(consttab)</tt></p>
<div class="indent"> Creation of Tcl constants (constant table)</div>
<p> <tt>%typemap(constcode)</tt></p>
<div class="indent"> Creation of Tcl constants (init function)</div>
<p> <tt>%typemap(memberin)</tt></p>
<div class="indent"> Setting of structure/class member data</div>
<p> <tt>%typemap(globalin)</tt></p>
<div class="indent"> Setting of C global variables</div>
<p> <tt>%typemap(check)</tt></p>
<div class="indent"> Checks function input values.</div>
<p> <tt>%typemap(default)</tt></p>
<div class="indent"> Set a default value for an argument (making it
optional).</div>
<p> <tt>%typemap(arginit)</tt></p>
<div class="indent"> Initialize an argument to a value before any
conversions occur.</div>
<p> Examples of these methods will appear shortly.</p>
<h3><a name="Tcl_nn37">37.7.3 Typemap variables</a></h3>
<p> Within typemap code, a number of special variables prefaced with a <tt>
$</tt> may appear. A full list of variables can be found in the &quot;<a href="#Typemaps">
Typemaps</a>&quot; chapter. This is a list of the most common variables:</p>
<p> <tt>$1</tt></p>
<div class="indent"> A C local variable corresponding to the actual type
specified in the <tt>%typemap</tt> directive. For input values, this is
a C local variable that's supposed to hold an argument value. For
output values, this is the raw result that's supposed to be returned to
Tcl.</div>
<p> <tt>$input</tt></p>
<div class="indent"> A <tt>Tcl_Obj *</tt> holding a raw Tcl object with
an argument or variable value.</div>
<p> <tt>$result</tt></p>
<div class="indent"> A <tt>Tcl_Obj *</tt> that holds the result to be
returned to Tcl.</div>
<p> <tt>$1_name</tt></p>
<div class="indent"> The parameter name that was matched.</div>
<p> <tt>$1_type</tt></p>
<div class="indent"> The actual C datatype matched by the typemap.</div>
<p> <tt>$1_ltype</tt></p>
<div class="indent"> An assignable version of the datatype matched by
the typemap (a type that can appear on the left-hand-side of a C
assignment operation). This type is stripped of qualifiers and may be
an altered version of <tt>$1_type</tt>. All arguments and local
variables in wrapper functions are declared using this type so that
their values can be properly assigned.</div>
<p> <tt>$symname</tt></p>
<div class="indent"> The Tcl name of the wrapper function being created.</div>
<h3><a name="Tcl_nn38">37.7.4 Converting a Tcl list to a char **</a></h3>
<p> A common problem in many C programs is the processing of command
line arguments, which are usually passed in an array of NULL terminated
strings. The following SWIG interface file allows a Tcl list to be used
as a <tt>char **</tt> object.</p>
<div class="code">
<pre>
%module argv
// This tells SWIG to treat char ** as a special case
%typemap(in) char ** {
Tcl_Obj **listobjv;
int nitems;
int i;
if (Tcl_ListObjGetElements(interp, $input, &amp;nitems, &amp;listobjv) == TCL_ERROR) {
return TCL_ERROR;
}
$1 = (char **) malloc((nitems+1)*sizeof(char *));
for (i = 0; i &lt; nitems; i++) {
$1[i] = Tcl_GetString(listobjv[i]);
}
$1[i] = 0;
}
// This gives SWIG some cleanup code that will get called after the function call
%typemap(freearg) char ** {
free($1);
}
// Now a test functions
%inline %{
int print_args(char **argv) {
int i = 0;
while (argv[i]) {
printf(&quot;argv[%d] = %s\n&quot;, i, argv[i]);
i++;
}
return i;
}
%}
%include &quot;tclsh.i&quot;
</pre>
</div>
<p> In Tcl:</p>
<div class="code">
<pre>
% print_args {John Guido Larry}
argv[0] = John
argv[1] = Guido
argv[2] = Larry
3
</pre>
</div>
<h3><a name="Tcl_nn39">37.7.5 Returning values in arguments</a></h3>
<p> The &quot;argout&quot; typemap can be used to return a value originating from
a function argument. For example :</p>
<div class="code">
<pre>
// A typemap defining how to return an argument by appending it to the result
%typemap(argout) double *outvalue {
Tcl_Obj *o = Tcl_NewDoubleObj($1);
Tcl_ListObjAppendElement(interp, $result, o);
}
// A typemap telling SWIG to ignore an argument for input
// However, we still need to pass a pointer to the C function
%typemap(in, numinputs=0) double *outvalue (double temp) {
$1 = &amp;temp;
}
// Now a function returning two values
int mypow(double a, double b, double *outvalue) {
if ((a &lt; 0) || (b &lt; 0)) return -1;
*outvalue = pow(a, b);
return 0;
};
</pre>
</div>
<p> When wrapped, SWIG matches the <tt>argout</tt> typemap to the &quot;<tt>
double *outvalue</tt>&quot; argument. The numinputs=0 specification tells
SWIG to simply ignore this argument when generating wrapper code. As a
result, a Tcl function using these typemaps will work like this :</p>
<div class="code">
<pre>
% mypow 2 3 # Returns two values, a status value and the result
0 8
%
</pre>
</div>
<h3><a name="Tcl_nn40">37.7.6 Useful functions</a></h3>
<p> The following tables provide some functions that may be useful in
writing Tcl typemaps.</p>
<p><b> Integers</b></p>
<div class="code">
<pre>
Tcl_Obj *Tcl_NewIntObj(int Value);
void Tcl_SetIntObj(Tcl_Obj *obj, int Value);
int Tcl_GetIntFromObj(Tcl_Interp *, Tcl_Obj *obj, int *ip);
</pre>
</div>
<p><b> Floating Point</b></p>
<div class="code">
<pre>
Tcl_Obj *Tcl_NewDoubleObj(double Value);
void Tcl_SetDoubleObj(Tcl_Obj *obj, double value);
int Tcl_GetDoubleFromObj(Tcl_Interp *, Tcl_Obj *o, double *dp);
</pre>
</div>
<p><b> Strings</b></p>
<div class="code">
<pre>
Tcl_Obj *Tcl_NewStringObj(char *str, int len);
char *Tcl_GetStringFromObj(Tcl_Obj *obj, int *len);
void Tcl_AppendToObj(Tcl_Obj *obj, char *str, int len);
</pre>
</div>
<p><b> Lists</b></p>
<div class="code">
<pre>
Tcl_Obj *Tcl_NewListObj(int objc, Tcl_Obj *objv);
int Tcl_ListObjAppendList(Tcl_Interp *, Tcl_Obj *listPtr, Tcl_Obj *elemListPtr);
int Tcl_ListObjAppendElement(Tcl_Interp *, Tcl_Obj *listPtr, Tcl_Obj *element);
int Tcl_ListObjGetElements(Tcl_Interp *, Tcl_Obj *listPtr, int *objcPtr,
Tcl_Obj ***objvPtr);
int Tcl_ListObjLength(Tcl_Interp *, Tcl_Obj *listPtr, int *intPtr);
int Tcl_ListObjIndex(Tcl_Interp *, Tcl_Obj *listPtr, int index,
Tcl_Obj_Obj **objptr);
int Tcl_ListObjReplace(Tcl_Interp *, Tcl_Obj *listPtr, int first, int count,
int objc, Tcl_Obj *objv);
</pre>
</div>
<p><b> Objects</b></p>
<div class="code">
<pre>
Tcl_Obj *Tcl_DuplicateObj(Tcl_Obj *obj);
void Tcl_IncrRefCount(Tcl_Obj *obj);
void Tcl_DecrRefCount(Tcl_Obj *obj);
int Tcl_IsShared(Tcl_Obj *obj);
</pre>
</div>
<h3><a name="Tcl_nn41">37.7.7 Standard typemaps</a></h3>
<p> The following typemaps show how to convert a few common kinds of
objects between Tcl and C (and to give a better idea of how typemaps
work)</p>
<p><b> Integer conversion</b></p>
<div class="code">
<pre>
%typemap(in) int, short, long {
int temp;
if (Tcl_GetIntFromObj(interp, $input, &amp;temp) == TCL_ERROR)
return TCL_ERROR;
$1 = ($1_ltype) temp;
}
</pre>
</div>
<br><div class="code">
<pre>
%typemap(out) int, short, long {
Tcl_SetIntObj($result, (int) $1);
}
</pre>
</div>
<p><b> Floating point conversion</b></p>
<div class="code">
<pre>
%typemap(in) float, double {
double temp;
if (Tcl_GetDoubleFromObj(interp, $input, &amp;temp) == TCL_ERROR)
return TCL_ERROR;
$1 = ($1_ltype) temp;
}
</pre>
</div>
<br><div class="code">
<pre>
%typemap(out) float, double {
Tcl_SetDoubleObj($result, $1);
}
</pre>
</div>
<p><b> String Conversion</b></p>
<div class="code">
<pre>
%typemap(in) char * {
int len;
$1 = Tcl_GetStringFromObj(interp, &amp;len);
}
</pre>
</div>
<br><div class="code">
<pre>
%typemap(out, noblock=1, fragment=&quot;SWIG_FromCharPtr&quot;) char *, const char * {
Tcl_SetObjResult(interp, SWIG_FromCharPtr((const char *)$1));
}
</pre>
</div>
<h3><a name="Tcl_nn42">37.7.8 Pointer handling</a></h3>
<p> SWIG pointers are mapped into Tcl strings containing the hexadecimal
value and type. The following functions can be used to create and read
pointer values.</p>
<p> <tt>int SWIG_ConvertPtr(Tcl_Obj *obj, void **ptr, swig_type_info
*ty, int flags)</tt></p>
<div class="indent"> Converts a Tcl object <tt>obj</tt> to a C pointer.
The result of the conversion is placed into the pointer located at <tt>
ptr</tt>. <tt>ty</tt> is a SWIG type descriptor structure. <tt>flags</tt>
is used to handle error checking and other aspects of conversion. It is
currently reserved for future expansion. Returns 0 on success and -1 on
error.</div>
<p> <tt>Tcl_Obj *SWIG_NewPointerObj(void *ptr, swig_type_info *ty, int
flags)</tt></p>
<div class="indent"> Creates a new Tcl pointer object. <tt>ptr</tt> is
the pointer to convert, <tt>ty</tt> is the SWIG type descriptor
structure that describes the type, and <tt>own</tt> is a flag reserved
for future expansion.</div>
<p> Both of these functions require the use of a special SWIG
type-descriptor structure. This structure contains information about
the mangled name of the datatype, type-equivalence information, as well
as information about converting pointer values under C++ inheritance.
For a type of <tt>Foo *</tt>, the type descriptor structure is usually
accessed as follows:</p>
<div class="indent">
<pre>
Foo *f;
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;f, SWIGTYPE_p_Foo, 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method '$symname', expecting type Foo&quot;);
}
Tcl_Obj *;
obj = SWIG_NewPointerObj(f, SWIGTYPE_p_Foo, 0);
</pre>
</div>
<p> In a typemap, the type descriptor should always be accessed using
the special typemap variable <tt>$1_descriptor</tt>. For example:</p>
<div class="indent">
<pre>
%typemap(in) Foo * {
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;$1, $1_descriptor, 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method '$symname', expecting type Foo&quot;);
}
}
</pre>
</div>
<p> If necessary, the descriptor for any type can be obtained using the <tt>
$descriptor()</tt> macro in a typemap. For example:</p>
<div class="indent">
<pre>
%typemap(in) Foo * {
if (!SWIG_IsOK(SWIG_ConvertPtr($input, (void **) &amp;$1, $descriptor(Foo *), 0))) {
SWIG_exception_fail(SWIG_TypeError, &quot;in method '$symname', expecting type Foo&quot;);
}
}
</pre>
</div>
<h2><a name="Tcl_nn43">37.8 Turning a SWIG module into a Tcl Package.</a>
</h2>
<p> SWIG generates all of the code necessary to create a Tcl extension
package. To set the package version use the <tt>-pkgversion</tt>
option. For example:</p>
<div class="code">
<pre>
% swig -tcl -pkgversion 2.3 example.i
</pre>
</div>
<p> After building the SWIG generated module, you need to execute the <tt>
pkg_mkIndex</tt> command inside tclsh. For example :</p>
<div class="code">
<pre>
unix &gt; tclsh
% pkg_mkIndex . example.so
% exit
</pre>
</div>
<p> This creates a file &quot;<tt>pkgIndex.tcl</tt>&quot; with information about
the package. To use your package, you now need to move it to its own
subdirectory which has the same name as the package. For example :</p>
<div class="code">
<pre>
./example/
pkgIndex.tcl # The file created by pkg_mkIndex
example.so # The SWIG generated module
</pre>
</div>
<p> Finally, assuming that you're not entirely confused at this point,
make sure that the example subdirectory is visible from the directories
contained in either the <tt>tcl_library</tt> or <tt>auto_path</tt>
variables. At this point you're ready to use the package as follows :</p>
<div class="code">
<pre>
unix &gt; tclsh
% package require example
% fact 4
24
%
</pre>
</div>
<p> If you're working with an example in the current directory and this
doesn't work, do this instead :</p>
<div class="code">
<pre>
unix &gt; tclsh
% lappend auto_path .
% package require example
% fact 4
24
</pre>
</div>
<p> As a final note, most SWIG examples do not yet use the <tt>package</tt>
commands. For simple extensions it may be easier just to use the <tt>
load</tt> command instead.</p>
<h2><a name="Tcl_nn44">37.9 Building new kinds of Tcl interfaces (in
Tcl)</a></h2>
<p> One of the most interesting aspects of Tcl and SWIG is that you can
create entirely new kinds of Tcl interfaces in Tcl using the low-level
SWIG accessor functions. For example, suppose you had a library of
helper functions to access arrays :</p>
<div class="code">
<pre>
/* File : array.i */
%module array
%inline %{
double *new_double(int size) {
return (double *) malloc(size*sizeof(double));
}
void delete_double(double *a) {
free(a);
}
double get_double(double *a, int index) {
return a[index];
}
void set_double(double *a, int index, double val) {
a[index] = val;
}
int *new_int(int size) {
return (int *) malloc(size*sizeof(int));
}
void delete_int(int *a) {
free(a);
}
int get_int(int *a, int index) {
return a[index];
}
int set_int(int *a, int index, int val) {
a[index] = val;
}
%}
</pre>
</div>
<p> While these could be called directly, we could also write a Tcl
script like this :</p>
<div class="code">
<pre>
proc Array {type size} {
set ptr [new_$type $size]
set code {
set method [lindex $args 0]
set parms [concat $ptr [lrange $args 1 end]]
switch $method {
get {return [eval &quot;get_$type $parms&quot;]}
set {return [eval &quot;set_$type $parms&quot;]}
delete {eval &quot;delete_$type $ptr; rename $ptr {}&quot;}
}
}
# Create a procedure
uplevel &quot;proc $ptr args {set ptr $ptr; set type $type;$code}&quot;
return $ptr
}
</pre>
</div>
<p> Our script allows easy array access as follows :</p>
<div class="code">
<pre>
set a [Array double 100] ;# Create a double [100]
for {set i 0} {$i &lt; 100} {incr i 1} { ;# Clear the array
$a set $i 0.0
}
$a set 3 3.1455 ;# Set an individual element
set b [$a get 10] ;# Retrieve an element
set ia [Array int 50] ;# Create an int[50]
for {set i 0} {$i &lt; 50} {incr i 1} { ;# Clear it
$ia set $i 0
}
$ia set 3 7 ;# Set an individual element
set ib [$ia get 10] ;# Get an individual element
$a delete ;# Destroy a
$ia delete ;# Destroy ia
</pre>
</div>
<p> The cool thing about this approach is that it makes a common
interface for two different types of arrays. In fact, if we were to add
more C datatypes to our wrapper file, the Tcl code would work with
those as well--without modification. If an unsupported datatype was
requested, the Tcl code would simply return with an error so there is
very little danger of blowing something up (although it is easily
accomplished with an out of bounds array access).</p>
<h3><a name="Tcl_nn45">37.9.1 Proxy classes</a></h3>
<p> A similar approach can be applied to proxy classes (also known as
shadow classes). The following example is provided by Erik Bierwagen
and Paul Saxe. To use it, run SWIG with the <tt>-noobject</tt> option
(which disables the builtin object oriented interface). When running
Tcl, simply source this file. Now, objects can be used in a more or
less natural fashion.</p>
<div class="code">
<pre>
# swig_c++.tcl
# Provides a simple object oriented interface using
# SWIG's low level interface.
#
proc new {objectType handle_r args} {
# Creates a new SWIG object of the given type,
# returning a handle in the variable &quot;handle_r&quot;.
#
# Also creates a procedure for the object and a trace on
# the handle variable that deletes the object when the
# handle variable is overwritten or unset
upvar $handle_r handle
#
# Create the new object
#
eval set handle \[new_$objectType $args\]
#
# Set up the object procedure
#
proc $handle {cmd args} &quot;eval ${objectType}_\$cmd $handle \$args&quot;
#
# And the trace ...
#
uplevel trace variable $handle_r uw &quot;{deleteObject $objectType $handle}&quot;
#
# Return the handle so that 'new' can be used as an argument to a procedure
#
return $handle
}
proc deleteObject {objectType handle name element op} {
#
# Check that the object handle has a reasonable form
#
if {![regexp {_[0-9a-f]*_(.+)_p} $handle]} {
error &quot;deleteObject: not a valid object handle: $handle&quot;
}
#
# Remove the object procedure
#
catch {rename $handle {}}
#
# Delete the object
#
delete_$objectType $handle
}
proc delete {handle_r} {
#
# A synonym for unset that is more familiar to C++ programmers
#
uplevel unset $handle_r
}
</pre>
</div>
<p> To use this file, we simply source it and execute commands such as
&quot;new&quot; and &quot;delete&quot; to manipulate objects. For example :</p>
<div class="code">
<pre>
// list.i
%module List
%{
#include &quot;list.h&quot;
%}
// Very simple C++ example
class List {
public:
List(); // Create a new list
~List(); // Destroy a list
int search(char *value);
void insert(char *); // Insert a new item into the list
void remove(char *); // Remove item from list
char *get(int n); // Get the nth item in the list
int length; // The current length of the list
static void print(List *l); // Print out the contents of the list
};
</pre>
</div>
<p> Now a Tcl script using the interface...</p>
<div class="code">
<pre>
load ./list.so list ; # Load the module
source swig_c++.tcl ; # Source the object file
new List l
$l insert Dave
$l insert John
$l insert Guido
$l remove Dave
puts $l length_get
delete l
</pre>
</div>
<p> The cool thing about this example is that it works with any C++
object wrapped by SWIG and requires no special compilation. Proof that
a short, but clever Tcl script can be combined with SWIG to do many
interesting things.</p>
<h2><a name="Tcl_nn46">37.10 Tcl/Tk Stubs</a></h2>
<p> For background information about the Tcl Stubs feature, see <a href="https://www.tcl.tk/doc/howto/stubs.html">
https://www.tcl.tk/doc/howto/stubs.html</a>.</p>
<p> As of SWIG 1.3.10, the generated C/C++ wrapper will use the Tcl
Stubs feature if compiled with <tt>-DUSE_TCL_STUBS</tt>.</p>
<p> As of SWIG 1.3.40, the generated C/C++ wrapper will use the Tk Stubs
feature if compiled with <tt>-DUSE_TK_STUBS</tt>.</p>
<p> By default SWIG sets the minimum Tcl version to support to the 8.4
as that's the minimum Tcl version we aim to support (since SWIG 4.1.0;
before this SWIG set it to 8.1, which was the first Tcl version with
the stubs mechanism). This minimum version is passed to <tt>
Tcl_InitStubs()</tt> and <tt>Tk_InitStubs()</tt>. You can override with
a specific version using <tt>-DSWIG_TCL_STUBS_VERSION=&quot;8.5&quot;</tt> or set
it to the Tcl version being compiled with using <tt>
-DSWIG_TCL_STUBS_VERSION=TCL_VERSION</tt>.</p>
<HR NOSHADE>
<h1><a name="C">38 SWIG and C as the target language</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#C_overview">Overview</a>
<ul>
<li><a href="#C_shortcomings">Known C++ Shortcomings in Generated C API</a>
</li>
</ul>
</li>
<li><a href="#C_preliminaries">Preliminaries</a>
<ul>
<li><a href="#C_running_swig">Running SWIG</a></li>
<li><a href="#C_commandline">Command line options</a></li>
<li><a href="#C_dynamic">Compiling a dynamic module</a></li>
<li><a href="#C_using_module">Using the generated module</a></li>
</ul>
</li>
<li><a href="#C_basic_c_wrapping">Basic C wrapping</a>
<ul>
<li><a href="#C_functions">Functions</a></li>
<li><a href="#C_variables">Variables</a></li>
</ul>
</li>
<li><a href="#C_basic_cpp_wrapping">Basic C++ wrapping</a>
<ul>
<li><a href="#C_enums">Enums</a></li>
<li><a href="#C_classes">Classes</a></li>
</ul>
</li>
<li><a href="#C_developer">Backend Developer Documentation</a></li>
<li><a href="#C_typemaps">Typemaps</a>
<ul>
<li><a href="#C_typemaps_walkthrough">C Typemaps, a Code Generation
Walkthrough</a>
<ul>
<li><a href="#C_walkthrough_interface">The Interface</a></li>
<li><a href="#C_walkthrough_wrapper">The Wrapper</a></li>
<li><a href="#C_walkthrough_proxy">The Proxy</a></li>
</ul>
</li>
</ul>
</li>
<li><a href="#C_exceptions">Exception handling</a></li>
<li><a href="#C_cxx_wrappers">C++ Wrappers</a>
<ul>
<li><a href="#C_additional_possibilities">Additional customization
possibilities</a></li>
<li><a href="#C_exception_handling">Exception handling</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> This chapter describes SWIG's support for creating ISO C wrappers.
This module has a special purpose and thus is different from most other
modules.</p>
<p><b> NOTE:</b> this module is still under development.</p>
<h2><a name="C_overview">38.1 Overview</a></h2>
<p> SWIG is normally used to provide access to C or C++ libraries from
target languages such as scripting languages or languages running on a
virtual machine. SWIG performs analysis of the input C/C++ library
header files from which it generates further code. For most target
languages this code consists of two layers; namely an intermediary C
code layer and a set of language specific proxy classes and functions
on top of the C code layer. We could also think of C as just another
target language supported by SWIG. The aim then is to generate a pure
ISO C interface to the input C or C++ library and hence the C target
language module.</p>
<p> With wrapper interfaces generated by SWIG, it is easy to use the
functionality of C++ libraries inside application code written in C.
This module may also be useful to generate custom APIs for a library,
to suit particular needs, e.g. to supply function calls with error
checking or to implement a &quot;design by contract&quot;.</p>
<p> Flattening C++ language constructs into a set of C-style functions
obviously comes with many limitations and inconveniences, but this
module is actually also capable of generating C++ wrappers defined
completely inline using the C functions, thus wrapping the original C++
library API in another, similar C++ API. Contrary to the natural
initial reaction, this is far from being completely pointless, as
wrapping C++ API in this way avoids all problems due to C++ ABI issues,
e.g. it is now possible to use the original C++ API using a different
C++ compiler, or a different version of the same compiler, or even the
same compiler, but with different compilation options affecting the
ABI. The C++ wrapper API is not identical to the original one, but
strives to be as close to it as possible.</p>
<h3><a name="C_shortcomings">38.1.1 Known C++ Shortcomings in Generated
C API</a></h3>
<ul>
<li>Enums with a context like class or namespace are broken</li>
<li>Global variables are not supported</li>
<li>Qualifiers are stripped</li>
<li>Vararg functions are not supported.</li>
</ul>
<h2><a name="C_preliminaries">38.2 Preliminaries</a></h2>
<h3><a name="C_running_swig">38.2.1 Running SWIG</a></h3>
<p> Consider the following simple example. Suppose we have an interface
file like:</p>
<div class="code">
<pre>
/* File: example.i */
%module test
%{
#include &quot;stuff.h&quot;
%}
int fact(int n);
</pre>
</div>
<p> To build a C module (C as the target language), run SWIG using the <tt>
-c</tt> option :</p>
<div class="shell">
<pre>
$ swig -c example.i
</pre>
</div>
<p> The above assumes C as the input language. If the input language is
C++ add the <tt>-c++</tt> option:</p>
<div class="shell">
<pre>
$ swig -c++ -c example.i
</pre>
</div>
<p> Note that <tt>-c</tt> is the option specifying the<b> target</b>
language and <tt>-c++</tt> controls what the<b> input</b> language is.</p>
<p> This will generate an <tt>example_wrap.c</tt> file or, in the latter
case, <tt>example_wrap.cxx</tt> file, along with <tt>example_wrap.h</tt>
(the same extension is used in both C and C++ cases for the last one).
The names of the files are derived from the name of the input file by
default, but can be changed using the <tt>-o</tt> and <tt>-oh</tt>
options common to all language modules.</p>
<p> The <tt>xxx_wrap.c</tt> file contains the wrapper functions, which
perform the main functionality of SWIG: each of the wrappers translates
the input arguments from C to C++, makes calls to the original
functions and marshals C++ output back to C data. The <tt>xxx_wrap.h</tt>
header file contains the declarations of these functions as well as
global variables.</p>
<h3><a name="C_commandline">38.2.2 Command line options</a></h3>
<p> The following table list the additional command line options
available for the C module. They can also be seen by using:</p>
<div class="shell">
<pre>
$ swig -c -help
</pre>
</div>
<table summary="C specific options">
<tr><th>C specific options</th></tr>
<tr><td>-namespace &lt;nspace&gt;</td><td>Generate wrappers with the prefix
based on the provided namespace, e.g. if the option value is <tt>
outer::inner</tt>, the prefix <tt>outer_inner_</tt> will be used. Notice
that this is different from using SWIG <tt>nspace</tt> feature, as it
applies the prefix to all the symbols, regardless of the namespace they
were actually declared in. Notably, this provides a way to export
instantiations of templates defined in the <tt>std</tt> namespace, such
as <tt>std::vector</tt>, using a custom prefix rather than <tt>std_</tt>
.</td></tr>
<tr><td>-nocxx</td><td>Don't generate C++ wrappers, even when the <tt>
-c++</tt> option is used. See <a href="#C_cxx_wrappers">C++ Wrappers</a>
section for more details.</td></tr>
<tr><td>-noexcept</td><td>generate wrappers with no support for
exception handling; see <a href="#C_exceptions">Exceptions</a> chapter
for more details</td></tr>
</table>
<h3><a name="C_dynamic">38.2.3 Compiling a dynamic module</a></h3>
<p> The next step is to build a dynamically loadable module, which we
can link to our application. For example, to do this using the <tt>gcc</tt>
compiler (Linux, MinGW, etc.):</p>
<div class="shell">
<pre>
$ swig -c example.i
$ gcc -fPIC -c example_wrap.c
$ gcc -shared example_wrap.o -o libexample.so
</pre>
</div>
<p> Or, for C++ input:</p>
<div class="shell">
<pre>
$ swig -c++ -c example.i
$ g++ -fPIC -c example_wrap.cxx
$ g++ -shared example_wrap.o -o libexample.so
</pre>
</div>
<p> Now the shared library module is ready to use. Note that the name of
the generated module is important: is should be prefixed with <tt>lib</tt>
on Unix, and have the specific extension, like <tt>.dll</tt> for
Windows or <tt>.so</tt> for Unix systems.</p>
<h3><a name="C_using_module">38.2.4 Using the generated module</a></h3>
<p> The simplest way to use the generated shared module is to link it to
the application code during the compilation stage. The process is
usually similar to this:</p>
<div class="shell">
<pre>
$ gcc runme.c -L. -lexample -o runme
</pre>
</div>
<p> This will compile the application code (<tt>runme.c</tt>) and link
it against the generated shared module. Following the <tt>-L</tt>
option is the path to the directory containing the shared module. The
output executable is ready to use. The last thing to do is to supply to
the operating system the information of location of our module. This is
system dependant, for instance Unix systems look for shared modules in
certain directories, like <tt>/usr/lib</tt>, and additionally we can
set the environment variable <tt>LD_LIBRARY_PATH</tt> (Unix) or <tt>
PATH</tt> (Windows) for other directories.</p>
<h2><a name="C_basic_c_wrapping">38.3 Basic C wrapping</a></h2>
<p> Wrapping C functions and variables is obviously performed in a
straightforward way. There is no need to perform type conversions, and
all language constructs can be preserved in their original form.
However, SWIG allows you to enhance the code with some additional
elements, for instance using a <tt>check</tt> typemap or <tt>%extend</tt>
directive.</p>
<p> It is also possible to output arbitrary additional code into the
generated header by using the <tt>%insert</tt> directive with <tt>
cheader</tt> section, e.g.</p>
<div class="code">
<pre>
%insert(&quot;cheader&quot;) %{
#include &quot;another.h&quot;
%}
</pre>
</div>
<h3><a name="C_functions">38.3.1 Functions</a></h3>
<p> For each C function declared in the interface file a wrapper
function with a prefix, required to make its name different from the
original one, is created. The prefix for the global functions is <tt>
module_</tt>, i.e. the name of the SWIG module followed by underscore,
by default. If <tt>-namespace</tt> option is used, the prefix
corresponding to the given fixed namespace is used instead. If <tt>
nspace</tt> feature is used, the prefix corresponding to the namespace
in which the function is defined is used -- note that, unlike with <tt>
-namespace</tt> option, this prefix can be different for different
functions. The wrapper function performs a call to the original
function, and returns its result.</p>
<p> For example, for function declaration in the module <tt>mymath</tt>:</p>
<div class="targetlang">
<pre>
int gcd(int x, int y);
</pre>
</div>
<p> The output is simply:</p>
<div class="targetlang">
<pre>
int mymath_gcd(int arg1, int arg2) {
int result;
result = gcd(arg1,arg2);
return result;
}
</pre>
</div>
<p> Now one might think, what's the use of creating such functions in C?
The answer is, you can apply special rules to the generated code. Take
for example constraint checking. You can write a &quot;check&quot; typemap in
your interface file:</p>
<div class="code">
<pre>
%typemap(check) int POSITIVE {
if ($1 &lt;= 0)
fprintf(stderr, &quot;Expected positive value for parameter $1 in $name.\n&quot;);
}
int gcd(int POSITIVE, int POSITIVE);
</pre>
</div>
<p> And now the generated result looks like:</p>
<div class="targetlang">
<pre>
int _wrap_gcd(int arg1, int arg2) {
int result;
{
if (arg1 &lt;= 0)
fprintf(stderr, &quot;Expected positive value for parameter arg1 in gcd.\n&quot;);
}
{
if (arg2 &lt;= 0)
fprintf(stderr, &quot;Expected positive value for parameter arg2 in gcd.\n&quot;);
}
result = gcd(arg1,arg2);
return result;
}
</pre>
</div>
<p> This time calling <tt>gcd</tt> with negative value argument will
trigger an error message. This can save you time writing all the
constraint checking code by hand.</p>
<h3><a name="C_variables">38.3.2 Variables</a></h3>
<p> Wrapping variables comes also without any special issues. All global
variables are directly accessible from application code. There is a
difference in the semantics of <tt>struct</tt> definition in C and C++.
When handling C <tt>struct</tt>, SWIG simply rewrites its declaration.
In C++ <tt>struct</tt> is handled as class declaration.</p>
<p> You can still apply some of the SWIG features when handling structs,
e.g. <tt>%extend</tt> directive. Suppose, you have a C struct
declaration:</p>
<div class="targetlang">
<pre>
typedef struct {
int x;
char *str;
} my_struct;
</pre>
</div>
<p> You can redefine it to have an additional fields, like:</p>
<div class="code">
<pre>
%extend my_struct {
double d;
};
</pre>
</div>
<p> In application code:</p>
<div class="targetlang">
<pre>
struct my_struct ms;
ms.x = 123;
ms.d = 123.123;
</pre>
</div>
<h2><a name="C_basic_cpp_wrapping">38.4 Basic C++ wrapping</a></h2>
<p> The main reason of having the C module in SWIG is to be able to
access C++ from C. In this chapter we will take a look at the rules of
wrapping elements of the C++ language.</p>
<p> By default, SWIG attempts to build a natural C interface to your
C/C++ code.</p>
<table BORDER summary="Generated C representation of C++">
<tr><th>C++ Type</th><th>SWIG C Translation</th></tr>
<tr><td>Class <tt>Example</tt></td><td>Empty structure <tt>Example</tt></td>
</tr>
<tr><td>Public, mutable member variable <tt>Foo Example::foo</tt></td><td>
<div class="targetlang">
<pre>
Example_foo_get(Example *e);
Example_foo_set(Example *e, Foo *f);
</pre>
</div></td></tr>
<tr><td>Public, immutable member variable <tt>Foo Example::bar</tt></td><td>
<div class="targetlang">
<pre>
Example_foo_get(Example *e);
</pre>
</div></td></tr>
</table>
<p> This section briefly covers the essential aspects of this wrapping.</p>
<h3><a name="C_enums">38.4.1 Enums</a></h3>
<p> C enums and unscoped C++ enums are simply copied to the generated
code and both the enum itself and its elements keep the same name as in
the original code unless <tt>-namespace</tt> option is used or <tt>
nspace</tt> feature is enabled, in which case the prefix corresponding
to the specified namespace is used.</p>
<p> For scoped C++11 enums, the enum name itself is used as an
additional prefix.</p>
<h3><a name="C_classes">38.4.2 Classes</a></h3>
<p> Consider the following example. We have a C++ class, and want to use
it from C code.</p>
<div class="targetlang">
<pre>
class Circle {
public:
double radius;
Circle(double r) : radius(r) { };
double area(void);
};
</pre>
</div>
<p> What we need to do is to create an object of the class, manipulate
it, and finally, destroy it. SWIG generates C functions for this
purpose each time a class declaration is encountered in the interface
file.</p>
<p> The first two generated functions are used to create and destroy
instances of class <tt>Circle</tt>. Such instances are represented on
the C side as pointers to special structs, called <tt>SwigObj</tt>.
They are all &quot;renamed&quot; (via typedef) to the original class names, so
that you can use the object instances on the C side using pointers
like:</p>
<div class="targetlang">
<pre>
Circle *circle;
</pre>
</div>
<p> The generated functions make calls to class' constructors and
destructors, respectively. They also do all the necessary things
required by the SWIG object management system in C.</p>
<div class="targetlang">
<pre>
Circle * Circle_new(double r);
void Circle_delete(Circle * self);
</pre>
</div>
<p> The class <tt>Circle</tt> has a public variable called <tt>radius</tt>
. SWIG generates a pair of setters and getters for each such variable:</p>
<div class="targetlang">
<pre>
void Circle_radius_set(Circle * self, double radius);
double Circle_radius_get(Circle * self);
</pre>
</div>
<p> For each public method, an appropriate function is generated:</p>
<div class="targetlang">
<pre>
double Circle_area(Circle * self);
</pre>
</div>
<p> You can see that in order to use the generated object we need to
provide a pointer to the object instance (struct <tt>Circle</tt> in
this case) as the first function argument. In fact, this struct is
basically wrapping pointer to the &quot;real&quot; C++ object.</p>
<p> Our application code could look like this:</p>
<div class="targetlang">
<pre>
Circle *c = Circle_new(1.5);
printf(&quot;radius: %f\narea: %f\n&quot;, Circle_radius_get(c), Circle_area(c));
Circle_delete(c);
</pre>
</div>
<p> After running this we'll get:</p>
<div class="shell">
<pre>
radius: 1.500000
area: 7.068583
</pre>
</div>
<h2><a name="C_developer">38.5 Backend Developer Documentation</a></h2>
<h2><a name="C_typemaps">38.6 Typemaps</a></h2>
<table BORDER summary="C Backend Typemaps">
<tr><th>Typemap</th><th>Used for</th></tr>
<tr><td><tt>ctype</tt></td><td>Provides types used for the C API and
<br> Typecasts wrapper functions return values in proxy functions<div class="targetlang">
<pre>
MyClass *MyClass_new(void) {
return (MyClass *)MyClass_new();
}
</pre>
</div></td></tr>
<tr><td><tt>in</tt></td><td>Mapping of wrapper functions parameters to
local C++ variables<div class="targetlang">
<pre>
SwigObj* MyClass_do(SwigObj *carg1) {
SomeCPPClass *arg1 = 0;
if (carg1)
arg1 = (SomeCPPClass*)carg1-&gt;obj
else
arg1 = 0;
}
</pre>
</div></td></tr>
<tr><td><tt>out</tt></td><td>Assigns wrapped function's return value to
a dedicated return variable, packaging it into SwigObj if necessary</td>
</tr>
<tr><td><tt>cppouttype</tt></td><td>Type of the result variable used for
the return value if the wrapped function is a C++ function</td></tr>
<tr><td><tt>cxxintype</tt></td><td>Defines the type for the parameters
of C++ wrapper functions corresponding to this type. By default is the
same as <tt>ctype</tt>, but may sometimes be different to make the
functions more convenient to use. For example, <tt>ctype</tt> for <tt>
std::string</tt> is <tt>const char*</tt>, but <tt>cxxintype</tt> typemap
for it is <tt>std::string const&amp;</tt>, i.e. even though the C++ string
passed as a raw pointer via C API, the C++ wrapper still accepts a C++
string. If this typemap is defined, <tt>cxxin</tt> should normally be
defined as well. If it is not defined, <tt>ctype</tt> is used.</td></tr>
<tr><td><tt>cxxouttype</tt></td><td>Similar to <tt>cxxintype</tt>, but
is used for the function return values and together with <tt>cxxout</tt>
typemap. Also defaults to <tt>ctype</tt> if not defined.</td></tr>
<tr><td><tt>cxxin</tt></td><td>Defines how to transform <tt>cxxintype</tt>
value to <tt>ctype</tt></td></tr>
<tr><td><tt>cxxout</tt></td><td>Defines how to transform <tt>ctype</tt>
value returned by a function to <tt>cxxouttype</tt></td></tr>
<tr><td><tt>cxxcode</tt></td><td>May contain arbitrary code that will be
injected in the declaration of the C++ wrapper class corresponding to
the given type. Ignored for non-class types. The special variable <tt>
$cxxclassname</tt> is replaced with the name of the class inside this
typemap expansion and <tt>$cclassptrname</tt> is replaced with the name
of the pointer type used to represent the class in C wrapper functions.</td>
</tr>
</table>
<h3><a name="C_typemaps_walkthrough">38.6.1 C Typemaps, a Code
Generation Walkthrough</a></h3>
<p> To get a better idea of which typemap is used for which generated
code, have a look at the following 'walk through'.</p>
<p> Let's assume we have the following C++ interface file, we'd like to
generate code for:</p>
<h4><a name="C_walkthrough_interface">38.6.1.1 The Interface</a></h4>
<div class="code">
<pre>
%module example
%inline
%{
class SomeClass{};
template &lt;typename T&gt; class SomeTemplateClass{};
SomeClass someFunction(SomeTemplateClass&lt;int&gt; &amp;someParameter, int simpleInt);
%}
%template (SomeIntTemplateClass) SomeTemplateClass&lt;int&gt;;
</pre>
</div>
<p> What we would like to generate as a C interface of this function
would be something like this:</p>
<div class="targetlang">
<pre>
// wrapper header file
typedef struct SwigObj_SomeClass SomeClass;
SomeClass * SomeClass_new();
void SomeClass_delete(SomeClass * carg1);
SomeClass* someFunction(SomeIntTemplateClass* carg1, int carg2);
typedef struct SwigObj_SomeIntTemplateClass SomeIntTemplateClass;
SomeIntTemplateClass * SomeIntTemplateClass_new();
void SomeIntTemplateClass_delete(SomeIntTemplateClass * carg1);
</pre>
</div>
<h4><a name="C_walkthrough_wrapper">38.6.1.2 The Wrapper</a></h4>
<p> We'll examine the generation of the wrapper function first.</p>
<div class="targetlang">
<pre>
SWIGEXPORTC SwigObj * module_someFunction(SwigObj * carg1, int carg2) {
SomeClass * cppresult;
SomeTemplateClass&lt; int &gt; *arg1 = 0 ;
int arg2 ;
SwigObj * result;
{
if (carg1)
arg1 = (SomeTemplateClass&lt; int &gt; *) carg1-&gt;obj;
else
arg1 = (SomeTemplateClass&lt; int &gt; *) 0;
}
arg2 = (int) carg2;
{
const SomeClass &amp;_result_ref = someFunction(*arg1,arg2);cppresult = (SomeClass*) &amp;_result_ref;
}
{
result = SWIG_create_object(cppresult, SWIG_STR(SomeClass));
}
return result;
}
</pre>
</div>
<p> It might be helpful to think of the way function calls are generated
as a composition of building blocks. A typical wrapper will be
composited with these [optional] blocks:</p>
<ol>
<li>Prototype</li>
<li>C return value variable</li>
<li>Local variables equal to the called C++ function's parameters</li>
<li>[C++ return value variable]</li>
<li>Assignment (extraction) of wrapper parameters to local parameter
copies</li>
<li>[Contract (e.g. constraints) checking]</li>
<li> C++ function call</li>
<li>[Exception handling]</li>
<li>[Assignment to C++ return value]</li>
<li>Assignment to C return value</li>
</ol>
<p> Let's go through it step by step and start with the wrapper
prototype</p>
<div class="targetlang">
<pre>
ctype ctype ctype
--------- --------- ---
SwigObj * module_someFunction(SwigObj * carg1, int carg2);
</pre>
</div>
<p> As first unit of the wrapper code, a variable to hold the return
value of the function is emitted to the wrapper's body</p>
<div class="targetlang">
<pre>
ctype
---------
SwigObj * result;
</pre>
</div>
<p> Now for each of the C++ function's arguments, a local variable with
the very same type is emitted to the wrapper's body.</p>
<div class="targetlang">
<pre>
SomeTemplateClass&lt; int &gt; *arg1 = 0 ;
int arg2 ;
</pre>
</div>
<p> If it's a C++ function that is wrapped (in this case it is), another
variable is emitted for the 'original' return value of the C++
function. At this point, we simply 'inject' behavior if it's a C++
function that is wrapped (in this case it obviously is).</p>
<div class="targetlang">
<pre>
cppouttype
-----------
SomeClass * cppresult;
</pre>
</div>
<p> Next, the values of the input parameters are assigned to the local
variables using the 'in' typemap.</p>
<div class="targetlang">
<pre>
{
if (carg1)
arg1 = (SomeTemplateClass&lt; int &gt; *) carg1-&gt;obj;
else
arg1 = (SomeTemplateClass&lt; int &gt; *) 0;
}
arg2 = (int) carg2;
</pre>
</div>
<p> A reasonable question would be: &quot;Why aren't the parameters assigned
in the declaration of their local counterparts?&quot; As seen above, for
complex types pointers have to be verified before extracting and
casting the actual data pointer from the provided SwigObj pointer. This
could easily become messy if it was done in the same line with the
local variable declaration.</p>
<p> At this point we are ready to call the C++ function with our
parameters.</p>
<div class="targetlang">
<pre>
{
const SomeClass &amp;_result_ref = someFunction(*arg1,arg2);cppresult = (SomeClass*) &amp;_result_ref;
}
</pre>
</div>
<p> Subsequently, the return value is assigned to the dedicated return
value variable using the 'out' typemap</p>
<div class="targetlang">
<pre>
{
result = SWIG_create_object(cppresult, SWIG_STR(SomeClass));
}
</pre>
</div>
<p> Finally, the return value variable is returned.</p>
<div class="targetlang">
<pre>
return result;
</pre>
</div>
<p> Note that typemaps may use <tt>$null</tt> special variable which
will be replaced with either <tt>0</tt> or nothing, depending on
whether the function has a non-void return value or not.</p>
<h4><a name="C_walkthrough_proxy">38.6.1.3 The Proxy</a></h4>
<p> Compared to the wrapper code generation, the header code is very
simple. Basically it contains just the declarations corresponding to
the definitions above.</p>
<div class="targetlang">
<pre>
// wrapper header file
typedef struct SwigObj_SomeClass SomeClass;
SomeClass * SomeClass_new();
void SomeClass_delete(SomeClass * carg1);
SomeClass* someFunction(SomeIntTemplateClass* carg1, int carg2);
typedef struct SwigObj_SomeIntTemplateClass SomeIntTemplateClass;
SomeIntTemplateClass * SomeIntTemplateClass_new();
void SomeIntTemplateClass_delete(SomeIntTemplateClass * carg1);
</pre>
</div>
<h2><a name="C_exceptions">38.7 Exception handling</a></h2>
<p> Any call to a C++ function may throw an exception, which cannot be
caught by C code. Instead, the special <tt>
SWIG_CException_get_pending()</tt> function must be called to check for
this. If it returns a non-null pointer, <tt>SWIG_CException_msg_get()</tt>
can be called to retrieve the error message associated with the
exception. Finally, <tt>SWIG_CException_reset_pending()</tt> must be
called to free the exception object and reset the current pending
exception. Note that exception handling is much simpler when using C++,
rather than C, wrappers, see sections 36.6.2.</p>
<h2><a name="C_cxx_wrappers">38.8 C++ Wrappers</a></h2>
<p> When <tt>-c++</tt> command line option is used (and <tt>-nocxx</tt>
one is not), the header file generated by SWIG will also contain the
declarations of C++ wrapper functions and classes mirroring the
original API. All C++ wrappers are fully inline, i.e. don't need to be
compiled separately, and are always defined inside the namespace (or
nested namespaces) specified by <tt>-namespace</tt> command-line option
or the namespace with the same name as the SWIG module name if this
option is not specified.</p>
<p> C++ wrappers try to provide a similar API to the original C++ API
being wrapped, notably any class <tt>Foo</tt> in the original API
appears as a class with the same name in the wrappers namespace, and
has the same, or similar, public methods. A class <tt>Bar</tt> deriving
from <tt>Foo</tt> also derives from it in the wrappers and so on. There
are some differences with the original API, however. Some of them are
due to fundamental limitations of the approach used, e.g.:</p>
<ul>
<li>Only template instantiations are present in the wrappers, not the
templates themselves.</li>
</ul>
<p> Other ones are due to things that could be supported but haven't
been implemented yet:</p>
<ul>
<li>Only single, not multiple, inheritance is currently supported.</li>
<li>Only enums using <tt>int</tt> (or smaller type) as underlying type
are supported.</li>
</ul>
<h3><a name="C_additional_possibilities">38.8.1 Additional customization
possibilities</a></h3>
<p> Generated C++ code can be customized by inserting custom code in the
following sections:</p>
<ul>
<li><tt>cxxheader</tt> for including additional headers and other
declarations in the global scope.</li>
<li><tt>cxxcode</tt> for additional code to appear after the
declarations of all wrapper classes, inside the module-specific
namespace.</li>
</ul>
<p> The following features are taken into account when generating C++
wrappers:</p>
<ul>
<li><tt>cxxignore</tt> May be set to skip generation of C++ wrappers for
the given function or class, while still generating C wrappers for
them.</li>
</ul>
<h3><a name="C_exception_handling">38.8.2 Exception handling</a></h3>
<p> Exception handling in C++ is more natural, as the exceptions are
re-thrown when using C++ wrappers and so can be caught, as objects of
the special <tt>SWIG_CException</tt> type, using the usual <tt>
try/catch</tt> statement. The objects of <tt>SWIG_CException</tt> class
have <tt>code()</tt> and <tt>msg()</tt> methods, with the latter
returning the error message associated with the exception.</p>
<p> If necessary, a custom exception type may be used instead of <tt>
SWIG_CException</tt>. To do this, a custom implementation of <tt>
swig_check()</tt> function, called to check for the pending exception
and throw the corresponding C++ exception if necessary, must be
provided and <tt>SWIG_swig_check_DEFINED</tt> preprocessor symbol must
be defined to prevent the default implementation of this function from
being compiled:</p>
<div class="code">
<pre>
%insert(cxxheader) %{
#ifndef SWIG_swig_check_DEFINED
#define SWIG_swig_check_DEFINED 1
#include &lt;stdexcept&gt;
class Exception : public std::runtime_error {
public:
explicit Exception(const char* msg) : std::runtime_error{msg} {}
};
inline void swig_check() {
if (auto* swig_ex = SWIG_CException_get_pending()) {
Exception const e{SWIG_CException_msg_get(swig_ex)};
SWIG_CException_reset_pending();
throw e;
}
}
template &lt;typename T&gt; T swig_check(T x) {
swig_check();
return x;
}
#endif // SWIG_swig_check_DEFINED
%}
</pre>
</div><HR NOSHADE>
<h1><a name="Mzscheme">39 SWIG and MzScheme/Racket</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#MzScheme_nn2">Creating native structures</a></li>
<li><a href="#MzScheme_simple">Simple example</a></li>
<li><a href="#MzScheme_external_docs">External documentation</a></li>
</ul>
</div>
<!-- INDEX -->
<p> This section contains information on SWIG's support of Racket,
formally known as MzScheme. SWIG works with Racket versions &lt; 8 (Racket
8 switched to be based on a different scheme interpreter and SWIG
hasn't been updated for this change).</p>
<h2><a name="MzScheme_nn2">39.1 Creating native structures</a></h2>
<p> Example interface file:</p>
<div class="code">
<pre>
/* define a macro for the struct creation */
%define handle_ptr(TYPE, NAME)
%typemap(argout) TYPE *NAME{
Scheme_Object *o = SWIG_NewStructFromPtr($1, $*1_mangle);
SWIG_APPEND_VALUE(o);
}
%typemap(in, numinputs=0) TYPE *NAME (TYPE temp) {
$1 = &amp;temp;
}
%enddef
/* setup the typemaps for the pointer to an output parameter cntrs */
handle_ptr(struct diag_cntrs, cntrs);
</pre>
</div>
<p> Then in scheme, you can use regular struct access procedures like</p>
<div class="code">
<pre>
; suppose a function created a struct foo as
; (define foo (make-diag-cntrs (#x1 #x2 #x3) (make-inspector))
; Then you can do
(format &quot;0x~x&quot; (diag-cntrs-field1 foo))
(format &quot;0x~x&quot; (diag-cntrs-field2 foo))
;etc...
</pre>
</div>
<h2><a name="MzScheme_simple">39.2 Simple example</a></h2>
<p> A few examples are available in the Examples/mzscheme directory. The
code and log of a session using SWIG below should help getting started.</p>
<p> C header file:</p>
<div class="code">
<pre>
// example.h
int fact(int n);
</pre>
</div>
<p> C source code:</p>
<div class="code">
<pre>
// File: example.c
#include &quot;example.h&quot;
int fact(int n) {
if (n &lt; 0) { /* This should probably return an error, but this is simpler */
return 0;
}
if (n == 0) {
return 1;
}
else {
/* testing for overflow would be a good idea here */
return n * fact(n-1);
}
}
</pre>
</div>
<p> SWIG interface file:</p>
<div class="code">
<pre>
/* File: example.i */
%module example
%{
#include &quot;example.h&quot;
%}
int fact(int n);
</pre>
</div>
<p> The session below using the above files is on an OS X machine, but
the points to be made are more general. On OS X, libtool is the tool
which creates libraries, which are named .dylib, rather than .so on
other unixes, or .dll on Windows.</p>
<div class="shell">
<pre>
% swig -mzscheme -declaremodule example.i
% gcc -c -m32 -o example.o example.c # force 32-bit object file (mzscheme is 32-bit only)
% libtool -dynamic -o libexample.dylib example.o # make it into a library
% ls # what've we got so far?
example.c example.o
example.h example_wrap.c
example.i libexample.dylib*
% mzc --cgc --cc example_wrap.c # compile the wrapping code
% LDFLAGS=&quot;-L. -lexample&quot; mzc --ld example_wrap.dylib example_wrap.o # ...and link it
% mzscheme -e '(path-&gt;string (build-path &quot;compiled&quot; &quot;native&quot; (system-library-subpath)))'
&quot;compiled/native/i386-macosx/3m&quot;
% mkdir -p compiled/native/i386-macosx/3m # move the extension library to a magic place
% mv example_wrap.dylib compiled/native/i386-macosx/3m/example_ss.dylib
% mzscheme
Welcome to MzScheme v4.2.4 [3m], Copyright (c) 2004-2010 PLT Scheme Inc.
&gt; (require &quot;example.ss&quot;)
&gt; (fact 5)
120
&gt; ^D
% echo 'It works!'
</pre>
</div>
<p> Some points of interest:</p>
<ul>
<li> This is on a 64-bit machine, so we have to include the -m32 option
when building the object file</li>
<li> If you want to declare a scheme module (and you probably do), it's
important that you include the -declaremodule option to swig (if you
miss this out, it'll appear to work, but fail later).</li>
<li> Use mzc to compile and then link the wrapped code. You'll probably
need to adjust the link flags to refer to the library you're wrapping
(you can either do this with an LDFLAGS declaration, as here, or with
multiple ++ldf options to mzc).</li>
<li> Create the directory with path (build-path &quot;compiled&quot; &quot;native&quot;
(system-library-subpath)) and move the freshly-generated .dylib to
there, changing its name to module-name_ss.dylib. After that, you can
REQUIRE the new module with (require &quot;module-name.ss&quot;).</li>
<li> The above requests mzc to create an extension using the CGC
garbage-collector. The alternative -- the 3m collector -- has generally
better performance, but work is still required for SWIG to emit code
which is compatible with it.</li>
</ul>
<h2><a name="MzScheme_external_docs">39.3 External documentation</a></h2>
<p> See the <a href="https://docs.racket-lang.org/inside/index.html">C
API</a> for more description of using the mechanism for adding
extensions. The main documentation is <a href="https://docs.racket-lang.org/">
here</a>.</p>
<p> Tip: mzc's --vv option is very useful for debugging the inevitable
library problems you'll encounter.</p>
<HR NOSHADE>
<h1><a name="Ocaml">40 SWIG and OCaml</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Ocaml_nn2">Preliminaries</a>
<ul>
<li><a href="#Ocaml_nn3">Running SWIG</a></li>
<li><a href="#Ocaml_nn4">Compiling the code</a></li>
<li><a href="#Ocaml_nn5">The camlp4 module</a></li>
<li><a href="#Ocaml_nn6">Using your module</a></li>
<li><a href="#Ocaml_nn7">Compilation problems and compiling with C++</a></li>
</ul>
</li>
<li><a href="#Ocaml_nn8">The low-level Ocaml/C interface</a>
<ul>
<li><a href="#Ocaml_nn9">The generated module</a></li>
<li><a href="#Ocaml_nn10">Enums</a>
<ul>
<li><a href="#Ocaml_nn11">Enum typing in Ocaml</a></li>
</ul>
</li>
<li><a href="#Ocaml_nn12">Arrays</a>
<ul>
<li><a href="#Ocaml_nn13">Simple types of bounded arrays</a></li>
<li><a href="#Ocaml_nn14">Complex and unbounded arrays</a></li>
<li><a href="#Ocaml_nn15">Using an object</a></li>
<li><a href="#Ocaml_nn16">Example typemap for a function taking float *
and int</a></li>
</ul>
</li>
<li><a href="#Ocaml_nn17">C++ Classes</a>
<ul>
<li><a href="#Ocaml_nn18">STL vector and string Example</a></li>
<li><a href="#Ocaml_nn19">C++ Class Example</a></li>
<li><a href="#Ocaml_nn20">Compiling the example</a></li>
<li><a href="#Ocaml_nn21">Sample Session</a></li>
</ul>
</li>
<li><a href="#Ocaml_nn22">Director Classes</a>
<ul>
<li><a href="#Ocaml_nn23">Director Introduction</a></li>
<li><a href="#Ocaml_nn24">Overriding Methods in Ocaml</a></li>
<li><a href="#Ocaml_nn25">Director Usage Example</a></li>
<li><a href="#Ocaml_nn26">Creating director objects</a></li>
<li><a href="#Ocaml_nn27">Typemaps for directors, directorin,
directorout, directorargout</a></li>
<li><a href="#Ocaml_nn28">directorin typemap</a></li>
<li><a href="#Ocaml_nn29">directorout typemap</a></li>
<li><a href="#Ocaml_nn30">directorargout typemap</a></li>
</ul>
</li>
<li><a href="#Ocaml_nn31">Exceptions</a></li>
</ul>
</li>
<li><a href="#Ocaml_nn32">Documentation Features</a>
<ul>
<li><a href="#Ocaml_nn33">Module docstring</a></li>
</ul>
</li>
</ul>
</div>
<!-- INDEX -->
<p> This chapter describes SWIG's support of Ocaml.</p>
<p> Ocaml was the second compiled, typed language to be added to SWIG.
Ocaml has a sophisticated type system, compile-time checking which
eliminates several classes of common programming errors, and good
native performance. While all of this is wonderful, there are
well-written C and C++ libraries that Ocaml users will want to take
advantage of as part of their arsenal (such as SSL and gdbm), as well
as their own mature C and C++ code. SWIG allows this code to be used in
a natural, type-safe way with Ocaml, by providing the necessary, but
repetitive glue code which creates and uses Ocaml values to communicate
with C and C++ code. In addition, SWIG also produces the needed Ocaml
source that binds variants, functions, classes, etc.</p>
<p> If you're not familiar with the Objective Caml language, you can
visit <a href="https://ocaml.org/">The Ocaml Website</a>.</p>
<h2><a name="Ocaml_nn2">40.1 Preliminaries</a></h2>
<p> SWIG is known to be compatible with OCaml 4.13.1 and above - older
versions are not regularly tested. Given the choice, you should use the
latest stable release. The SWIG Ocaml module has been tested on Linux
(x86, PPC, Sparc) and Cygwin on Windows. The best way to determine
whether your system will work is to compile the examples and test-suite
which come with SWIG. You can do this by running <tt>make check</tt>
from the SWIG root directory after installing SWIG. The Ocaml module
has been tested using the system's dynamic linking (the usual -lxxx
against libxxx.so, as well as with Gerd Stolpmann's <a href="http://download.camlcity.org/download/">
Dl package</a>. The ocaml_dynamic and ocaml_dynamic_cpp targets in the
file Examples/Makefile illustrate how to compile and link SWIG modules
that will be loaded dynamically. This has only been tested on Linux so
far.</p>
<h3><a name="Ocaml_nn3">40.1.1 Running SWIG</a></h3>
<p> The basics of getting a SWIG Ocaml module up and running can be seen
from one of SWIG's example Makefiles, but is also described here. To
build an Ocaml module, run SWIG using the <tt>-ocaml</tt> option.</p>
<div class="code">
<pre>
%swig -ocaml example.i
</pre>
</div>
<p>This will produce 3 files. The file <tt>example_wrap.c</tt> contains
all of the C code needed to build an Ocaml module. To build the module,
you will compile the file <tt>example_wrap.c</tt> with <tt>ocamlc</tt>
or <tt>ocamlopt</tt> to create the needed .o file. You will need to
compile the resulting .ml and .mli files as well, and do the final link
with -custom (not needed for native link).</p>
<h3><a name="Ocaml_nn4">40.1.2 Compiling the code</a></h3>
<p> The OCaml SWIG module now requires you to compile a module (<tt>Swig</tt>
) separately. In addition to aggregating common SWIG functionality, the
Swig module contains the data structure that represents C/C++ values.
This allows easier data sharing between modules if two or more are
combined, because the type of each SWIG'ed module's c_obj is derived
from Swig.c_obj_t. This also allows SWIG to acquire new conversions
painlessly, as well as giving the user more freedom with respect to
custom typing. Use <tt>ocamlc</tt> or <tt>ocamlopt</tt> to compile your
SWIG interface like:</p>
<div class="code">
<pre>
% swig -ocaml -co swig.mli ; swig -ocaml -co swig.ml
% ocamlc -c swig.mli ; ocamlc -c swig.ml
% ocamlc -c -ccopt &quot;-I/usr/include/foo&quot; example_wrap.c
% ocamlc -c example.mli
% ocamlc -c example.ml
</pre>
</div>
<p><tt>ocamlc</tt> is aware of .c files and knows how to handle them.
Unfortunately, it does not know about .cxx, .cc, or .cpp files, so when
SWIG is invoked in C++ mode, you must:</p>
<div class="code">
<pre>
% cp example_wrap.cxx example_wrap.cxx.c
% ocamlc -c ... -ccopt -xc++ example_wrap.cxx.c
% ...
</pre>
</div>
<h3><a name="Ocaml_nn5">40.1.3 The camlp4 module</a></h3>
<p> The camlp4 module (swigp4.ml -&gt; swigp4.cmo) contains a simple
rewriter which makes C++ code blend more seamlessly with objective caml
code. Its use is optional, but encouraged. The source file is included
in the Lib/ocaml directory of the SWIG source distribution. You can
checkout this file with <tt>&quot;swig -ocaml -co swigp4.ml&quot;</tt>. You
should compile the file with <tt>&quot;ocamlc -I `camlp4 -where` -pp
'camlp4o pa_extend.cmo q_MLast.cmo' -c swigp4.ml&quot;</tt></p>
<p> The basic principle of the module is to recognize certain non-caml
expressions and convert them for use with C++ code as interfaced by
SWIG. The camlp4 module is written to work with generated SWIG
interfaces, and probably isn't great to use with anything else.</p>
<p> Here are the main rewriting rules:</p>
<table border="1" summary="Rewriting rules">
<tr><th>Input</th><th>Rewritten to</th></tr>
<tr><td>f'( ... ) as in
<br> atoi'(&quot;0&quot;) or
<br> _exit'(0)</td><td>f(C_list [ ... ]) as in
<br> atoi (C_list [ C_string &quot;0&quot; ]) or
<br> _exit (C_list [ C_int 0 ])</td></tr>
<tr><td>object -&gt; method ( ... )</td><td>(invoke object) &quot;method&quot;
(C_list [ ... ])</td></tr>
<tr><td> object<i> 'binop</i> argument as in
<br> a '+= b</td><td> (invoke object) &quot;+=&quot; argument as in
<br> (invoke a) &quot;+=&quot; b</td><td></td></tr>
<tr><th colspan="2">Note that because camlp4 always recognizes &lt;&lt; and
&gt;&gt;, they are replaced by lsl and lsr in operator names.
<tr><td><i> 'unop</i> object as in
<br> '! a</td><td> (invoke a) &quot;!&quot; C_void</td></tr>
<tr><td><b> Smart pointer access like this</b>
<br> object '-&gt; method ( args )
<br></td><td> (invoke (invoke object &quot;-&gt;&quot; C_void))</td></tr>
<tr><td><b> Invoke syntax</b>
<br> object . '( ... )</td><td> (invoke object) &quot;()&quot; (C_list [ ... ])</td>
</tr>
<tr><td><b> Array syntax</b>
<br> object '[ 10 ]</td><td> (invoke object) &quot;[]&quot; (C_int 10)</td></tr>
<tr><td><b> Assignment syntax</b>
<br> let a = '10 and b = '&quot;foo&quot; and c = '1.0 and d = 'true</td><td> let
a = C_int 10 and b = C_string &quot;foo&quot; and c = C_double 1.0 and d = C_bool
true</td></tr>
<tr><td><b> Cast syntax</b>
<br> let a = _atoi '(&quot;2&quot;) as int
<br> let b = (getenv &quot;PATH&quot;) to string
<br> This works for int, string, float, bool</td><td> let a = get_int
(_atoi (C_string &quot;2&quot;))
<br> let b = C_string (getenv &quot;PATH&quot;)</td></tr>
</th></tr>
</table>
<h3><a name="Ocaml_nn6">40.1.4 Using your module</a></h3>
<p> You can test-drive your module by building a toplevel ocaml
interpreter. Consult the ocaml manual for details.</p>
<p> When linking any ocaml bytecode with your module, use the -custom
option to build your functions into the primitive list. This option is
not needed when you build native code.</p>
<h3><a name="Ocaml_nn7">40.1.5 Compilation problems and compiling with
C++</a></h3>
<p> Ocaml's C extension API unfortunately defines a <tt>value</tt>
typedef - this overly-generic name can collide with the C or C++ API
being wrapped if it defines a variable, function, member or type also
called <tt>value</tt>. Prior to SWIG 4.2.1, SWIG attempted to work
around this but the hack used itself caused problems so we've removed
it. The problem is not one SWIG created, nor one SWIG can really solve.</p>
<p> As mentioned earlier, C++ files need special handling to be compiled
with <tt>ocamlc</tt>. Other than that, C code that uses <tt>class</tt>
as a non-keyword, and C code that is too liberal with pointer types may
not compile under the C++ compiler. Most code meant to be compiled as
C++ will not have problems.</p>
<h2><a name="Ocaml_nn8">40.2 The low-level Ocaml/C interface</a></h2>
<p> In order to provide access to overloaded functions, and provide
sensible outputs from them, all C entities are represented as members
of the c_obj type:</p>
<p> In the code as seen by the typemap writer, there is a value,
swig_result, that always contains the current return data. It is a
list, and must be appended with the caml_list_append function, or with
functions and macros provided by objective caml.</p>
<div class="code">
<pre>
type c_obj =
C_void
| C_bool of bool
| C_char of char
| C_uchar of char
| C_short of int
| C_ushort of int
| C_int of int
| C_uint of int32
| C_int32 of int32
| C_int64 of int64
| C_float of float
| C_double of float
| C_ptr of int64 * int64
| C_array of c_obj array
| C_list of c_obj list
| C_obj of (string -&gt; c_obj -&gt; c_obj)
| C_string of string
| C_enum of c_enum_t
</pre>
</div>
<p> A few functions exist which generate and return these:</p>
<ul>
<li>caml_ptr_val receives a c_obj and returns a void *. This should be
used for all pointer purposes.</li>
<li>caml_long_val receives a c_obj and returns a long. This should be
used for most integral purposes.</li>
<li>caml_val_ptr receives a void * and returns a c_obj.</li>
<li>caml_val_bool receives a C int and returns a c_obj representing its
bool value.</li>
<li>caml_val_(u)?(char|short|int|long|float|double) receives an
appropriate C value and returns a c_obj representing it.</li>
<li>caml_val_string receives a char * and returns a string value.</li>
<li>caml_val_string_len receives a char * and a length and returns a
string value.</li>
<li>caml_val_obj receives a void * and an object type and returns a
C_obj, which contains a closure giving method access.</li>
</ul>
<p> Because of this style, a typemap can return any kind of value it
wants from a function. This enables out typemaps and inout typemaps to
work well. The one thing to remember about outputting values is that
you must append them to the return list with swig_result =
caml_list_append(swig_result, v).</p>
<p> This function will return a new list that has your element appended.
Upon return to caml space, the fnhelper function beautifies the result.
A list containing a single item degrades to only that item (i.e. [
C_int 3 ] -&gt; C_int 3), and a list containing more than one item is
wrapped in C_list (i.e. [ C_char 'a' ; C_char 'b' ] -&gt; C_list [ C_char
'a' ; C_char 'b' ]). This is in order to make return values easier to
handle when functions have only one return value, such as constructors,
and operators. In addition, string, pointer, and object values are
interchangeable with respect to caml_ptr_val, so you can allocate
memory as caml strings and still use the resulting pointers for C
purposes, even using them to construct simple objects on. Note, though,
that foreign C++ code does not respect the garbage collector, although
the SWIG interface does.</p>
<p> The wild card type that you can use in lots of different ways is
C_obj. It allows you to wrap any type of thing you like as an object
using the same mechanism that the ocaml module does. When evaluated in
caml_ptr_val, the returned value is the result of a call to the
object's &quot;&amp;&quot; operator, taken as a pointer.</p>
<p> You should only construct values using objective caml, or using the
functions caml_val_* functions provided as static functions to a SWIG
ocaml module, as well as the caml_list_* functions. These functions
provide everything a typemap needs to produce values. In addition,
value items pass through directly, but you must make your own type
signature for a function that uses value in this way.</p>
<h3><a name="Ocaml_nn9">40.2.1 The generated module</a></h3>
<p> The SWIG <tt>%module</tt> directive specifies the name of the Ocaml
module to be generated. If you specified `<tt>%module example</tt>',
then your Ocaml code will be accessible in the module Example. The
module name is always capitalized as is the ocaml convention. Note that
you must not use any Ocaml keyword to name your module. Remember that
the keywords are not the same as the C++ ones.</p>
<p> You can introduce extra code into the output wherever you like with
SWIG. These are the places you can introduce code:</p>
<table border="1" summary="Extra code sections">
<tr><td>&quot;header&quot;</td><td>This code is inserted near the beginning of the
C wrapper file, before any function definitions.</td></tr>
<tr><td>&quot;wrapper&quot;</td><td>This code is inserted in the function
definition section.</td></tr>
<tr><td>&quot;runtime&quot;</td><td>This code is inserted near the end of the C
wrapper file.</td></tr>
<tr><td>&quot;mli&quot;</td><td>This code is inserted into the caml interface
file. Special signatures should be inserted here.</td></tr>
<tr><td>&quot;ml&quot;</td><td>This code is inserted in the caml code defining the
interface to your C code. Special caml code, as well as any
initialization which should run when the module is loaded may be
inserted here.</td></tr>
<tr><td>&quot;classtemplate&quot;</td><td>The &quot;classtemplate&quot; place is special
because it describes the output SWIG will generate for class
definitions.</td></tr>
</table>
<h3><a name="Ocaml_nn10">40.2.2 Enums</a></h3>
<p> SWIG will wrap enumerations as polymorphic variants in the output
Ocaml code, as above in C_enum. In order to support all C++-style uses
of enums, the function int_to_enum and enum_to_int are provided for
ocaml code to produce and consume these values as integers. Other than
that, correct uses of enums will not have a problem. Since enum labels
may overlap between enums, the enum_to_int and int_to_enum functions
take an enum type label as an argument. Example:</p>
<div class="code">
<pre>
%module enum_test
%{
enum c_enum_type { a = 1, b, c = 4, d = 8 };
%}
enum c_enum_type { a = 1, b, c = 4, d = 8 };
</pre>
</div>
<p> The output mli contains:</p>
<div class="code">
<pre>
type c_enum_type = [
`unknown
| `c_enum_type
]
type c_enum_tag = [
`int of int
| `a
| `b
| `c
| `d
]
val int_to_enum c_enum_type -&gt; int -&gt; c_obj
val enum_to_int c_enum_type -&gt; c_obj -&gt; c_obj
</pre>
</div>
<p> So it's possible to do this:</p>
<div class="code">
<pre>
bash-2.05a$ ocamlmktop -custom enum_test_wrap.o enum_test.cmo -o enum_test_top
bash-2.05a$ ./enum_test_top
Objective Caml version 3.04
# open Enum_test ;;
# let x = C_enum `a ;;
val x : Enum_test.c_obj = C_enum `a
# enum_to_int `c_enum_type x ;;
- : Enum_test.c_obj = C_int 1
# int_to_enum `c_enum_type 4 ;;
- : Enum_test.c_obj = C_enum `c
</pre>
</div>
<h4><a name="Ocaml_nn11">40.2.2.1 Enum typing in Ocaml</a></h4>
<p> The ocaml SWIG module now has support for loading and using multiple
SWIG modules at the same time. This enhances modularity, but presents
problems when used with a language which assumes that each module's
types are complete at compile time. In order to achieve total soundness
enum types are now isolated per-module. The type issue matters when
values are shared between functions imported from different modules.
You must convert values to master values using the swig_val function
before sharing them with another module.</p>
<h3><a name="Ocaml_nn12">40.2.3 Arrays</a></h3>
<h4><a name="Ocaml_nn13">40.2.3.1 Simple types of bounded arrays</a></h4>
<p> SWIG has support for array types, but you generally will need to
provide a typemap to handle them. You can currently roll your own, or
expand some of the macros provided (but not included by default) with
the SWIG distribution.</p>
<p> By including &quot;carray.i&quot;, you will get access to some macros that
help you create typemaps for array types fairly easily.</p>
<p> <tt>%make_simple_array_typemap</tt> is the easiest way to get access
to arrays of simple types with known bounds in your code, but this only
works for arrays whose bounds are completely specified.</p>
<h4><a name="Ocaml_nn14">40.2.3.2 Complex and unbounded arrays</a></h4>
<p> Unfortunately, unbounded arrays and pointers can't be handled in a
completely general way by SWIG, because the end-condition of such an
array can't be predicted. In some cases, it will be by consent (e.g. an
array of four or more chars), sometimes by explicit length (char
*buffer, int len), and sometimes by sentinel value (0, -1, etc.). SWIG
can't predict which of these methods will be used in the array, so you
have to specify it for yourself in the form of a typemap.</p>
<h4><a name="Ocaml_nn15">40.2.3.3 Using an object</a></h4>
<p> It's possible to use C++ to your advantage by creating a simple
object that provides access to your array. This may be more desirable
in some cases, since the object can provide bounds checking, etc., that
prevents crashes.</p>
<p> Consider writing an object when the ending condition of your array
is complex, such as using a required sentinel, etc.</p>
<h4><a name="Ocaml_nn16">40.2.3.4 Example typemap for a function taking
float * and int</a></h4>
<p> This is a simple example <tt>in</tt> typemap for an array of float,
where the length of the array is specified as an extra parameter. Other
such typemaps will work similarly. In the example, the function
printfloats is called with a float array, and specified length. The
actual length reported in the len argument is the length of the array
passed from ocaml, making passing an array into this type of function
convenient.</p>
<table bgcolor="#dddddd" border="1" summary="float * and int typemap example">
<tr><th>
<center>tarray.i</center>
</th></tr>
<tr><td>
<pre>
%module tarray
%{
#include &lt;stdio.h&gt;
void printfloats( float *tab, int len ) {
int i;
for( i = 0; i &lt; len; i++ ) {
printf( &quot;%f &quot;, tab[i] );
}
printf( &quot;\n&quot; );
}
%}
%typemap(in) (float *tab, int len) {
int i;
/* $*1_type */
$2 = caml_array_len($input);
$1 = ($*1_type *)malloc( $2 * sizeof( float ) );
for( i = 0; i &lt; $2; i++ ) {
$1[i] = caml_double_val(caml_array_nth($input, i));
}
}
void printfloats( float *tab, int len );
</pre>
</td></tr>
<tr><th>Sample Run</th></tr>
<tr><td>
<pre>
# open Tarray ;;
# _printfloats (C_array [| C_double 1.0 ; C_double 3.0 ; C_double 5.6666 |]) ;;
1.000000 3.000000 5.666600
- : Tarray.c_obj = C_void
</pre>
</td></tr>
</table>
<h3><a name="Ocaml_nn17">40.2.4 C++ Classes</a></h3>
<p> C++ classes, along with structs and unions are represented by C_obj
(string -&gt; c_obj -&gt; c_obj) wrapped closures. These objects contain a
method list, and a type, which allow them to be used like C++ objects.
When passed into typemaps that use pointers, they degrade to pointers
through their &quot;&amp;&quot; method. Every method an object has is represented as
a string in the object's method table, and each method table exists in
memory only once. In addition to any other operators an object might
have, certain builtin ones are provided by SWIG: (all of these take no
arguments (C_void))</p>
<table summary="SWIG provided operators">
<tr><td>&quot;~&quot;</td><td>Delete this object</td></tr>
<tr><td>&quot;&amp;&quot;</td><td>Return an ordinary C_ptr value representing this
object's address</td></tr>
<tr><td>&quot;sizeof&quot;</td><td>If enabled with (&quot;sizeof&quot;=&quot;1&quot;) on the module
node, return the object's size in char.</td></tr>
<tr><td>&quot;:methods&quot;</td><td>Returns a list of strings containing the
names of the methods this object contains</td></tr>
<tr><td>&quot;:classof&quot;</td><td>Returns the name of the class this object
belongs to.</td></tr>
<tr><td>&quot;:parents&quot;</td><td>Returns a list of all direct parent classes
which have been wrapped by SWIG.</td></tr>
<tr><td>&quot;::[parent-class]&quot;</td><td>Returns a view of the object as the
indicated parent class. This is mainly used internally by the SWIG
module, but may be useful to client programs.</td></tr>
<tr><td>&quot;[member-variable]&quot;</td><td>Each member variable is wrapped as a
method with an optional parameter. Called with one argument, the member
variable is set to the value of the argument. With zero arguments, the
value is returned.</td></tr>
</table>
<p> Note that this string belongs to the wrapper object, and not the
underlying pointer, so using create_[x]_from_ptr alters the returned
value for the same object.</p>
<h4><a name="Ocaml_nn18">40.2.4.1 STL vector and string Example</a></h4>
<p> Standard typemaps are now provided for STL vector and string. More
are in the works. STL strings are passed just like normal strings, and
returned as strings. STL string references don't mutate the original
string, (which might be surprising), because Ocaml strings are mutable
but have fixed length. Instead, use multiple returns, as in the
argout_ref example.</p>
<table bgcolor="#dddddd" border="1" summary="STL vector and string example">
<tr><th>
<center>example.i</center>
</th></tr>
<tr><td>
<pre>
%module example
%{
#include &quot;example.h&quot;
%}
%include &lt;stl.i&gt;
namespace std {
%template(StringVector) std::vector &lt; string &gt;;
};
%include &quot;example.h&quot;
</pre>
</td></tr>
<tr><td><font size="-1"><i>This example is in Examples/ocaml/stl</i></font>
</td></tr>
</table>
<p> Since there's a makefile in that directory, the example is easy to
build.</p>
<p> Here's a sample transcript of an interactive session using a string
vector after making a toplevel (make toplevel). This example uses the
camlp4 module.</p>
<div class="code">
<pre>
bash-2.05a$ ./runme_top
Objective Caml version 3.06
Camlp4 Parsing version 3.06
# open Swig ;;
# open Example ;;
# let x = new_StringVector '() ;;
val x : Example.c_obj = C_obj &lt;fun&gt;
# x -&gt; &quot;:methods&quot; () ;;
- : Example.c_obj =
C_list
[C_string &quot;nop&quot;; C_string &quot;size&quot;; C_string &quot;empty&quot;; C_string &quot;clear&quot;;
C_string &quot;push_back&quot;; C_string &quot;[]&quot;; C_string &quot;=&quot;; C_string &quot;set&quot;;
C_string &quot;~&quot;; C_string &quot;&amp;&quot;; C_string &quot;:parents&quot;; C_string &quot;:classof&quot;;
C_string &quot;:methods&quot;]
# x -&gt; push_back (&quot;foo&quot;) ;;
- : Example.c_obj = C_void
# x -&gt; push_back (&quot;bar&quot;) ;;
- : Example.c_obj = C_void
# x -&gt; push_back (&quot;baz&quot;) ;;
- : Example.c_obj = C_void
# x '[1] ;;
- : Example.c_obj = C_string &quot;bar&quot;
# x -&gt; set (1, &quot;spam&quot;) ;;
- : Example.c_obj = C_void
# x '[1] ;;
- : Example.c_obj = C_string &quot;spam&quot;
# for i = 0 to (x -&gt; size() as int) - 1 do
print_endline ((x '[i to int]) as string)
done ;;
foo
bar
baz
- : unit = ()
#
</pre>
</div>
<h4><a name="Ocaml_nn19">40.2.4.2 C++ Class Example</a></h4>
<p> Here's a simple example using Trolltech's Qt Library:</p>
<table bgcolor="#dddddd" border="1" summary="Qt Library example">
<tr><th>
<center>qt.i</center>
</th></tr>
<tr><td>
<pre>
%module qt
%{
#include &lt;qapplication.h&gt;
#include &lt;qpushbutton.h&gt;
%}
class QApplication {
public:
QApplication( int argc, char **argv );
void exec();
};
class QPushButton {
public:
QPushButton( char *str, QWidget *w );
void resize( int x, int y );
void show();
};
</pre>
</td></tr>
</table>
<h4><a name="Ocaml_nn20">40.2.4.3 Compiling the example</a></h4>
<div class="code">
<pre>
$ QTPATH=/your/qt/path
$ for file in swig.mli swig.ml swigp4.ml ; do swig -ocaml -co $file ; done
$ ocamlc -c swig.mli ; ocamlc -c swig.ml
$ ocamlc -I `camlp4 -where` -pp &quot;camlp4o pa_extend.cmo q_MLast.cmo&quot; -c swigp4.ml
$ swig -ocaml -c++ -o qt_wrap.c qt.i
$ ocamlc -c -ccopt -xc++ -ccopt -g -g -ccopt -I$QTPATH/include qt_wrap.c
$ ocamlc -c qt.mli
$ ocamlc -c qt.ml
$ ocamlmktop -custom swig.cmo -I `camlp4 -where` \
camlp4o.cma swigp4.cmo qt_wrap.o qt.cmo -o qt_top -cclib \
-L$QTPATH/lib -cclib -lqt
</pre>
</div>
<h4><a name="Ocaml_nn21">40.2.4.4 Sample Session</a></h4>
<div class="code">
<pre>
bash-2.05a$ ./qt_top
Objective Caml version 3.06
Camlp4 Parsing version 3.06
# open Swig ;;
# open Qt ;;
# let a = new_QApplication '(0, 0) ;;
val a : Qt.c_obj = C_obj &lt;fun&gt;
# let hello = new_QPushButton '(&quot;hi&quot;, 0) ;;
val hello : Qt.c_obj = C_obj &lt;fun&gt;
# hello -&gt; resize (100, 30) ;;
- : Qt.c_obj = C_void
# hello -&gt; show () ;;
- : Qt.c_obj = C_void
# a -&gt; exec () ;;
</pre>
</div>
<p> Assuming you have a working installation of QT, you will see a
window containing the string &quot;hi&quot; in a button.</p>
<h3><a name="Ocaml_nn22">40.2.5 Director Classes</a></h3>
<h4><a name="Ocaml_nn23">40.2.5.1 Director Introduction</a></h4>
<p> Director classes are classes which allow Ocaml code to override the
public methods of a C++ object. This facility allows the user to use
C++ libraries that require a derived class to provide application
specific functionality in the context of an application or utility
framework.</p>
<p> You can turn on director classes by using an optional module
argument like this:</p>
<div class="code">
<pre>
%module(directors=&quot;1&quot;)
...
// Turn on the director class for a specific class like this:
%feature(&quot;director&quot;)
class foo {
...
};
</pre>
</div>
<h4><a name="Ocaml_nn24">40.2.5.2 Overriding Methods in Ocaml</a></h4>
<p> Because the Ocaml language module treats C++ method calls as calls
to a certain function, all you need to do is to define the function
that will handle the method calls in terms of the public methods of the
object, and any other relevant information. The function <tt>
new_derived_object</tt> uses a stub class to call your methods in place
of the ones provided by the underlying implementation. The object you
receive is the underlying object, so you are free to call any methods
you want from within your derived method. Note that calls to the
underlying object do not invoke Ocaml code. You need to handle that
yourself.</p>
<p> <tt>new_derived_object</tt> receives your function, the function
that creates the underlying object, and any constructor arguments, and
provides an object that you can use in any usual way. When C++ code
calls one of the object's methods, the object invokes the Ocaml
function as if it had been invoked from Ocaml, allowing any method
definitions to override the C++ ones.</p>
<p> In this example, I'll examine the objective caml code involved in
providing an overloaded class. This example is contained in
Examples/ocaml/shapes.</p>
<h4><a name="Ocaml_nn25">40.2.5.3 Director Usage Example</a></h4>
<table bgcolor="#dddddd" border="1" summary="Director usage example">
<tr><th>
<center>runme.ml</center>
</th></tr>
<tr><td>
<pre>
open Swig
open Example
...
let triangle_class pts ob meth args =
match meth with
&quot;cover&quot; -&gt;
(match args with
C_list [ x_arg ; y_arg ] -&gt;
let xa = x_arg as float
and ya = y_arg as float in
(point_in_triangle pts xa ya) to bool
| _ -&gt; raise (Failure &quot;cover needs two double arguments.&quot;))
| _ -&gt; (invoke ob) meth args ;;
...
let triangle =
new_derived_object
new_shape
(triangle_class ((0.0, 0.0), (0.5, 1.0), (1.0, 0.6)))
'() ;;
let _ = _draw_shape_coverage '(triangle, 60, 20) ;;
</pre>
</td></tr>
</table>
<p> This is the meat of what you need to do. The actual &quot;class&quot;
definition containing the overloaded method is defined in the function
triangle_class. This is a lot like the class definitions emitted by
SWIG, if you look at example.ml, which is generated when SWIG consumes
example.i. Basically, you are given the arguments as a c_obj and the
method name as a string, and you must intercept the method you are
interested in and provide whatever return value you need. Bear in mind
that the underlying C++ code needs the right return type, or an
exception will be thrown. This exception will generally be Failure, or
NotObject. You must call other ocaml methods that you rely on yourself.
Due to the way directors are implemented, method calls on your object
from with ocaml code will always invoke C++ methods even if they are
overridden in ocaml.</p>
<p> In the example, the draw_shape_coverage function plots the indicated
number of points as either covered (<tt>x</tt>) or uncovered ( )
between 0 and 1 on the X and Y axes. Your shape implementation can
provide any coverage map it likes, as long as it responds to the
&quot;cover&quot; method call with a boolean return (the underlying method
returns bool). This might allow a tricky shape implementation, such as
a boolean combination, to be expressed in a more effortless style in
ocaml, while leaving the &quot;engine&quot; part of the program in C++.</p>
<h4><a name="Ocaml_nn26">40.2.5.4 Creating director objects</a></h4>
<p> The definition of the actual object triangle can be described this
way:</p>
<div class="code">
<pre>
let triangle =
new_derived_object
new_shape
(triangle_class ((0.0, 0.0), (0.5, 1.0), (1.0, 0.0)))
'()
</pre>
</div>
<p> The first argument to <tt>new_derived_object</tt>, new_shape is the
method which returns a shape instance. This function will be invoked
with the third argument will be appended to the argument list [ C_void
]. In the example, the actual argument list is sent as (C_list [ C_void
; C_void ]). The augmented constructor for a director class needs the
first argument to determine whether it is being constructed as a
derived object, or as an object of the indicated type only (in this
case <tt>shape</tt>). The Second argument is a closure that will be
added to the final C_obj.</p>
<p> The actual object passed to the self parameter of the director
object will be a C_director_core, containing a c_obj option ref and a
c_obj. The c_obj provided is the same object that will be returned from
new_derived object, that is, the object exposing the overridden
methods. The other part is an option ref that will have its value
extracted before becoming the <tt>ob</tt> parameter of your class
closure. This ref will contain <tt>None</tt> if the C++ object
underlying is ever destroyed, and will consequently trigger an
exception when any method is called on the object after that point (the
actual raise is from an inner function used by new_derived_object, and
throws NotObject). This prevents a deleted C++ object from causing a
core dump, as long as the object is destroyed properly.</p>
<h4><a name="Ocaml_nn27">40.2.5.5 Typemaps for directors, directorin,
directorout, directorargout</a></h4>
<p> Special typemaps exist for use with directors, the <tt>directorin,
directorout, directorargout</tt> are used in place of <tt>in, out,
argout</tt> typemaps, except that their direction is reversed. They
provide for you to provide argout values, as well as a function return
value in the same way you provide function arguments, and to receive
arguments the same way you normally receive function returns.</p>
<h4><a name="Ocaml_nn28">40.2.5.6 directorin typemap</a></h4>
<p> The <tt>directorin</tt> typemap is used when you will receive
arguments from a call made by C++ code to you, therefore, values will
be translated from C++ to ocaml. You must provide some valid C_obj
value. This is the value your ocaml code receives when you are called.
In general, a simple <tt>directorin</tt> typemap can use the same body
as a simple <tt>out</tt> typemap.</p>
<h4><a name="Ocaml_nn29">40.2.5.7 directorout typemap</a></h4>
<p> The <tt>directorout</tt> typemap is used when you will send an
argument from your code back to the C++ caller. That is; directorout
specifies a function return conversion. You can usually use the same
body as an <tt>in</tt> typemap for the same type, except when there are
special requirements for object ownership, etc.</p>
<h4><a name="Ocaml_nn30">40.2.5.8 directorargout typemap</a></h4>
<p> C++ allows function arguments which are by pointer (*) and by
reference (&amp;) to receive a value from the called function, as well as
sending one there. Sometimes, this is the main purpose of the argument
given. <tt>directorargout</tt> typemaps allow your caml code to emulate
this by specifying additional return values to be put into the output
parameters. The SWIG ocaml module is a bit loose in order to make code
easier to write. In this case, your return to the caller must be a list
containing the normal function return first, followed by any argout
values in order. These argout values will be taken from the list and
assigned to the values to be returned to C++ through directorargout
typemaps. In the event that you don't specify all of the necessary
values, integral values will read zero, and struct or object returns
have undefined results.</p>
<h3><a name="Ocaml_nn31">40.2.6 Exceptions</a></h3>
<p> If an error occurs in a C or C++ function, you may want to convert
that error into an OCaml exception. To do this, you can use the <tt>
%exception</tt> directive. The <tt>%exception</tt> directive simply lets
you rewrite part of the generated wrapper code to include an error
check. It is detailed in full in the <a href="#Customization_exception">
Exception handling with %exception</a> section.</p>
<p> In C, a function often indicates an error by returning a status code
(e.g. a negative number or a NULL pointer). Here is a simple example of
how you might handle that:</p>
<div class="code">
<pre>
%exception malloc {
$action
if (result == NULL) {
caml_failwith(&quot;Not enough memory&quot;);
}
}
void *malloc(size_t nbytes);
</pre>
</div>
<p> In OCaml:</p>
<div class="code">
<pre>
# let a = _malloc (C_int 20000000000);;
Exception: Failure &quot;Not enough memory&quot;.
#
</pre>
</div>
<p> If a library provides some kind of general error handling framework,
you can also use that. For example:</p>
<div class="code">
<pre>
%exception {
$action
if (err_occurred()) {
caml_failwith(err_message());
}
}
</pre>
</div>
<p> If no declaration name is given to <tt>%exception</tt>, it is
applied to all wrapper functions. <tt>$action</tt> is a SWIG special
variable and is replaced by the C/C++ function call being wrapped.</p>
<p> C++ exceptions are also easy to handle. We can catch a C++ exception
and rethrow it as an OCaml exception like this:</p>
<div class="code">
<pre>
%exception getitem {
try {
$action
} catch (std::out_of_range &amp;e) {
caml_failwith(e.what());
}
}
class FooClass {
public:
int getitem(int index); // Exception handling added
...
};
</pre>
</div>
<p> The language-independent <tt>exception.i</tt> library file can also
be used to raise exceptions. See the <a href="#Library">SWIG Library</a>
chapter.</p>
<h2><a name="Ocaml_nn32">40.3 Documentation Features</a></h2>
<p> The features described in this section can be used to generate
documentation comments (colloquially referred to as &quot;docstrings&quot;) that
can be read by <a href="https://caml.inria.fr/pub/docs/manual-ocaml/ocamldoc.html">
OCamldoc</a>.</p>
<h3><a name="Ocaml_nn33">40.3.1 Module docstring</a></h3>
<p> The first documentation comment of an <tt>mli</tt> file is the
comment associated with the entire module. SWIG supports this by
setting an option of the <tt>%module</tt> directive. For example:</p>
<div class="code">
<pre>
%module(docstring=&quot;This is the example module's docstring&quot;) example
</pre>
</div>
<p> When you have more than just a line or so, you can retain the
readability of the <tt>%module</tt> directive by using a macro. For
example:</p>
<div class="code">
<pre>
%define DOCSTRING
&quot;The `XmlResource` class allows program resources defining menus,
controls on a panel, etc. to be loaded from an XML file.&quot;
%enddef
%module(docstring=DOCSTRING) xrc
</pre>
</div><HR NOSHADE>
<h1><a name="Extending">41 Extending SWIG to support new languages</a></h1>
<!-- INDEX -->
<div class="sectiontoc">
<ul>
<li><a href="#Extending_nn2">Introduction</a></li>
<li><a href="#Extending_nn3">Prerequisites</a></li>
<li><a href="#Extending_nn4">The Big Picture</a></li>
<li><a href="#Extending_nn5">Execution Model</a>
<ul>
<li><a href="#Extending_nn6">Preprocessing</a></li>
<li><a href="#Extending_nn7">Parsing</a></li>
<li><a href="#Extending_nn8">Parse Trees</a></li>
<li><a href="#Extending_nn9">Attribute namespaces</a></li>
<li><a href="#Extending_nn10">Symbol Tables</a></li>
<li><a href="#Extending_nn11">The %feature directive</a></li>
<li><a href="#Extending_nn12">Code Generation</a></li>
<li><a href="#Extending_nn13">SWIG and XML</a></li>
</ul>
</li>
<li><a href="#Extending_nn14">Primitive Data Structures</a>
<ul>
<li><a href="#Extending_nn15">Strings</a></li>
<li><a href="#Extending_nn16">Hashes</a></li>
<li><a href="#Extending_nn17">Lists</a></li>
<li><a href="#Extending_nn18">Common operations</a></li>
<li><a href="#Extending_nn19">Iterating over Lists and Hashes</a></li>
<li><a href="#Extending_nn20">I/O</a></li>
</ul>
</li>
<li><a href="#Extending_nn21">Navigating and manipulating parse trees</a>
</li>
<li><a href="#Extending_nn22">Working with attributes</a></li>
<li><a href="#Extending_nn23">Type system</a>
<ul>
<li><a href="#Extending_nn24">String encoding of types</a></li>
<li><a href="#Extending_nn25">Type construction</a></li>
<li><a href="#Extending_nn26">Type tests</a></li>
<li><a href="#Extending_nn27">Typedef and inheritance</a></li>
<li><a href="#Extending_nn28">Lvalues</a></li>
<li><a href="#Extending_nn29">Output functions</a></li>
</ul>
</li>
<li><a href="#Extending_nn30">Parameters</a></li>
<li><a href="#Extending_nn31">Writing a Language Module</a>
<ul>
<li><a href="#Extending_nn32">Execution model</a></li>
<li><a href="#Extending_starting_out">Starting out</a></li>
<li><a href="#Extending_nn34">Command line options</a></li>
<li><a href="#Extending_nn35">Configuration and preprocessing</a></li>
<li><a href="#Extending_nn36">Entry point to code generation</a></li>
<li><a href="#Extending_nn37">Module I/O and wrapper skeleton</a></li>
<li><a href="#Extending_nn38">Low-level code generators</a></li>
<li><a href="#Extending_configuration_files">Configuration files</a></li>
<li><a href="#Extending_nn40">Runtime support</a></li>
<li><a href="#Extending_nn41">Standard library files</a></li>
<li><a href="#Extending_nn42">User examples</a></li>
<li><a href="#Extending_test_suite">Test driven development and the
test-suite</a>
<ul>
<li><a href="#Extending_running_test_suite">Running the test-suite</a></li>
</ul>
</li>
<li><a href="#Extending_nn43">Documentation</a></li>
<li><a href="#Extending_coding_style_guidelines">Coding style guidelines</a>
</li>
<li><a href="#Extending_language_status">Target language status</a>
<ul>
<li><a href="#Extending_supported_status">Supported status</a></li>
<li><a href="#Extending_experimental_status">Experimental status</a></li>
<li><a href="#Extending_deprecated_status">Deprecated status</a></li>
</ul>
</li>
<li><a href="#Extending_prerequisites">Prerequisites for adding a new
language module to the SWIG distribution</a></li>
</ul>
</li>
<li><a href="#Extending_debugging_options">Debugging Options</a></li>
<li><a href="#Extending_nn46">Guide to parse tree nodes</a></li>
<li><a href="#Extending_further_info">Further Development Information</a>
</li>
</ul>
</div>
<!-- INDEX -->
<h2><a name="Extending_nn2">41.1 Introduction</a></h2>
<p> This chapter primarily describes SWIG's internal organization and
the process by which new target languages can be developed. It also
provides details for debugging and extending existing target languages
and functionality.</p>
<p> First, a brief word of warning---SWIG is continually evolving. The
information in this chapter is mostly up to date, but changes are
ongoing. Expect a few inconsistencies. Also, this chapter is not meant
to be a hand-holding tutorial. As a starting point, you should probably
look at one of SWIG's existing modules.</p>
<h2><a name="Extending_nn3">41.2 Prerequisites</a></h2>
<p> In order to extend SWIG, it is useful to have the following
background:</p>
<ul>
<li>An understanding of the C API for the target language.</li>
<li>A good grasp of the C++ type system.</li>
<li>An understanding of typemaps and some of SWIG's advanced features.</li>
<li>Some familiarity with writing C++ (language modules are currently
written in C++).</li>
</ul>
<p> Since SWIG is essentially a specialized C++ compiler, it may be
useful to have some prior experience with compiler design (perhaps even
a compilers course) to better understand certain parts of the system. A
number of books will also be useful. For example, &quot;The C Programming
Language&quot; by Kernighan and Ritchie (a.k.a, &quot;K&amp;R&quot;) and the C++ standard,
&quot;ISO/IEC 14882 Programming Languages - C++&quot; will be of great use.</p>
<p> Also, it is useful to keep in mind that SWIG primarily operates as
an extension of the C++<em> type</em> system. At first glance, this
might not be obvious, but almost all SWIG directives as well as the
low-level generation of wrapper code are driven by C++ datatypes.</p>
<h2><a name="Extending_nn4">41.3 The Big Picture</a></h2>
<p> SWIG is a special purpose compiler that parses C++ declarations to
generate wrapper code. To make this conversion possible, SWIG makes
three fundamental extensions to the C++ language:</p>
<ul>
<li><b>Typemaps</b>. Typemaps are used to define the
conversion/marshalling behavior of specific C++ datatypes. All type
conversion in SWIG is based on typemaps. Furthermore, the association
of typemaps to datatypes utilizes an advanced pattern matching
mechanism that is fully integrated with the C++ type system.</li>
<li><b>Declaration Annotation</b>. To customize wrapper code generation,
most declarations can be annotated with special features. For example,
you can make a variable read-only, you can ignore a declaration, you
can rename a member function, you can add exception handling, and so
forth. Virtually all of these customizations are built on top of a
low-level declaration annotator that can attach arbitrary attributes to
any declaration. Code generation modules can look for these attributes
to guide the wrapping process.</li>
<li><b>Class extension</b>. SWIG allows classes and structures to be
extended with new methods and attributes (the <tt>%extend</tt>
directive). This has the effect of altering the API in the target
language and can be used to generate OO interfaces to C libraries.</li>
</ul>
<p> It is important to emphasize that virtually all SWIG features reduce
to one of these three fundamental concepts. The type system and pattern
matching rules also play a critical role in making the system work. For
example, both typemaps and declaration annotation are based on pattern
matching and interact heavily with the underlying type system.</p>
<h2><a name="Extending_nn5">41.4 Execution Model</a></h2>
<p> When you run SWIG on an interface, processing is handled in stages
by a series of system components:</p>
<ul>
<li>An integrated C preprocessor reads a collection of configuration
files and the specified interface file into memory. The preprocessor
performs the usual functions including macro expansion and file
inclusion. However, the preprocessor also performs some transformations
of the interface. For instance, <tt>#define</tt> statements are
sometimes transformed into <tt>%constant</tt> declarations. In
addition, information related to file/line number tracking is inserted.</li>
<li>A C/C++ parser reads the preprocessed input and generates a full
parse tree of all of the SWIG directives and C declarations found. The
parser is responsible for many aspects of the system including
renaming, declaration annotation, and template expansion. However, the
parser does not produce any output nor does it interact with the target
language module as it runs. SWIG is not a one-pass compiler.</li>
<li>A type-checking pass is made. This adjusts all of the C++ typenames
to properly handle namespaces, typedefs, nested classes, and other
issues related to type scoping.</li>
<li>A semantic pass is made on the parse tree to collect information
related to properties of the C++ interface. For example, this pass
would determine whether or not a class allows a default constructor.</li>
<li>A code generation pass is made using a specific target language
module. This phase is responsible for generating the actual wrapper
code. All of SWIG's user-defined modules are invoked during this latter
stage of compilation.</li>
</ul>
<p> The next few sections briefly describe some of these stages.</p>
<h3><a name="Extending_nn6">41.4.1 Preprocessing</a></h3>
<p> The preprocessor plays a critical role in the SWIG implementation.
This is because a lot of SWIG's processing and internal configuration
is managed not by code written in C, but by configuration files in the
SWIG library. In fact, when you run SWIG, parsing starts with a small
interface file like this (note: this explains the cryptic error
messages that new users sometimes get when SWIG is misconfigured or
installed incorrectly):</p>
<div class="code">
<pre>
%include &quot;swig.swg&quot; // Global SWIG configuration
%include &quot;<em>langconfig.swg</em>&quot; // Language specific configuration
%include &quot;yourinterface.i&quot; // Your interface file
</pre>
</div>
<p> The <tt>swig.swg</tt> file contains global configuration
information. In addition, this file defines many of SWIG's standard
directives as macros. For instance, part of of <tt>swig.swg</tt> looks
like this:</p>
<div class="code">
<pre>
...
/* Code insertion directives such as %wrapper %{ ... %} */
#define %begin %insert(&quot;begin&quot;)
#define %runtime %insert(&quot;runtime&quot;)
#define %header %insert(&quot;header&quot;)
#define %wrapper %insert(&quot;wrapper&quot;)
#define %init %insert(&quot;init&quot;)
/* Access control directives */
#define %immutable %feature(&quot;immutable&quot;, &quot;1&quot;)
#define %mutable %feature(&quot;immutable&quot;)
/* Directives for callback functions */
#define %callback(x) %feature(&quot;callback&quot;) `x`;
#define %nocallback %feature(&quot;callback&quot;);
/* %ignore directive */
#define %ignore %rename($ignore)
#define %ignorewarn(x) %rename(&quot;$ignore:&quot; x)
...
</pre>
</div>
<p> The fact that most of the standard SWIG directives are macros is
intended to simplify the implementation of the internals. For instance,
rather than having to support dozens of special directives, it is
easier to have a few basic primitives such as <tt>%feature</tt> or <tt>
%insert</tt>.</p>
<p> The<em> <tt>langconfig.swg</tt></em> file is supplied by the target
language. This file contains language-specific configuration
information. More often than not, this file provides run-time wrapper
support code (e.g., the type-checker) as well as a collection of
typemaps that define the default wrapping behavior. Note: the name of
this file depends on the target language and is usually something like <tt>
python.swg</tt> or <tt>perl5.swg</tt>.</p>
<p> As a debugging aid, the text that SWIG feeds to its C++ parser can
be obtained by running <tt>swig -E interface.i</tt>. This option, like
the same option that regular C/C++ compilers support, generates the
preprocessed output and is useful for looking at how macros have been
expanded as well as everything else that goes into the low-level
construction of the wrapper code. Also, like a regular C/C++ compiler,
the preprocessed output can be generated in one invocation and then fed
back in with a second invocation. This is the approach that the <a href="#CCache">
CCache</a> tool uses as part of its strategy to speed up repeated builds
with the same inputs.</p>
<h3><a name="Extending_nn7">41.4.2 Parsing</a></h3>
<p> The current C++ parser handles a subset of C++. Most
incompatibilities with C are due to subtle aspects of how SWIG parses
declarations. Specifically, SWIG expects all C/C++ declarations to
follow this general form:</p>
<div class="diagram">
<pre>
<em>storage</em> <em>type</em> <em>declarator</em> <em>initializer</em>;
</pre>
</div>
<p> <tt><em>storage</em></tt> is a keyword such as <tt>extern</tt>, <tt>
static</tt>, <tt>typedef</tt>, or <tt>virtual</tt>. <tt><em>type</em></tt>
is a primitive datatype such as <tt>int</tt> or <tt>void</tt>. <tt><em>
type</em></tt> may be optionally qualified with a qualifier such as <tt>
const</tt> or <tt>volatile</tt>. <tt><em>declarator</em></tt> is a name
with additional type-construction modifiers attached to it (pointers,
arrays, references, functions, etc.). Examples of declarators include <tt>
*x</tt>, <tt>**x</tt>, <tt>x[20]</tt>, and <tt>(*x)(int, double)</tt>.
The <tt><em>initializer</em></tt> may be a value assigned using <tt>=</tt>
or body of code enclosed in braces <tt>{ ... }</tt>.</p>
<p> This declaration format covers most common C++ declarations.
However, the C++ standard is somewhat more flexible in the placement of
the parts. For example, it is technically legal, although uncommon to
write something like <tt>int typedef const a</tt> in your program. SWIG
simply doesn't bother to deal with this case.</p>
<p> The other significant difference between C++ and SWIG is in the
treatment of typenames. In C++, if you have a declaration like this,</p>
<div class="code">
<pre>
int blah(Foo *x, Bar *y);
</pre>
</div>
<p> it won't parse correctly unless <tt>Foo</tt> and <tt>Bar</tt> have
been previously defined as types either using a <tt>class</tt>
definition or a <tt>typedef</tt>. The reasons for this are subtle, but
this treatment of typenames is normally integrated at the level of the
C tokenizer---when a typename appears, a different token is returned to
the parser instead of an identifier.</p>
<p> SWIG does not operate in this manner--any legal identifier can be
used as a type name. The reason for this is primarily motivated by the
use of SWIG with partially defined data. Specifically, SWIG is supposed
to be easy to use on interfaces with missing type information.</p>
<p> Because of the different treatment of typenames, the most serious
limitation of the SWIG parser is that it can't process type
declarations where an extra (and unnecessary) grouping operator is
used. For example:</p>
<div class="code">
<pre>
int (x); /* A variable x */
int (y)(int); /* A function y */
</pre>
</div>
<p> The placing of extra parentheses in type declarations like this is
already recognized by the C++ community as a potential source of
strange programming errors. For example, Scott Meyers &quot;Effective STL&quot;
discusses this problem in a section on avoiding C++'s &quot;most vexing
parse.&quot;</p>
<p> The parser is also unable to handle declarations with no return type
or bare argument names. For example, in an old C program, you might see
things like this:</p>
<div class="code">
<pre>
foo(a, b) {
...
}
</pre>
</div>
<p> In this case, the return type as well as the types of the arguments
are taken by the C compiler to be an <tt>int</tt>. However, SWIG
interprets the above code as an abstract declarator for a function
returning a <tt>foo</tt> and taking types <tt>a</tt> and <tt>b</tt> as
arguments).</p>
<h3><a name="Extending_nn8">41.4.3 Parse Trees</a></h3>
<p> The SWIG parser produces a complete parse tree of the input file
before any wrapper code is actually generated. Each item in the tree is
known as a &quot;Node&quot;. Each node is identified by a symbolic tag.
Furthermore, a node may have an arbitrary number of children. The parse
tree structure and tag names of an interface can be displayed using <tt>
swig -debug-tags</tt>. For example:</p>
<div class="shell">
<pre>
$ <b>swig -c++ -python -debug-tags example.i</b>
. top (example.i:1)
. top . include (example.i:1)
. top . include . typemap (/r0/beazley/Projects/lib/swig1.3/swig.swg:71)
. top . include . typemap . typemapitem (/r0/beazley/Projects/lib/swig1.3/swig.swg:71)
. top . include . typemap (/r0/beazley/Projects/lib/swig1.3/swig.swg:83)
. top . include . typemap . typemapitem (/r0/beazley/Projects/lib/swig1.3/swig.swg:83)
. top . include (example.i:4)
. top . include . insert (/r0/beazley/Projects/lib/swig1.3/python/python.swg:7)
. top . include . insert (/r0/beazley/Projects/lib/swig1.3/python/python.swg:8)
. top . include . typemap (/r0/beazley/Projects/lib/swig1.3/python/python.swg:19)
...
. top . include (example.i:6)
. top . include . module (example.i:2)
. top . include . insert (example.i:6)
. top . include . include (example.i:9)
. top . include . include . class (example.h:3)
. top . include . include . class . access (example.h:4)
. top . include . include . class . constructor (example.h:7)
. top . include . include . class . destructor (example.h:10)
. top . include . include . class . cdecl (example.h:11)
. top . include . include . class . cdecl (example.h:11)
. top . include . include . class . cdecl (example.h:12)
. top . include . include . class . cdecl (example.h:13)
. top . include . include . class . cdecl (example.h:14)
. top . include . include . class . cdecl (example.h:15)
. top . include . include . class (example.h:18)
. top . include . include . class . access (example.h:19)
. top . include . include . class . cdecl (example.h:20)
. top . include . include . class . access (example.h:21)
. top . include . include . class . constructor (example.h:22)
. top . include . include . class . cdecl (example.h:23)
. top . include . include . class . cdecl (example.h:24)
. top . include . include . class (example.h:27)
. top . include . include . class . access (example.h:28)
. top . include . include . class . cdecl (example.h:29)
. top . include . include . class . access (example.h:30)
. top . include . include . class . constructor (example.h:31)
. top . include . include . class . cdecl (example.h:32)
. top . include . include . class . cdecl (example.h:33)
</pre>
</div>
<p> Even for the most simple interface, the parse tree structure is
larger than you might expect. For example, in the above output, a
substantial number of nodes are actually generated by the <tt>
python.swg</tt> configuration file which defines typemaps and other
directives. The contents of the user-supplied input file don't appear
until the end of the output.</p>
<p> The contents of each parse tree node consist of a collection of
attribute/value pairs. Internally, the nodes are simply represented by
hash tables. A display of the entire parse-tree structure can be
obtained using <tt>swig -debug-top &lt;n&gt;</tt>, where <tt>n</tt> is the
stage being processed. There are a number of other parse tree display
options, for example, <tt>swig -debug-module &lt;n&gt;</tt> will avoid
displaying system parse information and only display the parse tree
pertaining to the user's module at stage <tt>n</tt> of processing.
Adding the <tt>-debug-quiet</tt> option is recommended as it removes
some noise which is not usually needed, that is, the display of many
linked list pointers and symbol table pointers. See <a href="#Extending_debugging_options">
Debugging Options</a> for a full list.</p>
<div class="shell">
<pre>
$ swig -c++ -python -debug-module 1 -debug-quiet example.i
debug-module stage 1
+++ module ----------------------------------------
| name - &quot;example&quot;
|
+++ insert ----------------------------------------
| code - &quot;\n#include \&quot;example.h\&quot;\n&quot;
|
+++ include ----------------------------------------
| name - &quot;example.h&quot;
+++ class ----------------------------------------
| abstracts - 0x7f4f15182930
| allows_typedef - &quot;1&quot;
| kind - &quot;class&quot;
| name - &quot;Shape&quot;
| sym:name - &quot;Shape&quot;
+++ access ----------------------------------------
| kind - &quot;public&quot;
|
+++ constructor ----------------------------------------
| access - &quot;public&quot;
| code - &quot;{\n nshapes++;\n }&quot;
| decl - &quot;f().&quot;
| feature:new - &quot;1&quot;
| ismember - &quot;1&quot;
| name - &quot;Shape&quot;
| sym:name - &quot;Shape&quot;
|
+++ destructor ----------------------------------------
| access - &quot;public&quot;
| code - &quot;{\n nshapes--;\n }&quot;
| decl - &quot;f().&quot;
| ismember - &quot;1&quot;
| name - &quot;~Shape&quot;
| storage - &quot;virtual&quot;
| sym:name - &quot;~Shape&quot;
|
+++ cdecl ----------------------------------------
| access - &quot;public&quot;
| decl - &quot;&quot;
| ismember - &quot;1&quot;
| kind - &quot;variable&quot;
| name - &quot;x&quot;
| sym:name - &quot;x&quot;
| type - &quot;double&quot;
|
+++ cdecl ----------------------------------------
| access - &quot;public&quot;
| decl - &quot;&quot;
| ismember - &quot;1&quot;
| kind - &quot;variable&quot;
| name - &quot;y&quot;
| sym:name - &quot;y&quot;
| type - &quot;double&quot;
|
+++ cdecl ----------------------------------------
| access - &quot;public&quot;
| decl - &quot;f(double,double).&quot;
| ismember - &quot;1&quot;
| kind - &quot;function&quot;
| name - &quot;move&quot;
| parms - 'double dx,double dy'
| sym:name - &quot;move&quot;
| type - &quot;void&quot;
|
+++ cdecl ----------------------------------------
| abstract - &quot;1&quot;
| access - &quot;public&quot;
| decl - &quot;f().&quot;
| ismember - &quot;1&quot;
| kind - &quot;function&quot;
| name - &quot;area&quot;
| storage - &quot;virtual&quot;
| sym:name - &quot;area&quot;
| type - &quot;double&quot;
| value - &quot;0&quot;
| valuetype - &quot;int&quot;
|
+++ cdecl ----------------------------------------
| abstract - &quot;1&quot;
| access - &quot;public&quot;
| decl - &quot;f().&quot;
| ismember - &quot;1&quot;
| kind - &quot;function&quot;
| name - &quot;perimeter&quot;
| storage - &quot;virtual&quot;
| sym:name - &quot;perimeter&quot;
| type - &quot;double&quot;
| value - &quot;0&quot;
| valuetype - &quot;int&quot;
|
+++ cdecl ----------------------------------------
| access - &quot;public&quot;
| decl - &quot;&quot;
| ismember - &quot;1&quot;
| kind - &quot;variable&quot;
| name - &quot;nshapes&quot;
| storage - &quot;static&quot;
| sym:name - &quot;nshapes&quot;
| type - &quot;int&quot;
|
+++ class ----------------------------------------
| allows_typedef - &quot;1&quot;
| baselist - 0x7f4f15182ad0
| kind - &quot;class&quot;
| name - &quot;Circle&quot;
| privatebaselist - 0x7f4f15182b10
| protectedbaselist - 0x7f4f15182af0
| sym:name - &quot;Circle&quot;
+++ access ----------------------------------------
| kind - &quot;private&quot;
|
+++ cdecl ----------------------------------------
| access - &quot;private&quot;
| decl - &quot;&quot;
| ismember - &quot;1&quot;
| kind - &quot;variable&quot;
| name - &quot;radius&quot;
| type - &quot;double&quot;
|
+++ access ----------------------------------------
| kind - &quot;public&quot;
|
+++ constructor ----------------------------------------
| access - &quot;public&quot;
| code - &quot;{ }&quot;
| decl - &quot;f(double).&quot;
| feature:new - &quot;1&quot;
| ismember - &quot;1&quot;
| name - &quot;Circle&quot;
| parms - 'double r'
| sym:name - &quot;Circle&quot;
|
+++ cdecl ----------------------------------------
| access - &quot;public&quot;
| decl - &quot;f().&quot;
| ismember - &quot;1&quot;
| kind - &quot;function&quot;
| name - &quot;area&quot;
| storage - &quot;virtual&quot;
| sym:name - &quot;area&quot;
| type - &quot;double&quot;
|
+++ cdecl ----------------------------------------
| access - &quot;public&quot;
| decl - &quot;f().&quot;
| ismember - &quot;1&quot;
| kind - &quot;function&quot;
| name - &quot;perimeter&quot;
| storage - &quot;virtual&quot;
| sym:name - &quot;perimeter&quot;
| type - &quot;double&quot;
|
+++ class ----------------------------------------
| allows_typedef - &quot;1&quot;
| baselist - 0x7f4f15183830
| kind - &quot;class&quot;
| name - &quot;Square&quot;
| privatebaselist - 0x7f4f15183870
| protectedbaselist - 0x7f4f15183850
| sym:name - &quot;Square&quot;
+++ access ----------------------------------------
| kind - &quot;private&quot;
|
+++ cdecl ----------------------------------------
| access - &quot;private&quot;
| decl - &quot;&quot;
| ismember - &quot;1&quot;
| kind - &quot;variable&quot;
| name - &quot;width&quot;
| type - &quot;double&quot;
|
+++ access ----------------------------------------
| kind - &quot;public&quot;
|
+++ constructor ----------------------------------------
| access - &quot;public&quot;
| code - &quot;{ }&quot;
| decl - &quot;f(double).&quot;
| feature:new - &quot;1&quot;
| ismember - &quot;1&quot;
| name - &quot;Square&quot;
| parms - 'double w'
| sym:name - &quot;Square&quot;
|
+++ cdecl ----------------------------------------
| access - &quot;public&quot;
| decl - &quot;f().&quot;
| ismember - &quot;1&quot;
| kind - &quot;function&quot;
| name - &quot;area&quot;
| storage - &quot;virtual&quot;
| sym:name - &quot;area&quot;
| type - &quot;double&quot;
|
+++ cdecl ----------------------------------------
| access - &quot;public&quot;
| decl - &quot;f().&quot;
| ismember - &quot;1&quot;
| kind - &quot;function&quot;
| name - &quot;perimeter&quot;
| storage - &quot;virtual&quot;
| sym:name - &quot;perimeter&quot;
| type - &quot;double&quot;
</pre>
</div>
<h3><a name="Extending_nn9">41.4.4 Attribute namespaces</a></h3>
<p> Attributes of parse tree nodes are often prepended with a namespace
qualifier. For example, the attributes <tt>sym:name</tt> and <tt>
sym:symtab</tt> are attributes related to symbol table management and
are prefixed with <tt>sym:</tt>. As a general rule, only those
attributes which are directly related to the raw declaration appear
without a prefix (type, name, declarator, etc.).</p>
<p> Target language modules may add additional attributes to nodes to
assist the generation of wrapper code. The convention for doing this is
to place these attributes in a namespace that matches the name of the
target language. For example, <tt>python:foo</tt> or <tt>perl:foo</tt>.</p>
<h3><a name="Extending_nn10">41.4.5 Symbol Tables</a></h3>
<p> During parsing, all symbols are managed in the space of the target
language. The <tt>sym:name</tt> attribute of each node contains the
symbol name selected by the parser. Normally, <tt>sym:name</tt> and <tt>
name</tt> are the same. However, the <tt>%rename</tt> directive can be
used to change the value of <tt>sym:name</tt>. You can see the effect
of <tt>%rename</tt> by trying it on a simple interface and dumping the
parse tree. For example:</p>
<div class="code">
<pre>
%rename(foo_i) foo(int);
%rename(foo_d) foo(double);
void foo(int);
void foo(double);
void foo(Bar *b);
</pre>
</div>
<p> There are various <tt>debug-</tt> options that can be useful for
debugging and analysing the parse tree. For example, the <tt>debug-top
&lt;n&gt;</tt> or <tt>debug-module &lt;n&gt;</tt> options will dump the entire/top
of the parse tree or the module subtree at one of the four <tt>n</tt>
stages of processing. The parse tree can be viewed after the final
stage of processing by running SWIG:</p>
<div class="shell">
<pre>
$ swig -debug-top 1 -debug-quiet example.i
...
+++ cdecl ----------------------------------------
| decl - &quot;f(int).&quot;
| name - &quot;foo&quot;
| parms - int
| sym:name - &quot;foo_i&quot;
| type - &quot;void&quot;
|
+++ cdecl ----------------------------------------
| decl - &quot;f(double).&quot;
| name - &quot;foo&quot;
| parms - double
| sym:name - &quot;foo_d&quot;
| type - &quot;void&quot;
|
+++ cdecl ----------------------------------------
| decl - &quot;f(p.Bar).&quot;
| name - &quot;foo&quot;
| parms - Bar *
| sym:name - &quot;foo&quot;
| type - &quot;void&quot;
</pre>
</div>
<p> All symbol-related conflicts and complaints about overloading are
based on <tt>sym:name</tt> values. For instance, the following example
uses <tt>%rename</tt> in reverse to generate a name clash.</p>
<div class="code">
<pre>
%rename(foo) foo_i(int);
%rename(foo) foo_d(double);
void foo_i(int);
void foo_d(double);
void foo(Bar *b);
</pre>
</div>
<p> When you run SWIG on this you now get:</p>
<div class="shell">
<pre>
$ ./swig example.i
example.i:6. Overloaded declaration ignored. foo_d(double )
example.i:5. Previous declaration is foo_i(int )
example.i:7. Overloaded declaration ignored. foo(Bar *)
example.i:5. Previous declaration is foo_i(int )
</pre>
</div>
<h3><a name="Extending_nn11">41.4.6 The %feature directive</a></h3>
<p> A number of SWIG directives such as <tt>%exception</tt> are
implemented using the low-level <tt>%feature</tt> directive. For
example:</p>
<div class="code">
<pre>
%feature(&quot;except&quot;) getitem(int) {
try {
$action
} catch (badindex) {
...
}
}
...
class Foo {
public:
Object *getitem(int index) throws(badindex);
...
};
</pre>
</div>
<p> The behavior of <tt>%feature</tt> is very easy to describe--it
simply attaches a new attribute to any parse tree node that matches the
given prototype. When a feature is added, it shows up as an attribute
in the <tt>feature:</tt> namespace. You can see this when running with
the <tt>-debug-top 4 -debug-quiet</tt> option. For example:</p>
<div class="shell">
<pre>
+++ cdecl ----------------------------------------
| decl - &quot;f(int).p.&quot;
| feature:except - &quot;{\n try {\n $action\n } catc...&quot;
| name - &quot;getitem&quot;
| parms - int
| sym:name - &quot;getitem&quot;
| type - &quot;Object&quot;
|
</pre>
</div>
<p> Feature names are completely arbitrary and a target language module
can be programmed to respond to any feature name that it wants to
recognize. The data stored in a feature attribute is usually just a raw
unparsed string. For example, the exception code above is simply stored
without any modifications.</p>
<h3><a name="Extending_nn12">41.4.7 Code Generation</a></h3>
<p> Language modules work by defining handler functions that know how to
respond to different types of parse-tree nodes. These handlers simply
look at the attributes of each node in order to produce low-level code.</p>
<p> In reality, the generation of code is somewhat more subtle than
simply invoking handler functions. This is because parse-tree nodes
might be transformed. For example, suppose you are wrapping a class
like this:</p>
<div class="code">
<pre>
class Foo {
public:
virtual int *bar(int x);
};
</pre>
</div>
<p> When the parser constructs a node for the member <tt>bar</tt>, it
creates a raw &quot;cdecl&quot; node with the following attributes:</p>
<div class="diagram">
<pre>
nodeType : cdecl
name : bar
type : int
decl : f(int).p
parms : int x
storage : virtual
sym:name : bar
</pre>
</div>
<p> To produce wrapper code, this &quot;cdecl&quot; node undergoes a number of
transformations. First, the node is recognized as a function
declaration. This adjusts some of the type information--specifically,
the declarator is joined with the base datatype to produce this:</p>
<div class="diagram">
<pre>
nodeType : cdecl
name : bar
type : p.int &lt;-- Notice change in return type
decl : f(int).p
parms : int x
storage : virtual
sym:name : bar
</pre>
</div>
<p> Next, the context of the node indicates that the node is really a
member function. This produces a transformation to a low-level accessor
function like this:</p>
<div class="diagram">
<pre>
nodeType : cdecl
name : bar
type : int.p
decl : f(int).p
parms : Foo *self, int x &lt;-- Added parameter
storage : virtual
wrap:action : result = (arg1)-&gt;bar(arg2) &lt;-- Action code added
sym:name : Foo_bar &lt;-- Symbol name changed
</pre>
</div>
<p> In this transformation, notice how an additional parameter was added
to the parameter list and how the symbol name of the node has suddenly
changed into an accessor using the naming scheme described in the &quot;SWIG
Basics&quot; chapter. A small fragment of &quot;action&quot; code has also been
generated--notice how the <tt>wrap:action</tt> attribute defines the
access to the underlying method. The data in this transformed node is
then used to generate a wrapper.</p>
<p> Language modules work by registering handler functions for dealing
with various types of nodes at different stages of transformation. This
is done by inheriting from a special <tt>Language</tt> class and
defining a collection of virtual methods. For example, the Python
module defines a class as follows:</p>
<div class="code">
<pre>
class PYTHON : public Language {
protected:
public :
virtual void main(int, char *argv[]);
virtual int top(Node *);
virtual int functionWrapper(Node *);
virtual int constantWrapper(Node *);
virtual int variableWrapper(Node *);
virtual int nativeWrapper(Node *);
virtual int membervariableHandler(Node *);
virtual int memberconstantHandler(Node *);
virtual int memberfunctionHandler(Node *);
virtual int constructorHandler(Node *);
virtual int destructorHandler(Node *);
virtual int classHandler(Node *);
virtual int classforwardDeclaration(Node *);
virtual int insertDirective(Node *);
virtual int importDirective(Node *);
};
</pre>
</div>
<p> The role of these functions is described shortly.</p>
<h3><a name="Extending_nn13">41.4.8 SWIG and XML</a></h3>
<p> Much of SWIG's current parser design was originally motivated by
interest in using XML to represent SWIG parse trees. Although XML is
not currently used in any direct manner, the parse tree structure, use
of node tags, attributes, and attribute namespaces are all influenced
by aspects of XML parsing. Therefore, in trying to understand SWIG's
internal data structures, it may be useful to keep XML in the back of
your mind as a model.</p>
<p> In addition to the options beginning <tt>-debug-</tt> for dumping
out SWIG's parse tree in a simple text format, the parse tree can also
be dumped as XML. There are three options below, where <tt>file</tt>
specifies the name of a file for generating the resulting XML into:</p>
<ol>
<li>
<p> <tt>-xmlout &lt;file&gt;</tt></p>
<p> Use this XML option in addition to any target language options. The
parse tree is dumped after the final stage of processing, that is,
after the target language has finished processing. It is useful for
seeing the parse tree at stage 4 after any target language processing.
The parse tree dumped is very similar in content to that generated by <tt>
-debug-top 4</tt>.</p>
</li>
<li>
<p> <tt>-xml -o &lt;file&gt;</tt></p>
<p> Use this XML option without specifying a target language. The parse
tree that is dumped is the same as if there was a no target language
option. It is useful for seeing the parse tree at stage 3, which is the
same as stage 4, as there is no target language processing. The parse
tree dumped is very similar in content to that generated by <tt>
-debug-top 3</tt>.</p>
</li>
<li>
<p> <tt>-xml -xmllite -o &lt;file&gt;</tt></p>
<p> Same as above, except the addition of <tt>-xmllite</tt> reduces the
output by skipping some type information, that is, the typescope and
typetab nodes.</p>
</li>
</ol>
<p> The parse tree generated using <tt>-xmlout</tt> is much bigger than
that using <tt>-xml</tt> as the target languages process numerous extra
SWIG system files and also add to the parse tree quite considerably in
stage 4.</p>
<p> The XML is a dump of SWIG's internal parse tree and as such it is
subject to change at any time as and when SWIG's implementation
changes.</p>
<h2><a name="Extending_nn14">41.5 Primitive Data Structures</a></h2>
<p> Most of SWIG is constructed using three basic data structures:
strings, hashes, and lists. These data structures are dynamic in same
way as similar structures found in many scripting languages. For
instance, you can have containers (lists and hash tables) of mixed
types and certain operations are polymorphic.</p>
<p> This section briefly describes the basic structures so that later
sections of this chapter make more sense.</p>
<p> When describing the low-level API, the following type name
conventions are used:</p>
<ul>
<li><tt>String</tt>. A string object.</li>
<li><tt>Hash</tt>. A hash object.</li>
<li><tt>List</tt>. A list object.</li>
<li><tt>String_or_char</tt>. A string object or a <tt>char *</tt>.</li>
<li><tt>Object_or_char</tt>. An object or a <tt>char *</tt>.</li>
<li><tt>Object</tt>. Any object (string, hash, list, etc.)</li>
</ul>
<p> In most cases, other typenames in the source are aliases for one of
these primitive types. Specifically:</p>
<div class="code">
<pre>
typedef String SwigType;
typedef Hash Parm;
typedef Hash ParmList;
typedef Hash Node;
typedef Hash Symtab;
typedef Hash Typetab;
</pre>
</div>
<h3><a name="Extending_nn15">41.5.1 Strings</a></h3>
<p><b> <tt>String *NewString(const String_or_char *val)</tt></b></p>
<div class="indent"> Creates a new string with initial value <tt>val</tt>
. <tt>val</tt> may be a <tt>char *</tt> or another <tt>String</tt>
object. If you want to create an empty string, use &quot;&quot; for val.</div>
<p><b> <tt>String *NewStringf(const char *fmt, ...)</tt></b></p>
<div class="indent"> Creates a new string whose initial value is set
according to a C <tt>printf</tt> style format string in <tt>fmt</tt>.
Additional arguments follow depending on <tt>fmt</tt>.</div>
<p><b> <tt>String *Copy(String *s)</tt></b></p>
<div class="indent"> Make a copy of the string <tt>s</tt>.</div>
<p><b> <tt>void Delete(String *s)</tt></b></p>
<div class="indent"> Deletes <tt>s</tt>.</div>
<p><b> <tt>int Len(const String_or_char *s)</tt></b></p>
<div class="indent"> Returns the length of the string.</div>
<p><b> <tt>char *Char(const String_or_char *s)</tt></b></p>
<div class="indent"> Returns a pointer to the first character in a
string.</div>
<p><b> <tt>void Append(String *s, const String_or_char *t)</tt></b></p>
<div class="indent"> Appends <tt>t</tt> to the end of string <tt>s</tt>.</div>
<p><b> <tt>void Insert(String *s, int pos, const String_or_char *t)</tt></b>
</p>
<div class="indent"> Inserts <tt>t</tt> into <tt>s</tt> at position <tt>
pos</tt>. The contents of <tt>s</tt> are shifted accordingly. The
special value <tt>DOH_END</tt> can be used for <tt>pos</tt> to indicate
insertion at the end of the string (appending).</div>
<p><b> <tt>int Strcmp(const String_or_char *s, const String_or_char *t)</tt>
</b></p>
<div class="indent"> Compare strings <tt>s</tt> and <tt>t</tt>. Same as
the C <tt>strcmp()</tt> function.</div>
<p><b> <tt>int Strncmp(const String_or_char *s, const String_or_char *t,
int len)</tt></b></p>
<div class="indent"> Compare the first <tt>len</tt> characters of
strings <tt>s</tt> and <tt>t</tt>. Same as the C <tt>strncmp()</tt>
function.</div>
<p><b> <tt>char *Strstr(const String_or_char *s, const String_or_char
*pat)</tt></b></p>
<div class="indent"> Returns a pointer to the first occurrence of <tt>
pat</tt> in <tt>s</tt>. Same as the C <tt>strstr()</tt> function.</div>
<p><b> <tt>char *Strchr(const String_or_char *s, char ch)</tt></b></p>
<div class="indent"> Returns a pointer to the first occurrence of
character <tt>ch</tt> in <tt>s</tt>. Same as the C <tt>strchr()</tt>
function.</div>
<p><b> <tt>void Chop(String *s)</tt></b></p>
<div class="indent"> Chops trailing whitespace off the end of <tt>s</tt>
.</div>
<p><b> <tt>int Replace(String *s, const String_or_char *pat, const
String_or_char *rep, int flags)</tt></b></p>
<div class="indent">
<p> Replaces the pattern <tt>pat</tt> with <tt>rep</tt> in string <tt>s</tt>
. <tt>flags</tt> is a combination of the following flags:</p>
<div class="code">
<pre>
DOH_REPLACE_ANY - Replace all occurrences
DOH_REPLACE_ID - Valid C identifiers only
DOH_REPLACE_NOQUOTE - Don't replace in quoted strings
DOH_REPLACE_FIRST - Replace first occurrence only.
</pre>
</div>
<p> Returns the number of replacements made (if any).</p>
<p> At most one of <tt>DOH_REPLACE_ANY</tt> and <tt>DOH_REPLACE_FIRST</tt>
should be specified. <tt>DOH_REPLACE_ANY</tt> is the default if neither
is specified.</p>
</div>
<h3><a name="Extending_nn16">41.5.2 Hashes</a></h3>
<p><b> <tt>Hash *NewHash()</tt></b></p>
<div class="indent"> Creates a new empty hash table.</div>
<p><b> <tt>Hash *Copy(Hash *h)</tt></b></p>
<div class="indent"> Make a shallow copy of the hash <tt>h</tt>.</div>
<p><b> <tt>void Delete(Hash *h)</tt></b></p>
<div class="indent"> Deletes <tt>h</tt>.</div>
<p><b> <tt>int Len(Hash *h)</tt></b></p>
<div class="indent"> Returns the number of items in <tt>h</tt>.</div>
<p><b> <tt>Object *Getattr(Hash *h, const String_or_char *key)</tt></b></p>
<div class="indent"> Gets an object from <tt>h</tt>. <tt>key</tt> may be
a string or a simple <tt>char *</tt> string. Returns NULL if not found.</div>
<p><b> <tt>int Setattr(Hash *h, const String_or_char *key, const
Object_or_char *val)</tt></b></p>
<div class="indent"> Stores <tt>val</tt> in <tt>h</tt>. <tt>key</tt> may
be a string or a simple <tt>char *</tt>. If <tt>val</tt> is not a
standard object (String, Hash, or List) it is assumed to be a <tt>char
*</tt> in which case it is used to construct a <tt>String</tt> that is
stored in the hash. If <tt>val</tt> is NULL, the object is deleted.
Increases the reference count of <tt>val</tt>. Returns 1 if this
operation replaced an existing hash entry, 0 otherwise.</div>
<p><b> <tt>int Delattr(Hash *h, const String_or_char *key)</tt></b></p>
<div class="indent"> Deletes the hash item referenced by <tt>key</tt>.
Decreases the reference count on the corresponding object (if any).
Returns 1 if an object was removed, 0 otherwise.</div>
<p><b> <tt>List *Keys(Hash *h)</tt></b></p>
<div class="indent"> Returns the list of hash table keys.</div>
<p><b> <tt>List *SortedKeys(Hash *h, int (*cmp) (const DOH *, const DOH
*))</tt></b></p>
<div class="indent"> Returns the list of sorted hash table keys.</div>
<h3><a name="Extending_nn17">41.5.3 Lists</a></h3>
<p><b> <tt>List *NewList()</tt></b></p>
<div class="indent"> Creates a new empty list.</div>
<p><b> <tt>List *Copy(List *x)</tt></b></p>
<div class="indent"> Make a shallow copy of the List <tt>x</tt>.</div>
<p><b> <tt>void Delete(List *x)</tt></b></p>
<div class="indent"> Deletes <tt>x</tt>.</div>
<p><b> <tt>int Len(List *x)</tt></b></p>
<div class="indent"> Returns the number of items in <tt>x</tt>.</div>
<p><b> <tt>Object *Getitem(List *x, int n)</tt></b></p>
<div class="indent"> Returns an object from <tt>x</tt> with index <tt>n</tt>
. If <tt>n</tt> is beyond the end of the list, the last item is
returned. If <tt>n</tt> is negative, the first item is returned.</div>
<p><b> <tt>int *Setitem(List *x, int n, const Object_or_char *val)</tt></b>
</p>
<div class="indent"> Stores <tt>val</tt> in <tt>x</tt>. If <tt>val</tt>
is not a standard object (String, Hash, or List) it is assumed to be a <tt>
char *</tt> in which case it is used to construct a <tt>String</tt> that
is stored in the list. <tt>n</tt> must be in range. Otherwise, an
assertion will be raised.</div>
<p><b> <tt>int *Delitem(List *x, int n)</tt></b></p>
<div class="indent"> Deletes item <tt>n</tt> from the list, shifting
items down if necessary. To delete the last item in the list, use the
special value <tt>DOH_END</tt> for <tt>n</tt>.</div>
<p><b> <tt>void Append(List *x, const Object_or_char *t)</tt></b></p>
<div class="indent"> Appends <tt>t</tt> to the end of <tt>x</tt>. If <tt>
t</tt> is not a standard object, it is assumed to be a <tt>char *</tt>
and is used to create a String object.</div>
<p><b> <tt>void Insert(String *s, int pos, const Object_or_char *t)</tt></b>
</p>
<div class="indent"> Inserts <tt>t</tt> into <tt>s</tt> at position <tt>
pos</tt>. The contents of <tt>s</tt> are shifted accordingly. The
special value <tt>DOH_END</tt> can be used for <tt>pos</tt> to indicate
insertion at the end of the list (appending). If <tt>t</tt> is not a
standard object, it is assumed to be a <tt>char *</tt> and is used to
create a String object.</div>
<h3><a name="Extending_nn18">41.5.4 Common operations</a></h3>
The following operations are applicable to all datatypes.
<p><b> <tt>Object *Copy(Object *x)</tt></b></p>
<div class="indent"> Make a copy of the object <tt>x</tt>.</div>
<p><b> <tt>void Delete(Object *x)</tt></b></p>
<div class="indent"> Deletes <tt>x</tt>.</div>
<p><b> <tt>void Setfile(Object *x, String_or_char *f)</tt></b></p>
<div class="indent"> Sets the filename associated with <tt>x</tt>. Used
to track objects and report errors.</div>
<p><b> <tt>String *Getfile(Object *x)</tt></b></p>
<div class="indent"> Gets the filename associated with <tt>x</tt>.</div>
<p><b> <tt>void Setline(Object *x, int n)</tt></b></p>
<div class="indent"> Sets the line number associated with <tt>x</tt>.
Used to track objects and report errors.</div>
<p><b> <tt>int Getline(Object *x)</tt></b></p>
<div class="indent"> Gets the line number associated with <tt>x</tt>.</div>
<h3><a name="Extending_nn19">41.5.5 Iterating over Lists and Hashes</a></h3>
To iterate over the elements of a list or a hash table, the following
functions are used:
<p><b> <tt>Iterator First(Object *x)</tt></b></p>
<div class="indent"> Returns an iterator object that points to the first
item in a list or hash table. The <tt>item</tt> attribute of the
Iterator object is a pointer to the item. For hash tables, the <tt>key</tt>
attribute of the Iterator object additionally points to the
corresponding Hash table key. The <tt>item</tt> and <tt>key</tt>
attributes are NULL if the object contains no items or if there are no
more items.</div>
<p><b> <tt>Iterator Next(Iterator i)</tt></b></p>
<div class="indent">
<p>Returns an iterator that points to the next item in a list or hash
table. Here are two examples of iteration:</p>
<div class="code">
<pre>
List *l = (some list);
Iterator i;
for (i = First(l); i.item; i = Next(i)) {
Printf(stdout, &quot;%s\n&quot;, i.item);
}
Hash *h = (some hash);
Iterator j;
for (j = First(j); j.item; j= Next(j)) {
Printf(stdout, &quot;%s : %s\n&quot;, j.key, j.item);
}
</pre>
</div></div>
<h3><a name="Extending_nn20">41.5.6 I/O</a></h3>
Special I/O functions are used for all internal I/O. These operations
work on C <tt>FILE *</tt> objects, String objects, and special <tt>File</tt>
objects (which are merely a wrapper around <tt>FILE *</tt>).
<p><b> <tt>int Printf(String_or_FILE *f, const char *fmt, ...)</tt></b></p>
<div class="indent"> Formatted I/O. Same as the C <tt>fprintf()</tt>
function except that output can also be directed to a string object.
Note: the <tt>%s</tt> format specifier works with both strings and <tt>
char *</tt>. All other format operators have the same meaning.</div>
<p><b> <tt>int Printv(String_or_FILE *f, String_or_char *arg1, ...,
NULL)</tt></b></p>
<div class="indent"> Prints a variable number of strings arguments to
the output. The last argument to this function must be NULL. The other
arguments can either be <tt>char *</tt> or string objects.</div>
<p><b> <tt>int Putc(int ch, String_or_FILE *f)</tt></b></p>
<div class="indent"> Same as the C <tt>fputc()</tt> function.</div>
<p><b> <tt>int Write(String_or_FILE *f, void *buf, int len)</tt></b></p>
<div class="indent"> Same as the C <tt>write()</tt> function.</div>
<p><b> <tt>int Read(String_or_FILE *f, void *buf, int maxlen)</tt></b></p>
<div class="indent"> Same as the C <tt>read()</tt> function.</div>
<p><b> <tt>int Getc(String_or_FILE *f)</tt></b></p>
<div class="indent"> Same as the C <tt>fgetc()</tt> function.</div>
<p><b> <tt>int Ungetc(int ch, String_or_FILE *f)</tt></b></p>
<div class="indent"> Same as the C <tt>ungetc()</tt> function.</div>
<p><b> <tt>int Seek(String_or_FILE *f, int offset, int whence)</tt></b></p>
<div class="indent"> Same as the C <tt>seek()</tt> function. <tt>offset</tt>
is the number of bytes. <tt>whence</tt> is one of <tt>SEEK_SET</tt>, <tt>
SEEK_CUR</tt>, or <tt>SEEK_END</tt>..</div>
<p><b> <tt>long Tell(String_or_FILE *f)</tt></b></p>
<div class="indent"> Same as the C <tt>tell()</tt> function.</div>
<p><b> <tt>File *NewFile(const char *filename, const char *mode, List
*newfiles)</tt></b></p>
<div class="indent"> Create a File object using the <tt>fopen()</tt>
library call. This file differs from <tt>FILE *</tt> in that it can be
placed in the standard SWIG containers (lists, hashes, etc.). The <tt>
filename</tt> is added to the <tt>newfiles</tt> list if <tt>newfiles</tt>
is non-zero and the file was created successfully.</div>
<p><b> <tt>File *NewFileFromFile(FILE *f)</tt></b></p>
<div class="indent"> Create a File object wrapper around an existing <tt>
FILE *</tt> object.</div>
<p> There's no explicit function to close a file, just call <tt>
Delete(f)</tt> - this decreases the reference count, and the file will
be closed when the reference count reaches zero.</p>
<p> The use of the above I/O functions and strings play a critical role
in SWIG. It is common to see small code fragments of code generated
using code like this:</p>
<div class="code">
<pre>
/* Print into a string */
String *s = NewString(&quot;&quot;);
Printf(s, &quot;Hello\n&quot;);
for (i = 0; i &lt; 10; i++) {
Printf(s, &quot;%d\n&quot;, i);
}
...
/* Print string into a file */
Printf(f, &quot;%s\n&quot;, s);
</pre>
</div>
<p> Similarly, the preprocessor and parser all operate on string-files.</p>
<h2><a name="Extending_nn21">41.6 Navigating and manipulating parse
trees</a></h2>
Parse trees are built as collections of hash tables. Each node is a
hash table in which arbitrary attributes can be stored. Certain
attributes in the hash table provide links to other parse tree nodes.
The following macros can be used to move around the parse tree.
<p><b> <tt>String *nodeType(Node *n)</tt></b></p>
<div class="indent"> Returns the node type tag as a string. The returned
string indicates the type of parse tree node.</div>
<p><b> <tt>Node *nextSibling(Node *n)</tt></b></p>
<div class="indent"> Returns the next node in the parse tree. For
example, the next C declaration.</div>
<p><b> <tt>Node *previousSibling(Node *n)</tt></b></p>
<div class="indent"> Returns the previous node in the parse tree. For
example, the previous C declaration.</div>
<p><b> <tt>Node *firstChild(Node *n)</tt></b></p>
<div class="indent"> Returns the first child node. For example, if <tt>n</tt>
was a C++ class node, this would return the node for the first class
member.</div>
<p><b> <tt>Node *lastChild(Node *n)</tt></b></p>
<div class="indent"> Returns the last child node. You might use this if
you wanted to append a new node to the children of a class.</div>
<p><b> <tt>Node *parentNode(Node *n)</tt></b></p>
<div class="indent"> Returns the parent of node <tt>n</tt>. Use this to
move up the pass tree.</div>
<p> The following macros can be used to change all of the above
attributes. Normally, these functions are only used by the parser.
Changing them without knowing what you are doing is likely to be
dangerous.</p>
<p><b> <tt>void set_nodeType(Node *n, const String_or_char)</tt></b></p>
<div class="indent"> Change the node type. tree node.</div>
<p><b> <tt>void set_nextSibling(Node *n, Node *s)</tt></b></p>
<div class="indent"> Set the next sibling.</div>
<p><b> <tt>void set_previousSibling(Node *n, Node *s)</tt></b></p>
<div class="indent"> Set the previous sibling.</div>
<p><b> <tt>void set_firstChild(Node *n, Node *c)</tt></b></p>
<div class="indent"> Set the first child node.</div>
<p><b> <tt>void set_lastChild(Node *n, Node *c)</tt></b></p>
<div class="indent"> Set the last child node.</div>
<p><b> <tt>void set_parentNode(Node *n, Node *p)</tt></b></p>
<div class="indent"> Set the parent node.</div>
<p> The following utility functions are used to alter the parse tree (at
your own risk)</p>
<p><b> <tt>void appendChild(Node *parent, Node *child)</tt></b></p>
<div class="indent"> Append a child to <tt>parent</tt>. The appended
node becomes the last child.</div>
<p><b> <tt>void deleteNode(Node *node)</tt></b></p>
<div class="indent"> Deletes a node from the parse tree. Deletion
reconnects siblings and properly updates the parent so that sibling
nodes are unaffected.</div>
<h2><a name="Extending_nn22">41.7 Working with attributes</a></h2>
<p> Since parse tree nodes are just hash tables, attributes are accessed
using the <tt>Getattr()</tt>, <tt>Setattr()</tt>, and <tt>Delattr()</tt>
operations. For example:</p>
<div class="code">
<pre>
int functionHandler(Node *n) {
String *name = Getattr(n, &quot;name&quot;);
String *symname = Getattr(n, &quot;sym:name&quot;);
SwigType *type = Getattr(n, &quot;type&quot;);
...
}
</pre>
</div>
<p> New attributes can be freely attached to a node as needed. However,
when new attributes are attached during code generation, they should be
prepended with a namespace prefix. For example:</p>
<div class="code">
<pre>
...
Setattr(n, &quot;python:docstring&quot;, doc); /* Store docstring */
...
</pre>
</div>
<p> A quick way to check the value of an attribute is to use the <tt>
checkAttribute()</tt> function like this:</p>
<div class="code">
<pre>
if (checkAttribute(n, &quot;storage&quot;, &quot;virtual&quot;)) {
/* n is virtual */
...
}
</pre>
</div>
<p> Changing the values of existing attributes is allowed and is
sometimes done to implement node transformations. However, if a
function/method modifies a node, it is required to restore modified
attributes to their original values. To simplify the task of
saving/restoring attributes, the following functions are used:</p>
<p><b> <tt>int Swig_save(const char *ns, Node *n, const char *name1,
const char *name2, ..., NIL)</tt></b></p>
<div class="indent"> Saves a copy of attributes <tt>name1</tt>, <tt>
name2</tt>, etc. from node <tt>n</tt>. Copies of the attributes are
actually resaved in the node in a different namespace which is set by
the <tt>ns</tt> argument. For example, if you call <tt>Swig_save(&quot;foo&quot;,
n, &quot;type&quot;, NIL)</tt>, then the &quot;type&quot; attribute will be copied and
saved as &quot;foo:type&quot;. The namespace name itself is stored in the &quot;view&quot;
attribute of the node. If necessary, this can be examined to find out
where previous values of attributes might have been saved.</div>
<p><b> <tt>int Swig_restore(Node *n)</tt></b></p>
<div class="indent">
<p> Restores the attributes saved by the previous call to <tt>
Swig_save()</tt>. Those attributes that were supplied to <tt>Swig_save()</tt>
will be restored to their original values.</p>
<p> The <tt>Swig_save()</tt> and <tt>Swig_restore()</tt> functions must
always be used as a pair. That is, every call to <tt>Swig_save()</tt>
must have a matching call to <tt>Swig_restore()</tt>. Calls can be
nested if necessary. Here is an example that shows how the functions
might be used:</p>
<div class="code">
<pre>
int variableHandler(Node *n) {
Swig_save(&quot;variableHandler&quot;, n, &quot;type&quot;, &quot;sym:name&quot;, NIL);
String *symname = Getattr(n, &quot;sym:name&quot;);
SwigType *type = Getattr(n, &quot;type&quot;);
...
Append(symname, &quot;_global&quot;); // Change symbol name
SwigType_add_pointer(type); // Add pointer
...
generate wrappers
...
Swig_restore(n); // Restore original values
return SWIG_OK;
}
</pre>
</div></div>
<p><b> <tt>int Swig_require(const char *ns, Node *n, const char *name1,
const char *name2, ..., NIL)</tt></b></p>
<div class="indent"> This is an enhanced version of <tt>Swig_save()</tt>
that adds error checking. If an attribute name is not present in <tt>n</tt>
, a failed assertion results and SWIG terminates with a fatal error.
Optionally, if an attribute name is specified as &quot;*<em>name</em>&quot;, a
copy of the attribute is saved as with <tt>Swig_save()</tt>. If an
attribute is specified as &quot;?<em>name</em>&quot;, the attribute is optional. <tt>
Swig_restore()</tt> must always be called after using this function.</div>
<h2><a name="Extending_nn23">41.8 Type system</a></h2>
<p> SWIG implements the complete C++ type system including typedef,
inheritance, pointers, references, and pointers to members. A detailed
discussion of type theory is impossible here. However, let's cover the
highlights.</p>
<h3><a name="Extending_nn24">41.8.1 String encoding of types</a></h3>
<p> All types in SWIG consist of a base datatype and a collection of
type operators that are applied to the base. A base datatype is almost
always some kind of primitive type such as <tt>int</tt> or <tt>double</tt>
. The operators consist of things like pointers, references, arrays, and
so forth. Internally, types are represented as strings that are
constructed in a very precise manner. Here are some examples:</p>
<div class="diagram">
<pre>
C datatype SWIG encoding (strings)
----------------------------- --------------------------
int &quot;int&quot;
int * &quot;p.int&quot;
const int * &quot;p.q(const).int&quot;
int (*x)(int, double) &quot;p.f(int, double).int&quot;
int [20][30] &quot;a(20).a(30).int&quot;
int (F::*)(int) &quot;m(F).f(int).int&quot;
vector&lt;int&gt; * &quot;p.vector&lt;(int)&gt;&quot;
</pre>
</div>
<p> Reading the SWIG encoding is often easier than figuring out the C
code---just read it from left to right. For a type of &quot;p.f(int,
double).int&quot; is a &quot;pointer to a function(int, double) that returns
int&quot;.</p>
<p> The following operator encodings are used in type strings:</p>
<div class="diagram">
<pre>
Operator Meaning
------------------- -------------------------------
p. Pointer to
a(n). Array of dimension n
r. C++ reference
m(class). Member pointer to class
f(args). Function.
q(qlist). Qualifiers
</pre>
</div>
<p> In addition, type names may be parameterized by templates. This is
represented by enclosing the template parameters in <tt>&lt;( ... )&gt;</tt>.
Variable length arguments are represented by the special base type of <tt>
v(...)</tt>.</p>
<p> If you want to experiment with type encodings, the raw type strings
can be inserted into an interface file using backticks `` wherever a
type is expected. For instance, here is an extremely perverted example:</p>
<div class="diagram">
<pre>
`p.a(10).p.f(int, p.f(int).int)` foo(int, int (*x)(int));
</pre>
</div>
<p> This corresponds to the immediately obvious C declaration:</p>
<div class="diagram">
<pre>
(*(*foo(int, int (*)(int)))[10])(int, int (*)(int));
</pre>
</div>
<p> Aside from the potential use of this declaration on a C programming
quiz, it motivates the use of the special SWIG encoding of types. The
SWIG encoding is much easier to work with because types can be easily
examined, modified, and constructed using simple string operations
(comparison, substrings, concatenation, etc.). For example, in the
parser, a declaration like this</p>
<div class="code">
<pre>
int *a[30];
</pre>
</div>
<p> is processed in a few pieces. In this case, you have the base type &quot;<tt>
int</tt>&quot; and the declarator of type &quot;<tt>a(30).p.</tt>&quot;. To make the
final type, the two parts are just joined together using string
concatenation.</p>
<h3><a name="Extending_nn25">41.8.2 Type construction</a></h3>
<p> The following functions are used to construct types. You should use
these functions instead of trying to build the type strings yourself.</p>
<p><b> <tt>void SwigType_add_pointer(SwigType *ty)</tt></b></p>
<div class="indent"> Adds a pointer to <tt>ty</tt>.</div>
<p><b> <tt>void SwigType_del_pointer(SwigType *ty)</tt></b></p>
<div class="indent"> Removes a single pointer from <tt>ty</tt>.</div>
<p><b> <tt>void SwigType_add_reference(SwigType *ty)</tt></b></p>
<div class="indent"> Adds a reference to <tt>ty</tt>.</div>
<p><b> <tt>void SwigType_add_array(SwigType *ty, const String_or_char
*size)</tt></b></p>
<div class="indent"> Adds an array with dimension <tt>dim</tt> to <tt>ty</tt>
.</div>
<p><b> <tt>void SwigType_del_array(SwigType *ty)</tt></b></p>
<div class="indent"> Removes a single array dimension from <tt>ty</tt>.</div>
<p><b> <tt>int SwigType_array_ndim(SwigType *ty)</tt></b></p>
<div class="indent"> Returns number of array dimensions of <tt>ty</tt>.</div>
<p><b> <tt>String* SwigType_array_getdim(SwigType *ty, int n)</tt></b></p>
<div class="indent"> Returns <tt>n</tt>th array dimension of <tt>ty</tt>
.</div>
<p><b> <tt>void SwigType_array_setdim(SwigType *ty, int n, const
String_or_char *rep)</tt></b></p>
<div class="indent"> Sets <tt>n</tt>th array dimensions of <tt>ty</tt>
to <tt>rep</tt>.</div>
<p><b> <tt>void SwigType_add_qualifier(SwigType *ty, const
String_or_char *q)</tt></b></p>
<div class="indent"> Adds a type qualifier <tt>q</tt> to <tt>ty</tt>. <tt>
q</tt> is typically <tt>&quot;const&quot;</tt> or <tt>&quot;volatile&quot;</tt>.</div>
<p><b> <tt>void SwigType_add_memberpointer(SwigType *ty, const
String_or_char *cls)</tt></b></p>
<div class="indent"> Adds a pointer to a member of class <tt>cls</tt> to
<tt>ty</tt>.</div>
<p><b> <tt>void SwigType_add_function(SwigType *ty, ParmList *p)</tt></b>
</p>
<div class="indent"> Adds a function to <tt>ty</tt>. <tt>p</tt> is a
linked-list of parameter nodes as generated by the parser. See the
section on parameter lists for details about the representation.</div>
<p><b> <tt>void SwigType_add_template(SwigType *ty, ParmList *p)</tt></b>
</p>
<div class="indent"> Adds a template to <tt>ty</tt>. <tt>p</tt> is a
linked-list of parameter nodes as generated by the parser. See the
section on parameter lists for details about the representation.</div>
<p><b> <tt>SwigType *SwigType_pop(SwigType *ty)</tt></b></p>
<div class="indent"> Removes the last type constructor from <tt>ty</tt>
and returns it. <tt>ty</tt> is modified.</div>
<p><b> <tt>void SwigType_push(SwigType *ty, SwigType *op)</tt></b></p>
<div class="indent"> Pushes the type operators in <tt>op</tt> onto type <tt>
ty</tt>. The opposite of <tt>SwigType_pop()</tt>.</div>
<p><b> <tt>SwigType *SwigType_pop_arrays(SwigType *ty)</tt></b></p>
<div class="indent"> Removes all leading array operators from <tt>ty</tt>
and returns them. <tt>ty</tt> is modified. For example, if <tt>ty</tt>
is <tt>&quot;a(20).a(10).p.int&quot;</tt>, then this function would return <tt>
&quot;a(20).a(10).&quot;</tt> and modify <tt>ty</tt> so that it has the value <tt>
&quot;p.int&quot;</tt>.</div>
<p><b> <tt>SwigType *SwigType_pop_function(SwigType *ty)</tt></b></p>
<div class="indent"> Removes a function operator from <tt>ty</tt>
including any qualification. <tt>ty</tt> is modified. For example, if <tt>
ty</tt> is <tt>&quot;f(int).int&quot;</tt>, then this function would return <tt>
&quot;f(int).&quot;</tt> and modify <tt>ty</tt> so that it has the value <tt>&quot;int&quot;</tt>
.</div>
<p><b> <tt>SwigType *SwigType_base(SwigType *ty)</tt></b></p>
<div class="indent"> Returns the base type of a type. For example, if <tt>
ty</tt> is <tt>&quot;p.a(20).int&quot;</tt>, this function would return <tt>&quot;int&quot;</tt>
. <tt>ty</tt> is unmodified.</div>
<p><b> <tt>SwigType *SwigType_prefix(SwigType *ty)</tt></b></p>
<div class="indent"> Returns the prefix of a type. For example, if <tt>
ty</tt> is <tt>&quot;p.a(20).int&quot;</tt>, this function would return <tt>
&quot;p.a(20).&quot;</tt>. <tt>ty</tt> is unmodified.</div>
<h3><a name="Extending_nn26">41.8.3 Type tests</a></h3>
<p> The following functions can be used to test properties of a
datatype.</p>
<p><b> <tt>int SwigType_ispointer(SwigType *ty)</tt></b></p>
<div class="indent"> Checks if <tt>ty</tt> is a standard pointer.</div>
<p><b> <tt>int SwigType_ismemberpointer(SwigType *ty)</tt></b></p>
<div class="indent"> Checks if <tt>ty</tt> is a member pointer.</div>
<p><b> <tt>int SwigType_isreference(SwigType *ty)</tt></b></p>
<div class="indent"> Checks if <tt>ty</tt> is a C++ reference.</div>
<p><b> <tt>int SwigType_isarray(SwigType *ty)</tt></b></p>
<div class="indent"> Checks if <tt>ty</tt> is an array.</div>
<p><b> <tt>int SwigType_isfunction(SwigType *ty)</tt></b></p>
<div class="indent"> Checks if <tt>ty</tt> is a function.</div>
<p><b> <tt>int SwigType_isqualifier(SwigType *ty)</tt></b></p>
<div class="indent"> Checks if <tt>ty</tt> is a qualifier.</div>
<p><b> <tt>int SwigType_issimple(SwigType *ty)</tt></b></p>
<div class="indent"> Checks if <tt>ty</tt> is a simple type. No
operators applied.</div>
<p><b> <tt>int SwigType_isconst(SwigType *ty)</tt></b></p>
<div class="indent"> Checks if <tt>ty</tt> is a const type.</div>
<p><b> <tt>int SwigType_isvarargs(SwigType *ty)</tt></b></p>
<div class="indent"> Checks if <tt>ty</tt> is a varargs type.</div>
<p><b> <tt>int SwigType_istemplate(SwigType *ty)</tt></b></p>
<div class="indent"> Checks if <tt>ty</tt> is a templatized type.</div>
<h3><a name="Extending_nn27">41.8.4 Typedef and inheritance</a></h3>
<p> The behavior of <tt>typedef</tt> declaration is to introduce a type
alias. For instance, <tt>typedef int Integer</tt> makes the identifier <tt>
Integer</tt> an alias for <tt>int</tt>. The treatment of typedef in SWIG
is somewhat complicated due to the pattern matching rules that get
applied in typemaps and the fact that SWIG prefers to generate wrapper
code that closely matches the input to simplify debugging (a user will
see the typedef names used in their program instead of the low-level
primitive C datatypes).</p>
<p> To handle <tt>typedef</tt>, SWIG builds a collection of trees
containing typedef relations. For example,</p>
<div class="code">
<pre>
typedef int Integer;
typedef Integer *IntegerPtr;
typedef int Number;
typedef int Size;
</pre>
</div>
<p> produces two trees like this:</p>
<div class="diagram">
<pre>
int p.Integer
^ ^ ^ ^
/ | \ |
/ | \ |
Integer Size Number IntegerPtr
</pre>
</div>
<p> To resolve a single typedef relationship, the following function is
used:</p>
<p><b> <tt>SwigType *SwigType_typedef_resolve(SwigType *ty)</tt></b></p>
<div class="indent"> Checks if <tt>ty</tt> can be reduced to a new type
via typedef. If so, returns the new type. If not, returns NULL.</div>
<p> Typedefs are only resolved in simple typenames that appear in a
type. For example, the type base name and in function parameters. When
resolving types, the process starts in the leaf nodes and moves up the
tree towards the root. Here are a few examples that show how it works:</p>
<div class="diagram">
<pre>
Original type After typedef_resolve()
------------------------ -----------------------
Integer int
a(30).Integer int
p.IntegerPtr p.p.Integer
p.p.Integer p.p.int
</pre>
</div>
<p> For complicated types, the process can be quite involved. Here is
the reduction of a function pointer:</p>
<div class="diagram">
<pre>
p.f(Integer, p.IntegerPtr, Size).Integer : Start
p.f(Integer, p.IntegerPtr, Size).int
p.f(int, p.IntegerPtr, Size).int
p.f(int, p.p.Integer, Size).int
p.f(int, p.p.int, Size).int
p.f(int, p.p.int, int).int : End
</pre>
</div>
<p> Two types are equivalent if their full type reductions are the same.
The following function will fully reduce a datatype:</p>
<p><b> <tt>SwigType *SwigType_typedef_resolve_all(SwigType *ty)</tt></b></p>
<div class="indent"> Fully reduces <tt>ty</tt> according to typedef
rules. Resulting datatype will consist only of primitive typenames.</div>
<h3><a name="Extending_nn28">41.8.5 Lvalues</a></h3>
<p> When generating wrapper code, it is necessary to emit datatypes that
can be used on the left-hand side of an assignment operator (an
lvalue). However, not all C datatypes can be used in this
way---especially arrays and const-qualified types. To generate a type
that can be used as an lvalue, use the following function:</p>
<p><b> <tt>SwigType *SwigType_ltype(SwigType *ty)</tt></b></p>
<div class="indent"> Converts type <tt>ty</tt> to a type that can be
used as an lvalue in assignment. The resulting type is stripped of
qualifiers and arrays are converted to a pointers.</div>
<p> The creation of lvalues is fully aware of typedef and other aspects
of the type system. Therefore, the creation of an lvalue may result in
unexpected results. Here are a few examples:</p>
<div class="code">
<pre>
typedef double Matrix4[4][4];
Matrix4 x; // type = 'Matrix4', ltype='p.a(4).double'
typedef const char * Literal;
Literal y; // type = 'Literal', ltype='p.char'
</pre>
</div>
<h3><a name="Extending_nn29">41.8.6 Output functions</a></h3>
<p> The following functions produce strings that are suitable for
output.</p>
<p><b> <tt>String *SwigType_str(SwigType *ty, const String_or_char *id =
0)</tt></b></p>
<div class="indent"> Generates a C string for a datatype. <tt>id</tt> is
an optional declarator. For example, if <tt>ty</tt> is &quot;p.f(int).int&quot;
and <tt>id</tt> is &quot;foo&quot;, then this function produces &quot;<tt>int
(*foo)(int)</tt>&quot;. This function is used to convert string-encoded
types back into a form that is valid C syntax.</div>
<p><b> <tt>String *SwigType_lstr(SwigType *ty, const String_or_char *id
= 0)</tt></b></p>
<div class="indent"> This is the same as <tt>SwigType_str()</tt> except
that the result is generated from the type's lvalue (as generated from
SwigType_ltype).</div>
<p><b> <tt>String *SwigType_lcaststr(SwigType *ty, const String_or_char
*id = 0)</tt></b></p>
<div class="indent"> Generates a casting operation that converts from
type <tt>ty</tt> to its lvalue. <tt>id</tt> is an optional name to
include in the cast. For example, if <tt>ty</tt> is &quot;<tt>
q(const).p.char</tt>&quot; and <tt>id</tt> is &quot;<tt>foo</tt>&quot;, this function
produces the string &quot;<tt>(char *) foo</tt>&quot;.</div>
<p><b> <tt>String *SwigType_rcaststr(SwigType *ty, const String_or_char
*id = 0)</tt></b></p>
<div class="indent"> Generates a casting operation that converts from a
type's lvalue to a type equivalent to <tt>ty</tt>. <tt>id</tt> is an
optional name to include in the cast. For example, if <tt>ty</tt> is &quot;<tt>
q(const).p.char</tt>&quot; and <tt>id</tt> is &quot;<tt>foo</tt>&quot;, this function
produces the string &quot;<tt>(const char *) foo</tt>&quot;.</div>
<p><b> <tt>String *SwigType_manglestr(SwigType *ty)</tt></b></p>
<div class="indent"> Generates a mangled string encoding of type <tt>ty</tt>
. The mangled string only contains characters that are part of a valid C
identifier. The resulting string is used in various parts of SWIG, but
is most commonly associated with type-descriptor objects that appear in
wrappers (e.g., <tt>SWIGTYPE_p_double</tt>).</div>
<h2><a name="Extending_nn30">41.9 Parameters</a></h2>
<p> Several type-related functions involve parameter lists. These
include functions and templates. Parameter list are represented as a
list of nodes with the following attributes:</p>
<div class="diagram">
<pre>
&quot;type&quot; - Parameter type (required)
&quot;name&quot; - Parameter name (optional)
&quot;value&quot; - Initializer (optional)
</pre>
</div>
<p> Typically parameters are denoted in the source by using a typename
of <tt>Parm *</tt> or <tt>ParmList *</tt>. To walk a parameter list,
simply use code like this:</p>
<div class="diagram">
<pre>
Parm *parms;
Parm *p;
for (p = parms; p; p = nextSibling(p)) {
SwigType *type = Getattr(p, &quot;type&quot;);
String *name = Getattr(p, &quot;name&quot;);
String *value = Getattr(p, &quot;value&quot;);
...
}
</pre>
</div>
<p> Note: this code is exactly the same as what you would use to walk
parse tree nodes.</p>
<p> An empty list of parameters is denoted by a NULL pointer.</p>
<p> Since parameter lists are fairly common, the following utility
functions are provided to manipulate them:</p>
<p><b> <tt>Parm *CopyParm(Parm *p);</tt></b></p>
<div class="indent"> Copies a single parameter.</div>
<p><b> <tt>ParmList *CopyParmList(ParmList *p);</tt></b></p>
<div class="indent"> Copies an entire list of parameters.</div>
<p><b> <tt>int ParmList_len(ParmList *p);</tt></b></p>
<div class="indent"> Returns the number of parameters in a parameter
list.</div>
<p><b> <tt>String *ParmList_str(ParmList *p);</tt></b></p>
<div class="indent"> Converts a parameter list into a C string. For
example, produces a string like &quot;<tt>(int *p, int n, double x);</tt>&quot;.</div>
<p><b> <tt>String *ParmList_protostr(ParmList *p);</tt></b></p>
<div class="indent"> The same as <tt>ParmList_str()</tt> except that
parameter names are not included. Used to emit prototypes.</div>
<p><b> <tt>int ParmList_numrequired(ParmList *p);</tt></b></p>
<div class="indent"> Returns the number of required (non-optional)
arguments in <tt>p</tt>.</div>
<h2><a name="Extending_nn31">41.10 Writing a Language Module</a></h2>
<p> One of the easiest routes to supporting a new language module is to
copy an already supported language module implementation and modify it.
Be sure to choose a language that is similar in nature to the new
language. All language modules follow a similar structure and this
section briefly outlines the steps needed to create a bare-bones
language module from scratch. Since the code is relatively easy to
read, this section describes the creation of a minimal Python module.
You should be able to extrapolate this to other languages.</p>
<h3><a name="Extending_nn32">41.10.1 Execution model</a></h3>
<p> Code generation modules are defined by inheriting from the <tt>
Language</tt> class, currently defined in the <tt>Source/Modules</tt>
directory of SWIG. Starting from the parsing of command line options,
all aspects of code generation are controlled by different methods of
the <tt>Language</tt> that must be defined by your module.</p>
<h3><a name="Extending_starting_out">41.10.2 Starting out</a></h3>
<p> To define a new language module, first create a minimal
implementation using this example as a guide:</p>
<div class="code">
<pre>
#include &quot;swigmod.h&quot;
class PYTHON : public Language {
public:
virtual void main(int argc, char *argv[]) {
printf(&quot;I'm the Python module.\n&quot;);
}
virtual int top(Node *n) {
printf(&quot;Generating code.\n&quot;);
return SWIG_OK;
}
};
extern &quot;C&quot; Language *
swig_python(void) {
return new PYTHON();
}
</pre>
</div>
<p> The &quot;swigmod.h&quot; header file contains, among other things, the
declaration of the <tt>Language</tt> base class and so you should
include it at the top of your language module's source file. Similarly,
the &quot;swigconfig.h&quot; header file contains some other useful definitions
that you may need. Note that you should<em> not</em> include any header
files that are installed with the target language. That is to say, the
implementation of the SWIG Python module shouldn't have any
dependencies on the Python header files. The wrapper code generated by
SWIG will almost always depend on some language-specific C/C++ header
files, but SWIG itself does not.</p>
<p> Give your language class a reasonable name, usually the same as the
target language. By convention, these class names are all uppercase
(e.g. &quot;PYTHON&quot; for the Python language module) but this is not a
requirement. This class will ultimately consist of a number of
overrides of the virtual functions declared in the <tt>Language</tt>
base class, in addition to any language-specific member functions and
data you need. For now, just use the dummy implementations shown above.</p>
<p> The language module ends with a factory function, <tt>swig_python()</tt>
, that simply returns a new instance of the language class. As shown, it
should be declared with the <tt>extern &quot;C&quot;</tt> storage qualifier so
that it can be called from C code. It should also return a pointer to
the base class (<tt>Language</tt>) so that only the interface (and not
the implementation) of your language module is exposed to the rest of
SWIG.</p>
<p> Save the code for your language module in a file named &quot;<tt>
python.cxx</tt>&quot; and place this file in the <tt>Source/Modules</tt>
directory of the SWIG distribution. To ensure that your module is
compiled into SWIG along with the other language modules, modify the
file <tt>Source/Makefile.am</tt> to include the additional source
files. In addition, modify the file <tt>Source/Modules/swigmain.cxx</tt>
with an additional command line option that activates the module. Read
the source---it's straightforward.</p>
<p> Next, at the top level of the SWIG distribution, re-run the <tt>
autogen.sh</tt> script to regenerate the various build files:</p>
<div class="shell">
<pre>
$ <b>./autogen.sh</b>
</pre>
</div>
<p> Next re-run <tt>configure</tt> to regenerate all of the Makefiles:</p>
<div class="shell">
<pre>
$ <b>./configure</b>
</pre>
</div>
<p> Finally, rebuild SWIG with your module added:</p>
<div class="shell">
<pre>
$ <b>make</b>
</pre>
</div>
<p> Once it finishes compiling, try running SWIG with the command-line
option that activates your module. For example, <tt>swig -python foo.i</tt>
. The messages from your new module should appear.</p>
<h3><a name="Extending_nn34">41.10.3 Command line options</a></h3>
<p> When SWIG starts, the command line options are passed to your
language module. This occurs before any other processing occurs
(preprocessing, parsing, etc.). To capture the command line options,
simply use code similar to this:</p>
<div class="code">
<pre>
void Language::main(int argc, char *argv[]) {
for (int i = 1; i &lt; argc; i++) {
if (argv[i]) {
if (strcmp(argv[i], &quot;-interface&quot;) == 0) {
if (argv[i+1]) {
interface = NewString(argv[i+1]);
Swig_mark_arg(i);
Swig_mark_arg(i+1);
i++;
} else {
Swig_arg_error();
}
} else if (strcmp(argv[i], &quot;-globals&quot;) == 0) {
if (argv[i+1]) {
global_name = NewString(argv[i+1]);
Swig_mark_arg(i);
Swig_mark_arg(i+1);
i++;
} else {
Swig_arg_error();
}
} else if ((strcmp(argv[i], &quot;-proxy&quot;) == 0)) {
proxy_flag = 1;
Swig_mark_arg(i);
} else if (strcmp(argv[i], &quot;-keyword&quot;) == 0) {
use_kw = 1;
Swig_mark_arg(i);
} else if (strcmp(argv[i], &quot;-help&quot;) == 0) {
fputs(usage, stderr);
}
...
}
}
}
</pre>
</div>
<p> The exact set of options depends on what you want to do in your
module. Generally, you would use the options to change code generation
modes or to print diagnostic information.</p>
<p> If a module recognizes an option, it should always call <tt>
Swig_mark_arg()</tt> to mark the option as valid. If you forget to do
this, SWIG will terminate with an unrecognized command line option
error.</p>
<h3><a name="Extending_nn35">41.10.4 Configuration and preprocessing</a></h3>
<p> In addition to looking at command line options, the <tt>main()</tt>
method is responsible for some initial configuration of the SWIG
library and preprocessor. To do this, insert some code like this:</p>
<div class="code">
<pre>
void main(int argc, char *argv[]) {
... command line options ...
/* Set language-specific subdirectory in SWIG library */
SWIG_library_directory(&quot;python&quot;);
/* Set language-specific preprocessing symbol */
Preprocessor_define(&quot;SWIGPYTHON 1&quot;, 0);
/* Set language-specific configuration file */
SWIG_config_file(&quot;python.swg&quot;);
/* Set typemap language (historical) */
SWIG_typemap_lang(&quot;python&quot;);
}
</pre>
</div>
<p> The above code does several things--it registers the name of the
language module with the core, it supplies some preprocessor macro
definitions for use in input files (so that they can determine the
target language), and it registers a start-up file. In this case, the
file <tt>python.swg</tt> will be parsed before any part of the
user-supplied input file.</p>
<p> Before proceeding any further, create a directory for your module in
the SWIG library (The <tt>Lib</tt> directory). Now, create a
configuration file in the directory. For example, <tt>python.swg</tt>.</p>
<p> Just to review, your language module should now consist of two
files-- an implementation file <tt>python.cxx</tt> and a configuration
file <tt>python.swg</tt>.</p>
<h3><a name="Extending_nn36">41.10.5 Entry point to code generation</a></h3>
<p> SWIG is a multi-pass compiler. Once the <tt>main()</tt> method has
been invoked, the language module does not execute again until
preprocessing, parsing, and a variety of semantic analysis passes have
been performed. When the core is ready to start generating wrappers, it
invokes the <tt>top()</tt> method of your language class. The argument
to <tt>top</tt> is a single parse tree node that corresponds to the top
of the entire parse tree.</p>
<p> To get the code generation process started, the <tt>top()</tt>
procedure needs to do several things:</p>
<ul>
<li>Initialize the wrapper code output.</li>
<li>Set the module name.</li>
<li>Emit common initialization code.</li>
<li>Emit code for all of the child nodes.</li>
<li>Finalize the wrapper module and cleanup.</li>
</ul>
<p> An outline of <tt>top()</tt> might be as follows:</p>
<div class="code">
<pre>
int Python::top(Node *n) {
/* Get the module name */
String *module = Getattr(n, &quot;name&quot;);
/* Get the output file name */
String *outfile = Getattr(n, &quot;outfile&quot;);
/* Initialize I/O (see next section) */
...
/* Output module initialization code */
...
/* Emit code for children */
Language::top(n);
...
/* Cleanup files */
...
return SWIG_OK;
}
</pre>
</div>
<h3><a name="Extending_nn37">41.10.6 Module I/O and wrapper skeleton</a></h3>
<!-- please report bugs in this section to mgossage -->
<p> Within SWIG wrappers, there are five main sections. These are (in
order)</p>
<ul>
<li>begin: This section is a placeholder for users to put code at the
beginning of the C/C++ wrapper file.</li>
<li>runtime: This section has most of the common SWIG runtime code.</li>
<li>header: This section holds declarations and inclusions from the .i
file.</li>
<li>wrapper: This section holds all the wrapper code.</li>
<li>init: This section holds the module initialisation function (the
entry point for the interpreter).</li>
</ul>
<p> Different parts of the SWIG code will fill different sections, then
upon completion of the wrappering all the sections will be saved to the
wrapper file.</p>
<p> To perform this will require several additions to the code in
various places, such as:</p>
<div class="code">
<pre>
class PYTHON : public Language {
protected:
/* General DOH objects used for holding the strings */
File *f_begin;
File *f_runtime;
File *f_header;
File *f_wrappers;
File *f_init;
public:
...
};
int Python::top(Node *n) {
...
/* Initialize I/O */
f_begin = NewFile(outfile, &quot;w&quot;, SWIG_output_files());
if (!f_begin) {
FileErrorDisplay(outfile);
Exit(EXIT_FAILURE);
}
f_runtime = NewString(&quot;&quot;);
f_init = NewString(&quot;&quot;);
f_header = NewString(&quot;&quot;);
f_wrappers = NewString(&quot;&quot;);
/* Register file targets with the SWIG file handler */
Swig_register_filebyname(&quot;begin&quot;, f_begin);
Swig_register_filebyname(&quot;header&quot;, f_header);
Swig_register_filebyname(&quot;wrapper&quot;, f_wrappers);
Swig_register_filebyname(&quot;runtime&quot;, f_runtime);
Swig_register_filebyname(&quot;init&quot;, f_init);
/* Output module initialization code */
Swig_banner(f_begin);
...
/* Emit code for children */
Language::top(n);
...
/* Write all to the file */
Dump(f_runtime, f_begin);
Dump(f_header, f_begin);
Dump(f_wrappers, f_begin);
Wrapper_pretty_print(f_init, f_begin);
/* Cleanup files */
Delete(f_runtime);
Delete(f_header);
Delete(f_wrappers);
Delete(f_init);
Delete(f_begin);
return SWIG_OK;
}
</pre>
</div>
<p> Using this to process a file will generate a wrapper file, however
the wrapper will only consist of the common SWIG code as well as any
inline code which was written in the .i file. It does not contain any
wrappers for any of the functions or classes.</p>
<p> The code to generate the wrappers are the various member functions,
which currently have not been touched. We will look at <tt>
functionWrapper()</tt> as this is the most commonly used function. In
fact many of the other wrapper routines will call this to do their
work.</p>
<p> A simple modification to write some basic details to the wrapper
looks like this:</p>
<div class="code">
<pre>
int Python::functionWrapper(Node *n) {
/* Get some useful attributes of this function */
String *name = Getattr(n, &quot;sym:name&quot;);
SwigType *type = Getattr(n, &quot;type&quot;);
ParmList *parms = Getattr(n, &quot;parms&quot;);
String *parmstr= ParmList_str_defaultargs(parms); // to string
String *func = SwigType_str(type, NewStringf(&quot;%s(%s)&quot;, name, parmstr));
String *action = Getattr(n, &quot;wrap:action&quot;);
Printf(f_wrappers, &quot;functionWrapper : %s\n&quot;, func);
Printf(f_wrappers, &quot; action : %s\n&quot;, action);
return SWIG_OK;
}
</pre>
</div>
<p> This will now produce some useful information within your wrapper
file.</p>
<div class="shell">
<pre>
functionWrapper : void delete_Shape(Shape *self)
action : delete arg1;
functionWrapper : void Shape_x_set(Shape *self, double x)
action : if (arg1) (arg1)-&gt;x = arg2;
functionWrapper : double Shape_x_get(Shape *self)
action : result = (double) ((arg1)-&gt;x);
functionWrapper : void Shape_y_set(Shape *self, double y)
action : if (arg1) (arg1)-&gt;y = arg2;
...
</pre>
</div>
<h3><a name="Extending_nn38">41.10.7 Low-level code generators</a></h3>
<!-- please report bugs in this section to mgossage -->
<p> As ingenious as SWIG is, and despite all its capabilities and the
power of its parser, the Low-level code generation takes a lot of work
to write properly. Mainly because every language insists on its own
manner of interfacing to C/C++. To write the code generators you will
need a good understanding of how to manually write an interface to your
chosen language, so make sure you have your documentation handy.</p>
<p> At this point it is also probably a good idea to take a very simple
file (just one function), and try letting SWIG generate wrappers for
many different languages. Take a look at all of the wrappers generated,
and decide which one looks closest to the language you are trying to
wrap. This may help you to decide which code to look at.</p>
<p> In general most language wrappers look a little like this:</p>
<div class="code">
<pre>
/* wrapper for TYPE3 some_function(TYPE1, TYPE2); */
RETURN_TYPE _wrap_some_function(ARGS){
TYPE1 arg1;
TYPE2 arg2;
TYPE3 result;
if(ARG1 is not of TYPE1) goto fail;
arg1=(convert ARG1);
if(ARG2 is not of TYPE2) goto fail;
arg2=(convert ARG2);
result=some_function(arg1, arg2);
convert 'result' to whatever the language wants;
do any tidy up;
return ALL_OK;
fail:
do any tidy up;
return ERROR;
}
</pre>
</div>
<p> Yes, it is rather vague and not very clear. But each language works
differently so this will have to do for now.</p>
<p> Tackling this problem will be done in two stages:</p>
<ul>
<li>The skeleton: the function wrapper, and call, but without the
conversion</li>
<li>The conversion: converting the arguments to-from what the language
wants</li>
</ul>
<p> The first step will be done in the code, the second will be done in
typemaps.</p>
<p> Our first step will be to write the code for <tt>functionWrapper()</tt>
. What is shown below is<b> NOT</b> the solution, merely a step in the
right direction. There are a lot of issues to address.</p>
<ul>
<li>Variable length and default parameters</li>
<li>Typechecking and number of argument checks</li>
<li>Overloaded functions</li>
<li>Inout and Output only arguments</li>
</ul>
<div class="code">
<pre>
virtual int functionWrapper(Node *n) {
/* get useful attributes */
String *name = Getattr(n, &quot;sym:name&quot;);
SwigType *type = Getattr(n, &quot;type&quot;);
ParmList *parms = Getattr(n, &quot;parms&quot;);
...
/* create the wrapper object */
Wrapper *wrapper = NewWrapper();
/* create the wrapper function's name */
String *wname = Swig_name_wrapper(iname);
/* deal with overloading */
....
/* write the wrapper function definition */
Printv(wrapper-&gt;def, &quot;RETURN_TYPE &quot;, wname, &quot;(ARGS) {&quot;, NIL);
/* if any additional local variables are needed, add them now */
...
/* write the list of locals/arguments required */
emit_args(type, parms, wrapper);
/* check arguments */
...
/* write typemaps(in) */
....
/* write constraints */
....
/* Emit the function call */
emit_action(n, wrapper);
/* return value if necessary */
....
/* write typemaps(out) */
....
/* add cleanup code */
....
/* Close the function(ok) */
Printv(wrapper-&gt;code, &quot;return ALL_OK;\n&quot;, NIL);
/* add the failure cleanup code */
...
/* Close the function(error) */
Printv(wrapper-&gt;code, &quot;return ERROR;\n&quot;, &quot;}\n&quot;, NIL);
/* final substitutions if applicable */
...
/* Dump the function out */
Wrapper_print(wrapper, f_wrappers);
/* tidy up */
Delete(wname);
DelWrapper(wrapper);
return SWIG_OK;
}
</pre>
</div>
<p> Executing this code will produce wrappers which have our basic
skeleton but without the typemaps, there is still work to do.</p>
<h3><a name="Extending_configuration_files">41.10.8 Configuration files</a>
</h3>
<!-- please report bugs in this section to ttn -->
<p> At the time of this writing, SWIG supports nearly twenty languages,
which means that for continued sanity in maintaining the configuration
files, the language modules need to follow some conventions. These are
outlined here along with the admission that, yes it is ok to violate
these conventions in minor ways, as long as you know where to apply the
proper kludge to keep the overall system regular and running.
Engineering is the art of compromise, see...</p>
<p> Much of the maintenance regularity depends on choosing a suitable
nickname for your language module (and then using it in a controlled
way). Nicknames should be all lower case letters with an optional
numeric suffix (no underscores, no dashes, no spaces). Some examples
are: <tt>foo</tt>, <tt>bar</tt>, <tt>qux99</tt>.</p>
<p> The numeric suffix variant, as in the last example, is somewhat
tricky to work with because sometimes people expect to refer to the
language without this number but sometimes that number is extremely
relevant (especially when it corresponds to language implementation
versions with incompatible interfaces). New language modules that
unavoidably require a numeric suffix in their nickname should include
that number in all uses, or be prepared to kludge.</p>
<p> The nickname is used in four places:</p>
<table summary="nickname table">
<tr><td><b>usage</b></td><td><b>transform</b></td></tr>
<tr><td>&quot;skip&quot; tag</td><td>(none)</td></tr>
<tr><td>Examples/ subdir name</td><td>(none)</td></tr>
<tr><td>Examples/test-suite/ subdir name</td><td>(none)</td></tr>
<!-- add more uses here (remember to adjust header) -->
</table>
<p> As you can see, most usages are direct.</p>
<dl>
<dt><b> configure.ac</b></dt>
<dd> This file is processed by
<p> <a HREF="https://www.gnu.org/software/autoconf/">autoconf</a> to
generate the <tt>configure</tt> script. This is where you need to add
shell script fragments and autoconf macros to detect the presence of
whatever development support your language module requires, typically
directories where headers and libraries can be found, and/or utility
programs useful for integrating the generated wrapper code.</p>
<p> Use the <tt>AC_ARG_WITH</tt>, <tt>AC_MSG_CHECKING</tt>, <tt>AC_SUBST</tt>
macros and so forth (see other languages for examples). Avoid using the
<tt>[</tt> and <tt>]</tt> character in shell script fragments. The
variable names passed to <tt>AC_SUBST</tt> should begin with the
nickname, entirely upcased.</p>
<p> At the end of the new section is the place to put the aforementioned
nickname kludges (should they be needed). See Perl5 for examples of
what to do. [If this is still unclear after you've read the code, ping
me and I'll expand on this further. --ttn]</p>
</dd>
<dt><b> Makefile.in</b></dt>
<dd>
<p> Some of the variables AC_SUBSTituted are essential to the support of
your language module. Fashion these into a shell script &quot;test&quot; clause
and assign that to a skip tag using &quot;-z&quot; and &quot;-o&quot;:</p>
<div class="code"> <tt>skip-qux99 = [ -z &quot;@QUX99INCLUDE@&quot; -o -z
&quot;@QUX99LIBS&quot; ]</tt></div>
<p> This means if those vars should ever be empty, qux99 support should
be considered absent and so it would be a good idea to skip actions
that might rely on it.</p>
<p> Here is where you may also define an alias (but then you'll need to
kludge --- don't do this):</p>
<div class="code"> <tt>skip-qux = $(skip-qux99)</tt></div>
<p> Lastly, you need to modify each of <tt>check-aliveness</tt>, <tt>
check-examples</tt>, <tt>check-test-suite</tt> and <tt>lib-languages</tt>
(var). Use the nickname for these, not the alias. Note that you can do
this even before you have any tests or examples set up; the Makefile
rules do some sanity checking and skip around these kinds of problems.</p>
</dd>
<dt><b> Examples/Makefile.in</b></dt>
<dd> Nothing special here; see comments at the top of this file and look
to the existing languages for examples.</dd>
<dt><b> Examples/qux99/check.list</b></dt>
<dd> Do <tt>cp ../python/check.list .</tt> and modify to taste. One
subdir per line.</dd>
<dt><b> Lib/qux99/extra-install.list</b></dt>
<dd> If you add your language to the top-level Makefile.in var <tt>
lib-languages</tt>, then <tt>make install</tt> will install all <tt>*.i</tt>
and <tt>*.swg</tt> files from the language-specific subdirectory of <tt>
Lib</tt>. Use (optional) file <tt>extra-install.list</tt> in that
directory to name additional files to install (see ruby for example).</dd>
<dt><b> Source/Modules/Makefile.am</b></dt>
<dd> Add appropriate files to this Automake file. That's it!
<p> When you have modified these files, please make sure that the new
language module is completely ignored if it is not installed and
detected on a box, that is, <tt>make check-examples</tt> and <tt>make
check-test-suite</tt> politely displays the ignoring language message.</p>
</dd>
</dl>
<h3><a name="Extending_nn40">41.10.9 Runtime support</a></h3>
<p> Discuss the kinds of functions typically needed for SWIG runtime
support (e.g. <tt>SWIG_ConvertPtr()</tt> and <tt>SWIG_NewPointerObj()</tt>
) and the names of the SWIG files that implement those functions.</p>
<h3><a name="Extending_nn41">41.10.10 Standard library files</a></h3>
<p> The standard library files that most languages supply keeps growing
as SWIG matures. The following are the minimum that are usually
supported:</p>
<ul>
<li> typemaps.i</li>
<li> std_string.i</li>
<li> std_vector.i</li>
<li> stl.i</li>
</ul>
<p> Please copy these and modify for any new language.</p>
<h3><a name="Extending_nn42">41.10.11 User examples</a></h3>
<p> Each of the language modules provides one or more examples. These
examples are used to demonstrate different features of the language
module to SWIG end-users, but you'll find that they're useful during
development and testing of your language module as well. You can use
examples from the existing SWIG language modules for inspiration.</p>
<p> Each example is self-contained and consists of (at least) a <tt>
Makefile</tt>, a SWIG interface file for the example module, and a
'runme' script that demonstrates the functionality for that module. All
of these files are stored in the same subdirectory under the <tt>
Examples/[lang]</tt> directory. There are two classic examples which
should be the first to convert to a new language module. These are the
&quot;simple&quot; C example and the &quot;class&quot; C++ example. These can be found, for
example for Python, in <tt>Examples/python/simple</tt> and <tt>
Examples/python/class</tt>.</p>
<p> By default, all of the examples are built and run when the user
types <tt>make check</tt>. To ensure that your examples are
automatically run during this process, see the section on <a href="#Extending_configuration_files">
configuration files</a>.</p>
<h3><a name="Extending_test_suite">41.10.12 Test driven development and
the test-suite</a></h3>
<p> A test driven development approach is central to the improvement and
development of SWIG. Most modifications to SWIG are accompanied by
additional regression tests and checking all tests to ensure that no
regressions have been introduced.</p>
<p> The regression testing is carried out by the SWIG<i> test-suite</i>.
The test-suite consists of numerous testcase interface files in the <tt>
Examples/test-suite</tt> directory as well as target language specific
runtime tests in the <tt>Examples/test-suite/[lang]</tt> directory.
When a testcase is run, it will execute the following steps for each
testcase:</p>
<ol>
<li>Execute SWIG passing it the testcase interface file.</li>
<li>Compile the resulting generated C/C++ code with either the C or C++
compiler into object files.</li>
<li>Link the object files into a dynamic library (dll/shared object).</li>
<li>Compile any generated and any runtime test target language code with
the target language compiler, if the target language supports
compilation. This step thus does not apply to the interpreted
languages.</li>
<li>Execute a runtime test if one exists.</li>
</ol>
<p> For example, the<i> ret_by_value</i> testcase consists of two
components. The first component is the <tt>
Examples/test-suite/ret_by_value.i</tt> interface file. The name of the
SWIG module<b> must</b> always be the name of the testcase, so the <tt>
ret_by_value.i</tt> interface file thus begins with:</p>
<div class="code">
<pre>
%module ret_by_value
</pre>
</div>
<p> The testcase code will then follow the module declaration, usually
within a <tt>%inline %{ ... %}</tt> section for the majority of the
tests.</p>
<p> The second component is the optional runtime tests. Any runtime
tests are named using the following convention: <tt>
[testcase]_runme.[ext]</tt>, where <tt>[testcase]</tt> is the testcase
name and <tt>[ext]</tt> is the normal extension for the target language
file. In this case, the Java and Python target languages implement a
runtime test, so their files are respectively, <tt>
Examples/test-suite/java/ret_by_value_runme.java</tt> and <tt>
Examples/test-suite/python/ret_by_value_runme.py</tt>.</p>
<p> The goal of the test-suite is to test as much as possible in a<b>
silent</b> manner. This way any SWIG or compiler errors or warnings are
easily visible. Should there be any warnings, changes must be made to
either fix them (preferably) or suppress them. Compilation or runtime
errors result in a testcase failure and will be immediately visible. It
is therefore essential that the runtime tests are written in a manner
that displays nothing to stdout/stderr on success but error/exception
out with an error message on stderr on failure.</p>
<h4><a name="Extending_running_test_suite">41.10.12.1 Running the
test-suite</a></h4>
<p> In order for the test-suite to work for a particular target
language, the language must be correctly detected and configured during
the configure stage so that the correct Makefiles are generated. Most
development occurs on Linux, so usually it is a matter of installing
the development packages for the target language and simply configuring
as outlined <a href="#Extending_starting_out">earlier</a>.</p>
<p> If when running the test-suite commands that follow, you get a
message that the test was skipped, it indicates that the configure
stage is missing information in order to compile and run everything for
that language.</p>
<p> The test-suite can be run in a number of ways. The first group of
commands are for running multiple testcases in one run and should be
executed in the top level directory. To run the entire test-suite (can
take a long time):</p>
<div class="shell">
<pre>
make -k check-test-suite
</pre>
</div>
<p> To run the test-suite just for target language [lang], replace
[lang] with one of csharp, java, perl5, python, ruby, tcl etc:</p>
<div class="shell">
<pre>
make check-[lang]-test-suite
</pre>
</div>
<p> Note that if a runtime test is available, a message &quot;(with run
test)&quot; is displayed when run. For example:</p>
<div class="shell">
<pre>
$ make check-python-test-suite
checking python test-suite
checking python testcase argcargvtest (with run test)
checking python testcase python_autodoc
checking python testcase python_append (with run test)
checking python testcase callback (with run test)
</pre>
</div>
<p> The files generated on a previous run can be deleted using the clean
targets, either the whole test-suite or for a particular language:</p>
<div class="shell">
<pre>
make clean-test-suite
make clean-[lang]-test-suite
</pre>
</div>
<p> The test-suite can be run in a<i> partialcheck</i> mode where just
SWIG is executed, that is, the compile, link and running of the
testcases is not performed. Note that the partialcheck does not require
the target language to be correctly configured and detected and unlike
the other test-suite make targets, is never skipped. Once again, either
all the languages can be executed or just a chosen language:</p>
<div class="shell">
<pre>
make partialcheck-test-suite
make partialcheck-[lang]-test-suite
</pre>
</div>
<p> If your computer has more than one CPU, you are strongly advised to
use parallel make to speed up the execution speed. This can be done
with any of the make targets that execute more than one testcase. For
example, a dual core processor can efficiently use 2 parallel jobs:</p>
<div class="shell">
<pre>
make -j2 check-test-suite
make -j2 check-python-test-suite
make -j2 partialcheck-java-test-suite
</pre>
</div>
<p> The second group of commands are for running individual testcases
and should be executed in the appropriate target language directory, <tt>
Examples/test-suite/[lang]</tt>. Testcases can contain either C or C++
code and when one is written, a decision must be made as to which of
these input languages is to be used. Replace <tt>[testcase]</tt> in the
commands below with the name of the testcase.</p>
<p> For a C language testcase, add the testcase under the C_TEST_CASES
list in <tt>Examples/test-suite/common.mk</tt> and execute individually
as:</p>
<div class="shell">
<pre>
make -s [testcase].ctest
</pre>
</div>
<p> For a C++ language testcase, add the testcase under the
CPP_TEST_CASES list in <tt>Examples/test-suite/common.mk</tt> and
execute individually as:</p>
<div class="shell">
<pre>
make -s [testcase].cpptest
</pre>
</div>
<p> A third category of tests are C++ language testcases testing
multiple modules (the %import directive). These require more than one
shared library (dll/shared object) to be built and so are separated out
from the normal C++ testcases. Add the testcase under the
MULTI_CPP_TEST_CASES list in <tt>Examples/test-suite/common.mk</tt> and
execute individually as:</p>
<div class="shell">
<pre>
make -s [testcase].multicpptest
</pre>
</div>
<p> To delete the generated files, execute:</p>
<div class="shell">
<pre>
make -s [testcase].clean
</pre>
</div>
<p> If you would like to see the exact commands being executed, drop the
-s option:</p>
<div class="shell">
<pre>
make [testcase].ctest
make [testcase].cpptest
make [testcase].multicpptest
</pre>
</div>
<p> Some real examples of each:</p>
<div class="shell">
<pre>
make -s ret_by_value.clean
make -s ret_by_value.ctest
make -s bools.cpptest
make -s imports.multicpptest
</pre>
</div>
<p> Advanced usage of the test-suite facilitates running tools on some
of the five stages. The make variables <tt>SWIGTOOL</tt> and <tt>
RUNTOOL</tt> are used to specify a tool to respectively, invoke SWIG and
the execution of the runtime test. You are advised to view the <tt>
Examples/test-suite/common.mk</tt> file for details but for a short
summary, the classic usage is to use <a href="https://valgrind.org/">
Valgrind</a> for memory checking. For example, checking for memory leaks
when running the runtime test in the target language interpreter:</p>
<div class="shell">
<pre>
make ret_by_value.ctest RUNTOOL=&quot;valgrind --leak-check=full&quot;
</pre>
</div>
<p> This will probably make more sense if you look at the output of the
above as it will show the exact commands being executed. SWIG can be
analyzed for bad memory by first rebuilding swig with just the <tt>-g</tt>
option. Also define DOH_DEBUG_MEMORY_POOLS, see <a href="#Extending_further_info">
</a> section. SWIG can then be invoked via valgrind using:</p>
<div class="shell">
<pre>
make ret_by_value.ctest SWIGTOOL=&quot;valgrind --tool=memcheck --trace-children=yes&quot;
</pre>
</div>
<p> A debugger can also be invoked easily on an individual test, for
example gdb:</p>
<div class="shell">
<pre>
make ret_by_value.ctest RUNTOOL=&quot;gdb --args&quot;
</pre>
</div>
<p> SWIG reads the <tt>SWIG_FEATURES</tt> environment variable to obtain
options in addition to those passed on the command line. This is
particularly useful as the entire test-suite or a particular testcase
can be run customized by using additional arguments, for example the -O
optimization flag can be added in, as shown below for the bash shell:</p>
<div class="shell">
<pre>
env SWIG_FEATURES=-O make check-python-test-suite
</pre>
</div>
<p> The syntax for setting environment variables varies from one shell
to the next, but it also works as shown in the example below, where
some typemap debugging is added in:</p>
<div class="shell">
<pre>
make ret_by_value.ctest SWIG_FEATURES=&quot;-debug-tmsearch&quot;
</pre>
</div>
<p> There is also a special 'errors' test-suite which is a set of
regression tests checking SWIG warning and error messages. It can be
run in the same way as the other language test-suites, replacing [lang]
with errors, such as <tt>make check-errors-test-suite</tt>. The test
cases used and the way it works is described in <tt>
Examples/test-suite/errors/Makefile.in</tt>.</p>
<h3><a name="Extending_nn43">41.10.13 Documentation</a></h3>
<p> Don't forget to write end-user documentation for your language
module. Currently, each language module has a dedicated chapter You
shouldn't rehash things that are already covered in sufficient detail
in the <a href="#SWIG">SWIG Basics</a> and <a href="#SWIGPlus">SWIG and
C++</a> chapters. There is no fixed format for<em> what</em>, exactly,
you should document about your language module, but you'll obviously
want to cover issues that are unique to your language.</p>
<p> Some topics that you'll want to be sure to address include:</p>
<ul>
<li> Command line options unique to your language module.</li>
<li> Non-obvious mappings between C/C++ and target language concepts.
For example, if your target language provides a single floating point
type, it should be no big surprise to find that C/C++ <tt>float</tt>
and <tt>double</tt> types are mapped to it. On the other hand, if your
target language doesn't provide support for &quot;classes&quot; or something
similar, you'd want to discuss how C++ classes are handled.</li>
<li> How to compile the SWIG-generated wrapper code into shared
libraries that can actually be used. For some languages, there are
well-defined procedures for doing this, but for others it's an ad hoc
process. Provide as much detail as appropriate, and links to other
resources if available.</li>
</ul>
<h3><a name="Extending_coding_style_guidelines">41.10.14 Coding style
guidelines</a></h3>
<p> The coding guidelines for the C/C++ source code are pretty much K&amp;R
C style. The style can be inferred from the existing code base and is
largely dictated by the <tt>indent</tt> code beautifier tool set to K&amp;R
style. The code can formatted using the make targets in the Source
directory. Below is an example of how to format the emit.cxx file:</p>
<blockquote>
<pre>
$ cd Source
$ make beautify-file INDENTFILE=Modules/emit.cxx
</pre>
</blockquote>
<p> Of particular note is indentation is set to 2 spaces and a tab is
used instead of 8 spaces. The generated C/C++ code should also follow
this style as close as possible. However, tabs should be avoided as
unlike the SWIG developers, users will never have consistent tab
settings.</p>
<h3><a name="Extending_language_status">41.10.15 Target language status</a>
</h3>
<p> Target languages are given a status of either 'Supported',
'Experimental' or 'Deprecated' depending on their maturity as broadly
outlined in the <a href="#Introduction_target_languages">Target
language introduction</a>. This section provides more details on how
this status is given.</p>
<h4><a name="Extending_supported_status">41.10.15.1 Supported status</a></h4>
<p> A target language is given the 'Supported' status when</p>
<ul>
<li> It is in a mature, well functioning state.</li>
<li> It has its own comprehensive chapter in the documentation. The
level of documentation should be comprehensive and match the standard
of the other mature modules. Python and Java are good references.</li>
<li> It passes all of the main SWIG test-suite. The main test-suite is
defined by the tests in the C_TEST_CASES, CPP_TEST_CASES and
MULTI_CPP_TEST_CASES lists in Examples/test-suite/common.mk. All the
newer C++ standard tests need to work and are grouped together, such as
CPP11_TEST_CASES for C++11. These more 'modern' C++ standards are only
tested though if the compiler is detected as supporting the given
standard.</li>
<li> The test-suite must also include at least twenty wide-ranging
runtime tests. The most mature languages have a few hundred runtime
tests. Note that porting runtime tests from another language module is
a quick and easy way to achieve this.</li>
<li> It supports the vast majority of SWIG features. Some more advanced
features, such as, directors, full nested class support and target
language namespaces (nspace) may be unimplemented. A few support
libraries may be missing, for example, a small number of STL libraries.</li>
<li> It provides strong backwards compatibility between releases. Each
point release must aim to be fully backwards compatible. A point
release version is the 3rd version digit, so each of the x.y.* versions
should be backwards compatible. Backwards compatibility breakages can
occur in a new major or minor version if absolutely necessary and if
documented. A major or minor version is the first or second digit in
the three digit version.</li>
<li> Fixing unintended regressions in the Supported languages will be
given higher priority over the Experimental languages by the core SWIG
developers.</li>
<li> Examples must be available and run successfully.</li>
<li> The examples and test-suite must be fully functioning on the Github
Actions Continuous Integration platform.</li>
</ul>
<h4><a name="Extending_experimental_status">41.10.15.2 Experimental
status</a></h4>
<p> A target language is given the 'Experimental' status when</p>
<ul>
<li> It is of sub-standard quality, failing to meet the above
'Supported' status.</li>
<li> It is somewhere between the mid to mature stage of development.</li>
<li> It is in need of help to finish development.</li>
</ul>
<p> Some minimum requirements and notes about languages with the
'Experimental' status:</p>
<ul>
<li> Will at least implement basic functionality - support wrapping C
functions and simple C++ classes and templates.</li>
<li> Have its own documentation chapter containing a reasonable level of
detail. The documentation must provide enough basic functionality for a
user to get started.</li>
<li> Have fully functional examples of basic functionality (the simple
and class examples).</li>
<li> The test-suite must be implemented and include a few runtime tests
for both C and C++ test cases.</li>
<li> Failing tests must be put into one of the FAILING_CPP_TESTS or
FAILING_C_TESTS lists in the test-suite. This will ensure the
test-suite can be superficially made to pass by ignoring failing tests.
The number of tests in these lists should be no greater than half of
the number of tests in the full test-suite.</li>
<li> The examples and test-suite must also be fully functioning on the
Github Actions Continuous Integration platform. However, experimental
languages will be flagged as 'continue-on-error'. This means that pull
requests and normal development commits will not break the entire
Github Actions build should an experimental language fail.</li>
<li> Any new failed tests will be fixed on a 'best effort' basis by core
developers with no promises made.</li>
<li> If a language module has an official maintainer, then the
maintainer will be requested to focus on fixing test-suite regressions
and commit to migrating the module to become a 'Supported' module.</li>
<li> If a module does not have an official maintainer, then, as
maintenance will be on a 'best efforts' basis by the core maintainers,
no guarantees will be provided from one release to the next and
regressions may creep in.</li>
<li> Experimental target languages will have a (suppressible) warning
explaining the Experimental sub-standard status and encourage users to
help improve it.</li>
<li> No backwards compatibility is guaranteed as the module is
effectively 'in development'. If a language module has an official
maintainer, then a backwards compatibility guarantee may be provided at
the maintainer's discretion and should be documented as such.</li>
</ul>
<h4><a name="Extending_deprecated_status">41.10.15.3 Deprecated status</a>
</h4>
<p> Unfortunately target languages that once met 'Experimental' or
'Supported' status can become non-functional and simply bit rot due to
neglect or due to the language's C/C++ API evolving and changing over
time. If there is no language maintainer and the maintenance burden
becomes too much for the core SWIG developers to keep the test-suite
working with easily available versions of the target language, the
language is put into the 'Deprecated' status.</p>
<p> Changing the status to 'Deprecated' is an unfortunate step not done
lightly and only if recent versions cannot successfully run the
test-suite and examples on the Github Actions Continuous Integration
platform. As the target language would once have had a working set of
examples and test-suite, this usually only occurs when the more modern
operating system distributions available on Github can no longer run
any distribution supplied version of the target language.</p>
<p> Changing status to 'Deprecated' flags it for removal from SWIG in a
subsequent release (usually one SWIG release). This step becomes the
final plea for help from the community who use the target language. The
language will need updating by an interested community member to meet
the requirements of at least 'Experimental' status in order to prevent
removal. If you are a user of a 'Deprecated' target language and would
like to keep it available in future releases, please contact the SWIG
developers for details of how you can help.</p>
<h3><a name="Extending_prerequisites">41.10.16 Prerequisites for adding
a new language module to the SWIG distribution</a></h3>
<p> New target language modules can be included in SWIG and
contributions are encouraged for popular languages. In order to be
considered for inclusion, a language must at a minimum fit the
'Experimental' status described above.</p>
<p> Below are some practical steps that should help meet these
requirements.</p>
<ol>
<li> The &quot;simple&quot; example needs to be working to demonstrate basic C
code wrappers. Port the example from another language, such as from <tt>
Examples/python/simple</tt>.</li>
<li> The &quot;class&quot; example needs to be working to demonstrate basic C++
code wrappers. Port the example from another language, such as from <tt>
Examples/python/class</tt>.</li>
<li> Modify <tt>configure.ac</tt>, <tt>Makefile.in</tt> and <tt>
Examples/Makefile.in</tt> to run these examples. Please make sure that
if the new language is not installed properly on a box, <tt>make -k
check</tt> should still work by skipping the tests and examples for the
new language module.</li>
<li> Copying an existing language module and adapting the source for it
is likely to be the most efficient approach to fully developing a new
module as a number of corner cases are covered in the existing
implementations. The most advanced scripting languages are Python and
Ruby. The most advanced compiled target languages are Java and C#.</li>
<li> Get the <a href="#Extending_running_test_suite">test-suite</a>
running for the new language (<tt>make check-[lang]-test-suite</tt>).
While the test-suite tests many corner cases, we'd expect the majority
of it to work without much effort once the generated code is compiling
correctly for basic functionality as most of the corner cases are
covered in the SWIG core. Aim to first get one C and one C++ runtime
test running in the test-suite. Adding further runtime tests should be
a lot easier afterwards by porting existing runtime tests from another
language module.</li>
<li> The structure and contents of the html documentation chapter can be
copied and adapted from one of the other language modules.</li>
<li> Source code can be formatted correctly using the info in the <a href="#Extending_coding_style_guidelines">
coding style guidelines</a> section.</li>
<li> When ready, post a patch on Github, join the swig-devel mailing
list and email the SWIG developers with a demonstration of commitment
to maintaining the language module, certainly in the short term and
ideally long term.</li>
</ol>
<p> Once accepted into the official Git repository, development efforts
should concentrate on getting the entire test-suite to work in order to
migrate the language module to the 'Supported' status. Runtime tests
should be added for existing testcases and new test cases can be added
should there be an area not already covered by the existing tests.</p>
<h2><a name="Extending_debugging_options">41.11 Debugging Options</a></h2>
<p> There are various command line options which can aid debugging a
SWIG interface as well as debugging the development of a language
module. These are as follows:</p>
<div class="shell">
<pre>
-debug-classes - Display information about the classes found in the interface
-debug-module &lt;n&gt; - Display module parse tree at stages 1-4, &lt;n&gt; is a csv list of stages
-debug-symtabs - Display symbol tables information
-debug-symbols - Display target language symbols in the symbol tables
-debug-csymbols - Display C symbols in the symbol tables
-debug-lsymbols - Display target language layer symbols
-debug-quiet - Display less parse tree node debug info when using other -debug options
-debug-tags - Display information about the tags found in the interface
-debug-template - Display information for debugging templates
-debug-top &lt;n&gt; - Display entire parse tree at stages 1-4, &lt;n&gt; is a csv list of stages
-debug-typedef - Display information about the types and typedefs in the interface
-debug-typemap - Display information for debugging typemaps
-debug-tmsearch - Display typemap search debugging information
-debug-tmused - Display typemaps used debugging information
</pre>
</div>
<p> The complete list of command line options for SWIG are available by
running <tt>swig -help</tt>.</p>
<h2><a name="Extending_nn46">41.12 Guide to parse tree nodes</a></h2>
<p> This section describes the different parse tree nodes and their
attributes.</p>
<p><b> cdecl</b></p>
<p> Describes general C declarations including variables, functions, and
typedefs. A declaration is parsed as &quot;storage T D&quot; where storage is a
storage class, T is a base type, and D is a declarator.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Declarator name
&quot;type&quot; - Base type T
&quot;decl&quot; - Declarator type (abstract)
&quot;storage&quot; - Storage class (static, extern, typedef, etc.)
&quot;parms&quot; - Function parameters (if a function)
&quot;code&quot; - Function body code (if supplied)
&quot;value&quot; - Default value (if supplied)
</pre>
</div>
<p><b> constructor</b></p>
<p> C++ constructor declaration.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Name of constructor
&quot;parms&quot; - Parameters
&quot;decl&quot; - Declarator (function with parameters)
&quot;code&quot; - Function body code (if any)
&quot;feature:new&quot; - Set to indicate return of new object.
</pre>
</div>
<p><b> destructor</b></p>
<p> C++ destructor declaration.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Name of destructor
&quot;code&quot; - Function body code (if any)
&quot;storage&quot; - Storage class (set if virtual)
&quot;value&quot; - Default value (set if pure virtual).
</pre>
</div>
<p><b> access</b></p>
<p> C++ access change.</p>
<div class="diagram">
<pre>
&quot;kind&quot; - public, protected, private
</pre>
</div>
<p><b> constant</b></p>
<p> Constant created by %constant or #define.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Name of constant.
&quot;type&quot; - Base type.
&quot;value&quot; - Value.
&quot;storage&quot; - Set to %constant
&quot;feature:immutable&quot; - Set to indicate read-only
</pre>
</div>
<p><b> class</b></p>
<p> C++ class definition or C structure definition.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Name of the class.
&quot;kind&quot; - Class kind (&quot;struct&quot;, &quot;union&quot;, &quot;class&quot;)
&quot;symtab&quot; - Enclosing symbol table.
&quot;tdname&quot; - Typedef name. Use for typedef struct { ... } A.
&quot;abstract&quot; - Set if class has pure virtual methods.
&quot;baselist&quot; - List of base class names.
&quot;storage&quot; - Storage class (if any)
&quot;unnamed&quot; - Set if class is unnamed.
</pre>
</div>
<p><b> enum</b></p>
<p> Enumeration.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Name of the enum (if supplied).
&quot;storage&quot; - Storage class (if any)
&quot;tdname&quot; - Typedef name (typedef enum { ... } name).
&quot;unnamed&quot; - Set if enum is unnamed.
</pre>
</div>
<p><b> enumitem</b></p>
<p> Enumeration value.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Name of the enum value.
&quot;type&quot; - Type (integer or char)
&quot;value&quot; - Enum value (if given)
&quot;feature:immutable&quot; - Set to indicate read-only
</pre>
</div>
<p><b> namespace</b></p>
<p> C++ namespace.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Name of the namespace.
&quot;symtab&quot; - Symbol table for enclosed scope.
&quot;unnamed&quot; - Set if unnamed namespace
&quot;alias&quot; - Alias name. Set for namespace A = B;
</pre>
</div>
<p><b> using</b></p>
<p> C++ using directive.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Name of the object being referred to.
&quot;uname&quot; - Qualified name actually given to using.
&quot;node&quot; - Node being referenced.
&quot;namespace&quot; - Namespace name being reference (using namespace name)
</pre>
</div>
<p><b> classforward</b></p>
<p> A forward C++ class declaration.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Name of the class.
&quot;kind&quot; - Class kind (&quot;union&quot;, &quot;struct&quot;, &quot;class&quot;)
</pre>
</div>
<p><b> insert</b></p>
<p> Code insertion directive. For example, %{ ... %} or
%insert(section).</p>
<div class="diagram">
<pre>
&quot;code&quot; - Inserted code
&quot;section&quot; - Section name (&quot;header&quot;, &quot;wrapper&quot;, etc.)
</pre>
</div>
<p><b> top</b></p>
<p> Top of the parse tree.</p>
<div class="diagram">
<pre>
&quot;module&quot; - Module name
</pre>
</div>
<p><b> extend</b></p>
<p> %extend directive.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Module name
&quot;symtab&quot; - Symbol table of enclosed scope.
</pre>
</div>
<p><b> apply</b></p>
<p> %apply pattern { patternlist }.</p>
<div class="diagram">
<pre>
&quot;pattern&quot; - Source pattern.
&quot;symtab&quot; - Symbol table of enclosed scope.
</pre>
</div>
<p><b> clear</b></p>
<p> %clear patternlist;</p>
<div class="diagram">
<pre>
&quot;firstChild&quot; - Patterns to clear
</pre>
</div>
<p><b> include</b></p>
<p> %include directive.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Filename
&quot;firstChild&quot; - Children
</pre>
</div>
<p><b> import</b></p>
<p> %import directive.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Filename
&quot;firstChild&quot; - Children
</pre>
</div>
<p><b> module</b></p>
<p> %module directive.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Name of the module
</pre>
</div>
<p><b> typemap</b></p>
<p> %typemap directive.</p>
<div class="diagram">
<pre>
&quot;method&quot; - Typemap method name.
&quot;code&quot; - Typemap code.
&quot;kwargs&quot; - Keyword arguments (if any)
&quot;firstChild&quot; - Typemap patterns
</pre>
</div>
<p><b> typemapcopy</b></p>
<p> %typemap directive with copy.</p>
<div class="diagram">
<pre>
&quot;method&quot; - Typemap method name.
&quot;pattern&quot; - Typemap source pattern.
&quot;firstChild&quot; - Typemap patterns
</pre>
</div>
<p><b> typemapitem</b></p>
<p> %typemap pattern. Used with %apply, %clear, %typemap.</p>
<div class="diagram">
<pre>
&quot;pattern&quot; - Typemap pattern (a parameter list)
&quot;parms&quot; - Typemap parameters.
</pre>
</div>
<p><b> types</b></p>
<p> %types directive.</p>
<div class="diagram">
<pre>
&quot;parms&quot; - List of parameter types.
&quot;convcode&quot; - Code which replaces the default casting / conversion code
</pre>
</div>
<p><b> extern</b></p>
<p> extern &quot;X&quot; { ... } declaration.</p>
<div class="diagram">
<pre>
&quot;name&quot; - Name &quot;C&quot;, &quot;Fortran&quot;, etc.
</pre>
</div>
<h2><a name="Extending_further_info">41.13 Further Development
Information</a></h2>
<p> There is further documentation available on the internals of SWIG,
API documentation and debugging information. This is shipped with SWIG
in the <tt>Doc/Devel</tt> directory.</p>
</BODY>
</HTML>