Commit | Line | Data |
---|---|---|
86530b38 AT |
1 | <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> |
2 | <html> | |
3 | <head> | |
4 | <title>SWIG Basics</title> | |
5 | <link rel="stylesheet" type="text/css" href="style.css"> | |
6 | </head> | |
7 | ||
8 | <body bgcolor="#ffffff"> | |
9 | <H1><a name="SWIG"></a>5 SWIG Basics</H1> | |
10 | <!-- INDEX --> | |
11 | <div class="sectiontoc"> | |
12 | <ul> | |
13 | <li><a href="#SWIG_nn2">Running SWIG</a> | |
14 | <ul> | |
15 | <li><a href="#SWIG_nn3">Input format</a> | |
16 | <li><a href="#output">SWIG Output</a> | |
17 | <li><a href="#SWIG_nn5">Comments</a> | |
18 | <li><a href="#SWIG_nn6">C Preprocessor</a> | |
19 | <li><a href="#SWIG_nn7">SWIG Directives</a> | |
20 | <li><a href="#SWIG_nn8">Parser Limitations</a> | |
21 | </ul> | |
22 | <li><a href="#SWIG_nn9">Wrapping Simple C Declarations</a> | |
23 | <ul> | |
24 | <li><a href="#SWIG_nn10">Basic Type Handling</a> | |
25 | <li><a href="#SWIG_nn11">Global Variables</a> | |
26 | <li><a href="#SWIG_nn12">Constants</a> | |
27 | <li><a href="#SWIG_nn13">A brief word about <tt>const</tt></a> | |
28 | <li><a href="#SWIG_nn14">A cautionary tale of <tt>char *</tt></a> | |
29 | </ul> | |
30 | <li><a href="#SWIG_nn15">Pointers and complex objects</a> | |
31 | <ul> | |
32 | <li><a href="#SWIG_nn16">Simple pointers</a> | |
33 | <li><a href="#SWIG_nn17">Run time pointer type checking</a> | |
34 | <li><a href="#SWIG_nn18">Derived types, structs, and classes</a> | |
35 | <li><a href="#SWIG_nn19">Undefined datatypes</a> | |
36 | <li><a href="#SWIG_nn20">Typedef</a> | |
37 | </ul> | |
38 | <li><a href="#SWIG_nn21">Other Practicalities</a> | |
39 | <ul> | |
40 | <li><a href="#SWIG_nn22">Passing structures by value</a> | |
41 | <li><a href="#SWIG_nn23">Return by value</a> | |
42 | <li><a href="#SWIG_nn24">Linking to structure variables</a> | |
43 | <li><a href="#SWIG_nn25">Linking to <tt>char *</tt></a> | |
44 | <li><a href="#SWIG_nn26">Arrays</a> | |
45 | <li><a href="#SWIG_readonly_variables">Creating read-only variables</a> | |
46 | <li><a href="#SWIG_nn28">Renaming and ignoring declarations</a> | |
47 | <li><a href="#SWIG_default_args">Default/optional arguments</a> | |
48 | <li><a href="#SWIG_nn30">Pointers to functions and callbacks</a> | |
49 | </ul> | |
50 | <li><a href="#SWIG_nn31">Structures and unions</a> | |
51 | <ul> | |
52 | <li><a href="#SWIG_nn32">Typedef and structures</a> | |
53 | <li><a href="#SWIG_nn33">Character strings and structures</a> | |
54 | <li><a href="#SWIG_nn34">Array members</a> | |
55 | <li><a href="#SWIG_nn35">Structure data members</a> | |
56 | <li><a href="#SWIG_nn36">C constructors and destructors </a> | |
57 | <li><a href="#SWIG_adding_member_functions">Adding member functions to C structures</a> | |
58 | <li><a href="#SWIG_nn38">Nested structures</a> | |
59 | <li><a href="#SWIG_nn39">Other things to note about structure wrapping</a> | |
60 | </ul> | |
61 | <li><a href="#SWIG_nn40">Code Insertion</a> | |
62 | <ul> | |
63 | <li><a href="#SWIG_nn41">The output of SWIG</a> | |
64 | <li><a href="#SWIG_nn42">Code insertion blocks</a> | |
65 | <li><a href="#SWIG_nn43">Inlined code blocks</a> | |
66 | <li><a href="#SWIG_nn44">Initialization blocks</a> | |
67 | </ul> | |
68 | <li><a href="#SWIG_nn45">An Interface Building Strategy</a> | |
69 | <ul> | |
70 | <li><a href="#SWIG_nn46">Preparing a C program for SWIG</a> | |
71 | <li><a href="#SWIG_nn47">The SWIG interface file</a> | |
72 | <li><a href="#SWIG_nn48">Why use separate interface files?</a> | |
73 | <li><a href="#SWIG_nn49">Getting the right header files</a> | |
74 | <li><a href="#SWIG_nn50">What to do with main()</a> | |
75 | </ul> | |
76 | </ul> | |
77 | </div> | |
78 | <!-- INDEX --> | |
79 | ||
80 | ||
81 | ||
82 | <p> | |
83 | This chapter describes the basic operation of SWIG, the structure of its | |
84 | input files, and how it handles standard ANSI C declarations. C++ support is | |
85 | described in the next chapter. However, C++ programmers should still read this | |
86 | chapter to understand the basics. | |
87 | Specific details about each target language are described in later | |
88 | chapters. | |
89 | </p> | |
90 | ||
91 | <H2><a name="SWIG_nn2"></a>5.1 Running SWIG</H2> | |
92 | ||
93 | ||
94 | <p> | |
95 | To run SWIG, use the <tt>swig</tt> command with one or more of the | |
96 | following options and a filename like this: | |
97 | </p> | |
98 | ||
99 | <div class="shell"><pre> | |
100 | swig [ <em>options</em> ] filename | |
101 | ||
102 | -chicken Generate CHICKEN wrappers | |
103 | -csharp Generate C# wrappers | |
104 | -guile Generate Guile wrappers | |
105 | -java Generate Java wrappers | |
106 | -mzscheme Generate Mzscheme wrappers | |
107 | -ocaml Generate Ocaml wrappers | |
108 | -perl Generate Perl wrappers | |
109 | -php Generate PHP wrappers | |
110 | -pike Generate Pike wrappers | |
111 | -python Generate Python wrappers | |
112 | -ruby Generate Ruby wrappers | |
113 | -sexp Generate Lisp S-Expressions wrappers | |
114 | -tcl Generate Tcl wrappers | |
115 | -xml Generate XML wrappers | |
116 | -c++ Enable C++ parsing | |
117 | -D<em>symbol</em> Define a preprocessor symbol | |
118 | -Fstandard Display error/warning messages in commonly used format | |
119 | -Fmicrosoft Display error/warning messages in Microsoft format | |
120 | -help Display all options | |
121 | -I<em>dir</em> Add a directory to the file include path | |
122 | -l<em>file</em> Include a SWIG library file. | |
123 | -module <em>name</em> Set the name of the SWIG module | |
124 | -o <em>outfile</em> Name of output file | |
125 | -outdir <em>dir</em> Set language specific files output directory | |
126 | -swiglib Show location of SWIG library | |
127 | -version Show SWIG version number | |
128 | ||
129 | </pre></div> | |
130 | <p> | |
131 | This is a subset of commandline options. | |
132 | Additional options are also defined for each target language. A full list | |
133 | can be obtained by typing <tt>swig -help</tt> or <tt>swig | |
134 | -<em>lang</em> -help</tt>. | |
135 | </p> | |
136 | ||
137 | <H3><a name="SWIG_nn3"></a>5.1.1 Input format</H3> | |
138 | ||
139 | ||
140 | <p> | |
141 | As input, SWIG expects a file containing ANSI C/C++ declarations and | |
142 | special SWIG directives. More often than not, this is a special SWIG | |
143 | interface file which is usually denoted with a special <tt>.i</tt> or | |
144 | <tt>.swg</tt> suffix. In certain cases, SWIG can be used directly on | |
145 | raw header files or source files. However, this is not the most | |
146 | typical case and there are several reasons why you might not want to | |
147 | do this (described later). | |
148 | </p> | |
149 | ||
150 | <p> | |
151 | The most common format of a SWIG interface is as follows: | |
152 | </p> | |
153 | ||
154 | <div class="code"><pre> | |
155 | %module mymodule | |
156 | %{ | |
157 | #include "myheader.h" | |
158 | %} | |
159 | // Now list ANSI C/C++ declarations | |
160 | int foo; | |
161 | int bar(int x); | |
162 | ... | |
163 | </pre></div> | |
164 | <p> | |
165 | The name of the module is supplied using the special <tt>%module</tt> | |
166 | directive (or the <tt>-module</tt> command line option). This | |
167 | directive must appear at the beginning of the file and is used to name | |
168 | the resulting extension module (in addition, this name often defines | |
169 | a namespace in the target language). If the module name is supplied on the | |
170 | command line, it overrides the name specified with the | |
171 | <tt>%module</tt> directive. | |
172 | </p> | |
173 | ||
174 | <p> | |
175 | Everything in the <tt>%{ ... %}</tt> block is simply copied verbatim | |
176 | to the resulting wrapper file created by SWIG. This section is almost | |
177 | always used to include header files and other declarations that are | |
178 | required to make the generated wrapper code compile. It is important | |
179 | to emphasize that just because you include a declaration in a SWIG | |
180 | input file, that declaration does <em>not</em> automatically appear in | |
181 | the generated wrapper code---therefore you need to make sure you | |
182 | include the proper header files in the <tt>%{ ... %}</tt> section. It | |
183 | should be noted that the text enclosed in <tt>%{ ... %}</tt> is not | |
184 | parsed or interpreted by SWIG. The <tt>%{...%}</tt> syntax and | |
185 | semantics in SWIG is analogous to that of the declarations section | |
186 | used in input files to parser generation tools such as yacc or bison. | |
187 | </p> | |
188 | ||
189 | <H3><a name="output"></a>5.1.2 SWIG Output</H3> | |
190 | ||
191 | ||
192 | <p> | |
193 | The output of SWIG is a C/C++ file that contains all of the wrapper | |
194 | code needed to build an extension module. SWIG may generate some | |
195 | additional files depending on the target language. By default, an input file | |
196 | with the name <tt>file.i</tt> is transformed into a file | |
197 | <tt>file_wrap.c</tt> or <tt>file_wrap.cxx</tt> (depending on whether | |
198 | or not the <tt>-c++</tt> option has been used). The name of the | |
199 | output file can be changed using the <tt>-o</tt> option. In certain | |
200 | cases, file suffixes are used by the compiler to determine the source | |
201 | language (C, C++, etc.). Therefore, you have to use the | |
202 | <tt>-o</tt> option to change the suffix of the SWIG-generated wrapper | |
203 | file if you want something different than the default. For example: | |
204 | </p> | |
205 | ||
206 | <div class="shell"><pre> | |
207 | $ swig -c++ -python -o example_wrap.cpp example.i | |
208 | </pre></div> | |
209 | ||
210 | <p> | |
211 | The C/C++ output file created by SWIG often | |
212 | contains everything that is needed to construct a extension module | |
213 | for the target scripting language. SWIG is not a stub compiler nor is it | |
214 | usually necessary to edit the output file (and if you look at the output, | |
215 | you probably won't want to). To build the final extension module, the | |
216 | SWIG output file is compiled and linked with the rest of your C/C++ | |
217 | program to create a shared library. | |
218 | </p> | |
219 | ||
220 | <p> | |
221 | Many target languages will also generate proxy class files in the | |
222 | target language. The default output directory for these language | |
223 | specific files is the same directory as the generated C/C++ file. This can | |
224 | can be modified using the <tt>-outdir</tt> option. For example: | |
225 | </p> | |
226 | ||
227 | <div class="shell"><pre> | |
228 | $ swig -c++ -python -outdir pyfiles -o cppfiles/example_wrap.cpp example.i | |
229 | </pre></div> | |
230 | <p> | |
231 | If the directories <tt>cppfiles</tt> and <tt>pyfiles</tt> exist, the following | |
232 | will be generated:</p> | |
233 | <div class="shell"><pre> | |
234 | cppfiles/example_wrap.cpp | |
235 | pyfiles/example.py | |
236 | </pre></div> | |
237 | ||
238 | <H3><a name="SWIG_nn5"></a>5.1.3 Comments</H3> | |
239 | ||
240 | ||
241 | <p> | |
242 | C and C++ style comments may appear anywhere in interface files. In | |
243 | previous versions of SWIG, comments were used to generate | |
244 | documentation files. However, this feature is currently under repair | |
245 | and will reappear in a later SWIG release. | |
246 | </p> | |
247 | ||
248 | <H3><a name="SWIG_nn6"></a>5.1.4 C Preprocessor</H3> | |
249 | ||
250 | ||
251 | <p> | |
252 | Like C, SWIG preprocesses all input files through an enhanced version | |
253 | of the C preprocessor. All standard preprocessor features are | |
254 | supported including file inclusion, conditional compilation and | |
255 | macros. However, <tt>#include</tt> statements are ignored unless the | |
256 | <tt>-includeall</tt> command line option has been supplied. The | |
257 | reason for disabling includes is that SWIG is sometimes used to | |
258 | process raw C header files. In this case, you usually only want the | |
259 | extension module to include functions in the supplied header file | |
260 | rather than everything that might be included by that header file | |
261 | (i.e., system headers, C library functions, etc.). | |
262 | </p> | |
263 | ||
264 | <p> | |
265 | It should also be noted that the SWIG preprocessor skips all text | |
266 | enclosed inside a <tt>%{...%}</tt> block. In addition, the | |
267 | preprocessor includes a number of macro handling enhancements that | |
268 | make it more powerful than the normal C preprocessor. These | |
269 | extensions are described in the "<a href="Preprocessor.html#Preprocessor">Preprocessor</a>" chapter. | |
270 | </p> | |
271 | ||
272 | <H3><a name="SWIG_nn7"></a>5.1.5 SWIG Directives</H3> | |
273 | ||
274 | ||
275 | <p> | |
276 | Most of SWIG's operation is controlled by special directives that are | |
277 | always preceded by a "<tt>%</tt>" to distinguish them from normal C | |
278 | declarations. These directives are used to give SWIG hints or to alter | |
279 | SWIG's parsing behavior in some manner. | |
280 | </p> | |
281 | ||
282 | <p> | |
283 | Since SWIG directives are not legal C syntax, it is generally not | |
284 | possible to include them in header files. However, SWIG directives can be | |
285 | included in C header files using conditional compilation like this: | |
286 | </p> | |
287 | ||
288 | <div class="code"><pre> | |
289 | /* header.h --- Some header file */ | |
290 | ||
291 | /* SWIG directives -- only seen if SWIG is running */ | |
292 | #ifdef SWIG | |
293 | %module foo | |
294 | #endif | |
295 | </pre> | |
296 | </div> | |
297 | ||
298 | <p> | |
299 | <tt>SWIG</tt> is a special preprocessing symbol defined by SWIG when | |
300 | it is parsing an input file. | |
301 | </p> | |
302 | ||
303 | <H3><a name="SWIG_nn8"></a>5.1.6 Parser Limitations</H3> | |
304 | ||
305 | ||
306 | <p> | |
307 | Although SWIG can parse most C/C++ declarations, it does not | |
308 | provide a complete C/C++ parser implementation. Most of these | |
309 | limitations pertain to very complicated type declarations and certain | |
310 | advanced C++ features. Specifically, the following features are not | |
311 | currently supported: | |
312 | </p> | |
313 | ||
314 | <ul> | |
315 | <li>Non-conventional type declarations. | |
316 | For example, SWIG does not support declarations such as the following | |
317 | (even though this is legal C): | |
318 | ||
319 | <div class="code"> | |
320 | <pre> | |
321 | /* Non-conventional placement of storage specifier (extern) */ | |
322 | const int extern Number; | |
323 | ||
324 | /* Extra declarator grouping */ | |
325 | Matrix (foo); // A global variable | |
326 | ||
327 | /* Extra declarator grouping in parameters */ | |
328 | void bar(Spam (Grok)(Doh)); | |
329 | ||
330 | </pre> | |
331 | </div> | |
332 | ||
333 | <p> | |
334 | In practice, few (if any) C programmers actually write code like | |
335 | this since this style is never featured in programming books. However, | |
336 | if you're feeling particularly obfuscated, you can certainly break SWIG (although why would you want to?). | |
337 | </p> | |
338 | </li> | |
339 | ||
340 | <li>Running SWIG on C++ source files (what would appear in a .C or .cxx file) | |
341 | is not recommended. Even though SWIG can parse C++ class declarations, | |
342 | it ignores declarations that are decoupled from their | |
343 | original class definition (the declarations are parsed, but a lot of warning | |
344 | messages may be generated). For example: | |
345 | ||
346 | <div class="code"> | |
347 | <pre> | |
348 | /* Not supported by SWIG */ | |
349 | int foo::bar(int) { | |
350 | ... whatever ... | |
351 | } | |
352 | </pre> | |
353 | </div> | |
354 | </li> | |
355 | ||
356 | <li>Certain advanced features of C++ such as nested classes | |
357 | are not yet supported. Please see the section on using SWIG | |
358 | with C++ for more information. | |
359 | </ul> | |
360 | ||
361 | <p> | |
362 | In the event of a parsing error, conditional compilation can be used to skip | |
363 | offending code. For example: | |
364 | </p> | |
365 | ||
366 | <div class="code"> | |
367 | <pre> | |
368 | #ifndef SWIG | |
369 | ... some bad declarations ... | |
370 | #endif | |
371 | </pre> | |
372 | </div> | |
373 | <p> | |
374 | Alternatively, you can just delete the offending code from the interface file. | |
375 | </p> | |
376 | ||
377 | <p> | |
378 | One of the reasons why SWIG does not provide a full C++ parser | |
379 | implementation is that it has been designed to work with incomplete | |
380 | specifications and to be very permissive in its handling of C/C++ | |
381 | datatypes (e.g., SWIG can generate interfaces even when there are | |
382 | missing class declarations or opaque datatypes). Unfortunately, this | |
383 | approach makes it extremely difficult to implement certain parts of a | |
384 | C/C++ parser as most compilers use type information to assist in the | |
385 | parsing of more complex declarations (for the truly curious, the | |
386 | primary complication in the implementation is that the SWIG parser | |
387 | does not utilize a separate <em>typedef-name</em> terminal symbol as | |
388 | described on p. 234 of K&R). | |
389 | </p> | |
390 | ||
391 | <H2><a name="SWIG_nn9"></a>5.2 Wrapping Simple C Declarations</H2> | |
392 | ||
393 | ||
394 | <p> | |
395 | SWIG wraps simple C declarations by creating an interface that closely matches | |
396 | the way in which the declarations would be used in a C program. | |
397 | For example, consider the following interface file: | |
398 | </p> | |
399 | ||
400 | <div class="code"><pre> | |
401 | %module example | |
402 | ||
403 | %inline %{ | |
404 | extern double sin(double x); | |
405 | extern int strcmp(const char *, const char *); | |
406 | extern int Foo; | |
407 | %} | |
408 | #define STATUS 50 | |
409 | #define VERSION "1.1" | |
410 | </pre></div> | |
411 | <p> | |
412 | In this file, there are two functions <tt>sin()</tt> and <tt>strcmp()</tt>, | |
413 | a global variable <tt>Foo</tt>, and two constants <tt>STATUS</tt> and | |
414 | <tt>VERSION</tt>. When SWIG creates an extension module, these | |
415 | declarations are accessible as scripting language functions, variables, and | |
416 | constants respectively. For example, in Tcl: | |
417 | </p> | |
418 | ||
419 | <div class="targetlang"><pre> | |
420 | % sin 3 | |
421 | 5.2335956 | |
422 | % strcmp Dave Mike | |
423 | -1 | |
424 | % puts $Foo | |
425 | 42 | |
426 | % puts $STATUS | |
427 | 50 | |
428 | % puts $VERSION | |
429 | 1.1 | |
430 | </pre></div> | |
431 | <p> | |
432 | Or in Python: | |
433 | </p> | |
434 | ||
435 | <div class="targetlang"><pre> | |
436 | >>> example.sin(3) | |
437 | 5.2335956 | |
438 | >>> example.strcmp('Dave','Mike') | |
439 | -1 | |
440 | >>> print example.cvar.Foo | |
441 | 42 | |
442 | >>> print example.STATUS | |
443 | 50 | |
444 | >>> print example.VERSION | |
445 | 1.1 | |
446 | </pre></div> | |
447 | <p> | |
448 | Whenever possible, SWIG creates an interface that closely matches the underlying C/C++ | |
449 | code. However, due to subtle differences between languages, run-time | |
450 | environments, and semantics, it is not always possible to do so. The | |
451 | next few sections describes various aspects of this mapping. | |
452 | </p> | |
453 | ||
454 | <H3><a name="SWIG_nn10"></a>5.2.1 Basic Type Handling</H3> | |
455 | ||
456 | ||
457 | <p> | |
458 | In order to build an interface, SWIG has to convert C/C++ datatypes to | |
459 | equivalent types in the target language. Generally, | |
460 | scripting languages provide a more limited set of primitive types than C. | |
461 | Therefore, this conversion process involves a certain amount of type | |
462 | coercion. | |
463 | </p> | |
464 | ||
465 | <p> | |
466 | Most scripting languages provide a single integer type that is implemented using | |
467 | the <tt>int</tt> or <tt>long</tt> datatype in C. The following list shows | |
468 | all of the C datatypes that SWIG will convert to and from integers in the target language: | |
469 | </p> | |
470 | ||
471 | <div class="code"><pre> | |
472 | int | |
473 | short | |
474 | long | |
475 | unsigned | |
476 | signed | |
477 | unsigned short | |
478 | unsigned long | |
479 | unsigned char | |
480 | signed char | |
481 | bool | |
482 | </pre></div> | |
483 | ||
484 | <p> | |
485 | When an integral value is converted from C, a cast is used to convert it to | |
486 | the representation in the target language. | |
487 | Thus, a 16 bit short in C may be promoted to a 32 bit integer. When integers are | |
488 | converted in the other direction, the value is cast back into the original C type. | |
489 | If the value is too large to fit, it is silently truncated. | |
490 | <!-- Dave: Maybe we should fix this --> | |
491 | </p> | |
492 | ||
493 | <p> | |
494 | <tt>unsigned char</tt> and <tt>signed char</tt> are special cases that | |
495 | are handled as small 8-bit integers. Normally, the <tt>char</tt> | |
496 | datatype is mapped as a one-character ASCII string. </p> | |
497 | ||
498 | <p> | |
499 | The <tt>bool</tt> datatype is cast to and from an integer value of 0 | |
500 | and 1 unless the target language provides a special boolean type.</p> | |
501 | ||
502 | <p> | |
503 | Some care is required when working with large integer values. Most | |
504 | scripting languages use 32-bit integers so mapping a 64-bit long | |
505 | integer may lead to truncation errors. Similar problems may arise with | |
506 | 32 bit unsigned integers (which may appear as large negative | |
507 | numbers). As a rule of thumb, the <tt>int</tt> datatype and all | |
508 | variations of <tt>char</tt> and <tt>short</tt> datatypes are safe to | |
509 | use. For <tt>unsigned int</tt> and <tt>long</tt> datatypes, you will | |
510 | need to carefully check the correct operation of your program after | |
511 | it has been wrapped with SWIG. | |
512 | </p> | |
513 | ||
514 | <p> | |
515 | Although the SWIG parser supports the <tt>long long</tt> datatype, not | |
516 | all language modules support it. This is because <tt>long long</tt> | |
517 | usually exceeds the integer precision available in the target | |
518 | language. In certain modules such as Tcl and Perl5, <tt>long | |
519 | long</tt> integers are encoded as strings. This allows the full range | |
520 | of these numbers to be represented. However, it does not allow | |
521 | <tt>long long</tt> values to be used in arithmetic expressions. It | |
522 | should also be noted that although <tt>long long</tt> is part | |
523 | of the ISO C99 standard, it is not universally supported by all C | |
524 | compilers. Make sure you are using a compiler that supports <tt>long | |
525 | long</tt> before trying to use this type with SWIG. | |
526 | </p> | |
527 | ||
528 | <p> | |
529 | SWIG recognizes the following floating point types :</p> | |
530 | ||
531 | <div class="code"><pre> | |
532 | float | |
533 | double | |
534 | </pre></div> | |
535 | ||
536 | <p> | |
537 | Floating point numbers are mapped to and from the natural | |
538 | representation of floats in the target language. This is almost always | |
539 | a C <tt>double</tt>. The rarely used datatype of <tt>long double</tt> | |
540 | is not supported by SWIG.</p> | |
541 | ||
542 | <p> | |
543 | The <tt>char</tt> datatype is mapped into a NULL terminated ASCII | |
544 | string with a single character. When used in a scripting language it | |
545 | shows up as a tiny string containing the character value. When | |
546 | converting the value back into C, SWIG takes a character string | |
547 | from the scripting language and strips off the first character as the | |
548 | char value. Thus if the value "foo" is assigned to a | |
549 | <tt>char</tt> datatype, it gets the value `f'.</p> | |
550 | ||
551 | <p> | |
552 | The <tt>char *</tt> datatype is handled as a NULL-terminated ASCII | |
553 | string. SWIG maps this into a 8-bit character string in the target | |
554 | scripting language. SWIG converts character strings in the target | |
555 | language to NULL terminated strings before passing them into | |
556 | C/C++. The default handling of these strings does not allow them to | |
557 | have embedded NULL bytes. Therefore, the <tt>char *</tt> datatype is | |
558 | not generally suitable for passing binary data. However, it is | |
559 | possible to change this behavior by defining a SWIG typemap. See the chapter | |
560 | on <a href="Typemaps.html#Typemaps">Typemaps</a> for details about this. | |
561 | </p> | |
562 | ||
563 | <p> | |
564 | At this time, SWIG does not provide any special support for Unicode or | |
565 | wide-character strings (the C <tt>wchar_t</tt> type). This is a | |
566 | delicate topic that is poorly understood by many programmers and not | |
567 | implemented in a consistent manner across languages. For those | |
568 | scripting languages that provide Unicode support, Unicode strings are | |
569 | often available in an 8-bit representation such as UTF-8 that can be | |
570 | mapped to the <tt>char *</tt> type (in which case the SWIG interface | |
571 | will probably work). If the program you are wrapping uses Unicode, | |
572 | there is no guarantee that Unicode characters in the target language | |
573 | will use the same internal representation (e.g., UCS-2 vs. UCS-4). | |
574 | You may need to write some special conversion functions. | |
575 | </p> | |
576 | ||
577 | <H3><a name="SWIG_nn11"></a>5.2.2 Global Variables</H3> | |
578 | ||
579 | ||
580 | <p> | |
581 | Whenever possible, SWIG maps C/C++ global variables into scripting language | |
582 | variables. For example, | |
583 | </p> | |
584 | ||
585 | <div class="code"><pre> | |
586 | %module example | |
587 | double foo; | |
588 | ||
589 | </pre></div> | |
590 | <p> | |
591 | results in a scripting language variable like this: | |
592 | </p> | |
593 | ||
594 | <div class="code"><pre> | |
595 | # Tcl | |
596 | set foo [3.5] ;# Set foo to 3.5 | |
597 | puts $foo ;# Print the value of foo | |
598 | ||
599 | # Python | |
600 | cvar.foo = 3.5 # Set foo to 3.5 | |
601 | print cvar.foo # Print value of foo | |
602 | ||
603 | # Perl | |
604 | $foo = 3.5; # Set foo to 3.5 | |
605 | print $foo,"\n"; # Print value of foo | |
606 | ||
607 | # Ruby | |
608 | Module.foo = 3.5 # Set foo to 3.5 | |
609 | print Module.foo, "\n" # Print value of foo | |
610 | </pre></div> | |
611 | <p> | |
612 | Whenever the scripting language variable is used, the underlying C | |
613 | global variable is accessed. Although SWIG makes every | |
614 | attempt to make global variables work like scripting language | |
615 | variables, it is not always possible to do so. For instance, in | |
616 | Python, all global variables must be accessed through a special | |
617 | variable object known as <tt>cvar</tt> (shown above). In Ruby, variables are | |
618 | accessed as attributes of the module. Other languages may | |
619 | convert variables to a pair of accessor functions. For example, the | |
620 | Java module generates a pair of functions <tt>double get_foo()</tt> | |
621 | and <tt>set_foo(double val)</tt> that are used to manipulate the | |
622 | value. | |
623 | </p> | |
624 | ||
625 | <p> | |
626 | Finally, if a global variable has been declared as <tt>const</tt>, it | |
627 | only supports read-only access. Note: this behavior is new to SWIG-1.3. | |
628 | Earlier versions of SWIG incorrectly handled <tt>const</tt> and created | |
629 | constants instead. | |
630 | </p> | |
631 | ||
632 | <H3><a name="SWIG_nn12"></a>5.2.3 Constants</H3> | |
633 | ||
634 | ||
635 | <p> | |
636 | Constants can be created using <tt>#define</tt>, enumerations, | |
637 | or a special <tt>%constant</tt> directive. The following | |
638 | interface file shows a few valid constant declarations :</p> | |
639 | ||
640 | <div class="code"><pre> | |
641 | #define I_CONST 5 // An integer constant | |
642 | #define PI 3.14159 // A Floating point constant | |
643 | #define S_CONST "hello world" // A string constant | |
644 | #define NEWLINE '\n' // Character constant | |
645 | ||
646 | enum boolean {NO=0, YES=1}; | |
647 | enum months {JAN, FEB, MAR, APR, MAY, JUN, JUL, AUG, | |
648 | SEP, OCT, NOV, DEC}; | |
649 | %constant double BLAH = 42.37; | |
650 | #define F_CONST (double) 5 // A floating pointer constant with cast | |
651 | #define PI_4 PI/4 | |
652 | #define FLAGS 0x04 | 0x08 | 0x40 | |
653 | ||
654 | </pre></div> | |
655 | <p> | |
656 | In <tt>#define</tt> declarations, the type of a constant is inferred | |
657 | by syntax. For example, a number with a decimal point is assumed to be | |
658 | floating point. In addition, SWIG must be able to fully resolve all | |
659 | of the symbols used in a <tt>#define</tt> in order for a constant to | |
660 | actually be created. This restriction is necessary because | |
661 | <tt>#define</tt> is also used to define preprocessor macros that are | |
662 | definitely not meant to be part of the scripting language interface. | |
663 | For example: | |
664 | </p> | |
665 | ||
666 | <div class="code"> | |
667 | <pre> | |
668 | #define EXTERN extern | |
669 | ||
670 | EXTERN void foo(); | |
671 | </pre> | |
672 | </div> | |
673 | <p> | |
674 | In this case, you probably don't want to create a constant called | |
675 | <tt>EXTERN</tt> (what would the value be?). In general, | |
676 | SWIG will not create constants for macros unless the value can | |
677 | be completely determined by the preprocessor. For instance, in the above example, | |
678 | the declaration | |
679 | </p> | |
680 | ||
681 | <div class="code"> | |
682 | <pre> | |
683 | #define PI_4 PI/4 | |
684 | </pre> | |
685 | </div> | |
686 | <p> | |
687 | defines a constant because <tt>PI</tt> was already defined as a | |
688 | constant and the value is known. | |
689 | </p> | |
690 | ||
691 | <p> | |
692 | The use of constant expressions is allowed, but SWIG does not evaluate | |
693 | them. Rather, it passes them through to the output file and lets the C | |
694 | compiler perform the final evaluation (SWIG does perform a limited | |
695 | form of type-checking however).</p> | |
696 | ||
697 | <p> | |
698 | For enumerations, it is critical that the original enum definition be | |
699 | included somewhere in the interface file (either in a header file or | |
700 | in the <tt>%{,%}</tt> block). SWIG only translates the enumeration | |
701 | into code needed to add the constants to a scripting language. It | |
702 | needs the original enumeration declaration in order to get the correct | |
703 | enum values as assigned by the C compiler. | |
704 | </p> | |
705 | ||
706 | <p> | |
707 | The <tt>%constant</tt> directive is used to more precisely create | |
708 | constants corresponding to different C datatypes. Although it is not | |
709 | usually not needed for simple values, it is more useful when working | |
710 | with pointers and other more complex datatypes. Typically, <tt>%constant</tt> | |
711 | is only used when you want to add constants to the scripting language | |
712 | interface that are not defined in the original header file. | |
713 | </p> | |
714 | ||
715 | <H3><a name="SWIG_nn13"></a>5.2.4 A brief word about <tt>const</tt></H3> | |
716 | ||
717 | ||
718 | <p> | |
719 | A common confusion with C programming is the semantic meaning of the | |
720 | <tt>const</tt> qualifier in declarations--especially when it is mixed | |
721 | with pointers and other type modifiers. In fact, previous versions of SWIG | |
722 | handled <tt>const</tt> incorrectly--a situation that SWIG-1.3.7 and newer | |
723 | releases have fixed. | |
724 | </p> | |
725 | ||
726 | <p> | |
727 | Starting with SWIG-1.3, all variable declarations, regardless of any | |
728 | use of <tt>const</tt>, are wrapped as global variables. If a | |
729 | declaration happens to be declared as <tt>const</tt>, it is wrapped as | |
730 | a read-only variable. To tell if a variable is <tt>const</tt> or not, | |
731 | you need to look at the right-most occurrence of the <tt>const</tt> | |
732 | qualifier (that appears before the variable name). If the right-most | |
733 | <tt>const</tt> occurs after all other type modifiers (such as | |
734 | pointers), then the variable is <tt>const</tt>. Otherwise, it is not. | |
735 | </p> | |
736 | ||
737 | <p> | |
738 | Here are some examples of <tt>const</tt> declarations. | |
739 | </p> | |
740 | ||
741 | <div class="code"> | |
742 | <pre> | |
743 | const char a; // A constant character | |
744 | char const b; // A constant character (the same) | |
745 | char *const c; // A constant pointer to a character | |
746 | const char *const d; // A constant pointer to a constant character | |
747 | </pre> | |
748 | </div> | |
749 | <p> | |
750 | Here is an example of a declaration that is not <tt>const</tt>: | |
751 | </p> | |
752 | ||
753 | <div class="code"> | |
754 | <pre> | |
755 | const char *e; // A pointer to a constant character. The pointer | |
756 | // may be modified. | |
757 | </pre> | |
758 | </div> | |
759 | <p> | |
760 | In this case, the pointer <tt>e</tt> can change---it's only the value | |
761 | being pointed to that is read-only. | |
762 | </p> | |
763 | ||
764 | <p> | |
765 | <b>Compatibility Note:</b> One reason for changing SWIG to handle | |
766 | <tt>const</tt> declarations as read-only variables is that there are | |
767 | many situations where the value of a <tt>const</tt> variable might | |
768 | change. For example, a library might export a symbol as | |
769 | <tt>const</tt> in its public API to discourage modification, but still | |
770 | allow the value to change through some other kind of internal | |
771 | mechanism. Furthermore, programmers often overlook the fact that with | |
772 | a constant declaration like <tt>char *const</tt>, the underlying data | |
773 | being pointed to can be modified--it's only the pointer itself that is | |
774 | constant. In an embedded system, a <tt>const</tt> declaration might | |
775 | refer to a read-only memory address such as the location of a | |
776 | memory-mapped I/O device port (where the value changes, but writing to | |
777 | the port is not supported by the hardware). Rather than trying to | |
778 | build a bunch of special cases into the <tt>const</tt> qualifier, the | |
779 | new interpretation of <tt>const</tt> as "read-only" is simple and | |
780 | exactly matches the actual semantics of <tt>const</tt> in C/C++. If | |
781 | you really want to create a constant as in older versions of SWIG, use | |
782 | the <tt>%constant</tt> directive instead. For example: | |
783 | </p> | |
784 | ||
785 | <div class="code"> | |
786 | <pre> | |
787 | %constant double PI = 3.14159; | |
788 | </pre> | |
789 | </div> | |
790 | ||
791 | <p> | |
792 | or | |
793 | </p> | |
794 | ||
795 | <div class="code"> | |
796 | <pre> | |
797 | #ifdef SWIG | |
798 | #define const %constant | |
799 | #endif | |
800 | const double foo = 3.4; | |
801 | const double bar = 23.4; | |
802 | const int spam = 42; | |
803 | #ifdef SWIG | |
804 | #undef const | |
805 | #endif | |
806 | ... | |
807 | ||
808 | </pre> | |
809 | </div> | |
810 | ||
811 | <H3><a name="SWIG_nn14"></a>5.2.5 A cautionary tale of <tt>char *</tt></H3> | |
812 | ||
813 | ||
814 | <p> | |
815 | Before going any further, there is one bit of caution involving | |
816 | <tt>char *</tt> that must now be mentioned. When strings are passed | |
817 | from a scripting language to a C <tt>char *</tt>, the pointer usually | |
818 | points to string data stored inside the interpreter. It is almost | |
819 | always a really bad idea to modify this data. Furthermore, some | |
820 | languages may explicitly disallow it. For instance, in Python, | |
821 | strings are supposed be immutable. If you violate this, you will probably | |
822 | receive a vast amount of wrath when you unleash your module on the world. | |
823 | </p> | |
824 | ||
825 | <p> | |
826 | The primary source of problems are functions that might modify string data in place. | |
827 | A classic example would be a function like this: | |
828 | </p> | |
829 | ||
830 | <div class="code"> | |
831 | <pre> | |
832 | char *strcat(char *s, const char *t) | |
833 | </pre> | |
834 | </div> | |
835 | ||
836 | <p> | |
837 | Although SWIG will certainly generate a wrapper for this, its behavior | |
838 | will be undefined. In fact, it will probably cause your application | |
839 | to crash with a segmentation fault or other memory related problem. | |
840 | This is because <tt>s</tt> refers to some internal data in the target | |
841 | language---data that you shouldn't be touching. | |
842 | </p> | |
843 | ||
844 | <p> | |
845 | The bottom line: don't rely on <tt>char *</tt> for anything other than read-only | |
846 | input values. However, it must be noted that you could change the behavior of SWIG | |
847 | using <a href="Typemaps.html#Typemaps">typemaps</a>. | |
848 | </p> | |
849 | ||
850 | <H2><a name="SWIG_nn15"></a>5.3 Pointers and complex objects</H2> | |
851 | ||
852 | ||
853 | <p> | |
854 | Most C programs manipulate arrays, structures, and other types of objects. This section | |
855 | discusses the handling of these datatypes. | |
856 | </p> | |
857 | ||
858 | <H3><a name="SWIG_nn16"></a>5.3.1 Simple pointers</H3> | |
859 | ||
860 | ||
861 | <p> | |
862 | Pointers to primitive C datatypes such as </p> | |
863 | ||
864 | <div class="code"><pre> | |
865 | int * | |
866 | double *** | |
867 | char ** | |
868 | </pre></div> | |
869 | <p> | |
870 | are fully supported by SWIG. Rather than trying to convert the data being pointed to into a scripting | |
871 | representation, SWIG simply encodes the pointer itself into a | |
872 | representation that contains the actual value of the pointer and a type-tag. | |
873 | Thus, the SWIG representation of the above | |
874 | pointers (in Tcl), might look like this:</p> | |
875 | ||
876 | <div class="targetlang"><pre> | |
877 | _10081012_p_int | |
878 | _1008e124_ppp_double | |
879 | _f8ac_pp_char | |
880 | </pre></div> | |
881 | ||
882 | <p> | |
883 | A NULL pointer is represented by the string "NULL" or the value 0 | |
884 | encoded with type information.</p> | |
885 | ||
886 | <p> | |
887 | All pointers are treated as opaque objects by SWIG. Thus, a pointer | |
888 | may be returned by a function and passed around to other C functions | |
889 | as needed. For all practical purposes, the scripting language | |
890 | interface works in exactly the same way as you would use the | |
891 | pointer in a C program. The only difference is that there is no mechanism for | |
892 | dereferencing the pointer since this would require the target language | |
893 | to understand the memory layout of the underlying object. | |
894 | </p> | |
895 | ||
896 | <p> | |
897 | The scripting language representation of a pointer value should never be | |
898 | manipulated directly. Even though the values shown look like hexadecimal | |
899 | addresses, the numbers used may differ from the actual machine address (e.g., | |
900 | on little-endian machines, the digits may appear in reverse order). | |
901 | Furthermore, SWIG does not | |
902 | normally map pointers into high-level objects such as associative | |
903 | arrays or lists (for example, converting an | |
904 | <tt>int *</tt> into an list of integers). There are several reasons | |
905 | why SWIG does not do this:</p> | |
906 | ||
907 | <ul> | |
908 | <li>There is not enough information in a C declaration to properly map | |
909 | pointers into higher level constructs. For example, an <tt>int *</tt> | |
910 | may indeed be an array of integers, but if it contains ten million | |
911 | elements, converting it into a list object is probably a bad idea. | |
912 | </li> | |
913 | ||
914 | <li>The underlying semantics associated with a pointer is not known | |
915 | by SWIG. For instance, an <tt>int *</tt> might not be an array at all--perhaps it | |
916 | is an output value! | |
917 | </li> | |
918 | ||
919 | <li>By handling all pointers in a consistent manner, the implementation of SWIG is greatly | |
920 | simplified and less prone to error. | |
921 | </li> | |
922 | </ul> | |
923 | ||
924 | <H3><a name="SWIG_nn17"></a>5.3.2 Run time pointer type checking</H3> | |
925 | ||
926 | ||
927 | <p> | |
928 | By allowing pointers to be manipulated from a scripting language, extension modules | |
929 | effectively bypass compile-time type checking in the C/C++ | |
930 | compiler. To prevent errors, a type signature is encoded into all | |
931 | pointer values and is used to perform run-time type checking. This | |
932 | type-checking process is an integral part of SWIG and can not be | |
933 | disabled or modified without using typemaps (described in later | |
934 | chapters). | |
935 | </p> | |
936 | ||
937 | <p> | |
938 | Like C, <tt>void *</tt> matches any kind of pointer. Furthermore, | |
939 | <tt>NULL</tt> pointers can be passed to any function that expects to | |
940 | receive a pointer. Although this has the potential to cause a crash, | |
941 | <tt>NULL</tt> pointers are also sometimes used | |
942 | as sentinel values or to denote a missing/empty value. Therefore, | |
943 | SWIG leaves NULL pointer checking up to the application. | |
944 | </p> | |
945 | ||
946 | <H3><a name="SWIG_nn18"></a>5.3.3 Derived types, structs, and classes</H3> | |
947 | ||
948 | ||
949 | <p> | |
950 | For everything else (structs, classes, arrays, etc...) SWIG applies a | |
951 | very simple rule :</p> | |
952 | ||
953 | <center> | |
954 | <b>Everything else is a pointer</b> | |
955 | </center> | |
956 | ||
957 | <p> | |
958 | In other words, SWIG manipulates everything else by reference. This | |
959 | model makes sense because most C/C++ programs make heavy use of | |
960 | pointers and SWIG can use the type-checked pointer mechanism already | |
961 | present for handling pointers to basic datatypes.</p> | |
962 | ||
963 | <p> | |
964 | Although this probably sounds complicated, it's really quite | |
965 | simple. Suppose you have an interface file like this :</p> | |
966 | ||
967 | <div class="code"><pre> | |
968 | %module fileio | |
969 | FILE *fopen(char *, char *); | |
970 | int fclose(FILE *); | |
971 | unsigned fread(void *ptr, unsigned size, unsigned nobj, FILE *); | |
972 | unsigned fwrite(void *ptr, unsigned size, unsigned nobj, FILE *); | |
973 | void *malloc(int nbytes); | |
974 | void free(void *); | |
975 | ||
976 | </pre></div> | |
977 | ||
978 | <p> | |
979 | In this file, SWIG doesn't know what a <tt>FILE</tt> is, but since it's used | |
980 | as a pointer, so it doesn't really matter what it is. If you wrapped | |
981 | this module into Python, you can use the functions just like you | |
982 | expect :</p> | |
983 | ||
984 | <div class="targetlang"><pre> | |
985 | # Copy a file | |
986 | def filecopy(source,target): | |
987 | f1 = fopen(source,"r") | |
988 | f2 = fopen(target,"w") | |
989 | buffer = malloc(8192) | |
990 | nbytes = fread(buffer,8192,1,f1) | |
991 | while (nbytes > 0): | |
992 | fwrite(buffer,8192,1,f2) | |
993 | nbytes = fread(buffer,8192,1,f1) | |
994 | free(buffer) | |
995 | ||
996 | </pre></div> | |
997 | ||
998 | <p> | |
999 | In this case <tt>f1</tt>,<tt> f2</tt>, and <tt>buffer</tt> are all | |
1000 | opaque objects containing C pointers. It doesn't matter what value | |
1001 | they contain--our program works just fine without this knowledge.</p> | |
1002 | ||
1003 | <H3><a name="SWIG_nn19"></a>5.3.4 Undefined datatypes</H3> | |
1004 | ||
1005 | ||
1006 | <p> | |
1007 | When SWIG encounters an undeclared datatype, it automatically assumes | |
1008 | that it is a structure or class. For example, suppose the following | |
1009 | function appeared in a SWIG input file:</p> | |
1010 | ||
1011 | <div class="code"><pre> | |
1012 | void matrix_multiply(Matrix *a, Matrix *b, Matrix *c); | |
1013 | </pre></div> | |
1014 | ||
1015 | <p> | |
1016 | SWIG has no idea what a "<tt>Matrix</tt>" is. However, it is obviously | |
1017 | a pointer to something so SWIG generates a wrapper using its generic pointer | |
1018 | handling code. | |
1019 | </p> | |
1020 | ||
1021 | <p> | |
1022 | Unlike C or C++, SWIG does not actually care whether <tt>Matrix</tt> | |
1023 | has been previously defined in the interface file or not. This | |
1024 | allows SWIG to generate interfaces from | |
1025 | only partial or limited information. In some cases, you may not care | |
1026 | what a <tt>Matrix</tt> really is as long as you can pass an opaque reference to | |
1027 | one around in the scripting language interface. | |
1028 | </p> | |
1029 | ||
1030 | <p> | |
1031 | An important detail to mention is that SWIG will gladly generate | |
1032 | wrappers for an interface when there are unspecified type names. | |
1033 | However, <b>all unspecified types are internally handled as pointers | |
1034 | to structures or classes!</b> For example, consider the following declaration: | |
1035 | </p> | |
1036 | ||
1037 | <div class="code"> | |
1038 | <pre> | |
1039 | void foo(size_t num); | |
1040 | </pre> | |
1041 | </div> | |
1042 | ||
1043 | <p> | |
1044 | If <tt>size_t</tt> is undeclared, SWIG generates wrappers | |
1045 | that expect to receive a type of <tt>size_t *</tt> (this mapping is described shortly). | |
1046 | As a result, the scripting interface might behave strangely. For example: | |
1047 | </p> | |
1048 | ||
1049 | <div class="code"> | |
1050 | <pre> | |
1051 | foo(40); | |
1052 | TypeError: expected a _p_size_t. | |
1053 | </pre> | |
1054 | </div> | |
1055 | ||
1056 | <p> | |
1057 | The only way to fix this problem is to make sure you properly declare type names using | |
1058 | <tt>typedef</tt>. | |
1059 | </p> | |
1060 | ||
1061 | <!-- We might want to add an error reporting flag to swig --> | |
1062 | ||
1063 | <H3><a name="SWIG_nn20"></a>5.3.5 Typedef</H3> | |
1064 | ||
1065 | ||
1066 | <p> | |
1067 | Like C, <tt>typedef</tt> can be used to define new type names in SWIG. For example: | |
1068 | </p> | |
1069 | ||
1070 | <div class="code"><pre> | |
1071 | typedef unsigned int size_t; | |
1072 | </pre></div> | |
1073 | ||
1074 | <p> | |
1075 | <tt>typedef</tt> definitions appearing in a SWIG interface | |
1076 | are not propagated to the generated wrapper code. Therefore, they | |
1077 | either need to be defined in an included header file or placed in the | |
1078 | declarations section like this: | |
1079 | </p> | |
1080 | ||
1081 | <div class="code"> | |
1082 | <pre> | |
1083 | %{ | |
1084 | /* Include in the generated wrapper file */ | |
1085 | typedef unsigned int size_t; | |
1086 | %} | |
1087 | /* Tell SWIG about it */ | |
1088 | typedef unsigned int size_t; | |
1089 | </pre> | |
1090 | </div> | |
1091 | ||
1092 | <p> | |
1093 | or | |
1094 | </p> | |
1095 | ||
1096 | <div class="code"> | |
1097 | <pre> | |
1098 | %inline %{ | |
1099 | typedef unsigned int size_t; | |
1100 | %} | |
1101 | </pre> | |
1102 | </div> | |
1103 | ||
1104 | <p> | |
1105 | In certain cases, you might be able to include other header files to collect type information. | |
1106 | For example: | |
1107 | </p> | |
1108 | ||
1109 | <div class="code"> | |
1110 | <pre> | |
1111 | %module example | |
1112 | %import "sys/types.h" | |
1113 | </pre> | |
1114 | </div> | |
1115 | ||
1116 | <p> | |
1117 | In this case, you might run SWIG as follows: | |
1118 | </p> | |
1119 | ||
1120 | <div class="shell"> | |
1121 | <pre> | |
1122 | $ swig -I/usr/include -includeall example.i | |
1123 | </pre> | |
1124 | </div> | |
1125 | ||
1126 | <p> | |
1127 | It should be noted that your mileage will vary greatly here. | |
1128 | System headers are notoriously complicated and may rely upon a variety | |
1129 | of non-standard C coding extensions (e.g., such as special directives | |
1130 | to GCC). Unless you exactly specify the right include directories and | |
1131 | preprocessor symbols, this may not work correctly (you will have to | |
1132 | experiment). | |
1133 | </p> | |
1134 | ||
1135 | <p> | |
1136 | SWIG tracks <tt>typedef</tt> declarations and uses this information | |
1137 | for run-time type checking. For instance, if you use the above <tt>typedef</tt> and | |
1138 | had the following function declaration: | |
1139 | </p> | |
1140 | ||
1141 | <div class="code"> | |
1142 | <pre> | |
1143 | void foo(unsigned int *ptr); | |
1144 | </pre> | |
1145 | </div> | |
1146 | ||
1147 | <p> | |
1148 | The corresponding wrapper function will accept arguments of | |
1149 | type <tt>unsigned int *</tt> or <tt>size_t *</tt>. | |
1150 | </p> | |
1151 | ||
1152 | <H2><a name="SWIG_nn21"></a>5.4 Other Practicalities</H2> | |
1153 | ||
1154 | ||
1155 | <p> | |
1156 | So far, this chapter has presented almost everything you need to know to use SWIG | |
1157 | for simple interfaces. However, some C programs use idioms that are somewhat | |
1158 | more difficult to map to a scripting language interface. This section describes | |
1159 | some of these issues. | |
1160 | </p> | |
1161 | ||
1162 | <H3><a name="SWIG_nn22"></a>5.4.1 Passing structures by value</H3> | |
1163 | ||
1164 | ||
1165 | <p> | |
1166 | Sometimes a C function takes structure parameters that are passed | |
1167 | by value. For example, consider the following function: | |
1168 | </p> | |
1169 | ||
1170 | <div class="code"><pre> | |
1171 | double dot_product(Vector a, Vector b); | |
1172 | </pre></div> | |
1173 | ||
1174 | <p> | |
1175 | To deal with this, SWIG transforms the function to use pointers by | |
1176 | creating a wrapper equivalent to the following: | |
1177 | </p> | |
1178 | ||
1179 | <div class="code"><pre> | |
1180 | double wrap_dot_product(Vector *a, Vector *b) { | |
1181 | Vector x = *a; | |
1182 | Vector y = *b; | |
1183 | return dot_product(x,y); | |
1184 | } | |
1185 | </pre></div> | |
1186 | ||
1187 | <p> | |
1188 | In the target language, the <tt>dot_product()</tt> function now accepts pointers | |
1189 | to Vectors instead of Vectors. For the most part, this transformation | |
1190 | is transparent so you might not notice. | |
1191 | </p> | |
1192 | ||
1193 | <H3><a name="SWIG_nn23"></a>5.4.2 Return by value</H3> | |
1194 | ||
1195 | ||
1196 | <p> | |
1197 | C functions that return structures or classes datatypes by value are more difficult | |
1198 | to handle. Consider the following function:</p> | |
1199 | ||
1200 | <div class="code"><pre> | |
1201 | Vector cross_product(Vector v1, Vector v2); | |
1202 | </pre></div> | |
1203 | ||
1204 | <p> | |
1205 | This function wants to return <tt>Vector</tt>, but SWIG only really supports | |
1206 | pointers. As a result, SWIG creates a wrapper like this: | |
1207 | </p> | |
1208 | ||
1209 | <div class="code"><pre> | |
1210 | Vector *wrap_cross_product(Vector *v1, Vector *v2) { | |
1211 | Vector x = *v1; | |
1212 | Vector y = *v2; | |
1213 | Vector *result; | |
1214 | result = (Vector *) malloc(sizeof(Vector)); | |
1215 | *(result) = cross(x,y); | |
1216 | return result; | |
1217 | } | |
1218 | </pre></div> | |
1219 | ||
1220 | <p> | |
1221 | or if SWIG was run with the <tt>-c++</tt> option:</p> | |
1222 | ||
1223 | <div class="code"><pre> | |
1224 | Vector *wrap_cross(Vector *v1, Vector *v2) { | |
1225 | Vector x = *v1; | |
1226 | Vector y = *v2; | |
1227 | Vector *result = new Vector(cross(x,y)); // Uses default copy constructor | |
1228 | return result; | |
1229 | } | |
1230 | </pre></div> | |
1231 | ||
1232 | <p> | |
1233 | In both cases, SWIG allocates a new object and returns a reference to it. It | |
1234 | is up to the user to delete the returned object when it is no longer | |
1235 | in use. Clearly, this will leak memory if you are unaware of the implicit | |
1236 | memory allocation and don't take steps to free the result. That said, it should be | |
1237 | noted that some language modules can now automatically track newly created objects and | |
1238 | reclaim memory for you. Consult the documentation for each language module for more details. | |
1239 | </p> | |
1240 | ||
1241 | <p> | |
1242 | It should also be noted that the handling of pass/return by value in | |
1243 | C++ has some special cases. For example, the above code fragments | |
1244 | don't work correctly if <tt>Vector</tt> doesn't define a default | |
1245 | constructor. The section on SWIG and C++ has more information about this case. | |
1246 | </p> | |
1247 | ||
1248 | <H3><a name="SWIG_nn24"></a>5.4.3 Linking to structure variables</H3> | |
1249 | ||
1250 | ||
1251 | <p> | |
1252 | When global variables or class members involving structures are | |
1253 | encountered, SWIG handles them as pointers. For example, a global | |
1254 | variable like this</p> | |
1255 | ||
1256 | <div class="code"><pre> | |
1257 | Vector unit_i; | |
1258 | </pre></div> | |
1259 | ||
1260 | <p> | |
1261 | gets mapped to an underlying pair of set/get functions like this :</p> | |
1262 | ||
1263 | <div class="code"><pre> | |
1264 | Vector *unit_i_get() { | |
1265 | return &unit_i; | |
1266 | } | |
1267 | void unit_i_set(Vector *value) { | |
1268 | unit_i = *value; | |
1269 | } | |
1270 | </pre></div> | |
1271 | ||
1272 | <p> | |
1273 | Again some caution is in order. A global variable created in this | |
1274 | manner will show up as a pointer in the target scripting language. It | |
1275 | would be an extremely bad idea to free or destroy such a pointer. Also, | |
1276 | C++ classes must supply a properly defined copy constructor in order for | |
1277 | assignment to work correctly. | |
1278 | </p> | |
1279 | ||
1280 | <H3><a name="SWIG_nn25"></a>5.4.4 Linking to <tt>char *</tt></H3> | |
1281 | ||
1282 | ||
1283 | <p> | |
1284 | When a global variable of type <tt>char *</tt> appears, SWIG uses <tt>malloc()</tt> or | |
1285 | <tt>new</tt> to allocate memory for the new value. Specifically, if you have a variable | |
1286 | like this | |
1287 | </p> | |
1288 | ||
1289 | <div class="code"> | |
1290 | <pre> | |
1291 | char *foo; | |
1292 | </pre> | |
1293 | </div> | |
1294 | ||
1295 | <p> | |
1296 | SWIG generates the following code: | |
1297 | </p> | |
1298 | ||
1299 | <div class="code"> | |
1300 | <pre> | |
1301 | /* C mode */ | |
1302 | void foo_set(char *value) { | |
1303 | if (foo) free(foo); | |
1304 | foo = (char *) malloc(strlen(value)+1); | |
1305 | strcpy(foo,value); | |
1306 | } | |
1307 | ||
1308 | /* C++ mode. When -c++ option is used */ | |
1309 | void foo_set(char *value) { | |
1310 | if (foo) delete [] foo; | |
1311 | foo = new char[strlen(value)+1]; | |
1312 | strcpy(foo,value); | |
1313 | } | |
1314 | </pre> | |
1315 | </div> | |
1316 | ||
1317 | <p> | |
1318 | If this is not the behavior that you want, consider making the variable read-only using the | |
1319 | <tt>%immutable</tt> directive. Alternatively, you might write a short assist-function to set the value | |
1320 | exactly like you want. For example: | |
1321 | </p> | |
1322 | ||
1323 | <div class="code"> | |
1324 | <pre> | |
1325 | %inline %{ | |
1326 | void set_foo(char *value) { | |
1327 | strncpy(foo,value, 50); | |
1328 | } | |
1329 | %} | |
1330 | </pre> | |
1331 | </div> | |
1332 | ||
1333 | <p> | |
1334 | Note: If you write an assist function like this, you will have to call | |
1335 | it as a function from the target scripting language (it does not work | |
1336 | like a variable). For example, in Python you will have to write: | |
1337 | </p> | |
1338 | ||
1339 | <div class="targetlang"> | |
1340 | <pre> | |
1341 | >>> set_foo("Hello World") | |
1342 | </pre> | |
1343 | </div> | |
1344 | ||
1345 | <p> | |
1346 | A common mistake with <tt>char *</tt> variables is to link to a variable declared like this: | |
1347 | </p> | |
1348 | ||
1349 | <div class="code"> | |
1350 | <pre> | |
1351 | char *VERSION = "1.0"; | |
1352 | </pre> | |
1353 | </div> | |
1354 | ||
1355 | <p> | |
1356 | In this case, the variable will be readable, but any attempt to change | |
1357 | the value results in a segmentation or general protection fault. This | |
1358 | is due to the fact that SWIG is trying to release the old value using | |
1359 | <tt>free</tt> or <tt>delete</tt> when the string literal value currently assigned to the variable wasn't | |
1360 | allocated using <tt>malloc()</tt> or <tt>new</tt>. | |
1361 | To fix this behavior, you can | |
1362 | either mark the variable as read-only, write a typemap (as described in Chapter 6), | |
1363 | or write a special set function as shown. Another alternative is to declare the | |
1364 | variable as an array: | |
1365 | </p> | |
1366 | ||
1367 | <div class="code"> | |
1368 | <pre> | |
1369 | char VERSION[64] = "1.0"; | |
1370 | </pre> | |
1371 | </div> | |
1372 | ||
1373 | <p> | |
1374 | When variables of type <tt>const char *</tt> are declared, SWIG still generates functions for setting and | |
1375 | getting the value. However, the default behavior does <em>not</em> release the previous contents (resulting in | |
1376 | a possible memory leak). In fact, you may get a warning message such as this when wrapping such a variable: | |
1377 | </p> | |
1378 | ||
1379 | <div class="shell"> | |
1380 | <pre> | |
1381 | example.i:20. Typemap warning. Setting const char * variable may leak memory | |
1382 | </pre> | |
1383 | </div> | |
1384 | ||
1385 | <p> | |
1386 | The reason for this behavior is that <tt>const char *</tt> variables are often used to point to string literals. | |
1387 | For example: | |
1388 | </p> | |
1389 | ||
1390 | <div class="code"> | |
1391 | <pre> | |
1392 | const char *foo = "Hello World\n"; | |
1393 | </pre> | |
1394 | </div> | |
1395 | ||
1396 | <p> | |
1397 | Therefore, it's a really bad idea to call <tt>free()</tt> on such a | |
1398 | pointer. On the other hand, it <em>is</em> legal to change the | |
1399 | pointer to point to some other value. When setting a variable of this | |
1400 | type, SWIG allocates a new string (using malloc or new) and changes | |
1401 | the pointer to point to the new value. However, repeated | |
1402 | modifications of the value will result in a memory leak since the old | |
1403 | value is not released. | |
1404 | </p> | |
1405 | ||
1406 | ||
1407 | ||
1408 | ||
1409 | <H3><a name="SWIG_nn26"></a>5.4.5 Arrays</H3> | |
1410 | ||
1411 | ||
1412 | <p> | |
1413 | Arrays are fully supported by SWIG, but they are always handled as pointers instead | |
1414 | of mapping them to a special array object or list in the target language. Thus, the | |
1415 | following declarations :</p> | |
1416 | ||
1417 | <div class="code"><pre> | |
1418 | int foobar(int a[40]); | |
1419 | void grok(char *argv[]); | |
1420 | void transpose(double a[20][20]); | |
1421 | </pre></div> | |
1422 | ||
1423 | <p> | |
1424 | are processed as if they were really declared like this: | |
1425 | </p> | |
1426 | ||
1427 | <div class="code"><pre> | |
1428 | int foobar(int *a); | |
1429 | void grok(char **argv); | |
1430 | void transpose(double (*a)[20]); | |
1431 | </pre></div> | |
1432 | ||
1433 | <p> | |
1434 | Like C, SWIG does not perform array bounds checking. | |
1435 | It is up to the | |
1436 | user to make sure the pointer points a suitably allocated region of memory. | |
1437 | </p> | |
1438 | ||
1439 | <p> | |
1440 | Multi-dimensional arrays are transformed into a pointer to an array of one less | |
1441 | dimension. For example: | |
1442 | </p> | |
1443 | ||
1444 | <div class="code"> | |
1445 | <pre> | |
1446 | int [10]; // Maps to int * | |
1447 | int [10][20]; // Maps to int (*)[20] | |
1448 | int [10][20][30]; // Maps to int (*)[20][30] | |
1449 | </pre> | |
1450 | </div> | |
1451 | ||
1452 | <p> | |
1453 | It is important to note that in the C type system, a multidimensional | |
1454 | array <tt>a[][]</tt> is <b>NOT</b> equivalent to a single pointer | |
1455 | <tt>*a</tt> or a double pointer such as <tt>**a</tt>. Instead, a | |
1456 | pointer to an array is used (as shown above) where the actual value of | |
1457 | the pointer is the starting memory location of the array. The | |
1458 | reader is strongly advised to dust off their C book and re-read the | |
1459 | section on arrays before using them with SWIG. | |
1460 | </p> | |
1461 | ||
1462 | <p> | |
1463 | Array variables are supported, but are read-only by default. For example: | |
1464 | </p> | |
1465 | ||
1466 | <div class="code"> | |
1467 | <pre> | |
1468 | int a[100][200]; | |
1469 | </pre> | |
1470 | </div> | |
1471 | ||
1472 | <p> | |
1473 | In this case, reading the variable 'a' returns a pointer of type <tt>int (*)[200]</tt> | |
1474 | that points to the first element of the array <tt>&a[0][0]</tt>. Trying to modify 'a' results | |
1475 | in an error. This is because SWIG does not know how to copy data from the target | |
1476 | language into the array. To work around this limitation, you may want to write | |
1477 | a few simple assist functions like this: | |
1478 | </p> | |
1479 | ||
1480 | <div class="code"> | |
1481 | <pre> | |
1482 | %inline %{ | |
1483 | void a_set(int i, int j, int val) { | |
1484 | a[i][j] = val; | |
1485 | } | |
1486 | int a_get(int i, int j) { | |
1487 | return a[i][j]; | |
1488 | } | |
1489 | %} | |
1490 | </pre> | |
1491 | </div> | |
1492 | ||
1493 | <p> | |
1494 | To dynamically create arrays of various sizes and shapes, it may be useful to write | |
1495 | some helper functions in your interface. For example: | |
1496 | </p> | |
1497 | ||
1498 | <div class="code"> | |
1499 | <pre> | |
1500 | // Some array helpers | |
1501 | %inline %{ | |
1502 | /* Create any sort of [size] array */ | |
1503 | int *int_array(int size) { | |
1504 | return (int *) malloc(size*sizeof(int)); | |
1505 | } | |
1506 | /* Create a two-dimension array [size][10] */ | |
1507 | int (*int_array_10(int size))[10] { | |
1508 | return (int (*)[10]) malloc(size*10*sizeof(int)); | |
1509 | } | |
1510 | %} | |
1511 | </pre> | |
1512 | </div> | |
1513 | ||
1514 | <p> | |
1515 | Arrays of <tt>char</tt> are handled as a special case by SWIG. In this case, strings in the | |
1516 | target language can be stored in the array. For example, if you have a declaration like this, | |
1517 | </p> | |
1518 | ||
1519 | <div class="code"> | |
1520 | <pre> | |
1521 | char pathname[256]; | |
1522 | </pre> | |
1523 | </div> | |
1524 | ||
1525 | <p> | |
1526 | SWIG generates functions for both getting and setting the value that are equivalent to the following | |
1527 | code: | |
1528 | </p> | |
1529 | ||
1530 | <div class="code"> | |
1531 | <pre> | |
1532 | char *pathname_get() { | |
1533 | return pathname; | |
1534 | } | |
1535 | void pathname_set(char *value) { | |
1536 | strncpy(pathname,value,256); | |
1537 | } | |
1538 | </pre> | |
1539 | </div> | |
1540 | ||
1541 | <p> | |
1542 | In the target language, the value can be set like a normal variable. | |
1543 | </p> | |
1544 | ||
1545 | <H3><a name="SWIG_readonly_variables"></a>5.4.6 Creating read-only variables</H3> | |
1546 | ||
1547 | ||
1548 | <p> | |
1549 | A read-only variable can be created by using the <tt>%immutable</tt> | |
1550 | directive as shown :</p> | |
1551 | ||
1552 | <div class="code"><pre> | |
1553 | // File : interface.i | |
1554 | ||
1555 | int a; // Can read/write | |
1556 | %immutable; | |
1557 | int b,c,d // Read only variables | |
1558 | %mutable; | |
1559 | double x,y // read/write | |
1560 | </pre></div> | |
1561 | ||
1562 | <p> | |
1563 | The <tt>%immutable</tt> directive enables read-only mode until it is | |
1564 | explicitly disabled using the <tt>%mutable</tt> directive. As an alternative to turning | |
1565 | read-only mode off and on like this, individual declarations can also be tagged as | |
1566 | immutable. For example: | |
1567 | </p> | |
1568 | ||
1569 | <div class="code"><pre> | |
1570 | %immutable x; // Make x read-only | |
1571 | ... | |
1572 | double x; // Read-only (from earlier %immutable directive) | |
1573 | double y; // Read-write | |
1574 | ... | |
1575 | </pre></div> | |
1576 | ||
1577 | <p> | |
1578 | The <tt>%mutable</tt> and <tt>%immutable</tt> directives are actually | |
1579 | <a href="Customization.html#features">%feature directives</a> defined like this: | |
1580 | </p> | |
1581 | ||
1582 | <div class="code"><pre> | |
1583 | #define %immutable %feature("immutable") | |
1584 | #define %mutable %feature("immutable","") | |
1585 | </pre></div> | |
1586 | ||
1587 | <p> | |
1588 | If you wanted to make all wrapped variables read-only, barring one or two, it might be easier to take this approach: | |
1589 | </p> | |
1590 | ||
1591 | <div class="code"><pre> | |
1592 | %immutable; // Make all variables read-only | |
1593 | %feature("immutable","0") x; // except, make x read/write | |
1594 | ... | |
1595 | double x; | |
1596 | double y; | |
1597 | double z; | |
1598 | ... | |
1599 | </pre></div> | |
1600 | ||
1601 | <p> | |
1602 | Read-only variables are also created when declarations are declared as <tt>const</tt>. | |
1603 | For example: | |
1604 | </p> | |
1605 | ||
1606 | <div class="code"> | |
1607 | <pre> | |
1608 | const int foo; /* Read only variable */ | |
1609 | char * const version="1.0"; /* Read only variable */ | |
1610 | </pre></div> | |
1611 | ||
1612 | <p> | |
1613 | <b>Compatibility note:</b> Read-only access used to be controlled by a pair of directives | |
1614 | <tt>%readonly</tt> and <tt>%readwrite</tt>. Although these directives still work, they | |
1615 | generate a warning message. Simply change the directives to <tt>%immutable;</tt> and | |
1616 | <tt>%mutable;</tt> to silence the warning. Don't forget the extra semicolon! | |
1617 | </p> | |
1618 | ||
1619 | <H3><a name="SWIG_nn28"></a>5.4.7 Renaming and ignoring declarations</H3> | |
1620 | ||
1621 | ||
1622 | <p> | |
1623 | Normally, the name of a C declaration is used when that declaration is | |
1624 | wrapped into the target language. However, this may generate a | |
1625 | conflict with a keyword or already existing function in the scripting | |
1626 | language. To resolve a name conflict, you can use the <tt>%rename</tt> | |
1627 | directive as shown :</p> | |
1628 | ||
1629 | <div class="code"><pre> | |
1630 | // interface.i | |
1631 | ||
1632 | %rename(my_print) print; | |
1633 | extern void print(char *); | |
1634 | ||
1635 | %rename(foo) a_really_long_and_annoying_name; | |
1636 | extern int a_really_long_and_annoying_name; | |
1637 | ||
1638 | </pre></div> | |
1639 | ||
1640 | <p> | |
1641 | SWIG still calls the correct C function, but in this case the | |
1642 | function <tt>print()</tt> will really be called "<tt>my_print()</tt>" | |
1643 | in the target language. </p> | |
1644 | ||
1645 | <p> | |
1646 | The placement of the <tt>%rename</tt> directive is arbitrary as long as it appears | |
1647 | before the declarations to be renamed. A common technique is to write code for | |
1648 | wrapping a header file like this: | |
1649 | </p> | |
1650 | ||
1651 | <div class="code"><pre> | |
1652 | // interface.i | |
1653 | ||
1654 | %rename(my_print) print; | |
1655 | %rename(foo) a_really_long_and_annoying_name; | |
1656 | ||
1657 | %include "header.h" | |
1658 | </pre></div> | |
1659 | ||
1660 | <p> | |
1661 | <tt>%rename </tt>applies a renaming operation to all future | |
1662 | occurrences of a name. The renaming applies to functions, variables, | |
1663 | class and structure names, member functions, and member data. For | |
1664 | example, if you had two-dozen C++ classes, all with a member function | |
1665 | named `print' (which is a keyword in Python), you could rename them | |
1666 | all to `output' by specifying :</p> | |
1667 | ||
1668 | <div class="code"><pre> | |
1669 | %rename(output) print; // Rename all `print' functions to `output' | |
1670 | </pre></div> | |
1671 | ||
1672 | <p> | |
1673 | SWIG does not normally perform any checks to see if the functions it wraps are | |
1674 | already defined in the target scripting language. However, if you are | |
1675 | careful about namespaces and your use of modules, you can usually | |
1676 | avoid these problems.</p> | |
1677 | ||
1678 | <p> | |
1679 | Closely related to <tt>%rename</tt> is the <tt>%ignore</tt> directive. <tt>%ignore</tt> instructs SWIG | |
1680 | to ignore declarations that match a given identifier. For example: | |
1681 | </p> | |
1682 | ||
1683 | <div class="code"> | |
1684 | <pre> | |
1685 | %ignore print; // Ignore all declarations named print | |
1686 | %ignore _HAVE_FOO_H; // Ignore an include guard constant | |
1687 | ... | |
1688 | %include "foo.h" // Grab a header file | |
1689 | ... | |
1690 | </pre> | |
1691 | </div> | |
1692 | ||
1693 | <p> | |
1694 | One use of <tt>%ignore</tt> is to selectively remove certain declarations from a header file without having | |
1695 | to add conditional compilation to the header. However, it should be stressed that this only works for simple | |
1696 | declarations. If you need to remove a whole section of problematic code, the SWIG preprocessor should be used instead. | |
1697 | </p> | |
1698 | ||
1699 | <p> | |
1700 | More powerful variants of <tt>%rename</tt> and <tt>%ignore</tt> directives can be used to help | |
1701 | wrap C++ overloaded functions and methods or C++ methods which use default arguments. This is described in the | |
1702 | <a href="SWIGPlus.html#ambiguity_resolution_renaming">Ambiguity resolution and renaming</a> section in the C++ chapter. | |
1703 | </p> | |
1704 | ||
1705 | <p> | |
1706 | <b>Compatibility note: </b> Older versions of SWIG provided a special <tt>%name</tt> directive for renaming declarations. | |
1707 | For example: | |
1708 | </p> | |
1709 | ||
1710 | <div class="code"> | |
1711 | <pre> | |
1712 | %name(output) extern void print(char *); | |
1713 | </pre> | |
1714 | </div> | |
1715 | ||
1716 | <p> | |
1717 | This directive is still supported, but it is deprecated and should probably be avoided. The <tt>%rename</tt> | |
1718 | directive is more powerful and better supports wrapping of raw header file information. | |
1719 | </p> | |
1720 | ||
1721 | <H3><a name="SWIG_default_args"></a>5.4.8 Default/optional arguments</H3> | |
1722 | ||
1723 | ||
1724 | <p> | |
1725 | SWIG supports default arguments in both C and C++ code. For example: | |
1726 | </p> | |
1727 | ||
1728 | <div class="code"><pre> | |
1729 | int plot(double x, double y, int color=WHITE); | |
1730 | </pre></div> | |
1731 | ||
1732 | <p> | |
1733 | In this case, SWIG generates wrapper code where the | |
1734 | default arguments are optional in the target language. For example, this function could be | |
1735 | used in Tcl as follows :</p> | |
1736 | ||
1737 | <div class="targetlang"><pre> | |
1738 | % plot -3.4 7.5 # Use default value | |
1739 | % plot -3.4 7.5 10 # set color to 10 instead | |
1740 | ||
1741 | </pre></div> | |
1742 | ||
1743 | <p> | |
1744 | Although the ANSI C standard does not allow default arguments, default | |
1745 | arguments specified in a SWIG interface work with both C and C++. | |
1746 | </p> | |
1747 | ||
1748 | <p> | |
1749 | <b>Note:</b> There is a subtle semantic issue concerning the use | |
1750 | of default arguments and the SWIG generated wrapper code. When default | |
1751 | arguments are used in C code, the default values are emitted into the wrappers and the | |
1752 | function is invoked with a full set of arguments. This is different to when wrapping C++ | |
1753 | where an overloaded wrapper method is generated for each defaulted argument. | |
1754 | Please refer to the section on <a href="SWIGPlus.html#SWIGPlus_default_args">default arguments</a> | |
1755 | in the C++ chapter for further details. | |
1756 | </p> | |
1757 | ||
1758 | <H3><a name="SWIG_nn30"></a>5.4.9 Pointers to functions and callbacks</H3> | |
1759 | ||
1760 | ||
1761 | <p> | |
1762 | Occasionally, a C library may include functions that expect to receive | |
1763 | pointers to functions--possibly to serve as callbacks. SWIG | |
1764 | provides full support for function pointers provided that the callback | |
1765 | functions are defined in C and not in the target language. For example, | |
1766 | consider a function like this: | |
1767 | </p> | |
1768 | ||
1769 | <div class="code"><pre> | |
1770 | int binary_op(int a, int b, int (*op)(int,int)); | |
1771 | </pre></div> | |
1772 | ||
1773 | <p> | |
1774 | When you first wrap something like this into an extension module, you | |
1775 | may find the function to be impossible to use. For instance, in Python: | |
1776 | </p> | |
1777 | ||
1778 | <div class="targetlang"><pre> | |
1779 | >>> def add(x,y): | |
1780 | ... return x+y | |
1781 | ... | |
1782 | >>> binary_op(3,4,add) | |
1783 | Traceback (most recent call last): | |
1784 | File "<stdin>", line 1, in ? | |
1785 | TypeError: Type error. Expected _p_f_int_int__int | |
1786 | >>> | |
1787 | </pre> | |
1788 | </div> | |
1789 | ||
1790 | <p> | |
1791 | The reason for this error is that SWIG doesn't know how to map a scripting | |
1792 | language function into a C callback. However, existing C functions can | |
1793 | be used as arguments provided you install them as constants. | |
1794 | One way to do this is to use the <tt>%constant</tt> directive like this: | |
1795 | </p> | |
1796 | ||
1797 | <div class="code"><pre> | |
1798 | /* Function with a callback */ | |
1799 | int binary_op(int a, int b, int (*op)(int,int)); | |
1800 | ||
1801 | /* Some callback functions */ | |
1802 | %constant int add(int,int); | |
1803 | %constant int sub(int,int); | |
1804 | %constant int mul(int,int); | |
1805 | </pre></div> | |
1806 | ||
1807 | <p> | |
1808 | In this case, <tt>add</tt>, <tt>sub</tt>, and <tt>mul</tt> become function pointer | |
1809 | constants in the target scripting language. This allows you to use them as follows: | |
1810 | </p> | |
1811 | ||
1812 | <div class="targetlang"> | |
1813 | <pre> | |
1814 | >>> binary_op(3,4,add) | |
1815 | 7 | |
1816 | >>> binary_op(3,4,mul) | |
1817 | 12 | |
1818 | >>> | |
1819 | </pre> | |
1820 | </div> | |
1821 | ||
1822 | <p> | |
1823 | Unfortunately, by declaring the callback functions as constants, they are no longer accesible | |
1824 | as functions. For example: | |
1825 | </p> | |
1826 | ||
1827 | <div class="targetlang"> | |
1828 | <pre> | |
1829 | >>> add(3,4) | |
1830 | Traceback (most recent call last): | |
1831 | File "<stdin>", line 1, in ? | |
1832 | TypeError: object is not callable: '_ff020efc_p_f_int_int__int' | |
1833 | >>> | |
1834 | </pre> | |
1835 | </div> | |
1836 | ||
1837 | <p> | |
1838 | If you want to make a function available as both a callback function and a function, you | |
1839 | can use the <tt>%callback</tt> and <tt>%nocallback</tt> directives like this: | |
1840 | </p> | |
1841 | ||
1842 | <div class="code"> | |
1843 | <pre> | |
1844 | /* Function with a callback */ | |
1845 | int binary_op(int a, int b, int (*op)(int,int)); | |
1846 | ||
1847 | /* Some callback functions */ | |
1848 | %callback("%s_cb") | |
1849 | int add(int,int); | |
1850 | int sub(int,int); | |
1851 | int mul(int,int); | |
1852 | %nocallback | |
1853 | </pre></div> | |
1854 | ||
1855 | <p> | |
1856 | The argument to <tt>%callback</tt> is a printf-style format string that | |
1857 | specifies the naming convention for the callback constants (<tt>%s</tt> gets replaced | |
1858 | by the function name). The callback mode remains in effect until it is explicitly | |
1859 | disabled using <tt>%nocallback</tt>. When you do this, the interface now works as follows: | |
1860 | </p> | |
1861 | ||
1862 | <div class="targetlang"> | |
1863 | <pre> | |
1864 | >>> binary_op(3,4,add_cb) | |
1865 | 7 | |
1866 | >>> binary_op(3,4,mul_cb) | |
1867 | 12 | |
1868 | >>> add(3,4) | |
1869 | 7 | |
1870 | >>> mul(3,4) | |
1871 | 12 | |
1872 | </pre> | |
1873 | </div> | |
1874 | ||
1875 | <p> | |
1876 | Notice that when the function is used as a callback, special names | |
1877 | such as <tt>add_cb</tt> is used instead. To call the function | |
1878 | normally, just use the original function name such as <tt>add()</tt>. | |
1879 | </p> | |
1880 | ||
1881 | <p> | |
1882 | SWIG provides a number of extensions to standard C printf formatting | |
1883 | that may be useful in this context. For instance, the following | |
1884 | variation installs the callbacks as all upper-case constants such as | |
1885 | <tt>ADD</tt>, <tt>SUB</tt>, and <tt>MUL</tt>: | |
1886 | </p> | |
1887 | ||
1888 | <div class="code"><pre> | |
1889 | /* Some callback functions */ | |
1890 | %callback("%(upper)s") | |
1891 | int add(int,int); | |
1892 | int sub(int,int); | |
1893 | int mul(int,int); | |
1894 | %nocallback | |
1895 | </pre></div> | |
1896 | ||
1897 | <p> | |
1898 | A format string of <tt>"%(lower)s"</tt> converts all characters to lower-case. | |
1899 | A string of <tt>"%(title)s"</tt> capitalizes the first character and converts the | |
1900 | rest to lower case. | |
1901 | </p> | |
1902 | ||
1903 | <p> | |
1904 | And now, a final note about function pointer support. Although SWIG | |
1905 | does not normally allow callback functions to be written in the target language, this | |
1906 | can be accomplished with the use of typemaps and other advanced SWIG features. | |
1907 | This is described in a later chapter. | |
1908 | </p> | |
1909 | ||
1910 | <H2><a name="SWIG_nn31"></a>5.5 Structures and unions</H2> | |
1911 | ||
1912 | ||
1913 | <p> | |
1914 | This section describes the behavior of SWIG when processing ANSI C structures and union declarations. Extensions to | |
1915 | handle C++ are described in the next section. | |
1916 | </p> | |
1917 | ||
1918 | <p> | |
1919 | If SWIG encounters the definition of a structure or union, it | |
1920 | creates a set of accessor functions. Although SWIG does not need | |
1921 | structure definitions to build an interface, providing definitions | |
1922 | make it possible to access structure members. The accessor functions | |
1923 | generated by SWIG simply take a pointer to an object and allow access | |
1924 | to an individual member. For example, the declaration :</p> | |
1925 | ||
1926 | <div class="code"><pre> | |
1927 | struct Vector { | |
1928 | double x,y,z; | |
1929 | } | |
1930 | ||
1931 | </pre></div> | |
1932 | ||
1933 | <p> | |
1934 | gets transformed into the following set of accessor functions :</p> | |
1935 | ||
1936 | <div class="code"><pre> | |
1937 | double Vector_x_get(struct Vector *obj) { | |
1938 | return obj->x; | |
1939 | } | |
1940 | double Vector_y_get(struct Vector *obj) { | |
1941 | return obj->y; | |
1942 | } | |
1943 | double Vector_z_get(struct Vector *obj) { | |
1944 | return obj->z; | |
1945 | } | |
1946 | void Vector_x_set(struct Vector *obj, double value) { | |
1947 | obj->x = value; | |
1948 | } | |
1949 | void Vector_y_set(struct Vector *obj, double value) { | |
1950 | obj->y = value; | |
1951 | } | |
1952 | void Vector_z_set(struct Vector *obj, double value) { | |
1953 | obj->z = value; | |
1954 | } | |
1955 | </pre></div> | |
1956 | ||
1957 | <p> | |
1958 | In addition, SWIG creates default constructor and destructor functions if none are | |
1959 | defined in the interface. For example: | |
1960 | </p> | |
1961 | ||
1962 | <div class="code"><pre> | |
1963 | struct Vector *new_Vector() { | |
1964 | return (Vector *) calloc(1,sizeof(struct Vector)); | |
1965 | } | |
1966 | void delete_Vector(struct Vector *obj) { | |
1967 | free(obj); | |
1968 | } | |
1969 | </pre> | |
1970 | </div> | |
1971 | ||
1972 | <p> | |
1973 | Using these low-level accessor functions, an object can be minimally manipulated from the target | |
1974 | language using code like this: | |
1975 | </p> | |
1976 | ||
1977 | <div class="code"> | |
1978 | <pre> | |
1979 | v = new_Vector() | |
1980 | Vector_x_set(v,2) | |
1981 | Vector_y_set(v,10) | |
1982 | Vector_z_set(v,-5) | |
1983 | ... | |
1984 | delete_Vector(v) | |
1985 | </pre> | |
1986 | </div> | |
1987 | ||
1988 | <p> | |
1989 | However, most of SWIG's language modules also provide a high-level interface that is more convenient. Keep reading. | |
1990 | </p> | |
1991 | ||
1992 | <H3><a name="SWIG_nn32"></a>5.5.1 Typedef and structures</H3> | |
1993 | ||
1994 | ||
1995 | <p> | |
1996 | SWIG supports the following construct which is quite common in C | |
1997 | programs :</p> | |
1998 | ||
1999 | <div class="code"><pre> | |
2000 | typedef struct { | |
2001 | double x,y,z; | |
2002 | } Vector; | |
2003 | ||
2004 | </pre></div> | |
2005 | ||
2006 | <p> | |
2007 | When encountered, SWIG assumes that the name of the object is `Vector' | |
2008 | and creates accessor functions like before. The only difference is | |
2009 | that the use of <tt>typedef</tt> allows SWIG to drop the | |
2010 | <tt>struct</tt> keyword on its generated code. For example: | |
2011 | </p> | |
2012 | ||
2013 | <div class="code"> | |
2014 | <pre> | |
2015 | double Vector_x_get(Vector *obj) { | |
2016 | return obj->x; | |
2017 | } | |
2018 | </pre> | |
2019 | </div> | |
2020 | ||
2021 | <p> | |
2022 | If two different names are used like this :</p> | |
2023 | ||
2024 | <div class="code"><pre> | |
2025 | typedef struct vector_struct { | |
2026 | double x,y,z; | |
2027 | } Vector; | |
2028 | ||
2029 | </pre></div> | |
2030 | ||
2031 | <p> | |
2032 | the name <tt>Vector</tt> is used instead of <tt>vector_struct</tt> since | |
2033 | this is more typical C programming style. If declarations defined later in the interface use the type <tt>struct | |
2034 | vector_struct</tt>, SWIG knows that this is the same as | |
2035 | <tt>Vector</tt> and it generates the appropriate type-checking code. | |
2036 | </p> | |
2037 | ||
2038 | <H3><a name="SWIG_nn33"></a>5.5.2 Character strings and structures</H3> | |
2039 | ||
2040 | ||
2041 | <p> | |
2042 | Structures involving character strings require some care. SWIG assumes | |
2043 | that all members of type <tt>char *</tt> have been dynamically | |
2044 | allocated using <tt>malloc()</tt> and that they are NULL-terminated | |
2045 | ASCII strings. When such a member is modified, the previously contents | |
2046 | will be released, and the new contents allocated. For example :</p> | |
2047 | ||
2048 | <div class="code"><pre> | |
2049 | %module mymodule | |
2050 | ... | |
2051 | struct Foo { | |
2052 | char *name; | |
2053 | ... | |
2054 | } | |
2055 | ||
2056 | </pre></div> | |
2057 | ||
2058 | <p> | |
2059 | This results in the following accessor functions :</p> | |
2060 | ||
2061 | <div class="code"><pre> | |
2062 | char *Foo_name_get(Foo *obj) { | |
2063 | return Foo->name; | |
2064 | } | |
2065 | ||
2066 | char *Foo_name_set(Foo *obj, char *c) { | |
2067 | if (obj->name) free(obj->name); | |
2068 | obj->name = (char *) malloc(strlen(c)+1); | |
2069 | strcpy(obj->name,c); | |
2070 | return obj->name; | |
2071 | } | |
2072 | </pre></div> | |
2073 | ||
2074 | <p> | |
2075 | If this behavior differs from what you need in your applications, | |
2076 | the SWIG "memberin" typemap can be used to change it. See the | |
2077 | typemaps chapter for further details. | |
2078 | </p> | |
2079 | ||
2080 | <p> | |
2081 | Note: If the <tt>-c++</tt> option is used, <tt>new</tt> and <tt>delete</tt> are used to | |
2082 | perform memory allocation. | |
2083 | </p> | |
2084 | ||
2085 | <H3><a name="SWIG_nn34"></a>5.5.3 Array members</H3> | |
2086 | ||
2087 | ||
2088 | <p> | |
2089 | Arrays may appear as the members of structures, but they will be | |
2090 | read-only. SWIG will write an accessor function that returns the | |
2091 | pointer to the first element of the array, but will not write a | |
2092 | function to change the contents of the array itself. | |
2093 | When this | |
2094 | situation is detected, SWIG may generate a warning message such as the | |
2095 | following :</p> | |
2096 | ||
2097 | <div class="shell"><pre> | |
2098 | interface.i:116. Warning. Array member will be read-only | |
2099 | </pre></div> | |
2100 | ||
2101 | <p> | |
2102 | To eliminate the warning message, typemaps can be used, but this is | |
2103 | discussed in a later chapter. In many cases, the warning message is | |
2104 | harmless. | |
2105 | </p> | |
2106 | ||
2107 | <H3><a name="SWIG_nn35"></a>5.5.4 Structure data members</H3> | |
2108 | ||
2109 | ||
2110 | <p> | |
2111 | Occasionally, a structure will contain data members that are themselves structures. For example: | |
2112 | </p> | |
2113 | ||
2114 | <div class="code"> | |
2115 | <pre> | |
2116 | typedef struct Foo { | |
2117 | int x; | |
2118 | } Foo; | |
2119 | ||
2120 | typedef struct Bar { | |
2121 | int y; | |
2122 | Foo f; /* struct member */ | |
2123 | } Bar; | |
2124 | </pre> | |
2125 | </div> | |
2126 | ||
2127 | <p> | |
2128 | When a structure member is wrapped, it is always handled as a pointer. | |
2129 | For example: | |
2130 | </p> | |
2131 | ||
2132 | <div class="code"> | |
2133 | <pre> | |
2134 | Foo *Bar_f_get(Bar *b) { | |
2135 | return &b->f; | |
2136 | } | |
2137 | void Bar_f_set(Bar *b, Foo *value) { | |
2138 | b->f = *value; | |
2139 | } | |
2140 | </pre> | |
2141 | </div> | |
2142 | ||
2143 | <p> | |
2144 | The reasons for this are somewhat subtle but have to do with the | |
2145 | problem of modifying and accessing data inside the data member. For | |
2146 | example, suppose you wanted to modify the value of <tt>f.x</tt> | |
2147 | of a <tt>Bar</tt> object like this: | |
2148 | </p> | |
2149 | ||
2150 | <div class="code"> | |
2151 | <pre> | |
2152 | Bar *b; | |
2153 | b->f.x = 37; | |
2154 | </pre> | |
2155 | </div> | |
2156 | ||
2157 | <p> | |
2158 | Translating this assignment to function calls (as would be used inside the scripting | |
2159 | language interface) results in the following code: | |
2160 | </p> | |
2161 | ||
2162 | <div class="code"> | |
2163 | <pre> | |
2164 | Bar *b; | |
2165 | Foo_x_set(Bar_f_get(b),37); | |
2166 | </pre> | |
2167 | </div> | |
2168 | ||
2169 | <p> | |
2170 | In this code, if the <tt>Bar_f_get()</tt> function were to return a <tt>Foo</tt> instead of a | |
2171 | <tt>Foo *</tt>, then the resulting modification would be applied to a <em>copy</em> of <tt>f</tt> and not | |
2172 | the data member <tt>f</tt> itself. Clearly that's not what you want! | |
2173 | </p> | |
2174 | ||
2175 | <p> | |
2176 | It should be noted that this transformation to pointers only occurs if SWIG knows that a data member | |
2177 | is a structure or class. For instance, if you had a structure like this, | |
2178 | </p> | |
2179 | ||
2180 | <div class="code"> | |
2181 | <pre> | |
2182 | struct Foo { | |
2183 | WORD w; | |
2184 | }; | |
2185 | </pre> | |
2186 | </div> | |
2187 | ||
2188 | <p> | |
2189 | and nothing was known about <tt>WORD</tt>, then SWIG will generate more normal accessor functions | |
2190 | like this: | |
2191 | </p> | |
2192 | ||
2193 | <div class="code"> | |
2194 | <pre> | |
2195 | WORD Foo_w_get(Foo *f) { | |
2196 | return f->w; | |
2197 | } | |
2198 | void Foo_w_set(FOO *f, WORD value) { | |
2199 | f->w = value; | |
2200 | } | |
2201 | </pre> | |
2202 | </div> | |
2203 | ||
2204 | <p> | |
2205 | <b>Compatibility Note: </b> SWIG-1.3.11 and earlier releases transformed all non-primitive member datatypes | |
2206 | to pointers. Starting in SWIG-1.3.12, this transformation <em>only</em> occurs if a datatype is known to be a structure, | |
2207 | class, or union. This is unlikely to break existing code. However, if you need to tell SWIG that an undeclared | |
2208 | datatype is really a struct, simply use a forward struct declaration such as <tt>"struct Foo;"</tt>. | |
2209 | </p> | |
2210 | ||
2211 | <H3><a name="SWIG_nn36"></a>5.5.5 C constructors and destructors </H3> | |
2212 | ||
2213 | ||
2214 | <p> | |
2215 | When wrapping structures, it is generally useful to have a mechanism | |
2216 | for creating and destroying objects. If you don't do anything, SWIG | |
2217 | will automatically generate functions for creating and destroying | |
2218 | objects using <tt>malloc()</tt> and <tt>free()</tt>. Note: the use of | |
2219 | <tt>malloc()</tt> only applies when SWIG is used on C code (i.e., when the | |
2220 | <tt>-c++</tt> option is <em>not</em> supplied on the command line). C++ is handled | |
2221 | differently. | |
2222 | </p> | |
2223 | ||
2224 | <p> | |
2225 | If you don't want SWIG to generate constructors and destructors, you | |
2226 | can use the <tt>%nodefault</tt> directive or the <tt>-no_default</tt> | |
2227 | command line option. For example: | |
2228 | </p> | |
2229 | ||
2230 | <div class="shell"><pre> | |
2231 | swig -no_default example.i | |
2232 | </pre></div> | |
2233 | ||
2234 | <p> | |
2235 | or | |
2236 | </p> | |
2237 | ||
2238 | <div class="code"><pre> | |
2239 | %module foo | |
2240 | ... | |
2241 | %nodefault; // Don't create default constructors/destructors | |
2242 | ... declarations ... | |
2243 | %makedefault; // Reenable default constructors/destructors | |
2244 | </pre></div> | |
2245 | ||
2246 | <p> | |
2247 | If you need more precise control, <tt>%nodefault</tt> can selectively target individual structure | |
2248 | definitions. For example: | |
2249 | </p> | |
2250 | ||
2251 | <div class="code"> | |
2252 | <pre> | |
2253 | %nodefault Foo; // No default constructor/destructors for Foo | |
2254 | ... | |
2255 | struct Foo { // No default generated. | |
2256 | }; | |
2257 | ||
2258 | struct Bar { // Default constructor/destructor generated. | |
2259 | }; | |
2260 | </pre> | |
2261 | </div> | |
2262 | ||
2263 | <p> | |
2264 | <b>Compatibility note:</b> Prior to SWIG-1.3.7, SWIG did not generate default constructors | |
2265 | or destructors unless you explicitly turned them on using <tt>-make_default</tt>. | |
2266 | However, it appears that most users want to have constructor and destructor functions so it | |
2267 | has now been enabled as the default behavior. | |
2268 | </p> | |
2269 | ||
2270 | <H3><a name="SWIG_adding_member_functions"></a>5.5.6 Adding member functions to C structures</H3> | |
2271 | ||
2272 | ||
2273 | <p> | |
2274 | Most languages provide a mechanism for creating classes and | |
2275 | supporting object oriented programming. From a C standpoint, object | |
2276 | oriented programming really just boils down to the process of | |
2277 | attaching functions to structures. These functions normally operate | |
2278 | on an instance of the structure (or object). Although there is a | |
2279 | natural mapping of C++ to such a scheme, there is no direct mechanism | |
2280 | for utilizing it with C code. However, SWIG provides a special | |
2281 | <tt>%extend</tt> directive that makes it possible to attach | |
2282 | methods to C structures for purposes of building an object oriented | |
2283 | interface. Suppose you have a C header file with | |
2284 | the following declaration :</p> | |
2285 | ||
2286 | <div class="code"><pre> | |
2287 | /* file : vector.h */ | |
2288 | ... | |
2289 | typedef struct { | |
2290 | double x,y,z; | |
2291 | } Vector; | |
2292 | ||
2293 | </pre></div> | |
2294 | ||
2295 | <p> | |
2296 | You can make a <tt>Vector</tt> look alot like a class by writing a SWIG interface like this: | |
2297 | </p> | |
2298 | ||
2299 | <div class="code"><pre> | |
2300 | // file : vector.i | |
2301 | %module mymodule | |
2302 | %{ | |
2303 | #include "vector.h" | |
2304 | %} | |
2305 | ||
2306 | %include vector.h // Just grab original C header file | |
2307 | %extend Vector { // Attach these functions to struct Vector | |
2308 | Vector(double x, double y, double z) { | |
2309 | Vector *v; | |
2310 | v = (Vector *) malloc(sizeof(Vector)); | |
2311 | v->x = x; | |
2312 | v->y = y; | |
2313 | v->z = z; | |
2314 | return v; | |
2315 | } | |
2316 | ~Vector() { | |
2317 | free(self); | |
2318 | } | |
2319 | double magnitude() { | |
2320 | return sqrt(self->x*self->x+self->y*self->y+self->z*self->z); | |
2321 | } | |
2322 | void print() { | |
2323 | printf("Vector [%g, %g, %g]\n", self->x,self->y,self->z); | |
2324 | } | |
2325 | }; | |
2326 | ||
2327 | </pre></div> | |
2328 | ||
2329 | <p> | |
2330 | Now, when used with proxy classes in Python, you can do things like | |
2331 | this :</p> | |
2332 | ||
2333 | <div class="targetlang"><pre> | |
2334 | >>> v = Vector(3,4,0) # Create a new vector | |
2335 | >>> print v.magnitude() # Print magnitude | |
2336 | 5.0 | |
2337 | >>> v.print() # Print it out | |
2338 | [ 3, 4, 0 ] | |
2339 | >>> del v # Destroy it | |
2340 | </pre></div> | |
2341 | ||
2342 | <p> | |
2343 | The <tt>%extend</tt> directive can also be used inside the definition | |
2344 | of the Vector structure. For example:</p> | |
2345 | ||
2346 | <div class="code"><pre> | |
2347 | // file : vector.i | |
2348 | %module mymodule | |
2349 | %{ | |
2350 | #include "vector.h" | |
2351 | %} | |
2352 | ||
2353 | typedef struct { | |
2354 | double x,y,z; | |
2355 | %extend { | |
2356 | Vector(double x, double y, double z) { ... } | |
2357 | ~Vector() { ... } | |
2358 | ... | |
2359 | } | |
2360 | } Vector; | |
2361 | </pre></div> | |
2362 | ||
2363 | <p> | |
2364 | Finally, <tt>%extend</tt> can be used to access externally written | |
2365 | functions provided they follow the naming convention used in this | |
2366 | example :</p> | |
2367 | ||
2368 | <div class="code"><pre> | |
2369 | /* File : vector.c */ | |
2370 | /* Vector methods */ | |
2371 | #include "vector.h" | |
2372 | Vector *new_Vector(double x, double y, double z) { | |
2373 | Vector *v; | |
2374 | v = (Vector *) malloc(sizeof(Vector)); | |
2375 | v->x = x; | |
2376 | v->y = y; | |
2377 | v->z = z; | |
2378 | return v; | |
2379 | } | |
2380 | void delete_Vector(Vector *v) { | |
2381 | free(v); | |
2382 | } | |
2383 | ||
2384 | double Vector_magnitude(Vector *v) { | |
2385 | return sqrt(v->x*v->x+v->y*v->y+v->z*v->z); | |
2386 | } | |
2387 | ||
2388 | // File : vector.i | |
2389 | // Interface file | |
2390 | %module mymodule | |
2391 | %{ | |
2392 | #include "vector.h" | |
2393 | %} | |
2394 | ||
2395 | typedef struct { | |
2396 | double x,y,z; | |
2397 | %extend { | |
2398 | Vector(int,int,int); // This calls new_Vector() | |
2399 | ~Vector(); // This calls delete_Vector() | |
2400 | double magnitude(); // This will call Vector_magnitude() | |
2401 | ... | |
2402 | } | |
2403 | } Vector; | |
2404 | </pre> | |
2405 | </div> | |
2406 | ||
2407 | <p> | |
2408 | A little known feature of the <tt>%extend</tt> directive is that | |
2409 | it can also be used to add synthesized attributes or to modify the | |
2410 | behavior of existing data attributes. For example, suppose you wanted | |
2411 | to make <tt>magnitude</tt> a read-only attribute of <tt>Vector</tt> | |
2412 | instead of a method. To do this, you might write some code like this: | |
2413 | </p> | |
2414 | ||
2415 | <div class="code"> | |
2416 | <pre> | |
2417 | // Add a new attribute to Vector | |
2418 | %extend Vector { | |
2419 | const double magnitude; | |
2420 | } | |
2421 | // Now supply the implementation of the Vector_magnitude_get function | |
2422 | %{ | |
2423 | const double Vector_magnitude_get(Vector *v) { | |
2424 | return (const double) return sqrt(v->x*v->x+v->y*v->y+v->z*v->z); | |
2425 | } | |
2426 | %} | |
2427 | ||
2428 | </pre> | |
2429 | </div> | |
2430 | ||
2431 | <p> | |
2432 | Now, for all practial purposes, <tt>magnitude</tt> will appear like an attribute | |
2433 | of the object. | |
2434 | </p> | |
2435 | ||
2436 | <p> | |
2437 | A similar technique can also be used to work with problematic data members. | |
2438 | For example, consider this interface: | |
2439 | </p> | |
2440 | ||
2441 | <div class="code"> | |
2442 | <pre> | |
2443 | struct Person { | |
2444 | char name[50]; | |
2445 | ... | |
2446 | } | |
2447 | </pre> | |
2448 | </div> | |
2449 | ||
2450 | <p> | |
2451 | By default, the <tt>name</tt> attribute is read-only because SWIG does not | |
2452 | normally know how to modify arrays. However, you can rewrite the interface | |
2453 | as follows to change this: | |
2454 | </p> | |
2455 | ||
2456 | <div class="code"> | |
2457 | <pre> | |
2458 | struct Person { | |
2459 | %extend { | |
2460 | char *name; | |
2461 | } | |
2462 | ... | |
2463 | } | |
2464 | ||
2465 | // Specific implementation of set/get functions | |
2466 | %{ | |
2467 | char *Person_name_get(Person *p) { | |
2468 | return p->name; | |
2469 | } | |
2470 | void Person_name_set(Person *p, char *val) { | |
2471 | strncpy(p->name,val,50); | |
2472 | } | |
2473 | %} | |
2474 | </pre> | |
2475 | </div> | |
2476 | ||
2477 | <p> | |
2478 | Finally, it should be stressed that even though <tt>%extend</tt> | |
2479 | can be used to add new data members, these new members can not require | |
2480 | the allocation of additional storage in the object (e.g., their values must | |
2481 | be entirely synthesized from existing attributes of the structure). | |
2482 | </p> | |
2483 | ||
2484 | <p> | |
2485 | <b>Compatibility note:</b> The <tt>%extend</tt> directive is a new | |
2486 | name for the <tt>%addmethods</tt> directive. Since <tt>%addmethods</tt> could | |
2487 | be used to extend a structure with more than just methods, a more suitable | |
2488 | directive name has been chosen. | |
2489 | </p> | |
2490 | ||
2491 | <H3><a name="SWIG_nn38"></a>5.5.7 Nested structures</H3> | |
2492 | ||
2493 | ||
2494 | <p> | |
2495 | Occasionally, a C program will involve structures like this :</p> | |
2496 | ||
2497 | <div class="code"><pre> | |
2498 | typedef struct Object { | |
2499 | int objtype; | |
2500 | union { | |
2501 | int ivalue; | |
2502 | double dvalue; | |
2503 | char *strvalue; | |
2504 | void *ptrvalue; | |
2505 | } intRep; | |
2506 | } Object; | |
2507 | ||
2508 | </pre></div> | |
2509 | ||
2510 | <p> | |
2511 | When SWIG encounters this, it performs a structure splitting operation | |
2512 | that transforms the declaration into the equivalent of the | |
2513 | following:</p> | |
2514 | ||
2515 | <div class="code"><pre> | |
2516 | typedef union { | |
2517 | int ivalue; | |
2518 | double dvalue; | |
2519 | char *strvalue; | |
2520 | void *ptrvalue; | |
2521 | } Object_intRep; | |
2522 | ||
2523 | typedef struct Object { | |
2524 | int objType; | |
2525 | Object_intRep intRep; | |
2526 | } Object; | |
2527 | ||
2528 | </pre></div> | |
2529 | ||
2530 | <p> | |
2531 | SWIG will then create an <tt>Object_intRep</tt> structure for use inside | |
2532 | the interface file. Accessor functions will be created for both | |
2533 | structures. In this case, functions like this would be created :</p> | |
2534 | ||
2535 | <div class="code"><pre> | |
2536 | Object_intRep *Object_intRep_get(Object *o) { | |
2537 | return (Object_intRep *) &o->intRep; | |
2538 | } | |
2539 | int Object_intRep_ivalue_get(Object_intRep *o) { | |
2540 | return o->ivalue; | |
2541 | } | |
2542 | int Object_intRep_ivalue_set(Object_intRep *o, int value) { | |
2543 | return (o->ivalue = value); | |
2544 | } | |
2545 | double Object_intRep_dvalue_get(Object_intRep *o) { | |
2546 | return o->dvalue; | |
2547 | } | |
2548 | ... etc ... | |
2549 | ||
2550 | </pre></div> | |
2551 | ||
2552 | <p> | |
2553 | Although this process is a little hairy, it works like you would expect in the | |
2554 | target scripting language--especially when proxy classes are used. For instance, in Perl: | |
2555 | </p> | |
2556 | ||
2557 | <div class="targetlang"><pre> | |
2558 | # Perl5 script for accessing nested member | |
2559 | $o = CreateObject(); # Create an object somehow | |
2560 | $o->{intRep}->{ivalue} = 7 # Change value of o.intRep.ivalue | |
2561 | </pre></div> | |
2562 | ||
2563 | <p> | |
2564 | If you have a lot nested structure declarations, it is | |
2565 | advisable to double-check them after running SWIG. Although, | |
2566 | there is a good chance that they will work, you may have to | |
2567 | modify the interface file in certain cases. | |
2568 | ||
2569 | <H3><a name="SWIG_nn39"></a>5.5.8 Other things to note about structure wrapping</H3> | |
2570 | ||
2571 | ||
2572 | <p> | |
2573 | SWIG doesn't care if the declaration of a structure in a <tt>.i</tt> file exactly matches | |
2574 | that used in the underlying C code (except in the case of nested | |
2575 | structures). For this reason, there are no problems omitting | |
2576 | problematic members or simply omitting the structure definition | |
2577 | altogether. If you are happy passing pointers around, this can | |
2578 | be done without ever giving SWIG a structure definition.</p> | |
2579 | ||
2580 | <p> | |
2581 | Starting with SWIG1.3, a number of improvements have been made to SWIG's | |
2582 | code generator. Specifically, even though structure access has been described | |
2583 | in terms of high-level accessor functions such as this, | |
2584 | </p> | |
2585 | ||
2586 | <div class="code"> | |
2587 | <pre> | |
2588 | double Vector_x_get(Vector *v) { | |
2589 | return v->x; | |
2590 | } | |
2591 | </pre> | |
2592 | </div> | |
2593 | ||
2594 | <p> | |
2595 | most of the generated code is actually inlined directly into wrapper | |
2596 | functions. Therefore, no function <tt>Vector_x_get()</tt> actually | |
2597 | exists in the generated wrapper file. For example, when creating a Tcl module, | |
2598 | the following function is generated instead: | |
2599 | </p> | |
2600 | ||
2601 | <div class="code"> | |
2602 | <pre> | |
2603 | static int | |
2604 | _wrap_Vector_x_get(ClientData clientData, Tcl_Interp *interp, | |
2605 | int objc, Tcl_Obj *CONST objv[]) { | |
2606 | struct Vector *arg1 ; | |
2607 | double result ; | |
2608 | ||
2609 | if (SWIG_GetArgs(interp, objc, objv,"p:Vector_x_get self ",&arg0, | |
2610 | SWIGTYPE_p_Vector) == TCL_ERROR) | |
2611 | return TCL_ERROR; | |
2612 | result = (double ) (arg1->x); | |
2613 | Tcl_SetObjResult(interp,Tcl_NewDoubleObj((double) result)); | |
2614 | return TCL_OK; | |
2615 | } | |
2616 | </pre> | |
2617 | </div> | |
2618 | ||
2619 | <p> | |
2620 | The only exception to this rule are methods defined with <tt>%extend</tt>. In this | |
2621 | case, the added code is contained in a separate function. | |
2622 | </p> | |
2623 | ||
2624 | <p> | |
2625 | Finally, it is important to note that most language modules may choose to | |
2626 | build a more advanced interface. Although you may never use the low-level | |
2627 | interface described here, most of SWIG's language modules use it in | |
2628 | some way or another. | |
2629 | </p> | |
2630 | ||
2631 | <H2><a name="SWIG_nn40"></a>5.6 Code Insertion</H2> | |
2632 | ||
2633 | ||
2634 | <p> | |
2635 | Sometimes it is necessary to insert special code into the resulting | |
2636 | wrapper file generated by SWIG. For example, you may want to include | |
2637 | additional C code to perform initialization or other operations. | |
2638 | There are four common ways to insert code, but it's useful to know how the | |
2639 | output of SWIG is structured first.</p> | |
2640 | ||
2641 | <H3><a name="SWIG_nn41"></a>5.6.1 The output of SWIG</H3> | |
2642 | ||
2643 | ||
2644 | <p> | |
2645 | When SWIG creates its output file, it is broken up into four sections | |
2646 | corresponding to runtime code, headers, wrapper functions, and module | |
2647 | initialization code (in that order). | |
2648 | </p> | |
2649 | ||
2650 | <ul> | |
2651 | <li><b>Runtime code</b>. <br> | |
2652 | This code is internal to SWIG and is used to include | |
2653 | type-checking and other support functions that are used by the rest of the module. | |
2654 | </li> | |
2655 | ||
2656 | <li><b>Header section</b>. <br> | |
2657 | This is user-defined support code that has been included by | |
2658 | the <tt>%{ ... %}</tt> directive. Usually this consists of header files and | |
2659 | other helper functions. | |
2660 | </li> | |
2661 | ||
2662 | <li><b>Wrapper code</b>. <br> | |
2663 | These are the wrappers generated automatically by SWIG. | |
2664 | </li> | |
2665 | ||
2666 | <li><b>Module initialization</b>. <br> | |
2667 | The function generated by SWIG to initialize | |
2668 | the module upon loading. | |
2669 | </li> | |
2670 | </ul> | |
2671 | ||
2672 | <H3><a name="SWIG_nn42"></a>5.6.2 Code insertion blocks</H3> | |
2673 | ||
2674 | ||
2675 | <p> | |
2676 | Code is inserted into the appropriate code section by using one | |
2677 | of the following code insertion directives: | |
2678 | </p> | |
2679 | ||
2680 | <div class="code"> | |
2681 | <pre> | |
2682 | %runtime %{ | |
2683 | ... code in runtime section ... | |
2684 | %} | |
2685 | ||
2686 | %header %{ | |
2687 | ... code in header section ... | |
2688 | %} | |
2689 | ||
2690 | %wrapper %{ | |
2691 | ... code in wrapper section ... | |
2692 | %} | |
2693 | ||
2694 | %init %{ | |
2695 | ... code in init section ... | |
2696 | %} | |
2697 | </pre> | |
2698 | </div> | |
2699 | ||
2700 | <p> | |
2701 | The bare <tt>%{ ... %}</tt> directive is a shortcut that is the same as | |
2702 | <tt>%header %{ ... %}</tt>. | |
2703 | </p> | |
2704 | ||
2705 | <p> | |
2706 | Everything in a code insertion block is copied verbatim into the output file and is | |
2707 | not parsed by SWIG. Most SWIG input files have at least one such block to include header | |
2708 | files and support C code. Additional code blocks may be placed anywhere in a | |
2709 | SWIG file as needed. </p> | |
2710 | ||
2711 | <div class="code"><pre> | |
2712 | %module mymodule | |
2713 | %{ | |
2714 | #include "my_header.h" | |
2715 | %} | |
2716 | ... Declare functions here | |
2717 | %{ | |
2718 | ||
2719 | void some_extra_function() { | |
2720 | ... | |
2721 | } | |
2722 | %} | |
2723 | </pre></div> | |
2724 | ||
2725 | <p> | |
2726 | A common use for code blocks is to write "helper" functions. These | |
2727 | are functions that are used specifically for the purpose of building | |
2728 | an interface, but which are generally not visible to the normal C | |
2729 | program. For example :</p> | |
2730 | ||
2731 | <div class="code"><pre> | |
2732 | %{ | |
2733 | /* Create a new vector */ | |
2734 | static Vector *new_Vector() { | |
2735 | return (Vector *) malloc(sizeof(Vector)); | |
2736 | } | |
2737 | ||
2738 | %} | |
2739 | // Now wrap it | |
2740 | Vector *new_Vector(); | |
2741 | </pre></div> | |
2742 | ||
2743 | <H3><a name="SWIG_nn43"></a>5.6.3 Inlined code blocks</H3> | |
2744 | ||
2745 | ||
2746 | <p> | |
2747 | Since the process of writing helper functions is fairly common, | |
2748 | there is a special inlined form of code block that is used as follows | |
2749 | :</p> | |
2750 | ||
2751 | <div class="code"><pre> | |
2752 | %inline %{ | |
2753 | /* Create a new vector */ | |
2754 | Vector *new_Vector() { | |
2755 | return (Vector *) malloc(sizeof(Vector)); | |
2756 | } | |
2757 | %} | |
2758 | ||
2759 | </pre></div> | |
2760 | ||
2761 | <p> | |
2762 | The <tt>%inline</tt> directive inserts all of the code that follows | |
2763 | verbatim into the header portion of an interface file. The code is | |
2764 | then parsed by both the SWIG preprocessor and parser. | |
2765 | Thus, the above example creates a new command <tt>new_Vector</tt> using only one | |
2766 | declaration. Since the code inside an <tt>%inline %{ ... %}</tt> block | |
2767 | is given to both the C compiler and SWIG, it is illegal to include any | |
2768 | SWIG directives inside a <tt>%{ ... %}</tt> block.</p> | |
2769 | ||
2770 | <H3><a name="SWIG_nn44"></a>5.6.4 Initialization blocks</H3> | |
2771 | ||
2772 | ||
2773 | <p> | |
2774 | When code is included in the <tt>%init</tt> section, it is copied directly into the | |
2775 | module initialization function. For example, if you needed to perform some extra | |
2776 | initialization on module loading, you could write this: | |
2777 | </p> | |
2778 | ||
2779 | <div class="code"><pre> | |
2780 | %init %{ | |
2781 | init_variables(); | |
2782 | %} | |
2783 | </pre></div> | |
2784 | ||
2785 | <H2><a name="SWIG_nn45"></a>5.7 An Interface Building Strategy</H2> | |
2786 | ||
2787 | ||
2788 | <p> | |
2789 | This section describes the general approach for building interface | |
2790 | with SWIG. The specifics related to a particular scripting language | |
2791 | are found in later chapters.</p> | |
2792 | ||
2793 | <H3><a name="SWIG_nn46"></a>5.7.1 Preparing a C program for SWIG</H3> | |
2794 | ||
2795 | ||
2796 | <p> | |
2797 | SWIG doesn't require modifications to your C code, but if you feed it | |
2798 | a collection of raw C header files or source code, the results might | |
2799 | not be what you expect---in fact, they might be awful. Here's a series | |
2800 | of steps you can follow to make an interface for a C program :</p> | |
2801 | ||
2802 | <ul> | |
2803 | <li>Identify the functions that you want to wrap. It's probably not | |
2804 | necessary to access every single function in a C program--thus, a | |
2805 | little forethought can dramatically simplify the resulting scripting | |
2806 | language interface. C header files are particularly good source for | |
2807 | finding things to wrap. | |
2808 | ||
2809 | <li>Create a new interface file to describe the scripting language | |
2810 | interface to your program. | |
2811 | ||
2812 | <li>Copy the appropriate declarations into the interface file or use | |
2813 | SWIG's <tt>%include</tt> directive to process an entire C | |
2814 | source/header file. | |
2815 | ||
2816 | <li>Make sure everything in the interface file uses ANSI C/C++syntax. | |
2817 | ||
2818 | <li>Make sure all necessary `<tt>typedef</tt>' declarations and | |
2819 | type-information is available in the interface file. | |
2820 | ||
2821 | <li>If your program has a main() function, you may need to rename it | |
2822 | (read on). | |
2823 | ||
2824 | <li>Run SWIG and compile. | |
2825 | </ul> | |
2826 | ||
2827 | <p> | |
2828 | Although this may sound complicated, the process turns out to be | |
2829 | fairly easy once you get the hang of it. | |
2830 | </p> | |
2831 | ||
2832 | <p> | |
2833 | In the process of building an interface, SWIG may encounter syntax errors or | |
2834 | other problems. The best way to deal with this is to simply copy the offending | |
2835 | code into a separate interface file and edit it. However, the SWIG developers | |
2836 | have worked very hard to improve the SWIG parser--you should report parsing errors | |
2837 | to the <a href="http://www.swig.org/mail.html">swig-dev mailing list</a> or to the | |
2838 | <a href="http://www.swig.org/bugs.html">SWIG bug tracker</a>. | |
2839 | </p> | |
2840 | ||
2841 | <H3><a name="SWIG_nn47"></a>5.7.2 The SWIG interface file</H3> | |
2842 | ||
2843 | ||
2844 | <p> | |
2845 | The preferred method of using SWIG is to generate separate interface | |
2846 | file. Suppose you have the following C header file :</p> | |
2847 | ||
2848 | <div class="code"><pre> | |
2849 | /* File : header.h */ | |
2850 | ||
2851 | #include <stdio.h> | |
2852 | #include <math.h> | |
2853 | ||
2854 | extern int foo(double); | |
2855 | extern double bar(int, int); | |
2856 | extern void dump(FILE *f); | |
2857 | ||
2858 | </pre></div> | |
2859 | ||
2860 | <p> | |
2861 | A typical SWIG interface file for this header file would look like the | |
2862 | following :</p> | |
2863 | ||
2864 | <div class="code"><pre> | |
2865 | /* File : interface.i */ | |
2866 | %module mymodule | |
2867 | %{ | |
2868 | #include "header.h" | |
2869 | %} | |
2870 | extern int foo(double); | |
2871 | extern double bar(int, int); | |
2872 | extern void dump(FILE *f); | |
2873 | ||
2874 | </pre></div> | |
2875 | ||
2876 | <p> | |
2877 | Of course, in this case, our header file is pretty simple so we could | |
2878 | have made an interface file like this as well:</p> | |
2879 | ||
2880 | <div class="code"><pre> | |
2881 | /* File : interface.i */ | |
2882 | %module mymodule | |
2883 | %include header.h | |
2884 | </pre></div> | |
2885 | ||
2886 | <p> | |
2887 | Naturally, your mileage may vary.</p> | |
2888 | ||
2889 | <H3><a name="SWIG_nn48"></a>5.7.3 Why use separate interface files?</H3> | |
2890 | ||
2891 | ||
2892 | <p> | |
2893 | Although SWIG can parse many header files, it is more common to write a | |
2894 | special <tt>.i</tt> file defining the interface to a package. There | |
2895 | are several reasons why you might want to do this: | |
2896 | </p> | |
2897 | ||
2898 | <ul> | |
2899 | <li>It is rarely necessary to access every single function in a large | |
2900 | package. Many C functions might have little or no use in a scripted | |
2901 | environment. Therfore, why wrap them? | |
2902 | ||
2903 | <li>Separate interface files provide an opportunity to provide more | |
2904 | precise rules about how an interface is to be constructed. | |
2905 | ||
2906 | <li>Interface files can provide more structure and organization. | |
2907 | ||
2908 | <li>SWIG can't parse certain definitions that appear in header | |
2909 | files. Having a separate file allows you to eliminate or work around | |
2910 | these problems. | |
2911 | ||
2912 | <li>Interface files provide a more precise definition of what the interface | |
2913 | is. Users wanting to extend the system can go to the interface file | |
2914 | and immediately see what is available without having to dig it out of | |
2915 | header files. | |
2916 | </ul> | |
2917 | ||
2918 | <H3><a name="SWIG_nn49"></a>5.7.4 Getting the right header files</H3> | |
2919 | ||
2920 | ||
2921 | <p> | |
2922 | Sometimes, it is necessary to use certain header files in order for | |
2923 | the code generated by SWIG to compile properly. Make sure you | |
2924 | include certain header files by using a <tt>%{,%}</tt> block like this: | |
2925 | </p> | |
2926 | ||
2927 | <div class="code"><pre> | |
2928 | %module graphics | |
2929 | %{ | |
2930 | #include <GL/gl.h> | |
2931 | #include <GL/glu.h> | |
2932 | %} | |
2933 | ||
2934 | // Put rest of declarations here | |
2935 | ... | |
2936 | </pre></div> | |
2937 | ||
2938 | <H3><a name="SWIG_nn50"></a>5.7.5 What to do with main()</H3> | |
2939 | ||
2940 | ||
2941 | <p> | |
2942 | If your program defines a <tt>main()</tt> function, you may need to | |
2943 | get rid of it or rename it in order to use a scripting language. Most | |
2944 | scripting languages define their own <tt>main()</tt> procedure that | |
2945 | is called instead. <tt>main()</tt> also makes no sense when | |
2946 | working with dynamic loading. There are a few approaches to solving | |
2947 | the <tt>main()</tt> conflict :</p> | |
2948 | ||
2949 | <ul> | |
2950 | <li>Get rid of <tt>main()</tt> entirely. | |
2951 | ||
2952 | <li>Rename <tt>main()</tt> to something else. You can do this by | |
2953 | compiling your C program with an option like <tt>-Dmain=oldmain</tt>. | |
2954 | ||
2955 | <li>Use conditional compilation to only include <tt>main()</tt> when | |
2956 | not using a scripting language. | |
2957 | </ul> | |
2958 | ||
2959 | <p> | |
2960 | Getting rid of <tt>main()</tt> may cause potential initialization | |
2961 | problems of a program. To handle this problem, you may consider | |
2962 | writing a special function called <tt>program_init()</tt> that | |
2963 | initializes your program upon startup. This function could then be | |
2964 | called either from the scripting language as the first operation, or | |
2965 | when the SWIG generated module is loaded.</p> | |
2966 | ||
2967 | <p> | |
2968 | As a general note, many C programs only use the <tt>main()</tt> | |
2969 | function to parse command line options and to set parameters. However, | |
2970 | by using a scripting language, you are probably trying to create a | |
2971 | program that is more interactive. In many cases, the old | |
2972 | <tt>main()</tt> program can be completely replaced by a Perl, Python, | |
2973 | or Tcl script.</p> | |
2974 | ||
2975 | <p> | |
2976 | <b>Note:</b> If some cases, you might be inclined to create a | |
2977 | scripting language wrapper for <tt>main()</tt>. If you do this, the | |
2978 | compilation will probably work and your module might even load | |
2979 | correctly. The only trouble is that when you call your | |
2980 | <tt>main()</tt> wrapper, you will find that it actually invokes the | |
2981 | <tt>main()</tt> of the scripting language interpreter itself! This behavior | |
2982 | is a side effect of the symbol binding mechanism used in the dynamic linker. | |
2983 | The bottom line: don't do this. | |
2984 | </p> | |
2985 | ||
2986 | </body> | |
2987 | </html> |