Commit | Line | Data |
a0d0e21e |
1 | =head1 NAME |
2 | |
cb1a09d0 |
3 | perlembed - how to embed perl in your C program |
a0d0e21e |
4 | |
5 | =head1 DESCRIPTION |
6 | |
cb1a09d0 |
7 | =head2 PREAMBLE |
8 | |
9 | Do you want to: |
10 | |
11 | =over 5 |
12 | |
96dbc785 |
13 | =item B<Use C from Perl?> |
cb1a09d0 |
14 | |
15 | Read L<perlcall> and L<perlxs>. |
16 | |
96dbc785 |
17 | =item B<Use a UNIX program from Perl?> |
cb1a09d0 |
18 | |
5f05dabc |
19 | Read about back-quotes and about C<system> and C<exec> in L<perlfunc>. |
cb1a09d0 |
20 | |
96dbc785 |
21 | =item B<Use Perl from Perl?> |
cb1a09d0 |
22 | |
8a7dc658 |
23 | Read about L<perlfunc/do> and L<perlfunc/eval> and L<perlfunc/require> |
24 | and L<perlfunc/use>. |
cb1a09d0 |
25 | |
96dbc785 |
26 | =item B<Use C from C?> |
cb1a09d0 |
27 | |
28 | Rethink your design. |
29 | |
96dbc785 |
30 | =item B<Use Perl from C?> |
cb1a09d0 |
31 | |
32 | Read on... |
33 | |
34 | =back |
35 | |
36 | =head2 ROADMAP |
37 | |
38 | L<Compiling your C program> |
39 | |
a6006777 |
40 | There's one example in each of the eight sections: |
cb1a09d0 |
41 | |
42 | L<Adding a Perl interpreter to your C program> |
43 | |
44 | L<Calling a Perl subroutine from your C program> |
45 | |
46 | L<Evaluating a Perl statement from your C program> |
47 | |
48 | L<Performing Perl pattern matches and substitutions from your C program> |
49 | |
50 | L<Fiddling with the Perl stack from your C program> |
51 | |
a6006777 |
52 | L<Maintaining a persistent interpreter> |
53 | |
8ebc5c01 |
54 | L<Maintaining multiple interpreter instances> |
55 | |
96dbc785 |
56 | L<Using Perl modules, which themselves use C libraries, from your C program> |
57 | |
8a7dc658 |
58 | This documentation is Unix specific; if you have information about how |
59 | to embed Perl on other platforms, please send e-mail to |
60 | orwant@tpj.com. |
cb1a09d0 |
61 | |
62 | =head2 Compiling your C program |
63 | |
8a7dc658 |
64 | If you have trouble compiling the scripts in this documentation, |
65 | you're not alone. The cardinal rule: COMPILE THE PROGRAMS IN EXACTLY |
66 | THE SAME WAY THAT YOUR PERL WAS COMPILED. (Sorry for yelling.) |
cb1a09d0 |
67 | |
8a7dc658 |
68 | Also, every C program that uses Perl must link in the I<perl library>. |
cb1a09d0 |
69 | What's that, you ask? Perl is itself written in C; the perl library |
70 | is the collection of compiled C programs that were used to create your |
71 | perl executable (I</usr/bin/perl> or equivalent). (Corollary: you |
72 | can't use Perl from your C program unless Perl has been compiled on |
73 | your machine, or installed properly--that's why you shouldn't blithely |
74 | copy Perl executables from machine to machine without also copying the |
75 | I<lib> directory.) |
76 | |
8a7dc658 |
77 | When you use Perl from C, your C program will--usually--allocate, |
78 | "run", and deallocate a I<PerlInterpreter> object, which is defined by |
79 | the perl library. |
cb1a09d0 |
80 | |
81 | If your copy of Perl is recent enough to contain this documentation |
a6006777 |
82 | (version 5.002 or later), then the perl library (and I<EXTERN.h> and |
8a7dc658 |
83 | I<perl.h>, which you'll also need) will reside in a directory |
84 | that looks like this: |
cb1a09d0 |
85 | |
86 | /usr/local/lib/perl5/your_architecture_here/CORE |
87 | |
88 | or perhaps just |
89 | |
90 | /usr/local/lib/perl5/CORE |
91 | |
92 | or maybe something like |
93 | |
94 | /usr/opt/perl5/CORE |
95 | |
96 | Execute this statement for a hint about where to find CORE: |
97 | |
96dbc785 |
98 | perl -MConfig -e 'print $Config{archlib}' |
cb1a09d0 |
99 | |
8a7dc658 |
100 | Here's how you'd compile the example in the next section, |
101 | L<Adding a Perl interpreter to your C program>, on my Linux box: |
cb1a09d0 |
102 | |
8a7dc658 |
103 | % gcc -O2 -Dbool=char -DHAS_BOOL -I/usr/local/include |
104 | -I/usr/local/lib/perl5/i586-linux/5.003/CORE |
105 | -L/usr/local/lib/perl5/i586-linux/5.003/CORE |
106 | -o interp interp.c -lperl -lm |
cb1a09d0 |
107 | |
8a7dc658 |
108 | (That's all one line.) On my DEC Alpha running 5.00305, the incantation |
109 | is a bit different: |
110 | |
111 | % cc -O2 -Olimit 2900 -DSTANDARD_C -I/usr/local/include |
112 | -I/usr/local/lib/perl5/alpha-dec_osf/5.00305/CORE |
113 | -L/usr/local/lib/perl5/alpha-dec_osf/5.00305/CORE -L/usr/local/lib |
114 | -D__LANGUAGE_C__ -D_NO_PROTO -o interp interp.c -lperl -lm |
115 | |
116 | How can you figure out what to add? Assuming your Perl is post-5.001, |
117 | execute a C<perl -V> command and pay special attention to the "cc" and |
118 | "ccflags" information. |
119 | |
120 | You'll have to choose the appropriate compiler (I<cc>, I<gcc>, et al.) for |
121 | your machine: C<perl -MConfig -e 'print $Config{cc}'> will tell you what |
122 | to use. |
123 | |
124 | You'll also have to choose the appropriate library directory |
125 | (I</usr/local/lib/...>) for your machine. If your compiler complains |
126 | that certain functions are undefined, or that it can't locate |
127 | I<-lperl>, then you need to change the path following the C<-L>. If it |
128 | complains that it can't find I<EXTERN.h> and I<perl.h>, you need to |
129 | change the path following the C<-I>. |
cb1a09d0 |
130 | |
131 | You may have to add extra libraries as well. Which ones? |
96dbc785 |
132 | Perhaps those printed by |
133 | |
134 | perl -MConfig -e 'print $Config{libs}' |
135 | |
8a7dc658 |
136 | Provided your perl binary was properly configured and installed the |
137 | B<ExtUtils::Embed> module will determine all of this information for |
138 | you: |
96dbc785 |
139 | |
140 | % cc -o interp interp.c `perl -MExtUtils::Embed -e ccopts -e ldopts` |
141 | |
8a7dc658 |
142 | If the B<ExtUtils::Embed> module isn't part of your Perl distribution, |
143 | you can retrieve it from |
144 | http://www.perl.com/perl/CPAN/modules/by-module/ExtUtils::Embed. (If |
145 | this documentation came from your Perl distribution, then you're |
146 | running 5.004 or better and you already have it.) |
96dbc785 |
147 | |
8a7dc658 |
148 | The B<ExtUtils::Embed> kit on CPAN also contains all source code for |
149 | the examples in this document, tests, additional examples and other |
150 | information you may find useful. |
cb1a09d0 |
151 | |
152 | =head2 Adding a Perl interpreter to your C program |
153 | |
154 | In a sense, perl (the C program) is a good example of embedding Perl |
155 | (the language), so I'll demonstrate embedding with I<miniperlmain.c>, |
8a7dc658 |
156 | from the source distribution. Here's a bastardized, non-portable |
157 | version of I<miniperlmain.c> containing the essentials of embedding: |
cb1a09d0 |
158 | |
cb1a09d0 |
159 | #include <EXTERN.h> /* from the Perl distribution */ |
160 | #include <perl.h> /* from the Perl distribution */ |
96dbc785 |
161 | |
cb1a09d0 |
162 | static PerlInterpreter *my_perl; /*** The Perl interpreter ***/ |
96dbc785 |
163 | |
c07a80fd |
164 | int main(int argc, char **argv, char **env) |
cb1a09d0 |
165 | { |
166 | my_perl = perl_alloc(); |
167 | perl_construct(my_perl); |
96dbc785 |
168 | perl_parse(my_perl, NULL, argc, argv, (char **)NULL); |
cb1a09d0 |
169 | perl_run(my_perl); |
170 | perl_destruct(my_perl); |
171 | perl_free(my_perl); |
172 | } |
173 | |
8a7dc658 |
174 | Notice that we don't use the C<env> pointer. Normally handed to |
175 | C<perl_parse> as its final argument, C<env> here is replaced by |
176 | C<NULL>, which means that the current environment will be used. |
96dbc785 |
177 | |
cb1a09d0 |
178 | Now compile this program (I'll call it I<interp.c>) into an executable: |
179 | |
96dbc785 |
180 | % cc -o interp interp.c `perl -MExtUtils::Embed -e ccopts -e ldopts` |
cb1a09d0 |
181 | |
182 | After a successful compilation, you'll be able to use I<interp> just |
183 | like perl itself: |
184 | |
185 | % interp |
186 | print "Pretty Good Perl \n"; |
187 | print "10890 - 9801 is ", 10890 - 9801; |
188 | <CTRL-D> |
189 | Pretty Good Perl |
190 | 10890 - 9801 is 1089 |
191 | |
192 | or |
193 | |
194 | % interp -e 'printf("%x", 3735928559)' |
195 | deadbeef |
196 | |
197 | You can also read and execute Perl statements from a file while in the |
198 | midst of your C program, by placing the filename in I<argv[1]> before |
96dbc785 |
199 | calling I<perl_run()>. |
cb1a09d0 |
200 | |
201 | =head2 Calling a Perl subroutine from your C program |
202 | |
8ebc5c01 |
203 | To call individual Perl subroutines, you can use any of the B<perl_call_*> |
204 | functions documented in the L<perlcall> man page. |
205 | In this example we'll use I<perl_call_argv>. |
cb1a09d0 |
206 | |
207 | That's shown below, in a program I'll call I<showtime.c>. |
208 | |
cb1a09d0 |
209 | #include <EXTERN.h> |
96dbc785 |
210 | #include <perl.h> |
211 | |
212 | static PerlInterpreter *my_perl; |
213 | |
c07a80fd |
214 | int main(int argc, char **argv, char **env) |
cb1a09d0 |
215 | { |
8ebc5c01 |
216 | char *args[] = { NULL }; |
cb1a09d0 |
217 | my_perl = perl_alloc(); |
218 | perl_construct(my_perl); |
96dbc785 |
219 | |
220 | perl_parse(my_perl, NULL, argc, argv, NULL); |
221 | |
8ebc5c01 |
222 | /*** skipping perl_run() ***/ |
223 | |
224 | perl_call_argv("showtime", G_DISCARD | G_NOARGS, args); |
225 | |
cb1a09d0 |
226 | perl_destruct(my_perl); |
227 | perl_free(my_perl); |
228 | } |
229 | |
230 | where I<showtime> is a Perl subroutine that takes no arguments (that's the |
96dbc785 |
231 | I<G_NOARGS>) and for which I'll ignore the return value (that's the |
cb1a09d0 |
232 | I<G_DISCARD>). Those flags, and others, are discussed in L<perlcall>. |
233 | |
234 | I'll define the I<showtime> subroutine in a file called I<showtime.pl>: |
235 | |
236 | print "I shan't be printed."; |
96dbc785 |
237 | |
cb1a09d0 |
238 | sub showtime { |
239 | print time; |
240 | } |
241 | |
242 | Simple enough. Now compile and run: |
243 | |
96dbc785 |
244 | % cc -o showtime showtime.c `perl -MExtUtils::Embed -e ccopts -e ldopts` |
245 | |
cb1a09d0 |
246 | % showtime showtime.pl |
247 | 818284590 |
248 | |
249 | yielding the number of seconds that elapsed between January 1, 1970 |
8a7dc658 |
250 | (the beginning of the Unix epoch), and the moment I began writing this |
cb1a09d0 |
251 | sentence. |
252 | |
8a7dc658 |
253 | In this particular case we don't have to call I<perl_run>, but in |
254 | general it's considered good practice to ensure proper initialization |
255 | of library code, including execution of all object C<DESTROY> methods |
256 | and package C<END {}> blocks. |
8ebc5c01 |
257 | |
8a7dc658 |
258 | If you want to pass arguments to the Perl subroutine, you can add |
259 | strings to the C<NULL>-terminated C<args> list passed to |
260 | I<perl_call_argv>. For other data types, or to examine return values, |
261 | you'll need to manipulate the Perl stack. That's demonstrated in the |
262 | last section of this document: L<Fiddling with the Perl stack from |
263 | your C program>. |
cb1a09d0 |
264 | |
265 | =head2 Evaluating a Perl statement from your C program |
266 | |
8a7dc658 |
267 | One way to evaluate pieces of Perl code is to use |
268 | L<perlguts/perl_eval_sv()>. We've wrapped this inside our own |
269 | I<perl_eval()> function, which converts a command string to an SV, |
270 | passing this and the L<perlcall/G_DISCARD> flag to |
271 | L<perlguts/perl_eval_sv()>. |
cb1a09d0 |
272 | |
273 | Arguably, this is the only routine you'll ever need to execute |
274 | snippets of Perl code from within your C program. Your string can be |
8a7dc658 |
275 | as long as you wish; it can contain multiple statements; it can employ |
276 | L<perlfunc/use>, L<perlfunc/require> and L<perlfunc/do> to include |
277 | external Perl files. |
cb1a09d0 |
278 | |
96dbc785 |
279 | Our I<perl_eval()> lets us evaluate individual Perl strings, and then |
280 | extract variables for coercion into C types. The following program, |
cb1a09d0 |
281 | I<string.c>, executes three Perl strings, extracting an C<int> from |
282 | the first, a C<float> from the second, and a C<char *> from the third. |
283 | |
cb1a09d0 |
284 | #include <EXTERN.h> |
285 | #include <perl.h> |
96dbc785 |
286 | |
cb1a09d0 |
287 | static PerlInterpreter *my_perl; |
96dbc785 |
288 | |
a6006777 |
289 | I32 perl_eval(char *string) |
cb1a09d0 |
290 | { |
a6006777 |
291 | return perl_eval_sv(newSVpv(string,0), G_DISCARD); |
cb1a09d0 |
292 | } |
96dbc785 |
293 | |
c07a80fd |
294 | main (int argc, char **argv, char **env) |
cb1a09d0 |
295 | { |
a6006777 |
296 | char *embedding[] = { "", "-e", "0" }; |
cb1a09d0 |
297 | STRLEN length; |
96dbc785 |
298 | |
cb1a09d0 |
299 | my_perl = perl_alloc(); |
300 | perl_construct( my_perl ); |
96dbc785 |
301 | |
302 | perl_parse(my_perl, NULL, 3, embedding, NULL); |
8ebc5c01 |
303 | perl_run(my_perl); |
cb1a09d0 |
304 | /** Treat $a as an integer **/ |
305 | perl_eval("$a = 3; $a **= 2"); |
306 | printf("a = %d\n", SvIV(perl_get_sv("a", FALSE))); |
96dbc785 |
307 | |
cb1a09d0 |
308 | /** Treat $a as a float **/ |
309 | perl_eval("$a = 3.14; $a **= 2"); |
310 | printf("a = %f\n", SvNV(perl_get_sv("a", FALSE))); |
96dbc785 |
311 | |
cb1a09d0 |
312 | /** Treat $a as a string **/ |
313 | perl_eval("$a = 'rekcaH lreP rehtonA tsuJ'; $a = reverse($a); "); |
314 | printf("a = %s\n", SvPV(perl_get_sv("a", FALSE), length)); |
96dbc785 |
315 | |
cb1a09d0 |
316 | perl_destruct(my_perl); |
317 | perl_free(my_perl); |
318 | } |
319 | |
320 | All of those strange functions with I<sv> in their names help convert Perl scalars to C types. They're described in L<perlguts>. |
321 | |
322 | If you compile and run I<string.c>, you'll see the results of using |
323 | I<SvIV()> to create an C<int>, I<SvNV()> to create a C<float>, and |
324 | I<SvPV()> to create a string: |
325 | |
326 | a = 9 |
327 | a = 9.859600 |
328 | a = Just Another Perl Hacker |
329 | |
330 | |
331 | =head2 Performing Perl pattern matches and substitutions from your C program |
332 | |
333 | Our I<perl_eval()> lets us evaluate strings of Perl code, so we can |
334 | define some functions that use it to "specialize" in matches and |
335 | substitutions: I<match()>, I<substitute()>, and I<matches()>. |
336 | |
96dbc785 |
337 | char match(char *string, char *pattern); |
cb1a09d0 |
338 | |
8a7dc658 |
339 | Given a string and a pattern (e.g., C<m/clasp/> or C</\b\w*\b/>, which |
340 | in your C program might appear as "/\\b\\w*\\b/"), match() |
cb1a09d0 |
341 | returns 1 if the string matches the pattern and 0 otherwise. |
342 | |
cb1a09d0 |
343 | int substitute(char *string[], char *pattern); |
344 | |
8a7dc658 |
345 | Given a pointer to a string and an C<=~> operation (e.g., |
346 | C<s/bob/robert/g> or C<tr[A-Z][a-z]>), substitute() modifies the string |
347 | according to the operation, returning the number of substitutions |
348 | made. |
cb1a09d0 |
349 | |
350 | int matches(char *string, char *pattern, char **matches[]); |
351 | |
352 | Given a string, a pattern, and a pointer to an empty array of strings, |
8a7dc658 |
353 | matches() evaluates C<$string =~ $pattern> in an array context, and |
354 | fills in I<matches> with the array elements (allocating memory as it |
355 | does so), returning the number of matches found. |
cb1a09d0 |
356 | |
96dbc785 |
357 | Here's a sample program, I<match.c>, that uses all three (long lines have |
358 | been wrapped here): |
cb1a09d0 |
359 | |
cb1a09d0 |
360 | #include <EXTERN.h> |
361 | #include <perl.h> |
8a7dc658 |
362 | |
cb1a09d0 |
363 | static PerlInterpreter *my_perl; |
a6006777 |
364 | I32 perl_eval(char *string) |
cb1a09d0 |
365 | { |
a6006777 |
366 | return perl_eval_sv(newSVpv(string,0), G_DISCARD); |
cb1a09d0 |
367 | } |
cb1a09d0 |
368 | /** match(string, pattern) |
96dbc785 |
369 | ** |
370 | ** Used for matches in a scalar context. |
371 | ** |
372 | ** Returns 1 if the match was successful; 0 otherwise. |
373 | **/ |
374 | char match(char *string, char *pattern) |
cb1a09d0 |
375 | { |
376 | char *command; |
377 | command = malloc(sizeof(char) * strlen(string) + strlen(pattern) + 37); |
96dbc785 |
378 | sprintf(command, "$string = '%s'; $return = $string =~ %s", |
8ebc5c01 |
379 | string, pattern); |
cb1a09d0 |
380 | perl_eval(command); |
381 | free(command); |
382 | return SvIV(perl_get_sv("return", FALSE)); |
383 | } |
cb1a09d0 |
384 | /** substitute(string, pattern) |
96dbc785 |
385 | ** |
386 | ** Used for =~ operations that modify their left-hand side (s/// and tr///) |
387 | ** |
388 | ** Returns the number of successful matches, and |
389 | ** modifies the input string if there were any. |
390 | **/ |
391 | int substitute(char *string[], char *pattern) |
cb1a09d0 |
392 | { |
393 | char *command; |
394 | STRLEN length; |
395 | command = malloc(sizeof(char) * strlen(*string) + strlen(pattern) + 35); |
96dbc785 |
396 | sprintf(command, "$string = '%s'; $ret = ($string =~ %s)", |
8ebc5c01 |
397 | *string, pattern); |
398 | perl_eval(command); |
399 | free(command); |
400 | *string = SvPV(perl_get_sv("string", FALSE), length); |
401 | return SvIV(perl_get_sv("ret", FALSE)); |
cb1a09d0 |
402 | } |
cb1a09d0 |
403 | /** matches(string, pattern, matches) |
96dbc785 |
404 | ** |
405 | ** Used for matches in an array context. |
406 | ** |
407 | ** Returns the number of matches, |
408 | ** and fills in **matches with the matching substrings (allocates memory!) |
409 | **/ |
410 | int matches(char *string, char *pattern, char **match_list[]) |
cb1a09d0 |
411 | { |
412 | char *command; |
413 | SV *current_match; |
414 | AV *array; |
415 | I32 num_matches; |
416 | STRLEN length; |
417 | int i; |
cb1a09d0 |
418 | command = malloc(sizeof(char) * strlen(string) + strlen(pattern) + 38); |
96dbc785 |
419 | sprintf(command, "$string = '%s'; @array = ($string =~ %s)", |
8ebc5c01 |
420 | string, pattern); |
cb1a09d0 |
421 | perl_eval(command); |
422 | free(command); |
423 | array = perl_get_av("array", FALSE); |
424 | num_matches = av_len(array) + 1; /** assume $[ is 0 **/ |
96dbc785 |
425 | *match_list = (char **) malloc(sizeof(char *) * num_matches); |
426 | for (i = 0; i <= num_matches; i++) { |
cb1a09d0 |
427 | current_match = av_shift(array); |
96dbc785 |
428 | (*match_list)[i] = SvPV(current_match, length); |
cb1a09d0 |
429 | } |
430 | return num_matches; |
431 | } |
c07a80fd |
432 | main (int argc, char **argv, char **env) |
cb1a09d0 |
433 | { |
a6006777 |
434 | char *embedding[] = { "", "-e", "0" }; |
96dbc785 |
435 | char *text, **match_list; |
cb1a09d0 |
436 | int num_matches, i; |
437 | int j; |
cb1a09d0 |
438 | my_perl = perl_alloc(); |
439 | perl_construct( my_perl ); |
96dbc785 |
440 | perl_parse(my_perl, NULL, 3, embedding, NULL); |
8ebc5c01 |
441 | perl_run(my_perl); |
442 | |
cb1a09d0 |
443 | text = (char *) malloc(sizeof(char) * 486); /** A long string follows! **/ |
96dbc785 |
444 | sprintf(text, "%s", "When he is at a convenience store and the bill \ |
445 | comes to some amount like 76 cents, Maynard is aware that there is \ |
446 | something he *should* do, something that will enable him to get back \ |
447 | a quarter, but he has no idea *what*. He fumbles through his red \ |
448 | squeezey changepurse and gives the boy three extra pennies with his \ |
449 | dollar, hoping that he might luck into the correct amount. The boy \ |
450 | gives him back two of his own pennies and then the big shiny quarter \ |
451 | that is his prize. -RICHH"); |
452 | if (match(text, "m/quarter/")) /** Does text contain 'quarter'? **/ |
453 | printf("match: Text contains the word 'quarter'.\n\n"); |
454 | else |
455 | printf("match: Text doesn't contain the word 'quarter'.\n\n"); |
456 | if (match(text, "m/eighth/")) /** Does text contain 'eighth'? **/ |
457 | printf("match: Text contains the word 'eighth'.\n\n"); |
458 | else |
459 | printf("match: Text doesn't contain the word 'eighth'.\n\n"); |
460 | /** Match all occurrences of /wi../ **/ |
461 | num_matches = matches(text, "m/(wi..)/g", &match_list); |
462 | printf("matches: m/(wi..)/g found %d matches...\n", num_matches); |
463 | for (i = 0; i < num_matches; i++) |
464 | printf("match: %s\n", match_list[i]); |
cb1a09d0 |
465 | printf("\n"); |
466 | for (i = 0; i < num_matches; i++) { |
96dbc785 |
467 | free(match_list[i]); |
cb1a09d0 |
468 | } |
96dbc785 |
469 | free(match_list); |
470 | /** Remove all vowels from text **/ |
471 | num_matches = substitute(&text, "s/[aeiou]//gi"); |
cb1a09d0 |
472 | if (num_matches) { |
96dbc785 |
473 | printf("substitute: s/[aeiou]//gi...%d substitutions made.\n", |
8ebc5c01 |
474 | num_matches); |
cb1a09d0 |
475 | printf("Now text is: %s\n\n", text); |
476 | } |
96dbc785 |
477 | /** Attempt a substitution **/ |
478 | if (!substitute(&text, "s/Perl/C/")) { |
479 | printf("substitute: s/Perl/C...No substitution made.\n\n"); |
cb1a09d0 |
480 | } |
cb1a09d0 |
481 | free(text); |
cb1a09d0 |
482 | perl_destruct(my_perl); |
483 | perl_free(my_perl); |
484 | } |
485 | |
96dbc785 |
486 | which produces the output (again, long lines have been wrapped here) |
cb1a09d0 |
487 | |
8a7dc658 |
488 | match: Text contains the word 'quarter'. |
96dbc785 |
489 | |
8a7dc658 |
490 | match: Text doesn't contain the word 'eighth'. |
96dbc785 |
491 | |
8a7dc658 |
492 | matches: m/(wi..)/g found 2 matches... |
cb1a09d0 |
493 | match: will |
494 | match: with |
96dbc785 |
495 | |
8a7dc658 |
496 | substitute: s/[aeiou]//gi...139 substitutions made. |
96dbc785 |
497 | Now text is: Whn h s t cnvnnc str nd th bll cms t sm mnt lk 76 cnts, |
498 | Mynrd s wr tht thr s smthng h *shld* d, smthng tht wll nbl hm t gt bck |
499 | qrtr, bt h hs n d *wht*. H fmbls thrgh hs rd sqzy chngprs nd gvs th by |
500 | thr xtr pnns wth hs dllr, hpng tht h mght lck nt th crrct mnt. Th by gvs |
501 | hm bck tw f hs wn pnns nd thn th bg shny qrtr tht s hs prz. -RCHH |
502 | |
8a7dc658 |
503 | substitute: s/Perl/C...No substitution made. |
96dbc785 |
504 | |
cb1a09d0 |
505 | =head2 Fiddling with the Perl stack from your C program |
506 | |
507 | When trying to explain stacks, most computer science textbooks mumble |
508 | something about spring-loaded columns of cafeteria plates: the last |
509 | thing you pushed on the stack is the first thing you pop off. That'll |
510 | do for our purposes: your C program will push some arguments onto "the Perl |
511 | stack", shut its eyes while some magic happens, and then pop the |
512 | results--the return value of your Perl subroutine--off the stack. |
96dbc785 |
513 | |
cb1a09d0 |
514 | First you'll need to know how to convert between C types and Perl |
515 | types, with newSViv() and sv_setnv() and newAV() and all their |
516 | friends. They're described in L<perlguts>. |
517 | |
518 | Then you'll need to know how to manipulate the Perl stack. That's |
519 | described in L<perlcall>. |
520 | |
96dbc785 |
521 | Once you've understood those, embedding Perl in C is easy. |
cb1a09d0 |
522 | |
5f05dabc |
523 | Because C has no built-in function for integer exponentiation, let's |
cb1a09d0 |
524 | make Perl's ** operator available to it (this is less useful than it |
5f05dabc |
525 | sounds, because Perl implements ** with C's I<pow()> function). First |
cb1a09d0 |
526 | I'll create a stub exponentiation function in I<power.pl>: |
527 | |
528 | sub expo { |
529 | my ($a, $b) = @_; |
530 | return $a ** $b; |
531 | } |
532 | |
533 | Now I'll create a C program, I<power.c>, with a function |
534 | I<PerlPower()> that contains all the perlguts necessary to push the |
535 | two arguments into I<expo()> and to pop the return value out. Take a |
536 | deep breath... |
537 | |
cb1a09d0 |
538 | #include <EXTERN.h> |
539 | #include <perl.h> |
96dbc785 |
540 | |
cb1a09d0 |
541 | static PerlInterpreter *my_perl; |
96dbc785 |
542 | |
cb1a09d0 |
543 | static void |
544 | PerlPower(int a, int b) |
545 | { |
546 | dSP; /* initialize stack pointer */ |
547 | ENTER; /* everything created after here */ |
548 | SAVETMPS; /* ...is a temporary variable. */ |
549 | PUSHMARK(sp); /* remember the stack pointer */ |
550 | XPUSHs(sv_2mortal(newSViv(a))); /* push the base onto the stack */ |
551 | XPUSHs(sv_2mortal(newSViv(b))); /* push the exponent onto stack */ |
552 | PUTBACK; /* make local stack pointer global */ |
553 | perl_call_pv("expo", G_SCALAR); /* call the function */ |
554 | SPAGAIN; /* refresh stack pointer */ |
555 | /* pop the return value from stack */ |
556 | printf ("%d to the %dth power is %d.\n", a, b, POPi); |
96dbc785 |
557 | PUTBACK; |
cb1a09d0 |
558 | FREETMPS; /* free that return value */ |
559 | LEAVE; /* ...and the XPUSHed "mortal" args.*/ |
560 | } |
96dbc785 |
561 | |
562 | int main (int argc, char **argv, char **env) |
cb1a09d0 |
563 | { |
564 | char *my_argv[2]; |
96dbc785 |
565 | |
cb1a09d0 |
566 | my_perl = perl_alloc(); |
567 | perl_construct( my_perl ); |
96dbc785 |
568 | |
cb1a09d0 |
569 | my_argv[1] = (char *) malloc(10); |
570 | sprintf(my_argv[1], "power.pl"); |
96dbc785 |
571 | |
572 | perl_parse(my_perl, NULL, argc, my_argv, NULL); |
8ebc5c01 |
573 | perl_run(my_perl); |
96dbc785 |
574 | |
cb1a09d0 |
575 | PerlPower(3, 4); /*** Compute 3 ** 4 ***/ |
96dbc785 |
576 | |
cb1a09d0 |
577 | perl_destruct(my_perl); |
578 | perl_free(my_perl); |
579 | } |
96dbc785 |
580 | |
cb1a09d0 |
581 | |
582 | |
583 | Compile and run: |
584 | |
96dbc785 |
585 | % cc -o power power.c `perl -MExtUtils::Embed -e ccopts -e ldopts` |
586 | |
587 | % power |
cb1a09d0 |
588 | 3 to the 4th power is 81. |
589 | |
a6006777 |
590 | =head2 Maintaining a persistent interpreter |
591 | |
8a7dc658 |
592 | When developing interactive and/or potentially long-running |
593 | applications, it's a good idea to maintain a persistent interpreter |
594 | rather than allocating and constructing a new interpreter multiple |
595 | times. The major reason is speed: since Perl will only be loaded into |
596 | memory once. |
597 | |
598 | However, you have to be more cautious with namespace and variable |
599 | scoping when using a persistent interpreter. In previous examples |
600 | we've been using global variables in the default package C<main>. We |
601 | knew exactly what code would be run, and assumed we could avoid |
602 | variable collisions and outrageous symbol table growth. |
603 | |
604 | Let's say your application is a server that will occasionally run Perl |
605 | code from some arbitrary file. Your server has no way of knowing what |
606 | code it's going to run. Very dangerous. |
607 | |
608 | If the file is pulled in by C<perl_parse()>, compiled into a newly |
609 | constructed interpreter, and subsequently cleaned out with |
610 | C<perl_destruct()> afterwards, you're shielded from most namespace |
611 | troubles. |
612 | |
613 | One way to avoid namespace collisions in this scenario is to translate |
614 | the filename into a guaranteed-unique package name, and then compile |
615 | the code into that package using L<perlfunc/eval>. In the example |
616 | below, each file will only be compiled once. Or, the application |
617 | might choose to clean out the symbol table associated with the file |
618 | after it's no longer needed. Using L<perlcall/perl_call_argv>, We'll |
619 | call the subroutine C<Embed::Persistent::eval_file> which lives in the |
620 | file C<persistent.pl> and pass the filename and boolean cleanup/cache |
a6006777 |
621 | flag as arguments. |
622 | |
8a7dc658 |
623 | Note that the process will continue to grow for each file that it |
624 | uses. In addition, there might be C<AUTOLOAD>ed subroutines and other |
625 | conditions that cause Perl's symbol table to grow. You might want to |
626 | add some logic that keeps track of the process size, or restarts |
627 | itself after a certain number of requests, to ensure that memory |
628 | consumption is minimized. You'll also want to scope your variables |
629 | with L<perlfunc/my> whenever possible. |
a6006777 |
630 | |
631 | |
632 | package Embed::Persistent; |
633 | #persistent.pl |
634 | |
635 | use strict; |
636 | use vars '%Cache'; |
637 | |
a6006777 |
638 | sub valid_package_name { |
639 | my($string) = @_; |
640 | $string =~ s/([^A-Za-z0-9\/])/sprintf("_%2x",unpack("C",$1))/eg; |
641 | # second pass only for words starting with a digit |
642 | $string =~ s|/(\d)|sprintf("/_%2x",unpack("C",$1))|eg; |
643 | |
644 | # Dress it up as a real package name |
645 | $string =~ s|/|::|g; |
646 | return "Embed" . $string; |
647 | } |
648 | |
649 | #borrowed from Safe.pm |
650 | sub delete_package { |
651 | my $pkg = shift; |
652 | my ($stem, $leaf); |
653 | |
654 | no strict 'refs'; |
8ebc5c01 |
655 | $pkg = "main::$pkg\::"; # expand to full symbol table name |
a6006777 |
656 | ($stem, $leaf) = $pkg =~ m/(.*::)(\w+::)$/; |
657 | |
658 | my $stem_symtab = *{$stem}{HASH}; |
659 | |
660 | delete $stem_symtab->{$leaf}; |
661 | } |
662 | |
663 | sub eval_file { |
664 | my($filename, $delete) = @_; |
665 | my $package = valid_package_name($filename); |
666 | my $mtime = -M $filename; |
667 | if(defined $Cache{$package}{mtime} |
668 | && |
669 | $Cache{$package}{mtime} <= $mtime) |
670 | { |
8ebc5c01 |
671 | # we have compiled this subroutine already, |
672 | # it has not been updated on disk, nothing left to do |
673 | print STDERR "already compiled $package->handler\n"; |
a6006777 |
674 | } |
675 | else { |
8ebc5c01 |
676 | local *FH; |
677 | open FH, $filename or die "open '$filename' $!"; |
678 | local($/) = undef; |
679 | my $sub = <FH>; |
680 | close FH; |
a6006777 |
681 | |
8ebc5c01 |
682 | #wrap the code into a subroutine inside our unique package |
683 | my $eval = qq{package $package; sub handler { $sub; }}; |
684 | { |
685 | # hide our variables within this block |
686 | my($filename,$mtime,$package,$sub); |
687 | eval $eval; |
688 | } |
689 | die $@ if $@; |
a6006777 |
690 | |
8ebc5c01 |
691 | #cache it unless we're cleaning out each time |
692 | $Cache{$package}{mtime} = $mtime unless $delete; |
a6006777 |
693 | } |
694 | |
695 | eval {$package->handler;}; |
696 | die $@ if $@; |
697 | |
698 | delete_package($package) if $delete; |
699 | |
700 | #take a look if you want |
701 | #print Devel::Symdump->rnew($package)->as_string, $/; |
702 | } |
703 | |
704 | 1; |
705 | |
706 | __END__ |
707 | |
708 | /* persistent.c */ |
709 | #include <EXTERN.h> |
710 | #include <perl.h> |
711 | |
712 | /* 1 = clean out filename's symbol table after each request, 0 = don't */ |
713 | #ifndef DO_CLEAN |
714 | #define DO_CLEAN 0 |
715 | #endif |
716 | |
717 | static PerlInterpreter *perl = NULL; |
718 | |
719 | int |
720 | main(int argc, char **argv, char **env) |
721 | { |
722 | char *embedding[] = { "", "persistent.pl" }; |
723 | char *args[] = { "", DO_CLEAN, NULL }; |
724 | char filename [1024]; |
725 | int exitstatus = 0; |
726 | |
727 | if((perl = perl_alloc()) == NULL) { |
8ebc5c01 |
728 | fprintf(stderr, "no memory!"); |
729 | exit(1); |
a6006777 |
730 | } |
731 | perl_construct(perl); |
732 | |
733 | exitstatus = perl_parse(perl, NULL, 2, embedding, NULL); |
734 | |
735 | if(!exitstatus) { |
8ebc5c01 |
736 | exitstatus = perl_run(perl); |
a6006777 |
737 | |
8ebc5c01 |
738 | while(printf("Enter file name: ") && gets(filename)) { |
a6006777 |
739 | |
8ebc5c01 |
740 | /* call the subroutine, passing it the filename as an argument */ |
741 | args[0] = filename; |
742 | perl_call_argv("Embed::Persistent::eval_file", |
743 | G_DISCARD | G_EVAL, args); |
a6006777 |
744 | |
8ebc5c01 |
745 | /* check $@ */ |
746 | if(SvTRUE(GvSV(errgv))) |
747 | fprintf(stderr, "eval error: %s\n", SvPV(GvSV(errgv),na)); |
748 | } |
a6006777 |
749 | } |
750 | |
751 | perl_destruct_level = 0; |
752 | perl_destruct(perl); |
753 | perl_free(perl); |
754 | exit(exitstatus); |
755 | } |
756 | |
757 | |
758 | Now compile: |
759 | |
8a7dc658 |
760 | % cc -o persistent persistent.c `perl -MExtUtils::Embed -e ccopts -e ldopts` |
a6006777 |
761 | |
762 | Here's a example script file: |
763 | |
764 | #test.pl |
765 | my $string = "hello"; |
766 | foo($string); |
767 | |
768 | sub foo { |
769 | print "foo says: @_\n"; |
770 | } |
771 | |
772 | Now run: |
773 | |
774 | % persistent |
775 | Enter file name: test.pl |
776 | foo says: hello |
777 | Enter file name: test.pl |
778 | already compiled Embed::test_2epl->handler |
779 | foo says: hello |
780 | Enter file name: ^C |
781 | |
8ebc5c01 |
782 | =head2 Maintaining multiple interpreter instances |
783 | |
8a7dc658 |
784 | Some rare applications will need to create more than one interpreter |
785 | during a session. Such an application might sporadically decide to |
786 | release any resources associated with the interpreter. |
787 | |
788 | The program must take care to ensure that this takes place I<before> |
789 | the next interpreter is constructed. By default, the global variable |
790 | C<perl_destruct_level> is set to C<0>, since extra cleaning isn't |
791 | needed when a program has only one interpreter. |
792 | |
793 | Setting C<perl_destruct_level> to C<1> makes everything squeaky clean: |
794 | |
795 | perl_destruct_level = 1; |
8ebc5c01 |
796 | |
8ebc5c01 |
797 | while(1) { |
798 | ... |
799 | /* reset global variables here with perl_destruct_level = 1 */ |
8a7dc658 |
800 | perl_construct(my_perl); |
8ebc5c01 |
801 | ... |
802 | /* clean and reset _everything_ during perl_destruct */ |
8a7dc658 |
803 | perl_destruct(my_perl); |
8ebc5c01 |
804 | perl_free(my_perl); |
805 | ... |
806 | /* let's go do it again! */ |
807 | } |
808 | |
8a7dc658 |
809 | When I<perl_destruct()> is called, the interpreter's syntax parse tree |
810 | and symbol tables are cleaned up, and global variables are reset. |
8ebc5c01 |
811 | |
8a7dc658 |
812 | Now suppose we have more than one interpreter instance running at the |
813 | same time. This is feasible, but only if you used the |
814 | C<-DMULTIPLICITY> flag when building Perl. By default, that sets |
815 | C<perl_destruct_level> to C<1>. |
8ebc5c01 |
816 | |
817 | Let's give it a try: |
818 | |
819 | |
820 | #include <EXTERN.h> |
8a7dc658 |
821 | #include <perl.h> |
8ebc5c01 |
822 | |
823 | /* we're going to embed two interpreters */ |
824 | /* we're going to embed two interpreters */ |
825 | |
8ebc5c01 |
826 | #define SAY_HELLO "-e", "print qq(Hi, I'm $^X\n)" |
827 | |
8ebc5c01 |
828 | int main(int argc, char **argv, char **env) |
829 | { |
830 | PerlInterpreter |
831 | *one_perl = perl_alloc(), |
832 | *two_perl = perl_alloc(); |
833 | char *one_args[] = { "one_perl", SAY_HELLO }; |
834 | char *two_args[] = { "two_perl", SAY_HELLO }; |
835 | |
836 | perl_construct(one_perl); |
837 | perl_construct(two_perl); |
838 | |
839 | perl_parse(one_perl, NULL, 3, one_args, (char **)NULL); |
840 | perl_parse(two_perl, NULL, 3, two_args, (char **)NULL); |
841 | |
842 | perl_run(one_perl); |
843 | perl_run(two_perl); |
844 | |
845 | perl_destruct(one_perl); |
846 | perl_destruct(two_perl); |
847 | |
848 | perl_free(one_perl); |
849 | perl_free(two_perl); |
850 | } |
851 | |
852 | |
853 | Compile as usual: |
854 | |
855 | % cc -o multiplicity multiplicity.c `perl -MExtUtils::Embed -e ccopts -e ldopts` |
856 | |
857 | Run it, Run it: |
858 | |
859 | % multiplicity |
860 | Hi, I'm one_perl |
861 | Hi, I'm two_perl |
862 | |
96dbc785 |
863 | =head2 Using Perl modules, which themselves use C libraries, from your C program |
864 | |
865 | If you've played with the examples above and tried to embed a script |
866 | that I<use()>s a Perl module (such as I<Socket>) which itself uses a C or C++ library, |
867 | this probably happened: |
868 | |
869 | |
870 | Can't load module Socket, dynamic loading not available in this perl. |
871 | (You may need to build a new perl executable which either supports |
872 | dynamic loading or has the Socket module statically linked into it.) |
873 | |
874 | |
875 | What's wrong? |
876 | |
877 | Your interpreter doesn't know how to communicate with these extensions |
878 | on its own. A little glue will help. Up until now you've been |
879 | calling I<perl_parse()>, handing it NULL for the second argument: |
880 | |
881 | perl_parse(my_perl, NULL, argc, my_argv, NULL); |
882 | |
883 | That's where the glue code can be inserted to create the initial contact between |
884 | Perl and linked C/C++ routines. Let's take a look some pieces of I<perlmain.c> |
885 | to see how Perl does this: |
886 | |
887 | |
888 | #ifdef __cplusplus |
889 | # define EXTERN_C extern "C" |
890 | #else |
891 | # define EXTERN_C extern |
892 | #endif |
893 | |
894 | static void xs_init _((void)); |
895 | |
896 | EXTERN_C void boot_DynaLoader _((CV* cv)); |
897 | EXTERN_C void boot_Socket _((CV* cv)); |
898 | |
899 | |
900 | EXTERN_C void |
901 | xs_init() |
902 | { |
903 | char *file = __FILE__; |
904 | /* DynaLoader is a special case */ |
905 | newXS("DynaLoader::boot_DynaLoader", boot_DynaLoader, file); |
906 | newXS("Socket::bootstrap", boot_Socket, file); |
907 | } |
908 | |
909 | Simply put: for each extension linked with your Perl executable |
910 | (determined during its initial configuration on your |
911 | computer or when adding a new extension), |
912 | a Perl subroutine is created to incorporate the extension's |
913 | routines. Normally, that subroutine is named |
914 | I<Module::bootstrap()> and is invoked when you say I<use Module>. In |
915 | turn, this hooks into an XSUB, I<boot_Module>, which creates a Perl |
916 | counterpart for each of the extension's XSUBs. Don't worry about this |
917 | part; leave that to the I<xsubpp> and extension authors. If your |
918 | extension is dynamically loaded, DynaLoader creates I<Module::bootstrap()> |
919 | for you on the fly. In fact, if you have a working DynaLoader then there |
5f05dabc |
920 | is rarely any need to link in any other extensions statically. |
96dbc785 |
921 | |
922 | |
923 | Once you have this code, slap it into the second argument of I<perl_parse()>: |
924 | |
925 | |
926 | perl_parse(my_perl, xs_init, argc, my_argv, NULL); |
927 | |
928 | |
929 | Then compile: |
930 | |
8a7dc658 |
931 | % cc -o interp interp.c `perl -MExtUtils::Embed -e ccopts -e ldopts` |
96dbc785 |
932 | |
933 | % interp |
934 | use Socket; |
935 | use SomeDynamicallyLoadedModule; |
936 | |
937 | print "Now I can use extensions!\n"' |
938 | |
939 | B<ExtUtils::Embed> can also automate writing the I<xs_init> glue code. |
940 | |
8a7dc658 |
941 | % perl -MExtUtils::Embed -e xsinit -- -o perlxsi.c |
96dbc785 |
942 | % cc -c perlxsi.c `perl -MExtUtils::Embed -e ccopts` |
943 | % cc -c interp.c `perl -MExtUtils::Embed -e ccopts` |
8a7dc658 |
944 | % cc -o interp perlxsi.o interp.o `perl -MExtUtils::Embed -e ldopts` |
96dbc785 |
945 | |
946 | Consult L<perlxs> and L<perlguts> for more details. |
947 | |
948 | |
cb1a09d0 |
949 | =head1 MORAL |
950 | |
951 | You can sometimes I<write faster code> in C, but |
5f05dabc |
952 | you can always I<write code faster> in Perl. Because you can use |
cb1a09d0 |
953 | each from the other, combine them as you wish. |
954 | |
955 | |
956 | =head1 AUTHOR |
957 | |
8a7dc658 |
958 | Jon Orwant and F<E<lt>orwant@media.mit.eduE<gt>> and Doug MacEachern |
959 | F<E<lt>dougm@osf.orgE<gt>>, with small contributions from Tim Bunce, |
960 | Tom Christiansen, Hallvard Furuseth, Dov Grobgeld, and Ilya Zakharevich. |
961 | |
962 | Check out Doug's article on embedding in Volume 1, Issue 4 of The Perl |
963 | Journal. Info about TPJ is available from http://tpj.com. |
cb1a09d0 |
964 | |
8a7dc658 |
965 | February 1, 1997 |
cb1a09d0 |
966 | |
8a7dc658 |
967 | Some of this material is excerpted from Jon Orwant's book: I<Perl 5 |
968 | Interactive>, Waite Group Press, 1996 (ISBN 1-57169-064-6) and appears |
cb1a09d0 |
969 | courtesy of Waite Group Press. |
8a7dc658 |
970 | |
971 | =head1 COPYRIGHT |
972 | |
973 | Copyright (C) 1995, 1996, 1997 Doug MacEachern and Jon Orwant. All |
974 | Rights Reserved. |
975 | |
976 | Although destined for release with the standard Perl distribution, |
977 | this document is not public domain, nor is any of Perl and its |
978 | documentation. Permission is granted to freely distribute verbatim |
979 | copies of this document provided that no modifications outside of |
980 | formatting be made, and that this notice remain intact. You are |
981 | permitted and encouraged to use its code and derivatives thereof in |
982 | your own source code for fun or for profit as you see fit. |