1 const char pcrs_rcs[] = "$Id: pcrs.c,v 1.8 2001/06/29 21:45:41 oes Exp $";
3 /*********************************************************************
5 * File : $Source: /cvsroot/ijbswa/current/pcrs.c,v $
7 * Purpose : pcrs is a supplement to the brilliant pcre library by Philip
8 * Hazel (ph10@cam.ac.uk) and adds Perl-style substitution. That
9 * is, it mimics Perl's 's' operator.
11 * Currently, there's no documentation besides comments and the
14 * Note: In addition to perl's options, 'U' for ungreedy and 'T'
15 * for trivial (i.e.: ignore backrefs in the substitute) are
18 * Copyright : Written and Copyright (C) 2000, 2001 by Andreas S. Oesterhelt
19 * <andreas@oesterhelt.org>
21 * This program is free software; you can redistribute it
22 * and/or modify it under the terms of the GNU General
23 * Public License as published by the Free Software
24 * Foundation; either version 2 of the License, or (at
25 * your option) any later version.
27 * This program is distributed in the hope that it will
28 * be useful, but WITHOUT ANY WARRANTY; without even the
29 * implied warranty of MERCHANTABILITY or FITNESS FOR A
30 * PARTICULAR PURPOSE. See the GNU General Public
31 * License for more details.
33 * The GNU General Public License should be included with
34 * this file. If not, you can view it at
35 * http://www.gnu.org/copyleft/gpl.html
36 * or write to the Free Software Foundation, Inc., 59
37 * Temple Place - Suite 330, Boston, MA 02111-1307, USA.
41 * Revision 1.8 2001/06/29 21:45:41 oes
42 * Indentation, CRLF->LF, Tab-> Space
44 * Revision 1.7 2001/06/29 13:33:04 oes
45 * - Cleaned up, renamed and reordered functions,
48 * - Replaced globalflag with a general flags int
49 * that holds PCRS_GLOBAL, PCRS_SUCCESS, and PCRS_TRIVIAL
50 * - Introduced trivial option that will prevent pcrs
51 * from honouring backreferences in the substitute,
52 * which is useful for large substitutes that are
53 * red in from somewhere and saves the pain of escaping
55 * - Introduced convenience function pcrs_free_joblist()
56 * - Split pcrs_make_job() into pcrs_compile(), which still
57 * takes a complete s/// comand as argument and parses it,
58 * and a new function pcrs_make_job, which takes the
59 * three separate components. This should make for a
60 * much friendlier frontend.
61 * - Removed create_pcrs_job() which was useless
62 * - Fixed a bug in pcrs_execute
63 * - Success flag is now handled by pcrs instead of user
64 * - Removed logentry from cancelled commit
66 * Revision 1.6 2001/06/03 19:12:45 oes
69 * Revision 1.5 2001/05/29 09:50:24 jongfoster
70 * Unified blocklist/imagelist/permissionslist.
71 * File format is still under discussion, but the internal changes
74 * Also modified interceptor behaviour:
75 * - We now intercept all URLs beginning with one of the following
76 * prefixes (and *only* these prefixes):
78 * * http://ijbswa.sf.net/config/
79 * * http://ijbswa.sourceforge.net/config/
80 * - New interceptors "home page" - go to http://i.j.b/ to see it.
81 * - Internal changes so that intercepted and fast redirect pages
82 * are not replaced with an image.
83 * - Interceptors now have the option to send a binary page direct
84 * to the client. (i.e. ijb-send-banner uses this)
85 * - Implemented show-url-info interceptor. (Which is why I needed
86 * the above interceptors changes - a typical URL is
87 * "http://i.j.b/show-url-info?url=www.somesite.com/banner.gif".
88 * The previous mechanism would not have intercepted that, and
89 * if it had been intercepted then it then it would have replaced
92 * Revision 1.4 2001/05/25 14:12:40 oes
93 * Fixed bug: Empty substitutes now detected
95 * Revision 1.3 2001/05/25 11:03:55 oes
96 * Added sanity check for NULL jobs to pcrs_exec_substitution
98 * Revision 1.2 2001/05/22 18:46:04 oes
100 * - Enabled filtering banners by size rather than URL
101 * by adding patterns that replace all standard banner
102 * sizes with the "Junkbuster" gif to the re_filterfile
104 * - Enabled filtering WebBugs by providing a pattern
105 * which kills all 1x1 images
107 * - Added support for PCRE_UNGREEDY behaviour to pcrs,
108 * which is selected by the (nonstandard and therefore
109 * capital) letter 'U' in the option string.
110 * It causes the quantifiers to be ungreedy by default.
111 * Appending a ? turns back to greedy (!).
113 * - Added a new interceptor ijb-send-banner, which
114 * sends back the "Junkbuster" gif. Without imagelist or
115 * MSIE detection support, or if tinygif = 1, or the
116 * URL isn't recognized as an imageurl, a lame HTML
117 * explanation is sent instead.
119 * - Added new feature, which permits blocking remote
120 * script redirects and firing back a local redirect
122 * The feature is conditionally compiled, i.e. it
123 * can be disabled with --disable-fast-redirects,
124 * plus it must be activated by a "fast-redirects"
125 * line in the config file, has its own log level
126 * and of course wants to be displayed by show-proxy-args
127 * Note: Boy, all the #ifdefs in 1001 locations and
128 * all the fumbling with configure.in and acconfig.h
129 * were *way* more work than the feature itself :-(
131 * - Because a generic redirect template was needed for
132 * this, tinygif = 3 now uses the same.
134 * - Moved GIFs, and other static HTTP response templates
139 * - Removed some >400 CRs again (Jon, you really worked
142 * Revision 1.1.1.1 2001/05/15 13:59:02 oes
143 * Initial import of version 2.9.3 source tree
146 *********************************************************************/
152 const char pcrs_h_rcs[] = PCRS_H_VERSION;
155 /*********************************************************************
157 * Function : pcrs_compile_perl_options
159 * Description : This function parses a string containing the options to
160 * Perl's s/// operator. It returns an integer that is the
161 * pcre equivalent of the symbolic optstring.
162 * Since pcre doesn't know about Perl's 'g' (global) or pcrs',
163 * 'T' (trivial) options but pcrs needs them, the corresponding
164 * flags are set if 'g'or 'T' is encountered.
165 * Note: The 'T' and 'U' options do not conform to Perl.
168 * 1 : optstring = string with options in perl syntax
169 * 2 : flags = see description
171 * Returns : option integer suitable for pcre
173 *********************************************************************/
174 int pcrs_compile_perl_options(char *optstring, int *flags)
179 for (i=0; i < strlen(optstring); i++)
184 case 'g': *flags |= PCRS_GLOBAL; break;
185 case 'i': rc |= PCRE_CASELESS; break;
186 case 'm': rc |= PCRE_MULTILINE; break;
188 case 's': rc |= PCRE_DOTALL; break;
189 case 'x': rc |= PCRE_EXTENDED; break;
190 case 'U': rc |= PCRE_UNGREEDY; break;
191 case 'T': *flags |= PCRS_TRIVIAL; break;
200 /*********************************************************************
202 * Function : pcrs_compile_replacement
204 * Description : This function takes a Perl-style replacement (2nd argument
205 * to the s/// operator and returns a compiled pcrs_substitute,
206 * or NULL if memory allocation for the substitute structure
210 * 1 : replacement = replacement part of s/// operator
212 * 2 : errptr = pointer to an integer in which error
213 * conditions can be returned.
215 * Returns : pcrs_substitute data structure, or NULL if an
216 * error is encountered. In that case, *errptr has
219 *********************************************************************/
220 pcrs_substitute *pcrs_compile_replacement(char *replacement, int trivialflag, int *errptr)
222 int length, i, k = 0, l = 0, quoted = 0, idx;
223 char *text, *num_ptr, *numbers = "0123456789";
226 r = (pcrs_substitute *)malloc(sizeof(pcrs_substitute));
227 if (r == NULL) return NULL;
228 memset(r, '\0', sizeof(pcrs_substitute));
230 text = strdup(replacement); /* must be free()d by caller */
233 *errptr = PCRS_ERR_NOMEM;
238 length = strlen(replacement);
246 for (i=0; i < length; i++)
248 /* Backslash treatment */
249 if (replacement[i] == '\\')
253 text[k++] = replacement[i];
263 /* Dollar treatment */
264 if (replacement[i] == '$' && !quoted && i < length - 1)
266 if (strchr("0123456789&", replacement[i + 1]) == NULL)
268 text[k++] = replacement[i];
272 r->block_length[l] = k - r->block_offset[l];
274 if (replacement[i + 1] != '&')
276 while ((num_ptr = strchr(numbers, replacement[++i])) != NULL && i < length)
278 idx = num_ptr - numbers;
279 r->backref[l] = r->backref[l] * 10 + idx;
285 if (r->backref[l] < PCRS_MAX_SUBMATCHES)
286 r->backref_count[r->backref[l]] += 1;
288 r->block_offset[l] = k;
293 /* Plain char treatment */
294 text[k++] = replacement[i];
297 } /* -END- if (!trivialflag) */
302 r->block_length[l] = k - r->block_offset[l];
308 /*********************************************************************
310 * Function : pcrs_free_job
312 * Description : Frees the memory used by a pcrs_job struct and its
313 * dependant structures. Returns a pointer to the next
314 * job, if there was any, or NULL otherwise.
317 * 1 : job = pointer to the pcrs_job structure to be freed
319 * Returns : a pointer to the next job, if there was any, or
322 *********************************************************************/
323 pcrs_job *pcrs_free_job(pcrs_job *job)
334 if (job->pattern != NULL) free(job->pattern);
335 if (job->hints != NULL) free(job->hints);
336 if (job->substitute != NULL)
338 if (job->substitute->text != NULL) free(job->substitute->text);
339 free(job->substitute);
347 /*********************************************************************
349 * Function : pcrs_free_joblist
351 * Description : Iterates through a chained list of pcrs_job's and
352 * frees them using pcrs_free_job.
355 * 1 : joblist = pointer to the first pcrs_job structure to
360 *********************************************************************/
361 void pcrs_free_joblist(pcrs_job *joblist)
363 while ( NULL != (joblist = pcrs_free_job(joblist)) ) {};
370 /*********************************************************************
372 * Function : pcrs_compile_command
374 * Description : Parses a string with a Perl-style s/// command,
375 * calls pcrs_compile, and returns a corresponding
376 * pcrs_job, or NULL if parsing or compiling the job
380 * 1 : command = string with perl-style s/// command
381 * 2 : errptr = pointer to an integer in which error
382 * conditions can be returned.
384 * Returns : a corresponding pcrs_job data structure, or NULL
385 * if an error was encountered. In that case, *errptr
388 *********************************************************************/
389 pcrs_job *pcrs_compile_command(char *command, int *errptr)
391 int i, k, l, limit, quoted = FALSE;
399 * Tokenize the perl command
401 limit = strlen(command);
404 *errptr = PCRS_ERR_CMDSYNTAX;
409 delimiter = command[1];
412 tokens[l] = (char *) malloc(limit + 1);
414 for (i=0; i <= limit; i++)
417 if (command[i] == delimiter && !quoted)
424 tokens[0][k++] = '\0';
425 tokens[++l] = tokens[0] + k;
429 else if (command[i] == '\\' && !quoted && i+1 < limit && command[i+1] == delimiter)
434 tokens[0][k++] = command[i];
444 *errptr = PCRS_ERR_CMDSYNTAX;
449 newjob = pcrs_compile(tokens[1], tokens[2], tokens[3], errptr);
456 /*********************************************************************
458 * Function : pcrs_compile
460 * Description : Takes the three arguments to a perl s/// command
461 * and compiles a pcrs_job structure from them.
464 * 1 : pattern = string with perl-style pattern
465 * 2 : substitute = string with perl-style substitute
466 * 3 : options = string with perl-style options
467 * 4 : errptr = pointer to an integer in which error
468 * conditions can be returned.
470 * Returns : a corresponding pcrs_job data structure, or NULL
471 * if an error was encountered. In that case, *errptr
474 *********************************************************************/
475 pcrs_job *pcrs_compile(char *pattern, char *substitute, char *options, int *errptr)
483 * Handle NULL arguments
485 if (pattern == NULL) pattern = "";
486 if (substitute == NULL) substitute = "";
487 if (options == NULL) options = "";
491 * Get and init memory
493 if (NULL == (newjob = (pcrs_job *)malloc(sizeof(pcrs_job))))
495 *errptr = PCRS_ERR_NOMEM;
498 memset(newjob, '\0', sizeof(pcrs_job));
502 * Evaluate the options
504 newjob->options = pcrs_compile_perl_options(options, &flags);
505 newjob->flags = flags;
509 * Compile the pattern
511 newjob->pattern = pcre_compile(pattern, newjob->options, &error, errptr, NULL);
512 if (newjob->pattern == NULL)
514 pcrs_free_job(newjob);
520 * Generate hints. This has little overhead, since the
521 * hints will be NULL for a boring pattern anyway.
523 newjob->hints = pcre_study(newjob->pattern, 0, &error);
526 *errptr = PCRS_ERR_STUDY;
527 pcrs_free_job(newjob);
533 * Compile the substitute
535 if (NULL == (newjob->substitute = pcrs_compile_replacement(substitute, newjob->flags & PCRS_TRIVIAL, errptr)))
537 pcrs_free_job(newjob);
546 /*********************************************************************
548 * Function : pcrs_execute
550 * Description : Modify the subject by executing the regular substitution
551 * defined by the job. Since the result may be longer than
552 * the subject, its space requirements are precalculated in
553 * the matching phase and new memory is allocated accordingly.
554 * It is the caller's responsibility to free the result when
555 * it's no longer needed.
558 * 1 : job = the pcrs_job to be executed
559 * 2 : subject = the subject (== original) string
560 * 3 : subject_length = the subject's length
561 * INCLUDING the terminating zero, if string!
562 * 4 : result = char** for returning the result
563 * 5 : result_length = int* for returning the result's length
565 * Returns : the number of substitutions that were made. May be > 1
566 * if job->flags contained PCRS_GLOBAL
568 *********************************************************************/
569 int pcrs_execute(pcrs_job *job, char *subject, int subject_length, char **result, int *result_length)
571 int offsets[3 * PCRS_MAX_SUBMATCHES],
576 pcrs_match matches[PCRS_MAX_MATCHES];
584 if (job == NULL || job->pattern == NULL || job->substitute == NULL)
587 return(PCRS_ERR_BADJOB);
592 * Find the pattern and calculate the space
593 * requirements for the result (newsize)
595 newsize=subject_length;
597 while ((submatches = pcre_exec(job->pattern, job->hints, subject, subject_length, offset, 0, offsets, 3 * PCRS_MAX_SUBMATCHES)) > 0)
599 job->flags |= PCRS_SUCCESS;
600 matches[i].submatches = submatches;
601 for (k=0; k < submatches; k++)
603 matches[i].submatch_offset[k] = offsets[2 * k];
605 /* Note: Non-found optional submatches have length -1-(-1)==0 */
606 matches[i].submatch_length[k] = offsets[2 * k + 1] - offsets[2 * k];
608 /* reserve mem for each submatch as often as it is ref'd */
609 newsize += matches[i].submatch_length[k] * job->substitute->backref_count[k];
611 /* plus replacement text size minus match text size */
612 newsize += strlen(job->substitute->text) - matches[i].submatch_length[0];
614 /* Non-global search or limit reached? */
615 if (++i >= PCRS_MAX_MATCHES || !(job->flags & PCRS_GLOBAL) ) break;
617 /* Don't loop on empty matches */
618 if (offsets[1] == offset)
619 if (offset < subject_length)
623 /* Go find the next one */
627 /* Pass pcre error through if failiure */
628 if (submatches < -1) return submatches;
633 * Get memory for the result
635 if ((*result = (char *)malloc(newsize)) == NULL) /* must be free()d by caller */
637 return PCRS_ERR_NOMEM;
645 result_offset = *result;
647 for (i=0; i < matches_found; i++)
649 /* copy the chunk preceding the match */
650 memcpy(result_offset, subject + offset, matches[i].submatch_offset[0] - offset);
651 result_offset += matches[i].submatch_offset[0] - offset;
653 /* For every segment of the substitute.. */
654 for (k=0; k <= job->substitute->backrefs; k++)
656 /* ...copy its text.. */
657 memcpy(result_offset, job->substitute->text + job->substitute->block_offset[k], job->substitute->block_length[k]);
658 result_offset += job->substitute->block_length[k];
660 /* ..plus, if it's not the last chunk, i.e.: There *is* a backref.. */
661 if (k != job->substitute->backrefs
662 /* ..in legal range.. */
663 && job->substitute->backref[k] <= PCRS_MAX_SUBMATCHES
664 /* ..and referencing a nonempty match.. */
665 && matches[i].submatch_length[job->substitute->backref[k]] > 0)
667 /* ..copy the submatch that is ref'd. */
670 subject + matches[i].submatch_offset[job->substitute->backref[k]],
671 matches[i].submatch_length[job->substitute->backref[k]]
673 result_offset += matches[i].submatch_length[job->substitute->backref[k]];
676 offset = matches[i].submatch_offset[0] + matches[i].submatch_length[0];
680 memcpy(result_offset, subject + offset, subject_length - offset);
682 *result_length = newsize;
683 return matches_found;