1 const char pcrs_rcs[] = "$Id: pcrs.c,v 1.22 2006/12/24 17:34:20 fabiankeil Exp $";
3 /*********************************************************************
5 * File : $Source: /cvsroot/ijbswa/current/pcrs.c,v $
7 * Purpose : pcrs is a supplement to the pcre library by Philip Hazel
8 * <ph10@cam.ac.uk> and adds Perl-style substitution. That
9 * is, it mimics Perl's 's' operator. See pcrs(3) for details.
12 * Copyright : Written and Copyright (C) 2000, 2001 by Andreas S. Oesterhelt
13 * <andreas@oesterhelt.org>
15 * This program is free software; you can redistribute it
16 * and/or modify it under the terms of the GNU Lesser
17 * General Public License (LGPL), version 2.1, which should
18 * be included in this distribution (see LICENSE.txt), with
19 * the exception that the permission to replace that license
20 * with the GNU General Public License (GPL) given in section
21 * 3 is restricted to version 2 of the GPL.
23 * This program is distributed in the hope that it will
24 * be useful, but WITHOUT ANY WARRANTY; without even the
25 * implied warranty of MERCHANTABILITY or FITNESS FOR A
26 * PARTICULAR PURPOSE. See the license for more details.
28 * The GNU Lesser General Public License should be included
29 * with this file. If not, you can view it at
30 * http://www.gnu.org/licenses/lgpl.html
31 * or write to the Free Software Foundation, Inc., 59
32 * Temple Place - Suite 330, Boston, MA 02111-1307, USA.
36 * Revision 1.22 2006/12/24 17:34:20 fabiankeil
37 * Add pcrs_strerror() message for PCRE_ERROR_MATCHLIMIT
38 * and give a hint why an error code might be unknown.
40 * Catch NULL subjects early in pcrs_execute().
42 * Revision 1.21 2006/07/18 14:48:47 david__schmidt
43 * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
44 * with what was really the latest development (the v_3_0_branch branch)
46 * Revision 1.19.2.4 2005/05/07 21:50:55 david__schmidt
47 * A few memory leaks plugged (mostly on error paths)
49 * Revision 1.19.2.3 2003/12/04 12:32:45 oes
50 * Append a trailing nullbyte to result to facilitate string processing
52 * Revision 1.19.2.2 2002/10/08 16:22:28 oes
53 * Bugfix: Need to check validity of backreferences explicitly,
54 * because when max_matches are reached and matches is expanded,
55 * realloc() does not zero the memory. Fixes Bug # 606227
57 * Revision 1.19.2.1 2002/08/10 11:23:40 oes
58 * Include prce.h via project.h, where the appropriate
59 * source will have been selected
61 * Revision 1.19 2002/03/08 14:47:48 oes
64 * Revision 1.18 2002/03/08 14:17:14 oes
65 * Fixing -Wconversion warnings
67 * Revision 1.17 2002/03/08 13:45:48 oes
68 * Hiding internal functions
70 * Revision 1.16 2001/11/30 21:32:14 jongfoster
71 * Fixing signed/unsigned comparison (Andreas please check this!)
74 * Revision 1.15 2001/09/20 16:11:06 steudten
76 * Add casting for some string functions.
78 * Revision 1.14 2001/09/09 21:41:57 oes
79 * Fixing yet another silly bug
81 * Revision 1.13 2001/09/06 14:05:59 oes
84 * Revision 1.12 2001/08/18 11:35:00 oes
85 * - Introduced pcrs_strerror()
86 * - made some NULL arguments non-fatal
87 * - added support for \n \r \e \b \t \f \a \0 in substitute
88 * - made quoting adhere to standard rules
89 * - added warning for bad backrefs
90 * - added pcrs_execute_list()
92 * - bugfix & cosmetics
94 * Revision 1.11 2001/08/15 15:32:03 oes
95 * - Added support for Perl's special variables $+, $' and $`
96 * - Improved the substitute parser
97 * - Replaced the hard limit for the maximum number of matches
98 * by dynamic reallocation
100 * Revision 1.10 2001/08/05 13:13:11 jongfoster
101 * Making parameters "const" where possible.
103 * Revision 1.9 2001/07/18 17:27:00 oes
104 * Changed interface; Cosmetics
106 * Revision 1.8 2001/06/29 21:45:41 oes
107 * Indentation, CRLF->LF, Tab-> Space
109 * Revision 1.7 2001/06/29 13:33:04 oes
110 * - Cleaned up, renamed and reordered functions,
112 * - Removed my_strsep
113 * - Replaced globalflag with a general flags int
114 * that holds PCRS_GLOBAL, PCRS_SUCCESS, and PCRS_TRIVIAL
115 * - Introduced trivial option that will prevent pcrs
116 * from honouring backreferences in the substitute,
117 * which is useful for large substitutes that are
118 * red in from somewhere and saves the pain of escaping
120 * - Introduced convenience function pcrs_free_joblist()
121 * - Split pcrs_make_job() into pcrs_compile(), which still
122 * takes a complete s/// comand as argument and parses it,
123 * and a new function pcrs_make_job, which takes the
124 * three separate components. This should make for a
125 * much friendlier frontend.
126 * - Removed create_pcrs_job() which was useless
127 * - Fixed a bug in pcrs_execute
128 * - Success flag is now handled by pcrs instead of user
130 * Revision 1.6 2001/06/03 19:12:45 oes
133 * Revision 1.5 2001/05/29 09:50:24 jongfoster
134 * (Fixed one int -> size_t)
136 * Revision 1.4 2001/05/25 14:12:40 oes
137 * Fixed bug: Empty substitutes now detected
139 * Revision 1.3 2001/05/25 11:03:55 oes
140 * Added sanity check for NULL jobs to pcrs_exec_substitution
142 * Revision 1.2 2001/05/22 18:46:04 oes
144 * Added support for PCRE_UNGREEDY behaviour to pcrs,
145 * which is selected by the (nonstandard and therefore
146 * capital) letter 'U' in the option string.
147 * It causes the quantifiers to be ungreedy by default.
148 * Appending a ? turns back to greedy (!).
150 * Revision 1.1.1.1 2001/05/15 13:59:02 oes
151 * Initial import of version 2.9.3 source tree
154 *********************************************************************/
158 * Include project.h just so that the right pcre.h gets
159 * included from there
168 const char pcrs_h_rcs[] = PCRS_H_VERSION;
171 * Internal prototypes
174 static int pcrs_parse_perl_options(const char *optstring, int *flags);
175 static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag,
176 int capturecount, int *errptr);
178 /*********************************************************************
180 * Function : pcrs_strerror
182 * Description : Return a string describing a given error code.
185 * 1 : error = the error code
187 * Returns : char * to the descriptive string
189 *********************************************************************/
190 const char *pcrs_strerror(const int error)
196 /* Passed-through PCRE error: */
197 case PCRE_ERROR_NOMEMORY: return "(pcre:) No memory";
199 /* Shouldn't happen unless PCRE or PCRS bug, or user messed with compiled job: */
200 case PCRE_ERROR_NULL: return "(pcre:) NULL code or subject or ovector";
201 case PCRE_ERROR_BADOPTION: return "(pcre:) Unrecognized option bit";
202 case PCRE_ERROR_BADMAGIC: return "(pcre:) Bad magic number in code";
203 case PCRE_ERROR_UNKNOWN_NODE: return "(pcre:) Bad node in pattern";
205 /* Can't happen / not passed: */
206 case PCRE_ERROR_NOSUBSTRING: return "(pcre:) Fire in power supply";
207 case PCRE_ERROR_NOMATCH: return "(pcre:) Water in power supply";
209 #ifdef PCRE_ERROR_MATCHLIMIT
211 * Only reported by PCRE versions newer than our own.
213 case PCRE_ERROR_MATCHLIMIT: return "(pcre:) Match limit reached";
214 #endif /* def PCRE_ERROR_MATCHLIMIT */
217 case PCRS_ERR_NOMEM: return "(pcrs:) No memory";
218 case PCRS_ERR_CMDSYNTAX: return "(pcrs:) Syntax error while parsing command";
219 case PCRS_ERR_STUDY: return "(pcrs:) PCRE error while studying the pattern";
220 case PCRS_ERR_BADJOB: return "(pcrs:) Bad job - NULL job, pattern or substitute";
221 case PCRS_WARN_BADREF: return "(pcrs:) Backreference out of range";
224 * XXX: With the exception of PCRE_ERROR_MATCHLIMIT we
225 * only catch PCRE errors that can happen with our internal
226 * version. If Privoxy is linked against a newer
227 * PCRE version all bets are off ...
229 default: return "Unknown error. Privoxy out of sync with PCRE?";
232 /* error >= 0: No error */
233 return "(pcrs:) Everything's just fine. Thanks for asking.";
238 /*********************************************************************
240 * Function : pcrs_parse_perl_options
242 * Description : This function parses a string containing the options to
243 * Perl's s/// operator. It returns an integer that is the
244 * pcre equivalent of the symbolic optstring.
245 * Since pcre doesn't know about Perl's 'g' (global) or pcrs',
246 * 'T' (trivial) options but pcrs needs them, the corresponding
247 * flags are set if 'g'or 'T' is encountered.
248 * Note: The 'T' and 'U' options do not conform to Perl.
251 * 1 : optstring = string with options in perl syntax
252 * 2 : flags = see description
254 * Returns : option integer suitable for pcre
256 *********************************************************************/
257 static int pcrs_parse_perl_options(const char *optstring, int *flags)
263 if (NULL == optstring) return 0;
265 for (i = 0; i < strlen(optstring); i++)
269 case 'e': break; /* ToDo ;-) */
270 case 'g': *flags |= PCRS_GLOBAL; break;
271 case 'i': rc |= PCRE_CASELESS; break;
272 case 'm': rc |= PCRE_MULTILINE; break;
274 case 's': rc |= PCRE_DOTALL; break;
275 case 'x': rc |= PCRE_EXTENDED; break;
276 case 'U': rc |= PCRE_UNGREEDY; break;
277 case 'T': *flags |= PCRS_TRIVIAL; break;
286 /*********************************************************************
288 * Function : pcrs_compile_replacement
290 * Description : This function takes a Perl-style replacement (2nd argument
291 * to the s/// operator and returns a compiled pcrs_substitute,
292 * or NULL if memory allocation for the substitute structure
296 * 1 : replacement = replacement part of s/// operator
298 * 2 : trivialflag = Flag that causes backreferences to be
300 * 3 : capturecount = Number of capturing subpatterns in
301 * the pattern. Needed for $+ handling.
302 * 4 : errptr = pointer to an integer in which error
303 * conditions can be returned.
305 * Returns : pcrs_substitute data structure, or NULL if an
306 * error is encountered. In that case, *errptr has
309 *********************************************************************/
310 static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag, int capturecount, int *errptr)
317 i = k = l = quoted = 0;
322 if (NULL == replacement)
330 if (NULL == (r = (pcrs_substitute *)malloc(sizeof(pcrs_substitute))))
332 *errptr = PCRS_ERR_NOMEM;
335 memset(r, '\0', sizeof(pcrs_substitute));
337 length = strlen(replacement);
339 if (NULL == (text = (char *)malloc(length + 1)))
342 *errptr = PCRS_ERR_NOMEM;
345 memset(text, '\0', length + 1);
349 * In trivial mode, just copy the substitute text
353 text = strncpy(text, replacement, length + 1);
358 * Else, parse, cut out and record all backreferences
362 while (i < (int)length)
365 if (replacement[i] == '\\')
369 text[k++] = replacement[i++];
374 if (replacement[i+1] && strchr("tnrfae0", replacement[i+1]))
376 switch (replacement[++i])
412 if (replacement[i] == '$' && !quoted && i < (int)(length - 1))
414 char *symbol, symbols[] = "'`+&";
415 r->block_length[l] = (size_t)(k - r->block_offset[l]);
417 /* Numerical backreferences */
418 if (isdigit((int)replacement[i + 1]))
420 while (i < (int)length && isdigit((int)replacement[++i]))
422 r->backref[l] = r->backref[l] * 10 + replacement[i] - 48;
424 if (r->backref[l] > capturecount)
426 *errptr = PCRS_WARN_BADREF;
430 /* Symbolic backreferences: */
431 else if (NULL != (symbol = strchr(symbols, replacement[i + 1])))
434 if (symbol - symbols == 2) /* $+ */
436 r->backref[l] = capturecount;
438 else if (symbol - symbols == 3) /* $& */
444 r->backref[l] = PCRS_MAX_SUBMATCHES + 1 - (symbol - symbols);
449 /* Invalid backref -> plain '$' */
455 /* Valid and in range? -> record */
456 if (r->backref[l] < PCRS_MAX_SUBMATCHES + 2)
458 r->backref_count[r->backref[l]] += 1;
459 r->block_offset[++l] = k;
463 *errptr = PCRS_WARN_BADREF;
469 /* Plain chars are copied */
470 text[k++] = replacement[i++];
473 } /* -END- if (!trivialflag) */
480 r->block_length[l] = (size_t)(k - r->block_offset[l]);
487 /*********************************************************************
489 * Function : pcrs_free_job
491 * Description : Frees the memory used by a pcrs_job struct and its
492 * dependant structures.
495 * 1 : job = pointer to the pcrs_job structure to be freed
497 * Returns : a pointer to the next job, if there was any, or
500 *********************************************************************/
501 pcrs_job *pcrs_free_job(pcrs_job *job)
512 if (job->pattern != NULL) free(job->pattern);
513 if (job->hints != NULL) free(job->hints);
514 if (job->substitute != NULL)
516 if (job->substitute->text != NULL) free(job->substitute->text);
517 free(job->substitute);
526 /*********************************************************************
528 * Function : pcrs_free_joblist
530 * Description : Iterates through a chained list of pcrs_job's and
531 * frees them using pcrs_free_job.
534 * 1 : joblist = pointer to the first pcrs_job structure to
539 *********************************************************************/
540 void pcrs_free_joblist(pcrs_job *joblist)
542 while ( NULL != (joblist = pcrs_free_job(joblist)) ) {};
549 /*********************************************************************
551 * Function : pcrs_compile_command
553 * Description : Parses a string with a Perl-style s/// command,
554 * calls pcrs_compile, and returns a corresponding
555 * pcrs_job, or NULL if parsing or compiling the job
559 * 1 : command = string with perl-style s/// command
560 * 2 : errptr = pointer to an integer in which error
561 * conditions can be returned.
563 * Returns : a corresponding pcrs_job data structure, or NULL
564 * if an error was encountered. In that case, *errptr
567 *********************************************************************/
568 pcrs_job *pcrs_compile_command(const char *command, int *errptr)
570 int i, k, l, quoted = FALSE;
579 * Tokenize the perl command
581 limit = strlen(command);
584 *errptr = PCRS_ERR_CMDSYNTAX;
589 delimiter = command[1];
592 tokens[l] = (char *) malloc(limit + 1);
594 for (i = 0; i <= (int)limit; i++)
597 if (command[i] == delimiter && !quoted)
604 tokens[0][k++] = '\0';
605 tokens[++l] = tokens[0] + k;
609 else if (command[i] == '\\' && !quoted)
612 if (command[i+1] == delimiter) continue;
618 tokens[0][k++] = command[i];
626 *errptr = PCRS_ERR_CMDSYNTAX;
631 newjob = pcrs_compile(tokens[1], tokens[2], tokens[3], errptr);
638 /*********************************************************************
640 * Function : pcrs_compile
642 * Description : Takes the three arguments to a perl s/// command
643 * and compiles a pcrs_job structure from them.
646 * 1 : pattern = string with perl-style pattern
647 * 2 : substitute = string with perl-style substitute
648 * 3 : options = string with perl-style options
649 * 4 : errptr = pointer to an integer in which error
650 * conditions can be returned.
652 * Returns : a corresponding pcrs_job data structure, or NULL
653 * if an error was encountered. In that case, *errptr
656 *********************************************************************/
657 pcrs_job *pcrs_compile(const char *pattern, const char *substitute, const char *options, int *errptr)
667 * Handle NULL arguments
669 if (pattern == NULL) pattern = "";
670 if (substitute == NULL) substitute = "";
674 * Get and init memory
676 if (NULL == (newjob = (pcrs_job *)malloc(sizeof(pcrs_job))))
678 *errptr = PCRS_ERR_NOMEM;
681 memset(newjob, '\0', sizeof(pcrs_job));
685 * Evaluate the options
687 newjob->options = pcrs_parse_perl_options(options, &flags);
688 newjob->flags = flags;
692 * Compile the pattern
694 newjob->pattern = pcre_compile(pattern, newjob->options, &error, errptr, NULL);
695 if (newjob->pattern == NULL)
697 pcrs_free_job(newjob);
703 * Generate hints. This has little overhead, since the
704 * hints will be NULL for a boring pattern anyway.
706 newjob->hints = pcre_study(newjob->pattern, 0, &error);
709 *errptr = PCRS_ERR_STUDY;
710 pcrs_free_job(newjob);
716 * Determine the number of capturing subpatterns.
717 * This is needed for handling $+ in the substitute.
719 if (0 > (*errptr = pcre_fullinfo(newjob->pattern, newjob->hints, PCRE_INFO_CAPTURECOUNT, &capturecount)))
721 pcrs_free_job(newjob);
727 * Compile the substitute
729 if (NULL == (newjob->substitute = pcrs_compile_replacement(substitute, newjob->flags & PCRS_TRIVIAL, capturecount, errptr)))
731 pcrs_free_job(newjob);
740 /*********************************************************************
742 * Function : pcrs_execute_list
744 * Description : This is a multiple job wrapper for pcrs_execute().
745 * Apply the regular substitutions defined by the jobs in
746 * the joblist to the subject.
747 * The subject itself is left untouched, memory for the result
748 * is malloc()ed and it is the caller's responsibility to free
749 * the result when it's no longer needed.
751 * Note: For convenient string handling, a null byte is
752 * appended to the result. It does not count towards the
753 * result_length, though.
757 * 1 : joblist = the chained list of pcrs_jobs to be executed
758 * 2 : subject = the subject string
759 * 3 : subject_length = the subject's length
760 * 4 : result = char** for returning the result
761 * 5 : result_length = size_t* for returning the result's length
763 * Returns : On success, the number of substitutions that were made.
764 * May be > 1 if job->flags contained PCRS_GLOBAL
765 * On failure, the (negative) pcre error code describing the
766 * failure, which may be translated to text using pcrs_strerror().
768 *********************************************************************/
769 int pcrs_execute_list(pcrs_job *joblist, char *subject, size_t subject_length, char **result, size_t *result_length)
772 char *old, *new = NULL;
773 int hits, total_hits;
776 *result_length = subject_length;
777 hits = total_hits = 0;
779 for (job = joblist; job != NULL; job = job->next)
781 hits = pcrs_execute(job, old, *result_length, &new, result_length);
783 if (old != subject) free(old);
802 /*********************************************************************
804 * Function : pcrs_execute
806 * Description : Apply the regular substitution defined by the job to the
808 * The subject itself is left untouched, memory for the result
809 * is malloc()ed and it is the caller's responsibility to free
810 * the result when it's no longer needed.
812 * Note: For convenient string handling, a null byte is
813 * appended to the result. It does not count towards the
814 * result_length, though.
817 * 1 : job = the pcrs_job to be executed
818 * 2 : subject = the subject (== original) string
819 * 3 : subject_length = the subject's length
820 * 4 : result = char** for returning the result
821 * 5 : result_length = size_t* for returning the result's length
823 * Returns : On success, the number of substitutions that were made.
824 * May be > 1 if job->flags contained PCRS_GLOBAL
825 * On failure, the (negative) pcre error code describing the
826 * failure, which may be translated to text using pcrs_strerror().
828 *********************************************************************/
829 int pcrs_execute(pcrs_job *job, char *subject, size_t subject_length, char **result, size_t *result_length)
831 int offsets[3 * PCRS_MAX_SUBMATCHES],
836 max_matches = PCRS_MAX_MATCH_INIT;
838 pcrs_match *matches, *dummy;
844 * Sanity check & memory allocation
846 if (job == NULL || job->pattern == NULL || job->substitute == NULL || NULL == subject)
849 return(PCRS_ERR_BADJOB);
852 if (NULL == (matches = (pcrs_match *)malloc((size_t)max_matches * sizeof(pcrs_match))))
855 return(PCRS_ERR_NOMEM);
857 memset(matches, '\0', (size_t)max_matches * sizeof(pcrs_match));
861 * Find the pattern and calculate the space
862 * requirements for the result
864 newsize = subject_length;
866 while ((submatches = pcre_exec(job->pattern, job->hints, subject, (int)subject_length, offset, 0, offsets, 3 * PCRS_MAX_SUBMATCHES)) > 0)
868 job->flags |= PCRS_SUCCESS;
869 matches[i].submatches = submatches;
871 for (k = 0; k < submatches; k++)
873 matches[i].submatch_offset[k] = offsets[2 * k];
875 /* Note: Non-found optional submatches have length -1-(-1)==0 */
876 matches[i].submatch_length[k] = (size_t)(offsets[2 * k + 1] - offsets[2 * k]);
878 /* reserve mem for each submatch as often as it is ref'd */
879 newsize += matches[i].submatch_length[k] * (size_t)job->substitute->backref_count[k];
881 /* plus replacement text size minus match text size */
882 newsize += strlen(job->substitute->text) - matches[i].submatch_length[0];
884 /* chunk before match */
885 matches[i].submatch_offset[PCRS_MAX_SUBMATCHES] = 0;
886 matches[i].submatch_length[PCRS_MAX_SUBMATCHES] = (size_t)offsets[0];
887 newsize += (size_t)offsets[0] * (size_t)job->substitute->backref_count[PCRS_MAX_SUBMATCHES];
889 /* chunk after match */
890 matches[i].submatch_offset[PCRS_MAX_SUBMATCHES + 1] = offsets[1];
891 matches[i].submatch_length[PCRS_MAX_SUBMATCHES + 1] = subject_length - (size_t)offsets[1] - 1;
892 newsize += (subject_length - (size_t)offsets[1]) * (size_t)job->substitute->backref_count[PCRS_MAX_SUBMATCHES + 1];
894 /* Storage for matches exhausted? -> Extend! */
895 if (++i >= max_matches)
897 max_matches = (int)(max_matches * PCRS_MAX_MATCH_GROW);
898 if (NULL == (dummy = (pcrs_match *)realloc(matches, (size_t)max_matches * sizeof(pcrs_match))))
902 return(PCRS_ERR_NOMEM);
907 /* Non-global search or limit reached? */
908 if (!(job->flags & PCRS_GLOBAL)) break;
910 /* Don't loop on empty matches */
911 if (offsets[1] == offset)
912 if ((size_t)offset < subject_length)
916 /* Go find the next one */
920 /* Pass pcre error through if (bad) failiure */
921 if (submatches < PCRE_ERROR_NOMATCH)
930 * Get memory for the result (must be freed by caller!)
931 * and append terminating null byte.
933 if ((*result = (char *)malloc(newsize + 1)) == NULL)
936 return PCRS_ERR_NOMEM;
940 (*result)[newsize] = '\0';
948 result_offset = *result;
950 for (i = 0; i < matches_found; i++)
952 /* copy the chunk preceding the match */
953 memcpy(result_offset, subject + offset, (size_t)(matches[i].submatch_offset[0] - offset));
954 result_offset += matches[i].submatch_offset[0] - offset;
956 /* For every segment of the substitute.. */
957 for (k = 0; k <= job->substitute->backrefs; k++)
959 /* ...copy its text.. */
960 memcpy(result_offset, job->substitute->text + job->substitute->block_offset[k], job->substitute->block_length[k]);
961 result_offset += job->substitute->block_length[k];
963 /* ..plus, if it's not the last chunk, i.e.: There *is* a backref.. */
964 if (k != job->substitute->backrefs
965 /* ..in legal range.. */
966 && job->substitute->backref[k] < PCRS_MAX_SUBMATCHES + 2
967 /* ..and referencing a real submatch.. */
968 && job->substitute->backref[k] < matches[i].submatches
969 /* ..that is nonempty.. */
970 && matches[i].submatch_length[job->substitute->backref[k]] > 0)
972 /* ..copy the submatch that is ref'd. */
975 subject + matches[i].submatch_offset[job->substitute->backref[k]],
976 matches[i].submatch_length[job->substitute->backref[k]]
978 result_offset += matches[i].submatch_length[job->substitute->backref[k]];
981 offset = matches[i].submatch_offset[0] + (int)matches[i].submatch_length[0];
985 memcpy(result_offset, subject + offset, subject_length - (size_t)offset);
987 *result_length = newsize;
989 return matches_found;