1 const char pcrs_rcs[] = "$Id: pcrs.c,v 2.2 2002/09/04 15:52:02 oes Exp $";
3 /*********************************************************************
5 * File : $Source: /cvsroot/ijbswa/current/src/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 2.2 2002/09/04 15:52:02 oes
37 * Synced with the stable branch:
38 * Revision 1.19.2.1 2002/08/10 11:23:40 oes
39 * Include prce.h via project.h, where the appropriate
40 * source will have been selected
42 * Revision 2.1 2002/08/26 11:18:24 sarantis
45 * Revision 2.0 2002/06/04 14:34:21 jongfoster
46 * Moving source files to src/
48 * Revision 1.19 2002/03/08 14:47:48 oes
51 * Revision 1.18 2002/03/08 14:17:14 oes
52 * Fixing -Wconversion warnings
54 * Revision 1.17 2002/03/08 13:45:48 oes
55 * Hiding internal functions
57 * Revision 1.16 2001/11/30 21:32:14 jongfoster
58 * Fixing signed/unsigned comparison (Andreas please check this!)
61 * Revision 1.15 2001/09/20 16:11:06 steudten
63 * Add casting for some string functions.
65 * Revision 1.14 2001/09/09 21:41:57 oes
66 * Fixing yet another silly bug
68 * Revision 1.13 2001/09/06 14:05:59 oes
71 * Revision 1.12 2001/08/18 11:35:00 oes
72 * - Introduced pcrs_strerror()
73 * - made some NULL arguments non-fatal
74 * - added support for \n \r \e \b \t \f \a \0 in substitute
75 * - made quoting adhere to standard rules
76 * - added warning for bad backrefs
77 * - added pcrs_execute_list()
79 * - bugfix & cosmetics
81 * Revision 1.11 2001/08/15 15:32:03 oes
82 * - Added support for Perl's special variables $+, $' and $`
83 * - Improved the substitute parser
84 * - Replaced the hard limit for the maximum number of matches
85 * by dynamic reallocation
87 * Revision 1.10 2001/08/05 13:13:11 jongfoster
88 * Making parameters "const" where possible.
90 * Revision 1.9 2001/07/18 17:27:00 oes
91 * Changed interface; Cosmetics
93 * Revision 1.8 2001/06/29 21:45:41 oes
94 * Indentation, CRLF->LF, Tab-> Space
96 * Revision 1.7 2001/06/29 13:33:04 oes
97 * - Cleaned up, renamed and reordered functions,
100 * - Replaced globalflag with a general flags int
101 * that holds PCRS_GLOBAL, PCRS_SUCCESS, and PCRS_TRIVIAL
102 * - Introduced trivial option that will prevent pcrs
103 * from honouring backreferences in the substitute,
104 * which is useful for large substitutes that are
105 * red in from somewhere and saves the pain of escaping
107 * - Introduced convenience function pcrs_free_joblist()
108 * - Split pcrs_make_job() into pcrs_compile(), which still
109 * takes a complete s/// comand as argument and parses it,
110 * and a new function pcrs_make_job, which takes the
111 * three separate components. This should make for a
112 * much friendlier frontend.
113 * - Removed create_pcrs_job() which was useless
114 * - Fixed a bug in pcrs_execute
115 * - Success flag is now handled by pcrs instead of user
117 * Revision 1.6 2001/06/03 19:12:45 oes
120 * Revision 1.5 2001/05/29 09:50:24 jongfoster
121 * (Fixed one int -> size_t)
123 * Revision 1.4 2001/05/25 14:12:40 oes
124 * Fixed bug: Empty substitutes now detected
126 * Revision 1.3 2001/05/25 11:03:55 oes
127 * Added sanity check for NULL jobs to pcrs_exec_substitution
129 * Revision 1.2 2001/05/22 18:46:04 oes
131 * Added support for PCRE_UNGREEDY behaviour to pcrs,
132 * which is selected by the (nonstandard and therefore
133 * capital) letter 'U' in the option string.
134 * It causes the quantifiers to be ungreedy by default.
135 * Appending a ? turns back to greedy (!).
137 * Revision 1.1.1.1 2001/05/15 13:59:02 oes
138 * Initial import of version 2.9.3 source tree
141 *********************************************************************/
145 * Include project.h just so that the right pcre.h gets
146 * included from there
155 const char pcrs_h_rcs[] = PCRS_H_VERSION;
158 * Internal prototypes
161 static int pcrs_parse_perl_options(const char *optstring, int *flags);
162 static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag,
163 int capturecount, int *errptr);
165 /*********************************************************************
167 * Function : pcrs_strerror
169 * Description : Return a string describing a given error code.
172 * 1 : error = the error code
174 * Returns : char * to the descriptive string
176 *********************************************************************/
177 const char *pcrs_strerror(const int error)
183 /* Passed-through PCRE error: */
184 case PCRE_ERROR_NOMEMORY: return "(pcre:) No memory";
186 /* Shouldn't happen unless PCRE or PCRS bug, or user messed with compiled job: */
187 case PCRE_ERROR_NULL: return "(pcre:) NULL code or subject or ovector";
188 case PCRE_ERROR_BADOPTION: return "(pcre:) Unrecognized option bit";
189 case PCRE_ERROR_BADMAGIC: return "(pcre:) Bad magic number in code";
190 case PCRE_ERROR_UNKNOWN_NODE: return "(pcre:) Bad node in pattern";
192 /* Can't happen / not passed: */
193 case PCRE_ERROR_NOSUBSTRING: return "(pcre:) Fire in power supply";
194 case PCRE_ERROR_NOMATCH: return "(pcre:) Water in power supply";
197 case PCRS_ERR_NOMEM: return "(pcrs:) No memory";
198 case PCRS_ERR_CMDSYNTAX: return "(pcrs:) Syntax error while parsing command";
199 case PCRS_ERR_STUDY: return "(pcrs:) PCRE error while studying the pattern";
200 case PCRS_ERR_BADJOB: return "(pcrs:) Bad job - NULL job, pattern or substitute";
201 case PCRS_WARN_BADREF: return "(pcrs:) Backreference out of range";
204 default: return "Unknown error";
207 /* error >= 0: No error */
208 return "(pcrs:) Everything's just fine. Thanks for asking.";
213 /*********************************************************************
215 * Function : pcrs_parse_perl_options
217 * Description : This function parses a string containing the options to
218 * Perl's s/// operator. It returns an integer that is the
219 * pcre equivalent of the symbolic optstring.
220 * Since pcre doesn't know about Perl's 'g' (global) or pcrs',
221 * 'T' (trivial) options but pcrs needs them, the corresponding
222 * flags are set if 'g'or 'T' is encountered.
223 * Note: The 'T' and 'U' options do not conform to Perl.
226 * 1 : optstring = string with options in perl syntax
227 * 2 : flags = see description
229 * Returns : option integer suitable for pcre
231 *********************************************************************/
232 static int pcrs_parse_perl_options(const char *optstring, int *flags)
238 if (NULL == optstring) return 0;
240 for (i = 0; i < strlen(optstring); i++)
244 case 'e': break; /* ToDo ;-) */
245 case 'g': *flags |= PCRS_GLOBAL; break;
246 case 'i': rc |= PCRE_CASELESS; break;
247 case 'm': rc |= PCRE_MULTILINE; break;
249 case 's': rc |= PCRE_DOTALL; break;
250 case 'x': rc |= PCRE_EXTENDED; break;
251 case 'U': rc |= PCRE_UNGREEDY; break;
252 case 'T': *flags |= PCRS_TRIVIAL; break;
261 /*********************************************************************
263 * Function : pcrs_compile_replacement
265 * Description : This function takes a Perl-style replacement (2nd argument
266 * to the s/// operator and returns a compiled pcrs_substitute,
267 * or NULL if memory allocation for the substitute structure
271 * 1 : replacement = replacement part of s/// operator
273 * 2 : trivialflag = Flag that causes backreferences to be
275 * 3 : capturecount = Number of capturing subpatterns in
276 * the pattern. Needed for $+ handling.
277 * 4 : errptr = pointer to an integer in which error
278 * conditions can be returned.
280 * Returns : pcrs_substitute data structure, or NULL if an
281 * error is encountered. In that case, *errptr has
284 *********************************************************************/
285 static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag, int capturecount, int *errptr)
292 i = k = l = quoted = 0;
297 if (NULL == replacement)
305 if (NULL == (r = (pcrs_substitute *)malloc(sizeof(pcrs_substitute))))
307 *errptr = PCRS_ERR_NOMEM;
310 memset(r, '\0', sizeof(pcrs_substitute));
312 length = strlen(replacement);
314 if (NULL == (text = (char *)malloc(length + 1)))
317 *errptr = PCRS_ERR_NOMEM;
320 memset(text, '\0', length + 1);
324 * In trivial mode, just copy the substitute text
328 text = strncpy(text, replacement, length + 1);
333 * Else, parse, cut out and record all backreferences
337 while (i < (int)length)
340 if (replacement[i] == '\\')
344 text[k++] = replacement[i++];
349 if (replacement[i+1] && strchr("tnrfae0", replacement[i+1]))
351 switch (replacement[++i])
387 if (replacement[i] == '$' && !quoted && i < (int)(length - 1))
389 char *symbol, symbols[] = "'`+&";
390 r->block_length[l] = k - r->block_offset[l];
392 /* Numerical backreferences */
393 if (isdigit((int)replacement[i + 1]))
395 while (i < (int)length && isdigit((int)replacement[++i]))
397 r->backref[l] = r->backref[l] * 10 + replacement[i] - 48;
399 if (r->backref[l] > capturecount)
401 *errptr = PCRS_WARN_BADREF;
405 /* Symbolic backreferences: */
406 else if (NULL != (symbol = strchr(symbols, replacement[i + 1])))
409 if (symbol - symbols == 2) /* $+ */
411 r->backref[l] = capturecount;
413 else if (symbol - symbols == 3) /* $& */
419 r->backref[l] = PCRS_MAX_SUBMATCHES + 1 - (symbol - symbols);
424 /* Invalid backref -> plain '$' */
430 /* Valid and in range? -> record */
431 if (r->backref[l] < PCRS_MAX_SUBMATCHES + 2)
433 r->backref_count[r->backref[l]] += 1;
434 r->block_offset[++l] = k;
438 *errptr = PCRS_WARN_BADREF;
444 /* Plain chars are copied */
445 text[k++] = replacement[i++];
448 } /* -END- if (!trivialflag) */
455 r->block_length[l] = k - r->block_offset[l];
462 /*********************************************************************
464 * Function : pcrs_free_job
466 * Description : Frees the memory used by a pcrs_job struct and its
467 * dependant structures.
470 * 1 : job = pointer to the pcrs_job structure to be freed
472 * Returns : a pointer to the next job, if there was any, or
475 *********************************************************************/
476 pcrs_job *pcrs_free_job(pcrs_job *job)
487 if (job->pattern != NULL) free(job->pattern);
488 if (job->hints != NULL) free(job->hints);
489 if (job->substitute != NULL)
491 if (job->substitute->text != NULL) free(job->substitute->text);
492 free(job->substitute);
501 /*********************************************************************
503 * Function : pcrs_free_joblist
505 * Description : Iterates through a chained list of pcrs_job's and
506 * frees them using pcrs_free_job.
509 * 1 : joblist = pointer to the first pcrs_job structure to
514 *********************************************************************/
515 void pcrs_free_joblist(pcrs_job *joblist)
517 while ( NULL != (joblist = pcrs_free_job(joblist)) ) {};
524 /*********************************************************************
526 * Function : pcrs_compile_command
528 * Description : Parses a string with a Perl-style s/// command,
529 * calls pcrs_compile, and returns a corresponding
530 * pcrs_job, or NULL if parsing or compiling the job
534 * 1 : command = string with perl-style s/// command
535 * 2 : errptr = pointer to an integer in which error
536 * conditions can be returned.
538 * Returns : a corresponding pcrs_job data structure, or NULL
539 * if an error was encountered. In that case, *errptr
542 *********************************************************************/
543 pcrs_job *pcrs_compile_command(const char *command, int *errptr)
545 int i, k, l, quoted = FALSE;
554 * Tokenize the perl command
556 limit = strlen(command);
559 *errptr = PCRS_ERR_CMDSYNTAX;
564 delimiter = command[1];
567 tokens[l] = (char *) malloc(limit + 1);
569 for (i = 0; i <= (int)limit; i++)
572 if (command[i] == delimiter && !quoted)
579 tokens[0][k++] = '\0';
580 tokens[++l] = tokens[0] + k;
584 else if (command[i] == '\\' && !quoted)
587 if (command[i+1] == delimiter) continue;
593 tokens[0][k++] = command[i];
601 *errptr = PCRS_ERR_CMDSYNTAX;
606 newjob = pcrs_compile(tokens[1], tokens[2], tokens[3], errptr);
613 /*********************************************************************
615 * Function : pcrs_compile
617 * Description : Takes the three arguments to a perl s/// command
618 * and compiles a pcrs_job structure from them.
621 * 1 : pattern = string with perl-style pattern
622 * 2 : substitute = string with perl-style substitute
623 * 3 : options = string with perl-style options
624 * 4 : errptr = pointer to an integer in which error
625 * conditions can be returned.
627 * Returns : a corresponding pcrs_job data structure, or NULL
628 * if an error was encountered. In that case, *errptr
631 *********************************************************************/
632 pcrs_job *pcrs_compile(const char *pattern, const char *substitute, const char *options, int *errptr)
642 * Handle NULL arguments
644 if (pattern == NULL) pattern = "";
645 if (substitute == NULL) substitute = "";
649 * Get and init memory
651 if (NULL == (newjob = (pcrs_job *)malloc(sizeof(pcrs_job))))
653 *errptr = PCRS_ERR_NOMEM;
656 memset(newjob, '\0', sizeof(pcrs_job));
660 * Evaluate the options
662 newjob->options = pcrs_parse_perl_options(options, &flags);
663 newjob->flags = flags;
667 * Compile the pattern
669 newjob->pattern = pcre_compile(pattern, newjob->options, &error, errptr, NULL);
670 if (newjob->pattern == NULL)
672 pcrs_free_job(newjob);
678 * Generate hints. This has little overhead, since the
679 * hints will be NULL for a boring pattern anyway.
681 newjob->hints = pcre_study(newjob->pattern, 0, &error);
684 *errptr = PCRS_ERR_STUDY;
685 pcrs_free_job(newjob);
691 * Determine the number of capturing subpatterns.
692 * This is needed for handling $+ in the substitute.
694 if (0 > (*errptr = pcre_fullinfo(newjob->pattern, newjob->hints, PCRE_INFO_CAPTURECOUNT, &capturecount)))
696 pcrs_free_job(newjob);
702 * Compile the substitute
704 if (NULL == (newjob->substitute = pcrs_compile_replacement(substitute, newjob->flags & PCRS_TRIVIAL, capturecount, errptr)))
706 pcrs_free_job(newjob);
715 /*********************************************************************
717 * Function : pcrs_execute_list
719 * Description : This is a multiple job wrapper for pcrs_execute().
720 * Apply the regular substitutions defined by the jobs in
721 * the joblist to the subject.
722 * The subject itself is left untouched, memory for the result
723 * is malloc()ed and it is the caller's responsibility to free
724 * the result when it's no longer needed.
727 * 1 : joblist = the chained list of pcrs_jobs to be executed
728 * 2 : subject = the subject string
729 * 3 : subject_length = the subject's length
730 * INCLUDING the terminating zero, if string!
731 * 4 : result = char** for returning the result
732 * 5 : result_length = size_t* for returning the result's length
734 * Returns : On success, the number of substitutions that were made.
735 * May be > 1 if job->flags contained PCRS_GLOBAL
736 * On failure, the (negative) pcre error code describing the
737 * failure, which may be translated to text using pcrs_strerror().
739 *********************************************************************/
740 int pcrs_execute_list(pcrs_job *joblist, char *subject, size_t subject_length, char **result, size_t *result_length)
744 int hits, total_hits;
747 *result_length = subject_length;
748 hits = total_hits = 0;
750 for (job = joblist; job != NULL; job = job->next)
752 hits = pcrs_execute(job, old, *result_length, &new, result_length);
754 if (old != subject) free(old);
773 /*********************************************************************
775 * Function : pcrs_execute
777 * Description : Apply the regular substitution defined by the job to the
779 * The subject itself is left untouched, memory for the result
780 * is malloc()ed and it is the caller's responsibility to free
781 * the result when it's no longer needed.
784 * 1 : job = the pcrs_job to be executed
785 * 2 : subject = the subject (== original) string
786 * 3 : subject_length = the subject's length
787 * INCLUDING the terminating zero, if string!
788 * 4 : result = char** for returning the result
789 * 5 : result_length = size_t* for returning the result's length
791 * Returns : On success, the number of substitutions that were made.
792 * May be > 1 if job->flags contained PCRS_GLOBAL
793 * On failure, the (negative) pcre error code describing the
794 * failure, which may be translated to text using pcrs_strerror().
796 *********************************************************************/
797 int pcrs_execute(pcrs_job *job, char *subject, size_t subject_length, char **result, size_t *result_length)
799 int offsets[3 * PCRS_MAX_SUBMATCHES],
804 max_matches = PCRS_MAX_MATCH_INIT;
806 pcrs_match *matches, *dummy;
812 * Sanity check & memory allocation
814 if (job == NULL || job->pattern == NULL || job->substitute == NULL)
817 return(PCRS_ERR_BADJOB);
820 if (NULL == (matches = (pcrs_match *)malloc(max_matches * sizeof(pcrs_match))))
823 return(PCRS_ERR_NOMEM);
825 memset(matches, '\0', max_matches * sizeof(pcrs_match));
829 * Find the pattern and calculate the space
830 * requirements for the result
832 newsize = subject_length;
834 while ((submatches = pcre_exec(job->pattern, job->hints, subject, (int)subject_length, offset, 0, offsets, 3 * PCRS_MAX_SUBMATCHES)) > 0)
836 job->flags |= PCRS_SUCCESS;
837 matches[i].submatches = submatches;
839 for (k = 0; k < submatches; k++)
841 matches[i].submatch_offset[k] = offsets[2 * k];
843 /* Note: Non-found optional submatches have length -1-(-1)==0 */
844 matches[i].submatch_length[k] = offsets[2 * k + 1] - offsets[2 * k];
846 /* reserve mem for each submatch as often as it is ref'd */
847 newsize += matches[i].submatch_length[k] * job->substitute->backref_count[k];
849 /* plus replacement text size minus match text size */
850 newsize += strlen(job->substitute->text) - matches[i].submatch_length[0];
852 /* chunk before match */
853 matches[i].submatch_offset[PCRS_MAX_SUBMATCHES] = 0;
854 matches[i].submatch_length[PCRS_MAX_SUBMATCHES] = offsets[0];
855 newsize += offsets[0] * job->substitute->backref_count[PCRS_MAX_SUBMATCHES];
857 /* chunk after match */
858 matches[i].submatch_offset[PCRS_MAX_SUBMATCHES + 1] = offsets[1];
859 matches[i].submatch_length[PCRS_MAX_SUBMATCHES + 1] = subject_length - offsets[1] - 1;
860 newsize += (subject_length - offsets[1]) * job->substitute->backref_count[PCRS_MAX_SUBMATCHES + 1];
862 /* Storage for matches exhausted? -> Extend! */
863 if (++i >= max_matches)
865 max_matches = (int)(max_matches * PCRS_MAX_MATCH_GROW);
866 if (NULL == (dummy = (pcrs_match *)realloc(matches, max_matches * sizeof(pcrs_match))))
870 return(PCRS_ERR_NOMEM);
875 /* Non-global search or limit reached? */
876 if (!(job->flags & PCRS_GLOBAL)) break;
878 /* Don't loop on empty matches */
879 if (offsets[1] == offset)
880 if ((size_t)offset < subject_length)
884 /* Go find the next one */
888 /* Pass pcre error through if (bad) failure */
889 if (submatches < PCRE_ERROR_NOMATCH)
898 * Get memory for the result
900 if ((*result = (char *)malloc(newsize)) == NULL) /* must be free()d by caller */
903 return PCRS_ERR_NOMEM;
911 result_offset = *result;
913 for (i = 0; i < matches_found; i++)
915 /* copy the chunk preceding the match */
916 memcpy(result_offset, subject + offset, (size_t)matches[i].submatch_offset[0] - offset);
917 result_offset += matches[i].submatch_offset[0] - offset;
919 /* For every segment of the substitute.. */
920 for (k = 0; k <= job->substitute->backrefs; k++)
922 /* ...copy its text.. */
923 memcpy(result_offset, job->substitute->text + job->substitute->block_offset[k], job->substitute->block_length[k]);
924 result_offset += job->substitute->block_length[k];
926 /* ..plus, if it's not the last chunk, i.e.: There *is* a backref.. */
927 if (k != job->substitute->backrefs
928 /* ..in legal range.. */
929 && job->substitute->backref[k] < PCRS_MAX_SUBMATCHES + 2
930 /* ..and referencing a real submatch.. */
931 && job->substitute->backref[k] < matches[i].submatches
932 /* ..that is nonempty.. */
933 && matches[i].submatch_length[job->substitute->backref[k]] > 0)
935 /* ..copy the submatch that is ref'd. */
938 subject + matches[i].submatch_offset[job->substitute->backref[k]],
939 matches[i].submatch_length[job->substitute->backref[k]]
941 result_offset += matches[i].submatch_length[job->substitute->backref[k]];
944 offset = matches[i].submatch_offset[0] + matches[i].submatch_length[0];
948 memcpy(result_offset, subject + offset, subject_length - offset);
950 *result_length = newsize;
952 return matches_found;