1 const char pcrs_rcs[] = "$Id: pcrs.c,v 1.19 2002/03/08 14:47:48 oes Exp $";
3 /*********************************************************************
5 * File : $Source: /cvsroot/ijbswa/current/Attic/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 1.19 2002/03/08 14:47:48 oes
39 * Revision 1.18 2002/03/08 14:17:14 oes
40 * Fixing -Wconversion warnings
42 * Revision 1.17 2002/03/08 13:45:48 oes
43 * Hiding internal functions
45 * Revision 1.16 2001/11/30 21:32:14 jongfoster
46 * Fixing signed/unsigned comparison (Andreas please check this!)
49 * Revision 1.15 2001/09/20 16:11:06 steudten
51 * Add casting for some string functions.
53 * Revision 1.14 2001/09/09 21:41:57 oes
54 * Fixing yet another silly bug
56 * Revision 1.13 2001/09/06 14:05:59 oes
59 * Revision 1.12 2001/08/18 11:35:00 oes
60 * - Introduced pcrs_strerror()
61 * - made some NULL arguments non-fatal
62 * - added support for \n \r \e \b \t \f \a \0 in substitute
63 * - made quoting adhere to standard rules
64 * - added warning for bad backrefs
65 * - added pcrs_execute_list()
67 * - bugfix & cosmetics
69 * Revision 1.11 2001/08/15 15:32:03 oes
70 * - Added support for Perl's special variables $+, $' and $`
71 * - Improved the substitute parser
72 * - Replaced the hard limit for the maximum number of matches
73 * by dynamic reallocation
75 * Revision 1.10 2001/08/05 13:13:11 jongfoster
76 * Making parameters "const" where possible.
78 * Revision 1.9 2001/07/18 17:27:00 oes
79 * Changed interface; Cosmetics
81 * Revision 1.8 2001/06/29 21:45:41 oes
82 * Indentation, CRLF->LF, Tab-> Space
84 * Revision 1.7 2001/06/29 13:33:04 oes
85 * - Cleaned up, renamed and reordered functions,
88 * - Replaced globalflag with a general flags int
89 * that holds PCRS_GLOBAL, PCRS_SUCCESS, and PCRS_TRIVIAL
90 * - Introduced trivial option that will prevent pcrs
91 * from honouring backreferences in the substitute,
92 * which is useful for large substitutes that are
93 * red in from somewhere and saves the pain of escaping
95 * - Introduced convenience function pcrs_free_joblist()
96 * - Split pcrs_make_job() into pcrs_compile(), which still
97 * takes a complete s/// comand as argument and parses it,
98 * and a new function pcrs_make_job, which takes the
99 * three separate components. This should make for a
100 * much friendlier frontend.
101 * - Removed create_pcrs_job() which was useless
102 * - Fixed a bug in pcrs_execute
103 * - Success flag is now handled by pcrs instead of user
105 * Revision 1.6 2001/06/03 19:12:45 oes
108 * Revision 1.5 2001/05/29 09:50:24 jongfoster
109 * (Fixed one int -> size_t)
111 * Revision 1.4 2001/05/25 14:12:40 oes
112 * Fixed bug: Empty substitutes now detected
114 * Revision 1.3 2001/05/25 11:03:55 oes
115 * Added sanity check for NULL jobs to pcrs_exec_substitution
117 * Revision 1.2 2001/05/22 18:46:04 oes
119 * Added support for PCRE_UNGREEDY behaviour to pcrs,
120 * which is selected by the (nonstandard and therefore
121 * capital) letter 'U' in the option string.
122 * It causes the quantifiers to be ungreedy by default.
123 * Appending a ? turns back to greedy (!).
125 * Revision 1.1.1.1 2001/05/15 13:59:02 oes
126 * Initial import of version 2.9.3 source tree
129 *********************************************************************/
133 * Include project.h just so that the right pcre.h gets
134 * included from there
143 const char pcrs_h_rcs[] = PCRS_H_VERSION;
146 * Internal prototypes
149 static int pcrs_parse_perl_options(const char *optstring, int *flags);
150 static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag,
151 int capturecount, int *errptr);
153 /*********************************************************************
155 * Function : pcrs_strerror
157 * Description : Return a string describing a given error code.
160 * 1 : error = the error code
162 * Returns : char * to the descriptive string
164 *********************************************************************/
165 const char *pcrs_strerror(const int error)
171 /* Passed-through PCRE error: */
172 case PCRE_ERROR_NOMEMORY: return "(pcre:) No memory";
174 /* Shouldn't happen unless PCRE or PCRS bug, or user messed with compiled job: */
175 case PCRE_ERROR_NULL: return "(pcre:) NULL code or subject or ovector";
176 case PCRE_ERROR_BADOPTION: return "(pcre:) Unrecognized option bit";
177 case PCRE_ERROR_BADMAGIC: return "(pcre:) Bad magic number in code";
178 case PCRE_ERROR_UNKNOWN_NODE: return "(pcre:) Bad node in pattern";
180 /* Can't happen / not passed: */
181 case PCRE_ERROR_NOSUBSTRING: return "(pcre:) Fire in power supply";
182 case PCRE_ERROR_NOMATCH: return "(pcre:) Water in power supply";
185 case PCRS_ERR_NOMEM: return "(pcrs:) No memory";
186 case PCRS_ERR_CMDSYNTAX: return "(pcrs:) Syntax error while parsing command";
187 case PCRS_ERR_STUDY: return "(pcrs:) PCRE error while studying the pattern";
188 case PCRS_ERR_BADJOB: return "(pcrs:) Bad job - NULL job, pattern or substitute";
189 case PCRS_WARN_BADREF: return "(pcrs:) Backreference out of range";
192 default: return "Unknown error";
195 /* error >= 0: No error */
196 return "(pcrs:) Everything's just fine. Thanks for asking.";
201 /*********************************************************************
203 * Function : pcrs_parse_perl_options
205 * Description : This function parses a string containing the options to
206 * Perl's s/// operator. It returns an integer that is the
207 * pcre equivalent of the symbolic optstring.
208 * Since pcre doesn't know about Perl's 'g' (global) or pcrs',
209 * 'T' (trivial) options but pcrs needs them, the corresponding
210 * flags are set if 'g'or 'T' is encountered.
211 * Note: The 'T' and 'U' options do not conform to Perl.
214 * 1 : optstring = string with options in perl syntax
215 * 2 : flags = see description
217 * Returns : option integer suitable for pcre
219 *********************************************************************/
220 static int pcrs_parse_perl_options(const char *optstring, int *flags)
226 if (NULL == optstring) return 0;
228 for (i = 0; i < strlen(optstring); i++)
232 case 'e': break; /* ToDo ;-) */
233 case 'g': *flags |= PCRS_GLOBAL; break;
234 case 'i': rc |= PCRE_CASELESS; break;
235 case 'm': rc |= PCRE_MULTILINE; break;
237 case 's': rc |= PCRE_DOTALL; break;
238 case 'x': rc |= PCRE_EXTENDED; break;
239 case 'U': rc |= PCRE_UNGREEDY; break;
240 case 'T': *flags |= PCRS_TRIVIAL; break;
249 /*********************************************************************
251 * Function : pcrs_compile_replacement
253 * Description : This function takes a Perl-style replacement (2nd argument
254 * to the s/// operator and returns a compiled pcrs_substitute,
255 * or NULL if memory allocation for the substitute structure
259 * 1 : replacement = replacement part of s/// operator
261 * 2 : trivialflag = Flag that causes backreferences to be
263 * 3 : capturecount = Number of capturing subpatterns in
264 * the pattern. Needed for $+ handling.
265 * 4 : errptr = pointer to an integer in which error
266 * conditions can be returned.
268 * Returns : pcrs_substitute data structure, or NULL if an
269 * error is encountered. In that case, *errptr has
272 *********************************************************************/
273 static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag, int capturecount, int *errptr)
280 i = k = l = quoted = 0;
285 if (NULL == replacement)
293 if (NULL == (r = (pcrs_substitute *)malloc(sizeof(pcrs_substitute))))
295 *errptr = PCRS_ERR_NOMEM;
298 memset(r, '\0', sizeof(pcrs_substitute));
300 length = strlen(replacement);
302 if (NULL == (text = (char *)malloc(length + 1)))
305 *errptr = PCRS_ERR_NOMEM;
308 memset(text, '\0', length + 1);
312 * In trivial mode, just copy the substitute text
316 text = strncpy(text, replacement, length + 1);
321 * Else, parse, cut out and record all backreferences
325 while (i < (int)length)
328 if (replacement[i] == '\\')
332 text[k++] = replacement[i++];
337 if (replacement[i+1] && strchr("tnrfae0", replacement[i+1]))
339 switch (replacement[++i])
375 if (replacement[i] == '$' && !quoted && i < (int)(length - 1))
377 char *symbol, symbols[] = "'`+&";
378 r->block_length[l] = k - r->block_offset[l];
380 /* Numerical backreferences */
381 if (isdigit((int)replacement[i + 1]))
383 while (i < (int)length && isdigit((int)replacement[++i]))
385 r->backref[l] = r->backref[l] * 10 + replacement[i] - 48;
387 if (r->backref[l] > capturecount)
389 *errptr = PCRS_WARN_BADREF;
393 /* Symbolic backreferences: */
394 else if (NULL != (symbol = strchr(symbols, replacement[i + 1])))
397 if (symbol - symbols == 2) /* $+ */
399 r->backref[l] = capturecount;
401 else if (symbol - symbols == 3) /* $& */
407 r->backref[l] = PCRS_MAX_SUBMATCHES + 1 - (symbol - symbols);
412 /* Invalid backref -> plain '$' */
418 /* Valid and in range? -> record */
419 if (r->backref[l] < PCRS_MAX_SUBMATCHES + 2)
421 r->backref_count[r->backref[l]] += 1;
422 r->block_offset[++l] = k;
426 *errptr = PCRS_WARN_BADREF;
432 /* Plain chars are copied */
433 text[k++] = replacement[i++];
436 } /* -END- if (!trivialflag) */
443 r->block_length[l] = k - r->block_offset[l];
450 /*********************************************************************
452 * Function : pcrs_free_job
454 * Description : Frees the memory used by a pcrs_job struct and its
455 * dependant structures.
458 * 1 : job = pointer to the pcrs_job structure to be freed
460 * Returns : a pointer to the next job, if there was any, or
463 *********************************************************************/
464 pcrs_job *pcrs_free_job(pcrs_job *job)
475 if (job->pattern != NULL) free(job->pattern);
476 if (job->hints != NULL) free(job->hints);
477 if (job->substitute != NULL)
479 if (job->substitute->text != NULL) free(job->substitute->text);
480 free(job->substitute);
489 /*********************************************************************
491 * Function : pcrs_free_joblist
493 * Description : Iterates through a chained list of pcrs_job's and
494 * frees them using pcrs_free_job.
497 * 1 : joblist = pointer to the first pcrs_job structure to
502 *********************************************************************/
503 void pcrs_free_joblist(pcrs_job *joblist)
505 while ( NULL != (joblist = pcrs_free_job(joblist)) ) {};
512 /*********************************************************************
514 * Function : pcrs_compile_command
516 * Description : Parses a string with a Perl-style s/// command,
517 * calls pcrs_compile, and returns a corresponding
518 * pcrs_job, or NULL if parsing or compiling the job
522 * 1 : command = string with perl-style s/// command
523 * 2 : errptr = pointer to an integer in which error
524 * conditions can be returned.
526 * Returns : a corresponding pcrs_job data structure, or NULL
527 * if an error was encountered. In that case, *errptr
530 *********************************************************************/
531 pcrs_job *pcrs_compile_command(const char *command, int *errptr)
533 int i, k, l, quoted = FALSE;
542 * Tokenize the perl command
544 limit = strlen(command);
547 *errptr = PCRS_ERR_CMDSYNTAX;
552 delimiter = command[1];
555 tokens[l] = (char *) malloc(limit + 1);
557 for (i = 0; i <= (int)limit; i++)
560 if (command[i] == delimiter && !quoted)
567 tokens[0][k++] = '\0';
568 tokens[++l] = tokens[0] + k;
572 else if (command[i] == '\\' && !quoted)
575 if (command[i+1] == delimiter) continue;
581 tokens[0][k++] = command[i];
589 *errptr = PCRS_ERR_CMDSYNTAX;
594 newjob = pcrs_compile(tokens[1], tokens[2], tokens[3], errptr);
601 /*********************************************************************
603 * Function : pcrs_compile
605 * Description : Takes the three arguments to a perl s/// command
606 * and compiles a pcrs_job structure from them.
609 * 1 : pattern = string with perl-style pattern
610 * 2 : substitute = string with perl-style substitute
611 * 3 : options = string with perl-style options
612 * 4 : errptr = pointer to an integer in which error
613 * conditions can be returned.
615 * Returns : a corresponding pcrs_job data structure, or NULL
616 * if an error was encountered. In that case, *errptr
619 *********************************************************************/
620 pcrs_job *pcrs_compile(const char *pattern, const char *substitute, const char *options, int *errptr)
630 * Handle NULL arguments
632 if (pattern == NULL) pattern = "";
633 if (substitute == NULL) substitute = "";
637 * Get and init memory
639 if (NULL == (newjob = (pcrs_job *)malloc(sizeof(pcrs_job))))
641 *errptr = PCRS_ERR_NOMEM;
644 memset(newjob, '\0', sizeof(pcrs_job));
648 * Evaluate the options
650 newjob->options = pcrs_parse_perl_options(options, &flags);
651 newjob->flags = flags;
655 * Compile the pattern
657 newjob->pattern = pcre_compile(pattern, newjob->options, &error, errptr, NULL);
658 if (newjob->pattern == NULL)
660 pcrs_free_job(newjob);
666 * Generate hints. This has little overhead, since the
667 * hints will be NULL for a boring pattern anyway.
669 newjob->hints = pcre_study(newjob->pattern, 0, &error);
672 *errptr = PCRS_ERR_STUDY;
673 pcrs_free_job(newjob);
679 * Determine the number of capturing subpatterns.
680 * This is needed for handling $+ in the substitute.
682 if (0 > (*errptr = pcre_fullinfo(newjob->pattern, newjob->hints, PCRE_INFO_CAPTURECOUNT, &capturecount)))
684 pcrs_free_job(newjob);
690 * Compile the substitute
692 if (NULL == (newjob->substitute = pcrs_compile_replacement(substitute, newjob->flags & PCRS_TRIVIAL, capturecount, errptr)))
694 pcrs_free_job(newjob);
703 /*********************************************************************
705 * Function : pcrs_execute_list
707 * Description : This is a multiple job wrapper for pcrs_execute().
708 * Apply the regular substitutions defined by the jobs in
709 * the joblist to the subject.
710 * The subject itself is left untouched, memory for the result
711 * is malloc()ed and it is the caller's responsibility to free
712 * the result when it's no longer needed.
715 * 1 : joblist = the chained list of pcrs_jobs to be executed
716 * 2 : subject = the subject string
717 * 3 : subject_length = the subject's length
718 * INCLUDING the terminating zero, if string!
719 * 4 : result = char** for returning the result
720 * 5 : result_length = size_t* for returning the result's length
722 * Returns : On success, the number of substitutions that were made.
723 * May be > 1 if job->flags contained PCRS_GLOBAL
724 * On failiure, the (negative) pcre error code describing the
725 * failiure, which may be translated to text using pcrs_strerror().
727 *********************************************************************/
728 int pcrs_execute_list(pcrs_job *joblist, char *subject, size_t subject_length, char **result, size_t *result_length)
732 int hits, total_hits;
735 *result_length = subject_length;
736 hits = total_hits = 0;
738 for (job = joblist; job != NULL; job = job->next)
740 hits = pcrs_execute(job, old, *result_length, &new, result_length);
742 if (old != subject) free(old);
761 /*********************************************************************
763 * Function : pcrs_execute
765 * Description : Apply the regular substitution defined by the job to the
767 * The subject itself is left untouched, memory for the result
768 * is malloc()ed and it is the caller's responsibility to free
769 * the result when it's no longer needed.
772 * 1 : job = the pcrs_job to be executed
773 * 2 : subject = the subject (== original) string
774 * 3 : subject_length = the subject's length
775 * INCLUDING the terminating zero, if string!
776 * 4 : result = char** for returning the result
777 * 5 : result_length = size_t* for returning the result's length
779 * Returns : On success, the number of substitutions that were made.
780 * May be > 1 if job->flags contained PCRS_GLOBAL
781 * On failiure, the (negative) pcre error code describing the
782 * failiure, which may be translated to text using pcrs_strerror().
784 *********************************************************************/
785 int pcrs_execute(pcrs_job *job, char *subject, size_t subject_length, char **result, size_t *result_length)
787 int offsets[3 * PCRS_MAX_SUBMATCHES],
792 max_matches = PCRS_MAX_MATCH_INIT;
794 pcrs_match *matches, *dummy;
800 * Sanity check & memory allocation
802 if (job == NULL || job->pattern == NULL || job->substitute == NULL)
805 return(PCRS_ERR_BADJOB);
808 if (NULL == (matches = (pcrs_match *)malloc(max_matches * sizeof(pcrs_match))))
811 return(PCRS_ERR_NOMEM);
813 memset(matches, '\0', max_matches * sizeof(pcrs_match));
817 * Find the pattern and calculate the space
818 * requirements for the result
820 newsize = subject_length;
822 while ((submatches = pcre_exec(job->pattern, job->hints, subject, (int)subject_length, offset, 0, offsets, 3 * PCRS_MAX_SUBMATCHES)) > 0)
824 job->flags |= PCRS_SUCCESS;
825 matches[i].submatches = submatches;
827 for (k = 0; k < submatches; k++)
829 matches[i].submatch_offset[k] = offsets[2 * k];
831 /* Note: Non-found optional submatches have length -1-(-1)==0 */
832 matches[i].submatch_length[k] = offsets[2 * k + 1] - offsets[2 * k];
834 /* reserve mem for each submatch as often as it is ref'd */
835 newsize += matches[i].submatch_length[k] * job->substitute->backref_count[k];
837 /* plus replacement text size minus match text size */
838 newsize += strlen(job->substitute->text) - matches[i].submatch_length[0];
840 /* chunk before match */
841 matches[i].submatch_offset[PCRS_MAX_SUBMATCHES] = 0;
842 matches[i].submatch_length[PCRS_MAX_SUBMATCHES] = offsets[0];
843 newsize += offsets[0] * job->substitute->backref_count[PCRS_MAX_SUBMATCHES];
845 /* chunk after match */
846 matches[i].submatch_offset[PCRS_MAX_SUBMATCHES + 1] = offsets[1];
847 matches[i].submatch_length[PCRS_MAX_SUBMATCHES + 1] = subject_length - offsets[1] - 1;
848 newsize += (subject_length - offsets[1]) * job->substitute->backref_count[PCRS_MAX_SUBMATCHES + 1];
850 /* Storage for matches exhausted? -> Extend! */
851 if (++i >= max_matches)
853 max_matches = (int)(max_matches * PCRS_MAX_MATCH_GROW);
854 if (NULL == (dummy = (pcrs_match *)realloc(matches, max_matches * sizeof(pcrs_match))))
858 return(PCRS_ERR_NOMEM);
863 /* Non-global search or limit reached? */
864 if (!(job->flags & PCRS_GLOBAL)) break;
866 /* Don't loop on empty matches */
867 if (offsets[1] == offset)
868 if ((size_t)offset < subject_length)
872 /* Go find the next one */
876 /* Pass pcre error through if (bad) failiure */
877 if (submatches < PCRE_ERROR_NOMATCH)
886 * Get memory for the result
888 if ((*result = (char *)malloc(newsize)) == NULL) /* must be free()d by caller */
891 return PCRS_ERR_NOMEM;
899 result_offset = *result;
901 for (i = 0; i < matches_found; i++)
903 /* copy the chunk preceding the match */
904 memcpy(result_offset, subject + offset, (size_t)matches[i].submatch_offset[0] - offset);
905 result_offset += matches[i].submatch_offset[0] - offset;
907 /* For every segment of the substitute.. */
908 for (k = 0; k <= job->substitute->backrefs; k++)
910 /* ...copy its text.. */
911 memcpy(result_offset, job->substitute->text + job->substitute->block_offset[k], job->substitute->block_length[k]);
912 result_offset += job->substitute->block_length[k];
914 /* ..plus, if it's not the last chunk, i.e.: There *is* a backref.. */
915 if (k != job->substitute->backrefs
916 /* ..in legal range.. */
917 && job->substitute->backref[k] < PCRS_MAX_SUBMATCHES + 2
918 /* ..and referencing a nonempty match.. */
919 && matches[i].submatch_length[job->substitute->backref[k]] > 0)
921 /* ..copy the submatch that is ref'd. */
924 subject + matches[i].submatch_offset[job->substitute->backref[k]],
925 matches[i].submatch_length[job->substitute->backref[k]]
927 result_offset += matches[i].submatch_length[job->substitute->backref[k]];
930 offset = matches[i].submatch_offset[0] + matches[i].submatch_length[0];
934 memcpy(result_offset, subject + offset, subject_length - offset);
936 *result_length = newsize;
938 return matches_found;