1 const char pcrs_rcs[] = "$Id: pcrs.c,v 2.4 2003/01/21 02:49:27 david__schmidt 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 2.4 2003/01/21 02:49:27 david__schmidt
37 * Developer TODO 612294: src: C++ keyword as variable name
38 * I changed all ocurrences of 'new' to 'new_something' wherever I found
39 * one. I also brought up all the source files in MSDEV to see if I could
40 * spot any highlighted keywords that really were variables. Non-scientific,
41 * but at least I tried. :-)
43 * Revision 2.3 2002/10/08 16:25:30 oes
44 * Bugfix: Need to check validity of backreferences explicitly, because when max_matches are reached and matches is expanded, realloc() does not zero the memory. Fixes Bug # 606227
46 * Revision 2.2 2002/09/04 15:52:02 oes
47 * Synced with the stable branch:
48 * Revision 1.19.2.1 2002/08/10 11:23:40 oes
49 * Include prce.h via project.h, where the appropriate
50 * source will have been selected
52 * Revision 2.1 2002/08/26 11:18:24 sarantis
55 * Revision 2.0 2002/06/04 14:34:21 jongfoster
56 * Moving source files to src/
58 * Revision 1.19 2002/03/08 14:47:48 oes
61 * Revision 1.18 2002/03/08 14:17:14 oes
62 * Fixing -Wconversion warnings
64 * Revision 1.17 2002/03/08 13:45:48 oes
65 * Hiding internal functions
67 * Revision 1.16 2001/11/30 21:32:14 jongfoster
68 * Fixing signed/unsigned comparison (Andreas please check this!)
71 * Revision 1.15 2001/09/20 16:11:06 steudten
73 * Add casting for some string functions.
75 * Revision 1.14 2001/09/09 21:41:57 oes
76 * Fixing yet another silly bug
78 * Revision 1.13 2001/09/06 14:05:59 oes
81 * Revision 1.12 2001/08/18 11:35:00 oes
82 * - Introduced pcrs_strerror()
83 * - made some NULL arguments non-fatal
84 * - added support for \n \r \e \b \t \f \a \0 in substitute
85 * - made quoting adhere to standard rules
86 * - added warning for bad backrefs
87 * - added pcrs_execute_list()
89 * - bugfix & cosmetics
91 * Revision 1.11 2001/08/15 15:32:03 oes
92 * - Added support for Perl's special variables $+, $' and $`
93 * - Improved the substitute parser
94 * - Replaced the hard limit for the maximum number of matches
95 * by dynamic reallocation
97 * Revision 1.10 2001/08/05 13:13:11 jongfoster
98 * Making parameters "const" where possible.
100 * Revision 1.9 2001/07/18 17:27:00 oes
101 * Changed interface; Cosmetics
103 * Revision 1.8 2001/06/29 21:45:41 oes
104 * Indentation, CRLF->LF, Tab-> Space
106 * Revision 1.7 2001/06/29 13:33:04 oes
107 * - Cleaned up, renamed and reordered functions,
109 * - Removed my_strsep
110 * - Replaced globalflag with a general flags int
111 * that holds PCRS_GLOBAL, PCRS_SUCCESS, and PCRS_TRIVIAL
112 * - Introduced trivial option that will prevent pcrs
113 * from honouring backreferences in the substitute,
114 * which is useful for large substitutes that are
115 * red in from somewhere and saves the pain of escaping
117 * - Introduced convenience function pcrs_free_joblist()
118 * - Split pcrs_make_job() into pcrs_compile(), which still
119 * takes a complete s/// comand as argument and parses it,
120 * and a new function pcrs_make_job, which takes the
121 * three separate components. This should make for a
122 * much friendlier frontend.
123 * - Removed create_pcrs_job() which was useless
124 * - Fixed a bug in pcrs_execute
125 * - Success flag is now handled by pcrs instead of user
127 * Revision 1.6 2001/06/03 19:12:45 oes
130 * Revision 1.5 2001/05/29 09:50:24 jongfoster
131 * (Fixed one int -> size_t)
133 * Revision 1.4 2001/05/25 14:12:40 oes
134 * Fixed bug: Empty substitutes now detected
136 * Revision 1.3 2001/05/25 11:03:55 oes
137 * Added sanity check for NULL jobs to pcrs_exec_substitution
139 * Revision 1.2 2001/05/22 18:46:04 oes
141 * Added support for PCRE_UNGREEDY behaviour to pcrs,
142 * which is selected by the (nonstandard and therefore
143 * capital) letter 'U' in the option string.
144 * It causes the quantifiers to be ungreedy by default.
145 * Appending a ? turns back to greedy (!).
147 * Revision 1.1.1.1 2001/05/15 13:59:02 oes
148 * Initial import of version 2.9.3 source tree
151 *********************************************************************/
155 * Include project.h just so that the right pcre.h gets
156 * included from there
165 const char pcrs_h_rcs[] = PCRS_H_VERSION;
168 * Internal prototypes
171 static int pcrs_parse_perl_options(const char *optstring, int *flags);
172 static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag,
173 int capturecount, int *errptr);
175 /*********************************************************************
177 * Function : pcrs_strerror
179 * Description : Return a string describing a given error code.
182 * 1 : error = the error code
184 * Returns : char * to the descriptive string
186 *********************************************************************/
187 const char *pcrs_strerror(const int error)
193 /* Passed-through PCRE error: */
194 case PCRE_ERROR_NOMEMORY: return "(pcre:) No memory";
196 /* Shouldn't happen unless PCRE or PCRS bug, or user messed with compiled job: */
197 case PCRE_ERROR_NULL: return "(pcre:) NULL code or subject or ovector";
198 case PCRE_ERROR_BADOPTION: return "(pcre:) Unrecognized option bit";
199 case PCRE_ERROR_BADMAGIC: return "(pcre:) Bad magic number in code";
200 case PCRE_ERROR_UNKNOWN_NODE: return "(pcre:) Bad node in pattern";
202 /* Can't happen / not passed: */
203 case PCRE_ERROR_NOSUBSTRING: return "(pcre:) Fire in power supply";
204 case PCRE_ERROR_NOMATCH: return "(pcre:) Water in power supply";
207 case PCRS_ERR_NOMEM: return "(pcrs:) No memory";
208 case PCRS_ERR_CMDSYNTAX: return "(pcrs:) Syntax error while parsing command";
209 case PCRS_ERR_STUDY: return "(pcrs:) PCRE error while studying the pattern";
210 case PCRS_ERR_BADJOB: return "(pcrs:) Bad job - NULL job, pattern or substitute";
211 case PCRS_WARN_BADREF: return "(pcrs:) Backreference out of range";
214 default: return "Unknown error";
217 /* error >= 0: No error */
218 return "(pcrs:) Everything's just fine. Thanks for asking.";
223 /*********************************************************************
225 * Function : pcrs_parse_perl_options
227 * Description : This function parses a string containing the options to
228 * Perl's s/// operator. It returns an integer that is the
229 * pcre equivalent of the symbolic optstring.
230 * Since pcre doesn't know about Perl's 'g' (global) or pcrs',
231 * 'T' (trivial) options but pcrs needs them, the corresponding
232 * flags are set if 'g'or 'T' is encountered.
233 * Note: The 'T' and 'U' options do not conform to Perl.
236 * 1 : optstring = string with options in perl syntax
237 * 2 : flags = see description
239 * Returns : option integer suitable for pcre
241 *********************************************************************/
242 static int pcrs_parse_perl_options(const char *optstring, int *flags)
248 if (NULL == optstring) return 0;
250 for (i = 0; i < strlen(optstring); i++)
254 case 'e': break; /* ToDo ;-) */
255 case 'g': *flags |= PCRS_GLOBAL; break;
256 case 'i': rc |= PCRE_CASELESS; break;
257 case 'm': rc |= PCRE_MULTILINE; break;
259 case 's': rc |= PCRE_DOTALL; break;
260 case 'x': rc |= PCRE_EXTENDED; break;
261 case 'U': rc |= PCRE_UNGREEDY; break;
262 case 'T': *flags |= PCRS_TRIVIAL; break;
271 /*********************************************************************
273 * Function : pcrs_compile_replacement
275 * Description : This function takes a Perl-style replacement (2nd argument
276 * to the s/// operator and returns a compiled pcrs_substitute,
277 * or NULL if memory allocation for the substitute structure
281 * 1 : replacement = replacement part of s/// operator
283 * 2 : trivialflag = Flag that causes backreferences to be
285 * 3 : capturecount = Number of capturing subpatterns in
286 * the pattern. Needed for $+ handling.
287 * 4 : errptr = pointer to an integer in which error
288 * conditions can be returned.
290 * Returns : pcrs_substitute data structure, or NULL if an
291 * error is encountered. In that case, *errptr has
294 *********************************************************************/
295 static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag, int capturecount, int *errptr)
302 i = k = l = quoted = 0;
307 if (NULL == replacement)
315 if (NULL == (r = (pcrs_substitute *)malloc(sizeof(pcrs_substitute))))
317 *errptr = PCRS_ERR_NOMEM;
320 memset(r, '\0', sizeof(pcrs_substitute));
322 length = strlen(replacement);
324 if (NULL == (text = (char *)malloc(length + 1)))
327 *errptr = PCRS_ERR_NOMEM;
330 memset(text, '\0', length + 1);
334 * In trivial mode, just copy the substitute text
338 text = strncpy(text, replacement, length + 1);
343 * Else, parse, cut out and record all backreferences
347 while (i < (int)length)
350 if (replacement[i] == '\\')
354 text[k++] = replacement[i++];
359 if (replacement[i+1] && strchr("tnrfae0", replacement[i+1]))
361 switch (replacement[++i])
397 if (replacement[i] == '$' && !quoted && i < (int)(length - 1))
399 char *symbol, symbols[] = "'`+&";
400 r->block_length[l] = k - r->block_offset[l];
402 /* Numerical backreferences */
403 if (isdigit((int)replacement[i + 1]))
405 while (i < (int)length && isdigit((int)replacement[++i]))
407 r->backref[l] = r->backref[l] * 10 + replacement[i] - 48;
409 if (r->backref[l] > capturecount)
411 *errptr = PCRS_WARN_BADREF;
415 /* Symbolic backreferences: */
416 else if (NULL != (symbol = strchr(symbols, replacement[i + 1])))
419 if (symbol - symbols == 2) /* $+ */
421 r->backref[l] = capturecount;
423 else if (symbol - symbols == 3) /* $& */
429 r->backref[l] = PCRS_MAX_SUBMATCHES + 1 - (symbol - symbols);
434 /* Invalid backref -> plain '$' */
440 /* Valid and in range? -> record */
441 if (r->backref[l] < PCRS_MAX_SUBMATCHES + 2)
443 r->backref_count[r->backref[l]] += 1;
444 r->block_offset[++l] = k;
448 *errptr = PCRS_WARN_BADREF;
454 /* Plain chars are copied */
455 text[k++] = replacement[i++];
458 } /* -END- if (!trivialflag) */
465 r->block_length[l] = k - r->block_offset[l];
472 /*********************************************************************
474 * Function : pcrs_free_job
476 * Description : Frees the memory used by a pcrs_job struct and its
477 * dependant structures.
480 * 1 : job = pointer to the pcrs_job structure to be freed
482 * Returns : a pointer to the next job, if there was any, or
485 *********************************************************************/
486 pcrs_job *pcrs_free_job(pcrs_job *job)
497 if (job->pattern != NULL) free(job->pattern);
498 if (job->hints != NULL) free(job->hints);
499 if (job->substitute != NULL)
501 if (job->substitute->text != NULL) free(job->substitute->text);
502 free(job->substitute);
511 /*********************************************************************
513 * Function : pcrs_free_joblist
515 * Description : Iterates through a chained list of pcrs_job's and
516 * frees them using pcrs_free_job.
519 * 1 : joblist = pointer to the first pcrs_job structure to
524 *********************************************************************/
525 void pcrs_free_joblist(pcrs_job *joblist)
527 while ( NULL != (joblist = pcrs_free_job(joblist)) ) {};
534 /*********************************************************************
536 * Function : pcrs_compile_command
538 * Description : Parses a string with a Perl-style s/// command,
539 * calls pcrs_compile, and returns a corresponding
540 * pcrs_job, or NULL if parsing or compiling the job
544 * 1 : command = string with perl-style s/// command
545 * 2 : errptr = pointer to an integer in which error
546 * conditions can be returned.
548 * Returns : a corresponding pcrs_job data structure, or NULL
549 * if an error was encountered. In that case, *errptr
552 *********************************************************************/
553 pcrs_job *pcrs_compile_command(const char *command, int *errptr)
555 int i, k, l, quoted = FALSE;
564 * Tokenize the perl command
566 limit = strlen(command);
569 *errptr = PCRS_ERR_CMDSYNTAX;
574 delimiter = command[1];
577 tokens[l] = (char *) malloc(limit + 1);
579 for (i = 0; i <= (int)limit; i++)
582 if (command[i] == delimiter && !quoted)
589 tokens[0][k++] = '\0';
590 tokens[++l] = tokens[0] + k;
594 else if (command[i] == '\\' && !quoted)
597 if (command[i+1] == delimiter) continue;
603 tokens[0][k++] = command[i];
611 *errptr = PCRS_ERR_CMDSYNTAX;
616 newjob = pcrs_compile(tokens[1], tokens[2], tokens[3], errptr);
623 /*********************************************************************
625 * Function : pcrs_compile
627 * Description : Takes the three arguments to a perl s/// command
628 * and compiles a pcrs_job structure from them.
631 * 1 : pattern = string with perl-style pattern
632 * 2 : substitute = string with perl-style substitute
633 * 3 : options = string with perl-style options
634 * 4 : errptr = pointer to an integer in which error
635 * conditions can be returned.
637 * Returns : a corresponding pcrs_job data structure, or NULL
638 * if an error was encountered. In that case, *errptr
641 *********************************************************************/
642 pcrs_job *pcrs_compile(const char *pattern, const char *substitute, const char *options, int *errptr)
652 * Handle NULL arguments
654 if (pattern == NULL) pattern = "";
655 if (substitute == NULL) substitute = "";
659 * Get and init memory
661 if (NULL == (newjob = (pcrs_job *)malloc(sizeof(pcrs_job))))
663 *errptr = PCRS_ERR_NOMEM;
666 memset(newjob, '\0', sizeof(pcrs_job));
670 * Evaluate the options
672 newjob->options = pcrs_parse_perl_options(options, &flags);
673 newjob->flags = flags;
677 * Compile the pattern
679 newjob->pattern = pcre_compile(pattern, newjob->options, &error, errptr, NULL);
680 if (newjob->pattern == NULL)
682 pcrs_free_job(newjob);
688 * Generate hints. This has little overhead, since the
689 * hints will be NULL for a boring pattern anyway.
691 newjob->hints = pcre_study(newjob->pattern, 0, &error);
694 *errptr = PCRS_ERR_STUDY;
695 pcrs_free_job(newjob);
701 * Determine the number of capturing subpatterns.
702 * This is needed for handling $+ in the substitute.
704 if (0 > (*errptr = pcre_fullinfo(newjob->pattern, newjob->hints, PCRE_INFO_CAPTURECOUNT, &capturecount)))
706 pcrs_free_job(newjob);
712 * Compile the substitute
714 if (NULL == (newjob->substitute = pcrs_compile_replacement(substitute, newjob->flags & PCRS_TRIVIAL, capturecount, errptr)))
716 pcrs_free_job(newjob);
725 /*********************************************************************
727 * Function : pcrs_execute_list
729 * Description : This is a multiple job wrapper for pcrs_execute().
730 * Apply the regular substitutions defined by the jobs in
731 * the joblist to the subject.
732 * The subject itself is left untouched, memory for the result
733 * is malloc()ed and it is the caller's responsibility to free
734 * the result when it's no longer needed.
737 * 1 : joblist = the chained list of pcrs_jobs to be executed
738 * 2 : subject = the subject string
739 * 3 : subject_length = the subject's length
740 * INCLUDING the terminating zero, if string!
741 * 4 : result = char** for returning the result
742 * 5 : result_length = size_t* for returning the result's length
744 * Returns : On success, the number of substitutions that were made.
745 * May be > 1 if job->flags contained PCRS_GLOBAL
746 * On failure, the (negative) pcre error code describing the
747 * failure, which may be translated to text using pcrs_strerror().
749 *********************************************************************/
750 int pcrs_execute_list(pcrs_job *joblist, char *subject, size_t subject_length, char **result, size_t *result_length)
753 char *old_item, *new_item;
754 int hits, total_hits;
757 *result_length = subject_length;
758 hits = total_hits = 0;
760 for (job = joblist; job != NULL; job = job->next)
762 hits = pcrs_execute(job, old_item, *result_length, &new_item, result_length);
764 if (old_item != subject) free(old_item);
783 /*********************************************************************
785 * Function : pcrs_execute
787 * Description : Apply the regular substitution defined by the job to the
789 * The subject itself is left untouched, memory for the result
790 * is malloc()ed and it is the caller's responsibility to free
791 * the result when it's no longer needed.
794 * 1 : job = the pcrs_job to be executed
795 * 2 : subject = the subject (== original) string
796 * 3 : subject_length = the subject's length
797 * INCLUDING the terminating zero, if string!
798 * 4 : result = char** for returning the result
799 * 5 : result_length = size_t* for returning the result's length
801 * Returns : On success, the number of substitutions that were made.
802 * May be > 1 if job->flags contained PCRS_GLOBAL
803 * On failure, the (negative) pcre error code describing the
804 * failure, which may be translated to text using pcrs_strerror().
806 *********************************************************************/
807 int pcrs_execute(pcrs_job *job, char *subject, size_t subject_length, char **result, size_t *result_length)
809 int offsets[3 * PCRS_MAX_SUBMATCHES],
814 max_matches = PCRS_MAX_MATCH_INIT;
816 pcrs_match *matches, *dummy;
822 * Sanity check & memory allocation
824 if (job == NULL || job->pattern == NULL || job->substitute == NULL)
827 return(PCRS_ERR_BADJOB);
830 if (NULL == (matches = (pcrs_match *)malloc(max_matches * sizeof(pcrs_match))))
833 return(PCRS_ERR_NOMEM);
835 memset(matches, '\0', max_matches * sizeof(pcrs_match));
839 * Find the pattern and calculate the space
840 * requirements for the result
842 newsize = subject_length;
844 while ((submatches = pcre_exec(job->pattern, job->hints, subject, (int)subject_length, offset, 0, offsets, 3 * PCRS_MAX_SUBMATCHES)) > 0)
846 job->flags |= PCRS_SUCCESS;
847 matches[i].submatches = submatches;
849 for (k = 0; k < submatches; k++)
851 matches[i].submatch_offset[k] = offsets[2 * k];
853 /* Note: Non-found optional submatches have length -1-(-1)==0 */
854 matches[i].submatch_length[k] = offsets[2 * k + 1] - offsets[2 * k];
856 /* reserve mem for each submatch as often as it is ref'd */
857 newsize += matches[i].submatch_length[k] * job->substitute->backref_count[k];
859 /* plus replacement text size minus match text size */
860 newsize += strlen(job->substitute->text) - matches[i].submatch_length[0];
862 /* chunk before match */
863 matches[i].submatch_offset[PCRS_MAX_SUBMATCHES] = 0;
864 matches[i].submatch_length[PCRS_MAX_SUBMATCHES] = offsets[0];
865 newsize += offsets[0] * job->substitute->backref_count[PCRS_MAX_SUBMATCHES];
867 /* chunk after match */
868 matches[i].submatch_offset[PCRS_MAX_SUBMATCHES + 1] = offsets[1];
869 matches[i].submatch_length[PCRS_MAX_SUBMATCHES + 1] = subject_length - offsets[1] - 1;
870 newsize += (subject_length - offsets[1]) * job->substitute->backref_count[PCRS_MAX_SUBMATCHES + 1];
872 /* Storage for matches exhausted? -> Extend! */
873 if (++i >= max_matches)
875 max_matches = (int)(max_matches * PCRS_MAX_MATCH_GROW);
876 if (NULL == (dummy = (pcrs_match *)realloc(matches, max_matches * sizeof(pcrs_match))))
880 return(PCRS_ERR_NOMEM);
885 /* Non-global search or limit reached? */
886 if (!(job->flags & PCRS_GLOBAL)) break;
888 /* Don't loop on empty matches */
889 if (offsets[1] == offset)
890 if ((size_t)offset < subject_length)
894 /* Go find the next one */
898 /* Pass pcre error through if (bad) failure */
899 if (submatches < PCRE_ERROR_NOMATCH)
908 * Get memory for the result
910 if ((*result = (char *)malloc(newsize)) == NULL) /* must be free()d by caller */
913 return PCRS_ERR_NOMEM;
921 result_offset = *result;
923 for (i = 0; i < matches_found; i++)
925 /* copy the chunk preceding the match */
926 memcpy(result_offset, subject + offset, (size_t)matches[i].submatch_offset[0] - offset);
927 result_offset += matches[i].submatch_offset[0] - offset;
929 /* For every segment of the substitute.. */
930 for (k = 0; k <= job->substitute->backrefs; k++)
932 /* ...copy its text.. */
933 memcpy(result_offset, job->substitute->text + job->substitute->block_offset[k], job->substitute->block_length[k]);
934 result_offset += job->substitute->block_length[k];
936 /* ..plus, if it's not the last chunk, i.e.: There *is* a backref.. */
937 if (k != job->substitute->backrefs
938 /* ..in legal range.. */
939 && job->substitute->backref[k] < PCRS_MAX_SUBMATCHES + 2
940 /* ..and referencing a real submatch.. */
941 && job->substitute->backref[k] < matches[i].submatches
942 /* ..that is nonempty.. */
943 && matches[i].submatch_length[job->substitute->backref[k]] > 0)
945 /* ..copy the submatch that is ref'd. */
948 subject + matches[i].submatch_offset[job->substitute->backref[k]],
949 matches[i].submatch_length[job->substitute->backref[k]]
951 result_offset += matches[i].submatch_length[job->substitute->backref[k]];
954 offset = matches[i].submatch_offset[0] + matches[i].submatch_length[0];
958 memcpy(result_offset, subject + offset, subject_length - offset);
960 *result_length = newsize;
962 return matches_found;