1 const char pcrs_rcs[] = "$Id: pcrs.c,v 2.0 2002/06/04 14:34:21 jongfoster 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 2.0 2002/06/04 14:34:21 jongfoster
37 * Moving source files to src/
39 * Revision 1.19 2002/03/08 14:47:48 oes
42 * Revision 1.18 2002/03/08 14:17:14 oes
43 * Fixing -Wconversion warnings
45 * Revision 1.17 2002/03/08 13:45:48 oes
46 * Hiding internal functions
48 * Revision 1.16 2001/11/30 21:32:14 jongfoster
49 * Fixing signed/unsigned comparison (Andreas please check this!)
52 * Revision 1.15 2001/09/20 16:11:06 steudten
54 * Add casting for some string functions.
56 * Revision 1.14 2001/09/09 21:41:57 oes
57 * Fixing yet another silly bug
59 * Revision 1.13 2001/09/06 14:05:59 oes
62 * Revision 1.12 2001/08/18 11:35:00 oes
63 * - Introduced pcrs_strerror()
64 * - made some NULL arguments non-fatal
65 * - added support for \n \r \e \b \t \f \a \0 in substitute
66 * - made quoting adhere to standard rules
67 * - added warning for bad backrefs
68 * - added pcrs_execute_list()
70 * - bugfix & cosmetics
72 * Revision 1.11 2001/08/15 15:32:03 oes
73 * - Added support for Perl's special variables $+, $' and $`
74 * - Improved the substitute parser
75 * - Replaced the hard limit for the maximum number of matches
76 * by dynamic reallocation
78 * Revision 1.10 2001/08/05 13:13:11 jongfoster
79 * Making parameters "const" where possible.
81 * Revision 1.9 2001/07/18 17:27:00 oes
82 * Changed interface; Cosmetics
84 * Revision 1.8 2001/06/29 21:45:41 oes
85 * Indentation, CRLF->LF, Tab-> Space
87 * Revision 1.7 2001/06/29 13:33:04 oes
88 * - Cleaned up, renamed and reordered functions,
91 * - Replaced globalflag with a general flags int
92 * that holds PCRS_GLOBAL, PCRS_SUCCESS, and PCRS_TRIVIAL
93 * - Introduced trivial option that will prevent pcrs
94 * from honouring backreferences in the substitute,
95 * which is useful for large substitutes that are
96 * red in from somewhere and saves the pain of escaping
98 * - Introduced convenience function pcrs_free_joblist()
99 * - Split pcrs_make_job() into pcrs_compile(), which still
100 * takes a complete s/// comand as argument and parses it,
101 * and a new function pcrs_make_job, which takes the
102 * three separate components. This should make for a
103 * much friendlier frontend.
104 * - Removed create_pcrs_job() which was useless
105 * - Fixed a bug in pcrs_execute
106 * - Success flag is now handled by pcrs instead of user
108 * Revision 1.6 2001/06/03 19:12:45 oes
111 * Revision 1.5 2001/05/29 09:50:24 jongfoster
112 * (Fixed one int -> size_t)
114 * Revision 1.4 2001/05/25 14:12:40 oes
115 * Fixed bug: Empty substitutes now detected
117 * Revision 1.3 2001/05/25 11:03:55 oes
118 * Added sanity check for NULL jobs to pcrs_exec_substitution
120 * Revision 1.2 2001/05/22 18:46:04 oes
122 * Added support for PCRE_UNGREEDY behaviour to pcrs,
123 * which is selected by the (nonstandard and therefore
124 * capital) letter 'U' in the option string.
125 * It causes the quantifiers to be ungreedy by default.
126 * Appending a ? turns back to greedy (!).
128 * Revision 1.1.1.1 2001/05/15 13:59:02 oes
129 * Initial import of version 2.9.3 source tree
132 *********************************************************************/
141 const char pcrs_h_rcs[] = PCRS_H_VERSION;
144 * Internal prototypes
147 static int pcrs_parse_perl_options(const char *optstring, int *flags);
148 static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag,
149 int capturecount, int *errptr);
151 /*********************************************************************
153 * Function : pcrs_strerror
155 * Description : Return a string describing a given error code.
158 * 1 : error = the error code
160 * Returns : char * to the descriptive string
162 *********************************************************************/
163 const char *pcrs_strerror(const int error)
169 /* Passed-through PCRE error: */
170 case PCRE_ERROR_NOMEMORY: return "(pcre:) No memory";
172 /* Shouldn't happen unless PCRE or PCRS bug, or user messed with compiled job: */
173 case PCRE_ERROR_NULL: return "(pcre:) NULL code or subject or ovector";
174 case PCRE_ERROR_BADOPTION: return "(pcre:) Unrecognized option bit";
175 case PCRE_ERROR_BADMAGIC: return "(pcre:) Bad magic number in code";
176 case PCRE_ERROR_UNKNOWN_NODE: return "(pcre:) Bad node in pattern";
178 /* Can't happen / not passed: */
179 case PCRE_ERROR_NOSUBSTRING: return "(pcre:) Fire in power supply";
180 case PCRE_ERROR_NOMATCH: return "(pcre:) Water in power supply";
183 case PCRS_ERR_NOMEM: return "(pcrs:) No memory";
184 case PCRS_ERR_CMDSYNTAX: return "(pcrs:) Syntax error while parsing command";
185 case PCRS_ERR_STUDY: return "(pcrs:) PCRE error while studying the pattern";
186 case PCRS_ERR_BADJOB: return "(pcrs:) Bad job - NULL job, pattern or substitute";
187 case PCRS_WARN_BADREF: return "(pcrs:) Backreference out of range";
190 default: return "Unknown error";
193 /* error >= 0: No error */
194 return "(pcrs:) Everything's just fine. Thanks for asking.";
199 /*********************************************************************
201 * Function : pcrs_parse_perl_options
203 * Description : This function parses a string containing the options to
204 * Perl's s/// operator. It returns an integer that is the
205 * pcre equivalent of the symbolic optstring.
206 * Since pcre doesn't know about Perl's 'g' (global) or pcrs',
207 * 'T' (trivial) options but pcrs needs them, the corresponding
208 * flags are set if 'g'or 'T' is encountered.
209 * Note: The 'T' and 'U' options do not conform to Perl.
212 * 1 : optstring = string with options in perl syntax
213 * 2 : flags = see description
215 * Returns : option integer suitable for pcre
217 *********************************************************************/
218 static int pcrs_parse_perl_options(const char *optstring, int *flags)
224 if (NULL == optstring) return 0;
226 for (i = 0; i < strlen(optstring); i++)
230 case 'e': break; /* ToDo ;-) */
231 case 'g': *flags |= PCRS_GLOBAL; break;
232 case 'i': rc |= PCRE_CASELESS; break;
233 case 'm': rc |= PCRE_MULTILINE; break;
235 case 's': rc |= PCRE_DOTALL; break;
236 case 'x': rc |= PCRE_EXTENDED; break;
237 case 'U': rc |= PCRE_UNGREEDY; break;
238 case 'T': *flags |= PCRS_TRIVIAL; break;
247 /*********************************************************************
249 * Function : pcrs_compile_replacement
251 * Description : This function takes a Perl-style replacement (2nd argument
252 * to the s/// operator and returns a compiled pcrs_substitute,
253 * or NULL if memory allocation for the substitute structure
257 * 1 : replacement = replacement part of s/// operator
259 * 2 : trivialflag = Flag that causes backreferences to be
261 * 3 : capturecount = Number of capturing subpatterns in
262 * the pattern. Needed for $+ handling.
263 * 4 : errptr = pointer to an integer in which error
264 * conditions can be returned.
266 * Returns : pcrs_substitute data structure, or NULL if an
267 * error is encountered. In that case, *errptr has
270 *********************************************************************/
271 static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag, int capturecount, int *errptr)
278 i = k = l = quoted = 0;
283 if (NULL == replacement)
291 if (NULL == (r = (pcrs_substitute *)malloc(sizeof(pcrs_substitute))))
293 *errptr = PCRS_ERR_NOMEM;
296 memset(r, '\0', sizeof(pcrs_substitute));
298 length = strlen(replacement);
300 if (NULL == (text = (char *)malloc(length + 1)))
303 *errptr = PCRS_ERR_NOMEM;
306 memset(text, '\0', length + 1);
310 * In trivial mode, just copy the substitute text
314 text = strncpy(text, replacement, length + 1);
319 * Else, parse, cut out and record all backreferences
323 while (i < (int)length)
326 if (replacement[i] == '\\')
330 text[k++] = replacement[i++];
335 if (replacement[i+1] && strchr("tnrfae0", replacement[i+1]))
337 switch (replacement[++i])
373 if (replacement[i] == '$' && !quoted && i < (int)(length - 1))
375 char *symbol, symbols[] = "'`+&";
376 r->block_length[l] = k - r->block_offset[l];
378 /* Numerical backreferences */
379 if (isdigit((int)replacement[i + 1]))
381 while (i < (int)length && isdigit((int)replacement[++i]))
383 r->backref[l] = r->backref[l] * 10 + replacement[i] - 48;
385 if (r->backref[l] > capturecount)
387 *errptr = PCRS_WARN_BADREF;
391 /* Symbolic backreferences: */
392 else if (NULL != (symbol = strchr(symbols, replacement[i + 1])))
395 if (symbol - symbols == 2) /* $+ */
397 r->backref[l] = capturecount;
399 else if (symbol - symbols == 3) /* $& */
405 r->backref[l] = PCRS_MAX_SUBMATCHES + 1 - (symbol - symbols);
410 /* Invalid backref -> plain '$' */
416 /* Valid and in range? -> record */
417 if (r->backref[l] < PCRS_MAX_SUBMATCHES + 2)
419 r->backref_count[r->backref[l]] += 1;
420 r->block_offset[++l] = k;
424 *errptr = PCRS_WARN_BADREF;
430 /* Plain chars are copied */
431 text[k++] = replacement[i++];
434 } /* -END- if (!trivialflag) */
441 r->block_length[l] = k - r->block_offset[l];
448 /*********************************************************************
450 * Function : pcrs_free_job
452 * Description : Frees the memory used by a pcrs_job struct and its
453 * dependant structures.
456 * 1 : job = pointer to the pcrs_job structure to be freed
458 * Returns : a pointer to the next job, if there was any, or
461 *********************************************************************/
462 pcrs_job *pcrs_free_job(pcrs_job *job)
473 if (job->pattern != NULL) free(job->pattern);
474 if (job->hints != NULL) free(job->hints);
475 if (job->substitute != NULL)
477 if (job->substitute->text != NULL) free(job->substitute->text);
478 free(job->substitute);
487 /*********************************************************************
489 * Function : pcrs_free_joblist
491 * Description : Iterates through a chained list of pcrs_job's and
492 * frees them using pcrs_free_job.
495 * 1 : joblist = pointer to the first pcrs_job structure to
500 *********************************************************************/
501 void pcrs_free_joblist(pcrs_job *joblist)
503 while ( NULL != (joblist = pcrs_free_job(joblist)) ) {};
510 /*********************************************************************
512 * Function : pcrs_compile_command
514 * Description : Parses a string with a Perl-style s/// command,
515 * calls pcrs_compile, and returns a corresponding
516 * pcrs_job, or NULL if parsing or compiling the job
520 * 1 : command = string with perl-style s/// command
521 * 2 : errptr = pointer to an integer in which error
522 * conditions can be returned.
524 * Returns : a corresponding pcrs_job data structure, or NULL
525 * if an error was encountered. In that case, *errptr
528 *********************************************************************/
529 pcrs_job *pcrs_compile_command(const char *command, int *errptr)
531 int i, k, l, quoted = FALSE;
540 * Tokenize the perl command
542 limit = strlen(command);
545 *errptr = PCRS_ERR_CMDSYNTAX;
550 delimiter = command[1];
553 tokens[l] = (char *) malloc(limit + 1);
555 for (i = 0; i <= (int)limit; i++)
558 if (command[i] == delimiter && !quoted)
565 tokens[0][k++] = '\0';
566 tokens[++l] = tokens[0] + k;
570 else if (command[i] == '\\' && !quoted)
573 if (command[i+1] == delimiter) continue;
579 tokens[0][k++] = command[i];
587 *errptr = PCRS_ERR_CMDSYNTAX;
592 newjob = pcrs_compile(tokens[1], tokens[2], tokens[3], errptr);
599 /*********************************************************************
601 * Function : pcrs_compile
603 * Description : Takes the three arguments to a perl s/// command
604 * and compiles a pcrs_job structure from them.
607 * 1 : pattern = string with perl-style pattern
608 * 2 : substitute = string with perl-style substitute
609 * 3 : options = string with perl-style options
610 * 4 : errptr = pointer to an integer in which error
611 * conditions can be returned.
613 * Returns : a corresponding pcrs_job data structure, or NULL
614 * if an error was encountered. In that case, *errptr
617 *********************************************************************/
618 pcrs_job *pcrs_compile(const char *pattern, const char *substitute, const char *options, int *errptr)
628 * Handle NULL arguments
630 if (pattern == NULL) pattern = "";
631 if (substitute == NULL) substitute = "";
635 * Get and init memory
637 if (NULL == (newjob = (pcrs_job *)malloc(sizeof(pcrs_job))))
639 *errptr = PCRS_ERR_NOMEM;
642 memset(newjob, '\0', sizeof(pcrs_job));
646 * Evaluate the options
648 newjob->options = pcrs_parse_perl_options(options, &flags);
649 newjob->flags = flags;
653 * Compile the pattern
655 newjob->pattern = pcre_compile(pattern, newjob->options, &error, errptr, NULL);
656 if (newjob->pattern == NULL)
658 pcrs_free_job(newjob);
664 * Generate hints. This has little overhead, since the
665 * hints will be NULL for a boring pattern anyway.
667 newjob->hints = pcre_study(newjob->pattern, 0, &error);
670 *errptr = PCRS_ERR_STUDY;
671 pcrs_free_job(newjob);
677 * Determine the number of capturing subpatterns.
678 * This is needed for handling $+ in the substitute.
680 if (0 > (*errptr = pcre_fullinfo(newjob->pattern, newjob->hints, PCRE_INFO_CAPTURECOUNT, &capturecount)))
682 pcrs_free_job(newjob);
688 * Compile the substitute
690 if (NULL == (newjob->substitute = pcrs_compile_replacement(substitute, newjob->flags & PCRS_TRIVIAL, capturecount, errptr)))
692 pcrs_free_job(newjob);
701 /*********************************************************************
703 * Function : pcrs_execute_list
705 * Description : This is a multiple job wrapper for pcrs_execute().
706 * Apply the regular substitutions defined by the jobs in
707 * the joblist to the subject.
708 * The subject itself is left untouched, memory for the result
709 * is malloc()ed and it is the caller's responsibility to free
710 * the result when it's no longer needed.
713 * 1 : joblist = the chained list of pcrs_jobs to be executed
714 * 2 : subject = the subject string
715 * 3 : subject_length = the subject's length
716 * INCLUDING the terminating zero, if string!
717 * 4 : result = char** for returning the result
718 * 5 : result_length = size_t* for returning the result's length
720 * Returns : On success, the number of substitutions that were made.
721 * May be > 1 if job->flags contained PCRS_GLOBAL
722 * On failure, the (negative) pcre error code describing the
723 * failure, which may be translated to text using pcrs_strerror().
725 *********************************************************************/
726 int pcrs_execute_list(pcrs_job *joblist, char *subject, size_t subject_length, char **result, size_t *result_length)
730 int hits, total_hits;
733 *result_length = subject_length;
734 hits = total_hits = 0;
736 for (job = joblist; job != NULL; job = job->next)
738 hits = pcrs_execute(job, old, *result_length, &new, result_length);
740 if (old != subject) free(old);
759 /*********************************************************************
761 * Function : pcrs_execute
763 * Description : Apply the regular substitution defined by the job to the
765 * The subject itself is left untouched, memory for the result
766 * is malloc()ed and it is the caller's responsibility to free
767 * the result when it's no longer needed.
770 * 1 : job = the pcrs_job to be executed
771 * 2 : subject = the subject (== original) string
772 * 3 : subject_length = the subject's length
773 * INCLUDING the terminating zero, if string!
774 * 4 : result = char** for returning the result
775 * 5 : result_length = size_t* for returning the result's length
777 * Returns : On success, the number of substitutions that were made.
778 * May be > 1 if job->flags contained PCRS_GLOBAL
779 * On failure, the (negative) pcre error code describing the
780 * failure, which may be translated to text using pcrs_strerror().
782 *********************************************************************/
783 int pcrs_execute(pcrs_job *job, char *subject, size_t subject_length, char **result, size_t *result_length)
785 int offsets[3 * PCRS_MAX_SUBMATCHES],
790 max_matches = PCRS_MAX_MATCH_INIT;
792 pcrs_match *matches, *dummy;
798 * Sanity check & memory allocation
800 if (job == NULL || job->pattern == NULL || job->substitute == NULL)
803 return(PCRS_ERR_BADJOB);
806 if (NULL == (matches = (pcrs_match *)malloc(max_matches * sizeof(pcrs_match))))
809 return(PCRS_ERR_NOMEM);
811 memset(matches, '\0', max_matches * sizeof(pcrs_match));
815 * Find the pattern and calculate the space
816 * requirements for the result
818 newsize = subject_length;
820 while ((submatches = pcre_exec(job->pattern, job->hints, subject, (int)subject_length, offset, 0, offsets, 3 * PCRS_MAX_SUBMATCHES)) > 0)
822 job->flags |= PCRS_SUCCESS;
823 matches[i].submatches = submatches;
825 for (k = 0; k < submatches; k++)
827 matches[i].submatch_offset[k] = offsets[2 * k];
829 /* Note: Non-found optional submatches have length -1-(-1)==0 */
830 matches[i].submatch_length[k] = offsets[2 * k + 1] - offsets[2 * k];
832 /* reserve mem for each submatch as often as it is ref'd */
833 newsize += matches[i].submatch_length[k] * job->substitute->backref_count[k];
835 /* plus replacement text size minus match text size */
836 newsize += strlen(job->substitute->text) - matches[i].submatch_length[0];
838 /* chunk before match */
839 matches[i].submatch_offset[PCRS_MAX_SUBMATCHES] = 0;
840 matches[i].submatch_length[PCRS_MAX_SUBMATCHES] = offsets[0];
841 newsize += offsets[0] * job->substitute->backref_count[PCRS_MAX_SUBMATCHES];
843 /* chunk after match */
844 matches[i].submatch_offset[PCRS_MAX_SUBMATCHES + 1] = offsets[1];
845 matches[i].submatch_length[PCRS_MAX_SUBMATCHES + 1] = subject_length - offsets[1] - 1;
846 newsize += (subject_length - offsets[1]) * job->substitute->backref_count[PCRS_MAX_SUBMATCHES + 1];
848 /* Storage for matches exhausted? -> Extend! */
849 if (++i >= max_matches)
851 max_matches = (int)(max_matches * PCRS_MAX_MATCH_GROW);
852 if (NULL == (dummy = (pcrs_match *)realloc(matches, max_matches * sizeof(pcrs_match))))
856 return(PCRS_ERR_NOMEM);
861 /* Non-global search or limit reached? */
862 if (!(job->flags & PCRS_GLOBAL)) break;
864 /* Don't loop on empty matches */
865 if (offsets[1] == offset)
866 if ((size_t)offset < subject_length)
870 /* Go find the next one */
874 /* Pass pcre error through if (bad) failure */
875 if (submatches < PCRE_ERROR_NOMATCH)
884 * Get memory for the result
886 if ((*result = (char *)malloc(newsize)) == NULL) /* must be free()d by caller */
889 return PCRS_ERR_NOMEM;
897 result_offset = *result;
899 for (i = 0; i < matches_found; i++)
901 /* copy the chunk preceding the match */
902 memcpy(result_offset, subject + offset, (size_t)matches[i].submatch_offset[0] - offset);
903 result_offset += matches[i].submatch_offset[0] - offset;
905 /* For every segment of the substitute.. */
906 for (k = 0; k <= job->substitute->backrefs; k++)
908 /* ...copy its text.. */
909 memcpy(result_offset, job->substitute->text + job->substitute->block_offset[k], job->substitute->block_length[k]);
910 result_offset += job->substitute->block_length[k];
912 /* ..plus, if it's not the last chunk, i.e.: There *is* a backref.. */
913 if (k != job->substitute->backrefs
914 /* ..in legal range.. */
915 && job->substitute->backref[k] < PCRS_MAX_SUBMATCHES + 2
916 /* ..and referencing a nonempty match.. */
917 && matches[i].submatch_length[job->substitute->backref[k]] > 0)
919 /* ..copy the submatch that is ref'd. */
922 subject + matches[i].submatch_offset[job->substitute->backref[k]],
923 matches[i].submatch_length[job->substitute->backref[k]]
925 result_offset += matches[i].submatch_length[job->substitute->backref[k]];
928 offset = matches[i].submatch_offset[0] + matches[i].submatch_length[0];
932 memcpy(result_offset, subject + offset, subject_length - offset);
934 *result_length = newsize;
936 return matches_found;