1 /* GRegex -- regular expression API wrapper around PCRE.
3 * Copyright (C) 1999, 2000 Scott Wimer
4 * Copyright (C) 2004, Matthias Clasen <mclasen@redhat.com>
5 * Copyright (C) 2005 - 2007, Marco Barisione <marco@barisione.org>
7 * This library is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * This library is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with this library; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
30 #ifdef USE_SYSTEM_PCRE
33 #include "pcre/pcre.h"
36 /* PCRE 7.3 does not contain the definition of PCRE_ERROR_NULLWSLIMIT */
37 #ifndef PCRE_ERROR_NULLWSLIMIT
38 #define PCRE_ERROR_NULLWSLIMIT (-22)
43 /* Mask of all the possible values for GRegexCompileFlags. */
44 #define G_REGEX_COMPILE_MASK (G_REGEX_CASELESS | \
49 G_REGEX_DOLLAR_ENDONLY | \
52 G_REGEX_NO_AUTO_CAPTURE | \
55 G_REGEX_NEWLINE_CR | \
56 G_REGEX_NEWLINE_LF | \
59 /* Mask of all the possible values for GRegexMatchFlags. */
60 #define G_REGEX_MATCH_MASK (G_REGEX_MATCH_ANCHORED | \
61 G_REGEX_MATCH_NOTBOL | \
62 G_REGEX_MATCH_NOTEOL | \
63 G_REGEX_MATCH_NOTEMPTY | \
64 G_REGEX_MATCH_PARTIAL | \
65 G_REGEX_MATCH_NEWLINE_CR | \
66 G_REGEX_MATCH_NEWLINE_LF | \
67 G_REGEX_MATCH_NEWLINE_CRLF | \
68 G_REGEX_MATCH_NEWLINE_ANY)
70 /* if the string is in UTF-8 use g_utf8_ functions, else use
72 #define NEXT_CHAR(re, s) (((re)->compile_opts & PCRE_UTF8) ? \
73 g_utf8_next_char (s) : \
75 #define PREV_CHAR(re, s) (((re)->compile_opts & PCRE_UTF8) ? \
76 g_utf8_prev_char (s) : \
81 GRegex *regex; /* the regex */
82 GRegexMatchFlags match_opts; /* options used at match time on the regex */
83 gint matches; /* number of matching sub patterns */
84 gint pos; /* position in the string where last match left off */
85 gint *offsets; /* array of offsets paired 0,1 ; 2,3 ; 3,4 etc */
86 gint n_offsets; /* number of offsets */
87 gint *workspace; /* workspace for pcre_dfa_exec() */
88 gint n_workspace; /* number of workspace elements */
89 const gchar *string; /* string passed to the match function */
90 gssize string_len; /* length of string */
95 volatile gint ref_count; /* the ref count for the immutable part */
96 gchar *pattern; /* the pattern */
97 pcre *pcre_re; /* compiled form of the pattern */
98 GRegexCompileFlags compile_opts; /* options used at compile time on the pattern */
99 GRegexMatchFlags match_opts; /* options used at match time on the regex */
100 pcre_extra *extra; /* data stored when G_REGEX_OPTIMIZE is used */
103 /* TRUE if ret is an error code, FALSE otherwise. */
104 #define IS_PCRE_ERROR(ret) ((ret) < PCRE_ERROR_NOMATCH && (ret) != PCRE_ERROR_PARTIAL)
106 typedef struct _InterpolationData InterpolationData;
107 static gboolean interpolation_list_needs_match (GList *list);
108 static gboolean interpolate_replacement (const GMatchInfo *match_info,
111 static GList *split_replacement (const gchar *replacement,
113 static void free_interpolation_data (InterpolationData *data);
117 match_error (gint errcode)
121 case PCRE_ERROR_NOMATCH:
124 case PCRE_ERROR_NULL:
125 /* NULL argument, this should not happen in GRegex */
126 g_warning ("A NULL argument was passed to PCRE");
128 case PCRE_ERROR_BADOPTION:
129 return "bad options";
130 case PCRE_ERROR_BADMAGIC:
131 return _("corrupted object");
132 case PCRE_ERROR_UNKNOWN_OPCODE:
133 return N_("internal error or corrupted object");
134 case PCRE_ERROR_NOMEMORY:
135 return _("out of memory");
136 case PCRE_ERROR_NOSUBSTRING:
137 /* not used by pcre_exec() */
139 case PCRE_ERROR_MATCHLIMIT:
140 return _("backtracking limit reached");
141 case PCRE_ERROR_CALLOUT:
142 /* callouts are not implemented */
144 case PCRE_ERROR_BADUTF8:
145 case PCRE_ERROR_BADUTF8_OFFSET:
146 /* we do not check if strings are valid */
148 case PCRE_ERROR_PARTIAL:
151 case PCRE_ERROR_BADPARTIAL:
152 return _("the pattern contains items not supported for partial matching");
153 case PCRE_ERROR_INTERNAL:
154 return _("internal error");
155 case PCRE_ERROR_BADCOUNT:
156 /* negative ovecsize, this should not happen in GRegex */
157 g_warning ("A negative ovecsize was passed to PCRE");
159 case PCRE_ERROR_DFA_UITEM:
160 return _("the pattern contains items not supported for partial matching");
161 case PCRE_ERROR_DFA_UCOND:
162 return _("back references as conditions are not supported for partial matching");
163 case PCRE_ERROR_DFA_UMLIMIT:
164 /* the match_field field is not used in GRegex */
166 case PCRE_ERROR_DFA_WSSIZE:
167 /* handled expanding the workspace */
169 case PCRE_ERROR_DFA_RECURSE:
170 case PCRE_ERROR_RECURSIONLIMIT:
171 return _("recursion limit reached");
172 case PCRE_ERROR_NULLWSLIMIT:
173 return _("workspace limit for empty substrings reached");
174 case PCRE_ERROR_BADNEWLINE:
175 return _("invalid combination of newline flags");
179 return _("unknown error");
183 translate_compile_error (gint *errcode, const gchar **errmsg)
185 /* Compile errors are created adding 100 to the error code returned
187 * If errcode is known we put the translatable error message in
188 * erromsg. If errcode is unknown we put the generic
189 * G_REGEX_ERROR_COMPILE error code in errcode and keep the
190 * untranslated error message returned by PCRE.
191 * Note that there can be more PCRE errors with the same GRegexError
192 * and that some PCRE errors are useless for us.
198 case G_REGEX_ERROR_STRAY_BACKSLASH:
199 *errmsg = _("\\ at end of pattern");
201 case G_REGEX_ERROR_MISSING_CONTROL_CHAR:
202 *errmsg = _("\\c at end of pattern");
204 case G_REGEX_ERROR_UNRECOGNIZED_ESCAPE:
205 *errmsg = _("unrecognized character follows \\");
208 /* A number of Perl escapes are not handled by PCRE.
209 * Therefore it explicitly raises ERR37.
211 *errcode = G_REGEX_ERROR_UNRECOGNIZED_ESCAPE;
212 *errmsg = _("case-changing escapes (\\l, \\L, \\u, \\U) are not allowed here");
214 case G_REGEX_ERROR_QUANTIFIERS_OUT_OF_ORDER:
215 *errmsg = _("numbers out of order in {} quantifier");
217 case G_REGEX_ERROR_QUANTIFIER_TOO_BIG:
218 *errmsg = _("number too big in {} quantifier");
220 case G_REGEX_ERROR_UNTERMINATED_CHARACTER_CLASS:
221 *errmsg = _("missing terminating ] for character class");
223 case G_REGEX_ERROR_INVALID_ESCAPE_IN_CHARACTER_CLASS:
224 *errmsg = _("invalid escape sequence in character class");
226 case G_REGEX_ERROR_RANGE_OUT_OF_ORDER:
227 *errmsg = _("range out of order in character class");
229 case G_REGEX_ERROR_NOTHING_TO_REPEAT:
230 *errmsg = _("nothing to repeat");
232 case G_REGEX_ERROR_UNRECOGNIZED_CHARACTER:
233 *errmsg = _("unrecognized character after (?");
236 *errcode = G_REGEX_ERROR_UNRECOGNIZED_CHARACTER;
237 *errmsg = _("unrecognized character after (?<");
240 *errcode = G_REGEX_ERROR_UNRECOGNIZED_CHARACTER;
241 *errmsg = _("unrecognized character after (?P");
243 case G_REGEX_ERROR_POSIX_NAMED_CLASS_OUTSIDE_CLASS:
244 *errmsg = _("POSIX named classes are supported only within a class");
246 case G_REGEX_ERROR_UNMATCHED_PARENTHESIS:
247 *errmsg = _("missing terminating )");
250 *errcode = G_REGEX_ERROR_UNMATCHED_PARENTHESIS;
251 *errmsg = _(") without opening (");
254 *errcode = G_REGEX_ERROR_UNMATCHED_PARENTHESIS;
255 /* translators: '(?R' and '(?[+-]digits' are both meant as (groups of)
256 * sequences here, '(?-54' would be an example for the second group.
258 *errmsg = _("(?R or (?[+-]digits must be followed by )");
260 case G_REGEX_ERROR_INEXISTENT_SUBPATTERN_REFERENCE:
261 *errmsg = _("reference to non-existent subpattern");
263 case G_REGEX_ERROR_UNTERMINATED_COMMENT:
264 *errmsg = _("missing ) after comment");
266 case G_REGEX_ERROR_EXPRESSION_TOO_LARGE:
267 *errmsg = _("regular expression too large");
269 case G_REGEX_ERROR_MEMORY_ERROR:
270 *errmsg = _("failed to get memory");
272 case G_REGEX_ERROR_VARIABLE_LENGTH_LOOKBEHIND:
273 *errmsg = _("lookbehind assertion is not fixed length");
275 case G_REGEX_ERROR_MALFORMED_CONDITION:
276 *errmsg = _("malformed number or name after (?(");
278 case G_REGEX_ERROR_TOO_MANY_CONDITIONAL_BRANCHES:
279 *errmsg = _("conditional group contains more than two branches");
281 case G_REGEX_ERROR_ASSERTION_EXPECTED:
282 *errmsg = _("assertion expected after (?(");
284 case G_REGEX_ERROR_UNKNOWN_POSIX_CLASS_NAME:
285 *errmsg = _("unknown POSIX class name");
287 case G_REGEX_ERROR_POSIX_COLLATING_ELEMENTS_NOT_SUPPORTED:
288 *errmsg = _("POSIX collating elements are not supported");
290 case G_REGEX_ERROR_HEX_CODE_TOO_LARGE:
291 *errmsg = _("character value in \\x{...} sequence is too large");
293 case G_REGEX_ERROR_INVALID_CONDITION:
294 *errmsg = _("invalid condition (?(0)");
296 case G_REGEX_ERROR_SINGLE_BYTE_MATCH_IN_LOOKBEHIND:
297 *errmsg = _("\\C not allowed in lookbehind assertion");
299 case G_REGEX_ERROR_INFINITE_LOOP:
300 *errmsg = _("recursive call could loop indefinitely");
302 case G_REGEX_ERROR_MISSING_SUBPATTERN_NAME_TERMINATOR:
303 *errmsg = _("missing terminator in subpattern name");
305 case G_REGEX_ERROR_DUPLICATE_SUBPATTERN_NAME:
306 *errmsg = _("two named subpatterns have the same name");
308 case G_REGEX_ERROR_MALFORMED_PROPERTY:
309 *errmsg = _("malformed \\P or \\p sequence");
311 case G_REGEX_ERROR_UNKNOWN_PROPERTY:
312 *errmsg = _("unknown property name after \\P or \\p");
314 case G_REGEX_ERROR_SUBPATTERN_NAME_TOO_LONG:
315 *errmsg = _("subpattern name is too long (maximum 32 characters)");
317 case G_REGEX_ERROR_TOO_MANY_SUBPATTERNS:
318 *errmsg = _("too many named subpatterns (maximum 10,000)");
320 case G_REGEX_ERROR_INVALID_OCTAL_VALUE:
321 *errmsg = _("octal value is greater than \\377");
323 case G_REGEX_ERROR_TOO_MANY_BRANCHES_IN_DEFINE:
324 *errmsg = _("DEFINE group contains more than one branch");
326 case G_REGEX_ERROR_DEFINE_REPETION:
327 *errmsg = _("repeating a DEFINE group is not allowed");
329 case G_REGEX_ERROR_INCONSISTENT_NEWLINE_OPTIONS:
330 *errmsg = _("inconsistent NEWLINE options");
332 case G_REGEX_ERROR_MISSING_BACK_REFERENCE:
333 *errmsg = _("\\g is not followed by a braced name or an optionally "
334 "braced non-zero number");
337 *errcode = G_REGEX_ERROR_INTERNAL;
338 *errmsg = _("unexpected repeat");
341 *errcode = G_REGEX_ERROR_INTERNAL;
342 *errmsg = _("code overflow");
345 *errcode = G_REGEX_ERROR_INTERNAL;
346 *errmsg = _("overran compiling workspace");
349 *errcode = G_REGEX_ERROR_INTERNAL;
350 *errmsg = _("previously-checked referenced subpattern not found");
353 /* This should not happen as we never pass a NULL erroffset */
354 g_warning ("erroffset passed as NULL");
355 *errcode = G_REGEX_ERROR_COMPILE;
358 /* This should not happen as we check options before passing them
359 * to pcre_compile2() */
360 g_warning ("unknown option bit(s) set");
361 *errcode = G_REGEX_ERROR_COMPILE;
366 /* These errors should not happen as we are using an UTF8-enabled PCRE
367 * and we do not check if strings are valid */
368 g_warning ("%s", *errmsg);
369 *errcode = G_REGEX_ERROR_COMPILE;
372 *errcode = G_REGEX_ERROR_COMPILE;
379 match_info_new (const GRegex *regex,
386 GMatchInfo *match_info;
389 string_len = strlen (string);
391 match_info = g_new0 (GMatchInfo, 1);
392 match_info->regex = g_regex_ref ((GRegex *)regex);
393 match_info->string = string;
394 match_info->string_len = string_len;
395 match_info->matches = PCRE_ERROR_NOMATCH;
396 match_info->pos = start_position;
397 match_info->match_opts = match_options;
401 /* These values should be enough for most cases, if they are not
402 * enough g_regex_match_all_full() will expand them. */
403 match_info->n_offsets = 24;
404 match_info->n_workspace = 100;
405 match_info->workspace = g_new (gint, match_info->n_workspace);
410 pcre_fullinfo (regex->pcre_re, regex->extra,
411 PCRE_INFO_CAPTURECOUNT, &capture_count);
412 match_info->n_offsets = (capture_count + 1) * 3;
415 match_info->offsets = g_new0 (gint, match_info->n_offsets);
416 /* Set an invalid position for the previous match. */
417 match_info->offsets[0] = -1;
418 match_info->offsets[1] = -1;
424 * g_match_info_get_regex:
425 * @match_info: a #GMatchInfo
427 * Returns #GRegex object used in @match_info. It belongs to Glib
428 * and must not be freed. Use g_regex_ref() if you need to keep it
429 * after you free @match_info object.
431 * Returns: #GRegex object used in @match_info
436 g_match_info_get_regex (const GMatchInfo *match_info)
438 g_return_val_if_fail (match_info != NULL, NULL);
439 return match_info->regex;
443 * g_match_info_get_string:
444 * @match_info: a #GMatchInfo
446 * Returns the string searched with @match_info. This is the
447 * string passed to g_regex_match() or g_regex_replace() so
448 * you may not free it before calling this function.
450 * Returns: the string searched with @match_info
455 g_match_info_get_string (const GMatchInfo *match_info)
457 g_return_val_if_fail (match_info != NULL, NULL);
458 return match_info->string;
463 * @match_info: a #GMatchInfo
465 * Frees all the memory associated with the #GMatchInfo structure.
470 g_match_info_free (GMatchInfo *match_info)
474 g_regex_unref (match_info->regex);
475 g_free (match_info->offsets);
476 g_free (match_info->workspace);
483 * @match_info: a #GMatchInfo structure
484 * @error: location to store the error occuring, or %NULL to ignore errors
486 * Scans for the next match using the same parameters of the previous
487 * call to g_regex_match_full() or g_regex_match() that returned
490 * The match is done on the string passed to the match function, so you
491 * cannot free it before calling this function.
493 * Returns: %TRUE is the string matched, %FALSE otherwise
498 g_match_info_next (GMatchInfo *match_info,
502 gint prev_match_start;
505 g_return_val_if_fail (match_info != NULL, FALSE);
506 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
507 g_return_val_if_fail (match_info->pos >= 0, FALSE);
509 opts = match_info->regex->match_opts | match_info->match_opts;
511 prev_match_start = match_info->offsets[0];
512 prev_match_end = match_info->offsets[1];
514 match_info->matches = pcre_exec (match_info->regex->pcre_re,
515 match_info->regex->extra,
517 match_info->string_len,
519 match_info->regex->match_opts |
520 match_info->match_opts,
522 match_info->n_offsets);
523 if (IS_PCRE_ERROR (match_info->matches))
525 g_set_error (error, G_REGEX_ERROR, G_REGEX_ERROR_MATCH,
526 _("Error while matching regular expression %s: %s"),
527 match_info->regex->pattern, match_error (match_info->matches));
531 /* avoid infinite loops if the pattern is an empty string or something
533 if (match_info->pos == match_info->offsets[1])
535 if (match_info->pos > match_info->string_len)
537 /* we have reached the end of the string */
538 match_info->pos = -1;
539 match_info->matches = PCRE_ERROR_NOMATCH;
543 match_info->pos = NEXT_CHAR (match_info->regex,
544 &match_info->string[match_info->pos]) -
549 match_info->pos = match_info->offsets[1];
552 /* it's possibile to get two identical matches when we are matching
553 * empty strings, for instance if the pattern is "(?=[A-Z0-9])" and
554 * the string is "RegExTest" we have:
555 * - search at position 0: match from 0 to 0
556 * - search at position 1: match from 3 to 3
557 * - search at position 3: match from 3 to 3 (duplicate)
558 * - search at position 4: match from 5 to 5
559 * - search at position 5: match from 5 to 5 (duplicate)
560 * - search at position 6: no match -> stop
561 * so we have to ignore the duplicates.
562 * see bug #515944: http://bugzilla.gnome.org/show_bug.cgi?id=515944 */
563 if (match_info->matches >= 0 &&
564 prev_match_start == match_info->offsets[0] &&
565 prev_match_end == match_info->offsets[1])
567 /* ignore this match and search the next one */
568 return g_match_info_next (match_info, error);
571 return match_info->matches >= 0;
575 * g_match_info_matches:
576 * @match_info: a #GMatchInfo structure
578 * Returns whether the previous match operation succeeded.
580 * Returns: %TRUE if the previous match operation succeeded,
586 g_match_info_matches (const GMatchInfo *match_info)
588 g_return_val_if_fail (match_info != NULL, FALSE);
590 return match_info->matches >= 0;
594 * g_match_info_get_match_count:
595 * @match_info: a #GMatchInfo structure
597 * Retrieves the number of matched substrings (including substring 0,
598 * that is the whole matched text), so 1 is returned if the pattern
599 * has no substrings in it and 0 is returned if the match failed.
601 * If the last match was obtained using the DFA algorithm, that is
602 * using g_regex_match_all() or g_regex_match_all_full(), the retrieved
603 * count is not that of the number of capturing parentheses but that of
604 * the number of matched substrings.
606 * Returns: Number of matched substrings, or -1 if an error occurred
611 g_match_info_get_match_count (const GMatchInfo *match_info)
613 g_return_val_if_fail (match_info, -1);
615 if (match_info->matches == PCRE_ERROR_NOMATCH)
618 else if (match_info->matches < PCRE_ERROR_NOMATCH)
623 return match_info->matches;
627 * g_match_info_is_partial_match:
628 * @match_info: a #GMatchInfo structure
630 * Usually if the string passed to g_regex_match*() matches as far as
631 * it goes, but is too short to match the entire pattern, %FALSE is
632 * returned. There are circumstances where it might be helpful to
633 * distinguish this case from other cases in which there is no match.
635 * Consider, for example, an application where a human is required to
636 * type in data for a field with specific formatting requirements. An
637 * example might be a date in the form ddmmmyy, defined by the pattern
638 * "^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$".
639 * If the application sees the user’s keystrokes one by one, and can
640 * check that what has been typed so far is potentially valid, it is
641 * able to raise an error as soon as a mistake is made.
643 * GRegex supports the concept of partial matching by means of the
644 * #G_REGEX_MATCH_PARTIAL flag. When this is set the return code for
645 * g_regex_match() or g_regex_match_full() is, as usual, %TRUE
646 * for a complete match, %FALSE otherwise. But, when these functions
647 * return %FALSE, you can check if the match was partial calling
648 * g_match_info_is_partial_match().
650 * When using partial matching you cannot use g_match_info_fetch*().
652 * Because of the way certain internal optimizations are implemented
653 * the partial matching algorithm cannot be used with all patterns.
654 * So repeated single characters such as "a{2,4}" and repeated single
655 * meta-sequences such as "\d+" are not permitted if the maximum number
656 * of occurrences is greater than one. Optional items such as "\d?"
657 * (where the maximum is one) are permitted. Quantifiers with any values
658 * are permitted after parentheses, so the invalid examples above can be
659 * coded thus "(a){2,4}" and "(\d)+". If #G_REGEX_MATCH_PARTIAL is set
660 * for a pattern that does not conform to the restrictions, matching
661 * functions return an error.
663 * Returns: %TRUE if the match was partial, %FALSE otherwise
668 g_match_info_is_partial_match (const GMatchInfo *match_info)
670 g_return_val_if_fail (match_info != NULL, FALSE);
672 return match_info->matches == PCRE_ERROR_PARTIAL;
676 * g_match_info_expand_references:
677 * @match_info: a #GMatchInfo or %NULL
678 * @string_to_expand: the string to expand
679 * @error: location to store the error occuring, or %NULL to ignore errors
681 * Returns a new string containing the text in @string_to_expand with
682 * references and escape sequences expanded. References refer to the last
683 * match done with @string against @regex and have the same syntax used by
686 * The @string_to_expand must be UTF-8 encoded even if #G_REGEX_RAW was
687 * passed to g_regex_new().
689 * The backreferences are extracted from the string passed to the match
690 * function, so you cannot call this function after freeing the string.
692 * @match_info may be %NULL in which case @string_to_expand must not
693 * contain references. For instance "foo\n" does not refer to an actual
694 * pattern and '\n' merely will be replaced with \n character,
695 * while to expand "\0" (whole match) one needs the result of a match.
696 * Use g_regex_check_replacement() to find out whether @string_to_expand
697 * contains references.
699 * Returns: the expanded string, or %NULL if an error occurred
704 g_match_info_expand_references (const GMatchInfo *match_info,
705 const gchar *string_to_expand,
710 GError *tmp_error = NULL;
712 g_return_val_if_fail (string_to_expand != NULL, NULL);
713 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
715 list = split_replacement (string_to_expand, &tmp_error);
716 if (tmp_error != NULL)
718 g_propagate_error (error, tmp_error);
722 if (!match_info && interpolation_list_needs_match (list))
724 g_critical ("String '%s' contains references to the match, can't "
725 "expand references without GMatchInfo object",
730 result = g_string_sized_new (strlen (string_to_expand));
731 interpolate_replacement (match_info, result, list);
733 g_list_foreach (list, (GFunc)free_interpolation_data, NULL);
736 return g_string_free (result, FALSE);
740 * g_match_info_fetch:
741 * @match_info: #GMatchInfo structure
742 * @match_num: number of the sub expression
744 * Retrieves the text matching the @match_num<!-- -->'th capturing
745 * parentheses. 0 is the full text of the match, 1 is the first paren
746 * set, 2 the second, and so on.
748 * If @match_num is a valid sub pattern but it didn't match anything
749 * (e.g. sub pattern 1, matching "b" against "(a)?b") then an empty
750 * string is returned.
752 * If the match was obtained using the DFA algorithm, that is using
753 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
754 * string is not that of a set of parentheses but that of a matched
755 * substring. Substrings are matched in reverse order of length, so
756 * 0 is the longest match.
758 * The string is fetched from the string passed to the match function,
759 * so you cannot call this function after freeing the string.
761 * Returns: The matched substring, or %NULL if an error occurred.
762 * You have to free the string yourself
767 g_match_info_fetch (const GMatchInfo *match_info,
770 /* we cannot use pcre_get_substring() because it allocates the
771 * string using pcre_malloc(). */
775 g_return_val_if_fail (match_info != NULL, NULL);
776 g_return_val_if_fail (match_num >= 0, NULL);
778 /* match_num does not exist or it didn't matched, i.e. matching "b"
779 * against "(a)?b" then group 0 is empty. */
780 if (!g_match_info_fetch_pos (match_info, match_num, &start, &end))
782 else if (start == -1)
783 match = g_strdup ("");
785 match = g_strndup (&match_info->string[start], end - start);
791 * g_match_info_fetch_pos:
792 * @match_info: #GMatchInfo structure
793 * @match_num: number of the sub expression
794 * @start_pos: pointer to location where to store the start position
795 * @end_pos: pointer to location where to store the end position
797 * Retrieves the position of the @match_num<!-- -->'th capturing
798 * parentheses. 0 is the full text of the match, 1 is the first
799 * paren set, 2 the second, and so on.
801 * If @match_num is a valid sub pattern but it didn't match anything
802 * (e.g. sub pattern 1, matching "b" against "(a)?b") then @start_pos
803 * and @end_pos are set to -1 and %TRUE is returned.
805 * If the match was obtained using the DFA algorithm, that is using
806 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
807 * position is not that of a set of parentheses but that of a matched
808 * substring. Substrings are matched in reverse order of length, so
809 * 0 is the longest match.
811 * Returns: %TRUE if the position was fetched, %FALSE otherwise. If
812 * the position cannot be fetched, @start_pos and @end_pos are left
818 g_match_info_fetch_pos (const GMatchInfo *match_info,
823 g_return_val_if_fail (match_info != NULL, FALSE);
824 g_return_val_if_fail (match_num >= 0, FALSE);
826 /* make sure the sub expression number they're requesting is less than
827 * the total number of sub expressions that were matched. */
828 if (match_num >= match_info->matches)
831 if (start_pos != NULL)
832 *start_pos = match_info->offsets[2 * match_num];
835 *end_pos = match_info->offsets[2 * match_num + 1];
841 * Returns number of first matched subpattern with name @name.
842 * There may be more than one in case when DUPNAMES is used,
843 * and not all subpatterns with that name match;
844 * pcre_get_stringnumber() does not work in that case.
847 get_matched_substring_number (const GMatchInfo *match_info,
854 if (!(match_info->regex->compile_opts & G_REGEX_DUPNAMES))
855 return pcre_get_stringnumber (match_info->regex->pcre_re, name);
857 /* This code is copied from pcre_get.c: get_first_set() */
858 entrysize = pcre_get_stringtable_entries (match_info->regex->pcre_re,
866 for (entry = (guchar*) first; entry <= (guchar*) last; entry += entrysize)
868 gint n = (entry[0] << 8) + entry[1];
869 if (match_info->offsets[n*2] >= 0)
873 return (first[0] << 8) + first[1];
877 * g_match_info_fetch_named:
878 * @match_info: #GMatchInfo structure
879 * @name: name of the subexpression
881 * Retrieves the text matching the capturing parentheses named @name.
883 * If @name is a valid sub pattern name but it didn't match anything
884 * (e.g. sub pattern "X", matching "b" against "(?P<X>a)?b")
885 * then an empty string is returned.
887 * The string is fetched from the string passed to the match function,
888 * so you cannot call this function after freeing the string.
890 * Returns: The matched substring, or %NULL if an error occurred.
891 * You have to free the string yourself
896 g_match_info_fetch_named (const GMatchInfo *match_info,
899 /* we cannot use pcre_get_named_substring() because it allocates the
900 * string using pcre_malloc(). */
903 g_return_val_if_fail (match_info != NULL, NULL);
904 g_return_val_if_fail (name != NULL, NULL);
906 num = get_matched_substring_number (match_info, name);
910 return g_match_info_fetch (match_info, num);
914 * g_match_info_fetch_named_pos:
915 * @match_info: #GMatchInfo structure
916 * @name: name of the subexpression
917 * @start_pos: pointer to location where to store the start position
918 * @end_pos: pointer to location where to store the end position
920 * Retrieves the position of the capturing parentheses named @name.
922 * If @name is a valid sub pattern name but it didn't match anything
923 * (e.g. sub pattern "X", matching "b" against "(?P<X>a)?b")
924 * then @start_pos and @end_pos are set to -1 and %TRUE is returned.
926 * Returns: %TRUE if the position was fetched, %FALSE otherwise. If
927 * the position cannot be fetched, @start_pos and @end_pos are left
933 g_match_info_fetch_named_pos (const GMatchInfo *match_info,
940 g_return_val_if_fail (match_info != NULL, FALSE);
941 g_return_val_if_fail (name != NULL, FALSE);
943 num = get_matched_substring_number (match_info, name);
947 return g_match_info_fetch_pos (match_info, num, start_pos, end_pos);
951 * g_match_info_fetch_all:
952 * @match_info: a #GMatchInfo structure
954 * Bundles up pointers to each of the matching substrings from a match
955 * and stores them in an array of gchar pointers. The first element in
956 * the returned array is the match number 0, i.e. the entire matched
959 * If a sub pattern didn't match anything (e.g. sub pattern 1, matching
960 * "b" against "(a)?b") then an empty string is inserted.
962 * If the last match was obtained using the DFA algorithm, that is using
963 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
964 * strings are not that matched by sets of parentheses but that of the
965 * matched substring. Substrings are matched in reverse order of length,
966 * so the first one is the longest match.
968 * The strings are fetched from the string passed to the match function,
969 * so you cannot call this function after freeing the string.
971 * Returns: a %NULL-terminated array of gchar * pointers. It must be
972 * freed using g_strfreev(). If the previous match failed %NULL is
978 g_match_info_fetch_all (const GMatchInfo *match_info)
980 /* we cannot use pcre_get_substring_list() because the returned value
981 * isn't suitable for g_strfreev(). */
985 g_return_val_if_fail (match_info != NULL, NULL);
987 if (match_info->matches < 0)
990 result = g_new (gchar *, match_info->matches + 1);
991 for (i = 0; i < match_info->matches; i++)
992 result[i] = g_match_info_fetch (match_info, i);
1002 g_regex_error_quark (void)
1004 static GQuark error_quark = 0;
1006 if (error_quark == 0)
1007 error_quark = g_quark_from_static_string ("g-regex-error-quark");
1016 * Increases reference count of @regex by 1.
1023 g_regex_ref (GRegex *regex)
1025 g_return_val_if_fail (regex != NULL, NULL);
1026 g_atomic_int_inc (®ex->ref_count);
1034 * Decreases reference count of @regex by 1. When reference count drops
1035 * to zero, it frees all the memory associated with the regex structure.
1040 g_regex_unref (GRegex *regex)
1042 g_return_if_fail (regex != NULL);
1044 if (g_atomic_int_exchange_and_add (®ex->ref_count, -1) - 1 == 0)
1046 g_free (regex->pattern);
1047 if (regex->pcre_re != NULL)
1048 pcre_free (regex->pcre_re);
1049 if (regex->extra != NULL)
1050 pcre_free (regex->extra);
1057 * @pattern: the regular expression
1058 * @compile_options: compile options for the regular expression, or 0
1059 * @match_options: match options for the regular expression, or 0
1060 * @error: return location for a #GError
1062 * Compiles the regular expression to an internal form, and does
1063 * the initial setup of the #GRegex structure.
1065 * Returns: a #GRegex structure. Call g_regex_unref() when you
1071 g_regex_new (const gchar *pattern,
1072 GRegexCompileFlags compile_options,
1073 GRegexMatchFlags match_options,
1078 const gchar *errmsg;
1081 gboolean optimize = FALSE;
1082 static gboolean initialized = FALSE;
1083 unsigned long int pcre_compile_options;
1085 g_return_val_if_fail (pattern != NULL, NULL);
1086 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
1087 g_return_val_if_fail ((compile_options & ~G_REGEX_COMPILE_MASK) == 0, NULL);
1088 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
1095 pcre_config (PCRE_CONFIG_UTF8, &support);
1098 msg = N_("PCRE library is compiled without UTF8 support");
1099 g_critical ("%s", msg);
1100 g_set_error_literal (error, G_REGEX_ERROR, G_REGEX_ERROR_COMPILE, gettext (msg));
1104 pcre_config (PCRE_CONFIG_UNICODE_PROPERTIES, &support);
1107 msg = N_("PCRE library is compiled without UTF8 properties support");
1108 g_critical ("%s", msg);
1109 g_set_error_literal (error, G_REGEX_ERROR, G_REGEX_ERROR_COMPILE, gettext (msg));
1116 /* G_REGEX_OPTIMIZE has the same numeric value of PCRE_NO_UTF8_CHECK,
1117 * as we do not need to wrap PCRE_NO_UTF8_CHECK. */
1118 if (compile_options & G_REGEX_OPTIMIZE)
1121 /* In GRegex the string are, by default, UTF-8 encoded. PCRE
1122 * instead uses UTF-8 only if required with PCRE_UTF8. */
1123 if (compile_options & G_REGEX_RAW)
1126 compile_options &= ~G_REGEX_RAW;
1131 compile_options |= PCRE_UTF8 | PCRE_NO_UTF8_CHECK;
1132 match_options |= PCRE_NO_UTF8_CHECK;
1135 /* PCRE_NEWLINE_ANY is the default for the internal PCRE but
1136 * not for the system one. */
1137 if (!(compile_options & G_REGEX_NEWLINE_CR) &&
1138 !(compile_options & G_REGEX_NEWLINE_LF))
1140 compile_options |= PCRE_NEWLINE_ANY;
1143 /* compile the pattern */
1144 re = pcre_compile2 (pattern, compile_options, &errcode,
1145 &errmsg, &erroffset, NULL);
1147 /* if the compilation failed, set the error member and return
1153 /* Translate the PCRE error code to GRegexError and use a translated
1154 * error message if possible */
1155 translate_compile_error (&errcode, &errmsg);
1157 /* PCRE uses byte offsets but we want to show character offsets */
1158 erroffset = g_utf8_pointer_to_offset (pattern, &pattern[erroffset]);
1160 tmp_error = g_error_new (G_REGEX_ERROR, errcode,
1161 _("Error while compiling regular "
1162 "expression %s at char %d: %s"),
1163 pattern, erroffset, errmsg);
1164 g_propagate_error (error, tmp_error);
1169 /* For options set at the beginning of the pattern, pcre puts them into
1170 * compile options, e.g. "(?i)foo" will make the pcre structure store
1171 * PCRE_CASELESS even though it wasn't explicitly given for compilation. */
1172 pcre_fullinfo (re, NULL, PCRE_INFO_OPTIONS, &pcre_compile_options);
1173 compile_options = pcre_compile_options;
1175 if (!(compile_options & G_REGEX_DUPNAMES))
1177 gboolean jchanged = FALSE;
1178 pcre_fullinfo (re, NULL, PCRE_INFO_JCHANGED, &jchanged);
1180 compile_options |= G_REGEX_DUPNAMES;
1183 regex = g_new0 (GRegex, 1);
1184 regex->ref_count = 1;
1185 regex->pattern = g_strdup (pattern);
1186 regex->pcre_re = re;
1187 regex->compile_opts = compile_options;
1188 regex->match_opts = match_options;
1192 regex->extra = pcre_study (regex->pcre_re, 0, &errmsg);
1195 GError *tmp_error = g_error_new (G_REGEX_ERROR,
1196 G_REGEX_ERROR_OPTIMIZE,
1197 _("Error while optimizing "
1198 "regular expression %s: %s"),
1201 g_propagate_error (error, tmp_error);
1210 * g_regex_get_pattern:
1211 * @regex: a #GRegex structure
1213 * Gets the pattern string associated with @regex, i.e. a copy of
1214 * the string passed to g_regex_new().
1216 * Returns: the pattern of @regex
1221 g_regex_get_pattern (const GRegex *regex)
1223 g_return_val_if_fail (regex != NULL, NULL);
1225 return regex->pattern;
1229 * g_regex_get_max_backref:
1232 * Returns the number of the highest back reference
1233 * in the pattern, or 0 if the pattern does not contain
1236 * Returns: the number of the highest back reference
1241 g_regex_get_max_backref (const GRegex *regex)
1245 pcre_fullinfo (regex->pcre_re, regex->extra,
1246 PCRE_INFO_BACKREFMAX, &value);
1252 * g_regex_get_capture_count:
1255 * Returns the number of capturing subpatterns in the pattern.
1257 * Returns: the number of capturing subpatterns
1262 g_regex_get_capture_count (const GRegex *regex)
1266 pcre_fullinfo (regex->pcre_re, regex->extra,
1267 PCRE_INFO_CAPTURECOUNT, &value);
1273 * g_regex_match_simple:
1274 * @pattern: the regular expression
1275 * @string: the string to scan for matches
1276 * @compile_options: compile options for the regular expression, or 0
1277 * @match_options: match options, or 0
1279 * Scans for a match in @string for @pattern.
1281 * This function is equivalent to g_regex_match() but it does not
1282 * require to compile the pattern with g_regex_new(), avoiding some
1283 * lines of code when you need just to do a match without extracting
1284 * substrings, capture counts, and so on.
1286 * If this function is to be called on the same @pattern more than
1287 * once, it's more efficient to compile the pattern once with
1288 * g_regex_new() and then use g_regex_match().
1290 * Returns: %TRUE if the string matched, %FALSE otherwise
1295 g_regex_match_simple (const gchar *pattern,
1296 const gchar *string,
1297 GRegexCompileFlags compile_options,
1298 GRegexMatchFlags match_options)
1303 regex = g_regex_new (pattern, compile_options, 0, NULL);
1306 result = g_regex_match_full (regex, string, -1, 0, match_options, NULL, NULL);
1307 g_regex_unref (regex);
1313 * @regex: a #GRegex structure from g_regex_new()
1314 * @string: the string to scan for matches
1315 * @match_options: match options
1316 * @match_info: pointer to location where to store the #GMatchInfo,
1317 * or %NULL if you do not need it
1319 * Scans for a match in string for the pattern in @regex.
1320 * The @match_options are combined with the match options specified
1321 * when the @regex structure was created, letting you have more
1322 * flexibility in reusing #GRegex structures.
1324 * A #GMatchInfo structure, used to get information on the match,
1325 * is stored in @match_info if not %NULL. Note that if @match_info
1326 * is not %NULL then it is created even if the function returns %FALSE,
1327 * i.e. you must free it regardless if regular expression actually matched.
1329 * To retrieve all the non-overlapping matches of the pattern in
1330 * string you can use g_match_info_next().
1334 * print_uppercase_words (const gchar *string)
1336 * /* Print all uppercase-only words. */
1338 * GMatchInfo *match_info;
1340 * regex = g_regex_new ("[A-Z]+", 0, 0, NULL);
1341 * g_regex_match (regex, string, 0, &match_info);
1342 * while (g_match_info_matches (match_info))
1344 * gchar *word = g_match_info_fetch (match_info, 0);
1345 * g_print ("Found: %s\n", word);
1347 * g_match_info_next (match_info, NULL);
1349 * g_match_info_free (match_info);
1350 * g_regex_unref (regex);
1354 * Returns: %TRUE is the string matched, %FALSE otherwise
1359 g_regex_match (const GRegex *regex,
1360 const gchar *string,
1361 GRegexMatchFlags match_options,
1362 GMatchInfo **match_info)
1364 return g_regex_match_full (regex, string, -1, 0, match_options,
1369 * g_regex_match_full:
1370 * @regex: a #GRegex structure from g_regex_new()
1371 * @string: the string to scan for matches
1372 * @string_len: the length of @string, or -1 if @string is nul-terminated
1373 * @start_position: starting index of the string to match
1374 * @match_options: match options
1375 * @match_info: pointer to location where to store the #GMatchInfo,
1376 * or %NULL if you do not need it
1377 * @error: location to store the error occuring, or %NULL to ignore errors
1379 * Scans for a match in string for the pattern in @regex.
1380 * The @match_options are combined with the match options specified
1381 * when the @regex structure was created, letting you have more
1382 * flexibility in reusing #GRegex structures.
1384 * Setting @start_position differs from just passing over a shortened
1385 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
1386 * that begins with any kind of lookbehind assertion, such as "\b".
1388 * A #GMatchInfo structure, used to get information on the match, is
1389 * stored in @match_info if not %NULL. Note that if @match_info is
1390 * not %NULL then it is created even if the function returns %FALSE,
1391 * i.e. you must free it regardless if regular expression actually
1394 * @string is not copied and is used in #GMatchInfo internally. If
1395 * you use any #GMatchInfo method (except g_match_info_free()) after
1396 * freeing or modifying @string then the behaviour is undefined.
1398 * To retrieve all the non-overlapping matches of the pattern in
1399 * string you can use g_match_info_next().
1403 * print_uppercase_words (const gchar *string)
1405 * /* Print all uppercase-only words. */
1407 * GMatchInfo *match_info;
1408 * GError *error = NULL;
1410 * regex = g_regex_new ("[A-Z]+", 0, 0, NULL);
1411 * g_regex_match_full (regex, string, -1, 0, 0, &match_info, &error);
1412 * while (g_match_info_matches (match_info))
1414 * gchar *word = g_match_info_fetch (match_info, 0);
1415 * g_print ("Found: %s\n", word);
1417 * g_match_info_next (match_info, &error);
1419 * g_match_info_free (match_info);
1420 * g_regex_unref (regex);
1421 * if (error != NULL)
1423 * g_printerr ("Error while matching: %s\n", error->message);
1424 * g_error_free (error);
1429 * Returns: %TRUE is the string matched, %FALSE otherwise
1434 g_regex_match_full (const GRegex *regex,
1435 const gchar *string,
1437 gint start_position,
1438 GRegexMatchFlags match_options,
1439 GMatchInfo **match_info,
1445 g_return_val_if_fail (regex != NULL, FALSE);
1446 g_return_val_if_fail (string != NULL, FALSE);
1447 g_return_val_if_fail (start_position >= 0, FALSE);
1448 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
1449 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, FALSE);
1451 info = match_info_new (regex, string, string_len, start_position,
1452 match_options, FALSE);
1453 match_ok = g_match_info_next (info, error);
1454 if (match_info != NULL)
1457 g_match_info_free (info);
1463 * g_regex_match_all:
1464 * @regex: a #GRegex structure from g_regex_new()
1465 * @string: the string to scan for matches
1466 * @match_options: match options
1467 * @match_info: pointer to location where to store the #GMatchInfo,
1468 * or %NULL if you do not need it
1470 * Using the standard algorithm for regular expression matching only
1471 * the longest match in the string is retrieved. This function uses
1472 * a different algorithm so it can retrieve all the possible matches.
1473 * For more documentation see g_regex_match_all_full().
1475 * A #GMatchInfo structure, used to get information on the match, is
1476 * stored in @match_info if not %NULL. Note that if @match_info is
1477 * not %NULL then it is created even if the function returns %FALSE,
1478 * i.e. you must free it regardless if regular expression actually
1481 * Returns: %TRUE is the string matched, %FALSE otherwise
1486 g_regex_match_all (const GRegex *regex,
1487 const gchar *string,
1488 GRegexMatchFlags match_options,
1489 GMatchInfo **match_info)
1491 return g_regex_match_all_full (regex, string, -1, 0, match_options,
1496 * g_regex_match_all_full:
1497 * @regex: a #GRegex structure from g_regex_new()
1498 * @string: the string to scan for matches
1499 * @string_len: the length of @string, or -1 if @string is nul-terminated
1500 * @start_position: starting index of the string to match
1501 * @match_options: match options
1502 * @match_info: pointer to location where to store the #GMatchInfo,
1503 * or %NULL if you do not need it
1504 * @error: location to store the error occuring, or %NULL to ignore errors
1506 * Using the standard algorithm for regular expression matching only
1507 * the longest match in the string is retrieved, it is not possibile
1508 * to obtain all the available matches. For instance matching
1509 * "<a> <b> <c>" against the pattern "<.*>"
1510 * you get "<a> <b> <c>".
1512 * This function uses a different algorithm (called DFA, i.e. deterministic
1513 * finite automaton), so it can retrieve all the possible matches, all
1514 * starting at the same point in the string. For instance matching
1515 * "<a> <b> <c>" against the pattern "<.*>"
1516 * you would obtain three matches: "<a> <b> <c>",
1517 * "<a> <b>" and "<a>".
1519 * The number of matched strings is retrieved using
1520 * g_match_info_get_match_count(). To obtain the matched strings and
1521 * their position you can use, respectively, g_match_info_fetch() and
1522 * g_match_info_fetch_pos(). Note that the strings are returned in
1523 * reverse order of length; that is, the longest matching string is
1526 * Note that the DFA algorithm is slower than the standard one and it
1527 * is not able to capture substrings, so backreferences do not work.
1529 * Setting @start_position differs from just passing over a shortened
1530 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
1531 * that begins with any kind of lookbehind assertion, such as "\b".
1533 * A #GMatchInfo structure, used to get information on the match, is
1534 * stored in @match_info if not %NULL. Note that if @match_info is
1535 * not %NULL then it is created even if the function returns %FALSE,
1536 * i.e. you must free it regardless if regular expression actually
1539 * Returns: %TRUE is the string matched, %FALSE otherwise
1544 g_regex_match_all_full (const GRegex *regex,
1545 const gchar *string,
1547 gint start_position,
1548 GRegexMatchFlags match_options,
1549 GMatchInfo **match_info,
1555 g_return_val_if_fail (regex != NULL, FALSE);
1556 g_return_val_if_fail (string != NULL, FALSE);
1557 g_return_val_if_fail (start_position >= 0, FALSE);
1558 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
1559 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, FALSE);
1561 info = match_info_new (regex, string, string_len, start_position,
1562 match_options, TRUE);
1568 info->matches = pcre_dfa_exec (regex->pcre_re, regex->extra,
1569 info->string, info->string_len,
1571 regex->match_opts | match_options,
1572 info->offsets, info->n_offsets,
1573 info->workspace, info->n_workspace);
1574 if (info->matches == PCRE_ERROR_DFA_WSSIZE)
1576 /* info->workspace is too small. */
1577 info->n_workspace *= 2;
1578 info->workspace = g_realloc (info->workspace,
1579 info->n_workspace * sizeof (gint));
1582 else if (info->matches == 0)
1584 /* info->offsets is too small. */
1585 info->n_offsets *= 2;
1586 info->offsets = g_realloc (info->offsets,
1587 info->n_offsets * sizeof (gint));
1590 else if (IS_PCRE_ERROR (info->matches))
1592 g_set_error (error, G_REGEX_ERROR, G_REGEX_ERROR_MATCH,
1593 _("Error while matching regular expression %s: %s"),
1594 regex->pattern, match_error (info->matches));
1598 /* set info->pos to -1 so that a call to g_match_info_next() fails. */
1601 if (match_info != NULL)
1604 g_match_info_free (info);
1606 return info->matches >= 0;
1610 * g_regex_get_string_number:
1611 * @regex: #GRegex structure
1612 * @name: name of the subexpression
1614 * Retrieves the number of the subexpression named @name.
1616 * Returns: The number of the subexpression or -1 if @name
1622 g_regex_get_string_number (const GRegex *regex,
1627 g_return_val_if_fail (regex != NULL, -1);
1628 g_return_val_if_fail (name != NULL, -1);
1630 num = pcre_get_stringnumber (regex->pcre_re, name);
1631 if (num == PCRE_ERROR_NOSUBSTRING)
1638 * g_regex_split_simple:
1639 * @pattern: the regular expression
1640 * @string: the string to scan for matches
1641 * @compile_options: compile options for the regular expression, or 0
1642 * @match_options: match options, or 0
1644 * Breaks the string on the pattern, and returns an array of
1645 * the tokens. If the pattern contains capturing parentheses,
1646 * then the text for each of the substrings will also be returned.
1647 * If the pattern does not match anywhere in the string, then the
1648 * whole string is returned as the first token.
1650 * This function is equivalent to g_regex_split() but it does
1651 * not require to compile the pattern with g_regex_new(), avoiding
1652 * some lines of code when you need just to do a split without
1653 * extracting substrings, capture counts, and so on.
1655 * If this function is to be called on the same @pattern more than
1656 * once, it's more efficient to compile the pattern once with
1657 * g_regex_new() and then use g_regex_split().
1659 * As a special case, the result of splitting the empty string ""
1660 * is an empty vector, not a vector containing a single string.
1661 * The reason for this special case is that being able to represent
1662 * a empty vector is typically more useful than consistent handling
1663 * of empty elements. If you do need to represent empty elements,
1664 * you'll need to check for the empty string before calling this
1667 * A pattern that can match empty strings splits @string into
1668 * separate characters wherever it matches the empty string between
1669 * characters. For example splitting "ab c" using as a separator
1670 * "\s*", you will get "a", "b" and "c".
1672 * Returns: a %NULL-terminated array of strings. Free it using g_strfreev()
1677 g_regex_split_simple (const gchar *pattern,
1678 const gchar *string,
1679 GRegexCompileFlags compile_options,
1680 GRegexMatchFlags match_options)
1685 regex = g_regex_new (pattern, compile_options, 0, NULL);
1688 result = g_regex_split_full (regex, string, -1, 0, match_options, 0, NULL);
1689 g_regex_unref (regex);
1695 * @regex: a #GRegex structure
1696 * @string: the string to split with the pattern
1697 * @match_options: match time option flags
1699 * Breaks the string on the pattern, and returns an array of the tokens.
1700 * If the pattern contains capturing parentheses, then the text for each
1701 * of the substrings will also be returned. If the pattern does not match
1702 * anywhere in the string, then the whole string is returned as the first
1705 * As a special case, the result of splitting the empty string "" is an
1706 * empty vector, not a vector containing a single string. The reason for
1707 * this special case is that being able to represent a empty vector is
1708 * typically more useful than consistent handling of empty elements. If
1709 * you do need to represent empty elements, you'll need to check for the
1710 * empty string before calling this function.
1712 * A pattern that can match empty strings splits @string into separate
1713 * characters wherever it matches the empty string between characters.
1714 * For example splitting "ab c" using as a separator "\s*", you will get
1717 * Returns: a %NULL-terminated gchar ** array. Free it using g_strfreev()
1722 g_regex_split (const GRegex *regex,
1723 const gchar *string,
1724 GRegexMatchFlags match_options)
1726 return g_regex_split_full (regex, string, -1, 0,
1727 match_options, 0, NULL);
1731 * g_regex_split_full:
1732 * @regex: a #GRegex structure
1733 * @string: the string to split with the pattern
1734 * @string_len: the length of @string, or -1 if @string is nul-terminated
1735 * @start_position: starting index of the string to match
1736 * @match_options: match time option flags
1737 * @max_tokens: the maximum number of tokens to split @string into.
1738 * If this is less than 1, the string is split completely
1739 * @error: return location for a #GError
1741 * Breaks the string on the pattern, and returns an array of the tokens.
1742 * If the pattern contains capturing parentheses, then the text for each
1743 * of the substrings will also be returned. If the pattern does not match
1744 * anywhere in the string, then the whole string is returned as the first
1747 * As a special case, the result of splitting the empty string "" is an
1748 * empty vector, not a vector containing a single string. The reason for
1749 * this special case is that being able to represent a empty vector is
1750 * typically more useful than consistent handling of empty elements. If
1751 * you do need to represent empty elements, you'll need to check for the
1752 * empty string before calling this function.
1754 * A pattern that can match empty strings splits @string into separate
1755 * characters wherever it matches the empty string between characters.
1756 * For example splitting "ab c" using as a separator "\s*", you will get
1759 * Setting @start_position differs from just passing over a shortened
1760 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
1761 * that begins with any kind of lookbehind assertion, such as "\b".
1763 * Returns: a %NULL-terminated gchar ** array. Free it using g_strfreev()
1768 g_regex_split_full (const GRegex *regex,
1769 const gchar *string,
1771 gint start_position,
1772 GRegexMatchFlags match_options,
1776 GError *tmp_error = NULL;
1777 GMatchInfo *match_info;
1782 /* position of the last separator. */
1783 gint last_separator_end;
1784 /* was the last match 0 bytes long? */
1785 gboolean last_match_is_empty;
1786 /* the returned array of char **s */
1787 gchar **string_list;
1789 g_return_val_if_fail (regex != NULL, NULL);
1790 g_return_val_if_fail (string != NULL, NULL);
1791 g_return_val_if_fail (start_position >= 0, NULL);
1792 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
1793 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
1795 if (max_tokens <= 0)
1796 max_tokens = G_MAXINT;
1799 string_len = strlen (string);
1801 /* zero-length string */
1802 if (string_len - start_position == 0)
1803 return g_new0 (gchar *, 1);
1805 if (max_tokens == 1)
1807 string_list = g_new0 (gchar *, 2);
1808 string_list[0] = g_strndup (&string[start_position],
1809 string_len - start_position);
1815 last_separator_end = start_position;
1816 last_match_is_empty = FALSE;
1818 match_ok = g_regex_match_full (regex, string, string_len, start_position,
1819 match_options, &match_info, &tmp_error);
1820 while (tmp_error == NULL)
1824 last_match_is_empty =
1825 (match_info->offsets[0] == match_info->offsets[1]);
1827 /* we need to skip empty separators at the same position of the end
1828 * of another separator. e.g. the string is "a b" and the separator
1829 * is " *", so from 1 to 2 we have a match and at position 2 we have
1830 * an empty match. */
1831 if (last_separator_end != match_info->offsets[1])
1836 token = g_strndup (string + last_separator_end,
1837 match_info->offsets[0] - last_separator_end);
1838 list = g_list_prepend (list, token);
1841 /* if there were substrings, these need to be added to
1843 match_count = g_match_info_get_match_count (match_info);
1844 if (match_count > 1)
1846 for (i = 1; i < match_count; i++)
1847 list = g_list_prepend (list, g_match_info_fetch (match_info, i));
1853 /* if there was no match, copy to end of string. */
1854 if (!last_match_is_empty)
1856 gchar *token = g_strndup (string + last_separator_end,
1857 match_info->string_len - last_separator_end);
1858 list = g_list_prepend (list, token);
1860 /* no more tokens, end the loop. */
1864 /* -1 to leave room for the last part. */
1865 if (token_count >= max_tokens - 1)
1867 /* we have reached the maximum number of tokens, so we copy
1868 * the remaining part of the string. */
1869 if (last_match_is_empty)
1871 /* the last match was empty, so we have moved one char
1872 * after the real position to avoid empty matches at the
1874 match_info->pos = PREV_CHAR (regex, &string[match_info->pos]) - string;
1876 /* the if is needed in the case we have terminated the available
1877 * tokens, but we are at the end of the string, so there are no
1878 * characters left to copy. */
1879 if (string_len > match_info->pos)
1881 gchar *token = g_strndup (string + match_info->pos,
1882 string_len - match_info->pos);
1883 list = g_list_prepend (list, token);
1889 last_separator_end = match_info->pos;
1890 if (last_match_is_empty)
1891 /* if the last match was empty, g_match_info_next() has moved
1892 * forward to avoid infinite loops, but we still need to copy that
1894 last_separator_end = PREV_CHAR (regex, &string[last_separator_end]) - string;
1896 match_ok = g_match_info_next (match_info, &tmp_error);
1898 g_match_info_free (match_info);
1899 if (tmp_error != NULL)
1901 g_propagate_error (error, tmp_error);
1902 g_list_foreach (list, (GFunc)g_free, NULL);
1904 match_info->pos = -1;
1908 string_list = g_new (gchar *, g_list_length (list) + 1);
1910 for (last = g_list_last (list); last; last = g_list_previous (last))
1911 string_list[i++] = last->data;
1912 string_list[i] = NULL;
1921 REPL_TYPE_CHARACTER,
1922 REPL_TYPE_SYMBOLIC_REFERENCE,
1923 REPL_TYPE_NUMERIC_REFERENCE,
1924 REPL_TYPE_CHANGE_CASE
1929 CHANGE_CASE_NONE = 1 << 0,
1930 CHANGE_CASE_UPPER = 1 << 1,
1931 CHANGE_CASE_LOWER = 1 << 2,
1932 CHANGE_CASE_UPPER_SINGLE = 1 << 3,
1933 CHANGE_CASE_LOWER_SINGLE = 1 << 4,
1934 CHANGE_CASE_SINGLE_MASK = CHANGE_CASE_UPPER_SINGLE | CHANGE_CASE_LOWER_SINGLE,
1935 CHANGE_CASE_LOWER_MASK = CHANGE_CASE_LOWER | CHANGE_CASE_LOWER_SINGLE,
1936 CHANGE_CASE_UPPER_MASK = CHANGE_CASE_UPPER | CHANGE_CASE_UPPER_SINGLE
1939 struct _InterpolationData
1945 ChangeCase change_case;
1949 free_interpolation_data (InterpolationData *data)
1951 g_free (data->text);
1955 static const gchar *
1956 expand_escape (const gchar *replacement,
1958 InterpolationData *data,
1963 const gchar *error_detail;
1965 GError *tmp_error = NULL;
1973 data->type = REPL_TYPE_CHARACTER;
1978 data->type = REPL_TYPE_CHARACTER;
1983 data->type = REPL_TYPE_CHARACTER;
1988 data->type = REPL_TYPE_CHARACTER;
1993 data->type = REPL_TYPE_CHARACTER;
1998 data->type = REPL_TYPE_CHARACTER;
2003 data->type = REPL_TYPE_CHARACTER;
2008 data->type = REPL_TYPE_CHARACTER;
2018 h = g_ascii_xdigit_value (*p);
2021 error_detail = _("hexadecimal digit or '}' expected");
2032 for (i = 0; i < 2; i++)
2034 h = g_ascii_xdigit_value (*p);
2037 error_detail = _("hexadecimal digit expected");
2044 data->type = REPL_TYPE_STRING;
2045 data->text = g_new0 (gchar, 8);
2046 g_unichar_to_utf8 (x, data->text);
2050 data->type = REPL_TYPE_CHANGE_CASE;
2051 data->change_case = CHANGE_CASE_LOWER_SINGLE;
2055 data->type = REPL_TYPE_CHANGE_CASE;
2056 data->change_case = CHANGE_CASE_UPPER_SINGLE;
2060 data->type = REPL_TYPE_CHANGE_CASE;
2061 data->change_case = CHANGE_CASE_LOWER;
2065 data->type = REPL_TYPE_CHANGE_CASE;
2066 data->change_case = CHANGE_CASE_UPPER;
2070 data->type = REPL_TYPE_CHANGE_CASE;
2071 data->change_case = CHANGE_CASE_NONE;
2077 error_detail = _("missing '<' in symbolic reference");
2086 error_detail = _("unfinished symbolic reference");
2093 error_detail = _("zero-length symbolic reference");
2096 if (g_ascii_isdigit (*q))
2101 h = g_ascii_digit_value (*q);
2104 error_detail = _("digit expected");
2113 data->type = REPL_TYPE_NUMERIC_REFERENCE;
2120 if (!g_ascii_isalnum (*r))
2122 error_detail = _("illegal symbolic reference");
2129 data->text = g_strndup (q, p - q);
2130 data->type = REPL_TYPE_SYMBOLIC_REFERENCE;
2135 /* if \0 is followed by a number is an octal number representing a
2136 * character, else it is a numeric reference. */
2137 if (g_ascii_digit_value (*g_utf8_next_char (p)) >= 0)
2140 p = g_utf8_next_char (p);
2153 for (i = 0; i < 3; i++)
2155 h = g_ascii_digit_value (*p);
2165 if (i == 2 && base == 10)
2171 if (base == 8 || i == 3)
2173 data->type = REPL_TYPE_STRING;
2174 data->text = g_new0 (gchar, 8);
2175 g_unichar_to_utf8 (x, data->text);
2179 data->type = REPL_TYPE_NUMERIC_REFERENCE;
2184 error_detail = _("stray final '\\'");
2188 error_detail = _("unknown escape sequence");
2195 /* G_GSSIZE_FORMAT doesn't work with gettext, so we use %lu */
2196 tmp_error = g_error_new (G_REGEX_ERROR,
2197 G_REGEX_ERROR_REPLACE,
2198 _("Error while parsing replacement "
2199 "text \"%s\" at char %lu: %s"),
2201 (gulong)(p - replacement),
2203 g_propagate_error (error, tmp_error);
2209 split_replacement (const gchar *replacement,
2213 InterpolationData *data;
2214 const gchar *p, *start;
2216 start = p = replacement;
2221 data = g_new0 (InterpolationData, 1);
2222 start = p = expand_escape (replacement, p, data, error);
2225 g_list_foreach (list, (GFunc)free_interpolation_data, NULL);
2227 free_interpolation_data (data);
2231 list = g_list_prepend (list, data);
2236 if (*p == '\\' || *p == '\0')
2240 data = g_new0 (InterpolationData, 1);
2241 data->text = g_strndup (start, p - start);
2242 data->type = REPL_TYPE_STRING;
2243 list = g_list_prepend (list, data);
2249 return g_list_reverse (list);
2252 /* Change the case of c based on change_case. */
2253 #define CHANGE_CASE(c, change_case) \
2254 (((change_case) & CHANGE_CASE_LOWER_MASK) ? \
2255 g_unichar_tolower (c) : \
2256 g_unichar_toupper (c))
2259 string_append (GString *string,
2261 ChangeCase *change_case)
2265 if (text[0] == '\0')
2268 if (*change_case == CHANGE_CASE_NONE)
2270 g_string_append (string, text);
2272 else if (*change_case & CHANGE_CASE_SINGLE_MASK)
2274 c = g_utf8_get_char (text);
2275 g_string_append_unichar (string, CHANGE_CASE (c, *change_case));
2276 g_string_append (string, g_utf8_next_char (text));
2277 *change_case = CHANGE_CASE_NONE;
2281 while (*text != '\0')
2283 c = g_utf8_get_char (text);
2284 g_string_append_unichar (string, CHANGE_CASE (c, *change_case));
2285 text = g_utf8_next_char (text);
2291 interpolate_replacement (const GMatchInfo *match_info,
2296 InterpolationData *idata;
2298 ChangeCase change_case = CHANGE_CASE_NONE;
2300 for (list = data; list; list = list->next)
2303 switch (idata->type)
2305 case REPL_TYPE_STRING:
2306 string_append (result, idata->text, &change_case);
2308 case REPL_TYPE_CHARACTER:
2309 g_string_append_c (result, CHANGE_CASE (idata->c, change_case));
2310 if (change_case & CHANGE_CASE_SINGLE_MASK)
2311 change_case = CHANGE_CASE_NONE;
2313 case REPL_TYPE_NUMERIC_REFERENCE:
2314 match = g_match_info_fetch (match_info, idata->num);
2317 string_append (result, match, &change_case);
2321 case REPL_TYPE_SYMBOLIC_REFERENCE:
2322 match = g_match_info_fetch_named (match_info, idata->text);
2325 string_append (result, match, &change_case);
2329 case REPL_TYPE_CHANGE_CASE:
2330 change_case = idata->change_case;
2338 /* whether actual match_info is needed for replacement, i.e.
2339 * whether there are references
2342 interpolation_list_needs_match (GList *list)
2344 while (list != NULL)
2346 InterpolationData *data = list->data;
2348 if (data->type == REPL_TYPE_SYMBOLIC_REFERENCE ||
2349 data->type == REPL_TYPE_NUMERIC_REFERENCE)
2362 * @regex: a #GRegex structure
2363 * @string: the string to perform matches against
2364 * @string_len: the length of @string, or -1 if @string is nul-terminated
2365 * @start_position: starting index of the string to match
2366 * @replacement: text to replace each match with
2367 * @match_options: options for the match
2368 * @error: location to store the error occuring, or %NULL to ignore errors
2370 * Replaces all occurances of the pattern in @regex with the
2371 * replacement text. Backreferences of the form '\number' or
2372 * '\g<number>' in the replacement text are interpolated by the
2373 * number-th captured subexpression of the match, '\g<name>' refers
2374 * to the captured subexpression with the given name. '\0' refers to the
2375 * complete match, but '\0' followed by a number is the octal representation
2376 * of a character. To include a literal '\' in the replacement, write '\\'.
2377 * There are also escapes that changes the case of the following text:
2380 * <varlistentry><term>\l</term>
2382 * <para>Convert to lower case the next character</para>
2385 * <varlistentry><term>\u</term>
2387 * <para>Convert to upper case the next character</para>
2390 * <varlistentry><term>\L</term>
2392 * <para>Convert to lower case till \E</para>
2395 * <varlistentry><term>\U</term>
2397 * <para>Convert to upper case till \E</para>
2400 * <varlistentry><term>\E</term>
2402 * <para>End case modification</para>
2407 * If you do not need to use backreferences use g_regex_replace_literal().
2409 * The @replacement string must be UTF-8 encoded even if #G_REGEX_RAW was
2410 * passed to g_regex_new(). If you want to use not UTF-8 encoded stings
2411 * you can use g_regex_replace_literal().
2413 * Setting @start_position differs from just passing over a shortened
2414 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern that
2415 * begins with any kind of lookbehind assertion, such as "\b".
2417 * Returns: a newly allocated string containing the replacements
2422 g_regex_replace (const GRegex *regex,
2423 const gchar *string,
2425 gint start_position,
2426 const gchar *replacement,
2427 GRegexMatchFlags match_options,
2432 GError *tmp_error = NULL;
2434 g_return_val_if_fail (regex != NULL, NULL);
2435 g_return_val_if_fail (string != NULL, NULL);
2436 g_return_val_if_fail (start_position >= 0, NULL);
2437 g_return_val_if_fail (replacement != NULL, NULL);
2438 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
2439 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2441 list = split_replacement (replacement, &tmp_error);
2442 if (tmp_error != NULL)
2444 g_propagate_error (error, tmp_error);
2448 result = g_regex_replace_eval (regex,
2449 string, string_len, start_position,
2451 interpolate_replacement,
2454 if (tmp_error != NULL)
2455 g_propagate_error (error, tmp_error);
2457 g_list_foreach (list, (GFunc)free_interpolation_data, NULL);
2464 literal_replacement (const GMatchInfo *match_info,
2468 g_string_append (result, data);
2473 * g_regex_replace_literal:
2474 * @regex: a #GRegex structure
2475 * @string: the string to perform matches against
2476 * @string_len: the length of @string, or -1 if @string is nul-terminated
2477 * @start_position: starting index of the string to match
2478 * @replacement: text to replace each match with
2479 * @match_options: options for the match
2480 * @error: location to store the error occuring, or %NULL to ignore errors
2482 * Replaces all occurances of the pattern in @regex with the
2483 * replacement text. @replacement is replaced literally, to
2484 * include backreferences use g_regex_replace().
2486 * Setting @start_position differs from just passing over a
2487 * shortened string and setting #G_REGEX_MATCH_NOTBOL in the
2488 * case of a pattern that begins with any kind of lookbehind
2489 * assertion, such as "\b".
2491 * Returns: a newly allocated string containing the replacements
2496 g_regex_replace_literal (const GRegex *regex,
2497 const gchar *string,
2499 gint start_position,
2500 const gchar *replacement,
2501 GRegexMatchFlags match_options,
2504 g_return_val_if_fail (replacement != NULL, NULL);
2505 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2507 return g_regex_replace_eval (regex,
2508 string, string_len, start_position,
2510 literal_replacement,
2511 (gpointer)replacement,
2516 * g_regex_replace_eval:
2517 * @regex: a #GRegex structure from g_regex_new()
2518 * @string: string to perform matches against
2519 * @string_len: the length of @string, or -1 if @string is nul-terminated
2520 * @start_position: starting index of the string to match
2521 * @match_options: options for the match
2522 * @eval: a function to call for each match
2523 * @user_data: user data to pass to the function
2524 * @error: location to store the error occuring, or %NULL to ignore errors
2526 * Replaces occurances of the pattern in regex with the output of
2527 * @eval for that occurance.
2529 * Setting @start_position differs from just passing over a shortened
2530 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
2531 * that begins with any kind of lookbehind assertion, such as "\b".
2533 * Returns: a newly allocated string containing the replacements
2538 g_regex_replace_eval (const GRegex *regex,
2539 const gchar *string,
2541 gint start_position,
2542 GRegexMatchFlags match_options,
2543 GRegexEvalCallback eval,
2547 GMatchInfo *match_info;
2550 gboolean done = FALSE;
2551 GError *tmp_error = NULL;
2553 g_return_val_if_fail (regex != NULL, NULL);
2554 g_return_val_if_fail (string != NULL, NULL);
2555 g_return_val_if_fail (start_position >= 0, NULL);
2556 g_return_val_if_fail (eval != NULL, NULL);
2557 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2560 string_len = strlen (string);
2562 result = g_string_sized_new (string_len);
2564 /* run down the string making matches. */
2565 g_regex_match_full (regex, string, string_len, start_position,
2566 match_options, &match_info, &tmp_error);
2567 while (!done && g_match_info_matches (match_info))
2569 g_string_append_len (result,
2571 match_info->offsets[0] - str_pos);
2572 done = (*eval) (match_info, result, user_data);
2573 str_pos = match_info->offsets[1];
2574 g_match_info_next (match_info, &tmp_error);
2576 g_match_info_free (match_info);
2577 if (tmp_error != NULL)
2579 g_propagate_error (error, tmp_error);
2580 g_string_free (result, TRUE);
2584 g_string_append_len (result, string + str_pos, string_len - str_pos);
2585 return g_string_free (result, FALSE);
2589 * g_regex_check_replacement:
2590 * @replacement: the replacement string
2591 * @has_references: location to store information about
2592 * references in @replacement or %NULL
2593 * @error: location to store error
2595 * Checks whether @replacement is a valid replacement string
2596 * (see g_regex_replace()), i.e. that all escape sequences in
2599 * If @has_references is not %NULL then @replacement is checked
2600 * for pattern references. For instance, replacement text 'foo\n'
2601 * does not contain references and may be evaluated without information
2602 * about actual match, but '\0\1' (whole match followed by first
2603 * subpattern) requires valid #GMatchInfo object.
2605 * Returns: whether @replacement is a valid replacement string
2610 g_regex_check_replacement (const gchar *replacement,
2611 gboolean *has_references,
2617 list = split_replacement (replacement, &tmp);
2621 g_propagate_error (error, tmp);
2626 *has_references = interpolation_list_needs_match (list);
2628 g_list_foreach (list, (GFunc) free_interpolation_data, NULL);
2635 * g_regex_escape_string:
2636 * @string: the string to escape
2637 * @length: the length of @string, or -1 if @string is nul-terminated
2639 * Escapes the special characters used for regular expressions
2640 * in @string, for instance "a.b*c" becomes "a\.b\*c". This
2641 * function is useful to dynamically generate regular expressions.
2643 * @string can contain nul characters that are replaced with "\0",
2644 * in this case remember to specify the correct length of @string
2647 * Returns: a newly-allocated escaped string
2652 g_regex_escape_string (const gchar *string,
2656 const char *p, *piece_start, *end;
2658 g_return_val_if_fail (string != NULL, NULL);
2661 length = strlen (string);
2663 end = string + length;
2664 p = piece_start = string;
2665 escaped = g_string_sized_new (length + 1);
2686 if (p != piece_start)
2687 /* copy the previous piece. */
2688 g_string_append_len (escaped, piece_start, p - piece_start);
2689 g_string_append_c (escaped, '\\');
2691 g_string_append_c (escaped, '0');
2693 g_string_append_c (escaped, *p);
2697 p = g_utf8_next_char (p);
2702 if (piece_start < end)
2703 g_string_append_len (escaped, piece_start, end - piece_start);
2705 return g_string_free (escaped, FALSE);
2708 #define __G_REGEX_C__
2709 #include "galiasdef.c"