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
26 #ifdef USE_SYSTEM_PCRE
29 #include "pcre/pcre.h"
36 #include "gmessages.h"
37 #include "gstrfuncs.h"
43 * @title: Perl-compatible regular expressions
44 * @short_description: matches strings against regular expressions
45 * @see_also: <xref linkend="glib-regex-syntax"/>
47 * The <function>g_regex_*()</function> functions implement regular
48 * expression pattern matching using syntax and semantics similar to
49 * Perl regular expression.
51 * Some functions accept a @start_position argument, setting it differs
52 * from just passing over a shortened string and setting #G_REGEX_MATCH_NOTBOL
53 * in the case of a pattern that begins with any kind of lookbehind assertion.
54 * For example, consider the pattern "\Biss\B" which finds occurrences of "iss"
55 * in the middle of words. ("\B" matches only if the current position in the
56 * subject is not a word boundary.) When applied to the string "Mississipi"
57 * from the fourth byte, namely "issipi", it does not match, because "\B" is
58 * always false at the start of the subject, which is deemed to be a word
59 * boundary. However, if the entire string is passed , but with
60 * @start_position set to 4, it finds the second occurrence of "iss" because
61 * it is able to look behind the starting point to discover that it is
62 * preceded by a letter.
64 * Note that, unless you set the #G_REGEX_RAW flag, all the strings passed
65 * to these functions must be encoded in UTF-8. The lengths and the positions
66 * inside the strings are in bytes and not in characters, so, for instance,
67 * "\xc3\xa0" (i.e. "à") is two bytes long but it is treated as a
68 * single character. If you set #G_REGEX_RAW the strings can be non-valid
69 * UTF-8 strings and a byte is treated as a character, so "\xc3\xa0" is two
70 * bytes and two characters long.
72 * When matching a pattern, "\n" matches only against a "\n" character in
73 * the string, and "\r" matches only a "\r" character. To match any newline
74 * sequence use "\R". This particular group matches either the two-character
75 * sequence CR + LF ("\r\n"), or one of the single characters LF (linefeed,
76 * U+000A, "\n"), VT vertical tab, U+000B, "\v"), FF (formfeed, U+000C, "\f"),
77 * CR (carriage return, U+000D, "\r"), NEL (next line, U+0085), LS (line
78 * separator, U+2028), or PS (paragraph separator, U+2029).
80 * The behaviour of the dot, circumflex, and dollar metacharacters are
81 * affected by newline characters, the default is to recognize any newline
82 * character (the same characters recognized by "\R"). This can be changed
83 * with #G_REGEX_NEWLINE_CR, #G_REGEX_NEWLINE_LF and #G_REGEX_NEWLINE_CRLF
84 * compile options, and with #G_REGEX_MATCH_NEWLINE_ANY,
85 * #G_REGEX_MATCH_NEWLINE_CR, #G_REGEX_MATCH_NEWLINE_LF and
86 * #G_REGEX_MATCH_NEWLINE_CRLF match options. These settings are also
87 * relevant when compiling a pattern if #G_REGEX_EXTENDED is set, and an
88 * unescaped "#" outside a character class is encountered. This indicates
89 * a comment that lasts until after the next newline.
91 * When setting the %G_REGEX_JAVASCRIPT_COMPAT flag, pattern syntax and pattern
92 * matching is changed to be compatible with the way that regular expressions
93 * work in JavaScript. More precisely, a lonely ']' character in the pattern
94 * is a syntax error; the '\x' escape only allows 0 to 2 hexadecimal digits, and
95 * you must use the '\u' escape sequence with 4 hex digits to specify a unicode
96 * codepoint instead of '\x' or 'x{....}'. If '\x' or '\u' are not followed by
97 * the specified number of hex digits, they match 'x' and 'u' literally; also
98 * '\U' always matches 'U' instead of being an error in the pattern. Finally,
99 * pattern matching is modified so that back references to an unset subpattern
100 * group produces a match with the empty string instead of an error. See
101 * <ulink>man:pcreapi(3)<ulink> for more information.
103 * Creating and manipulating the same #GRegex structure from different
104 * threads is not a problem as #GRegex does not modify its internal
105 * state between creation and destruction, on the other hand #GMatchInfo
108 * The regular expressions low-level functionalities are obtained through
109 * the excellent <ulink url="http://www.pcre.org/">PCRE</ulink> library
110 * written by Philip Hazel.
113 /* Mask of all the possible values for GRegexCompileFlags. */
114 #define G_REGEX_COMPILE_MASK (G_REGEX_CASELESS | \
115 G_REGEX_MULTILINE | \
119 G_REGEX_DOLLAR_ENDONLY | \
122 G_REGEX_NO_AUTO_CAPTURE | \
124 G_REGEX_FIRSTLINE | \
126 G_REGEX_NEWLINE_CR | \
127 G_REGEX_NEWLINE_LF | \
128 G_REGEX_NEWLINE_CRLF | \
129 G_REGEX_NEWLINE_ANYCRLF | \
130 G_REGEX_BSR_ANYCRLF | \
131 G_REGEX_JAVASCRIPT_COMPAT)
133 /* Mask of all GRegexCompileFlags values that are (not) passed trough to PCRE */
134 #define G_REGEX_COMPILE_PCRE_MASK (G_REGEX_COMPILE_MASK & ~G_REGEX_COMPILE_NONPCRE_MASK)
135 #define G_REGEX_COMPILE_NONPCRE_MASK (G_REGEX_RAW | \
138 /* Mask of all the possible values for GRegexMatchFlags. */
139 #define G_REGEX_MATCH_MASK (G_REGEX_MATCH_ANCHORED | \
140 G_REGEX_MATCH_NOTBOL | \
141 G_REGEX_MATCH_NOTEOL | \
142 G_REGEX_MATCH_NOTEMPTY | \
143 G_REGEX_MATCH_PARTIAL | \
144 G_REGEX_MATCH_NEWLINE_CR | \
145 G_REGEX_MATCH_NEWLINE_LF | \
146 G_REGEX_MATCH_NEWLINE_CRLF | \
147 G_REGEX_MATCH_NEWLINE_ANY | \
148 G_REGEX_MATCH_NEWLINE_ANYCRLF | \
149 G_REGEX_MATCH_BSR_ANYCRLF | \
150 G_REGEX_MATCH_BSR_ANY | \
151 G_REGEX_MATCH_PARTIAL_SOFT | \
152 G_REGEX_MATCH_PARTIAL_HARD | \
153 G_REGEX_MATCH_NOTEMPTY_ATSTART)
155 /* we rely on these flags having the same values */
156 G_STATIC_ASSERT (G_REGEX_CASELESS == PCRE_CASELESS);
157 G_STATIC_ASSERT (G_REGEX_MULTILINE == PCRE_MULTILINE);
158 G_STATIC_ASSERT (G_REGEX_DOTALL == PCRE_DOTALL);
159 G_STATIC_ASSERT (G_REGEX_EXTENDED == PCRE_EXTENDED);
160 G_STATIC_ASSERT (G_REGEX_ANCHORED == PCRE_ANCHORED);
161 G_STATIC_ASSERT (G_REGEX_DOLLAR_ENDONLY == PCRE_DOLLAR_ENDONLY);
162 G_STATIC_ASSERT (G_REGEX_UNGREEDY == PCRE_UNGREEDY);
163 G_STATIC_ASSERT (G_REGEX_NO_AUTO_CAPTURE == PCRE_NO_AUTO_CAPTURE);
164 G_STATIC_ASSERT (G_REGEX_FIRSTLINE == PCRE_FIRSTLINE);
165 G_STATIC_ASSERT (G_REGEX_DUPNAMES == PCRE_DUPNAMES);
166 G_STATIC_ASSERT (G_REGEX_NEWLINE_CR == PCRE_NEWLINE_CR);
167 G_STATIC_ASSERT (G_REGEX_NEWLINE_LF == PCRE_NEWLINE_LF);
168 G_STATIC_ASSERT (G_REGEX_NEWLINE_CRLF == PCRE_NEWLINE_CRLF);
169 G_STATIC_ASSERT (G_REGEX_NEWLINE_ANYCRLF == PCRE_NEWLINE_ANYCRLF);
170 G_STATIC_ASSERT (G_REGEX_BSR_ANYCRLF == PCRE_BSR_ANYCRLF);
171 G_STATIC_ASSERT (G_REGEX_JAVASCRIPT_COMPAT == PCRE_JAVASCRIPT_COMPAT);
173 G_STATIC_ASSERT (G_REGEX_MATCH_ANCHORED == PCRE_ANCHORED);
174 G_STATIC_ASSERT (G_REGEX_MATCH_NOTBOL == PCRE_NOTBOL);
175 G_STATIC_ASSERT (G_REGEX_MATCH_NOTEOL == PCRE_NOTEOL);
176 G_STATIC_ASSERT (G_REGEX_MATCH_NOTEMPTY == PCRE_NOTEMPTY);
177 G_STATIC_ASSERT (G_REGEX_MATCH_PARTIAL == PCRE_PARTIAL);
178 G_STATIC_ASSERT (G_REGEX_MATCH_NEWLINE_CR == PCRE_NEWLINE_CR);
179 G_STATIC_ASSERT (G_REGEX_MATCH_NEWLINE_LF == PCRE_NEWLINE_LF);
180 G_STATIC_ASSERT (G_REGEX_MATCH_NEWLINE_CRLF == PCRE_NEWLINE_CRLF);
181 G_STATIC_ASSERT (G_REGEX_MATCH_NEWLINE_ANY == PCRE_NEWLINE_ANY);
182 G_STATIC_ASSERT (G_REGEX_MATCH_NEWLINE_ANYCRLF == PCRE_NEWLINE_ANYCRLF);
183 G_STATIC_ASSERT (G_REGEX_MATCH_BSR_ANYCRLF == PCRE_BSR_ANYCRLF);
184 G_STATIC_ASSERT (G_REGEX_MATCH_BSR_ANY == PCRE_BSR_UNICODE);
185 G_STATIC_ASSERT (G_REGEX_MATCH_PARTIAL_SOFT == PCRE_PARTIAL_SOFT);
186 G_STATIC_ASSERT (G_REGEX_MATCH_PARTIAL_HARD == PCRE_PARTIAL_HARD);
187 G_STATIC_ASSERT (G_REGEX_MATCH_NOTEMPTY_ATSTART == PCRE_NOTEMPTY_ATSTART);
189 /* These PCRE flags are unused or not exposed publically in GRegexFlags, so
190 * it should be ok to reuse them for different things.
192 G_STATIC_ASSERT (G_REGEX_OPTIMIZE == PCRE_NO_UTF8_CHECK);
193 G_STATIC_ASSERT (G_REGEX_RAW == PCRE_UTF8);
195 /* if the string is in UTF-8 use g_utf8_ functions, else use
197 #define NEXT_CHAR(re, s) (((re)->compile_opts & G_REGEX_RAW) ? \
199 g_utf8_next_char (s))
200 #define PREV_CHAR(re, s) (((re)->compile_opts & G_REGEX_RAW) ? \
202 g_utf8_prev_char (s))
206 volatile gint ref_count; /* the ref count */
207 GRegex *regex; /* the regex */
208 GRegexMatchFlags match_opts; /* options used at match time on the regex */
209 gint matches; /* number of matching sub patterns */
210 gint pos; /* position in the string where last match left off */
211 gint n_offsets; /* number of offsets */
212 gint *offsets; /* array of offsets paired 0,1 ; 2,3 ; 3,4 etc */
213 gint *workspace; /* workspace for pcre_dfa_exec() */
214 gint n_workspace; /* number of workspace elements */
215 const gchar *string; /* string passed to the match function */
216 gssize string_len; /* length of string */
221 volatile gint ref_count; /* the ref count for the immutable part */
222 gchar *pattern; /* the pattern */
223 pcre *pcre_re; /* compiled form of the pattern */
224 GRegexCompileFlags compile_opts; /* options used at compile time on the pattern */
225 GRegexMatchFlags match_opts; /* options used at match time on the regex */
226 pcre_extra *extra; /* data stored when G_REGEX_OPTIMIZE is used */
229 /* TRUE if ret is an error code, FALSE otherwise. */
230 #define IS_PCRE_ERROR(ret) ((ret) < PCRE_ERROR_NOMATCH && (ret) != PCRE_ERROR_PARTIAL)
232 typedef struct _InterpolationData InterpolationData;
233 static gboolean interpolation_list_needs_match (GList *list);
234 static gboolean interpolate_replacement (const GMatchInfo *match_info,
237 static GList *split_replacement (const gchar *replacement,
239 static void free_interpolation_data (InterpolationData *data);
243 match_error (gint errcode)
247 case PCRE_ERROR_NOMATCH:
250 case PCRE_ERROR_NULL:
251 /* NULL argument, this should not happen in GRegex */
252 g_warning ("A NULL argument was passed to PCRE");
254 case PCRE_ERROR_BADOPTION:
255 return "bad options";
256 case PCRE_ERROR_BADMAGIC:
257 return _("corrupted object");
258 case PCRE_ERROR_UNKNOWN_OPCODE:
259 return N_("internal error or corrupted object");
260 case PCRE_ERROR_NOMEMORY:
261 return _("out of memory");
262 case PCRE_ERROR_NOSUBSTRING:
263 /* not used by pcre_exec() */
265 case PCRE_ERROR_MATCHLIMIT:
266 return _("backtracking limit reached");
267 case PCRE_ERROR_CALLOUT:
268 /* callouts are not implemented */
270 case PCRE_ERROR_BADUTF8:
271 case PCRE_ERROR_BADUTF8_OFFSET:
272 /* we do not check if strings are valid */
274 case PCRE_ERROR_PARTIAL:
277 case PCRE_ERROR_BADPARTIAL:
278 return _("the pattern contains items not supported for partial matching");
279 case PCRE_ERROR_INTERNAL:
280 return _("internal error");
281 case PCRE_ERROR_BADCOUNT:
282 /* negative ovecsize, this should not happen in GRegex */
283 g_warning ("A negative ovecsize was passed to PCRE");
285 case PCRE_ERROR_DFA_UITEM:
286 return _("the pattern contains items not supported for partial matching");
287 case PCRE_ERROR_DFA_UCOND:
288 return _("back references as conditions are not supported for partial matching");
289 case PCRE_ERROR_DFA_UMLIMIT:
290 /* the match_field field is not used in GRegex */
292 case PCRE_ERROR_DFA_WSSIZE:
293 /* handled expanding the workspace */
295 case PCRE_ERROR_DFA_RECURSE:
296 case PCRE_ERROR_RECURSIONLIMIT:
297 return _("recursion limit reached");
298 case PCRE_ERROR_BADNEWLINE:
299 return _("invalid combination of newline flags");
300 case PCRE_ERROR_BADOFFSET:
301 return _("bad offset");
302 case PCRE_ERROR_SHORTUTF8:
303 return _("short utf8");
304 case PCRE_ERROR_RECURSELOOP:
305 return _("recursion loop");
309 return _("unknown error");
313 translate_compile_error (gint *errcode, const gchar **errmsg)
315 /* Compile errors are created adding 100 to the error code returned
317 * If errcode is known we put the translatable error message in
318 * erromsg. If errcode is unknown we put the generic
319 * G_REGEX_ERROR_COMPILE error code in errcode and keep the
320 * untranslated error message returned by PCRE.
321 * Note that there can be more PCRE errors with the same GRegexError
322 * and that some PCRE errors are useless for us.
328 case G_REGEX_ERROR_STRAY_BACKSLASH:
329 *errmsg = _("\\ at end of pattern");
331 case G_REGEX_ERROR_MISSING_CONTROL_CHAR:
332 *errmsg = _("\\c at end of pattern");
334 case G_REGEX_ERROR_UNRECOGNIZED_ESCAPE:
335 *errmsg = _("unrecognized character follows \\");
337 case G_REGEX_ERROR_QUANTIFIERS_OUT_OF_ORDER:
338 *errmsg = _("numbers out of order in {} quantifier");
340 case G_REGEX_ERROR_QUANTIFIER_TOO_BIG:
341 *errmsg = _("number too big in {} quantifier");
343 case G_REGEX_ERROR_UNTERMINATED_CHARACTER_CLASS:
344 *errmsg = _("missing terminating ] for character class");
346 case G_REGEX_ERROR_INVALID_ESCAPE_IN_CHARACTER_CLASS:
347 *errmsg = _("invalid escape sequence in character class");
349 case G_REGEX_ERROR_RANGE_OUT_OF_ORDER:
350 *errmsg = _("range out of order in character class");
352 case G_REGEX_ERROR_NOTHING_TO_REPEAT:
353 *errmsg = _("nothing to repeat");
355 case 111: /* internal error: unexpected repeat */
356 *errcode = G_REGEX_ERROR_INTERNAL;
357 *errmsg = _("unexpected repeat");
359 case G_REGEX_ERROR_UNRECOGNIZED_CHARACTER:
360 *errmsg = _("unrecognized character after (? or (?-");
362 case G_REGEX_ERROR_POSIX_NAMED_CLASS_OUTSIDE_CLASS:
363 *errmsg = _("POSIX named classes are supported only within a class");
365 case G_REGEX_ERROR_UNMATCHED_PARENTHESIS:
366 *errmsg = _("missing terminating )");
368 case G_REGEX_ERROR_INEXISTENT_SUBPATTERN_REFERENCE:
369 *errmsg = _("reference to non-existent subpattern");
371 case G_REGEX_ERROR_UNTERMINATED_COMMENT:
372 *errmsg = _("missing ) after comment");
374 case G_REGEX_ERROR_EXPRESSION_TOO_LARGE:
375 *errmsg = _("regular expression is too large");
377 case G_REGEX_ERROR_MEMORY_ERROR:
378 *errmsg = _("failed to get memory");
380 case 122: /* unmatched parentheses */
381 *errcode = G_REGEX_ERROR_UNMATCHED_PARENTHESIS;
382 *errmsg = _(") without opening (");
384 case 123: /* internal error: code overflow */
385 *errcode = G_REGEX_ERROR_INTERNAL;
386 *errmsg = _("code overflow");
388 case 124: /* "unrecognized character after (?<\0 */
389 *errcode = G_REGEX_ERROR_UNRECOGNIZED_CHARACTER;
390 *errmsg = _("unrecognized character after (?<");
392 case G_REGEX_ERROR_VARIABLE_LENGTH_LOOKBEHIND:
393 *errmsg = _("lookbehind assertion is not fixed length");
395 case G_REGEX_ERROR_MALFORMED_CONDITION:
396 *errmsg = _("malformed number or name after (?(");
398 case G_REGEX_ERROR_TOO_MANY_CONDITIONAL_BRANCHES:
399 *errmsg = _("conditional group contains more than two branches");
401 case G_REGEX_ERROR_ASSERTION_EXPECTED:
402 *errmsg = _("assertion expected after (?(");
405 *errcode = G_REGEX_ERROR_UNMATCHED_PARENTHESIS;
406 /* translators: '(?R' and '(?[+-]digits' are both meant as (groups of)
407 * sequences here, '(?-54' would be an example for the second group.
409 *errmsg = _("(?R or (?[+-]digits must be followed by )");
411 case G_REGEX_ERROR_UNKNOWN_POSIX_CLASS_NAME:
412 *errmsg = _("unknown POSIX class name");
414 case G_REGEX_ERROR_POSIX_COLLATING_ELEMENTS_NOT_SUPPORTED:
415 *errmsg = _("POSIX collating elements are not supported");
417 case G_REGEX_ERROR_HEX_CODE_TOO_LARGE:
418 *errmsg = _("character value in \\x{...} sequence is too large");
420 case G_REGEX_ERROR_INVALID_CONDITION:
421 *errmsg = _("invalid condition (?(0)");
423 case G_REGEX_ERROR_SINGLE_BYTE_MATCH_IN_LOOKBEHIND:
424 *errmsg = _("\\C not allowed in lookbehind assertion");
426 case 137: /* PCRE does not support \\L, \\l, \\N{name}, \\U, or \\u\0 */
427 /* A number of Perl escapes are not handled by PCRE.
428 * Therefore it explicitly raises ERR37.
430 *errcode = G_REGEX_ERROR_UNRECOGNIZED_ESCAPE;
431 *errmsg = _("escapes \\L, \\l, \\N{name}, \\U, and \\u are not supported");
433 case G_REGEX_ERROR_INFINITE_LOOP:
434 *errmsg = _("recursive call could loop indefinitely");
436 case 141: /* unrecognized character after (?P\0 */
437 *errcode = G_REGEX_ERROR_UNRECOGNIZED_CHARACTER;
438 *errmsg = _("unrecognized character after (?P");
440 case G_REGEX_ERROR_MISSING_SUBPATTERN_NAME_TERMINATOR:
441 *errmsg = _("missing terminator in subpattern name");
443 case G_REGEX_ERROR_DUPLICATE_SUBPATTERN_NAME:
444 *errmsg = _("two named subpatterns have the same name");
446 case G_REGEX_ERROR_MALFORMED_PROPERTY:
447 *errmsg = _("malformed \\P or \\p sequence");
449 case G_REGEX_ERROR_UNKNOWN_PROPERTY:
450 *errmsg = _("unknown property name after \\P or \\p");
452 case G_REGEX_ERROR_SUBPATTERN_NAME_TOO_LONG:
453 *errmsg = _("subpattern name is too long (maximum 32 characters)");
455 case G_REGEX_ERROR_TOO_MANY_SUBPATTERNS:
456 *errmsg = _("too many named subpatterns (maximum 10,000)");
458 case G_REGEX_ERROR_INVALID_OCTAL_VALUE:
459 *errmsg = _("octal value is greater than \\377");
461 case 152: /* internal error: overran compiling workspace */
462 *errcode = G_REGEX_ERROR_INTERNAL;
463 *errmsg = _("overran compiling workspace");
465 case 153: /* internal error: previously-checked referenced subpattern not found */
466 *errcode = G_REGEX_ERROR_INTERNAL;
467 *errmsg = _("previously-checked referenced subpattern not found");
469 case G_REGEX_ERROR_TOO_MANY_BRANCHES_IN_DEFINE:
470 *errmsg = _("DEFINE group contains more than one branch");
472 case G_REGEX_ERROR_INCONSISTENT_NEWLINE_OPTIONS:
473 *errmsg = _("inconsistent NEWLINE options");
475 case G_REGEX_ERROR_MISSING_BACK_REFERENCE:
476 *errmsg = _("\\g is not followed by a braced, angle-bracketed, or quoted name or "
477 "number, or by a plain number");
479 case G_REGEX_ERROR_INVALID_RELATIVE_REFERENCE:
480 *errmsg = _("a numbered reference must not be zero");
482 case G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_FORBIDDEN:
483 *errmsg = _("an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)");
485 case G_REGEX_ERROR_UNKNOWN_BACKTRACKING_CONTROL_VERB:
486 *errmsg = _("(*VERB) not recognized");
488 case G_REGEX_ERROR_NUMBER_TOO_BIG:
489 *errmsg = _("number is too big");
491 case G_REGEX_ERROR_MISSING_SUBPATTERN_NAME:
492 *errmsg = _("missing subpattern name after (?&");
494 case G_REGEX_ERROR_MISSING_DIGIT:
495 *errmsg = _("digit expected after (?+");
497 case G_REGEX_ERROR_INVALID_DATA_CHARACTER:
498 *errmsg = _("] is an invalid data character in JavaScript compatibility mode");
500 case G_REGEX_ERROR_EXTRA_SUBPATTERN_NAME:
501 *errmsg = _("different names for subpatterns of the same number are not allowed");
503 case G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_REQUIRED:
504 *errmsg = _("(*MARK) must have an argument");
506 case G_REGEX_ERROR_INVALID_CONTROL_CHAR:
507 *errmsg = _( "\\c must be followed by an ASCII character");
509 case G_REGEX_ERROR_MISSING_NAME:
510 *errmsg = _("\\k is not followed by a braced, angle-bracketed, or quoted name");
512 case G_REGEX_ERROR_NOT_SUPPORTED_IN_CLASS:
513 *errmsg = _("\\N is not supported in a class");
515 case G_REGEX_ERROR_TOO_MANY_FORWARD_REFERENCES:
516 *errmsg = _("too many forward references");
518 case G_REGEX_ERROR_NAME_TOO_LONG:
519 *errmsg = _("name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)");
522 case 116: /* erroffset passed as NULL */
523 /* This should not happen as we never pass a NULL erroffset */
524 g_warning ("erroffset passed as NULL");
525 *errcode = G_REGEX_ERROR_COMPILE;
527 case 117: /* unknown option bit(s) set */
528 /* This should not happen as we check options before passing them
529 * to pcre_compile2() */
530 g_warning ("unknown option bit(s) set");
531 *errcode = G_REGEX_ERROR_COMPILE;
533 case 132: /* this version of PCRE is compiled without UTF support */
534 case 144: /* invalid UTF-8 string */
535 case 145: /* support for \\P, \\p, and \\X has not been compiled */
536 case 167: /* this version of PCRE is not compiled with Unicode property support */
537 case 173: /* disallowed Unicode code point (>= 0xd800 && <= 0xdfff) */
538 case 174: /* invalid UTF-16 string */
539 /* These errors should not happen as we are using an UTF-8 and UCP-enabled PCRE
540 * and we do not check if strings are valid */
541 case 170: /* internal error: unknown opcode in find_fixedlength() */
542 *errcode = G_REGEX_ERROR_INTERNAL;
546 *errcode = G_REGEX_ERROR_COMPILE;
553 match_info_new (const GRegex *regex,
560 GMatchInfo *match_info;
563 string_len = strlen (string);
565 match_info = g_new0 (GMatchInfo, 1);
566 match_info->ref_count = 1;
567 match_info->regex = g_regex_ref ((GRegex *)regex);
568 match_info->string = string;
569 match_info->string_len = string_len;
570 match_info->matches = PCRE_ERROR_NOMATCH;
571 match_info->pos = start_position;
572 match_info->match_opts = match_options;
576 /* These values should be enough for most cases, if they are not
577 * enough g_regex_match_all_full() will expand them. */
578 match_info->n_offsets = 24;
579 match_info->n_workspace = 100;
580 match_info->workspace = g_new (gint, match_info->n_workspace);
585 pcre_fullinfo (regex->pcre_re, regex->extra,
586 PCRE_INFO_CAPTURECOUNT, &capture_count);
587 match_info->n_offsets = (capture_count + 1) * 3;
590 match_info->offsets = g_new0 (gint, match_info->n_offsets);
591 /* Set an invalid position for the previous match. */
592 match_info->offsets[0] = -1;
593 match_info->offsets[1] = -1;
599 * g_match_info_get_regex:
600 * @match_info: a #GMatchInfo
602 * Returns #GRegex object used in @match_info. It belongs to Glib
603 * and must not be freed. Use g_regex_ref() if you need to keep it
604 * after you free @match_info object.
606 * Returns: #GRegex object used in @match_info
611 g_match_info_get_regex (const GMatchInfo *match_info)
613 g_return_val_if_fail (match_info != NULL, NULL);
614 return match_info->regex;
618 * g_match_info_get_string:
619 * @match_info: a #GMatchInfo
621 * Returns the string searched with @match_info. This is the
622 * string passed to g_regex_match() or g_regex_replace() so
623 * you may not free it before calling this function.
625 * Returns: the string searched with @match_info
630 g_match_info_get_string (const GMatchInfo *match_info)
632 g_return_val_if_fail (match_info != NULL, NULL);
633 return match_info->string;
638 * @match_info: a #GMatchInfo
640 * Increases reference count of @match_info by 1.
642 * Returns: @match_info
647 g_match_info_ref (GMatchInfo *match_info)
649 g_return_val_if_fail (match_info != NULL, NULL);
650 g_atomic_int_inc (&match_info->ref_count);
655 * g_match_info_unref:
656 * @match_info: a #GMatchInfo
658 * Decreases reference count of @match_info by 1. When reference count drops
659 * to zero, it frees all the memory associated with the match_info structure.
664 g_match_info_unref (GMatchInfo *match_info)
666 if (g_atomic_int_dec_and_test (&match_info->ref_count))
668 g_regex_unref (match_info->regex);
669 g_free (match_info->offsets);
670 g_free (match_info->workspace);
677 * @match_info: (allow-none): a #GMatchInfo, or %NULL
679 * If @match_info is not %NULL, calls g_match_info_unref(); otherwise does
685 g_match_info_free (GMatchInfo *match_info)
687 if (match_info == NULL)
690 g_match_info_unref (match_info);
695 * @match_info: a #GMatchInfo structure
696 * @error: location to store the error occurring, or %NULL to ignore errors
698 * Scans for the next match using the same parameters of the previous
699 * call to g_regex_match_full() or g_regex_match() that returned
702 * The match is done on the string passed to the match function, so you
703 * cannot free it before calling this function.
705 * Returns: %TRUE is the string matched, %FALSE otherwise
710 g_match_info_next (GMatchInfo *match_info,
713 gint prev_match_start;
716 g_return_val_if_fail (match_info != NULL, FALSE);
717 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
718 g_return_val_if_fail (match_info->pos >= 0, FALSE);
720 prev_match_start = match_info->offsets[0];
721 prev_match_end = match_info->offsets[1];
723 if (match_info->pos > match_info->string_len)
725 /* we have reached the end of the string */
726 match_info->pos = -1;
727 match_info->matches = PCRE_ERROR_NOMATCH;
731 match_info->matches = pcre_exec (match_info->regex->pcre_re,
732 match_info->regex->extra,
734 match_info->string_len,
736 match_info->regex->match_opts | match_info->match_opts,
738 match_info->n_offsets);
739 if (IS_PCRE_ERROR (match_info->matches))
741 g_set_error (error, G_REGEX_ERROR, G_REGEX_ERROR_MATCH,
742 _("Error while matching regular expression %s: %s"),
743 match_info->regex->pattern, match_error (match_info->matches));
747 /* avoid infinite loops if the pattern is an empty string or something
749 if (match_info->pos == match_info->offsets[1])
751 if (match_info->pos > match_info->string_len)
753 /* we have reached the end of the string */
754 match_info->pos = -1;
755 match_info->matches = PCRE_ERROR_NOMATCH;
759 match_info->pos = NEXT_CHAR (match_info->regex,
760 &match_info->string[match_info->pos]) -
765 match_info->pos = match_info->offsets[1];
768 /* it's possible to get two identical matches when we are matching
769 * empty strings, for instance if the pattern is "(?=[A-Z0-9])" and
770 * the string is "RegExTest" we have:
771 * - search at position 0: match from 0 to 0
772 * - search at position 1: match from 3 to 3
773 * - search at position 3: match from 3 to 3 (duplicate)
774 * - search at position 4: match from 5 to 5
775 * - search at position 5: match from 5 to 5 (duplicate)
776 * - search at position 6: no match -> stop
777 * so we have to ignore the duplicates.
778 * see bug #515944: http://bugzilla.gnome.org/show_bug.cgi?id=515944 */
779 if (match_info->matches >= 0 &&
780 prev_match_start == match_info->offsets[0] &&
781 prev_match_end == match_info->offsets[1])
783 /* ignore this match and search the next one */
784 return g_match_info_next (match_info, error);
787 return match_info->matches >= 0;
791 * g_match_info_matches:
792 * @match_info: a #GMatchInfo structure
794 * Returns whether the previous match operation succeeded.
796 * Returns: %TRUE if the previous match operation succeeded,
802 g_match_info_matches (const GMatchInfo *match_info)
804 g_return_val_if_fail (match_info != NULL, FALSE);
806 return match_info->matches >= 0;
810 * g_match_info_get_match_count:
811 * @match_info: a #GMatchInfo structure
813 * Retrieves the number of matched substrings (including substring 0,
814 * that is the whole matched text), so 1 is returned if the pattern
815 * has no substrings in it and 0 is returned if the match failed.
817 * If the last match was obtained using the DFA algorithm, that is
818 * using g_regex_match_all() or g_regex_match_all_full(), the retrieved
819 * count is not that of the number of capturing parentheses but that of
820 * the number of matched substrings.
822 * Returns: Number of matched substrings, or -1 if an error occurred
827 g_match_info_get_match_count (const GMatchInfo *match_info)
829 g_return_val_if_fail (match_info, -1);
831 if (match_info->matches == PCRE_ERROR_NOMATCH)
834 else if (match_info->matches < PCRE_ERROR_NOMATCH)
839 return match_info->matches;
843 * g_match_info_is_partial_match:
844 * @match_info: a #GMatchInfo structure
846 * Usually if the string passed to g_regex_match*() matches as far as
847 * it goes, but is too short to match the entire pattern, %FALSE is
848 * returned. There are circumstances where it might be helpful to
849 * distinguish this case from other cases in which there is no match.
851 * Consider, for example, an application where a human is required to
852 * type in data for a field with specific formatting requirements. An
853 * example might be a date in the form ddmmmyy, defined by the pattern
854 * "^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$".
855 * If the application sees the user’s keystrokes one by one, and can
856 * check that what has been typed so far is potentially valid, it is
857 * able to raise an error as soon as a mistake is made.
859 * GRegex supports the concept of partial matching by means of the
860 * #G_REGEX_MATCH_PARTIAL_SOFT and #G_REGEX_MATCH_PARTIAL_HARD flags.
861 * When they are used, the return code for
862 * g_regex_match() or g_regex_match_full() is, as usual, %TRUE
863 * for a complete match, %FALSE otherwise. But, when these functions
864 * return %FALSE, you can check if the match was partial calling
865 * g_match_info_is_partial_match().
867 * The difference between #G_REGEX_MATCH_PARTIAL_SOFT and
868 * #G_REGEX_MATCH_PARTIAL_HARD is that when a partial match is encountered
869 * with #G_REGEX_MATCH_PARTIAL_SOFT, matching continues to search for a
870 * possible complete match, while with #G_REGEX_MATCH_PARTIAL_HARD matching
871 * stops at the partial match.
872 * When both #G_REGEX_MATCH_PARTIAL_SOFT and #G_REGEX_MATCH_PARTIAL_HARD
873 * are set, the latter takes precedence.
874 * See <ulink>man:pcrepartial</ulink> for more information on partial matching.
876 * Because of the way certain internal optimizations are implemented
877 * the partial matching algorithm cannot be used with all patterns.
878 * So repeated single characters such as "a{2,4}" and repeated single
879 * meta-sequences such as "\d+" are not permitted if the maximum number
880 * of occurrences is greater than one. Optional items such as "\d?"
881 * (where the maximum is one) are permitted. Quantifiers with any values
882 * are permitted after parentheses, so the invalid examples above can be
883 * coded thus "(a){2,4}" and "(\d)+". If #G_REGEX_MATCH_PARTIAL or
884 * #G_REGEX_MATCH_PARTIAL_HARD is set
885 * for a pattern that does not conform to the restrictions, matching
886 * functions return an error.
888 * Returns: %TRUE if the match was partial, %FALSE otherwise
893 g_match_info_is_partial_match (const GMatchInfo *match_info)
895 g_return_val_if_fail (match_info != NULL, FALSE);
897 return match_info->matches == PCRE_ERROR_PARTIAL;
901 * g_match_info_expand_references:
902 * @match_info: (allow-none): a #GMatchInfo or %NULL
903 * @string_to_expand: the string to expand
904 * @error: location to store the error occurring, or %NULL to ignore errors
906 * Returns a new string containing the text in @string_to_expand with
907 * references and escape sequences expanded. References refer to the last
908 * match done with @string against @regex and have the same syntax used by
911 * The @string_to_expand must be UTF-8 encoded even if #G_REGEX_RAW was
912 * passed to g_regex_new().
914 * The backreferences are extracted from the string passed to the match
915 * function, so you cannot call this function after freeing the string.
917 * @match_info may be %NULL in which case @string_to_expand must not
918 * contain references. For instance "foo\n" does not refer to an actual
919 * pattern and '\n' merely will be replaced with \n character,
920 * while to expand "\0" (whole match) one needs the result of a match.
921 * Use g_regex_check_replacement() to find out whether @string_to_expand
922 * contains references.
924 * Returns: (allow-none): the expanded string, or %NULL if an error occurred
929 g_match_info_expand_references (const GMatchInfo *match_info,
930 const gchar *string_to_expand,
935 GError *tmp_error = NULL;
937 g_return_val_if_fail (string_to_expand != NULL, NULL);
938 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
940 list = split_replacement (string_to_expand, &tmp_error);
941 if (tmp_error != NULL)
943 g_propagate_error (error, tmp_error);
947 if (!match_info && interpolation_list_needs_match (list))
949 g_critical ("String '%s' contains references to the match, can't "
950 "expand references without GMatchInfo object",
955 result = g_string_sized_new (strlen (string_to_expand));
956 interpolate_replacement (match_info, result, list);
958 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
960 return g_string_free (result, FALSE);
964 * g_match_info_fetch:
965 * @match_info: #GMatchInfo structure
966 * @match_num: number of the sub expression
968 * Retrieves the text matching the @match_num<!-- -->'th capturing
969 * parentheses. 0 is the full text of the match, 1 is the first paren
970 * set, 2 the second, and so on.
972 * If @match_num is a valid sub pattern but it didn't match anything
973 * (e.g. sub pattern 1, matching "b" against "(a)?b") then an empty
974 * string is returned.
976 * If the match was obtained using the DFA algorithm, that is using
977 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
978 * string is not that of a set of parentheses but that of a matched
979 * substring. Substrings are matched in reverse order of length, so
980 * 0 is the longest match.
982 * The string is fetched from the string passed to the match function,
983 * so you cannot call this function after freeing the string.
985 * Returns: (allow-none): The matched substring, or %NULL if an error
986 * occurred. You have to free the string yourself
991 g_match_info_fetch (const GMatchInfo *match_info,
994 /* we cannot use pcre_get_substring() because it allocates the
995 * string using pcre_malloc(). */
999 g_return_val_if_fail (match_info != NULL, NULL);
1000 g_return_val_if_fail (match_num >= 0, NULL);
1002 /* match_num does not exist or it didn't matched, i.e. matching "b"
1003 * against "(a)?b" then group 0 is empty. */
1004 if (!g_match_info_fetch_pos (match_info, match_num, &start, &end))
1006 else if (start == -1)
1007 match = g_strdup ("");
1009 match = g_strndup (&match_info->string[start], end - start);
1015 * g_match_info_fetch_pos:
1016 * @match_info: #GMatchInfo structure
1017 * @match_num: number of the sub expression
1018 * @start_pos: (out) (allow-none): pointer to location where to store
1019 * the start position, or %NULL
1020 * @end_pos: (out) (allow-none): pointer to location where to store
1021 * the end position, or %NULL
1023 * Retrieves the position in bytes of the @match_num<!-- -->'th capturing
1024 * parentheses. 0 is the full text of the match, 1 is the first
1025 * paren set, 2 the second, and so on.
1027 * If @match_num is a valid sub pattern but it didn't match anything
1028 * (e.g. sub pattern 1, matching "b" against "(a)?b") then @start_pos
1029 * and @end_pos are set to -1 and %TRUE is returned.
1031 * If the match was obtained using the DFA algorithm, that is using
1032 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
1033 * position is not that of a set of parentheses but that of a matched
1034 * substring. Substrings are matched in reverse order of length, so
1035 * 0 is the longest match.
1037 * Returns: %TRUE if the position was fetched, %FALSE otherwise. If
1038 * the position cannot be fetched, @start_pos and @end_pos are left
1044 g_match_info_fetch_pos (const GMatchInfo *match_info,
1049 g_return_val_if_fail (match_info != NULL, FALSE);
1050 g_return_val_if_fail (match_num >= 0, FALSE);
1052 /* make sure the sub expression number they're requesting is less than
1053 * the total number of sub expressions that were matched. */
1054 if (match_num >= match_info->matches)
1057 if (start_pos != NULL)
1058 *start_pos = match_info->offsets[2 * match_num];
1060 if (end_pos != NULL)
1061 *end_pos = match_info->offsets[2 * match_num + 1];
1067 * Returns number of first matched subpattern with name @name.
1068 * There may be more than one in case when DUPNAMES is used,
1069 * and not all subpatterns with that name match;
1070 * pcre_get_stringnumber() does not work in that case.
1073 get_matched_substring_number (const GMatchInfo *match_info,
1077 gchar *first, *last;
1080 if (!(match_info->regex->compile_opts & G_REGEX_DUPNAMES))
1081 return pcre_get_stringnumber (match_info->regex->pcre_re, name);
1083 /* This code is copied from pcre_get.c: get_first_set() */
1084 entrysize = pcre_get_stringtable_entries (match_info->regex->pcre_re,
1092 for (entry = (guchar*) first; entry <= (guchar*) last; entry += entrysize)
1094 gint n = (entry[0] << 8) + entry[1];
1095 if (match_info->offsets[n*2] >= 0)
1099 return (first[0] << 8) + first[1];
1103 * g_match_info_fetch_named:
1104 * @match_info: #GMatchInfo structure
1105 * @name: name of the subexpression
1107 * Retrieves the text matching the capturing parentheses named @name.
1109 * If @name is a valid sub pattern name but it didn't match anything
1110 * (e.g. sub pattern "X", matching "b" against "(?P<X>a)?b")
1111 * then an empty string is returned.
1113 * The string is fetched from the string passed to the match function,
1114 * so you cannot call this function after freeing the string.
1116 * Returns: (allow-none): The matched substring, or %NULL if an error
1117 * occurred. You have to free the string yourself
1122 g_match_info_fetch_named (const GMatchInfo *match_info,
1125 /* we cannot use pcre_get_named_substring() because it allocates the
1126 * string using pcre_malloc(). */
1129 g_return_val_if_fail (match_info != NULL, NULL);
1130 g_return_val_if_fail (name != NULL, NULL);
1132 num = get_matched_substring_number (match_info, name);
1136 return g_match_info_fetch (match_info, num);
1140 * g_match_info_fetch_named_pos:
1141 * @match_info: #GMatchInfo structure
1142 * @name: name of the subexpression
1143 * @start_pos: (out) (allow-none): pointer to location where to store
1144 * the start position, or %NULL
1145 * @end_pos: (out) (allow-none): pointer to location where to store
1146 * the end position, or %NULL
1148 * Retrieves the position in bytes of the capturing parentheses named @name.
1150 * If @name is a valid sub pattern name but it didn't match anything
1151 * (e.g. sub pattern "X", matching "b" against "(?P<X>a)?b")
1152 * then @start_pos and @end_pos are set to -1 and %TRUE is returned.
1154 * Returns: %TRUE if the position was fetched, %FALSE otherwise.
1155 * If the position cannot be fetched, @start_pos and @end_pos
1156 * are left unchanged.
1161 g_match_info_fetch_named_pos (const GMatchInfo *match_info,
1168 g_return_val_if_fail (match_info != NULL, FALSE);
1169 g_return_val_if_fail (name != NULL, FALSE);
1171 num = get_matched_substring_number (match_info, name);
1175 return g_match_info_fetch_pos (match_info, num, start_pos, end_pos);
1179 * g_match_info_fetch_all:
1180 * @match_info: a #GMatchInfo structure
1182 * Bundles up pointers to each of the matching substrings from a match
1183 * and stores them in an array of gchar pointers. The first element in
1184 * the returned array is the match number 0, i.e. the entire matched
1187 * If a sub pattern didn't match anything (e.g. sub pattern 1, matching
1188 * "b" against "(a)?b") then an empty string is inserted.
1190 * If the last match was obtained using the DFA algorithm, that is using
1191 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
1192 * strings are not that matched by sets of parentheses but that of the
1193 * matched substring. Substrings are matched in reverse order of length,
1194 * so the first one is the longest match.
1196 * The strings are fetched from the string passed to the match function,
1197 * so you cannot call this function after freeing the string.
1199 * Returns: (allow-none): a %NULL-terminated array of gchar * pointers.
1200 * It must be freed using g_strfreev(). If the previous match failed
1206 g_match_info_fetch_all (const GMatchInfo *match_info)
1208 /* we cannot use pcre_get_substring_list() because the returned value
1209 * isn't suitable for g_strfreev(). */
1213 g_return_val_if_fail (match_info != NULL, NULL);
1215 if (match_info->matches < 0)
1218 result = g_new (gchar *, match_info->matches + 1);
1219 for (i = 0; i < match_info->matches; i++)
1220 result[i] = g_match_info_fetch (match_info, i);
1230 g_regex_error_quark (void)
1232 static GQuark error_quark = 0;
1234 if (error_quark == 0)
1235 error_quark = g_quark_from_static_string ("g-regex-error-quark");
1244 * Increases reference count of @regex by 1.
1251 g_regex_ref (GRegex *regex)
1253 g_return_val_if_fail (regex != NULL, NULL);
1254 g_atomic_int_inc (®ex->ref_count);
1262 * Decreases reference count of @regex by 1. When reference count drops
1263 * to zero, it frees all the memory associated with the regex structure.
1268 g_regex_unref (GRegex *regex)
1270 g_return_if_fail (regex != NULL);
1272 if (g_atomic_int_dec_and_test (®ex->ref_count))
1274 g_free (regex->pattern);
1275 if (regex->pcre_re != NULL)
1276 pcre_free (regex->pcre_re);
1277 if (regex->extra != NULL)
1278 pcre_free (regex->extra);
1285 * @pattern: the regular expression
1286 * @compile_options: compile options for the regular expression, or 0
1287 * @match_options: match options for the regular expression, or 0
1288 * @error: return location for a #GError
1290 * Compiles the regular expression to an internal form, and does
1291 * the initial setup of the #GRegex structure.
1293 * Returns: a #GRegex structure. Call g_regex_unref() when you
1299 g_regex_new (const gchar *pattern,
1300 GRegexCompileFlags compile_options,
1301 GRegexMatchFlags match_options,
1306 const gchar *errmsg;
1309 gboolean optimize = FALSE;
1310 static volatile gsize initialised = 0;
1311 unsigned long int pcre_compile_options;
1312 GRegexCompileFlags nonpcre_compile_options;
1314 g_return_val_if_fail (pattern != NULL, NULL);
1315 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
1316 g_return_val_if_fail ((compile_options & ~G_REGEX_COMPILE_MASK) == 0, NULL);
1317 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
1319 if (g_once_init_enter (&initialised))
1321 int supports_utf8, supports_ucp;
1323 pcre_config (PCRE_CONFIG_UTF8, &supports_utf8);
1325 g_critical (_("PCRE library is compiled without UTF8 support"));
1327 pcre_config (PCRE_CONFIG_UNICODE_PROPERTIES, &supports_ucp);
1329 g_critical (_("PCRE library is compiled without UTF8 properties support"));
1331 g_once_init_leave (&initialised, supports_utf8 && supports_ucp ? 1 : 2);
1334 if (G_UNLIKELY (initialised != 1))
1336 g_set_error_literal (error, G_REGEX_ERROR, G_REGEX_ERROR_COMPILE,
1337 _("PCRE library is compiled with incompatible options"));
1341 nonpcre_compile_options = compile_options & G_REGEX_COMPILE_NONPCRE_MASK;
1343 /* G_REGEX_OPTIMIZE has the same numeric value of PCRE_NO_UTF8_CHECK,
1344 * as we do not need to wrap PCRE_NO_UTF8_CHECK. */
1345 if (compile_options & G_REGEX_OPTIMIZE)
1348 /* In GRegex the string are, by default, UTF-8 encoded. PCRE
1349 * instead uses UTF-8 only if required with PCRE_UTF8. */
1350 if (compile_options & G_REGEX_RAW)
1353 compile_options &= ~G_REGEX_RAW;
1358 compile_options |= PCRE_UTF8 | PCRE_NO_UTF8_CHECK;
1359 match_options |= PCRE_NO_UTF8_CHECK;
1362 /* PCRE_NEWLINE_ANY is the default for the internal PCRE but
1363 * not for the system one. */
1364 if (!(compile_options & G_REGEX_NEWLINE_CR) &&
1365 !(compile_options & G_REGEX_NEWLINE_LF))
1367 compile_options |= PCRE_NEWLINE_ANY;
1370 compile_options |= PCRE_UCP;
1372 /* PCRE_BSR_UNICODE is the default for the internal PCRE but
1373 * possibly not for the system one.
1375 if (~compile_options & G_REGEX_BSR_ANYCRLF)
1376 compile_options |= PCRE_BSR_UNICODE;
1378 /* compile the pattern */
1379 re = pcre_compile2 (pattern, compile_options, &errcode,
1380 &errmsg, &erroffset, NULL);
1382 /* if the compilation failed, set the error member and return
1388 /* Translate the PCRE error code to GRegexError and use a translated
1389 * error message if possible */
1390 translate_compile_error (&errcode, &errmsg);
1392 /* PCRE uses byte offsets but we want to show character offsets */
1393 erroffset = g_utf8_pointer_to_offset (pattern, &pattern[erroffset]);
1395 tmp_error = g_error_new (G_REGEX_ERROR, errcode,
1396 _("Error while compiling regular "
1397 "expression %s at char %d: %s"),
1398 pattern, erroffset, errmsg);
1399 g_propagate_error (error, tmp_error);
1404 /* For options set at the beginning of the pattern, pcre puts them into
1405 * compile options, e.g. "(?i)foo" will make the pcre structure store
1406 * PCRE_CASELESS even though it wasn't explicitly given for compilation. */
1407 pcre_fullinfo (re, NULL, PCRE_INFO_OPTIONS, &pcre_compile_options);
1408 compile_options = pcre_compile_options & G_REGEX_COMPILE_PCRE_MASK;
1410 /* Don't leak PCRE_NEWLINE_ANY, which is part of PCRE_NEWLINE_ANYCRLF */
1411 if ((pcre_compile_options & PCRE_NEWLINE_ANYCRLF) != PCRE_NEWLINE_ANYCRLF)
1412 compile_options &= ~PCRE_NEWLINE_ANY;
1414 compile_options |= nonpcre_compile_options;
1416 if (!(compile_options & G_REGEX_DUPNAMES))
1418 gboolean jchanged = FALSE;
1419 pcre_fullinfo (re, NULL, PCRE_INFO_JCHANGED, &jchanged);
1421 compile_options |= G_REGEX_DUPNAMES;
1424 regex = g_new0 (GRegex, 1);
1425 regex->ref_count = 1;
1426 regex->pattern = g_strdup (pattern);
1427 regex->pcre_re = re;
1428 regex->compile_opts = compile_options;
1429 regex->match_opts = match_options;
1433 regex->extra = pcre_study (regex->pcre_re, 0, &errmsg);
1436 GError *tmp_error = g_error_new (G_REGEX_ERROR,
1437 G_REGEX_ERROR_OPTIMIZE,
1438 _("Error while optimizing "
1439 "regular expression %s: %s"),
1442 g_propagate_error (error, tmp_error);
1444 g_regex_unref (regex);
1453 * g_regex_get_pattern:
1454 * @regex: a #GRegex structure
1456 * Gets the pattern string associated with @regex, i.e. a copy of
1457 * the string passed to g_regex_new().
1459 * Returns: the pattern of @regex
1464 g_regex_get_pattern (const GRegex *regex)
1466 g_return_val_if_fail (regex != NULL, NULL);
1468 return regex->pattern;
1472 * g_regex_get_max_backref:
1475 * Returns the number of the highest back reference
1476 * in the pattern, or 0 if the pattern does not contain
1479 * Returns: the number of the highest back reference
1484 g_regex_get_max_backref (const GRegex *regex)
1488 pcre_fullinfo (regex->pcre_re, regex->extra,
1489 PCRE_INFO_BACKREFMAX, &value);
1495 * g_regex_get_capture_count:
1498 * Returns the number of capturing subpatterns in the pattern.
1500 * Returns: the number of capturing subpatterns
1505 g_regex_get_capture_count (const GRegex *regex)
1509 pcre_fullinfo (regex->pcre_re, regex->extra,
1510 PCRE_INFO_CAPTURECOUNT, &value);
1516 * g_regex_get_has_cr_or_lf:
1517 * @regex: a #GRegex structure
1519 * Checks whether the pattern contains explicit CR or LF references.
1521 * Returns: %TRUE if the pattern contains explicit CR or LF references
1526 g_regex_get_has_cr_or_lf (const GRegex *regex)
1530 pcre_fullinfo (regex->pcre_re, regex->extra,
1531 PCRE_INFO_HASCRORLF, &value);
1537 * g_regex_get_compile_flags:
1540 * Returns the compile options that @regex was created with.
1542 * Returns: flags from #GRegexCompileFlags
1547 g_regex_get_compile_flags (const GRegex *regex)
1549 g_return_val_if_fail (regex != NULL, 0);
1551 return regex->compile_opts;
1555 * g_regex_get_match_flags:
1558 * Returns the match options that @regex was created with.
1560 * Returns: flags from #GRegexMatchFlags
1565 g_regex_get_match_flags (const GRegex *regex)
1567 g_return_val_if_fail (regex != NULL, 0);
1569 return regex->match_opts & G_REGEX_MATCH_MASK;
1573 * g_regex_match_simple:
1574 * @pattern: the regular expression
1575 * @string: the string to scan for matches
1576 * @compile_options: compile options for the regular expression, or 0
1577 * @match_options: match options, or 0
1579 * Scans for a match in @string for @pattern.
1581 * This function is equivalent to g_regex_match() but it does not
1582 * require to compile the pattern with g_regex_new(), avoiding some
1583 * lines of code when you need just to do a match without extracting
1584 * substrings, capture counts, and so on.
1586 * If this function is to be called on the same @pattern more than
1587 * once, it's more efficient to compile the pattern once with
1588 * g_regex_new() and then use g_regex_match().
1590 * Returns: %TRUE if the string matched, %FALSE otherwise
1595 g_regex_match_simple (const gchar *pattern,
1596 const gchar *string,
1597 GRegexCompileFlags compile_options,
1598 GRegexMatchFlags match_options)
1603 regex = g_regex_new (pattern, compile_options, 0, NULL);
1606 result = g_regex_match_full (regex, string, -1, 0, match_options, NULL, NULL);
1607 g_regex_unref (regex);
1613 * @regex: a #GRegex structure from g_regex_new()
1614 * @string: the string to scan for matches
1615 * @match_options: match options
1616 * @match_info: (out) (allow-none): pointer to location where to store
1617 * the #GMatchInfo, or %NULL if you do not need it
1619 * Scans for a match in string for the pattern in @regex.
1620 * The @match_options are combined with the match options specified
1621 * when the @regex structure was created, letting you have more
1622 * flexibility in reusing #GRegex structures.
1624 * A #GMatchInfo structure, used to get information on the match,
1625 * is stored in @match_info if not %NULL. Note that if @match_info
1626 * is not %NULL then it is created even if the function returns %FALSE,
1627 * i.e. you must free it regardless if regular expression actually matched.
1629 * To retrieve all the non-overlapping matches of the pattern in
1630 * string you can use g_match_info_next().
1634 * print_uppercase_words (const gchar *string)
1636 * /* Print all uppercase-only words. */
1638 * GMatchInfo *match_info;
1640 * regex = g_regex_new ("[A-Z]+", 0, 0, NULL);
1641 * g_regex_match (regex, string, 0, &match_info);
1642 * while (g_match_info_matches (match_info))
1644 * gchar *word = g_match_info_fetch (match_info, 0);
1645 * g_print ("Found: %s\n", word);
1647 * g_match_info_next (match_info, NULL);
1649 * g_match_info_free (match_info);
1650 * g_regex_unref (regex);
1654 * @string is not copied and is used in #GMatchInfo internally. If
1655 * you use any #GMatchInfo method (except g_match_info_free()) after
1656 * freeing or modifying @string then the behaviour is undefined.
1658 * Returns: %TRUE is the string matched, %FALSE otherwise
1663 g_regex_match (const GRegex *regex,
1664 const gchar *string,
1665 GRegexMatchFlags match_options,
1666 GMatchInfo **match_info)
1668 return g_regex_match_full (regex, string, -1, 0, match_options,
1673 * g_regex_match_full:
1674 * @regex: a #GRegex structure from g_regex_new()
1675 * @string: (array length=string_len): the string to scan for matches
1676 * @string_len: the length of @string, or -1 if @string is nul-terminated
1677 * @start_position: starting index of the string to match
1678 * @match_options: match options
1679 * @match_info: (out) (allow-none): pointer to location where to store
1680 * the #GMatchInfo, or %NULL if you do not need it
1681 * @error: location to store the error occurring, or %NULL to ignore errors
1683 * Scans for a match in string for the pattern in @regex.
1684 * The @match_options are combined with the match options specified
1685 * when the @regex structure was created, letting you have more
1686 * flexibility in reusing #GRegex structures.
1688 * Setting @start_position differs from just passing over a shortened
1689 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
1690 * that begins with any kind of lookbehind assertion, such as "\b".
1692 * A #GMatchInfo structure, used to get information on the match, is
1693 * stored in @match_info if not %NULL. Note that if @match_info is
1694 * not %NULL then it is created even if the function returns %FALSE,
1695 * i.e. you must free it regardless if regular expression actually
1698 * @string is not copied and is used in #GMatchInfo internally. If
1699 * you use any #GMatchInfo method (except g_match_info_free()) after
1700 * freeing or modifying @string then the behaviour is undefined.
1702 * To retrieve all the non-overlapping matches of the pattern in
1703 * string you can use g_match_info_next().
1707 * print_uppercase_words (const gchar *string)
1709 * /* Print all uppercase-only words. */
1711 * GMatchInfo *match_info;
1712 * GError *error = NULL;
1714 * regex = g_regex_new ("[A-Z]+", 0, 0, NULL);
1715 * g_regex_match_full (regex, string, -1, 0, 0, &match_info, &error);
1716 * while (g_match_info_matches (match_info))
1718 * gchar *word = g_match_info_fetch (match_info, 0);
1719 * g_print ("Found: %s\n", word);
1721 * g_match_info_next (match_info, &error);
1723 * g_match_info_free (match_info);
1724 * g_regex_unref (regex);
1725 * if (error != NULL)
1727 * g_printerr ("Error while matching: %s\n", error->message);
1728 * g_error_free (error);
1733 * Returns: %TRUE is the string matched, %FALSE otherwise
1738 g_regex_match_full (const GRegex *regex,
1739 const gchar *string,
1741 gint start_position,
1742 GRegexMatchFlags match_options,
1743 GMatchInfo **match_info,
1749 g_return_val_if_fail (regex != NULL, FALSE);
1750 g_return_val_if_fail (string != NULL, FALSE);
1751 g_return_val_if_fail (start_position >= 0, FALSE);
1752 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
1753 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, FALSE);
1755 info = match_info_new (regex, string, string_len, start_position,
1756 match_options, FALSE);
1757 match_ok = g_match_info_next (info, error);
1758 if (match_info != NULL)
1761 g_match_info_free (info);
1767 * g_regex_match_all:
1768 * @regex: a #GRegex structure from g_regex_new()
1769 * @string: the string to scan for matches
1770 * @match_options: match options
1771 * @match_info: (out) (allow-none): pointer to location where to store
1772 * the #GMatchInfo, or %NULL if you do not need it
1774 * Using the standard algorithm for regular expression matching only
1775 * the longest match in the string is retrieved. This function uses
1776 * a different algorithm so it can retrieve all the possible matches.
1777 * For more documentation see g_regex_match_all_full().
1779 * A #GMatchInfo structure, used to get information on the match, is
1780 * stored in @match_info if not %NULL. Note that if @match_info is
1781 * not %NULL then it is created even if the function returns %FALSE,
1782 * i.e. you must free it regardless if regular expression actually
1785 * @string is not copied and is used in #GMatchInfo internally. If
1786 * you use any #GMatchInfo method (except g_match_info_free()) after
1787 * freeing or modifying @string then the behaviour is undefined.
1789 * Returns: %TRUE is the string matched, %FALSE otherwise
1794 g_regex_match_all (const GRegex *regex,
1795 const gchar *string,
1796 GRegexMatchFlags match_options,
1797 GMatchInfo **match_info)
1799 return g_regex_match_all_full (regex, string, -1, 0, match_options,
1804 * g_regex_match_all_full:
1805 * @regex: a #GRegex structure from g_regex_new()
1806 * @string: (array length=string_len): the string to scan for matches
1807 * @string_len: the length of @string, or -1 if @string is nul-terminated
1808 * @start_position: starting index of the string to match
1809 * @match_options: match options
1810 * @match_info: (out) (allow-none): pointer to location where to store
1811 * the #GMatchInfo, or %NULL if you do not need it
1812 * @error: location to store the error occurring, or %NULL to ignore errors
1814 * Using the standard algorithm for regular expression matching only
1815 * the longest match in the string is retrieved, it is not possible
1816 * to obtain all the available matches. For instance matching
1817 * "<a> <b> <c>" against the pattern "<.*>"
1818 * you get "<a> <b> <c>".
1820 * This function uses a different algorithm (called DFA, i.e. deterministic
1821 * finite automaton), so it can retrieve all the possible matches, all
1822 * starting at the same point in the string. For instance matching
1823 * "<a> <b> <c>" against the pattern "<.*>"
1824 * you would obtain three matches: "<a> <b> <c>",
1825 * "<a> <b>" and "<a>".
1827 * The number of matched strings is retrieved using
1828 * g_match_info_get_match_count(). To obtain the matched strings and
1829 * their position you can use, respectively, g_match_info_fetch() and
1830 * g_match_info_fetch_pos(). Note that the strings are returned in
1831 * reverse order of length; that is, the longest matching string is
1834 * Note that the DFA algorithm is slower than the standard one and it
1835 * is not able to capture substrings, so backreferences do not work.
1837 * Setting @start_position differs from just passing over a shortened
1838 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
1839 * that begins with any kind of lookbehind assertion, such as "\b".
1841 * A #GMatchInfo structure, used to get information on the match, is
1842 * stored in @match_info if not %NULL. Note that if @match_info is
1843 * not %NULL then it is created even if the function returns %FALSE,
1844 * i.e. you must free it regardless if regular expression actually
1847 * @string is not copied and is used in #GMatchInfo internally. If
1848 * you use any #GMatchInfo method (except g_match_info_free()) after
1849 * freeing or modifying @string then the behaviour is undefined.
1851 * Returns: %TRUE is the string matched, %FALSE otherwise
1856 g_regex_match_all_full (const GRegex *regex,
1857 const gchar *string,
1859 gint start_position,
1860 GRegexMatchFlags match_options,
1861 GMatchInfo **match_info,
1867 g_return_val_if_fail (regex != NULL, FALSE);
1868 g_return_val_if_fail (string != NULL, FALSE);
1869 g_return_val_if_fail (start_position >= 0, FALSE);
1870 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
1871 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, FALSE);
1873 info = match_info_new (regex, string, string_len, start_position,
1874 match_options, TRUE);
1880 info->matches = pcre_dfa_exec (regex->pcre_re, regex->extra,
1881 info->string, info->string_len,
1883 regex->match_opts | match_options,
1884 info->offsets, info->n_offsets,
1885 info->workspace, info->n_workspace);
1886 if (info->matches == PCRE_ERROR_DFA_WSSIZE)
1888 /* info->workspace is too small. */
1889 info->n_workspace *= 2;
1890 info->workspace = g_realloc (info->workspace,
1891 info->n_workspace * sizeof (gint));
1894 else if (info->matches == 0)
1896 /* info->offsets is too small. */
1897 info->n_offsets *= 2;
1898 info->offsets = g_realloc (info->offsets,
1899 info->n_offsets * sizeof (gint));
1902 else if (IS_PCRE_ERROR (info->matches))
1904 g_set_error (error, G_REGEX_ERROR, G_REGEX_ERROR_MATCH,
1905 _("Error while matching regular expression %s: %s"),
1906 regex->pattern, match_error (info->matches));
1910 /* set info->pos to -1 so that a call to g_match_info_next() fails. */
1913 if (match_info != NULL)
1916 g_match_info_free (info);
1918 return info->matches >= 0;
1922 * g_regex_get_string_number:
1923 * @regex: #GRegex structure
1924 * @name: name of the subexpression
1926 * Retrieves the number of the subexpression named @name.
1928 * Returns: The number of the subexpression or -1 if @name
1934 g_regex_get_string_number (const GRegex *regex,
1939 g_return_val_if_fail (regex != NULL, -1);
1940 g_return_val_if_fail (name != NULL, -1);
1942 num = pcre_get_stringnumber (regex->pcre_re, name);
1943 if (num == PCRE_ERROR_NOSUBSTRING)
1950 * g_regex_split_simple:
1951 * @pattern: the regular expression
1952 * @string: the string to scan for matches
1953 * @compile_options: compile options for the regular expression, or 0
1954 * @match_options: match options, or 0
1956 * Breaks the string on the pattern, and returns an array of
1957 * the tokens. If the pattern contains capturing parentheses,
1958 * then the text for each of the substrings will also be returned.
1959 * If the pattern does not match anywhere in the string, then the
1960 * whole string is returned as the first token.
1962 * This function is equivalent to g_regex_split() but it does
1963 * not require to compile the pattern with g_regex_new(), avoiding
1964 * some lines of code when you need just to do a split without
1965 * extracting substrings, capture counts, and so on.
1967 * If this function is to be called on the same @pattern more than
1968 * once, it's more efficient to compile the pattern once with
1969 * g_regex_new() and then use g_regex_split().
1971 * As a special case, the result of splitting the empty string ""
1972 * is an empty vector, not a vector containing a single string.
1973 * The reason for this special case is that being able to represent
1974 * a empty vector is typically more useful than consistent handling
1975 * of empty elements. If you do need to represent empty elements,
1976 * you'll need to check for the empty string before calling this
1979 * A pattern that can match empty strings splits @string into
1980 * separate characters wherever it matches the empty string between
1981 * characters. For example splitting "ab c" using as a separator
1982 * "\s*", you will get "a", "b" and "c".
1984 * Returns: a %NULL-terminated array of strings. Free it using g_strfreev()
1989 g_regex_split_simple (const gchar *pattern,
1990 const gchar *string,
1991 GRegexCompileFlags compile_options,
1992 GRegexMatchFlags match_options)
1997 regex = g_regex_new (pattern, compile_options, 0, NULL);
2001 result = g_regex_split_full (regex, string, -1, 0, match_options, 0, NULL);
2002 g_regex_unref (regex);
2008 * @regex: a #GRegex structure
2009 * @string: the string to split with the pattern
2010 * @match_options: match time option flags
2012 * Breaks the string on the pattern, and returns an array of the tokens.
2013 * If the pattern contains capturing parentheses, then the text for each
2014 * of the substrings will also be returned. If the pattern does not match
2015 * anywhere in the string, then the whole string is returned as the first
2018 * As a special case, the result of splitting the empty string "" is an
2019 * empty vector, not a vector containing a single string. The reason for
2020 * this special case is that being able to represent a empty vector is
2021 * typically more useful than consistent handling of empty elements. If
2022 * you do need to represent empty elements, you'll need to check for the
2023 * empty string before calling this function.
2025 * A pattern that can match empty strings splits @string into separate
2026 * characters wherever it matches the empty string between characters.
2027 * For example splitting "ab c" using as a separator "\s*", you will get
2030 * Returns: a %NULL-terminated gchar ** array. Free it using g_strfreev()
2035 g_regex_split (const GRegex *regex,
2036 const gchar *string,
2037 GRegexMatchFlags match_options)
2039 return g_regex_split_full (regex, string, -1, 0,
2040 match_options, 0, NULL);
2044 * g_regex_split_full:
2045 * @regex: a #GRegex structure
2046 * @string: (array length=string_len): the string to split with the pattern
2047 * @string_len: the length of @string, or -1 if @string is nul-terminated
2048 * @start_position: starting index of the string to match
2049 * @match_options: match time option flags
2050 * @max_tokens: the maximum number of tokens to split @string into.
2051 * If this is less than 1, the string is split completely
2052 * @error: return location for a #GError
2054 * Breaks the string on the pattern, and returns an array of the tokens.
2055 * If the pattern contains capturing parentheses, then the text for each
2056 * of the substrings will also be returned. If the pattern does not match
2057 * anywhere in the string, then the whole string is returned as the first
2060 * As a special case, the result of splitting the empty string "" is an
2061 * empty vector, not a vector containing a single string. The reason for
2062 * this special case is that being able to represent a empty vector is
2063 * typically more useful than consistent handling of empty elements. If
2064 * you do need to represent empty elements, you'll need to check for the
2065 * empty string before calling this function.
2067 * A pattern that can match empty strings splits @string into separate
2068 * characters wherever it matches the empty string between characters.
2069 * For example splitting "ab c" using as a separator "\s*", you will get
2072 * Setting @start_position differs from just passing over a shortened
2073 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
2074 * that begins with any kind of lookbehind assertion, such as "\b".
2076 * Returns: a %NULL-terminated gchar ** array. Free it using g_strfreev()
2081 g_regex_split_full (const GRegex *regex,
2082 const gchar *string,
2084 gint start_position,
2085 GRegexMatchFlags match_options,
2089 GError *tmp_error = NULL;
2090 GMatchInfo *match_info;
2095 /* position of the last separator. */
2096 gint last_separator_end;
2097 /* was the last match 0 bytes long? */
2098 gboolean last_match_is_empty;
2099 /* the returned array of char **s */
2100 gchar **string_list;
2102 g_return_val_if_fail (regex != NULL, NULL);
2103 g_return_val_if_fail (string != NULL, NULL);
2104 g_return_val_if_fail (start_position >= 0, NULL);
2105 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
2106 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2108 if (max_tokens <= 0)
2109 max_tokens = G_MAXINT;
2112 string_len = strlen (string);
2114 /* zero-length string */
2115 if (string_len - start_position == 0)
2116 return g_new0 (gchar *, 1);
2118 if (max_tokens == 1)
2120 string_list = g_new0 (gchar *, 2);
2121 string_list[0] = g_strndup (&string[start_position],
2122 string_len - start_position);
2128 last_separator_end = start_position;
2129 last_match_is_empty = FALSE;
2131 match_ok = g_regex_match_full (regex, string, string_len, start_position,
2132 match_options, &match_info, &tmp_error);
2134 while (tmp_error == NULL)
2138 last_match_is_empty =
2139 (match_info->offsets[0] == match_info->offsets[1]);
2141 /* we need to skip empty separators at the same position of the end
2142 * of another separator. e.g. the string is "a b" and the separator
2143 * is " *", so from 1 to 2 we have a match and at position 2 we have
2144 * an empty match. */
2145 if (last_separator_end != match_info->offsets[1])
2150 token = g_strndup (string + last_separator_end,
2151 match_info->offsets[0] - last_separator_end);
2152 list = g_list_prepend (list, token);
2155 /* if there were substrings, these need to be added to
2157 match_count = g_match_info_get_match_count (match_info);
2158 if (match_count > 1)
2160 for (i = 1; i < match_count; i++)
2161 list = g_list_prepend (list, g_match_info_fetch (match_info, i));
2167 /* if there was no match, copy to end of string. */
2168 if (!last_match_is_empty)
2170 gchar *token = g_strndup (string + last_separator_end,
2171 match_info->string_len - last_separator_end);
2172 list = g_list_prepend (list, token);
2174 /* no more tokens, end the loop. */
2178 /* -1 to leave room for the last part. */
2179 if (token_count >= max_tokens - 1)
2181 /* we have reached the maximum number of tokens, so we copy
2182 * the remaining part of the string. */
2183 if (last_match_is_empty)
2185 /* the last match was empty, so we have moved one char
2186 * after the real position to avoid empty matches at the
2188 match_info->pos = PREV_CHAR (regex, &string[match_info->pos]) - string;
2190 /* the if is needed in the case we have terminated the available
2191 * tokens, but we are at the end of the string, so there are no
2192 * characters left to copy. */
2193 if (string_len > match_info->pos)
2195 gchar *token = g_strndup (string + match_info->pos,
2196 string_len - match_info->pos);
2197 list = g_list_prepend (list, token);
2203 last_separator_end = match_info->pos;
2204 if (last_match_is_empty)
2205 /* if the last match was empty, g_match_info_next() has moved
2206 * forward to avoid infinite loops, but we still need to copy that
2208 last_separator_end = PREV_CHAR (regex, &string[last_separator_end]) - string;
2210 match_ok = g_match_info_next (match_info, &tmp_error);
2212 g_match_info_free (match_info);
2213 if (tmp_error != NULL)
2215 g_propagate_error (error, tmp_error);
2216 g_list_free_full (list, g_free);
2217 match_info->pos = -1;
2221 string_list = g_new (gchar *, g_list_length (list) + 1);
2223 for (last = g_list_last (list); last; last = g_list_previous (last))
2224 string_list[i++] = last->data;
2225 string_list[i] = NULL;
2234 REPL_TYPE_CHARACTER,
2235 REPL_TYPE_SYMBOLIC_REFERENCE,
2236 REPL_TYPE_NUMERIC_REFERENCE,
2237 REPL_TYPE_CHANGE_CASE
2242 CHANGE_CASE_NONE = 1 << 0,
2243 CHANGE_CASE_UPPER = 1 << 1,
2244 CHANGE_CASE_LOWER = 1 << 2,
2245 CHANGE_CASE_UPPER_SINGLE = 1 << 3,
2246 CHANGE_CASE_LOWER_SINGLE = 1 << 4,
2247 CHANGE_CASE_SINGLE_MASK = CHANGE_CASE_UPPER_SINGLE | CHANGE_CASE_LOWER_SINGLE,
2248 CHANGE_CASE_LOWER_MASK = CHANGE_CASE_LOWER | CHANGE_CASE_LOWER_SINGLE,
2249 CHANGE_CASE_UPPER_MASK = CHANGE_CASE_UPPER | CHANGE_CASE_UPPER_SINGLE
2252 struct _InterpolationData
2258 ChangeCase change_case;
2262 free_interpolation_data (InterpolationData *data)
2264 g_free (data->text);
2268 static const gchar *
2269 expand_escape (const gchar *replacement,
2271 InterpolationData *data,
2276 const gchar *error_detail;
2278 GError *tmp_error = NULL;
2286 data->type = REPL_TYPE_CHARACTER;
2291 data->type = REPL_TYPE_CHARACTER;
2296 data->type = REPL_TYPE_CHARACTER;
2301 data->type = REPL_TYPE_CHARACTER;
2306 data->type = REPL_TYPE_CHARACTER;
2311 data->type = REPL_TYPE_CHARACTER;
2316 data->type = REPL_TYPE_CHARACTER;
2321 data->type = REPL_TYPE_CHARACTER;
2331 h = g_ascii_xdigit_value (*p);
2334 error_detail = _("hexadecimal digit or '}' expected");
2345 for (i = 0; i < 2; i++)
2347 h = g_ascii_xdigit_value (*p);
2350 error_detail = _("hexadecimal digit expected");
2357 data->type = REPL_TYPE_STRING;
2358 data->text = g_new0 (gchar, 8);
2359 g_unichar_to_utf8 (x, data->text);
2363 data->type = REPL_TYPE_CHANGE_CASE;
2364 data->change_case = CHANGE_CASE_LOWER_SINGLE;
2368 data->type = REPL_TYPE_CHANGE_CASE;
2369 data->change_case = CHANGE_CASE_UPPER_SINGLE;
2373 data->type = REPL_TYPE_CHANGE_CASE;
2374 data->change_case = CHANGE_CASE_LOWER;
2378 data->type = REPL_TYPE_CHANGE_CASE;
2379 data->change_case = CHANGE_CASE_UPPER;
2383 data->type = REPL_TYPE_CHANGE_CASE;
2384 data->change_case = CHANGE_CASE_NONE;
2390 error_detail = _("missing '<' in symbolic reference");
2399 error_detail = _("unfinished symbolic reference");
2406 error_detail = _("zero-length symbolic reference");
2409 if (g_ascii_isdigit (*q))
2414 h = g_ascii_digit_value (*q);
2417 error_detail = _("digit expected");
2426 data->type = REPL_TYPE_NUMERIC_REFERENCE;
2433 if (!g_ascii_isalnum (*r))
2435 error_detail = _("illegal symbolic reference");
2442 data->text = g_strndup (q, p - q);
2443 data->type = REPL_TYPE_SYMBOLIC_REFERENCE;
2448 /* if \0 is followed by a number is an octal number representing a
2449 * character, else it is a numeric reference. */
2450 if (g_ascii_digit_value (*g_utf8_next_char (p)) >= 0)
2453 p = g_utf8_next_char (p);
2466 for (i = 0; i < 3; i++)
2468 h = g_ascii_digit_value (*p);
2478 if (i == 2 && base == 10)
2484 if (base == 8 || i == 3)
2486 data->type = REPL_TYPE_STRING;
2487 data->text = g_new0 (gchar, 8);
2488 g_unichar_to_utf8 (x, data->text);
2492 data->type = REPL_TYPE_NUMERIC_REFERENCE;
2497 error_detail = _("stray final '\\'");
2501 error_detail = _("unknown escape sequence");
2508 /* G_GSSIZE_FORMAT doesn't work with gettext, so we use %lu */
2509 tmp_error = g_error_new (G_REGEX_ERROR,
2510 G_REGEX_ERROR_REPLACE,
2511 _("Error while parsing replacement "
2512 "text \"%s\" at char %lu: %s"),
2514 (gulong)(p - replacement),
2516 g_propagate_error (error, tmp_error);
2522 split_replacement (const gchar *replacement,
2526 InterpolationData *data;
2527 const gchar *p, *start;
2529 start = p = replacement;
2534 data = g_new0 (InterpolationData, 1);
2535 start = p = expand_escape (replacement, p, data, error);
2538 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
2539 free_interpolation_data (data);
2543 list = g_list_prepend (list, data);
2548 if (*p == '\\' || *p == '\0')
2552 data = g_new0 (InterpolationData, 1);
2553 data->text = g_strndup (start, p - start);
2554 data->type = REPL_TYPE_STRING;
2555 list = g_list_prepend (list, data);
2561 return g_list_reverse (list);
2564 /* Change the case of c based on change_case. */
2565 #define CHANGE_CASE(c, change_case) \
2566 (((change_case) & CHANGE_CASE_LOWER_MASK) ? \
2567 g_unichar_tolower (c) : \
2568 g_unichar_toupper (c))
2571 string_append (GString *string,
2573 ChangeCase *change_case)
2577 if (text[0] == '\0')
2580 if (*change_case == CHANGE_CASE_NONE)
2582 g_string_append (string, text);
2584 else if (*change_case & CHANGE_CASE_SINGLE_MASK)
2586 c = g_utf8_get_char (text);
2587 g_string_append_unichar (string, CHANGE_CASE (c, *change_case));
2588 g_string_append (string, g_utf8_next_char (text));
2589 *change_case = CHANGE_CASE_NONE;
2593 while (*text != '\0')
2595 c = g_utf8_get_char (text);
2596 g_string_append_unichar (string, CHANGE_CASE (c, *change_case));
2597 text = g_utf8_next_char (text);
2603 interpolate_replacement (const GMatchInfo *match_info,
2608 InterpolationData *idata;
2610 ChangeCase change_case = CHANGE_CASE_NONE;
2612 for (list = data; list; list = list->next)
2615 switch (idata->type)
2617 case REPL_TYPE_STRING:
2618 string_append (result, idata->text, &change_case);
2620 case REPL_TYPE_CHARACTER:
2621 g_string_append_c (result, CHANGE_CASE (idata->c, change_case));
2622 if (change_case & CHANGE_CASE_SINGLE_MASK)
2623 change_case = CHANGE_CASE_NONE;
2625 case REPL_TYPE_NUMERIC_REFERENCE:
2626 match = g_match_info_fetch (match_info, idata->num);
2629 string_append (result, match, &change_case);
2633 case REPL_TYPE_SYMBOLIC_REFERENCE:
2634 match = g_match_info_fetch_named (match_info, idata->text);
2637 string_append (result, match, &change_case);
2641 case REPL_TYPE_CHANGE_CASE:
2642 change_case = idata->change_case;
2650 /* whether actual match_info is needed for replacement, i.e.
2651 * whether there are references
2654 interpolation_list_needs_match (GList *list)
2656 while (list != NULL)
2658 InterpolationData *data = list->data;
2660 if (data->type == REPL_TYPE_SYMBOLIC_REFERENCE ||
2661 data->type == REPL_TYPE_NUMERIC_REFERENCE)
2674 * @regex: a #GRegex structure
2675 * @string: (array length=string_len): the string to perform matches against
2676 * @string_len: the length of @string, or -1 if @string is nul-terminated
2677 * @start_position: starting index of the string to match
2678 * @replacement: text to replace each match with
2679 * @match_options: options for the match
2680 * @error: location to store the error occurring, or %NULL to ignore errors
2682 * Replaces all occurrences of the pattern in @regex with the
2683 * replacement text. Backreferences of the form '\number' or
2684 * '\g<number>' in the replacement text are interpolated by the
2685 * number-th captured subexpression of the match, '\g<name>' refers
2686 * to the captured subexpression with the given name. '\0' refers to the
2687 * complete match, but '\0' followed by a number is the octal representation
2688 * of a character. To include a literal '\' in the replacement, write '\\'.
2689 * There are also escapes that changes the case of the following text:
2692 * <varlistentry><term>\l</term>
2694 * <para>Convert to lower case the next character</para>
2697 * <varlistentry><term>\u</term>
2699 * <para>Convert to upper case the next character</para>
2702 * <varlistentry><term>\L</term>
2704 * <para>Convert to lower case till \E</para>
2707 * <varlistentry><term>\U</term>
2709 * <para>Convert to upper case till \E</para>
2712 * <varlistentry><term>\E</term>
2714 * <para>End case modification</para>
2719 * If you do not need to use backreferences use g_regex_replace_literal().
2721 * The @replacement string must be UTF-8 encoded even if #G_REGEX_RAW was
2722 * passed to g_regex_new(). If you want to use not UTF-8 encoded stings
2723 * you can use g_regex_replace_literal().
2725 * Setting @start_position differs from just passing over a shortened
2726 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern that
2727 * begins with any kind of lookbehind assertion, such as "\b".
2729 * Returns: a newly allocated string containing the replacements
2734 g_regex_replace (const GRegex *regex,
2735 const gchar *string,
2737 gint start_position,
2738 const gchar *replacement,
2739 GRegexMatchFlags match_options,
2744 GError *tmp_error = NULL;
2746 g_return_val_if_fail (regex != NULL, NULL);
2747 g_return_val_if_fail (string != NULL, NULL);
2748 g_return_val_if_fail (start_position >= 0, NULL);
2749 g_return_val_if_fail (replacement != NULL, NULL);
2750 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
2751 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2753 list = split_replacement (replacement, &tmp_error);
2754 if (tmp_error != NULL)
2756 g_propagate_error (error, tmp_error);
2760 result = g_regex_replace_eval (regex,
2761 string, string_len, start_position,
2763 interpolate_replacement,
2766 if (tmp_error != NULL)
2767 g_propagate_error (error, tmp_error);
2769 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
2775 literal_replacement (const GMatchInfo *match_info,
2779 g_string_append (result, data);
2784 * g_regex_replace_literal:
2785 * @regex: a #GRegex structure
2786 * @string: (array length=string_len): the string to perform matches against
2787 * @string_len: the length of @string, or -1 if @string is nul-terminated
2788 * @start_position: starting index of the string to match
2789 * @replacement: text to replace each match with
2790 * @match_options: options for the match
2791 * @error: location to store the error occurring, or %NULL to ignore errors
2793 * Replaces all occurrences of the pattern in @regex with the
2794 * replacement text. @replacement is replaced literally, to
2795 * include backreferences use g_regex_replace().
2797 * Setting @start_position differs from just passing over a
2798 * shortened string and setting #G_REGEX_MATCH_NOTBOL in the
2799 * case of a pattern that begins with any kind of lookbehind
2800 * assertion, such as "\b".
2802 * Returns: a newly allocated string containing the replacements
2807 g_regex_replace_literal (const GRegex *regex,
2808 const gchar *string,
2810 gint start_position,
2811 const gchar *replacement,
2812 GRegexMatchFlags match_options,
2815 g_return_val_if_fail (replacement != NULL, NULL);
2816 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2818 return g_regex_replace_eval (regex,
2819 string, string_len, start_position,
2821 literal_replacement,
2822 (gpointer)replacement,
2827 * g_regex_replace_eval:
2828 * @regex: a #GRegex structure from g_regex_new()
2829 * @string: (array length=string_len): string to perform matches against
2830 * @string_len: the length of @string, or -1 if @string is nul-terminated
2831 * @start_position: starting index of the string to match
2832 * @match_options: options for the match
2833 * @eval: a function to call for each match
2834 * @user_data: user data to pass to the function
2835 * @error: location to store the error occurring, or %NULL to ignore errors
2837 * Replaces occurrences of the pattern in regex with the output of
2838 * @eval for that occurrence.
2840 * Setting @start_position differs from just passing over a shortened
2841 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
2842 * that begins with any kind of lookbehind assertion, such as "\b".
2844 * The following example uses g_regex_replace_eval() to replace multiple
2848 * eval_cb (const GMatchInfo *info,
2855 * match = g_match_info_fetch (info, 0);
2856 * r = g_hash_table_lookup ((GHashTable *)data, match);
2857 * g_string_append (res, r);
2869 * h = g_hash_table_new (g_str_hash, g_str_equal);
2871 * g_hash_table_insert (h, "1", "ONE");
2872 * g_hash_table_insert (h, "2", "TWO");
2873 * g_hash_table_insert (h, "3", "THREE");
2874 * g_hash_table_insert (h, "4", "FOUR");
2876 * reg = g_regex_new ("1|2|3|4", 0, 0, NULL);
2877 * res = g_regex_replace_eval (reg, text, -1, 0, 0, eval_cb, h, NULL);
2878 * g_hash_table_destroy (h);
2883 * Returns: a newly allocated string containing the replacements
2888 g_regex_replace_eval (const GRegex *regex,
2889 const gchar *string,
2891 gint start_position,
2892 GRegexMatchFlags match_options,
2893 GRegexEvalCallback eval,
2897 GMatchInfo *match_info;
2900 gboolean done = FALSE;
2901 GError *tmp_error = NULL;
2903 g_return_val_if_fail (regex != NULL, NULL);
2904 g_return_val_if_fail (string != NULL, NULL);
2905 g_return_val_if_fail (start_position >= 0, NULL);
2906 g_return_val_if_fail (eval != NULL, NULL);
2907 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2910 string_len = strlen (string);
2912 result = g_string_sized_new (string_len);
2914 /* run down the string making matches. */
2915 g_regex_match_full (regex, string, string_len, start_position,
2916 match_options, &match_info, &tmp_error);
2917 while (!done && g_match_info_matches (match_info))
2919 g_string_append_len (result,
2921 match_info->offsets[0] - str_pos);
2922 done = (*eval) (match_info, result, user_data);
2923 str_pos = match_info->offsets[1];
2924 g_match_info_next (match_info, &tmp_error);
2926 g_match_info_free (match_info);
2927 if (tmp_error != NULL)
2929 g_propagate_error (error, tmp_error);
2930 g_string_free (result, TRUE);
2934 g_string_append_len (result, string + str_pos, string_len - str_pos);
2935 return g_string_free (result, FALSE);
2939 * g_regex_check_replacement:
2940 * @replacement: the replacement string
2941 * @has_references: (out) (allow-none): location to store information about
2942 * references in @replacement or %NULL
2943 * @error: location to store error
2945 * Checks whether @replacement is a valid replacement string
2946 * (see g_regex_replace()), i.e. that all escape sequences in
2949 * If @has_references is not %NULL then @replacement is checked
2950 * for pattern references. For instance, replacement text 'foo\n'
2951 * does not contain references and may be evaluated without information
2952 * about actual match, but '\0\1' (whole match followed by first
2953 * subpattern) requires valid #GMatchInfo object.
2955 * Returns: whether @replacement is a valid replacement string
2960 g_regex_check_replacement (const gchar *replacement,
2961 gboolean *has_references,
2967 list = split_replacement (replacement, &tmp);
2971 g_propagate_error (error, tmp);
2976 *has_references = interpolation_list_needs_match (list);
2978 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
2984 * g_regex_escape_nul:
2985 * @string: the string to escape
2986 * @length: the length of @string
2988 * Escapes the nul characters in @string to "\x00". It can be used
2989 * to compile a regex with embedded nul characters.
2991 * For completeness, @length can be -1 for a nul-terminated string.
2992 * In this case the output string will be of course equal to @string.
2994 * Returns: a newly-allocated escaped string
2999 g_regex_escape_nul (const gchar *string,
3003 const gchar *p, *piece_start, *end;
3006 g_return_val_if_fail (string != NULL, NULL);
3009 return g_strdup (string);
3011 end = string + length;
3012 p = piece_start = string;
3013 escaped = g_string_sized_new (length + 1);
3021 if (p != piece_start)
3023 /* copy the previous piece. */
3024 g_string_append_len (escaped, piece_start, p - piece_start);
3026 if ((backslashes & 1) == 0)
3027 g_string_append_c (escaped, '\\');
3028 g_string_append_c (escaped, 'x');
3029 g_string_append_c (escaped, '0');
3030 g_string_append_c (escaped, '0');
3040 p = g_utf8_next_char (p);
3045 if (piece_start < end)
3046 g_string_append_len (escaped, piece_start, end - piece_start);
3048 return g_string_free (escaped, FALSE);
3052 * g_regex_escape_string:
3053 * @string: (array length=length): the string to escape
3054 * @length: the length of @string, or -1 if @string is nul-terminated
3056 * Escapes the special characters used for regular expressions
3057 * in @string, for instance "a.b*c" becomes "a\.b\*c". This
3058 * function is useful to dynamically generate regular expressions.
3060 * @string can contain nul characters that are replaced with "\0",
3061 * in this case remember to specify the correct length of @string
3064 * Returns: a newly-allocated escaped string
3069 g_regex_escape_string (const gchar *string,
3073 const char *p, *piece_start, *end;
3075 g_return_val_if_fail (string != NULL, NULL);
3078 length = strlen (string);
3080 end = string + length;
3081 p = piece_start = string;
3082 escaped = g_string_sized_new (length + 1);
3103 if (p != piece_start)
3104 /* copy the previous piece. */
3105 g_string_append_len (escaped, piece_start, p - piece_start);
3106 g_string_append_c (escaped, '\\');
3108 g_string_append_c (escaped, '0');
3110 g_string_append_c (escaped, *p);
3114 p = g_utf8_next_char (p);
3119 if (piece_start < end)
3120 g_string_append_len (escaped, piece_start, end - piece_start);
3122 return g_string_free (escaped, FALSE);