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)");
521 case G_REGEX_ERROR_CHARACTER_VALUE_TOO_LARGE:
522 *errmsg = _("character value in \\u.... sequence is too large");
525 case 116: /* erroffset passed as NULL */
526 /* This should not happen as we never pass a NULL erroffset */
527 g_warning ("erroffset passed as NULL");
528 *errcode = G_REGEX_ERROR_COMPILE;
530 case 117: /* unknown option bit(s) set */
531 /* This should not happen as we check options before passing them
532 * to pcre_compile2() */
533 g_warning ("unknown option bit(s) set");
534 *errcode = G_REGEX_ERROR_COMPILE;
536 case 132: /* this version of PCRE is compiled without UTF support */
537 case 144: /* invalid UTF-8 string */
538 case 145: /* support for \\P, \\p, and \\X has not been compiled */
539 case 167: /* this version of PCRE is not compiled with Unicode property support */
540 case 173: /* disallowed Unicode code point (>= 0xd800 && <= 0xdfff) */
541 case 174: /* invalid UTF-16 string */
542 /* These errors should not happen as we are using an UTF-8 and UCP-enabled PCRE
543 * and we do not check if strings are valid */
544 case 170: /* internal error: unknown opcode in find_fixedlength() */
545 *errcode = G_REGEX_ERROR_INTERNAL;
549 *errcode = G_REGEX_ERROR_COMPILE;
556 match_info_new (const GRegex *regex,
563 GMatchInfo *match_info;
566 string_len = strlen (string);
568 match_info = g_new0 (GMatchInfo, 1);
569 match_info->ref_count = 1;
570 match_info->regex = g_regex_ref ((GRegex *)regex);
571 match_info->string = string;
572 match_info->string_len = string_len;
573 match_info->matches = PCRE_ERROR_NOMATCH;
574 match_info->pos = start_position;
575 match_info->match_opts = match_options;
579 /* These values should be enough for most cases, if they are not
580 * enough g_regex_match_all_full() will expand them. */
581 match_info->n_offsets = 24;
582 match_info->n_workspace = 100;
583 match_info->workspace = g_new (gint, match_info->n_workspace);
588 pcre_fullinfo (regex->pcre_re, regex->extra,
589 PCRE_INFO_CAPTURECOUNT, &capture_count);
590 match_info->n_offsets = (capture_count + 1) * 3;
593 match_info->offsets = g_new0 (gint, match_info->n_offsets);
594 /* Set an invalid position for the previous match. */
595 match_info->offsets[0] = -1;
596 match_info->offsets[1] = -1;
602 * g_match_info_get_regex:
603 * @match_info: a #GMatchInfo
605 * Returns #GRegex object used in @match_info. It belongs to Glib
606 * and must not be freed. Use g_regex_ref() if you need to keep it
607 * after you free @match_info object.
609 * Returns: #GRegex object used in @match_info
614 g_match_info_get_regex (const GMatchInfo *match_info)
616 g_return_val_if_fail (match_info != NULL, NULL);
617 return match_info->regex;
621 * g_match_info_get_string:
622 * @match_info: a #GMatchInfo
624 * Returns the string searched with @match_info. This is the
625 * string passed to g_regex_match() or g_regex_replace() so
626 * you may not free it before calling this function.
628 * Returns: the string searched with @match_info
633 g_match_info_get_string (const GMatchInfo *match_info)
635 g_return_val_if_fail (match_info != NULL, NULL);
636 return match_info->string;
641 * @match_info: a #GMatchInfo
643 * Increases reference count of @match_info by 1.
645 * Returns: @match_info
650 g_match_info_ref (GMatchInfo *match_info)
652 g_return_val_if_fail (match_info != NULL, NULL);
653 g_atomic_int_inc (&match_info->ref_count);
658 * g_match_info_unref:
659 * @match_info: a #GMatchInfo
661 * Decreases reference count of @match_info by 1. When reference count drops
662 * to zero, it frees all the memory associated with the match_info structure.
667 g_match_info_unref (GMatchInfo *match_info)
669 if (g_atomic_int_dec_and_test (&match_info->ref_count))
671 g_regex_unref (match_info->regex);
672 g_free (match_info->offsets);
673 g_free (match_info->workspace);
680 * @match_info: (allow-none): a #GMatchInfo, or %NULL
682 * If @match_info is not %NULL, calls g_match_info_unref(); otherwise does
688 g_match_info_free (GMatchInfo *match_info)
690 if (match_info == NULL)
693 g_match_info_unref (match_info);
698 * @match_info: a #GMatchInfo structure
699 * @error: location to store the error occurring, or %NULL to ignore errors
701 * Scans for the next match using the same parameters of the previous
702 * call to g_regex_match_full() or g_regex_match() that returned
705 * The match is done on the string passed to the match function, so you
706 * cannot free it before calling this function.
708 * Returns: %TRUE is the string matched, %FALSE otherwise
713 g_match_info_next (GMatchInfo *match_info,
716 gint prev_match_start;
719 g_return_val_if_fail (match_info != NULL, FALSE);
720 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
721 g_return_val_if_fail (match_info->pos >= 0, FALSE);
723 prev_match_start = match_info->offsets[0];
724 prev_match_end = match_info->offsets[1];
726 if (match_info->pos > match_info->string_len)
728 /* we have reached the end of the string */
729 match_info->pos = -1;
730 match_info->matches = PCRE_ERROR_NOMATCH;
734 match_info->matches = pcre_exec (match_info->regex->pcre_re,
735 match_info->regex->extra,
737 match_info->string_len,
739 match_info->regex->match_opts | match_info->match_opts,
741 match_info->n_offsets);
742 if (IS_PCRE_ERROR (match_info->matches))
744 g_set_error (error, G_REGEX_ERROR, G_REGEX_ERROR_MATCH,
745 _("Error while matching regular expression %s: %s"),
746 match_info->regex->pattern, match_error (match_info->matches));
750 /* avoid infinite loops if the pattern is an empty string or something
752 if (match_info->pos == match_info->offsets[1])
754 if (match_info->pos > match_info->string_len)
756 /* we have reached the end of the string */
757 match_info->pos = -1;
758 match_info->matches = PCRE_ERROR_NOMATCH;
762 match_info->pos = NEXT_CHAR (match_info->regex,
763 &match_info->string[match_info->pos]) -
768 match_info->pos = match_info->offsets[1];
771 /* it's possible to get two identical matches when we are matching
772 * empty strings, for instance if the pattern is "(?=[A-Z0-9])" and
773 * the string is "RegExTest" we have:
774 * - search at position 0: match from 0 to 0
775 * - search at position 1: match from 3 to 3
776 * - search at position 3: match from 3 to 3 (duplicate)
777 * - search at position 4: match from 5 to 5
778 * - search at position 5: match from 5 to 5 (duplicate)
779 * - search at position 6: no match -> stop
780 * so we have to ignore the duplicates.
781 * see bug #515944: http://bugzilla.gnome.org/show_bug.cgi?id=515944 */
782 if (match_info->matches >= 0 &&
783 prev_match_start == match_info->offsets[0] &&
784 prev_match_end == match_info->offsets[1])
786 /* ignore this match and search the next one */
787 return g_match_info_next (match_info, error);
790 return match_info->matches >= 0;
794 * g_match_info_matches:
795 * @match_info: a #GMatchInfo structure
797 * Returns whether the previous match operation succeeded.
799 * Returns: %TRUE if the previous match operation succeeded,
805 g_match_info_matches (const GMatchInfo *match_info)
807 g_return_val_if_fail (match_info != NULL, FALSE);
809 return match_info->matches >= 0;
813 * g_match_info_get_match_count:
814 * @match_info: a #GMatchInfo structure
816 * Retrieves the number of matched substrings (including substring 0,
817 * that is the whole matched text), so 1 is returned if the pattern
818 * has no substrings in it and 0 is returned if the match failed.
820 * If the last match was obtained using the DFA algorithm, that is
821 * using g_regex_match_all() or g_regex_match_all_full(), the retrieved
822 * count is not that of the number of capturing parentheses but that of
823 * the number of matched substrings.
825 * Returns: Number of matched substrings, or -1 if an error occurred
830 g_match_info_get_match_count (const GMatchInfo *match_info)
832 g_return_val_if_fail (match_info, -1);
834 if (match_info->matches == PCRE_ERROR_NOMATCH)
837 else if (match_info->matches < PCRE_ERROR_NOMATCH)
842 return match_info->matches;
846 * g_match_info_is_partial_match:
847 * @match_info: a #GMatchInfo structure
849 * Usually if the string passed to g_regex_match*() matches as far as
850 * it goes, but is too short to match the entire pattern, %FALSE is
851 * returned. There are circumstances where it might be helpful to
852 * distinguish this case from other cases in which there is no match.
854 * Consider, for example, an application where a human is required to
855 * type in data for a field with specific formatting requirements. An
856 * example might be a date in the form ddmmmyy, defined by the pattern
857 * "^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$".
858 * If the application sees the user’s keystrokes one by one, and can
859 * check that what has been typed so far is potentially valid, it is
860 * able to raise an error as soon as a mistake is made.
862 * GRegex supports the concept of partial matching by means of the
863 * #G_REGEX_MATCH_PARTIAL_SOFT and #G_REGEX_MATCH_PARTIAL_HARD flags.
864 * When they are used, the return code for
865 * g_regex_match() or g_regex_match_full() is, as usual, %TRUE
866 * for a complete match, %FALSE otherwise. But, when these functions
867 * return %FALSE, you can check if the match was partial calling
868 * g_match_info_is_partial_match().
870 * The difference between #G_REGEX_MATCH_PARTIAL_SOFT and
871 * #G_REGEX_MATCH_PARTIAL_HARD is that when a partial match is encountered
872 * with #G_REGEX_MATCH_PARTIAL_SOFT, matching continues to search for a
873 * possible complete match, while with #G_REGEX_MATCH_PARTIAL_HARD matching
874 * stops at the partial match.
875 * When both #G_REGEX_MATCH_PARTIAL_SOFT and #G_REGEX_MATCH_PARTIAL_HARD
876 * are set, the latter takes precedence.
877 * See <ulink>man:pcrepartial</ulink> for more information on partial matching.
879 * Because of the way certain internal optimizations are implemented
880 * the partial matching algorithm cannot be used with all patterns.
881 * So repeated single characters such as "a{2,4}" and repeated single
882 * meta-sequences such as "\d+" are not permitted if the maximum number
883 * of occurrences is greater than one. Optional items such as "\d?"
884 * (where the maximum is one) are permitted. Quantifiers with any values
885 * are permitted after parentheses, so the invalid examples above can be
886 * coded thus "(a){2,4}" and "(\d)+". If #G_REGEX_MATCH_PARTIAL or
887 * #G_REGEX_MATCH_PARTIAL_HARD is set
888 * for a pattern that does not conform to the restrictions, matching
889 * functions return an error.
891 * Returns: %TRUE if the match was partial, %FALSE otherwise
896 g_match_info_is_partial_match (const GMatchInfo *match_info)
898 g_return_val_if_fail (match_info != NULL, FALSE);
900 return match_info->matches == PCRE_ERROR_PARTIAL;
904 * g_match_info_expand_references:
905 * @match_info: (allow-none): a #GMatchInfo or %NULL
906 * @string_to_expand: the string to expand
907 * @error: location to store the error occurring, or %NULL to ignore errors
909 * Returns a new string containing the text in @string_to_expand with
910 * references and escape sequences expanded. References refer to the last
911 * match done with @string against @regex and have the same syntax used by
914 * The @string_to_expand must be UTF-8 encoded even if #G_REGEX_RAW was
915 * passed to g_regex_new().
917 * The backreferences are extracted from the string passed to the match
918 * function, so you cannot call this function after freeing the string.
920 * @match_info may be %NULL in which case @string_to_expand must not
921 * contain references. For instance "foo\n" does not refer to an actual
922 * pattern and '\n' merely will be replaced with \n character,
923 * while to expand "\0" (whole match) one needs the result of a match.
924 * Use g_regex_check_replacement() to find out whether @string_to_expand
925 * contains references.
927 * Returns: (allow-none): the expanded string, or %NULL if an error occurred
932 g_match_info_expand_references (const GMatchInfo *match_info,
933 const gchar *string_to_expand,
938 GError *tmp_error = NULL;
940 g_return_val_if_fail (string_to_expand != NULL, NULL);
941 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
943 list = split_replacement (string_to_expand, &tmp_error);
944 if (tmp_error != NULL)
946 g_propagate_error (error, tmp_error);
950 if (!match_info && interpolation_list_needs_match (list))
952 g_critical ("String '%s' contains references to the match, can't "
953 "expand references without GMatchInfo object",
958 result = g_string_sized_new (strlen (string_to_expand));
959 interpolate_replacement (match_info, result, list);
961 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
963 return g_string_free (result, FALSE);
967 * g_match_info_fetch:
968 * @match_info: #GMatchInfo structure
969 * @match_num: number of the sub expression
971 * Retrieves the text matching the @match_num<!-- -->'th capturing
972 * parentheses. 0 is the full text of the match, 1 is the first paren
973 * set, 2 the second, and so on.
975 * If @match_num is a valid sub pattern but it didn't match anything
976 * (e.g. sub pattern 1, matching "b" against "(a)?b") then an empty
977 * string is returned.
979 * If the match was obtained using the DFA algorithm, that is using
980 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
981 * string is not that of a set of parentheses but that of a matched
982 * substring. Substrings are matched in reverse order of length, so
983 * 0 is the longest match.
985 * The string is fetched from the string passed to the match function,
986 * so you cannot call this function after freeing the string.
988 * Returns: (allow-none): The matched substring, or %NULL if an error
989 * occurred. You have to free the string yourself
994 g_match_info_fetch (const GMatchInfo *match_info,
997 /* we cannot use pcre_get_substring() because it allocates the
998 * string using pcre_malloc(). */
1002 g_return_val_if_fail (match_info != NULL, NULL);
1003 g_return_val_if_fail (match_num >= 0, NULL);
1005 /* match_num does not exist or it didn't matched, i.e. matching "b"
1006 * against "(a)?b" then group 0 is empty. */
1007 if (!g_match_info_fetch_pos (match_info, match_num, &start, &end))
1009 else if (start == -1)
1010 match = g_strdup ("");
1012 match = g_strndup (&match_info->string[start], end - start);
1018 * g_match_info_fetch_pos:
1019 * @match_info: #GMatchInfo structure
1020 * @match_num: number of the sub expression
1021 * @start_pos: (out) (allow-none): pointer to location where to store
1022 * the start position, or %NULL
1023 * @end_pos: (out) (allow-none): pointer to location where to store
1024 * the end position, or %NULL
1026 * Retrieves the position in bytes of the @match_num<!-- -->'th capturing
1027 * parentheses. 0 is the full text of the match, 1 is the first
1028 * paren set, 2 the second, and so on.
1030 * If @match_num is a valid sub pattern but it didn't match anything
1031 * (e.g. sub pattern 1, matching "b" against "(a)?b") then @start_pos
1032 * and @end_pos are set to -1 and %TRUE is returned.
1034 * If the match was obtained using the DFA algorithm, that is using
1035 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
1036 * position is not that of a set of parentheses but that of a matched
1037 * substring. Substrings are matched in reverse order of length, so
1038 * 0 is the longest match.
1040 * Returns: %TRUE if the position was fetched, %FALSE otherwise. If
1041 * the position cannot be fetched, @start_pos and @end_pos are left
1047 g_match_info_fetch_pos (const GMatchInfo *match_info,
1052 g_return_val_if_fail (match_info != NULL, FALSE);
1053 g_return_val_if_fail (match_num >= 0, FALSE);
1055 /* make sure the sub expression number they're requesting is less than
1056 * the total number of sub expressions that were matched. */
1057 if (match_num >= match_info->matches)
1060 if (start_pos != NULL)
1061 *start_pos = match_info->offsets[2 * match_num];
1063 if (end_pos != NULL)
1064 *end_pos = match_info->offsets[2 * match_num + 1];
1070 * Returns number of first matched subpattern with name @name.
1071 * There may be more than one in case when DUPNAMES is used,
1072 * and not all subpatterns with that name match;
1073 * pcre_get_stringnumber() does not work in that case.
1076 get_matched_substring_number (const GMatchInfo *match_info,
1080 gchar *first, *last;
1083 if (!(match_info->regex->compile_opts & G_REGEX_DUPNAMES))
1084 return pcre_get_stringnumber (match_info->regex->pcre_re, name);
1086 /* This code is copied from pcre_get.c: get_first_set() */
1087 entrysize = pcre_get_stringtable_entries (match_info->regex->pcre_re,
1095 for (entry = (guchar*) first; entry <= (guchar*) last; entry += entrysize)
1097 gint n = (entry[0] << 8) + entry[1];
1098 if (match_info->offsets[n*2] >= 0)
1102 return (first[0] << 8) + first[1];
1106 * g_match_info_fetch_named:
1107 * @match_info: #GMatchInfo structure
1108 * @name: name of the subexpression
1110 * Retrieves the text matching the capturing parentheses named @name.
1112 * If @name is a valid sub pattern name but it didn't match anything
1113 * (e.g. sub pattern "X", matching "b" against "(?P<X>a)?b")
1114 * then an empty string is returned.
1116 * The string is fetched from the string passed to the match function,
1117 * so you cannot call this function after freeing the string.
1119 * Returns: (allow-none): The matched substring, or %NULL if an error
1120 * occurred. You have to free the string yourself
1125 g_match_info_fetch_named (const GMatchInfo *match_info,
1128 /* we cannot use pcre_get_named_substring() because it allocates the
1129 * string using pcre_malloc(). */
1132 g_return_val_if_fail (match_info != NULL, NULL);
1133 g_return_val_if_fail (name != NULL, NULL);
1135 num = get_matched_substring_number (match_info, name);
1139 return g_match_info_fetch (match_info, num);
1143 * g_match_info_fetch_named_pos:
1144 * @match_info: #GMatchInfo structure
1145 * @name: name of the subexpression
1146 * @start_pos: (out) (allow-none): pointer to location where to store
1147 * the start position, or %NULL
1148 * @end_pos: (out) (allow-none): pointer to location where to store
1149 * the end position, or %NULL
1151 * Retrieves the position in bytes of the capturing parentheses named @name.
1153 * If @name is a valid sub pattern name but it didn't match anything
1154 * (e.g. sub pattern "X", matching "b" against "(?P<X>a)?b")
1155 * then @start_pos and @end_pos are set to -1 and %TRUE is returned.
1157 * Returns: %TRUE if the position was fetched, %FALSE otherwise.
1158 * If the position cannot be fetched, @start_pos and @end_pos
1159 * are left unchanged.
1164 g_match_info_fetch_named_pos (const GMatchInfo *match_info,
1171 g_return_val_if_fail (match_info != NULL, FALSE);
1172 g_return_val_if_fail (name != NULL, FALSE);
1174 num = get_matched_substring_number (match_info, name);
1178 return g_match_info_fetch_pos (match_info, num, start_pos, end_pos);
1182 * g_match_info_fetch_all:
1183 * @match_info: a #GMatchInfo structure
1185 * Bundles up pointers to each of the matching substrings from a match
1186 * and stores them in an array of gchar pointers. The first element in
1187 * the returned array is the match number 0, i.e. the entire matched
1190 * If a sub pattern didn't match anything (e.g. sub pattern 1, matching
1191 * "b" against "(a)?b") then an empty string is inserted.
1193 * If the last match was obtained using the DFA algorithm, that is using
1194 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
1195 * strings are not that matched by sets of parentheses but that of the
1196 * matched substring. Substrings are matched in reverse order of length,
1197 * so the first one is the longest match.
1199 * The strings are fetched from the string passed to the match function,
1200 * so you cannot call this function after freeing the string.
1202 * Returns: (transfer full): a %NULL-terminated array of gchar *
1203 * pointers. It must be freed using g_strfreev(). If the previous
1204 * match failed %NULL is returned
1209 g_match_info_fetch_all (const GMatchInfo *match_info)
1211 /* we cannot use pcre_get_substring_list() because the returned value
1212 * isn't suitable for g_strfreev(). */
1216 g_return_val_if_fail (match_info != NULL, NULL);
1218 if (match_info->matches < 0)
1221 result = g_new (gchar *, match_info->matches + 1);
1222 for (i = 0; i < match_info->matches; i++)
1223 result[i] = g_match_info_fetch (match_info, i);
1233 g_regex_error_quark (void)
1235 static GQuark error_quark = 0;
1237 if (error_quark == 0)
1238 error_quark = g_quark_from_static_string ("g-regex-error-quark");
1247 * Increases reference count of @regex by 1.
1254 g_regex_ref (GRegex *regex)
1256 g_return_val_if_fail (regex != NULL, NULL);
1257 g_atomic_int_inc (®ex->ref_count);
1265 * Decreases reference count of @regex by 1. When reference count drops
1266 * to zero, it frees all the memory associated with the regex structure.
1271 g_regex_unref (GRegex *regex)
1273 g_return_if_fail (regex != NULL);
1275 if (g_atomic_int_dec_and_test (®ex->ref_count))
1277 g_free (regex->pattern);
1278 if (regex->pcre_re != NULL)
1279 pcre_free (regex->pcre_re);
1280 if (regex->extra != NULL)
1281 pcre_free (regex->extra);
1288 * @pattern: the regular expression
1289 * @compile_options: compile options for the regular expression, or 0
1290 * @match_options: match options for the regular expression, or 0
1291 * @error: return location for a #GError
1293 * Compiles the regular expression to an internal form, and does
1294 * the initial setup of the #GRegex structure.
1296 * Returns: a #GRegex structure. Call g_regex_unref() when you
1302 g_regex_new (const gchar *pattern,
1303 GRegexCompileFlags compile_options,
1304 GRegexMatchFlags match_options,
1309 const gchar *errmsg;
1312 gboolean optimize = FALSE;
1313 static volatile gsize initialised = 0;
1314 unsigned long int pcre_compile_options;
1315 GRegexCompileFlags nonpcre_compile_options;
1317 g_return_val_if_fail (pattern != NULL, NULL);
1318 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
1319 g_return_val_if_fail ((compile_options & ~G_REGEX_COMPILE_MASK) == 0, NULL);
1320 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
1322 if (g_once_init_enter (&initialised))
1324 int supports_utf8, supports_ucp;
1326 pcre_config (PCRE_CONFIG_UTF8, &supports_utf8);
1328 g_critical (_("PCRE library is compiled without UTF8 support"));
1330 pcre_config (PCRE_CONFIG_UNICODE_PROPERTIES, &supports_ucp);
1332 g_critical (_("PCRE library is compiled without UTF8 properties support"));
1334 g_once_init_leave (&initialised, supports_utf8 && supports_ucp ? 1 : 2);
1337 if (G_UNLIKELY (initialised != 1))
1339 g_set_error_literal (error, G_REGEX_ERROR, G_REGEX_ERROR_COMPILE,
1340 _("PCRE library is compiled with incompatible options"));
1344 nonpcre_compile_options = compile_options & G_REGEX_COMPILE_NONPCRE_MASK;
1346 /* G_REGEX_OPTIMIZE has the same numeric value of PCRE_NO_UTF8_CHECK,
1347 * as we do not need to wrap PCRE_NO_UTF8_CHECK. */
1348 if (compile_options & G_REGEX_OPTIMIZE)
1351 /* In GRegex the string are, by default, UTF-8 encoded. PCRE
1352 * instead uses UTF-8 only if required with PCRE_UTF8. */
1353 if (compile_options & G_REGEX_RAW)
1356 compile_options &= ~G_REGEX_RAW;
1361 compile_options |= PCRE_UTF8 | PCRE_NO_UTF8_CHECK;
1362 match_options |= PCRE_NO_UTF8_CHECK;
1365 /* PCRE_NEWLINE_ANY is the default for the internal PCRE but
1366 * not for the system one. */
1367 if (!(compile_options & G_REGEX_NEWLINE_CR) &&
1368 !(compile_options & G_REGEX_NEWLINE_LF))
1370 compile_options |= PCRE_NEWLINE_ANY;
1373 compile_options |= PCRE_UCP;
1375 /* PCRE_BSR_UNICODE is the default for the internal PCRE but
1376 * possibly not for the system one.
1378 if (~compile_options & G_REGEX_BSR_ANYCRLF)
1379 compile_options |= PCRE_BSR_UNICODE;
1381 /* compile the pattern */
1382 re = pcre_compile2 (pattern, compile_options, &errcode,
1383 &errmsg, &erroffset, NULL);
1385 /* if the compilation failed, set the error member and return
1391 /* Translate the PCRE error code to GRegexError and use a translated
1392 * error message if possible */
1393 translate_compile_error (&errcode, &errmsg);
1395 /* PCRE uses byte offsets but we want to show character offsets */
1396 erroffset = g_utf8_pointer_to_offset (pattern, &pattern[erroffset]);
1398 tmp_error = g_error_new (G_REGEX_ERROR, errcode,
1399 _("Error while compiling regular "
1400 "expression %s at char %d: %s"),
1401 pattern, erroffset, errmsg);
1402 g_propagate_error (error, tmp_error);
1407 /* For options set at the beginning of the pattern, pcre puts them into
1408 * compile options, e.g. "(?i)foo" will make the pcre structure store
1409 * PCRE_CASELESS even though it wasn't explicitly given for compilation. */
1410 pcre_fullinfo (re, NULL, PCRE_INFO_OPTIONS, &pcre_compile_options);
1411 compile_options = pcre_compile_options & G_REGEX_COMPILE_PCRE_MASK;
1413 /* Don't leak PCRE_NEWLINE_ANY, which is part of PCRE_NEWLINE_ANYCRLF */
1414 if ((pcre_compile_options & PCRE_NEWLINE_ANYCRLF) != PCRE_NEWLINE_ANYCRLF)
1415 compile_options &= ~PCRE_NEWLINE_ANY;
1417 compile_options |= nonpcre_compile_options;
1419 if (!(compile_options & G_REGEX_DUPNAMES))
1421 gboolean jchanged = FALSE;
1422 pcre_fullinfo (re, NULL, PCRE_INFO_JCHANGED, &jchanged);
1424 compile_options |= G_REGEX_DUPNAMES;
1427 regex = g_new0 (GRegex, 1);
1428 regex->ref_count = 1;
1429 regex->pattern = g_strdup (pattern);
1430 regex->pcre_re = re;
1431 regex->compile_opts = compile_options;
1432 regex->match_opts = match_options;
1436 regex->extra = pcre_study (regex->pcre_re, 0, &errmsg);
1439 GError *tmp_error = g_error_new (G_REGEX_ERROR,
1440 G_REGEX_ERROR_OPTIMIZE,
1441 _("Error while optimizing "
1442 "regular expression %s: %s"),
1445 g_propagate_error (error, tmp_error);
1447 g_regex_unref (regex);
1456 * g_regex_get_pattern:
1457 * @regex: a #GRegex structure
1459 * Gets the pattern string associated with @regex, i.e. a copy of
1460 * the string passed to g_regex_new().
1462 * Returns: the pattern of @regex
1467 g_regex_get_pattern (const GRegex *regex)
1469 g_return_val_if_fail (regex != NULL, NULL);
1471 return regex->pattern;
1475 * g_regex_get_max_backref:
1478 * Returns the number of the highest back reference
1479 * in the pattern, or 0 if the pattern does not contain
1482 * Returns: the number of the highest back reference
1487 g_regex_get_max_backref (const GRegex *regex)
1491 pcre_fullinfo (regex->pcre_re, regex->extra,
1492 PCRE_INFO_BACKREFMAX, &value);
1498 * g_regex_get_capture_count:
1501 * Returns the number of capturing subpatterns in the pattern.
1503 * Returns: the number of capturing subpatterns
1508 g_regex_get_capture_count (const GRegex *regex)
1512 pcre_fullinfo (regex->pcre_re, regex->extra,
1513 PCRE_INFO_CAPTURECOUNT, &value);
1519 * g_regex_get_has_cr_or_lf:
1520 * @regex: a #GRegex structure
1522 * Checks whether the pattern contains explicit CR or LF references.
1524 * Returns: %TRUE if the pattern contains explicit CR or LF references
1529 g_regex_get_has_cr_or_lf (const GRegex *regex)
1533 pcre_fullinfo (regex->pcre_re, regex->extra,
1534 PCRE_INFO_HASCRORLF, &value);
1540 * g_regex_get_compile_flags:
1543 * Returns the compile options that @regex was created with.
1545 * Returns: flags from #GRegexCompileFlags
1550 g_regex_get_compile_flags (const GRegex *regex)
1552 g_return_val_if_fail (regex != NULL, 0);
1554 return regex->compile_opts;
1558 * g_regex_get_match_flags:
1561 * Returns the match options that @regex was created with.
1563 * Returns: flags from #GRegexMatchFlags
1568 g_regex_get_match_flags (const GRegex *regex)
1570 g_return_val_if_fail (regex != NULL, 0);
1572 return regex->match_opts & G_REGEX_MATCH_MASK;
1576 * g_regex_match_simple:
1577 * @pattern: the regular expression
1578 * @string: the string to scan for matches
1579 * @compile_options: compile options for the regular expression, or 0
1580 * @match_options: match options, or 0
1582 * Scans for a match in @string for @pattern.
1584 * This function is equivalent to g_regex_match() but it does not
1585 * require to compile the pattern with g_regex_new(), avoiding some
1586 * lines of code when you need just to do a match without extracting
1587 * substrings, capture counts, and so on.
1589 * If this function is to be called on the same @pattern more than
1590 * once, it's more efficient to compile the pattern once with
1591 * g_regex_new() and then use g_regex_match().
1593 * Returns: %TRUE if the string matched, %FALSE otherwise
1598 g_regex_match_simple (const gchar *pattern,
1599 const gchar *string,
1600 GRegexCompileFlags compile_options,
1601 GRegexMatchFlags match_options)
1606 regex = g_regex_new (pattern, compile_options, 0, NULL);
1609 result = g_regex_match_full (regex, string, -1, 0, match_options, NULL, NULL);
1610 g_regex_unref (regex);
1616 * @regex: a #GRegex structure from g_regex_new()
1617 * @string: the string to scan for matches
1618 * @match_options: match options
1619 * @match_info: (out) (allow-none): pointer to location where to store
1620 * the #GMatchInfo, or %NULL if you do not need it
1622 * Scans for a match in string for the pattern in @regex.
1623 * The @match_options are combined with the match options specified
1624 * when the @regex structure was created, letting you have more
1625 * flexibility in reusing #GRegex structures.
1627 * A #GMatchInfo structure, used to get information on the match,
1628 * is stored in @match_info if not %NULL. Note that if @match_info
1629 * is not %NULL then it is created even if the function returns %FALSE,
1630 * i.e. you must free it regardless if regular expression actually matched.
1632 * To retrieve all the non-overlapping matches of the pattern in
1633 * string you can use g_match_info_next().
1637 * print_uppercase_words (const gchar *string)
1639 * /* Print all uppercase-only words. */
1641 * GMatchInfo *match_info;
1643 * regex = g_regex_new ("[A-Z]+", 0, 0, NULL);
1644 * g_regex_match (regex, string, 0, &match_info);
1645 * while (g_match_info_matches (match_info))
1647 * gchar *word = g_match_info_fetch (match_info, 0);
1648 * g_print ("Found: %s\n", word);
1650 * g_match_info_next (match_info, NULL);
1652 * g_match_info_free (match_info);
1653 * g_regex_unref (regex);
1657 * @string is not copied and is used in #GMatchInfo internally. If
1658 * you use any #GMatchInfo method (except g_match_info_free()) after
1659 * freeing or modifying @string then the behaviour is undefined.
1661 * Returns: %TRUE is the string matched, %FALSE otherwise
1666 g_regex_match (const GRegex *regex,
1667 const gchar *string,
1668 GRegexMatchFlags match_options,
1669 GMatchInfo **match_info)
1671 return g_regex_match_full (regex, string, -1, 0, match_options,
1676 * g_regex_match_full:
1677 * @regex: a #GRegex structure from g_regex_new()
1678 * @string: (array length=string_len): the string to scan for matches
1679 * @string_len: the length of @string, or -1 if @string is nul-terminated
1680 * @start_position: starting index of the string to match
1681 * @match_options: match options
1682 * @match_info: (out) (allow-none): pointer to location where to store
1683 * the #GMatchInfo, or %NULL if you do not need it
1684 * @error: location to store the error occurring, or %NULL to ignore errors
1686 * Scans for a match in string for the pattern in @regex.
1687 * The @match_options are combined with the match options specified
1688 * when the @regex structure was created, letting you have more
1689 * flexibility in reusing #GRegex structures.
1691 * Setting @start_position differs from just passing over a shortened
1692 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
1693 * that begins with any kind of lookbehind assertion, such as "\b".
1695 * A #GMatchInfo structure, used to get information on the match, is
1696 * stored in @match_info if not %NULL. Note that if @match_info is
1697 * not %NULL then it is created even if the function returns %FALSE,
1698 * i.e. you must free it regardless if regular expression actually
1701 * @string is not copied and is used in #GMatchInfo internally. If
1702 * you use any #GMatchInfo method (except g_match_info_free()) after
1703 * freeing or modifying @string then the behaviour is undefined.
1705 * To retrieve all the non-overlapping matches of the pattern in
1706 * string you can use g_match_info_next().
1710 * print_uppercase_words (const gchar *string)
1712 * /* Print all uppercase-only words. */
1714 * GMatchInfo *match_info;
1715 * GError *error = NULL;
1717 * regex = g_regex_new ("[A-Z]+", 0, 0, NULL);
1718 * g_regex_match_full (regex, string, -1, 0, 0, &match_info, &error);
1719 * while (g_match_info_matches (match_info))
1721 * gchar *word = g_match_info_fetch (match_info, 0);
1722 * g_print ("Found: %s\n", word);
1724 * g_match_info_next (match_info, &error);
1726 * g_match_info_free (match_info);
1727 * g_regex_unref (regex);
1728 * if (error != NULL)
1730 * g_printerr ("Error while matching: %s\n", error->message);
1731 * g_error_free (error);
1736 * Returns: %TRUE is the string matched, %FALSE otherwise
1741 g_regex_match_full (const GRegex *regex,
1742 const gchar *string,
1744 gint start_position,
1745 GRegexMatchFlags match_options,
1746 GMatchInfo **match_info,
1752 g_return_val_if_fail (regex != NULL, FALSE);
1753 g_return_val_if_fail (string != NULL, FALSE);
1754 g_return_val_if_fail (start_position >= 0, FALSE);
1755 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
1756 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, FALSE);
1758 info = match_info_new (regex, string, string_len, start_position,
1759 match_options, FALSE);
1760 match_ok = g_match_info_next (info, error);
1761 if (match_info != NULL)
1764 g_match_info_free (info);
1770 * g_regex_match_all:
1771 * @regex: a #GRegex structure from g_regex_new()
1772 * @string: the string to scan for matches
1773 * @match_options: match options
1774 * @match_info: (out) (allow-none): pointer to location where to store
1775 * the #GMatchInfo, or %NULL if you do not need it
1777 * Using the standard algorithm for regular expression matching only
1778 * the longest match in the string is retrieved. This function uses
1779 * a different algorithm so it can retrieve all the possible matches.
1780 * For more documentation see g_regex_match_all_full().
1782 * A #GMatchInfo structure, used to get information on the match, is
1783 * stored in @match_info if not %NULL. Note that if @match_info is
1784 * not %NULL then it is created even if the function returns %FALSE,
1785 * i.e. you must free it regardless if regular expression actually
1788 * @string is not copied and is used in #GMatchInfo internally. If
1789 * you use any #GMatchInfo method (except g_match_info_free()) after
1790 * freeing or modifying @string then the behaviour is undefined.
1792 * Returns: %TRUE is the string matched, %FALSE otherwise
1797 g_regex_match_all (const GRegex *regex,
1798 const gchar *string,
1799 GRegexMatchFlags match_options,
1800 GMatchInfo **match_info)
1802 return g_regex_match_all_full (regex, string, -1, 0, match_options,
1807 * g_regex_match_all_full:
1808 * @regex: a #GRegex structure from g_regex_new()
1809 * @string: (array length=string_len): the string to scan for matches
1810 * @string_len: the length of @string, or -1 if @string is nul-terminated
1811 * @start_position: starting index of the string to match
1812 * @match_options: match options
1813 * @match_info: (out) (allow-none): pointer to location where to store
1814 * the #GMatchInfo, or %NULL if you do not need it
1815 * @error: location to store the error occurring, or %NULL to ignore errors
1817 * Using the standard algorithm for regular expression matching only
1818 * the longest match in the string is retrieved, it is not possible
1819 * to obtain all the available matches. For instance matching
1820 * "<a> <b> <c>" against the pattern "<.*>"
1821 * you get "<a> <b> <c>".
1823 * This function uses a different algorithm (called DFA, i.e. deterministic
1824 * finite automaton), so it can retrieve all the possible matches, all
1825 * starting at the same point in the string. For instance matching
1826 * "<a> <b> <c>" against the pattern "<.*>"
1827 * you would obtain three matches: "<a> <b> <c>",
1828 * "<a> <b>" and "<a>".
1830 * The number of matched strings is retrieved using
1831 * g_match_info_get_match_count(). To obtain the matched strings and
1832 * their position you can use, respectively, g_match_info_fetch() and
1833 * g_match_info_fetch_pos(). Note that the strings are returned in
1834 * reverse order of length; that is, the longest matching string is
1837 * Note that the DFA algorithm is slower than the standard one and it
1838 * is not able to capture substrings, so backreferences do not work.
1840 * Setting @start_position differs from just passing over a shortened
1841 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
1842 * that begins with any kind of lookbehind assertion, such as "\b".
1844 * A #GMatchInfo structure, used to get information on the match, is
1845 * stored in @match_info if not %NULL. Note that if @match_info is
1846 * not %NULL then it is created even if the function returns %FALSE,
1847 * i.e. you must free it regardless if regular expression actually
1850 * @string is not copied and is used in #GMatchInfo internally. If
1851 * you use any #GMatchInfo method (except g_match_info_free()) after
1852 * freeing or modifying @string then the behaviour is undefined.
1854 * Returns: %TRUE is the string matched, %FALSE otherwise
1859 g_regex_match_all_full (const GRegex *regex,
1860 const gchar *string,
1862 gint start_position,
1863 GRegexMatchFlags match_options,
1864 GMatchInfo **match_info,
1870 g_return_val_if_fail (regex != NULL, FALSE);
1871 g_return_val_if_fail (string != NULL, FALSE);
1872 g_return_val_if_fail (start_position >= 0, FALSE);
1873 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
1874 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, FALSE);
1876 info = match_info_new (regex, string, string_len, start_position,
1877 match_options, TRUE);
1883 info->matches = pcre_dfa_exec (regex->pcre_re, regex->extra,
1884 info->string, info->string_len,
1886 regex->match_opts | match_options,
1887 info->offsets, info->n_offsets,
1888 info->workspace, info->n_workspace);
1889 if (info->matches == PCRE_ERROR_DFA_WSSIZE)
1891 /* info->workspace is too small. */
1892 info->n_workspace *= 2;
1893 info->workspace = g_realloc (info->workspace,
1894 info->n_workspace * sizeof (gint));
1897 else if (info->matches == 0)
1899 /* info->offsets is too small. */
1900 info->n_offsets *= 2;
1901 info->offsets = g_realloc (info->offsets,
1902 info->n_offsets * sizeof (gint));
1905 else if (IS_PCRE_ERROR (info->matches))
1907 g_set_error (error, G_REGEX_ERROR, G_REGEX_ERROR_MATCH,
1908 _("Error while matching regular expression %s: %s"),
1909 regex->pattern, match_error (info->matches));
1913 /* set info->pos to -1 so that a call to g_match_info_next() fails. */
1916 if (match_info != NULL)
1919 g_match_info_free (info);
1921 return info->matches >= 0;
1925 * g_regex_get_string_number:
1926 * @regex: #GRegex structure
1927 * @name: name of the subexpression
1929 * Retrieves the number of the subexpression named @name.
1931 * Returns: The number of the subexpression or -1 if @name
1937 g_regex_get_string_number (const GRegex *regex,
1942 g_return_val_if_fail (regex != NULL, -1);
1943 g_return_val_if_fail (name != NULL, -1);
1945 num = pcre_get_stringnumber (regex->pcre_re, name);
1946 if (num == PCRE_ERROR_NOSUBSTRING)
1953 * g_regex_split_simple:
1954 * @pattern: the regular expression
1955 * @string: the string to scan for matches
1956 * @compile_options: compile options for the regular expression, or 0
1957 * @match_options: match options, or 0
1959 * Breaks the string on the pattern, and returns an array of
1960 * the tokens. If the pattern contains capturing parentheses,
1961 * then the text for each of the substrings will also be returned.
1962 * If the pattern does not match anywhere in the string, then the
1963 * whole string is returned as the first token.
1965 * This function is equivalent to g_regex_split() but it does
1966 * not require to compile the pattern with g_regex_new(), avoiding
1967 * some lines of code when you need just to do a split without
1968 * extracting substrings, capture counts, and so on.
1970 * If this function is to be called on the same @pattern more than
1971 * once, it's more efficient to compile the pattern once with
1972 * g_regex_new() and then use g_regex_split().
1974 * As a special case, the result of splitting the empty string ""
1975 * is an empty vector, not a vector containing a single string.
1976 * The reason for this special case is that being able to represent
1977 * a empty vector is typically more useful than consistent handling
1978 * of empty elements. If you do need to represent empty elements,
1979 * you'll need to check for the empty string before calling this
1982 * A pattern that can match empty strings splits @string into
1983 * separate characters wherever it matches the empty string between
1984 * characters. For example splitting "ab c" using as a separator
1985 * "\s*", you will get "a", "b" and "c".
1987 * Returns: (transfer full): a %NULL-terminated array of strings. Free
1988 * it using g_strfreev()
1993 g_regex_split_simple (const gchar *pattern,
1994 const gchar *string,
1995 GRegexCompileFlags compile_options,
1996 GRegexMatchFlags match_options)
2001 regex = g_regex_new (pattern, compile_options, 0, NULL);
2005 result = g_regex_split_full (regex, string, -1, 0, match_options, 0, NULL);
2006 g_regex_unref (regex);
2012 * @regex: a #GRegex structure
2013 * @string: the string to split with the pattern
2014 * @match_options: match time option flags
2016 * Breaks the string on the pattern, and returns an array of the tokens.
2017 * If the pattern contains capturing parentheses, then the text for each
2018 * of the substrings will also be returned. If the pattern does not match
2019 * anywhere in the string, then the whole string is returned as the first
2022 * As a special case, the result of splitting the empty string "" is an
2023 * empty vector, not a vector containing a single string. The reason for
2024 * this special case is that being able to represent a empty vector is
2025 * typically more useful than consistent handling of empty elements. If
2026 * you do need to represent empty elements, you'll need to check for the
2027 * empty string before calling this function.
2029 * A pattern that can match empty strings splits @string into separate
2030 * characters wherever it matches the empty string between characters.
2031 * For example splitting "ab c" using as a separator "\s*", you will get
2034 * Returns: (transfer full): a %NULL-terminated gchar ** array. Free
2035 * it using g_strfreev()
2040 g_regex_split (const GRegex *regex,
2041 const gchar *string,
2042 GRegexMatchFlags match_options)
2044 return g_regex_split_full (regex, string, -1, 0,
2045 match_options, 0, NULL);
2049 * g_regex_split_full:
2050 * @regex: a #GRegex structure
2051 * @string: (array length=string_len): the string to split with the pattern
2052 * @string_len: the length of @string, or -1 if @string is nul-terminated
2053 * @start_position: starting index of the string to match
2054 * @match_options: match time option flags
2055 * @max_tokens: the maximum number of tokens to split @string into.
2056 * If this is less than 1, the string is split completely
2057 * @error: return location for a #GError
2059 * Breaks the string on the pattern, and returns an array of the tokens.
2060 * If the pattern contains capturing parentheses, then the text for each
2061 * of the substrings will also be returned. If the pattern does not match
2062 * anywhere in the string, then the whole string is returned as the first
2065 * As a special case, the result of splitting the empty string "" is an
2066 * empty vector, not a vector containing a single string. The reason for
2067 * this special case is that being able to represent a empty vector is
2068 * typically more useful than consistent handling of empty elements. If
2069 * you do need to represent empty elements, you'll need to check for the
2070 * empty string before calling this function.
2072 * A pattern that can match empty strings splits @string into separate
2073 * characters wherever it matches the empty string between characters.
2074 * For example splitting "ab c" using as a separator "\s*", you will get
2077 * Setting @start_position differs from just passing over a shortened
2078 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
2079 * that begins with any kind of lookbehind assertion, such as "\b".
2081 * Returns: (transfer full): a %NULL-terminated gchar ** array. Free
2082 * it using g_strfreev()
2087 g_regex_split_full (const GRegex *regex,
2088 const gchar *string,
2090 gint start_position,
2091 GRegexMatchFlags match_options,
2095 GError *tmp_error = NULL;
2096 GMatchInfo *match_info;
2101 /* position of the last separator. */
2102 gint last_separator_end;
2103 /* was the last match 0 bytes long? */
2104 gboolean last_match_is_empty;
2105 /* the returned array of char **s */
2106 gchar **string_list;
2108 g_return_val_if_fail (regex != NULL, NULL);
2109 g_return_val_if_fail (string != NULL, NULL);
2110 g_return_val_if_fail (start_position >= 0, NULL);
2111 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
2112 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2114 if (max_tokens <= 0)
2115 max_tokens = G_MAXINT;
2118 string_len = strlen (string);
2120 /* zero-length string */
2121 if (string_len - start_position == 0)
2122 return g_new0 (gchar *, 1);
2124 if (max_tokens == 1)
2126 string_list = g_new0 (gchar *, 2);
2127 string_list[0] = g_strndup (&string[start_position],
2128 string_len - start_position);
2134 last_separator_end = start_position;
2135 last_match_is_empty = FALSE;
2137 match_ok = g_regex_match_full (regex, string, string_len, start_position,
2138 match_options, &match_info, &tmp_error);
2140 while (tmp_error == NULL)
2144 last_match_is_empty =
2145 (match_info->offsets[0] == match_info->offsets[1]);
2147 /* we need to skip empty separators at the same position of the end
2148 * of another separator. e.g. the string is "a b" and the separator
2149 * is " *", so from 1 to 2 we have a match and at position 2 we have
2150 * an empty match. */
2151 if (last_separator_end != match_info->offsets[1])
2156 token = g_strndup (string + last_separator_end,
2157 match_info->offsets[0] - last_separator_end);
2158 list = g_list_prepend (list, token);
2161 /* if there were substrings, these need to be added to
2163 match_count = g_match_info_get_match_count (match_info);
2164 if (match_count > 1)
2166 for (i = 1; i < match_count; i++)
2167 list = g_list_prepend (list, g_match_info_fetch (match_info, i));
2173 /* if there was no match, copy to end of string. */
2174 if (!last_match_is_empty)
2176 gchar *token = g_strndup (string + last_separator_end,
2177 match_info->string_len - last_separator_end);
2178 list = g_list_prepend (list, token);
2180 /* no more tokens, end the loop. */
2184 /* -1 to leave room for the last part. */
2185 if (token_count >= max_tokens - 1)
2187 /* we have reached the maximum number of tokens, so we copy
2188 * the remaining part of the string. */
2189 if (last_match_is_empty)
2191 /* the last match was empty, so we have moved one char
2192 * after the real position to avoid empty matches at the
2194 match_info->pos = PREV_CHAR (regex, &string[match_info->pos]) - string;
2196 /* the if is needed in the case we have terminated the available
2197 * tokens, but we are at the end of the string, so there are no
2198 * characters left to copy. */
2199 if (string_len > match_info->pos)
2201 gchar *token = g_strndup (string + match_info->pos,
2202 string_len - match_info->pos);
2203 list = g_list_prepend (list, token);
2209 last_separator_end = match_info->pos;
2210 if (last_match_is_empty)
2211 /* if the last match was empty, g_match_info_next() has moved
2212 * forward to avoid infinite loops, but we still need to copy that
2214 last_separator_end = PREV_CHAR (regex, &string[last_separator_end]) - string;
2216 match_ok = g_match_info_next (match_info, &tmp_error);
2218 g_match_info_free (match_info);
2219 if (tmp_error != NULL)
2221 g_propagate_error (error, tmp_error);
2222 g_list_free_full (list, g_free);
2223 match_info->pos = -1;
2227 string_list = g_new (gchar *, g_list_length (list) + 1);
2229 for (last = g_list_last (list); last; last = g_list_previous (last))
2230 string_list[i++] = last->data;
2231 string_list[i] = NULL;
2240 REPL_TYPE_CHARACTER,
2241 REPL_TYPE_SYMBOLIC_REFERENCE,
2242 REPL_TYPE_NUMERIC_REFERENCE,
2243 REPL_TYPE_CHANGE_CASE
2248 CHANGE_CASE_NONE = 1 << 0,
2249 CHANGE_CASE_UPPER = 1 << 1,
2250 CHANGE_CASE_LOWER = 1 << 2,
2251 CHANGE_CASE_UPPER_SINGLE = 1 << 3,
2252 CHANGE_CASE_LOWER_SINGLE = 1 << 4,
2253 CHANGE_CASE_SINGLE_MASK = CHANGE_CASE_UPPER_SINGLE | CHANGE_CASE_LOWER_SINGLE,
2254 CHANGE_CASE_LOWER_MASK = CHANGE_CASE_LOWER | CHANGE_CASE_LOWER_SINGLE,
2255 CHANGE_CASE_UPPER_MASK = CHANGE_CASE_UPPER | CHANGE_CASE_UPPER_SINGLE
2258 struct _InterpolationData
2264 ChangeCase change_case;
2268 free_interpolation_data (InterpolationData *data)
2270 g_free (data->text);
2274 static const gchar *
2275 expand_escape (const gchar *replacement,
2277 InterpolationData *data,
2282 const gchar *error_detail;
2284 GError *tmp_error = NULL;
2292 data->type = REPL_TYPE_CHARACTER;
2297 data->type = REPL_TYPE_CHARACTER;
2302 data->type = REPL_TYPE_CHARACTER;
2307 data->type = REPL_TYPE_CHARACTER;
2312 data->type = REPL_TYPE_CHARACTER;
2317 data->type = REPL_TYPE_CHARACTER;
2322 data->type = REPL_TYPE_CHARACTER;
2327 data->type = REPL_TYPE_CHARACTER;
2337 h = g_ascii_xdigit_value (*p);
2340 error_detail = _("hexadecimal digit or '}' expected");
2351 for (i = 0; i < 2; i++)
2353 h = g_ascii_xdigit_value (*p);
2356 error_detail = _("hexadecimal digit expected");
2363 data->type = REPL_TYPE_STRING;
2364 data->text = g_new0 (gchar, 8);
2365 g_unichar_to_utf8 (x, data->text);
2369 data->type = REPL_TYPE_CHANGE_CASE;
2370 data->change_case = CHANGE_CASE_LOWER_SINGLE;
2374 data->type = REPL_TYPE_CHANGE_CASE;
2375 data->change_case = CHANGE_CASE_UPPER_SINGLE;
2379 data->type = REPL_TYPE_CHANGE_CASE;
2380 data->change_case = CHANGE_CASE_LOWER;
2384 data->type = REPL_TYPE_CHANGE_CASE;
2385 data->change_case = CHANGE_CASE_UPPER;
2389 data->type = REPL_TYPE_CHANGE_CASE;
2390 data->change_case = CHANGE_CASE_NONE;
2396 error_detail = _("missing '<' in symbolic reference");
2405 error_detail = _("unfinished symbolic reference");
2412 error_detail = _("zero-length symbolic reference");
2415 if (g_ascii_isdigit (*q))
2420 h = g_ascii_digit_value (*q);
2423 error_detail = _("digit expected");
2432 data->type = REPL_TYPE_NUMERIC_REFERENCE;
2439 if (!g_ascii_isalnum (*r))
2441 error_detail = _("illegal symbolic reference");
2448 data->text = g_strndup (q, p - q);
2449 data->type = REPL_TYPE_SYMBOLIC_REFERENCE;
2454 /* if \0 is followed by a number is an octal number representing a
2455 * character, else it is a numeric reference. */
2456 if (g_ascii_digit_value (*g_utf8_next_char (p)) >= 0)
2459 p = g_utf8_next_char (p);
2472 for (i = 0; i < 3; i++)
2474 h = g_ascii_digit_value (*p);
2484 if (i == 2 && base == 10)
2490 if (base == 8 || i == 3)
2492 data->type = REPL_TYPE_STRING;
2493 data->text = g_new0 (gchar, 8);
2494 g_unichar_to_utf8 (x, data->text);
2498 data->type = REPL_TYPE_NUMERIC_REFERENCE;
2503 error_detail = _("stray final '\\'");
2507 error_detail = _("unknown escape sequence");
2514 /* G_GSSIZE_FORMAT doesn't work with gettext, so we use %lu */
2515 tmp_error = g_error_new (G_REGEX_ERROR,
2516 G_REGEX_ERROR_REPLACE,
2517 _("Error while parsing replacement "
2518 "text \"%s\" at char %lu: %s"),
2520 (gulong)(p - replacement),
2522 g_propagate_error (error, tmp_error);
2528 split_replacement (const gchar *replacement,
2532 InterpolationData *data;
2533 const gchar *p, *start;
2535 start = p = replacement;
2540 data = g_new0 (InterpolationData, 1);
2541 start = p = expand_escape (replacement, p, data, error);
2544 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
2545 free_interpolation_data (data);
2549 list = g_list_prepend (list, data);
2554 if (*p == '\\' || *p == '\0')
2558 data = g_new0 (InterpolationData, 1);
2559 data->text = g_strndup (start, p - start);
2560 data->type = REPL_TYPE_STRING;
2561 list = g_list_prepend (list, data);
2567 return g_list_reverse (list);
2570 /* Change the case of c based on change_case. */
2571 #define CHANGE_CASE(c, change_case) \
2572 (((change_case) & CHANGE_CASE_LOWER_MASK) ? \
2573 g_unichar_tolower (c) : \
2574 g_unichar_toupper (c))
2577 string_append (GString *string,
2579 ChangeCase *change_case)
2583 if (text[0] == '\0')
2586 if (*change_case == CHANGE_CASE_NONE)
2588 g_string_append (string, text);
2590 else if (*change_case & CHANGE_CASE_SINGLE_MASK)
2592 c = g_utf8_get_char (text);
2593 g_string_append_unichar (string, CHANGE_CASE (c, *change_case));
2594 g_string_append (string, g_utf8_next_char (text));
2595 *change_case = CHANGE_CASE_NONE;
2599 while (*text != '\0')
2601 c = g_utf8_get_char (text);
2602 g_string_append_unichar (string, CHANGE_CASE (c, *change_case));
2603 text = g_utf8_next_char (text);
2609 interpolate_replacement (const GMatchInfo *match_info,
2614 InterpolationData *idata;
2616 ChangeCase change_case = CHANGE_CASE_NONE;
2618 for (list = data; list; list = list->next)
2621 switch (idata->type)
2623 case REPL_TYPE_STRING:
2624 string_append (result, idata->text, &change_case);
2626 case REPL_TYPE_CHARACTER:
2627 g_string_append_c (result, CHANGE_CASE (idata->c, change_case));
2628 if (change_case & CHANGE_CASE_SINGLE_MASK)
2629 change_case = CHANGE_CASE_NONE;
2631 case REPL_TYPE_NUMERIC_REFERENCE:
2632 match = g_match_info_fetch (match_info, idata->num);
2635 string_append (result, match, &change_case);
2639 case REPL_TYPE_SYMBOLIC_REFERENCE:
2640 match = g_match_info_fetch_named (match_info, idata->text);
2643 string_append (result, match, &change_case);
2647 case REPL_TYPE_CHANGE_CASE:
2648 change_case = idata->change_case;
2656 /* whether actual match_info is needed for replacement, i.e.
2657 * whether there are references
2660 interpolation_list_needs_match (GList *list)
2662 while (list != NULL)
2664 InterpolationData *data = list->data;
2666 if (data->type == REPL_TYPE_SYMBOLIC_REFERENCE ||
2667 data->type == REPL_TYPE_NUMERIC_REFERENCE)
2680 * @regex: a #GRegex structure
2681 * @string: (array length=string_len): the string to perform matches against
2682 * @string_len: the length of @string, or -1 if @string is nul-terminated
2683 * @start_position: starting index of the string to match
2684 * @replacement: text to replace each match with
2685 * @match_options: options for the match
2686 * @error: location to store the error occurring, or %NULL to ignore errors
2688 * Replaces all occurrences of the pattern in @regex with the
2689 * replacement text. Backreferences of the form '\number' or
2690 * '\g<number>' in the replacement text are interpolated by the
2691 * number-th captured subexpression of the match, '\g<name>' refers
2692 * to the captured subexpression with the given name. '\0' refers to the
2693 * complete match, but '\0' followed by a number is the octal representation
2694 * of a character. To include a literal '\' in the replacement, write '\\'.
2695 * There are also escapes that changes the case of the following text:
2698 * <varlistentry><term>\l</term>
2700 * <para>Convert to lower case the next character</para>
2703 * <varlistentry><term>\u</term>
2705 * <para>Convert to upper case the next character</para>
2708 * <varlistentry><term>\L</term>
2710 * <para>Convert to lower case till \E</para>
2713 * <varlistentry><term>\U</term>
2715 * <para>Convert to upper case till \E</para>
2718 * <varlistentry><term>\E</term>
2720 * <para>End case modification</para>
2725 * If you do not need to use backreferences use g_regex_replace_literal().
2727 * The @replacement string must be UTF-8 encoded even if #G_REGEX_RAW was
2728 * passed to g_regex_new(). If you want to use not UTF-8 encoded stings
2729 * you can use g_regex_replace_literal().
2731 * Setting @start_position differs from just passing over a shortened
2732 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern that
2733 * begins with any kind of lookbehind assertion, such as "\b".
2735 * Returns: a newly allocated string containing the replacements
2740 g_regex_replace (const GRegex *regex,
2741 const gchar *string,
2743 gint start_position,
2744 const gchar *replacement,
2745 GRegexMatchFlags match_options,
2750 GError *tmp_error = NULL;
2752 g_return_val_if_fail (regex != NULL, NULL);
2753 g_return_val_if_fail (string != NULL, NULL);
2754 g_return_val_if_fail (start_position >= 0, NULL);
2755 g_return_val_if_fail (replacement != NULL, NULL);
2756 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
2757 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2759 list = split_replacement (replacement, &tmp_error);
2760 if (tmp_error != NULL)
2762 g_propagate_error (error, tmp_error);
2766 result = g_regex_replace_eval (regex,
2767 string, string_len, start_position,
2769 interpolate_replacement,
2772 if (tmp_error != NULL)
2773 g_propagate_error (error, tmp_error);
2775 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
2781 literal_replacement (const GMatchInfo *match_info,
2785 g_string_append (result, data);
2790 * g_regex_replace_literal:
2791 * @regex: a #GRegex structure
2792 * @string: (array length=string_len): the string to perform matches against
2793 * @string_len: the length of @string, or -1 if @string is nul-terminated
2794 * @start_position: starting index of the string to match
2795 * @replacement: text to replace each match with
2796 * @match_options: options for the match
2797 * @error: location to store the error occurring, or %NULL to ignore errors
2799 * Replaces all occurrences of the pattern in @regex with the
2800 * replacement text. @replacement is replaced literally, to
2801 * include backreferences use g_regex_replace().
2803 * Setting @start_position differs from just passing over a
2804 * shortened string and setting #G_REGEX_MATCH_NOTBOL in the
2805 * case of a pattern that begins with any kind of lookbehind
2806 * assertion, such as "\b".
2808 * Returns: a newly allocated string containing the replacements
2813 g_regex_replace_literal (const GRegex *regex,
2814 const gchar *string,
2816 gint start_position,
2817 const gchar *replacement,
2818 GRegexMatchFlags match_options,
2821 g_return_val_if_fail (replacement != NULL, NULL);
2822 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2824 return g_regex_replace_eval (regex,
2825 string, string_len, start_position,
2827 literal_replacement,
2828 (gpointer)replacement,
2833 * g_regex_replace_eval:
2834 * @regex: a #GRegex structure from g_regex_new()
2835 * @string: (array length=string_len): string to perform matches against
2836 * @string_len: the length of @string, or -1 if @string is nul-terminated
2837 * @start_position: starting index of the string to match
2838 * @match_options: options for the match
2839 * @eval: a function to call for each match
2840 * @user_data: user data to pass to the function
2841 * @error: location to store the error occurring, or %NULL to ignore errors
2843 * Replaces occurrences of the pattern in regex with the output of
2844 * @eval for that occurrence.
2846 * Setting @start_position differs from just passing over a shortened
2847 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
2848 * that begins with any kind of lookbehind assertion, such as "\b".
2850 * The following example uses g_regex_replace_eval() to replace multiple
2854 * eval_cb (const GMatchInfo *info,
2861 * match = g_match_info_fetch (info, 0);
2862 * r = g_hash_table_lookup ((GHashTable *)data, match);
2863 * g_string_append (res, r);
2875 * h = g_hash_table_new (g_str_hash, g_str_equal);
2877 * g_hash_table_insert (h, "1", "ONE");
2878 * g_hash_table_insert (h, "2", "TWO");
2879 * g_hash_table_insert (h, "3", "THREE");
2880 * g_hash_table_insert (h, "4", "FOUR");
2882 * reg = g_regex_new ("1|2|3|4", 0, 0, NULL);
2883 * res = g_regex_replace_eval (reg, text, -1, 0, 0, eval_cb, h, NULL);
2884 * g_hash_table_destroy (h);
2889 * Returns: a newly allocated string containing the replacements
2894 g_regex_replace_eval (const GRegex *regex,
2895 const gchar *string,
2897 gint start_position,
2898 GRegexMatchFlags match_options,
2899 GRegexEvalCallback eval,
2903 GMatchInfo *match_info;
2906 gboolean done = FALSE;
2907 GError *tmp_error = NULL;
2909 g_return_val_if_fail (regex != NULL, NULL);
2910 g_return_val_if_fail (string != NULL, NULL);
2911 g_return_val_if_fail (start_position >= 0, NULL);
2912 g_return_val_if_fail (eval != NULL, NULL);
2913 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2916 string_len = strlen (string);
2918 result = g_string_sized_new (string_len);
2920 /* run down the string making matches. */
2921 g_regex_match_full (regex, string, string_len, start_position,
2922 match_options, &match_info, &tmp_error);
2923 while (!done && g_match_info_matches (match_info))
2925 g_string_append_len (result,
2927 match_info->offsets[0] - str_pos);
2928 done = (*eval) (match_info, result, user_data);
2929 str_pos = match_info->offsets[1];
2930 g_match_info_next (match_info, &tmp_error);
2932 g_match_info_free (match_info);
2933 if (tmp_error != NULL)
2935 g_propagate_error (error, tmp_error);
2936 g_string_free (result, TRUE);
2940 g_string_append_len (result, string + str_pos, string_len - str_pos);
2941 return g_string_free (result, FALSE);
2945 * g_regex_check_replacement:
2946 * @replacement: the replacement string
2947 * @has_references: (out) (allow-none): location to store information about
2948 * references in @replacement or %NULL
2949 * @error: location to store error
2951 * Checks whether @replacement is a valid replacement string
2952 * (see g_regex_replace()), i.e. that all escape sequences in
2955 * If @has_references is not %NULL then @replacement is checked
2956 * for pattern references. For instance, replacement text 'foo\n'
2957 * does not contain references and may be evaluated without information
2958 * about actual match, but '\0\1' (whole match followed by first
2959 * subpattern) requires valid #GMatchInfo object.
2961 * Returns: whether @replacement is a valid replacement string
2966 g_regex_check_replacement (const gchar *replacement,
2967 gboolean *has_references,
2973 list = split_replacement (replacement, &tmp);
2977 g_propagate_error (error, tmp);
2982 *has_references = interpolation_list_needs_match (list);
2984 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
2990 * g_regex_escape_nul:
2991 * @string: the string to escape
2992 * @length: the length of @string
2994 * Escapes the nul characters in @string to "\x00". It can be used
2995 * to compile a regex with embedded nul characters.
2997 * For completeness, @length can be -1 for a nul-terminated string.
2998 * In this case the output string will be of course equal to @string.
3000 * Returns: a newly-allocated escaped string
3005 g_regex_escape_nul (const gchar *string,
3009 const gchar *p, *piece_start, *end;
3012 g_return_val_if_fail (string != NULL, NULL);
3015 return g_strdup (string);
3017 end = string + length;
3018 p = piece_start = string;
3019 escaped = g_string_sized_new (length + 1);
3027 if (p != piece_start)
3029 /* copy the previous piece. */
3030 g_string_append_len (escaped, piece_start, p - piece_start);
3032 if ((backslashes & 1) == 0)
3033 g_string_append_c (escaped, '\\');
3034 g_string_append_c (escaped, 'x');
3035 g_string_append_c (escaped, '0');
3036 g_string_append_c (escaped, '0');
3046 p = g_utf8_next_char (p);
3051 if (piece_start < end)
3052 g_string_append_len (escaped, piece_start, end - piece_start);
3054 return g_string_free (escaped, FALSE);
3058 * g_regex_escape_string:
3059 * @string: (array length=length): the string to escape
3060 * @length: the length of @string, or -1 if @string is nul-terminated
3062 * Escapes the special characters used for regular expressions
3063 * in @string, for instance "a.b*c" becomes "a\.b\*c". This
3064 * function is useful to dynamically generate regular expressions.
3066 * @string can contain nul characters that are replaced with "\0",
3067 * in this case remember to specify the correct length of @string
3070 * Returns: a newly-allocated escaped string
3075 g_regex_escape_string (const gchar *string,
3079 const char *p, *piece_start, *end;
3081 g_return_val_if_fail (string != NULL, NULL);
3084 length = strlen (string);
3086 end = string + length;
3087 p = piece_start = string;
3088 escaped = g_string_sized_new (length + 1);
3109 if (p != piece_start)
3110 /* copy the previous piece. */
3111 g_string_append_len (escaped, piece_start, p - piece_start);
3112 g_string_append_c (escaped, '\\');
3114 g_string_append_c (escaped, '0');
3116 g_string_append_c (escaped, *p);
3120 p = g_utf8_next_char (p);
3125 if (piece_start < end)
3126 g_string_append_len (escaped, piece_start, end - piece_start);
3128 return g_string_free (escaped, FALSE);