?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
regex: Add NEWLINE_ANYCRLF match option
[glib.git] / glib / gregex.c
blobadc1f7dbcbf02e205abe39bed7f6c2d779a97078
1 /* GRegex -- regular expression API wrapper around PCRE.
2 *
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>
6 *
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.
11 *
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.
16 *
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
20 */
21
22 #include "config.h"
23
24 #include <string.h>
25
26 #ifdef USE_SYSTEM_PCRE
27 #include <pcre.h>
28 #else
29 #include "pcre/pcre.h"
30 #endif
31
32 #include "gtypes.h"
33 #include "gregex.h"
34 #include "glibintl.h"
35 #include "glist.h"
36 #include "gmessages.h"
37 #include "gstrfuncs.h"
38 #include "gatomic.h"
39 #include "gthread.h"
40
41 /**
42 * SECTION:gregex
43 * @title: Perl-compatible regular expressions
44 * @short_description: matches strings against regular expressions
45 * @see_also: <xref linkend="glib-regex-syntax"/>
46 *
47 * The <function>g_regex_*()</function> functions implement regular
48 * expression pattern matching using syntax and semantics similar to
49 * Perl regular expression.
50 *
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.
63 *
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. "&agrave;") 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.
71 *
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).
79 *
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.
90 *
91 * Creating and manipulating the same #GRegex structure from different
92 * threads is not a problem as #GRegex does not modify its internal
93 * state between creation and destruction, on the other hand #GMatchInfo
94 * is not threadsafe.
95 *
96 * The regular expressions low-level functionalities are obtained through
97 * the excellent <ulink url="http://www.pcre.org/">PCRE</ulink> library
98 * written by Philip Hazel.
99 */
100
101 /* Mask of all the possible values for GRegexCompileFlags. */
102 #define G_REGEX_COMPILE_MASK (G_REGEX_CASELESS | \
103 G_REGEX_MULTILINE | \
104 G_REGEX_DOTALL | \
105 G_REGEX_EXTENDED | \
106 G_REGEX_ANCHORED | \
107 G_REGEX_DOLLAR_ENDONLY | \
108 G_REGEX_UNGREEDY | \
109 G_REGEX_RAW | \
110 G_REGEX_NO_AUTO_CAPTURE | \
111 G_REGEX_OPTIMIZE | \
112 G_REGEX_DUPNAMES | \
113 G_REGEX_NEWLINE_CR | \
114 G_REGEX_NEWLINE_LF | \
115 G_REGEX_NEWLINE_CRLF | \
116 G_REGEX_NEWLINE_ANYCRLF)
117
118 /* Mask of all the possible values for GRegexMatchFlags. */
119 #define G_REGEX_MATCH_MASK (G_REGEX_MATCH_ANCHORED | \
120 G_REGEX_MATCH_NOTBOL | \
121 G_REGEX_MATCH_NOTEOL | \
122 G_REGEX_MATCH_NOTEMPTY | \
123 G_REGEX_MATCH_PARTIAL | \
124 G_REGEX_MATCH_NEWLINE_CR | \
125 G_REGEX_MATCH_NEWLINE_LF | \
126 G_REGEX_MATCH_NEWLINE_CRLF | \
127 G_REGEX_MATCH_NEWLINE_ANY | \
128 G_REGEX_MATCH_NEWLINE_ANYCRLF)
129
130 /* we rely on these flags having the same values */
131 G_STATIC_ASSERT (G_REGEX_CASELESS == PCRE_CASELESS);
132 G_STATIC_ASSERT (G_REGEX_MULTILINE == PCRE_MULTILINE);
133 G_STATIC_ASSERT (G_REGEX_DOTALL == PCRE_DOTALL);
134 G_STATIC_ASSERT (G_REGEX_EXTENDED == PCRE_EXTENDED);
135 G_STATIC_ASSERT (G_REGEX_ANCHORED == PCRE_ANCHORED);
136 G_STATIC_ASSERT (G_REGEX_DOLLAR_ENDONLY == PCRE_DOLLAR_ENDONLY);
137 G_STATIC_ASSERT (G_REGEX_UNGREEDY == PCRE_UNGREEDY);
138 G_STATIC_ASSERT (G_REGEX_NO_AUTO_CAPTURE == PCRE_NO_AUTO_CAPTURE);
139 G_STATIC_ASSERT (G_REGEX_DUPNAMES == PCRE_DUPNAMES);
140 G_STATIC_ASSERT (G_REGEX_NEWLINE_CR == PCRE_NEWLINE_CR);
141 G_STATIC_ASSERT (G_REGEX_NEWLINE_LF == PCRE_NEWLINE_LF);
142 G_STATIC_ASSERT (G_REGEX_NEWLINE_CRLF == PCRE_NEWLINE_CRLF);
143 G_STATIC_ASSERT (G_REGEX_NEWLINE_ANYCRLF == PCRE_NEWLINE_ANYCRLF);
144
145 G_STATIC_ASSERT (G_REGEX_MATCH_ANCHORED == PCRE_ANCHORED);
146 G_STATIC_ASSERT (G_REGEX_MATCH_NOTBOL == PCRE_NOTBOL);
147 G_STATIC_ASSERT (G_REGEX_MATCH_NOTEOL == PCRE_NOTEOL);
148 G_STATIC_ASSERT (G_REGEX_MATCH_NOTEMPTY == PCRE_NOTEMPTY);
149 G_STATIC_ASSERT (G_REGEX_MATCH_PARTIAL == PCRE_PARTIAL);
150 G_STATIC_ASSERT (G_REGEX_MATCH_NEWLINE_CR == PCRE_NEWLINE_CR);
151 G_STATIC_ASSERT (G_REGEX_MATCH_NEWLINE_LF == PCRE_NEWLINE_LF);
152 G_STATIC_ASSERT (G_REGEX_MATCH_NEWLINE_CRLF == PCRE_NEWLINE_CRLF);
153 G_STATIC_ASSERT (G_REGEX_MATCH_NEWLINE_ANY == PCRE_NEWLINE_ANY);
154 G_STATIC_ASSERT (G_REGEX_MATCH_NEWLINE_ANYCRLF == PCRE_NEWLINE_ANYCRLF);
155
156 /* if the string is in UTF-8 use g_utf8_ functions, else use
157 * use just +/- 1. */
158 #define NEXT_CHAR(re, s) (((re)->compile_opts & PCRE_UTF8) ? \
159 g_utf8_next_char (s) : \
160 ((s) + 1))
161 #define PREV_CHAR(re, s) (((re)->compile_opts & PCRE_UTF8) ? \
162 g_utf8_prev_char (s) : \
163 ((s) - 1))
164
165 struct _GMatchInfo
166 {
167 volatile gint ref_count; /* the ref count */
168 GRegex *regex; /* the regex */
169 GRegexMatchFlags match_opts; /* options used at match time on the regex */
170 gint matches; /* number of matching sub patterns */
171 gint pos; /* position in the string where last match left off */
172 gint n_offsets; /* number of offsets */
173 gint *offsets; /* array of offsets paired 0,1 ; 2,3 ; 3,4 etc */
174 gint *workspace; /* workspace for pcre_dfa_exec() */
175 gint n_workspace; /* number of workspace elements */
176 const gchar *string; /* string passed to the match function */
177 gssize string_len; /* length of string */
178 };
179
180 struct _GRegex
181 {
182 volatile gint ref_count; /* the ref count for the immutable part */
183 gchar *pattern; /* the pattern */
184 pcre *pcre_re; /* compiled form of the pattern */
185 GRegexCompileFlags compile_opts; /* options used at compile time on the pattern */
186 GRegexMatchFlags match_opts; /* options used at match time on the regex */
187 pcre_extra *extra; /* data stored when G_REGEX_OPTIMIZE is used */
188 };
189
190 /* TRUE if ret is an error code, FALSE otherwise. */
191 #define IS_PCRE_ERROR(ret) ((ret) < PCRE_ERROR_NOMATCH && (ret) != PCRE_ERROR_PARTIAL)
192
193 typedef struct _InterpolationData InterpolationData;
194 static gboolean interpolation_list_needs_match (GList *list);
195 static gboolean interpolate_replacement (const GMatchInfo *match_info,
196 GString *result,
197 gpointer data);
198 static GList *split_replacement (const gchar *replacement,
199 GError **error);
200 static void free_interpolation_data (InterpolationData *data);
201
202
203 static const gchar *
204 match_error (gint errcode)
205 {
206 switch (errcode)
207 {
208 case PCRE_ERROR_NOMATCH:
209 /* not an error */
210 break;
211 case PCRE_ERROR_NULL:
212 /* NULL argument, this should not happen in GRegex */
213 g_warning ("A NULL argument was passed to PCRE");
214 break;
215 case PCRE_ERROR_BADOPTION:
216 return "bad options";
217 case PCRE_ERROR_BADMAGIC:
218 return _("corrupted object");
219 case PCRE_ERROR_UNKNOWN_OPCODE:
220 return N_("internal error or corrupted object");
221 case PCRE_ERROR_NOMEMORY:
222 return _("out of memory");
223 case PCRE_ERROR_NOSUBSTRING:
224 /* not used by pcre_exec() */
225 break;
226 case PCRE_ERROR_MATCHLIMIT:
227 return _("backtracking limit reached");
228 case PCRE_ERROR_CALLOUT:
229 /* callouts are not implemented */
230 break;
231 case PCRE_ERROR_BADUTF8:
232 case PCRE_ERROR_BADUTF8_OFFSET:
233 /* we do not check if strings are valid */
234 break;
235 case PCRE_ERROR_PARTIAL:
236 /* not an error */
237 break;
238 case PCRE_ERROR_BADPARTIAL:
239 return _("the pattern contains items not supported for partial matching");
240 case PCRE_ERROR_INTERNAL:
241 return _("internal error");
242 case PCRE_ERROR_BADCOUNT:
243 /* negative ovecsize, this should not happen in GRegex */
244 g_warning ("A negative ovecsize was passed to PCRE");
245 break;
246 case PCRE_ERROR_DFA_UITEM:
247 return _("the pattern contains items not supported for partial matching");
248 case PCRE_ERROR_DFA_UCOND:
249 return _("back references as conditions are not supported for partial matching");
250 case PCRE_ERROR_DFA_UMLIMIT:
251 /* the match_field field is not used in GRegex */
252 break;
253 case PCRE_ERROR_DFA_WSSIZE:
254 /* handled expanding the workspace */
255 break;
256 case PCRE_ERROR_DFA_RECURSE:
257 case PCRE_ERROR_RECURSIONLIMIT:
258 return _("recursion limit reached");
259 case PCRE_ERROR_NULLWSLIMIT:
260 return _("workspace limit for empty substrings reached");
261 case PCRE_ERROR_BADNEWLINE:
262 return _("invalid combination of newline flags");
263 case PCRE_ERROR_BADOFFSET:
264 return _("bad offset");
265 case PCRE_ERROR_SHORTUTF8:
266 return _("short utf8");
267 case PCRE_ERROR_RECURSELOOP:
268 return _("recursion loop");
269 default:
270 break;
271 }
272 return _("unknown error");
273 }
274
275 static void
276 translate_compile_error (gint *errcode, const gchar **errmsg)
277 {
278 /* Compile errors are created adding 100 to the error code returned
279 * by PCRE.
280 * If errcode is known we put the translatable error message in
281 * erromsg. If errcode is unknown we put the generic
282 * G_REGEX_ERROR_COMPILE error code in errcode and keep the
283 * untranslated error message returned by PCRE.
284 * Note that there can be more PCRE errors with the same GRegexError
285 * and that some PCRE errors are useless for us.
286 */
287 *errcode += 100;
288
289 switch (*errcode)
290 {
291 case G_REGEX_ERROR_STRAY_BACKSLASH:
292 *errmsg = _("\\ at end of pattern");
293 break;
294 case G_REGEX_ERROR_MISSING_CONTROL_CHAR:
295 *errmsg = _("\\c at end of pattern");
296 break;
297 case G_REGEX_ERROR_UNRECOGNIZED_ESCAPE:
298 *errmsg = _("unrecognized character follows \\");
299 break;
300 case G_REGEX_ERROR_QUANTIFIERS_OUT_OF_ORDER:
301 *errmsg = _("numbers out of order in {} quantifier");
302 break;
303 case G_REGEX_ERROR_QUANTIFIER_TOO_BIG:
304 *errmsg = _("number too big in {} quantifier");
305 break;
306 case G_REGEX_ERROR_UNTERMINATED_CHARACTER_CLASS:
307 *errmsg = _("missing terminating ] for character class");
308 break;
309 case G_REGEX_ERROR_INVALID_ESCAPE_IN_CHARACTER_CLASS:
310 *errmsg = _("invalid escape sequence in character class");
311 break;
312 case G_REGEX_ERROR_RANGE_OUT_OF_ORDER:
313 *errmsg = _("range out of order in character class");
314 break;
315 case G_REGEX_ERROR_NOTHING_TO_REPEAT:
316 *errmsg = _("nothing to repeat");
317 break;
318 case 111: /* internal error: unexpected repeat */
319 *errcode = G_REGEX_ERROR_INTERNAL;
320 *errmsg = _("unexpected repeat");
321 break;
322 case G_REGEX_ERROR_UNRECOGNIZED_CHARACTER:
323 *errmsg = _("unrecognized character after (? or (?-");
324 break;
325 case G_REGEX_ERROR_POSIX_NAMED_CLASS_OUTSIDE_CLASS:
326 *errmsg = _("POSIX named classes are supported only within a class");
327 break;
328 case G_REGEX_ERROR_UNMATCHED_PARENTHESIS:
329 *errmsg = _("missing terminating )");
330 break;
331 case G_REGEX_ERROR_INEXISTENT_SUBPATTERN_REFERENCE:
332 *errmsg = _("reference to non-existent subpattern");
333 break;
334 case G_REGEX_ERROR_UNTERMINATED_COMMENT:
335 *errmsg = _("missing ) after comment");
336 break;
337 case G_REGEX_ERROR_EXPRESSION_TOO_LARGE:
338 *errmsg = _("regular expression is too large");
339 break;
340 case G_REGEX_ERROR_MEMORY_ERROR:
341 *errmsg = _("failed to get memory");
342 break;
343 case 122: /* unmatched parentheses */
344 *errcode = G_REGEX_ERROR_UNMATCHED_PARENTHESIS;
345 *errmsg = _(") without opening (");
346 break;
347 case 123: /* internal error: code overflow */
348 *errcode = G_REGEX_ERROR_INTERNAL;
349 *errmsg = _("code overflow");
350 break;
351 case 124: /* "unrecognized character after (?<\0 */
352 *errcode = G_REGEX_ERROR_UNRECOGNIZED_CHARACTER;
353 *errmsg = _("unrecognized character after (?<");
354 break;
355 case G_REGEX_ERROR_VARIABLE_LENGTH_LOOKBEHIND:
356 *errmsg = _("lookbehind assertion is not fixed length");
357 break;
358 case G_REGEX_ERROR_MALFORMED_CONDITION:
359 *errmsg = _("malformed number or name after (?(");
360 break;
361 case G_REGEX_ERROR_TOO_MANY_CONDITIONAL_BRANCHES:
362 *errmsg = _("conditional group contains more than two branches");
363 break;
364 case G_REGEX_ERROR_ASSERTION_EXPECTED:
365 *errmsg = _("assertion expected after (?(");
366 break;
367 case 129:
368 *errcode = G_REGEX_ERROR_UNMATCHED_PARENTHESIS;
369 /* translators: '(?R' and '(?[+-]digits' are both meant as (groups of)
370 * sequences here, '(?-54' would be an example for the second group.
371 */
372 *errmsg = _("(?R or (?[+-]digits must be followed by )");
373 break;
374 case G_REGEX_ERROR_UNKNOWN_POSIX_CLASS_NAME:
375 *errmsg = _("unknown POSIX class name");
376 break;
377 case G_REGEX_ERROR_POSIX_COLLATING_ELEMENTS_NOT_SUPPORTED:
378 *errmsg = _("POSIX collating elements are not supported");
379 break;
380 case G_REGEX_ERROR_HEX_CODE_TOO_LARGE:
381 *errmsg = _("character value in \\x{...} sequence is too large");
382 break;
383 case G_REGEX_ERROR_INVALID_CONDITION:
384 *errmsg = _("invalid condition (?(0)");
385 break;
386 case G_REGEX_ERROR_SINGLE_BYTE_MATCH_IN_LOOKBEHIND:
387 *errmsg = _("\\C not allowed in lookbehind assertion");
388 break;
389 case 137: /* PCRE does not support \\L, \\l, \\N{name}, \\U, or \\u\0 */
390 /* A number of Perl escapes are not handled by PCRE.
391 * Therefore it explicitly raises ERR37.
392 */
393 *errcode = G_REGEX_ERROR_UNRECOGNIZED_ESCAPE;
394 *errmsg = _("escapes \\L, \\l, \\N{name}, \\U, and \\u are not supported");
395 break;
396 case G_REGEX_ERROR_INFINITE_LOOP:
397 *errmsg = _("recursive call could loop indefinitely");
398 break;
399 case 141: /* unrecognized character after (?P\0 */
400 *errcode = G_REGEX_ERROR_UNRECOGNIZED_CHARACTER;
401 *errmsg = _("unrecognized character after (?P");
402 break;
403 case G_REGEX_ERROR_MISSING_SUBPATTERN_NAME_TERMINATOR:
404 *errmsg = _("missing terminator in subpattern name");
405 break;
406 case G_REGEX_ERROR_DUPLICATE_SUBPATTERN_NAME:
407 *errmsg = _("two named subpatterns have the same name");
408 break;
409 case G_REGEX_ERROR_MALFORMED_PROPERTY:
410 *errmsg = _("malformed \\P or \\p sequence");
411 break;
412 case G_REGEX_ERROR_UNKNOWN_PROPERTY:
413 *errmsg = _("unknown property name after \\P or \\p");
414 break;
415 case G_REGEX_ERROR_SUBPATTERN_NAME_TOO_LONG:
416 *errmsg = _("subpattern name is too long (maximum 32 characters)");
417 break;
418 case G_REGEX_ERROR_TOO_MANY_SUBPATTERNS:
419 *errmsg = _("too many named subpatterns (maximum 10,000)");
420 break;
421 case G_REGEX_ERROR_INVALID_OCTAL_VALUE:
422 *errmsg = _("octal value is greater than \\377");
423 break;
424 case 152: /* internal error: overran compiling workspace */
425 *errcode = G_REGEX_ERROR_INTERNAL;
426 *errmsg = _("overran compiling workspace");
427 break;
428 case 153: /* internal error: previously-checked referenced subpattern not found */
429 *errcode = G_REGEX_ERROR_INTERNAL;
430 *errmsg = _("previously-checked referenced subpattern not found");
431 break;
432 case G_REGEX_ERROR_TOO_MANY_BRANCHES_IN_DEFINE:
433 *errmsg = _("DEFINE group contains more than one branch");
434 break;
435 case G_REGEX_ERROR_INCONSISTENT_NEWLINE_OPTIONS:
436 *errmsg = _("inconsistent NEWLINE options");
437 break;
438 case G_REGEX_ERROR_MISSING_BACK_REFERENCE:
439 *errmsg = _("\\g is not followed by a braced, angle-bracketed, or quoted name or "
440 "number, or by a plain number");
441 break;
442 case G_REGEX_ERROR_INVALID_RELATIVE_REFERENCE:
443 *errmsg = _("a numbered reference must not be zero");
444 break;
445 case G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_FORBIDDEN:
446 *errmsg = _("an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)");
447 break;
448 case G_REGEX_ERROR_UNKNOWN_BACKTRACKING_CONTROL_VERB:
449 *errmsg = _("(*VERB) not recognized");
450 break;
451 case G_REGEX_ERROR_NUMBER_TOO_BIG:
452 *errmsg = _("number is too bug");
453 break;
454 case G_REGEX_ERROR_MISSING_SUBPATTERN_NAME:
455 *errmsg = _("missing subpattern name after (?&");
456 break;
457 case G_REGEX_ERROR_MISSING_DIGIT:
458 *errmsg = _("digit expected after (?+");
459 break;
460 case G_REGEX_ERROR_EXTRA_SUBPATTERN_NAME:
461 *errmsg = _("different names for subpatterns of the same number are not allowed");
462 break;
463 case G_REGEX_ERROR_BACKTRACKING_CONTROL_VERB_ARGUMENT_REQUIRED:
464 *errmsg = _("(*MARK) must have an argument");
465 break;
466 case G_REGEX_ERROR_INVALID_CONTROL_CHAR:
467 *errmsg = _( "\\c must be followed by an ASCII character");
468 break;
469 case G_REGEX_ERROR_MISSING_NAME:
470 *errmsg = _("\\k is not followed by a braced, angle-bracketed, or quoted name");
471 break;
472 case G_REGEX_ERROR_NOT_SUPPORTED_IN_CLASS:
473 *errmsg = _("\\N is not supported in a class");
474 break;
475 case G_REGEX_ERROR_TOO_MANY_FORWARD_REFERENCES:
476 *errmsg = _("too many forward references");
477 break;
478 case G_REGEX_ERROR_NAME_TOO_LONG:
479 *errmsg = _("name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)");
480 break;
481
482 case 116: /* erroffset passed as NULL */
483 /* This should not happen as we never pass a NULL erroffset */
484 g_warning ("erroffset passed as NULL");
485 *errcode = G_REGEX_ERROR_COMPILE;
486 break;
487 case 117: /* unknown option bit(s) set */
488 /* This should not happen as we check options before passing them
489 * to pcre_compile2() */
490 g_warning ("unknown option bit(s) set");
491 *errcode = G_REGEX_ERROR_COMPILE;
492 break;
493 case 132: /* this version of PCRE is compiled without UTF support */
494 case 144: /* invalid UTF-8 string */
495 case 145: /* support for \\P, \\p, and \\X has not been compiled */
496 case 167: /* this version of PCRE is not compiled with Unicode property support */
497 case 173: /* disallowed Unicode code point (>= 0xd800 && <= 0xdfff) */
498 case 174: /* invalid UTF-16 string */
499 /* These errors should not happen as we are using an UTF-8 and UCP-enabled PCRE
500 * and we do not check if strings are valid */
501 case 164: /* ] is an invalid data character in JavaScript compatibility mode */
502 /* This should not happen as we don't use PCRE_JAVASCRIPT_COMPAT */
503 g_warning ("%s", *errmsg);
504 *errcode = G_REGEX_ERROR_COMPILE;
505 break;
506 case 170: /* internal error: unknown opcode in find_fixedlength() */
507 *errcode = G_REGEX_ERROR_INTERNAL;
508 break;
509
510 default:
511 *errcode = G_REGEX_ERROR_COMPILE;
512 }
513 }
514
515 /* GMatchInfo */
516
517 static GMatchInfo *
518 match_info_new (const GRegex *regex,
519 const gchar *string,
520 gint string_len,
521 gint start_position,
522 gint match_options,
523 gboolean is_dfa)
524 {
525 GMatchInfo *match_info;
526
527 if (string_len < 0)
528 string_len = strlen (string);
529
530 match_info = g_new0 (GMatchInfo, 1);
531 match_info->ref_count = 1;
532 match_info->regex = g_regex_ref ((GRegex *)regex);
533 match_info->string = string;
534 match_info->string_len = string_len;
535 match_info->matches = PCRE_ERROR_NOMATCH;
536 match_info->pos = start_position;
537 match_info->match_opts = match_options;
538
539 if (is_dfa)
540 {
541 /* These values should be enough for most cases, if they are not
542 * enough g_regex_match_all_full() will expand them. */
543 match_info->n_offsets = 24;
544 match_info->n_workspace = 100;
545 match_info->workspace = g_new (gint, match_info->n_workspace);
546 }
547 else
548 {
549 gint capture_count;
550 pcre_fullinfo (regex->pcre_re, regex->extra,
551 PCRE_INFO_CAPTURECOUNT, &capture_count);
552 match_info->n_offsets = (capture_count + 1) * 3;
553 }
554
555 match_info->offsets = g_new0 (gint, match_info->n_offsets);
556 /* Set an invalid position for the previous match. */
557 match_info->offsets[0] = -1;
558 match_info->offsets[1] = -1;
559
560 return match_info;
561 }
562
563 /**
564 * g_match_info_get_regex:
565 * @match_info: a #GMatchInfo
566 *
567 * Returns #GRegex object used in @match_info. It belongs to Glib
568 * and must not be freed. Use g_regex_ref() if you need to keep it
569 * after you free @match_info object.
570 *
571 * Returns: #GRegex object used in @match_info
572 *
573 * Since: 2.14
574 */
575 GRegex *
576 g_match_info_get_regex (const GMatchInfo *match_info)
577 {
578 g_return_val_if_fail (match_info != NULL, NULL);
579 return match_info->regex;
580 }
581
582 /**
583 * g_match_info_get_string:
584 * @match_info: a #GMatchInfo
585 *
586 * Returns the string searched with @match_info. This is the
587 * string passed to g_regex_match() or g_regex_replace() so
588 * you may not free it before calling this function.
589 *
590 * Returns: the string searched with @match_info
591 *
592 * Since: 2.14
593 */
594 const gchar *
595 g_match_info_get_string (const GMatchInfo *match_info)
596 {
597 g_return_val_if_fail (match_info != NULL, NULL);
598 return match_info->string;
599 }
600
601 /**
602 * g_match_info_ref:
603 * @match_info: a #GMatchInfo
604 *
605 * Increases reference count of @match_info by 1.
606 *
607 * Returns: @match_info
608 *
609 * Since: 2.30
610 */
611 GMatchInfo *
612 g_match_info_ref (GMatchInfo *match_info)
613 {
614 g_return_val_if_fail (match_info != NULL, NULL);
615 g_atomic_int_inc (&match_info->ref_count);
616 return match_info;
617 }
618
619 /**
620 * g_match_info_unref:
621 * @match_info: a #GMatchInfo
622 *
623 * Decreases reference count of @match_info by 1. When reference count drops
624 * to zero, it frees all the memory associated with the match_info structure.
625 *
626 * Since: 2.30
627 */
628 void
629 g_match_info_unref (GMatchInfo *match_info)
630 {
631 if (g_atomic_int_dec_and_test (&match_info->ref_count))
632 {
633 g_regex_unref (match_info->regex);
634 g_free (match_info->offsets);
635 g_free (match_info->workspace);
636 g_free (match_info);
637 }
638 }
639
640 /**
641 * g_match_info_free:
642 * @match_info: (allow-none): a #GMatchInfo, or %NULL
643 *
644 * If @match_info is not %NULL, calls g_match_info_unref(); otherwise does
645 * nothing.
646 *
647 * Since: 2.14
648 */
649 void
650 g_match_info_free (GMatchInfo *match_info)
651 {
652 if (match_info == NULL)
653 return;
654
655 g_match_info_unref (match_info);
656 }
657
658 /**
659 * g_match_info_next:
660 * @match_info: a #GMatchInfo structure
661 * @error: location to store the error occurring, or %NULL to ignore errors
662 *
663 * Scans for the next match using the same parameters of the previous
664 * call to g_regex_match_full() or g_regex_match() that returned
665 * @match_info.
666 *
667 * The match is done on the string passed to the match function, so you
668 * cannot free it before calling this function.
669 *
670 * Returns: %TRUE is the string matched, %FALSE otherwise
671 *
672 * Since: 2.14
673 */
674 gboolean
675 g_match_info_next (GMatchInfo *match_info,
676 GError **error)
677 {
678 gint prev_match_start;
679 gint prev_match_end;
680
681 g_return_val_if_fail (match_info != NULL, FALSE);
682 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
683 g_return_val_if_fail (match_info->pos >= 0, FALSE);
684
685 prev_match_start = match_info->offsets[0];
686 prev_match_end = match_info->offsets[1];
687
688 if (match_info->pos > match_info->string_len)
689 {
690 /* we have reached the end of the string */
691 match_info->pos = -1;
692 match_info->matches = PCRE_ERROR_NOMATCH;
693 return FALSE;
694 }
695
696 match_info->matches = pcre_exec (match_info->regex->pcre_re,
697 match_info->regex->extra,
698 match_info->string,
699 match_info->string_len,
700 match_info->pos,
701 match_info->regex->match_opts | match_info->match_opts,
702 match_info->offsets,
703 match_info->n_offsets);
704 if (IS_PCRE_ERROR (match_info->matches))
705 {
706 g_set_error (error, G_REGEX_ERROR, G_REGEX_ERROR_MATCH,
707 _("Error while matching regular expression %s: %s"),
708 match_info->regex->pattern, match_error (match_info->matches));
709 return FALSE;
710 }
711
712 /* avoid infinite loops if the pattern is an empty string or something
713 * equivalent */
714 if (match_info->pos == match_info->offsets[1])
715 {
716 if (match_info->pos > match_info->string_len)
717 {
718 /* we have reached the end of the string */
719 match_info->pos = -1;
720 match_info->matches = PCRE_ERROR_NOMATCH;
721 return FALSE;
722 }
723
724 match_info->pos = NEXT_CHAR (match_info->regex,
725 &match_info->string[match_info->pos]) -
726 match_info->string;
727 }
728 else
729 {
730 match_info->pos = match_info->offsets[1];
731 }
732
733 /* it's possible to get two identical matches when we are matching
734 * empty strings, for instance if the pattern is "(?=[A-Z0-9])" and
735 * the string is "RegExTest" we have:
736 * - search at position 0: match from 0 to 0
737 * - search at position 1: match from 3 to 3
738 * - search at position 3: match from 3 to 3 (duplicate)
739 * - search at position 4: match from 5 to 5
740 * - search at position 5: match from 5 to 5 (duplicate)
741 * - search at position 6: no match -> stop
742 * so we have to ignore the duplicates.
743 * see bug #515944: http://bugzilla.gnome.org/show_bug.cgi?id=515944 */
744 if (match_info->matches >= 0 &&
745 prev_match_start == match_info->offsets[0] &&
746 prev_match_end == match_info->offsets[1])
747 {
748 /* ignore this match and search the next one */
749 return g_match_info_next (match_info, error);
750 }
751
752 return match_info->matches >= 0;
753 }
754
755 /**
756 * g_match_info_matches:
757 * @match_info: a #GMatchInfo structure
758 *
759 * Returns whether the previous match operation succeeded.
760 *
761 * Returns: %TRUE if the previous match operation succeeded,
762 * %FALSE otherwise
763 *
764 * Since: 2.14
765 */
766 gboolean
767 g_match_info_matches (const GMatchInfo *match_info)
768 {
769 g_return_val_if_fail (match_info != NULL, FALSE);
770
771 return match_info->matches >= 0;
772 }
773
774 /**
775 * g_match_info_get_match_count:
776 * @match_info: a #GMatchInfo structure
777 *
778 * Retrieves the number of matched substrings (including substring 0,
779 * that is the whole matched text), so 1 is returned if the pattern
780 * has no substrings in it and 0 is returned if the match failed.
781 *
782 * If the last match was obtained using the DFA algorithm, that is
783 * using g_regex_match_all() or g_regex_match_all_full(), the retrieved
784 * count is not that of the number of capturing parentheses but that of
785 * the number of matched substrings.
786 *
787 * Returns: Number of matched substrings, or -1 if an error occurred
788 *
789 * Since: 2.14
790 */
791 gint
792 g_match_info_get_match_count (const GMatchInfo *match_info)
793 {
794 g_return_val_if_fail (match_info, -1);
795
796 if (match_info->matches == PCRE_ERROR_NOMATCH)
797 /* no match */
798 return 0;
799 else if (match_info->matches < PCRE_ERROR_NOMATCH)
800 /* error */
801 return -1;
802 else
803 /* match */
804 return match_info->matches;
805 }
806
807 /**
808 * g_match_info_is_partial_match:
809 * @match_info: a #GMatchInfo structure
810 *
811 * Usually if the string passed to g_regex_match*() matches as far as
812 * it goes, but is too short to match the entire pattern, %FALSE is
813 * returned. There are circumstances where it might be helpful to
814 * distinguish this case from other cases in which there is no match.
815 *
816 * Consider, for example, an application where a human is required to
817 * type in data for a field with specific formatting requirements. An
818 * example might be a date in the form ddmmmyy, defined by the pattern
819 * "^\d?\d(jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\d\d$".
820 * If the application sees the user’s keystrokes one by one, and can
821 * check that what has been typed so far is potentially valid, it is
822 * able to raise an error as soon as a mistake is made.
823 *
824 * GRegex supports the concept of partial matching by means of the
825 * #G_REGEX_MATCH_PARTIAL flag. When this is set the return code for
826 * g_regex_match() or g_regex_match_full() is, as usual, %TRUE
827 * for a complete match, %FALSE otherwise. But, when these functions
828 * return %FALSE, you can check if the match was partial calling
829 * g_match_info_is_partial_match().
830 *
831 * When using partial matching you cannot use g_match_info_fetch*().
832 *
833 * Because of the way certain internal optimizations are implemented
834 * the partial matching algorithm cannot be used with all patterns.
835 * So repeated single characters such as "a{2,4}" and repeated single
836 * meta-sequences such as "\d+" are not permitted if the maximum number
837 * of occurrences is greater than one. Optional items such as "\d?"
838 * (where the maximum is one) are permitted. Quantifiers with any values
839 * are permitted after parentheses, so the invalid examples above can be
840 * coded thus "(a){2,4}" and "(\d)+". If #G_REGEX_MATCH_PARTIAL is set
841 * for a pattern that does not conform to the restrictions, matching
842 * functions return an error.
843 *
844 * Returns: %TRUE if the match was partial, %FALSE otherwise
845 *
846 * Since: 2.14
847 */
848 gboolean
849 g_match_info_is_partial_match (const GMatchInfo *match_info)
850 {
851 g_return_val_if_fail (match_info != NULL, FALSE);
852
853 return match_info->matches == PCRE_ERROR_PARTIAL;
854 }
855
856 /**
857 * g_match_info_expand_references:
858 * @match_info: (allow-none): a #GMatchInfo or %NULL
859 * @string_to_expand: the string to expand
860 * @error: location to store the error occurring, or %NULL to ignore errors
861 *
862 * Returns a new string containing the text in @string_to_expand with
863 * references and escape sequences expanded. References refer to the last
864 * match done with @string against @regex and have the same syntax used by
865 * g_regex_replace().
866 *
867 * The @string_to_expand must be UTF-8 encoded even if #G_REGEX_RAW was
868 * passed to g_regex_new().
869 *
870 * The backreferences are extracted from the string passed to the match
871 * function, so you cannot call this function after freeing the string.
872 *
873 * @match_info may be %NULL in which case @string_to_expand must not
874 * contain references. For instance "foo\n" does not refer to an actual
875 * pattern and '\n' merely will be replaced with \n character,
876 * while to expand "\0" (whole match) one needs the result of a match.
877 * Use g_regex_check_replacement() to find out whether @string_to_expand
878 * contains references.
879 *
880 * Returns: (allow-none): the expanded string, or %NULL if an error occurred
881 *
882 * Since: 2.14
883 */
884 gchar *
885 g_match_info_expand_references (const GMatchInfo *match_info,
886 const gchar *string_to_expand,
887 GError **error)
888 {
889 GString *result;
890 GList *list;
891 GError *tmp_error = NULL;
892
893 g_return_val_if_fail (string_to_expand != NULL, NULL);
894 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
895
896 list = split_replacement (string_to_expand, &tmp_error);
897 if (tmp_error != NULL)
898 {
899 g_propagate_error (error, tmp_error);
900 return NULL;
901 }
902
903 if (!match_info && interpolation_list_needs_match (list))
904 {
905 g_critical ("String '%s' contains references to the match, can't "
906 "expand references without GMatchInfo object",
907 string_to_expand);
908 return NULL;
909 }
910
911 result = g_string_sized_new (strlen (string_to_expand));
912 interpolate_replacement (match_info, result, list);
913
914 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
915
916 return g_string_free (result, FALSE);
917 }
918
919 /**
920 * g_match_info_fetch:
921 * @match_info: #GMatchInfo structure
922 * @match_num: number of the sub expression
923 *
924 * Retrieves the text matching the @match_num<!-- -->'th capturing
925 * parentheses. 0 is the full text of the match, 1 is the first paren
926 * set, 2 the second, and so on.
927 *
928 * If @match_num is a valid sub pattern but it didn't match anything
929 * (e.g. sub pattern 1, matching "b" against "(a)?b") then an empty
930 * string is returned.
931 *
932 * If the match was obtained using the DFA algorithm, that is using
933 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
934 * string is not that of a set of parentheses but that of a matched
935 * substring. Substrings are matched in reverse order of length, so
936 * 0 is the longest match.
937 *
938 * The string is fetched from the string passed to the match function,
939 * so you cannot call this function after freeing the string.
940 *
941 * Returns: (allow-none): The matched substring, or %NULL if an error
942 * occurred. You have to free the string yourself
943 *
944 * Since: 2.14
945 */
946 gchar *
947 g_match_info_fetch (const GMatchInfo *match_info,
948 gint match_num)
949 {
950 /* we cannot use pcre_get_substring() because it allocates the
951 * string using pcre_malloc(). */
952 gchar *match = NULL;
953 gint start, end;
954
955 g_return_val_if_fail (match_info != NULL, NULL);
956 g_return_val_if_fail (match_num >= 0, NULL);
957
958 /* match_num does not exist or it didn't matched, i.e. matching "b"
959 * against "(a)?b" then group 0 is empty. */
960 if (!g_match_info_fetch_pos (match_info, match_num, &start, &end))
961 match = NULL;
962 else if (start == -1)
963 match = g_strdup ("");
964 else
965 match = g_strndup (&match_info->string[start], end - start);
966
967 return match;
968 }
969
970 /**
971 * g_match_info_fetch_pos:
972 * @match_info: #GMatchInfo structure
973 * @match_num: number of the sub expression
974 * @start_pos: (out) (allow-none): pointer to location where to store
975 * the start position, or %NULL
976 * @end_pos: (out) (allow-none): pointer to location where to store
977 * the end position, or %NULL
978 *
979 * Retrieves the position in bytes of the @match_num<!-- -->'th capturing
980 * parentheses. 0 is the full text of the match, 1 is the first
981 * paren set, 2 the second, and so on.
982 *
983 * If @match_num is a valid sub pattern but it didn't match anything
984 * (e.g. sub pattern 1, matching "b" against "(a)?b") then @start_pos
985 * and @end_pos are set to -1 and %TRUE is returned.
986 *
987 * If the match was obtained using the DFA algorithm, that is using
988 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
989 * position is not that of a set of parentheses but that of a matched
990 * substring. Substrings are matched in reverse order of length, so
991 * 0 is the longest match.
992 *
993 * Returns: %TRUE if the position was fetched, %FALSE otherwise. If
994 * the position cannot be fetched, @start_pos and @end_pos are left
995 * unchanged
996 *
997 * Since: 2.14
998 */
999 gboolean
1000 g_match_info_fetch_pos (const GMatchInfo *match_info,
1001 gint match_num,
1002 gint *start_pos,
1003 gint *end_pos)
1004 {
1005 g_return_val_if_fail (match_info != NULL, FALSE);
1006 g_return_val_if_fail (match_num >= 0, FALSE);
1007
1008 /* make sure the sub expression number they're requesting is less than
1009 * the total number of sub expressions that were matched. */
1010 if (match_num >= match_info->matches)
1011 return FALSE;
1012
1013 if (start_pos != NULL)
1014 *start_pos = match_info->offsets[2 * match_num];
1015
1016 if (end_pos != NULL)
1017 *end_pos = match_info->offsets[2 * match_num + 1];
1018
1019 return TRUE;
1020 }
1021
1022 /*
1023 * Returns number of first matched subpattern with name @name.
1024 * There may be more than one in case when DUPNAMES is used,
1025 * and not all subpatterns with that name match;
1026 * pcre_get_stringnumber() does not work in that case.
1027 */
1028 static gint
1029 get_matched_substring_number (const GMatchInfo *match_info,
1030 const gchar *name)
1031 {
1032 gint entrysize;
1033 gchar *first, *last;
1034 guchar *entry;
1035
1036 if (!(match_info->regex->compile_opts & G_REGEX_DUPNAMES))
1037 return pcre_get_stringnumber (match_info->regex->pcre_re, name);
1038
1039 /* This code is copied from pcre_get.c: get_first_set() */
1040 entrysize = pcre_get_stringtable_entries (match_info->regex->pcre_re,
1041 name,
1042 &first,
1043 &last);
1044
1045 if (entrysize <= 0)
1046 return entrysize;
1047
1048 for (entry = (guchar*) first; entry <= (guchar*) last; entry += entrysize)
1049 {
1050 gint n = (entry[0] << 8) + entry[1];
1051 if (match_info->offsets[n*2] >= 0)
1052 return n;
1053 }
1054
1055 return (first[0] << 8) + first[1];
1056 }
1057
1058 /**
1059 * g_match_info_fetch_named:
1060 * @match_info: #GMatchInfo structure
1061 * @name: name of the subexpression
1062 *
1063 * Retrieves the text matching the capturing parentheses named @name.
1064 *
1065 * If @name is a valid sub pattern name but it didn't match anything
1066 * (e.g. sub pattern "X", matching "b" against "(?P&lt;X&gt;a)?b")
1067 * then an empty string is returned.
1068 *
1069 * The string is fetched from the string passed to the match function,
1070 * so you cannot call this function after freeing the string.
1071 *
1072 * Returns: (allow-none): The matched substring, or %NULL if an error
1073 * occurred. You have to free the string yourself
1074 *
1075 * Since: 2.14
1076 */
1077 gchar *
1078 g_match_info_fetch_named (const GMatchInfo *match_info,
1079 const gchar *name)
1080 {
1081 /* we cannot use pcre_get_named_substring() because it allocates the
1082 * string using pcre_malloc(). */
1083 gint num;
1084
1085 g_return_val_if_fail (match_info != NULL, NULL);
1086 g_return_val_if_fail (name != NULL, NULL);
1087
1088 num = get_matched_substring_number (match_info, name);
1089 if (num < 0)
1090 return NULL;
1091 else
1092 return g_match_info_fetch (match_info, num);
1093 }
1094
1095 /**
1096 * g_match_info_fetch_named_pos:
1097 * @match_info: #GMatchInfo structure
1098 * @name: name of the subexpression
1099 * @start_pos: (out) (allow-none): pointer to location where to store
1100 * the start position, or %NULL
1101 * @end_pos: (out) (allow-none): pointer to location where to store
1102 * the end position, or %NULL
1103 *
1104 * Retrieves the position in bytes of the capturing parentheses named @name.
1105 *
1106 * If @name is a valid sub pattern name but it didn't match anything
1107 * (e.g. sub pattern "X", matching "b" against "(?P&lt;X&gt;a)?b")
1108 * then @start_pos and @end_pos are set to -1 and %TRUE is returned.
1109 *
1110 * Returns: %TRUE if the position was fetched, %FALSE otherwise.
1111 * If the position cannot be fetched, @start_pos and @end_pos
1112 * are left unchanged.
1113 *
1114 * Since: 2.14
1115 */
1116 gboolean
1117 g_match_info_fetch_named_pos (const GMatchInfo *match_info,
1118 const gchar *name,
1119 gint *start_pos,
1120 gint *end_pos)
1121 {
1122 gint num;
1123
1124 g_return_val_if_fail (match_info != NULL, FALSE);
1125 g_return_val_if_fail (name != NULL, FALSE);
1126
1127 num = get_matched_substring_number (match_info, name);
1128 if (num < 0)
1129 return FALSE;
1130
1131 return g_match_info_fetch_pos (match_info, num, start_pos, end_pos);
1132 }
1133
1134 /**
1135 * g_match_info_fetch_all:
1136 * @match_info: a #GMatchInfo structure
1137 *
1138 * Bundles up pointers to each of the matching substrings from a match
1139 * and stores them in an array of gchar pointers. The first element in
1140 * the returned array is the match number 0, i.e. the entire matched
1141 * text.
1142 *
1143 * If a sub pattern didn't match anything (e.g. sub pattern 1, matching
1144 * "b" against "(a)?b") then an empty string is inserted.
1145 *
1146 * If the last match was obtained using the DFA algorithm, that is using
1147 * g_regex_match_all() or g_regex_match_all_full(), the retrieved
1148 * strings are not that matched by sets of parentheses but that of the
1149 * matched substring. Substrings are matched in reverse order of length,
1150 * so the first one is the longest match.
1151 *
1152 * The strings are fetched from the string passed to the match function,
1153 * so you cannot call this function after freeing the string.
1154 *
1155 * Returns: (allow-none): a %NULL-terminated array of gchar * pointers.
1156 * It must be freed using g_strfreev(). If the previous match failed
1157 * %NULL is returned
1158 *
1159 * Since: 2.14
1160 */
1161 gchar **
1162 g_match_info_fetch_all (const GMatchInfo *match_info)
1163 {
1164 /* we cannot use pcre_get_substring_list() because the returned value
1165 * isn't suitable for g_strfreev(). */
1166 gchar **result;
1167 gint i;
1168
1169 g_return_val_if_fail (match_info != NULL, NULL);
1170
1171 if (match_info->matches < 0)
1172 return NULL;
1173
1174 result = g_new (gchar *, match_info->matches + 1);
1175 for (i = 0; i < match_info->matches; i++)
1176 result[i] = g_match_info_fetch (match_info, i);
1177 result[i] = NULL;
1178
1179 return result;
1180 }
1181
1182
1183 /* GRegex */
1184
1185 GQuark
1186 g_regex_error_quark (void)
1187 {
1188 static GQuark error_quark = 0;
1189
1190 if (error_quark == 0)
1191 error_quark = g_quark_from_static_string ("g-regex-error-quark");
1192
1193 return error_quark;
1194 }
1195
1196 /**
1197 * g_regex_ref:
1198 * @regex: a #GRegex
1199 *
1200 * Increases reference count of @regex by 1.
1201 *
1202 * Returns: @regex
1203 *
1204 * Since: 2.14
1205 */
1206 GRegex *
1207 g_regex_ref (GRegex *regex)
1208 {
1209 g_return_val_if_fail (regex != NULL, NULL);
1210 g_atomic_int_inc (&regex->ref_count);
1211 return regex;
1212 }
1213
1214 /**
1215 * g_regex_unref:
1216 * @regex: a #GRegex
1217 *
1218 * Decreases reference count of @regex by 1. When reference count drops
1219 * to zero, it frees all the memory associated with the regex structure.
1220 *
1221 * Since: 2.14
1222 */
1223 void
1224 g_regex_unref (GRegex *regex)
1225 {
1226 g_return_if_fail (regex != NULL);
1227
1228 if (g_atomic_int_dec_and_test (&regex->ref_count))
1229 {
1230 g_free (regex->pattern);
1231 if (regex->pcre_re != NULL)
1232 pcre_free (regex->pcre_re);
1233 if (regex->extra != NULL)
1234 pcre_free (regex->extra);
1235 g_free (regex);
1236 }
1237 }
1238
1239 /**
1240 * g_regex_new:
1241 * @pattern: the regular expression
1242 * @compile_options: compile options for the regular expression, or 0
1243 * @match_options: match options for the regular expression, or 0
1244 * @error: return location for a #GError
1245 *
1246 * Compiles the regular expression to an internal form, and does
1247 * the initial setup of the #GRegex structure.
1248 *
1249 * Returns: a #GRegex structure. Call g_regex_unref() when you
1250 * are done with it
1251 *
1252 * Since: 2.14
1253 */
1254 GRegex *
1255 g_regex_new (const gchar *pattern,
1256 GRegexCompileFlags compile_options,
1257 GRegexMatchFlags match_options,
1258 GError **error)
1259 {
1260 GRegex *regex;
1261 pcre *re;
1262 const gchar *errmsg;
1263 gint erroffset;
1264 gint errcode;
1265 gboolean optimize = FALSE;
1266 static gsize initialised;
1267 unsigned long int pcre_compile_options;
1268
1269 g_return_val_if_fail (pattern != NULL, NULL);
1270 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
1271 g_return_val_if_fail ((compile_options & ~G_REGEX_COMPILE_MASK) == 0, NULL);
1272 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
1273
1274 if (g_once_init_enter (&initialised))
1275 {
1276 gint support;
1277 const gchar *msg;
1278
1279 pcre_config (PCRE_CONFIG_UTF8, &support);
1280 if (!support)
1281 {
1282 msg = N_("PCRE library is compiled without UTF8 support");
1283 g_critical ("%s", msg);
1284 g_set_error_literal (error, G_REGEX_ERROR, G_REGEX_ERROR_COMPILE, gettext (msg));
1285 return NULL;
1286 }
1287
1288 pcre_config (PCRE_CONFIG_UNICODE_PROPERTIES, &support);
1289 if (!support)
1290 {
1291 msg = N_("PCRE library is compiled without UTF8 properties support");
1292 g_critical ("%s", msg);
1293 g_set_error_literal (error, G_REGEX_ERROR, G_REGEX_ERROR_COMPILE, gettext (msg));
1294 return NULL;
1295 }
1296
1297 g_once_init_leave (&initialised, TRUE);
1298 }
1299
1300 /* G_REGEX_OPTIMIZE has the same numeric value of PCRE_NO_UTF8_CHECK,
1301 * as we do not need to wrap PCRE_NO_UTF8_CHECK. */
1302 if (compile_options & G_REGEX_OPTIMIZE)
1303 optimize = TRUE;
1304
1305 /* In GRegex the string are, by default, UTF-8 encoded. PCRE
1306 * instead uses UTF-8 only if required with PCRE_UTF8. */
1307 if (compile_options & G_REGEX_RAW)
1308 {
1309 /* disable utf-8 */
1310 compile_options &= ~G_REGEX_RAW;
1311 }
1312 else
1313 {
1314 /* enable utf-8 */
1315 compile_options |= PCRE_UTF8 | PCRE_NO_UTF8_CHECK;
1316 match_options |= PCRE_NO_UTF8_CHECK;
1317 }
1318
1319 /* PCRE_NEWLINE_ANY is the default for the internal PCRE but
1320 * not for the system one. */
1321 if (!(compile_options & G_REGEX_NEWLINE_CR) &&
1322 !(compile_options & G_REGEX_NEWLINE_LF))
1323 {
1324 compile_options |= PCRE_NEWLINE_ANY;
1325 }
1326
1327 compile_options |= PCRE_UCP;
1328
1329 /* compile the pattern */
1330 re = pcre_compile2 (pattern, compile_options, &errcode,
1331 &errmsg, &erroffset, NULL);
1332
1333 /* if the compilation failed, set the error member and return
1334 * immediately */
1335 if (re == NULL)
1336 {
1337 GError *tmp_error;
1338
1339 /* Translate the PCRE error code to GRegexError and use a translated
1340 * error message if possible */
1341 translate_compile_error (&errcode, &errmsg);
1342
1343 /* PCRE uses byte offsets but we want to show character offsets */
1344 erroffset = g_utf8_pointer_to_offset (pattern, &pattern[erroffset]);
1345
1346 tmp_error = g_error_new (G_REGEX_ERROR, errcode,
1347 _("Error while compiling regular "
1348 "expression %s at char %d: %s"),
1349 pattern, erroffset, errmsg);
1350 g_propagate_error (error, tmp_error);
1351
1352 return NULL;
1353 }
1354
1355 /* For options set at the beginning of the pattern, pcre puts them into
1356 * compile options, e.g. "(?i)foo" will make the pcre structure store
1357 * PCRE_CASELESS even though it wasn't explicitly given for compilation. */
1358 pcre_fullinfo (re, NULL, PCRE_INFO_OPTIONS, &pcre_compile_options);
1359 compile_options = pcre_compile_options;
1360
1361 if (!(compile_options & G_REGEX_DUPNAMES))
1362 {
1363 gboolean jchanged = FALSE;
1364 pcre_fullinfo (re, NULL, PCRE_INFO_JCHANGED, &jchanged);
1365 if (jchanged)
1366 compile_options |= G_REGEX_DUPNAMES;
1367 }
1368
1369 regex = g_new0 (GRegex, 1);
1370 regex->ref_count = 1;
1371 regex->pattern = g_strdup (pattern);
1372 regex->pcre_re = re;
1373 regex->compile_opts = compile_options;
1374 regex->match_opts = match_options;
1375
1376 if (optimize)
1377 {
1378 regex->extra = pcre_study (regex->pcre_re, 0, &errmsg);
1379 if (errmsg != NULL)
1380 {
1381 GError *tmp_error = g_error_new (G_REGEX_ERROR,
1382 G_REGEX_ERROR_OPTIMIZE,
1383 _("Error while optimizing "
1384 "regular expression %s: %s"),
1385 regex->pattern,
1386 errmsg);
1387 g_propagate_error (error, tmp_error);
1388
1389 g_regex_unref (regex);
1390 return NULL;
1391 }
1392 }
1393
1394 return regex;
1395 }
1396
1397 /**
1398 * g_regex_get_pattern:
1399 * @regex: a #GRegex structure
1400 *
1401 * Gets the pattern string associated with @regex, i.e. a copy of
1402 * the string passed to g_regex_new().
1403 *
1404 * Returns: the pattern of @regex
1405 *
1406 * Since: 2.14
1407 */
1408 const gchar *
1409 g_regex_get_pattern (const GRegex *regex)
1410 {
1411 g_return_val_if_fail (regex != NULL, NULL);
1412
1413 return regex->pattern;
1414 }
1415
1416 /**
1417 * g_regex_get_max_backref:
1418 * @regex: a #GRegex
1419 *
1420 * Returns the number of the highest back reference
1421 * in the pattern, or 0 if the pattern does not contain
1422 * back references.
1423 *
1424 * Returns: the number of the highest back reference
1425 *
1426 * Since: 2.14
1427 */
1428 gint
1429 g_regex_get_max_backref (const GRegex *regex)
1430 {
1431 gint value;
1432
1433 pcre_fullinfo (regex->pcre_re, regex->extra,
1434 PCRE_INFO_BACKREFMAX, &value);
1435
1436 return value;
1437 }
1438
1439 /**
1440 * g_regex_get_capture_count:
1441 * @regex: a #GRegex
1442 *
1443 * Returns the number of capturing subpatterns in the pattern.
1444 *
1445 * Returns: the number of capturing subpatterns
1446 *
1447 * Since: 2.14
1448 */
1449 gint
1450 g_regex_get_capture_count (const GRegex *regex)
1451 {
1452 gint value;
1453
1454 pcre_fullinfo (regex->pcre_re, regex->extra,
1455 PCRE_INFO_CAPTURECOUNT, &value);
1456
1457 return value;
1458 }
1459
1460 /**
1461 * g_regex_get_compile_flags:
1462 * @regex: a #GRegex
1463 *
1464 * Returns the compile options that @regex was created with.
1465 *
1466 * Returns: flags from #GRegexCompileFlags
1467 *
1468 * Since: 2.26
1469 */
1470 GRegexCompileFlags
1471 g_regex_get_compile_flags (const GRegex *regex)
1472 {
1473 g_return_val_if_fail (regex != NULL, 0);
1474
1475 return regex->compile_opts;
1476 }
1477
1478 /**
1479 * g_regex_get_match_flags:
1480 * @regex: a #GRegex
1481 *
1482 * Returns the match options that @regex was created with.
1483 *
1484 * Returns: flags from #GRegexMatchFlags
1485 *
1486 * Since: 2.26
1487 */
1488 GRegexMatchFlags
1489 g_regex_get_match_flags (const GRegex *regex)
1490 {
1491 g_return_val_if_fail (regex != NULL, 0);
1492
1493 return regex->match_opts;
1494 }
1495
1496 /**
1497 * g_regex_match_simple:
1498 * @pattern: the regular expression
1499 * @string: the string to scan for matches
1500 * @compile_options: compile options for the regular expression, or 0
1501 * @match_options: match options, or 0
1502 *
1503 * Scans for a match in @string for @pattern.
1504 *
1505 * This function is equivalent to g_regex_match() but it does not
1506 * require to compile the pattern with g_regex_new(), avoiding some
1507 * lines of code when you need just to do a match without extracting
1508 * substrings, capture counts, and so on.
1509 *
1510 * If this function is to be called on the same @pattern more than
1511 * once, it's more efficient to compile the pattern once with
1512 * g_regex_new() and then use g_regex_match().
1513 *
1514 * Returns: %TRUE if the string matched, %FALSE otherwise
1515 *
1516 * Since: 2.14
1517 */
1518 gboolean
1519 g_regex_match_simple (const gchar *pattern,
1520 const gchar *string,
1521 GRegexCompileFlags compile_options,
1522 GRegexMatchFlags match_options)
1523 {
1524 GRegex *regex;
1525 gboolean result;
1526
1527 regex = g_regex_new (pattern, compile_options, 0, NULL);
1528 if (!regex)
1529 return FALSE;
1530 result = g_regex_match_full (regex, string, -1, 0, match_options, NULL, NULL);
1531 g_regex_unref (regex);
1532 return result;
1533 }
1534
1535 /**
1536 * g_regex_match:
1537 * @regex: a #GRegex structure from g_regex_new()
1538 * @string: the string to scan for matches
1539 * @match_options: match options
1540 * @match_info: (out) (allow-none): pointer to location where to store
1541 * the #GMatchInfo, or %NULL if you do not need it
1542 *
1543 * Scans for a match in string for the pattern in @regex.
1544 * The @match_options are combined with the match options specified
1545 * when the @regex structure was created, letting you have more
1546 * flexibility in reusing #GRegex structures.
1547 *
1548 * A #GMatchInfo structure, used to get information on the match,
1549 * is stored in @match_info if not %NULL. Note that if @match_info
1550 * is not %NULL then it is created even if the function returns %FALSE,
1551 * i.e. you must free it regardless if regular expression actually matched.
1552 *
1553 * To retrieve all the non-overlapping matches of the pattern in
1554 * string you can use g_match_info_next().
1555 *
1556 * |[
1557 * static void
1558 * print_uppercase_words (const gchar *string)
1559 * {
1560 * /&ast; Print all uppercase-only words. &ast;/
1561 * GRegex *regex;
1562 * GMatchInfo *match_info;
1563 * &nbsp;
1564 * regex = g_regex_new ("[A-Z]+", 0, 0, NULL);
1565 * g_regex_match (regex, string, 0, &amp;match_info);
1566 * while (g_match_info_matches (match_info))
1567 * {
1568 * gchar *word = g_match_info_fetch (match_info, 0);
1569 * g_print ("Found: %s\n", word);
1570 * g_free (word);
1571 * g_match_info_next (match_info, NULL);
1572 * }
1573 * g_match_info_free (match_info);
1574 * g_regex_unref (regex);
1575 * }
1576 * ]|
1577 *
1578 * @string is not copied and is used in #GMatchInfo internally. If
1579 * you use any #GMatchInfo method (except g_match_info_free()) after
1580 * freeing or modifying @string then the behaviour is undefined.
1581 *
1582 * Returns: %TRUE is the string matched, %FALSE otherwise
1583 *
1584 * Since: 2.14
1585 */
1586 gboolean
1587 g_regex_match (const GRegex *regex,
1588 const gchar *string,
1589 GRegexMatchFlags match_options,
1590 GMatchInfo **match_info)
1591 {
1592 return g_regex_match_full (regex, string, -1, 0, match_options,
1593 match_info, NULL);
1594 }
1595
1596 /**
1597 * g_regex_match_full:
1598 * @regex: a #GRegex structure from g_regex_new()
1599 * @string: (array length=string_len): the string to scan for matches
1600 * @string_len: the length of @string, or -1 if @string is nul-terminated
1601 * @start_position: starting index of the string to match
1602 * @match_options: match options
1603 * @match_info: (out) (allow-none): pointer to location where to store
1604 * the #GMatchInfo, or %NULL if you do not need it
1605 * @error: location to store the error occurring, or %NULL to ignore errors
1606 *
1607 * Scans for a match in string for the pattern in @regex.
1608 * The @match_options are combined with the match options specified
1609 * when the @regex structure was created, letting you have more
1610 * flexibility in reusing #GRegex structures.
1611 *
1612 * Setting @start_position differs from just passing over a shortened
1613 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
1614 * that begins with any kind of lookbehind assertion, such as "\b".
1615 *
1616 * A #GMatchInfo structure, used to get information on the match, is
1617 * stored in @match_info if not %NULL. Note that if @match_info is
1618 * not %NULL then it is created even if the function returns %FALSE,
1619 * i.e. you must free it regardless if regular expression actually
1620 * matched.
1621 *
1622 * @string is not copied and is used in #GMatchInfo internally. If
1623 * you use any #GMatchInfo method (except g_match_info_free()) after
1624 * freeing or modifying @string then the behaviour is undefined.
1625 *
1626 * To retrieve all the non-overlapping matches of the pattern in
1627 * string you can use g_match_info_next().
1628 *
1629 * |[
1630 * static void
1631 * print_uppercase_words (const gchar *string)
1632 * {
1633 * /&ast; Print all uppercase-only words. &ast;/
1634 * GRegex *regex;
1635 * GMatchInfo *match_info;
1636 * GError *error = NULL;
1637 * &nbsp;
1638 * regex = g_regex_new ("[A-Z]+", 0, 0, NULL);
1639 * g_regex_match_full (regex, string, -1, 0, 0, &amp;match_info, &amp;error);
1640 * while (g_match_info_matches (match_info))
1641 * {
1642 * gchar *word = g_match_info_fetch (match_info, 0);
1643 * g_print ("Found: %s\n", word);
1644 * g_free (word);
1645 * g_match_info_next (match_info, &amp;error);
1646 * }
1647 * g_match_info_free (match_info);
1648 * g_regex_unref (regex);
1649 * if (error != NULL)
1650 * {
1651 * g_printerr ("Error while matching: %s\n", error->message);
1652 * g_error_free (error);
1653 * }
1654 * }
1655 * ]|
1656 *
1657 * Returns: %TRUE is the string matched, %FALSE otherwise
1658 *
1659 * Since: 2.14
1660 */
1661 gboolean
1662 g_regex_match_full (const GRegex *regex,
1663 const gchar *string,
1664 gssize string_len,
1665 gint start_position,
1666 GRegexMatchFlags match_options,
1667 GMatchInfo **match_info,
1668 GError **error)
1669 {
1670 GMatchInfo *info;
1671 gboolean match_ok;
1672
1673 g_return_val_if_fail (regex != NULL, FALSE);
1674 g_return_val_if_fail (string != NULL, FALSE);
1675 g_return_val_if_fail (start_position >= 0, FALSE);
1676 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
1677 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, FALSE);
1678
1679 info = match_info_new (regex, string, string_len, start_position,
1680 match_options, FALSE);
1681 match_ok = g_match_info_next (info, error);
1682 if (match_info != NULL)
1683 *match_info = info;
1684 else
1685 g_match_info_free (info);
1686
1687 return match_ok;
1688 }
1689
1690 /**
1691 * g_regex_match_all:
1692 * @regex: a #GRegex structure from g_regex_new()
1693 * @string: the string to scan for matches
1694 * @match_options: match options
1695 * @match_info: (out) (allow-none): pointer to location where to store
1696 * the #GMatchInfo, or %NULL if you do not need it
1697 *
1698 * Using the standard algorithm for regular expression matching only
1699 * the longest match in the string is retrieved. This function uses
1700 * a different algorithm so it can retrieve all the possible matches.
1701 * For more documentation see g_regex_match_all_full().
1702 *
1703 * A #GMatchInfo structure, used to get information on the match, is
1704 * stored in @match_info if not %NULL. Note that if @match_info is
1705 * not %NULL then it is created even if the function returns %FALSE,
1706 * i.e. you must free it regardless if regular expression actually
1707 * matched.
1708 *
1709 * @string is not copied and is used in #GMatchInfo internally. If
1710 * you use any #GMatchInfo method (except g_match_info_free()) after
1711 * freeing or modifying @string then the behaviour is undefined.
1712 *
1713 * Returns: %TRUE is the string matched, %FALSE otherwise
1714 *
1715 * Since: 2.14
1716 */
1717 gboolean
1718 g_regex_match_all (const GRegex *regex,
1719 const gchar *string,
1720 GRegexMatchFlags match_options,
1721 GMatchInfo **match_info)
1722 {
1723 return g_regex_match_all_full (regex, string, -1, 0, match_options,
1724 match_info, NULL);
1725 }
1726
1727 /**
1728 * g_regex_match_all_full:
1729 * @regex: a #GRegex structure from g_regex_new()
1730 * @string: (array length=string_len): the string to scan for matches
1731 * @string_len: the length of @string, or -1 if @string is nul-terminated
1732 * @start_position: starting index of the string to match
1733 * @match_options: match options
1734 * @match_info: (out) (allow-none): pointer to location where to store
1735 * the #GMatchInfo, or %NULL if you do not need it
1736 * @error: location to store the error occurring, or %NULL to ignore errors
1737 *
1738 * Using the standard algorithm for regular expression matching only
1739 * the longest match in the string is retrieved, it is not possible
1740 * to obtain all the available matches. For instance matching
1741 * "&lt;a&gt; &lt;b&gt; &lt;c&gt;" against the pattern "&lt;.*&gt;"
1742 * you get "&lt;a&gt; &lt;b&gt; &lt;c&gt;".
1743 *
1744 * This function uses a different algorithm (called DFA, i.e. deterministic
1745 * finite automaton), so it can retrieve all the possible matches, all
1746 * starting at the same point in the string. For instance matching
1747 * "&lt;a&gt; &lt;b&gt; &lt;c&gt;" against the pattern "&lt;.*&gt;"
1748 * you would obtain three matches: "&lt;a&gt; &lt;b&gt; &lt;c&gt;",
1749 * "&lt;a&gt; &lt;b&gt;" and "&lt;a&gt;".
1750 *
1751 * The number of matched strings is retrieved using
1752 * g_match_info_get_match_count(). To obtain the matched strings and
1753 * their position you can use, respectively, g_match_info_fetch() and
1754 * g_match_info_fetch_pos(). Note that the strings are returned in
1755 * reverse order of length; that is, the longest matching string is
1756 * given first.
1757 *
1758 * Note that the DFA algorithm is slower than the standard one and it
1759 * is not able to capture substrings, so backreferences do not work.
1760 *
1761 * Setting @start_position differs from just passing over a shortened
1762 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
1763 * that begins with any kind of lookbehind assertion, such as "\b".
1764 *
1765 * A #GMatchInfo structure, used to get information on the match, is
1766 * stored in @match_info if not %NULL. Note that if @match_info is
1767 * not %NULL then it is created even if the function returns %FALSE,
1768 * i.e. you must free it regardless if regular expression actually
1769 * matched.
1770 *
1771 * @string is not copied and is used in #GMatchInfo internally. If
1772 * you use any #GMatchInfo method (except g_match_info_free()) after
1773 * freeing or modifying @string then the behaviour is undefined.
1774 *
1775 * Returns: %TRUE is the string matched, %FALSE otherwise
1776 *
1777 * Since: 2.14
1778 */
1779 gboolean
1780 g_regex_match_all_full (const GRegex *regex,
1781 const gchar *string,
1782 gssize string_len,
1783 gint start_position,
1784 GRegexMatchFlags match_options,
1785 GMatchInfo **match_info,
1786 GError **error)
1787 {
1788 GMatchInfo *info;
1789 gboolean done;
1790
1791 g_return_val_if_fail (regex != NULL, FALSE);
1792 g_return_val_if_fail (string != NULL, FALSE);
1793 g_return_val_if_fail (start_position >= 0, FALSE);
1794 g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
1795 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, FALSE);
1796
1797 info = match_info_new (regex, string, string_len, start_position,
1798 match_options, TRUE);
1799
1800 done = FALSE;
1801 while (!done)
1802 {
1803 done = TRUE;
1804 info->matches = pcre_dfa_exec (regex->pcre_re, regex->extra,
1805 info->string, info->string_len,
1806 info->pos,
1807 regex->match_opts | match_options,
1808 info->offsets, info->n_offsets,
1809 info->workspace, info->n_workspace);
1810 if (info->matches == PCRE_ERROR_DFA_WSSIZE)
1811 {
1812 /* info->workspace is too small. */
1813 info->n_workspace *= 2;
1814 info->workspace = g_realloc (info->workspace,
1815 info->n_workspace * sizeof (gint));
1816 done = FALSE;
1817 }
1818 else if (info->matches == 0)
1819 {
1820 /* info->offsets is too small. */
1821 info->n_offsets *= 2;
1822 info->offsets = g_realloc (info->offsets,
1823 info->n_offsets * sizeof (gint));
1824 done = FALSE;
1825 }
1826 else if (IS_PCRE_ERROR (info->matches))
1827 {
1828 g_set_error (error, G_REGEX_ERROR, G_REGEX_ERROR_MATCH,
1829 _("Error while matching regular expression %s: %s"),
1830 regex->pattern, match_error (info->matches));
1831 }
1832 }
1833
1834 /* set info->pos to -1 so that a call to g_match_info_next() fails. */
1835 info->pos = -1;
1836
1837 if (match_info != NULL)
1838 *match_info = info;
1839 else
1840 g_match_info_free (info);
1841
1842 return info->matches >= 0;
1843 }
1844
1845 /**
1846 * g_regex_get_string_number:
1847 * @regex: #GRegex structure
1848 * @name: name of the subexpression
1849 *
1850 * Retrieves the number of the subexpression named @name.
1851 *
1852 * Returns: The number of the subexpression or -1 if @name
1853 * does not exists
1854 *
1855 * Since: 2.14
1856 */
1857 gint
1858 g_regex_get_string_number (const GRegex *regex,
1859 const gchar *name)
1860 {
1861 gint num;
1862
1863 g_return_val_if_fail (regex != NULL, -1);
1864 g_return_val_if_fail (name != NULL, -1);
1865
1866 num = pcre_get_stringnumber (regex->pcre_re, name);
1867 if (num == PCRE_ERROR_NOSUBSTRING)
1868 num = -1;
1869
1870 return num;
1871 }
1872
1873 /**
1874 * g_regex_split_simple:
1875 * @pattern: the regular expression
1876 * @string: the string to scan for matches
1877 * @compile_options: compile options for the regular expression, or 0
1878 * @match_options: match options, or 0
1879 *
1880 * Breaks the string on the pattern, and returns an array of
1881 * the tokens. If the pattern contains capturing parentheses,
1882 * then the text for each of the substrings will also be returned.
1883 * If the pattern does not match anywhere in the string, then the
1884 * whole string is returned as the first token.
1885 *
1886 * This function is equivalent to g_regex_split() but it does
1887 * not require to compile the pattern with g_regex_new(), avoiding
1888 * some lines of code when you need just to do a split without
1889 * extracting substrings, capture counts, and so on.
1890 *
1891 * If this function is to be called on the same @pattern more than
1892 * once, it's more efficient to compile the pattern once with
1893 * g_regex_new() and then use g_regex_split().
1894 *
1895 * As a special case, the result of splitting the empty string ""
1896 * is an empty vector, not a vector containing a single string.
1897 * The reason for this special case is that being able to represent
1898 * a empty vector is typically more useful than consistent handling
1899 * of empty elements. If you do need to represent empty elements,
1900 * you'll need to check for the empty string before calling this
1901 * function.
1902 *
1903 * A pattern that can match empty strings splits @string into
1904 * separate characters wherever it matches the empty string between
1905 * characters. For example splitting "ab c" using as a separator
1906 * "\s*", you will get "a", "b" and "c".
1907 *
1908 * Returns: a %NULL-terminated array of strings. Free it using g_strfreev()
1909 *
1910 * Since: 2.14
1911 **/
1912 gchar **
1913 g_regex_split_simple (const gchar *pattern,
1914 const gchar *string,
1915 GRegexCompileFlags compile_options,
1916 GRegexMatchFlags match_options)
1917 {
1918 GRegex *regex;
1919 gchar **result;
1920
1921 regex = g_regex_new (pattern, compile_options, 0, NULL);
1922 if (!regex)
1923 return NULL;
1924
1925 result = g_regex_split_full (regex, string, -1, 0, match_options, 0, NULL);
1926 g_regex_unref (regex);
1927 return result;
1928 }
1929
1930 /**
1931 * g_regex_split:
1932 * @regex: a #GRegex structure
1933 * @string: the string to split with the pattern
1934 * @match_options: match time option flags
1935 *
1936 * Breaks the string on the pattern, and returns an array of the tokens.
1937 * If the pattern contains capturing parentheses, then the text for each
1938 * of the substrings will also be returned. If the pattern does not match
1939 * anywhere in the string, then the whole string is returned as the first
1940 * token.
1941 *
1942 * As a special case, the result of splitting the empty string "" is an
1943 * empty vector, not a vector containing a single string. The reason for
1944 * this special case is that being able to represent a empty vector is
1945 * typically more useful than consistent handling of empty elements. If
1946 * you do need to represent empty elements, you'll need to check for the
1947 * empty string before calling this function.
1948 *
1949 * A pattern that can match empty strings splits @string into separate
1950 * characters wherever it matches the empty string between characters.
1951 * For example splitting "ab c" using as a separator "\s*", you will get
1952 * "a", "b" and "c".
1953 *
1954 * Returns: a %NULL-terminated gchar ** array. Free it using g_strfreev()
1955 *
1956 * Since: 2.14
1957 **/
1958 gchar **
1959 g_regex_split (const GRegex *regex,
1960 const gchar *string,
1961 GRegexMatchFlags match_options)
1962 {
1963 return g_regex_split_full (regex, string, -1, 0,
1964 match_options, 0, NULL);
1965 }
1966
1967 /**
1968 * g_regex_split_full:
1969 * @regex: a #GRegex structure
1970 * @string: (array length=string_len): the string to split with the pattern
1971 * @string_len: the length of @string, or -1 if @string is nul-terminated
1972 * @start_position: starting index of the string to match
1973 * @match_options: match time option flags
1974 * @max_tokens: the maximum number of tokens to split @string into.
1975 * If this is less than 1, the string is split completely
1976 * @error: return location for a #GError
1977 *
1978 * Breaks the string on the pattern, and returns an array of the tokens.
1979 * If the pattern contains capturing parentheses, then the text for each
1980 * of the substrings will also be returned. If the pattern does not match
1981 * anywhere in the string, then the whole string is returned as the first
1982 * token.
1983 *
1984 * As a special case, the result of splitting the empty string "" is an
1985 * empty vector, not a vector containing a single string. The reason for
1986 * this special case is that being able to represent a empty vector is
1987 * typically more useful than consistent handling of empty elements. If
1988 * you do need to represent empty elements, you'll need to check for the
1989 * empty string before calling this function.
1990 *
1991 * A pattern that can match empty strings splits @string into separate
1992 * characters wherever it matches the empty string between characters.
1993 * For example splitting "ab c" using as a separator "\s*", you will get
1994 * "a", "b" and "c".
1995 *
1996 * Setting @start_position differs from just passing over a shortened
1997 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
1998 * that begins with any kind of lookbehind assertion, such as "\b".
1999 *
2000 * Returns: a %NULL-terminated gchar ** array. Free it using g_strfreev()
2001 *
2002 * Since: 2.14
2003 **/
2004 gchar **
2005 g_regex_split_full (const GRegex *regex,
2006 const gchar *string,
2007 gssize string_len,
2008 gint start_position,
2009 GRegexMatchFlags match_options,
2010 gint max_tokens,
2011 GError **error)
2012 {
2013 GError *tmp_error = NULL;
2014 GMatchInfo *match_info;
2015 GList *list, *last;
2016 gint i;
2017 gint token_count;
2018 gboolean match_ok;
2019 /* position of the last separator. */
2020 gint last_separator_end;
2021 /* was the last match 0 bytes long? */
2022 gboolean last_match_is_empty;
2023 /* the returned array of char **s */
2024 gchar **string_list;
2025
2026 g_return_val_if_fail (regex != NULL, NULL);
2027 g_return_val_if_fail (string != NULL, NULL);
2028 g_return_val_if_fail (start_position >= 0, NULL);
2029 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
2030 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2031
2032 if (max_tokens <= 0)
2033 max_tokens = G_MAXINT;
2034
2035 if (string_len < 0)
2036 string_len = strlen (string);
2037
2038 /* zero-length string */
2039 if (string_len - start_position == 0)
2040 return g_new0 (gchar *, 1);
2041
2042 if (max_tokens == 1)
2043 {
2044 string_list = g_new0 (gchar *, 2);
2045 string_list[0] = g_strndup (&string[start_position],
2046 string_len - start_position);
2047 return string_list;
2048 }
2049
2050 list = NULL;
2051 token_count = 0;
2052 last_separator_end = start_position;
2053 last_match_is_empty = FALSE;
2054
2055 match_ok = g_regex_match_full (regex, string, string_len, start_position,
2056 match_options, &match_info, &tmp_error);
2057
2058 while (tmp_error == NULL)
2059 {
2060 if (match_ok)
2061 {
2062 last_match_is_empty =
2063 (match_info->offsets[0] == match_info->offsets[1]);
2064
2065 /* we need to skip empty separators at the same position of the end
2066 * of another separator. e.g. the string is "a b" and the separator
2067 * is " *", so from 1 to 2 we have a match and at position 2 we have
2068 * an empty match. */
2069 if (last_separator_end != match_info->offsets[1])
2070 {
2071 gchar *token;
2072 gint match_count;
2073
2074 token = g_strndup (string + last_separator_end,
2075 match_info->offsets[0] - last_separator_end);
2076 list = g_list_prepend (list, token);
2077 token_count++;
2078
2079 /* if there were substrings, these need to be added to
2080 * the list. */
2081 match_count = g_match_info_get_match_count (match_info);
2082 if (match_count > 1)
2083 {
2084 for (i = 1; i < match_count; i++)
2085 list = g_list_prepend (list, g_match_info_fetch (match_info, i));
2086 }
2087 }
2088 }
2089 else
2090 {
2091 /* if there was no match, copy to end of string. */
2092 if (!last_match_is_empty)
2093 {
2094 gchar *token = g_strndup (string + last_separator_end,
2095 match_info->string_len - last_separator_end);
2096 list = g_list_prepend (list, token);
2097 }
2098 /* no more tokens, end the loop. */
2099 break;
2100 }
2101
2102 /* -1 to leave room for the last part. */
2103 if (token_count >= max_tokens - 1)
2104 {
2105 /* we have reached the maximum number of tokens, so we copy
2106 * the remaining part of the string. */
2107 if (last_match_is_empty)
2108 {
2109 /* the last match was empty, so we have moved one char
2110 * after the real position to avoid empty matches at the
2111 * same position. */
2112 match_info->pos = PREV_CHAR (regex, &string[match_info->pos]) - string;
2113 }
2114 /* the if is needed in the case we have terminated the available
2115 * tokens, but we are at the end of the string, so there are no
2116 * characters left to copy. */
2117 if (string_len > match_info->pos)
2118 {
2119 gchar *token = g_strndup (string + match_info->pos,
2120 string_len - match_info->pos);
2121 list = g_list_prepend (list, token);
2122 }
2123 /* end the loop. */
2124 break;
2125 }
2126
2127 last_separator_end = match_info->pos;
2128 if (last_match_is_empty)
2129 /* if the last match was empty, g_match_info_next() has moved
2130 * forward to avoid infinite loops, but we still need to copy that
2131 * character. */
2132 last_separator_end = PREV_CHAR (regex, &string[last_separator_end]) - string;
2133
2134 match_ok = g_match_info_next (match_info, &tmp_error);
2135 }
2136 g_match_info_free (match_info);
2137 if (tmp_error != NULL)
2138 {
2139 g_propagate_error (error, tmp_error);
2140 g_list_free_full (list, g_free);
2141 match_info->pos = -1;
2142 return NULL;
2143 }
2144
2145 string_list = g_new (gchar *, g_list_length (list) + 1);
2146 i = 0;
2147 for (last = g_list_last (list); last; last = g_list_previous (last))
2148 string_list[i++] = last->data;
2149 string_list[i] = NULL;
2150 g_list_free (list);
2151
2152 return string_list;
2153 }
2154
2155 enum
2156 {
2157 REPL_TYPE_STRING,
2158 REPL_TYPE_CHARACTER,
2159 REPL_TYPE_SYMBOLIC_REFERENCE,
2160 REPL_TYPE_NUMERIC_REFERENCE,
2161 REPL_TYPE_CHANGE_CASE
2162 };
2163
2164 typedef enum
2165 {
2166 CHANGE_CASE_NONE = 1 << 0,
2167 CHANGE_CASE_UPPER = 1 << 1,
2168 CHANGE_CASE_LOWER = 1 << 2,
2169 CHANGE_CASE_UPPER_SINGLE = 1 << 3,
2170 CHANGE_CASE_LOWER_SINGLE = 1 << 4,
2171 CHANGE_CASE_SINGLE_MASK = CHANGE_CASE_UPPER_SINGLE | CHANGE_CASE_LOWER_SINGLE,
2172 CHANGE_CASE_LOWER_MASK = CHANGE_CASE_LOWER | CHANGE_CASE_LOWER_SINGLE,
2173 CHANGE_CASE_UPPER_MASK = CHANGE_CASE_UPPER | CHANGE_CASE_UPPER_SINGLE
2174 } ChangeCase;
2175
2176 struct _InterpolationData
2177 {
2178 gchar *text;
2179 gint type;
2180 gint num;
2181 gchar c;
2182 ChangeCase change_case;
2183 };
2184
2185 static void
2186 free_interpolation_data (InterpolationData *data)
2187 {
2188 g_free (data->text);
2189 g_free (data);
2190 }
2191
2192 static const gchar *
2193 expand_escape (const gchar *replacement,
2194 const gchar *p,
2195 InterpolationData *data,
2196 GError **error)
2197 {
2198 const gchar *q, *r;
2199 gint x, d, h, i;
2200 const gchar *error_detail;
2201 gint base = 0;
2202 GError *tmp_error = NULL;
2203
2204 p++;
2205 switch (*p)
2206 {
2207 case 't':
2208 p++;
2209 data->c = '\t';
2210 data->type = REPL_TYPE_CHARACTER;
2211 break;
2212 case 'n':
2213 p++;
2214 data->c = '\n';
2215 data->type = REPL_TYPE_CHARACTER;
2216 break;
2217 case 'v':
2218 p++;
2219 data->c = '\v';
2220 data->type = REPL_TYPE_CHARACTER;
2221 break;
2222 case 'r':
2223 p++;
2224 data->c = '\r';
2225 data->type = REPL_TYPE_CHARACTER;
2226 break;
2227 case 'f':
2228 p++;
2229 data->c = '\f';
2230 data->type = REPL_TYPE_CHARACTER;
2231 break;
2232 case 'a':
2233 p++;
2234 data->c = '\a';
2235 data->type = REPL_TYPE_CHARACTER;
2236 break;
2237 case 'b':
2238 p++;
2239 data->c = '\b';
2240 data->type = REPL_TYPE_CHARACTER;
2241 break;
2242 case '\\':
2243 p++;
2244 data->c = '\\';
2245 data->type = REPL_TYPE_CHARACTER;
2246 break;
2247 case 'x':
2248 p++;
2249 x = 0;
2250 if (*p == '{')
2251 {
2252 p++;
2253 do
2254 {
2255 h = g_ascii_xdigit_value (*p);
2256 if (h < 0)
2257 {
2258 error_detail = _("hexadecimal digit or '}' expected");
2259 goto error;
2260 }
2261 x = x * 16 + h;
2262 p++;
2263 }
2264 while (*p != '}');
2265 p++;
2266 }
2267 else
2268 {
2269 for (i = 0; i < 2; i++)
2270 {
2271 h = g_ascii_xdigit_value (*p);
2272 if (h < 0)
2273 {
2274 error_detail = _("hexadecimal digit expected");
2275 goto error;
2276 }
2277 x = x * 16 + h;
2278 p++;
2279 }
2280 }
2281 data->type = REPL_TYPE_STRING;
2282 data->text = g_new0 (gchar, 8);
2283 g_unichar_to_utf8 (x, data->text);
2284 break;
2285 case 'l':
2286 p++;
2287 data->type = REPL_TYPE_CHANGE_CASE;
2288 data->change_case = CHANGE_CASE_LOWER_SINGLE;
2289 break;
2290 case 'u':
2291 p++;
2292 data->type = REPL_TYPE_CHANGE_CASE;
2293 data->change_case = CHANGE_CASE_UPPER_SINGLE;
2294 break;
2295 case 'L':
2296 p++;
2297 data->type = REPL_TYPE_CHANGE_CASE;
2298 data->change_case = CHANGE_CASE_LOWER;
2299 break;
2300 case 'U':
2301 p++;
2302 data->type = REPL_TYPE_CHANGE_CASE;
2303 data->change_case = CHANGE_CASE_UPPER;
2304 break;
2305 case 'E':
2306 p++;
2307 data->type = REPL_TYPE_CHANGE_CASE;
2308 data->change_case = CHANGE_CASE_NONE;
2309 break;
2310 case 'g':
2311 p++;
2312 if (*p != '<')
2313 {
2314 error_detail = _("missing '<' in symbolic reference");
2315 goto error;
2316 }
2317 q = p + 1;
2318 do
2319 {
2320 p++;
2321 if (!*p)
2322 {
2323 error_detail = _("unfinished symbolic reference");
2324 goto error;
2325 }
2326 }
2327 while (*p != '>');
2328 if (p - q == 0)
2329 {
2330 error_detail = _("zero-length symbolic reference");
2331 goto error;
2332 }
2333 if (g_ascii_isdigit (*q))
2334 {
2335 x = 0;
2336 do
2337 {
2338 h = g_ascii_digit_value (*q);
2339 if (h < 0)
2340 {
2341 error_detail = _("digit expected");
2342 p = q;
2343 goto error;
2344 }
2345 x = x * 10 + h;
2346 q++;
2347 }
2348 while (q != p);
2349 data->num = x;
2350 data->type = REPL_TYPE_NUMERIC_REFERENCE;
2351 }
2352 else
2353 {
2354 r = q;
2355 do
2356 {
2357 if (!g_ascii_isalnum (*r))
2358 {
2359 error_detail = _("illegal symbolic reference");
2360 p = r;
2361 goto error;
2362 }
2363 r++;
2364 }
2365 while (r != p);
2366 data->text = g_strndup (q, p - q);
2367 data->type = REPL_TYPE_SYMBOLIC_REFERENCE;
2368 }
2369 p++;
2370 break;
2371 case '0':
2372 /* if \0 is followed by a number is an octal number representing a
2373 * character, else it is a numeric reference. */
2374 if (g_ascii_digit_value (*g_utf8_next_char (p)) >= 0)
2375 {
2376 base = 8;
2377 p = g_utf8_next_char (p);
2378 }
2379 case '1':
2380 case '2':
2381 case '3':
2382 case '4':
2383 case '5':
2384 case '6':
2385 case '7':
2386 case '8':
2387 case '9':
2388 x = 0;
2389 d = 0;
2390 for (i = 0; i < 3; i++)
2391 {
2392 h = g_ascii_digit_value (*p);
2393 if (h < 0)
2394 break;
2395 if (h > 7)
2396 {
2397 if (base == 8)
2398 break;
2399 else
2400 base = 10;
2401 }
2402 if (i == 2 && base == 10)
2403 break;
2404 x = x * 8 + h;
2405 d = d * 10 + h;
2406 p++;
2407 }
2408 if (base == 8 || i == 3)
2409 {
2410 data->type = REPL_TYPE_STRING;
2411 data->text = g_new0 (gchar, 8);
2412 g_unichar_to_utf8 (x, data->text);
2413 }
2414 else
2415 {
2416 data->type = REPL_TYPE_NUMERIC_REFERENCE;
2417 data->num = d;
2418 }
2419 break;
2420 case 0:
2421 error_detail = _("stray final '\\'");
2422 goto error;
2423 break;
2424 default:
2425 error_detail = _("unknown escape sequence");
2426 goto error;
2427 }
2428
2429 return p;
2430
2431 error:
2432 /* G_GSSIZE_FORMAT doesn't work with gettext, so we use %lu */
2433 tmp_error = g_error_new (G_REGEX_ERROR,
2434 G_REGEX_ERROR_REPLACE,
2435 _("Error while parsing replacement "
2436 "text \"%s\" at char %lu: %s"),
2437 replacement,
2438 (gulong)(p - replacement),
2439 error_detail);
2440 g_propagate_error (error, tmp_error);
2441
2442 return NULL;
2443 }
2444
2445 static GList *
2446 split_replacement (const gchar *replacement,
2447 GError **error)
2448 {
2449 GList *list = NULL;
2450 InterpolationData *data;
2451 const gchar *p, *start;
2452
2453 start = p = replacement;
2454 while (*p)
2455 {
2456 if (*p == '\\')
2457 {
2458 data = g_new0 (InterpolationData, 1);
2459 start = p = expand_escape (replacement, p, data, error);
2460 if (p == NULL)
2461 {
2462 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
2463 free_interpolation_data (data);
2464
2465 return NULL;
2466 }
2467 list = g_list_prepend (list, data);
2468 }
2469 else
2470 {
2471 p++;
2472 if (*p == '\\' || *p == '\0')
2473 {
2474 if (p - start > 0)
2475 {
2476 data = g_new0 (InterpolationData, 1);
2477 data->text = g_strndup (start, p - start);
2478 data->type = REPL_TYPE_STRING;
2479 list = g_list_prepend (list, data);
2480 }
2481 }
2482 }
2483 }
2484
2485 return g_list_reverse (list);
2486 }
2487
2488 /* Change the case of c based on change_case. */
2489 #define CHANGE_CASE(c, change_case) \
2490 (((change_case) & CHANGE_CASE_LOWER_MASK) ? \
2491 g_unichar_tolower (c) : \
2492 g_unichar_toupper (c))
2493
2494 static void
2495 string_append (GString *string,
2496 const gchar *text,
2497 ChangeCase *change_case)
2498 {
2499 gunichar c;
2500
2501 if (text[0] == '\0')
2502 return;
2503
2504 if (*change_case == CHANGE_CASE_NONE)
2505 {
2506 g_string_append (string, text);
2507 }
2508 else if (*change_case & CHANGE_CASE_SINGLE_MASK)
2509 {
2510 c = g_utf8_get_char (text);
2511 g_string_append_unichar (string, CHANGE_CASE (c, *change_case));
2512 g_string_append (string, g_utf8_next_char (text));
2513 *change_case = CHANGE_CASE_NONE;
2514 }
2515 else
2516 {
2517 while (*text != '\0')
2518 {
2519 c = g_utf8_get_char (text);
2520 g_string_append_unichar (string, CHANGE_CASE (c, *change_case));
2521 text = g_utf8_next_char (text);
2522 }
2523 }
2524 }
2525
2526 static gboolean
2527 interpolate_replacement (const GMatchInfo *match_info,
2528 GString *result,
2529 gpointer data)
2530 {
2531 GList *list;
2532 InterpolationData *idata;
2533 gchar *match;
2534 ChangeCase change_case = CHANGE_CASE_NONE;
2535
2536 for (list = data; list; list = list->next)
2537 {
2538 idata = list->data;
2539 switch (idata->type)
2540 {
2541 case REPL_TYPE_STRING:
2542 string_append (result, idata->text, &change_case);
2543 break;
2544 case REPL_TYPE_CHARACTER:
2545 g_string_append_c (result, CHANGE_CASE (idata->c, change_case));
2546 if (change_case & CHANGE_CASE_SINGLE_MASK)
2547 change_case = CHANGE_CASE_NONE;
2548 break;
2549 case REPL_TYPE_NUMERIC_REFERENCE:
2550 match = g_match_info_fetch (match_info, idata->num);
2551 if (match)
2552 {
2553 string_append (result, match, &change_case);
2554 g_free (match);
2555 }
2556 break;
2557 case REPL_TYPE_SYMBOLIC_REFERENCE:
2558 match = g_match_info_fetch_named (match_info, idata->text);
2559 if (match)
2560 {
2561 string_append (result, match, &change_case);
2562 g_free (match);
2563 }
2564 break;
2565 case REPL_TYPE_CHANGE_CASE:
2566 change_case = idata->change_case;
2567 break;
2568 }
2569 }
2570
2571 return FALSE;
2572 }
2573
2574 /* whether actual match_info is needed for replacement, i.e.
2575 * whether there are references
2576 */
2577 static gboolean
2578 interpolation_list_needs_match (GList *list)
2579 {
2580 while (list != NULL)
2581 {
2582 InterpolationData *data = list->data;
2583
2584 if (data->type == REPL_TYPE_SYMBOLIC_REFERENCE ||
2585 data->type == REPL_TYPE_NUMERIC_REFERENCE)
2586 {
2587 return TRUE;
2588 }
2589
2590 list = list->next;
2591 }
2592
2593 return FALSE;
2594 }
2595
2596 /**
2597 * g_regex_replace:
2598 * @regex: a #GRegex structure
2599 * @string: (array length=string_len): the string to perform matches against
2600 * @string_len: the length of @string, or -1 if @string is nul-terminated
2601 * @start_position: starting index of the string to match
2602 * @replacement: text to replace each match with
2603 * @match_options: options for the match
2604 * @error: location to store the error occurring, or %NULL to ignore errors
2605 *
2606 * Replaces all occurrences of the pattern in @regex with the
2607 * replacement text. Backreferences of the form '\number' or
2608 * '\g&lt;number&gt;' in the replacement text are interpolated by the
2609 * number-th captured subexpression of the match, '\g&lt;name&gt;' refers
2610 * to the captured subexpression with the given name. '\0' refers to the
2611 * complete match, but '\0' followed by a number is the octal representation
2612 * of a character. To include a literal '\' in the replacement, write '\\'.
2613 * There are also escapes that changes the case of the following text:
2614 *
2615 * <variablelist>
2616 * <varlistentry><term>\l</term>
2617 * <listitem>
2618 * <para>Convert to lower case the next character</para>
2619 * </listitem>
2620 * </varlistentry>
2621 * <varlistentry><term>\u</term>
2622 * <listitem>
2623 * <para>Convert to upper case the next character</para>
2624 * </listitem>
2625 * </varlistentry>
2626 * <varlistentry><term>\L</term>
2627 * <listitem>
2628 * <para>Convert to lower case till \E</para>
2629 * </listitem>
2630 * </varlistentry>
2631 * <varlistentry><term>\U</term>
2632 * <listitem>
2633 * <para>Convert to upper case till \E</para>
2634 * </listitem>
2635 * </varlistentry>
2636 * <varlistentry><term>\E</term>
2637 * <listitem>
2638 * <para>End case modification</para>
2639 * </listitem>
2640 * </varlistentry>
2641 * </variablelist>
2642 *
2643 * If you do not need to use backreferences use g_regex_replace_literal().
2644 *
2645 * The @replacement string must be UTF-8 encoded even if #G_REGEX_RAW was
2646 * passed to g_regex_new(). If you want to use not UTF-8 encoded stings
2647 * you can use g_regex_replace_literal().
2648 *
2649 * Setting @start_position differs from just passing over a shortened
2650 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern that
2651 * begins with any kind of lookbehind assertion, such as "\b".
2652 *
2653 * Returns: a newly allocated string containing the replacements
2654 *
2655 * Since: 2.14
2656 */
2657 gchar *
2658 g_regex_replace (const GRegex *regex,
2659 const gchar *string,
2660 gssize string_len,
2661 gint start_position,
2662 const gchar *replacement,
2663 GRegexMatchFlags match_options,
2664 GError **error)
2665 {
2666 gchar *result;
2667 GList *list;
2668 GError *tmp_error = NULL;
2669
2670 g_return_val_if_fail (regex != NULL, NULL);
2671 g_return_val_if_fail (string != NULL, NULL);
2672 g_return_val_if_fail (start_position >= 0, NULL);
2673 g_return_val_if_fail (replacement != NULL, NULL);
2674 g_return_val_if_fail (error == NULL || *error == NULL, NULL);
2675 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2676
2677 list = split_replacement (replacement, &tmp_error);
2678 if (tmp_error != NULL)
2679 {
2680 g_propagate_error (error, tmp_error);
2681 return NULL;
2682 }
2683
2684 result = g_regex_replace_eval (regex,
2685 string, string_len, start_position,
2686 match_options,
2687 interpolate_replacement,
2688 (gpointer)list,
2689 &tmp_error);
2690 if (tmp_error != NULL)
2691 g_propagate_error (error, tmp_error);
2692
2693 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
2694
2695 return result;
2696 }
2697
2698 static gboolean
2699 literal_replacement (const GMatchInfo *match_info,
2700 GString *result,
2701 gpointer data)
2702 {
2703 g_string_append (result, data);
2704 return FALSE;
2705 }
2706
2707 /**
2708 * g_regex_replace_literal:
2709 * @regex: a #GRegex structure
2710 * @string: (array length=string_len): the string to perform matches against
2711 * @string_len: the length of @string, or -1 if @string is nul-terminated
2712 * @start_position: starting index of the string to match
2713 * @replacement: text to replace each match with
2714 * @match_options: options for the match
2715 * @error: location to store the error occurring, or %NULL to ignore errors
2716 *
2717 * Replaces all occurrences of the pattern in @regex with the
2718 * replacement text. @replacement is replaced literally, to
2719 * include backreferences use g_regex_replace().
2720 *
2721 * Setting @start_position differs from just passing over a
2722 * shortened string and setting #G_REGEX_MATCH_NOTBOL in the
2723 * case of a pattern that begins with any kind of lookbehind
2724 * assertion, such as "\b".
2725 *
2726 * Returns: a newly allocated string containing the replacements
2727 *
2728 * Since: 2.14
2729 */
2730 gchar *
2731 g_regex_replace_literal (const GRegex *regex,
2732 const gchar *string,
2733 gssize string_len,
2734 gint start_position,
2735 const gchar *replacement,
2736 GRegexMatchFlags match_options,
2737 GError **error)
2738 {
2739 g_return_val_if_fail (replacement != NULL, NULL);
2740 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2741
2742 return g_regex_replace_eval (regex,
2743 string, string_len, start_position,
2744 match_options,
2745 literal_replacement,
2746 (gpointer)replacement,
2747 error);
2748 }
2749
2750 /**
2751 * g_regex_replace_eval:
2752 * @regex: a #GRegex structure from g_regex_new()
2753 * @string: (array length=string_len): string to perform matches against
2754 * @string_len: the length of @string, or -1 if @string is nul-terminated
2755 * @start_position: starting index of the string to match
2756 * @match_options: options for the match
2757 * @eval: a function to call for each match
2758 * @user_data: user data to pass to the function
2759 * @error: location to store the error occurring, or %NULL to ignore errors
2760 *
2761 * Replaces occurrences of the pattern in regex with the output of
2762 * @eval for that occurrence.
2763 *
2764 * Setting @start_position differs from just passing over a shortened
2765 * string and setting #G_REGEX_MATCH_NOTBOL in the case of a pattern
2766 * that begins with any kind of lookbehind assertion, such as "\b".
2767 *
2768 * The following example uses g_regex_replace_eval() to replace multiple
2769 * strings at once:
2770 * |[
2771 * static gboolean
2772 * eval_cb (const GMatchInfo *info,
2773 * GString *res,
2774 * gpointer data)
2775 * {
2776 * gchar *match;
2777 * gchar *r;
2778 *
2779 * match = g_match_info_fetch (info, 0);
2780 * r = g_hash_table_lookup ((GHashTable *)data, match);
2781 * g_string_append (res, r);
2782 * g_free (match);
2783 *
2784 * return FALSE;
2785 * }
2786 *
2787 * /&ast; ... &ast;/
2788 *
2789 * GRegex *reg;
2790 * GHashTable *h;
2791 * gchar *res;
2792 *
2793 * h = g_hash_table_new (g_str_hash, g_str_equal);
2794 *
2795 * g_hash_table_insert (h, "1", "ONE");
2796 * g_hash_table_insert (h, "2", "TWO");
2797 * g_hash_table_insert (h, "3", "THREE");
2798 * g_hash_table_insert (h, "4", "FOUR");
2799 *
2800 * reg = g_regex_new ("1|2|3|4", 0, 0, NULL);
2801 * res = g_regex_replace_eval (reg, text, -1, 0, 0, eval_cb, h, NULL);
2802 * g_hash_table_destroy (h);
2803 *
2804 * /&ast; ... &ast;/
2805 * ]|
2806 *
2807 * Returns: a newly allocated string containing the replacements
2808 *
2809 * Since: 2.14
2810 */
2811 gchar *
2812 g_regex_replace_eval (const GRegex *regex,
2813 const gchar *string,
2814 gssize string_len,
2815 gint start_position,
2816 GRegexMatchFlags match_options,
2817 GRegexEvalCallback eval,
2818 gpointer user_data,
2819 GError **error)
2820 {
2821 GMatchInfo *match_info;
2822 GString *result;
2823 gint str_pos = 0;
2824 gboolean done = FALSE;
2825 GError *tmp_error = NULL;
2826
2827 g_return_val_if_fail (regex != NULL, NULL);
2828 g_return_val_if_fail (string != NULL, NULL);
2829 g_return_val_if_fail (start_position >= 0, NULL);
2830 g_return_val_if_fail (eval != NULL, NULL);
2831 g_return_val_if_fail ((match_options & ~G_REGEX_MATCH_MASK) == 0, NULL);
2832
2833 if (string_len < 0)
2834 string_len = strlen (string);
2835
2836 result = g_string_sized_new (string_len);
2837
2838 /* run down the string making matches. */
2839 g_regex_match_full (regex, string, string_len, start_position,
2840 match_options, &match_info, &tmp_error);
2841 while (!done && g_match_info_matches (match_info))
2842 {
2843 g_string_append_len (result,
2844 string + str_pos,
2845 match_info->offsets[0] - str_pos);
2846 done = (*eval) (match_info, result, user_data);
2847 str_pos = match_info->offsets[1];
2848 g_match_info_next (match_info, &tmp_error);
2849 }
2850 g_match_info_free (match_info);
2851 if (tmp_error != NULL)
2852 {
2853 g_propagate_error (error, tmp_error);
2854 g_string_free (result, TRUE);
2855 return NULL;
2856 }
2857
2858 g_string_append_len (result, string + str_pos, string_len - str_pos);
2859 return g_string_free (result, FALSE);
2860 }
2861
2862 /**
2863 * g_regex_check_replacement:
2864 * @replacement: the replacement string
2865 * @has_references: (out) (allow-none): location to store information about
2866 * references in @replacement or %NULL
2867 * @error: location to store error
2868 *
2869 * Checks whether @replacement is a valid replacement string
2870 * (see g_regex_replace()), i.e. that all escape sequences in
2871 * it are valid.
2872 *
2873 * If @has_references is not %NULL then @replacement is checked
2874 * for pattern references. For instance, replacement text 'foo\n'
2875 * does not contain references and may be evaluated without information
2876 * about actual match, but '\0\1' (whole match followed by first
2877 * subpattern) requires valid #GMatchInfo object.
2878 *
2879 * Returns: whether @replacement is a valid replacement string
2880 *
2881 * Since: 2.14
2882 */
2883 gboolean
2884 g_regex_check_replacement (const gchar *replacement,
2885 gboolean *has_references,
2886 GError **error)
2887 {
2888 GList *list;
2889 GError *tmp = NULL;
2890
2891 list = split_replacement (replacement, &tmp);
2892
2893 if (tmp)
2894 {
2895 g_propagate_error (error, tmp);
2896 return FALSE;
2897 }
2898
2899 if (has_references)
2900 *has_references = interpolation_list_needs_match (list);
2901
2902 g_list_free_full (list, (GDestroyNotify) free_interpolation_data);
2903
2904 return TRUE;
2905 }
2906
2907 /**
2908 * g_regex_escape_nul:
2909 * @string: the string to escape
2910 * @length: the length of @string
2911 *
2912 * Escapes the nul characters in @string to "\x00". It can be used
2913 * to compile a regex with embedded nul characters.
2914 *
2915 * For completeness, @length can be -1 for a nul-terminated string.
2916 * In this case the output string will be of course equal to @string.
2917 *
2918 * Returns: a newly-allocated escaped string
2919 *
2920 * Since: 2.30
2921 */
2922 gchar *
2923 g_regex_escape_nul (const gchar *string,
2924 gint length)
2925 {
2926 GString *escaped;
2927 const gchar *p, *piece_start, *end;
2928 gint backslashes;
2929
2930 g_return_val_if_fail (string != NULL, NULL);
2931
2932 if (length < 0)
2933 return g_strdup (string);
2934
2935 end = string + length;
2936 p = piece_start = string;
2937 escaped = g_string_sized_new (length + 1);
2938
2939 backslashes = 0;
2940 while (p < end)
2941 {
2942 switch (*p)
2943 {
2944 case '\0':
2945 if (p != piece_start)
2946 {
2947 /* copy the previous piece. */
2948 g_string_append_len (escaped, piece_start, p - piece_start);
2949 }
2950 if ((backslashes & 1) == 0)
2951 g_string_append_c (escaped, '\\');
2952 g_string_append_c (escaped, 'x');
2953 g_string_append_c (escaped, '0');
2954 g_string_append_c (escaped, '0');
2955 piece_start = ++p;
2956 backslashes = 0;
2957 break;
2958 case '\\':
2959 backslashes++;
2960 ++p;
2961 break;
2962 default:
2963 backslashes = 0;
2964 p = g_utf8_next_char (p);
2965 break;
2966 }
2967 }
2968
2969 if (piece_start < end)
2970 g_string_append_len (escaped, piece_start, end - piece_start);
2971
2972 return g_string_free (escaped, FALSE);
2973 }
2974
2975 /**
2976 * g_regex_escape_string:
2977 * @string: (array length=length): the string to escape
2978 * @length: the length of @string, or -1 if @string is nul-terminated
2979 *
2980 * Escapes the special characters used for regular expressions
2981 * in @string, for instance "a.b*c" becomes "a\.b\*c". This
2982 * function is useful to dynamically generate regular expressions.
2983 *
2984 * @string can contain nul characters that are replaced with "\0",
2985 * in this case remember to specify the correct length of @string
2986 * in @length.
2987 *
2988 * Returns: a newly-allocated escaped string
2989 *
2990 * Since: 2.14
2991 */
2992 gchar *
2993 g_regex_escape_string (const gchar *string,
2994 gint length)
2995 {
2996 GString *escaped;
2997 const char *p, *piece_start, *end;
2998
2999 g_return_val_if_fail (string != NULL, NULL);
3000
3001 if (length < 0)
3002 length = strlen (string);
3003
3004 end = string + length;
3005 p = piece_start = string;
3006 escaped = g_string_sized_new (length + 1);
3007
3008 while (p < end)
3009 {
3010 switch (*p)
3011 {
3012 case '\0':
3013 case '\\':
3014 case '|':
3015 case '(':
3016 case ')':
3017 case '[':
3018 case ']':
3019 case '{':
3020 case '}':
3021 case '^':
3022 case '$':
3023 case '*':
3024 case '+':
3025 case '?':
3026 case '.':
3027 if (p != piece_start)
3028 /* copy the previous piece. */
3029 g_string_append_len (escaped, piece_start, p - piece_start);
3030 g_string_append_c (escaped, '\\');
3031 if (*p == '\0')
3032 g_string_append_c (escaped, '0');
3033 else
3034 g_string_append_c (escaped, *p);
3035 piece_start = ++p;
3036 break;
3037 default:
3038 p = g_utf8_next_char (p);
3039 break;
3040 }
3041 }
3042
3043 if (piece_start < end)
3044 g_string_append_len (escaped, piece_start, end - piece_start);
3045
3046 return g_string_free (escaped, FALSE);
3047 }
glib mirror