1 | /* Definitions for data structures and routines for the regular |
2 | expression library. |
3 | Copyright (C) 1985, 1989-2024 Free Software Foundation, Inc. |
4 | This file is part of the GNU C Library. |
5 | |
6 | The GNU C Library is free software; you can redistribute it and/or |
7 | modify it under the terms of the GNU Lesser General Public |
8 | License as published by the Free Software Foundation; either |
9 | version 2.1 of the License, or (at your option) any later version. |
10 | |
11 | The GNU C Library is distributed in the hope that it will be useful, |
12 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
14 | Lesser General Public License for more details. |
15 | |
16 | You should have received a copy of the GNU Lesser General Public |
17 | License along with the GNU C Library; if not, see |
18 | <https://www.gnu.org/licenses/>. */ |
19 | |
20 | #ifndef _REGEX_H |
21 | #define _REGEX_H 1 |
22 | |
23 | #include <sys/types.h> |
24 | |
25 | /* Allow the use in C++ code. */ |
26 | #ifdef __cplusplus |
27 | extern "C" { |
28 | #endif |
29 | |
30 | /* Define __USE_GNU to declare GNU extensions that violate the |
31 | POSIX name space rules. */ |
32 | #ifdef _GNU_SOURCE |
33 | # define __USE_GNU 1 |
34 | #endif |
35 | |
36 | #ifdef _REGEX_LARGE_OFFSETS |
37 | |
38 | /* Use types and values that are wide enough to represent signed and |
39 | unsigned byte offsets in memory. This currently works only when |
40 | the regex code is used outside of the GNU C library; it is not yet |
41 | supported within glibc itself, and glibc users should not define |
42 | _REGEX_LARGE_OFFSETS. */ |
43 | |
44 | /* The type of object sizes. */ |
45 | typedef size_t __re_size_t; |
46 | |
47 | /* The type of object sizes, in places where the traditional code |
48 | uses unsigned long int. */ |
49 | typedef size_t __re_long_size_t; |
50 | |
51 | #else |
52 | |
53 | /* The traditional GNU regex implementation mishandles strings longer |
54 | than INT_MAX. */ |
55 | typedef unsigned int __re_size_t; |
56 | typedef unsigned long int __re_long_size_t; |
57 | |
58 | #endif |
59 | |
60 | /* The following two types have to be signed and unsigned integer type |
61 | wide enough to hold a value of a pointer. For most ANSI compilers |
62 | ptrdiff_t and size_t should be likely OK. Still size of these two |
63 | types is 2 for Microsoft C. Ugh... */ |
64 | typedef long int s_reg_t; |
65 | typedef unsigned long int active_reg_t; |
66 | |
67 | /* The following bits are used to determine the regexp syntax we |
68 | recognize. The set/not-set meanings are chosen so that Emacs syntax |
69 | remains the value 0. The bits are given in alphabetical order, and |
70 | the definitions shifted by one from the previous bit; thus, when we |
71 | add or remove a bit, only one other definition need change. */ |
72 | typedef unsigned long int reg_syntax_t; |
73 | |
74 | #ifdef __USE_GNU |
75 | /* If this bit is not set, then \ inside a bracket expression is literal. |
76 | If set, then such a \ quotes the following character. */ |
77 | # define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1) |
78 | |
79 | /* If this bit is not set, then + and ? are operators, and \+ and \? are |
80 | literals. |
81 | If set, then \+ and \? are operators and + and ? are literals. */ |
82 | # define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1) |
83 | |
84 | /* If this bit is set, then character classes are supported. They are: |
85 | [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:], |
86 | [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:]. |
87 | If not set, then character classes are not supported. */ |
88 | # define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1) |
89 | |
90 | /* If this bit is set, then ^ and $ are always anchors (outside bracket |
91 | expressions, of course). |
92 | If this bit is not set, then it depends: |
93 | ^ is an anchor if it is at the beginning of a regular |
94 | expression or after an open-group or an alternation operator; |
95 | $ is an anchor if it is at the end of a regular expression, or |
96 | before a close-group or an alternation operator. |
97 | |
98 | This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because |
99 | POSIX draft 11.2 says that * etc. in leading positions is undefined. |
100 | We already implemented a previous draft which made those constructs |
101 | invalid, though, so we haven't changed the code back. */ |
102 | # define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1) |
103 | |
104 | /* If this bit is set, then special characters are always special |
105 | regardless of where they are in the pattern. |
106 | If this bit is not set, then special characters are special only in |
107 | some contexts; otherwise they are ordinary. Specifically, |
108 | * + ? and intervals are only special when not after the beginning, |
109 | open-group, or alternation operator. */ |
110 | # define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1) |
111 | |
112 | /* If this bit is set, then *, +, ?, and { cannot be first in an re or |
113 | immediately after an alternation or begin-group operator. */ |
114 | # define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1) |
115 | |
116 | /* If this bit is set, then . matches newline. |
117 | If not set, then it doesn't. */ |
118 | # define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1) |
119 | |
120 | /* If this bit is set, then . doesn't match NUL. |
121 | If not set, then it does. */ |
122 | # define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1) |
123 | |
124 | /* If this bit is set, nonmatching lists [^...] do not match newline. |
125 | If not set, they do. */ |
126 | # define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1) |
127 | |
128 | /* If this bit is set, either \{...\} or {...} defines an |
129 | interval, depending on RE_NO_BK_BRACES. |
130 | If not set, \{, \}, {, and } are literals. */ |
131 | # define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1) |
132 | |
133 | /* If this bit is set, +, ? and | aren't recognized as operators. |
134 | If not set, they are. */ |
135 | # define RE_LIMITED_OPS (RE_INTERVALS << 1) |
136 | |
137 | /* If this bit is set, newline is an alternation operator. |
138 | If not set, newline is literal. */ |
139 | # define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1) |
140 | |
141 | /* If this bit is set, then '{...}' defines an interval, and \{ and \} |
142 | are literals. |
143 | If not set, then '\{...\}' defines an interval. */ |
144 | # define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1) |
145 | |
146 | /* If this bit is set, (...) defines a group, and \( and \) are literals. |
147 | If not set, \(...\) defines a group, and ( and ) are literals. */ |
148 | # define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1) |
149 | |
150 | /* If this bit is set, then \<digit> matches <digit>. |
151 | If not set, then \<digit> is a back-reference. */ |
152 | # define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1) |
153 | |
154 | /* If this bit is set, then | is an alternation operator, and \| is literal. |
155 | If not set, then \| is an alternation operator, and | is literal. */ |
156 | # define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1) |
157 | |
158 | /* If this bit is set, then an ending range point collating higher |
159 | than the starting range point, as in [z-a], is invalid. |
160 | If not set, then when ending range point collates higher than the |
161 | starting range point, the range is ignored. */ |
162 | # define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1) |
163 | |
164 | /* If this bit is set, then an unmatched ) is ordinary. |
165 | If not set, then an unmatched ) is invalid. */ |
166 | # define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1) |
167 | |
168 | /* If this bit is set, succeed as soon as we match the whole pattern, |
169 | without further backtracking. */ |
170 | # define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1) |
171 | |
172 | /* If this bit is set, do not process the GNU regex operators. |
173 | If not set, then the GNU regex operators are recognized. */ |
174 | # define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1) |
175 | |
176 | /* If this bit is set, turn on internal regex debugging. |
177 | If not set, and debugging was on, turn it off. |
178 | This only works if regex.c is compiled -DDEBUG. |
179 | We define this bit always, so that all that's needed to turn on |
180 | debugging is to recompile regex.c; the calling code can always have |
181 | this bit set, and it won't affect anything in the normal case. */ |
182 | # define RE_DEBUG (RE_NO_GNU_OPS << 1) |
183 | |
184 | /* If this bit is set, a syntactically invalid interval is treated as |
185 | a string of ordinary characters. For example, the ERE 'a{1' is |
186 | treated as 'a\{1'. */ |
187 | # define RE_INVALID_INTERVAL_ORD (RE_DEBUG << 1) |
188 | |
189 | /* If this bit is set, then ignore case when matching. |
190 | If not set, then case is significant. */ |
191 | # define RE_ICASE (RE_INVALID_INTERVAL_ORD << 1) |
192 | |
193 | /* This bit is used internally like RE_CONTEXT_INDEP_ANCHORS but only |
194 | for ^, because it is difficult to scan the regex backwards to find |
195 | whether ^ should be special. */ |
196 | # define RE_CARET_ANCHORS_HERE (RE_ICASE << 1) |
197 | |
198 | /* If this bit is set, then \{ cannot be first in a regex or |
199 | immediately after an alternation, open-group or \} operator. */ |
200 | # define RE_CONTEXT_INVALID_DUP (RE_CARET_ANCHORS_HERE << 1) |
201 | |
202 | /* If this bit is set, then no_sub will be set to 1 during |
203 | re_compile_pattern. */ |
204 | # define RE_NO_SUB (RE_CONTEXT_INVALID_DUP << 1) |
205 | #endif |
206 | |
207 | /* This global variable defines the particular regexp syntax to use (for |
208 | some interfaces). When a regexp is compiled, the syntax used is |
209 | stored in the pattern buffer, so changing this does not affect |
210 | already-compiled regexps. */ |
211 | extern reg_syntax_t re_syntax_options; |
212 | |
213 | #ifdef __USE_GNU |
214 | /* Define combinations of the above bits for the standard possibilities. |
215 | (The [[[ comments delimit what gets put into the Texinfo file, so |
216 | don't delete them!) */ |
217 | /* [[[begin syntaxes]]] */ |
218 | # define RE_SYNTAX_EMACS 0 |
219 | |
220 | # define RE_SYNTAX_AWK \ |
221 | (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \ |
222 | | RE_NO_BK_PARENS | RE_NO_BK_REFS \ |
223 | | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \ |
224 | | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \ |
225 | | RE_CHAR_CLASSES \ |
226 | | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS) |
227 | |
228 | # define RE_SYNTAX_GNU_AWK \ |
229 | ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \ |
230 | | RE_INVALID_INTERVAL_ORD) \ |
231 | & ~(RE_DOT_NOT_NULL | RE_CONTEXT_INDEP_OPS \ |
232 | | RE_CONTEXT_INVALID_OPS )) |
233 | |
234 | # define RE_SYNTAX_POSIX_AWK \ |
235 | (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \ |
236 | | RE_INTERVALS | RE_NO_GNU_OPS \ |
237 | | RE_INVALID_INTERVAL_ORD) |
238 | |
239 | # define RE_SYNTAX_GREP \ |
240 | ((RE_SYNTAX_POSIX_BASIC | RE_NEWLINE_ALT) \ |
241 | & ~(RE_CONTEXT_INVALID_DUP | RE_DOT_NOT_NULL)) |
242 | |
243 | # define RE_SYNTAX_EGREP \ |
244 | ((RE_SYNTAX_POSIX_EXTENDED | RE_INVALID_INTERVAL_ORD | RE_NEWLINE_ALT) \ |
245 | & ~(RE_CONTEXT_INVALID_OPS | RE_DOT_NOT_NULL)) |
246 | |
247 | /* POSIX grep -E behavior is no longer incompatible with GNU. */ |
248 | # define RE_SYNTAX_POSIX_EGREP \ |
249 | RE_SYNTAX_EGREP |
250 | |
251 | /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */ |
252 | # define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC |
253 | |
254 | # define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC |
255 | |
256 | /* Syntax bits common to both basic and extended POSIX regex syntax. */ |
257 | # define _RE_SYNTAX_POSIX_COMMON \ |
258 | (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \ |
259 | | RE_INTERVALS | RE_NO_EMPTY_RANGES) |
260 | |
261 | # define RE_SYNTAX_POSIX_BASIC \ |
262 | (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM | RE_CONTEXT_INVALID_DUP) |
263 | |
264 | /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes |
265 | RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this |
266 | isn't minimal, since other operators, such as \`, aren't disabled. */ |
267 | # define RE_SYNTAX_POSIX_MINIMAL_BASIC \ |
268 | (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS) |
269 | |
270 | # define RE_SYNTAX_POSIX_EXTENDED \ |
271 | (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ |
272 | | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \ |
273 | | RE_NO_BK_PARENS | RE_NO_BK_VBAR \ |
274 | | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD) |
275 | |
276 | /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is |
277 | removed and RE_NO_BK_REFS is added. */ |
278 | # define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \ |
279 | (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ |
280 | | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \ |
281 | | RE_NO_BK_PARENS | RE_NO_BK_REFS \ |
282 | | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD) |
283 | /* [[[end syntaxes]]] */ |
284 | |
285 | /* Maximum number of duplicates an interval can allow. POSIX-conforming |
286 | systems might define this in <limits.h>, but we want our |
287 | value, so remove any previous define. */ |
288 | # ifdef _REGEX_INCLUDE_LIMITS_H |
289 | # include <limits.h> |
290 | # endif |
291 | # ifdef RE_DUP_MAX |
292 | # undef RE_DUP_MAX |
293 | # endif |
294 | |
295 | /* RE_DUP_MAX is 2**15 - 1 because an earlier implementation stored |
296 | the counter as a 2-byte signed integer. This is no longer true, so |
297 | RE_DUP_MAX could be increased to (INT_MAX / 10 - 1), or to |
298 | ((SIZE_MAX - 9) / 10) if _REGEX_LARGE_OFFSETS is defined. |
299 | However, there would be a huge performance problem if someone |
300 | actually used a pattern like a\{214748363\}, so RE_DUP_MAX retains |
301 | its historical value. */ |
302 | # define RE_DUP_MAX (0x7fff) |
303 | #endif |
304 | |
305 | |
306 | /* POSIX 'cflags' bits (i.e., information for 'regcomp'). */ |
307 | |
308 | /* If this bit is set, then use extended regular expression syntax. |
309 | If not set, then use basic regular expression syntax. */ |
310 | #define REG_EXTENDED 1 |
311 | |
312 | /* If this bit is set, then ignore case when matching. |
313 | If not set, then case is significant. */ |
314 | #define REG_ICASE (1 << 1) |
315 | |
316 | /* If this bit is set, then anchors do not match at newline |
317 | characters in the string. |
318 | If not set, then anchors do match at newlines. */ |
319 | #define REG_NEWLINE (1 << 2) |
320 | |
321 | /* If this bit is set, then report only success or fail in regexec. |
322 | If not set, then returns differ between not matching and errors. */ |
323 | #define REG_NOSUB (1 << 3) |
324 | |
325 | |
326 | /* POSIX 'eflags' bits (i.e., information for regexec). */ |
327 | |
328 | /* If this bit is set, then the beginning-of-line operator doesn't match |
329 | the beginning of the string (presumably because it's not the |
330 | beginning of a line). |
331 | If not set, then the beginning-of-line operator does match the |
332 | beginning of the string. */ |
333 | #define REG_NOTBOL 1 |
334 | |
335 | /* Like REG_NOTBOL, except for the end-of-line. */ |
336 | #define REG_NOTEOL (1 << 1) |
337 | |
338 | /* Use PMATCH[0] to delimit the start and end of the search in the |
339 | buffer. */ |
340 | #define REG_STARTEND (1 << 2) |
341 | |
342 | |
343 | /* If any error codes are removed, changed, or added, update the |
344 | '__re_error_msgid' table in regcomp.c. */ |
345 | |
346 | typedef enum |
347 | { |
348 | _REG_ENOSYS = -1, /* This will never happen for this implementation. */ |
349 | _REG_NOERROR = 0, /* Success. */ |
350 | _REG_NOMATCH, /* Didn't find a match (for regexec). */ |
351 | |
352 | /* POSIX regcomp return error codes. (In the order listed in the |
353 | standard.) */ |
354 | _REG_BADPAT, /* Invalid pattern. */ |
355 | _REG_ECOLLATE, /* Invalid collating element. */ |
356 | _REG_ECTYPE, /* Invalid character class name. */ |
357 | _REG_EESCAPE, /* Trailing backslash. */ |
358 | _REG_ESUBREG, /* Invalid back reference. */ |
359 | _REG_EBRACK, /* Unmatched left bracket. */ |
360 | _REG_EPAREN, /* Parenthesis imbalance. */ |
361 | _REG_EBRACE, /* Unmatched \{. */ |
362 | _REG_BADBR, /* Invalid contents of \{\}. */ |
363 | _REG_ERANGE, /* Invalid range end. */ |
364 | _REG_ESPACE, /* Ran out of memory. */ |
365 | _REG_BADRPT, /* No preceding re for repetition op. */ |
366 | |
367 | /* Error codes we've added. */ |
368 | _REG_EEND, /* Premature end. */ |
369 | _REG_ESIZE, /* Too large (e.g., repeat count too large). */ |
370 | _REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */ |
371 | } reg_errcode_t; |
372 | |
373 | #if defined _XOPEN_SOURCE || defined __USE_XOPEN2K |
374 | # define REG_ENOSYS _REG_ENOSYS |
375 | #endif |
376 | #define REG_NOERROR _REG_NOERROR |
377 | #define REG_NOMATCH _REG_NOMATCH |
378 | #define REG_BADPAT _REG_BADPAT |
379 | #define REG_ECOLLATE _REG_ECOLLATE |
380 | #define REG_ECTYPE _REG_ECTYPE |
381 | #define REG_EESCAPE _REG_EESCAPE |
382 | #define REG_ESUBREG _REG_ESUBREG |
383 | #define REG_EBRACK _REG_EBRACK |
384 | #define REG_EPAREN _REG_EPAREN |
385 | #define REG_EBRACE _REG_EBRACE |
386 | #define REG_BADBR _REG_BADBR |
387 | #define REG_ERANGE _REG_ERANGE |
388 | #define REG_ESPACE _REG_ESPACE |
389 | #define REG_BADRPT _REG_BADRPT |
390 | #define REG_EEND _REG_EEND |
391 | #define REG_ESIZE _REG_ESIZE |
392 | #define REG_ERPAREN _REG_ERPAREN |
393 | |
394 | /* This data structure represents a compiled pattern. Before calling |
395 | the pattern compiler, the fields 'buffer', 'allocated', 'fastmap', |
396 | and 'translate' can be set. After the pattern has been compiled, |
397 | the fields 're_nsub', 'not_bol' and 'not_eol' are available. All |
398 | other fields are private to the regex routines. */ |
399 | |
400 | #ifndef RE_TRANSLATE_TYPE |
401 | # define __RE_TRANSLATE_TYPE unsigned char * |
402 | # ifdef __USE_GNU |
403 | # define RE_TRANSLATE_TYPE __RE_TRANSLATE_TYPE |
404 | # endif |
405 | #endif |
406 | |
407 | #ifdef __USE_GNU |
408 | # define __REPB_PREFIX(name) name |
409 | #else |
410 | # define __REPB_PREFIX(name) __##name |
411 | #endif |
412 | |
413 | struct re_pattern_buffer |
414 | { |
415 | /* Space that holds the compiled pattern. The type |
416 | 'struct re_dfa_t' is private and is not declared here. */ |
417 | struct re_dfa_t *__REPB_PREFIX(buffer); |
418 | |
419 | /* Number of bytes to which 'buffer' points. */ |
420 | __re_long_size_t __REPB_PREFIX(allocated); |
421 | |
422 | /* Number of bytes actually used in 'buffer'. */ |
423 | __re_long_size_t __REPB_PREFIX(used); |
424 | |
425 | /* Syntax setting with which the pattern was compiled. */ |
426 | reg_syntax_t __REPB_PREFIX(syntax); |
427 | |
428 | /* Pointer to a fastmap, if any, otherwise zero. re_search uses the |
429 | fastmap, if there is one, to skip over impossible starting points |
430 | for matches. */ |
431 | char *__REPB_PREFIX(fastmap); |
432 | |
433 | /* Either a translate table to apply to all characters before |
434 | comparing them, or zero for no translation. The translation is |
435 | applied to a pattern when it is compiled and to a string when it |
436 | is matched. */ |
437 | __RE_TRANSLATE_TYPE __REPB_PREFIX(translate); |
438 | |
439 | /* Number of subexpressions found by the compiler. */ |
440 | size_t re_nsub; |
441 | |
442 | /* Zero if this pattern cannot match the empty string, one else. |
443 | Well, in truth it's used only in 're_search_2', to see whether or |
444 | not we should use the fastmap, so we don't set this absolutely |
445 | perfectly; see 're_compile_fastmap' (the "duplicate" case). */ |
446 | unsigned __REPB_PREFIX(can_be_null) : 1; |
447 | |
448 | /* If REGS_UNALLOCATED, allocate space in the 'regs' structure |
449 | for 'max (RE_NREGS, re_nsub + 1)' groups. |
450 | If REGS_REALLOCATE, reallocate space if necessary. |
451 | If REGS_FIXED, use what's there. */ |
452 | #ifdef __USE_GNU |
453 | # define REGS_UNALLOCATED 0 |
454 | # define REGS_REALLOCATE 1 |
455 | # define REGS_FIXED 2 |
456 | #endif |
457 | unsigned __REPB_PREFIX(regs_allocated) : 2; |
458 | |
459 | /* Set to zero when 're_compile_pattern' compiles a pattern; set to |
460 | one by 're_compile_fastmap' if it updates the fastmap. */ |
461 | unsigned __REPB_PREFIX(fastmap_accurate) : 1; |
462 | |
463 | /* If set, 're_match_2' does not return information about |
464 | subexpressions. */ |
465 | unsigned __REPB_PREFIX(no_sub) : 1; |
466 | |
467 | /* If set, a beginning-of-line anchor doesn't match at the beginning |
468 | of the string. */ |
469 | unsigned __REPB_PREFIX(not_bol) : 1; |
470 | |
471 | /* Similarly for an end-of-line anchor. */ |
472 | unsigned __REPB_PREFIX(not_eol) : 1; |
473 | |
474 | /* If true, an anchor at a newline matches. */ |
475 | unsigned __REPB_PREFIX(newline_anchor) : 1; |
476 | }; |
477 | |
478 | typedef struct re_pattern_buffer regex_t; |
479 | |
480 | /* Type for byte offsets within the string. POSIX mandates this. */ |
481 | #ifdef _REGEX_LARGE_OFFSETS |
482 | /* POSIX 1003.1-2008 requires that regoff_t be at least as wide as |
483 | ptrdiff_t and ssize_t. We don't know of any hosts where ptrdiff_t |
484 | is wider than ssize_t, so ssize_t is safe. ptrdiff_t is not |
485 | visible here, so use ssize_t. */ |
486 | typedef ssize_t regoff_t; |
487 | #else |
488 | /* The traditional GNU regex implementation mishandles strings longer |
489 | than INT_MAX. */ |
490 | typedef int regoff_t; |
491 | #endif |
492 | |
493 | |
494 | #ifdef __USE_GNU |
495 | /* This is the structure we store register match data in. See |
496 | regex.texinfo for a full description of what registers match. */ |
497 | struct re_registers |
498 | { |
499 | __re_size_t num_regs; |
500 | regoff_t *start; |
501 | regoff_t *end; |
502 | }; |
503 | |
504 | |
505 | /* If 'regs_allocated' is REGS_UNALLOCATED in the pattern buffer, |
506 | 're_match_2' returns information about at least this many registers |
507 | the first time a 'regs' structure is passed. */ |
508 | # ifndef RE_NREGS |
509 | # define RE_NREGS 30 |
510 | # endif |
511 | #endif |
512 | |
513 | |
514 | /* POSIX specification for registers. Aside from the different names than |
515 | 're_registers', POSIX uses an array of structures, instead of a |
516 | structure of arrays. */ |
517 | typedef struct |
518 | { |
519 | regoff_t rm_so; /* Byte offset from string's start to substring's start. */ |
520 | regoff_t rm_eo; /* Byte offset from string's start to substring's end. */ |
521 | } regmatch_t; |
522 | |
523 | /* Declarations for routines. */ |
524 | |
525 | #ifndef _REGEX_NELTS |
526 | # if (defined __STDC_VERSION__ && 199901L <= __STDC_VERSION__ \ |
527 | && !defined __STDC_NO_VLA__) |
528 | # define _REGEX_NELTS(n) n |
529 | # else |
530 | # define _REGEX_NELTS(n) |
531 | # endif |
532 | #endif |
533 | |
534 | #if defined __GNUC__ && 4 < __GNUC__ + (6 <= __GNUC_MINOR__) |
535 | # pragma GCC diagnostic push |
536 | # pragma GCC diagnostic ignored "-Wvla" |
537 | #endif |
538 | |
539 | #ifndef _Attr_access_ |
540 | # ifdef __attr_access |
541 | # define _Attr_access_(arg) __attr_access (arg) |
542 | # elif defined __GNUC__ && 10 <= __GNUC__ |
543 | # define _Attr_access_(x) __attribute__ ((__access__ x)) |
544 | # else |
545 | # define _Attr_access_(x) |
546 | # endif |
547 | #endif |
548 | |
549 | #ifdef __USE_GNU |
550 | /* Sets the current default syntax to SYNTAX, and return the old syntax. |
551 | You can also simply assign to the 're_syntax_options' variable. */ |
552 | extern reg_syntax_t re_set_syntax (reg_syntax_t __syntax); |
553 | |
554 | /* Compile the regular expression PATTERN, with length LENGTH |
555 | and syntax given by the global 're_syntax_options', into the buffer |
556 | BUFFER. Return NULL if successful, and an error string if not. |
557 | |
558 | To free the allocated storage, you must call 'regfree' on BUFFER. |
559 | Note that the translate table must either have been initialized by |
560 | 'regcomp', with a malloc'ed value, or set to NULL before calling |
561 | 'regfree'. */ |
562 | extern const char *re_compile_pattern (const char *__pattern, size_t __length, |
563 | struct re_pattern_buffer *__buffer) |
564 | _Attr_access_ ((__read_only__, 1, 2)); |
565 | |
566 | |
567 | /* Compile a fastmap for the compiled pattern in BUFFER; used to |
568 | accelerate searches. Return 0 if successful and -2 if was an |
569 | internal error. */ |
570 | extern int re_compile_fastmap (struct re_pattern_buffer *__buffer); |
571 | |
572 | |
573 | /* Search in the string STRING (with length LENGTH) for the pattern |
574 | compiled into BUFFER. Start searching at position START, for RANGE |
575 | characters. Return the starting position of the match, -1 for no |
576 | match, or -2 for an internal error. Also return register |
577 | information in REGS (if REGS and BUFFER->no_sub are nonzero). */ |
578 | extern regoff_t re_search (struct re_pattern_buffer *__buffer, |
579 | const char *__String, regoff_t __length, |
580 | regoff_t __start, regoff_t __range, |
581 | struct re_registers *__regs) |
582 | _Attr_access_ ((__read_only__, 2, 3)); |
583 | |
584 | |
585 | /* Like 're_search', but search in the concatenation of STRING1 and |
586 | STRING2. Also, stop searching at index START + STOP. */ |
587 | extern regoff_t re_search_2 (struct re_pattern_buffer *__buffer, |
588 | const char *__string1, regoff_t __length1, |
589 | const char *__string2, regoff_t __length2, |
590 | regoff_t __start, regoff_t __range, |
591 | struct re_registers *__regs, |
592 | regoff_t __stop) |
593 | _Attr_access_ ((__read_only__, 2, 3)) |
594 | _Attr_access_ ((__read_only__, 4, 5)); |
595 | |
596 | |
597 | /* Like 're_search', but return how many characters in STRING the regexp |
598 | in BUFFER matched, starting at position START. */ |
599 | extern regoff_t re_match (struct re_pattern_buffer *__buffer, |
600 | const char *__String, regoff_t __length, |
601 | regoff_t __start, struct re_registers *__regs) |
602 | _Attr_access_ ((__read_only__, 2, 3)); |
603 | |
604 | |
605 | /* Relates to 're_match' as 're_search_2' relates to 're_search'. */ |
606 | extern regoff_t re_match_2 (struct re_pattern_buffer *__buffer, |
607 | const char *__string1, regoff_t __length1, |
608 | const char *__string2, regoff_t __length2, |
609 | regoff_t __start, struct re_registers *__regs, |
610 | regoff_t __stop) |
611 | _Attr_access_ ((__read_only__, 2, 3)) |
612 | _Attr_access_ ((__read_only__, 4, 5)); |
613 | |
614 | |
615 | /* Set REGS to hold NUM_REGS registers, storing them in STARTS and |
616 | ENDS. Subsequent matches using BUFFER and REGS will use this memory |
617 | for recording register information. STARTS and ENDS must be |
618 | allocated with malloc, and must each be at least 'NUM_REGS * sizeof |
619 | (regoff_t)' bytes long. |
620 | |
621 | If NUM_REGS == 0, then subsequent matches should allocate their own |
622 | register data. |
623 | |
624 | Unless this function is called, the first search or match using |
625 | BUFFER will allocate its own register data, without |
626 | freeing the old data. */ |
627 | extern void re_set_registers (struct re_pattern_buffer *__buffer, |
628 | struct re_registers *__regs, |
629 | __re_size_t __num_regs, |
630 | regoff_t *__starts, regoff_t *__ends); |
631 | #endif /* Use GNU */ |
632 | |
633 | #if defined _REGEX_RE_COMP || (defined _LIBC && defined __USE_MISC) |
634 | /* 4.2 bsd compatibility. */ |
635 | extern char *re_comp (const char *); |
636 | extern int re_exec (const char *); |
637 | #endif |
638 | |
639 | /* For plain 'restrict', use glibc's __restrict if defined. |
640 | Otherwise, GCC 2.95 and later have "__restrict"; C99 compilers have |
641 | "restrict", and "configure" may have defined "restrict". |
642 | Other compilers use __restrict, __restrict__, and _Restrict, and |
643 | 'configure' might #define 'restrict' to those words, so pick a |
644 | different name. */ |
645 | #ifndef _Restrict_ |
646 | # if defined __restrict \ |
647 | || 2 < __GNUC__ + (95 <= __GNUC_MINOR__) \ |
648 | || __clang_major__ >= 3 |
649 | # define _Restrict_ __restrict |
650 | # elif 199901L <= __STDC_VERSION__ || defined restrict |
651 | # define _Restrict_ restrict |
652 | # else |
653 | # define _Restrict_ |
654 | # endif |
655 | #endif |
656 | /* For the ISO C99 syntax |
657 | array_name[restrict] |
658 | use glibc's __restrict_arr if available. |
659 | Otherwise, GCC 3.1 and clang support this syntax (but not in C++ mode). |
660 | Other ISO C99 compilers support it as well. */ |
661 | #ifndef _Restrict_arr_ |
662 | # ifdef __restrict_arr |
663 | # define _Restrict_arr_ __restrict_arr |
664 | # elif ((199901L <= __STDC_VERSION__ \ |
665 | || 3 < __GNUC__ + (1 <= __GNUC_MINOR__) \ |
666 | || __clang_major__ >= 3) \ |
667 | && !defined __cplusplus) |
668 | # define _Restrict_arr_ _Restrict_ |
669 | # else |
670 | # define _Restrict_arr_ |
671 | # endif |
672 | #endif |
673 | |
674 | /* POSIX compatibility. */ |
675 | extern int regcomp (regex_t *_Restrict_ __preg, |
676 | const char *_Restrict_ __pattern, |
677 | int __cflags); |
678 | |
679 | extern int regexec (const regex_t *_Restrict_ __preg, |
680 | const char *_Restrict_ __String, size_t __nmatch, |
681 | regmatch_t __pmatch[_Restrict_arr_ |
682 | _REGEX_NELTS (__nmatch)], |
683 | int __eflags); |
684 | |
685 | extern size_t regerror (int __errcode, const regex_t *_Restrict_ __preg, |
686 | char *_Restrict_ __errbuf, size_t __errbuf_size) |
687 | _Attr_access_ ((__write_only__, 3, 4)); |
688 | |
689 | extern void regfree (regex_t *__preg); |
690 | |
691 | #if defined __GNUC__ && 4 < __GNUC__ + (6 <= __GNUC_MINOR__) |
692 | # pragma GCC diagnostic pop |
693 | #endif |
694 | |
695 | #ifdef __cplusplus |
696 | } |
697 | #endif /* C++ */ |
698 | |
699 | #endif /* regex.h */ |
700 | |