Commit | Line | Data |
---|---|---|
eb4a14ed LC |
1 | /* Definitions for data structures and routines for the regular |
2 | expression library. | |
3 | Copyright (C) 1985, 1989-1993, 1995-1998, 2000-2003, 2005-2006, 2009-2012 | |
4 | Free Software Foundation, Inc. | |
5 | This file is part of the GNU C Library. | |
6 | ||
7 | This program is free software; you can redistribute it and/or modify | |
8 | it under the terms of the GNU Lesser General Public License as published by | |
9 | the Free Software Foundation; either version 2, or (at your option) | |
10 | any later version. | |
11 | ||
12 | This program 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 | |
15 | GNU Lesser General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU Lesser General Public License along | |
18 | with this program; if not, write to the Free Software Foundation, | |
19 | Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ | |
20 | ||
21 | #ifndef _REGEX_H | |
22 | #define _REGEX_H 1 | |
23 | ||
24 | #include <sys/types.h> | |
25 | ||
26 | /* Allow the use in C++ code. */ | |
27 | #ifdef __cplusplus | |
28 | extern "C" { | |
29 | #endif | |
30 | ||
31 | /* Define __USE_GNU_REGEX to declare GNU extensions that violate the | |
32 | POSIX name space rules. */ | |
33 | #undef __USE_GNU_REGEX | |
34 | #if (defined _GNU_SOURCE \ | |
35 | || (!defined _POSIX_C_SOURCE && !defined _POSIX_SOURCE \ | |
36 | && !defined _XOPEN_SOURCE)) | |
37 | # define __USE_GNU_REGEX 1 | |
38 | #endif | |
39 | ||
40 | #ifdef _REGEX_LARGE_OFFSETS | |
41 | ||
42 | /* Use types and values that are wide enough to represent signed and | |
43 | unsigned byte offsets in memory. This currently works only when | |
44 | the regex code is used outside of the GNU C library; it is not yet | |
45 | supported within glibc itself, and glibc users should not define | |
46 | _REGEX_LARGE_OFFSETS. */ | |
47 | ||
48 | /* The type of the offset of a byte within a string. | |
49 | For historical reasons POSIX 1003.1-2004 requires that regoff_t be | |
50 | at least as wide as off_t. However, many common POSIX platforms set | |
51 | regoff_t to the more-sensible ssize_t and the Open Group has | |
52 | signalled its intention to change the requirement to be that | |
53 | regoff_t be at least as wide as ptrdiff_t and ssize_t; see XBD ERN | |
54 | 60 (2005-08-25). We don't know of any hosts where ssize_t or | |
55 | ptrdiff_t is wider than ssize_t, so ssize_t is safe. */ | |
56 | typedef ssize_t regoff_t; | |
57 | ||
58 | /* The type of nonnegative object indexes. Traditionally, GNU regex | |
59 | uses 'int' for these. Code that uses __re_idx_t should work | |
60 | regardless of whether the type is signed. */ | |
61 | typedef size_t __re_idx_t; | |
62 | ||
63 | /* The type of object sizes. */ | |
64 | typedef size_t __re_size_t; | |
65 | ||
66 | /* The type of object sizes, in places where the traditional code | |
67 | uses unsigned long int. */ | |
68 | typedef size_t __re_long_size_t; | |
69 | ||
70 | #else | |
71 | ||
72 | /* Use types that are binary-compatible with the traditional GNU regex | |
73 | implementation, which mishandles strings longer than INT_MAX. */ | |
74 | ||
75 | typedef int regoff_t; | |
76 | typedef int __re_idx_t; | |
77 | typedef unsigned int __re_size_t; | |
78 | typedef unsigned long int __re_long_size_t; | |
79 | ||
80 | #endif | |
81 | ||
82 | /* The following two types have to be signed and unsigned integer type | |
83 | wide enough to hold a value of a pointer. For most ANSI compilers | |
84 | ptrdiff_t and size_t should be likely OK. Still size of these two | |
85 | types is 2 for Microsoft C. Ugh... */ | |
86 | typedef long int s_reg_t; | |
87 | typedef unsigned long int active_reg_t; | |
88 | ||
89 | /* The following bits are used to determine the regexp syntax we | |
90 | recognize. The set/not-set meanings are chosen so that Emacs syntax | |
91 | remains the value 0. The bits are given in alphabetical order, and | |
92 | the definitions shifted by one from the previous bit; thus, when we | |
93 | add or remove a bit, only one other definition need change. */ | |
94 | typedef unsigned long int reg_syntax_t; | |
95 | ||
96 | #ifdef __USE_GNU_REGEX | |
97 | ||
98 | /* If this bit is not set, then \ inside a bracket expression is literal. | |
99 | If set, then such a \ quotes the following character. */ | |
100 | # define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1) | |
101 | ||
102 | /* If this bit is not set, then + and ? are operators, and \+ and \? are | |
103 | literals. | |
104 | If set, then \+ and \? are operators and + and ? are literals. */ | |
105 | # define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1) | |
106 | ||
107 | /* If this bit is set, then character classes are supported. They are: | |
108 | [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:], | |
109 | [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:]. | |
110 | If not set, then character classes are not supported. */ | |
111 | # define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1) | |
112 | ||
113 | /* If this bit is set, then ^ and $ are always anchors (outside bracket | |
114 | expressions, of course). | |
115 | If this bit is not set, then it depends: | |
116 | ^ is an anchor if it is at the beginning of a regular | |
117 | expression or after an open-group or an alternation operator; | |
118 | $ is an anchor if it is at the end of a regular expression, or | |
119 | before a close-group or an alternation operator. | |
120 | ||
121 | This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because | |
122 | POSIX draft 11.2 says that * etc. in leading positions is undefined. | |
123 | We already implemented a previous draft which made those constructs | |
124 | invalid, though, so we haven't changed the code back. */ | |
125 | # define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1) | |
126 | ||
127 | /* If this bit is set, then special characters are always special | |
128 | regardless of where they are in the pattern. | |
129 | If this bit is not set, then special characters are special only in | |
130 | some contexts; otherwise they are ordinary. Specifically, | |
131 | * + ? and intervals are only special when not after the beginning, | |
132 | open-group, or alternation operator. */ | |
133 | # define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1) | |
134 | ||
135 | /* If this bit is set, then *, +, ?, and { cannot be first in an re or | |
136 | immediately after an alternation or begin-group operator. */ | |
137 | # define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1) | |
138 | ||
139 | /* If this bit is set, then . matches newline. | |
140 | If not set, then it doesn't. */ | |
141 | # define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1) | |
142 | ||
143 | /* If this bit is set, then . doesn't match NUL. | |
144 | If not set, then it does. */ | |
145 | # define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1) | |
146 | ||
147 | /* If this bit is set, nonmatching lists [^...] do not match newline. | |
148 | If not set, they do. */ | |
149 | # define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1) | |
150 | ||
151 | /* If this bit is set, either \{...\} or {...} defines an | |
152 | interval, depending on RE_NO_BK_BRACES. | |
153 | If not set, \{, \}, {, and } are literals. */ | |
154 | # define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1) | |
155 | ||
156 | /* If this bit is set, +, ? and | aren't recognized as operators. | |
157 | If not set, they are. */ | |
158 | # define RE_LIMITED_OPS (RE_INTERVALS << 1) | |
159 | ||
160 | /* If this bit is set, newline is an alternation operator. | |
161 | If not set, newline is literal. */ | |
162 | # define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1) | |
163 | ||
164 | /* If this bit is set, then '{...}' defines an interval, and \{ and \} | |
165 | are literals. | |
166 | If not set, then '\{...\}' defines an interval. */ | |
167 | # define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1) | |
168 | ||
169 | /* If this bit is set, (...) defines a group, and \( and \) are literals. | |
170 | If not set, \(...\) defines a group, and ( and ) are literals. */ | |
171 | # define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1) | |
172 | ||
173 | /* If this bit is set, then \<digit> matches <digit>. | |
174 | If not set, then \<digit> is a back-reference. */ | |
175 | # define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1) | |
176 | ||
177 | /* If this bit is set, then | is an alternation operator, and \| is literal. | |
178 | If not set, then \| is an alternation operator, and | is literal. */ | |
179 | # define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1) | |
180 | ||
181 | /* If this bit is set, then an ending range point collating higher | |
182 | than the starting range point, as in [z-a], is invalid. | |
183 | If not set, then when ending range point collates higher than the | |
184 | starting range point, the range is ignored. */ | |
185 | # define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1) | |
186 | ||
187 | /* If this bit is set, then an unmatched ) is ordinary. | |
188 | If not set, then an unmatched ) is invalid. */ | |
189 | # define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1) | |
190 | ||
191 | /* If this bit is set, succeed as soon as we match the whole pattern, | |
192 | without further backtracking. */ | |
193 | # define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1) | |
194 | ||
195 | /* If this bit is set, do not process the GNU regex operators. | |
196 | If not set, then the GNU regex operators are recognized. */ | |
197 | # define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1) | |
198 | ||
199 | /* If this bit is set, turn on internal regex debugging. | |
200 | If not set, and debugging was on, turn it off. | |
201 | This only works if regex.c is compiled -DDEBUG. | |
202 | We define this bit always, so that all that's needed to turn on | |
203 | debugging is to recompile regex.c; the calling code can always have | |
204 | this bit set, and it won't affect anything in the normal case. */ | |
205 | # define RE_DEBUG (RE_NO_GNU_OPS << 1) | |
206 | ||
207 | /* If this bit is set, a syntactically invalid interval is treated as | |
208 | a string of ordinary characters. For example, the ERE 'a{1' is | |
209 | treated as 'a\{1'. */ | |
210 | # define RE_INVALID_INTERVAL_ORD (RE_DEBUG << 1) | |
211 | ||
212 | /* If this bit is set, then ignore case when matching. | |
213 | If not set, then case is significant. */ | |
214 | # define RE_ICASE (RE_INVALID_INTERVAL_ORD << 1) | |
215 | ||
216 | /* This bit is used internally like RE_CONTEXT_INDEP_ANCHORS but only | |
217 | for ^, because it is difficult to scan the regex backwards to find | |
218 | whether ^ should be special. */ | |
219 | # define RE_CARET_ANCHORS_HERE (RE_ICASE << 1) | |
220 | ||
221 | /* If this bit is set, then \{ cannot be first in a regex or | |
222 | immediately after an alternation, open-group or \} operator. */ | |
223 | # define RE_CONTEXT_INVALID_DUP (RE_CARET_ANCHORS_HERE << 1) | |
224 | ||
225 | /* If this bit is set, then no_sub will be set to 1 during | |
226 | re_compile_pattern. */ | |
227 | # define RE_NO_SUB (RE_CONTEXT_INVALID_DUP << 1) | |
228 | ||
229 | #endif /* defined __USE_GNU_REGEX */ | |
230 | ||
231 | /* This global variable defines the particular regexp syntax to use (for | |
232 | some interfaces). When a regexp is compiled, the syntax used is | |
233 | stored in the pattern buffer, so changing this does not affect | |
234 | already-compiled regexps. */ | |
235 | extern reg_syntax_t re_syntax_options; | |
236 | \f | |
237 | #ifdef __USE_GNU_REGEX | |
238 | /* Define combinations of the above bits for the standard possibilities. | |
239 | (The [[[ comments delimit what gets put into the Texinfo file, so | |
240 | don't delete them!) */ | |
241 | /* [[[begin syntaxes]]] */ | |
242 | # define RE_SYNTAX_EMACS 0 | |
243 | ||
244 | # define RE_SYNTAX_AWK \ | |
245 | (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \ | |
246 | | RE_NO_BK_PARENS | RE_NO_BK_REFS \ | |
247 | | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \ | |
248 | | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \ | |
249 | | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS) | |
250 | ||
251 | # define RE_SYNTAX_GNU_AWK \ | |
252 | ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DEBUG) \ | |
253 | & ~(RE_DOT_NOT_NULL | RE_INTERVALS | RE_CONTEXT_INDEP_OPS \ | |
254 | | RE_CONTEXT_INVALID_OPS )) | |
255 | ||
256 | # define RE_SYNTAX_POSIX_AWK \ | |
257 | (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \ | |
258 | | RE_INTERVALS | RE_NO_GNU_OPS) | |
259 | ||
260 | # define RE_SYNTAX_GREP \ | |
261 | (RE_BK_PLUS_QM | RE_CHAR_CLASSES \ | |
262 | | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \ | |
263 | | RE_NEWLINE_ALT) | |
264 | ||
265 | # define RE_SYNTAX_EGREP \ | |
266 | (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \ | |
267 | | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \ | |
268 | | RE_NEWLINE_ALT | RE_NO_BK_PARENS \ | |
269 | | RE_NO_BK_VBAR) | |
270 | ||
271 | # define RE_SYNTAX_POSIX_EGREP \ | |
272 | (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES \ | |
273 | | RE_INVALID_INTERVAL_ORD) | |
274 | ||
275 | /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */ | |
276 | # define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC | |
277 | ||
278 | # define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC | |
279 | ||
280 | /* Syntax bits common to both basic and extended POSIX regex syntax. */ | |
281 | # define _RE_SYNTAX_POSIX_COMMON \ | |
282 | (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \ | |
283 | | RE_INTERVALS | RE_NO_EMPTY_RANGES) | |
284 | ||
285 | # define RE_SYNTAX_POSIX_BASIC \ | |
286 | (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM | RE_CONTEXT_INVALID_DUP) | |
287 | ||
288 | /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes | |
289 | RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this | |
290 | isn't minimal, since other operators, such as \`, aren't disabled. */ | |
291 | # define RE_SYNTAX_POSIX_MINIMAL_BASIC \ | |
292 | (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS) | |
293 | ||
294 | # define RE_SYNTAX_POSIX_EXTENDED \ | |
295 | (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ | |
296 | | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \ | |
297 | | RE_NO_BK_PARENS | RE_NO_BK_VBAR \ | |
298 | | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD) | |
299 | ||
300 | /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is | |
301 | removed and RE_NO_BK_REFS is added. */ | |
302 | # define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \ | |
303 | (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ | |
304 | | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \ | |
305 | | RE_NO_BK_PARENS | RE_NO_BK_REFS \ | |
306 | | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD) | |
307 | /* [[[end syntaxes]]] */ | |
308 | ||
309 | #endif /* defined __USE_GNU_REGEX */ | |
310 | \f | |
311 | #ifdef __USE_GNU_REGEX | |
312 | ||
313 | /* Maximum number of duplicates an interval can allow. POSIX-conforming | |
314 | systems might define this in <limits.h>, but we want our | |
315 | value, so remove any previous define. */ | |
316 | # ifdef RE_DUP_MAX | |
317 | # undef RE_DUP_MAX | |
318 | # endif | |
319 | ||
320 | /* RE_DUP_MAX is 2**15 - 1 because an earlier implementation stored | |
321 | the counter as a 2-byte signed integer. This is no longer true, so | |
322 | RE_DUP_MAX could be increased to (INT_MAX / 10 - 1), or to | |
323 | ((SIZE_MAX - 2) / 10 - 1) if _REGEX_LARGE_OFFSETS is defined. | |
324 | However, there would be a huge performance problem if someone | |
325 | actually used a pattern like a\{214748363\}, so RE_DUP_MAX retains | |
326 | its historical value. */ | |
327 | # define RE_DUP_MAX (0x7fff) | |
328 | ||
329 | #endif /* defined __USE_GNU_REGEX */ | |
330 | ||
331 | ||
332 | /* POSIX 'cflags' bits (i.e., information for 'regcomp'). */ | |
333 | ||
334 | /* If this bit is set, then use extended regular expression syntax. | |
335 | If not set, then use basic regular expression syntax. */ | |
336 | #define REG_EXTENDED 1 | |
337 | ||
338 | /* If this bit is set, then ignore case when matching. | |
339 | If not set, then case is significant. */ | |
340 | #define REG_ICASE (1 << 1) | |
341 | ||
342 | /* If this bit is set, then anchors do not match at newline | |
343 | characters in the string. | |
344 | If not set, then anchors do match at newlines. */ | |
345 | #define REG_NEWLINE (1 << 2) | |
346 | ||
347 | /* If this bit is set, then report only success or fail in regexec. | |
348 | If not set, then returns differ between not matching and errors. */ | |
349 | #define REG_NOSUB (1 << 3) | |
350 | ||
351 | ||
352 | /* POSIX 'eflags' bits (i.e., information for regexec). */ | |
353 | ||
354 | /* If this bit is set, then the beginning-of-line operator doesn't match | |
355 | the beginning of the string (presumably because it's not the | |
356 | beginning of a line). | |
357 | If not set, then the beginning-of-line operator does match the | |
358 | beginning of the string. */ | |
359 | #define REG_NOTBOL 1 | |
360 | ||
361 | /* Like REG_NOTBOL, except for the end-of-line. */ | |
362 | #define REG_NOTEOL (1 << 1) | |
363 | ||
364 | /* Use PMATCH[0] to delimit the start and end of the search in the | |
365 | buffer. */ | |
366 | #define REG_STARTEND (1 << 2) | |
367 | ||
368 | ||
369 | /* If any error codes are removed, changed, or added, update the | |
370 | '__re_error_msgid' table in regcomp.c. */ | |
371 | ||
372 | typedef enum | |
373 | { | |
374 | _REG_ENOSYS = -1, /* This will never happen for this implementation. */ | |
375 | _REG_NOERROR = 0, /* Success. */ | |
376 | _REG_NOMATCH, /* Didn't find a match (for regexec). */ | |
377 | ||
378 | /* POSIX regcomp return error codes. (In the order listed in the | |
379 | standard.) */ | |
380 | _REG_BADPAT, /* Invalid pattern. */ | |
381 | _REG_ECOLLATE, /* Invalid collating element. */ | |
382 | _REG_ECTYPE, /* Invalid character class name. */ | |
383 | _REG_EESCAPE, /* Trailing backslash. */ | |
384 | _REG_ESUBREG, /* Invalid back reference. */ | |
385 | _REG_EBRACK, /* Unmatched left bracket. */ | |
386 | _REG_EPAREN, /* Parenthesis imbalance. */ | |
387 | _REG_EBRACE, /* Unmatched \{. */ | |
388 | _REG_BADBR, /* Invalid contents of \{\}. */ | |
389 | _REG_ERANGE, /* Invalid range end. */ | |
390 | _REG_ESPACE, /* Ran out of memory. */ | |
391 | _REG_BADRPT, /* No preceding re for repetition op. */ | |
392 | ||
393 | /* Error codes we've added. */ | |
394 | _REG_EEND, /* Premature end. */ | |
395 | _REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */ | |
396 | _REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */ | |
397 | } reg_errcode_t; | |
398 | ||
399 | #ifdef _XOPEN_SOURCE | |
400 | # define REG_ENOSYS _REG_ENOSYS | |
401 | #endif | |
402 | #define REG_NOERROR _REG_NOERROR | |
403 | #define REG_NOMATCH _REG_NOMATCH | |
404 | #define REG_BADPAT _REG_BADPAT | |
405 | #define REG_ECOLLATE _REG_ECOLLATE | |
406 | #define REG_ECTYPE _REG_ECTYPE | |
407 | #define REG_EESCAPE _REG_EESCAPE | |
408 | #define REG_ESUBREG _REG_ESUBREG | |
409 | #define REG_EBRACK _REG_EBRACK | |
410 | #define REG_EPAREN _REG_EPAREN | |
411 | #define REG_EBRACE _REG_EBRACE | |
412 | #define REG_BADBR _REG_BADBR | |
413 | #define REG_ERANGE _REG_ERANGE | |
414 | #define REG_ESPACE _REG_ESPACE | |
415 | #define REG_BADRPT _REG_BADRPT | |
416 | #define REG_EEND _REG_EEND | |
417 | #define REG_ESIZE _REG_ESIZE | |
418 | #define REG_ERPAREN _REG_ERPAREN | |
419 | \f | |
420 | /* struct re_pattern_buffer normally uses member names like 'buffer' | |
421 | that POSIX does not allow. In POSIX mode these members have names | |
422 | with leading 're_' (e.g., 're_buffer'). */ | |
423 | #ifdef __USE_GNU_REGEX | |
424 | # define _REG_RE_NAME(id) id | |
425 | # define _REG_RM_NAME(id) id | |
426 | #else | |
427 | # define _REG_RE_NAME(id) re_##id | |
428 | # define _REG_RM_NAME(id) rm_##id | |
429 | #endif | |
430 | ||
431 | /* The user can specify the type of the re_translate member by | |
432 | defining the macro RE_TRANSLATE_TYPE, which defaults to unsigned | |
433 | char *. This pollutes the POSIX name space, so in POSIX mode just | |
434 | use unsigned char *. */ | |
435 | #ifdef __USE_GNU_REGEX | |
436 | # ifndef RE_TRANSLATE_TYPE | |
437 | # define RE_TRANSLATE_TYPE unsigned char * | |
438 | # endif | |
439 | # define REG_TRANSLATE_TYPE RE_TRANSLATE_TYPE | |
440 | #else | |
441 | # define REG_TRANSLATE_TYPE unsigned char * | |
442 | #endif | |
443 | ||
444 | /* This data structure represents a compiled pattern. Before calling | |
445 | the pattern compiler, the fields 'buffer', 'allocated', 'fastmap', | |
446 | 'translate', and 'no_sub' can be set. After the pattern has been | |
447 | compiled, the 're_nsub' field is available. All other fields are | |
448 | private to the regex routines. */ | |
449 | ||
450 | struct re_pattern_buffer | |
451 | { | |
452 | /* Space that holds the compiled pattern. It is declared as | |
453 | 'unsigned char *' because its elements are sometimes used as | |
454 | array indexes. */ | |
455 | unsigned char *_REG_RE_NAME (buffer); | |
456 | ||
457 | /* Number of bytes to which 'buffer' points. */ | |
458 | __re_long_size_t _REG_RE_NAME (allocated); | |
459 | ||
460 | /* Number of bytes actually used in 'buffer'. */ | |
461 | __re_long_size_t _REG_RE_NAME (used); | |
462 | ||
463 | /* Syntax setting with which the pattern was compiled. */ | |
464 | reg_syntax_t _REG_RE_NAME (syntax); | |
465 | ||
466 | /* Pointer to a fastmap, if any, otherwise zero. re_search uses the | |
467 | fastmap, if there is one, to skip over impossible starting points | |
468 | for matches. */ | |
469 | char *_REG_RE_NAME (fastmap); | |
470 | ||
471 | /* Either a translate table to apply to all characters before | |
472 | comparing them, or zero for no translation. The translation is | |
473 | applied to a pattern when it is compiled and to a string when it | |
474 | is matched. */ | |
475 | REG_TRANSLATE_TYPE _REG_RE_NAME (translate); | |
476 | ||
477 | /* Number of subexpressions found by the compiler. */ | |
478 | size_t re_nsub; | |
479 | ||
480 | /* Zero if this pattern cannot match the empty string, one else. | |
481 | Well, in truth it's used only in 're_search_2', to see whether or | |
482 | not we should use the fastmap, so we don't set this absolutely | |
483 | perfectly; see 're_compile_fastmap' (the "duplicate" case). */ | |
484 | unsigned int _REG_RE_NAME (can_be_null) : 1; | |
485 | ||
486 | /* If REGS_UNALLOCATED, allocate space in the 'regs' structure | |
487 | for 'max (RE_NREGS, re_nsub + 1)' groups. | |
488 | If REGS_REALLOCATE, reallocate space if necessary. | |
489 | If REGS_FIXED, use what's there. */ | |
490 | #ifdef __USE_GNU_REGEX | |
491 | # define REGS_UNALLOCATED 0 | |
492 | # define REGS_REALLOCATE 1 | |
493 | # define REGS_FIXED 2 | |
494 | #endif | |
495 | unsigned int _REG_RE_NAME (regs_allocated) : 2; | |
496 | ||
497 | /* Set to zero when 're_compile_pattern' compiles a pattern; set to | |
498 | one by 're_compile_fastmap' if it updates the fastmap. */ | |
499 | unsigned int _REG_RE_NAME (fastmap_accurate) : 1; | |
500 | ||
501 | /* If set, 're_match_2' does not return information about | |
502 | subexpressions. */ | |
503 | unsigned int _REG_RE_NAME (no_sub) : 1; | |
504 | ||
505 | /* If set, a beginning-of-line anchor doesn't match at the beginning | |
506 | of the string. */ | |
507 | unsigned int _REG_RE_NAME (not_bol) : 1; | |
508 | ||
509 | /* Similarly for an end-of-line anchor. */ | |
510 | unsigned int _REG_RE_NAME (not_eol) : 1; | |
511 | ||
512 | /* If true, an anchor at a newline matches. */ | |
513 | unsigned int _REG_RE_NAME (newline_anchor) : 1; | |
514 | ||
515 | /* [[[end pattern_buffer]]] */ | |
516 | }; | |
517 | ||
518 | typedef struct re_pattern_buffer regex_t; | |
519 | \f | |
520 | /* This is the structure we store register match data in. See | |
521 | regex.texinfo for a full description of what registers match. */ | |
522 | struct re_registers | |
523 | { | |
524 | __re_size_t _REG_RM_NAME (num_regs); | |
525 | regoff_t *_REG_RM_NAME (start); | |
526 | regoff_t *_REG_RM_NAME (end); | |
527 | }; | |
528 | ||
529 | ||
530 | /* If 'regs_allocated' is REGS_UNALLOCATED in the pattern buffer, | |
531 | 're_match_2' returns information about at least this many registers | |
532 | the first time a 'regs' structure is passed. */ | |
533 | #if !defined RE_NREGS && defined __USE_GNU_REGEX | |
534 | # define RE_NREGS 30 | |
535 | #endif | |
536 | ||
537 | ||
538 | /* POSIX specification for registers. Aside from the different names than | |
539 | 're_registers', POSIX uses an array of structures, instead of a | |
540 | structure of arrays. */ | |
541 | typedef struct | |
542 | { | |
543 | regoff_t rm_so; /* Byte offset from string's start to substring's start. */ | |
544 | regoff_t rm_eo; /* Byte offset from string's start to substring's end. */ | |
545 | } regmatch_t; | |
546 | \f | |
547 | /* Declarations for routines. */ | |
548 | ||
549 | /* Sets the current default syntax to SYNTAX, and return the old syntax. | |
550 | You can also simply assign to the 're_syntax_options' variable. */ | |
551 | extern reg_syntax_t re_set_syntax (reg_syntax_t __syntax); | |
552 | ||
553 | /* Compile the regular expression PATTERN, with length LENGTH | |
554 | and syntax given by the global 're_syntax_options', into the buffer | |
555 | BUFFER. Return NULL if successful, and an error string if not. */ | |
556 | extern const char *re_compile_pattern (const char *__pattern, size_t __length, | |
557 | struct re_pattern_buffer *__buffer); | |
558 | ||
559 | ||
560 | /* Compile a fastmap for the compiled pattern in BUFFER; used to | |
561 | accelerate searches. Return 0 if successful and -2 if was an | |
562 | internal error. */ | |
563 | extern int re_compile_fastmap (struct re_pattern_buffer *__buffer); | |
564 | ||
565 | ||
566 | /* Search in the string STRING (with length LENGTH) for the pattern | |
567 | compiled into BUFFER. Start searching at position START, for RANGE | |
568 | characters. Return the starting position of the match, -1 for no | |
569 | match, or -2 for an internal error. Also return register | |
570 | information in REGS (if REGS and BUFFER->no_sub are nonzero). */ | |
571 | extern regoff_t re_search (struct re_pattern_buffer *__buffer, | |
572 | const char *__string, __re_idx_t __length, | |
573 | __re_idx_t __start, regoff_t __range, | |
574 | struct re_registers *__regs); | |
575 | ||
576 | ||
577 | /* Like 're_search', but search in the concatenation of STRING1 and | |
578 | STRING2. Also, stop searching at index START + STOP. */ | |
579 | extern regoff_t re_search_2 (struct re_pattern_buffer *__buffer, | |
580 | const char *__string1, __re_idx_t __length1, | |
581 | const char *__string2, __re_idx_t __length2, | |
582 | __re_idx_t __start, regoff_t __range, | |
583 | struct re_registers *__regs, | |
584 | __re_idx_t __stop); | |
585 | ||
586 | ||
587 | /* Like 're_search', but return how many characters in STRING the regexp | |
588 | in BUFFER matched, starting at position START. */ | |
589 | extern regoff_t re_match (struct re_pattern_buffer *__buffer, | |
590 | const char *__string, __re_idx_t __length, | |
591 | __re_idx_t __start, struct re_registers *__regs); | |
592 | ||
593 | ||
594 | /* Relates to 're_match' as 're_search_2' relates to 're_search'. */ | |
595 | extern regoff_t re_match_2 (struct re_pattern_buffer *__buffer, | |
596 | const char *__string1, __re_idx_t __length1, | |
597 | const char *__string2, __re_idx_t __length2, | |
598 | __re_idx_t __start, struct re_registers *__regs, | |
599 | __re_idx_t __stop); | |
600 | ||
601 | ||
602 | /* Set REGS to hold NUM_REGS registers, storing them in STARTS and | |
603 | ENDS. Subsequent matches using BUFFER and REGS will use this memory | |
604 | for recording register information. STARTS and ENDS must be | |
605 | allocated with malloc, and must each be at least 'NUM_REGS * sizeof | |
606 | (regoff_t)' bytes long. | |
607 | ||
608 | If NUM_REGS == 0, then subsequent matches should allocate their own | |
609 | register data. | |
610 | ||
611 | Unless this function is called, the first search or match using | |
612 | BUFFER will allocate its own register data, without freeing the old | |
613 | data. */ | |
614 | extern void re_set_registers (struct re_pattern_buffer *__buffer, | |
615 | struct re_registers *__regs, | |
616 | __re_size_t __num_regs, | |
617 | regoff_t *__starts, regoff_t *__ends); | |
618 | ||
619 | #if defined _REGEX_RE_COMP || defined _LIBC | |
620 | # ifndef _CRAY | |
621 | /* 4.2 bsd compatibility. */ | |
622 | extern char *re_comp (const char *); | |
623 | extern int re_exec (const char *); | |
624 | # endif | |
625 | #endif | |
626 | ||
627 | /* GCC 2.95 and later have "__restrict"; C99 compilers have | |
628 | "restrict", and "configure" may have defined "restrict". | |
629 | Other compilers use __restrict, __restrict__, and _Restrict, and | |
630 | 'configure' might #define 'restrict' to those words, so pick a | |
631 | different name. */ | |
632 | #ifndef _Restrict_ | |
633 | # if 199901L <= __STDC_VERSION__ | |
634 | # define _Restrict_ restrict | |
635 | # elif 2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__) | |
636 | # define _Restrict_ __restrict | |
637 | # else | |
638 | # define _Restrict_ | |
639 | # endif | |
640 | #endif | |
641 | /* gcc 3.1 and up support the [restrict] syntax. Don't trust | |
642 | sys/cdefs.h's definition of __restrict_arr, though, as it | |
643 | mishandles gcc -ansi -pedantic. */ | |
644 | #ifndef _Restrict_arr_ | |
645 | # if ((199901L <= __STDC_VERSION__ \ | |
646 | || ((3 < __GNUC__ || (3 == __GNUC__ && 1 <= __GNUC_MINOR__)) \ | |
647 | && !defined __STRICT_ANSI__)) \ | |
648 | && !defined __GNUG__) | |
649 | # define _Restrict_arr_ _Restrict_ | |
650 | # else | |
651 | # define _Restrict_arr_ | |
652 | # endif | |
653 | #endif | |
654 | ||
655 | /* POSIX compatibility. */ | |
656 | extern int regcomp (regex_t *_Restrict_ __preg, | |
657 | const char *_Restrict_ __pattern, | |
658 | int __cflags); | |
659 | ||
660 | extern int regexec (const regex_t *_Restrict_ __preg, | |
661 | const char *_Restrict_ __string, size_t __nmatch, | |
662 | regmatch_t __pmatch[_Restrict_arr_], | |
663 | int __eflags); | |
664 | ||
665 | extern size_t regerror (int __errcode, const regex_t *_Restrict_ __preg, | |
666 | char *_Restrict_ __errbuf, size_t __errbuf_size); | |
667 | ||
668 | extern void regfree (regex_t *__preg); | |
669 | ||
670 | ||
671 | #ifdef __cplusplus | |
672 | } | |
673 | #endif /* C++ */ | |
674 | ||
675 | #endif /* regex.h */ |