Merge remote-tracking branch 'origin/stable-2.0'
[bpt/guile.git] / lib / regex.h
1 /* Definitions for data structures and routines for the regular
2 expression library.
3 Copyright (C) 1985, 1989-1993, 1995-1998, 2000-2003, 2005-2012
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, see <http://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 nonnegative object indexes. Traditionally, GNU regex
45 uses 'int' for these. Code that uses __re_idx_t should work
46 regardless of whether the type is signed. */
47 typedef size_t __re_idx_t;
48
49 /* The type of object sizes. */
50 typedef size_t __re_size_t;
51
52 /* The type of object sizes, in places where the traditional code
53 uses unsigned long int. */
54 typedef size_t __re_long_size_t;
55
56 #else
57
58 /* The traditional GNU regex implementation mishandles strings longer
59 than INT_MAX. */
60 typedef int __re_idx_t;
61 typedef unsigned int __re_size_t;
62 typedef unsigned long int __re_long_size_t;
63
64 #endif
65
66 /* The following two types have to be signed and unsigned integer type
67 wide enough to hold a value of a pointer. For most ANSI compilers
68 ptrdiff_t and size_t should be likely OK. Still size of these two
69 types is 2 for Microsoft C. Ugh... */
70 typedef long int s_reg_t;
71 typedef unsigned long int active_reg_t;
72
73 /* The following bits are used to determine the regexp syntax we
74 recognize. The set/not-set meanings are chosen so that Emacs syntax
75 remains the value 0. The bits are given in alphabetical order, and
76 the definitions shifted by one from the previous bit; thus, when we
77 add or remove a bit, only one other definition need change. */
78 typedef unsigned long int reg_syntax_t;
79
80 #ifdef __USE_GNU
81 /* If this bit is not set, then \ inside a bracket expression is literal.
82 If set, then such a \ quotes the following character. */
83 # define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1)
84
85 /* If this bit is not set, then + and ? are operators, and \+ and \? are
86 literals.
87 If set, then \+ and \? are operators and + and ? are literals. */
88 # define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1)
89
90 /* If this bit is set, then character classes are supported. They are:
91 [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:],
92 [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:].
93 If not set, then character classes are not supported. */
94 # define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1)
95
96 /* If this bit is set, then ^ and $ are always anchors (outside bracket
97 expressions, of course).
98 If this bit is not set, then it depends:
99 ^ is an anchor if it is at the beginning of a regular
100 expression or after an open-group or an alternation operator;
101 $ is an anchor if it is at the end of a regular expression, or
102 before a close-group or an alternation operator.
103
104 This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because
105 POSIX draft 11.2 says that * etc. in leading positions is undefined.
106 We already implemented a previous draft which made those constructs
107 invalid, though, so we haven't changed the code back. */
108 # define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1)
109
110 /* If this bit is set, then special characters are always special
111 regardless of where they are in the pattern.
112 If this bit is not set, then special characters are special only in
113 some contexts; otherwise they are ordinary. Specifically,
114 * + ? and intervals are only special when not after the beginning,
115 open-group, or alternation operator. */
116 # define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1)
117
118 /* If this bit is set, then *, +, ?, and { cannot be first in an re or
119 immediately after an alternation or begin-group operator. */
120 # define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1)
121
122 /* If this bit is set, then . matches newline.
123 If not set, then it doesn't. */
124 # define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1)
125
126 /* If this bit is set, then . doesn't match NUL.
127 If not set, then it does. */
128 # define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1)
129
130 /* If this bit is set, nonmatching lists [^...] do not match newline.
131 If not set, they do. */
132 # define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1)
133
134 /* If this bit is set, either \{...\} or {...} defines an
135 interval, depending on RE_NO_BK_BRACES.
136 If not set, \{, \}, {, and } are literals. */
137 # define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1)
138
139 /* If this bit is set, +, ? and | aren't recognized as operators.
140 If not set, they are. */
141 # define RE_LIMITED_OPS (RE_INTERVALS << 1)
142
143 /* If this bit is set, newline is an alternation operator.
144 If not set, newline is literal. */
145 # define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1)
146
147 /* If this bit is set, then '{...}' defines an interval, and \{ and \}
148 are literals.
149 If not set, then '\{...\}' defines an interval. */
150 # define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1)
151
152 /* If this bit is set, (...) defines a group, and \( and \) are literals.
153 If not set, \(...\) defines a group, and ( and ) are literals. */
154 # define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1)
155
156 /* If this bit is set, then \<digit> matches <digit>.
157 If not set, then \<digit> is a back-reference. */
158 # define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1)
159
160 /* If this bit is set, then | is an alternation operator, and \| is literal.
161 If not set, then \| is an alternation operator, and | is literal. */
162 # define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1)
163
164 /* If this bit is set, then an ending range point collating higher
165 than the starting range point, as in [z-a], is invalid.
166 If not set, then when ending range point collates higher than the
167 starting range point, the range is ignored. */
168 # define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1)
169
170 /* If this bit is set, then an unmatched ) is ordinary.
171 If not set, then an unmatched ) is invalid. */
172 # define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1)
173
174 /* If this bit is set, succeed as soon as we match the whole pattern,
175 without further backtracking. */
176 # define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1)
177
178 /* If this bit is set, do not process the GNU regex operators.
179 If not set, then the GNU regex operators are recognized. */
180 # define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1)
181
182 /* If this bit is set, turn on internal regex debugging.
183 If not set, and debugging was on, turn it off.
184 This only works if regex.c is compiled -DDEBUG.
185 We define this bit always, so that all that's needed to turn on
186 debugging is to recompile regex.c; the calling code can always have
187 this bit set, and it won't affect anything in the normal case. */
188 # define RE_DEBUG (RE_NO_GNU_OPS << 1)
189
190 /* If this bit is set, a syntactically invalid interval is treated as
191 a string of ordinary characters. For example, the ERE 'a{1' is
192 treated as 'a\{1'. */
193 # define RE_INVALID_INTERVAL_ORD (RE_DEBUG << 1)
194
195 /* If this bit is set, then ignore case when matching.
196 If not set, then case is significant. */
197 # define RE_ICASE (RE_INVALID_INTERVAL_ORD << 1)
198
199 /* This bit is used internally like RE_CONTEXT_INDEP_ANCHORS but only
200 for ^, because it is difficult to scan the regex backwards to find
201 whether ^ should be special. */
202 # define RE_CARET_ANCHORS_HERE (RE_ICASE << 1)
203
204 /* If this bit is set, then \{ cannot be first in a regex or
205 immediately after an alternation, open-group or \} operator. */
206 # define RE_CONTEXT_INVALID_DUP (RE_CARET_ANCHORS_HERE << 1)
207
208 /* If this bit is set, then no_sub will be set to 1 during
209 re_compile_pattern. */
210 # define RE_NO_SUB (RE_CONTEXT_INVALID_DUP << 1)
211 #endif
212
213 /* This global variable defines the particular regexp syntax to use (for
214 some interfaces). When a regexp is compiled, the syntax used is
215 stored in the pattern buffer, so changing this does not affect
216 already-compiled regexps. */
217 extern reg_syntax_t re_syntax_options;
218 \f
219 #ifdef __USE_GNU
220 /* Define combinations of the above bits for the standard possibilities.
221 (The [[[ comments delimit what gets put into the Texinfo file, so
222 don't delete them!) */
223 /* [[[begin syntaxes]]] */
224 # define RE_SYNTAX_EMACS 0
225
226 # define RE_SYNTAX_AWK \
227 (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \
228 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
229 | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \
230 | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \
231 | RE_CHAR_CLASSES \
232 | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS)
233
234 # define RE_SYNTAX_GNU_AWK \
235 ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
236 | RE_INVALID_INTERVAL_ORD) \
237 & ~(RE_DOT_NOT_NULL | RE_CONTEXT_INDEP_OPS \
238 | RE_CONTEXT_INVALID_OPS ))
239
240 # define RE_SYNTAX_POSIX_AWK \
241 (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
242 | RE_INTERVALS | RE_NO_GNU_OPS \
243 | RE_INVALID_INTERVAL_ORD)
244
245 # define RE_SYNTAX_GREP \
246 (RE_BK_PLUS_QM | RE_CHAR_CLASSES \
247 | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \
248 | RE_NEWLINE_ALT)
249
250 # define RE_SYNTAX_EGREP \
251 (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \
252 | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \
253 | RE_NEWLINE_ALT | RE_NO_BK_PARENS \
254 | RE_NO_BK_VBAR)
255
256 # define RE_SYNTAX_POSIX_EGREP \
257 (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES \
258 | RE_INVALID_INTERVAL_ORD)
259
260 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */
261 # define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
262
263 # define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
264
265 /* Syntax bits common to both basic and extended POSIX regex syntax. */
266 # define _RE_SYNTAX_POSIX_COMMON \
267 (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \
268 | RE_INTERVALS | RE_NO_EMPTY_RANGES)
269
270 # define RE_SYNTAX_POSIX_BASIC \
271 (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM | RE_CONTEXT_INVALID_DUP)
272
273 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
274 RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this
275 isn't minimal, since other operators, such as \`, aren't disabled. */
276 # define RE_SYNTAX_POSIX_MINIMAL_BASIC \
277 (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
278
279 # define RE_SYNTAX_POSIX_EXTENDED \
280 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
281 | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \
282 | RE_NO_BK_PARENS | RE_NO_BK_VBAR \
283 | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD)
284
285 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is
286 removed and RE_NO_BK_REFS is added. */
287 # define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \
288 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
289 | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \
290 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
291 | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD)
292 /* [[[end syntaxes]]] */
293
294 /* Maximum number of duplicates an interval can allow. POSIX-conforming
295 systems might define this in <limits.h>, but we want our
296 value, so remove any previous define. */
297 # ifdef _REGEX_INCLUDE_LIMITS_H
298 # include <limits.h>
299 # endif
300 # ifdef RE_DUP_MAX
301 # undef RE_DUP_MAX
302 # endif
303
304 /* RE_DUP_MAX is 2**15 - 1 because an earlier implementation stored
305 the counter as a 2-byte signed integer. This is no longer true, so
306 RE_DUP_MAX could be increased to (INT_MAX / 10 - 1), or to
307 ((SIZE_MAX - 9) / 10) if _REGEX_LARGE_OFFSETS is defined.
308 However, there would be a huge performance problem if someone
309 actually used a pattern like a\{214748363\}, so RE_DUP_MAX retains
310 its historical value. */
311 # define RE_DUP_MAX (0x7fff)
312 #endif
313
314
315 /* POSIX 'cflags' bits (i.e., information for 'regcomp'). */
316
317 /* If this bit is set, then use extended regular expression syntax.
318 If not set, then use basic regular expression syntax. */
319 #define REG_EXTENDED 1
320
321 /* If this bit is set, then ignore case when matching.
322 If not set, then case is significant. */
323 #define REG_ICASE (1 << 1)
324
325 /* If this bit is set, then anchors do not match at newline
326 characters in the string.
327 If not set, then anchors do match at newlines. */
328 #define REG_NEWLINE (1 << 2)
329
330 /* If this bit is set, then report only success or fail in regexec.
331 If not set, then returns differ between not matching and errors. */
332 #define REG_NOSUB (1 << 3)
333
334
335 /* POSIX 'eflags' bits (i.e., information for regexec). */
336
337 /* If this bit is set, then the beginning-of-line operator doesn't match
338 the beginning of the string (presumably because it's not the
339 beginning of a line).
340 If not set, then the beginning-of-line operator does match the
341 beginning of the string. */
342 #define REG_NOTBOL 1
343
344 /* Like REG_NOTBOL, except for the end-of-line. */
345 #define REG_NOTEOL (1 << 1)
346
347 /* Use PMATCH[0] to delimit the start and end of the search in the
348 buffer. */
349 #define REG_STARTEND (1 << 2)
350
351
352 /* If any error codes are removed, changed, or added, update the
353 '__re_error_msgid' table in regcomp.c. */
354
355 typedef enum
356 {
357 _REG_ENOSYS = -1, /* This will never happen for this implementation. */
358 _REG_NOERROR = 0, /* Success. */
359 _REG_NOMATCH, /* Didn't find a match (for regexec). */
360
361 /* POSIX regcomp return error codes. (In the order listed in the
362 standard.) */
363 _REG_BADPAT, /* Invalid pattern. */
364 _REG_ECOLLATE, /* Invalid collating element. */
365 _REG_ECTYPE, /* Invalid character class name. */
366 _REG_EESCAPE, /* Trailing backslash. */
367 _REG_ESUBREG, /* Invalid back reference. */
368 _REG_EBRACK, /* Unmatched left bracket. */
369 _REG_EPAREN, /* Parenthesis imbalance. */
370 _REG_EBRACE, /* Unmatched \{. */
371 _REG_BADBR, /* Invalid contents of \{\}. */
372 _REG_ERANGE, /* Invalid range end. */
373 _REG_ESPACE, /* Ran out of memory. */
374 _REG_BADRPT, /* No preceding re for repetition op. */
375
376 /* Error codes we've added. */
377 _REG_EEND, /* Premature end. */
378 _REG_ESIZE, /* Too large (e.g., repeat count too large). */
379 _REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */
380 } reg_errcode_t;
381
382 #if defined _XOPEN_SOURCE || defined __USE_XOPEN2K
383 # define REG_ENOSYS _REG_ENOSYS
384 #endif
385 #define REG_NOERROR _REG_NOERROR
386 #define REG_NOMATCH _REG_NOMATCH
387 #define REG_BADPAT _REG_BADPAT
388 #define REG_ECOLLATE _REG_ECOLLATE
389 #define REG_ECTYPE _REG_ECTYPE
390 #define REG_EESCAPE _REG_EESCAPE
391 #define REG_ESUBREG _REG_ESUBREG
392 #define REG_EBRACK _REG_EBRACK
393 #define REG_EPAREN _REG_EPAREN
394 #define REG_EBRACE _REG_EBRACE
395 #define REG_BADBR _REG_BADBR
396 #define REG_ERANGE _REG_ERANGE
397 #define REG_ESPACE _REG_ESPACE
398 #define REG_BADRPT _REG_BADRPT
399 #define REG_EEND _REG_EEND
400 #define REG_ESIZE _REG_ESIZE
401 #define REG_ERPAREN _REG_ERPAREN
402 \f
403 /* This data structure represents a compiled pattern. Before calling
404 the pattern compiler, the fields 'buffer', 'allocated', 'fastmap',
405 and 'translate' can be set. After the pattern has been compiled,
406 the fields 're_nsub', 'not_bol' and 'not_eol' are available. All
407 other fields are private to the regex routines. */
408
409 #ifndef RE_TRANSLATE_TYPE
410 # define __RE_TRANSLATE_TYPE unsigned char *
411 # ifdef __USE_GNU
412 # define RE_TRANSLATE_TYPE __RE_TRANSLATE_TYPE
413 # endif
414 #endif
415
416 #ifdef __USE_GNU
417 # define __REPB_PREFIX(name) name
418 #else
419 # define __REPB_PREFIX(name) __##name
420 #endif
421
422 struct re_pattern_buffer
423 {
424 /* Space that holds the compiled pattern. The type
425 'struct re_dfa_t' is private and is not declared here. */
426 struct re_dfa_t *__REPB_PREFIX(buffer);
427
428 /* Number of bytes to which 'buffer' points. */
429 __re_long_size_t __REPB_PREFIX(allocated);
430
431 /* Number of bytes actually used in 'buffer'. */
432 __re_long_size_t __REPB_PREFIX(used);
433
434 /* Syntax setting with which the pattern was compiled. */
435 reg_syntax_t __REPB_PREFIX(syntax);
436
437 /* Pointer to a fastmap, if any, otherwise zero. re_search uses the
438 fastmap, if there is one, to skip over impossible starting points
439 for matches. */
440 char *__REPB_PREFIX(fastmap);
441
442 /* Either a translate table to apply to all characters before
443 comparing them, or zero for no translation. The translation is
444 applied to a pattern when it is compiled and to a string when it
445 is matched. */
446 __RE_TRANSLATE_TYPE __REPB_PREFIX(translate);
447
448 /* Number of subexpressions found by the compiler. */
449 size_t re_nsub;
450
451 /* Zero if this pattern cannot match the empty string, one else.
452 Well, in truth it's used only in 're_search_2', to see whether or
453 not we should use the fastmap, so we don't set this absolutely
454 perfectly; see 're_compile_fastmap' (the "duplicate" case). */
455 unsigned __REPB_PREFIX(can_be_null) : 1;
456
457 /* If REGS_UNALLOCATED, allocate space in the 'regs' structure
458 for 'max (RE_NREGS, re_nsub + 1)' groups.
459 If REGS_REALLOCATE, reallocate space if necessary.
460 If REGS_FIXED, use what's there. */
461 #ifdef __USE_GNU
462 # define REGS_UNALLOCATED 0
463 # define REGS_REALLOCATE 1
464 # define REGS_FIXED 2
465 #endif
466 unsigned __REPB_PREFIX(regs_allocated) : 2;
467
468 /* Set to zero when 're_compile_pattern' compiles a pattern; set to
469 one by 're_compile_fastmap' if it updates the fastmap. */
470 unsigned __REPB_PREFIX(fastmap_accurate) : 1;
471
472 /* If set, 're_match_2' does not return information about
473 subexpressions. */
474 unsigned __REPB_PREFIX(no_sub) : 1;
475
476 /* If set, a beginning-of-line anchor doesn't match at the beginning
477 of the string. */
478 unsigned __REPB_PREFIX(not_bol) : 1;
479
480 /* Similarly for an end-of-line anchor. */
481 unsigned __REPB_PREFIX(not_eol) : 1;
482
483 /* If true, an anchor at a newline matches. */
484 unsigned __REPB_PREFIX(newline_anchor) : 1;
485 };
486
487 typedef struct re_pattern_buffer regex_t;
488 \f
489 /* Type for byte offsets within the string. POSIX mandates this. */
490 #ifdef _REGEX_LARGE_OFFSETS
491 /* POSIX 1003.1-2008 requires that regoff_t be at least as wide as
492 ptrdiff_t and ssize_t. We don't know of any hosts where ptrdiff_t
493 is wider than ssize_t, so ssize_t is safe. */
494 typedef ssize_t regoff_t;
495 #else
496 /* The traditional GNU regex implementation mishandles strings longer
497 than INT_MAX. */
498 typedef int regoff_t;
499 #endif
500
501
502 #ifdef __USE_GNU
503 /* This is the structure we store register match data in. See
504 regex.texinfo for a full description of what registers match. */
505 struct re_registers
506 {
507 __re_size_t num_regs;
508 regoff_t *start;
509 regoff_t *end;
510 };
511
512
513 /* If 'regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
514 're_match_2' returns information about at least this many registers
515 the first time a 'regs' structure is passed. */
516 # ifndef RE_NREGS
517 # define RE_NREGS 30
518 # endif
519 #endif
520
521
522 /* POSIX specification for registers. Aside from the different names than
523 're_registers', POSIX uses an array of structures, instead of a
524 structure of arrays. */
525 typedef struct
526 {
527 regoff_t rm_so; /* Byte offset from string's start to substring's start. */
528 regoff_t rm_eo; /* Byte offset from string's start to substring's end. */
529 } regmatch_t;
530 \f
531 /* Declarations for routines. */
532
533 #ifdef __USE_GNU
534 /* Sets the current default syntax to SYNTAX, and return the old syntax.
535 You can also simply assign to the 're_syntax_options' variable. */
536 extern reg_syntax_t re_set_syntax (reg_syntax_t __syntax);
537
538 /* Compile the regular expression PATTERN, with length LENGTH
539 and syntax given by the global 're_syntax_options', into the buffer
540 BUFFER. Return NULL if successful, and an error string if not.
541
542 To free the allocated storage, you must call 'regfree' on BUFFER.
543 Note that the translate table must either have been initialised by
544 'regcomp', with a malloc'ed value, or set to NULL before calling
545 'regfree'. */
546 extern const char *re_compile_pattern (const char *__pattern, size_t __length,
547 struct re_pattern_buffer *__buffer);
548
549
550 /* Compile a fastmap for the compiled pattern in BUFFER; used to
551 accelerate searches. Return 0 if successful and -2 if was an
552 internal error. */
553 extern int re_compile_fastmap (struct re_pattern_buffer *__buffer);
554
555
556 /* Search in the string STRING (with length LENGTH) for the pattern
557 compiled into BUFFER. Start searching at position START, for RANGE
558 characters. Return the starting position of the match, -1 for no
559 match, or -2 for an internal error. Also return register
560 information in REGS (if REGS and BUFFER->no_sub are nonzero). */
561 extern regoff_t re_search (struct re_pattern_buffer *__buffer,
562 const char *__string, __re_idx_t __length,
563 __re_idx_t __start, regoff_t __range,
564 struct re_registers *__regs);
565
566
567 /* Like 're_search', but search in the concatenation of STRING1 and
568 STRING2. Also, stop searching at index START + STOP. */
569 extern regoff_t re_search_2 (struct re_pattern_buffer *__buffer,
570 const char *__string1, __re_idx_t __length1,
571 const char *__string2, __re_idx_t __length2,
572 __re_idx_t __start, regoff_t __range,
573 struct re_registers *__regs,
574 __re_idx_t __stop);
575
576
577 /* Like 're_search', but return how many characters in STRING the regexp
578 in BUFFER matched, starting at position START. */
579 extern regoff_t re_match (struct re_pattern_buffer *__buffer,
580 const char *__string, __re_idx_t __length,
581 __re_idx_t __start, struct re_registers *__regs);
582
583
584 /* Relates to 're_match' as 're_search_2' relates to 're_search'. */
585 extern regoff_t re_match_2 (struct re_pattern_buffer *__buffer,
586 const char *__string1, __re_idx_t __length1,
587 const char *__string2, __re_idx_t __length2,
588 __re_idx_t __start, struct re_registers *__regs,
589 __re_idx_t __stop);
590
591
592 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
593 ENDS. Subsequent matches using BUFFER and REGS will use this memory
594 for recording register information. STARTS and ENDS must be
595 allocated with malloc, and must each be at least 'NUM_REGS * sizeof
596 (regoff_t)' bytes long.
597
598 If NUM_REGS == 0, then subsequent matches should allocate their own
599 register data.
600
601 Unless this function is called, the first search or match using
602 BUFFER will allocate its own register data, without
603 freeing the old data. */
604 extern void re_set_registers (struct re_pattern_buffer *__buffer,
605 struct re_registers *__regs,
606 __re_size_t __num_regs,
607 regoff_t *__starts, regoff_t *__ends);
608 #endif /* Use GNU */
609
610 #if defined _REGEX_RE_COMP || (defined _LIBC && defined __USE_BSD)
611 # ifndef _CRAY
612 /* 4.2 bsd compatibility. */
613 extern char *re_comp (const char *);
614 extern int re_exec (const char *);
615 # endif
616 #endif
617
618 /* GCC 2.95 and later have "__restrict"; C99 compilers have
619 "restrict", and "configure" may have defined "restrict".
620 Other compilers use __restrict, __restrict__, and _Restrict, and
621 'configure' might #define 'restrict' to those words, so pick a
622 different name. */
623 #ifndef _Restrict_
624 # if 199901L <= __STDC_VERSION__
625 # define _Restrict_ restrict
626 # elif 2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__)
627 # define _Restrict_ __restrict
628 # else
629 # define _Restrict_
630 # endif
631 #endif
632 /* gcc 3.1 and up support the [restrict] syntax. Don't trust
633 sys/cdefs.h's definition of __restrict_arr, though, as it
634 mishandles gcc -ansi -pedantic. */
635 #ifndef _Restrict_arr_
636 # if ((199901L <= __STDC_VERSION__ \
637 || ((3 < __GNUC__ || (3 == __GNUC__ && 1 <= __GNUC_MINOR__)) \
638 && !defined __STRICT_ANSI__)) \
639 && !defined __GNUG__)
640 # define _Restrict_arr_ _Restrict_
641 # else
642 # define _Restrict_arr_
643 # endif
644 #endif
645
646 /* POSIX compatibility. */
647 extern int regcomp (regex_t *_Restrict_ __preg,
648 const char *_Restrict_ __pattern,
649 int __cflags);
650
651 extern int regexec (const regex_t *_Restrict_ __preg,
652 const char *_Restrict_ __string, size_t __nmatch,
653 regmatch_t __pmatch[_Restrict_arr_],
654 int __eflags);
655
656 extern size_t regerror (int __errcode, const regex_t *_Restrict_ __preg,
657 char *_Restrict_ __errbuf, size_t __errbuf_size);
658
659 extern void regfree (regex_t *__preg);
660
661
662 #ifdef __cplusplus
663 }
664 #endif /* C++ */
665
666 #endif /* regex.h */