file tst-timer.c was initially added on branch fedora-branch.
[glibc/history.git] / posix / regex.h
blob807c404ec24e2aa316d6e3061a63bc07f32e179a
1 /* Definitions for data structures and routines for the regular
2 expression library.
3 Copyright (C) 1985,1989-93,1995-98,2000,2001,2002,2003,2005,2006
4 Free Software Foundation, Inc.
5 This file is part of the GNU C Library.
7 The GNU C Library is free software; you can redistribute it and/or
8 modify it under the terms of the GNU Lesser General Public
9 License as published by the Free Software Foundation; either
10 version 2.1 of the License, or (at your option) any later version.
12 The GNU C Library is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 Lesser General Public License for more details.
17 You should have received a copy of the GNU Lesser General Public
18 License along with the GNU C Library; if not, write to the Free
19 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
20 02111-1307 USA. */
22 #ifndef _REGEX_H
23 #define _REGEX_H 1
25 #include <sys/types.h>
27 /* Allow the use in C++ code. */
28 #ifdef __cplusplus
29 extern "C" {
30 #endif
32 /* The following two types have to be signed and unsigned integer type
33 wide enough to hold a value of a pointer. For most ANSI compilers
34 ptrdiff_t and size_t should be likely OK. Still size of these two
35 types is 2 for Microsoft C. Ugh... */
36 typedef long int s_reg_t;
37 typedef unsigned long int active_reg_t;
39 /* The following bits are used to determine the regexp syntax we
40 recognize. The set/not-set meanings are chosen so that Emacs syntax
41 remains the value 0. The bits are given in alphabetical order, and
42 the definitions shifted by one from the previous bit; thus, when we
43 add or remove a bit, only one other definition need change. */
44 typedef unsigned long int reg_syntax_t;
46 /* If this bit is not set, then \ inside a bracket expression is literal.
47 If set, then such a \ quotes the following character. */
48 #define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1)
50 /* If this bit is not set, then + and ? are operators, and \+ and \? are
51 literals.
52 If set, then \+ and \? are operators and + and ? are literals. */
53 #define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1)
55 /* If this bit is set, then character classes are supported. They are:
56 [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:],
57 [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:].
58 If not set, then character classes are not supported. */
59 #define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1)
61 /* If this bit is set, then ^ and $ are always anchors (outside bracket
62 expressions, of course).
63 If this bit is not set, then it depends:
64 ^ is an anchor if it is at the beginning of a regular
65 expression or after an open-group or an alternation operator;
66 $ is an anchor if it is at the end of a regular expression, or
67 before a close-group or an alternation operator.
69 This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because
70 POSIX draft 11.2 says that * etc. in leading positions is undefined.
71 We already implemented a previous draft which made those constructs
72 invalid, though, so we haven't changed the code back. */
73 #define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1)
75 /* If this bit is set, then special characters are always special
76 regardless of where they are in the pattern.
77 If this bit is not set, then special characters are special only in
78 some contexts; otherwise they are ordinary. Specifically,
79 * + ? and intervals are only special when not after the beginning,
80 open-group, or alternation operator. */
81 #define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1)
83 /* If this bit is set, then *, +, ?, and { cannot be first in an re or
84 immediately after an alternation or begin-group operator. */
85 #define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1)
87 /* If this bit is set, then . matches newline.
88 If not set, then it doesn't. */
89 #define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1)
91 /* If this bit is set, then . doesn't match NUL.
92 If not set, then it does. */
93 #define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1)
95 /* If this bit is set, nonmatching lists [^...] do not match newline.
96 If not set, they do. */
97 #define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1)
99 /* If this bit is set, either \{...\} or {...} defines an
100 interval, depending on RE_NO_BK_BRACES.
101 If not set, \{, \}, {, and } are literals. */
102 #define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1)
104 /* If this bit is set, +, ? and | aren't recognized as operators.
105 If not set, they are. */
106 #define RE_LIMITED_OPS (RE_INTERVALS << 1)
108 /* If this bit is set, newline is an alternation operator.
109 If not set, newline is literal. */
110 #define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1)
112 /* If this bit is set, then `{...}' defines an interval, and \{ and \}
113 are literals.
114 If not set, then `\{...\}' defines an interval. */
115 #define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1)
117 /* If this bit is set, (...) defines a group, and \( and \) are literals.
118 If not set, \(...\) defines a group, and ( and ) are literals. */
119 #define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1)
121 /* If this bit is set, then \<digit> matches <digit>.
122 If not set, then \<digit> is a back-reference. */
123 #define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1)
125 /* If this bit is set, then | is an alternation operator, and \| is literal.
126 If not set, then \| is an alternation operator, and | is literal. */
127 #define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1)
129 /* If this bit is set, then an ending range point collating higher
130 than the starting range point, as in [z-a], is invalid.
131 If not set, then when ending range point collates higher than the
132 starting range point, the range is ignored. */
133 #define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1)
135 /* If this bit is set, then an unmatched ) is ordinary.
136 If not set, then an unmatched ) is invalid. */
137 #define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1)
139 /* If this bit is set, succeed as soon as we match the whole pattern,
140 without further backtracking. */
141 #define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1)
143 /* If this bit is set, do not process the GNU regex operators.
144 If not set, then the GNU regex operators are recognized. */
145 #define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1)
147 /* If this bit is set, turn on internal regex debugging.
148 If not set, and debugging was on, turn it off.
149 This only works if regex.c is compiled -DDEBUG.
150 We define this bit always, so that all that's needed to turn on
151 debugging is to recompile regex.c; the calling code can always have
152 this bit set, and it won't affect anything in the normal case. */
153 #define RE_DEBUG (RE_NO_GNU_OPS << 1)
155 /* If this bit is set, a syntactically invalid interval is treated as
156 a string of ordinary characters. For example, the ERE 'a{1' is
157 treated as 'a\{1'. */
158 #define RE_INVALID_INTERVAL_ORD (RE_DEBUG << 1)
160 /* If this bit is set, then ignore case when matching.
161 If not set, then case is significant. */
162 #define RE_ICASE (RE_INVALID_INTERVAL_ORD << 1)
164 /* This bit is used internally like RE_CONTEXT_INDEP_ANCHORS but only
165 for ^, because it is difficult to scan the regex backwards to find
166 whether ^ should be special. */
167 #define RE_CARET_ANCHORS_HERE (RE_ICASE << 1)
169 /* If this bit is set, then \{ cannot be first in an bre or
170 immediately after an alternation or begin-group operator. */
171 #define RE_CONTEXT_INVALID_DUP (RE_CARET_ANCHORS_HERE << 1)
173 /* If this bit is set, then no_sub will be set to 1 during
174 re_compile_pattern. */
175 #define RE_NO_SUB (RE_CONTEXT_INVALID_DUP << 1)
177 /* This global variable defines the particular regexp syntax to use (for
178 some interfaces). When a regexp is compiled, the syntax used is
179 stored in the pattern buffer, so changing this does not affect
180 already-compiled regexps. */
181 extern reg_syntax_t re_syntax_options;
183 /* Define combinations of the above bits for the standard possibilities.
184 (The [[[ comments delimit what gets put into the Texinfo file, so
185 don't delete them!) */
186 /* [[[begin syntaxes]]] */
187 #define RE_SYNTAX_EMACS 0
189 #define RE_SYNTAX_AWK \
190 (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \
191 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
192 | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \
193 | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \
194 | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS)
196 #define RE_SYNTAX_GNU_AWK \
197 ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DEBUG) \
198 & ~(RE_DOT_NOT_NULL | RE_INTERVALS | RE_CONTEXT_INDEP_OPS \
199 | RE_CONTEXT_INVALID_OPS ))
201 #define RE_SYNTAX_POSIX_AWK \
202 (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
203 | RE_INTERVALS | RE_NO_GNU_OPS)
205 #define RE_SYNTAX_GREP \
206 (RE_BK_PLUS_QM | RE_CHAR_CLASSES \
207 | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \
208 | RE_NEWLINE_ALT)
210 #define RE_SYNTAX_EGREP \
211 (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \
212 | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \
213 | RE_NEWLINE_ALT | RE_NO_BK_PARENS \
214 | RE_NO_BK_VBAR)
216 #define RE_SYNTAX_POSIX_EGREP \
217 (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES \
218 | RE_INVALID_INTERVAL_ORD)
220 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */
221 #define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
223 #define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
225 /* Syntax bits common to both basic and extended POSIX regex syntax. */
226 #define _RE_SYNTAX_POSIX_COMMON \
227 (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \
228 | RE_INTERVALS | RE_NO_EMPTY_RANGES)
230 #define RE_SYNTAX_POSIX_BASIC \
231 (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM | RE_CONTEXT_INVALID_DUP)
233 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
234 RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this
235 isn't minimal, since other operators, such as \`, aren't disabled. */
236 #define RE_SYNTAX_POSIX_MINIMAL_BASIC \
237 (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
239 #define RE_SYNTAX_POSIX_EXTENDED \
240 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
241 | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \
242 | RE_NO_BK_PARENS | RE_NO_BK_VBAR \
243 | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD)
245 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is
246 removed and RE_NO_BK_REFS is added. */
247 #define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \
248 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
249 | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \
250 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
251 | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD)
252 /* [[[end syntaxes]]] */
254 /* Maximum number of duplicates an interval can allow. Some systems
255 (erroneously) define this in other header files, but we want our
256 value, so remove any previous define. */
257 #ifdef RE_DUP_MAX
258 # undef RE_DUP_MAX
259 #endif
260 /* If sizeof(int) == 2, then ((1 << 15) - 1) overflows. */
261 #define RE_DUP_MAX (0x7fff)
264 /* POSIX `cflags' bits (i.e., information for `regcomp'). */
266 /* If this bit is set, then use extended regular expression syntax.
267 If not set, then use basic regular expression syntax. */
268 #define REG_EXTENDED 1
270 /* If this bit is set, then ignore case when matching.
271 If not set, then case is significant. */
272 #define REG_ICASE (REG_EXTENDED << 1)
274 /* If this bit is set, then anchors do not match at newline
275 characters in the string.
276 If not set, then anchors do match at newlines. */
277 #define REG_NEWLINE (REG_ICASE << 1)
279 /* If this bit is set, then report only success or fail in regexec.
280 If not set, then returns differ between not matching and errors. */
281 #define REG_NOSUB (REG_NEWLINE << 1)
284 /* POSIX `eflags' bits (i.e., information for regexec). */
286 /* If this bit is set, then the beginning-of-line operator doesn't match
287 the beginning of the string (presumably because it's not the
288 beginning of a line).
289 If not set, then the beginning-of-line operator does match the
290 beginning of the string. */
291 #define REG_NOTBOL 1
293 /* Like REG_NOTBOL, except for the end-of-line. */
294 #define REG_NOTEOL (1 << 1)
296 /* Use PMATCH[0] to delimit the start and end of the search in the
297 buffer. */
298 #define REG_STARTEND (1 << 2)
301 /* If any error codes are removed, changed, or added, update the
302 `re_error_msg' table in regex.c. */
303 typedef enum
305 #ifdef _XOPEN_SOURCE
306 REG_ENOSYS = -1, /* This will never happen for this implementation. */
307 #endif
309 REG_NOERROR = 0, /* Success. */
310 REG_NOMATCH, /* Didn't find a match (for regexec). */
312 /* POSIX regcomp return error codes. (In the order listed in the
313 standard.) */
314 REG_BADPAT, /* Invalid pattern. */
315 REG_ECOLLATE, /* Inalid collating element. */
316 REG_ECTYPE, /* Invalid character class name. */
317 REG_EESCAPE, /* Trailing backslash. */
318 REG_ESUBREG, /* Invalid back reference. */
319 REG_EBRACK, /* Unmatched left bracket. */
320 REG_EPAREN, /* Parenthesis imbalance. */
321 REG_EBRACE, /* Unmatched \{. */
322 REG_BADBR, /* Invalid contents of \{\}. */
323 REG_ERANGE, /* Invalid range end. */
324 REG_ESPACE, /* Ran out of memory. */
325 REG_BADRPT, /* No preceding re for repetition op. */
327 /* Error codes we've added. */
328 REG_EEND, /* Premature end. */
329 REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */
330 REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */
331 } reg_errcode_t;
333 /* This data structure represents a compiled pattern. Before calling
334 the pattern compiler, the fields `buffer', `allocated', `fastmap',
335 `translate', and `no_sub' can be set. After the pattern has been
336 compiled, the `re_nsub' field is available. All other fields are
337 private to the regex routines. */
339 #ifndef RE_TRANSLATE_TYPE
340 # define RE_TRANSLATE_TYPE unsigned char *
341 #endif
343 struct re_pattern_buffer
345 /* Space that holds the compiled pattern. It is declared as
346 `unsigned char *' because its elements are sometimes used as
347 array indexes. */
348 unsigned char *buffer;
350 /* Number of bytes to which `buffer' points. */
351 unsigned long int allocated;
353 /* Number of bytes actually used in `buffer'. */
354 unsigned long int used;
356 /* Syntax setting with which the pattern was compiled. */
357 reg_syntax_t syntax;
359 /* Pointer to a fastmap, if any, otherwise zero. re_search uses the
360 fastmap, if there is one, to skip over impossible starting points
361 for matches. */
362 char *fastmap;
364 /* Either a translate table to apply to all characters before
365 comparing them, or zero for no translation. The translation is
366 applied to a pattern when it is compiled and to a string when it
367 is matched. */
368 RE_TRANSLATE_TYPE translate;
370 /* Number of subexpressions found by the compiler. */
371 size_t re_nsub;
373 /* Zero if this pattern cannot match the empty string, one else.
374 Well, in truth it's used only in `re_search_2', to see whether or
375 not we should use the fastmap, so we don't set this absolutely
376 perfectly; see `re_compile_fastmap' (the `duplicate' case). */
377 unsigned can_be_null : 1;
379 /* If REGS_UNALLOCATED, allocate space in the `regs' structure
380 for `max (RE_NREGS, re_nsub + 1)' groups.
381 If REGS_REALLOCATE, reallocate space if necessary.
382 If REGS_FIXED, use what's there. */
383 #define REGS_UNALLOCATED 0
384 #define REGS_REALLOCATE 1
385 #define REGS_FIXED 2
386 unsigned regs_allocated : 2;
388 /* Set to zero when `regex_compile' compiles a pattern; set to one
389 by `re_compile_fastmap' if it updates the fastmap. */
390 unsigned fastmap_accurate : 1;
392 /* If set, `re_match_2' does not return information about
393 subexpressions. */
394 unsigned no_sub : 1;
396 /* If set, a beginning-of-line anchor doesn't match at the beginning
397 of the string. */
398 unsigned not_bol : 1;
400 /* Similarly for an end-of-line anchor. */
401 unsigned not_eol : 1;
403 /* If true, an anchor at a newline matches. */
404 unsigned newline_anchor : 1;
407 typedef struct re_pattern_buffer regex_t;
409 /* Type for byte offsets within the string. POSIX mandates this. */
410 typedef int regoff_t;
413 /* This is the structure we store register match data in. See
414 regex.texinfo for a full description of what registers match. */
415 struct re_registers
417 unsigned num_regs;
418 regoff_t *start;
419 regoff_t *end;
423 /* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
424 `re_match_2' returns information about at least this many registers
425 the first time a `regs' structure is passed. */
426 #ifndef RE_NREGS
427 # define RE_NREGS 30
428 #endif
431 /* POSIX specification for registers. Aside from the different names than
432 `re_registers', POSIX uses an array of structures, instead of a
433 structure of arrays. */
434 typedef struct
436 regoff_t rm_so; /* Byte offset from string's start to substring's start. */
437 regoff_t rm_eo; /* Byte offset from string's start to substring's end. */
438 } regmatch_t;
440 /* Declarations for routines. */
442 /* Sets the current default syntax to SYNTAX, and return the old syntax.
443 You can also simply assign to the `re_syntax_options' variable. */
444 extern reg_syntax_t re_set_syntax (reg_syntax_t __syntax);
446 /* Compile the regular expression PATTERN, with length LENGTH
447 and syntax given by the global `re_syntax_options', into the buffer
448 BUFFER. Return NULL if successful, and an error string if not. */
449 extern const char *re_compile_pattern (const char *__pattern, size_t __length,
450 struct re_pattern_buffer *__buffer);
453 /* Compile a fastmap for the compiled pattern in BUFFER; used to
454 accelerate searches. Return 0 if successful and -2 if was an
455 internal error. */
456 extern int re_compile_fastmap (struct re_pattern_buffer *__buffer);
459 /* Search in the string STRING (with length LENGTH) for the pattern
460 compiled into BUFFER. Start searching at position START, for RANGE
461 characters. Return the starting position of the match, -1 for no
462 match, or -2 for an internal error. Also return register
463 information in REGS (if REGS and BUFFER->no_sub are nonzero). */
464 extern int re_search (struct re_pattern_buffer *__buffer, const char *__string,
465 int __length, int __start, int __range,
466 struct re_registers *__regs);
469 /* Like `re_search', but search in the concatenation of STRING1 and
470 STRING2. Also, stop searching at index START + STOP. */
471 extern int re_search_2 (struct re_pattern_buffer *__buffer,
472 const char *__string1, int __length1,
473 const char *__string2, int __length2, int __start,
474 int __range, struct re_registers *__regs, int __stop);
477 /* Like `re_search', but return how many characters in STRING the regexp
478 in BUFFER matched, starting at position START. */
479 extern int re_match (struct re_pattern_buffer *__buffer, const char *__string,
480 int __length, int __start, struct re_registers *__regs);
483 /* Relates to `re_match' as `re_search_2' relates to `re_search'. */
484 extern int re_match_2 (struct re_pattern_buffer *__buffer,
485 const char *__string1, int __length1,
486 const char *__string2, int __length2, int __start,
487 struct re_registers *__regs, int __stop);
490 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
491 ENDS. Subsequent matches using BUFFER and REGS will use this memory
492 for recording register information. STARTS and ENDS must be
493 allocated with malloc, and must each be at least `NUM_REGS * sizeof
494 (regoff_t)' bytes long.
496 If NUM_REGS == 0, then subsequent matches should allocate their own
497 register data.
499 Unless this function is called, the first search or match using
500 PATTERN_BUFFER will allocate its own register data, without
501 freeing the old data. */
502 extern void re_set_registers (struct re_pattern_buffer *__buffer,
503 struct re_registers *__regs,
504 unsigned int __num_regs,
505 regoff_t *__starts, regoff_t *__ends);
507 #if defined _REGEX_RE_COMP || defined _LIBC
508 # ifndef _CRAY
509 /* 4.2 bsd compatibility. */
510 extern char *re_comp (const char *);
511 extern int re_exec (const char *);
512 # endif
513 #endif
515 /* GCC 2.95 and later have "__restrict"; C99 compilers have
516 "restrict", and "configure" may have defined "restrict". */
517 #ifndef __restrict
518 # if ! (2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__))
519 # if defined restrict || 199901L <= __STDC_VERSION__
520 # define __restrict restrict
521 # else
522 # define __restrict
523 # endif
524 # endif
525 #endif
526 /* gcc 3.1 and up support the [restrict] syntax. */
527 #ifndef __restrict_arr
528 # if (__GNUC__ > 3 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)) \
529 && !defined __GNUG__
530 # define __restrict_arr __restrict
531 # else
532 # define __restrict_arr
533 # endif
534 #endif
536 /* POSIX compatibility. */
537 extern int regcomp (regex_t *__restrict __preg,
538 const char *__restrict __pattern,
539 int __cflags);
541 extern int regexec (const regex_t *__restrict __preg,
542 const char *__restrict __string, size_t __nmatch,
543 regmatch_t __pmatch[__restrict_arr],
544 int __eflags);
546 extern size_t regerror (int __errcode, const regex_t *__restrict __preg,
547 char *__restrict __errbuf, size_t __errbuf_size);
549 extern void regfree (regex_t *__preg);
552 #ifdef __cplusplus
554 #endif /* C++ */
556 #endif /* regex.h */