4 /* Definitions for data structures and routines for the regular
6 Copyright (C) 1985,1989-93,1995-98,2000,2001,2002,2003,2005,2006,2008
7 Free Software Foundation, Inc.
8 This file is part of the GNU C Library.
10 The GNU C Library is free software; you can redistribute it and/or
11 modify it under the terms of the GNU Lesser General Public
12 License as published by the Free Software Foundation; either
13 version 2.1 of the License, or (at your option) any later version.
15 The GNU C Library is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 Lesser General Public License for more details.
20 You should have received a copy of the GNU Lesser General Public
21 License along with the GNU C Library; if not, see
22 <http://www.gnu.org/licenses/>. */
31 #ifdef HAVE_SYS_TYPES_H
32 #include <sys/types.h>
39 /* Allow the use in C++ code. */
44 #define regcomp git_regcomp
45 #define regexec git_regexec
46 #define regerror git_regerror
47 #define regfree git_regfree
49 /* The following two types have to be signed and unsigned integer type
50 wide enough to hold a value of a pointer. For most ANSI compilers
51 ptrdiff_t and size_t should be likely OK. Still size of these two
52 types is 2 for Microsoft C. Ugh... */
53 typedef long int s_reg_t
;
54 typedef unsigned long int active_reg_t
;
56 /* The following bits are used to determine the regexp syntax we
57 recognize. The set/not-set meanings are chosen so that Emacs syntax
58 remains the value 0. The bits are given in alphabetical order, and
59 the definitions shifted by one from the previous bit; thus, when we
60 add or remove a bit, only one other definition need change. */
61 typedef unsigned long int reg_syntax_t
;
64 /* If this bit is not set, then \ inside a bracket expression is literal.
65 If set, then such a \ quotes the following character. */
66 # define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1)
68 /* If this bit is not set, then + and ? are operators, and \+ and \? are
70 If set, then \+ and \? are operators and + and ? are literals. */
71 # define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1)
73 /* If this bit is set, then character classes are supported. They are:
74 [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:],
75 [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:].
76 If not set, then character classes are not supported. */
77 # define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1)
79 /* If this bit is set, then ^ and $ are always anchors (outside bracket
80 expressions, of course).
81 If this bit is not set, then it depends:
82 ^ is an anchor if it is at the beginning of a regular
83 expression or after an open-group or an alternation operator;
84 $ is an anchor if it is at the end of a regular expression, or
85 before a close-group or an alternation operator.
87 This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because
88 POSIX draft 11.2 says that * etc. in leading positions is undefined.
89 We already implemented a previous draft which made those constructs
90 invalid, though, so we haven't changed the code back. */
91 # define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1)
93 /* If this bit is set, then special characters are always special
94 regardless of where they are in the pattern.
95 If this bit is not set, then special characters are special only in
96 some contexts; otherwise they are ordinary. Specifically,
97 * + ? and intervals are only special when not after the beginning,
98 open-group, or alternation operator. */
99 # define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1)
101 /* If this bit is set, then *, +, ?, and { cannot be first in an re or
102 immediately after an alternation or begin-group operator. */
103 # define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1)
105 /* If this bit is set, then . matches newline.
106 If not set, then it doesn't. */
107 # define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1)
109 /* If this bit is set, then . doesn't match NUL.
110 If not set, then it does. */
111 # define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1)
113 /* If this bit is set, nonmatching lists [^...] do not match newline.
114 If not set, they do. */
115 # define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1)
117 /* If this bit is set, either \{...\} or {...} defines an
118 interval, depending on RE_NO_BK_BRACES.
119 If not set, \{, \}, {, and } are literals. */
120 # define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1)
122 /* If this bit is set, +, ? and | aren't recognized as operators.
123 If not set, they are. */
124 # define RE_LIMITED_OPS (RE_INTERVALS << 1)
126 /* If this bit is set, newline is an alternation operator.
127 If not set, newline is literal. */
128 # define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1)
130 /* If this bit is set, then `{...}' defines an interval, and \{ and \}
132 If not set, then `\{...\}' defines an interval. */
133 # define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1)
135 /* If this bit is set, (...) defines a group, and \( and \) are literals.
136 If not set, \(...\) defines a group, and ( and ) are literals. */
137 # define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1)
139 /* If this bit is set, then \<digit> matches <digit>.
140 If not set, then \<digit> is a back-reference. */
141 # define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1)
143 /* If this bit is set, then | is an alternation operator, and \| is literal.
144 If not set, then \| is an alternation operator, and | is literal. */
145 # define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1)
147 /* If this bit is set, then an ending range point collating higher
148 than the starting range point, as in [z-a], is invalid.
149 If not set, then when ending range point collates higher than the
150 starting range point, the range is ignored. */
151 # define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1)
153 /* If this bit is set, then an unmatched ) is ordinary.
154 If not set, then an unmatched ) is invalid. */
155 # define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1)
157 /* If this bit is set, succeed as soon as we match the whole pattern,
158 without further backtracking. */
159 # define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1)
161 /* If this bit is set, do not process the GNU regex operators.
162 If not set, then the GNU regex operators are recognized. */
163 # define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1)
165 /* If this bit is set, a syntactically invalid interval is treated as
166 a string of ordinary characters. For example, the ERE 'a{1' is
167 treated as 'a\{1'. */
168 # define RE_INVALID_INTERVAL_ORD (RE_NO_GNU_OPS << 1)
170 /* If this bit is set, then ignore case when matching.
171 If not set, then case is significant. */
172 # define RE_ICASE (RE_INVALID_INTERVAL_ORD << 1)
174 /* This bit is used internally like RE_CONTEXT_INDEP_ANCHORS but only
175 for ^, because it is difficult to scan the regex backwards to find
176 whether ^ should be special. */
177 # define RE_CARET_ANCHORS_HERE (RE_ICASE << 1)
179 /* If this bit is set, then \{ cannot be first in an bre or
180 immediately after an alternation or begin-group operator. */
181 # define RE_CONTEXT_INVALID_DUP (RE_CARET_ANCHORS_HERE << 1)
183 /* If this bit is set, then no_sub will be set to 1 during
184 re_compile_pattern. */
185 #define RE_NO_SUB (RE_CONTEXT_INVALID_DUP << 1)
188 /* This global variable defines the particular regexp syntax to use (for
189 some interfaces). When a regexp is compiled, the syntax used is
190 stored in the pattern buffer, so changing this does not affect
191 already-compiled regexps. */
192 extern reg_syntax_t re_syntax_options
;
195 /* Define combinations of the above bits for the standard possibilities.
196 (The [[[ comments delimit what gets put into the Texinfo file, so
197 don't delete them!) */
198 /* [[[begin syntaxes]]] */
199 #define RE_SYNTAX_EMACS 0
201 #define RE_SYNTAX_AWK \
202 (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \
203 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
204 | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \
205 | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \
206 | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS)
208 #define RE_SYNTAX_GNU_AWK \
209 ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
210 | RE_INVALID_INTERVAL_ORD) \
211 & ~(RE_DOT_NOT_NULL | RE_CONTEXT_INDEP_OPS \
212 | RE_CONTEXT_INVALID_OPS ))
214 #define RE_SYNTAX_POSIX_AWK \
215 (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
216 | RE_INTERVALS | RE_NO_GNU_OPS \
217 | RE_INVALID_INTERVAL_ORD)
219 #define RE_SYNTAX_GREP \
220 (RE_BK_PLUS_QM | RE_CHAR_CLASSES \
221 | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \
224 #define RE_SYNTAX_EGREP \
225 (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \
226 | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \
227 | RE_NEWLINE_ALT | RE_NO_BK_PARENS \
230 #define RE_SYNTAX_POSIX_EGREP \
231 (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES \
232 | RE_INVALID_INTERVAL_ORD)
234 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */
235 #define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
237 #define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
239 /* Syntax bits common to both basic and extended POSIX regex syntax. */
240 #define _RE_SYNTAX_POSIX_COMMON \
241 (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \
242 | RE_INTERVALS | RE_NO_EMPTY_RANGES)
244 #define RE_SYNTAX_POSIX_BASIC \
245 (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM | RE_CONTEXT_INVALID_DUP)
247 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
248 RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this
249 isn't minimal, since other operators, such as \`, aren't disabled. */
250 #define RE_SYNTAX_POSIX_MINIMAL_BASIC \
251 (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
253 #define RE_SYNTAX_POSIX_EXTENDED \
254 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
255 | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \
256 | RE_NO_BK_PARENS | RE_NO_BK_VBAR \
257 | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD)
259 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is
260 removed and RE_NO_BK_REFS is added. */
261 #define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \
262 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
263 | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \
264 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
265 | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD)
266 /* [[[end syntaxes]]] */
268 /* Maximum number of duplicates an interval can allow. Some systems
269 (erroneously) define this in other header files, but we want our
270 value, so remove any previous define. */
274 /* If sizeof(int) == 2, then ((1 << 15) - 1) overflows. */
275 # define RE_DUP_MAX (0x7fff)
279 /* POSIX `cflags' bits (i.e., information for `regcomp'). */
281 /* If this bit is set, then use extended regular expression syntax.
282 If not set, then use basic regular expression syntax. */
283 #define REG_EXTENDED 1
285 /* If this bit is set, then ignore case when matching.
286 If not set, then case is significant. */
287 #define REG_ICASE (REG_EXTENDED << 1)
289 /* If this bit is set, then anchors do not match at newline
290 characters in the string.
291 If not set, then anchors do match at newlines. */
292 #define REG_NEWLINE (REG_ICASE << 1)
294 /* If this bit is set, then report only success or fail in regexec.
295 If not set, then returns differ between not matching and errors. */
296 #define REG_NOSUB (REG_NEWLINE << 1)
299 /* POSIX `eflags' bits (i.e., information for regexec). */
301 /* If this bit is set, then the beginning-of-line operator doesn't match
302 the beginning of the string (presumably because it's not the
303 beginning of a line).
304 If not set, then the beginning-of-line operator does match the
305 beginning of the string. */
308 /* Like REG_NOTBOL, except for the end-of-line. */
309 #define REG_NOTEOL (1 << 1)
311 /* Use PMATCH[0] to delimit the start and end of the search in the
313 #define REG_STARTEND (1 << 2)
316 /* If any error codes are removed, changed, or added, update the
317 `re_error_msg' table in regex.c. */
320 #if defined _XOPEN_SOURCE || defined __USE_XOPEN2K
321 REG_ENOSYS
= -1, /* This will never happen for this implementation. */
324 REG_NOERROR
= 0, /* Success. */
325 REG_NOMATCH
, /* Didn't find a match (for regexec). */
327 /* POSIX regcomp return error codes. (In the order listed in the
329 REG_BADPAT
, /* Invalid pattern. */
330 REG_ECOLLATE
, /* Invalid collating element. */
331 REG_ECTYPE
, /* Invalid character class name. */
332 REG_EESCAPE
, /* Trailing backslash. */
333 REG_ESUBREG
, /* Invalid back reference. */
334 REG_EBRACK
, /* Unmatched left bracket. */
335 REG_EPAREN
, /* Parenthesis imbalance. */
336 REG_EBRACE
, /* Unmatched \{. */
337 REG_BADBR
, /* Invalid contents of \{\}. */
338 REG_ERANGE
, /* Invalid range end. */
339 REG_ESPACE
, /* Ran out of memory. */
340 REG_BADRPT
, /* No preceding re for repetition op. */
342 /* Error codes we've added. */
343 REG_EEND
, /* Premature end. */
344 REG_ESIZE
, /* Compiled pattern bigger than 2^16 bytes. */
345 REG_ERPAREN
/* Unmatched ) or \); not returned from regcomp. */
348 /* This data structure represents a compiled pattern. Before calling
349 the pattern compiler, the fields `buffer', `allocated', `fastmap',
350 `translate', and `no_sub' can be set. After the pattern has been
351 compiled, the `re_nsub' field is available. All other fields are
352 private to the regex routines. */
354 #ifndef RE_TRANSLATE_TYPE
355 # define __RE_TRANSLATE_TYPE unsigned char *
357 # define RE_TRANSLATE_TYPE __RE_TRANSLATE_TYPE
362 # define __REPB_PREFIX(name) name
364 # define __REPB_PREFIX(name) __##name
367 struct re_pattern_buffer
369 /* Space that holds the compiled pattern. It is declared as
370 `unsigned char *' because its elements are sometimes used as
372 unsigned char *__REPB_PREFIX(buffer
);
374 /* Number of bytes to which `buffer' points. */
375 unsigned long int __REPB_PREFIX(allocated
);
377 /* Number of bytes actually used in `buffer'. */
378 unsigned long int __REPB_PREFIX(used
);
380 /* Syntax setting with which the pattern was compiled. */
381 reg_syntax_t
__REPB_PREFIX(syntax
);
383 /* Pointer to a fastmap, if any, otherwise zero. re_search uses the
384 fastmap, if there is one, to skip over impossible starting points
386 char *__REPB_PREFIX(fastmap
);
388 /* Either a translate table to apply to all characters before
389 comparing them, or zero for no translation. The translation is
390 applied to a pattern when it is compiled and to a string when it
392 __RE_TRANSLATE_TYPE
__REPB_PREFIX(translate
);
394 /* Number of subexpressions found by the compiler. */
397 /* Zero if this pattern cannot match the empty string, one else.
398 Well, in truth it's used only in `re_search_2', to see whether or
399 not we should use the fastmap, so we don't set this absolutely
400 perfectly; see `re_compile_fastmap' (the `duplicate' case). */
401 unsigned __REPB_PREFIX(can_be_null
) : 1;
403 /* If REGS_UNALLOCATED, allocate space in the `regs' structure
404 for `max (RE_NREGS, re_nsub + 1)' groups.
405 If REGS_REALLOCATE, reallocate space if necessary.
406 If REGS_FIXED, use what's there. */
408 # define REGS_UNALLOCATED 0
409 # define REGS_REALLOCATE 1
410 # define REGS_FIXED 2
412 unsigned __REPB_PREFIX(regs_allocated
) : 2;
414 /* Set to zero when `regex_compile' compiles a pattern; set to one
415 by `re_compile_fastmap' if it updates the fastmap. */
416 unsigned __REPB_PREFIX(fastmap_accurate
) : 1;
418 /* If set, `re_match_2' does not return information about
420 unsigned __REPB_PREFIX(no_sub
) : 1;
422 /* If set, a beginning-of-line anchor doesn't match at the beginning
424 unsigned __REPB_PREFIX(not_bol
) : 1;
426 /* Similarly for an end-of-line anchor. */
427 unsigned __REPB_PREFIX(not_eol
) : 1;
429 /* If true, an anchor at a newline matches. */
430 unsigned __REPB_PREFIX(newline_anchor
) : 1;
433 typedef struct re_pattern_buffer regex_t
;
435 /* Type for byte offsets within the string. POSIX mandates this. */
436 typedef int regoff_t
;
440 /* This is the structure we store register match data in. See
441 regex.texinfo for a full description of what registers match. */
450 /* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
451 `re_match_2' returns information about at least this many registers
452 the first time a `regs' structure is passed. */
459 /* POSIX specification for registers. Aside from the different names than
460 `re_registers', POSIX uses an array of structures, instead of a
461 structure of arrays. */
464 regoff_t rm_so
; /* Byte offset from string's start to substring's start. */
465 regoff_t rm_eo
; /* Byte offset from string's start to substring's end. */
468 /* Declarations for routines. */
471 /* Sets the current default syntax to SYNTAX, and return the old syntax.
472 You can also simply assign to the `re_syntax_options' variable. */
473 extern reg_syntax_t
re_set_syntax (reg_syntax_t __syntax
);
475 /* Compile the regular expression PATTERN, with length LENGTH
476 and syntax given by the global `re_syntax_options', into the buffer
477 BUFFER. Return NULL if successful, and an error string if not. */
478 extern const char *re_compile_pattern (const char *__pattern
, size_t __length
,
479 struct re_pattern_buffer
*__buffer
);
482 /* Compile a fastmap for the compiled pattern in BUFFER; used to
483 accelerate searches. Return 0 if successful and -2 if was an
485 extern int re_compile_fastmap (struct re_pattern_buffer
*__buffer
);
488 /* Search in the string STRING (with length LENGTH) for the pattern
489 compiled into BUFFER. Start searching at position START, for RANGE
490 characters. Return the starting position of the match, -1 for no
491 match, or -2 for an internal error. Also return register
492 information in REGS (if REGS and BUFFER->no_sub are nonzero). */
493 extern int re_search (struct re_pattern_buffer
*__buffer
, const char *__cstring
,
494 int __length
, int __start
, int __range
,
495 struct re_registers
*__regs
);
498 /* Like `re_search', but search in the concatenation of STRING1 and
499 STRING2. Also, stop searching at index START + STOP. */
500 extern int re_search_2 (struct re_pattern_buffer
*__buffer
,
501 const char *__string1
, int __length1
,
502 const char *__string2
, int __length2
, int __start
,
503 int __range
, struct re_registers
*__regs
, int __stop
);
506 /* Like `re_search', but return how many characters in STRING the regexp
507 in BUFFER matched, starting at position START. */
508 extern int re_match (struct re_pattern_buffer
*__buffer
, const char *__cstring
,
509 int __length
, int __start
, struct re_registers
*__regs
);
512 /* Relates to `re_match' as `re_search_2' relates to `re_search'. */
513 extern int re_match_2 (struct re_pattern_buffer
*__buffer
,
514 const char *__string1
, int __length1
,
515 const char *__string2
, int __length2
, int __start
,
516 struct re_registers
*__regs
, int __stop
);
519 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
520 ENDS. Subsequent matches using BUFFER and REGS will use this memory
521 for recording register information. STARTS and ENDS must be
522 allocated with malloc, and must each be at least `NUM_REGS * sizeof
523 (regoff_t)' bytes long.
525 If NUM_REGS == 0, then subsequent matches should allocate their own
528 Unless this function is called, the first search or match using
529 PATTERN_BUFFER will allocate its own register data, without
530 freeing the old data. */
531 extern void re_set_registers (struct re_pattern_buffer
*__buffer
,
532 struct re_registers
*__regs
,
533 unsigned int __num_regs
,
534 regoff_t
*__starts
, regoff_t
*__ends
);
537 #if defined _REGEX_RE_COMP || (defined _LIBC && defined __USE_BSD)
539 /* 4.2 bsd compatibility. */
540 extern char *re_comp (const char *);
541 extern int re_exec (const char *);
545 /* GCC 2.95 and later have "__restrict"; C99 compilers have
546 "restrict", and "configure" may have defined "restrict". */
548 # if ! (2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__))
549 # if defined restrict || 199901L <= __STDC_VERSION__
550 # define __restrict restrict
556 /* gcc 3.1 and up support the [restrict] syntax. */
557 #ifndef __restrict_arr
558 # if (__GNUC__ > 3 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)) \
560 # define __restrict_arr __restrict
562 # define __restrict_arr
566 /* POSIX compatibility. */
567 extern int regcomp (regex_t
*__restrict __preg
,
568 const char *__restrict __pattern
,
571 extern int regexec (const regex_t
*__restrict __preg
,
572 const char *__restrict __cstring
, size_t __nmatch
,
573 regmatch_t __pmatch
[__restrict_arr
],
576 extern size_t regerror (int __errcode
, const regex_t
*__restrict __preg
,
577 char *__restrict __errbuf
, size_t __errbuf_size
);
579 extern void regfree (regex_t
*__preg
);