1 /* Definitions for data structures and routines for the regular
3 Copyright (C) 1985,1989-93,1995-98,2000,2001,2002,2003,2005,2006,2008,2011
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, see
19 <http://www.gnu.org/licenses/>. */
28 #ifdef HAVE_SYS_TYPES_H
29 #include <sys/types.h>
36 /* Allow the use in C++ code. */
41 /* The following two types have to be signed and unsigned integer type
42 wide enough to hold a value of a pointer. For most ANSI compilers
43 ptrdiff_t and size_t should be likely OK. Still size of these two
44 types is 2 for Microsoft C. Ugh... */
45 typedef long int s_reg_t;
46 typedef unsigned long int active_reg_t;
48 /* The following bits are used to determine the regexp syntax we
49 recognize. The set/not-set meanings are chosen so that Emacs syntax
50 remains the value 0. The bits are given in alphabetical order, and
51 the definitions shifted by one from the previous bit; thus, when we
52 add or remove a bit, only one other definition need change. */
53 typedef unsigned long int reg_syntax_t;
56 /* If this bit is not set, then \ inside a bracket expression is literal.
57 If set, then such a \ quotes the following character. */
58 # define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1)
60 /* If this bit is not set, then + and ? are operators, and \+ and \? are
62 If set, then \+ and \? are operators and + and ? are literals. */
63 # define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1)
65 /* If this bit is set, then character classes are supported. They are:
66 [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:],
67 [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:].
68 If not set, then character classes are not supported. */
69 # define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1)
71 /* If this bit is set, then ^ and $ are always anchors (outside bracket
72 expressions, of course).
73 If this bit is not set, then it depends:
74 ^ is an anchor if it is at the beginning of a regular
75 expression or after an open-group or an alternation operator;
76 $ is an anchor if it is at the end of a regular expression, or
77 before a close-group or an alternation operator.
79 This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because
80 POSIX draft 11.2 says that * etc. in leading positions is undefined.
81 We already implemented a previous draft which made those constructs
82 invalid, though, so we haven't changed the code back. */
83 # define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1)
85 /* If this bit is set, then special characters are always special
86 regardless of where they are in the pattern.
87 If this bit is not set, then special characters are special only in
88 some contexts; otherwise they are ordinary. Specifically,
89 * + ? and intervals are only special when not after the beginning,
90 open-group, or alternation operator. */
91 # define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1)
93 /* If this bit is set, then *, +, ?, and { cannot be first in an re or
94 immediately after an alternation or begin-group operator. */
95 # define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1)
97 /* If this bit is set, then . matches newline.
98 If not set, then it doesn't. */
99 # define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1)
101 /* If this bit is set, then . doesn't match NUL.
102 If not set, then it does. */
103 # define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1)
105 /* If this bit is set, nonmatching lists [^...] do not match newline.
106 If not set, they do. */
107 # define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1)
109 /* If this bit is set, either \{...\} or {...} defines an
110 interval, depending on RE_NO_BK_BRACES.
111 If not set, \{, \}, {, and } are literals. */
112 # define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1)
114 /* If this bit is set, +, ? and | aren't recognized as operators.
115 If not set, they are. */
116 # define RE_LIMITED_OPS (RE_INTERVALS << 1)
118 /* If this bit is set, newline is an alternation operator.
119 If not set, newline is literal. */
120 # define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1)
122 /* If this bit is set, then `{...}' defines an interval, and \{ and \}
124 If not set, then `\{...\}' defines an interval. */
125 # define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1)
127 /* If this bit is set, (...) defines a group, and \( and \) are literals.
128 If not set, \(...\) defines a group, and ( and ) are literals. */
129 # define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1)
131 /* If this bit is set, then \<digit> matches <digit>.
132 If not set, then \<digit> is a back-reference. */
133 # define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1)
135 /* If this bit is set, then | is an alternation operator, and \| is literal.
136 If not set, then \| is an alternation operator, and | is literal. */
137 # define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1)
139 /* If this bit is set, then an ending range point collating higher
140 than the starting range point, as in [z-a], is invalid.
141 If not set, then when ending range point collates higher than the
142 starting range point, the range is ignored. */
143 # define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1)
145 /* If this bit is set, then an unmatched ) is ordinary.
146 If not set, then an unmatched ) is invalid. */
147 # define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1)
149 /* If this bit is set, succeed as soon as we match the whole pattern,
150 without further backtracking. */
151 # define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1)
153 /* If this bit is set, do not process the GNU regex operators.
154 If not set, then the GNU regex operators are recognized. */
155 # define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1)
157 /* If this bit is set, turn on internal regex debugging.
158 If not set, and debugging was on, turn it off.
159 This only works if regex.c is compiled -DDEBUG.
160 We define this bit always, so that all that's needed to turn on
161 debugging is to recompile regex.c; the calling code can always have
162 this bit set, and it won't affect anything in the normal case. */
163 # define RE_DEBUG (RE_NO_GNU_OPS << 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_DEBUG << 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 \
207 | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS)
209 #define RE_SYNTAX_GNU_AWK \
210 ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
211 | RE_INVALID_INTERVAL_ORD) \
212 & ~(RE_DOT_NOT_NULL | RE_CONTEXT_INDEP_OPS \
213 | RE_CONTEXT_INVALID_OPS ))
215 #define RE_SYNTAX_POSIX_AWK \
216 (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
217 | RE_INTERVALS | RE_NO_GNU_OPS \
218 | RE_INVALID_INTERVAL_ORD)
220 #define RE_SYNTAX_GREP \
221 (RE_BK_PLUS_QM | RE_CHAR_CLASSES \
222 | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \
225 #define RE_SYNTAX_EGREP \
226 (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \
227 | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \
228 | RE_NEWLINE_ALT | RE_NO_BK_PARENS \
231 #define RE_SYNTAX_POSIX_EGREP \
232 (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES \
233 | RE_INVALID_INTERVAL_ORD)
235 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */
236 #define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
238 #define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
240 /* Syntax bits common to both basic and extended POSIX regex syntax. */
241 #define _RE_SYNTAX_POSIX_COMMON \
242 (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \
243 | RE_INTERVALS | RE_NO_EMPTY_RANGES)
245 #define RE_SYNTAX_POSIX_BASIC \
246 (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM | RE_CONTEXT_INVALID_DUP)
248 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
249 RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this
250 isn't minimal, since other operators, such as \`, aren't disabled. */
251 #define RE_SYNTAX_POSIX_MINIMAL_BASIC \
252 (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
254 #define RE_SYNTAX_POSIX_EXTENDED \
255 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
256 | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \
257 | RE_NO_BK_PARENS | RE_NO_BK_VBAR \
258 | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD)
260 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is
261 removed and RE_NO_BK_REFS is added. */
262 #define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \
263 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
264 | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \
265 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
266 | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD)
267 /* [[[end syntaxes]]] */
269 /* Maximum number of duplicates an interval can allow. Some systems
270 (erroneously) define this in other header files, but we want our
271 value, so remove any previous define. */
275 /* If sizeof(int) == 2, then ((1 << 15) - 1) overflows. */
276 # define RE_DUP_MAX (0x7fff)
280 /* POSIX `cflags' bits (i.e., information for `regcomp'). */
282 /* If this bit is set, then use extended regular expression syntax.
283 If not set, then use basic regular expression syntax. */
284 #define REG_EXTENDED 1
286 /* If this bit is set, then ignore case when matching.
287 If not set, then case is significant. */
288 #define REG_ICASE (REG_EXTENDED << 1)
290 /* If this bit is set, then anchors do not match at newline
291 characters in the string.
292 If not set, then anchors do match at newlines. */
293 #define REG_NEWLINE (REG_ICASE << 1)
295 /* If this bit is set, then report only success or fail in regexec.
296 If not set, then returns differ between not matching and errors. */
297 #define REG_NOSUB (REG_NEWLINE << 1)
300 /* POSIX `eflags' bits (i.e., information for regexec). */
302 /* If this bit is set, then the beginning-of-line operator doesn't match
303 the beginning of the string (presumably because it's not the
304 beginning of a line).
305 If not set, then the beginning-of-line operator does match the
306 beginning of the string. */
309 /* Like REG_NOTBOL, except for the end-of-line. */
310 #define REG_NOTEOL (1 << 1)
312 /* Use PMATCH[0] to delimit the start and end of the search in the
314 #define REG_STARTEND (1 << 2)
317 /* If any error codes are removed, changed, or added, update the
318 `re_error_msg' table in regex.c. */
321 #if defined _XOPEN_SOURCE || defined __USE_XOPEN2K
322 REG_ENOSYS = -1, /* This will never happen for this implementation. */
325 REG_NOERROR = 0, /* Success. */
326 REG_NOMATCH, /* Didn't find a match (for regexec). */
328 /* POSIX regcomp return error codes. (In the order listed in the
330 REG_BADPAT, /* Invalid pattern. */
331 REG_ECOLLATE, /* Inalid collating element. */
332 REG_ECTYPE, /* Invalid character class name. */
333 REG_EESCAPE, /* Trailing backslash. */
334 REG_ESUBREG, /* Invalid back reference. */
335 REG_EBRACK, /* Unmatched left bracket. */
336 REG_EPAREN, /* Parenthesis imbalance. */
337 REG_EBRACE, /* Unmatched \{. */
338 REG_BADBR, /* Invalid contents of \{\}. */
339 REG_ERANGE, /* Invalid range end. */
340 REG_ESPACE, /* Ran out of memory. */
341 REG_BADRPT, /* No preceding re for repetition op. */
343 /* Error codes we've added. */
344 REG_EEND, /* Premature end. */
345 REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */
346 REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */
349 /* This data structure represents a compiled pattern. Before calling
350 the pattern compiler, the fields `buffer', `allocated', `fastmap',
351 and `translate' can be set. After the pattern has been compiled,
352 the fields `re_nsub', `not_bol' and `not_eol' are available. All
353 other fields are private to the regex routines. */
355 #ifndef RE_TRANSLATE_TYPE
356 # define __RE_TRANSLATE_TYPE unsigned char *
358 # define RE_TRANSLATE_TYPE __RE_TRANSLATE_TYPE
363 # define __REPB_PREFIX(name) name
365 # define __REPB_PREFIX(name) __##name
368 struct re_pattern_buffer
370 /* Space that holds the compiled pattern. It is declared as
371 `unsigned char *' because its elements are sometimes used as
373 unsigned char *__REPB_PREFIX(buffer);
375 /* Number of bytes to which `buffer' points. */
376 unsigned long int __REPB_PREFIX(allocated);
378 /* Number of bytes actually used in `buffer'. */
379 unsigned long int __REPB_PREFIX(used);
381 /* Syntax setting with which the pattern was compiled. */
382 reg_syntax_t __REPB_PREFIX(syntax);
384 /* Pointer to a fastmap, if any, otherwise zero. re_search uses the
385 fastmap, if there is one, to skip over impossible starting points
387 char *__REPB_PREFIX(fastmap);
389 /* Either a translate table to apply to all characters before
390 comparing them, or zero for no translation. The translation is
391 applied to a pattern when it is compiled and to a string when it
393 __RE_TRANSLATE_TYPE __REPB_PREFIX(translate);
395 /* Number of subexpressions found by the compiler. */
398 /* Zero if this pattern cannot match the empty string, one else.
399 Well, in truth it's used only in `re_search_2', to see whether or
400 not we should use the fastmap, so we don't set this absolutely
401 perfectly; see `re_compile_fastmap' (the `duplicate' case). */
402 unsigned __REPB_PREFIX(can_be_null) : 1;
404 /* If REGS_UNALLOCATED, allocate space in the `regs' structure
405 for `max (RE_NREGS, re_nsub + 1)' groups.
406 If REGS_REALLOCATE, reallocate space if necessary.
407 If REGS_FIXED, use what's there. */
409 # define REGS_UNALLOCATED 0
410 # define REGS_REALLOCATE 1
411 # define REGS_FIXED 2
413 unsigned __REPB_PREFIX(regs_allocated) : 2;
415 /* Set to zero when `regex_compile' compiles a pattern; set to one
416 by `re_compile_fastmap' if it updates the fastmap. */
417 unsigned __REPB_PREFIX(fastmap_accurate) : 1;
419 /* If set, `re_match_2' does not return information about
421 unsigned __REPB_PREFIX(no_sub) : 1;
423 /* If set, a beginning-of-line anchor doesn't match at the beginning
425 unsigned __REPB_PREFIX(not_bol) : 1;
427 /* Similarly for an end-of-line anchor. */
428 unsigned __REPB_PREFIX(not_eol) : 1;
430 /* If true, an anchor at a newline matches. */
431 unsigned __REPB_PREFIX(newline_anchor) : 1;
434 typedef struct re_pattern_buffer regex_t;
436 /* Type for byte offsets within the string. POSIX mandates this. */
437 typedef int regoff_t;
441 /* This is the structure we store register match data in. See
442 regex.texinfo for a full description of what registers match. */
451 /* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
452 `re_match_2' returns information about at least this many registers
453 the first time a `regs' structure is passed. */
460 /* POSIX specification for registers. Aside from the different names than
461 `re_registers', POSIX uses an array of structures, instead of a
462 structure of arrays. */
465 regoff_t rm_so; /* Byte offset from string's start to substring's start. */
466 regoff_t rm_eo; /* Byte offset from string's start to substring's end. */
469 /* Declarations for routines. */
472 /* Sets the current default syntax to SYNTAX, and return the old syntax.
473 You can also simply assign to the `re_syntax_options' variable. */
474 extern reg_syntax_t re_set_syntax (reg_syntax_t __syntax);
476 /* Compile the regular expression PATTERN, with length LENGTH
477 and syntax given by the global `re_syntax_options', into the buffer
478 BUFFER. Return NULL if successful, and an error string if not.
480 To free the allocated storage, you must call `regfree' on BUFFER.
481 Note that the translate table must either have been initialised by
482 `regcomp', with a malloc'ed value, or set to NULL before calling
484 extern const char *re_compile_pattern (const char *__pattern, size_t __length,
485 struct re_pattern_buffer *__buffer);
488 /* Compile a fastmap for the compiled pattern in BUFFER; used to
489 accelerate searches. Return 0 if successful and -2 if was an
491 extern int re_compile_fastmap (struct re_pattern_buffer *__buffer);
494 /* Search in the string STRING (with length LENGTH) for the pattern
495 compiled into BUFFER. Start searching at position START, for RANGE
496 characters. Return the starting position of the match, -1 for no
497 match, or -2 for an internal error. Also return register
498 information in REGS (if REGS and BUFFER->no_sub are nonzero). */
499 extern int re_search (struct re_pattern_buffer *__buffer, const char *__cstring,
500 int __length, int __start, int __range,
501 struct re_registers *__regs);
504 /* Like `re_search', but search in the concatenation of STRING1 and
505 STRING2. Also, stop searching at index START + STOP. */
506 extern int re_search_2 (struct re_pattern_buffer *__buffer,
507 const char *__string1, int __length1,
508 const char *__string2, int __length2, int __start,
509 int __range, struct re_registers *__regs, int __stop);
512 /* Like `re_search', but return how many characters in STRING the regexp
513 in BUFFER matched, starting at position START. */
514 extern int re_match (struct re_pattern_buffer *__buffer, const char *__cstring,
515 int __length, int __start, struct re_registers *__regs);
518 /* Relates to `re_match' as `re_search_2' relates to `re_search'. */
519 extern int re_match_2 (struct re_pattern_buffer *__buffer,
520 const char *__string1, int __length1,
521 const char *__string2, int __length2, int __start,
522 struct re_registers *__regs, int __stop);
525 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
526 ENDS. Subsequent matches using BUFFER and REGS will use this memory
527 for recording register information. STARTS and ENDS must be
528 allocated with malloc, and must each be at least `NUM_REGS * sizeof
529 (regoff_t)' bytes long.
531 If NUM_REGS == 0, then subsequent matches should allocate their own
534 Unless this function is called, the first search or match using
535 PATTERN_BUFFER will allocate its own register data, without
536 freeing the old data. */
537 extern void re_set_registers (struct re_pattern_buffer *__buffer,
538 struct re_registers *__regs,
539 unsigned int __num_regs,
540 regoff_t *__starts, regoff_t *__ends);
543 #if defined _REGEX_RE_COMP || (defined _LIBC && defined __USE_BSD)
545 /* 4.2 bsd compatibility. */
546 extern char *re_comp (const char *);
547 extern int re_exec (const char *);
551 /* GCC 2.95 and later have "__restrict"; C99 compilers have
552 "restrict", and "configure" may have defined "restrict". */
554 # if ! (2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__))
555 # if defined restrict || 199901L <= __STDC_VERSION__
556 # define __restrict restrict
562 /* gcc 3.1 and up support the [restrict] syntax. */
563 #ifndef __restrict_arr
564 # if (__GNUC__ > 3 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)) \
566 # define __restrict_arr __restrict
568 # define __restrict_arr
572 /* POSIX compatibility. */
573 extern int regcomp (regex_t *__restrict __preg,
574 const char *__restrict __pattern,
577 extern int regexec (const regex_t *__restrict __preg,
578 const char *__restrict __cstring, size_t __nmatch,
579 regmatch_t __pmatch[__restrict_arr],
582 extern size_t regerror (int __errcode, const regex_t *__restrict __preg,
583 char *__restrict __errbuf, size_t __errbuf_size);
585 extern void regfree (regex_t *__preg);