*** empty log message ***
[emacs.git] / src / regex.h
blobef4284cdce2894645bd705d1a41314fb21cf9d70
1 /* Definitions for data structures and routines for the regular
2 expression library, version 0.12.
4 Copyright (C) 1985,89,90,91,92,93,95,2000 Free Software Foundation, Inc.
6 This program is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2, or (at your option)
9 any later version.
11 This program is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with this program; if not, write to the Free Software
18 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
19 USA. */
21 #ifndef _REGEX_H
22 #define _REGEX_H 1
24 /* Allow the use in C++ code. */
25 #ifdef __cplusplus
26 extern "C" {
27 #endif
29 /* POSIX says that <sys/types.h> must be included (by the caller) before
30 <regex.h>. */
32 #if !defined _POSIX_C_SOURCE && !defined _POSIX_SOURCE && defined VMS
33 /* VMS doesn't have `size_t' in <sys/types.h>, even though POSIX says it
34 should be there. */
35 # include <stddef.h>
36 #endif
38 /* The following bits are used to determine the regexp syntax we
39 recognize. The set/not-set meanings where historically chosen so
40 that Emacs syntax had the value 0.
41 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, then *?, +? and ?? match non greedily. */
148 #define RE_FRUGAL (RE_NO_GNU_OPS << 1)
150 /* If this bit is set, then (?:...) is treated as a shy group. */
151 #define RE_SHY_GROUPS (RE_FRUGAL << 1)
153 /* If this bit is set, ^ and $ only match at beg/end of buffer. */
154 #define RE_NO_NEWLINE_ANCHOR (RE_SHY_GROUPS << 1)
156 /* If this bit is set, turn on internal regex debugging.
157 If not set, and debugging was on, turn it off.
158 This only works if regex.c is compiled -DDEBUG.
159 We define this bit always, so that all that's needed to turn on
160 debugging is to recompile regex.c; the calling code can always have
161 this bit set, and it won't affect anything in the normal case. */
162 #define RE_DEBUG (RE_NO_NEWLINE_ANCHOR << 1)
164 /* This global variable defines the particular regexp syntax to use (for
165 some interfaces). When a regexp is compiled, the syntax used is
166 stored in the pattern buffer, so changing this does not affect
167 already-compiled regexps. */
168 extern reg_syntax_t re_syntax_options;
170 #ifdef emacs
171 /* In Emacs, this is the string or buffer in which we
172 are matching. It is used for looking up syntax properties. */
173 extern Lisp_Object re_match_object;
174 #endif
177 /* Define combinations of the above bits for the standard possibilities.
178 (The [[[ comments delimit what gets put into the Texinfo file, so
179 don't delete them!) */
180 /* [[[begin syntaxes]]] */
181 #define RE_SYNTAX_EMACS \
182 (RE_CHAR_CLASSES | RE_INTERVALS | RE_SHY_GROUPS | RE_FRUGAL)
184 #define RE_SYNTAX_AWK \
185 (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \
186 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
187 | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \
188 | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \
189 | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS)
191 #define RE_SYNTAX_GNU_AWK \
192 ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DEBUG) \
193 & ~(RE_DOT_NOT_NULL | RE_INTERVALS | RE_CONTEXT_INDEP_OPS))
195 #define RE_SYNTAX_POSIX_AWK \
196 (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
197 | RE_INTERVALS | RE_NO_GNU_OPS)
199 #define RE_SYNTAX_GREP \
200 (RE_BK_PLUS_QM | RE_CHAR_CLASSES \
201 | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \
202 | RE_NEWLINE_ALT)
204 #define RE_SYNTAX_EGREP \
205 (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \
206 | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \
207 | RE_NEWLINE_ALT | RE_NO_BK_PARENS \
208 | RE_NO_BK_VBAR)
210 #define RE_SYNTAX_POSIX_EGREP \
211 (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES)
213 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */
214 #define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
216 #define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
218 /* Syntax bits common to both basic and extended POSIX regex syntax. */
219 #define _RE_SYNTAX_POSIX_COMMON \
220 (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \
221 | RE_INTERVALS | RE_NO_EMPTY_RANGES)
223 #define RE_SYNTAX_POSIX_BASIC \
224 (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM)
226 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
227 RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this
228 isn't minimal, since other operators, such as \`, aren't disabled. */
229 #define RE_SYNTAX_POSIX_MINIMAL_BASIC \
230 (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
232 #define RE_SYNTAX_POSIX_EXTENDED \
233 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
234 | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \
235 | RE_NO_BK_PARENS | RE_NO_BK_VBAR \
236 | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD)
238 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is
239 removed and RE_NO_BK_REFS is added. */
240 #define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \
241 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
242 | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \
243 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
244 | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD)
245 /* [[[end syntaxes]]] */
247 /* Maximum number of duplicates an interval can allow. Some systems
248 (erroneously) define this in other header files, but we want our
249 value, so remove any previous define. */
250 #ifdef RE_DUP_MAX
251 # undef RE_DUP_MAX
252 #endif
253 /* If sizeof(int) == 2, then ((1 << 15) - 1) overflows. */
254 #define RE_DUP_MAX (0x7fff)
257 /* POSIX `cflags' bits (i.e., information for `regcomp'). */
259 /* If this bit is set, then use extended regular expression syntax.
260 If not set, then use basic regular expression syntax. */
261 #define REG_EXTENDED 1
263 /* If this bit is set, then ignore case when matching.
264 If not set, then case is significant. */
265 #define REG_ICASE (REG_EXTENDED << 1)
267 /* If this bit is set, then anchors do not match at newline
268 characters in the string.
269 If not set, then anchors do match at newlines. */
270 #define REG_NEWLINE (REG_ICASE << 1)
272 /* If this bit is set, then report only success or fail in regexec.
273 If not set, then returns differ between not matching and errors. */
274 #define REG_NOSUB (REG_NEWLINE << 1)
277 /* POSIX `eflags' bits (i.e., information for regexec). */
279 /* If this bit is set, then the beginning-of-line operator doesn't match
280 the beginning of the string (presumably because it's not the
281 beginning of a line).
282 If not set, then the beginning-of-line operator does match the
283 beginning of the string. */
284 #define REG_NOTBOL 1
286 /* Like REG_NOTBOL, except for the end-of-line. */
287 #define REG_NOTEOL (1 << 1)
290 /* If any error codes are removed, changed, or added, update the
291 `re_error_msg' table in regex.c. */
292 typedef enum
294 #ifdef _XOPEN_SOURCE
295 REG_ENOSYS = -1, /* This will never happen for this implementation. */
296 #endif
298 REG_NOERROR = 0, /* Success. */
299 REG_NOMATCH, /* Didn't find a match (for regexec). */
301 /* POSIX regcomp return error codes. (In the order listed in the
302 standard.) */
303 REG_BADPAT, /* Invalid pattern. */
304 REG_ECOLLATE, /* Not implemented. */
305 REG_ECTYPE, /* Invalid character class name. */
306 REG_EESCAPE, /* Trailing backslash. */
307 REG_ESUBREG, /* Invalid back reference. */
308 REG_EBRACK, /* Unmatched left bracket. */
309 REG_EPAREN, /* Parenthesis imbalance. */
310 REG_EBRACE, /* Unmatched \{. */
311 REG_BADBR, /* Invalid contents of \{\}. */
312 REG_ERANGE, /* Invalid range end. */
313 REG_ESPACE, /* Ran out of memory. */
314 REG_BADRPT, /* No preceding re for repetition op. */
316 /* Error codes we've added. */
317 REG_EEND, /* Premature end. */
318 REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */
319 REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */
320 } reg_errcode_t;
322 /* This data structure represents a compiled pattern. Before calling
323 the pattern compiler, the fields `buffer', `allocated', `fastmap',
324 `translate', and `no_sub' can be set. After the pattern has been
325 compiled, the `re_nsub' field is available. All other fields are
326 private to the regex routines. */
328 #ifndef RE_TRANSLATE_TYPE
329 # define RE_TRANSLATE_TYPE char *
330 #endif
332 struct re_pattern_buffer
334 /* [[[begin pattern_buffer]]] */
335 /* Space that holds the compiled pattern. It is declared as
336 `unsigned char *' because its elements are
337 sometimes used as array indexes. */
338 unsigned char *buffer;
340 /* Number of bytes to which `buffer' points. */
341 size_t allocated;
343 /* Number of bytes actually used in `buffer'. */
344 size_t used;
346 /* Syntax setting with which the pattern was compiled. */
347 reg_syntax_t syntax;
349 /* Pointer to a fastmap, if any, otherwise zero. re_search uses
350 the fastmap, if there is one, to skip over impossible
351 starting points for matches. */
352 char *fastmap;
354 /* Either a translate table to apply to all characters before
355 comparing them, or zero for no translation. The translation
356 is applied to a pattern when it is compiled and to a string
357 when it is matched. */
358 RE_TRANSLATE_TYPE translate;
360 /* Number of subexpressions found by the compiler. */
361 size_t re_nsub;
363 /* Zero if this pattern cannot match the empty string, one else.
364 Well, in truth it's used only in `re_search_2', to see
365 whether or not we should use the fastmap, so we don't set
366 this absolutely perfectly; see `re_compile_fastmap'. */
367 unsigned can_be_null : 1;
369 /* If REGS_UNALLOCATED, allocate space in the `regs' structure
370 for `max (RE_NREGS, re_nsub + 1)' groups.
371 If REGS_REALLOCATE, reallocate space if necessary.
372 If REGS_FIXED, use what's there. */
373 #define REGS_UNALLOCATED 0
374 #define REGS_REALLOCATE 1
375 #define REGS_FIXED 2
376 unsigned regs_allocated : 2;
378 /* Set to zero when `regex_compile' compiles a pattern; set to one
379 by `re_compile_fastmap' if it updates the fastmap. */
380 unsigned fastmap_accurate : 1;
382 /* If set, `re_match_2' does not return information about
383 subexpressions. */
384 unsigned no_sub : 1;
386 /* If set, a beginning-of-line anchor doesn't match at the
387 beginning of the string. */
388 unsigned not_bol : 1;
390 /* Similarly for an end-of-line anchor. */
391 unsigned not_eol : 1;
393 #ifdef emacs
394 /* If true, multi-byte form in the `buffer' should be recognized as a
395 multibyte character. */
396 unsigned multibyte : 1;
397 #endif
399 /* [[[end pattern_buffer]]] */
402 typedef struct re_pattern_buffer regex_t;
404 /* Type for byte offsets within the string. POSIX mandates this. */
405 typedef int regoff_t;
408 /* This is the structure we store register match data in. See
409 regex.texinfo for a full description of what registers match. */
410 struct re_registers
412 unsigned num_regs;
413 regoff_t *start;
414 regoff_t *end;
418 /* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
419 `re_match_2' returns information about at least this many registers
420 the first time a `regs' structure is passed. */
421 #ifndef RE_NREGS
422 # define RE_NREGS 30
423 #endif
426 /* POSIX specification for registers. Aside from the different names than
427 `re_registers', POSIX uses an array of structures, instead of a
428 structure of arrays. */
429 typedef struct
431 regoff_t rm_so; /* Byte offset from string's start to substring's start. */
432 regoff_t rm_eo; /* Byte offset from string's start to substring's end. */
433 } regmatch_t;
435 /* Declarations for routines. */
437 /* To avoid duplicating every routine declaration -- once with a
438 prototype (if we are ANSI), and once without (if we aren't) -- we
439 use the following macro to declare argument types. This
440 unfortunately clutters up the declarations a bit, but I think it's
441 worth it. */
443 #if defined __STDC__ || defined PROTOTYPES
445 # define _RE_ARGS(args) args
447 #else /* not __STDC__ || PROTOTYPES */
449 # define _RE_ARGS(args) ()
451 #endif /* not __STDC__ || PROTOTYPES */
453 /* Sets the current default syntax to SYNTAX, and return the old syntax.
454 You can also simply assign to the `re_syntax_options' variable. */
455 extern reg_syntax_t re_set_syntax _RE_ARGS ((reg_syntax_t syntax));
457 /* Compile the regular expression PATTERN, with length LENGTH
458 and syntax given by the global `re_syntax_options', into the buffer
459 BUFFER. Return NULL if successful, and an error string if not. */
460 extern const char *re_compile_pattern
461 _RE_ARGS ((const char *pattern, size_t length,
462 struct re_pattern_buffer *buffer));
465 /* Compile a fastmap for the compiled pattern in BUFFER; used to
466 accelerate searches. Return 0 if successful and -2 if was an
467 internal error. */
468 extern int re_compile_fastmap _RE_ARGS ((struct re_pattern_buffer *buffer));
471 /* Search in the string STRING (with length LENGTH) for the pattern
472 compiled into BUFFER. Start searching at position START, for RANGE
473 characters. Return the starting position of the match, -1 for no
474 match, or -2 for an internal error. Also return register
475 information in REGS (if REGS and BUFFER->no_sub are nonzero). */
476 extern int re_search
477 _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string,
478 int length, int start, int range, struct re_registers *regs));
481 /* Like `re_search', but search in the concatenation of STRING1 and
482 STRING2. Also, stop searching at index START + STOP. */
483 extern int re_search_2
484 _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1,
485 int length1, const char *string2, int length2,
486 int start, int range, struct re_registers *regs, int stop));
489 /* Like `re_search', but return how many characters in STRING the regexp
490 in BUFFER matched, starting at position START. */
491 extern int re_match
492 _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string,
493 int length, int start, struct re_registers *regs));
496 /* Relates to `re_match' as `re_search_2' relates to `re_search'. */
497 extern int re_match_2
498 _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1,
499 int length1, const char *string2, int length2,
500 int start, struct re_registers *regs, int stop));
503 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
504 ENDS. Subsequent matches using BUFFER and REGS will use this memory
505 for recording register information. STARTS and ENDS must be
506 allocated with malloc, and must each be at least `NUM_REGS * sizeof
507 (regoff_t)' bytes long.
509 If NUM_REGS == 0, then subsequent matches should allocate their own
510 register data.
512 Unless this function is called, the first search or match using
513 PATTERN_BUFFER will allocate its own register data, without
514 freeing the old data. */
515 extern void re_set_registers
516 _RE_ARGS ((struct re_pattern_buffer *buffer, struct re_registers *regs,
517 unsigned num_regs, regoff_t *starts, regoff_t *ends));
519 #if defined _REGEX_RE_COMP || defined _LIBC
520 # ifndef _CRAY
521 /* 4.2 bsd compatibility. */
522 extern char *re_comp _RE_ARGS ((const char *));
523 extern int re_exec _RE_ARGS ((const char *));
524 # endif
525 #endif
527 /* GCC 2.95 and later have "__restrict"; C99 compilers have
528 "restrict", and "configure" may have defined "restrict". */
529 #ifndef __restrict
530 # if ! (2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__))
531 # if defined restrict || 199901L <= __STDC_VERSION__
532 # define __restrict restrict
533 # else
534 # define __restrict
535 # endif
536 # endif
537 #endif
538 /* For now unconditionally define __restrict_arr to expand to nothing.
539 Ideally we would have a test for the compiler which allows defining
540 it to restrict. */
541 #define __restrict_arr
543 /* POSIX compatibility. */
544 extern int regcomp _RE_ARGS ((regex_t *__restrict __preg,
545 const char *__restrict __pattern,
546 int __cflags));
548 extern int regexec _RE_ARGS ((const regex_t *__restrict __preg,
549 const char *__restrict __string, size_t __nmatch,
550 regmatch_t __pmatch[__restrict_arr],
551 int __eflags));
553 extern size_t regerror _RE_ARGS ((int __errcode, const regex_t *__preg,
554 char *__errbuf, size_t __errbuf_size));
556 extern void regfree _RE_ARGS ((regex_t *__preg));
559 #ifdef __cplusplus
561 #endif /* C++ */
563 #endif /* regex.h */
566 Local variables:
567 make-backup-files: t
568 version-control: t
569 trim-versions-without-asking: nil
570 End: