1 /* Definitions for data structures and routines for the regular
2 expression library, version 0.12.
3 Copyright (C) 1985,89,90,91,92,93,95,96,97 Free Software Foundation, Inc.
5 This file is part of the GNU C Library. Its master source is NOT part of
6 the C library, however. The master source lives in /gd/gnu/lib.
8 The GNU C Library is free software; you can redistribute it and/or
9 modify it under the terms of the GNU Library General Public License as
10 published by the Free Software Foundation; either version 2 of the
11 License, or (at your option) any later version.
13 The GNU C Library is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 Library General Public License for more details.
18 You should have received a copy of the GNU Library General Public
19 License along with the GNU C Library; see the file COPYING.LIB. If not,
20 write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
21 Boston, MA 02111-1307, USA. */
26 /* Allow the use in C++ code. */
31 /* POSIX says that <sys/types.h> must be included (by the caller) before
34 #if !defined _POSIX_C_SOURCE && !defined _POSIX_SOURCE && defined VMS
35 /* VMS doesn't have `size_t' in <sys/types.h>, even though POSIX says it
40 /* The following two types have to be signed and unsigned integer type
41 wide enough to hold a value of a pointer. For most ANSI compilers
42 ptrdiff_t and size_t should be likely OK. Still size of these two
43 types is 2 for Microsoft C. Ugh... */
44 typedef long int s_reg_t
;
45 typedef unsigned long int active_reg_t
;
47 /* The following bits are used to determine the regexp syntax we
48 recognize. The set/not-set meanings are chosen so that Emacs syntax
49 remains the value 0. The bits are given in alphabetical order, and
50 the definitions shifted by one from the previous bit; thus, when we
51 add or remove a bit, only one other definition need change. */
52 typedef unsigned long int reg_syntax_t
;
54 /* If this bit is not set, then \ inside a bracket expression is literal.
55 If set, then such a \ quotes the following character. */
56 #define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1)
58 /* If this bit is not set, then + and ? are operators, and \+ and \? are
60 If set, then \+ and \? are operators and + and ? are literals. */
61 #define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1)
63 /* If this bit is set, then character classes are supported. They are:
64 [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:],
65 [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:].
66 If not set, then character classes are not supported. */
67 #define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1)
69 /* If this bit is set, then ^ and $ are always anchors (outside bracket
70 expressions, of course).
71 If this bit is not set, then it depends:
72 ^ is an anchor if it is at the beginning of a regular
73 expression or after an open-group or an alternation operator;
74 $ is an anchor if it is at the end of a regular expression, or
75 before a close-group or an alternation operator.
77 This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because
78 POSIX draft 11.2 says that * etc. in leading positions is undefined.
79 We already implemented a previous draft which made those constructs
80 invalid, though, so we haven't changed the code back. */
81 #define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1)
83 /* If this bit is set, then special characters are always special
84 regardless of where they are in the pattern.
85 If this bit is not set, then special characters are special only in
86 some contexts; otherwise they are ordinary. Specifically,
87 * + ? and intervals are only special when not after the beginning,
88 open-group, or alternation operator. */
89 #define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1)
91 /* If this bit is set, then *, +, ?, and { cannot be first in an re or
92 immediately after an alternation or begin-group operator. */
93 #define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1)
95 /* If this bit is set, then . matches newline.
96 If not set, then it doesn't. */
97 #define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1)
99 /* If this bit is set, then . doesn't match NUL.
100 If not set, then it does. */
101 #define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1)
103 /* If this bit is set, nonmatching lists [^...] do not match newline.
104 If not set, they do. */
105 #define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1)
107 /* If this bit is set, either \{...\} or {...} defines an
108 interval, depending on RE_NO_BK_BRACES.
109 If not set, \{, \}, {, and } are literals. */
110 #define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1)
112 /* If this bit is set, +, ? and | aren't recognized as operators.
113 If not set, they are. */
114 #define RE_LIMITED_OPS (RE_INTERVALS << 1)
116 /* If this bit is set, newline is an alternation operator.
117 If not set, newline is literal. */
118 #define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1)
120 /* If this bit is set, then `{...}' defines an interval, and \{ and \}
122 If not set, then `\{...\}' defines an interval. */
123 #define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1)
125 /* If this bit is set, (...) defines a group, and \( and \) are literals.
126 If not set, \(...\) defines a group, and ( and ) are literals. */
127 #define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1)
129 /* If this bit is set, then \<digit> matches <digit>.
130 If not set, then \<digit> is a back-reference. */
131 #define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1)
133 /* If this bit is set, then | is an alternation operator, and \| is literal.
134 If not set, then \| is an alternation operator, and | is literal. */
135 #define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1)
137 /* If this bit is set, then an ending range point collating higher
138 than the starting range point, as in [z-a], is invalid.
139 If not set, then when ending range point collates higher than the
140 starting range point, the range is ignored. */
141 #define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1)
143 /* If this bit is set, then an unmatched ) is ordinary.
144 If not set, then an unmatched ) is invalid. */
145 #define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1)
147 /* If this bit is set, succeed as soon as we match the whole pattern,
148 without further backtracking. */
149 #define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1)
151 /* If this bit is set, do not process the GNU regex operators.
152 If not set, then the GNU regex operators are recognized. */
153 #define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1)
155 /* If this bit is set, turn on internal regex debugging.
156 If not set, and debugging was on, turn it off.
157 This only works if regex.c is compiled -DDEBUG.
158 We define this bit always, so that all that's needed to turn on
159 debugging is to recompile regex.c; the calling code can always have
160 this bit set, and it won't affect anything in the normal case. */
161 #define RE_DEBUG (RE_NO_GNU_OPS << 1)
163 /* This global variable defines the particular regexp syntax to use (for
164 some interfaces). When a regexp is compiled, the syntax used is
165 stored in the pattern buffer, so changing this does not affect
166 already-compiled regexps. */
167 extern reg_syntax_t re_syntax_options
;
169 /* Define combinations of the above bits for the standard possibilities.
170 (The [[[ comments delimit what gets put into the Texinfo file, so
171 don't delete them!) */
172 /* [[[begin syntaxes]]] */
173 #define RE_SYNTAX_EMACS 0
175 #define RE_SYNTAX_AWK \
176 (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \
177 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
178 | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \
179 | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \
180 | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS)
182 #define RE_SYNTAX_GNU_AWK \
183 ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DEBUG) \
184 & ~(RE_DOT_NOT_NULL | RE_INTERVALS | RE_CONTEXT_INDEP_OPS))
186 #define RE_SYNTAX_POSIX_AWK \
187 (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
188 | RE_INTERVALS | RE_NO_GNU_OPS)
190 #define RE_SYNTAX_GREP \
191 (RE_BK_PLUS_QM | RE_CHAR_CLASSES \
192 | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \
195 #define RE_SYNTAX_EGREP \
196 (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \
197 | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \
198 | RE_NEWLINE_ALT | RE_NO_BK_PARENS \
201 #define RE_SYNTAX_POSIX_EGREP \
202 (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES)
204 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */
205 #define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
207 #define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
209 /* Syntax bits common to both basic and extended POSIX regex syntax. */
210 #define _RE_SYNTAX_POSIX_COMMON \
211 (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \
212 | RE_INTERVALS | RE_NO_EMPTY_RANGES)
214 #define RE_SYNTAX_POSIX_BASIC \
215 (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM)
217 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
218 RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this
219 isn't minimal, since other operators, such as \`, aren't disabled. */
220 #define RE_SYNTAX_POSIX_MINIMAL_BASIC \
221 (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
223 #define RE_SYNTAX_POSIX_EXTENDED \
224 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
225 | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \
226 | RE_NO_BK_PARENS | RE_NO_BK_VBAR \
227 | RE_UNMATCHED_RIGHT_PAREN_ORD)
229 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INVALID_OPS
230 replaces RE_CONTEXT_INDEP_OPS and RE_NO_BK_REFS is added. */
231 #define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \
232 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
233 | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \
234 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
235 | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD)
236 /* [[[end syntaxes]]] */
238 /* Maximum number of duplicates an interval can allow. Some systems
239 (erroneously) define this in other header files, but we want our
240 value, so remove any previous define. */
244 /* If sizeof(int) == 2, then ((1 << 15) - 1) overflows. */
245 #define RE_DUP_MAX (0x7fff)
248 /* POSIX `cflags' bits (i.e., information for `regcomp'). */
250 /* If this bit is set, then use extended regular expression syntax.
251 If not set, then use basic regular expression syntax. */
252 #define REG_EXTENDED 1
254 /* If this bit is set, then ignore case when matching.
255 If not set, then case is significant. */
256 #define REG_ICASE (REG_EXTENDED << 1)
258 /* If this bit is set, then anchors do not match at newline
259 characters in the string.
260 If not set, then anchors do match at newlines. */
261 #define REG_NEWLINE (REG_ICASE << 1)
263 /* If this bit is set, then report only success or fail in regexec.
264 If not set, then returns differ between not matching and errors. */
265 #define REG_NOSUB (REG_NEWLINE << 1)
268 /* POSIX `eflags' bits (i.e., information for regexec). */
270 /* If this bit is set, then the beginning-of-line operator doesn't match
271 the beginning of the string (presumably because it's not the
272 beginning of a line).
273 If not set, then the beginning-of-line operator does match the
274 beginning of the string. */
277 /* Like REG_NOTBOL, except for the end-of-line. */
278 #define REG_NOTEOL (1 << 1)
281 /* If any error codes are removed, changed, or added, update the
282 `re_error_msg' table in regex.c. */
285 #if (_XOPEN_SOURCE - 0) == 500
286 REG_NOSYS
= -1, /* This will never happen for this implementation. */
289 REG_NOERROR
= 0, /* Success. */
290 REG_NOMATCH
, /* Didn't find a match (for regexec). */
292 /* POSIX regcomp return error codes. (In the order listed in the
294 REG_BADPAT
, /* Invalid pattern. */
295 REG_ECOLLATE
, /* Not implemented. */
296 REG_ECTYPE
, /* Invalid character class name. */
297 REG_EESCAPE
, /* Trailing backslash. */
298 REG_ESUBREG
, /* Invalid back reference. */
299 REG_EBRACK
, /* Unmatched left bracket. */
300 REG_EPAREN
, /* Parenthesis imbalance. */
301 REG_EBRACE
, /* Unmatched \{. */
302 REG_BADBR
, /* Invalid contents of \{\}. */
303 REG_ERANGE
, /* Invalid range end. */
304 REG_ESPACE
, /* Ran out of memory. */
305 REG_BADRPT
, /* No preceding re for repetition op. */
307 /* Error codes we've added. */
308 REG_EEND
, /* Premature end. */
309 REG_ESIZE
, /* Compiled pattern bigger than 2^16 bytes. */
310 REG_ERPAREN
/* Unmatched ) or \); not returned from regcomp. */
313 /* This data structure represents a compiled pattern. Before calling
314 the pattern compiler, the fields `buffer', `allocated', `fastmap',
315 `translate', and `no_sub' can be set. After the pattern has been
316 compiled, the `re_nsub' field is available. All other fields are
317 private to the regex routines. */
319 #ifndef RE_TRANSLATE_TYPE
320 # define RE_TRANSLATE_TYPE char *
323 struct re_pattern_buffer
325 /* [[[begin pattern_buffer]]] */
326 /* Space that holds the compiled pattern. It is declared as
327 `unsigned char *' because its elements are
328 sometimes used as array indexes. */
329 unsigned char *buffer
;
331 /* Number of bytes to which `buffer' points. */
332 unsigned long int allocated
;
334 /* Number of bytes actually used in `buffer'. */
335 unsigned long int used
;
337 /* Syntax setting with which the pattern was compiled. */
340 /* Pointer to a fastmap, if any, otherwise zero. re_search uses
341 the fastmap, if there is one, to skip over impossible
342 starting points for matches. */
345 /* Either a translate table to apply to all characters before
346 comparing them, or zero for no translation. The translation
347 is applied to a pattern when it is compiled and to a string
348 when it is matched. */
349 RE_TRANSLATE_TYPE translate
;
351 /* Number of subexpressions found by the compiler. */
354 /* Zero if this pattern cannot match the empty string, one else.
355 Well, in truth it's used only in `re_search_2', to see
356 whether or not we should use the fastmap, so we don't set
357 this absolutely perfectly; see `re_compile_fastmap' (the
358 `duplicate' case). */
359 unsigned can_be_null
: 1;
361 /* If REGS_UNALLOCATED, allocate space in the `regs' structure
362 for `max (RE_NREGS, re_nsub + 1)' groups.
363 If REGS_REALLOCATE, reallocate space if necessary.
364 If REGS_FIXED, use what's there. */
365 #define REGS_UNALLOCATED 0
366 #define REGS_REALLOCATE 1
368 unsigned regs_allocated
: 2;
370 /* Set to zero when `regex_compile' compiles a pattern; set to one
371 by `re_compile_fastmap' if it updates the fastmap. */
372 unsigned fastmap_accurate
: 1;
374 /* If set, `re_match_2' does not return information about
378 /* If set, a beginning-of-line anchor doesn't match at the
379 beginning of the string. */
380 unsigned not_bol
: 1;
382 /* Similarly for an end-of-line anchor. */
383 unsigned not_eol
: 1;
385 /* If true, an anchor at a newline matches. */
386 unsigned newline_anchor
: 1;
388 /* [[[end pattern_buffer]]] */
391 typedef struct re_pattern_buffer regex_t
;
393 /* Type for byte offsets within the string. POSIX mandates this. */
394 typedef int regoff_t
;
397 /* This is the structure we store register match data in. See
398 regex.texinfo for a full description of what registers match. */
407 /* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
408 `re_match_2' returns information about at least this many registers
409 the first time a `regs' structure is passed. */
415 /* POSIX specification for registers. Aside from the different names than
416 `re_registers', POSIX uses an array of structures, instead of a
417 structure of arrays. */
420 regoff_t rm_so
; /* Byte offset from string's start to substring's start. */
421 regoff_t rm_eo
; /* Byte offset from string's start to substring's end. */
424 /* Declarations for routines. */
426 /* To avoid duplicating every routine declaration -- once with a
427 prototype (if we are ANSI), and once without (if we aren't) -- we
428 use the following macro to declare argument types. This
429 unfortunately clutters up the declarations a bit, but I think it's
434 # define _RE_ARGS(args) args
436 #else /* not __STDC__ */
438 # define _RE_ARGS(args) ()
440 #endif /* not __STDC__ */
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
_RE_ARGS ((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
450 _RE_ARGS ((const char *pattern
, size_t length
,
451 struct re_pattern_buffer
*buffer
));
454 /* Compile a fastmap for the compiled pattern in BUFFER; used to
455 accelerate searches. Return 0 if successful and -2 if was an
457 extern int re_compile_fastmap
_RE_ARGS ((struct re_pattern_buffer
*buffer
));
460 /* Search in the string STRING (with length LENGTH) for the pattern
461 compiled into BUFFER. Start searching at position START, for RANGE
462 characters. Return the starting position of the match, -1 for no
463 match, or -2 for an internal error. Also return register
464 information in REGS (if REGS and BUFFER->no_sub are nonzero). */
466 _RE_ARGS ((struct re_pattern_buffer
*buffer
, const char *string
,
467 int length
, int start
, int range
, struct re_registers
*regs
));
470 /* Like `re_search', but search in the concatenation of STRING1 and
471 STRING2. Also, stop searching at index START + STOP. */
472 extern int re_search_2
473 _RE_ARGS ((struct re_pattern_buffer
*buffer
, const char *string1
,
474 int length1
, const char *string2
, int length2
,
475 int start
, int range
, struct re_registers
*regs
, int stop
));
478 /* Like `re_search', but return how many characters in STRING the regexp
479 in BUFFER matched, starting at position START. */
481 _RE_ARGS ((struct re_pattern_buffer
*buffer
, const char *string
,
482 int length
, int start
, struct re_registers
*regs
));
485 /* Relates to `re_match' as `re_search_2' relates to `re_search'. */
486 extern int re_match_2
487 _RE_ARGS ((struct re_pattern_buffer
*buffer
, const char *string1
,
488 int length1
, const char *string2
, int length2
,
489 int start
, struct re_registers
*regs
, int stop
));
492 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
493 ENDS. Subsequent matches using BUFFER and REGS will use this memory
494 for recording register information. STARTS and ENDS must be
495 allocated with malloc, and must each be at least `NUM_REGS * sizeof
496 (regoff_t)' bytes long.
498 If NUM_REGS == 0, then subsequent matches should allocate their own
501 Unless this function is called, the first search or match using
502 PATTERN_BUFFER will allocate its own register data, without
503 freeing the old data. */
504 extern void re_set_registers
505 _RE_ARGS ((struct re_pattern_buffer
*buffer
, struct re_registers
*regs
,
506 unsigned num_regs
, regoff_t
*starts
, regoff_t
*ends
));
508 #ifdef _REGEX_RE_COMP
510 /* 4.2 bsd compatibility. */
511 extern char *re_comp
_RE_ARGS ((const char *));
512 extern int re_exec
_RE_ARGS ((const char *));
516 /* POSIX compatibility. */
517 extern int regcomp
_RE_ARGS ((regex_t
*preg
, const char *pattern
, int cflags
));
519 _RE_ARGS ((const regex_t
*preg
, const char *string
, size_t nmatch
,
520 regmatch_t pmatch
[], int eflags
));
521 extern size_t regerror
522 _RE_ARGS ((int errcode
, const regex_t
*preg
, char *errbuf
,
523 size_t errbuf_size
));
524 extern void regfree
_RE_ARGS ((regex_t
*preg
));
537 trim-versions-without-asking: nil