1 /* Definitions for data structures and routines for the regular
2 expression library, version 0.12.
3 Copyright (C) 1985,1989-1993,1995-1998, 2000 Free Software Foundation, Inc.
5 This program is free software; you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published by
7 the Free Software Foundation; either version 2, or (at your option)
10 This program is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 GNU General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with this program; if not, write to the Free Software Foundation,
17 Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
22 /* Allow the use in C++ code. */
27 /* POSIX says that <sys/types.h> must be included (by the caller) before
30 #if !defined _POSIX_C_SOURCE && !defined _POSIX_SOURCE && defined VMS
31 /* VMS doesn't have `size_t' in <sys/types.h>, even though POSIX says it
36 /* The following two types have to be signed and unsigned integer type
37 wide enough to hold a value of a pointer. For most ANSI compilers
38 ptrdiff_t and size_t should be likely OK. Still size of these two
39 types is 2 for Microsoft C. Ugh... */
40 typedef long int s_reg_t;
41 typedef unsigned long int active_reg_t;
43 /* The following bits are used to determine the regexp syntax we
44 recognize. The set/not-set meanings are chosen so that Emacs syntax
45 remains the value 0. The bits are given in alphabetical order, and
46 the definitions shifted by one from the previous bit; thus, when we
47 add or remove a bit, only one other definition need change. */
48 typedef unsigned long int reg_syntax_t;
50 /* If this bit is not set, then \ inside a bracket expression is literal.
51 If set, then such a \ quotes the following character. */
52 #define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1)
54 /* If this bit is not set, then + and ? are operators, and \+ and \? are
56 If set, then \+ and \? are operators and + and ? are literals. */
57 #define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1)
59 /* If this bit is set, then character classes are supported. They are:
60 [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:],
61 [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:].
62 If not set, then character classes are not supported. */
63 #define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1)
65 /* If this bit is set, then ^ and $ are always anchors (outside bracket
66 expressions, of course).
67 If this bit is not set, then it depends:
68 ^ is an anchor if it is at the beginning of a regular
69 expression or after an open-group or an alternation operator;
70 $ is an anchor if it is at the end of a regular expression, or
71 before a close-group or an alternation operator.
73 This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because
74 POSIX draft 11.2 says that * etc. in leading positions is undefined.
75 We already implemented a previous draft which made those constructs
76 invalid, though, so we haven't changed the code back. */
77 #define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1)
79 /* If this bit is set, then special characters are always special
80 regardless of where they are in the pattern.
81 If this bit is not set, then special characters are special only in
82 some contexts; otherwise they are ordinary. Specifically,
83 * + ? and intervals are only special when not after the beginning,
84 open-group, or alternation operator. */
85 #define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1)
87 /* If this bit is set, then *, +, ?, and { cannot be first in an re or
88 immediately after an alternation or begin-group operator. */
89 #define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1)
91 /* If this bit is set, then . matches newline.
92 If not set, then it doesn't. */
93 #define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1)
95 /* If this bit is set, then . doesn't match NUL.
96 If not set, then it does. */
97 #define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1)
99 /* If this bit is set, nonmatching lists [^...] do not match newline.
100 If not set, they do. */
101 #define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1)
103 /* If this bit is set, either \{...\} or {...} defines an
104 interval, depending on RE_NO_BK_BRACES.
105 If not set, \{, \}, {, and } are literals. */
106 #define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1)
108 /* If this bit is set, +, ? and | aren't recognized as operators.
109 If not set, they are. */
110 #define RE_LIMITED_OPS (RE_INTERVALS << 1)
112 /* If this bit is set, newline is an alternation operator.
113 If not set, newline is literal. */
114 #define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1)
116 /* If this bit is set, then `{...}' defines an interval, and \{ and \}
118 If not set, then `\{...\}' defines an interval. */
119 #define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1)
121 /* If this bit is set, (...) defines a group, and \( and \) are literals.
122 If not set, \(...\) defines a group, and ( and ) are literals. */
123 #define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1)
125 /* If this bit is set, then \<digit> matches <digit>.
126 If not set, then \<digit> is a back-reference. */
127 #define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1)
129 /* If this bit is set, then | is an alternation operator, and \| is literal.
130 If not set, then \| is an alternation operator, and | is literal. */
131 #define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1)
133 /* If this bit is set, then an ending range point collating higher
134 than the starting range point, as in [z-a], is invalid.
135 If not set, then when ending range point collates higher than the
136 starting range point, the range is ignored. */
137 #define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1)
139 /* If this bit is set, then an unmatched ) is ordinary.
140 If not set, then an unmatched ) is invalid. */
141 #define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1)
143 /* If this bit is set, succeed as soon as we match the whole pattern,
144 without further backtracking. */
145 #define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1)
147 /* If this bit is set, do not process the GNU regex operators.
148 If not set, then the GNU regex operators are recognized. */
149 #define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1)
151 /* If this bit is set, turn on internal regex debugging.
152 If not set, and debugging was on, turn it off.
153 This only works if regex.c is compiled -DDEBUG.
154 We define this bit always, so that all that's needed to turn on
155 debugging is to recompile regex.c; the calling code can always have
156 this bit set, and it won't affect anything in the normal case. */
157 #define RE_DEBUG (RE_NO_GNU_OPS << 1)
159 /* If this bit is set, a syntactically invalid interval is treated as
160 a string of ordinary characters. For example, the ERE 'a{1' is
161 treated as 'a\{1'. */
162 #define RE_INVALID_INTERVAL_ORD (RE_DEBUG << 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 /* Define combinations of the above bits for the standard possibilities.
171 (The [[[ comments delimit what gets put into the Texinfo file, so
172 don't delete them!) */
173 /* [[[begin syntaxes]]] */
174 #define RE_SYNTAX_EMACS 0
176 #define RE_SYNTAX_AWK \
177 (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \
178 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
179 | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \
180 | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \
181 | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS)
183 #define RE_SYNTAX_GNU_AWK \
184 ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DEBUG) \
185 & ~(RE_DOT_NOT_NULL | RE_INTERVALS | RE_CONTEXT_INDEP_OPS))
187 #define RE_SYNTAX_POSIX_AWK \
188 (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
189 | RE_INTERVALS | RE_NO_GNU_OPS)
191 #define RE_SYNTAX_GREP \
192 (RE_BK_PLUS_QM | RE_CHAR_CLASSES \
193 | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \
196 #define RE_SYNTAX_EGREP \
197 (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \
198 | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \
199 | RE_NEWLINE_ALT | RE_NO_BK_PARENS \
202 #define RE_SYNTAX_POSIX_EGREP \
203 (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES \
204 | RE_INVALID_INTERVAL_ORD)
206 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */
207 #define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
209 #define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
211 /* Syntax bits common to both basic and extended POSIX regex syntax. */
212 #define _RE_SYNTAX_POSIX_COMMON \
213 (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \
214 | RE_INTERVALS | RE_NO_EMPTY_RANGES)
216 #define RE_SYNTAX_POSIX_BASIC \
217 (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM)
219 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
220 RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this
221 isn't minimal, since other operators, such as \`, aren't disabled. */
222 #define RE_SYNTAX_POSIX_MINIMAL_BASIC \
223 (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
225 #define RE_SYNTAX_POSIX_EXTENDED \
226 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
227 | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \
228 | RE_NO_BK_PARENS | RE_NO_BK_VBAR \
229 | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD)
231 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is
232 removed and RE_NO_BK_REFS is added. */
233 #define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \
234 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
235 | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \
236 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
237 | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD)
238 /* [[[end syntaxes]]] */
240 /* Maximum number of duplicates an interval can allow. Some systems
241 (erroneously) define this in other header files, but we want our
242 value, so remove any previous define. */
246 /* If sizeof(int) == 2, then ((1 << 15) - 1) overflows. */
247 #define RE_DUP_MAX (0x7fff)
250 /* POSIX `cflags' bits (i.e., information for `regcomp'). */
252 /* If this bit is set, then use extended regular expression syntax.
253 If not set, then use basic regular expression syntax. */
254 #define REG_EXTENDED 1
256 /* If this bit is set, then ignore case when matching.
257 If not set, then case is significant. */
258 #define REG_ICASE (REG_EXTENDED << 1)
260 /* If this bit is set, then anchors do not match at newline
261 characters in the string.
262 If not set, then anchors do match at newlines. */
263 #define REG_NEWLINE (REG_ICASE << 1)
265 /* If this bit is set, then report only success or fail in regexec.
266 If not set, then returns differ between not matching and errors. */
267 #define REG_NOSUB (REG_NEWLINE << 1)
270 /* POSIX `eflags' bits (i.e., information for regexec). */
272 /* If this bit is set, then the beginning-of-line operator doesn't match
273 the beginning of the string (presumably because it's not the
274 beginning of a line).
275 If not set, then the beginning-of-line operator does match the
276 beginning of the string. */
279 /* Like REG_NOTBOL, except for the end-of-line. */
280 #define REG_NOTEOL (1 << 1)
283 /* If any error codes are removed, changed, or added, update the
284 `re_error_msg' table in regex.c. */
288 REG_ENOSYS = -1, /* This will never happen for this implementation. */
291 REG_NOERROR = 0, /* Success. */
292 REG_NOMATCH, /* Didn't find a match (for regexec). */
294 /* POSIX regcomp return error codes. (In the order listed in the
296 REG_BADPAT, /* Invalid pattern. */
297 REG_ECOLLATE, /* Not implemented. */
298 REG_ECTYPE, /* Invalid character class name. */
299 REG_EESCAPE, /* Trailing backslash. */
300 REG_ESUBREG, /* Invalid back reference. */
301 REG_EBRACK, /* Unmatched left bracket. */
302 REG_EPAREN, /* Parenthesis imbalance. */
303 REG_EBRACE, /* Unmatched \{. */
304 REG_BADBR, /* Invalid contents of \{\}. */
305 REG_ERANGE, /* Invalid range end. */
306 REG_ESPACE, /* Ran out of memory. */
307 REG_BADRPT, /* No preceding re for repetition op. */
309 /* Error codes we've added. */
310 REG_EEND, /* Premature end. */
311 REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */
312 REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */
315 /* This data structure represents a compiled pattern. Before calling
316 the pattern compiler, the fields `buffer', `allocated', `fastmap',
317 `translate', and `no_sub' can be set. After the pattern has been
318 compiled, the `re_nsub' field is available. All other fields are
319 private to the regex routines. */
321 #ifndef RE_TRANSLATE_TYPE
322 # define RE_TRANSLATE_TYPE char *
325 struct re_pattern_buffer
327 /* [[[begin pattern_buffer]]] */
328 /* Space that holds the compiled pattern. It is declared as
329 `unsigned char *' because its elements are
330 sometimes used as array indexes. */
331 unsigned char *buffer;
333 /* Number of bytes to which `buffer' points. */
334 unsigned long int allocated;
336 /* Number of bytes actually used in `buffer'. */
337 unsigned long int used;
339 /* Syntax setting with which the pattern was compiled. */
342 /* Pointer to a fastmap, if any, otherwise zero. re_search uses
343 the fastmap, if there is one, to skip over impossible
344 starting points for matches. */
347 /* Either a translate table to apply to all characters before
348 comparing them, or zero for no translation. The translation
349 is applied to a pattern when it is compiled and to a string
350 when it is matched. */
351 RE_TRANSLATE_TYPE translate;
353 /* Number of subexpressions found by the compiler. */
356 /* Zero if this pattern cannot match the empty string, one else.
357 Well, in truth it's used only in `re_search_2', to see
358 whether or not we should use the fastmap, so we don't set
359 this absolutely perfectly; see `re_compile_fastmap' (the
360 `duplicate' case). */
361 unsigned can_be_null : 1;
363 /* If REGS_UNALLOCATED, allocate space in the `regs' structure
364 for `max (RE_NREGS, re_nsub + 1)' groups.
365 If REGS_REALLOCATE, reallocate space if necessary.
366 If REGS_FIXED, use what's there. */
367 #define REGS_UNALLOCATED 0
368 #define REGS_REALLOCATE 1
370 unsigned regs_allocated : 2;
372 /* Set to zero when `regex_compile' compiles a pattern; set to one
373 by `re_compile_fastmap' if it updates the fastmap. */
374 unsigned fastmap_accurate : 1;
376 /* If set, `re_match_2' does not return information about
380 /* If set, a beginning-of-line anchor doesn't match at the
381 beginning of the string. */
382 unsigned not_bol : 1;
384 /* Similarly for an end-of-line anchor. */
385 unsigned not_eol : 1;
387 /* If true, an anchor at a newline matches. */
388 unsigned newline_anchor : 1;
390 /* [[[end pattern_buffer]]] */
393 typedef struct re_pattern_buffer regex_t;
395 /* Type for byte offsets within the string. POSIX mandates this. */
396 typedef int regoff_t;
399 /* This is the structure we store register match data in. See
400 regex.texinfo for a full description of what registers match. */
409 /* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
410 `re_match_2' returns information about at least this many registers
411 the first time a `regs' structure is passed. */
417 /* POSIX specification for registers. Aside from the different names than
418 `re_registers', POSIX uses an array of structures, instead of a
419 structure of arrays. */
422 regoff_t rm_so; /* Byte offset from string's start to substring's start. */
423 regoff_t rm_eo; /* Byte offset from string's start to substring's end. */
426 /* Declarations for routines. */
428 /* To avoid duplicating every routine declaration -- once with a
429 prototype (if we are ANSI), and once without (if we aren't) -- we
430 use the following macro to declare argument types. This
431 unfortunately clutters up the declarations a bit, but I think it's
436 # define _RE_ARGS(args) args
438 #else /* not __STDC__ */
440 # define _RE_ARGS(args) ()
442 #endif /* not __STDC__ */
444 /* Sets the current default syntax to SYNTAX, and return the old syntax.
445 You can also simply assign to the `re_syntax_options' variable. */
446 extern reg_syntax_t re_set_syntax _RE_ARGS ((reg_syntax_t syntax));
448 /* Compile the regular expression PATTERN, with length LENGTH
449 and syntax given by the global `re_syntax_options', into the buffer
450 BUFFER. Return NULL if successful, and an error string if not. */
451 extern const char *re_compile_pattern
452 _RE_ARGS ((const char *pattern, size_t length,
453 struct re_pattern_buffer *buffer));
456 /* Compile a fastmap for the compiled pattern in BUFFER; used to
457 accelerate searches. Return 0 if successful and -2 if was an
459 extern int re_compile_fastmap _RE_ARGS ((struct re_pattern_buffer *buffer));
462 /* Search in the string STRING (with length LENGTH) for the pattern
463 compiled into BUFFER. Start searching at position START, for RANGE
464 characters. Return the starting position of the match, -1 for no
465 match, or -2 for an internal error. Also return register
466 information in REGS (if REGS and BUFFER->no_sub are nonzero). */
468 _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string,
469 int length, int start, int range, struct re_registers *regs));
472 /* Like `re_search', but search in the concatenation of STRING1 and
473 STRING2. Also, stop searching at index START + STOP. */
474 extern int re_search_2
475 _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1,
476 int length1, const char *string2, int length2,
477 int start, int range, struct re_registers *regs, int stop));
480 /* Like `re_search', but return how many characters in STRING the regexp
481 in BUFFER matched, starting at position START. */
483 _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string,
484 int length, int start, struct re_registers *regs));
487 /* Relates to `re_match' as `re_search_2' relates to `re_search'. */
488 extern int re_match_2
489 _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1,
490 int length1, const char *string2, int length2,
491 int start, struct re_registers *regs, int stop));
494 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
495 ENDS. Subsequent matches using BUFFER and REGS will use this memory
496 for recording register information. STARTS and ENDS must be
497 allocated with malloc, and must each be at least `NUM_REGS * sizeof
498 (regoff_t)' bytes long.
500 If NUM_REGS == 0, then subsequent matches should allocate their own
503 Unless this function is called, the first search or match using
504 PATTERN_BUFFER will allocate its own register data, without
505 freeing the old data. */
506 extern void re_set_registers
507 _RE_ARGS ((struct re_pattern_buffer *buffer, struct re_registers *regs,
508 unsigned num_regs, regoff_t *starts, regoff_t *ends));
510 #if defined _REGEX_RE_COMP || defined _LIBC
512 /* 4.2 bsd compatibility. */
513 extern char *re_comp _RE_ARGS ((const char *));
514 extern int re_exec _RE_ARGS ((const char *));
518 /* GCC 2.95 and later have "__restrict"; C99 compilers have
519 "restrict", and "configure" may have defined "restrict". */
521 # if ! (2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__))
522 # if defined restrict || 199901L <= __STDC_VERSION__
523 # define __restrict restrict
529 /* For now unconditionally define __restrict_arr to expand to nothing.
530 Ideally we would have a test for the compiler which allows defining
532 #define __restrict_arr
534 /* POSIX compatibility. */
535 extern int regcomp _RE_ARGS ((regex_t *__restrict __preg,
536 const char *__restrict __pattern,
539 extern int regexec _RE_ARGS ((const regex_t *__restrict __preg,
540 const char *__restrict __string, size_t __nmatch,
541 regmatch_t __pmatch[__restrict_arr],
544 extern size_t regerror _RE_ARGS ((int __errcode, const regex_t *__preg,
545 char *__errbuf, size_t __errbuf_size));
547 extern void regfree _RE_ARGS ((regex_t *__preg));
560 trim-versions-without-asking: nil