1 /* Definitions for data structures and routines for the regular
2 expression library, version 0.12.
4 Copyright (C) 1985-2022 Free Software Foundation, Inc.
6 This file is part of the GNU C Library. Its master source is NOT part of
7 the C library, however. The master source lives in /gd/gnu/lib.
9 The GNU C Library is free software; you can redistribute it and/or
10 modify it under the terms of the GNU Lesser General Public
11 License as published by the Free Software Foundation; either
12 version 2.1 of the License, or (at your option) any later version.
14 The GNU C Library is distributed in the hope that it will be useful,
15 but WITHOUT ANY WARRANTY; without even the implied warranty of
16 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 Lesser General Public License for more details.
19 You should have received a copy of the GNU Lesser General Public
20 License along with the GNU C Library; if not, write to the Free
21 Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
27 /* Allow the use in C++ code. */
32 /* POSIX says that <sys/types.h> must be included (by the caller) before
35 #if !defined _POSIX_C_SOURCE && !defined _POSIX_SOURCE && defined VMS
36 /* VMS doesn't have `size_t' in <sys/types.h>, even though POSIX says it
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
;
55 /* If this bit is not set, then \ inside a bracket expression is literal.
56 If set, then such a \ quotes the following character. */
57 #define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1)
59 /* If this bit is not set, then + and ? are operators, and \+ and \? are
61 If set, then \+ and \? are operators and + and ? are literals. */
62 #define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1)
64 /* If this bit is set, then character classes are supported. They are:
65 [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:],
66 [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:].
67 If not set, then character classes are not supported. */
68 #define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1)
70 /* If this bit is set, then ^ and $ are always anchors (outside bracket
71 expressions, of course).
72 If this bit is not set, then it depends:
73 ^ is an anchor if it is at the beginning of a regular
74 expression or after an open-group or an alternation operator;
75 $ is an anchor if it is at the end of a regular expression, or
76 before a close-group or an alternation operator.
78 This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because
79 POSIX draft 11.2 says that * etc. in leading positions is undefined.
80 We already implemented a previous draft which made those constructs
81 invalid, though, so we haven't changed the code back. */
82 #define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1)
84 /* If this bit is set, then special characters are always special
85 regardless of where they are in the pattern.
86 If this bit is not set, then special characters are special only in
87 some contexts; otherwise they are ordinary. Specifically,
88 * + ? and intervals are only special when not after the beginning,
89 open-group, or alternation operator. */
90 #define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1)
92 /* If this bit is set, then *, +, ?, and { cannot be first in an re or
93 immediately after an alternation or begin-group operator. */
94 #define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1)
96 /* If this bit is set, then . matches newline.
97 If not set, then it doesn't. */
98 #define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1)
100 /* If this bit is set, then . doesn't match NUL.
101 If not set, then it does. */
102 #define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1)
104 /* If this bit is set, nonmatching lists [^...] do not match newline.
105 If not set, they do. */
106 #define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1)
108 /* If this bit is set, either \{...\} or {...} defines an
109 interval, depending on RE_NO_BK_BRACES.
110 If not set, \{, \}, {, and } are literals. */
111 #define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1)
113 /* If this bit is set, +, ? and | aren't recognized as operators.
114 If not set, they are. */
115 #define RE_LIMITED_OPS (RE_INTERVALS << 1)
117 /* If this bit is set, newline is an alternation operator.
118 If not set, newline is literal. */
119 #define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1)
121 /* If this bit is set, then `{...}' defines an interval, and \{ and \}
123 If not set, then `\{...\}' defines an interval. */
124 #define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1)
126 /* If this bit is set, (...) defines a group, and \( and \) are literals.
127 If not set, \(...\) defines a group, and ( and ) are literals. */
128 #define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1)
130 /* If this bit is set, then \<digit> matches <digit>.
131 If not set, then \<digit> is a back-reference. */
132 #define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1)
134 /* If this bit is set, then | is an alternation operator, and \| is literal.
135 If not set, then \| is an alternation operator, and | is literal. */
136 #define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1)
138 /* If this bit is set, then an ending range point collating higher
139 than the starting range point, as in [z-a], is invalid.
140 If not set, then when ending range point collates higher than the
141 starting range point, the range is ignored. */
142 #define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1)
144 /* If this bit is set, then an unmatched ) is ordinary.
145 If not set, then an unmatched ) is invalid. */
146 #define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1)
148 /* If this bit is set, succeed as soon as we match the whole pattern,
149 without further backtracking. */
150 #define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1)
152 /* If this bit is set, do not process the GNU regex operators.
153 If not set, then the GNU regex operators are recognized. */
154 #define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 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_GNU_OPS << 1)
164 /* If this bit is set, a syntactically invalid interval is treated as
165 a string of ordinary characters. For example, the ERE 'a{1' is
166 treated as 'a\{1'. */
167 #define RE_INVALID_INTERVAL_ORD (RE_DEBUG << 1)
169 /* This global variable defines the particular regexp syntax to use (for
170 some interfaces). When a regexp is compiled, the syntax used is
171 stored in the pattern buffer, so changing this does not affect
172 already-compiled regexps. */
173 extern reg_syntax_t re_syntax_options
;
175 /* Define combinations of the above bits for the standard possibilities.
176 (The [[[ comments delimit what gets put into the Texinfo file, so
177 don't delete them!) */
178 /* [[[begin syntaxes]]] */
179 #define RE_SYNTAX_EMACS 0
181 #define RE_SYNTAX_AWK \
182 (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \
183 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
184 | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \
185 | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \
186 | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS)
188 #define RE_SYNTAX_GNU_AWK \
189 ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DEBUG) \
190 & ~(RE_DOT_NOT_NULL | RE_INTERVALS | RE_CONTEXT_INDEP_OPS))
192 #define RE_SYNTAX_POSIX_AWK \
193 (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
194 | RE_INTERVALS | RE_NO_GNU_OPS)
196 #define RE_SYNTAX_GREP \
197 (RE_BK_PLUS_QM | RE_CHAR_CLASSES \
198 | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \
201 #define RE_SYNTAX_EGREP \
202 (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \
203 | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \
204 | RE_NEWLINE_ALT | RE_NO_BK_PARENS \
207 #define RE_SYNTAX_POSIX_EGREP \
208 (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES \
209 | RE_INVALID_INTERVAL_ORD)
211 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */
212 #define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
214 #define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
216 /* Syntax bits common to both basic and extended POSIX regex syntax. */
217 #define _RE_SYNTAX_POSIX_COMMON \
218 (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \
219 | RE_INTERVALS | RE_NO_EMPTY_RANGES)
221 #define RE_SYNTAX_POSIX_BASIC \
222 (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM)
224 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
225 RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this
226 isn't minimal, since other operators, such as \`, aren't disabled. */
227 #define RE_SYNTAX_POSIX_MINIMAL_BASIC \
228 (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
230 #define RE_SYNTAX_POSIX_EXTENDED \
231 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
232 | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \
233 | RE_NO_BK_PARENS | RE_NO_BK_VBAR \
234 | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD)
236 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is
237 removed and RE_NO_BK_REFS is added. */
238 #define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \
239 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
240 | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \
241 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
242 | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD)
243 /* [[[end syntaxes]]] */
245 /* Maximum number of duplicates an interval can allow. Some systems
246 (erroneously) define this in other header files, but we want our
247 value, so remove any previous define. */
251 /* If sizeof(int) == 2, then ((1 << 15) - 1) overflows. */
252 #define RE_DUP_MAX (0x7fff)
255 /* POSIX `cflags' bits (i.e., information for `regcomp'). */
257 /* If this bit is set, then use extended regular expression syntax.
258 If not set, then use basic regular expression syntax. */
259 #define REG_EXTENDED 1
261 /* If this bit is set, then ignore case when matching.
262 If not set, then case is significant. */
263 #define REG_ICASE (REG_EXTENDED << 1)
265 /* If this bit is set, then anchors do not match at newline
266 characters in the string.
267 If not set, then anchors do match at newlines. */
268 #define REG_NEWLINE (REG_ICASE << 1)
270 /* If this bit is set, then report only success or fail in regexec.
271 If not set, then returns differ between not matching and errors. */
272 #define REG_NOSUB (REG_NEWLINE << 1)
275 /* POSIX `eflags' bits (i.e., information for regexec). */
277 /* If this bit is set, then the beginning-of-line operator doesn't match
278 the beginning of the string (presumably because it's not the
279 beginning of a line).
280 If not set, then the beginning-of-line operator does match the
281 beginning of the string. */
284 /* Like REG_NOTBOL, except for the end-of-line. */
285 #define REG_NOTEOL (1 << 1)
288 /* If any error codes are removed, changed, or added, update the
289 `re_error_msg' table in regex.c. */
293 REG_ENOSYS
= -1, /* This will never happen for this implementation. */
296 REG_NOERROR
= 0, /* Success. */
297 REG_NOMATCH
, /* Didn't find a match (for regexec). */
299 /* POSIX regcomp return error codes. (In the order listed in the
301 REG_BADPAT
, /* Invalid pattern. */
302 REG_ECOLLATE
, /* Not implemented. */
303 REG_ECTYPE
, /* Invalid character class name. */
304 REG_EESCAPE
, /* Trailing backslash. */
305 REG_ESUBREG
, /* Invalid back reference. */
306 REG_EBRACK
, /* Unmatched left bracket. */
307 REG_EPAREN
, /* Parenthesis imbalance. */
308 REG_EBRACE
, /* Unmatched \{. */
309 REG_BADBR
, /* Invalid contents of \{\}. */
310 REG_ERANGE
, /* Invalid range end. */
311 REG_ESPACE
, /* Ran out of memory. */
312 REG_BADRPT
, /* No preceding re for repetition op. */
314 /* Error codes we've added. */
315 REG_EEND
, /* Premature end. */
316 REG_ESIZE
, /* Compiled pattern bigger than 2^16 bytes. */
317 REG_ERPAREN
/* Unmatched ) or \); not returned from regcomp. */
320 /* This data structure represents a compiled pattern. Before calling
321 the pattern compiler, the fields `buffer', `allocated', `fastmap',
322 `translate', and `no_sub' can be set. After the pattern has been
323 compiled, the `re_nsub' field is available. All other fields are
324 private to the regex routines. */
326 #ifndef RE_TRANSLATE_TYPE
327 # define RE_TRANSLATE_TYPE char *
330 struct re_pattern_buffer
332 /* [[[begin pattern_buffer]]] */
333 /* Space that holds the compiled pattern. It is declared as
334 `unsigned char *' because its elements are
335 sometimes used as array indexes. */
336 unsigned char *buffer
;
338 /* Number of bytes to which `buffer' points. */
339 unsigned long int allocated
;
341 /* Number of bytes actually used in `buffer'. */
342 unsigned long int used
;
344 /* Syntax setting with which the pattern was compiled. */
347 /* Pointer to a fastmap, if any, otherwise zero. re_search uses
348 the fastmap, if there is one, to skip over impossible
349 starting points for matches. */
352 /* Either a translate table to apply to all characters before
353 comparing them, or zero for no translation. The translation
354 is applied to a pattern when it is compiled and to a string
355 when it is matched. */
356 RE_TRANSLATE_TYPE translate
;
358 /* Number of subexpressions found by the compiler. */
361 /* Zero if this pattern cannot match the empty string, one else.
362 Well, in truth it's used only in `re_search_2', to see
363 whether or not we should use the fastmap, so we don't set
364 this absolutely perfectly; see `re_compile_fastmap' (the
365 `duplicate' case). */
366 unsigned can_be_null
: 1;
368 /* If REGS_UNALLOCATED, allocate space in the `regs' structure
369 for `max (RE_NREGS, re_nsub + 1)' groups.
370 If REGS_REALLOCATE, reallocate space if necessary.
371 If REGS_FIXED, use what's there. */
372 #define REGS_UNALLOCATED 0
373 #define REGS_REALLOCATE 1
375 unsigned regs_allocated
: 2;
377 /* Set to zero when `regex_compile' compiles a pattern; set to one
378 by `re_compile_fastmap' if it updates the fastmap. */
379 unsigned fastmap_accurate
: 1;
381 /* If set, `re_match_2' does not return information about
385 /* If set, a beginning-of-line anchor doesn't match at the
386 beginning of the string. */
387 unsigned not_bol
: 1;
389 /* Similarly for an end-of-line anchor. */
390 unsigned not_eol
: 1;
392 /* If true, an anchor at a newline matches. */
393 unsigned newline_anchor
: 1;
395 /* [[[end pattern_buffer]]] */
398 typedef struct re_pattern_buffer regex_t
;
400 /* Type for byte offsets within the string. POSIX mandates this. */
401 typedef int regoff_t
;
404 /* This is the structure we store register match data in. See
405 regex.texinfo for a full description of what registers match. */
414 /* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
415 `re_match_2' returns information about at least this many registers
416 the first time a `regs' structure is passed. */
422 /* POSIX specification for registers. Aside from the different names than
423 `re_registers', POSIX uses an array of structures, instead of a
424 structure of arrays. */
427 regoff_t rm_so
; /* Byte offset from string's start to substring's start. */
428 regoff_t rm_eo
; /* Byte offset from string's start to substring's end. */
431 /* Declarations for routines. */
433 /* To avoid duplicating every routine declaration -- once with a
434 prototype (if we are ANSI), and once without (if we aren't) -- we
435 use the following macro to declare argument types. This
436 unfortunately clutters up the declarations a bit, but I think it's
439 /* Sets the current default syntax to SYNTAX, and return the old syntax.
440 You can also simply assign to the `re_syntax_options' variable. */
441 extern reg_syntax_t
re_set_syntax (reg_syntax_t syntax
);
443 /* Compile the regular expression PATTERN, with length LENGTH
444 and syntax given by the global `re_syntax_options', into the buffer
445 BUFFER. Return NULL if successful, and an error string if not. */
446 extern const char *re_compile_pattern (const char *pattern
, size_t length
,
447 struct re_pattern_buffer
*buffer
);
450 /* Compile a fastmap for the compiled pattern in BUFFER; used to
451 accelerate searches. Return 0 if successful and -2 if was an
453 extern int re_compile_fastmap (struct re_pattern_buffer
*buffer
);
456 /* Search in the string STRING (with length LENGTH) for the pattern
457 compiled into BUFFER. Start searching at position START, for RANGE
458 characters. Return the starting position of the match, -1 for no
459 match, or -2 for an internal error. Also return register
460 information in REGS (if REGS and BUFFER->no_sub are nonzero). */
461 extern int re_search (struct re_pattern_buffer
*buffer
, const char *string
,
462 int length
, int start
, int range
,
463 struct re_registers
*regs
);
466 /* Like `re_search', but search in the concatenation of STRING1 and
467 STRING2. Also, stop searching at index START + STOP. */
468 extern int re_search_2 (struct re_pattern_buffer
*buffer
, const char *string1
,
469 int length1
, const char *string2
, int length2
,
470 int start
, int range
, struct re_registers
*regs
,
474 /* Like `re_search', but return how many characters in STRING the regexp
475 in BUFFER matched, starting at position START. */
476 extern int re_match (struct re_pattern_buffer
*buffer
, const char *string
,
477 int length
, int start
, struct re_registers
*regs
);
480 /* Relates to `re_match' as `re_search_2' relates to `re_search'. */
481 extern int re_match_2 (struct re_pattern_buffer
*buffer
, const char *string1
,
482 int length1
, const char *string2
, int length2
,
483 int start
, struct re_registers
*regs
, int stop
);
486 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
487 ENDS. Subsequent matches using BUFFER and REGS will use this memory
488 for recording register information. STARTS and ENDS must be
489 allocated with malloc, and must each be at least `NUM_REGS * sizeof
490 (regoff_t)' bytes long.
492 If NUM_REGS == 0, then subsequent matches should allocate their own
495 Unless this function is called, the first search or match using
496 PATTERN_BUFFER will allocate its own register data, without
497 freeing the old data. */
498 extern void re_set_registers (struct re_pattern_buffer
*buffer
,
499 struct re_registers
*regs
,
500 unsigned num_regs
, regoff_t
*starts
,
503 #if defined _REGEX_RE_COMP || defined _LIBC
505 /* 4.2 bsd compatibility. */
506 extern char *re_comp (const char *);
507 extern int re_exec (const char *);
511 /* GCC 2.95 and later have "__restrict"; C99 compilers have
512 "restrict", and "configure" may have defined "restrict". */
514 # if ! (2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__))
515 # if defined restrict || 199901L <= __STDC_VERSION__
516 # define __restrict restrict
523 /* GCC 3.1 and later support declaring arrays as non-overlapping
524 using the syntax array_name[restrict] */
525 #ifndef __restrict_arr
526 # if ! (3 < __GNUC__ || (3 == __GNUC__ && 1 <= __GNUC_MINOR__)) || defined (__GNUG__)
527 # define __restrict_arr
529 # define __restrict_arr __restrict
533 /* POSIX compatibility. */
534 extern int regcomp (regex_t
*__restrict __preg
,
535 const char *__restrict __pattern
,
541 extern int regexec (const regex_t
*__restrict __preg
,
542 const char *__restrict __string
, size_t __nmatch
,
543 regmatch_t __pmatch
[__restrict_arr
],
546 extern size_t regerror (int __errcode
, const regex_t
*__preg
,
547 char *__errbuf
, size_t __errbuf_size
);
549 extern void regfree (regex_t
*__preg
);
562 trim-versions-without-asking: nil