2007-10-10 H.J. Lu <hongjiu.lu@intel.com>
[binutils.git] / include / xregex2.h
blobd3d0da14a9878b40291ca954b8d524f0e6e8a343
1 /* Definitions for data structures and routines for the regular
2 expression library, version 0.12.
4 Copyright (C) 1985, 1989, 1990, 1991, 1992, 1993, 1995, 1996, 1997,
5 1998, 2000, 2005 Free Software Foundation, Inc.
7 This file is part of the GNU C Library. Its master source is NOT part of
8 the C library, however. The master source lives in /gd/gnu/lib.
10 The GNU C Library is free software; you can redistribute it and/or
11 modify it under the terms of the GNU Lesser General Public
12 License as published by the Free Software Foundation; either
13 version 2.1 of the License, or (at your option) any later version.
15 The GNU C Library is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 Lesser General Public License for more details.
20 You should have received a copy of the GNU Lesser General Public
21 License along with the GNU C Library; if not, write to the Free
22 Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
23 02110-1301 USA. */
25 #ifndef _REGEX_H
26 #define _REGEX_H 1
28 /* Allow the use in C++ code. */
29 #ifdef __cplusplus
30 extern "C" {
31 #endif
33 /* POSIX says that <sys/types.h> must be included (by the caller) before
34 <regex.h>. */
36 #if !defined _POSIX_C_SOURCE && !defined _POSIX_SOURCE && defined VMS
37 /* VMS doesn't have `size_t' in <sys/types.h>, even though POSIX says it
38 should be there. */
39 # include <stddef.h>
40 #endif
42 /* The following two types have to be signed and unsigned integer type
43 wide enough to hold a value of a pointer. For most ANSI compilers
44 ptrdiff_t and size_t should be likely OK. Still size of these two
45 types is 2 for Microsoft C. Ugh... */
46 typedef long int s_reg_t;
47 typedef unsigned long int active_reg_t;
49 /* The following bits are used to determine the regexp syntax we
50 recognize. The set/not-set meanings are chosen so that Emacs syntax
51 remains the value 0. The bits are given in alphabetical order, and
52 the definitions shifted by one from the previous bit; thus, when we
53 add or remove a bit, only one other definition need change. */
54 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
61 literals.
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 \}
123 are literals.
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 /* This global variable defines the particular regexp syntax to use (for
171 some interfaces). When a regexp is compiled, the syntax used is
172 stored in the pattern buffer, so changing this does not affect
173 already-compiled regexps. */
174 extern reg_syntax_t re_syntax_options;
176 /* Define combinations of the above bits for the standard possibilities.
177 (The [[[ comments delimit what gets put into the Texinfo file, so
178 don't delete them!) */
179 /* [[[begin syntaxes]]] */
180 #define RE_SYNTAX_EMACS 0
182 #define RE_SYNTAX_AWK \
183 (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \
184 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
185 | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \
186 | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \
187 | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS)
189 #define RE_SYNTAX_GNU_AWK \
190 ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DEBUG) \
191 & ~(RE_DOT_NOT_NULL | RE_INTERVALS | RE_CONTEXT_INDEP_OPS))
193 #define RE_SYNTAX_POSIX_AWK \
194 (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
195 | RE_INTERVALS | RE_NO_GNU_OPS)
197 #define RE_SYNTAX_GREP \
198 (RE_BK_PLUS_QM | RE_CHAR_CLASSES \
199 | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \
200 | RE_NEWLINE_ALT)
202 #define RE_SYNTAX_EGREP \
203 (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \
204 | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \
205 | RE_NEWLINE_ALT | RE_NO_BK_PARENS \
206 | RE_NO_BK_VBAR)
208 #define RE_SYNTAX_POSIX_EGREP \
209 (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES \
210 | RE_INVALID_INTERVAL_ORD)
212 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */
213 #define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
215 #define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
217 /* Syntax bits common to both basic and extended POSIX regex syntax. */
218 #define _RE_SYNTAX_POSIX_COMMON \
219 (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \
220 | RE_INTERVALS | RE_NO_EMPTY_RANGES)
222 #define RE_SYNTAX_POSIX_BASIC \
223 (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM)
225 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
226 RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this
227 isn't minimal, since other operators, such as \`, aren't disabled. */
228 #define RE_SYNTAX_POSIX_MINIMAL_BASIC \
229 (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
231 #define RE_SYNTAX_POSIX_EXTENDED \
232 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
233 | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \
234 | RE_NO_BK_PARENS | RE_NO_BK_VBAR \
235 | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD)
237 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is
238 removed and RE_NO_BK_REFS is added. */
239 #define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \
240 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
241 | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \
242 | RE_NO_BK_PARENS | RE_NO_BK_REFS \
243 | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD)
244 /* [[[end syntaxes]]] */
246 /* Maximum number of duplicates an interval can allow. Some systems
247 (erroneously) define this in other header files, but we want our
248 value, so remove any previous define. */
249 #ifdef RE_DUP_MAX
250 # undef RE_DUP_MAX
251 #endif
252 /* If sizeof(int) == 2, then ((1 << 15) - 1) overflows. */
253 #define RE_DUP_MAX (0x7fff)
256 /* POSIX `cflags' bits (i.e., information for `regcomp'). */
258 /* If this bit is set, then use extended regular expression syntax.
259 If not set, then use basic regular expression syntax. */
260 #define REG_EXTENDED 1
262 /* If this bit is set, then ignore case when matching.
263 If not set, then case is significant. */
264 #define REG_ICASE (REG_EXTENDED << 1)
266 /* If this bit is set, then anchors do not match at newline
267 characters in the string.
268 If not set, then anchors do match at newlines. */
269 #define REG_NEWLINE (REG_ICASE << 1)
271 /* If this bit is set, then report only success or fail in regexec.
272 If not set, then returns differ between not matching and errors. */
273 #define REG_NOSUB (REG_NEWLINE << 1)
276 /* POSIX `eflags' bits (i.e., information for regexec). */
278 /* If this bit is set, then the beginning-of-line operator doesn't match
279 the beginning of the string (presumably because it's not the
280 beginning of a line).
281 If not set, then the beginning-of-line operator does match the
282 beginning of the string. */
283 #define REG_NOTBOL 1
285 /* Like REG_NOTBOL, except for the end-of-line. */
286 #define REG_NOTEOL (1 << 1)
289 /* If any error codes are removed, changed, or added, update the
290 `re_error_msg' table in regex.c. */
291 typedef enum
293 #ifdef _XOPEN_SOURCE
294 REG_ENOSYS = -1, /* This will never happen for this implementation. */
295 #endif
297 REG_NOERROR = 0, /* Success. */
298 REG_NOMATCH, /* Didn't find a match (for regexec). */
300 /* POSIX regcomp return error codes. (In the order listed in the
301 standard.) */
302 REG_BADPAT, /* Invalid pattern. */
303 REG_ECOLLATE, /* Not implemented. */
304 REG_ECTYPE, /* Invalid character class name. */
305 REG_EESCAPE, /* Trailing backslash. */
306 REG_ESUBREG, /* Invalid back reference. */
307 REG_EBRACK, /* Unmatched left bracket. */
308 REG_EPAREN, /* Parenthesis imbalance. */
309 REG_EBRACE, /* Unmatched \{. */
310 REG_BADBR, /* Invalid contents of \{\}. */
311 REG_ERANGE, /* Invalid range end. */
312 REG_ESPACE, /* Ran out of memory. */
313 REG_BADRPT, /* No preceding re for repetition op. */
315 /* Error codes we've added. */
316 REG_EEND, /* Premature end. */
317 REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */
318 REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */
319 } reg_errcode_t;
321 /* This data structure represents a compiled pattern. Before calling
322 the pattern compiler, the fields `buffer', `allocated', `fastmap',
323 `translate', and `no_sub' can be set. After the pattern has been
324 compiled, the `re_nsub' field is available. All other fields are
325 private to the regex routines. */
327 #ifndef RE_TRANSLATE_TYPE
328 # define RE_TRANSLATE_TYPE char *
329 #endif
331 struct re_pattern_buffer
333 /* [[[begin pattern_buffer]]] */
334 /* Space that holds the compiled pattern. It is declared as
335 `unsigned char *' because its elements are
336 sometimes used as array indexes. */
337 unsigned char *buffer;
339 /* Number of bytes to which `buffer' points. */
340 unsigned long int allocated;
342 /* Number of bytes actually used in `buffer'. */
343 unsigned long int used;
345 /* Syntax setting with which the pattern was compiled. */
346 reg_syntax_t syntax;
348 /* Pointer to a fastmap, if any, otherwise zero. re_search uses
349 the fastmap, if there is one, to skip over impossible
350 starting points for matches. */
351 char *fastmap;
353 /* Either a translate table to apply to all characters before
354 comparing them, or zero for no translation. The translation
355 is applied to a pattern when it is compiled and to a string
356 when it is matched. */
357 RE_TRANSLATE_TYPE translate;
359 /* Number of subexpressions found by the compiler. */
360 size_t re_nsub;
362 /* Zero if this pattern cannot match the empty string, one else.
363 Well, in truth it's used only in `re_search_2', to see
364 whether or not we should use the fastmap, so we don't set
365 this absolutely perfectly; see `re_compile_fastmap' (the
366 `duplicate' case). */
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 /* If true, an anchor at a newline matches. */
394 unsigned newline_anchor : 1;
396 /* [[[end pattern_buffer]]] */
399 typedef struct re_pattern_buffer regex_t;
401 /* Type for byte offsets within the string. POSIX mandates this. */
402 typedef int regoff_t;
405 /* This is the structure we store register match data in. See
406 regex.texinfo for a full description of what registers match. */
407 struct re_registers
409 unsigned num_regs;
410 regoff_t *start;
411 regoff_t *end;
415 /* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
416 `re_match_2' returns information about at least this many registers
417 the first time a `regs' structure is passed. */
418 #ifndef RE_NREGS
419 # define RE_NREGS 30
420 #endif
423 /* POSIX specification for registers. Aside from the different names than
424 `re_registers', POSIX uses an array of structures, instead of a
425 structure of arrays. */
426 typedef struct
428 regoff_t rm_so; /* Byte offset from string's start to substring's start. */
429 regoff_t rm_eo; /* Byte offset from string's start to substring's end. */
430 } regmatch_t;
432 /* Declarations for routines. */
434 /* To avoid duplicating every routine declaration -- once with a
435 prototype (if we are ANSI), and once without (if we aren't) -- we
436 use the following macro to declare argument types. This
437 unfortunately clutters up the declarations a bit, but I think it's
438 worth it. */
440 /* Sets the current default syntax to SYNTAX, and return the old syntax.
441 You can also simply assign to the `re_syntax_options' variable. */
442 extern reg_syntax_t re_set_syntax (reg_syntax_t syntax);
444 /* Compile the regular expression PATTERN, with length LENGTH
445 and syntax given by the global `re_syntax_options', into the buffer
446 BUFFER. Return NULL if successful, and an error string if not. */
447 extern const char *re_compile_pattern (const char *pattern, size_t length,
448 struct re_pattern_buffer *buffer);
451 /* Compile a fastmap for the compiled pattern in BUFFER; used to
452 accelerate searches. Return 0 if successful and -2 if was an
453 internal error. */
454 extern int re_compile_fastmap (struct re_pattern_buffer *buffer);
457 /* Search in the string STRING (with length LENGTH) for the pattern
458 compiled into BUFFER. Start searching at position START, for RANGE
459 characters. Return the starting position of the match, -1 for no
460 match, or -2 for an internal error. Also return register
461 information in REGS (if REGS and BUFFER->no_sub are nonzero). */
462 extern int re_search (struct re_pattern_buffer *buffer, const char *string,
463 int length, int start, int range,
464 struct re_registers *regs);
467 /* Like `re_search', but search in the concatenation of STRING1 and
468 STRING2. Also, stop searching at index START + STOP. */
469 extern int re_search_2 (struct re_pattern_buffer *buffer, const char *string1,
470 int length1, const char *string2, int length2,
471 int start, int range, struct re_registers *regs,
472 int stop);
475 /* Like `re_search', but return how many characters in STRING the regexp
476 in BUFFER matched, starting at position START. */
477 extern int re_match (struct re_pattern_buffer *buffer, const char *string,
478 int length, int start, struct re_registers *regs);
481 /* Relates to `re_match' as `re_search_2' relates to `re_search'. */
482 extern int re_match_2 (struct re_pattern_buffer *buffer, const char *string1,
483 int length1, const char *string2, int length2,
484 int start, struct re_registers *regs, int stop);
487 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
488 ENDS. Subsequent matches using BUFFER and REGS will use this memory
489 for recording register information. STARTS and ENDS must be
490 allocated with malloc, and must each be at least `NUM_REGS * sizeof
491 (regoff_t)' bytes long.
493 If NUM_REGS == 0, then subsequent matches should allocate their own
494 register data.
496 Unless this function is called, the first search or match using
497 PATTERN_BUFFER will allocate its own register data, without
498 freeing the old data. */
499 extern void re_set_registers (struct re_pattern_buffer *buffer,
500 struct re_registers *regs,
501 unsigned num_regs, regoff_t *starts,
502 regoff_t *ends);
504 #if defined _REGEX_RE_COMP || defined _LIBC
505 # ifndef _CRAY
506 /* 4.2 bsd compatibility. */
507 extern char *re_comp (const char *);
508 extern int re_exec (const char *);
509 # endif
510 #endif
512 /* GCC 2.95 and later have "__restrict"; C99 compilers have
513 "restrict", and "configure" may have defined "restrict". */
514 #ifndef __restrict
515 # if ! (2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__))
516 # if defined restrict || 199901L <= __STDC_VERSION__
517 # define __restrict restrict
518 # else
519 # define __restrict
520 # endif
521 # endif
522 #endif
524 /* GCC 3.1 and later support declaring arrays as non-overlapping
525 using the syntax array_name[restrict] */
526 #ifndef __restrict_arr
527 # if ! (3 < __GNUC__ || (3 == __GNUC__ && 1 <= __GNUC_MINOR__)) || defined (__GNUG__)
528 # define __restrict_arr
529 # else
530 # define __restrict_arr __restrict
531 # endif
532 #endif
534 /* POSIX compatibility. */
535 extern int regcomp (regex_t *__restrict __preg,
536 const char *__restrict __pattern,
537 int __cflags);
539 #if (__GNUC__)
540 __extension__
541 #endif
542 extern int regexec (const regex_t *__restrict __preg,
543 const char *__restrict __string, size_t __nmatch,
544 regmatch_t __pmatch[__restrict_arr],
545 int __eflags);
547 extern size_t regerror (int __errcode, const regex_t *__preg,
548 char *__errbuf, size_t __errbuf_size);
550 extern void regfree (regex_t *__preg);
553 #ifdef __cplusplus
555 #endif /* C++ */
557 #endif /* regex.h */
560 Local variables:
561 make-backup-files: t
562 version-control: t
563 trim-versions-without-asking: nil
564 End: