HCoop / bpt / emacs.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
(CCL_WRITE_CHAR): Check variable `extra_bytes'.
[bpt/emacs.git] / src / regex.c
1 /* Extended regular expression matching and search library, version
2 0.12. (Implements POSIX draft P1003.2/D11.2, except for some of the
3 internationalization features.)
4
5 Copyright (C) 1993,94,95,96,97,98,99,2000 Free Software Foundation, Inc.
6
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2, or (at your option)
10 any later version.
11
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
16
17 You should have received a copy of the GNU General Public License
18 along with this program; if not, write to the Free Software
19 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
20 USA. */
21
22 /* TODO:
23 - structure the opcode space into opcode+flag.
24 - merge with glibc's regex.[ch].
25 - replace (succeed_n + jump_n + set_number_at) with something that doesn't
26 need to modify the compiled regexp so that re_match can be reentrant.
27 - get rid of on_failure_jump_smart by doing the optimization in re_comp
28 rather than at run-time, so that re_match can be reentrant.
29 */
30
31 /* AIX requires this to be the first thing in the file. */
32 #if defined _AIX && !defined REGEX_MALLOC
33 #pragma alloca
34 #endif
35
36 #undef _GNU_SOURCE
37 #define _GNU_SOURCE
38
39 #ifdef HAVE_CONFIG_H
40 # include <config.h>
41 #endif
42
43 #if defined STDC_HEADERS && !defined emacs
44 # include <stddef.h>
45 #else
46 /* We need this for `regex.h', and perhaps for the Emacs include files. */
47 # include <sys/types.h>
48 #endif
49
50 /* Whether to use ISO C Amendment 1 wide char functions.
51 Those should not be used for Emacs since it uses its own. */
52 #define WIDE_CHAR_SUPPORT \
53 (defined _LIBC || HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC && !emacs)
54
55 /* For platform which support the ISO C amendement 1 functionality we
56 support user defined character classes. */
57 #if WIDE_CHAR_SUPPORT
58 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
59 # include <wchar.h>
60 # include <wctype.h>
61 #endif
62
63 #ifdef _LIBC
64 /* We have to keep the namespace clean. */
65 # define regfree(preg) __regfree (preg)
66 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
67 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
68 # define regerror(errcode, preg, errbuf, errbuf_size) \
69 __regerror(errcode, preg, errbuf, errbuf_size)
70 # define re_set_registers(bu, re, nu, st, en) \
71 __re_set_registers (bu, re, nu, st, en)
72 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
73 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
74 # define re_match(bufp, string, size, pos, regs) \
75 __re_match (bufp, string, size, pos, regs)
76 # define re_search(bufp, string, size, startpos, range, regs) \
77 __re_search (bufp, string, size, startpos, range, regs)
78 # define re_compile_pattern(pattern, length, bufp) \
79 __re_compile_pattern (pattern, length, bufp)
80 # define re_set_syntax(syntax) __re_set_syntax (syntax)
81 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
82 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
83 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
84
85 /* Make sure we call libc's function even if the user overrides them. */
86 # define btowc __btowc
87 # define iswctype __iswctype
88 # define wctype __wctype
89
90 # define WEAK_ALIAS(a,b) weak_alias (a, b)
91
92 /* We are also using some library internals. */
93 # include <locale/localeinfo.h>
94 # include <locale/elem-hash.h>
95 # include <langinfo.h>
96 #else
97 # define WEAK_ALIAS(a,b)
98 #endif
99
100 /* This is for other GNU distributions with internationalized messages. */
101 #if HAVE_LIBINTL_H || defined _LIBC
102 # include <libintl.h>
103 #else
104 # define gettext(msgid) (msgid)
105 #endif
106
107 #ifndef gettext_noop
108 /* This define is so xgettext can find the internationalizable
109 strings. */
110 # define gettext_noop(String) String
111 #endif
112
113 /* The `emacs' switch turns on certain matching commands
114 that make sense only in Emacs. */
115 #ifdef emacs
116
117 # include "lisp.h"
118 # include "buffer.h"
119
120 /* Make syntax table lookup grant data in gl_state. */
121 # define SYNTAX_ENTRY_VIA_PROPERTY
122
123 # include "syntax.h"
124 # include "charset.h"
125 # include "category.h"
126
127 # define malloc xmalloc
128 # define realloc xrealloc
129 # define free xfree
130
131 /* Converts the pointer to the char to BEG-based offset from the start. */
132 # define PTR_TO_OFFSET(d) POS_AS_IN_BUFFER (POINTER_TO_OFFSET (d))
133 # define POS_AS_IN_BUFFER(p) ((p) + (NILP (re_match_object) || BUFFERP (re_match_object)))
134
135 # define RE_MULTIBYTE_P(bufp) ((bufp)->multibyte)
136 # define RE_STRING_CHAR(p, s) \
137 (multibyte ? (STRING_CHAR (p, s)) : (*(p)))
138 # define RE_STRING_CHAR_AND_LENGTH(p, s, len) \
139 (multibyte ? (STRING_CHAR_AND_LENGTH (p, s, len)) : ((len) = 1, *(p)))
140
141 /* Set C a (possibly multibyte) character before P. P points into a
142 string which is the virtual concatenation of STR1 (which ends at
143 END1) or STR2 (which ends at END2). */
144 # define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
145 do { \
146 if (multibyte) \
147 { \
148 re_char *dtemp = (p) == (str2) ? (end1) : (p); \
149 re_char *dlimit = ((p) > (str2) && (p) <= (end2)) ? (str2) : (str1); \
150 while (dtemp-- > dlimit && !CHAR_HEAD_P (*dtemp)); \
151 c = STRING_CHAR (dtemp, (p) - dtemp); \
152 } \
153 else \
154 (c = ((p) == (str2) ? (end1) : (p))[-1]); \
155 } while (0)
156
157
158 #else /* not emacs */
159
160 /* If we are not linking with Emacs proper,
161 we can't use the relocating allocator
162 even if config.h says that we can. */
163 # undef REL_ALLOC
164
165 # if defined STDC_HEADERS || defined _LIBC
166 # include <stdlib.h>
167 # else
168 char *malloc ();
169 char *realloc ();
170 # endif
171
172 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
173 If nothing else has been done, use the method below. */
174 # ifdef INHIBIT_STRING_HEADER
175 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
176 # if !defined bzero && !defined bcopy
177 # undef INHIBIT_STRING_HEADER
178 # endif
179 # endif
180 # endif
181
182 /* This is the normal way of making sure we have memcpy, memcmp and bzero.
183 This is used in most programs--a few other programs avoid this
184 by defining INHIBIT_STRING_HEADER. */
185 # ifndef INHIBIT_STRING_HEADER
186 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
187 # include <string.h>
188 # ifndef bzero
189 # ifndef _LIBC
190 # define bzero(s, n) (memset (s, '\0', n), (s))
191 # else
192 # define bzero(s, n) __bzero (s, n)
193 # endif
194 # endif
195 # else
196 # include <strings.h>
197 # ifndef memcmp
198 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
199 # endif
200 # ifndef memcpy
201 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
202 # endif
203 # endif
204 # endif
205
206 /* Define the syntax stuff for \<, \>, etc. */
207
208 /* Sword must be nonzero for the wordchar pattern commands in re_match_2. */
209 enum syntaxcode { Swhitespace = 0, Sword = 1 };
210
211 # ifdef SWITCH_ENUM_BUG
212 # define SWITCH_ENUM_CAST(x) ((int)(x))
213 # else
214 # define SWITCH_ENUM_CAST(x) (x)
215 # endif
216
217 /* Dummy macros for non-Emacs environments. */
218 # define BASE_LEADING_CODE_P(c) (0)
219 # define CHAR_CHARSET(c) 0
220 # define CHARSET_LEADING_CODE_BASE(c) 0
221 # define MAX_MULTIBYTE_LENGTH 1
222 # define RE_MULTIBYTE_P(x) 0
223 # define WORD_BOUNDARY_P(c1, c2) (0)
224 # define CHAR_HEAD_P(p) (1)
225 # define SINGLE_BYTE_CHAR_P(c) (1)
226 # define SAME_CHARSET_P(c1, c2) (1)
227 # define MULTIBYTE_FORM_LENGTH(p, s) (1)
228 # define STRING_CHAR(p, s) (*(p))
229 # define RE_STRING_CHAR STRING_CHAR
230 # define CHAR_STRING(c, s) (*(s) = (c), 1)
231 # define STRING_CHAR_AND_LENGTH(p, s, actual_len) ((actual_len) = 1, *(p))
232 # define RE_STRING_CHAR_AND_LENGTH STRING_CHAR_AND_LENGTH
233 # define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
234 (c = ((p) == (str2) ? *((end1) - 1) : *((p) - 1)))
235 # define MAKE_CHAR(charset, c1, c2) (c1)
236 #endif /* not emacs */
237
238 #ifndef RE_TRANSLATE
239 # define RE_TRANSLATE(TBL, C) ((unsigned char)(TBL)[C])
240 # define RE_TRANSLATE_P(TBL) (TBL)
241 #endif
242 \f
243 /* Get the interface, including the syntax bits. */
244 #include "regex.h"
245
246 /* isalpha etc. are used for the character classes. */
247 #include <ctype.h>
248
249 #ifdef emacs
250
251 /* 1 if C is an ASCII character. */
252 # define IS_REAL_ASCII(c) ((c) < 0200)
253
254 /* 1 if C is a unibyte character. */
255 # define ISUNIBYTE(c) (SINGLE_BYTE_CHAR_P ((c)))
256
257 /* The Emacs definitions should not be directly affected by locales. */
258
259 /* In Emacs, these are only used for single-byte characters. */
260 # define ISDIGIT(c) ((c) >= '0' && (c) <= '9')
261 # define ISCNTRL(c) ((c) < ' ')
262 # define ISXDIGIT(c) (((c) >= '0' && (c) <= '9') \
263 || ((c) >= 'a' && (c) <= 'f') \
264 || ((c) >= 'A' && (c) <= 'F'))
265
266 /* This is only used for single-byte characters. */
267 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
268
269 /* The rest must handle multibyte characters. */
270
271 # define ISGRAPH(c) (SINGLE_BYTE_CHAR_P (c) \
272 ? (c) > ' ' && !((c) >= 0177 && (c) <= 0237) \
273 : 1)
274
275 # define ISPRINT(c) (SINGLE_BYTE_CHAR_P (c) \
276 ? (c) >= ' ' && !((c) >= 0177 && (c) <= 0237) \
277 : 1)
278
279 # define ISALNUM(c) (IS_REAL_ASCII (c) \
280 ? (((c) >= 'a' && (c) <= 'z') \
281 || ((c) >= 'A' && (c) <= 'Z') \
282 || ((c) >= '0' && (c) <= '9')) \
283 : SYNTAX (c) == Sword)
284
285 # define ISALPHA(c) (IS_REAL_ASCII (c) \
286 ? (((c) >= 'a' && (c) <= 'z') \
287 || ((c) >= 'A' && (c) <= 'Z')) \
288 : SYNTAX (c) == Sword)
289
290 # define ISLOWER(c) (LOWERCASEP (c))
291
292 # define ISPUNCT(c) (IS_REAL_ASCII (c) \
293 ? ((c) > ' ' && (c) < 0177 \
294 && !(((c) >= 'a' && (c) <= 'z') \
295 || ((c) >= 'A' && (c) <= 'Z') \
296 || ((c) >= '0' && (c) <= '9'))) \
297 : SYNTAX (c) != Sword)
298
299 # define ISSPACE(c) (SYNTAX (c) == Swhitespace)
300
301 # define ISUPPER(c) (UPPERCASEP (c))
302
303 # define ISWORD(c) (SYNTAX (c) == Sword)
304
305 #else /* not emacs */
306
307 /* Jim Meyering writes:
308
309 "... Some ctype macros are valid only for character codes that
310 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
311 using /bin/cc or gcc but without giving an ansi option). So, all
312 ctype uses should be through macros like ISPRINT... If
313 STDC_HEADERS is defined, then autoconf has verified that the ctype
314 macros don't need to be guarded with references to isascii. ...
315 Defining isascii to 1 should let any compiler worth its salt
316 eliminate the && through constant folding."
317 Solaris defines some of these symbols so we must undefine them first. */
318
319 # undef ISASCII
320 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
321 # define ISASCII(c) 1
322 # else
323 # define ISASCII(c) isascii(c)
324 # endif
325
326 /* 1 if C is an ASCII character. */
327 # define IS_REAL_ASCII(c) ((c) < 0200)
328
329 /* This distinction is not meaningful, except in Emacs. */
330 # define ISUNIBYTE(c) 1
331
332 # ifdef isblank
333 # define ISBLANK(c) (ISASCII (c) && isblank (c))
334 # else
335 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
336 # endif
337 # ifdef isgraph
338 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
339 # else
340 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
341 # endif
342
343 # undef ISPRINT
344 # define ISPRINT(c) (ISASCII (c) && isprint (c))
345 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
346 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
347 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
348 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
349 # define ISLOWER(c) (ISASCII (c) && islower (c))
350 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
351 # define ISSPACE(c) (ISASCII (c) && isspace (c))
352 # define ISUPPER(c) (ISASCII (c) && isupper (c))
353 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
354
355 # define ISWORD(c) ISALPHA(c)
356
357 # ifdef _tolower
358 # define TOLOWER(c) _tolower(c)
359 # else
360 # define TOLOWER(c) tolower(c)
361 # endif
362
363 /* How many characters in the character set. */
364 # define CHAR_SET_SIZE 256
365
366 # ifdef SYNTAX_TABLE
367
368 extern char *re_syntax_table;
369
370 # else /* not SYNTAX_TABLE */
371
372 static char re_syntax_table[CHAR_SET_SIZE];
373
374 static void
375 init_syntax_once ()
376 {
377 register int c;
378 static int done = 0;
379
380 if (done)
381 return;
382
383 bzero (re_syntax_table, sizeof re_syntax_table);
384
385 for (c = 0; c < CHAR_SET_SIZE; ++c)
386 if (ISALNUM (c))
387 re_syntax_table[c] = Sword;
388
389 re_syntax_table['_'] = Sword;
390
391 done = 1;
392 }
393
394 # endif /* not SYNTAX_TABLE */
395
396 # define SYNTAX(c) re_syntax_table[(c)]
397
398 #endif /* not emacs */
399 \f
400 #ifndef NULL
401 # define NULL (void *)0
402 #endif
403
404 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
405 since ours (we hope) works properly with all combinations of
406 machines, compilers, `char' and `unsigned char' argument types.
407 (Per Bothner suggested the basic approach.) */
408 #undef SIGN_EXTEND_CHAR
409 #if __STDC__
410 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
411 #else /* not __STDC__ */
412 /* As in Harbison and Steele. */
413 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
414 #endif
415 \f
416 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
417 use `alloca' instead of `malloc'. This is because using malloc in
418 re_search* or re_match* could cause memory leaks when C-g is used in
419 Emacs; also, malloc is slower and causes storage fragmentation. On
420 the other hand, malloc is more portable, and easier to debug.
421
422 Because we sometimes use alloca, some routines have to be macros,
423 not functions -- `alloca'-allocated space disappears at the end of the
424 function it is called in. */
425
426 #ifdef REGEX_MALLOC
427
428 # define REGEX_ALLOCATE malloc
429 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
430 # define REGEX_FREE free
431
432 #else /* not REGEX_MALLOC */
433
434 /* Emacs already defines alloca, sometimes. */
435 # ifndef alloca
436
437 /* Make alloca work the best possible way. */
438 # ifdef __GNUC__
439 # define alloca __builtin_alloca
440 # else /* not __GNUC__ */
441 # if HAVE_ALLOCA_H
442 # include <alloca.h>
443 # endif /* HAVE_ALLOCA_H */
444 # endif /* not __GNUC__ */
445
446 # endif /* not alloca */
447
448 # define REGEX_ALLOCATE alloca
449
450 /* Assumes a `char *destination' variable. */
451 # define REGEX_REALLOCATE(source, osize, nsize) \
452 (destination = (char *) alloca (nsize), \
453 memcpy (destination, source, osize))
454
455 /* No need to do anything to free, after alloca. */
456 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
457
458 #endif /* not REGEX_MALLOC */
459
460 /* Define how to allocate the failure stack. */
461
462 #if defined REL_ALLOC && defined REGEX_MALLOC
463
464 # define REGEX_ALLOCATE_STACK(size) \
465 r_alloc (&failure_stack_ptr, (size))
466 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
467 r_re_alloc (&failure_stack_ptr, (nsize))
468 # define REGEX_FREE_STACK(ptr) \
469 r_alloc_free (&failure_stack_ptr)
470
471 #else /* not using relocating allocator */
472
473 # ifdef REGEX_MALLOC
474
475 # define REGEX_ALLOCATE_STACK malloc
476 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
477 # define REGEX_FREE_STACK free
478
479 # else /* not REGEX_MALLOC */
480
481 # define REGEX_ALLOCATE_STACK alloca
482
483 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
484 REGEX_REALLOCATE (source, osize, nsize)
485 /* No need to explicitly free anything. */
486 # define REGEX_FREE_STACK(arg) ((void)0)
487
488 # endif /* not REGEX_MALLOC */
489 #endif /* not using relocating allocator */
490
491
492 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
493 `string1' or just past its end. This works if PTR is NULL, which is
494 a good thing. */
495 #define FIRST_STRING_P(ptr) \
496 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
497
498 /* (Re)Allocate N items of type T using malloc, or fail. */
499 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
500 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
501 #define RETALLOC_IF(addr, n, t) \
502 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
503 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
504
505 #define BYTEWIDTH 8 /* In bits. */
506
507 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
508
509 #undef MAX
510 #undef MIN
511 #define MAX(a, b) ((a) > (b) ? (a) : (b))
512 #define MIN(a, b) ((a) < (b) ? (a) : (b))
513
514 /* Type of source-pattern and string chars. */
515 typedef const unsigned char re_char;
516
517 typedef char boolean;
518 #define false 0
519 #define true 1
520
521 static int re_match_2_internal _RE_ARGS ((struct re_pattern_buffer *bufp,
522 re_char *string1, int size1,
523 re_char *string2, int size2,
524 int pos,
525 struct re_registers *regs,
526 int stop));
527 \f
528 /* These are the command codes that appear in compiled regular
529 expressions. Some opcodes are followed by argument bytes. A
530 command code can specify any interpretation whatsoever for its
531 arguments. Zero bytes may appear in the compiled regular expression. */
532
533 typedef enum
534 {
535 no_op = 0,
536
537 /* Succeed right away--no more backtracking. */
538 succeed,
539
540 /* Followed by one byte giving n, then by n literal bytes. */
541 exactn,
542
543 /* Matches any (more or less) character. */
544 anychar,
545
546 /* Matches any one char belonging to specified set. First
547 following byte is number of bitmap bytes. Then come bytes
548 for a bitmap saying which chars are in. Bits in each byte
549 are ordered low-bit-first. A character is in the set if its
550 bit is 1. A character too large to have a bit in the map is
551 automatically not in the set.
552
553 If the length byte has the 0x80 bit set, then that stuff
554 is followed by a range table:
555 2 bytes of flags for character sets (low 8 bits, high 8 bits)
556 See RANGE_TABLE_WORK_BITS below.
557 2 bytes, the number of pairs that follow (upto 32767)
558 pairs, each 2 multibyte characters,
559 each multibyte character represented as 3 bytes. */
560 charset,
561
562 /* Same parameters as charset, but match any character that is
563 not one of those specified. */
564 charset_not,
565
566 /* Start remembering the text that is matched, for storing in a
567 register. Followed by one byte with the register number, in
568 the range 0 to one less than the pattern buffer's re_nsub
569 field. */
570 start_memory,
571
572 /* Stop remembering the text that is matched and store it in a
573 memory register. Followed by one byte with the register
574 number, in the range 0 to one less than `re_nsub' in the
575 pattern buffer. */
576 stop_memory,
577
578 /* Match a duplicate of something remembered. Followed by one
579 byte containing the register number. */
580 duplicate,
581
582 /* Fail unless at beginning of line. */
583 begline,
584
585 /* Fail unless at end of line. */
586 endline,
587
588 /* Succeeds if at beginning of buffer (if emacs) or at beginning
589 of string to be matched (if not). */
590 begbuf,
591
592 /* Analogously, for end of buffer/string. */
593 endbuf,
594
595 /* Followed by two byte relative address to which to jump. */
596 jump,
597
598 /* Followed by two-byte relative address of place to resume at
599 in case of failure. */
600 on_failure_jump,
601
602 /* Like on_failure_jump, but pushes a placeholder instead of the
603 current string position when executed. */
604 on_failure_keep_string_jump,
605
606 /* Just like `on_failure_jump', except that it checks that we
607 don't get stuck in an infinite loop (matching an empty string
608 indefinitely). */
609 on_failure_jump_loop,
610
611 /* Just like `on_failure_jump_loop', except that it checks for
612 a different kind of loop (the kind that shows up with non-greedy
613 operators). This operation has to be immediately preceded
614 by a `no_op'. */
615 on_failure_jump_nastyloop,
616
617 /* A smart `on_failure_jump' used for greedy * and + operators.
618 It analyses the loop before which it is put and if the
619 loop does not require backtracking, it changes itself to
620 `on_failure_keep_string_jump' and short-circuits the loop,
621 else it just defaults to changing itself into `on_failure_jump'.
622 It assumes that it is pointing to just past a `jump'. */
623 on_failure_jump_smart,
624
625 /* Followed by two-byte relative address and two-byte number n.
626 After matching N times, jump to the address upon failure.
627 Does not work if N starts at 0: use on_failure_jump_loop
628 instead. */
629 succeed_n,
630
631 /* Followed by two-byte relative address, and two-byte number n.
632 Jump to the address N times, then fail. */
633 jump_n,
634
635 /* Set the following two-byte relative address to the
636 subsequent two-byte number. The address *includes* the two
637 bytes of number. */
638 set_number_at,
639
640 wordbeg, /* Succeeds if at word beginning. */
641 wordend, /* Succeeds if at word end. */
642
643 wordbound, /* Succeeds if at a word boundary. */
644 notwordbound, /* Succeeds if not at a word boundary. */
645
646 /* Matches any character whose syntax is specified. Followed by
647 a byte which contains a syntax code, e.g., Sword. */
648 syntaxspec,
649
650 /* Matches any character whose syntax is not that specified. */
651 notsyntaxspec
652
653 #ifdef emacs
654 ,before_dot, /* Succeeds if before point. */
655 at_dot, /* Succeeds if at point. */
656 after_dot, /* Succeeds if after point. */
657
658 /* Matches any character whose category-set contains the specified
659 category. The operator is followed by a byte which contains a
660 category code (mnemonic ASCII character). */
661 categoryspec,
662
663 /* Matches any character whose category-set does not contain the
664 specified category. The operator is followed by a byte which
665 contains the category code (mnemonic ASCII character). */
666 notcategoryspec
667 #endif /* emacs */
668 } re_opcode_t;
669 \f
670 /* Common operations on the compiled pattern. */
671
672 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
673
674 #define STORE_NUMBER(destination, number) \
675 do { \
676 (destination)[0] = (number) & 0377; \
677 (destination)[1] = (number) >> 8; \
678 } while (0)
679
680 /* Same as STORE_NUMBER, except increment DESTINATION to
681 the byte after where the number is stored. Therefore, DESTINATION
682 must be an lvalue. */
683
684 #define STORE_NUMBER_AND_INCR(destination, number) \
685 do { \
686 STORE_NUMBER (destination, number); \
687 (destination) += 2; \
688 } while (0)
689
690 /* Put into DESTINATION a number stored in two contiguous bytes starting
691 at SOURCE. */
692
693 #define EXTRACT_NUMBER(destination, source) \
694 do { \
695 (destination) = *(source) & 0377; \
696 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
697 } while (0)
698
699 #ifdef DEBUG
700 static void extract_number _RE_ARGS ((int *dest, re_char *source));
701 static void
702 extract_number (dest, source)
703 int *dest;
704 re_char *source;
705 {
706 int temp = SIGN_EXTEND_CHAR (*(source + 1));
707 *dest = *source & 0377;
708 *dest += temp << 8;
709 }
710
711 # ifndef EXTRACT_MACROS /* To debug the macros. */
712 # undef EXTRACT_NUMBER
713 # define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
714 # endif /* not EXTRACT_MACROS */
715
716 #endif /* DEBUG */
717
718 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
719 SOURCE must be an lvalue. */
720
721 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
722 do { \
723 EXTRACT_NUMBER (destination, source); \
724 (source) += 2; \
725 } while (0)
726
727 #ifdef DEBUG
728 static void extract_number_and_incr _RE_ARGS ((int *destination,
729 re_char **source));
730 static void
731 extract_number_and_incr (destination, source)
732 int *destination;
733 re_char **source;
734 {
735 extract_number (destination, *source);
736 *source += 2;
737 }
738
739 # ifndef EXTRACT_MACROS
740 # undef EXTRACT_NUMBER_AND_INCR
741 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
742 extract_number_and_incr (&dest, &src)
743 # endif /* not EXTRACT_MACROS */
744
745 #endif /* DEBUG */
746 \f
747 /* Store a multibyte character in three contiguous bytes starting
748 DESTINATION, and increment DESTINATION to the byte after where the
749 character is stored. Therefore, DESTINATION must be an lvalue. */
750
751 #define STORE_CHARACTER_AND_INCR(destination, character) \
752 do { \
753 (destination)[0] = (character) & 0377; \
754 (destination)[1] = ((character) >> 8) & 0377; \
755 (destination)[2] = (character) >> 16; \
756 (destination) += 3; \
757 } while (0)
758
759 /* Put into DESTINATION a character stored in three contiguous bytes
760 starting at SOURCE. */
761
762 #define EXTRACT_CHARACTER(destination, source) \
763 do { \
764 (destination) = ((source)[0] \
765 | ((source)[1] << 8) \
766 | ((source)[2] << 16)); \
767 } while (0)
768
769
770 /* Macros for charset. */
771
772 /* Size of bitmap of charset P in bytes. P is a start of charset,
773 i.e. *P is (re_opcode_t) charset or (re_opcode_t) charset_not. */
774 #define CHARSET_BITMAP_SIZE(p) ((p)[1] & 0x7F)
775
776 /* Nonzero if charset P has range table. */
777 #define CHARSET_RANGE_TABLE_EXISTS_P(p) ((p)[1] & 0x80)
778
779 /* Return the address of range table of charset P. But not the start
780 of table itself, but the before where the number of ranges is
781 stored. `2 +' means to skip re_opcode_t and size of bitmap,
782 and the 2 bytes of flags at the start of the range table. */
783 #define CHARSET_RANGE_TABLE(p) (&(p)[4 + CHARSET_BITMAP_SIZE (p)])
784
785 /* Extract the bit flags that start a range table. */
786 #define CHARSET_RANGE_TABLE_BITS(p) \
787 ((p)[2 + CHARSET_BITMAP_SIZE (p)] \
788 + (p)[3 + CHARSET_BITMAP_SIZE (p)] * 0x100)
789
790 /* Test if C is listed in the bitmap of charset P. */
791 #define CHARSET_LOOKUP_BITMAP(p, c) \
792 ((c) < CHARSET_BITMAP_SIZE (p) * BYTEWIDTH \
793 && (p)[2 + (c) / BYTEWIDTH] & (1 << ((c) % BYTEWIDTH)))
794
795 /* Return the address of end of RANGE_TABLE. COUNT is number of
796 ranges (which is a pair of (start, end)) in the RANGE_TABLE. `* 2'
797 is start of range and end of range. `* 3' is size of each start
798 and end. */
799 #define CHARSET_RANGE_TABLE_END(range_table, count) \
800 ((range_table) + (count) * 2 * 3)
801
802 /* Test if C is in RANGE_TABLE. A flag NOT is negated if C is in.
803 COUNT is number of ranges in RANGE_TABLE. */
804 #define CHARSET_LOOKUP_RANGE_TABLE_RAW(not, c, range_table, count) \
805 do \
806 { \
807 re_wchar_t range_start, range_end; \
808 re_char *p; \
809 re_char *range_table_end \
810 = CHARSET_RANGE_TABLE_END ((range_table), (count)); \
811 \
812 for (p = (range_table); p < range_table_end; p += 2 * 3) \
813 { \
814 EXTRACT_CHARACTER (range_start, p); \
815 EXTRACT_CHARACTER (range_end, p + 3); \
816 \
817 if (range_start <= (c) && (c) <= range_end) \
818 { \
819 (not) = !(not); \
820 break; \
821 } \
822 } \
823 } \
824 while (0)
825
826 /* Test if C is in range table of CHARSET. The flag NOT is negated if
827 C is listed in it. */
828 #define CHARSET_LOOKUP_RANGE_TABLE(not, c, charset) \
829 do \
830 { \
831 /* Number of ranges in range table. */ \
832 int count; \
833 re_char *range_table = CHARSET_RANGE_TABLE (charset); \
834 \
835 EXTRACT_NUMBER_AND_INCR (count, range_table); \
836 CHARSET_LOOKUP_RANGE_TABLE_RAW ((not), (c), range_table, count); \
837 } \
838 while (0)
839 \f
840 /* If DEBUG is defined, Regex prints many voluminous messages about what
841 it is doing (if the variable `debug' is nonzero). If linked with the
842 main program in `iregex.c', you can enter patterns and strings
843 interactively. And if linked with the main program in `main.c' and
844 the other test files, you can run the already-written tests. */
845
846 #ifdef DEBUG
847
848 /* We use standard I/O for debugging. */
849 # include <stdio.h>
850
851 /* It is useful to test things that ``must'' be true when debugging. */
852 # include <assert.h>
853
854 static int debug = -100000;
855
856 # define DEBUG_STATEMENT(e) e
857 # define DEBUG_PRINT1(x) if (debug > 0) printf (x)
858 # define DEBUG_PRINT2(x1, x2) if (debug > 0) printf (x1, x2)
859 # define DEBUG_PRINT3(x1, x2, x3) if (debug > 0) printf (x1, x2, x3)
860 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug > 0) printf (x1, x2, x3, x4)
861 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
862 if (debug > 0) print_partial_compiled_pattern (s, e)
863 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
864 if (debug > 0) print_double_string (w, s1, sz1, s2, sz2)
865
866
867 /* Print the fastmap in human-readable form. */
868
869 void
870 print_fastmap (fastmap)
871 char *fastmap;
872 {
873 unsigned was_a_range = 0;
874 unsigned i = 0;
875
876 while (i < (1 << BYTEWIDTH))
877 {
878 if (fastmap[i++])
879 {
880 was_a_range = 0;
881 putchar (i - 1);
882 while (i < (1 << BYTEWIDTH) && fastmap[i])
883 {
884 was_a_range = 1;
885 i++;
886 }
887 if (was_a_range)
888 {
889 printf ("-");
890 putchar (i - 1);
891 }
892 }
893 }
894 putchar ('\n');
895 }
896
897
898 /* Print a compiled pattern string in human-readable form, starting at
899 the START pointer into it and ending just before the pointer END. */
900
901 void
902 print_partial_compiled_pattern (start, end)
903 re_char *start;
904 re_char *end;
905 {
906 int mcnt, mcnt2;
907 re_char *p = start;
908 re_char *pend = end;
909
910 if (start == NULL)
911 {
912 printf ("(null)\n");
913 return;
914 }
915
916 /* Loop over pattern commands. */
917 while (p < pend)
918 {
919 printf ("%d:\t", p - start);
920
921 switch ((re_opcode_t) *p++)
922 {
923 case no_op:
924 printf ("/no_op");
925 break;
926
927 case succeed:
928 printf ("/succeed");
929 break;
930
931 case exactn:
932 mcnt = *p++;
933 printf ("/exactn/%d", mcnt);
934 do
935 {
936 putchar ('/');
937 putchar (*p++);
938 }
939 while (--mcnt);
940 break;
941
942 case start_memory:
943 printf ("/start_memory/%d", *p++);
944 break;
945
946 case stop_memory:
947 printf ("/stop_memory/%d", *p++);
948 break;
949
950 case duplicate:
951 printf ("/duplicate/%d", *p++);
952 break;
953
954 case anychar:
955 printf ("/anychar");
956 break;
957
958 case charset:
959 case charset_not:
960 {
961 register int c, last = -100;
962 register int in_range = 0;
963 int length = CHARSET_BITMAP_SIZE (p - 1);
964 int has_range_table = CHARSET_RANGE_TABLE_EXISTS_P (p - 1);
965
966 printf ("/charset [%s",
967 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
968
969 assert (p + *p < pend);
970
971 for (c = 0; c < 256; c++)
972 if (c / 8 < length
973 && (p[1 + (c/8)] & (1 << (c % 8))))
974 {
975 /* Are we starting a range? */
976 if (last + 1 == c && ! in_range)
977 {
978 putchar ('-');
979 in_range = 1;
980 }
981 /* Have we broken a range? */
982 else if (last + 1 != c && in_range)
983 {
984 putchar (last);
985 in_range = 0;
986 }
987
988 if (! in_range)
989 putchar (c);
990
991 last = c;
992 }
993
994 if (in_range)
995 putchar (last);
996
997 putchar (']');
998
999 p += 1 + length;
1000
1001 if (has_range_table)
1002 {
1003 int count;
1004 printf ("has-range-table");
1005
1006 /* ??? Should print the range table; for now, just skip it. */
1007 p += 2; /* skip range table bits */
1008 EXTRACT_NUMBER_AND_INCR (count, p);
1009 p = CHARSET_RANGE_TABLE_END (p, count);
1010 }
1011 }
1012 break;
1013
1014 case begline:
1015 printf ("/begline");
1016 break;
1017
1018 case endline:
1019 printf ("/endline");
1020 break;
1021
1022 case on_failure_jump:
1023 extract_number_and_incr (&mcnt, &p);
1024 printf ("/on_failure_jump to %d", p + mcnt - start);
1025 break;
1026
1027 case on_failure_keep_string_jump:
1028 extract_number_and_incr (&mcnt, &p);
1029 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
1030 break;
1031
1032 case on_failure_jump_nastyloop:
1033 extract_number_and_incr (&mcnt, &p);
1034 printf ("/on_failure_jump_nastyloop to %d", p + mcnt - start);
1035 break;
1036
1037 case on_failure_jump_loop:
1038 extract_number_and_incr (&mcnt, &p);
1039 printf ("/on_failure_jump_loop to %d", p + mcnt - start);
1040 break;
1041
1042 case on_failure_jump_smart:
1043 extract_number_and_incr (&mcnt, &p);
1044 printf ("/on_failure_jump_smart to %d", p + mcnt - start);
1045 break;
1046
1047 case jump:
1048 extract_number_and_incr (&mcnt, &p);
1049 printf ("/jump to %d", p + mcnt - start);
1050 break;
1051
1052 case succeed_n:
1053 extract_number_and_incr (&mcnt, &p);
1054 extract_number_and_incr (&mcnt2, &p);
1055 printf ("/succeed_n to %d, %d times", p - 2 + mcnt - start, mcnt2);
1056 break;
1057
1058 case jump_n:
1059 extract_number_and_incr (&mcnt, &p);
1060 extract_number_and_incr (&mcnt2, &p);
1061 printf ("/jump_n to %d, %d times", p - 2 + mcnt - start, mcnt2);
1062 break;
1063
1064 case set_number_at:
1065 extract_number_and_incr (&mcnt, &p);
1066 extract_number_and_incr (&mcnt2, &p);
1067 printf ("/set_number_at location %d to %d", p - 2 + mcnt - start, mcnt2);
1068 break;
1069
1070 case wordbound:
1071 printf ("/wordbound");
1072 break;
1073
1074 case notwordbound:
1075 printf ("/notwordbound");
1076 break;
1077
1078 case wordbeg:
1079 printf ("/wordbeg");
1080 break;
1081
1082 case wordend:
1083 printf ("/wordend");
1084
1085 case syntaxspec:
1086 printf ("/syntaxspec");
1087 mcnt = *p++;
1088 printf ("/%d", mcnt);
1089 break;
1090
1091 case notsyntaxspec:
1092 printf ("/notsyntaxspec");
1093 mcnt = *p++;
1094 printf ("/%d", mcnt);
1095 break;
1096
1097 # ifdef emacs
1098 case before_dot:
1099 printf ("/before_dot");
1100 break;
1101
1102 case at_dot:
1103 printf ("/at_dot");
1104 break;
1105
1106 case after_dot:
1107 printf ("/after_dot");
1108 break;
1109
1110 case categoryspec:
1111 printf ("/categoryspec");
1112 mcnt = *p++;
1113 printf ("/%d", mcnt);
1114 break;
1115
1116 case notcategoryspec:
1117 printf ("/notcategoryspec");
1118 mcnt = *p++;
1119 printf ("/%d", mcnt);
1120 break;
1121 # endif /* emacs */
1122
1123 case begbuf:
1124 printf ("/begbuf");
1125 break;
1126
1127 case endbuf:
1128 printf ("/endbuf");
1129 break;
1130
1131 default:
1132 printf ("?%d", *(p-1));
1133 }
1134
1135 putchar ('\n');
1136 }
1137
1138 printf ("%d:\tend of pattern.\n", p - start);
1139 }
1140
1141
1142 void
1143 print_compiled_pattern (bufp)
1144 struct re_pattern_buffer *bufp;
1145 {
1146 re_char *buffer = bufp->buffer;
1147
1148 print_partial_compiled_pattern (buffer, buffer + bufp->used);
1149 printf ("%ld bytes used/%ld bytes allocated.\n",
1150 bufp->used, bufp->allocated);
1151
1152 if (bufp->fastmap_accurate && bufp->fastmap)
1153 {
1154 printf ("fastmap: ");
1155 print_fastmap (bufp->fastmap);
1156 }
1157
1158 printf ("re_nsub: %d\t", bufp->re_nsub);
1159 printf ("regs_alloc: %d\t", bufp->regs_allocated);
1160 printf ("can_be_null: %d\t", bufp->can_be_null);
1161 printf ("no_sub: %d\t", bufp->no_sub);
1162 printf ("not_bol: %d\t", bufp->not_bol);
1163 printf ("not_eol: %d\t", bufp->not_eol);
1164 printf ("syntax: %lx\n", bufp->syntax);
1165 fflush (stdout);
1166 /* Perhaps we should print the translate table? */
1167 }
1168
1169
1170 void
1171 print_double_string (where, string1, size1, string2, size2)
1172 re_char *where;
1173 re_char *string1;
1174 re_char *string2;
1175 int size1;
1176 int size2;
1177 {
1178 int this_char;
1179
1180 if (where == NULL)
1181 printf ("(null)");
1182 else
1183 {
1184 if (FIRST_STRING_P (where))
1185 {
1186 for (this_char = where - string1; this_char < size1; this_char++)
1187 putchar (string1[this_char]);
1188
1189 where = string2;
1190 }
1191
1192 for (this_char = where - string2; this_char < size2; this_char++)
1193 putchar (string2[this_char]);
1194 }
1195 }
1196
1197 #else /* not DEBUG */
1198
1199 # undef assert
1200 # define assert(e)
1201
1202 # define DEBUG_STATEMENT(e)
1203 # define DEBUG_PRINT1(x)
1204 # define DEBUG_PRINT2(x1, x2)
1205 # define DEBUG_PRINT3(x1, x2, x3)
1206 # define DEBUG_PRINT4(x1, x2, x3, x4)
1207 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1208 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1209
1210 #endif /* not DEBUG */
1211 \f
1212 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1213 also be assigned to arbitrarily: each pattern buffer stores its own
1214 syntax, so it can be changed between regex compilations. */
1215 /* This has no initializer because initialized variables in Emacs
1216 become read-only after dumping. */
1217 reg_syntax_t re_syntax_options;
1218
1219
1220 /* Specify the precise syntax of regexps for compilation. This provides
1221 for compatibility for various utilities which historically have
1222 different, incompatible syntaxes.
1223
1224 The argument SYNTAX is a bit mask comprised of the various bits
1225 defined in regex.h. We return the old syntax. */
1226
1227 reg_syntax_t
1228 re_set_syntax (syntax)
1229 reg_syntax_t syntax;
1230 {
1231 reg_syntax_t ret = re_syntax_options;
1232
1233 re_syntax_options = syntax;
1234 return ret;
1235 }
1236 WEAK_ALIAS (__re_set_syntax, re_set_syntax)
1237 \f
1238 /* This table gives an error message for each of the error codes listed
1239 in regex.h. Obviously the order here has to be same as there.
1240 POSIX doesn't require that we do anything for REG_NOERROR,
1241 but why not be nice? */
1242
1243 static const char *re_error_msgid[] =
1244 {
1245 gettext_noop ("Success"), /* REG_NOERROR */
1246 gettext_noop ("No match"), /* REG_NOMATCH */
1247 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
1248 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
1249 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
1250 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
1251 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
1252 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
1253 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
1254 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
1255 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
1256 gettext_noop ("Invalid range end"), /* REG_ERANGE */
1257 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
1258 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
1259 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
1260 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
1261 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
1262 };
1263 \f
1264 /* Avoiding alloca during matching, to placate r_alloc. */
1265
1266 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1267 searching and matching functions should not call alloca. On some
1268 systems, alloca is implemented in terms of malloc, and if we're
1269 using the relocating allocator routines, then malloc could cause a
1270 relocation, which might (if the strings being searched are in the
1271 ralloc heap) shift the data out from underneath the regexp
1272 routines.
1273
1274 Here's another reason to avoid allocation: Emacs
1275 processes input from X in a signal handler; processing X input may
1276 call malloc; if input arrives while a matching routine is calling
1277 malloc, then we're scrod. But Emacs can't just block input while
1278 calling matching routines; then we don't notice interrupts when
1279 they come in. So, Emacs blocks input around all regexp calls
1280 except the matching calls, which it leaves unprotected, in the
1281 faith that they will not malloc. */
1282
1283 /* Normally, this is fine. */
1284 #define MATCH_MAY_ALLOCATE
1285
1286 /* When using GNU C, we are not REALLY using the C alloca, no matter
1287 what config.h may say. So don't take precautions for it. */
1288 #ifdef __GNUC__
1289 # undef C_ALLOCA
1290 #endif
1291
1292 /* The match routines may not allocate if (1) they would do it with malloc
1293 and (2) it's not safe for them to use malloc.
1294 Note that if REL_ALLOC is defined, matching would not use malloc for the
1295 failure stack, but we would still use it for the register vectors;
1296 so REL_ALLOC should not affect this. */
1297 #if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1298 # undef MATCH_MAY_ALLOCATE
1299 #endif
1300
1301 \f
1302 /* Failure stack declarations and macros; both re_compile_fastmap and
1303 re_match_2 use a failure stack. These have to be macros because of
1304 REGEX_ALLOCATE_STACK. */
1305
1306
1307 /* Approximate number of failure points for which to initially allocate space
1308 when matching. If this number is exceeded, we allocate more
1309 space, so it is not a hard limit. */
1310 #ifndef INIT_FAILURE_ALLOC
1311 # define INIT_FAILURE_ALLOC 20
1312 #endif
1313
1314 /* Roughly the maximum number of failure points on the stack. Would be
1315 exactly that if always used TYPICAL_FAILURE_SIZE items each time we failed.
1316 This is a variable only so users of regex can assign to it; we never
1317 change it ourselves. */
1318 # if defined MATCH_MAY_ALLOCATE
1319 /* Note that 4400 was enough to cause a crash on Alpha OSF/1,
1320 whose default stack limit is 2mb. In order for a larger
1321 value to work reliably, you have to try to make it accord
1322 with the process stack limit. */
1323 size_t re_max_failures = 40000;
1324 # else
1325 size_t re_max_failures = 4000;
1326 # endif
1327
1328 union fail_stack_elt
1329 {
1330 re_char *pointer;
1331 /* This should be the biggest `int' that's no bigger than a pointer. */
1332 long integer;
1333 };
1334
1335 typedef union fail_stack_elt fail_stack_elt_t;
1336
1337 typedef struct
1338 {
1339 fail_stack_elt_t *stack;
1340 size_t size;
1341 size_t avail; /* Offset of next open position. */
1342 size_t frame; /* Offset of the cur constructed frame. */
1343 } fail_stack_type;
1344
1345 #define FAIL_STACK_EMPTY() (fail_stack.frame == 0)
1346 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1347
1348
1349 /* Define macros to initialize and free the failure stack.
1350 Do `return -2' if the alloc fails. */
1351
1352 #ifdef MATCH_MAY_ALLOCATE
1353 # define INIT_FAIL_STACK() \
1354 do { \
1355 fail_stack.stack = (fail_stack_elt_t *) \
1356 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * TYPICAL_FAILURE_SIZE \
1357 * sizeof (fail_stack_elt_t)); \
1358 \
1359 if (fail_stack.stack == NULL) \
1360 return -2; \
1361 \
1362 fail_stack.size = INIT_FAILURE_ALLOC; \
1363 fail_stack.avail = 0; \
1364 fail_stack.frame = 0; \
1365 } while (0)
1366
1367 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1368 #else
1369 # define INIT_FAIL_STACK() \
1370 do { \
1371 fail_stack.avail = 0; \
1372 fail_stack.frame = 0; \
1373 } while (0)
1374
1375 # define RESET_FAIL_STACK() ((void)0)
1376 #endif
1377
1378
1379 /* Double the size of FAIL_STACK, up to a limit
1380 which allows approximately `re_max_failures' items.
1381
1382 Return 1 if succeeds, and 0 if either ran out of memory
1383 allocating space for it or it was already too large.
1384
1385 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1386
1387 /* Factor to increase the failure stack size by
1388 when we increase it.
1389 This used to be 2, but 2 was too wasteful
1390 because the old discarded stacks added up to as much space
1391 were as ultimate, maximum-size stack. */
1392 #define FAIL_STACK_GROWTH_FACTOR 4
1393
1394 #define GROW_FAIL_STACK(fail_stack) \
1395 (((fail_stack).size * sizeof (fail_stack_elt_t) \
1396 >= re_max_failures * TYPICAL_FAILURE_SIZE) \
1397 ? 0 \
1398 : ((fail_stack).stack \
1399 = (fail_stack_elt_t *) \
1400 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1401 (fail_stack).size * sizeof (fail_stack_elt_t), \
1402 MIN (re_max_failures * TYPICAL_FAILURE_SIZE, \
1403 ((fail_stack).size * sizeof (fail_stack_elt_t) \
1404 * FAIL_STACK_GROWTH_FACTOR))), \
1405 \
1406 (fail_stack).stack == NULL \
1407 ? 0 \
1408 : ((fail_stack).size \
1409 = (MIN (re_max_failures * TYPICAL_FAILURE_SIZE, \
1410 ((fail_stack).size * sizeof (fail_stack_elt_t) \
1411 * FAIL_STACK_GROWTH_FACTOR)) \
1412 / sizeof (fail_stack_elt_t)), \
1413 1)))
1414
1415
1416 /* Push a pointer value onto the failure stack.
1417 Assumes the variable `fail_stack'. Probably should only
1418 be called from within `PUSH_FAILURE_POINT'. */
1419 #define PUSH_FAILURE_POINTER(item) \
1420 fail_stack.stack[fail_stack.avail++].pointer = (item)
1421
1422 /* This pushes an integer-valued item onto the failure stack.
1423 Assumes the variable `fail_stack'. Probably should only
1424 be called from within `PUSH_FAILURE_POINT'. */
1425 #define PUSH_FAILURE_INT(item) \
1426 fail_stack.stack[fail_stack.avail++].integer = (item)
1427
1428 /* Push a fail_stack_elt_t value onto the failure stack.
1429 Assumes the variable `fail_stack'. Probably should only
1430 be called from within `PUSH_FAILURE_POINT'. */
1431 #define PUSH_FAILURE_ELT(item) \
1432 fail_stack.stack[fail_stack.avail++] = (item)
1433
1434 /* These three POP... operations complement the three PUSH... operations.
1435 All assume that `fail_stack' is nonempty. */
1436 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1437 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1438 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1439
1440 /* Individual items aside from the registers. */
1441 #define NUM_NONREG_ITEMS 3
1442
1443 /* Used to examine the stack (to detect infinite loops). */
1444 #define FAILURE_PAT(h) fail_stack.stack[(h) - 1].pointer
1445 #define FAILURE_STR(h) (fail_stack.stack[(h) - 2].pointer)
1446 #define NEXT_FAILURE_HANDLE(h) fail_stack.stack[(h) - 3].integer
1447 #define TOP_FAILURE_HANDLE() fail_stack.frame
1448
1449
1450 #define ENSURE_FAIL_STACK(space) \
1451 while (REMAINING_AVAIL_SLOTS <= space) { \
1452 if (!GROW_FAIL_STACK (fail_stack)) \
1453 return -2; \
1454 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", (fail_stack).size);\
1455 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1456 }
1457
1458 /* Push register NUM onto the stack. */
1459 #define PUSH_FAILURE_REG(num) \
1460 do { \
1461 char *destination; \
1462 ENSURE_FAIL_STACK(3); \
1463 DEBUG_PRINT4 (" Push reg %d (spanning %p -> %p)\n", \
1464 num, regstart[num], regend[num]); \
1465 PUSH_FAILURE_POINTER (regstart[num]); \
1466 PUSH_FAILURE_POINTER (regend[num]); \
1467 PUSH_FAILURE_INT (num); \
1468 } while (0)
1469
1470 /* Change the counter's value to VAL, but make sure that it will
1471 be reset when backtracking. */
1472 #define PUSH_NUMBER(ptr,val) \
1473 do { \
1474 char *destination; \
1475 int c; \
1476 ENSURE_FAIL_STACK(3); \
1477 EXTRACT_NUMBER (c, ptr); \
1478 DEBUG_PRINT4 (" Push number %p = %d -> %d\n", ptr, c, val); \
1479 PUSH_FAILURE_INT (c); \
1480 PUSH_FAILURE_POINTER (ptr); \
1481 PUSH_FAILURE_INT (-1); \
1482 STORE_NUMBER (ptr, val); \
1483 } while (0)
1484
1485 /* Pop a saved register off the stack. */
1486 #define POP_FAILURE_REG_OR_COUNT() \
1487 do { \
1488 int reg = POP_FAILURE_INT (); \
1489 if (reg == -1) \
1490 { \
1491 /* It's a counter. */ \
1492 /* Here, we discard `const', making re_match non-reentrant. */ \
1493 unsigned char *ptr = (unsigned char*) POP_FAILURE_POINTER (); \
1494 reg = POP_FAILURE_INT (); \
1495 STORE_NUMBER (ptr, reg); \
1496 DEBUG_PRINT3 (" Pop counter %p = %d\n", ptr, reg); \
1497 } \
1498 else \
1499 { \
1500 regend[reg] = POP_FAILURE_POINTER (); \
1501 regstart[reg] = POP_FAILURE_POINTER (); \
1502 DEBUG_PRINT4 (" Pop reg %d (spanning %p -> %p)\n", \
1503 reg, regstart[reg], regend[reg]); \
1504 } \
1505 } while (0)
1506
1507 /* Check that we are not stuck in an infinite loop. */
1508 #define CHECK_INFINITE_LOOP(pat_cur, string_place) \
1509 do { \
1510 int failure = TOP_FAILURE_HANDLE(); \
1511 /* Check for infinite matching loops */ \
1512 while (failure > 0 && \
1513 (FAILURE_STR (failure) == string_place \
1514 || FAILURE_STR (failure) == NULL)) \
1515 { \
1516 assert (FAILURE_PAT (failure) >= bufp->buffer \
1517 && FAILURE_PAT (failure) <= bufp->buffer + bufp->used); \
1518 if (FAILURE_PAT (failure) == pat_cur) \
1519 goto fail; \
1520 DEBUG_PRINT2 (" Other pattern: %p\n", FAILURE_PAT (failure)); \
1521 failure = NEXT_FAILURE_HANDLE(failure); \
1522 } \
1523 DEBUG_PRINT2 (" Other string: %p\n", FAILURE_STR (failure)); \
1524 } while (0)
1525
1526 /* Push the information about the state we will need
1527 if we ever fail back to it.
1528
1529 Requires variables fail_stack, regstart, regend and
1530 num_regs be declared. GROW_FAIL_STACK requires `destination' be
1531 declared.
1532
1533 Does `return FAILURE_CODE' if runs out of memory. */
1534
1535 #define PUSH_FAILURE_POINT(pattern, string_place) \
1536 do { \
1537 char *destination; \
1538 /* Must be int, so when we don't save any registers, the arithmetic \
1539 of 0 + -1 isn't done as unsigned. */ \
1540 \
1541 DEBUG_STATEMENT (nfailure_points_pushed++); \
1542 DEBUG_PRINT1 ("\nPUSH_FAILURE_POINT:\n"); \
1543 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail); \
1544 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1545 \
1546 ENSURE_FAIL_STACK (NUM_NONREG_ITEMS); \
1547 \
1548 DEBUG_PRINT1 ("\n"); \
1549 \
1550 DEBUG_PRINT2 (" Push frame index: %d\n", fail_stack.frame); \
1551 PUSH_FAILURE_INT (fail_stack.frame); \
1552 \
1553 DEBUG_PRINT2 (" Push string %p: `", string_place); \
1554 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, size2);\
1555 DEBUG_PRINT1 ("'\n"); \
1556 PUSH_FAILURE_POINTER (string_place); \
1557 \
1558 DEBUG_PRINT2 (" Push pattern %p: ", pattern); \
1559 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern, pend); \
1560 PUSH_FAILURE_POINTER (pattern); \
1561 \
1562 /* Close the frame by moving the frame pointer past it. */ \
1563 fail_stack.frame = fail_stack.avail; \
1564 } while (0)
1565
1566 /* Estimate the size of data pushed by a typical failure stack entry.
1567 An estimate is all we need, because all we use this for
1568 is to choose a limit for how big to make the failure stack. */
1569
1570 #define TYPICAL_FAILURE_SIZE 20
1571
1572 /* How many items can still be added to the stack without overflowing it. */
1573 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1574
1575
1576 /* Pops what PUSH_FAIL_STACK pushes.
1577
1578 We restore into the parameters, all of which should be lvalues:
1579 STR -- the saved data position.
1580 PAT -- the saved pattern position.
1581 REGSTART, REGEND -- arrays of string positions.
1582
1583 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1584 `pend', `string1', `size1', `string2', and `size2'. */
1585
1586 #define POP_FAILURE_POINT(str, pat) \
1587 do { \
1588 assert (!FAIL_STACK_EMPTY ()); \
1589 \
1590 /* Remove failure points and point to how many regs pushed. */ \
1591 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1592 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1593 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1594 \
1595 /* Pop the saved registers. */ \
1596 while (fail_stack.frame < fail_stack.avail) \
1597 POP_FAILURE_REG_OR_COUNT (); \
1598 \
1599 pat = POP_FAILURE_POINTER (); \
1600 DEBUG_PRINT2 (" Popping pattern %p: ", pat); \
1601 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1602 \
1603 /* If the saved string location is NULL, it came from an \
1604 on_failure_keep_string_jump opcode, and we want to throw away the \
1605 saved NULL, thus retaining our current position in the string. */ \
1606 str = POP_FAILURE_POINTER (); \
1607 DEBUG_PRINT2 (" Popping string %p: `", str); \
1608 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1609 DEBUG_PRINT1 ("'\n"); \
1610 \
1611 fail_stack.frame = POP_FAILURE_INT (); \
1612 DEBUG_PRINT2 (" Popping frame index: %d\n", fail_stack.frame); \
1613 \
1614 assert (fail_stack.avail >= 0); \
1615 assert (fail_stack.frame <= fail_stack.avail); \
1616 \
1617 DEBUG_STATEMENT (nfailure_points_popped++); \
1618 } while (0) /* POP_FAILURE_POINT */
1619
1620
1621 \f
1622 /* Registers are set to a sentinel when they haven't yet matched. */
1623 #define REG_UNSET(e) ((e) == NULL)
1624 \f
1625 /* Subroutine declarations and macros for regex_compile. */
1626
1627 static reg_errcode_t regex_compile _RE_ARGS ((re_char *pattern, size_t size,
1628 reg_syntax_t syntax,
1629 struct re_pattern_buffer *bufp));
1630 static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg));
1631 static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1632 int arg1, int arg2));
1633 static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1634 int arg, unsigned char *end));
1635 static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1636 int arg1, int arg2, unsigned char *end));
1637 static boolean at_begline_loc_p _RE_ARGS ((re_char *pattern,
1638 re_char *p,
1639 reg_syntax_t syntax));
1640 static boolean at_endline_loc_p _RE_ARGS ((re_char *p,
1641 re_char *pend,
1642 reg_syntax_t syntax));
1643 static re_char *skip_one_char _RE_ARGS ((re_char *p));
1644 static int analyse_first _RE_ARGS ((re_char *p, re_char *pend,
1645 char *fastmap, const int multibyte));
1646
1647 /* Fetch the next character in the uncompiled pattern---translating it
1648 if necessary. */
1649 #define PATFETCH(c) \
1650 do { \
1651 PATFETCH_RAW (c); \
1652 c = TRANSLATE (c); \
1653 } while (0)
1654
1655 /* Fetch the next character in the uncompiled pattern, with no
1656 translation. */
1657 #define PATFETCH_RAW(c) \
1658 do { \
1659 int len; \
1660 if (p == pend) return REG_EEND; \
1661 c = RE_STRING_CHAR_AND_LENGTH (p, pend - p, len); \
1662 p += len; \
1663 } while (0)
1664
1665
1666 /* If `translate' is non-null, return translate[D], else just D. We
1667 cast the subscript to translate because some data is declared as
1668 `char *', to avoid warnings when a string constant is passed. But
1669 when we use a character as a subscript we must make it unsigned. */
1670 #ifndef TRANSLATE
1671 # define TRANSLATE(d) \
1672 (RE_TRANSLATE_P (translate) ? RE_TRANSLATE (translate, (d)) : (d))
1673 #endif
1674
1675
1676 /* Macros for outputting the compiled pattern into `buffer'. */
1677
1678 /* If the buffer isn't allocated when it comes in, use this. */
1679 #define INIT_BUF_SIZE 32
1680
1681 /* Make sure we have at least N more bytes of space in buffer. */
1682 #define GET_BUFFER_SPACE(n) \
1683 while ((size_t) (b - bufp->buffer + (n)) > bufp->allocated) \
1684 EXTEND_BUFFER ()
1685
1686 /* Make sure we have one more byte of buffer space and then add C to it. */
1687 #define BUF_PUSH(c) \
1688 do { \
1689 GET_BUFFER_SPACE (1); \
1690 *b++ = (unsigned char) (c); \
1691 } while (0)
1692
1693
1694 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1695 #define BUF_PUSH_2(c1, c2) \
1696 do { \
1697 GET_BUFFER_SPACE (2); \
1698 *b++ = (unsigned char) (c1); \
1699 *b++ = (unsigned char) (c2); \
1700 } while (0)
1701
1702
1703 /* As with BUF_PUSH_2, except for three bytes. */
1704 #define BUF_PUSH_3(c1, c2, c3) \
1705 do { \
1706 GET_BUFFER_SPACE (3); \
1707 *b++ = (unsigned char) (c1); \
1708 *b++ = (unsigned char) (c2); \
1709 *b++ = (unsigned char) (c3); \
1710 } while (0)
1711
1712
1713 /* Store a jump with opcode OP at LOC to location TO. We store a
1714 relative address offset by the three bytes the jump itself occupies. */
1715 #define STORE_JUMP(op, loc, to) \
1716 store_op1 (op, loc, (to) - (loc) - 3)
1717
1718 /* Likewise, for a two-argument jump. */
1719 #define STORE_JUMP2(op, loc, to, arg) \
1720 store_op2 (op, loc, (to) - (loc) - 3, arg)
1721
1722 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1723 #define INSERT_JUMP(op, loc, to) \
1724 insert_op1 (op, loc, (to) - (loc) - 3, b)
1725
1726 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1727 #define INSERT_JUMP2(op, loc, to, arg) \
1728 insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
1729
1730
1731 /* This is not an arbitrary limit: the arguments which represent offsets
1732 into the pattern are two bytes long. So if 2^16 bytes turns out to
1733 be too small, many things would have to change. */
1734 /* Any other compiler which, like MSC, has allocation limit below 2^16
1735 bytes will have to use approach similar to what was done below for
1736 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1737 reallocating to 0 bytes. Such thing is not going to work too well.
1738 You have been warned!! */
1739 #if defined _MSC_VER && !defined WIN32
1740 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes. */
1741 # define MAX_BUF_SIZE 65500L
1742 #else
1743 # define MAX_BUF_SIZE (1L << 16)
1744 #endif
1745
1746 /* Extend the buffer by twice its current size via realloc and
1747 reset the pointers that pointed into the old block to point to the
1748 correct places in the new one. If extending the buffer results in it
1749 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1750 #if __BOUNDED_POINTERS__
1751 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
1752 # define MOVE_BUFFER_POINTER(P) \
1753 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
1754 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
1755 else \
1756 { \
1757 SET_HIGH_BOUND (b); \
1758 SET_HIGH_BOUND (begalt); \
1759 if (fixup_alt_jump) \
1760 SET_HIGH_BOUND (fixup_alt_jump); \
1761 if (laststart) \
1762 SET_HIGH_BOUND (laststart); \
1763 if (pending_exact) \
1764 SET_HIGH_BOUND (pending_exact); \
1765 }
1766 #else
1767 # define MOVE_BUFFER_POINTER(P) (P) += incr
1768 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
1769 #endif
1770 #define EXTEND_BUFFER() \
1771 do { \
1772 re_char *old_buffer = bufp->buffer; \
1773 if (bufp->allocated == MAX_BUF_SIZE) \
1774 return REG_ESIZE; \
1775 bufp->allocated <<= 1; \
1776 if (bufp->allocated > MAX_BUF_SIZE) \
1777 bufp->allocated = MAX_BUF_SIZE; \
1778 RETALLOC (bufp->buffer, bufp->allocated, unsigned char); \
1779 if (bufp->buffer == NULL) \
1780 return REG_ESPACE; \
1781 /* If the buffer moved, move all the pointers into it. */ \
1782 if (old_buffer != bufp->buffer) \
1783 { \
1784 int incr = bufp->buffer - old_buffer; \
1785 MOVE_BUFFER_POINTER (b); \
1786 MOVE_BUFFER_POINTER (begalt); \
1787 if (fixup_alt_jump) \
1788 MOVE_BUFFER_POINTER (fixup_alt_jump); \
1789 if (laststart) \
1790 MOVE_BUFFER_POINTER (laststart); \
1791 if (pending_exact) \
1792 MOVE_BUFFER_POINTER (pending_exact); \
1793 } \
1794 ELSE_EXTEND_BUFFER_HIGH_BOUND \
1795 } while (0)
1796
1797
1798 /* Since we have one byte reserved for the register number argument to
1799 {start,stop}_memory, the maximum number of groups we can report
1800 things about is what fits in that byte. */
1801 #define MAX_REGNUM 255
1802
1803 /* But patterns can have more than `MAX_REGNUM' registers. We just
1804 ignore the excess. */
1805 typedef unsigned regnum_t;
1806
1807
1808 /* Macros for the compile stack. */
1809
1810 /* Since offsets can go either forwards or backwards, this type needs to
1811 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1812 /* int may be not enough when sizeof(int) == 2. */
1813 typedef long pattern_offset_t;
1814
1815 typedef struct
1816 {
1817 pattern_offset_t begalt_offset;
1818 pattern_offset_t fixup_alt_jump;
1819 pattern_offset_t laststart_offset;
1820 regnum_t regnum;
1821 } compile_stack_elt_t;
1822
1823
1824 typedef struct
1825 {
1826 compile_stack_elt_t *stack;
1827 unsigned size;
1828 unsigned avail; /* Offset of next open position. */
1829 } compile_stack_type;
1830
1831
1832 #define INIT_COMPILE_STACK_SIZE 32
1833
1834 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1835 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1836
1837 /* The next available element. */
1838 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1839
1840
1841 /* Structure to manage work area for range table. */
1842 struct range_table_work_area
1843 {
1844 int *table; /* actual work area. */
1845 int allocated; /* allocated size for work area in bytes. */
1846 int used; /* actually used size in words. */
1847 int bits; /* flag to record character classes */
1848 };
1849
1850 /* Make sure that WORK_AREA can hold more N multibyte characters. */
1851 #define EXTEND_RANGE_TABLE_WORK_AREA(work_area, n) \
1852 do { \
1853 if (((work_area).used + (n)) * sizeof (int) > (work_area).allocated) \
1854 { \
1855 (work_area).allocated += 16 * sizeof (int); \
1856 if ((work_area).table) \
1857 (work_area).table \
1858 = (int *) realloc ((work_area).table, (work_area).allocated); \
1859 else \
1860 (work_area).table \
1861 = (int *) malloc ((work_area).allocated); \
1862 if ((work_area).table == 0) \
1863 FREE_STACK_RETURN (REG_ESPACE); \
1864 } \
1865 } while (0)
1866
1867 #define SET_RANGE_TABLE_WORK_AREA_BIT(work_area, bit) \
1868 (work_area).bits |= (bit)
1869
1870 /* Bits used to implement the multibyte-part of the various character classes
1871 such as [:alnum:] in a charset's range table. */
1872 #define BIT_WORD 0x1
1873 #define BIT_LOWER 0x2
1874 #define BIT_PUNCT 0x4
1875 #define BIT_SPACE 0x8
1876 #define BIT_UPPER 0x10
1877 #define BIT_MULTIBYTE 0x20
1878
1879 /* Set a range (RANGE_START, RANGE_END) to WORK_AREA. */
1880 #define SET_RANGE_TABLE_WORK_AREA(work_area, range_start, range_end) \
1881 do { \
1882 EXTEND_RANGE_TABLE_WORK_AREA ((work_area), 2); \
1883 (work_area).table[(work_area).used++] = (range_start); \
1884 (work_area).table[(work_area).used++] = (range_end); \
1885 } while (0)
1886
1887 /* Free allocated memory for WORK_AREA. */
1888 #define FREE_RANGE_TABLE_WORK_AREA(work_area) \
1889 do { \
1890 if ((work_area).table) \
1891 free ((work_area).table); \
1892 } while (0)
1893
1894 #define CLEAR_RANGE_TABLE_WORK_USED(work_area) ((work_area).used = 0, (work_area).bits = 0)
1895 #define RANGE_TABLE_WORK_USED(work_area) ((work_area).used)
1896 #define RANGE_TABLE_WORK_BITS(work_area) ((work_area).bits)
1897 #define RANGE_TABLE_WORK_ELT(work_area, i) ((work_area).table[i])
1898
1899
1900 /* Set the bit for character C in a list. */
1901 #define SET_LIST_BIT(c) (b[((c)) / BYTEWIDTH] |= 1 << ((c) % BYTEWIDTH))
1902
1903
1904 /* Get the next unsigned number in the uncompiled pattern. */
1905 #define GET_UNSIGNED_NUMBER(num) \
1906 do { if (p != pend) \
1907 { \
1908 PATFETCH (c); \
1909 while ('0' <= c && c <= '9') \
1910 { \
1911 if (num < 0) \
1912 num = 0; \
1913 num = num * 10 + c - '0'; \
1914 if (p == pend) \
1915 break; \
1916 PATFETCH (c); \
1917 } \
1918 } \
1919 } while (0)
1920
1921 #if WIDE_CHAR_SUPPORT
1922 /* The GNU C library provides support for user-defined character classes
1923 and the functions from ISO C amendement 1. */
1924 # ifdef CHARCLASS_NAME_MAX
1925 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
1926 # else
1927 /* This shouldn't happen but some implementation might still have this
1928 problem. Use a reasonable default value. */
1929 # define CHAR_CLASS_MAX_LENGTH 256
1930 # endif
1931 typedef wctype_t re_wctype_t;
1932 typedef wchar_t re_wchar_t;
1933 # define re_wctype wctype
1934 # define re_iswctype iswctype
1935 # define re_wctype_to_bit(cc) 0
1936 #else
1937 # define CHAR_CLASS_MAX_LENGTH 9 /* Namely, `multibyte'. */
1938 # define btowc(c) c
1939
1940 /* Character classes. */
1941 typedef enum { RECC_ERROR = 0,
1942 RECC_ALNUM, RECC_ALPHA, RECC_WORD,
1943 RECC_GRAPH, RECC_PRINT,
1944 RECC_LOWER, RECC_UPPER,
1945 RECC_PUNCT, RECC_CNTRL,
1946 RECC_DIGIT, RECC_XDIGIT,
1947 RECC_BLANK, RECC_SPACE,
1948 RECC_MULTIBYTE, RECC_NONASCII,
1949 RECC_ASCII, RECC_UNIBYTE
1950 } re_wctype_t;
1951
1952 typedef int re_wchar_t;
1953
1954 /* Map a string to the char class it names (if any). */
1955 static re_wctype_t
1956 re_wctype (string)
1957 re_char *string;
1958 {
1959 if (STREQ (string, "alnum")) return RECC_ALNUM;
1960 else if (STREQ (string, "alpha")) return RECC_ALPHA;
1961 else if (STREQ (string, "word")) return RECC_WORD;
1962 else if (STREQ (string, "ascii")) return RECC_ASCII;
1963 else if (STREQ (string, "nonascii")) return RECC_NONASCII;
1964 else if (STREQ (string, "graph")) return RECC_GRAPH;
1965 else if (STREQ (string, "lower")) return RECC_LOWER;
1966 else if (STREQ (string, "print")) return RECC_PRINT;
1967 else if (STREQ (string, "punct")) return RECC_PUNCT;
1968 else if (STREQ (string, "space")) return RECC_SPACE;
1969 else if (STREQ (string, "upper")) return RECC_UPPER;
1970 else if (STREQ (string, "unibyte")) return RECC_UNIBYTE;
1971 else if (STREQ (string, "multibyte")) return RECC_MULTIBYTE;
1972 else if (STREQ (string, "digit")) return RECC_DIGIT;
1973 else if (STREQ (string, "xdigit")) return RECC_XDIGIT;
1974 else if (STREQ (string, "cntrl")) return RECC_CNTRL;
1975 else if (STREQ (string, "blank")) return RECC_BLANK;
1976 else return 0;
1977 }
1978
1979 /* True iff CH is in the char class CC. */
1980 static boolean
1981 re_iswctype (ch, cc)
1982 int ch;
1983 re_wctype_t cc;
1984 {
1985 switch (cc)
1986 {
1987 case RECC_ALNUM: return ISALNUM (ch);
1988 case RECC_ALPHA: return ISALPHA (ch);
1989 case RECC_BLANK: return ISBLANK (ch);
1990 case RECC_CNTRL: return ISCNTRL (ch);
1991 case RECC_DIGIT: return ISDIGIT (ch);
1992 case RECC_GRAPH: return ISGRAPH (ch);
1993 case RECC_LOWER: return ISLOWER (ch);
1994 case RECC_PRINT: return ISPRINT (ch);
1995 case RECC_PUNCT: return ISPUNCT (ch);
1996 case RECC_SPACE: return ISSPACE (ch);
1997 case RECC_UPPER: return ISUPPER (ch);
1998 case RECC_XDIGIT: return ISXDIGIT (ch);
1999 case RECC_ASCII: return IS_REAL_ASCII (ch);
2000 case RECC_NONASCII: return !IS_REAL_ASCII (ch);
2001 case RECC_UNIBYTE: return ISUNIBYTE (ch);
2002 case RECC_MULTIBYTE: return !ISUNIBYTE (ch);
2003 case RECC_WORD: return ISWORD (ch);
2004 case RECC_ERROR: return false;
2005 default:
2006 abort();
2007 }
2008 }
2009
2010 /* Return a bit-pattern to use in the range-table bits to match multibyte
2011 chars of class CC. */
2012 static int
2013 re_wctype_to_bit (cc)
2014 re_wctype_t cc;
2015 {
2016 switch (cc)
2017 {
2018 case RECC_NONASCII: case RECC_PRINT: case RECC_GRAPH:
2019 case RECC_MULTIBYTE: return BIT_MULTIBYTE;
2020 case RECC_ALPHA: case RECC_ALNUM: case RECC_WORD: return BIT_WORD;
2021 case RECC_LOWER: return BIT_LOWER;
2022 case RECC_UPPER: return BIT_UPPER;
2023 case RECC_PUNCT: return BIT_PUNCT;
2024 case RECC_SPACE: return BIT_SPACE;
2025 case RECC_ASCII: case RECC_DIGIT: case RECC_XDIGIT: case RECC_CNTRL:
2026 case RECC_BLANK: case RECC_UNIBYTE: case RECC_ERROR: return 0;
2027 default:
2028 abort();
2029 }
2030 }
2031 #endif
2032
2033 /* Explicit quit checking is only used on NTemacs. */
2034 #if defined WINDOWSNT && defined emacs && defined QUIT
2035 extern int immediate_quit;
2036 # define IMMEDIATE_QUIT_CHECK \
2037 do { \
2038 if (immediate_quit) QUIT; \
2039 } while (0)
2040 #else
2041 # define IMMEDIATE_QUIT_CHECK ((void)0)
2042 #endif
2043 \f
2044 #ifndef MATCH_MAY_ALLOCATE
2045
2046 /* If we cannot allocate large objects within re_match_2_internal,
2047 we make the fail stack and register vectors global.
2048 The fail stack, we grow to the maximum size when a regexp
2049 is compiled.
2050 The register vectors, we adjust in size each time we
2051 compile a regexp, according to the number of registers it needs. */
2052
2053 static fail_stack_type fail_stack;
2054
2055 /* Size with which the following vectors are currently allocated.
2056 That is so we can make them bigger as needed,
2057 but never make them smaller. */
2058 static int regs_allocated_size;
2059
2060 static re_char ** regstart, ** regend;
2061 static re_char **best_regstart, **best_regend;
2062
2063 /* Make the register vectors big enough for NUM_REGS registers,
2064 but don't make them smaller. */
2065
2066 static
2067 regex_grow_registers (num_regs)
2068 int num_regs;
2069 {
2070 if (num_regs > regs_allocated_size)
2071 {
2072 RETALLOC_IF (regstart, num_regs, re_char *);
2073 RETALLOC_IF (regend, num_regs, re_char *);
2074 RETALLOC_IF (best_regstart, num_regs, re_char *);
2075 RETALLOC_IF (best_regend, num_regs, re_char *);
2076
2077 regs_allocated_size = num_regs;
2078 }
2079 }
2080
2081 #endif /* not MATCH_MAY_ALLOCATE */
2082 \f
2083 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
2084 compile_stack,
2085 regnum_t regnum));
2086
2087 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2088 Returns one of error codes defined in `regex.h', or zero for success.
2089
2090 Assumes the `allocated' (and perhaps `buffer') and `translate'
2091 fields are set in BUFP on entry.
2092
2093 If it succeeds, results are put in BUFP (if it returns an error, the
2094 contents of BUFP are undefined):
2095 `buffer' is the compiled pattern;
2096 `syntax' is set to SYNTAX;
2097 `used' is set to the length of the compiled pattern;
2098 `fastmap_accurate' is zero;
2099 `re_nsub' is the number of subexpressions in PATTERN;
2100 `not_bol' and `not_eol' are zero;
2101
2102 The `fastmap' field is neither examined nor set. */
2103
2104 /* Insert the `jump' from the end of last alternative to "here".
2105 The space for the jump has already been allocated. */
2106 #define FIXUP_ALT_JUMP() \
2107 do { \
2108 if (fixup_alt_jump) \
2109 STORE_JUMP (jump, fixup_alt_jump, b); \
2110 } while (0)
2111
2112
2113 /* Return, freeing storage we allocated. */
2114 #define FREE_STACK_RETURN(value) \
2115 do { \
2116 FREE_RANGE_TABLE_WORK_AREA (range_table_work); \
2117 free (compile_stack.stack); \
2118 return value; \
2119 } while (0)
2120
2121 static reg_errcode_t
2122 regex_compile (pattern, size, syntax, bufp)
2123 re_char *pattern;
2124 size_t size;
2125 reg_syntax_t syntax;
2126 struct re_pattern_buffer *bufp;
2127 {
2128 /* We fetch characters from PATTERN here. */
2129 register re_wchar_t c, c1;
2130
2131 /* A random temporary spot in PATTERN. */
2132 re_char *p1;
2133
2134 /* Points to the end of the buffer, where we should append. */
2135 register unsigned char *b;
2136
2137 /* Keeps track of unclosed groups. */
2138 compile_stack_type compile_stack;
2139
2140 /* Points to the current (ending) position in the pattern. */
2141 #ifdef AIX
2142 /* `const' makes AIX compiler fail. */
2143 unsigned char *p = pattern;
2144 #else
2145 re_char *p = pattern;
2146 #endif
2147 re_char *pend = pattern + size;
2148
2149 /* How to translate the characters in the pattern. */
2150 RE_TRANSLATE_TYPE translate = bufp->translate;
2151
2152 /* Address of the count-byte of the most recently inserted `exactn'
2153 command. This makes it possible to tell if a new exact-match
2154 character can be added to that command or if the character requires
2155 a new `exactn' command. */
2156 unsigned char *pending_exact = 0;
2157
2158 /* Address of start of the most recently finished expression.
2159 This tells, e.g., postfix * where to find the start of its
2160 operand. Reset at the beginning of groups and alternatives. */
2161 unsigned char *laststart = 0;
2162
2163 /* Address of beginning of regexp, or inside of last group. */
2164 unsigned char *begalt;
2165
2166 /* Place in the uncompiled pattern (i.e., the {) to
2167 which to go back if the interval is invalid. */
2168 re_char *beg_interval;
2169
2170 /* Address of the place where a forward jump should go to the end of
2171 the containing expression. Each alternative of an `or' -- except the
2172 last -- ends with a forward jump of this sort. */
2173 unsigned char *fixup_alt_jump = 0;
2174
2175 /* Counts open-groups as they are encountered. Remembered for the
2176 matching close-group on the compile stack, so the same register
2177 number is put in the stop_memory as the start_memory. */
2178 regnum_t regnum = 0;
2179
2180 /* Work area for range table of charset. */
2181 struct range_table_work_area range_table_work;
2182
2183 /* If the object matched can contain multibyte characters. */
2184 const boolean multibyte = RE_MULTIBYTE_P (bufp);
2185
2186 #ifdef DEBUG
2187 debug++;
2188 DEBUG_PRINT1 ("\nCompiling pattern: ");
2189 if (debug > 0)
2190 {
2191 unsigned debug_count;
2192
2193 for (debug_count = 0; debug_count < size; debug_count++)
2194 putchar (pattern[debug_count]);
2195 putchar ('\n');
2196 }
2197 #endif /* DEBUG */
2198
2199 /* Initialize the compile stack. */
2200 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
2201 if (compile_stack.stack == NULL)
2202 return REG_ESPACE;
2203
2204 compile_stack.size = INIT_COMPILE_STACK_SIZE;
2205 compile_stack.avail = 0;
2206
2207 range_table_work.table = 0;
2208 range_table_work.allocated = 0;
2209
2210 /* Initialize the pattern buffer. */
2211 bufp->syntax = syntax;
2212 bufp->fastmap_accurate = 0;
2213 bufp->not_bol = bufp->not_eol = 0;
2214
2215 /* Set `used' to zero, so that if we return an error, the pattern
2216 printer (for debugging) will think there's no pattern. We reset it
2217 at the end. */
2218 bufp->used = 0;
2219
2220 /* Always count groups, whether or not bufp->no_sub is set. */
2221 bufp->re_nsub = 0;
2222
2223 #if !defined emacs && !defined SYNTAX_TABLE
2224 /* Initialize the syntax table. */
2225 init_syntax_once ();
2226 #endif
2227
2228 if (bufp->allocated == 0)
2229 {
2230 if (bufp->buffer)
2231 { /* If zero allocated, but buffer is non-null, try to realloc
2232 enough space. This loses if buffer's address is bogus, but
2233 that is the user's responsibility. */
2234 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
2235 }
2236 else
2237 { /* Caller did not allocate a buffer. Do it for them. */
2238 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
2239 }
2240 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
2241
2242 bufp->allocated = INIT_BUF_SIZE;
2243 }
2244
2245 begalt = b = bufp->buffer;
2246
2247 /* Loop through the uncompiled pattern until we're at the end. */
2248 while (p != pend)
2249 {
2250 PATFETCH (c);
2251
2252 switch (c)
2253 {
2254 case '^':
2255 {
2256 if ( /* If at start of pattern, it's an operator. */
2257 p == pattern + 1
2258 /* If context independent, it's an operator. */
2259 || syntax & RE_CONTEXT_INDEP_ANCHORS
2260 /* Otherwise, depends on what's come before. */
2261 || at_begline_loc_p (pattern, p, syntax))
2262 BUF_PUSH ((syntax & RE_NO_NEWLINE_ANCHOR) ? begbuf : begline);
2263 else
2264 goto normal_char;
2265 }
2266 break;
2267
2268
2269 case '$':
2270 {
2271 if ( /* If at end of pattern, it's an operator. */
2272 p == pend
2273 /* If context independent, it's an operator. */
2274 || syntax & RE_CONTEXT_INDEP_ANCHORS
2275 /* Otherwise, depends on what's next. */
2276 || at_endline_loc_p (p, pend, syntax))
2277 BUF_PUSH ((syntax & RE_NO_NEWLINE_ANCHOR) ? endbuf : endline);
2278 else
2279 goto normal_char;
2280 }
2281 break;
2282
2283
2284 case '+':
2285 case '?':
2286 if ((syntax & RE_BK_PLUS_QM)
2287 || (syntax & RE_LIMITED_OPS))
2288 goto normal_char;
2289 handle_plus:
2290 case '*':
2291 /* If there is no previous pattern... */
2292 if (!laststart)
2293 {
2294 if (syntax & RE_CONTEXT_INVALID_OPS)
2295 FREE_STACK_RETURN (REG_BADRPT);
2296 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
2297 goto normal_char;
2298 }
2299
2300 {
2301 /* 1 means zero (many) matches is allowed. */
2302 boolean zero_times_ok = 0, many_times_ok = 0;
2303 boolean greedy = 1;
2304
2305 /* If there is a sequence of repetition chars, collapse it
2306 down to just one (the right one). We can't combine
2307 interval operators with these because of, e.g., `a{2}*',
2308 which should only match an even number of `a's. */
2309
2310 for (;;)
2311 {
2312 if ((syntax & RE_FRUGAL)
2313 && c == '?' && (zero_times_ok || many_times_ok))
2314 greedy = 0;
2315 else
2316 {
2317 zero_times_ok |= c != '+';
2318 many_times_ok |= c != '?';
2319 }
2320
2321 if (p == pend)
2322 break;
2323 else if (*p == '*'
2324 || (!(syntax & RE_BK_PLUS_QM)
2325 && (*p == '+' || *p == '?')))
2326 ;
2327 else if (syntax & RE_BK_PLUS_QM && *p == '\\')
2328 {
2329 if (p+1 == pend)
2330 FREE_STACK_RETURN (REG_EESCAPE);
2331 if (p[1] == '+' || p[1] == '?')
2332 PATFETCH (c); /* Gobble up the backslash. */
2333 else
2334 break;
2335 }
2336 else
2337 break;
2338 /* If we get here, we found another repeat character. */
2339 PATFETCH (c);
2340 }
2341
2342 /* Star, etc. applied to an empty pattern is equivalent
2343 to an empty pattern. */
2344 if (!laststart || laststart == b)
2345 break;
2346
2347 /* Now we know whether or not zero matches is allowed
2348 and also whether or not two or more matches is allowed. */
2349 if (greedy)
2350 {
2351 if (many_times_ok)
2352 {
2353 boolean simple = skip_one_char (laststart) == b;
2354 unsigned int startoffset = 0;
2355 re_opcode_t ofj =
2356 /* Check if the loop can match the empty string. */
2357 (simple || !analyse_first (laststart, b, NULL, 0)) ?
2358 on_failure_jump : on_failure_jump_loop;
2359 assert (skip_one_char (laststart) <= b);
2360
2361 if (!zero_times_ok && simple)
2362 { /* Since simple * loops can be made faster by using
2363 on_failure_keep_string_jump, we turn simple P+
2364 into PP* if P is simple. */
2365 unsigned char *p1, *p2;
2366 startoffset = b - laststart;
2367 GET_BUFFER_SPACE (startoffset);
2368 p1 = b; p2 = laststart;
2369 while (p2 < p1)
2370 *b++ = *p2++;
2371 zero_times_ok = 1;
2372 }
2373
2374 GET_BUFFER_SPACE (6);
2375 if (!zero_times_ok)
2376 /* A + loop. */
2377 STORE_JUMP (ofj, b, b + 6);
2378 else
2379 /* Simple * loops can use on_failure_keep_string_jump
2380 depending on what follows. But since we don't know
2381 that yet, we leave the decision up to
2382 on_failure_jump_smart. */
2383 INSERT_JUMP (simple ? on_failure_jump_smart : ofj,
2384 laststart + startoffset, b + 6);
2385 b += 3;
2386 STORE_JUMP (jump, b, laststart + startoffset);
2387 b += 3;
2388 }
2389 else
2390 {
2391 /* A simple ? pattern. */
2392 assert (zero_times_ok);
2393 GET_BUFFER_SPACE (3);
2394 INSERT_JUMP (on_failure_jump, laststart, b + 3);
2395 b += 3;
2396 }
2397 }
2398 else /* not greedy */
2399 { /* I wish the greedy and non-greedy cases could be merged. */
2400
2401 GET_BUFFER_SPACE (7); /* We might use less. */
2402 if (many_times_ok)
2403 {
2404 boolean emptyp = analyse_first (laststart, b, NULL, 0);
2405
2406 /* The non-greedy multiple match looks like a repeat..until:
2407 we only need a conditional jump at the end of the loop */
2408 if (emptyp) BUF_PUSH (no_op);
2409 STORE_JUMP (emptyp ? on_failure_jump_nastyloop
2410 : on_failure_jump, b, laststart);
2411 b += 3;
2412 if (zero_times_ok)
2413 {
2414 /* The repeat...until naturally matches one or more.
2415 To also match zero times, we need to first jump to
2416 the end of the loop (its conditional jump). */
2417 INSERT_JUMP (jump, laststart, b);
2418 b += 3;
2419 }
2420 }
2421 else
2422 {
2423 /* non-greedy a?? */
2424 INSERT_JUMP (jump, laststart, b + 3);
2425 b += 3;
2426 INSERT_JUMP (on_failure_jump, laststart, laststart + 6);
2427 b += 3;
2428 }
2429 }
2430 }
2431 pending_exact = 0;
2432 break;
2433
2434
2435 case '.':
2436 laststart = b;
2437 BUF_PUSH (anychar);
2438 break;
2439
2440
2441 case '[':
2442 {
2443 CLEAR_RANGE_TABLE_WORK_USED (range_table_work);
2444
2445 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2446
2447 /* Ensure that we have enough space to push a charset: the
2448 opcode, the length count, and the bitset; 34 bytes in all. */
2449 GET_BUFFER_SPACE (34);
2450
2451 laststart = b;
2452
2453 /* We test `*p == '^' twice, instead of using an if
2454 statement, so we only need one BUF_PUSH. */
2455 BUF_PUSH (*p == '^' ? charset_not : charset);
2456 if (*p == '^')
2457 p++;
2458
2459 /* Remember the first position in the bracket expression. */
2460 p1 = p;
2461
2462 /* Push the number of bytes in the bitmap. */
2463 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
2464
2465 /* Clear the whole map. */
2466 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
2467
2468 /* charset_not matches newline according to a syntax bit. */
2469 if ((re_opcode_t) b[-2] == charset_not
2470 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2471 SET_LIST_BIT ('\n');
2472
2473 /* Read in characters and ranges, setting map bits. */
2474 for (;;)
2475 {
2476 boolean escaped_char = false;
2477 const unsigned char *p2 = p;
2478
2479 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2480
2481 PATFETCH (c);
2482
2483 /* \ might escape characters inside [...] and [^...]. */
2484 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2485 {
2486 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2487
2488 PATFETCH (c);
2489 escaped_char = true;
2490 }
2491 else
2492 {
2493 /* Could be the end of the bracket expression. If it's
2494 not (i.e., when the bracket expression is `[]' so
2495 far), the ']' character bit gets set way below. */
2496 if (c == ']' && p2 != p1)
2497 break;
2498 }
2499
2500 /* What should we do for the character which is
2501 greater than 0x7F, but not BASE_LEADING_CODE_P?
2502 XXX */
2503
2504 /* See if we're at the beginning of a possible character
2505 class. */
2506
2507 if (!escaped_char &&
2508 syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2509 {
2510 /* Leave room for the null. */
2511 unsigned char str[CHAR_CLASS_MAX_LENGTH + 1];
2512 const unsigned char *class_beg;
2513
2514 PATFETCH (c);
2515 c1 = 0;
2516 class_beg = p;
2517
2518 /* If pattern is `[[:'. */
2519 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2520
2521 for (;;)
2522 {
2523 PATFETCH (c);
2524 if ((c == ':' && *p == ']') || p == pend)
2525 break;
2526 if (c1 < CHAR_CLASS_MAX_LENGTH)
2527 str[c1++] = c;
2528 else
2529 /* This is in any case an invalid class name. */
2530 str[0] = '\0';
2531 }
2532 str[c1] = '\0';
2533
2534 /* If isn't a word bracketed by `[:' and `:]':
2535 undo the ending character, the letters, and
2536 leave the leading `:' and `[' (but set bits for
2537 them). */
2538 if (c == ':' && *p == ']')
2539 {
2540 int ch;
2541 re_wctype_t cc;
2542
2543 cc = re_wctype (str);
2544
2545 if (cc == 0)
2546 FREE_STACK_RETURN (REG_ECTYPE);
2547
2548 /* Throw away the ] at the end of the character
2549 class. */
2550 PATFETCH (c);
2551
2552 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2553
2554 /* Most character classes in a multibyte match
2555 just set a flag. Exceptions are is_blank,
2556 is_digit, is_cntrl, and is_xdigit, since
2557 they can only match ASCII characters. We
2558 don't need to handle them for multibyte.
2559 They are distinguished by a negative wctype. */
2560
2561 if (multibyte)
2562 SET_RANGE_TABLE_WORK_AREA_BIT (range_table_work,
2563 re_wctype_to_bit (cc));
2564
2565 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
2566 {
2567 int translated = TRANSLATE (ch);
2568 if (re_iswctype (btowc (ch), cc))
2569 SET_LIST_BIT (translated);
2570 }
2571
2572 /* Repeat the loop. */
2573 continue;
2574 }
2575 else
2576 {
2577 /* Go back to right after the "[:". */
2578 p = class_beg;
2579 SET_LIST_BIT ('[');
2580
2581 /* Because the `:' may starts the range, we
2582 can't simply set bit and repeat the loop.
2583 Instead, just set it to C and handle below. */
2584 c = ':';
2585 }
2586 }
2587
2588 if (p < pend && p[0] == '-' && p[1] != ']')
2589 {
2590
2591 /* Discard the `-'. */
2592 PATFETCH (c1);
2593
2594 /* Fetch the character which ends the range. */
2595 PATFETCH (c1);
2596
2597 if (SINGLE_BYTE_CHAR_P (c))
2598 {
2599 if (! SINGLE_BYTE_CHAR_P (c1))
2600 {
2601 /* Handle a range starting with a
2602 character of less than 256, and ending
2603 with a character of not less than 256.
2604 Split that into two ranges, the low one
2605 ending at 0377, and the high one
2606 starting at the smallest character in
2607 the charset of C1 and ending at C1. */
2608 int charset = CHAR_CHARSET (c1);
2609 int c2 = MAKE_CHAR (charset, 0, 0);
2610
2611 SET_RANGE_TABLE_WORK_AREA (range_table_work,
2612 c2, c1);
2613 c1 = 0377;
2614 }
2615 }
2616 else if (!SAME_CHARSET_P (c, c1))
2617 FREE_STACK_RETURN (REG_ERANGE);
2618 }
2619 else
2620 /* Range from C to C. */
2621 c1 = c;
2622
2623 /* Set the range ... */
2624 if (SINGLE_BYTE_CHAR_P (c))
2625 /* ... into bitmap. */
2626 {
2627 re_wchar_t this_char;
2628 int range_start = c, range_end = c1;
2629
2630 /* If the start is after the end, the range is empty. */
2631 if (range_start > range_end)
2632 {
2633 if (syntax & RE_NO_EMPTY_RANGES)
2634 FREE_STACK_RETURN (REG_ERANGE);
2635 /* Else, repeat the loop. */
2636 }
2637 else
2638 {
2639 for (this_char = range_start; this_char <= range_end;
2640 this_char++)
2641 SET_LIST_BIT (TRANSLATE (this_char));
2642 }
2643 }
2644 else
2645 /* ... into range table. */
2646 SET_RANGE_TABLE_WORK_AREA (range_table_work, c, c1);
2647 }
2648
2649 /* Discard any (non)matching list bytes that are all 0 at the
2650 end of the map. Decrease the map-length byte too. */
2651 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2652 b[-1]--;
2653 b += b[-1];
2654
2655 /* Build real range table from work area. */
2656 if (RANGE_TABLE_WORK_USED (range_table_work)
2657 || RANGE_TABLE_WORK_BITS (range_table_work))
2658 {
2659 int i;
2660 int used = RANGE_TABLE_WORK_USED (range_table_work);
2661
2662 /* Allocate space for COUNT + RANGE_TABLE. Needs two
2663 bytes for flags, two for COUNT, and three bytes for
2664 each character. */
2665 GET_BUFFER_SPACE (4 + used * 3);
2666
2667 /* Indicate the existence of range table. */
2668 laststart[1] |= 0x80;
2669
2670 /* Store the character class flag bits into the range table.
2671 If not in emacs, these flag bits are always 0. */
2672 *b++ = RANGE_TABLE_WORK_BITS (range_table_work) & 0xff;
2673 *b++ = RANGE_TABLE_WORK_BITS (range_table_work) >> 8;
2674
2675 STORE_NUMBER_AND_INCR (b, used / 2);
2676 for (i = 0; i < used; i++)
2677 STORE_CHARACTER_AND_INCR
2678 (b, RANGE_TABLE_WORK_ELT (range_table_work, i));
2679 }
2680 }
2681 break;
2682
2683
2684 case '(':
2685 if (syntax & RE_NO_BK_PARENS)
2686 goto handle_open;
2687 else
2688 goto normal_char;
2689
2690
2691 case ')':
2692 if (syntax & RE_NO_BK_PARENS)
2693 goto handle_close;
2694 else
2695 goto normal_char;
2696
2697
2698 case '\n':
2699 if (syntax & RE_NEWLINE_ALT)
2700 goto handle_alt;
2701 else
2702 goto normal_char;
2703
2704
2705 case '|':
2706 if (syntax & RE_NO_BK_VBAR)
2707 goto handle_alt;
2708 else
2709 goto normal_char;
2710
2711
2712 case '{':
2713 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2714 goto handle_interval;
2715 else
2716 goto normal_char;
2717
2718
2719 case '\\':
2720 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2721
2722 /* Do not translate the character after the \, so that we can
2723 distinguish, e.g., \B from \b, even if we normally would
2724 translate, e.g., B to b. */
2725 PATFETCH_RAW (c);
2726
2727 switch (c)
2728 {
2729 case '(':
2730 if (syntax & RE_NO_BK_PARENS)
2731 goto normal_backslash;
2732
2733 handle_open:
2734 {
2735 int shy = 0;
2736 if (p+1 < pend)
2737 {
2738 /* Look for a special (?...) construct */
2739 if ((syntax & RE_SHY_GROUPS) && *p == '?')
2740 {
2741 PATFETCH (c); /* Gobble up the '?'. */
2742 PATFETCH (c);
2743 switch (c)
2744 {
2745 case ':': shy = 1; break;
2746 default:
2747 /* Only (?:...) is supported right now. */
2748 FREE_STACK_RETURN (REG_BADPAT);
2749 }
2750 }
2751 }
2752
2753 if (!shy)
2754 {
2755 bufp->re_nsub++;
2756 regnum++;
2757 }
2758
2759 if (COMPILE_STACK_FULL)
2760 {
2761 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2762 compile_stack_elt_t);
2763 if (compile_stack.stack == NULL) return REG_ESPACE;
2764
2765 compile_stack.size <<= 1;
2766 }
2767
2768 /* These are the values to restore when we hit end of this
2769 group. They are all relative offsets, so that if the
2770 whole pattern moves because of realloc, they will still
2771 be valid. */
2772 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2773 COMPILE_STACK_TOP.fixup_alt_jump
2774 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2775 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2776 COMPILE_STACK_TOP.regnum = shy ? -regnum : regnum;
2777
2778 /* Do not push a
2779 start_memory for groups beyond the last one we can
2780 represent in the compiled pattern. */
2781 if (regnum <= MAX_REGNUM && !shy)
2782 BUF_PUSH_2 (start_memory, regnum);
2783
2784 compile_stack.avail++;
2785
2786 fixup_alt_jump = 0;
2787 laststart = 0;
2788 begalt = b;
2789 /* If we've reached MAX_REGNUM groups, then this open
2790 won't actually generate any code, so we'll have to
2791 clear pending_exact explicitly. */
2792 pending_exact = 0;
2793 break;
2794 }
2795
2796 case ')':
2797 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2798
2799 if (COMPILE_STACK_EMPTY)
2800 {
2801 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2802 goto normal_backslash;
2803 else
2804 FREE_STACK_RETURN (REG_ERPAREN);
2805 }
2806
2807 handle_close:
2808 FIXUP_ALT_JUMP ();
2809
2810 /* See similar code for backslashed left paren above. */
2811 if (COMPILE_STACK_EMPTY)
2812 {
2813 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2814 goto normal_char;
2815 else
2816 FREE_STACK_RETURN (REG_ERPAREN);
2817 }
2818
2819 /* Since we just checked for an empty stack above, this
2820 ``can't happen''. */
2821 assert (compile_stack.avail != 0);
2822 {
2823 /* We don't just want to restore into `regnum', because
2824 later groups should continue to be numbered higher,
2825 as in `(ab)c(de)' -- the second group is #2. */
2826 regnum_t this_group_regnum;
2827
2828 compile_stack.avail--;
2829 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2830 fixup_alt_jump
2831 = COMPILE_STACK_TOP.fixup_alt_jump
2832 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2833 : 0;
2834 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2835 this_group_regnum = COMPILE_STACK_TOP.regnum;
2836 /* If we've reached MAX_REGNUM groups, then this open
2837 won't actually generate any code, so we'll have to
2838 clear pending_exact explicitly. */
2839 pending_exact = 0;
2840
2841 /* We're at the end of the group, so now we know how many
2842 groups were inside this one. */
2843 if (this_group_regnum <= MAX_REGNUM && this_group_regnum > 0)
2844 BUF_PUSH_2 (stop_memory, this_group_regnum);
2845 }
2846 break;
2847
2848
2849 case '|': /* `\|'. */
2850 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2851 goto normal_backslash;
2852 handle_alt:
2853 if (syntax & RE_LIMITED_OPS)
2854 goto normal_char;
2855
2856 /* Insert before the previous alternative a jump which
2857 jumps to this alternative if the former fails. */
2858 GET_BUFFER_SPACE (3);
2859 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2860 pending_exact = 0;
2861 b += 3;
2862
2863 /* The alternative before this one has a jump after it
2864 which gets executed if it gets matched. Adjust that
2865 jump so it will jump to this alternative's analogous
2866 jump (put in below, which in turn will jump to the next
2867 (if any) alternative's such jump, etc.). The last such
2868 jump jumps to the correct final destination. A picture:
2869 _____ _____
2870 | | | |
2871 | v | v
2872 a | b | c
2873
2874 If we are at `b', then fixup_alt_jump right now points to a
2875 three-byte space after `a'. We'll put in the jump, set
2876 fixup_alt_jump to right after `b', and leave behind three
2877 bytes which we'll fill in when we get to after `c'. */
2878
2879 FIXUP_ALT_JUMP ();
2880
2881 /* Mark and leave space for a jump after this alternative,
2882 to be filled in later either by next alternative or
2883 when know we're at the end of a series of alternatives. */
2884 fixup_alt_jump = b;
2885 GET_BUFFER_SPACE (3);
2886 b += 3;
2887
2888 laststart = 0;
2889 begalt = b;
2890 break;
2891
2892
2893 case '{':
2894 /* If \{ is a literal. */
2895 if (!(syntax & RE_INTERVALS)
2896 /* If we're at `\{' and it's not the open-interval
2897 operator. */
2898 || (syntax & RE_NO_BK_BRACES))
2899 goto normal_backslash;
2900
2901 handle_interval:
2902 {
2903 /* If got here, then the syntax allows intervals. */
2904
2905 /* At least (most) this many matches must be made. */
2906 int lower_bound = 0, upper_bound = -1;
2907
2908 beg_interval = p;
2909
2910 if (p == pend)
2911 FREE_STACK_RETURN (REG_EBRACE);
2912
2913 GET_UNSIGNED_NUMBER (lower_bound);
2914
2915 if (c == ',')
2916 GET_UNSIGNED_NUMBER (upper_bound);
2917 else
2918 /* Interval such as `{1}' => match exactly once. */
2919 upper_bound = lower_bound;
2920
2921 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2922 || (upper_bound >= 0 && lower_bound > upper_bound))
2923 FREE_STACK_RETURN (REG_BADBR);
2924
2925 if (!(syntax & RE_NO_BK_BRACES))
2926 {
2927 if (c != '\\')
2928 FREE_STACK_RETURN (REG_BADBR);
2929
2930 PATFETCH (c);
2931 }
2932
2933 if (c != '}')
2934 FREE_STACK_RETURN (REG_BADBR);
2935
2936 /* We just parsed a valid interval. */
2937
2938 /* If it's invalid to have no preceding re. */
2939 if (!laststart)
2940 {
2941 if (syntax & RE_CONTEXT_INVALID_OPS)
2942 FREE_STACK_RETURN (REG_BADRPT);
2943 else if (syntax & RE_CONTEXT_INDEP_OPS)
2944 laststart = b;
2945 else
2946 goto unfetch_interval;
2947 }
2948
2949 if (upper_bound == 0)
2950 /* If the upper bound is zero, just drop the sub pattern
2951 altogether. */
2952 b = laststart;
2953 else if (lower_bound == 1 && upper_bound == 1)
2954 /* Just match it once: nothing to do here. */
2955 ;
2956
2957 /* Otherwise, we have a nontrivial interval. When
2958 we're all done, the pattern will look like:
2959 set_number_at <jump count> <upper bound>
2960 set_number_at <succeed_n count> <lower bound>
2961 succeed_n <after jump addr> <succeed_n count>
2962 <body of loop>
2963 jump_n <succeed_n addr> <jump count>
2964 (The upper bound and `jump_n' are omitted if
2965 `upper_bound' is 1, though.) */
2966 else
2967 { /* If the upper bound is > 1, we need to insert
2968 more at the end of the loop. */
2969 unsigned int nbytes = (upper_bound < 0 ? 3
2970 : upper_bound > 1 ? 5 : 0);
2971 unsigned int startoffset = 0;
2972
2973 GET_BUFFER_SPACE (20); /* We might use less. */
2974
2975 if (lower_bound == 0)
2976 {
2977 /* A succeed_n that starts with 0 is really a
2978 a simple on_failure_jump_loop. */
2979 INSERT_JUMP (on_failure_jump_loop, laststart,
2980 b + 3 + nbytes);
2981 b += 3;
2982 }
2983 else
2984 {
2985 /* Initialize lower bound of the `succeed_n', even
2986 though it will be set during matching by its
2987 attendant `set_number_at' (inserted next),
2988 because `re_compile_fastmap' needs to know.
2989 Jump to the `jump_n' we might insert below. */
2990 INSERT_JUMP2 (succeed_n, laststart,
2991 b + 5 + nbytes,
2992 lower_bound);
2993 b += 5;
2994
2995 /* Code to initialize the lower bound. Insert
2996 before the `succeed_n'. The `5' is the last two
2997 bytes of this `set_number_at', plus 3 bytes of
2998 the following `succeed_n'. */
2999 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
3000 b += 5;
3001 startoffset += 5;
3002 }
3003
3004 if (upper_bound < 0)
3005 {
3006 /* A negative upper bound stands for infinity,
3007 in which case it degenerates to a plain jump. */
3008 STORE_JUMP (jump, b, laststart + startoffset);
3009 b += 3;
3010 }
3011 else if (upper_bound > 1)
3012 { /* More than one repetition is allowed, so
3013 append a backward jump to the `succeed_n'
3014 that starts this interval.
3015
3016 When we've reached this during matching,
3017 we'll have matched the interval once, so
3018 jump back only `upper_bound - 1' times. */
3019 STORE_JUMP2 (jump_n, b, laststart + startoffset,
3020 upper_bound - 1);
3021 b += 5;
3022
3023 /* The location we want to set is the second
3024 parameter of the `jump_n'; that is `b-2' as
3025 an absolute address. `laststart' will be
3026 the `set_number_at' we're about to insert;
3027 `laststart+3' the number to set, the source
3028 for the relative address. But we are
3029 inserting into the middle of the pattern --
3030 so everything is getting moved up by 5.
3031 Conclusion: (b - 2) - (laststart + 3) + 5,
3032 i.e., b - laststart.
3033
3034 We insert this at the beginning of the loop
3035 so that if we fail during matching, we'll
3036 reinitialize the bounds. */
3037 insert_op2 (set_number_at, laststart, b - laststart,
3038 upper_bound - 1, b);
3039 b += 5;
3040 }
3041 }
3042 pending_exact = 0;
3043 beg_interval = NULL;
3044 }
3045 break;
3046
3047 unfetch_interval:
3048 /* If an invalid interval, match the characters as literals. */
3049 assert (beg_interval);
3050 p = beg_interval;
3051 beg_interval = NULL;
3052
3053 /* normal_char and normal_backslash need `c'. */
3054 c = '{';
3055
3056 if (!(syntax & RE_NO_BK_BRACES))
3057 {
3058 assert (p > pattern && p[-1] == '\\');
3059 goto normal_backslash;
3060 }
3061 else
3062 goto normal_char;
3063
3064 #ifdef emacs
3065 /* There is no way to specify the before_dot and after_dot
3066 operators. rms says this is ok. --karl */
3067 case '=':
3068 BUF_PUSH (at_dot);
3069 break;
3070
3071 case 's':
3072 laststart = b;
3073 PATFETCH (c);
3074 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
3075 break;
3076
3077 case 'S':
3078 laststart = b;
3079 PATFETCH (c);
3080 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
3081 break;
3082
3083 case 'c':
3084 laststart = b;
3085 PATFETCH_RAW (c);
3086 BUF_PUSH_2 (categoryspec, c);
3087 break;
3088
3089 case 'C':
3090 laststart = b;
3091 PATFETCH_RAW (c);
3092 BUF_PUSH_2 (notcategoryspec, c);
3093 break;
3094 #endif /* emacs */
3095
3096
3097 case 'w':
3098 if (syntax & RE_NO_GNU_OPS)
3099 goto normal_char;
3100 laststart = b;
3101 BUF_PUSH_2 (syntaxspec, Sword);
3102 break;
3103
3104
3105 case 'W':
3106 if (syntax & RE_NO_GNU_OPS)
3107 goto normal_char;
3108 laststart = b;
3109 BUF_PUSH_2 (notsyntaxspec, Sword);
3110 break;
3111
3112
3113 case '<':
3114 if (syntax & RE_NO_GNU_OPS)
3115 goto normal_char;
3116 BUF_PUSH (wordbeg);
3117 break;
3118
3119 case '>':
3120 if (syntax & RE_NO_GNU_OPS)
3121 goto normal_char;
3122 BUF_PUSH (wordend);
3123 break;
3124
3125 case 'b':
3126 if (syntax & RE_NO_GNU_OPS)
3127 goto normal_char;
3128 BUF_PUSH (wordbound);
3129 break;
3130
3131 case 'B':
3132 if (syntax & RE_NO_GNU_OPS)
3133 goto normal_char;
3134 BUF_PUSH (notwordbound);
3135 break;
3136
3137 case '`':
3138 if (syntax & RE_NO_GNU_OPS)
3139 goto normal_char;
3140 BUF_PUSH (begbuf);
3141 break;
3142
3143 case '\'':
3144 if (syntax & RE_NO_GNU_OPS)
3145 goto normal_char;
3146 BUF_PUSH (endbuf);
3147 break;
3148
3149 case '1': case '2': case '3': case '4': case '5':
3150 case '6': case '7': case '8': case '9':
3151 {
3152 regnum_t reg;
3153
3154 if (syntax & RE_NO_BK_REFS)
3155 goto normal_backslash;
3156
3157 reg = c - '0';
3158
3159 /* Can't back reference to a subexpression before its end. */
3160 if (reg > regnum || group_in_compile_stack (compile_stack, reg))
3161 FREE_STACK_RETURN (REG_ESUBREG);
3162
3163 laststart = b;
3164 BUF_PUSH_2 (duplicate, reg);
3165 }
3166 break;
3167
3168
3169 case '+':
3170 case '?':
3171 if (syntax & RE_BK_PLUS_QM)
3172 goto handle_plus;
3173 else
3174 goto normal_backslash;
3175
3176 default:
3177 normal_backslash:
3178 /* You might think it would be useful for \ to mean
3179 not to translate; but if we don't translate it
3180 it will never match anything. */
3181 c = TRANSLATE (c);
3182 goto normal_char;
3183 }
3184 break;
3185
3186
3187 default:
3188 /* Expects the character in `c'. */
3189 normal_char:
3190 /* If no exactn currently being built. */
3191 if (!pending_exact
3192
3193 /* If last exactn not at current position. */
3194 || pending_exact + *pending_exact + 1 != b
3195
3196 /* We have only one byte following the exactn for the count. */
3197 || *pending_exact >= (1 << BYTEWIDTH) - MAX_MULTIBYTE_LENGTH
3198
3199 /* If followed by a repetition operator. */
3200 || (p != pend && (*p == '*' || *p == '^'))
3201 || ((syntax & RE_BK_PLUS_QM)
3202 ? p + 1 < pend && *p == '\\' && (p[1] == '+' || p[1] == '?')
3203 : p != pend && (*p == '+' || *p == '?'))
3204 || ((syntax & RE_INTERVALS)
3205 && ((syntax & RE_NO_BK_BRACES)
3206 ? p != pend && *p == '{'
3207 : p + 1 < pend && p[0] == '\\' && p[1] == '{')))
3208 {
3209 /* Start building a new exactn. */
3210
3211 laststart = b;
3212
3213 BUF_PUSH_2 (exactn, 0);
3214 pending_exact = b - 1;
3215 }
3216
3217 GET_BUFFER_SPACE (MAX_MULTIBYTE_LENGTH);
3218 {
3219 int len;
3220
3221 if (multibyte)
3222 len = CHAR_STRING (c, b);
3223 else
3224 *b = c, len = 1;
3225 b += len;
3226 (*pending_exact) += len;
3227 }
3228
3229 break;
3230 } /* switch (c) */
3231 } /* while p != pend */
3232
3233
3234 /* Through the pattern now. */
3235
3236 FIXUP_ALT_JUMP ();
3237
3238 if (!COMPILE_STACK_EMPTY)
3239 FREE_STACK_RETURN (REG_EPAREN);
3240
3241 /* If we don't want backtracking, force success
3242 the first time we reach the end of the compiled pattern. */
3243 if (syntax & RE_NO_POSIX_BACKTRACKING)
3244 BUF_PUSH (succeed);
3245
3246 free (compile_stack.stack);
3247
3248 /* We have succeeded; set the length of the buffer. */
3249 bufp->used = b - bufp->buffer;
3250
3251 #ifdef DEBUG
3252 if (debug > 0)
3253 {
3254 re_compile_fastmap (bufp);
3255 DEBUG_PRINT1 ("\nCompiled pattern: \n");
3256 print_compiled_pattern (bufp);
3257 }
3258 debug--;
3259 #endif /* DEBUG */
3260
3261 #ifndef MATCH_MAY_ALLOCATE
3262 /* Initialize the failure stack to the largest possible stack. This
3263 isn't necessary unless we're trying to avoid calling alloca in
3264 the search and match routines. */
3265 {
3266 int num_regs = bufp->re_nsub + 1;
3267
3268 if (fail_stack.size < re_max_failures * TYPICAL_FAILURE_SIZE)
3269 {
3270 fail_stack.size = re_max_failures * TYPICAL_FAILURE_SIZE;
3271
3272 if (! fail_stack.stack)
3273 fail_stack.stack
3274 = (fail_stack_elt_t *) malloc (fail_stack.size
3275 * sizeof (fail_stack_elt_t));
3276 else
3277 fail_stack.stack
3278 = (fail_stack_elt_t *) realloc (fail_stack.stack,
3279 (fail_stack.size
3280 * sizeof (fail_stack_elt_t)));
3281 }
3282
3283 regex_grow_registers (num_regs);
3284 }
3285 #endif /* not MATCH_MAY_ALLOCATE */
3286
3287 return REG_NOERROR;
3288 } /* regex_compile */
3289 \f
3290 /* Subroutines for `regex_compile'. */
3291
3292 /* Store OP at LOC followed by two-byte integer parameter ARG. */
3293
3294 static void
3295 store_op1 (op, loc, arg)
3296 re_opcode_t op;
3297 unsigned char *loc;
3298 int arg;
3299 {
3300 *loc = (unsigned char) op;
3301 STORE_NUMBER (loc + 1, arg);
3302 }
3303
3304
3305 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
3306
3307 static void
3308 store_op2 (op, loc, arg1, arg2)
3309 re_opcode_t op;
3310 unsigned char *loc;
3311 int arg1, arg2;
3312 {
3313 *loc = (unsigned char) op;
3314 STORE_NUMBER (loc + 1, arg1);
3315 STORE_NUMBER (loc + 3, arg2);
3316 }
3317
3318
3319 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
3320 for OP followed by two-byte integer parameter ARG. */
3321
3322 static void
3323 insert_op1 (op, loc, arg, end)
3324 re_opcode_t op;
3325 unsigned char *loc;
3326 int arg;
3327 unsigned char *end;
3328 {
3329 register unsigned char *pfrom = end;
3330 register unsigned char *pto = end + 3;
3331
3332 while (pfrom != loc)
3333 *--pto = *--pfrom;
3334
3335 store_op1 (op, loc, arg);
3336 }
3337
3338
3339 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
3340
3341 static void
3342 insert_op2 (op, loc, arg1, arg2, end)
3343 re_opcode_t op;
3344 unsigned char *loc;
3345 int arg1, arg2;
3346 unsigned char *end;
3347 {
3348 register unsigned char *pfrom = end;
3349 register unsigned char *pto = end + 5;
3350
3351 while (pfrom != loc)
3352 *--pto = *--pfrom;
3353
3354 store_op2 (op, loc, arg1, arg2);
3355 }
3356
3357
3358 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
3359 after an alternative or a begin-subexpression. We assume there is at
3360 least one character before the ^. */
3361
3362 static boolean
3363 at_begline_loc_p (pattern, p, syntax)
3364 re_char *pattern, *p;
3365 reg_syntax_t syntax;
3366 {
3367 re_char *prev = p - 2;
3368 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
3369
3370 return
3371 /* After a subexpression? */
3372 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
3373 /* After an alternative? */
3374 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash))
3375 /* After a shy subexpression? */
3376 || ((syntax & RE_SHY_GROUPS) && prev - 2 >= pattern
3377 && prev[-1] == '?' && prev[-2] == '('
3378 && (syntax & RE_NO_BK_PARENS
3379 || (prev - 3 >= pattern && prev[-3] == '\\')));
3380 }
3381
3382
3383 /* The dual of at_begline_loc_p. This one is for $. We assume there is
3384 at least one character after the $, i.e., `P < PEND'. */
3385
3386 static boolean
3387 at_endline_loc_p (p, pend, syntax)
3388 re_char *p, *pend;
3389 reg_syntax_t syntax;
3390 {
3391 re_char *next = p;
3392 boolean next_backslash = *next == '\\';
3393 re_char *next_next = p + 1 < pend ? p + 1 : 0;
3394
3395 return
3396 /* Before a subexpression? */
3397 (syntax & RE_NO_BK_PARENS ? *next == ')'
3398 : next_backslash && next_next && *next_next == ')')
3399 /* Before an alternative? */
3400 || (syntax & RE_NO_BK_VBAR ? *next == '|'
3401 : next_backslash && next_next && *next_next == '|');
3402 }
3403
3404
3405 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
3406 false if it's not. */
3407
3408 static boolean
3409 group_in_compile_stack (compile_stack, regnum)
3410 compile_stack_type compile_stack;
3411 regnum_t regnum;
3412 {
3413 int this_element;
3414
3415 for (this_element = compile_stack.avail - 1;
3416 this_element >= 0;
3417 this_element--)
3418 if (compile_stack.stack[this_element].regnum == regnum)
3419 return true;
3420
3421 return false;
3422 }
3423 \f
3424 /* analyse_first.
3425 If fastmap is non-NULL, go through the pattern and fill fastmap
3426 with all the possible leading chars. If fastmap is NULL, don't
3427 bother filling it up (obviously) and only return whether the
3428 pattern could potentially match the empty string.
3429
3430 Return 1 if p..pend might match the empty string.
3431 Return 0 if p..pend matches at least one char.
3432 Return -1 if fastmap was not updated accurately. */
3433
3434 static int
3435 analyse_first (p, pend, fastmap, multibyte)
3436 re_char *p, *pend;
3437 char *fastmap;
3438 const int multibyte;
3439 {
3440 int j, k;
3441 boolean not;
3442
3443 /* If all elements for base leading-codes in fastmap is set, this
3444 flag is set true. */
3445 boolean match_any_multibyte_characters = false;
3446
3447 assert (p);
3448
3449 /* The loop below works as follows:
3450 - It has a working-list kept in the PATTERN_STACK and which basically
3451 starts by only containing a pointer to the first operation.
3452 - If the opcode we're looking at is a match against some set of
3453 chars, then we add those chars to the fastmap and go on to the
3454 next work element from the worklist (done via `break').
3455 - If the opcode is a control operator on the other hand, we either
3456 ignore it (if it's meaningless at this point, such as `start_memory')
3457 or execute it (if it's a jump). If the jump has several destinations
3458 (i.e. `on_failure_jump'), then we push the other destination onto the
3459 worklist.
3460 We guarantee termination by ignoring backward jumps (more or less),
3461 so that `p' is monotonically increasing. More to the point, we
3462 never set `p' (or push) anything `<= p1'. */
3463
3464 while (p < pend)
3465 {
3466 /* `p1' is used as a marker of how far back a `on_failure_jump'
3467 can go without being ignored. It is normally equal to `p'
3468 (which prevents any backward `on_failure_jump') except right
3469 after a plain `jump', to allow patterns such as:
3470 0: jump 10
3471 3..9: <body>
3472 10: on_failure_jump 3
3473 as used for the *? operator. */
3474 re_char *p1 = p;
3475
3476 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3477 {
3478 case succeed:
3479 return 1;
3480 continue;
3481
3482 case duplicate:
3483 /* If the first character has to match a backreference, that means
3484 that the group was empty (since it already matched). Since this
3485 is the only case that interests us here, we can assume that the
3486 backreference must match the empty string. */
3487 p++;
3488 continue;
3489
3490
3491 /* Following are the cases which match a character. These end
3492 with `break'. */
3493
3494 case exactn:
3495 if (fastmap)
3496 {
3497 int c = RE_STRING_CHAR (p + 1, pend - p);
3498
3499 if (SINGLE_BYTE_CHAR_P (c))
3500 fastmap[c] = 1;
3501 else
3502 fastmap[p[1]] = 1;
3503 }
3504 break;
3505
3506
3507 case anychar:
3508 /* We could put all the chars except for \n (and maybe \0)
3509 but we don't bother since it is generally not worth it. */
3510 if (!fastmap) break;
3511 return -1;
3512
3513
3514 case charset_not:
3515 /* Chars beyond end of bitmap are possible matches.
3516 All the single-byte codes can occur in multibyte buffers.
3517 So any that are not listed in the charset
3518 are possible matches, even in multibyte buffers. */
3519 if (!fastmap) break;
3520 for (j = CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH;
3521 j < (1 << BYTEWIDTH); j++)
3522 fastmap[j] = 1;
3523 /* Fallthrough */
3524 case charset:
3525 if (!fastmap) break;
3526 not = (re_opcode_t) *(p - 1) == charset_not;
3527 for (j = CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH - 1, p++;
3528 j >= 0; j--)
3529 if (!!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) ^ not)
3530 fastmap[j] = 1;
3531
3532 if ((not && multibyte)
3533 /* Any character set can possibly contain a character
3534 which doesn't match the specified set of characters. */
3535 || (CHARSET_RANGE_TABLE_EXISTS_P (&p[-2])
3536 && CHARSET_RANGE_TABLE_BITS (&p[-2]) != 0))
3537 /* If we can match a character class, we can match
3538 any character set. */
3539 {
3540 set_fastmap_for_multibyte_characters:
3541 if (match_any_multibyte_characters == false)
3542 {
3543 for (j = 0x80; j < 0xA0; j++) /* XXX */
3544 if (BASE_LEADING_CODE_P (j))
3545 fastmap[j] = 1;
3546 match_any_multibyte_characters = true;
3547 }
3548 }
3549
3550 else if (!not && CHARSET_RANGE_TABLE_EXISTS_P (&p[-2])
3551 && match_any_multibyte_characters == false)
3552 {
3553 /* Set fastmap[I] 1 where I is a base leading code of each
3554 multibyte character in the range table. */
3555 int c, count;
3556
3557 /* Make P points the range table. `+ 2' is to skip flag
3558 bits for a character class. */
3559 p += CHARSET_BITMAP_SIZE (&p[-2]) + 2;
3560
3561 /* Extract the number of ranges in range table into COUNT. */
3562 EXTRACT_NUMBER_AND_INCR (count, p);
3563 for (; count > 0; count--, p += 2 * 3) /* XXX */
3564 {
3565 /* Extract the start of each range. */
3566 EXTRACT_CHARACTER (c, p);
3567 j = CHAR_CHARSET (c);
3568 fastmap[CHARSET_LEADING_CODE_BASE (j)] = 1;
3569 }
3570 }
3571 break;
3572
3573 case syntaxspec:
3574 case notsyntaxspec:
3575 if (!fastmap) break;
3576 #ifndef emacs
3577 not = (re_opcode_t)p[-1] == notsyntaxspec;
3578 k = *p++;
3579 for (j = 0; j < (1 << BYTEWIDTH); j++)
3580 if ((SYNTAX (j) == (enum syntaxcode) k) ^ not)
3581 fastmap[j] = 1;
3582 break;
3583 #else /* emacs */
3584 /* This match depends on text properties. These end with
3585 aborting optimizations. */
3586 return -1;
3587
3588 case categoryspec:
3589 case notcategoryspec:
3590 if (!fastmap) break;
3591 not = (re_opcode_t)p[-1] == notcategoryspec;
3592 k = *p++;
3593 for (j = 0; j < (1 << BYTEWIDTH); j++)
3594 if ((CHAR_HAS_CATEGORY (j, k)) ^ not)
3595 fastmap[j] = 1;
3596
3597 if (multibyte)
3598 /* Any character set can possibly contain a character
3599 whose category is K (or not). */
3600 goto set_fastmap_for_multibyte_characters;
3601 break;
3602
3603 /* All cases after this match the empty string. These end with
3604 `continue'. */
3605
3606 case before_dot:
3607 case at_dot:
3608 case after_dot:
3609 #endif /* !emacs */
3610 case no_op:
3611 case begline:
3612 case endline:
3613 case begbuf:
3614 case endbuf:
3615 case wordbound:
3616 case notwordbound:
3617 case wordbeg:
3618 case wordend:
3619 continue;
3620
3621
3622 case jump:
3623 EXTRACT_NUMBER_AND_INCR (j, p);
3624 if (j < 0)
3625 /* Backward jumps can only go back to code that we've already
3626 visited. `re_compile' should make sure this is true. */
3627 break;
3628 p += j;
3629 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p))
3630 {
3631 case on_failure_jump:
3632 case on_failure_keep_string_jump:
3633 case on_failure_jump_loop:
3634 case on_failure_jump_nastyloop:
3635 case on_failure_jump_smart:
3636 p++;
3637 break;
3638 default:
3639 continue;
3640 };
3641 /* Keep `p1' to allow the `on_failure_jump' we are jumping to
3642 to jump back to "just after here". */
3643 /* Fallthrough */
3644
3645 case on_failure_jump:
3646 case on_failure_keep_string_jump:
3647 case on_failure_jump_nastyloop:
3648 case on_failure_jump_loop:
3649 case on_failure_jump_smart:
3650 EXTRACT_NUMBER_AND_INCR (j, p);
3651 if (p + j <= p1)
3652 ; /* Backward jump to be ignored. */
3653 else
3654 { /* We have to look down both arms.
3655 We first go down the "straight" path so as to minimize
3656 stack usage when going through alternatives. */
3657 int r = analyse_first (p, pend, fastmap, multibyte);
3658 if (r) return r;
3659 p += j;
3660 }
3661 continue;
3662
3663
3664 case jump_n:
3665 /* This code simply does not properly handle forward jump_n. */
3666 DEBUG_STATEMENT (EXTRACT_NUMBER (j, p); assert (j < 0));
3667 p += 4;
3668 /* jump_n can either jump or fall through. The (backward) jump
3669 case has already been handled, so we only need to look at the
3670 fallthrough case. */
3671 continue;
3672
3673 case succeed_n:
3674 /* If N == 0, it should be an on_failure_jump_loop instead. */
3675 DEBUG_STATEMENT (EXTRACT_NUMBER (j, p + 2); assert (j > 0));
3676 p += 4;
3677 /* We only care about one iteration of the loop, so we don't
3678 need to consider the case where this behaves like an
3679 on_failure_jump. */
3680 continue;
3681
3682
3683 case set_number_at:
3684 p += 4;
3685 continue;
3686
3687
3688 case start_memory:
3689 case stop_memory:
3690 p += 1;
3691 continue;
3692
3693
3694 default:
3695 abort (); /* We have listed all the cases. */
3696 } /* switch *p++ */
3697
3698 /* Getting here means we have found the possible starting
3699 characters for one path of the pattern -- and that the empty
3700 string does not match. We need not follow this path further. */
3701 return 0;
3702 } /* while p */
3703
3704 /* We reached the end without matching anything. */
3705 return 1;
3706
3707 } /* analyse_first */
3708 \f
3709 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3710 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3711 characters can start a string that matches the pattern. This fastmap
3712 is used by re_search to skip quickly over impossible starting points.
3713
3714 Character codes above (1 << BYTEWIDTH) are not represented in the
3715 fastmap, but the leading codes are represented. Thus, the fastmap
3716 indicates which character sets could start a match.
3717
3718 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3719 area as BUFP->fastmap.
3720
3721 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3722 the pattern buffer.
3723
3724 Returns 0 if we succeed, -2 if an internal error. */
3725
3726 int
3727 re_compile_fastmap (bufp)
3728 struct re_pattern_buffer *bufp;
3729 {
3730 char *fastmap = bufp->fastmap;
3731 int analysis;
3732
3733 assert (fastmap && bufp->buffer);
3734
3735 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
3736 bufp->fastmap_accurate = 1; /* It will be when we're done. */
3737
3738 analysis = analyse_first (bufp->buffer, bufp->buffer + bufp->used,
3739 fastmap, RE_MULTIBYTE_P (bufp));
3740 bufp->can_be_null = (analysis != 0);
3741 return 0;
3742 } /* re_compile_fastmap */
3743 \f
3744 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3745 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3746 this memory for recording register information. STARTS and ENDS
3747 must be allocated using the malloc library routine, and must each
3748 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3749
3750 If NUM_REGS == 0, then subsequent matches should allocate their own
3751 register data.
3752
3753 Unless this function is called, the first search or match using
3754 PATTERN_BUFFER will allocate its own register data, without
3755 freeing the old data. */
3756
3757 void
3758 re_set_registers (bufp, regs, num_regs, starts, ends)
3759 struct re_pattern_buffer *bufp;
3760 struct re_registers *regs;
3761 unsigned num_regs;
3762 regoff_t *starts, *ends;
3763 {
3764 if (num_regs)
3765 {
3766 bufp->regs_allocated = REGS_REALLOCATE;
3767 regs->num_regs = num_regs;
3768 regs->start = starts;
3769 regs->end = ends;
3770 }
3771 else
3772 {
3773 bufp->regs_allocated = REGS_UNALLOCATED;
3774 regs->num_regs = 0;
3775 regs->start = regs->end = (regoff_t *) 0;
3776 }
3777 }
3778 WEAK_ALIAS (__re_set_registers, re_set_registers)
3779 \f
3780 /* Searching routines. */
3781
3782 /* Like re_search_2, below, but only one string is specified, and
3783 doesn't let you say where to stop matching. */
3784
3785 int
3786 re_search (bufp, string, size, startpos, range, regs)
3787 struct re_pattern_buffer *bufp;
3788 const char *string;
3789 int size, startpos, range;
3790 struct re_registers *regs;
3791 {
3792 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3793 regs, size);
3794 }
3795 WEAK_ALIAS (__re_search, re_search)
3796
3797 /* End address of virtual concatenation of string. */
3798 #define STOP_ADDR_VSTRING(P) \
3799 (((P) >= size1 ? string2 + size2 : string1 + size1))
3800
3801 /* Address of POS in the concatenation of virtual string. */
3802 #define POS_ADDR_VSTRING(POS) \
3803 (((POS) >= size1 ? string2 - size1 : string1) + (POS))
3804
3805 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3806 virtual concatenation of STRING1 and STRING2, starting first at index
3807 STARTPOS, then at STARTPOS + 1, and so on.
3808
3809 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3810
3811 RANGE is how far to scan while trying to match. RANGE = 0 means try
3812 only at STARTPOS; in general, the last start tried is STARTPOS +
3813 RANGE.
3814
3815 In REGS, return the indices of the virtual concatenation of STRING1
3816 and STRING2 that matched the entire BUFP->buffer and its contained
3817 subexpressions.
3818
3819 Do not consider matching one past the index STOP in the virtual
3820 concatenation of STRING1 and STRING2.
3821
3822 We return either the position in the strings at which the match was
3823 found, -1 if no match, or -2 if error (such as failure
3824 stack overflow). */
3825
3826 int
3827 re_search_2 (bufp, str1, size1, str2, size2, startpos, range, regs, stop)
3828 struct re_pattern_buffer *bufp;
3829 const char *str1, *str2;
3830 int size1, size2;
3831 int startpos;
3832 int range;
3833 struct re_registers *regs;
3834 int stop;
3835 {
3836 int val;
3837 re_char *string1 = (re_char*) str1;
3838 re_char *string2 = (re_char*) str2;
3839 register char *fastmap = bufp->fastmap;
3840 register RE_TRANSLATE_TYPE translate = bufp->translate;
3841 int total_size = size1 + size2;
3842 int endpos = startpos + range;
3843 boolean anchored_start;
3844
3845 /* Nonzero if we have to concern multibyte character. */
3846 const boolean multibyte = RE_MULTIBYTE_P (bufp);
3847
3848 /* Check for out-of-range STARTPOS. */
3849 if (startpos < 0 || startpos > total_size)
3850 return -1;
3851
3852 /* Fix up RANGE if it might eventually take us outside
3853 the virtual concatenation of STRING1 and STRING2.
3854 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3855 if (endpos < 0)
3856 range = 0 - startpos;
3857 else if (endpos > total_size)
3858 range = total_size - startpos;
3859
3860 /* If the search isn't to be a backwards one, don't waste time in a
3861 search for a pattern anchored at beginning of buffer. */
3862 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3863 {
3864 if (startpos > 0)
3865 return -1;
3866 else
3867 range = 0;
3868 }
3869
3870 #ifdef emacs
3871 /* In a forward search for something that starts with \=.
3872 don't keep searching past point. */
3873 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3874 {
3875 range = PT_BYTE - BEGV_BYTE - startpos;
3876 if (range < 0)
3877 return -1;
3878 }
3879 #endif /* emacs */
3880
3881 /* Update the fastmap now if not correct already. */
3882 if (fastmap && !bufp->fastmap_accurate)
3883 re_compile_fastmap (bufp);
3884
3885 /* See whether the pattern is anchored. */
3886 anchored_start = (bufp->buffer[0] == begline);
3887
3888 #ifdef emacs
3889 gl_state.object = re_match_object;
3890 {
3891 int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (POS_AS_IN_BUFFER (startpos));
3892
3893 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, charpos, 1);
3894 }
3895 #endif
3896
3897 /* Loop through the string, looking for a place to start matching. */
3898 for (;;)
3899 {
3900 /* If the pattern is anchored,
3901 skip quickly past places we cannot match.
3902 We don't bother to treat startpos == 0 specially
3903 because that case doesn't repeat. */
3904 if (anchored_start && startpos > 0)
3905 {
3906 if (! ((startpos <= size1 ? string1[startpos - 1]
3907 : string2[startpos - size1 - 1])
3908 == '\n'))
3909 goto advance;
3910 }
3911
3912 /* If a fastmap is supplied, skip quickly over characters that
3913 cannot be the start of a match. If the pattern can match the
3914 null string, however, we don't need to skip characters; we want
3915 the first null string. */
3916 if (fastmap && startpos < total_size && !bufp->can_be_null)
3917 {
3918 register re_char *d;
3919 register re_wchar_t buf_ch;
3920
3921 d = POS_ADDR_VSTRING (startpos);
3922
3923 if (range > 0) /* Searching forwards. */
3924 {
3925 register int lim = 0;
3926 int irange = range;
3927
3928 if (startpos < size1 && startpos + range >= size1)
3929 lim = range - (size1 - startpos);
3930
3931 /* Written out as an if-else to avoid testing `translate'
3932 inside the loop. */
3933 if (RE_TRANSLATE_P (translate))
3934 {
3935 if (multibyte)
3936 while (range > lim)
3937 {
3938 int buf_charlen;
3939
3940 buf_ch = STRING_CHAR_AND_LENGTH (d, range - lim,
3941 buf_charlen);
3942
3943 buf_ch = RE_TRANSLATE (translate, buf_ch);
3944 if (buf_ch >= 0400
3945 || fastmap[buf_ch])
3946 break;
3947
3948 range -= buf_charlen;
3949 d += buf_charlen;
3950 }
3951 else
3952 while (range > lim
3953 && !fastmap[RE_TRANSLATE (translate, *d)])
3954 {
3955 d++;
3956 range--;
3957 }
3958 }
3959 else
3960 while (range > lim && !fastmap[*d])
3961 {
3962 d++;
3963 range--;
3964 }
3965
3966 startpos += irange - range;
3967 }
3968 else /* Searching backwards. */
3969 {
3970 int room = (startpos >= size1
3971 ? size2 + size1 - startpos
3972 : size1 - startpos);
3973 buf_ch = RE_STRING_CHAR (d, room);
3974 buf_ch = TRANSLATE (buf_ch);
3975
3976 if (! (buf_ch >= 0400
3977 || fastmap[buf_ch]))
3978 goto advance;
3979 }
3980 }
3981
3982 /* If can't match the null string, and that's all we have left, fail. */
3983 if (range >= 0 && startpos == total_size && fastmap
3984 && !bufp->can_be_null)
3985 return -1;
3986
3987 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3988 startpos, regs, stop);
3989 #ifndef REGEX_MALLOC
3990 # ifdef C_ALLOCA
3991 alloca (0);
3992 # endif
3993 #endif
3994
3995 if (val >= 0)
3996 return startpos;
3997
3998 if (val == -2)
3999 return -2;
4000
4001 advance:
4002 if (!range)
4003 break;
4004 else if (range > 0)
4005 {
4006 /* Update STARTPOS to the next character boundary. */
4007 if (multibyte)
4008 {
4009 re_char *p = POS_ADDR_VSTRING (startpos);
4010 re_char *pend = STOP_ADDR_VSTRING (startpos);
4011 int len = MULTIBYTE_FORM_LENGTH (p, pend - p);
4012
4013 range -= len;
4014 if (range < 0)
4015 break;
4016 startpos += len;
4017 }
4018 else
4019 {
4020 range--;
4021 startpos++;
4022 }
4023 }
4024 else
4025 {
4026 range++;
4027 startpos--;
4028
4029 /* Update STARTPOS to the previous character boundary. */
4030 if (multibyte)
4031 {
4032 re_char *p = POS_ADDR_VSTRING (startpos);
4033 int len = 0;
4034
4035 /* Find the head of multibyte form. */
4036 while (!CHAR_HEAD_P (*p))
4037 p--, len++;
4038
4039 /* Adjust it. */
4040 #if 0 /* XXX */
4041 if (MULTIBYTE_FORM_LENGTH (p, len + 1) != (len + 1))
4042 ;
4043 else
4044 #endif
4045 {
4046 range += len;
4047 if (range > 0)
4048 break;
4049
4050 startpos -= len;
4051 }
4052 }
4053 }
4054 }
4055 return -1;
4056 } /* re_search_2 */
4057 WEAK_ALIAS (__re_search_2, re_search_2)
4058 \f
4059 /* Declarations and macros for re_match_2. */
4060
4061 static int bcmp_translate _RE_ARGS((re_char *s1, re_char *s2,
4062 register int len,
4063 RE_TRANSLATE_TYPE translate,
4064 const int multibyte));
4065
4066 /* This converts PTR, a pointer into one of the search strings `string1'
4067 and `string2' into an offset from the beginning of that string. */
4068 #define POINTER_TO_OFFSET(ptr) \
4069 (FIRST_STRING_P (ptr) \
4070 ? ((regoff_t) ((ptr) - string1)) \
4071 : ((regoff_t) ((ptr) - string2 + size1)))
4072
4073 /* Call before fetching a character with *d. This switches over to
4074 string2 if necessary.
4075 Check re_match_2_internal for a discussion of why end_match_2 might
4076 not be within string2 (but be equal to end_match_1 instead). */
4077 #define PREFETCH() \
4078 while (d == dend) \
4079 { \
4080 /* End of string2 => fail. */ \
4081 if (dend == end_match_2) \
4082 goto fail; \
4083 /* End of string1 => advance to string2. */ \
4084 d = string2; \
4085 dend = end_match_2; \
4086 }
4087
4088 /* Call before fetching a char with *d if you already checked other limits.
4089 This is meant for use in lookahead operations like wordend, etc..
4090 where we might need to look at parts of the string that might be
4091 outside of the LIMITs (i.e past `stop'). */
4092 #define PREFETCH_NOLIMIT() \
4093 if (d == end1) \
4094 { \
4095 d = string2; \
4096 dend = end_match_2; \
4097 } \
4098
4099 /* Test if at very beginning or at very end of the virtual concatenation
4100 of `string1' and `string2'. If only one string, it's `string2'. */
4101 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
4102 #define AT_STRINGS_END(d) ((d) == end2)
4103
4104
4105 /* Test if D points to a character which is word-constituent. We have
4106 two special cases to check for: if past the end of string1, look at
4107 the first character in string2; and if before the beginning of
4108 string2, look at the last character in string1. */
4109 #define WORDCHAR_P(d) \
4110 (SYNTAX ((d) == end1 ? *string2 \
4111 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
4112 == Sword)
4113
4114 /* Disabled due to a compiler bug -- see comment at case wordbound */
4115
4116 /* The comment at case wordbound is following one, but we don't use
4117 AT_WORD_BOUNDARY anymore to support multibyte form.
4118
4119 The DEC Alpha C compiler 3.x generates incorrect code for the
4120 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4121 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4122 macro and introducing temporary variables works around the bug. */
4123
4124 #if 0
4125 /* Test if the character before D and the one at D differ with respect
4126 to being word-constituent. */
4127 #define AT_WORD_BOUNDARY(d) \
4128 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
4129 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
4130 #endif
4131
4132 /* Free everything we malloc. */
4133 #ifdef MATCH_MAY_ALLOCATE
4134 # define FREE_VAR(var) if (var) { REGEX_FREE (var); var = NULL; } else
4135 # define FREE_VARIABLES() \
4136 do { \
4137 REGEX_FREE_STACK (fail_stack.stack); \
4138 FREE_VAR (regstart); \
4139 FREE_VAR (regend); \
4140 FREE_VAR (best_regstart); \
4141 FREE_VAR (best_regend); \
4142 } while (0)
4143 #else
4144 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
4145 #endif /* not MATCH_MAY_ALLOCATE */
4146
4147 \f
4148 /* Optimization routines. */
4149
4150 /* If the operation is a match against one or more chars,
4151 return a pointer to the next operation, else return NULL. */
4152 static re_char *
4153 skip_one_char (p)
4154 re_char *p;
4155 {
4156 switch (SWITCH_ENUM_CAST (*p++))
4157 {
4158 case anychar:
4159 break;
4160
4161 case exactn:
4162 p += *p + 1;
4163 break;
4164
4165 case charset_not:
4166 case charset:
4167 if (CHARSET_RANGE_TABLE_EXISTS_P (p - 1))
4168 {
4169 int mcnt;
4170 p = CHARSET_RANGE_TABLE (p - 1);
4171 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4172 p = CHARSET_RANGE_TABLE_END (p, mcnt);
4173 }
4174 else
4175 p += 1 + CHARSET_BITMAP_SIZE (p - 1);
4176 break;
4177
4178 case syntaxspec:
4179 case notsyntaxspec:
4180 #ifdef emacs
4181 case categoryspec:
4182 case notcategoryspec:
4183 #endif /* emacs */
4184 p++;
4185 break;
4186
4187 default:
4188 p = NULL;
4189 }
4190 return p;
4191 }
4192
4193
4194 /* Jump over non-matching operations. */
4195 static unsigned char *
4196 skip_noops (p, pend)
4197 unsigned char *p, *pend;
4198 {
4199 int mcnt;
4200 while (p < pend)
4201 {
4202 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p))
4203 {
4204 case start_memory:
4205 case stop_memory:
4206 p += 2; break;
4207 case no_op:
4208 p += 1; break;
4209 case jump:
4210 p += 1;
4211 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4212 p += mcnt;
4213 break;
4214 default:
4215 return p;
4216 }
4217 }
4218 assert (p == pend);
4219 return p;
4220 }
4221
4222 /* Non-zero if "p1 matches something" implies "p2 fails". */
4223 static int
4224 mutually_exclusive_p (bufp, p1, p2)
4225 struct re_pattern_buffer *bufp;
4226 unsigned char *p1, *p2;
4227 {
4228 re_opcode_t op2;
4229 const boolean multibyte = RE_MULTIBYTE_P (bufp);
4230 unsigned char *pend = bufp->buffer + bufp->used;
4231
4232 assert (p1 >= bufp->buffer && p1 < pend
4233 && p2 >= bufp->buffer && p2 <= pend);
4234
4235 /* Skip over open/close-group commands.
4236 If what follows this loop is a ...+ construct,
4237 look at what begins its body, since we will have to
4238 match at least one of that. */
4239 p2 = skip_noops (p2, pend);
4240 /* The same skip can be done for p1, except that this function
4241 is only used in the case where p1 is a simple match operator. */
4242 /* p1 = skip_noops (p1, pend); */
4243
4244 assert (p1 >= bufp->buffer && p1 < pend
4245 && p2 >= bufp->buffer && p2 <= pend);
4246
4247 op2 = p2 == pend ? succeed : *p2;
4248
4249 switch (SWITCH_ENUM_CAST (op2))
4250 {
4251 case succeed:
4252 case endbuf:
4253 /* If we're at the end of the pattern, we can change. */
4254 if (skip_one_char (p1))
4255 {
4256 DEBUG_PRINT1 (" End of pattern: fast loop.\n");
4257 return 1;
4258 }
4259 break;
4260
4261 case endline:
4262 case exactn:
4263 {
4264 register re_wchar_t c
4265 = (re_opcode_t) *p2 == endline ? '\n'
4266 : RE_STRING_CHAR(p2 + 2, pend - p2 - 2);
4267
4268 if ((re_opcode_t) *p1 == exactn)
4269 {
4270 if (c != RE_STRING_CHAR (p1 + 2, pend - p1 - 2))
4271 {
4272 DEBUG_PRINT3 (" '%c' != '%c' => fast loop.\n", c, p1[2]);
4273 return 1;
4274 }
4275 }
4276
4277 else if ((re_opcode_t) *p1 == charset
4278 || (re_opcode_t) *p1 == charset_not)
4279 {
4280 int not = (re_opcode_t) *p1 == charset_not;
4281
4282 /* Test if C is listed in charset (or charset_not)
4283 at `p1'. */
4284 if (SINGLE_BYTE_CHAR_P (c))
4285 {
4286 if (c < CHARSET_BITMAP_SIZE (p1) * BYTEWIDTH
4287 && p1[2 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4288 not = !not;
4289 }
4290 else if (CHARSET_RANGE_TABLE_EXISTS_P (p1))
4291 CHARSET_LOOKUP_RANGE_TABLE (not, c, p1);
4292
4293 /* `not' is equal to 1 if c would match, which means
4294 that we can't change to pop_failure_jump. */
4295 if (!not)
4296 {
4297 DEBUG_PRINT1 (" No match => fast loop.\n");
4298 return 1;
4299 }
4300 }
4301 else if ((re_opcode_t) *p1 == anychar
4302 && c == '\n')
4303 {
4304 DEBUG_PRINT1 (" . != \\n => fast loop.\n");
4305 return 1;
4306 }
4307 }
4308 break;
4309
4310 case charset:
4311 case charset_not:
4312 {
4313 if ((re_opcode_t) *p1 == exactn)
4314 /* Reuse the code above. */
4315 return mutually_exclusive_p (bufp, p2, p1);
4316
4317
4318 /* It is hard to list up all the character in charset
4319 P2 if it includes multibyte character. Give up in
4320 such case. */
4321 else if (!multibyte || !CHARSET_RANGE_TABLE_EXISTS_P (p2))
4322 {
4323 /* Now, we are sure that P2 has no range table.
4324 So, for the size of bitmap in P2, `p2[1]' is
4325 enough. But P1 may have range table, so the
4326 size of bitmap table of P1 is extracted by
4327 using macro `CHARSET_BITMAP_SIZE'.
4328
4329 Since we know that all the character listed in
4330 P2 is ASCII, it is enough to test only bitmap
4331 table of P1. */
4332
4333 if (*p1 == *p2)
4334 {
4335 int idx;
4336 /* We win if the charset inside the loop
4337 has no overlap with the one after the loop. */
4338 for (idx = 0;
4339 (idx < (int) p2[1]
4340 && idx < CHARSET_BITMAP_SIZE (p1));
4341 idx++)
4342 if ((p2[2 + idx] & p1[2 + idx]) != 0)
4343 break;
4344
4345 if (idx == p2[1]
4346 || idx == CHARSET_BITMAP_SIZE (p1))
4347 {
4348 DEBUG_PRINT1 (" No match => fast loop.\n");
4349 return 1;
4350 }
4351 }
4352 else if ((re_opcode_t) *p1 == charset
4353 || (re_opcode_t) *p1 == charset_not)
4354 {
4355 int idx;
4356 /* We win if the charset_not inside the loop lists
4357 every character listed in the charset after. */
4358 for (idx = 0; idx < (int) p2[1]; idx++)
4359 if (! (p2[2 + idx] == 0
4360 || (idx < CHARSET_BITMAP_SIZE (p1)
4361 && ((p2[2 + idx] & ~ p1[2 + idx]) == 0))))
4362 break;
4363
4364 if (idx == p2[1])
4365 {
4366 DEBUG_PRINT1 (" No match => fast loop.\n");
4367 return 1;
4368 }
4369 }
4370 }
4371 }
4372
4373 case wordend:
4374 case notsyntaxspec:
4375 return ((re_opcode_t) *p1 == syntaxspec
4376 && p1[1] == (op2 == wordend ? Sword : p2[1]));
4377
4378 case wordbeg:
4379 case syntaxspec:
4380 return ((re_opcode_t) *p1 == notsyntaxspec
4381 && p1[1] == (op2 == wordend ? Sword : p2[1]));
4382
4383 case wordbound:
4384 return (((re_opcode_t) *p1 == notsyntaxspec
4385 || (re_opcode_t) *p1 == syntaxspec)
4386 && p1[1] == Sword);
4387
4388 #ifdef emacs
4389 case categoryspec:
4390 return ((re_opcode_t) *p1 == notcategoryspec && p1[1] == p2[1]);
4391 case notcategoryspec:
4392 return ((re_opcode_t) *p1 == categoryspec && p1[1] == p2[1]);
4393 #endif /* emacs */
4394
4395 default:
4396 ;
4397 }
4398
4399 /* Safe default. */
4400 return 0;
4401 }
4402
4403 \f
4404 /* Matching routines. */
4405
4406 #ifndef emacs /* Emacs never uses this. */
4407 /* re_match is like re_match_2 except it takes only a single string. */
4408
4409 int
4410 re_match (bufp, string, size, pos, regs)
4411 struct re_pattern_buffer *bufp;
4412 const char *string;
4413 int size, pos;
4414 struct re_registers *regs;
4415 {
4416 int result = re_match_2_internal (bufp, NULL, 0, (re_char*) string, size,
4417 pos, regs, size);
4418 # if defined C_ALLOCA && !defined REGEX_MALLOC
4419 alloca (0);
4420 # endif
4421 return result;
4422 }
4423 WEAK_ALIAS (__re_match, re_match)
4424 #endif /* not emacs */
4425
4426 #ifdef emacs
4427 /* In Emacs, this is the string or buffer in which we
4428 are matching. It is used for looking up syntax properties. */
4429 Lisp_Object re_match_object;
4430 #endif
4431
4432 /* re_match_2 matches the compiled pattern in BUFP against the
4433 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
4434 and SIZE2, respectively). We start matching at POS, and stop
4435 matching at STOP.
4436
4437 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
4438 store offsets for the substring each group matched in REGS. See the
4439 documentation for exactly how many groups we fill.
4440
4441 We return -1 if no match, -2 if an internal error (such as the
4442 failure stack overflowing). Otherwise, we return the length of the
4443 matched substring. */
4444
4445 int
4446 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
4447 struct re_pattern_buffer *bufp;
4448 const char *string1, *string2;
4449 int size1, size2;
4450 int pos;
4451 struct re_registers *regs;
4452 int stop;
4453 {
4454 int result;
4455
4456 #ifdef emacs
4457 int charpos;
4458 gl_state.object = re_match_object;
4459 charpos = SYNTAX_TABLE_BYTE_TO_CHAR (POS_AS_IN_BUFFER (pos));
4460 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, charpos, 1);
4461 #endif
4462
4463 result = re_match_2_internal (bufp, (re_char*) string1, size1,
4464 (re_char*) string2, size2,
4465 pos, regs, stop);
4466 #if defined C_ALLOCA && !defined REGEX_MALLOC
4467 alloca (0);
4468 #endif
4469 return result;
4470 }
4471 WEAK_ALIAS (__re_match_2, re_match_2)
4472
4473 /* This is a separate function so that we can force an alloca cleanup
4474 afterwards. */
4475 static int
4476 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
4477 struct re_pattern_buffer *bufp;
4478 re_char *string1, *string2;
4479 int size1, size2;
4480 int pos;
4481 struct re_registers *regs;
4482 int stop;
4483 {
4484 /* General temporaries. */
4485 int mcnt;
4486 size_t reg;
4487 boolean not;
4488
4489 /* Just past the end of the corresponding string. */
4490 re_char *end1, *end2;
4491
4492 /* Pointers into string1 and string2, just past the last characters in
4493 each to consider matching. */
4494 re_char *end_match_1, *end_match_2;
4495
4496 /* Where we are in the data, and the end of the current string. */
4497 re_char *d, *dend;
4498
4499 /* Used sometimes to remember where we were before starting matching
4500 an operator so that we can go back in case of failure. This "atomic"
4501 behavior of matching opcodes is indispensable to the correctness
4502 of the on_failure_keep_string_jump optimization. */
4503 re_char *dfail;
4504
4505 /* Where we are in the pattern, and the end of the pattern. */
4506 re_char *p = bufp->buffer;
4507 re_char *pend = p + bufp->used;
4508
4509 /* We use this to map every character in the string. */
4510 RE_TRANSLATE_TYPE translate = bufp->translate;
4511
4512 /* Nonzero if we have to concern multibyte character. */
4513 const boolean multibyte = RE_MULTIBYTE_P (bufp);
4514
4515 /* Failure point stack. Each place that can handle a failure further
4516 down the line pushes a failure point on this stack. It consists of
4517 regstart, and regend for all registers corresponding to
4518 the subexpressions we're currently inside, plus the number of such
4519 registers, and, finally, two char *'s. The first char * is where
4520 to resume scanning the pattern; the second one is where to resume
4521 scanning the strings. */
4522 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
4523 fail_stack_type fail_stack;
4524 #endif
4525 #ifdef DEBUG
4526 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
4527 #endif
4528
4529 #if defined REL_ALLOC && defined REGEX_MALLOC
4530 /* This holds the pointer to the failure stack, when
4531 it is allocated relocatably. */
4532 fail_stack_elt_t *failure_stack_ptr;
4533 #endif
4534
4535 /* We fill all the registers internally, independent of what we
4536 return, for use in backreferences. The number here includes
4537 an element for register zero. */
4538 size_t num_regs = bufp->re_nsub + 1;
4539
4540 /* Information on the contents of registers. These are pointers into
4541 the input strings; they record just what was matched (on this
4542 attempt) by a subexpression part of the pattern, that is, the
4543 regnum-th regstart pointer points to where in the pattern we began
4544 matching and the regnum-th regend points to right after where we
4545 stopped matching the regnum-th subexpression. (The zeroth register
4546 keeps track of what the whole pattern matches.) */
4547 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4548 re_char **regstart, **regend;
4549 #endif
4550
4551 /* The following record the register info as found in the above
4552 variables when we find a match better than any we've seen before.
4553 This happens as we backtrack through the failure points, which in
4554 turn happens only if we have not yet matched the entire string. */
4555 unsigned best_regs_set = false;
4556 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4557 re_char **best_regstart, **best_regend;
4558 #endif
4559
4560 /* Logically, this is `best_regend[0]'. But we don't want to have to
4561 allocate space for that if we're not allocating space for anything
4562 else (see below). Also, we never need info about register 0 for
4563 any of the other register vectors, and it seems rather a kludge to
4564 treat `best_regend' differently than the rest. So we keep track of
4565 the end of the best match so far in a separate variable. We
4566 initialize this to NULL so that when we backtrack the first time
4567 and need to test it, it's not garbage. */
4568 re_char *match_end = NULL;
4569
4570 #ifdef DEBUG
4571 /* Counts the total number of registers pushed. */
4572 unsigned num_regs_pushed = 0;
4573 #endif
4574
4575 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
4576
4577 INIT_FAIL_STACK ();
4578
4579 #ifdef MATCH_MAY_ALLOCATE
4580 /* Do not bother to initialize all the register variables if there are
4581 no groups in the pattern, as it takes a fair amount of time. If
4582 there are groups, we include space for register 0 (the whole
4583 pattern), even though we never use it, since it simplifies the
4584 array indexing. We should fix this. */
4585 if (bufp->re_nsub)
4586 {
4587 regstart = REGEX_TALLOC (num_regs, re_char *);
4588 regend = REGEX_TALLOC (num_regs, re_char *);
4589 best_regstart = REGEX_TALLOC (num_regs, re_char *);
4590 best_regend = REGEX_TALLOC (num_regs, re_char *);
4591
4592 if (!(regstart && regend && best_regstart && best_regend))
4593 {
4594 FREE_VARIABLES ();
4595 return -2;
4596 }
4597 }
4598 else
4599 {
4600 /* We must initialize all our variables to NULL, so that
4601 `FREE_VARIABLES' doesn't try to free them. */
4602 regstart = regend = best_regstart = best_regend = NULL;
4603 }
4604 #endif /* MATCH_MAY_ALLOCATE */
4605
4606 /* The starting position is bogus. */
4607 if (pos < 0 || pos > size1 + size2)
4608 {
4609 FREE_VARIABLES ();
4610 return -1;
4611 }
4612
4613 /* Initialize subexpression text positions to -1 to mark ones that no
4614 start_memory/stop_memory has been seen for. Also initialize the
4615 register information struct. */
4616 for (reg = 1; reg < num_regs; reg++)
4617 regstart[reg] = regend[reg] = NULL;
4618
4619 /* We move `string1' into `string2' if the latter's empty -- but not if
4620 `string1' is null. */
4621 if (size2 == 0 && string1 != NULL)
4622 {
4623 string2 = string1;
4624 size2 = size1;
4625 string1 = 0;
4626 size1 = 0;
4627 }
4628 end1 = string1 + size1;
4629 end2 = string2 + size2;
4630
4631 /* `p' scans through the pattern as `d' scans through the data.
4632 `dend' is the end of the input string that `d' points within. `d'
4633 is advanced into the following input string whenever necessary, but
4634 this happens before fetching; therefore, at the beginning of the
4635 loop, `d' can be pointing at the end of a string, but it cannot
4636 equal `string2'. */
4637 if (pos >= size1)
4638 {
4639 /* Only match within string2. */
4640 d = string2 + pos - size1;
4641 dend = end_match_2 = string2 + stop - size1;
4642 end_match_1 = end1; /* Just to give it a value. */
4643 }
4644 else
4645 {
4646 if (stop < size1)
4647 {
4648 /* Only match within string1. */
4649 end_match_1 = string1 + stop;
4650 /* BEWARE!
4651 When we reach end_match_1, PREFETCH normally switches to string2.
4652 But in the present case, this means that just doing a PREFETCH
4653 makes us jump from `stop' to `gap' within the string.
4654 What we really want here is for the search to stop as
4655 soon as we hit end_match_1. That's why we set end_match_2
4656 to end_match_1 (since PREFETCH fails as soon as we hit
4657 end_match_2). */
4658 end_match_2 = end_match_1;
4659 }
4660 else
4661 { /* It's important to use this code when stop == size so that
4662 moving `d' from end1 to string2 will not prevent the d == dend
4663 check from catching the end of string. */
4664 end_match_1 = end1;
4665 end_match_2 = string2 + stop - size1;
4666 }
4667 d = string1 + pos;
4668 dend = end_match_1;
4669 }
4670
4671 DEBUG_PRINT1 ("The compiled pattern is: ");
4672 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
4673 DEBUG_PRINT1 ("The string to match is: `");
4674 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
4675 DEBUG_PRINT1 ("'\n");
4676
4677 /* This loops over pattern commands. It exits by returning from the
4678 function if the match is complete, or it drops through if the match
4679 fails at this starting point in the input data. */
4680 for (;;)
4681 {
4682 DEBUG_PRINT2 ("\n%p: ", p);
4683
4684 if (p == pend)
4685 { /* End of pattern means we might have succeeded. */
4686 DEBUG_PRINT1 ("end of pattern ... ");
4687
4688 /* If we haven't matched the entire string, and we want the
4689 longest match, try backtracking. */
4690 if (d != end_match_2)
4691 {
4692 /* 1 if this match ends in the same string (string1 or string2)
4693 as the best previous match. */
4694 boolean same_str_p = (FIRST_STRING_P (match_end)
4695 == FIRST_STRING_P (d));
4696 /* 1 if this match is the best seen so far. */
4697 boolean best_match_p;
4698
4699 /* AIX compiler got confused when this was combined
4700 with the previous declaration. */
4701 if (same_str_p)
4702 best_match_p = d > match_end;
4703 else
4704 best_match_p = !FIRST_STRING_P (d);
4705
4706 DEBUG_PRINT1 ("backtracking.\n");
4707
4708 if (!FAIL_STACK_EMPTY ())
4709 { /* More failure points to try. */
4710
4711 /* If exceeds best match so far, save it. */
4712 if (!best_regs_set || best_match_p)
4713 {
4714 best_regs_set = true;
4715 match_end = d;
4716
4717 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
4718
4719 for (reg = 1; reg < num_regs; reg++)
4720 {
4721 best_regstart[reg] = regstart[reg];
4722 best_regend[reg] = regend[reg];
4723 }
4724 }
4725 goto fail;
4726 }
4727
4728 /* If no failure points, don't restore garbage. And if
4729 last match is real best match, don't restore second
4730 best one. */
4731 else if (best_regs_set && !best_match_p)
4732 {
4733 restore_best_regs:
4734 /* Restore best match. It may happen that `dend ==
4735 end_match_1' while the restored d is in string2.
4736 For example, the pattern `x.*y.*z' against the
4737 strings `x-' and `y-z-', if the two strings are
4738 not consecutive in memory. */
4739 DEBUG_PRINT1 ("Restoring best registers.\n");
4740
4741 d = match_end;
4742 dend = ((d >= string1 && d <= end1)
4743 ? end_match_1 : end_match_2);
4744
4745 for (reg = 1; reg < num_regs; reg++)
4746 {
4747 regstart[reg] = best_regstart[reg];
4748 regend[reg] = best_regend[reg];
4749 }
4750 }
4751 } /* d != end_match_2 */
4752
4753 succeed_label:
4754 DEBUG_PRINT1 ("Accepting match.\n");
4755
4756 /* If caller wants register contents data back, do it. */
4757 if (regs && !bufp->no_sub)
4758 {
4759 /* Have the register data arrays been allocated? */
4760 if (bufp->regs_allocated == REGS_UNALLOCATED)
4761 { /* No. So allocate them with malloc. We need one
4762 extra element beyond `num_regs' for the `-1' marker
4763 GNU code uses. */
4764 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
4765 regs->start = TALLOC (regs->num_regs, regoff_t);
4766 regs->end = TALLOC (regs->num_regs, regoff_t);
4767 if (regs->start == NULL || regs->end == NULL)
4768 {
4769 FREE_VARIABLES ();
4770 return -2;
4771 }
4772 bufp->regs_allocated = REGS_REALLOCATE;
4773 }
4774 else if (bufp->regs_allocated == REGS_REALLOCATE)
4775 { /* Yes. If we need more elements than were already
4776 allocated, reallocate them. If we need fewer, just
4777 leave it alone. */
4778 if (regs->num_regs < num_regs + 1)
4779 {
4780 regs->num_regs = num_regs + 1;
4781 RETALLOC (regs->start, regs->num_regs, regoff_t);
4782 RETALLOC (regs->end, regs->num_regs, regoff_t);
4783 if (regs->start == NULL || regs->end == NULL)
4784 {
4785 FREE_VARIABLES ();
4786 return -2;
4787 }
4788 }
4789 }
4790 else
4791 {
4792 /* These braces fend off a "empty body in an else-statement"
4793 warning under GCC when assert expands to nothing. */
4794 assert (bufp->regs_allocated == REGS_FIXED);
4795 }
4796
4797 /* Convert the pointer data in `regstart' and `regend' to
4798 indices. Register zero has to be set differently,
4799 since we haven't kept track of any info for it. */
4800 if (regs->num_regs > 0)
4801 {
4802 regs->start[0] = pos;
4803 regs->end[0] = POINTER_TO_OFFSET (d);
4804 }
4805
4806 /* Go through the first `min (num_regs, regs->num_regs)'
4807 registers, since that is all we initialized. */
4808 for (reg = 1; reg < MIN (num_regs, regs->num_regs); reg++)
4809 {
4810 if (REG_UNSET (regstart[reg]) || REG_UNSET (regend[reg]))
4811 regs->start[reg] = regs->end[reg] = -1;
4812 else
4813 {
4814 regs->start[reg]
4815 = (regoff_t) POINTER_TO_OFFSET (regstart[reg]);
4816 regs->end[reg]
4817 = (regoff_t) POINTER_TO_OFFSET (regend[reg]);
4818 }
4819 }
4820
4821 /* If the regs structure we return has more elements than
4822 were in the pattern, set the extra elements to -1. If
4823 we (re)allocated the registers, this is the case,
4824 because we always allocate enough to have at least one
4825 -1 at the end. */
4826 for (reg = num_regs; reg < regs->num_regs; reg++)
4827 regs->start[reg] = regs->end[reg] = -1;
4828 } /* regs && !bufp->no_sub */
4829
4830 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
4831 nfailure_points_pushed, nfailure_points_popped,
4832 nfailure_points_pushed - nfailure_points_popped);
4833 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
4834
4835 mcnt = POINTER_TO_OFFSET (d) - pos;
4836
4837 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
4838
4839 FREE_VARIABLES ();
4840 return mcnt;
4841 }
4842
4843 /* Otherwise match next pattern command. */
4844 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4845 {
4846 /* Ignore these. Used to ignore the n of succeed_n's which
4847 currently have n == 0. */
4848 case no_op:
4849 DEBUG_PRINT1 ("EXECUTING no_op.\n");
4850 break;
4851
4852 case succeed:
4853 DEBUG_PRINT1 ("EXECUTING succeed.\n");
4854 goto succeed_label;
4855
4856 /* Match the next n pattern characters exactly. The following
4857 byte in the pattern defines n, and the n bytes after that
4858 are the characters to match. */
4859 case exactn:
4860 mcnt = *p++;
4861 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
4862
4863 /* Remember the start point to rollback upon failure. */
4864 dfail = d;
4865
4866 /* This is written out as an if-else so we don't waste time
4867 testing `translate' inside the loop. */
4868 if (RE_TRANSLATE_P (translate))
4869 {
4870 if (multibyte)
4871 do
4872 {
4873 int pat_charlen, buf_charlen;
4874 unsigned int pat_ch, buf_ch;
4875
4876 PREFETCH ();
4877 pat_ch = STRING_CHAR_AND_LENGTH (p, pend - p, pat_charlen);
4878 buf_ch = STRING_CHAR_AND_LENGTH (d, dend - d, buf_charlen);
4879
4880 if (RE_TRANSLATE (translate, buf_ch)
4881 != pat_ch)
4882 {
4883 d = dfail;
4884 goto fail;
4885 }
4886
4887 p += pat_charlen;
4888 d += buf_charlen;
4889 mcnt -= pat_charlen;
4890 }
4891 while (mcnt > 0);
4892 else
4893 do
4894 {
4895 PREFETCH ();
4896 if (RE_TRANSLATE (translate, *d) != *p++)
4897 {
4898 d = dfail;
4899 goto fail;
4900 }
4901 d++;
4902 }
4903 while (--mcnt);
4904 }
4905 else
4906 {
4907 do
4908 {
4909 PREFETCH ();
4910 if (*d++ != *p++)
4911 {
4912 d = dfail;
4913 goto fail;
4914 }
4915 }
4916 while (--mcnt);
4917 }
4918 break;
4919
4920
4921 /* Match any character except possibly a newline or a null. */
4922 case anychar:
4923 {
4924 int buf_charlen;
4925 re_wchar_t buf_ch;
4926
4927 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4928
4929 PREFETCH ();
4930 buf_ch = RE_STRING_CHAR_AND_LENGTH (d, dend - d, buf_charlen);
4931 buf_ch = TRANSLATE (buf_ch);
4932
4933 if ((!(bufp->syntax & RE_DOT_NEWLINE)
4934 && buf_ch == '\n')
4935 || ((bufp->syntax & RE_DOT_NOT_NULL)
4936 && buf_ch == '\000'))
4937 goto fail;
4938
4939 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4940 d += buf_charlen;
4941 }
4942 break;
4943
4944
4945 case charset:
4946 case charset_not:
4947 {
4948 register unsigned int c;
4949 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4950 int len;
4951
4952 /* Start of actual range_table, or end of bitmap if there is no
4953 range table. */
4954 re_char *range_table;
4955
4956 /* Nonzero if there is a range table. */
4957 int range_table_exists;
4958
4959 /* Number of ranges of range table. This is not included
4960 in the initial byte-length of the command. */
4961 int count = 0;
4962
4963 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4964
4965 range_table_exists = CHARSET_RANGE_TABLE_EXISTS_P (&p[-1]);
4966
4967 if (range_table_exists)
4968 {
4969 range_table = CHARSET_RANGE_TABLE (&p[-1]); /* Past the bitmap. */
4970 EXTRACT_NUMBER_AND_INCR (count, range_table);
4971 }
4972
4973 PREFETCH ();
4974 c = RE_STRING_CHAR_AND_LENGTH (d, dend - d, len);
4975 c = TRANSLATE (c); /* The character to match. */
4976
4977 if (SINGLE_BYTE_CHAR_P (c))
4978 { /* Lookup bitmap. */
4979 /* Cast to `unsigned' instead of `unsigned char' in
4980 case the bit list is a full 32 bytes long. */
4981 if (c < (unsigned) (CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH)
4982 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4983 not = !not;
4984 }
4985 #ifdef emacs
4986 else if (range_table_exists)
4987 {
4988 int class_bits = CHARSET_RANGE_TABLE_BITS (&p[-1]);
4989
4990 if ( (class_bits & BIT_LOWER && ISLOWER (c))
4991 | (class_bits & BIT_MULTIBYTE)
4992 | (class_bits & BIT_PUNCT && ISPUNCT (c))
4993 | (class_bits & BIT_SPACE && ISSPACE (c))
4994 | (class_bits & BIT_UPPER && ISUPPER (c))
4995 | (class_bits & BIT_WORD && ISWORD (c)))
4996 not = !not;
4997 else
4998 CHARSET_LOOKUP_RANGE_TABLE_RAW (not, c, range_table, count);
4999 }
5000 #endif /* emacs */
5001
5002 if (range_table_exists)
5003 p = CHARSET_RANGE_TABLE_END (range_table, count);
5004 else
5005 p += CHARSET_BITMAP_SIZE (&p[-1]) + 1;
5006
5007 if (!not) goto fail;
5008
5009 d += len;
5010 break;
5011 }
5012
5013
5014 /* The beginning of a group is represented by start_memory.
5015 The argument is the register number. The text
5016 matched within the group is recorded (in the internal
5017 registers data structure) under the register number. */
5018 case start_memory:
5019 DEBUG_PRINT2 ("EXECUTING start_memory %d:\n", *p);
5020
5021 /* In case we need to undo this operation (via backtracking). */
5022 PUSH_FAILURE_REG ((unsigned int)*p);
5023
5024 regstart[*p] = d;
5025 regend[*p] = NULL; /* probably unnecessary. -sm */
5026 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
5027
5028 /* Move past the register number and inner group count. */
5029 p += 1;
5030 break;
5031
5032
5033 /* The stop_memory opcode represents the end of a group. Its
5034 argument is the same as start_memory's: the register number. */
5035 case stop_memory:
5036 DEBUG_PRINT2 ("EXECUTING stop_memory %d:\n", *p);
5037
5038 assert (!REG_UNSET (regstart[*p]));
5039 /* Strictly speaking, there should be code such as:
5040
5041 assert (REG_UNSET (regend[*p]));
5042 PUSH_FAILURE_REGSTOP ((unsigned int)*p);
5043
5044 But the only info to be pushed is regend[*p] and it is known to
5045 be UNSET, so there really isn't anything to push.
5046 Not pushing anything, on the other hand deprives us from the
5047 guarantee that regend[*p] is UNSET since undoing this operation
5048 will not reset its value properly. This is not important since
5049 the value will only be read on the next start_memory or at
5050 the very end and both events can only happen if this stop_memory
5051 is *not* undone. */
5052
5053 regend[*p] = d;
5054 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
5055
5056 /* Move past the register number and the inner group count. */
5057 p += 1;
5058 break;
5059
5060
5061 /* \<digit> has been turned into a `duplicate' command which is
5062 followed by the numeric value of <digit> as the register number. */
5063 case duplicate:
5064 {
5065 register re_char *d2, *dend2;
5066 int regno = *p++; /* Get which register to match against. */
5067 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
5068
5069 /* Can't back reference a group which we've never matched. */
5070 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
5071 goto fail;
5072
5073 /* Where in input to try to start matching. */
5074 d2 = regstart[regno];
5075
5076 /* Remember the start point to rollback upon failure. */
5077 dfail = d;
5078
5079 /* Where to stop matching; if both the place to start and
5080 the place to stop matching are in the same string, then
5081 set to the place to stop, otherwise, for now have to use
5082 the end of the first string. */
5083
5084 dend2 = ((FIRST_STRING_P (regstart[regno])
5085 == FIRST_STRING_P (regend[regno]))
5086 ? regend[regno] : end_match_1);
5087 for (;;)
5088 {
5089 /* If necessary, advance to next segment in register
5090 contents. */
5091 while (d2 == dend2)
5092 {
5093 if (dend2 == end_match_2) break;
5094 if (dend2 == regend[regno]) break;
5095
5096 /* End of string1 => advance to string2. */
5097 d2 = string2;
5098 dend2 = regend[regno];
5099 }
5100 /* At end of register contents => success */
5101 if (d2 == dend2) break;
5102
5103 /* If necessary, advance to next segment in data. */
5104 PREFETCH ();
5105
5106 /* How many characters left in this segment to match. */
5107 mcnt = dend - d;
5108
5109 /* Want how many consecutive characters we can match in
5110 one shot, so, if necessary, adjust the count. */
5111 if (mcnt > dend2 - d2)
5112 mcnt = dend2 - d2;
5113
5114 /* Compare that many; failure if mismatch, else move
5115 past them. */
5116 if (RE_TRANSLATE_P (translate)
5117 ? bcmp_translate (d, d2, mcnt, translate, multibyte)
5118 : memcmp (d, d2, mcnt))
5119 {
5120 d = dfail;
5121 goto fail;
5122 }
5123 d += mcnt, d2 += mcnt;
5124 }
5125 }
5126 break;
5127
5128
5129 /* begline matches the empty string at the beginning of the string
5130 (unless `not_bol' is set in `bufp'), and after newlines. */
5131 case begline:
5132 DEBUG_PRINT1 ("EXECUTING begline.\n");
5133
5134 if (AT_STRINGS_BEG (d))
5135 {
5136 if (!bufp->not_bol) break;
5137 }
5138 else
5139 {
5140 unsigned char c;
5141 GET_CHAR_BEFORE_2 (c, d, string1, end1, string2, end2);
5142 if (c == '\n')
5143 break;
5144 }
5145 /* In all other cases, we fail. */
5146 goto fail;
5147
5148
5149 /* endline is the dual of begline. */
5150 case endline:
5151 DEBUG_PRINT1 ("EXECUTING endline.\n");
5152
5153 if (AT_STRINGS_END (d))
5154 {
5155 if (!bufp->not_eol) break;
5156 }
5157 else
5158 {
5159 PREFETCH_NOLIMIT ();
5160 if (*d == '\n')
5161 break;
5162 }
5163 goto fail;
5164
5165
5166 /* Match at the very beginning of the data. */
5167 case begbuf:
5168 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
5169 if (AT_STRINGS_BEG (d))
5170 break;
5171 goto fail;
5172
5173
5174 /* Match at the very end of the data. */
5175 case endbuf:
5176 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
5177 if (AT_STRINGS_END (d))
5178 break;
5179 goto fail;
5180
5181
5182 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
5183 pushes NULL as the value for the string on the stack. Then
5184 `POP_FAILURE_POINT' will keep the current value for the
5185 string, instead of restoring it. To see why, consider
5186 matching `foo\nbar' against `.*\n'. The .* matches the foo;
5187 then the . fails against the \n. But the next thing we want
5188 to do is match the \n against the \n; if we restored the
5189 string value, we would be back at the foo.
5190
5191 Because this is used only in specific cases, we don't need to
5192 check all the things that `on_failure_jump' does, to make
5193 sure the right things get saved on the stack. Hence we don't
5194 share its code. The only reason to push anything on the
5195 stack at all is that otherwise we would have to change
5196 `anychar's code to do something besides goto fail in this
5197 case; that seems worse than this. */
5198 case on_failure_keep_string_jump:
5199 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5200 DEBUG_PRINT3 ("EXECUTING on_failure_keep_string_jump %d (to %p):\n",
5201 mcnt, p + mcnt);
5202
5203 PUSH_FAILURE_POINT (p - 3, NULL);
5204 break;
5205
5206 /* A nasty loop is introduced by the non-greedy *? and +?.
5207 With such loops, the stack only ever contains one failure point
5208 at a time, so that a plain on_failure_jump_loop kind of
5209 cycle detection cannot work. Worse yet, such a detection
5210 can not only fail to detect a cycle, but it can also wrongly
5211 detect a cycle (between different instantiations of the same
5212 loop.
5213 So the method used for those nasty loops is a little different:
5214 We use a special cycle-detection-stack-frame which is pushed
5215 when the on_failure_jump_nastyloop failure-point is *popped*.
5216 This special frame thus marks the beginning of one iteration
5217 through the loop and we can hence easily check right here
5218 whether something matched between the beginning and the end of
5219 the loop. */
5220 case on_failure_jump_nastyloop:
5221 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5222 DEBUG_PRINT3 ("EXECUTING on_failure_jump_nastyloop %d (to %p):\n",
5223 mcnt, p + mcnt);
5224
5225 assert ((re_opcode_t)p[-4] == no_op);
5226 CHECK_INFINITE_LOOP (p - 4, d);
5227 PUSH_FAILURE_POINT (p - 3, d);
5228 break;
5229
5230
5231 /* Simple loop detecting on_failure_jump: just check on the
5232 failure stack if the same spot was already hit earlier. */
5233 case on_failure_jump_loop:
5234 on_failure:
5235 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5236 DEBUG_PRINT3 ("EXECUTING on_failure_jump_loop %d (to %p):\n",
5237 mcnt, p + mcnt);
5238
5239 CHECK_INFINITE_LOOP (p - 3, d);
5240 PUSH_FAILURE_POINT (p - 3, d);
5241 break;
5242
5243
5244 /* Uses of on_failure_jump:
5245
5246 Each alternative starts with an on_failure_jump that points
5247 to the beginning of the next alternative. Each alternative
5248 except the last ends with a jump that in effect jumps past
5249 the rest of the alternatives. (They really jump to the
5250 ending jump of the following alternative, because tensioning
5251 these jumps is a hassle.)
5252
5253 Repeats start with an on_failure_jump that points past both
5254 the repetition text and either the following jump or
5255 pop_failure_jump back to this on_failure_jump. */
5256 case on_failure_jump:
5257 IMMEDIATE_QUIT_CHECK;
5258 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5259 DEBUG_PRINT3 ("EXECUTING on_failure_jump %d (to %p):\n",
5260 mcnt, p + mcnt);
5261
5262 PUSH_FAILURE_POINT (p -3, d);
5263 break;
5264
5265 /* This operation is used for greedy *.
5266 Compare the beginning of the repeat with what in the
5267 pattern follows its end. If we can establish that there
5268 is nothing that they would both match, i.e., that we
5269 would have to backtrack because of (as in, e.g., `a*a')
5270 then we can use a non-backtracking loop based on
5271 on_failure_keep_string_jump instead of on_failure_jump. */
5272 case on_failure_jump_smart:
5273 IMMEDIATE_QUIT_CHECK;
5274 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5275 DEBUG_PRINT3 ("EXECUTING on_failure_jump_smart %d (to %p).\n",
5276 mcnt, p + mcnt);
5277 {
5278 re_char *p1 = p; /* Next operation. */
5279 /* Here, we discard `const', making re_match non-reentrant. */
5280 unsigned char *p2 = (unsigned char*) p + mcnt; /* Jump dest. */
5281 unsigned char *p3 = (unsigned char*) p - 3; /* opcode location. */
5282
5283 p -= 3; /* Reset so that we will re-execute the
5284 instruction once it's been changed. */
5285
5286 EXTRACT_NUMBER (mcnt, p2 - 2);
5287
5288 /* Ensure this is a indeed the trivial kind of loop
5289 we are expecting. */
5290 assert (skip_one_char (p1) == p2 - 3);
5291 assert ((re_opcode_t) p2[-3] == jump && p2 + mcnt == p);
5292 DEBUG_STATEMENT (debug += 2);
5293 if (mutually_exclusive_p (bufp, p1, p2))
5294 {
5295 /* Use a fast `on_failure_keep_string_jump' loop. */
5296 DEBUG_PRINT1 (" smart exclusive => fast loop.\n");
5297 *p3 = (unsigned char) on_failure_keep_string_jump;
5298 STORE_NUMBER (p2 - 2, mcnt + 3);
5299 }
5300 else
5301 {
5302 /* Default to a safe `on_failure_jump' loop. */
5303 DEBUG_PRINT1 (" smart default => slow loop.\n");
5304 *p3 = (unsigned char) on_failure_jump;
5305 }
5306 DEBUG_STATEMENT (debug -= 2);
5307 }
5308 break;
5309
5310 /* Unconditionally jump (without popping any failure points). */
5311 case jump:
5312 unconditional_jump:
5313 IMMEDIATE_QUIT_CHECK;
5314 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
5315 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
5316 p += mcnt; /* Do the jump. */
5317 DEBUG_PRINT2 ("(to %p).\n", p);
5318 break;
5319
5320
5321 /* Have to succeed matching what follows at least n times.
5322 After that, handle like `on_failure_jump'. */
5323 case succeed_n:
5324 /* Signedness doesn't matter since we only compare MCNT to 0. */
5325 EXTRACT_NUMBER (mcnt, p + 2);
5326 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
5327
5328 /* Originally, mcnt is how many times we HAVE to succeed. */
5329 if (mcnt != 0)
5330 {
5331 /* Here, we discard `const', making re_match non-reentrant. */
5332 unsigned char *p2 = (unsigned char*) p + 2; /* counter loc. */
5333 mcnt--;
5334 p += 4;
5335 PUSH_NUMBER (p2, mcnt);
5336 }
5337 else
5338 /* The two bytes encoding mcnt == 0 are two no_op opcodes. */
5339 goto on_failure;
5340 break;
5341
5342 case jump_n:
5343 /* Signedness doesn't matter since we only compare MCNT to 0. */
5344 EXTRACT_NUMBER (mcnt, p + 2);
5345 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
5346
5347 /* Originally, this is how many times we CAN jump. */
5348 if (mcnt != 0)
5349 {
5350 /* Here, we discard `const', making re_match non-reentrant. */
5351 unsigned char *p2 = (unsigned char*) p + 2; /* counter loc. */
5352 mcnt--;
5353 PUSH_NUMBER (p2, mcnt);
5354 goto unconditional_jump;
5355 }
5356 /* If don't have to jump any more, skip over the rest of command. */
5357 else
5358 p += 4;
5359 break;
5360
5361 case set_number_at:
5362 {
5363 unsigned char *p2; /* Location of the counter. */
5364 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
5365
5366 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5367 /* Here, we discard `const', making re_match non-reentrant. */
5368 p2 = (unsigned char*) p + mcnt;
5369 /* Signedness doesn't matter since we only copy MCNT's bits . */
5370 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5371 DEBUG_PRINT3 (" Setting %p to %d.\n", p2, mcnt);
5372 PUSH_NUMBER (p2, mcnt);
5373 break;
5374 }
5375
5376 case wordbound:
5377 case notwordbound:
5378 not = (re_opcode_t) *(p - 1) == notwordbound;
5379 DEBUG_PRINT2 ("EXECUTING %swordbound.\n", not?"not":"");
5380
5381 /* We SUCCEED (or FAIL) in one of the following cases: */
5382
5383 /* Case 1: D is at the beginning or the end of string. */
5384 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5385 not = !not;
5386 else
5387 {
5388 /* C1 is the character before D, S1 is the syntax of C1, C2
5389 is the character at D, and S2 is the syntax of C2. */
5390 re_wchar_t c1, c2;
5391 int s1, s2;
5392 #ifdef emacs
5393 int offset = PTR_TO_OFFSET (d - 1);
5394 int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
5395 UPDATE_SYNTAX_TABLE (charpos);
5396 #endif
5397 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
5398 s1 = SYNTAX (c1);
5399 #ifdef emacs
5400 UPDATE_SYNTAX_TABLE_FORWARD (charpos + 1);
5401 #endif
5402 PREFETCH_NOLIMIT ();
5403 c2 = RE_STRING_CHAR (d, dend - d);
5404 s2 = SYNTAX (c2);
5405
5406 if (/* Case 2: Only one of S1 and S2 is Sword. */
5407 ((s1 == Sword) != (s2 == Sword))
5408 /* Case 3: Both of S1 and S2 are Sword, and macro
5409 WORD_BOUNDARY_P (C1, C2) returns nonzero. */
5410 || ((s1 == Sword) && WORD_BOUNDARY_P (c1, c2)))
5411 not = !not;
5412 }
5413 if (not)
5414 break;
5415 else
5416 goto fail;
5417
5418 case wordbeg:
5419 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
5420
5421 /* We FAIL in one of the following cases: */
5422
5423 /* Case 1: D is at the end of string. */
5424 if (AT_STRINGS_END (d))
5425 goto fail;
5426 else
5427 {
5428 /* C1 is the character before D, S1 is the syntax of C1, C2
5429 is the character at D, and S2 is the syntax of C2. */
5430 re_wchar_t c1, c2;
5431 int s1, s2;
5432 #ifdef emacs
5433 int offset = PTR_TO_OFFSET (d);
5434 int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
5435 UPDATE_SYNTAX_TABLE (charpos);
5436 #endif
5437 PREFETCH ();
5438 c2 = RE_STRING_CHAR (d, dend - d);
5439 s2 = SYNTAX (c2);
5440
5441 /* Case 2: S2 is not Sword. */
5442 if (s2 != Sword)
5443 goto fail;
5444
5445 /* Case 3: D is not at the beginning of string ... */
5446 if (!AT_STRINGS_BEG (d))
5447 {
5448 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
5449 #ifdef emacs
5450 UPDATE_SYNTAX_TABLE_BACKWARD (charpos - 1);
5451 #endif
5452 s1 = SYNTAX (c1);
5453
5454 /* ... and S1 is Sword, and WORD_BOUNDARY_P (C1, C2)
5455 returns 0. */
5456 if ((s1 == Sword) && !WORD_BOUNDARY_P (c1, c2))
5457 goto fail;
5458 }
5459 }
5460 break;
5461
5462 case wordend:
5463 DEBUG_PRINT1 ("EXECUTING wordend.\n");
5464
5465 /* We FAIL in one of the following cases: */
5466
5467 /* Case 1: D is at the beginning of string. */
5468 if (AT_STRINGS_BEG (d))
5469 goto fail;
5470 else
5471 {
5472 /* C1 is the character before D, S1 is the syntax of C1, C2
5473 is the character at D, and S2 is the syntax of C2. */
5474 re_wchar_t c1, c2;
5475 int s1, s2;
5476 #ifdef emacs
5477 int offset = PTR_TO_OFFSET (d) - 1;
5478 int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
5479 UPDATE_SYNTAX_TABLE (charpos);
5480 #endif
5481 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
5482 s1 = SYNTAX (c1);
5483
5484 /* Case 2: S1 is not Sword. */
5485 if (s1 != Sword)
5486 goto fail;
5487
5488 /* Case 3: D is not at the end of string ... */
5489 if (!AT_STRINGS_END (d))
5490 {
5491 PREFETCH_NOLIMIT ();
5492 c2 = RE_STRING_CHAR (d, dend - d);
5493 #ifdef emacs
5494 UPDATE_SYNTAX_TABLE_FORWARD (charpos);
5495 #endif
5496 s2 = SYNTAX (c2);
5497
5498 /* ... and S2 is Sword, and WORD_BOUNDARY_P (C1, C2)
5499 returns 0. */
5500 if ((s2 == Sword) && !WORD_BOUNDARY_P (c1, c2))
5501 goto fail;
5502 }
5503 }
5504 break;
5505
5506 case syntaxspec:
5507 case notsyntaxspec:
5508 not = (re_opcode_t) *(p - 1) == notsyntaxspec;
5509 mcnt = *p++;
5510 DEBUG_PRINT3 ("EXECUTING %ssyntaxspec %d.\n", not?"not":"", mcnt);
5511 PREFETCH ();
5512 #ifdef emacs
5513 {
5514 int offset = PTR_TO_OFFSET (d);
5515 int pos1 = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
5516 UPDATE_SYNTAX_TABLE (pos1);
5517 }
5518 #endif
5519 {
5520 int len;
5521 re_wchar_t c;
5522
5523 c = RE_STRING_CHAR_AND_LENGTH (d, dend - d, len);
5524
5525 if ((SYNTAX (c) != (enum syntaxcode) mcnt) ^ not)
5526 goto fail;
5527 d += len;
5528 }
5529 break;
5530
5531 #ifdef emacs
5532 case before_dot:
5533 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
5534 if (PTR_BYTE_POS (d) >= PT_BYTE)
5535 goto fail;
5536 break;
5537
5538 case at_dot:
5539 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
5540 if (PTR_BYTE_POS (d) != PT_BYTE)
5541 goto fail;
5542 break;
5543
5544 case after_dot:
5545 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
5546 if (PTR_BYTE_POS (d) <= PT_BYTE)
5547 goto fail;
5548 break;
5549
5550 case categoryspec:
5551 case notcategoryspec:
5552 not = (re_opcode_t) *(p - 1) == notcategoryspec;
5553 mcnt = *p++;
5554 DEBUG_PRINT3 ("EXECUTING %scategoryspec %d.\n", not?"not":"", mcnt);
5555 PREFETCH ();
5556 {
5557 int len;
5558 re_wchar_t c;
5559
5560 c = RE_STRING_CHAR_AND_LENGTH (d, dend - d, len);
5561
5562 if ((!CHAR_HAS_CATEGORY (c, mcnt)) ^ not)
5563 goto fail;
5564 d += len;
5565 }
5566 break;
5567
5568 #endif /* emacs */
5569
5570 default:
5571 abort ();
5572 }
5573 continue; /* Successfully executed one pattern command; keep going. */
5574
5575
5576 /* We goto here if a matching operation fails. */
5577 fail:
5578 IMMEDIATE_QUIT_CHECK;
5579 if (!FAIL_STACK_EMPTY ())
5580 {
5581 re_char *str, *pat;
5582 /* A restart point is known. Restore to that state. */
5583 DEBUG_PRINT1 ("\nFAIL:\n");
5584 POP_FAILURE_POINT (str, pat);
5585 switch (SWITCH_ENUM_CAST ((re_opcode_t) *pat++))
5586 {
5587 case on_failure_keep_string_jump:
5588 assert (str == NULL);
5589 goto continue_failure_jump;
5590
5591 case on_failure_jump_nastyloop:
5592 assert ((re_opcode_t)pat[-2] == no_op);
5593 PUSH_FAILURE_POINT (pat - 2, str);
5594 /* Fallthrough */
5595
5596 case on_failure_jump_loop:
5597 case on_failure_jump:
5598 case succeed_n:
5599 d = str;
5600 continue_failure_jump:
5601 EXTRACT_NUMBER_AND_INCR (mcnt, pat);
5602 p = pat + mcnt;
5603 break;
5604
5605 case no_op:
5606 /* A special frame used for nastyloops. */
5607 goto fail;
5608
5609 default:
5610 abort();
5611 }
5612
5613 assert (p >= bufp->buffer && p <= pend);
5614
5615 if (d >= string1 && d <= end1)
5616 dend = end_match_1;
5617 }
5618 else
5619 break; /* Matching at this starting point really fails. */
5620 } /* for (;;) */
5621
5622 if (best_regs_set)
5623 goto restore_best_regs;
5624
5625 FREE_VARIABLES ();
5626
5627 return -1; /* Failure to match. */
5628 } /* re_match_2 */
5629 \f
5630 /* Subroutine definitions for re_match_2. */
5631
5632 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5633 bytes; nonzero otherwise. */
5634
5635 static int
5636 bcmp_translate (s1, s2, len, translate, multibyte)
5637 re_char *s1, *s2;
5638 register int len;
5639 RE_TRANSLATE_TYPE translate;
5640 const int multibyte;
5641 {
5642 register re_char *p1 = s1, *p2 = s2;
5643 re_char *p1_end = s1 + len;
5644 re_char *p2_end = s2 + len;
5645
5646 /* FIXME: Checking both p1 and p2 presumes that the two strings might have
5647 different lengths, but relying on a single `len' would break this. -sm */
5648 while (p1 < p1_end && p2 < p2_end)
5649 {
5650 int p1_charlen, p2_charlen;
5651 re_wchar_t p1_ch, p2_ch;
5652
5653 p1_ch = RE_STRING_CHAR_AND_LENGTH (p1, p1_end - p1, p1_charlen);
5654 p2_ch = RE_STRING_CHAR_AND_LENGTH (p2, p2_end - p2, p2_charlen);
5655
5656 if (RE_TRANSLATE (translate, p1_ch)
5657 != RE_TRANSLATE (translate, p2_ch))
5658 return 1;
5659
5660 p1 += p1_charlen, p2 += p2_charlen;
5661 }
5662
5663 if (p1 != p1_end || p2 != p2_end)
5664 return 1;
5665
5666 return 0;
5667 }
5668 \f
5669 /* Entry points for GNU code. */
5670
5671 /* re_compile_pattern is the GNU regular expression compiler: it
5672 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5673 Returns 0 if the pattern was valid, otherwise an error string.
5674
5675 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5676 are set in BUFP on entry.
5677
5678 We call regex_compile to do the actual compilation. */
5679
5680 const char *
5681 re_compile_pattern (pattern, length, bufp)
5682 const char *pattern;
5683 size_t length;
5684 struct re_pattern_buffer *bufp;
5685 {
5686 reg_errcode_t ret;
5687
5688 /* GNU code is written to assume at least RE_NREGS registers will be set
5689 (and at least one extra will be -1). */
5690 bufp->regs_allocated = REGS_UNALLOCATED;
5691
5692 /* And GNU code determines whether or not to get register information
5693 by passing null for the REGS argument to re_match, etc., not by
5694 setting no_sub. */
5695 bufp->no_sub = 0;
5696
5697 ret = regex_compile ((re_char*) pattern, length, re_syntax_options, bufp);
5698
5699 if (!ret)
5700 return NULL;
5701 return gettext (re_error_msgid[(int) ret]);
5702 }
5703 WEAK_ALIAS (__re_compile_pattern, re_compile_pattern)
5704 \f
5705 /* Entry points compatible with 4.2 BSD regex library. We don't define
5706 them unless specifically requested. */
5707
5708 #if defined _REGEX_RE_COMP || defined _LIBC
5709
5710 /* BSD has one and only one pattern buffer. */
5711 static struct re_pattern_buffer re_comp_buf;
5712
5713 char *
5714 # ifdef _LIBC
5715 /* Make these definitions weak in libc, so POSIX programs can redefine
5716 these names if they don't use our functions, and still use
5717 regcomp/regexec below without link errors. */
5718 weak_function
5719 # endif
5720 re_comp (s)
5721 const char *s;
5722 {
5723 reg_errcode_t ret;
5724
5725 if (!s)
5726 {
5727 if (!re_comp_buf.buffer)
5728 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5729 return (char *) gettext ("No previous regular expression");
5730 return 0;
5731 }
5732
5733 if (!re_comp_buf.buffer)
5734 {
5735 re_comp_buf.buffer = (unsigned char *) malloc (200);
5736 if (re_comp_buf.buffer == NULL)
5737 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5738 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
5739 re_comp_buf.allocated = 200;
5740
5741 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5742 if (re_comp_buf.fastmap == NULL)
5743 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5744 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
5745 }
5746
5747 /* Since `re_exec' always passes NULL for the `regs' argument, we
5748 don't need to initialize the pattern buffer fields which affect it. */
5749
5750 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5751
5752 if (!ret)
5753 return NULL;
5754
5755 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5756 return (char *) gettext (re_error_msgid[(int) ret]);
5757 }
5758
5759
5760 int
5761 # ifdef _LIBC
5762 weak_function
5763 # endif
5764 re_exec (s)
5765 const char *s;
5766 {
5767 const int len = strlen (s);
5768 return
5769 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5770 }
5771 #endif /* _REGEX_RE_COMP */
5772 \f
5773 /* POSIX.2 functions. Don't define these for Emacs. */
5774
5775 #ifndef emacs
5776
5777 /* regcomp takes a regular expression as a string and compiles it.
5778
5779 PREG is a regex_t *. We do not expect any fields to be initialized,
5780 since POSIX says we shouldn't. Thus, we set
5781
5782 `buffer' to the compiled pattern;
5783 `used' to the length of the compiled pattern;
5784 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5785 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5786 RE_SYNTAX_POSIX_BASIC;
5787 `fastmap' to an allocated space for the fastmap;
5788 `fastmap_accurate' to zero;
5789 `re_nsub' to the number of subexpressions in PATTERN.
5790
5791 PATTERN is the address of the pattern string.
5792
5793 CFLAGS is a series of bits which affect compilation.
5794
5795 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5796 use POSIX basic syntax.
5797
5798 If REG_NEWLINE is set, then . and [^...] don't match newline.
5799 Also, regexec will try a match beginning after every newline.
5800
5801 If REG_ICASE is set, then we considers upper- and lowercase
5802 versions of letters to be equivalent when matching.
5803
5804 If REG_NOSUB is set, then when PREG is passed to regexec, that
5805 routine will report only success or failure, and nothing about the
5806 registers.
5807
5808 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
5809 the return codes and their meanings.) */
5810
5811 int
5812 regcomp (preg, pattern, cflags)
5813 regex_t *preg;
5814 const char *pattern;
5815 int cflags;
5816 {
5817 reg_errcode_t ret;
5818 reg_syntax_t syntax
5819 = (cflags & REG_EXTENDED) ?
5820 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5821
5822 /* regex_compile will allocate the space for the compiled pattern. */
5823 preg->buffer = 0;
5824 preg->allocated = 0;
5825 preg->used = 0;
5826
5827 /* Try to allocate space for the fastmap. */
5828 preg->fastmap = (char *) malloc (1 << BYTEWIDTH);
5829
5830 if (cflags & REG_ICASE)
5831 {
5832 unsigned i;
5833
5834 preg->translate
5835 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
5836 * sizeof (*(RE_TRANSLATE_TYPE)0));
5837 if (preg->translate == NULL)
5838 return (int) REG_ESPACE;
5839
5840 /* Map uppercase characters to corresponding lowercase ones. */
5841 for (i = 0; i < CHAR_SET_SIZE; i++)
5842 preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i;
5843 }
5844 else
5845 preg->translate = NULL;
5846
5847 /* If REG_NEWLINE is set, newlines are treated differently. */
5848 if (cflags & REG_NEWLINE)
5849 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5850 syntax &= ~RE_DOT_NEWLINE;
5851 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5852 }
5853 else
5854 syntax |= RE_NO_NEWLINE_ANCHOR;
5855
5856 preg->no_sub = !!(cflags & REG_NOSUB);
5857
5858 /* POSIX says a null character in the pattern terminates it, so we
5859 can use strlen here in compiling the pattern. */
5860 ret = regex_compile ((re_char*) pattern, strlen (pattern), syntax, preg);
5861
5862 /* POSIX doesn't distinguish between an unmatched open-group and an
5863 unmatched close-group: both are REG_EPAREN. */
5864 if (ret == REG_ERPAREN)
5865 ret = REG_EPAREN;
5866
5867 if (ret == REG_NOERROR && preg->fastmap)
5868 { /* Compute the fastmap now, since regexec cannot modify the pattern
5869 buffer. */
5870 re_compile_fastmap (preg);
5871 if (preg->can_be_null)
5872 { /* The fastmap can't be used anyway. */
5873 free (preg->fastmap);
5874 preg->fastmap = NULL;
5875 }
5876 }
5877 return (int) ret;
5878 }
5879 WEAK_ALIAS (__regcomp, regcomp)
5880
5881
5882 /* regexec searches for a given pattern, specified by PREG, in the
5883 string STRING.
5884
5885 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5886 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5887 least NMATCH elements, and we set them to the offsets of the
5888 corresponding matched substrings.
5889
5890 EFLAGS specifies `execution flags' which affect matching: if
5891 REG_NOTBOL is set, then ^ does not match at the beginning of the
5892 string; if REG_NOTEOL is set, then $ does not match at the end.
5893
5894 We return 0 if we find a match and REG_NOMATCH if not. */
5895
5896 int
5897 regexec (preg, string, nmatch, pmatch, eflags)
5898 const regex_t *preg;
5899 const char *string;
5900 size_t nmatch;
5901 regmatch_t pmatch[];
5902 int eflags;
5903 {
5904 int ret;
5905 struct re_registers regs;
5906 regex_t private_preg;
5907 int len = strlen (string);
5908 boolean want_reg_info = !preg->no_sub && nmatch > 0 && pmatch;
5909
5910 private_preg = *preg;
5911
5912 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5913 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5914
5915 /* The user has told us exactly how many registers to return
5916 information about, via `nmatch'. We have to pass that on to the
5917 matching routines. */
5918 private_preg.regs_allocated = REGS_FIXED;
5919
5920 if (want_reg_info)
5921 {
5922 regs.num_regs = nmatch;
5923 regs.start = TALLOC (nmatch * 2, regoff_t);
5924 if (regs.start == NULL)
5925 return (int) REG_NOMATCH;
5926 regs.end = regs.start + nmatch;
5927 }
5928
5929 /* Instead of using not_eol to implement REG_NOTEOL, we could simply
5930 pass (&private_preg, string, len + 1, 0, len, ...) pretending the string
5931 was a little bit longer but still only matching the real part.
5932 This works because the `endline' will check for a '\n' and will find a
5933 '\0', correctly deciding that this is not the end of a line.
5934 But it doesn't work out so nicely for REG_NOTBOL, since we don't have
5935 a convenient '\0' there. For all we know, the string could be preceded
5936 by '\n' which would throw things off. */
5937
5938 /* Perform the searching operation. */
5939 ret = re_search (&private_preg, string, len,
5940 /* start: */ 0, /* range: */ len,
5941 want_reg_info ? &regs : (struct re_registers *) 0);
5942
5943 /* Copy the register information to the POSIX structure. */
5944 if (want_reg_info)
5945 {
5946 if (ret >= 0)
5947 {
5948 unsigned r;
5949
5950 for (r = 0; r < nmatch; r++)
5951 {
5952 pmatch[r].rm_so = regs.start[r];
5953 pmatch[r].rm_eo = regs.end[r];
5954 }
5955 }
5956
5957 /* If we needed the temporary register info, free the space now. */
5958 free (regs.start);
5959 }
5960
5961 /* We want zero return to mean success, unlike `re_search'. */
5962 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5963 }
5964 WEAK_ALIAS (__regexec, regexec)
5965
5966
5967 /* Returns a message corresponding to an error code, ERRCODE, returned
5968 from either regcomp or regexec. We don't use PREG here. */
5969
5970 size_t
5971 regerror (errcode, preg, errbuf, errbuf_size)
5972 int errcode;
5973 const regex_t *preg;
5974 char *errbuf;
5975 size_t errbuf_size;
5976 {
5977 const char *msg;
5978 size_t msg_size;
5979
5980 if (errcode < 0
5981 || errcode >= (sizeof (re_error_msgid) / sizeof (re_error_msgid[0])))
5982 /* Only error codes returned by the rest of the code should be passed
5983 to this routine. If we are given anything else, or if other regex
5984 code generates an invalid error code, then the program has a bug.
5985 Dump core so we can fix it. */
5986 abort ();
5987
5988 msg = gettext (re_error_msgid[errcode]);
5989
5990 msg_size = strlen (msg) + 1; /* Includes the null. */
5991
5992 if (errbuf_size != 0)
5993 {
5994 if (msg_size > errbuf_size)
5995 {
5996 strncpy (errbuf, msg, errbuf_size - 1);
5997 errbuf[errbuf_size - 1] = 0;
5998 }
5999 else
6000 strcpy (errbuf, msg);
6001 }
6002
6003 return msg_size;
6004 }
6005 WEAK_ALIAS (__regerror, regerror)
6006
6007
6008 /* Free dynamically allocated space used by PREG. */
6009
6010 void
6011 regfree (preg)
6012 regex_t *preg;
6013 {
6014 if (preg->buffer != NULL)
6015 free (preg->buffer);
6016 preg->buffer = NULL;
6017
6018 preg->allocated = 0;
6019 preg->used = 0;
6020
6021 if (preg->fastmap != NULL)
6022 free (preg->fastmap);
6023 preg->fastmap = NULL;
6024 preg->fastmap_accurate = 0;
6025
6026 if (preg->translate != NULL)
6027 free (preg->translate);
6028 preg->translate = NULL;
6029 }
6030 WEAK_ALIAS (__regfree, regfree)
6031
6032 #endif /* not emacs */
Unnamed repository; edit this file 'description' to name the repository.