aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorRoland McGrath1995-05-20 16:40:11 +0000
committerRoland McGrath1995-05-20 16:40:11 +0000
commitfa9a63c5c25233d6758d38a5f6a6c3202af41345 (patch)
treeaa4a1b70c9c84bd46c16c2c6fb7f99d0e72baf03
parenteb9780b030287af9625e516affdea18f77ee7a29 (diff)
downloademacs-fa9a63c5c25233d6758d38a5f6a6c3202af41345.tar.gz
emacs-fa9a63c5c25233d6758d38a5f6a6c3202af41345.zip
Add `#ifdef _LIBC' in a few places, so this can be compiled in libc.
Diffstat
-rw-r--r--src/regex.c5389
1 files changed, 5389 insertions, 0 deletions
diff --git a/src/regex.c b/src/regex.c
index 5d11a635514..5a6d3a157dc 100644
--- a/src/regex.c
+++ b/src/regex.c
@@ -5,3 +5,5392 @@
5 5
6 Copyright (C) 1993, 1994, 1995 Free Software Foundation, Inc. 6 Copyright (C) 1993, 1994, 1995 Free Software Foundation, Inc.
7 7
8 This program is free software; you can redistribute it and/or modify
9 it under the terms of the GNU General Public License as published by
10 the Free Software Foundation; either version 2, or (at your option)
11 any later version.
12
13 This program is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
17
18 You should have received a copy of the GNU General Public License
19 along with this program; if not, write to the Free Software
20 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
21
22/* AIX requires this to be the first thing in the file. */
23#if defined (_AIX) && !defined (REGEX_MALLOC)
24 #pragma alloca
25#endif
26
27#define _GNU_SOURCE
28
29#ifdef HAVE_CONFIG_H
30#include <config.h>
31#endif
32
33/* We need this for `regex.h', and perhaps for the Emacs include files. */
34#include <sys/types.h>
35
36/* This is for other GNU distributions with internationalized messages. */
37#if HAVE_LIBINTL_H || defined (_LIBC)
38# include <libintl.h>
39#else
40# define gettext(msgid) (msgid)
41#endif
42
43/* The `emacs' switch turns on certain matching commands
44 that make sense only in Emacs. */
45#ifdef emacs
46
47#include "lisp.h"
48#include "buffer.h"
49#include "syntax.h"
50
51#else /* not emacs */
52
53/* If we are not linking with Emacs proper,
54 we can't use the relocating allocator
55 even if config.h says that we can. */
56#undef REL_ALLOC
57
58#if defined (STDC_HEADERS) || defined (_LIBC)
59#include <stdlib.h>
60#else
61char *malloc ();
62char *realloc ();
63#endif
64
65/* We used to test for `BSTRING' here, but only GCC and Emacs define
66 `BSTRING', as far as I know, and neither of them use this code. */
67#ifndef INHIBIT_STRING_HEADER
68#if HAVE_STRING_H || STDC_HEADERS || defined (_LIBC)
69#include <string.h>
70#ifndef bcmp
71#define bcmp(s1, s2, n) memcmp ((s1), (s2), (n))
72#endif
73#ifndef bcopy
74#define bcopy(s, d, n) memcpy ((d), (s), (n))
75#endif
76#ifndef bzero
77#define bzero(s, n) memset ((s), 0, (n))
78#endif
79#else
80#include <strings.h>
81#endif
82#endif
83
84/* Define the syntax stuff for \<, \>, etc. */
85
86/* This must be nonzero for the wordchar and notwordchar pattern
87 commands in re_match_2. */
88#ifndef Sword
89#define Sword 1
90#endif
91
92#ifdef SWITCH_ENUM_BUG
93#define SWITCH_ENUM_CAST(x) ((int)(x))
94#else
95#define SWITCH_ENUM_CAST(x) (x)
96#endif
97
98#ifdef SYNTAX_TABLE
99
100extern char *re_syntax_table;
101
102#else /* not SYNTAX_TABLE */
103
104/* How many characters in the character set. */
105#define CHAR_SET_SIZE 256
106
107static char re_syntax_table[CHAR_SET_SIZE];
108
109static void
110init_syntax_once ()
111{
112 register int c;
113 static int done = 0;
114
115 if (done)
116 return;
117
118 bzero (re_syntax_table, sizeof re_syntax_table);
119
120 for (c = 'a'; c <= 'z'; c++)
121 re_syntax_table[c] = Sword;
122
123 for (c = 'A'; c <= 'Z'; c++)
124 re_syntax_table[c] = Sword;
125
126 for (c = '0'; c <= '9'; c++)
127 re_syntax_table[c] = Sword;
128
129 re_syntax_table['_'] = Sword;
130
131 done = 1;
132}
133
134#endif /* not SYNTAX_TABLE */
135
136#define SYNTAX(c) re_syntax_table[c]
137
138#endif /* not emacs */
139
140/* Get the interface, including the syntax bits. */
141#include "regex.h"
142
143/* isalpha etc. are used for the character classes. */
144#include <ctype.h>
145
146/* Jim Meyering writes:
147
148 "... Some ctype macros are valid only for character codes that
149 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
150 using /bin/cc or gcc but without giving an ansi option). So, all
151 ctype uses should be through macros like ISPRINT... If
152 STDC_HEADERS is defined, then autoconf has verified that the ctype
153 macros don't need to be guarded with references to isascii. ...
154 Defining isascii to 1 should let any compiler worth its salt
155 eliminate the && through constant folding." */
156
157#if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII))
158#define ISASCII(c) 1
159#else
160#define ISASCII(c) isascii(c)
161#endif
162
163#ifdef isblank
164#define ISBLANK(c) (ISASCII (c) && isblank (c))
165#else
166#define ISBLANK(c) ((c) == ' ' || (c) == '\t')
167#endif
168#ifdef isgraph
169#define ISGRAPH(c) (ISASCII (c) && isgraph (c))
170#else
171#define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
172#endif
173
174#define ISPRINT(c) (ISASCII (c) && isprint (c))
175#define ISDIGIT(c) (ISASCII (c) && isdigit (c))
176#define ISALNUM(c) (ISASCII (c) && isalnum (c))
177#define ISALPHA(c) (ISASCII (c) && isalpha (c))
178#define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
179#define ISLOWER(c) (ISASCII (c) && islower (c))
180#define ISPUNCT(c) (ISASCII (c) && ispunct (c))
181#define ISSPACE(c) (ISASCII (c) && isspace (c))
182#define ISUPPER(c) (ISASCII (c) && isupper (c))
183#define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
184
185#ifndef NULL
186#define NULL 0
187#endif
188
189/* We remove any previous definition of `SIGN_EXTEND_CHAR',
190 since ours (we hope) works properly with all combinations of
191 machines, compilers, `char' and `unsigned char' argument types.
192 (Per Bothner suggested the basic approach.) */
193#undef SIGN_EXTEND_CHAR
194#if __STDC__
195#define SIGN_EXTEND_CHAR(c) ((signed char) (c))
196#else /* not __STDC__ */
197/* As in Harbison and Steele. */
198#define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
199#endif
200
201/* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
202 use `alloca' instead of `malloc'. This is because using malloc in
203 re_search* or re_match* could cause memory leaks when C-g is used in
204 Emacs; also, malloc is slower and causes storage fragmentation. On
205 the other hand, malloc is more portable, and easier to debug.
206
207 Because we sometimes use alloca, some routines have to be macros,
208 not functions -- `alloca'-allocated space disappears at the end of the
209 function it is called in. */
210
211#ifdef REGEX_MALLOC
212
213#define REGEX_ALLOCATE malloc
214#define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
215#define REGEX_FREE free
216
217#else /* not REGEX_MALLOC */
218
219/* Emacs already defines alloca, sometimes. */
220#ifndef alloca
221
222/* Make alloca work the best possible way. */
223#ifdef __GNUC__
224#define alloca __builtin_alloca
225#else /* not __GNUC__ */
226#if HAVE_ALLOCA_H
227#include <alloca.h>
228#else /* not __GNUC__ or HAVE_ALLOCA_H */
229#ifndef _AIX /* Already did AIX, up at the top. */
230char *alloca ();
231#endif /* not _AIX */
232#endif /* not HAVE_ALLOCA_H */
233#endif /* not __GNUC__ */
234
235#endif /* not alloca */
236
237#define REGEX_ALLOCATE alloca
238
239/* Assumes a `char *destination' variable. */
240#define REGEX_REALLOCATE(source, osize, nsize) \
241 (destination = (char *) alloca (nsize), \
242 bcopy (source, destination, osize), \
243 destination)
244
245/* No need to do anything to free, after alloca. */
246#define REGEX_FREE(arg) (0)
247
248#endif /* not REGEX_MALLOC */
249
250/* Define how to allocate the failure stack. */
251
252#ifdef REL_ALLOC
253#define REGEX_ALLOCATE_STACK(size) \
254 r_alloc (&failure_stack_ptr, (size))
255#define REGEX_REALLOCATE_STACK(source, osize, nsize) \
256 r_re_alloc (&failure_stack_ptr, (nsize))
257#define REGEX_FREE_STACK(ptr) \
258 r_alloc_free (&failure_stack_ptr)
259
260#else /* not REL_ALLOC */
261
262#ifdef REGEX_MALLOC
263
264#define REGEX_ALLOCATE_STACK malloc
265#define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
266#define REGEX_FREE_STACK free
267
268#else /* not REGEX_MALLOC */
269
270#define REGEX_ALLOCATE_STACK alloca
271
272#define REGEX_REALLOCATE_STACK(source, osize, nsize) \
273 REGEX_REALLOCATE (source, osize, nsize)
274/* No need to explicitly free anything. */
275#define REGEX_FREE_STACK(arg)
276
277#endif /* not REGEX_MALLOC */
278#endif /* not REL_ALLOC */
279
280
281/* True if `size1' is non-NULL and PTR is pointing anywhere inside
282 `string1' or just past its end. This works if PTR is NULL, which is
283 a good thing. */
284#define FIRST_STRING_P(ptr) \
285 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
286
287/* (Re)Allocate N items of type T using malloc, or fail. */
288#define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
289#define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
290#define RETALLOC_IF(addr, n, t) \
291 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
292#define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
293
294#define BYTEWIDTH 8 /* In bits. */
295
296#define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
297
298#undef MAX
299#undef MIN
300#define MAX(a, b) ((a) > (b) ? (a) : (b))
301#define MIN(a, b) ((a) < (b) ? (a) : (b))
302
303typedef char boolean;
304#define false 0
305#define true 1
306
307static int re_match_2_internal ();
308
309/* These are the command codes that appear in compiled regular
310 expressions. Some opcodes are followed by argument bytes. A
311 command code can specify any interpretation whatsoever for its
312 arguments. Zero bytes may appear in the compiled regular expression. */
313
314typedef enum
315{
316 no_op = 0,
317
318 /* Succeed right away--no more backtracking. */
319 succeed,
320
321 /* Followed by one byte giving n, then by n literal bytes. */
322 exactn,
323
324 /* Matches any (more or less) character. */
325 anychar,
326
327 /* Matches any one char belonging to specified set. First
328 following byte is number of bitmap bytes. Then come bytes
329 for a bitmap saying which chars are in. Bits in each byte
330 are ordered low-bit-first. A character is in the set if its
331 bit is 1. A character too large to have a bit in the map is
332 automatically not in the set. */
333 charset,
334
335 /* Same parameters as charset, but match any character that is
336 not one of those specified. */
337 charset_not,
338
339 /* Start remembering the text that is matched, for storing in a
340 register. Followed by one byte with the register number, in
341 the range 0 to one less than the pattern buffer's re_nsub
342 field. Then followed by one byte with the number of groups
343 inner to this one. (This last has to be part of the
344 start_memory only because we need it in the on_failure_jump
345 of re_match_2.) */
346 start_memory,
347
348 /* Stop remembering the text that is matched and store it in a
349 memory register. Followed by one byte with the register
350 number, in the range 0 to one less than `re_nsub' in the
351 pattern buffer, and one byte with the number of inner groups,
352 just like `start_memory'. (We need the number of inner
353 groups here because we don't have any easy way of finding the
354 corresponding start_memory when we're at a stop_memory.) */
355 stop_memory,
356
357 /* Match a duplicate of something remembered. Followed by one
358 byte containing the register number. */
359 duplicate,
360
361 /* Fail unless at beginning of line. */
362 begline,
363
364 /* Fail unless at end of line. */
365 endline,
366
367 /* Succeeds if at beginning of buffer (if emacs) or at beginning
368 of string to be matched (if not). */
369 begbuf,
370
371 /* Analogously, for end of buffer/string. */
372 endbuf,
373
374 /* Followed by two byte relative address to which to jump. */
375 jump,
376
377 /* Same as jump, but marks the end of an alternative. */
378 jump_past_alt,
379
380 /* Followed by two-byte relative address of place to resume at
381 in case of failure. */
382 on_failure_jump,
383
384 /* Like on_failure_jump, but pushes a placeholder instead of the
385 current string position when executed. */
386 on_failure_keep_string_jump,
387
388 /* Throw away latest failure point and then jump to following
389 two-byte relative address. */
390 pop_failure_jump,
391
392 /* Change to pop_failure_jump if know won't have to backtrack to
393 match; otherwise change to jump. This is used to jump
394 back to the beginning of a repeat. If what follows this jump
395 clearly won't match what the repeat does, such that we can be
396 sure that there is no use backtracking out of repetitions
397 already matched, then we change it to a pop_failure_jump.
398 Followed by two-byte address. */
399 maybe_pop_jump,
400
401 /* Jump to following two-byte address, and push a dummy failure
402 point. This failure point will be thrown away if an attempt
403 is made to use it for a failure. A `+' construct makes this
404 before the first repeat. Also used as an intermediary kind
405 of jump when compiling an alternative. */
406 dummy_failure_jump,
407
408 /* Push a dummy failure point and continue. Used at the end of
409 alternatives. */
410 push_dummy_failure,
411
412 /* Followed by two-byte relative address and two-byte number n.
413 After matching N times, jump to the address upon failure. */
414 succeed_n,
415
416 /* Followed by two-byte relative address, and two-byte number n.
417 Jump to the address N times, then fail. */
418 jump_n,
419
420 /* Set the following two-byte relative address to the
421 subsequent two-byte number. The address *includes* the two
422 bytes of number. */
423 set_number_at,
424
425 wordchar, /* Matches any word-constituent character. */
426 notwordchar, /* Matches any char that is not a word-constituent. */
427
428 wordbeg, /* Succeeds if at word beginning. */
429 wordend, /* Succeeds if at word end. */
430
431 wordbound, /* Succeeds if at a word boundary. */
432 notwordbound /* Succeeds if not at a word boundary. */
433
434#ifdef emacs
435 ,before_dot, /* Succeeds if before point. */
436 at_dot, /* Succeeds if at point. */
437 after_dot, /* Succeeds if after point. */
438
439 /* Matches any character whose syntax is specified. Followed by
440 a byte which contains a syntax code, e.g., Sword. */
441 syntaxspec,
442
443 /* Matches any character whose syntax is not that specified. */
444 notsyntaxspec
445#endif /* emacs */
446} re_opcode_t;
447
448/* Common operations on the compiled pattern. */
449
450/* Store NUMBER in two contiguous bytes starting at DESTINATION. */
451
452#define STORE_NUMBER(destination, number) \
453 do { \
454 (destination)[0] = (number) & 0377; \
455 (destination)[1] = (number) >> 8; \
456 } while (0)
457
458/* Same as STORE_NUMBER, except increment DESTINATION to
459 the byte after where the number is stored. Therefore, DESTINATION
460 must be an lvalue. */
461
462#define STORE_NUMBER_AND_INCR(destination, number) \
463 do { \
464 STORE_NUMBER (destination, number); \
465 (destination) += 2; \
466 } while (0)
467
468/* Put into DESTINATION a number stored in two contiguous bytes starting
469 at SOURCE. */
470
471#define EXTRACT_NUMBER(destination, source) \
472 do { \
473 (destination) = *(source) & 0377; \
474 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
475 } while (0)
476
477#ifdef DEBUG
478static void
479extract_number (dest, source)
480 int *dest;
481 unsigned char *source;
482{
483 int temp = SIGN_EXTEND_CHAR (*(source + 1));
484 *dest = *source & 0377;
485 *dest += temp << 8;
486}
487
488#ifndef EXTRACT_MACROS /* To debug the macros. */
489#undef EXTRACT_NUMBER
490#define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
491#endif /* not EXTRACT_MACROS */
492
493#endif /* DEBUG */
494
495/* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
496 SOURCE must be an lvalue. */
497
498#define EXTRACT_NUMBER_AND_INCR(destination, source) \
499 do { \
500 EXTRACT_NUMBER (destination, source); \
501 (source) += 2; \
502 } while (0)
503
504#ifdef DEBUG
505static void
506extract_number_and_incr (destination, source)
507 int *destination;
508 unsigned char **source;
509{
510 extract_number (destination, *source);
511 *source += 2;
512}
513
514#ifndef EXTRACT_MACROS
515#undef EXTRACT_NUMBER_AND_INCR
516#define EXTRACT_NUMBER_AND_INCR(dest, src) \
517 extract_number_and_incr (&dest, &src)
518#endif /* not EXTRACT_MACROS */
519
520#endif /* DEBUG */
521
522/* If DEBUG is defined, Regex prints many voluminous messages about what
523 it is doing (if the variable `debug' is nonzero). If linked with the
524 main program in `iregex.c', you can enter patterns and strings
525 interactively. And if linked with the main program in `main.c' and
526 the other test files, you can run the already-written tests. */
527
528#ifdef DEBUG
529
530/* We use standard I/O for debugging. */
531#include <stdio.h>
532
533/* It is useful to test things that ``must'' be true when debugging. */
534#include <assert.h>
535
536static int debug = 0;
537
538#define DEBUG_STATEMENT(e) e
539#define DEBUG_PRINT1(x) if (debug) printf (x)
540#define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
541#define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
542#define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
543#define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
544 if (debug) print_partial_compiled_pattern (s, e)
545#define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
546 if (debug) print_double_string (w, s1, sz1, s2, sz2)
547
548
549/* Print the fastmap in human-readable form. */
550
551void
552print_fastmap (fastmap)
553 char *fastmap;
554{
555 unsigned was_a_range = 0;
556 unsigned i = 0;
557
558 while (i < (1 << BYTEWIDTH))
559 {
560 if (fastmap[i++])
561 {
562 was_a_range = 0;
563 putchar (i - 1);
564 while (i < (1 << BYTEWIDTH) && fastmap[i])
565 {
566 was_a_range = 1;
567 i++;
568 }
569 if (was_a_range)
570 {
571 printf ("-");
572 putchar (i - 1);
573 }
574 }
575 }
576 putchar ('\n');
577}
578
579
580/* Print a compiled pattern string in human-readable form, starting at
581 the START pointer into it and ending just before the pointer END. */
582
583void
584print_partial_compiled_pattern (start, end)
585 unsigned char *start;
586 unsigned char *end;
587{
588 int mcnt, mcnt2;
589 unsigned char *p = start;
590 unsigned char *pend = end;
591
592 if (start == NULL)
593 {
594 printf ("(null)\n");
595 return;
596 }
597
598 /* Loop over pattern commands. */
599 while (p < pend)
600 {
601 printf ("%d:\t", p - start);
602
603 switch ((re_opcode_t) *p++)
604 {
605 case no_op:
606 printf ("/no_op");
607 break;
608
609 case exactn:
610 mcnt = *p++;
611 printf ("/exactn/%d", mcnt);
612 do
613 {
614 putchar ('/');
615 putchar (*p++);
616 }
617 while (--mcnt);
618 break;
619
620 case start_memory:
621 mcnt = *p++;
622 printf ("/start_memory/%d/%d", mcnt, *p++);
623 break;
624
625 case stop_memory:
626 mcnt = *p++;
627 printf ("/stop_memory/%d/%d", mcnt, *p++);
628 break;
629
630 case duplicate:
631 printf ("/duplicate/%d", *p++);
632 break;
633
634 case anychar:
635 printf ("/anychar");
636 break;
637
638 case charset:
639 case charset_not:
640 {
641 register int c, last = -100;
642 register int in_range = 0;
643
644 printf ("/charset [%s",
645 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
646
647 assert (p + *p < pend);
648
649 for (c = 0; c < 256; c++)
650 if (c / 8 < *p
651 && (p[1 + (c/8)] & (1 << (c % 8))))
652 {
653 /* Are we starting a range? */
654 if (last + 1 == c && ! in_range)
655 {
656 putchar ('-');
657 in_range = 1;
658 }
659 /* Have we broken a range? */
660 else if (last + 1 != c && in_range)
661 {
662 putchar (last);
663 in_range = 0;
664 }
665
666 if (! in_range)
667 putchar (c);
668
669 last = c;
670 }
671
672 if (in_range)
673 putchar (last);
674
675 putchar (']');
676
677 p += 1 + *p;
678 }
679 break;
680
681 case begline:
682 printf ("/begline");
683 break;
684
685 case endline:
686 printf ("/endline");
687 break;
688
689 case on_failure_jump:
690 extract_number_and_incr (&mcnt, &p);
691 printf ("/on_failure_jump to %d", p + mcnt - start);
692 break;
693
694 case on_failure_keep_string_jump:
695 extract_number_and_incr (&mcnt, &p);
696 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
697 break;
698
699 case dummy_failure_jump:
700 extract_number_and_incr (&mcnt, &p);
701 printf ("/dummy_failure_jump to %d", p + mcnt - start);
702 break;
703
704 case push_dummy_failure:
705 printf ("/push_dummy_failure");
706 break;
707
708 case maybe_pop_jump:
709 extract_number_and_incr (&mcnt, &p);
710 printf ("/maybe_pop_jump to %d", p + mcnt - start);
711 break;
712
713 case pop_failure_jump:
714 extract_number_and_incr (&mcnt, &p);
715 printf ("/pop_failure_jump to %d", p + mcnt - start);
716 break;
717
718 case jump_past_alt:
719 extract_number_and_incr (&mcnt, &p);
720 printf ("/jump_past_alt to %d", p + mcnt - start);
721 break;
722
723 case jump:
724 extract_number_and_incr (&mcnt, &p);
725 printf ("/jump to %d", p + mcnt - start);
726 break;
727
728 case succeed_n:
729 extract_number_and_incr (&mcnt, &p);
730 extract_number_and_incr (&mcnt2, &p);
731 printf ("/succeed_n to %d, %d times", p + mcnt - start, mcnt2);
732 break;
733
734 case jump_n:
735 extract_number_and_incr (&mcnt, &p);
736 extract_number_and_incr (&mcnt2, &p);
737 printf ("/jump_n to %d, %d times", p + mcnt - start, mcnt2);
738 break;
739
740 case set_number_at:
741 extract_number_and_incr (&mcnt, &p);
742 extract_number_and_incr (&mcnt2, &p);
743 printf ("/set_number_at location %d to %d", p + mcnt - start, mcnt2);
744 break;
745
746 case wordbound:
747 printf ("/wordbound");
748 break;
749
750 case notwordbound:
751 printf ("/notwordbound");
752 break;
753
754 case wordbeg:
755 printf ("/wordbeg");
756 break;
757
758 case wordend:
759 printf ("/wordend");
760
761#ifdef emacs
762 case before_dot:
763 printf ("/before_dot");
764 break;
765
766 case at_dot:
767 printf ("/at_dot");
768 break;
769
770 case after_dot:
771 printf ("/after_dot");
772 break;
773
774 case syntaxspec:
775 printf ("/syntaxspec");
776 mcnt = *p++;
777 printf ("/%d", mcnt);
778 break;
779
780 case notsyntaxspec:
781 printf ("/notsyntaxspec");
782 mcnt = *p++;
783 printf ("/%d", mcnt);
784 break;
785#endif /* emacs */
786
787 case wordchar:
788 printf ("/wordchar");
789 break;
790
791 case notwordchar:
792 printf ("/notwordchar");
793 break;
794
795 case begbuf:
796 printf ("/begbuf");
797 break;
798
799 case endbuf:
800 printf ("/endbuf");
801 break;
802
803 default:
804 printf ("?%d", *(p-1));
805 }
806
807 putchar ('\n');
808 }
809
810 printf ("%d:\tend of pattern.\n", p - start);
811}
812
813
814void
815print_compiled_pattern (bufp)
816 struct re_pattern_buffer *bufp;
817{
818 unsigned char *buffer = bufp->buffer;
819
820 print_partial_compiled_pattern (buffer, buffer + bufp->used);
821 printf ("%d bytes used/%d bytes allocated.\n", bufp->used, bufp->allocated);
822
823 if (bufp->fastmap_accurate && bufp->fastmap)
824 {
825 printf ("fastmap: ");
826 print_fastmap (bufp->fastmap);
827 }
828
829 printf ("re_nsub: %d\t", bufp->re_nsub);
830 printf ("regs_alloc: %d\t", bufp->regs_allocated);
831 printf ("can_be_null: %d\t", bufp->can_be_null);
832 printf ("newline_anchor: %d\n", bufp->newline_anchor);
833 printf ("no_sub: %d\t", bufp->no_sub);
834 printf ("not_bol: %d\t", bufp->not_bol);
835 printf ("not_eol: %d\t", bufp->not_eol);
836 printf ("syntax: %d\n", bufp->syntax);
837 /* Perhaps we should print the translate table? */
838}
839
840
841void
842print_double_string (where, string1, size1, string2, size2)
843 const char *where;
844 const char *string1;
845 const char *string2;
846 int size1;
847 int size2;
848{
849 unsigned this_char;
850
851 if (where == NULL)
852 printf ("(null)");
853 else
854 {
855 if (FIRST_STRING_P (where))
856 {
857 for (this_char = where - string1; this_char < size1; this_char++)
858 putchar (string1[this_char]);
859
860 where = string2;
861 }
862
863 for (this_char = where - string2; this_char < size2; this_char++)
864 putchar (string2[this_char]);
865 }
866}
867
868#else /* not DEBUG */
869
870#undef assert
871#define assert(e)
872
873#define DEBUG_STATEMENT(e)
874#define DEBUG_PRINT1(x)
875#define DEBUG_PRINT2(x1, x2)
876#define DEBUG_PRINT3(x1, x2, x3)
877#define DEBUG_PRINT4(x1, x2, x3, x4)
878#define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
879#define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
880
881#endif /* not DEBUG */
882
883/* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
884 also be assigned to arbitrarily: each pattern buffer stores its own
885 syntax, so it can be changed between regex compilations. */
886/* This has no initializer because initialized variables in Emacs
887 become read-only after dumping. */
888reg_syntax_t re_syntax_options;
889
890
891/* Specify the precise syntax of regexps for compilation. This provides
892 for compatibility for various utilities which historically have
893 different, incompatible syntaxes.
894
895 The argument SYNTAX is a bit mask comprised of the various bits
896 defined in regex.h. We return the old syntax. */
897
898reg_syntax_t
899re_set_syntax (syntax)
900 reg_syntax_t syntax;
901{
902 reg_syntax_t ret = re_syntax_options;
903
904 re_syntax_options = syntax;
905 return ret;
906}
907
908/* This table gives an error message for each of the error codes listed
909 in regex.h. Obviously the order here has to be same as there.
910 POSIX doesn't require that we do anything for REG_NOERROR,
911 but why not be nice? */
912
913static const char *re_error_msgid[] =
914 { "Success", /* REG_NOERROR */
915 "No match", /* REG_NOMATCH */
916 "Invalid regular expression", /* REG_BADPAT */
917 "Invalid collation character", /* REG_ECOLLATE */
918 "Invalid character class name", /* REG_ECTYPE */
919 "Trailing backslash", /* REG_EESCAPE */
920 "Invalid back reference", /* REG_ESUBREG */
921 "Unmatched [ or [^", /* REG_EBRACK */
922 "Unmatched ( or \\(", /* REG_EPAREN */
923 "Unmatched \\{", /* REG_EBRACE */
924 "Invalid content of \\{\\}", /* REG_BADBR */
925 "Invalid range end", /* REG_ERANGE */
926 "Memory exhausted", /* REG_ESPACE */
927 "Invalid preceding regular expression", /* REG_BADRPT */
928 "Premature end of regular expression", /* REG_EEND */
929 "Regular expression too big", /* REG_ESIZE */
930 "Unmatched ) or \\)", /* REG_ERPAREN */
931 };
932
933/* Avoiding alloca during matching, to placate r_alloc. */
934
935/* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
936 searching and matching functions should not call alloca. On some
937 systems, alloca is implemented in terms of malloc, and if we're
938 using the relocating allocator routines, then malloc could cause a
939 relocation, which might (if the strings being searched are in the
940 ralloc heap) shift the data out from underneath the regexp
941 routines.
942
943 Here's another reason to avoid allocation: Emacs
944 processes input from X in a signal handler; processing X input may
945 call malloc; if input arrives while a matching routine is calling
946 malloc, then we're scrod. But Emacs can't just block input while
947 calling matching routines; then we don't notice interrupts when
948 they come in. So, Emacs blocks input around all regexp calls
949 except the matching calls, which it leaves unprotected, in the
950 faith that they will not malloc. */
951
952/* Normally, this is fine. */
953#define MATCH_MAY_ALLOCATE
954
955/* When using GNU C, we are not REALLY using the C alloca, no matter
956 what config.h may say. So don't take precautions for it. */
957#ifdef __GNUC__
958#undef C_ALLOCA
959#endif
960
961/* The match routines may not allocate if (1) they would do it with malloc
962 and (2) it's not safe for them to use malloc.
963 Note that if REL_ALLOC is defined, matching would not use malloc for the
964 failure stack, but we would still use it for the register vectors;
965 so REL_ALLOC should not affect this. */
966#if (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (emacs)
967#undef MATCH_MAY_ALLOCATE
968#endif
969
970
971/* Failure stack declarations and macros; both re_compile_fastmap and
972 re_match_2 use a failure stack. These have to be macros because of
973 REGEX_ALLOCATE_STACK. */
974
975
976/* Number of failure points for which to initially allocate space
977 when matching. If this number is exceeded, we allocate more
978 space, so it is not a hard limit. */
979#ifndef INIT_FAILURE_ALLOC
980#define INIT_FAILURE_ALLOC 5
981#endif
982
983/* Roughly the maximum number of failure points on the stack. Would be
984 exactly that if always used MAX_FAILURE_SPACE each time we failed.
985 This is a variable only so users of regex can assign to it; we never
986 change it ourselves. */
987#if defined (MATCH_MAY_ALLOCATE)
988int re_max_failures = 200000;
989#else
990int re_max_failures = 2000;
991#endif
992
993union fail_stack_elt
994{
995 unsigned char *pointer;
996 int integer;
997};
998
999typedef union fail_stack_elt fail_stack_elt_t;
1000
1001typedef struct
1002{
1003 fail_stack_elt_t *stack;
1004 unsigned size;
1005 unsigned avail; /* Offset of next open position. */
1006} fail_stack_type;
1007
1008#define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1009#define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1010#define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1011
1012
1013/* Define macros to initialize and free the failure stack.
1014 Do `return -2' if the alloc fails. */
1015
1016#ifdef MATCH_MAY_ALLOCATE
1017#define INIT_FAIL_STACK() \
1018 do { \
1019 fail_stack.stack = (fail_stack_elt_t *) \
1020 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1021 \
1022 if (fail_stack.stack == NULL) \
1023 return -2; \
1024 \
1025 fail_stack.size = INIT_FAILURE_ALLOC; \
1026 fail_stack.avail = 0; \
1027 } while (0)
1028
1029#define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1030#else
1031#define INIT_FAIL_STACK() \
1032 do { \
1033 fail_stack.avail = 0; \
1034 } while (0)
1035
1036#define RESET_FAIL_STACK()
1037#endif
1038
1039
1040/* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1041
1042 Return 1 if succeeds, and 0 if either ran out of memory
1043 allocating space for it or it was already too large.
1044
1045 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1046
1047#define DOUBLE_FAIL_STACK(fail_stack) \
1048 ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \
1049 ? 0 \
1050 : ((fail_stack).stack = (fail_stack_elt_t *) \
1051 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1052 (fail_stack).size * sizeof (fail_stack_elt_t), \
1053 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1054 \
1055 (fail_stack).stack == NULL \
1056 ? 0 \
1057 : ((fail_stack).size <<= 1, \
1058 1)))
1059
1060
1061/* Push pointer POINTER on FAIL_STACK.
1062 Return 1 if was able to do so and 0 if ran out of memory allocating
1063 space to do so. */
1064#define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1065 ((FAIL_STACK_FULL () \
1066 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1067 ? 0 \
1068 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1069 1))
1070
1071/* Push a pointer value onto the failure stack.
1072 Assumes the variable `fail_stack'. Probably should only
1073 be called from within `PUSH_FAILURE_POINT'. */
1074#define PUSH_FAILURE_POINTER(item) \
1075 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1076
1077/* This pushes an integer-valued item onto the failure stack.
1078 Assumes the variable `fail_stack'. Probably should only
1079 be called from within `PUSH_FAILURE_POINT'. */
1080#define PUSH_FAILURE_INT(item) \
1081 fail_stack.stack[fail_stack.avail++].integer = (item)
1082
1083/* Push a fail_stack_elt_t value onto the failure stack.
1084 Assumes the variable `fail_stack'. Probably should only
1085 be called from within `PUSH_FAILURE_POINT'. */
1086#define PUSH_FAILURE_ELT(item) \
1087 fail_stack.stack[fail_stack.avail++] = (item)
1088
1089/* These three POP... operations complement the three PUSH... operations.
1090 All assume that `fail_stack' is nonempty. */
1091#define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1092#define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1093#define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1094
1095/* Used to omit pushing failure point id's when we're not debugging. */
1096#ifdef DEBUG
1097#define DEBUG_PUSH PUSH_FAILURE_INT
1098#define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1099#else
1100#define DEBUG_PUSH(item)
1101#define DEBUG_POP(item_addr)
1102#endif
1103
1104
1105/* Push the information about the state we will need
1106 if we ever fail back to it.
1107
1108 Requires variables fail_stack, regstart, regend, reg_info, and
1109 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be
1110 declared.
1111
1112 Does `return FAILURE_CODE' if runs out of memory. */
1113
1114#define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1115 do { \
1116 char *destination; \
1117 /* Must be int, so when we don't save any registers, the arithmetic \
1118 of 0 + -1 isn't done as unsigned. */ \
1119 int this_reg; \
1120 \
1121 DEBUG_STATEMENT (failure_id++); \
1122 DEBUG_STATEMENT (nfailure_points_pushed++); \
1123 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1124 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1125 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1126 \
1127 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \
1128 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1129 \
1130 /* Ensure we have enough space allocated for what we will push. */ \
1131 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1132 { \
1133 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1134 return failure_code; \
1135 \
1136 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1137 (fail_stack).size); \
1138 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1139 } \
1140 \
1141 /* Push the info, starting with the registers. */ \
1142 DEBUG_PRINT1 ("\n"); \
1143 \
1144 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1145 this_reg++) \
1146 { \
1147 DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \
1148 DEBUG_STATEMENT (num_regs_pushed++); \
1149 \
1150 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1151 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1152 \
1153 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1154 PUSH_FAILURE_POINTER (regend[this_reg]); \
1155 \
1156 DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \
1157 DEBUG_PRINT2 (" match_null=%d", \
1158 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1159 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1160 DEBUG_PRINT2 (" matched_something=%d", \
1161 MATCHED_SOMETHING (reg_info[this_reg])); \
1162 DEBUG_PRINT2 (" ever_matched=%d", \
1163 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1164 DEBUG_PRINT1 ("\n"); \
1165 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1166 } \
1167 \
1168 DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\
1169 PUSH_FAILURE_INT (lowest_active_reg); \
1170 \
1171 DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\
1172 PUSH_FAILURE_INT (highest_active_reg); \
1173 \
1174 DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \
1175 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1176 PUSH_FAILURE_POINTER (pattern_place); \
1177 \
1178 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \
1179 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1180 size2); \
1181 DEBUG_PRINT1 ("'\n"); \
1182 PUSH_FAILURE_POINTER (string_place); \
1183 \
1184 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1185 DEBUG_PUSH (failure_id); \
1186 } while (0)
1187
1188/* This is the number of items that are pushed and popped on the stack
1189 for each register. */
1190#define NUM_REG_ITEMS 3
1191
1192/* Individual items aside from the registers. */
1193#ifdef DEBUG
1194#define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1195#else
1196#define NUM_NONREG_ITEMS 4
1197#endif
1198
1199/* We push at most this many items on the stack. */
1200#define MAX_FAILURE_ITEMS ((num_regs - 1) * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1201
1202/* We actually push this many items. */
1203#define NUM_FAILURE_ITEMS \
1204 ((highest_active_reg - lowest_active_reg + 1) * NUM_REG_ITEMS \
1205 + NUM_NONREG_ITEMS)
1206
1207/* How many items can still be added to the stack without overflowing it. */
1208#define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1209
1210
1211/* Pops what PUSH_FAIL_STACK pushes.
1212
1213 We restore into the parameters, all of which should be lvalues:
1214 STR -- the saved data position.
1215 PAT -- the saved pattern position.
1216 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1217 REGSTART, REGEND -- arrays of string positions.
1218 REG_INFO -- array of information about each subexpression.
1219
1220 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1221 `pend', `string1', `size1', `string2', and `size2'. */
1222
1223#define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1224{ \
1225 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \
1226 int this_reg; \
1227 const unsigned char *string_temp; \
1228 \
1229 assert (!FAIL_STACK_EMPTY ()); \
1230 \
1231 /* Remove failure points and point to how many regs pushed. */ \
1232 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1233 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1234 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1235 \
1236 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1237 \
1238 DEBUG_POP (&failure_id); \
1239 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1240 \
1241 /* If the saved string location is NULL, it came from an \
1242 on_failure_keep_string_jump opcode, and we want to throw away the \
1243 saved NULL, thus retaining our current position in the string. */ \
1244 string_temp = POP_FAILURE_POINTER (); \
1245 if (string_temp != NULL) \
1246 str = (const char *) string_temp; \
1247 \
1248 DEBUG_PRINT2 (" Popping string 0x%x: `", str); \
1249 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1250 DEBUG_PRINT1 ("'\n"); \
1251 \
1252 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1253 DEBUG_PRINT2 (" Popping pattern 0x%x: ", pat); \
1254 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1255 \
1256 /* Restore register info. */ \
1257 high_reg = (unsigned) POP_FAILURE_INT (); \
1258 DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \
1259 \
1260 low_reg = (unsigned) POP_FAILURE_INT (); \
1261 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \
1262 \
1263 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1264 { \
1265 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \
1266 \
1267 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1268 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \
1269 \
1270 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1271 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1272 \
1273 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1274 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1275 } \
1276 \
1277 set_regs_matched_done = 0; \
1278 DEBUG_STATEMENT (nfailure_points_popped++); \
1279} /* POP_FAILURE_POINT */
1280
1281
1282
1283/* Structure for per-register (a.k.a. per-group) information.
1284 Other register information, such as the
1285 starting and ending positions (which are addresses), and the list of
1286 inner groups (which is a bits list) are maintained in separate
1287 variables.
1288
1289 We are making a (strictly speaking) nonportable assumption here: that
1290 the compiler will pack our bit fields into something that fits into
1291 the type of `word', i.e., is something that fits into one item on the
1292 failure stack. */
1293
1294typedef union
1295{
1296 fail_stack_elt_t word;
1297 struct
1298 {
1299 /* This field is one if this group can match the empty string,
1300 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1301#define MATCH_NULL_UNSET_VALUE 3
1302 unsigned match_null_string_p : 2;
1303 unsigned is_active : 1;
1304 unsigned matched_something : 1;
1305 unsigned ever_matched_something : 1;
1306 } bits;
1307} register_info_type;
1308
1309#define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1310#define IS_ACTIVE(R) ((R).bits.is_active)
1311#define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1312#define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1313
1314
1315/* Call this when have matched a real character; it sets `matched' flags
1316 for the subexpressions which we are currently inside. Also records
1317 that those subexprs have matched. */
1318#define SET_REGS_MATCHED() \
1319 do \
1320 { \
1321 if (!set_regs_matched_done) \
1322 { \
1323 unsigned r; \
1324 set_regs_matched_done = 1; \
1325 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1326 { \
1327 MATCHED_SOMETHING (reg_info[r]) \
1328 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1329 = 1; \
1330 } \
1331 } \
1332 } \
1333 while (0)
1334
1335/* Registers are set to a sentinel when they haven't yet matched. */
1336static char reg_unset_dummy;
1337#define REG_UNSET_VALUE (&reg_unset_dummy)
1338#define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1339
1340/* Subroutine declarations and macros for regex_compile. */
1341
1342static void store_op1 (), store_op2 ();
1343static void insert_op1 (), insert_op2 ();
1344static boolean at_begline_loc_p (), at_endline_loc_p ();
1345static boolean group_in_compile_stack ();
1346static reg_errcode_t compile_range ();
1347
1348/* Fetch the next character in the uncompiled pattern---translating it
1349 if necessary. Also cast from a signed character in the constant
1350 string passed to us by the user to an unsigned char that we can use
1351 as an array index (in, e.g., `translate'). */
1352#define PATFETCH(c) \
1353 do {if (p == pend) return REG_EEND; \
1354 c = (unsigned char) *p++; \
1355 if (translate) c = translate[c]; \
1356 } while (0)
1357
1358/* Fetch the next character in the uncompiled pattern, with no
1359 translation. */
1360#define PATFETCH_RAW(c) \
1361 do {if (p == pend) return REG_EEND; \
1362 c = (unsigned char) *p++; \
1363 } while (0)
1364
1365/* Go backwards one character in the pattern. */
1366#define PATUNFETCH p--
1367
1368
1369/* If `translate' is non-null, return translate[D], else just D. We
1370 cast the subscript to translate because some data is declared as
1371 `char *', to avoid warnings when a string constant is passed. But
1372 when we use a character as a subscript we must make it unsigned. */
1373#define TRANSLATE(d) (translate ? translate[(unsigned char) (d)] : (d))
1374
1375
1376/* Macros for outputting the compiled pattern into `buffer'. */
1377
1378/* If the buffer isn't allocated when it comes in, use this. */
1379#define INIT_BUF_SIZE 32
1380
1381/* Make sure we have at least N more bytes of space in buffer. */
1382#define GET_BUFFER_SPACE(n) \
1383 while (b - bufp->buffer + (n) > bufp->allocated) \
1384 EXTEND_BUFFER ()
1385
1386/* Make sure we have one more byte of buffer space and then add C to it. */
1387#define BUF_PUSH(c) \
1388 do { \
1389 GET_BUFFER_SPACE (1); \
1390 *b++ = (unsigned char) (c); \
1391 } while (0)
1392
1393
1394/* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1395#define BUF_PUSH_2(c1, c2) \
1396 do { \
1397 GET_BUFFER_SPACE (2); \
1398 *b++ = (unsigned char) (c1); \
1399 *b++ = (unsigned char) (c2); \
1400 } while (0)
1401
1402
1403/* As with BUF_PUSH_2, except for three bytes. */
1404#define BUF_PUSH_3(c1, c2, c3) \
1405 do { \
1406 GET_BUFFER_SPACE (3); \
1407 *b++ = (unsigned char) (c1); \
1408 *b++ = (unsigned char) (c2); \
1409 *b++ = (unsigned char) (c3); \
1410 } while (0)
1411
1412
1413/* Store a jump with opcode OP at LOC to location TO. We store a
1414 relative address offset by the three bytes the jump itself occupies. */
1415#define STORE_JUMP(op, loc, to) \
1416 store_op1 (op, loc, (to) - (loc) - 3)
1417
1418/* Likewise, for a two-argument jump. */
1419#define STORE_JUMP2(op, loc, to, arg) \
1420 store_op2 (op, loc, (to) - (loc) - 3, arg)
1421
1422/* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1423#define INSERT_JUMP(op, loc, to) \
1424 insert_op1 (op, loc, (to) - (loc) - 3, b)
1425
1426/* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1427#define INSERT_JUMP2(op, loc, to, arg) \
1428 insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
1429
1430
1431/* This is not an arbitrary limit: the arguments which represent offsets
1432 into the pattern are two bytes long. So if 2^16 bytes turns out to
1433 be too small, many things would have to change. */
1434#define MAX_BUF_SIZE (1L << 16)
1435
1436
1437/* Extend the buffer by twice its current size via realloc and
1438 reset the pointers that pointed into the old block to point to the
1439 correct places in the new one. If extending the buffer results in it
1440 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1441#define EXTEND_BUFFER() \
1442 do { \
1443 unsigned char *old_buffer = bufp->buffer; \
1444 if (bufp->allocated == MAX_BUF_SIZE) \
1445 return REG_ESIZE; \
1446 bufp->allocated <<= 1; \
1447 if (bufp->allocated > MAX_BUF_SIZE) \
1448 bufp->allocated = MAX_BUF_SIZE; \
1449 bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\
1450 if (bufp->buffer == NULL) \
1451 return REG_ESPACE; \
1452 /* If the buffer moved, move all the pointers into it. */ \
1453 if (old_buffer != bufp->buffer) \
1454 { \
1455 b = (b - old_buffer) + bufp->buffer; \
1456 begalt = (begalt - old_buffer) + bufp->buffer; \
1457 if (fixup_alt_jump) \
1458 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1459 if (laststart) \
1460 laststart = (laststart - old_buffer) + bufp->buffer; \
1461 if (pending_exact) \
1462 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1463 } \
1464 } while (0)
1465
1466
1467/* Since we have one byte reserved for the register number argument to
1468 {start,stop}_memory, the maximum number of groups we can report
1469 things about is what fits in that byte. */
1470#define MAX_REGNUM 255
1471
1472/* But patterns can have more than `MAX_REGNUM' registers. We just
1473 ignore the excess. */
1474typedef unsigned regnum_t;
1475
1476
1477/* Macros for the compile stack. */
1478
1479/* Since offsets can go either forwards or backwards, this type needs to
1480 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1481typedef int pattern_offset_t;
1482
1483typedef struct
1484{
1485 pattern_offset_t begalt_offset;
1486 pattern_offset_t fixup_alt_jump;
1487 pattern_offset_t inner_group_offset;
1488 pattern_offset_t laststart_offset;
1489 regnum_t regnum;
1490} compile_stack_elt_t;
1491
1492
1493typedef struct
1494{
1495 compile_stack_elt_t *stack;
1496 unsigned size;
1497 unsigned avail; /* Offset of next open position. */
1498} compile_stack_type;
1499
1500
1501#define INIT_COMPILE_STACK_SIZE 32
1502
1503#define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1504#define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1505
1506/* The next available element. */
1507#define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1508
1509
1510/* Set the bit for character C in a list. */
1511#define SET_LIST_BIT(c) \
1512 (b[((unsigned char) (c)) / BYTEWIDTH] \
1513 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1514
1515
1516/* Get the next unsigned number in the uncompiled pattern. */
1517#define GET_UNSIGNED_NUMBER(num) \
1518 { if (p != pend) \
1519 { \
1520 PATFETCH (c); \
1521 while (ISDIGIT (c)) \
1522 { \
1523 if (num < 0) \
1524 num = 0; \
1525 num = num * 10 + c - '0'; \
1526 if (p == pend) \
1527 break; \
1528 PATFETCH (c); \
1529 } \
1530 } \
1531 }
1532
1533#define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1534
1535#define IS_CHAR_CLASS(string) \
1536 (STREQ (string, "alpha") || STREQ (string, "upper") \
1537 || STREQ (string, "lower") || STREQ (string, "digit") \
1538 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1539 || STREQ (string, "space") || STREQ (string, "print") \
1540 || STREQ (string, "punct") || STREQ (string, "graph") \
1541 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1542
1543#ifndef MATCH_MAY_ALLOCATE
1544
1545/* If we cannot allocate large objects within re_match_2_internal,
1546 we make the fail stack and register vectors global.
1547 The fail stack, we grow to the maximum size when a regexp
1548 is compiled.
1549 The register vectors, we adjust in size each time we
1550 compile a regexp, according to the number of registers it needs. */
1551
1552static fail_stack_type fail_stack;
1553
1554/* Size with which the following vectors are currently allocated.
1555 That is so we can make them bigger as needed,
1556 but never make them smaller. */
1557static int regs_allocated_size;
1558
1559static const char ** regstart, ** regend;
1560static const char ** old_regstart, ** old_regend;
1561static const char **best_regstart, **best_regend;
1562static register_info_type *reg_info;
1563static const char **reg_dummy;
1564static register_info_type *reg_info_dummy;
1565
1566/* Make the register vectors big enough for NUM_REGS registers,
1567 but don't make them smaller. */
1568
1569static
1570regex_grow_registers (num_regs)
1571 int num_regs;
1572{
1573 if (num_regs > regs_allocated_size)
1574 {
1575 RETALLOC_IF (regstart, num_regs, const char *);
1576 RETALLOC_IF (regend, num_regs, const char *);
1577 RETALLOC_IF (old_regstart, num_regs, const char *);
1578 RETALLOC_IF (old_regend, num_regs, const char *);
1579 RETALLOC_IF (best_regstart, num_regs, const char *);
1580 RETALLOC_IF (best_regend, num_regs, const char *);
1581 RETALLOC_IF (reg_info, num_regs, register_info_type);
1582 RETALLOC_IF (reg_dummy, num_regs, const char *);
1583 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1584
1585 regs_allocated_size = num_regs;
1586 }
1587}
1588
1589#endif /* not MATCH_MAY_ALLOCATE */
1590
1591/* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1592 Returns one of error codes defined in `regex.h', or zero for success.
1593
1594 Assumes the `allocated' (and perhaps `buffer') and `translate'
1595 fields are set in BUFP on entry.
1596
1597 If it succeeds, results are put in BUFP (if it returns an error, the
1598 contents of BUFP are undefined):
1599 `buffer' is the compiled pattern;
1600 `syntax' is set to SYNTAX;
1601 `used' is set to the length of the compiled pattern;
1602 `fastmap_accurate' is zero;
1603 `re_nsub' is the number of subexpressions in PATTERN;
1604 `not_bol' and `not_eol' are zero;
1605
1606 The `fastmap' and `newline_anchor' fields are neither
1607 examined nor set. */
1608
1609/* Return, freeing storage we allocated. */
1610#define FREE_STACK_RETURN(value) \
1611 return (free (compile_stack.stack), value)
1612
1613static reg_errcode_t
1614regex_compile (pattern, size, syntax, bufp)
1615 const char *pattern;
1616 int size;
1617 reg_syntax_t syntax;
1618 struct re_pattern_buffer *bufp;
1619{
1620 /* We fetch characters from PATTERN here. Even though PATTERN is
1621 `char *' (i.e., signed), we declare these variables as unsigned, so
1622 they can be reliably used as array indices. */
1623 register unsigned char c, c1;
1624
1625 /* A random temporary spot in PATTERN. */
1626 const char *p1;
1627
1628 /* Points to the end of the buffer, where we should append. */
1629 register unsigned char *b;
1630
1631 /* Keeps track of unclosed groups. */
1632 compile_stack_type compile_stack;
1633
1634 /* Points to the current (ending) position in the pattern. */
1635 const char *p = pattern;
1636 const char *pend = pattern + size;
1637
1638 /* How to translate the characters in the pattern. */
1639 char *translate = bufp->translate;
1640
1641 /* Address of the count-byte of the most recently inserted `exactn'
1642 command. This makes it possible to tell if a new exact-match
1643 character can be added to that command or if the character requires
1644 a new `exactn' command. */
1645 unsigned char *pending_exact = 0;
1646
1647 /* Address of start of the most recently finished expression.
1648 This tells, e.g., postfix * where to find the start of its
1649 operand. Reset at the beginning of groups and alternatives. */
1650 unsigned char *laststart = 0;
1651
1652 /* Address of beginning of regexp, or inside of last group. */
1653 unsigned char *begalt;
1654
1655 /* Place in the uncompiled pattern (i.e., the {) to
1656 which to go back if the interval is invalid. */
1657 const char *beg_interval;
1658
1659 /* Address of the place where a forward jump should go to the end of
1660 the containing expression. Each alternative of an `or' -- except the
1661 last -- ends with a forward jump of this sort. */
1662 unsigned char *fixup_alt_jump = 0;
1663
1664 /* Counts open-groups as they are encountered. Remembered for the
1665 matching close-group on the compile stack, so the same register
1666 number is put in the stop_memory as the start_memory. */
1667 regnum_t regnum = 0;
1668
1669#ifdef DEBUG
1670 DEBUG_PRINT1 ("\nCompiling pattern: ");
1671 if (debug)
1672 {
1673 unsigned debug_count;
1674
1675 for (debug_count = 0; debug_count < size; debug_count++)
1676 putchar (pattern[debug_count]);
1677 putchar ('\n');
1678 }
1679#endif /* DEBUG */
1680
1681 /* Initialize the compile stack. */
1682 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1683 if (compile_stack.stack == NULL)
1684 return REG_ESPACE;
1685
1686 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1687 compile_stack.avail = 0;
1688
1689 /* Initialize the pattern buffer. */
1690 bufp->syntax = syntax;
1691 bufp->fastmap_accurate = 0;
1692 bufp->not_bol = bufp->not_eol = 0;
1693
1694 /* Set `used' to zero, so that if we return an error, the pattern
1695 printer (for debugging) will think there's no pattern. We reset it
1696 at the end. */
1697 bufp->used = 0;
1698
1699 /* Always count groups, whether or not bufp->no_sub is set. */
1700 bufp->re_nsub = 0;
1701
1702#if !defined (emacs) && !defined (SYNTAX_TABLE)
1703 /* Initialize the syntax table. */
1704 init_syntax_once ();
1705#endif
1706
1707 if (bufp->allocated == 0)
1708 {
1709 if (bufp->buffer)
1710 { /* If zero allocated, but buffer is non-null, try to realloc
1711 enough space. This loses if buffer's address is bogus, but
1712 that is the user's responsibility. */
1713 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1714 }
1715 else
1716 { /* Caller did not allocate a buffer. Do it for them. */
1717 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1718 }
1719 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1720
1721 bufp->allocated = INIT_BUF_SIZE;
1722 }
1723
1724 begalt = b = bufp->buffer;
1725
1726 /* Loop through the uncompiled pattern until we're at the end. */
1727 while (p != pend)
1728 {
1729 PATFETCH (c);
1730
1731 switch (c)
1732 {
1733 case '^':
1734 {
1735 if ( /* If at start of pattern, it's an operator. */
1736 p == pattern + 1
1737 /* If context independent, it's an operator. */
1738 || syntax & RE_CONTEXT_INDEP_ANCHORS
1739 /* Otherwise, depends on what's come before. */
1740 || at_begline_loc_p (pattern, p, syntax))
1741 BUF_PUSH (begline);
1742 else
1743 goto normal_char;
1744 }
1745 break;
1746
1747
1748 case '$':
1749 {
1750 if ( /* If at end of pattern, it's an operator. */
1751 p == pend
1752 /* If context independent, it's an operator. */
1753 || syntax & RE_CONTEXT_INDEP_ANCHORS
1754 /* Otherwise, depends on what's next. */
1755 || at_endline_loc_p (p, pend, syntax))
1756 BUF_PUSH (endline);
1757 else
1758 goto normal_char;
1759 }
1760 break;
1761
1762
1763 case '+':
1764 case '?':
1765 if ((syntax & RE_BK_PLUS_QM)
1766 || (syntax & RE_LIMITED_OPS))
1767 goto normal_char;
1768 handle_plus:
1769 case '*':
1770 /* If there is no previous pattern... */
1771 if (!laststart)
1772 {
1773 if (syntax & RE_CONTEXT_INVALID_OPS)
1774 FREE_STACK_RETURN (REG_BADRPT);
1775 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1776 goto normal_char;
1777 }
1778
1779 {
1780 /* Are we optimizing this jump? */
1781 boolean keep_string_p = false;
1782
1783 /* 1 means zero (many) matches is allowed. */
1784 char zero_times_ok = 0, many_times_ok = 0;
1785
1786 /* If there is a sequence of repetition chars, collapse it
1787 down to just one (the right one). We can't combine
1788 interval operators with these because of, e.g., `a{2}*',
1789 which should only match an even number of `a's. */
1790
1791 for (;;)
1792 {
1793 zero_times_ok |= c != '+';
1794 many_times_ok |= c != '?';
1795
1796 if (p == pend)
1797 break;
1798
1799 PATFETCH (c);
1800
1801 if (c == '*'
1802 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
1803 ;
1804
1805 else if (syntax & RE_BK_PLUS_QM && c == '\\')
1806 {
1807 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1808
1809 PATFETCH (c1);
1810 if (!(c1 == '+' || c1 == '?'))
1811 {
1812 PATUNFETCH;
1813 PATUNFETCH;
1814 break;
1815 }
1816
1817 c = c1;
1818 }
1819 else
1820 {
1821 PATUNFETCH;
1822 break;
1823 }
1824
1825 /* If we get here, we found another repeat character. */
1826 }
1827
1828 /* Star, etc. applied to an empty pattern is equivalent
1829 to an empty pattern. */
1830 if (!laststart)
1831 break;
1832
1833 /* Now we know whether or not zero matches is allowed
1834 and also whether or not two or more matches is allowed. */
1835 if (many_times_ok)
1836 { /* More than one repetition is allowed, so put in at the
1837 end a backward relative jump from `b' to before the next
1838 jump we're going to put in below (which jumps from
1839 laststart to after this jump).
1840
1841 But if we are at the `*' in the exact sequence `.*\n',
1842 insert an unconditional jump backwards to the .,
1843 instead of the beginning of the loop. This way we only
1844 push a failure point once, instead of every time
1845 through the loop. */
1846 assert (p - 1 > pattern);
1847
1848 /* Allocate the space for the jump. */
1849 GET_BUFFER_SPACE (3);
1850
1851 /* We know we are not at the first character of the pattern,
1852 because laststart was nonzero. And we've already
1853 incremented `p', by the way, to be the character after
1854 the `*'. Do we have to do something analogous here
1855 for null bytes, because of RE_DOT_NOT_NULL? */
1856 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
1857 && zero_times_ok
1858 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
1859 && !(syntax & RE_DOT_NEWLINE))
1860 { /* We have .*\n. */
1861 STORE_JUMP (jump, b, laststart);
1862 keep_string_p = true;
1863 }
1864 else
1865 /* Anything else. */
1866 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
1867
1868 /* We've added more stuff to the buffer. */
1869 b += 3;
1870 }
1871
1872 /* On failure, jump from laststart to b + 3, which will be the
1873 end of the buffer after this jump is inserted. */
1874 GET_BUFFER_SPACE (3);
1875 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
1876 : on_failure_jump,
1877 laststart, b + 3);
1878 pending_exact = 0;
1879 b += 3;
1880
1881 if (!zero_times_ok)
1882 {
1883 /* At least one repetition is required, so insert a
1884 `dummy_failure_jump' before the initial
1885 `on_failure_jump' instruction of the loop. This
1886 effects a skip over that instruction the first time
1887 we hit that loop. */
1888 GET_BUFFER_SPACE (3);
1889 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
1890 b += 3;
1891 }
1892 }
1893 break;
1894
1895
1896 case '.':
1897 laststart = b;
1898 BUF_PUSH (anychar);
1899 break;
1900
1901
1902 case '[':
1903 {
1904 boolean had_char_class = false;
1905
1906 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1907
1908 /* Ensure that we have enough space to push a charset: the
1909 opcode, the length count, and the bitset; 34 bytes in all. */
1910 GET_BUFFER_SPACE (34);
1911
1912 laststart = b;
1913
1914 /* We test `*p == '^' twice, instead of using an if
1915 statement, so we only need one BUF_PUSH. */
1916 BUF_PUSH (*p == '^' ? charset_not : charset);
1917 if (*p == '^')
1918 p++;
1919
1920 /* Remember the first position in the bracket expression. */
1921 p1 = p;
1922
1923 /* Push the number of bytes in the bitmap. */
1924 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
1925
1926 /* Clear the whole map. */
1927 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
1928
1929 /* charset_not matches newline according to a syntax bit. */
1930 if ((re_opcode_t) b[-2] == charset_not
1931 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
1932 SET_LIST_BIT ('\n');
1933
1934 /* Read in characters and ranges, setting map bits. */
1935 for (;;)
1936 {
1937 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1938
1939 PATFETCH (c);
1940
1941 /* \ might escape characters inside [...] and [^...]. */
1942 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
1943 {
1944 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1945
1946 PATFETCH (c1);
1947 SET_LIST_BIT (c1);
1948 continue;
1949 }
1950
1951 /* Could be the end of the bracket expression. If it's
1952 not (i.e., when the bracket expression is `[]' so
1953 far), the ']' character bit gets set way below. */
1954 if (c == ']' && p != p1 + 1)
1955 break;
1956
1957 /* Look ahead to see if it's a range when the last thing
1958 was a character class. */
1959 if (had_char_class && c == '-' && *p != ']')
1960 FREE_STACK_RETURN (REG_ERANGE);
1961
1962 /* Look ahead to see if it's a range when the last thing
1963 was a character: if this is a hyphen not at the
1964 beginning or the end of a list, then it's the range
1965 operator. */
1966 if (c == '-'
1967 && !(p - 2 >= pattern && p[-2] == '[')
1968 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
1969 && *p != ']')
1970 {
1971 reg_errcode_t ret
1972 = compile_range (&p, pend, translate, syntax, b);
1973 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
1974 }
1975
1976 else if (p[0] == '-' && p[1] != ']')
1977 { /* This handles ranges made up of characters only. */
1978 reg_errcode_t ret;
1979
1980 /* Move past the `-'. */
1981 PATFETCH (c1);
1982
1983 ret = compile_range (&p, pend, translate, syntax, b);
1984 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
1985 }
1986
1987 /* See if we're at the beginning of a possible character
1988 class. */
1989
1990 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
1991 { /* Leave room for the null. */
1992 char str[CHAR_CLASS_MAX_LENGTH + 1];
1993
1994 PATFETCH (c);
1995 c1 = 0;
1996
1997 /* If pattern is `[[:'. */
1998 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1999
2000 for (;;)
2001 {
2002 PATFETCH (c);
2003 if (c == ':' || c == ']' || p == pend
2004 || c1 == CHAR_CLASS_MAX_LENGTH)
2005 break;
2006 str[c1++] = c;
2007 }
2008 str[c1] = '\0';
2009
2010 /* If isn't a word bracketed by `[:' and:`]':
2011 undo the ending character, the letters, and leave
2012 the leading `:' and `[' (but set bits for them). */
2013 if (c == ':' && *p == ']')
2014 {
2015 int ch;
2016 boolean is_alnum = STREQ (str, "alnum");
2017 boolean is_alpha = STREQ (str, "alpha");
2018 boolean is_blank = STREQ (str, "blank");
2019 boolean is_cntrl = STREQ (str, "cntrl");
2020 boolean is_digit = STREQ (str, "digit");
2021 boolean is_graph = STREQ (str, "graph");
2022 boolean is_lower = STREQ (str, "lower");
2023 boolean is_print = STREQ (str, "print");
2024 boolean is_punct = STREQ (str, "punct");
2025 boolean is_space = STREQ (str, "space");
2026 boolean is_upper = STREQ (str, "upper");
2027 boolean is_xdigit = STREQ (str, "xdigit");
2028
2029 if (!IS_CHAR_CLASS (str))
2030 FREE_STACK_RETURN (REG_ECTYPE);
2031
2032 /* Throw away the ] at the end of the character
2033 class. */
2034 PATFETCH (c);
2035
2036 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2037
2038 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2039 {
2040 /* This was split into 3 if's to
2041 avoid an arbitrary limit in some compiler. */
2042 if ( (is_alnum && ISALNUM (ch))
2043 || (is_alpha && ISALPHA (ch))
2044 || (is_blank && ISBLANK (ch))
2045 || (is_cntrl && ISCNTRL (ch)))
2046 SET_LIST_BIT (ch);
2047 if ( (is_digit && ISDIGIT (ch))
2048 || (is_graph && ISGRAPH (ch))
2049 || (is_lower && ISLOWER (ch))
2050 || (is_print && ISPRINT (ch)))
2051 SET_LIST_BIT (ch);
2052 if ( (is_punct && ISPUNCT (ch))
2053 || (is_space && ISSPACE (ch))
2054 || (is_upper && ISUPPER (ch))
2055 || (is_xdigit && ISXDIGIT (ch)))
2056 SET_LIST_BIT (ch);
2057 }
2058 had_char_class = true;
2059 }
2060 else
2061 {
2062 c1++;
2063 while (c1--)
2064 PATUNFETCH;
2065 SET_LIST_BIT ('[');
2066 SET_LIST_BIT (':');
2067 had_char_class = false;
2068 }
2069 }
2070 else
2071 {
2072 had_char_class = false;
2073 SET_LIST_BIT (c);
2074 }
2075 }
2076
2077 /* Discard any (non)matching list bytes that are all 0 at the
2078 end of the map. Decrease the map-length byte too. */
2079 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2080 b[-1]--;
2081 b += b[-1];
2082 }
2083 break;
2084
2085
2086 case '(':
2087 if (syntax & RE_NO_BK_PARENS)
2088 goto handle_open;
2089 else
2090 goto normal_char;
2091
2092
2093 case ')':
2094 if (syntax & RE_NO_BK_PARENS)
2095 goto handle_close;
2096 else
2097 goto normal_char;
2098
2099
2100 case '\n':
2101 if (syntax & RE_NEWLINE_ALT)
2102 goto handle_alt;
2103 else
2104 goto normal_char;
2105
2106
2107 case '|':
2108 if (syntax & RE_NO_BK_VBAR)
2109 goto handle_alt;
2110 else
2111 goto normal_char;
2112
2113
2114 case '{':
2115 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2116 goto handle_interval;
2117 else
2118 goto normal_char;
2119
2120
2121 case '\\':
2122 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2123
2124 /* Do not translate the character after the \, so that we can
2125 distinguish, e.g., \B from \b, even if we normally would
2126 translate, e.g., B to b. */
2127 PATFETCH_RAW (c);
2128
2129 switch (c)
2130 {
2131 case '(':
2132 if (syntax & RE_NO_BK_PARENS)
2133 goto normal_backslash;
2134
2135 handle_open:
2136 bufp->re_nsub++;
2137 regnum++;
2138
2139 if (COMPILE_STACK_FULL)
2140 {
2141 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2142 compile_stack_elt_t);
2143 if (compile_stack.stack == NULL) return REG_ESPACE;
2144
2145 compile_stack.size <<= 1;
2146 }
2147
2148 /* These are the values to restore when we hit end of this
2149 group. They are all relative offsets, so that if the
2150 whole pattern moves because of realloc, they will still
2151 be valid. */
2152 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2153 COMPILE_STACK_TOP.fixup_alt_jump
2154 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2155 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2156 COMPILE_STACK_TOP.regnum = regnum;
2157
2158 /* We will eventually replace the 0 with the number of
2159 groups inner to this one. But do not push a
2160 start_memory for groups beyond the last one we can
2161 represent in the compiled pattern. */
2162 if (regnum <= MAX_REGNUM)
2163 {
2164 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2165 BUF_PUSH_3 (start_memory, regnum, 0);
2166 }
2167
2168 compile_stack.avail++;
2169
2170 fixup_alt_jump = 0;
2171 laststart = 0;
2172 begalt = b;
2173 /* If we've reached MAX_REGNUM groups, then this open
2174 won't actually generate any code, so we'll have to
2175 clear pending_exact explicitly. */
2176 pending_exact = 0;
2177 break;
2178
2179
2180 case ')':
2181 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2182
2183 if (COMPILE_STACK_EMPTY)
2184 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2185 goto normal_backslash;
2186 else
2187 FREE_STACK_RETURN (REG_ERPAREN);
2188
2189 handle_close:
2190 if (fixup_alt_jump)
2191 { /* Push a dummy failure point at the end of the
2192 alternative for a possible future
2193 `pop_failure_jump' to pop. See comments at
2194 `push_dummy_failure' in `re_match_2'. */
2195 BUF_PUSH (push_dummy_failure);
2196
2197 /* We allocated space for this jump when we assigned
2198 to `fixup_alt_jump', in the `handle_alt' case below. */
2199 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2200 }
2201
2202 /* See similar code for backslashed left paren above. */
2203 if (COMPILE_STACK_EMPTY)
2204 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2205 goto normal_char;
2206 else
2207 FREE_STACK_RETURN (REG_ERPAREN);
2208
2209 /* Since we just checked for an empty stack above, this
2210 ``can't happen''. */
2211 assert (compile_stack.avail != 0);
2212 {
2213 /* We don't just want to restore into `regnum', because
2214 later groups should continue to be numbered higher,
2215 as in `(ab)c(de)' -- the second group is #2. */
2216 regnum_t this_group_regnum;
2217
2218 compile_stack.avail--;
2219 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2220 fixup_alt_jump
2221 = COMPILE_STACK_TOP.fixup_alt_jump
2222 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2223 : 0;
2224 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2225 this_group_regnum = COMPILE_STACK_TOP.regnum;
2226 /* If we've reached MAX_REGNUM groups, then this open
2227 won't actually generate any code, so we'll have to
2228 clear pending_exact explicitly. */
2229 pending_exact = 0;
2230
2231 /* We're at the end of the group, so now we know how many
2232 groups were inside this one. */
2233 if (this_group_regnum <= MAX_REGNUM)
2234 {
2235 unsigned char *inner_group_loc
2236 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2237
2238 *inner_group_loc = regnum - this_group_regnum;
2239 BUF_PUSH_3 (stop_memory, this_group_regnum,
2240 regnum - this_group_regnum);
2241 }
2242 }
2243 break;
2244
2245
2246 case '|': /* `\|'. */
2247 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2248 goto normal_backslash;
2249 handle_alt:
2250 if (syntax & RE_LIMITED_OPS)
2251 goto normal_char;
2252
2253 /* Insert before the previous alternative a jump which
2254 jumps to this alternative if the former fails. */
2255 GET_BUFFER_SPACE (3);
2256 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2257 pending_exact = 0;
2258 b += 3;
2259
2260 /* The alternative before this one has a jump after it
2261 which gets executed if it gets matched. Adjust that
2262 jump so it will jump to this alternative's analogous
2263 jump (put in below, which in turn will jump to the next
2264 (if any) alternative's such jump, etc.). The last such
2265 jump jumps to the correct final destination. A picture:
2266 _____ _____
2267 | | | |
2268 | v | v
2269 a | b | c
2270
2271 If we are at `b', then fixup_alt_jump right now points to a
2272 three-byte space after `a'. We'll put in the jump, set
2273 fixup_alt_jump to right after `b', and leave behind three
2274 bytes which we'll fill in when we get to after `c'. */
2275
2276 if (fixup_alt_jump)
2277 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2278
2279 /* Mark and leave space for a jump after this alternative,
2280 to be filled in later either by next alternative or
2281 when know we're at the end of a series of alternatives. */
2282 fixup_alt_jump = b;
2283 GET_BUFFER_SPACE (3);
2284 b += 3;
2285
2286 laststart = 0;
2287 begalt = b;
2288 break;
2289
2290
2291 case '{':
2292 /* If \{ is a literal. */
2293 if (!(syntax & RE_INTERVALS)
2294 /* If we're at `\{' and it's not the open-interval
2295 operator. */
2296 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2297 || (p - 2 == pattern && p == pend))
2298 goto normal_backslash;
2299
2300 handle_interval:
2301 {
2302 /* If got here, then the syntax allows intervals. */
2303
2304 /* At least (most) this many matches must be made. */
2305 int lower_bound = -1, upper_bound = -1;
2306
2307 beg_interval = p - 1;
2308
2309 if (p == pend)
2310 {
2311 if (syntax & RE_NO_BK_BRACES)
2312 goto unfetch_interval;
2313 else
2314 FREE_STACK_RETURN (REG_EBRACE);
2315 }
2316
2317 GET_UNSIGNED_NUMBER (lower_bound);
2318
2319 if (c == ',')
2320 {
2321 GET_UNSIGNED_NUMBER (upper_bound);
2322 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2323 }
2324 else
2325 /* Interval such as `{1}' => match exactly once. */
2326 upper_bound = lower_bound;
2327
2328 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2329 || lower_bound > upper_bound)
2330 {
2331 if (syntax & RE_NO_BK_BRACES)
2332 goto unfetch_interval;
2333 else
2334 FREE_STACK_RETURN (REG_BADBR);
2335 }
2336
2337 if (!(syntax & RE_NO_BK_BRACES))
2338 {
2339 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2340
2341 PATFETCH (c);
2342 }
2343
2344 if (c != '}')
2345 {
2346 if (syntax & RE_NO_BK_BRACES)
2347 goto unfetch_interval;
2348 else
2349 FREE_STACK_RETURN (REG_BADBR);
2350 }
2351
2352 /* We just parsed a valid interval. */
2353
2354 /* If it's invalid to have no preceding re. */
2355 if (!laststart)
2356 {
2357 if (syntax & RE_CONTEXT_INVALID_OPS)
2358 FREE_STACK_RETURN (REG_BADRPT);
2359 else if (syntax & RE_CONTEXT_INDEP_OPS)
2360 laststart = b;
2361 else
2362 goto unfetch_interval;
2363 }
2364
2365 /* If the upper bound is zero, don't want to succeed at
2366 all; jump from `laststart' to `b + 3', which will be
2367 the end of the buffer after we insert the jump. */
2368 if (upper_bound == 0)
2369 {
2370 GET_BUFFER_SPACE (3);
2371 INSERT_JUMP (jump, laststart, b + 3);
2372 b += 3;
2373 }
2374
2375 /* Otherwise, we have a nontrivial interval. When
2376 we're all done, the pattern will look like:
2377 set_number_at <jump count> <upper bound>
2378 set_number_at <succeed_n count> <lower bound>
2379 succeed_n <after jump addr> <succeed_n count>
2380 <body of loop>
2381 jump_n <succeed_n addr> <jump count>
2382 (The upper bound and `jump_n' are omitted if
2383 `upper_bound' is 1, though.) */
2384 else
2385 { /* If the upper bound is > 1, we need to insert
2386 more at the end of the loop. */
2387 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2388
2389 GET_BUFFER_SPACE (nbytes);
2390
2391 /* Initialize lower bound of the `succeed_n', even
2392 though it will be set during matching by its
2393 attendant `set_number_at' (inserted next),
2394 because `re_compile_fastmap' needs to know.
2395 Jump to the `jump_n' we might insert below. */
2396 INSERT_JUMP2 (succeed_n, laststart,
2397 b + 5 + (upper_bound > 1) * 5,
2398 lower_bound);
2399 b += 5;
2400
2401 /* Code to initialize the lower bound. Insert
2402 before the `succeed_n'. The `5' is the last two
2403 bytes of this `set_number_at', plus 3 bytes of
2404 the following `succeed_n'. */
2405 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2406 b += 5;
2407
2408 if (upper_bound > 1)
2409 { /* More than one repetition is allowed, so
2410 append a backward jump to the `succeed_n'
2411 that starts this interval.
2412
2413 When we've reached this during matching,
2414 we'll have matched the interval once, so
2415 jump back only `upper_bound - 1' times. */
2416 STORE_JUMP2 (jump_n, b, laststart + 5,
2417 upper_bound - 1);
2418 b += 5;
2419
2420 /* The location we want to set is the second
2421 parameter of the `jump_n'; that is `b-2' as
2422 an absolute address. `laststart' will be
2423 the `set_number_at' we're about to insert;
2424 `laststart+3' the number to set, the source
2425 for the relative address. But we are
2426 inserting into the middle of the pattern --
2427 so everything is getting moved up by 5.
2428 Conclusion: (b - 2) - (laststart + 3) + 5,
2429 i.e., b - laststart.
2430
2431 We insert this at the beginning of the loop
2432 so that if we fail during matching, we'll
2433 reinitialize the bounds. */
2434 insert_op2 (set_number_at, laststart, b - laststart,
2435 upper_bound - 1, b);
2436 b += 5;
2437 }
2438 }
2439 pending_exact = 0;
2440 beg_interval = NULL;
2441 }
2442 break;
2443
2444 unfetch_interval:
2445 /* If an invalid interval, match the characters as literals. */
2446 assert (beg_interval);
2447 p = beg_interval;
2448 beg_interval = NULL;
2449
2450 /* normal_char and normal_backslash need `c'. */
2451 PATFETCH (c);
2452
2453 if (!(syntax & RE_NO_BK_BRACES))
2454 {
2455 if (p > pattern && p[-1] == '\\')
2456 goto normal_backslash;
2457 }
2458 goto normal_char;
2459
2460#ifdef emacs
2461 /* There is no way to specify the before_dot and after_dot
2462 operators. rms says this is ok. --karl */
2463 case '=':
2464 BUF_PUSH (at_dot);
2465 break;
2466
2467 case 's':
2468 laststart = b;
2469 PATFETCH (c);
2470 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2471 break;
2472
2473 case 'S':
2474 laststart = b;
2475 PATFETCH (c);
2476 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2477 break;
2478#endif /* emacs */
2479
2480
2481 case 'w':
2482 laststart = b;
2483 BUF_PUSH (wordchar);
2484 break;
2485
2486
2487 case 'W':
2488 laststart = b;
2489 BUF_PUSH (notwordchar);
2490 break;
2491
2492
2493 case '<':
2494 BUF_PUSH (wordbeg);
2495 break;
2496
2497 case '>':
2498 BUF_PUSH (wordend);
2499 break;
2500
2501 case 'b':
2502 BUF_PUSH (wordbound);
2503 break;
2504
2505 case 'B':
2506 BUF_PUSH (notwordbound);
2507 break;
2508
2509 case '`':
2510 BUF_PUSH (begbuf);
2511 break;
2512
2513 case '\'':
2514 BUF_PUSH (endbuf);
2515 break;
2516
2517 case '1': case '2': case '3': case '4': case '5':
2518 case '6': case '7': case '8': case '9':
2519 if (syntax & RE_NO_BK_REFS)
2520 goto normal_char;
2521
2522 c1 = c - '0';
2523
2524 if (c1 > regnum)
2525 FREE_STACK_RETURN (REG_ESUBREG);
2526
2527 /* Can't back reference to a subexpression if inside of it. */
2528 if (group_in_compile_stack (compile_stack, c1))
2529 goto normal_char;
2530
2531 laststart = b;
2532 BUF_PUSH_2 (duplicate, c1);
2533 break;
2534
2535
2536 case '+':
2537 case '?':
2538 if (syntax & RE_BK_PLUS_QM)
2539 goto handle_plus;
2540 else
2541 goto normal_backslash;
2542
2543 default:
2544 normal_backslash:
2545 /* You might think it would be useful for \ to mean
2546 not to translate; but if we don't translate it
2547 it will never match anything. */
2548 c = TRANSLATE (c);
2549 goto normal_char;
2550 }
2551 break;
2552
2553
2554 default:
2555 /* Expects the character in `c'. */
2556 normal_char:
2557 /* If no exactn currently being built. */
2558 if (!pending_exact
2559
2560 /* If last exactn not at current position. */
2561 || pending_exact + *pending_exact + 1 != b
2562
2563 /* We have only one byte following the exactn for the count. */
2564 || *pending_exact == (1 << BYTEWIDTH) - 1
2565
2566 /* If followed by a repetition operator. */
2567 || *p == '*' || *p == '^'
2568 || ((syntax & RE_BK_PLUS_QM)
2569 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2570 : (*p == '+' || *p == '?'))
2571 || ((syntax & RE_INTERVALS)
2572 && ((syntax & RE_NO_BK_BRACES)
2573 ? *p == '{'
2574 : (p[0] == '\\' && p[1] == '{'))))
2575 {
2576 /* Start building a new exactn. */
2577
2578 laststart = b;
2579
2580 BUF_PUSH_2 (exactn, 0);
2581 pending_exact = b - 1;
2582 }
2583
2584 BUF_PUSH (c);
2585 (*pending_exact)++;
2586 break;
2587 } /* switch (c) */
2588 } /* while p != pend */
2589
2590
2591 /* Through the pattern now. */
2592
2593 if (fixup_alt_jump)
2594 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2595
2596 if (!COMPILE_STACK_EMPTY)
2597 FREE_STACK_RETURN (REG_EPAREN);
2598
2599 /* If we don't want backtracking, force success
2600 the first time we reach the end of the compiled pattern. */
2601 if (syntax & RE_NO_POSIX_BACKTRACKING)
2602 BUF_PUSH (succeed);
2603
2604 free (compile_stack.stack);
2605
2606 /* We have succeeded; set the length of the buffer. */
2607 bufp->used = b - bufp->buffer;
2608
2609#ifdef DEBUG
2610 if (debug)
2611 {
2612 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2613 print_compiled_pattern (bufp);
2614 }
2615#endif /* DEBUG */
2616
2617#ifndef MATCH_MAY_ALLOCATE
2618 /* Initialize the failure stack to the largest possible stack. This
2619 isn't necessary unless we're trying to avoid calling alloca in
2620 the search and match routines. */
2621 {
2622 int num_regs = bufp->re_nsub + 1;
2623
2624 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2625 is strictly greater than re_max_failures, the largest possible stack
2626 is 2 * re_max_failures failure points. */
2627 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2628 {
2629 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2630
2631#ifdef emacs
2632 if (! fail_stack.stack)
2633 fail_stack.stack
2634 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2635 * sizeof (fail_stack_elt_t));
2636 else
2637 fail_stack.stack
2638 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2639 (fail_stack.size
2640 * sizeof (fail_stack_elt_t)));
2641#else /* not emacs */
2642 if (! fail_stack.stack)
2643 fail_stack.stack
2644 = (fail_stack_elt_t *) malloc (fail_stack.size
2645 * sizeof (fail_stack_elt_t));
2646 else
2647 fail_stack.stack
2648 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2649 (fail_stack.size
2650 * sizeof (fail_stack_elt_t)));
2651#endif /* not emacs */
2652 }
2653
2654 regex_grow_registers (num_regs);
2655 }
2656#endif /* not MATCH_MAY_ALLOCATE */
2657
2658 return REG_NOERROR;
2659} /* regex_compile */
2660
2661/* Subroutines for `regex_compile'. */
2662
2663/* Store OP at LOC followed by two-byte integer parameter ARG. */
2664
2665static void
2666store_op1 (op, loc, arg)
2667 re_opcode_t op;
2668 unsigned char *loc;
2669 int arg;
2670{
2671 *loc = (unsigned char) op;
2672 STORE_NUMBER (loc + 1, arg);
2673}
2674
2675
2676/* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2677
2678static void
2679store_op2 (op, loc, arg1, arg2)
2680 re_opcode_t op;
2681 unsigned char *loc;
2682 int arg1, arg2;
2683{
2684 *loc = (unsigned char) op;
2685 STORE_NUMBER (loc + 1, arg1);
2686 STORE_NUMBER (loc + 3, arg2);
2687}
2688
2689
2690/* Copy the bytes from LOC to END to open up three bytes of space at LOC
2691 for OP followed by two-byte integer parameter ARG. */
2692
2693static void
2694insert_op1 (op, loc, arg, end)
2695 re_opcode_t op;
2696 unsigned char *loc;
2697 int arg;
2698 unsigned char *end;
2699{
2700 register unsigned char *pfrom = end;
2701 register unsigned char *pto = end + 3;
2702
2703 while (pfrom != loc)
2704 *--pto = *--pfrom;
2705
2706 store_op1 (op, loc, arg);
2707}
2708
2709
2710/* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2711
2712static void
2713insert_op2 (op, loc, arg1, arg2, end)
2714 re_opcode_t op;
2715 unsigned char *loc;
2716 int arg1, arg2;
2717 unsigned char *end;
2718{
2719 register unsigned char *pfrom = end;
2720 register unsigned char *pto = end + 5;
2721
2722 while (pfrom != loc)
2723 *--pto = *--pfrom;
2724
2725 store_op2 (op, loc, arg1, arg2);
2726}
2727
2728
2729/* P points to just after a ^ in PATTERN. Return true if that ^ comes
2730 after an alternative or a begin-subexpression. We assume there is at
2731 least one character before the ^. */
2732
2733static boolean
2734at_begline_loc_p (pattern, p, syntax)
2735 const char *pattern, *p;
2736 reg_syntax_t syntax;
2737{
2738 const char *prev = p - 2;
2739 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
2740
2741 return
2742 /* After a subexpression? */
2743 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
2744 /* After an alternative? */
2745 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
2746}
2747
2748
2749/* The dual of at_begline_loc_p. This one is for $. We assume there is
2750 at least one character after the $, i.e., `P < PEND'. */
2751
2752static boolean
2753at_endline_loc_p (p, pend, syntax)
2754 const char *p, *pend;
2755 int syntax;
2756{
2757 const char *next = p;
2758 boolean next_backslash = *next == '\\';
2759 const char *next_next = p + 1 < pend ? p + 1 : NULL;
2760
2761 return
2762 /* Before a subexpression? */
2763 (syntax & RE_NO_BK_PARENS ? *next == ')'
2764 : next_backslash && next_next && *next_next == ')')
2765 /* Before an alternative? */
2766 || (syntax & RE_NO_BK_VBAR ? *next == '|'
2767 : next_backslash && next_next && *next_next == '|');
2768}
2769
2770
2771/* Returns true if REGNUM is in one of COMPILE_STACK's elements and
2772 false if it's not. */
2773
2774static boolean
2775group_in_compile_stack (compile_stack, regnum)
2776 compile_stack_type compile_stack;
2777 regnum_t regnum;
2778{
2779 int this_element;
2780
2781 for (this_element = compile_stack.avail - 1;
2782 this_element >= 0;
2783 this_element--)
2784 if (compile_stack.stack[this_element].regnum == regnum)
2785 return true;
2786
2787 return false;
2788}
2789
2790
2791/* Read the ending character of a range (in a bracket expression) from the
2792 uncompiled pattern *P_PTR (which ends at PEND). We assume the
2793 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
2794 Then we set the translation of all bits between the starting and
2795 ending characters (inclusive) in the compiled pattern B.
2796
2797 Return an error code.
2798
2799 We use these short variable names so we can use the same macros as
2800 `regex_compile' itself. */
2801
2802static reg_errcode_t
2803compile_range (p_ptr, pend, translate, syntax, b)
2804 const char **p_ptr, *pend;
2805 char *translate;
2806 reg_syntax_t syntax;
2807 unsigned char *b;
2808{
2809 unsigned this_char;
2810
2811 const char *p = *p_ptr;
2812 int range_start, range_end;
2813
2814 if (p == pend)
2815 return REG_ERANGE;
2816
2817 /* Even though the pattern is a signed `char *', we need to fetch
2818 with unsigned char *'s; if the high bit of the pattern character
2819 is set, the range endpoints will be negative if we fetch using a
2820 signed char *.
2821
2822 We also want to fetch the endpoints without translating them; the
2823 appropriate translation is done in the bit-setting loop below. */
2824 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
2825 range_start = ((const unsigned char *) p)[-2];
2826 range_end = ((const unsigned char *) p)[0];
2827
2828 /* Have to increment the pointer into the pattern string, so the
2829 caller isn't still at the ending character. */
2830 (*p_ptr)++;
2831
2832 /* If the start is after the end, the range is empty. */
2833 if (range_start > range_end)
2834 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
2835
2836 /* Here we see why `this_char' has to be larger than an `unsigned
2837 char' -- the range is inclusive, so if `range_end' == 0xff
2838 (assuming 8-bit characters), we would otherwise go into an infinite
2839 loop, since all characters <= 0xff. */
2840 for (this_char = range_start; this_char <= range_end; this_char++)
2841 {
2842 SET_LIST_BIT (TRANSLATE (this_char));
2843 }
2844
2845 return REG_NOERROR;
2846}
2847
2848/* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
2849 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
2850 characters can start a string that matches the pattern. This fastmap
2851 is used by re_search to skip quickly over impossible starting points.
2852
2853 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
2854 area as BUFP->fastmap.
2855
2856 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
2857 the pattern buffer.
2858
2859 Returns 0 if we succeed, -2 if an internal error. */
2860
2861int
2862re_compile_fastmap (bufp)
2863 struct re_pattern_buffer *bufp;
2864{
2865 int j, k;
2866#ifdef MATCH_MAY_ALLOCATE
2867 fail_stack_type fail_stack;
2868#endif
2869#ifndef REGEX_MALLOC
2870 char *destination;
2871#endif
2872 /* We don't push any register information onto the failure stack. */
2873 unsigned num_regs = 0;
2874
2875 register char *fastmap = bufp->fastmap;
2876 unsigned char *pattern = bufp->buffer;
2877 unsigned long size = bufp->used;
2878 unsigned char *p = pattern;
2879 register unsigned char *pend = pattern + size;
2880
2881 /* This holds the pointer to the failure stack, when
2882 it is allocated relocatably. */
2883 fail_stack_elt_t *failure_stack_ptr;
2884
2885 /* Assume that each path through the pattern can be null until
2886 proven otherwise. We set this false at the bottom of switch
2887 statement, to which we get only if a particular path doesn't
2888 match the empty string. */
2889 boolean path_can_be_null = true;
2890
2891 /* We aren't doing a `succeed_n' to begin with. */
2892 boolean succeed_n_p = false;
2893
2894 assert (fastmap != NULL && p != NULL);
2895
2896 INIT_FAIL_STACK ();
2897 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
2898 bufp->fastmap_accurate = 1; /* It will be when we're done. */
2899 bufp->can_be_null = 0;
2900
2901 while (1)
2902 {
2903 if (p == pend || *p == succeed)
2904 {
2905 /* We have reached the (effective) end of pattern. */
2906 if (!FAIL_STACK_EMPTY ())
2907 {
2908 bufp->can_be_null |= path_can_be_null;
2909
2910 /* Reset for next path. */
2911 path_can_be_null = true;
2912
2913 p = fail_stack.stack[--fail_stack.avail].pointer;
2914
2915 continue;
2916 }
2917 else
2918 break;
2919 }
2920
2921 /* We should never be about to go beyond the end of the pattern. */
2922 assert (p < pend);
2923
2924 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
2925 {
2926
2927 /* I guess the idea here is to simply not bother with a fastmap
2928 if a backreference is used, since it's too hard to figure out
2929 the fastmap for the corresponding group. Setting
2930 `can_be_null' stops `re_search_2' from using the fastmap, so
2931 that is all we do. */
2932 case duplicate:
2933 bufp->can_be_null = 1;
2934 goto done;
2935
2936
2937 /* Following are the cases which match a character. These end
2938 with `break'. */
2939
2940 case exactn:
2941 fastmap[p[1]] = 1;
2942 break;
2943
2944
2945 case charset:
2946 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
2947 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
2948 fastmap[j] = 1;
2949 break;
2950
2951
2952 case charset_not:
2953 /* Chars beyond end of map must be allowed. */
2954 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
2955 fastmap[j] = 1;
2956
2957 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
2958 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
2959 fastmap[j] = 1;
2960 break;
2961
2962
2963 case wordchar:
2964 for (j = 0; j < (1 << BYTEWIDTH); j++)
2965 if (SYNTAX (j) == Sword)
2966 fastmap[j] = 1;
2967 break;
2968
2969
2970 case notwordchar:
2971 for (j = 0; j < (1 << BYTEWIDTH); j++)
2972 if (SYNTAX (j) != Sword)
2973 fastmap[j] = 1;
2974 break;
2975
2976
2977 case anychar:
2978 {
2979 int fastmap_newline = fastmap['\n'];
2980
2981 /* `.' matches anything ... */
2982 for (j = 0; j < (1 << BYTEWIDTH); j++)
2983 fastmap[j] = 1;
2984
2985 /* ... except perhaps newline. */
2986 if (!(bufp->syntax & RE_DOT_NEWLINE))
2987 fastmap['\n'] = fastmap_newline;
2988
2989 /* Return if we have already set `can_be_null'; if we have,
2990 then the fastmap is irrelevant. Something's wrong here. */
2991 else if (bufp->can_be_null)
2992 goto done;
2993
2994 /* Otherwise, have to check alternative paths. */
2995 break;
2996 }
2997
2998#ifdef emacs
2999 case syntaxspec:
3000 k = *p++;
3001 for (j = 0; j < (1 << BYTEWIDTH); j++)
3002 if (SYNTAX (j) == (enum syntaxcode) k)
3003 fastmap[j] = 1;
3004 break;
3005
3006
3007 case notsyntaxspec:
3008 k = *p++;
3009 for (j = 0; j < (1 << BYTEWIDTH); j++)
3010 if (SYNTAX (j) != (enum syntaxcode) k)
3011 fastmap[j] = 1;
3012 break;
3013
3014
3015 /* All cases after this match the empty string. These end with
3016 `continue'. */
3017
3018
3019 case before_dot:
3020 case at_dot:
3021 case after_dot:
3022 continue;
3023#endif /* not emacs */
3024
3025
3026 case no_op:
3027 case begline:
3028 case endline:
3029 case begbuf:
3030 case endbuf:
3031 case wordbound:
3032 case notwordbound:
3033 case wordbeg:
3034 case wordend:
3035 case push_dummy_failure:
3036 continue;
3037
3038
3039 case jump_n:
3040 case pop_failure_jump:
3041 case maybe_pop_jump:
3042 case jump:
3043 case jump_past_alt:
3044 case dummy_failure_jump:
3045 EXTRACT_NUMBER_AND_INCR (j, p);
3046 p += j;
3047 if (j > 0)
3048 continue;
3049
3050 /* Jump backward implies we just went through the body of a
3051 loop and matched nothing. Opcode jumped to should be
3052 `on_failure_jump' or `succeed_n'. Just treat it like an
3053 ordinary jump. For a * loop, it has pushed its failure
3054 point already; if so, discard that as redundant. */
3055 if ((re_opcode_t) *p != on_failure_jump
3056 && (re_opcode_t) *p != succeed_n)
3057 continue;
3058
3059 p++;
3060 EXTRACT_NUMBER_AND_INCR (j, p);
3061 p += j;
3062
3063 /* If what's on the stack is where we are now, pop it. */
3064 if (!FAIL_STACK_EMPTY ()
3065 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3066 fail_stack.avail--;
3067
3068 continue;
3069
3070
3071 case on_failure_jump:
3072 case on_failure_keep_string_jump:
3073 handle_on_failure_jump:
3074 EXTRACT_NUMBER_AND_INCR (j, p);
3075
3076 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3077 end of the pattern. We don't want to push such a point,
3078 since when we restore it above, entering the switch will
3079 increment `p' past the end of the pattern. We don't need
3080 to push such a point since we obviously won't find any more
3081 fastmap entries beyond `pend'. Such a pattern can match
3082 the null string, though. */
3083 if (p + j < pend)
3084 {
3085 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3086 {
3087 RESET_FAIL_STACK ();
3088 return -2;
3089 }
3090 }
3091 else
3092 bufp->can_be_null = 1;
3093
3094 if (succeed_n_p)
3095 {
3096 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3097 succeed_n_p = false;
3098 }
3099
3100 continue;
3101
3102
3103 case succeed_n:
3104 /* Get to the number of times to succeed. */
3105 p += 2;
3106
3107 /* Increment p past the n for when k != 0. */
3108 EXTRACT_NUMBER_AND_INCR (k, p);
3109 if (k == 0)
3110 {
3111 p -= 4;
3112 succeed_n_p = true; /* Spaghetti code alert. */
3113 goto handle_on_failure_jump;
3114 }
3115 continue;
3116
3117
3118 case set_number_at:
3119 p += 4;
3120 continue;
3121
3122
3123 case start_memory:
3124 case stop_memory:
3125 p += 2;
3126 continue;
3127
3128
3129 default:
3130 abort (); /* We have listed all the cases. */
3131 } /* switch *p++ */
3132
3133 /* Getting here means we have found the possible starting
3134 characters for one path of the pattern -- and that the empty
3135 string does not match. We need not follow this path further.
3136 Instead, look at the next alternative (remembered on the
3137 stack), or quit if no more. The test at the top of the loop
3138 does these things. */
3139 path_can_be_null = false;
3140 p = pend;
3141 } /* while p */
3142
3143 /* Set `can_be_null' for the last path (also the first path, if the
3144 pattern is empty). */
3145 bufp->can_be_null |= path_can_be_null;
3146
3147 done:
3148 RESET_FAIL_STACK ();
3149 return 0;
3150} /* re_compile_fastmap */
3151
3152/* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3153 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3154 this memory for recording register information. STARTS and ENDS
3155 must be allocated using the malloc library routine, and must each
3156 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3157
3158 If NUM_REGS == 0, then subsequent matches should allocate their own
3159 register data.
3160
3161 Unless this function is called, the first search or match using
3162 PATTERN_BUFFER will allocate its own register data, without
3163 freeing the old data. */
3164
3165void
3166re_set_registers (bufp, regs, num_regs, starts, ends)
3167 struct re_pattern_buffer *bufp;
3168 struct re_registers *regs;
3169 unsigned num_regs;
3170 regoff_t *starts, *ends;
3171{
3172 if (num_regs)
3173 {
3174 bufp->regs_allocated = REGS_REALLOCATE;
3175 regs->num_regs = num_regs;
3176 regs->start = starts;
3177 regs->end = ends;
3178 }
3179 else
3180 {
3181 bufp->regs_allocated = REGS_UNALLOCATED;
3182 regs->num_regs = 0;
3183 regs->start = regs->end = (regoff_t *) 0;
3184 }
3185}
3186
3187/* Searching routines. */
3188
3189/* Like re_search_2, below, but only one string is specified, and
3190 doesn't let you say where to stop matching. */
3191
3192int
3193re_search (bufp, string, size, startpos, range, regs)
3194 struct re_pattern_buffer *bufp;
3195 const char *string;
3196 int size, startpos, range;
3197 struct re_registers *regs;
3198{
3199 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3200 regs, size);
3201}
3202
3203
3204/* Using the compiled pattern in BUFP->buffer, first tries to match the
3205 virtual concatenation of STRING1 and STRING2, starting first at index
3206 STARTPOS, then at STARTPOS + 1, and so on.
3207
3208 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3209
3210 RANGE is how far to scan while trying to match. RANGE = 0 means try
3211 only at STARTPOS; in general, the last start tried is STARTPOS +
3212 RANGE.
3213
3214 In REGS, return the indices of the virtual concatenation of STRING1
3215 and STRING2 that matched the entire BUFP->buffer and its contained
3216 subexpressions.
3217
3218 Do not consider matching one past the index STOP in the virtual
3219 concatenation of STRING1 and STRING2.
3220
3221 We return either the position in the strings at which the match was
3222 found, -1 if no match, or -2 if error (such as failure
3223 stack overflow). */
3224
3225int
3226re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3227 struct re_pattern_buffer *bufp;
3228 const char *string1, *string2;
3229 int size1, size2;
3230 int startpos;
3231 int range;
3232 struct re_registers *regs;
3233 int stop;
3234{
3235 int val;
3236 register char *fastmap = bufp->fastmap;
3237 register char *translate = bufp->translate;
3238 int total_size = size1 + size2;
3239 int endpos = startpos + range;
3240
3241 /* Check for out-of-range STARTPOS. */
3242 if (startpos < 0 || startpos > total_size)
3243 return -1;
3244
3245 /* Fix up RANGE if it might eventually take us outside
3246 the virtual concatenation of STRING1 and STRING2. */
3247 if (endpos < -1)
3248 range = -1 - startpos;
3249 else if (endpos > total_size)
3250 range = total_size - startpos;
3251
3252 /* If the search isn't to be a backwards one, don't waste time in a
3253 search for a pattern that must be anchored. */
3254 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3255 {
3256 if (startpos > 0)
3257 return -1;
3258 else
3259 range = 1;
3260 }
3261
3262 /* Update the fastmap now if not correct already. */
3263 if (fastmap && !bufp->fastmap_accurate)
3264 if (re_compile_fastmap (bufp) == -2)
3265 return -2;
3266
3267 /* Loop through the string, looking for a place to start matching. */
3268 for (;;)
3269 {
3270 /* If a fastmap is supplied, skip quickly over characters that
3271 cannot be the start of a match. If the pattern can match the
3272 null string, however, we don't need to skip characters; we want
3273 the first null string. */
3274 if (fastmap && startpos < total_size && !bufp->can_be_null)
3275 {
3276 if (range > 0) /* Searching forwards. */
3277 {
3278 register const char *d;
3279 register int lim = 0;
3280 int irange = range;
3281
3282 if (startpos < size1 && startpos + range >= size1)
3283 lim = range - (size1 - startpos);
3284
3285 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3286
3287 /* Written out as an if-else to avoid testing `translate'
3288 inside the loop. */
3289 if (translate)
3290 while (range > lim
3291 && !fastmap[(unsigned char)
3292 translate[(unsigned char) *d++]])
3293 range--;
3294 else
3295 while (range > lim && !fastmap[(unsigned char) *d++])
3296 range--;
3297
3298 startpos += irange - range;
3299 }
3300 else /* Searching backwards. */
3301 {
3302 register char c = (size1 == 0 || startpos >= size1
3303 ? string2[startpos - size1]
3304 : string1[startpos]);
3305
3306 if (!fastmap[(unsigned char) TRANSLATE (c)])
3307 goto advance;
3308 }
3309 }
3310
3311 /* If can't match the null string, and that's all we have left, fail. */
3312 if (range >= 0 && startpos == total_size && fastmap
3313 && !bufp->can_be_null)
3314 return -1;
3315
3316 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3317 startpos, regs, stop);
3318#ifndef REGEX_MALLOC
3319#ifdef C_ALLOCA
3320 alloca (0);
3321#endif
3322#endif
3323
3324 if (val >= 0)
3325 return startpos;
3326
3327 if (val == -2)
3328 return -2;
3329
3330 advance:
3331 if (!range)
3332 break;
3333 else if (range > 0)
3334 {
3335 range--;
3336 startpos++;
3337 }
3338 else
3339 {
3340 range++;
3341 startpos--;
3342 }
3343 }
3344 return -1;
3345} /* re_search_2 */
3346
3347/* Declarations and macros for re_match_2. */
3348
3349static int bcmp_translate ();
3350static boolean alt_match_null_string_p (),
3351 common_op_match_null_string_p (),
3352 group_match_null_string_p ();
3353
3354/* This converts PTR, a pointer into one of the search strings `string1'
3355 and `string2' into an offset from the beginning of that string. */
3356#define POINTER_TO_OFFSET(ptr) \
3357 (FIRST_STRING_P (ptr) \
3358 ? ((regoff_t) ((ptr) - string1)) \
3359 : ((regoff_t) ((ptr) - string2 + size1)))
3360
3361/* Macros for dealing with the split strings in re_match_2. */
3362
3363#define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3364
3365/* Call before fetching a character with *d. This switches over to
3366 string2 if necessary. */
3367#define PREFETCH() \
3368 while (d == dend) \
3369 { \
3370 /* End of string2 => fail. */ \
3371 if (dend == end_match_2) \
3372 goto fail; \
3373 /* End of string1 => advance to string2. */ \
3374 d = string2; \
3375 dend = end_match_2; \
3376 }
3377
3378
3379/* Test if at very beginning or at very end of the virtual concatenation
3380 of `string1' and `string2'. If only one string, it's `string2'. */
3381#define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3382#define AT_STRINGS_END(d) ((d) == end2)
3383
3384
3385/* Test if D points to a character which is word-constituent. We have
3386 two special cases to check for: if past the end of string1, look at
3387 the first character in string2; and if before the beginning of
3388 string2, look at the last character in string1. */
3389#define WORDCHAR_P(d) \
3390 (SYNTAX ((d) == end1 ? *string2 \
3391 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3392 == Sword)
3393
3394/* Test if the character before D and the one at D differ with respect
3395 to being word-constituent. */
3396#define AT_WORD_BOUNDARY(d) \
3397 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3398 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3399
3400
3401/* Free everything we malloc. */
3402#ifdef MATCH_MAY_ALLOCATE
3403#define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3404#define FREE_VARIABLES() \
3405 do { \
3406 REGEX_FREE_STACK (fail_stack.stack); \
3407 FREE_VAR (regstart); \
3408 FREE_VAR (regend); \
3409 FREE_VAR (old_regstart); \
3410 FREE_VAR (old_regend); \
3411 FREE_VAR (best_regstart); \
3412 FREE_VAR (best_regend); \
3413 FREE_VAR (reg_info); \
3414 FREE_VAR (reg_dummy); \
3415 FREE_VAR (reg_info_dummy); \
3416 } while (0)
3417#else
3418#define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3419#endif /* not MATCH_MAY_ALLOCATE */
3420
3421/* These values must meet several constraints. They must not be valid
3422 register values; since we have a limit of 255 registers (because
3423 we use only one byte in the pattern for the register number), we can
3424 use numbers larger than 255. They must differ by 1, because of
3425 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3426 be larger than the value for the highest register, so we do not try
3427 to actually save any registers when none are active. */
3428#define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3429#define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3430
3431/* Matching routines. */
3432
3433#ifndef emacs /* Emacs never uses this. */
3434/* re_match is like re_match_2 except it takes only a single string. */
3435
3436int
3437re_match (bufp, string, size, pos, regs)
3438 struct re_pattern_buffer *bufp;
3439 const char *string;
3440 int size, pos;
3441 struct re_registers *regs;
3442{
3443 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3444 pos, regs, size);
3445 alloca (0);
3446 return result;
3447}
3448#endif /* not emacs */
3449
3450
3451/* re_match_2 matches the compiled pattern in BUFP against the
3452 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3453 and SIZE2, respectively). We start matching at POS, and stop
3454 matching at STOP.
3455
3456 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3457 store offsets for the substring each group matched in REGS. See the
3458 documentation for exactly how many groups we fill.
3459
3460 We return -1 if no match, -2 if an internal error (such as the
3461 failure stack overflowing). Otherwise, we return the length of the
3462 matched substring. */
3463
3464int
3465re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3466 struct re_pattern_buffer *bufp;
3467 const char *string1, *string2;
3468 int size1, size2;
3469 int pos;
3470 struct re_registers *regs;
3471 int stop;
3472{
3473 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3474 pos, regs, stop);
3475 alloca (0);
3476 return result;
3477}
3478
3479/* This is a separate function so that we can force an alloca cleanup
3480 afterwards. */
3481static int
3482re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3483 struct re_pattern_buffer *bufp;
3484 const char *string1, *string2;
3485 int size1, size2;
3486 int pos;
3487 struct re_registers *regs;
3488 int stop;
3489{
3490 /* General temporaries. */
3491 int mcnt;
3492 unsigned char *p1;
3493
3494 /* Just past the end of the corresponding string. */
3495 const char *end1, *end2;
3496
3497 /* Pointers into string1 and string2, just past the last characters in
3498 each to consider matching. */
3499 const char *end_match_1, *end_match_2;
3500
3501 /* Where we are in the data, and the end of the current string. */
3502 const char *d, *dend;
3503
3504 /* Where we are in the pattern, and the end of the pattern. */
3505 unsigned char *p = bufp->buffer;
3506 register unsigned char *pend = p + bufp->used;
3507
3508 /* Mark the opcode just after a start_memory, so we can test for an
3509 empty subpattern when we get to the stop_memory. */
3510 unsigned char *just_past_start_mem = 0;
3511
3512 /* We use this to map every character in the string. */
3513 char *translate = bufp->translate;
3514
3515 /* Failure point stack. Each place that can handle a failure further
3516 down the line pushes a failure point on this stack. It consists of
3517 restart, regend, and reg_info for all registers corresponding to
3518 the subexpressions we're currently inside, plus the number of such
3519 registers, and, finally, two char *'s. The first char * is where
3520 to resume scanning the pattern; the second one is where to resume
3521 scanning the strings. If the latter is zero, the failure point is
3522 a ``dummy''; if a failure happens and the failure point is a dummy,
3523 it gets discarded and the next next one is tried. */
3524#ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3525 fail_stack_type fail_stack;
3526#endif
3527#ifdef DEBUG
3528 static unsigned failure_id = 0;
3529 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3530#endif
3531
3532 /* This holds the pointer to the failure stack, when
3533 it is allocated relocatably. */
3534 fail_stack_elt_t *failure_stack_ptr;
3535
3536 /* We fill all the registers internally, independent of what we
3537 return, for use in backreferences. The number here includes
3538 an element for register zero. */
3539 unsigned num_regs = bufp->re_nsub + 1;
3540
3541 /* The currently active registers. */
3542 unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3543 unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3544
3545 /* Information on the contents of registers. These are pointers into
3546 the input strings; they record just what was matched (on this
3547 attempt) by a subexpression part of the pattern, that is, the
3548 regnum-th regstart pointer points to where in the pattern we began
3549 matching and the regnum-th regend points to right after where we
3550 stopped matching the regnum-th subexpression. (The zeroth register
3551 keeps track of what the whole pattern matches.) */
3552#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3553 const char **regstart, **regend;
3554#endif
3555
3556 /* If a group that's operated upon by a repetition operator fails to
3557 match anything, then the register for its start will need to be
3558 restored because it will have been set to wherever in the string we
3559 are when we last see its open-group operator. Similarly for a
3560 register's end. */
3561#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3562 const char **old_regstart, **old_regend;
3563#endif
3564
3565 /* The is_active field of reg_info helps us keep track of which (possibly
3566 nested) subexpressions we are currently in. The matched_something
3567 field of reg_info[reg_num] helps us tell whether or not we have
3568 matched any of the pattern so far this time through the reg_num-th
3569 subexpression. These two fields get reset each time through any
3570 loop their register is in. */
3571#ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3572 register_info_type *reg_info;
3573#endif
3574
3575 /* The following record the register info as found in the above
3576 variables when we find a match better than any we've seen before.
3577 This happens as we backtrack through the failure points, which in
3578 turn happens only if we have not yet matched the entire string. */
3579 unsigned best_regs_set = false;
3580#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3581 const char **best_regstart, **best_regend;
3582#endif
3583
3584 /* Logically, this is `best_regend[0]'. But we don't want to have to
3585 allocate space for that if we're not allocating space for anything
3586 else (see below). Also, we never need info about register 0 for
3587 any of the other register vectors, and it seems rather a kludge to
3588 treat `best_regend' differently than the rest. So we keep track of
3589 the end of the best match so far in a separate variable. We
3590 initialize this to NULL so that when we backtrack the first time
3591 and need to test it, it's not garbage. */
3592 const char *match_end = NULL;
3593
3594 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3595 int set_regs_matched_done = 0;
3596
3597 /* Used when we pop values we don't care about. */
3598#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3599 const char **reg_dummy;
3600 register_info_type *reg_info_dummy;
3601#endif
3602
3603#ifdef DEBUG
3604 /* Counts the total number of registers pushed. */
3605 unsigned num_regs_pushed = 0;
3606#endif
3607
3608 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3609
3610 INIT_FAIL_STACK ();
3611
3612#ifdef MATCH_MAY_ALLOCATE
3613 /* Do not bother to initialize all the register variables if there are
3614 no groups in the pattern, as it takes a fair amount of time. If
3615 there are groups, we include space for register 0 (the whole
3616 pattern), even though we never use it, since it simplifies the
3617 array indexing. We should fix this. */
3618 if (bufp->re_nsub)
3619 {
3620 regstart = REGEX_TALLOC (num_regs, const char *);
3621 regend = REGEX_TALLOC (num_regs, const char *);
3622 old_regstart = REGEX_TALLOC (num_regs, const char *);
3623 old_regend = REGEX_TALLOC (num_regs, const char *);
3624 best_regstart = REGEX_TALLOC (num_regs, const char *);
3625 best_regend = REGEX_TALLOC (num_regs, const char *);
3626 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3627 reg_dummy = REGEX_TALLOC (num_regs, const char *);
3628 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3629
3630 if (!(regstart && regend && old_regstart && old_regend && reg_info
3631 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3632 {
3633 FREE_VARIABLES ();
3634 return -2;
3635 }
3636 }
3637 else
3638 {
3639 /* We must initialize all our variables to NULL, so that
3640 `FREE_VARIABLES' doesn't try to free them. */
3641 regstart = regend = old_regstart = old_regend = best_regstart
3642 = best_regend = reg_dummy = NULL;
3643 reg_info = reg_info_dummy = (register_info_type *) NULL;
3644 }
3645#endif /* MATCH_MAY_ALLOCATE */
3646
3647 /* The starting position is bogus. */
3648 if (pos < 0 || pos > size1 + size2)
3649 {
3650 FREE_VARIABLES ();
3651 return -1;
3652 }
3653
3654 /* Initialize subexpression text positions to -1 to mark ones that no
3655 start_memory/stop_memory has been seen for. Also initialize the
3656 register information struct. */
3657 for (mcnt = 1; mcnt < num_regs; mcnt++)
3658 {
3659 regstart[mcnt] = regend[mcnt]
3660 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3661
3662 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3663 IS_ACTIVE (reg_info[mcnt]) = 0;
3664 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3665 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3666 }
3667
3668 /* We move `string1' into `string2' if the latter's empty -- but not if
3669 `string1' is null. */
3670 if (size2 == 0 && string1 != NULL)
3671 {
3672 string2 = string1;
3673 size2 = size1;
3674 string1 = 0;
3675 size1 = 0;
3676 }
3677 end1 = string1 + size1;
3678 end2 = string2 + size2;
3679
3680 /* Compute where to stop matching, within the two strings. */
3681 if (stop <= size1)
3682 {
3683 end_match_1 = string1 + stop;
3684 end_match_2 = string2;
3685 }
3686 else
3687 {
3688 end_match_1 = end1;
3689 end_match_2 = string2 + stop - size1;
3690 }
3691
3692 /* `p' scans through the pattern as `d' scans through the data.
3693 `dend' is the end of the input string that `d' points within. `d'
3694 is advanced into the following input string whenever necessary, but
3695 this happens before fetching; therefore, at the beginning of the
3696 loop, `d' can be pointing at the end of a string, but it cannot
3697 equal `string2'. */
3698 if (size1 > 0 && pos <= size1)
3699 {
3700 d = string1 + pos;
3701 dend = end_match_1;
3702 }
3703 else
3704 {
3705 d = string2 + pos - size1;
3706 dend = end_match_2;
3707 }
3708
3709 DEBUG_PRINT1 ("The compiled pattern is: ");
3710 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
3711 DEBUG_PRINT1 ("The string to match is: `");
3712 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
3713 DEBUG_PRINT1 ("'\n");
3714
3715 /* This loops over pattern commands. It exits by returning from the
3716 function if the match is complete, or it drops through if the match
3717 fails at this starting point in the input data. */
3718 for (;;)
3719 {
3720 DEBUG_PRINT2 ("\n0x%x: ", p);
3721
3722 if (p == pend)
3723 { /* End of pattern means we might have succeeded. */
3724 DEBUG_PRINT1 ("end of pattern ... ");
3725
3726 /* If we haven't matched the entire string, and we want the
3727 longest match, try backtracking. */
3728 if (d != end_match_2)
3729 {
3730 /* 1 if this match ends in the same string (string1 or string2)
3731 as the best previous match. */
3732 boolean same_str_p = (FIRST_STRING_P (match_end)
3733 == MATCHING_IN_FIRST_STRING);
3734 /* 1 if this match is the best seen so far. */
3735 boolean best_match_p;
3736
3737 /* AIX compiler got confused when this was combined
3738 with the previous declaration. */
3739 if (same_str_p)
3740 best_match_p = d > match_end;
3741 else
3742 best_match_p = !MATCHING_IN_FIRST_STRING;
3743
3744 DEBUG_PRINT1 ("backtracking.\n");
3745
3746 if (!FAIL_STACK_EMPTY ())
3747 { /* More failure points to try. */
3748
3749 /* If exceeds best match so far, save it. */
3750 if (!best_regs_set || best_match_p)
3751 {
3752 best_regs_set = true;
3753 match_end = d;
3754
3755 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
3756
3757 for (mcnt = 1; mcnt < num_regs; mcnt++)
3758 {
3759 best_regstart[mcnt] = regstart[mcnt];
3760 best_regend[mcnt] = regend[mcnt];
3761 }
3762 }
3763 goto fail;
3764 }
3765
3766 /* If no failure points, don't restore garbage. And if
3767 last match is real best match, don't restore second
3768 best one. */
3769 else if (best_regs_set && !best_match_p)
3770 {
3771 restore_best_regs:
3772 /* Restore best match. It may happen that `dend ==
3773 end_match_1' while the restored d is in string2.
3774 For example, the pattern `x.*y.*z' against the
3775 strings `x-' and `y-z-', if the two strings are
3776 not consecutive in memory. */
3777 DEBUG_PRINT1 ("Restoring best registers.\n");
3778
3779 d = match_end;
3780 dend = ((d >= string1 && d <= end1)
3781 ? end_match_1 : end_match_2);
3782
3783 for (mcnt = 1; mcnt < num_regs; mcnt++)
3784 {
3785 regstart[mcnt] = best_regstart[mcnt];
3786 regend[mcnt] = best_regend[mcnt];
3787 }
3788 }
3789 } /* d != end_match_2 */
3790
3791 succeed_label:
3792 DEBUG_PRINT1 ("Accepting match.\n");
3793
3794 /* If caller wants register contents data back, do it. */
3795 if (regs && !bufp->no_sub)
3796 {
3797 /* Have the register data arrays been allocated? */
3798 if (bufp->regs_allocated == REGS_UNALLOCATED)
3799 { /* No. So allocate them with malloc. We need one
3800 extra element beyond `num_regs' for the `-1' marker
3801 GNU code uses. */
3802 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
3803 regs->start = TALLOC (regs->num_regs, regoff_t);
3804 regs->end = TALLOC (regs->num_regs, regoff_t);
3805 if (regs->start == NULL || regs->end == NULL)
3806 {
3807 FREE_VARIABLES ();
3808 return -2;
3809 }
3810 bufp->regs_allocated = REGS_REALLOCATE;
3811 }
3812 else if (bufp->regs_allocated == REGS_REALLOCATE)
3813 { /* Yes. If we need more elements than were already
3814 allocated, reallocate them. If we need fewer, just
3815 leave it alone. */
3816 if (regs->num_regs < num_regs + 1)
3817 {
3818 regs->num_regs = num_regs + 1;
3819 RETALLOC (regs->start, regs->num_regs, regoff_t);
3820 RETALLOC (regs->end, regs->num_regs, regoff_t);
3821 if (regs->start == NULL || regs->end == NULL)
3822 {
3823 FREE_VARIABLES ();
3824 return -2;
3825 }
3826 }
3827 }
3828 else
3829 {
3830 /* These braces fend off a "empty body in an else-statement"
3831 warning under GCC when assert expands to nothing. */
3832 assert (bufp->regs_allocated == REGS_FIXED);
3833 }
3834
3835 /* Convert the pointer data in `regstart' and `regend' to
3836 indices. Register zero has to be set differently,
3837 since we haven't kept track of any info for it. */
3838 if (regs->num_regs > 0)
3839 {
3840 regs->start[0] = pos;
3841 regs->end[0] = (MATCHING_IN_FIRST_STRING
3842 ? ((regoff_t) (d - string1))
3843 : ((regoff_t) (d - string2 + size1)));
3844 }
3845
3846 /* Go through the first `min (num_regs, regs->num_regs)'
3847 registers, since that is all we initialized. */
3848 for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++)
3849 {
3850 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
3851 regs->start[mcnt] = regs->end[mcnt] = -1;
3852 else
3853 {
3854 regs->start[mcnt]
3855 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
3856 regs->end[mcnt]
3857 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
3858 }
3859 }
3860
3861 /* If the regs structure we return has more elements than
3862 were in the pattern, set the extra elements to -1. If
3863 we (re)allocated the registers, this is the case,
3864 because we always allocate enough to have at least one
3865 -1 at the end. */
3866 for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++)
3867 regs->start[mcnt] = regs->end[mcnt] = -1;
3868 } /* regs && !bufp->no_sub */
3869
3870 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
3871 nfailure_points_pushed, nfailure_points_popped,
3872 nfailure_points_pushed - nfailure_points_popped);
3873 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
3874
3875 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
3876 ? string1
3877 : string2 - size1);
3878
3879 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
3880
3881 FREE_VARIABLES ();
3882 return mcnt;
3883 }
3884
3885 /* Otherwise match next pattern command. */
3886 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3887 {
3888 /* Ignore these. Used to ignore the n of succeed_n's which
3889 currently have n == 0. */
3890 case no_op:
3891 DEBUG_PRINT1 ("EXECUTING no_op.\n");
3892 break;
3893
3894 case succeed:
3895 DEBUG_PRINT1 ("EXECUTING succeed.\n");
3896 goto succeed_label;
3897
3898 /* Match the next n pattern characters exactly. The following
3899 byte in the pattern defines n, and the n bytes after that
3900 are the characters to match. */
3901 case exactn:
3902 mcnt = *p++;
3903 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
3904
3905 /* This is written out as an if-else so we don't waste time
3906 testing `translate' inside the loop. */
3907 if (translate)
3908 {
3909 do
3910 {
3911 PREFETCH ();
3912 if (translate[(unsigned char) *d++] != (char) *p++)
3913 goto fail;
3914 }
3915 while (--mcnt);
3916 }
3917 else
3918 {
3919 do
3920 {
3921 PREFETCH ();
3922 if (*d++ != (char) *p++) goto fail;
3923 }
3924 while (--mcnt);
3925 }
3926 SET_REGS_MATCHED ();
3927 break;
3928
3929
3930 /* Match any character except possibly a newline or a null. */
3931 case anychar:
3932 DEBUG_PRINT1 ("EXECUTING anychar.\n");
3933
3934 PREFETCH ();
3935
3936 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
3937 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
3938 goto fail;
3939
3940 SET_REGS_MATCHED ();
3941 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
3942 d++;
3943 break;
3944
3945
3946 case charset:
3947 case charset_not:
3948 {
3949 register unsigned char c;
3950 boolean not = (re_opcode_t) *(p - 1) == charset_not;
3951
3952 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
3953
3954 PREFETCH ();
3955 c = TRANSLATE (*d); /* The character to match. */
3956
3957 /* Cast to `unsigned' instead of `unsigned char' in case the
3958 bit list is a full 32 bytes long. */
3959 if (c < (unsigned) (*p * BYTEWIDTH)
3960 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
3961 not = !not;
3962
3963 p += 1 + *p;
3964
3965 if (!not) goto fail;
3966
3967 SET_REGS_MATCHED ();
3968 d++;
3969 break;
3970 }
3971
3972
3973 /* The beginning of a group is represented by start_memory.
3974 The arguments are the register number in the next byte, and the
3975 number of groups inner to this one in the next. The text
3976 matched within the group is recorded (in the internal
3977 registers data structure) under the register number. */
3978 case start_memory:
3979 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
3980
3981 /* Find out if this group can match the empty string. */
3982 p1 = p; /* To send to group_match_null_string_p. */
3983
3984 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
3985 REG_MATCH_NULL_STRING_P (reg_info[*p])
3986 = group_match_null_string_p (&p1, pend, reg_info);
3987
3988 /* Save the position in the string where we were the last time
3989 we were at this open-group operator in case the group is
3990 operated upon by a repetition operator, e.g., with `(a*)*b'
3991 against `ab'; then we want to ignore where we are now in
3992 the string in case this attempt to match fails. */
3993 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
3994 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
3995 : regstart[*p];
3996 DEBUG_PRINT2 (" old_regstart: %d\n",
3997 POINTER_TO_OFFSET (old_regstart[*p]));
3998
3999 regstart[*p] = d;
4000 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4001
4002 IS_ACTIVE (reg_info[*p]) = 1;
4003 MATCHED_SOMETHING (reg_info[*p]) = 0;
4004
4005 /* Clear this whenever we change the register activity status. */
4006 set_regs_matched_done = 0;
4007
4008 /* This is the new highest active register. */
4009 highest_active_reg = *p;
4010
4011 /* If nothing was active before, this is the new lowest active
4012 register. */
4013 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4014 lowest_active_reg = *p;
4015
4016 /* Move past the register number and inner group count. */
4017 p += 2;
4018 just_past_start_mem = p;
4019
4020 break;
4021
4022
4023 /* The stop_memory opcode represents the end of a group. Its
4024 arguments are the same as start_memory's: the register
4025 number, and the number of inner groups. */
4026 case stop_memory:
4027 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4028
4029 /* We need to save the string position the last time we were at
4030 this close-group operator in case the group is operated
4031 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4032 against `aba'; then we want to ignore where we are now in
4033 the string in case this attempt to match fails. */
4034 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4035 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4036 : regend[*p];
4037 DEBUG_PRINT2 (" old_regend: %d\n",
4038 POINTER_TO_OFFSET (old_regend[*p]));
4039
4040 regend[*p] = d;
4041 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4042
4043 /* This register isn't active anymore. */
4044 IS_ACTIVE (reg_info[*p]) = 0;
4045
4046 /* Clear this whenever we change the register activity status. */
4047 set_regs_matched_done = 0;
4048
4049 /* If this was the only register active, nothing is active
4050 anymore. */
4051 if (lowest_active_reg == highest_active_reg)
4052 {
4053 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4054 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4055 }
4056 else
4057 { /* We must scan for the new highest active register, since
4058 it isn't necessarily one less than now: consider
4059 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4060 new highest active register is 1. */
4061 unsigned char r = *p - 1;
4062 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4063 r--;
4064
4065 /* If we end up at register zero, that means that we saved
4066 the registers as the result of an `on_failure_jump', not
4067 a `start_memory', and we jumped to past the innermost
4068 `stop_memory'. For example, in ((.)*) we save
4069 registers 1 and 2 as a result of the *, but when we pop
4070 back to the second ), we are at the stop_memory 1.
4071 Thus, nothing is active. */
4072 if (r == 0)
4073 {
4074 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4075 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4076 }
4077 else
4078 highest_active_reg = r;
4079 }
4080
4081 /* If just failed to match something this time around with a
4082 group that's operated on by a repetition operator, try to
4083 force exit from the ``loop'', and restore the register
4084 information for this group that we had before trying this
4085 last match. */
4086 if ((!MATCHED_SOMETHING (reg_info[*p])
4087 || just_past_start_mem == p - 1)
4088 && (p + 2) < pend)
4089 {
4090 boolean is_a_jump_n = false;
4091
4092 p1 = p + 2;
4093 mcnt = 0;
4094 switch ((re_opcode_t) *p1++)
4095 {
4096 case jump_n:
4097 is_a_jump_n = true;
4098 case pop_failure_jump:
4099 case maybe_pop_jump:
4100 case jump:
4101 case dummy_failure_jump:
4102 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4103 if (is_a_jump_n)
4104 p1 += 2;
4105 break;
4106
4107 default:
4108 /* do nothing */ ;
4109 }
4110 p1 += mcnt;
4111
4112 /* If the next operation is a jump backwards in the pattern
4113 to an on_failure_jump right before the start_memory
4114 corresponding to this stop_memory, exit from the loop
4115 by forcing a failure after pushing on the stack the
4116 on_failure_jump's jump in the pattern, and d. */
4117 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4118 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4119 {
4120 /* If this group ever matched anything, then restore
4121 what its registers were before trying this last
4122 failed match, e.g., with `(a*)*b' against `ab' for
4123 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4124 against `aba' for regend[3].
4125
4126 Also restore the registers for inner groups for,
4127 e.g., `((a*)(b*))*' against `aba' (register 3 would
4128 otherwise get trashed). */
4129
4130 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4131 {
4132 unsigned r;
4133
4134 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4135
4136 /* Restore this and inner groups' (if any) registers. */
4137 for (r = *p; r < *p + *(p + 1); r++)
4138 {
4139 regstart[r] = old_regstart[r];
4140
4141 /* xx why this test? */
4142 if (old_regend[r] >= regstart[r])
4143 regend[r] = old_regend[r];
4144 }
4145 }
4146 p1++;
4147 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4148 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4149
4150 goto fail;
4151 }
4152 }
4153
4154 /* Move past the register number and the inner group count. */
4155 p += 2;
4156 break;
4157
4158
4159 /* \<digit> has been turned into a `duplicate' command which is
4160 followed by the numeric value of <digit> as the register number. */
4161 case duplicate:
4162 {
4163 register const char *d2, *dend2;
4164 int regno = *p++; /* Get which register to match against. */
4165 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4166
4167 /* Can't back reference a group which we've never matched. */
4168 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4169 goto fail;
4170
4171 /* Where in input to try to start matching. */
4172 d2 = regstart[regno];
4173
4174 /* Where to stop matching; if both the place to start and
4175 the place to stop matching are in the same string, then
4176 set to the place to stop, otherwise, for now have to use
4177 the end of the first string. */
4178
4179 dend2 = ((FIRST_STRING_P (regstart[regno])
4180 == FIRST_STRING_P (regend[regno]))
4181 ? regend[regno] : end_match_1);
4182 for (;;)
4183 {
4184 /* If necessary, advance to next segment in register
4185 contents. */
4186 while (d2 == dend2)
4187 {
4188 if (dend2 == end_match_2) break;
4189 if (dend2 == regend[regno]) break;
4190
4191 /* End of string1 => advance to string2. */
4192 d2 = string2;
4193 dend2 = regend[regno];
4194 }
4195 /* At end of register contents => success */
4196 if (d2 == dend2) break;
4197
4198 /* If necessary, advance to next segment in data. */
4199 PREFETCH ();
4200
4201 /* How many characters left in this segment to match. */
4202 mcnt = dend - d;
4203
4204 /* Want how many consecutive characters we can match in
4205 one shot, so, if necessary, adjust the count. */
4206 if (mcnt > dend2 - d2)
4207 mcnt = dend2 - d2;
4208
4209 /* Compare that many; failure if mismatch, else move
4210 past them. */
4211 if (translate
4212 ? bcmp_translate (d, d2, mcnt, translate)
4213 : bcmp (d, d2, mcnt))
4214 goto fail;
4215 d += mcnt, d2 += mcnt;
4216
4217 /* Do this because we've match some characters. */
4218 SET_REGS_MATCHED ();
4219 }
4220 }
4221 break;
4222
4223
4224 /* begline matches the empty string at the beginning of the string
4225 (unless `not_bol' is set in `bufp'), and, if
4226 `newline_anchor' is set, after newlines. */
4227 case begline:
4228 DEBUG_PRINT1 ("EXECUTING begline.\n");
4229
4230 if (AT_STRINGS_BEG (d))
4231 {
4232 if (!bufp->not_bol) break;
4233 }
4234 else if (d[-1] == '\n' && bufp->newline_anchor)
4235 {
4236 break;
4237 }
4238 /* In all other cases, we fail. */
4239 goto fail;
4240
4241
4242 /* endline is the dual of begline. */
4243 case endline:
4244 DEBUG_PRINT1 ("EXECUTING endline.\n");
4245
4246 if (AT_STRINGS_END (d))
4247 {
4248 if (!bufp->not_eol) break;
4249 }
4250
4251 /* We have to ``prefetch'' the next character. */
4252 else if ((d == end1 ? *string2 : *d) == '\n'
4253 && bufp->newline_anchor)
4254 {
4255 break;
4256 }
4257 goto fail;
4258
4259
4260 /* Match at the very beginning of the data. */
4261 case begbuf:
4262 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4263 if (AT_STRINGS_BEG (d))
4264 break;
4265 goto fail;
4266
4267
4268 /* Match at the very end of the data. */
4269 case endbuf:
4270 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4271 if (AT_STRINGS_END (d))
4272 break;
4273 goto fail;
4274
4275
4276 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4277 pushes NULL as the value for the string on the stack. Then
4278 `pop_failure_point' will keep the current value for the
4279 string, instead of restoring it. To see why, consider
4280 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4281 then the . fails against the \n. But the next thing we want
4282 to do is match the \n against the \n; if we restored the
4283 string value, we would be back at the foo.
4284
4285 Because this is used only in specific cases, we don't need to
4286 check all the things that `on_failure_jump' does, to make
4287 sure the right things get saved on the stack. Hence we don't
4288 share its code. The only reason to push anything on the
4289 stack at all is that otherwise we would have to change
4290 `anychar's code to do something besides goto fail in this
4291 case; that seems worse than this. */
4292 case on_failure_keep_string_jump:
4293 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4294
4295 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4296 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4297
4298 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4299 break;
4300
4301
4302 /* Uses of on_failure_jump:
4303
4304 Each alternative starts with an on_failure_jump that points
4305 to the beginning of the next alternative. Each alternative
4306 except the last ends with a jump that in effect jumps past
4307 the rest of the alternatives. (They really jump to the
4308 ending jump of the following alternative, because tensioning
4309 these jumps is a hassle.)
4310
4311 Repeats start with an on_failure_jump that points past both
4312 the repetition text and either the following jump or
4313 pop_failure_jump back to this on_failure_jump. */
4314 case on_failure_jump:
4315 on_failure:
4316 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4317
4318 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4319 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4320
4321 /* If this on_failure_jump comes right before a group (i.e.,
4322 the original * applied to a group), save the information
4323 for that group and all inner ones, so that if we fail back
4324 to this point, the group's information will be correct.
4325 For example, in \(a*\)*\1, we need the preceding group,
4326 and in \(\(a*\)b*\)\2, we need the inner group. */
4327
4328 /* We can't use `p' to check ahead because we push
4329 a failure point to `p + mcnt' after we do this. */
4330 p1 = p;
4331
4332 /* We need to skip no_op's before we look for the
4333 start_memory in case this on_failure_jump is happening as
4334 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4335 against aba. */
4336 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4337 p1++;
4338
4339 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4340 {
4341 /* We have a new highest active register now. This will
4342 get reset at the start_memory we are about to get to,
4343 but we will have saved all the registers relevant to
4344 this repetition op, as described above. */
4345 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4346 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4347 lowest_active_reg = *(p1 + 1);
4348 }
4349
4350 DEBUG_PRINT1 (":\n");
4351 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4352 break;
4353
4354
4355 /* A smart repeat ends with `maybe_pop_jump'.
4356 We change it to either `pop_failure_jump' or `jump'. */
4357 case maybe_pop_jump:
4358 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4359 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4360 {
4361 register unsigned char *p2 = p;
4362
4363 /* Compare the beginning of the repeat with what in the
4364 pattern follows its end. If we can establish that there
4365 is nothing that they would both match, i.e., that we
4366 would have to backtrack because of (as in, e.g., `a*a')
4367 then we can change to pop_failure_jump, because we'll
4368 never have to backtrack.
4369
4370 This is not true in the case of alternatives: in
4371 `(a|ab)*' we do need to backtrack to the `ab' alternative
4372 (e.g., if the string was `ab'). But instead of trying to
4373 detect that here, the alternative has put on a dummy
4374 failure point which is what we will end up popping. */
4375
4376 /* Skip over open/close-group commands.
4377 If what follows this loop is a ...+ construct,
4378 look at what begins its body, since we will have to
4379 match at least one of that. */
4380 while (1)
4381 {
4382 if (p2 + 2 < pend
4383 && ((re_opcode_t) *p2 == stop_memory
4384 || (re_opcode_t) *p2 == start_memory))
4385 p2 += 3;
4386 else if (p2 + 6 < pend
4387 && (re_opcode_t) *p2 == dummy_failure_jump)
4388 p2 += 6;
4389 else
4390 break;
4391 }
4392
4393 p1 = p + mcnt;
4394 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4395 to the `maybe_finalize_jump' of this case. Examine what
4396 follows. */
4397
4398 /* If we're at the end of the pattern, we can change. */
4399 if (p2 == pend)
4400 {
4401 /* Consider what happens when matching ":\(.*\)"
4402 against ":/". I don't really understand this code
4403 yet. */
4404 p[-3] = (unsigned char) pop_failure_jump;
4405 DEBUG_PRINT1
4406 (" End of pattern: change to `pop_failure_jump'.\n");
4407 }
4408
4409 else if ((re_opcode_t) *p2 == exactn
4410 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4411 {
4412 register unsigned char c
4413 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4414
4415 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4416 {
4417 p[-3] = (unsigned char) pop_failure_jump;
4418 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4419 c, p1[5]);
4420 }
4421
4422 else if ((re_opcode_t) p1[3] == charset
4423 || (re_opcode_t) p1[3] == charset_not)
4424 {
4425 int not = (re_opcode_t) p1[3] == charset_not;
4426
4427 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4428 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4429 not = !not;
4430
4431 /* `not' is equal to 1 if c would match, which means
4432 that we can't change to pop_failure_jump. */
4433 if (!not)
4434 {
4435 p[-3] = (unsigned char) pop_failure_jump;
4436 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4437 }
4438 }
4439 }
4440 else if ((re_opcode_t) *p2 == charset)
4441 {
4442#ifdef DEBUG
4443 register unsigned char c
4444 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4445#endif
4446
4447 if ((re_opcode_t) p1[3] == exactn
4448 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[4]
4449 && (p2[1 + p1[4] / BYTEWIDTH]
4450 & (1 << (p1[4] % BYTEWIDTH)))))
4451 {
4452 p[-3] = (unsigned char) pop_failure_jump;
4453 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4454 c, p1[5]);
4455 }
4456
4457 else if ((re_opcode_t) p1[3] == charset_not)
4458 {
4459 int idx;
4460 /* We win if the charset_not inside the loop
4461 lists every character listed in the charset after. */
4462 for (idx = 0; idx < (int) p2[1]; idx++)
4463 if (! (p2[2 + idx] == 0
4464 || (idx < (int) p1[4]
4465 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4466 break;
4467
4468 if (idx == p2[1])
4469 {
4470 p[-3] = (unsigned char) pop_failure_jump;
4471 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4472 }
4473 }
4474 else if ((re_opcode_t) p1[3] == charset)
4475 {
4476 int idx;
4477 /* We win if the charset inside the loop
4478 has no overlap with the one after the loop. */
4479 for (idx = 0;
4480 idx < (int) p2[1] && idx < (int) p1[4];
4481 idx++)
4482 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4483 break;
4484
4485 if (idx == p2[1] || idx == p1[4])
4486 {
4487 p[-3] = (unsigned char) pop_failure_jump;
4488 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4489 }
4490 }
4491 }
4492 }
4493 p -= 2; /* Point at relative address again. */
4494 if ((re_opcode_t) p[-1] != pop_failure_jump)
4495 {
4496 p[-1] = (unsigned char) jump;
4497 DEBUG_PRINT1 (" Match => jump.\n");
4498 goto unconditional_jump;
4499 }
4500 /* Note fall through. */
4501
4502
4503 /* The end of a simple repeat has a pop_failure_jump back to
4504 its matching on_failure_jump, where the latter will push a
4505 failure point. The pop_failure_jump takes off failure
4506 points put on by this pop_failure_jump's matching
4507 on_failure_jump; we got through the pattern to here from the
4508 matching on_failure_jump, so didn't fail. */
4509 case pop_failure_jump:
4510 {
4511 /* We need to pass separate storage for the lowest and
4512 highest registers, even though we don't care about the
4513 actual values. Otherwise, we will restore only one
4514 register from the stack, since lowest will == highest in
4515 `pop_failure_point'. */
4516 unsigned dummy_low_reg, dummy_high_reg;
4517 unsigned char *pdummy;
4518 const char *sdummy;
4519
4520 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4521 POP_FAILURE_POINT (sdummy, pdummy,
4522 dummy_low_reg, dummy_high_reg,
4523 reg_dummy, reg_dummy, reg_info_dummy);
4524 }
4525 /* Note fall through. */
4526
4527
4528 /* Unconditionally jump (without popping any failure points). */
4529 case jump:
4530 unconditional_jump:
4531 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4532 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4533 p += mcnt; /* Do the jump. */
4534 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4535 break;
4536
4537
4538 /* We need this opcode so we can detect where alternatives end
4539 in `group_match_null_string_p' et al. */
4540 case jump_past_alt:
4541 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4542 goto unconditional_jump;
4543
4544
4545 /* Normally, the on_failure_jump pushes a failure point, which
4546 then gets popped at pop_failure_jump. We will end up at
4547 pop_failure_jump, also, and with a pattern of, say, `a+', we
4548 are skipping over the on_failure_jump, so we have to push
4549 something meaningless for pop_failure_jump to pop. */
4550 case dummy_failure_jump:
4551 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4552 /* It doesn't matter what we push for the string here. What
4553 the code at `fail' tests is the value for the pattern. */
4554 PUSH_FAILURE_POINT (0, 0, -2);
4555 goto unconditional_jump;
4556
4557
4558 /* At the end of an alternative, we need to push a dummy failure
4559 point in case we are followed by a `pop_failure_jump', because
4560 we don't want the failure point for the alternative to be
4561 popped. For example, matching `(a|ab)*' against `aab'
4562 requires that we match the `ab' alternative. */
4563 case push_dummy_failure:
4564 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4565 /* See comments just above at `dummy_failure_jump' about the
4566 two zeroes. */
4567 PUSH_FAILURE_POINT (0, 0, -2);
4568 break;
4569
4570 /* Have to succeed matching what follows at least n times.
4571 After that, handle like `on_failure_jump'. */
4572 case succeed_n:
4573 EXTRACT_NUMBER (mcnt, p + 2);
4574 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4575
4576 assert (mcnt >= 0);
4577 /* Originally, this is how many times we HAVE to succeed. */
4578 if (mcnt > 0)
4579 {
4580 mcnt--;
4581 p += 2;
4582 STORE_NUMBER_AND_INCR (p, mcnt);
4583 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt);
4584 }
4585 else if (mcnt == 0)
4586 {
4587 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4588 p[2] = (unsigned char) no_op;
4589 p[3] = (unsigned char) no_op;
4590 goto on_failure;
4591 }
4592 break;
4593
4594 case jump_n:
4595 EXTRACT_NUMBER (mcnt, p + 2);
4596 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4597
4598 /* Originally, this is how many times we CAN jump. */
4599 if (mcnt)
4600 {
4601 mcnt--;
4602 STORE_NUMBER (p + 2, mcnt);
4603 goto unconditional_jump;
4604 }
4605 /* If don't have to jump any more, skip over the rest of command. */
4606 else
4607 p += 4;
4608 break;
4609
4610 case set_number_at:
4611 {
4612 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4613
4614 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4615 p1 = p + mcnt;
4616 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4617 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4618 STORE_NUMBER (p1, mcnt);
4619 break;
4620 }
4621
4622 case wordbound:
4623 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4624 if (AT_WORD_BOUNDARY (d))
4625 break;
4626 goto fail;
4627
4628 case notwordbound:
4629 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4630 if (AT_WORD_BOUNDARY (d))
4631 goto fail;
4632 break;
4633
4634 case wordbeg:
4635 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
4636 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
4637 break;
4638 goto fail;
4639
4640 case wordend:
4641 DEBUG_PRINT1 ("EXECUTING wordend.\n");
4642 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
4643 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
4644 break;
4645 goto fail;
4646
4647#ifdef emacs
4648 case before_dot:
4649 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
4650 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
4651 goto fail;
4652 break;
4653
4654 case at_dot:
4655 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
4656 if (PTR_CHAR_POS ((unsigned char *) d) != point)
4657 goto fail;
4658 break;
4659
4660 case after_dot:
4661 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
4662 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
4663 goto fail;
4664 break;
4665#if 0 /* not emacs19 */
4666 case at_dot:
4667 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
4668 if (PTR_CHAR_POS ((unsigned char *) d) + 1 != point)
4669 goto fail;
4670 break;
4671#endif /* not emacs19 */
4672
4673 case syntaxspec:
4674 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
4675 mcnt = *p++;
4676 goto matchsyntax;
4677
4678 case wordchar:
4679 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
4680 mcnt = (int) Sword;
4681 matchsyntax:
4682 PREFETCH ();
4683 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
4684 d++;
4685 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
4686 goto fail;
4687 SET_REGS_MATCHED ();
4688 break;
4689
4690 case notsyntaxspec:
4691 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
4692 mcnt = *p++;
4693 goto matchnotsyntax;
4694
4695 case notwordchar:
4696 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
4697 mcnt = (int) Sword;
4698 matchnotsyntax:
4699 PREFETCH ();
4700 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
4701 d++;
4702 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
4703 goto fail;
4704 SET_REGS_MATCHED ();
4705 break;
4706
4707#else /* not emacs */
4708 case wordchar:
4709 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
4710 PREFETCH ();
4711 if (!WORDCHAR_P (d))
4712 goto fail;
4713 SET_REGS_MATCHED ();
4714 d++;
4715 break;
4716
4717 case notwordchar:
4718 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
4719 PREFETCH ();
4720 if (WORDCHAR_P (d))
4721 goto fail;
4722 SET_REGS_MATCHED ();
4723 d++;
4724 break;
4725#endif /* not emacs */
4726
4727 default:
4728 abort ();
4729 }
4730 continue; /* Successfully executed one pattern command; keep going. */
4731
4732
4733 /* We goto here if a matching operation fails. */
4734 fail:
4735 if (!FAIL_STACK_EMPTY ())
4736 { /* A restart point is known. Restore to that state. */
4737 DEBUG_PRINT1 ("\nFAIL:\n");
4738 POP_FAILURE_POINT (d, p,
4739 lowest_active_reg, highest_active_reg,
4740 regstart, regend, reg_info);
4741
4742 /* If this failure point is a dummy, try the next one. */
4743 if (!p)
4744 goto fail;
4745
4746 /* If we failed to the end of the pattern, don't examine *p. */
4747 assert (p <= pend);
4748 if (p < pend)
4749 {
4750 boolean is_a_jump_n = false;
4751
4752 /* If failed to a backwards jump that's part of a repetition
4753 loop, need to pop this failure point and use the next one. */
4754 switch ((re_opcode_t) *p)
4755 {
4756 case jump_n:
4757 is_a_jump_n = true;
4758 case maybe_pop_jump:
4759 case pop_failure_jump:
4760 case jump:
4761 p1 = p + 1;
4762 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4763 p1 += mcnt;
4764
4765 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
4766 || (!is_a_jump_n
4767 && (re_opcode_t) *p1 == on_failure_jump))
4768 goto fail;
4769 break;
4770 default:
4771 /* do nothing */ ;
4772 }
4773 }
4774
4775 if (d >= string1 && d <= end1)
4776 dend = end_match_1;
4777 }
4778 else
4779 break; /* Matching at this starting point really fails. */
4780 } /* for (;;) */
4781
4782 if (best_regs_set)
4783 goto restore_best_regs;
4784
4785 FREE_VARIABLES ();
4786
4787 return -1; /* Failure to match. */
4788} /* re_match_2 */
4789
4790/* Subroutine definitions for re_match_2. */
4791
4792
4793/* We are passed P pointing to a register number after a start_memory.
4794
4795 Return true if the pattern up to the corresponding stop_memory can
4796 match the empty string, and false otherwise.
4797
4798 If we find the matching stop_memory, sets P to point to one past its number.
4799 Otherwise, sets P to an undefined byte less than or equal to END.
4800
4801 We don't handle duplicates properly (yet). */
4802
4803static boolean
4804group_match_null_string_p (p, end, reg_info)
4805 unsigned char **p, *end;
4806 register_info_type *reg_info;
4807{
4808 int mcnt;
4809 /* Point to after the args to the start_memory. */
4810 unsigned char *p1 = *p + 2;
4811
4812 while (p1 < end)
4813 {
4814 /* Skip over opcodes that can match nothing, and return true or
4815 false, as appropriate, when we get to one that can't, or to the
4816 matching stop_memory. */
4817
4818 switch ((re_opcode_t) *p1)
4819 {
4820 /* Could be either a loop or a series of alternatives. */
4821 case on_failure_jump:
4822 p1++;
4823 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4824
4825 /* If the next operation is not a jump backwards in the
4826 pattern. */
4827
4828 if (mcnt >= 0)
4829 {
4830 /* Go through the on_failure_jumps of the alternatives,
4831 seeing if any of the alternatives cannot match nothing.
4832 The last alternative starts with only a jump,
4833 whereas the rest start with on_failure_jump and end
4834 with a jump, e.g., here is the pattern for `a|b|c':
4835
4836 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
4837 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
4838 /exactn/1/c
4839
4840 So, we have to first go through the first (n-1)
4841 alternatives and then deal with the last one separately. */
4842
4843
4844 /* Deal with the first (n-1) alternatives, which start
4845 with an on_failure_jump (see above) that jumps to right
4846 past a jump_past_alt. */
4847
4848 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
4849 {
4850 /* `mcnt' holds how many bytes long the alternative
4851 is, including the ending `jump_past_alt' and
4852 its number. */
4853
4854 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
4855 reg_info))
4856 return false;
4857
4858 /* Move to right after this alternative, including the
4859 jump_past_alt. */
4860 p1 += mcnt;
4861
4862 /* Break if it's the beginning of an n-th alternative
4863 that doesn't begin with an on_failure_jump. */
4864 if ((re_opcode_t) *p1 != on_failure_jump)
4865 break;
4866
4867 /* Still have to check that it's not an n-th
4868 alternative that starts with an on_failure_jump. */
4869 p1++;
4870 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4871 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
4872 {
4873 /* Get to the beginning of the n-th alternative. */
4874 p1 -= 3;
4875 break;
4876 }
4877 }
4878
4879 /* Deal with the last alternative: go back and get number
4880 of the `jump_past_alt' just before it. `mcnt' contains
4881 the length of the alternative. */
4882 EXTRACT_NUMBER (mcnt, p1 - 2);
4883
4884 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
4885 return false;
4886
4887 p1 += mcnt; /* Get past the n-th alternative. */
4888 } /* if mcnt > 0 */
4889 break;
4890
4891
4892 case stop_memory:
4893 assert (p1[1] == **p);
4894 *p = p1 + 2;
4895 return true;
4896
4897
4898 default:
4899 if (!common_op_match_null_string_p (&p1, end, reg_info))
4900 return false;
4901 }
4902 } /* while p1 < end */
4903
4904 return false;
4905} /* group_match_null_string_p */
4906
4907
4908/* Similar to group_match_null_string_p, but doesn't deal with alternatives:
4909 It expects P to be the first byte of a single alternative and END one
4910 byte past the last. The alternative can contain groups. */
4911
4912static boolean
4913alt_match_null_string_p (p, end, reg_info)
4914 unsigned char *p, *end;
4915 register_info_type *reg_info;
4916{
4917 int mcnt;
4918 unsigned char *p1 = p;
4919
4920 while (p1 < end)
4921 {
4922 /* Skip over opcodes that can match nothing, and break when we get
4923 to one that can't. */
4924
4925 switch ((re_opcode_t) *p1)
4926 {
4927 /* It's a loop. */
4928 case on_failure_jump:
4929 p1++;
4930 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4931 p1 += mcnt;
4932 break;
4933
4934 default:
4935 if (!common_op_match_null_string_p (&p1, end, reg_info))
4936 return false;
4937 }
4938 } /* while p1 < end */
4939
4940 return true;
4941} /* alt_match_null_string_p */
4942
4943
4944/* Deals with the ops common to group_match_null_string_p and
4945 alt_match_null_string_p.
4946
4947 Sets P to one after the op and its arguments, if any. */
4948
4949static boolean
4950common_op_match_null_string_p (p, end, reg_info)
4951 unsigned char **p, *end;
4952 register_info_type *reg_info;
4953{
4954 int mcnt;
4955 boolean ret;
4956 int reg_no;
4957 unsigned char *p1 = *p;
4958
4959 switch ((re_opcode_t) *p1++)
4960 {
4961 case no_op:
4962 case begline:
4963 case endline:
4964 case begbuf:
4965 case endbuf:
4966 case wordbeg:
4967 case wordend:
4968 case wordbound:
4969 case notwordbound:
4970#ifdef emacs
4971 case before_dot:
4972 case at_dot:
4973 case after_dot:
4974#endif
4975 break;
4976
4977 case start_memory:
4978 reg_no = *p1;
4979 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
4980 ret = group_match_null_string_p (&p1, end, reg_info);
4981
4982 /* Have to set this here in case we're checking a group which
4983 contains a group and a back reference to it. */
4984
4985 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
4986 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
4987
4988 if (!ret)
4989 return false;
4990 break;
4991
4992 /* If this is an optimized succeed_n for zero times, make the jump. */
4993 case jump:
4994 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4995 if (mcnt >= 0)
4996 p1 += mcnt;
4997 else
4998 return false;
4999 break;
5000
5001 case succeed_n:
5002 /* Get to the number of times to succeed. */
5003 p1 += 2;
5004 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5005
5006 if (mcnt == 0)
5007 {
5008 p1 -= 4;
5009 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5010 p1 += mcnt;
5011 }
5012 else
5013 return false;
5014 break;
5015
5016 case duplicate:
5017 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5018 return false;
5019 break;
5020
5021 case set_number_at:
5022 p1 += 4;
5023
5024 default:
5025 /* All other opcodes mean we cannot match the empty string. */
5026 return false;
5027 }
5028
5029 *p = p1;
5030 return true;
5031} /* common_op_match_null_string_p */
5032
5033
5034/* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5035 bytes; nonzero otherwise. */
5036
5037static int
5038bcmp_translate (s1, s2, len, translate)
5039 unsigned char *s1, *s2;
5040 register int len;
5041 char *translate;
5042{
5043 register unsigned char *p1 = s1, *p2 = s2;
5044 while (len)
5045 {
5046 if (translate[*p1++] != translate[*p2++]) return 1;
5047 len--;
5048 }
5049 return 0;
5050}
5051
5052/* Entry points for GNU code. */
5053
5054/* re_compile_pattern is the GNU regular expression compiler: it
5055 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5056 Returns 0 if the pattern was valid, otherwise an error string.
5057
5058 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5059 are set in BUFP on entry.
5060
5061 We call regex_compile to do the actual compilation. */
5062
5063const char *
5064re_compile_pattern (pattern, length, bufp)
5065 const char *pattern;
5066 int length;
5067 struct re_pattern_buffer *bufp;
5068{
5069 reg_errcode_t ret;
5070
5071 /* GNU code is written to assume at least RE_NREGS registers will be set
5072 (and at least one extra will be -1). */
5073 bufp->regs_allocated = REGS_UNALLOCATED;
5074
5075 /* And GNU code determines whether or not to get register information
5076 by passing null for the REGS argument to re_match, etc., not by
5077 setting no_sub. */
5078 bufp->no_sub = 0;
5079
5080 /* Match anchors at newline. */
5081 bufp->newline_anchor = 1;
5082
5083 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5084
5085 if (!ret)
5086 return NULL;
5087 return gettext (re_error_msgid[(int) ret]);
5088}
5089
5090/* Entry points compatible with 4.2 BSD regex library. We don't define
5091 them unless specifically requested. */
5092
5093#ifdef _REGEX_RE_COMP
5094
5095/* BSD has one and only one pattern buffer. */
5096static struct re_pattern_buffer re_comp_buf;
5097
5098char *
5099re_comp (s)
5100 const char *s;
5101{
5102 reg_errcode_t ret;
5103
5104 if (!s)
5105 {
5106 if (!re_comp_buf.buffer)
5107 return gettext ("No previous regular expression");
5108 return 0;
5109 }
5110
5111 if (!re_comp_buf.buffer)
5112 {
5113 re_comp_buf.buffer = (unsigned char *) malloc (200);
5114 if (re_comp_buf.buffer == NULL)
5115 return gettext (re_error_msgid[(int) REG_ESPACE]);
5116 re_comp_buf.allocated = 200;
5117
5118 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5119 if (re_comp_buf.fastmap == NULL)
5120 return gettext (re_error_msgid[(int) REG_ESPACE]);
5121 }
5122
5123 /* Since `re_exec' always passes NULL for the `regs' argument, we
5124 don't need to initialize the pattern buffer fields which affect it. */
5125
5126 /* Match anchors at newlines. */
5127 re_comp_buf.newline_anchor = 1;
5128
5129 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5130
5131 if (!ret)
5132 return NULL;
5133
5134 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5135 return (char *) gettext (re_error_msgid[(int) ret]);
5136}
5137
5138
5139int
5140re_exec (s)
5141 const char *s;
5142{
5143 const int len = strlen (s);
5144 return
5145 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5146}
5147#endif /* _REGEX_RE_COMP */
5148
5149/* POSIX.2 functions. Don't define these for Emacs. */
5150
5151#ifndef emacs
5152
5153/* regcomp takes a regular expression as a string and compiles it.
5154
5155 PREG is a regex_t *. We do not expect any fields to be initialized,
5156 since POSIX says we shouldn't. Thus, we set
5157
5158 `buffer' to the compiled pattern;
5159 `used' to the length of the compiled pattern;
5160 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5161 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5162 RE_SYNTAX_POSIX_BASIC;
5163 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5164 `fastmap' and `fastmap_accurate' to zero;
5165 `re_nsub' to the number of subexpressions in PATTERN.
5166
5167 PATTERN is the address of the pattern string.
5168
5169 CFLAGS is a series of bits which affect compilation.
5170
5171 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5172 use POSIX basic syntax.
5173
5174 If REG_NEWLINE is set, then . and [^...] don't match newline.
5175 Also, regexec will try a match beginning after every newline.
5176
5177 If REG_ICASE is set, then we considers upper- and lowercase
5178 versions of letters to be equivalent when matching.
5179
5180 If REG_NOSUB is set, then when PREG is passed to regexec, that
5181 routine will report only success or failure, and nothing about the
5182 registers.
5183
5184 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
5185 the return codes and their meanings.) */
5186
5187int
5188regcomp (preg, pattern, cflags)
5189 regex_t *preg;
5190 const char *pattern;
5191 int cflags;
5192{
5193 reg_errcode_t ret;
5194 unsigned syntax
5195 = (cflags & REG_EXTENDED) ?
5196 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5197
5198 /* regex_compile will allocate the space for the compiled pattern. */
5199 preg->buffer = 0;
5200 preg->allocated = 0;
5201 preg->used = 0;
5202
5203 /* Don't bother to use a fastmap when searching. This simplifies the
5204 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
5205 characters after newlines into the fastmap. This way, we just try
5206 every character. */
5207 preg->fastmap = 0;
5208
5209 if (cflags & REG_ICASE)
5210 {
5211 unsigned i;
5212
5213 preg->translate = (char *) malloc (CHAR_SET_SIZE);
5214 if (preg->translate == NULL)
5215 return (int) REG_ESPACE;
5216
5217 /* Map uppercase characters to corresponding lowercase ones. */
5218 for (i = 0; i < CHAR_SET_SIZE; i++)
5219 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
5220 }
5221 else
5222 preg->translate = NULL;
5223
5224 /* If REG_NEWLINE is set, newlines are treated differently. */
5225 if (cflags & REG_NEWLINE)
5226 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5227 syntax &= ~RE_DOT_NEWLINE;
5228 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5229 /* It also changes the matching behavior. */
5230 preg->newline_anchor = 1;
5231 }
5232 else
5233 preg->newline_anchor = 0;
5234
5235 preg->no_sub = !!(cflags & REG_NOSUB);
5236
5237 /* POSIX says a null character in the pattern terminates it, so we
5238 can use strlen here in compiling the pattern. */
5239 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5240
5241 /* POSIX doesn't distinguish between an unmatched open-group and an
5242 unmatched close-group: both are REG_EPAREN. */
5243 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5244
5245 return (int) ret;
5246}
5247
5248
5249/* regexec searches for a given pattern, specified by PREG, in the
5250 string STRING.
5251
5252 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5253 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5254 least NMATCH elements, and we set them to the offsets of the
5255 corresponding matched substrings.
5256
5257 EFLAGS specifies `execution flags' which affect matching: if
5258 REG_NOTBOL is set, then ^ does not match at the beginning of the
5259 string; if REG_NOTEOL is set, then $ does not match at the end.
5260
5261 We return 0 if we find a match and REG_NOMATCH if not. */
5262
5263int
5264regexec (preg, string, nmatch, pmatch, eflags)
5265 const regex_t *preg;
5266 const char *string;
5267 size_t nmatch;
5268 regmatch_t pmatch[];
5269 int eflags;
5270{
5271 int ret;
5272 struct re_registers regs;
5273 regex_t private_preg;
5274 int len = strlen (string);
5275 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5276
5277 private_preg = *preg;
5278
5279 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5280 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5281
5282 /* The user has told us exactly how many registers to return
5283 information about, via `nmatch'. We have to pass that on to the
5284 matching routines. */
5285 private_preg.regs_allocated = REGS_FIXED;
5286
5287 if (want_reg_info)
5288 {
5289 regs.num_regs = nmatch;
5290 regs.start = TALLOC (nmatch, regoff_t);
5291 regs.end = TALLOC (nmatch, regoff_t);
5292 if (regs.start == NULL || regs.end == NULL)
5293 return (int) REG_NOMATCH;
5294 }
5295
5296 /* Perform the searching operation. */
5297 ret = re_search (&private_preg, string, len,
5298 /* start: */ 0, /* range: */ len,
5299 want_reg_info ? &regs : (struct re_registers *) 0);
5300
5301 /* Copy the register information to the POSIX structure. */
5302 if (want_reg_info)
5303 {
5304 if (ret >= 0)
5305 {
5306 unsigned r;
5307
5308 for (r = 0; r < nmatch; r++)
5309 {
5310 pmatch[r].rm_so = regs.start[r];
5311 pmatch[r].rm_eo = regs.end[r];
5312 }
5313 }
5314
5315 /* If we needed the temporary register info, free the space now. */
5316 free (regs.start);
5317 free (regs.end);
5318 }
5319
5320 /* We want zero return to mean success, unlike `re_search'. */
5321 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5322}
5323
5324
5325/* Returns a message corresponding to an error code, ERRCODE, returned
5326 from either regcomp or regexec. We don't use PREG here. */
5327
5328size_t
5329regerror (errcode, preg, errbuf, errbuf_size)
5330 int errcode;
5331 const regex_t *preg;
5332 char *errbuf;
5333 size_t errbuf_size;
5334{
5335 const char *msg;
5336 size_t msg_size;
5337
5338 if (errcode < 0
5339 || errcode >= (sizeof (re_error_msgid) / sizeof (re_error_msgid[0])))
5340 /* Only error codes returned by the rest of the code should be passed
5341 to this routine. If we are given anything else, or if other regex
5342 code generates an invalid error code, then the program has a bug.
5343 Dump core so we can fix it. */
5344 abort ();
5345
5346 msg = gettext (re_error_msgid[errcode]);
5347
5348 msg_size = strlen (msg) + 1; /* Includes the null. */
5349
5350 if (errbuf_size != 0)
5351 {
5352 if (msg_size > errbuf_size)
5353 {
5354 strncpy (errbuf, msg, errbuf_size - 1);
5355 errbuf[errbuf_size - 1] = 0;
5356 }
5357 else
5358 strcpy (errbuf, msg);
5359 }
5360
5361 return msg_size;
5362}
5363
5364
5365/* Free dynamically allocated space used by PREG. */
5366
5367void
5368regfree (preg)
5369 regex_t *preg;
5370{
5371 if (preg->buffer != NULL)
5372 free (preg->buffer);
5373 preg->buffer = NULL;
5374
5375 preg->allocated = 0;
5376 preg->used = 0;
5377
5378 if (preg->fastmap != NULL)
5379 free (preg->fastmap);
5380 preg->fastmap = NULL;
5381 preg->fastmap_accurate = 0;
5382
5383 if (preg->translate != NULL)
5384 free (preg->translate);
5385 preg->translate = NULL;
5386}
5387
5388#endif /* not emacs */
5389
5390/*
5391Local variables:
5392make-backup-files: t
5393version-control: t
5394trim-versions-without-asking: nil
5395End:
5396*/
generated by cgit v1.2.1 (git 2.18.0) at 2026-04-24 14:59:37 +0000