1 #ifndef lint /* SRE: lint *hates* this code! */
2 /* This is the GNU regular expression library.
3 * I've modified it for portability and clarity (!).
4 * "SRE" comments flag the changes I've made.
5 * Thanks to the GNU folks for providing this code!
6 * SRE, Sun Jan 7 09:48:26 1996
7 */
8
9 /* Extended regular expression matching and search library,
10 version 0.12.
11 (Implements POSIX draft P10003.2/D11.2, except for
12 internationalization features.)
13
14 Copyright (C) 1993 Free Software Foundation, Inc.
15
16 This program is free software; you can redistribute it and/or modify
17 it under the terms of the GNU General Public License as published by
18 the Free Software Foundation; either version 2, or (at your option)
19 any later version.
20
21 This program is distributed in the hope that it will be useful,
22 but WITHOUT ANY WARRANTY; without even the implied warranty of
23 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
24 GNU General Public License for more details.
25
26 You should have received a copy of the GNU General Public License
27 along with this program; if not, write to the Free Software
28 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
29
30 /* AIX requires this to be the first thing in the file. */
31 #if defined (_AIX) && !defined (REGEX_MALLOC)
32 #pragma alloca
33 #endif
34
35 #define _GNU_SOURCE
36
37 /* We need this for `regex.h', and perhaps for the Emacs include files. */
38 #include <sys/types.h>
39
40 #ifdef HAVE_CONFIG_H
41 #include "config.h"
42 #endif
43
44 /* SRE: removed ifdef emacs; removed if HAVE_STRING_H; removed STDC_HEADERS */
45 #include <string.h>
46 #include <stdlib.h>
47
48 #ifndef bcmp
49 #define bcmp(s1, s2, n) memcmp ((s1), (s2), (n))
50 #endif
51 #ifndef bcopy
52 #define bcopy(s, d, n) memcpy ((d), (s), (n))
53 #endif
54 #ifndef bzero
55 #define bzero(s, n) memset ((s), 0, (n))
56 #endif
57
58
59 /* Define the syntax stuff for \<, \>, etc. */
60
61 /* This must be nonzero for the wordchar and notwordchar pattern
62 commands in re_match_2. */
63 #ifndef Sword
64 #define Sword 1
65 #endif
66
67 #ifdef SYNTAX_TABLE
68
69 extern char *re_syntax_table;
70
71 #else /* not SYNTAX_TABLE */
72
73 /* How many characters in the character set. */
74 #define CHAR_SET_SIZE 256
75
76 static char re_syntax_table[CHAR_SET_SIZE];
77
78 static void
init_syntax_once()79 init_syntax_once ()
80 {
81 register int c;
82 static int done = 0;
83
84 if (done)
85 return;
86
87 bzero (re_syntax_table, sizeof re_syntax_table);
88
89 for (c = 'a'; c <= 'z'; c++)
90 re_syntax_table[c] = Sword;
91
92 for (c = 'A'; c <= 'Z'; c++)
93 re_syntax_table[c] = Sword;
94
95 for (c = '0'; c <= '9'; c++)
96 re_syntax_table[c] = Sword;
97
98 re_syntax_table['_'] = Sword;
99
100 done = 1;
101 }
102
103 #endif /* not SYNTAX_TABLE */
104
105 #define SYNTAX(c) re_syntax_table[c]
106
107
108
109 /* Get the interface, including the syntax bits. */
110 #include "gnuregex.h"
111
112 /* isalpha etc. are used for the character classes. */
113 #include <ctype.h>
114
115 #ifndef isascii
116 #define isascii(c) 1
117 #endif
118
119 #ifdef isblank
120 #define ISBLANK(c) (isascii (c) && isblank (c))
121 #else
122 #define ISBLANK(c) ((c) == ' ' || (c) == '\t')
123 #endif
124 #ifdef isgraph
125 #define ISGRAPH(c) (isascii (c) && isgraph (c))
126 #else
127 #define ISGRAPH(c) (isascii (c) && isprint (c) && !isspace (c))
128 #endif
129
130 #define ISPRINT(c) (isascii (c) && isprint (c))
131 #define ISDIGIT(c) (isascii (c) && isdigit (c))
132 #define ISALNUM(c) (isascii (c) && isalnum (c))
133 #define ISALPHA(c) (isascii (c) && isalpha (c))
134 #define ISCNTRL(c) (isascii (c) && iscntrl (c))
135 #define ISLOWER(c) (isascii (c) && islower (c))
136 #define ISPUNCT(c) (isascii (c) && ispunct (c))
137 #define ISSPACE(c) (isascii (c) && isspace (c))
138 #define ISUPPER(c) (isascii (c) && isupper (c))
139 #define ISXDIGIT(c) (isascii (c) && isxdigit (c))
140
141 #ifndef NULL
142 #define NULL 0
143 #endif
144
145 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
146 since ours (we hope) works properly with all combinations of
147 machines, compilers, `char' and `unsigned char' argument types.
148 (Per Bothner suggested the basic approach.) */
149 #undef SIGN_EXTEND_CHAR
150 #if __STDC__
151 #define SIGN_EXTEND_CHAR(c) ((signed char) (c))
152 #else /* not __STDC__ */
153 /* As in Harbison and Steele. */
154 #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
155 #endif
156
157 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
158 use `alloca' instead of `malloc'. This is because using malloc in
159 re_search* or re_match* could cause memory leaks when C-g is used in
160 Emacs; also, malloc is slower and causes storage fragmentation. On
161 the other hand, malloc is more portable, and easier to debug.
162
163 Because we sometimes use alloca, some routines have to be macros,
164 not functions -- `alloca'-allocated space disappears at the end of the
165 function it is called in. */
166
167 #define REGEX_MALLOC /* SRE -- alloca not portable? */
168
169 #ifdef REGEX_MALLOC
170
171 #define REGEX_ALLOCATE malloc
172 #define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
173
174 #else /* not REGEX_MALLOC */
175
176 /* Emacs already defines alloca, sometimes. */
177 #ifndef alloca
178
179 /* Make alloca work the best possible way. */
180 #ifdef __GNUC__
181 #define alloca __builtin_alloca
182 #else /* not __GNUC__ */
183 #if HAVE_ALLOCA_H
184 #include <alloca.h>
185 #else /* not __GNUC__ or HAVE_ALLOCA_H */
186 #ifndef _AIX /* Already did AIX, up at the top. */
187 char *alloca ();
188 #endif /* not _AIX */
189 #endif /* not HAVE_ALLOCA_H */
190 #endif /* not __GNUC__ */
191
192 #endif /* not alloca */
193
194 #define REGEX_ALLOCATE alloca
195
196 /* Assumes a `char *destination' variable. */
197 #define REGEX_REALLOCATE(source, osize, nsize) \
198 (destination = (char *) alloca (nsize), \
199 bcopy (source, destination, osize), \
200 destination)
201
202 #endif /* not REGEX_MALLOC */
203
204
205 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
206 `string1' or just past its end. This works if PTR is NULL, which is
207 a good thing. */
208 #define FIRST_STRING_P(ptr) \
209 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
210
211 /* (Re)Allocate N items of type T using malloc, or fail. */
212 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
213 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
214 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
215
216 #define BYTEWIDTH 8 /* In bits. */
217
218 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
219
220 #define MAX(a, b) ((a) > (b) ? (a) : (b))
221 #define MIN(a, b) ((a) < (b) ? (a) : (b))
222
223 typedef char boolean;
224 #define false 0
225 #define true 1
226
227 /* These are the command codes that appear in compiled regular
228 expressions. Some opcodes are followed by argument bytes. A
229 command code can specify any interpretation whatsoever for its
230 arguments. Zero bytes may appear in the compiled regular expression.
231
232 The value of `exactn' is needed in search.c (search_buffer) in Emacs.
233 So regex.h defines a symbol `RE_EXACTN_VALUE' to be 1; the value of
234 `exactn' we use here must also be 1. */
235
236 typedef enum
237 {
238 no_op = 0,
239
240 /* Followed by one byte giving n, then by n literal bytes. */
241 exactn = 1,
242
243 /* Matches any (more or less) character. */
244 anychar,
245
246 /* Matches any one char belonging to specified set. First
247 following byte is number of bitmap bytes. Then come bytes
248 for a bitmap saying which chars are in. Bits in each byte
249 are ordered low-bit-first. A character is in the set if its
250 bit is 1. A character too large to have a bit in the map is
251 automatically not in the set. */
252 charset,
253
254 /* Same parameters as charset, but match any character that is
255 not one of those specified. */
256 charset_not,
257
258 /* Start remembering the text that is matched, for storing in a
259 register. Followed by one byte with the register number, in
260 the range 0 to one less than the pattern buffer's re_nsub
261 field. Then followed by one byte with the number of groups
262 inner to this one. (This last has to be part of the
263 start_memory only because we need it in the on_failure_jump
264 of re_match_2.) */
265 start_memory,
266
267 /* Stop remembering the text that is matched and store it in a
268 memory register. Followed by one byte with the register
269 number, in the range 0 to one less than `re_nsub' in the
270 pattern buffer, and one byte with the number of inner groups,
271 just like `start_memory'. (We need the number of inner
272 groups here because we don't have any easy way of finding the
273 corresponding start_memory when we're at a stop_memory.) */
274 stop_memory,
275
276 /* Match a duplicate of something remembered. Followed by one
277 byte containing the register number. */
278 duplicate,
279
280 /* Fail unless at beginning of line. */
281 begline,
282
283 /* Fail unless at end of line. */
284 endline,
285
286 /* Succeeds if at beginning of buffer (if emacs) or at beginning
287 of string to be matched (if not). */
288 begbuf,
289
290 /* Analogously, for end of buffer/string. */
291 endbuf,
292
293 /* Followed by two byte relative address to which to jump. */
294 jump,
295
296 /* Same as jump, but marks the end of an alternative. */
297 jump_past_alt,
298
299 /* Followed by two-byte relative address of place to resume at
300 in case of failure. */
301 on_failure_jump,
302
303 /* Like on_failure_jump, but pushes a placeholder instead of the
304 current string position when executed. */
305 on_failure_keep_string_jump,
306
307 /* Throw away latest failure point and then jump to following
308 two-byte relative address. */
309 pop_failure_jump,
310
311 /* Change to pop_failure_jump if know won't have to backtrack to
312 match; otherwise change to jump. This is used to jump
313 back to the beginning of a repeat. If what follows this jump
314 clearly won't match what the repeat does, such that we can be
315 sure that there is no use backtracking out of repetitions
316 already matched, then we change it to a pop_failure_jump.
317 Followed by two-byte address. */
318 maybe_pop_jump,
319
320 /* Jump to following two-byte address, and push a dummy failure
321 point. This failure point will be thrown away if an attempt
322 is made to use it for a failure. A `+' construct makes this
323 before the first repeat. Also used as an intermediary kind
324 of jump when compiling an alternative. */
325 dummy_failure_jump,
326
327 /* Push a dummy failure point and continue. Used at the end of
328 alternatives. */
329 push_dummy_failure,
330
331 /* Followed by two-byte relative address and two-byte number n.
332 After matching N times, jump to the address upon failure. */
333 succeed_n,
334
335 /* Followed by two-byte relative address, and two-byte number n.
336 Jump to the address N times, then fail. */
337 jump_n,
338
339 /* Set the following two-byte relative address to the
340 subsequent two-byte number. The address *includes* the two
341 bytes of number. */
342 set_number_at,
343
344 wordchar, /* Matches any word-constituent character. */
345 notwordchar, /* Matches any char that is not a word-constituent. */
346
347 wordbeg, /* Succeeds if at word beginning. */
348 wordend, /* Succeeds if at word end. */
349
350 wordbound, /* Succeeds if at a word boundary. */
351 notwordbound /* Succeeds if not at a word boundary. */
352
353 #ifdef emacs
354 ,before_dot, /* Succeeds if before point. */
355 at_dot, /* Succeeds if at point. */
356 after_dot, /* Succeeds if after point. */
357
358 /* Matches any character whose syntax is specified. Followed by
359 a byte which contains a syntax code, e.g., Sword. */
360 syntaxspec,
361
362 /* Matches any character whose syntax is not that specified. */
363 notsyntaxspec
364 #endif /* emacs */
365 } re_opcode_t;
366
367 /* Common operations on the compiled pattern. */
368
369 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
370
371 #define STORE_NUMBER(destination, number) \
372 do { \
373 (destination)[0] = (number) & 0377; \
374 (destination)[1] = (number) >> 8; \
375 } while (0)
376
377 /* Same as STORE_NUMBER, except increment DESTINATION to
378 the byte after where the number is stored. Therefore, DESTINATION
379 must be an lvalue. */
380
381 #define STORE_NUMBER_AND_INCR(destination, number) \
382 do { \
383 STORE_NUMBER (destination, number); \
384 (destination) += 2; \
385 } while (0)
386
387 /* Put into DESTINATION a number stored in two contiguous bytes starting
388 at SOURCE. */
389
390 #define EXTRACT_NUMBER(destination, source) \
391 do { \
392 (destination) = *(source) & 0377; \
393 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
394 } while (0)
395
396 #ifdef DEBUG
397 static void
extract_number(dest,source)398 extract_number (dest, source)
399 int *dest;
400 unsigned char *source;
401 {
402 int temp = SIGN_EXTEND_CHAR (*(source + 1));
403 *dest = *source & 0377;
404 *dest += temp << 8;
405 }
406
407 #ifndef EXTRACT_MACROS /* To debug the macros. */
408 #undef EXTRACT_NUMBER
409 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
410 #endif /* not EXTRACT_MACROS */
411
412 #endif /* DEBUG */
413
414 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
415 SOURCE must be an lvalue. */
416
417 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
418 do { \
419 EXTRACT_NUMBER (destination, source); \
420 (source) += 2; \
421 } while (0)
422
423 #ifdef DEBUG
424 static void
extract_number_and_incr(destination,source)425 extract_number_and_incr (destination, source)
426 int *destination;
427 unsigned char **source;
428 {
429 extract_number (destination, *source);
430 *source += 2;
431 }
432
433 #ifndef EXTRACT_MACROS
434 #undef EXTRACT_NUMBER_AND_INCR
435 #define EXTRACT_NUMBER_AND_INCR(dest, src) \
436 extract_number_and_incr (&dest, &src)
437 #endif /* not EXTRACT_MACROS */
438
439 #endif /* DEBUG */
440
441 /* If DEBUG is defined, Regex prints many voluminous messages about what
442 it is doing (if the variable `debug' is nonzero). If linked with the
443 main program in `iregex.c', you can enter patterns and strings
444 interactively. And if linked with the main program in `main.c' and
445 the other test files, you can run the already-written tests. */
446
447 #ifdef DEBUG
448
449 /* We use standard I/O for debugging. */
450 #include <stdio.h>
451
452 /* It is useful to test things that ``must'' be true when debugging. */
453 #include <assert.h>
454
455 static int debug = 0;
456
457 #define DEBUG_STATEMENT(e) e
458 #define DEBUG_PRINT1(x) if (debug) printf (x)
459 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
460 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
461 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
462 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
463 if (debug) print_partial_compiled_pattern (s, e)
464 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
465 if (debug) print_double_string (w, s1, sz1, s2, sz2)
466
467
468 extern void printchar ();
469
470 /* Print the fastmap in human-readable form. */
471
472 void
print_fastmap(fastmap)473 print_fastmap (fastmap)
474 char *fastmap;
475 {
476 unsigned was_a_range = 0;
477 unsigned i = 0;
478
479 while (i < (1 << BYTEWIDTH))
480 {
481 if (fastmap[i++])
482 {
483 was_a_range = 0;
484 printchar (i - 1);
485 while (i < (1 << BYTEWIDTH) && fastmap[i])
486 {
487 was_a_range = 1;
488 i++;
489 }
490 if (was_a_range)
491 {
492 printf ("-");
493 printchar (i - 1);
494 }
495 }
496 }
497 putchar ('\n');
498 }
499
500
501 /* Print a compiled pattern string in human-readable form, starting at
502 the START pointer into it and ending just before the pointer END. */
503
504 void
print_partial_compiled_pattern(start,end)505 print_partial_compiled_pattern (start, end)
506 unsigned char *start;
507 unsigned char *end;
508 {
509 int mcnt, mcnt2;
510 unsigned char *p = start;
511 unsigned char *pend = end;
512
513 if (start == NULL)
514 {
515 printf ("(null)\n");
516 return;
517 }
518
519 /* Loop over pattern commands. */
520 while (p < pend)
521 {
522 switch ((re_opcode_t) *p++)
523 {
524 case no_op:
525 printf ("/no_op");
526 break;
527
528 case exactn:
529 mcnt = *p++;
530 printf ("/exactn/%d", mcnt);
531 do
532 {
533 putchar ('/');
534 printchar (*p++);
535 }
536 while (--mcnt);
537 break;
538
539 case start_memory:
540 mcnt = *p++;
541 printf ("/start_memory/%d/%d", mcnt, *p++);
542 break;
543
544 case stop_memory:
545 mcnt = *p++;
546 printf ("/stop_memory/%d/%d", mcnt, *p++);
547 break;
548
549 case duplicate:
550 printf ("/duplicate/%d", *p++);
551 break;
552
553 case anychar:
554 printf ("/anychar");
555 break;
556
557 case charset:
558 case charset_not:
559 {
560 register int c;
561
562 printf ("/charset%s",
563 (re_opcode_t) *(p - 1) == charset_not ? "_not" : "");
564
565 assert (p + *p < pend);
566
567 for (c = 0; c < *p; c++)
568 {
569 unsigned bit;
570 unsigned char map_byte = p[1 + c];
571
572 putchar ('/');
573
574 for (bit = 0; bit < BYTEWIDTH; bit++)
575 if (map_byte & (1 << bit))
576 printchar (c * BYTEWIDTH + bit);
577 }
578 p += 1 + *p;
579 break;
580 }
581
582 case begline:
583 printf ("/begline");
584 break;
585
586 case endline:
587 printf ("/endline");
588 break;
589
590 case on_failure_jump:
591 extract_number_and_incr (&mcnt, &p);
592 printf ("/on_failure_jump/0/%d", mcnt);
593 break;
594
595 case on_failure_keep_string_jump:
596 extract_number_and_incr (&mcnt, &p);
597 printf ("/on_failure_keep_string_jump/0/%d", mcnt);
598 break;
599
600 case dummy_failure_jump:
601 extract_number_and_incr (&mcnt, &p);
602 printf ("/dummy_failure_jump/0/%d", mcnt);
603 break;
604
605 case push_dummy_failure:
606 printf ("/push_dummy_failure");
607 break;
608
609 case maybe_pop_jump:
610 extract_number_and_incr (&mcnt, &p);
611 printf ("/maybe_pop_jump/0/%d", mcnt);
612 break;
613
614 case pop_failure_jump:
615 extract_number_and_incr (&mcnt, &p);
616 printf ("/pop_failure_jump/0/%d", mcnt);
617 break;
618
619 case jump_past_alt:
620 extract_number_and_incr (&mcnt, &p);
621 printf ("/jump_past_alt/0/%d", mcnt);
622 break;
623
624 case jump:
625 extract_number_and_incr (&mcnt, &p);
626 printf ("/jump/0/%d", mcnt);
627 break;
628
629 case succeed_n:
630 extract_number_and_incr (&mcnt, &p);
631 extract_number_and_incr (&mcnt2, &p);
632 printf ("/succeed_n/0/%d/0/%d", mcnt, mcnt2);
633 break;
634
635 case jump_n:
636 extract_number_and_incr (&mcnt, &p);
637 extract_number_and_incr (&mcnt2, &p);
638 printf ("/jump_n/0/%d/0/%d", mcnt, mcnt2);
639 break;
640
641 case set_number_at:
642 extract_number_and_incr (&mcnt, &p);
643 extract_number_and_incr (&mcnt2, &p);
644 printf ("/set_number_at/0/%d/0/%d", mcnt, mcnt2);
645 break;
646
647 case wordbound:
648 printf ("/wordbound");
649 break;
650
651 case notwordbound:
652 printf ("/notwordbound");
653 break;
654
655 case wordbeg:
656 printf ("/wordbeg");
657 break;
658
659 case wordend:
660 printf ("/wordend");
661
662 #ifdef emacs
663 case before_dot:
664 printf ("/before_dot");
665 break;
666
667 case at_dot:
668 printf ("/at_dot");
669 break;
670
671 case after_dot:
672 printf ("/after_dot");
673 break;
674
675 case syntaxspec:
676 printf ("/syntaxspec");
677 mcnt = *p++;
678 printf ("/%d", mcnt);
679 break;
680
681 case notsyntaxspec:
682 printf ("/notsyntaxspec");
683 mcnt = *p++;
684 printf ("/%d", mcnt);
685 break;
686 #endif /* emacs */
687
688 case wordchar:
689 printf ("/wordchar");
690 break;
691
692 case notwordchar:
693 printf ("/notwordchar");
694 break;
695
696 case begbuf:
697 printf ("/begbuf");
698 break;
699
700 case endbuf:
701 printf ("/endbuf");
702 break;
703
704 default:
705 printf ("?%d", *(p-1));
706 }
707 }
708 printf ("/\n");
709 }
710
711
712 void
print_compiled_pattern(bufp)713 print_compiled_pattern (bufp)
714 struct re_pattern_buffer *bufp;
715 {
716 unsigned char *buffer = bufp->buffer;
717
718 print_partial_compiled_pattern (buffer, buffer + bufp->used);
719 printf ("%d bytes used/%d bytes allocated.\n", bufp->used, bufp->allocated);
720
721 if (bufp->fastmap_accurate && bufp->fastmap)
722 {
723 printf ("fastmap: ");
724 print_fastmap (bufp->fastmap);
725 }
726
727 printf ("re_nsub: %d\t", bufp->re_nsub);
728 printf ("regs_alloc: %d\t", bufp->regs_allocated);
729 printf ("can_be_null: %d\t", bufp->can_be_null);
730 printf ("newline_anchor: %d\n", bufp->newline_anchor);
731 printf ("no_sub: %d\t", bufp->no_sub);
732 printf ("not_bol: %d\t", bufp->not_bol);
733 printf ("not_eol: %d\t", bufp->not_eol);
734 printf ("syntax: %d\n", bufp->syntax);
735 /* Perhaps we should print the translate table? */
736 }
737
738
739 void
print_double_string(where,string1,size1,string2,size2)740 print_double_string (where, string1, size1, string2, size2)
741 char *where;
742 char *string1;
743 char *string2;
744 int size1;
745 int size2;
746 {
747 unsigned this_char;
748
749 if (where == NULL)
750 printf ("(null)");
751 else
752 {
753 if (FIRST_STRING_P (where))
754 {
755 for (this_char = where - string1; this_char < size1; this_char++)
756 printchar (string1[this_char]);
757
758 where = string2;
759 }
760
761 for (this_char = where - string2; this_char < size2; this_char++)
762 printchar (string2[this_char]);
763 }
764 }
765
766 #else /* not DEBUG */
767
768 #undef assert
769 #define assert(e)
770
771 #define DEBUG_STATEMENT(e)
772 #define DEBUG_PRINT1(x)
773 #define DEBUG_PRINT2(x1, x2)
774 #define DEBUG_PRINT3(x1, x2, x3)
775 #define DEBUG_PRINT4(x1, x2, x3, x4)
776 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
777 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
778
779 #endif /* not DEBUG */
780
781 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
782 also be assigned to arbitrarily: each pattern buffer stores its own
783 syntax, so it can be changed between regex compilations. */
784 reg_syntax_t re_syntax_options = RE_SYNTAX_EMACS;
785
786
787 /* Specify the precise syntax of regexps for compilation. This provides
788 for compatibility for various utilities which historically have
789 different, incompatible syntaxes.
790
791 The argument SYNTAX is a bit mask comprised of the various bits
792 defined in regex.h. We return the old syntax. */
793
794 reg_syntax_t
re_set_syntax(syntax)795 re_set_syntax (syntax)
796 reg_syntax_t syntax;
797 {
798 reg_syntax_t ret = re_syntax_options;
799
800 re_syntax_options = syntax;
801 return ret;
802 }
803
804 /* This table gives an error message for each of the error codes listed
805 in regex.h. Obviously the order here has to be same as there. */
806
807 static char *re_error_msg[] =
808 { NULL, /* REG_NOERROR */
809 "No match", /* REG_NOMATCH */
810 "Invalid regular expression", /* REG_BADPAT */
811 "Invalid collation character", /* REG_ECOLLATE */
812 "Invalid character class name", /* REG_ECTYPE */
813 "Trailing backslash", /* REG_EESCAPE */
814 "Invalid back reference", /* REG_ESUBREG */
815 "Unmatched [ or [^", /* REG_EBRACK */
816 "Unmatched ( or \\(", /* REG_EPAREN */
817 "Unmatched \\{", /* REG_EBRACE */
818 "Invalid content of \\{\\}", /* REG_BADBR */
819 "Invalid range end", /* REG_ERANGE */
820 "Memory exhausted", /* REG_ESPACE */
821 "Invalid preceding regular expression", /* REG_BADRPT */
822 "Premature end of regular expression", /* REG_EEND */
823 "Regular expression too big", /* REG_ESIZE */
824 "Unmatched ) or \\)", /* REG_ERPAREN */
825 };
826
827 /* Subroutine declarations and macros for regex_compile. */
828
829 static void store_op1 (), store_op2 ();
830 static void insert_op1 (), insert_op2 ();
831 static boolean at_begline_loc_p (), at_endline_loc_p ();
832 static boolean group_in_compile_stack ();
833 static reg_errcode_t compile_range ();
834
835 /* Fetch the next character in the uncompiled pattern---translating it
836 if necessary. Also cast from a signed character in the constant
837 string passed to us by the user to an unsigned char that we can use
838 as an array index (in, e.g., `translate'). */
839 #define PATFETCH(c) \
840 do {if (p == pend) return REG_EEND; \
841 c = (unsigned char) *p++; \
842 if (translate) c = translate[c]; \
843 } while (0)
844
845 /* Fetch the next character in the uncompiled pattern, with no
846 translation. */
847 #define PATFETCH_RAW(c) \
848 do {if (p == pend) return REG_EEND; \
849 c = (unsigned char) *p++; \
850 } while (0)
851
852 /* Go backwards one character in the pattern. */
853 #define PATUNFETCH p--
854
855
856 /* If `translate' is non-null, return translate[D], else just D. We
857 cast the subscript to translate because some data is declared as
858 `char *', to avoid warnings when a string constant is passed. But
859 when we use a character as a subscript we must make it unsigned. */
860 #define TRANSLATE(d) (translate ? translate[(unsigned char) (d)] : (d))
861
862
863 /* Macros for outputting the compiled pattern into `buffer'. */
864
865 /* If the buffer isn't allocated when it comes in, use this. */
866 #define INIT_BUF_SIZE 32
867
868 /* Make sure we have at least N more bytes of space in buffer. */
869 #define GET_BUFFER_SPACE(n) \
870 while (b - bufp->buffer + (n) > bufp->allocated) \
871 EXTEND_BUFFER ()
872
873 /* Make sure we have one more byte of buffer space and then add C to it. */
874 #define BUF_PUSH(c) \
875 do { \
876 GET_BUFFER_SPACE (1); \
877 *b++ = (unsigned char) (c); \
878 } while (0)
879
880
881 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
882 #define BUF_PUSH_2(c1, c2) \
883 do { \
884 GET_BUFFER_SPACE (2); \
885 *b++ = (unsigned char) (c1); \
886 *b++ = (unsigned char) (c2); \
887 } while (0)
888
889
890 /* As with BUF_PUSH_2, except for three bytes. */
891 #define BUF_PUSH_3(c1, c2, c3) \
892 do { \
893 GET_BUFFER_SPACE (3); \
894 *b++ = (unsigned char) (c1); \
895 *b++ = (unsigned char) (c2); \
896 *b++ = (unsigned char) (c3); \
897 } while (0)
898
899
900 /* Store a jump with opcode OP at LOC to location TO. We store a
901 relative address offset by the three bytes the jump itself occupies. */
902 #define STORE_JUMP(op, loc, to) \
903 store_op1 (op, loc, (to) - (loc) - 3)
904
905 /* Likewise, for a two-argument jump. */
906 #define STORE_JUMP2(op, loc, to, arg) \
907 store_op2 (op, loc, (to) - (loc) - 3, arg)
908
909 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
910 #define INSERT_JUMP(op, loc, to) \
911 insert_op1 (op, loc, (to) - (loc) - 3, b)
912
913 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
914 #define INSERT_JUMP2(op, loc, to, arg) \
915 insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
916
917
918 /* This is not an arbitrary limit: the arguments which represent offsets
919 into the pattern are two bytes long. So if 2^16 bytes turns out to
920 be too small, many things would have to change. */
921 #define MAX_BUF_SIZE (1L << 16)
922
923
924 /* Extend the buffer by twice its current size via realloc and
925 reset the pointers that pointed into the old block to point to the
926 correct places in the new one. If extending the buffer results in it
927 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
928 #define EXTEND_BUFFER() \
929 do { \
930 unsigned char *old_buffer = bufp->buffer; \
931 if (bufp->allocated == MAX_BUF_SIZE) \
932 return REG_ESIZE; \
933 bufp->allocated <<= 1; \
934 if (bufp->allocated > MAX_BUF_SIZE) \
935 bufp->allocated = MAX_BUF_SIZE; \
936 bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\
937 if (bufp->buffer == NULL) \
938 return REG_ESPACE; \
939 /* If the buffer moved, move all the pointers into it. */ \
940 if (old_buffer != bufp->buffer) \
941 { \
942 b = (b - old_buffer) + bufp->buffer; \
943 begalt = (begalt - old_buffer) + bufp->buffer; \
944 if (fixup_alt_jump) \
945 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
946 if (laststart) \
947 laststart = (laststart - old_buffer) + bufp->buffer; \
948 if (pending_exact) \
949 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
950 } \
951 } while (0)
952
953
954 /* Since we have one byte reserved for the register number argument to
955 {start,stop}_memory, the maximum number of groups we can report
956 things about is what fits in that byte. */
957 #define MAX_REGNUM 255
958
959 /* But patterns can have more than `MAX_REGNUM' registers. We just
960 ignore the excess. */
961 typedef unsigned regnum_t;
962
963
964 /* Macros for the compile stack. */
965
966 /* Since offsets can go either forwards or backwards, this type needs to
967 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
968 typedef int pattern_offset_t;
969
970 typedef struct
971 {
972 pattern_offset_t begalt_offset;
973 pattern_offset_t fixup_alt_jump;
974 pattern_offset_t inner_group_offset;
975 pattern_offset_t laststart_offset;
976 regnum_t regnum;
977 } compile_stack_elt_t;
978
979
980 typedef struct
981 {
982 compile_stack_elt_t *stack;
983 unsigned size;
984 unsigned avail; /* Offset of next open position. */
985 } compile_stack_type;
986
987
988 #define INIT_COMPILE_STACK_SIZE 32
989
990 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
991 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
992
993 /* The next available element. */
994 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
995
996
997 /* Set the bit for character C in a list. */
998 #define SET_LIST_BIT(c) \
999 (b[((unsigned char) (c)) / BYTEWIDTH] \
1000 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1001
1002
1003 /* Get the next unsigned number in the uncompiled pattern. */
1004 #define GET_UNSIGNED_NUMBER(num) \
1005 { if (p != pend) \
1006 { \
1007 PATFETCH (c); \
1008 while (ISDIGIT (c)) \
1009 { \
1010 if (num < 0) \
1011 num = 0; \
1012 num = num * 10 + c - '0'; \
1013 if (p == pend) \
1014 break; \
1015 PATFETCH (c); \
1016 } \
1017 } \
1018 }
1019
1020 #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1021
1022 #define IS_CHAR_CLASS(string) \
1023 (STREQ (string, "alpha") || STREQ (string, "upper") \
1024 || STREQ (string, "lower") || STREQ (string, "digit") \
1025 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1026 || STREQ (string, "space") || STREQ (string, "print") \
1027 || STREQ (string, "punct") || STREQ (string, "graph") \
1028 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1029
1030 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1031 Returns one of error codes defined in `regex.h', or zero for success.
1032
1033 Assumes the `allocated' (and perhaps `buffer') and `translate'
1034 fields are set in BUFP on entry.
1035
1036 If it succeeds, results are put in BUFP (if it returns an error, the
1037 contents of BUFP are undefined):
1038 `buffer' is the compiled pattern;
1039 `syntax' is set to SYNTAX;
1040 `used' is set to the length of the compiled pattern;
1041 `fastmap_accurate' is zero;
1042 `re_nsub' is the number of subexpressions in PATTERN;
1043 `not_bol' and `not_eol' are zero;
1044
1045 The `fastmap' and `newline_anchor' fields are neither
1046 examined nor set. */
1047
1048 static reg_errcode_t
regex_compile(pattern,size,syntax,bufp)1049 regex_compile (pattern, size, syntax, bufp)
1050 char *pattern;
1051 int size;
1052 reg_syntax_t syntax;
1053 struct re_pattern_buffer *bufp;
1054 {
1055 /* We fetch characters from PATTERN here. Even though PATTERN is
1056 `char *' (i.e., signed), we declare these variables as unsigned, so
1057 they can be reliably used as array indices. */
1058 register unsigned char c, c1;
1059
1060 /* A random tempory spot in PATTERN. */
1061 char *p1;
1062
1063 /* Points to the end of the buffer, where we should append. */
1064 register unsigned char *b;
1065
1066 /* Keeps track of unclosed groups. */
1067 compile_stack_type compile_stack;
1068
1069 /* Points to the current (ending) position in the pattern. */
1070 char *p = pattern;
1071 char *pend = pattern + size;
1072
1073 /* How to translate the characters in the pattern. */
1074 char *translate = bufp->translate;
1075
1076 /* Address of the count-byte of the most recently inserted `exactn'
1077 command. This makes it possible to tell if a new exact-match
1078 character can be added to that command or if the character requires
1079 a new `exactn' command. */
1080 unsigned char *pending_exact = 0;
1081
1082 /* Address of start of the most recently finished expression.
1083 This tells, e.g., postfix * where to find the start of its
1084 operand. Reset at the beginning of groups and alternatives. */
1085 unsigned char *laststart = 0;
1086
1087 /* Address of beginning of regexp, or inside of last group. */
1088 unsigned char *begalt;
1089
1090 /* Place in the uncompiled pattern (i.e., the {) to
1091 which to go back if the interval is invalid. */
1092 char *beg_interval;
1093
1094 /* Address of the place where a forward jump should go to the end of
1095 the containing expression. Each alternative of an `or' -- except the
1096 last -- ends with a forward jump of this sort. */
1097 unsigned char *fixup_alt_jump = 0;
1098
1099 /* Counts open-groups as they are encountered. Remembered for the
1100 matching close-group on the compile stack, so the same register
1101 number is put in the stop_memory as the start_memory. */
1102 regnum_t regnum = 0;
1103
1104 #ifdef DEBUG
1105 DEBUG_PRINT1 ("\nCompiling pattern: ");
1106 if (debug)
1107 {
1108 unsigned debug_count;
1109
1110 for (debug_count = 0; debug_count < size; debug_count++)
1111 printchar (pattern[debug_count]);
1112 putchar ('\n');
1113 }
1114 #endif /* DEBUG */
1115
1116 /* Initialize the compile stack. */
1117 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1118 if (compile_stack.stack == NULL)
1119 return REG_ESPACE;
1120
1121 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1122 compile_stack.avail = 0;
1123
1124 /* Initialize the pattern buffer. */
1125 bufp->syntax = syntax;
1126 bufp->fastmap_accurate = 0;
1127 bufp->not_bol = bufp->not_eol = 0;
1128
1129 /* Set `used' to zero, so that if we return an error, the pattern
1130 printer (for debugging) will think there's no pattern. We reset it
1131 at the end. */
1132 bufp->used = 0;
1133
1134 /* Always count groups, whether or not bufp->no_sub is set. */
1135 bufp->re_nsub = 0;
1136
1137 #if !defined (emacs) && !defined (SYNTAX_TABLE)
1138 /* Initialize the syntax table. */
1139 init_syntax_once ();
1140 #endif
1141
1142 if (bufp->allocated == 0)
1143 {
1144 if (bufp->buffer)
1145 { /* If zero allocated, but buffer is non-null, try to realloc
1146 enough space. This loses if buffer's address is bogus, but
1147 that is the user's responsibility. */
1148 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1149 }
1150 else
1151 { /* Caller did not allocate a buffer. Do it for them. */
1152 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1153 }
1154 if (!bufp->buffer) return REG_ESPACE;
1155
1156 bufp->allocated = INIT_BUF_SIZE;
1157 }
1158
1159 begalt = b = bufp->buffer;
1160
1161 /* Loop through the uncompiled pattern until we're at the end. */
1162 while (p != pend)
1163 {
1164 PATFETCH (c);
1165
1166 switch (c)
1167 {
1168 case '^':
1169 {
1170 if ( /* If at start of pattern, it's an operator. */
1171 p == pattern + 1
1172 /* If context independent, it's an operator. */
1173 || syntax & RE_CONTEXT_INDEP_ANCHORS
1174 /* Otherwise, depends on what's come before. */
1175 || at_begline_loc_p (pattern, p, syntax))
1176 BUF_PUSH (begline);
1177 else
1178 goto normal_char;
1179 }
1180 break;
1181
1182
1183 case '$':
1184 {
1185 if ( /* If at end of pattern, it's an operator. */
1186 p == pend
1187 /* If context independent, it's an operator. */
1188 || syntax & RE_CONTEXT_INDEP_ANCHORS
1189 /* Otherwise, depends on what's next. */
1190 || at_endline_loc_p (p, pend, syntax))
1191 BUF_PUSH (endline);
1192 else
1193 goto normal_char;
1194 }
1195 break;
1196
1197
1198 case '+':
1199 case '?':
1200 if ((syntax & RE_BK_PLUS_QM)
1201 || (syntax & RE_LIMITED_OPS))
1202 goto normal_char;
1203 handle_plus:
1204 case '*':
1205 /* If there is no previous pattern... */
1206 if (!laststart)
1207 {
1208 if (syntax & RE_CONTEXT_INVALID_OPS)
1209 return REG_BADRPT;
1210 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1211 goto normal_char;
1212 }
1213
1214 {
1215 /* Are we optimizing this jump? */
1216 boolean keep_string_p = false;
1217
1218 /* 1 means zero (many) matches is allowed. */
1219 char zero_times_ok = 0, many_times_ok = 0;
1220
1221 /* If there is a sequence of repetition chars, collapse it
1222 down to just one (the right one). We can't combine
1223 interval operators with these because of, e.g., `a{2}*',
1224 which should only match an even number of `a's. */
1225
1226 for (;;)
1227 {
1228 zero_times_ok |= c != '+';
1229 many_times_ok |= c != '?';
1230
1231 if (p == pend)
1232 break;
1233
1234 PATFETCH (c);
1235
1236 if (c == '*'
1237 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
1238 ;
1239
1240 else if (syntax & RE_BK_PLUS_QM && c == '\\')
1241 {
1242 if (p == pend) return REG_EESCAPE;
1243
1244 PATFETCH (c1);
1245 if (!(c1 == '+' || c1 == '?'))
1246 {
1247 PATUNFETCH;
1248 PATUNFETCH;
1249 break;
1250 }
1251
1252 c = c1;
1253 }
1254 else
1255 {
1256 PATUNFETCH;
1257 break;
1258 }
1259
1260 /* If we get here, we found another repeat character. */
1261 }
1262
1263 /* Star, etc. applied to an empty pattern is equivalent
1264 to an empty pattern. */
1265 if (!laststart)
1266 break;
1267
1268 /* Now we know whether or not zero matches is allowed
1269 and also whether or not two or more matches is allowed. */
1270 if (many_times_ok)
1271 { /* More than one repetition is allowed, so put in at the
1272 end a backward relative jump from `b' to before the next
1273 jump we're going to put in below (which jumps from
1274 laststart to after this jump).
1275
1276 But if we are at the `*' in the exact sequence `.*\n',
1277 insert an unconditional jump backwards to the .,
1278 instead of the beginning of the loop. This way we only
1279 push a failure point once, instead of every time
1280 through the loop. */
1281 assert (p - 1 > pattern);
1282
1283 /* Allocate the space for the jump. */
1284 GET_BUFFER_SPACE (3);
1285
1286 /* We know we are not at the first character of the pattern,
1287 because laststart was nonzero. And we've already
1288 incremented `p', by the way, to be the character after
1289 the `*'. Do we have to do something analogous here
1290 for null bytes, because of RE_DOT_NOT_NULL? */
1291 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
1292 && zero_times_ok
1293 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
1294 && !(syntax & RE_DOT_NEWLINE))
1295 { /* We have .*\n. */
1296 STORE_JUMP (jump, b, laststart);
1297 keep_string_p = true;
1298 }
1299 else
1300 /* Anything else. */
1301 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
1302
1303 /* We've added more stuff to the buffer. */
1304 b += 3;
1305 }
1306
1307 /* On failure, jump from laststart to b + 3, which will be the
1308 end of the buffer after this jump is inserted. */
1309 GET_BUFFER_SPACE (3);
1310 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
1311 : on_failure_jump,
1312 laststart, b + 3);
1313 pending_exact = 0;
1314 b += 3;
1315
1316 if (!zero_times_ok)
1317 {
1318 /* At least one repetition is required, so insert a
1319 `dummy_failure_jump' before the initial
1320 `on_failure_jump' instruction of the loop. This
1321 effects a skip over that instruction the first time
1322 we hit that loop. */
1323 GET_BUFFER_SPACE (3);
1324 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
1325 b += 3;
1326 }
1327 }
1328 break;
1329
1330
1331 case '.':
1332 laststart = b;
1333 BUF_PUSH (anychar);
1334 break;
1335
1336
1337 case '[':
1338 {
1339 boolean had_char_class = false;
1340
1341 if (p == pend) return REG_EBRACK;
1342
1343 /* Ensure that we have enough space to push a charset: the
1344 opcode, the length count, and the bitset; 34 bytes in all. */
1345 GET_BUFFER_SPACE (34);
1346
1347 laststart = b;
1348
1349 /* We test `*p == '^' twice, instead of using an if
1350 statement, so we only need one BUF_PUSH. */
1351 BUF_PUSH (*p == '^' ? charset_not : charset);
1352 if (*p == '^')
1353 p++;
1354
1355 /* Remember the first position in the bracket expression. */
1356 p1 = p;
1357
1358 /* Push the number of bytes in the bitmap. */
1359 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
1360
1361 /* Clear the whole map. */
1362 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
1363
1364 /* charset_not matches newline according to a syntax bit. */
1365 if ((re_opcode_t) b[-2] == charset_not
1366 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
1367 SET_LIST_BIT ('\n');
1368
1369 /* Read in characters and ranges, setting map bits. */
1370 for (;;)
1371 {
1372 if (p == pend) return REG_EBRACK;
1373
1374 PATFETCH (c);
1375
1376 /* \ might escape characters inside [...] and [^...]. */
1377 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
1378 {
1379 if (p == pend) return REG_EESCAPE;
1380
1381 PATFETCH (c1);
1382 SET_LIST_BIT (c1);
1383 continue;
1384 }
1385
1386 /* Could be the end of the bracket expression. If it's
1387 not (i.e., when the bracket expression is `[]' so
1388 far), the ']' character bit gets set way below. */
1389 if (c == ']' && p != p1 + 1)
1390 break;
1391
1392 /* Look ahead to see if it's a range when the last thing
1393 was a character class. */
1394 if (had_char_class && c == '-' && *p != ']')
1395 return REG_ERANGE;
1396
1397 /* Look ahead to see if it's a range when the last thing
1398 was a character: if this is a hyphen not at the
1399 beginning or the end of a list, then it's the range
1400 operator. */
1401 if (c == '-'
1402 && !(p - 2 >= pattern && p[-2] == '[')
1403 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
1404 && *p != ']')
1405 {
1406 reg_errcode_t ret
1407 = compile_range (&p, pend, translate, syntax, b);
1408 if (ret != REG_NOERROR) return ret;
1409 }
1410
1411 else if (p[0] == '-' && p[1] != ']')
1412 { /* This handles ranges made up of characters only. */
1413 reg_errcode_t ret;
1414
1415 /* Move past the `-'. */
1416 PATFETCH (c1);
1417
1418 ret = compile_range (&p, pend, translate, syntax, b);
1419 if (ret != REG_NOERROR) return ret;
1420 }
1421
1422 /* See if we're at the beginning of a possible character
1423 class. */
1424
1425 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
1426 { /* Leave room for the null. */
1427 char str[CHAR_CLASS_MAX_LENGTH + 1];
1428
1429 PATFETCH (c);
1430 c1 = 0;
1431
1432 /* If pattern is `[[:'. */
1433 if (p == pend) return REG_EBRACK;
1434
1435 for (;;)
1436 {
1437 PATFETCH (c);
1438 if (c == ':' || c == ']' || p == pend
1439 || c1 == CHAR_CLASS_MAX_LENGTH)
1440 break;
1441 str[c1++] = c;
1442 }
1443 str[c1] = '\0';
1444
1445 /* If isn't a word bracketed by `[:' and:`]':
1446 undo the ending character, the letters, and leave
1447 the leading `:' and `[' (but set bits for them). */
1448 if (c == ':' && *p == ']')
1449 {
1450 int ch;
1451 boolean is_alnum = STREQ (str, "alnum");
1452 boolean is_alpha = STREQ (str, "alpha");
1453 boolean is_blank = STREQ (str, "blank");
1454 boolean is_cntrl = STREQ (str, "cntrl");
1455 boolean is_digit = STREQ (str, "digit");
1456 boolean is_graph = STREQ (str, "graph");
1457 boolean is_lower = STREQ (str, "lower");
1458 boolean is_print = STREQ (str, "print");
1459 boolean is_punct = STREQ (str, "punct");
1460 boolean is_space = STREQ (str, "space");
1461 boolean is_upper = STREQ (str, "upper");
1462 boolean is_xdigit = STREQ (str, "xdigit");
1463
1464 if (!IS_CHAR_CLASS (str)) return REG_ECTYPE;
1465
1466 /* Throw away the ] at the end of the character
1467 class. */
1468 PATFETCH (c);
1469
1470 if (p == pend) return REG_EBRACK;
1471
1472 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
1473 {
1474 if ( (is_alnum && ISALNUM (ch))
1475 || (is_alpha && ISALPHA (ch))
1476 || (is_blank && ISBLANK (ch))
1477 || (is_cntrl && ISCNTRL (ch))
1478 || (is_digit && ISDIGIT (ch))
1479 || (is_graph && ISGRAPH (ch))
1480 || (is_lower && ISLOWER (ch))
1481 || (is_print && ISPRINT (ch))
1482 || (is_punct && ISPUNCT (ch))
1483 || (is_space && ISSPACE (ch))
1484 || (is_upper && ISUPPER (ch))
1485 || (is_xdigit && ISXDIGIT (ch)))
1486 SET_LIST_BIT (ch);
1487 }
1488 had_char_class = true;
1489 }
1490 else
1491 {
1492 c1++;
1493 while (c1--)
1494 PATUNFETCH;
1495 SET_LIST_BIT ('[');
1496 SET_LIST_BIT (':');
1497 had_char_class = false;
1498 }
1499 }
1500 else
1501 {
1502 had_char_class = false;
1503 SET_LIST_BIT (c);
1504 }
1505 }
1506
1507 /* Discard any (non)matching list bytes that are all 0 at the
1508 end of the map. Decrease the map-length byte too. */
1509 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
1510 b[-1]--;
1511 b += b[-1];
1512 }
1513 break;
1514
1515
1516 case '(':
1517 if (syntax & RE_NO_BK_PARENS)
1518 goto handle_open;
1519 else
1520 goto normal_char;
1521
1522
1523 case ')':
1524 if (syntax & RE_NO_BK_PARENS)
1525 goto handle_close;
1526 else
1527 goto normal_char;
1528
1529
1530 case '\n':
1531 if (syntax & RE_NEWLINE_ALT)
1532 goto handle_alt;
1533 else
1534 goto normal_char;
1535
1536
1537 case '|':
1538 if (syntax & RE_NO_BK_VBAR)
1539 goto handle_alt;
1540 else
1541 goto normal_char;
1542
1543
1544 case '{':
1545 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
1546 goto handle_interval;
1547 else
1548 goto normal_char;
1549
1550
1551 case '\\':
1552 if (p == pend) return REG_EESCAPE;
1553
1554 /* Do not translate the character after the \, so that we can
1555 distinguish, e.g., \B from \b, even if we normally would
1556 translate, e.g., B to b. */
1557 PATFETCH_RAW (c);
1558
1559 switch (c)
1560 {
1561 case '(':
1562 if (syntax & RE_NO_BK_PARENS)
1563 goto normal_backslash;
1564
1565 handle_open:
1566 bufp->re_nsub++;
1567 regnum++;
1568
1569 if (COMPILE_STACK_FULL)
1570 {
1571 RETALLOC (compile_stack.stack, compile_stack.size << 1,
1572 compile_stack_elt_t);
1573 if (compile_stack.stack == NULL) return REG_ESPACE;
1574
1575 compile_stack.size <<= 1;
1576 }
1577
1578 /* These are the values to restore when we hit end of this
1579 group. They are all relative offsets, so that if the
1580 whole pattern moves because of realloc, they will still
1581 be valid. */
1582 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
1583 COMPILE_STACK_TOP.fixup_alt_jump
1584 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
1585 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
1586 COMPILE_STACK_TOP.regnum = regnum;
1587
1588 /* We will eventually replace the 0 with the number of
1589 groups inner to this one. But do not push a
1590 start_memory for groups beyond the last one we can
1591 represent in the compiled pattern. */
1592 if (regnum <= MAX_REGNUM)
1593 {
1594 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
1595 BUF_PUSH_3 (start_memory, regnum, 0);
1596 }
1597
1598 compile_stack.avail++;
1599
1600 fixup_alt_jump = 0;
1601 laststart = 0;
1602 begalt = b;
1603 /* If we've reached MAX_REGNUM groups, then this open
1604 won't actually generate any code, so we'll have to
1605 clear pending_exact explicitly. */
1606 pending_exact = 0;
1607 break;
1608
1609
1610 case ')':
1611 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
1612
1613 if (COMPILE_STACK_EMPTY)
1614 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
1615 goto normal_backslash;
1616 else
1617 return REG_ERPAREN;
1618
1619 handle_close:
1620 if (fixup_alt_jump)
1621 { /* Push a dummy failure point at the end of the
1622 alternative for a possible future
1623 `pop_failure_jump' to pop. See comments at
1624 `push_dummy_failure' in `re_match_2'. */
1625 BUF_PUSH (push_dummy_failure);
1626
1627 /* We allocated space for this jump when we assigned
1628 to `fixup_alt_jump', in the `handle_alt' case below. */
1629 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
1630 }
1631
1632 /* See similar code for backslashed left paren above. */
1633 if (COMPILE_STACK_EMPTY)
1634 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
1635 goto normal_char;
1636 else
1637 return REG_ERPAREN;
1638
1639 /* Since we just checked for an empty stack above, this
1640 ``can't happen''. */
1641 assert (compile_stack.avail != 0);
1642 {
1643 /* We don't just want to restore into `regnum', because
1644 later groups should continue to be numbered higher,
1645 as in `(ab)c(de)' -- the second group is #2. */
1646 regnum_t this_group_regnum;
1647
1648 compile_stack.avail--;
1649 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
1650 fixup_alt_jump
1651 = COMPILE_STACK_TOP.fixup_alt_jump
1652 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
1653 : 0;
1654 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
1655 this_group_regnum = COMPILE_STACK_TOP.regnum;
1656 /* If we've reached MAX_REGNUM groups, then this open
1657 won't actually generate any code, so we'll have to
1658 clear pending_exact explicitly. */
1659 pending_exact = 0;
1660
1661 /* We're at the end of the group, so now we know how many
1662 groups were inside this one. */
1663 if (this_group_regnum <= MAX_REGNUM)
1664 {
1665 unsigned char *inner_group_loc
1666 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
1667
1668 *inner_group_loc = regnum - this_group_regnum;
1669 BUF_PUSH_3 (stop_memory, this_group_regnum,
1670 regnum - this_group_regnum);
1671 }
1672 }
1673 break;
1674
1675
1676 case '|': /* `\|'. */
1677 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
1678 goto normal_backslash;
1679 handle_alt:
1680 if (syntax & RE_LIMITED_OPS)
1681 goto normal_char;
1682
1683 /* Insert before the previous alternative a jump which
1684 jumps to this alternative if the former fails. */
1685 GET_BUFFER_SPACE (3);
1686 INSERT_JUMP (on_failure_jump, begalt, b + 6);
1687 pending_exact = 0;
1688 b += 3;
1689
1690 /* The alternative before this one has a jump after it
1691 which gets executed if it gets matched. Adjust that
1692 jump so it will jump to this alternative's analogous
1693 jump (put in below, which in turn will jump to the next
1694 (if any) alternative's such jump, etc.). The last such
1695 jump jumps to the correct final destination. A picture:
1696 _____ _____
1697 | | | |
1698 | v | v
1699 a | b | c
1700
1701 If we are at `b', then fixup_alt_jump right now points to a
1702 three-byte space after `a'. We'll put in the jump, set
1703 fixup_alt_jump to right after `b', and leave behind three
1704 bytes which we'll fill in when we get to after `c'. */
1705
1706 if (fixup_alt_jump)
1707 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
1708
1709 /* Mark and leave space for a jump after this alternative,
1710 to be filled in later either by next alternative or
1711 when know we're at the end of a series of alternatives. */
1712 fixup_alt_jump = b;
1713 GET_BUFFER_SPACE (3);
1714 b += 3;
1715
1716 laststart = 0;
1717 begalt = b;
1718 break;
1719
1720
1721 case '{':
1722 /* If \{ is a literal. */
1723 if (!(syntax & RE_INTERVALS)
1724 /* If we're at `\{' and it's not the open-interval
1725 operator. */
1726 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
1727 || (p - 2 == pattern && p == pend))
1728 goto normal_backslash;
1729
1730 handle_interval:
1731 {
1732 /* If got here, then the syntax allows intervals. */
1733
1734 /* At least (most) this many matches must be made. */
1735 int lower_bound = -1, upper_bound = -1;
1736
1737 beg_interval = p - 1;
1738
1739 if (p == pend)
1740 {
1741 if (syntax & RE_NO_BK_BRACES)
1742 goto unfetch_interval;
1743 else
1744 return REG_EBRACE;
1745 }
1746
1747 GET_UNSIGNED_NUMBER (lower_bound);
1748
1749 if (c == ',')
1750 {
1751 GET_UNSIGNED_NUMBER (upper_bound);
1752 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
1753 }
1754 else
1755 /* Interval such as `{1}' => match exactly once. */
1756 upper_bound = lower_bound;
1757
1758 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
1759 || lower_bound > upper_bound)
1760 {
1761 if (syntax & RE_NO_BK_BRACES)
1762 goto unfetch_interval;
1763 else
1764 return REG_BADBR;
1765 }
1766
1767 if (!(syntax & RE_NO_BK_BRACES))
1768 {
1769 if (c != '\\') return REG_EBRACE;
1770
1771 PATFETCH (c);
1772 }
1773
1774 if (c != '}')
1775 {
1776 if (syntax & RE_NO_BK_BRACES)
1777 goto unfetch_interval;
1778 else
1779 return REG_BADBR;
1780 }
1781
1782 /* We just parsed a valid interval. */
1783
1784 /* If it's invalid to have no preceding re. */
1785 if (!laststart)
1786 {
1787 if (syntax & RE_CONTEXT_INVALID_OPS)
1788 return REG_BADRPT;
1789 else if (syntax & RE_CONTEXT_INDEP_OPS)
1790 laststart = b;
1791 else
1792 goto unfetch_interval;
1793 }
1794
1795 /* If the upper bound is zero, don't want to succeed at
1796 all; jump from `laststart' to `b + 3', which will be
1797 the end of the buffer after we insert the jump. */
1798 if (upper_bound == 0)
1799 {
1800 GET_BUFFER_SPACE (3);
1801 INSERT_JUMP (jump, laststart, b + 3);
1802 b += 3;
1803 }
1804
1805 /* Otherwise, we have a nontrivial interval. When
1806 we're all done, the pattern will look like:
1807 set_number_at <jump count> <upper bound>
1808 set_number_at <succeed_n count> <lower bound>
1809 succeed_n <after jump addr> <succed_n count>
1810 <body of loop>
1811 jump_n <succeed_n addr> <jump count>
1812 (The upper bound and `jump_n' are omitted if
1813 `upper_bound' is 1, though.) */
1814 else
1815 { /* If the upper bound is > 1, we need to insert
1816 more at the end of the loop. */
1817 unsigned nbytes = 10 + (upper_bound > 1) * 10;
1818
1819 GET_BUFFER_SPACE (nbytes);
1820
1821 /* Initialize lower bound of the `succeed_n', even
1822 though it will be set during matching by its
1823 attendant `set_number_at' (inserted next),
1824 because `re_compile_fastmap' needs to know.
1825 Jump to the `jump_n' we might insert below. */
1826 INSERT_JUMP2 (succeed_n, laststart,
1827 b + 5 + (upper_bound > 1) * 5,
1828 lower_bound);
1829 b += 5;
1830
1831 /* Code to initialize the lower bound. Insert
1832 before the `succeed_n'. The `5' is the last two
1833 bytes of this `set_number_at', plus 3 bytes of
1834 the following `succeed_n'. */
1835 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
1836 b += 5;
1837
1838 if (upper_bound > 1)
1839 { /* More than one repetition is allowed, so
1840 append a backward jump to the `succeed_n'
1841 that starts this interval.
1842
1843 When we've reached this during matching,
1844 we'll have matched the interval once, so
1845 jump back only `upper_bound - 1' times. */
1846 STORE_JUMP2 (jump_n, b, laststart + 5,
1847 upper_bound - 1);
1848 b += 5;
1849
1850 /* The location we want to set is the second
1851 parameter of the `jump_n'; that is `b-2' as
1852 an absolute address. `laststart' will be
1853 the `set_number_at' we're about to insert;
1854 `laststart+3' the number to set, the source
1855 for the relative address. But we are
1856 inserting into the middle of the pattern --
1857 so everything is getting moved up by 5.
1858 Conclusion: (b - 2) - (laststart + 3) + 5,
1859 i.e., b - laststart.
1860
1861 We insert this at the beginning of the loop
1862 so that if we fail during matching, we'll
1863 reinitialize the bounds. */
1864 insert_op2 (set_number_at, laststart, b - laststart,
1865 upper_bound - 1, b);
1866 b += 5;
1867 }
1868 }
1869 pending_exact = 0;
1870 beg_interval = NULL;
1871 }
1872 break;
1873
1874 unfetch_interval:
1875 /* If an invalid interval, match the characters as literals. */
1876 assert (beg_interval);
1877 p = beg_interval;
1878 beg_interval = NULL;
1879
1880 /* normal_char and normal_backslash need `c'. */
1881 PATFETCH (c);
1882
1883 if (!(syntax & RE_NO_BK_BRACES))
1884 {
1885 if (p > pattern && p[-1] == '\\')
1886 goto normal_backslash;
1887 }
1888 goto normal_char;
1889
1890 #ifdef emacs
1891 /* There is no way to specify the before_dot and after_dot
1892 operators. rms says this is ok. --karl */
1893 case '=':
1894 BUF_PUSH (at_dot);
1895 break;
1896
1897 case 's':
1898 laststart = b;
1899 PATFETCH (c);
1900 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
1901 break;
1902
1903 case 'S':
1904 laststart = b;
1905 PATFETCH (c);
1906 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
1907 break;
1908 #endif /* emacs */
1909
1910
1911 case 'w':
1912 laststart = b;
1913 BUF_PUSH (wordchar);
1914 break;
1915
1916
1917 case 'W':
1918 laststart = b;
1919 BUF_PUSH (notwordchar);
1920 break;
1921
1922
1923 case '<':
1924 BUF_PUSH (wordbeg);
1925 break;
1926
1927 case '>':
1928 BUF_PUSH (wordend);
1929 break;
1930
1931 case 'b':
1932 BUF_PUSH (wordbound);
1933 break;
1934
1935 case 'B':
1936 BUF_PUSH (notwordbound);
1937 break;
1938
1939 case '`':
1940 BUF_PUSH (begbuf);
1941 break;
1942
1943 case '\'':
1944 BUF_PUSH (endbuf);
1945 break;
1946
1947 case '1': case '2': case '3': case '4': case '5':
1948 case '6': case '7': case '8': case '9':
1949 if (syntax & RE_NO_BK_REFS)
1950 goto normal_char;
1951
1952 c1 = c - '0';
1953
1954 if (c1 > regnum)
1955 return REG_ESUBREG;
1956
1957 /* Can't back reference to a subexpression if inside of it. */
1958 if (group_in_compile_stack (compile_stack, c1))
1959 goto normal_char;
1960
1961 laststart = b;
1962 BUF_PUSH_2 (duplicate, c1);
1963 break;
1964
1965
1966 case '+':
1967 case '?':
1968 if (syntax & RE_BK_PLUS_QM)
1969 goto handle_plus;
1970 else
1971 goto normal_backslash;
1972
1973 default:
1974 normal_backslash:
1975 /* You might think it would be useful for \ to mean
1976 not to translate; but if we don't translate it
1977 it will never match anything. */
1978 c = TRANSLATE (c);
1979 goto normal_char;
1980 }
1981 break;
1982
1983
1984 default:
1985 /* Expects the character in `c'. */
1986 normal_char:
1987 /* If no exactn currently being built. */
1988 if (!pending_exact
1989
1990 /* If last exactn not at current position. */
1991 || pending_exact + *pending_exact + 1 != b
1992
1993 /* We have only one byte following the exactn for the count. */
1994 || *pending_exact == (1 << BYTEWIDTH) - 1
1995
1996 /* If followed by a repetition operator. */
1997 || *p == '*' || *p == '^'
1998 || ((syntax & RE_BK_PLUS_QM)
1999 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2000 : (*p == '+' || *p == '?'))
2001 || ((syntax & RE_INTERVALS)
2002 && ((syntax & RE_NO_BK_BRACES)
2003 ? *p == '{'
2004 : (p[0] == '\\' && p[1] == '{'))))
2005 {
2006 /* Start building a new exactn. */
2007
2008 laststart = b;
2009
2010 BUF_PUSH_2 (exactn, 0);
2011 pending_exact = b - 1;
2012 }
2013
2014 BUF_PUSH (c);
2015 (*pending_exact)++;
2016 break;
2017 } /* switch (c) */
2018 } /* while p != pend */
2019
2020
2021 /* Through the pattern now. */
2022
2023 if (fixup_alt_jump)
2024 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2025
2026 if (!COMPILE_STACK_EMPTY)
2027 return REG_EPAREN;
2028
2029 free (compile_stack.stack);
2030
2031 /* We have succeeded; set the length of the buffer. */
2032 bufp->used = b - bufp->buffer;
2033
2034 #ifdef DEBUG
2035 if (debug)
2036 {
2037 DEBUG_PRINT1 ("\nCompiled pattern: ");
2038 print_compiled_pattern (bufp);
2039 }
2040 #endif /* DEBUG */
2041
2042 return REG_NOERROR;
2043 } /* regex_compile */
2044
2045 /* Subroutines for `regex_compile'. */
2046
2047 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2048
2049 static void
store_op1(op,loc,arg)2050 store_op1 (op, loc, arg)
2051 re_opcode_t op;
2052 unsigned char *loc;
2053 int arg;
2054 {
2055 *loc = (unsigned char) op;
2056 STORE_NUMBER (loc + 1, arg);
2057 }
2058
2059
2060 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2061
2062 static void
store_op2(op,loc,arg1,arg2)2063 store_op2 (op, loc, arg1, arg2)
2064 re_opcode_t op;
2065 unsigned char *loc;
2066 int arg1, arg2;
2067 {
2068 *loc = (unsigned char) op;
2069 STORE_NUMBER (loc + 1, arg1);
2070 STORE_NUMBER (loc + 3, arg2);
2071 }
2072
2073
2074 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
2075 for OP followed by two-byte integer parameter ARG. */
2076
2077 static void
insert_op1(op,loc,arg,end)2078 insert_op1 (op, loc, arg, end)
2079 re_opcode_t op;
2080 unsigned char *loc;
2081 int arg;
2082 unsigned char *end;
2083 {
2084 register unsigned char *pfrom = end;
2085 register unsigned char *pto = end + 3;
2086
2087 while (pfrom != loc)
2088 *--pto = *--pfrom;
2089
2090 store_op1 (op, loc, arg);
2091 }
2092
2093
2094 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2095
2096 static void
insert_op2(op,loc,arg1,arg2,end)2097 insert_op2 (op, loc, arg1, arg2, end)
2098 re_opcode_t op;
2099 unsigned char *loc;
2100 int arg1, arg2;
2101 unsigned char *end;
2102 {
2103 register unsigned char *pfrom = end;
2104 register unsigned char *pto = end + 5;
2105
2106 while (pfrom != loc)
2107 *--pto = *--pfrom;
2108
2109 store_op2 (op, loc, arg1, arg2);
2110 }
2111
2112
2113 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
2114 after an alternative or a begin-subexpression. We assume there is at
2115 least one character before the ^. */
2116
2117 static boolean
at_begline_loc_p(pattern,p,syntax)2118 at_begline_loc_p (pattern, p, syntax)
2119 char *pattern, *p;
2120 reg_syntax_t syntax;
2121 {
2122 char *prev = p - 2;
2123 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
2124
2125 return
2126 /* After a subexpression? */
2127 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
2128 /* After an alternative? */
2129 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
2130 }
2131
2132
2133 /* The dual of at_begline_loc_p. This one is for $. We assume there is
2134 at least one character after the $, i.e., `P < PEND'. */
2135
2136 static boolean
at_endline_loc_p(p,pend,syntax)2137 at_endline_loc_p (p, pend, syntax)
2138 char *p, *pend;
2139 int syntax;
2140 {
2141 char *next = p;
2142 boolean next_backslash = *next == '\\';
2143 char *next_next = p + 1 < pend ? p + 1 : NULL;
2144
2145 return
2146 /* Before a subexpression? */
2147 (syntax & RE_NO_BK_PARENS ? *next == ')'
2148 : next_backslash && next_next && *next_next == ')')
2149 /* Before an alternative? */
2150 || (syntax & RE_NO_BK_VBAR ? *next == '|'
2151 : next_backslash && next_next && *next_next == '|');
2152 }
2153
2154
2155 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
2156 false if it's not. */
2157
2158 static boolean
group_in_compile_stack(compile_stack,regnum)2159 group_in_compile_stack (compile_stack, regnum)
2160 compile_stack_type compile_stack;
2161 regnum_t regnum;
2162 {
2163 int this_element;
2164
2165 for (this_element = compile_stack.avail - 1;
2166 this_element >= 0;
2167 this_element--)
2168 if (compile_stack.stack[this_element].regnum == regnum)
2169 return true;
2170
2171 return false;
2172 }
2173
2174
2175 /* Read the ending character of a range (in a bracket expression) from the
2176 uncompiled pattern *P_PTR (which ends at PEND). We assume the
2177 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
2178 Then we set the translation of all bits between the starting and
2179 ending characters (inclusive) in the compiled pattern B.
2180
2181 Return an error code.
2182
2183 We use these short variable names so we can use the same macros as
2184 `regex_compile' itself. */
2185
2186 static reg_errcode_t
compile_range(p_ptr,pend,translate,syntax,b)2187 compile_range (p_ptr, pend, translate, syntax, b)
2188 char **p_ptr, *pend;
2189 char *translate;
2190 reg_syntax_t syntax;
2191 unsigned char *b;
2192 {
2193 unsigned this_char;
2194
2195 char *p = *p_ptr;
2196 int range_start, range_end;
2197
2198 if (p == pend)
2199 return REG_ERANGE;
2200
2201 /* Even though the pattern is a signed `char *', we need to fetch
2202 with unsigned char *'s; if the high bit of the pattern character
2203 is set, the range endpoints will be negative if we fetch using a
2204 signed char *.
2205
2206 We also want to fetch the endpoints without translating them; the
2207 appropriate translation is done in the bit-setting loop below. */
2208 range_start = ((unsigned char *) p)[-2];
2209 range_end = ((unsigned char *) p)[0];
2210
2211 /* Have to increment the pointer into the pattern string, so the
2212 caller isn't still at the ending character. */
2213 (*p_ptr)++;
2214
2215 /* If the start is after the end, the range is empty. */
2216 if (range_start > range_end)
2217 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
2218
2219 /* Here we see why `this_char' has to be larger than an `unsigned
2220 char' -- the range is inclusive, so if `range_end' == 0xff
2221 (assuming 8-bit characters), we would otherwise go into an infinite
2222 loop, since all characters <= 0xff. */
2223 for (this_char = range_start; this_char <= range_end; this_char++)
2224 {
2225 SET_LIST_BIT (TRANSLATE (this_char));
2226 }
2227
2228 return REG_NOERROR;
2229 }
2230
2231 /* Failure stack declarations and macros; both re_compile_fastmap and
2232 re_match_2 use a failure stack. These have to be macros because of
2233 REGEX_ALLOCATE. */
2234
2235
2236 /* Number of failure points for which to initially allocate space
2237 when matching. If this number is exceeded, we allocate more
2238 space, so it is not a hard limit. */
2239 #ifndef INIT_FAILURE_ALLOC
2240 #define INIT_FAILURE_ALLOC 5
2241 #endif
2242
2243 /* Roughly the maximum number of failure points on the stack. Would be
2244 exactly that if always used MAX_FAILURE_SPACE each time we failed.
2245 This is a variable only so users of regex can assign to it; we never
2246 change it ourselves. */
2247 int re_max_failures = 2000;
2248
2249 typedef unsigned char *fail_stack_elt_t;
2250
2251 typedef struct
2252 {
2253 fail_stack_elt_t *stack;
2254 unsigned size;
2255 unsigned avail; /* Offset of next open position. */
2256 } fail_stack_type;
2257
2258 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
2259 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
2260 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
2261 #define FAIL_STACK_TOP() (fail_stack.stack[fail_stack.avail])
2262
2263
2264 /* Initialize `fail_stack'. Do `return -2' if the alloc fails. */
2265
2266 #define INIT_FAIL_STACK() \
2267 do { \
2268 fail_stack.stack = (fail_stack_elt_t *) \
2269 REGEX_ALLOCATE (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
2270 \
2271 if (fail_stack.stack == NULL) \
2272 return -2; \
2273 \
2274 fail_stack.size = INIT_FAILURE_ALLOC; \
2275 fail_stack.avail = 0; \
2276 } while (0)
2277
2278
2279 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
2280
2281 Return 1 if succeeds, and 0 if either ran out of memory
2282 allocating space for it or it was already too large.
2283
2284 REGEX_REALLOCATE requires `destination' be declared. */
2285
2286 #define DOUBLE_FAIL_STACK(fail_stack) \
2287 ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \
2288 ? 0 \
2289 : ((fail_stack).stack = (fail_stack_elt_t *) \
2290 REGEX_REALLOCATE ((fail_stack).stack, \
2291 (fail_stack).size * sizeof (fail_stack_elt_t), \
2292 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
2293 \
2294 (fail_stack).stack == NULL \
2295 ? 0 \
2296 : ((fail_stack).size <<= 1, \
2297 1)))
2298
2299
2300 /* Push PATTERN_OP on FAIL_STACK.
2301
2302 Return 1 if was able to do so and 0 if ran out of memory allocating
2303 space to do so. */
2304 #define PUSH_PATTERN_OP(pattern_op, fail_stack) \
2305 ((FAIL_STACK_FULL () \
2306 && !DOUBLE_FAIL_STACK (fail_stack)) \
2307 ? 0 \
2308 : ((fail_stack).stack[(fail_stack).avail++] = pattern_op, \
2309 1))
2310
2311 /* This pushes an item onto the failure stack. Must be a four-byte
2312 value. Assumes the variable `fail_stack'. Probably should only
2313 be called from within `PUSH_FAILURE_POINT'. */
2314 #define PUSH_FAILURE_ITEM(item) \
2315 fail_stack.stack[fail_stack.avail++] = (fail_stack_elt_t) item
2316
2317 /* The complement operation. Assumes `fail_stack' is nonempty. */
2318 #define POP_FAILURE_ITEM() fail_stack.stack[--fail_stack.avail]
2319
2320 /* Used to omit pushing failure point id's when we're not debugging. */
2321 #ifdef DEBUG
2322 #define DEBUG_PUSH PUSH_FAILURE_ITEM
2323 #define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_ITEM ()
2324 #else
2325 #define DEBUG_PUSH(item)
2326 #define DEBUG_POP(item_addr)
2327 #endif
2328
2329
2330 /* Push the information about the state we will need
2331 if we ever fail back to it.
2332
2333 Requires variables fail_stack, regstart, regend, reg_info, and
2334 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be
2335 declared.
2336
2337 Does `return FAILURE_CODE' if runs out of memory. */
2338
2339 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
2340 do { \
2341 char *destination; \
2342 /* Must be int, so when we don't save any registers, the arithmetic \
2343 of 0 + -1 isn't done as unsigned. */ \
2344 int this_reg; \
2345 \
2346 DEBUG_STATEMENT (failure_id++); \
2347 DEBUG_STATEMENT (nfailure_points_pushed++); \
2348 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
2349 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
2350 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
2351 \
2352 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \
2353 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
2354 \
2355 /* Ensure we have enough space allocated for what we will push. */ \
2356 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
2357 { \
2358 if (!DOUBLE_FAIL_STACK (fail_stack)) \
2359 return failure_code; \
2360 \
2361 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
2362 (fail_stack).size); \
2363 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
2364 } \
2365 \
2366 /* Push the info, starting with the registers. */ \
2367 DEBUG_PRINT1 ("\n"); \
2368 \
2369 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
2370 this_reg++) \
2371 { \
2372 DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \
2373 DEBUG_STATEMENT (num_regs_pushed++); \
2374 \
2375 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
2376 PUSH_FAILURE_ITEM (regstart[this_reg]); \
2377 \
2378 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
2379 PUSH_FAILURE_ITEM (regend[this_reg]); \
2380 \
2381 DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \
2382 DEBUG_PRINT2 (" match_null=%d", \
2383 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
2384 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
2385 DEBUG_PRINT2 (" matched_something=%d", \
2386 MATCHED_SOMETHING (reg_info[this_reg])); \
2387 DEBUG_PRINT2 (" ever_matched=%d", \
2388 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
2389 DEBUG_PRINT1 ("\n"); \
2390 PUSH_FAILURE_ITEM (reg_info[this_reg].word); \
2391 } \
2392 \
2393 DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\
2394 PUSH_FAILURE_ITEM ((unsigned long)lowest_active_reg); \
2395 \
2396 DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\
2397 PUSH_FAILURE_ITEM ((unsigned long)highest_active_reg); \
2398 \
2399 DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \
2400 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
2401 PUSH_FAILURE_ITEM (pattern_place); \
2402 \
2403 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \
2404 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
2405 size2); \
2406 DEBUG_PRINT1 ("'\n"); \
2407 PUSH_FAILURE_ITEM (string_place); \
2408 \
2409 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
2410 DEBUG_PUSH (failure_id); \
2411 } while (0)
2412
2413 /* This is the number of items that are pushed and popped on the stack
2414 for each register. */
2415 #define NUM_REG_ITEMS 3
2416
2417 /* Individual items aside from the registers. */
2418 #ifdef DEBUG
2419 #define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
2420 #else
2421 #define NUM_NONREG_ITEMS 4
2422 #endif
2423
2424 /* We push at most this many items on the stack. */
2425 #define MAX_FAILURE_ITEMS ((num_regs - 1) * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
2426
2427 /* We actually push this many items. */
2428 #define NUM_FAILURE_ITEMS \
2429 ((highest_active_reg - lowest_active_reg + 1) * NUM_REG_ITEMS \
2430 + NUM_NONREG_ITEMS)
2431
2432 /* How many items can still be added to the stack without overflowing it. */
2433 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
2434
2435
2436 /* Pops what PUSH_FAIL_STACK pushes.
2437
2438 We restore into the parameters, all of which should be lvalues:
2439 STR -- the saved data position.
2440 PAT -- the saved pattern position.
2441 LOW_REG, HIGH_REG -- the highest and lowest active registers.
2442 REGSTART, REGEND -- arrays of string positions.
2443 REG_INFO -- array of information about each subexpression.
2444
2445 Also assumes the variables `fail_stack' and (if debugging), `bufp',
2446 `pend', `string1', `size1', `string2', and `size2'. */
2447
2448 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
2449 { \
2450 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \
2451 unsigned long this_reg; \
2452 unsigned char *string_temp; \
2453 \
2454 assert (!FAIL_STACK_EMPTY ()); \
2455 \
2456 /* Remove failure points and point to how many regs pushed. */ \
2457 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
2458 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
2459 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
2460 \
2461 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
2462 \
2463 DEBUG_POP (&failure_id); \
2464 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
2465 \
2466 /* If the saved string location is NULL, it came from an \
2467 on_failure_keep_string_jump opcode, and we want to throw away the \
2468 saved NULL, thus retaining our current position in the string. */ \
2469 string_temp = POP_FAILURE_ITEM (); \
2470 if (string_temp != NULL) \
2471 str = (char *) string_temp; \
2472 \
2473 DEBUG_PRINT2 (" Popping string 0x%x: `", str); \
2474 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
2475 DEBUG_PRINT1 ("'\n"); \
2476 \
2477 pat = (unsigned char *) POP_FAILURE_ITEM (); \
2478 DEBUG_PRINT2 (" Popping pattern 0x%x: ", pat); \
2479 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
2480 \
2481 /* Restore register info. */ \
2482 high_reg = (unsigned long) POP_FAILURE_ITEM (); \
2483 DEBUG_PRINT2 (" Popping high active reg: %u\n", high_reg); \
2484 \
2485 low_reg = (unsigned long) POP_FAILURE_ITEM (); \
2486 DEBUG_PRINT2 (" Popping low active reg: %u\n", low_reg); \
2487 \
2488 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
2489 { \
2490 DEBUG_PRINT2 (" Popping reg: %u\n", this_reg); \
2491 \
2492 reg_info[this_reg].word = POP_FAILURE_ITEM (); \
2493 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \
2494 \
2495 regend[this_reg] = (char *) POP_FAILURE_ITEM (); \
2496 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
2497 \
2498 regstart[this_reg] = (char *) POP_FAILURE_ITEM (); \
2499 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
2500 } \
2501 \
2502 DEBUG_STATEMENT (nfailure_points_popped++); \
2503 } /* POP_FAILURE_POINT */
2504
2505 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
2506 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
2507 characters can start a string that matches the pattern. This fastmap
2508 is used by re_search to skip quickly over impossible starting points.
2509
2510 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
2511 area as BUFP->fastmap.
2512
2513 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
2514 the pattern buffer.
2515
2516 Returns 0 if we succeed, -2 if an internal error. */
2517
2518 int
re_compile_fastmap(bufp)2519 re_compile_fastmap (bufp)
2520 struct re_pattern_buffer *bufp;
2521 {
2522 int j, k;
2523 fail_stack_type fail_stack;
2524 #ifndef REGEX_MALLOC
2525 char *destination;
2526 #endif
2527 /* We don't push any register information onto the failure stack. */
2528 unsigned num_regs = 0;
2529
2530 register char *fastmap = bufp->fastmap;
2531 unsigned char *pattern = bufp->buffer;
2532 unsigned long size = bufp->used;
2533 unsigned char *p = pattern;
2534 register unsigned char *pend = pattern + size;
2535
2536 /* Assume that each path through the pattern can be null until
2537 proven otherwise. We set this false at the bottom of switch
2538 statement, to which we get only if a particular path doesn't
2539 match the empty string. */
2540 boolean path_can_be_null = true;
2541
2542 /* We aren't doing a `succeed_n' to begin with. */
2543 boolean succeed_n_p = false;
2544
2545 assert (fastmap != NULL && p != NULL);
2546
2547 INIT_FAIL_STACK ();
2548 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
2549 bufp->fastmap_accurate = 1; /* It will be when we're done. */
2550 bufp->can_be_null = 0;
2551
2552 while (p != pend || !FAIL_STACK_EMPTY ())
2553 {
2554 if (p == pend)
2555 {
2556 bufp->can_be_null |= path_can_be_null;
2557
2558 /* Reset for next path. */
2559 path_can_be_null = true;
2560
2561 p = fail_stack.stack[--fail_stack.avail];
2562 }
2563
2564 /* We should never be about to go beyond the end of the pattern. */
2565 assert (p < pend);
2566
2567 #ifdef SWITCH_ENUM_BUG
2568 switch ((int) ((re_opcode_t) *p++))
2569 #else
2570 switch ((re_opcode_t) *p++)
2571 #endif
2572 {
2573
2574 /* I guess the idea here is to simply not bother with a fastmap
2575 if a backreference is used, since it's too hard to figure out
2576 the fastmap for the corresponding group. Setting
2577 `can_be_null' stops `re_search_2' from using the fastmap, so
2578 that is all we do. */
2579 case duplicate:
2580 bufp->can_be_null = 1;
2581 return 0;
2582
2583
2584 /* Following are the cases which match a character. These end
2585 with `break'. */
2586
2587 case exactn:
2588 fastmap[p[1]] = 1;
2589 break;
2590
2591
2592 case charset:
2593 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
2594 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
2595 fastmap[j] = 1;
2596 break;
2597
2598
2599 case charset_not:
2600 /* Chars beyond end of map must be allowed. */
2601 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
2602 fastmap[j] = 1;
2603
2604 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
2605 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
2606 fastmap[j] = 1;
2607 break;
2608
2609
2610 case wordchar:
2611 for (j = 0; j < (1 << BYTEWIDTH); j++)
2612 if (SYNTAX (j) == Sword)
2613 fastmap[j] = 1;
2614 break;
2615
2616
2617 case notwordchar:
2618 for (j = 0; j < (1 << BYTEWIDTH); j++)
2619 if (SYNTAX (j) != Sword)
2620 fastmap[j] = 1;
2621 break;
2622
2623
2624 case anychar:
2625 /* `.' matches anything ... */
2626 for (j = 0; j < (1 << BYTEWIDTH); j++)
2627 fastmap[j] = 1;
2628
2629 /* ... except perhaps newline. */
2630 if (!(bufp->syntax & RE_DOT_NEWLINE))
2631 fastmap['\n'] = 0;
2632
2633 /* Return if we have already set `can_be_null'; if we have,
2634 then the fastmap is irrelevant. Something's wrong here. */
2635 else if (bufp->can_be_null)
2636 return 0;
2637
2638 /* Otherwise, have to check alternative paths. */
2639 break;
2640
2641
2642 #ifdef emacs
2643 case syntaxspec:
2644 k = *p++;
2645 for (j = 0; j < (1 << BYTEWIDTH); j++)
2646 if (SYNTAX (j) == (enum syntaxcode) k)
2647 fastmap[j] = 1;
2648 break;
2649
2650
2651 case notsyntaxspec:
2652 k = *p++;
2653 for (j = 0; j < (1 << BYTEWIDTH); j++)
2654 if (SYNTAX (j) != (enum syntaxcode) k)
2655 fastmap[j] = 1;
2656 break;
2657
2658
2659 /* All cases after this match the empty string. These end with
2660 `continue'. */
2661
2662
2663 case before_dot:
2664 case at_dot:
2665 case after_dot:
2666 continue;
2667 #endif /* not emacs */
2668
2669
2670 case no_op:
2671 case begline:
2672 case endline:
2673 case begbuf:
2674 case endbuf:
2675 case wordbound:
2676 case notwordbound:
2677 case wordbeg:
2678 case wordend:
2679 case push_dummy_failure:
2680 continue;
2681
2682
2683 case jump_n:
2684 case pop_failure_jump:
2685 case maybe_pop_jump:
2686 case jump:
2687 case jump_past_alt:
2688 case dummy_failure_jump:
2689 EXTRACT_NUMBER_AND_INCR (j, p);
2690 p += j;
2691 if (j > 0)
2692 continue;
2693
2694 /* Jump backward implies we just went through the body of a
2695 loop and matched nothing. Opcode jumped to should be
2696 `on_failure_jump' or `succeed_n'. Just treat it like an
2697 ordinary jump. For a * loop, it has pushed its failure
2698 point already; if so, discard that as redundant. */
2699 if ((re_opcode_t) *p != on_failure_jump
2700 && (re_opcode_t) *p != succeed_n)
2701 continue;
2702
2703 p++;
2704 EXTRACT_NUMBER_AND_INCR (j, p);
2705 p += j;
2706
2707 /* If what's on the stack is where we are now, pop it. */
2708 if (!FAIL_STACK_EMPTY ()
2709 && fail_stack.stack[fail_stack.avail - 1] == p)
2710 fail_stack.avail--;
2711
2712 continue;
2713
2714
2715 case on_failure_jump:
2716 case on_failure_keep_string_jump:
2717 handle_on_failure_jump:
2718 EXTRACT_NUMBER_AND_INCR (j, p);
2719
2720 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
2721 end of the pattern. We don't want to push such a point,
2722 since when we restore it above, entering the switch will
2723 increment `p' past the end of the pattern. We don't need
2724 to push such a point since we obviously won't find any more
2725 fastmap entries beyond `pend'. Such a pattern can match
2726 the null string, though. */
2727 if (p + j < pend)
2728 {
2729 if (!PUSH_PATTERN_OP (p + j, fail_stack))
2730 return -2;
2731 }
2732 else
2733 bufp->can_be_null = 1;
2734
2735 if (succeed_n_p)
2736 {
2737 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
2738 succeed_n_p = false;
2739 }
2740
2741 continue;
2742
2743
2744 case succeed_n:
2745 /* Get to the number of times to succeed. */
2746 p += 2;
2747
2748 /* Increment p past the n for when k != 0. */
2749 EXTRACT_NUMBER_AND_INCR (k, p);
2750 if (k == 0)
2751 {
2752 p -= 4;
2753 succeed_n_p = true; /* Spaghetti code alert. */
2754 goto handle_on_failure_jump;
2755 }
2756 continue;
2757
2758
2759 case set_number_at:
2760 p += 4;
2761 continue;
2762
2763
2764 case start_memory:
2765 case stop_memory:
2766 p += 2;
2767 continue;
2768
2769
2770 default:
2771 abort (); /* We have listed all the cases. */
2772 } /* switch *p++ */
2773
2774 /* Getting here means we have found the possible starting
2775 characters for one path of the pattern -- and that the empty
2776 string does not match. We need not follow this path further.
2777 Instead, look at the next alternative (remembered on the
2778 stack), or quit if no more. The test at the top of the loop
2779 does these things. */
2780 path_can_be_null = false;
2781 p = pend;
2782 } /* while p */
2783
2784 /* Set `can_be_null' for the last path (also the first path, if the
2785 pattern is empty). */
2786 bufp->can_be_null |= path_can_be_null;
2787 return 0;
2788 } /* re_compile_fastmap */
2789
2790 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
2791 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
2792 this memory for recording register information. STARTS and ENDS
2793 must be allocated using the malloc library routine, and must each
2794 be at least NUM_REGS * sizeof (regoff_t) bytes long.
2795
2796 If NUM_REGS == 0, then subsequent matches should allocate their own
2797 register data.
2798
2799 Unless this function is called, the first search or match using
2800 PATTERN_BUFFER will allocate its own register data, without
2801 freeing the old data. */
2802
2803 void
re_set_registers(bufp,regs,num_regs,starts,ends)2804 re_set_registers (bufp, regs, num_regs, starts, ends)
2805 struct re_pattern_buffer *bufp;
2806 struct re_registers *regs;
2807 unsigned num_regs;
2808 regoff_t *starts, *ends;
2809 {
2810 if (num_regs)
2811 {
2812 bufp->regs_allocated = REGS_REALLOCATE;
2813 regs->num_regs = num_regs;
2814 regs->start = starts;
2815 regs->end = ends;
2816 }
2817 else
2818 {
2819 bufp->regs_allocated = REGS_UNALLOCATED;
2820 regs->num_regs = 0;
2821 regs->start = regs->end = (regoff_t) 0;
2822 }
2823 }
2824
2825 /* Searching routines. */
2826
2827 /* Like re_search_2, below, but only one string is specified, and
2828 doesn't let you say where to stop matching. */
2829
2830 int
re_search(bufp,string,size,startpos,range,regs)2831 re_search (bufp, string, size, startpos, range, regs)
2832 struct re_pattern_buffer *bufp;
2833 char *string;
2834 int size, startpos, range;
2835 struct re_registers *regs;
2836 {
2837 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
2838 regs, size);
2839 }
2840
2841
2842 /* Using the compiled pattern in BUFP->buffer, first tries to match the
2843 virtual concatenation of STRING1 and STRING2, starting first at index
2844 STARTPOS, then at STARTPOS + 1, and so on.
2845
2846 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
2847
2848 RANGE is how far to scan while trying to match. RANGE = 0 means try
2849 only at STARTPOS; in general, the last start tried is STARTPOS +
2850 RANGE.
2851
2852 In REGS, return the indices of the virtual concatenation of STRING1
2853 and STRING2 that matched the entire BUFP->buffer and its contained
2854 subexpressions.
2855
2856 Do not consider matching one past the index STOP in the virtual
2857 concatenation of STRING1 and STRING2.
2858
2859 We return either the position in the strings at which the match was
2860 found, -1 if no match, or -2 if error (such as failure
2861 stack overflow). */
2862
2863 int
re_search_2(bufp,string1,size1,string2,size2,startpos,range,regs,stop)2864 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
2865 struct re_pattern_buffer *bufp;
2866 char *string1, *string2;
2867 int size1, size2;
2868 int startpos;
2869 int range;
2870 struct re_registers *regs;
2871 int stop;
2872 {
2873 int val;
2874 register char *fastmap = bufp->fastmap;
2875 register char *translate = bufp->translate;
2876 int total_size = size1 + size2;
2877 int endpos = startpos + range;
2878
2879 /* Check for out-of-range STARTPOS. */
2880 if (startpos < 0 || startpos > total_size)
2881 return -1;
2882
2883 /* Fix up RANGE if it might eventually take us outside
2884 the virtual concatenation of STRING1 and STRING2. */
2885 if (endpos < -1)
2886 range = -1 - startpos;
2887 else if (endpos > total_size)
2888 range = total_size - startpos;
2889
2890 /* If the search isn't to be a backwards one, don't waste time in a
2891 search for a pattern that must be anchored. */
2892 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
2893 {
2894 if (startpos > 0)
2895 return -1;
2896 else
2897 range = 1;
2898 }
2899
2900 /* Update the fastmap now if not correct already. */
2901 if (fastmap && !bufp->fastmap_accurate)
2902 if (re_compile_fastmap (bufp) == -2)
2903 return -2;
2904
2905 /* Loop through the string, looking for a place to start matching. */
2906 for (;;)
2907 {
2908 /* If a fastmap is supplied, skip quickly over characters that
2909 cannot be the start of a match. If the pattern can match the
2910 null string, however, we don't need to skip characters; we want
2911 the first null string. */
2912 if (fastmap && startpos < total_size && !bufp->can_be_null)
2913 {
2914 if (range > 0) /* Searching forwards. */
2915 {
2916 register char *d;
2917 register int lim = 0;
2918 int irange = range;
2919
2920 if (startpos < size1 && startpos + range >= size1)
2921 lim = range - (size1 - startpos);
2922
2923 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
2924
2925 /* Written out as an if-else to avoid testing `translate'
2926 inside the loop. */
2927 if (translate)
2928 while (range > lim
2929 && !fastmap[(unsigned char)
2930 translate[(unsigned char) *d++]])
2931 range--;
2932 else
2933 while (range > lim && !fastmap[(unsigned char) *d++])
2934 range--;
2935
2936 startpos += irange - range;
2937 }
2938 else /* Searching backwards. */
2939 {
2940 register char c = (size1 == 0 || startpos >= size1
2941 ? string2[startpos - size1]
2942 : string1[startpos]);
2943
2944 if (!fastmap[(unsigned char) TRANSLATE (c)])
2945 goto advance;
2946 }
2947 }
2948
2949 /* If can't match the null string, and that's all we have left, fail. */
2950 if (range >= 0 && startpos == total_size && fastmap
2951 && !bufp->can_be_null)
2952 return -1;
2953
2954 val = re_match_2 (bufp, string1, size1, string2, size2,
2955 startpos, regs, stop);
2956 if (val >= 0)
2957 return startpos;
2958
2959 if (val == -2)
2960 return -2;
2961
2962 advance:
2963 if (!range)
2964 break;
2965 else if (range > 0)
2966 {
2967 range--;
2968 startpos++;
2969 }
2970 else
2971 {
2972 range++;
2973 startpos--;
2974 }
2975 }
2976 return -1;
2977 } /* re_search_2 */
2978
2979 /* Declarations and macros for re_match_2. */
2980
2981 static int bcmp_translate ();
2982 static boolean alt_match_null_string_p (),
2983 common_op_match_null_string_p (),
2984 group_match_null_string_p ();
2985
2986 /* Structure for per-register (a.k.a. per-group) information.
2987 This must not be longer than one word, because we push this value
2988 onto the failure stack. Other register information, such as the
2989 starting and ending positions (which are addresses), and the list of
2990 inner groups (which is a bits list) are maintained in separate
2991 variables.
2992
2993 We are making a (strictly speaking) nonportable assumption here: that
2994 the compiler will pack our bit fields into something that fits into
2995 the type of `word', i.e., is something that fits into one item on the
2996 failure stack. */
2997 typedef union
2998 {
2999 fail_stack_elt_t word;
3000 struct
3001 {
3002 /* This field is one if this group can match the empty string,
3003 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
3004 #define MATCH_NULL_UNSET_VALUE 3
3005 unsigned match_null_string_p : 2;
3006 unsigned is_active : 1;
3007 unsigned matched_something : 1;
3008 unsigned ever_matched_something : 1;
3009 } bits;
3010 } register_info_type;
3011
3012 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
3013 #define IS_ACTIVE(R) ((R).bits.is_active)
3014 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
3015 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
3016
3017
3018 /* Call this when have matched a real character; it sets `matched' flags
3019 for the subexpressions which we are currently inside. Also records
3020 that those subexprs have matched. */
3021 #define SET_REGS_MATCHED() \
3022 do \
3023 { \
3024 unsigned r; \
3025 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
3026 { \
3027 MATCHED_SOMETHING (reg_info[r]) \
3028 = EVER_MATCHED_SOMETHING (reg_info[r]) \
3029 = 1; \
3030 } \
3031 } \
3032 while (0)
3033
3034
3035 /* This converts PTR, a pointer into one of the search strings `string1'
3036 and `string2' into an offset from the beginning of that string. */
3037 #define POINTER_TO_OFFSET(ptr) \
3038 (FIRST_STRING_P (ptr) ? (ptr) - string1 : (ptr) - string2 + size1)
3039
3040 /* Registers are set to a sentinel when they haven't yet matched. */
3041 #define REG_UNSET_VALUE ((char *) -1)
3042 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
3043
3044
3045 /* Macros for dealing with the split strings in re_match_2. */
3046
3047 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3048
3049 /* Call before fetching a character with *d. This switches over to
3050 string2 if necessary. */
3051 #define PREFETCH() \
3052 while (d == dend) \
3053 { \
3054 /* End of string2 => fail. */ \
3055 if (dend == end_match_2) \
3056 goto fail; \
3057 /* End of string1 => advance to string2. */ \
3058 d = string2; \
3059 dend = end_match_2; \
3060 }
3061
3062
3063 /* Test if at very beginning or at very end of the virtual concatenation
3064 of `string1' and `string2'. If only one string, it's `string2'. */
3065 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3066 #define AT_STRINGS_END(d) ((d) == end2)
3067
3068
3069 /* Test if D points to a character which is word-constituent. We have
3070 two special cases to check for: if past the end of string1, look at
3071 the first character in string2; and if before the beginning of
3072 string2, look at the last character in string1. */
3073 #define WORDCHAR_P(d) \
3074 (SYNTAX ((d) == end1 ? *string2 \
3075 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3076 == Sword)
3077
3078 /* Test if the character before D and the one at D differ with respect
3079 to being word-constituent. */
3080 #define AT_WORD_BOUNDARY(d) \
3081 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3082 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3083
3084
3085 /* Free everything we malloc. */
3086 #ifdef REGEX_MALLOC
3087 #define FREE_VAR(var) if (var) free (var); var = NULL
3088 #define FREE_VARIABLES() \
3089 do { \
3090 FREE_VAR (fail_stack.stack); \
3091 FREE_VAR (regstart); \
3092 FREE_VAR (regend); \
3093 FREE_VAR (old_regstart); \
3094 FREE_VAR (old_regend); \
3095 FREE_VAR (best_regstart); \
3096 FREE_VAR (best_regend); \
3097 FREE_VAR (reg_info); \
3098 FREE_VAR (reg_dummy); \
3099 FREE_VAR (reg_info_dummy); \
3100 } while (0)
3101 #else /* not REGEX_MALLOC */
3102 /* Some MIPS systems (at least) want this to free alloca'd storage. */
3103 #define FREE_VARIABLES() alloca (0)
3104 #endif /* not REGEX_MALLOC */
3105
3106
3107 /* These values must meet several constraints. They must not be valid
3108 register values; since we have a limit of 255 registers (because
3109 we use only one byte in the pattern for the register number), we can
3110 use numbers larger than 255. They must differ by 1, because of
3111 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3112 be larger than the value for the highest register, so we do not try
3113 to actually save any registers when none are active. */
3114 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3115 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3116
3117 /* Matching routines. */
3118
3119 #ifndef emacs /* Emacs never uses this. */
3120 /* re_match is like re_match_2 except it takes only a single string. */
3121
3122 int
re_match(bufp,string,size,pos,regs)3123 re_match (bufp, string, size, pos, regs)
3124 struct re_pattern_buffer *bufp;
3125 char *string;
3126 int size, pos;
3127 struct re_registers *regs;
3128 {
3129 return re_match_2 (bufp, NULL, 0, string, size, pos, regs, size);
3130 }
3131 #endif /* not emacs */
3132
3133
3134 /* re_match_2 matches the compiled pattern in BUFP against the
3135 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3136 and SIZE2, respectively). We start matching at POS, and stop
3137 matching at STOP.
3138
3139 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3140 store offsets for the substring each group matched in REGS. See the
3141 documentation for exactly how many groups we fill.
3142
3143 We return -1 if no match, -2 if an internal error (such as the
3144 failure stack overflowing). Otherwise, we return the length of the
3145 matched substring. */
3146
3147 int
re_match_2(bufp,string1,size1,string2,size2,pos,regs,stop)3148 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3149 struct re_pattern_buffer *bufp;
3150 char *string1, *string2;
3151 int size1, size2;
3152 int pos;
3153 struct re_registers *regs;
3154 int stop;
3155 {
3156 /* General temporaries. */
3157 int mcnt;
3158 unsigned char *p1;
3159
3160 /* Just past the end of the corresponding string. */
3161 char *end1, *end2;
3162
3163 /* Pointers into string1 and string2, just past the last characters in
3164 each to consider matching. */
3165 char *end_match_1, *end_match_2;
3166
3167 /* Where we are in the data, and the end of the current string. */
3168 char *d, *dend;
3169
3170 /* Where we are in the pattern, and the end of the pattern. */
3171 unsigned char *p = bufp->buffer;
3172 register unsigned char *pend = p + bufp->used;
3173
3174 /* We use this to map every character in the string. */
3175 char *translate = bufp->translate;
3176
3177 /* Failure point stack. Each place that can handle a failure further
3178 down the line pushes a failure point on this stack. It consists of
3179 restart, regend, and reg_info for all registers corresponding to
3180 the subexpressions we're currently inside, plus the number of such
3181 registers, and, finally, two char *'s. The first char * is where
3182 to resume scanning the pattern; the second one is where to resume
3183 scanning the strings. If the latter is zero, the failure point is
3184 a ``dummy''; if a failure happens and the failure point is a dummy,
3185 it gets discarded and the next next one is tried. */
3186 fail_stack_type fail_stack;
3187 #ifdef DEBUG
3188 static unsigned failure_id = 0;
3189 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3190 #endif
3191
3192 /* We fill all the registers internally, independent of what we
3193 return, for use in backreferences. The number here includes
3194 an element for register zero. */
3195 unsigned num_regs = bufp->re_nsub + 1;
3196
3197 /* The currently active registers. */
3198 unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3199 unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3200
3201 /* Information on the contents of registers. These are pointers into
3202 the input strings; they record just what was matched (on this
3203 attempt) by a subexpression part of the pattern, that is, the
3204 regnum-th regstart pointer points to where in the pattern we began
3205 matching and the regnum-th regend points to right after where we
3206 stopped matching the regnum-th subexpression. (The zeroth register
3207 keeps track of what the whole pattern matches.) */
3208 char **regstart, **regend;
3209
3210 /* If a group that's operated upon by a repetition operator fails to
3211 match anything, then the register for its start will need to be
3212 restored because it will have been set to wherever in the string we
3213 are when we last see its open-group operator. Similarly for a
3214 register's end. */
3215 char **old_regstart, **old_regend;
3216
3217 /* The is_active field of reg_info helps us keep track of which (possibly
3218 nested) subexpressions we are currently in. The matched_something
3219 field of reg_info[reg_num] helps us tell whether or not we have
3220 matched any of the pattern so far this time through the reg_num-th
3221 subexpression. These two fields get reset each time through any
3222 loop their register is in. */
3223 register_info_type *reg_info;
3224
3225 /* The following record the register info as found in the above
3226 variables when we find a match better than any we've seen before.
3227 This happens as we backtrack through the failure points, which in
3228 turn happens only if we have not yet matched the entire string. */
3229 unsigned best_regs_set = false;
3230 char **best_regstart, **best_regend;
3231
3232 /* Logically, this is `best_regend[0]'. But we don't want to have to
3233 allocate space for that if we're not allocating space for anything
3234 else (see below). Also, we never need info about register 0 for
3235 any of the other register vectors, and it seems rather a kludge to
3236 treat `best_regend' differently than the rest. So we keep track of
3237 the end of the best match so far in a separate variable. We
3238 initialize this to NULL so that when we backtrack the first time
3239 and need to test it, it's not garbage. */
3240 char *match_end = NULL;
3241
3242 /* Used when we pop values we don't care about. */
3243 char **reg_dummy;
3244 register_info_type *reg_info_dummy;
3245
3246 #ifdef DEBUG
3247 /* Counts the total number of registers pushed. */
3248 unsigned num_regs_pushed = 0;
3249 #endif
3250
3251 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3252
3253 INIT_FAIL_STACK ();
3254
3255 /* Do not bother to initialize all the register variables if there are
3256 no groups in the pattern, as it takes a fair amount of time. If
3257 there are groups, we include space for register 0 (the whole
3258 pattern), even though we never use it, since it simplifies the
3259 array indexing. We should fix this. */
3260 if (bufp->re_nsub)
3261 {
3262 regstart = REGEX_TALLOC (num_regs, char *);
3263 regend = REGEX_TALLOC (num_regs, char *);
3264 old_regstart = REGEX_TALLOC (num_regs, char *);
3265 old_regend = REGEX_TALLOC (num_regs, char *);
3266 best_regstart = REGEX_TALLOC (num_regs, char *);
3267 best_regend = REGEX_TALLOC (num_regs, char *);
3268 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3269 reg_dummy = REGEX_TALLOC (num_regs, char *);
3270 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3271
3272 if (!(regstart && regend && old_regstart && old_regend && reg_info
3273 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3274 {
3275 FREE_VARIABLES ();
3276 return -2;
3277 }
3278 }
3279 #ifdef REGEX_MALLOC
3280 else
3281 {
3282 /* We must initialize all our variables to NULL, so that
3283 `FREE_VARIABLES' doesn't try to free them. */
3284 regstart = regend = old_regstart = old_regend = best_regstart
3285 = best_regend = reg_dummy = NULL;
3286 reg_info = reg_info_dummy = (register_info_type *) NULL;
3287 }
3288 #endif /* REGEX_MALLOC */
3289
3290 /* The starting position is bogus. */
3291 if (pos < 0 || pos > size1 + size2)
3292 {
3293 FREE_VARIABLES ();
3294 return -1;
3295 }
3296
3297 /* Initialize subexpression text positions to -1 to mark ones that no
3298 start_memory/stop_memory has been seen for. Also initialize the
3299 register information struct. */
3300 for (mcnt = 1; mcnt < num_regs; mcnt++)
3301 {
3302 regstart[mcnt] = regend[mcnt]
3303 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3304
3305 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3306 IS_ACTIVE (reg_info[mcnt]) = 0;
3307 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3308 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3309 }
3310
3311 /* We move `string1' into `string2' if the latter's empty -- but not if
3312 `string1' is null. */
3313 if (size2 == 0 && string1 != NULL)
3314 {
3315 string2 = string1;
3316 size2 = size1;
3317 string1 = 0;
3318 size1 = 0;
3319 }
3320 end1 = string1 + size1;
3321 end2 = string2 + size2;
3322
3323 /* Compute where to stop matching, within the two strings. */
3324 if (stop <= size1)
3325 {
3326 end_match_1 = string1 + stop;
3327 end_match_2 = string2;
3328 }
3329 else
3330 {
3331 end_match_1 = end1;
3332 end_match_2 = string2 + stop - size1;
3333 }
3334
3335 /* `p' scans through the pattern as `d' scans through the data.
3336 `dend' is the end of the input string that `d' points within. `d'
3337 is advanced into the following input string whenever necessary, but
3338 this happens before fetching; therefore, at the beginning of the
3339 loop, `d' can be pointing at the end of a string, but it cannot
3340 equal `string2'. */
3341 if (size1 > 0 && pos <= size1)
3342 {
3343 d = string1 + pos;
3344 dend = end_match_1;
3345 }
3346 else
3347 {
3348 d = string2 + pos - size1;
3349 dend = end_match_2;
3350 }
3351
3352 DEBUG_PRINT1 ("The compiled pattern is: ");
3353 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
3354 DEBUG_PRINT1 ("The string to match is: `");
3355 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
3356 DEBUG_PRINT1 ("'\n");
3357
3358 /* This loops over pattern commands. It exits by returning from the
3359 function if the match is complete, or it drops through if the match
3360 fails at this starting point in the input data. */
3361 for (;;)
3362 {
3363 DEBUG_PRINT2 ("\n0x%x: ", p);
3364
3365 if (p == pend)
3366 { /* End of pattern means we might have succeeded. */
3367 DEBUG_PRINT1 ("end of pattern ... ");
3368
3369 /* If we haven't matched the entire string, and we want the
3370 longest match, try backtracking. */
3371 if (d != end_match_2)
3372 {
3373 DEBUG_PRINT1 ("backtracking.\n");
3374
3375 if (!FAIL_STACK_EMPTY ())
3376 { /* More failure points to try. */
3377 boolean same_str_p = (FIRST_STRING_P (match_end)
3378 == MATCHING_IN_FIRST_STRING);
3379
3380 /* If exceeds best match so far, save it. */
3381 if (!best_regs_set
3382 || (same_str_p && d > match_end)
3383 || (!same_str_p && !MATCHING_IN_FIRST_STRING))
3384 {
3385 best_regs_set = true;
3386 match_end = d;
3387
3388 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
3389
3390 for (mcnt = 1; mcnt < num_regs; mcnt++)
3391 {
3392 best_regstart[mcnt] = regstart[mcnt];
3393 best_regend[mcnt] = regend[mcnt];
3394 }
3395 }
3396 goto fail;
3397 }
3398
3399 /* If no failure points, don't restore garbage. */
3400 else if (best_regs_set)
3401 {
3402 restore_best_regs:
3403 /* Restore best match. It may happen that `dend ==
3404 end_match_1' while the restored d is in string2.
3405 For example, the pattern `x.*y.*z' against the
3406 strings `x-' and `y-z-', if the two strings are
3407 not consecutive in memory. */
3408 DEBUG_PRINT1 ("Restoring best registers.\n");
3409
3410 d = match_end;
3411 dend = ((d >= string1 && d <= end1)
3412 ? end_match_1 : end_match_2);
3413
3414 for (mcnt = 1; mcnt < num_regs; mcnt++)
3415 {
3416 regstart[mcnt] = best_regstart[mcnt];
3417 regend[mcnt] = best_regend[mcnt];
3418 }
3419 }
3420 } /* d != end_match_2 */
3421
3422 DEBUG_PRINT1 ("Accepting match.\n");
3423
3424 /* If caller wants register contents data back, do it. */
3425 if (regs && !bufp->no_sub)
3426 {
3427 /* Have the register data arrays been allocated? */
3428 if (bufp->regs_allocated == REGS_UNALLOCATED)
3429 { /* No. So allocate them with malloc. We need one
3430 extra element beyond `num_regs' for the `-1' marker
3431 GNU code uses. */
3432 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
3433 regs->start = TALLOC (regs->num_regs, regoff_t);
3434 regs->end = TALLOC (regs->num_regs, regoff_t);
3435 if (regs->start == NULL || regs->end == NULL)
3436 return -2;
3437 bufp->regs_allocated = REGS_REALLOCATE;
3438 }
3439 else if (bufp->regs_allocated == REGS_REALLOCATE)
3440 { /* Yes. If we need more elements than were already
3441 allocated, reallocate them. If we need fewer, just
3442 leave it alone. */
3443 if (regs->num_regs < num_regs + 1)
3444 {
3445 regs->num_regs = num_regs + 1;
3446 RETALLOC (regs->start, regs->num_regs, regoff_t);
3447 RETALLOC (regs->end, regs->num_regs, regoff_t);
3448 if (regs->start == NULL || regs->end == NULL)
3449 return -2;
3450 }
3451 }
3452 else
3453 assert (bufp->regs_allocated == REGS_FIXED);
3454
3455 /* Convert the pointer data in `regstart' and `regend' to
3456 indices. Register zero has to be set differently,
3457 since we haven't kept track of any info for it. */
3458 if (regs->num_regs > 0)
3459 {
3460 regs->start[0] = pos;
3461 regs->end[0] = (MATCHING_IN_FIRST_STRING ? d - string1
3462 : d - string2 + size1);
3463 }
3464
3465 /* Go through the first `min (num_regs, regs->num_regs)'
3466 registers, since that is all we initialized. */
3467 for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++)
3468 {
3469 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
3470 regs->start[mcnt] = regs->end[mcnt] = -1;
3471 else
3472 {
3473 regs->start[mcnt] = POINTER_TO_OFFSET (regstart[mcnt]);
3474 regs->end[mcnt] = POINTER_TO_OFFSET (regend[mcnt]);
3475 }
3476 }
3477
3478 /* If the regs structure we return has more elements than
3479 were in the pattern, set the extra elements to -1. If
3480 we (re)allocated the registers, this is the case,
3481 because we always allocate enough to have at least one
3482 -1 at the end. */
3483 for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++)
3484 regs->start[mcnt] = regs->end[mcnt] = -1;
3485 } /* regs && !bufp->no_sub */
3486
3487 FREE_VARIABLES ();
3488 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
3489 nfailure_points_pushed, nfailure_points_popped,
3490 nfailure_points_pushed - nfailure_points_popped);
3491 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
3492
3493 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
3494 ? string1
3495 : string2 - size1);
3496
3497 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
3498
3499 return mcnt;
3500 }
3501
3502 /* Otherwise match next pattern command. */
3503 #ifdef SWITCH_ENUM_BUG
3504 switch ((int) ((re_opcode_t) *p++))
3505 #else
3506 switch ((re_opcode_t) *p++)
3507 #endif
3508 {
3509 /* Ignore these. Used to ignore the n of succeed_n's which
3510 currently have n == 0. */
3511 case no_op:
3512 DEBUG_PRINT1 ("EXECUTING no_op.\n");
3513 break;
3514
3515
3516 /* Match the next n pattern characters exactly. The following
3517 byte in the pattern defines n, and the n bytes after that
3518 are the characters to match. */
3519 case exactn:
3520 mcnt = *p++;
3521 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
3522
3523 /* This is written out as an if-else so we don't waste time
3524 testing `translate' inside the loop. */
3525 if (translate)
3526 {
3527 do
3528 {
3529 PREFETCH ();
3530 if (translate[(unsigned char) *d++] != (char) *p++)
3531 goto fail;
3532 }
3533 while (--mcnt);
3534 }
3535 else
3536 {
3537 do
3538 {
3539 PREFETCH ();
3540 if (*d++ != (char) *p++) goto fail;
3541 }
3542 while (--mcnt);
3543 }
3544 SET_REGS_MATCHED ();
3545 break;
3546
3547
3548 /* Match any character except possibly a newline or a null. */
3549 case anychar:
3550 DEBUG_PRINT1 ("EXECUTING anychar.\n");
3551
3552 PREFETCH ();
3553
3554 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
3555 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
3556 goto fail;
3557
3558 SET_REGS_MATCHED ();
3559 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
3560 d++;
3561 break;
3562
3563
3564 case charset:
3565 case charset_not:
3566 {
3567 register unsigned char c;
3568 boolean not = (re_opcode_t) *(p - 1) == charset_not;
3569
3570 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
3571
3572 PREFETCH ();
3573 c = TRANSLATE (*d); /* The character to match. */
3574
3575 /* Cast to `unsigned' instead of `unsigned char' in case the
3576 bit list is a full 32 bytes long. */
3577 if (c < (unsigned) (*p * BYTEWIDTH)
3578 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
3579 not = !not;
3580
3581 p += 1 + *p;
3582
3583 if (!not) goto fail;
3584
3585 SET_REGS_MATCHED ();
3586 d++;
3587 break;
3588 }
3589
3590
3591 /* The beginning of a group is represented by start_memory.
3592 The arguments are the register number in the next byte, and the
3593 number of groups inner to this one in the next. The text
3594 matched within the group is recorded (in the internal
3595 registers data structure) under the register number. */
3596 case start_memory:
3597 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
3598
3599 /* Find out if this group can match the empty string. */
3600 p1 = p; /* To send to group_match_null_string_p. */
3601
3602 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
3603 REG_MATCH_NULL_STRING_P (reg_info[*p])
3604 = group_match_null_string_p (&p1, pend, reg_info);
3605
3606 /* Save the position in the string where we were the last time
3607 we were at this open-group operator in case the group is
3608 operated upon by a repetition operator, e.g., with `(a*)*b'
3609 against `ab'; then we want to ignore where we are now in
3610 the string in case this attempt to match fails. */
3611 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
3612 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
3613 : regstart[*p];
3614 DEBUG_PRINT2 (" old_regstart: %d\n",
3615 POINTER_TO_OFFSET (old_regstart[*p]));
3616
3617 regstart[*p] = d;
3618 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
3619
3620 IS_ACTIVE (reg_info[*p]) = 1;
3621 MATCHED_SOMETHING (reg_info[*p]) = 0;
3622
3623 /* This is the new highest active register. */
3624 highest_active_reg = *p;
3625
3626 /* If nothing was active before, this is the new lowest active
3627 register. */
3628 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
3629 lowest_active_reg = *p;
3630
3631 /* Move past the register number and inner group count. */
3632 p += 2;
3633 break;
3634
3635
3636 /* The stop_memory opcode represents the end of a group. Its
3637 arguments are the same as start_memory's: the register
3638 number, and the number of inner groups. */
3639 case stop_memory:
3640 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
3641
3642 /* We need to save the string position the last time we were at
3643 this close-group operator in case the group is operated
3644 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
3645 against `aba'; then we want to ignore where we are now in
3646 the string in case this attempt to match fails. */
3647 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
3648 ? REG_UNSET (regend[*p]) ? d : regend[*p]
3649 : regend[*p];
3650 DEBUG_PRINT2 (" old_regend: %d\n",
3651 POINTER_TO_OFFSET (old_regend[*p]));
3652
3653 regend[*p] = d;
3654 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
3655
3656 /* This register isn't active anymore. */
3657 IS_ACTIVE (reg_info[*p]) = 0;
3658
3659 /* If this was the only register active, nothing is active
3660 anymore. */
3661 if (lowest_active_reg == highest_active_reg)
3662 {
3663 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3664 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3665 }
3666 else
3667 { /* We must scan for the new highest active register, since
3668 it isn't necessarily one less than now: consider
3669 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
3670 new highest active register is 1. */
3671 unsigned char r = *p - 1;
3672 while (r > 0 && !IS_ACTIVE (reg_info[r]))
3673 r--;
3674
3675 /* If we end up at register zero, that means that we saved
3676 the registers as the result of an `on_failure_jump', not
3677 a `start_memory', and we jumped to past the innermost
3678 `stop_memory'. For example, in ((.)*) we save
3679 registers 1 and 2 as a result of the *, but when we pop
3680 back to the second ), we are at the stop_memory 1.
3681 Thus, nothing is active. */
3682 if (r == 0)
3683 {
3684 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3685 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3686 }
3687 else
3688 highest_active_reg = r;
3689 }
3690
3691 /* If just failed to match something this time around with a
3692 group that's operated on by a repetition operator, try to
3693 force exit from the ``loop'', and restore the register
3694 information for this group that we had before trying this
3695 last match. */
3696 if ((!MATCHED_SOMETHING (reg_info[*p])
3697 || (re_opcode_t) p[-3] == start_memory)
3698 && (p + 2) < pend)
3699 {
3700 boolean is_a_jump_n = false;
3701
3702 p1 = p + 2;
3703 mcnt = 0;
3704 switch ((re_opcode_t) *p1++)
3705 {
3706 case jump_n:
3707 is_a_jump_n = true;
3708 case pop_failure_jump:
3709 case maybe_pop_jump:
3710 case jump:
3711 case dummy_failure_jump:
3712 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
3713 if (is_a_jump_n)
3714 p1 += 2;
3715 break;
3716
3717 default:
3718 /* do nothing */ ;
3719 }
3720 p1 += mcnt;
3721
3722 /* If the next operation is a jump backwards in the pattern
3723 to an on_failure_jump right before the start_memory
3724 corresponding to this stop_memory, exit from the loop
3725 by forcing a failure after pushing on the stack the
3726 on_failure_jump's jump in the pattern, and d. */
3727 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
3728 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
3729 {
3730 /* If this group ever matched anything, then restore
3731 what its registers were before trying this last
3732 failed match, e.g., with `(a*)*b' against `ab' for
3733 regstart[1], and, e.g., with `((a*)*(b*)*)*'
3734 against `aba' for regend[3].
3735
3736 Also restore the registers for inner groups for,
3737 e.g., `((a*)(b*))*' against `aba' (register 3 would
3738 otherwise get trashed). */
3739
3740 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
3741 {
3742 unsigned r;
3743
3744 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
3745
3746 /* Restore this and inner groups' (if any) registers. */
3747 for (r = *p; r < *p + *(p + 1); r++)
3748 {
3749 regstart[r] = old_regstart[r];
3750
3751 /* xx why this test? */
3752 if ((unsigned long) old_regend[r] >= (unsigned long) regstart[r])
3753 regend[r] = old_regend[r];
3754 }
3755 }
3756 p1++;
3757 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
3758 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
3759
3760 goto fail;
3761 }
3762 }
3763
3764 /* Move past the register number and the inner group count. */
3765 p += 2;
3766 break;
3767
3768
3769 /* \<digit> has been turned into a `duplicate' command which is
3770 followed by the numeric value of <digit> as the register number. */
3771 case duplicate:
3772 {
3773 register char *d2, *dend2;
3774 int regno = *p++; /* Get which register to match against. */
3775 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
3776
3777 /* Can't back reference a group which we've never matched. */
3778 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
3779 goto fail;
3780
3781 /* Where in input to try to start matching. */
3782 d2 = regstart[regno];
3783
3784 /* Where to stop matching; if both the place to start and
3785 the place to stop matching are in the same string, then
3786 set to the place to stop, otherwise, for now have to use
3787 the end of the first string. */
3788
3789 dend2 = ((FIRST_STRING_P (regstart[regno])
3790 == FIRST_STRING_P (regend[regno]))
3791 ? regend[regno] : end_match_1);
3792 for (;;)
3793 {
3794 /* If necessary, advance to next segment in register
3795 contents. */
3796 while (d2 == dend2)
3797 {
3798 if (dend2 == end_match_2) break;
3799 if (dend2 == regend[regno]) break;
3800
3801 /* End of string1 => advance to string2. */
3802 d2 = string2;
3803 dend2 = regend[regno];
3804 }
3805 /* At end of register contents => success */
3806 if (d2 == dend2) break;
3807
3808 /* If necessary, advance to next segment in data. */
3809 PREFETCH ();
3810
3811 /* How many characters left in this segment to match. */
3812 mcnt = dend - d;
3813
3814 /* Want how many consecutive characters we can match in
3815 one shot, so, if necessary, adjust the count. */
3816 if (mcnt > dend2 - d2)
3817 mcnt = dend2 - d2;
3818
3819 /* Compare that many; failure if mismatch, else move
3820 past them. */
3821 if (translate
3822 ? bcmp_translate (d, d2, mcnt, translate)
3823 : bcmp (d, d2, mcnt))
3824 goto fail;
3825 d += mcnt, d2 += mcnt;
3826 }
3827 }
3828 break;
3829
3830
3831 /* begline matches the empty string at the beginning of the string
3832 (unless `not_bol' is set in `bufp'), and, if
3833 `newline_anchor' is set, after newlines. */
3834 case begline:
3835 DEBUG_PRINT1 ("EXECUTING begline.\n");
3836
3837 if (AT_STRINGS_BEG (d))
3838 {
3839 if (!bufp->not_bol) break;
3840 }
3841 else if (d[-1] == '\n' && bufp->newline_anchor)
3842 {
3843 break;
3844 }
3845 /* In all other cases, we fail. */
3846 goto fail;
3847
3848
3849 /* endline is the dual of begline. */
3850 case endline:
3851 DEBUG_PRINT1 ("EXECUTING endline.\n");
3852
3853 if (AT_STRINGS_END (d))
3854 {
3855 if (!bufp->not_eol) break;
3856 }
3857
3858 /* We have to ``prefetch'' the next character. */
3859 else if ((d == end1 ? *string2 : *d) == '\n'
3860 && bufp->newline_anchor)
3861 {
3862 break;
3863 }
3864 goto fail;
3865
3866
3867 /* Match at the very beginning of the data. */
3868 case begbuf:
3869 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
3870 if (AT_STRINGS_BEG (d))
3871 break;
3872 goto fail;
3873
3874
3875 /* Match at the very end of the data. */
3876 case endbuf:
3877 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
3878 if (AT_STRINGS_END (d))
3879 break;
3880 goto fail;
3881
3882
3883 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
3884 pushes NULL as the value for the string on the stack. Then
3885 `pop_failure_point' will keep the current value for the
3886 string, instead of restoring it. To see why, consider
3887 matching `foo\nbar' against `.*\n'. The .* matches the foo;
3888 then the . fails against the \n. But the next thing we want
3889 to do is match the \n against the \n; if we restored the
3890 string value, we would be back at the foo.
3891
3892 Because this is used only in specific cases, we don't need to
3893 check all the things that `on_failure_jump' does, to make
3894 sure the right things get saved on the stack. Hence we don't
3895 share its code. The only reason to push anything on the
3896 stack at all is that otherwise we would have to change
3897 `anychar's code to do something besides goto fail in this
3898 case; that seems worse than this. */
3899 case on_failure_keep_string_jump:
3900 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
3901
3902 EXTRACT_NUMBER_AND_INCR (mcnt, p);
3903 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
3904
3905 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
3906 break;
3907
3908
3909 /* Uses of on_failure_jump:
3910
3911 Each alternative starts with an on_failure_jump that points
3912 to the beginning of the next alternative. Each alternative
3913 except the last ends with a jump that in effect jumps past
3914 the rest of the alternatives. (They really jump to the
3915 ending jump of the following alternative, because tensioning
3916 these jumps is a hassle.)
3917
3918 Repeats start with an on_failure_jump that points past both
3919 the repetition text and either the following jump or
3920 pop_failure_jump back to this on_failure_jump. */
3921 case on_failure_jump:
3922 on_failure:
3923 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
3924
3925 EXTRACT_NUMBER_AND_INCR (mcnt, p);
3926 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
3927
3928 /* If this on_failure_jump comes right before a group (i.e.,
3929 the original * applied to a group), save the information
3930 for that group and all inner ones, so that if we fail back
3931 to this point, the group's information will be correct.
3932 For example, in \(a*\)*\1, we need the preceding group,
3933 and in \(\(a*\)b*\)\2, we need the inner group. */
3934
3935 /* We can't use `p' to check ahead because we push
3936 a failure point to `p + mcnt' after we do this. */
3937 p1 = p;
3938
3939 /* We need to skip no_op's before we look for the
3940 start_memory in case this on_failure_jump is happening as
3941 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
3942 against aba. */
3943 while (p1 < pend && (re_opcode_t) *p1 == no_op)
3944 p1++;
3945
3946 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
3947 {
3948 /* We have a new highest active register now. This will
3949 get reset at the start_memory we are about to get to,
3950 but we will have saved all the registers relevant to
3951 this repetition op, as described above. */
3952 highest_active_reg = *(p1 + 1) + *(p1 + 2);
3953 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
3954 lowest_active_reg = *(p1 + 1);
3955 }
3956
3957 DEBUG_PRINT1 (":\n");
3958 PUSH_FAILURE_POINT (p + mcnt, d, -2);
3959 break;
3960
3961
3962 /* A smart repeat ends with `maybe_pop_jump'.
3963 We change it to either `pop_failure_jump' or `jump'. */
3964 case maybe_pop_jump:
3965 EXTRACT_NUMBER_AND_INCR (mcnt, p);
3966 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
3967 {
3968 register unsigned char *p2 = p;
3969
3970 /* Compare the beginning of the repeat with what in the
3971 pattern follows its end. If we can establish that there
3972 is nothing that they would both match, i.e., that we
3973 would have to backtrack because of (as in, e.g., `a*a')
3974 then we can change to pop_failure_jump, because we'll
3975 never have to backtrack.
3976
3977 This is not true in the case of alternatives: in
3978 `(a|ab)*' we do need to backtrack to the `ab' alternative
3979 (e.g., if the string was `ab'). But instead of trying to
3980 detect that here, the alternative has put on a dummy
3981 failure point which is what we will end up popping. */
3982
3983 /* Skip over open/close-group commands. */
3984 while (p2 + 2 < pend
3985 && ((re_opcode_t) *p2 == stop_memory
3986 || (re_opcode_t) *p2 == start_memory))
3987 p2 += 3; /* Skip over args, too. */
3988
3989 /* If we're at the end of the pattern, we can change. */
3990 if (p2 == pend)
3991 {
3992 /* Consider what happens when matching ":\(.*\)"
3993 against ":/". I don't really understand this code
3994 yet. */
3995 p[-3] = (unsigned char) pop_failure_jump;
3996 DEBUG_PRINT1
3997 (" End of pattern: change to `pop_failure_jump'.\n");
3998 }
3999
4000 else if ((re_opcode_t) *p2 == exactn
4001 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4002 {
4003 register unsigned char c
4004 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4005 p1 = p + mcnt;
4006
4007 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4008 to the `maybe_finalize_jump' of this case. Examine what
4009 follows. */
4010 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4011 {
4012 p[-3] = (unsigned char) pop_failure_jump;
4013 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4014 c, p1[5]);
4015 }
4016
4017 else if ((re_opcode_t) p1[3] == charset
4018 || (re_opcode_t) p1[3] == charset_not)
4019 {
4020 int not = (re_opcode_t) p1[3] == charset_not;
4021
4022 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4023 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4024 not = !not;
4025
4026 /* `not' is equal to 1 if c would match, which means
4027 that we can't change to pop_failure_jump. */
4028 if (!not)
4029 {
4030 p[-3] = (unsigned char) pop_failure_jump;
4031 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4032 }
4033 }
4034 }
4035 }
4036 p -= 2; /* Point at relative address again. */
4037 if ((re_opcode_t) p[-1] != pop_failure_jump)
4038 {
4039 p[-1] = (unsigned char) jump;
4040 DEBUG_PRINT1 (" Match => jump.\n");
4041 goto unconditional_jump;
4042 }
4043 /* Note fall through. */
4044
4045
4046 /* The end of a simple repeat has a pop_failure_jump back to
4047 its matching on_failure_jump, where the latter will push a
4048 failure point. The pop_failure_jump takes off failure
4049 points put on by this pop_failure_jump's matching
4050 on_failure_jump; we got through the pattern to here from the
4051 matching on_failure_jump, so didn't fail. */
4052 case pop_failure_jump:
4053 {
4054 /* We need to pass separate storage for the lowest and
4055 highest registers, even though we don't care about the
4056 actual values. Otherwise, we will restore only one
4057 register from the stack, since lowest will == highest in
4058 `pop_failure_point'. */
4059 unsigned dummy_low_reg, dummy_high_reg;
4060 unsigned char *pdummy;
4061 char *sdummy;
4062
4063 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4064 POP_FAILURE_POINT (sdummy, pdummy,
4065 dummy_low_reg, dummy_high_reg,
4066 reg_dummy, reg_dummy, reg_info_dummy);
4067 }
4068 /* Note fall through. */
4069
4070
4071 /* Unconditionally jump (without popping any failure points). */
4072 case jump:
4073 unconditional_jump:
4074 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4075 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4076 p += mcnt; /* Do the jump. */
4077 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4078 break;
4079
4080
4081 /* We need this opcode so we can detect where alternatives end
4082 in `group_match_null_string_p' et al. */
4083 case jump_past_alt:
4084 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4085 goto unconditional_jump;
4086
4087
4088 /* Normally, the on_failure_jump pushes a failure point, which
4089 then gets popped at pop_failure_jump. We will end up at
4090 pop_failure_jump, also, and with a pattern of, say, `a+', we
4091 are skipping over the on_failure_jump, so we have to push
4092 something meaningless for pop_failure_jump to pop. */
4093 case dummy_failure_jump:
4094 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4095 /* It doesn't matter what we push for the string here. What
4096 the code at `fail' tests is the value for the pattern. */
4097 PUSH_FAILURE_POINT (0, 0, -2);
4098 goto unconditional_jump;
4099
4100
4101 /* At the end of an alternative, we need to push a dummy failure
4102 point in case we are followed by a `pop_failure_jump', because
4103 we don't want the failure point for the alternative to be
4104 popped. For example, matching `(a|ab)*' against `aab'
4105 requires that we match the `ab' alternative. */
4106 case push_dummy_failure:
4107 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4108 /* See comments just above at `dummy_failure_jump' about the
4109 two zeroes. */
4110 PUSH_FAILURE_POINT (0, 0, -2);
4111 break;
4112
4113 /* Have to succeed matching what follows at least n times.
4114 After that, handle like `on_failure_jump'. */
4115 case succeed_n:
4116 EXTRACT_NUMBER (mcnt, p + 2);
4117 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4118
4119 assert (mcnt >= 0);
4120 /* Originally, this is how many times we HAVE to succeed. */
4121 if (mcnt > 0)
4122 {
4123 mcnt--;
4124 p += 2;
4125 STORE_NUMBER_AND_INCR (p, mcnt);
4126 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt);
4127 }
4128 else if (mcnt == 0)
4129 {
4130 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4131 p[2] = (unsigned char) no_op;
4132 p[3] = (unsigned char) no_op;
4133 goto on_failure;
4134 }
4135 break;
4136
4137 case jump_n:
4138 EXTRACT_NUMBER (mcnt, p + 2);
4139 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4140
4141 /* Originally, this is how many times we CAN jump. */
4142 if (mcnt)
4143 {
4144 mcnt--;
4145 STORE_NUMBER (p + 2, mcnt);
4146 goto unconditional_jump;
4147 }
4148 /* If don't have to jump any more, skip over the rest of command. */
4149 else
4150 p += 4;
4151 break;
4152
4153 case set_number_at:
4154 {
4155 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4156
4157 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4158 p1 = p + mcnt;
4159 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4160 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4161 STORE_NUMBER (p1, mcnt);
4162 break;
4163 }
4164
4165 case wordbound:
4166 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4167 if (AT_WORD_BOUNDARY (d))
4168 break;
4169 goto fail;
4170
4171 case notwordbound:
4172 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4173 if (AT_WORD_BOUNDARY (d))
4174 goto fail;
4175 break;
4176
4177 case wordbeg:
4178 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
4179 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
4180 break;
4181 goto fail;
4182
4183 case wordend:
4184 DEBUG_PRINT1 ("EXECUTING wordend.\n");
4185 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
4186 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
4187 break;
4188 goto fail;
4189
4190 #ifdef emacs
4191 #ifdef emacs19
4192 case before_dot:
4193 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
4194 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
4195 goto fail;
4196 break;
4197
4198 case at_dot:
4199 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
4200 if (PTR_CHAR_POS ((unsigned char *) d) != point)
4201 goto fail;
4202 break;
4203
4204 case after_dot:
4205 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
4206 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
4207 goto fail;
4208 break;
4209 #else /* not emacs19 */
4210 case at_dot:
4211 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
4212 if (PTR_CHAR_POS ((unsigned char *) d) + 1 != point)
4213 goto fail;
4214 break;
4215 #endif /* not emacs19 */
4216
4217 case syntaxspec:
4218 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
4219 mcnt = *p++;
4220 goto matchsyntax;
4221
4222 case wordchar:
4223 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
4224 mcnt = (int) Sword;
4225 matchsyntax:
4226 PREFETCH ();
4227 if (SYNTAX (*d++) != (enum syntaxcode) mcnt)
4228 goto fail;
4229 SET_REGS_MATCHED ();
4230 break;
4231
4232 case notsyntaxspec:
4233 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
4234 mcnt = *p++;
4235 goto matchnotsyntax;
4236
4237 case notwordchar:
4238 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
4239 mcnt = (int) Sword;
4240 matchnotsyntax:
4241 PREFETCH ();
4242 if (SYNTAX (*d++) == (enum syntaxcode) mcnt)
4243 goto fail;
4244 SET_REGS_MATCHED ();
4245 break;
4246
4247 #else /* not emacs */
4248 case wordchar:
4249 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
4250 PREFETCH ();
4251 if (!WORDCHAR_P (d))
4252 goto fail;
4253 SET_REGS_MATCHED ();
4254 d++;
4255 break;
4256
4257 case notwordchar:
4258 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
4259 PREFETCH ();
4260 if (WORDCHAR_P (d))
4261 goto fail;
4262 SET_REGS_MATCHED ();
4263 d++;
4264 break;
4265 #endif /* not emacs */
4266
4267 default:
4268 abort ();
4269 }
4270 continue; /* Successfully executed one pattern command; keep going. */
4271
4272
4273 /* We goto here if a matching operation fails. */
4274 fail:
4275 if (!FAIL_STACK_EMPTY ())
4276 { /* A restart point is known. Restore to that state. */
4277 DEBUG_PRINT1 ("\nFAIL:\n");
4278 POP_FAILURE_POINT (d, p,
4279 lowest_active_reg, highest_active_reg,
4280 regstart, regend, reg_info);
4281
4282 /* If this failure point is a dummy, try the next one. */
4283 if (!p)
4284 goto fail;
4285
4286 /* If we failed to the end of the pattern, don't examine *p. */
4287 assert (p <= pend);
4288 if (p < pend)
4289 {
4290 boolean is_a_jump_n = false;
4291
4292 /* If failed to a backwards jump that's part of a repetition
4293 loop, need to pop this failure point and use the next one. */
4294 switch ((re_opcode_t) *p)
4295 {
4296 case jump_n:
4297 is_a_jump_n = true;
4298 case maybe_pop_jump:
4299 case pop_failure_jump:
4300 case jump:
4301 p1 = p + 1;
4302 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4303 p1 += mcnt;
4304
4305 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
4306 || (!is_a_jump_n
4307 && (re_opcode_t) *p1 == on_failure_jump))
4308 goto fail;
4309 break;
4310 default:
4311 /* do nothing */ ;
4312 }
4313 }
4314
4315 if (d >= string1 && d <= end1)
4316 dend = end_match_1;
4317 }
4318 else
4319 break; /* Matching at this starting point really fails. */
4320 } /* for (;;) */
4321
4322 if (best_regs_set)
4323 goto restore_best_regs;
4324
4325 FREE_VARIABLES ();
4326
4327 return -1; /* Failure to match. */
4328 } /* re_match_2 */
4329
4330 /* Subroutine definitions for re_match_2. */
4331
4332
4333 /* We are passed P pointing to a register number after a start_memory.
4334
4335 Return true if the pattern up to the corresponding stop_memory can
4336 match the empty string, and false otherwise.
4337
4338 If we find the matching stop_memory, sets P to point to one past its number.
4339 Otherwise, sets P to an undefined byte less than or equal to END.
4340
4341 We don't handle duplicates properly (yet). */
4342
4343 static boolean
group_match_null_string_p(p,end,reg_info)4344 group_match_null_string_p (p, end, reg_info)
4345 unsigned char **p, *end;
4346 register_info_type *reg_info;
4347 {
4348 int mcnt;
4349 /* Point to after the args to the start_memory. */
4350 unsigned char *p1 = *p + 2;
4351
4352 while (p1 < end)
4353 {
4354 /* Skip over opcodes that can match nothing, and return true or
4355 false, as appropriate, when we get to one that can't, or to the
4356 matching stop_memory. */
4357
4358 switch ((re_opcode_t) *p1)
4359 {
4360 /* Could be either a loop or a series of alternatives. */
4361 case on_failure_jump:
4362 p1++;
4363 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4364
4365 /* If the next operation is not a jump backwards in the
4366 pattern. */
4367
4368 if (mcnt >= 0)
4369 {
4370 /* Go through the on_failure_jumps of the alternatives,
4371 seeing if any of the alternatives cannot match nothing.
4372 The last alternative starts with only a jump,
4373 whereas the rest start with on_failure_jump and end
4374 with a jump, e.g., here is the pattern for `a|b|c':
4375
4376 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
4377 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
4378 /exactn/1/c
4379
4380 So, we have to first go through the first (n-1)
4381 alternatives and then deal with the last one separately. */
4382
4383
4384 /* Deal with the first (n-1) alternatives, which start
4385 with an on_failure_jump (see above) that jumps to right
4386 past a jump_past_alt. */
4387
4388 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
4389 {
4390 /* `mcnt' holds how many bytes long the alternative
4391 is, including the ending `jump_past_alt' and
4392 its number. */
4393
4394 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
4395 reg_info))
4396 return false;
4397
4398 /* Move to right after this alternative, including the
4399 jump_past_alt. */
4400 p1 += mcnt;
4401
4402 /* Break if it's the beginning of an n-th alternative
4403 that doesn't begin with an on_failure_jump. */
4404 if ((re_opcode_t) *p1 != on_failure_jump)
4405 break;
4406
4407 /* Still have to check that it's not an n-th
4408 alternative that starts with an on_failure_jump. */
4409 p1++;
4410 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4411 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
4412 {
4413 /* Get to the beginning of the n-th alternative. */
4414 p1 -= 3;
4415 break;
4416 }
4417 }
4418
4419 /* Deal with the last alternative: go back and get number
4420 of the `jump_past_alt' just before it. `mcnt' contains
4421 the length of the alternative. */
4422 EXTRACT_NUMBER (mcnt, p1 - 2);
4423
4424 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
4425 return false;
4426
4427 p1 += mcnt; /* Get past the n-th alternative. */
4428 } /* if mcnt > 0 */
4429 break;
4430
4431
4432 case stop_memory:
4433 assert (p1[1] == **p);
4434 *p = p1 + 2;
4435 return true;
4436
4437
4438 default:
4439 if (!common_op_match_null_string_p (&p1, end, reg_info))
4440 return false;
4441 }
4442 } /* while p1 < end */
4443
4444 return false;
4445 } /* group_match_null_string_p */
4446
4447
4448 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
4449 It expects P to be the first byte of a single alternative and END one
4450 byte past the last. The alternative can contain groups. */
4451
4452 static boolean
alt_match_null_string_p(p,end,reg_info)4453 alt_match_null_string_p (p, end, reg_info)
4454 unsigned char *p, *end;
4455 register_info_type *reg_info;
4456 {
4457 int mcnt;
4458 unsigned char *p1 = p;
4459
4460 while (p1 < end)
4461 {
4462 /* Skip over opcodes that can match nothing, and break when we get
4463 to one that can't. */
4464
4465 switch ((re_opcode_t) *p1)
4466 {
4467 /* It's a loop. */
4468 case on_failure_jump:
4469 p1++;
4470 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4471 p1 += mcnt;
4472 break;
4473
4474 default:
4475 if (!common_op_match_null_string_p (&p1, end, reg_info))
4476 return false;
4477 }
4478 } /* while p1 < end */
4479
4480 return true;
4481 } /* alt_match_null_string_p */
4482
4483
4484 /* Deals with the ops common to group_match_null_string_p and
4485 alt_match_null_string_p.
4486
4487 Sets P to one after the op and its arguments, if any. */
4488
4489 static boolean
common_op_match_null_string_p(p,end,reg_info)4490 common_op_match_null_string_p (p, end, reg_info)
4491 unsigned char **p, *end;
4492 register_info_type *reg_info;
4493 {
4494 int mcnt;
4495 boolean ret;
4496 int reg_no;
4497 unsigned char *p1 = *p;
4498
4499 switch ((re_opcode_t) *p1++)
4500 {
4501 case no_op:
4502 case begline:
4503 case endline:
4504 case begbuf:
4505 case endbuf:
4506 case wordbeg:
4507 case wordend:
4508 case wordbound:
4509 case notwordbound:
4510 #ifdef emacs
4511 case before_dot:
4512 case at_dot:
4513 case after_dot:
4514 #endif
4515 break;
4516
4517 case start_memory:
4518 reg_no = *p1;
4519 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
4520 ret = group_match_null_string_p (&p1, end, reg_info);
4521
4522 /* Have to set this here in case we're checking a group which
4523 contains a group and a back reference to it. */
4524
4525 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
4526 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
4527
4528 if (!ret)
4529 return false;
4530 break;
4531
4532 /* If this is an optimized succeed_n for zero times, make the jump. */
4533 case jump:
4534 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4535 if (mcnt >= 0)
4536 p1 += mcnt;
4537 else
4538 return false;
4539 break;
4540
4541 case succeed_n:
4542 /* Get to the number of times to succeed. */
4543 p1 += 2;
4544 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4545
4546 if (mcnt == 0)
4547 {
4548 p1 -= 4;
4549 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4550 p1 += mcnt;
4551 }
4552 else
4553 return false;
4554 break;
4555
4556 case duplicate:
4557 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
4558 return false;
4559 break;
4560
4561 case set_number_at:
4562 p1 += 4;
4563
4564 default:
4565 /* All other opcodes mean we cannot match the empty string. */
4566 return false;
4567 }
4568
4569 *p = p1;
4570 return true;
4571 } /* common_op_match_null_string_p */
4572
4573
4574 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
4575 bytes; nonzero otherwise. */
4576
4577 static int
bcmp_translate(s1,s2,len,translate)4578 bcmp_translate (s1, s2, len, translate)
4579 unsigned char *s1, *s2;
4580 register int len;
4581 char *translate;
4582 {
4583 register unsigned char *p1 = s1, *p2 = s2;
4584 while (len)
4585 {
4586 if (translate[*p1++] != translate[*p2++]) return 1;
4587 len--;
4588 }
4589 return 0;
4590 }
4591
4592 /* Entry points for GNU code. */
4593
4594 /* re_compile_pattern is the GNU regular expression compiler: it
4595 compiles PATTERN (of length SIZE) and puts the result in BUFP.
4596 Returns 0 if the pattern was valid, otherwise an error string.
4597
4598 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
4599 are set in BUFP on entry.
4600
4601 We call regex_compile to do the actual compilation. */
4602
4603 char *
re_compile_pattern(char * pattern,int length,struct re_pattern_buffer * bufp)4604 re_compile_pattern (char *pattern, int length, struct re_pattern_buffer *bufp)
4605 {
4606 reg_errcode_t ret;
4607
4608 /* GNU code is written to assume at least RE_NREGS registers will be set
4609 (and at least one extra will be -1). */
4610 bufp->regs_allocated = REGS_UNALLOCATED;
4611
4612 /* And GNU code determines whether or not to get register information
4613 by passing null for the REGS argument to re_match, etc., not by
4614 setting no_sub. */
4615 bufp->no_sub = 0;
4616
4617 /* Match anchors at newline. */
4618 bufp->newline_anchor = 1;
4619
4620 ret = regex_compile (pattern, length, re_syntax_options, bufp);
4621
4622 return re_error_msg[(int) ret];
4623 }
4624
4625 /* Entry points compatible with 4.2 BSD regex library. We don't define
4626 them if this is an Emacs or POSIX compilation. */
4627
4628 #if !defined (emacs) && !defined (_POSIX_SOURCE)
4629
4630 /* BSD has one and only one pattern buffer. */
4631 static struct re_pattern_buffer re_comp_buf;
4632
4633 char *
re_comp(s)4634 re_comp (s)
4635 char *s;
4636 {
4637 reg_errcode_t ret;
4638
4639 if (!s)
4640 {
4641 if (!re_comp_buf.buffer)
4642 return "No previous regular expression";
4643 return 0;
4644 }
4645
4646 if (!re_comp_buf.buffer)
4647 {
4648 re_comp_buf.buffer = (unsigned char *) malloc (200);
4649 if (re_comp_buf.buffer == NULL)
4650 return "Memory exhausted";
4651 re_comp_buf.allocated = 200;
4652
4653 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
4654 if (re_comp_buf.fastmap == NULL)
4655 return "Memory exhausted";
4656 }
4657
4658 /* Since `re_exec' always passes NULL for the `regs' argument, we
4659 don't need to initialize the pattern buffer fields which affect it. */
4660
4661 /* Match anchors at newlines. */
4662 re_comp_buf.newline_anchor = 1;
4663
4664 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
4665
4666 /* Yes, we're discarding `const' here. */
4667 return (char *) re_error_msg[(int) ret];
4668 }
4669
4670
4671 int
re_exec(s)4672 re_exec (s)
4673 char *s;
4674 {
4675 int len = strlen (s);
4676 return
4677 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
4678 }
4679 #endif /* not emacs and not _POSIX_SOURCE */
4680
4681 /* POSIX.2 functions. Don't define these for Emacs. */
4682
4683 #ifndef emacs
4684
4685 /* regcomp takes a regular expression as a string and compiles it.
4686
4687 PREG is a regex_t *. We do not expect any fields to be initialized,
4688 since POSIX says we shouldn't. Thus, we set
4689
4690 `buffer' to the compiled pattern;
4691 `used' to the length of the compiled pattern;
4692 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
4693 REG_EXTENDED bit in CFLAGS is set; otherwise, to
4694 RE_SYNTAX_POSIX_BASIC;
4695 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
4696 `fastmap' and `fastmap_accurate' to zero;
4697 `re_nsub' to the number of subexpressions in PATTERN.
4698
4699 PATTERN is the address of the pattern string.
4700
4701 CFLAGS is a series of bits which affect compilation.
4702
4703 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
4704 use POSIX basic syntax.
4705
4706 If REG_NEWLINE is set, then . and [^...] don't match newline.
4707 Also, regexec will try a match beginning after every newline.
4708
4709 If REG_ICASE is set, then we considers upper- and lowercase
4710 versions of letters to be equivalent when matching.
4711
4712 If REG_NOSUB is set, then when PREG is passed to regexec, that
4713 routine will report only success or failure, and nothing about the
4714 registers.
4715
4716 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
4717 the return codes and their meanings.) */
4718
4719 int
regcomp(preg,pattern,cflags)4720 regcomp (preg, pattern, cflags)
4721 regex_t *preg;
4722 char *pattern;
4723 int cflags;
4724 {
4725 reg_errcode_t ret;
4726 unsigned syntax
4727 = (cflags & REG_EXTENDED) ?
4728 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
4729
4730 /* regex_compile will allocate the space for the compiled pattern. */
4731 preg->buffer = 0;
4732 preg->allocated = 0;
4733
4734 /* Don't bother to use a fastmap when searching. This simplifies the
4735 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
4736 characters after newlines into the fastmap. This way, we just try
4737 every character. */
4738 preg->fastmap = 0;
4739
4740 if (cflags & REG_ICASE)
4741 {
4742 unsigned i;
4743
4744 preg->translate = (char *) malloc (CHAR_SET_SIZE);
4745 if (preg->translate == NULL)
4746 return (int) REG_ESPACE;
4747
4748 /* Map uppercase characters to corresponding lowercase ones. */
4749 for (i = 0; i < CHAR_SET_SIZE; i++)
4750 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
4751 }
4752 else
4753 preg->translate = NULL;
4754
4755 /* If REG_NEWLINE is set, newlines are treated differently. */
4756 if (cflags & REG_NEWLINE)
4757 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
4758 syntax &= ~RE_DOT_NEWLINE;
4759 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
4760 /* It also changes the matching behavior. */
4761 preg->newline_anchor = 1;
4762 }
4763 else
4764 preg->newline_anchor = 0;
4765
4766 preg->no_sub = !!(cflags & REG_NOSUB);
4767
4768 /* POSIX says a null character in the pattern terminates it, so we
4769 can use strlen here in compiling the pattern. */
4770 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
4771
4772 /* POSIX doesn't distinguish between an unmatched open-group and an
4773 unmatched close-group: both are REG_EPAREN. */
4774 if (ret == REG_ERPAREN) ret = REG_EPAREN;
4775
4776 return (int) ret;
4777 }
4778
4779
4780 /* regexec searches for a given pattern, specified by PREG, in the
4781 string STRING.
4782
4783 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
4784 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
4785 least NMATCH elements, and we set them to the offsets of the
4786 corresponding matched substrings.
4787
4788 EFLAGS specifies `execution flags' which affect matching: if
4789 REG_NOTBOL is set, then ^ does not match at the beginning of the
4790 string; if REG_NOTEOL is set, then $ does not match at the end.
4791
4792 We return 0 if we find a match and REG_NOMATCH if not. */
4793
4794 int
regexec(preg,string,nmatch,pmatch,eflags)4795 regexec (preg, string, nmatch, pmatch, eflags)
4796 regex_t *preg;
4797 char *string;
4798 size_t nmatch;
4799 regmatch_t pmatch[];
4800 int eflags;
4801 {
4802 int ret;
4803 struct re_registers regs;
4804 regex_t private_preg;
4805 int len = strlen (string);
4806 boolean want_reg_info = !preg->no_sub && nmatch > 0;
4807
4808 private_preg = *preg;
4809
4810 private_preg.not_bol = !!(eflags & REG_NOTBOL);
4811 private_preg.not_eol = !!(eflags & REG_NOTEOL);
4812
4813 /* The user has told us exactly how many registers to return
4814 information about, via `nmatch'. We have to pass that on to the
4815 matching routines. */
4816 private_preg.regs_allocated = REGS_FIXED;
4817
4818 if (want_reg_info)
4819 {
4820 regs.num_regs = nmatch;
4821 regs.start = TALLOC (nmatch, regoff_t);
4822 regs.end = TALLOC (nmatch, regoff_t);
4823 if (regs.start == NULL || regs.end == NULL)
4824 return (int) REG_NOMATCH;
4825 }
4826
4827 /* Perform the searching operation. */
4828 ret = re_search (&private_preg, string, len,
4829 /* start: */ 0, /* range: */ len,
4830 want_reg_info ? ®s : (struct re_registers *) 0);
4831
4832 /* Copy the register information to the POSIX structure. */
4833 if (want_reg_info)
4834 {
4835 if (ret >= 0)
4836 {
4837 unsigned r;
4838
4839 for (r = 0; r < nmatch; r++)
4840 {
4841 pmatch[r].rm_so = regs.start[r];
4842 pmatch[r].rm_eo = regs.end[r];
4843 }
4844 }
4845
4846 /* If we needed the temporary register info, free the space now. */
4847 free (regs.start);
4848 free (regs.end);
4849 }
4850
4851 /* We want zero return to mean success, unlike `re_search'. */
4852 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
4853 }
4854
4855
4856 /* Returns a message corresponding to an error code, ERRCODE, returned
4857 from either regcomp or regexec. We don't use PREG here. */
4858
4859 size_t
regerror(errcode,preg,errbuf,errbuf_size)4860 regerror (errcode, preg, errbuf, errbuf_size)
4861 int errcode;
4862 regex_t *preg;
4863 char *errbuf;
4864 size_t errbuf_size;
4865 {
4866 char *msg;
4867 size_t msg_size;
4868
4869 if (errcode < 0
4870 || errcode >= (sizeof (re_error_msg) / sizeof (re_error_msg[0])))
4871 /* Only error codes returned by the rest of the code should be passed
4872 to this routine. If we are given anything else, or if other regex
4873 code generates an invalid error code, then the program has a bug.
4874 Dump core so we can fix it. */
4875 abort ();
4876
4877 msg = re_error_msg[errcode];
4878
4879 /* POSIX doesn't require that we do anything in this case, but why
4880 not be nice. */
4881 if (! msg)
4882 msg = "Success";
4883
4884 msg_size = strlen (msg) + 1; /* Includes the null. */
4885
4886 if (errbuf_size != 0)
4887 {
4888 if (msg_size > errbuf_size)
4889 {
4890 strncpy (errbuf, msg, errbuf_size - 1);
4891 errbuf[errbuf_size - 1] = 0;
4892 }
4893 else
4894 strcpy (errbuf, msg);
4895 }
4896
4897 return msg_size;
4898 }
4899
4900
4901 /* Free dynamically allocated space used by PREG. */
4902
4903 void
regfree(preg)4904 regfree (preg)
4905 regex_t *preg;
4906 {
4907 if (preg->buffer != NULL)
4908 free (preg->buffer);
4909 preg->buffer = NULL;
4910
4911 preg->allocated = 0;
4912 preg->used = 0;
4913
4914 if (preg->fastmap != NULL)
4915 free (preg->fastmap);
4916 preg->fastmap = NULL;
4917 preg->fastmap_accurate = 0;
4918
4919 if (preg->translate != NULL)
4920 free (preg->translate);
4921 preg->translate = NULL;
4922 }
4923
4924 #endif /* not emacs */
4925
4926 /*
4927 Local variables:
4928 make-backup-files: t
4929 version-control: t
4930 trim-versions-without-asking: nil
4931 End:
4932 */
4933 #endif /* not lint */
4934