1 /* Extended regular expression matching and search library,
3 * (Implements POSIX draft P1003.2/D11.2, except for some of the
4 * internationalization features.)
6 * Copyright (C) 1993, 1994, 1995, 1996, 1997 Free Software Foundation, Inc.
8 * This file is part of the GNU C Library. Its master source is NOT part of
9 * the C library, however. The master source lives in /gd/gnu/lib.
11 * The GNU C Library is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Library General Public License as
13 * published by the Free Software Foundation; either version 2 of the
14 * License, or (at your option) any later version.
16 * The GNU C Library is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Library General Public License for more details.
21 * You should have received a copy of the GNU Library General Public
22 * License along with the GNU C Library; see the file COPYING.LIB. If not,
23 * write to the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
24 * Boston, MA 02110-1301, USA.
30 * Use _regex.h instead of regex.h. tlr, 1999-01-06
31 * Make REGEX_MALLOC depend on HAVE_ALLOCA &c.
33 * Don't switch on regex debugging when debugging mutt.
37 /* The following doesn't mix too well with autoconfiguring
38 * the use of alloca. So let's disable it for AIX.
43 /* AIX requires this to be the first thing in the file. */
44 # if defined (_AIX) && !defined (REGEX_MALLOC)
59 /* On OS X 10.5.x, wide char functions are inlined by default breaking
60 * --without-wc-funcs compilation
63 #define _DONT_USE_CTYPE_INLINE_
66 #if (defined(HAVE_ALLOCA_H) && !defined(_AIX))
70 #if (!defined(HAVE_ALLOCA) || defined(_AIX))
74 #if defined(STDC_HEADERS) && !defined(emacs)
77 /* We need this for `regex.h', and perhaps for the Emacs include files. */
78 #include <sys/types.h>
81 /* For platform which support the ISO C amendement 1 functionality we
82 support user defined character classes. */
86 #if defined(HAVE_WCTYPE_H) && defined(HAVE_WC_FUNCS)
90 /* This is for other GNU distributions with internationalized messages. */
91 #if HAVE_LIBINTL_H || defined (_LIBC)
94 # define gettext(msgid) (msgid)
98 /* This define is so xgettext can find the internationalizable
100 #define gettext_noop(String) String
103 /* The `emacs' switch turns on certain matching commands
104 that make sense only in Emacs. */
111 #else /* not emacs */
113 /* If we are not linking with Emacs proper,
114 we can't use the relocating allocator
115 even if config.h says that we can. */
118 #if defined (STDC_HEADERS) || defined (_LIBC)
121 char *malloc (); /* __MEM_CHECKED__ */
122 char *realloc (); /* __MEM_CHECKED__ */
125 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
126 If nothing else has been done, use the method below. */
127 #ifdef INHIBIT_STRING_HEADER
128 #if !(defined (HAVE_BZERO) && defined (HAVE_BCOPY))
129 #if !defined (bzero) && !defined (bcopy)
130 #undef INHIBIT_STRING_HEADER
135 /* This is the normal way of making sure we have a bcopy and a bzero.
136 This is used in most programs--a few other programs avoid this
137 by defining INHIBIT_STRING_HEADER. */
138 #ifndef INHIBIT_STRING_HEADER
139 #if defined (HAVE_STRING_H) || defined (STDC_HEADERS) || defined (_LIBC)
142 #define bcmp(s1, s2, n) memcmp ((s1), (s2), (n))
145 #define bcopy(s, d, n) memcpy ((d), (s), (n))
148 #define bzero(s, n) memset ((s), 0, (n))
155 /* Define the syntax stuff for \<, \>, etc. */
157 /* This must be nonzero for the wordchar and notwordchar pattern
158 commands in re_match_2. */
163 #ifdef SWITCH_ENUM_BUG
164 #define SWITCH_ENUM_CAST(x) ((int)(x))
166 #define SWITCH_ENUM_CAST(x) (x)
171 extern char *re_syntax_table;
173 #else /* not SYNTAX_TABLE */
175 /* How many characters in the character set. */
176 #define CHAR_SET_SIZE 256
178 static char re_syntax_table[CHAR_SET_SIZE];
189 bzero (re_syntax_table, sizeof re_syntax_table);
191 for (c = 'a'; c <= 'z'; c++)
192 re_syntax_table[c] = Sword;
194 for (c = 'A'; c <= 'Z'; c++)
195 re_syntax_table[c] = Sword;
197 for (c = '0'; c <= '9'; c++)
198 re_syntax_table[c] = Sword;
200 re_syntax_table['_'] = Sword;
205 #endif /* not SYNTAX_TABLE */
207 #define SYNTAX(c) re_syntax_table[c]
209 #endif /* not emacs */
211 /* Get the interface, including the syntax bits. */
213 /* Changed to fit into mutt - tlr, 1999-01-06 */
217 /* isalpha etc. are used for the character classes. */
220 /* Jim Meyering writes:
222 "... Some ctype macros are valid only for character codes that
223 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
224 using /bin/cc or gcc but without giving an ansi option). So, all
225 ctype uses should be through macros like ISPRINT... If
226 STDC_HEADERS is defined, then autoconf has verified that the ctype
227 macros don't need to be guarded with references to isascii. ...
228 Defining isascii to 1 should let any compiler worth its salt
229 eliminate the && through constant folding." */
231 #if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII))
234 #define ISASCII(c) isascii(c)
238 #define ISBLANK(c) (ISASCII (c) && isblank (c))
240 #define ISBLANK(c) ((c) == ' ' || (c) == '\t')
243 #define ISGRAPH(c) (ISASCII (c) && isgraph (c))
245 #define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
248 #define ISPRINT(c) (ISASCII (c) && isprint (c))
249 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
250 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
251 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
252 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
253 #define ISLOWER(c) (ISASCII (c) && islower (c))
254 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
255 #define ISSPACE(c) (ISASCII (c) && isspace (c))
256 #define ISUPPER(c) (ISASCII (c) && isupper (c))
257 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
260 #define NULL (void *)0
263 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
264 since ours (we hope) works properly with all combinations of
265 machines, compilers, `char' and `unsigned char' argument types.
266 (Per Bothner suggested the basic approach.) */
267 #undef SIGN_EXTEND_CHAR
269 #define SIGN_EXTEND_CHAR(c) ((signed char) (c))
270 #else /* not __STDC__ */
271 /* As in Harbison and Steele. */
272 #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
275 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
276 use `alloca' instead of `malloc'. This is because using malloc in
277 re_search* or re_match* could cause memory leaks when C-g is used in
278 Emacs; also, malloc is slower and causes storage fragmentation. On
279 the other hand, malloc is more portable, and easier to debug.
281 Because we sometimes use alloca, some routines have to be macros,
282 not functions -- `alloca'-allocated space disappears at the end of the
283 function it is called in. */
287 #define REGEX_ALLOCATE malloc
288 #define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
289 #define REGEX_FREE free
291 #else /* not REGEX_MALLOC */
293 /* Emacs already defines alloca, sometimes. */
296 /* Make alloca work the best possible way. */
298 #define alloca __builtin_alloca
299 #else /* not __GNUC__ */
302 #else /* not __GNUC__ or HAVE_ALLOCA_H */
303 #if 0 /* It is a bad idea to declare alloca. We always cast the result. */
304 #ifndef _AIX /* Already did AIX, up at the top. */
306 #endif /* not _AIX */
308 #endif /* not HAVE_ALLOCA_H */
309 #endif /* not __GNUC__ */
311 #endif /* not alloca */
313 #define REGEX_ALLOCATE alloca
315 /* Assumes a `char *destination' variable. */
316 #define REGEX_REALLOCATE(source, osize, nsize) \
317 (destination = (char *) alloca (nsize), \
318 bcopy (source, destination, osize), \
321 /* No need to do anything to free, after alloca. */
322 #define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
324 #endif /* not REGEX_MALLOC */
326 /* Define how to allocate the failure stack. */
328 #if defined (REL_ALLOC) && defined (REGEX_MALLOC)
330 #define REGEX_ALLOCATE_STACK(size) \
331 r_alloc (&failure_stack_ptr, (size))
332 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
333 r_re_alloc (&failure_stack_ptr, (nsize))
334 #define REGEX_FREE_STACK(ptr) \
335 r_alloc_free (&failure_stack_ptr)
337 #else /* not using relocating allocator */
341 #define REGEX_ALLOCATE_STACK malloc
342 #define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
343 #define REGEX_FREE_STACK free
345 #else /* not REGEX_MALLOC */
347 #define REGEX_ALLOCATE_STACK alloca
349 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
350 REGEX_REALLOCATE (source, osize, nsize)
351 /* No need to explicitly free anything. */
352 #define REGEX_FREE_STACK(arg)
354 #endif /* not REGEX_MALLOC */
355 #endif /* not using relocating allocator */
358 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
359 `string1' or just past its end. This works if PTR is NULL, which is
361 #define FIRST_STRING_P(ptr) \
362 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
364 /* (Re)Allocate N items of type T using malloc, or fail. */
365 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
366 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
367 #define RETALLOC_IF(addr, n, t) \
368 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
369 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
371 #define BYTEWIDTH 8 /* In bits. */
373 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
377 #define MAX(a, b) ((a) > (b) ? (a) : (b))
378 #define MIN(a, b) ((a) < (b) ? (a) : (b))
380 typedef char boolean;
384 static int re_match_2_internal ();
386 /* These are the command codes that appear in compiled regular
387 expressions. Some opcodes are followed by argument bytes. A
388 command code can specify any interpretation whatsoever for its
389 arguments. Zero bytes may appear in the compiled regular expression. */
395 /* Succeed right away--no more backtracking. */
398 /* Followed by one byte giving n, then by n literal bytes. */
401 /* Matches any (more or less) character. */
404 /* Matches any one char belonging to specified set. First
405 following byte is number of bitmap bytes. Then come bytes
406 for a bitmap saying which chars are in. Bits in each byte
407 are ordered low-bit-first. A character is in the set if its
408 bit is 1. A character too large to have a bit in the map is
409 automatically not in the set. */
412 /* Same parameters as charset, but match any character that is
413 not one of those specified. */
416 /* Start remembering the text that is matched, for storing in a
417 register. Followed by one byte with the register number, in
418 the range 0 to one less than the pattern buffer's re_nsub
419 field. Then followed by one byte with the number of groups
420 inner to this one. (This last has to be part of the
421 start_memory only because we need it in the on_failure_jump
425 /* Stop remembering the text that is matched and store it in a
426 memory register. Followed by one byte with the register
427 number, in the range 0 to one less than `re_nsub' in the
428 pattern buffer, and one byte with the number of inner groups,
429 just like `start_memory'. (We need the number of inner
430 groups here because we don't have any easy way of finding the
431 corresponding start_memory when we're at a stop_memory.) */
434 /* Match a duplicate of something remembered. Followed by one
435 byte containing the register number. */
438 /* Fail unless at beginning of line. */
441 /* Fail unless at end of line. */
444 /* Succeeds if at beginning of buffer (if emacs) or at beginning
445 of string to be matched (if not). */
448 /* Analogously, for end of buffer/string. */
451 /* Followed by two byte relative address to which to jump. */
454 /* Same as jump, but marks the end of an alternative. */
457 /* Followed by two-byte relative address of place to resume at
458 in case of failure. */
461 /* Like on_failure_jump, but pushes a placeholder instead of the
462 current string position when executed. */
463 on_failure_keep_string_jump,
465 /* Throw away latest failure point and then jump to following
466 two-byte relative address. */
469 /* Change to pop_failure_jump if know won't have to backtrack to
470 match; otherwise change to jump. This is used to jump
471 back to the beginning of a repeat. If what follows this jump
472 clearly won't match what the repeat does, such that we can be
473 sure that there is no use backtracking out of repetitions
474 already matched, then we change it to a pop_failure_jump.
475 Followed by two-byte address. */
478 /* Jump to following two-byte address, and push a dummy failure
479 point. This failure point will be thrown away if an attempt
480 is made to use it for a failure. A `+' construct makes this
481 before the first repeat. Also used as an intermediary kind
482 of jump when compiling an alternative. */
485 /* Push a dummy failure point and continue. Used at the end of
489 /* Followed by two-byte relative address and two-byte number n.
490 After matching N times, jump to the address upon failure. */
493 /* Followed by two-byte relative address, and two-byte number n.
494 Jump to the address N times, then fail. */
497 /* Set the following two-byte relative address to the
498 subsequent two-byte number. The address *includes* the two
502 wordchar, /* Matches any word-constituent character. */
503 notwordchar, /* Matches any char that is not a word-constituent. */
505 wordbeg, /* Succeeds if at word beginning. */
506 wordend, /* Succeeds if at word end. */
508 wordbound, /* Succeeds if at a word boundary. */
509 notwordbound /* Succeeds if not at a word boundary. */
512 ,before_dot, /* Succeeds if before point. */
513 at_dot, /* Succeeds if at point. */
514 after_dot, /* Succeeds if after point. */
516 /* Matches any character whose syntax is specified. Followed by
517 a byte which contains a syntax code, e.g., Sword. */
520 /* Matches any character whose syntax is not that specified. */
525 /* Common operations on the compiled pattern. */
527 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
529 #define STORE_NUMBER(destination, number) \
531 (destination)[0] = (number) & 0377; \
532 (destination)[1] = (number) >> 8; \
535 /* Same as STORE_NUMBER, except increment DESTINATION to
536 the byte after where the number is stored. Therefore, DESTINATION
537 must be an lvalue. */
539 #define STORE_NUMBER_AND_INCR(destination, number) \
541 STORE_NUMBER (destination, number); \
542 (destination) += 2; \
545 /* Put into DESTINATION a number stored in two contiguous bytes starting
548 #define EXTRACT_NUMBER(destination, source) \
550 (destination) = *(source) & 0377; \
551 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
555 static void extract_number _RE_ARGS ((int *dest, unsigned char *source));
557 extract_number (dest, source)
559 unsigned char *source;
561 int temp = SIGN_EXTEND_CHAR (*(source + 1));
562 *dest = *source & 0377;
566 #ifndef EXTRACT_MACROS /* To debug the macros. */
567 #undef EXTRACT_NUMBER
568 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
569 #endif /* not EXTRACT_MACROS */
573 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
574 SOURCE must be an lvalue. */
576 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
578 EXTRACT_NUMBER (destination, source); \
583 static void extract_number_and_incr _RE_ARGS ((int *destination,
584 unsigned char **source));
586 extract_number_and_incr (destination, source)
588 unsigned char **source;
590 extract_number (destination, *source);
594 #ifndef EXTRACT_MACROS
595 #undef EXTRACT_NUMBER_AND_INCR
596 #define EXTRACT_NUMBER_AND_INCR(dest, src) \
597 extract_number_and_incr (&dest, &src)
598 #endif /* not EXTRACT_MACROS */
602 /* If DEBUG is defined, Regex prints many voluminous messages about what
603 it is doing (if the variable `debug' is nonzero). If linked with the
604 main program in `iregex.c', you can enter patterns and strings
605 interactively. And if linked with the main program in `main.c' and
606 the other test files, you can run the already-written tests. */
610 /* We use standard I/O for debugging. */
613 /* It is useful to test things that ``must'' be true when debugging. */
616 static int debug = 0;
618 #define DEBUG_STATEMENT(e) e
619 #define DEBUG_PRINT1(x) if (debug) printf (x)
620 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
621 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
622 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
623 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
624 if (debug) print_partial_compiled_pattern (s, e)
625 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
626 if (debug) print_double_string (w, s1, sz1, s2, sz2)
629 /* Print the fastmap in human-readable form. */
632 print_fastmap (fastmap)
635 unsigned was_a_range = 0;
638 while (i < (1 << BYTEWIDTH))
644 while (i < (1 << BYTEWIDTH) && fastmap[i])
660 /* Print a compiled pattern string in human-readable form, starting at
661 the START pointer into it and ending just before the pointer END. */
664 print_partial_compiled_pattern (start, end)
665 unsigned char *start;
670 unsigned char *p = start;
671 unsigned char *pend = end;
679 /* Loop over pattern commands. */
682 printf ("%d:\t", p - start);
684 switch ((re_opcode_t) *p++)
692 printf ("/exactn/%d", mcnt);
703 printf ("/start_memory/%d/%d", mcnt, *p++);
708 printf ("/stop_memory/%d/%d", mcnt, *p++);
712 printf ("/duplicate/%d", *p++);
722 register int c, last = -100;
723 register int in_range = 0;
725 printf ("/charset [%s",
726 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
728 assert (p + *p < pend);
730 for (c = 0; c < 256; c++)
732 && (p[1 + (c/8)] & (1 << (c % 8))))
734 /* Are we starting a range? */
735 if (last + 1 == c && ! in_range)
740 /* Have we broken a range? */
741 else if (last + 1 != c && in_range)
770 case on_failure_jump:
771 extract_number_and_incr (&mcnt, &p);
772 printf ("/on_failure_jump to %d", p + mcnt - start);
775 case on_failure_keep_string_jump:
776 extract_number_and_incr (&mcnt, &p);
777 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
780 case dummy_failure_jump:
781 extract_number_and_incr (&mcnt, &p);
782 printf ("/dummy_failure_jump to %d", p + mcnt - start);
785 case push_dummy_failure:
786 printf ("/push_dummy_failure");
790 extract_number_and_incr (&mcnt, &p);
791 printf ("/maybe_pop_jump to %d", p + mcnt - start);
794 case pop_failure_jump:
795 extract_number_and_incr (&mcnt, &p);
796 printf ("/pop_failure_jump to %d", p + mcnt - start);
800 extract_number_and_incr (&mcnt, &p);
801 printf ("/jump_past_alt to %d", p + mcnt - start);
805 extract_number_and_incr (&mcnt, &p);
806 printf ("/jump to %d", p + mcnt - start);
810 extract_number_and_incr (&mcnt, &p);
812 extract_number_and_incr (&mcnt2, &p);
813 printf ("/succeed_n to %d, %d times", p1 - start, mcnt2);
817 extract_number_and_incr (&mcnt, &p);
819 extract_number_and_incr (&mcnt2, &p);
820 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
824 extract_number_and_incr (&mcnt, &p);
826 extract_number_and_incr (&mcnt2, &p);
827 printf ("/set_number_at location %d to %d", p1 - start, mcnt2);
831 printf ("/wordbound");
835 printf ("/notwordbound");
847 printf ("/before_dot");
855 printf ("/after_dot");
859 printf ("/syntaxspec");
861 printf ("/%d", mcnt);
865 printf ("/notsyntaxspec");
867 printf ("/%d", mcnt);
872 printf ("/wordchar");
876 printf ("/notwordchar");
888 printf ("?%d", *(p-1));
894 printf ("%d:\tend of pattern.\n", p - start);
899 print_compiled_pattern (bufp)
900 struct re_pattern_buffer *bufp;
902 unsigned char *buffer = bufp->buffer;
904 print_partial_compiled_pattern (buffer, buffer + bufp->used);
905 printf ("%ld bytes used/%ld bytes allocated.\n",
906 bufp->used, bufp->allocated);
908 if (bufp->fastmap_accurate && bufp->fastmap)
910 printf ("fastmap: ");
911 print_fastmap (bufp->fastmap);
914 printf ("re_nsub: %d\t", bufp->re_nsub);
915 printf ("regs_alloc: %d\t", bufp->regs_allocated);
916 printf ("can_be_null: %d\t", bufp->can_be_null);
917 printf ("newline_anchor: %d\n", bufp->newline_anchor);
918 printf ("no_sub: %d\t", bufp->no_sub);
919 printf ("not_bol: %d\t", bufp->not_bol);
920 printf ("not_eol: %d\t", bufp->not_eol);
921 printf ("syntax: %lx\n", bufp->syntax);
922 /* Perhaps we should print the translate table? */
927 print_double_string (where, string1, size1, string2, size2)
940 if (FIRST_STRING_P (where))
942 for (this_char = where - string1; this_char < size1; this_char++)
943 putchar (string1[this_char]);
948 for (this_char = where - string2; this_char < size2; this_char++)
949 putchar (string2[this_char]);
960 #else /* not DEBUG */
965 #define DEBUG_STATEMENT(e)
966 #define DEBUG_PRINT1(x)
967 #define DEBUG_PRINT2(x1, x2)
968 #define DEBUG_PRINT3(x1, x2, x3)
969 #define DEBUG_PRINT4(x1, x2, x3, x4)
970 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
971 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
973 #endif /* not DEBUG */
975 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
976 also be assigned to arbitrarily: each pattern buffer stores its own
977 syntax, so it can be changed between regex compilations. */
978 /* This has no initializer because initialized variables in Emacs
979 become read-only after dumping. */
980 reg_syntax_t re_syntax_options;
983 /* Specify the precise syntax of regexps for compilation. This provides
984 for compatibility for various utilities which historically have
985 different, incompatible syntaxes.
987 The argument SYNTAX is a bit mask comprised of the various bits
988 defined in regex.h. We return the old syntax. */
991 re_set_syntax (syntax)
994 reg_syntax_t ret = re_syntax_options;
996 re_syntax_options = syntax;
998 if (syntax & RE_DEBUG)
1000 else if (debug) /* was on but now is not */
1006 /* This table gives an error message for each of the error codes listed
1007 in regex.h. Obviously the order here has to be same as there.
1008 POSIX doesn't require that we do anything for REG_NOERROR,
1009 but why not be nice? */
1011 static const char *re_error_msgid[] =
1013 gettext_noop ("Success"), /* REG_NOERROR */
1014 gettext_noop ("No match"), /* REG_NOMATCH */
1015 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
1016 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
1017 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
1018 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
1019 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
1020 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
1021 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
1022 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
1023 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
1024 gettext_noop ("Invalid range end"), /* REG_ERANGE */
1025 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
1026 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
1027 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
1028 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
1029 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
1032 /* Avoiding alloca during matching, to placate r_alloc. */
1034 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1035 searching and matching functions should not call alloca. On some
1036 systems, alloca is implemented in terms of malloc, and if we're
1037 using the relocating allocator routines, then malloc could cause a
1038 relocation, which might (if the strings being searched are in the
1039 ralloc heap) shift the data out from underneath the regexp
1042 Here's another reason to avoid allocation: Emacs
1043 processes input from X in a signal handler; processing X input may
1044 call malloc; if input arrives while a matching routine is calling
1045 malloc, then we're scrod. But Emacs can't just block input while
1046 calling matching routines; then we don't notice interrupts when
1047 they come in. So, Emacs blocks input around all regexp calls
1048 except the matching calls, which it leaves unprotected, in the
1049 faith that they will not malloc. */
1051 /* Normally, this is fine. */
1052 #define MATCH_MAY_ALLOCATE
1054 /* When using GNU C, we are not REALLY using the C alloca, no matter
1055 what config.h may say. So don't take precautions for it. */
1060 /* The match routines may not allocate if (1) they would do it with malloc
1061 and (2) it's not safe for them to use malloc.
1062 Note that if REL_ALLOC is defined, matching would not use malloc for the
1063 failure stack, but we would still use it for the register vectors;
1064 so REL_ALLOC should not affect this. */
1065 #if (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (emacs)
1066 #undef MATCH_MAY_ALLOCATE
1070 /* Failure stack declarations and macros; both re_compile_fastmap and
1071 re_match_2 use a failure stack. These have to be macros because of
1072 REGEX_ALLOCATE_STACK. */
1075 /* Number of failure points for which to initially allocate space
1076 when matching. If this number is exceeded, we allocate more
1077 space, so it is not a hard limit. */
1078 #ifndef INIT_FAILURE_ALLOC
1079 #define INIT_FAILURE_ALLOC 5
1082 /* Roughly the maximum number of failure points on the stack. Would be
1083 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1084 This is a variable only so users of regex can assign to it; we never
1085 change it ourselves. */
1089 #if defined (MATCH_MAY_ALLOCATE)
1090 /* 4400 was enough to cause a crash on Alpha OSF/1,
1091 whose default stack limit is 2mb. */
1092 long int re_max_failures = 4000;
1094 long int re_max_failures = 2000;
1097 union fail_stack_elt
1099 unsigned char *pointer;
1103 typedef union fail_stack_elt fail_stack_elt_t;
1107 fail_stack_elt_t *stack;
1108 unsigned long int size;
1109 unsigned long int avail; /* Offset of next open position. */
1112 #else /* not INT_IS_16BIT */
1114 #if defined (MATCH_MAY_ALLOCATE)
1115 /* 4400 was enough to cause a crash on Alpha OSF/1,
1116 whose default stack limit is 2mb. */
1117 int re_max_failures = 20000;
1119 int re_max_failures = 2000;
1122 union fail_stack_elt
1124 unsigned char *pointer;
1128 typedef union fail_stack_elt fail_stack_elt_t;
1132 fail_stack_elt_t *stack;
1134 unsigned avail; /* Offset of next open position. */
1137 #endif /* INT_IS_16BIT */
1139 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1140 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1141 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1144 /* Define macros to initialize and free the failure stack.
1145 Do `return -2' if the alloc fails. */
1147 #ifdef MATCH_MAY_ALLOCATE
1148 #define INIT_FAIL_STACK() \
1150 fail_stack.stack = (fail_stack_elt_t *) \
1151 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1153 if (fail_stack.stack == NULL) \
1156 fail_stack.size = INIT_FAILURE_ALLOC; \
1157 fail_stack.avail = 0; \
1160 #define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1162 #define INIT_FAIL_STACK() \
1164 fail_stack.avail = 0; \
1167 #define RESET_FAIL_STACK()
1171 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1173 Return 1 if succeeds, and 0 if either ran out of memory
1174 allocating space for it or it was already too large.
1176 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1178 #define DOUBLE_FAIL_STACK(fail_stack) \
1179 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1181 : ((fail_stack).stack = (fail_stack_elt_t *) \
1182 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1183 (fail_stack).size * sizeof (fail_stack_elt_t), \
1184 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1186 (fail_stack).stack == NULL \
1188 : ((fail_stack).size <<= 1, \
1192 /* Push pointer POINTER on FAIL_STACK.
1193 Return 1 if was able to do so and 0 if ran out of memory allocating
1195 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1196 ((FAIL_STACK_FULL () \
1197 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1199 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1202 /* Push a pointer value onto the failure stack.
1203 Assumes the variable `fail_stack'. Probably should only
1204 be called from within `PUSH_FAILURE_POINT'. */
1205 #define PUSH_FAILURE_POINTER(item) \
1206 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1208 /* This pushes an integer-valued item onto the failure stack.
1209 Assumes the variable `fail_stack'. Probably should only
1210 be called from within `PUSH_FAILURE_POINT'. */
1211 #define PUSH_FAILURE_INT(item) \
1212 fail_stack.stack[fail_stack.avail++].integer = (item)
1214 /* Push a fail_stack_elt_t value onto the failure stack.
1215 Assumes the variable `fail_stack'. Probably should only
1216 be called from within `PUSH_FAILURE_POINT'. */
1217 #define PUSH_FAILURE_ELT(item) \
1218 fail_stack.stack[fail_stack.avail++] = (item)
1220 /* These three POP... operations complement the three PUSH... operations.
1221 All assume that `fail_stack' is nonempty. */
1222 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1223 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1224 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1226 /* Used to omit pushing failure point id's when we're not debugging. */
1228 #define DEBUG_PUSH PUSH_FAILURE_INT
1229 #define DEBUG_POP(item_addr) (item_addr)->integer = POP_FAILURE_INT ()
1231 #define DEBUG_PUSH(item)
1232 #define DEBUG_POP(item_addr)
1236 /* Push the information about the state we will need
1237 if we ever fail back to it.
1239 Requires variables fail_stack, regstart, regend, reg_info, and
1240 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be
1243 Does `return FAILURE_CODE' if runs out of memory. */
1245 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1247 char *destination; \
1248 /* Must be int, so when we don't save any registers, the arithmetic \
1249 of 0 + -1 isn't done as unsigned. */ \
1250 /* Can't be int, since there is not a shred of a guarantee that int \
1251 is wide enough to hold a value of something to which pointer can \
1255 DEBUG_STATEMENT (failure_id++); \
1256 DEBUG_STATEMENT (nfailure_points_pushed++); \
1257 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1258 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1259 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1261 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \
1262 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1264 /* Ensure we have enough space allocated for what we will push. */ \
1265 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1267 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1268 return failure_code; \
1270 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1271 (fail_stack).size); \
1272 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1275 /* Push the info, starting with the registers. */ \
1276 DEBUG_PRINT1 ("\n"); \
1279 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1282 DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \
1283 DEBUG_STATEMENT (num_regs_pushed++); \
1285 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1286 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1288 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1289 PUSH_FAILURE_POINTER (regend[this_reg]); \
1291 DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \
1292 DEBUG_PRINT2 (" match_null=%d", \
1293 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1294 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1295 DEBUG_PRINT2 (" matched_something=%d", \
1296 MATCHED_SOMETHING (reg_info[this_reg])); \
1297 DEBUG_PRINT2 (" ever_matched=%d", \
1298 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1299 DEBUG_PRINT1 ("\n"); \
1300 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1303 DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\
1304 PUSH_FAILURE_INT (lowest_active_reg); \
1306 DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\
1307 PUSH_FAILURE_INT (highest_active_reg); \
1309 DEBUG_PRINT2 (" Pushing pattern 0x%x:\n", pattern_place); \
1310 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1311 PUSH_FAILURE_POINTER (pattern_place); \
1313 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \
1314 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1316 DEBUG_PRINT1 ("'\n"); \
1317 PUSH_FAILURE_POINTER (string_place); \
1319 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1320 DEBUG_PUSH (failure_id); \
1323 /* This is the number of items that are pushed and popped on the stack
1324 for each register. */
1325 #define NUM_REG_ITEMS 3
1327 /* Individual items aside from the registers. */
1329 #define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1331 #define NUM_NONREG_ITEMS 4
1334 /* We push at most this many items on the stack. */
1335 /* We used to use (num_regs - 1), which is the number of registers
1336 this regexp will save; but that was changed to 5
1337 to avoid stack overflow for a regexp with lots of parens. */
1338 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1340 /* We actually push this many items. */
1341 #define NUM_FAILURE_ITEMS \
1343 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1347 /* How many items can still be added to the stack without overflowing it. */
1348 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1351 /* Pops what PUSH_FAIL_STACK pushes.
1353 We restore into the parameters, all of which should be lvalues:
1354 STR -- the saved data position.
1355 PAT -- the saved pattern position.
1356 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1357 REGSTART, REGEND -- arrays of string positions.
1358 REG_INFO -- array of information about each subexpression.
1360 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1361 `pend', `string1', `size1', `string2', and `size2'. */
1363 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1365 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \
1367 const unsigned char *string_temp; \
1369 assert (!FAIL_STACK_EMPTY ()); \
1371 /* Remove failure points and point to how many regs pushed. */ \
1372 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1373 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1374 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1376 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1378 DEBUG_POP (&failure_id); \
1379 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1381 /* If the saved string location is NULL, it came from an \
1382 on_failure_keep_string_jump opcode, and we want to throw away the \
1383 saved NULL, thus retaining our current position in the string. */ \
1384 string_temp = POP_FAILURE_POINTER (); \
1385 if (string_temp != NULL) \
1386 str = (const char *) string_temp; \
1388 DEBUG_PRINT2 (" Popping string 0x%x: `", str); \
1389 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1390 DEBUG_PRINT1 ("'\n"); \
1392 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1393 DEBUG_PRINT2 (" Popping pattern 0x%x:\n", pat); \
1394 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1396 /* Restore register info. */ \
1397 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1398 DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \
1400 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1401 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \
1404 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1406 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \
1408 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1409 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \
1411 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1412 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1414 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1415 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1419 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1421 reg_info[this_reg].word.integer = 0; \
1422 regend[this_reg] = 0; \
1423 regstart[this_reg] = 0; \
1425 highest_active_reg = high_reg; \
1428 set_regs_matched_done = 0; \
1429 DEBUG_STATEMENT (nfailure_points_popped++); \
1430 } /* POP_FAILURE_POINT */
1434 /* Structure for per-register (a.k.a. per-group) information.
1435 Other register information, such as the
1436 starting and ending positions (which are addresses), and the list of
1437 inner groups (which is a bits list) are maintained in separate
1440 We are making a (strictly speaking) nonportable assumption here: that
1441 the compiler will pack our bit fields into something that fits into
1442 the type of `word', i.e., is something that fits into one item on the
1446 /* Declarations and macros for re_match_2. */
1450 fail_stack_elt_t word;
1453 /* This field is one if this group can match the empty string,
1454 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1455 #define MATCH_NULL_UNSET_VALUE 3
1456 unsigned match_null_string_p : 2;
1457 unsigned is_active : 1;
1458 unsigned matched_something : 1;
1459 unsigned ever_matched_something : 1;
1461 } register_info_type;
1463 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1464 #define IS_ACTIVE(R) ((R).bits.is_active)
1465 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1466 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1469 /* Call this when have matched a real character; it sets `matched' flags
1470 for the subexpressions which we are currently inside. Also records
1471 that those subexprs have matched. */
1472 #define SET_REGS_MATCHED() \
1475 if (!set_regs_matched_done) \
1478 set_regs_matched_done = 1; \
1479 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1481 MATCHED_SOMETHING (reg_info[r]) \
1482 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1489 /* Registers are set to a sentinel when they haven't yet matched. */
1490 static char reg_unset_dummy;
1491 #define REG_UNSET_VALUE (®_unset_dummy)
1492 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1494 /* Subroutine declarations and macros for regex_compile. */
1496 static reg_errcode_t regex_compile _RE_ARGS ((const char *pattern, size_t size,
1497 reg_syntax_t syntax,
1498 struct re_pattern_buffer *bufp));
1499 static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg));
1500 static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1501 int arg1, int arg2));
1502 static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1503 int arg, unsigned char *end));
1504 static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1505 int arg1, int arg2, unsigned char *end));
1506 static boolean at_begline_loc_p _RE_ARGS ((const char *pattern, const char *p,
1507 reg_syntax_t syntax));
1508 static boolean at_endline_loc_p _RE_ARGS ((const char *p, const char *pend,
1509 reg_syntax_t syntax));
1510 static reg_errcode_t compile_range _RE_ARGS ((const char **p_ptr,
1513 reg_syntax_t syntax,
1516 /* Fetch the next character in the uncompiled pattern---translating it
1517 if necessary. Also cast from a signed character in the constant
1518 string passed to us by the user to an unsigned char that we can use
1519 as an array index (in, e.g., `translate'). */
1521 #define PATFETCH(c) \
1522 do {if (p == pend) return REG_EEND; \
1523 c = (unsigned char) *p++; \
1524 if (translate) c = (unsigned char) translate[c]; \
1528 /* Fetch the next character in the uncompiled pattern, with no
1530 #define PATFETCH_RAW(c) \
1531 do {if (p == pend) return REG_EEND; \
1532 c = (unsigned char) *p++; \
1535 /* Go backwards one character in the pattern. */
1536 #define PATUNFETCH p--
1539 /* If `translate' is non-null, return translate[D], else just D. We
1540 cast the subscript to translate because some data is declared as
1541 `char *', to avoid warnings when a string constant is passed. But
1542 when we use a character as a subscript we must make it unsigned. */
1544 #define TRANSLATE(d) \
1545 (translate ? (char) translate[(unsigned char) (d)] : (d))
1549 /* Macros for outputting the compiled pattern into `buffer'. */
1551 /* If the buffer isn't allocated when it comes in, use this. */
1552 #define INIT_BUF_SIZE 32
1554 /* Make sure we have at least N more bytes of space in buffer. */
1555 #define GET_BUFFER_SPACE(n) \
1556 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1559 /* Make sure we have one more byte of buffer space and then add C to it. */
1560 #define BUF_PUSH(c) \
1562 GET_BUFFER_SPACE (1); \
1563 *b++ = (unsigned char) (c); \
1567 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1568 #define BUF_PUSH_2(c1, c2) \
1570 GET_BUFFER_SPACE (2); \
1571 *b++ = (unsigned char) (c1); \
1572 *b++ = (unsigned char) (c2); \
1576 /* As with BUF_PUSH_2, except for three bytes. */
1577 #define BUF_PUSH_3(c1, c2, c3) \
1579 GET_BUFFER_SPACE (3); \
1580 *b++ = (unsigned char) (c1); \
1581 *b++ = (unsigned char) (c2); \
1582 *b++ = (unsigned char) (c3); \
1586 /* Store a jump with opcode OP at LOC to location TO. We store a
1587 relative address offset by the three bytes the jump itself occupies. */
1588 #define STORE_JUMP(op, loc, to) \
1589 store_op1 (op, loc, (int) ((to) - (loc) - 3))
1591 /* Likewise, for a two-argument jump. */
1592 #define STORE_JUMP2(op, loc, to, arg) \
1593 store_op2 (op, loc, (int) ((to) - (loc) - 3), arg)
1595 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1596 #define INSERT_JUMP(op, loc, to) \
1597 insert_op1 (op, loc, (int) ((to) - (loc) - 3), b)
1599 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1600 #define INSERT_JUMP2(op, loc, to, arg) \
1601 insert_op2 (op, loc, (int) ((to) - (loc) - 3), arg, b)
1604 /* This is not an arbitrary limit: the arguments which represent offsets
1605 into the pattern are two bytes long. So if 2^16 bytes turns out to
1606 be too small, many things would have to change. */
1607 /* Any other compiler which, like MSC, has allocation limit below 2^16
1608 bytes will have to use approach similar to what was done below for
1609 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1610 reallocating to 0 bytes. Such thing is not going to work too well.
1611 You have been warned!! */
1612 #if defined(_MSC_VER) && !defined(WIN32)
1613 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
1614 The REALLOC define eliminates a flurry of conversion warnings,
1615 but is not required. */
1616 #define MAX_BUF_SIZE 65500L
1617 #define REALLOC(p,s) realloc ((p), (size_t) (s))
1619 #define MAX_BUF_SIZE (1L << 16)
1620 #define REALLOC(p,s) realloc ((p), (s))
1623 /* Extend the buffer by twice its current size via realloc and
1624 reset the pointers that pointed into the old block to point to the
1625 correct places in the new one. If extending the buffer results in it
1626 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1627 #define EXTEND_BUFFER() \
1629 unsigned char *old_buffer = bufp->buffer; \
1630 if (bufp->allocated == MAX_BUF_SIZE) \
1632 bufp->allocated <<= 1; \
1633 if (bufp->allocated > MAX_BUF_SIZE) \
1634 bufp->allocated = MAX_BUF_SIZE; \
1635 bufp->buffer = (unsigned char *) REALLOC (bufp->buffer, bufp->allocated);\
1636 if (bufp->buffer == NULL) \
1637 return REG_ESPACE; \
1638 /* If the buffer moved, move all the pointers into it. */ \
1639 if (old_buffer != bufp->buffer) \
1641 b = (b - old_buffer) + bufp->buffer; \
1642 begalt = (begalt - old_buffer) + bufp->buffer; \
1643 if (fixup_alt_jump) \
1644 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1646 laststart = (laststart - old_buffer) + bufp->buffer; \
1647 if (pending_exact) \
1648 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1653 /* Since we have one byte reserved for the register number argument to
1654 {start,stop}_memory, the maximum number of groups we can report
1655 things about is what fits in that byte. */
1656 #define MAX_REGNUM 255
1658 /* But patterns can have more than `MAX_REGNUM' registers. We just
1659 ignore the excess. */
1660 typedef unsigned regnum_t;
1663 /* Macros for the compile stack. */
1665 /* Since offsets can go either forwards or backwards, this type needs to
1666 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1667 /* int may be not enough when sizeof(int) == 2. */
1668 typedef long pattern_offset_t;
1672 pattern_offset_t begalt_offset;
1673 pattern_offset_t fixup_alt_jump;
1674 pattern_offset_t inner_group_offset;
1675 pattern_offset_t laststart_offset;
1677 } compile_stack_elt_t;
1682 compile_stack_elt_t *stack;
1684 unsigned avail; /* Offset of next open position. */
1685 } compile_stack_type;
1688 #define INIT_COMPILE_STACK_SIZE 32
1690 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1691 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1693 /* The next available element. */
1694 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1697 /* Set the bit for character C in a list. */
1698 #define SET_LIST_BIT(c) \
1699 (b[((unsigned char) (c)) / BYTEWIDTH] \
1700 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1703 /* Get the next unsigned number in the uncompiled pattern. */
1704 #define GET_UNSIGNED_NUMBER(num) \
1708 while (ISDIGIT (c)) \
1712 num = num * 10 + c - '0'; \
1720 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
1721 /* The GNU C library provides support for user-defined character classes
1722 and the functions from ISO C amendement 1. */
1723 # ifdef CHARCLASS_NAME_MAX
1724 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
1726 /* This shouldn't happen but some implementation might still have this
1727 problem. Use a reasonable default value. */
1728 # define CHAR_CLASS_MAX_LENGTH 256
1731 # define IS_CHAR_CLASS(string) wctype (string)
1733 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1735 # define IS_CHAR_CLASS(string) \
1736 (STREQ (string, "alpha") || STREQ (string, "upper") \
1737 || STREQ (string, "lower") || STREQ (string, "digit") \
1738 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1739 || STREQ (string, "space") || STREQ (string, "print") \
1740 || STREQ (string, "punct") || STREQ (string, "graph") \
1741 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1744 #ifndef MATCH_MAY_ALLOCATE
1746 /* If we cannot allocate large objects within re_match_2_internal,
1747 we make the fail stack and register vectors global.
1748 The fail stack, we grow to the maximum size when a regexp
1750 The register vectors, we adjust in size each time we
1751 compile a regexp, according to the number of registers it needs. */
1753 static fail_stack_type fail_stack;
1755 /* Size with which the following vectors are currently allocated.
1756 That is so we can make them bigger as needed,
1757 but never make them smaller. */
1758 static int regs_allocated_size;
1760 static const char ** regstart, ** regend;
1761 static const char ** old_regstart, ** old_regend;
1762 static const char **best_regstart, **best_regend;
1763 static register_info_type *reg_info;
1764 static const char **reg_dummy;
1765 static register_info_type *reg_info_dummy;
1767 /* Make the register vectors big enough for NUM_REGS registers,
1768 but don't make them smaller. */
1771 regex_grow_registers (num_regs)
1774 if (num_regs > regs_allocated_size)
1776 RETALLOC_IF (regstart, num_regs, const char *);
1777 RETALLOC_IF (regend, num_regs, const char *);
1778 RETALLOC_IF (old_regstart, num_regs, const char *);
1779 RETALLOC_IF (old_regend, num_regs, const char *);
1780 RETALLOC_IF (best_regstart, num_regs, const char *);
1781 RETALLOC_IF (best_regend, num_regs, const char *);
1782 RETALLOC_IF (reg_info, num_regs, register_info_type);
1783 RETALLOC_IF (reg_dummy, num_regs, const char *);
1784 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1786 regs_allocated_size = num_regs;
1790 #endif /* not MATCH_MAY_ALLOCATE */
1792 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
1796 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1797 Returns one of error codes defined in `regex.h', or zero for success.
1799 Assumes the `allocated' (and perhaps `buffer') and `translate'
1800 fields are set in BUFP on entry.
1802 If it succeeds, results are put in BUFP (if it returns an error, the
1803 contents of BUFP are undefined):
1804 `buffer' is the compiled pattern;
1805 `syntax' is set to SYNTAX;
1806 `used' is set to the length of the compiled pattern;
1807 `fastmap_accurate' is zero;
1808 `re_nsub' is the number of subexpressions in PATTERN;
1809 `not_bol' and `not_eol' are zero;
1811 The `fastmap' and `newline_anchor' fields are neither
1812 examined nor set. */
1814 /* Return, freeing storage we allocated. */
1815 #define FREE_STACK_RETURN(value) \
1816 return (free (compile_stack.stack), value) /* __MEM_CHECKED__ */
1818 static reg_errcode_t
1819 regex_compile (pattern, size, syntax, bufp)
1820 const char *pattern;
1822 reg_syntax_t syntax;
1823 struct re_pattern_buffer *bufp;
1825 /* We fetch characters from PATTERN here. Even though PATTERN is
1826 `char *' (i.e., signed), we declare these variables as unsigned, so
1827 they can be reliably used as array indices. */
1828 register unsigned char c, c1;
1830 /* A random temporary spot in PATTERN. */
1833 /* Points to the end of the buffer, where we should append. */
1834 register unsigned char *b;
1836 /* Keeps track of unclosed groups. */
1837 compile_stack_type compile_stack;
1839 /* Points to the current (ending) position in the pattern. */
1840 const char *p = pattern;
1841 const char *pend = pattern + size;
1843 /* How to translate the characters in the pattern. */
1844 RE_TRANSLATE_TYPE translate = bufp->translate;
1846 /* Address of the count-byte of the most recently inserted `exactn'
1847 command. This makes it possible to tell if a new exact-match
1848 character can be added to that command or if the character requires
1849 a new `exactn' command. */
1850 unsigned char *pending_exact = 0;
1852 /* Address of start of the most recently finished expression.
1853 This tells, e.g., postfix * where to find the start of its
1854 operand. Reset at the beginning of groups and alternatives. */
1855 unsigned char *laststart = 0;
1857 /* Address of beginning of regexp, or inside of last group. */
1858 unsigned char *begalt;
1860 /* Place in the uncompiled pattern (i.e., the {) to
1861 which to go back if the interval is invalid. */
1862 const char *beg_interval;
1864 /* Address of the place where a forward jump should go to the end of
1865 the containing expression. Each alternative of an `or' -- except the
1866 last -- ends with a forward jump of this sort. */
1867 unsigned char *fixup_alt_jump = 0;
1869 /* Counts open-groups as they are encountered. Remembered for the
1870 matching close-group on the compile stack, so the same register
1871 number is put in the stop_memory as the start_memory. */
1872 regnum_t regnum = 0;
1875 DEBUG_PRINT1 ("\nCompiling pattern: ");
1878 unsigned debug_count;
1880 for (debug_count = 0; debug_count < size; debug_count++)
1881 putchar (pattern[debug_count]);
1886 /* Initialize the compile stack. */
1887 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1888 if (compile_stack.stack == NULL)
1891 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1892 compile_stack.avail = 0;
1894 /* Initialize the pattern buffer. */
1895 bufp->syntax = syntax;
1896 bufp->fastmap_accurate = 0;
1897 bufp->not_bol = bufp->not_eol = 0;
1899 /* Set `used' to zero, so that if we return an error, the pattern
1900 printer (for debugging) will think there's no pattern. We reset it
1904 /* Always count groups, whether or not bufp->no_sub is set. */
1907 #if !defined (emacs) && !defined (SYNTAX_TABLE)
1908 /* Initialize the syntax table. */
1909 init_syntax_once ();
1912 if (bufp->allocated == 0)
1915 { /* If zero allocated, but buffer is non-null, try to realloc
1916 enough space. This loses if buffer's address is bogus, but
1917 that is the user's responsibility. */
1918 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1921 { /* Caller did not allocate a buffer. Do it for them. */
1922 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1924 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1926 bufp->allocated = INIT_BUF_SIZE;
1929 begalt = b = bufp->buffer;
1931 /* Loop through the uncompiled pattern until we're at the end. */
1940 if ( /* If at start of pattern, it's an operator. */
1942 /* If context independent, it's an operator. */
1943 || syntax & RE_CONTEXT_INDEP_ANCHORS
1944 /* Otherwise, depends on what's come before. */
1945 || at_begline_loc_p (pattern, p, syntax))
1955 if ( /* If at end of pattern, it's an operator. */
1957 /* If context independent, it's an operator. */
1958 || syntax & RE_CONTEXT_INDEP_ANCHORS
1959 /* Otherwise, depends on what's next. */
1960 || at_endline_loc_p (p, pend, syntax))
1970 if ((syntax & RE_BK_PLUS_QM)
1971 || (syntax & RE_LIMITED_OPS))
1975 /* If there is no previous pattern... */
1978 if (syntax & RE_CONTEXT_INVALID_OPS)
1979 FREE_STACK_RETURN (REG_BADRPT);
1980 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1985 /* Are we optimizing this jump? */
1986 boolean keep_string_p = false;
1988 /* 1 means zero (many) matches is allowed. */
1989 char zero_times_ok = 0, many_times_ok = 0;
1991 /* If there is a sequence of repetition chars, collapse it
1992 down to just one (the right one). We can't combine
1993 interval operators with these because of, e.g., `a{2}*',
1994 which should only match an even number of `a's. */
1998 zero_times_ok |= c != '+';
1999 many_times_ok |= c != '?';
2007 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
2010 else if (syntax & RE_BK_PLUS_QM && c == '\\')
2012 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2015 if (!(c1 == '+' || c1 == '?'))
2030 /* If we get here, we found another repeat character. */
2033 /* Star, etc. applied to an empty pattern is equivalent
2034 to an empty pattern. */
2038 /* Now we know whether or not zero matches is allowed
2039 and also whether or not two or more matches is allowed. */
2041 { /* More than one repetition is allowed, so put in at the
2042 end a backward relative jump from `b' to before the next
2043 jump we're going to put in below (which jumps from
2044 laststart to after this jump).
2046 But if we are at the `*' in the exact sequence `.*\n',
2047 insert an unconditional jump backwards to the .,
2048 instead of the beginning of the loop. This way we only
2049 push a failure point once, instead of every time
2050 through the loop. */
2051 assert (p - 1 > pattern);
2053 /* Allocate the space for the jump. */
2054 GET_BUFFER_SPACE (3);
2056 /* We know we are not at the first character of the pattern,
2057 because laststart was nonzero. And we've already
2058 incremented `p', by the way, to be the character after
2059 the `*'. Do we have to do something analogous here
2060 for null bytes, because of RE_DOT_NOT_NULL? */
2061 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2063 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2064 && !(syntax & RE_DOT_NEWLINE))
2065 { /* We have .*\n. */
2066 STORE_JUMP (jump, b, laststart);
2067 keep_string_p = true;
2070 /* Anything else. */
2071 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
2073 /* We've added more stuff to the buffer. */
2077 /* On failure, jump from laststart to b + 3, which will be the
2078 end of the buffer after this jump is inserted. */
2079 GET_BUFFER_SPACE (3);
2080 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2088 /* At least one repetition is required, so insert a
2089 `dummy_failure_jump' before the initial
2090 `on_failure_jump' instruction of the loop. This
2091 effects a skip over that instruction the first time
2092 we hit that loop. */
2093 GET_BUFFER_SPACE (3);
2094 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
2109 boolean had_char_class = false;
2111 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2113 /* Ensure that we have enough space to push a charset: the
2114 opcode, the length count, and the bitset; 34 bytes in all. */
2115 GET_BUFFER_SPACE (34);
2119 /* We test `*p == '^' twice, instead of using an if
2120 statement, so we only need one BUF_PUSH. */
2121 BUF_PUSH (*p == '^' ? charset_not : charset);
2125 /* Remember the first position in the bracket expression. */
2128 /* Push the number of bytes in the bitmap. */
2129 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
2131 /* Clear the whole map. */
2132 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
2134 /* charset_not matches newline according to a syntax bit. */
2135 if ((re_opcode_t) b[-2] == charset_not
2136 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2137 SET_LIST_BIT ('\n');
2139 /* Read in characters and ranges, setting map bits. */
2142 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2146 /* \ might escape characters inside [...] and [^...]. */
2147 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2149 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2156 /* Could be the end of the bracket expression. If it's
2157 not (i.e., when the bracket expression is `[]' so
2158 far), the ']' character bit gets set way below. */
2159 if (c == ']' && p != p1 + 1)
2162 /* Look ahead to see if it's a range when the last thing
2163 was a character class. */
2164 if (had_char_class && c == '-' && *p != ']')
2165 FREE_STACK_RETURN (REG_ERANGE);
2167 /* Look ahead to see if it's a range when the last thing
2168 was a character: if this is a hyphen not at the
2169 beginning or the end of a list, then it's the range
2172 && !(p - 2 >= pattern && p[-2] == '[')
2173 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2177 = compile_range (&p, pend, translate, syntax, b);
2178 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2181 else if (p[0] == '-' && p[1] != ']')
2182 { /* This handles ranges made up of characters only. */
2185 /* Move past the `-'. */
2188 ret = compile_range (&p, pend, translate, syntax, b);
2189 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2192 /* See if we're at the beginning of a possible character
2195 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2196 { /* Leave room for the null. */
2197 char str[CHAR_CLASS_MAX_LENGTH + 1];
2202 /* If pattern is `[[:'. */
2203 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2208 if (c == ':' || c == ']' || p == pend
2209 || (unsigned int)c1 == CHAR_CLASS_MAX_LENGTH)
2215 /* If isn't a word bracketed by `[:' and:`]':
2216 undo the ending character, the letters, and leave
2217 the leading `:' and `[' (but set bits for them). */
2218 if (c == ':' && *p == ']')
2220 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
2221 boolean is_lower = STREQ (str, "lower");
2222 boolean is_upper = STREQ (str, "upper");
2228 FREE_STACK_RETURN (REG_ECTYPE);
2230 /* Throw away the ] at the end of the character
2234 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2236 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
2238 if (iswctype (btowc (ch), wt))
2241 if (translate && (is_upper || is_lower)
2242 && (ISUPPER (ch) || ISLOWER (ch)))
2246 had_char_class = true;
2249 boolean is_alnum = STREQ (str, "alnum");
2250 boolean is_alpha = STREQ (str, "alpha");
2251 boolean is_blank = STREQ (str, "blank");
2252 boolean is_cntrl = STREQ (str, "cntrl");
2253 boolean is_digit = STREQ (str, "digit");
2254 boolean is_graph = STREQ (str, "graph");
2255 boolean is_lower = STREQ (str, "lower");
2256 boolean is_print = STREQ (str, "print");
2257 boolean is_punct = STREQ (str, "punct");
2258 boolean is_space = STREQ (str, "space");
2259 boolean is_upper = STREQ (str, "upper");
2260 boolean is_xdigit = STREQ (str, "xdigit");
2262 if (!IS_CHAR_CLASS (str))
2263 FREE_STACK_RETURN (REG_ECTYPE);
2265 /* Throw away the ] at the end of the character
2269 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2271 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2273 /* This was split into 3 if's to
2274 avoid an arbitrary limit in some compiler. */
2275 if ( (is_alnum && ISALNUM (ch))
2276 || (is_alpha && ISALPHA (ch))
2277 || (is_blank && ISBLANK (ch))
2278 || (is_cntrl && ISCNTRL (ch)))
2280 if ( (is_digit && ISDIGIT (ch))
2281 || (is_graph && ISGRAPH (ch))
2282 || (is_lower && ISLOWER (ch))
2283 || (is_print && ISPRINT (ch)))
2285 if ( (is_punct && ISPUNCT (ch))
2286 || (is_space && ISSPACE (ch))
2287 || (is_upper && ISUPPER (ch))
2288 || (is_xdigit && ISXDIGIT (ch)))
2290 if ( translate && (is_upper || is_lower)
2291 && (ISUPPER (ch) || ISLOWER (ch)))
2294 had_char_class = true;
2295 #endif /* libc || wctype.h */
2304 had_char_class = false;
2309 had_char_class = false;
2314 /* Discard any (non)matching list bytes that are all 0 at the
2315 end of the map. Decrease the map-length byte too. */
2316 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2324 if (syntax & RE_NO_BK_PARENS)
2331 if (syntax & RE_NO_BK_PARENS)
2338 if (syntax & RE_NEWLINE_ALT)
2345 if (syntax & RE_NO_BK_VBAR)
2352 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2353 goto handle_interval;
2359 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2361 /* Do not translate the character after the \, so that we can
2362 distinguish, e.g., \B from \b, even if we normally would
2363 translate, e.g., B to b. */
2369 if (syntax & RE_NO_BK_PARENS)
2370 goto normal_backslash;
2376 if (COMPILE_STACK_FULL)
2378 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2379 compile_stack_elt_t);
2380 if (compile_stack.stack == NULL) return REG_ESPACE;
2382 compile_stack.size <<= 1;
2385 /* These are the values to restore when we hit end of this
2386 group. They are all relative offsets, so that if the
2387 whole pattern moves because of realloc, they will still
2389 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2390 COMPILE_STACK_TOP.fixup_alt_jump
2391 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2392 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2393 COMPILE_STACK_TOP.regnum = regnum;
2395 /* We will eventually replace the 0 with the number of
2396 groups inner to this one. But do not push a
2397 start_memory for groups beyond the last one we can
2398 represent in the compiled pattern. */
2399 if (regnum <= MAX_REGNUM)
2401 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2402 BUF_PUSH_3 (start_memory, regnum, 0);
2405 compile_stack.avail++;
2410 /* If we've reached MAX_REGNUM groups, then this open
2411 won't actually generate any code, so we'll have to
2412 clear pending_exact explicitly. */
2418 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2420 if (COMPILE_STACK_EMPTY)
2422 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2423 goto normal_backslash;
2425 FREE_STACK_RETURN (REG_ERPAREN);
2430 { /* Push a dummy failure point at the end of the
2431 alternative for a possible future
2432 `pop_failure_jump' to pop. See comments at
2433 `push_dummy_failure' in `re_match_2'. */
2434 BUF_PUSH (push_dummy_failure);
2436 /* We allocated space for this jump when we assigned
2437 to `fixup_alt_jump', in the `handle_alt' case below. */
2438 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2441 /* See similar code for backslashed left paren above. */
2442 if (COMPILE_STACK_EMPTY)
2444 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2447 FREE_STACK_RETURN (REG_ERPAREN);
2450 /* Since we just checked for an empty stack above, this
2451 ``can't happen''. */
2452 assert (compile_stack.avail != 0);
2454 /* We don't just want to restore into `regnum', because
2455 later groups should continue to be numbered higher,
2456 as in `(ab)c(de)' -- the second group is #2. */
2457 regnum_t this_group_regnum;
2459 compile_stack.avail--;
2460 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2462 = COMPILE_STACK_TOP.fixup_alt_jump
2463 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2465 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2466 this_group_regnum = COMPILE_STACK_TOP.regnum;
2467 /* If we've reached MAX_REGNUM groups, then this open
2468 won't actually generate any code, so we'll have to
2469 clear pending_exact explicitly. */
2472 /* We're at the end of the group, so now we know how many
2473 groups were inside this one. */
2474 if (this_group_regnum <= MAX_REGNUM)
2476 unsigned char *inner_group_loc
2477 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2479 *inner_group_loc = regnum - this_group_regnum;
2480 BUF_PUSH_3 (stop_memory, this_group_regnum,
2481 regnum - this_group_regnum);
2487 case '|': /* `\|'. */
2488 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2489 goto normal_backslash;
2491 if (syntax & RE_LIMITED_OPS)
2494 /* Insert before the previous alternative a jump which
2495 jumps to this alternative if the former fails. */
2496 GET_BUFFER_SPACE (3);
2497 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2501 /* The alternative before this one has a jump after it
2502 which gets executed if it gets matched. Adjust that
2503 jump so it will jump to this alternative's analogous
2504 jump (put in below, which in turn will jump to the next
2505 (if any) alternative's such jump, etc.). The last such
2506 jump jumps to the correct final destination. A picture:
2512 If we are at `b', then fixup_alt_jump right now points to a
2513 three-byte space after `a'. We'll put in the jump, set
2514 fixup_alt_jump to right after `b', and leave behind three
2515 bytes which we'll fill in when we get to after `c'. */
2518 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2520 /* Mark and leave space for a jump after this alternative,
2521 to be filled in later either by next alternative or
2522 when know we're at the end of a series of alternatives. */
2524 GET_BUFFER_SPACE (3);
2533 /* If \{ is a literal. */
2534 if (!(syntax & RE_INTERVALS)
2535 /* If we're at `\{' and it's not the open-interval
2537 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2538 || (p - 2 == pattern && p == pend))
2539 goto normal_backslash;
2543 /* If got here, then the syntax allows intervals. */
2545 /* At least (most) this many matches must be made. */
2546 int lower_bound = -1, upper_bound = -1;
2548 beg_interval = p - 1;
2552 if (syntax & RE_NO_BK_BRACES)
2553 goto unfetch_interval;
2555 FREE_STACK_RETURN (REG_EBRACE);
2558 GET_UNSIGNED_NUMBER (lower_bound);
2562 GET_UNSIGNED_NUMBER (upper_bound);
2563 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2566 /* Interval such as `{1}' => match exactly once. */
2567 upper_bound = lower_bound;
2569 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2570 || lower_bound > upper_bound)
2572 if (syntax & RE_NO_BK_BRACES)
2573 goto unfetch_interval;
2575 FREE_STACK_RETURN (REG_BADBR);
2578 if (!(syntax & RE_NO_BK_BRACES))
2580 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2587 if (syntax & RE_NO_BK_BRACES)
2588 goto unfetch_interval;
2590 FREE_STACK_RETURN (REG_BADBR);
2593 /* We just parsed a valid interval. */
2595 /* If it's invalid to have no preceding re. */
2598 if (syntax & RE_CONTEXT_INVALID_OPS)
2599 FREE_STACK_RETURN (REG_BADRPT);
2600 else if (syntax & RE_CONTEXT_INDEP_OPS)
2603 goto unfetch_interval;
2606 /* If the upper bound is zero, don't want to succeed at
2607 all; jump from `laststart' to `b + 3', which will be
2608 the end of the buffer after we insert the jump. */
2609 if (upper_bound == 0)
2611 GET_BUFFER_SPACE (3);
2612 INSERT_JUMP (jump, laststart, b + 3);
2616 /* Otherwise, we have a nontrivial interval. When
2617 we're all done, the pattern will look like:
2618 set_number_at <jump count> <upper bound>
2619 set_number_at <succeed_n count> <lower bound>
2620 succeed_n <after jump addr> <succeed_n count>
2622 jump_n <succeed_n addr> <jump count>
2623 (The upper bound and `jump_n' are omitted if
2624 `upper_bound' is 1, though.) */
2626 { /* If the upper bound is > 1, we need to insert
2627 more at the end of the loop. */
2628 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2630 GET_BUFFER_SPACE (nbytes);
2632 /* Initialize lower bound of the `succeed_n', even
2633 though it will be set during matching by its
2634 attendant `set_number_at' (inserted next),
2635 because `re_compile_fastmap' needs to know.
2636 Jump to the `jump_n' we might insert below. */
2637 INSERT_JUMP2 (succeed_n, laststart,
2638 b + 5 + (upper_bound > 1) * 5,
2642 /* Code to initialize the lower bound. Insert
2643 before the `succeed_n'. The `5' is the last two
2644 bytes of this `set_number_at', plus 3 bytes of
2645 the following `succeed_n'. */
2646 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2649 if (upper_bound > 1)
2650 { /* More than one repetition is allowed, so
2651 append a backward jump to the `succeed_n'
2652 that starts this interval.
2654 When we've reached this during matching,
2655 we'll have matched the interval once, so
2656 jump back only `upper_bound - 1' times. */
2657 STORE_JUMP2 (jump_n, b, laststart + 5,
2661 /* The location we want to set is the second
2662 parameter of the `jump_n'; that is `b-2' as
2663 an absolute address. `laststart' will be
2664 the `set_number_at' we're about to insert;
2665 `laststart+3' the number to set, the source
2666 for the relative address. But we are
2667 inserting into the middle of the pattern --
2668 so everything is getting moved up by 5.
2669 Conclusion: (b - 2) - (laststart + 3) + 5,
2670 i.e., b - laststart.
2672 We insert this at the beginning of the loop
2673 so that if we fail during matching, we'll
2674 reinitialize the bounds. */
2675 insert_op2 (set_number_at, laststart, b - laststart,
2676 upper_bound - 1, b);
2681 beg_interval = NULL;
2686 /* If an invalid interval, match the characters as literals. */
2687 assert (beg_interval);
2689 beg_interval = NULL;
2691 /* normal_char and normal_backslash need `c'. */
2694 if (!(syntax & RE_NO_BK_BRACES))
2696 if (p > pattern && p[-1] == '\\')
2697 goto normal_backslash;
2702 /* There is no way to specify the before_dot and after_dot
2703 operators. rms says this is ok. --karl */
2711 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2717 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2723 if (re_syntax_options & RE_NO_GNU_OPS)
2726 BUF_PUSH (wordchar);
2731 if (re_syntax_options & RE_NO_GNU_OPS)
2734 BUF_PUSH (notwordchar);
2739 if (re_syntax_options & RE_NO_GNU_OPS)
2745 if (re_syntax_options & RE_NO_GNU_OPS)
2751 if (re_syntax_options & RE_NO_GNU_OPS)
2753 BUF_PUSH (wordbound);
2757 if (re_syntax_options & RE_NO_GNU_OPS)
2759 BUF_PUSH (notwordbound);
2763 if (re_syntax_options & RE_NO_GNU_OPS)
2769 if (re_syntax_options & RE_NO_GNU_OPS)
2774 case '1': case '2': case '3': case '4': case '5':
2775 case '6': case '7': case '8': case '9':
2776 if (syntax & RE_NO_BK_REFS)
2782 FREE_STACK_RETURN (REG_ESUBREG);
2784 /* Can't back reference to a subexpression if inside of it. */
2785 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
2789 BUF_PUSH_2 (duplicate, c1);
2795 if (syntax & RE_BK_PLUS_QM)
2798 goto normal_backslash;
2802 /* You might think it would be useful for \ to mean
2803 not to translate; but if we don't translate it
2804 it will never match anything. */
2812 /* Expects the character in `c'. */
2814 /* If no exactn currently being built. */
2817 /* If last exactn not at current position. */
2818 || pending_exact + *pending_exact + 1 != b
2820 /* We have only one byte following the exactn for the count. */
2821 || *pending_exact == (1 << BYTEWIDTH) - 1
2823 /* If followed by a repetition operator. */
2824 || *p == '*' || *p == '^'
2825 || ((syntax & RE_BK_PLUS_QM)
2826 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2827 : (*p == '+' || *p == '?'))
2828 || ((syntax & RE_INTERVALS)
2829 && ((syntax & RE_NO_BK_BRACES)
2831 : (p[0] == '\\' && p[1] == '{'))))
2833 /* Start building a new exactn. */
2837 BUF_PUSH_2 (exactn, 0);
2838 pending_exact = b - 1;
2845 } /* while p != pend */
2848 /* Through the pattern now. */
2851 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2853 if (!COMPILE_STACK_EMPTY)
2854 FREE_STACK_RETURN (REG_EPAREN);
2856 /* If we don't want backtracking, force success
2857 the first time we reach the end of the compiled pattern. */
2858 if (syntax & RE_NO_POSIX_BACKTRACKING)
2861 free (compile_stack.stack); /* __MEM_CHECKED__ */
2863 /* We have succeeded; set the length of the buffer. */
2864 bufp->used = b - bufp->buffer;
2869 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2870 print_compiled_pattern (bufp);
2874 #ifndef MATCH_MAY_ALLOCATE
2875 /* Initialize the failure stack to the largest possible stack. This
2876 isn't necessary unless we're trying to avoid calling alloca in
2877 the search and match routines. */
2879 int num_regs = bufp->re_nsub + 1;
2881 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2882 is strictly greater than re_max_failures, the largest possible stack
2883 is 2 * re_max_failures failure points. */
2884 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2886 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2889 if (! fail_stack.stack)
2891 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2892 * sizeof (fail_stack_elt_t));
2895 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2897 * sizeof (fail_stack_elt_t)));
2898 #else /* not emacs */
2899 if (! fail_stack.stack)
2901 = (fail_stack_elt_t *) malloc (fail_stack.size /* __MEM_CHECKED__ */
2902 * sizeof (fail_stack_elt_t));
2905 = (fail_stack_elt_t *) realloc (fail_stack.stack, /* __MEM_CHECKED__ */
2907 * sizeof (fail_stack_elt_t)));
2908 #endif /* not emacs */
2911 regex_grow_registers (num_regs);
2913 #endif /* not MATCH_MAY_ALLOCATE */
2916 } /* regex_compile */
2918 /* Subroutines for `regex_compile'. */
2920 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2923 store_op1 (op, loc, arg)
2928 *loc = (unsigned char) op;
2929 STORE_NUMBER (loc + 1, arg);
2933 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2936 store_op2 (op, loc, arg1, arg2)
2941 *loc = (unsigned char) op;
2942 STORE_NUMBER (loc + 1, arg1);
2943 STORE_NUMBER (loc + 3, arg2);
2947 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
2948 for OP followed by two-byte integer parameter ARG. */
2951 insert_op1 (op, loc, arg, end)
2957 register unsigned char *pfrom = end;
2958 register unsigned char *pto = end + 3;
2960 while (pfrom != loc)
2963 store_op1 (op, loc, arg);
2967 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2970 insert_op2 (op, loc, arg1, arg2, end)
2976 register unsigned char *pfrom = end;
2977 register unsigned char *pto = end + 5;
2979 while (pfrom != loc)
2982 store_op2 (op, loc, arg1, arg2);
2986 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
2987 after an alternative or a begin-subexpression. We assume there is at
2988 least one character before the ^. */
2991 at_begline_loc_p (pattern, p, syntax)
2992 const char *pattern, *p;
2993 reg_syntax_t syntax;
2995 const char *prev = p - 2;
2996 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
2999 /* After a subexpression? */
3000 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
3001 /* After an alternative? */
3002 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
3006 /* The dual of at_begline_loc_p. This one is for $. We assume there is
3007 at least one character after the $, i.e., `P < PEND'. */
3010 at_endline_loc_p (p, pend, syntax)
3011 const char *p, *pend;
3012 reg_syntax_t syntax;
3014 const char *next = p;
3015 boolean next_backslash = *next == '\\';
3016 const char *next_next = p + 1 < pend ? p + 1 : 0;
3019 /* Before a subexpression? */
3020 (syntax & RE_NO_BK_PARENS ? *next == ')'
3021 : next_backslash && next_next && *next_next == ')')
3022 /* Before an alternative? */
3023 || (syntax & RE_NO_BK_VBAR ? *next == '|'
3024 : next_backslash && next_next && *next_next == '|');
3028 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
3029 false if it's not. */
3032 group_in_compile_stack (compile_stack, regnum)
3033 compile_stack_type compile_stack;
3038 for (this_element = compile_stack.avail - 1;
3041 if (compile_stack.stack[this_element].regnum == regnum)
3048 /* Read the ending character of a range (in a bracket expression) from the
3049 uncompiled pattern *P_PTR (which ends at PEND). We assume the
3050 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
3051 Then we set the translation of all bits between the starting and
3052 ending characters (inclusive) in the compiled pattern B.
3054 Return an error code.
3056 We use these short variable names so we can use the same macros as
3057 `regex_compile' itself. */
3059 static reg_errcode_t
3060 compile_range (p_ptr, pend, translate, syntax, b)
3061 const char **p_ptr, *pend;
3062 RE_TRANSLATE_TYPE translate;
3063 reg_syntax_t syntax;
3068 const char *p = *p_ptr;
3069 unsigned int range_start, range_end;
3074 /* Even though the pattern is a signed `char *', we need to fetch
3075 with unsigned char *'s; if the high bit of the pattern character
3076 is set, the range endpoints will be negative if we fetch using a
3079 We also want to fetch the endpoints without translating them; the
3080 appropriate translation is done in the bit-setting loop below. */
3081 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
3082 range_start = ((const unsigned char *) p)[-2];
3083 range_end = ((const unsigned char *) p)[0];
3085 /* Have to increment the pointer into the pattern string, so the
3086 caller isn't still at the ending character. */
3089 /* If the start is after the end, the range is empty. */
3090 if (range_start > range_end)
3091 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
3093 /* Here we see why `this_char' has to be larger than an `unsigned
3094 char' -- the range is inclusive, so if `range_end' == 0xff
3095 (assuming 8-bit characters), we would otherwise go into an infinite
3096 loop, since all characters <= 0xff. */
3097 for (this_char = range_start; this_char <= range_end; this_char++)
3099 SET_LIST_BIT (TRANSLATE (this_char));
3105 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3106 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3107 characters can start a string that matches the pattern. This fastmap
3108 is used by re_search to skip quickly over impossible starting points.
3110 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3111 area as BUFP->fastmap.
3113 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3116 Returns 0 if we succeed, -2 if an internal error. */
3119 re_compile_fastmap (bufp)
3120 struct re_pattern_buffer *bufp;
3123 #ifdef MATCH_MAY_ALLOCATE
3124 fail_stack_type fail_stack;
3126 #ifndef REGEX_MALLOC
3129 register char *fastmap = bufp->fastmap;
3130 unsigned char *pattern = bufp->buffer;
3131 unsigned char *p = pattern;
3132 register unsigned char *pend = pattern + bufp->used;
3135 /* This holds the pointer to the failure stack, when
3136 it is allocated relocatably. */
3137 fail_stack_elt_t *failure_stack_ptr;
3140 /* Assume that each path through the pattern can be null until
3141 proven otherwise. We set this false at the bottom of switch
3142 statement, to which we get only if a particular path doesn't
3143 match the empty string. */
3144 boolean path_can_be_null = true;
3146 /* We aren't doing a `succeed_n' to begin with. */
3147 boolean succeed_n_p = false;
3149 assert (fastmap != NULL && p != NULL);
3152 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
3153 bufp->fastmap_accurate = 1; /* It will be when we're done. */
3154 bufp->can_be_null = 0;
3158 if (p == pend || *p == succeed)
3160 /* We have reached the (effective) end of pattern. */
3161 if (!FAIL_STACK_EMPTY ())
3163 bufp->can_be_null |= path_can_be_null;
3165 /* Reset for next path. */
3166 path_can_be_null = true;
3168 p = fail_stack.stack[--fail_stack.avail].pointer;
3176 /* We should never be about to go beyond the end of the pattern. */
3179 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3182 /* I guess the idea here is to simply not bother with a fastmap
3183 if a backreference is used, since it's too hard to figure out
3184 the fastmap for the corresponding group. Setting
3185 `can_be_null' stops `re_search_2' from using the fastmap, so
3186 that is all we do. */
3188 bufp->can_be_null = 1;
3192 /* Following are the cases which match a character. These end
3201 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3202 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3208 /* Chars beyond end of map must be allowed. */
3209 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3212 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3213 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3219 for (j = 0; j < (1 << BYTEWIDTH); j++)
3220 if (SYNTAX (j) == Sword)
3226 for (j = 0; j < (1 << BYTEWIDTH); j++)
3227 if (SYNTAX (j) != Sword)
3234 int fastmap_newline = fastmap['\n'];
3236 /* `.' matches anything ... */
3237 for (j = 0; j < (1 << BYTEWIDTH); j++)
3240 /* ... except perhaps newline. */
3241 if (!(bufp->syntax & RE_DOT_NEWLINE))
3242 fastmap['\n'] = fastmap_newline;
3244 /* Return if we have already set `can_be_null'; if we have,
3245 then the fastmap is irrelevant. Something's wrong here. */
3246 else if (bufp->can_be_null)
3249 /* Otherwise, have to check alternative paths. */
3256 for (j = 0; j < (1 << BYTEWIDTH); j++)
3257 if (SYNTAX (j) == (enum syntaxcode) k)
3264 for (j = 0; j < (1 << BYTEWIDTH); j++)
3265 if (SYNTAX (j) != (enum syntaxcode) k)
3270 /* All cases after this match the empty string. These end with
3290 case push_dummy_failure:
3295 case pop_failure_jump:
3296 case maybe_pop_jump:
3299 case dummy_failure_jump:
3300 EXTRACT_NUMBER_AND_INCR (j, p);
3305 /* Jump backward implies we just went through the body of a
3306 loop and matched nothing. Opcode jumped to should be
3307 `on_failure_jump' or `succeed_n'. Just treat it like an
3308 ordinary jump. For a * loop, it has pushed its failure
3309 point already; if so, discard that as redundant. */
3310 if ((re_opcode_t) *p != on_failure_jump
3311 && (re_opcode_t) *p != succeed_n)
3315 EXTRACT_NUMBER_AND_INCR (j, p);
3318 /* If what's on the stack is where we are now, pop it. */
3319 if (!FAIL_STACK_EMPTY ()
3320 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3326 case on_failure_jump:
3327 case on_failure_keep_string_jump:
3328 handle_on_failure_jump:
3329 EXTRACT_NUMBER_AND_INCR (j, p);
3331 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3332 end of the pattern. We don't want to push such a point,
3333 since when we restore it above, entering the switch will
3334 increment `p' past the end of the pattern. We don't need
3335 to push such a point since we obviously won't find any more
3336 fastmap entries beyond `pend'. Such a pattern can match
3337 the null string, though. */
3340 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3342 RESET_FAIL_STACK ();
3347 bufp->can_be_null = 1;
3351 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3352 succeed_n_p = false;
3359 /* Get to the number of times to succeed. */
3362 /* Increment p past the n for when k != 0. */
3363 EXTRACT_NUMBER_AND_INCR (k, p);
3367 succeed_n_p = true; /* Spaghetti code alert. */
3368 goto handle_on_failure_jump;
3385 abort (); /* We have listed all the cases. */
3388 /* Getting here means we have found the possible starting
3389 characters for one path of the pattern -- and that the empty
3390 string does not match. We need not follow this path further.
3391 Instead, look at the next alternative (remembered on the
3392 stack), or quit if no more. The test at the top of the loop
3393 does these things. */
3394 path_can_be_null = false;
3398 /* Set `can_be_null' for the last path (also the first path, if the
3399 pattern is empty). */
3400 bufp->can_be_null |= path_can_be_null;
3403 RESET_FAIL_STACK ();
3405 } /* re_compile_fastmap */
3407 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3408 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3409 this memory for recording register information. STARTS and ENDS
3410 must be allocated using the malloc library routine, and must each
3411 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3413 If NUM_REGS == 0, then subsequent matches should allocate their own
3416 Unless this function is called, the first search or match using
3417 PATTERN_BUFFER will allocate its own register data, without
3418 freeing the old data. */
3421 re_set_registers (bufp, regs, num_regs, starts, ends)
3422 struct re_pattern_buffer *bufp;
3423 struct re_registers *regs;
3425 regoff_t *starts, *ends;
3429 bufp->regs_allocated = REGS_REALLOCATE;
3430 regs->num_regs = num_regs;
3431 regs->start = starts;
3436 bufp->regs_allocated = REGS_UNALLOCATED;
3438 regs->start = regs->end = (regoff_t *) 0;
3442 /* Searching routines. */
3444 /* Like re_search_2, below, but only one string is specified, and
3445 doesn't let you say where to stop matching. */
3448 re_search (bufp, string, size, startpos, range, regs)
3449 struct re_pattern_buffer *bufp;
3451 int size, startpos, range;
3452 struct re_registers *regs;
3454 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3459 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3460 virtual concatenation of STRING1 and STRING2, starting first at index
3461 STARTPOS, then at STARTPOS + 1, and so on.
3463 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3465 RANGE is how far to scan while trying to match. RANGE = 0 means try
3466 only at STARTPOS; in general, the last start tried is STARTPOS +
3469 In REGS, return the indices of the virtual concatenation of STRING1
3470 and STRING2 that matched the entire BUFP->buffer and its contained
3473 Do not consider matching one past the index STOP in the virtual
3474 concatenation of STRING1 and STRING2.
3476 We return either the position in the strings at which the match was
3477 found, -1 if no match, or -2 if error (such as failure
3481 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3482 struct re_pattern_buffer *bufp;
3483 const char *string1, *string2;
3487 struct re_registers *regs;
3491 register char *fastmap = bufp->fastmap;
3492 register RE_TRANSLATE_TYPE translate = bufp->translate;
3493 int total_size = size1 + size2;
3494 int endpos = startpos + range;
3496 /* Check for out-of-range STARTPOS. */
3497 if (startpos < 0 || startpos > total_size)
3500 /* Fix up RANGE if it might eventually take us outside
3501 the virtual concatenation of STRING1 and STRING2.
3502 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3504 range = 0 - startpos;
3505 else if (endpos > total_size)
3506 range = total_size - startpos;
3508 /* If the search isn't to be a backwards one, don't waste time in a
3509 search for a pattern that must be anchored. */
3510 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3519 /* In a forward search for something that starts with \=.
3520 don't keep searching past point. */
3521 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3523 range = PT - startpos;
3529 /* Update the fastmap now if not correct already. */
3530 if (fastmap && !bufp->fastmap_accurate)
3531 if (re_compile_fastmap (bufp) == -2)
3534 /* Loop through the string, looking for a place to start matching. */
3537 /* If a fastmap is supplied, skip quickly over characters that
3538 cannot be the start of a match. If the pattern can match the
3539 null string, however, we don't need to skip characters; we want
3540 the first null string. */
3541 if (fastmap && startpos < total_size && !bufp->can_be_null)
3543 if (range > 0) /* Searching forwards. */
3545 register const char *d;
3546 register int lim = 0;
3549 if (startpos < size1 && startpos + range >= size1)
3550 lim = range - (size1 - startpos);
3552 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3554 /* Written out as an if-else to avoid testing `translate'
3558 && !fastmap[(unsigned char)
3559 translate[(unsigned char) *d++]])
3562 while (range > lim && !fastmap[(unsigned char) *d++])
3565 startpos += irange - range;
3567 else /* Searching backwards. */
3569 register char c = (size1 == 0 || startpos >= size1
3570 ? string2[startpos - size1]
3571 : string1[startpos]);
3573 if (!fastmap[(unsigned char) TRANSLATE (c)])
3578 /* If can't match the null string, and that's all we have left, fail. */
3579 if (range >= 0 && startpos == total_size && fastmap
3580 && !bufp->can_be_null)
3583 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3584 startpos, regs, stop);
3585 #ifndef REGEX_MALLOC
3614 /* This converts PTR, a pointer into one of the search strings `string1'
3615 and `string2' into an offset from the beginning of that string. */
3616 #define POINTER_TO_OFFSET(ptr) \
3617 (FIRST_STRING_P (ptr) \
3618 ? ((regoff_t) ((ptr) - string1)) \
3619 : ((regoff_t) ((ptr) - string2 + size1)))
3621 /* Macros for dealing with the split strings in re_match_2. */
3623 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3625 /* Call before fetching a character with *d. This switches over to
3626 string2 if necessary. */
3627 #define PREFETCH() \
3630 /* End of string2 => fail. */ \
3631 if (dend == end_match_2) \
3633 /* End of string1 => advance to string2. */ \
3635 dend = end_match_2; \
3639 /* Test if at very beginning or at very end of the virtual concatenation
3640 of `string1' and `string2'. If only one string, it's `string2'. */
3641 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3642 #define AT_STRINGS_END(d) ((d) == end2)
3645 /* Test if D points to a character which is word-constituent. We have
3646 two special cases to check for: if past the end of string1, look at
3647 the first character in string2; and if before the beginning of
3648 string2, look at the last character in string1. */
3649 #define WORDCHAR_P(d) \
3650 (SYNTAX ((d) == end1 ? *string2 \
3651 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3654 /* Disabled due to a compiler bug -- see comment at case wordbound */
3656 /* Test if the character before D and the one at D differ with respect
3657 to being word-constituent. */
3658 #define AT_WORD_BOUNDARY(d) \
3659 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3660 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3663 /* Free everything we malloc. */
3664 #ifdef MATCH_MAY_ALLOCATE
3665 #define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3666 #define FREE_VARIABLES() \
3668 REGEX_FREE_STACK (fail_stack.stack); \
3669 FREE_VAR (regstart); \
3670 FREE_VAR (regend); \
3671 FREE_VAR (old_regstart); \
3672 FREE_VAR (old_regend); \
3673 FREE_VAR (best_regstart); \
3674 FREE_VAR (best_regend); \
3675 FREE_VAR (reg_info); \
3676 FREE_VAR (reg_dummy); \
3677 FREE_VAR (reg_info_dummy); \
3680 #define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3681 #endif /* not MATCH_MAY_ALLOCATE */
3683 /* These values must meet several constraints. They must not be valid
3684 register values; since we have a limit of 255 registers (because
3685 we use only one byte in the pattern for the register number), we can
3686 use numbers larger than 255. They must differ by 1, because of
3687 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3688 be larger than the value for the highest register, so we do not try
3689 to actually save any registers when none are active. */
3690 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3691 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3693 /* Matching routines. */
3695 #ifndef emacs /* Emacs never uses this. */
3696 /* re_match is like re_match_2 except it takes only a single string. */
3699 re_match (bufp, string, size, pos, regs)
3700 struct re_pattern_buffer *bufp;
3703 struct re_registers *regs;
3705 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3707 #ifndef REGEX_MALLOC
3714 #endif /* not emacs */
3716 static boolean group_match_null_string_p _RE_ARGS ((unsigned char **p,
3718 register_info_type *reg_info));
3719 static boolean alt_match_null_string_p _RE_ARGS ((unsigned char *p,
3721 register_info_type *reg_info));
3722 static boolean common_op_match_null_string_p _RE_ARGS ((unsigned char **p,
3724 register_info_type *reg_info));
3725 static int bcmp_translate _RE_ARGS ((const char *s1, const char *s2,
3726 int len, char *translate));
3728 /* re_match_2 matches the compiled pattern in BUFP against the
3729 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3730 and SIZE2, respectively). We start matching at POS, and stop
3733 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3734 store offsets for the substring each group matched in REGS. See the
3735 documentation for exactly how many groups we fill.
3737 We return -1 if no match, -2 if an internal error (such as the
3738 failure stack overflowing). Otherwise, we return the length of the
3739 matched substring. */
3742 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3743 struct re_pattern_buffer *bufp;
3744 const char *string1, *string2;
3747 struct re_registers *regs;
3750 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3752 #ifndef REGEX_MALLOC
3760 /* This is a separate function so that we can force an alloca cleanup
3763 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3764 struct re_pattern_buffer *bufp;
3765 const char *string1, *string2;
3768 struct re_registers *regs;
3771 /* General temporaries. */
3775 /* Just past the end of the corresponding string. */
3776 const char *end1, *end2;
3778 /* Pointers into string1 and string2, just past the last characters in
3779 each to consider matching. */
3780 const char *end_match_1, *end_match_2;
3782 /* Where we are in the data, and the end of the current string. */
3783 const char *d, *dend;
3785 /* Where we are in the pattern, and the end of the pattern. */
3786 unsigned char *p = bufp->buffer;
3787 register unsigned char *pend = p + bufp->used;
3789 /* Mark the opcode just after a start_memory, so we can test for an
3790 empty subpattern when we get to the stop_memory. */
3791 unsigned char *just_past_start_mem = 0;
3793 /* We use this to map every character in the string. */
3794 RE_TRANSLATE_TYPE translate = bufp->translate;
3796 /* Failure point stack. Each place that can handle a failure further
3797 down the line pushes a failure point on this stack. It consists of
3798 restart, regend, and reg_info for all registers corresponding to
3799 the subexpressions we're currently inside, plus the number of such
3800 registers, and, finally, two char *'s. The first char * is where
3801 to resume scanning the pattern; the second one is where to resume
3802 scanning the strings. If the latter is zero, the failure point is
3803 a ``dummy''; if a failure happens and the failure point is a dummy,
3804 it gets discarded and the next next one is tried. */
3805 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3806 fail_stack_type fail_stack;
3809 static unsigned failure_id = 0;
3810 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3814 /* This holds the pointer to the failure stack, when
3815 it is allocated relocatably. */
3816 fail_stack_elt_t *failure_stack_ptr;
3819 /* We fill all the registers internally, independent of what we
3820 return, for use in backreferences. The number here includes
3821 an element for register zero. */
3822 size_t num_regs = bufp->re_nsub + 1;
3824 /* The currently active registers. */
3825 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3826 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3828 /* Information on the contents of registers. These are pointers into
3829 the input strings; they record just what was matched (on this
3830 attempt) by a subexpression part of the pattern, that is, the
3831 regnum-th regstart pointer points to where in the pattern we began
3832 matching and the regnum-th regend points to right after where we
3833 stopped matching the regnum-th subexpression. (The zeroth register
3834 keeps track of what the whole pattern matches.) */
3835 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3836 const char **regstart, **regend;
3839 /* If a group that's operated upon by a repetition operator fails to
3840 match anything, then the register for its start will need to be
3841 restored because it will have been set to wherever in the string we
3842 are when we last see its open-group operator. Similarly for a
3844 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3845 const char **old_regstart, **old_regend;
3848 /* The is_active field of reg_info helps us keep track of which (possibly
3849 nested) subexpressions we are currently in. The matched_something
3850 field of reg_info[reg_num] helps us tell whether or not we have
3851 matched any of the pattern so far this time through the reg_num-th
3852 subexpression. These two fields get reset each time through any
3853 loop their register is in. */
3854 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3855 register_info_type *reg_info;
3858 /* The following record the register info as found in the above
3859 variables when we find a match better than any we've seen before.
3860 This happens as we backtrack through the failure points, which in
3861 turn happens only if we have not yet matched the entire string. */
3862 unsigned best_regs_set = false;
3863 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3864 const char **best_regstart, **best_regend;
3867 /* Logically, this is `best_regend[0]'. But we don't want to have to
3868 allocate space for that if we're not allocating space for anything
3869 else (see below). Also, we never need info about register 0 for
3870 any of the other register vectors, and it seems rather a kludge to
3871 treat `best_regend' differently than the rest. So we keep track of
3872 the end of the best match so far in a separate variable. We
3873 initialize this to NULL so that when we backtrack the first time
3874 and need to test it, it's not garbage. */
3875 const char *match_end = NULL;
3877 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3878 int set_regs_matched_done = 0;
3880 /* Used when we pop values we don't care about. */
3881 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3882 const char **reg_dummy;
3883 register_info_type *reg_info_dummy;
3887 /* Counts the total number of registers pushed. */
3888 unsigned num_regs_pushed = 0;
3891 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3895 #ifdef MATCH_MAY_ALLOCATE
3896 /* Do not bother to initialize all the register variables if there are
3897 no groups in the pattern, as it takes a fair amount of time. If
3898 there are groups, we include space for register 0 (the whole
3899 pattern), even though we never use it, since it simplifies the
3900 array indexing. We should fix this. */
3903 regstart = REGEX_TALLOC (num_regs, const char *);
3904 regend = REGEX_TALLOC (num_regs, const char *);
3905 old_regstart = REGEX_TALLOC (num_regs, const char *);
3906 old_regend = REGEX_TALLOC (num_regs, const char *);
3907 best_regstart = REGEX_TALLOC (num_regs, const char *);
3908 best_regend = REGEX_TALLOC (num_regs, const char *);
3909 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3910 reg_dummy = REGEX_TALLOC (num_regs, const char *);
3911 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3913 if (!(regstart && regend && old_regstart && old_regend && reg_info
3914 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3922 /* We must initialize all our variables to NULL, so that
3923 `FREE_VARIABLES' doesn't try to free them. */
3924 regstart = regend = old_regstart = old_regend = best_regstart
3925 = best_regend = reg_dummy = NULL;
3926 reg_info = reg_info_dummy = (register_info_type *) NULL;
3928 #endif /* MATCH_MAY_ALLOCATE */
3930 /* The starting position is bogus. */
3931 if (pos < 0 || pos > size1 + size2)
3937 /* Initialize subexpression text positions to -1 to mark ones that no
3938 start_memory/stop_memory has been seen for. Also initialize the
3939 register information struct. */
3940 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
3942 regstart[mcnt] = regend[mcnt]
3943 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3945 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3946 IS_ACTIVE (reg_info[mcnt]) = 0;
3947 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3948 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3951 /* We move `string1' into `string2' if the latter's empty -- but not if
3952 `string1' is null. */
3953 if (size2 == 0 && string1 != NULL)
3960 end1 = string1 + size1;
3961 end2 = string2 + size2;
3963 /* Compute where to stop matching, within the two strings. */
3966 end_match_1 = string1 + stop;
3967 end_match_2 = string2;
3972 end_match_2 = string2 + stop - size1;
3975 /* `p' scans through the pattern as `d' scans through the data.
3976 `dend' is the end of the input string that `d' points within. `d'
3977 is advanced into the following input string whenever necessary, but
3978 this happens before fetching; therefore, at the beginning of the
3979 loop, `d' can be pointing at the end of a string, but it cannot
3981 if (size1 > 0 && pos <= size1)
3988 d = string2 + pos - size1;
3992 DEBUG_PRINT1 ("The compiled pattern is:\n");
3993 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
3994 DEBUG_PRINT1 ("The string to match is: `");
3995 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
3996 DEBUG_PRINT1 ("'\n");
3998 /* This loops over pattern commands. It exits by returning from the
3999 function if the match is complete, or it drops through if the match
4000 fails at this starting point in the input data. */
4004 DEBUG_PRINT2 ("\n%p: ", p);
4006 DEBUG_PRINT2 ("\n0x%x: ", p);
4010 { /* End of pattern means we might have succeeded. */
4011 DEBUG_PRINT1 ("end of pattern ... ");
4013 /* If we haven't matched the entire string, and we want the
4014 longest match, try backtracking. */
4015 if (d != end_match_2)
4017 /* 1 if this match ends in the same string (string1 or string2)
4018 as the best previous match. */
4019 boolean same_str_p = (FIRST_STRING_P (match_end)
4020 == MATCHING_IN_FIRST_STRING);
4021 /* 1 if this match is the best seen so far. */
4022 boolean best_match_p;
4024 /* AIX compiler got confused when this was combined
4025 with the previous declaration. */
4027 best_match_p = d > match_end;
4029 best_match_p = !MATCHING_IN_FIRST_STRING;
4031 DEBUG_PRINT1 ("backtracking.\n");
4033 if (!FAIL_STACK_EMPTY ())
4034 { /* More failure points to try. */
4036 /* If exceeds best match so far, save it. */
4037 if (!best_regs_set || best_match_p)
4039 best_regs_set = true;
4042 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
4044 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4046 best_regstart[mcnt] = regstart[mcnt];
4047 best_regend[mcnt] = regend[mcnt];
4053 /* If no failure points, don't restore garbage. And if
4054 last match is real best match, don't restore second
4056 else if (best_regs_set && !best_match_p)
4059 /* Restore best match. It may happen that `dend ==
4060 end_match_1' while the restored d is in string2.
4061 For example, the pattern `x.*y.*z' against the
4062 strings `x-' and `y-z-', if the two strings are
4063 not consecutive in memory. */
4064 DEBUG_PRINT1 ("Restoring best registers.\n");
4067 dend = ((d >= string1 && d <= end1)
4068 ? end_match_1 : end_match_2);
4070 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4072 regstart[mcnt] = best_regstart[mcnt];
4073 regend[mcnt] = best_regend[mcnt];
4076 } /* d != end_match_2 */
4079 DEBUG_PRINT1 ("Accepting match.\n");
4081 /* If caller wants register contents data back, do it. */
4082 if (regs && !bufp->no_sub)
4084 /* Have the register data arrays been allocated? */
4085 if (bufp->regs_allocated == REGS_UNALLOCATED)
4086 { /* No. So allocate them with malloc. We need one
4087 extra element beyond `num_regs' for the `-1' marker
4089 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
4090 regs->start = TALLOC (regs->num_regs, regoff_t);
4091 regs->end = TALLOC (regs->num_regs, regoff_t);
4092 if (regs->start == NULL || regs->end == NULL)
4097 bufp->regs_allocated = REGS_REALLOCATE;
4099 else if (bufp->regs_allocated == REGS_REALLOCATE)
4100 { /* Yes. If we need more elements than were already
4101 allocated, reallocate them. If we need fewer, just
4103 if (regs->num_regs < num_regs + 1)
4105 regs->num_regs = num_regs + 1;
4106 RETALLOC (regs->start, regs->num_regs, regoff_t);
4107 RETALLOC (regs->end, regs->num_regs, regoff_t);
4108 if (regs->start == NULL || regs->end == NULL)
4117 /* These braces fend off a "empty body in an else-statement"
4118 warning under GCC when assert expands to nothing. */
4119 assert (bufp->regs_allocated == REGS_FIXED);
4122 /* Convert the pointer data in `regstart' and `regend' to
4123 indices. Register zero has to be set differently,
4124 since we haven't kept track of any info for it. */
4125 if (regs->num_regs > 0)
4127 regs->start[0] = pos;
4128 regs->end[0] = (MATCHING_IN_FIRST_STRING
4129 ? ((regoff_t) (d - string1))
4130 : ((regoff_t) (d - string2 + size1)));
4133 /* Go through the first `min (num_regs, regs->num_regs)'
4134 registers, since that is all we initialized. */
4135 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
4138 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
4139 regs->start[mcnt] = regs->end[mcnt] = -1;
4143 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
4145 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
4149 /* If the regs structure we return has more elements than
4150 were in the pattern, set the extra elements to -1. If
4151 we (re)allocated the registers, this is the case,
4152 because we always allocate enough to have at least one
4154 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
4155 regs->start[mcnt] = regs->end[mcnt] = -1;
4156 } /* regs && !bufp->no_sub */
4158 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
4159 nfailure_points_pushed, nfailure_points_popped,
4160 nfailure_points_pushed - nfailure_points_popped);
4161 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
4163 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
4167 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
4173 /* Otherwise match next pattern command. */
4174 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4176 /* Ignore these. Used to ignore the n of succeed_n's which
4177 currently have n == 0. */
4179 DEBUG_PRINT1 ("EXECUTING no_op.\n");
4183 DEBUG_PRINT1 ("EXECUTING succeed.\n");
4186 /* Match the next n pattern characters exactly. The following
4187 byte in the pattern defines n, and the n bytes after that
4188 are the characters to match. */
4191 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
4193 /* This is written out as an if-else so we don't waste time
4194 testing `translate' inside the loop. */
4200 if ((unsigned char) translate[(unsigned char) *d++]
4201 != (unsigned char) *p++)
4211 if (*d++ != (char) *p++) goto fail;
4215 SET_REGS_MATCHED ();
4219 /* Match any character except possibly a newline or a null. */
4221 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4225 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
4226 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4229 SET_REGS_MATCHED ();
4230 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4238 register unsigned char c;
4239 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4241 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4244 c = TRANSLATE (*d); /* The character to match. */
4246 /* Cast to `unsigned' instead of `unsigned char' in case the
4247 bit list is a full 32 bytes long. */
4248 if (c < (unsigned) (*p * BYTEWIDTH)
4249 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4254 if (!not) goto fail;
4256 SET_REGS_MATCHED ();
4262 /* The beginning of a group is represented by start_memory.
4263 The arguments are the register number in the next byte, and the
4264 number of groups inner to this one in the next. The text
4265 matched within the group is recorded (in the internal
4266 registers data structure) under the register number. */
4268 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4270 /* Find out if this group can match the empty string. */
4271 p1 = p; /* To send to group_match_null_string_p. */
4273 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4274 REG_MATCH_NULL_STRING_P (reg_info[*p])
4275 = group_match_null_string_p (&p1, pend, reg_info);
4277 /* Save the position in the string where we were the last time
4278 we were at this open-group operator in case the group is
4279 operated upon by a repetition operator, e.g., with `(a*)*b'
4280 against `ab'; then we want to ignore where we are now in
4281 the string in case this attempt to match fails. */
4282 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4283 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4285 DEBUG_PRINT2 (" old_regstart: %d\n",
4286 POINTER_TO_OFFSET (old_regstart[*p]));
4289 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4291 IS_ACTIVE (reg_info[*p]) = 1;
4292 MATCHED_SOMETHING (reg_info[*p]) = 0;
4294 /* Clear this whenever we change the register activity status. */
4295 set_regs_matched_done = 0;
4297 /* This is the new highest active register. */
4298 highest_active_reg = *p;
4300 /* If nothing was active before, this is the new lowest active
4302 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4303 lowest_active_reg = *p;
4305 /* Move past the register number and inner group count. */
4307 just_past_start_mem = p;
4312 /* The stop_memory opcode represents the end of a group. Its
4313 arguments are the same as start_memory's: the register
4314 number, and the number of inner groups. */
4316 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4318 /* We need to save the string position the last time we were at
4319 this close-group operator in case the group is operated
4320 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4321 against `aba'; then we want to ignore where we are now in
4322 the string in case this attempt to match fails. */
4323 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4324 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4326 DEBUG_PRINT2 (" old_regend: %d\n",
4327 POINTER_TO_OFFSET (old_regend[*p]));
4330 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4332 /* This register isn't active anymore. */
4333 IS_ACTIVE (reg_info[*p]) = 0;
4335 /* Clear this whenever we change the register activity status. */
4336 set_regs_matched_done = 0;
4338 /* If this was the only register active, nothing is active
4340 if (lowest_active_reg == highest_active_reg)
4342 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4343 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4346 { /* We must scan for the new highest active register, since
4347 it isn't necessarily one less than now: consider
4348 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4349 new highest active register is 1. */
4350 unsigned char r = *p - 1;
4351 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4354 /* If we end up at register zero, that means that we saved
4355 the registers as the result of an `on_failure_jump', not
4356 a `start_memory', and we jumped to past the innermost
4357 `stop_memory'. For example, in ((.)*) we save
4358 registers 1 and 2 as a result of the *, but when we pop
4359 back to the second ), we are at the stop_memory 1.
4360 Thus, nothing is active. */
4363 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4364 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4367 highest_active_reg = r;
4370 /* If just failed to match something this time around with a
4371 group that's operated on by a repetition operator, try to
4372 force exit from the ``loop'', and restore the register
4373 information for this group that we had before trying this
4375 if ((!MATCHED_SOMETHING (reg_info[*p])
4376 || just_past_start_mem == p - 1)
4379 boolean is_a_jump_n = false;
4383 switch ((re_opcode_t) *p1++)
4387 case pop_failure_jump:
4388 case maybe_pop_jump:
4390 case dummy_failure_jump:
4391 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4401 /* If the next operation is a jump backwards in the pattern
4402 to an on_failure_jump right before the start_memory
4403 corresponding to this stop_memory, exit from the loop
4404 by forcing a failure after pushing on the stack the
4405 on_failure_jump's jump in the pattern, and d. */
4406 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4407 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4409 /* If this group ever matched anything, then restore
4410 what its registers were before trying this last
4411 failed match, e.g., with `(a*)*b' against `ab' for
4412 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4413 against `aba' for regend[3].
4415 Also restore the registers for inner groups for,
4416 e.g., `((a*)(b*))*' against `aba' (register 3 would
4417 otherwise get trashed). */
4419 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4423 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4425 /* Restore this and inner groups' (if any) registers. */
4426 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
4429 regstart[r] = old_regstart[r];
4431 /* xx why this test? */
4432 if (old_regend[r] >= regstart[r])
4433 regend[r] = old_regend[r];
4437 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4438 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4444 /* Move past the register number and the inner group count. */
4449 /* \<digit> has been turned into a `duplicate' command which is
4450 followed by the numeric value of <digit> as the register number. */
4453 register const char *d2, *dend2;
4454 int regno = *p++; /* Get which register to match against. */
4455 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4457 /* Can't back reference a group which we've never matched. */
4458 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4461 /* Where in input to try to start matching. */
4462 d2 = regstart[regno];
4464 /* Where to stop matching; if both the place to start and
4465 the place to stop matching are in the same string, then
4466 set to the place to stop, otherwise, for now have to use
4467 the end of the first string. */
4469 dend2 = ((FIRST_STRING_P (regstart[regno])
4470 == FIRST_STRING_P (regend[regno]))
4471 ? regend[regno] : end_match_1);
4474 /* If necessary, advance to next segment in register
4478 if (dend2 == end_match_2) break;
4479 if (dend2 == regend[regno]) break;
4481 /* End of string1 => advance to string2. */
4483 dend2 = regend[regno];
4485 /* At end of register contents => success */
4486 if (d2 == dend2) break;
4488 /* If necessary, advance to next segment in data. */
4491 /* How many characters left in this segment to match. */
4494 /* Want how many consecutive characters we can match in
4495 one shot, so, if necessary, adjust the count. */
4496 if (mcnt > dend2 - d2)
4499 /* Compare that many; failure if mismatch, else move
4502 ? bcmp_translate (d, d2, mcnt, translate)
4503 : bcmp (d, d2, mcnt))
4505 d += mcnt, d2 += mcnt;
4507 /* Do this because we've match some characters. */
4508 SET_REGS_MATCHED ();
4514 /* begline matches the empty string at the beginning of the string
4515 (unless `not_bol' is set in `bufp'), and, if
4516 `newline_anchor' is set, after newlines. */
4518 DEBUG_PRINT1 ("EXECUTING begline.\n");
4520 if (AT_STRINGS_BEG (d))
4522 if (!bufp->not_bol) break;
4524 else if (d[-1] == '\n' && bufp->newline_anchor)
4528 /* In all other cases, we fail. */
4532 /* endline is the dual of begline. */
4534 DEBUG_PRINT1 ("EXECUTING endline.\n");
4536 if (AT_STRINGS_END (d))
4538 if (!bufp->not_eol) break;
4541 /* We have to ``prefetch'' the next character. */
4542 else if ((d == end1 ? *string2 : *d) == '\n'
4543 && bufp->newline_anchor)
4550 /* Match at the very beginning of the data. */
4552 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4553 if (AT_STRINGS_BEG (d))
4558 /* Match at the very end of the data. */
4560 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4561 if (AT_STRINGS_END (d))
4566 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4567 pushes NULL as the value for the string on the stack. Then
4568 `pop_failure_point' will keep the current value for the
4569 string, instead of restoring it. To see why, consider
4570 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4571 then the . fails against the \n. But the next thing we want
4572 to do is match the \n against the \n; if we restored the
4573 string value, we would be back at the foo.
4575 Because this is used only in specific cases, we don't need to
4576 check all the things that `on_failure_jump' does, to make
4577 sure the right things get saved on the stack. Hence we don't
4578 share its code. The only reason to push anything on the
4579 stack at all is that otherwise we would have to change
4580 `anychar's code to do something besides goto fail in this
4581 case; that seems worse than this. */
4582 case on_failure_keep_string_jump:
4583 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4585 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4587 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
4589 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4592 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4596 /* Uses of on_failure_jump:
4598 Each alternative starts with an on_failure_jump that points
4599 to the beginning of the next alternative. Each alternative
4600 except the last ends with a jump that in effect jumps past
4601 the rest of the alternatives. (They really jump to the
4602 ending jump of the following alternative, because tensioning
4603 these jumps is a hassle.)
4605 Repeats start with an on_failure_jump that points past both
4606 the repetition text and either the following jump or
4607 pop_failure_jump back to this on_failure_jump. */
4608 case on_failure_jump:
4610 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4612 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4614 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
4616 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4619 /* If this on_failure_jump comes right before a group (i.e.,
4620 the original * applied to a group), save the information
4621 for that group and all inner ones, so that if we fail back
4622 to this point, the group's information will be correct.
4623 For example, in \(a*\)*\1, we need the preceding group,
4624 and in \(zz\(a*\)b*\)\2, we need the inner group. */
4626 /* We can't use `p' to check ahead because we push
4627 a failure point to `p + mcnt' after we do this. */
4630 /* We need to skip no_op's before we look for the
4631 start_memory in case this on_failure_jump is happening as
4632 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4634 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4637 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4639 /* We have a new highest active register now. This will
4640 get reset at the start_memory we are about to get to,
4641 but we will have saved all the registers relevant to
4642 this repetition op, as described above. */
4643 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4644 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4645 lowest_active_reg = *(p1 + 1);
4648 DEBUG_PRINT1 (":\n");
4649 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4653 /* A smart repeat ends with `maybe_pop_jump'.
4654 We change it to either `pop_failure_jump' or `jump'. */
4655 case maybe_pop_jump:
4656 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4657 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4659 register unsigned char *p2 = p;
4661 /* Compare the beginning of the repeat with what in the
4662 pattern follows its end. If we can establish that there
4663 is nothing that they would both match, i.e., that we
4664 would have to backtrack because of (as in, e.g., `a*a')
4665 then we can change to pop_failure_jump, because we'll
4666 never have to backtrack.
4668 This is not true in the case of alternatives: in
4669 `(a|ab)*' we do need to backtrack to the `ab' alternative
4670 (e.g., if the string was `ab'). But instead of trying to
4671 detect that here, the alternative has put on a dummy
4672 failure point which is what we will end up popping. */
4674 /* Skip over open/close-group commands.
4675 If what follows this loop is a ...+ construct,
4676 look at what begins its body, since we will have to
4677 match at least one of that. */
4681 && ((re_opcode_t) *p2 == stop_memory
4682 || (re_opcode_t) *p2 == start_memory))
4684 else if (p2 + 6 < pend
4685 && (re_opcode_t) *p2 == dummy_failure_jump)
4692 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4693 to the `maybe_finalize_jump' of this case. Examine what
4696 /* If we're at the end of the pattern, we can change. */
4699 /* Consider what happens when matching ":\(.*\)"
4700 against ":/". I don't really understand this code
4702 p[-3] = (unsigned char) pop_failure_jump;
4704 (" End of pattern: change to `pop_failure_jump'.\n");
4707 else if ((re_opcode_t) *p2 == exactn
4708 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4710 register unsigned char c
4711 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4713 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4715 p[-3] = (unsigned char) pop_failure_jump;
4716 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4720 else if ((re_opcode_t) p1[3] == charset
4721 || (re_opcode_t) p1[3] == charset_not)
4723 int not = (re_opcode_t) p1[3] == charset_not;
4725 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4726 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4729 /* `not' is equal to 1 if c would match, which means
4730 that we can't change to pop_failure_jump. */
4733 p[-3] = (unsigned char) pop_failure_jump;
4734 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4738 else if ((re_opcode_t) *p2 == charset)
4741 register unsigned char c
4742 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4746 if ((re_opcode_t) p1[3] == exactn
4747 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
4748 && (p2[2 + p1[5] / BYTEWIDTH]
4749 & (1 << (p1[5] % BYTEWIDTH)))))
4751 if ((re_opcode_t) p1[3] == exactn
4752 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[4]
4753 && (p2[2 + p1[4] / BYTEWIDTH]
4754 & (1 << (p1[4] % BYTEWIDTH)))))
4757 p[-3] = (unsigned char) pop_failure_jump;
4758 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4762 else if ((re_opcode_t) p1[3] == charset_not)
4765 /* We win if the charset_not inside the loop
4766 lists every character listed in the charset after. */
4767 for (idx = 0; idx < (int) p2[1]; idx++)
4768 if (! (p2[2 + idx] == 0
4769 || (idx < (int) p1[4]
4770 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4775 p[-3] = (unsigned char) pop_failure_jump;
4776 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4779 else if ((re_opcode_t) p1[3] == charset)
4782 /* We win if the charset inside the loop
4783 has no overlap with the one after the loop. */
4785 idx < (int) p2[1] && idx < (int) p1[4];
4787 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4790 if (idx == p2[1] || idx == p1[4])
4792 p[-3] = (unsigned char) pop_failure_jump;
4793 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4798 p -= 2; /* Point at relative address again. */
4799 if ((re_opcode_t) p[-1] != pop_failure_jump)
4801 p[-1] = (unsigned char) jump;
4802 DEBUG_PRINT1 (" Match => jump.\n");
4803 goto unconditional_jump;
4805 /* Note fall through. */
4808 /* The end of a simple repeat has a pop_failure_jump back to
4809 its matching on_failure_jump, where the latter will push a
4810 failure point. The pop_failure_jump takes off failure
4811 points put on by this pop_failure_jump's matching
4812 on_failure_jump; we got through the pattern to here from the
4813 matching on_failure_jump, so didn't fail. */
4814 case pop_failure_jump:
4816 /* We need to pass separate storage for the lowest and
4817 highest registers, even though we don't care about the
4818 actual values. Otherwise, we will restore only one
4819 register from the stack, since lowest will == highest in
4820 `pop_failure_point'. */
4821 active_reg_t dummy_low_reg, dummy_high_reg;
4822 unsigned char *pdummy;
4825 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4826 POP_FAILURE_POINT (sdummy, pdummy,
4827 dummy_low_reg, dummy_high_reg,
4828 reg_dummy, reg_dummy, reg_info_dummy);
4830 /* Note fall through. */
4834 DEBUG_PRINT2 ("\n%p: ", p);
4836 DEBUG_PRINT2 ("\n0x%x: ", p);
4838 /* Note fall through. */
4840 /* Unconditionally jump (without popping any failure points). */
4842 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4843 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4844 p += mcnt; /* Do the jump. */
4846 DEBUG_PRINT2 ("(to %p).\n", p);
4848 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4853 /* We need this opcode so we can detect where alternatives end
4854 in `group_match_null_string_p' et al. */
4856 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4857 goto unconditional_jump;
4860 /* Normally, the on_failure_jump pushes a failure point, which
4861 then gets popped at pop_failure_jump. We will end up at
4862 pop_failure_jump, also, and with a pattern of, say, `a+', we
4863 are skipping over the on_failure_jump, so we have to push
4864 something meaningless for pop_failure_jump to pop. */
4865 case dummy_failure_jump:
4866 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4867 /* It doesn't matter what we push for the string here. What
4868 the code at `fail' tests is the value for the pattern. */
4869 PUSH_FAILURE_POINT (0, 0, -2);
4870 goto unconditional_jump;
4873 /* At the end of an alternative, we need to push a dummy failure
4874 point in case we are followed by a `pop_failure_jump', because
4875 we don't want the failure point for the alternative to be
4876 popped. For example, matching `(a|ab)*' against `aab'
4877 requires that we match the `ab' alternative. */
4878 case push_dummy_failure:
4879 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4880 /* See comments just above at `dummy_failure_jump' about the
4882 PUSH_FAILURE_POINT (0, 0, -2);
4885 /* Have to succeed matching what follows at least n times.
4886 After that, handle like `on_failure_jump'. */
4888 EXTRACT_NUMBER (mcnt, p + 2);
4889 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4892 /* Originally, this is how many times we HAVE to succeed. */
4897 STORE_NUMBER_AND_INCR (p, mcnt);
4899 DEBUG_PRINT3 (" Setting %p to %d.\n", p - 2, mcnt);
4901 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - 2, mcnt);
4907 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n", p+2);
4909 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4911 p[2] = (unsigned char) no_op;
4912 p[3] = (unsigned char) no_op;
4918 EXTRACT_NUMBER (mcnt, p + 2);
4919 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4921 /* Originally, this is how many times we CAN jump. */
4925 STORE_NUMBER (p + 2, mcnt);
4927 DEBUG_PRINT3 (" Setting %p to %d.\n", p + 2, mcnt);
4929 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + 2, mcnt);
4931 goto unconditional_jump;
4933 /* If don't have to jump any more, skip over the rest of command. */
4940 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4942 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4944 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4946 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
4948 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4950 STORE_NUMBER (p1, mcnt);
4955 /* The DEC Alpha C compiler 3.x generates incorrect code for the
4956 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4957 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4958 macro and introducing temporary variables works around the bug. */
4961 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4962 if (AT_WORD_BOUNDARY (d))
4967 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4968 if (AT_WORD_BOUNDARY (d))
4974 boolean prevchar, thischar;
4976 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4977 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4980 prevchar = WORDCHAR_P (d - 1);
4981 thischar = WORDCHAR_P (d);
4982 if (prevchar != thischar)
4989 boolean prevchar, thischar;
4991 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4992 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4995 prevchar = WORDCHAR_P (d - 1);
4996 thischar = WORDCHAR_P (d);
4997 if (prevchar != thischar)
5004 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
5005 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
5010 DEBUG_PRINT1 ("EXECUTING wordend.\n");
5011 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
5012 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
5018 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
5019 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
5024 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
5025 if (PTR_CHAR_POS ((unsigned char *) d) != point)
5030 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
5031 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
5036 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
5041 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
5045 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5047 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
5049 SET_REGS_MATCHED ();
5053 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
5055 goto matchnotsyntax;
5058 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
5062 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5064 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
5066 SET_REGS_MATCHED ();
5069 #else /* not emacs */
5071 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
5073 if (!WORDCHAR_P (d))
5075 SET_REGS_MATCHED ();
5080 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
5084 SET_REGS_MATCHED ();
5087 #endif /* not emacs */
5092 continue; /* Successfully executed one pattern command; keep going. */
5095 /* We goto here if a matching operation fails. */
5097 if (!FAIL_STACK_EMPTY ())
5098 { /* A restart point is known. Restore to that state. */
5099 DEBUG_PRINT1 ("\nFAIL:\n");
5100 POP_FAILURE_POINT (d, p,
5101 lowest_active_reg, highest_active_reg,
5102 regstart, regend, reg_info);
5104 /* If this failure point is a dummy, try the next one. */
5108 /* If we failed to the end of the pattern, don't examine *p. */
5112 boolean is_a_jump_n = false;
5114 /* If failed to a backwards jump that's part of a repetition
5115 loop, need to pop this failure point and use the next one. */
5116 switch ((re_opcode_t) *p)
5120 case maybe_pop_jump:
5121 case pop_failure_jump:
5124 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5127 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
5129 && (re_opcode_t) *p1 == on_failure_jump))
5137 if (d >= string1 && d <= end1)
5141 break; /* Matching at this starting point really fails. */
5145 goto restore_best_regs;
5149 return -1; /* Failure to match. */
5152 /* Subroutine definitions for re_match_2. */
5155 /* We are passed P pointing to a register number after a start_memory.
5157 Return true if the pattern up to the corresponding stop_memory can
5158 match the empty string, and false otherwise.
5160 If we find the matching stop_memory, sets P to point to one past its number.
5161 Otherwise, sets P to an undefined byte less than or equal to END.
5163 We don't handle duplicates properly (yet). */
5166 group_match_null_string_p (p, end, reg_info)
5167 unsigned char **p, *end;
5168 register_info_type *reg_info;
5171 /* Point to after the args to the start_memory. */
5172 unsigned char *p1 = *p + 2;
5176 /* Skip over opcodes that can match nothing, and return true or
5177 false, as appropriate, when we get to one that can't, or to the
5178 matching stop_memory. */
5180 switch ((re_opcode_t) *p1)
5182 /* Could be either a loop or a series of alternatives. */
5183 case on_failure_jump:
5185 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5187 /* If the next operation is not a jump backwards in the
5192 /* Go through the on_failure_jumps of the alternatives,
5193 seeing if any of the alternatives cannot match nothing.
5194 The last alternative starts with only a jump,
5195 whereas the rest start with on_failure_jump and end
5196 with a jump, e.g., here is the pattern for `a|b|c':
5198 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
5199 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
5202 So, we have to first go through the first (n-1)
5203 alternatives and then deal with the last one separately. */
5206 /* Deal with the first (n-1) alternatives, which start
5207 with an on_failure_jump (see above) that jumps to right
5208 past a jump_past_alt. */
5210 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
5212 /* `mcnt' holds how many bytes long the alternative
5213 is, including the ending `jump_past_alt' and
5216 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
5220 /* Move to right after this alternative, including the
5224 /* Break if it's the beginning of an n-th alternative
5225 that doesn't begin with an on_failure_jump. */
5226 if ((re_opcode_t) *p1 != on_failure_jump)
5229 /* Still have to check that it's not an n-th
5230 alternative that starts with an on_failure_jump. */
5232 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5233 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
5235 /* Get to the beginning of the n-th alternative. */
5241 /* Deal with the last alternative: go back and get number
5242 of the `jump_past_alt' just before it. `mcnt' contains
5243 the length of the alternative. */
5244 EXTRACT_NUMBER (mcnt, p1 - 2);
5246 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
5249 p1 += mcnt; /* Get past the n-th alternative. */
5255 assert (p1[1] == **p);
5261 if (!common_op_match_null_string_p (&p1, end, reg_info))
5264 } /* while p1 < end */
5267 } /* group_match_null_string_p */
5270 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5271 It expects P to be the first byte of a single alternative and END one
5272 byte past the last. The alternative can contain groups. */
5275 alt_match_null_string_p (p, end, reg_info)
5276 unsigned char *p, *end;
5277 register_info_type *reg_info;
5280 unsigned char *p1 = p;
5284 /* Skip over opcodes that can match nothing, and break when we get
5285 to one that can't. */
5287 switch ((re_opcode_t) *p1)
5290 case on_failure_jump:
5292 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5297 if (!common_op_match_null_string_p (&p1, end, reg_info))
5300 } /* while p1 < end */
5303 } /* alt_match_null_string_p */
5306 /* Deals with the ops common to group_match_null_string_p and
5307 alt_match_null_string_p.
5309 Sets P to one after the op and its arguments, if any. */
5312 common_op_match_null_string_p (p, end, reg_info)
5313 unsigned char **p, *end;
5314 register_info_type *reg_info;
5319 unsigned char *p1 = *p;
5321 switch ((re_opcode_t) *p1++)
5341 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5342 ret = group_match_null_string_p (&p1, end, reg_info);
5344 /* Have to set this here in case we're checking a group which
5345 contains a group and a back reference to it. */
5347 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5348 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5354 /* If this is an optimized succeed_n for zero times, make the jump. */
5356 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5364 /* Get to the number of times to succeed. */
5366 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5371 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5379 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5387 /* All other opcodes mean we cannot match the empty string. */
5393 } /* common_op_match_null_string_p */
5396 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5397 bytes; nonzero otherwise. */
5400 bcmp_translate (s1, s2, len, translate)
5401 const char *s1, *s2;
5403 RE_TRANSLATE_TYPE translate;
5405 register const unsigned char *p1 = (const unsigned char *) s1;
5406 register const unsigned char *p2 = (const unsigned char *) s2;
5409 if (translate[*p1++] != translate[*p2++]) return 1;
5415 /* Entry points for GNU code. */
5417 /* re_compile_pattern is the GNU regular expression compiler: it
5418 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5419 Returns 0 if the pattern was valid, otherwise an error string.
5421 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5422 are set in BUFP on entry.
5424 We call regex_compile to do the actual compilation. */
5427 re_compile_pattern (pattern, length, bufp)
5428 const char *pattern;
5430 struct re_pattern_buffer *bufp;
5434 /* GNU code is written to assume at least RE_NREGS registers will be set
5435 (and at least one extra will be -1). */
5436 bufp->regs_allocated = REGS_UNALLOCATED;
5438 /* And GNU code determines whether or not to get register information
5439 by passing null for the REGS argument to re_match, etc., not by
5443 /* Match anchors at newline. */
5444 bufp->newline_anchor = 1;
5446 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5450 return gettext (re_error_msgid[(int) ret]);
5453 /* Entry points compatible with 4.2 BSD regex library. We don't define
5454 them unless specifically requested. */
5456 #if defined (_REGEX_RE_COMP) || defined (_LIBC)
5458 /* BSD has one and only one pattern buffer. */
5459 static struct re_pattern_buffer re_comp_buf;
5463 /* Make these definitions weak in libc, so POSIX programs can redefine
5464 these names if they don't use our functions, and still use
5465 regcomp/regexec below without link errors. */
5475 if (!re_comp_buf.buffer)
5476 return gettext ("No previous regular expression");
5480 if (!re_comp_buf.buffer)
5482 re_comp_buf.buffer = (unsigned char *) malloc (200); /* __MEM_CHECKED__ */
5483 if (re_comp_buf.buffer == NULL)
5484 return gettext (re_error_msgid[(int) REG_ESPACE]);
5485 re_comp_buf.allocated = 200;
5487 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH); /* __MEM_CHECKED__ */
5488 if (re_comp_buf.fastmap == NULL)
5489 return gettext (re_error_msgid[(int) REG_ESPACE]);
5492 /* Since `re_exec' always passes NULL for the `regs' argument, we
5493 don't need to initialize the pattern buffer fields which affect it. */
5495 /* Match anchors at newlines. */
5496 re_comp_buf.newline_anchor = 1;
5498 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5503 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5504 return (char *) gettext (re_error_msgid[(int) ret]);
5515 const int len = strlen (s);
5517 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5520 #endif /* _REGEX_RE_COMP */
5522 /* POSIX.2 functions. Don't define these for Emacs. */
5526 /* regcomp takes a regular expression as a string and compiles it.
5528 PREG is a regex_t *. We do not expect any fields to be initialized,
5529 since POSIX says we shouldn't. Thus, we set
5531 `buffer' to the compiled pattern;
5532 `used' to the length of the compiled pattern;
5533 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5534 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5535 RE_SYNTAX_POSIX_BASIC;
5536 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5537 `fastmap' and `fastmap_accurate' to zero;
5538 `re_nsub' to the number of subexpressions in PATTERN.
5540 PATTERN is the address of the pattern string.
5542 CFLAGS is a series of bits which affect compilation.
5544 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5545 use POSIX basic syntax.
5547 If REG_NEWLINE is set, then . and [^...] don't match newline.
5548 Also, regexec will try a match beginning after every newline.
5550 If REG_ICASE is set, then we considers upper- and lowercase
5551 versions of letters to be equivalent when matching.
5553 If REG_NOSUB is set, then when PREG is passed to regexec, that
5554 routine will report only success or failure, and nothing about the
5557 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
5558 the return codes and their meanings.) */
5561 regcomp (preg, pattern, cflags)
5563 const char *pattern;
5568 = (cflags & REG_EXTENDED) ?
5569 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5571 /* regex_compile will allocate the space for the compiled pattern. */
5573 preg->allocated = 0;
5576 /* Don't bother to use a fastmap when searching. This simplifies the
5577 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
5578 characters after newlines into the fastmap. This way, we just try
5582 if (cflags & REG_ICASE)
5587 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE /* __MEM_CHECKED__ */
5588 * sizeof (*(RE_TRANSLATE_TYPE)0));
5589 if (preg->translate == NULL)
5590 return (int) REG_ESPACE;
5592 /* Map uppercase characters to corresponding lowercase ones. */
5593 for (i = 0; i < CHAR_SET_SIZE; i++)
5594 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
5597 preg->translate = NULL;
5599 /* If REG_NEWLINE is set, newlines are treated differently. */
5600 if (cflags & REG_NEWLINE)
5601 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5602 syntax &= ~RE_DOT_NEWLINE;
5603 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5604 /* It also changes the matching behavior. */
5605 preg->newline_anchor = 1;
5608 preg->newline_anchor = 0;
5610 preg->no_sub = !!(cflags & REG_NOSUB);
5612 /* POSIX says a null character in the pattern terminates it, so we
5613 can use strlen here in compiling the pattern. */
5614 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5616 /* POSIX doesn't distinguish between an unmatched open-group and an
5617 unmatched close-group: both are REG_EPAREN. */
5618 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5624 /* regexec searches for a given pattern, specified by PREG, in the
5627 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5628 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5629 least NMATCH elements, and we set them to the offsets of the
5630 corresponding matched substrings.
5632 EFLAGS specifies `execution flags' which affect matching: if
5633 REG_NOTBOL is set, then ^ does not match at the beginning of the
5634 string; if REG_NOTEOL is set, then $ does not match at the end.
5636 We return 0 if we find a match and REG_NOMATCH if not. */
5639 regexec (preg, string, nmatch, pmatch, eflags)
5640 const regex_t *preg;
5643 regmatch_t pmatch[];
5647 struct re_registers regs;
5648 regex_t private_preg;
5649 int len = strlen (string);
5650 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5652 private_preg = *preg;
5654 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5655 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5657 /* The user has told us exactly how many registers to return
5658 information about, via `nmatch'. We have to pass that on to the
5659 matching routines. */
5660 private_preg.regs_allocated = REGS_FIXED;
5664 regs.num_regs = nmatch;
5665 regs.start = TALLOC (nmatch, regoff_t);
5666 regs.end = TALLOC (nmatch, regoff_t);
5667 if (regs.start == NULL || regs.end == NULL)
5668 return (int) REG_NOMATCH;
5671 /* Perform the searching operation. */
5672 ret = re_search (&private_preg, string, len,
5673 /* start: */ 0, /* range: */ len,
5674 want_reg_info ? ®s : (struct re_registers *) 0);
5676 /* Copy the register information to the POSIX structure. */
5683 for (r = 0; r < nmatch; r++)
5685 pmatch[r].rm_so = regs.start[r];
5686 pmatch[r].rm_eo = regs.end[r];
5690 /* If we needed the temporary register info, free the space now. */
5691 free (regs.start); /* __MEM_CHECKED__ */
5692 free (regs.end); /* __MEM_CHECKED__ */
5695 /* We want zero return to mean success, unlike `re_search'. */
5696 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5700 /* Returns a message corresponding to an error code, ERRCODE, returned
5701 from either regcomp or regexec. We don't use PREG here. */
5704 regerror (errcode, preg, errbuf, errbuf_size)
5706 const regex_t *preg;
5714 || errcode >= (int) (sizeof (re_error_msgid)
5715 / sizeof (re_error_msgid[0])))
5716 /* Only error codes returned by the rest of the code should be passed
5717 to this routine. If we are given anything else, or if other regex
5718 code generates an invalid error code, then the program has a bug.
5719 Dump core so we can fix it. */
5722 msg = gettext (re_error_msgid[errcode]);
5724 msg_size = strlen (msg) + 1; /* Includes the null. */
5726 if (errbuf_size != 0)
5728 if (msg_size > errbuf_size)
5730 strncpy (errbuf, msg, errbuf_size - 1);
5731 errbuf[errbuf_size - 1] = 0;
5734 strcpy (errbuf, msg); /* __STRCPY_CHECKED__ */
5741 /* Free dynamically allocated space used by PREG. */
5747 if (preg->buffer != NULL)
5748 free (preg->buffer); /* __MEM_CHECKED__ */
5749 preg->buffer = NULL;
5751 preg->allocated = 0;
5754 if (preg->fastmap != NULL)
5755 free (preg->fastmap); /* __MEM_CHECKED__ */
5756 preg->fastmap = NULL;
5757 preg->fastmap_accurate = 0;
5759 if (preg->translate != NULL)
5760 free (preg->translate); /* __MEM_CHECKED__ */
5761 preg->translate = NULL;
5764 #endif /* not emacs */
projects / software / mutt-debian.git / blob