1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
5 Copyright (C) 1993, 94, 95, 96, 97, 98 Free Software Foundation, Inc.
7 NOTE: The canonical source of this file is maintained with the
8 GNU C Library. Bugs can be reported to bug-glibc@prep.ai.mit.edu.
10 This program is free software; you can redistribute it and/or modify it
11 under the terms of the GNU General Public License as published by the
12 Free Software Foundation; either version 2, or (at your option) any
15 This program is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 GNU General Public License for more details.
20 You should have received a copy of the GNU General Public License
21 along with this program; if not, write to the Free Software Foundation,
22 Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
24 /* AIX requires this to be the first thing in the file. */
25 #if defined _AIX && !defined REGEX_MALLOC
37 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
38 # define PARAMS(args) args
40 # define PARAMS(args) ()
42 #endif /* Not PARAMS. */
44 #if defined STDC_HEADERS && !defined emacs
47 /* We need this for `gnu-regex.h', and perhaps for the Emacs include files. */
48 # include <sys/types.h>
51 /* For platform which support the ISO C amendement 1 functionality we
52 support user defined character classes. */
53 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
54 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
59 /* This is for other GNU distributions with internationalized messages. */
60 #if HAVE_LIBINTL_H || defined _LIBC
63 # define gettext(msgid) (msgid)
67 /* This define is so xgettext can find the internationalizable
69 # define gettext_noop(String) String
72 /* The `emacs' switch turns on certain matching commands
73 that make sense only in Emacs. */
82 /* If we are not linking with Emacs proper,
83 we can't use the relocating allocator
84 even if config.h says that we can. */
87 # if defined STDC_HEADERS || defined _LIBC
94 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
95 If nothing else has been done, use the method below. */
96 # ifdef INHIBIT_STRING_HEADER
97 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
98 # if !defined bzero && !defined bcopy
99 # undef INHIBIT_STRING_HEADER
104 /* This is the normal way of making sure we have a bcopy and a bzero.
105 This is used in most programs--a few other programs avoid this
106 by defining INHIBIT_STRING_HEADER. */
107 # ifndef INHIBIT_STRING_HEADER
108 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
112 # define bzero(s, n) (memset (s, '\0', n), (s))
114 # define bzero(s, n) __bzero (s, n)
118 # include <strings.h>
120 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
123 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
128 /* Define the syntax stuff for \<, \>, etc. */
130 /* This must be nonzero for the wordchar and notwordchar pattern
131 commands in re_match_2. */
136 # ifdef SWITCH_ENUM_BUG
137 # define SWITCH_ENUM_CAST(x) ((int)(x))
139 # define SWITCH_ENUM_CAST(x) (x)
142 /* How many characters in the character set. */
143 # define CHAR_SET_SIZE 256
145 /* CYGNUS LOCAL: define _REGEX_RE_COMP to get BSD style re_comp and re_exec */
146 #define _REGEX_RE_COMP
150 extern char *re_syntax_table;
152 # else /* not SYNTAX_TABLE */
154 static char re_syntax_table[CHAR_SET_SIZE];
165 bzero (re_syntax_table, sizeof re_syntax_table);
167 for (c = 'a'; c <= 'z'; c++)
168 re_syntax_table[c] = Sword;
170 for (c = 'A'; c <= 'Z'; c++)
171 re_syntax_table[c] = Sword;
173 for (c = '0'; c <= '9'; c++)
174 re_syntax_table[c] = Sword;
176 re_syntax_table['_'] = Sword;
181 # endif /* not SYNTAX_TABLE */
183 # define SYNTAX(c) re_syntax_table[c]
185 #endif /* not emacs */
187 /* Get the interface, including the syntax bits. */
188 /* CYGNUS LOCAL: call it gnu-regex.h, not regex.h, to avoid name conflicts */
189 #include "gnu-regex.h"
191 /* isalpha etc. are used for the character classes. */
194 /* Jim Meyering writes:
196 "... Some ctype macros are valid only for character codes that
197 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
198 using /bin/cc or gcc but without giving an ansi option). So, all
199 ctype uses should be through macros like ISPRINT... If
200 STDC_HEADERS is defined, then autoconf has verified that the ctype
201 macros don't need to be guarded with references to isascii. ...
202 Defining isascii to 1 should let any compiler worth its salt
203 eliminate the && through constant folding."
204 Solaris defines some of these symbols so we must undefine them first. */
207 #if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
208 # define ISASCII(c) 1
210 # define ISASCII(c) isascii(c)
214 # define ISBLANK(c) (ISASCII (c) && isblank (c))
216 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
219 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
221 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
225 #define ISPRINT(c) (ISASCII (c) && isprint (c))
226 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
227 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
228 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
229 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
230 #define ISLOWER(c) (ISASCII (c) && islower (c))
231 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
232 #define ISSPACE(c) (ISASCII (c) && isspace (c))
233 #define ISUPPER(c) (ISASCII (c) && isupper (c))
234 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
237 # define NULL (void *)0
240 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
241 since ours (we hope) works properly with all combinations of
242 machines, compilers, `char' and `unsigned char' argument types.
243 (Per Bothner suggested the basic approach.) */
244 #undef SIGN_EXTEND_CHAR
246 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
247 #else /* not __STDC__ */
248 /* As in Harbison and Steele. */
249 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
252 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
253 use `alloca' instead of `malloc'. This is because using malloc in
254 re_search* or re_match* could cause memory leaks when C-g is used in
255 Emacs; also, malloc is slower and causes storage fragmentation. On
256 the other hand, malloc is more portable, and easier to debug.
258 Because we sometimes use alloca, some routines have to be macros,
259 not functions -- `alloca'-allocated space disappears at the end of the
260 function it is called in. */
264 # define REGEX_ALLOCATE malloc
265 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
266 # define REGEX_FREE free
268 #else /* not REGEX_MALLOC */
270 /* Emacs already defines alloca, sometimes. */
273 /* Make alloca work the best possible way. */
275 # define alloca __builtin_alloca
276 # else /* not __GNUC__ */
279 # endif /* HAVE_ALLOCA_H */
280 # endif /* not __GNUC__ */
282 # endif /* not alloca */
284 # define REGEX_ALLOCATE alloca
286 /* Assumes a `char *destination' variable. */
287 # define REGEX_REALLOCATE(source, osize, nsize) \
288 (destination = (char *) alloca (nsize), \
289 memcpy (destination, source, osize))
291 /* No need to do anything to free, after alloca. */
292 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
294 #endif /* not REGEX_MALLOC */
296 /* Define how to allocate the failure stack. */
298 #if defined REL_ALLOC && defined REGEX_MALLOC
300 # define REGEX_ALLOCATE_STACK(size) \
301 r_alloc (&failure_stack_ptr, (size))
302 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
303 r_re_alloc (&failure_stack_ptr, (nsize))
304 # define REGEX_FREE_STACK(ptr) \
305 r_alloc_free (&failure_stack_ptr)
307 #else /* not using relocating allocator */
311 # define REGEX_ALLOCATE_STACK malloc
312 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
313 # define REGEX_FREE_STACK free
315 # else /* not REGEX_MALLOC */
317 # define REGEX_ALLOCATE_STACK alloca
319 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
320 REGEX_REALLOCATE (source, osize, nsize)
321 /* No need to explicitly free anything. */
322 # define REGEX_FREE_STACK(arg)
324 # endif /* not REGEX_MALLOC */
325 #endif /* not using relocating allocator */
328 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
329 `string1' or just past its end. This works if PTR is NULL, which is
331 #define FIRST_STRING_P(ptr) \
332 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
334 /* (Re)Allocate N items of type T using malloc, or fail. */
335 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
336 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
337 #define RETALLOC_IF(addr, n, t) \
338 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
339 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
341 #define BYTEWIDTH 8 /* In bits. */
343 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
347 #define MAX(a, b) ((a) > (b) ? (a) : (b))
348 #define MIN(a, b) ((a) < (b) ? (a) : (b))
350 typedef char boolean;
354 static int re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
355 const char *string1, int size1,
356 const char *string2, int size2,
358 struct re_registers *regs,
361 /* These are the command codes that appear in compiled regular
362 expressions. Some opcodes are followed by argument bytes. A
363 command code can specify any interpretation whatsoever for its
364 arguments. Zero bytes may appear in the compiled regular expression. */
370 /* Succeed right away--no more backtracking. */
373 /* Followed by one byte giving n, then by n literal bytes. */
376 /* Matches any (more or less) character. */
379 /* Matches any one char belonging to specified set. First
380 following byte is number of bitmap bytes. Then come bytes
381 for a bitmap saying which chars are in. Bits in each byte
382 are ordered low-bit-first. A character is in the set if its
383 bit is 1. A character too large to have a bit in the map is
384 automatically not in the set. */
387 /* Same parameters as charset, but match any character that is
388 not one of those specified. */
391 /* Start remembering the text that is matched, for storing in a
392 register. Followed by one byte with the register number, in
393 the range 0 to one less than the pattern buffer's re_nsub
394 field. Then followed by one byte with the number of groups
395 inner to this one. (This last has to be part of the
396 start_memory only because we need it in the on_failure_jump
400 /* Stop remembering the text that is matched and store it in a
401 memory register. Followed by one byte with the register
402 number, in the range 0 to one less than `re_nsub' in the
403 pattern buffer, and one byte with the number of inner groups,
404 just like `start_memory'. (We need the number of inner
405 groups here because we don't have any easy way of finding the
406 corresponding start_memory when we're at a stop_memory.) */
409 /* Match a duplicate of something remembered. Followed by one
410 byte containing the register number. */
413 /* Fail unless at beginning of line. */
416 /* Fail unless at end of line. */
419 /* Succeeds if at beginning of buffer (if emacs) or at beginning
420 of string to be matched (if not). */
423 /* Analogously, for end of buffer/string. */
426 /* Followed by two byte relative address to which to jump. */
429 /* Same as jump, but marks the end of an alternative. */
432 /* Followed by two-byte relative address of place to resume at
433 in case of failure. */
436 /* Like on_failure_jump, but pushes a placeholder instead of the
437 current string position when executed. */
438 on_failure_keep_string_jump,
440 /* Throw away latest failure point and then jump to following
441 two-byte relative address. */
444 /* Change to pop_failure_jump if know won't have to backtrack to
445 match; otherwise change to jump. This is used to jump
446 back to the beginning of a repeat. If what follows this jump
447 clearly won't match what the repeat does, such that we can be
448 sure that there is no use backtracking out of repetitions
449 already matched, then we change it to a pop_failure_jump.
450 Followed by two-byte address. */
453 /* Jump to following two-byte address, and push a dummy failure
454 point. This failure point will be thrown away if an attempt
455 is made to use it for a failure. A `+' construct makes this
456 before the first repeat. Also used as an intermediary kind
457 of jump when compiling an alternative. */
460 /* Push a dummy failure point and continue. Used at the end of
464 /* Followed by two-byte relative address and two-byte number n.
465 After matching N times, jump to the address upon failure. */
468 /* Followed by two-byte relative address, and two-byte number n.
469 Jump to the address N times, then fail. */
472 /* Set the following two-byte relative address to the
473 subsequent two-byte number. The address *includes* the two
477 wordchar, /* Matches any word-constituent character. */
478 notwordchar, /* Matches any char that is not a word-constituent. */
480 wordbeg, /* Succeeds if at word beginning. */
481 wordend, /* Succeeds if at word end. */
483 wordbound, /* Succeeds if at a word boundary. */
484 notwordbound /* Succeeds if not at a word boundary. */
487 ,before_dot, /* Succeeds if before point. */
488 at_dot, /* Succeeds if at point. */
489 after_dot, /* Succeeds if after point. */
491 /* Matches any character whose syntax is specified. Followed by
492 a byte which contains a syntax code, e.g., Sword. */
495 /* Matches any character whose syntax is not that specified. */
500 /* Common operations on the compiled pattern. */
502 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
504 #define STORE_NUMBER(destination, number) \
506 (destination)[0] = (number) & 0377; \
507 (destination)[1] = (number) >> 8; \
510 /* Same as STORE_NUMBER, except increment DESTINATION to
511 the byte after where the number is stored. Therefore, DESTINATION
512 must be an lvalue. */
514 #define STORE_NUMBER_AND_INCR(destination, number) \
516 STORE_NUMBER (destination, number); \
517 (destination) += 2; \
520 /* Put into DESTINATION a number stored in two contiguous bytes starting
523 #define EXTRACT_NUMBER(destination, source) \
525 (destination) = *(source) & 0377; \
526 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
530 static void extract_number _RE_ARGS ((int *dest, unsigned char *source));
532 extract_number (dest, source)
534 unsigned char *source;
536 int temp = SIGN_EXTEND_CHAR (*(source + 1));
537 *dest = *source & 0377;
541 # ifndef EXTRACT_MACROS /* To debug the macros. */
542 # undef EXTRACT_NUMBER
543 # define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
544 # endif /* not EXTRACT_MACROS */
548 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
549 SOURCE must be an lvalue. */
551 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
553 EXTRACT_NUMBER (destination, source); \
558 static void extract_number_and_incr _RE_ARGS ((int *destination,
559 unsigned char **source));
561 extract_number_and_incr (destination, source)
563 unsigned char **source;
565 extract_number (destination, *source);
569 # ifndef EXTRACT_MACROS
570 # undef EXTRACT_NUMBER_AND_INCR
571 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
572 extract_number_and_incr (&dest, &src)
573 # endif /* not EXTRACT_MACROS */
577 /* If DEBUG is defined, Regex prints many voluminous messages about what
578 it is doing (if the variable `debug' is nonzero). If linked with the
579 main program in `iregex.c', you can enter patterns and strings
580 interactively. And if linked with the main program in `main.c' and
581 the other test files, you can run the already-written tests. */
585 /* We use standard I/O for debugging. */
588 /* It is useful to test things that ``must'' be true when debugging. */
591 static int debug = 0;
593 # define DEBUG_STATEMENT(e) e
594 # define DEBUG_PRINT1(x) if (debug) printf (x)
595 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
596 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
597 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
598 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
599 if (debug) print_partial_compiled_pattern (s, e)
600 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
601 if (debug) print_double_string (w, s1, sz1, s2, sz2)
604 /* Print the fastmap in human-readable form. */
607 print_fastmap (fastmap)
610 unsigned was_a_range = 0;
613 while (i < (1 << BYTEWIDTH))
619 while (i < (1 << BYTEWIDTH) && fastmap[i])
635 /* Print a compiled pattern string in human-readable form, starting at
636 the START pointer into it and ending just before the pointer END. */
639 print_partial_compiled_pattern (start, end)
640 unsigned char *start;
645 unsigned char *p = start;
646 unsigned char *pend = end;
654 /* Loop over pattern commands. */
657 printf ("%d:\t", p - start);
659 switch ((re_opcode_t) *p++)
667 printf ("/exactn/%d", mcnt);
678 printf ("/start_memory/%d/%d", mcnt, *p++);
683 printf ("/stop_memory/%d/%d", mcnt, *p++);
687 printf ("/duplicate/%d", *p++);
697 register int c, last = -100;
698 register int in_range = 0;
700 printf ("/charset [%s",
701 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
703 assert (p + *p < pend);
705 for (c = 0; c < 256; c++)
707 && (p[1 + (c/8)] & (1 << (c % 8))))
709 /* Are we starting a range? */
710 if (last + 1 == c && ! in_range)
715 /* Have we broken a range? */
716 else if (last + 1 != c && in_range)
745 case on_failure_jump:
746 extract_number_and_incr (&mcnt, &p);
747 printf ("/on_failure_jump to %d", p + mcnt - start);
750 case on_failure_keep_string_jump:
751 extract_number_and_incr (&mcnt, &p);
752 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
755 case dummy_failure_jump:
756 extract_number_and_incr (&mcnt, &p);
757 printf ("/dummy_failure_jump to %d", p + mcnt - start);
760 case push_dummy_failure:
761 printf ("/push_dummy_failure");
765 extract_number_and_incr (&mcnt, &p);
766 printf ("/maybe_pop_jump to %d", p + mcnt - start);
769 case pop_failure_jump:
770 extract_number_and_incr (&mcnt, &p);
771 printf ("/pop_failure_jump to %d", p + mcnt - start);
775 extract_number_and_incr (&mcnt, &p);
776 printf ("/jump_past_alt to %d", p + mcnt - start);
780 extract_number_and_incr (&mcnt, &p);
781 printf ("/jump to %d", p + mcnt - start);
785 extract_number_and_incr (&mcnt, &p);
787 extract_number_and_incr (&mcnt2, &p);
788 printf ("/succeed_n to %d, %d times", p1 - start, mcnt2);
792 extract_number_and_incr (&mcnt, &p);
794 extract_number_and_incr (&mcnt2, &p);
795 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
799 extract_number_and_incr (&mcnt, &p);
801 extract_number_and_incr (&mcnt2, &p);
802 printf ("/set_number_at location %d to %d", p1 - start, mcnt2);
806 printf ("/wordbound");
810 printf ("/notwordbound");
822 printf ("/before_dot");
830 printf ("/after_dot");
834 printf ("/syntaxspec");
836 printf ("/%d", mcnt);
840 printf ("/notsyntaxspec");
842 printf ("/%d", mcnt);
847 printf ("/wordchar");
851 printf ("/notwordchar");
863 printf ("?%d", *(p-1));
869 printf ("%d:\tend of pattern.\n", p - start);
874 print_compiled_pattern (bufp)
875 struct re_pattern_buffer *bufp;
877 unsigned char *buffer = bufp->buffer;
879 print_partial_compiled_pattern (buffer, buffer + bufp->used);
880 printf ("%ld bytes used/%ld bytes allocated.\n",
881 bufp->used, bufp->allocated);
883 if (bufp->fastmap_accurate && bufp->fastmap)
885 printf ("fastmap: ");
886 print_fastmap (bufp->fastmap);
889 printf ("re_nsub: %d\t", bufp->re_nsub);
890 printf ("regs_alloc: %d\t", bufp->regs_allocated);
891 printf ("can_be_null: %d\t", bufp->can_be_null);
892 printf ("newline_anchor: %d\n", bufp->newline_anchor);
893 printf ("no_sub: %d\t", bufp->no_sub);
894 printf ("not_bol: %d\t", bufp->not_bol);
895 printf ("not_eol: %d\t", bufp->not_eol);
896 printf ("syntax: %lx\n", bufp->syntax);
897 /* Perhaps we should print the translate table? */
902 print_double_string (where, string1, size1, string2, size2)
915 if (FIRST_STRING_P (where))
917 for (this_char = where - string1; this_char < size1; this_char++)
918 putchar (string1[this_char]);
923 for (this_char = where - string2; this_char < size2; this_char++)
924 putchar (string2[this_char]);
935 #else /* not DEBUG */
940 # define DEBUG_STATEMENT(e)
941 # define DEBUG_PRINT1(x)
942 # define DEBUG_PRINT2(x1, x2)
943 # define DEBUG_PRINT3(x1, x2, x3)
944 # define DEBUG_PRINT4(x1, x2, x3, x4)
945 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
946 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
948 #endif /* not DEBUG */
950 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
951 also be assigned to arbitrarily: each pattern buffer stores its own
952 syntax, so it can be changed between regex compilations. */
953 /* This has no initializer because initialized variables in Emacs
954 become read-only after dumping. */
955 reg_syntax_t re_syntax_options;
958 /* Specify the precise syntax of regexps for compilation. This provides
959 for compatibility for various utilities which historically have
960 different, incompatible syntaxes.
962 The argument SYNTAX is a bit mask comprised of the various bits
963 defined in gnu-regex.h. We return the old syntax. */
966 re_set_syntax (syntax)
969 reg_syntax_t ret = re_syntax_options;
971 re_syntax_options = syntax;
973 if (syntax & RE_DEBUG)
975 else if (debug) /* was on but now is not */
981 weak_alias (__re_set_syntax, re_set_syntax)
984 /* This table gives an error message for each of the error codes listed
985 in gnu-regex.h. Obviously the order here has to be same as there.
986 POSIX doesn't require that we do anything for REG_NOERROR,
987 but why not be nice? */
989 static const char *re_error_msgid[] =
991 gettext_noop ("Success"), /* REG_NOERROR */
992 gettext_noop ("No match"), /* REG_NOMATCH */
993 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
994 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
995 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
996 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
997 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
998 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
999 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
1000 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
1001 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
1002 gettext_noop ("Invalid range end"), /* REG_ERANGE */
1003 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
1004 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
1005 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
1006 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
1007 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
1010 /* Avoiding alloca during matching, to placate r_alloc. */
1012 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1013 searching and matching functions should not call alloca. On some
1014 systems, alloca is implemented in terms of malloc, and if we're
1015 using the relocating allocator routines, then malloc could cause a
1016 relocation, which might (if the strings being searched are in the
1017 ralloc heap) shift the data out from underneath the regexp
1020 Here's another reason to avoid allocation: Emacs
1021 processes input from X in a signal handler; processing X input may
1022 call malloc; if input arrives while a matching routine is calling
1023 malloc, then we're scrod. But Emacs can't just block input while
1024 calling matching routines; then we don't notice interrupts when
1025 they come in. So, Emacs blocks input around all regexp calls
1026 except the matching calls, which it leaves unprotected, in the
1027 faith that they will not malloc. */
1029 /* Normally, this is fine. */
1030 #define MATCH_MAY_ALLOCATE
1032 /* When using GNU C, we are not REALLY using the C alloca, no matter
1033 what config.h may say. So don't take precautions for it. */
1038 /* The match routines may not allocate if (1) they would do it with malloc
1039 and (2) it's not safe for them to use malloc.
1040 Note that if REL_ALLOC is defined, matching would not use malloc for the
1041 failure stack, but we would still use it for the register vectors;
1042 so REL_ALLOC should not affect this. */
1043 #if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1044 # undef MATCH_MAY_ALLOCATE
1048 /* Failure stack declarations and macros; both re_compile_fastmap and
1049 re_match_2 use a failure stack. These have to be macros because of
1050 REGEX_ALLOCATE_STACK. */
1053 /* Number of failure points for which to initially allocate space
1054 when matching. If this number is exceeded, we allocate more
1055 space, so it is not a hard limit. */
1056 #ifndef INIT_FAILURE_ALLOC
1057 # define INIT_FAILURE_ALLOC 5
1060 /* Roughly the maximum number of failure points on the stack. Would be
1061 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1062 This is a variable only so users of regex can assign to it; we never
1063 change it ourselves. */
1067 # if defined MATCH_MAY_ALLOCATE
1068 /* 4400 was enough to cause a crash on Alpha OSF/1,
1069 whose default stack limit is 2mb. */
1070 long int re_max_failures = 4000;
1072 long int re_max_failures = 2000;
1075 union fail_stack_elt
1077 unsigned char *pointer;
1081 typedef union fail_stack_elt fail_stack_elt_t;
1085 fail_stack_elt_t *stack;
1086 unsigned long int size;
1087 unsigned long int avail; /* Offset of next open position. */
1090 #else /* not INT_IS_16BIT */
1092 # if defined MATCH_MAY_ALLOCATE
1093 /* 4400 was enough to cause a crash on Alpha OSF/1,
1094 whose default stack limit is 2mb. */
1095 int re_max_failures = 20000;
1097 int re_max_failures = 2000;
1100 union fail_stack_elt
1102 unsigned char *pointer;
1106 typedef union fail_stack_elt fail_stack_elt_t;
1110 fail_stack_elt_t *stack;
1112 unsigned avail; /* Offset of next open position. */
1115 #endif /* INT_IS_16BIT */
1117 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1118 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1119 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1122 /* Define macros to initialize and free the failure stack.
1123 Do `return -2' if the alloc fails. */
1125 #ifdef MATCH_MAY_ALLOCATE
1126 # define INIT_FAIL_STACK() \
1128 fail_stack.stack = (fail_stack_elt_t *) \
1129 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1131 if (fail_stack.stack == NULL) \
1134 fail_stack.size = INIT_FAILURE_ALLOC; \
1135 fail_stack.avail = 0; \
1138 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1140 # define INIT_FAIL_STACK() \
1142 fail_stack.avail = 0; \
1145 # define RESET_FAIL_STACK()
1149 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1151 Return 1 if succeeds, and 0 if either ran out of memory
1152 allocating space for it or it was already too large.
1154 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1156 #define DOUBLE_FAIL_STACK(fail_stack) \
1157 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1159 : ((fail_stack).stack = (fail_stack_elt_t *) \
1160 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1161 (fail_stack).size * sizeof (fail_stack_elt_t), \
1162 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1164 (fail_stack).stack == NULL \
1166 : ((fail_stack).size <<= 1, \
1170 /* Push pointer POINTER on FAIL_STACK.
1171 Return 1 if was able to do so and 0 if ran out of memory allocating
1173 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1174 ((FAIL_STACK_FULL () \
1175 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1177 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1180 /* Push a pointer value onto the failure stack.
1181 Assumes the variable `fail_stack'. Probably should only
1182 be called from within `PUSH_FAILURE_POINT'. */
1183 #define PUSH_FAILURE_POINTER(item) \
1184 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1186 /* This pushes an integer-valued item onto the failure stack.
1187 Assumes the variable `fail_stack'. Probably should only
1188 be called from within `PUSH_FAILURE_POINT'. */
1189 #define PUSH_FAILURE_INT(item) \
1190 fail_stack.stack[fail_stack.avail++].integer = (item)
1192 /* Push a fail_stack_elt_t value onto the failure stack.
1193 Assumes the variable `fail_stack'. Probably should only
1194 be called from within `PUSH_FAILURE_POINT'. */
1195 #define PUSH_FAILURE_ELT(item) \
1196 fail_stack.stack[fail_stack.avail++] = (item)
1198 /* These three POP... operations complement the three PUSH... operations.
1199 All assume that `fail_stack' is nonempty. */
1200 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1201 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1202 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1204 /* Used to omit pushing failure point id's when we're not debugging. */
1206 # define DEBUG_PUSH PUSH_FAILURE_INT
1207 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1209 # define DEBUG_PUSH(item)
1210 # define DEBUG_POP(item_addr)
1214 /* Push the information about the state we will need
1215 if we ever fail back to it.
1217 Requires variables fail_stack, regstart, regend, reg_info, and
1218 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1221 Does `return FAILURE_CODE' if runs out of memory. */
1223 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1225 char *destination; \
1226 /* Must be int, so when we don't save any registers, the arithmetic \
1227 of 0 + -1 isn't done as unsigned. */ \
1228 /* Can't be int, since there is not a shred of a guarantee that int \
1229 is wide enough to hold a value of something to which pointer can \
1231 active_reg_t this_reg; \
1233 DEBUG_STATEMENT (failure_id++); \
1234 DEBUG_STATEMENT (nfailure_points_pushed++); \
1235 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1236 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1237 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1239 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1240 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1242 /* Ensure we have enough space allocated for what we will push. */ \
1243 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1245 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1246 return failure_code; \
1248 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1249 (fail_stack).size); \
1250 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1253 /* Push the info, starting with the registers. */ \
1254 DEBUG_PRINT1 ("\n"); \
1257 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1260 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1261 DEBUG_STATEMENT (num_regs_pushed++); \
1263 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1264 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1266 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1267 PUSH_FAILURE_POINTER (regend[this_reg]); \
1269 DEBUG_PRINT2 (" info: %p\n ", \
1270 reg_info[this_reg].word.pointer); \
1271 DEBUG_PRINT2 (" match_null=%d", \
1272 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1273 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1274 DEBUG_PRINT2 (" matched_something=%d", \
1275 MATCHED_SOMETHING (reg_info[this_reg])); \
1276 DEBUG_PRINT2 (" ever_matched=%d", \
1277 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1278 DEBUG_PRINT1 ("\n"); \
1279 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1282 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1283 PUSH_FAILURE_INT (lowest_active_reg); \
1285 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1286 PUSH_FAILURE_INT (highest_active_reg); \
1288 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1289 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1290 PUSH_FAILURE_POINTER (pattern_place); \
1292 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1293 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1295 DEBUG_PRINT1 ("'\n"); \
1296 PUSH_FAILURE_POINTER (string_place); \
1298 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1299 DEBUG_PUSH (failure_id); \
1302 /* This is the number of items that are pushed and popped on the stack
1303 for each register. */
1304 #define NUM_REG_ITEMS 3
1306 /* Individual items aside from the registers. */
1308 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1310 # define NUM_NONREG_ITEMS 4
1313 /* We push at most this many items on the stack. */
1314 /* We used to use (num_regs - 1), which is the number of registers
1315 this regexp will save; but that was changed to 5
1316 to avoid stack overflow for a regexp with lots of parens. */
1317 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1319 /* We actually push this many items. */
1320 #define NUM_FAILURE_ITEMS \
1322 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1326 /* How many items can still be added to the stack without overflowing it. */
1327 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1330 /* Pops what PUSH_FAIL_STACK pushes.
1332 We restore into the parameters, all of which should be lvalues:
1333 STR -- the saved data position.
1334 PAT -- the saved pattern position.
1335 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1336 REGSTART, REGEND -- arrays of string positions.
1337 REG_INFO -- array of information about each subexpression.
1339 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1340 `pend', `string1', `size1', `string2', and `size2'. */
1342 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1344 DEBUG_STATEMENT (unsigned failure_id;) \
1345 active_reg_t this_reg; \
1346 const unsigned char *string_temp; \
1348 assert (!FAIL_STACK_EMPTY ()); \
1350 /* Remove failure points and point to how many regs pushed. */ \
1351 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1352 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1353 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1355 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1357 DEBUG_POP (&failure_id); \
1358 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1360 /* If the saved string location is NULL, it came from an \
1361 on_failure_keep_string_jump opcode, and we want to throw away the \
1362 saved NULL, thus retaining our current position in the string. */ \
1363 string_temp = POP_FAILURE_POINTER (); \
1364 if (string_temp != NULL) \
1365 str = (const char *) string_temp; \
1367 DEBUG_PRINT2 (" Popping string %p: `", str); \
1368 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1369 DEBUG_PRINT1 ("'\n"); \
1371 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1372 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1373 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1375 /* Restore register info. */ \
1376 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1377 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1379 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1380 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1383 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1385 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1387 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1388 DEBUG_PRINT2 (" info: %p\n", \
1389 reg_info[this_reg].word.pointer); \
1391 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1392 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1394 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1395 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1399 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1401 reg_info[this_reg].word.integer = 0; \
1402 regend[this_reg] = 0; \
1403 regstart[this_reg] = 0; \
1405 highest_active_reg = high_reg; \
1408 set_regs_matched_done = 0; \
1409 DEBUG_STATEMENT (nfailure_points_popped++); \
1410 } /* POP_FAILURE_POINT */
1414 /* Structure for per-register (a.k.a. per-group) information.
1415 Other register information, such as the
1416 starting and ending positions (which are addresses), and the list of
1417 inner groups (which is a bits list) are maintained in separate
1420 We are making a (strictly speaking) nonportable assumption here: that
1421 the compiler will pack our bit fields into something that fits into
1422 the type of `word', i.e., is something that fits into one item on the
1426 /* Declarations and macros for re_match_2. */
1430 fail_stack_elt_t word;
1433 /* This field is one if this group can match the empty string,
1434 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1435 #define MATCH_NULL_UNSET_VALUE 3
1436 unsigned match_null_string_p : 2;
1437 unsigned is_active : 1;
1438 unsigned matched_something : 1;
1439 unsigned ever_matched_something : 1;
1441 } register_info_type;
1443 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1444 #define IS_ACTIVE(R) ((R).bits.is_active)
1445 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1446 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1449 /* Call this when have matched a real character; it sets `matched' flags
1450 for the subexpressions which we are currently inside. Also records
1451 that those subexprs have matched. */
1452 #define SET_REGS_MATCHED() \
1455 if (!set_regs_matched_done) \
1458 set_regs_matched_done = 1; \
1459 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1461 MATCHED_SOMETHING (reg_info[r]) \
1462 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1469 /* Registers are set to a sentinel when they haven't yet matched. */
1470 static char reg_unset_dummy;
1471 #define REG_UNSET_VALUE (®_unset_dummy)
1472 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1474 /* Subroutine declarations and macros for regex_compile. */
1476 static reg_errcode_t regex_compile _RE_ARGS ((const char *pattern, size_t size,
1477 reg_syntax_t syntax,
1478 struct re_pattern_buffer *bufp));
1479 static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg));
1480 static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1481 int arg1, int arg2));
1482 static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1483 int arg, unsigned char *end));
1484 static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1485 int arg1, int arg2, unsigned char *end));
1486 static boolean at_begline_loc_p _RE_ARGS ((const char *pattern, const char *p,
1487 reg_syntax_t syntax));
1488 static boolean at_endline_loc_p _RE_ARGS ((const char *p, const char *pend,
1489 reg_syntax_t syntax));
1490 static reg_errcode_t compile_range _RE_ARGS ((const char **p_ptr,
1493 reg_syntax_t syntax,
1496 /* Fetch the next character in the uncompiled pattern---translating it
1497 if necessary. Also cast from a signed character in the constant
1498 string passed to us by the user to an unsigned char that we can use
1499 as an array index (in, e.g., `translate'). */
1501 # define PATFETCH(c) \
1502 do {if (p == pend) return REG_EEND; \
1503 c = (unsigned char) *p++; \
1504 if (translate) c = (unsigned char) translate[c]; \
1508 /* Fetch the next character in the uncompiled pattern, with no
1510 #define PATFETCH_RAW(c) \
1511 do {if (p == pend) return REG_EEND; \
1512 c = (unsigned char) *p++; \
1515 /* Go backwards one character in the pattern. */
1516 #define PATUNFETCH p--
1519 /* If `translate' is non-null, return translate[D], else just D. We
1520 cast the subscript to translate because some data is declared as
1521 `char *', to avoid warnings when a string constant is passed. But
1522 when we use a character as a subscript we must make it unsigned. */
1524 # define TRANSLATE(d) \
1525 (translate ? (char) translate[(unsigned char) (d)] : (d))
1529 /* Macros for outputting the compiled pattern into `buffer'. */
1531 /* If the buffer isn't allocated when it comes in, use this. */
1532 #define INIT_BUF_SIZE 32
1534 /* Make sure we have at least N more bytes of space in buffer. */
1535 #define GET_BUFFER_SPACE(n) \
1536 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1539 /* Make sure we have one more byte of buffer space and then add C to it. */
1540 #define BUF_PUSH(c) \
1542 GET_BUFFER_SPACE (1); \
1543 *b++ = (unsigned char) (c); \
1547 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1548 #define BUF_PUSH_2(c1, c2) \
1550 GET_BUFFER_SPACE (2); \
1551 *b++ = (unsigned char) (c1); \
1552 *b++ = (unsigned char) (c2); \
1556 /* As with BUF_PUSH_2, except for three bytes. */
1557 #define BUF_PUSH_3(c1, c2, c3) \
1559 GET_BUFFER_SPACE (3); \
1560 *b++ = (unsigned char) (c1); \
1561 *b++ = (unsigned char) (c2); \
1562 *b++ = (unsigned char) (c3); \
1566 /* Store a jump with opcode OP at LOC to location TO. We store a
1567 relative address offset by the three bytes the jump itself occupies. */
1568 #define STORE_JUMP(op, loc, to) \
1569 store_op1 (op, loc, (int) ((to) - (loc) - 3))
1571 /* Likewise, for a two-argument jump. */
1572 #define STORE_JUMP2(op, loc, to, arg) \
1573 store_op2 (op, loc, (int) ((to) - (loc) - 3), arg)
1575 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1576 #define INSERT_JUMP(op, loc, to) \
1577 insert_op1 (op, loc, (int) ((to) - (loc) - 3), b)
1579 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1580 #define INSERT_JUMP2(op, loc, to, arg) \
1581 insert_op2 (op, loc, (int) ((to) - (loc) - 3), arg, b)
1584 /* This is not an arbitrary limit: the arguments which represent offsets
1585 into the pattern are two bytes long. So if 2^16 bytes turns out to
1586 be too small, many things would have to change. */
1587 /* Any other compiler which, like MSC, has allocation limit below 2^16
1588 bytes will have to use approach similar to what was done below for
1589 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1590 reallocating to 0 bytes. Such thing is not going to work too well.
1591 You have been warned!! */
1592 #if defined _MSC_VER && !defined WIN32
1593 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
1594 The REALLOC define eliminates a flurry of conversion warnings,
1595 but is not required. */
1596 # define MAX_BUF_SIZE 65500L
1597 # define REALLOC(p,s) realloc ((p), (size_t) (s))
1599 # define MAX_BUF_SIZE (1L << 16)
1600 # define REALLOC(p,s) realloc ((p), (s))
1603 /* Extend the buffer by twice its current size via realloc and
1604 reset the pointers that pointed into the old block to point to the
1605 correct places in the new one. If extending the buffer results in it
1606 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1607 #define EXTEND_BUFFER() \
1609 unsigned char *old_buffer = bufp->buffer; \
1610 if (bufp->allocated == MAX_BUF_SIZE) \
1612 bufp->allocated <<= 1; \
1613 if (bufp->allocated > MAX_BUF_SIZE) \
1614 bufp->allocated = MAX_BUF_SIZE; \
1615 bufp->buffer = (unsigned char *) REALLOC (bufp->buffer, bufp->allocated);\
1616 if (bufp->buffer == NULL) \
1617 return REG_ESPACE; \
1618 /* If the buffer moved, move all the pointers into it. */ \
1619 if (old_buffer != bufp->buffer) \
1621 b = (b - old_buffer) + bufp->buffer; \
1622 begalt = (begalt - old_buffer) + bufp->buffer; \
1623 if (fixup_alt_jump) \
1624 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1626 laststart = (laststart - old_buffer) + bufp->buffer; \
1627 if (pending_exact) \
1628 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1633 /* Since we have one byte reserved for the register number argument to
1634 {start,stop}_memory, the maximum number of groups we can report
1635 things about is what fits in that byte. */
1636 #define MAX_REGNUM 255
1638 /* But patterns can have more than `MAX_REGNUM' registers. We just
1639 ignore the excess. */
1640 typedef unsigned regnum_t;
1643 /* Macros for the compile stack. */
1645 /* Since offsets can go either forwards or backwards, this type needs to
1646 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1647 /* int may be not enough when sizeof(int) == 2. */
1648 typedef long pattern_offset_t;
1652 pattern_offset_t begalt_offset;
1653 pattern_offset_t fixup_alt_jump;
1654 pattern_offset_t inner_group_offset;
1655 pattern_offset_t laststart_offset;
1657 } compile_stack_elt_t;
1662 compile_stack_elt_t *stack;
1664 unsigned avail; /* Offset of next open position. */
1665 } compile_stack_type;
1668 #define INIT_COMPILE_STACK_SIZE 32
1670 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1671 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1673 /* The next available element. */
1674 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1677 /* Set the bit for character C in a list. */
1678 #define SET_LIST_BIT(c) \
1679 (b[((unsigned char) (c)) / BYTEWIDTH] \
1680 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1683 /* Get the next unsigned number in the uncompiled pattern. */
1684 #define GET_UNSIGNED_NUMBER(num) \
1688 while (ISDIGIT (c)) \
1692 num = num * 10 + c - '0'; \
1700 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
1701 /* The GNU C library provides support for user-defined character classes
1702 and the functions from ISO C amendement 1. */
1703 # ifdef CHARCLASS_NAME_MAX
1704 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
1706 /* This shouldn't happen but some implementation might still have this
1707 problem. Use a reasonable default value. */
1708 # define CHAR_CLASS_MAX_LENGTH 256
1712 # define IS_CHAR_CLASS(string) __wctype (string)
1714 # define IS_CHAR_CLASS(string) wctype (string)
1717 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1719 # define IS_CHAR_CLASS(string) \
1720 (STREQ (string, "alpha") || STREQ (string, "upper") \
1721 || STREQ (string, "lower") || STREQ (string, "digit") \
1722 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1723 || STREQ (string, "space") || STREQ (string, "print") \
1724 || STREQ (string, "punct") || STREQ (string, "graph") \
1725 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1728 #ifndef MATCH_MAY_ALLOCATE
1730 /* If we cannot allocate large objects within re_match_2_internal,
1731 we make the fail stack and register vectors global.
1732 The fail stack, we grow to the maximum size when a regexp
1734 The register vectors, we adjust in size each time we
1735 compile a regexp, according to the number of registers it needs. */
1737 static fail_stack_type fail_stack;
1739 /* Size with which the following vectors are currently allocated.
1740 That is so we can make them bigger as needed,
1741 but never make them smaller. */
1742 static int regs_allocated_size;
1744 static const char ** regstart, ** regend;
1745 static const char ** old_regstart, ** old_regend;
1746 static const char **best_regstart, **best_regend;
1747 static register_info_type *reg_info;
1748 static const char **reg_dummy;
1749 static register_info_type *reg_info_dummy;
1751 /* Make the register vectors big enough for NUM_REGS registers,
1752 but don't make them smaller. */
1755 regex_grow_registers (num_regs)
1758 if (num_regs > regs_allocated_size)
1760 RETALLOC_IF (regstart, num_regs, const char *);
1761 RETALLOC_IF (regend, num_regs, const char *);
1762 RETALLOC_IF (old_regstart, num_regs, const char *);
1763 RETALLOC_IF (old_regend, num_regs, const char *);
1764 RETALLOC_IF (best_regstart, num_regs, const char *);
1765 RETALLOC_IF (best_regend, num_regs, const char *);
1766 RETALLOC_IF (reg_info, num_regs, register_info_type);
1767 RETALLOC_IF (reg_dummy, num_regs, const char *);
1768 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1770 regs_allocated_size = num_regs;
1774 #endif /* not MATCH_MAY_ALLOCATE */
1776 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
1780 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1781 Returns one of error codes defined in `gnu-regex.h', or zero for success.
1783 Assumes the `allocated' (and perhaps `buffer') and `translate'
1784 fields are set in BUFP on entry.
1786 If it succeeds, results are put in BUFP (if it returns an error, the
1787 contents of BUFP are undefined):
1788 `buffer' is the compiled pattern;
1789 `syntax' is set to SYNTAX;
1790 `used' is set to the length of the compiled pattern;
1791 `fastmap_accurate' is zero;
1792 `re_nsub' is the number of subexpressions in PATTERN;
1793 `not_bol' and `not_eol' are zero;
1795 The `fastmap' and `newline_anchor' fields are neither
1796 examined nor set. */
1798 /* Return, freeing storage we allocated. */
1799 #define FREE_STACK_RETURN(value) \
1800 return (free (compile_stack.stack), value)
1802 static reg_errcode_t
1803 regex_compile (pattern, size, syntax, bufp)
1804 const char *pattern;
1806 reg_syntax_t syntax;
1807 struct re_pattern_buffer *bufp;
1809 /* We fetch characters from PATTERN here. Even though PATTERN is
1810 `char *' (i.e., signed), we declare these variables as unsigned, so
1811 they can be reliably used as array indices. */
1812 register unsigned char c, c1;
1814 /* A random temporary spot in PATTERN. */
1817 /* Points to the end of the buffer, where we should append. */
1818 register unsigned char *b;
1820 /* Keeps track of unclosed groups. */
1821 compile_stack_type compile_stack;
1823 /* Points to the current (ending) position in the pattern. */
1824 const char *p = pattern;
1825 const char *pend = pattern + size;
1827 /* How to translate the characters in the pattern. */
1828 RE_TRANSLATE_TYPE translate = bufp->translate;
1830 /* Address of the count-byte of the most recently inserted `exactn'
1831 command. This makes it possible to tell if a new exact-match
1832 character can be added to that command or if the character requires
1833 a new `exactn' command. */
1834 unsigned char *pending_exact = 0;
1836 /* Address of start of the most recently finished expression.
1837 This tells, e.g., postfix * where to find the start of its
1838 operand. Reset at the beginning of groups and alternatives. */
1839 unsigned char *laststart = 0;
1841 /* Address of beginning of regexp, or inside of last group. */
1842 unsigned char *begalt;
1844 /* Place in the uncompiled pattern (i.e., the {) to
1845 which to go back if the interval is invalid. */
1846 const char *beg_interval;
1848 /* Address of the place where a forward jump should go to the end of
1849 the containing expression. Each alternative of an `or' -- except the
1850 last -- ends with a forward jump of this sort. */
1851 unsigned char *fixup_alt_jump = 0;
1853 /* Counts open-groups as they are encountered. Remembered for the
1854 matching close-group on the compile stack, so the same register
1855 number is put in the stop_memory as the start_memory. */
1856 regnum_t regnum = 0;
1859 DEBUG_PRINT1 ("\nCompiling pattern: ");
1862 unsigned debug_count;
1864 for (debug_count = 0; debug_count < size; debug_count++)
1865 putchar (pattern[debug_count]);
1870 /* Initialize the compile stack. */
1871 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1872 if (compile_stack.stack == NULL)
1875 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1876 compile_stack.avail = 0;
1878 /* Initialize the pattern buffer. */
1879 bufp->syntax = syntax;
1880 bufp->fastmap_accurate = 0;
1881 bufp->not_bol = bufp->not_eol = 0;
1883 /* Set `used' to zero, so that if we return an error, the pattern
1884 printer (for debugging) will think there's no pattern. We reset it
1888 /* Always count groups, whether or not bufp->no_sub is set. */
1891 #if !defined emacs && !defined SYNTAX_TABLE
1892 /* Initialize the syntax table. */
1893 init_syntax_once ();
1896 if (bufp->allocated == 0)
1899 { /* If zero allocated, but buffer is non-null, try to realloc
1900 enough space. This loses if buffer's address is bogus, but
1901 that is the user's responsibility. */
1902 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1905 { /* Caller did not allocate a buffer. Do it for them. */
1906 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1908 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1910 bufp->allocated = INIT_BUF_SIZE;
1913 begalt = b = bufp->buffer;
1915 /* Loop through the uncompiled pattern until we're at the end. */
1924 if ( /* If at start of pattern, it's an operator. */
1926 /* If context independent, it's an operator. */
1927 || syntax & RE_CONTEXT_INDEP_ANCHORS
1928 /* Otherwise, depends on what's come before. */
1929 || at_begline_loc_p (pattern, p, syntax))
1939 if ( /* If at end of pattern, it's an operator. */
1941 /* If context independent, it's an operator. */
1942 || syntax & RE_CONTEXT_INDEP_ANCHORS
1943 /* Otherwise, depends on what's next. */
1944 || at_endline_loc_p (p, pend, syntax))
1954 if ((syntax & RE_BK_PLUS_QM)
1955 || (syntax & RE_LIMITED_OPS))
1959 /* If there is no previous pattern... */
1962 if (syntax & RE_CONTEXT_INVALID_OPS)
1963 FREE_STACK_RETURN (REG_BADRPT);
1964 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1969 /* Are we optimizing this jump? */
1970 boolean keep_string_p = false;
1972 /* 1 means zero (many) matches is allowed. */
1973 char zero_times_ok = 0, many_times_ok = 0;
1975 /* If there is a sequence of repetition chars, collapse it
1976 down to just one (the right one). We can't combine
1977 interval operators with these because of, e.g., `a{2}*',
1978 which should only match an even number of `a's. */
1982 zero_times_ok |= c != '+';
1983 many_times_ok |= c != '?';
1991 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
1994 else if (syntax & RE_BK_PLUS_QM && c == '\\')
1996 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1999 if (!(c1 == '+' || c1 == '?'))
2014 /* If we get here, we found another repeat character. */
2017 /* Star, etc. applied to an empty pattern is equivalent
2018 to an empty pattern. */
2022 /* Now we know whether or not zero matches is allowed
2023 and also whether or not two or more matches is allowed. */
2025 { /* More than one repetition is allowed, so put in at the
2026 end a backward relative jump from `b' to before the next
2027 jump we're going to put in below (which jumps from
2028 laststart to after this jump).
2030 But if we are at the `*' in the exact sequence `.*\n',
2031 insert an unconditional jump backwards to the .,
2032 instead of the beginning of the loop. This way we only
2033 push a failure point once, instead of every time
2034 through the loop. */
2035 assert (p - 1 > pattern);
2037 /* Allocate the space for the jump. */
2038 GET_BUFFER_SPACE (3);
2040 /* We know we are not at the first character of the pattern,
2041 because laststart was nonzero. And we've already
2042 incremented `p', by the way, to be the character after
2043 the `*'. Do we have to do something analogous here
2044 for null bytes, because of RE_DOT_NOT_NULL? */
2045 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2047 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2048 && !(syntax & RE_DOT_NEWLINE))
2049 { /* We have .*\n. */
2050 STORE_JUMP (jump, b, laststart);
2051 keep_string_p = true;
2054 /* Anything else. */
2055 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
2057 /* We've added more stuff to the buffer. */
2061 /* On failure, jump from laststart to b + 3, which will be the
2062 end of the buffer after this jump is inserted. */
2063 GET_BUFFER_SPACE (3);
2064 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2072 /* At least one repetition is required, so insert a
2073 `dummy_failure_jump' before the initial
2074 `on_failure_jump' instruction of the loop. This
2075 effects a skip over that instruction the first time
2076 we hit that loop. */
2077 GET_BUFFER_SPACE (3);
2078 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
2093 boolean had_char_class = false;
2095 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2097 /* Ensure that we have enough space to push a charset: the
2098 opcode, the length count, and the bitset; 34 bytes in all. */
2099 GET_BUFFER_SPACE (34);
2103 /* We test `*p == '^' twice, instead of using an if
2104 statement, so we only need one BUF_PUSH. */
2105 BUF_PUSH (*p == '^' ? charset_not : charset);
2109 /* Remember the first position in the bracket expression. */
2112 /* Push the number of bytes in the bitmap. */
2113 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
2115 /* Clear the whole map. */
2116 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
2118 /* charset_not matches newline according to a syntax bit. */
2119 if ((re_opcode_t) b[-2] == charset_not
2120 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2121 SET_LIST_BIT ('\n');
2123 /* Read in characters and ranges, setting map bits. */
2126 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2130 /* \ might escape characters inside [...] and [^...]. */
2131 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2133 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2140 /* Could be the end of the bracket expression. If it's
2141 not (i.e., when the bracket expression is `[]' so
2142 far), the ']' character bit gets set way below. */
2143 if (c == ']' && p != p1 + 1)
2146 /* Look ahead to see if it's a range when the last thing
2147 was a character class. */
2148 if (had_char_class && c == '-' && *p != ']')
2149 FREE_STACK_RETURN (REG_ERANGE);
2151 /* Look ahead to see if it's a range when the last thing
2152 was a character: if this is a hyphen not at the
2153 beginning or the end of a list, then it's the range
2156 && !(p - 2 >= pattern && p[-2] == '[')
2157 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2161 = compile_range (&p, pend, translate, syntax, b);
2162 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2165 else if (p[0] == '-' && p[1] != ']')
2166 { /* This handles ranges made up of characters only. */
2169 /* Move past the `-'. */
2172 ret = compile_range (&p, pend, translate, syntax, b);
2173 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2176 /* See if we're at the beginning of a possible character
2179 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2180 { /* Leave room for the null. */
2181 char str[CHAR_CLASS_MAX_LENGTH + 1];
2186 /* If pattern is `[[:'. */
2187 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2192 if ((c == ':' && *p == ']') || p == pend
2193 || c1 == CHAR_CLASS_MAX_LENGTH)
2199 /* If isn't a word bracketed by `[:' and `:]':
2200 undo the ending character, the letters, and leave
2201 the leading `:' and `[' (but set bits for them). */
2202 if (c == ':' && *p == ']')
2204 /* CYGNUS LOCAL: Skip this code if we don't have btowc(). btowc() is */
2205 /* defined in the 1994 Amendment 1 to ISO C and may not be present on */
2206 /* systems where we have wchar.h and wctype.h. */
2207 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H && defined HAVE_BTOWC)
2208 boolean is_lower = STREQ (str, "lower");
2209 boolean is_upper = STREQ (str, "upper");
2213 wt = IS_CHAR_CLASS (str);
2215 FREE_STACK_RETURN (REG_ECTYPE);
2217 /* Throw away the ] at the end of the character
2221 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2223 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
2226 if (__iswctype (__btowc (ch), wt))
2229 if (iswctype (btowc (ch), wt))
2233 if (translate && (is_upper || is_lower)
2234 && (ISUPPER (ch) || ISLOWER (ch)))
2238 had_char_class = true;
2241 boolean is_alnum = STREQ (str, "alnum");
2242 boolean is_alpha = STREQ (str, "alpha");
2243 boolean is_blank = STREQ (str, "blank");
2244 boolean is_cntrl = STREQ (str, "cntrl");
2245 boolean is_digit = STREQ (str, "digit");
2246 boolean is_graph = STREQ (str, "graph");
2247 boolean is_lower = STREQ (str, "lower");
2248 boolean is_print = STREQ (str, "print");
2249 boolean is_punct = STREQ (str, "punct");
2250 boolean is_space = STREQ (str, "space");
2251 boolean is_upper = STREQ (str, "upper");
2252 boolean is_xdigit = STREQ (str, "xdigit");
2254 if (!IS_CHAR_CLASS (str))
2255 FREE_STACK_RETURN (REG_ECTYPE);
2257 /* Throw away the ] at the end of the character
2261 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2263 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2265 /* This was split into 3 if's to
2266 avoid an arbitrary limit in some compiler. */
2267 if ( (is_alnum && ISALNUM (ch))
2268 || (is_alpha && ISALPHA (ch))
2269 || (is_blank && ISBLANK (ch))
2270 || (is_cntrl && ISCNTRL (ch)))
2272 if ( (is_digit && ISDIGIT (ch))
2273 || (is_graph && ISGRAPH (ch))
2274 || (is_lower && ISLOWER (ch))
2275 || (is_print && ISPRINT (ch)))
2277 if ( (is_punct && ISPUNCT (ch))
2278 || (is_space && ISSPACE (ch))
2279 || (is_upper && ISUPPER (ch))
2280 || (is_xdigit && ISXDIGIT (ch)))
2282 if ( translate && (is_upper || is_lower)
2283 && (ISUPPER (ch) || ISLOWER (ch)))
2286 had_char_class = true;
2287 #endif /* libc || wctype.h */
2296 had_char_class = false;
2301 had_char_class = false;
2306 /* Discard any (non)matching list bytes that are all 0 at the
2307 end of the map. Decrease the map-length byte too. */
2308 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2316 if (syntax & RE_NO_BK_PARENS)
2323 if (syntax & RE_NO_BK_PARENS)
2330 if (syntax & RE_NEWLINE_ALT)
2337 if (syntax & RE_NO_BK_VBAR)
2344 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2345 goto handle_interval;
2351 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2353 /* Do not translate the character after the \, so that we can
2354 distinguish, e.g., \B from \b, even if we normally would
2355 translate, e.g., B to b. */
2361 if (syntax & RE_NO_BK_PARENS)
2362 goto normal_backslash;
2368 if (COMPILE_STACK_FULL)
2370 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2371 compile_stack_elt_t);
2372 if (compile_stack.stack == NULL) return REG_ESPACE;
2374 compile_stack.size <<= 1;
2377 /* These are the values to restore when we hit end of this
2378 group. They are all relative offsets, so that if the
2379 whole pattern moves because of realloc, they will still
2381 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2382 COMPILE_STACK_TOP.fixup_alt_jump
2383 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2384 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2385 COMPILE_STACK_TOP.regnum = regnum;
2387 /* We will eventually replace the 0 with the number of
2388 groups inner to this one. But do not push a
2389 start_memory for groups beyond the last one we can
2390 represent in the compiled pattern. */
2391 if (regnum <= MAX_REGNUM)
2393 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2394 BUF_PUSH_3 (start_memory, regnum, 0);
2397 compile_stack.avail++;
2402 /* If we've reached MAX_REGNUM groups, then this open
2403 won't actually generate any code, so we'll have to
2404 clear pending_exact explicitly. */
2410 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2412 if (COMPILE_STACK_EMPTY)
2414 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2415 goto normal_backslash;
2417 FREE_STACK_RETURN (REG_ERPAREN);
2422 { /* Push a dummy failure point at the end of the
2423 alternative for a possible future
2424 `pop_failure_jump' to pop. See comments at
2425 `push_dummy_failure' in `re_match_2'. */
2426 BUF_PUSH (push_dummy_failure);
2428 /* We allocated space for this jump when we assigned
2429 to `fixup_alt_jump', in the `handle_alt' case below. */
2430 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2433 /* See similar code for backslashed left paren above. */
2434 if (COMPILE_STACK_EMPTY)
2436 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2439 FREE_STACK_RETURN (REG_ERPAREN);
2442 /* Since we just checked for an empty stack above, this
2443 ``can't happen''. */
2444 assert (compile_stack.avail != 0);
2446 /* We don't just want to restore into `regnum', because
2447 later groups should continue to be numbered higher,
2448 as in `(ab)c(de)' -- the second group is #2. */
2449 regnum_t this_group_regnum;
2451 compile_stack.avail--;
2452 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2454 = COMPILE_STACK_TOP.fixup_alt_jump
2455 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2457 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2458 this_group_regnum = COMPILE_STACK_TOP.regnum;
2459 /* If we've reached MAX_REGNUM groups, then this open
2460 won't actually generate any code, so we'll have to
2461 clear pending_exact explicitly. */
2464 /* We're at the end of the group, so now we know how many
2465 groups were inside this one. */
2466 if (this_group_regnum <= MAX_REGNUM)
2468 unsigned char *inner_group_loc
2469 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2471 *inner_group_loc = regnum - this_group_regnum;
2472 BUF_PUSH_3 (stop_memory, this_group_regnum,
2473 regnum - this_group_regnum);
2479 case '|': /* `\|'. */
2480 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2481 goto normal_backslash;
2483 if (syntax & RE_LIMITED_OPS)
2486 /* Insert before the previous alternative a jump which
2487 jumps to this alternative if the former fails. */
2488 GET_BUFFER_SPACE (3);
2489 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2493 /* The alternative before this one has a jump after it
2494 which gets executed if it gets matched. Adjust that
2495 jump so it will jump to this alternative's analogous
2496 jump (put in below, which in turn will jump to the next
2497 (if any) alternative's such jump, etc.). The last such
2498 jump jumps to the correct final destination. A picture:
2504 If we are at `b', then fixup_alt_jump right now points to a
2505 three-byte space after `a'. We'll put in the jump, set
2506 fixup_alt_jump to right after `b', and leave behind three
2507 bytes which we'll fill in when we get to after `c'. */
2510 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2512 /* Mark and leave space for a jump after this alternative,
2513 to be filled in later either by next alternative or
2514 when know we're at the end of a series of alternatives. */
2516 GET_BUFFER_SPACE (3);
2525 /* If \{ is a literal. */
2526 if (!(syntax & RE_INTERVALS)
2527 /* If we're at `\{' and it's not the open-interval
2529 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2530 || (p - 2 == pattern && p == pend))
2531 goto normal_backslash;
2535 /* If got here, then the syntax allows intervals. */
2537 /* At least (most) this many matches must be made. */
2538 int lower_bound = -1, upper_bound = -1;
2540 beg_interval = p - 1;
2544 if (syntax & RE_NO_BK_BRACES)
2545 goto unfetch_interval;
2547 FREE_STACK_RETURN (REG_EBRACE);
2550 GET_UNSIGNED_NUMBER (lower_bound);
2554 GET_UNSIGNED_NUMBER (upper_bound);
2555 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2558 /* Interval such as `{1}' => match exactly once. */
2559 upper_bound = lower_bound;
2561 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2562 || lower_bound > upper_bound)
2564 if (syntax & RE_NO_BK_BRACES)
2565 goto unfetch_interval;
2567 FREE_STACK_RETURN (REG_BADBR);
2570 if (!(syntax & RE_NO_BK_BRACES))
2572 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2579 if (syntax & RE_NO_BK_BRACES)
2580 goto unfetch_interval;
2582 FREE_STACK_RETURN (REG_BADBR);
2585 /* We just parsed a valid interval. */
2587 /* If it's invalid to have no preceding re. */
2590 if (syntax & RE_CONTEXT_INVALID_OPS)
2591 FREE_STACK_RETURN (REG_BADRPT);
2592 else if (syntax & RE_CONTEXT_INDEP_OPS)
2595 goto unfetch_interval;
2598 /* If the upper bound is zero, don't want to succeed at
2599 all; jump from `laststart' to `b + 3', which will be
2600 the end of the buffer after we insert the jump. */
2601 if (upper_bound == 0)
2603 GET_BUFFER_SPACE (3);
2604 INSERT_JUMP (jump, laststart, b + 3);
2608 /* Otherwise, we have a nontrivial interval. When
2609 we're all done, the pattern will look like:
2610 set_number_at <jump count> <upper bound>
2611 set_number_at <succeed_n count> <lower bound>
2612 succeed_n <after jump addr> <succeed_n count>
2614 jump_n <succeed_n addr> <jump count>
2615 (The upper bound and `jump_n' are omitted if
2616 `upper_bound' is 1, though.) */
2618 { /* If the upper bound is > 1, we need to insert
2619 more at the end of the loop. */
2620 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2622 GET_BUFFER_SPACE (nbytes);
2624 /* Initialize lower bound of the `succeed_n', even
2625 though it will be set during matching by its
2626 attendant `set_number_at' (inserted next),
2627 because `re_compile_fastmap' needs to know.
2628 Jump to the `jump_n' we might insert below. */
2629 INSERT_JUMP2 (succeed_n, laststart,
2630 b + 5 + (upper_bound > 1) * 5,
2634 /* Code to initialize the lower bound. Insert
2635 before the `succeed_n'. The `5' is the last two
2636 bytes of this `set_number_at', plus 3 bytes of
2637 the following `succeed_n'. */
2638 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2641 if (upper_bound > 1)
2642 { /* More than one repetition is allowed, so
2643 append a backward jump to the `succeed_n'
2644 that starts this interval.
2646 When we've reached this during matching,
2647 we'll have matched the interval once, so
2648 jump back only `upper_bound - 1' times. */
2649 STORE_JUMP2 (jump_n, b, laststart + 5,
2653 /* The location we want to set is the second
2654 parameter of the `jump_n'; that is `b-2' as
2655 an absolute address. `laststart' will be
2656 the `set_number_at' we're about to insert;
2657 `laststart+3' the number to set, the source
2658 for the relative address. But we are
2659 inserting into the middle of the pattern --
2660 so everything is getting moved up by 5.
2661 Conclusion: (b - 2) - (laststart + 3) + 5,
2662 i.e., b - laststart.
2664 We insert this at the beginning of the loop
2665 so that if we fail during matching, we'll
2666 reinitialize the bounds. */
2667 insert_op2 (set_number_at, laststart, b - laststart,
2668 upper_bound - 1, b);
2673 beg_interval = NULL;
2678 /* If an invalid interval, match the characters as literals. */
2679 assert (beg_interval);
2681 beg_interval = NULL;
2683 /* normal_char and normal_backslash need `c'. */
2686 if (!(syntax & RE_NO_BK_BRACES))
2688 if (p > pattern && p[-1] == '\\')
2689 goto normal_backslash;
2694 /* There is no way to specify the before_dot and after_dot
2695 operators. rms says this is ok. --karl */
2703 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2709 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2715 if (syntax & RE_NO_GNU_OPS)
2718 BUF_PUSH (wordchar);
2723 if (syntax & RE_NO_GNU_OPS)
2726 BUF_PUSH (notwordchar);
2731 if (syntax & RE_NO_GNU_OPS)
2737 if (syntax & RE_NO_GNU_OPS)
2743 if (syntax & RE_NO_GNU_OPS)
2745 BUF_PUSH (wordbound);
2749 if (syntax & RE_NO_GNU_OPS)
2751 BUF_PUSH (notwordbound);
2755 if (syntax & RE_NO_GNU_OPS)
2761 if (syntax & RE_NO_GNU_OPS)
2766 case '1': case '2': case '3': case '4': case '5':
2767 case '6': case '7': case '8': case '9':
2768 if (syntax & RE_NO_BK_REFS)
2774 FREE_STACK_RETURN (REG_ESUBREG);
2776 /* Can't back reference to a subexpression if inside of it. */
2777 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
2781 BUF_PUSH_2 (duplicate, c1);
2787 if (syntax & RE_BK_PLUS_QM)
2790 goto normal_backslash;
2794 /* You might think it would be useful for \ to mean
2795 not to translate; but if we don't translate it
2796 it will never match anything. */
2804 /* Expects the character in `c'. */
2806 /* If no exactn currently being built. */
2809 /* If last exactn not at current position. */
2810 || pending_exact + *pending_exact + 1 != b
2812 /* We have only one byte following the exactn for the count. */
2813 || *pending_exact == (1 << BYTEWIDTH) - 1
2815 /* If followed by a repetition operator. */
2816 || *p == '*' || *p == '^'
2817 || ((syntax & RE_BK_PLUS_QM)
2818 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2819 : (*p == '+' || *p == '?'))
2820 || ((syntax & RE_INTERVALS)
2821 && ((syntax & RE_NO_BK_BRACES)
2823 : (p[0] == '\\' && p[1] == '{'))))
2825 /* Start building a new exactn. */
2829 BUF_PUSH_2 (exactn, 0);
2830 pending_exact = b - 1;
2837 } /* while p != pend */
2840 /* Through the pattern now. */
2843 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2845 if (!COMPILE_STACK_EMPTY)
2846 FREE_STACK_RETURN (REG_EPAREN);
2848 /* If we don't want backtracking, force success
2849 the first time we reach the end of the compiled pattern. */
2850 if (syntax & RE_NO_POSIX_BACKTRACKING)
2853 free (compile_stack.stack);
2855 /* We have succeeded; set the length of the buffer. */
2856 bufp->used = b - bufp->buffer;
2861 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2862 print_compiled_pattern (bufp);
2866 #ifndef MATCH_MAY_ALLOCATE
2867 /* Initialize the failure stack to the largest possible stack. This
2868 isn't necessary unless we're trying to avoid calling alloca in
2869 the search and match routines. */
2871 int num_regs = bufp->re_nsub + 1;
2873 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2874 is strictly greater than re_max_failures, the largest possible stack
2875 is 2 * re_max_failures failure points. */
2876 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2878 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2881 if (! fail_stack.stack)
2883 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2884 * sizeof (fail_stack_elt_t));
2887 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2889 * sizeof (fail_stack_elt_t)));
2890 # else /* not emacs */
2891 if (! fail_stack.stack)
2893 = (fail_stack_elt_t *) malloc (fail_stack.size
2894 * sizeof (fail_stack_elt_t));
2897 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2899 * sizeof (fail_stack_elt_t)));
2900 # endif /* not emacs */
2903 regex_grow_registers (num_regs);
2905 #endif /* not MATCH_MAY_ALLOCATE */
2908 } /* regex_compile */
2910 /* Subroutines for `regex_compile'. */
2912 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2915 store_op1 (op, loc, arg)
2920 *loc = (unsigned char) op;
2921 STORE_NUMBER (loc + 1, arg);
2925 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2928 store_op2 (op, loc, arg1, arg2)
2933 *loc = (unsigned char) op;
2934 STORE_NUMBER (loc + 1, arg1);
2935 STORE_NUMBER (loc + 3, arg2);
2939 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
2940 for OP followed by two-byte integer parameter ARG. */
2943 insert_op1 (op, loc, arg, end)
2949 register unsigned char *pfrom = end;
2950 register unsigned char *pto = end + 3;
2952 while (pfrom != loc)
2955 store_op1 (op, loc, arg);
2959 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2962 insert_op2 (op, loc, arg1, arg2, end)
2968 register unsigned char *pfrom = end;
2969 register unsigned char *pto = end + 5;
2971 while (pfrom != loc)
2974 store_op2 (op, loc, arg1, arg2);
2978 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
2979 after an alternative or a begin-subexpression. We assume there is at
2980 least one character before the ^. */
2983 at_begline_loc_p (pattern, p, syntax)
2984 const char *pattern, *p;
2985 reg_syntax_t syntax;
2987 const char *prev = p - 2;
2988 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
2991 /* After a subexpression? */
2992 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
2993 /* After an alternative? */
2994 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
2998 /* The dual of at_begline_loc_p. This one is for $. We assume there is
2999 at least one character after the $, i.e., `P < PEND'. */
3002 at_endline_loc_p (p, pend, syntax)
3003 const char *p, *pend;
3004 reg_syntax_t syntax;
3006 const char *next = p;
3007 boolean next_backslash = *next == '\\';
3008 const char *next_next = p + 1 < pend ? p + 1 : 0;
3011 /* Before a subexpression? */
3012 (syntax & RE_NO_BK_PARENS ? *next == ')'
3013 : next_backslash && next_next && *next_next == ')')
3014 /* Before an alternative? */
3015 || (syntax & RE_NO_BK_VBAR ? *next == '|'
3016 : next_backslash && next_next && *next_next == '|');
3020 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
3021 false if it's not. */
3024 group_in_compile_stack (compile_stack, regnum)
3025 compile_stack_type compile_stack;
3030 for (this_element = compile_stack.avail - 1;
3033 if (compile_stack.stack[this_element].regnum == regnum)
3040 /* Read the ending character of a range (in a bracket expression) from the
3041 uncompiled pattern *P_PTR (which ends at PEND). We assume the
3042 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
3043 Then we set the translation of all bits between the starting and
3044 ending characters (inclusive) in the compiled pattern B.
3046 Return an error code.
3048 We use these short variable names so we can use the same macros as
3049 `regex_compile' itself. */
3051 static reg_errcode_t
3052 compile_range (p_ptr, pend, translate, syntax, b)
3053 const char **p_ptr, *pend;
3054 RE_TRANSLATE_TYPE translate;
3055 reg_syntax_t syntax;
3060 const char *p = *p_ptr;
3061 unsigned int range_start, range_end;
3066 /* Even though the pattern is a signed `char *', we need to fetch
3067 with unsigned char *'s; if the high bit of the pattern character
3068 is set, the range endpoints will be negative if we fetch using a
3071 We also want to fetch the endpoints without translating them; the
3072 appropriate translation is done in the bit-setting loop below. */
3073 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
3074 range_start = ((const unsigned char *) p)[-2];
3075 range_end = ((const unsigned char *) p)[0];
3077 /* Have to increment the pointer into the pattern string, so the
3078 caller isn't still at the ending character. */
3081 /* If the start is after the end, the range is empty. */
3082 if (range_start > range_end)
3083 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
3085 /* Here we see why `this_char' has to be larger than an `unsigned
3086 char' -- the range is inclusive, so if `range_end' == 0xff
3087 (assuming 8-bit characters), we would otherwise go into an infinite
3088 loop, since all characters <= 0xff. */
3089 for (this_char = range_start; this_char <= range_end; this_char++)
3091 SET_LIST_BIT (TRANSLATE (this_char));
3097 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3098 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3099 characters can start a string that matches the pattern. This fastmap
3100 is used by re_search to skip quickly over impossible starting points.
3102 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3103 area as BUFP->fastmap.
3105 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3108 Returns 0 if we succeed, -2 if an internal error. */
3111 re_compile_fastmap (bufp)
3112 struct re_pattern_buffer *bufp;
3115 #ifdef MATCH_MAY_ALLOCATE
3116 fail_stack_type fail_stack;
3118 #ifndef REGEX_MALLOC
3122 register char *fastmap = bufp->fastmap;
3123 unsigned char *pattern = bufp->buffer;
3124 unsigned char *p = pattern;
3125 register unsigned char *pend = pattern + bufp->used;
3128 /* This holds the pointer to the failure stack, when
3129 it is allocated relocatably. */
3130 fail_stack_elt_t *failure_stack_ptr;
3133 /* Assume that each path through the pattern can be null until
3134 proven otherwise. We set this false at the bottom of switch
3135 statement, to which we get only if a particular path doesn't
3136 match the empty string. */
3137 boolean path_can_be_null = true;
3139 /* We aren't doing a `succeed_n' to begin with. */
3140 boolean succeed_n_p = false;
3142 assert (fastmap != NULL && p != NULL);
3145 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
3146 bufp->fastmap_accurate = 1; /* It will be when we're done. */
3147 bufp->can_be_null = 0;
3151 if (p == pend || *p == succeed)
3153 /* We have reached the (effective) end of pattern. */
3154 if (!FAIL_STACK_EMPTY ())
3156 bufp->can_be_null |= path_can_be_null;
3158 /* Reset for next path. */
3159 path_can_be_null = true;
3161 p = fail_stack.stack[--fail_stack.avail].pointer;
3169 /* We should never be about to go beyond the end of the pattern. */
3172 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3175 /* I guess the idea here is to simply not bother with a fastmap
3176 if a backreference is used, since it's too hard to figure out
3177 the fastmap for the corresponding group. Setting
3178 `can_be_null' stops `re_search_2' from using the fastmap, so
3179 that is all we do. */
3181 bufp->can_be_null = 1;
3185 /* Following are the cases which match a character. These end
3194 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3195 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3201 /* Chars beyond end of map must be allowed. */
3202 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3205 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3206 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3212 for (j = 0; j < (1 << BYTEWIDTH); j++)
3213 if (SYNTAX (j) == Sword)
3219 for (j = 0; j < (1 << BYTEWIDTH); j++)
3220 if (SYNTAX (j) != Sword)
3227 int fastmap_newline = fastmap['\n'];
3229 /* `.' matches anything ... */
3230 for (j = 0; j < (1 << BYTEWIDTH); j++)
3233 /* ... except perhaps newline. */
3234 if (!(bufp->syntax & RE_DOT_NEWLINE))
3235 fastmap['\n'] = fastmap_newline;
3237 /* Return if we have already set `can_be_null'; if we have,
3238 then the fastmap is irrelevant. Something's wrong here. */
3239 else if (bufp->can_be_null)
3242 /* Otherwise, have to check alternative paths. */
3249 for (j = 0; j < (1 << BYTEWIDTH); j++)
3250 if (SYNTAX (j) == (enum syntaxcode) k)
3257 for (j = 0; j < (1 << BYTEWIDTH); j++)
3258 if (SYNTAX (j) != (enum syntaxcode) k)
3263 /* All cases after this match the empty string. These end with
3283 case push_dummy_failure:
3288 case pop_failure_jump:
3289 case maybe_pop_jump:
3292 case dummy_failure_jump:
3293 EXTRACT_NUMBER_AND_INCR (j, p);
3298 /* Jump backward implies we just went through the body of a
3299 loop and matched nothing. Opcode jumped to should be
3300 `on_failure_jump' or `succeed_n'. Just treat it like an
3301 ordinary jump. For a * loop, it has pushed its failure
3302 point already; if so, discard that as redundant. */
3303 if ((re_opcode_t) *p != on_failure_jump
3304 && (re_opcode_t) *p != succeed_n)
3308 EXTRACT_NUMBER_AND_INCR (j, p);
3311 /* If what's on the stack is where we are now, pop it. */
3312 if (!FAIL_STACK_EMPTY ()
3313 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3319 case on_failure_jump:
3320 case on_failure_keep_string_jump:
3321 handle_on_failure_jump:
3322 EXTRACT_NUMBER_AND_INCR (j, p);
3324 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3325 end of the pattern. We don't want to push such a point,
3326 since when we restore it above, entering the switch will
3327 increment `p' past the end of the pattern. We don't need
3328 to push such a point since we obviously won't find any more
3329 fastmap entries beyond `pend'. Such a pattern can match
3330 the null string, though. */
3333 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3335 RESET_FAIL_STACK ();
3340 bufp->can_be_null = 1;
3344 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3345 succeed_n_p = false;
3352 /* Get to the number of times to succeed. */
3355 /* Increment p past the n for when k != 0. */
3356 EXTRACT_NUMBER_AND_INCR (k, p);
3360 succeed_n_p = true; /* Spaghetti code alert. */
3361 goto handle_on_failure_jump;
3378 abort (); /* We have listed all the cases. */
3381 /* Getting here means we have found the possible starting
3382 characters for one path of the pattern -- and that the empty
3383 string does not match. We need not follow this path further.
3384 Instead, look at the next alternative (remembered on the
3385 stack), or quit if no more. The test at the top of the loop
3386 does these things. */
3387 path_can_be_null = false;
3391 /* Set `can_be_null' for the last path (also the first path, if the
3392 pattern is empty). */
3393 bufp->can_be_null |= path_can_be_null;
3396 RESET_FAIL_STACK ();
3398 } /* re_compile_fastmap */
3400 weak_alias (__re_compile_fastmap, re_compile_fastmap)
3403 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3404 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3405 this memory for recording register information. STARTS and ENDS
3406 must be allocated using the malloc library routine, and must each
3407 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3409 If NUM_REGS == 0, then subsequent matches should allocate their own
3412 Unless this function is called, the first search or match using
3413 PATTERN_BUFFER will allocate its own register data, without
3414 freeing the old data. */
3417 re_set_registers (bufp, regs, num_regs, starts, ends)
3418 struct re_pattern_buffer *bufp;
3419 struct re_registers *regs;
3421 regoff_t *starts, *ends;
3425 bufp->regs_allocated = REGS_REALLOCATE;
3426 regs->num_regs = num_regs;
3427 regs->start = starts;
3432 bufp->regs_allocated = REGS_UNALLOCATED;
3434 regs->start = regs->end = (regoff_t *) 0;
3438 weak_alias (__re_set_registers, re_set_registers)
3441 /* Searching routines. */
3443 /* Like re_search_2, below, but only one string is specified, and
3444 doesn't let you say where to stop matching. */
3447 re_search (bufp, string, size, startpos, range, regs)
3448 struct re_pattern_buffer *bufp;
3450 int size, startpos, range;
3451 struct re_registers *regs;
3453 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3457 weak_alias (__re_search, re_search)
3461 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3462 virtual concatenation of STRING1 and STRING2, starting first at index
3463 STARTPOS, then at STARTPOS + 1, and so on.
3465 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3467 RANGE is how far to scan while trying to match. RANGE = 0 means try
3468 only at STARTPOS; in general, the last start tried is STARTPOS +
3471 In REGS, return the indices of the virtual concatenation of STRING1
3472 and STRING2 that matched the entire BUFP->buffer and its contained
3475 Do not consider matching one past the index STOP in the virtual
3476 concatenation of STRING1 and STRING2.
3478 We return either the position in the strings at which the match was
3479 found, -1 if no match, or -2 if error (such as failure
3483 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3484 struct re_pattern_buffer *bufp;
3485 const char *string1, *string2;
3489 struct re_registers *regs;
3493 register char *fastmap = bufp->fastmap;
3494 register RE_TRANSLATE_TYPE translate = bufp->translate;
3495 int total_size = size1 + size2;
3496 int endpos = startpos + range;
3498 /* Check for out-of-range STARTPOS. */
3499 if (startpos < 0 || startpos > total_size)
3502 /* Fix up RANGE if it might eventually take us outside
3503 the virtual concatenation of STRING1 and STRING2.
3504 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3506 range = 0 - startpos;
3507 else if (endpos > total_size)
3508 range = total_size - startpos;
3510 /* If the search isn't to be a backwards one, don't waste time in a
3511 search for a pattern that must be anchored. */
3512 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3521 /* In a forward search for something that starts with \=.
3522 don't keep searching past point. */
3523 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3525 range = PT - startpos;
3531 /* Update the fastmap now if not correct already. */
3532 if (fastmap && !bufp->fastmap_accurate)
3533 if (re_compile_fastmap (bufp) == -2)
3536 /* Loop through the string, looking for a place to start matching. */
3539 /* If a fastmap is supplied, skip quickly over characters that
3540 cannot be the start of a match. If the pattern can match the
3541 null string, however, we don't need to skip characters; we want
3542 the first null string. */
3543 if (fastmap && startpos < total_size && !bufp->can_be_null)
3545 if (range > 0) /* Searching forwards. */
3547 register const char *d;
3548 register int lim = 0;
3551 if (startpos < size1 && startpos + range >= size1)
3552 lim = range - (size1 - startpos);
3554 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3556 /* Written out as an if-else to avoid testing `translate'
3560 && !fastmap[(unsigned char)
3561 translate[(unsigned char) *d++]])
3564 while (range > lim && !fastmap[(unsigned char) *d++])
3567 startpos += irange - range;
3569 else /* Searching backwards. */
3571 register char c = (size1 == 0 || startpos >= size1
3572 ? string2[startpos - size1]
3573 : string1[startpos]);
3575 if (!fastmap[(unsigned char) TRANSLATE (c)])
3580 /* If can't match the null string, and that's all we have left, fail. */
3581 if (range >= 0 && startpos == total_size && fastmap
3582 && !bufp->can_be_null)
3585 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3586 startpos, regs, stop);
3587 #ifndef REGEX_MALLOC
3616 weak_alias (__re_search_2, re_search_2)
3619 /* This converts PTR, a pointer into one of the search strings `string1'
3620 and `string2' into an offset from the beginning of that string. */
3621 #define POINTER_TO_OFFSET(ptr) \
3622 (FIRST_STRING_P (ptr) \
3623 ? ((regoff_t) ((ptr) - string1)) \
3624 : ((regoff_t) ((ptr) - string2 + size1)))
3626 /* Macros for dealing with the split strings in re_match_2. */
3628 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3630 /* Call before fetching a character with *d. This switches over to
3631 string2 if necessary. */
3632 #define PREFETCH() \
3635 /* End of string2 => fail. */ \
3636 if (dend == end_match_2) \
3638 /* End of string1 => advance to string2. */ \
3640 dend = end_match_2; \
3644 /* Test if at very beginning or at very end of the virtual concatenation
3645 of `string1' and `string2'. If only one string, it's `string2'. */
3646 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3647 #define AT_STRINGS_END(d) ((d) == end2)
3650 /* Test if D points to a character which is word-constituent. We have
3651 two special cases to check for: if past the end of string1, look at
3652 the first character in string2; and if before the beginning of
3653 string2, look at the last character in string1. */
3654 #define WORDCHAR_P(d) \
3655 (SYNTAX ((d) == end1 ? *string2 \
3656 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3659 /* Disabled due to a compiler bug -- see comment at case wordbound */
3661 /* Test if the character before D and the one at D differ with respect
3662 to being word-constituent. */
3663 #define AT_WORD_BOUNDARY(d) \
3664 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3665 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3668 /* Free everything we malloc. */
3669 #ifdef MATCH_MAY_ALLOCATE
3670 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3671 # define FREE_VARIABLES() \
3673 REGEX_FREE_STACK (fail_stack.stack); \
3674 FREE_VAR (regstart); \
3675 FREE_VAR (regend); \
3676 FREE_VAR (old_regstart); \
3677 FREE_VAR (old_regend); \
3678 FREE_VAR (best_regstart); \
3679 FREE_VAR (best_regend); \
3680 FREE_VAR (reg_info); \
3681 FREE_VAR (reg_dummy); \
3682 FREE_VAR (reg_info_dummy); \
3685 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3686 #endif /* not MATCH_MAY_ALLOCATE */
3688 /* These values must meet several constraints. They must not be valid
3689 register values; since we have a limit of 255 registers (because
3690 we use only one byte in the pattern for the register number), we can
3691 use numbers larger than 255. They must differ by 1, because of
3692 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3693 be larger than the value for the highest register, so we do not try
3694 to actually save any registers when none are active. */
3695 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3696 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3698 /* Matching routines. */
3700 #ifndef emacs /* Emacs never uses this. */
3701 /* re_match is like re_match_2 except it takes only a single string. */
3704 re_match (bufp, string, size, pos, regs)
3705 struct re_pattern_buffer *bufp;
3708 struct re_registers *regs;
3710 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3712 # ifndef REGEX_MALLOC
3720 weak_alias (__re_match, re_match)
3722 #endif /* not emacs */
3724 static boolean group_match_null_string_p _RE_ARGS ((unsigned char **p,
3726 register_info_type *reg_info));
3727 static boolean alt_match_null_string_p _RE_ARGS ((unsigned char *p,
3729 register_info_type *reg_info));
3730 static boolean common_op_match_null_string_p _RE_ARGS ((unsigned char **p,
3732 register_info_type *reg_info));
3733 static int bcmp_translate _RE_ARGS ((const char *s1, const char *s2,
3734 int len, char *translate));
3736 /* re_match_2 matches the compiled pattern in BUFP against the
3737 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3738 and SIZE2, respectively). We start matching at POS, and stop
3741 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3742 store offsets for the substring each group matched in REGS. See the
3743 documentation for exactly how many groups we fill.
3745 We return -1 if no match, -2 if an internal error (such as the
3746 failure stack overflowing). Otherwise, we return the length of the
3747 matched substring. */
3750 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3751 struct re_pattern_buffer *bufp;
3752 const char *string1, *string2;
3755 struct re_registers *regs;
3758 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3760 #ifndef REGEX_MALLOC
3768 weak_alias (__re_match_2, re_match_2)
3771 /* This is a separate function so that we can force an alloca cleanup
3774 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3775 struct re_pattern_buffer *bufp;
3776 const char *string1, *string2;
3779 struct re_registers *regs;
3782 /* General temporaries. */
3786 /* Just past the end of the corresponding string. */
3787 const char *end1, *end2;
3789 /* Pointers into string1 and string2, just past the last characters in
3790 each to consider matching. */
3791 const char *end_match_1, *end_match_2;
3793 /* Where we are in the data, and the end of the current string. */
3794 const char *d, *dend;
3796 /* Where we are in the pattern, and the end of the pattern. */
3797 unsigned char *p = bufp->buffer;
3798 register unsigned char *pend = p + bufp->used;
3800 /* Mark the opcode just after a start_memory, so we can test for an
3801 empty subpattern when we get to the stop_memory. */
3802 unsigned char *just_past_start_mem = 0;
3804 /* We use this to map every character in the string. */
3805 RE_TRANSLATE_TYPE translate = bufp->translate;
3807 /* Failure point stack. Each place that can handle a failure further
3808 down the line pushes a failure point on this stack. It consists of
3809 restart, regend, and reg_info for all registers corresponding to
3810 the subexpressions we're currently inside, plus the number of such
3811 registers, and, finally, two char *'s. The first char * is where
3812 to resume scanning the pattern; the second one is where to resume
3813 scanning the strings. If the latter is zero, the failure point is
3814 a ``dummy''; if a failure happens and the failure point is a dummy,
3815 it gets discarded and the next next one is tried. */
3816 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3817 fail_stack_type fail_stack;
3820 static unsigned failure_id = 0;
3821 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3825 /* This holds the pointer to the failure stack, when
3826 it is allocated relocatably. */
3827 fail_stack_elt_t *failure_stack_ptr;
3830 /* We fill all the registers internally, independent of what we
3831 return, for use in backreferences. The number here includes
3832 an element for register zero. */
3833 size_t num_regs = bufp->re_nsub + 1;
3835 /* The currently active registers. */
3836 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3837 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3839 /* Information on the contents of registers. These are pointers into
3840 the input strings; they record just what was matched (on this
3841 attempt) by a subexpression part of the pattern, that is, the
3842 regnum-th regstart pointer points to where in the pattern we began
3843 matching and the regnum-th regend points to right after where we
3844 stopped matching the regnum-th subexpression. (The zeroth register
3845 keeps track of what the whole pattern matches.) */
3846 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3847 const char **regstart, **regend;
3850 /* If a group that's operated upon by a repetition operator fails to
3851 match anything, then the register for its start will need to be
3852 restored because it will have been set to wherever in the string we
3853 are when we last see its open-group operator. Similarly for a
3855 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3856 const char **old_regstart, **old_regend;
3859 /* The is_active field of reg_info helps us keep track of which (possibly
3860 nested) subexpressions we are currently in. The matched_something
3861 field of reg_info[reg_num] helps us tell whether or not we have
3862 matched any of the pattern so far this time through the reg_num-th
3863 subexpression. These two fields get reset each time through any
3864 loop their register is in. */
3865 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3866 register_info_type *reg_info;
3869 /* The following record the register info as found in the above
3870 variables when we find a match better than any we've seen before.
3871 This happens as we backtrack through the failure points, which in
3872 turn happens only if we have not yet matched the entire string. */
3873 unsigned best_regs_set = false;
3874 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3875 const char **best_regstart, **best_regend;
3878 /* Logically, this is `best_regend[0]'. But we don't want to have to
3879 allocate space for that if we're not allocating space for anything
3880 else (see below). Also, we never need info about register 0 for
3881 any of the other register vectors, and it seems rather a kludge to
3882 treat `best_regend' differently than the rest. So we keep track of
3883 the end of the best match so far in a separate variable. We
3884 initialize this to NULL so that when we backtrack the first time
3885 and need to test it, it's not garbage. */
3886 const char *match_end = NULL;
3888 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3889 int set_regs_matched_done = 0;
3891 /* Used when we pop values we don't care about. */
3892 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3893 const char **reg_dummy;
3894 register_info_type *reg_info_dummy;
3898 /* Counts the total number of registers pushed. */
3899 unsigned num_regs_pushed = 0;
3902 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3906 #ifdef MATCH_MAY_ALLOCATE
3907 /* Do not bother to initialize all the register variables if there are
3908 no groups in the pattern, as it takes a fair amount of time. If
3909 there are groups, we include space for register 0 (the whole
3910 pattern), even though we never use it, since it simplifies the
3911 array indexing. We should fix this. */
3914 regstart = REGEX_TALLOC (num_regs, const char *);
3915 regend = REGEX_TALLOC (num_regs, const char *);
3916 old_regstart = REGEX_TALLOC (num_regs, const char *);
3917 old_regend = REGEX_TALLOC (num_regs, const char *);
3918 best_regstart = REGEX_TALLOC (num_regs, const char *);
3919 best_regend = REGEX_TALLOC (num_regs, const char *);
3920 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3921 reg_dummy = REGEX_TALLOC (num_regs, const char *);
3922 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3924 if (!(regstart && regend && old_regstart && old_regend && reg_info
3925 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3933 /* We must initialize all our variables to NULL, so that
3934 `FREE_VARIABLES' doesn't try to free them. */
3935 regstart = regend = old_regstart = old_regend = best_regstart
3936 = best_regend = reg_dummy = NULL;
3937 reg_info = reg_info_dummy = (register_info_type *) NULL;
3939 #endif /* MATCH_MAY_ALLOCATE */
3941 /* The starting position is bogus. */
3942 if (pos < 0 || pos > size1 + size2)
3948 /* Initialize subexpression text positions to -1 to mark ones that no
3949 start_memory/stop_memory has been seen for. Also initialize the
3950 register information struct. */
3951 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
3953 regstart[mcnt] = regend[mcnt]
3954 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3956 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3957 IS_ACTIVE (reg_info[mcnt]) = 0;
3958 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3959 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3962 /* We move `string1' into `string2' if the latter's empty -- but not if
3963 `string1' is null. */
3964 if (size2 == 0 && string1 != NULL)
3971 end1 = string1 + size1;
3972 end2 = string2 + size2;
3974 /* Compute where to stop matching, within the two strings. */
3977 end_match_1 = string1 + stop;
3978 end_match_2 = string2;
3983 end_match_2 = string2 + stop - size1;
3986 /* `p' scans through the pattern as `d' scans through the data.
3987 `dend' is the end of the input string that `d' points within. `d'
3988 is advanced into the following input string whenever necessary, but
3989 this happens before fetching; therefore, at the beginning of the
3990 loop, `d' can be pointing at the end of a string, but it cannot
3992 if (size1 > 0 && pos <= size1)
3999 d = string2 + pos - size1;
4003 DEBUG_PRINT1 ("The compiled pattern is:\n");
4004 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
4005 DEBUG_PRINT1 ("The string to match is: `");
4006 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
4007 DEBUG_PRINT1 ("'\n");
4009 /* This loops over pattern commands. It exits by returning from the
4010 function if the match is complete, or it drops through if the match
4011 fails at this starting point in the input data. */
4015 DEBUG_PRINT2 ("\n%p: ", p);
4017 DEBUG_PRINT2 ("\n0x%x: ", p);
4021 { /* End of pattern means we might have succeeded. */
4022 DEBUG_PRINT1 ("end of pattern ... ");
4024 /* If we haven't matched the entire string, and we want the
4025 longest match, try backtracking. */
4026 if (d != end_match_2)
4028 /* 1 if this match ends in the same string (string1 or string2)
4029 as the best previous match. */
4030 boolean same_str_p = (FIRST_STRING_P (match_end)
4031 == MATCHING_IN_FIRST_STRING);
4032 /* 1 if this match is the best seen so far. */
4033 boolean best_match_p;
4035 /* AIX compiler got confused when this was combined
4036 with the previous declaration. */
4038 best_match_p = d > match_end;
4040 best_match_p = !MATCHING_IN_FIRST_STRING;
4042 DEBUG_PRINT1 ("backtracking.\n");
4044 if (!FAIL_STACK_EMPTY ())
4045 { /* More failure points to try. */
4047 /* If exceeds best match so far, save it. */
4048 if (!best_regs_set || best_match_p)
4050 best_regs_set = true;
4053 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
4055 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4057 best_regstart[mcnt] = regstart[mcnt];
4058 best_regend[mcnt] = regend[mcnt];
4064 /* If no failure points, don't restore garbage. And if
4065 last match is real best match, don't restore second
4067 else if (best_regs_set && !best_match_p)
4070 /* Restore best match. It may happen that `dend ==
4071 end_match_1' while the restored d is in string2.
4072 For example, the pattern `x.*y.*z' against the
4073 strings `x-' and `y-z-', if the two strings are
4074 not consecutive in memory. */
4075 DEBUG_PRINT1 ("Restoring best registers.\n");
4078 dend = ((d >= string1 && d <= end1)
4079 ? end_match_1 : end_match_2);
4081 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4083 regstart[mcnt] = best_regstart[mcnt];
4084 regend[mcnt] = best_regend[mcnt];
4087 } /* d != end_match_2 */
4090 DEBUG_PRINT1 ("Accepting match.\n");
4092 /* If caller wants register contents data back, do it. */
4093 if (regs && !bufp->no_sub)
4095 /* Have the register data arrays been allocated? */
4096 if (bufp->regs_allocated == REGS_UNALLOCATED)
4097 { /* No. So allocate them with malloc. We need one
4098 extra element beyond `num_regs' for the `-1' marker
4100 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
4101 regs->start = TALLOC (regs->num_regs, regoff_t);
4102 regs->end = TALLOC (regs->num_regs, regoff_t);
4103 if (regs->start == NULL || regs->end == NULL)
4108 bufp->regs_allocated = REGS_REALLOCATE;
4110 else if (bufp->regs_allocated == REGS_REALLOCATE)
4111 { /* Yes. If we need more elements than were already
4112 allocated, reallocate them. If we need fewer, just
4114 if (regs->num_regs < num_regs + 1)
4116 regs->num_regs = num_regs + 1;
4117 RETALLOC (regs->start, regs->num_regs, regoff_t);
4118 RETALLOC (regs->end, regs->num_regs, regoff_t);
4119 if (regs->start == NULL || regs->end == NULL)
4128 /* These braces fend off a "empty body in an else-statement"
4129 warning under GCC when assert expands to nothing. */
4130 assert (bufp->regs_allocated == REGS_FIXED);
4133 /* Convert the pointer data in `regstart' and `regend' to
4134 indices. Register zero has to be set differently,
4135 since we haven't kept track of any info for it. */
4136 if (regs->num_regs > 0)
4138 regs->start[0] = pos;
4139 regs->end[0] = (MATCHING_IN_FIRST_STRING
4140 ? ((regoff_t) (d - string1))
4141 : ((regoff_t) (d - string2 + size1)));
4144 /* Go through the first `min (num_regs, regs->num_regs)'
4145 registers, since that is all we initialized. */
4146 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
4149 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
4150 regs->start[mcnt] = regs->end[mcnt] = -1;
4154 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
4156 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
4160 /* If the regs structure we return has more elements than
4161 were in the pattern, set the extra elements to -1. If
4162 we (re)allocated the registers, this is the case,
4163 because we always allocate enough to have at least one
4165 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
4166 regs->start[mcnt] = regs->end[mcnt] = -1;
4167 } /* regs && !bufp->no_sub */
4169 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
4170 nfailure_points_pushed, nfailure_points_popped,
4171 nfailure_points_pushed - nfailure_points_popped);
4172 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
4174 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
4178 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
4184 /* Otherwise match next pattern command. */
4185 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4187 /* Ignore these. Used to ignore the n of succeed_n's which
4188 currently have n == 0. */
4190 DEBUG_PRINT1 ("EXECUTING no_op.\n");
4194 DEBUG_PRINT1 ("EXECUTING succeed.\n");
4197 /* Match the next n pattern characters exactly. The following
4198 byte in the pattern defines n, and the n bytes after that
4199 are the characters to match. */
4202 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
4204 /* This is written out as an if-else so we don't waste time
4205 testing `translate' inside the loop. */
4211 if ((unsigned char) translate[(unsigned char) *d++]
4212 != (unsigned char) *p++)
4222 if (*d++ != (char) *p++) goto fail;
4226 SET_REGS_MATCHED ();
4230 /* Match any character except possibly a newline or a null. */
4232 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4236 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
4237 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4240 SET_REGS_MATCHED ();
4241 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4249 register unsigned char c;
4250 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4252 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4255 c = TRANSLATE (*d); /* The character to match. */
4257 /* Cast to `unsigned' instead of `unsigned char' in case the
4258 bit list is a full 32 bytes long. */
4259 if (c < (unsigned) (*p * BYTEWIDTH)
4260 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4265 if (!not) goto fail;
4267 SET_REGS_MATCHED ();
4273 /* The beginning of a group is represented by start_memory.
4274 The arguments are the register number in the next byte, and the
4275 number of groups inner to this one in the next. The text
4276 matched within the group is recorded (in the internal
4277 registers data structure) under the register number. */
4279 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4281 /* Find out if this group can match the empty string. */
4282 p1 = p; /* To send to group_match_null_string_p. */
4284 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4285 REG_MATCH_NULL_STRING_P (reg_info[*p])
4286 = group_match_null_string_p (&p1, pend, reg_info);
4288 /* Save the position in the string where we were the last time
4289 we were at this open-group operator in case the group is
4290 operated upon by a repetition operator, e.g., with `(a*)*b'
4291 against `ab'; then we want to ignore where we are now in
4292 the string in case this attempt to match fails. */
4293 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4294 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4296 DEBUG_PRINT2 (" old_regstart: %d\n",
4297 POINTER_TO_OFFSET (old_regstart[*p]));
4300 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4302 IS_ACTIVE (reg_info[*p]) = 1;
4303 MATCHED_SOMETHING (reg_info[*p]) = 0;
4305 /* Clear this whenever we change the register activity status. */
4306 set_regs_matched_done = 0;
4308 /* This is the new highest active register. */
4309 highest_active_reg = *p;
4311 /* If nothing was active before, this is the new lowest active
4313 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4314 lowest_active_reg = *p;
4316 /* Move past the register number and inner group count. */
4318 just_past_start_mem = p;
4323 /* The stop_memory opcode represents the end of a group. Its
4324 arguments are the same as start_memory's: the register
4325 number, and the number of inner groups. */
4327 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4329 /* We need to save the string position the last time we were at
4330 this close-group operator in case the group is operated
4331 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4332 against `aba'; then we want to ignore where we are now in
4333 the string in case this attempt to match fails. */
4334 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4335 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4337 DEBUG_PRINT2 (" old_regend: %d\n",
4338 POINTER_TO_OFFSET (old_regend[*p]));
4341 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4343 /* This register isn't active anymore. */
4344 IS_ACTIVE (reg_info[*p]) = 0;
4346 /* Clear this whenever we change the register activity status. */
4347 set_regs_matched_done = 0;
4349 /* If this was the only register active, nothing is active
4351 if (lowest_active_reg == highest_active_reg)
4353 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4354 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4357 { /* We must scan for the new highest active register, since
4358 it isn't necessarily one less than now: consider
4359 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4360 new highest active register is 1. */
4361 unsigned char r = *p - 1;
4362 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4365 /* If we end up at register zero, that means that we saved
4366 the registers as the result of an `on_failure_jump', not
4367 a `start_memory', and we jumped to past the innermost
4368 `stop_memory'. For example, in ((.)*) we save
4369 registers 1 and 2 as a result of the *, but when we pop
4370 back to the second ), we are at the stop_memory 1.
4371 Thus, nothing is active. */
4374 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4375 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4378 highest_active_reg = r;
4381 /* If just failed to match something this time around with a
4382 group that's operated on by a repetition operator, try to
4383 force exit from the ``loop'', and restore the register
4384 information for this group that we had before trying this
4386 if ((!MATCHED_SOMETHING (reg_info[*p])
4387 || just_past_start_mem == p - 1)
4390 boolean is_a_jump_n = false;
4394 switch ((re_opcode_t) *p1++)
4398 case pop_failure_jump:
4399 case maybe_pop_jump:
4401 case dummy_failure_jump:
4402 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4412 /* If the next operation is a jump backwards in the pattern
4413 to an on_failure_jump right before the start_memory
4414 corresponding to this stop_memory, exit from the loop
4415 by forcing a failure after pushing on the stack the
4416 on_failure_jump's jump in the pattern, and d. */
4417 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4418 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4420 /* If this group ever matched anything, then restore
4421 what its registers were before trying this last
4422 failed match, e.g., with `(a*)*b' against `ab' for
4423 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4424 against `aba' for regend[3].
4426 Also restore the registers for inner groups for,
4427 e.g., `((a*)(b*))*' against `aba' (register 3 would
4428 otherwise get trashed). */
4430 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4434 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4436 /* Restore this and inner groups' (if any) registers. */
4437 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
4440 regstart[r] = old_regstart[r];
4442 /* xx why this test? */
4443 if (old_regend[r] >= regstart[r])
4444 regend[r] = old_regend[r];
4448 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4449 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4455 /* Move past the register number and the inner group count. */
4460 /* \<digit> has been turned into a `duplicate' command which is
4461 followed by the numeric value of <digit> as the register number. */
4464 register const char *d2, *dend2;
4465 int regno = *p++; /* Get which register to match against. */
4466 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4468 /* Can't back reference a group which we've never matched. */
4469 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4472 /* Where in input to try to start matching. */
4473 d2 = regstart[regno];
4475 /* Where to stop matching; if both the place to start and
4476 the place to stop matching are in the same string, then
4477 set to the place to stop, otherwise, for now have to use
4478 the end of the first string. */
4480 dend2 = ((FIRST_STRING_P (regstart[regno])
4481 == FIRST_STRING_P (regend[regno]))
4482 ? regend[regno] : end_match_1);
4485 /* If necessary, advance to next segment in register
4489 if (dend2 == end_match_2) break;
4490 if (dend2 == regend[regno]) break;
4492 /* End of string1 => advance to string2. */
4494 dend2 = regend[regno];
4496 /* At end of register contents => success */
4497 if (d2 == dend2) break;
4499 /* If necessary, advance to next segment in data. */
4502 /* How many characters left in this segment to match. */
4505 /* Want how many consecutive characters we can match in
4506 one shot, so, if necessary, adjust the count. */
4507 if (mcnt > dend2 - d2)
4510 /* Compare that many; failure if mismatch, else move
4513 ? bcmp_translate (d, d2, mcnt, translate)
4514 : memcmp (d, d2, mcnt))
4516 d += mcnt, d2 += mcnt;
4518 /* Do this because we've match some characters. */
4519 SET_REGS_MATCHED ();
4525 /* begline matches the empty string at the beginning of the string
4526 (unless `not_bol' is set in `bufp'), and, if
4527 `newline_anchor' is set, after newlines. */
4529 DEBUG_PRINT1 ("EXECUTING begline.\n");
4531 if (AT_STRINGS_BEG (d))
4533 if (!bufp->not_bol) break;
4535 else if (d[-1] == '\n' && bufp->newline_anchor)
4539 /* In all other cases, we fail. */
4543 /* endline is the dual of begline. */
4545 DEBUG_PRINT1 ("EXECUTING endline.\n");
4547 if (AT_STRINGS_END (d))
4549 if (!bufp->not_eol) break;
4552 /* We have to ``prefetch'' the next character. */
4553 else if ((d == end1 ? *string2 : *d) == '\n'
4554 && bufp->newline_anchor)
4561 /* Match at the very beginning of the data. */
4563 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4564 if (AT_STRINGS_BEG (d))
4569 /* Match at the very end of the data. */
4571 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4572 if (AT_STRINGS_END (d))
4577 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4578 pushes NULL as the value for the string on the stack. Then
4579 `pop_failure_point' will keep the current value for the
4580 string, instead of restoring it. To see why, consider
4581 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4582 then the . fails against the \n. But the next thing we want
4583 to do is match the \n against the \n; if we restored the
4584 string value, we would be back at the foo.
4586 Because this is used only in specific cases, we don't need to
4587 check all the things that `on_failure_jump' does, to make
4588 sure the right things get saved on the stack. Hence we don't
4589 share its code. The only reason to push anything on the
4590 stack at all is that otherwise we would have to change
4591 `anychar's code to do something besides goto fail in this
4592 case; that seems worse than this. */
4593 case on_failure_keep_string_jump:
4594 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4596 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4598 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
4600 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4603 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4607 /* Uses of on_failure_jump:
4609 Each alternative starts with an on_failure_jump that points
4610 to the beginning of the next alternative. Each alternative
4611 except the last ends with a jump that in effect jumps past
4612 the rest of the alternatives. (They really jump to the
4613 ending jump of the following alternative, because tensioning
4614 these jumps is a hassle.)
4616 Repeats start with an on_failure_jump that points past both
4617 the repetition text and either the following jump or
4618 pop_failure_jump back to this on_failure_jump. */
4619 case on_failure_jump:
4621 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4623 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4625 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
4627 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4630 /* If this on_failure_jump comes right before a group (i.e.,
4631 the original * applied to a group), save the information
4632 for that group and all inner ones, so that if we fail back
4633 to this point, the group's information will be correct.
4634 For example, in \(a*\)*\1, we need the preceding group,
4635 and in \(zz\(a*\)b*\)\2, we need the inner group. */
4637 /* We can't use `p' to check ahead because we push
4638 a failure point to `p + mcnt' after we do this. */
4641 /* We need to skip no_op's before we look for the
4642 start_memory in case this on_failure_jump is happening as
4643 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4645 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4648 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4650 /* We have a new highest active register now. This will
4651 get reset at the start_memory we are about to get to,
4652 but we will have saved all the registers relevant to
4653 this repetition op, as described above. */
4654 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4655 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4656 lowest_active_reg = *(p1 + 1);
4659 DEBUG_PRINT1 (":\n");
4660 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4664 /* A smart repeat ends with `maybe_pop_jump'.
4665 We change it to either `pop_failure_jump' or `jump'. */
4666 case maybe_pop_jump:
4667 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4668 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4670 register unsigned char *p2 = p;
4672 /* Compare the beginning of the repeat with what in the
4673 pattern follows its end. If we can establish that there
4674 is nothing that they would both match, i.e., that we
4675 would have to backtrack because of (as in, e.g., `a*a')
4676 then we can change to pop_failure_jump, because we'll
4677 never have to backtrack.
4679 This is not true in the case of alternatives: in
4680 `(a|ab)*' we do need to backtrack to the `ab' alternative
4681 (e.g., if the string was `ab'). But instead of trying to
4682 detect that here, the alternative has put on a dummy
4683 failure point which is what we will end up popping. */
4685 /* Skip over open/close-group commands.
4686 If what follows this loop is a ...+ construct,
4687 look at what begins its body, since we will have to
4688 match at least one of that. */
4692 && ((re_opcode_t) *p2 == stop_memory
4693 || (re_opcode_t) *p2 == start_memory))
4695 else if (p2 + 6 < pend
4696 && (re_opcode_t) *p2 == dummy_failure_jump)
4703 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4704 to the `maybe_finalize_jump' of this case. Examine what
4707 /* If we're at the end of the pattern, we can change. */
4710 /* Consider what happens when matching ":\(.*\)"
4711 against ":/". I don't really understand this code
4713 p[-3] = (unsigned char) pop_failure_jump;
4715 (" End of pattern: change to `pop_failure_jump'.\n");
4718 else if ((re_opcode_t) *p2 == exactn
4719 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4721 register unsigned char c
4722 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4724 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4726 p[-3] = (unsigned char) pop_failure_jump;
4727 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4731 else if ((re_opcode_t) p1[3] == charset
4732 || (re_opcode_t) p1[3] == charset_not)
4734 int not = (re_opcode_t) p1[3] == charset_not;
4736 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4737 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4740 /* `not' is equal to 1 if c would match, which means
4741 that we can't change to pop_failure_jump. */
4744 p[-3] = (unsigned char) pop_failure_jump;
4745 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4749 else if ((re_opcode_t) *p2 == charset)
4752 register unsigned char c
4753 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4757 if ((re_opcode_t) p1[3] == exactn
4758 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
4759 && (p2[2 + p1[5] / BYTEWIDTH]
4760 & (1 << (p1[5] % BYTEWIDTH)))))
4762 if ((re_opcode_t) p1[3] == exactn
4763 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[4]
4764 && (p2[2 + p1[4] / BYTEWIDTH]
4765 & (1 << (p1[4] % BYTEWIDTH)))))
4768 p[-3] = (unsigned char) pop_failure_jump;
4769 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4773 else if ((re_opcode_t) p1[3] == charset_not)
4776 /* We win if the charset_not inside the loop
4777 lists every character listed in the charset after. */
4778 for (idx = 0; idx < (int) p2[1]; idx++)
4779 if (! (p2[2 + idx] == 0
4780 || (idx < (int) p1[4]
4781 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4786 p[-3] = (unsigned char) pop_failure_jump;
4787 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4790 else if ((re_opcode_t) p1[3] == charset)
4793 /* We win if the charset inside the loop
4794 has no overlap with the one after the loop. */
4796 idx < (int) p2[1] && idx < (int) p1[4];
4798 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4801 if (idx == p2[1] || idx == p1[4])
4803 p[-3] = (unsigned char) pop_failure_jump;
4804 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4809 p -= 2; /* Point at relative address again. */
4810 if ((re_opcode_t) p[-1] != pop_failure_jump)
4812 p[-1] = (unsigned char) jump;
4813 DEBUG_PRINT1 (" Match => jump.\n");
4814 goto unconditional_jump;
4816 /* Note fall through. */
4819 /* The end of a simple repeat has a pop_failure_jump back to
4820 its matching on_failure_jump, where the latter will push a
4821 failure point. The pop_failure_jump takes off failure
4822 points put on by this pop_failure_jump's matching
4823 on_failure_jump; we got through the pattern to here from the
4824 matching on_failure_jump, so didn't fail. */
4825 case pop_failure_jump:
4827 /* We need to pass separate storage for the lowest and
4828 highest registers, even though we don't care about the
4829 actual values. Otherwise, we will restore only one
4830 register from the stack, since lowest will == highest in
4831 `pop_failure_point'. */
4832 active_reg_t dummy_low_reg, dummy_high_reg;
4833 unsigned char *pdummy;
4836 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4837 POP_FAILURE_POINT (sdummy, pdummy,
4838 dummy_low_reg, dummy_high_reg,
4839 reg_dummy, reg_dummy, reg_info_dummy);
4841 /* Note fall through. */
4845 DEBUG_PRINT2 ("\n%p: ", p);
4847 DEBUG_PRINT2 ("\n0x%x: ", p);
4849 /* Note fall through. */
4851 /* Unconditionally jump (without popping any failure points). */
4853 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4854 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4855 p += mcnt; /* Do the jump. */
4857 DEBUG_PRINT2 ("(to %p).\n", p);
4859 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4864 /* We need this opcode so we can detect where alternatives end
4865 in `group_match_null_string_p' et al. */
4867 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4868 goto unconditional_jump;
4871 /* Normally, the on_failure_jump pushes a failure point, which
4872 then gets popped at pop_failure_jump. We will end up at
4873 pop_failure_jump, also, and with a pattern of, say, `a+', we
4874 are skipping over the on_failure_jump, so we have to push
4875 something meaningless for pop_failure_jump to pop. */
4876 case dummy_failure_jump:
4877 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4878 /* It doesn't matter what we push for the string here. What
4879 the code at `fail' tests is the value for the pattern. */
4880 PUSH_FAILURE_POINT (NULL, NULL, -2);
4881 goto unconditional_jump;
4884 /* At the end of an alternative, we need to push a dummy failure
4885 point in case we are followed by a `pop_failure_jump', because
4886 we don't want the failure point for the alternative to be
4887 popped. For example, matching `(a|ab)*' against `aab'
4888 requires that we match the `ab' alternative. */
4889 case push_dummy_failure:
4890 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4891 /* See comments just above at `dummy_failure_jump' about the
4893 PUSH_FAILURE_POINT (NULL, NULL, -2);
4896 /* Have to succeed matching what follows at least n times.
4897 After that, handle like `on_failure_jump'. */
4899 EXTRACT_NUMBER (mcnt, p + 2);
4900 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4903 /* Originally, this is how many times we HAVE to succeed. */
4908 STORE_NUMBER_AND_INCR (p, mcnt);
4910 DEBUG_PRINT3 (" Setting %p to %d.\n", p - 2, mcnt);
4912 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - 2, mcnt);
4918 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n", p+2);
4920 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4922 p[2] = (unsigned char) no_op;
4923 p[3] = (unsigned char) no_op;
4929 EXTRACT_NUMBER (mcnt, p + 2);
4930 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4932 /* Originally, this is how many times we CAN jump. */
4936 STORE_NUMBER (p + 2, mcnt);
4938 DEBUG_PRINT3 (" Setting %p to %d.\n", p + 2, mcnt);
4940 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + 2, mcnt);
4942 goto unconditional_jump;
4944 /* If don't have to jump any more, skip over the rest of command. */
4951 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4953 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4955 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4957 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
4959 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4961 STORE_NUMBER (p1, mcnt);
4966 /* The DEC Alpha C compiler 3.x generates incorrect code for the
4967 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4968 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4969 macro and introducing temporary variables works around the bug. */
4972 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4973 if (AT_WORD_BOUNDARY (d))
4978 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4979 if (AT_WORD_BOUNDARY (d))
4985 boolean prevchar, thischar;
4987 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4988 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4991 prevchar = WORDCHAR_P (d - 1);
4992 thischar = WORDCHAR_P (d);
4993 if (prevchar != thischar)
5000 boolean prevchar, thischar;
5002 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
5003 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5006 prevchar = WORDCHAR_P (d - 1);
5007 thischar = WORDCHAR_P (d);
5008 if (prevchar != thischar)
5015 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
5016 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
5021 DEBUG_PRINT1 ("EXECUTING wordend.\n");
5022 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
5023 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
5029 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
5030 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
5035 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
5036 if (PTR_CHAR_POS ((unsigned char *) d) != point)
5041 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
5042 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
5047 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
5052 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
5056 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5058 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
5060 SET_REGS_MATCHED ();
5064 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
5066 goto matchnotsyntax;
5069 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
5073 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5075 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
5077 SET_REGS_MATCHED ();
5080 #else /* not emacs */
5082 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
5084 if (!WORDCHAR_P (d))
5086 SET_REGS_MATCHED ();
5091 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
5095 SET_REGS_MATCHED ();
5098 #endif /* not emacs */
5103 continue; /* Successfully executed one pattern command; keep going. */
5106 /* We goto here if a matching operation fails. */
5108 if (!FAIL_STACK_EMPTY ())
5109 { /* A restart point is known. Restore to that state. */
5110 DEBUG_PRINT1 ("\nFAIL:\n");
5111 POP_FAILURE_POINT (d, p,
5112 lowest_active_reg, highest_active_reg,
5113 regstart, regend, reg_info);
5115 /* If this failure point is a dummy, try the next one. */
5119 /* If we failed to the end of the pattern, don't examine *p. */
5123 boolean is_a_jump_n = false;
5125 /* If failed to a backwards jump that's part of a repetition
5126 loop, need to pop this failure point and use the next one. */
5127 switch ((re_opcode_t) *p)
5131 case maybe_pop_jump:
5132 case pop_failure_jump:
5135 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5138 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
5140 && (re_opcode_t) *p1 == on_failure_jump))
5148 if (d >= string1 && d <= end1)
5152 break; /* Matching at this starting point really fails. */
5156 goto restore_best_regs;
5160 return -1; /* Failure to match. */
5163 /* Subroutine definitions for re_match_2. */
5166 /* We are passed P pointing to a register number after a start_memory.
5168 Return true if the pattern up to the corresponding stop_memory can
5169 match the empty string, and false otherwise.
5171 If we find the matching stop_memory, sets P to point to one past its number.
5172 Otherwise, sets P to an undefined byte less than or equal to END.
5174 We don't handle duplicates properly (yet). */
5177 group_match_null_string_p (p, end, reg_info)
5178 unsigned char **p, *end;
5179 register_info_type *reg_info;
5182 /* Point to after the args to the start_memory. */
5183 unsigned char *p1 = *p + 2;
5187 /* Skip over opcodes that can match nothing, and return true or
5188 false, as appropriate, when we get to one that can't, or to the
5189 matching stop_memory. */
5191 switch ((re_opcode_t) *p1)
5193 /* Could be either a loop or a series of alternatives. */
5194 case on_failure_jump:
5196 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5198 /* If the next operation is not a jump backwards in the
5203 /* Go through the on_failure_jumps of the alternatives,
5204 seeing if any of the alternatives cannot match nothing.
5205 The last alternative starts with only a jump,
5206 whereas the rest start with on_failure_jump and end
5207 with a jump, e.g., here is the pattern for `a|b|c':
5209 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
5210 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
5213 So, we have to first go through the first (n-1)
5214 alternatives and then deal with the last one separately. */
5217 /* Deal with the first (n-1) alternatives, which start
5218 with an on_failure_jump (see above) that jumps to right
5219 past a jump_past_alt. */
5221 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
5223 /* `mcnt' holds how many bytes long the alternative
5224 is, including the ending `jump_past_alt' and
5227 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
5231 /* Move to right after this alternative, including the
5235 /* Break if it's the beginning of an n-th alternative
5236 that doesn't begin with an on_failure_jump. */
5237 if ((re_opcode_t) *p1 != on_failure_jump)
5240 /* Still have to check that it's not an n-th
5241 alternative that starts with an on_failure_jump. */
5243 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5244 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
5246 /* Get to the beginning of the n-th alternative. */
5252 /* Deal with the last alternative: go back and get number
5253 of the `jump_past_alt' just before it. `mcnt' contains
5254 the length of the alternative. */
5255 EXTRACT_NUMBER (mcnt, p1 - 2);
5257 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
5260 p1 += mcnt; /* Get past the n-th alternative. */
5266 assert (p1[1] == **p);
5272 if (!common_op_match_null_string_p (&p1, end, reg_info))
5275 } /* while p1 < end */
5278 } /* group_match_null_string_p */
5281 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5282 It expects P to be the first byte of a single alternative and END one
5283 byte past the last. The alternative can contain groups. */
5286 alt_match_null_string_p (p, end, reg_info)
5287 unsigned char *p, *end;
5288 register_info_type *reg_info;
5291 unsigned char *p1 = p;
5295 /* Skip over opcodes that can match nothing, and break when we get
5296 to one that can't. */
5298 switch ((re_opcode_t) *p1)
5301 case on_failure_jump:
5303 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5308 if (!common_op_match_null_string_p (&p1, end, reg_info))
5311 } /* while p1 < end */
5314 } /* alt_match_null_string_p */
5317 /* Deals with the ops common to group_match_null_string_p and
5318 alt_match_null_string_p.
5320 Sets P to one after the op and its arguments, if any. */
5323 common_op_match_null_string_p (p, end, reg_info)
5324 unsigned char **p, *end;
5325 register_info_type *reg_info;
5330 unsigned char *p1 = *p;
5332 switch ((re_opcode_t) *p1++)
5352 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5353 ret = group_match_null_string_p (&p1, end, reg_info);
5355 /* Have to set this here in case we're checking a group which
5356 contains a group and a back reference to it. */
5358 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5359 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5365 /* If this is an optimized succeed_n for zero times, make the jump. */
5367 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5375 /* Get to the number of times to succeed. */
5377 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5382 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5390 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5398 /* All other opcodes mean we cannot match the empty string. */
5404 } /* common_op_match_null_string_p */
5407 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5408 bytes; nonzero otherwise. */
5411 bcmp_translate (s1, s2, len, translate)
5412 const char *s1, *s2;
5414 RE_TRANSLATE_TYPE translate;
5416 register const unsigned char *p1 = (const unsigned char *) s1;
5417 register const unsigned char *p2 = (const unsigned char *) s2;
5420 if (translate[*p1++] != translate[*p2++]) return 1;
5426 /* Entry points for GNU code. */
5428 /* re_compile_pattern is the GNU regular expression compiler: it
5429 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5430 Returns 0 if the pattern was valid, otherwise an error string.
5432 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5433 are set in BUFP on entry.
5435 We call regex_compile to do the actual compilation. */
5438 re_compile_pattern (pattern, length, bufp)
5439 const char *pattern;
5441 struct re_pattern_buffer *bufp;
5445 /* GNU code is written to assume at least RE_NREGS registers will be set
5446 (and at least one extra will be -1). */
5447 bufp->regs_allocated = REGS_UNALLOCATED;
5449 /* And GNU code determines whether or not to get register information
5450 by passing null for the REGS argument to re_match, etc., not by
5454 /* Match anchors at newline. */
5455 bufp->newline_anchor = 1;
5457 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5461 return gettext (re_error_msgid[(int) ret]);
5464 weak_alias (__re_compile_pattern, re_compile_pattern)
5467 /* Entry points compatible with 4.2 BSD regex library. We don't define
5468 them unless specifically requested. */
5470 #if defined _REGEX_RE_COMP || defined _LIBC
5472 /* BSD has one and only one pattern buffer. */
5473 static struct re_pattern_buffer re_comp_buf;
5477 /* Make these definitions weak in libc, so POSIX programs can redefine
5478 these names if they don't use our functions, and still use
5479 regcomp/regexec below without link errors. */
5489 if (!re_comp_buf.buffer)
5490 return gettext ("No previous regular expression");
5494 if (!re_comp_buf.buffer)
5496 re_comp_buf.buffer = (unsigned char *) malloc (200);
5497 if (re_comp_buf.buffer == NULL)
5498 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
5499 re_comp_buf.allocated = 200;
5501 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5502 if (re_comp_buf.fastmap == NULL)
5503 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
5506 /* Since `re_exec' always passes NULL for the `regs' argument, we
5507 don't need to initialize the pattern buffer fields which affect it. */
5509 /* Match anchors at newlines. */
5510 re_comp_buf.newline_anchor = 1;
5512 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5517 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5518 return (char *) gettext (re_error_msgid[(int) ret]);
5529 const int len = strlen (s);
5531 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5534 #endif /* _REGEX_RE_COMP */
5536 /* POSIX.2 functions. Don't define these for Emacs. */
5540 /* regcomp takes a regular expression as a string and compiles it.
5542 PREG is a regex_t *. We do not expect any fields to be initialized,
5543 since POSIX says we shouldn't. Thus, we set
5545 `buffer' to the compiled pattern;
5546 `used' to the length of the compiled pattern;
5547 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5548 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5549 RE_SYNTAX_POSIX_BASIC;
5550 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5551 `fastmap' and `fastmap_accurate' to zero;
5552 `re_nsub' to the number of subexpressions in PATTERN.
5554 PATTERN is the address of the pattern string.
5556 CFLAGS is a series of bits which affect compilation.
5558 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5559 use POSIX basic syntax.
5561 If REG_NEWLINE is set, then . and [^...] don't match newline.
5562 Also, regexec will try a match beginning after every newline.
5564 If REG_ICASE is set, then we considers upper- and lowercase
5565 versions of letters to be equivalent when matching.
5567 If REG_NOSUB is set, then when PREG is passed to regexec, that
5568 routine will report only success or failure, and nothing about the
5571 It returns 0 if it succeeds, nonzero if it doesn't. (See gnu-regex.h for
5572 the return codes and their meanings.) */
5575 regcomp (preg, pattern, cflags)
5577 const char *pattern;
5582 = (cflags & REG_EXTENDED) ?
5583 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5585 /* regex_compile will allocate the space for the compiled pattern. */
5587 preg->allocated = 0;
5590 /* Don't bother to use a fastmap when searching. This simplifies the
5591 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
5592 characters after newlines into the fastmap. This way, we just try
5596 if (cflags & REG_ICASE)
5601 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
5602 * sizeof (*(RE_TRANSLATE_TYPE)0));
5603 if (preg->translate == NULL)
5604 return (int) REG_ESPACE;
5606 /* Map uppercase characters to corresponding lowercase ones. */
5607 for (i = 0; i < CHAR_SET_SIZE; i++)
5608 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
5611 preg->translate = NULL;
5613 /* If REG_NEWLINE is set, newlines are treated differently. */
5614 if (cflags & REG_NEWLINE)
5615 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5616 syntax &= ~RE_DOT_NEWLINE;
5617 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5618 /* It also changes the matching behavior. */
5619 preg->newline_anchor = 1;
5622 preg->newline_anchor = 0;
5624 preg->no_sub = !!(cflags & REG_NOSUB);
5626 /* POSIX says a null character in the pattern terminates it, so we
5627 can use strlen here in compiling the pattern. */
5628 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5630 /* POSIX doesn't distinguish between an unmatched open-group and an
5631 unmatched close-group: both are REG_EPAREN. */
5632 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5637 weak_alias (__regcomp, regcomp)
5641 /* regexec searches for a given pattern, specified by PREG, in the
5644 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5645 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5646 least NMATCH elements, and we set them to the offsets of the
5647 corresponding matched substrings.
5649 EFLAGS specifies `execution flags' which affect matching: if
5650 REG_NOTBOL is set, then ^ does not match at the beginning of the
5651 string; if REG_NOTEOL is set, then $ does not match at the end.
5653 We return 0 if we find a match and REG_NOMATCH if not. */
5656 regexec (preg, string, nmatch, pmatch, eflags)
5657 const regex_t *preg;
5660 regmatch_t pmatch[];
5664 struct re_registers regs;
5665 regex_t private_preg;
5666 int len = strlen (string);
5667 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5669 private_preg = *preg;
5671 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5672 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5674 /* The user has told us exactly how many registers to return
5675 information about, via `nmatch'. We have to pass that on to the
5676 matching routines. */
5677 private_preg.regs_allocated = REGS_FIXED;
5681 regs.num_regs = nmatch;
5682 regs.start = TALLOC (nmatch, regoff_t);
5683 regs.end = TALLOC (nmatch, regoff_t);
5684 if (regs.start == NULL || regs.end == NULL)
5685 return (int) REG_NOMATCH;
5688 /* Perform the searching operation. */
5689 ret = re_search (&private_preg, string, len,
5690 /* start: */ 0, /* range: */ len,
5691 want_reg_info ? ®s : (struct re_registers *) 0);
5693 /* Copy the register information to the POSIX structure. */
5700 for (r = 0; r < nmatch; r++)
5702 pmatch[r].rm_so = regs.start[r];
5703 pmatch[r].rm_eo = regs.end[r];
5707 /* If we needed the temporary register info, free the space now. */
5712 /* We want zero return to mean success, unlike `re_search'. */
5713 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5716 weak_alias (__regexec, regexec)
5720 /* Returns a message corresponding to an error code, ERRCODE, returned
5721 from either regcomp or regexec. We don't use PREG here. */
5724 __regerror (errcode, preg, errbuf, errbuf_size)
5726 const regex_t *preg;
5734 || errcode >= (int) (sizeof (re_error_msgid)
5735 / sizeof (re_error_msgid[0])))
5736 /* Only error codes returned by the rest of the code should be passed
5737 to this routine. If we are given anything else, or if other regex
5738 code generates an invalid error code, then the program has a bug.
5739 Dump core so we can fix it. */
5742 msg = gettext (re_error_msgid[errcode]);
5744 msg_size = strlen (msg) + 1; /* Includes the null. */
5746 if (errbuf_size != 0)
5748 if (msg_size > errbuf_size)
5750 #if defined HAVE_MEMPCPY || defined _LIBC
5751 *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
5753 memcpy (errbuf, msg, errbuf_size - 1);
5754 errbuf[errbuf_size - 1] = 0;
5758 memcpy (errbuf, msg, msg_size);
5764 weak_alias (__regerror, regerror)
5768 /* Free dynamically allocated space used by PREG. */
5774 if (preg->buffer != NULL)
5775 free (preg->buffer);
5776 preg->buffer = NULL;
5778 preg->allocated = 0;
5781 if (preg->fastmap != NULL)
5782 free (preg->fastmap);
5783 preg->fastmap = NULL;
5784 preg->fastmap_accurate = 0;
5786 if (preg->translate != NULL)
5787 free (preg->translate);
5788 preg->translate = NULL;
5791 weak_alias (__regfree, regfree)
5794 #endif /* not emacs */
projects / platform / upstream / binutils.git / blob