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 GNU C Library.
8 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
22 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
25 /* AIX requires this to be the first thing in the file. */
26 #if defined _AIX && !defined REGEX_MALLOC
38 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
39 # define PARAMS(args) args
41 # define PARAMS(args) ()
43 #endif /* Not PARAMS. */
45 #if defined STDC_HEADERS && !defined emacs
48 /* We need this for `regex.h', and perhaps for the Emacs include files. */
49 # include <sys/types.h>
52 /* For platform which support the ISO C amendement 1 functionality we
53 support user defined character classes. */
54 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_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
110 # if !defined bzero && !defined _LIBC
111 # define bzero(s, n) (memset (s, '\0', n), (s))
114 # include <strings.h>
116 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
119 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
124 /* Define the syntax stuff for \<, \>, etc. */
126 /* This must be nonzero for the wordchar and notwordchar pattern
127 commands in re_match_2. */
132 # ifdef SWITCH_ENUM_BUG
133 # define SWITCH_ENUM_CAST(x) ((int)(x))
135 # define SWITCH_ENUM_CAST(x) (x)
138 /* How many characters in the character set. */
139 # define CHAR_SET_SIZE 256
143 extern char *re_syntax_table;
145 # else /* not SYNTAX_TABLE */
147 static char re_syntax_table[CHAR_SET_SIZE];
158 bzero (re_syntax_table, sizeof re_syntax_table);
160 for (c = 'a'; c <= 'z'; c++)
161 re_syntax_table[c] = Sword;
163 for (c = 'A'; c <= 'Z'; c++)
164 re_syntax_table[c] = Sword;
166 for (c = '0'; c <= '9'; c++)
167 re_syntax_table[c] = Sword;
169 re_syntax_table['_'] = Sword;
174 # endif /* not SYNTAX_TABLE */
176 # define SYNTAX(c) re_syntax_table[c]
178 #endif /* not emacs */
180 /* Get the interface, including the syntax bits. */
183 /* isalpha etc. are used for the character classes. */
186 /* Jim Meyering writes:
188 "... Some ctype macros are valid only for character codes that
189 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
190 using /bin/cc or gcc but without giving an ansi option). So, all
191 ctype uses should be through macros like ISPRINT... If
192 STDC_HEADERS is defined, then autoconf has verified that the ctype
193 macros don't need to be guarded with references to isascii. ...
194 Defining isascii to 1 should let any compiler worth its salt
195 eliminate the && through constant folding." */
197 #if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
198 # define ISASCII(c) 1
200 # define ISASCII(c) isascii(c)
204 # define ISBLANK(c) (ISASCII (c) && isblank (c))
206 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
209 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
211 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
214 #define ISPRINT(c) (ISASCII (c) && isprint (c))
215 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
216 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
217 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
218 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
219 #define ISLOWER(c) (ISASCII (c) && islower (c))
220 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
221 #define ISSPACE(c) (ISASCII (c) && isspace (c))
222 #define ISUPPER(c) (ISASCII (c) && isupper (c))
223 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
226 # define NULL (void *)0
229 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
230 since ours (we hope) works properly with all combinations of
231 machines, compilers, `char' and `unsigned char' argument types.
232 (Per Bothner suggested the basic approach.) */
233 #undef SIGN_EXTEND_CHAR
235 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
236 #else /* not __STDC__ */
237 /* As in Harbison and Steele. */
238 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
241 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
242 use `alloca' instead of `malloc'. This is because using malloc in
243 re_search* or re_match* could cause memory leaks when C-g is used in
244 Emacs; also, malloc is slower and causes storage fragmentation. On
245 the other hand, malloc is more portable, and easier to debug.
247 Because we sometimes use alloca, some routines have to be macros,
248 not functions -- `alloca'-allocated space disappears at the end of the
249 function it is called in. */
253 # define REGEX_ALLOCATE malloc
254 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
255 # define REGEX_FREE free
257 #else /* not REGEX_MALLOC */
259 /* Emacs already defines alloca, sometimes. */
262 /* Make alloca work the best possible way. */
264 # define alloca __builtin_alloca
265 # else /* not __GNUC__ */
268 # endif /* HAVE_ALLOCA_H */
269 # endif /* not __GNUC__ */
271 # endif /* not alloca */
273 # define REGEX_ALLOCATE alloca
275 /* Assumes a `char *destination' variable. */
276 # define REGEX_REALLOCATE(source, osize, nsize) \
277 (destination = (char *) alloca (nsize), \
278 memcpy (destination, source, osize))
280 /* No need to do anything to free, after alloca. */
281 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
283 #endif /* not REGEX_MALLOC */
285 /* Define how to allocate the failure stack. */
287 #if defined REL_ALLOC && defined REGEX_MALLOC
289 # define REGEX_ALLOCATE_STACK(size) \
290 r_alloc (&failure_stack_ptr, (size))
291 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
292 r_re_alloc (&failure_stack_ptr, (nsize))
293 # define REGEX_FREE_STACK(ptr) \
294 r_alloc_free (&failure_stack_ptr)
296 #else /* not using relocating allocator */
300 # define REGEX_ALLOCATE_STACK malloc
301 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
302 # define REGEX_FREE_STACK free
304 # else /* not REGEX_MALLOC */
306 # define REGEX_ALLOCATE_STACK alloca
308 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
309 REGEX_REALLOCATE (source, osize, nsize)
310 /* No need to explicitly free anything. */
311 # define REGEX_FREE_STACK(arg)
313 # endif /* not REGEX_MALLOC */
314 #endif /* not using relocating allocator */
317 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
318 `string1' or just past its end. This works if PTR is NULL, which is
320 #define FIRST_STRING_P(ptr) \
321 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
323 /* (Re)Allocate N items of type T using malloc, or fail. */
324 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
325 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
326 #define RETALLOC_IF(addr, n, t) \
327 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
328 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
330 #define BYTEWIDTH 8 /* In bits. */
332 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
336 #define MAX(a, b) ((a) > (b) ? (a) : (b))
337 #define MIN(a, b) ((a) < (b) ? (a) : (b))
339 typedef char boolean;
343 static int re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
344 const char *string1, int size1,
345 const char *string2, int size2,
347 struct re_registers *regs,
350 /* These are the command codes that appear in compiled regular
351 expressions. Some opcodes are followed by argument bytes. A
352 command code can specify any interpretation whatsoever for its
353 arguments. Zero bytes may appear in the compiled regular expression. */
359 /* Succeed right away--no more backtracking. */
362 /* Followed by one byte giving n, then by n literal bytes. */
365 /* Matches any (more or less) character. */
368 /* Matches any one char belonging to specified set. First
369 following byte is number of bitmap bytes. Then come bytes
370 for a bitmap saying which chars are in. Bits in each byte
371 are ordered low-bit-first. A character is in the set if its
372 bit is 1. A character too large to have a bit in the map is
373 automatically not in the set. */
376 /* Same parameters as charset, but match any character that is
377 not one of those specified. */
380 /* Start remembering the text that is matched, for storing in a
381 register. Followed by one byte with the register number, in
382 the range 0 to one less than the pattern buffer's re_nsub
383 field. Then followed by one byte with the number of groups
384 inner to this one. (This last has to be part of the
385 start_memory only because we need it in the on_failure_jump
389 /* Stop remembering the text that is matched and store it in a
390 memory register. Followed by one byte with the register
391 number, in the range 0 to one less than `re_nsub' in the
392 pattern buffer, and one byte with the number of inner groups,
393 just like `start_memory'. (We need the number of inner
394 groups here because we don't have any easy way of finding the
395 corresponding start_memory when we're at a stop_memory.) */
398 /* Match a duplicate of something remembered. Followed by one
399 byte containing the register number. */
402 /* Fail unless at beginning of line. */
405 /* Fail unless at end of line. */
408 /* Succeeds if at beginning of buffer (if emacs) or at beginning
409 of string to be matched (if not). */
412 /* Analogously, for end of buffer/string. */
415 /* Followed by two byte relative address to which to jump. */
418 /* Same as jump, but marks the end of an alternative. */
421 /* Followed by two-byte relative address of place to resume at
422 in case of failure. */
425 /* Like on_failure_jump, but pushes a placeholder instead of the
426 current string position when executed. */
427 on_failure_keep_string_jump,
429 /* Throw away latest failure point and then jump to following
430 two-byte relative address. */
433 /* Change to pop_failure_jump if know won't have to backtrack to
434 match; otherwise change to jump. This is used to jump
435 back to the beginning of a repeat. If what follows this jump
436 clearly won't match what the repeat does, such that we can be
437 sure that there is no use backtracking out of repetitions
438 already matched, then we change it to a pop_failure_jump.
439 Followed by two-byte address. */
442 /* Jump to following two-byte address, and push a dummy failure
443 point. This failure point will be thrown away if an attempt
444 is made to use it for a failure. A `+' construct makes this
445 before the first repeat. Also used as an intermediary kind
446 of jump when compiling an alternative. */
449 /* Push a dummy failure point and continue. Used at the end of
453 /* Followed by two-byte relative address and two-byte number n.
454 After matching N times, jump to the address upon failure. */
457 /* Followed by two-byte relative address, and two-byte number n.
458 Jump to the address N times, then fail. */
461 /* Set the following two-byte relative address to the
462 subsequent two-byte number. The address *includes* the two
466 wordchar, /* Matches any word-constituent character. */
467 notwordchar, /* Matches any char that is not a word-constituent. */
469 wordbeg, /* Succeeds if at word beginning. */
470 wordend, /* Succeeds if at word end. */
472 wordbound, /* Succeeds if at a word boundary. */
473 notwordbound /* Succeeds if not at a word boundary. */
476 ,before_dot, /* Succeeds if before point. */
477 at_dot, /* Succeeds if at point. */
478 after_dot, /* Succeeds if after point. */
480 /* Matches any character whose syntax is specified. Followed by
481 a byte which contains a syntax code, e.g., Sword. */
484 /* Matches any character whose syntax is not that specified. */
489 /* Common operations on the compiled pattern. */
491 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
493 #define STORE_NUMBER(destination, number) \
495 (destination)[0] = (number) & 0377; \
496 (destination)[1] = (number) >> 8; \
499 /* Same as STORE_NUMBER, except increment DESTINATION to
500 the byte after where the number is stored. Therefore, DESTINATION
501 must be an lvalue. */
503 #define STORE_NUMBER_AND_INCR(destination, number) \
505 STORE_NUMBER (destination, number); \
506 (destination) += 2; \
509 /* Put into DESTINATION a number stored in two contiguous bytes starting
512 #define EXTRACT_NUMBER(destination, source) \
514 (destination) = *(source) & 0377; \
515 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
519 static void extract_number _RE_ARGS ((int *dest, unsigned char *source));
521 extract_number (dest, source)
523 unsigned char *source;
525 int temp = SIGN_EXTEND_CHAR (*(source + 1));
526 *dest = *source & 0377;
530 # ifndef EXTRACT_MACROS /* To debug the macros. */
531 # undef EXTRACT_NUMBER
532 # define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
533 # endif /* not EXTRACT_MACROS */
537 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
538 SOURCE must be an lvalue. */
540 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
542 EXTRACT_NUMBER (destination, source); \
547 static void extract_number_and_incr _RE_ARGS ((int *destination,
548 unsigned char **source));
550 extract_number_and_incr (destination, source)
552 unsigned char **source;
554 extract_number (destination, *source);
558 # ifndef EXTRACT_MACROS
559 # undef EXTRACT_NUMBER_AND_INCR
560 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
561 extract_number_and_incr (&dest, &src)
562 # endif /* not EXTRACT_MACROS */
566 /* If DEBUG is defined, Regex prints many voluminous messages about what
567 it is doing (if the variable `debug' is nonzero). If linked with the
568 main program in `iregex.c', you can enter patterns and strings
569 interactively. And if linked with the main program in `main.c' and
570 the other test files, you can run the already-written tests. */
574 /* We use standard I/O for debugging. */
577 /* It is useful to test things that ``must'' be true when debugging. */
580 static int debug = 0;
582 # define DEBUG_STATEMENT(e) e
583 # define DEBUG_PRINT1(x) if (debug) printf (x)
584 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
585 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
586 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
587 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
588 if (debug) print_partial_compiled_pattern (s, e)
589 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
590 if (debug) print_double_string (w, s1, sz1, s2, sz2)
593 /* Print the fastmap in human-readable form. */
596 print_fastmap (fastmap)
599 unsigned was_a_range = 0;
602 while (i < (1 << BYTEWIDTH))
608 while (i < (1 << BYTEWIDTH) && fastmap[i])
624 /* Print a compiled pattern string in human-readable form, starting at
625 the START pointer into it and ending just before the pointer END. */
628 print_partial_compiled_pattern (start, end)
629 unsigned char *start;
634 unsigned char *p = start;
635 unsigned char *pend = end;
643 /* Loop over pattern commands. */
646 printf ("%d:\t", p - start);
648 switch ((re_opcode_t) *p++)
656 printf ("/exactn/%d", mcnt);
667 printf ("/start_memory/%d/%d", mcnt, *p++);
672 printf ("/stop_memory/%d/%d", mcnt, *p++);
676 printf ("/duplicate/%d", *p++);
686 register int c, last = -100;
687 register int in_range = 0;
689 printf ("/charset [%s",
690 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
692 assert (p + *p < pend);
694 for (c = 0; c < 256; c++)
696 && (p[1 + (c/8)] & (1 << (c % 8))))
698 /* Are we starting a range? */
699 if (last + 1 == c && ! in_range)
704 /* Have we broken a range? */
705 else if (last + 1 != c && in_range)
734 case on_failure_jump:
735 extract_number_and_incr (&mcnt, &p);
736 printf ("/on_failure_jump to %d", p + mcnt - start);
739 case on_failure_keep_string_jump:
740 extract_number_and_incr (&mcnt, &p);
741 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
744 case dummy_failure_jump:
745 extract_number_and_incr (&mcnt, &p);
746 printf ("/dummy_failure_jump to %d", p + mcnt - start);
749 case push_dummy_failure:
750 printf ("/push_dummy_failure");
754 extract_number_and_incr (&mcnt, &p);
755 printf ("/maybe_pop_jump to %d", p + mcnt - start);
758 case pop_failure_jump:
759 extract_number_and_incr (&mcnt, &p);
760 printf ("/pop_failure_jump to %d", p + mcnt - start);
764 extract_number_and_incr (&mcnt, &p);
765 printf ("/jump_past_alt to %d", p + mcnt - start);
769 extract_number_and_incr (&mcnt, &p);
770 printf ("/jump to %d", p + mcnt - start);
774 extract_number_and_incr (&mcnt, &p);
776 extract_number_and_incr (&mcnt2, &p);
777 printf ("/succeed_n to %d, %d times", p1 - start, mcnt2);
781 extract_number_and_incr (&mcnt, &p);
783 extract_number_and_incr (&mcnt2, &p);
784 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
788 extract_number_and_incr (&mcnt, &p);
790 extract_number_and_incr (&mcnt2, &p);
791 printf ("/set_number_at location %d to %d", p1 - start, mcnt2);
795 printf ("/wordbound");
799 printf ("/notwordbound");
811 printf ("/before_dot");
819 printf ("/after_dot");
823 printf ("/syntaxspec");
825 printf ("/%d", mcnt);
829 printf ("/notsyntaxspec");
831 printf ("/%d", mcnt);
836 printf ("/wordchar");
840 printf ("/notwordchar");
852 printf ("?%d", *(p-1));
858 printf ("%d:\tend of pattern.\n", p - start);
863 print_compiled_pattern (bufp)
864 struct re_pattern_buffer *bufp;
866 unsigned char *buffer = bufp->buffer;
868 print_partial_compiled_pattern (buffer, buffer + bufp->used);
869 printf ("%ld bytes used/%ld bytes allocated.\n",
870 bufp->used, bufp->allocated);
872 if (bufp->fastmap_accurate && bufp->fastmap)
874 printf ("fastmap: ");
875 print_fastmap (bufp->fastmap);
878 printf ("re_nsub: %d\t", bufp->re_nsub);
879 printf ("regs_alloc: %d\t", bufp->regs_allocated);
880 printf ("can_be_null: %d\t", bufp->can_be_null);
881 printf ("newline_anchor: %d\n", bufp->newline_anchor);
882 printf ("no_sub: %d\t", bufp->no_sub);
883 printf ("not_bol: %d\t", bufp->not_bol);
884 printf ("not_eol: %d\t", bufp->not_eol);
885 printf ("syntax: %lx\n", bufp->syntax);
886 /* Perhaps we should print the translate table? */
891 print_double_string (where, string1, size1, string2, size2)
904 if (FIRST_STRING_P (where))
906 for (this_char = where - string1; this_char < size1; this_char++)
907 putchar (string1[this_char]);
912 for (this_char = where - string2; this_char < size2; this_char++)
913 putchar (string2[this_char]);
924 #else /* not DEBUG */
929 # define DEBUG_STATEMENT(e)
930 # define DEBUG_PRINT1(x)
931 # define DEBUG_PRINT2(x1, x2)
932 # define DEBUG_PRINT3(x1, x2, x3)
933 # define DEBUG_PRINT4(x1, x2, x3, x4)
934 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
935 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
937 #endif /* not DEBUG */
939 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
940 also be assigned to arbitrarily: each pattern buffer stores its own
941 syntax, so it can be changed between regex compilations. */
942 /* This has no initializer because initialized variables in Emacs
943 become read-only after dumping. */
944 reg_syntax_t re_syntax_options;
947 /* Specify the precise syntax of regexps for compilation. This provides
948 for compatibility for various utilities which historically have
949 different, incompatible syntaxes.
951 The argument SYNTAX is a bit mask comprised of the various bits
952 defined in regex.h. We return the old syntax. */
955 re_set_syntax (syntax)
958 reg_syntax_t ret = re_syntax_options;
960 re_syntax_options = syntax;
962 if (syntax & RE_DEBUG)
964 else if (debug) /* was on but now is not */
970 /* This table gives an error message for each of the error codes listed
971 in regex.h. Obviously the order here has to be same as there.
972 POSIX doesn't require that we do anything for REG_NOERROR,
973 but why not be nice? */
975 static const char *re_error_msgid[] =
977 gettext_noop ("Success"), /* REG_NOERROR */
978 gettext_noop ("No match"), /* REG_NOMATCH */
979 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
980 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
981 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
982 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
983 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
984 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
985 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
986 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
987 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
988 gettext_noop ("Invalid range end"), /* REG_ERANGE */
989 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
990 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
991 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
992 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
993 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
996 /* Avoiding alloca during matching, to placate r_alloc. */
998 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
999 searching and matching functions should not call alloca. On some
1000 systems, alloca is implemented in terms of malloc, and if we're
1001 using the relocating allocator routines, then malloc could cause a
1002 relocation, which might (if the strings being searched are in the
1003 ralloc heap) shift the data out from underneath the regexp
1006 Here's another reason to avoid allocation: Emacs
1007 processes input from X in a signal handler; processing X input may
1008 call malloc; if input arrives while a matching routine is calling
1009 malloc, then we're scrod. But Emacs can't just block input while
1010 calling matching routines; then we don't notice interrupts when
1011 they come in. So, Emacs blocks input around all regexp calls
1012 except the matching calls, which it leaves unprotected, in the
1013 faith that they will not malloc. */
1015 /* Normally, this is fine. */
1016 #define MATCH_MAY_ALLOCATE
1018 /* When using GNU C, we are not REALLY using the C alloca, no matter
1019 what config.h may say. So don't take precautions for it. */
1024 /* The match routines may not allocate if (1) they would do it with malloc
1025 and (2) it's not safe for them to use malloc.
1026 Note that if REL_ALLOC is defined, matching would not use malloc for the
1027 failure stack, but we would still use it for the register vectors;
1028 so REL_ALLOC should not affect this. */
1029 #if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1030 # undef MATCH_MAY_ALLOCATE
1034 /* Failure stack declarations and macros; both re_compile_fastmap and
1035 re_match_2 use a failure stack. These have to be macros because of
1036 REGEX_ALLOCATE_STACK. */
1039 /* Number of failure points for which to initially allocate space
1040 when matching. If this number is exceeded, we allocate more
1041 space, so it is not a hard limit. */
1042 #ifndef INIT_FAILURE_ALLOC
1043 # define INIT_FAILURE_ALLOC 5
1046 /* Roughly the maximum number of failure points on the stack. Would be
1047 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1048 This is a variable only so users of regex can assign to it; we never
1049 change it ourselves. */
1053 # if defined MATCH_MAY_ALLOCATE
1054 /* 4400 was enough to cause a crash on Alpha OSF/1,
1055 whose default stack limit is 2mb. */
1056 long int re_max_failures = 4000;
1058 long int re_max_failures = 2000;
1061 union fail_stack_elt
1063 unsigned char *pointer;
1067 typedef union fail_stack_elt fail_stack_elt_t;
1071 fail_stack_elt_t *stack;
1072 unsigned long int size;
1073 unsigned long int avail; /* Offset of next open position. */
1076 #else /* not INT_IS_16BIT */
1078 # if defined MATCH_MAY_ALLOCATE
1079 /* 4400 was enough to cause a crash on Alpha OSF/1,
1080 whose default stack limit is 2mb. */
1081 int re_max_failures = 20000;
1083 int re_max_failures = 2000;
1086 union fail_stack_elt
1088 unsigned char *pointer;
1092 typedef union fail_stack_elt fail_stack_elt_t;
1096 fail_stack_elt_t *stack;
1098 unsigned avail; /* Offset of next open position. */
1101 #endif /* INT_IS_16BIT */
1103 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1104 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1105 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1108 /* Define macros to initialize and free the failure stack.
1109 Do `return -2' if the alloc fails. */
1111 #ifdef MATCH_MAY_ALLOCATE
1112 # define INIT_FAIL_STACK() \
1114 fail_stack.stack = (fail_stack_elt_t *) \
1115 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1117 if (fail_stack.stack == NULL) \
1120 fail_stack.size = INIT_FAILURE_ALLOC; \
1121 fail_stack.avail = 0; \
1124 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1126 # define INIT_FAIL_STACK() \
1128 fail_stack.avail = 0; \
1131 # define RESET_FAIL_STACK()
1135 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1137 Return 1 if succeeds, and 0 if either ran out of memory
1138 allocating space for it or it was already too large.
1140 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1142 #define DOUBLE_FAIL_STACK(fail_stack) \
1143 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1145 : ((fail_stack).stack = (fail_stack_elt_t *) \
1146 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1147 (fail_stack).size * sizeof (fail_stack_elt_t), \
1148 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1150 (fail_stack).stack == NULL \
1152 : ((fail_stack).size <<= 1, \
1156 /* Push pointer POINTER on FAIL_STACK.
1157 Return 1 if was able to do so and 0 if ran out of memory allocating
1159 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1160 ((FAIL_STACK_FULL () \
1161 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1163 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1166 /* Push a pointer value onto the failure stack.
1167 Assumes the variable `fail_stack'. Probably should only
1168 be called from within `PUSH_FAILURE_POINT'. */
1169 #define PUSH_FAILURE_POINTER(item) \
1170 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1172 /* This pushes an integer-valued item onto the failure stack.
1173 Assumes the variable `fail_stack'. Probably should only
1174 be called from within `PUSH_FAILURE_POINT'. */
1175 #define PUSH_FAILURE_INT(item) \
1176 fail_stack.stack[fail_stack.avail++].integer = (item)
1178 /* Push a fail_stack_elt_t value onto the failure stack.
1179 Assumes the variable `fail_stack'. Probably should only
1180 be called from within `PUSH_FAILURE_POINT'. */
1181 #define PUSH_FAILURE_ELT(item) \
1182 fail_stack.stack[fail_stack.avail++] = (item)
1184 /* These three POP... operations complement the three PUSH... operations.
1185 All assume that `fail_stack' is nonempty. */
1186 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1187 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1188 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1190 /* Used to omit pushing failure point id's when we're not debugging. */
1192 # define DEBUG_PUSH PUSH_FAILURE_INT
1193 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1195 # define DEBUG_PUSH(item)
1196 # define DEBUG_POP(item_addr)
1200 /* Push the information about the state we will need
1201 if we ever fail back to it.
1203 Requires variables fail_stack, regstart, regend, reg_info, and
1204 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1207 Does `return FAILURE_CODE' if runs out of memory. */
1209 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1211 char *destination; \
1212 /* Must be int, so when we don't save any registers, the arithmetic \
1213 of 0 + -1 isn't done as unsigned. */ \
1214 /* Can't be int, since there is not a shred of a guarantee that int \
1215 is wide enough to hold a value of something to which pointer can \
1217 active_reg_t this_reg; \
1219 DEBUG_STATEMENT (failure_id++); \
1220 DEBUG_STATEMENT (nfailure_points_pushed++); \
1221 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1222 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1223 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1225 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1226 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1228 /* Ensure we have enough space allocated for what we will push. */ \
1229 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1231 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1232 return failure_code; \
1234 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1235 (fail_stack).size); \
1236 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1239 /* Push the info, starting with the registers. */ \
1240 DEBUG_PRINT1 ("\n"); \
1243 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1246 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1247 DEBUG_STATEMENT (num_regs_pushed++); \
1249 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1250 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1252 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1253 PUSH_FAILURE_POINTER (regend[this_reg]); \
1255 DEBUG_PRINT2 (" info: %p\n ", \
1256 reg_info[this_reg].word.pointer); \
1257 DEBUG_PRINT2 (" match_null=%d", \
1258 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1259 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1260 DEBUG_PRINT2 (" matched_something=%d", \
1261 MATCHED_SOMETHING (reg_info[this_reg])); \
1262 DEBUG_PRINT2 (" ever_matched=%d", \
1263 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1264 DEBUG_PRINT1 ("\n"); \
1265 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1268 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1269 PUSH_FAILURE_INT (lowest_active_reg); \
1271 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1272 PUSH_FAILURE_INT (highest_active_reg); \
1274 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1275 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1276 PUSH_FAILURE_POINTER (pattern_place); \
1278 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1279 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1281 DEBUG_PRINT1 ("'\n"); \
1282 PUSH_FAILURE_POINTER (string_place); \
1284 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1285 DEBUG_PUSH (failure_id); \
1288 /* This is the number of items that are pushed and popped on the stack
1289 for each register. */
1290 #define NUM_REG_ITEMS 3
1292 /* Individual items aside from the registers. */
1294 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1296 # define NUM_NONREG_ITEMS 4
1299 /* We push at most this many items on the stack. */
1300 /* We used to use (num_regs - 1), which is the number of registers
1301 this regexp will save; but that was changed to 5
1302 to avoid stack overflow for a regexp with lots of parens. */
1303 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1305 /* We actually push this many items. */
1306 #define NUM_FAILURE_ITEMS \
1308 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1312 /* How many items can still be added to the stack without overflowing it. */
1313 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1316 /* Pops what PUSH_FAIL_STACK pushes.
1318 We restore into the parameters, all of which should be lvalues:
1319 STR -- the saved data position.
1320 PAT -- the saved pattern position.
1321 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1322 REGSTART, REGEND -- arrays of string positions.
1323 REG_INFO -- array of information about each subexpression.
1325 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1326 `pend', `string1', `size1', `string2', and `size2'. */
1328 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1330 DEBUG_STATEMENT (unsigned failure_id;) \
1331 active_reg_t this_reg; \
1332 const unsigned char *string_temp; \
1334 assert (!FAIL_STACK_EMPTY ()); \
1336 /* Remove failure points and point to how many regs pushed. */ \
1337 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1338 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1339 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1341 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1343 DEBUG_POP (&failure_id); \
1344 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1346 /* If the saved string location is NULL, it came from an \
1347 on_failure_keep_string_jump opcode, and we want to throw away the \
1348 saved NULL, thus retaining our current position in the string. */ \
1349 string_temp = POP_FAILURE_POINTER (); \
1350 if (string_temp != NULL) \
1351 str = (const char *) string_temp; \
1353 DEBUG_PRINT2 (" Popping string %p: `", str); \
1354 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1355 DEBUG_PRINT1 ("'\n"); \
1357 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1358 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1359 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1361 /* Restore register info. */ \
1362 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1363 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1365 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1366 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1369 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1371 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1373 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1374 DEBUG_PRINT2 (" info: %p\n", \
1375 reg_info[this_reg].word.pointer); \
1377 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1378 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1380 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1381 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1385 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1387 reg_info[this_reg].word.integer = 0; \
1388 regend[this_reg] = 0; \
1389 regstart[this_reg] = 0; \
1391 highest_active_reg = high_reg; \
1394 set_regs_matched_done = 0; \
1395 DEBUG_STATEMENT (nfailure_points_popped++); \
1396 } /* POP_FAILURE_POINT */
1400 /* Structure for per-register (a.k.a. per-group) information.
1401 Other register information, such as the
1402 starting and ending positions (which are addresses), and the list of
1403 inner groups (which is a bits list) are maintained in separate
1406 We are making a (strictly speaking) nonportable assumption here: that
1407 the compiler will pack our bit fields into something that fits into
1408 the type of `word', i.e., is something that fits into one item on the
1412 /* Declarations and macros for re_match_2. */
1416 fail_stack_elt_t word;
1419 /* This field is one if this group can match the empty string,
1420 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1421 #define MATCH_NULL_UNSET_VALUE 3
1422 unsigned match_null_string_p : 2;
1423 unsigned is_active : 1;
1424 unsigned matched_something : 1;
1425 unsigned ever_matched_something : 1;
1427 } register_info_type;
1429 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1430 #define IS_ACTIVE(R) ((R).bits.is_active)
1431 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1432 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1435 /* Call this when have matched a real character; it sets `matched' flags
1436 for the subexpressions which we are currently inside. Also records
1437 that those subexprs have matched. */
1438 #define SET_REGS_MATCHED() \
1441 if (!set_regs_matched_done) \
1444 set_regs_matched_done = 1; \
1445 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1447 MATCHED_SOMETHING (reg_info[r]) \
1448 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1455 /* Registers are set to a sentinel when they haven't yet matched. */
1456 static char reg_unset_dummy;
1457 #define REG_UNSET_VALUE (®_unset_dummy)
1458 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1460 /* Subroutine declarations and macros for regex_compile. */
1462 static reg_errcode_t regex_compile _RE_ARGS ((const char *pattern, size_t size,
1463 reg_syntax_t syntax,
1464 struct re_pattern_buffer *bufp));
1465 static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg));
1466 static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1467 int arg1, int arg2));
1468 static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1469 int arg, unsigned char *end));
1470 static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1471 int arg1, int arg2, unsigned char *end));
1472 static boolean at_begline_loc_p _RE_ARGS ((const char *pattern, const char *p,
1473 reg_syntax_t syntax));
1474 static boolean at_endline_loc_p _RE_ARGS ((const char *p, const char *pend,
1475 reg_syntax_t syntax));
1476 static reg_errcode_t compile_range _RE_ARGS ((const char **p_ptr,
1479 reg_syntax_t syntax,
1482 /* Fetch the next character in the uncompiled pattern---translating it
1483 if necessary. Also cast from a signed character in the constant
1484 string passed to us by the user to an unsigned char that we can use
1485 as an array index (in, e.g., `translate'). */
1487 # define PATFETCH(c) \
1488 do {if (p == pend) return REG_EEND; \
1489 c = (unsigned char) *p++; \
1490 if (translate) c = (unsigned char) translate[c]; \
1494 /* Fetch the next character in the uncompiled pattern, with no
1496 #define PATFETCH_RAW(c) \
1497 do {if (p == pend) return REG_EEND; \
1498 c = (unsigned char) *p++; \
1501 /* Go backwards one character in the pattern. */
1502 #define PATUNFETCH p--
1505 /* If `translate' is non-null, return translate[D], else just D. We
1506 cast the subscript to translate because some data is declared as
1507 `char *', to avoid warnings when a string constant is passed. But
1508 when we use a character as a subscript we must make it unsigned. */
1510 # define TRANSLATE(d) \
1511 (translate ? (char) translate[(unsigned char) (d)] : (d))
1515 /* Macros for outputting the compiled pattern into `buffer'. */
1517 /* If the buffer isn't allocated when it comes in, use this. */
1518 #define INIT_BUF_SIZE 32
1520 /* Make sure we have at least N more bytes of space in buffer. */
1521 #define GET_BUFFER_SPACE(n) \
1522 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1525 /* Make sure we have one more byte of buffer space and then add C to it. */
1526 #define BUF_PUSH(c) \
1528 GET_BUFFER_SPACE (1); \
1529 *b++ = (unsigned char) (c); \
1533 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1534 #define BUF_PUSH_2(c1, c2) \
1536 GET_BUFFER_SPACE (2); \
1537 *b++ = (unsigned char) (c1); \
1538 *b++ = (unsigned char) (c2); \
1542 /* As with BUF_PUSH_2, except for three bytes. */
1543 #define BUF_PUSH_3(c1, c2, c3) \
1545 GET_BUFFER_SPACE (3); \
1546 *b++ = (unsigned char) (c1); \
1547 *b++ = (unsigned char) (c2); \
1548 *b++ = (unsigned char) (c3); \
1552 /* Store a jump with opcode OP at LOC to location TO. We store a
1553 relative address offset by the three bytes the jump itself occupies. */
1554 #define STORE_JUMP(op, loc, to) \
1555 store_op1 (op, loc, (int) ((to) - (loc) - 3))
1557 /* Likewise, for a two-argument jump. */
1558 #define STORE_JUMP2(op, loc, to, arg) \
1559 store_op2 (op, loc, (int) ((to) - (loc) - 3), arg)
1561 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1562 #define INSERT_JUMP(op, loc, to) \
1563 insert_op1 (op, loc, (int) ((to) - (loc) - 3), b)
1565 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1566 #define INSERT_JUMP2(op, loc, to, arg) \
1567 insert_op2 (op, loc, (int) ((to) - (loc) - 3), arg, b)
1570 /* This is not an arbitrary limit: the arguments which represent offsets
1571 into the pattern are two bytes long. So if 2^16 bytes turns out to
1572 be too small, many things would have to change. */
1573 /* Any other compiler which, like MSC, has allocation limit below 2^16
1574 bytes will have to use approach similar to what was done below for
1575 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1576 reallocating to 0 bytes. Such thing is not going to work too well.
1577 You have been warned!! */
1578 #if defined _MSC_VER && !defined WIN32
1579 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
1580 The REALLOC define eliminates a flurry of conversion warnings,
1581 but is not required. */
1582 # define MAX_BUF_SIZE 65500L
1583 # define REALLOC(p,s) realloc ((p), (size_t) (s))
1585 # define MAX_BUF_SIZE (1L << 16)
1586 # define REALLOC(p,s) realloc ((p), (s))
1589 /* Extend the buffer by twice its current size via realloc and
1590 reset the pointers that pointed into the old block to point to the
1591 correct places in the new one. If extending the buffer results in it
1592 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1593 #define EXTEND_BUFFER() \
1595 unsigned char *old_buffer = bufp->buffer; \
1596 if (bufp->allocated == MAX_BUF_SIZE) \
1598 bufp->allocated <<= 1; \
1599 if (bufp->allocated > MAX_BUF_SIZE) \
1600 bufp->allocated = MAX_BUF_SIZE; \
1601 bufp->buffer = (unsigned char *) REALLOC (bufp->buffer, bufp->allocated);\
1602 if (bufp->buffer == NULL) \
1603 return REG_ESPACE; \
1604 /* If the buffer moved, move all the pointers into it. */ \
1605 if (old_buffer != bufp->buffer) \
1607 b = (b - old_buffer) + bufp->buffer; \
1608 begalt = (begalt - old_buffer) + bufp->buffer; \
1609 if (fixup_alt_jump) \
1610 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1612 laststart = (laststart - old_buffer) + bufp->buffer; \
1613 if (pending_exact) \
1614 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1619 /* Since we have one byte reserved for the register number argument to
1620 {start,stop}_memory, the maximum number of groups we can report
1621 things about is what fits in that byte. */
1622 #define MAX_REGNUM 255
1624 /* But patterns can have more than `MAX_REGNUM' registers. We just
1625 ignore the excess. */
1626 typedef unsigned regnum_t;
1629 /* Macros for the compile stack. */
1631 /* Since offsets can go either forwards or backwards, this type needs to
1632 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1633 /* int may be not enough when sizeof(int) == 2. */
1634 typedef long pattern_offset_t;
1638 pattern_offset_t begalt_offset;
1639 pattern_offset_t fixup_alt_jump;
1640 pattern_offset_t inner_group_offset;
1641 pattern_offset_t laststart_offset;
1643 } compile_stack_elt_t;
1648 compile_stack_elt_t *stack;
1650 unsigned avail; /* Offset of next open position. */
1651 } compile_stack_type;
1654 #define INIT_COMPILE_STACK_SIZE 32
1656 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1657 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1659 /* The next available element. */
1660 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1663 /* Set the bit for character C in a list. */
1664 #define SET_LIST_BIT(c) \
1665 (b[((unsigned char) (c)) / BYTEWIDTH] \
1666 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1669 /* Get the next unsigned number in the uncompiled pattern. */
1670 #define GET_UNSIGNED_NUMBER(num) \
1674 while (ISDIGIT (c)) \
1678 num = num * 10 + c - '0'; \
1686 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
1687 /* The GNU C library provides support for user-defined character classes
1688 and the functions from ISO C amendement 1. */
1689 # ifdef CHARCLASS_NAME_MAX
1690 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
1692 /* This shouldn't happen but some implementation might still have this
1693 problem. Use a reasonable default value. */
1694 # define CHAR_CLASS_MAX_LENGTH 256
1697 # define IS_CHAR_CLASS(string) wctype (string)
1699 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1701 # define IS_CHAR_CLASS(string) \
1702 (STREQ (string, "alpha") || STREQ (string, "upper") \
1703 || STREQ (string, "lower") || STREQ (string, "digit") \
1704 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1705 || STREQ (string, "space") || STREQ (string, "print") \
1706 || STREQ (string, "punct") || STREQ (string, "graph") \
1707 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1710 #ifndef MATCH_MAY_ALLOCATE
1712 /* If we cannot allocate large objects within re_match_2_internal,
1713 we make the fail stack and register vectors global.
1714 The fail stack, we grow to the maximum size when a regexp
1716 The register vectors, we adjust in size each time we
1717 compile a regexp, according to the number of registers it needs. */
1719 static fail_stack_type fail_stack;
1721 /* Size with which the following vectors are currently allocated.
1722 That is so we can make them bigger as needed,
1723 but never make them smaller. */
1724 static int regs_allocated_size;
1726 static const char ** regstart, ** regend;
1727 static const char ** old_regstart, ** old_regend;
1728 static const char **best_regstart, **best_regend;
1729 static register_info_type *reg_info;
1730 static const char **reg_dummy;
1731 static register_info_type *reg_info_dummy;
1733 /* Make the register vectors big enough for NUM_REGS registers,
1734 but don't make them smaller. */
1737 regex_grow_registers (num_regs)
1740 if (num_regs > regs_allocated_size)
1742 RETALLOC_IF (regstart, num_regs, const char *);
1743 RETALLOC_IF (regend, num_regs, const char *);
1744 RETALLOC_IF (old_regstart, num_regs, const char *);
1745 RETALLOC_IF (old_regend, num_regs, const char *);
1746 RETALLOC_IF (best_regstart, num_regs, const char *);
1747 RETALLOC_IF (best_regend, num_regs, const char *);
1748 RETALLOC_IF (reg_info, num_regs, register_info_type);
1749 RETALLOC_IF (reg_dummy, num_regs, const char *);
1750 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1752 regs_allocated_size = num_regs;
1756 #endif /* not MATCH_MAY_ALLOCATE */
1758 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
1762 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1763 Returns one of error codes defined in `regex.h', or zero for success.
1765 Assumes the `allocated' (and perhaps `buffer') and `translate'
1766 fields are set in BUFP on entry.
1768 If it succeeds, results are put in BUFP (if it returns an error, the
1769 contents of BUFP are undefined):
1770 `buffer' is the compiled pattern;
1771 `syntax' is set to SYNTAX;
1772 `used' is set to the length of the compiled pattern;
1773 `fastmap_accurate' is zero;
1774 `re_nsub' is the number of subexpressions in PATTERN;
1775 `not_bol' and `not_eol' are zero;
1777 The `fastmap' and `newline_anchor' fields are neither
1778 examined nor set. */
1780 /* Return, freeing storage we allocated. */
1781 #define FREE_STACK_RETURN(value) \
1782 return (free (compile_stack.stack), value)
1784 static reg_errcode_t
1785 regex_compile (pattern, size, syntax, bufp)
1786 const char *pattern;
1788 reg_syntax_t syntax;
1789 struct re_pattern_buffer *bufp;
1791 /* We fetch characters from PATTERN here. Even though PATTERN is
1792 `char *' (i.e., signed), we declare these variables as unsigned, so
1793 they can be reliably used as array indices. */
1794 register unsigned char c, c1;
1796 /* A random temporary spot in PATTERN. */
1799 /* Points to the end of the buffer, where we should append. */
1800 register unsigned char *b;
1802 /* Keeps track of unclosed groups. */
1803 compile_stack_type compile_stack;
1805 /* Points to the current (ending) position in the pattern. */
1806 const char *p = pattern;
1807 const char *pend = pattern + size;
1809 /* How to translate the characters in the pattern. */
1810 RE_TRANSLATE_TYPE translate = bufp->translate;
1812 /* Address of the count-byte of the most recently inserted `exactn'
1813 command. This makes it possible to tell if a new exact-match
1814 character can be added to that command or if the character requires
1815 a new `exactn' command. */
1816 unsigned char *pending_exact = 0;
1818 /* Address of start of the most recently finished expression.
1819 This tells, e.g., postfix * where to find the start of its
1820 operand. Reset at the beginning of groups and alternatives. */
1821 unsigned char *laststart = 0;
1823 /* Address of beginning of regexp, or inside of last group. */
1824 unsigned char *begalt;
1826 /* Place in the uncompiled pattern (i.e., the {) to
1827 which to go back if the interval is invalid. */
1828 const char *beg_interval;
1830 /* Address of the place where a forward jump should go to the end of
1831 the containing expression. Each alternative of an `or' -- except the
1832 last -- ends with a forward jump of this sort. */
1833 unsigned char *fixup_alt_jump = 0;
1835 /* Counts open-groups as they are encountered. Remembered for the
1836 matching close-group on the compile stack, so the same register
1837 number is put in the stop_memory as the start_memory. */
1838 regnum_t regnum = 0;
1841 DEBUG_PRINT1 ("\nCompiling pattern: ");
1844 unsigned debug_count;
1846 for (debug_count = 0; debug_count < size; debug_count++)
1847 putchar (pattern[debug_count]);
1852 /* Initialize the compile stack. */
1853 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1854 if (compile_stack.stack == NULL)
1857 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1858 compile_stack.avail = 0;
1860 /* Initialize the pattern buffer. */
1861 bufp->syntax = syntax;
1862 bufp->fastmap_accurate = 0;
1863 bufp->not_bol = bufp->not_eol = 0;
1865 /* Set `used' to zero, so that if we return an error, the pattern
1866 printer (for debugging) will think there's no pattern. We reset it
1870 /* Always count groups, whether or not bufp->no_sub is set. */
1873 #if !defined emacs && !defined SYNTAX_TABLE
1874 /* Initialize the syntax table. */
1875 init_syntax_once ();
1878 if (bufp->allocated == 0)
1881 { /* If zero allocated, but buffer is non-null, try to realloc
1882 enough space. This loses if buffer's address is bogus, but
1883 that is the user's responsibility. */
1884 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1887 { /* Caller did not allocate a buffer. Do it for them. */
1888 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1890 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1892 bufp->allocated = INIT_BUF_SIZE;
1895 begalt = b = bufp->buffer;
1897 /* Loop through the uncompiled pattern until we're at the end. */
1906 if ( /* If at start of pattern, it's an operator. */
1908 /* If context independent, it's an operator. */
1909 || syntax & RE_CONTEXT_INDEP_ANCHORS
1910 /* Otherwise, depends on what's come before. */
1911 || at_begline_loc_p (pattern, p, syntax))
1921 if ( /* If at end of pattern, it's an operator. */
1923 /* If context independent, it's an operator. */
1924 || syntax & RE_CONTEXT_INDEP_ANCHORS
1925 /* Otherwise, depends on what's next. */
1926 || at_endline_loc_p (p, pend, syntax))
1936 if ((syntax & RE_BK_PLUS_QM)
1937 || (syntax & RE_LIMITED_OPS))
1941 /* If there is no previous pattern... */
1944 if (syntax & RE_CONTEXT_INVALID_OPS)
1945 FREE_STACK_RETURN (REG_BADRPT);
1946 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1951 /* Are we optimizing this jump? */
1952 boolean keep_string_p = false;
1954 /* 1 means zero (many) matches is allowed. */
1955 char zero_times_ok = 0, many_times_ok = 0;
1957 /* If there is a sequence of repetition chars, collapse it
1958 down to just one (the right one). We can't combine
1959 interval operators with these because of, e.g., `a{2}*',
1960 which should only match an even number of `a's. */
1964 zero_times_ok |= c != '+';
1965 many_times_ok |= c != '?';
1973 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
1976 else if (syntax & RE_BK_PLUS_QM && c == '\\')
1978 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1981 if (!(c1 == '+' || c1 == '?'))
1996 /* If we get here, we found another repeat character. */
1999 /* Star, etc. applied to an empty pattern is equivalent
2000 to an empty pattern. */
2004 /* Now we know whether or not zero matches is allowed
2005 and also whether or not two or more matches is allowed. */
2007 { /* More than one repetition is allowed, so put in at the
2008 end a backward relative jump from `b' to before the next
2009 jump we're going to put in below (which jumps from
2010 laststart to after this jump).
2012 But if we are at the `*' in the exact sequence `.*\n',
2013 insert an unconditional jump backwards to the .,
2014 instead of the beginning of the loop. This way we only
2015 push a failure point once, instead of every time
2016 through the loop. */
2017 assert (p - 1 > pattern);
2019 /* Allocate the space for the jump. */
2020 GET_BUFFER_SPACE (3);
2022 /* We know we are not at the first character of the pattern,
2023 because laststart was nonzero. And we've already
2024 incremented `p', by the way, to be the character after
2025 the `*'. Do we have to do something analogous here
2026 for null bytes, because of RE_DOT_NOT_NULL? */
2027 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2029 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2030 && !(syntax & RE_DOT_NEWLINE))
2031 { /* We have .*\n. */
2032 STORE_JUMP (jump, b, laststart);
2033 keep_string_p = true;
2036 /* Anything else. */
2037 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
2039 /* We've added more stuff to the buffer. */
2043 /* On failure, jump from laststart to b + 3, which will be the
2044 end of the buffer after this jump is inserted. */
2045 GET_BUFFER_SPACE (3);
2046 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2054 /* At least one repetition is required, so insert a
2055 `dummy_failure_jump' before the initial
2056 `on_failure_jump' instruction of the loop. This
2057 effects a skip over that instruction the first time
2058 we hit that loop. */
2059 GET_BUFFER_SPACE (3);
2060 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
2075 boolean had_char_class = false;
2077 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2079 /* Ensure that we have enough space to push a charset: the
2080 opcode, the length count, and the bitset; 34 bytes in all. */
2081 GET_BUFFER_SPACE (34);
2085 /* We test `*p == '^' twice, instead of using an if
2086 statement, so we only need one BUF_PUSH. */
2087 BUF_PUSH (*p == '^' ? charset_not : charset);
2091 /* Remember the first position in the bracket expression. */
2094 /* Push the number of bytes in the bitmap. */
2095 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
2097 /* Clear the whole map. */
2098 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
2100 /* charset_not matches newline according to a syntax bit. */
2101 if ((re_opcode_t) b[-2] == charset_not
2102 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2103 SET_LIST_BIT ('\n');
2105 /* Read in characters and ranges, setting map bits. */
2108 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2112 /* \ might escape characters inside [...] and [^...]. */
2113 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2115 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2122 /* Could be the end of the bracket expression. If it's
2123 not (i.e., when the bracket expression is `[]' so
2124 far), the ']' character bit gets set way below. */
2125 if (c == ']' && p != p1 + 1)
2128 /* Look ahead to see if it's a range when the last thing
2129 was a character class. */
2130 if (had_char_class && c == '-' && *p != ']')
2131 FREE_STACK_RETURN (REG_ERANGE);
2133 /* Look ahead to see if it's a range when the last thing
2134 was a character: if this is a hyphen not at the
2135 beginning or the end of a list, then it's the range
2138 && !(p - 2 >= pattern && p[-2] == '[')
2139 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2143 = compile_range (&p, pend, translate, syntax, b);
2144 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2147 else if (p[0] == '-' && p[1] != ']')
2148 { /* This handles ranges made up of characters only. */
2151 /* Move past the `-'. */
2154 ret = compile_range (&p, pend, translate, syntax, b);
2155 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2158 /* See if we're at the beginning of a possible character
2161 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2162 { /* Leave room for the null. */
2163 char str[CHAR_CLASS_MAX_LENGTH + 1];
2168 /* If pattern is `[[:'. */
2169 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2174 if (c == ':' || c == ']' || p == pend
2175 || c1 == CHAR_CLASS_MAX_LENGTH)
2181 /* If isn't a word bracketed by `[:' and:`]':
2182 undo the ending character, the letters, and leave
2183 the leading `:' and `[' (but set bits for them). */
2184 if (c == ':' && *p == ']')
2186 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
2187 boolean is_lower = STREQ (str, "lower");
2188 boolean is_upper = STREQ (str, "upper");
2194 FREE_STACK_RETURN (REG_ECTYPE);
2196 /* Throw away the ] at the end of the character
2200 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2202 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
2204 if (iswctype (btowc (ch), wt))
2207 if (translate && (is_upper || is_lower)
2208 && (ISUPPER (ch) || ISLOWER (ch)))
2212 had_char_class = true;
2215 boolean is_alnum = STREQ (str, "alnum");
2216 boolean is_alpha = STREQ (str, "alpha");
2217 boolean is_blank = STREQ (str, "blank");
2218 boolean is_cntrl = STREQ (str, "cntrl");
2219 boolean is_digit = STREQ (str, "digit");
2220 boolean is_graph = STREQ (str, "graph");
2221 boolean is_lower = STREQ (str, "lower");
2222 boolean is_print = STREQ (str, "print");
2223 boolean is_punct = STREQ (str, "punct");
2224 boolean is_space = STREQ (str, "space");
2225 boolean is_upper = STREQ (str, "upper");
2226 boolean is_xdigit = STREQ (str, "xdigit");
2228 if (!IS_CHAR_CLASS (str))
2229 FREE_STACK_RETURN (REG_ECTYPE);
2231 /* Throw away the ] at the end of the character
2235 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2237 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2239 /* This was split into 3 if's to
2240 avoid an arbitrary limit in some compiler. */
2241 if ( (is_alnum && ISALNUM (ch))
2242 || (is_alpha && ISALPHA (ch))
2243 || (is_blank && ISBLANK (ch))
2244 || (is_cntrl && ISCNTRL (ch)))
2246 if ( (is_digit && ISDIGIT (ch))
2247 || (is_graph && ISGRAPH (ch))
2248 || (is_lower && ISLOWER (ch))
2249 || (is_print && ISPRINT (ch)))
2251 if ( (is_punct && ISPUNCT (ch))
2252 || (is_space && ISSPACE (ch))
2253 || (is_upper && ISUPPER (ch))
2254 || (is_xdigit && ISXDIGIT (ch)))
2256 if ( translate && (is_upper || is_lower)
2257 && (ISUPPER (ch) || ISLOWER (ch)))
2260 had_char_class = true;
2261 #endif /* libc || wctype.h */
2270 had_char_class = false;
2275 had_char_class = false;
2280 /* Discard any (non)matching list bytes that are all 0 at the
2281 end of the map. Decrease the map-length byte too. */
2282 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2290 if (syntax & RE_NO_BK_PARENS)
2297 if (syntax & RE_NO_BK_PARENS)
2304 if (syntax & RE_NEWLINE_ALT)
2311 if (syntax & RE_NO_BK_VBAR)
2318 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2319 goto handle_interval;
2325 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2327 /* Do not translate the character after the \, so that we can
2328 distinguish, e.g., \B from \b, even if we normally would
2329 translate, e.g., B to b. */
2335 if (syntax & RE_NO_BK_PARENS)
2336 goto normal_backslash;
2342 if (COMPILE_STACK_FULL)
2344 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2345 compile_stack_elt_t);
2346 if (compile_stack.stack == NULL) return REG_ESPACE;
2348 compile_stack.size <<= 1;
2351 /* These are the values to restore when we hit end of this
2352 group. They are all relative offsets, so that if the
2353 whole pattern moves because of realloc, they will still
2355 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2356 COMPILE_STACK_TOP.fixup_alt_jump
2357 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2358 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2359 COMPILE_STACK_TOP.regnum = regnum;
2361 /* We will eventually replace the 0 with the number of
2362 groups inner to this one. But do not push a
2363 start_memory for groups beyond the last one we can
2364 represent in the compiled pattern. */
2365 if (regnum <= MAX_REGNUM)
2367 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2368 BUF_PUSH_3 (start_memory, regnum, 0);
2371 compile_stack.avail++;
2376 /* If we've reached MAX_REGNUM groups, then this open
2377 won't actually generate any code, so we'll have to
2378 clear pending_exact explicitly. */
2384 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2386 if (COMPILE_STACK_EMPTY)
2388 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2389 goto normal_backslash;
2391 FREE_STACK_RETURN (REG_ERPAREN);
2396 { /* Push a dummy failure point at the end of the
2397 alternative for a possible future
2398 `pop_failure_jump' to pop. See comments at
2399 `push_dummy_failure' in `re_match_2'. */
2400 BUF_PUSH (push_dummy_failure);
2402 /* We allocated space for this jump when we assigned
2403 to `fixup_alt_jump', in the `handle_alt' case below. */
2404 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2407 /* See similar code for backslashed left paren above. */
2408 if (COMPILE_STACK_EMPTY)
2410 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2413 FREE_STACK_RETURN (REG_ERPAREN);
2416 /* Since we just checked for an empty stack above, this
2417 ``can't happen''. */
2418 assert (compile_stack.avail != 0);
2420 /* We don't just want to restore into `regnum', because
2421 later groups should continue to be numbered higher,
2422 as in `(ab)c(de)' -- the second group is #2. */
2423 regnum_t this_group_regnum;
2425 compile_stack.avail--;
2426 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2428 = COMPILE_STACK_TOP.fixup_alt_jump
2429 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2431 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2432 this_group_regnum = COMPILE_STACK_TOP.regnum;
2433 /* If we've reached MAX_REGNUM groups, then this open
2434 won't actually generate any code, so we'll have to
2435 clear pending_exact explicitly. */
2438 /* We're at the end of the group, so now we know how many
2439 groups were inside this one. */
2440 if (this_group_regnum <= MAX_REGNUM)
2442 unsigned char *inner_group_loc
2443 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2445 *inner_group_loc = regnum - this_group_regnum;
2446 BUF_PUSH_3 (stop_memory, this_group_regnum,
2447 regnum - this_group_regnum);
2453 case '|': /* `\|'. */
2454 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2455 goto normal_backslash;
2457 if (syntax & RE_LIMITED_OPS)
2460 /* Insert before the previous alternative a jump which
2461 jumps to this alternative if the former fails. */
2462 GET_BUFFER_SPACE (3);
2463 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2467 /* The alternative before this one has a jump after it
2468 which gets executed if it gets matched. Adjust that
2469 jump so it will jump to this alternative's analogous
2470 jump (put in below, which in turn will jump to the next
2471 (if any) alternative's such jump, etc.). The last such
2472 jump jumps to the correct final destination. A picture:
2478 If we are at `b', then fixup_alt_jump right now points to a
2479 three-byte space after `a'. We'll put in the jump, set
2480 fixup_alt_jump to right after `b', and leave behind three
2481 bytes which we'll fill in when we get to after `c'. */
2484 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2486 /* Mark and leave space for a jump after this alternative,
2487 to be filled in later either by next alternative or
2488 when know we're at the end of a series of alternatives. */
2490 GET_BUFFER_SPACE (3);
2499 /* If \{ is a literal. */
2500 if (!(syntax & RE_INTERVALS)
2501 /* If we're at `\{' and it's not the open-interval
2503 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2504 || (p - 2 == pattern && p == pend))
2505 goto normal_backslash;
2509 /* If got here, then the syntax allows intervals. */
2511 /* At least (most) this many matches must be made. */
2512 int lower_bound = -1, upper_bound = -1;
2514 beg_interval = p - 1;
2518 if (syntax & RE_NO_BK_BRACES)
2519 goto unfetch_interval;
2521 FREE_STACK_RETURN (REG_EBRACE);
2524 GET_UNSIGNED_NUMBER (lower_bound);
2528 GET_UNSIGNED_NUMBER (upper_bound);
2529 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2532 /* Interval such as `{1}' => match exactly once. */
2533 upper_bound = lower_bound;
2535 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2536 || lower_bound > upper_bound)
2538 if (syntax & RE_NO_BK_BRACES)
2539 goto unfetch_interval;
2541 FREE_STACK_RETURN (REG_BADBR);
2544 if (!(syntax & RE_NO_BK_BRACES))
2546 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2553 if (syntax & RE_NO_BK_BRACES)
2554 goto unfetch_interval;
2556 FREE_STACK_RETURN (REG_BADBR);
2559 /* We just parsed a valid interval. */
2561 /* If it's invalid to have no preceding re. */
2564 if (syntax & RE_CONTEXT_INVALID_OPS)
2565 FREE_STACK_RETURN (REG_BADRPT);
2566 else if (syntax & RE_CONTEXT_INDEP_OPS)
2569 goto unfetch_interval;
2572 /* If the upper bound is zero, don't want to succeed at
2573 all; jump from `laststart' to `b + 3', which will be
2574 the end of the buffer after we insert the jump. */
2575 if (upper_bound == 0)
2577 GET_BUFFER_SPACE (3);
2578 INSERT_JUMP (jump, laststart, b + 3);
2582 /* Otherwise, we have a nontrivial interval. When
2583 we're all done, the pattern will look like:
2584 set_number_at <jump count> <upper bound>
2585 set_number_at <succeed_n count> <lower bound>
2586 succeed_n <after jump addr> <succeed_n count>
2588 jump_n <succeed_n addr> <jump count>
2589 (The upper bound and `jump_n' are omitted if
2590 `upper_bound' is 1, though.) */
2592 { /* If the upper bound is > 1, we need to insert
2593 more at the end of the loop. */
2594 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2596 GET_BUFFER_SPACE (nbytes);
2598 /* Initialize lower bound of the `succeed_n', even
2599 though it will be set during matching by its
2600 attendant `set_number_at' (inserted next),
2601 because `re_compile_fastmap' needs to know.
2602 Jump to the `jump_n' we might insert below. */
2603 INSERT_JUMP2 (succeed_n, laststart,
2604 b + 5 + (upper_bound > 1) * 5,
2608 /* Code to initialize the lower bound. Insert
2609 before the `succeed_n'. The `5' is the last two
2610 bytes of this `set_number_at', plus 3 bytes of
2611 the following `succeed_n'. */
2612 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2615 if (upper_bound > 1)
2616 { /* More than one repetition is allowed, so
2617 append a backward jump to the `succeed_n'
2618 that starts this interval.
2620 When we've reached this during matching,
2621 we'll have matched the interval once, so
2622 jump back only `upper_bound - 1' times. */
2623 STORE_JUMP2 (jump_n, b, laststart + 5,
2627 /* The location we want to set is the second
2628 parameter of the `jump_n'; that is `b-2' as
2629 an absolute address. `laststart' will be
2630 the `set_number_at' we're about to insert;
2631 `laststart+3' the number to set, the source
2632 for the relative address. But we are
2633 inserting into the middle of the pattern --
2634 so everything is getting moved up by 5.
2635 Conclusion: (b - 2) - (laststart + 3) + 5,
2636 i.e., b - laststart.
2638 We insert this at the beginning of the loop
2639 so that if we fail during matching, we'll
2640 reinitialize the bounds. */
2641 insert_op2 (set_number_at, laststart, b - laststart,
2642 upper_bound - 1, b);
2647 beg_interval = NULL;
2652 /* If an invalid interval, match the characters as literals. */
2653 assert (beg_interval);
2655 beg_interval = NULL;
2657 /* normal_char and normal_backslash need `c'. */
2660 if (!(syntax & RE_NO_BK_BRACES))
2662 if (p > pattern && p[-1] == '\\')
2663 goto normal_backslash;
2668 /* There is no way to specify the before_dot and after_dot
2669 operators. rms says this is ok. --karl */
2677 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2683 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2689 if (re_syntax_options & RE_NO_GNU_OPS)
2692 BUF_PUSH (wordchar);
2697 if (re_syntax_options & RE_NO_GNU_OPS)
2700 BUF_PUSH (notwordchar);
2705 if (re_syntax_options & RE_NO_GNU_OPS)
2711 if (re_syntax_options & RE_NO_GNU_OPS)
2717 if (re_syntax_options & RE_NO_GNU_OPS)
2719 BUF_PUSH (wordbound);
2723 if (re_syntax_options & RE_NO_GNU_OPS)
2725 BUF_PUSH (notwordbound);
2729 if (re_syntax_options & RE_NO_GNU_OPS)
2735 if (re_syntax_options & RE_NO_GNU_OPS)
2740 case '1': case '2': case '3': case '4': case '5':
2741 case '6': case '7': case '8': case '9':
2742 if (syntax & RE_NO_BK_REFS)
2748 FREE_STACK_RETURN (REG_ESUBREG);
2750 /* Can't back reference to a subexpression if inside of it. */
2751 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
2755 BUF_PUSH_2 (duplicate, c1);
2761 if (syntax & RE_BK_PLUS_QM)
2764 goto normal_backslash;
2768 /* You might think it would be useful for \ to mean
2769 not to translate; but if we don't translate it
2770 it will never match anything. */
2778 /* Expects the character in `c'. */
2780 /* If no exactn currently being built. */
2783 /* If last exactn not at current position. */
2784 || pending_exact + *pending_exact + 1 != b
2786 /* We have only one byte following the exactn for the count. */
2787 || *pending_exact == (1 << BYTEWIDTH) - 1
2789 /* If followed by a repetition operator. */
2790 || *p == '*' || *p == '^'
2791 || ((syntax & RE_BK_PLUS_QM)
2792 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2793 : (*p == '+' || *p == '?'))
2794 || ((syntax & RE_INTERVALS)
2795 && ((syntax & RE_NO_BK_BRACES)
2797 : (p[0] == '\\' && p[1] == '{'))))
2799 /* Start building a new exactn. */
2803 BUF_PUSH_2 (exactn, 0);
2804 pending_exact = b - 1;
2811 } /* while p != pend */
2814 /* Through the pattern now. */
2817 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2819 if (!COMPILE_STACK_EMPTY)
2820 FREE_STACK_RETURN (REG_EPAREN);
2822 /* If we don't want backtracking, force success
2823 the first time we reach the end of the compiled pattern. */
2824 if (syntax & RE_NO_POSIX_BACKTRACKING)
2827 free (compile_stack.stack);
2829 /* We have succeeded; set the length of the buffer. */
2830 bufp->used = b - bufp->buffer;
2835 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2836 print_compiled_pattern (bufp);
2840 #ifndef MATCH_MAY_ALLOCATE
2841 /* Initialize the failure stack to the largest possible stack. This
2842 isn't necessary unless we're trying to avoid calling alloca in
2843 the search and match routines. */
2845 int num_regs = bufp->re_nsub + 1;
2847 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2848 is strictly greater than re_max_failures, the largest possible stack
2849 is 2 * re_max_failures failure points. */
2850 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2852 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2855 if (! fail_stack.stack)
2857 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2858 * sizeof (fail_stack_elt_t));
2861 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2863 * sizeof (fail_stack_elt_t)));
2864 # else /* not emacs */
2865 if (! fail_stack.stack)
2867 = (fail_stack_elt_t *) malloc (fail_stack.size
2868 * sizeof (fail_stack_elt_t));
2871 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2873 * sizeof (fail_stack_elt_t)));
2874 # endif /* not emacs */
2877 regex_grow_registers (num_regs);
2879 #endif /* not MATCH_MAY_ALLOCATE */
2882 } /* regex_compile */
2884 /* Subroutines for `regex_compile'. */
2886 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2889 store_op1 (op, loc, arg)
2894 *loc = (unsigned char) op;
2895 STORE_NUMBER (loc + 1, arg);
2899 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2902 store_op2 (op, loc, arg1, arg2)
2907 *loc = (unsigned char) op;
2908 STORE_NUMBER (loc + 1, arg1);
2909 STORE_NUMBER (loc + 3, arg2);
2913 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
2914 for OP followed by two-byte integer parameter ARG. */
2917 insert_op1 (op, loc, arg, end)
2923 register unsigned char *pfrom = end;
2924 register unsigned char *pto = end + 3;
2926 while (pfrom != loc)
2929 store_op1 (op, loc, arg);
2933 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2936 insert_op2 (op, loc, arg1, arg2, end)
2942 register unsigned char *pfrom = end;
2943 register unsigned char *pto = end + 5;
2945 while (pfrom != loc)
2948 store_op2 (op, loc, arg1, arg2);
2952 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
2953 after an alternative or a begin-subexpression. We assume there is at
2954 least one character before the ^. */
2957 at_begline_loc_p (pattern, p, syntax)
2958 const char *pattern, *p;
2959 reg_syntax_t syntax;
2961 const char *prev = p - 2;
2962 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
2965 /* After a subexpression? */
2966 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
2967 /* After an alternative? */
2968 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
2972 /* The dual of at_begline_loc_p. This one is for $. We assume there is
2973 at least one character after the $, i.e., `P < PEND'. */
2976 at_endline_loc_p (p, pend, syntax)
2977 const char *p, *pend;
2978 reg_syntax_t syntax;
2980 const char *next = p;
2981 boolean next_backslash = *next == '\\';
2982 const char *next_next = p + 1 < pend ? p + 1 : 0;
2985 /* Before a subexpression? */
2986 (syntax & RE_NO_BK_PARENS ? *next == ')'
2987 : next_backslash && next_next && *next_next == ')')
2988 /* Before an alternative? */
2989 || (syntax & RE_NO_BK_VBAR ? *next == '|'
2990 : next_backslash && next_next && *next_next == '|');
2994 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
2995 false if it's not. */
2998 group_in_compile_stack (compile_stack, regnum)
2999 compile_stack_type compile_stack;
3004 for (this_element = compile_stack.avail - 1;
3007 if (compile_stack.stack[this_element].regnum == regnum)
3014 /* Read the ending character of a range (in a bracket expression) from the
3015 uncompiled pattern *P_PTR (which ends at PEND). We assume the
3016 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
3017 Then we set the translation of all bits between the starting and
3018 ending characters (inclusive) in the compiled pattern B.
3020 Return an error code.
3022 We use these short variable names so we can use the same macros as
3023 `regex_compile' itself. */
3025 static reg_errcode_t
3026 compile_range (p_ptr, pend, translate, syntax, b)
3027 const char **p_ptr, *pend;
3028 RE_TRANSLATE_TYPE translate;
3029 reg_syntax_t syntax;
3034 const char *p = *p_ptr;
3035 unsigned int range_start, range_end;
3040 /* Even though the pattern is a signed `char *', we need to fetch
3041 with unsigned char *'s; if the high bit of the pattern character
3042 is set, the range endpoints will be negative if we fetch using a
3045 We also want to fetch the endpoints without translating them; the
3046 appropriate translation is done in the bit-setting loop below. */
3047 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
3048 range_start = ((const unsigned char *) p)[-2];
3049 range_end = ((const unsigned char *) p)[0];
3051 /* Have to increment the pointer into the pattern string, so the
3052 caller isn't still at the ending character. */
3055 /* If the start is after the end, the range is empty. */
3056 if (range_start > range_end)
3057 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
3059 /* Here we see why `this_char' has to be larger than an `unsigned
3060 char' -- the range is inclusive, so if `range_end' == 0xff
3061 (assuming 8-bit characters), we would otherwise go into an infinite
3062 loop, since all characters <= 0xff. */
3063 for (this_char = range_start; this_char <= range_end; this_char++)
3065 SET_LIST_BIT (TRANSLATE (this_char));
3071 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3072 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3073 characters can start a string that matches the pattern. This fastmap
3074 is used by re_search to skip quickly over impossible starting points.
3076 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3077 area as BUFP->fastmap.
3079 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3082 Returns 0 if we succeed, -2 if an internal error. */
3085 re_compile_fastmap (bufp)
3086 struct re_pattern_buffer *bufp;
3089 #ifdef MATCH_MAY_ALLOCATE
3090 fail_stack_type fail_stack;
3092 #ifndef REGEX_MALLOC
3096 register char *fastmap = bufp->fastmap;
3097 unsigned char *pattern = bufp->buffer;
3098 unsigned char *p = pattern;
3099 register unsigned char *pend = pattern + bufp->used;
3102 /* This holds the pointer to the failure stack, when
3103 it is allocated relocatably. */
3104 fail_stack_elt_t *failure_stack_ptr;
3107 /* Assume that each path through the pattern can be null until
3108 proven otherwise. We set this false at the bottom of switch
3109 statement, to which we get only if a particular path doesn't
3110 match the empty string. */
3111 boolean path_can_be_null = true;
3113 /* We aren't doing a `succeed_n' to begin with. */
3114 boolean succeed_n_p = false;
3116 assert (fastmap != NULL && p != NULL);
3119 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
3120 bufp->fastmap_accurate = 1; /* It will be when we're done. */
3121 bufp->can_be_null = 0;
3125 if (p == pend || *p == succeed)
3127 /* We have reached the (effective) end of pattern. */
3128 if (!FAIL_STACK_EMPTY ())
3130 bufp->can_be_null |= path_can_be_null;
3132 /* Reset for next path. */
3133 path_can_be_null = true;
3135 p = fail_stack.stack[--fail_stack.avail].pointer;
3143 /* We should never be about to go beyond the end of the pattern. */
3146 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3149 /* I guess the idea here is to simply not bother with a fastmap
3150 if a backreference is used, since it's too hard to figure out
3151 the fastmap for the corresponding group. Setting
3152 `can_be_null' stops `re_search_2' from using the fastmap, so
3153 that is all we do. */
3155 bufp->can_be_null = 1;
3159 /* Following are the cases which match a character. These end
3168 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3169 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3175 /* Chars beyond end of map must be allowed. */
3176 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3179 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3180 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3186 for (j = 0; j < (1 << BYTEWIDTH); j++)
3187 if (SYNTAX (j) == Sword)
3193 for (j = 0; j < (1 << BYTEWIDTH); j++)
3194 if (SYNTAX (j) != Sword)
3201 int fastmap_newline = fastmap['\n'];
3203 /* `.' matches anything ... */
3204 for (j = 0; j < (1 << BYTEWIDTH); j++)
3207 /* ... except perhaps newline. */
3208 if (!(bufp->syntax & RE_DOT_NEWLINE))
3209 fastmap['\n'] = fastmap_newline;
3211 /* Return if we have already set `can_be_null'; if we have,
3212 then the fastmap is irrelevant. Something's wrong here. */
3213 else if (bufp->can_be_null)
3216 /* Otherwise, have to check alternative paths. */
3223 for (j = 0; j < (1 << BYTEWIDTH); j++)
3224 if (SYNTAX (j) == (enum syntaxcode) k)
3231 for (j = 0; j < (1 << BYTEWIDTH); j++)
3232 if (SYNTAX (j) != (enum syntaxcode) k)
3237 /* All cases after this match the empty string. These end with
3257 case push_dummy_failure:
3262 case pop_failure_jump:
3263 case maybe_pop_jump:
3266 case dummy_failure_jump:
3267 EXTRACT_NUMBER_AND_INCR (j, p);
3272 /* Jump backward implies we just went through the body of a
3273 loop and matched nothing. Opcode jumped to should be
3274 `on_failure_jump' or `succeed_n'. Just treat it like an
3275 ordinary jump. For a * loop, it has pushed its failure
3276 point already; if so, discard that as redundant. */
3277 if ((re_opcode_t) *p != on_failure_jump
3278 && (re_opcode_t) *p != succeed_n)
3282 EXTRACT_NUMBER_AND_INCR (j, p);
3285 /* If what's on the stack is where we are now, pop it. */
3286 if (!FAIL_STACK_EMPTY ()
3287 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3293 case on_failure_jump:
3294 case on_failure_keep_string_jump:
3295 handle_on_failure_jump:
3296 EXTRACT_NUMBER_AND_INCR (j, p);
3298 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3299 end of the pattern. We don't want to push such a point,
3300 since when we restore it above, entering the switch will
3301 increment `p' past the end of the pattern. We don't need
3302 to push such a point since we obviously won't find any more
3303 fastmap entries beyond `pend'. Such a pattern can match
3304 the null string, though. */
3307 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3309 RESET_FAIL_STACK ();
3314 bufp->can_be_null = 1;
3318 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3319 succeed_n_p = false;
3326 /* Get to the number of times to succeed. */
3329 /* Increment p past the n for when k != 0. */
3330 EXTRACT_NUMBER_AND_INCR (k, p);
3334 succeed_n_p = true; /* Spaghetti code alert. */
3335 goto handle_on_failure_jump;
3352 abort (); /* We have listed all the cases. */
3355 /* Getting here means we have found the possible starting
3356 characters for one path of the pattern -- and that the empty
3357 string does not match. We need not follow this path further.
3358 Instead, look at the next alternative (remembered on the
3359 stack), or quit if no more. The test at the top of the loop
3360 does these things. */
3361 path_can_be_null = false;
3365 /* Set `can_be_null' for the last path (also the first path, if the
3366 pattern is empty). */
3367 bufp->can_be_null |= path_can_be_null;
3370 RESET_FAIL_STACK ();
3372 } /* re_compile_fastmap */
3374 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3375 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3376 this memory for recording register information. STARTS and ENDS
3377 must be allocated using the malloc library routine, and must each
3378 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3380 If NUM_REGS == 0, then subsequent matches should allocate their own
3383 Unless this function is called, the first search or match using
3384 PATTERN_BUFFER will allocate its own register data, without
3385 freeing the old data. */
3388 re_set_registers (bufp, regs, num_regs, starts, ends)
3389 struct re_pattern_buffer *bufp;
3390 struct re_registers *regs;
3392 regoff_t *starts, *ends;
3396 bufp->regs_allocated = REGS_REALLOCATE;
3397 regs->num_regs = num_regs;
3398 regs->start = starts;
3403 bufp->regs_allocated = REGS_UNALLOCATED;
3405 regs->start = regs->end = (regoff_t *) 0;
3409 /* Searching routines. */
3411 /* Like re_search_2, below, but only one string is specified, and
3412 doesn't let you say where to stop matching. */
3415 re_search (bufp, string, size, startpos, range, regs)
3416 struct re_pattern_buffer *bufp;
3418 int size, startpos, range;
3419 struct re_registers *regs;
3421 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3426 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3427 virtual concatenation of STRING1 and STRING2, starting first at index
3428 STARTPOS, then at STARTPOS + 1, and so on.
3430 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3432 RANGE is how far to scan while trying to match. RANGE = 0 means try
3433 only at STARTPOS; in general, the last start tried is STARTPOS +
3436 In REGS, return the indices of the virtual concatenation of STRING1
3437 and STRING2 that matched the entire BUFP->buffer and its contained
3440 Do not consider matching one past the index STOP in the virtual
3441 concatenation of STRING1 and STRING2.
3443 We return either the position in the strings at which the match was
3444 found, -1 if no match, or -2 if error (such as failure
3448 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3449 struct re_pattern_buffer *bufp;
3450 const char *string1, *string2;
3454 struct re_registers *regs;
3458 register char *fastmap = bufp->fastmap;
3459 register RE_TRANSLATE_TYPE translate = bufp->translate;
3460 int total_size = size1 + size2;
3461 int endpos = startpos + range;
3463 /* Check for out-of-range STARTPOS. */
3464 if (startpos < 0 || startpos > total_size)
3467 /* Fix up RANGE if it might eventually take us outside
3468 the virtual concatenation of STRING1 and STRING2.
3469 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3471 range = 0 - startpos;
3472 else if (endpos > total_size)
3473 range = total_size - startpos;
3475 /* If the search isn't to be a backwards one, don't waste time in a
3476 search for a pattern that must be anchored. */
3477 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3486 /* In a forward search for something that starts with \=.
3487 don't keep searching past point. */
3488 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3490 range = PT - startpos;
3496 /* Update the fastmap now if not correct already. */
3497 if (fastmap && !bufp->fastmap_accurate)
3498 if (re_compile_fastmap (bufp) == -2)
3501 /* Loop through the string, looking for a place to start matching. */
3504 /* If a fastmap is supplied, skip quickly over characters that
3505 cannot be the start of a match. If the pattern can match the
3506 null string, however, we don't need to skip characters; we want
3507 the first null string. */
3508 if (fastmap && startpos < total_size && !bufp->can_be_null)
3510 if (range > 0) /* Searching forwards. */
3512 register const char *d;
3513 register int lim = 0;
3516 if (startpos < size1 && startpos + range >= size1)
3517 lim = range - (size1 - startpos);
3519 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3521 /* Written out as an if-else to avoid testing `translate'
3525 && !fastmap[(unsigned char)
3526 translate[(unsigned char) *d++]])
3529 while (range > lim && !fastmap[(unsigned char) *d++])
3532 startpos += irange - range;
3534 else /* Searching backwards. */
3536 register char c = (size1 == 0 || startpos >= size1
3537 ? string2[startpos - size1]
3538 : string1[startpos]);
3540 if (!fastmap[(unsigned char) TRANSLATE (c)])
3545 /* If can't match the null string, and that's all we have left, fail. */
3546 if (range >= 0 && startpos == total_size && fastmap
3547 && !bufp->can_be_null)
3550 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3551 startpos, regs, stop);
3552 #ifndef REGEX_MALLOC
3581 /* This converts PTR, a pointer into one of the search strings `string1'
3582 and `string2' into an offset from the beginning of that string. */
3583 #define POINTER_TO_OFFSET(ptr) \
3584 (FIRST_STRING_P (ptr) \
3585 ? ((regoff_t) ((ptr) - string1)) \
3586 : ((regoff_t) ((ptr) - string2 + size1)))
3588 /* Macros for dealing with the split strings in re_match_2. */
3590 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3592 /* Call before fetching a character with *d. This switches over to
3593 string2 if necessary. */
3594 #define PREFETCH() \
3597 /* End of string2 => fail. */ \
3598 if (dend == end_match_2) \
3600 /* End of string1 => advance to string2. */ \
3602 dend = end_match_2; \
3606 /* Test if at very beginning or at very end of the virtual concatenation
3607 of `string1' and `string2'. If only one string, it's `string2'. */
3608 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3609 #define AT_STRINGS_END(d) ((d) == end2)
3612 /* Test if D points to a character which is word-constituent. We have
3613 two special cases to check for: if past the end of string1, look at
3614 the first character in string2; and if before the beginning of
3615 string2, look at the last character in string1. */
3616 #define WORDCHAR_P(d) \
3617 (SYNTAX ((d) == end1 ? *string2 \
3618 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3621 /* Disabled due to a compiler bug -- see comment at case wordbound */
3623 /* Test if the character before D and the one at D differ with respect
3624 to being word-constituent. */
3625 #define AT_WORD_BOUNDARY(d) \
3626 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3627 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3630 /* Free everything we malloc. */
3631 #ifdef MATCH_MAY_ALLOCATE
3632 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3633 # define FREE_VARIABLES() \
3635 REGEX_FREE_STACK (fail_stack.stack); \
3636 FREE_VAR (regstart); \
3637 FREE_VAR (regend); \
3638 FREE_VAR (old_regstart); \
3639 FREE_VAR (old_regend); \
3640 FREE_VAR (best_regstart); \
3641 FREE_VAR (best_regend); \
3642 FREE_VAR (reg_info); \
3643 FREE_VAR (reg_dummy); \
3644 FREE_VAR (reg_info_dummy); \
3647 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3648 #endif /* not MATCH_MAY_ALLOCATE */
3650 /* These values must meet several constraints. They must not be valid
3651 register values; since we have a limit of 255 registers (because
3652 we use only one byte in the pattern for the register number), we can
3653 use numbers larger than 255. They must differ by 1, because of
3654 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3655 be larger than the value for the highest register, so we do not try
3656 to actually save any registers when none are active. */
3657 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3658 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3660 /* Matching routines. */
3662 #ifndef emacs /* Emacs never uses this. */
3663 /* re_match is like re_match_2 except it takes only a single string. */
3666 re_match (bufp, string, size, pos, regs)
3667 struct re_pattern_buffer *bufp;
3670 struct re_registers *regs;
3672 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3674 # ifndef REGEX_MALLOC
3681 #endif /* not emacs */
3683 static boolean group_match_null_string_p _RE_ARGS ((unsigned char **p,
3685 register_info_type *reg_info));
3686 static boolean alt_match_null_string_p _RE_ARGS ((unsigned char *p,
3688 register_info_type *reg_info));
3689 static boolean common_op_match_null_string_p _RE_ARGS ((unsigned char **p,
3691 register_info_type *reg_info));
3692 static int bcmp_translate _RE_ARGS ((const char *s1, const char *s2,
3693 int len, char *translate));
3695 /* re_match_2 matches the compiled pattern in BUFP against the
3696 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3697 and SIZE2, respectively). We start matching at POS, and stop
3700 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3701 store offsets for the substring each group matched in REGS. See the
3702 documentation for exactly how many groups we fill.
3704 We return -1 if no match, -2 if an internal error (such as the
3705 failure stack overflowing). Otherwise, we return the length of the
3706 matched substring. */
3709 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3710 struct re_pattern_buffer *bufp;
3711 const char *string1, *string2;
3714 struct re_registers *regs;
3717 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3719 #ifndef REGEX_MALLOC
3727 /* This is a separate function so that we can force an alloca cleanup
3730 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3731 struct re_pattern_buffer *bufp;
3732 const char *string1, *string2;
3735 struct re_registers *regs;
3738 /* General temporaries. */
3742 /* Just past the end of the corresponding string. */
3743 const char *end1, *end2;
3745 /* Pointers into string1 and string2, just past the last characters in
3746 each to consider matching. */
3747 const char *end_match_1, *end_match_2;
3749 /* Where we are in the data, and the end of the current string. */
3750 const char *d, *dend;
3752 /* Where we are in the pattern, and the end of the pattern. */
3753 unsigned char *p = bufp->buffer;
3754 register unsigned char *pend = p + bufp->used;
3756 /* Mark the opcode just after a start_memory, so we can test for an
3757 empty subpattern when we get to the stop_memory. */
3758 unsigned char *just_past_start_mem = 0;
3760 /* We use this to map every character in the string. */
3761 RE_TRANSLATE_TYPE translate = bufp->translate;
3763 /* Failure point stack. Each place that can handle a failure further
3764 down the line pushes a failure point on this stack. It consists of
3765 restart, regend, and reg_info for all registers corresponding to
3766 the subexpressions we're currently inside, plus the number of such
3767 registers, and, finally, two char *'s. The first char * is where
3768 to resume scanning the pattern; the second one is where to resume
3769 scanning the strings. If the latter is zero, the failure point is
3770 a ``dummy''; if a failure happens and the failure point is a dummy,
3771 it gets discarded and the next next one is tried. */
3772 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3773 fail_stack_type fail_stack;
3776 static unsigned failure_id = 0;
3777 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3781 /* This holds the pointer to the failure stack, when
3782 it is allocated relocatably. */
3783 fail_stack_elt_t *failure_stack_ptr;
3786 /* We fill all the registers internally, independent of what we
3787 return, for use in backreferences. The number here includes
3788 an element for register zero. */
3789 size_t num_regs = bufp->re_nsub + 1;
3791 /* The currently active registers. */
3792 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3793 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3795 /* Information on the contents of registers. These are pointers into
3796 the input strings; they record just what was matched (on this
3797 attempt) by a subexpression part of the pattern, that is, the
3798 regnum-th regstart pointer points to where in the pattern we began
3799 matching and the regnum-th regend points to right after where we
3800 stopped matching the regnum-th subexpression. (The zeroth register
3801 keeps track of what the whole pattern matches.) */
3802 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3803 const char **regstart, **regend;
3806 /* If a group that's operated upon by a repetition operator fails to
3807 match anything, then the register for its start will need to be
3808 restored because it will have been set to wherever in the string we
3809 are when we last see its open-group operator. Similarly for a
3811 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3812 const char **old_regstart, **old_regend;
3815 /* The is_active field of reg_info helps us keep track of which (possibly
3816 nested) subexpressions we are currently in. The matched_something
3817 field of reg_info[reg_num] helps us tell whether or not we have
3818 matched any of the pattern so far this time through the reg_num-th
3819 subexpression. These two fields get reset each time through any
3820 loop their register is in. */
3821 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3822 register_info_type *reg_info;
3825 /* The following record the register info as found in the above
3826 variables when we find a match better than any we've seen before.
3827 This happens as we backtrack through the failure points, which in
3828 turn happens only if we have not yet matched the entire string. */
3829 unsigned best_regs_set = false;
3830 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3831 const char **best_regstart, **best_regend;
3834 /* Logically, this is `best_regend[0]'. But we don't want to have to
3835 allocate space for that if we're not allocating space for anything
3836 else (see below). Also, we never need info about register 0 for
3837 any of the other register vectors, and it seems rather a kludge to
3838 treat `best_regend' differently than the rest. So we keep track of
3839 the end of the best match so far in a separate variable. We
3840 initialize this to NULL so that when we backtrack the first time
3841 and need to test it, it's not garbage. */
3842 const char *match_end = NULL;
3844 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3845 int set_regs_matched_done = 0;
3847 /* Used when we pop values we don't care about. */
3848 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3849 const char **reg_dummy;
3850 register_info_type *reg_info_dummy;
3854 /* Counts the total number of registers pushed. */
3855 unsigned num_regs_pushed = 0;
3858 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3862 #ifdef MATCH_MAY_ALLOCATE
3863 /* Do not bother to initialize all the register variables if there are
3864 no groups in the pattern, as it takes a fair amount of time. If
3865 there are groups, we include space for register 0 (the whole
3866 pattern), even though we never use it, since it simplifies the
3867 array indexing. We should fix this. */
3870 regstart = REGEX_TALLOC (num_regs, const char *);
3871 regend = REGEX_TALLOC (num_regs, const char *);
3872 old_regstart = REGEX_TALLOC (num_regs, const char *);
3873 old_regend = REGEX_TALLOC (num_regs, const char *);
3874 best_regstart = REGEX_TALLOC (num_regs, const char *);
3875 best_regend = REGEX_TALLOC (num_regs, const char *);
3876 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3877 reg_dummy = REGEX_TALLOC (num_regs, const char *);
3878 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3880 if (!(regstart && regend && old_regstart && old_regend && reg_info
3881 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3889 /* We must initialize all our variables to NULL, so that
3890 `FREE_VARIABLES' doesn't try to free them. */
3891 regstart = regend = old_regstart = old_regend = best_regstart
3892 = best_regend = reg_dummy = NULL;
3893 reg_info = reg_info_dummy = (register_info_type *) NULL;
3895 #endif /* MATCH_MAY_ALLOCATE */
3897 /* The starting position is bogus. */
3898 if (pos < 0 || pos > size1 + size2)
3904 /* Initialize subexpression text positions to -1 to mark ones that no
3905 start_memory/stop_memory has been seen for. Also initialize the
3906 register information struct. */
3907 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
3909 regstart[mcnt] = regend[mcnt]
3910 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3912 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3913 IS_ACTIVE (reg_info[mcnt]) = 0;
3914 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3915 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3918 /* We move `string1' into `string2' if the latter's empty -- but not if
3919 `string1' is null. */
3920 if (size2 == 0 && string1 != NULL)
3927 end1 = string1 + size1;
3928 end2 = string2 + size2;
3930 /* Compute where to stop matching, within the two strings. */
3933 end_match_1 = string1 + stop;
3934 end_match_2 = string2;
3939 end_match_2 = string2 + stop - size1;
3942 /* `p' scans through the pattern as `d' scans through the data.
3943 `dend' is the end of the input string that `d' points within. `d'
3944 is advanced into the following input string whenever necessary, but
3945 this happens before fetching; therefore, at the beginning of the
3946 loop, `d' can be pointing at the end of a string, but it cannot
3948 if (size1 > 0 && pos <= size1)
3955 d = string2 + pos - size1;
3959 DEBUG_PRINT1 ("The compiled pattern is:\n");
3960 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
3961 DEBUG_PRINT1 ("The string to match is: `");
3962 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
3963 DEBUG_PRINT1 ("'\n");
3965 /* This loops over pattern commands. It exits by returning from the
3966 function if the match is complete, or it drops through if the match
3967 fails at this starting point in the input data. */
3971 DEBUG_PRINT2 ("\n%p: ", p);
3973 DEBUG_PRINT2 ("\n0x%x: ", p);
3977 { /* End of pattern means we might have succeeded. */
3978 DEBUG_PRINT1 ("end of pattern ... ");
3980 /* If we haven't matched the entire string, and we want the
3981 longest match, try backtracking. */
3982 if (d != end_match_2)
3984 /* 1 if this match ends in the same string (string1 or string2)
3985 as the best previous match. */
3986 boolean same_str_p = (FIRST_STRING_P (match_end)
3987 == MATCHING_IN_FIRST_STRING);
3988 /* 1 if this match is the best seen so far. */
3989 boolean best_match_p;
3991 /* AIX compiler got confused when this was combined
3992 with the previous declaration. */
3994 best_match_p = d > match_end;
3996 best_match_p = !MATCHING_IN_FIRST_STRING;
3998 DEBUG_PRINT1 ("backtracking.\n");
4000 if (!FAIL_STACK_EMPTY ())
4001 { /* More failure points to try. */
4003 /* If exceeds best match so far, save it. */
4004 if (!best_regs_set || best_match_p)
4006 best_regs_set = true;
4009 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
4011 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4013 best_regstart[mcnt] = regstart[mcnt];
4014 best_regend[mcnt] = regend[mcnt];
4020 /* If no failure points, don't restore garbage. And if
4021 last match is real best match, don't restore second
4023 else if (best_regs_set && !best_match_p)
4026 /* Restore best match. It may happen that `dend ==
4027 end_match_1' while the restored d is in string2.
4028 For example, the pattern `x.*y.*z' against the
4029 strings `x-' and `y-z-', if the two strings are
4030 not consecutive in memory. */
4031 DEBUG_PRINT1 ("Restoring best registers.\n");
4034 dend = ((d >= string1 && d <= end1)
4035 ? end_match_1 : end_match_2);
4037 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4039 regstart[mcnt] = best_regstart[mcnt];
4040 regend[mcnt] = best_regend[mcnt];
4043 } /* d != end_match_2 */
4046 DEBUG_PRINT1 ("Accepting match.\n");
4048 /* If caller wants register contents data back, do it. */
4049 if (regs && !bufp->no_sub)
4051 /* Have the register data arrays been allocated? */
4052 if (bufp->regs_allocated == REGS_UNALLOCATED)
4053 { /* No. So allocate them with malloc. We need one
4054 extra element beyond `num_regs' for the `-1' marker
4056 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
4057 regs->start = TALLOC (regs->num_regs, regoff_t);
4058 regs->end = TALLOC (regs->num_regs, regoff_t);
4059 if (regs->start == NULL || regs->end == NULL)
4064 bufp->regs_allocated = REGS_REALLOCATE;
4066 else if (bufp->regs_allocated == REGS_REALLOCATE)
4067 { /* Yes. If we need more elements than were already
4068 allocated, reallocate them. If we need fewer, just
4070 if (regs->num_regs < num_regs + 1)
4072 regs->num_regs = num_regs + 1;
4073 RETALLOC (regs->start, regs->num_regs, regoff_t);
4074 RETALLOC (regs->end, regs->num_regs, regoff_t);
4075 if (regs->start == NULL || regs->end == NULL)
4084 /* These braces fend off a "empty body in an else-statement"
4085 warning under GCC when assert expands to nothing. */
4086 assert (bufp->regs_allocated == REGS_FIXED);
4089 /* Convert the pointer data in `regstart' and `regend' to
4090 indices. Register zero has to be set differently,
4091 since we haven't kept track of any info for it. */
4092 if (regs->num_regs > 0)
4094 regs->start[0] = pos;
4095 regs->end[0] = (MATCHING_IN_FIRST_STRING
4096 ? ((regoff_t) (d - string1))
4097 : ((regoff_t) (d - string2 + size1)));
4100 /* Go through the first `min (num_regs, regs->num_regs)'
4101 registers, since that is all we initialized. */
4102 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
4105 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
4106 regs->start[mcnt] = regs->end[mcnt] = -1;
4110 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
4112 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
4116 /* If the regs structure we return has more elements than
4117 were in the pattern, set the extra elements to -1. If
4118 we (re)allocated the registers, this is the case,
4119 because we always allocate enough to have at least one
4121 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
4122 regs->start[mcnt] = regs->end[mcnt] = -1;
4123 } /* regs && !bufp->no_sub */
4125 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
4126 nfailure_points_pushed, nfailure_points_popped,
4127 nfailure_points_pushed - nfailure_points_popped);
4128 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
4130 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
4134 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
4140 /* Otherwise match next pattern command. */
4141 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4143 /* Ignore these. Used to ignore the n of succeed_n's which
4144 currently have n == 0. */
4146 DEBUG_PRINT1 ("EXECUTING no_op.\n");
4150 DEBUG_PRINT1 ("EXECUTING succeed.\n");
4153 /* Match the next n pattern characters exactly. The following
4154 byte in the pattern defines n, and the n bytes after that
4155 are the characters to match. */
4158 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
4160 /* This is written out as an if-else so we don't waste time
4161 testing `translate' inside the loop. */
4167 if ((unsigned char) translate[(unsigned char) *d++]
4168 != (unsigned char) *p++)
4178 if (*d++ != (char) *p++) goto fail;
4182 SET_REGS_MATCHED ();
4186 /* Match any character except possibly a newline or a null. */
4188 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4192 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
4193 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4196 SET_REGS_MATCHED ();
4197 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4205 register unsigned char c;
4206 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4208 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4211 c = TRANSLATE (*d); /* The character to match. */
4213 /* Cast to `unsigned' instead of `unsigned char' in case the
4214 bit list is a full 32 bytes long. */
4215 if (c < (unsigned) (*p * BYTEWIDTH)
4216 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4221 if (!not) goto fail;
4223 SET_REGS_MATCHED ();
4229 /* The beginning of a group is represented by start_memory.
4230 The arguments are the register number in the next byte, and the
4231 number of groups inner to this one in the next. The text
4232 matched within the group is recorded (in the internal
4233 registers data structure) under the register number. */
4235 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4237 /* Find out if this group can match the empty string. */
4238 p1 = p; /* To send to group_match_null_string_p. */
4240 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4241 REG_MATCH_NULL_STRING_P (reg_info[*p])
4242 = group_match_null_string_p (&p1, pend, reg_info);
4244 /* Save the position in the string where we were the last time
4245 we were at this open-group operator in case the group is
4246 operated upon by a repetition operator, e.g., with `(a*)*b'
4247 against `ab'; then we want to ignore where we are now in
4248 the string in case this attempt to match fails. */
4249 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4250 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4252 DEBUG_PRINT2 (" old_regstart: %d\n",
4253 POINTER_TO_OFFSET (old_regstart[*p]));
4256 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4258 IS_ACTIVE (reg_info[*p]) = 1;
4259 MATCHED_SOMETHING (reg_info[*p]) = 0;
4261 /* Clear this whenever we change the register activity status. */
4262 set_regs_matched_done = 0;
4264 /* This is the new highest active register. */
4265 highest_active_reg = *p;
4267 /* If nothing was active before, this is the new lowest active
4269 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4270 lowest_active_reg = *p;
4272 /* Move past the register number and inner group count. */
4274 just_past_start_mem = p;
4279 /* The stop_memory opcode represents the end of a group. Its
4280 arguments are the same as start_memory's: the register
4281 number, and the number of inner groups. */
4283 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4285 /* We need to save the string position the last time we were at
4286 this close-group operator in case the group is operated
4287 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4288 against `aba'; then we want to ignore where we are now in
4289 the string in case this attempt to match fails. */
4290 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4291 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4293 DEBUG_PRINT2 (" old_regend: %d\n",
4294 POINTER_TO_OFFSET (old_regend[*p]));
4297 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4299 /* This register isn't active anymore. */
4300 IS_ACTIVE (reg_info[*p]) = 0;
4302 /* Clear this whenever we change the register activity status. */
4303 set_regs_matched_done = 0;
4305 /* If this was the only register active, nothing is active
4307 if (lowest_active_reg == highest_active_reg)
4309 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4310 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4313 { /* We must scan for the new highest active register, since
4314 it isn't necessarily one less than now: consider
4315 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4316 new highest active register is 1. */
4317 unsigned char r = *p - 1;
4318 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4321 /* If we end up at register zero, that means that we saved
4322 the registers as the result of an `on_failure_jump', not
4323 a `start_memory', and we jumped to past the innermost
4324 `stop_memory'. For example, in ((.)*) we save
4325 registers 1 and 2 as a result of the *, but when we pop
4326 back to the second ), we are at the stop_memory 1.
4327 Thus, nothing is active. */
4330 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4331 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4334 highest_active_reg = r;
4337 /* If just failed to match something this time around with a
4338 group that's operated on by a repetition operator, try to
4339 force exit from the ``loop'', and restore the register
4340 information for this group that we had before trying this
4342 if ((!MATCHED_SOMETHING (reg_info[*p])
4343 || just_past_start_mem == p - 1)
4346 boolean is_a_jump_n = false;
4350 switch ((re_opcode_t) *p1++)
4354 case pop_failure_jump:
4355 case maybe_pop_jump:
4357 case dummy_failure_jump:
4358 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4368 /* If the next operation is a jump backwards in the pattern
4369 to an on_failure_jump right before the start_memory
4370 corresponding to this stop_memory, exit from the loop
4371 by forcing a failure after pushing on the stack the
4372 on_failure_jump's jump in the pattern, and d. */
4373 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4374 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4376 /* If this group ever matched anything, then restore
4377 what its registers were before trying this last
4378 failed match, e.g., with `(a*)*b' against `ab' for
4379 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4380 against `aba' for regend[3].
4382 Also restore the registers for inner groups for,
4383 e.g., `((a*)(b*))*' against `aba' (register 3 would
4384 otherwise get trashed). */
4386 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4390 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4392 /* Restore this and inner groups' (if any) registers. */
4393 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
4396 regstart[r] = old_regstart[r];
4398 /* xx why this test? */
4399 if (old_regend[r] >= regstart[r])
4400 regend[r] = old_regend[r];
4404 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4405 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4411 /* Move past the register number and the inner group count. */
4416 /* \<digit> has been turned into a `duplicate' command which is
4417 followed by the numeric value of <digit> as the register number. */
4420 register const char *d2, *dend2;
4421 int regno = *p++; /* Get which register to match against. */
4422 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4424 /* Can't back reference a group which we've never matched. */
4425 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4428 /* Where in input to try to start matching. */
4429 d2 = regstart[regno];
4431 /* Where to stop matching; if both the place to start and
4432 the place to stop matching are in the same string, then
4433 set to the place to stop, otherwise, for now have to use
4434 the end of the first string. */
4436 dend2 = ((FIRST_STRING_P (regstart[regno])
4437 == FIRST_STRING_P (regend[regno]))
4438 ? regend[regno] : end_match_1);
4441 /* If necessary, advance to next segment in register
4445 if (dend2 == end_match_2) break;
4446 if (dend2 == regend[regno]) break;
4448 /* End of string1 => advance to string2. */
4450 dend2 = regend[regno];
4452 /* At end of register contents => success */
4453 if (d2 == dend2) break;
4455 /* If necessary, advance to next segment in data. */
4458 /* How many characters left in this segment to match. */
4461 /* Want how many consecutive characters we can match in
4462 one shot, so, if necessary, adjust the count. */
4463 if (mcnt > dend2 - d2)
4466 /* Compare that many; failure if mismatch, else move
4469 ? bcmp_translate (d, d2, mcnt, translate)
4470 : memcmp (d, d2, mcnt))
4472 d += mcnt, d2 += mcnt;
4474 /* Do this because we've match some characters. */
4475 SET_REGS_MATCHED ();
4481 /* begline matches the empty string at the beginning of the string
4482 (unless `not_bol' is set in `bufp'), and, if
4483 `newline_anchor' is set, after newlines. */
4485 DEBUG_PRINT1 ("EXECUTING begline.\n");
4487 if (AT_STRINGS_BEG (d))
4489 if (!bufp->not_bol) break;
4491 else if (d[-1] == '\n' && bufp->newline_anchor)
4495 /* In all other cases, we fail. */
4499 /* endline is the dual of begline. */
4501 DEBUG_PRINT1 ("EXECUTING endline.\n");
4503 if (AT_STRINGS_END (d))
4505 if (!bufp->not_eol) break;
4508 /* We have to ``prefetch'' the next character. */
4509 else if ((d == end1 ? *string2 : *d) == '\n'
4510 && bufp->newline_anchor)
4517 /* Match at the very beginning of the data. */
4519 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4520 if (AT_STRINGS_BEG (d))
4525 /* Match at the very end of the data. */
4527 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4528 if (AT_STRINGS_END (d))
4533 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4534 pushes NULL as the value for the string on the stack. Then
4535 `pop_failure_point' will keep the current value for the
4536 string, instead of restoring it. To see why, consider
4537 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4538 then the . fails against the \n. But the next thing we want
4539 to do is match the \n against the \n; if we restored the
4540 string value, we would be back at the foo.
4542 Because this is used only in specific cases, we don't need to
4543 check all the things that `on_failure_jump' does, to make
4544 sure the right things get saved on the stack. Hence we don't
4545 share its code. The only reason to push anything on the
4546 stack at all is that otherwise we would have to change
4547 `anychar's code to do something besides goto fail in this
4548 case; that seems worse than this. */
4549 case on_failure_keep_string_jump:
4550 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4552 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4554 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
4556 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4559 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4563 /* Uses of on_failure_jump:
4565 Each alternative starts with an on_failure_jump that points
4566 to the beginning of the next alternative. Each alternative
4567 except the last ends with a jump that in effect jumps past
4568 the rest of the alternatives. (They really jump to the
4569 ending jump of the following alternative, because tensioning
4570 these jumps is a hassle.)
4572 Repeats start with an on_failure_jump that points past both
4573 the repetition text and either the following jump or
4574 pop_failure_jump back to this on_failure_jump. */
4575 case on_failure_jump:
4577 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4579 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4581 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
4583 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4586 /* If this on_failure_jump comes right before a group (i.e.,
4587 the original * applied to a group), save the information
4588 for that group and all inner ones, so that if we fail back
4589 to this point, the group's information will be correct.
4590 For example, in \(a*\)*\1, we need the preceding group,
4591 and in \(zz\(a*\)b*\)\2, we need the inner group. */
4593 /* We can't use `p' to check ahead because we push
4594 a failure point to `p + mcnt' after we do this. */
4597 /* We need to skip no_op's before we look for the
4598 start_memory in case this on_failure_jump is happening as
4599 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4601 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4604 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4606 /* We have a new highest active register now. This will
4607 get reset at the start_memory we are about to get to,
4608 but we will have saved all the registers relevant to
4609 this repetition op, as described above. */
4610 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4611 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4612 lowest_active_reg = *(p1 + 1);
4615 DEBUG_PRINT1 (":\n");
4616 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4620 /* A smart repeat ends with `maybe_pop_jump'.
4621 We change it to either `pop_failure_jump' or `jump'. */
4622 case maybe_pop_jump:
4623 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4624 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4626 register unsigned char *p2 = p;
4628 /* Compare the beginning of the repeat with what in the
4629 pattern follows its end. If we can establish that there
4630 is nothing that they would both match, i.e., that we
4631 would have to backtrack because of (as in, e.g., `a*a')
4632 then we can change to pop_failure_jump, because we'll
4633 never have to backtrack.
4635 This is not true in the case of alternatives: in
4636 `(a|ab)*' we do need to backtrack to the `ab' alternative
4637 (e.g., if the string was `ab'). But instead of trying to
4638 detect that here, the alternative has put on a dummy
4639 failure point which is what we will end up popping. */
4641 /* Skip over open/close-group commands.
4642 If what follows this loop is a ...+ construct,
4643 look at what begins its body, since we will have to
4644 match at least one of that. */
4648 && ((re_opcode_t) *p2 == stop_memory
4649 || (re_opcode_t) *p2 == start_memory))
4651 else if (p2 + 6 < pend
4652 && (re_opcode_t) *p2 == dummy_failure_jump)
4659 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4660 to the `maybe_finalize_jump' of this case. Examine what
4663 /* If we're at the end of the pattern, we can change. */
4666 /* Consider what happens when matching ":\(.*\)"
4667 against ":/". I don't really understand this code
4669 p[-3] = (unsigned char) pop_failure_jump;
4671 (" End of pattern: change to `pop_failure_jump'.\n");
4674 else if ((re_opcode_t) *p2 == exactn
4675 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4677 register unsigned char c
4678 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4680 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4682 p[-3] = (unsigned char) pop_failure_jump;
4683 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4687 else if ((re_opcode_t) p1[3] == charset
4688 || (re_opcode_t) p1[3] == charset_not)
4690 int not = (re_opcode_t) p1[3] == charset_not;
4692 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4693 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4696 /* `not' is equal to 1 if c would match, which means
4697 that we can't change to pop_failure_jump. */
4700 p[-3] = (unsigned char) pop_failure_jump;
4701 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4705 else if ((re_opcode_t) *p2 == charset)
4708 register unsigned char c
4709 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4713 if ((re_opcode_t) p1[3] == exactn
4714 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
4715 && (p2[2 + p1[5] / BYTEWIDTH]
4716 & (1 << (p1[5] % BYTEWIDTH)))))
4718 if ((re_opcode_t) p1[3] == exactn
4719 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[4]
4720 && (p2[2 + p1[4] / BYTEWIDTH]
4721 & (1 << (p1[4] % BYTEWIDTH)))))
4724 p[-3] = (unsigned char) pop_failure_jump;
4725 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4729 else if ((re_opcode_t) p1[3] == charset_not)
4732 /* We win if the charset_not inside the loop
4733 lists every character listed in the charset after. */
4734 for (idx = 0; idx < (int) p2[1]; idx++)
4735 if (! (p2[2 + idx] == 0
4736 || (idx < (int) p1[4]
4737 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4742 p[-3] = (unsigned char) pop_failure_jump;
4743 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4746 else if ((re_opcode_t) p1[3] == charset)
4749 /* We win if the charset inside the loop
4750 has no overlap with the one after the loop. */
4752 idx < (int) p2[1] && idx < (int) p1[4];
4754 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4757 if (idx == p2[1] || idx == p1[4])
4759 p[-3] = (unsigned char) pop_failure_jump;
4760 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4765 p -= 2; /* Point at relative address again. */
4766 if ((re_opcode_t) p[-1] != pop_failure_jump)
4768 p[-1] = (unsigned char) jump;
4769 DEBUG_PRINT1 (" Match => jump.\n");
4770 goto unconditional_jump;
4772 /* Note fall through. */
4775 /* The end of a simple repeat has a pop_failure_jump back to
4776 its matching on_failure_jump, where the latter will push a
4777 failure point. The pop_failure_jump takes off failure
4778 points put on by this pop_failure_jump's matching
4779 on_failure_jump; we got through the pattern to here from the
4780 matching on_failure_jump, so didn't fail. */
4781 case pop_failure_jump:
4783 /* We need to pass separate storage for the lowest and
4784 highest registers, even though we don't care about the
4785 actual values. Otherwise, we will restore only one
4786 register from the stack, since lowest will == highest in
4787 `pop_failure_point'. */
4788 active_reg_t dummy_low_reg, dummy_high_reg;
4789 unsigned char *pdummy;
4792 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4793 POP_FAILURE_POINT (sdummy, pdummy,
4794 dummy_low_reg, dummy_high_reg,
4795 reg_dummy, reg_dummy, reg_info_dummy);
4797 /* Note fall through. */
4801 DEBUG_PRINT2 ("\n%p: ", p);
4803 DEBUG_PRINT2 ("\n0x%x: ", p);
4805 /* Note fall through. */
4807 /* Unconditionally jump (without popping any failure points). */
4809 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4810 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4811 p += mcnt; /* Do the jump. */
4813 DEBUG_PRINT2 ("(to %p).\n", p);
4815 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4820 /* We need this opcode so we can detect where alternatives end
4821 in `group_match_null_string_p' et al. */
4823 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4824 goto unconditional_jump;
4827 /* Normally, the on_failure_jump pushes a failure point, which
4828 then gets popped at pop_failure_jump. We will end up at
4829 pop_failure_jump, also, and with a pattern of, say, `a+', we
4830 are skipping over the on_failure_jump, so we have to push
4831 something meaningless for pop_failure_jump to pop. */
4832 case dummy_failure_jump:
4833 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4834 /* It doesn't matter what we push for the string here. What
4835 the code at `fail' tests is the value for the pattern. */
4836 PUSH_FAILURE_POINT (NULL, NULL, -2);
4837 goto unconditional_jump;
4840 /* At the end of an alternative, we need to push a dummy failure
4841 point in case we are followed by a `pop_failure_jump', because
4842 we don't want the failure point for the alternative to be
4843 popped. For example, matching `(a|ab)*' against `aab'
4844 requires that we match the `ab' alternative. */
4845 case push_dummy_failure:
4846 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4847 /* See comments just above at `dummy_failure_jump' about the
4849 PUSH_FAILURE_POINT (NULL, NULL, -2);
4852 /* Have to succeed matching what follows at least n times.
4853 After that, handle like `on_failure_jump'. */
4855 EXTRACT_NUMBER (mcnt, p + 2);
4856 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4859 /* Originally, this is how many times we HAVE to succeed. */
4864 STORE_NUMBER_AND_INCR (p, mcnt);
4866 DEBUG_PRINT3 (" Setting %p to %d.\n", p - 2, mcnt);
4868 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - 2, mcnt);
4874 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n", p+2);
4876 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4878 p[2] = (unsigned char) no_op;
4879 p[3] = (unsigned char) no_op;
4885 EXTRACT_NUMBER (mcnt, p + 2);
4886 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4888 /* Originally, this is how many times we CAN jump. */
4892 STORE_NUMBER (p + 2, mcnt);
4894 DEBUG_PRINT3 (" Setting %p to %d.\n", p + 2, mcnt);
4896 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + 2, mcnt);
4898 goto unconditional_jump;
4900 /* If don't have to jump any more, skip over the rest of command. */
4907 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4909 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4911 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4913 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
4915 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4917 STORE_NUMBER (p1, mcnt);
4922 /* The DEC Alpha C compiler 3.x generates incorrect code for the
4923 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4924 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4925 macro and introducing temporary variables works around the bug. */
4928 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4929 if (AT_WORD_BOUNDARY (d))
4934 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4935 if (AT_WORD_BOUNDARY (d))
4941 boolean prevchar, thischar;
4943 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4944 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4947 prevchar = WORDCHAR_P (d - 1);
4948 thischar = WORDCHAR_P (d);
4949 if (prevchar != thischar)
4956 boolean prevchar, thischar;
4958 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4959 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4962 prevchar = WORDCHAR_P (d - 1);
4963 thischar = WORDCHAR_P (d);
4964 if (prevchar != thischar)
4971 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
4972 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
4977 DEBUG_PRINT1 ("EXECUTING wordend.\n");
4978 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
4979 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
4985 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
4986 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
4991 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
4992 if (PTR_CHAR_POS ((unsigned char *) d) != point)
4997 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
4998 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
5003 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
5008 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
5012 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5014 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
5016 SET_REGS_MATCHED ();
5020 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
5022 goto matchnotsyntax;
5025 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
5029 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5031 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
5033 SET_REGS_MATCHED ();
5036 #else /* not emacs */
5038 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
5040 if (!WORDCHAR_P (d))
5042 SET_REGS_MATCHED ();
5047 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
5051 SET_REGS_MATCHED ();
5054 #endif /* not emacs */
5059 continue; /* Successfully executed one pattern command; keep going. */
5062 /* We goto here if a matching operation fails. */
5064 if (!FAIL_STACK_EMPTY ())
5065 { /* A restart point is known. Restore to that state. */
5066 DEBUG_PRINT1 ("\nFAIL:\n");
5067 POP_FAILURE_POINT (d, p,
5068 lowest_active_reg, highest_active_reg,
5069 regstart, regend, reg_info);
5071 /* If this failure point is a dummy, try the next one. */
5075 /* If we failed to the end of the pattern, don't examine *p. */
5079 boolean is_a_jump_n = false;
5081 /* If failed to a backwards jump that's part of a repetition
5082 loop, need to pop this failure point and use the next one. */
5083 switch ((re_opcode_t) *p)
5087 case maybe_pop_jump:
5088 case pop_failure_jump:
5091 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5094 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
5096 && (re_opcode_t) *p1 == on_failure_jump))
5104 if (d >= string1 && d <= end1)
5108 break; /* Matching at this starting point really fails. */
5112 goto restore_best_regs;
5116 return -1; /* Failure to match. */
5119 /* Subroutine definitions for re_match_2. */
5122 /* We are passed P pointing to a register number after a start_memory.
5124 Return true if the pattern up to the corresponding stop_memory can
5125 match the empty string, and false otherwise.
5127 If we find the matching stop_memory, sets P to point to one past its number.
5128 Otherwise, sets P to an undefined byte less than or equal to END.
5130 We don't handle duplicates properly (yet). */
5133 group_match_null_string_p (p, end, reg_info)
5134 unsigned char **p, *end;
5135 register_info_type *reg_info;
5138 /* Point to after the args to the start_memory. */
5139 unsigned char *p1 = *p + 2;
5143 /* Skip over opcodes that can match nothing, and return true or
5144 false, as appropriate, when we get to one that can't, or to the
5145 matching stop_memory. */
5147 switch ((re_opcode_t) *p1)
5149 /* Could be either a loop or a series of alternatives. */
5150 case on_failure_jump:
5152 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5154 /* If the next operation is not a jump backwards in the
5159 /* Go through the on_failure_jumps of the alternatives,
5160 seeing if any of the alternatives cannot match nothing.
5161 The last alternative starts with only a jump,
5162 whereas the rest start with on_failure_jump and end
5163 with a jump, e.g., here is the pattern for `a|b|c':
5165 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
5166 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
5169 So, we have to first go through the first (n-1)
5170 alternatives and then deal with the last one separately. */
5173 /* Deal with the first (n-1) alternatives, which start
5174 with an on_failure_jump (see above) that jumps to right
5175 past a jump_past_alt. */
5177 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
5179 /* `mcnt' holds how many bytes long the alternative
5180 is, including the ending `jump_past_alt' and
5183 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
5187 /* Move to right after this alternative, including the
5191 /* Break if it's the beginning of an n-th alternative
5192 that doesn't begin with an on_failure_jump. */
5193 if ((re_opcode_t) *p1 != on_failure_jump)
5196 /* Still have to check that it's not an n-th
5197 alternative that starts with an on_failure_jump. */
5199 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5200 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
5202 /* Get to the beginning of the n-th alternative. */
5208 /* Deal with the last alternative: go back and get number
5209 of the `jump_past_alt' just before it. `mcnt' contains
5210 the length of the alternative. */
5211 EXTRACT_NUMBER (mcnt, p1 - 2);
5213 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
5216 p1 += mcnt; /* Get past the n-th alternative. */
5222 assert (p1[1] == **p);
5228 if (!common_op_match_null_string_p (&p1, end, reg_info))
5231 } /* while p1 < end */
5234 } /* group_match_null_string_p */
5237 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5238 It expects P to be the first byte of a single alternative and END one
5239 byte past the last. The alternative can contain groups. */
5242 alt_match_null_string_p (p, end, reg_info)
5243 unsigned char *p, *end;
5244 register_info_type *reg_info;
5247 unsigned char *p1 = p;
5251 /* Skip over opcodes that can match nothing, and break when we get
5252 to one that can't. */
5254 switch ((re_opcode_t) *p1)
5257 case on_failure_jump:
5259 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5264 if (!common_op_match_null_string_p (&p1, end, reg_info))
5267 } /* while p1 < end */
5270 } /* alt_match_null_string_p */
5273 /* Deals with the ops common to group_match_null_string_p and
5274 alt_match_null_string_p.
5276 Sets P to one after the op and its arguments, if any. */
5279 common_op_match_null_string_p (p, end, reg_info)
5280 unsigned char **p, *end;
5281 register_info_type *reg_info;
5286 unsigned char *p1 = *p;
5288 switch ((re_opcode_t) *p1++)
5308 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5309 ret = group_match_null_string_p (&p1, end, reg_info);
5311 /* Have to set this here in case we're checking a group which
5312 contains a group and a back reference to it. */
5314 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5315 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5321 /* If this is an optimized succeed_n for zero times, make the jump. */
5323 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5331 /* Get to the number of times to succeed. */
5333 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5338 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5346 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5354 /* All other opcodes mean we cannot match the empty string. */
5360 } /* common_op_match_null_string_p */
5363 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5364 bytes; nonzero otherwise. */
5367 bcmp_translate (s1, s2, len, translate)
5368 const char *s1, *s2;
5370 RE_TRANSLATE_TYPE translate;
5372 register const unsigned char *p1 = (const unsigned char *) s1;
5373 register const unsigned char *p2 = (const unsigned char *) s2;
5376 if (translate[*p1++] != translate[*p2++]) return 1;
5382 /* Entry points for GNU code. */
5384 /* re_compile_pattern is the GNU regular expression compiler: it
5385 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5386 Returns 0 if the pattern was valid, otherwise an error string.
5388 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5389 are set in BUFP on entry.
5391 We call regex_compile to do the actual compilation. */
5394 re_compile_pattern (pattern, length, bufp)
5395 const char *pattern;
5397 struct re_pattern_buffer *bufp;
5401 /* GNU code is written to assume at least RE_NREGS registers will be set
5402 (and at least one extra will be -1). */
5403 bufp->regs_allocated = REGS_UNALLOCATED;
5405 /* And GNU code determines whether or not to get register information
5406 by passing null for the REGS argument to re_match, etc., not by
5410 /* Match anchors at newline. */
5411 bufp->newline_anchor = 1;
5413 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5417 return gettext (re_error_msgid[(int) ret]);
5420 /* Entry points compatible with 4.2 BSD regex library. We don't define
5421 them unless specifically requested. */
5423 #if defined _REGEX_RE_COMP || defined _LIBC
5425 /* BSD has one and only one pattern buffer. */
5426 static struct re_pattern_buffer re_comp_buf;
5430 /* Make these definitions weak in libc, so POSIX programs can redefine
5431 these names if they don't use our functions, and still use
5432 regcomp/regexec below without link errors. */
5442 if (!re_comp_buf.buffer)
5443 return gettext ("No previous regular expression");
5447 if (!re_comp_buf.buffer)
5449 re_comp_buf.buffer = (unsigned char *) malloc (200);
5450 if (re_comp_buf.buffer == NULL)
5451 return gettext (re_error_msgid[(int) REG_ESPACE]);
5452 re_comp_buf.allocated = 200;
5454 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5455 if (re_comp_buf.fastmap == NULL)
5456 return gettext (re_error_msgid[(int) REG_ESPACE]);
5459 /* Since `re_exec' always passes NULL for the `regs' argument, we
5460 don't need to initialize the pattern buffer fields which affect it. */
5462 /* Match anchors at newlines. */
5463 re_comp_buf.newline_anchor = 1;
5465 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5470 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5471 return (char *) gettext (re_error_msgid[(int) ret]);
5482 const int len = strlen (s);
5484 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5487 #endif /* _REGEX_RE_COMP */
5489 /* POSIX.2 functions. Don't define these for Emacs. */
5493 /* regcomp takes a regular expression as a string and compiles it.
5495 PREG is a regex_t *. We do not expect any fields to be initialized,
5496 since POSIX says we shouldn't. Thus, we set
5498 `buffer' to the compiled pattern;
5499 `used' to the length of the compiled pattern;
5500 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5501 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5502 RE_SYNTAX_POSIX_BASIC;
5503 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5504 `fastmap' and `fastmap_accurate' to zero;
5505 `re_nsub' to the number of subexpressions in PATTERN.
5507 PATTERN is the address of the pattern string.
5509 CFLAGS is a series of bits which affect compilation.
5511 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5512 use POSIX basic syntax.
5514 If REG_NEWLINE is set, then . and [^...] don't match newline.
5515 Also, regexec will try a match beginning after every newline.
5517 If REG_ICASE is set, then we considers upper- and lowercase
5518 versions of letters to be equivalent when matching.
5520 If REG_NOSUB is set, then when PREG is passed to regexec, that
5521 routine will report only success or failure, and nothing about the
5524 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
5525 the return codes and their meanings.) */
5528 regcomp (preg, pattern, cflags)
5530 const char *pattern;
5535 = (cflags & REG_EXTENDED) ?
5536 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5538 /* regex_compile will allocate the space for the compiled pattern. */
5540 preg->allocated = 0;
5543 /* Don't bother to use a fastmap when searching. This simplifies the
5544 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
5545 characters after newlines into the fastmap. This way, we just try
5549 if (cflags & REG_ICASE)
5554 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
5555 * sizeof (*(RE_TRANSLATE_TYPE)0));
5556 if (preg->translate == NULL)
5557 return (int) REG_ESPACE;
5559 /* Map uppercase characters to corresponding lowercase ones. */
5560 for (i = 0; i < CHAR_SET_SIZE; i++)
5561 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
5564 preg->translate = NULL;
5566 /* If REG_NEWLINE is set, newlines are treated differently. */
5567 if (cflags & REG_NEWLINE)
5568 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5569 syntax &= ~RE_DOT_NEWLINE;
5570 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5571 /* It also changes the matching behavior. */
5572 preg->newline_anchor = 1;
5575 preg->newline_anchor = 0;
5577 preg->no_sub = !!(cflags & REG_NOSUB);
5579 /* POSIX says a null character in the pattern terminates it, so we
5580 can use strlen here in compiling the pattern. */
5581 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5583 /* POSIX doesn't distinguish between an unmatched open-group and an
5584 unmatched close-group: both are REG_EPAREN. */
5585 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5591 /* regexec searches for a given pattern, specified by PREG, in the
5594 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5595 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5596 least NMATCH elements, and we set them to the offsets of the
5597 corresponding matched substrings.
5599 EFLAGS specifies `execution flags' which affect matching: if
5600 REG_NOTBOL is set, then ^ does not match at the beginning of the
5601 string; if REG_NOTEOL is set, then $ does not match at the end.
5603 We return 0 if we find a match and REG_NOMATCH if not. */
5606 regexec (preg, string, nmatch, pmatch, eflags)
5607 const regex_t *preg;
5610 regmatch_t pmatch[];
5614 struct re_registers regs;
5615 regex_t private_preg;
5616 int len = strlen (string);
5617 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5619 private_preg = *preg;
5621 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5622 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5624 /* The user has told us exactly how many registers to return
5625 information about, via `nmatch'. We have to pass that on to the
5626 matching routines. */
5627 private_preg.regs_allocated = REGS_FIXED;
5631 regs.num_regs = nmatch;
5632 regs.start = TALLOC (nmatch, regoff_t);
5633 regs.end = TALLOC (nmatch, regoff_t);
5634 if (regs.start == NULL || regs.end == NULL)
5635 return (int) REG_NOMATCH;
5638 /* Perform the searching operation. */
5639 ret = re_search (&private_preg, string, len,
5640 /* start: */ 0, /* range: */ len,
5641 want_reg_info ? ®s : (struct re_registers *) 0);
5643 /* Copy the register information to the POSIX structure. */
5650 for (r = 0; r < nmatch; r++)
5652 pmatch[r].rm_so = regs.start[r];
5653 pmatch[r].rm_eo = regs.end[r];
5657 /* If we needed the temporary register info, free the space now. */
5662 /* We want zero return to mean success, unlike `re_search'. */
5663 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5667 /* Returns a message corresponding to an error code, ERRCODE, returned
5668 from either regcomp or regexec. We don't use PREG here. */
5671 regerror (errcode, preg, errbuf, errbuf_size)
5673 const regex_t *preg;
5681 || errcode >= (int) (sizeof (re_error_msgid)
5682 / sizeof (re_error_msgid[0])))
5683 /* Only error codes returned by the rest of the code should be passed
5684 to this routine. If we are given anything else, or if other regex
5685 code generates an invalid error code, then the program has a bug.
5686 Dump core so we can fix it. */
5689 msg = gettext (re_error_msgid[errcode]);
5691 msg_size = strlen (msg) + 1; /* Includes the null. */
5693 if (errbuf_size != 0)
5695 if (msg_size > errbuf_size)
5697 #if defined HAVE_MEMPCPY || defined _LIBC
5698 *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
5700 memcpy (errbuf, msg, errbuf_size - 1);
5701 errbuf[errbuf_size - 1] = 0;
5705 memcpy (errbuf, msg, msg_size);
5712 /* Free dynamically allocated space used by PREG. */
5718 if (preg->buffer != NULL)
5719 free (preg->buffer);
5720 preg->buffer = NULL;
5722 preg->allocated = 0;
5725 if (preg->fastmap != NULL)
5726 free (preg->fastmap);
5727 preg->fastmap = NULL;
5728 preg->fastmap_accurate = 0;
5730 if (preg->translate != NULL)
5731 free (preg->translate);
5732 preg->translate = NULL;
5735 #endif /* not emacs */
projects / gnulib.git / blob