1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P10003.2/D11.2, except for
4 internationalization features.)
6 Copyright (C) 1993, 1994, 1995, 1996, 1997 Free Software Foundation, Inc.
8 This file is part of the GNU C Library. Its master source is NOT part of
9 the C library, however. The master source lives in /gd/gnu/lib.
11 The GNU C Library is free software; you can redistribute it and/or
12 modify it under the terms of the GNU Library General Public License as
13 published by the Free Software Foundation; either version 2 of the
14 License, or (at your option) any later version.
16 The GNU C Library is distributed in the hope that it will be useful,
17 but WITHOUT ANY WARRANTY; without even the implied warranty of
18 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 Library General Public License for more details.
21 You should have received a copy of the GNU Library General Public
22 License along with the GNU C Library; see the file COPYING.LIB. If not,
23 write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
24 Boston, MA 02111-1307, USA. */
26 /* AIX requires this to be the first thing in the file. */
27 #if defined (_AIX) && !defined (REGEX_MALLOC)
38 /* We need this for `regex.h', and perhaps for the Emacs include files. */
39 #include <sys/types.h>
41 /* This is for other GNU distributions with internationalized messages. */
42 #if HAVE_LIBINTL_H || defined (_LIBC)
45 # define gettext(msgid) (msgid)
49 /* This define is so xgettext can find the internationalizable
51 #define gettext_noop(String) String
54 /* The `emacs' switch turns on certain matching commands
55 that make sense only in Emacs. */
64 /* If we are not linking with Emacs proper,
65 we can't use the relocating allocator
66 even if config.h says that we can. */
69 #if defined (STDC_HEADERS) || defined (_LIBC)
76 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
77 If nothing else has been done, use the method below. */
78 #ifdef INHIBIT_STRING_HEADER
79 #if !(defined (HAVE_BZERO) && defined (HAVE_BCOPY))
80 #if !defined (bzero) && !defined (bcopy)
81 #undef INHIBIT_STRING_HEADER
86 /* This is the normal way of making sure we have a bcopy and a bzero.
87 This is used in most programs--a few other programs avoid this
88 by defining INHIBIT_STRING_HEADER. */
89 #ifndef INHIBIT_STRING_HEADER
90 #if defined (HAVE_STRING_H) || defined (STDC_HEADERS) || defined (_LIBC)
93 #define bcmp(s1, s2, n) memcmp ((s1), (s2), (n))
96 #define bcopy(s, d, n) memcpy ((d), (s), (n))
99 #define bzero(s, n) memset ((s), 0, (n))
106 /* Define the syntax stuff for \<, \>, etc. */
108 /* This must be nonzero for the wordchar and notwordchar pattern
109 commands in re_match_2. */
114 #ifdef SWITCH_ENUM_BUG
115 #define SWITCH_ENUM_CAST(x) ((int)(x))
117 #define SWITCH_ENUM_CAST(x) (x)
122 extern char *re_syntax_table;
124 #else /* not SYNTAX_TABLE */
126 /* How many characters in the character set. */
127 #define CHAR_SET_SIZE 256
129 static char re_syntax_table[CHAR_SET_SIZE];
140 bzero (re_syntax_table, sizeof re_syntax_table);
142 for (c = 'a'; c <= 'z'; c++)
143 re_syntax_table[c] = Sword;
145 for (c = 'A'; c <= 'Z'; c++)
146 re_syntax_table[c] = Sword;
148 for (c = '0'; c <= '9'; c++)
149 re_syntax_table[c] = Sword;
151 re_syntax_table['_'] = Sword;
156 #endif /* not SYNTAX_TABLE */
158 #define SYNTAX(c) re_syntax_table[c]
160 #endif /* not emacs */
162 /* Get the interface, including the syntax bits. */
165 /* isalpha etc. are used for the character classes. */
168 /* Jim Meyering writes:
170 "... Some ctype macros are valid only for character codes that
171 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
172 using /bin/cc or gcc but without giving an ansi option). So, all
173 ctype uses should be through macros like ISPRINT... If
174 STDC_HEADERS is defined, then autoconf has verified that the ctype
175 macros don't need to be guarded with references to isascii. ...
176 Defining isascii to 1 should let any compiler worth its salt
177 eliminate the && through constant folding." */
179 #if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII))
182 #define ISASCII(c) isascii(c)
186 #define ISBLANK(c) (ISASCII (c) && isblank (c))
188 #define ISBLANK(c) ((c) == ' ' || (c) == '\t')
191 #define ISGRAPH(c) (ISASCII (c) && isgraph (c))
193 #define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
196 #define ISPRINT(c) (ISASCII (c) && isprint (c))
197 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
198 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
199 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
200 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
201 #define ISLOWER(c) (ISASCII (c) && islower (c))
202 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
203 #define ISSPACE(c) (ISASCII (c) && isspace (c))
204 #define ISUPPER(c) (ISASCII (c) && isupper (c))
205 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
208 #define NULL (void *)0
211 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
212 since ours (we hope) works properly with all combinations of
213 machines, compilers, `char' and `unsigned char' argument types.
214 (Per Bothner suggested the basic approach.) */
215 #undef SIGN_EXTEND_CHAR
217 #define SIGN_EXTEND_CHAR(c) ((signed char) (c))
218 #else /* not __STDC__ */
219 /* As in Harbison and Steele. */
220 #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
223 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
224 use `alloca' instead of `malloc'. This is because using malloc in
225 re_search* or re_match* could cause memory leaks when C-g is used in
226 Emacs; also, malloc is slower and causes storage fragmentation. On
227 the other hand, malloc is more portable, and easier to debug.
229 Because we sometimes use alloca, some routines have to be macros,
230 not functions -- `alloca'-allocated space disappears at the end of the
231 function it is called in. */
235 #define REGEX_ALLOCATE malloc
236 #define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
237 #define REGEX_FREE free
239 #else /* not REGEX_MALLOC */
241 /* Emacs already defines alloca, sometimes. */
244 /* Make alloca work the best possible way. */
246 #define alloca __builtin_alloca
247 #else /* not __GNUC__ */
250 #else /* not __GNUC__ or HAVE_ALLOCA_H */
251 #if 0 /* It is a bad idea to declare alloca. We always cast the result. */
252 #ifndef _AIX /* Already did AIX, up at the top. */
254 #endif /* not _AIX */
256 #endif /* not HAVE_ALLOCA_H */
257 #endif /* not __GNUC__ */
259 #endif /* not alloca */
261 #define REGEX_ALLOCATE alloca
263 /* Assumes a `char *destination' variable. */
264 #define REGEX_REALLOCATE(source, osize, nsize) \
265 (destination = (char *) alloca (nsize), \
266 bcopy (source, destination, osize), \
269 /* No need to do anything to free, after alloca. */
270 #define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
272 #endif /* not REGEX_MALLOC */
274 /* Define how to allocate the failure stack. */
276 #if defined (REL_ALLOC) && defined (REGEX_MALLOC)
278 #define REGEX_ALLOCATE_STACK(size) \
279 r_alloc (&failure_stack_ptr, (size))
280 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
281 r_re_alloc (&failure_stack_ptr, (nsize))
282 #define REGEX_FREE_STACK(ptr) \
283 r_alloc_free (&failure_stack_ptr)
285 #else /* not using relocating allocator */
289 #define REGEX_ALLOCATE_STACK malloc
290 #define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
291 #define REGEX_FREE_STACK free
293 #else /* not REGEX_MALLOC */
295 #define REGEX_ALLOCATE_STACK alloca
297 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
298 REGEX_REALLOCATE (source, osize, nsize)
299 /* No need to explicitly free anything. */
300 #define REGEX_FREE_STACK(arg)
302 #endif /* not REGEX_MALLOC */
303 #endif /* not using relocating allocator */
306 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
307 `string1' or just past its end. This works if PTR is NULL, which is
309 #define FIRST_STRING_P(ptr) \
310 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
312 /* (Re)Allocate N items of type T using malloc, or fail. */
313 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
314 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
315 #define RETALLOC_IF(addr, n, t) \
316 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
317 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
319 #define BYTEWIDTH 8 /* In bits. */
321 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
325 #define MAX(a, b) ((a) > (b) ? (a) : (b))
326 #define MIN(a, b) ((a) < (b) ? (a) : (b))
328 typedef char boolean;
332 static int re_match_2_internal ();
334 /* These are the command codes that appear in compiled regular
335 expressions. Some opcodes are followed by argument bytes. A
336 command code can specify any interpretation whatsoever for its
337 arguments. Zero bytes may appear in the compiled regular expression. */
343 /* Succeed right away--no more backtracking. */
346 /* Followed by one byte giving n, then by n literal bytes. */
349 /* Matches any (more or less) character. */
352 /* Matches any one char belonging to specified set. First
353 following byte is number of bitmap bytes. Then come bytes
354 for a bitmap saying which chars are in. Bits in each byte
355 are ordered low-bit-first. A character is in the set if its
356 bit is 1. A character too large to have a bit in the map is
357 automatically not in the set. */
360 /* Same parameters as charset, but match any character that is
361 not one of those specified. */
364 /* Start remembering the text that is matched, for storing in a
365 register. Followed by one byte with the register number, in
366 the range 0 to one less than the pattern buffer's re_nsub
367 field. Then followed by one byte with the number of groups
368 inner to this one. (This last has to be part of the
369 start_memory only because we need it in the on_failure_jump
373 /* Stop remembering the text that is matched and store it in a
374 memory register. Followed by one byte with the register
375 number, in the range 0 to one less than `re_nsub' in the
376 pattern buffer, and one byte with the number of inner groups,
377 just like `start_memory'. (We need the number of inner
378 groups here because we don't have any easy way of finding the
379 corresponding start_memory when we're at a stop_memory.) */
382 /* Match a duplicate of something remembered. Followed by one
383 byte containing the register number. */
386 /* Fail unless at beginning of line. */
389 /* Fail unless at end of line. */
392 /* Succeeds if at beginning of buffer (if emacs) or at beginning
393 of string to be matched (if not). */
396 /* Analogously, for end of buffer/string. */
399 /* Followed by two byte relative address to which to jump. */
402 /* Same as jump, but marks the end of an alternative. */
405 /* Followed by two-byte relative address of place to resume at
406 in case of failure. */
409 /* Like on_failure_jump, but pushes a placeholder instead of the
410 current string position when executed. */
411 on_failure_keep_string_jump,
413 /* Throw away latest failure point and then jump to following
414 two-byte relative address. */
417 /* Change to pop_failure_jump if know won't have to backtrack to
418 match; otherwise change to jump. This is used to jump
419 back to the beginning of a repeat. If what follows this jump
420 clearly won't match what the repeat does, such that we can be
421 sure that there is no use backtracking out of repetitions
422 already matched, then we change it to a pop_failure_jump.
423 Followed by two-byte address. */
426 /* Jump to following two-byte address, and push a dummy failure
427 point. This failure point will be thrown away if an attempt
428 is made to use it for a failure. A `+' construct makes this
429 before the first repeat. Also used as an intermediary kind
430 of jump when compiling an alternative. */
433 /* Push a dummy failure point and continue. Used at the end of
437 /* Followed by two-byte relative address and two-byte number n.
438 After matching N times, jump to the address upon failure. */
441 /* Followed by two-byte relative address, and two-byte number n.
442 Jump to the address N times, then fail. */
445 /* Set the following two-byte relative address to the
446 subsequent two-byte number. The address *includes* the two
450 wordchar, /* Matches any word-constituent character. */
451 notwordchar, /* Matches any char that is not a word-constituent. */
453 wordbeg, /* Succeeds if at word beginning. */
454 wordend, /* Succeeds if at word end. */
456 wordbound, /* Succeeds if at a word boundary. */
457 notwordbound /* Succeeds if not at a word boundary. */
460 ,before_dot, /* Succeeds if before point. */
461 at_dot, /* Succeeds if at point. */
462 after_dot, /* Succeeds if after point. */
464 /* Matches any character whose syntax is specified. Followed by
465 a byte which contains a syntax code, e.g., Sword. */
468 /* Matches any character whose syntax is not that specified. */
473 /* Common operations on the compiled pattern. */
475 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
477 #define STORE_NUMBER(destination, number) \
479 (destination)[0] = (number) & 0377; \
480 (destination)[1] = (number) >> 8; \
483 /* Same as STORE_NUMBER, except increment DESTINATION to
484 the byte after where the number is stored. Therefore, DESTINATION
485 must be an lvalue. */
487 #define STORE_NUMBER_AND_INCR(destination, number) \
489 STORE_NUMBER (destination, number); \
490 (destination) += 2; \
493 /* Put into DESTINATION a number stored in two contiguous bytes starting
496 #define EXTRACT_NUMBER(destination, source) \
498 (destination) = *(source) & 0377; \
499 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
504 extract_number (dest, source)
506 unsigned char *source;
508 int temp = SIGN_EXTEND_CHAR (*(source + 1));
509 *dest = *source & 0377;
513 #ifndef EXTRACT_MACROS /* To debug the macros. */
514 #undef EXTRACT_NUMBER
515 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
516 #endif /* not EXTRACT_MACROS */
520 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
521 SOURCE must be an lvalue. */
523 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
525 EXTRACT_NUMBER (destination, source); \
531 extract_number_and_incr (destination, source)
533 unsigned char **source;
535 extract_number (destination, *source);
539 #ifndef EXTRACT_MACROS
540 #undef EXTRACT_NUMBER_AND_INCR
541 #define EXTRACT_NUMBER_AND_INCR(dest, src) \
542 extract_number_and_incr (&dest, &src)
543 #endif /* not EXTRACT_MACROS */
547 /* If DEBUG is defined, Regex prints many voluminous messages about what
548 it is doing (if the variable `debug' is nonzero). If linked with the
549 main program in `iregex.c', you can enter patterns and strings
550 interactively. And if linked with the main program in `main.c' and
551 the other test files, you can run the already-written tests. */
555 /* We use standard I/O for debugging. */
558 /* It is useful to test things that ``must'' be true when debugging. */
561 static int debug = 0;
563 #define DEBUG_STATEMENT(e) e
564 #define DEBUG_PRINT1(x) if (debug) printf (x)
565 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
566 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
567 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
568 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
569 if (debug) print_partial_compiled_pattern (s, e)
570 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
571 if (debug) print_double_string (w, s1, sz1, s2, sz2)
574 /* Print the fastmap in human-readable form. */
577 print_fastmap (fastmap)
580 unsigned was_a_range = 0;
583 while (i < (1 << BYTEWIDTH))
589 while (i < (1 << BYTEWIDTH) && fastmap[i])
605 /* Print a compiled pattern string in human-readable form, starting at
606 the START pointer into it and ending just before the pointer END. */
609 print_partial_compiled_pattern (start, end)
610 unsigned char *start;
614 unsigned char *p = start;
615 unsigned char *pend = end;
623 /* Loop over pattern commands. */
626 printf ("%d:\t", p - start);
628 switch ((re_opcode_t) *p++)
636 printf ("/exactn/%d", mcnt);
647 printf ("/start_memory/%d/%d", mcnt, *p++);
652 printf ("/stop_memory/%d/%d", mcnt, *p++);
656 printf ("/duplicate/%d", *p++);
666 register int c, last = -100;
667 register int in_range = 0;
669 printf ("/charset [%s",
670 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
672 assert (p + *p < pend);
674 for (c = 0; c < 256; c++)
676 && (p[1 + (c/8)] & (1 << (c % 8))))
678 /* Are we starting a range? */
679 if (last + 1 == c && ! in_range)
684 /* Have we broken a range? */
685 else if (last + 1 != c && in_range)
714 case on_failure_jump:
715 extract_number_and_incr (&mcnt, &p);
716 printf ("/on_failure_jump to %d", p + mcnt - start);
719 case on_failure_keep_string_jump:
720 extract_number_and_incr (&mcnt, &p);
721 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
724 case dummy_failure_jump:
725 extract_number_and_incr (&mcnt, &p);
726 printf ("/dummy_failure_jump to %d", p + mcnt - start);
729 case push_dummy_failure:
730 printf ("/push_dummy_failure");
734 extract_number_and_incr (&mcnt, &p);
735 printf ("/maybe_pop_jump to %d", p + mcnt - start);
738 case pop_failure_jump:
739 extract_number_and_incr (&mcnt, &p);
740 printf ("/pop_failure_jump to %d", p + mcnt - start);
744 extract_number_and_incr (&mcnt, &p);
745 printf ("/jump_past_alt to %d", p + mcnt - start);
749 extract_number_and_incr (&mcnt, &p);
750 printf ("/jump to %d", p + mcnt - start);
754 extract_number_and_incr (&mcnt, &p);
755 extract_number_and_incr (&mcnt2, &p);
756 printf ("/succeed_n to %d, %d times", p + mcnt - start, mcnt2);
760 extract_number_and_incr (&mcnt, &p);
761 extract_number_and_incr (&mcnt2, &p);
762 printf ("/jump_n to %d, %d times", p + mcnt - start, mcnt2);
766 extract_number_and_incr (&mcnt, &p);
767 extract_number_and_incr (&mcnt2, &p);
768 printf ("/set_number_at location %d to %d", p + mcnt - start, mcnt2);
772 printf ("/wordbound");
776 printf ("/notwordbound");
788 printf ("/before_dot");
796 printf ("/after_dot");
800 printf ("/syntaxspec");
802 printf ("/%d", mcnt);
806 printf ("/notsyntaxspec");
808 printf ("/%d", mcnt);
813 printf ("/wordchar");
817 printf ("/notwordchar");
829 printf ("?%d", *(p-1));
835 printf ("%d:\tend of pattern.\n", p - start);
840 print_compiled_pattern (bufp)
841 struct re_pattern_buffer *bufp;
843 unsigned char *buffer = bufp->buffer;
845 print_partial_compiled_pattern (buffer, buffer + bufp->used);
846 printf ("%d bytes used/%d bytes allocated.\n", bufp->used, bufp->allocated);
848 if (bufp->fastmap_accurate && bufp->fastmap)
850 printf ("fastmap: ");
851 print_fastmap (bufp->fastmap);
854 printf ("re_nsub: %d\t", bufp->re_nsub);
855 printf ("regs_alloc: %d\t", bufp->regs_allocated);
856 printf ("can_be_null: %d\t", bufp->can_be_null);
857 printf ("newline_anchor: %d\n", bufp->newline_anchor);
858 printf ("no_sub: %d\t", bufp->no_sub);
859 printf ("not_bol: %d\t", bufp->not_bol);
860 printf ("not_eol: %d\t", bufp->not_eol);
861 printf ("syntax: %d\n", bufp->syntax);
862 /* Perhaps we should print the translate table? */
867 print_double_string (where, string1, size1, string2, size2)
880 if (FIRST_STRING_P (where))
882 for (this_char = where - string1; this_char < size1; this_char++)
883 putchar (string1[this_char]);
888 for (this_char = where - string2; this_char < size2; this_char++)
889 putchar (string2[this_char]);
893 #else /* not DEBUG */
898 #define DEBUG_STATEMENT(e)
899 #define DEBUG_PRINT1(x)
900 #define DEBUG_PRINT2(x1, x2)
901 #define DEBUG_PRINT3(x1, x2, x3)
902 #define DEBUG_PRINT4(x1, x2, x3, x4)
903 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
904 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
906 #endif /* not DEBUG */
908 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
909 also be assigned to arbitrarily: each pattern buffer stores its own
910 syntax, so it can be changed between regex compilations. */
911 /* This has no initializer because initialized variables in Emacs
912 become read-only after dumping. */
913 reg_syntax_t re_syntax_options;
916 /* Specify the precise syntax of regexps for compilation. This provides
917 for compatibility for various utilities which historically have
918 different, incompatible syntaxes.
920 The argument SYNTAX is a bit mask comprised of the various bits
921 defined in regex.h. We return the old syntax. */
924 re_set_syntax (syntax)
927 reg_syntax_t ret = re_syntax_options;
929 re_syntax_options = syntax;
933 /* This table gives an error message for each of the error codes listed
934 in regex.h. Obviously the order here has to be same as there.
935 POSIX doesn't require that we do anything for REG_NOERROR,
936 but why not be nice? */
938 static const char *re_error_msgid[] =
940 gettext_noop ("Success"), /* REG_NOERROR */
941 gettext_noop ("No match"), /* REG_NOMATCH */
942 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
943 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
944 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
945 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
946 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
947 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
948 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
949 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
950 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
951 gettext_noop ("Invalid range end"), /* REG_ERANGE */
952 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
953 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
954 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
955 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
956 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
959 /* Avoiding alloca during matching, to placate r_alloc. */
961 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
962 searching and matching functions should not call alloca. On some
963 systems, alloca is implemented in terms of malloc, and if we're
964 using the relocating allocator routines, then malloc could cause a
965 relocation, which might (if the strings being searched are in the
966 ralloc heap) shift the data out from underneath the regexp
969 Here's another reason to avoid allocation: Emacs
970 processes input from X in a signal handler; processing X input may
971 call malloc; if input arrives while a matching routine is calling
972 malloc, then we're scrod. But Emacs can't just block input while
973 calling matching routines; then we don't notice interrupts when
974 they come in. So, Emacs blocks input around all regexp calls
975 except the matching calls, which it leaves unprotected, in the
976 faith that they will not malloc. */
978 /* Normally, this is fine. */
979 #define MATCH_MAY_ALLOCATE
981 /* When using GNU C, we are not REALLY using the C alloca, no matter
982 what config.h may say. So don't take precautions for it. */
987 /* The match routines may not allocate if (1) they would do it with malloc
988 and (2) it's not safe for them to use malloc.
989 Note that if REL_ALLOC is defined, matching would not use malloc for the
990 failure stack, but we would still use it for the register vectors;
991 so REL_ALLOC should not affect this. */
992 #if (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (emacs)
993 #undef MATCH_MAY_ALLOCATE
997 /* Failure stack declarations and macros; both re_compile_fastmap and
998 re_match_2 use a failure stack. These have to be macros because of
999 REGEX_ALLOCATE_STACK. */
1002 /* Number of failure points for which to initially allocate space
1003 when matching. If this number is exceeded, we allocate more
1004 space, so it is not a hard limit. */
1005 #ifndef INIT_FAILURE_ALLOC
1006 #define INIT_FAILURE_ALLOC 5
1009 /* Roughly the maximum number of failure points on the stack. Would be
1010 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1011 This is a variable only so users of regex can assign to it; we never
1012 change it ourselves. */
1013 #if defined (MATCH_MAY_ALLOCATE)
1014 /* 4400 was enough to cause a crash on Alpha OSF/1,
1015 whose default stack limit is 2mb. */
1016 int re_max_failures = 4000;
1018 int re_max_failures = 2000;
1021 union fail_stack_elt
1023 unsigned char *pointer;
1027 typedef union fail_stack_elt fail_stack_elt_t;
1031 fail_stack_elt_t *stack;
1033 unsigned avail; /* Offset of next open position. */
1036 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1037 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1038 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1041 /* Define macros to initialize and free the failure stack.
1042 Do `return -2' if the alloc fails. */
1044 #ifdef MATCH_MAY_ALLOCATE
1045 #define INIT_FAIL_STACK() \
1047 fail_stack.stack = (fail_stack_elt_t *) \
1048 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1050 if (fail_stack.stack == NULL) \
1053 fail_stack.size = INIT_FAILURE_ALLOC; \
1054 fail_stack.avail = 0; \
1057 #define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1059 #define INIT_FAIL_STACK() \
1061 fail_stack.avail = 0; \
1064 #define RESET_FAIL_STACK()
1068 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1070 Return 1 if succeeds, and 0 if either ran out of memory
1071 allocating space for it or it was already too large.
1073 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1075 #define DOUBLE_FAIL_STACK(fail_stack) \
1076 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1078 : ((fail_stack).stack = (fail_stack_elt_t *) \
1079 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1080 (fail_stack).size * sizeof (fail_stack_elt_t), \
1081 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1083 (fail_stack).stack == NULL \
1085 : ((fail_stack).size <<= 1, \
1089 /* Push pointer POINTER on FAIL_STACK.
1090 Return 1 if was able to do so and 0 if ran out of memory allocating
1092 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1093 ((FAIL_STACK_FULL () \
1094 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1096 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1099 /* Push a pointer value onto the failure stack.
1100 Assumes the variable `fail_stack'. Probably should only
1101 be called from within `PUSH_FAILURE_POINT'. */
1102 #define PUSH_FAILURE_POINTER(item) \
1103 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1105 /* This pushes an integer-valued item onto the failure stack.
1106 Assumes the variable `fail_stack'. Probably should only
1107 be called from within `PUSH_FAILURE_POINT'. */
1108 #define PUSH_FAILURE_INT(item) \
1109 fail_stack.stack[fail_stack.avail++].integer = (item)
1111 /* Push a fail_stack_elt_t value onto the failure stack.
1112 Assumes the variable `fail_stack'. Probably should only
1113 be called from within `PUSH_FAILURE_POINT'. */
1114 #define PUSH_FAILURE_ELT(item) \
1115 fail_stack.stack[fail_stack.avail++] = (item)
1117 /* These three POP... operations complement the three PUSH... operations.
1118 All assume that `fail_stack' is nonempty. */
1119 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1120 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1121 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1123 /* Used to omit pushing failure point id's when we're not debugging. */
1125 #define DEBUG_PUSH PUSH_FAILURE_INT
1126 #define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1128 #define DEBUG_PUSH(item)
1129 #define DEBUG_POP(item_addr)
1133 /* Push the information about the state we will need
1134 if we ever fail back to it.
1136 Requires variables fail_stack, regstart, regend, reg_info, and
1137 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be
1140 Does `return FAILURE_CODE' if runs out of memory. */
1142 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1144 char *destination; \
1145 /* Must be int, so when we don't save any registers, the arithmetic \
1146 of 0 + -1 isn't done as unsigned. */ \
1147 unsigned this_reg; \
1149 DEBUG_STATEMENT (failure_id++); \
1150 DEBUG_STATEMENT (nfailure_points_pushed++); \
1151 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1152 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1153 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1155 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \
1156 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1158 /* Ensure we have enough space allocated for what we will push. */ \
1159 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1161 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1162 return failure_code; \
1164 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1165 (fail_stack).size); \
1166 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1169 /* Push the info, starting with the registers. */ \
1170 DEBUG_PRINT1 ("\n"); \
1173 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1176 DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \
1177 DEBUG_STATEMENT (num_regs_pushed++); \
1179 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1180 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1182 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1183 PUSH_FAILURE_POINTER (regend[this_reg]); \
1185 DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \
1186 DEBUG_PRINT2 (" match_null=%d", \
1187 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1188 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1189 DEBUG_PRINT2 (" matched_something=%d", \
1190 MATCHED_SOMETHING (reg_info[this_reg])); \
1191 DEBUG_PRINT2 (" ever_matched=%d", \
1192 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1193 DEBUG_PRINT1 ("\n"); \
1194 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1197 DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\
1198 PUSH_FAILURE_INT (lowest_active_reg); \
1200 DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\
1201 PUSH_FAILURE_INT (highest_active_reg); \
1203 DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \
1204 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1205 PUSH_FAILURE_POINTER (pattern_place); \
1207 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \
1208 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1210 DEBUG_PRINT1 ("'\n"); \
1211 PUSH_FAILURE_POINTER (string_place); \
1213 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1214 DEBUG_PUSH (failure_id); \
1217 /* This is the number of items that are pushed and popped on the stack
1218 for each register. */
1219 #define NUM_REG_ITEMS 3
1221 /* Individual items aside from the registers. */
1223 #define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1225 #define NUM_NONREG_ITEMS 4
1228 /* We push at most this many items on the stack. */
1229 /* We used to use (num_regs - 1), which is the number of registers
1230 this regexp will save; but that was changed to 5
1231 to avoid stack overflow for a regexp with lots of parens. */
1232 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1234 /* We actually push this many items. */
1235 #define NUM_FAILURE_ITEMS \
1237 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1241 /* How many items can still be added to the stack without overflowing it. */
1242 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1245 /* Pops what PUSH_FAIL_STACK pushes.
1247 We restore into the parameters, all of which should be lvalues:
1248 STR -- the saved data position.
1249 PAT -- the saved pattern position.
1250 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1251 REGSTART, REGEND -- arrays of string positions.
1252 REG_INFO -- array of information about each subexpression.
1254 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1255 `pend', `string1', `size1', `string2', and `size2'. */
1257 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1259 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \
1260 unsigned this_reg; \
1261 const unsigned char *string_temp; \
1263 assert (!FAIL_STACK_EMPTY ()); \
1265 /* Remove failure points and point to how many regs pushed. */ \
1266 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1267 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1268 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1270 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1272 DEBUG_POP (&failure_id); \
1273 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1275 /* If the saved string location is NULL, it came from an \
1276 on_failure_keep_string_jump opcode, and we want to throw away the \
1277 saved NULL, thus retaining our current position in the string. */ \
1278 string_temp = POP_FAILURE_POINTER (); \
1279 if (string_temp != NULL) \
1280 str = (const char *) string_temp; \
1282 DEBUG_PRINT2 (" Popping string 0x%x: `", str); \
1283 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1284 DEBUG_PRINT1 ("'\n"); \
1286 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1287 DEBUG_PRINT2 (" Popping pattern 0x%x: ", pat); \
1288 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1290 /* Restore register info. */ \
1291 high_reg = (unsigned) POP_FAILURE_INT (); \
1292 DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \
1294 low_reg = (unsigned) POP_FAILURE_INT (); \
1295 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \
1298 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1300 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \
1302 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1303 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \
1305 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1306 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1308 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1309 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1313 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1315 reg_info[this_reg].word.integer = 0; \
1316 regend[this_reg] = 0; \
1317 regstart[this_reg] = 0; \
1319 highest_active_reg = high_reg; \
1322 set_regs_matched_done = 0; \
1323 DEBUG_STATEMENT (nfailure_points_popped++); \
1324 } /* POP_FAILURE_POINT */
1328 /* Structure for per-register (a.k.a. per-group) information.
1329 Other register information, such as the
1330 starting and ending positions (which are addresses), and the list of
1331 inner groups (which is a bits list) are maintained in separate
1334 We are making a (strictly speaking) nonportable assumption here: that
1335 the compiler will pack our bit fields into something that fits into
1336 the type of `word', i.e., is something that fits into one item on the
1341 fail_stack_elt_t word;
1344 /* This field is one if this group can match the empty string,
1345 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1346 #define MATCH_NULL_UNSET_VALUE 3
1347 unsigned match_null_string_p : 2;
1348 unsigned is_active : 1;
1349 unsigned matched_something : 1;
1350 unsigned ever_matched_something : 1;
1352 } register_info_type;
1354 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1355 #define IS_ACTIVE(R) ((R).bits.is_active)
1356 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1357 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1360 /* Call this when have matched a real character; it sets `matched' flags
1361 for the subexpressions which we are currently inside. Also records
1362 that those subexprs have matched. */
1363 #define SET_REGS_MATCHED() \
1366 if (!set_regs_matched_done) \
1369 set_regs_matched_done = 1; \
1370 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1372 MATCHED_SOMETHING (reg_info[r]) \
1373 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1380 /* Registers are set to a sentinel when they haven't yet matched. */
1381 static char reg_unset_dummy;
1382 #define REG_UNSET_VALUE (®_unset_dummy)
1383 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1385 /* Subroutine declarations and macros for regex_compile. */
1387 static void store_op1 (), store_op2 ();
1388 static void insert_op1 (), insert_op2 ();
1389 static boolean at_begline_loc_p (), at_endline_loc_p ();
1390 static boolean group_in_compile_stack ();
1391 static reg_errcode_t compile_range ();
1393 /* Fetch the next character in the uncompiled pattern---translating it
1394 if necessary. Also cast from a signed character in the constant
1395 string passed to us by the user to an unsigned char that we can use
1396 as an array index (in, e.g., `translate'). */
1398 #define PATFETCH(c) \
1399 do {if (p == pend) return REG_EEND; \
1400 c = (unsigned char) *p++; \
1401 if (translate) c = (unsigned char) translate[c]; \
1405 /* Fetch the next character in the uncompiled pattern, with no
1407 #define PATFETCH_RAW(c) \
1408 do {if (p == pend) return REG_EEND; \
1409 c = (unsigned char) *p++; \
1412 /* Go backwards one character in the pattern. */
1413 #define PATUNFETCH p--
1416 /* If `translate' is non-null, return translate[D], else just D. We
1417 cast the subscript to translate because some data is declared as
1418 `char *', to avoid warnings when a string constant is passed. But
1419 when we use a character as a subscript we must make it unsigned. */
1421 #define TRANSLATE(d) \
1422 (translate ? (char) translate[(unsigned char) (d)] : (d))
1426 /* Macros for outputting the compiled pattern into `buffer'. */
1428 /* If the buffer isn't allocated when it comes in, use this. */
1429 #define INIT_BUF_SIZE 32
1431 /* Make sure we have at least N more bytes of space in buffer. */
1432 #define GET_BUFFER_SPACE(n) \
1433 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1436 /* Make sure we have one more byte of buffer space and then add C to it. */
1437 #define BUF_PUSH(c) \
1439 GET_BUFFER_SPACE (1); \
1440 *b++ = (unsigned char) (c); \
1444 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1445 #define BUF_PUSH_2(c1, c2) \
1447 GET_BUFFER_SPACE (2); \
1448 *b++ = (unsigned char) (c1); \
1449 *b++ = (unsigned char) (c2); \
1453 /* As with BUF_PUSH_2, except for three bytes. */
1454 #define BUF_PUSH_3(c1, c2, c3) \
1456 GET_BUFFER_SPACE (3); \
1457 *b++ = (unsigned char) (c1); \
1458 *b++ = (unsigned char) (c2); \
1459 *b++ = (unsigned char) (c3); \
1463 /* Store a jump with opcode OP at LOC to location TO. We store a
1464 relative address offset by the three bytes the jump itself occupies. */
1465 #define STORE_JUMP(op, loc, to) \
1466 store_op1 (op, loc, (to) - (loc) - 3)
1468 /* Likewise, for a two-argument jump. */
1469 #define STORE_JUMP2(op, loc, to, arg) \
1470 store_op2 (op, loc, (to) - (loc) - 3, arg)
1472 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1473 #define INSERT_JUMP(op, loc, to) \
1474 insert_op1 (op, loc, (to) - (loc) - 3, b)
1476 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1477 #define INSERT_JUMP2(op, loc, to, arg) \
1478 insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
1481 /* This is not an arbitrary limit: the arguments which represent offsets
1482 into the pattern are two bytes long. So if 2^16 bytes turns out to
1483 be too small, many things would have to change. */
1484 #define MAX_BUF_SIZE (1L << 16)
1487 /* Extend the buffer by twice its current size via realloc and
1488 reset the pointers that pointed into the old block to point to the
1489 correct places in the new one. If extending the buffer results in it
1490 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1491 #define EXTEND_BUFFER() \
1493 unsigned char *old_buffer = bufp->buffer; \
1494 if (bufp->allocated == MAX_BUF_SIZE) \
1496 bufp->allocated <<= 1; \
1497 if (bufp->allocated > MAX_BUF_SIZE) \
1498 bufp->allocated = MAX_BUF_SIZE; \
1499 bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\
1500 if (bufp->buffer == NULL) \
1501 return REG_ESPACE; \
1502 /* If the buffer moved, move all the pointers into it. */ \
1503 if (old_buffer != bufp->buffer) \
1505 b = (b - old_buffer) + bufp->buffer; \
1506 begalt = (begalt - old_buffer) + bufp->buffer; \
1507 if (fixup_alt_jump) \
1508 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1510 laststart = (laststart - old_buffer) + bufp->buffer; \
1511 if (pending_exact) \
1512 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1517 /* Since we have one byte reserved for the register number argument to
1518 {start,stop}_memory, the maximum number of groups we can report
1519 things about is what fits in that byte. */
1520 #define MAX_REGNUM 255
1522 /* But patterns can have more than `MAX_REGNUM' registers. We just
1523 ignore the excess. */
1524 typedef unsigned regnum_t;
1527 /* Macros for the compile stack. */
1529 /* Since offsets can go either forwards or backwards, this type needs to
1530 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1531 typedef int pattern_offset_t;
1535 pattern_offset_t begalt_offset;
1536 pattern_offset_t fixup_alt_jump;
1537 pattern_offset_t inner_group_offset;
1538 pattern_offset_t laststart_offset;
1540 } compile_stack_elt_t;
1545 compile_stack_elt_t *stack;
1547 unsigned avail; /* Offset of next open position. */
1548 } compile_stack_type;
1551 #define INIT_COMPILE_STACK_SIZE 32
1553 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1554 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1556 /* The next available element. */
1557 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1560 /* Set the bit for character C in a list. */
1561 #define SET_LIST_BIT(c) \
1562 (b[((unsigned char) (c)) / BYTEWIDTH] \
1563 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1566 /* Get the next unsigned number in the uncompiled pattern. */
1567 #define GET_UNSIGNED_NUMBER(num) \
1571 while (ISDIGIT (c)) \
1575 num = num * 10 + c - '0'; \
1583 #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1585 #define IS_CHAR_CLASS(string) \
1586 (STREQ (string, "alpha") || STREQ (string, "upper") \
1587 || STREQ (string, "lower") || STREQ (string, "digit") \
1588 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1589 || STREQ (string, "space") || STREQ (string, "print") \
1590 || STREQ (string, "punct") || STREQ (string, "graph") \
1591 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1593 #ifndef MATCH_MAY_ALLOCATE
1595 /* If we cannot allocate large objects within re_match_2_internal,
1596 we make the fail stack and register vectors global.
1597 The fail stack, we grow to the maximum size when a regexp
1599 The register vectors, we adjust in size each time we
1600 compile a regexp, according to the number of registers it needs. */
1602 static fail_stack_type fail_stack;
1604 /* Size with which the following vectors are currently allocated.
1605 That is so we can make them bigger as needed,
1606 but never make them smaller. */
1607 static int regs_allocated_size;
1609 static const char ** regstart, ** regend;
1610 static const char ** old_regstart, ** old_regend;
1611 static const char **best_regstart, **best_regend;
1612 static register_info_type *reg_info;
1613 static const char **reg_dummy;
1614 static register_info_type *reg_info_dummy;
1616 /* Make the register vectors big enough for NUM_REGS registers,
1617 but don't make them smaller. */
1620 regex_grow_registers (num_regs)
1623 if (num_regs > regs_allocated_size)
1625 RETALLOC_IF (regstart, num_regs, const char *);
1626 RETALLOC_IF (regend, num_regs, const char *);
1627 RETALLOC_IF (old_regstart, num_regs, const char *);
1628 RETALLOC_IF (old_regend, num_regs, const char *);
1629 RETALLOC_IF (best_regstart, num_regs, const char *);
1630 RETALLOC_IF (best_regend, num_regs, const char *);
1631 RETALLOC_IF (reg_info, num_regs, register_info_type);
1632 RETALLOC_IF (reg_dummy, num_regs, const char *);
1633 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1635 regs_allocated_size = num_regs;
1639 #endif /* not MATCH_MAY_ALLOCATE */
1641 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1642 Returns one of error codes defined in `regex.h', or zero for success.
1644 Assumes the `allocated' (and perhaps `buffer') and `translate'
1645 fields are set in BUFP on entry.
1647 If it succeeds, results are put in BUFP (if it returns an error, the
1648 contents of BUFP are undefined):
1649 `buffer' is the compiled pattern;
1650 `syntax' is set to SYNTAX;
1651 `used' is set to the length of the compiled pattern;
1652 `fastmap_accurate' is zero;
1653 `re_nsub' is the number of subexpressions in PATTERN;
1654 `not_bol' and `not_eol' are zero;
1656 The `fastmap' and `newline_anchor' fields are neither
1657 examined nor set. */
1659 /* Return, freeing storage we allocated. */
1660 #define FREE_STACK_RETURN(value) \
1661 return (free (compile_stack.stack), value)
1663 static reg_errcode_t
1664 regex_compile (pattern, size, syntax, bufp)
1665 const char *pattern;
1667 reg_syntax_t syntax;
1668 struct re_pattern_buffer *bufp;
1670 /* We fetch characters from PATTERN here. Even though PATTERN is
1671 `char *' (i.e., signed), we declare these variables as unsigned, so
1672 they can be reliably used as array indices. */
1673 register unsigned char c, c1;
1675 /* A random temporary spot in PATTERN. */
1678 /* Points to the end of the buffer, where we should append. */
1679 register unsigned char *b;
1681 /* Keeps track of unclosed groups. */
1682 compile_stack_type compile_stack;
1684 /* Points to the current (ending) position in the pattern. */
1685 const char *p = pattern;
1686 const char *pend = pattern + size;
1688 /* How to translate the characters in the pattern. */
1689 RE_TRANSLATE_TYPE translate = bufp->translate;
1691 /* Address of the count-byte of the most recently inserted `exactn'
1692 command. This makes it possible to tell if a new exact-match
1693 character can be added to that command or if the character requires
1694 a new `exactn' command. */
1695 unsigned char *pending_exact = 0;
1697 /* Address of start of the most recently finished expression.
1698 This tells, e.g., postfix * where to find the start of its
1699 operand. Reset at the beginning of groups and alternatives. */
1700 unsigned char *laststart = 0;
1702 /* Address of beginning of regexp, or inside of last group. */
1703 unsigned char *begalt;
1705 /* Place in the uncompiled pattern (i.e., the {) to
1706 which to go back if the interval is invalid. */
1707 const char *beg_interval;
1709 /* Address of the place where a forward jump should go to the end of
1710 the containing expression. Each alternative of an `or' -- except the
1711 last -- ends with a forward jump of this sort. */
1712 unsigned char *fixup_alt_jump = 0;
1714 /* Counts open-groups as they are encountered. Remembered for the
1715 matching close-group on the compile stack, so the same register
1716 number is put in the stop_memory as the start_memory. */
1717 regnum_t regnum = 0;
1720 DEBUG_PRINT1 ("\nCompiling pattern: ");
1723 unsigned debug_count;
1725 for (debug_count = 0; debug_count < size; debug_count++)
1726 putchar (pattern[debug_count]);
1731 /* Initialize the compile stack. */
1732 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1733 if (compile_stack.stack == NULL)
1736 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1737 compile_stack.avail = 0;
1739 /* Initialize the pattern buffer. */
1740 bufp->syntax = syntax;
1741 bufp->fastmap_accurate = 0;
1742 bufp->not_bol = bufp->not_eol = 0;
1744 /* Set `used' to zero, so that if we return an error, the pattern
1745 printer (for debugging) will think there's no pattern. We reset it
1749 /* Always count groups, whether or not bufp->no_sub is set. */
1752 #if !defined (emacs) && !defined (SYNTAX_TABLE)
1753 /* Initialize the syntax table. */
1754 init_syntax_once ();
1757 if (bufp->allocated == 0)
1760 { /* If zero allocated, but buffer is non-null, try to realloc
1761 enough space. This loses if buffer's address is bogus, but
1762 that is the user's responsibility. */
1763 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1766 { /* Caller did not allocate a buffer. Do it for them. */
1767 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1769 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1771 bufp->allocated = INIT_BUF_SIZE;
1774 begalt = b = bufp->buffer;
1776 /* Loop through the uncompiled pattern until we're at the end. */
1785 if ( /* If at start of pattern, it's an operator. */
1787 /* If context independent, it's an operator. */
1788 || syntax & RE_CONTEXT_INDEP_ANCHORS
1789 /* Otherwise, depends on what's come before. */
1790 || at_begline_loc_p (pattern, p, syntax))
1800 if ( /* If at end of pattern, it's an operator. */
1802 /* If context independent, it's an operator. */
1803 || syntax & RE_CONTEXT_INDEP_ANCHORS
1804 /* Otherwise, depends on what's next. */
1805 || at_endline_loc_p (p, pend, syntax))
1815 if ((syntax & RE_BK_PLUS_QM)
1816 || (syntax & RE_LIMITED_OPS))
1820 /* If there is no previous pattern... */
1823 if (syntax & RE_CONTEXT_INVALID_OPS)
1824 FREE_STACK_RETURN (REG_BADRPT);
1825 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1830 /* Are we optimizing this jump? */
1831 boolean keep_string_p = false;
1833 /* 1 means zero (many) matches is allowed. */
1834 char zero_times_ok = 0, many_times_ok = 0;
1836 /* If there is a sequence of repetition chars, collapse it
1837 down to just one (the right one). We can't combine
1838 interval operators with these because of, e.g., `a{2}*',
1839 which should only match an even number of `a's. */
1843 zero_times_ok |= c != '+';
1844 many_times_ok |= c != '?';
1852 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
1855 else if (syntax & RE_BK_PLUS_QM && c == '\\')
1857 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1860 if (!(c1 == '+' || c1 == '?'))
1875 /* If we get here, we found another repeat character. */
1878 /* Star, etc. applied to an empty pattern is equivalent
1879 to an empty pattern. */
1883 /* Now we know whether or not zero matches is allowed
1884 and also whether or not two or more matches is allowed. */
1886 { /* More than one repetition is allowed, so put in at the
1887 end a backward relative jump from `b' to before the next
1888 jump we're going to put in below (which jumps from
1889 laststart to after this jump).
1891 But if we are at the `*' in the exact sequence `.*\n',
1892 insert an unconditional jump backwards to the .,
1893 instead of the beginning of the loop. This way we only
1894 push a failure point once, instead of every time
1895 through the loop. */
1896 assert (p - 1 > pattern);
1898 /* Allocate the space for the jump. */
1899 GET_BUFFER_SPACE (3);
1901 /* We know we are not at the first character of the pattern,
1902 because laststart was nonzero. And we've already
1903 incremented `p', by the way, to be the character after
1904 the `*'. Do we have to do something analogous here
1905 for null bytes, because of RE_DOT_NOT_NULL? */
1906 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
1908 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
1909 && !(syntax & RE_DOT_NEWLINE))
1910 { /* We have .*\n. */
1911 STORE_JUMP (jump, b, laststart);
1912 keep_string_p = true;
1915 /* Anything else. */
1916 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
1918 /* We've added more stuff to the buffer. */
1922 /* On failure, jump from laststart to b + 3, which will be the
1923 end of the buffer after this jump is inserted. */
1924 GET_BUFFER_SPACE (3);
1925 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
1933 /* At least one repetition is required, so insert a
1934 `dummy_failure_jump' before the initial
1935 `on_failure_jump' instruction of the loop. This
1936 effects a skip over that instruction the first time
1937 we hit that loop. */
1938 GET_BUFFER_SPACE (3);
1939 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
1954 boolean had_char_class = false;
1956 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1958 /* Ensure that we have enough space to push a charset: the
1959 opcode, the length count, and the bitset; 34 bytes in all. */
1960 GET_BUFFER_SPACE (34);
1964 /* We test `*p == '^' twice, instead of using an if
1965 statement, so we only need one BUF_PUSH. */
1966 BUF_PUSH (*p == '^' ? charset_not : charset);
1970 /* Remember the first position in the bracket expression. */
1973 /* Push the number of bytes in the bitmap. */
1974 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
1976 /* Clear the whole map. */
1977 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
1979 /* charset_not matches newline according to a syntax bit. */
1980 if ((re_opcode_t) b[-2] == charset_not
1981 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
1982 SET_LIST_BIT ('\n');
1984 /* Read in characters and ranges, setting map bits. */
1987 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1991 /* \ might escape characters inside [...] and [^...]. */
1992 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
1994 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2001 /* Could be the end of the bracket expression. If it's
2002 not (i.e., when the bracket expression is `[]' so
2003 far), the ']' character bit gets set way below. */
2004 if (c == ']' && p != p1 + 1)
2007 /* Look ahead to see if it's a range when the last thing
2008 was a character class. */
2009 if (had_char_class && c == '-' && *p != ']')
2010 FREE_STACK_RETURN (REG_ERANGE);
2012 /* Look ahead to see if it's a range when the last thing
2013 was a character: if this is a hyphen not at the
2014 beginning or the end of a list, then it's the range
2017 && !(p - 2 >= pattern && p[-2] == '[')
2018 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2022 = compile_range (&p, pend, translate, syntax, b);
2023 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2026 else if (p[0] == '-' && p[1] != ']')
2027 { /* This handles ranges made up of characters only. */
2030 /* Move past the `-'. */
2033 ret = compile_range (&p, pend, translate, syntax, b);
2034 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2037 /* See if we're at the beginning of a possible character
2040 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2041 { /* Leave room for the null. */
2042 char str[CHAR_CLASS_MAX_LENGTH + 1];
2047 /* If pattern is `[[:'. */
2048 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2053 if (c == ':' || c == ']' || p == pend
2054 || c1 == CHAR_CLASS_MAX_LENGTH)
2060 /* If isn't a word bracketed by `[:' and:`]':
2061 undo the ending character, the letters, and leave
2062 the leading `:' and `[' (but set bits for them). */
2063 if (c == ':' && *p == ']')
2066 boolean is_alnum = STREQ (str, "alnum");
2067 boolean is_alpha = STREQ (str, "alpha");
2068 boolean is_blank = STREQ (str, "blank");
2069 boolean is_cntrl = STREQ (str, "cntrl");
2070 boolean is_digit = STREQ (str, "digit");
2071 boolean is_graph = STREQ (str, "graph");
2072 boolean is_lower = STREQ (str, "lower");
2073 boolean is_print = STREQ (str, "print");
2074 boolean is_punct = STREQ (str, "punct");
2075 boolean is_space = STREQ (str, "space");
2076 boolean is_upper = STREQ (str, "upper");
2077 boolean is_xdigit = STREQ (str, "xdigit");
2079 if (!IS_CHAR_CLASS (str))
2080 FREE_STACK_RETURN (REG_ECTYPE);
2082 /* Throw away the ] at the end of the character
2086 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2088 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2090 /* This was split into 3 if's to
2091 avoid an arbitrary limit in some compiler. */
2092 if ( (is_alnum && ISALNUM (ch))
2093 || (is_alpha && ISALPHA (ch))
2094 || (is_blank && ISBLANK (ch))
2095 || (is_cntrl && ISCNTRL (ch)))
2097 if ( (is_digit && ISDIGIT (ch))
2098 || (is_graph && ISGRAPH (ch))
2099 || (is_lower && ISLOWER (ch))
2100 || (is_print && ISPRINT (ch)))
2102 if ( (is_punct && ISPUNCT (ch))
2103 || (is_space && ISSPACE (ch))
2104 || (is_upper && ISUPPER (ch))
2105 || (is_xdigit && ISXDIGIT (ch)))
2108 had_char_class = true;
2117 had_char_class = false;
2122 had_char_class = false;
2127 /* Discard any (non)matching list bytes that are all 0 at the
2128 end of the map. Decrease the map-length byte too. */
2129 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2137 if (syntax & RE_NO_BK_PARENS)
2144 if (syntax & RE_NO_BK_PARENS)
2151 if (syntax & RE_NEWLINE_ALT)
2158 if (syntax & RE_NO_BK_VBAR)
2165 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2166 goto handle_interval;
2172 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2174 /* Do not translate the character after the \, so that we can
2175 distinguish, e.g., \B from \b, even if we normally would
2176 translate, e.g., B to b. */
2182 if (syntax & RE_NO_BK_PARENS)
2183 goto normal_backslash;
2189 if (COMPILE_STACK_FULL)
2191 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2192 compile_stack_elt_t);
2193 if (compile_stack.stack == NULL) return REG_ESPACE;
2195 compile_stack.size <<= 1;
2198 /* These are the values to restore when we hit end of this
2199 group. They are all relative offsets, so that if the
2200 whole pattern moves because of realloc, they will still
2202 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2203 COMPILE_STACK_TOP.fixup_alt_jump
2204 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2205 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2206 COMPILE_STACK_TOP.regnum = regnum;
2208 /* We will eventually replace the 0 with the number of
2209 groups inner to this one. But do not push a
2210 start_memory for groups beyond the last one we can
2211 represent in the compiled pattern. */
2212 if (regnum <= MAX_REGNUM)
2214 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2215 BUF_PUSH_3 (start_memory, regnum, 0);
2218 compile_stack.avail++;
2223 /* If we've reached MAX_REGNUM groups, then this open
2224 won't actually generate any code, so we'll have to
2225 clear pending_exact explicitly. */
2231 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2233 if (COMPILE_STACK_EMPTY)
2234 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2235 goto normal_backslash;
2237 FREE_STACK_RETURN (REG_ERPAREN);
2241 { /* Push a dummy failure point at the end of the
2242 alternative for a possible future
2243 `pop_failure_jump' to pop. See comments at
2244 `push_dummy_failure' in `re_match_2'. */
2245 BUF_PUSH (push_dummy_failure);
2247 /* We allocated space for this jump when we assigned
2248 to `fixup_alt_jump', in the `handle_alt' case below. */
2249 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2252 /* See similar code for backslashed left paren above. */
2253 if (COMPILE_STACK_EMPTY)
2254 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2257 FREE_STACK_RETURN (REG_ERPAREN);
2259 /* Since we just checked for an empty stack above, this
2260 ``can't happen''. */
2261 assert (compile_stack.avail != 0);
2263 /* We don't just want to restore into `regnum', because
2264 later groups should continue to be numbered higher,
2265 as in `(ab)c(de)' -- the second group is #2. */
2266 regnum_t this_group_regnum;
2268 compile_stack.avail--;
2269 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2271 = COMPILE_STACK_TOP.fixup_alt_jump
2272 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2274 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2275 this_group_regnum = COMPILE_STACK_TOP.regnum;
2276 /* If we've reached MAX_REGNUM groups, then this open
2277 won't actually generate any code, so we'll have to
2278 clear pending_exact explicitly. */
2281 /* We're at the end of the group, so now we know how many
2282 groups were inside this one. */
2283 if (this_group_regnum <= MAX_REGNUM)
2285 unsigned char *inner_group_loc
2286 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2288 *inner_group_loc = regnum - this_group_regnum;
2289 BUF_PUSH_3 (stop_memory, this_group_regnum,
2290 regnum - this_group_regnum);
2296 case '|': /* `\|'. */
2297 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2298 goto normal_backslash;
2300 if (syntax & RE_LIMITED_OPS)
2303 /* Insert before the previous alternative a jump which
2304 jumps to this alternative if the former fails. */
2305 GET_BUFFER_SPACE (3);
2306 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2310 /* The alternative before this one has a jump after it
2311 which gets executed if it gets matched. Adjust that
2312 jump so it will jump to this alternative's analogous
2313 jump (put in below, which in turn will jump to the next
2314 (if any) alternative's such jump, etc.). The last such
2315 jump jumps to the correct final destination. A picture:
2321 If we are at `b', then fixup_alt_jump right now points to a
2322 three-byte space after `a'. We'll put in the jump, set
2323 fixup_alt_jump to right after `b', and leave behind three
2324 bytes which we'll fill in when we get to after `c'. */
2327 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2329 /* Mark and leave space for a jump after this alternative,
2330 to be filled in later either by next alternative or
2331 when know we're at the end of a series of alternatives. */
2333 GET_BUFFER_SPACE (3);
2342 /* If \{ is a literal. */
2343 if (!(syntax & RE_INTERVALS)
2344 /* If we're at `\{' and it's not the open-interval
2346 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2347 || (p - 2 == pattern && p == pend))
2348 goto normal_backslash;
2352 /* If got here, then the syntax allows intervals. */
2354 /* At least (most) this many matches must be made. */
2355 int lower_bound = -1, upper_bound = -1;
2357 beg_interval = p - 1;
2361 if (syntax & RE_NO_BK_BRACES)
2362 goto unfetch_interval;
2364 FREE_STACK_RETURN (REG_EBRACE);
2367 GET_UNSIGNED_NUMBER (lower_bound);
2371 GET_UNSIGNED_NUMBER (upper_bound);
2372 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2375 /* Interval such as `{1}' => match exactly once. */
2376 upper_bound = lower_bound;
2378 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2379 || lower_bound > upper_bound)
2381 if (syntax & RE_NO_BK_BRACES)
2382 goto unfetch_interval;
2384 FREE_STACK_RETURN (REG_BADBR);
2387 if (!(syntax & RE_NO_BK_BRACES))
2389 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2396 if (syntax & RE_NO_BK_BRACES)
2397 goto unfetch_interval;
2399 FREE_STACK_RETURN (REG_BADBR);
2402 /* We just parsed a valid interval. */
2404 /* If it's invalid to have no preceding re. */
2407 if (syntax & RE_CONTEXT_INVALID_OPS)
2408 FREE_STACK_RETURN (REG_BADRPT);
2409 else if (syntax & RE_CONTEXT_INDEP_OPS)
2412 goto unfetch_interval;
2415 /* If the upper bound is zero, don't want to succeed at
2416 all; jump from `laststart' to `b + 3', which will be
2417 the end of the buffer after we insert the jump. */
2418 if (upper_bound == 0)
2420 GET_BUFFER_SPACE (3);
2421 INSERT_JUMP (jump, laststart, b + 3);
2425 /* Otherwise, we have a nontrivial interval. When
2426 we're all done, the pattern will look like:
2427 set_number_at <jump count> <upper bound>
2428 set_number_at <succeed_n count> <lower bound>
2429 succeed_n <after jump addr> <succeed_n count>
2431 jump_n <succeed_n addr> <jump count>
2432 (The upper bound and `jump_n' are omitted if
2433 `upper_bound' is 1, though.) */
2435 { /* If the upper bound is > 1, we need to insert
2436 more at the end of the loop. */
2437 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2439 GET_BUFFER_SPACE (nbytes);
2441 /* Initialize lower bound of the `succeed_n', even
2442 though it will be set during matching by its
2443 attendant `set_number_at' (inserted next),
2444 because `re_compile_fastmap' needs to know.
2445 Jump to the `jump_n' we might insert below. */
2446 INSERT_JUMP2 (succeed_n, laststart,
2447 b + 5 + (upper_bound > 1) * 5,
2451 /* Code to initialize the lower bound. Insert
2452 before the `succeed_n'. The `5' is the last two
2453 bytes of this `set_number_at', plus 3 bytes of
2454 the following `succeed_n'. */
2455 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2458 if (upper_bound > 1)
2459 { /* More than one repetition is allowed, so
2460 append a backward jump to the `succeed_n'
2461 that starts this interval.
2463 When we've reached this during matching,
2464 we'll have matched the interval once, so
2465 jump back only `upper_bound - 1' times. */
2466 STORE_JUMP2 (jump_n, b, laststart + 5,
2470 /* The location we want to set is the second
2471 parameter of the `jump_n'; that is `b-2' as
2472 an absolute address. `laststart' will be
2473 the `set_number_at' we're about to insert;
2474 `laststart+3' the number to set, the source
2475 for the relative address. But we are
2476 inserting into the middle of the pattern --
2477 so everything is getting moved up by 5.
2478 Conclusion: (b - 2) - (laststart + 3) + 5,
2479 i.e., b - laststart.
2481 We insert this at the beginning of the loop
2482 so that if we fail during matching, we'll
2483 reinitialize the bounds. */
2484 insert_op2 (set_number_at, laststart, b - laststart,
2485 upper_bound - 1, b);
2490 beg_interval = NULL;
2495 /* If an invalid interval, match the characters as literals. */
2496 assert (beg_interval);
2498 beg_interval = NULL;
2500 /* normal_char and normal_backslash need `c'. */
2503 if (!(syntax & RE_NO_BK_BRACES))
2505 if (p > pattern && p[-1] == '\\')
2506 goto normal_backslash;
2511 /* There is no way to specify the before_dot and after_dot
2512 operators. rms says this is ok. --karl */
2520 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2526 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2533 BUF_PUSH (wordchar);
2539 BUF_PUSH (notwordchar);
2552 BUF_PUSH (wordbound);
2556 BUF_PUSH (notwordbound);
2567 case '1': case '2': case '3': case '4': case '5':
2568 case '6': case '7': case '8': case '9':
2569 if (syntax & RE_NO_BK_REFS)
2575 FREE_STACK_RETURN (REG_ESUBREG);
2577 /* Can't back reference to a subexpression if inside of it. */
2578 if (group_in_compile_stack (compile_stack, c1))
2582 BUF_PUSH_2 (duplicate, c1);
2588 if (syntax & RE_BK_PLUS_QM)
2591 goto normal_backslash;
2595 /* You might think it would be useful for \ to mean
2596 not to translate; but if we don't translate it
2597 it will never match anything. */
2605 /* Expects the character in `c'. */
2607 /* If no exactn currently being built. */
2610 /* If last exactn not at current position. */
2611 || pending_exact + *pending_exact + 1 != b
2613 /* We have only one byte following the exactn for the count. */
2614 || *pending_exact == (1 << BYTEWIDTH) - 1
2616 /* If followed by a repetition operator. */
2617 || *p == '*' || *p == '^'
2618 || ((syntax & RE_BK_PLUS_QM)
2619 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2620 : (*p == '+' || *p == '?'))
2621 || ((syntax & RE_INTERVALS)
2622 && ((syntax & RE_NO_BK_BRACES)
2624 : (p[0] == '\\' && p[1] == '{'))))
2626 /* Start building a new exactn. */
2630 BUF_PUSH_2 (exactn, 0);
2631 pending_exact = b - 1;
2638 } /* while p != pend */
2641 /* Through the pattern now. */
2644 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2646 if (!COMPILE_STACK_EMPTY)
2647 FREE_STACK_RETURN (REG_EPAREN);
2649 /* If we don't want backtracking, force success
2650 the first time we reach the end of the compiled pattern. */
2651 if (syntax & RE_NO_POSIX_BACKTRACKING)
2654 free (compile_stack.stack);
2656 /* We have succeeded; set the length of the buffer. */
2657 bufp->used = b - bufp->buffer;
2662 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2663 print_compiled_pattern (bufp);
2667 #ifndef MATCH_MAY_ALLOCATE
2668 /* Initialize the failure stack to the largest possible stack. This
2669 isn't necessary unless we're trying to avoid calling alloca in
2670 the search and match routines. */
2672 int num_regs = bufp->re_nsub + 1;
2674 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2675 is strictly greater than re_max_failures, the largest possible stack
2676 is 2 * re_max_failures failure points. */
2677 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2679 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2682 if (! fail_stack.stack)
2684 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2685 * sizeof (fail_stack_elt_t));
2688 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2690 * sizeof (fail_stack_elt_t)));
2691 #else /* not emacs */
2692 if (! fail_stack.stack)
2694 = (fail_stack_elt_t *) malloc (fail_stack.size
2695 * sizeof (fail_stack_elt_t));
2698 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2700 * sizeof (fail_stack_elt_t)));
2701 #endif /* not emacs */
2704 regex_grow_registers (num_regs);
2706 #endif /* not MATCH_MAY_ALLOCATE */
2709 } /* regex_compile */
2711 /* Subroutines for `regex_compile'. */
2713 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2716 store_op1 (op, loc, arg)
2721 *loc = (unsigned char) op;
2722 STORE_NUMBER (loc + 1, arg);
2726 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2729 store_op2 (op, loc, arg1, arg2)
2734 *loc = (unsigned char) op;
2735 STORE_NUMBER (loc + 1, arg1);
2736 STORE_NUMBER (loc + 3, arg2);
2740 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
2741 for OP followed by two-byte integer parameter ARG. */
2744 insert_op1 (op, loc, arg, end)
2750 register unsigned char *pfrom = end;
2751 register unsigned char *pto = end + 3;
2753 while (pfrom != loc)
2756 store_op1 (op, loc, arg);
2760 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2763 insert_op2 (op, loc, arg1, arg2, end)
2769 register unsigned char *pfrom = end;
2770 register unsigned char *pto = end + 5;
2772 while (pfrom != loc)
2775 store_op2 (op, loc, arg1, arg2);
2779 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
2780 after an alternative or a begin-subexpression. We assume there is at
2781 least one character before the ^. */
2784 at_begline_loc_p (pattern, p, syntax)
2785 const char *pattern, *p;
2786 reg_syntax_t syntax;
2788 const char *prev = p - 2;
2789 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
2792 /* After a subexpression? */
2793 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
2794 /* After an alternative? */
2795 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
2799 /* The dual of at_begline_loc_p. This one is for $. We assume there is
2800 at least one character after the $, i.e., `P < PEND'. */
2803 at_endline_loc_p (p, pend, syntax)
2804 const char *p, *pend;
2807 const char *next = p;
2808 boolean next_backslash = *next == '\\';
2809 const char *next_next = p + 1 < pend ? p + 1 : 0;
2812 /* Before a subexpression? */
2813 (syntax & RE_NO_BK_PARENS ? *next == ')'
2814 : next_backslash && next_next && *next_next == ')')
2815 /* Before an alternative? */
2816 || (syntax & RE_NO_BK_VBAR ? *next == '|'
2817 : next_backslash && next_next && *next_next == '|');
2821 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
2822 false if it's not. */
2825 group_in_compile_stack (compile_stack, regnum)
2826 compile_stack_type compile_stack;
2831 for (this_element = compile_stack.avail - 1;
2834 if (compile_stack.stack[this_element].regnum == regnum)
2841 /* Read the ending character of a range (in a bracket expression) from the
2842 uncompiled pattern *P_PTR (which ends at PEND). We assume the
2843 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
2844 Then we set the translation of all bits between the starting and
2845 ending characters (inclusive) in the compiled pattern B.
2847 Return an error code.
2849 We use these short variable names so we can use the same macros as
2850 `regex_compile' itself. */
2852 static reg_errcode_t
2853 compile_range (p_ptr, pend, translate, syntax, b)
2854 const char **p_ptr, *pend;
2855 RE_TRANSLATE_TYPE translate;
2856 reg_syntax_t syntax;
2861 const char *p = *p_ptr;
2862 unsigned int range_start, range_end;
2867 /* Even though the pattern is a signed `char *', we need to fetch
2868 with unsigned char *'s; if the high bit of the pattern character
2869 is set, the range endpoints will be negative if we fetch using a
2872 We also want to fetch the endpoints without translating them; the
2873 appropriate translation is done in the bit-setting loop below. */
2874 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
2875 range_start = ((const unsigned char *) p)[-2];
2876 range_end = ((const unsigned char *) p)[0];
2878 /* Have to increment the pointer into the pattern string, so the
2879 caller isn't still at the ending character. */
2882 /* If the start is after the end, the range is empty. */
2883 if (range_start > range_end)
2884 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
2886 /* Here we see why `this_char' has to be larger than an `unsigned
2887 char' -- the range is inclusive, so if `range_end' == 0xff
2888 (assuming 8-bit characters), we would otherwise go into an infinite
2889 loop, since all characters <= 0xff. */
2890 for (this_char = range_start; this_char <= range_end; this_char++)
2892 SET_LIST_BIT (TRANSLATE (this_char));
2898 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
2899 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
2900 characters can start a string that matches the pattern. This fastmap
2901 is used by re_search to skip quickly over impossible starting points.
2903 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
2904 area as BUFP->fastmap.
2906 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
2909 Returns 0 if we succeed, -2 if an internal error. */
2912 re_compile_fastmap (bufp)
2913 struct re_pattern_buffer *bufp;
2916 #ifdef MATCH_MAY_ALLOCATE
2917 fail_stack_type fail_stack;
2919 #ifndef REGEX_MALLOC
2922 /* We don't push any register information onto the failure stack. */
2923 unsigned num_regs = 0;
2925 register char *fastmap = bufp->fastmap;
2926 unsigned char *pattern = bufp->buffer;
2927 unsigned long size = bufp->used;
2928 unsigned char *p = pattern;
2929 register unsigned char *pend = pattern + size;
2931 /* This holds the pointer to the failure stack, when
2932 it is allocated relocatably. */
2933 fail_stack_elt_t *failure_stack_ptr;
2935 /* Assume that each path through the pattern can be null until
2936 proven otherwise. We set this false at the bottom of switch
2937 statement, to which we get only if a particular path doesn't
2938 match the empty string. */
2939 boolean path_can_be_null = true;
2941 /* We aren't doing a `succeed_n' to begin with. */
2942 boolean succeed_n_p = false;
2944 assert (fastmap != NULL && p != NULL);
2947 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
2948 bufp->fastmap_accurate = 1; /* It will be when we're done. */
2949 bufp->can_be_null = 0;
2953 if (p == pend || *p == succeed)
2955 /* We have reached the (effective) end of pattern. */
2956 if (!FAIL_STACK_EMPTY ())
2958 bufp->can_be_null |= path_can_be_null;
2960 /* Reset for next path. */
2961 path_can_be_null = true;
2963 p = fail_stack.stack[--fail_stack.avail].pointer;
2971 /* We should never be about to go beyond the end of the pattern. */
2974 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
2977 /* I guess the idea here is to simply not bother with a fastmap
2978 if a backreference is used, since it's too hard to figure out
2979 the fastmap for the corresponding group. Setting
2980 `can_be_null' stops `re_search_2' from using the fastmap, so
2981 that is all we do. */
2983 bufp->can_be_null = 1;
2987 /* Following are the cases which match a character. These end
2996 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
2997 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3003 /* Chars beyond end of map must be allowed. */
3004 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3007 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3008 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3014 for (j = 0; j < (1 << BYTEWIDTH); j++)
3015 if (SYNTAX (j) == Sword)
3021 for (j = 0; j < (1 << BYTEWIDTH); j++)
3022 if (SYNTAX (j) != Sword)
3029 int fastmap_newline = fastmap['\n'];
3031 /* `.' matches anything ... */
3032 for (j = 0; j < (1 << BYTEWIDTH); j++)
3035 /* ... except perhaps newline. */
3036 if (!(bufp->syntax & RE_DOT_NEWLINE))
3037 fastmap['\n'] = fastmap_newline;
3039 /* Return if we have already set `can_be_null'; if we have,
3040 then the fastmap is irrelevant. Something's wrong here. */
3041 else if (bufp->can_be_null)
3044 /* Otherwise, have to check alternative paths. */
3051 for (j = 0; j < (1 << BYTEWIDTH); j++)
3052 if (SYNTAX (j) == (enum syntaxcode) k)
3059 for (j = 0; j < (1 << BYTEWIDTH); j++)
3060 if (SYNTAX (j) != (enum syntaxcode) k)
3065 /* All cases after this match the empty string. These end with
3085 case push_dummy_failure:
3090 case pop_failure_jump:
3091 case maybe_pop_jump:
3094 case dummy_failure_jump:
3095 EXTRACT_NUMBER_AND_INCR (j, p);
3100 /* Jump backward implies we just went through the body of a
3101 loop and matched nothing. Opcode jumped to should be
3102 `on_failure_jump' or `succeed_n'. Just treat it like an
3103 ordinary jump. For a * loop, it has pushed its failure
3104 point already; if so, discard that as redundant. */
3105 if ((re_opcode_t) *p != on_failure_jump
3106 && (re_opcode_t) *p != succeed_n)
3110 EXTRACT_NUMBER_AND_INCR (j, p);
3113 /* If what's on the stack is where we are now, pop it. */
3114 if (!FAIL_STACK_EMPTY ()
3115 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3121 case on_failure_jump:
3122 case on_failure_keep_string_jump:
3123 handle_on_failure_jump:
3124 EXTRACT_NUMBER_AND_INCR (j, p);
3126 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3127 end of the pattern. We don't want to push such a point,
3128 since when we restore it above, entering the switch will
3129 increment `p' past the end of the pattern. We don't need
3130 to push such a point since we obviously won't find any more
3131 fastmap entries beyond `pend'. Such a pattern can match
3132 the null string, though. */
3135 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3137 RESET_FAIL_STACK ();
3142 bufp->can_be_null = 1;
3146 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3147 succeed_n_p = false;
3154 /* Get to the number of times to succeed. */
3157 /* Increment p past the n for when k != 0. */
3158 EXTRACT_NUMBER_AND_INCR (k, p);
3162 succeed_n_p = true; /* Spaghetti code alert. */
3163 goto handle_on_failure_jump;
3180 abort (); /* We have listed all the cases. */
3183 /* Getting here means we have found the possible starting
3184 characters for one path of the pattern -- and that the empty
3185 string does not match. We need not follow this path further.
3186 Instead, look at the next alternative (remembered on the
3187 stack), or quit if no more. The test at the top of the loop
3188 does these things. */
3189 path_can_be_null = false;
3193 /* Set `can_be_null' for the last path (also the first path, if the
3194 pattern is empty). */
3195 bufp->can_be_null |= path_can_be_null;
3198 RESET_FAIL_STACK ();
3200 } /* re_compile_fastmap */
3202 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3203 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3204 this memory for recording register information. STARTS and ENDS
3205 must be allocated using the malloc library routine, and must each
3206 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3208 If NUM_REGS == 0, then subsequent matches should allocate their own
3211 Unless this function is called, the first search or match using
3212 PATTERN_BUFFER will allocate its own register data, without
3213 freeing the old data. */
3216 re_set_registers (bufp, regs, num_regs, starts, ends)
3217 struct re_pattern_buffer *bufp;
3218 struct re_registers *regs;
3220 regoff_t *starts, *ends;
3224 bufp->regs_allocated = REGS_REALLOCATE;
3225 regs->num_regs = num_regs;
3226 regs->start = starts;
3231 bufp->regs_allocated = REGS_UNALLOCATED;
3233 regs->start = regs->end = (regoff_t *) 0;
3237 /* Searching routines. */
3239 /* Like re_search_2, below, but only one string is specified, and
3240 doesn't let you say where to stop matching. */
3243 re_search (bufp, string, size, startpos, range, regs)
3244 struct re_pattern_buffer *bufp;
3246 int size, startpos, range;
3247 struct re_registers *regs;
3249 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3254 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3255 virtual concatenation of STRING1 and STRING2, starting first at index
3256 STARTPOS, then at STARTPOS + 1, and so on.
3258 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3260 RANGE is how far to scan while trying to match. RANGE = 0 means try
3261 only at STARTPOS; in general, the last start tried is STARTPOS +
3264 In REGS, return the indices of the virtual concatenation of STRING1
3265 and STRING2 that matched the entire BUFP->buffer and its contained
3268 Do not consider matching one past the index STOP in the virtual
3269 concatenation of STRING1 and STRING2.
3271 We return either the position in the strings at which the match was
3272 found, -1 if no match, or -2 if error (such as failure
3276 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3277 struct re_pattern_buffer *bufp;
3278 const char *string1, *string2;
3282 struct re_registers *regs;
3286 register char *fastmap = bufp->fastmap;
3287 register RE_TRANSLATE_TYPE translate = bufp->translate;
3288 int total_size = size1 + size2;
3289 int endpos = startpos + range;
3291 /* Check for out-of-range STARTPOS. */
3292 if (startpos < 0 || startpos > total_size)
3295 /* Fix up RANGE if it might eventually take us outside
3296 the virtual concatenation of STRING1 and STRING2.
3297 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3299 range = 0 - startpos;
3300 else if (endpos > total_size)
3301 range = total_size - startpos;
3303 /* If the search isn't to be a backwards one, don't waste time in a
3304 search for a pattern that must be anchored. */
3305 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3314 /* In a forward search for something that starts with \=.
3315 don't keep searching past point. */
3316 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3318 range = PT - startpos;
3324 /* Update the fastmap now if not correct already. */
3325 if (fastmap && !bufp->fastmap_accurate)
3326 if (re_compile_fastmap (bufp) == -2)
3329 /* Loop through the string, looking for a place to start matching. */
3332 /* If a fastmap is supplied, skip quickly over characters that
3333 cannot be the start of a match. If the pattern can match the
3334 null string, however, we don't need to skip characters; we want
3335 the first null string. */
3336 if (fastmap && startpos < total_size && !bufp->can_be_null)
3338 if (range > 0) /* Searching forwards. */
3340 register const char *d;
3341 register int lim = 0;
3344 if (startpos < size1 && startpos + range >= size1)
3345 lim = range - (size1 - startpos);
3347 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3349 /* Written out as an if-else to avoid testing `translate'
3353 && !fastmap[(unsigned char)
3354 translate[(unsigned char) *d++]])
3357 while (range > lim && !fastmap[(unsigned char) *d++])
3360 startpos += irange - range;
3362 else /* Searching backwards. */
3364 register char c = (size1 == 0 || startpos >= size1
3365 ? string2[startpos - size1]
3366 : string1[startpos]);
3368 if (!fastmap[(unsigned char) TRANSLATE (c)])
3373 /* If can't match the null string, and that's all we have left, fail. */
3374 if (range >= 0 && startpos == total_size && fastmap
3375 && !bufp->can_be_null)
3378 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3379 startpos, regs, stop);
3380 #ifndef REGEX_MALLOC
3409 /* Declarations and macros for re_match_2. */
3411 static int bcmp_translate ();
3412 static boolean alt_match_null_string_p (),
3413 common_op_match_null_string_p (),
3414 group_match_null_string_p ();
3416 /* This converts PTR, a pointer into one of the search strings `string1'
3417 and `string2' into an offset from the beginning of that string. */
3418 #define POINTER_TO_OFFSET(ptr) \
3419 (FIRST_STRING_P (ptr) \
3420 ? ((regoff_t) ((ptr) - string1)) \
3421 : ((regoff_t) ((ptr) - string2 + size1)))
3423 /* Macros for dealing with the split strings in re_match_2. */
3425 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3427 /* Call before fetching a character with *d. This switches over to
3428 string2 if necessary. */
3429 #define PREFETCH() \
3432 /* End of string2 => fail. */ \
3433 if (dend == end_match_2) \
3435 /* End of string1 => advance to string2. */ \
3437 dend = end_match_2; \
3441 /* Test if at very beginning or at very end of the virtual concatenation
3442 of `string1' and `string2'. If only one string, it's `string2'. */
3443 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3444 #define AT_STRINGS_END(d) ((d) == end2)
3447 /* Test if D points to a character which is word-constituent. We have
3448 two special cases to check for: if past the end of string1, look at
3449 the first character in string2; and if before the beginning of
3450 string2, look at the last character in string1. */
3451 #define WORDCHAR_P(d) \
3452 (SYNTAX ((d) == end1 ? *string2 \
3453 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3456 /* Disabled due to a compiler bug -- see comment at case wordbound */
3458 /* Test if the character before D and the one at D differ with respect
3459 to being word-constituent. */
3460 #define AT_WORD_BOUNDARY(d) \
3461 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3462 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3465 /* Free everything we malloc. */
3466 #ifdef MATCH_MAY_ALLOCATE
3467 #define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3468 #define FREE_VARIABLES() \
3470 REGEX_FREE_STACK (fail_stack.stack); \
3471 FREE_VAR (regstart); \
3472 FREE_VAR (regend); \
3473 FREE_VAR (old_regstart); \
3474 FREE_VAR (old_regend); \
3475 FREE_VAR (best_regstart); \
3476 FREE_VAR (best_regend); \
3477 FREE_VAR (reg_info); \
3478 FREE_VAR (reg_dummy); \
3479 FREE_VAR (reg_info_dummy); \
3482 #define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3483 #endif /* not MATCH_MAY_ALLOCATE */
3485 /* These values must meet several constraints. They must not be valid
3486 register values; since we have a limit of 255 registers (because
3487 we use only one byte in the pattern for the register number), we can
3488 use numbers larger than 255. They must differ by 1, because of
3489 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3490 be larger than the value for the highest register, so we do not try
3491 to actually save any registers when none are active. */
3492 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3493 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3495 /* Matching routines. */
3497 #ifndef emacs /* Emacs never uses this. */
3498 /* re_match is like re_match_2 except it takes only a single string. */
3501 re_match (bufp, string, size, pos, regs)
3502 struct re_pattern_buffer *bufp;
3505 struct re_registers *regs;
3507 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3512 #endif /* not emacs */
3515 /* re_match_2 matches the compiled pattern in BUFP against the
3516 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3517 and SIZE2, respectively). We start matching at POS, and stop
3520 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3521 store offsets for the substring each group matched in REGS. See the
3522 documentation for exactly how many groups we fill.
3524 We return -1 if no match, -2 if an internal error (such as the
3525 failure stack overflowing). Otherwise, we return the length of the
3526 matched substring. */
3529 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3530 struct re_pattern_buffer *bufp;
3531 const char *string1, *string2;
3534 struct re_registers *regs;
3537 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3543 /* This is a separate function so that we can force an alloca cleanup
3546 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3547 struct re_pattern_buffer *bufp;
3548 const char *string1, *string2;
3551 struct re_registers *regs;
3554 /* General temporaries. */
3558 /* Just past the end of the corresponding string. */
3559 const char *end1, *end2;
3561 /* Pointers into string1 and string2, just past the last characters in
3562 each to consider matching. */
3563 const char *end_match_1, *end_match_2;
3565 /* Where we are in the data, and the end of the current string. */
3566 const char *d, *dend;
3568 /* Where we are in the pattern, and the end of the pattern. */
3569 unsigned char *p = bufp->buffer;
3570 register unsigned char *pend = p + bufp->used;
3572 /* Mark the opcode just after a start_memory, so we can test for an
3573 empty subpattern when we get to the stop_memory. */
3574 unsigned char *just_past_start_mem = 0;
3576 /* We use this to map every character in the string. */
3577 RE_TRANSLATE_TYPE translate = bufp->translate;
3579 /* Failure point stack. Each place that can handle a failure further
3580 down the line pushes a failure point on this stack. It consists of
3581 restart, regend, and reg_info for all registers corresponding to
3582 the subexpressions we're currently inside, plus the number of such
3583 registers, and, finally, two char *'s. The first char * is where
3584 to resume scanning the pattern; the second one is where to resume
3585 scanning the strings. If the latter is zero, the failure point is
3586 a ``dummy''; if a failure happens and the failure point is a dummy,
3587 it gets discarded and the next next one is tried. */
3588 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3589 fail_stack_type fail_stack;
3592 static unsigned failure_id = 0;
3593 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3596 /* This holds the pointer to the failure stack, when
3597 it is allocated relocatably. */
3598 fail_stack_elt_t *failure_stack_ptr;
3600 /* We fill all the registers internally, independent of what we
3601 return, for use in backreferences. The number here includes
3602 an element for register zero. */
3603 unsigned num_regs = bufp->re_nsub + 1;
3605 /* The currently active registers. */
3606 unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3607 unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3609 /* Information on the contents of registers. These are pointers into
3610 the input strings; they record just what was matched (on this
3611 attempt) by a subexpression part of the pattern, that is, the
3612 regnum-th regstart pointer points to where in the pattern we began
3613 matching and the regnum-th regend points to right after where we
3614 stopped matching the regnum-th subexpression. (The zeroth register
3615 keeps track of what the whole pattern matches.) */
3616 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3617 const char **regstart, **regend;
3620 /* If a group that's operated upon by a repetition operator fails to
3621 match anything, then the register for its start will need to be
3622 restored because it will have been set to wherever in the string we
3623 are when we last see its open-group operator. Similarly for a
3625 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3626 const char **old_regstart, **old_regend;
3629 /* The is_active field of reg_info helps us keep track of which (possibly
3630 nested) subexpressions we are currently in. The matched_something
3631 field of reg_info[reg_num] helps us tell whether or not we have
3632 matched any of the pattern so far this time through the reg_num-th
3633 subexpression. These two fields get reset each time through any
3634 loop their register is in. */
3635 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3636 register_info_type *reg_info;
3639 /* The following record the register info as found in the above
3640 variables when we find a match better than any we've seen before.
3641 This happens as we backtrack through the failure points, which in
3642 turn happens only if we have not yet matched the entire string. */
3643 unsigned best_regs_set = false;
3644 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3645 const char **best_regstart, **best_regend;
3648 /* Logically, this is `best_regend[0]'. But we don't want to have to
3649 allocate space for that if we're not allocating space for anything
3650 else (see below). Also, we never need info about register 0 for
3651 any of the other register vectors, and it seems rather a kludge to
3652 treat `best_regend' differently than the rest. So we keep track of
3653 the end of the best match so far in a separate variable. We
3654 initialize this to NULL so that when we backtrack the first time
3655 and need to test it, it's not garbage. */
3656 const char *match_end = NULL;
3658 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3659 int set_regs_matched_done = 0;
3661 /* Used when we pop values we don't care about. */
3662 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3663 const char **reg_dummy;
3664 register_info_type *reg_info_dummy;
3668 /* Counts the total number of registers pushed. */
3669 unsigned num_regs_pushed = 0;
3672 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3676 #ifdef MATCH_MAY_ALLOCATE
3677 /* Do not bother to initialize all the register variables if there are
3678 no groups in the pattern, as it takes a fair amount of time. If
3679 there are groups, we include space for register 0 (the whole
3680 pattern), even though we never use it, since it simplifies the
3681 array indexing. We should fix this. */
3684 regstart = REGEX_TALLOC (num_regs, const char *);
3685 regend = REGEX_TALLOC (num_regs, const char *);
3686 old_regstart = REGEX_TALLOC (num_regs, const char *);
3687 old_regend = REGEX_TALLOC (num_regs, const char *);
3688 best_regstart = REGEX_TALLOC (num_regs, const char *);
3689 best_regend = REGEX_TALLOC (num_regs, const char *);
3690 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3691 reg_dummy = REGEX_TALLOC (num_regs, const char *);
3692 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3694 if (!(regstart && regend && old_regstart && old_regend && reg_info
3695 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3703 /* We must initialize all our variables to NULL, so that
3704 `FREE_VARIABLES' doesn't try to free them. */
3705 regstart = regend = old_regstart = old_regend = best_regstart
3706 = best_regend = reg_dummy = NULL;
3707 reg_info = reg_info_dummy = (register_info_type *) NULL;
3709 #endif /* MATCH_MAY_ALLOCATE */
3711 /* The starting position is bogus. */
3712 if (pos < 0 || pos > size1 + size2)
3718 /* Initialize subexpression text positions to -1 to mark ones that no
3719 start_memory/stop_memory has been seen for. Also initialize the
3720 register information struct. */
3721 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
3723 regstart[mcnt] = regend[mcnt]
3724 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3726 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3727 IS_ACTIVE (reg_info[mcnt]) = 0;
3728 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3729 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3732 /* We move `string1' into `string2' if the latter's empty -- but not if
3733 `string1' is null. */
3734 if (size2 == 0 && string1 != NULL)
3741 end1 = string1 + size1;
3742 end2 = string2 + size2;
3744 /* Compute where to stop matching, within the two strings. */
3747 end_match_1 = string1 + stop;
3748 end_match_2 = string2;
3753 end_match_2 = string2 + stop - size1;
3756 /* `p' scans through the pattern as `d' scans through the data.
3757 `dend' is the end of the input string that `d' points within. `d'
3758 is advanced into the following input string whenever necessary, but
3759 this happens before fetching; therefore, at the beginning of the
3760 loop, `d' can be pointing at the end of a string, but it cannot
3762 if (size1 > 0 && pos <= size1)
3769 d = string2 + pos - size1;
3773 DEBUG_PRINT1 ("The compiled pattern is: ");
3774 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
3775 DEBUG_PRINT1 ("The string to match is: `");
3776 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
3777 DEBUG_PRINT1 ("'\n");
3779 /* This loops over pattern commands. It exits by returning from the
3780 function if the match is complete, or it drops through if the match
3781 fails at this starting point in the input data. */
3784 DEBUG_PRINT2 ("\n0x%x: ", p);
3787 { /* End of pattern means we might have succeeded. */
3788 DEBUG_PRINT1 ("end of pattern ... ");
3790 /* If we haven't matched the entire string, and we want the
3791 longest match, try backtracking. */
3792 if (d != end_match_2)
3794 /* 1 if this match ends in the same string (string1 or string2)
3795 as the best previous match. */
3796 boolean same_str_p = (FIRST_STRING_P (match_end)
3797 == MATCHING_IN_FIRST_STRING);
3798 /* 1 if this match is the best seen so far. */
3799 boolean best_match_p;
3801 /* AIX compiler got confused when this was combined
3802 with the previous declaration. */
3804 best_match_p = d > match_end;
3806 best_match_p = !MATCHING_IN_FIRST_STRING;
3808 DEBUG_PRINT1 ("backtracking.\n");
3810 if (!FAIL_STACK_EMPTY ())
3811 { /* More failure points to try. */
3813 /* If exceeds best match so far, save it. */
3814 if (!best_regs_set || best_match_p)
3816 best_regs_set = true;
3819 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
3821 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
3823 best_regstart[mcnt] = regstart[mcnt];
3824 best_regend[mcnt] = regend[mcnt];
3830 /* If no failure points, don't restore garbage. And if
3831 last match is real best match, don't restore second
3833 else if (best_regs_set && !best_match_p)
3836 /* Restore best match. It may happen that `dend ==
3837 end_match_1' while the restored d is in string2.
3838 For example, the pattern `x.*y.*z' against the
3839 strings `x-' and `y-z-', if the two strings are
3840 not consecutive in memory. */
3841 DEBUG_PRINT1 ("Restoring best registers.\n");
3844 dend = ((d >= string1 && d <= end1)
3845 ? end_match_1 : end_match_2);
3847 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
3849 regstart[mcnt] = best_regstart[mcnt];
3850 regend[mcnt] = best_regend[mcnt];
3853 } /* d != end_match_2 */
3856 DEBUG_PRINT1 ("Accepting match.\n");
3858 /* If caller wants register contents data back, do it. */
3859 if (regs && !bufp->no_sub)
3861 /* Have the register data arrays been allocated? */
3862 if (bufp->regs_allocated == REGS_UNALLOCATED)
3863 { /* No. So allocate them with malloc. We need one
3864 extra element beyond `num_regs' for the `-1' marker
3866 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
3867 regs->start = TALLOC (regs->num_regs, regoff_t);
3868 regs->end = TALLOC (regs->num_regs, regoff_t);
3869 if (regs->start == NULL || regs->end == NULL)
3874 bufp->regs_allocated = REGS_REALLOCATE;
3876 else if (bufp->regs_allocated == REGS_REALLOCATE)
3877 { /* Yes. If we need more elements than were already
3878 allocated, reallocate them. If we need fewer, just
3880 if (regs->num_regs < num_regs + 1)
3882 regs->num_regs = num_regs + 1;
3883 RETALLOC (regs->start, regs->num_regs, regoff_t);
3884 RETALLOC (regs->end, regs->num_regs, regoff_t);
3885 if (regs->start == NULL || regs->end == NULL)
3894 /* These braces fend off a "empty body in an else-statement"
3895 warning under GCC when assert expands to nothing. */
3896 assert (bufp->regs_allocated == REGS_FIXED);
3899 /* Convert the pointer data in `regstart' and `regend' to
3900 indices. Register zero has to be set differently,
3901 since we haven't kept track of any info for it. */
3902 if (regs->num_regs > 0)
3904 regs->start[0] = pos;
3905 regs->end[0] = (MATCHING_IN_FIRST_STRING
3906 ? ((regoff_t) (d - string1))
3907 : ((regoff_t) (d - string2 + size1)));
3910 /* Go through the first `min (num_regs, regs->num_regs)'
3911 registers, since that is all we initialized. */
3912 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
3915 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
3916 regs->start[mcnt] = regs->end[mcnt] = -1;
3920 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
3922 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
3926 /* If the regs structure we return has more elements than
3927 were in the pattern, set the extra elements to -1. If
3928 we (re)allocated the registers, this is the case,
3929 because we always allocate enough to have at least one
3931 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
3932 regs->start[mcnt] = regs->end[mcnt] = -1;
3933 } /* regs && !bufp->no_sub */
3935 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
3936 nfailure_points_pushed, nfailure_points_popped,
3937 nfailure_points_pushed - nfailure_points_popped);
3938 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
3940 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
3944 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
3950 /* Otherwise match next pattern command. */
3951 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3953 /* Ignore these. Used to ignore the n of succeed_n's which
3954 currently have n == 0. */
3956 DEBUG_PRINT1 ("EXECUTING no_op.\n");
3960 DEBUG_PRINT1 ("EXECUTING succeed.\n");
3963 /* Match the next n pattern characters exactly. The following
3964 byte in the pattern defines n, and the n bytes after that
3965 are the characters to match. */
3968 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
3970 /* This is written out as an if-else so we don't waste time
3971 testing `translate' inside the loop. */
3977 if ((unsigned char) translate[(unsigned char) *d++]
3978 != (unsigned char) *p++)
3988 if (*d++ != (char) *p++) goto fail;
3992 SET_REGS_MATCHED ();
3996 /* Match any character except possibly a newline or a null. */
3998 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4002 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
4003 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4006 SET_REGS_MATCHED ();
4007 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4015 register unsigned char c;
4016 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4018 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4021 c = TRANSLATE (*d); /* The character to match. */
4023 /* Cast to `unsigned' instead of `unsigned char' in case the
4024 bit list is a full 32 bytes long. */
4025 if (c < (unsigned) (*p * BYTEWIDTH)
4026 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4031 if (!not) goto fail;
4033 SET_REGS_MATCHED ();
4039 /* The beginning of a group is represented by start_memory.
4040 The arguments are the register number in the next byte, and the
4041 number of groups inner to this one in the next. The text
4042 matched within the group is recorded (in the internal
4043 registers data structure) under the register number. */
4045 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4047 /* Find out if this group can match the empty string. */
4048 p1 = p; /* To send to group_match_null_string_p. */
4050 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4051 REG_MATCH_NULL_STRING_P (reg_info[*p])
4052 = group_match_null_string_p (&p1, pend, reg_info);
4054 /* Save the position in the string where we were the last time
4055 we were at this open-group operator in case the group is
4056 operated upon by a repetition operator, e.g., with `(a*)*b'
4057 against `ab'; then we want to ignore where we are now in
4058 the string in case this attempt to match fails. */
4059 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4060 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4062 DEBUG_PRINT2 (" old_regstart: %d\n",
4063 POINTER_TO_OFFSET (old_regstart[*p]));
4066 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4068 IS_ACTIVE (reg_info[*p]) = 1;
4069 MATCHED_SOMETHING (reg_info[*p]) = 0;
4071 /* Clear this whenever we change the register activity status. */
4072 set_regs_matched_done = 0;
4074 /* This is the new highest active register. */
4075 highest_active_reg = *p;
4077 /* If nothing was active before, this is the new lowest active
4079 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4080 lowest_active_reg = *p;
4082 /* Move past the register number and inner group count. */
4084 just_past_start_mem = p;
4089 /* The stop_memory opcode represents the end of a group. Its
4090 arguments are the same as start_memory's: the register
4091 number, and the number of inner groups. */
4093 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4095 /* We need to save the string position the last time we were at
4096 this close-group operator in case the group is operated
4097 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4098 against `aba'; then we want to ignore where we are now in
4099 the string in case this attempt to match fails. */
4100 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4101 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4103 DEBUG_PRINT2 (" old_regend: %d\n",
4104 POINTER_TO_OFFSET (old_regend[*p]));
4107 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4109 /* This register isn't active anymore. */
4110 IS_ACTIVE (reg_info[*p]) = 0;
4112 /* Clear this whenever we change the register activity status. */
4113 set_regs_matched_done = 0;
4115 /* If this was the only register active, nothing is active
4117 if (lowest_active_reg == highest_active_reg)
4119 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4120 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4123 { /* We must scan for the new highest active register, since
4124 it isn't necessarily one less than now: consider
4125 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4126 new highest active register is 1. */
4127 unsigned char r = *p - 1;
4128 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4131 /* If we end up at register zero, that means that we saved
4132 the registers as the result of an `on_failure_jump', not
4133 a `start_memory', and we jumped to past the innermost
4134 `stop_memory'. For example, in ((.)*) we save
4135 registers 1 and 2 as a result of the *, but when we pop
4136 back to the second ), we are at the stop_memory 1.
4137 Thus, nothing is active. */
4140 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4141 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4144 highest_active_reg = r;
4147 /* If just failed to match something this time around with a
4148 group that's operated on by a repetition operator, try to
4149 force exit from the ``loop'', and restore the register
4150 information for this group that we had before trying this
4152 if ((!MATCHED_SOMETHING (reg_info[*p])
4153 || just_past_start_mem == p - 1)
4156 boolean is_a_jump_n = false;
4160 switch ((re_opcode_t) *p1++)
4164 case pop_failure_jump:
4165 case maybe_pop_jump:
4167 case dummy_failure_jump:
4168 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4178 /* If the next operation is a jump backwards in the pattern
4179 to an on_failure_jump right before the start_memory
4180 corresponding to this stop_memory, exit from the loop
4181 by forcing a failure after pushing on the stack the
4182 on_failure_jump's jump in the pattern, and d. */
4183 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4184 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4186 /* If this group ever matched anything, then restore
4187 what its registers were before trying this last
4188 failed match, e.g., with `(a*)*b' against `ab' for
4189 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4190 against `aba' for regend[3].
4192 Also restore the registers for inner groups for,
4193 e.g., `((a*)(b*))*' against `aba' (register 3 would
4194 otherwise get trashed). */
4196 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4200 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4202 /* Restore this and inner groups' (if any) registers. */
4203 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
4206 regstart[r] = old_regstart[r];
4208 /* xx why this test? */
4209 if (old_regend[r] >= regstart[r])
4210 regend[r] = old_regend[r];
4214 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4215 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4221 /* Move past the register number and the inner group count. */
4226 /* \<digit> has been turned into a `duplicate' command which is
4227 followed by the numeric value of <digit> as the register number. */
4230 register const char *d2, *dend2;
4231 int regno = *p++; /* Get which register to match against. */
4232 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4234 /* Can't back reference a group which we've never matched. */
4235 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4238 /* Where in input to try to start matching. */
4239 d2 = regstart[regno];
4241 /* Where to stop matching; if both the place to start and
4242 the place to stop matching are in the same string, then
4243 set to the place to stop, otherwise, for now have to use
4244 the end of the first string. */
4246 dend2 = ((FIRST_STRING_P (regstart[regno])
4247 == FIRST_STRING_P (regend[regno]))
4248 ? regend[regno] : end_match_1);
4251 /* If necessary, advance to next segment in register
4255 if (dend2 == end_match_2) break;
4256 if (dend2 == regend[regno]) break;
4258 /* End of string1 => advance to string2. */
4260 dend2 = regend[regno];
4262 /* At end of register contents => success */
4263 if (d2 == dend2) break;
4265 /* If necessary, advance to next segment in data. */
4268 /* How many characters left in this segment to match. */
4271 /* Want how many consecutive characters we can match in
4272 one shot, so, if necessary, adjust the count. */
4273 if (mcnt > dend2 - d2)
4276 /* Compare that many; failure if mismatch, else move
4279 ? bcmp_translate (d, d2, mcnt, translate)
4280 : bcmp (d, d2, mcnt))
4282 d += mcnt, d2 += mcnt;
4284 /* Do this because we've match some characters. */
4285 SET_REGS_MATCHED ();
4291 /* begline matches the empty string at the beginning of the string
4292 (unless `not_bol' is set in `bufp'), and, if
4293 `newline_anchor' is set, after newlines. */
4295 DEBUG_PRINT1 ("EXECUTING begline.\n");
4297 if (AT_STRINGS_BEG (d))
4299 if (!bufp->not_bol) break;
4301 else if (d[-1] == '\n' && bufp->newline_anchor)
4305 /* In all other cases, we fail. */
4309 /* endline is the dual of begline. */
4311 DEBUG_PRINT1 ("EXECUTING endline.\n");
4313 if (AT_STRINGS_END (d))
4315 if (!bufp->not_eol) break;
4318 /* We have to ``prefetch'' the next character. */
4319 else if ((d == end1 ? *string2 : *d) == '\n'
4320 && bufp->newline_anchor)
4327 /* Match at the very beginning of the data. */
4329 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4330 if (AT_STRINGS_BEG (d))
4335 /* Match at the very end of the data. */
4337 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4338 if (AT_STRINGS_END (d))
4343 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4344 pushes NULL as the value for the string on the stack. Then
4345 `pop_failure_point' will keep the current value for the
4346 string, instead of restoring it. To see why, consider
4347 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4348 then the . fails against the \n. But the next thing we want
4349 to do is match the \n against the \n; if we restored the
4350 string value, we would be back at the foo.
4352 Because this is used only in specific cases, we don't need to
4353 check all the things that `on_failure_jump' does, to make
4354 sure the right things get saved on the stack. Hence we don't
4355 share its code. The only reason to push anything on the
4356 stack at all is that otherwise we would have to change
4357 `anychar's code to do something besides goto fail in this
4358 case; that seems worse than this. */
4359 case on_failure_keep_string_jump:
4360 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4362 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4363 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4365 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4369 /* Uses of on_failure_jump:
4371 Each alternative starts with an on_failure_jump that points
4372 to the beginning of the next alternative. Each alternative
4373 except the last ends with a jump that in effect jumps past
4374 the rest of the alternatives. (They really jump to the
4375 ending jump of the following alternative, because tensioning
4376 these jumps is a hassle.)
4378 Repeats start with an on_failure_jump that points past both
4379 the repetition text and either the following jump or
4380 pop_failure_jump back to this on_failure_jump. */
4381 case on_failure_jump:
4383 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4385 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4386 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4388 /* If this on_failure_jump comes right before a group (i.e.,
4389 the original * applied to a group), save the information
4390 for that group and all inner ones, so that if we fail back
4391 to this point, the group's information will be correct.
4392 For example, in \(a*\)*\1, we need the preceding group,
4393 and in \(zz\(a*\)b*\)\2, we need the inner group. */
4395 /* We can't use `p' to check ahead because we push
4396 a failure point to `p + mcnt' after we do this. */
4399 /* We need to skip no_op's before we look for the
4400 start_memory in case this on_failure_jump is happening as
4401 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4403 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4406 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4408 /* We have a new highest active register now. This will
4409 get reset at the start_memory we are about to get to,
4410 but we will have saved all the registers relevant to
4411 this repetition op, as described above. */
4412 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4413 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4414 lowest_active_reg = *(p1 + 1);
4417 DEBUG_PRINT1 (":\n");
4418 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4422 /* A smart repeat ends with `maybe_pop_jump'.
4423 We change it to either `pop_failure_jump' or `jump'. */
4424 case maybe_pop_jump:
4425 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4426 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4428 register unsigned char *p2 = p;
4430 /* Compare the beginning of the repeat with what in the
4431 pattern follows its end. If we can establish that there
4432 is nothing that they would both match, i.e., that we
4433 would have to backtrack because of (as in, e.g., `a*a')
4434 then we can change to pop_failure_jump, because we'll
4435 never have to backtrack.
4437 This is not true in the case of alternatives: in
4438 `(a|ab)*' we do need to backtrack to the `ab' alternative
4439 (e.g., if the string was `ab'). But instead of trying to
4440 detect that here, the alternative has put on a dummy
4441 failure point which is what we will end up popping. */
4443 /* Skip over open/close-group commands.
4444 If what follows this loop is a ...+ construct,
4445 look at what begins its body, since we will have to
4446 match at least one of that. */
4450 && ((re_opcode_t) *p2 == stop_memory
4451 || (re_opcode_t) *p2 == start_memory))
4453 else if (p2 + 6 < pend
4454 && (re_opcode_t) *p2 == dummy_failure_jump)
4461 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4462 to the `maybe_finalize_jump' of this case. Examine what
4465 /* If we're at the end of the pattern, we can change. */
4468 /* Consider what happens when matching ":\(.*\)"
4469 against ":/". I don't really understand this code
4471 p[-3] = (unsigned char) pop_failure_jump;
4473 (" End of pattern: change to `pop_failure_jump'.\n");
4476 else if ((re_opcode_t) *p2 == exactn
4477 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4479 register unsigned char c
4480 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4482 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4484 p[-3] = (unsigned char) pop_failure_jump;
4485 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4489 else if ((re_opcode_t) p1[3] == charset
4490 || (re_opcode_t) p1[3] == charset_not)
4492 int not = (re_opcode_t) p1[3] == charset_not;
4494 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4495 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4498 /* `not' is equal to 1 if c would match, which means
4499 that we can't change to pop_failure_jump. */
4502 p[-3] = (unsigned char) pop_failure_jump;
4503 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4507 else if ((re_opcode_t) *p2 == charset)
4510 register unsigned char c
4511 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4514 if ((re_opcode_t) p1[3] == exactn
4515 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
4516 && (p2[2 + p1[5] / BYTEWIDTH]
4517 & (1 << (p1[5] % BYTEWIDTH)))))
4519 p[-3] = (unsigned char) pop_failure_jump;
4520 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4524 else if ((re_opcode_t) p1[3] == charset_not)
4527 /* We win if the charset_not inside the loop
4528 lists every character listed in the charset after. */
4529 for (idx = 0; idx < (int) p2[1]; idx++)
4530 if (! (p2[2 + idx] == 0
4531 || (idx < (int) p1[4]
4532 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4537 p[-3] = (unsigned char) pop_failure_jump;
4538 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4541 else if ((re_opcode_t) p1[3] == charset)
4544 /* We win if the charset inside the loop
4545 has no overlap with the one after the loop. */
4547 idx < (int) p2[1] && idx < (int) p1[4];
4549 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4552 if (idx == p2[1] || idx == p1[4])
4554 p[-3] = (unsigned char) pop_failure_jump;
4555 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4560 p -= 2; /* Point at relative address again. */
4561 if ((re_opcode_t) p[-1] != pop_failure_jump)
4563 p[-1] = (unsigned char) jump;
4564 DEBUG_PRINT1 (" Match => jump.\n");
4565 goto unconditional_jump;
4567 /* Note fall through. */
4570 /* The end of a simple repeat has a pop_failure_jump back to
4571 its matching on_failure_jump, where the latter will push a
4572 failure point. The pop_failure_jump takes off failure
4573 points put on by this pop_failure_jump's matching
4574 on_failure_jump; we got through the pattern to here from the
4575 matching on_failure_jump, so didn't fail. */
4576 case pop_failure_jump:
4578 /* We need to pass separate storage for the lowest and
4579 highest registers, even though we don't care about the
4580 actual values. Otherwise, we will restore only one
4581 register from the stack, since lowest will == highest in
4582 `pop_failure_point'. */
4583 unsigned dummy_low_reg, dummy_high_reg;
4584 unsigned char *pdummy;
4587 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4588 POP_FAILURE_POINT (sdummy, pdummy,
4589 dummy_low_reg, dummy_high_reg,
4590 reg_dummy, reg_dummy, reg_info_dummy);
4592 /* Note fall through. */
4595 /* Unconditionally jump (without popping any failure points). */
4598 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4599 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4600 p += mcnt; /* Do the jump. */
4601 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4605 /* We need this opcode so we can detect where alternatives end
4606 in `group_match_null_string_p' et al. */
4608 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4609 goto unconditional_jump;
4612 /* Normally, the on_failure_jump pushes a failure point, which
4613 then gets popped at pop_failure_jump. We will end up at
4614 pop_failure_jump, also, and with a pattern of, say, `a+', we
4615 are skipping over the on_failure_jump, so we have to push
4616 something meaningless for pop_failure_jump to pop. */
4617 case dummy_failure_jump:
4618 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4619 /* It doesn't matter what we push for the string here. What
4620 the code at `fail' tests is the value for the pattern. */
4621 PUSH_FAILURE_POINT (0, 0, -2);
4622 goto unconditional_jump;
4625 /* At the end of an alternative, we need to push a dummy failure
4626 point in case we are followed by a `pop_failure_jump', because
4627 we don't want the failure point for the alternative to be
4628 popped. For example, matching `(a|ab)*' against `aab'
4629 requires that we match the `ab' alternative. */
4630 case push_dummy_failure:
4631 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4632 /* See comments just above at `dummy_failure_jump' about the
4634 PUSH_FAILURE_POINT (0, 0, -2);
4637 /* Have to succeed matching what follows at least n times.
4638 After that, handle like `on_failure_jump'. */
4640 EXTRACT_NUMBER (mcnt, p + 2);
4641 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4644 /* Originally, this is how many times we HAVE to succeed. */
4649 STORE_NUMBER_AND_INCR (p, mcnt);
4650 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt);
4654 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4655 p[2] = (unsigned char) no_op;
4656 p[3] = (unsigned char) no_op;
4662 EXTRACT_NUMBER (mcnt, p + 2);
4663 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4665 /* Originally, this is how many times we CAN jump. */
4669 STORE_NUMBER (p + 2, mcnt);
4670 goto unconditional_jump;
4672 /* If don't have to jump any more, skip over the rest of command. */
4679 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4681 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4683 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4684 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4685 STORE_NUMBER (p1, mcnt);
4690 /* The DEC Alpha C compiler 3.x generates incorrect code for the
4691 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4692 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4693 macro and introducing temporary variables works around the bug. */
4696 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4697 if (AT_WORD_BOUNDARY (d))
4702 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4703 if (AT_WORD_BOUNDARY (d))
4709 boolean prevchar, thischar;
4711 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4712 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4715 prevchar = WORDCHAR_P (d - 1);
4716 thischar = WORDCHAR_P (d);
4717 if (prevchar != thischar)
4724 boolean prevchar, thischar;
4726 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4727 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4730 prevchar = WORDCHAR_P (d - 1);
4731 thischar = WORDCHAR_P (d);
4732 if (prevchar != thischar)
4739 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
4740 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
4745 DEBUG_PRINT1 ("EXECUTING wordend.\n");
4746 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
4747 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
4753 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
4754 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
4759 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
4760 if (PTR_CHAR_POS ((unsigned char *) d) != point)
4765 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
4766 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
4771 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
4776 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
4780 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
4782 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
4784 SET_REGS_MATCHED ();
4788 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
4790 goto matchnotsyntax;
4793 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
4797 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
4799 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
4801 SET_REGS_MATCHED ();
4804 #else /* not emacs */
4806 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
4808 if (!WORDCHAR_P (d))
4810 SET_REGS_MATCHED ();
4815 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
4819 SET_REGS_MATCHED ();
4822 #endif /* not emacs */
4827 continue; /* Successfully executed one pattern command; keep going. */
4830 /* We goto here if a matching operation fails. */
4832 if (!FAIL_STACK_EMPTY ())
4833 { /* A restart point is known. Restore to that state. */
4834 DEBUG_PRINT1 ("\nFAIL:\n");
4835 POP_FAILURE_POINT (d, p,
4836 lowest_active_reg, highest_active_reg,
4837 regstart, regend, reg_info);
4839 /* If this failure point is a dummy, try the next one. */
4843 /* If we failed to the end of the pattern, don't examine *p. */
4847 boolean is_a_jump_n = false;
4849 /* If failed to a backwards jump that's part of a repetition
4850 loop, need to pop this failure point and use the next one. */
4851 switch ((re_opcode_t) *p)
4855 case maybe_pop_jump:
4856 case pop_failure_jump:
4859 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4862 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
4864 && (re_opcode_t) *p1 == on_failure_jump))
4872 if (d >= string1 && d <= end1)
4876 break; /* Matching at this starting point really fails. */
4880 goto restore_best_regs;
4884 return -1; /* Failure to match. */
4887 /* Subroutine definitions for re_match_2. */
4890 /* We are passed P pointing to a register number after a start_memory.
4892 Return true if the pattern up to the corresponding stop_memory can
4893 match the empty string, and false otherwise.
4895 If we find the matching stop_memory, sets P to point to one past its number.
4896 Otherwise, sets P to an undefined byte less than or equal to END.
4898 We don't handle duplicates properly (yet). */
4901 group_match_null_string_p (p, end, reg_info)
4902 unsigned char **p, *end;
4903 register_info_type *reg_info;
4906 /* Point to after the args to the start_memory. */
4907 unsigned char *p1 = *p + 2;
4911 /* Skip over opcodes that can match nothing, and return true or
4912 false, as appropriate, when we get to one that can't, or to the
4913 matching stop_memory. */
4915 switch ((re_opcode_t) *p1)
4917 /* Could be either a loop or a series of alternatives. */
4918 case on_failure_jump:
4920 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4922 /* If the next operation is not a jump backwards in the
4927 /* Go through the on_failure_jumps of the alternatives,
4928 seeing if any of the alternatives cannot match nothing.
4929 The last alternative starts with only a jump,
4930 whereas the rest start with on_failure_jump and end
4931 with a jump, e.g., here is the pattern for `a|b|c':
4933 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
4934 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
4937 So, we have to first go through the first (n-1)
4938 alternatives and then deal with the last one separately. */
4941 /* Deal with the first (n-1) alternatives, which start
4942 with an on_failure_jump (see above) that jumps to right
4943 past a jump_past_alt. */
4945 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
4947 /* `mcnt' holds how many bytes long the alternative
4948 is, including the ending `jump_past_alt' and
4951 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
4955 /* Move to right after this alternative, including the
4959 /* Break if it's the beginning of an n-th alternative
4960 that doesn't begin with an on_failure_jump. */
4961 if ((re_opcode_t) *p1 != on_failure_jump)
4964 /* Still have to check that it's not an n-th
4965 alternative that starts with an on_failure_jump. */
4967 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4968 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
4970 /* Get to the beginning of the n-th alternative. */
4976 /* Deal with the last alternative: go back and get number
4977 of the `jump_past_alt' just before it. `mcnt' contains
4978 the length of the alternative. */
4979 EXTRACT_NUMBER (mcnt, p1 - 2);
4981 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
4984 p1 += mcnt; /* Get past the n-th alternative. */
4990 assert (p1[1] == **p);
4996 if (!common_op_match_null_string_p (&p1, end, reg_info))
4999 } /* while p1 < end */
5002 } /* group_match_null_string_p */
5005 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5006 It expects P to be the first byte of a single alternative and END one
5007 byte past the last. The alternative can contain groups. */
5010 alt_match_null_string_p (p, end, reg_info)
5011 unsigned char *p, *end;
5012 register_info_type *reg_info;
5015 unsigned char *p1 = p;
5019 /* Skip over opcodes that can match nothing, and break when we get
5020 to one that can't. */
5022 switch ((re_opcode_t) *p1)
5025 case on_failure_jump:
5027 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5032 if (!common_op_match_null_string_p (&p1, end, reg_info))
5035 } /* while p1 < end */
5038 } /* alt_match_null_string_p */
5041 /* Deals with the ops common to group_match_null_string_p and
5042 alt_match_null_string_p.
5044 Sets P to one after the op and its arguments, if any. */
5047 common_op_match_null_string_p (p, end, reg_info)
5048 unsigned char **p, *end;
5049 register_info_type *reg_info;
5054 unsigned char *p1 = *p;
5056 switch ((re_opcode_t) *p1++)
5076 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5077 ret = group_match_null_string_p (&p1, end, reg_info);
5079 /* Have to set this here in case we're checking a group which
5080 contains a group and a back reference to it. */
5082 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5083 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5089 /* If this is an optimized succeed_n for zero times, make the jump. */
5091 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5099 /* Get to the number of times to succeed. */
5101 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5106 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5114 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5122 /* All other opcodes mean we cannot match the empty string. */
5128 } /* common_op_match_null_string_p */
5131 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5132 bytes; nonzero otherwise. */
5135 bcmp_translate (s1, s2, len, translate)
5136 unsigned char *s1, *s2;
5138 RE_TRANSLATE_TYPE translate;
5140 register unsigned char *p1 = s1, *p2 = s2;
5143 if (translate[*p1++] != translate[*p2++]) return 1;
5149 /* Entry points for GNU code. */
5151 /* re_compile_pattern is the GNU regular expression compiler: it
5152 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5153 Returns 0 if the pattern was valid, otherwise an error string.
5155 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5156 are set in BUFP on entry.
5158 We call regex_compile to do the actual compilation. */
5161 re_compile_pattern (pattern, length, bufp)
5162 const char *pattern;
5164 struct re_pattern_buffer *bufp;
5168 /* GNU code is written to assume at least RE_NREGS registers will be set
5169 (and at least one extra will be -1). */
5170 bufp->regs_allocated = REGS_UNALLOCATED;
5172 /* And GNU code determines whether or not to get register information
5173 by passing null for the REGS argument to re_match, etc., not by
5177 /* Match anchors at newline. */
5178 bufp->newline_anchor = 1;
5180 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5184 return gettext (re_error_msgid[(int) ret]);
5187 /* Entry points compatible with 4.2 BSD regex library. We don't define
5188 them unless specifically requested. */
5190 #if defined (_REGEX_RE_COMP) || defined (_LIBC)
5192 /* BSD has one and only one pattern buffer. */
5193 static struct re_pattern_buffer re_comp_buf;
5195 char * weak_function
5203 if (!re_comp_buf.buffer)
5204 return gettext ("No previous regular expression");
5208 if (!re_comp_buf.buffer)
5210 re_comp_buf.buffer = (unsigned char *) malloc (200);
5211 if (re_comp_buf.buffer == NULL)
5212 return gettext (re_error_msgid[(int) REG_ESPACE]);
5213 re_comp_buf.allocated = 200;
5215 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5216 if (re_comp_buf.fastmap == NULL)
5217 return gettext (re_error_msgid[(int) REG_ESPACE]);
5220 /* Since `re_exec' always passes NULL for the `regs' argument, we
5221 don't need to initialize the pattern buffer fields which affect it. */
5223 /* Match anchors at newlines. */
5224 re_comp_buf.newline_anchor = 1;
5226 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5231 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5232 return (char *) gettext (re_error_msgid[(int) ret]);
5240 const int len = strlen (s);
5242 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5245 #endif /* _REGEX_RE_COMP */
5247 /* POSIX.2 functions. Don't define these for Emacs. */
5251 /* regcomp takes a regular expression as a string and compiles it.
5253 PREG is a regex_t *. We do not expect any fields to be initialized,
5254 since POSIX says we shouldn't. Thus, we set
5256 `buffer' to the compiled pattern;
5257 `used' to the length of the compiled pattern;
5258 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5259 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5260 RE_SYNTAX_POSIX_BASIC;
5261 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5262 `fastmap' and `fastmap_accurate' to zero;
5263 `re_nsub' to the number of subexpressions in PATTERN.
5265 PATTERN is the address of the pattern string.
5267 CFLAGS is a series of bits which affect compilation.
5269 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5270 use POSIX basic syntax.
5272 If REG_NEWLINE is set, then . and [^...] don't match newline.
5273 Also, regexec will try a match beginning after every newline.
5275 If REG_ICASE is set, then we considers upper- and lowercase
5276 versions of letters to be equivalent when matching.
5278 If REG_NOSUB is set, then when PREG is passed to regexec, that
5279 routine will report only success or failure, and nothing about the
5282 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
5283 the return codes and their meanings.) */
5286 regcomp (preg, pattern, cflags)
5288 const char *pattern;
5293 = (cflags & REG_EXTENDED) ?
5294 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5296 /* regex_compile will allocate the space for the compiled pattern. */
5298 preg->allocated = 0;
5301 /* Don't bother to use a fastmap when searching. This simplifies the
5302 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
5303 characters after newlines into the fastmap. This way, we just try
5307 if (cflags & REG_ICASE)
5312 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
5313 * sizeof (*(RE_TRANSLATE_TYPE)0));
5314 if (preg->translate == NULL)
5315 return (int) REG_ESPACE;
5317 /* Map uppercase characters to corresponding lowercase ones. */
5318 for (i = 0; i < CHAR_SET_SIZE; i++)
5319 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
5322 preg->translate = NULL;
5324 /* If REG_NEWLINE is set, newlines are treated differently. */
5325 if (cflags & REG_NEWLINE)
5326 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5327 syntax &= ~RE_DOT_NEWLINE;
5328 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5329 /* It also changes the matching behavior. */
5330 preg->newline_anchor = 1;
5333 preg->newline_anchor = 0;
5335 preg->no_sub = !!(cflags & REG_NOSUB);
5337 /* POSIX says a null character in the pattern terminates it, so we
5338 can use strlen here in compiling the pattern. */
5339 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5341 /* POSIX doesn't distinguish between an unmatched open-group and an
5342 unmatched close-group: both are REG_EPAREN. */
5343 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5349 /* regexec searches for a given pattern, specified by PREG, in the
5352 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5353 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5354 least NMATCH elements, and we set them to the offsets of the
5355 corresponding matched substrings.
5357 EFLAGS specifies `execution flags' which affect matching: if
5358 REG_NOTBOL is set, then ^ does not match at the beginning of the
5359 string; if REG_NOTEOL is set, then $ does not match at the end.
5361 We return 0 if we find a match and REG_NOMATCH if not. */
5364 regexec (preg, string, nmatch, pmatch, eflags)
5365 const regex_t *preg;
5368 regmatch_t pmatch[];
5372 struct re_registers regs;
5373 regex_t private_preg;
5374 int len = strlen (string);
5375 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5377 private_preg = *preg;
5379 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5380 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5382 /* The user has told us exactly how many registers to return
5383 information about, via `nmatch'. We have to pass that on to the
5384 matching routines. */
5385 private_preg.regs_allocated = REGS_FIXED;
5389 regs.num_regs = nmatch;
5390 regs.start = TALLOC (nmatch, regoff_t);
5391 regs.end = TALLOC (nmatch, regoff_t);
5392 if (regs.start == NULL || regs.end == NULL)
5393 return (int) REG_NOMATCH;
5396 /* Perform the searching operation. */
5397 ret = re_search (&private_preg, string, len,
5398 /* start: */ 0, /* range: */ len,
5399 want_reg_info ? ®s : (struct re_registers *) 0);
5401 /* Copy the register information to the POSIX structure. */
5408 for (r = 0; r < nmatch; r++)
5410 pmatch[r].rm_so = regs.start[r];
5411 pmatch[r].rm_eo = regs.end[r];
5415 /* If we needed the temporary register info, free the space now. */
5420 /* We want zero return to mean success, unlike `re_search'. */
5421 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5425 /* Returns a message corresponding to an error code, ERRCODE, returned
5426 from either regcomp or regexec. We don't use PREG here. */
5429 regerror (errcode, preg, errbuf, errbuf_size)
5431 const regex_t *preg;
5439 || errcode >= (int) (sizeof (re_error_msgid)
5440 / sizeof (re_error_msgid[0])))
5441 /* Only error codes returned by the rest of the code should be passed
5442 to this routine. If we are given anything else, or if other regex
5443 code generates an invalid error code, then the program has a bug.
5444 Dump core so we can fix it. */
5447 msg = gettext (re_error_msgid[errcode]);
5449 msg_size = strlen (msg) + 1; /* Includes the null. */
5451 if (errbuf_size != 0)
5453 if (msg_size > errbuf_size)
5455 strncpy (errbuf, msg, errbuf_size - 1);
5456 errbuf[errbuf_size - 1] = 0;
5459 strcpy (errbuf, msg);
5466 /* Free dynamically allocated space used by PREG. */
5472 if (preg->buffer != NULL)
5473 free (preg->buffer);
5474 preg->buffer = NULL;
5476 preg->allocated = 0;
5479 if (preg->fastmap != NULL)
5480 free (preg->fastmap);
5481 preg->fastmap = NULL;
5482 preg->fastmap_accurate = 0;
5484 if (preg->translate != NULL)
5485 free (preg->translate);
5486 preg->translate = NULL;
5489 #endif /* not emacs */
projects / platform / upstream / glibc.git / blob