1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
6 Copyright (C) 1993, 1994, 1995, 1996, 1997 Free Software Foundation, Inc.
8 This program is free software; you can redistribute it and/or modify
9 it under the terms of the GNU General Public License as published by
10 the Free Software Foundation; either version 2, or (at your option)
13 This program is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with this program; if not, write to the Free Software Foundation,
20 Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
22 /* AIX requires this to be the first thing in the file. */
23 #if defined (_AIX) && !defined (REGEX_MALLOC)
34 #if defined(STDC_HEADERS) && !defined(emacs)
37 /* We need this for `regex.h', and perhaps for the Emacs include files. */
38 #include <sys/types.h>
41 /* For platform which support the ISO C amendement 1 functionality we
42 support user defined character classes. */
43 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
48 /* This is for other GNU distributions with internationalized messages. */
49 #if HAVE_LIBINTL_H || defined (_LIBC)
52 # define gettext(msgid) (msgid)
56 /* This define is so xgettext can find the internationalizable
58 #define gettext_noop(String) String
61 /* The `emacs' switch turns on certain matching commands
62 that make sense only in Emacs. */
71 /* If we are not linking with Emacs proper,
72 we can't use the relocating allocator
73 even if config.h says that we can. */
76 #if defined (STDC_HEADERS) || defined (_LIBC)
83 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
84 If nothing else has been done, use the method below. */
85 #ifdef INHIBIT_STRING_HEADER
86 #if !(defined (HAVE_BZERO) && defined (HAVE_BCOPY))
87 #if !defined (bzero) && !defined (bcopy)
88 #undef INHIBIT_STRING_HEADER
93 /* This is the normal way of making sure we have a bcopy and a bzero.
94 This is used in most programs--a few other programs avoid this
95 by defining INHIBIT_STRING_HEADER. */
96 #ifndef INHIBIT_STRING_HEADER
97 #if defined (HAVE_STRING_H) || defined (STDC_HEADERS) || defined (_LIBC) || defined(_WIN32)
100 #define bcmp(s1, s2, n) memcmp ((s1), (s2), (n))
103 #define bcopy(s, d, n) memcpy ((d), (s), (n))
106 #define bzero(s, n) memset ((s), 0, (n))
113 /* Define the syntax stuff for \<, \>, etc. */
115 /* This must be nonzero for the wordchar and notwordchar pattern
116 commands in re_match_2. */
121 #ifdef SWITCH_ENUM_BUG
122 #define SWITCH_ENUM_CAST(x) ((int)(x))
124 #define SWITCH_ENUM_CAST(x) (x)
129 extern char *re_syntax_table;
131 #else /* not SYNTAX_TABLE */
133 /* How many characters in the character set. */
134 #define CHAR_SET_SIZE 256
136 static char re_syntax_table[CHAR_SET_SIZE];
147 bzero (re_syntax_table, sizeof re_syntax_table);
149 for (c = 'a'; c <= 'z'; c++)
150 re_syntax_table[c] = Sword;
152 for (c = 'A'; c <= 'Z'; c++)
153 re_syntax_table[c] = Sword;
155 for (c = '0'; c <= '9'; c++)
156 re_syntax_table[c] = Sword;
158 re_syntax_table['_'] = Sword;
163 #endif /* not SYNTAX_TABLE */
165 #define SYNTAX(c) re_syntax_table[c]
167 #endif /* not emacs */
169 /* Get the interface, including the syntax bits. */
170 #include "ghcRegex.h"
172 /* isalpha etc. are used for the character classes. */
175 /* Jim Meyering writes:
177 "... Some ctype macros are valid only for character codes that
178 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
179 using /bin/cc or gcc but without giving an ansi option). So, all
180 ctype uses should be through macros like ISPRINT... If
181 STDC_HEADERS is defined, then autoconf has verified that the ctype
182 macros don't need to be guarded with references to isascii. ...
183 Defining isascii to 1 should let any compiler worth its salt
184 eliminate the && through constant folding." */
186 #if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII))
189 #define ISASCII(c) isascii(c)
193 #define ISBLANK(c) (ISASCII (c) && isblank (c))
195 #define ISBLANK(c) ((c) == ' ' || (c) == '\t')
198 #define ISGRAPH(c) (ISASCII (c) && isgraph (c))
200 #define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
203 #define ISPRINT(c) (ISASCII (c) && isprint (c))
204 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
205 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
206 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
207 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
208 #define ISLOWER(c) (ISASCII (c) && islower (c))
209 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
210 #define ISSPACE(c) (ISASCII (c) && isspace (c))
211 #define ISUPPER(c) (ISASCII (c) && isupper (c))
212 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
215 #define NULL (void *)0
218 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
219 since ours (we hope) works properly with all combinations of
220 machines, compilers, `char' and `unsigned char' argument types.
221 (Per Bothner suggested the basic approach.) */
222 #undef SIGN_EXTEND_CHAR
224 #define SIGN_EXTEND_CHAR(c) ((signed char) (c))
225 #else /* not __STDC__ */
226 /* As in Harbison and Steele. */
227 #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
230 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
231 use `alloca' instead of `malloc'. This is because using malloc in
232 re_search* or re_match* could cause memory leaks when C-g is used in
233 Emacs; also, malloc is slower and causes storage fragmentation. On
234 the other hand, malloc is more portable, and easier to debug.
236 Because we sometimes use alloca, some routines have to be macros,
237 not functions -- `alloca'-allocated space disappears at the end of the
238 function it is called in. */
242 #define REGEX_ALLOCATE malloc
243 #define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
244 #define REGEX_FREE free
246 #else /* not REGEX_MALLOC */
248 /* Emacs already defines alloca, sometimes. */
251 /* Make alloca work the best possible way. */
253 #define alloca __builtin_alloca
254 #else /* not __GNUC__ */
257 #else /* not __GNUC__ or HAVE_ALLOCA_H */
258 #if 0 /* It is a bad idea to declare alloca. We always cast the result. */
259 #ifndef _AIX /* Already did AIX, up at the top. */
261 #endif /* not _AIX */
263 #endif /* not HAVE_ALLOCA_H */
264 #endif /* not __GNUC__ */
266 #endif /* not alloca */
268 #define REGEX_ALLOCATE alloca
270 /* Assumes a `char *destination' variable. */
271 #define REGEX_REALLOCATE(source, osize, nsize) \
272 (destination = (char *) alloca (nsize), \
273 bcopy (source, destination, osize), \
276 /* No need to do anything to free, after alloca. */
277 #define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
279 #endif /* not REGEX_MALLOC */
281 /* Define how to allocate the failure stack. */
283 #if defined (REL_ALLOC) && defined (REGEX_MALLOC)
285 #define REGEX_ALLOCATE_STACK(size) \
286 r_alloc (&failure_stack_ptr, (size))
287 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
288 r_re_alloc (&failure_stack_ptr, (nsize))
289 #define REGEX_FREE_STACK(ptr) \
290 r_alloc_free (&failure_stack_ptr)
292 #else /* not using relocating allocator */
296 #define REGEX_ALLOCATE_STACK malloc
297 #define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
298 #define REGEX_FREE_STACK free
300 #else /* not REGEX_MALLOC */
302 #define REGEX_ALLOCATE_STACK alloca
304 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
305 REGEX_REALLOCATE (source, osize, nsize)
306 /* No need to explicitly free anything. */
307 #define REGEX_FREE_STACK(arg)
309 #endif /* not REGEX_MALLOC */
310 #endif /* not using relocating allocator */
313 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
314 `string1' or just past its end. This works if PTR is NULL, which is
316 #define FIRST_STRING_P(ptr) \
317 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
319 /* (Re)Allocate N items of type T using malloc, or fail. */
320 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
321 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
322 #define RETALLOC_IF(addr, n, t) \
323 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
324 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
326 #define BYTEWIDTH 8 /* In bits. */
328 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
332 #define MAX(a, b) ((a) > (b) ? (a) : (b))
333 #define MIN(a, b) ((a) < (b) ? (a) : (b))
335 typedef char boolean;
339 static int re_match_2_internal ();
341 /* These are the command codes that appear in compiled regular
342 expressions. Some opcodes are followed by argument bytes. A
343 command code can specify any interpretation whatsoever for its
344 arguments. Zero bytes may appear in the compiled regular expression. */
350 /* Succeed right away--no more backtracking. */
353 /* Followed by one byte giving n, then by n literal bytes. */
356 /* Matches any (more or less) character. */
359 /* Matches any one char belonging to specified set. First
360 following byte is number of bitmap bytes. Then come bytes
361 for a bitmap saying which chars are in. Bits in each byte
362 are ordered low-bit-first. A character is in the set if its
363 bit is 1. A character too large to have a bit in the map is
364 automatically not in the set. */
367 /* Same parameters as charset, but match any character that is
368 not one of those specified. */
371 /* Start remembering the text that is matched, for storing in a
372 register. Followed by one byte with the register number, in
373 the range 0 to one less than the pattern buffer's re_nsub
374 field. Then followed by one byte with the number of groups
375 inner to this one. (This last has to be part of the
376 start_memory only because we need it in the on_failure_jump
380 /* Stop remembering the text that is matched and store it in a
381 memory register. Followed by one byte with the register
382 number, in the range 0 to one less than `re_nsub' in the
383 pattern buffer, and one byte with the number of inner groups,
384 just like `start_memory'. (We need the number of inner
385 groups here because we don't have any easy way of finding the
386 corresponding start_memory when we're at a stop_memory.) */
389 /* Match a duplicate of something remembered. Followed by one
390 byte containing the register number. */
393 /* Fail unless at beginning of line. */
396 /* Fail unless at end of line. */
399 /* Succeeds if at beginning of buffer (if emacs) or at beginning
400 of string to be matched (if not). */
403 /* Analogously, for end of buffer/string. */
406 /* Followed by two byte relative address to which to jump. */
409 /* Same as jump, but marks the end of an alternative. */
412 /* Followed by two-byte relative address of place to resume at
413 in case of failure. */
416 /* Like on_failure_jump, but pushes a placeholder instead of the
417 current string position when executed. */
418 on_failure_keep_string_jump,
420 /* Throw away latest failure point and then jump to following
421 two-byte relative address. */
424 /* Change to pop_failure_jump if know won't have to backtrack to
425 match; otherwise change to jump. This is used to jump
426 back to the beginning of a repeat. If what follows this jump
427 clearly won't match what the repeat does, such that we can be
428 sure that there is no use backtracking out of repetitions
429 already matched, then we change it to a pop_failure_jump.
430 Followed by two-byte address. */
433 /* Jump to following two-byte address, and push a dummy failure
434 point. This failure point will be thrown away if an attempt
435 is made to use it for a failure. A `+' construct makes this
436 before the first repeat. Also used as an intermediary kind
437 of jump when compiling an alternative. */
440 /* Push a dummy failure point and continue. Used at the end of
444 /* Followed by two-byte relative address and two-byte number n.
445 After matching N times, jump to the address upon failure. */
448 /* Followed by two-byte relative address, and two-byte number n.
449 Jump to the address N times, then fail. */
452 /* Set the following two-byte relative address to the
453 subsequent two-byte number. The address *includes* the two
457 wordchar, /* Matches any word-constituent character. */
458 notwordchar, /* Matches any char that is not a word-constituent. */
460 wordbeg, /* Succeeds if at word beginning. */
461 wordend, /* Succeeds if at word end. */
463 wordbound, /* Succeeds if at a word boundary. */
464 notwordbound /* Succeeds if not at a word boundary. */
467 ,before_dot, /* Succeeds if before point. */
468 at_dot, /* Succeeds if at point. */
469 after_dot, /* Succeeds if after point. */
471 /* Matches any character whose syntax is specified. Followed by
472 a byte which contains a syntax code, e.g., Sword. */
475 /* Matches any character whose syntax is not that specified. */
480 /* Common operations on the compiled pattern. */
482 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
484 #define STORE_NUMBER(destination, number) \
486 (destination)[0] = (number) & 0377; \
487 (destination)[1] = (number) >> 8; \
490 /* Same as STORE_NUMBER, except increment DESTINATION to
491 the byte after where the number is stored. Therefore, DESTINATION
492 must be an lvalue. */
494 #define STORE_NUMBER_AND_INCR(destination, number) \
496 STORE_NUMBER (destination, number); \
497 (destination) += 2; \
500 /* Put into DESTINATION a number stored in two contiguous bytes starting
503 #define EXTRACT_NUMBER(destination, source) \
505 (destination) = *(source) & 0377; \
506 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
510 static void extract_number _RE_ARGS ((int *dest, unsigned char *source));
512 extract_number (dest, source)
514 unsigned char *source;
516 int temp = SIGN_EXTEND_CHAR (*(source + 1));
517 *dest = *source & 0377;
521 #ifndef EXTRACT_MACROS /* To debug the macros. */
522 #undef EXTRACT_NUMBER
523 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
524 #endif /* not EXTRACT_MACROS */
528 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
529 SOURCE must be an lvalue. */
531 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
533 EXTRACT_NUMBER (destination, source); \
538 static void extract_number_and_incr _RE_ARGS ((int *destination,
539 unsigned char **source));
541 extract_number_and_incr (destination, source)
543 unsigned char **source;
545 extract_number (destination, *source);
549 #ifndef EXTRACT_MACROS
550 #undef EXTRACT_NUMBER_AND_INCR
551 #define EXTRACT_NUMBER_AND_INCR(dest, src) \
552 extract_number_and_incr (&dest, &src)
553 #endif /* not EXTRACT_MACROS */
557 /* If DEBUG is defined, Regex prints many voluminous messages about what
558 it is doing (if the variable `debug' is nonzero). If linked with the
559 main program in `iregex.c', you can enter patterns and strings
560 interactively. And if linked with the main program in `main.c' and
561 the other test files, you can run the already-written tests. */
565 /* We use standard I/O for debugging. */
568 /* It is useful to test things that ``must'' be true when debugging. */
571 static int debug = 0;
573 #define DEBUG_STATEMENT(e) e
574 #define DEBUG_PRINT1(x) if (debug) printf (x)
575 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
576 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
577 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
578 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
579 if (debug) print_partial_compiled_pattern (s, e)
580 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
581 if (debug) print_double_string (w, s1, sz1, s2, sz2)
584 /* Print the fastmap in human-readable form. */
587 print_fastmap (fastmap)
590 unsigned was_a_range = 0;
593 while (i < (1 << BYTEWIDTH))
599 while (i < (1 << BYTEWIDTH) && fastmap[i])
615 /* Print a compiled pattern string in human-readable form, starting at
616 the START pointer into it and ending just before the pointer END. */
619 print_partial_compiled_pattern (start, end)
620 unsigned char *start;
625 unsigned char *p = start;
626 unsigned char *pend = end;
634 /* Loop over pattern commands. */
637 printf ("%d:\t", p - start);
639 switch ((re_opcode_t) *p++)
647 printf ("/exactn/%d", mcnt);
658 printf ("/start_memory/%d/%d", mcnt, *p++);
663 printf ("/stop_memory/%d/%d", mcnt, *p++);
667 printf ("/duplicate/%d", *p++);
677 register int c, last = -100;
678 register int in_range = 0;
680 printf ("/charset [%s",
681 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
683 assert (p + *p < pend);
685 for (c = 0; c < 256; c++)
687 && (p[1 + (c/8)] & (1 << (c % 8))))
689 /* Are we starting a range? */
690 if (last + 1 == c && ! in_range)
695 /* Have we broken a range? */
696 else if (last + 1 != c && in_range)
725 case on_failure_jump:
726 extract_number_and_incr (&mcnt, &p);
727 printf ("/on_failure_jump to %d", p + mcnt - start);
730 case on_failure_keep_string_jump:
731 extract_number_and_incr (&mcnt, &p);
732 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
735 case dummy_failure_jump:
736 extract_number_and_incr (&mcnt, &p);
737 printf ("/dummy_failure_jump to %d", p + mcnt - start);
740 case push_dummy_failure:
741 printf ("/push_dummy_failure");
745 extract_number_and_incr (&mcnt, &p);
746 printf ("/maybe_pop_jump to %d", p + mcnt - start);
749 case pop_failure_jump:
750 extract_number_and_incr (&mcnt, &p);
751 printf ("/pop_failure_jump to %d", p + mcnt - start);
755 extract_number_and_incr (&mcnt, &p);
756 printf ("/jump_past_alt to %d", p + mcnt - start);
760 extract_number_and_incr (&mcnt, &p);
761 printf ("/jump to %d", p + mcnt - start);
765 extract_number_and_incr (&mcnt, &p);
767 extract_number_and_incr (&mcnt2, &p);
768 printf ("/succeed_n to %d, %d times", p1 - start, mcnt2);
772 extract_number_and_incr (&mcnt, &p);
774 extract_number_and_incr (&mcnt2, &p);
775 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
779 extract_number_and_incr (&mcnt, &p);
781 extract_number_and_incr (&mcnt2, &p);
782 printf ("/set_number_at location %d to %d", p1 - start, mcnt2);
786 printf ("/wordbound");
790 printf ("/notwordbound");
802 printf ("/before_dot");
810 printf ("/after_dot");
814 printf ("/syntaxspec");
816 printf ("/%d", mcnt);
820 printf ("/notsyntaxspec");
822 printf ("/%d", mcnt);
827 printf ("/wordchar");
831 printf ("/notwordchar");
843 printf ("?%d", *(p-1));
849 printf ("%d:\tend of pattern.\n", p - start);
854 print_compiled_pattern (bufp)
855 struct re_pattern_buffer *bufp;
857 unsigned char *buffer = bufp->buffer;
859 print_partial_compiled_pattern (buffer, buffer + bufp->used);
860 printf ("%ld bytes used/%ld bytes allocated.\n",
861 bufp->used, bufp->allocated);
863 if (bufp->fastmap_accurate && bufp->fastmap)
865 printf ("fastmap: ");
866 print_fastmap (bufp->fastmap);
869 printf ("re_nsub: %d\t", bufp->re_nsub);
870 printf ("regs_alloc: %d\t", bufp->regs_allocated);
871 printf ("can_be_null: %d\t", bufp->can_be_null);
872 printf ("newline_anchor: %d\n", bufp->newline_anchor);
873 printf ("no_sub: %d\t", bufp->no_sub);
874 printf ("not_bol: %d\t", bufp->not_bol);
875 printf ("not_eol: %d\t", bufp->not_eol);
876 printf ("syntax: %lx\n", bufp->syntax);
877 /* Perhaps we should print the translate table? */
882 print_double_string (where, string1, size1, string2, size2)
895 if (FIRST_STRING_P (where))
897 for (this_char = where - string1; this_char < size1; this_char++)
898 putchar (string1[this_char]);
903 for (this_char = where - string2; this_char < size2; this_char++)
904 putchar (string2[this_char]);
915 #else /* not DEBUG */
920 #define DEBUG_STATEMENT(e)
921 #define DEBUG_PRINT1(x)
922 #define DEBUG_PRINT2(x1, x2)
923 #define DEBUG_PRINT3(x1, x2, x3)
924 #define DEBUG_PRINT4(x1, x2, x3, x4)
925 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
926 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
928 #endif /* not DEBUG */
930 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
931 also be assigned to arbitrarily: each pattern buffer stores its own
932 syntax, so it can be changed between regex compilations. */
933 /* This has no initializer because initialized variables in Emacs
934 become read-only after dumping. */
935 reg_syntax_t re_syntax_options;
938 /* Specify the precise syntax of regexps for compilation. This provides
939 for compatibility for various utilities which historically have
940 different, incompatible syntaxes.
942 The argument SYNTAX is a bit mask comprised of the various bits
943 defined in regex.h. We return the old syntax. */
946 re_set_syntax (syntax)
949 reg_syntax_t ret = re_syntax_options;
951 re_syntax_options = syntax;
953 if (syntax & RE_DEBUG)
955 else if (debug) /* was on but now is not */
961 /* This table gives an error message for each of the error codes listed
962 in regex.h. Obviously the order here has to be same as there.
963 POSIX doesn't require that we do anything for REG_NOERROR,
964 but why not be nice? */
966 static const char *re_error_msgid[] =
968 gettext_noop ("Success"), /* REG_NOERROR */
969 gettext_noop ("No match"), /* REG_NOMATCH */
970 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
971 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
972 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
973 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
974 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
975 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
976 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
977 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
978 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
979 gettext_noop ("Invalid range end"), /* REG_ERANGE */
980 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
981 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
982 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
983 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
984 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
987 /* Avoiding alloca during matching, to placate r_alloc. */
989 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
990 searching and matching functions should not call alloca. On some
991 systems, alloca is implemented in terms of malloc, and if we're
992 using the relocating allocator routines, then malloc could cause a
993 relocation, which might (if the strings being searched are in the
994 ralloc heap) shift the data out from underneath the regexp
997 Here's another reason to avoid allocation: Emacs
998 processes input from X in a signal handler; processing X input may
999 call malloc; if input arrives while a matching routine is calling
1000 malloc, then we're scrod. But Emacs can't just block input while
1001 calling matching routines; then we don't notice interrupts when
1002 they come in. So, Emacs blocks input around all regexp calls
1003 except the matching calls, which it leaves unprotected, in the
1004 faith that they will not malloc. */
1006 /* Normally, this is fine. */
1007 #define MATCH_MAY_ALLOCATE
1009 /* When using GNU C, we are not REALLY using the C alloca, no matter
1010 what config.h may say. So don't take precautions for it. */
1015 /* The match routines may not allocate if (1) they would do it with malloc
1016 and (2) it's not safe for them to use malloc.
1017 Note that if REL_ALLOC is defined, matching would not use malloc for the
1018 failure stack, but we would still use it for the register vectors;
1019 so REL_ALLOC should not affect this. */
1020 #if (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (emacs)
1021 #undef MATCH_MAY_ALLOCATE
1025 /* Failure stack declarations and macros; both re_compile_fastmap and
1026 re_match_2 use a failure stack. These have to be macros because of
1027 REGEX_ALLOCATE_STACK. */
1030 /* Number of failure points for which to initially allocate space
1031 when matching. If this number is exceeded, we allocate more
1032 space, so it is not a hard limit. */
1033 #ifndef INIT_FAILURE_ALLOC
1034 #define INIT_FAILURE_ALLOC 5
1037 /* Roughly the maximum number of failure points on the stack. Would be
1038 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1039 This is a variable only so users of regex can assign to it; we never
1040 change it ourselves. */
1044 #if defined (MATCH_MAY_ALLOCATE)
1045 /* 4400 was enough to cause a crash on Alpha OSF/1,
1046 whose default stack limit is 2mb. */
1047 long int re_max_failures = 4000;
1049 long int re_max_failures = 2000;
1052 union fail_stack_elt
1054 unsigned char *pointer;
1058 typedef union fail_stack_elt fail_stack_elt_t;
1062 fail_stack_elt_t *stack;
1063 unsigned long int size;
1064 unsigned long int avail; /* Offset of next open position. */
1067 #else /* not INT_IS_16BIT */
1069 #if defined (MATCH_MAY_ALLOCATE)
1070 /* 4400 was enough to cause a crash on Alpha OSF/1,
1071 whose default stack limit is 2mb. */
1072 int re_max_failures = 20000;
1074 int re_max_failures = 2000;
1077 union fail_stack_elt
1079 unsigned char *pointer;
1083 typedef union fail_stack_elt fail_stack_elt_t;
1087 fail_stack_elt_t *stack;
1089 unsigned avail; /* Offset of next open position. */
1092 #endif /* INT_IS_16BIT */
1094 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1095 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1096 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1099 /* Define macros to initialize and free the failure stack.
1100 Do `return -2' if the alloc fails. */
1102 #ifdef MATCH_MAY_ALLOCATE
1103 #define INIT_FAIL_STACK() \
1105 fail_stack.stack = (fail_stack_elt_t *) \
1106 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1108 if (fail_stack.stack == NULL) \
1111 fail_stack.size = INIT_FAILURE_ALLOC; \
1112 fail_stack.avail = 0; \
1115 #define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1117 #define INIT_FAIL_STACK() \
1119 fail_stack.avail = 0; \
1122 #define RESET_FAIL_STACK()
1126 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1128 Return 1 if succeeds, and 0 if either ran out of memory
1129 allocating space for it or it was already too large.
1131 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1133 #define DOUBLE_FAIL_STACK(fail_stack) \
1134 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1136 : ((fail_stack).stack = (fail_stack_elt_t *) \
1137 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1138 (fail_stack).size * sizeof (fail_stack_elt_t), \
1139 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1141 (fail_stack).stack == NULL \
1143 : ((fail_stack).size <<= 1, \
1147 /* Push pointer POINTER on FAIL_STACK.
1148 Return 1 if was able to do so and 0 if ran out of memory allocating
1150 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1151 ((FAIL_STACK_FULL () \
1152 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1154 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1157 /* Push a pointer value onto the failure stack.
1158 Assumes the variable `fail_stack'. Probably should only
1159 be called from within `PUSH_FAILURE_POINT'. */
1160 #define PUSH_FAILURE_POINTER(item) \
1161 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1163 /* This pushes an integer-valued item onto the failure stack.
1164 Assumes the variable `fail_stack'. Probably should only
1165 be called from within `PUSH_FAILURE_POINT'. */
1166 #define PUSH_FAILURE_INT(item) \
1167 fail_stack.stack[fail_stack.avail++].integer = (item)
1169 /* Push a fail_stack_elt_t value onto the failure stack.
1170 Assumes the variable `fail_stack'. Probably should only
1171 be called from within `PUSH_FAILURE_POINT'. */
1172 #define PUSH_FAILURE_ELT(item) \
1173 fail_stack.stack[fail_stack.avail++] = (item)
1175 /* These three POP... operations complement the three PUSH... operations.
1176 All assume that `fail_stack' is nonempty. */
1177 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1178 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1179 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1181 /* Used to omit pushing failure point id's when we're not debugging. */
1183 #define DEBUG_PUSH PUSH_FAILURE_INT
1184 #define DEBUG_POP(item_addr) (item_addr)->integer = POP_FAILURE_INT ()
1186 #define DEBUG_PUSH(item)
1187 #define DEBUG_POP(item_addr)
1191 /* Push the information about the state we will need
1192 if we ever fail back to it.
1194 Requires variables fail_stack, regstart, regend, reg_info, and
1195 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be
1198 Does `return FAILURE_CODE' if runs out of memory. */
1200 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1202 char *destination; \
1203 /* Must be int, so when we don't save any registers, the arithmetic \
1204 of 0 + -1 isn't done as unsigned. */ \
1205 /* Can't be int, since there is not a shred of a guarantee that int \
1206 is wide enough to hold a value of something to which pointer can \
1210 DEBUG_STATEMENT (failure_id++); \
1211 DEBUG_STATEMENT (nfailure_points_pushed++); \
1212 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1213 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1214 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1216 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \
1217 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1219 /* Ensure we have enough space allocated for what we will push. */ \
1220 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1222 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1223 return failure_code; \
1225 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1226 (fail_stack).size); \
1227 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1230 /* Push the info, starting with the registers. */ \
1231 DEBUG_PRINT1 ("\n"); \
1234 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1237 DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \
1238 DEBUG_STATEMENT (num_regs_pushed++); \
1240 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1241 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1243 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1244 PUSH_FAILURE_POINTER (regend[this_reg]); \
1246 DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \
1247 DEBUG_PRINT2 (" match_null=%d", \
1248 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1249 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1250 DEBUG_PRINT2 (" matched_something=%d", \
1251 MATCHED_SOMETHING (reg_info[this_reg])); \
1252 DEBUG_PRINT2 (" ever_matched=%d", \
1253 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1254 DEBUG_PRINT1 ("\n"); \
1255 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1258 DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\
1259 PUSH_FAILURE_INT (lowest_active_reg); \
1261 DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\
1262 PUSH_FAILURE_INT (highest_active_reg); \
1264 DEBUG_PRINT2 (" Pushing pattern 0x%x:\n", pattern_place); \
1265 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1266 PUSH_FAILURE_POINTER (pattern_place); \
1268 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \
1269 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1271 DEBUG_PRINT1 ("'\n"); \
1272 PUSH_FAILURE_POINTER (string_place); \
1274 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1275 DEBUG_PUSH (failure_id); \
1278 /* This is the number of items that are pushed and popped on the stack
1279 for each register. */
1280 #define NUM_REG_ITEMS 3
1282 /* Individual items aside from the registers. */
1284 #define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1286 #define NUM_NONREG_ITEMS 4
1289 /* We push at most this many items on the stack. */
1290 /* We used to use (num_regs - 1), which is the number of registers
1291 this regexp will save; but that was changed to 5
1292 to avoid stack overflow for a regexp with lots of parens. */
1293 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1295 /* We actually push this many items. */
1296 #define NUM_FAILURE_ITEMS \
1298 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1302 /* How many items can still be added to the stack without overflowing it. */
1303 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1306 /* Pops what PUSH_FAIL_STACK pushes.
1308 We restore into the parameters, all of which should be lvalues:
1309 STR -- the saved data position.
1310 PAT -- the saved pattern position.
1311 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1312 REGSTART, REGEND -- arrays of string positions.
1313 REG_INFO -- array of information about each subexpression.
1315 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1316 `pend', `string1', `size1', `string2', and `size2'. */
1318 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1320 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \
1322 const unsigned char *string_temp; \
1324 assert (!FAIL_STACK_EMPTY ()); \
1326 /* Remove failure points and point to how many regs pushed. */ \
1327 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1328 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1329 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1331 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1333 DEBUG_POP (&failure_id); \
1334 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1336 /* If the saved string location is NULL, it came from an \
1337 on_failure_keep_string_jump opcode, and we want to throw away the \
1338 saved NULL, thus retaining our current position in the string. */ \
1339 string_temp = POP_FAILURE_POINTER (); \
1340 if (string_temp != NULL) \
1341 str = (const char *) string_temp; \
1343 DEBUG_PRINT2 (" Popping string 0x%x: `", str); \
1344 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1345 DEBUG_PRINT1 ("'\n"); \
1347 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1348 DEBUG_PRINT2 (" Popping pattern 0x%x:\n", pat); \
1349 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1351 /* Restore register info. */ \
1352 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1353 DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \
1355 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1356 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \
1359 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1361 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \
1363 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1364 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \
1366 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1367 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1369 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1370 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1374 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1376 reg_info[this_reg].word.integer = 0; \
1377 regend[this_reg] = 0; \
1378 regstart[this_reg] = 0; \
1380 highest_active_reg = high_reg; \
1383 set_regs_matched_done = 0; \
1384 DEBUG_STATEMENT (nfailure_points_popped++); \
1385 } /* POP_FAILURE_POINT */
1389 /* Structure for per-register (a.k.a. per-group) information.
1390 Other register information, such as the
1391 starting and ending positions (which are addresses), and the list of
1392 inner groups (which is a bits list) are maintained in separate
1395 We are making a (strictly speaking) nonportable assumption here: that
1396 the compiler will pack our bit fields into something that fits into
1397 the type of `word', i.e., is something that fits into one item on the
1401 /* Declarations and macros for re_match_2. */
1405 fail_stack_elt_t word;
1408 /* This field is one if this group can match the empty string,
1409 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1410 #define MATCH_NULL_UNSET_VALUE 3
1411 unsigned match_null_string_p : 2;
1412 unsigned is_active : 1;
1413 unsigned matched_something : 1;
1414 unsigned ever_matched_something : 1;
1416 } register_info_type;
1418 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1419 #define IS_ACTIVE(R) ((R).bits.is_active)
1420 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1421 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1424 /* Call this when have matched a real character; it sets `matched' flags
1425 for the subexpressions which we are currently inside. Also records
1426 that those subexprs have matched. */
1427 #define SET_REGS_MATCHED() \
1430 if (!set_regs_matched_done) \
1433 set_regs_matched_done = 1; \
1434 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1436 MATCHED_SOMETHING (reg_info[r]) \
1437 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1444 /* Registers are set to a sentinel when they haven't yet matched. */
1445 static char reg_unset_dummy;
1446 #define REG_UNSET_VALUE (®_unset_dummy)
1447 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1449 /* Subroutine declarations and macros for regex_compile. */
1451 static reg_errcode_t regex_compile _RE_ARGS ((const char *pattern, size_t size,
1452 reg_syntax_t syntax,
1453 struct re_pattern_buffer *bufp));
1454 static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg));
1455 static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1456 int arg1, int arg2));
1457 static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1458 int arg, unsigned char *end));
1459 static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1460 int arg1, int arg2, unsigned char *end));
1461 static boolean at_begline_loc_p _RE_ARGS ((const char *pattern, const char *p,
1462 reg_syntax_t syntax));
1463 static boolean at_endline_loc_p _RE_ARGS ((const char *p, const char *pend,
1464 reg_syntax_t syntax));
1465 static reg_errcode_t compile_range _RE_ARGS ((const char **p_ptr,
1468 reg_syntax_t syntax,
1471 /* Fetch the next character in the uncompiled pattern---translating it
1472 if necessary. Also cast from a signed character in the constant
1473 string passed to us by the user to an unsigned char that we can use
1474 as an array index (in, e.g., `translate'). */
1476 #define PATFETCH(c) \
1477 do {if (p == pend) return REG_EEND; \
1478 c = (unsigned char) *p++; \
1479 if (translate) c = (unsigned char) translate[c]; \
1483 /* Fetch the next character in the uncompiled pattern, with no
1485 #define PATFETCH_RAW(c) \
1486 do {if (p == pend) return REG_EEND; \
1487 c = (unsigned char) *p++; \
1490 /* Go backwards one character in the pattern. */
1491 #define PATUNFETCH p--
1494 /* If `translate' is non-null, return translate[D], else just D. We
1495 cast the subscript to translate because some data is declared as
1496 `char *', to avoid warnings when a string constant is passed. But
1497 when we use a character as a subscript we must make it unsigned. */
1499 #define TRANSLATE(d) \
1500 (translate ? (char) translate[(unsigned char) (d)] : (d))
1504 /* Macros for outputting the compiled pattern into `buffer'. */
1506 /* If the buffer isn't allocated when it comes in, use this. */
1507 #define INIT_BUF_SIZE 32
1509 /* Make sure we have at least N more bytes of space in buffer. */
1510 #define GET_BUFFER_SPACE(n) \
1511 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1514 /* Make sure we have one more byte of buffer space and then add C to it. */
1515 #define BUF_PUSH(c) \
1517 GET_BUFFER_SPACE (1); \
1518 *b++ = (unsigned char) (c); \
1522 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1523 #define BUF_PUSH_2(c1, c2) \
1525 GET_BUFFER_SPACE (2); \
1526 *b++ = (unsigned char) (c1); \
1527 *b++ = (unsigned char) (c2); \
1531 /* As with BUF_PUSH_2, except for three bytes. */
1532 #define BUF_PUSH_3(c1, c2, c3) \
1534 GET_BUFFER_SPACE (3); \
1535 *b++ = (unsigned char) (c1); \
1536 *b++ = (unsigned char) (c2); \
1537 *b++ = (unsigned char) (c3); \
1541 /* Store a jump with opcode OP at LOC to location TO. We store a
1542 relative address offset by the three bytes the jump itself occupies. */
1543 #define STORE_JUMP(op, loc, to) \
1544 store_op1 (op, loc, (int) ((to) - (loc) - 3))
1546 /* Likewise, for a two-argument jump. */
1547 #define STORE_JUMP2(op, loc, to, arg) \
1548 store_op2 (op, loc, (int) ((to) - (loc) - 3), arg)
1550 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1551 #define INSERT_JUMP(op, loc, to) \
1552 insert_op1 (op, loc, (int) ((to) - (loc) - 3), b)
1554 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1555 #define INSERT_JUMP2(op, loc, to, arg) \
1556 insert_op2 (op, loc, (int) ((to) - (loc) - 3), arg, b)
1559 /* This is not an arbitrary limit: the arguments which represent offsets
1560 into the pattern are two bytes long. So if 2^16 bytes turns out to
1561 be too small, many things would have to change. */
1562 /* Any other compiler which, like MSC, has allocation limit below 2^16
1563 bytes will have to use approach similar to what was done below for
1564 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1565 reallocating to 0 bytes. Such thing is not going to work too well.
1566 You have been warned!! */
1567 #if defined(_MSC_VER) && !defined(WIN32)
1568 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
1569 The REALLOC define eliminates a flurry of conversion warnings,
1570 but is not required. */
1571 #define MAX_BUF_SIZE 65500L
1572 #define REALLOC(p,s) realloc ((p), (size_t) (s))
1574 #define MAX_BUF_SIZE (1L << 16)
1575 #define REALLOC(p,s) realloc ((p), (s))
1578 /* Extend the buffer by twice its current size via realloc and
1579 reset the pointers that pointed into the old block to point to the
1580 correct places in the new one. If extending the buffer results in it
1581 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1582 #define EXTEND_BUFFER() \
1584 unsigned char *old_buffer = bufp->buffer; \
1585 if (bufp->allocated == MAX_BUF_SIZE) \
1587 bufp->allocated <<= 1; \
1588 if (bufp->allocated > MAX_BUF_SIZE) \
1589 bufp->allocated = MAX_BUF_SIZE; \
1590 bufp->buffer = (unsigned char *) REALLOC (bufp->buffer, bufp->allocated);\
1591 if (bufp->buffer == NULL) \
1592 return REG_ESPACE; \
1593 /* If the buffer moved, move all the pointers into it. */ \
1594 if (old_buffer != bufp->buffer) \
1596 b = (b - old_buffer) + bufp->buffer; \
1597 begalt = (begalt - old_buffer) + bufp->buffer; \
1598 if (fixup_alt_jump) \
1599 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1601 laststart = (laststart - old_buffer) + bufp->buffer; \
1602 if (pending_exact) \
1603 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1608 /* Since we have one byte reserved for the register number argument to
1609 {start,stop}_memory, the maximum number of groups we can report
1610 things about is what fits in that byte. */
1611 #define MAX_REGNUM 255
1613 /* But patterns can have more than `MAX_REGNUM' registers. We just
1614 ignore the excess. */
1615 typedef unsigned regnum_t;
1618 /* Macros for the compile stack. */
1620 /* Since offsets can go either forwards or backwards, this type needs to
1621 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1622 /* int may be not enough when sizeof(int) == 2. */
1623 typedef long pattern_offset_t;
1627 pattern_offset_t begalt_offset;
1628 pattern_offset_t fixup_alt_jump;
1629 pattern_offset_t inner_group_offset;
1630 pattern_offset_t laststart_offset;
1632 } compile_stack_elt_t;
1637 compile_stack_elt_t *stack;
1639 unsigned avail; /* Offset of next open position. */
1640 } compile_stack_type;
1643 #define INIT_COMPILE_STACK_SIZE 32
1645 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1646 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1648 /* The next available element. */
1649 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1652 /* Set the bit for character C in a list. */
1653 #define SET_LIST_BIT(c) \
1654 (b[((unsigned char) (c)) / BYTEWIDTH] \
1655 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1658 /* Get the next unsigned number in the uncompiled pattern. */
1659 #define GET_UNSIGNED_NUMBER(num) \
1663 while (ISDIGIT (c)) \
1667 num = num * 10 + c - '0'; \
1675 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
1676 /* The GNU C library provides support for user-defined character classes
1677 and the functions from ISO C amendement 1. */
1678 # ifdef CHARCLASS_NAME_MAX
1679 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
1681 /* This shouldn't happen but some implementation might still have this
1682 problem. Use a reasonable default value. */
1683 # define CHAR_CLASS_MAX_LENGTH 256
1686 # define IS_CHAR_CLASS(string) wctype (string)
1688 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1690 # define IS_CHAR_CLASS(string) \
1691 (STREQ (string, "alpha") || STREQ (string, "upper") \
1692 || STREQ (string, "lower") || STREQ (string, "digit") \
1693 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1694 || STREQ (string, "space") || STREQ (string, "print") \
1695 || STREQ (string, "punct") || STREQ (string, "graph") \
1696 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1699 #ifndef MATCH_MAY_ALLOCATE
1701 /* If we cannot allocate large objects within re_match_2_internal,
1702 we make the fail stack and register vectors global.
1703 The fail stack, we grow to the maximum size when a regexp
1705 The register vectors, we adjust in size each time we
1706 compile a regexp, according to the number of registers it needs. */
1708 static fail_stack_type fail_stack;
1710 /* Size with which the following vectors are currently allocated.
1711 That is so we can make them bigger as needed,
1712 but never make them smaller. */
1713 static int regs_allocated_size;
1715 static const char ** regstart, ** regend;
1716 static const char ** old_regstart, ** old_regend;
1717 static const char **best_regstart, **best_regend;
1718 static register_info_type *reg_info;
1719 static const char **reg_dummy;
1720 static register_info_type *reg_info_dummy;
1722 /* Make the register vectors big enough for NUM_REGS registers,
1723 but don't make them smaller. */
1726 regex_grow_registers (num_regs)
1729 if (num_regs > regs_allocated_size)
1731 RETALLOC_IF (regstart, num_regs, const char *);
1732 RETALLOC_IF (regend, num_regs, const char *);
1733 RETALLOC_IF (old_regstart, num_regs, const char *);
1734 RETALLOC_IF (old_regend, num_regs, const char *);
1735 RETALLOC_IF (best_regstart, num_regs, const char *);
1736 RETALLOC_IF (best_regend, num_regs, const char *);
1737 RETALLOC_IF (reg_info, num_regs, register_info_type);
1738 RETALLOC_IF (reg_dummy, num_regs, const char *);
1739 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1741 regs_allocated_size = num_regs;
1745 #endif /* not MATCH_MAY_ALLOCATE */
1747 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
1751 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1752 Returns one of error codes defined in `regex.h', or zero for success.
1754 Assumes the `allocated' (and perhaps `buffer') and `translate'
1755 fields are set in BUFP on entry.
1757 If it succeeds, results are put in BUFP (if it returns an error, the
1758 contents of BUFP are undefined):
1759 `buffer' is the compiled pattern;
1760 `syntax' is set to SYNTAX;
1761 `used' is set to the length of the compiled pattern;
1762 `fastmap_accurate' is zero;
1763 `re_nsub' is the number of subexpressions in PATTERN;
1764 `not_bol' and `not_eol' are zero;
1766 The `fastmap' and `newline_anchor' fields are neither
1767 examined nor set. */
1769 /* Return, freeing storage we allocated. */
1770 #define FREE_STACK_RETURN(value) \
1771 return (free (compile_stack.stack), value)
1773 static reg_errcode_t
1774 regex_compile (pattern, size, syntax, bufp)
1775 const char *pattern;
1777 reg_syntax_t syntax;
1778 struct re_pattern_buffer *bufp;
1780 /* We fetch characters from PATTERN here. Even though PATTERN is
1781 `char *' (i.e., signed), we declare these variables as unsigned, so
1782 they can be reliably used as array indices. */
1783 register unsigned char c, c1;
1785 /* A random temporary spot in PATTERN. */
1788 /* Points to the end of the buffer, where we should append. */
1789 register unsigned char *b;
1791 /* Keeps track of unclosed groups. */
1792 compile_stack_type compile_stack;
1794 /* Points to the current (ending) position in the pattern. */
1795 const char *p = pattern;
1796 const char *pend = pattern + size;
1798 /* How to translate the characters in the pattern. */
1799 RE_TRANSLATE_TYPE translate = bufp->translate;
1801 /* Address of the count-byte of the most recently inserted `exactn'
1802 command. This makes it possible to tell if a new exact-match
1803 character can be added to that command or if the character requires
1804 a new `exactn' command. */
1805 unsigned char *pending_exact = 0;
1807 /* Address of start of the most recently finished expression.
1808 This tells, e.g., postfix * where to find the start of its
1809 operand. Reset at the beginning of groups and alternatives. */
1810 unsigned char *laststart = 0;
1812 /* Address of beginning of regexp, or inside of last group. */
1813 unsigned char *begalt;
1815 /* Place in the uncompiled pattern (i.e., the {) to
1816 which to go back if the interval is invalid. */
1817 const char *beg_interval;
1819 /* Address of the place where a forward jump should go to the end of
1820 the containing expression. Each alternative of an `or' -- except the
1821 last -- ends with a forward jump of this sort. */
1822 unsigned char *fixup_alt_jump = 0;
1824 /* Counts open-groups as they are encountered. Remembered for the
1825 matching close-group on the compile stack, so the same register
1826 number is put in the stop_memory as the start_memory. */
1827 regnum_t regnum = 0;
1830 DEBUG_PRINT1 ("\nCompiling pattern: ");
1833 unsigned debug_count;
1835 for (debug_count = 0; debug_count < size; debug_count++)
1836 putchar (pattern[debug_count]);
1841 /* Initialize the compile stack. */
1842 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1843 if (compile_stack.stack == NULL)
1846 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1847 compile_stack.avail = 0;
1849 /* Initialize the pattern buffer. */
1850 bufp->syntax = syntax;
1851 bufp->fastmap_accurate = 0;
1852 bufp->not_bol = bufp->not_eol = 0;
1854 /* Set `used' to zero, so that if we return an error, the pattern
1855 printer (for debugging) will think there's no pattern. We reset it
1859 /* Always count groups, whether or not bufp->no_sub is set. */
1862 #if !defined (emacs) && !defined (SYNTAX_TABLE)
1863 /* Initialize the syntax table. */
1864 init_syntax_once ();
1867 if (bufp->allocated == 0)
1870 { /* If zero allocated, but buffer is non-null, try to realloc
1871 enough space. This loses if buffer's address is bogus, but
1872 that is the user's responsibility. */
1873 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1876 { /* Caller did not allocate a buffer. Do it for them. */
1877 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1879 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1881 bufp->allocated = INIT_BUF_SIZE;
1884 begalt = b = bufp->buffer;
1886 /* Loop through the uncompiled pattern until we're at the end. */
1895 if ( /* If at start of pattern, it's an operator. */
1897 /* If context independent, it's an operator. */
1898 || syntax & RE_CONTEXT_INDEP_ANCHORS
1899 /* Otherwise, depends on what's come before. */
1900 || at_begline_loc_p (pattern, p, syntax))
1910 if ( /* If at end of pattern, it's an operator. */
1912 /* If context independent, it's an operator. */
1913 || syntax & RE_CONTEXT_INDEP_ANCHORS
1914 /* Otherwise, depends on what's next. */
1915 || at_endline_loc_p (p, pend, syntax))
1925 if ((syntax & RE_BK_PLUS_QM)
1926 || (syntax & RE_LIMITED_OPS))
1930 /* If there is no previous pattern... */
1933 if (syntax & RE_CONTEXT_INVALID_OPS)
1934 FREE_STACK_RETURN (REG_BADRPT);
1935 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1940 /* Are we optimizing this jump? */
1941 boolean keep_string_p = false;
1943 /* 1 means zero (many) matches is allowed. */
1944 char zero_times_ok = 0, many_times_ok = 0;
1946 /* If there is a sequence of repetition chars, collapse it
1947 down to just one (the right one). We can't combine
1948 interval operators with these because of, e.g., `a{2}*',
1949 which should only match an even number of `a's. */
1953 zero_times_ok |= c != '+';
1954 many_times_ok |= c != '?';
1962 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
1965 else if (syntax & RE_BK_PLUS_QM && c == '\\')
1967 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1970 if (!(c1 == '+' || c1 == '?'))
1985 /* If we get here, we found another repeat character. */
1988 /* Star, etc. applied to an empty pattern is equivalent
1989 to an empty pattern. */
1993 /* Now we know whether or not zero matches is allowed
1994 and also whether or not two or more matches is allowed. */
1996 { /* More than one repetition is allowed, so put in at the
1997 end a backward relative jump from `b' to before the next
1998 jump we're going to put in below (which jumps from
1999 laststart to after this jump).
2001 But if we are at the `*' in the exact sequence `.*\n',
2002 insert an unconditional jump backwards to the .,
2003 instead of the beginning of the loop. This way we only
2004 push a failure point once, instead of every time
2005 through the loop. */
2006 assert (p - 1 > pattern);
2008 /* Allocate the space for the jump. */
2009 GET_BUFFER_SPACE (3);
2011 /* We know we are not at the first character of the pattern,
2012 because laststart was nonzero. And we've already
2013 incremented `p', by the way, to be the character after
2014 the `*'. Do we have to do something analogous here
2015 for null bytes, because of RE_DOT_NOT_NULL? */
2016 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2018 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2019 && !(syntax & RE_DOT_NEWLINE))
2020 { /* We have .*\n. */
2021 STORE_JUMP (jump, b, laststart);
2022 keep_string_p = true;
2025 /* Anything else. */
2026 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
2028 /* We've added more stuff to the buffer. */
2032 /* On failure, jump from laststart to b + 3, which will be the
2033 end of the buffer after this jump is inserted. */
2034 GET_BUFFER_SPACE (3);
2035 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2043 /* At least one repetition is required, so insert a
2044 `dummy_failure_jump' before the initial
2045 `on_failure_jump' instruction of the loop. This
2046 effects a skip over that instruction the first time
2047 we hit that loop. */
2048 GET_BUFFER_SPACE (3);
2049 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
2064 boolean had_char_class = false;
2066 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2068 /* Ensure that we have enough space to push a charset: the
2069 opcode, the length count, and the bitset; 34 bytes in all. */
2070 GET_BUFFER_SPACE (34);
2074 /* We test `*p == '^' twice, instead of using an if
2075 statement, so we only need one BUF_PUSH. */
2076 BUF_PUSH (*p == '^' ? charset_not : charset);
2080 /* Remember the first position in the bracket expression. */
2083 /* Push the number of bytes in the bitmap. */
2084 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
2086 /* Clear the whole map. */
2087 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
2089 /* charset_not matches newline according to a syntax bit. */
2090 if ((re_opcode_t) b[-2] == charset_not
2091 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2092 SET_LIST_BIT ('\n');
2094 /* Read in characters and ranges, setting map bits. */
2097 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2101 /* \ might escape characters inside [...] and [^...]. */
2102 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2104 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2111 /* Could be the end of the bracket expression. If it's
2112 not (i.e., when the bracket expression is `[]' so
2113 far), the ']' character bit gets set way below. */
2114 if (c == ']' && p != p1 + 1)
2117 /* Look ahead to see if it's a range when the last thing
2118 was a character class. */
2119 if (had_char_class && c == '-' && *p != ']')
2120 FREE_STACK_RETURN (REG_ERANGE);
2122 /* Look ahead to see if it's a range when the last thing
2123 was a character: if this is a hyphen not at the
2124 beginning or the end of a list, then it's the range
2127 && !(p - 2 >= pattern && p[-2] == '[')
2128 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2132 = compile_range (&p, pend, translate, syntax, b);
2133 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2136 else if (p[0] == '-' && p[1] != ']')
2137 { /* This handles ranges made up of characters only. */
2140 /* Move past the `-'. */
2143 ret = compile_range (&p, pend, translate, syntax, b);
2144 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2147 /* See if we're at the beginning of a possible character
2150 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2151 { /* Leave room for the null. */
2152 char str[CHAR_CLASS_MAX_LENGTH + 1];
2157 /* If pattern is `[[:'. */
2158 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2163 if (c == ':' || c == ']' || p == pend
2164 || c1 == CHAR_CLASS_MAX_LENGTH)
2170 /* If isn't a word bracketed by `[:' and:`]':
2171 undo the ending character, the letters, and leave
2172 the leading `:' and `[' (but set bits for them). */
2173 if (c == ':' && *p == ']')
2175 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
2176 boolean is_lower = STREQ (str, "lower");
2177 boolean is_upper = STREQ (str, "upper");
2183 FREE_STACK_RETURN (REG_ECTYPE);
2185 /* Throw away the ] at the end of the character
2189 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2191 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
2193 if (iswctype (btowc (ch), wt))
2196 if (translate && (is_upper || is_lower)
2197 && (ISUPPER (ch) || ISLOWER (ch)))
2201 had_char_class = true;
2204 boolean is_alnum = STREQ (str, "alnum");
2205 boolean is_alpha = STREQ (str, "alpha");
2206 boolean is_blank = STREQ (str, "blank");
2207 boolean is_cntrl = STREQ (str, "cntrl");
2208 boolean is_digit = STREQ (str, "digit");
2209 boolean is_graph = STREQ (str, "graph");
2210 boolean is_lower = STREQ (str, "lower");
2211 boolean is_print = STREQ (str, "print");
2212 boolean is_punct = STREQ (str, "punct");
2213 boolean is_space = STREQ (str, "space");
2214 boolean is_upper = STREQ (str, "upper");
2215 boolean is_xdigit = STREQ (str, "xdigit");
2217 if (!IS_CHAR_CLASS (str))
2218 FREE_STACK_RETURN (REG_ECTYPE);
2220 /* Throw away the ] at the end of the character
2224 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2226 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2228 /* This was split into 3 if's to
2229 avoid an arbitrary limit in some compiler. */
2230 if ( (is_alnum && ISALNUM (ch))
2231 || (is_alpha && ISALPHA (ch))
2232 || (is_blank && ISBLANK (ch))
2233 || (is_cntrl && ISCNTRL (ch)))
2235 if ( (is_digit && ISDIGIT (ch))
2236 || (is_graph && ISGRAPH (ch))
2237 || (is_lower && ISLOWER (ch))
2238 || (is_print && ISPRINT (ch)))
2240 if ( (is_punct && ISPUNCT (ch))
2241 || (is_space && ISSPACE (ch))
2242 || (is_upper && ISUPPER (ch))
2243 || (is_xdigit && ISXDIGIT (ch)))
2245 if ( translate && (is_upper || is_lower)
2246 && (ISUPPER (ch) || ISLOWER (ch)))
2249 had_char_class = true;
2250 #endif /* libc || wctype.h */
2259 had_char_class = false;
2264 had_char_class = false;
2269 /* Discard any (non)matching list bytes that are all 0 at the
2270 end of the map. Decrease the map-length byte too. */
2271 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2279 if (syntax & RE_NO_BK_PARENS)
2286 if (syntax & RE_NO_BK_PARENS)
2293 if (syntax & RE_NEWLINE_ALT)
2300 if (syntax & RE_NO_BK_VBAR)
2307 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2308 goto handle_interval;
2314 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2316 /* Do not translate the character after the \, so that we can
2317 distinguish, e.g., \B from \b, even if we normally would
2318 translate, e.g., B to b. */
2324 if (syntax & RE_NO_BK_PARENS)
2325 goto normal_backslash;
2331 if (COMPILE_STACK_FULL)
2333 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2334 compile_stack_elt_t);
2335 if (compile_stack.stack == NULL) return REG_ESPACE;
2337 compile_stack.size <<= 1;
2340 /* These are the values to restore when we hit end of this
2341 group. They are all relative offsets, so that if the
2342 whole pattern moves because of realloc, they will still
2344 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2345 COMPILE_STACK_TOP.fixup_alt_jump
2346 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2347 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2348 COMPILE_STACK_TOP.regnum = regnum;
2350 /* We will eventually replace the 0 with the number of
2351 groups inner to this one. But do not push a
2352 start_memory for groups beyond the last one we can
2353 represent in the compiled pattern. */
2354 if (regnum <= MAX_REGNUM)
2356 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2357 BUF_PUSH_3 (start_memory, regnum, 0);
2360 compile_stack.avail++;
2365 /* If we've reached MAX_REGNUM groups, then this open
2366 won't actually generate any code, so we'll have to
2367 clear pending_exact explicitly. */
2373 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2375 if (COMPILE_STACK_EMPTY)
2376 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2377 goto normal_backslash;
2379 FREE_STACK_RETURN (REG_ERPAREN);
2383 { /* Push a dummy failure point at the end of the
2384 alternative for a possible future
2385 `pop_failure_jump' to pop. See comments at
2386 `push_dummy_failure' in `re_match_2'. */
2387 BUF_PUSH (push_dummy_failure);
2389 /* We allocated space for this jump when we assigned
2390 to `fixup_alt_jump', in the `handle_alt' case below. */
2391 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2394 /* See similar code for backslashed left paren above. */
2395 if (COMPILE_STACK_EMPTY)
2396 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2399 FREE_STACK_RETURN (REG_ERPAREN);
2401 /* Since we just checked for an empty stack above, this
2402 ``can't happen''. */
2403 assert (compile_stack.avail != 0);
2405 /* We don't just want to restore into `regnum', because
2406 later groups should continue to be numbered higher,
2407 as in `(ab)c(de)' -- the second group is #2. */
2408 regnum_t this_group_regnum;
2410 compile_stack.avail--;
2411 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2413 = COMPILE_STACK_TOP.fixup_alt_jump
2414 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2416 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2417 this_group_regnum = COMPILE_STACK_TOP.regnum;
2418 /* If we've reached MAX_REGNUM groups, then this open
2419 won't actually generate any code, so we'll have to
2420 clear pending_exact explicitly. */
2423 /* We're at the end of the group, so now we know how many
2424 groups were inside this one. */
2425 if (this_group_regnum <= MAX_REGNUM)
2427 unsigned char *inner_group_loc
2428 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2430 *inner_group_loc = regnum - this_group_regnum;
2431 BUF_PUSH_3 (stop_memory, this_group_regnum,
2432 regnum - this_group_regnum);
2438 case '|': /* `\|'. */
2439 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2440 goto normal_backslash;
2442 if (syntax & RE_LIMITED_OPS)
2445 /* Insert before the previous alternative a jump which
2446 jumps to this alternative if the former fails. */
2447 GET_BUFFER_SPACE (3);
2448 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2452 /* The alternative before this one has a jump after it
2453 which gets executed if it gets matched. Adjust that
2454 jump so it will jump to this alternative's analogous
2455 jump (put in below, which in turn will jump to the next
2456 (if any) alternative's such jump, etc.). The last such
2457 jump jumps to the correct final destination. A picture:
2463 If we are at `b', then fixup_alt_jump right now points to a
2464 three-byte space after `a'. We'll put in the jump, set
2465 fixup_alt_jump to right after `b', and leave behind three
2466 bytes which we'll fill in when we get to after `c'. */
2469 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2471 /* Mark and leave space for a jump after this alternative,
2472 to be filled in later either by next alternative or
2473 when know we're at the end of a series of alternatives. */
2475 GET_BUFFER_SPACE (3);
2484 /* If \{ is a literal. */
2485 if (!(syntax & RE_INTERVALS)
2486 /* If we're at `\{' and it's not the open-interval
2488 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2489 || (p - 2 == pattern && p == pend))
2490 goto normal_backslash;
2494 /* If got here, then the syntax allows intervals. */
2496 /* At least (most) this many matches must be made. */
2497 int lower_bound = -1, upper_bound = -1;
2499 beg_interval = p - 1;
2503 if (syntax & RE_NO_BK_BRACES)
2504 goto unfetch_interval;
2506 FREE_STACK_RETURN (REG_EBRACE);
2509 GET_UNSIGNED_NUMBER (lower_bound);
2513 GET_UNSIGNED_NUMBER (upper_bound);
2514 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2517 /* Interval such as `{1}' => match exactly once. */
2518 upper_bound = lower_bound;
2520 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2521 || lower_bound > upper_bound)
2523 if (syntax & RE_NO_BK_BRACES)
2524 goto unfetch_interval;
2526 FREE_STACK_RETURN (REG_BADBR);
2529 if (!(syntax & RE_NO_BK_BRACES))
2531 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2538 if (syntax & RE_NO_BK_BRACES)
2539 goto unfetch_interval;
2541 FREE_STACK_RETURN (REG_BADBR);
2544 /* We just parsed a valid interval. */
2546 /* If it's invalid to have no preceding re. */
2549 if (syntax & RE_CONTEXT_INVALID_OPS)
2550 FREE_STACK_RETURN (REG_BADRPT);
2551 else if (syntax & RE_CONTEXT_INDEP_OPS)
2554 goto unfetch_interval;
2557 /* If the upper bound is zero, don't want to succeed at
2558 all; jump from `laststart' to `b + 3', which will be
2559 the end of the buffer after we insert the jump. */
2560 if (upper_bound == 0)
2562 GET_BUFFER_SPACE (3);
2563 INSERT_JUMP (jump, laststart, b + 3);
2567 /* Otherwise, we have a nontrivial interval. When
2568 we're all done, the pattern will look like:
2569 set_number_at <jump count> <upper bound>
2570 set_number_at <succeed_n count> <lower bound>
2571 succeed_n <after jump addr> <succeed_n count>
2573 jump_n <succeed_n addr> <jump count>
2574 (The upper bound and `jump_n' are omitted if
2575 `upper_bound' is 1, though.) */
2577 { /* If the upper bound is > 1, we need to insert
2578 more at the end of the loop. */
2579 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2581 GET_BUFFER_SPACE (nbytes);
2583 /* Initialize lower bound of the `succeed_n', even
2584 though it will be set during matching by its
2585 attendant `set_number_at' (inserted next),
2586 because `re_compile_fastmap' needs to know.
2587 Jump to the `jump_n' we might insert below. */
2588 INSERT_JUMP2 (succeed_n, laststart,
2589 b + 5 + (upper_bound > 1) * 5,
2593 /* Code to initialize the lower bound. Insert
2594 before the `succeed_n'. The `5' is the last two
2595 bytes of this `set_number_at', plus 3 bytes of
2596 the following `succeed_n'. */
2597 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2600 if (upper_bound > 1)
2601 { /* More than one repetition is allowed, so
2602 append a backward jump to the `succeed_n'
2603 that starts this interval.
2605 When we've reached this during matching,
2606 we'll have matched the interval once, so
2607 jump back only `upper_bound - 1' times. */
2608 STORE_JUMP2 (jump_n, b, laststart + 5,
2612 /* The location we want to set is the second
2613 parameter of the `jump_n'; that is `b-2' as
2614 an absolute address. `laststart' will be
2615 the `set_number_at' we're about to insert;
2616 `laststart+3' the number to set, the source
2617 for the relative address. But we are
2618 inserting into the middle of the pattern --
2619 so everything is getting moved up by 5.
2620 Conclusion: (b - 2) - (laststart + 3) + 5,
2621 i.e., b - laststart.
2623 We insert this at the beginning of the loop
2624 so that if we fail during matching, we'll
2625 reinitialize the bounds. */
2626 insert_op2 (set_number_at, laststart, b - laststart,
2627 upper_bound - 1, b);
2632 beg_interval = NULL;
2637 /* If an invalid interval, match the characters as literals. */
2638 assert (beg_interval);
2640 beg_interval = NULL;
2642 /* normal_char and normal_backslash need `c'. */
2645 if (!(syntax & RE_NO_BK_BRACES))
2647 if (p > pattern && p[-1] == '\\')
2648 goto normal_backslash;
2653 /* There is no way to specify the before_dot and after_dot
2654 operators. rms says this is ok. --karl */
2662 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2668 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2674 if (re_syntax_options & RE_NO_GNU_OPS)
2677 BUF_PUSH (wordchar);
2682 if (re_syntax_options & RE_NO_GNU_OPS)
2685 BUF_PUSH (notwordchar);
2690 if (re_syntax_options & RE_NO_GNU_OPS)
2696 if (re_syntax_options & RE_NO_GNU_OPS)
2702 if (re_syntax_options & RE_NO_GNU_OPS)
2704 BUF_PUSH (wordbound);
2708 if (re_syntax_options & RE_NO_GNU_OPS)
2710 BUF_PUSH (notwordbound);
2714 if (re_syntax_options & RE_NO_GNU_OPS)
2720 if (re_syntax_options & RE_NO_GNU_OPS)
2725 case '1': case '2': case '3': case '4': case '5':
2726 case '6': case '7': case '8': case '9':
2727 if (syntax & RE_NO_BK_REFS)
2733 FREE_STACK_RETURN (REG_ESUBREG);
2735 /* Can't back reference to a subexpression if inside of it. */
2736 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
2740 BUF_PUSH_2 (duplicate, c1);
2746 if (syntax & RE_BK_PLUS_QM)
2749 goto normal_backslash;
2753 /* You might think it would be useful for \ to mean
2754 not to translate; but if we don't translate it
2755 it will never match anything. */
2763 /* Expects the character in `c'. */
2765 /* If no exactn currently being built. */
2768 /* If last exactn not at current position. */
2769 || pending_exact + *pending_exact + 1 != b
2771 /* We have only one byte following the exactn for the count. */
2772 || *pending_exact == (1 << BYTEWIDTH) - 1
2774 /* If followed by a repetition operator. */
2775 || *p == '*' || *p == '^'
2776 || ((syntax & RE_BK_PLUS_QM)
2777 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2778 : (*p == '+' || *p == '?'))
2779 || ((syntax & RE_INTERVALS)
2780 && ((syntax & RE_NO_BK_BRACES)
2782 : (p[0] == '\\' && p[1] == '{'))))
2784 /* Start building a new exactn. */
2788 BUF_PUSH_2 (exactn, 0);
2789 pending_exact = b - 1;
2796 } /* while p != pend */
2799 /* Through the pattern now. */
2802 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2804 if (!COMPILE_STACK_EMPTY)
2805 FREE_STACK_RETURN (REG_EPAREN);
2807 /* If we don't want backtracking, force success
2808 the first time we reach the end of the compiled pattern. */
2809 if (syntax & RE_NO_POSIX_BACKTRACKING)
2812 free (compile_stack.stack);
2814 /* We have succeeded; set the length of the buffer. */
2815 bufp->used = b - bufp->buffer;
2820 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2821 print_compiled_pattern (bufp);
2825 #ifndef MATCH_MAY_ALLOCATE
2826 /* Initialize the failure stack to the largest possible stack. This
2827 isn't necessary unless we're trying to avoid calling alloca in
2828 the search and match routines. */
2830 int num_regs = bufp->re_nsub + 1;
2832 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2833 is strictly greater than re_max_failures, the largest possible stack
2834 is 2 * re_max_failures failure points. */
2835 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2837 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2840 if (! fail_stack.stack)
2842 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2843 * sizeof (fail_stack_elt_t));
2846 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2848 * sizeof (fail_stack_elt_t)));
2849 #else /* not emacs */
2850 if (! fail_stack.stack)
2852 = (fail_stack_elt_t *) malloc (fail_stack.size
2853 * sizeof (fail_stack_elt_t));
2856 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2858 * sizeof (fail_stack_elt_t)));
2859 #endif /* not emacs */
2862 regex_grow_registers (num_regs);
2864 #endif /* not MATCH_MAY_ALLOCATE */
2867 } /* regex_compile */
2869 /* Subroutines for `regex_compile'. */
2871 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2874 store_op1 (op, loc, arg)
2879 *loc = (unsigned char) op;
2880 STORE_NUMBER (loc + 1, arg);
2884 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2887 store_op2 (op, loc, arg1, arg2)
2892 *loc = (unsigned char) op;
2893 STORE_NUMBER (loc + 1, arg1);
2894 STORE_NUMBER (loc + 3, arg2);
2898 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
2899 for OP followed by two-byte integer parameter ARG. */
2902 insert_op1 (op, loc, arg, end)
2908 register unsigned char *pfrom = end;
2909 register unsigned char *pto = end + 3;
2911 while (pfrom != loc)
2914 store_op1 (op, loc, arg);
2918 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2921 insert_op2 (op, loc, arg1, arg2, end)
2927 register unsigned char *pfrom = end;
2928 register unsigned char *pto = end + 5;
2930 while (pfrom != loc)
2933 store_op2 (op, loc, arg1, arg2);
2937 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
2938 after an alternative or a begin-subexpression. We assume there is at
2939 least one character before the ^. */
2942 at_begline_loc_p (pattern, p, syntax)
2943 const char *pattern, *p;
2944 reg_syntax_t syntax;
2946 const char *prev = p - 2;
2947 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
2950 /* After a subexpression? */
2951 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
2952 /* After an alternative? */
2953 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
2957 /* The dual of at_begline_loc_p. This one is for $. We assume there is
2958 at least one character after the $, i.e., `P < PEND'. */
2961 at_endline_loc_p (p, pend, syntax)
2962 const char *p, *pend;
2963 reg_syntax_t syntax;
2965 const char *next = p;
2966 boolean next_backslash = *next == '\\';
2967 const char *next_next = p + 1 < pend ? p + 1 : 0;
2970 /* Before a subexpression? */
2971 (syntax & RE_NO_BK_PARENS ? *next == ')'
2972 : next_backslash && next_next && *next_next == ')')
2973 /* Before an alternative? */
2974 || (syntax & RE_NO_BK_VBAR ? *next == '|'
2975 : next_backslash && next_next && *next_next == '|');
2979 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
2980 false if it's not. */
2983 group_in_compile_stack (compile_stack, regnum)
2984 compile_stack_type compile_stack;
2989 for (this_element = compile_stack.avail - 1;
2992 if (compile_stack.stack[this_element].regnum == regnum)
2999 /* Read the ending character of a range (in a bracket expression) from the
3000 uncompiled pattern *P_PTR (which ends at PEND). We assume the
3001 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
3002 Then we set the translation of all bits between the starting and
3003 ending characters (inclusive) in the compiled pattern B.
3005 Return an error code.
3007 We use these short variable names so we can use the same macros as
3008 `regex_compile' itself. */
3010 static reg_errcode_t
3011 compile_range (p_ptr, pend, translate, syntax, b)
3012 const char **p_ptr, *pend;
3013 RE_TRANSLATE_TYPE translate;
3014 reg_syntax_t syntax;
3019 const char *p = *p_ptr;
3020 unsigned int range_start, range_end;
3025 /* Even though the pattern is a signed `char *', we need to fetch
3026 with unsigned char *'s; if the high bit of the pattern character
3027 is set, the range endpoints will be negative if we fetch using a
3030 We also want to fetch the endpoints without translating them; the
3031 appropriate translation is done in the bit-setting loop below. */
3032 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
3033 range_start = ((const unsigned char *) p)[-2];
3034 range_end = ((const unsigned char *) p)[0];
3036 /* Have to increment the pointer into the pattern string, so the
3037 caller isn't still at the ending character. */
3040 /* If the start is after the end, the range is empty. */
3041 if (range_start > range_end)
3042 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
3044 /* Here we see why `this_char' has to be larger than an `unsigned
3045 char' -- the range is inclusive, so if `range_end' == 0xff
3046 (assuming 8-bit characters), we would otherwise go into an infinite
3047 loop, since all characters <= 0xff. */
3048 for (this_char = range_start; this_char <= range_end; this_char++)
3050 SET_LIST_BIT (TRANSLATE (this_char));
3056 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3057 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3058 characters can start a string that matches the pattern. This fastmap
3059 is used by re_search to skip quickly over impossible starting points.
3061 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3062 area as BUFP->fastmap.
3064 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3067 Returns 0 if we succeed, -2 if an internal error. */
3070 re_compile_fastmap (bufp)
3071 struct re_pattern_buffer *bufp;
3074 #ifdef MATCH_MAY_ALLOCATE
3075 fail_stack_type fail_stack;
3077 #ifndef REGEX_MALLOC
3080 /* We don't push any register information onto the failure stack. */
3081 unsigned num_regs = 0;
3083 register char *fastmap = bufp->fastmap;
3084 unsigned char *pattern = bufp->buffer;
3085 unsigned char *p = pattern;
3086 register unsigned char *pend = pattern + bufp->used;
3089 /* This holds the pointer to the failure stack, when
3090 it is allocated relocatably. */
3091 fail_stack_elt_t *failure_stack_ptr;
3094 /* Assume that each path through the pattern can be null until
3095 proven otherwise. We set this false at the bottom of switch
3096 statement, to which we get only if a particular path doesn't
3097 match the empty string. */
3098 boolean path_can_be_null = true;
3100 /* We aren't doing a `succeed_n' to begin with. */
3101 boolean succeed_n_p = false;
3103 assert (fastmap != NULL && p != NULL);
3106 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
3107 bufp->fastmap_accurate = 1; /* It will be when we're done. */
3108 bufp->can_be_null = 0;
3112 if (p == pend || *p == succeed)
3114 /* We have reached the (effective) end of pattern. */
3115 if (!FAIL_STACK_EMPTY ())
3117 bufp->can_be_null |= path_can_be_null;
3119 /* Reset for next path. */
3120 path_can_be_null = true;
3122 p = fail_stack.stack[--fail_stack.avail].pointer;
3130 /* We should never be about to go beyond the end of the pattern. */
3133 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3136 /* I guess the idea here is to simply not bother with a fastmap
3137 if a backreference is used, since it's too hard to figure out
3138 the fastmap for the corresponding group. Setting
3139 `can_be_null' stops `re_search_2' from using the fastmap, so
3140 that is all we do. */
3142 bufp->can_be_null = 1;
3146 /* Following are the cases which match a character. These end
3155 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3156 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3162 /* Chars beyond end of map must be allowed. */
3163 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3166 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3167 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3173 for (j = 0; j < (1 << BYTEWIDTH); j++)
3174 if (SYNTAX (j) == Sword)
3180 for (j = 0; j < (1 << BYTEWIDTH); j++)
3181 if (SYNTAX (j) != Sword)
3188 int fastmap_newline = fastmap['\n'];
3190 /* `.' matches anything ... */
3191 for (j = 0; j < (1 << BYTEWIDTH); j++)
3194 /* ... except perhaps newline. */
3195 if (!(bufp->syntax & RE_DOT_NEWLINE))
3196 fastmap['\n'] = fastmap_newline;
3198 /* Return if we have already set `can_be_null'; if we have,
3199 then the fastmap is irrelevant. Something's wrong here. */
3200 else if (bufp->can_be_null)
3203 /* Otherwise, have to check alternative paths. */
3210 for (j = 0; j < (1 << BYTEWIDTH); j++)
3211 if (SYNTAX (j) == (enum syntaxcode) k)
3218 for (j = 0; j < (1 << BYTEWIDTH); j++)
3219 if (SYNTAX (j) != (enum syntaxcode) k)
3224 /* All cases after this match the empty string. These end with
3244 case push_dummy_failure:
3249 case pop_failure_jump:
3250 case maybe_pop_jump:
3253 case dummy_failure_jump:
3254 EXTRACT_NUMBER_AND_INCR (j, p);
3259 /* Jump backward implies we just went through the body of a
3260 loop and matched nothing. Opcode jumped to should be
3261 `on_failure_jump' or `succeed_n'. Just treat it like an
3262 ordinary jump. For a * loop, it has pushed its failure
3263 point already; if so, discard that as redundant. */
3264 if ((re_opcode_t) *p != on_failure_jump
3265 && (re_opcode_t) *p != succeed_n)
3269 EXTRACT_NUMBER_AND_INCR (j, p);
3272 /* If what's on the stack is where we are now, pop it. */
3273 if (!FAIL_STACK_EMPTY ()
3274 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3280 case on_failure_jump:
3281 case on_failure_keep_string_jump:
3282 handle_on_failure_jump:
3283 EXTRACT_NUMBER_AND_INCR (j, p);
3285 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3286 end of the pattern. We don't want to push such a point,
3287 since when we restore it above, entering the switch will
3288 increment `p' past the end of the pattern. We don't need
3289 to push such a point since we obviously won't find any more
3290 fastmap entries beyond `pend'. Such a pattern can match
3291 the null string, though. */
3294 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3296 RESET_FAIL_STACK ();
3301 bufp->can_be_null = 1;
3305 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3306 succeed_n_p = false;
3313 /* Get to the number of times to succeed. */
3316 /* Increment p past the n for when k != 0. */
3317 EXTRACT_NUMBER_AND_INCR (k, p);
3321 succeed_n_p = true; /* Spaghetti code alert. */
3322 goto handle_on_failure_jump;
3339 abort (); /* We have listed all the cases. */
3342 /* Getting here means we have found the possible starting
3343 characters for one path of the pattern -- and that the empty
3344 string does not match. We need not follow this path further.
3345 Instead, look at the next alternative (remembered on the
3346 stack), or quit if no more. The test at the top of the loop
3347 does these things. */
3348 path_can_be_null = false;
3352 /* Set `can_be_null' for the last path (also the first path, if the
3353 pattern is empty). */
3354 bufp->can_be_null |= path_can_be_null;
3357 RESET_FAIL_STACK ();
3359 } /* re_compile_fastmap */
3361 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3362 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3363 this memory for recording register information. STARTS and ENDS
3364 must be allocated using the malloc library routine, and must each
3365 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3367 If NUM_REGS == 0, then subsequent matches should allocate their own
3370 Unless this function is called, the first search or match using
3371 PATTERN_BUFFER will allocate its own register data, without
3372 freeing the old data. */
3375 re_set_registers (bufp, regs, num_regs, starts, ends)
3376 struct re_pattern_buffer *bufp;
3377 struct re_registers *regs;
3379 regoff_t *starts, *ends;
3383 bufp->regs_allocated = REGS_REALLOCATE;
3384 regs->num_regs = num_regs;
3385 regs->start = starts;
3390 bufp->regs_allocated = REGS_UNALLOCATED;
3392 regs->start = regs->end = (regoff_t *) 0;
3396 /* Searching routines. */
3398 /* Like re_search_2, below, but only one string is specified, and
3399 doesn't let you say where to stop matching. */
3402 re_search (bufp, string, size, startpos, range, regs)
3403 struct re_pattern_buffer *bufp;
3405 int size, startpos, range;
3406 struct re_registers *regs;
3408 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3413 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3414 virtual concatenation of STRING1 and STRING2, starting first at index
3415 STARTPOS, then at STARTPOS + 1, and so on.
3417 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3419 RANGE is how far to scan while trying to match. RANGE = 0 means try
3420 only at STARTPOS; in general, the last start tried is STARTPOS +
3423 In REGS, return the indices of the virtual concatenation of STRING1
3424 and STRING2 that matched the entire BUFP->buffer and its contained
3427 Do not consider matching one past the index STOP in the virtual
3428 concatenation of STRING1 and STRING2.
3430 We return either the position in the strings at which the match was
3431 found, -1 if no match, or -2 if error (such as failure
3435 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3436 struct re_pattern_buffer *bufp;
3437 const char *string1, *string2;
3441 struct re_registers *regs;
3445 register char *fastmap = bufp->fastmap;
3446 register RE_TRANSLATE_TYPE translate = bufp->translate;
3447 int total_size = size1 + size2;
3448 int endpos = startpos + range;
3450 /* Check for out-of-range STARTPOS. */
3451 if (startpos < 0 || startpos > total_size)
3454 /* Fix up RANGE if it might eventually take us outside
3455 the virtual concatenation of STRING1 and STRING2.
3456 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3458 range = 0 - startpos;
3459 else if (endpos > total_size)
3460 range = total_size - startpos;
3462 /* If the search isn't to be a backwards one, don't waste time in a
3463 search for a pattern that must be anchored. */
3464 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3473 /* In a forward search for something that starts with \=.
3474 don't keep searching past point. */
3475 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3477 range = PT - startpos;
3483 /* Update the fastmap now if not correct already. */
3484 if (fastmap && !bufp->fastmap_accurate)
3485 if (re_compile_fastmap (bufp) == -2)
3488 /* Loop through the string, looking for a place to start matching. */
3491 /* If a fastmap is supplied, skip quickly over characters that
3492 cannot be the start of a match. If the pattern can match the
3493 null string, however, we don't need to skip characters; we want
3494 the first null string. */
3495 if (fastmap && startpos < total_size && !bufp->can_be_null)
3497 if (range > 0) /* Searching forwards. */
3499 register const char *d;
3500 register int lim = 0;
3503 if (startpos < size1 && startpos + range >= size1)
3504 lim = range - (size1 - startpos);
3506 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3508 /* Written out as an if-else to avoid testing `translate'
3512 && !fastmap[(unsigned char)
3513 translate[(unsigned char) *d++]])
3516 while (range > lim && !fastmap[(unsigned char) *d++])
3519 startpos += irange - range;
3521 else /* Searching backwards. */
3523 register char c = (size1 == 0 || startpos >= size1
3524 ? string2[startpos - size1]
3525 : string1[startpos]);
3527 if (!fastmap[(unsigned char) TRANSLATE (c)])
3532 /* If can't match the null string, and that's all we have left, fail. */
3533 if (range >= 0 && startpos == total_size && fastmap
3534 && !bufp->can_be_null)
3537 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3538 startpos, regs, stop);
3539 #ifndef REGEX_MALLOC
3568 /* This converts PTR, a pointer into one of the search strings `string1'
3569 and `string2' into an offset from the beginning of that string. */
3570 #define POINTER_TO_OFFSET(ptr) \
3571 (FIRST_STRING_P (ptr) \
3572 ? ((regoff_t) ((ptr) - string1)) \
3573 : ((regoff_t) ((ptr) - string2 + size1)))
3575 /* Macros for dealing with the split strings in re_match_2. */
3577 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3579 /* Call before fetching a character with *d. This switches over to
3580 string2 if necessary. */
3581 #define PREFETCH() \
3584 /* End of string2 => fail. */ \
3585 if (dend == end_match_2) \
3587 /* End of string1 => advance to string2. */ \
3589 dend = end_match_2; \
3593 /* Test if at very beginning or at very end of the virtual concatenation
3594 of `string1' and `string2'. If only one string, it's `string2'. */
3595 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3596 #define AT_STRINGS_END(d) ((d) == end2)
3599 /* Test if D points to a character which is word-constituent. We have
3600 two special cases to check for: if past the end of string1, look at
3601 the first character in string2; and if before the beginning of
3602 string2, look at the last character in string1. */
3603 #define WORDCHAR_P(d) \
3604 (SYNTAX ((d) == end1 ? *string2 \
3605 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3608 /* Disabled due to a compiler bug -- see comment at case wordbound */
3610 /* Test if the character before D and the one at D differ with respect
3611 to being word-constituent. */
3612 #define AT_WORD_BOUNDARY(d) \
3613 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3614 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3617 /* Free everything we malloc. */
3618 #ifdef MATCH_MAY_ALLOCATE
3619 #define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3620 #define FREE_VARIABLES() \
3622 REGEX_FREE_STACK (fail_stack.stack); \
3623 FREE_VAR (regstart); \
3624 FREE_VAR (regend); \
3625 FREE_VAR (old_regstart); \
3626 FREE_VAR (old_regend); \
3627 FREE_VAR (best_regstart); \
3628 FREE_VAR (best_regend); \
3629 FREE_VAR (reg_info); \
3630 FREE_VAR (reg_dummy); \
3631 FREE_VAR (reg_info_dummy); \
3634 #define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3635 #endif /* not MATCH_MAY_ALLOCATE */
3637 /* These values must meet several constraints. They must not be valid
3638 register values; since we have a limit of 255 registers (because
3639 we use only one byte in the pattern for the register number), we can
3640 use numbers larger than 255. They must differ by 1, because of
3641 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3642 be larger than the value for the highest register, so we do not try
3643 to actually save any registers when none are active. */
3644 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3645 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3647 /* Matching routines. */
3649 #ifndef emacs /* Emacs never uses this. */
3650 /* re_match is like re_match_2 except it takes only a single string. */
3653 re_match (bufp, string, size, pos, regs)
3654 struct re_pattern_buffer *bufp;
3657 struct re_registers *regs;
3659 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3661 #ifndef REGEX_MALLOC
3668 #endif /* not emacs */
3670 static boolean group_match_null_string_p _RE_ARGS ((unsigned char **p,
3672 register_info_type *reg_info));
3673 static boolean alt_match_null_string_p _RE_ARGS ((unsigned char *p,
3675 register_info_type *reg_info));
3676 static boolean common_op_match_null_string_p _RE_ARGS ((unsigned char **p,
3678 register_info_type *reg_info));
3679 static int bcmp_translate _RE_ARGS ((const char *s1, const char *s2,
3680 int len, char *translate));
3682 /* re_match_2 matches the compiled pattern in BUFP against the
3683 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3684 and SIZE2, respectively). We start matching at POS, and stop
3687 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3688 store offsets for the substring each group matched in REGS. See the
3689 documentation for exactly how many groups we fill.
3691 We return -1 if no match, -2 if an internal error (such as the
3692 failure stack overflowing). Otherwise, we return the length of the
3693 matched substring. */
3696 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3697 struct re_pattern_buffer *bufp;
3698 const char *string1, *string2;
3701 struct re_registers *regs;
3704 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3706 #ifndef REGEX_MALLOC
3714 /* This is a separate function so that we can force an alloca cleanup
3717 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3718 struct re_pattern_buffer *bufp;
3719 const char *string1, *string2;
3722 struct re_registers *regs;
3725 /* General temporaries. */
3729 /* Just past the end of the corresponding string. */
3730 const char *end1, *end2;
3732 /* Pointers into string1 and string2, just past the last characters in
3733 each to consider matching. */
3734 const char *end_match_1, *end_match_2;
3736 /* Where we are in the data, and the end of the current string. */
3737 const char *d, *dend;
3739 /* Where we are in the pattern, and the end of the pattern. */
3740 unsigned char *p = bufp->buffer;
3741 register unsigned char *pend = p + bufp->used;
3743 /* Mark the opcode just after a start_memory, so we can test for an
3744 empty subpattern when we get to the stop_memory. */
3745 unsigned char *just_past_start_mem = 0;
3747 /* We use this to map every character in the string. */
3748 RE_TRANSLATE_TYPE translate = bufp->translate;
3750 /* Failure point stack. Each place that can handle a failure further
3751 down the line pushes a failure point on this stack. It consists of
3752 restart, regend, and reg_info for all registers corresponding to
3753 the subexpressions we're currently inside, plus the number of such
3754 registers, and, finally, two char *'s. The first char * is where
3755 to resume scanning the pattern; the second one is where to resume
3756 scanning the strings. If the latter is zero, the failure point is
3757 a ``dummy''; if a failure happens and the failure point is a dummy,
3758 it gets discarded and the next next one is tried. */
3759 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3760 fail_stack_type fail_stack;
3763 static unsigned failure_id = 0;
3764 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3768 /* This holds the pointer to the failure stack, when
3769 it is allocated relocatably. */
3770 fail_stack_elt_t *failure_stack_ptr;
3773 /* We fill all the registers internally, independent of what we
3774 return, for use in backreferences. The number here includes
3775 an element for register zero. */
3776 size_t num_regs = bufp->re_nsub + 1;
3778 /* The currently active registers. */
3779 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3780 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3782 /* Information on the contents of registers. These are pointers into
3783 the input strings; they record just what was matched (on this
3784 attempt) by a subexpression part of the pattern, that is, the
3785 regnum-th regstart pointer points to where in the pattern we began
3786 matching and the regnum-th regend points to right after where we
3787 stopped matching the regnum-th subexpression. (The zeroth register
3788 keeps track of what the whole pattern matches.) */
3789 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3790 const char **regstart, **regend;
3793 /* If a group that's operated upon by a repetition operator fails to
3794 match anything, then the register for its start will need to be
3795 restored because it will have been set to wherever in the string we
3796 are when we last see its open-group operator. Similarly for a
3798 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3799 const char **old_regstart, **old_regend;
3802 /* The is_active field of reg_info helps us keep track of which (possibly
3803 nested) subexpressions we are currently in. The matched_something
3804 field of reg_info[reg_num] helps us tell whether or not we have
3805 matched any of the pattern so far this time through the reg_num-th
3806 subexpression. These two fields get reset each time through any
3807 loop their register is in. */
3808 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3809 register_info_type *reg_info;
3812 /* The following record the register info as found in the above
3813 variables when we find a match better than any we've seen before.
3814 This happens as we backtrack through the failure points, which in
3815 turn happens only if we have not yet matched the entire string. */
3816 unsigned best_regs_set = false;
3817 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3818 const char **best_regstart, **best_regend;
3821 /* Logically, this is `best_regend[0]'. But we don't want to have to
3822 allocate space for that if we're not allocating space for anything
3823 else (see below). Also, we never need info about register 0 for
3824 any of the other register vectors, and it seems rather a kludge to
3825 treat `best_regend' differently than the rest. So we keep track of
3826 the end of the best match so far in a separate variable. We
3827 initialize this to NULL so that when we backtrack the first time
3828 and need to test it, it's not garbage. */
3829 const char *match_end = NULL;
3831 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3832 int set_regs_matched_done = 0;
3834 /* Used when we pop values we don't care about. */
3835 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3836 const char **reg_dummy;
3837 register_info_type *reg_info_dummy;
3841 /* Counts the total number of registers pushed. */
3842 unsigned num_regs_pushed = 0;
3845 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3849 #ifdef MATCH_MAY_ALLOCATE
3850 /* Do not bother to initialize all the register variables if there are
3851 no groups in the pattern, as it takes a fair amount of time. If
3852 there are groups, we include space for register 0 (the whole
3853 pattern), even though we never use it, since it simplifies the
3854 array indexing. We should fix this. */
3857 regstart = REGEX_TALLOC (num_regs, const char *);
3858 regend = REGEX_TALLOC (num_regs, const char *);
3859 old_regstart = REGEX_TALLOC (num_regs, const char *);
3860 old_regend = REGEX_TALLOC (num_regs, const char *);
3861 best_regstart = REGEX_TALLOC (num_regs, const char *);
3862 best_regend = REGEX_TALLOC (num_regs, const char *);
3863 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3864 reg_dummy = REGEX_TALLOC (num_regs, const char *);
3865 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3867 if (!(regstart && regend && old_regstart && old_regend && reg_info
3868 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3876 /* We must initialize all our variables to NULL, so that
3877 `FREE_VARIABLES' doesn't try to free them. */
3878 regstart = regend = old_regstart = old_regend = best_regstart
3879 = best_regend = reg_dummy = NULL;
3880 reg_info = reg_info_dummy = (register_info_type *) NULL;
3882 #endif /* MATCH_MAY_ALLOCATE */
3884 /* The starting position is bogus. */
3885 if (pos < 0 || pos > size1 + size2)
3891 /* Initialize subexpression text positions to -1 to mark ones that no
3892 start_memory/stop_memory has been seen for. Also initialize the
3893 register information struct. */
3894 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
3896 regstart[mcnt] = regend[mcnt]
3897 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3899 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3900 IS_ACTIVE (reg_info[mcnt]) = 0;
3901 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3902 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3905 /* We move `string1' into `string2' if the latter's empty -- but not if
3906 `string1' is null. */
3907 if (size2 == 0 && string1 != NULL)
3914 end1 = string1 + size1;
3915 end2 = string2 + size2;
3917 /* Compute where to stop matching, within the two strings. */
3920 end_match_1 = string1 + stop;
3921 end_match_2 = string2;
3926 end_match_2 = string2 + stop - size1;
3929 /* `p' scans through the pattern as `d' scans through the data.
3930 `dend' is the end of the input string that `d' points within. `d'
3931 is advanced into the following input string whenever necessary, but
3932 this happens before fetching; therefore, at the beginning of the
3933 loop, `d' can be pointing at the end of a string, but it cannot
3935 if (size1 > 0 && pos <= size1)
3942 d = string2 + pos - size1;
3946 DEBUG_PRINT1 ("The compiled pattern is:\n");
3947 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
3948 DEBUG_PRINT1 ("The string to match is: `");
3949 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
3950 DEBUG_PRINT1 ("'\n");
3952 /* This loops over pattern commands. It exits by returning from the
3953 function if the match is complete, or it drops through if the match
3954 fails at this starting point in the input data. */
3958 DEBUG_PRINT2 ("\n%p: ", p);
3960 DEBUG_PRINT2 ("\n0x%x: ", p);
3964 { /* End of pattern means we might have succeeded. */
3965 DEBUG_PRINT1 ("end of pattern ... ");
3967 /* If we haven't matched the entire string, and we want the
3968 longest match, try backtracking. */
3969 if (d != end_match_2)
3971 /* 1 if this match ends in the same string (string1 or string2)
3972 as the best previous match. */
3973 boolean same_str_p = (FIRST_STRING_P (match_end)
3974 == MATCHING_IN_FIRST_STRING);
3975 /* 1 if this match is the best seen so far. */
3976 boolean best_match_p;
3978 /* AIX compiler got confused when this was combined
3979 with the previous declaration. */
3981 best_match_p = d > match_end;
3983 best_match_p = !MATCHING_IN_FIRST_STRING;
3985 DEBUG_PRINT1 ("backtracking.\n");
3987 if (!FAIL_STACK_EMPTY ())
3988 { /* More failure points to try. */
3990 /* If exceeds best match so far, save it. */
3991 if (!best_regs_set || best_match_p)
3993 best_regs_set = true;
3996 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
3998 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4000 best_regstart[mcnt] = regstart[mcnt];
4001 best_regend[mcnt] = regend[mcnt];
4007 /* If no failure points, don't restore garbage. And if
4008 last match is real best match, don't restore second
4010 else if (best_regs_set && !best_match_p)
4013 /* Restore best match. It may happen that `dend ==
4014 end_match_1' while the restored d is in string2.
4015 For example, the pattern `x.*y.*z' against the
4016 strings `x-' and `y-z-', if the two strings are
4017 not consecutive in memory. */
4018 DEBUG_PRINT1 ("Restoring best registers.\n");
4021 dend = ((d >= string1 && d <= end1)
4022 ? end_match_1 : end_match_2);
4024 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4026 regstart[mcnt] = best_regstart[mcnt];
4027 regend[mcnt] = best_regend[mcnt];
4030 } /* d != end_match_2 */
4033 DEBUG_PRINT1 ("Accepting match.\n");
4035 /* If caller wants register contents data back, do it. */
4036 if (regs && !bufp->no_sub)
4038 /* Have the register data arrays been allocated? */
4039 if (bufp->regs_allocated == REGS_UNALLOCATED)
4040 { /* No. So allocate them with malloc. We need one
4041 extra element beyond `num_regs' for the `-1' marker
4043 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
4044 regs->start = TALLOC (regs->num_regs, regoff_t);
4045 regs->end = TALLOC (regs->num_regs, regoff_t);
4046 if (regs->start == NULL || regs->end == NULL)
4051 bufp->regs_allocated = REGS_REALLOCATE;
4053 else if (bufp->regs_allocated == REGS_REALLOCATE)
4054 { /* Yes. If we need more elements than were already
4055 allocated, reallocate them. If we need fewer, just
4057 if (regs->num_regs < num_regs + 1)
4059 regs->num_regs = num_regs + 1;
4060 RETALLOC (regs->start, regs->num_regs, regoff_t);
4061 RETALLOC (regs->end, regs->num_regs, regoff_t);
4062 if (regs->start == NULL || regs->end == NULL)
4071 /* These braces fend off a "empty body in an else-statement"
4072 warning under GCC when assert expands to nothing. */
4073 assert (bufp->regs_allocated == REGS_FIXED);
4076 /* Convert the pointer data in `regstart' and `regend' to
4077 indices. Register zero has to be set differently,
4078 since we haven't kept track of any info for it. */
4079 if (regs->num_regs > 0)
4081 regs->start[0] = pos;
4082 regs->end[0] = (MATCHING_IN_FIRST_STRING
4083 ? ((regoff_t) (d - string1))
4084 : ((regoff_t) (d - string2 + size1)));
4087 /* Go through the first `min (num_regs, regs->num_regs)'
4088 registers, since that is all we initialized. */
4089 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
4092 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
4093 regs->start[mcnt] = regs->end[mcnt] = -1;
4097 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
4099 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
4103 /* If the regs structure we return has more elements than
4104 were in the pattern, set the extra elements to -1. If
4105 we (re)allocated the registers, this is the case,
4106 because we always allocate enough to have at least one
4108 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
4109 regs->start[mcnt] = regs->end[mcnt] = -1;
4110 } /* regs && !bufp->no_sub */
4112 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
4113 nfailure_points_pushed, nfailure_points_popped,
4114 nfailure_points_pushed - nfailure_points_popped);
4115 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
4117 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
4121 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
4127 /* Otherwise match next pattern command. */
4128 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4130 /* Ignore these. Used to ignore the n of succeed_n's which
4131 currently have n == 0. */
4133 DEBUG_PRINT1 ("EXECUTING no_op.\n");
4137 DEBUG_PRINT1 ("EXECUTING succeed.\n");
4140 /* Match the next n pattern characters exactly. The following
4141 byte in the pattern defines n, and the n bytes after that
4142 are the characters to match. */
4145 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
4147 /* This is written out as an if-else so we don't waste time
4148 testing `translate' inside the loop. */
4154 if ((unsigned char) translate[(unsigned char) *d++]
4155 != (unsigned char) *p++)
4165 if (*d++ != (char) *p++) goto fail;
4169 SET_REGS_MATCHED ();
4173 /* Match any character except possibly a newline or a null. */
4175 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4179 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
4180 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4183 SET_REGS_MATCHED ();
4184 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4192 register unsigned char c;
4193 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4195 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4198 c = TRANSLATE (*d); /* The character to match. */
4200 /* Cast to `unsigned' instead of `unsigned char' in case the
4201 bit list is a full 32 bytes long. */
4202 if (c < (unsigned) (*p * BYTEWIDTH)
4203 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4208 if (!not) goto fail;
4210 SET_REGS_MATCHED ();
4216 /* The beginning of a group is represented by start_memory.
4217 The arguments are the register number in the next byte, and the
4218 number of groups inner to this one in the next. The text
4219 matched within the group is recorded (in the internal
4220 registers data structure) under the register number. */
4222 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4224 /* Find out if this group can match the empty string. */
4225 p1 = p; /* To send to group_match_null_string_p. */
4227 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4228 REG_MATCH_NULL_STRING_P (reg_info[*p])
4229 = group_match_null_string_p (&p1, pend, reg_info);
4231 /* Save the position in the string where we were the last time
4232 we were at this open-group operator in case the group is
4233 operated upon by a repetition operator, e.g., with `(a*)*b'
4234 against `ab'; then we want to ignore where we are now in
4235 the string in case this attempt to match fails. */
4236 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4237 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4239 DEBUG_PRINT2 (" old_regstart: %d\n",
4240 POINTER_TO_OFFSET (old_regstart[*p]));
4243 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4245 IS_ACTIVE (reg_info[*p]) = 1;
4246 MATCHED_SOMETHING (reg_info[*p]) = 0;
4248 /* Clear this whenever we change the register activity status. */
4249 set_regs_matched_done = 0;
4251 /* This is the new highest active register. */
4252 highest_active_reg = *p;
4254 /* If nothing was active before, this is the new lowest active
4256 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4257 lowest_active_reg = *p;
4259 /* Move past the register number and inner group count. */
4261 just_past_start_mem = p;
4266 /* The stop_memory opcode represents the end of a group. Its
4267 arguments are the same as start_memory's: the register
4268 number, and the number of inner groups. */
4270 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4272 /* We need to save the string position the last time we were at
4273 this close-group operator in case the group is operated
4274 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4275 against `aba'; then we want to ignore where we are now in
4276 the string in case this attempt to match fails. */
4277 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4278 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4280 DEBUG_PRINT2 (" old_regend: %d\n",
4281 POINTER_TO_OFFSET (old_regend[*p]));
4284 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4286 /* This register isn't active anymore. */
4287 IS_ACTIVE (reg_info[*p]) = 0;
4289 /* Clear this whenever we change the register activity status. */
4290 set_regs_matched_done = 0;
4292 /* If this was the only register active, nothing is active
4294 if (lowest_active_reg == highest_active_reg)
4296 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4297 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4300 { /* We must scan for the new highest active register, since
4301 it isn't necessarily one less than now: consider
4302 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4303 new highest active register is 1. */
4304 unsigned char r = *p - 1;
4305 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4308 /* If we end up at register zero, that means that we saved
4309 the registers as the result of an `on_failure_jump', not
4310 a `start_memory', and we jumped to past the innermost
4311 `stop_memory'. For example, in ((.)*) we save
4312 registers 1 and 2 as a result of the *, but when we pop
4313 back to the second ), we are at the stop_memory 1.
4314 Thus, nothing is active. */
4317 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4318 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4321 highest_active_reg = r;
4324 /* If just failed to match something this time around with a
4325 group that's operated on by a repetition operator, try to
4326 force exit from the ``loop'', and restore the register
4327 information for this group that we had before trying this
4329 if ((!MATCHED_SOMETHING (reg_info[*p])
4330 || just_past_start_mem == p - 1)
4333 boolean is_a_jump_n = false;
4337 switch ((re_opcode_t) *p1++)
4341 case pop_failure_jump:
4342 case maybe_pop_jump:
4344 case dummy_failure_jump:
4345 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4355 /* If the next operation is a jump backwards in the pattern
4356 to an on_failure_jump right before the start_memory
4357 corresponding to this stop_memory, exit from the loop
4358 by forcing a failure after pushing on the stack the
4359 on_failure_jump's jump in the pattern, and d. */
4360 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4361 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4363 /* If this group ever matched anything, then restore
4364 what its registers were before trying this last
4365 failed match, e.g., with `(a*)*b' against `ab' for
4366 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4367 against `aba' for regend[3].
4369 Also restore the registers for inner groups for,
4370 e.g., `((a*)(b*))*' against `aba' (register 3 would
4371 otherwise get trashed). */
4373 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4377 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4379 /* Restore this and inner groups' (if any) registers. */
4380 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
4383 regstart[r] = old_regstart[r];
4385 /* xx why this test? */
4386 if (old_regend[r] >= regstart[r])
4387 regend[r] = old_regend[r];
4391 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4392 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4398 /* Move past the register number and the inner group count. */
4403 /* \<digit> has been turned into a `duplicate' command which is
4404 followed by the numeric value of <digit> as the register number. */
4407 register const char *d2, *dend2;
4408 int regno = *p++; /* Get which register to match against. */
4409 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4411 /* Can't back reference a group which we've never matched. */
4412 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4415 /* Where in input to try to start matching. */
4416 d2 = regstart[regno];
4418 /* Where to stop matching; if both the place to start and
4419 the place to stop matching are in the same string, then
4420 set to the place to stop, otherwise, for now have to use
4421 the end of the first string. */
4423 dend2 = ((FIRST_STRING_P (regstart[regno])
4424 == FIRST_STRING_P (regend[regno]))
4425 ? regend[regno] : end_match_1);
4428 /* If necessary, advance to next segment in register
4432 if (dend2 == end_match_2) break;
4433 if (dend2 == regend[regno]) break;
4435 /* End of string1 => advance to string2. */
4437 dend2 = regend[regno];
4439 /* At end of register contents => success */
4440 if (d2 == dend2) break;
4442 /* If necessary, advance to next segment in data. */
4445 /* How many characters left in this segment to match. */
4448 /* Want how many consecutive characters we can match in
4449 one shot, so, if necessary, adjust the count. */
4450 if (mcnt > dend2 - d2)
4453 /* Compare that many; failure if mismatch, else move
4456 ? bcmp_translate (d, d2, mcnt, translate)
4457 : bcmp (d, d2, mcnt))
4459 d += mcnt, d2 += mcnt;
4461 /* Do this because we've match some characters. */
4462 SET_REGS_MATCHED ();
4468 /* begline matches the empty string at the beginning of the string
4469 (unless `not_bol' is set in `bufp'), and, if
4470 `newline_anchor' is set, after newlines. */
4472 DEBUG_PRINT1 ("EXECUTING begline.\n");
4474 if (AT_STRINGS_BEG (d))
4476 if (!bufp->not_bol) break;
4478 else if (d[-1] == '\n' && bufp->newline_anchor)
4482 /* In all other cases, we fail. */
4486 /* endline is the dual of begline. */
4488 DEBUG_PRINT1 ("EXECUTING endline.\n");
4490 if (AT_STRINGS_END (d))
4492 if (!bufp->not_eol) break;
4495 /* We have to ``prefetch'' the next character. */
4496 else if ((d == end1 ? *string2 : *d) == '\n'
4497 && bufp->newline_anchor)
4504 /* Match at the very beginning of the data. */
4506 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4507 if (AT_STRINGS_BEG (d))
4512 /* Match at the very end of the data. */
4514 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4515 if (AT_STRINGS_END (d))
4520 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4521 pushes NULL as the value for the string on the stack. Then
4522 `pop_failure_point' will keep the current value for the
4523 string, instead of restoring it. To see why, consider
4524 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4525 then the . fails against the \n. But the next thing we want
4526 to do is match the \n against the \n; if we restored the
4527 string value, we would be back at the foo.
4529 Because this is used only in specific cases, we don't need to
4530 check all the things that `on_failure_jump' does, to make
4531 sure the right things get saved on the stack. Hence we don't
4532 share its code. The only reason to push anything on the
4533 stack at all is that otherwise we would have to change
4534 `anychar's code to do something besides goto fail in this
4535 case; that seems worse than this. */
4536 case on_failure_keep_string_jump:
4537 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4539 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4541 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
4543 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4546 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4550 /* Uses of on_failure_jump:
4552 Each alternative starts with an on_failure_jump that points
4553 to the beginning of the next alternative. Each alternative
4554 except the last ends with a jump that in effect jumps past
4555 the rest of the alternatives. (They really jump to the
4556 ending jump of the following alternative, because tensioning
4557 these jumps is a hassle.)
4559 Repeats start with an on_failure_jump that points past both
4560 the repetition text and either the following jump or
4561 pop_failure_jump back to this on_failure_jump. */
4562 case on_failure_jump:
4564 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4566 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4568 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
4570 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4573 /* If this on_failure_jump comes right before a group (i.e.,
4574 the original * applied to a group), save the information
4575 for that group and all inner ones, so that if we fail back
4576 to this point, the group's information will be correct.
4577 For example, in \(a*\)*\1, we need the preceding group,
4578 and in \(zz\(a*\)b*\)\2, we need the inner group. */
4580 /* We can't use `p' to check ahead because we push
4581 a failure point to `p + mcnt' after we do this. */
4584 /* We need to skip no_op's before we look for the
4585 start_memory in case this on_failure_jump is happening as
4586 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4588 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4591 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4593 /* We have a new highest active register now. This will
4594 get reset at the start_memory we are about to get to,
4595 but we will have saved all the registers relevant to
4596 this repetition op, as described above. */
4597 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4598 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4599 lowest_active_reg = *(p1 + 1);
4602 DEBUG_PRINT1 (":\n");
4603 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4607 /* A smart repeat ends with `maybe_pop_jump'.
4608 We change it to either `pop_failure_jump' or `jump'. */
4609 case maybe_pop_jump:
4610 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4611 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4613 register unsigned char *p2 = p;
4615 /* Compare the beginning of the repeat with what in the
4616 pattern follows its end. If we can establish that there
4617 is nothing that they would both match, i.e., that we
4618 would have to backtrack because of (as in, e.g., `a*a')
4619 then we can change to pop_failure_jump, because we'll
4620 never have to backtrack.
4622 This is not true in the case of alternatives: in
4623 `(a|ab)*' we do need to backtrack to the `ab' alternative
4624 (e.g., if the string was `ab'). But instead of trying to
4625 detect that here, the alternative has put on a dummy
4626 failure point which is what we will end up popping. */
4628 /* Skip over open/close-group commands.
4629 If what follows this loop is a ...+ construct,
4630 look at what begins its body, since we will have to
4631 match at least one of that. */
4635 && ((re_opcode_t) *p2 == stop_memory
4636 || (re_opcode_t) *p2 == start_memory))
4638 else if (p2 + 6 < pend
4639 && (re_opcode_t) *p2 == dummy_failure_jump)
4646 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4647 to the `maybe_finalize_jump' of this case. Examine what
4650 /* If we're at the end of the pattern, we can change. */
4653 /* Consider what happens when matching ":\(.*\)"
4654 against ":/". I don't really understand this code
4656 p[-3] = (unsigned char) pop_failure_jump;
4658 (" End of pattern: change to `pop_failure_jump'.\n");
4661 else if ((re_opcode_t) *p2 == exactn
4662 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4664 register unsigned char c
4665 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4667 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4669 p[-3] = (unsigned char) pop_failure_jump;
4670 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4674 else if ((re_opcode_t) p1[3] == charset
4675 || (re_opcode_t) p1[3] == charset_not)
4677 int not = (re_opcode_t) p1[3] == charset_not;
4679 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4680 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4683 /* `not' is equal to 1 if c would match, which means
4684 that we can't change to pop_failure_jump. */
4687 p[-3] = (unsigned char) pop_failure_jump;
4688 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4692 else if ((re_opcode_t) *p2 == charset)
4695 register unsigned char c
4696 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4700 if ((re_opcode_t) p1[3] == exactn
4701 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
4702 && (p2[2 + p1[5] / BYTEWIDTH]
4703 & (1 << (p1[5] % BYTEWIDTH)))))
4705 if ((re_opcode_t) p1[3] == exactn
4706 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[4]
4707 && (p2[2 + p1[4] / BYTEWIDTH]
4708 & (1 << (p1[4] % BYTEWIDTH)))))
4711 p[-3] = (unsigned char) pop_failure_jump;
4712 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4716 else if ((re_opcode_t) p1[3] == charset_not)
4719 /* We win if the charset_not inside the loop
4720 lists every character listed in the charset after. */
4721 for (idx = 0; idx < (int) p2[1]; idx++)
4722 if (! (p2[2 + idx] == 0
4723 || (idx < (int) p1[4]
4724 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4729 p[-3] = (unsigned char) pop_failure_jump;
4730 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4733 else if ((re_opcode_t) p1[3] == charset)
4736 /* We win if the charset inside the loop
4737 has no overlap with the one after the loop. */
4739 idx < (int) p2[1] && idx < (int) p1[4];
4741 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4744 if (idx == p2[1] || idx == p1[4])
4746 p[-3] = (unsigned char) pop_failure_jump;
4747 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4752 p -= 2; /* Point at relative address again. */
4753 if ((re_opcode_t) p[-1] != pop_failure_jump)
4755 p[-1] = (unsigned char) jump;
4756 DEBUG_PRINT1 (" Match => jump.\n");
4757 goto unconditional_jump;
4759 /* Note fall through. */
4762 /* The end of a simple repeat has a pop_failure_jump back to
4763 its matching on_failure_jump, where the latter will push a
4764 failure point. The pop_failure_jump takes off failure
4765 points put on by this pop_failure_jump's matching
4766 on_failure_jump; we got through the pattern to here from the
4767 matching on_failure_jump, so didn't fail. */
4768 case pop_failure_jump:
4770 /* We need to pass separate storage for the lowest and
4771 highest registers, even though we don't care about the
4772 actual values. Otherwise, we will restore only one
4773 register from the stack, since lowest will == highest in
4774 `pop_failure_point'. */
4775 active_reg_t dummy_low_reg, dummy_high_reg;
4776 unsigned char *pdummy;
4779 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4780 POP_FAILURE_POINT (sdummy, pdummy,
4781 dummy_low_reg, dummy_high_reg,
4782 reg_dummy, reg_dummy, reg_info_dummy);
4784 /* Note fall through. */
4788 DEBUG_PRINT2 ("\n%p: ", p);
4790 DEBUG_PRINT2 ("\n0x%x: ", p);
4792 /* Note fall through. */
4794 /* Unconditionally jump (without popping any failure points). */
4796 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4797 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4798 p += mcnt; /* Do the jump. */
4800 DEBUG_PRINT2 ("(to %p).\n", p);
4802 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4807 /* We need this opcode so we can detect where alternatives end
4808 in `group_match_null_string_p' et al. */
4810 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4811 goto unconditional_jump;
4814 /* Normally, the on_failure_jump pushes a failure point, which
4815 then gets popped at pop_failure_jump. We will end up at
4816 pop_failure_jump, also, and with a pattern of, say, `a+', we
4817 are skipping over the on_failure_jump, so we have to push
4818 something meaningless for pop_failure_jump to pop. */
4819 case dummy_failure_jump:
4820 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4821 /* It doesn't matter what we push for the string here. What
4822 the code at `fail' tests is the value for the pattern. */
4823 PUSH_FAILURE_POINT (0, 0, -2);
4824 goto unconditional_jump;
4827 /* At the end of an alternative, we need to push a dummy failure
4828 point in case we are followed by a `pop_failure_jump', because
4829 we don't want the failure point for the alternative to be
4830 popped. For example, matching `(a|ab)*' against `aab'
4831 requires that we match the `ab' alternative. */
4832 case push_dummy_failure:
4833 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4834 /* See comments just above at `dummy_failure_jump' about the
4836 PUSH_FAILURE_POINT (0, 0, -2);
4839 /* Have to succeed matching what follows at least n times.
4840 After that, handle like `on_failure_jump'. */
4842 EXTRACT_NUMBER (mcnt, p + 2);
4843 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4846 /* Originally, this is how many times we HAVE to succeed. */
4851 STORE_NUMBER_AND_INCR (p, mcnt);
4853 DEBUG_PRINT3 (" Setting %p to %d.\n", p - 2, mcnt);
4855 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - 2, mcnt);
4861 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n", p+2);
4863 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4865 p[2] = (unsigned char) no_op;
4866 p[3] = (unsigned char) no_op;
4872 EXTRACT_NUMBER (mcnt, p + 2);
4873 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4875 /* Originally, this is how many times we CAN jump. */
4879 STORE_NUMBER (p + 2, mcnt);
4881 DEBUG_PRINT3 (" Setting %p to %d.\n", p + 2, mcnt);
4883 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + 2, mcnt);
4885 goto unconditional_jump;
4887 /* If don't have to jump any more, skip over the rest of command. */
4894 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4896 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4898 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4900 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
4902 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4904 STORE_NUMBER (p1, mcnt);
4909 /* The DEC Alpha C compiler 3.x generates incorrect code for the
4910 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4911 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4912 macro and introducing temporary variables works around the bug. */
4915 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4916 if (AT_WORD_BOUNDARY (d))
4921 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4922 if (AT_WORD_BOUNDARY (d))
4928 boolean prevchar, thischar;
4930 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4931 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4934 prevchar = WORDCHAR_P (d - 1);
4935 thischar = WORDCHAR_P (d);
4936 if (prevchar != thischar)
4943 boolean prevchar, thischar;
4945 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4946 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4949 prevchar = WORDCHAR_P (d - 1);
4950 thischar = WORDCHAR_P (d);
4951 if (prevchar != thischar)
4958 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
4959 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
4964 DEBUG_PRINT1 ("EXECUTING wordend.\n");
4965 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
4966 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
4972 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
4973 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
4978 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
4979 if (PTR_CHAR_POS ((unsigned char *) d) != point)
4984 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
4985 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
4990 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
4995 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
4999 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5001 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
5003 SET_REGS_MATCHED ();
5007 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
5009 goto matchnotsyntax;
5012 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
5016 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5018 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
5020 SET_REGS_MATCHED ();
5023 #else /* not emacs */
5025 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
5027 if (!WORDCHAR_P (d))
5029 SET_REGS_MATCHED ();
5034 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
5038 SET_REGS_MATCHED ();
5041 #endif /* not emacs */
5046 continue; /* Successfully executed one pattern command; keep going. */
5049 /* We goto here if a matching operation fails. */
5051 if (!FAIL_STACK_EMPTY ())
5052 { /* A restart point is known. Restore to that state. */
5053 DEBUG_PRINT1 ("\nFAIL:\n");
5054 POP_FAILURE_POINT (d, p,
5055 lowest_active_reg, highest_active_reg,
5056 regstart, regend, reg_info);
5058 /* If this failure point is a dummy, try the next one. */
5062 /* If we failed to the end of the pattern, don't examine *p. */
5066 boolean is_a_jump_n = false;
5068 /* If failed to a backwards jump that's part of a repetition
5069 loop, need to pop this failure point and use the next one. */
5070 switch ((re_opcode_t) *p)
5074 case maybe_pop_jump:
5075 case pop_failure_jump:
5078 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5081 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
5083 && (re_opcode_t) *p1 == on_failure_jump))
5091 if (d >= string1 && d <= end1)
5095 break; /* Matching at this starting point really fails. */
5099 goto restore_best_regs;
5103 return -1; /* Failure to match. */
5106 /* Subroutine definitions for re_match_2. */
5109 /* We are passed P pointing to a register number after a start_memory.
5111 Return true if the pattern up to the corresponding stop_memory can
5112 match the empty string, and false otherwise.
5114 If we find the matching stop_memory, sets P to point to one past its number.
5115 Otherwise, sets P to an undefined byte less than or equal to END.
5117 We don't handle duplicates properly (yet). */
5120 group_match_null_string_p (p, end, reg_info)
5121 unsigned char **p, *end;
5122 register_info_type *reg_info;
5125 /* Point to after the args to the start_memory. */
5126 unsigned char *p1 = *p + 2;
5130 /* Skip over opcodes that can match nothing, and return true or
5131 false, as appropriate, when we get to one that can't, or to the
5132 matching stop_memory. */
5134 switch ((re_opcode_t) *p1)
5136 /* Could be either a loop or a series of alternatives. */
5137 case on_failure_jump:
5139 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5141 /* If the next operation is not a jump backwards in the
5146 /* Go through the on_failure_jumps of the alternatives,
5147 seeing if any of the alternatives cannot match nothing.
5148 The last alternative starts with only a jump,
5149 whereas the rest start with on_failure_jump and end
5150 with a jump, e.g., here is the pattern for `a|b|c':
5152 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
5153 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
5156 So, we have to first go through the first (n-1)
5157 alternatives and then deal with the last one separately. */
5160 /* Deal with the first (n-1) alternatives, which start
5161 with an on_failure_jump (see above) that jumps to right
5162 past a jump_past_alt. */
5164 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
5166 /* `mcnt' holds how many bytes long the alternative
5167 is, including the ending `jump_past_alt' and
5170 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
5174 /* Move to right after this alternative, including the
5178 /* Break if it's the beginning of an n-th alternative
5179 that doesn't begin with an on_failure_jump. */
5180 if ((re_opcode_t) *p1 != on_failure_jump)
5183 /* Still have to check that it's not an n-th
5184 alternative that starts with an on_failure_jump. */
5186 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5187 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
5189 /* Get to the beginning of the n-th alternative. */
5195 /* Deal with the last alternative: go back and get number
5196 of the `jump_past_alt' just before it. `mcnt' contains
5197 the length of the alternative. */
5198 EXTRACT_NUMBER (mcnt, p1 - 2);
5200 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
5203 p1 += mcnt; /* Get past the n-th alternative. */
5209 assert (p1[1] == **p);
5215 if (!common_op_match_null_string_p (&p1, end, reg_info))
5218 } /* while p1 < end */
5221 } /* group_match_null_string_p */
5224 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5225 It expects P to be the first byte of a single alternative and END one
5226 byte past the last. The alternative can contain groups. */
5229 alt_match_null_string_p (p, end, reg_info)
5230 unsigned char *p, *end;
5231 register_info_type *reg_info;
5234 unsigned char *p1 = p;
5238 /* Skip over opcodes that can match nothing, and break when we get
5239 to one that can't. */
5241 switch ((re_opcode_t) *p1)
5244 case on_failure_jump:
5246 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5251 if (!common_op_match_null_string_p (&p1, end, reg_info))
5254 } /* while p1 < end */
5257 } /* alt_match_null_string_p */
5260 /* Deals with the ops common to group_match_null_string_p and
5261 alt_match_null_string_p.
5263 Sets P to one after the op and its arguments, if any. */
5266 common_op_match_null_string_p (p, end, reg_info)
5267 unsigned char **p, *end;
5268 register_info_type *reg_info;
5273 unsigned char *p1 = *p;
5275 switch ((re_opcode_t) *p1++)
5295 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5296 ret = group_match_null_string_p (&p1, end, reg_info);
5298 /* Have to set this here in case we're checking a group which
5299 contains a group and a back reference to it. */
5301 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5302 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5308 /* If this is an optimized succeed_n for zero times, make the jump. */
5310 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5318 /* Get to the number of times to succeed. */
5320 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5325 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5333 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5341 /* All other opcodes mean we cannot match the empty string. */
5347 } /* common_op_match_null_string_p */
5350 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5351 bytes; nonzero otherwise. */
5354 bcmp_translate (s1, s2, len, translate)
5355 const char *s1, *s2;
5357 RE_TRANSLATE_TYPE translate;
5359 register const unsigned char *p1 = (const unsigned char *) s1;
5360 register const unsigned char *p2 = (const unsigned char *) s2;
5363 if (translate[*p1++] != translate[*p2++]) return 1;
5369 /* Entry points for GNU code. */
5371 /* re_compile_pattern is the GNU regular expression compiler: it
5372 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5373 Returns 0 if the pattern was valid, otherwise an error string.
5375 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5376 are set in BUFP on entry.
5378 We call regex_compile to do the actual compilation. */
5381 re_compile_pattern (pattern, length, bufp)
5382 const char *pattern;
5384 struct re_pattern_buffer *bufp;
5388 /* GNU code is written to assume at least RE_NREGS registers will be set
5389 (and at least one extra will be -1). */
5390 bufp->regs_allocated = REGS_UNALLOCATED;
5392 /* And GNU code determines whether or not to get register information
5393 by passing null for the REGS argument to re_match, etc., not by
5397 /* Match anchors at newline. */
5398 bufp->newline_anchor = 1;
5400 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5404 return gettext (re_error_msgid[(int) ret]);
5407 /* Entry points compatible with 4.2 BSD regex library. We don't define
5408 them unless specifically requested. */
5410 #if defined (_REGEX_RE_COMP) || defined (_LIBC)
5412 /* BSD has one and only one pattern buffer. */
5413 static struct re_pattern_buffer re_comp_buf;
5417 /* Make these definitions weak in libc, so POSIX programs can redefine
5418 these names if they don't use our functions, and still use
5419 regcomp/regexec below without link errors. */
5429 if (!re_comp_buf.buffer)
5430 return gettext ("No previous regular expression");
5434 if (!re_comp_buf.buffer)
5436 re_comp_buf.buffer = (unsigned char *) malloc (200);
5437 if (re_comp_buf.buffer == NULL)
5438 return gettext (re_error_msgid[(int) REG_ESPACE]);
5439 re_comp_buf.allocated = 200;
5441 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5442 if (re_comp_buf.fastmap == NULL)
5443 return gettext (re_error_msgid[(int) REG_ESPACE]);
5446 /* Since `re_exec' always passes NULL for the `regs' argument, we
5447 don't need to initialize the pattern buffer fields which affect it. */
5449 /* Match anchors at newlines. */
5450 re_comp_buf.newline_anchor = 1;
5452 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5457 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5458 return (char *) gettext (re_error_msgid[(int) ret]);
5469 const int len = strlen (s);
5471 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5474 #endif /* _REGEX_RE_COMP */
5476 /* POSIX.2 functions. Don't define these for Emacs. */
5480 /* regcomp takes a regular expression as a string and compiles it.
5482 PREG is a regex_t *. We do not expect any fields to be initialized,
5483 since POSIX says we shouldn't. Thus, we set
5485 `buffer' to the compiled pattern;
5486 `used' to the length of the compiled pattern;
5487 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5488 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5489 RE_SYNTAX_POSIX_BASIC;
5490 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5491 `fastmap' and `fastmap_accurate' to zero;
5492 `re_nsub' to the number of subexpressions in PATTERN.
5494 PATTERN is the address of the pattern string.
5496 CFLAGS is a series of bits which affect compilation.
5498 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5499 use POSIX basic syntax.
5501 If REG_NEWLINE is set, then . and [^...] don't match newline.
5502 Also, regexec will try a match beginning after every newline.
5504 If REG_ICASE is set, then we considers upper- and lowercase
5505 versions of letters to be equivalent when matching.
5507 If REG_NOSUB is set, then when PREG is passed to regexec, that
5508 routine will report only success or failure, and nothing about the
5511 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
5512 the return codes and their meanings.) */
5515 regcomp (preg, pattern, cflags)
5517 const char *pattern;
5522 = (cflags & REG_EXTENDED) ?
5523 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5525 /* regex_compile will allocate the space for the compiled pattern. */
5527 preg->allocated = 0;
5530 /* Don't bother to use a fastmap when searching. This simplifies the
5531 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
5532 characters after newlines into the fastmap. This way, we just try
5536 if (cflags & REG_ICASE)
5541 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
5542 * sizeof (*(RE_TRANSLATE_TYPE)0));
5543 if (preg->translate == NULL)
5544 return (int) REG_ESPACE;
5546 /* Map uppercase characters to corresponding lowercase ones. */
5547 for (i = 0; i < CHAR_SET_SIZE; i++)
5548 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
5551 preg->translate = NULL;
5553 /* If REG_NEWLINE is set, newlines are treated differently. */
5554 if (cflags & REG_NEWLINE)
5555 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5556 syntax &= ~RE_DOT_NEWLINE;
5557 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5558 /* It also changes the matching behavior. */
5559 preg->newline_anchor = 1;
5562 preg->newline_anchor = 0;
5564 preg->no_sub = !!(cflags & REG_NOSUB);
5566 /* POSIX says a null character in the pattern terminates it, so we
5567 can use strlen here in compiling the pattern. */
5568 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5570 /* POSIX doesn't distinguish between an unmatched open-group and an
5571 unmatched close-group: both are REG_EPAREN. */
5572 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5578 /* regexec searches for a given pattern, specified by PREG, in the
5581 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5582 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5583 least NMATCH elements, and we set them to the offsets of the
5584 corresponding matched substrings.
5586 EFLAGS specifies `execution flags' which affect matching: if
5587 REG_NOTBOL is set, then ^ does not match at the beginning of the
5588 string; if REG_NOTEOL is set, then $ does not match at the end.
5590 We return 0 if we find a match and REG_NOMATCH if not. */
5593 regexec (preg, string, nmatch, pmatch, eflags)
5594 const regex_t *preg;
5597 regmatch_t pmatch[];
5601 struct re_registers regs;
5602 regex_t private_preg;
5603 int len = strlen (string);
5604 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5606 private_preg = *preg;
5608 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5609 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5611 /* The user has told us exactly how many registers to return
5612 information about, via `nmatch'. We have to pass that on to the
5613 matching routines. */
5614 private_preg.regs_allocated = REGS_FIXED;
5618 regs.num_regs = nmatch;
5619 regs.start = TALLOC (nmatch, regoff_t);
5620 regs.end = TALLOC (nmatch, regoff_t);
5621 if (regs.start == NULL || regs.end == NULL)
5622 return (int) REG_NOMATCH;
5625 /* Perform the searching operation. */
5626 ret = re_search (&private_preg, string, len,
5627 /* start: */ 0, /* range: */ len,
5628 want_reg_info ? ®s : (struct re_registers *) 0);
5630 /* Copy the register information to the POSIX structure. */
5637 for (r = 0; r < nmatch; r++)
5639 pmatch[r].rm_so = regs.start[r];
5640 pmatch[r].rm_eo = regs.end[r];
5644 /* If we needed the temporary register info, free the space now. */
5649 /* We want zero return to mean success, unlike `re_search'. */
5650 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5654 /* Returns a message corresponding to an error code, ERRCODE, returned
5655 from either regcomp or regexec. We don't use PREG here. */
5658 regerror (errcode, preg, errbuf, errbuf_size)
5660 const regex_t *preg;
5668 || errcode >= (int) (sizeof (re_error_msgid)
5669 / sizeof (re_error_msgid[0])))
5670 /* Only error codes returned by the rest of the code should be passed
5671 to this routine. If we are given anything else, or if other regex
5672 code generates an invalid error code, then the program has a bug.
5673 Dump core so we can fix it. */
5676 msg = gettext (re_error_msgid[errcode]);
5678 msg_size = strlen (msg) + 1; /* Includes the null. */
5680 if (errbuf_size != 0)
5682 if (msg_size > errbuf_size)
5684 strncpy (errbuf, msg, errbuf_size - 1);
5685 errbuf[errbuf_size - 1] = 0;
5688 strcpy (errbuf, msg);
5695 /* Free dynamically allocated space used by PREG. */
5701 if (preg->buffer != NULL)
5702 free (preg->buffer);
5703 preg->buffer = NULL;
5705 preg->allocated = 0;
5708 if (preg->fastmap != NULL)
5709 free (preg->fastmap);
5710 preg->fastmap = NULL;
5711 preg->fastmap_accurate = 0;
5713 if (preg->translate != NULL)
5714 free (preg->translate);
5715 preg->translate = NULL;
5718 #endif /* not emacs */
projects / ghc-hetmet.git / blob