1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P10003.2/D11.2, except for
4 internationalization features.)
6 Copyright (C) 1993, 1994, 1995, 1996 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
20 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
23 /* AIX requires this to be the first thing in the file. */
24 #if defined (_AIX) && !defined (REGEX_MALLOC)
35 /* We need this for `regex.h', and perhaps for the Emacs include files. */
36 #include <sys/types.h>
38 /* This is for other GNU distributions with internationalized messages. */
39 #if HAVE_LIBINTL_H || defined (_LIBC)
42 # define gettext(msgid) (msgid)
46 /* This define is so xgettext can find the internationalizable
48 #define gettext_noop(String) String
51 /* The `emacs' switch turns on certain matching commands
52 that make sense only in Emacs. */
61 /* If we are not linking with Emacs proper,
62 we can't use the relocating allocator
63 even if config.h says that we can. */
66 #if defined (STDC_HEADERS) || defined (_LIBC)
73 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
74 If nothing else has been done, use the method below. */
75 #ifdef INHIBIT_STRING_HEADER
76 #if !(defined (HAVE_BZERO) && defined (HAVE_BCOPY))
77 #if !defined (bzero) && !defined (bcopy)
78 #undef INHIBIT_STRING_HEADER
83 /* This is the normal way of making sure we have a bcopy and a bzero.
84 This is used in most programs--a few other programs avoid this
85 by defining INHIBIT_STRING_HEADER. */
86 #ifndef INHIBIT_STRING_HEADER
87 #if defined (HAVE_STRING_H) || defined (STDC_HEADERS) || defined (_LIBC)
90 #define bcmp(s1, s2, n) memcmp ((s1), (s2), (n))
93 #define bcopy(s, d, n) memcpy ((d), (s), (n))
96 #define bzero(s, n) memset ((s), 0, (n))
103 /* Define the syntax stuff for \<, \>, etc. */
105 /* This must be nonzero for the wordchar and notwordchar pattern
106 commands in re_match_2. */
111 #ifdef SWITCH_ENUM_BUG
112 #define SWITCH_ENUM_CAST(x) ((int)(x))
114 #define SWITCH_ENUM_CAST(x) (x)
119 extern char *re_syntax_table;
121 #else /* not SYNTAX_TABLE */
123 /* How many characters in the character set. */
124 #define CHAR_SET_SIZE 256
126 static char re_syntax_table[CHAR_SET_SIZE];
137 bzero (re_syntax_table, sizeof re_syntax_table);
139 for (c = 'a'; c <= 'z'; c++)
140 re_syntax_table[c] = Sword;
142 for (c = 'A'; c <= 'Z'; c++)
143 re_syntax_table[c] = Sword;
145 for (c = '0'; c <= '9'; c++)
146 re_syntax_table[c] = Sword;
148 re_syntax_table['_'] = Sword;
153 #endif /* not SYNTAX_TABLE */
155 #define SYNTAX(c) re_syntax_table[c]
157 #endif /* not emacs */
159 /* Get the interface, including the syntax bits. */
162 /* isalpha etc. are used for the character classes. */
165 /* Jim Meyering writes:
167 "... Some ctype macros are valid only for character codes that
168 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
169 using /bin/cc or gcc but without giving an ansi option). So, all
170 ctype uses should be through macros like ISPRINT... If
171 STDC_HEADERS is defined, then autoconf has verified that the ctype
172 macros don't need to be guarded with references to isascii. ...
173 Defining isascii to 1 should let any compiler worth its salt
174 eliminate the && through constant folding." */
176 #if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII))
179 #define ISASCII(c) isascii(c)
183 #define ISBLANK(c) (ISASCII (c) && isblank (c))
185 #define ISBLANK(c) ((c) == ' ' || (c) == '\t')
188 #define ISGRAPH(c) (ISASCII (c) && isgraph (c))
190 #define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
193 #define ISPRINT(c) (ISASCII (c) && isprint (c))
194 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
195 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
196 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
197 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
198 #define ISLOWER(c) (ISASCII (c) && islower (c))
199 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
200 #define ISSPACE(c) (ISASCII (c) && isspace (c))
201 #define ISUPPER(c) (ISASCII (c) && isupper (c))
202 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
205 #define NULL (void *)0
208 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
209 since ours (we hope) works properly with all combinations of
210 machines, compilers, `char' and `unsigned char' argument types.
211 (Per Bothner suggested the basic approach.) */
212 #undef SIGN_EXTEND_CHAR
214 #define SIGN_EXTEND_CHAR(c) ((signed char) (c))
215 #else /* not __STDC__ */
216 /* As in Harbison and Steele. */
217 #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
220 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
221 use `alloca' instead of `malloc'. This is because using malloc in
222 re_search* or re_match* could cause memory leaks when C-g is used in
223 Emacs; also, malloc is slower and causes storage fragmentation. On
224 the other hand, malloc is more portable, and easier to debug.
226 Because we sometimes use alloca, some routines have to be macros,
227 not functions -- `alloca'-allocated space disappears at the end of the
228 function it is called in. */
232 #define REGEX_ALLOCATE malloc
233 #define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
234 #define REGEX_FREE free
236 #else /* not REGEX_MALLOC */
238 /* Emacs already defines alloca, sometimes. */
241 /* Make alloca work the best possible way. */
243 #define alloca __builtin_alloca
244 #else /* not __GNUC__ */
247 #else /* not __GNUC__ or HAVE_ALLOCA_H */
248 #if 0 /* It is a bad idea to declare alloca. We always cast the result. */
249 #ifndef _AIX /* Already did AIX, up at the top. */
251 #endif /* not _AIX */
253 #endif /* not HAVE_ALLOCA_H */
254 #endif /* not __GNUC__ */
256 #endif /* not alloca */
258 #define REGEX_ALLOCATE alloca
260 /* Assumes a `char *destination' variable. */
261 #define REGEX_REALLOCATE(source, osize, nsize) \
262 (destination = (char *) alloca (nsize), \
263 bcopy (source, destination, osize), \
266 /* No need to do anything to free, after alloca. */
267 #define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
269 #endif /* not REGEX_MALLOC */
271 /* Define how to allocate the failure stack. */
273 #if defined (REL_ALLOC) && defined (REGEX_MALLOC)
275 #define REGEX_ALLOCATE_STACK(size) \
276 r_alloc (&failure_stack_ptr, (size))
277 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
278 r_re_alloc (&failure_stack_ptr, (nsize))
279 #define REGEX_FREE_STACK(ptr) \
280 r_alloc_free (&failure_stack_ptr)
282 #else /* not using relocating allocator */
286 #define REGEX_ALLOCATE_STACK malloc
287 #define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
288 #define REGEX_FREE_STACK free
290 #else /* not REGEX_MALLOC */
292 #define REGEX_ALLOCATE_STACK alloca
294 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
295 REGEX_REALLOCATE (source, osize, nsize)
296 /* No need to explicitly free anything. */
297 #define REGEX_FREE_STACK(arg)
299 #endif /* not REGEX_MALLOC */
300 #endif /* not using relocating allocator */
303 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
304 `string1' or just past its end. This works if PTR is NULL, which is
306 #define FIRST_STRING_P(ptr) \
307 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
309 /* (Re)Allocate N items of type T using malloc, or fail. */
310 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
311 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
312 #define RETALLOC_IF(addr, n, t) \
313 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
314 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
316 #define BYTEWIDTH 8 /* In bits. */
318 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
322 #define MAX(a, b) ((a) > (b) ? (a) : (b))
323 #define MIN(a, b) ((a) < (b) ? (a) : (b))
325 typedef char boolean;
329 static int re_match_2_internal ();
331 /* These are the command codes that appear in compiled regular
332 expressions. Some opcodes are followed by argument bytes. A
333 command code can specify any interpretation whatsoever for its
334 arguments. Zero bytes may appear in the compiled regular expression. */
340 /* Succeed right away--no more backtracking. */
343 /* Followed by one byte giving n, then by n literal bytes. */
346 /* Matches any (more or less) character. */
349 /* Matches any one char belonging to specified set. First
350 following byte is number of bitmap bytes. Then come bytes
351 for a bitmap saying which chars are in. Bits in each byte
352 are ordered low-bit-first. A character is in the set if its
353 bit is 1. A character too large to have a bit in the map is
354 automatically not in the set. */
357 /* Same parameters as charset, but match any character that is
358 not one of those specified. */
361 /* Start remembering the text that is matched, for storing in a
362 register. Followed by one byte with the register number, in
363 the range 0 to one less than the pattern buffer's re_nsub
364 field. Then followed by one byte with the number of groups
365 inner to this one. (This last has to be part of the
366 start_memory only because we need it in the on_failure_jump
370 /* Stop remembering the text that is matched and store it in a
371 memory register. Followed by one byte with the register
372 number, in the range 0 to one less than `re_nsub' in the
373 pattern buffer, and one byte with the number of inner groups,
374 just like `start_memory'. (We need the number of inner
375 groups here because we don't have any easy way of finding the
376 corresponding start_memory when we're at a stop_memory.) */
379 /* Match a duplicate of something remembered. Followed by one
380 byte containing the register number. */
383 /* Fail unless at beginning of line. */
386 /* Fail unless at end of line. */
389 /* Succeeds if at beginning of buffer (if emacs) or at beginning
390 of string to be matched (if not). */
393 /* Analogously, for end of buffer/string. */
396 /* Followed by two byte relative address to which to jump. */
399 /* Same as jump, but marks the end of an alternative. */
402 /* Followed by two-byte relative address of place to resume at
403 in case of failure. */
406 /* Like on_failure_jump, but pushes a placeholder instead of the
407 current string position when executed. */
408 on_failure_keep_string_jump,
410 /* Throw away latest failure point and then jump to following
411 two-byte relative address. */
414 /* Change to pop_failure_jump if know won't have to backtrack to
415 match; otherwise change to jump. This is used to jump
416 back to the beginning of a repeat. If what follows this jump
417 clearly won't match what the repeat does, such that we can be
418 sure that there is no use backtracking out of repetitions
419 already matched, then we change it to a pop_failure_jump.
420 Followed by two-byte address. */
423 /* Jump to following two-byte address, and push a dummy failure
424 point. This failure point will be thrown away if an attempt
425 is made to use it for a failure. A `+' construct makes this
426 before the first repeat. Also used as an intermediary kind
427 of jump when compiling an alternative. */
430 /* Push a dummy failure point and continue. Used at the end of
434 /* Followed by two-byte relative address and two-byte number n.
435 After matching N times, jump to the address upon failure. */
438 /* Followed by two-byte relative address, and two-byte number n.
439 Jump to the address N times, then fail. */
442 /* Set the following two-byte relative address to the
443 subsequent two-byte number. The address *includes* the two
447 wordchar, /* Matches any word-constituent character. */
448 notwordchar, /* Matches any char that is not a word-constituent. */
450 wordbeg, /* Succeeds if at word beginning. */
451 wordend, /* Succeeds if at word end. */
453 wordbound, /* Succeeds if at a word boundary. */
454 notwordbound /* Succeeds if not at a word boundary. */
457 ,before_dot, /* Succeeds if before point. */
458 at_dot, /* Succeeds if at point. */
459 after_dot, /* Succeeds if after point. */
461 /* Matches any character whose syntax is specified. Followed by
462 a byte which contains a syntax code, e.g., Sword. */
465 /* Matches any character whose syntax is not that specified. */
470 /* Common operations on the compiled pattern. */
472 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
474 #define STORE_NUMBER(destination, number) \
476 (destination)[0] = (number) & 0377; \
477 (destination)[1] = (number) >> 8; \
480 /* Same as STORE_NUMBER, except increment DESTINATION to
481 the byte after where the number is stored. Therefore, DESTINATION
482 must be an lvalue. */
484 #define STORE_NUMBER_AND_INCR(destination, number) \
486 STORE_NUMBER (destination, number); \
487 (destination) += 2; \
490 /* Put into DESTINATION a number stored in two contiguous bytes starting
493 #define EXTRACT_NUMBER(destination, source) \
495 (destination) = *(source) & 0377; \
496 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
501 extract_number (dest, source)
503 unsigned char *source;
505 int temp = SIGN_EXTEND_CHAR (*(source + 1));
506 *dest = *source & 0377;
510 #ifndef EXTRACT_MACROS /* To debug the macros. */
511 #undef EXTRACT_NUMBER
512 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
513 #endif /* not EXTRACT_MACROS */
517 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
518 SOURCE must be an lvalue. */
520 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
522 EXTRACT_NUMBER (destination, source); \
528 extract_number_and_incr (destination, source)
530 unsigned char **source;
532 extract_number (destination, *source);
536 #ifndef EXTRACT_MACROS
537 #undef EXTRACT_NUMBER_AND_INCR
538 #define EXTRACT_NUMBER_AND_INCR(dest, src) \
539 extract_number_and_incr (&dest, &src)
540 #endif /* not EXTRACT_MACROS */
544 /* If DEBUG is defined, Regex prints many voluminous messages about what
545 it is doing (if the variable `debug' is nonzero). If linked with the
546 main program in `iregex.c', you can enter patterns and strings
547 interactively. And if linked with the main program in `main.c' and
548 the other test files, you can run the already-written tests. */
552 /* We use standard I/O for debugging. */
555 /* It is useful to test things that ``must'' be true when debugging. */
558 static int debug = 0;
560 #define DEBUG_STATEMENT(e) e
561 #define DEBUG_PRINT1(x) if (debug) printf (x)
562 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
563 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
564 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
565 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
566 if (debug) print_partial_compiled_pattern (s, e)
567 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
568 if (debug) print_double_string (w, s1, sz1, s2, sz2)
571 /* Print the fastmap in human-readable form. */
574 print_fastmap (fastmap)
577 unsigned was_a_range = 0;
580 while (i < (1 << BYTEWIDTH))
586 while (i < (1 << BYTEWIDTH) && fastmap[i])
602 /* Print a compiled pattern string in human-readable form, starting at
603 the START pointer into it and ending just before the pointer END. */
606 print_partial_compiled_pattern (start, end)
607 unsigned char *start;
611 unsigned char *p = start;
612 unsigned char *pend = end;
620 /* Loop over pattern commands. */
623 printf ("%d:\t", p - start);
625 switch ((re_opcode_t) *p++)
633 printf ("/exactn/%d", mcnt);
644 printf ("/start_memory/%d/%d", mcnt, *p++);
649 printf ("/stop_memory/%d/%d", mcnt, *p++);
653 printf ("/duplicate/%d", *p++);
663 register int c, last = -100;
664 register int in_range = 0;
666 printf ("/charset [%s",
667 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
669 assert (p + *p < pend);
671 for (c = 0; c < 256; c++)
673 && (p[1 + (c/8)] & (1 << (c % 8))))
675 /* Are we starting a range? */
676 if (last + 1 == c && ! in_range)
681 /* Have we broken a range? */
682 else if (last + 1 != c && in_range)
711 case on_failure_jump:
712 extract_number_and_incr (&mcnt, &p);
713 printf ("/on_failure_jump to %d", p + mcnt - start);
716 case on_failure_keep_string_jump:
717 extract_number_and_incr (&mcnt, &p);
718 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
721 case dummy_failure_jump:
722 extract_number_and_incr (&mcnt, &p);
723 printf ("/dummy_failure_jump to %d", p + mcnt - start);
726 case push_dummy_failure:
727 printf ("/push_dummy_failure");
731 extract_number_and_incr (&mcnt, &p);
732 printf ("/maybe_pop_jump to %d", p + mcnt - start);
735 case pop_failure_jump:
736 extract_number_and_incr (&mcnt, &p);
737 printf ("/pop_failure_jump to %d", p + mcnt - start);
741 extract_number_and_incr (&mcnt, &p);
742 printf ("/jump_past_alt to %d", p + mcnt - start);
746 extract_number_and_incr (&mcnt, &p);
747 printf ("/jump to %d", p + mcnt - start);
751 extract_number_and_incr (&mcnt, &p);
752 extract_number_and_incr (&mcnt2, &p);
753 printf ("/succeed_n to %d, %d times", p + mcnt - start, mcnt2);
757 extract_number_and_incr (&mcnt, &p);
758 extract_number_and_incr (&mcnt2, &p);
759 printf ("/jump_n to %d, %d times", p + mcnt - start, mcnt2);
763 extract_number_and_incr (&mcnt, &p);
764 extract_number_and_incr (&mcnt2, &p);
765 printf ("/set_number_at location %d to %d", p + mcnt - start, mcnt2);
769 printf ("/wordbound");
773 printf ("/notwordbound");
785 printf ("/before_dot");
793 printf ("/after_dot");
797 printf ("/syntaxspec");
799 printf ("/%d", mcnt);
803 printf ("/notsyntaxspec");
805 printf ("/%d", mcnt);
810 printf ("/wordchar");
814 printf ("/notwordchar");
826 printf ("?%d", *(p-1));
832 printf ("%d:\tend of pattern.\n", p - start);
837 print_compiled_pattern (bufp)
838 struct re_pattern_buffer *bufp;
840 unsigned char *buffer = bufp->buffer;
842 print_partial_compiled_pattern (buffer, buffer + bufp->used);
843 printf ("%d bytes used/%d bytes allocated.\n", bufp->used, bufp->allocated);
845 if (bufp->fastmap_accurate && bufp->fastmap)
847 printf ("fastmap: ");
848 print_fastmap (bufp->fastmap);
851 printf ("re_nsub: %d\t", bufp->re_nsub);
852 printf ("regs_alloc: %d\t", bufp->regs_allocated);
853 printf ("can_be_null: %d\t", bufp->can_be_null);
854 printf ("newline_anchor: %d\n", bufp->newline_anchor);
855 printf ("no_sub: %d\t", bufp->no_sub);
856 printf ("not_bol: %d\t", bufp->not_bol);
857 printf ("not_eol: %d\t", bufp->not_eol);
858 printf ("syntax: %d\n", bufp->syntax);
859 /* Perhaps we should print the translate table? */
864 print_double_string (where, string1, size1, string2, size2)
877 if (FIRST_STRING_P (where))
879 for (this_char = where - string1; this_char < size1; this_char++)
880 putchar (string1[this_char]);
885 for (this_char = where - string2; this_char < size2; this_char++)
886 putchar (string2[this_char]);
890 #else /* not DEBUG */
895 #define DEBUG_STATEMENT(e)
896 #define DEBUG_PRINT1(x)
897 #define DEBUG_PRINT2(x1, x2)
898 #define DEBUG_PRINT3(x1, x2, x3)
899 #define DEBUG_PRINT4(x1, x2, x3, x4)
900 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
901 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
903 #endif /* not DEBUG */
905 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
906 also be assigned to arbitrarily: each pattern buffer stores its own
907 syntax, so it can be changed between regex compilations. */
908 /* This has no initializer because initialized variables in Emacs
909 become read-only after dumping. */
910 reg_syntax_t re_syntax_options;
913 /* Specify the precise syntax of regexps for compilation. This provides
914 for compatibility for various utilities which historically have
915 different, incompatible syntaxes.
917 The argument SYNTAX is a bit mask comprised of the various bits
918 defined in regex.h. We return the old syntax. */
921 re_set_syntax (syntax)
924 reg_syntax_t ret = re_syntax_options;
926 re_syntax_options = syntax;
930 /* This table gives an error message for each of the error codes listed
931 in regex.h. Obviously the order here has to be same as there.
932 POSIX doesn't require that we do anything for REG_NOERROR,
933 but why not be nice? */
935 static const char *re_error_msgid[] =
937 gettext_noop ("Success"), /* REG_NOERROR */
938 gettext_noop ("No match"), /* REG_NOMATCH */
939 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
940 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
941 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
942 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
943 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
944 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
945 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
946 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
947 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
948 gettext_noop ("Invalid range end"), /* REG_ERANGE */
949 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
950 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
951 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
952 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
953 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
956 /* Avoiding alloca during matching, to placate r_alloc. */
958 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
959 searching and matching functions should not call alloca. On some
960 systems, alloca is implemented in terms of malloc, and if we're
961 using the relocating allocator routines, then malloc could cause a
962 relocation, which might (if the strings being searched are in the
963 ralloc heap) shift the data out from underneath the regexp
966 Here's another reason to avoid allocation: Emacs
967 processes input from X in a signal handler; processing X input may
968 call malloc; if input arrives while a matching routine is calling
969 malloc, then we're scrod. But Emacs can't just block input while
970 calling matching routines; then we don't notice interrupts when
971 they come in. So, Emacs blocks input around all regexp calls
972 except the matching calls, which it leaves unprotected, in the
973 faith that they will not malloc. */
975 /* Normally, this is fine. */
976 #define MATCH_MAY_ALLOCATE
978 /* When using GNU C, we are not REALLY using the C alloca, no matter
979 what config.h may say. So don't take precautions for it. */
984 /* The match routines may not allocate if (1) they would do it with malloc
985 and (2) it's not safe for them to use malloc.
986 Note that if REL_ALLOC is defined, matching would not use malloc for the
987 failure stack, but we would still use it for the register vectors;
988 so REL_ALLOC should not affect this. */
989 #if (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (emacs)
990 #undef MATCH_MAY_ALLOCATE
994 /* Failure stack declarations and macros; both re_compile_fastmap and
995 re_match_2 use a failure stack. These have to be macros because of
996 REGEX_ALLOCATE_STACK. */
999 /* Number of failure points for which to initially allocate space
1000 when matching. If this number is exceeded, we allocate more
1001 space, so it is not a hard limit. */
1002 #ifndef INIT_FAILURE_ALLOC
1003 #define INIT_FAILURE_ALLOC 5
1006 /* Roughly the maximum number of failure points on the stack. Would be
1007 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1008 This is a variable only so users of regex can assign to it; we never
1009 change it ourselves. */
1010 #if defined (MATCH_MAY_ALLOCATE)
1011 /* 4400 was enough to cause a crash on Alpha OSF/1,
1012 whose default stack limit is 2mb. */
1013 int re_max_failures = 4000;
1015 int re_max_failures = 2000;
1018 union fail_stack_elt
1020 unsigned char *pointer;
1024 typedef union fail_stack_elt fail_stack_elt_t;
1028 fail_stack_elt_t *stack;
1030 unsigned avail; /* Offset of next open position. */
1033 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1034 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1035 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1038 /* Define macros to initialize and free the failure stack.
1039 Do `return -2' if the alloc fails. */
1041 #ifdef MATCH_MAY_ALLOCATE
1042 #define INIT_FAIL_STACK() \
1044 fail_stack.stack = (fail_stack_elt_t *) \
1045 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1047 if (fail_stack.stack == NULL) \
1050 fail_stack.size = INIT_FAILURE_ALLOC; \
1051 fail_stack.avail = 0; \
1054 #define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1056 #define INIT_FAIL_STACK() \
1058 fail_stack.avail = 0; \
1061 #define RESET_FAIL_STACK()
1065 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1067 Return 1 if succeeds, and 0 if either ran out of memory
1068 allocating space for it or it was already too large.
1070 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1072 #define DOUBLE_FAIL_STACK(fail_stack) \
1073 ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \
1075 : ((fail_stack).stack = (fail_stack_elt_t *) \
1076 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1077 (fail_stack).size * sizeof (fail_stack_elt_t), \
1078 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1080 (fail_stack).stack == NULL \
1082 : ((fail_stack).size <<= 1, \
1086 /* Push pointer POINTER on FAIL_STACK.
1087 Return 1 if was able to do so and 0 if ran out of memory allocating
1089 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1090 ((FAIL_STACK_FULL () \
1091 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1093 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1096 /* Push a pointer value onto the failure stack.
1097 Assumes the variable `fail_stack'. Probably should only
1098 be called from within `PUSH_FAILURE_POINT'. */
1099 #define PUSH_FAILURE_POINTER(item) \
1100 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1102 /* This pushes an integer-valued item onto the failure stack.
1103 Assumes the variable `fail_stack'. Probably should only
1104 be called from within `PUSH_FAILURE_POINT'. */
1105 #define PUSH_FAILURE_INT(item) \
1106 fail_stack.stack[fail_stack.avail++].integer = (item)
1108 /* Push a fail_stack_elt_t value onto the failure stack.
1109 Assumes the variable `fail_stack'. Probably should only
1110 be called from within `PUSH_FAILURE_POINT'. */
1111 #define PUSH_FAILURE_ELT(item) \
1112 fail_stack.stack[fail_stack.avail++] = (item)
1114 /* These three POP... operations complement the three PUSH... operations.
1115 All assume that `fail_stack' is nonempty. */
1116 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1117 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1118 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1120 /* Used to omit pushing failure point id's when we're not debugging. */
1122 #define DEBUG_PUSH PUSH_FAILURE_INT
1123 #define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1125 #define DEBUG_PUSH(item)
1126 #define DEBUG_POP(item_addr)
1130 /* Push the information about the state we will need
1131 if we ever fail back to it.
1133 Requires variables fail_stack, regstart, regend, reg_info, and
1134 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be
1137 Does `return FAILURE_CODE' if runs out of memory. */
1139 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1141 char *destination; \
1142 /* Must be int, so when we don't save any registers, the arithmetic \
1143 of 0 + -1 isn't done as unsigned. */ \
1146 DEBUG_STATEMENT (failure_id++); \
1147 DEBUG_STATEMENT (nfailure_points_pushed++); \
1148 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1149 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1150 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1152 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \
1153 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1155 /* Ensure we have enough space allocated for what we will push. */ \
1156 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1158 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1159 return failure_code; \
1161 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1162 (fail_stack).size); \
1163 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1166 /* Push the info, starting with the registers. */ \
1167 DEBUG_PRINT1 ("\n"); \
1170 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1173 DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \
1174 DEBUG_STATEMENT (num_regs_pushed++); \
1176 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1177 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1179 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1180 PUSH_FAILURE_POINTER (regend[this_reg]); \
1182 DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \
1183 DEBUG_PRINT2 (" match_null=%d", \
1184 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1185 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1186 DEBUG_PRINT2 (" matched_something=%d", \
1187 MATCHED_SOMETHING (reg_info[this_reg])); \
1188 DEBUG_PRINT2 (" ever_matched=%d", \
1189 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1190 DEBUG_PRINT1 ("\n"); \
1191 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1194 DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\
1195 PUSH_FAILURE_INT (lowest_active_reg); \
1197 DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\
1198 PUSH_FAILURE_INT (highest_active_reg); \
1200 DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \
1201 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1202 PUSH_FAILURE_POINTER (pattern_place); \
1204 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \
1205 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1207 DEBUG_PRINT1 ("'\n"); \
1208 PUSH_FAILURE_POINTER (string_place); \
1210 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1211 DEBUG_PUSH (failure_id); \
1214 /* This is the number of items that are pushed and popped on the stack
1215 for each register. */
1216 #define NUM_REG_ITEMS 3
1218 /* Individual items aside from the registers. */
1220 #define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1222 #define NUM_NONREG_ITEMS 4
1225 /* We push at most this many items on the stack. */
1226 /* We used to use (num_regs - 1), which is the number of registers
1227 this regexp will save; but that was changed to 5
1228 to avoid stack overflow for a regexp with lots of parens. */
1229 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1231 /* We actually push this many items. */
1232 #define NUM_FAILURE_ITEMS \
1234 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1238 /* How many items can still be added to the stack without overflowing it. */
1239 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1242 /* Pops what PUSH_FAIL_STACK pushes.
1244 We restore into the parameters, all of which should be lvalues:
1245 STR -- the saved data position.
1246 PAT -- the saved pattern position.
1247 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1248 REGSTART, REGEND -- arrays of string positions.
1249 REG_INFO -- array of information about each subexpression.
1251 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1252 `pend', `string1', `size1', `string2', and `size2'. */
1254 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1256 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \
1258 const unsigned char *string_temp; \
1260 assert (!FAIL_STACK_EMPTY ()); \
1262 /* Remove failure points and point to how many regs pushed. */ \
1263 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1264 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1265 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1267 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1269 DEBUG_POP (&failure_id); \
1270 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1272 /* If the saved string location is NULL, it came from an \
1273 on_failure_keep_string_jump opcode, and we want to throw away the \
1274 saved NULL, thus retaining our current position in the string. */ \
1275 string_temp = POP_FAILURE_POINTER (); \
1276 if (string_temp != NULL) \
1277 str = (const char *) string_temp; \
1279 DEBUG_PRINT2 (" Popping string 0x%x: `", str); \
1280 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1281 DEBUG_PRINT1 ("'\n"); \
1283 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1284 DEBUG_PRINT2 (" Popping pattern 0x%x: ", pat); \
1285 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1287 /* Restore register info. */ \
1288 high_reg = (unsigned) POP_FAILURE_INT (); \
1289 DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \
1291 low_reg = (unsigned) POP_FAILURE_INT (); \
1292 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \
1295 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1297 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \
1299 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1300 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \
1302 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1303 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1305 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1306 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1310 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1312 reg_info[this_reg].word.integer = 0; \
1313 regend[this_reg] = 0; \
1314 regstart[this_reg] = 0; \
1316 highest_active_reg = high_reg; \
1319 set_regs_matched_done = 0; \
1320 DEBUG_STATEMENT (nfailure_points_popped++); \
1321 } /* POP_FAILURE_POINT */
1325 /* Structure for per-register (a.k.a. per-group) information.
1326 Other register information, such as the
1327 starting and ending positions (which are addresses), and the list of
1328 inner groups (which is a bits list) are maintained in separate
1331 We are making a (strictly speaking) nonportable assumption here: that
1332 the compiler will pack our bit fields into something that fits into
1333 the type of `word', i.e., is something that fits into one item on the
1338 fail_stack_elt_t word;
1341 /* This field is one if this group can match the empty string,
1342 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1343 #define MATCH_NULL_UNSET_VALUE 3
1344 unsigned match_null_string_p : 2;
1345 unsigned is_active : 1;
1346 unsigned matched_something : 1;
1347 unsigned ever_matched_something : 1;
1349 } register_info_type;
1351 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1352 #define IS_ACTIVE(R) ((R).bits.is_active)
1353 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1354 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1357 /* Call this when have matched a real character; it sets `matched' flags
1358 for the subexpressions which we are currently inside. Also records
1359 that those subexprs have matched. */
1360 #define SET_REGS_MATCHED() \
1363 if (!set_regs_matched_done) \
1366 set_regs_matched_done = 1; \
1367 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1369 MATCHED_SOMETHING (reg_info[r]) \
1370 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1377 /* Registers are set to a sentinel when they haven't yet matched. */
1378 static char reg_unset_dummy;
1379 #define REG_UNSET_VALUE (®_unset_dummy)
1380 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1382 /* Subroutine declarations and macros for regex_compile. */
1384 static void store_op1 (), store_op2 ();
1385 static void insert_op1 (), insert_op2 ();
1386 static boolean at_begline_loc_p (), at_endline_loc_p ();
1387 static boolean group_in_compile_stack ();
1388 static reg_errcode_t compile_range ();
1390 /* Fetch the next character in the uncompiled pattern---translating it
1391 if necessary. Also cast from a signed character in the constant
1392 string passed to us by the user to an unsigned char that we can use
1393 as an array index (in, e.g., `translate'). */
1395 #define PATFETCH(c) \
1396 do {if (p == pend) return REG_EEND; \
1397 c = (unsigned char) *p++; \
1398 if (translate) c = (unsigned char) translate[c]; \
1402 /* Fetch the next character in the uncompiled pattern, with no
1404 #define PATFETCH_RAW(c) \
1405 do {if (p == pend) return REG_EEND; \
1406 c = (unsigned char) *p++; \
1409 /* Go backwards one character in the pattern. */
1410 #define PATUNFETCH p--
1413 /* If `translate' is non-null, return translate[D], else just D. We
1414 cast the subscript to translate because some data is declared as
1415 `char *', to avoid warnings when a string constant is passed. But
1416 when we use a character as a subscript we must make it unsigned. */
1418 #define TRANSLATE(d) \
1419 (translate ? (char) translate[(unsigned char) (d)] : (d))
1423 /* Macros for outputting the compiled pattern into `buffer'. */
1425 /* If the buffer isn't allocated when it comes in, use this. */
1426 #define INIT_BUF_SIZE 32
1428 /* Make sure we have at least N more bytes of space in buffer. */
1429 #define GET_BUFFER_SPACE(n) \
1430 while (b - bufp->buffer + (n) > bufp->allocated) \
1433 /* Make sure we have one more byte of buffer space and then add C to it. */
1434 #define BUF_PUSH(c) \
1436 GET_BUFFER_SPACE (1); \
1437 *b++ = (unsigned char) (c); \
1441 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1442 #define BUF_PUSH_2(c1, c2) \
1444 GET_BUFFER_SPACE (2); \
1445 *b++ = (unsigned char) (c1); \
1446 *b++ = (unsigned char) (c2); \
1450 /* As with BUF_PUSH_2, except for three bytes. */
1451 #define BUF_PUSH_3(c1, c2, c3) \
1453 GET_BUFFER_SPACE (3); \
1454 *b++ = (unsigned char) (c1); \
1455 *b++ = (unsigned char) (c2); \
1456 *b++ = (unsigned char) (c3); \
1460 /* Store a jump with opcode OP at LOC to location TO. We store a
1461 relative address offset by the three bytes the jump itself occupies. */
1462 #define STORE_JUMP(op, loc, to) \
1463 store_op1 (op, loc, (to) - (loc) - 3)
1465 /* Likewise, for a two-argument jump. */
1466 #define STORE_JUMP2(op, loc, to, arg) \
1467 store_op2 (op, loc, (to) - (loc) - 3, arg)
1469 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1470 #define INSERT_JUMP(op, loc, to) \
1471 insert_op1 (op, loc, (to) - (loc) - 3, b)
1473 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1474 #define INSERT_JUMP2(op, loc, to, arg) \
1475 insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
1478 /* This is not an arbitrary limit: the arguments which represent offsets
1479 into the pattern are two bytes long. So if 2^16 bytes turns out to
1480 be too small, many things would have to change. */
1481 #define MAX_BUF_SIZE (1L << 16)
1484 /* Extend the buffer by twice its current size via realloc and
1485 reset the pointers that pointed into the old block to point to the
1486 correct places in the new one. If extending the buffer results in it
1487 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1488 #define EXTEND_BUFFER() \
1490 unsigned char *old_buffer = bufp->buffer; \
1491 if (bufp->allocated == MAX_BUF_SIZE) \
1493 bufp->allocated <<= 1; \
1494 if (bufp->allocated > MAX_BUF_SIZE) \
1495 bufp->allocated = MAX_BUF_SIZE; \
1496 bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\
1497 if (bufp->buffer == NULL) \
1498 return REG_ESPACE; \
1499 /* If the buffer moved, move all the pointers into it. */ \
1500 if (old_buffer != bufp->buffer) \
1502 b = (b - old_buffer) + bufp->buffer; \
1503 begalt = (begalt - old_buffer) + bufp->buffer; \
1504 if (fixup_alt_jump) \
1505 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1507 laststart = (laststart - old_buffer) + bufp->buffer; \
1508 if (pending_exact) \
1509 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1514 /* Since we have one byte reserved for the register number argument to
1515 {start,stop}_memory, the maximum number of groups we can report
1516 things about is what fits in that byte. */
1517 #define MAX_REGNUM 255
1519 /* But patterns can have more than `MAX_REGNUM' registers. We just
1520 ignore the excess. */
1521 typedef unsigned regnum_t;
1524 /* Macros for the compile stack. */
1526 /* Since offsets can go either forwards or backwards, this type needs to
1527 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1528 typedef int pattern_offset_t;
1532 pattern_offset_t begalt_offset;
1533 pattern_offset_t fixup_alt_jump;
1534 pattern_offset_t inner_group_offset;
1535 pattern_offset_t laststart_offset;
1537 } compile_stack_elt_t;
1542 compile_stack_elt_t *stack;
1544 unsigned avail; /* Offset of next open position. */
1545 } compile_stack_type;
1548 #define INIT_COMPILE_STACK_SIZE 32
1550 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1551 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1553 /* The next available element. */
1554 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1557 /* Set the bit for character C in a list. */
1558 #define SET_LIST_BIT(c) \
1559 (b[((unsigned char) (c)) / BYTEWIDTH] \
1560 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1563 /* Get the next unsigned number in the uncompiled pattern. */
1564 #define GET_UNSIGNED_NUMBER(num) \
1568 while (ISDIGIT (c)) \
1572 num = num * 10 + c - '0'; \
1580 #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1582 #define IS_CHAR_CLASS(string) \
1583 (STREQ (string, "alpha") || STREQ (string, "upper") \
1584 || STREQ (string, "lower") || STREQ (string, "digit") \
1585 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1586 || STREQ (string, "space") || STREQ (string, "print") \
1587 || STREQ (string, "punct") || STREQ (string, "graph") \
1588 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1590 #ifndef MATCH_MAY_ALLOCATE
1592 /* If we cannot allocate large objects within re_match_2_internal,
1593 we make the fail stack and register vectors global.
1594 The fail stack, we grow to the maximum size when a regexp
1596 The register vectors, we adjust in size each time we
1597 compile a regexp, according to the number of registers it needs. */
1599 static fail_stack_type fail_stack;
1601 /* Size with which the following vectors are currently allocated.
1602 That is so we can make them bigger as needed,
1603 but never make them smaller. */
1604 static int regs_allocated_size;
1606 static const char ** regstart, ** regend;
1607 static const char ** old_regstart, ** old_regend;
1608 static const char **best_regstart, **best_regend;
1609 static register_info_type *reg_info;
1610 static const char **reg_dummy;
1611 static register_info_type *reg_info_dummy;
1613 /* Make the register vectors big enough for NUM_REGS registers,
1614 but don't make them smaller. */
1617 regex_grow_registers (num_regs)
1620 if (num_regs > regs_allocated_size)
1622 RETALLOC_IF (regstart, num_regs, const char *);
1623 RETALLOC_IF (regend, num_regs, const char *);
1624 RETALLOC_IF (old_regstart, num_regs, const char *);
1625 RETALLOC_IF (old_regend, num_regs, const char *);
1626 RETALLOC_IF (best_regstart, num_regs, const char *);
1627 RETALLOC_IF (best_regend, num_regs, const char *);
1628 RETALLOC_IF (reg_info, num_regs, register_info_type);
1629 RETALLOC_IF (reg_dummy, num_regs, const char *);
1630 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1632 regs_allocated_size = num_regs;
1636 #endif /* not MATCH_MAY_ALLOCATE */
1638 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1639 Returns one of error codes defined in `regex.h', or zero for success.
1641 Assumes the `allocated' (and perhaps `buffer') and `translate'
1642 fields are set in BUFP on entry.
1644 If it succeeds, results are put in BUFP (if it returns an error, the
1645 contents of BUFP are undefined):
1646 `buffer' is the compiled pattern;
1647 `syntax' is set to SYNTAX;
1648 `used' is set to the length of the compiled pattern;
1649 `fastmap_accurate' is zero;
1650 `re_nsub' is the number of subexpressions in PATTERN;
1651 `not_bol' and `not_eol' are zero;
1653 The `fastmap' and `newline_anchor' fields are neither
1654 examined nor set. */
1656 /* Return, freeing storage we allocated. */
1657 #define FREE_STACK_RETURN(value) \
1658 return (free (compile_stack.stack), value)
1660 static reg_errcode_t
1661 regex_compile (pattern, size, syntax, bufp)
1662 const char *pattern;
1664 reg_syntax_t syntax;
1665 struct re_pattern_buffer *bufp;
1667 /* We fetch characters from PATTERN here. Even though PATTERN is
1668 `char *' (i.e., signed), we declare these variables as unsigned, so
1669 they can be reliably used as array indices. */
1670 register unsigned char c, c1;
1672 /* A random temporary spot in PATTERN. */
1675 /* Points to the end of the buffer, where we should append. */
1676 register unsigned char *b;
1678 /* Keeps track of unclosed groups. */
1679 compile_stack_type compile_stack;
1681 /* Points to the current (ending) position in the pattern. */
1682 const char *p = pattern;
1683 const char *pend = pattern + size;
1685 /* How to translate the characters in the pattern. */
1686 RE_TRANSLATE_TYPE translate = bufp->translate;
1688 /* Address of the count-byte of the most recently inserted `exactn'
1689 command. This makes it possible to tell if a new exact-match
1690 character can be added to that command or if the character requires
1691 a new `exactn' command. */
1692 unsigned char *pending_exact = 0;
1694 /* Address of start of the most recently finished expression.
1695 This tells, e.g., postfix * where to find the start of its
1696 operand. Reset at the beginning of groups and alternatives. */
1697 unsigned char *laststart = 0;
1699 /* Address of beginning of regexp, or inside of last group. */
1700 unsigned char *begalt;
1702 /* Place in the uncompiled pattern (i.e., the {) to
1703 which to go back if the interval is invalid. */
1704 const char *beg_interval;
1706 /* Address of the place where a forward jump should go to the end of
1707 the containing expression. Each alternative of an `or' -- except the
1708 last -- ends with a forward jump of this sort. */
1709 unsigned char *fixup_alt_jump = 0;
1711 /* Counts open-groups as they are encountered. Remembered for the
1712 matching close-group on the compile stack, so the same register
1713 number is put in the stop_memory as the start_memory. */
1714 regnum_t regnum = 0;
1717 DEBUG_PRINT1 ("\nCompiling pattern: ");
1720 unsigned debug_count;
1722 for (debug_count = 0; debug_count < size; debug_count++)
1723 putchar (pattern[debug_count]);
1728 /* Initialize the compile stack. */
1729 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1730 if (compile_stack.stack == NULL)
1733 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1734 compile_stack.avail = 0;
1736 /* Initialize the pattern buffer. */
1737 bufp->syntax = syntax;
1738 bufp->fastmap_accurate = 0;
1739 bufp->not_bol = bufp->not_eol = 0;
1741 /* Set `used' to zero, so that if we return an error, the pattern
1742 printer (for debugging) will think there's no pattern. We reset it
1746 /* Always count groups, whether or not bufp->no_sub is set. */
1749 #if !defined (emacs) && !defined (SYNTAX_TABLE)
1750 /* Initialize the syntax table. */
1751 init_syntax_once ();
1754 if (bufp->allocated == 0)
1757 { /* If zero allocated, but buffer is non-null, try to realloc
1758 enough space. This loses if buffer's address is bogus, but
1759 that is the user's responsibility. */
1760 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1763 { /* Caller did not allocate a buffer. Do it for them. */
1764 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1766 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1768 bufp->allocated = INIT_BUF_SIZE;
1771 begalt = b = bufp->buffer;
1773 /* Loop through the uncompiled pattern until we're at the end. */
1782 if ( /* If at start of pattern, it's an operator. */
1784 /* If context independent, it's an operator. */
1785 || syntax & RE_CONTEXT_INDEP_ANCHORS
1786 /* Otherwise, depends on what's come before. */
1787 || at_begline_loc_p (pattern, p, syntax))
1797 if ( /* If at end of pattern, it's an operator. */
1799 /* If context independent, it's an operator. */
1800 || syntax & RE_CONTEXT_INDEP_ANCHORS
1801 /* Otherwise, depends on what's next. */
1802 || at_endline_loc_p (p, pend, syntax))
1812 if ((syntax & RE_BK_PLUS_QM)
1813 || (syntax & RE_LIMITED_OPS))
1817 /* If there is no previous pattern... */
1820 if (syntax & RE_CONTEXT_INVALID_OPS)
1821 FREE_STACK_RETURN (REG_BADRPT);
1822 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1827 /* Are we optimizing this jump? */
1828 boolean keep_string_p = false;
1830 /* 1 means zero (many) matches is allowed. */
1831 char zero_times_ok = 0, many_times_ok = 0;
1833 /* If there is a sequence of repetition chars, collapse it
1834 down to just one (the right one). We can't combine
1835 interval operators with these because of, e.g., `a{2}*',
1836 which should only match an even number of `a's. */
1840 zero_times_ok |= c != '+';
1841 many_times_ok |= c != '?';
1849 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
1852 else if (syntax & RE_BK_PLUS_QM && c == '\\')
1854 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1857 if (!(c1 == '+' || c1 == '?'))
1872 /* If we get here, we found another repeat character. */
1875 /* Star, etc. applied to an empty pattern is equivalent
1876 to an empty pattern. */
1880 /* Now we know whether or not zero matches is allowed
1881 and also whether or not two or more matches is allowed. */
1883 { /* More than one repetition is allowed, so put in at the
1884 end a backward relative jump from `b' to before the next
1885 jump we're going to put in below (which jumps from
1886 laststart to after this jump).
1888 But if we are at the `*' in the exact sequence `.*\n',
1889 insert an unconditional jump backwards to the .,
1890 instead of the beginning of the loop. This way we only
1891 push a failure point once, instead of every time
1892 through the loop. */
1893 assert (p - 1 > pattern);
1895 /* Allocate the space for the jump. */
1896 GET_BUFFER_SPACE (3);
1898 /* We know we are not at the first character of the pattern,
1899 because laststart was nonzero. And we've already
1900 incremented `p', by the way, to be the character after
1901 the `*'. Do we have to do something analogous here
1902 for null bytes, because of RE_DOT_NOT_NULL? */
1903 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
1905 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
1906 && !(syntax & RE_DOT_NEWLINE))
1907 { /* We have .*\n. */
1908 STORE_JUMP (jump, b, laststart);
1909 keep_string_p = true;
1912 /* Anything else. */
1913 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
1915 /* We've added more stuff to the buffer. */
1919 /* On failure, jump from laststart to b + 3, which will be the
1920 end of the buffer after this jump is inserted. */
1921 GET_BUFFER_SPACE (3);
1922 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
1930 /* At least one repetition is required, so insert a
1931 `dummy_failure_jump' before the initial
1932 `on_failure_jump' instruction of the loop. This
1933 effects a skip over that instruction the first time
1934 we hit that loop. */
1935 GET_BUFFER_SPACE (3);
1936 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
1951 boolean had_char_class = false;
1953 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1955 /* Ensure that we have enough space to push a charset: the
1956 opcode, the length count, and the bitset; 34 bytes in all. */
1957 GET_BUFFER_SPACE (34);
1961 /* We test `*p == '^' twice, instead of using an if
1962 statement, so we only need one BUF_PUSH. */
1963 BUF_PUSH (*p == '^' ? charset_not : charset);
1967 /* Remember the first position in the bracket expression. */
1970 /* Push the number of bytes in the bitmap. */
1971 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
1973 /* Clear the whole map. */
1974 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
1976 /* charset_not matches newline according to a syntax bit. */
1977 if ((re_opcode_t) b[-2] == charset_not
1978 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
1979 SET_LIST_BIT ('\n');
1981 /* Read in characters and ranges, setting map bits. */
1984 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1988 /* \ might escape characters inside [...] and [^...]. */
1989 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
1991 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1998 /* Could be the end of the bracket expression. If it's
1999 not (i.e., when the bracket expression is `[]' so
2000 far), the ']' character bit gets set way below. */
2001 if (c == ']' && p != p1 + 1)
2004 /* Look ahead to see if it's a range when the last thing
2005 was a character class. */
2006 if (had_char_class && c == '-' && *p != ']')
2007 FREE_STACK_RETURN (REG_ERANGE);
2009 /* Look ahead to see if it's a range when the last thing
2010 was a character: if this is a hyphen not at the
2011 beginning or the end of a list, then it's the range
2014 && !(p - 2 >= pattern && p[-2] == '[')
2015 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2019 = compile_range (&p, pend, translate, syntax, b);
2020 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2023 else if (p[0] == '-' && p[1] != ']')
2024 { /* This handles ranges made up of characters only. */
2027 /* Move past the `-'. */
2030 ret = compile_range (&p, pend, translate, syntax, b);
2031 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2034 /* See if we're at the beginning of a possible character
2037 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2038 { /* Leave room for the null. */
2039 char str[CHAR_CLASS_MAX_LENGTH + 1];
2044 /* If pattern is `[[:'. */
2045 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2050 if (c == ':' || c == ']' || p == pend
2051 || c1 == CHAR_CLASS_MAX_LENGTH)
2057 /* If isn't a word bracketed by `[:' and:`]':
2058 undo the ending character, the letters, and leave
2059 the leading `:' and `[' (but set bits for them). */
2060 if (c == ':' && *p == ']')
2063 boolean is_alnum = STREQ (str, "alnum");
2064 boolean is_alpha = STREQ (str, "alpha");
2065 boolean is_blank = STREQ (str, "blank");
2066 boolean is_cntrl = STREQ (str, "cntrl");
2067 boolean is_digit = STREQ (str, "digit");
2068 boolean is_graph = STREQ (str, "graph");
2069 boolean is_lower = STREQ (str, "lower");
2070 boolean is_print = STREQ (str, "print");
2071 boolean is_punct = STREQ (str, "punct");
2072 boolean is_space = STREQ (str, "space");
2073 boolean is_upper = STREQ (str, "upper");
2074 boolean is_xdigit = STREQ (str, "xdigit");
2076 if (!IS_CHAR_CLASS (str))
2077 FREE_STACK_RETURN (REG_ECTYPE);
2079 /* Throw away the ] at the end of the character
2083 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2085 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2087 /* This was split into 3 if's to
2088 avoid an arbitrary limit in some compiler. */
2089 if ( (is_alnum && ISALNUM (ch))
2090 || (is_alpha && ISALPHA (ch))
2091 || (is_blank && ISBLANK (ch))
2092 || (is_cntrl && ISCNTRL (ch)))
2094 if ( (is_digit && ISDIGIT (ch))
2095 || (is_graph && ISGRAPH (ch))
2096 || (is_lower && ISLOWER (ch))
2097 || (is_print && ISPRINT (ch)))
2099 if ( (is_punct && ISPUNCT (ch))
2100 || (is_space && ISSPACE (ch))
2101 || (is_upper && ISUPPER (ch))
2102 || (is_xdigit && ISXDIGIT (ch)))
2105 had_char_class = true;
2114 had_char_class = false;
2119 had_char_class = false;
2124 /* Discard any (non)matching list bytes that are all 0 at the
2125 end of the map. Decrease the map-length byte too. */
2126 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2134 if (syntax & RE_NO_BK_PARENS)
2141 if (syntax & RE_NO_BK_PARENS)
2148 if (syntax & RE_NEWLINE_ALT)
2155 if (syntax & RE_NO_BK_VBAR)
2162 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2163 goto handle_interval;
2169 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2171 /* Do not translate the character after the \, so that we can
2172 distinguish, e.g., \B from \b, even if we normally would
2173 translate, e.g., B to b. */
2179 if (syntax & RE_NO_BK_PARENS)
2180 goto normal_backslash;
2186 if (COMPILE_STACK_FULL)
2188 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2189 compile_stack_elt_t);
2190 if (compile_stack.stack == NULL) return REG_ESPACE;
2192 compile_stack.size <<= 1;
2195 /* These are the values to restore when we hit end of this
2196 group. They are all relative offsets, so that if the
2197 whole pattern moves because of realloc, they will still
2199 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2200 COMPILE_STACK_TOP.fixup_alt_jump
2201 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2202 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2203 COMPILE_STACK_TOP.regnum = regnum;
2205 /* We will eventually replace the 0 with the number of
2206 groups inner to this one. But do not push a
2207 start_memory for groups beyond the last one we can
2208 represent in the compiled pattern. */
2209 if (regnum <= MAX_REGNUM)
2211 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2212 BUF_PUSH_3 (start_memory, regnum, 0);
2215 compile_stack.avail++;
2220 /* If we've reached MAX_REGNUM groups, then this open
2221 won't actually generate any code, so we'll have to
2222 clear pending_exact explicitly. */
2228 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2230 if (COMPILE_STACK_EMPTY)
2231 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2232 goto normal_backslash;
2234 FREE_STACK_RETURN (REG_ERPAREN);
2238 { /* Push a dummy failure point at the end of the
2239 alternative for a possible future
2240 `pop_failure_jump' to pop. See comments at
2241 `push_dummy_failure' in `re_match_2'. */
2242 BUF_PUSH (push_dummy_failure);
2244 /* We allocated space for this jump when we assigned
2245 to `fixup_alt_jump', in the `handle_alt' case below. */
2246 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2249 /* See similar code for backslashed left paren above. */
2250 if (COMPILE_STACK_EMPTY)
2251 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2254 FREE_STACK_RETURN (REG_ERPAREN);
2256 /* Since we just checked for an empty stack above, this
2257 ``can't happen''. */
2258 assert (compile_stack.avail != 0);
2260 /* We don't just want to restore into `regnum', because
2261 later groups should continue to be numbered higher,
2262 as in `(ab)c(de)' -- the second group is #2. */
2263 regnum_t this_group_regnum;
2265 compile_stack.avail--;
2266 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2268 = COMPILE_STACK_TOP.fixup_alt_jump
2269 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2271 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2272 this_group_regnum = COMPILE_STACK_TOP.regnum;
2273 /* If we've reached MAX_REGNUM groups, then this open
2274 won't actually generate any code, so we'll have to
2275 clear pending_exact explicitly. */
2278 /* We're at the end of the group, so now we know how many
2279 groups were inside this one. */
2280 if (this_group_regnum <= MAX_REGNUM)
2282 unsigned char *inner_group_loc
2283 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2285 *inner_group_loc = regnum - this_group_regnum;
2286 BUF_PUSH_3 (stop_memory, this_group_regnum,
2287 regnum - this_group_regnum);
2293 case '|': /* `\|'. */
2294 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2295 goto normal_backslash;
2297 if (syntax & RE_LIMITED_OPS)
2300 /* Insert before the previous alternative a jump which
2301 jumps to this alternative if the former fails. */
2302 GET_BUFFER_SPACE (3);
2303 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2307 /* The alternative before this one has a jump after it
2308 which gets executed if it gets matched. Adjust that
2309 jump so it will jump to this alternative's analogous
2310 jump (put in below, which in turn will jump to the next
2311 (if any) alternative's such jump, etc.). The last such
2312 jump jumps to the correct final destination. A picture:
2318 If we are at `b', then fixup_alt_jump right now points to a
2319 three-byte space after `a'. We'll put in the jump, set
2320 fixup_alt_jump to right after `b', and leave behind three
2321 bytes which we'll fill in when we get to after `c'. */
2324 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2326 /* Mark and leave space for a jump after this alternative,
2327 to be filled in later either by next alternative or
2328 when know we're at the end of a series of alternatives. */
2330 GET_BUFFER_SPACE (3);
2339 /* If \{ is a literal. */
2340 if (!(syntax & RE_INTERVALS)
2341 /* If we're at `\{' and it's not the open-interval
2343 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2344 || (p - 2 == pattern && p == pend))
2345 goto normal_backslash;
2349 /* If got here, then the syntax allows intervals. */
2351 /* At least (most) this many matches must be made. */
2352 int lower_bound = -1, upper_bound = -1;
2354 beg_interval = p - 1;
2358 if (syntax & RE_NO_BK_BRACES)
2359 goto unfetch_interval;
2361 FREE_STACK_RETURN (REG_EBRACE);
2364 GET_UNSIGNED_NUMBER (lower_bound);
2368 GET_UNSIGNED_NUMBER (upper_bound);
2369 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2372 /* Interval such as `{1}' => match exactly once. */
2373 upper_bound = lower_bound;
2375 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2376 || lower_bound > upper_bound)
2378 if (syntax & RE_NO_BK_BRACES)
2379 goto unfetch_interval;
2381 FREE_STACK_RETURN (REG_BADBR);
2384 if (!(syntax & RE_NO_BK_BRACES))
2386 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2393 if (syntax & RE_NO_BK_BRACES)
2394 goto unfetch_interval;
2396 FREE_STACK_RETURN (REG_BADBR);
2399 /* We just parsed a valid interval. */
2401 /* If it's invalid to have no preceding re. */
2404 if (syntax & RE_CONTEXT_INVALID_OPS)
2405 FREE_STACK_RETURN (REG_BADRPT);
2406 else if (syntax & RE_CONTEXT_INDEP_OPS)
2409 goto unfetch_interval;
2412 /* If the upper bound is zero, don't want to succeed at
2413 all; jump from `laststart' to `b + 3', which will be
2414 the end of the buffer after we insert the jump. */
2415 if (upper_bound == 0)
2417 GET_BUFFER_SPACE (3);
2418 INSERT_JUMP (jump, laststart, b + 3);
2422 /* Otherwise, we have a nontrivial interval. When
2423 we're all done, the pattern will look like:
2424 set_number_at <jump count> <upper bound>
2425 set_number_at <succeed_n count> <lower bound>
2426 succeed_n <after jump addr> <succeed_n count>
2428 jump_n <succeed_n addr> <jump count>
2429 (The upper bound and `jump_n' are omitted if
2430 `upper_bound' is 1, though.) */
2432 { /* If the upper bound is > 1, we need to insert
2433 more at the end of the loop. */
2434 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2436 GET_BUFFER_SPACE (nbytes);
2438 /* Initialize lower bound of the `succeed_n', even
2439 though it will be set during matching by its
2440 attendant `set_number_at' (inserted next),
2441 because `re_compile_fastmap' needs to know.
2442 Jump to the `jump_n' we might insert below. */
2443 INSERT_JUMP2 (succeed_n, laststart,
2444 b + 5 + (upper_bound > 1) * 5,
2448 /* Code to initialize the lower bound. Insert
2449 before the `succeed_n'. The `5' is the last two
2450 bytes of this `set_number_at', plus 3 bytes of
2451 the following `succeed_n'. */
2452 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2455 if (upper_bound > 1)
2456 { /* More than one repetition is allowed, so
2457 append a backward jump to the `succeed_n'
2458 that starts this interval.
2460 When we've reached this during matching,
2461 we'll have matched the interval once, so
2462 jump back only `upper_bound - 1' times. */
2463 STORE_JUMP2 (jump_n, b, laststart + 5,
2467 /* The location we want to set is the second
2468 parameter of the `jump_n'; that is `b-2' as
2469 an absolute address. `laststart' will be
2470 the `set_number_at' we're about to insert;
2471 `laststart+3' the number to set, the source
2472 for the relative address. But we are
2473 inserting into the middle of the pattern --
2474 so everything is getting moved up by 5.
2475 Conclusion: (b - 2) - (laststart + 3) + 5,
2476 i.e., b - laststart.
2478 We insert this at the beginning of the loop
2479 so that if we fail during matching, we'll
2480 reinitialize the bounds. */
2481 insert_op2 (set_number_at, laststart, b - laststart,
2482 upper_bound - 1, b);
2487 beg_interval = NULL;
2492 /* If an invalid interval, match the characters as literals. */
2493 assert (beg_interval);
2495 beg_interval = NULL;
2497 /* normal_char and normal_backslash need `c'. */
2500 if (!(syntax & RE_NO_BK_BRACES))
2502 if (p > pattern && p[-1] == '\\')
2503 goto normal_backslash;
2508 /* There is no way to specify the before_dot and after_dot
2509 operators. rms says this is ok. --karl */
2517 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2523 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2530 BUF_PUSH (wordchar);
2536 BUF_PUSH (notwordchar);
2549 BUF_PUSH (wordbound);
2553 BUF_PUSH (notwordbound);
2564 case '1': case '2': case '3': case '4': case '5':
2565 case '6': case '7': case '8': case '9':
2566 if (syntax & RE_NO_BK_REFS)
2572 FREE_STACK_RETURN (REG_ESUBREG);
2574 /* Can't back reference to a subexpression if inside of it. */
2575 if (group_in_compile_stack (compile_stack, c1))
2579 BUF_PUSH_2 (duplicate, c1);
2585 if (syntax & RE_BK_PLUS_QM)
2588 goto normal_backslash;
2592 /* You might think it would be useful for \ to mean
2593 not to translate; but if we don't translate it
2594 it will never match anything. */
2602 /* Expects the character in `c'. */
2604 /* If no exactn currently being built. */
2607 /* If last exactn not at current position. */
2608 || pending_exact + *pending_exact + 1 != b
2610 /* We have only one byte following the exactn for the count. */
2611 || *pending_exact == (1 << BYTEWIDTH) - 1
2613 /* If followed by a repetition operator. */
2614 || *p == '*' || *p == '^'
2615 || ((syntax & RE_BK_PLUS_QM)
2616 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2617 : (*p == '+' || *p == '?'))
2618 || ((syntax & RE_INTERVALS)
2619 && ((syntax & RE_NO_BK_BRACES)
2621 : (p[0] == '\\' && p[1] == '{'))))
2623 /* Start building a new exactn. */
2627 BUF_PUSH_2 (exactn, 0);
2628 pending_exact = b - 1;
2635 } /* while p != pend */
2638 /* Through the pattern now. */
2641 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2643 if (!COMPILE_STACK_EMPTY)
2644 FREE_STACK_RETURN (REG_EPAREN);
2646 /* If we don't want backtracking, force success
2647 the first time we reach the end of the compiled pattern. */
2648 if (syntax & RE_NO_POSIX_BACKTRACKING)
2651 free (compile_stack.stack);
2653 /* We have succeeded; set the length of the buffer. */
2654 bufp->used = b - bufp->buffer;
2659 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2660 print_compiled_pattern (bufp);
2664 #ifndef MATCH_MAY_ALLOCATE
2665 /* Initialize the failure stack to the largest possible stack. This
2666 isn't necessary unless we're trying to avoid calling alloca in
2667 the search and match routines. */
2669 int num_regs = bufp->re_nsub + 1;
2671 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2672 is strictly greater than re_max_failures, the largest possible stack
2673 is 2 * re_max_failures failure points. */
2674 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2676 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2679 if (! fail_stack.stack)
2681 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2682 * sizeof (fail_stack_elt_t));
2685 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2687 * sizeof (fail_stack_elt_t)));
2688 #else /* not emacs */
2689 if (! fail_stack.stack)
2691 = (fail_stack_elt_t *) malloc (fail_stack.size
2692 * sizeof (fail_stack_elt_t));
2695 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2697 * sizeof (fail_stack_elt_t)));
2698 #endif /* not emacs */
2701 regex_grow_registers (num_regs);
2703 #endif /* not MATCH_MAY_ALLOCATE */
2706 } /* regex_compile */
2708 /* Subroutines for `regex_compile'. */
2710 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2713 store_op1 (op, loc, arg)
2718 *loc = (unsigned char) op;
2719 STORE_NUMBER (loc + 1, arg);
2723 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2726 store_op2 (op, loc, arg1, arg2)
2731 *loc = (unsigned char) op;
2732 STORE_NUMBER (loc + 1, arg1);
2733 STORE_NUMBER (loc + 3, arg2);
2737 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
2738 for OP followed by two-byte integer parameter ARG. */
2741 insert_op1 (op, loc, arg, end)
2747 register unsigned char *pfrom = end;
2748 register unsigned char *pto = end + 3;
2750 while (pfrom != loc)
2753 store_op1 (op, loc, arg);
2757 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2760 insert_op2 (op, loc, arg1, arg2, end)
2766 register unsigned char *pfrom = end;
2767 register unsigned char *pto = end + 5;
2769 while (pfrom != loc)
2772 store_op2 (op, loc, arg1, arg2);
2776 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
2777 after an alternative or a begin-subexpression. We assume there is at
2778 least one character before the ^. */
2781 at_begline_loc_p (pattern, p, syntax)
2782 const char *pattern, *p;
2783 reg_syntax_t syntax;
2785 const char *prev = p - 2;
2786 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
2789 /* After a subexpression? */
2790 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
2791 /* After an alternative? */
2792 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
2796 /* The dual of at_begline_loc_p. This one is for $. We assume there is
2797 at least one character after the $, i.e., `P < PEND'. */
2800 at_endline_loc_p (p, pend, syntax)
2801 const char *p, *pend;
2804 const char *next = p;
2805 boolean next_backslash = *next == '\\';
2806 const char *next_next = p + 1 < pend ? p + 1 : 0;
2809 /* Before a subexpression? */
2810 (syntax & RE_NO_BK_PARENS ? *next == ')'
2811 : next_backslash && next_next && *next_next == ')')
2812 /* Before an alternative? */
2813 || (syntax & RE_NO_BK_VBAR ? *next == '|'
2814 : next_backslash && next_next && *next_next == '|');
2818 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
2819 false if it's not. */
2822 group_in_compile_stack (compile_stack, regnum)
2823 compile_stack_type compile_stack;
2828 for (this_element = compile_stack.avail - 1;
2831 if (compile_stack.stack[this_element].regnum == regnum)
2838 /* Read the ending character of a range (in a bracket expression) from the
2839 uncompiled pattern *P_PTR (which ends at PEND). We assume the
2840 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
2841 Then we set the translation of all bits between the starting and
2842 ending characters (inclusive) in the compiled pattern B.
2844 Return an error code.
2846 We use these short variable names so we can use the same macros as
2847 `regex_compile' itself. */
2849 static reg_errcode_t
2850 compile_range (p_ptr, pend, translate, syntax, b)
2851 const char **p_ptr, *pend;
2852 RE_TRANSLATE_TYPE translate;
2853 reg_syntax_t syntax;
2858 const char *p = *p_ptr;
2859 int range_start, range_end;
2864 /* Even though the pattern is a signed `char *', we need to fetch
2865 with unsigned char *'s; if the high bit of the pattern character
2866 is set, the range endpoints will be negative if we fetch using a
2869 We also want to fetch the endpoints without translating them; the
2870 appropriate translation is done in the bit-setting loop below. */
2871 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
2872 range_start = ((const unsigned char *) p)[-2];
2873 range_end = ((const unsigned char *) p)[0];
2875 /* Have to increment the pointer into the pattern string, so the
2876 caller isn't still at the ending character. */
2879 /* If the start is after the end, the range is empty. */
2880 if (range_start > range_end)
2881 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
2883 /* Here we see why `this_char' has to be larger than an `unsigned
2884 char' -- the range is inclusive, so if `range_end' == 0xff
2885 (assuming 8-bit characters), we would otherwise go into an infinite
2886 loop, since all characters <= 0xff. */
2887 for (this_char = range_start; this_char <= range_end; this_char++)
2889 SET_LIST_BIT (TRANSLATE (this_char));
2895 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
2896 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
2897 characters can start a string that matches the pattern. This fastmap
2898 is used by re_search to skip quickly over impossible starting points.
2900 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
2901 area as BUFP->fastmap.
2903 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
2906 Returns 0 if we succeed, -2 if an internal error. */
2909 re_compile_fastmap (bufp)
2910 struct re_pattern_buffer *bufp;
2913 #ifdef MATCH_MAY_ALLOCATE
2914 fail_stack_type fail_stack;
2916 #ifndef REGEX_MALLOC
2919 /* We don't push any register information onto the failure stack. */
2920 unsigned num_regs = 0;
2922 register char *fastmap = bufp->fastmap;
2923 unsigned char *pattern = bufp->buffer;
2924 unsigned long size = bufp->used;
2925 unsigned char *p = pattern;
2926 register unsigned char *pend = pattern + size;
2928 /* This holds the pointer to the failure stack, when
2929 it is allocated relocatably. */
2930 fail_stack_elt_t *failure_stack_ptr;
2932 /* Assume that each path through the pattern can be null until
2933 proven otherwise. We set this false at the bottom of switch
2934 statement, to which we get only if a particular path doesn't
2935 match the empty string. */
2936 boolean path_can_be_null = true;
2938 /* We aren't doing a `succeed_n' to begin with. */
2939 boolean succeed_n_p = false;
2941 assert (fastmap != NULL && p != NULL);
2944 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
2945 bufp->fastmap_accurate = 1; /* It will be when we're done. */
2946 bufp->can_be_null = 0;
2950 if (p == pend || *p == succeed)
2952 /* We have reached the (effective) end of pattern. */
2953 if (!FAIL_STACK_EMPTY ())
2955 bufp->can_be_null |= path_can_be_null;
2957 /* Reset for next path. */
2958 path_can_be_null = true;
2960 p = fail_stack.stack[--fail_stack.avail].pointer;
2968 /* We should never be about to go beyond the end of the pattern. */
2971 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
2974 /* I guess the idea here is to simply not bother with a fastmap
2975 if a backreference is used, since it's too hard to figure out
2976 the fastmap for the corresponding group. Setting
2977 `can_be_null' stops `re_search_2' from using the fastmap, so
2978 that is all we do. */
2980 bufp->can_be_null = 1;
2984 /* Following are the cases which match a character. These end
2993 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
2994 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3000 /* Chars beyond end of map must be allowed. */
3001 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3004 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3005 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3011 for (j = 0; j < (1 << BYTEWIDTH); j++)
3012 if (SYNTAX (j) == Sword)
3018 for (j = 0; j < (1 << BYTEWIDTH); j++)
3019 if (SYNTAX (j) != Sword)
3026 int fastmap_newline = fastmap['\n'];
3028 /* `.' matches anything ... */
3029 for (j = 0; j < (1 << BYTEWIDTH); j++)
3032 /* ... except perhaps newline. */
3033 if (!(bufp->syntax & RE_DOT_NEWLINE))
3034 fastmap['\n'] = fastmap_newline;
3036 /* Return if we have already set `can_be_null'; if we have,
3037 then the fastmap is irrelevant. Something's wrong here. */
3038 else if (bufp->can_be_null)
3041 /* Otherwise, have to check alternative paths. */
3048 for (j = 0; j < (1 << BYTEWIDTH); j++)
3049 if (SYNTAX (j) == (enum syntaxcode) k)
3056 for (j = 0; j < (1 << BYTEWIDTH); j++)
3057 if (SYNTAX (j) != (enum syntaxcode) k)
3062 /* All cases after this match the empty string. These end with
3082 case push_dummy_failure:
3087 case pop_failure_jump:
3088 case maybe_pop_jump:
3091 case dummy_failure_jump:
3092 EXTRACT_NUMBER_AND_INCR (j, p);
3097 /* Jump backward implies we just went through the body of a
3098 loop and matched nothing. Opcode jumped to should be
3099 `on_failure_jump' or `succeed_n'. Just treat it like an
3100 ordinary jump. For a * loop, it has pushed its failure
3101 point already; if so, discard that as redundant. */
3102 if ((re_opcode_t) *p != on_failure_jump
3103 && (re_opcode_t) *p != succeed_n)
3107 EXTRACT_NUMBER_AND_INCR (j, p);
3110 /* If what's on the stack is where we are now, pop it. */
3111 if (!FAIL_STACK_EMPTY ()
3112 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3118 case on_failure_jump:
3119 case on_failure_keep_string_jump:
3120 handle_on_failure_jump:
3121 EXTRACT_NUMBER_AND_INCR (j, p);
3123 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3124 end of the pattern. We don't want to push such a point,
3125 since when we restore it above, entering the switch will
3126 increment `p' past the end of the pattern. We don't need
3127 to push such a point since we obviously won't find any more
3128 fastmap entries beyond `pend'. Such a pattern can match
3129 the null string, though. */
3132 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3134 RESET_FAIL_STACK ();
3139 bufp->can_be_null = 1;
3143 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3144 succeed_n_p = false;
3151 /* Get to the number of times to succeed. */
3154 /* Increment p past the n for when k != 0. */
3155 EXTRACT_NUMBER_AND_INCR (k, p);
3159 succeed_n_p = true; /* Spaghetti code alert. */
3160 goto handle_on_failure_jump;
3177 abort (); /* We have listed all the cases. */
3180 /* Getting here means we have found the possible starting
3181 characters for one path of the pattern -- and that the empty
3182 string does not match. We need not follow this path further.
3183 Instead, look at the next alternative (remembered on the
3184 stack), or quit if no more. The test at the top of the loop
3185 does these things. */
3186 path_can_be_null = false;
3190 /* Set `can_be_null' for the last path (also the first path, if the
3191 pattern is empty). */
3192 bufp->can_be_null |= path_can_be_null;
3195 RESET_FAIL_STACK ();
3197 } /* re_compile_fastmap */
3199 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3200 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3201 this memory for recording register information. STARTS and ENDS
3202 must be allocated using the malloc library routine, and must each
3203 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3205 If NUM_REGS == 0, then subsequent matches should allocate their own
3208 Unless this function is called, the first search or match using
3209 PATTERN_BUFFER will allocate its own register data, without
3210 freeing the old data. */
3213 re_set_registers (bufp, regs, num_regs, starts, ends)
3214 struct re_pattern_buffer *bufp;
3215 struct re_registers *regs;
3217 regoff_t *starts, *ends;
3221 bufp->regs_allocated = REGS_REALLOCATE;
3222 regs->num_regs = num_regs;
3223 regs->start = starts;
3228 bufp->regs_allocated = REGS_UNALLOCATED;
3230 regs->start = regs->end = (regoff_t *) 0;
3234 /* Searching routines. */
3236 /* Like re_search_2, below, but only one string is specified, and
3237 doesn't let you say where to stop matching. */
3240 re_search (bufp, string, size, startpos, range, regs)
3241 struct re_pattern_buffer *bufp;
3243 int size, startpos, range;
3244 struct re_registers *regs;
3246 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3251 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3252 virtual concatenation of STRING1 and STRING2, starting first at index
3253 STARTPOS, then at STARTPOS + 1, and so on.
3255 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3257 RANGE is how far to scan while trying to match. RANGE = 0 means try
3258 only at STARTPOS; in general, the last start tried is STARTPOS +
3261 In REGS, return the indices of the virtual concatenation of STRING1
3262 and STRING2 that matched the entire BUFP->buffer and its contained
3265 Do not consider matching one past the index STOP in the virtual
3266 concatenation of STRING1 and STRING2.
3268 We return either the position in the strings at which the match was
3269 found, -1 if no match, or -2 if error (such as failure
3273 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3274 struct re_pattern_buffer *bufp;
3275 const char *string1, *string2;
3279 struct re_registers *regs;
3283 register char *fastmap = bufp->fastmap;
3284 register RE_TRANSLATE_TYPE translate = bufp->translate;
3285 int total_size = size1 + size2;
3286 int endpos = startpos + range;
3288 /* Check for out-of-range STARTPOS. */
3289 if (startpos < 0 || startpos > total_size)
3292 /* Fix up RANGE if it might eventually take us outside
3293 the virtual concatenation of STRING1 and STRING2.
3294 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3296 range = 0 - startpos;
3297 else if (endpos > total_size)
3298 range = total_size - startpos;
3300 /* If the search isn't to be a backwards one, don't waste time in a
3301 search for a pattern that must be anchored. */
3302 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3311 /* In a forward search for something that starts with \=.
3312 don't keep searching past point. */
3313 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3315 range = PT - startpos;
3321 /* Update the fastmap now if not correct already. */
3322 if (fastmap && !bufp->fastmap_accurate)
3323 if (re_compile_fastmap (bufp) == -2)
3326 /* Loop through the string, looking for a place to start matching. */
3329 /* If a fastmap is supplied, skip quickly over characters that
3330 cannot be the start of a match. If the pattern can match the
3331 null string, however, we don't need to skip characters; we want
3332 the first null string. */
3333 if (fastmap && startpos < total_size && !bufp->can_be_null)
3335 if (range > 0) /* Searching forwards. */
3337 register const char *d;
3338 register int lim = 0;
3341 if (startpos < size1 && startpos + range >= size1)
3342 lim = range - (size1 - startpos);
3344 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3346 /* Written out as an if-else to avoid testing `translate'
3350 && !fastmap[(unsigned char)
3351 translate[(unsigned char) *d++]])
3354 while (range > lim && !fastmap[(unsigned char) *d++])
3357 startpos += irange - range;
3359 else /* Searching backwards. */
3361 register char c = (size1 == 0 || startpos >= size1
3362 ? string2[startpos - size1]
3363 : string1[startpos]);
3365 if (!fastmap[(unsigned char) TRANSLATE (c)])
3370 /* If can't match the null string, and that's all we have left, fail. */
3371 if (range >= 0 && startpos == total_size && fastmap
3372 && !bufp->can_be_null)
3375 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3376 startpos, regs, stop);
3377 #ifndef REGEX_MALLOC
3406 /* Declarations and macros for re_match_2. */
3408 static int bcmp_translate ();
3409 static boolean alt_match_null_string_p (),
3410 common_op_match_null_string_p (),
3411 group_match_null_string_p ();
3413 /* This converts PTR, a pointer into one of the search strings `string1'
3414 and `string2' into an offset from the beginning of that string. */
3415 #define POINTER_TO_OFFSET(ptr) \
3416 (FIRST_STRING_P (ptr) \
3417 ? ((regoff_t) ((ptr) - string1)) \
3418 : ((regoff_t) ((ptr) - string2 + size1)))
3420 /* Macros for dealing with the split strings in re_match_2. */
3422 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3424 /* Call before fetching a character with *d. This switches over to
3425 string2 if necessary. */
3426 #define PREFETCH() \
3429 /* End of string2 => fail. */ \
3430 if (dend == end_match_2) \
3432 /* End of string1 => advance to string2. */ \
3434 dend = end_match_2; \
3438 /* Test if at very beginning or at very end of the virtual concatenation
3439 of `string1' and `string2'. If only one string, it's `string2'. */
3440 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3441 #define AT_STRINGS_END(d) ((d) == end2)
3444 /* Test if D points to a character which is word-constituent. We have
3445 two special cases to check for: if past the end of string1, look at
3446 the first character in string2; and if before the beginning of
3447 string2, look at the last character in string1. */
3448 #define WORDCHAR_P(d) \
3449 (SYNTAX ((d) == end1 ? *string2 \
3450 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3453 /* Disabled due to a compiler bug -- see comment at case wordbound */
3455 /* Test if the character before D and the one at D differ with respect
3456 to being word-constituent. */
3457 #define AT_WORD_BOUNDARY(d) \
3458 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3459 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3462 /* Free everything we malloc. */
3463 #ifdef MATCH_MAY_ALLOCATE
3464 #define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3465 #define FREE_VARIABLES() \
3467 REGEX_FREE_STACK (fail_stack.stack); \
3468 FREE_VAR (regstart); \
3469 FREE_VAR (regend); \
3470 FREE_VAR (old_regstart); \
3471 FREE_VAR (old_regend); \
3472 FREE_VAR (best_regstart); \
3473 FREE_VAR (best_regend); \
3474 FREE_VAR (reg_info); \
3475 FREE_VAR (reg_dummy); \
3476 FREE_VAR (reg_info_dummy); \
3479 #define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3480 #endif /* not MATCH_MAY_ALLOCATE */
3482 /* These values must meet several constraints. They must not be valid
3483 register values; since we have a limit of 255 registers (because
3484 we use only one byte in the pattern for the register number), we can
3485 use numbers larger than 255. They must differ by 1, because of
3486 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3487 be larger than the value for the highest register, so we do not try
3488 to actually save any registers when none are active. */
3489 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3490 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3492 /* Matching routines. */
3494 #ifndef emacs /* Emacs never uses this. */
3495 /* re_match is like re_match_2 except it takes only a single string. */
3498 re_match (bufp, string, size, pos, regs)
3499 struct re_pattern_buffer *bufp;
3502 struct re_registers *regs;
3504 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3509 #endif /* not emacs */
3512 /* re_match_2 matches the compiled pattern in BUFP against the
3513 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3514 and SIZE2, respectively). We start matching at POS, and stop
3517 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3518 store offsets for the substring each group matched in REGS. See the
3519 documentation for exactly how many groups we fill.
3521 We return -1 if no match, -2 if an internal error (such as the
3522 failure stack overflowing). Otherwise, we return the length of the
3523 matched substring. */
3526 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3527 struct re_pattern_buffer *bufp;
3528 const char *string1, *string2;
3531 struct re_registers *regs;
3534 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3540 /* This is a separate function so that we can force an alloca cleanup
3543 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3544 struct re_pattern_buffer *bufp;
3545 const char *string1, *string2;
3548 struct re_registers *regs;
3551 /* General temporaries. */
3555 /* Just past the end of the corresponding string. */
3556 const char *end1, *end2;
3558 /* Pointers into string1 and string2, just past the last characters in
3559 each to consider matching. */
3560 const char *end_match_1, *end_match_2;
3562 /* Where we are in the data, and the end of the current string. */
3563 const char *d, *dend;
3565 /* Where we are in the pattern, and the end of the pattern. */
3566 unsigned char *p = bufp->buffer;
3567 register unsigned char *pend = p + bufp->used;
3569 /* Mark the opcode just after a start_memory, so we can test for an
3570 empty subpattern when we get to the stop_memory. */
3571 unsigned char *just_past_start_mem = 0;
3573 /* We use this to map every character in the string. */
3574 RE_TRANSLATE_TYPE translate = bufp->translate;
3576 /* Failure point stack. Each place that can handle a failure further
3577 down the line pushes a failure point on this stack. It consists of
3578 restart, regend, and reg_info for all registers corresponding to
3579 the subexpressions we're currently inside, plus the number of such
3580 registers, and, finally, two char *'s. The first char * is where
3581 to resume scanning the pattern; the second one is where to resume
3582 scanning the strings. If the latter is zero, the failure point is
3583 a ``dummy''; if a failure happens and the failure point is a dummy,
3584 it gets discarded and the next next one is tried. */
3585 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3586 fail_stack_type fail_stack;
3589 static unsigned failure_id = 0;
3590 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3593 /* This holds the pointer to the failure stack, when
3594 it is allocated relocatably. */
3595 fail_stack_elt_t *failure_stack_ptr;
3597 /* We fill all the registers internally, independent of what we
3598 return, for use in backreferences. The number here includes
3599 an element for register zero. */
3600 unsigned num_regs = bufp->re_nsub + 1;
3602 /* The currently active registers. */
3603 unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3604 unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3606 /* Information on the contents of registers. These are pointers into
3607 the input strings; they record just what was matched (on this
3608 attempt) by a subexpression part of the pattern, that is, the
3609 regnum-th regstart pointer points to where in the pattern we began
3610 matching and the regnum-th regend points to right after where we
3611 stopped matching the regnum-th subexpression. (The zeroth register
3612 keeps track of what the whole pattern matches.) */
3613 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3614 const char **regstart, **regend;
3617 /* If a group that's operated upon by a repetition operator fails to
3618 match anything, then the register for its start will need to be
3619 restored because it will have been set to wherever in the string we
3620 are when we last see its open-group operator. Similarly for a
3622 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3623 const char **old_regstart, **old_regend;
3626 /* The is_active field of reg_info helps us keep track of which (possibly
3627 nested) subexpressions we are currently in. The matched_something
3628 field of reg_info[reg_num] helps us tell whether or not we have
3629 matched any of the pattern so far this time through the reg_num-th
3630 subexpression. These two fields get reset each time through any
3631 loop their register is in. */
3632 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3633 register_info_type *reg_info;
3636 /* The following record the register info as found in the above
3637 variables when we find a match better than any we've seen before.
3638 This happens as we backtrack through the failure points, which in
3639 turn happens only if we have not yet matched the entire string. */
3640 unsigned best_regs_set = false;
3641 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3642 const char **best_regstart, **best_regend;
3645 /* Logically, this is `best_regend[0]'. But we don't want to have to
3646 allocate space for that if we're not allocating space for anything
3647 else (see below). Also, we never need info about register 0 for
3648 any of the other register vectors, and it seems rather a kludge to
3649 treat `best_regend' differently than the rest. So we keep track of
3650 the end of the best match so far in a separate variable. We
3651 initialize this to NULL so that when we backtrack the first time
3652 and need to test it, it's not garbage. */
3653 const char *match_end = NULL;
3655 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3656 int set_regs_matched_done = 0;
3658 /* Used when we pop values we don't care about. */
3659 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3660 const char **reg_dummy;
3661 register_info_type *reg_info_dummy;
3665 /* Counts the total number of registers pushed. */
3666 unsigned num_regs_pushed = 0;
3669 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3673 #ifdef MATCH_MAY_ALLOCATE
3674 /* Do not bother to initialize all the register variables if there are
3675 no groups in the pattern, as it takes a fair amount of time. If
3676 there are groups, we include space for register 0 (the whole
3677 pattern), even though we never use it, since it simplifies the
3678 array indexing. We should fix this. */
3681 regstart = REGEX_TALLOC (num_regs, const char *);
3682 regend = REGEX_TALLOC (num_regs, const char *);
3683 old_regstart = REGEX_TALLOC (num_regs, const char *);
3684 old_regend = REGEX_TALLOC (num_regs, const char *);
3685 best_regstart = REGEX_TALLOC (num_regs, const char *);
3686 best_regend = REGEX_TALLOC (num_regs, const char *);
3687 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3688 reg_dummy = REGEX_TALLOC (num_regs, const char *);
3689 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3691 if (!(regstart && regend && old_regstart && old_regend && reg_info
3692 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3700 /* We must initialize all our variables to NULL, so that
3701 `FREE_VARIABLES' doesn't try to free them. */
3702 regstart = regend = old_regstart = old_regend = best_regstart
3703 = best_regend = reg_dummy = NULL;
3704 reg_info = reg_info_dummy = (register_info_type *) NULL;
3706 #endif /* MATCH_MAY_ALLOCATE */
3708 /* The starting position is bogus. */
3709 if (pos < 0 || pos > size1 + size2)
3715 /* Initialize subexpression text positions to -1 to mark ones that no
3716 start_memory/stop_memory has been seen for. Also initialize the
3717 register information struct. */
3718 for (mcnt = 1; mcnt < num_regs; mcnt++)
3720 regstart[mcnt] = regend[mcnt]
3721 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3723 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3724 IS_ACTIVE (reg_info[mcnt]) = 0;
3725 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3726 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3729 /* We move `string1' into `string2' if the latter's empty -- but not if
3730 `string1' is null. */
3731 if (size2 == 0 && string1 != NULL)
3738 end1 = string1 + size1;
3739 end2 = string2 + size2;
3741 /* Compute where to stop matching, within the two strings. */
3744 end_match_1 = string1 + stop;
3745 end_match_2 = string2;
3750 end_match_2 = string2 + stop - size1;
3753 /* `p' scans through the pattern as `d' scans through the data.
3754 `dend' is the end of the input string that `d' points within. `d'
3755 is advanced into the following input string whenever necessary, but
3756 this happens before fetching; therefore, at the beginning of the
3757 loop, `d' can be pointing at the end of a string, but it cannot
3759 if (size1 > 0 && pos <= size1)
3766 d = string2 + pos - size1;
3770 DEBUG_PRINT1 ("The compiled pattern is: ");
3771 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
3772 DEBUG_PRINT1 ("The string to match is: `");
3773 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
3774 DEBUG_PRINT1 ("'\n");
3776 /* This loops over pattern commands. It exits by returning from the
3777 function if the match is complete, or it drops through if the match
3778 fails at this starting point in the input data. */
3781 DEBUG_PRINT2 ("\n0x%x: ", p);
3784 { /* End of pattern means we might have succeeded. */
3785 DEBUG_PRINT1 ("end of pattern ... ");
3787 /* If we haven't matched the entire string, and we want the
3788 longest match, try backtracking. */
3789 if (d != end_match_2)
3791 /* 1 if this match ends in the same string (string1 or string2)
3792 as the best previous match. */
3793 boolean same_str_p = (FIRST_STRING_P (match_end)
3794 == MATCHING_IN_FIRST_STRING);
3795 /* 1 if this match is the best seen so far. */
3796 boolean best_match_p;
3798 /* AIX compiler got confused when this was combined
3799 with the previous declaration. */
3801 best_match_p = d > match_end;
3803 best_match_p = !MATCHING_IN_FIRST_STRING;
3805 DEBUG_PRINT1 ("backtracking.\n");
3807 if (!FAIL_STACK_EMPTY ())
3808 { /* More failure points to try. */
3810 /* If exceeds best match so far, save it. */
3811 if (!best_regs_set || best_match_p)
3813 best_regs_set = true;
3816 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
3818 for (mcnt = 1; mcnt < num_regs; mcnt++)
3820 best_regstart[mcnt] = regstart[mcnt];
3821 best_regend[mcnt] = regend[mcnt];
3827 /* If no failure points, don't restore garbage. And if
3828 last match is real best match, don't restore second
3830 else if (best_regs_set && !best_match_p)
3833 /* Restore best match. It may happen that `dend ==
3834 end_match_1' while the restored d is in string2.
3835 For example, the pattern `x.*y.*z' against the
3836 strings `x-' and `y-z-', if the two strings are
3837 not consecutive in memory. */
3838 DEBUG_PRINT1 ("Restoring best registers.\n");
3841 dend = ((d >= string1 && d <= end1)
3842 ? end_match_1 : end_match_2);
3844 for (mcnt = 1; mcnt < num_regs; mcnt++)
3846 regstart[mcnt] = best_regstart[mcnt];
3847 regend[mcnt] = best_regend[mcnt];
3850 } /* d != end_match_2 */
3853 DEBUG_PRINT1 ("Accepting match.\n");
3855 /* If caller wants register contents data back, do it. */
3856 if (regs && !bufp->no_sub)
3858 /* Have the register data arrays been allocated? */
3859 if (bufp->regs_allocated == REGS_UNALLOCATED)
3860 { /* No. So allocate them with malloc. We need one
3861 extra element beyond `num_regs' for the `-1' marker
3863 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
3864 regs->start = TALLOC (regs->num_regs, regoff_t);
3865 regs->end = TALLOC (regs->num_regs, regoff_t);
3866 if (regs->start == NULL || regs->end == NULL)
3871 bufp->regs_allocated = REGS_REALLOCATE;
3873 else if (bufp->regs_allocated == REGS_REALLOCATE)
3874 { /* Yes. If we need more elements than were already
3875 allocated, reallocate them. If we need fewer, just
3877 if (regs->num_regs < num_regs + 1)
3879 regs->num_regs = num_regs + 1;
3880 RETALLOC (regs->start, regs->num_regs, regoff_t);
3881 RETALLOC (regs->end, regs->num_regs, regoff_t);
3882 if (regs->start == NULL || regs->end == NULL)
3891 /* These braces fend off a "empty body in an else-statement"
3892 warning under GCC when assert expands to nothing. */
3893 assert (bufp->regs_allocated == REGS_FIXED);
3896 /* Convert the pointer data in `regstart' and `regend' to
3897 indices. Register zero has to be set differently,
3898 since we haven't kept track of any info for it. */
3899 if (regs->num_regs > 0)
3901 regs->start[0] = pos;
3902 regs->end[0] = (MATCHING_IN_FIRST_STRING
3903 ? ((regoff_t) (d - string1))
3904 : ((regoff_t) (d - string2 + size1)));
3907 /* Go through the first `min (num_regs, regs->num_regs)'
3908 registers, since that is all we initialized. */
3909 for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++)
3911 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
3912 regs->start[mcnt] = regs->end[mcnt] = -1;
3916 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
3918 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
3922 /* If the regs structure we return has more elements than
3923 were in the pattern, set the extra elements to -1. If
3924 we (re)allocated the registers, this is the case,
3925 because we always allocate enough to have at least one
3927 for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++)
3928 regs->start[mcnt] = regs->end[mcnt] = -1;
3929 } /* regs && !bufp->no_sub */
3931 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
3932 nfailure_points_pushed, nfailure_points_popped,
3933 nfailure_points_pushed - nfailure_points_popped);
3934 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
3936 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
3940 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
3946 /* Otherwise match next pattern command. */
3947 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3949 /* Ignore these. Used to ignore the n of succeed_n's which
3950 currently have n == 0. */
3952 DEBUG_PRINT1 ("EXECUTING no_op.\n");
3956 DEBUG_PRINT1 ("EXECUTING succeed.\n");
3959 /* Match the next n pattern characters exactly. The following
3960 byte in the pattern defines n, and the n bytes after that
3961 are the characters to match. */
3964 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
3966 /* This is written out as an if-else so we don't waste time
3967 testing `translate' inside the loop. */
3973 if ((unsigned char) translate[(unsigned char) *d++]
3974 != (unsigned char) *p++)
3984 if (*d++ != (char) *p++) goto fail;
3988 SET_REGS_MATCHED ();
3992 /* Match any character except possibly a newline or a null. */
3994 DEBUG_PRINT1 ("EXECUTING anychar.\n");
3998 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
3999 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4002 SET_REGS_MATCHED ();
4003 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4011 register unsigned char c;
4012 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4014 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4017 c = TRANSLATE (*d); /* The character to match. */
4019 /* Cast to `unsigned' instead of `unsigned char' in case the
4020 bit list is a full 32 bytes long. */
4021 if (c < (unsigned) (*p * BYTEWIDTH)
4022 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4027 if (!not) goto fail;
4029 SET_REGS_MATCHED ();
4035 /* The beginning of a group is represented by start_memory.
4036 The arguments are the register number in the next byte, and the
4037 number of groups inner to this one in the next. The text
4038 matched within the group is recorded (in the internal
4039 registers data structure) under the register number. */
4041 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4043 /* Find out if this group can match the empty string. */
4044 p1 = p; /* To send to group_match_null_string_p. */
4046 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4047 REG_MATCH_NULL_STRING_P (reg_info[*p])
4048 = group_match_null_string_p (&p1, pend, reg_info);
4050 /* Save the position in the string where we were the last time
4051 we were at this open-group operator in case the group is
4052 operated upon by a repetition operator, e.g., with `(a*)*b'
4053 against `ab'; then we want to ignore where we are now in
4054 the string in case this attempt to match fails. */
4055 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4056 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4058 DEBUG_PRINT2 (" old_regstart: %d\n",
4059 POINTER_TO_OFFSET (old_regstart[*p]));
4062 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4064 IS_ACTIVE (reg_info[*p]) = 1;
4065 MATCHED_SOMETHING (reg_info[*p]) = 0;
4067 /* Clear this whenever we change the register activity status. */
4068 set_regs_matched_done = 0;
4070 /* This is the new highest active register. */
4071 highest_active_reg = *p;
4073 /* If nothing was active before, this is the new lowest active
4075 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4076 lowest_active_reg = *p;
4078 /* Move past the register number and inner group count. */
4080 just_past_start_mem = p;
4085 /* The stop_memory opcode represents the end of a group. Its
4086 arguments are the same as start_memory's: the register
4087 number, and the number of inner groups. */
4089 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4091 /* We need to save the string position the last time we were at
4092 this close-group operator in case the group is operated
4093 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4094 against `aba'; then we want to ignore where we are now in
4095 the string in case this attempt to match fails. */
4096 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4097 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4099 DEBUG_PRINT2 (" old_regend: %d\n",
4100 POINTER_TO_OFFSET (old_regend[*p]));
4103 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4105 /* This register isn't active anymore. */
4106 IS_ACTIVE (reg_info[*p]) = 0;
4108 /* Clear this whenever we change the register activity status. */
4109 set_regs_matched_done = 0;
4111 /* If this was the only register active, nothing is active
4113 if (lowest_active_reg == highest_active_reg)
4115 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4116 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4119 { /* We must scan for the new highest active register, since
4120 it isn't necessarily one less than now: consider
4121 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4122 new highest active register is 1. */
4123 unsigned char r = *p - 1;
4124 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4127 /* If we end up at register zero, that means that we saved
4128 the registers as the result of an `on_failure_jump', not
4129 a `start_memory', and we jumped to past the innermost
4130 `stop_memory'. For example, in ((.)*) we save
4131 registers 1 and 2 as a result of the *, but when we pop
4132 back to the second ), we are at the stop_memory 1.
4133 Thus, nothing is active. */
4136 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4137 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4140 highest_active_reg = r;
4143 /* If just failed to match something this time around with a
4144 group that's operated on by a repetition operator, try to
4145 force exit from the ``loop'', and restore the register
4146 information for this group that we had before trying this
4148 if ((!MATCHED_SOMETHING (reg_info[*p])
4149 || just_past_start_mem == p - 1)
4152 boolean is_a_jump_n = false;
4156 switch ((re_opcode_t) *p1++)
4160 case pop_failure_jump:
4161 case maybe_pop_jump:
4163 case dummy_failure_jump:
4164 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4174 /* If the next operation is a jump backwards in the pattern
4175 to an on_failure_jump right before the start_memory
4176 corresponding to this stop_memory, exit from the loop
4177 by forcing a failure after pushing on the stack the
4178 on_failure_jump's jump in the pattern, and d. */
4179 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4180 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4182 /* If this group ever matched anything, then restore
4183 what its registers were before trying this last
4184 failed match, e.g., with `(a*)*b' against `ab' for
4185 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4186 against `aba' for regend[3].
4188 Also restore the registers for inner groups for,
4189 e.g., `((a*)(b*))*' against `aba' (register 3 would
4190 otherwise get trashed). */
4192 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4196 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4198 /* Restore this and inner groups' (if any) registers. */
4199 for (r = *p; r < *p + *(p + 1); r++)
4201 regstart[r] = old_regstart[r];
4203 /* xx why this test? */
4204 if (old_regend[r] >= regstart[r])
4205 regend[r] = old_regend[r];
4209 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4210 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4216 /* Move past the register number and the inner group count. */
4221 /* \<digit> has been turned into a `duplicate' command which is
4222 followed by the numeric value of <digit> as the register number. */
4225 register const char *d2, *dend2;
4226 int regno = *p++; /* Get which register to match against. */
4227 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4229 /* Can't back reference a group which we've never matched. */
4230 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4233 /* Where in input to try to start matching. */
4234 d2 = regstart[regno];
4236 /* Where to stop matching; if both the place to start and
4237 the place to stop matching are in the same string, then
4238 set to the place to stop, otherwise, for now have to use
4239 the end of the first string. */
4241 dend2 = ((FIRST_STRING_P (regstart[regno])
4242 == FIRST_STRING_P (regend[regno]))
4243 ? regend[regno] : end_match_1);
4246 /* If necessary, advance to next segment in register
4250 if (dend2 == end_match_2) break;
4251 if (dend2 == regend[regno]) break;
4253 /* End of string1 => advance to string2. */
4255 dend2 = regend[regno];
4257 /* At end of register contents => success */
4258 if (d2 == dend2) break;
4260 /* If necessary, advance to next segment in data. */
4263 /* How many characters left in this segment to match. */
4266 /* Want how many consecutive characters we can match in
4267 one shot, so, if necessary, adjust the count. */
4268 if (mcnt > dend2 - d2)
4271 /* Compare that many; failure if mismatch, else move
4274 ? bcmp_translate (d, d2, mcnt, translate)
4275 : bcmp (d, d2, mcnt))
4277 d += mcnt, d2 += mcnt;
4279 /* Do this because we've match some characters. */
4280 SET_REGS_MATCHED ();
4286 /* begline matches the empty string at the beginning of the string
4287 (unless `not_bol' is set in `bufp'), and, if
4288 `newline_anchor' is set, after newlines. */
4290 DEBUG_PRINT1 ("EXECUTING begline.\n");
4292 if (AT_STRINGS_BEG (d))
4294 if (!bufp->not_bol) break;
4296 else if (d[-1] == '\n' && bufp->newline_anchor)
4300 /* In all other cases, we fail. */
4304 /* endline is the dual of begline. */
4306 DEBUG_PRINT1 ("EXECUTING endline.\n");
4308 if (AT_STRINGS_END (d))
4310 if (!bufp->not_eol) break;
4313 /* We have to ``prefetch'' the next character. */
4314 else if ((d == end1 ? *string2 : *d) == '\n'
4315 && bufp->newline_anchor)
4322 /* Match at the very beginning of the data. */
4324 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4325 if (AT_STRINGS_BEG (d))
4330 /* Match at the very end of the data. */
4332 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4333 if (AT_STRINGS_END (d))
4338 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4339 pushes NULL as the value for the string on the stack. Then
4340 `pop_failure_point' will keep the current value for the
4341 string, instead of restoring it. To see why, consider
4342 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4343 then the . fails against the \n. But the next thing we want
4344 to do is match the \n against the \n; if we restored the
4345 string value, we would be back at the foo.
4347 Because this is used only in specific cases, we don't need to
4348 check all the things that `on_failure_jump' does, to make
4349 sure the right things get saved on the stack. Hence we don't
4350 share its code. The only reason to push anything on the
4351 stack at all is that otherwise we would have to change
4352 `anychar's code to do something besides goto fail in this
4353 case; that seems worse than this. */
4354 case on_failure_keep_string_jump:
4355 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4357 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4358 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4360 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4364 /* Uses of on_failure_jump:
4366 Each alternative starts with an on_failure_jump that points
4367 to the beginning of the next alternative. Each alternative
4368 except the last ends with a jump that in effect jumps past
4369 the rest of the alternatives. (They really jump to the
4370 ending jump of the following alternative, because tensioning
4371 these jumps is a hassle.)
4373 Repeats start with an on_failure_jump that points past both
4374 the repetition text and either the following jump or
4375 pop_failure_jump back to this on_failure_jump. */
4376 case on_failure_jump:
4378 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4380 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4381 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4383 /* If this on_failure_jump comes right before a group (i.e.,
4384 the original * applied to a group), save the information
4385 for that group and all inner ones, so that if we fail back
4386 to this point, the group's information will be correct.
4387 For example, in \(a*\)*\1, we need the preceding group,
4388 and in \(zz\(a*\)b*\)\2, we need the inner group. */
4390 /* We can't use `p' to check ahead because we push
4391 a failure point to `p + mcnt' after we do this. */
4394 /* We need to skip no_op's before we look for the
4395 start_memory in case this on_failure_jump is happening as
4396 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4398 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4401 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4403 /* We have a new highest active register now. This will
4404 get reset at the start_memory we are about to get to,
4405 but we will have saved all the registers relevant to
4406 this repetition op, as described above. */
4407 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4408 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4409 lowest_active_reg = *(p1 + 1);
4412 DEBUG_PRINT1 (":\n");
4413 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4417 /* A smart repeat ends with `maybe_pop_jump'.
4418 We change it to either `pop_failure_jump' or `jump'. */
4419 case maybe_pop_jump:
4420 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4421 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4423 register unsigned char *p2 = p;
4425 /* Compare the beginning of the repeat with what in the
4426 pattern follows its end. If we can establish that there
4427 is nothing that they would both match, i.e., that we
4428 would have to backtrack because of (as in, e.g., `a*a')
4429 then we can change to pop_failure_jump, because we'll
4430 never have to backtrack.
4432 This is not true in the case of alternatives: in
4433 `(a|ab)*' we do need to backtrack to the `ab' alternative
4434 (e.g., if the string was `ab'). But instead of trying to
4435 detect that here, the alternative has put on a dummy
4436 failure point which is what we will end up popping. */
4438 /* Skip over open/close-group commands.
4439 If what follows this loop is a ...+ construct,
4440 look at what begins its body, since we will have to
4441 match at least one of that. */
4445 && ((re_opcode_t) *p2 == stop_memory
4446 || (re_opcode_t) *p2 == start_memory))
4448 else if (p2 + 6 < pend
4449 && (re_opcode_t) *p2 == dummy_failure_jump)
4456 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4457 to the `maybe_finalize_jump' of this case. Examine what
4460 /* If we're at the end of the pattern, we can change. */
4463 /* Consider what happens when matching ":\(.*\)"
4464 against ":/". I don't really understand this code
4466 p[-3] = (unsigned char) pop_failure_jump;
4468 (" End of pattern: change to `pop_failure_jump'.\n");
4471 else if ((re_opcode_t) *p2 == exactn
4472 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4474 register unsigned char c
4475 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4477 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4479 p[-3] = (unsigned char) pop_failure_jump;
4480 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4484 else if ((re_opcode_t) p1[3] == charset
4485 || (re_opcode_t) p1[3] == charset_not)
4487 int not = (re_opcode_t) p1[3] == charset_not;
4489 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4490 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4493 /* `not' is equal to 1 if c would match, which means
4494 that we can't change to pop_failure_jump. */
4497 p[-3] = (unsigned char) pop_failure_jump;
4498 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4502 else if ((re_opcode_t) *p2 == charset)
4505 register unsigned char c
4506 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4509 if ((re_opcode_t) p1[3] == exactn
4510 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
4511 && (p2[2 + p1[5] / BYTEWIDTH]
4512 & (1 << (p1[5] % BYTEWIDTH)))))
4514 p[-3] = (unsigned char) pop_failure_jump;
4515 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4519 else if ((re_opcode_t) p1[3] == charset_not)
4522 /* We win if the charset_not inside the loop
4523 lists every character listed in the charset after. */
4524 for (idx = 0; idx < (int) p2[1]; idx++)
4525 if (! (p2[2 + idx] == 0
4526 || (idx < (int) p1[4]
4527 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4532 p[-3] = (unsigned char) pop_failure_jump;
4533 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4536 else if ((re_opcode_t) p1[3] == charset)
4539 /* We win if the charset inside the loop
4540 has no overlap with the one after the loop. */
4542 idx < (int) p2[1] && idx < (int) p1[4];
4544 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4547 if (idx == p2[1] || idx == p1[4])
4549 p[-3] = (unsigned char) pop_failure_jump;
4550 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4555 p -= 2; /* Point at relative address again. */
4556 if ((re_opcode_t) p[-1] != pop_failure_jump)
4558 p[-1] = (unsigned char) jump;
4559 DEBUG_PRINT1 (" Match => jump.\n");
4560 goto unconditional_jump;
4562 /* Note fall through. */
4565 /* The end of a simple repeat has a pop_failure_jump back to
4566 its matching on_failure_jump, where the latter will push a
4567 failure point. The pop_failure_jump takes off failure
4568 points put on by this pop_failure_jump's matching
4569 on_failure_jump; we got through the pattern to here from the
4570 matching on_failure_jump, so didn't fail. */
4571 case pop_failure_jump:
4573 /* We need to pass separate storage for the lowest and
4574 highest registers, even though we don't care about the
4575 actual values. Otherwise, we will restore only one
4576 register from the stack, since lowest will == highest in
4577 `pop_failure_point'. */
4578 unsigned dummy_low_reg, dummy_high_reg;
4579 unsigned char *pdummy;
4582 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4583 POP_FAILURE_POINT (sdummy, pdummy,
4584 dummy_low_reg, dummy_high_reg,
4585 reg_dummy, reg_dummy, reg_info_dummy);
4587 /* Note fall through. */
4590 /* Unconditionally jump (without popping any failure points). */
4593 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4594 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4595 p += mcnt; /* Do the jump. */
4596 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4600 /* We need this opcode so we can detect where alternatives end
4601 in `group_match_null_string_p' et al. */
4603 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4604 goto unconditional_jump;
4607 /* Normally, the on_failure_jump pushes a failure point, which
4608 then gets popped at pop_failure_jump. We will end up at
4609 pop_failure_jump, also, and with a pattern of, say, `a+', we
4610 are skipping over the on_failure_jump, so we have to push
4611 something meaningless for pop_failure_jump to pop. */
4612 case dummy_failure_jump:
4613 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4614 /* It doesn't matter what we push for the string here. What
4615 the code at `fail' tests is the value for the pattern. */
4616 PUSH_FAILURE_POINT (0, 0, -2);
4617 goto unconditional_jump;
4620 /* At the end of an alternative, we need to push a dummy failure
4621 point in case we are followed by a `pop_failure_jump', because
4622 we don't want the failure point for the alternative to be
4623 popped. For example, matching `(a|ab)*' against `aab'
4624 requires that we match the `ab' alternative. */
4625 case push_dummy_failure:
4626 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4627 /* See comments just above at `dummy_failure_jump' about the
4629 PUSH_FAILURE_POINT (0, 0, -2);
4632 /* Have to succeed matching what follows at least n times.
4633 After that, handle like `on_failure_jump'. */
4635 EXTRACT_NUMBER (mcnt, p + 2);
4636 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4639 /* Originally, this is how many times we HAVE to succeed. */
4644 STORE_NUMBER_AND_INCR (p, mcnt);
4645 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt);
4649 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4650 p[2] = (unsigned char) no_op;
4651 p[3] = (unsigned char) no_op;
4657 EXTRACT_NUMBER (mcnt, p + 2);
4658 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4660 /* Originally, this is how many times we CAN jump. */
4664 STORE_NUMBER (p + 2, mcnt);
4665 goto unconditional_jump;
4667 /* If don't have to jump any more, skip over the rest of command. */
4674 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4676 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4678 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4679 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4680 STORE_NUMBER (p1, mcnt);
4685 /* The DEC Alpha C compiler 3.x generates incorrect code for the
4686 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4687 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4688 macro and introducing temporary variables works around the bug. */
4691 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4692 if (AT_WORD_BOUNDARY (d))
4697 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4698 if (AT_WORD_BOUNDARY (d))
4704 boolean prevchar, thischar;
4706 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4707 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4710 prevchar = WORDCHAR_P (d - 1);
4711 thischar = WORDCHAR_P (d);
4712 if (prevchar != thischar)
4719 boolean prevchar, thischar;
4721 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4722 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4725 prevchar = WORDCHAR_P (d - 1);
4726 thischar = WORDCHAR_P (d);
4727 if (prevchar != thischar)
4734 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
4735 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
4740 DEBUG_PRINT1 ("EXECUTING wordend.\n");
4741 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
4742 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
4748 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
4749 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
4754 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
4755 if (PTR_CHAR_POS ((unsigned char *) d) != point)
4760 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
4761 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
4766 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
4771 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
4775 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
4777 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
4779 SET_REGS_MATCHED ();
4783 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
4785 goto matchnotsyntax;
4788 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
4792 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
4794 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
4796 SET_REGS_MATCHED ();
4799 #else /* not emacs */
4801 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
4803 if (!WORDCHAR_P (d))
4805 SET_REGS_MATCHED ();
4810 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
4814 SET_REGS_MATCHED ();
4817 #endif /* not emacs */
4822 continue; /* Successfully executed one pattern command; keep going. */
4825 /* We goto here if a matching operation fails. */
4827 if (!FAIL_STACK_EMPTY ())
4828 { /* A restart point is known. Restore to that state. */
4829 DEBUG_PRINT1 ("\nFAIL:\n");
4830 POP_FAILURE_POINT (d, p,
4831 lowest_active_reg, highest_active_reg,
4832 regstart, regend, reg_info);
4834 /* If this failure point is a dummy, try the next one. */
4838 /* If we failed to the end of the pattern, don't examine *p. */
4842 boolean is_a_jump_n = false;
4844 /* If failed to a backwards jump that's part of a repetition
4845 loop, need to pop this failure point and use the next one. */
4846 switch ((re_opcode_t) *p)
4850 case maybe_pop_jump:
4851 case pop_failure_jump:
4854 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4857 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
4859 && (re_opcode_t) *p1 == on_failure_jump))
4867 if (d >= string1 && d <= end1)
4871 break; /* Matching at this starting point really fails. */
4875 goto restore_best_regs;
4879 return -1; /* Failure to match. */
4882 /* Subroutine definitions for re_match_2. */
4885 /* We are passed P pointing to a register number after a start_memory.
4887 Return true if the pattern up to the corresponding stop_memory can
4888 match the empty string, and false otherwise.
4890 If we find the matching stop_memory, sets P to point to one past its number.
4891 Otherwise, sets P to an undefined byte less than or equal to END.
4893 We don't handle duplicates properly (yet). */
4896 group_match_null_string_p (p, end, reg_info)
4897 unsigned char **p, *end;
4898 register_info_type *reg_info;
4901 /* Point to after the args to the start_memory. */
4902 unsigned char *p1 = *p + 2;
4906 /* Skip over opcodes that can match nothing, and return true or
4907 false, as appropriate, when we get to one that can't, or to the
4908 matching stop_memory. */
4910 switch ((re_opcode_t) *p1)
4912 /* Could be either a loop or a series of alternatives. */
4913 case on_failure_jump:
4915 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4917 /* If the next operation is not a jump backwards in the
4922 /* Go through the on_failure_jumps of the alternatives,
4923 seeing if any of the alternatives cannot match nothing.
4924 The last alternative starts with only a jump,
4925 whereas the rest start with on_failure_jump and end
4926 with a jump, e.g., here is the pattern for `a|b|c':
4928 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
4929 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
4932 So, we have to first go through the first (n-1)
4933 alternatives and then deal with the last one separately. */
4936 /* Deal with the first (n-1) alternatives, which start
4937 with an on_failure_jump (see above) that jumps to right
4938 past a jump_past_alt. */
4940 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
4942 /* `mcnt' holds how many bytes long the alternative
4943 is, including the ending `jump_past_alt' and
4946 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
4950 /* Move to right after this alternative, including the
4954 /* Break if it's the beginning of an n-th alternative
4955 that doesn't begin with an on_failure_jump. */
4956 if ((re_opcode_t) *p1 != on_failure_jump)
4959 /* Still have to check that it's not an n-th
4960 alternative that starts with an on_failure_jump. */
4962 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4963 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
4965 /* Get to the beginning of the n-th alternative. */
4971 /* Deal with the last alternative: go back and get number
4972 of the `jump_past_alt' just before it. `mcnt' contains
4973 the length of the alternative. */
4974 EXTRACT_NUMBER (mcnt, p1 - 2);
4976 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
4979 p1 += mcnt; /* Get past the n-th alternative. */
4985 assert (p1[1] == **p);
4991 if (!common_op_match_null_string_p (&p1, end, reg_info))
4994 } /* while p1 < end */
4997 } /* group_match_null_string_p */
5000 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5001 It expects P to be the first byte of a single alternative and END one
5002 byte past the last. The alternative can contain groups. */
5005 alt_match_null_string_p (p, end, reg_info)
5006 unsigned char *p, *end;
5007 register_info_type *reg_info;
5010 unsigned char *p1 = p;
5014 /* Skip over opcodes that can match nothing, and break when we get
5015 to one that can't. */
5017 switch ((re_opcode_t) *p1)
5020 case on_failure_jump:
5022 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5027 if (!common_op_match_null_string_p (&p1, end, reg_info))
5030 } /* while p1 < end */
5033 } /* alt_match_null_string_p */
5036 /* Deals with the ops common to group_match_null_string_p and
5037 alt_match_null_string_p.
5039 Sets P to one after the op and its arguments, if any. */
5042 common_op_match_null_string_p (p, end, reg_info)
5043 unsigned char **p, *end;
5044 register_info_type *reg_info;
5049 unsigned char *p1 = *p;
5051 switch ((re_opcode_t) *p1++)
5071 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5072 ret = group_match_null_string_p (&p1, end, reg_info);
5074 /* Have to set this here in case we're checking a group which
5075 contains a group and a back reference to it. */
5077 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5078 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5084 /* If this is an optimized succeed_n for zero times, make the jump. */
5086 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5094 /* Get to the number of times to succeed. */
5096 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5101 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5109 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5117 /* All other opcodes mean we cannot match the empty string. */
5123 } /* common_op_match_null_string_p */
5126 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5127 bytes; nonzero otherwise. */
5130 bcmp_translate (s1, s2, len, translate)
5131 unsigned char *s1, *s2;
5133 RE_TRANSLATE_TYPE translate;
5135 register unsigned char *p1 = s1, *p2 = s2;
5138 if (translate[*p1++] != translate[*p2++]) return 1;
5144 /* Entry points for GNU code. */
5146 /* re_compile_pattern is the GNU regular expression compiler: it
5147 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5148 Returns 0 if the pattern was valid, otherwise an error string.
5150 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5151 are set in BUFP on entry.
5153 We call regex_compile to do the actual compilation. */
5156 re_compile_pattern (pattern, length, bufp)
5157 const char *pattern;
5159 struct re_pattern_buffer *bufp;
5163 /* GNU code is written to assume at least RE_NREGS registers will be set
5164 (and at least one extra will be -1). */
5165 bufp->regs_allocated = REGS_UNALLOCATED;
5167 /* And GNU code determines whether or not to get register information
5168 by passing null for the REGS argument to re_match, etc., not by
5172 /* Match anchors at newline. */
5173 bufp->newline_anchor = 1;
5175 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5179 return gettext (re_error_msgid[(int) ret]);
5182 /* Entry points compatible with 4.2 BSD regex library. We don't define
5183 them unless specifically requested. */
5185 #if defined (_REGEX_RE_COMP) || defined (_LIBC)
5187 /* BSD has one and only one pattern buffer. */
5188 static struct re_pattern_buffer re_comp_buf;
5192 /* Make these definitions weak in libc, so POSIX programs can redefine
5193 these names if they don't use our functions, and still use
5194 regcomp/regexec below without link errors. */
5204 if (!re_comp_buf.buffer)
5205 return gettext ("No previous regular expression");
5209 if (!re_comp_buf.buffer)
5211 re_comp_buf.buffer = (unsigned char *) malloc (200);
5212 if (re_comp_buf.buffer == NULL)
5213 return gettext (re_error_msgid[(int) REG_ESPACE]);
5214 re_comp_buf.allocated = 200;
5216 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5217 if (re_comp_buf.fastmap == NULL)
5218 return gettext (re_error_msgid[(int) REG_ESPACE]);
5221 /* Since `re_exec' always passes NULL for the `regs' argument, we
5222 don't need to initialize the pattern buffer fields which affect it. */
5224 /* Match anchors at newlines. */
5225 re_comp_buf.newline_anchor = 1;
5227 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5232 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5233 return (char *) gettext (re_error_msgid[(int) ret]);
5244 const int len = strlen (s);
5246 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5248 #endif /* _REGEX_RE_COMP */
5250 /* POSIX.2 functions. Don't define these for Emacs. */
5254 /* regcomp takes a regular expression as a string and compiles it.
5256 PREG is a regex_t *. We do not expect any fields to be initialized,
5257 since POSIX says we shouldn't. Thus, we set
5259 `buffer' to the compiled pattern;
5260 `used' to the length of the compiled pattern;
5261 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5262 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5263 RE_SYNTAX_POSIX_BASIC;
5264 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5265 `fastmap' and `fastmap_accurate' to zero;
5266 `re_nsub' to the number of subexpressions in PATTERN.
5268 PATTERN is the address of the pattern string.
5270 CFLAGS is a series of bits which affect compilation.
5272 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5273 use POSIX basic syntax.
5275 If REG_NEWLINE is set, then . and [^...] don't match newline.
5276 Also, regexec will try a match beginning after every newline.
5278 If REG_ICASE is set, then we considers upper- and lowercase
5279 versions of letters to be equivalent when matching.
5281 If REG_NOSUB is set, then when PREG is passed to regexec, that
5282 routine will report only success or failure, and nothing about the
5285 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
5286 the return codes and their meanings.) */
5289 regcomp (preg, pattern, cflags)
5291 const char *pattern;
5296 = (cflags & REG_EXTENDED) ?
5297 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5299 /* regex_compile will allocate the space for the compiled pattern. */
5301 preg->allocated = 0;
5304 /* Don't bother to use a fastmap when searching. This simplifies the
5305 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
5306 characters after newlines into the fastmap. This way, we just try
5310 if (cflags & REG_ICASE)
5315 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
5316 * sizeof (*(RE_TRANSLATE_TYPE)0));
5317 if (preg->translate == NULL)
5318 return (int) REG_ESPACE;
5320 /* Map uppercase characters to corresponding lowercase ones. */
5321 for (i = 0; i < CHAR_SET_SIZE; i++)
5322 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
5325 preg->translate = NULL;
5327 /* If REG_NEWLINE is set, newlines are treated differently. */
5328 if (cflags & REG_NEWLINE)
5329 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5330 syntax &= ~RE_DOT_NEWLINE;
5331 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5332 /* It also changes the matching behavior. */
5333 preg->newline_anchor = 1;
5336 preg->newline_anchor = 0;
5338 preg->no_sub = !!(cflags & REG_NOSUB);
5340 /* POSIX says a null character in the pattern terminates it, so we
5341 can use strlen here in compiling the pattern. */
5342 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5344 /* POSIX doesn't distinguish between an unmatched open-group and an
5345 unmatched close-group: both are REG_EPAREN. */
5346 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5352 /* regexec searches for a given pattern, specified by PREG, in the
5355 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5356 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5357 least NMATCH elements, and we set them to the offsets of the
5358 corresponding matched substrings.
5360 EFLAGS specifies `execution flags' which affect matching: if
5361 REG_NOTBOL is set, then ^ does not match at the beginning of the
5362 string; if REG_NOTEOL is set, then $ does not match at the end.
5364 We return 0 if we find a match and REG_NOMATCH if not. */
5367 regexec (preg, string, nmatch, pmatch, eflags)
5368 const regex_t *preg;
5371 regmatch_t pmatch[];
5375 struct re_registers regs;
5376 regex_t private_preg;
5377 int len = strlen (string);
5378 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5380 private_preg = *preg;
5382 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5383 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5385 /* The user has told us exactly how many registers to return
5386 information about, via `nmatch'. We have to pass that on to the
5387 matching routines. */
5388 private_preg.regs_allocated = REGS_FIXED;
5392 regs.num_regs = nmatch;
5393 regs.start = TALLOC (nmatch, regoff_t);
5394 regs.end = TALLOC (nmatch, regoff_t);
5395 if (regs.start == NULL || regs.end == NULL)
5396 return (int) REG_NOMATCH;
5399 /* Perform the searching operation. */
5400 ret = re_search (&private_preg, string, len,
5401 /* start: */ 0, /* range: */ len,
5402 want_reg_info ? ®s : (struct re_registers *) 0);
5404 /* Copy the register information to the POSIX structure. */
5411 for (r = 0; r < nmatch; r++)
5413 pmatch[r].rm_so = regs.start[r];
5414 pmatch[r].rm_eo = regs.end[r];
5418 /* If we needed the temporary register info, free the space now. */
5423 /* We want zero return to mean success, unlike `re_search'. */
5424 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5428 /* Returns a message corresponding to an error code, ERRCODE, returned
5429 from either regcomp or regexec. We don't use PREG here. */
5432 regerror (errcode, preg, errbuf, errbuf_size)
5434 const regex_t *preg;
5442 || errcode >= (sizeof (re_error_msgid) / sizeof (re_error_msgid[0])))
5443 /* Only error codes returned by the rest of the code should be passed
5444 to this routine. If we are given anything else, or if other regex
5445 code generates an invalid error code, then the program has a bug.
5446 Dump core so we can fix it. */
5449 msg = gettext (re_error_msgid[errcode]);
5451 msg_size = strlen (msg) + 1; /* Includes the null. */
5453 if (errbuf_size != 0)
5455 if (msg_size > errbuf_size)
5457 strncpy (errbuf, msg, errbuf_size - 1);
5458 errbuf[errbuf_size - 1] = 0;
5461 strcpy (errbuf, msg);
5468 /* Free dynamically allocated space used by PREG. */
5474 if (preg->buffer != NULL)
5475 free (preg->buffer);
5476 preg->buffer = NULL;
5478 preg->allocated = 0;
5481 if (preg->fastmap != NULL)
5482 free (preg->fastmap);
5483 preg->fastmap = NULL;
5484 preg->fastmap_accurate = 0;
5486 if (preg->translate != NULL)
5487 free (preg->translate);
5488 preg->translate = NULL;
5491 #endif /* not emacs */
projects / gnulib.git / blob