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 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 int re_max_failures = 20000;
1013 int re_max_failures = 2000;
1016 union fail_stack_elt
1018 unsigned char *pointer;
1022 typedef union fail_stack_elt fail_stack_elt_t;
1026 fail_stack_elt_t *stack;
1028 unsigned avail; /* Offset of next open position. */
1031 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1032 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1033 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1036 /* Define macros to initialize and free the failure stack.
1037 Do `return -2' if the alloc fails. */
1039 #ifdef MATCH_MAY_ALLOCATE
1040 #define INIT_FAIL_STACK() \
1042 fail_stack.stack = (fail_stack_elt_t *) \
1043 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1045 if (fail_stack.stack == NULL) \
1048 fail_stack.size = INIT_FAILURE_ALLOC; \
1049 fail_stack.avail = 0; \
1052 #define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1054 #define INIT_FAIL_STACK() \
1056 fail_stack.avail = 0; \
1059 #define RESET_FAIL_STACK()
1063 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1065 Return 1 if succeeds, and 0 if either ran out of memory
1066 allocating space for it or it was already too large.
1068 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1070 #define DOUBLE_FAIL_STACK(fail_stack) \
1071 ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \
1073 : ((fail_stack).stack = (fail_stack_elt_t *) \
1074 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1075 (fail_stack).size * sizeof (fail_stack_elt_t), \
1076 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1078 (fail_stack).stack == NULL \
1080 : ((fail_stack).size <<= 1, \
1084 /* Push pointer POINTER on FAIL_STACK.
1085 Return 1 if was able to do so and 0 if ran out of memory allocating
1087 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1088 ((FAIL_STACK_FULL () \
1089 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1091 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1094 /* Push a pointer value onto the failure stack.
1095 Assumes the variable `fail_stack'. Probably should only
1096 be called from within `PUSH_FAILURE_POINT'. */
1097 #define PUSH_FAILURE_POINTER(item) \
1098 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1100 /* This pushes an integer-valued item onto the failure stack.
1101 Assumes the variable `fail_stack'. Probably should only
1102 be called from within `PUSH_FAILURE_POINT'. */
1103 #define PUSH_FAILURE_INT(item) \
1104 fail_stack.stack[fail_stack.avail++].integer = (item)
1106 /* Push a fail_stack_elt_t value onto the failure stack.
1107 Assumes the variable `fail_stack'. Probably should only
1108 be called from within `PUSH_FAILURE_POINT'. */
1109 #define PUSH_FAILURE_ELT(item) \
1110 fail_stack.stack[fail_stack.avail++] = (item)
1112 /* These three POP... operations complement the three PUSH... operations.
1113 All assume that `fail_stack' is nonempty. */
1114 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1115 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1116 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1118 /* Used to omit pushing failure point id's when we're not debugging. */
1120 #define DEBUG_PUSH PUSH_FAILURE_INT
1121 #define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1123 #define DEBUG_PUSH(item)
1124 #define DEBUG_POP(item_addr)
1128 /* Push the information about the state we will need
1129 if we ever fail back to it.
1131 Requires variables fail_stack, regstart, regend, reg_info, and
1132 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be
1135 Does `return FAILURE_CODE' if runs out of memory. */
1137 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1139 char *destination; \
1140 /* Must be int, so when we don't save any registers, the arithmetic \
1141 of 0 + -1 isn't done as unsigned. */ \
1144 DEBUG_STATEMENT (failure_id++); \
1145 DEBUG_STATEMENT (nfailure_points_pushed++); \
1146 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1147 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1148 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1150 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \
1151 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1153 /* Ensure we have enough space allocated for what we will push. */ \
1154 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1156 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1157 return failure_code; \
1159 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1160 (fail_stack).size); \
1161 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1164 /* Push the info, starting with the registers. */ \
1165 DEBUG_PRINT1 ("\n"); \
1168 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1171 DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \
1172 DEBUG_STATEMENT (num_regs_pushed++); \
1174 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1175 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1177 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1178 PUSH_FAILURE_POINTER (regend[this_reg]); \
1180 DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \
1181 DEBUG_PRINT2 (" match_null=%d", \
1182 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1183 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1184 DEBUG_PRINT2 (" matched_something=%d", \
1185 MATCHED_SOMETHING (reg_info[this_reg])); \
1186 DEBUG_PRINT2 (" ever_matched=%d", \
1187 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1188 DEBUG_PRINT1 ("\n"); \
1189 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1192 DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\
1193 PUSH_FAILURE_INT (lowest_active_reg); \
1195 DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\
1196 PUSH_FAILURE_INT (highest_active_reg); \
1198 DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \
1199 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1200 PUSH_FAILURE_POINTER (pattern_place); \
1202 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \
1203 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1205 DEBUG_PRINT1 ("'\n"); \
1206 PUSH_FAILURE_POINTER (string_place); \
1208 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1209 DEBUG_PUSH (failure_id); \
1212 /* This is the number of items that are pushed and popped on the stack
1213 for each register. */
1214 #define NUM_REG_ITEMS 3
1216 /* Individual items aside from the registers. */
1218 #define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1220 #define NUM_NONREG_ITEMS 4
1223 /* We push at most this many items on the stack. */
1224 /* We used to use (num_regs - 1), which is the number of registers
1225 this regexp will save; but that was changed to 5
1226 to avoid stack overflow for a regexp with lots of parens. */
1227 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1229 /* We actually push this many items. */
1230 #define NUM_FAILURE_ITEMS \
1232 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1236 /* How many items can still be added to the stack without overflowing it. */
1237 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1240 /* Pops what PUSH_FAIL_STACK pushes.
1242 We restore into the parameters, all of which should be lvalues:
1243 STR -- the saved data position.
1244 PAT -- the saved pattern position.
1245 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1246 REGSTART, REGEND -- arrays of string positions.
1247 REG_INFO -- array of information about each subexpression.
1249 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1250 `pend', `string1', `size1', `string2', and `size2'. */
1252 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1254 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \
1256 const unsigned char *string_temp; \
1258 assert (!FAIL_STACK_EMPTY ()); \
1260 /* Remove failure points and point to how many regs pushed. */ \
1261 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1262 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1263 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1265 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1267 DEBUG_POP (&failure_id); \
1268 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1270 /* If the saved string location is NULL, it came from an \
1271 on_failure_keep_string_jump opcode, and we want to throw away the \
1272 saved NULL, thus retaining our current position in the string. */ \
1273 string_temp = POP_FAILURE_POINTER (); \
1274 if (string_temp != NULL) \
1275 str = (const char *) string_temp; \
1277 DEBUG_PRINT2 (" Popping string 0x%x: `", str); \
1278 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1279 DEBUG_PRINT1 ("'\n"); \
1281 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1282 DEBUG_PRINT2 (" Popping pattern 0x%x: ", pat); \
1283 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1285 /* Restore register info. */ \
1286 high_reg = (unsigned) POP_FAILURE_INT (); \
1287 DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \
1289 low_reg = (unsigned) POP_FAILURE_INT (); \
1290 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \
1293 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1295 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \
1297 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1298 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \
1300 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1301 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1303 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1304 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1308 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1310 reg_info[this_reg].word.integer = 0; \
1311 regend[this_reg] = 0; \
1312 regstart[this_reg] = 0; \
1314 highest_active_reg = high_reg; \
1317 set_regs_matched_done = 0; \
1318 DEBUG_STATEMENT (nfailure_points_popped++); \
1319 } /* POP_FAILURE_POINT */
1323 /* Structure for per-register (a.k.a. per-group) information.
1324 Other register information, such as the
1325 starting and ending positions (which are addresses), and the list of
1326 inner groups (which is a bits list) are maintained in separate
1329 We are making a (strictly speaking) nonportable assumption here: that
1330 the compiler will pack our bit fields into something that fits into
1331 the type of `word', i.e., is something that fits into one item on the
1336 fail_stack_elt_t word;
1339 /* This field is one if this group can match the empty string,
1340 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1341 #define MATCH_NULL_UNSET_VALUE 3
1342 unsigned match_null_string_p : 2;
1343 unsigned is_active : 1;
1344 unsigned matched_something : 1;
1345 unsigned ever_matched_something : 1;
1347 } register_info_type;
1349 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1350 #define IS_ACTIVE(R) ((R).bits.is_active)
1351 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1352 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1355 /* Call this when have matched a real character; it sets `matched' flags
1356 for the subexpressions which we are currently inside. Also records
1357 that those subexprs have matched. */
1358 #define SET_REGS_MATCHED() \
1361 if (!set_regs_matched_done) \
1364 set_regs_matched_done = 1; \
1365 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1367 MATCHED_SOMETHING (reg_info[r]) \
1368 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1375 /* Registers are set to a sentinel when they haven't yet matched. */
1376 static char reg_unset_dummy;
1377 #define REG_UNSET_VALUE (®_unset_dummy)
1378 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1380 /* Subroutine declarations and macros for regex_compile. */
1382 static void store_op1 (), store_op2 ();
1383 static void insert_op1 (), insert_op2 ();
1384 static boolean at_begline_loc_p (), at_endline_loc_p ();
1385 static boolean group_in_compile_stack ();
1386 static reg_errcode_t compile_range ();
1388 /* Fetch the next character in the uncompiled pattern---translating it
1389 if necessary. Also cast from a signed character in the constant
1390 string passed to us by the user to an unsigned char that we can use
1391 as an array index (in, e.g., `translate'). */
1393 #define PATFETCH(c) \
1394 do {if (p == pend) return REG_EEND; \
1395 c = (unsigned char) *p++; \
1396 if (translate) c = (unsigned char) translate[c]; \
1400 /* Fetch the next character in the uncompiled pattern, with no
1402 #define PATFETCH_RAW(c) \
1403 do {if (p == pend) return REG_EEND; \
1404 c = (unsigned char) *p++; \
1407 /* Go backwards one character in the pattern. */
1408 #define PATUNFETCH p--
1411 /* If `translate' is non-null, return translate[D], else just D. We
1412 cast the subscript to translate because some data is declared as
1413 `char *', to avoid warnings when a string constant is passed. But
1414 when we use a character as a subscript we must make it unsigned. */
1416 #define TRANSLATE(d) \
1417 (translate ? (char) translate[(unsigned char) (d)] : (d))
1421 /* Macros for outputting the compiled pattern into `buffer'. */
1423 /* If the buffer isn't allocated when it comes in, use this. */
1424 #define INIT_BUF_SIZE 32
1426 /* Make sure we have at least N more bytes of space in buffer. */
1427 #define GET_BUFFER_SPACE(n) \
1428 while (b - bufp->buffer + (n) > bufp->allocated) \
1431 /* Make sure we have one more byte of buffer space and then add C to it. */
1432 #define BUF_PUSH(c) \
1434 GET_BUFFER_SPACE (1); \
1435 *b++ = (unsigned char) (c); \
1439 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1440 #define BUF_PUSH_2(c1, c2) \
1442 GET_BUFFER_SPACE (2); \
1443 *b++ = (unsigned char) (c1); \
1444 *b++ = (unsigned char) (c2); \
1448 /* As with BUF_PUSH_2, except for three bytes. */
1449 #define BUF_PUSH_3(c1, c2, c3) \
1451 GET_BUFFER_SPACE (3); \
1452 *b++ = (unsigned char) (c1); \
1453 *b++ = (unsigned char) (c2); \
1454 *b++ = (unsigned char) (c3); \
1458 /* Store a jump with opcode OP at LOC to location TO. We store a
1459 relative address offset by the three bytes the jump itself occupies. */
1460 #define STORE_JUMP(op, loc, to) \
1461 store_op1 (op, loc, (to) - (loc) - 3)
1463 /* Likewise, for a two-argument jump. */
1464 #define STORE_JUMP2(op, loc, to, arg) \
1465 store_op2 (op, loc, (to) - (loc) - 3, arg)
1467 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1468 #define INSERT_JUMP(op, loc, to) \
1469 insert_op1 (op, loc, (to) - (loc) - 3, b)
1471 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1472 #define INSERT_JUMP2(op, loc, to, arg) \
1473 insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
1476 /* This is not an arbitrary limit: the arguments which represent offsets
1477 into the pattern are two bytes long. So if 2^16 bytes turns out to
1478 be too small, many things would have to change. */
1479 #define MAX_BUF_SIZE (1L << 16)
1482 /* Extend the buffer by twice its current size via realloc and
1483 reset the pointers that pointed into the old block to point to the
1484 correct places in the new one. If extending the buffer results in it
1485 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1486 #define EXTEND_BUFFER() \
1488 unsigned char *old_buffer = bufp->buffer; \
1489 if (bufp->allocated == MAX_BUF_SIZE) \
1491 bufp->allocated <<= 1; \
1492 if (bufp->allocated > MAX_BUF_SIZE) \
1493 bufp->allocated = MAX_BUF_SIZE; \
1494 bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\
1495 if (bufp->buffer == NULL) \
1496 return REG_ESPACE; \
1497 /* If the buffer moved, move all the pointers into it. */ \
1498 if (old_buffer != bufp->buffer) \
1500 b = (b - old_buffer) + bufp->buffer; \
1501 begalt = (begalt - old_buffer) + bufp->buffer; \
1502 if (fixup_alt_jump) \
1503 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1505 laststart = (laststart - old_buffer) + bufp->buffer; \
1506 if (pending_exact) \
1507 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1512 /* Since we have one byte reserved for the register number argument to
1513 {start,stop}_memory, the maximum number of groups we can report
1514 things about is what fits in that byte. */
1515 #define MAX_REGNUM 255
1517 /* But patterns can have more than `MAX_REGNUM' registers. We just
1518 ignore the excess. */
1519 typedef unsigned regnum_t;
1522 /* Macros for the compile stack. */
1524 /* Since offsets can go either forwards or backwards, this type needs to
1525 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1526 typedef int pattern_offset_t;
1530 pattern_offset_t begalt_offset;
1531 pattern_offset_t fixup_alt_jump;
1532 pattern_offset_t inner_group_offset;
1533 pattern_offset_t laststart_offset;
1535 } compile_stack_elt_t;
1540 compile_stack_elt_t *stack;
1542 unsigned avail; /* Offset of next open position. */
1543 } compile_stack_type;
1546 #define INIT_COMPILE_STACK_SIZE 32
1548 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1549 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1551 /* The next available element. */
1552 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1555 /* Set the bit for character C in a list. */
1556 #define SET_LIST_BIT(c) \
1557 (b[((unsigned char) (c)) / BYTEWIDTH] \
1558 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1561 /* Get the next unsigned number in the uncompiled pattern. */
1562 #define GET_UNSIGNED_NUMBER(num) \
1566 while (ISDIGIT (c)) \
1570 num = num * 10 + c - '0'; \
1578 #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1580 #define IS_CHAR_CLASS(string) \
1581 (STREQ (string, "alpha") || STREQ (string, "upper") \
1582 || STREQ (string, "lower") || STREQ (string, "digit") \
1583 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1584 || STREQ (string, "space") || STREQ (string, "print") \
1585 || STREQ (string, "punct") || STREQ (string, "graph") \
1586 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1588 #ifndef MATCH_MAY_ALLOCATE
1590 /* If we cannot allocate large objects within re_match_2_internal,
1591 we make the fail stack and register vectors global.
1592 The fail stack, we grow to the maximum size when a regexp
1594 The register vectors, we adjust in size each time we
1595 compile a regexp, according to the number of registers it needs. */
1597 static fail_stack_type fail_stack;
1599 /* Size with which the following vectors are currently allocated.
1600 That is so we can make them bigger as needed,
1601 but never make them smaller. */
1602 static int regs_allocated_size;
1604 static const char ** regstart, ** regend;
1605 static const char ** old_regstart, ** old_regend;
1606 static const char **best_regstart, **best_regend;
1607 static register_info_type *reg_info;
1608 static const char **reg_dummy;
1609 static register_info_type *reg_info_dummy;
1611 /* Make the register vectors big enough for NUM_REGS registers,
1612 but don't make them smaller. */
1615 regex_grow_registers (num_regs)
1618 if (num_regs > regs_allocated_size)
1620 RETALLOC_IF (regstart, num_regs, const char *);
1621 RETALLOC_IF (regend, num_regs, const char *);
1622 RETALLOC_IF (old_regstart, num_regs, const char *);
1623 RETALLOC_IF (old_regend, num_regs, const char *);
1624 RETALLOC_IF (best_regstart, num_regs, const char *);
1625 RETALLOC_IF (best_regend, num_regs, const char *);
1626 RETALLOC_IF (reg_info, num_regs, register_info_type);
1627 RETALLOC_IF (reg_dummy, num_regs, const char *);
1628 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1630 regs_allocated_size = num_regs;
1634 #endif /* not MATCH_MAY_ALLOCATE */
1636 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1637 Returns one of error codes defined in `regex.h', or zero for success.
1639 Assumes the `allocated' (and perhaps `buffer') and `translate'
1640 fields are set in BUFP on entry.
1642 If it succeeds, results are put in BUFP (if it returns an error, the
1643 contents of BUFP are undefined):
1644 `buffer' is the compiled pattern;
1645 `syntax' is set to SYNTAX;
1646 `used' is set to the length of the compiled pattern;
1647 `fastmap_accurate' is zero;
1648 `re_nsub' is the number of subexpressions in PATTERN;
1649 `not_bol' and `not_eol' are zero;
1651 The `fastmap' and `newline_anchor' fields are neither
1652 examined nor set. */
1654 /* Return, freeing storage we allocated. */
1655 #define FREE_STACK_RETURN(value) \
1656 return (free (compile_stack.stack), value)
1658 static reg_errcode_t
1659 regex_compile (pattern, size, syntax, bufp)
1660 const char *pattern;
1662 reg_syntax_t syntax;
1663 struct re_pattern_buffer *bufp;
1665 /* We fetch characters from PATTERN here. Even though PATTERN is
1666 `char *' (i.e., signed), we declare these variables as unsigned, so
1667 they can be reliably used as array indices. */
1668 register unsigned char c, c1;
1670 /* A random temporary spot in PATTERN. */
1673 /* Points to the end of the buffer, where we should append. */
1674 register unsigned char *b;
1676 /* Keeps track of unclosed groups. */
1677 compile_stack_type compile_stack;
1679 /* Points to the current (ending) position in the pattern. */
1680 const char *p = pattern;
1681 const char *pend = pattern + size;
1683 /* How to translate the characters in the pattern. */
1684 RE_TRANSLATE_TYPE translate = bufp->translate;
1686 /* Address of the count-byte of the most recently inserted `exactn'
1687 command. This makes it possible to tell if a new exact-match
1688 character can be added to that command or if the character requires
1689 a new `exactn' command. */
1690 unsigned char *pending_exact = 0;
1692 /* Address of start of the most recently finished expression.
1693 This tells, e.g., postfix * where to find the start of its
1694 operand. Reset at the beginning of groups and alternatives. */
1695 unsigned char *laststart = 0;
1697 /* Address of beginning of regexp, or inside of last group. */
1698 unsigned char *begalt;
1700 /* Place in the uncompiled pattern (i.e., the {) to
1701 which to go back if the interval is invalid. */
1702 const char *beg_interval;
1704 /* Address of the place where a forward jump should go to the end of
1705 the containing expression. Each alternative of an `or' -- except the
1706 last -- ends with a forward jump of this sort. */
1707 unsigned char *fixup_alt_jump = 0;
1709 /* Counts open-groups as they are encountered. Remembered for the
1710 matching close-group on the compile stack, so the same register
1711 number is put in the stop_memory as the start_memory. */
1712 regnum_t regnum = 0;
1715 DEBUG_PRINT1 ("\nCompiling pattern: ");
1718 unsigned debug_count;
1720 for (debug_count = 0; debug_count < size; debug_count++)
1721 putchar (pattern[debug_count]);
1726 /* Initialize the compile stack. */
1727 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1728 if (compile_stack.stack == NULL)
1731 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1732 compile_stack.avail = 0;
1734 /* Initialize the pattern buffer. */
1735 bufp->syntax = syntax;
1736 bufp->fastmap_accurate = 0;
1737 bufp->not_bol = bufp->not_eol = 0;
1739 /* Set `used' to zero, so that if we return an error, the pattern
1740 printer (for debugging) will think there's no pattern. We reset it
1744 /* Always count groups, whether or not bufp->no_sub is set. */
1747 #if !defined (emacs) && !defined (SYNTAX_TABLE)
1748 /* Initialize the syntax table. */
1749 init_syntax_once ();
1752 if (bufp->allocated == 0)
1755 { /* If zero allocated, but buffer is non-null, try to realloc
1756 enough space. This loses if buffer's address is bogus, but
1757 that is the user's responsibility. */
1758 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1761 { /* Caller did not allocate a buffer. Do it for them. */
1762 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1764 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1766 bufp->allocated = INIT_BUF_SIZE;
1769 begalt = b = bufp->buffer;
1771 /* Loop through the uncompiled pattern until we're at the end. */
1780 if ( /* If at start of pattern, it's an operator. */
1782 /* If context independent, it's an operator. */
1783 || syntax & RE_CONTEXT_INDEP_ANCHORS
1784 /* Otherwise, depends on what's come before. */
1785 || at_begline_loc_p (pattern, p, syntax))
1795 if ( /* If at end of pattern, it's an operator. */
1797 /* If context independent, it's an operator. */
1798 || syntax & RE_CONTEXT_INDEP_ANCHORS
1799 /* Otherwise, depends on what's next. */
1800 || at_endline_loc_p (p, pend, syntax))
1810 if ((syntax & RE_BK_PLUS_QM)
1811 || (syntax & RE_LIMITED_OPS))
1815 /* If there is no previous pattern... */
1818 if (syntax & RE_CONTEXT_INVALID_OPS)
1819 FREE_STACK_RETURN (REG_BADRPT);
1820 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1825 /* Are we optimizing this jump? */
1826 boolean keep_string_p = false;
1828 /* 1 means zero (many) matches is allowed. */
1829 char zero_times_ok = 0, many_times_ok = 0;
1831 /* If there is a sequence of repetition chars, collapse it
1832 down to just one (the right one). We can't combine
1833 interval operators with these because of, e.g., `a{2}*',
1834 which should only match an even number of `a's. */
1838 zero_times_ok |= c != '+';
1839 many_times_ok |= c != '?';
1847 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
1850 else if (syntax & RE_BK_PLUS_QM && c == '\\')
1852 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1855 if (!(c1 == '+' || c1 == '?'))
1870 /* If we get here, we found another repeat character. */
1873 /* Star, etc. applied to an empty pattern is equivalent
1874 to an empty pattern. */
1878 /* Now we know whether or not zero matches is allowed
1879 and also whether or not two or more matches is allowed. */
1881 { /* More than one repetition is allowed, so put in at the
1882 end a backward relative jump from `b' to before the next
1883 jump we're going to put in below (which jumps from
1884 laststart to after this jump).
1886 But if we are at the `*' in the exact sequence `.*\n',
1887 insert an unconditional jump backwards to the .,
1888 instead of the beginning of the loop. This way we only
1889 push a failure point once, instead of every time
1890 through the loop. */
1891 assert (p - 1 > pattern);
1893 /* Allocate the space for the jump. */
1894 GET_BUFFER_SPACE (3);
1896 /* We know we are not at the first character of the pattern,
1897 because laststart was nonzero. And we've already
1898 incremented `p', by the way, to be the character after
1899 the `*'. Do we have to do something analogous here
1900 for null bytes, because of RE_DOT_NOT_NULL? */
1901 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
1903 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
1904 && !(syntax & RE_DOT_NEWLINE))
1905 { /* We have .*\n. */
1906 STORE_JUMP (jump, b, laststart);
1907 keep_string_p = true;
1910 /* Anything else. */
1911 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
1913 /* We've added more stuff to the buffer. */
1917 /* On failure, jump from laststart to b + 3, which will be the
1918 end of the buffer after this jump is inserted. */
1919 GET_BUFFER_SPACE (3);
1920 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
1928 /* At least one repetition is required, so insert a
1929 `dummy_failure_jump' before the initial
1930 `on_failure_jump' instruction of the loop. This
1931 effects a skip over that instruction the first time
1932 we hit that loop. */
1933 GET_BUFFER_SPACE (3);
1934 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
1949 boolean had_char_class = false;
1951 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1953 /* Ensure that we have enough space to push a charset: the
1954 opcode, the length count, and the bitset; 34 bytes in all. */
1955 GET_BUFFER_SPACE (34);
1959 /* We test `*p == '^' twice, instead of using an if
1960 statement, so we only need one BUF_PUSH. */
1961 BUF_PUSH (*p == '^' ? charset_not : charset);
1965 /* Remember the first position in the bracket expression. */
1968 /* Push the number of bytes in the bitmap. */
1969 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
1971 /* Clear the whole map. */
1972 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
1974 /* charset_not matches newline according to a syntax bit. */
1975 if ((re_opcode_t) b[-2] == charset_not
1976 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
1977 SET_LIST_BIT ('\n');
1979 /* Read in characters and ranges, setting map bits. */
1982 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1986 /* \ might escape characters inside [...] and [^...]. */
1987 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
1989 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1996 /* Could be the end of the bracket expression. If it's
1997 not (i.e., when the bracket expression is `[]' so
1998 far), the ']' character bit gets set way below. */
1999 if (c == ']' && p != p1 + 1)
2002 /* Look ahead to see if it's a range when the last thing
2003 was a character class. */
2004 if (had_char_class && c == '-' && *p != ']')
2005 FREE_STACK_RETURN (REG_ERANGE);
2007 /* Look ahead to see if it's a range when the last thing
2008 was a character: if this is a hyphen not at the
2009 beginning or the end of a list, then it's the range
2012 && !(p - 2 >= pattern && p[-2] == '[')
2013 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2017 = compile_range (&p, pend, translate, syntax, b);
2018 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2021 else if (p[0] == '-' && p[1] != ']')
2022 { /* This handles ranges made up of characters only. */
2025 /* Move past the `-'. */
2028 ret = compile_range (&p, pend, translate, syntax, b);
2029 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2032 /* See if we're at the beginning of a possible character
2035 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2036 { /* Leave room for the null. */
2037 char str[CHAR_CLASS_MAX_LENGTH + 1];
2042 /* If pattern is `[[:'. */
2043 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2048 if (c == ':' || c == ']' || p == pend
2049 || c1 == CHAR_CLASS_MAX_LENGTH)
2055 /* If isn't a word bracketed by `[:' and:`]':
2056 undo the ending character, the letters, and leave
2057 the leading `:' and `[' (but set bits for them). */
2058 if (c == ':' && *p == ']')
2061 boolean is_alnum = STREQ (str, "alnum");
2062 boolean is_alpha = STREQ (str, "alpha");
2063 boolean is_blank = STREQ (str, "blank");
2064 boolean is_cntrl = STREQ (str, "cntrl");
2065 boolean is_digit = STREQ (str, "digit");
2066 boolean is_graph = STREQ (str, "graph");
2067 boolean is_lower = STREQ (str, "lower");
2068 boolean is_print = STREQ (str, "print");
2069 boolean is_punct = STREQ (str, "punct");
2070 boolean is_space = STREQ (str, "space");
2071 boolean is_upper = STREQ (str, "upper");
2072 boolean is_xdigit = STREQ (str, "xdigit");
2074 if (!IS_CHAR_CLASS (str))
2075 FREE_STACK_RETURN (REG_ECTYPE);
2077 /* Throw away the ] at the end of the character
2081 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2083 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2085 /* This was split into 3 if's to
2086 avoid an arbitrary limit in some compiler. */
2087 if ( (is_alnum && ISALNUM (ch))
2088 || (is_alpha && ISALPHA (ch))
2089 || (is_blank && ISBLANK (ch))
2090 || (is_cntrl && ISCNTRL (ch)))
2092 if ( (is_digit && ISDIGIT (ch))
2093 || (is_graph && ISGRAPH (ch))
2094 || (is_lower && ISLOWER (ch))
2095 || (is_print && ISPRINT (ch)))
2097 if ( (is_punct && ISPUNCT (ch))
2098 || (is_space && ISSPACE (ch))
2099 || (is_upper && ISUPPER (ch))
2100 || (is_xdigit && ISXDIGIT (ch)))
2103 had_char_class = true;
2112 had_char_class = false;
2117 had_char_class = false;
2122 /* Discard any (non)matching list bytes that are all 0 at the
2123 end of the map. Decrease the map-length byte too. */
2124 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2132 if (syntax & RE_NO_BK_PARENS)
2139 if (syntax & RE_NO_BK_PARENS)
2146 if (syntax & RE_NEWLINE_ALT)
2153 if (syntax & RE_NO_BK_VBAR)
2160 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2161 goto handle_interval;
2167 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2169 /* Do not translate the character after the \, so that we can
2170 distinguish, e.g., \B from \b, even if we normally would
2171 translate, e.g., B to b. */
2177 if (syntax & RE_NO_BK_PARENS)
2178 goto normal_backslash;
2184 if (COMPILE_STACK_FULL)
2186 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2187 compile_stack_elt_t);
2188 if (compile_stack.stack == NULL) return REG_ESPACE;
2190 compile_stack.size <<= 1;
2193 /* These are the values to restore when we hit end of this
2194 group. They are all relative offsets, so that if the
2195 whole pattern moves because of realloc, they will still
2197 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2198 COMPILE_STACK_TOP.fixup_alt_jump
2199 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2200 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2201 COMPILE_STACK_TOP.regnum = regnum;
2203 /* We will eventually replace the 0 with the number of
2204 groups inner to this one. But do not push a
2205 start_memory for groups beyond the last one we can
2206 represent in the compiled pattern. */
2207 if (regnum <= MAX_REGNUM)
2209 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2210 BUF_PUSH_3 (start_memory, regnum, 0);
2213 compile_stack.avail++;
2218 /* If we've reached MAX_REGNUM groups, then this open
2219 won't actually generate any code, so we'll have to
2220 clear pending_exact explicitly. */
2226 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2228 if (COMPILE_STACK_EMPTY)
2229 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2230 goto normal_backslash;
2232 FREE_STACK_RETURN (REG_ERPAREN);
2236 { /* Push a dummy failure point at the end of the
2237 alternative for a possible future
2238 `pop_failure_jump' to pop. See comments at
2239 `push_dummy_failure' in `re_match_2'. */
2240 BUF_PUSH (push_dummy_failure);
2242 /* We allocated space for this jump when we assigned
2243 to `fixup_alt_jump', in the `handle_alt' case below. */
2244 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2247 /* See similar code for backslashed left paren above. */
2248 if (COMPILE_STACK_EMPTY)
2249 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2252 FREE_STACK_RETURN (REG_ERPAREN);
2254 /* Since we just checked for an empty stack above, this
2255 ``can't happen''. */
2256 assert (compile_stack.avail != 0);
2258 /* We don't just want to restore into `regnum', because
2259 later groups should continue to be numbered higher,
2260 as in `(ab)c(de)' -- the second group is #2. */
2261 regnum_t this_group_regnum;
2263 compile_stack.avail--;
2264 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2266 = COMPILE_STACK_TOP.fixup_alt_jump
2267 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2269 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2270 this_group_regnum = COMPILE_STACK_TOP.regnum;
2271 /* If we've reached MAX_REGNUM groups, then this open
2272 won't actually generate any code, so we'll have to
2273 clear pending_exact explicitly. */
2276 /* We're at the end of the group, so now we know how many
2277 groups were inside this one. */
2278 if (this_group_regnum <= MAX_REGNUM)
2280 unsigned char *inner_group_loc
2281 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2283 *inner_group_loc = regnum - this_group_regnum;
2284 BUF_PUSH_3 (stop_memory, this_group_regnum,
2285 regnum - this_group_regnum);
2291 case '|': /* `\|'. */
2292 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2293 goto normal_backslash;
2295 if (syntax & RE_LIMITED_OPS)
2298 /* Insert before the previous alternative a jump which
2299 jumps to this alternative if the former fails. */
2300 GET_BUFFER_SPACE (3);
2301 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2305 /* The alternative before this one has a jump after it
2306 which gets executed if it gets matched. Adjust that
2307 jump so it will jump to this alternative's analogous
2308 jump (put in below, which in turn will jump to the next
2309 (if any) alternative's such jump, etc.). The last such
2310 jump jumps to the correct final destination. A picture:
2316 If we are at `b', then fixup_alt_jump right now points to a
2317 three-byte space after `a'. We'll put in the jump, set
2318 fixup_alt_jump to right after `b', and leave behind three
2319 bytes which we'll fill in when we get to after `c'. */
2322 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2324 /* Mark and leave space for a jump after this alternative,
2325 to be filled in later either by next alternative or
2326 when know we're at the end of a series of alternatives. */
2328 GET_BUFFER_SPACE (3);
2337 /* If \{ is a literal. */
2338 if (!(syntax & RE_INTERVALS)
2339 /* If we're at `\{' and it's not the open-interval
2341 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2342 || (p - 2 == pattern && p == pend))
2343 goto normal_backslash;
2347 /* If got here, then the syntax allows intervals. */
2349 /* At least (most) this many matches must be made. */
2350 int lower_bound = -1, upper_bound = -1;
2352 beg_interval = p - 1;
2356 if (syntax & RE_NO_BK_BRACES)
2357 goto unfetch_interval;
2359 FREE_STACK_RETURN (REG_EBRACE);
2362 GET_UNSIGNED_NUMBER (lower_bound);
2366 GET_UNSIGNED_NUMBER (upper_bound);
2367 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2370 /* Interval such as `{1}' => match exactly once. */
2371 upper_bound = lower_bound;
2373 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2374 || lower_bound > upper_bound)
2376 if (syntax & RE_NO_BK_BRACES)
2377 goto unfetch_interval;
2379 FREE_STACK_RETURN (REG_BADBR);
2382 if (!(syntax & RE_NO_BK_BRACES))
2384 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2391 if (syntax & RE_NO_BK_BRACES)
2392 goto unfetch_interval;
2394 FREE_STACK_RETURN (REG_BADBR);
2397 /* We just parsed a valid interval. */
2399 /* If it's invalid to have no preceding re. */
2402 if (syntax & RE_CONTEXT_INVALID_OPS)
2403 FREE_STACK_RETURN (REG_BADRPT);
2404 else if (syntax & RE_CONTEXT_INDEP_OPS)
2407 goto unfetch_interval;
2410 /* If the upper bound is zero, don't want to succeed at
2411 all; jump from `laststart' to `b + 3', which will be
2412 the end of the buffer after we insert the jump. */
2413 if (upper_bound == 0)
2415 GET_BUFFER_SPACE (3);
2416 INSERT_JUMP (jump, laststart, b + 3);
2420 /* Otherwise, we have a nontrivial interval. When
2421 we're all done, the pattern will look like:
2422 set_number_at <jump count> <upper bound>
2423 set_number_at <succeed_n count> <lower bound>
2424 succeed_n <after jump addr> <succeed_n count>
2426 jump_n <succeed_n addr> <jump count>
2427 (The upper bound and `jump_n' are omitted if
2428 `upper_bound' is 1, though.) */
2430 { /* If the upper bound is > 1, we need to insert
2431 more at the end of the loop. */
2432 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2434 GET_BUFFER_SPACE (nbytes);
2436 /* Initialize lower bound of the `succeed_n', even
2437 though it will be set during matching by its
2438 attendant `set_number_at' (inserted next),
2439 because `re_compile_fastmap' needs to know.
2440 Jump to the `jump_n' we might insert below. */
2441 INSERT_JUMP2 (succeed_n, laststart,
2442 b + 5 + (upper_bound > 1) * 5,
2446 /* Code to initialize the lower bound. Insert
2447 before the `succeed_n'. The `5' is the last two
2448 bytes of this `set_number_at', plus 3 bytes of
2449 the following `succeed_n'. */
2450 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2453 if (upper_bound > 1)
2454 { /* More than one repetition is allowed, so
2455 append a backward jump to the `succeed_n'
2456 that starts this interval.
2458 When we've reached this during matching,
2459 we'll have matched the interval once, so
2460 jump back only `upper_bound - 1' times. */
2461 STORE_JUMP2 (jump_n, b, laststart + 5,
2465 /* The location we want to set is the second
2466 parameter of the `jump_n'; that is `b-2' as
2467 an absolute address. `laststart' will be
2468 the `set_number_at' we're about to insert;
2469 `laststart+3' the number to set, the source
2470 for the relative address. But we are
2471 inserting into the middle of the pattern --
2472 so everything is getting moved up by 5.
2473 Conclusion: (b - 2) - (laststart + 3) + 5,
2474 i.e., b - laststart.
2476 We insert this at the beginning of the loop
2477 so that if we fail during matching, we'll
2478 reinitialize the bounds. */
2479 insert_op2 (set_number_at, laststart, b - laststart,
2480 upper_bound - 1, b);
2485 beg_interval = NULL;
2490 /* If an invalid interval, match the characters as literals. */
2491 assert (beg_interval);
2493 beg_interval = NULL;
2495 /* normal_char and normal_backslash need `c'. */
2498 if (!(syntax & RE_NO_BK_BRACES))
2500 if (p > pattern && p[-1] == '\\')
2501 goto normal_backslash;
2506 /* There is no way to specify the before_dot and after_dot
2507 operators. rms says this is ok. --karl */
2515 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2521 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2528 BUF_PUSH (wordchar);
2534 BUF_PUSH (notwordchar);
2547 BUF_PUSH (wordbound);
2551 BUF_PUSH (notwordbound);
2562 case '1': case '2': case '3': case '4': case '5':
2563 case '6': case '7': case '8': case '9':
2564 if (syntax & RE_NO_BK_REFS)
2570 FREE_STACK_RETURN (REG_ESUBREG);
2572 /* Can't back reference to a subexpression if inside of it. */
2573 if (group_in_compile_stack (compile_stack, c1))
2577 BUF_PUSH_2 (duplicate, c1);
2583 if (syntax & RE_BK_PLUS_QM)
2586 goto normal_backslash;
2590 /* You might think it would be useful for \ to mean
2591 not to translate; but if we don't translate it
2592 it will never match anything. */
2600 /* Expects the character in `c'. */
2602 /* If no exactn currently being built. */
2605 /* If last exactn not at current position. */
2606 || pending_exact + *pending_exact + 1 != b
2608 /* We have only one byte following the exactn for the count. */
2609 || *pending_exact == (1 << BYTEWIDTH) - 1
2611 /* If followed by a repetition operator. */
2612 || *p == '*' || *p == '^'
2613 || ((syntax & RE_BK_PLUS_QM)
2614 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2615 : (*p == '+' || *p == '?'))
2616 || ((syntax & RE_INTERVALS)
2617 && ((syntax & RE_NO_BK_BRACES)
2619 : (p[0] == '\\' && p[1] == '{'))))
2621 /* Start building a new exactn. */
2625 BUF_PUSH_2 (exactn, 0);
2626 pending_exact = b - 1;
2633 } /* while p != pend */
2636 /* Through the pattern now. */
2639 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2641 if (!COMPILE_STACK_EMPTY)
2642 FREE_STACK_RETURN (REG_EPAREN);
2644 /* If we don't want backtracking, force success
2645 the first time we reach the end of the compiled pattern. */
2646 if (syntax & RE_NO_POSIX_BACKTRACKING)
2649 free (compile_stack.stack);
2651 /* We have succeeded; set the length of the buffer. */
2652 bufp->used = b - bufp->buffer;
2657 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2658 print_compiled_pattern (bufp);
2662 #ifndef MATCH_MAY_ALLOCATE
2663 /* Initialize the failure stack to the largest possible stack. This
2664 isn't necessary unless we're trying to avoid calling alloca in
2665 the search and match routines. */
2667 int num_regs = bufp->re_nsub + 1;
2669 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2670 is strictly greater than re_max_failures, the largest possible stack
2671 is 2 * re_max_failures failure points. */
2672 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2674 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2677 if (! fail_stack.stack)
2679 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2680 * sizeof (fail_stack_elt_t));
2683 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2685 * sizeof (fail_stack_elt_t)));
2686 #else /* not emacs */
2687 if (! fail_stack.stack)
2689 = (fail_stack_elt_t *) malloc (fail_stack.size
2690 * sizeof (fail_stack_elt_t));
2693 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2695 * sizeof (fail_stack_elt_t)));
2696 #endif /* not emacs */
2699 regex_grow_registers (num_regs);
2701 #endif /* not MATCH_MAY_ALLOCATE */
2704 } /* regex_compile */
2706 /* Subroutines for `regex_compile'. */
2708 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2711 store_op1 (op, loc, arg)
2716 *loc = (unsigned char) op;
2717 STORE_NUMBER (loc + 1, arg);
2721 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2724 store_op2 (op, loc, arg1, arg2)
2729 *loc = (unsigned char) op;
2730 STORE_NUMBER (loc + 1, arg1);
2731 STORE_NUMBER (loc + 3, arg2);
2735 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
2736 for OP followed by two-byte integer parameter ARG. */
2739 insert_op1 (op, loc, arg, end)
2745 register unsigned char *pfrom = end;
2746 register unsigned char *pto = end + 3;
2748 while (pfrom != loc)
2751 store_op1 (op, loc, arg);
2755 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2758 insert_op2 (op, loc, arg1, arg2, end)
2764 register unsigned char *pfrom = end;
2765 register unsigned char *pto = end + 5;
2767 while (pfrom != loc)
2770 store_op2 (op, loc, arg1, arg2);
2774 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
2775 after an alternative or a begin-subexpression. We assume there is at
2776 least one character before the ^. */
2779 at_begline_loc_p (pattern, p, syntax)
2780 const char *pattern, *p;
2781 reg_syntax_t syntax;
2783 const char *prev = p - 2;
2784 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
2787 /* After a subexpression? */
2788 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
2789 /* After an alternative? */
2790 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
2794 /* The dual of at_begline_loc_p. This one is for $. We assume there is
2795 at least one character after the $, i.e., `P < PEND'. */
2798 at_endline_loc_p (p, pend, syntax)
2799 const char *p, *pend;
2802 const char *next = p;
2803 boolean next_backslash = *next == '\\';
2804 const char *next_next = p + 1 < pend ? p + 1 : 0;
2807 /* Before a subexpression? */
2808 (syntax & RE_NO_BK_PARENS ? *next == ')'
2809 : next_backslash && next_next && *next_next == ')')
2810 /* Before an alternative? */
2811 || (syntax & RE_NO_BK_VBAR ? *next == '|'
2812 : next_backslash && next_next && *next_next == '|');
2816 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
2817 false if it's not. */
2820 group_in_compile_stack (compile_stack, regnum)
2821 compile_stack_type compile_stack;
2826 for (this_element = compile_stack.avail - 1;
2829 if (compile_stack.stack[this_element].regnum == regnum)
2836 /* Read the ending character of a range (in a bracket expression) from the
2837 uncompiled pattern *P_PTR (which ends at PEND). We assume the
2838 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
2839 Then we set the translation of all bits between the starting and
2840 ending characters (inclusive) in the compiled pattern B.
2842 Return an error code.
2844 We use these short variable names so we can use the same macros as
2845 `regex_compile' itself. */
2847 static reg_errcode_t
2848 compile_range (p_ptr, pend, translate, syntax, b)
2849 const char **p_ptr, *pend;
2850 RE_TRANSLATE_TYPE translate;
2851 reg_syntax_t syntax;
2856 const char *p = *p_ptr;
2857 int range_start, range_end;
2862 /* Even though the pattern is a signed `char *', we need to fetch
2863 with unsigned char *'s; if the high bit of the pattern character
2864 is set, the range endpoints will be negative if we fetch using a
2867 We also want to fetch the endpoints without translating them; the
2868 appropriate translation is done in the bit-setting loop below. */
2869 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
2870 range_start = ((const unsigned char *) p)[-2];
2871 range_end = ((const unsigned char *) p)[0];
2873 /* Have to increment the pointer into the pattern string, so the
2874 caller isn't still at the ending character. */
2877 /* If the start is after the end, the range is empty. */
2878 if (range_start > range_end)
2879 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
2881 /* Here we see why `this_char' has to be larger than an `unsigned
2882 char' -- the range is inclusive, so if `range_end' == 0xff
2883 (assuming 8-bit characters), we would otherwise go into an infinite
2884 loop, since all characters <= 0xff. */
2885 for (this_char = range_start; this_char <= range_end; this_char++)
2887 SET_LIST_BIT (TRANSLATE (this_char));
2893 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
2894 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
2895 characters can start a string that matches the pattern. This fastmap
2896 is used by re_search to skip quickly over impossible starting points.
2898 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
2899 area as BUFP->fastmap.
2901 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
2904 Returns 0 if we succeed, -2 if an internal error. */
2907 re_compile_fastmap (bufp)
2908 struct re_pattern_buffer *bufp;
2911 #ifdef MATCH_MAY_ALLOCATE
2912 fail_stack_type fail_stack;
2914 #ifndef REGEX_MALLOC
2917 /* We don't push any register information onto the failure stack. */
2918 unsigned num_regs = 0;
2920 register char *fastmap = bufp->fastmap;
2921 unsigned char *pattern = bufp->buffer;
2922 unsigned long size = bufp->used;
2923 unsigned char *p = pattern;
2924 register unsigned char *pend = pattern + size;
2926 /* This holds the pointer to the failure stack, when
2927 it is allocated relocatably. */
2928 fail_stack_elt_t *failure_stack_ptr;
2930 /* Assume that each path through the pattern can be null until
2931 proven otherwise. We set this false at the bottom of switch
2932 statement, to which we get only if a particular path doesn't
2933 match the empty string. */
2934 boolean path_can_be_null = true;
2936 /* We aren't doing a `succeed_n' to begin with. */
2937 boolean succeed_n_p = false;
2939 assert (fastmap != NULL && p != NULL);
2942 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
2943 bufp->fastmap_accurate = 1; /* It will be when we're done. */
2944 bufp->can_be_null = 0;
2948 if (p == pend || *p == succeed)
2950 /* We have reached the (effective) end of pattern. */
2951 if (!FAIL_STACK_EMPTY ())
2953 bufp->can_be_null |= path_can_be_null;
2955 /* Reset for next path. */
2956 path_can_be_null = true;
2958 p = fail_stack.stack[--fail_stack.avail].pointer;
2966 /* We should never be about to go beyond the end of the pattern. */
2969 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
2972 /* I guess the idea here is to simply not bother with a fastmap
2973 if a backreference is used, since it's too hard to figure out
2974 the fastmap for the corresponding group. Setting
2975 `can_be_null' stops `re_search_2' from using the fastmap, so
2976 that is all we do. */
2978 bufp->can_be_null = 1;
2982 /* Following are the cases which match a character. These end
2991 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
2992 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
2998 /* Chars beyond end of map must be allowed. */
2999 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3002 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3003 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3009 for (j = 0; j < (1 << BYTEWIDTH); j++)
3010 if (SYNTAX (j) == Sword)
3016 for (j = 0; j < (1 << BYTEWIDTH); j++)
3017 if (SYNTAX (j) != Sword)
3024 int fastmap_newline = fastmap['\n'];
3026 /* `.' matches anything ... */
3027 for (j = 0; j < (1 << BYTEWIDTH); j++)
3030 /* ... except perhaps newline. */
3031 if (!(bufp->syntax & RE_DOT_NEWLINE))
3032 fastmap['\n'] = fastmap_newline;
3034 /* Return if we have already set `can_be_null'; if we have,
3035 then the fastmap is irrelevant. Something's wrong here. */
3036 else if (bufp->can_be_null)
3039 /* Otherwise, have to check alternative paths. */
3046 for (j = 0; j < (1 << BYTEWIDTH); j++)
3047 if (SYNTAX (j) == (enum syntaxcode) k)
3054 for (j = 0; j < (1 << BYTEWIDTH); j++)
3055 if (SYNTAX (j) != (enum syntaxcode) k)
3060 /* All cases after this match the empty string. These end with
3080 case push_dummy_failure:
3085 case pop_failure_jump:
3086 case maybe_pop_jump:
3089 case dummy_failure_jump:
3090 EXTRACT_NUMBER_AND_INCR (j, p);
3095 /* Jump backward implies we just went through the body of a
3096 loop and matched nothing. Opcode jumped to should be
3097 `on_failure_jump' or `succeed_n'. Just treat it like an
3098 ordinary jump. For a * loop, it has pushed its failure
3099 point already; if so, discard that as redundant. */
3100 if ((re_opcode_t) *p != on_failure_jump
3101 && (re_opcode_t) *p != succeed_n)
3105 EXTRACT_NUMBER_AND_INCR (j, p);
3108 /* If what's on the stack is where we are now, pop it. */
3109 if (!FAIL_STACK_EMPTY ()
3110 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3116 case on_failure_jump:
3117 case on_failure_keep_string_jump:
3118 handle_on_failure_jump:
3119 EXTRACT_NUMBER_AND_INCR (j, p);
3121 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3122 end of the pattern. We don't want to push such a point,
3123 since when we restore it above, entering the switch will
3124 increment `p' past the end of the pattern. We don't need
3125 to push such a point since we obviously won't find any more
3126 fastmap entries beyond `pend'. Such a pattern can match
3127 the null string, though. */
3130 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3132 RESET_FAIL_STACK ();
3137 bufp->can_be_null = 1;
3141 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3142 succeed_n_p = false;
3149 /* Get to the number of times to succeed. */
3152 /* Increment p past the n for when k != 0. */
3153 EXTRACT_NUMBER_AND_INCR (k, p);
3157 succeed_n_p = true; /* Spaghetti code alert. */
3158 goto handle_on_failure_jump;
3175 abort (); /* We have listed all the cases. */
3178 /* Getting here means we have found the possible starting
3179 characters for one path of the pattern -- and that the empty
3180 string does not match. We need not follow this path further.
3181 Instead, look at the next alternative (remembered on the
3182 stack), or quit if no more. The test at the top of the loop
3183 does these things. */
3184 path_can_be_null = false;
3188 /* Set `can_be_null' for the last path (also the first path, if the
3189 pattern is empty). */
3190 bufp->can_be_null |= path_can_be_null;
3193 RESET_FAIL_STACK ();
3195 } /* re_compile_fastmap */
3197 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3198 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3199 this memory for recording register information. STARTS and ENDS
3200 must be allocated using the malloc library routine, and must each
3201 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3203 If NUM_REGS == 0, then subsequent matches should allocate their own
3206 Unless this function is called, the first search or match using
3207 PATTERN_BUFFER will allocate its own register data, without
3208 freeing the old data. */
3211 re_set_registers (bufp, regs, num_regs, starts, ends)
3212 struct re_pattern_buffer *bufp;
3213 struct re_registers *regs;
3215 regoff_t *starts, *ends;
3219 bufp->regs_allocated = REGS_REALLOCATE;
3220 regs->num_regs = num_regs;
3221 regs->start = starts;
3226 bufp->regs_allocated = REGS_UNALLOCATED;
3228 regs->start = regs->end = (regoff_t *) 0;
3232 /* Searching routines. */
3234 /* Like re_search_2, below, but only one string is specified, and
3235 doesn't let you say where to stop matching. */
3238 re_search (bufp, string, size, startpos, range, regs)
3239 struct re_pattern_buffer *bufp;
3241 int size, startpos, range;
3242 struct re_registers *regs;
3244 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3249 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3250 virtual concatenation of STRING1 and STRING2, starting first at index
3251 STARTPOS, then at STARTPOS + 1, and so on.
3253 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3255 RANGE is how far to scan while trying to match. RANGE = 0 means try
3256 only at STARTPOS; in general, the last start tried is STARTPOS +
3259 In REGS, return the indices of the virtual concatenation of STRING1
3260 and STRING2 that matched the entire BUFP->buffer and its contained
3263 Do not consider matching one past the index STOP in the virtual
3264 concatenation of STRING1 and STRING2.
3266 We return either the position in the strings at which the match was
3267 found, -1 if no match, or -2 if error (such as failure
3271 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3272 struct re_pattern_buffer *bufp;
3273 const char *string1, *string2;
3277 struct re_registers *regs;
3281 register char *fastmap = bufp->fastmap;
3282 register RE_TRANSLATE_TYPE translate = bufp->translate;
3283 int total_size = size1 + size2;
3284 int endpos = startpos + range;
3286 /* Check for out-of-range STARTPOS. */
3287 if (startpos < 0 || startpos > total_size)
3290 /* Fix up RANGE if it might eventually take us outside
3291 the virtual concatenation of STRING1 and STRING2.
3292 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3294 range = 0 - startpos;
3295 else if (endpos > total_size)
3296 range = total_size - startpos;
3298 /* If the search isn't to be a backwards one, don't waste time in a
3299 search for a pattern that must be anchored. */
3300 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3309 /* In a forward search for something that starts with \=.
3310 don't keep searching past point. */
3311 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3313 range = PT - startpos;
3319 /* Update the fastmap now if not correct already. */
3320 if (fastmap && !bufp->fastmap_accurate)
3321 if (re_compile_fastmap (bufp) == -2)
3324 /* Loop through the string, looking for a place to start matching. */
3327 /* If a fastmap is supplied, skip quickly over characters that
3328 cannot be the start of a match. If the pattern can match the
3329 null string, however, we don't need to skip characters; we want
3330 the first null string. */
3331 if (fastmap && startpos < total_size && !bufp->can_be_null)
3333 if (range > 0) /* Searching forwards. */
3335 register const char *d;
3336 register int lim = 0;
3339 if (startpos < size1 && startpos + range >= size1)
3340 lim = range - (size1 - startpos);
3342 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3344 /* Written out as an if-else to avoid testing `translate'
3348 && !fastmap[(unsigned char)
3349 translate[(unsigned char) *d++]])
3352 while (range > lim && !fastmap[(unsigned char) *d++])
3355 startpos += irange - range;
3357 else /* Searching backwards. */
3359 register char c = (size1 == 0 || startpos >= size1
3360 ? string2[startpos - size1]
3361 : string1[startpos]);
3363 if (!fastmap[(unsigned char) TRANSLATE (c)])
3368 /* If can't match the null string, and that's all we have left, fail. */
3369 if (range >= 0 && startpos == total_size && fastmap
3370 && !bufp->can_be_null)
3373 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3374 startpos, regs, stop);
3375 #ifndef REGEX_MALLOC
3404 /* Declarations and macros for re_match_2. */
3406 static int bcmp_translate ();
3407 static boolean alt_match_null_string_p (),
3408 common_op_match_null_string_p (),
3409 group_match_null_string_p ();
3411 /* This converts PTR, a pointer into one of the search strings `string1'
3412 and `string2' into an offset from the beginning of that string. */
3413 #define POINTER_TO_OFFSET(ptr) \
3414 (FIRST_STRING_P (ptr) \
3415 ? ((regoff_t) ((ptr) - string1)) \
3416 : ((regoff_t) ((ptr) - string2 + size1)))
3418 /* Macros for dealing with the split strings in re_match_2. */
3420 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3422 /* Call before fetching a character with *d. This switches over to
3423 string2 if necessary. */
3424 #define PREFETCH() \
3427 /* End of string2 => fail. */ \
3428 if (dend == end_match_2) \
3430 /* End of string1 => advance to string2. */ \
3432 dend = end_match_2; \
3436 /* Test if at very beginning or at very end of the virtual concatenation
3437 of `string1' and `string2'. If only one string, it's `string2'. */
3438 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3439 #define AT_STRINGS_END(d) ((d) == end2)
3442 /* Test if D points to a character which is word-constituent. We have
3443 two special cases to check for: if past the end of string1, look at
3444 the first character in string2; and if before the beginning of
3445 string2, look at the last character in string1. */
3446 #define WORDCHAR_P(d) \
3447 (SYNTAX ((d) == end1 ? *string2 \
3448 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3451 /* Disabled due to a compiler bug -- see comment at case wordbound */
3453 /* Test if the character before D and the one at D differ with respect
3454 to being word-constituent. */
3455 #define AT_WORD_BOUNDARY(d) \
3456 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3457 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3460 /* Free everything we malloc. */
3461 #ifdef MATCH_MAY_ALLOCATE
3462 #define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3463 #define FREE_VARIABLES() \
3465 REGEX_FREE_STACK (fail_stack.stack); \
3466 FREE_VAR (regstart); \
3467 FREE_VAR (regend); \
3468 FREE_VAR (old_regstart); \
3469 FREE_VAR (old_regend); \
3470 FREE_VAR (best_regstart); \
3471 FREE_VAR (best_regend); \
3472 FREE_VAR (reg_info); \
3473 FREE_VAR (reg_dummy); \
3474 FREE_VAR (reg_info_dummy); \
3477 #define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3478 #endif /* not MATCH_MAY_ALLOCATE */
3480 /* These values must meet several constraints. They must not be valid
3481 register values; since we have a limit of 255 registers (because
3482 we use only one byte in the pattern for the register number), we can
3483 use numbers larger than 255. They must differ by 1, because of
3484 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3485 be larger than the value for the highest register, so we do not try
3486 to actually save any registers when none are active. */
3487 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3488 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3490 /* Matching routines. */
3492 #ifndef emacs /* Emacs never uses this. */
3493 /* re_match is like re_match_2 except it takes only a single string. */
3496 re_match (bufp, string, size, pos, regs)
3497 struct re_pattern_buffer *bufp;
3500 struct re_registers *regs;
3502 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3507 #endif /* not emacs */
3510 /* re_match_2 matches the compiled pattern in BUFP against the
3511 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3512 and SIZE2, respectively). We start matching at POS, and stop
3515 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3516 store offsets for the substring each group matched in REGS. See the
3517 documentation for exactly how many groups we fill.
3519 We return -1 if no match, -2 if an internal error (such as the
3520 failure stack overflowing). Otherwise, we return the length of the
3521 matched substring. */
3524 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3525 struct re_pattern_buffer *bufp;
3526 const char *string1, *string2;
3529 struct re_registers *regs;
3532 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3538 /* This is a separate function so that we can force an alloca cleanup
3541 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3542 struct re_pattern_buffer *bufp;
3543 const char *string1, *string2;
3546 struct re_registers *regs;
3549 /* General temporaries. */
3553 /* Just past the end of the corresponding string. */
3554 const char *end1, *end2;
3556 /* Pointers into string1 and string2, just past the last characters in
3557 each to consider matching. */
3558 const char *end_match_1, *end_match_2;
3560 /* Where we are in the data, and the end of the current string. */
3561 const char *d, *dend;
3563 /* Where we are in the pattern, and the end of the pattern. */
3564 unsigned char *p = bufp->buffer;
3565 register unsigned char *pend = p + bufp->used;
3567 /* Mark the opcode just after a start_memory, so we can test for an
3568 empty subpattern when we get to the stop_memory. */
3569 unsigned char *just_past_start_mem = 0;
3571 /* We use this to map every character in the string. */
3572 RE_TRANSLATE_TYPE translate = bufp->translate;
3574 /* Failure point stack. Each place that can handle a failure further
3575 down the line pushes a failure point on this stack. It consists of
3576 restart, regend, and reg_info for all registers corresponding to
3577 the subexpressions we're currently inside, plus the number of such
3578 registers, and, finally, two char *'s. The first char * is where
3579 to resume scanning the pattern; the second one is where to resume
3580 scanning the strings. If the latter is zero, the failure point is
3581 a ``dummy''; if a failure happens and the failure point is a dummy,
3582 it gets discarded and the next next one is tried. */
3583 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3584 fail_stack_type fail_stack;
3587 static unsigned failure_id = 0;
3588 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3591 /* This holds the pointer to the failure stack, when
3592 it is allocated relocatably. */
3593 fail_stack_elt_t *failure_stack_ptr;
3595 /* We fill all the registers internally, independent of what we
3596 return, for use in backreferences. The number here includes
3597 an element for register zero. */
3598 unsigned num_regs = bufp->re_nsub + 1;
3600 /* The currently active registers. */
3601 unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3602 unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3604 /* Information on the contents of registers. These are pointers into
3605 the input strings; they record just what was matched (on this
3606 attempt) by a subexpression part of the pattern, that is, the
3607 regnum-th regstart pointer points to where in the pattern we began
3608 matching and the regnum-th regend points to right after where we
3609 stopped matching the regnum-th subexpression. (The zeroth register
3610 keeps track of what the whole pattern matches.) */
3611 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3612 const char **regstart, **regend;
3615 /* If a group that's operated upon by a repetition operator fails to
3616 match anything, then the register for its start will need to be
3617 restored because it will have been set to wherever in the string we
3618 are when we last see its open-group operator. Similarly for a
3620 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3621 const char **old_regstart, **old_regend;
3624 /* The is_active field of reg_info helps us keep track of which (possibly
3625 nested) subexpressions we are currently in. The matched_something
3626 field of reg_info[reg_num] helps us tell whether or not we have
3627 matched any of the pattern so far this time through the reg_num-th
3628 subexpression. These two fields get reset each time through any
3629 loop their register is in. */
3630 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3631 register_info_type *reg_info;
3634 /* The following record the register info as found in the above
3635 variables when we find a match better than any we've seen before.
3636 This happens as we backtrack through the failure points, which in
3637 turn happens only if we have not yet matched the entire string. */
3638 unsigned best_regs_set = false;
3639 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3640 const char **best_regstart, **best_regend;
3643 /* Logically, this is `best_regend[0]'. But we don't want to have to
3644 allocate space for that if we're not allocating space for anything
3645 else (see below). Also, we never need info about register 0 for
3646 any of the other register vectors, and it seems rather a kludge to
3647 treat `best_regend' differently than the rest. So we keep track of
3648 the end of the best match so far in a separate variable. We
3649 initialize this to NULL so that when we backtrack the first time
3650 and need to test it, it's not garbage. */
3651 const char *match_end = NULL;
3653 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3654 int set_regs_matched_done = 0;
3656 /* Used when we pop values we don't care about. */
3657 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3658 const char **reg_dummy;
3659 register_info_type *reg_info_dummy;
3663 /* Counts the total number of registers pushed. */
3664 unsigned num_regs_pushed = 0;
3667 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3671 #ifdef MATCH_MAY_ALLOCATE
3672 /* Do not bother to initialize all the register variables if there are
3673 no groups in the pattern, as it takes a fair amount of time. If
3674 there are groups, we include space for register 0 (the whole
3675 pattern), even though we never use it, since it simplifies the
3676 array indexing. We should fix this. */
3679 regstart = REGEX_TALLOC (num_regs, const char *);
3680 regend = REGEX_TALLOC (num_regs, const char *);
3681 old_regstart = REGEX_TALLOC (num_regs, const char *);
3682 old_regend = REGEX_TALLOC (num_regs, const char *);
3683 best_regstart = REGEX_TALLOC (num_regs, const char *);
3684 best_regend = REGEX_TALLOC (num_regs, const char *);
3685 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3686 reg_dummy = REGEX_TALLOC (num_regs, const char *);
3687 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3689 if (!(regstart && regend && old_regstart && old_regend && reg_info
3690 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3698 /* We must initialize all our variables to NULL, so that
3699 `FREE_VARIABLES' doesn't try to free them. */
3700 regstart = regend = old_regstart = old_regend = best_regstart
3701 = best_regend = reg_dummy = NULL;
3702 reg_info = reg_info_dummy = (register_info_type *) NULL;
3704 #endif /* MATCH_MAY_ALLOCATE */
3706 /* The starting position is bogus. */
3707 if (pos < 0 || pos > size1 + size2)
3713 /* Initialize subexpression text positions to -1 to mark ones that no
3714 start_memory/stop_memory has been seen for. Also initialize the
3715 register information struct. */
3716 for (mcnt = 1; mcnt < num_regs; mcnt++)
3718 regstart[mcnt] = regend[mcnt]
3719 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3721 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3722 IS_ACTIVE (reg_info[mcnt]) = 0;
3723 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3724 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3727 /* We move `string1' into `string2' if the latter's empty -- but not if
3728 `string1' is null. */
3729 if (size2 == 0 && string1 != NULL)
3736 end1 = string1 + size1;
3737 end2 = string2 + size2;
3739 /* Compute where to stop matching, within the two strings. */
3742 end_match_1 = string1 + stop;
3743 end_match_2 = string2;
3748 end_match_2 = string2 + stop - size1;
3751 /* `p' scans through the pattern as `d' scans through the data.
3752 `dend' is the end of the input string that `d' points within. `d'
3753 is advanced into the following input string whenever necessary, but
3754 this happens before fetching; therefore, at the beginning of the
3755 loop, `d' can be pointing at the end of a string, but it cannot
3757 if (size1 > 0 && pos <= size1)
3764 d = string2 + pos - size1;
3768 DEBUG_PRINT1 ("The compiled pattern is: ");
3769 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
3770 DEBUG_PRINT1 ("The string to match is: `");
3771 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
3772 DEBUG_PRINT1 ("'\n");
3774 /* This loops over pattern commands. It exits by returning from the
3775 function if the match is complete, or it drops through if the match
3776 fails at this starting point in the input data. */
3779 DEBUG_PRINT2 ("\n0x%x: ", p);
3782 { /* End of pattern means we might have succeeded. */
3783 DEBUG_PRINT1 ("end of pattern ... ");
3785 /* If we haven't matched the entire string, and we want the
3786 longest match, try backtracking. */
3787 if (d != end_match_2)
3789 /* 1 if this match ends in the same string (string1 or string2)
3790 as the best previous match. */
3791 boolean same_str_p = (FIRST_STRING_P (match_end)
3792 == MATCHING_IN_FIRST_STRING);
3793 /* 1 if this match is the best seen so far. */
3794 boolean best_match_p;
3796 /* AIX compiler got confused when this was combined
3797 with the previous declaration. */
3799 best_match_p = d > match_end;
3801 best_match_p = !MATCHING_IN_FIRST_STRING;
3803 DEBUG_PRINT1 ("backtracking.\n");
3805 if (!FAIL_STACK_EMPTY ())
3806 { /* More failure points to try. */
3808 /* If exceeds best match so far, save it. */
3809 if (!best_regs_set || best_match_p)
3811 best_regs_set = true;
3814 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
3816 for (mcnt = 1; mcnt < num_regs; mcnt++)
3818 best_regstart[mcnt] = regstart[mcnt];
3819 best_regend[mcnt] = regend[mcnt];
3825 /* If no failure points, don't restore garbage. And if
3826 last match is real best match, don't restore second
3828 else if (best_regs_set && !best_match_p)
3831 /* Restore best match. It may happen that `dend ==
3832 end_match_1' while the restored d is in string2.
3833 For example, the pattern `x.*y.*z' against the
3834 strings `x-' and `y-z-', if the two strings are
3835 not consecutive in memory. */
3836 DEBUG_PRINT1 ("Restoring best registers.\n");
3839 dend = ((d >= string1 && d <= end1)
3840 ? end_match_1 : end_match_2);
3842 for (mcnt = 1; mcnt < num_regs; mcnt++)
3844 regstart[mcnt] = best_regstart[mcnt];
3845 regend[mcnt] = best_regend[mcnt];
3848 } /* d != end_match_2 */
3851 DEBUG_PRINT1 ("Accepting match.\n");
3853 /* If caller wants register contents data back, do it. */
3854 if (regs && !bufp->no_sub)
3856 /* Have the register data arrays been allocated? */
3857 if (bufp->regs_allocated == REGS_UNALLOCATED)
3858 { /* No. So allocate them with malloc. We need one
3859 extra element beyond `num_regs' for the `-1' marker
3861 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
3862 regs->start = TALLOC (regs->num_regs, regoff_t);
3863 regs->end = TALLOC (regs->num_regs, regoff_t);
3864 if (regs->start == NULL || regs->end == NULL)
3869 bufp->regs_allocated = REGS_REALLOCATE;
3871 else if (bufp->regs_allocated == REGS_REALLOCATE)
3872 { /* Yes. If we need more elements than were already
3873 allocated, reallocate them. If we need fewer, just
3875 if (regs->num_regs < num_regs + 1)
3877 regs->num_regs = num_regs + 1;
3878 RETALLOC (regs->start, regs->num_regs, regoff_t);
3879 RETALLOC (regs->end, regs->num_regs, regoff_t);
3880 if (regs->start == NULL || regs->end == NULL)
3889 /* These braces fend off a "empty body in an else-statement"
3890 warning under GCC when assert expands to nothing. */
3891 assert (bufp->regs_allocated == REGS_FIXED);
3894 /* Convert the pointer data in `regstart' and `regend' to
3895 indices. Register zero has to be set differently,
3896 since we haven't kept track of any info for it. */
3897 if (regs->num_regs > 0)
3899 regs->start[0] = pos;
3900 regs->end[0] = (MATCHING_IN_FIRST_STRING
3901 ? ((regoff_t) (d - string1))
3902 : ((regoff_t) (d - string2 + size1)));
3905 /* Go through the first `min (num_regs, regs->num_regs)'
3906 registers, since that is all we initialized. */
3907 for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++)
3909 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
3910 regs->start[mcnt] = regs->end[mcnt] = -1;
3914 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
3916 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
3920 /* If the regs structure we return has more elements than
3921 were in the pattern, set the extra elements to -1. If
3922 we (re)allocated the registers, this is the case,
3923 because we always allocate enough to have at least one
3925 for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++)
3926 regs->start[mcnt] = regs->end[mcnt] = -1;
3927 } /* regs && !bufp->no_sub */
3929 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
3930 nfailure_points_pushed, nfailure_points_popped,
3931 nfailure_points_pushed - nfailure_points_popped);
3932 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
3934 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
3938 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
3944 /* Otherwise match next pattern command. */
3945 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3947 /* Ignore these. Used to ignore the n of succeed_n's which
3948 currently have n == 0. */
3950 DEBUG_PRINT1 ("EXECUTING no_op.\n");
3954 DEBUG_PRINT1 ("EXECUTING succeed.\n");
3957 /* Match the next n pattern characters exactly. The following
3958 byte in the pattern defines n, and the n bytes after that
3959 are the characters to match. */
3962 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
3964 /* This is written out as an if-else so we don't waste time
3965 testing `translate' inside the loop. */
3971 if ((unsigned char) translate[(unsigned char) *d++]
3972 != (unsigned char) *p++)
3982 if (*d++ != (char) *p++) goto fail;
3986 SET_REGS_MATCHED ();
3990 /* Match any character except possibly a newline or a null. */
3992 DEBUG_PRINT1 ("EXECUTING anychar.\n");
3996 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
3997 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4000 SET_REGS_MATCHED ();
4001 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4009 register unsigned char c;
4010 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4012 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4015 c = TRANSLATE (*d); /* The character to match. */
4017 /* Cast to `unsigned' instead of `unsigned char' in case the
4018 bit list is a full 32 bytes long. */
4019 if (c < (unsigned) (*p * BYTEWIDTH)
4020 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4025 if (!not) goto fail;
4027 SET_REGS_MATCHED ();
4033 /* The beginning of a group is represented by start_memory.
4034 The arguments are the register number in the next byte, and the
4035 number of groups inner to this one in the next. The text
4036 matched within the group is recorded (in the internal
4037 registers data structure) under the register number. */
4039 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4041 /* Find out if this group can match the empty string. */
4042 p1 = p; /* To send to group_match_null_string_p. */
4044 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4045 REG_MATCH_NULL_STRING_P (reg_info[*p])
4046 = group_match_null_string_p (&p1, pend, reg_info);
4048 /* Save the position in the string where we were the last time
4049 we were at this open-group operator in case the group is
4050 operated upon by a repetition operator, e.g., with `(a*)*b'
4051 against `ab'; then we want to ignore where we are now in
4052 the string in case this attempt to match fails. */
4053 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4054 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4056 DEBUG_PRINT2 (" old_regstart: %d\n",
4057 POINTER_TO_OFFSET (old_regstart[*p]));
4060 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4062 IS_ACTIVE (reg_info[*p]) = 1;
4063 MATCHED_SOMETHING (reg_info[*p]) = 0;
4065 /* Clear this whenever we change the register activity status. */
4066 set_regs_matched_done = 0;
4068 /* This is the new highest active register. */
4069 highest_active_reg = *p;
4071 /* If nothing was active before, this is the new lowest active
4073 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4074 lowest_active_reg = *p;
4076 /* Move past the register number and inner group count. */
4078 just_past_start_mem = p;
4083 /* The stop_memory opcode represents the end of a group. Its
4084 arguments are the same as start_memory's: the register
4085 number, and the number of inner groups. */
4087 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4089 /* We need to save the string position the last time we were at
4090 this close-group operator in case the group is operated
4091 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4092 against `aba'; then we want to ignore where we are now in
4093 the string in case this attempt to match fails. */
4094 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4095 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4097 DEBUG_PRINT2 (" old_regend: %d\n",
4098 POINTER_TO_OFFSET (old_regend[*p]));
4101 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4103 /* This register isn't active anymore. */
4104 IS_ACTIVE (reg_info[*p]) = 0;
4106 /* Clear this whenever we change the register activity status. */
4107 set_regs_matched_done = 0;
4109 /* If this was the only register active, nothing is active
4111 if (lowest_active_reg == highest_active_reg)
4113 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4114 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4117 { /* We must scan for the new highest active register, since
4118 it isn't necessarily one less than now: consider
4119 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4120 new highest active register is 1. */
4121 unsigned char r = *p - 1;
4122 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4125 /* If we end up at register zero, that means that we saved
4126 the registers as the result of an `on_failure_jump', not
4127 a `start_memory', and we jumped to past the innermost
4128 `stop_memory'. For example, in ((.)*) we save
4129 registers 1 and 2 as a result of the *, but when we pop
4130 back to the second ), we are at the stop_memory 1.
4131 Thus, nothing is active. */
4134 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4135 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4138 highest_active_reg = r;
4141 /* If just failed to match something this time around with a
4142 group that's operated on by a repetition operator, try to
4143 force exit from the ``loop'', and restore the register
4144 information for this group that we had before trying this
4146 if ((!MATCHED_SOMETHING (reg_info[*p])
4147 || just_past_start_mem == p - 1)
4150 boolean is_a_jump_n = false;
4154 switch ((re_opcode_t) *p1++)
4158 case pop_failure_jump:
4159 case maybe_pop_jump:
4161 case dummy_failure_jump:
4162 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4172 /* If the next operation is a jump backwards in the pattern
4173 to an on_failure_jump right before the start_memory
4174 corresponding to this stop_memory, exit from the loop
4175 by forcing a failure after pushing on the stack the
4176 on_failure_jump's jump in the pattern, and d. */
4177 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4178 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4180 /* If this group ever matched anything, then restore
4181 what its registers were before trying this last
4182 failed match, e.g., with `(a*)*b' against `ab' for
4183 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4184 against `aba' for regend[3].
4186 Also restore the registers for inner groups for,
4187 e.g., `((a*)(b*))*' against `aba' (register 3 would
4188 otherwise get trashed). */
4190 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4194 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4196 /* Restore this and inner groups' (if any) registers. */
4197 for (r = *p; r < *p + *(p + 1); r++)
4199 regstart[r] = old_regstart[r];
4201 /* xx why this test? */
4202 if (old_regend[r] >= regstart[r])
4203 regend[r] = old_regend[r];
4207 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4208 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4214 /* Move past the register number and the inner group count. */
4219 /* \<digit> has been turned into a `duplicate' command which is
4220 followed by the numeric value of <digit> as the register number. */
4223 register const char *d2, *dend2;
4224 int regno = *p++; /* Get which register to match against. */
4225 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4227 /* Can't back reference a group which we've never matched. */
4228 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4231 /* Where in input to try to start matching. */
4232 d2 = regstart[regno];
4234 /* Where to stop matching; if both the place to start and
4235 the place to stop matching are in the same string, then
4236 set to the place to stop, otherwise, for now have to use
4237 the end of the first string. */
4239 dend2 = ((FIRST_STRING_P (regstart[regno])
4240 == FIRST_STRING_P (regend[regno]))
4241 ? regend[regno] : end_match_1);
4244 /* If necessary, advance to next segment in register
4248 if (dend2 == end_match_2) break;
4249 if (dend2 == regend[regno]) break;
4251 /* End of string1 => advance to string2. */
4253 dend2 = regend[regno];
4255 /* At end of register contents => success */
4256 if (d2 == dend2) break;
4258 /* If necessary, advance to next segment in data. */
4261 /* How many characters left in this segment to match. */
4264 /* Want how many consecutive characters we can match in
4265 one shot, so, if necessary, adjust the count. */
4266 if (mcnt > dend2 - d2)
4269 /* Compare that many; failure if mismatch, else move
4272 ? bcmp_translate (d, d2, mcnt, translate)
4273 : bcmp (d, d2, mcnt))
4275 d += mcnt, d2 += mcnt;
4277 /* Do this because we've match some characters. */
4278 SET_REGS_MATCHED ();
4284 /* begline matches the empty string at the beginning of the string
4285 (unless `not_bol' is set in `bufp'), and, if
4286 `newline_anchor' is set, after newlines. */
4288 DEBUG_PRINT1 ("EXECUTING begline.\n");
4290 if (AT_STRINGS_BEG (d))
4292 if (!bufp->not_bol) break;
4294 else if (d[-1] == '\n' && bufp->newline_anchor)
4298 /* In all other cases, we fail. */
4302 /* endline is the dual of begline. */
4304 DEBUG_PRINT1 ("EXECUTING endline.\n");
4306 if (AT_STRINGS_END (d))
4308 if (!bufp->not_eol) break;
4311 /* We have to ``prefetch'' the next character. */
4312 else if ((d == end1 ? *string2 : *d) == '\n'
4313 && bufp->newline_anchor)
4320 /* Match at the very beginning of the data. */
4322 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4323 if (AT_STRINGS_BEG (d))
4328 /* Match at the very end of the data. */
4330 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4331 if (AT_STRINGS_END (d))
4336 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4337 pushes NULL as the value for the string on the stack. Then
4338 `pop_failure_point' will keep the current value for the
4339 string, instead of restoring it. To see why, consider
4340 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4341 then the . fails against the \n. But the next thing we want
4342 to do is match the \n against the \n; if we restored the
4343 string value, we would be back at the foo.
4345 Because this is used only in specific cases, we don't need to
4346 check all the things that `on_failure_jump' does, to make
4347 sure the right things get saved on the stack. Hence we don't
4348 share its code. The only reason to push anything on the
4349 stack at all is that otherwise we would have to change
4350 `anychar's code to do something besides goto fail in this
4351 case; that seems worse than this. */
4352 case on_failure_keep_string_jump:
4353 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4355 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4356 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4358 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4362 /* Uses of on_failure_jump:
4364 Each alternative starts with an on_failure_jump that points
4365 to the beginning of the next alternative. Each alternative
4366 except the last ends with a jump that in effect jumps past
4367 the rest of the alternatives. (They really jump to the
4368 ending jump of the following alternative, because tensioning
4369 these jumps is a hassle.)
4371 Repeats start with an on_failure_jump that points past both
4372 the repetition text and either the following jump or
4373 pop_failure_jump back to this on_failure_jump. */
4374 case on_failure_jump:
4376 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4378 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4379 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4381 /* If this on_failure_jump comes right before a group (i.e.,
4382 the original * applied to a group), save the information
4383 for that group and all inner ones, so that if we fail back
4384 to this point, the group's information will be correct.
4385 For example, in \(a*\)*\1, we need the preceding group,
4386 and in \(zz\(a*\)b*\)\2, we need the inner group. */
4388 /* We can't use `p' to check ahead because we push
4389 a failure point to `p + mcnt' after we do this. */
4392 /* We need to skip no_op's before we look for the
4393 start_memory in case this on_failure_jump is happening as
4394 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4396 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4399 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4401 /* We have a new highest active register now. This will
4402 get reset at the start_memory we are about to get to,
4403 but we will have saved all the registers relevant to
4404 this repetition op, as described above. */
4405 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4406 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4407 lowest_active_reg = *(p1 + 1);
4410 DEBUG_PRINT1 (":\n");
4411 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4415 /* A smart repeat ends with `maybe_pop_jump'.
4416 We change it to either `pop_failure_jump' or `jump'. */
4417 case maybe_pop_jump:
4418 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4419 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4421 register unsigned char *p2 = p;
4423 /* Compare the beginning of the repeat with what in the
4424 pattern follows its end. If we can establish that there
4425 is nothing that they would both match, i.e., that we
4426 would have to backtrack because of (as in, e.g., `a*a')
4427 then we can change to pop_failure_jump, because we'll
4428 never have to backtrack.
4430 This is not true in the case of alternatives: in
4431 `(a|ab)*' we do need to backtrack to the `ab' alternative
4432 (e.g., if the string was `ab'). But instead of trying to
4433 detect that here, the alternative has put on a dummy
4434 failure point which is what we will end up popping. */
4436 /* Skip over open/close-group commands.
4437 If what follows this loop is a ...+ construct,
4438 look at what begins its body, since we will have to
4439 match at least one of that. */
4443 && ((re_opcode_t) *p2 == stop_memory
4444 || (re_opcode_t) *p2 == start_memory))
4446 else if (p2 + 6 < pend
4447 && (re_opcode_t) *p2 == dummy_failure_jump)
4454 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4455 to the `maybe_finalize_jump' of this case. Examine what
4458 /* If we're at the end of the pattern, we can change. */
4461 /* Consider what happens when matching ":\(.*\)"
4462 against ":/". I don't really understand this code
4464 p[-3] = (unsigned char) pop_failure_jump;
4466 (" End of pattern: change to `pop_failure_jump'.\n");
4469 else if ((re_opcode_t) *p2 == exactn
4470 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4472 register unsigned char c
4473 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4475 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4477 p[-3] = (unsigned char) pop_failure_jump;
4478 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4482 else if ((re_opcode_t) p1[3] == charset
4483 || (re_opcode_t) p1[3] == charset_not)
4485 int not = (re_opcode_t) p1[3] == charset_not;
4487 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4488 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4491 /* `not' is equal to 1 if c would match, which means
4492 that we can't change to pop_failure_jump. */
4495 p[-3] = (unsigned char) pop_failure_jump;
4496 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4500 else if ((re_opcode_t) *p2 == charset)
4503 register unsigned char c
4504 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4507 if ((re_opcode_t) p1[3] == exactn
4508 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
4509 && (p2[2 + p1[5] / BYTEWIDTH]
4510 & (1 << (p1[5] % BYTEWIDTH)))))
4512 p[-3] = (unsigned char) pop_failure_jump;
4513 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4517 else if ((re_opcode_t) p1[3] == charset_not)
4520 /* We win if the charset_not inside the loop
4521 lists every character listed in the charset after. */
4522 for (idx = 0; idx < (int) p2[1]; idx++)
4523 if (! (p2[2 + idx] == 0
4524 || (idx < (int) p1[4]
4525 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4530 p[-3] = (unsigned char) pop_failure_jump;
4531 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4534 else if ((re_opcode_t) p1[3] == charset)
4537 /* We win if the charset inside the loop
4538 has no overlap with the one after the loop. */
4540 idx < (int) p2[1] && idx < (int) p1[4];
4542 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4545 if (idx == p2[1] || idx == p1[4])
4547 p[-3] = (unsigned char) pop_failure_jump;
4548 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4553 p -= 2; /* Point at relative address again. */
4554 if ((re_opcode_t) p[-1] != pop_failure_jump)
4556 p[-1] = (unsigned char) jump;
4557 DEBUG_PRINT1 (" Match => jump.\n");
4558 goto unconditional_jump;
4560 /* Note fall through. */
4563 /* The end of a simple repeat has a pop_failure_jump back to
4564 its matching on_failure_jump, where the latter will push a
4565 failure point. The pop_failure_jump takes off failure
4566 points put on by this pop_failure_jump's matching
4567 on_failure_jump; we got through the pattern to here from the
4568 matching on_failure_jump, so didn't fail. */
4569 case pop_failure_jump:
4571 /* We need to pass separate storage for the lowest and
4572 highest registers, even though we don't care about the
4573 actual values. Otherwise, we will restore only one
4574 register from the stack, since lowest will == highest in
4575 `pop_failure_point'. */
4576 unsigned dummy_low_reg, dummy_high_reg;
4577 unsigned char *pdummy;
4580 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4581 POP_FAILURE_POINT (sdummy, pdummy,
4582 dummy_low_reg, dummy_high_reg,
4583 reg_dummy, reg_dummy, reg_info_dummy);
4585 /* Note fall through. */
4588 /* Unconditionally jump (without popping any failure points). */
4591 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4592 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4593 p += mcnt; /* Do the jump. */
4594 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4598 /* We need this opcode so we can detect where alternatives end
4599 in `group_match_null_string_p' et al. */
4601 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4602 goto unconditional_jump;
4605 /* Normally, the on_failure_jump pushes a failure point, which
4606 then gets popped at pop_failure_jump. We will end up at
4607 pop_failure_jump, also, and with a pattern of, say, `a+', we
4608 are skipping over the on_failure_jump, so we have to push
4609 something meaningless for pop_failure_jump to pop. */
4610 case dummy_failure_jump:
4611 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4612 /* It doesn't matter what we push for the string here. What
4613 the code at `fail' tests is the value for the pattern. */
4614 PUSH_FAILURE_POINT (0, 0, -2);
4615 goto unconditional_jump;
4618 /* At the end of an alternative, we need to push a dummy failure
4619 point in case we are followed by a `pop_failure_jump', because
4620 we don't want the failure point for the alternative to be
4621 popped. For example, matching `(a|ab)*' against `aab'
4622 requires that we match the `ab' alternative. */
4623 case push_dummy_failure:
4624 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4625 /* See comments just above at `dummy_failure_jump' about the
4627 PUSH_FAILURE_POINT (0, 0, -2);
4630 /* Have to succeed matching what follows at least n times.
4631 After that, handle like `on_failure_jump'. */
4633 EXTRACT_NUMBER (mcnt, p + 2);
4634 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4637 /* Originally, this is how many times we HAVE to succeed. */
4642 STORE_NUMBER_AND_INCR (p, mcnt);
4643 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt);
4647 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4648 p[2] = (unsigned char) no_op;
4649 p[3] = (unsigned char) no_op;
4655 EXTRACT_NUMBER (mcnt, p + 2);
4656 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4658 /* Originally, this is how many times we CAN jump. */
4662 STORE_NUMBER (p + 2, mcnt);
4663 goto unconditional_jump;
4665 /* If don't have to jump any more, skip over the rest of command. */
4672 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4674 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4676 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4677 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4678 STORE_NUMBER (p1, mcnt);
4683 /* The DEC Alpha C compiler 3.x generates incorrect code for the
4684 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4685 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4686 macro and introducing temporary variables works around the bug. */
4689 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4690 if (AT_WORD_BOUNDARY (d))
4695 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4696 if (AT_WORD_BOUNDARY (d))
4702 boolean prevchar, thischar;
4704 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4705 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4708 prevchar = WORDCHAR_P (d - 1);
4709 thischar = WORDCHAR_P (d);
4710 if (prevchar != thischar)
4717 boolean prevchar, thischar;
4719 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4720 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4723 prevchar = WORDCHAR_P (d - 1);
4724 thischar = WORDCHAR_P (d);
4725 if (prevchar != thischar)
4732 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
4733 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
4738 DEBUG_PRINT1 ("EXECUTING wordend.\n");
4739 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
4740 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
4746 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
4747 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
4752 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
4753 if (PTR_CHAR_POS ((unsigned char *) d) != point)
4758 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
4759 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
4764 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
4769 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
4773 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
4775 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
4777 SET_REGS_MATCHED ();
4781 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
4783 goto matchnotsyntax;
4786 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
4790 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
4792 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
4794 SET_REGS_MATCHED ();
4797 #else /* not emacs */
4799 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
4801 if (!WORDCHAR_P (d))
4803 SET_REGS_MATCHED ();
4808 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
4812 SET_REGS_MATCHED ();
4815 #endif /* not emacs */
4820 continue; /* Successfully executed one pattern command; keep going. */
4823 /* We goto here if a matching operation fails. */
4825 if (!FAIL_STACK_EMPTY ())
4826 { /* A restart point is known. Restore to that state. */
4827 DEBUG_PRINT1 ("\nFAIL:\n");
4828 POP_FAILURE_POINT (d, p,
4829 lowest_active_reg, highest_active_reg,
4830 regstart, regend, reg_info);
4832 /* If this failure point is a dummy, try the next one. */
4836 /* If we failed to the end of the pattern, don't examine *p. */
4840 boolean is_a_jump_n = false;
4842 /* If failed to a backwards jump that's part of a repetition
4843 loop, need to pop this failure point and use the next one. */
4844 switch ((re_opcode_t) *p)
4848 case maybe_pop_jump:
4849 case pop_failure_jump:
4852 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4855 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
4857 && (re_opcode_t) *p1 == on_failure_jump))
4865 if (d >= string1 && d <= end1)
4869 break; /* Matching at this starting point really fails. */
4873 goto restore_best_regs;
4877 return -1; /* Failure to match. */
4880 /* Subroutine definitions for re_match_2. */
4883 /* We are passed P pointing to a register number after a start_memory.
4885 Return true if the pattern up to the corresponding stop_memory can
4886 match the empty string, and false otherwise.
4888 If we find the matching stop_memory, sets P to point to one past its number.
4889 Otherwise, sets P to an undefined byte less than or equal to END.
4891 We don't handle duplicates properly (yet). */
4894 group_match_null_string_p (p, end, reg_info)
4895 unsigned char **p, *end;
4896 register_info_type *reg_info;
4899 /* Point to after the args to the start_memory. */
4900 unsigned char *p1 = *p + 2;
4904 /* Skip over opcodes that can match nothing, and return true or
4905 false, as appropriate, when we get to one that can't, or to the
4906 matching stop_memory. */
4908 switch ((re_opcode_t) *p1)
4910 /* Could be either a loop or a series of alternatives. */
4911 case on_failure_jump:
4913 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4915 /* If the next operation is not a jump backwards in the
4920 /* Go through the on_failure_jumps of the alternatives,
4921 seeing if any of the alternatives cannot match nothing.
4922 The last alternative starts with only a jump,
4923 whereas the rest start with on_failure_jump and end
4924 with a jump, e.g., here is the pattern for `a|b|c':
4926 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
4927 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
4930 So, we have to first go through the first (n-1)
4931 alternatives and then deal with the last one separately. */
4934 /* Deal with the first (n-1) alternatives, which start
4935 with an on_failure_jump (see above) that jumps to right
4936 past a jump_past_alt. */
4938 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
4940 /* `mcnt' holds how many bytes long the alternative
4941 is, including the ending `jump_past_alt' and
4944 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
4948 /* Move to right after this alternative, including the
4952 /* Break if it's the beginning of an n-th alternative
4953 that doesn't begin with an on_failure_jump. */
4954 if ((re_opcode_t) *p1 != on_failure_jump)
4957 /* Still have to check that it's not an n-th
4958 alternative that starts with an on_failure_jump. */
4960 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4961 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
4963 /* Get to the beginning of the n-th alternative. */
4969 /* Deal with the last alternative: go back and get number
4970 of the `jump_past_alt' just before it. `mcnt' contains
4971 the length of the alternative. */
4972 EXTRACT_NUMBER (mcnt, p1 - 2);
4974 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
4977 p1 += mcnt; /* Get past the n-th alternative. */
4983 assert (p1[1] == **p);
4989 if (!common_op_match_null_string_p (&p1, end, reg_info))
4992 } /* while p1 < end */
4995 } /* group_match_null_string_p */
4998 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
4999 It expects P to be the first byte of a single alternative and END one
5000 byte past the last. The alternative can contain groups. */
5003 alt_match_null_string_p (p, end, reg_info)
5004 unsigned char *p, *end;
5005 register_info_type *reg_info;
5008 unsigned char *p1 = p;
5012 /* Skip over opcodes that can match nothing, and break when we get
5013 to one that can't. */
5015 switch ((re_opcode_t) *p1)
5018 case on_failure_jump:
5020 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5025 if (!common_op_match_null_string_p (&p1, end, reg_info))
5028 } /* while p1 < end */
5031 } /* alt_match_null_string_p */
5034 /* Deals with the ops common to group_match_null_string_p and
5035 alt_match_null_string_p.
5037 Sets P to one after the op and its arguments, if any. */
5040 common_op_match_null_string_p (p, end, reg_info)
5041 unsigned char **p, *end;
5042 register_info_type *reg_info;
5047 unsigned char *p1 = *p;
5049 switch ((re_opcode_t) *p1++)
5069 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5070 ret = group_match_null_string_p (&p1, end, reg_info);
5072 /* Have to set this here in case we're checking a group which
5073 contains a group and a back reference to it. */
5075 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5076 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5082 /* If this is an optimized succeed_n for zero times, make the jump. */
5084 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5092 /* Get to the number of times to succeed. */
5094 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5099 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5107 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5115 /* All other opcodes mean we cannot match the empty string. */
5121 } /* common_op_match_null_string_p */
5124 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5125 bytes; nonzero otherwise. */
5128 bcmp_translate (s1, s2, len, translate)
5129 unsigned char *s1, *s2;
5131 RE_TRANSLATE_TYPE translate;
5133 register unsigned char *p1 = s1, *p2 = s2;
5136 if (translate[*p1++] != translate[*p2++]) return 1;
5142 /* Entry points for GNU code. */
5144 /* re_compile_pattern is the GNU regular expression compiler: it
5145 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5146 Returns 0 if the pattern was valid, otherwise an error string.
5148 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5149 are set in BUFP on entry.
5151 We call regex_compile to do the actual compilation. */
5154 re_compile_pattern (pattern, length, bufp)
5155 const char *pattern;
5157 struct re_pattern_buffer *bufp;
5161 /* GNU code is written to assume at least RE_NREGS registers will be set
5162 (and at least one extra will be -1). */
5163 bufp->regs_allocated = REGS_UNALLOCATED;
5165 /* And GNU code determines whether or not to get register information
5166 by passing null for the REGS argument to re_match, etc., not by
5170 /* Match anchors at newline. */
5171 bufp->newline_anchor = 1;
5173 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5177 return gettext (re_error_msgid[(int) ret]);
5180 /* Entry points compatible with 4.2 BSD regex library. We don't define
5181 them unless specifically requested. */
5183 #ifdef _REGEX_RE_COMP
5185 /* BSD has one and only one pattern buffer. */
5186 static struct re_pattern_buffer re_comp_buf;
5196 if (!re_comp_buf.buffer)
5197 return gettext ("No previous regular expression");
5201 if (!re_comp_buf.buffer)
5203 re_comp_buf.buffer = (unsigned char *) malloc (200);
5204 if (re_comp_buf.buffer == NULL)
5205 return gettext (re_error_msgid[(int) REG_ESPACE]);
5206 re_comp_buf.allocated = 200;
5208 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5209 if (re_comp_buf.fastmap == NULL)
5210 return gettext (re_error_msgid[(int) REG_ESPACE]);
5213 /* Since `re_exec' always passes NULL for the `regs' argument, we
5214 don't need to initialize the pattern buffer fields which affect it. */
5216 /* Match anchors at newlines. */
5217 re_comp_buf.newline_anchor = 1;
5219 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5224 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5225 return (char *) gettext (re_error_msgid[(int) ret]);
5233 const int len = strlen (s);
5235 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5237 #endif /* _REGEX_RE_COMP */
5239 /* POSIX.2 functions. Don't define these for Emacs. */
5243 /* regcomp takes a regular expression as a string and compiles it.
5245 PREG is a regex_t *. We do not expect any fields to be initialized,
5246 since POSIX says we shouldn't. Thus, we set
5248 `buffer' to the compiled pattern;
5249 `used' to the length of the compiled pattern;
5250 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5251 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5252 RE_SYNTAX_POSIX_BASIC;
5253 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5254 `fastmap' and `fastmap_accurate' to zero;
5255 `re_nsub' to the number of subexpressions in PATTERN.
5257 PATTERN is the address of the pattern string.
5259 CFLAGS is a series of bits which affect compilation.
5261 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5262 use POSIX basic syntax.
5264 If REG_NEWLINE is set, then . and [^...] don't match newline.
5265 Also, regexec will try a match beginning after every newline.
5267 If REG_ICASE is set, then we considers upper- and lowercase
5268 versions of letters to be equivalent when matching.
5270 If REG_NOSUB is set, then when PREG is passed to regexec, that
5271 routine will report only success or failure, and nothing about the
5274 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
5275 the return codes and their meanings.) */
5278 regcomp (preg, pattern, cflags)
5280 const char *pattern;
5285 = (cflags & REG_EXTENDED) ?
5286 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5288 /* regex_compile will allocate the space for the compiled pattern. */
5290 preg->allocated = 0;
5293 /* Don't bother to use a fastmap when searching. This simplifies the
5294 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
5295 characters after newlines into the fastmap. This way, we just try
5299 if (cflags & REG_ICASE)
5304 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
5305 * sizeof (*(RE_TRANSLATE_TYPE)0));
5306 if (preg->translate == NULL)
5307 return (int) REG_ESPACE;
5309 /* Map uppercase characters to corresponding lowercase ones. */
5310 for (i = 0; i < CHAR_SET_SIZE; i++)
5311 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
5314 preg->translate = NULL;
5316 /* If REG_NEWLINE is set, newlines are treated differently. */
5317 if (cflags & REG_NEWLINE)
5318 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5319 syntax &= ~RE_DOT_NEWLINE;
5320 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5321 /* It also changes the matching behavior. */
5322 preg->newline_anchor = 1;
5325 preg->newline_anchor = 0;
5327 preg->no_sub = !!(cflags & REG_NOSUB);
5329 /* POSIX says a null character in the pattern terminates it, so we
5330 can use strlen here in compiling the pattern. */
5331 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5333 /* POSIX doesn't distinguish between an unmatched open-group and an
5334 unmatched close-group: both are REG_EPAREN. */
5335 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5341 /* regexec searches for a given pattern, specified by PREG, in the
5344 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5345 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5346 least NMATCH elements, and we set them to the offsets of the
5347 corresponding matched substrings.
5349 EFLAGS specifies `execution flags' which affect matching: if
5350 REG_NOTBOL is set, then ^ does not match at the beginning of the
5351 string; if REG_NOTEOL is set, then $ does not match at the end.
5353 We return 0 if we find a match and REG_NOMATCH if not. */
5356 regexec (preg, string, nmatch, pmatch, eflags)
5357 const regex_t *preg;
5360 regmatch_t pmatch[];
5364 struct re_registers regs;
5365 regex_t private_preg;
5366 int len = strlen (string);
5367 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5369 private_preg = *preg;
5371 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5372 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5374 /* The user has told us exactly how many registers to return
5375 information about, via `nmatch'. We have to pass that on to the
5376 matching routines. */
5377 private_preg.regs_allocated = REGS_FIXED;
5381 regs.num_regs = nmatch;
5382 regs.start = TALLOC (nmatch, regoff_t);
5383 regs.end = TALLOC (nmatch, regoff_t);
5384 if (regs.start == NULL || regs.end == NULL)
5385 return (int) REG_NOMATCH;
5388 /* Perform the searching operation. */
5389 ret = re_search (&private_preg, string, len,
5390 /* start: */ 0, /* range: */ len,
5391 want_reg_info ? ®s : (struct re_registers *) 0);
5393 /* Copy the register information to the POSIX structure. */
5400 for (r = 0; r < nmatch; r++)
5402 pmatch[r].rm_so = regs.start[r];
5403 pmatch[r].rm_eo = regs.end[r];
5407 /* If we needed the temporary register info, free the space now. */
5412 /* We want zero return to mean success, unlike `re_search'. */
5413 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5417 /* Returns a message corresponding to an error code, ERRCODE, returned
5418 from either regcomp or regexec. We don't use PREG here. */
5421 regerror (errcode, preg, errbuf, errbuf_size)
5423 const regex_t *preg;
5431 || errcode >= (sizeof (re_error_msgid) / sizeof (re_error_msgid[0])))
5432 /* Only error codes returned by the rest of the code should be passed
5433 to this routine. If we are given anything else, or if other regex
5434 code generates an invalid error code, then the program has a bug.
5435 Dump core so we can fix it. */
5438 msg = gettext (re_error_msgid[errcode]);
5440 msg_size = strlen (msg) + 1; /* Includes the null. */
5442 if (errbuf_size != 0)
5444 if (msg_size > errbuf_size)
5446 strncpy (errbuf, msg, errbuf_size - 1);
5447 errbuf[errbuf_size - 1] = 0;
5450 strcpy (errbuf, msg);
5457 /* Free dynamically allocated space used by PREG. */
5463 if (preg->buffer != NULL)
5464 free (preg->buffer);
5465 preg->buffer = NULL;
5467 preg->allocated = 0;
5470 if (preg->fastmap != NULL)
5471 free (preg->fastmap);
5472 preg->fastmap = NULL;
5473 preg->fastmap_accurate = 0;
5475 if (preg->translate != NULL)
5476 free (preg->translate);
5477 preg->translate = NULL;
5480 #endif /* not emacs */
projects / gnulib.git / blob