1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
5 Copyright (C) 1993, 94, 95, 96, 97, 98 Free Software Foundation, Inc.
7 NOTE: The canonical source of this file is maintained with the
8 GNU C Library. Bugs can be reported to bug-glibc@prep.ai.mit.edu.
10 This program is free software; you can redistribute it and/or modify it
11 under the terms of the GNU General Public License as published by the
12 Free Software Foundation; either version 2, or (at your option) any
15 This program is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 GNU General Public License for more details.
20 You should have received a copy of the GNU General Public License
21 along with this program; if not, write to the Free Software Foundation,
22 Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
24 /* AIX requires this to be the first thing in the file. */
25 #if defined _AIX && !defined REGEX_MALLOC
37 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
38 # define PARAMS(args) args
40 # define PARAMS(args) ()
42 #endif /* Not PARAMS. */
44 #if defined STDC_HEADERS && !defined emacs
47 /* We need this for `gnu-regex.h', and perhaps for the Emacs include files. */
48 # include <sys/types.h>
51 /* For platform which support the ISO C amendement 1 functionality we
52 support user defined character classes. */
53 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
54 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
59 /* This is for other GNU distributions with internationalized messages. */
60 #if HAVE_LIBINTL_H || defined _LIBC
63 # define gettext(msgid) (msgid)
67 /* This define is so xgettext can find the internationalizable
69 # define gettext_noop(String) String
72 /* The `emacs' switch turns on certain matching commands
73 that make sense only in Emacs. */
82 /* If we are not linking with Emacs proper,
83 we can't use the relocating allocator
84 even if config.h says that we can. */
87 # if defined STDC_HEADERS || defined _LIBC
94 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
95 If nothing else has been done, use the method below. */
96 # ifdef INHIBIT_STRING_HEADER
97 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
98 # if !defined bzero && !defined bcopy
99 # undef INHIBIT_STRING_HEADER
104 /* This is the normal way of making sure we have a bcopy and a bzero.
105 This is used in most programs--a few other programs avoid this
106 by defining INHIBIT_STRING_HEADER. */
107 # ifndef INHIBIT_STRING_HEADER
108 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
112 # define bzero(s, n) (memset (s, '\0', n), (s))
114 # define bzero(s, n) __bzero (s, n)
118 # include <strings.h>
120 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
123 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
128 /* Define the syntax stuff for \<, \>, etc. */
130 /* This must be nonzero for the wordchar and notwordchar pattern
131 commands in re_match_2. */
136 # ifdef SWITCH_ENUM_BUG
137 # define SWITCH_ENUM_CAST(x) ((int)(x))
139 # define SWITCH_ENUM_CAST(x) (x)
142 /* How many characters in the character set. */
143 # define CHAR_SET_SIZE 256
145 /* GDB LOCAL: define _REGEX_RE_COMP to get BSD style re_comp and re_exec */
146 #ifndef _REGEX_RE_COMP
147 #define _REGEX_RE_COMP
152 extern char *re_syntax_table;
154 # else /* not SYNTAX_TABLE */
156 static char re_syntax_table[CHAR_SET_SIZE];
167 bzero (re_syntax_table, sizeof re_syntax_table);
169 for (c = 'a'; c <= 'z'; c++)
170 re_syntax_table[c] = Sword;
172 for (c = 'A'; c <= 'Z'; c++)
173 re_syntax_table[c] = Sword;
175 for (c = '0'; c <= '9'; c++)
176 re_syntax_table[c] = Sword;
178 re_syntax_table['_'] = Sword;
183 # endif /* not SYNTAX_TABLE */
185 # define SYNTAX(c) re_syntax_table[c]
187 #endif /* not emacs */
189 /* Get the interface, including the syntax bits. */
190 /* CYGNUS LOCAL: call it gnu-regex.h, not regex.h, to avoid name conflicts */
191 #include "gnu-regex.h"
193 /* isalpha etc. are used for the character classes. */
196 /* Jim Meyering writes:
198 "... Some ctype macros are valid only for character codes that
199 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
200 using /bin/cc or gcc but without giving an ansi option). So, all
201 ctype uses should be through macros like ISPRINT... If
202 STDC_HEADERS is defined, then autoconf has verified that the ctype
203 macros don't need to be guarded with references to isascii. ...
204 Defining isascii to 1 should let any compiler worth its salt
205 eliminate the && through constant folding."
206 Solaris defines some of these symbols so we must undefine them first. */
209 #if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
210 # define ISASCII(c) 1
212 # define ISASCII(c) isascii(c)
216 # define ISBLANK(c) (ISASCII (c) && isblank (c))
218 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
221 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
223 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
227 #define ISPRINT(c) (ISASCII (c) && isprint (c))
228 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
229 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
230 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
231 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
232 #define ISLOWER(c) (ISASCII (c) && islower (c))
233 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
234 #define ISSPACE(c) (ISASCII (c) && isspace (c))
235 #define ISUPPER(c) (ISASCII (c) && isupper (c))
236 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
239 # define NULL (void *)0
242 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
243 since ours (we hope) works properly with all combinations of
244 machines, compilers, `char' and `unsigned char' argument types.
245 (Per Bothner suggested the basic approach.) */
246 #undef SIGN_EXTEND_CHAR
248 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
249 #else /* not __STDC__ */
250 /* As in Harbison and Steele. */
251 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
254 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
255 use `alloca' instead of `malloc'. This is because using malloc in
256 re_search* or re_match* could cause memory leaks when C-g is used in
257 Emacs; also, malloc is slower and causes storage fragmentation. On
258 the other hand, malloc is more portable, and easier to debug.
260 Because we sometimes use alloca, some routines have to be macros,
261 not functions -- `alloca'-allocated space disappears at the end of the
262 function it is called in. */
266 # define REGEX_ALLOCATE malloc
267 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
268 # define REGEX_FREE free
270 #else /* not REGEX_MALLOC */
272 /* Emacs already defines alloca, sometimes. */
275 /* Make alloca work the best possible way. */
277 # define alloca __builtin_alloca
278 # else /* not __GNUC__ */
281 # endif /* HAVE_ALLOCA_H */
282 # endif /* not __GNUC__ */
284 # endif /* not alloca */
286 # define REGEX_ALLOCATE alloca
288 /* Assumes a `char *destination' variable. */
289 # define REGEX_REALLOCATE(source, osize, nsize) \
290 (destination = (char *) alloca (nsize), \
291 memcpy (destination, source, osize))
293 /* No need to do anything to free, after alloca. */
294 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
296 #endif /* not REGEX_MALLOC */
298 /* Define how to allocate the failure stack. */
300 #if defined REL_ALLOC && defined REGEX_MALLOC
302 # define REGEX_ALLOCATE_STACK(size) \
303 r_alloc (&failure_stack_ptr, (size))
304 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
305 r_re_alloc (&failure_stack_ptr, (nsize))
306 # define REGEX_FREE_STACK(ptr) \
307 r_alloc_free (&failure_stack_ptr)
309 #else /* not using relocating allocator */
313 # define REGEX_ALLOCATE_STACK malloc
314 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
315 # define REGEX_FREE_STACK free
317 # else /* not REGEX_MALLOC */
319 # define REGEX_ALLOCATE_STACK alloca
321 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
322 REGEX_REALLOCATE (source, osize, nsize)
323 /* No need to explicitly free anything. */
324 # define REGEX_FREE_STACK(arg)
326 # endif /* not REGEX_MALLOC */
327 #endif /* not using relocating allocator */
330 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
331 `string1' or just past its end. This works if PTR is NULL, which is
333 #define FIRST_STRING_P(ptr) \
334 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
336 /* (Re)Allocate N items of type T using malloc, or fail. */
337 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
338 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
339 #define RETALLOC_IF(addr, n, t) \
340 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
341 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
343 #define BYTEWIDTH 8 /* In bits. */
345 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
349 #define MAX(a, b) ((a) > (b) ? (a) : (b))
350 #define MIN(a, b) ((a) < (b) ? (a) : (b))
352 typedef char boolean;
356 static int re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
357 const char *string1, int size1,
358 const char *string2, int size2,
360 struct re_registers *regs,
363 /* These are the command codes that appear in compiled regular
364 expressions. Some opcodes are followed by argument bytes. A
365 command code can specify any interpretation whatsoever for its
366 arguments. Zero bytes may appear in the compiled regular expression. */
372 /* Succeed right away--no more backtracking. */
375 /* Followed by one byte giving n, then by n literal bytes. */
378 /* Matches any (more or less) character. */
381 /* Matches any one char belonging to specified set. First
382 following byte is number of bitmap bytes. Then come bytes
383 for a bitmap saying which chars are in. Bits in each byte
384 are ordered low-bit-first. A character is in the set if its
385 bit is 1. A character too large to have a bit in the map is
386 automatically not in the set. */
389 /* Same parameters as charset, but match any character that is
390 not one of those specified. */
393 /* Start remembering the text that is matched, for storing in a
394 register. Followed by one byte with the register number, in
395 the range 0 to one less than the pattern buffer's re_nsub
396 field. Then followed by one byte with the number of groups
397 inner to this one. (This last has to be part of the
398 start_memory only because we need it in the on_failure_jump
402 /* Stop remembering the text that is matched and store it in a
403 memory register. Followed by one byte with the register
404 number, in the range 0 to one less than `re_nsub' in the
405 pattern buffer, and one byte with the number of inner groups,
406 just like `start_memory'. (We need the number of inner
407 groups here because we don't have any easy way of finding the
408 corresponding start_memory when we're at a stop_memory.) */
411 /* Match a duplicate of something remembered. Followed by one
412 byte containing the register number. */
415 /* Fail unless at beginning of line. */
418 /* Fail unless at end of line. */
421 /* Succeeds if at beginning of buffer (if emacs) or at beginning
422 of string to be matched (if not). */
425 /* Analogously, for end of buffer/string. */
428 /* Followed by two byte relative address to which to jump. */
431 /* Same as jump, but marks the end of an alternative. */
434 /* Followed by two-byte relative address of place to resume at
435 in case of failure. */
438 /* Like on_failure_jump, but pushes a placeholder instead of the
439 current string position when executed. */
440 on_failure_keep_string_jump,
442 /* Throw away latest failure point and then jump to following
443 two-byte relative address. */
446 /* Change to pop_failure_jump if know won't have to backtrack to
447 match; otherwise change to jump. This is used to jump
448 back to the beginning of a repeat. If what follows this jump
449 clearly won't match what the repeat does, such that we can be
450 sure that there is no use backtracking out of repetitions
451 already matched, then we change it to a pop_failure_jump.
452 Followed by two-byte address. */
455 /* Jump to following two-byte address, and push a dummy failure
456 point. This failure point will be thrown away if an attempt
457 is made to use it for a failure. A `+' construct makes this
458 before the first repeat. Also used as an intermediary kind
459 of jump when compiling an alternative. */
462 /* Push a dummy failure point and continue. Used at the end of
466 /* Followed by two-byte relative address and two-byte number n.
467 After matching N times, jump to the address upon failure. */
470 /* Followed by two-byte relative address, and two-byte number n.
471 Jump to the address N times, then fail. */
474 /* Set the following two-byte relative address to the
475 subsequent two-byte number. The address *includes* the two
479 wordchar, /* Matches any word-constituent character. */
480 notwordchar, /* Matches any char that is not a word-constituent. */
482 wordbeg, /* Succeeds if at word beginning. */
483 wordend, /* Succeeds if at word end. */
485 wordbound, /* Succeeds if at a word boundary. */
486 notwordbound /* Succeeds if not at a word boundary. */
489 ,before_dot, /* Succeeds if before point. */
490 at_dot, /* Succeeds if at point. */
491 after_dot, /* Succeeds if after point. */
493 /* Matches any character whose syntax is specified. Followed by
494 a byte which contains a syntax code, e.g., Sword. */
497 /* Matches any character whose syntax is not that specified. */
502 /* Common operations on the compiled pattern. */
504 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
506 #define STORE_NUMBER(destination, number) \
508 (destination)[0] = (number) & 0377; \
509 (destination)[1] = (number) >> 8; \
512 /* Same as STORE_NUMBER, except increment DESTINATION to
513 the byte after where the number is stored. Therefore, DESTINATION
514 must be an lvalue. */
516 #define STORE_NUMBER_AND_INCR(destination, number) \
518 STORE_NUMBER (destination, number); \
519 (destination) += 2; \
522 /* Put into DESTINATION a number stored in two contiguous bytes starting
525 #define EXTRACT_NUMBER(destination, source) \
527 (destination) = *(source) & 0377; \
528 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
532 static void extract_number _RE_ARGS ((int *dest, unsigned char *source));
534 extract_number (dest, source)
536 unsigned char *source;
538 int temp = SIGN_EXTEND_CHAR (*(source + 1));
539 *dest = *source & 0377;
543 # ifndef EXTRACT_MACROS /* To debug the macros. */
544 # undef EXTRACT_NUMBER
545 # define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
546 # endif /* not EXTRACT_MACROS */
550 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
551 SOURCE must be an lvalue. */
553 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
555 EXTRACT_NUMBER (destination, source); \
560 static void extract_number_and_incr _RE_ARGS ((int *destination,
561 unsigned char **source));
563 extract_number_and_incr (destination, source)
565 unsigned char **source;
567 extract_number (destination, *source);
571 # ifndef EXTRACT_MACROS
572 # undef EXTRACT_NUMBER_AND_INCR
573 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
574 extract_number_and_incr (&dest, &src)
575 # endif /* not EXTRACT_MACROS */
579 /* If DEBUG is defined, Regex prints many voluminous messages about what
580 it is doing (if the variable `debug' is nonzero). If linked with the
581 main program in `iregex.c', you can enter patterns and strings
582 interactively. And if linked with the main program in `main.c' and
583 the other test files, you can run the already-written tests. */
587 /* We use standard I/O for debugging. */
590 /* It is useful to test things that ``must'' be true when debugging. */
593 static int debug = 0;
595 # define DEBUG_STATEMENT(e) e
596 # define DEBUG_PRINT1(x) if (debug) printf (x)
597 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
598 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
599 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
600 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
601 if (debug) print_partial_compiled_pattern (s, e)
602 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
603 if (debug) print_double_string (w, s1, sz1, s2, sz2)
606 /* Print the fastmap in human-readable form. */
609 print_fastmap (fastmap)
612 unsigned was_a_range = 0;
615 while (i < (1 << BYTEWIDTH))
621 while (i < (1 << BYTEWIDTH) && fastmap[i])
637 /* Print a compiled pattern string in human-readable form, starting at
638 the START pointer into it and ending just before the pointer END. */
641 print_partial_compiled_pattern (start, end)
642 unsigned char *start;
647 unsigned char *p = start;
648 unsigned char *pend = end;
656 /* Loop over pattern commands. */
659 printf ("%d:\t", p - start);
661 switch ((re_opcode_t) *p++)
669 printf ("/exactn/%d", mcnt);
680 printf ("/start_memory/%d/%d", mcnt, *p++);
685 printf ("/stop_memory/%d/%d", mcnt, *p++);
689 printf ("/duplicate/%d", *p++);
699 register int c, last = -100;
700 register int in_range = 0;
702 printf ("/charset [%s",
703 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
705 assert (p + *p < pend);
707 for (c = 0; c < 256; c++)
709 && (p[1 + (c/8)] & (1 << (c % 8))))
711 /* Are we starting a range? */
712 if (last + 1 == c && ! in_range)
717 /* Have we broken a range? */
718 else if (last + 1 != c && in_range)
747 case on_failure_jump:
748 extract_number_and_incr (&mcnt, &p);
749 printf ("/on_failure_jump to %d", p + mcnt - start);
752 case on_failure_keep_string_jump:
753 extract_number_and_incr (&mcnt, &p);
754 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
757 case dummy_failure_jump:
758 extract_number_and_incr (&mcnt, &p);
759 printf ("/dummy_failure_jump to %d", p + mcnt - start);
762 case push_dummy_failure:
763 printf ("/push_dummy_failure");
767 extract_number_and_incr (&mcnt, &p);
768 printf ("/maybe_pop_jump to %d", p + mcnt - start);
771 case pop_failure_jump:
772 extract_number_and_incr (&mcnt, &p);
773 printf ("/pop_failure_jump to %d", p + mcnt - start);
777 extract_number_and_incr (&mcnt, &p);
778 printf ("/jump_past_alt to %d", p + mcnt - start);
782 extract_number_and_incr (&mcnt, &p);
783 printf ("/jump to %d", p + mcnt - start);
787 extract_number_and_incr (&mcnt, &p);
789 extract_number_and_incr (&mcnt2, &p);
790 printf ("/succeed_n to %d, %d times", p1 - start, mcnt2);
794 extract_number_and_incr (&mcnt, &p);
796 extract_number_and_incr (&mcnt2, &p);
797 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
801 extract_number_and_incr (&mcnt, &p);
803 extract_number_and_incr (&mcnt2, &p);
804 printf ("/set_number_at location %d to %d", p1 - start, mcnt2);
808 printf ("/wordbound");
812 printf ("/notwordbound");
824 printf ("/before_dot");
832 printf ("/after_dot");
836 printf ("/syntaxspec");
838 printf ("/%d", mcnt);
842 printf ("/notsyntaxspec");
844 printf ("/%d", mcnt);
849 printf ("/wordchar");
853 printf ("/notwordchar");
865 printf ("?%d", *(p-1));
871 printf ("%d:\tend of pattern.\n", p - start);
876 print_compiled_pattern (bufp)
877 struct re_pattern_buffer *bufp;
879 unsigned char *buffer = bufp->buffer;
881 print_partial_compiled_pattern (buffer, buffer + bufp->used);
882 printf ("%ld bytes used/%ld bytes allocated.\n",
883 bufp->used, bufp->allocated);
885 if (bufp->fastmap_accurate && bufp->fastmap)
887 printf ("fastmap: ");
888 print_fastmap (bufp->fastmap);
891 printf ("re_nsub: %d\t", bufp->re_nsub);
892 printf ("regs_alloc: %d\t", bufp->regs_allocated);
893 printf ("can_be_null: %d\t", bufp->can_be_null);
894 printf ("newline_anchor: %d\n", bufp->newline_anchor);
895 printf ("no_sub: %d\t", bufp->no_sub);
896 printf ("not_bol: %d\t", bufp->not_bol);
897 printf ("not_eol: %d\t", bufp->not_eol);
898 printf ("syntax: %lx\n", bufp->syntax);
899 /* Perhaps we should print the translate table? */
904 print_double_string (where, string1, size1, string2, size2)
917 if (FIRST_STRING_P (where))
919 for (this_char = where - string1; this_char < size1; this_char++)
920 putchar (string1[this_char]);
925 for (this_char = where - string2; this_char < size2; this_char++)
926 putchar (string2[this_char]);
937 #else /* not DEBUG */
942 # define DEBUG_STATEMENT(e)
943 # define DEBUG_PRINT1(x)
944 # define DEBUG_PRINT2(x1, x2)
945 # define DEBUG_PRINT3(x1, x2, x3)
946 # define DEBUG_PRINT4(x1, x2, x3, x4)
947 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
948 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
950 #endif /* not DEBUG */
952 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
953 also be assigned to arbitrarily: each pattern buffer stores its own
954 syntax, so it can be changed between regex compilations. */
955 /* This has no initializer because initialized variables in Emacs
956 become read-only after dumping. */
957 reg_syntax_t re_syntax_options;
960 /* Specify the precise syntax of regexps for compilation. This provides
961 for compatibility for various utilities which historically have
962 different, incompatible syntaxes.
964 The argument SYNTAX is a bit mask comprised of the various bits
965 defined in gnu-regex.h. We return the old syntax. */
968 re_set_syntax (syntax)
971 reg_syntax_t ret = re_syntax_options;
973 re_syntax_options = syntax;
975 if (syntax & RE_DEBUG)
977 else if (debug) /* was on but now is not */
983 weak_alias (__re_set_syntax, re_set_syntax)
986 /* This table gives an error message for each of the error codes listed
987 in gnu-regex.h. Obviously the order here has to be same as there.
988 POSIX doesn't require that we do anything for REG_NOERROR,
989 but why not be nice? */
991 static const char *re_error_msgid[] =
993 gettext_noop ("Success"), /* REG_NOERROR */
994 gettext_noop ("No match"), /* REG_NOMATCH */
995 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
996 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
997 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
998 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
999 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
1000 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
1001 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
1002 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
1003 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
1004 gettext_noop ("Invalid range end"), /* REG_ERANGE */
1005 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
1006 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
1007 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
1008 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
1009 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
1012 /* Avoiding alloca during matching, to placate r_alloc. */
1014 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1015 searching and matching functions should not call alloca. On some
1016 systems, alloca is implemented in terms of malloc, and if we're
1017 using the relocating allocator routines, then malloc could cause a
1018 relocation, which might (if the strings being searched are in the
1019 ralloc heap) shift the data out from underneath the regexp
1022 Here's another reason to avoid allocation: Emacs
1023 processes input from X in a signal handler; processing X input may
1024 call malloc; if input arrives while a matching routine is calling
1025 malloc, then we're scrod. But Emacs can't just block input while
1026 calling matching routines; then we don't notice interrupts when
1027 they come in. So, Emacs blocks input around all regexp calls
1028 except the matching calls, which it leaves unprotected, in the
1029 faith that they will not malloc. */
1031 /* Normally, this is fine. */
1032 #define MATCH_MAY_ALLOCATE
1034 /* When using GNU C, we are not REALLY using the C alloca, no matter
1035 what config.h may say. So don't take precautions for it. */
1040 /* The match routines may not allocate if (1) they would do it with malloc
1041 and (2) it's not safe for them to use malloc.
1042 Note that if REL_ALLOC is defined, matching would not use malloc for the
1043 failure stack, but we would still use it for the register vectors;
1044 so REL_ALLOC should not affect this. */
1045 #if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1046 # undef MATCH_MAY_ALLOCATE
1050 /* Failure stack declarations and macros; both re_compile_fastmap and
1051 re_match_2 use a failure stack. These have to be macros because of
1052 REGEX_ALLOCATE_STACK. */
1055 /* Number of failure points for which to initially allocate space
1056 when matching. If this number is exceeded, we allocate more
1057 space, so it is not a hard limit. */
1058 #ifndef INIT_FAILURE_ALLOC
1059 # define INIT_FAILURE_ALLOC 5
1062 /* Roughly the maximum number of failure points on the stack. Would be
1063 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1064 This is a variable only so users of regex can assign to it; we never
1065 change it ourselves. */
1069 # if defined MATCH_MAY_ALLOCATE
1070 /* 4400 was enough to cause a crash on Alpha OSF/1,
1071 whose default stack limit is 2mb. */
1072 long int re_max_failures = 4000;
1074 long int re_max_failures = 2000;
1077 union fail_stack_elt
1079 unsigned char *pointer;
1083 typedef union fail_stack_elt fail_stack_elt_t;
1087 fail_stack_elt_t *stack;
1088 unsigned long int size;
1089 unsigned long int avail; /* Offset of next open position. */
1092 #else /* not INT_IS_16BIT */
1094 # if defined MATCH_MAY_ALLOCATE
1095 /* 4400 was enough to cause a crash on Alpha OSF/1,
1096 whose default stack limit is 2mb. */
1097 int re_max_failures = 20000;
1099 int re_max_failures = 2000;
1102 union fail_stack_elt
1104 unsigned char *pointer;
1108 typedef union fail_stack_elt fail_stack_elt_t;
1112 fail_stack_elt_t *stack;
1114 unsigned avail; /* Offset of next open position. */
1117 #endif /* INT_IS_16BIT */
1119 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1120 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1121 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1124 /* Define macros to initialize and free the failure stack.
1125 Do `return -2' if the alloc fails. */
1127 #ifdef MATCH_MAY_ALLOCATE
1128 # define INIT_FAIL_STACK() \
1130 fail_stack.stack = (fail_stack_elt_t *) \
1131 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1133 if (fail_stack.stack == NULL) \
1136 fail_stack.size = INIT_FAILURE_ALLOC; \
1137 fail_stack.avail = 0; \
1140 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1142 # define INIT_FAIL_STACK() \
1144 fail_stack.avail = 0; \
1147 # define RESET_FAIL_STACK()
1151 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1153 Return 1 if succeeds, and 0 if either ran out of memory
1154 allocating space for it or it was already too large.
1156 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1158 #define DOUBLE_FAIL_STACK(fail_stack) \
1159 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1161 : ((fail_stack).stack = (fail_stack_elt_t *) \
1162 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1163 (fail_stack).size * sizeof (fail_stack_elt_t), \
1164 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1166 (fail_stack).stack == NULL \
1168 : ((fail_stack).size <<= 1, \
1172 /* Push pointer POINTER on FAIL_STACK.
1173 Return 1 if was able to do so and 0 if ran out of memory allocating
1175 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1176 ((FAIL_STACK_FULL () \
1177 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1179 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1182 /* Push a pointer value onto the failure stack.
1183 Assumes the variable `fail_stack'. Probably should only
1184 be called from within `PUSH_FAILURE_POINT'. */
1185 #define PUSH_FAILURE_POINTER(item) \
1186 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1188 /* This pushes an integer-valued item onto the failure stack.
1189 Assumes the variable `fail_stack'. Probably should only
1190 be called from within `PUSH_FAILURE_POINT'. */
1191 #define PUSH_FAILURE_INT(item) \
1192 fail_stack.stack[fail_stack.avail++].integer = (item)
1194 /* Push a fail_stack_elt_t value onto the failure stack.
1195 Assumes the variable `fail_stack'. Probably should only
1196 be called from within `PUSH_FAILURE_POINT'. */
1197 #define PUSH_FAILURE_ELT(item) \
1198 fail_stack.stack[fail_stack.avail++] = (item)
1200 /* These three POP... operations complement the three PUSH... operations.
1201 All assume that `fail_stack' is nonempty. */
1202 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1203 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1204 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1206 /* Used to omit pushing failure point id's when we're not debugging. */
1208 # define DEBUG_PUSH PUSH_FAILURE_INT
1209 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1211 # define DEBUG_PUSH(item)
1212 # define DEBUG_POP(item_addr)
1216 /* Push the information about the state we will need
1217 if we ever fail back to it.
1219 Requires variables fail_stack, regstart, regend, reg_info, and
1220 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1223 Does `return FAILURE_CODE' if runs out of memory. */
1225 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1227 char *destination; \
1228 /* Must be int, so when we don't save any registers, the arithmetic \
1229 of 0 + -1 isn't done as unsigned. */ \
1230 /* Can't be int, since there is not a shred of a guarantee that int \
1231 is wide enough to hold a value of something to which pointer can \
1233 active_reg_t this_reg; \
1235 DEBUG_STATEMENT (failure_id++); \
1236 DEBUG_STATEMENT (nfailure_points_pushed++); \
1237 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1238 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1239 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1241 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1242 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1244 /* Ensure we have enough space allocated for what we will push. */ \
1245 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1247 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1248 return failure_code; \
1250 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1251 (fail_stack).size); \
1252 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1255 /* Push the info, starting with the registers. */ \
1256 DEBUG_PRINT1 ("\n"); \
1259 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1262 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1263 DEBUG_STATEMENT (num_regs_pushed++); \
1265 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1266 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1268 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1269 PUSH_FAILURE_POINTER (regend[this_reg]); \
1271 DEBUG_PRINT2 (" info: %p\n ", \
1272 reg_info[this_reg].word.pointer); \
1273 DEBUG_PRINT2 (" match_null=%d", \
1274 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1275 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1276 DEBUG_PRINT2 (" matched_something=%d", \
1277 MATCHED_SOMETHING (reg_info[this_reg])); \
1278 DEBUG_PRINT2 (" ever_matched=%d", \
1279 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1280 DEBUG_PRINT1 ("\n"); \
1281 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1284 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1285 PUSH_FAILURE_INT (lowest_active_reg); \
1287 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1288 PUSH_FAILURE_INT (highest_active_reg); \
1290 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1291 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1292 PUSH_FAILURE_POINTER (pattern_place); \
1294 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1295 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1297 DEBUG_PRINT1 ("'\n"); \
1298 PUSH_FAILURE_POINTER (string_place); \
1300 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1301 DEBUG_PUSH (failure_id); \
1304 /* This is the number of items that are pushed and popped on the stack
1305 for each register. */
1306 #define NUM_REG_ITEMS 3
1308 /* Individual items aside from the registers. */
1310 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1312 # define NUM_NONREG_ITEMS 4
1315 /* We push at most this many items on the stack. */
1316 /* We used to use (num_regs - 1), which is the number of registers
1317 this regexp will save; but that was changed to 5
1318 to avoid stack overflow for a regexp with lots of parens. */
1319 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1321 /* We actually push this many items. */
1322 #define NUM_FAILURE_ITEMS \
1324 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1328 /* How many items can still be added to the stack without overflowing it. */
1329 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1332 /* Pops what PUSH_FAIL_STACK pushes.
1334 We restore into the parameters, all of which should be lvalues:
1335 STR -- the saved data position.
1336 PAT -- the saved pattern position.
1337 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1338 REGSTART, REGEND -- arrays of string positions.
1339 REG_INFO -- array of information about each subexpression.
1341 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1342 `pend', `string1', `size1', `string2', and `size2'. */
1344 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1346 DEBUG_STATEMENT (unsigned failure_id;) \
1347 active_reg_t this_reg; \
1348 const unsigned char *string_temp; \
1350 assert (!FAIL_STACK_EMPTY ()); \
1352 /* Remove failure points and point to how many regs pushed. */ \
1353 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1354 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1355 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1357 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1359 DEBUG_POP (&failure_id); \
1360 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1362 /* If the saved string location is NULL, it came from an \
1363 on_failure_keep_string_jump opcode, and we want to throw away the \
1364 saved NULL, thus retaining our current position in the string. */ \
1365 string_temp = POP_FAILURE_POINTER (); \
1366 if (string_temp != NULL) \
1367 str = (const char *) string_temp; \
1369 DEBUG_PRINT2 (" Popping string %p: `", str); \
1370 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1371 DEBUG_PRINT1 ("'\n"); \
1373 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1374 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1375 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1377 /* Restore register info. */ \
1378 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1379 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1381 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1382 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1385 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1387 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1389 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1390 DEBUG_PRINT2 (" info: %p\n", \
1391 reg_info[this_reg].word.pointer); \
1393 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1394 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1396 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1397 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1401 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1403 reg_info[this_reg].word.integer = 0; \
1404 regend[this_reg] = 0; \
1405 regstart[this_reg] = 0; \
1407 highest_active_reg = high_reg; \
1410 set_regs_matched_done = 0; \
1411 DEBUG_STATEMENT (nfailure_points_popped++); \
1412 } /* POP_FAILURE_POINT */
1416 /* Structure for per-register (a.k.a. per-group) information.
1417 Other register information, such as the
1418 starting and ending positions (which are addresses), and the list of
1419 inner groups (which is a bits list) are maintained in separate
1422 We are making a (strictly speaking) nonportable assumption here: that
1423 the compiler will pack our bit fields into something that fits into
1424 the type of `word', i.e., is something that fits into one item on the
1428 /* Declarations and macros for re_match_2. */
1432 fail_stack_elt_t word;
1435 /* This field is one if this group can match the empty string,
1436 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1437 #define MATCH_NULL_UNSET_VALUE 3
1438 unsigned match_null_string_p : 2;
1439 unsigned is_active : 1;
1440 unsigned matched_something : 1;
1441 unsigned ever_matched_something : 1;
1443 } register_info_type;
1445 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1446 #define IS_ACTIVE(R) ((R).bits.is_active)
1447 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1448 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1451 /* Call this when have matched a real character; it sets `matched' flags
1452 for the subexpressions which we are currently inside. Also records
1453 that those subexprs have matched. */
1454 #define SET_REGS_MATCHED() \
1457 if (!set_regs_matched_done) \
1460 set_regs_matched_done = 1; \
1461 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1463 MATCHED_SOMETHING (reg_info[r]) \
1464 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1471 /* Registers are set to a sentinel when they haven't yet matched. */
1472 static char reg_unset_dummy;
1473 #define REG_UNSET_VALUE (®_unset_dummy)
1474 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1476 /* Subroutine declarations and macros for regex_compile. */
1478 static reg_errcode_t regex_compile _RE_ARGS ((const char *pattern, size_t size,
1479 reg_syntax_t syntax,
1480 struct re_pattern_buffer *bufp));
1481 static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg));
1482 static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1483 int arg1, int arg2));
1484 static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1485 int arg, unsigned char *end));
1486 static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1487 int arg1, int arg2, unsigned char *end));
1488 static boolean at_begline_loc_p _RE_ARGS ((const char *pattern, const char *p,
1489 reg_syntax_t syntax));
1490 static boolean at_endline_loc_p _RE_ARGS ((const char *p, const char *pend,
1491 reg_syntax_t syntax));
1492 static reg_errcode_t compile_range _RE_ARGS ((const char **p_ptr,
1495 reg_syntax_t syntax,
1498 /* Fetch the next character in the uncompiled pattern---translating it
1499 if necessary. Also cast from a signed character in the constant
1500 string passed to us by the user to an unsigned char that we can use
1501 as an array index (in, e.g., `translate'). */
1503 # define PATFETCH(c) \
1504 do {if (p == pend) return REG_EEND; \
1505 c = (unsigned char) *p++; \
1506 if (translate) c = (unsigned char) translate[c]; \
1510 /* Fetch the next character in the uncompiled pattern, with no
1512 #define PATFETCH_RAW(c) \
1513 do {if (p == pend) return REG_EEND; \
1514 c = (unsigned char) *p++; \
1517 /* Go backwards one character in the pattern. */
1518 #define PATUNFETCH p--
1521 /* If `translate' is non-null, return translate[D], else just D. We
1522 cast the subscript to translate because some data is declared as
1523 `char *', to avoid warnings when a string constant is passed. But
1524 when we use a character as a subscript we must make it unsigned. */
1526 # define TRANSLATE(d) \
1527 (translate ? (char) translate[(unsigned char) (d)] : (d))
1531 /* Macros for outputting the compiled pattern into `buffer'. */
1533 /* If the buffer isn't allocated when it comes in, use this. */
1534 #define INIT_BUF_SIZE 32
1536 /* Make sure we have at least N more bytes of space in buffer. */
1537 #define GET_BUFFER_SPACE(n) \
1538 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1541 /* Make sure we have one more byte of buffer space and then add C to it. */
1542 #define BUF_PUSH(c) \
1544 GET_BUFFER_SPACE (1); \
1545 *b++ = (unsigned char) (c); \
1549 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1550 #define BUF_PUSH_2(c1, c2) \
1552 GET_BUFFER_SPACE (2); \
1553 *b++ = (unsigned char) (c1); \
1554 *b++ = (unsigned char) (c2); \
1558 /* As with BUF_PUSH_2, except for three bytes. */
1559 #define BUF_PUSH_3(c1, c2, c3) \
1561 GET_BUFFER_SPACE (3); \
1562 *b++ = (unsigned char) (c1); \
1563 *b++ = (unsigned char) (c2); \
1564 *b++ = (unsigned char) (c3); \
1568 /* Store a jump with opcode OP at LOC to location TO. We store a
1569 relative address offset by the three bytes the jump itself occupies. */
1570 #define STORE_JUMP(op, loc, to) \
1571 store_op1 (op, loc, (int) ((to) - (loc) - 3))
1573 /* Likewise, for a two-argument jump. */
1574 #define STORE_JUMP2(op, loc, to, arg) \
1575 store_op2 (op, loc, (int) ((to) - (loc) - 3), arg)
1577 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1578 #define INSERT_JUMP(op, loc, to) \
1579 insert_op1 (op, loc, (int) ((to) - (loc) - 3), b)
1581 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1582 #define INSERT_JUMP2(op, loc, to, arg) \
1583 insert_op2 (op, loc, (int) ((to) - (loc) - 3), arg, b)
1586 /* This is not an arbitrary limit: the arguments which represent offsets
1587 into the pattern are two bytes long. So if 2^16 bytes turns out to
1588 be too small, many things would have to change. */
1589 /* Any other compiler which, like MSC, has allocation limit below 2^16
1590 bytes will have to use approach similar to what was done below for
1591 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1592 reallocating to 0 bytes. Such thing is not going to work too well.
1593 You have been warned!! */
1594 #if defined _MSC_VER && !defined WIN32
1595 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
1596 The REALLOC define eliminates a flurry of conversion warnings,
1597 but is not required. */
1598 # define MAX_BUF_SIZE 65500L
1599 # define REALLOC(p,s) realloc ((p), (size_t) (s))
1601 # define MAX_BUF_SIZE (1L << 16)
1602 # define REALLOC(p,s) realloc ((p), (s))
1605 /* Extend the buffer by twice its current size via realloc and
1606 reset the pointers that pointed into the old block to point to the
1607 correct places in the new one. If extending the buffer results in it
1608 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1609 #define EXTEND_BUFFER() \
1611 unsigned char *old_buffer = bufp->buffer; \
1612 if (bufp->allocated == MAX_BUF_SIZE) \
1614 bufp->allocated <<= 1; \
1615 if (bufp->allocated > MAX_BUF_SIZE) \
1616 bufp->allocated = MAX_BUF_SIZE; \
1617 bufp->buffer = (unsigned char *) REALLOC (bufp->buffer, bufp->allocated);\
1618 if (bufp->buffer == NULL) \
1619 return REG_ESPACE; \
1620 /* If the buffer moved, move all the pointers into it. */ \
1621 if (old_buffer != bufp->buffer) \
1623 b = (b - old_buffer) + bufp->buffer; \
1624 begalt = (begalt - old_buffer) + bufp->buffer; \
1625 if (fixup_alt_jump) \
1626 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1628 laststart = (laststart - old_buffer) + bufp->buffer; \
1629 if (pending_exact) \
1630 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1635 /* Since we have one byte reserved for the register number argument to
1636 {start,stop}_memory, the maximum number of groups we can report
1637 things about is what fits in that byte. */
1638 #define MAX_REGNUM 255
1640 /* But patterns can have more than `MAX_REGNUM' registers. We just
1641 ignore the excess. */
1642 typedef unsigned regnum_t;
1645 /* Macros for the compile stack. */
1647 /* Since offsets can go either forwards or backwards, this type needs to
1648 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1649 /* int may be not enough when sizeof(int) == 2. */
1650 typedef long pattern_offset_t;
1654 pattern_offset_t begalt_offset;
1655 pattern_offset_t fixup_alt_jump;
1656 pattern_offset_t inner_group_offset;
1657 pattern_offset_t laststart_offset;
1659 } compile_stack_elt_t;
1664 compile_stack_elt_t *stack;
1666 unsigned avail; /* Offset of next open position. */
1667 } compile_stack_type;
1670 #define INIT_COMPILE_STACK_SIZE 32
1672 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1673 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1675 /* The next available element. */
1676 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1679 /* Set the bit for character C in a list. */
1680 #define SET_LIST_BIT(c) \
1681 (b[((unsigned char) (c)) / BYTEWIDTH] \
1682 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1685 /* Get the next unsigned number in the uncompiled pattern. */
1686 #define GET_UNSIGNED_NUMBER(num) \
1690 while (ISDIGIT (c)) \
1694 num = num * 10 + c - '0'; \
1702 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
1703 /* The GNU C library provides support for user-defined character classes
1704 and the functions from ISO C amendement 1. */
1705 # ifdef CHARCLASS_NAME_MAX
1706 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
1708 /* This shouldn't happen but some implementation might still have this
1709 problem. Use a reasonable default value. */
1710 # define CHAR_CLASS_MAX_LENGTH 256
1714 # define IS_CHAR_CLASS(string) __wctype (string)
1716 # define IS_CHAR_CLASS(string) wctype (string)
1719 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1721 # define IS_CHAR_CLASS(string) \
1722 (STREQ (string, "alpha") || STREQ (string, "upper") \
1723 || STREQ (string, "lower") || STREQ (string, "digit") \
1724 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1725 || STREQ (string, "space") || STREQ (string, "print") \
1726 || STREQ (string, "punct") || STREQ (string, "graph") \
1727 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1730 #ifndef MATCH_MAY_ALLOCATE
1732 /* If we cannot allocate large objects within re_match_2_internal,
1733 we make the fail stack and register vectors global.
1734 The fail stack, we grow to the maximum size when a regexp
1736 The register vectors, we adjust in size each time we
1737 compile a regexp, according to the number of registers it needs. */
1739 static fail_stack_type fail_stack;
1741 /* Size with which the following vectors are currently allocated.
1742 That is so we can make them bigger as needed,
1743 but never make them smaller. */
1744 static int regs_allocated_size;
1746 static const char ** regstart, ** regend;
1747 static const char ** old_regstart, ** old_regend;
1748 static const char **best_regstart, **best_regend;
1749 static register_info_type *reg_info;
1750 static const char **reg_dummy;
1751 static register_info_type *reg_info_dummy;
1753 /* Make the register vectors big enough for NUM_REGS registers,
1754 but don't make them smaller. */
1757 regex_grow_registers (num_regs)
1760 if (num_regs > regs_allocated_size)
1762 RETALLOC_IF (regstart, num_regs, const char *);
1763 RETALLOC_IF (regend, num_regs, const char *);
1764 RETALLOC_IF (old_regstart, num_regs, const char *);
1765 RETALLOC_IF (old_regend, num_regs, const char *);
1766 RETALLOC_IF (best_regstart, num_regs, const char *);
1767 RETALLOC_IF (best_regend, num_regs, const char *);
1768 RETALLOC_IF (reg_info, num_regs, register_info_type);
1769 RETALLOC_IF (reg_dummy, num_regs, const char *);
1770 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1772 regs_allocated_size = num_regs;
1776 #endif /* not MATCH_MAY_ALLOCATE */
1778 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
1782 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1783 Returns one of error codes defined in `gnu-regex.h', or zero for success.
1785 Assumes the `allocated' (and perhaps `buffer') and `translate'
1786 fields are set in BUFP on entry.
1788 If it succeeds, results are put in BUFP (if it returns an error, the
1789 contents of BUFP are undefined):
1790 `buffer' is the compiled pattern;
1791 `syntax' is set to SYNTAX;
1792 `used' is set to the length of the compiled pattern;
1793 `fastmap_accurate' is zero;
1794 `re_nsub' is the number of subexpressions in PATTERN;
1795 `not_bol' and `not_eol' are zero;
1797 The `fastmap' and `newline_anchor' fields are neither
1798 examined nor set. */
1800 /* Return, freeing storage we allocated. */
1801 #define FREE_STACK_RETURN(value) \
1802 return (free (compile_stack.stack), value)
1804 static reg_errcode_t
1805 regex_compile (pattern, size, syntax, bufp)
1806 const char *pattern;
1808 reg_syntax_t syntax;
1809 struct re_pattern_buffer *bufp;
1811 /* We fetch characters from PATTERN here. Even though PATTERN is
1812 `char *' (i.e., signed), we declare these variables as unsigned, so
1813 they can be reliably used as array indices. */
1814 register unsigned char c, c1;
1816 /* A random temporary spot in PATTERN. */
1819 /* Points to the end of the buffer, where we should append. */
1820 register unsigned char *b;
1822 /* Keeps track of unclosed groups. */
1823 compile_stack_type compile_stack;
1825 /* Points to the current (ending) position in the pattern. */
1826 const char *p = pattern;
1827 const char *pend = pattern + size;
1829 /* How to translate the characters in the pattern. */
1830 RE_TRANSLATE_TYPE translate = bufp->translate;
1832 /* Address of the count-byte of the most recently inserted `exactn'
1833 command. This makes it possible to tell if a new exact-match
1834 character can be added to that command or if the character requires
1835 a new `exactn' command. */
1836 unsigned char *pending_exact = 0;
1838 /* Address of start of the most recently finished expression.
1839 This tells, e.g., postfix * where to find the start of its
1840 operand. Reset at the beginning of groups and alternatives. */
1841 unsigned char *laststart = 0;
1843 /* Address of beginning of regexp, or inside of last group. */
1844 unsigned char *begalt;
1846 /* Place in the uncompiled pattern (i.e., the {) to
1847 which to go back if the interval is invalid. */
1848 const char *beg_interval;
1850 /* Address of the place where a forward jump should go to the end of
1851 the containing expression. Each alternative of an `or' -- except the
1852 last -- ends with a forward jump of this sort. */
1853 unsigned char *fixup_alt_jump = 0;
1855 /* Counts open-groups as they are encountered. Remembered for the
1856 matching close-group on the compile stack, so the same register
1857 number is put in the stop_memory as the start_memory. */
1858 regnum_t regnum = 0;
1861 DEBUG_PRINT1 ("\nCompiling pattern: ");
1864 unsigned debug_count;
1866 for (debug_count = 0; debug_count < size; debug_count++)
1867 putchar (pattern[debug_count]);
1872 /* Initialize the compile stack. */
1873 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1874 if (compile_stack.stack == NULL)
1877 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1878 compile_stack.avail = 0;
1880 /* Initialize the pattern buffer. */
1881 bufp->syntax = syntax;
1882 bufp->fastmap_accurate = 0;
1883 bufp->not_bol = bufp->not_eol = 0;
1885 /* Set `used' to zero, so that if we return an error, the pattern
1886 printer (for debugging) will think there's no pattern. We reset it
1890 /* Always count groups, whether or not bufp->no_sub is set. */
1893 #if !defined emacs && !defined SYNTAX_TABLE
1894 /* Initialize the syntax table. */
1895 init_syntax_once ();
1898 if (bufp->allocated == 0)
1901 { /* If zero allocated, but buffer is non-null, try to realloc
1902 enough space. This loses if buffer's address is bogus, but
1903 that is the user's responsibility. */
1904 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1907 { /* Caller did not allocate a buffer. Do it for them. */
1908 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1910 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1912 bufp->allocated = INIT_BUF_SIZE;
1915 begalt = b = bufp->buffer;
1917 /* Loop through the uncompiled pattern until we're at the end. */
1926 if ( /* If at start of pattern, it's an operator. */
1928 /* If context independent, it's an operator. */
1929 || syntax & RE_CONTEXT_INDEP_ANCHORS
1930 /* Otherwise, depends on what's come before. */
1931 || at_begline_loc_p (pattern, p, syntax))
1941 if ( /* If at end of pattern, it's an operator. */
1943 /* If context independent, it's an operator. */
1944 || syntax & RE_CONTEXT_INDEP_ANCHORS
1945 /* Otherwise, depends on what's next. */
1946 || at_endline_loc_p (p, pend, syntax))
1956 if ((syntax & RE_BK_PLUS_QM)
1957 || (syntax & RE_LIMITED_OPS))
1961 /* If there is no previous pattern... */
1964 if (syntax & RE_CONTEXT_INVALID_OPS)
1965 FREE_STACK_RETURN (REG_BADRPT);
1966 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1971 /* Are we optimizing this jump? */
1972 boolean keep_string_p = false;
1974 /* 1 means zero (many) matches is allowed. */
1975 char zero_times_ok = 0, many_times_ok = 0;
1977 /* If there is a sequence of repetition chars, collapse it
1978 down to just one (the right one). We can't combine
1979 interval operators with these because of, e.g., `a{2}*',
1980 which should only match an even number of `a's. */
1984 zero_times_ok |= c != '+';
1985 many_times_ok |= c != '?';
1993 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
1996 else if (syntax & RE_BK_PLUS_QM && c == '\\')
1998 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2001 if (!(c1 == '+' || c1 == '?'))
2016 /* If we get here, we found another repeat character. */
2019 /* Star, etc. applied to an empty pattern is equivalent
2020 to an empty pattern. */
2024 /* Now we know whether or not zero matches is allowed
2025 and also whether or not two or more matches is allowed. */
2027 { /* More than one repetition is allowed, so put in at the
2028 end a backward relative jump from `b' to before the next
2029 jump we're going to put in below (which jumps from
2030 laststart to after this jump).
2032 But if we are at the `*' in the exact sequence `.*\n',
2033 insert an unconditional jump backwards to the .,
2034 instead of the beginning of the loop. This way we only
2035 push a failure point once, instead of every time
2036 through the loop. */
2037 assert (p - 1 > pattern);
2039 /* Allocate the space for the jump. */
2040 GET_BUFFER_SPACE (3);
2042 /* We know we are not at the first character of the pattern,
2043 because laststart was nonzero. And we've already
2044 incremented `p', by the way, to be the character after
2045 the `*'. Do we have to do something analogous here
2046 for null bytes, because of RE_DOT_NOT_NULL? */
2047 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2049 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2050 && !(syntax & RE_DOT_NEWLINE))
2051 { /* We have .*\n. */
2052 STORE_JUMP (jump, b, laststart);
2053 keep_string_p = true;
2056 /* Anything else. */
2057 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
2059 /* We've added more stuff to the buffer. */
2063 /* On failure, jump from laststart to b + 3, which will be the
2064 end of the buffer after this jump is inserted. */
2065 GET_BUFFER_SPACE (3);
2066 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2074 /* At least one repetition is required, so insert a
2075 `dummy_failure_jump' before the initial
2076 `on_failure_jump' instruction of the loop. This
2077 effects a skip over that instruction the first time
2078 we hit that loop. */
2079 GET_BUFFER_SPACE (3);
2080 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
2095 boolean had_char_class = false;
2097 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2099 /* Ensure that we have enough space to push a charset: the
2100 opcode, the length count, and the bitset; 34 bytes in all. */
2101 GET_BUFFER_SPACE (34);
2105 /* We test `*p == '^' twice, instead of using an if
2106 statement, so we only need one BUF_PUSH. */
2107 BUF_PUSH (*p == '^' ? charset_not : charset);
2111 /* Remember the first position in the bracket expression. */
2114 /* Push the number of bytes in the bitmap. */
2115 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
2117 /* Clear the whole map. */
2118 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
2120 /* charset_not matches newline according to a syntax bit. */
2121 if ((re_opcode_t) b[-2] == charset_not
2122 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2123 SET_LIST_BIT ('\n');
2125 /* Read in characters and ranges, setting map bits. */
2128 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2132 /* \ might escape characters inside [...] and [^...]. */
2133 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2135 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2142 /* Could be the end of the bracket expression. If it's
2143 not (i.e., when the bracket expression is `[]' so
2144 far), the ']' character bit gets set way below. */
2145 if (c == ']' && p != p1 + 1)
2148 /* Look ahead to see if it's a range when the last thing
2149 was a character class. */
2150 if (had_char_class && c == '-' && *p != ']')
2151 FREE_STACK_RETURN (REG_ERANGE);
2153 /* Look ahead to see if it's a range when the last thing
2154 was a character: if this is a hyphen not at the
2155 beginning or the end of a list, then it's the range
2158 && !(p - 2 >= pattern && p[-2] == '[')
2159 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2163 = compile_range (&p, pend, translate, syntax, b);
2164 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2167 else if (p[0] == '-' && p[1] != ']')
2168 { /* This handles ranges made up of characters only. */
2171 /* Move past the `-'. */
2174 ret = compile_range (&p, pend, translate, syntax, b);
2175 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2178 /* See if we're at the beginning of a possible character
2181 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2182 { /* Leave room for the null. */
2183 char str[CHAR_CLASS_MAX_LENGTH + 1];
2188 /* If pattern is `[[:'. */
2189 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2194 if ((c == ':' && *p == ']') || p == pend
2195 || c1 == CHAR_CLASS_MAX_LENGTH)
2201 /* If isn't a word bracketed by `[:' and `:]':
2202 undo the ending character, the letters, and leave
2203 the leading `:' and `[' (but set bits for them). */
2204 if (c == ':' && *p == ']')
2206 /* CYGNUS LOCAL: Skip this code if we don't have btowc(). btowc() is */
2207 /* defined in the 1994 Amendment 1 to ISO C and may not be present on */
2208 /* systems where we have wchar.h and wctype.h. */
2209 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H && defined HAVE_BTOWC)
2210 boolean is_lower = STREQ (str, "lower");
2211 boolean is_upper = STREQ (str, "upper");
2215 wt = IS_CHAR_CLASS (str);
2217 FREE_STACK_RETURN (REG_ECTYPE);
2219 /* Throw away the ] at the end of the character
2223 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2225 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
2228 if (__iswctype (__btowc (ch), wt))
2231 if (iswctype (btowc (ch), wt))
2235 if (translate && (is_upper || is_lower)
2236 && (ISUPPER (ch) || ISLOWER (ch)))
2240 had_char_class = true;
2243 boolean is_alnum = STREQ (str, "alnum");
2244 boolean is_alpha = STREQ (str, "alpha");
2245 boolean is_blank = STREQ (str, "blank");
2246 boolean is_cntrl = STREQ (str, "cntrl");
2247 boolean is_digit = STREQ (str, "digit");
2248 boolean is_graph = STREQ (str, "graph");
2249 boolean is_lower = STREQ (str, "lower");
2250 boolean is_print = STREQ (str, "print");
2251 boolean is_punct = STREQ (str, "punct");
2252 boolean is_space = STREQ (str, "space");
2253 boolean is_upper = STREQ (str, "upper");
2254 boolean is_xdigit = STREQ (str, "xdigit");
2256 if (!IS_CHAR_CLASS (str))
2257 FREE_STACK_RETURN (REG_ECTYPE);
2259 /* Throw away the ] at the end of the character
2263 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2265 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2267 /* This was split into 3 if's to
2268 avoid an arbitrary limit in some compiler. */
2269 if ( (is_alnum && ISALNUM (ch))
2270 || (is_alpha && ISALPHA (ch))
2271 || (is_blank && ISBLANK (ch))
2272 || (is_cntrl && ISCNTRL (ch)))
2274 if ( (is_digit && ISDIGIT (ch))
2275 || (is_graph && ISGRAPH (ch))
2276 || (is_lower && ISLOWER (ch))
2277 || (is_print && ISPRINT (ch)))
2279 if ( (is_punct && ISPUNCT (ch))
2280 || (is_space && ISSPACE (ch))
2281 || (is_upper && ISUPPER (ch))
2282 || (is_xdigit && ISXDIGIT (ch)))
2284 if ( translate && (is_upper || is_lower)
2285 && (ISUPPER (ch) || ISLOWER (ch)))
2288 had_char_class = true;
2289 #endif /* libc || wctype.h */
2298 had_char_class = false;
2303 had_char_class = false;
2308 /* Discard any (non)matching list bytes that are all 0 at the
2309 end of the map. Decrease the map-length byte too. */
2310 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2318 if (syntax & RE_NO_BK_PARENS)
2325 if (syntax & RE_NO_BK_PARENS)
2332 if (syntax & RE_NEWLINE_ALT)
2339 if (syntax & RE_NO_BK_VBAR)
2346 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2347 goto handle_interval;
2353 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2355 /* Do not translate the character after the \, so that we can
2356 distinguish, e.g., \B from \b, even if we normally would
2357 translate, e.g., B to b. */
2363 if (syntax & RE_NO_BK_PARENS)
2364 goto normal_backslash;
2370 if (COMPILE_STACK_FULL)
2372 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2373 compile_stack_elt_t);
2374 if (compile_stack.stack == NULL) return REG_ESPACE;
2376 compile_stack.size <<= 1;
2379 /* These are the values to restore when we hit end of this
2380 group. They are all relative offsets, so that if the
2381 whole pattern moves because of realloc, they will still
2383 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2384 COMPILE_STACK_TOP.fixup_alt_jump
2385 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2386 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2387 COMPILE_STACK_TOP.regnum = regnum;
2389 /* We will eventually replace the 0 with the number of
2390 groups inner to this one. But do not push a
2391 start_memory for groups beyond the last one we can
2392 represent in the compiled pattern. */
2393 if (regnum <= MAX_REGNUM)
2395 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2396 BUF_PUSH_3 (start_memory, regnum, 0);
2399 compile_stack.avail++;
2404 /* If we've reached MAX_REGNUM groups, then this open
2405 won't actually generate any code, so we'll have to
2406 clear pending_exact explicitly. */
2412 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2414 if (COMPILE_STACK_EMPTY)
2416 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2417 goto normal_backslash;
2419 FREE_STACK_RETURN (REG_ERPAREN);
2424 { /* Push a dummy failure point at the end of the
2425 alternative for a possible future
2426 `pop_failure_jump' to pop. See comments at
2427 `push_dummy_failure' in `re_match_2'. */
2428 BUF_PUSH (push_dummy_failure);
2430 /* We allocated space for this jump when we assigned
2431 to `fixup_alt_jump', in the `handle_alt' case below. */
2432 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2435 /* See similar code for backslashed left paren above. */
2436 if (COMPILE_STACK_EMPTY)
2438 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2441 FREE_STACK_RETURN (REG_ERPAREN);
2444 /* Since we just checked for an empty stack above, this
2445 ``can't happen''. */
2446 assert (compile_stack.avail != 0);
2448 /* We don't just want to restore into `regnum', because
2449 later groups should continue to be numbered higher,
2450 as in `(ab)c(de)' -- the second group is #2. */
2451 regnum_t this_group_regnum;
2453 compile_stack.avail--;
2454 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2456 = COMPILE_STACK_TOP.fixup_alt_jump
2457 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2459 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2460 this_group_regnum = COMPILE_STACK_TOP.regnum;
2461 /* If we've reached MAX_REGNUM groups, then this open
2462 won't actually generate any code, so we'll have to
2463 clear pending_exact explicitly. */
2466 /* We're at the end of the group, so now we know how many
2467 groups were inside this one. */
2468 if (this_group_regnum <= MAX_REGNUM)
2470 unsigned char *inner_group_loc
2471 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2473 *inner_group_loc = regnum - this_group_regnum;
2474 BUF_PUSH_3 (stop_memory, this_group_regnum,
2475 regnum - this_group_regnum);
2481 case '|': /* `\|'. */
2482 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2483 goto normal_backslash;
2485 if (syntax & RE_LIMITED_OPS)
2488 /* Insert before the previous alternative a jump which
2489 jumps to this alternative if the former fails. */
2490 GET_BUFFER_SPACE (3);
2491 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2495 /* The alternative before this one has a jump after it
2496 which gets executed if it gets matched. Adjust that
2497 jump so it will jump to this alternative's analogous
2498 jump (put in below, which in turn will jump to the next
2499 (if any) alternative's such jump, etc.). The last such
2500 jump jumps to the correct final destination. A picture:
2506 If we are at `b', then fixup_alt_jump right now points to a
2507 three-byte space after `a'. We'll put in the jump, set
2508 fixup_alt_jump to right after `b', and leave behind three
2509 bytes which we'll fill in when we get to after `c'. */
2512 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2514 /* Mark and leave space for a jump after this alternative,
2515 to be filled in later either by next alternative or
2516 when know we're at the end of a series of alternatives. */
2518 GET_BUFFER_SPACE (3);
2527 /* If \{ is a literal. */
2528 if (!(syntax & RE_INTERVALS)
2529 /* If we're at `\{' and it's not the open-interval
2531 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2532 || (p - 2 == pattern && p == pend))
2533 goto normal_backslash;
2537 /* If got here, then the syntax allows intervals. */
2539 /* At least (most) this many matches must be made. */
2540 int lower_bound = -1, upper_bound = -1;
2542 beg_interval = p - 1;
2546 if (syntax & RE_NO_BK_BRACES)
2547 goto unfetch_interval;
2549 FREE_STACK_RETURN (REG_EBRACE);
2552 GET_UNSIGNED_NUMBER (lower_bound);
2556 GET_UNSIGNED_NUMBER (upper_bound);
2557 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2560 /* Interval such as `{1}' => match exactly once. */
2561 upper_bound = lower_bound;
2563 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2564 || lower_bound > upper_bound)
2566 if (syntax & RE_NO_BK_BRACES)
2567 goto unfetch_interval;
2569 FREE_STACK_RETURN (REG_BADBR);
2572 if (!(syntax & RE_NO_BK_BRACES))
2574 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2581 if (syntax & RE_NO_BK_BRACES)
2582 goto unfetch_interval;
2584 FREE_STACK_RETURN (REG_BADBR);
2587 /* We just parsed a valid interval. */
2589 /* If it's invalid to have no preceding re. */
2592 if (syntax & RE_CONTEXT_INVALID_OPS)
2593 FREE_STACK_RETURN (REG_BADRPT);
2594 else if (syntax & RE_CONTEXT_INDEP_OPS)
2597 goto unfetch_interval;
2600 /* If the upper bound is zero, don't want to succeed at
2601 all; jump from `laststart' to `b + 3', which will be
2602 the end of the buffer after we insert the jump. */
2603 if (upper_bound == 0)
2605 GET_BUFFER_SPACE (3);
2606 INSERT_JUMP (jump, laststart, b + 3);
2610 /* Otherwise, we have a nontrivial interval. When
2611 we're all done, the pattern will look like:
2612 set_number_at <jump count> <upper bound>
2613 set_number_at <succeed_n count> <lower bound>
2614 succeed_n <after jump addr> <succeed_n count>
2616 jump_n <succeed_n addr> <jump count>
2617 (The upper bound and `jump_n' are omitted if
2618 `upper_bound' is 1, though.) */
2620 { /* If the upper bound is > 1, we need to insert
2621 more at the end of the loop. */
2622 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2624 GET_BUFFER_SPACE (nbytes);
2626 /* Initialize lower bound of the `succeed_n', even
2627 though it will be set during matching by its
2628 attendant `set_number_at' (inserted next),
2629 because `re_compile_fastmap' needs to know.
2630 Jump to the `jump_n' we might insert below. */
2631 INSERT_JUMP2 (succeed_n, laststart,
2632 b + 5 + (upper_bound > 1) * 5,
2636 /* Code to initialize the lower bound. Insert
2637 before the `succeed_n'. The `5' is the last two
2638 bytes of this `set_number_at', plus 3 bytes of
2639 the following `succeed_n'. */
2640 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2643 if (upper_bound > 1)
2644 { /* More than one repetition is allowed, so
2645 append a backward jump to the `succeed_n'
2646 that starts this interval.
2648 When we've reached this during matching,
2649 we'll have matched the interval once, so
2650 jump back only `upper_bound - 1' times. */
2651 STORE_JUMP2 (jump_n, b, laststart + 5,
2655 /* The location we want to set is the second
2656 parameter of the `jump_n'; that is `b-2' as
2657 an absolute address. `laststart' will be
2658 the `set_number_at' we're about to insert;
2659 `laststart+3' the number to set, the source
2660 for the relative address. But we are
2661 inserting into the middle of the pattern --
2662 so everything is getting moved up by 5.
2663 Conclusion: (b - 2) - (laststart + 3) + 5,
2664 i.e., b - laststart.
2666 We insert this at the beginning of the loop
2667 so that if we fail during matching, we'll
2668 reinitialize the bounds. */
2669 insert_op2 (set_number_at, laststart, b - laststart,
2670 upper_bound - 1, b);
2675 beg_interval = NULL;
2680 /* If an invalid interval, match the characters as literals. */
2681 assert (beg_interval);
2683 beg_interval = NULL;
2685 /* normal_char and normal_backslash need `c'. */
2688 if (!(syntax & RE_NO_BK_BRACES))
2690 if (p > pattern && p[-1] == '\\')
2691 goto normal_backslash;
2696 /* There is no way to specify the before_dot and after_dot
2697 operators. rms says this is ok. --karl */
2705 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2711 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2717 if (syntax & RE_NO_GNU_OPS)
2720 BUF_PUSH (wordchar);
2725 if (syntax & RE_NO_GNU_OPS)
2728 BUF_PUSH (notwordchar);
2733 if (syntax & RE_NO_GNU_OPS)
2739 if (syntax & RE_NO_GNU_OPS)
2745 if (syntax & RE_NO_GNU_OPS)
2747 BUF_PUSH (wordbound);
2751 if (syntax & RE_NO_GNU_OPS)
2753 BUF_PUSH (notwordbound);
2757 if (syntax & RE_NO_GNU_OPS)
2763 if (syntax & RE_NO_GNU_OPS)
2768 case '1': case '2': case '3': case '4': case '5':
2769 case '6': case '7': case '8': case '9':
2770 if (syntax & RE_NO_BK_REFS)
2776 FREE_STACK_RETURN (REG_ESUBREG);
2778 /* Can't back reference to a subexpression if inside of it. */
2779 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
2783 BUF_PUSH_2 (duplicate, c1);
2789 if (syntax & RE_BK_PLUS_QM)
2792 goto normal_backslash;
2796 /* You might think it would be useful for \ to mean
2797 not to translate; but if we don't translate it
2798 it will never match anything. */
2806 /* Expects the character in `c'. */
2808 /* If no exactn currently being built. */
2811 /* If last exactn not at current position. */
2812 || pending_exact + *pending_exact + 1 != b
2814 /* We have only one byte following the exactn for the count. */
2815 || *pending_exact == (1 << BYTEWIDTH) - 1
2817 /* If followed by a repetition operator. */
2818 || *p == '*' || *p == '^'
2819 || ((syntax & RE_BK_PLUS_QM)
2820 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2821 : (*p == '+' || *p == '?'))
2822 || ((syntax & RE_INTERVALS)
2823 && ((syntax & RE_NO_BK_BRACES)
2825 : (p[0] == '\\' && p[1] == '{'))))
2827 /* Start building a new exactn. */
2831 BUF_PUSH_2 (exactn, 0);
2832 pending_exact = b - 1;
2839 } /* while p != pend */
2842 /* Through the pattern now. */
2845 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2847 if (!COMPILE_STACK_EMPTY)
2848 FREE_STACK_RETURN (REG_EPAREN);
2850 /* If we don't want backtracking, force success
2851 the first time we reach the end of the compiled pattern. */
2852 if (syntax & RE_NO_POSIX_BACKTRACKING)
2855 free (compile_stack.stack);
2857 /* We have succeeded; set the length of the buffer. */
2858 bufp->used = b - bufp->buffer;
2863 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2864 print_compiled_pattern (bufp);
2868 #ifndef MATCH_MAY_ALLOCATE
2869 /* Initialize the failure stack to the largest possible stack. This
2870 isn't necessary unless we're trying to avoid calling alloca in
2871 the search and match routines. */
2873 int num_regs = bufp->re_nsub + 1;
2875 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2876 is strictly greater than re_max_failures, the largest possible stack
2877 is 2 * re_max_failures failure points. */
2878 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2880 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2883 if (! fail_stack.stack)
2885 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2886 * sizeof (fail_stack_elt_t));
2889 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2891 * sizeof (fail_stack_elt_t)));
2892 # else /* not emacs */
2893 if (! fail_stack.stack)
2895 = (fail_stack_elt_t *) malloc (fail_stack.size
2896 * sizeof (fail_stack_elt_t));
2899 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2901 * sizeof (fail_stack_elt_t)));
2902 # endif /* not emacs */
2905 regex_grow_registers (num_regs);
2907 #endif /* not MATCH_MAY_ALLOCATE */
2910 } /* regex_compile */
2912 /* Subroutines for `regex_compile'. */
2914 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2917 store_op1 (op, loc, arg)
2922 *loc = (unsigned char) op;
2923 STORE_NUMBER (loc + 1, arg);
2927 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2930 store_op2 (op, loc, arg1, arg2)
2935 *loc = (unsigned char) op;
2936 STORE_NUMBER (loc + 1, arg1);
2937 STORE_NUMBER (loc + 3, arg2);
2941 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
2942 for OP followed by two-byte integer parameter ARG. */
2945 insert_op1 (op, loc, arg, end)
2951 register unsigned char *pfrom = end;
2952 register unsigned char *pto = end + 3;
2954 while (pfrom != loc)
2957 store_op1 (op, loc, arg);
2961 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2964 insert_op2 (op, loc, arg1, arg2, end)
2970 register unsigned char *pfrom = end;
2971 register unsigned char *pto = end + 5;
2973 while (pfrom != loc)
2976 store_op2 (op, loc, arg1, arg2);
2980 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
2981 after an alternative or a begin-subexpression. We assume there is at
2982 least one character before the ^. */
2985 at_begline_loc_p (pattern, p, syntax)
2986 const char *pattern, *p;
2987 reg_syntax_t syntax;
2989 const char *prev = p - 2;
2990 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
2993 /* After a subexpression? */
2994 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
2995 /* After an alternative? */
2996 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
3000 /* The dual of at_begline_loc_p. This one is for $. We assume there is
3001 at least one character after the $, i.e., `P < PEND'. */
3004 at_endline_loc_p (p, pend, syntax)
3005 const char *p, *pend;
3006 reg_syntax_t syntax;
3008 const char *next = p;
3009 boolean next_backslash = *next == '\\';
3010 const char *next_next = p + 1 < pend ? p + 1 : 0;
3013 /* Before a subexpression? */
3014 (syntax & RE_NO_BK_PARENS ? *next == ')'
3015 : next_backslash && next_next && *next_next == ')')
3016 /* Before an alternative? */
3017 || (syntax & RE_NO_BK_VBAR ? *next == '|'
3018 : next_backslash && next_next && *next_next == '|');
3022 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
3023 false if it's not. */
3026 group_in_compile_stack (compile_stack, regnum)
3027 compile_stack_type compile_stack;
3032 for (this_element = compile_stack.avail - 1;
3035 if (compile_stack.stack[this_element].regnum == regnum)
3042 /* Read the ending character of a range (in a bracket expression) from the
3043 uncompiled pattern *P_PTR (which ends at PEND). We assume the
3044 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
3045 Then we set the translation of all bits between the starting and
3046 ending characters (inclusive) in the compiled pattern B.
3048 Return an error code.
3050 We use these short variable names so we can use the same macros as
3051 `regex_compile' itself. */
3053 static reg_errcode_t
3054 compile_range (p_ptr, pend, translate, syntax, b)
3055 const char **p_ptr, *pend;
3056 RE_TRANSLATE_TYPE translate;
3057 reg_syntax_t syntax;
3062 const char *p = *p_ptr;
3063 unsigned int range_start, range_end;
3068 /* Even though the pattern is a signed `char *', we need to fetch
3069 with unsigned char *'s; if the high bit of the pattern character
3070 is set, the range endpoints will be negative if we fetch using a
3073 We also want to fetch the endpoints without translating them; the
3074 appropriate translation is done in the bit-setting loop below. */
3075 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
3076 range_start = ((const unsigned char *) p)[-2];
3077 range_end = ((const unsigned char *) p)[0];
3079 /* Have to increment the pointer into the pattern string, so the
3080 caller isn't still at the ending character. */
3083 /* If the start is after the end, the range is empty. */
3084 if (range_start > range_end)
3085 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
3087 /* Here we see why `this_char' has to be larger than an `unsigned
3088 char' -- the range is inclusive, so if `range_end' == 0xff
3089 (assuming 8-bit characters), we would otherwise go into an infinite
3090 loop, since all characters <= 0xff. */
3091 for (this_char = range_start; this_char <= range_end; this_char++)
3093 SET_LIST_BIT (TRANSLATE (this_char));
3099 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3100 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3101 characters can start a string that matches the pattern. This fastmap
3102 is used by re_search to skip quickly over impossible starting points.
3104 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3105 area as BUFP->fastmap.
3107 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3110 Returns 0 if we succeed, -2 if an internal error. */
3113 re_compile_fastmap (bufp)
3114 struct re_pattern_buffer *bufp;
3117 #ifdef MATCH_MAY_ALLOCATE
3118 fail_stack_type fail_stack;
3120 #ifndef REGEX_MALLOC
3124 register char *fastmap = bufp->fastmap;
3125 unsigned char *pattern = bufp->buffer;
3126 unsigned char *p = pattern;
3127 register unsigned char *pend = pattern + bufp->used;
3130 /* This holds the pointer to the failure stack, when
3131 it is allocated relocatably. */
3132 fail_stack_elt_t *failure_stack_ptr;
3135 /* Assume that each path through the pattern can be null until
3136 proven otherwise. We set this false at the bottom of switch
3137 statement, to which we get only if a particular path doesn't
3138 match the empty string. */
3139 boolean path_can_be_null = true;
3141 /* We aren't doing a `succeed_n' to begin with. */
3142 boolean succeed_n_p = false;
3144 assert (fastmap != NULL && p != NULL);
3147 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
3148 bufp->fastmap_accurate = 1; /* It will be when we're done. */
3149 bufp->can_be_null = 0;
3153 if (p == pend || *p == succeed)
3155 /* We have reached the (effective) end of pattern. */
3156 if (!FAIL_STACK_EMPTY ())
3158 bufp->can_be_null |= path_can_be_null;
3160 /* Reset for next path. */
3161 path_can_be_null = true;
3163 p = fail_stack.stack[--fail_stack.avail].pointer;
3171 /* We should never be about to go beyond the end of the pattern. */
3174 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3177 /* I guess the idea here is to simply not bother with a fastmap
3178 if a backreference is used, since it's too hard to figure out
3179 the fastmap for the corresponding group. Setting
3180 `can_be_null' stops `re_search_2' from using the fastmap, so
3181 that is all we do. */
3183 bufp->can_be_null = 1;
3187 /* Following are the cases which match a character. These end
3196 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3197 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3203 /* Chars beyond end of map must be allowed. */
3204 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3207 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3208 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3214 for (j = 0; j < (1 << BYTEWIDTH); j++)
3215 if (SYNTAX (j) == Sword)
3221 for (j = 0; j < (1 << BYTEWIDTH); j++)
3222 if (SYNTAX (j) != Sword)
3229 int fastmap_newline = fastmap['\n'];
3231 /* `.' matches anything ... */
3232 for (j = 0; j < (1 << BYTEWIDTH); j++)
3235 /* ... except perhaps newline. */
3236 if (!(bufp->syntax & RE_DOT_NEWLINE))
3237 fastmap['\n'] = fastmap_newline;
3239 /* Return if we have already set `can_be_null'; if we have,
3240 then the fastmap is irrelevant. Something's wrong here. */
3241 else if (bufp->can_be_null)
3244 /* Otherwise, have to check alternative paths. */
3251 for (j = 0; j < (1 << BYTEWIDTH); j++)
3252 if (SYNTAX (j) == (enum syntaxcode) k)
3259 for (j = 0; j < (1 << BYTEWIDTH); j++)
3260 if (SYNTAX (j) != (enum syntaxcode) k)
3265 /* All cases after this match the empty string. These end with
3285 case push_dummy_failure:
3290 case pop_failure_jump:
3291 case maybe_pop_jump:
3294 case dummy_failure_jump:
3295 EXTRACT_NUMBER_AND_INCR (j, p);
3300 /* Jump backward implies we just went through the body of a
3301 loop and matched nothing. Opcode jumped to should be
3302 `on_failure_jump' or `succeed_n'. Just treat it like an
3303 ordinary jump. For a * loop, it has pushed its failure
3304 point already; if so, discard that as redundant. */
3305 if ((re_opcode_t) *p != on_failure_jump
3306 && (re_opcode_t) *p != succeed_n)
3310 EXTRACT_NUMBER_AND_INCR (j, p);
3313 /* If what's on the stack is where we are now, pop it. */
3314 if (!FAIL_STACK_EMPTY ()
3315 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3321 case on_failure_jump:
3322 case on_failure_keep_string_jump:
3323 handle_on_failure_jump:
3324 EXTRACT_NUMBER_AND_INCR (j, p);
3326 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3327 end of the pattern. We don't want to push such a point,
3328 since when we restore it above, entering the switch will
3329 increment `p' past the end of the pattern. We don't need
3330 to push such a point since we obviously won't find any more
3331 fastmap entries beyond `pend'. Such a pattern can match
3332 the null string, though. */
3335 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3337 RESET_FAIL_STACK ();
3342 bufp->can_be_null = 1;
3346 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3347 succeed_n_p = false;
3354 /* Get to the number of times to succeed. */
3357 /* Increment p past the n for when k != 0. */
3358 EXTRACT_NUMBER_AND_INCR (k, p);
3362 succeed_n_p = true; /* Spaghetti code alert. */
3363 goto handle_on_failure_jump;
3380 abort (); /* We have listed all the cases. */
3383 /* Getting here means we have found the possible starting
3384 characters for one path of the pattern -- and that the empty
3385 string does not match. We need not follow this path further.
3386 Instead, look at the next alternative (remembered on the
3387 stack), or quit if no more. The test at the top of the loop
3388 does these things. */
3389 path_can_be_null = false;
3393 /* Set `can_be_null' for the last path (also the first path, if the
3394 pattern is empty). */
3395 bufp->can_be_null |= path_can_be_null;
3398 RESET_FAIL_STACK ();
3400 } /* re_compile_fastmap */
3402 weak_alias (__re_compile_fastmap, re_compile_fastmap)
3405 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3406 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3407 this memory for recording register information. STARTS and ENDS
3408 must be allocated using the malloc library routine, and must each
3409 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3411 If NUM_REGS == 0, then subsequent matches should allocate their own
3414 Unless this function is called, the first search or match using
3415 PATTERN_BUFFER will allocate its own register data, without
3416 freeing the old data. */
3419 re_set_registers (bufp, regs, num_regs, starts, ends)
3420 struct re_pattern_buffer *bufp;
3421 struct re_registers *regs;
3423 regoff_t *starts, *ends;
3427 bufp->regs_allocated = REGS_REALLOCATE;
3428 regs->num_regs = num_regs;
3429 regs->start = starts;
3434 bufp->regs_allocated = REGS_UNALLOCATED;
3436 regs->start = regs->end = (regoff_t *) 0;
3440 weak_alias (__re_set_registers, re_set_registers)
3443 /* Searching routines. */
3445 /* Like re_search_2, below, but only one string is specified, and
3446 doesn't let you say where to stop matching. */
3449 re_search (bufp, string, size, startpos, range, regs)
3450 struct re_pattern_buffer *bufp;
3452 int size, startpos, range;
3453 struct re_registers *regs;
3455 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3459 weak_alias (__re_search, re_search)
3463 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3464 virtual concatenation of STRING1 and STRING2, starting first at index
3465 STARTPOS, then at STARTPOS + 1, and so on.
3467 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3469 RANGE is how far to scan while trying to match. RANGE = 0 means try
3470 only at STARTPOS; in general, the last start tried is STARTPOS +
3473 In REGS, return the indices of the virtual concatenation of STRING1
3474 and STRING2 that matched the entire BUFP->buffer and its contained
3477 Do not consider matching one past the index STOP in the virtual
3478 concatenation of STRING1 and STRING2.
3480 We return either the position in the strings at which the match was
3481 found, -1 if no match, or -2 if error (such as failure
3485 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3486 struct re_pattern_buffer *bufp;
3487 const char *string1, *string2;
3491 struct re_registers *regs;
3495 register char *fastmap = bufp->fastmap;
3496 register RE_TRANSLATE_TYPE translate = bufp->translate;
3497 int total_size = size1 + size2;
3498 int endpos = startpos + range;
3500 /* Check for out-of-range STARTPOS. */
3501 if (startpos < 0 || startpos > total_size)
3504 /* Fix up RANGE if it might eventually take us outside
3505 the virtual concatenation of STRING1 and STRING2.
3506 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3508 range = 0 - startpos;
3509 else if (endpos > total_size)
3510 range = total_size - startpos;
3512 /* If the search isn't to be a backwards one, don't waste time in a
3513 search for a pattern that must be anchored. */
3514 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3523 /* In a forward search for something that starts with \=.
3524 don't keep searching past point. */
3525 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3527 range = PT - startpos;
3533 /* Update the fastmap now if not correct already. */
3534 if (fastmap && !bufp->fastmap_accurate)
3535 if (re_compile_fastmap (bufp) == -2)
3538 /* Loop through the string, looking for a place to start matching. */
3541 /* If a fastmap is supplied, skip quickly over characters that
3542 cannot be the start of a match. If the pattern can match the
3543 null string, however, we don't need to skip characters; we want
3544 the first null string. */
3545 if (fastmap && startpos < total_size && !bufp->can_be_null)
3547 if (range > 0) /* Searching forwards. */
3549 register const char *d;
3550 register int lim = 0;
3553 if (startpos < size1 && startpos + range >= size1)
3554 lim = range - (size1 - startpos);
3556 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3558 /* Written out as an if-else to avoid testing `translate'
3562 && !fastmap[(unsigned char)
3563 translate[(unsigned char) *d++]])
3566 while (range > lim && !fastmap[(unsigned char) *d++])
3569 startpos += irange - range;
3571 else /* Searching backwards. */
3573 register char c = (size1 == 0 || startpos >= size1
3574 ? string2[startpos - size1]
3575 : string1[startpos]);
3577 if (!fastmap[(unsigned char) TRANSLATE (c)])
3582 /* If can't match the null string, and that's all we have left, fail. */
3583 if (range >= 0 && startpos == total_size && fastmap
3584 && !bufp->can_be_null)
3587 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3588 startpos, regs, stop);
3589 #ifndef REGEX_MALLOC
3618 weak_alias (__re_search_2, re_search_2)
3621 /* This converts PTR, a pointer into one of the search strings `string1'
3622 and `string2' into an offset from the beginning of that string. */
3623 #define POINTER_TO_OFFSET(ptr) \
3624 (FIRST_STRING_P (ptr) \
3625 ? ((regoff_t) ((ptr) - string1)) \
3626 : ((regoff_t) ((ptr) - string2 + size1)))
3628 /* Macros for dealing with the split strings in re_match_2. */
3630 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3632 /* Call before fetching a character with *d. This switches over to
3633 string2 if necessary. */
3634 #define PREFETCH() \
3637 /* End of string2 => fail. */ \
3638 if (dend == end_match_2) \
3640 /* End of string1 => advance to string2. */ \
3642 dend = end_match_2; \
3646 /* Test if at very beginning or at very end of the virtual concatenation
3647 of `string1' and `string2'. If only one string, it's `string2'. */
3648 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3649 #define AT_STRINGS_END(d) ((d) == end2)
3652 /* Test if D points to a character which is word-constituent. We have
3653 two special cases to check for: if past the end of string1, look at
3654 the first character in string2; and if before the beginning of
3655 string2, look at the last character in string1. */
3656 #define WORDCHAR_P(d) \
3657 (SYNTAX ((d) == end1 ? *string2 \
3658 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3661 /* Disabled due to a compiler bug -- see comment at case wordbound */
3663 /* Test if the character before D and the one at D differ with respect
3664 to being word-constituent. */
3665 #define AT_WORD_BOUNDARY(d) \
3666 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3667 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3670 /* Free everything we malloc. */
3671 #ifdef MATCH_MAY_ALLOCATE
3672 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3673 # define FREE_VARIABLES() \
3675 REGEX_FREE_STACK (fail_stack.stack); \
3676 FREE_VAR (regstart); \
3677 FREE_VAR (regend); \
3678 FREE_VAR (old_regstart); \
3679 FREE_VAR (old_regend); \
3680 FREE_VAR (best_regstart); \
3681 FREE_VAR (best_regend); \
3682 FREE_VAR (reg_info); \
3683 FREE_VAR (reg_dummy); \
3684 FREE_VAR (reg_info_dummy); \
3687 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3688 #endif /* not MATCH_MAY_ALLOCATE */
3690 /* These values must meet several constraints. They must not be valid
3691 register values; since we have a limit of 255 registers (because
3692 we use only one byte in the pattern for the register number), we can
3693 use numbers larger than 255. They must differ by 1, because of
3694 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3695 be larger than the value for the highest register, so we do not try
3696 to actually save any registers when none are active. */
3697 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3698 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3700 /* Matching routines. */
3702 #ifndef emacs /* Emacs never uses this. */
3703 /* re_match is like re_match_2 except it takes only a single string. */
3706 re_match (bufp, string, size, pos, regs)
3707 struct re_pattern_buffer *bufp;
3710 struct re_registers *regs;
3712 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3714 # ifndef REGEX_MALLOC
3722 weak_alias (__re_match, re_match)
3724 #endif /* not emacs */
3726 static boolean group_match_null_string_p _RE_ARGS ((unsigned char **p,
3728 register_info_type *reg_info));
3729 static boolean alt_match_null_string_p _RE_ARGS ((unsigned char *p,
3731 register_info_type *reg_info));
3732 static boolean common_op_match_null_string_p _RE_ARGS ((unsigned char **p,
3734 register_info_type *reg_info));
3735 static int bcmp_translate _RE_ARGS ((const char *s1, const char *s2,
3736 int len, char *translate));
3738 /* re_match_2 matches the compiled pattern in BUFP against the
3739 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3740 and SIZE2, respectively). We start matching at POS, and stop
3743 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3744 store offsets for the substring each group matched in REGS. See the
3745 documentation for exactly how many groups we fill.
3747 We return -1 if no match, -2 if an internal error (such as the
3748 failure stack overflowing). Otherwise, we return the length of the
3749 matched substring. */
3752 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3753 struct re_pattern_buffer *bufp;
3754 const char *string1, *string2;
3757 struct re_registers *regs;
3760 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3762 #ifndef REGEX_MALLOC
3770 weak_alias (__re_match_2, re_match_2)
3773 /* This is a separate function so that we can force an alloca cleanup
3776 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3777 struct re_pattern_buffer *bufp;
3778 const char *string1, *string2;
3781 struct re_registers *regs;
3784 /* General temporaries. */
3788 /* Just past the end of the corresponding string. */
3789 const char *end1, *end2;
3791 /* Pointers into string1 and string2, just past the last characters in
3792 each to consider matching. */
3793 const char *end_match_1, *end_match_2;
3795 /* Where we are in the data, and the end of the current string. */
3796 const char *d, *dend;
3798 /* Where we are in the pattern, and the end of the pattern. */
3799 unsigned char *p = bufp->buffer;
3800 register unsigned char *pend = p + bufp->used;
3802 /* Mark the opcode just after a start_memory, so we can test for an
3803 empty subpattern when we get to the stop_memory. */
3804 unsigned char *just_past_start_mem = 0;
3806 /* We use this to map every character in the string. */
3807 RE_TRANSLATE_TYPE translate = bufp->translate;
3809 /* Failure point stack. Each place that can handle a failure further
3810 down the line pushes a failure point on this stack. It consists of
3811 restart, regend, and reg_info for all registers corresponding to
3812 the subexpressions we're currently inside, plus the number of such
3813 registers, and, finally, two char *'s. The first char * is where
3814 to resume scanning the pattern; the second one is where to resume
3815 scanning the strings. If the latter is zero, the failure point is
3816 a ``dummy''; if a failure happens and the failure point is a dummy,
3817 it gets discarded and the next next one is tried. */
3818 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3819 fail_stack_type fail_stack;
3822 static unsigned failure_id = 0;
3823 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3827 /* This holds the pointer to the failure stack, when
3828 it is allocated relocatably. */
3829 fail_stack_elt_t *failure_stack_ptr;
3832 /* We fill all the registers internally, independent of what we
3833 return, for use in backreferences. The number here includes
3834 an element for register zero. */
3835 size_t num_regs = bufp->re_nsub + 1;
3837 /* The currently active registers. */
3838 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3839 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3841 /* Information on the contents of registers. These are pointers into
3842 the input strings; they record just what was matched (on this
3843 attempt) by a subexpression part of the pattern, that is, the
3844 regnum-th regstart pointer points to where in the pattern we began
3845 matching and the regnum-th regend points to right after where we
3846 stopped matching the regnum-th subexpression. (The zeroth register
3847 keeps track of what the whole pattern matches.) */
3848 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3849 const char **regstart, **regend;
3852 /* If a group that's operated upon by a repetition operator fails to
3853 match anything, then the register for its start will need to be
3854 restored because it will have been set to wherever in the string we
3855 are when we last see its open-group operator. Similarly for a
3857 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3858 const char **old_regstart, **old_regend;
3861 /* The is_active field of reg_info helps us keep track of which (possibly
3862 nested) subexpressions we are currently in. The matched_something
3863 field of reg_info[reg_num] helps us tell whether or not we have
3864 matched any of the pattern so far this time through the reg_num-th
3865 subexpression. These two fields get reset each time through any
3866 loop their register is in. */
3867 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3868 register_info_type *reg_info;
3871 /* The following record the register info as found in the above
3872 variables when we find a match better than any we've seen before.
3873 This happens as we backtrack through the failure points, which in
3874 turn happens only if we have not yet matched the entire string. */
3875 unsigned best_regs_set = false;
3876 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3877 const char **best_regstart, **best_regend;
3880 /* Logically, this is `best_regend[0]'. But we don't want to have to
3881 allocate space for that if we're not allocating space for anything
3882 else (see below). Also, we never need info about register 0 for
3883 any of the other register vectors, and it seems rather a kludge to
3884 treat `best_regend' differently than the rest. So we keep track of
3885 the end of the best match so far in a separate variable. We
3886 initialize this to NULL so that when we backtrack the first time
3887 and need to test it, it's not garbage. */
3888 const char *match_end = NULL;
3890 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3891 int set_regs_matched_done = 0;
3893 /* Used when we pop values we don't care about. */
3894 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3895 const char **reg_dummy;
3896 register_info_type *reg_info_dummy;
3900 /* Counts the total number of registers pushed. */
3901 unsigned num_regs_pushed = 0;
3904 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3908 #ifdef MATCH_MAY_ALLOCATE
3909 /* Do not bother to initialize all the register variables if there are
3910 no groups in the pattern, as it takes a fair amount of time. If
3911 there are groups, we include space for register 0 (the whole
3912 pattern), even though we never use it, since it simplifies the
3913 array indexing. We should fix this. */
3916 regstart = REGEX_TALLOC (num_regs, const char *);
3917 regend = REGEX_TALLOC (num_regs, const char *);
3918 old_regstart = REGEX_TALLOC (num_regs, const char *);
3919 old_regend = REGEX_TALLOC (num_regs, const char *);
3920 best_regstart = REGEX_TALLOC (num_regs, const char *);
3921 best_regend = REGEX_TALLOC (num_regs, const char *);
3922 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3923 reg_dummy = REGEX_TALLOC (num_regs, const char *);
3924 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3926 if (!(regstart && regend && old_regstart && old_regend && reg_info
3927 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3935 /* We must initialize all our variables to NULL, so that
3936 `FREE_VARIABLES' doesn't try to free them. */
3937 regstart = regend = old_regstart = old_regend = best_regstart
3938 = best_regend = reg_dummy = NULL;
3939 reg_info = reg_info_dummy = (register_info_type *) NULL;
3941 #endif /* MATCH_MAY_ALLOCATE */
3943 /* The starting position is bogus. */
3944 if (pos < 0 || pos > size1 + size2)
3950 /* Initialize subexpression text positions to -1 to mark ones that no
3951 start_memory/stop_memory has been seen for. Also initialize the
3952 register information struct. */
3953 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
3955 regstart[mcnt] = regend[mcnt]
3956 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3958 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3959 IS_ACTIVE (reg_info[mcnt]) = 0;
3960 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3961 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3964 /* We move `string1' into `string2' if the latter's empty -- but not if
3965 `string1' is null. */
3966 if (size2 == 0 && string1 != NULL)
3973 end1 = string1 + size1;
3974 end2 = string2 + size2;
3976 /* Compute where to stop matching, within the two strings. */
3979 end_match_1 = string1 + stop;
3980 end_match_2 = string2;
3985 end_match_2 = string2 + stop - size1;
3988 /* `p' scans through the pattern as `d' scans through the data.
3989 `dend' is the end of the input string that `d' points within. `d'
3990 is advanced into the following input string whenever necessary, but
3991 this happens before fetching; therefore, at the beginning of the
3992 loop, `d' can be pointing at the end of a string, but it cannot
3994 if (size1 > 0 && pos <= size1)
4001 d = string2 + pos - size1;
4005 DEBUG_PRINT1 ("The compiled pattern is:\n");
4006 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
4007 DEBUG_PRINT1 ("The string to match is: `");
4008 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
4009 DEBUG_PRINT1 ("'\n");
4011 /* This loops over pattern commands. It exits by returning from the
4012 function if the match is complete, or it drops through if the match
4013 fails at this starting point in the input data. */
4017 DEBUG_PRINT2 ("\n%p: ", p);
4019 DEBUG_PRINT2 ("\n0x%x: ", p);
4023 { /* End of pattern means we might have succeeded. */
4024 DEBUG_PRINT1 ("end of pattern ... ");
4026 /* If we haven't matched the entire string, and we want the
4027 longest match, try backtracking. */
4028 if (d != end_match_2)
4030 /* 1 if this match ends in the same string (string1 or string2)
4031 as the best previous match. */
4032 boolean same_str_p = (FIRST_STRING_P (match_end)
4033 == MATCHING_IN_FIRST_STRING);
4034 /* 1 if this match is the best seen so far. */
4035 boolean best_match_p;
4037 /* AIX compiler got confused when this was combined
4038 with the previous declaration. */
4040 best_match_p = d > match_end;
4042 best_match_p = !MATCHING_IN_FIRST_STRING;
4044 DEBUG_PRINT1 ("backtracking.\n");
4046 if (!FAIL_STACK_EMPTY ())
4047 { /* More failure points to try. */
4049 /* If exceeds best match so far, save it. */
4050 if (!best_regs_set || best_match_p)
4052 best_regs_set = true;
4055 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
4057 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4059 best_regstart[mcnt] = regstart[mcnt];
4060 best_regend[mcnt] = regend[mcnt];
4066 /* If no failure points, don't restore garbage. And if
4067 last match is real best match, don't restore second
4069 else if (best_regs_set && !best_match_p)
4072 /* Restore best match. It may happen that `dend ==
4073 end_match_1' while the restored d is in string2.
4074 For example, the pattern `x.*y.*z' against the
4075 strings `x-' and `y-z-', if the two strings are
4076 not consecutive in memory. */
4077 DEBUG_PRINT1 ("Restoring best registers.\n");
4080 dend = ((d >= string1 && d <= end1)
4081 ? end_match_1 : end_match_2);
4083 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4085 regstart[mcnt] = best_regstart[mcnt];
4086 regend[mcnt] = best_regend[mcnt];
4089 } /* d != end_match_2 */
4092 DEBUG_PRINT1 ("Accepting match.\n");
4094 /* If caller wants register contents data back, do it. */
4095 if (regs && !bufp->no_sub)
4097 /* Have the register data arrays been allocated? */
4098 if (bufp->regs_allocated == REGS_UNALLOCATED)
4099 { /* No. So allocate them with malloc. We need one
4100 extra element beyond `num_regs' for the `-1' marker
4102 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
4103 regs->start = TALLOC (regs->num_regs, regoff_t);
4104 regs->end = TALLOC (regs->num_regs, regoff_t);
4105 if (regs->start == NULL || regs->end == NULL)
4110 bufp->regs_allocated = REGS_REALLOCATE;
4112 else if (bufp->regs_allocated == REGS_REALLOCATE)
4113 { /* Yes. If we need more elements than were already
4114 allocated, reallocate them. If we need fewer, just
4116 if (regs->num_regs < num_regs + 1)
4118 regs->num_regs = num_regs + 1;
4119 RETALLOC (regs->start, regs->num_regs, regoff_t);
4120 RETALLOC (regs->end, regs->num_regs, regoff_t);
4121 if (regs->start == NULL || regs->end == NULL)
4130 /* These braces fend off a "empty body in an else-statement"
4131 warning under GCC when assert expands to nothing. */
4132 assert (bufp->regs_allocated == REGS_FIXED);
4135 /* Convert the pointer data in `regstart' and `regend' to
4136 indices. Register zero has to be set differently,
4137 since we haven't kept track of any info for it. */
4138 if (regs->num_regs > 0)
4140 regs->start[0] = pos;
4141 regs->end[0] = (MATCHING_IN_FIRST_STRING
4142 ? ((regoff_t) (d - string1))
4143 : ((regoff_t) (d - string2 + size1)));
4146 /* Go through the first `min (num_regs, regs->num_regs)'
4147 registers, since that is all we initialized. */
4148 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
4151 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
4152 regs->start[mcnt] = regs->end[mcnt] = -1;
4156 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
4158 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
4162 /* If the regs structure we return has more elements than
4163 were in the pattern, set the extra elements to -1. If
4164 we (re)allocated the registers, this is the case,
4165 because we always allocate enough to have at least one
4167 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
4168 regs->start[mcnt] = regs->end[mcnt] = -1;
4169 } /* regs && !bufp->no_sub */
4171 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
4172 nfailure_points_pushed, nfailure_points_popped,
4173 nfailure_points_pushed - nfailure_points_popped);
4174 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
4176 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
4180 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
4186 /* Otherwise match next pattern command. */
4187 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4189 /* Ignore these. Used to ignore the n of succeed_n's which
4190 currently have n == 0. */
4192 DEBUG_PRINT1 ("EXECUTING no_op.\n");
4196 DEBUG_PRINT1 ("EXECUTING succeed.\n");
4199 /* Match the next n pattern characters exactly. The following
4200 byte in the pattern defines n, and the n bytes after that
4201 are the characters to match. */
4204 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
4206 /* This is written out as an if-else so we don't waste time
4207 testing `translate' inside the loop. */
4213 if ((unsigned char) translate[(unsigned char) *d++]
4214 != (unsigned char) *p++)
4224 if (*d++ != (char) *p++) goto fail;
4228 SET_REGS_MATCHED ();
4232 /* Match any character except possibly a newline or a null. */
4234 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4238 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
4239 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4242 SET_REGS_MATCHED ();
4243 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4251 register unsigned char c;
4252 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4254 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4257 c = TRANSLATE (*d); /* The character to match. */
4259 /* Cast to `unsigned' instead of `unsigned char' in case the
4260 bit list is a full 32 bytes long. */
4261 if (c < (unsigned) (*p * BYTEWIDTH)
4262 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4267 if (!not) goto fail;
4269 SET_REGS_MATCHED ();
4275 /* The beginning of a group is represented by start_memory.
4276 The arguments are the register number in the next byte, and the
4277 number of groups inner to this one in the next. The text
4278 matched within the group is recorded (in the internal
4279 registers data structure) under the register number. */
4281 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4283 /* Find out if this group can match the empty string. */
4284 p1 = p; /* To send to group_match_null_string_p. */
4286 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4287 REG_MATCH_NULL_STRING_P (reg_info[*p])
4288 = group_match_null_string_p (&p1, pend, reg_info);
4290 /* Save the position in the string where we were the last time
4291 we were at this open-group operator in case the group is
4292 operated upon by a repetition operator, e.g., with `(a*)*b'
4293 against `ab'; then we want to ignore where we are now in
4294 the string in case this attempt to match fails. */
4295 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4296 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4298 DEBUG_PRINT2 (" old_regstart: %d\n",
4299 POINTER_TO_OFFSET (old_regstart[*p]));
4302 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4304 IS_ACTIVE (reg_info[*p]) = 1;
4305 MATCHED_SOMETHING (reg_info[*p]) = 0;
4307 /* Clear this whenever we change the register activity status. */
4308 set_regs_matched_done = 0;
4310 /* This is the new highest active register. */
4311 highest_active_reg = *p;
4313 /* If nothing was active before, this is the new lowest active
4315 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4316 lowest_active_reg = *p;
4318 /* Move past the register number and inner group count. */
4320 just_past_start_mem = p;
4325 /* The stop_memory opcode represents the end of a group. Its
4326 arguments are the same as start_memory's: the register
4327 number, and the number of inner groups. */
4329 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4331 /* We need to save the string position the last time we were at
4332 this close-group operator in case the group is operated
4333 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4334 against `aba'; then we want to ignore where we are now in
4335 the string in case this attempt to match fails. */
4336 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4337 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4339 DEBUG_PRINT2 (" old_regend: %d\n",
4340 POINTER_TO_OFFSET (old_regend[*p]));
4343 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4345 /* This register isn't active anymore. */
4346 IS_ACTIVE (reg_info[*p]) = 0;
4348 /* Clear this whenever we change the register activity status. */
4349 set_regs_matched_done = 0;
4351 /* If this was the only register active, nothing is active
4353 if (lowest_active_reg == highest_active_reg)
4355 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4356 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4359 { /* We must scan for the new highest active register, since
4360 it isn't necessarily one less than now: consider
4361 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4362 new highest active register is 1. */
4363 unsigned char r = *p - 1;
4364 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4367 /* If we end up at register zero, that means that we saved
4368 the registers as the result of an `on_failure_jump', not
4369 a `start_memory', and we jumped to past the innermost
4370 `stop_memory'. For example, in ((.)*) we save
4371 registers 1 and 2 as a result of the *, but when we pop
4372 back to the second ), we are at the stop_memory 1.
4373 Thus, nothing is active. */
4376 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4377 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4380 highest_active_reg = r;
4383 /* If just failed to match something this time around with a
4384 group that's operated on by a repetition operator, try to
4385 force exit from the ``loop'', and restore the register
4386 information for this group that we had before trying this
4388 if ((!MATCHED_SOMETHING (reg_info[*p])
4389 || just_past_start_mem == p - 1)
4392 boolean is_a_jump_n = false;
4396 switch ((re_opcode_t) *p1++)
4400 case pop_failure_jump:
4401 case maybe_pop_jump:
4403 case dummy_failure_jump:
4404 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4414 /* If the next operation is a jump backwards in the pattern
4415 to an on_failure_jump right before the start_memory
4416 corresponding to this stop_memory, exit from the loop
4417 by forcing a failure after pushing on the stack the
4418 on_failure_jump's jump in the pattern, and d. */
4419 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4420 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4422 /* If this group ever matched anything, then restore
4423 what its registers were before trying this last
4424 failed match, e.g., with `(a*)*b' against `ab' for
4425 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4426 against `aba' for regend[3].
4428 Also restore the registers for inner groups for,
4429 e.g., `((a*)(b*))*' against `aba' (register 3 would
4430 otherwise get trashed). */
4432 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4436 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4438 /* Restore this and inner groups' (if any) registers. */
4439 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
4442 regstart[r] = old_regstart[r];
4444 /* xx why this test? */
4445 if (old_regend[r] >= regstart[r])
4446 regend[r] = old_regend[r];
4450 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4451 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4457 /* Move past the register number and the inner group count. */
4462 /* \<digit> has been turned into a `duplicate' command which is
4463 followed by the numeric value of <digit> as the register number. */
4466 register const char *d2, *dend2;
4467 int regno = *p++; /* Get which register to match against. */
4468 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4470 /* Can't back reference a group which we've never matched. */
4471 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4474 /* Where in input to try to start matching. */
4475 d2 = regstart[regno];
4477 /* Where to stop matching; if both the place to start and
4478 the place to stop matching are in the same string, then
4479 set to the place to stop, otherwise, for now have to use
4480 the end of the first string. */
4482 dend2 = ((FIRST_STRING_P (regstart[regno])
4483 == FIRST_STRING_P (regend[regno]))
4484 ? regend[regno] : end_match_1);
4487 /* If necessary, advance to next segment in register
4491 if (dend2 == end_match_2) break;
4492 if (dend2 == regend[regno]) break;
4494 /* End of string1 => advance to string2. */
4496 dend2 = regend[regno];
4498 /* At end of register contents => success */
4499 if (d2 == dend2) break;
4501 /* If necessary, advance to next segment in data. */
4504 /* How many characters left in this segment to match. */
4507 /* Want how many consecutive characters we can match in
4508 one shot, so, if necessary, adjust the count. */
4509 if (mcnt > dend2 - d2)
4512 /* Compare that many; failure if mismatch, else move
4515 ? bcmp_translate (d, d2, mcnt, translate)
4516 : memcmp (d, d2, mcnt))
4518 d += mcnt, d2 += mcnt;
4520 /* Do this because we've match some characters. */
4521 SET_REGS_MATCHED ();
4527 /* begline matches the empty string at the beginning of the string
4528 (unless `not_bol' is set in `bufp'), and, if
4529 `newline_anchor' is set, after newlines. */
4531 DEBUG_PRINT1 ("EXECUTING begline.\n");
4533 if (AT_STRINGS_BEG (d))
4535 if (!bufp->not_bol) break;
4537 else if (d[-1] == '\n' && bufp->newline_anchor)
4541 /* In all other cases, we fail. */
4545 /* endline is the dual of begline. */
4547 DEBUG_PRINT1 ("EXECUTING endline.\n");
4549 if (AT_STRINGS_END (d))
4551 if (!bufp->not_eol) break;
4554 /* We have to ``prefetch'' the next character. */
4555 else if ((d == end1 ? *string2 : *d) == '\n'
4556 && bufp->newline_anchor)
4563 /* Match at the very beginning of the data. */
4565 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4566 if (AT_STRINGS_BEG (d))
4571 /* Match at the very end of the data. */
4573 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4574 if (AT_STRINGS_END (d))
4579 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4580 pushes NULL as the value for the string on the stack. Then
4581 `pop_failure_point' will keep the current value for the
4582 string, instead of restoring it. To see why, consider
4583 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4584 then the . fails against the \n. But the next thing we want
4585 to do is match the \n against the \n; if we restored the
4586 string value, we would be back at the foo.
4588 Because this is used only in specific cases, we don't need to
4589 check all the things that `on_failure_jump' does, to make
4590 sure the right things get saved on the stack. Hence we don't
4591 share its code. The only reason to push anything on the
4592 stack at all is that otherwise we would have to change
4593 `anychar's code to do something besides goto fail in this
4594 case; that seems worse than this. */
4595 case on_failure_keep_string_jump:
4596 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4598 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4600 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
4602 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4605 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4609 /* Uses of on_failure_jump:
4611 Each alternative starts with an on_failure_jump that points
4612 to the beginning of the next alternative. Each alternative
4613 except the last ends with a jump that in effect jumps past
4614 the rest of the alternatives. (They really jump to the
4615 ending jump of the following alternative, because tensioning
4616 these jumps is a hassle.)
4618 Repeats start with an on_failure_jump that points past both
4619 the repetition text and either the following jump or
4620 pop_failure_jump back to this on_failure_jump. */
4621 case on_failure_jump:
4623 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4625 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4627 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
4629 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4632 /* If this on_failure_jump comes right before a group (i.e.,
4633 the original * applied to a group), save the information
4634 for that group and all inner ones, so that if we fail back
4635 to this point, the group's information will be correct.
4636 For example, in \(a*\)*\1, we need the preceding group,
4637 and in \(zz\(a*\)b*\)\2, we need the inner group. */
4639 /* We can't use `p' to check ahead because we push
4640 a failure point to `p + mcnt' after we do this. */
4643 /* We need to skip no_op's before we look for the
4644 start_memory in case this on_failure_jump is happening as
4645 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4647 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4650 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4652 /* We have a new highest active register now. This will
4653 get reset at the start_memory we are about to get to,
4654 but we will have saved all the registers relevant to
4655 this repetition op, as described above. */
4656 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4657 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4658 lowest_active_reg = *(p1 + 1);
4661 DEBUG_PRINT1 (":\n");
4662 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4666 /* A smart repeat ends with `maybe_pop_jump'.
4667 We change it to either `pop_failure_jump' or `jump'. */
4668 case maybe_pop_jump:
4669 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4670 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4672 register unsigned char *p2 = p;
4674 /* Compare the beginning of the repeat with what in the
4675 pattern follows its end. If we can establish that there
4676 is nothing that they would both match, i.e., that we
4677 would have to backtrack because of (as in, e.g., `a*a')
4678 then we can change to pop_failure_jump, because we'll
4679 never have to backtrack.
4681 This is not true in the case of alternatives: in
4682 `(a|ab)*' we do need to backtrack to the `ab' alternative
4683 (e.g., if the string was `ab'). But instead of trying to
4684 detect that here, the alternative has put on a dummy
4685 failure point which is what we will end up popping. */
4687 /* Skip over open/close-group commands.
4688 If what follows this loop is a ...+ construct,
4689 look at what begins its body, since we will have to
4690 match at least one of that. */
4694 && ((re_opcode_t) *p2 == stop_memory
4695 || (re_opcode_t) *p2 == start_memory))
4697 else if (p2 + 6 < pend
4698 && (re_opcode_t) *p2 == dummy_failure_jump)
4705 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4706 to the `maybe_finalize_jump' of this case. Examine what
4709 /* If we're at the end of the pattern, we can change. */
4712 /* Consider what happens when matching ":\(.*\)"
4713 against ":/". I don't really understand this code
4715 p[-3] = (unsigned char) pop_failure_jump;
4717 (" End of pattern: change to `pop_failure_jump'.\n");
4720 else if ((re_opcode_t) *p2 == exactn
4721 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4723 register unsigned char c
4724 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4726 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4728 p[-3] = (unsigned char) pop_failure_jump;
4729 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4733 else if ((re_opcode_t) p1[3] == charset
4734 || (re_opcode_t) p1[3] == charset_not)
4736 int not = (re_opcode_t) p1[3] == charset_not;
4738 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4739 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4742 /* `not' is equal to 1 if c would match, which means
4743 that we can't change to pop_failure_jump. */
4746 p[-3] = (unsigned char) pop_failure_jump;
4747 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4751 else if ((re_opcode_t) *p2 == charset)
4754 register unsigned char c
4755 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4759 if ((re_opcode_t) p1[3] == exactn
4760 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
4761 && (p2[2 + p1[5] / BYTEWIDTH]
4762 & (1 << (p1[5] % BYTEWIDTH)))))
4764 if ((re_opcode_t) p1[3] == exactn
4765 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[4]
4766 && (p2[2 + p1[4] / BYTEWIDTH]
4767 & (1 << (p1[4] % BYTEWIDTH)))))
4770 p[-3] = (unsigned char) pop_failure_jump;
4771 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4775 else if ((re_opcode_t) p1[3] == charset_not)
4778 /* We win if the charset_not inside the loop
4779 lists every character listed in the charset after. */
4780 for (idx = 0; idx < (int) p2[1]; idx++)
4781 if (! (p2[2 + idx] == 0
4782 || (idx < (int) p1[4]
4783 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4788 p[-3] = (unsigned char) pop_failure_jump;
4789 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4792 else if ((re_opcode_t) p1[3] == charset)
4795 /* We win if the charset inside the loop
4796 has no overlap with the one after the loop. */
4798 idx < (int) p2[1] && idx < (int) p1[4];
4800 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4803 if (idx == p2[1] || idx == p1[4])
4805 p[-3] = (unsigned char) pop_failure_jump;
4806 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4811 p -= 2; /* Point at relative address again. */
4812 if ((re_opcode_t) p[-1] != pop_failure_jump)
4814 p[-1] = (unsigned char) jump;
4815 DEBUG_PRINT1 (" Match => jump.\n");
4816 goto unconditional_jump;
4818 /* Note fall through. */
4821 /* The end of a simple repeat has a pop_failure_jump back to
4822 its matching on_failure_jump, where the latter will push a
4823 failure point. The pop_failure_jump takes off failure
4824 points put on by this pop_failure_jump's matching
4825 on_failure_jump; we got through the pattern to here from the
4826 matching on_failure_jump, so didn't fail. */
4827 case pop_failure_jump:
4829 /* We need to pass separate storage for the lowest and
4830 highest registers, even though we don't care about the
4831 actual values. Otherwise, we will restore only one
4832 register from the stack, since lowest will == highest in
4833 `pop_failure_point'. */
4834 active_reg_t dummy_low_reg, dummy_high_reg;
4835 unsigned char *pdummy;
4838 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4839 POP_FAILURE_POINT (sdummy, pdummy,
4840 dummy_low_reg, dummy_high_reg,
4841 reg_dummy, reg_dummy, reg_info_dummy);
4843 /* Note fall through. */
4847 DEBUG_PRINT2 ("\n%p: ", p);
4849 DEBUG_PRINT2 ("\n0x%x: ", p);
4851 /* Note fall through. */
4853 /* Unconditionally jump (without popping any failure points). */
4855 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4856 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4857 p += mcnt; /* Do the jump. */
4859 DEBUG_PRINT2 ("(to %p).\n", p);
4861 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4866 /* We need this opcode so we can detect where alternatives end
4867 in `group_match_null_string_p' et al. */
4869 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4870 goto unconditional_jump;
4873 /* Normally, the on_failure_jump pushes a failure point, which
4874 then gets popped at pop_failure_jump. We will end up at
4875 pop_failure_jump, also, and with a pattern of, say, `a+', we
4876 are skipping over the on_failure_jump, so we have to push
4877 something meaningless for pop_failure_jump to pop. */
4878 case dummy_failure_jump:
4879 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4880 /* It doesn't matter what we push for the string here. What
4881 the code at `fail' tests is the value for the pattern. */
4882 PUSH_FAILURE_POINT (NULL, NULL, -2);
4883 goto unconditional_jump;
4886 /* At the end of an alternative, we need to push a dummy failure
4887 point in case we are followed by a `pop_failure_jump', because
4888 we don't want the failure point for the alternative to be
4889 popped. For example, matching `(a|ab)*' against `aab'
4890 requires that we match the `ab' alternative. */
4891 case push_dummy_failure:
4892 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4893 /* See comments just above at `dummy_failure_jump' about the
4895 PUSH_FAILURE_POINT (NULL, NULL, -2);
4898 /* Have to succeed matching what follows at least n times.
4899 After that, handle like `on_failure_jump'. */
4901 EXTRACT_NUMBER (mcnt, p + 2);
4902 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4905 /* Originally, this is how many times we HAVE to succeed. */
4910 STORE_NUMBER_AND_INCR (p, mcnt);
4912 DEBUG_PRINT3 (" Setting %p to %d.\n", p - 2, mcnt);
4914 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - 2, mcnt);
4920 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n", p+2);
4922 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4924 p[2] = (unsigned char) no_op;
4925 p[3] = (unsigned char) no_op;
4931 EXTRACT_NUMBER (mcnt, p + 2);
4932 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4934 /* Originally, this is how many times we CAN jump. */
4938 STORE_NUMBER (p + 2, mcnt);
4940 DEBUG_PRINT3 (" Setting %p to %d.\n", p + 2, mcnt);
4942 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + 2, mcnt);
4944 goto unconditional_jump;
4946 /* If don't have to jump any more, skip over the rest of command. */
4953 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4955 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4957 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4959 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
4961 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4963 STORE_NUMBER (p1, mcnt);
4968 /* The DEC Alpha C compiler 3.x generates incorrect code for the
4969 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4970 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4971 macro and introducing temporary variables works around the bug. */
4974 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4975 if (AT_WORD_BOUNDARY (d))
4980 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4981 if (AT_WORD_BOUNDARY (d))
4987 boolean prevchar, thischar;
4989 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4990 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
4993 prevchar = WORDCHAR_P (d - 1);
4994 thischar = WORDCHAR_P (d);
4995 if (prevchar != thischar)
5002 boolean prevchar, thischar;
5004 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
5005 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5008 prevchar = WORDCHAR_P (d - 1);
5009 thischar = WORDCHAR_P (d);
5010 if (prevchar != thischar)
5017 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
5018 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
5023 DEBUG_PRINT1 ("EXECUTING wordend.\n");
5024 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
5025 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
5031 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
5032 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
5037 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
5038 if (PTR_CHAR_POS ((unsigned char *) d) != point)
5043 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
5044 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
5049 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
5054 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
5058 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5060 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
5062 SET_REGS_MATCHED ();
5066 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
5068 goto matchnotsyntax;
5071 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
5075 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5077 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
5079 SET_REGS_MATCHED ();
5082 #else /* not emacs */
5084 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
5086 if (!WORDCHAR_P (d))
5088 SET_REGS_MATCHED ();
5093 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
5097 SET_REGS_MATCHED ();
5100 #endif /* not emacs */
5105 continue; /* Successfully executed one pattern command; keep going. */
5108 /* We goto here if a matching operation fails. */
5110 if (!FAIL_STACK_EMPTY ())
5111 { /* A restart point is known. Restore to that state. */
5112 DEBUG_PRINT1 ("\nFAIL:\n");
5113 POP_FAILURE_POINT (d, p,
5114 lowest_active_reg, highest_active_reg,
5115 regstart, regend, reg_info);
5117 /* If this failure point is a dummy, try the next one. */
5121 /* If we failed to the end of the pattern, don't examine *p. */
5125 boolean is_a_jump_n = false;
5127 /* If failed to a backwards jump that's part of a repetition
5128 loop, need to pop this failure point and use the next one. */
5129 switch ((re_opcode_t) *p)
5133 case maybe_pop_jump:
5134 case pop_failure_jump:
5137 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5140 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
5142 && (re_opcode_t) *p1 == on_failure_jump))
5150 if (d >= string1 && d <= end1)
5154 break; /* Matching at this starting point really fails. */
5158 goto restore_best_regs;
5162 return -1; /* Failure to match. */
5165 /* Subroutine definitions for re_match_2. */
5168 /* We are passed P pointing to a register number after a start_memory.
5170 Return true if the pattern up to the corresponding stop_memory can
5171 match the empty string, and false otherwise.
5173 If we find the matching stop_memory, sets P to point to one past its number.
5174 Otherwise, sets P to an undefined byte less than or equal to END.
5176 We don't handle duplicates properly (yet). */
5179 group_match_null_string_p (p, end, reg_info)
5180 unsigned char **p, *end;
5181 register_info_type *reg_info;
5184 /* Point to after the args to the start_memory. */
5185 unsigned char *p1 = *p + 2;
5189 /* Skip over opcodes that can match nothing, and return true or
5190 false, as appropriate, when we get to one that can't, or to the
5191 matching stop_memory. */
5193 switch ((re_opcode_t) *p1)
5195 /* Could be either a loop or a series of alternatives. */
5196 case on_failure_jump:
5198 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5200 /* If the next operation is not a jump backwards in the
5205 /* Go through the on_failure_jumps of the alternatives,
5206 seeing if any of the alternatives cannot match nothing.
5207 The last alternative starts with only a jump,
5208 whereas the rest start with on_failure_jump and end
5209 with a jump, e.g., here is the pattern for `a|b|c':
5211 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
5212 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
5215 So, we have to first go through the first (n-1)
5216 alternatives and then deal with the last one separately. */
5219 /* Deal with the first (n-1) alternatives, which start
5220 with an on_failure_jump (see above) that jumps to right
5221 past a jump_past_alt. */
5223 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
5225 /* `mcnt' holds how many bytes long the alternative
5226 is, including the ending `jump_past_alt' and
5229 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
5233 /* Move to right after this alternative, including the
5237 /* Break if it's the beginning of an n-th alternative
5238 that doesn't begin with an on_failure_jump. */
5239 if ((re_opcode_t) *p1 != on_failure_jump)
5242 /* Still have to check that it's not an n-th
5243 alternative that starts with an on_failure_jump. */
5245 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5246 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
5248 /* Get to the beginning of the n-th alternative. */
5254 /* Deal with the last alternative: go back and get number
5255 of the `jump_past_alt' just before it. `mcnt' contains
5256 the length of the alternative. */
5257 EXTRACT_NUMBER (mcnt, p1 - 2);
5259 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
5262 p1 += mcnt; /* Get past the n-th alternative. */
5268 assert (p1[1] == **p);
5274 if (!common_op_match_null_string_p (&p1, end, reg_info))
5277 } /* while p1 < end */
5280 } /* group_match_null_string_p */
5283 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5284 It expects P to be the first byte of a single alternative and END one
5285 byte past the last. The alternative can contain groups. */
5288 alt_match_null_string_p (p, end, reg_info)
5289 unsigned char *p, *end;
5290 register_info_type *reg_info;
5293 unsigned char *p1 = p;
5297 /* Skip over opcodes that can match nothing, and break when we get
5298 to one that can't. */
5300 switch ((re_opcode_t) *p1)
5303 case on_failure_jump:
5305 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5310 if (!common_op_match_null_string_p (&p1, end, reg_info))
5313 } /* while p1 < end */
5316 } /* alt_match_null_string_p */
5319 /* Deals with the ops common to group_match_null_string_p and
5320 alt_match_null_string_p.
5322 Sets P to one after the op and its arguments, if any. */
5325 common_op_match_null_string_p (p, end, reg_info)
5326 unsigned char **p, *end;
5327 register_info_type *reg_info;
5332 unsigned char *p1 = *p;
5334 switch ((re_opcode_t) *p1++)
5354 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5355 ret = group_match_null_string_p (&p1, end, reg_info);
5357 /* Have to set this here in case we're checking a group which
5358 contains a group and a back reference to it. */
5360 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5361 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5367 /* If this is an optimized succeed_n for zero times, make the jump. */
5369 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5377 /* Get to the number of times to succeed. */
5379 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5384 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5392 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5400 /* All other opcodes mean we cannot match the empty string. */
5406 } /* common_op_match_null_string_p */
5409 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5410 bytes; nonzero otherwise. */
5413 bcmp_translate (s1, s2, len, translate)
5414 const char *s1, *s2;
5416 RE_TRANSLATE_TYPE translate;
5418 register const unsigned char *p1 = (const unsigned char *) s1;
5419 register const unsigned char *p2 = (const unsigned char *) s2;
5422 if (translate[*p1++] != translate[*p2++]) return 1;
5428 /* Entry points for GNU code. */
5430 /* re_compile_pattern is the GNU regular expression compiler: it
5431 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5432 Returns 0 if the pattern was valid, otherwise an error string.
5434 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5435 are set in BUFP on entry.
5437 We call regex_compile to do the actual compilation. */
5440 re_compile_pattern (pattern, length, bufp)
5441 const char *pattern;
5443 struct re_pattern_buffer *bufp;
5447 /* GNU code is written to assume at least RE_NREGS registers will be set
5448 (and at least one extra will be -1). */
5449 bufp->regs_allocated = REGS_UNALLOCATED;
5451 /* And GNU code determines whether or not to get register information
5452 by passing null for the REGS argument to re_match, etc., not by
5456 /* Match anchors at newline. */
5457 bufp->newline_anchor = 1;
5459 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5463 return gettext (re_error_msgid[(int) ret]);
5466 weak_alias (__re_compile_pattern, re_compile_pattern)
5469 /* Entry points compatible with 4.2 BSD regex library. We don't define
5470 them unless specifically requested. */
5472 #if defined _REGEX_RE_COMP || defined _LIBC
5474 /* BSD has one and only one pattern buffer. */
5475 static struct re_pattern_buffer re_comp_buf;
5479 /* Make these definitions weak in libc, so POSIX programs can redefine
5480 these names if they don't use our functions, and still use
5481 regcomp/regexec below without link errors. */
5491 if (!re_comp_buf.buffer)
5492 return gettext ("No previous regular expression");
5496 if (!re_comp_buf.buffer)
5498 re_comp_buf.buffer = (unsigned char *) malloc (200);
5499 if (re_comp_buf.buffer == NULL)
5500 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
5501 re_comp_buf.allocated = 200;
5503 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5504 if (re_comp_buf.fastmap == NULL)
5505 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
5508 /* Since `re_exec' always passes NULL for the `regs' argument, we
5509 don't need to initialize the pattern buffer fields which affect it. */
5511 /* Match anchors at newlines. */
5512 re_comp_buf.newline_anchor = 1;
5514 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5519 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5520 return (char *) gettext (re_error_msgid[(int) ret]);
5531 const int len = strlen (s);
5533 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5536 #endif /* _REGEX_RE_COMP */
5538 /* POSIX.2 functions. Don't define these for Emacs. */
5542 /* regcomp takes a regular expression as a string and compiles it.
5544 PREG is a regex_t *. We do not expect any fields to be initialized,
5545 since POSIX says we shouldn't. Thus, we set
5547 `buffer' to the compiled pattern;
5548 `used' to the length of the compiled pattern;
5549 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5550 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5551 RE_SYNTAX_POSIX_BASIC;
5552 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5553 `fastmap' and `fastmap_accurate' to zero;
5554 `re_nsub' to the number of subexpressions in PATTERN.
5556 PATTERN is the address of the pattern string.
5558 CFLAGS is a series of bits which affect compilation.
5560 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5561 use POSIX basic syntax.
5563 If REG_NEWLINE is set, then . and [^...] don't match newline.
5564 Also, regexec will try a match beginning after every newline.
5566 If REG_ICASE is set, then we considers upper- and lowercase
5567 versions of letters to be equivalent when matching.
5569 If REG_NOSUB is set, then when PREG is passed to regexec, that
5570 routine will report only success or failure, and nothing about the
5573 It returns 0 if it succeeds, nonzero if it doesn't. (See gnu-regex.h for
5574 the return codes and their meanings.) */
5577 regcomp (preg, pattern, cflags)
5579 const char *pattern;
5584 = (cflags & REG_EXTENDED) ?
5585 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5587 /* regex_compile will allocate the space for the compiled pattern. */
5589 preg->allocated = 0;
5592 /* Don't bother to use a fastmap when searching. This simplifies the
5593 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
5594 characters after newlines into the fastmap. This way, we just try
5598 if (cflags & REG_ICASE)
5603 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
5604 * sizeof (*(RE_TRANSLATE_TYPE)0));
5605 if (preg->translate == NULL)
5606 return (int) REG_ESPACE;
5608 /* Map uppercase characters to corresponding lowercase ones. */
5609 for (i = 0; i < CHAR_SET_SIZE; i++)
5610 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
5613 preg->translate = NULL;
5615 /* If REG_NEWLINE is set, newlines are treated differently. */
5616 if (cflags & REG_NEWLINE)
5617 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5618 syntax &= ~RE_DOT_NEWLINE;
5619 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5620 /* It also changes the matching behavior. */
5621 preg->newline_anchor = 1;
5624 preg->newline_anchor = 0;
5626 preg->no_sub = !!(cflags & REG_NOSUB);
5628 /* POSIX says a null character in the pattern terminates it, so we
5629 can use strlen here in compiling the pattern. */
5630 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5632 /* POSIX doesn't distinguish between an unmatched open-group and an
5633 unmatched close-group: both are REG_EPAREN. */
5634 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5639 weak_alias (__regcomp, regcomp)
5643 /* regexec searches for a given pattern, specified by PREG, in the
5646 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5647 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5648 least NMATCH elements, and we set them to the offsets of the
5649 corresponding matched substrings.
5651 EFLAGS specifies `execution flags' which affect matching: if
5652 REG_NOTBOL is set, then ^ does not match at the beginning of the
5653 string; if REG_NOTEOL is set, then $ does not match at the end.
5655 We return 0 if we find a match and REG_NOMATCH if not. */
5658 regexec (preg, string, nmatch, pmatch, eflags)
5659 const regex_t *preg;
5662 regmatch_t pmatch[];
5666 struct re_registers regs;
5667 regex_t private_preg;
5668 int len = strlen (string);
5669 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5671 private_preg = *preg;
5673 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5674 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5676 /* The user has told us exactly how many registers to return
5677 information about, via `nmatch'. We have to pass that on to the
5678 matching routines. */
5679 private_preg.regs_allocated = REGS_FIXED;
5683 regs.num_regs = nmatch;
5684 regs.start = TALLOC (nmatch, regoff_t);
5685 regs.end = TALLOC (nmatch, regoff_t);
5686 if (regs.start == NULL || regs.end == NULL)
5687 return (int) REG_NOMATCH;
5690 /* Perform the searching operation. */
5691 ret = re_search (&private_preg, string, len,
5692 /* start: */ 0, /* range: */ len,
5693 want_reg_info ? ®s : (struct re_registers *) 0);
5695 /* Copy the register information to the POSIX structure. */
5702 for (r = 0; r < nmatch; r++)
5704 pmatch[r].rm_so = regs.start[r];
5705 pmatch[r].rm_eo = regs.end[r];
5709 /* If we needed the temporary register info, free the space now. */
5714 /* We want zero return to mean success, unlike `re_search'. */
5715 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5718 weak_alias (__regexec, regexec)
5722 /* Returns a message corresponding to an error code, ERRCODE, returned
5723 from either regcomp or regexec. We don't use PREG here. */
5726 __regerror (errcode, preg, errbuf, errbuf_size)
5728 const regex_t *preg;
5736 || errcode >= (int) (sizeof (re_error_msgid)
5737 / sizeof (re_error_msgid[0])))
5738 /* Only error codes returned by the rest of the code should be passed
5739 to this routine. If we are given anything else, or if other regex
5740 code generates an invalid error code, then the program has a bug.
5741 Dump core so we can fix it. */
5744 msg = gettext (re_error_msgid[errcode]);
5746 msg_size = strlen (msg) + 1; /* Includes the null. */
5748 if (errbuf_size != 0)
5750 if (msg_size > errbuf_size)
5752 #if defined HAVE_MEMPCPY || defined _LIBC
5753 *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
5755 memcpy (errbuf, msg, errbuf_size - 1);
5756 errbuf[errbuf_size - 1] = 0;
5760 memcpy (errbuf, msg, msg_size);
5766 weak_alias (__regerror, regerror)
5770 /* Free dynamically allocated space used by PREG. */
5776 if (preg->buffer != NULL)
5777 free (preg->buffer);
5778 preg->buffer = NULL;
5780 preg->allocated = 0;
5783 if (preg->fastmap != NULL)
5784 free (preg->fastmap);
5785 preg->fastmap = NULL;
5786 preg->fastmap_accurate = 0;
5788 if (preg->translate != NULL)
5789 free (preg->translate);
5790 preg->translate = NULL;
5793 weak_alias (__regfree, regfree)
5796 #endif /* not emacs */
projects / external / binutils.git / blob