1 /* *INDENT-OFF* */ /* keep in sync with glibc */
2 /* Extended regular expression matching and search library,
4 (Implements POSIX draft P1003.2/D11.2, except for some of the
5 internationalization features.)
6 Copyright 1993, 1994, 1995, 1996, 1998, 1999, 2000
7 Free Software Foundation, Inc.
9 NOTE: The canonical source of this file is maintained with the
10 GNU C Library. Bugs can be reported to bug-glibc@gnu.org.
12 This program is free software; you can redistribute it and/or modify it
13 under the terms of the GNU General Public License as published by the
14 Free Software Foundation; either version 2, or (at your option) any
17 This program is distributed in the hope that it will be useful,
18 but WITHOUT ANY WARRANTY; without even the implied warranty of
19 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 GNU General Public License for more details.
22 You should have received a copy of the GNU General Public License
23 along with this program; if not, write to the Free Software Foundation,
24 Inc., 59 Temple Place - Suite 330,
25 Boston, MA 02111-1307, USA. */
27 /* AIX requires this to be the first thing in the file. */
28 #if defined _AIX && !defined REGEX_MALLOC
40 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
41 # define PARAMS(args) args
43 # define PARAMS(args) ()
45 #endif /* Not PARAMS. */
47 #if defined STDC_HEADERS && !defined emacs
50 /* We need this for `gnu-regex.h', and perhaps for the Emacs include files. */
51 # include <sys/types.h>
54 /* For platform which support the ISO C amendement 1 functionality we
55 support user defined character classes. */
56 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H)
57 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
62 /* This is for other GNU distributions with internationalized messages. */
63 /* CYGNUS LOCAL: ../intl will handle this for us */
67 # define gettext(msgid) (msgid)
71 /* This define is so xgettext can find the internationalizable
73 # define gettext_noop(String) String
76 /* The `emacs' switch turns on certain matching commands
77 that make sense only in Emacs. */
86 /* If we are not linking with Emacs proper,
87 we can't use the relocating allocator
88 even if config.h says that we can. */
91 # if defined STDC_HEADERS || defined _LIBC
98 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
99 If nothing else has been done, use the method below. */
100 # ifdef INHIBIT_STRING_HEADER
101 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
102 # if !defined bzero && !defined bcopy
103 # undef INHIBIT_STRING_HEADER
108 /* This is the normal way of making sure we have a bcopy and a bzero.
109 This is used in most programs--a few other programs avoid this
110 by defining INHIBIT_STRING_HEADER. */
111 # ifndef INHIBIT_STRING_HEADER
112 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
116 # define bzero(s, n) (memset (s, '\0', n), (s))
118 # define bzero(s, n) __bzero (s, n)
122 # include <strings.h>
124 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
127 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
132 /* Define the syntax stuff for \<, \>, etc. */
134 /* This must be nonzero for the wordchar and notwordchar pattern
135 commands in re_match_2. */
140 # ifdef SWITCH_ENUM_BUG
141 # define SWITCH_ENUM_CAST(x) ((int)(x))
143 # define SWITCH_ENUM_CAST(x) (x)
146 /* How many characters in the character set. */
147 # define CHAR_SET_SIZE 256
149 /* GDB LOCAL: define _REGEX_RE_COMP to get BSD style re_comp and re_exec */
150 #ifndef _REGEX_RE_COMP
151 #define _REGEX_RE_COMP
156 extern char *re_syntax_table;
158 # else /* not SYNTAX_TABLE */
160 static char re_syntax_table[CHAR_SET_SIZE];
171 bzero (re_syntax_table, sizeof re_syntax_table);
173 for (c = 'a'; c <= 'z'; c++)
174 re_syntax_table[c] = Sword;
176 for (c = 'A'; c <= 'Z'; c++)
177 re_syntax_table[c] = Sword;
179 for (c = '0'; c <= '9'; c++)
180 re_syntax_table[c] = Sword;
182 re_syntax_table['_'] = Sword;
187 # endif /* not SYNTAX_TABLE */
189 # define SYNTAX(c) re_syntax_table[c]
191 #endif /* not emacs */
193 /* Get the interface, including the syntax bits. */
194 /* CYGNUS LOCAL: call it gnu-regex.h, not regex.h, to avoid name conflicts */
195 #include "gnu-regex.h"
197 /* isalpha etc. are used for the character classes. */
200 /* Jim Meyering writes:
202 "... Some ctype macros are valid only for character codes that
203 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
204 using /bin/cc or gcc but without giving an ansi option). So, all
205 ctype uses should be through macros like ISPRINT... If
206 STDC_HEADERS is defined, then autoconf has verified that the ctype
207 macros don't need to be guarded with references to isascii. ...
208 Defining isascii to 1 should let any compiler worth its salt
209 eliminate the && through constant folding."
210 Solaris defines some of these symbols so we must undefine them first. */
213 #if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
214 # define ISASCII(c) 1
216 # define ISASCII(c) isascii(c)
220 # define ISBLANK(c) (ISASCII (c) && isblank (c))
222 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
225 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
227 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
231 #define ISPRINT(c) (ISASCII (c) && isprint (c))
232 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
233 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
234 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
235 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
236 #define ISLOWER(c) (ISASCII (c) && islower (c))
237 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
238 #define ISSPACE(c) (ISASCII (c) && isspace (c))
239 #define ISUPPER(c) (ISASCII (c) && isupper (c))
240 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
243 # define NULL (void *)0
246 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
247 since ours (we hope) works properly with all combinations of
248 machines, compilers, `char' and `unsigned char' argument types.
249 (Per Bothner suggested the basic approach.) */
250 #undef SIGN_EXTEND_CHAR
252 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
253 #else /* not __STDC__ */
254 /* As in Harbison and Steele. */
255 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
258 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
259 use `alloca' instead of `malloc'. This is because using malloc in
260 re_search* or re_match* could cause memory leaks when C-g is used in
261 Emacs; also, malloc is slower and causes storage fragmentation. On
262 the other hand, malloc is more portable, and easier to debug.
264 Because we sometimes use alloca, some routines have to be macros,
265 not functions -- `alloca'-allocated space disappears at the end of the
266 function it is called in. */
270 # define REGEX_ALLOCATE malloc
271 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
272 # define REGEX_FREE free
274 #else /* not REGEX_MALLOC */
276 /* Emacs already defines alloca, sometimes. */
279 /* Make alloca work the best possible way. */
281 # define alloca __builtin_alloca
282 # else /* not __GNUC__ */
285 # endif /* HAVE_ALLOCA_H */
286 # endif /* not __GNUC__ */
288 # endif /* not alloca */
290 # define REGEX_ALLOCATE alloca
292 /* Assumes a `char *destination' variable. */
293 # define REGEX_REALLOCATE(source, osize, nsize) \
294 (destination = (char *) alloca (nsize), \
295 memcpy (destination, source, osize))
297 /* No need to do anything to free, after alloca. */
298 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
300 #endif /* not REGEX_MALLOC */
302 /* Define how to allocate the failure stack. */
304 #if defined REL_ALLOC && defined REGEX_MALLOC
306 # define REGEX_ALLOCATE_STACK(size) \
307 r_alloc (&failure_stack_ptr, (size))
308 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
309 r_re_alloc (&failure_stack_ptr, (nsize))
310 # define REGEX_FREE_STACK(ptr) \
311 r_alloc_free (&failure_stack_ptr)
313 #else /* not using relocating allocator */
317 # define REGEX_ALLOCATE_STACK malloc
318 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
319 # define REGEX_FREE_STACK free
321 # else /* not REGEX_MALLOC */
323 # define REGEX_ALLOCATE_STACK alloca
325 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
326 REGEX_REALLOCATE (source, osize, nsize)
327 /* No need to explicitly free anything. */
328 # define REGEX_FREE_STACK(arg)
330 # endif /* not REGEX_MALLOC */
331 #endif /* not using relocating allocator */
334 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
335 `string1' or just past its end. This works if PTR is NULL, which is
337 #define FIRST_STRING_P(ptr) \
338 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
340 /* (Re)Allocate N items of type T using malloc, or fail. */
341 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
342 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
343 #define RETALLOC_IF(addr, n, t) \
344 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
345 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
347 #define BYTEWIDTH 8 /* In bits. */
349 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
353 #define MAX(a, b) ((a) > (b) ? (a) : (b))
354 #define MIN(a, b) ((a) < (b) ? (a) : (b))
356 typedef char boolean;
360 static int re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
361 const char *string1, int size1,
362 const char *string2, int size2,
364 struct re_registers *regs,
367 /* These are the command codes that appear in compiled regular
368 expressions. Some opcodes are followed by argument bytes. A
369 command code can specify any interpretation whatsoever for its
370 arguments. Zero bytes may appear in the compiled regular expression. */
376 /* Succeed right away--no more backtracking. */
379 /* Followed by one byte giving n, then by n literal bytes. */
382 /* Matches any (more or less) character. */
385 /* Matches any one char belonging to specified set. First
386 following byte is number of bitmap bytes. Then come bytes
387 for a bitmap saying which chars are in. Bits in each byte
388 are ordered low-bit-first. A character is in the set if its
389 bit is 1. A character too large to have a bit in the map is
390 automatically not in the set. */
393 /* Same parameters as charset, but match any character that is
394 not one of those specified. */
397 /* Start remembering the text that is matched, for storing in a
398 register. Followed by one byte with the register number, in
399 the range 0 to one less than the pattern buffer's re_nsub
400 field. Then followed by one byte with the number of groups
401 inner to this one. (This last has to be part of the
402 start_memory only because we need it in the on_failure_jump
406 /* Stop remembering the text that is matched and store it in a
407 memory register. Followed by one byte with the register
408 number, in the range 0 to one less than `re_nsub' in the
409 pattern buffer, and one byte with the number of inner groups,
410 just like `start_memory'. (We need the number of inner
411 groups here because we don't have any easy way of finding the
412 corresponding start_memory when we're at a stop_memory.) */
415 /* Match a duplicate of something remembered. Followed by one
416 byte containing the register number. */
419 /* Fail unless at beginning of line. */
422 /* Fail unless at end of line. */
425 /* Succeeds if at beginning of buffer (if emacs) or at beginning
426 of string to be matched (if not). */
429 /* Analogously, for end of buffer/string. */
432 /* Followed by two byte relative address to which to jump. */
435 /* Same as jump, but marks the end of an alternative. */
438 /* Followed by two-byte relative address of place to resume at
439 in case of failure. */
442 /* Like on_failure_jump, but pushes a placeholder instead of the
443 current string position when executed. */
444 on_failure_keep_string_jump,
446 /* Throw away latest failure point and then jump to following
447 two-byte relative address. */
450 /* Change to pop_failure_jump if know won't have to backtrack to
451 match; otherwise change to jump. This is used to jump
452 back to the beginning of a repeat. If what follows this jump
453 clearly won't match what the repeat does, such that we can be
454 sure that there is no use backtracking out of repetitions
455 already matched, then we change it to a pop_failure_jump.
456 Followed by two-byte address. */
459 /* Jump to following two-byte address, and push a dummy failure
460 point. This failure point will be thrown away if an attempt
461 is made to use it for a failure. A `+' construct makes this
462 before the first repeat. Also used as an intermediary kind
463 of jump when compiling an alternative. */
466 /* Push a dummy failure point and continue. Used at the end of
470 /* Followed by two-byte relative address and two-byte number n.
471 After matching N times, jump to the address upon failure. */
474 /* Followed by two-byte relative address, and two-byte number n.
475 Jump to the address N times, then fail. */
478 /* Set the following two-byte relative address to the
479 subsequent two-byte number. The address *includes* the two
483 wordchar, /* Matches any word-constituent character. */
484 notwordchar, /* Matches any char that is not a word-constituent. */
486 wordbeg, /* Succeeds if at word beginning. */
487 wordend, /* Succeeds if at word end. */
489 wordbound, /* Succeeds if at a word boundary. */
490 notwordbound /* Succeeds if not at a word boundary. */
493 ,before_dot, /* Succeeds if before point. */
494 at_dot, /* Succeeds if at point. */
495 after_dot, /* Succeeds if after point. */
497 /* Matches any character whose syntax is specified. Followed by
498 a byte which contains a syntax code, e.g., Sword. */
501 /* Matches any character whose syntax is not that specified. */
506 /* Common operations on the compiled pattern. */
508 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
510 #define STORE_NUMBER(destination, number) \
512 (destination)[0] = (number) & 0377; \
513 (destination)[1] = (number) >> 8; \
516 /* Same as STORE_NUMBER, except increment DESTINATION to
517 the byte after where the number is stored. Therefore, DESTINATION
518 must be an lvalue. */
520 #define STORE_NUMBER_AND_INCR(destination, number) \
522 STORE_NUMBER (destination, number); \
523 (destination) += 2; \
526 /* Put into DESTINATION a number stored in two contiguous bytes starting
529 #define EXTRACT_NUMBER(destination, source) \
531 (destination) = *(source) & 0377; \
532 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
536 static void extract_number _RE_ARGS ((int *dest, unsigned char *source));
538 extract_number (dest, source)
540 unsigned char *source;
542 int temp = SIGN_EXTEND_CHAR (*(source + 1));
543 *dest = *source & 0377;
547 # ifndef EXTRACT_MACROS /* To debug the macros. */
548 # undef EXTRACT_NUMBER
549 # define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
550 # endif /* not EXTRACT_MACROS */
554 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
555 SOURCE must be an lvalue. */
557 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
559 EXTRACT_NUMBER (destination, source); \
564 static void extract_number_and_incr _RE_ARGS ((int *destination,
565 unsigned char **source));
567 extract_number_and_incr (destination, source)
569 unsigned char **source;
571 extract_number (destination, *source);
575 # ifndef EXTRACT_MACROS
576 # undef EXTRACT_NUMBER_AND_INCR
577 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
578 extract_number_and_incr (&dest, &src)
579 # endif /* not EXTRACT_MACROS */
583 /* If DEBUG is defined, Regex prints many voluminous messages about what
584 it is doing (if the variable `debug' is nonzero). If linked with the
585 main program in `iregex.c', you can enter patterns and strings
586 interactively. And if linked with the main program in `main.c' and
587 the other test files, you can run the already-written tests. */
591 /* We use standard I/O for debugging. */
594 /* It is useful to test things that ``must'' be true when debugging. */
597 static int debug = 0;
599 # define DEBUG_STATEMENT(e) e
600 # define DEBUG_PRINT1(x) if (debug) printf (x)
601 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
602 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
603 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
604 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
605 if (debug) print_partial_compiled_pattern (s, e)
606 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
607 if (debug) print_double_string (w, s1, sz1, s2, sz2)
610 /* Print the fastmap in human-readable form. */
613 print_fastmap (fastmap)
616 unsigned was_a_range = 0;
619 while (i < (1 << BYTEWIDTH))
625 while (i < (1 << BYTEWIDTH) && fastmap[i])
641 /* Print a compiled pattern string in human-readable form, starting at
642 the START pointer into it and ending just before the pointer END. */
645 print_partial_compiled_pattern (start, end)
646 unsigned char *start;
651 unsigned char *p = start;
652 unsigned char *pend = end;
660 /* Loop over pattern commands. */
663 printf ("%d:\t", p - start);
665 switch ((re_opcode_t) *p++)
673 printf ("/exactn/%d", mcnt);
684 printf ("/start_memory/%d/%d", mcnt, *p++);
689 printf ("/stop_memory/%d/%d", mcnt, *p++);
693 printf ("/duplicate/%d", *p++);
703 register int c, last = -100;
704 register int in_range = 0;
706 printf ("/charset [%s",
707 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
709 assert (p + *p < pend);
711 for (c = 0; c < 256; c++)
713 && (p[1 + (c/8)] & (1 << (c % 8))))
715 /* Are we starting a range? */
716 if (last + 1 == c && ! in_range)
721 /* Have we broken a range? */
722 else if (last + 1 != c && in_range)
751 case on_failure_jump:
752 extract_number_and_incr (&mcnt, &p);
753 printf ("/on_failure_jump to %d", p + mcnt - start);
756 case on_failure_keep_string_jump:
757 extract_number_and_incr (&mcnt, &p);
758 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
761 case dummy_failure_jump:
762 extract_number_and_incr (&mcnt, &p);
763 printf ("/dummy_failure_jump to %d", p + mcnt - start);
766 case push_dummy_failure:
767 printf ("/push_dummy_failure");
771 extract_number_and_incr (&mcnt, &p);
772 printf ("/maybe_pop_jump to %d", p + mcnt - start);
775 case pop_failure_jump:
776 extract_number_and_incr (&mcnt, &p);
777 printf ("/pop_failure_jump to %d", p + mcnt - start);
781 extract_number_and_incr (&mcnt, &p);
782 printf ("/jump_past_alt to %d", p + mcnt - start);
786 extract_number_and_incr (&mcnt, &p);
787 printf ("/jump to %d", p + mcnt - start);
791 extract_number_and_incr (&mcnt, &p);
793 extract_number_and_incr (&mcnt2, &p);
794 printf ("/succeed_n to %d, %d times", p1 - start, mcnt2);
798 extract_number_and_incr (&mcnt, &p);
800 extract_number_and_incr (&mcnt2, &p);
801 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
805 extract_number_and_incr (&mcnt, &p);
807 extract_number_and_incr (&mcnt2, &p);
808 printf ("/set_number_at location %d to %d", p1 - start, mcnt2);
812 printf ("/wordbound");
816 printf ("/notwordbound");
828 printf ("/before_dot");
836 printf ("/after_dot");
840 printf ("/syntaxspec");
842 printf ("/%d", mcnt);
846 printf ("/notsyntaxspec");
848 printf ("/%d", mcnt);
853 printf ("/wordchar");
857 printf ("/notwordchar");
869 printf ("?%d", *(p-1));
875 printf ("%d:\tend of pattern.\n", p - start);
880 print_compiled_pattern (bufp)
881 struct re_pattern_buffer *bufp;
883 unsigned char *buffer = bufp->buffer;
885 print_partial_compiled_pattern (buffer, buffer + bufp->used);
886 printf ("%ld bytes used/%ld bytes allocated.\n",
887 bufp->used, bufp->allocated);
889 if (bufp->fastmap_accurate && bufp->fastmap)
891 printf ("fastmap: ");
892 print_fastmap (bufp->fastmap);
895 printf ("re_nsub: %d\t", bufp->re_nsub);
896 printf ("regs_alloc: %d\t", bufp->regs_allocated);
897 printf ("can_be_null: %d\t", bufp->can_be_null);
898 printf ("newline_anchor: %d\n", bufp->newline_anchor);
899 printf ("no_sub: %d\t", bufp->no_sub);
900 printf ("not_bol: %d\t", bufp->not_bol);
901 printf ("not_eol: %d\t", bufp->not_eol);
902 printf ("syntax: %lx\n", bufp->syntax);
903 /* Perhaps we should print the translate table? */
908 print_double_string (where, string1, size1, string2, size2)
921 if (FIRST_STRING_P (where))
923 for (this_char = where - string1; this_char < size1; this_char++)
924 putchar (string1[this_char]);
929 for (this_char = where - string2; this_char < size2; this_char++)
930 putchar (string2[this_char]);
941 #else /* not DEBUG */
946 # define DEBUG_STATEMENT(e)
947 # define DEBUG_PRINT1(x)
948 # define DEBUG_PRINT2(x1, x2)
949 # define DEBUG_PRINT3(x1, x2, x3)
950 # define DEBUG_PRINT4(x1, x2, x3, x4)
951 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
952 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
954 #endif /* not DEBUG */
956 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
957 also be assigned to arbitrarily: each pattern buffer stores its own
958 syntax, so it can be changed between regex compilations. */
959 /* This has no initializer because initialized variables in Emacs
960 become read-only after dumping. */
961 reg_syntax_t re_syntax_options;
964 /* Specify the precise syntax of regexps for compilation. This provides
965 for compatibility for various utilities which historically have
966 different, incompatible syntaxes.
968 The argument SYNTAX is a bit mask comprised of the various bits
969 defined in gnu-regex.h. We return the old syntax. */
972 re_set_syntax (syntax)
975 reg_syntax_t ret = re_syntax_options;
977 re_syntax_options = syntax;
979 if (syntax & RE_DEBUG)
981 else if (debug) /* was on but now is not */
987 weak_alias (__re_set_syntax, re_set_syntax)
990 /* This table gives an error message for each of the error codes listed
991 in gnu-regex.h. Obviously the order here has to be same as there.
992 POSIX doesn't require that we do anything for REG_NOERROR,
993 but why not be nice? */
995 static const char *re_error_msgid[] =
997 gettext_noop ("Success"), /* REG_NOERROR */
998 gettext_noop ("No match"), /* REG_NOMATCH */
999 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
1000 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
1001 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
1002 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
1003 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
1004 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
1005 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
1006 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
1007 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
1008 gettext_noop ("Invalid range end"), /* REG_ERANGE */
1009 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
1010 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
1011 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
1012 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
1013 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
1016 /* Avoiding alloca during matching, to placate r_alloc. */
1018 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1019 searching and matching functions should not call alloca. On some
1020 systems, alloca is implemented in terms of malloc, and if we're
1021 using the relocating allocator routines, then malloc could cause a
1022 relocation, which might (if the strings being searched are in the
1023 ralloc heap) shift the data out from underneath the regexp
1026 Here's another reason to avoid allocation: Emacs
1027 processes input from X in a signal handler; processing X input may
1028 call malloc; if input arrives while a matching routine is calling
1029 malloc, then we're scrod. But Emacs can't just block input while
1030 calling matching routines; then we don't notice interrupts when
1031 they come in. So, Emacs blocks input around all regexp calls
1032 except the matching calls, which it leaves unprotected, in the
1033 faith that they will not malloc. */
1035 /* Normally, this is fine. */
1036 #define MATCH_MAY_ALLOCATE
1038 /* When using GNU C, we are not REALLY using the C alloca, no matter
1039 what config.h may say. So don't take precautions for it. */
1044 /* The match routines may not allocate if (1) they would do it with malloc
1045 and (2) it's not safe for them to use malloc.
1046 Note that if REL_ALLOC is defined, matching would not use malloc for the
1047 failure stack, but we would still use it for the register vectors;
1048 so REL_ALLOC should not affect this. */
1049 #if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1050 # undef MATCH_MAY_ALLOCATE
1054 /* Failure stack declarations and macros; both re_compile_fastmap and
1055 re_match_2 use a failure stack. These have to be macros because of
1056 REGEX_ALLOCATE_STACK. */
1059 /* Number of failure points for which to initially allocate space
1060 when matching. If this number is exceeded, we allocate more
1061 space, so it is not a hard limit. */
1062 #ifndef INIT_FAILURE_ALLOC
1063 # define INIT_FAILURE_ALLOC 5
1066 /* Roughly the maximum number of failure points on the stack. Would be
1067 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1068 This is a variable only so users of regex can assign to it; we never
1069 change it ourselves. */
1073 # if defined MATCH_MAY_ALLOCATE
1074 /* 4400 was enough to cause a crash on Alpha OSF/1,
1075 whose default stack limit is 2mb. */
1076 long int re_max_failures = 4000;
1078 long int re_max_failures = 2000;
1081 union fail_stack_elt
1083 unsigned char *pointer;
1087 typedef union fail_stack_elt fail_stack_elt_t;
1091 fail_stack_elt_t *stack;
1092 unsigned long int size;
1093 unsigned long int avail; /* Offset of next open position. */
1096 #else /* not INT_IS_16BIT */
1098 # if defined MATCH_MAY_ALLOCATE
1099 /* 4400 was enough to cause a crash on Alpha OSF/1,
1100 whose default stack limit is 2mb. */
1101 int re_max_failures = 20000;
1103 int re_max_failures = 2000;
1106 union fail_stack_elt
1108 unsigned char *pointer;
1112 typedef union fail_stack_elt fail_stack_elt_t;
1116 fail_stack_elt_t *stack;
1118 unsigned avail; /* Offset of next open position. */
1121 #endif /* INT_IS_16BIT */
1123 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1124 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1125 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1128 /* Define macros to initialize and free the failure stack.
1129 Do `return -2' if the alloc fails. */
1131 #ifdef MATCH_MAY_ALLOCATE
1132 # define INIT_FAIL_STACK() \
1134 fail_stack.stack = (fail_stack_elt_t *) \
1135 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1137 if (fail_stack.stack == NULL) \
1140 fail_stack.size = INIT_FAILURE_ALLOC; \
1141 fail_stack.avail = 0; \
1144 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1146 # define INIT_FAIL_STACK() \
1148 fail_stack.avail = 0; \
1151 # define RESET_FAIL_STACK()
1155 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1157 Return 1 if succeeds, and 0 if either ran out of memory
1158 allocating space for it or it was already too large.
1160 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1162 #define DOUBLE_FAIL_STACK(fail_stack) \
1163 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1165 : ((fail_stack).stack = (fail_stack_elt_t *) \
1166 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1167 (fail_stack).size * sizeof (fail_stack_elt_t), \
1168 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1170 (fail_stack).stack == NULL \
1172 : ((fail_stack).size <<= 1, \
1176 /* Push pointer POINTER on FAIL_STACK.
1177 Return 1 if was able to do so and 0 if ran out of memory allocating
1179 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1180 ((FAIL_STACK_FULL () \
1181 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1183 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1186 /* Push a pointer value onto the failure stack.
1187 Assumes the variable `fail_stack'. Probably should only
1188 be called from within `PUSH_FAILURE_POINT'. */
1189 #define PUSH_FAILURE_POINTER(item) \
1190 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1192 /* This pushes an integer-valued item onto the failure stack.
1193 Assumes the variable `fail_stack'. Probably should only
1194 be called from within `PUSH_FAILURE_POINT'. */
1195 #define PUSH_FAILURE_INT(item) \
1196 fail_stack.stack[fail_stack.avail++].integer = (item)
1198 /* Push a fail_stack_elt_t value onto the failure stack.
1199 Assumes the variable `fail_stack'. Probably should only
1200 be called from within `PUSH_FAILURE_POINT'. */
1201 #define PUSH_FAILURE_ELT(item) \
1202 fail_stack.stack[fail_stack.avail++] = (item)
1204 /* These three POP... operations complement the three PUSH... operations.
1205 All assume that `fail_stack' is nonempty. */
1206 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1207 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1208 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1210 /* Used to omit pushing failure point id's when we're not debugging. */
1212 # define DEBUG_PUSH PUSH_FAILURE_INT
1213 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1215 # define DEBUG_PUSH(item)
1216 # define DEBUG_POP(item_addr)
1220 /* Push the information about the state we will need
1221 if we ever fail back to it.
1223 Requires variables fail_stack, regstart, regend, reg_info, and
1224 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1227 Does `return FAILURE_CODE' if runs out of memory. */
1229 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1231 char *destination; \
1232 /* Must be int, so when we don't save any registers, the arithmetic \
1233 of 0 + -1 isn't done as unsigned. */ \
1234 /* Can't be int, since there is not a shred of a guarantee that int \
1235 is wide enough to hold a value of something to which pointer can \
1237 active_reg_t this_reg; \
1239 DEBUG_STATEMENT (failure_id++); \
1240 DEBUG_STATEMENT (nfailure_points_pushed++); \
1241 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1242 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1243 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1245 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1246 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1248 /* Ensure we have enough space allocated for what we will push. */ \
1249 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1251 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1252 return failure_code; \
1254 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1255 (fail_stack).size); \
1256 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1259 /* Push the info, starting with the registers. */ \
1260 DEBUG_PRINT1 ("\n"); \
1263 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1266 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1267 DEBUG_STATEMENT (num_regs_pushed++); \
1269 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1270 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1272 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1273 PUSH_FAILURE_POINTER (regend[this_reg]); \
1275 DEBUG_PRINT2 (" info: %p\n ", \
1276 reg_info[this_reg].word.pointer); \
1277 DEBUG_PRINT2 (" match_null=%d", \
1278 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1279 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1280 DEBUG_PRINT2 (" matched_something=%d", \
1281 MATCHED_SOMETHING (reg_info[this_reg])); \
1282 DEBUG_PRINT2 (" ever_matched=%d", \
1283 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1284 DEBUG_PRINT1 ("\n"); \
1285 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1288 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1289 PUSH_FAILURE_INT (lowest_active_reg); \
1291 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1292 PUSH_FAILURE_INT (highest_active_reg); \
1294 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1295 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1296 PUSH_FAILURE_POINTER (pattern_place); \
1298 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1299 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1301 DEBUG_PRINT1 ("'\n"); \
1302 PUSH_FAILURE_POINTER (string_place); \
1304 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1305 DEBUG_PUSH (failure_id); \
1308 /* This is the number of items that are pushed and popped on the stack
1309 for each register. */
1310 #define NUM_REG_ITEMS 3
1312 /* Individual items aside from the registers. */
1314 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1316 # define NUM_NONREG_ITEMS 4
1319 /* We push at most this many items on the stack. */
1320 /* We used to use (num_regs - 1), which is the number of registers
1321 this regexp will save; but that was changed to 5
1322 to avoid stack overflow for a regexp with lots of parens. */
1323 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1325 /* We actually push this many items. */
1326 #define NUM_FAILURE_ITEMS \
1328 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1332 /* How many items can still be added to the stack without overflowing it. */
1333 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1336 /* Pops what PUSH_FAIL_STACK pushes.
1338 We restore into the parameters, all of which should be lvalues:
1339 STR -- the saved data position.
1340 PAT -- the saved pattern position.
1341 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1342 REGSTART, REGEND -- arrays of string positions.
1343 REG_INFO -- array of information about each subexpression.
1345 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1346 `pend', `string1', `size1', `string2', and `size2'. */
1348 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1350 DEBUG_STATEMENT (unsigned failure_id;) \
1351 active_reg_t this_reg; \
1352 const unsigned char *string_temp; \
1354 assert (!FAIL_STACK_EMPTY ()); \
1356 /* Remove failure points and point to how many regs pushed. */ \
1357 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1358 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1359 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1361 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1363 DEBUG_POP (&failure_id); \
1364 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1366 /* If the saved string location is NULL, it came from an \
1367 on_failure_keep_string_jump opcode, and we want to throw away the \
1368 saved NULL, thus retaining our current position in the string. */ \
1369 string_temp = POP_FAILURE_POINTER (); \
1370 if (string_temp != NULL) \
1371 str = (const char *) string_temp; \
1373 DEBUG_PRINT2 (" Popping string %p: `", str); \
1374 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1375 DEBUG_PRINT1 ("'\n"); \
1377 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1378 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1379 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1381 /* Restore register info. */ \
1382 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1383 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1385 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1386 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1389 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1391 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1393 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1394 DEBUG_PRINT2 (" info: %p\n", \
1395 reg_info[this_reg].word.pointer); \
1397 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1398 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1400 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1401 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1405 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1407 reg_info[this_reg].word.integer = 0; \
1408 regend[this_reg] = 0; \
1409 regstart[this_reg] = 0; \
1411 highest_active_reg = high_reg; \
1414 set_regs_matched_done = 0; \
1415 DEBUG_STATEMENT (nfailure_points_popped++); \
1416 } /* POP_FAILURE_POINT */
1420 /* Structure for per-register (a.k.a. per-group) information.
1421 Other register information, such as the
1422 starting and ending positions (which are addresses), and the list of
1423 inner groups (which is a bits list) are maintained in separate
1426 We are making a (strictly speaking) nonportable assumption here: that
1427 the compiler will pack our bit fields into something that fits into
1428 the type of `word', i.e., is something that fits into one item on the
1432 /* Declarations and macros for re_match_2. */
1436 fail_stack_elt_t word;
1439 /* This field is one if this group can match the empty string,
1440 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1441 #define MATCH_NULL_UNSET_VALUE 3
1442 unsigned match_null_string_p : 2;
1443 unsigned is_active : 1;
1444 unsigned matched_something : 1;
1445 unsigned ever_matched_something : 1;
1447 } register_info_type;
1449 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1450 #define IS_ACTIVE(R) ((R).bits.is_active)
1451 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1452 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1455 /* Call this when have matched a real character; it sets `matched' flags
1456 for the subexpressions which we are currently inside. Also records
1457 that those subexprs have matched. */
1458 #define SET_REGS_MATCHED() \
1461 if (!set_regs_matched_done) \
1464 set_regs_matched_done = 1; \
1465 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1467 MATCHED_SOMETHING (reg_info[r]) \
1468 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1475 /* Registers are set to a sentinel when they haven't yet matched. */
1476 static char reg_unset_dummy;
1477 #define REG_UNSET_VALUE (®_unset_dummy)
1478 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1480 /* Subroutine declarations and macros for regex_compile. */
1482 static reg_errcode_t regex_compile _RE_ARGS ((const char *pattern, size_t size,
1483 reg_syntax_t syntax,
1484 struct re_pattern_buffer *bufp));
1485 static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg));
1486 static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1487 int arg1, int arg2));
1488 static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1489 int arg, unsigned char *end));
1490 static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1491 int arg1, int arg2, unsigned char *end));
1492 static boolean at_begline_loc_p _RE_ARGS ((const char *pattern, const char *p,
1493 reg_syntax_t syntax));
1494 static boolean at_endline_loc_p _RE_ARGS ((const char *p, const char *pend,
1495 reg_syntax_t syntax));
1496 static reg_errcode_t compile_range _RE_ARGS ((const char **p_ptr,
1499 reg_syntax_t syntax,
1502 /* Fetch the next character in the uncompiled pattern---translating it
1503 if necessary. Also cast from a signed character in the constant
1504 string passed to us by the user to an unsigned char that we can use
1505 as an array index (in, e.g., `translate'). */
1507 # define PATFETCH(c) \
1508 do {if (p == pend) return REG_EEND; \
1509 c = (unsigned char) *p++; \
1510 if (translate) c = (unsigned char) translate[c]; \
1514 /* Fetch the next character in the uncompiled pattern, with no
1516 #define PATFETCH_RAW(c) \
1517 do {if (p == pend) return REG_EEND; \
1518 c = (unsigned char) *p++; \
1521 /* Go backwards one character in the pattern. */
1522 #define PATUNFETCH p--
1525 /* If `translate' is non-null, return translate[D], else just D. We
1526 cast the subscript to translate because some data is declared as
1527 `char *', to avoid warnings when a string constant is passed. But
1528 when we use a character as a subscript we must make it unsigned. */
1530 # define TRANSLATE(d) \
1531 (translate ? (char) translate[(unsigned char) (d)] : (d))
1535 /* Macros for outputting the compiled pattern into `buffer'. */
1537 /* If the buffer isn't allocated when it comes in, use this. */
1538 #define INIT_BUF_SIZE 32
1540 /* Make sure we have at least N more bytes of space in buffer. */
1541 #define GET_BUFFER_SPACE(n) \
1542 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1545 /* Make sure we have one more byte of buffer space and then add C to it. */
1546 #define BUF_PUSH(c) \
1548 GET_BUFFER_SPACE (1); \
1549 *b++ = (unsigned char) (c); \
1553 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1554 #define BUF_PUSH_2(c1, c2) \
1556 GET_BUFFER_SPACE (2); \
1557 *b++ = (unsigned char) (c1); \
1558 *b++ = (unsigned char) (c2); \
1562 /* As with BUF_PUSH_2, except for three bytes. */
1563 #define BUF_PUSH_3(c1, c2, c3) \
1565 GET_BUFFER_SPACE (3); \
1566 *b++ = (unsigned char) (c1); \
1567 *b++ = (unsigned char) (c2); \
1568 *b++ = (unsigned char) (c3); \
1572 /* Store a jump with opcode OP at LOC to location TO. We store a
1573 relative address offset by the three bytes the jump itself occupies. */
1574 #define STORE_JUMP(op, loc, to) \
1575 store_op1 (op, loc, (int) ((to) - (loc) - 3))
1577 /* Likewise, for a two-argument jump. */
1578 #define STORE_JUMP2(op, loc, to, arg) \
1579 store_op2 (op, loc, (int) ((to) - (loc) - 3), arg)
1581 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1582 #define INSERT_JUMP(op, loc, to) \
1583 insert_op1 (op, loc, (int) ((to) - (loc) - 3), b)
1585 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1586 #define INSERT_JUMP2(op, loc, to, arg) \
1587 insert_op2 (op, loc, (int) ((to) - (loc) - 3), arg, b)
1590 /* This is not an arbitrary limit: the arguments which represent offsets
1591 into the pattern are two bytes long. So if 2^16 bytes turns out to
1592 be too small, many things would have to change. */
1593 /* Any other compiler which, like MSC, has allocation limit below 2^16
1594 bytes will have to use approach similar to what was done below for
1595 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1596 reallocating to 0 bytes. Such thing is not going to work too well.
1597 You have been warned!! */
1598 #if defined _MSC_VER && !defined WIN32
1599 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
1600 The REALLOC define eliminates a flurry of conversion warnings,
1601 but is not required. */
1602 # define MAX_BUF_SIZE 65500L
1603 # define REALLOC(p,s) realloc ((p), (size_t) (s))
1605 # define MAX_BUF_SIZE (1L << 16)
1606 # define REALLOC(p,s) realloc ((p), (s))
1609 /* Extend the buffer by twice its current size via realloc and
1610 reset the pointers that pointed into the old block to point to the
1611 correct places in the new one. If extending the buffer results in it
1612 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1613 #define EXTEND_BUFFER() \
1615 unsigned char *old_buffer = bufp->buffer; \
1616 if (bufp->allocated == MAX_BUF_SIZE) \
1618 bufp->allocated <<= 1; \
1619 if (bufp->allocated > MAX_BUF_SIZE) \
1620 bufp->allocated = MAX_BUF_SIZE; \
1621 bufp->buffer = (unsigned char *) REALLOC (bufp->buffer, bufp->allocated);\
1622 if (bufp->buffer == NULL) \
1623 return REG_ESPACE; \
1624 /* If the buffer moved, move all the pointers into it. */ \
1625 if (old_buffer != bufp->buffer) \
1627 b = (b - old_buffer) + bufp->buffer; \
1628 begalt = (begalt - old_buffer) + bufp->buffer; \
1629 if (fixup_alt_jump) \
1630 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1632 laststart = (laststart - old_buffer) + bufp->buffer; \
1633 if (pending_exact) \
1634 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1639 /* Since we have one byte reserved for the register number argument to
1640 {start,stop}_memory, the maximum number of groups we can report
1641 things about is what fits in that byte. */
1642 #define MAX_REGNUM 255
1644 /* But patterns can have more than `MAX_REGNUM' registers. We just
1645 ignore the excess. */
1646 typedef unsigned regnum_t;
1649 /* Macros for the compile stack. */
1651 /* Since offsets can go either forwards or backwards, this type needs to
1652 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1653 /* int may be not enough when sizeof(int) == 2. */
1654 typedef long pattern_offset_t;
1658 pattern_offset_t begalt_offset;
1659 pattern_offset_t fixup_alt_jump;
1660 pattern_offset_t inner_group_offset;
1661 pattern_offset_t laststart_offset;
1663 } compile_stack_elt_t;
1668 compile_stack_elt_t *stack;
1670 unsigned avail; /* Offset of next open position. */
1671 } compile_stack_type;
1674 #define INIT_COMPILE_STACK_SIZE 32
1676 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1677 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1679 /* The next available element. */
1680 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1683 /* Set the bit for character C in a list. */
1684 #define SET_LIST_BIT(c) \
1685 (b[((unsigned char) (c)) / BYTEWIDTH] \
1686 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1689 /* Get the next unsigned number in the uncompiled pattern. */
1690 #define GET_UNSIGNED_NUMBER(num) \
1694 while (ISDIGIT (c)) \
1698 num = num * 10 + c - '0'; \
1706 /* Use this only if they have btowc(), since wctype() is used below
1707 together with btowc(). btowc() is defined in the 1994 Amendment 1
1708 to ISO C and may not be present on systems where we have wchar.h
1710 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H && defined HAVE_BTOWC)
1711 /* The GNU C library provides support for user-defined character classes
1712 and the functions from ISO C amendement 1. */
1713 # ifdef CHARCLASS_NAME_MAX
1714 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
1716 /* This shouldn't happen but some implementation might still have this
1717 problem. Use a reasonable default value. */
1718 # define CHAR_CLASS_MAX_LENGTH 256
1722 # define IS_CHAR_CLASS(string) __wctype (string)
1724 # define IS_CHAR_CLASS(string) wctype (string)
1727 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1729 # define IS_CHAR_CLASS(string) \
1730 (STREQ (string, "alpha") || STREQ (string, "upper") \
1731 || STREQ (string, "lower") || STREQ (string, "digit") \
1732 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1733 || STREQ (string, "space") || STREQ (string, "print") \
1734 || STREQ (string, "punct") || STREQ (string, "graph") \
1735 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1738 #ifndef MATCH_MAY_ALLOCATE
1740 /* If we cannot allocate large objects within re_match_2_internal,
1741 we make the fail stack and register vectors global.
1742 The fail stack, we grow to the maximum size when a regexp
1744 The register vectors, we adjust in size each time we
1745 compile a regexp, according to the number of registers it needs. */
1747 static fail_stack_type fail_stack;
1749 /* Size with which the following vectors are currently allocated.
1750 That is so we can make them bigger as needed,
1751 but never make them smaller. */
1752 static int regs_allocated_size;
1754 static const char ** regstart, ** regend;
1755 static const char ** old_regstart, ** old_regend;
1756 static const char **best_regstart, **best_regend;
1757 static register_info_type *reg_info;
1758 static const char **reg_dummy;
1759 static register_info_type *reg_info_dummy;
1761 /* Make the register vectors big enough for NUM_REGS registers,
1762 but don't make them smaller. */
1765 regex_grow_registers (num_regs)
1768 if (num_regs > regs_allocated_size)
1770 RETALLOC_IF (regstart, num_regs, const char *);
1771 RETALLOC_IF (regend, num_regs, const char *);
1772 RETALLOC_IF (old_regstart, num_regs, const char *);
1773 RETALLOC_IF (old_regend, num_regs, const char *);
1774 RETALLOC_IF (best_regstart, num_regs, const char *);
1775 RETALLOC_IF (best_regend, num_regs, const char *);
1776 RETALLOC_IF (reg_info, num_regs, register_info_type);
1777 RETALLOC_IF (reg_dummy, num_regs, const char *);
1778 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1780 regs_allocated_size = num_regs;
1784 #endif /* not MATCH_MAY_ALLOCATE */
1786 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
1790 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1791 Returns one of error codes defined in `gnu-regex.h', or zero for success.
1793 Assumes the `allocated' (and perhaps `buffer') and `translate'
1794 fields are set in BUFP on entry.
1796 If it succeeds, results are put in BUFP (if it returns an error, the
1797 contents of BUFP are undefined):
1798 `buffer' is the compiled pattern;
1799 `syntax' is set to SYNTAX;
1800 `used' is set to the length of the compiled pattern;
1801 `fastmap_accurate' is zero;
1802 `re_nsub' is the number of subexpressions in PATTERN;
1803 `not_bol' and `not_eol' are zero;
1805 The `fastmap' and `newline_anchor' fields are neither
1806 examined nor set. */
1808 /* Return, freeing storage we allocated. */
1809 #define FREE_STACK_RETURN(value) \
1810 return (free (compile_stack.stack), value)
1812 static reg_errcode_t
1813 regex_compile (pattern, size, syntax, bufp)
1814 const char *pattern;
1816 reg_syntax_t syntax;
1817 struct re_pattern_buffer *bufp;
1819 /* We fetch characters from PATTERN here. Even though PATTERN is
1820 `char *' (i.e., signed), we declare these variables as unsigned, so
1821 they can be reliably used as array indices. */
1822 register unsigned char c, c1;
1824 /* A random temporary spot in PATTERN. */
1827 /* Points to the end of the buffer, where we should append. */
1828 register unsigned char *b;
1830 /* Keeps track of unclosed groups. */
1831 compile_stack_type compile_stack;
1833 /* Points to the current (ending) position in the pattern. */
1834 const char *p = pattern;
1835 const char *pend = pattern + size;
1837 /* How to translate the characters in the pattern. */
1838 RE_TRANSLATE_TYPE translate = bufp->translate;
1840 /* Address of the count-byte of the most recently inserted `exactn'
1841 command. This makes it possible to tell if a new exact-match
1842 character can be added to that command or if the character requires
1843 a new `exactn' command. */
1844 unsigned char *pending_exact = 0;
1846 /* Address of start of the most recently finished expression.
1847 This tells, e.g., postfix * where to find the start of its
1848 operand. Reset at the beginning of groups and alternatives. */
1849 unsigned char *laststart = 0;
1851 /* Address of beginning of regexp, or inside of last group. */
1852 unsigned char *begalt;
1854 /* Place in the uncompiled pattern (i.e., the {) to
1855 which to go back if the interval is invalid. */
1856 const char *beg_interval;
1858 /* Address of the place where a forward jump should go to the end of
1859 the containing expression. Each alternative of an `or' -- except the
1860 last -- ends with a forward jump of this sort. */
1861 unsigned char *fixup_alt_jump = 0;
1863 /* Counts open-groups as they are encountered. Remembered for the
1864 matching close-group on the compile stack, so the same register
1865 number is put in the stop_memory as the start_memory. */
1866 regnum_t regnum = 0;
1869 DEBUG_PRINT1 ("\nCompiling pattern: ");
1872 unsigned debug_count;
1874 for (debug_count = 0; debug_count < size; debug_count++)
1875 putchar (pattern[debug_count]);
1880 /* Initialize the compile stack. */
1881 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1882 if (compile_stack.stack == NULL)
1885 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1886 compile_stack.avail = 0;
1888 /* Initialize the pattern buffer. */
1889 bufp->syntax = syntax;
1890 bufp->fastmap_accurate = 0;
1891 bufp->not_bol = bufp->not_eol = 0;
1893 /* Set `used' to zero, so that if we return an error, the pattern
1894 printer (for debugging) will think there's no pattern. We reset it
1898 /* Always count groups, whether or not bufp->no_sub is set. */
1901 #if !defined emacs && !defined SYNTAX_TABLE
1902 /* Initialize the syntax table. */
1903 init_syntax_once ();
1906 if (bufp->allocated == 0)
1909 { /* If zero allocated, but buffer is non-null, try to realloc
1910 enough space. This loses if buffer's address is bogus, but
1911 that is the user's responsibility. */
1912 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1915 { /* Caller did not allocate a buffer. Do it for them. */
1916 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1918 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1920 bufp->allocated = INIT_BUF_SIZE;
1923 begalt = b = bufp->buffer;
1925 /* Loop through the uncompiled pattern until we're at the end. */
1934 if ( /* If at start of pattern, it's an operator. */
1936 /* If context independent, it's an operator. */
1937 || syntax & RE_CONTEXT_INDEP_ANCHORS
1938 /* Otherwise, depends on what's come before. */
1939 || at_begline_loc_p (pattern, p, syntax))
1949 if ( /* If at end of pattern, it's an operator. */
1951 /* If context independent, it's an operator. */
1952 || syntax & RE_CONTEXT_INDEP_ANCHORS
1953 /* Otherwise, depends on what's next. */
1954 || at_endline_loc_p (p, pend, syntax))
1964 if ((syntax & RE_BK_PLUS_QM)
1965 || (syntax & RE_LIMITED_OPS))
1969 /* If there is no previous pattern... */
1972 if (syntax & RE_CONTEXT_INVALID_OPS)
1973 FREE_STACK_RETURN (REG_BADRPT);
1974 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1979 /* Are we optimizing this jump? */
1980 boolean keep_string_p = false;
1982 /* 1 means zero (many) matches is allowed. */
1983 char zero_times_ok = 0, many_times_ok = 0;
1985 /* If there is a sequence of repetition chars, collapse it
1986 down to just one (the right one). We can't combine
1987 interval operators with these because of, e.g., `a{2}*',
1988 which should only match an even number of `a's. */
1992 zero_times_ok |= c != '+';
1993 many_times_ok |= c != '?';
2001 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
2004 else if (syntax & RE_BK_PLUS_QM && c == '\\')
2006 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2009 if (!(c1 == '+' || c1 == '?'))
2024 /* If we get here, we found another repeat character. */
2027 /* Star, etc. applied to an empty pattern is equivalent
2028 to an empty pattern. */
2032 /* Now we know whether or not zero matches is allowed
2033 and also whether or not two or more matches is allowed. */
2035 { /* More than one repetition is allowed, so put in at the
2036 end a backward relative jump from `b' to before the next
2037 jump we're going to put in below (which jumps from
2038 laststart to after this jump).
2040 But if we are at the `*' in the exact sequence `.*\n',
2041 insert an unconditional jump backwards to the .,
2042 instead of the beginning of the loop. This way we only
2043 push a failure point once, instead of every time
2044 through the loop. */
2045 assert (p - 1 > pattern);
2047 /* Allocate the space for the jump. */
2048 GET_BUFFER_SPACE (3);
2050 /* We know we are not at the first character of the pattern,
2051 because laststart was nonzero. And we've already
2052 incremented `p', by the way, to be the character after
2053 the `*'. Do we have to do something analogous here
2054 for null bytes, because of RE_DOT_NOT_NULL? */
2055 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2057 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2058 && !(syntax & RE_DOT_NEWLINE))
2059 { /* We have .*\n. */
2060 STORE_JUMP (jump, b, laststart);
2061 keep_string_p = true;
2064 /* Anything else. */
2065 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
2067 /* We've added more stuff to the buffer. */
2071 /* On failure, jump from laststart to b + 3, which will be the
2072 end of the buffer after this jump is inserted. */
2073 GET_BUFFER_SPACE (3);
2074 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2082 /* At least one repetition is required, so insert a
2083 `dummy_failure_jump' before the initial
2084 `on_failure_jump' instruction of the loop. This
2085 effects a skip over that instruction the first time
2086 we hit that loop. */
2087 GET_BUFFER_SPACE (3);
2088 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
2103 boolean had_char_class = false;
2105 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2107 /* Ensure that we have enough space to push a charset: the
2108 opcode, the length count, and the bitset; 34 bytes in all. */
2109 GET_BUFFER_SPACE (34);
2113 /* We test `*p == '^' twice, instead of using an if
2114 statement, so we only need one BUF_PUSH. */
2115 BUF_PUSH (*p == '^' ? charset_not : charset);
2119 /* Remember the first position in the bracket expression. */
2122 /* Push the number of bytes in the bitmap. */
2123 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
2125 /* Clear the whole map. */
2126 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
2128 /* charset_not matches newline according to a syntax bit. */
2129 if ((re_opcode_t) b[-2] == charset_not
2130 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2131 SET_LIST_BIT ('\n');
2133 /* Read in characters and ranges, setting map bits. */
2136 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2140 /* \ might escape characters inside [...] and [^...]. */
2141 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2143 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2150 /* Could be the end of the bracket expression. If it's
2151 not (i.e., when the bracket expression is `[]' so
2152 far), the ']' character bit gets set way below. */
2153 if (c == ']' && p != p1 + 1)
2156 /* Look ahead to see if it's a range when the last thing
2157 was a character class. */
2158 if (had_char_class && c == '-' && *p != ']')
2159 FREE_STACK_RETURN (REG_ERANGE);
2161 /* Look ahead to see if it's a range when the last thing
2162 was a character: if this is a hyphen not at the
2163 beginning or the end of a list, then it's the range
2166 && !(p - 2 >= pattern && p[-2] == '[')
2167 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2171 = compile_range (&p, pend, translate, syntax, b);
2172 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2175 else if (p[0] == '-' && p[1] != ']')
2176 { /* This handles ranges made up of characters only. */
2179 /* Move past the `-'. */
2182 ret = compile_range (&p, pend, translate, syntax, b);
2183 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2186 /* See if we're at the beginning of a possible character
2189 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2190 { /* Leave room for the null. */
2191 char str[CHAR_CLASS_MAX_LENGTH + 1];
2196 /* If pattern is `[[:'. */
2197 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2202 if ((c == ':' && *p == ']') || p == pend
2203 || c1 == CHAR_CLASS_MAX_LENGTH)
2209 /* If isn't a word bracketed by `[:' and `:]':
2210 undo the ending character, the letters, and leave
2211 the leading `:' and `[' (but set bits for them). */
2212 if (c == ':' && *p == ']')
2214 /* CYGNUS LOCAL: Skip this code if we don't have btowc(). btowc() is */
2215 /* defined in the 1994 Amendment 1 to ISO C and may not be present on */
2216 /* systems where we have wchar.h and wctype.h. */
2217 #if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H && defined HAVE_BTOWC)
2218 boolean is_lower = STREQ (str, "lower");
2219 boolean is_upper = STREQ (str, "upper");
2223 wt = IS_CHAR_CLASS (str);
2225 FREE_STACK_RETURN (REG_ECTYPE);
2227 /* Throw away the ] at the end of the character
2231 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2233 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
2236 if (__iswctype (__btowc (ch), wt))
2239 if (iswctype (btowc (ch), wt))
2243 if (translate && (is_upper || is_lower)
2244 && (ISUPPER (ch) || ISLOWER (ch)))
2248 had_char_class = true;
2251 boolean is_alnum = STREQ (str, "alnum");
2252 boolean is_alpha = STREQ (str, "alpha");
2253 boolean is_blank = STREQ (str, "blank");
2254 boolean is_cntrl = STREQ (str, "cntrl");
2255 boolean is_digit = STREQ (str, "digit");
2256 boolean is_graph = STREQ (str, "graph");
2257 boolean is_lower = STREQ (str, "lower");
2258 boolean is_print = STREQ (str, "print");
2259 boolean is_punct = STREQ (str, "punct");
2260 boolean is_space = STREQ (str, "space");
2261 boolean is_upper = STREQ (str, "upper");
2262 boolean is_xdigit = STREQ (str, "xdigit");
2264 if (!IS_CHAR_CLASS (str))
2265 FREE_STACK_RETURN (REG_ECTYPE);
2267 /* Throw away the ] at the end of the character
2271 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2273 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2275 /* This was split into 3 if's to
2276 avoid an arbitrary limit in some compiler. */
2277 if ( (is_alnum && ISALNUM (ch))
2278 || (is_alpha && ISALPHA (ch))
2279 || (is_blank && ISBLANK (ch))
2280 || (is_cntrl && ISCNTRL (ch)))
2282 if ( (is_digit && ISDIGIT (ch))
2283 || (is_graph && ISGRAPH (ch))
2284 || (is_lower && ISLOWER (ch))
2285 || (is_print && ISPRINT (ch)))
2287 if ( (is_punct && ISPUNCT (ch))
2288 || (is_space && ISSPACE (ch))
2289 || (is_upper && ISUPPER (ch))
2290 || (is_xdigit && ISXDIGIT (ch)))
2292 if ( translate && (is_upper || is_lower)
2293 && (ISUPPER (ch) || ISLOWER (ch)))
2296 had_char_class = true;
2297 #endif /* libc || wctype.h */
2306 had_char_class = false;
2311 had_char_class = false;
2316 /* Discard any (non)matching list bytes that are all 0 at the
2317 end of the map. Decrease the map-length byte too. */
2318 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2326 if (syntax & RE_NO_BK_PARENS)
2333 if (syntax & RE_NO_BK_PARENS)
2340 if (syntax & RE_NEWLINE_ALT)
2347 if (syntax & RE_NO_BK_VBAR)
2354 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2355 goto handle_interval;
2361 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2363 /* Do not translate the character after the \, so that we can
2364 distinguish, e.g., \B from \b, even if we normally would
2365 translate, e.g., B to b. */
2371 if (syntax & RE_NO_BK_PARENS)
2372 goto normal_backslash;
2378 if (COMPILE_STACK_FULL)
2380 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2381 compile_stack_elt_t);
2382 if (compile_stack.stack == NULL) return REG_ESPACE;
2384 compile_stack.size <<= 1;
2387 /* These are the values to restore when we hit end of this
2388 group. They are all relative offsets, so that if the
2389 whole pattern moves because of realloc, they will still
2391 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2392 COMPILE_STACK_TOP.fixup_alt_jump
2393 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2394 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2395 COMPILE_STACK_TOP.regnum = regnum;
2397 /* We will eventually replace the 0 with the number of
2398 groups inner to this one. But do not push a
2399 start_memory for groups beyond the last one we can
2400 represent in the compiled pattern. */
2401 if (regnum <= MAX_REGNUM)
2403 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2404 BUF_PUSH_3 (start_memory, regnum, 0);
2407 compile_stack.avail++;
2412 /* If we've reached MAX_REGNUM groups, then this open
2413 won't actually generate any code, so we'll have to
2414 clear pending_exact explicitly. */
2420 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2422 if (COMPILE_STACK_EMPTY)
2424 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2425 goto normal_backslash;
2427 FREE_STACK_RETURN (REG_ERPAREN);
2432 { /* Push a dummy failure point at the end of the
2433 alternative for a possible future
2434 `pop_failure_jump' to pop. See comments at
2435 `push_dummy_failure' in `re_match_2'. */
2436 BUF_PUSH (push_dummy_failure);
2438 /* We allocated space for this jump when we assigned
2439 to `fixup_alt_jump', in the `handle_alt' case below. */
2440 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2443 /* See similar code for backslashed left paren above. */
2444 if (COMPILE_STACK_EMPTY)
2446 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2449 FREE_STACK_RETURN (REG_ERPAREN);
2452 /* Since we just checked for an empty stack above, this
2453 ``can't happen''. */
2454 assert (compile_stack.avail != 0);
2456 /* We don't just want to restore into `regnum', because
2457 later groups should continue to be numbered higher,
2458 as in `(ab)c(de)' -- the second group is #2. */
2459 regnum_t this_group_regnum;
2461 compile_stack.avail--;
2462 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2464 = COMPILE_STACK_TOP.fixup_alt_jump
2465 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2467 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2468 this_group_regnum = COMPILE_STACK_TOP.regnum;
2469 /* If we've reached MAX_REGNUM groups, then this open
2470 won't actually generate any code, so we'll have to
2471 clear pending_exact explicitly. */
2474 /* We're at the end of the group, so now we know how many
2475 groups were inside this one. */
2476 if (this_group_regnum <= MAX_REGNUM)
2478 unsigned char *inner_group_loc
2479 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2481 *inner_group_loc = regnum - this_group_regnum;
2482 BUF_PUSH_3 (stop_memory, this_group_regnum,
2483 regnum - this_group_regnum);
2489 case '|': /* `\|'. */
2490 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2491 goto normal_backslash;
2493 if (syntax & RE_LIMITED_OPS)
2496 /* Insert before the previous alternative a jump which
2497 jumps to this alternative if the former fails. */
2498 GET_BUFFER_SPACE (3);
2499 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2503 /* The alternative before this one has a jump after it
2504 which gets executed if it gets matched. Adjust that
2505 jump so it will jump to this alternative's analogous
2506 jump (put in below, which in turn will jump to the next
2507 (if any) alternative's such jump, etc.). The last such
2508 jump jumps to the correct final destination. A picture:
2514 If we are at `b', then fixup_alt_jump right now points to a
2515 three-byte space after `a'. We'll put in the jump, set
2516 fixup_alt_jump to right after `b', and leave behind three
2517 bytes which we'll fill in when we get to after `c'. */
2520 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2522 /* Mark and leave space for a jump after this alternative,
2523 to be filled in later either by next alternative or
2524 when know we're at the end of a series of alternatives. */
2526 GET_BUFFER_SPACE (3);
2535 /* If \{ is a literal. */
2536 if (!(syntax & RE_INTERVALS)
2537 /* If we're at `\{' and it's not the open-interval
2539 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2540 || (p - 2 == pattern && p == pend))
2541 goto normal_backslash;
2545 /* If got here, then the syntax allows intervals. */
2547 /* At least (most) this many matches must be made. */
2548 int lower_bound = -1, upper_bound = -1;
2550 beg_interval = p - 1;
2554 if (syntax & RE_NO_BK_BRACES)
2555 goto unfetch_interval;
2557 FREE_STACK_RETURN (REG_EBRACE);
2560 GET_UNSIGNED_NUMBER (lower_bound);
2564 GET_UNSIGNED_NUMBER (upper_bound);
2565 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2568 /* Interval such as `{1}' => match exactly once. */
2569 upper_bound = lower_bound;
2571 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2572 || lower_bound > upper_bound)
2574 if (syntax & RE_NO_BK_BRACES)
2575 goto unfetch_interval;
2577 FREE_STACK_RETURN (REG_BADBR);
2580 if (!(syntax & RE_NO_BK_BRACES))
2582 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2589 if (syntax & RE_NO_BK_BRACES)
2590 goto unfetch_interval;
2592 FREE_STACK_RETURN (REG_BADBR);
2595 /* We just parsed a valid interval. */
2597 /* If it's invalid to have no preceding re. */
2600 if (syntax & RE_CONTEXT_INVALID_OPS)
2601 FREE_STACK_RETURN (REG_BADRPT);
2602 else if (syntax & RE_CONTEXT_INDEP_OPS)
2605 goto unfetch_interval;
2608 /* If the upper bound is zero, don't want to succeed at
2609 all; jump from `laststart' to `b + 3', which will be
2610 the end of the buffer after we insert the jump. */
2611 if (upper_bound == 0)
2613 GET_BUFFER_SPACE (3);
2614 INSERT_JUMP (jump, laststart, b + 3);
2618 /* Otherwise, we have a nontrivial interval. When
2619 we're all done, the pattern will look like:
2620 set_number_at <jump count> <upper bound>
2621 set_number_at <succeed_n count> <lower bound>
2622 succeed_n <after jump addr> <succeed_n count>
2624 jump_n <succeed_n addr> <jump count>
2625 (The upper bound and `jump_n' are omitted if
2626 `upper_bound' is 1, though.) */
2628 { /* If the upper bound is > 1, we need to insert
2629 more at the end of the loop. */
2630 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2632 GET_BUFFER_SPACE (nbytes);
2634 /* Initialize lower bound of the `succeed_n', even
2635 though it will be set during matching by its
2636 attendant `set_number_at' (inserted next),
2637 because `re_compile_fastmap' needs to know.
2638 Jump to the `jump_n' we might insert below. */
2639 INSERT_JUMP2 (succeed_n, laststart,
2640 b + 5 + (upper_bound > 1) * 5,
2644 /* Code to initialize the lower bound. Insert
2645 before the `succeed_n'. The `5' is the last two
2646 bytes of this `set_number_at', plus 3 bytes of
2647 the following `succeed_n'. */
2648 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2651 if (upper_bound > 1)
2652 { /* More than one repetition is allowed, so
2653 append a backward jump to the `succeed_n'
2654 that starts this interval.
2656 When we've reached this during matching,
2657 we'll have matched the interval once, so
2658 jump back only `upper_bound - 1' times. */
2659 STORE_JUMP2 (jump_n, b, laststart + 5,
2663 /* The location we want to set is the second
2664 parameter of the `jump_n'; that is `b-2' as
2665 an absolute address. `laststart' will be
2666 the `set_number_at' we're about to insert;
2667 `laststart+3' the number to set, the source
2668 for the relative address. But we are
2669 inserting into the middle of the pattern --
2670 so everything is getting moved up by 5.
2671 Conclusion: (b - 2) - (laststart + 3) + 5,
2672 i.e., b - laststart.
2674 We insert this at the beginning of the loop
2675 so that if we fail during matching, we'll
2676 reinitialize the bounds. */
2677 insert_op2 (set_number_at, laststart, b - laststart,
2678 upper_bound - 1, b);
2683 beg_interval = NULL;
2688 /* If an invalid interval, match the characters as literals. */
2689 assert (beg_interval);
2691 beg_interval = NULL;
2693 /* normal_char and normal_backslash need `c'. */
2696 if (!(syntax & RE_NO_BK_BRACES))
2698 if (p > pattern && p[-1] == '\\')
2699 goto normal_backslash;
2704 /* There is no way to specify the before_dot and after_dot
2705 operators. rms says this is ok. --karl */
2713 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2719 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2725 if (syntax & RE_NO_GNU_OPS)
2728 BUF_PUSH (wordchar);
2733 if (syntax & RE_NO_GNU_OPS)
2736 BUF_PUSH (notwordchar);
2741 if (syntax & RE_NO_GNU_OPS)
2747 if (syntax & RE_NO_GNU_OPS)
2753 if (syntax & RE_NO_GNU_OPS)
2755 BUF_PUSH (wordbound);
2759 if (syntax & RE_NO_GNU_OPS)
2761 BUF_PUSH (notwordbound);
2765 if (syntax & RE_NO_GNU_OPS)
2771 if (syntax & RE_NO_GNU_OPS)
2776 case '1': case '2': case '3': case '4': case '5':
2777 case '6': case '7': case '8': case '9':
2778 if (syntax & RE_NO_BK_REFS)
2784 FREE_STACK_RETURN (REG_ESUBREG);
2786 /* Can't back reference to a subexpression if inside of it. */
2787 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
2791 BUF_PUSH_2 (duplicate, c1);
2797 if (syntax & RE_BK_PLUS_QM)
2800 goto normal_backslash;
2804 /* You might think it would be useful for \ to mean
2805 not to translate; but if we don't translate it
2806 it will never match anything. */
2814 /* Expects the character in `c'. */
2816 /* If no exactn currently being built. */
2819 /* If last exactn not at current position. */
2820 || pending_exact + *pending_exact + 1 != b
2822 /* We have only one byte following the exactn for the count. */
2823 || *pending_exact == (1 << BYTEWIDTH) - 1
2825 /* If followed by a repetition operator. */
2826 || *p == '*' || *p == '^'
2827 || ((syntax & RE_BK_PLUS_QM)
2828 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2829 : (*p == '+' || *p == '?'))
2830 || ((syntax & RE_INTERVALS)
2831 && ((syntax & RE_NO_BK_BRACES)
2833 : (p[0] == '\\' && p[1] == '{'))))
2835 /* Start building a new exactn. */
2839 BUF_PUSH_2 (exactn, 0);
2840 pending_exact = b - 1;
2847 } /* while p != pend */
2850 /* Through the pattern now. */
2853 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2855 if (!COMPILE_STACK_EMPTY)
2856 FREE_STACK_RETURN (REG_EPAREN);
2858 /* If we don't want backtracking, force success
2859 the first time we reach the end of the compiled pattern. */
2860 if (syntax & RE_NO_POSIX_BACKTRACKING)
2863 free (compile_stack.stack);
2865 /* We have succeeded; set the length of the buffer. */
2866 bufp->used = b - bufp->buffer;
2871 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2872 print_compiled_pattern (bufp);
2876 #ifndef MATCH_MAY_ALLOCATE
2877 /* Initialize the failure stack to the largest possible stack. This
2878 isn't necessary unless we're trying to avoid calling alloca in
2879 the search and match routines. */
2881 int num_regs = bufp->re_nsub + 1;
2883 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2884 is strictly greater than re_max_failures, the largest possible stack
2885 is 2 * re_max_failures failure points. */
2886 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2888 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2891 if (! fail_stack.stack)
2893 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2894 * sizeof (fail_stack_elt_t));
2897 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2899 * sizeof (fail_stack_elt_t)));
2900 # else /* not emacs */
2901 if (! fail_stack.stack)
2903 = (fail_stack_elt_t *) malloc (fail_stack.size
2904 * sizeof (fail_stack_elt_t));
2907 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2909 * sizeof (fail_stack_elt_t)));
2910 # endif /* not emacs */
2913 regex_grow_registers (num_regs);
2915 #endif /* not MATCH_MAY_ALLOCATE */
2918 } /* regex_compile */
2920 /* Subroutines for `regex_compile'. */
2922 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2925 store_op1 (op, loc, arg)
2930 *loc = (unsigned char) op;
2931 STORE_NUMBER (loc + 1, arg);
2935 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2938 store_op2 (op, loc, arg1, arg2)
2943 *loc = (unsigned char) op;
2944 STORE_NUMBER (loc + 1, arg1);
2945 STORE_NUMBER (loc + 3, arg2);
2949 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
2950 for OP followed by two-byte integer parameter ARG. */
2953 insert_op1 (op, loc, arg, end)
2959 register unsigned char *pfrom = end;
2960 register unsigned char *pto = end + 3;
2962 while (pfrom != loc)
2965 store_op1 (op, loc, arg);
2969 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2972 insert_op2 (op, loc, arg1, arg2, end)
2978 register unsigned char *pfrom = end;
2979 register unsigned char *pto = end + 5;
2981 while (pfrom != loc)
2984 store_op2 (op, loc, arg1, arg2);
2988 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
2989 after an alternative or a begin-subexpression. We assume there is at
2990 least one character before the ^. */
2993 at_begline_loc_p (pattern, p, syntax)
2994 const char *pattern, *p;
2995 reg_syntax_t syntax;
2997 const char *prev = p - 2;
2998 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
3001 /* After a subexpression? */
3002 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
3003 /* After an alternative? */
3004 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
3008 /* The dual of at_begline_loc_p. This one is for $. We assume there is
3009 at least one character after the $, i.e., `P < PEND'. */
3012 at_endline_loc_p (p, pend, syntax)
3013 const char *p, *pend;
3014 reg_syntax_t syntax;
3016 const char *next = p;
3017 boolean next_backslash = *next == '\\';
3018 const char *next_next = p + 1 < pend ? p + 1 : 0;
3021 /* Before a subexpression? */
3022 (syntax & RE_NO_BK_PARENS ? *next == ')'
3023 : next_backslash && next_next && *next_next == ')')
3024 /* Before an alternative? */
3025 || (syntax & RE_NO_BK_VBAR ? *next == '|'
3026 : next_backslash && next_next && *next_next == '|');
3030 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
3031 false if it's not. */
3034 group_in_compile_stack (compile_stack, regnum)
3035 compile_stack_type compile_stack;
3040 for (this_element = compile_stack.avail - 1;
3043 if (compile_stack.stack[this_element].regnum == regnum)
3050 /* Read the ending character of a range (in a bracket expression) from the
3051 uncompiled pattern *P_PTR (which ends at PEND). We assume the
3052 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
3053 Then we set the translation of all bits between the starting and
3054 ending characters (inclusive) in the compiled pattern B.
3056 Return an error code.
3058 We use these short variable names so we can use the same macros as
3059 `regex_compile' itself. */
3061 static reg_errcode_t
3062 compile_range (p_ptr, pend, translate, syntax, b)
3063 const char **p_ptr, *pend;
3064 RE_TRANSLATE_TYPE translate;
3065 reg_syntax_t syntax;
3070 const char *p = *p_ptr;
3071 unsigned int range_start, range_end;
3076 /* Even though the pattern is a signed `char *', we need to fetch
3077 with unsigned char *'s; if the high bit of the pattern character
3078 is set, the range endpoints will be negative if we fetch using a
3081 We also want to fetch the endpoints without translating them; the
3082 appropriate translation is done in the bit-setting loop below. */
3083 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
3084 range_start = ((const unsigned char *) p)[-2];
3085 range_end = ((const unsigned char *) p)[0];
3087 /* Have to increment the pointer into the pattern string, so the
3088 caller isn't still at the ending character. */
3091 /* If the start is after the end, the range is empty. */
3092 if (range_start > range_end)
3093 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
3095 /* Here we see why `this_char' has to be larger than an `unsigned
3096 char' -- the range is inclusive, so if `range_end' == 0xff
3097 (assuming 8-bit characters), we would otherwise go into an infinite
3098 loop, since all characters <= 0xff. */
3099 for (this_char = range_start; this_char <= range_end; this_char++)
3101 SET_LIST_BIT (TRANSLATE (this_char));
3107 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3108 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3109 characters can start a string that matches the pattern. This fastmap
3110 is used by re_search to skip quickly over impossible starting points.
3112 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3113 area as BUFP->fastmap.
3115 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3118 Returns 0 if we succeed, -2 if an internal error. */
3121 re_compile_fastmap (bufp)
3122 struct re_pattern_buffer *bufp;
3125 #ifdef MATCH_MAY_ALLOCATE
3126 fail_stack_type fail_stack;
3128 #ifndef REGEX_MALLOC
3132 register char *fastmap = bufp->fastmap;
3133 unsigned char *pattern = bufp->buffer;
3134 unsigned char *p = pattern;
3135 register unsigned char *pend = pattern + bufp->used;
3138 /* This holds the pointer to the failure stack, when
3139 it is allocated relocatably. */
3140 fail_stack_elt_t *failure_stack_ptr;
3143 /* Assume that each path through the pattern can be null until
3144 proven otherwise. We set this false at the bottom of switch
3145 statement, to which we get only if a particular path doesn't
3146 match the empty string. */
3147 boolean path_can_be_null = true;
3149 /* We aren't doing a `succeed_n' to begin with. */
3150 boolean succeed_n_p = false;
3152 assert (fastmap != NULL && p != NULL);
3155 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
3156 bufp->fastmap_accurate = 1; /* It will be when we're done. */
3157 bufp->can_be_null = 0;
3161 if (p == pend || *p == succeed)
3163 /* We have reached the (effective) end of pattern. */
3164 if (!FAIL_STACK_EMPTY ())
3166 bufp->can_be_null |= path_can_be_null;
3168 /* Reset for next path. */
3169 path_can_be_null = true;
3171 p = fail_stack.stack[--fail_stack.avail].pointer;
3179 /* We should never be about to go beyond the end of the pattern. */
3182 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3185 /* I guess the idea here is to simply not bother with a fastmap
3186 if a backreference is used, since it's too hard to figure out
3187 the fastmap for the corresponding group. Setting
3188 `can_be_null' stops `re_search_2' from using the fastmap, so
3189 that is all we do. */
3191 bufp->can_be_null = 1;
3195 /* Following are the cases which match a character. These end
3204 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3205 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3211 /* Chars beyond end of map must be allowed. */
3212 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3215 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3216 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3222 for (j = 0; j < (1 << BYTEWIDTH); j++)
3223 if (SYNTAX (j) == Sword)
3229 for (j = 0; j < (1 << BYTEWIDTH); j++)
3230 if (SYNTAX (j) != Sword)
3237 int fastmap_newline = fastmap['\n'];
3239 /* `.' matches anything ... */
3240 for (j = 0; j < (1 << BYTEWIDTH); j++)
3243 /* ... except perhaps newline. */
3244 if (!(bufp->syntax & RE_DOT_NEWLINE))
3245 fastmap['\n'] = fastmap_newline;
3247 /* Return if we have already set `can_be_null'; if we have,
3248 then the fastmap is irrelevant. Something's wrong here. */
3249 else if (bufp->can_be_null)
3252 /* Otherwise, have to check alternative paths. */
3259 for (j = 0; j < (1 << BYTEWIDTH); j++)
3260 if (SYNTAX (j) == (enum syntaxcode) k)
3267 for (j = 0; j < (1 << BYTEWIDTH); j++)
3268 if (SYNTAX (j) != (enum syntaxcode) k)
3273 /* All cases after this match the empty string. These end with
3293 case push_dummy_failure:
3298 case pop_failure_jump:
3299 case maybe_pop_jump:
3302 case dummy_failure_jump:
3303 EXTRACT_NUMBER_AND_INCR (j, p);
3308 /* Jump backward implies we just went through the body of a
3309 loop and matched nothing. Opcode jumped to should be
3310 `on_failure_jump' or `succeed_n'. Just treat it like an
3311 ordinary jump. For a * loop, it has pushed its failure
3312 point already; if so, discard that as redundant. */
3313 if ((re_opcode_t) *p != on_failure_jump
3314 && (re_opcode_t) *p != succeed_n)
3318 EXTRACT_NUMBER_AND_INCR (j, p);
3321 /* If what's on the stack is where we are now, pop it. */
3322 if (!FAIL_STACK_EMPTY ()
3323 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3329 case on_failure_jump:
3330 case on_failure_keep_string_jump:
3331 handle_on_failure_jump:
3332 EXTRACT_NUMBER_AND_INCR (j, p);
3334 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3335 end of the pattern. We don't want to push such a point,
3336 since when we restore it above, entering the switch will
3337 increment `p' past the end of the pattern. We don't need
3338 to push such a point since we obviously won't find any more
3339 fastmap entries beyond `pend'. Such a pattern can match
3340 the null string, though. */
3343 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3345 RESET_FAIL_STACK ();
3350 bufp->can_be_null = 1;
3354 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3355 succeed_n_p = false;
3362 /* Get to the number of times to succeed. */
3365 /* Increment p past the n for when k != 0. */
3366 EXTRACT_NUMBER_AND_INCR (k, p);
3370 succeed_n_p = true; /* Spaghetti code alert. */
3371 goto handle_on_failure_jump;
3388 abort (); /* We have listed all the cases. */
3391 /* Getting here means we have found the possible starting
3392 characters for one path of the pattern -- and that the empty
3393 string does not match. We need not follow this path further.
3394 Instead, look at the next alternative (remembered on the
3395 stack), or quit if no more. The test at the top of the loop
3396 does these things. */
3397 path_can_be_null = false;
3401 /* Set `can_be_null' for the last path (also the first path, if the
3402 pattern is empty). */
3403 bufp->can_be_null |= path_can_be_null;
3406 RESET_FAIL_STACK ();
3408 } /* re_compile_fastmap */
3410 weak_alias (__re_compile_fastmap, re_compile_fastmap)
3413 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3414 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3415 this memory for recording register information. STARTS and ENDS
3416 must be allocated using the malloc library routine, and must each
3417 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3419 If NUM_REGS == 0, then subsequent matches should allocate their own
3422 Unless this function is called, the first search or match using
3423 PATTERN_BUFFER will allocate its own register data, without
3424 freeing the old data. */
3427 re_set_registers (bufp, regs, num_regs, starts, ends)
3428 struct re_pattern_buffer *bufp;
3429 struct re_registers *regs;
3431 regoff_t *starts, *ends;
3435 bufp->regs_allocated = REGS_REALLOCATE;
3436 regs->num_regs = num_regs;
3437 regs->start = starts;
3442 bufp->regs_allocated = REGS_UNALLOCATED;
3444 regs->start = regs->end = (regoff_t *) 0;
3448 weak_alias (__re_set_registers, re_set_registers)
3451 /* Searching routines. */
3453 /* Like re_search_2, below, but only one string is specified, and
3454 doesn't let you say where to stop matching. */
3457 re_search (bufp, string, size, startpos, range, regs)
3458 struct re_pattern_buffer *bufp;
3460 int size, startpos, range;
3461 struct re_registers *regs;
3463 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3467 weak_alias (__re_search, re_search)
3471 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3472 virtual concatenation of STRING1 and STRING2, starting first at index
3473 STARTPOS, then at STARTPOS + 1, and so on.
3475 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3477 RANGE is how far to scan while trying to match. RANGE = 0 means try
3478 only at STARTPOS; in general, the last start tried is STARTPOS +
3481 In REGS, return the indices of the virtual concatenation of STRING1
3482 and STRING2 that matched the entire BUFP->buffer and its contained
3485 Do not consider matching one past the index STOP in the virtual
3486 concatenation of STRING1 and STRING2.
3488 We return either the position in the strings at which the match was
3489 found, -1 if no match, or -2 if error (such as failure
3493 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3494 struct re_pattern_buffer *bufp;
3495 const char *string1, *string2;
3499 struct re_registers *regs;
3503 register char *fastmap = bufp->fastmap;
3504 register RE_TRANSLATE_TYPE translate = bufp->translate;
3505 int total_size = size1 + size2;
3506 int endpos = startpos + range;
3508 /* Check for out-of-range STARTPOS. */
3509 if (startpos < 0 || startpos > total_size)
3512 /* Fix up RANGE if it might eventually take us outside
3513 the virtual concatenation of STRING1 and STRING2.
3514 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3516 range = 0 - startpos;
3517 else if (endpos > total_size)
3518 range = total_size - startpos;
3520 /* If the search isn't to be a backwards one, don't waste time in a
3521 search for a pattern that must be anchored. */
3522 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3531 /* In a forward search for something that starts with \=.
3532 don't keep searching past point. */
3533 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3535 range = PT - startpos;
3541 /* Update the fastmap now if not correct already. */
3542 if (fastmap && !bufp->fastmap_accurate)
3543 if (re_compile_fastmap (bufp) == -2)
3546 /* Loop through the string, looking for a place to start matching. */
3549 /* If a fastmap is supplied, skip quickly over characters that
3550 cannot be the start of a match. If the pattern can match the
3551 null string, however, we don't need to skip characters; we want
3552 the first null string. */
3553 if (fastmap && startpos < total_size && !bufp->can_be_null)
3555 if (range > 0) /* Searching forwards. */
3557 register const char *d;
3558 register int lim = 0;
3561 if (startpos < size1 && startpos + range >= size1)
3562 lim = range - (size1 - startpos);
3564 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3566 /* Written out as an if-else to avoid testing `translate'
3570 && !fastmap[(unsigned char)
3571 translate[(unsigned char) *d++]])
3574 while (range > lim && !fastmap[(unsigned char) *d++])
3577 startpos += irange - range;
3579 else /* Searching backwards. */
3581 register char c = (size1 == 0 || startpos >= size1
3582 ? string2[startpos - size1]
3583 : string1[startpos]);
3585 if (!fastmap[(unsigned char) TRANSLATE (c)])
3590 /* If can't match the null string, and that's all we have left, fail. */
3591 if (range >= 0 && startpos == total_size && fastmap
3592 && !bufp->can_be_null)
3595 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3596 startpos, regs, stop);
3597 #ifndef REGEX_MALLOC
3626 weak_alias (__re_search_2, re_search_2)
3629 /* This converts PTR, a pointer into one of the search strings `string1'
3630 and `string2' into an offset from the beginning of that string. */
3631 #define POINTER_TO_OFFSET(ptr) \
3632 (FIRST_STRING_P (ptr) \
3633 ? ((regoff_t) ((ptr) - string1)) \
3634 : ((regoff_t) ((ptr) - string2 + size1)))
3636 /* Macros for dealing with the split strings in re_match_2. */
3638 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3640 /* Call before fetching a character with *d. This switches over to
3641 string2 if necessary. */
3642 #define PREFETCH() \
3645 /* End of string2 => fail. */ \
3646 if (dend == end_match_2) \
3648 /* End of string1 => advance to string2. */ \
3650 dend = end_match_2; \
3654 /* Test if at very beginning or at very end of the virtual concatenation
3655 of `string1' and `string2'. If only one string, it's `string2'. */
3656 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3657 #define AT_STRINGS_END(d) ((d) == end2)
3660 /* Test if D points to a character which is word-constituent. We have
3661 two special cases to check for: if past the end of string1, look at
3662 the first character in string2; and if before the beginning of
3663 string2, look at the last character in string1. */
3664 #define WORDCHAR_P(d) \
3665 (SYNTAX ((d) == end1 ? *string2 \
3666 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3669 /* Disabled due to a compiler bug -- see comment at case wordbound */
3671 /* Test if the character before D and the one at D differ with respect
3672 to being word-constituent. */
3673 #define AT_WORD_BOUNDARY(d) \
3674 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3675 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3678 /* Free everything we malloc. */
3679 #ifdef MATCH_MAY_ALLOCATE
3680 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3681 # define FREE_VARIABLES() \
3683 REGEX_FREE_STACK (fail_stack.stack); \
3684 FREE_VAR (regstart); \
3685 FREE_VAR (regend); \
3686 FREE_VAR (old_regstart); \
3687 FREE_VAR (old_regend); \
3688 FREE_VAR (best_regstart); \
3689 FREE_VAR (best_regend); \
3690 FREE_VAR (reg_info); \
3691 FREE_VAR (reg_dummy); \
3692 FREE_VAR (reg_info_dummy); \
3695 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3696 #endif /* not MATCH_MAY_ALLOCATE */
3698 /* These values must meet several constraints. They must not be valid
3699 register values; since we have a limit of 255 registers (because
3700 we use only one byte in the pattern for the register number), we can
3701 use numbers larger than 255. They must differ by 1, because of
3702 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3703 be larger than the value for the highest register, so we do not try
3704 to actually save any registers when none are active. */
3705 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3706 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3708 /* Matching routines. */
3710 #ifndef emacs /* Emacs never uses this. */
3711 /* re_match is like re_match_2 except it takes only a single string. */
3714 re_match (bufp, string, size, pos, regs)
3715 struct re_pattern_buffer *bufp;
3718 struct re_registers *regs;
3720 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3722 # ifndef REGEX_MALLOC
3730 weak_alias (__re_match, re_match)
3732 #endif /* not emacs */
3734 static boolean group_match_null_string_p _RE_ARGS ((unsigned char **p,
3736 register_info_type *reg_info));
3737 static boolean alt_match_null_string_p _RE_ARGS ((unsigned char *p,
3739 register_info_type *reg_info));
3740 static boolean common_op_match_null_string_p _RE_ARGS ((unsigned char **p,
3742 register_info_type *reg_info));
3743 static int bcmp_translate _RE_ARGS ((const char *s1, const char *s2,
3744 int len, char *translate));
3746 /* re_match_2 matches the compiled pattern in BUFP against the
3747 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3748 and SIZE2, respectively). We start matching at POS, and stop
3751 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3752 store offsets for the substring each group matched in REGS. See the
3753 documentation for exactly how many groups we fill.
3755 We return -1 if no match, -2 if an internal error (such as the
3756 failure stack overflowing). Otherwise, we return the length of the
3757 matched substring. */
3760 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3761 struct re_pattern_buffer *bufp;
3762 const char *string1, *string2;
3765 struct re_registers *regs;
3768 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3770 #ifndef REGEX_MALLOC
3778 weak_alias (__re_match_2, re_match_2)
3781 /* This is a separate function so that we can force an alloca cleanup
3784 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3785 struct re_pattern_buffer *bufp;
3786 const char *string1, *string2;
3789 struct re_registers *regs;
3792 /* General temporaries. */
3796 /* Just past the end of the corresponding string. */
3797 const char *end1, *end2;
3799 /* Pointers into string1 and string2, just past the last characters in
3800 each to consider matching. */
3801 const char *end_match_1, *end_match_2;
3803 /* Where we are in the data, and the end of the current string. */
3804 const char *d, *dend;
3806 /* Where we are in the pattern, and the end of the pattern. */
3807 unsigned char *p = bufp->buffer;
3808 register unsigned char *pend = p + bufp->used;
3810 /* Mark the opcode just after a start_memory, so we can test for an
3811 empty subpattern when we get to the stop_memory. */
3812 unsigned char *just_past_start_mem = 0;
3814 /* We use this to map every character in the string. */
3815 RE_TRANSLATE_TYPE translate = bufp->translate;
3817 /* Failure point stack. Each place that can handle a failure further
3818 down the line pushes a failure point on this stack. It consists of
3819 restart, regend, and reg_info for all registers corresponding to
3820 the subexpressions we're currently inside, plus the number of such
3821 registers, and, finally, two char *'s. The first char * is where
3822 to resume scanning the pattern; the second one is where to resume
3823 scanning the strings. If the latter is zero, the failure point is
3824 a ``dummy''; if a failure happens and the failure point is a dummy,
3825 it gets discarded and the next next one is tried. */
3826 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3827 fail_stack_type fail_stack;
3830 static unsigned failure_id = 0;
3831 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3835 /* This holds the pointer to the failure stack, when
3836 it is allocated relocatably. */
3837 fail_stack_elt_t *failure_stack_ptr;
3840 /* We fill all the registers internally, independent of what we
3841 return, for use in backreferences. The number here includes
3842 an element for register zero. */
3843 size_t num_regs = bufp->re_nsub + 1;
3845 /* The currently active registers. */
3846 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3847 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3849 /* Information on the contents of registers. These are pointers into
3850 the input strings; they record just what was matched (on this
3851 attempt) by a subexpression part of the pattern, that is, the
3852 regnum-th regstart pointer points to where in the pattern we began
3853 matching and the regnum-th regend points to right after where we
3854 stopped matching the regnum-th subexpression. (The zeroth register
3855 keeps track of what the whole pattern matches.) */
3856 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3857 const char **regstart, **regend;
3860 /* If a group that's operated upon by a repetition operator fails to
3861 match anything, then the register for its start will need to be
3862 restored because it will have been set to wherever in the string we
3863 are when we last see its open-group operator. Similarly for a
3865 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3866 const char **old_regstart, **old_regend;
3869 /* The is_active field of reg_info helps us keep track of which (possibly
3870 nested) subexpressions we are currently in. The matched_something
3871 field of reg_info[reg_num] helps us tell whether or not we have
3872 matched any of the pattern so far this time through the reg_num-th
3873 subexpression. These two fields get reset each time through any
3874 loop their register is in. */
3875 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3876 register_info_type *reg_info;
3879 /* The following record the register info as found in the above
3880 variables when we find a match better than any we've seen before.
3881 This happens as we backtrack through the failure points, which in
3882 turn happens only if we have not yet matched the entire string. */
3883 unsigned best_regs_set = false;
3884 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3885 const char **best_regstart, **best_regend;
3888 /* Logically, this is `best_regend[0]'. But we don't want to have to
3889 allocate space for that if we're not allocating space for anything
3890 else (see below). Also, we never need info about register 0 for
3891 any of the other register vectors, and it seems rather a kludge to
3892 treat `best_regend' differently than the rest. So we keep track of
3893 the end of the best match so far in a separate variable. We
3894 initialize this to NULL so that when we backtrack the first time
3895 and need to test it, it's not garbage. */
3896 const char *match_end = NULL;
3898 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3899 int set_regs_matched_done = 0;
3901 /* Used when we pop values we don't care about. */
3902 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3903 const char **reg_dummy;
3904 register_info_type *reg_info_dummy;
3908 /* Counts the total number of registers pushed. */
3909 unsigned num_regs_pushed = 0;
3912 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3916 #ifdef MATCH_MAY_ALLOCATE
3917 /* Do not bother to initialize all the register variables if there are
3918 no groups in the pattern, as it takes a fair amount of time. If
3919 there are groups, we include space for register 0 (the whole
3920 pattern), even though we never use it, since it simplifies the
3921 array indexing. We should fix this. */
3924 regstart = REGEX_TALLOC (num_regs, const char *);
3925 regend = REGEX_TALLOC (num_regs, const char *);
3926 old_regstart = REGEX_TALLOC (num_regs, const char *);
3927 old_regend = REGEX_TALLOC (num_regs, const char *);
3928 best_regstart = REGEX_TALLOC (num_regs, const char *);
3929 best_regend = REGEX_TALLOC (num_regs, const char *);
3930 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3931 reg_dummy = REGEX_TALLOC (num_regs, const char *);
3932 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3934 if (!(regstart && regend && old_regstart && old_regend && reg_info
3935 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3943 /* We must initialize all our variables to NULL, so that
3944 `FREE_VARIABLES' doesn't try to free them. */
3945 regstart = regend = old_regstart = old_regend = best_regstart
3946 = best_regend = reg_dummy = NULL;
3947 reg_info = reg_info_dummy = (register_info_type *) NULL;
3949 #endif /* MATCH_MAY_ALLOCATE */
3951 /* The starting position is bogus. */
3952 if (pos < 0 || pos > size1 + size2)
3958 /* Initialize subexpression text positions to -1 to mark ones that no
3959 start_memory/stop_memory has been seen for. Also initialize the
3960 register information struct. */
3961 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
3963 regstart[mcnt] = regend[mcnt]
3964 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3966 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3967 IS_ACTIVE (reg_info[mcnt]) = 0;
3968 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3969 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3972 /* We move `string1' into `string2' if the latter's empty -- but not if
3973 `string1' is null. */
3974 if (size2 == 0 && string1 != NULL)
3981 end1 = string1 + size1;
3982 end2 = string2 + size2;
3984 /* Compute where to stop matching, within the two strings. */
3987 end_match_1 = string1 + stop;
3988 end_match_2 = string2;
3993 end_match_2 = string2 + stop - size1;
3996 /* `p' scans through the pattern as `d' scans through the data.
3997 `dend' is the end of the input string that `d' points within. `d'
3998 is advanced into the following input string whenever necessary, but
3999 this happens before fetching; therefore, at the beginning of the
4000 loop, `d' can be pointing at the end of a string, but it cannot
4002 if (size1 > 0 && pos <= size1)
4009 d = string2 + pos - size1;
4013 DEBUG_PRINT1 ("The compiled pattern is:\n");
4014 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
4015 DEBUG_PRINT1 ("The string to match is: `");
4016 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
4017 DEBUG_PRINT1 ("'\n");
4019 /* This loops over pattern commands. It exits by returning from the
4020 function if the match is complete, or it drops through if the match
4021 fails at this starting point in the input data. */
4025 DEBUG_PRINT2 ("\n%p: ", p);
4027 DEBUG_PRINT2 ("\n0x%x: ", p);
4031 { /* End of pattern means we might have succeeded. */
4032 DEBUG_PRINT1 ("end of pattern ... ");
4034 /* If we haven't matched the entire string, and we want the
4035 longest match, try backtracking. */
4036 if (d != end_match_2)
4038 /* 1 if this match ends in the same string (string1 or string2)
4039 as the best previous match. */
4040 boolean same_str_p = (FIRST_STRING_P (match_end)
4041 == MATCHING_IN_FIRST_STRING);
4042 /* 1 if this match is the best seen so far. */
4043 boolean best_match_p;
4045 /* AIX compiler got confused when this was combined
4046 with the previous declaration. */
4048 best_match_p = d > match_end;
4050 best_match_p = !MATCHING_IN_FIRST_STRING;
4052 DEBUG_PRINT1 ("backtracking.\n");
4054 if (!FAIL_STACK_EMPTY ())
4055 { /* More failure points to try. */
4057 /* If exceeds best match so far, save it. */
4058 if (!best_regs_set || best_match_p)
4060 best_regs_set = true;
4063 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
4065 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4067 best_regstart[mcnt] = regstart[mcnt];
4068 best_regend[mcnt] = regend[mcnt];
4074 /* If no failure points, don't restore garbage. And if
4075 last match is real best match, don't restore second
4077 else if (best_regs_set && !best_match_p)
4080 /* Restore best match. It may happen that `dend ==
4081 end_match_1' while the restored d is in string2.
4082 For example, the pattern `x.*y.*z' against the
4083 strings `x-' and `y-z-', if the two strings are
4084 not consecutive in memory. */
4085 DEBUG_PRINT1 ("Restoring best registers.\n");
4088 dend = ((d >= string1 && d <= end1)
4089 ? end_match_1 : end_match_2);
4091 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4093 regstart[mcnt] = best_regstart[mcnt];
4094 regend[mcnt] = best_regend[mcnt];
4097 } /* d != end_match_2 */
4100 DEBUG_PRINT1 ("Accepting match.\n");
4102 /* If caller wants register contents data back, do it. */
4103 if (regs && !bufp->no_sub)
4105 /* Have the register data arrays been allocated? */
4106 if (bufp->regs_allocated == REGS_UNALLOCATED)
4107 { /* No. So allocate them with malloc. We need one
4108 extra element beyond `num_regs' for the `-1' marker
4110 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
4111 regs->start = TALLOC (regs->num_regs, regoff_t);
4112 regs->end = TALLOC (regs->num_regs, regoff_t);
4113 if (regs->start == NULL || regs->end == NULL)
4118 bufp->regs_allocated = REGS_REALLOCATE;
4120 else if (bufp->regs_allocated == REGS_REALLOCATE)
4121 { /* Yes. If we need more elements than were already
4122 allocated, reallocate them. If we need fewer, just
4124 if (regs->num_regs < num_regs + 1)
4126 regs->num_regs = num_regs + 1;
4127 RETALLOC (regs->start, regs->num_regs, regoff_t);
4128 RETALLOC (regs->end, regs->num_regs, regoff_t);
4129 if (regs->start == NULL || regs->end == NULL)
4138 /* These braces fend off a "empty body in an else-statement"
4139 warning under GCC when assert expands to nothing. */
4140 assert (bufp->regs_allocated == REGS_FIXED);
4143 /* Convert the pointer data in `regstart' and `regend' to
4144 indices. Register zero has to be set differently,
4145 since we haven't kept track of any info for it. */
4146 if (regs->num_regs > 0)
4148 regs->start[0] = pos;
4149 regs->end[0] = (MATCHING_IN_FIRST_STRING
4150 ? ((regoff_t) (d - string1))
4151 : ((regoff_t) (d - string2 + size1)));
4154 /* Go through the first `min (num_regs, regs->num_regs)'
4155 registers, since that is all we initialized. */
4156 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
4159 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
4160 regs->start[mcnt] = regs->end[mcnt] = -1;
4164 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
4166 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
4170 /* If the regs structure we return has more elements than
4171 were in the pattern, set the extra elements to -1. If
4172 we (re)allocated the registers, this is the case,
4173 because we always allocate enough to have at least one
4175 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
4176 regs->start[mcnt] = regs->end[mcnt] = -1;
4177 } /* regs && !bufp->no_sub */
4179 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
4180 nfailure_points_pushed, nfailure_points_popped,
4181 nfailure_points_pushed - nfailure_points_popped);
4182 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
4184 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
4188 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
4194 /* Otherwise match next pattern command. */
4195 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4197 /* Ignore these. Used to ignore the n of succeed_n's which
4198 currently have n == 0. */
4200 DEBUG_PRINT1 ("EXECUTING no_op.\n");
4204 DEBUG_PRINT1 ("EXECUTING succeed.\n");
4207 /* Match the next n pattern characters exactly. The following
4208 byte in the pattern defines n, and the n bytes after that
4209 are the characters to match. */
4212 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
4214 /* This is written out as an if-else so we don't waste time
4215 testing `translate' inside the loop. */
4221 if ((unsigned char) translate[(unsigned char) *d++]
4222 != (unsigned char) *p++)
4232 if (*d++ != (char) *p++) goto fail;
4236 SET_REGS_MATCHED ();
4240 /* Match any character except possibly a newline or a null. */
4242 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4246 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
4247 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4250 SET_REGS_MATCHED ();
4251 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4259 register unsigned char c;
4260 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4262 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4265 c = TRANSLATE (*d); /* The character to match. */
4267 /* Cast to `unsigned' instead of `unsigned char' in case the
4268 bit list is a full 32 bytes long. */
4269 if (c < (unsigned) (*p * BYTEWIDTH)
4270 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4275 if (!not) goto fail;
4277 SET_REGS_MATCHED ();
4283 /* The beginning of a group is represented by start_memory.
4284 The arguments are the register number in the next byte, and the
4285 number of groups inner to this one in the next. The text
4286 matched within the group is recorded (in the internal
4287 registers data structure) under the register number. */
4289 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4291 /* Find out if this group can match the empty string. */
4292 p1 = p; /* To send to group_match_null_string_p. */
4294 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4295 REG_MATCH_NULL_STRING_P (reg_info[*p])
4296 = group_match_null_string_p (&p1, pend, reg_info);
4298 /* Save the position in the string where we were the last time
4299 we were at this open-group operator in case the group is
4300 operated upon by a repetition operator, e.g., with `(a*)*b'
4301 against `ab'; then we want to ignore where we are now in
4302 the string in case this attempt to match fails. */
4303 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4304 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4306 DEBUG_PRINT2 (" old_regstart: %d\n",
4307 POINTER_TO_OFFSET (old_regstart[*p]));
4310 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4312 IS_ACTIVE (reg_info[*p]) = 1;
4313 MATCHED_SOMETHING (reg_info[*p]) = 0;
4315 /* Clear this whenever we change the register activity status. */
4316 set_regs_matched_done = 0;
4318 /* This is the new highest active register. */
4319 highest_active_reg = *p;
4321 /* If nothing was active before, this is the new lowest active
4323 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4324 lowest_active_reg = *p;
4326 /* Move past the register number and inner group count. */
4328 just_past_start_mem = p;
4333 /* The stop_memory opcode represents the end of a group. Its
4334 arguments are the same as start_memory's: the register
4335 number, and the number of inner groups. */
4337 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4339 /* We need to save the string position the last time we were at
4340 this close-group operator in case the group is operated
4341 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4342 against `aba'; then we want to ignore where we are now in
4343 the string in case this attempt to match fails. */
4344 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4345 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4347 DEBUG_PRINT2 (" old_regend: %d\n",
4348 POINTER_TO_OFFSET (old_regend[*p]));
4351 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4353 /* This register isn't active anymore. */
4354 IS_ACTIVE (reg_info[*p]) = 0;
4356 /* Clear this whenever we change the register activity status. */
4357 set_regs_matched_done = 0;
4359 /* If this was the only register active, nothing is active
4361 if (lowest_active_reg == highest_active_reg)
4363 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4364 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4367 { /* We must scan for the new highest active register, since
4368 it isn't necessarily one less than now: consider
4369 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4370 new highest active register is 1. */
4371 unsigned char r = *p - 1;
4372 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4375 /* If we end up at register zero, that means that we saved
4376 the registers as the result of an `on_failure_jump', not
4377 a `start_memory', and we jumped to past the innermost
4378 `stop_memory'. For example, in ((.)*) we save
4379 registers 1 and 2 as a result of the *, but when we pop
4380 back to the second ), we are at the stop_memory 1.
4381 Thus, nothing is active. */
4384 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4385 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4388 highest_active_reg = r;
4391 /* If just failed to match something this time around with a
4392 group that's operated on by a repetition operator, try to
4393 force exit from the ``loop'', and restore the register
4394 information for this group that we had before trying this
4396 if ((!MATCHED_SOMETHING (reg_info[*p])
4397 || just_past_start_mem == p - 1)
4400 boolean is_a_jump_n = false;
4404 switch ((re_opcode_t) *p1++)
4408 case pop_failure_jump:
4409 case maybe_pop_jump:
4411 case dummy_failure_jump:
4412 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4422 /* If the next operation is a jump backwards in the pattern
4423 to an on_failure_jump right before the start_memory
4424 corresponding to this stop_memory, exit from the loop
4425 by forcing a failure after pushing on the stack the
4426 on_failure_jump's jump in the pattern, and d. */
4427 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4428 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4430 /* If this group ever matched anything, then restore
4431 what its registers were before trying this last
4432 failed match, e.g., with `(a*)*b' against `ab' for
4433 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4434 against `aba' for regend[3].
4436 Also restore the registers for inner groups for,
4437 e.g., `((a*)(b*))*' against `aba' (register 3 would
4438 otherwise get trashed). */
4440 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4444 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4446 /* Restore this and inner groups' (if any) registers. */
4447 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
4450 regstart[r] = old_regstart[r];
4452 /* xx why this test? */
4453 if (old_regend[r] >= regstart[r])
4454 regend[r] = old_regend[r];
4458 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4459 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4465 /* Move past the register number and the inner group count. */
4470 /* \<digit> has been turned into a `duplicate' command which is
4471 followed by the numeric value of <digit> as the register number. */
4474 register const char *d2, *dend2;
4475 int regno = *p++; /* Get which register to match against. */
4476 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4478 /* Can't back reference a group which we've never matched. */
4479 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4482 /* Where in input to try to start matching. */
4483 d2 = regstart[regno];
4485 /* Where to stop matching; if both the place to start and
4486 the place to stop matching are in the same string, then
4487 set to the place to stop, otherwise, for now have to use
4488 the end of the first string. */
4490 dend2 = ((FIRST_STRING_P (regstart[regno])
4491 == FIRST_STRING_P (regend[regno]))
4492 ? regend[regno] : end_match_1);
4495 /* If necessary, advance to next segment in register
4499 if (dend2 == end_match_2) break;
4500 if (dend2 == regend[regno]) break;
4502 /* End of string1 => advance to string2. */
4504 dend2 = regend[regno];
4506 /* At end of register contents => success */
4507 if (d2 == dend2) break;
4509 /* If necessary, advance to next segment in data. */
4512 /* How many characters left in this segment to match. */
4515 /* Want how many consecutive characters we can match in
4516 one shot, so, if necessary, adjust the count. */
4517 if (mcnt > dend2 - d2)
4520 /* Compare that many; failure if mismatch, else move
4523 ? bcmp_translate (d, d2, mcnt, translate)
4524 : memcmp (d, d2, mcnt))
4526 d += mcnt, d2 += mcnt;
4528 /* Do this because we've match some characters. */
4529 SET_REGS_MATCHED ();
4535 /* begline matches the empty string at the beginning of the string
4536 (unless `not_bol' is set in `bufp'), and, if
4537 `newline_anchor' is set, after newlines. */
4539 DEBUG_PRINT1 ("EXECUTING begline.\n");
4541 if (AT_STRINGS_BEG (d))
4543 if (!bufp->not_bol) break;
4545 else if (d[-1] == '\n' && bufp->newline_anchor)
4549 /* In all other cases, we fail. */
4553 /* endline is the dual of begline. */
4555 DEBUG_PRINT1 ("EXECUTING endline.\n");
4557 if (AT_STRINGS_END (d))
4559 if (!bufp->not_eol) break;
4562 /* We have to ``prefetch'' the next character. */
4563 else if ((d == end1 ? *string2 : *d) == '\n'
4564 && bufp->newline_anchor)
4571 /* Match at the very beginning of the data. */
4573 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4574 if (AT_STRINGS_BEG (d))
4579 /* Match at the very end of the data. */
4581 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4582 if (AT_STRINGS_END (d))
4587 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4588 pushes NULL as the value for the string on the stack. Then
4589 `pop_failure_point' will keep the current value for the
4590 string, instead of restoring it. To see why, consider
4591 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4592 then the . fails against the \n. But the next thing we want
4593 to do is match the \n against the \n; if we restored the
4594 string value, we would be back at the foo.
4596 Because this is used only in specific cases, we don't need to
4597 check all the things that `on_failure_jump' does, to make
4598 sure the right things get saved on the stack. Hence we don't
4599 share its code. The only reason to push anything on the
4600 stack at all is that otherwise we would have to change
4601 `anychar's code to do something besides goto fail in this
4602 case; that seems worse than this. */
4603 case on_failure_keep_string_jump:
4604 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4606 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4608 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
4610 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4613 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4617 /* Uses of on_failure_jump:
4619 Each alternative starts with an on_failure_jump that points
4620 to the beginning of the next alternative. Each alternative
4621 except the last ends with a jump that in effect jumps past
4622 the rest of the alternatives. (They really jump to the
4623 ending jump of the following alternative, because tensioning
4624 these jumps is a hassle.)
4626 Repeats start with an on_failure_jump that points past both
4627 the repetition text and either the following jump or
4628 pop_failure_jump back to this on_failure_jump. */
4629 case on_failure_jump:
4631 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4633 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4635 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
4637 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4640 /* If this on_failure_jump comes right before a group (i.e.,
4641 the original * applied to a group), save the information
4642 for that group and all inner ones, so that if we fail back
4643 to this point, the group's information will be correct.
4644 For example, in \(a*\)*\1, we need the preceding group,
4645 and in \(zz\(a*\)b*\)\2, we need the inner group. */
4647 /* We can't use `p' to check ahead because we push
4648 a failure point to `p + mcnt' after we do this. */
4651 /* We need to skip no_op's before we look for the
4652 start_memory in case this on_failure_jump is happening as
4653 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4655 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4658 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4660 /* We have a new highest active register now. This will
4661 get reset at the start_memory we are about to get to,
4662 but we will have saved all the registers relevant to
4663 this repetition op, as described above. */
4664 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4665 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4666 lowest_active_reg = *(p1 + 1);
4669 DEBUG_PRINT1 (":\n");
4670 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4674 /* A smart repeat ends with `maybe_pop_jump'.
4675 We change it to either `pop_failure_jump' or `jump'. */
4676 case maybe_pop_jump:
4677 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4678 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4680 register unsigned char *p2 = p;
4682 /* Compare the beginning of the repeat with what in the
4683 pattern follows its end. If we can establish that there
4684 is nothing that they would both match, i.e., that we
4685 would have to backtrack because of (as in, e.g., `a*a')
4686 then we can change to pop_failure_jump, because we'll
4687 never have to backtrack.
4689 This is not true in the case of alternatives: in
4690 `(a|ab)*' we do need to backtrack to the `ab' alternative
4691 (e.g., if the string was `ab'). But instead of trying to
4692 detect that here, the alternative has put on a dummy
4693 failure point which is what we will end up popping. */
4695 /* Skip over open/close-group commands.
4696 If what follows this loop is a ...+ construct,
4697 look at what begins its body, since we will have to
4698 match at least one of that. */
4702 && ((re_opcode_t) *p2 == stop_memory
4703 || (re_opcode_t) *p2 == start_memory))
4705 else if (p2 + 6 < pend
4706 && (re_opcode_t) *p2 == dummy_failure_jump)
4713 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4714 to the `maybe_finalize_jump' of this case. Examine what
4717 /* If we're at the end of the pattern, we can change. */
4720 /* Consider what happens when matching ":\(.*\)"
4721 against ":/". I don't really understand this code
4723 p[-3] = (unsigned char) pop_failure_jump;
4725 (" End of pattern: change to `pop_failure_jump'.\n");
4728 else if ((re_opcode_t) *p2 == exactn
4729 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4731 register unsigned char c
4732 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4734 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4736 p[-3] = (unsigned char) pop_failure_jump;
4737 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4741 else if ((re_opcode_t) p1[3] == charset
4742 || (re_opcode_t) p1[3] == charset_not)
4744 int not = (re_opcode_t) p1[3] == charset_not;
4746 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4747 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4750 /* `not' is equal to 1 if c would match, which means
4751 that we can't change to pop_failure_jump. */
4754 p[-3] = (unsigned char) pop_failure_jump;
4755 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4759 else if ((re_opcode_t) *p2 == charset)
4762 register unsigned char c
4763 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4767 if ((re_opcode_t) p1[3] == exactn
4768 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
4769 && (p2[2 + p1[5] / BYTEWIDTH]
4770 & (1 << (p1[5] % BYTEWIDTH)))))
4772 if ((re_opcode_t) p1[3] == exactn
4773 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[4]
4774 && (p2[2 + p1[4] / BYTEWIDTH]
4775 & (1 << (p1[4] % BYTEWIDTH)))))
4778 p[-3] = (unsigned char) pop_failure_jump;
4779 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4783 else if ((re_opcode_t) p1[3] == charset_not)
4786 /* We win if the charset_not inside the loop
4787 lists every character listed in the charset after. */
4788 for (idx = 0; idx < (int) p2[1]; idx++)
4789 if (! (p2[2 + idx] == 0
4790 || (idx < (int) p1[4]
4791 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4796 p[-3] = (unsigned char) pop_failure_jump;
4797 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4800 else if ((re_opcode_t) p1[3] == charset)
4803 /* We win if the charset inside the loop
4804 has no overlap with the one after the loop. */
4806 idx < (int) p2[1] && idx < (int) p1[4];
4808 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4811 if (idx == p2[1] || idx == p1[4])
4813 p[-3] = (unsigned char) pop_failure_jump;
4814 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4819 p -= 2; /* Point at relative address again. */
4820 if ((re_opcode_t) p[-1] != pop_failure_jump)
4822 p[-1] = (unsigned char) jump;
4823 DEBUG_PRINT1 (" Match => jump.\n");
4824 goto unconditional_jump;
4826 /* Note fall through. */
4829 /* The end of a simple repeat has a pop_failure_jump back to
4830 its matching on_failure_jump, where the latter will push a
4831 failure point. The pop_failure_jump takes off failure
4832 points put on by this pop_failure_jump's matching
4833 on_failure_jump; we got through the pattern to here from the
4834 matching on_failure_jump, so didn't fail. */
4835 case pop_failure_jump:
4837 /* We need to pass separate storage for the lowest and
4838 highest registers, even though we don't care about the
4839 actual values. Otherwise, we will restore only one
4840 register from the stack, since lowest will == highest in
4841 `pop_failure_point'. */
4842 active_reg_t dummy_low_reg, dummy_high_reg;
4843 unsigned char *pdummy;
4846 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4847 POP_FAILURE_POINT (sdummy, pdummy,
4848 dummy_low_reg, dummy_high_reg,
4849 reg_dummy, reg_dummy, reg_info_dummy);
4851 /* Note fall through. */
4855 DEBUG_PRINT2 ("\n%p: ", p);
4857 DEBUG_PRINT2 ("\n0x%x: ", p);
4859 /* Note fall through. */
4861 /* Unconditionally jump (without popping any failure points). */
4863 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4864 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4865 p += mcnt; /* Do the jump. */
4867 DEBUG_PRINT2 ("(to %p).\n", p);
4869 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4874 /* We need this opcode so we can detect where alternatives end
4875 in `group_match_null_string_p' et al. */
4877 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4878 goto unconditional_jump;
4881 /* Normally, the on_failure_jump pushes a failure point, which
4882 then gets popped at pop_failure_jump. We will end up at
4883 pop_failure_jump, also, and with a pattern of, say, `a+', we
4884 are skipping over the on_failure_jump, so we have to push
4885 something meaningless for pop_failure_jump to pop. */
4886 case dummy_failure_jump:
4887 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4888 /* It doesn't matter what we push for the string here. What
4889 the code at `fail' tests is the value for the pattern. */
4890 PUSH_FAILURE_POINT (NULL, NULL, -2);
4891 goto unconditional_jump;
4894 /* At the end of an alternative, we need to push a dummy failure
4895 point in case we are followed by a `pop_failure_jump', because
4896 we don't want the failure point for the alternative to be
4897 popped. For example, matching `(a|ab)*' against `aab'
4898 requires that we match the `ab' alternative. */
4899 case push_dummy_failure:
4900 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4901 /* See comments just above at `dummy_failure_jump' about the
4903 PUSH_FAILURE_POINT (NULL, NULL, -2);
4906 /* Have to succeed matching what follows at least n times.
4907 After that, handle like `on_failure_jump'. */
4909 EXTRACT_NUMBER (mcnt, p + 2);
4910 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4913 /* Originally, this is how many times we HAVE to succeed. */
4918 STORE_NUMBER_AND_INCR (p, mcnt);
4920 DEBUG_PRINT3 (" Setting %p to %d.\n", p - 2, mcnt);
4922 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - 2, mcnt);
4928 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n", p+2);
4930 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4932 p[2] = (unsigned char) no_op;
4933 p[3] = (unsigned char) no_op;
4939 EXTRACT_NUMBER (mcnt, p + 2);
4940 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4942 /* Originally, this is how many times we CAN jump. */
4946 STORE_NUMBER (p + 2, mcnt);
4948 DEBUG_PRINT3 (" Setting %p to %d.\n", p + 2, mcnt);
4950 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + 2, mcnt);
4952 goto unconditional_jump;
4954 /* If don't have to jump any more, skip over the rest of command. */
4961 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4963 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4965 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4967 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
4969 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4971 STORE_NUMBER (p1, mcnt);
4976 /* The DEC Alpha C compiler 3.x generates incorrect code for the
4977 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
4978 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
4979 macro and introducing temporary variables works around the bug. */
4982 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4983 if (AT_WORD_BOUNDARY (d))
4988 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4989 if (AT_WORD_BOUNDARY (d))
4995 boolean prevchar, thischar;
4997 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4998 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5001 prevchar = WORDCHAR_P (d - 1);
5002 thischar = WORDCHAR_P (d);
5003 if (prevchar != thischar)
5010 boolean prevchar, thischar;
5012 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
5013 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5016 prevchar = WORDCHAR_P (d - 1);
5017 thischar = WORDCHAR_P (d);
5018 if (prevchar != thischar)
5025 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
5026 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
5031 DEBUG_PRINT1 ("EXECUTING wordend.\n");
5032 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
5033 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
5039 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
5040 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
5045 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
5046 if (PTR_CHAR_POS ((unsigned char *) d) != point)
5051 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
5052 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
5057 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
5062 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
5066 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5068 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
5070 SET_REGS_MATCHED ();
5074 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
5076 goto matchnotsyntax;
5079 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
5083 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5085 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
5087 SET_REGS_MATCHED ();
5090 #else /* not emacs */
5092 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
5094 if (!WORDCHAR_P (d))
5096 SET_REGS_MATCHED ();
5101 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
5105 SET_REGS_MATCHED ();
5108 #endif /* not emacs */
5113 continue; /* Successfully executed one pattern command; keep going. */
5116 /* We goto here if a matching operation fails. */
5118 if (!FAIL_STACK_EMPTY ())
5119 { /* A restart point is known. Restore to that state. */
5120 DEBUG_PRINT1 ("\nFAIL:\n");
5121 POP_FAILURE_POINT (d, p,
5122 lowest_active_reg, highest_active_reg,
5123 regstart, regend, reg_info);
5125 /* If this failure point is a dummy, try the next one. */
5129 /* If we failed to the end of the pattern, don't examine *p. */
5133 boolean is_a_jump_n = false;
5135 /* If failed to a backwards jump that's part of a repetition
5136 loop, need to pop this failure point and use the next one. */
5137 switch ((re_opcode_t) *p)
5141 case maybe_pop_jump:
5142 case pop_failure_jump:
5145 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5148 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
5150 && (re_opcode_t) *p1 == on_failure_jump))
5158 if (d >= string1 && d <= end1)
5162 break; /* Matching at this starting point really fails. */
5166 goto restore_best_regs;
5170 return -1; /* Failure to match. */
5173 /* Subroutine definitions for re_match_2. */
5176 /* We are passed P pointing to a register number after a start_memory.
5178 Return true if the pattern up to the corresponding stop_memory can
5179 match the empty string, and false otherwise.
5181 If we find the matching stop_memory, sets P to point to one past its number.
5182 Otherwise, sets P to an undefined byte less than or equal to END.
5184 We don't handle duplicates properly (yet). */
5187 group_match_null_string_p (p, end, reg_info)
5188 unsigned char **p, *end;
5189 register_info_type *reg_info;
5192 /* Point to after the args to the start_memory. */
5193 unsigned char *p1 = *p + 2;
5197 /* Skip over opcodes that can match nothing, and return true or
5198 false, as appropriate, when we get to one that can't, or to the
5199 matching stop_memory. */
5201 switch ((re_opcode_t) *p1)
5203 /* Could be either a loop or a series of alternatives. */
5204 case on_failure_jump:
5206 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5208 /* If the next operation is not a jump backwards in the
5213 /* Go through the on_failure_jumps of the alternatives,
5214 seeing if any of the alternatives cannot match nothing.
5215 The last alternative starts with only a jump,
5216 whereas the rest start with on_failure_jump and end
5217 with a jump, e.g., here is the pattern for `a|b|c':
5219 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
5220 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
5223 So, we have to first go through the first (n-1)
5224 alternatives and then deal with the last one separately. */
5227 /* Deal with the first (n-1) alternatives, which start
5228 with an on_failure_jump (see above) that jumps to right
5229 past a jump_past_alt. */
5231 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
5233 /* `mcnt' holds how many bytes long the alternative
5234 is, including the ending `jump_past_alt' and
5237 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
5241 /* Move to right after this alternative, including the
5245 /* Break if it's the beginning of an n-th alternative
5246 that doesn't begin with an on_failure_jump. */
5247 if ((re_opcode_t) *p1 != on_failure_jump)
5250 /* Still have to check that it's not an n-th
5251 alternative that starts with an on_failure_jump. */
5253 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5254 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
5256 /* Get to the beginning of the n-th alternative. */
5262 /* Deal with the last alternative: go back and get number
5263 of the `jump_past_alt' just before it. `mcnt' contains
5264 the length of the alternative. */
5265 EXTRACT_NUMBER (mcnt, p1 - 2);
5267 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
5270 p1 += mcnt; /* Get past the n-th alternative. */
5276 assert (p1[1] == **p);
5282 if (!common_op_match_null_string_p (&p1, end, reg_info))
5285 } /* while p1 < end */
5288 } /* group_match_null_string_p */
5291 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5292 It expects P to be the first byte of a single alternative and END one
5293 byte past the last. The alternative can contain groups. */
5296 alt_match_null_string_p (p, end, reg_info)
5297 unsigned char *p, *end;
5298 register_info_type *reg_info;
5301 unsigned char *p1 = p;
5305 /* Skip over opcodes that can match nothing, and break when we get
5306 to one that can't. */
5308 switch ((re_opcode_t) *p1)
5311 case on_failure_jump:
5313 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5318 if (!common_op_match_null_string_p (&p1, end, reg_info))
5321 } /* while p1 < end */
5324 } /* alt_match_null_string_p */
5327 /* Deals with the ops common to group_match_null_string_p and
5328 alt_match_null_string_p.
5330 Sets P to one after the op and its arguments, if any. */
5333 common_op_match_null_string_p (p, end, reg_info)
5334 unsigned char **p, *end;
5335 register_info_type *reg_info;
5340 unsigned char *p1 = *p;
5342 switch ((re_opcode_t) *p1++)
5362 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5363 ret = group_match_null_string_p (&p1, end, reg_info);
5365 /* Have to set this here in case we're checking a group which
5366 contains a group and a back reference to it. */
5368 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5369 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5375 /* If this is an optimized succeed_n for zero times, make the jump. */
5377 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5385 /* Get to the number of times to succeed. */
5387 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5392 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5400 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5408 /* All other opcodes mean we cannot match the empty string. */
5414 } /* common_op_match_null_string_p */
5417 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5418 bytes; nonzero otherwise. */
5421 bcmp_translate (s1, s2, len, translate)
5422 const char *s1, *s2;
5424 RE_TRANSLATE_TYPE translate;
5426 register const unsigned char *p1 = (const unsigned char *) s1;
5427 register const unsigned char *p2 = (const unsigned char *) s2;
5430 if (translate[*p1++] != translate[*p2++]) return 1;
5436 /* Entry points for GNU code. */
5438 /* re_compile_pattern is the GNU regular expression compiler: it
5439 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5440 Returns 0 if the pattern was valid, otherwise an error string.
5442 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5443 are set in BUFP on entry.
5445 We call regex_compile to do the actual compilation. */
5448 re_compile_pattern (pattern, length, bufp)
5449 const char *pattern;
5451 struct re_pattern_buffer *bufp;
5455 /* GNU code is written to assume at least RE_NREGS registers will be set
5456 (and at least one extra will be -1). */
5457 bufp->regs_allocated = REGS_UNALLOCATED;
5459 /* And GNU code determines whether or not to get register information
5460 by passing null for the REGS argument to re_match, etc., not by
5464 /* Match anchors at newline. */
5465 bufp->newline_anchor = 1;
5467 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5471 return gettext (re_error_msgid[(int) ret]);
5474 weak_alias (__re_compile_pattern, re_compile_pattern)
5477 /* Entry points compatible with 4.2 BSD regex library. We don't define
5478 them unless specifically requested. */
5480 #if defined _REGEX_RE_COMP || defined _LIBC
5482 /* BSD has one and only one pattern buffer. */
5483 static struct re_pattern_buffer re_comp_buf;
5487 /* Make these definitions weak in libc, so POSIX programs can redefine
5488 these names if they don't use our functions, and still use
5489 regcomp/regexec below without link errors. */
5499 if (!re_comp_buf.buffer)
5500 return gettext ("No previous regular expression");
5504 if (!re_comp_buf.buffer)
5506 re_comp_buf.buffer = (unsigned char *) malloc (200);
5507 if (re_comp_buf.buffer == NULL)
5508 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
5509 re_comp_buf.allocated = 200;
5511 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5512 if (re_comp_buf.fastmap == NULL)
5513 return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
5516 /* Since `re_exec' always passes NULL for the `regs' argument, we
5517 don't need to initialize the pattern buffer fields which affect it. */
5519 /* Match anchors at newlines. */
5520 re_comp_buf.newline_anchor = 1;
5522 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5527 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5528 return (char *) gettext (re_error_msgid[(int) ret]);
5539 const int len = strlen (s);
5541 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5544 #endif /* _REGEX_RE_COMP */
5546 /* POSIX.2 functions. Don't define these for Emacs. */
5550 /* regcomp takes a regular expression as a string and compiles it.
5552 PREG is a regex_t *. We do not expect any fields to be initialized,
5553 since POSIX says we shouldn't. Thus, we set
5555 `buffer' to the compiled pattern;
5556 `used' to the length of the compiled pattern;
5557 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5558 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5559 RE_SYNTAX_POSIX_BASIC;
5560 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5561 `fastmap' and `fastmap_accurate' to zero;
5562 `re_nsub' to the number of subexpressions in PATTERN.
5564 PATTERN is the address of the pattern string.
5566 CFLAGS is a series of bits which affect compilation.
5568 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5569 use POSIX basic syntax.
5571 If REG_NEWLINE is set, then . and [^...] don't match newline.
5572 Also, regexec will try a match beginning after every newline.
5574 If REG_ICASE is set, then we considers upper- and lowercase
5575 versions of letters to be equivalent when matching.
5577 If REG_NOSUB is set, then when PREG is passed to regexec, that
5578 routine will report only success or failure, and nothing about the
5581 It returns 0 if it succeeds, nonzero if it doesn't. (See gnu-regex.h for
5582 the return codes and their meanings.) */
5585 regcomp (preg, pattern, cflags)
5587 const char *pattern;
5592 = (cflags & REG_EXTENDED) ?
5593 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5595 /* regex_compile will allocate the space for the compiled pattern. */
5597 preg->allocated = 0;
5600 /* Don't bother to use a fastmap when searching. This simplifies the
5601 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
5602 characters after newlines into the fastmap. This way, we just try
5606 if (cflags & REG_ICASE)
5611 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
5612 * sizeof (*(RE_TRANSLATE_TYPE)0));
5613 if (preg->translate == NULL)
5614 return (int) REG_ESPACE;
5616 /* Map uppercase characters to corresponding lowercase ones. */
5617 for (i = 0; i < CHAR_SET_SIZE; i++)
5618 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
5621 preg->translate = NULL;
5623 /* If REG_NEWLINE is set, newlines are treated differently. */
5624 if (cflags & REG_NEWLINE)
5625 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5626 syntax &= ~RE_DOT_NEWLINE;
5627 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5628 /* It also changes the matching behavior. */
5629 preg->newline_anchor = 1;
5632 preg->newline_anchor = 0;
5634 preg->no_sub = !!(cflags & REG_NOSUB);
5636 /* POSIX says a null character in the pattern terminates it, so we
5637 can use strlen here in compiling the pattern. */
5638 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5640 /* POSIX doesn't distinguish between an unmatched open-group and an
5641 unmatched close-group: both are REG_EPAREN. */
5642 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5647 weak_alias (__regcomp, regcomp)
5651 /* regexec searches for a given pattern, specified by PREG, in the
5654 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5655 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5656 least NMATCH elements, and we set them to the offsets of the
5657 corresponding matched substrings.
5659 EFLAGS specifies `execution flags' which affect matching: if
5660 REG_NOTBOL is set, then ^ does not match at the beginning of the
5661 string; if REG_NOTEOL is set, then $ does not match at the end.
5663 We return 0 if we find a match and REG_NOMATCH if not. */
5666 regexec (preg, string, nmatch, pmatch, eflags)
5667 const regex_t *preg;
5670 regmatch_t pmatch[];
5674 struct re_registers regs;
5675 regex_t private_preg;
5676 int len = strlen (string);
5677 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5679 private_preg = *preg;
5681 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5682 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5684 /* The user has told us exactly how many registers to return
5685 information about, via `nmatch'. We have to pass that on to the
5686 matching routines. */
5687 private_preg.regs_allocated = REGS_FIXED;
5691 regs.num_regs = nmatch;
5692 regs.start = TALLOC (nmatch, regoff_t);
5693 regs.end = TALLOC (nmatch, regoff_t);
5694 if (regs.start == NULL || regs.end == NULL)
5695 return (int) REG_NOMATCH;
5698 /* Perform the searching operation. */
5699 ret = re_search (&private_preg, string, len,
5700 /* start: */ 0, /* range: */ len,
5701 want_reg_info ? ®s : (struct re_registers *) 0);
5703 /* Copy the register information to the POSIX structure. */
5710 for (r = 0; r < nmatch; r++)
5712 pmatch[r].rm_so = regs.start[r];
5713 pmatch[r].rm_eo = regs.end[r];
5717 /* If we needed the temporary register info, free the space now. */
5722 /* We want zero return to mean success, unlike `re_search'. */
5723 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5726 weak_alias (__regexec, regexec)
5730 /* Returns a message corresponding to an error code, ERRCODE, returned
5731 from either regcomp or regexec. We don't use PREG here. */
5734 regerror (errcode, preg, errbuf, errbuf_size)
5736 const regex_t *preg;
5744 || errcode >= (int) (sizeof (re_error_msgid)
5745 / sizeof (re_error_msgid[0])))
5746 /* Only error codes returned by the rest of the code should be passed
5747 to this routine. If we are given anything else, or if other regex
5748 code generates an invalid error code, then the program has a bug.
5749 Dump core so we can fix it. */
5752 msg = gettext (re_error_msgid[errcode]);
5754 msg_size = strlen (msg) + 1; /* Includes the null. */
5756 if (errbuf_size != 0)
5758 if (msg_size > errbuf_size)
5760 #if defined HAVE_MEMPCPY || defined _LIBC
5761 *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
5763 memcpy (errbuf, msg, errbuf_size - 1);
5764 errbuf[errbuf_size - 1] = 0;
5768 memcpy (errbuf, msg, msg_size);
5774 weak_alias (__regerror, regerror)
5778 /* Free dynamically allocated space used by PREG. */
5784 if (preg->buffer != NULL)
5785 free (preg->buffer);
5786 preg->buffer = NULL;
5788 preg->allocated = 0;
5791 if (preg->fastmap != NULL)
5792 free (preg->fastmap);
5793 preg->fastmap = NULL;
5794 preg->fastmap_accurate = 0;
5796 if (preg->translate != NULL)
5797 free (preg->translate);
5798 preg->translate = NULL;
5801 weak_alias (__regfree, regfree)
5804 #endif /* not emacs */
projects / platform / upstream / binutils.git / blob