| blob | 522b0eedf442743348c82a3c37b3ac1c54a7134d |
1 /* Extended regular expression matching and search library,
2 version 0.12.
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
5 Copyright (C) 1993-1999, 2000, 2001, 2002 Free Software Foundation, Inc.
6 This file is part of the GNU C Library.
8 The GNU C Library is free software; you can redistribute it and/or
9 modify it under the terms of the GNU Lesser General Public
10 License as published by the Free Software Foundation; either
11 version 2.1 of the License, or (at your option) any later version.
13 The GNU C Library is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 Lesser General Public License for more details.
18 You should have received a copy of the GNU Lesser General Public
19 License along with the GNU C Library; if not, write to the Free
20 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
21 02111-1307 USA. */
23 /* This file has been modified for usage in libiberty. It includes "xregex.h"
24 instead of <regex.h>. The "xregex.h" header file renames all external
25 routines with an "x" prefix so they do not collide with the native regex
26 routines or with other components regex routines. */
27 /* AIX requires this to be the first thing in the file. */
28 #if defined _AIX && !defined __GNUC__ && !defined REGEX_MALLOC
29 #pragma alloca
30 #endif
32 #undef _GNU_SOURCE
33 #define _GNU_SOURCE
35 #ifdef HAVE_CONFIG_H
36 # include <config.h>
37 #endif
39 #include <ansidecl.h>
41 #ifndef PARAMS
42 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
43 # define PARAMS(args) args
44 # else
45 # define PARAMS(args) ()
49 #ifndef INSIDE_RECURSION
51 # if defined STDC_HEADERS && !defined emacs
52 # include <stddef.h>
53 # else
54 /* We need this for `regex.h', and perhaps for the Emacs include files. */
55 # include <sys/types.h>
56 # endif
58 # define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
60 /* For platform which support the ISO C amendement 1 functionality we
61 support user defined character classes. */
62 # if defined _LIBC || WIDE_CHAR_SUPPORT
63 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
64 # include <wchar.h>
65 # include <wctype.h>
66 # endif
68 # ifdef _LIBC
69 /* We have to keep the namespace clean. */
70 # define regfree(preg) __regfree (preg)
71 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
72 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
73 # define regerror(errcode, preg, errbuf, errbuf_size) \
74 __regerror(errcode, preg, errbuf, errbuf_size)
75 # define re_set_registers(bu, re, nu, st, en) \
76 __re_set_registers (bu, re, nu, st, en)
77 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
78 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
79 # define re_match(bufp, string, size, pos, regs) \
80 __re_match (bufp, string, size, pos, regs)
81 # define re_search(bufp, string, size, startpos, range, regs) \
82 __re_search (bufp, string, size, startpos, range, regs)
83 # define re_compile_pattern(pattern, length, bufp) \
84 __re_compile_pattern (pattern, length, bufp)
85 # define re_set_syntax(syntax) __re_set_syntax (syntax)
86 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
87 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
88 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
90 # define btowc __btowc
92 /* We are also using some library internals. */
93 # include <locale/localeinfo.h>
94 # include <locale/elem-hash.h>
95 # include <langinfo.h>
96 # include <locale/coll-lookup.h>
97 # endif
99 /* This is for other GNU distributions with internationalized messages. */
100 # if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
101 # include <libintl.h>
102 # ifdef _LIBC
103 # undef gettext
105 # endif
106 # else
107 # define gettext(msgid) (msgid)
108 # endif
110 # ifndef gettext_noop
111 /* This define is so xgettext can find the internationalizable
112 strings. */
113 # define gettext_noop(String) String
114 # endif
116 /* The `emacs' switch turns on certain matching commands
117 that make sense only in Emacs. */
118 # ifdef emacs
126 /* If we are not linking with Emacs proper,
127 we can't use the relocating allocator
128 even if config.h says that we can. */
129 # undef REL_ALLOC
131 # if defined STDC_HEADERS || defined _LIBC
132 # include <stdlib.h>
133 # else
136 # endif
138 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
139 If nothing else has been done, use the method below. */
140 # ifdef INHIBIT_STRING_HEADER
141 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
142 # if !defined bzero && !defined bcopy
143 # undef INHIBIT_STRING_HEADER
144 # endif
145 # endif
146 # endif
148 /* This is the normal way of making sure we have a bcopy and a bzero.
149 This is used in most programs--a few other programs avoid this
150 by defining INHIBIT_STRING_HEADER. */
151 # ifndef INHIBIT_STRING_HEADER
152 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
153 # include <string.h>
154 # ifndef bzero
155 # ifndef _LIBC
157 # else
158 # define bzero(s, n) __bzero (s, n)
159 # endif
160 # endif
161 # else
162 # include <strings.h>
163 # ifndef memcmp
164 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
165 # endif
166 # ifndef memcpy
167 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
168 # endif
169 # endif
170 # endif
172 /* Define the syntax stuff for \<, \>, etc. */
174 /* This must be nonzero for the wordchar and notwordchar pattern
175 commands in re_match_2. */
176 # ifndef Sword
177 # define Sword 1
178 # endif
180 # ifdef SWITCH_ENUM_BUG
181 # define SWITCH_ENUM_CAST(x) ((int)(x))
182 # else
183 # define SWITCH_ENUM_CAST(x) (x)
184 # endif
188 # if defined _LIBC || HAVE_LIMITS_H
189 # include <limits.h>
190 # endif
192 # ifndef MB_LEN_MAX
193 # define MB_LEN_MAX 1
194 # endif
195 \f
196 /* Get the interface, including the syntax bits. */
199 /* isalpha etc. are used for the character classes. */
200 # include <ctype.h>
202 /* Jim Meyering writes:
204 "... Some ctype macros are valid only for character codes that
205 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
206 using /bin/cc or gcc but without giving an ansi option). So, all
207 ctype uses should be through macros like ISPRINT... If
208 STDC_HEADERS is defined, then autoconf has verified that the ctype
209 macros don't need to be guarded with references to isascii. ...
210 Defining isascii to 1 should let any compiler worth its salt
211 eliminate the && through constant folding."
212 Solaris defines some of these symbols so we must undefine them first. */
214 # undef ISASCII
215 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
216 # define ISASCII(c) 1
217 # else
218 # define ISASCII(c) isascii(c)
219 # endif
221 # ifdef isblank
222 # define ISBLANK(c) (ISASCII (c) && isblank (c))
223 # else
225 # endif
226 # ifdef isgraph
227 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
228 # else
229 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
230 # endif
232 # undef ISPRINT
233 # define ISPRINT(c) (ISASCII (c) && isprint (c))
234 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
235 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
236 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
237 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
238 # define ISLOWER(c) (ISASCII (c) && islower (c))
239 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
240 # define ISSPACE(c) (ISASCII (c) && isspace (c))
241 # define ISUPPER(c) (ISASCII (c) && isupper (c))
242 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
244 # ifdef _tolower
245 # define TOLOWER(c) _tolower(c)
246 # else
247 # define TOLOWER(c) tolower(c)
248 # endif
250 # ifndef NULL
251 # define NULL (void *)0
252 # endif
254 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
255 since ours (we hope) works properly with all combinations of
256 machines, compilers, `char' and `unsigned char' argument types.
257 (Per Bothner suggested the basic approach.) */
258 # undef SIGN_EXTEND_CHAR
259 # if __STDC__
260 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
262 /* As in Harbison and Steele. */
263 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
264 # endif
265 \f
266 # ifndef emacs
267 /* How many characters in the character set. */
268 # define CHAR_SET_SIZE 256
270 # ifdef SYNTAX_TABLE
280 static void
282 {
297 }
301 # define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
304 \f
305 /* Integer type for pointers. */
306 # if !defined _LIBC && !defined HAVE_UINTPTR_T
308 # endif
310 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
311 use `alloca' instead of `malloc'. This is because using malloc in
312 re_search* or re_match* could cause memory leaks when C-g is used in
313 Emacs; also, malloc is slower and causes storage fragmentation. On
314 the other hand, malloc is more portable, and easier to debug.
316 Because we sometimes use alloca, some routines have to be macros,
317 not functions -- `alloca'-allocated space disappears at the end of the
318 function it is called in. */
320 # ifdef REGEX_MALLOC
322 # define REGEX_ALLOCATE malloc
323 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
324 # define REGEX_FREE free
328 /* Emacs already defines alloca, sometimes. */
329 # ifndef alloca
331 /* Make alloca work the best possible way. */
332 # ifdef __GNUC__
333 # define alloca __builtin_alloca
335 # if HAVE_ALLOCA_H
336 # include <alloca.h>
342 # define REGEX_ALLOCATE alloca
344 /* Assumes a `char *destination' variable. */
345 # define REGEX_REALLOCATE(source, osize, nsize) \
346 (destination = (char *) alloca (nsize), \
347 memcpy (destination, source, osize))
349 /* No need to do anything to free, after alloca. */
354 /* Define how to allocate the failure stack. */
356 # if defined REL_ALLOC && defined REGEX_MALLOC
358 # define REGEX_ALLOCATE_STACK(size) \
359 r_alloc (&failure_stack_ptr, (size))
360 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
361 r_re_alloc (&failure_stack_ptr, (nsize))
362 # define REGEX_FREE_STACK(ptr) \
363 r_alloc_free (&failure_stack_ptr)
367 # ifdef REGEX_MALLOC
369 # define REGEX_ALLOCATE_STACK malloc
370 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
371 # define REGEX_FREE_STACK free
375 # define REGEX_ALLOCATE_STACK alloca
377 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
378 REGEX_REALLOCATE (source, osize, nsize)
379 /* No need to explicitly free anything. */
380 # define REGEX_FREE_STACK(arg)
386 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
387 `string1' or just past its end. This works if PTR is NULL, which is
388 a good thing. */
389 # define FIRST_STRING_P(ptr) \
390 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
392 /* (Re)Allocate N items of type T using malloc, or fail. */
393 # define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
394 # define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
395 # define RETALLOC_IF(addr, n, t) \
396 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
397 # define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
401 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
403 # undef MAX
404 # undef MIN
405 # define MAX(a, b) ((a) > (b) ? (a) : (b))
406 # define MIN(a, b) ((a) < (b) ? (a) : (b))
409 # define false 0
410 # define true 1
413 reg_syntax_t syntax,
429 #ifdef MBS_SUPPORT
431 reg_syntax_t syntax,
450 #endif
451 \f
452 /* These are the command codes that appear in compiled regular
453 expressions. Some opcodes are followed by argument bytes. A
454 command code can specify any interpretation whatsoever for its
455 arguments. Zero bytes may appear in the compiled regular expression. */
458 {
461 /* Succeed right away--no more backtracking. */
462 succeed,
464 /* Followed by one byte giving n, then by n literal bytes. */
465 exactn,
467 # ifdef MBS_SUPPORT
468 /* Same as exactn, but contains binary data. */
469 exactn_bin,
470 # endif
472 /* Matches any (more or less) character. */
473 anychar,
475 /* Matches any one char belonging to specified set. First
476 following byte is number of bitmap bytes. Then come bytes
477 for a bitmap saying which chars are in. Bits in each byte
478 are ordered low-bit-first. A character is in the set if its
479 bit is 1. A character too large to have a bit in the map is
480 automatically not in the set. */
481 /* ifdef MBS_SUPPORT, following element is length of character
482 classes, length of collating symbols, length of equivalence
483 classes, length of character ranges, and length of characters.
484 Next, character class element, collating symbols elements,
485 equivalence class elements, range elements, and character
486 elements follow.
487 See regex_compile function. */
488 charset,
490 /* Same parameters as charset, but match any character that is
491 not one of those specified. */
492 charset_not,
494 /* Start remembering the text that is matched, for storing in a
495 register. Followed by one byte with the register number, in
496 the range 0 to one less than the pattern buffer's re_nsub
497 field. Then followed by one byte with the number of groups
498 inner to this one. (This last has to be part of the
499 start_memory only because we need it in the on_failure_jump
500 of re_match_2.) */
501 start_memory,
503 /* Stop remembering the text that is matched and store it in a
504 memory register. Followed by one byte with the register
505 number, in the range 0 to one less than `re_nsub' in the
506 pattern buffer, and one byte with the number of inner groups,
507 just like `start_memory'. (We need the number of inner
508 groups here because we don't have any easy way of finding the
509 corresponding start_memory when we're at a stop_memory.) */
510 stop_memory,
512 /* Match a duplicate of something remembered. Followed by one
513 byte containing the register number. */
514 duplicate,
516 /* Fail unless at beginning of line. */
517 begline,
519 /* Fail unless at end of line. */
520 endline,
522 /* Succeeds if at beginning of buffer (if emacs) or at beginning
523 of string to be matched (if not). */
524 begbuf,
526 /* Analogously, for end of buffer/string. */
527 endbuf,
529 /* Followed by two byte relative address to which to jump. */
530 jump,
532 /* Same as jump, but marks the end of an alternative. */
533 jump_past_alt,
535 /* Followed by two-byte relative address of place to resume at
536 in case of failure. */
537 /* ifdef MBS_SUPPORT, the size of address is 1. */
538 on_failure_jump,
540 /* Like on_failure_jump, but pushes a placeholder instead of the
541 current string position when executed. */
542 on_failure_keep_string_jump,
544 /* Throw away latest failure point and then jump to following
545 two-byte relative address. */
546 /* ifdef MBS_SUPPORT, the size of address is 1. */
547 pop_failure_jump,
549 /* Change to pop_failure_jump if know won't have to backtrack to
550 match; otherwise change to jump. This is used to jump
551 back to the beginning of a repeat. If what follows this jump
552 clearly won't match what the repeat does, such that we can be
553 sure that there is no use backtracking out of repetitions
554 already matched, then we change it to a pop_failure_jump.
555 Followed by two-byte address. */
556 /* ifdef MBS_SUPPORT, the size of address is 1. */
557 maybe_pop_jump,
559 /* Jump to following two-byte address, and push a dummy failure
560 point. This failure point will be thrown away if an attempt
561 is made to use it for a failure. A `+' construct makes this
562 before the first repeat. Also used as an intermediary kind
563 of jump when compiling an alternative. */
564 /* ifdef MBS_SUPPORT, the size of address is 1. */
565 dummy_failure_jump,
567 /* Push a dummy failure point and continue. Used at the end of
568 alternatives. */
569 push_dummy_failure,
571 /* Followed by two-byte relative address and two-byte number n.
572 After matching N times, jump to the address upon failure. */
573 /* ifdef MBS_SUPPORT, the size of address is 1. */
574 succeed_n,
576 /* Followed by two-byte relative address, and two-byte number n.
577 Jump to the address N times, then fail. */
578 /* ifdef MBS_SUPPORT, the size of address is 1. */
579 jump_n,
581 /* Set the following two-byte relative address to the
582 subsequent two-byte number. The address *includes* the two
583 bytes of number. */
584 /* ifdef MBS_SUPPORT, the size of address is 1. */
585 set_number_at,
594 notwordbound /* Succeeds if not at a word boundary. */
596 # ifdef emacs
601 /* Matches any character whose syntax is specified. Followed by
602 a byte which contains a syntax code, e.g., Sword. */
603 syntaxspec,
605 /* Matches any character whose syntax is not that specified. */
606 notsyntaxspec
610 \f
612 #ifdef BYTE
613 # define CHAR_T char
614 # define UCHAR_T unsigned char
615 # define COMPILED_BUFFER_VAR bufp->buffer
616 # define OFFSET_ADDRESS_SIZE 2
617 # if defined (__STDC__) || defined (ALMOST_STDC) || defined (HAVE_STRINGIZE)
618 # define PREFIX(name) byte_##name
619 # else
621 # endif
622 # define ARG_PREFIX(name) name
623 # define PUT_CHAR(c) putchar (c)
624 #else
625 # ifdef WCHAR
626 # define CHAR_T wchar_t
627 # define UCHAR_T wchar_t
628 # define COMPILED_BUFFER_VAR wc_buffer
630 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
631 # if defined (__STDC__) || defined (ALMOST_STDC) || defined (HAVE_STRINGIZE)
632 # define PREFIX(name) wcs_##name
633 # define ARG_PREFIX(name) c##name
634 # else
637 # endif
638 /* Should we use wide stream?? */
640 # define TRUE 1
641 # define FALSE 0
642 # else
643 # ifdef MBS_SUPPORT
644 # define WCHAR
645 # define INSIDE_RECURSION
647 # undef INSIDE_RECURSION
648 # endif
649 # define BYTE
650 # define INSIDE_RECURSION
652 # undef INSIDE_RECURSION
653 # endif
654 #endif
656 #ifdef INSIDE_RECURSION
657 /* Common operations on the compiled pattern. */
659 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
660 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
662 # ifdef WCHAR
663 # define STORE_NUMBER(destination, number) \
664 do { \
665 *(destination) = (UCHAR_T)(number); \
666 } while (0)
668 # define STORE_NUMBER(destination, number) \
669 do { \
670 (destination)[0] = (number) & 0377; \
671 (destination)[1] = (number) >> 8; \
672 } while (0)
675 /* Same as STORE_NUMBER, except increment DESTINATION to
676 the byte after where the number is stored. Therefore, DESTINATION
677 must be an lvalue. */
678 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
680 # define STORE_NUMBER_AND_INCR(destination, number) \
681 do { \
682 STORE_NUMBER (destination, number); \
683 (destination) += OFFSET_ADDRESS_SIZE; \
684 } while (0)
686 /* Put into DESTINATION a number stored in two contiguous bytes starting
687 at SOURCE. */
688 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
690 # ifdef WCHAR
691 # define EXTRACT_NUMBER(destination, source) \
692 do { \
693 (destination) = *(source); \
694 } while (0)
696 # define EXTRACT_NUMBER(destination, source) \
697 do { \
698 (destination) = *(source) & 0377; \
699 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
700 } while (0)
701 # endif
703 # ifdef DEBUG
705 static void
709 {
710 # ifdef WCHAR
716 # endif
717 }
720 # undef EXTRACT_NUMBER
721 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
726 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
727 SOURCE must be an lvalue. */
729 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
730 do { \
731 EXTRACT_NUMBER (destination, source); \
732 (source) += OFFSET_ADDRESS_SIZE; \
733 } while (0)
735 # ifdef DEBUG
738 static void
742 {
745 }
747 # ifndef EXTRACT_MACROS
748 # undef EXTRACT_NUMBER_AND_INCR
749 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
750 PREFIX(extract_number_and_incr) (&dest, &src)
755 \f
757 /* If DEBUG is defined, Regex prints many voluminous messages about what
758 it is doing (if the variable `debug' is nonzero). If linked with the
759 main program in `iregex.c', you can enter patterns and strings
760 interactively. And if linked with the main program in `main.c' and
761 the other test files, you can run the already-written tests. */
763 # ifdef DEBUG
765 # ifndef DEFINED_ONCE
767 /* We use standard I/O for debugging. */
768 # include <stdio.h>
770 /* It is useful to test things that ``must'' be true when debugging. */
771 # include <assert.h>
775 # define DEBUG_STATEMENT(e) e
776 # define DEBUG_PRINT1(x) if (debug) printf (x)
777 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
778 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
779 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
782 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
783 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
784 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
785 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
788 /* Print the fastmap in human-readable form. */
790 # ifndef DEFINED_ONCE
791 void
794 {
799 {
801 {
805 {
807 i++;
808 }
810 {
813 }
814 }
815 }
817 }
821 /* Print a compiled pattern string in human-readable form, starting at
822 the START pointer into it and ending just before the pointer END. */
824 void
828 {
835 {
838 }
840 /* Loop over pattern commands. */
842 {
843 # ifdef _LIBC
845 # else
847 # endif
850 {
858 do
859 {
862 }
866 # ifdef MBS_SUPPORT
870 do
871 {
873 }
898 {
899 # ifdef WCHAR
910 {
916 }
919 {
925 }
928 {
932 }
937 # else
949 {
950 /* Are we starting a range? */
952 {
955 }
956 /* Have we broken a range? */
958 {
961 }
967 }
976 }
989 # ifdef _LIBC
991 # else
993 # endif
998 # ifdef _LIBC
1000 # else
1003 # endif
1008 # ifdef _LIBC
1010 # else
1012 # endif
1021 # ifdef _LIBC
1023 # else
1025 # endif
1030 # ifdef _LIBC
1032 # else
1034 # endif
1039 # ifdef _LIBC
1041 # else
1043 # endif
1048 # ifdef _LIBC
1050 # else
1052 # endif
1059 # ifdef _LIBC
1061 # else
1064 # endif
1078 # ifdef _LIBC
1080 # else
1083 # endif
1102 # ifdef emacs
1146 }
1149 }
1151 # ifdef _LIBC
1153 # else
1155 # endif
1156 }
1159 void
1162 {
1171 {
1174 }
1176 # ifdef _LIBC
1178 # else
1180 # endif
1188 /* Perhaps we should print the translate table? */
1189 }
1192 void
1199 {
1204 else
1205 {
1209 {
1214 }
1218 {
1221 {
1224 }
1225 }
1226 }
1227 }
1229 # ifndef DEFINED_ONCE
1230 void
1233 {
1235 }
1236 # endif
1240 # ifndef DEFINED_ONCE
1241 # undef assert
1242 # define assert(e)
1244 # define DEBUG_STATEMENT(e)
1245 # define DEBUG_PRINT1(x)
1246 # define DEBUG_PRINT2(x1, x2)
1247 # define DEBUG_PRINT3(x1, x2, x3)
1248 # define DEBUG_PRINT4(x1, x2, x3, x4)
1250 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1251 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1255 \f
1257 # ifdef WCHAR
1258 /* This convert a multibyte string to a wide character string.
1259 And write their correspondances to offset_buffer(see below)
1260 and write whether each wchar_t is binary data to is_binary.
1261 This assume invalid multibyte sequences as binary data.
1262 We assume offset_buffer and is_binary is already allocated
1263 enough space. */
1268 static size_t
1274 /* It hold correspondances between src(char string) and
1275 dest(wchar_t string) for optimization.
1276 e.g. src = "xxxyzz"
1277 dest = {'X', 'Y', 'Z'}
1278 (each "xxx", "y" and "zz" represent one multibyte character
1279 corresponding to 'X', 'Y' and 'Z'.)
1280 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1281 = {0, 3, 4, 6}
1282 */
1285 {
1295 /* Initialize the conversion state. */
1301 {
1302 #ifdef _LIBC
1304 #else
1306 #endif
1309 /* failed to convert. maybe src contains binary data.
1310 So we consume 1 byte manualy. */
1311 {
1315 }
1316 else
1318 /* In sjis encoding, we use yen sign as escape character in
1319 place of reverse solidus. So we convert 0x5c(yen sign in
1320 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1321 solidus in UCS2). */
1326 }
1328 /* Fill remain of the buffer with sentinel. */
1333 }
1339 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1340 also be assigned to arbitrarily: each pattern buffer stores its own
1341 syntax, so it can be changed between regex compilations. */
1342 /* This has no initializer because initialized variables in Emacs
1343 become read-only after dumping. */
1344 reg_syntax_t re_syntax_options;
1347 /* Specify the precise syntax of regexps for compilation. This provides
1348 for compatibility for various utilities which historically have
1349 different, incompatible syntaxes.
1351 The argument SYNTAX is a bit mask comprised of the various bits
1352 defined in regex.h. We return the old syntax. */
1354 reg_syntax_t
1356 reg_syntax_t syntax;
1357 {
1361 # ifdef DEBUG
1368 }
1369 # ifdef _LIBC
1371 # endif
1372 \f
1373 /* This table gives an error message for each of the error codes listed
1374 in regex.h. Obviously the order here has to be same as there.
1375 POSIX doesn't require that we do anything for REG_NOERROR,
1376 but why not be nice? */
1379 {
1397 };
1398 \f
1401 #ifndef DEFINED_ONCE
1402 /* Avoiding alloca during matching, to placate r_alloc. */
1404 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1405 searching and matching functions should not call alloca. On some
1406 systems, alloca is implemented in terms of malloc, and if we're
1407 using the relocating allocator routines, then malloc could cause a
1408 relocation, which might (if the strings being searched are in the
1409 ralloc heap) shift the data out from underneath the regexp
1410 routines.
1412 Here's another reason to avoid allocation: Emacs
1413 processes input from X in a signal handler; processing X input may
1414 call malloc; if input arrives while a matching routine is calling
1415 malloc, then we're scrod. But Emacs can't just block input while
1416 calling matching routines; then we don't notice interrupts when
1417 they come in. So, Emacs blocks input around all regexp calls
1418 except the matching calls, which it leaves unprotected, in the
1419 faith that they will not malloc. */
1421 /* Normally, this is fine. */
1422 # define MATCH_MAY_ALLOCATE
1424 /* When using GNU C, we are not REALLY using the C alloca, no matter
1425 what config.h may say. So don't take precautions for it. */
1426 # ifdef __GNUC__
1427 # undef C_ALLOCA
1428 # endif
1430 /* The match routines may not allocate if (1) they would do it with malloc
1431 and (2) it's not safe for them to use malloc.
1432 Note that if REL_ALLOC is defined, matching would not use malloc for the
1433 failure stack, but we would still use it for the register vectors;
1434 so REL_ALLOC should not affect this. */
1435 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1436 # undef MATCH_MAY_ALLOCATE
1437 # endif
1439 \f
1440 #ifdef INSIDE_RECURSION
1441 /* Failure stack declarations and macros; both re_compile_fastmap and
1442 re_match_2 use a failure stack. These have to be macros because of
1443 REGEX_ALLOCATE_STACK. */
1446 /* Number of failure points for which to initially allocate space
1447 when matching. If this number is exceeded, we allocate more
1448 space, so it is not a hard limit. */
1449 # ifndef INIT_FAILURE_ALLOC
1450 # define INIT_FAILURE_ALLOC 5
1451 # endif
1453 /* Roughly the maximum number of failure points on the stack. Would be
1454 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1455 This is a variable only so users of regex can assign to it; we never
1456 change it ourselves. */
1458 # ifdef INT_IS_16BIT
1460 # ifndef DEFINED_ONCE
1461 # if defined MATCH_MAY_ALLOCATE
1462 /* 4400 was enough to cause a crash on Alpha OSF/1,
1463 whose default stack limit is 2mb. */
1465 # else
1467 # endif
1468 # endif
1471 {
1474 };
1479 {
1487 # ifndef DEFINED_ONCE
1488 # if defined MATCH_MAY_ALLOCATE
1489 /* 4400 was enough to cause a crash on Alpha OSF/1,
1490 whose default stack limit is 2mb. */
1492 # else
1494 # endif
1495 # endif
1498 {
1501 };
1506 {
1514 # ifndef DEFINED_ONCE
1515 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1516 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1517 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1518 # endif
1521 /* Define macros to initialize and free the failure stack.
1522 Do `return -2' if the alloc fails. */
1524 # ifdef MATCH_MAY_ALLOCATE
1525 # define INIT_FAIL_STACK() \
1526 do { \
1527 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1528 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1529 \
1530 if (fail_stack.stack == NULL) \
1531 return -2; \
1532 \
1533 fail_stack.size = INIT_FAILURE_ALLOC; \
1534 fail_stack.avail = 0; \
1535 } while (0)
1537 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1538 # else
1539 # define INIT_FAIL_STACK() \
1540 do { \
1541 fail_stack.avail = 0; \
1542 } while (0)
1544 # define RESET_FAIL_STACK()
1545 # endif
1548 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1550 Return 1 if succeeds, and 0 if either ran out of memory
1551 allocating space for it or it was already too large.
1553 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1555 # define DOUBLE_FAIL_STACK(fail_stack) \
1556 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1557 ? 0 \
1558 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1559 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1560 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1561 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1562 \
1563 (fail_stack).stack == NULL \
1564 ? 0 \
1565 : ((fail_stack).size <<= 1, \
1566 1)))
1569 /* Push pointer POINTER on FAIL_STACK.
1570 Return 1 if was able to do so and 0 if ran out of memory allocating
1571 space to do so. */
1572 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1573 ((FAIL_STACK_FULL () \
1574 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1575 ? 0 \
1576 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1577 1))
1579 /* Push a pointer value onto the failure stack.
1580 Assumes the variable `fail_stack'. Probably should only
1581 be called from within `PUSH_FAILURE_POINT'. */
1582 # define PUSH_FAILURE_POINTER(item) \
1583 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1585 /* This pushes an integer-valued item onto the failure stack.
1586 Assumes the variable `fail_stack'. Probably should only
1587 be called from within `PUSH_FAILURE_POINT'. */
1588 # define PUSH_FAILURE_INT(item) \
1589 fail_stack.stack[fail_stack.avail++].integer = (item)
1591 /* Push a fail_stack_elt_t value onto the failure stack.
1592 Assumes the variable `fail_stack'. Probably should only
1593 be called from within `PUSH_FAILURE_POINT'. */
1594 # define PUSH_FAILURE_ELT(item) \
1595 fail_stack.stack[fail_stack.avail++] = (item)
1597 /* These three POP... operations complement the three PUSH... operations.
1598 All assume that `fail_stack' is nonempty. */
1599 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1600 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1601 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1603 /* Used to omit pushing failure point id's when we're not debugging. */
1604 # ifdef DEBUG
1605 # define DEBUG_PUSH PUSH_FAILURE_INT
1606 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1607 # else
1608 # define DEBUG_PUSH(item)
1609 # define DEBUG_POP(item_addr)
1610 # endif
1613 /* Push the information about the state we will need
1614 if we ever fail back to it.
1616 Requires variables fail_stack, regstart, regend, reg_info, and
1617 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1618 be declared.
1620 Does `return FAILURE_CODE' if runs out of memory. */
1622 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1623 do { \
1624 char *destination; \
1625 /* Must be int, so when we don't save any registers, the arithmetic \
1627 /* Can't be int, since there is not a shred of a guarantee that int \
1628 is wide enough to hold a value of something to which pointer can \
1630 active_reg_t this_reg; \
1631 \
1632 DEBUG_STATEMENT (failure_id++); \
1633 DEBUG_STATEMENT (nfailure_points_pushed++); \
1637 \
1640 \
1642 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1643 { \
1644 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1645 return failure_code; \
1646 \
1648 (fail_stack).size); \
1650 } \
1651 \
1654 \
1655 if (1) \
1656 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1657 this_reg++) \
1658 { \
1660 DEBUG_STATEMENT (num_regs_pushed++); \
1661 \
1663 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1664 \
1666 PUSH_FAILURE_POINTER (regend[this_reg]); \
1667 \
1669 reg_info[this_reg].word.pointer); \
1671 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1674 MATCHED_SOMETHING (reg_info[this_reg])); \
1676 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1678 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1679 } \
1680 \
1682 PUSH_FAILURE_INT (lowest_active_reg); \
1683 \
1685 PUSH_FAILURE_INT (highest_active_reg); \
1686 \
1688 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1689 PUSH_FAILURE_POINTER (pattern_place); \
1690 \
1692 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1693 size2); \
1695 PUSH_FAILURE_POINTER (string_place); \
1696 \
1698 DEBUG_PUSH (failure_id); \
1699 } while (0)
1701 # ifndef DEFINED_ONCE
1702 /* This is the number of items that are pushed and popped on the stack
1703 for each register. */
1704 # define NUM_REG_ITEMS 3
1706 /* Individual items aside from the registers. */
1707 # ifdef DEBUG
1709 # else
1710 # define NUM_NONREG_ITEMS 4
1711 # endif
1713 /* We push at most this many items on the stack. */
1714 /* We used to use (num_regs - 1), which is the number of registers
1715 this regexp will save; but that was changed to 5
1716 to avoid stack overflow for a regexp with lots of parens. */
1717 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1719 /* We actually push this many items. */
1720 # define NUM_FAILURE_ITEMS \
1721 (((0 \
1722 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1723 * NUM_REG_ITEMS) \
1724 + NUM_NONREG_ITEMS)
1726 /* How many items can still be added to the stack without overflowing it. */
1727 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1731 /* Pops what PUSH_FAIL_STACK pushes.
1733 We restore into the parameters, all of which should be lvalues:
1734 STR -- the saved data position.
1735 PAT -- the saved pattern position.
1736 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1737 REGSTART, REGEND -- arrays of string positions.
1738 REG_INFO -- array of information about each subexpression.
1740 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1741 `pend', `string1', `size1', `string2', and `size2'. */
1742 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1743 { \
1744 DEBUG_STATEMENT (unsigned failure_id;) \
1745 active_reg_t this_reg; \
1746 const UCHAR_T *string_temp; \
1747 \
1748 assert (!FAIL_STACK_EMPTY ()); \
1749 \
1754 \
1755 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1756 \
1757 DEBUG_POP (&failure_id); \
1759 \
1760 /* If the saved string location is NULL, it came from an \
1761 on_failure_keep_string_jump opcode, and we want to throw away the \
1763 string_temp = POP_FAILURE_POINTER (); \
1764 if (string_temp != NULL) \
1765 str = (const CHAR_T *) string_temp; \
1766 \
1768 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1770 \
1771 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1773 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1774 \
1776 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1778 \
1779 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1781 \
1782 if (1) \
1783 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1784 { \
1786 \
1787 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1789 reg_info[this_reg].word.pointer); \
1790 \
1791 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1793 \
1794 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1796 } \
1797 else \
1798 { \
1799 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1800 { \
1801 reg_info[this_reg].word.integer = 0; \
1802 regend[this_reg] = 0; \
1803 regstart[this_reg] = 0; \
1804 } \
1805 highest_active_reg = high_reg; \
1806 } \
1807 \
1808 set_regs_matched_done = 0; \
1809 DEBUG_STATEMENT (nfailure_points_popped++); \
1811 \f
1812 /* Structure for per-register (a.k.a. per-group) information.
1813 Other register information, such as the
1814 starting and ending positions (which are addresses), and the list of
1815 inner groups (which is a bits list) are maintained in separate
1816 variables.
1818 We are making a (strictly speaking) nonportable assumption here: that
1819 the compiler will pack our bit fields into something that fits into
1820 the type of `word', i.e., is something that fits into one item on the
1821 failure stack. */
1824 /* Declarations and macros for re_match_2. */
1827 {
1829 struct
1830 {
1831 /* This field is one if this group can match the empty string,
1832 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1833 # define MATCH_NULL_UNSET_VALUE 3
1841 # ifndef DEFINED_ONCE
1842 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1843 # define IS_ACTIVE(R) ((R).bits.is_active)
1844 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1845 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1848 /* Call this when have matched a real character; it sets `matched' flags
1849 for the subexpressions which we are currently inside. Also records
1850 that those subexprs have matched. */
1851 # define SET_REGS_MATCHED() \
1852 do \
1853 { \
1854 if (!set_regs_matched_done) \
1855 { \
1856 active_reg_t r; \
1857 set_regs_matched_done = 1; \
1858 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1859 { \
1860 MATCHED_SOMETHING (reg_info[r]) \
1861 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1862 = 1; \
1863 } \
1864 } \
1865 } \
1866 while (0)
1869 /* Registers are set to a sentinel when they haven't yet matched. */
1871 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1872 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1874 /* Subroutine declarations and macros for regex_compile. */
1884 reg_syntax_t syntax));
1887 reg_syntax_t syntax));
1888 # ifdef WCHAR
1893 reg_syntax_t syntax,
1902 reg_syntax_t syntax,
1906 /* Fetch the next character in the uncompiled pattern---translating it
1907 if necessary. Also cast from a signed character in the constant
1908 string passed to us by the user to an unsigned char that we can use
1909 as an array index (in, e.g., `translate'). */
1910 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1911 because it is impossible to allocate 4GB array for some encodings
1912 which have 4 byte character_set like UCS4. */
1913 # ifndef PATFETCH
1914 # ifdef WCHAR
1915 # define PATFETCH(c) \
1916 do {if (p == pend) return REG_EEND; \
1917 c = (UCHAR_T) *p++; \
1918 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1919 } while (0)
1921 # define PATFETCH(c) \
1922 do {if (p == pend) return REG_EEND; \
1923 c = (unsigned char) *p++; \
1924 if (translate) c = (unsigned char) translate[c]; \
1925 } while (0)
1927 # endif
1929 /* Fetch the next character in the uncompiled pattern, with no
1930 translation. */
1931 # define PATFETCH_RAW(c) \
1932 do {if (p == pend) return REG_EEND; \
1933 c = (UCHAR_T) *p++; \
1934 } while (0)
1936 /* Go backwards one character in the pattern. */
1937 # define PATUNFETCH p--
1940 /* If `translate' is non-null, return translate[D], else just D. We
1941 cast the subscript to translate because some data is declared as
1942 `char *', to avoid warnings when a string constant is passed. But
1943 when we use a character as a subscript we must make it unsigned. */
1944 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1945 because it is impossible to allocate 4GB array for some encodings
1946 which have 4 byte character_set like UCS4. */
1948 # ifndef TRANSLATE
1949 # ifdef WCHAR
1950 # define TRANSLATE(d) \
1951 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1952 ? (char) translate[(unsigned char) (d)] : (d))
1954 # define TRANSLATE(d) \
1955 (translate ? (char) translate[(unsigned char) (d)] : (d))
1957 # endif
1960 /* Macros for outputting the compiled pattern into `buffer'. */
1962 /* If the buffer isn't allocated when it comes in, use this. */
1963 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1965 /* Make sure we have at least N more bytes of space in buffer. */
1966 # ifdef WCHAR
1967 # define GET_BUFFER_SPACE(n) \
1968 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
1969 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
1970 EXTEND_BUFFER ()
1972 # define GET_BUFFER_SPACE(n) \
1973 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1974 EXTEND_BUFFER ()
1977 /* Make sure we have one more byte of buffer space and then add C to it. */
1978 # define BUF_PUSH(c) \
1979 do { \
1980 GET_BUFFER_SPACE (1); \
1981 *b++ = (UCHAR_T) (c); \
1982 } while (0)
1985 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1986 # define BUF_PUSH_2(c1, c2) \
1987 do { \
1988 GET_BUFFER_SPACE (2); \
1989 *b++ = (UCHAR_T) (c1); \
1990 *b++ = (UCHAR_T) (c2); \
1991 } while (0)
1994 /* As with BUF_PUSH_2, except for three bytes. */
1995 # define BUF_PUSH_3(c1, c2, c3) \
1996 do { \
1997 GET_BUFFER_SPACE (3); \
1998 *b++ = (UCHAR_T) (c1); \
1999 *b++ = (UCHAR_T) (c2); \
2000 *b++ = (UCHAR_T) (c3); \
2001 } while (0)
2003 /* Store a jump with opcode OP at LOC to location TO. We store a
2004 relative address offset by the three bytes the jump itself occupies. */
2005 # define STORE_JUMP(op, loc, to) \
2006 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
2008 /* Likewise, for a two-argument jump. */
2009 # define STORE_JUMP2(op, loc, to, arg) \
2010 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
2012 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
2013 # define INSERT_JUMP(op, loc, to) \
2014 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
2016 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
2017 # define INSERT_JUMP2(op, loc, to, arg) \
2018 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
2019 arg, b)
2021 /* This is not an arbitrary limit: the arguments which represent offsets
2022 into the pattern are two bytes long. So if 2^16 bytes turns out to
2023 be too small, many things would have to change. */
2024 /* Any other compiler which, like MSC, has allocation limit below 2^16
2025 bytes will have to use approach similar to what was done below for
2026 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
2027 reallocating to 0 bytes. Such thing is not going to work too well.
2028 You have been warned!! */
2029 # ifndef DEFINED_ONCE
2030 # if defined _MSC_VER && !defined WIN32
2031 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2032 The REALLOC define eliminates a flurry of conversion warnings,
2033 but is not required. */
2034 # define MAX_BUF_SIZE 65500L
2035 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2036 # else
2037 # define MAX_BUF_SIZE (1L << 16)
2038 # define REALLOC(p,s) realloc ((p), (s))
2039 # endif
2041 /* Extend the buffer by twice its current size via realloc and
2042 reset the pointers that pointed into the old block to point to the
2043 correct places in the new one. If extending the buffer results in it
2044 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2045 # if __BOUNDED_POINTERS__
2046 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2047 # define MOVE_BUFFER_POINTER(P) \
2048 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2049 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2050 else \
2051 { \
2052 SET_HIGH_BOUND (b); \
2053 SET_HIGH_BOUND (begalt); \
2054 if (fixup_alt_jump) \
2055 SET_HIGH_BOUND (fixup_alt_jump); \
2056 if (laststart) \
2057 SET_HIGH_BOUND (laststart); \
2058 if (pending_exact) \
2059 SET_HIGH_BOUND (pending_exact); \
2060 }
2061 # else
2062 # define MOVE_BUFFER_POINTER(P) (P) += incr
2063 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2064 # endif
2067 # ifdef WCHAR
2068 # define EXTEND_BUFFER() \
2069 do { \
2070 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2071 int wchar_count; \
2072 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2073 return REG_ESIZE; \
2074 bufp->allocated <<= 1; \
2075 if (bufp->allocated > MAX_BUF_SIZE) \
2076 bufp->allocated = MAX_BUF_SIZE; \
2078 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2079 if (wchar_count == 0) wchar_count = 1; \
2081 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2082 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2083 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2084 if (COMPILED_BUFFER_VAR == NULL) \
2085 return REG_ESPACE; \
2087 if (old_buffer != COMPILED_BUFFER_VAR) \
2088 { \
2089 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2090 MOVE_BUFFER_POINTER (b); \
2091 MOVE_BUFFER_POINTER (begalt); \
2092 if (fixup_alt_jump) \
2093 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2094 if (laststart) \
2095 MOVE_BUFFER_POINTER (laststart); \
2096 if (pending_exact) \
2097 MOVE_BUFFER_POINTER (pending_exact); \
2098 } \
2099 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2100 } while (0)
2102 # define EXTEND_BUFFER() \
2103 do { \
2104 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2105 if (bufp->allocated == MAX_BUF_SIZE) \
2106 return REG_ESIZE; \
2107 bufp->allocated <<= 1; \
2108 if (bufp->allocated > MAX_BUF_SIZE) \
2109 bufp->allocated = MAX_BUF_SIZE; \
2110 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2111 bufp->allocated); \
2112 if (COMPILED_BUFFER_VAR == NULL) \
2113 return REG_ESPACE; \
2115 if (old_buffer != COMPILED_BUFFER_VAR) \
2116 { \
2117 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2118 MOVE_BUFFER_POINTER (b); \
2119 MOVE_BUFFER_POINTER (begalt); \
2120 if (fixup_alt_jump) \
2121 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2122 if (laststart) \
2123 MOVE_BUFFER_POINTER (laststart); \
2124 if (pending_exact) \
2125 MOVE_BUFFER_POINTER (pending_exact); \
2126 } \
2127 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2128 } while (0)
2131 # ifndef DEFINED_ONCE
2132 /* Since we have one byte reserved for the register number argument to
2133 {start,stop}_memory, the maximum number of groups we can report
2134 things about is what fits in that byte. */
2135 # define MAX_REGNUM 255
2137 /* But patterns can have more than `MAX_REGNUM' registers. We just
2138 ignore the excess. */
2142 /* Macros for the compile stack. */
2144 /* Since offsets can go either forwards or backwards, this type needs to
2145 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2146 /* int may be not enough when sizeof(int) == 2. */
2150 {
2151 pattern_offset_t begalt_offset;
2152 pattern_offset_t fixup_alt_jump;
2153 pattern_offset_t inner_group_offset;
2154 pattern_offset_t laststart_offset;
2155 regnum_t regnum;
2160 {
2167 # define INIT_COMPILE_STACK_SIZE 32
2169 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2170 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2172 /* The next available element. */
2173 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2177 /* Set the bit for character C in a list. */
2178 # ifndef DEFINED_ONCE
2179 # define SET_LIST_BIT(c) \
2180 (b[((unsigned char) (c)) / BYTEWIDTH] \
2181 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2184 /* Get the next unsigned number in the uncompiled pattern. */
2185 # define GET_UNSIGNED_NUMBER(num) \
2186 { \
2187 while (p != pend) \
2188 { \
2189 PATFETCH (c); \
2191 break; \
2192 if (num <= RE_DUP_MAX) \
2193 { \
2194 if (num < 0) \
2195 num = 0; \
2197 } \
2198 } \
2199 }
2201 # ifndef DEFINED_ONCE
2202 # if defined _LIBC || WIDE_CHAR_SUPPORT
2203 /* The GNU C library provides support for user-defined character classes
2204 and the functions from ISO C amendement 1. */
2205 # ifdef CHARCLASS_NAME_MAX
2206 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2207 # else
2208 /* This shouldn't happen but some implementation might still have this
2209 problem. Use a reasonable default value. */
2210 # define CHAR_CLASS_MAX_LENGTH 256
2211 # endif
2213 # ifdef _LIBC
2214 # define IS_CHAR_CLASS(string) __wctype (string)
2215 # else
2216 # define IS_CHAR_CLASS(string) wctype (string)
2217 # endif
2218 # else
2221 # define IS_CHAR_CLASS(string) \
2228 # endif
2230 \f
2231 # ifndef MATCH_MAY_ALLOCATE
2233 /* If we cannot allocate large objects within re_match_2_internal,
2234 we make the fail stack and register vectors global.
2235 The fail stack, we grow to the maximum size when a regexp
2236 is compiled.
2237 The register vectors, we adjust in size each time we
2238 compile a regexp, according to the number of registers it needs. */
2242 /* Size with which the following vectors are currently allocated.
2243 That is so we can make them bigger as needed,
2244 but never make them smaller. */
2245 # ifdef DEFINED_ONCE
2257 /* Make the register vectors big enough for NUM_REGS registers,
2258 but don't make them smaller. */
2260 static void
2263 {
2265 {
2277 }
2278 }
2281 \f
2282 # ifndef DEFINED_ONCE
2284 compile_stack,
2285 regnum_t regnum));
2288 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2289 Returns one of error codes defined in `regex.h', or zero for success.
2291 Assumes the `allocated' (and perhaps `buffer') and `translate'
2292 fields are set in BUFP on entry.
2294 If it succeeds, results are put in BUFP (if it returns an error, the
2295 contents of BUFP are undefined):
2296 `buffer' is the compiled pattern;
2297 `syntax' is set to SYNTAX;
2298 `used' is set to the length of the compiled pattern;
2299 `fastmap_accurate' is zero;
2300 `re_nsub' is the number of subexpressions in PATTERN;
2301 `not_bol' and `not_eol' are zero;
2303 The `fastmap' and `newline_anchor' fields are neither
2304 examined nor set. */
2306 /* Return, freeing storage we allocated. */
2307 # ifdef WCHAR
2308 # define FREE_STACK_RETURN(value) \
2309 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2310 # else
2311 # define FREE_STACK_RETURN(value) \
2312 return (free (compile_stack.stack), value)
2315 static reg_errcode_t
2319 reg_syntax_t syntax;
2321 {
2322 /* We fetch characters from PATTERN here. Even though PATTERN is
2323 `char *' (i.e., signed), we declare these variables as unsigned, so
2324 they can be reliably used as array indices. */
2327 #ifdef WCHAR
2328 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2331 /* offset buffer for optimization. See convert_mbs_to_wc. */
2333 /* It hold whether each wchar_t is binary data or not. */
2335 /* A flag whether exactn is handling binary data or not. */
2339 /* A random temporary spot in PATTERN. */
2342 /* Points to the end of the buffer, where we should append. */
2345 /* Keeps track of unclosed groups. */
2346 compile_stack_type compile_stack;
2348 /* Points to the current (ending) position in the pattern. */
2349 #ifdef WCHAR
2357 /* How to translate the characters in the pattern. */
2360 /* Address of the count-byte of the most recently inserted `exactn'
2361 command. This makes it possible to tell if a new exact-match
2362 character can be added to that command or if the character requires
2363 a new `exactn' command. */
2366 /* Address of start of the most recently finished expression.
2367 This tells, e.g., postfix * where to find the start of its
2368 operand. Reset at the beginning of groups and alternatives. */
2371 /* Address of beginning of regexp, or inside of last group. */
2374 /* Address of the place where a forward jump should go to the end of
2375 the containing expression. Each alternative of an `or' -- except the
2376 last -- ends with a forward jump of this sort. */
2379 /* Counts open-groups as they are encountered. Remembered for the
2380 matching close-group on the compile stack, so the same register
2381 number is put in the stop_memory as the start_memory. */
2384 #ifdef WCHAR
2385 /* Initialize the wchar_t PATTERN and offset_buffer. */
2390 {
2395 }
2400 {
2405 }
2406 #endif
2408 #ifdef DEBUG
2411 {
2417 }
2420 /* Initialize the compile stack. */
2423 {
2424 #ifdef WCHAR
2428 #endif
2430 }
2435 /* Initialize the pattern buffer. */
2440 /* Set `used' to zero, so that if we return an error, the pattern
2441 printer (for debugging) will think there's no pattern. We reset it
2442 at the end. */
2445 /* Always count groups, whether or not bufp->no_sub is set. */
2448 #if !defined emacs && !defined SYNTAX_TABLE
2449 /* Initialize the syntax table. */
2451 #endif
2454 {
2457 enough space. This loses if buffer's address is bogus, but
2458 that is the user's responsibility. */
2459 #ifdef WCHAR
2460 /* Free bufp->buffer and allocate an array for wchar_t pattern
2461 buffer. */
2464 UCHAR_T);
2465 #else
2468 }
2469 else
2472 UCHAR_T);
2473 }
2476 #ifdef WCHAR
2480 }
2481 #ifdef WCHAR
2482 else
2484 #endif
2488 /* Loop through the uncompiled pattern until we're at the end. */
2490 {
2494 {
2496 {
2499 /* If context independent, it's an operator. */
2501 /* Otherwise, depends on what's come before. */
2504 else
2506 }
2511 {
2513 p == pend
2514 /* If context independent, it's an operator. */
2516 /* Otherwise, depends on what's next. */
2519 else
2521 }
2530 handle_plus:
2532 /* If there is no previous pattern... */
2534 {
2539 }
2541 {
2542 /* Are we optimizing this jump? */
2545 /* 1 means zero (many) matches is allowed. */
2548 /* If there is a sequence of repetition chars, collapse it
2549 down to just one (the right one). We can't combine
2550 interval operators with these because of, e.g., `a{2}*',
2551 which should only match an even number of `a's. */
2554 {
2565 ;
2568 {
2573 {
2574 PATUNFETCH;
2575 PATUNFETCH;
2577 }
2580 }
2581 else
2582 {
2583 PATUNFETCH;
2585 }
2587 /* If we get here, we found another repeat character. */
2588 }
2590 /* Star, etc. applied to an empty pattern is equivalent
2591 to an empty pattern. */
2595 /* Now we know whether or not zero matches is allowed
2596 and also whether or not two or more matches is allowed. */
2599 end a backward relative jump from `b' to before the next
2600 jump we're going to put in below (which jumps from
2601 laststart to after this jump).
2603 But if we are at the `*' in the exact sequence `.*\n',
2604 insert an unconditional jump backwards to the .,
2605 instead of the beginning of the loop. This way we only
2606 push a failure point once, instead of every time
2607 through the loop. */
2610 /* Allocate the space for the jump. */
2613 /* We know we are not at the first character of the pattern,
2614 because laststart was nonzero. And we've already
2615 incremented `p', by the way, to be the character after
2616 the `*'. Do we have to do something analogous here
2617 for null bytes, because of RE_DOT_NOT_NULL? */
2619 && zero_times_ok
2625 }
2626 else
2627 /* Anything else. */
2631 /* We've added more stuff to the buffer. */
2633 }
2635 /* On failure, jump from laststart to b + 3, which will be the
2636 end of the buffer after this jump is inserted. */
2637 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2638 'b + 3'. */
2647 {
2648 /* At least one repetition is required, so insert a
2649 `dummy_failure_jump' before the initial
2650 `on_failure_jump' instruction of the loop. This
2651 effects a skip over that instruction the first time
2652 we hit that loop. */
2657 }
2658 }
2669 {
2671 #ifdef WCHAR
2673 #else
2675 #endif
2678 #ifdef WCHAR
2679 /* We assume a charset(_not) structure as a wchar_t array.
2680 charset[0] = (re_opcode_t) charset(_not)
2681 charset[1] = l (= length of char_classes)
2682 charset[2] = m (= length of collating_symbols)
2683 charset[3] = n (= length of equivalence_classes)
2684 charset[4] = o (= length of char_ranges)
2685 charset[5] = p (= length of chars)
2687 charset[6] = char_class (wctype_t)
2688 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2689 ...
2690 charset[l+5] = char_class (wctype_t)
2692 charset[l+6] = collating_symbol (wchar_t)
2693 ...
2694 charset[l+m+5] = collating_symbol (wchar_t)
2695 ifdef _LIBC we use the index if
2696 _NL_COLLATE_SYMB_EXTRAMB instead of
2697 wchar_t string.
2699 charset[l+m+6] = equivalence_classes (wchar_t)
2700 ...
2701 charset[l+m+n+5] = equivalence_classes (wchar_t)
2702 ifdef _LIBC we use the index in
2703 _NL_COLLATE_WEIGHT instead of
2704 wchar_t string.
2706 charset[l+m+n+6] = range_start
2707 charset[l+m+n+7] = range_end
2708 ...
2709 charset[l+m+n+2o+4] = range_start
2710 charset[l+m+n+2o+5] = range_end
2711 ifdef _LIBC we use the value looked up
2712 in _NL_COLLATE_COLLSEQ instead of
2713 wchar_t character.
2715 charset[l+m+n+2o+6] = char
2716 ...
2717 charset[l+m+n+2o+p+5] = char
2719 */
2721 /* We need at least 6 spaces: the opcode, the length of
2722 char_classes, the length of collating_symbols, the length of
2723 equivalence_classes, the length of char_ranges, the length of
2724 chars. */
2727 /* Save b as laststart. And We use laststart as the pointer
2728 to the first element of the charset here.
2729 In other words, laststart[i] indicates charset[i]. */
2732 /* We test `*p == '^' twice, instead of using an if
2733 statement, so we only need one BUF_PUSH. */
2736 p++;
2738 /* Push the length of char_classes, the length of
2739 collating_symbols, the length of equivalence_classes, the
2740 length of char_ranges and the length of chars. */
2744 /* Remember the first position in the bracket expression. */
2747 /* charset_not matches newline according to a syntax bit. */
2750 {
2753 }
2755 /* Read in characters and ranges, setting map bits. */
2757 {
2762 /* \ might escape characters inside [...] and [^...]. */
2764 {
2772 }
2774 /* Could be the end of the bracket expression. If it's
2775 not (i.e., when the bracket expression is `[]' so
2776 far), the ']' character bit gets set way below. */
2780 /* Look ahead to see if it's a range when the last thing
2781 was a character class. */
2785 /* Look ahead to see if it's a range when the last thing
2786 was a character: if this is a hyphen not at the
2787 beginning or the end of a list, then it's the range
2788 operator. */
2793 {
2794 reg_errcode_t ret;
2795 /* Allocate the space for range_start and range_end. */
2797 /* Update the pointer to indicate end of buffer. */
2803 }
2806 reg_errcode_t ret;
2808 /* Move past the `-'. */
2810 /* Allocate the space for range_start and range_end. */
2812 /* Update the pointer to indicate end of buffer. */
2815 laststart);
2818 }
2820 /* See if we're at the beginning of a possible character
2821 class. */
2829 /* If pattern is `[[:'. */
2833 {
2839 else
2840 /* This is in any case an invalid class name. */
2842 }
2845 /* If isn't a word bracketed by `[:' and `:]':
2846 undo the ending character, the letters, and leave
2847 the leading `:' and `[' (but store them as character). */
2849 {
2853 /* Query the character class as wctype_t. */
2858 /* Throw away the ] at the end of the character
2859 class. */
2864 /* Allocate the space for character class. */
2866 /* Update the pointer to indicate end of buffer. */
2868 /* Move data which follow character classes
2869 not to violate the data. */
2876 /* Store the character class. */
2878 /* Update length of char_classes */
2882 }
2883 else
2884 {
2885 c1++;
2887 PATUNFETCH;
2893 }
2894 }
2897 {
2900 # ifdef _LIBC
2903 # endif
2907 /* If pattern is `[[=' or '[[.'. */
2911 {
2917 else
2918 /* This is in any case an invalid class name. */
2920 }
2924 {
2926 /* If we have no collation data we use the default
2927 collation in which each character is in a class
2928 by itself. It also means that ASCII is the
2929 character set and therefore we cannot have character
2930 with more than one byte in the multibyte
2931 representation. */
2933 /* If not defined _LIBC, we push the name and
2934 `\0' for the sake of matching performance. */
2937 # ifdef _LIBC
2940 # endif
2941 {
2944 }
2945 # ifdef _LIBC
2946 else
2947 {
2954 /* This #include defines a local function! */
2955 # include <locale/weightwc.h>
2958 {
2959 /* We push the index for equivalence class. */
2964 _NL_COLLATE_TABLEWC);
2967 _NL_COLLATE_WEIGHTWC);
2970 _NL_COLLATE_EXTRAWC);
2973 _NL_COLLATE_INDIRECTWC);
2977 /* This is no valid character. */
2981 }
2983 {
2984 /* We push collation sequence value
2985 for collating symbol. */
2995 /* We have to convert the name to a single-byte
2996 string. This is possible since the names
2997 consist of ASCII characters and the internal
2998 representation is UCS4. */
3002 table_size =
3004 _NL_COLLATE_SYMB_HASH_SIZEMB);
3007 _NL_COLLATE_SYMB_TABLEMB);
3010 _NL_COLLATE_SYMB_EXTRAMB);
3012 /* Locate the character in the hashing table. */
3019 {
3020 /* First compare the hashing value. */
3026 {
3027 /* Yep, this is the entry. */
3031 }
3033 /* Next entry. */
3035 }
3038 {
3039 /* Compute the index of the byte sequence
3040 in the table. */
3042 /* Adjust for the alignment. */
3046 }
3048 {
3049 /* No valid character. Match it as a
3050 single byte character. */
3053 /* Update the length of characters */
3057 /* Throw away the ] at the end of the
3058 collating symbol. */
3060 /* exit from the switch block. */
3062 }
3063 else
3065 }
3067 }
3068 # endif
3069 /* Throw away the ] at the end of the equivalence
3070 class (or collating symbol). */
3073 /* Allocate the space for the equivalence class
3074 (or collating symbol) (and '\0' if needed). */
3076 /* Update the pointer to indicate end of buffer. */
3081 /* Calculate the offset of char_ranges,
3082 which is next to equivalence_classes. */
3085 /* Insert space. */
3088 /* Write the equivalence_class and \0. */
3092 /* Update the length of equivalence_classes. */
3095 }
3098 /* Calculate the offset of the equivalence_classes,
3099 which is next to collating_symbols. */
3101 /* Insert space and write the collationg_symbol
3102 and \0. */
3107 /* In re_match_2_internal if range_start < -1, we
3108 assume -range_start is the offset of the
3109 collating symbol which is specified as
3110 the character of the range start. So we assign
3111 -(laststart[1] + laststart[2] + 6) to
3112 range_start. */
3114 /* Update the length of collating_symbol. */
3117 }
3118 }
3119 else
3120 {
3121 c1++;
3123 PATUNFETCH;
3129 }
3130 }
3131 else
3132 {
3137 }
3138 }
3141 /* Ensure that we have enough space to push a charset: the
3142 opcode, the length count, and the bitset; 34 bytes in all. */
3147 /* We test `*p == '^' twice, instead of using an if
3148 statement, so we only need one BUF_PUSH. */
3151 p++;
3153 /* Remember the first position in the bracket expression. */
3156 /* Push the number of bytes in the bitmap. */
3159 /* Clear the whole map. */
3162 /* charset_not matches newline according to a syntax bit. */
3167 /* Read in characters and ranges, setting map bits. */
3169 {
3174 /* \ might escape characters inside [...] and [^...]. */
3176 {
3183 }
3185 /* Could be the end of the bracket expression. If it's
3186 not (i.e., when the bracket expression is `[]' so
3187 far), the ']' character bit gets set way below. */
3191 /* Look ahead to see if it's a range when the last thing
3192 was a character class. */
3196 /* Look ahead to see if it's a range when the last thing
3197 was a character: if this is a hyphen not at the
3198 beginning or the end of a list, then it's the range
3199 operator. */
3204 {
3205 reg_errcode_t ret
3210 }
3214 reg_errcode_t ret;
3216 /* Move past the `-'. */
3222 }
3224 /* See if we're at the beginning of a possible character
3225 class. */
3234 /* If pattern is `[[:'. */
3238 {
3244 else
3245 /* This is in any case an invalid class name. */
3247 }
3250 /* If isn't a word bracketed by `[:' and `:]':
3251 undo the ending character, the letters, and leave
3252 the leading `:' and `[' (but set bits for them). */
3254 {
3255 # if defined _LIBC || WIDE_CHAR_SUPPORT
3265 /* Throw away the ] at the end of the character
3266 class. */
3272 {
3273 # ifdef _LIBC
3276 # else
3279 # endif
3284 }
3287 # else
3305 /* Throw away the ] at the end of the character
3306 class. */
3312 {
3313 /* This was split into 3 if's to
3314 avoid an arbitrary limit in some compiler. */
3333 }
3336 }
3337 else
3338 {
3339 c1++;
3341 PATUNFETCH;
3346 }
3347 }
3349 {
3351 # ifdef _LIBC
3354 # endif
3359 /* If pattern is `[[='. */
3363 {
3369 else
3370 /* This is in any case an invalid class name. */
3372 }
3376 {
3377 /* If we have no collation data we use the default
3378 collation in which each character is in a class
3379 by itself. It also means that ASCII is the
3380 character set and therefore we cannot have character
3381 with more than one byte in the multibyte
3382 representation. */
3383 # ifdef _LIBC
3385 # endif
3386 {
3390 /* Throw away the ] at the end of the equivalence
3391 class. */
3394 /* Set the bit for the character. */
3396 }
3397 # ifdef _LIBC
3398 else
3399 {
3400 /* Try to match the byte sequence in `str' against
3401 those known to the collate implementation.
3402 First find out whether the bytes in `str' are
3403 actually from exactly one character. */
3412 /* This #include defines a local function! */
3413 # include <locale/weight.h>
3426 /* This is no valid character. */
3429 /* Throw away the ] at the end of the equivalence
3430 class. */
3433 /* Now we have to go throught the whole table
3434 and find all characters which have the same
3435 first level weight.
3437 XXX Note that this is not entirely correct.
3438 we would have to match multibyte sequences
3439 but this is not possible with the current
3440 implementation. */
3442 /* XXX This test would have to be changed if we
3443 would allow matching multibyte sequences. */
3445 {
3449 /* Test whether the lenghts match. */
3451 {
3452 /* They do. New compare the bytes of
3453 the weight. */
3462 /* They match. Mark the character as
3463 acceptable. */
3465 }
3466 }
3467 }
3468 # endif
3470 }
3471 else
3472 {
3473 c1++;
3475 PATUNFETCH;
3480 }
3481 }
3483 {
3485 # ifdef _LIBC
3488 # endif
3493 /* If pattern is `[[.'. */
3497 {
3503 else
3504 /* This is in any case an invalid class name. */
3506 }
3510 {
3511 /* If we have no collation data we use the default
3512 collation in which each character is the name
3513 for its own class which contains only the one
3514 character. It also means that ASCII is the
3515 character set and therefore we cannot have character
3516 with more than one byte in the multibyte
3517 representation. */
3518 # ifdef _LIBC
3520 # endif
3521 {
3525 /* Throw away the ] at the end of the equivalence
3526 class. */
3529 /* Set the bit for the character. */
3532 }
3533 # ifdef _LIBC
3534 else
3535 {
3536 /* Try to match the byte sequence in `str' against
3537 those known to the collate implementation.
3538 First find out whether the bytes in `str' are
3539 actually from exactly one character. */
3548 table_size =
3550 _NL_COLLATE_SYMB_HASH_SIZEMB);
3553 _NL_COLLATE_SYMB_TABLEMB);
3556 _NL_COLLATE_SYMB_EXTRAMB);
3558 /* Locate the character in the hashing table. */
3565 {
3566 /* First compare the hashing value. */
3573 {
3574 /* Yep, this is the entry. */
3578 }
3580 /* Next entry. */
3582 }
3585 /* This is no valid character. */
3588 /* Throw away the ] at the end of the equivalence
3589 class. */
3592 /* Now add the multibyte character(s) we found
3593 to the accept list.
3595 XXX Note that this is not entirely correct.
3596 we would have to match multibyte sequences
3597 but this is not possible with the current
3598 implementation. Also, we have to match
3599 collating symbols, which expand to more than
3600 one file, as a whole and not allow the
3601 individual bytes. */
3606 {
3609 }
3610 }
3611 # endif
3613 }
3614 else
3615 {
3616 c1++;
3618 PATUNFETCH;
3623 }
3624 }
3625 else
3626 {
3630 }
3631 }
3633 /* Discard any (non)matching list bytes that are all 0 at the
3634 end of the map. Decrease the map-length byte too. */
3639 }
3646 else
3653 else
3660 else
3667 else
3674 else
3681 /* Do not translate the character after the \, so that we can
3682 distinguish, e.g., \B from \b, even if we normally would
3683 translate, e.g., B to b. */
3687 {
3692 handle_open:
3694 regnum++;
3697 {
3699 compile_stack_elt_t);
3703 }
3705 /* These are the values to restore when we hit end of this
3706 group. They are all relative offsets, so that if the
3707 whole pattern moves because of realloc, they will still
3708 be valid. */
3710 COMPILE_STACK_TOP.fixup_alt_jump
3715 /* We will eventually replace the 0 with the number of
3716 groups inner to this one. But do not push a
3717 start_memory for groups beyond the last one we can
3718 represent in the compiled pattern. */
3720 {
3724 }
3731 /* If we've reached MAX_REGNUM groups, then this open
3732 won't actually generate any code, so we'll have to
3733 clear pending_exact explicitly. */
3742 {
3745 else
3747 }
3749 handle_close:
3752 alternative for a possible future
3753 `pop_failure_jump' to pop. See comments at
3754 `push_dummy_failure' in `re_match_2'. */
3757 /* We allocated space for this jump when we assigned
3758 to `fixup_alt_jump', in the `handle_alt' case below. */
3760 }
3762 /* See similar code for backslashed left paren above. */
3764 {
3767 else
3769 }
3771 /* Since we just checked for an empty stack above, this
3772 ``can't happen''. */
3774 {
3775 /* We don't just want to restore into `regnum', because
3776 later groups should continue to be numbered higher,
3777 as in `(ab)c(de)' -- the second group is #2. */
3778 regnum_t this_group_regnum;
3782 fixup_alt_jump
3788 /* If we've reached MAX_REGNUM groups, then this open
3789 won't actually generate any code, so we'll have to
3790 clear pending_exact explicitly. */
3793 /* We're at the end of the group, so now we know how many
3794 groups were inside this one. */
3796 {
3797 UCHAR_T *inner_group_loc
3803 }
3804 }
3811 handle_alt:
3815 /* Insert before the previous alternative a jump which
3816 jumps to this alternative if the former fails. */
3823 /* The alternative before this one has a jump after it
3824 which gets executed if it gets matched. Adjust that
3825 jump so it will jump to this alternative's analogous
3826 jump (put in below, which in turn will jump to the next
3827 (if any) alternative's such jump, etc.). The last such
3828 jump jumps to the correct final destination. A picture:
3829 _____ _____
3830 | | | |
3831 | v | v
3832 a | b | c
3834 If we are at `b', then fixup_alt_jump right now points to a
3835 three-byte space after `a'. We'll put in the jump, set
3836 fixup_alt_jump to right after `b', and leave behind three
3837 bytes which we'll fill in when we get to after `c'. */
3842 /* Mark and leave space for a jump after this alternative,
3843 to be filled in later either by next alternative or
3844 when know we're at the end of a series of alternatives. */
3855 /* If \{ is a literal. */
3857 /* If we're at `\{' and it's not the open-interval
3858 operator. */
3862 handle_interval:
3863 {
3864 /* If got here, then the syntax allows intervals. */
3866 /* At least (most) this many matches must be made. */
3869 /* Place in the uncompiled pattern (i.e., just after
3870 the '{') to go back to if the interval is invalid. */
3879 {
3883 }
3884 else
3885 /* Interval such as `{1}' => match exactly once. */
3892 {
3896 }
3901 /* If it's invalid to have no preceding re. */
3903 {
3909 else
3911 }
3913 /* We just parsed a valid interval. */
3918 /* If the upper bound is zero, don't want to succeed at
3919 all; jump from `laststart' to `b + 3', which will be
3920 the end of the buffer after we insert the jump. */
3921 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3922 instead of 'b + 3'. */
3924 {
3929 }
3931 /* Otherwise, we have a nontrivial interval. When
3932 we're all done, the pattern will look like:
3933 set_number_at <jump count> <upper bound>
3934 set_number_at <succeed_n count> <lower bound>
3935 succeed_n <after jump addr> <succeed_n count>
3936 <body of loop>
3937 jump_n <succeed_n addr> <jump count>
3938 (The upper bound and `jump_n' are omitted if
3939 `upper_bound' is 1, though.) */
3940 else
3942 more at the end of the loop. */
3948 /* Initialize lower bound of the `succeed_n', even
3949 though it will be set during matching by its
3950 attendant `set_number_at' (inserted next),
3951 because `re_compile_fastmap' needs to know.
3952 Jump to the `jump_n' we might insert below. */
3959 /* Code to initialize the lower bound. Insert
3960 before the `succeed_n'. The `5' is the last two
3961 bytes of this `set_number_at', plus 3 bytes of
3962 the following `succeed_n'. */
3963 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3964 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3965 of the following `succeed_n'. */
3972 append a backward jump to the `succeed_n'
3973 that starts this interval.
3975 When we've reached this during matching,
3976 we'll have matched the interval once, so
3977 jump back only `upper_bound - 1' times. */
3983 /* The location we want to set is the second
3984 parameter of the `jump_n'; that is `b-2' as
3985 an absolute address. `laststart' will be
3986 the `set_number_at' we're about to insert;
3987 `laststart+3' the number to set, the source
3988 for the relative address. But we are
3989 inserting into the middle of the pattern --
3990 so everything is getting moved up by 5.
3991 Conclusion: (b - 2) - (laststart + 3) + 5,
3992 i.e., b - laststart.
3994 We insert this at the beginning of the loop
3995 so that if we fail during matching, we'll
3996 reinitialize the bounds. */
4001 }
4002 }
4006 invalid_interval:
4009 unfetch_interval:
4010 /* Match the characters as literals. */
4015 else
4017 }
4019 #ifdef emacs
4020 /* There is no way to specify the before_dot and after_dot
4021 operators. rms says this is ok. --karl */
4102 /* Can't back reference to a subexpression if inside of it. */
4115 else
4119 normal_backslash:
4120 /* You might think it would be useful for \ to mean
4121 not to translate; but if we don't translate it
4122 it will never match anything. */
4125 }
4130 /* Expects the character in `c'. */
4131 normal_char:
4132 /* If no exactn currently being built. */
4134 #ifdef WCHAR
4135 /* If last exactn handle binary(or character) and
4136 new exactn handle character(or binary). */
4140 /* If last exactn not at current position. */
4143 /* We have only one byte following the exactn for the count. */
4146 /* If followed by a repetition operator. */
4155 {
4156 /* Start building a new exactn. */
4160 #ifdef WCHAR
4161 /* Is this exactn binary data or character? */
4165 else
4167 #else
4171 }
4180 /* Through the pattern now. */
4188 /* If we don't want backtracking, force success
4189 the first time we reach the end of the compiled pattern. */
4193 #ifdef WCHAR
4197 #endif
4200 /* We have succeeded; set the length of the buffer. */
4201 #ifdef WCHAR
4203 #else
4205 #endif
4207 #ifdef DEBUG
4209 {
4212 }
4215 #ifndef MATCH_MAY_ALLOCATE
4216 /* Initialize the failure stack to the largest possible stack. This
4217 isn't necessary unless we're trying to avoid calling alloca in
4218 the search and match routines. */
4219 {
4222 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4223 is strictly greater than re_max_failures, the largest possible stack
4224 is 2 * re_max_failures failure points. */
4226 {
4229 # ifdef emacs
4231 fail_stack.stack
4234 else
4235 fail_stack.stack
4241 fail_stack.stack
4244 else
4245 fail_stack.stack
4250 }
4253 }
4259 /* Subroutines for `regex_compile'. */
4261 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4262 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4264 static void
4266 re_opcode_t op;
4269 {
4272 }
4275 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4276 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4278 static void
4280 re_opcode_t op;
4283 {
4287 }
4290 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4291 for OP followed by two-byte integer parameter ARG. */
4292 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4294 static void
4296 re_opcode_t op;
4300 {
4308 }
4311 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4312 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4314 static void
4316 re_opcode_t op;
4320 {
4328 }
4331 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4332 after an alternative or a begin-subexpression. We assume there is at
4333 least one character before the ^. */
4335 static boolean
4338 reg_syntax_t syntax;
4339 {
4343 return
4344 /* After a subexpression? */
4346 /* After an alternative? */
4348 }
4351 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4352 at least one character after the $, i.e., `P < PEND'. */
4354 static boolean
4357 reg_syntax_t syntax;
4358 {
4363 return
4364 /* Before a subexpression? */
4367 /* Before an alternative? */
4370 }
4374 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4375 false if it's not. */
4377 static boolean
4379 compile_stack_type compile_stack;
4380 regnum_t regnum;
4381 {
4386 this_element--)
4391 }
4394 #ifdef INSIDE_RECURSION
4396 #ifdef WCHAR
4397 /* This insert space, which size is "num", into the pattern at "loc".
4398 "end" must point the end of the allocated buffer. */
4399 static void
4404 {
4410 }
4413 #ifdef WCHAR
4414 static reg_errcode_t
4416 char_set)
4417 CHAR_T range_start_char;
4420 RE_TRANSLATE_TYPE translate;
4421 reg_syntax_t syntax;
4422 {
4425 reg_errcode_t ret;
4426 # ifdef _LIBC
4429 # endif
4433 # ifdef _LIBC
4436 {
4438 _NL_COLLATE_COLLSEQWC);
4443 {
4444 /* range_start is a collating symbol. */
4446 /* Retreive the index and get collation sequence value. */
4449 }
4450 else
4455 /* Report an error if the range is empty and the syntax prohibits
4456 this. */
4460 /* Insert space to the end of the char_ranges. */
4465 }
4466 else
4467 # endif
4468 {
4470 range_start_char;
4472 /* Report an error if the range is empty and the syntax prohibits
4473 this. */
4477 /* Insert space to the end of the char_ranges. */
4482 }
4483 /* Have to increment the pointer into the pattern string, so the
4484 caller isn't still at the ending character. */
4488 }
4490 /* Read the ending character of a range (in a bracket expression) from the
4491 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4492 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4493 Then we set the translation of all bits between the starting and
4494 ending characters (inclusive) in the compiled pattern B.
4496 Return an error code.
4498 We use these short variable names so we can use the same macros as
4499 `regex_compile' itself. */
4501 static reg_errcode_t
4505 RE_TRANSLATE_TYPE translate;
4506 reg_syntax_t syntax;
4508 {
4511 reg_errcode_t ret;
4512 # if _LIBC
4516 # else
4518 # endif
4523 /* Have to increment the pointer into the pattern string, so the
4524 caller isn't still at the ending character. */
4527 /* Report an error if the range is empty and the syntax prohibits this. */
4530 # if _LIBC
4532 _NL_COLLATE_COLLSEQMB);
4537 {
4541 {
4544 }
4545 }
4546 # else
4547 /* Here we see why `this_char' has to be larger than an `unsigned
4548 char' -- we would otherwise go into an infinite loop, since all
4549 characters <= 0xff. */
4551 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4552 and some compilers cast it to int implicitly, so following for_loop
4553 may fall to (almost) infinite loop.
4554 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4555 To avoid this, we cast p[0] to unsigned int and truncate it. */
4559 {
4562 }
4563 # endif
4566 }
4568 \f
4569 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4570 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4571 characters can start a string that matches the pattern. This fastmap
4572 is used by re_search to skip quickly over impossible starting points.
4574 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4575 area as BUFP->fastmap.
4577 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4578 the pattern buffer.
4580 Returns 0 if we succeed, -2 if an internal error. */
4582 #ifdef WCHAR
4583 /* local function for re_compile_fastmap.
4584 truncate wchar_t character to char. */
4587 static unsigned char
4589 CHAR_T c;
4590 {
4595 # ifdef _LIBC
4597 # else
4599 # endif
4601 }
4604 static int
4607 {
4609 #ifdef MATCH_MAY_ALLOCATE
4611 #endif
4612 #ifndef REGEX_MALLOC
4614 #endif
4618 #ifdef WCHAR
4619 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4620 pattern to (char*) in regex_compile. */
4629 #ifdef REL_ALLOC
4630 /* This holds the pointer to the failure stack, when
4631 it is allocated relocatably. */
4633 #endif
4635 /* Assume that each path through the pattern can be null until
4636 proven otherwise. We set this false at the bottom of switch
4637 statement, to which we get only if a particular path doesn't
4638 match the empty string. */
4641 /* We aren't doing a `succeed_n' to begin with. */
4652 {
4654 {
4655 /* We have reached the (effective) end of pattern. */
4657 {
4660 /* Reset for next path. */
4666 }
4667 else
4669 }
4671 /* We should never be about to go beyond the end of the pattern. */
4675 {
4677 /* I guess the idea here is to simply not bother with a fastmap
4678 if a backreference is used, since it's too hard to figure out
4679 the fastmap for the corresponding group. Setting
4680 `can_be_null' stops `re_search_2' from using the fastmap, so
4681 that is all we do. */
4687 /* Following are the cases which match a character. These end
4688 with `break'. */
4690 #ifdef WCHAR
4699 #ifdef MBS_SUPPORT
4703 #endif
4705 #ifdef WCHAR
4706 /* It is hard to distinguish fastmap from (multi byte) characters
4707 which depends on current locale. */
4723 /* Chars beyond end of map must be allowed. */
4748 {
4751 /* `.' matches anything ... */
4755 /* ... except perhaps newline. */
4759 /* Return if we have already set `can_be_null'; if we have,
4760 then the fastmap is irrelevant. Something's wrong here. */
4764 /* Otherwise, have to check alternative paths. */
4766 }
4768 #ifdef emacs
4785 /* All cases after this match the empty string. These end with
4786 `continue'. */
4820 /* Jump backward implies we just went through the body of a
4821 loop and matched nothing. Opcode jumped to should be
4822 `on_failure_jump' or `succeed_n'. Just treat it like an
4823 ordinary jump. For a * loop, it has pushed its failure
4824 point already; if so, discard that as redundant. */
4829 p++;
4833 /* If what's on the stack is where we are now, pop it. */
4843 handle_on_failure_jump:
4846 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4847 end of the pattern. We don't want to push such a point,
4848 since when we restore it above, entering the switch will
4849 increment `p' past the end of the pattern. We don't need
4850 to push such a point since we obviously won't find any more
4851 fastmap entries beyond `pend'. Such a pattern can match
4852 the null string, though. */
4854 {
4856 {
4859 }
4860 }
4861 else
4865 {
4868 }
4874 /* Get to the number of times to succeed. */
4877 /* Increment p past the n for when k != 0. */
4880 {
4884 }
4903 /* Getting here means we have found the possible starting
4904 characters for one path of the pattern -- and that the empty
4905 string does not match. We need not follow this path further.
4906 Instead, look at the next alternative (remembered on the
4907 stack), or quit if no more. The test at the top of the loop
4908 does these things. */
4913 /* Set `can_be_null' for the last path (also the first path, if the
4914 pattern is empty). */
4917 done:
4920 }
4924 int
4927 {
4928 # ifdef MBS_SUPPORT
4931 else
4932 # endif
4935 #ifdef _LIBC
4937 #endif
4938 \f
4940 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4941 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4942 this memory for recording register information. STARTS and ENDS
4943 must be allocated using the malloc library routine, and must each
4944 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4946 If NUM_REGS == 0, then subsequent matches should allocate their own
4947 register data.
4949 Unless this function is called, the first search or match using
4950 PATTERN_BUFFER will allocate its own register data, without
4951 freeing the old data. */
4953 void
4959 {
4961 {
4966 }
4967 else
4968 {
4972 }
4973 }
4974 #ifdef _LIBC
4976 #endif
4977 \f
4978 /* Searching routines. */
4980 /* Like re_search_2, below, but only one string is specified, and
4981 doesn't let you say where to stop matching. */
4983 int
4989 {
4992 }
4993 #ifdef _LIBC
4995 #endif
4998 /* Using the compiled pattern in BUFP->buffer, first tries to match the
4999 virtual concatenation of STRING1 and STRING2, starting first at index
5000 STARTPOS, then at STARTPOS + 1, and so on.
5002 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
5004 RANGE is how far to scan while trying to match. RANGE = 0 means try
5005 only at STARTPOS; in general, the last start tried is STARTPOS +
5006 RANGE.
5008 In REGS, return the indices of the virtual concatenation of STRING1
5009 and STRING2 that matched the entire BUFP->buffer and its contained
5010 subexpressions.
5012 Do not consider matching one past the index STOP in the virtual
5013 concatenation of STRING1 and STRING2.
5015 We return either the position in the strings at which the match was
5016 found, -1 if no match, or -2 if error (such as failure
5017 stack overflow). */
5019 int
5028 {
5029 # ifdef MBS_SUPPORT
5033 else
5034 # endif
5038 #ifdef _LIBC
5040 #endif
5044 #ifdef INSIDE_RECURSION
5046 #ifdef MATCH_MAY_ALLOCATE
5047 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
5048 #else
5049 # define FREE_VAR(var) if (var) free (var); var = NULL
5050 #endif
5052 #ifdef WCHAR
5053 # define MAX_ALLOCA_SIZE 2000
5055 # define FREE_WCS_BUFFERS() \
5056 do { \
5057 if (size1 > MAX_ALLOCA_SIZE) \
5058 { \
5059 free (wcs_string1); \
5060 free (mbs_offset1); \
5061 } \
5062 else \
5063 { \
5064 FREE_VAR (wcs_string1); \
5065 FREE_VAR (mbs_offset1); \
5066 } \
5067 if (size2 > MAX_ALLOCA_SIZE) \
5068 { \
5069 free (wcs_string2); \
5070 free (mbs_offset2); \
5071 } \
5072 else \
5073 { \
5074 FREE_VAR (wcs_string2); \
5075 FREE_VAR (mbs_offset2); \
5076 } \
5077 } while (0)
5079 #endif
5082 static int
5092 {
5098 #ifdef WCHAR
5099 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5101 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5103 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5105 /* They hold whether each wchar_t is binary data or not. */
5109 /* Check for out-of-range STARTPOS. */
5113 /* Fix up RANGE if it might eventually take us outside
5114 the virtual concatenation of STRING1 and STRING2.
5115 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5121 /* If the search isn't to be a backwards one, don't waste time in a
5122 search for a pattern that must be anchored. */
5125 /* `begline' is like `begbuf' if it cannot match at newlines. */
5128 {
5131 else
5133 }
5135 #ifdef emacs
5136 /* In a forward search for something that starts with \=.
5137 don't keep searching past point. */
5139 {
5143 }
5146 /* Update the fastmap now if not correct already. */
5151 #ifdef WCHAR
5152 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5153 fill them with converted string. */
5155 {
5157 {
5161 }
5162 else
5163 {
5167 }
5169 {
5171 {
5175 }
5176 else
5177 {
5181 }
5183 }
5189 else
5191 }
5193 {
5195 {
5199 }
5200 else
5201 {
5205 }
5207 {
5211 else
5214 }
5220 else
5222 }
5226 /* Loop through the string, looking for a place to start matching. */
5228 {
5229 /* If a fastmap is supplied, skip quickly over characters that
5230 cannot be the start of a match. If the pattern can match the
5231 null string, however, we don't need to skip characters; we want
5232 the first null string. */
5234 {
5236 {
5246 /* Written out as an if-else to avoid testing `translate'
5247 inside the loop. */
5252 range--;
5253 else
5255 range--;
5258 }
5260 {
5267 }
5268 }
5270 /* If can't match the null string, and that's all we have left, fail. */
5273 {
5274 #ifdef WCHAR
5276 #endif
5278 }
5280 #ifdef WCHAR
5291 #ifndef REGEX_MALLOC
5292 # ifdef C_ALLOCA
5294 # endif
5295 #endif
5298 {
5299 #ifdef WCHAR
5301 #endif
5303 }
5306 {
5307 #ifdef WCHAR
5309 #endif
5311 }
5313 advance:
5317 {
5318 range--;
5319 startpos++;
5320 }
5321 else
5322 {
5323 range++;
5324 startpos--;
5325 }
5326 }
5327 #ifdef WCHAR
5329 #endif
5331 }
5333 #ifdef WCHAR
5334 /* This converts PTR, a pointer into one of the search wchar_t strings
5335 `string1' and `string2' into an multibyte string offset from the
5336 beginning of that string. We use mbs_offset to optimize.
5337 See convert_mbs_to_wcs. */
5338 # define POINTER_TO_OFFSET(ptr) \
5339 (FIRST_STRING_P (ptr) \
5340 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5341 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5342 + csize1)))
5344 /* This converts PTR, a pointer into one of the search strings `string1'
5345 and `string2' into an offset from the beginning of that string. */
5346 # define POINTER_TO_OFFSET(ptr) \
5347 (FIRST_STRING_P (ptr) \
5348 ? ((regoff_t) ((ptr) - string1)) \
5349 : ((regoff_t) ((ptr) - string2 + size1)))
5352 /* Macros for dealing with the split strings in re_match_2. */
5354 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5356 /* Call before fetching a character with *d. This switches over to
5357 string2 if necessary. */
5358 #define PREFETCH() \
5359 while (d == dend) \
5360 { \
5362 if (dend == end_match_2) \
5363 goto fail; \
5365 d = string2; \
5366 dend = end_match_2; \
5367 }
5369 /* Test if at very beginning or at very end of the virtual concatenation
5370 of `string1' and `string2'. If only one string, it's `string2'. */
5371 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5372 #define AT_STRINGS_END(d) ((d) == end2)
5375 /* Test if D points to a character which is word-constituent. We have
5376 two special cases to check for: if past the end of string1, look at
5377 the first character in string2; and if before the beginning of
5378 string2, look at the last character in string1. */
5379 #ifdef WCHAR
5380 /* Use internationalized API instead of SYNTAX. */
5381 # define WORDCHAR_P(d) \
5382 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5383 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5384 || ((d) == end1 ? *string2 \
5387 # define WORDCHAR_P(d) \
5388 (SYNTAX ((d) == end1 ? *string2 \
5389 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5390 == Sword)
5393 /* Disabled due to a compiler bug -- see comment at case wordbound */
5394 #if 0
5395 /* Test if the character before D and the one at D differ with respect
5396 to being word-constituent. */
5397 #define AT_WORD_BOUNDARY(d) \
5398 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5399 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5400 #endif
5402 /* Free everything we malloc. */
5403 #ifdef MATCH_MAY_ALLOCATE
5404 # ifdef WCHAR
5405 # define FREE_VARIABLES() \
5406 do { \
5407 REGEX_FREE_STACK (fail_stack.stack); \
5408 FREE_VAR (regstart); \
5409 FREE_VAR (regend); \
5410 FREE_VAR (old_regstart); \
5411 FREE_VAR (old_regend); \
5412 FREE_VAR (best_regstart); \
5413 FREE_VAR (best_regend); \
5414 FREE_VAR (reg_info); \
5415 FREE_VAR (reg_dummy); \
5416 FREE_VAR (reg_info_dummy); \
5417 if (!cant_free_wcs_buf) \
5418 { \
5419 FREE_VAR (string1); \
5420 FREE_VAR (string2); \
5421 FREE_VAR (mbs_offset1); \
5422 FREE_VAR (mbs_offset2); \
5423 } \
5424 } while (0)
5426 # define FREE_VARIABLES() \
5427 do { \
5428 REGEX_FREE_STACK (fail_stack.stack); \
5429 FREE_VAR (regstart); \
5430 FREE_VAR (regend); \
5431 FREE_VAR (old_regstart); \
5432 FREE_VAR (old_regend); \
5433 FREE_VAR (best_regstart); \
5434 FREE_VAR (best_regend); \
5435 FREE_VAR (reg_info); \
5436 FREE_VAR (reg_dummy); \
5437 FREE_VAR (reg_info_dummy); \
5438 } while (0)
5440 #else
5441 # ifdef WCHAR
5442 # define FREE_VARIABLES() \
5443 do { \
5444 if (!cant_free_wcs_buf) \
5445 { \
5446 FREE_VAR (string1); \
5447 FREE_VAR (string2); \
5448 FREE_VAR (mbs_offset1); \
5449 FREE_VAR (mbs_offset2); \
5450 } \
5451 } while (0)
5457 /* These values must meet several constraints. They must not be valid
5458 register values; since we have a limit of 255 registers (because
5459 we use only one byte in the pattern for the register number), we can
5460 use numbers larger than 255. They must differ by 1, because of
5461 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5462 be larger than the value for the highest register, so we do not try
5463 to actually save any registers when none are active. */
5464 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5465 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5466 \f
5468 /* Matching routines. */
5471 /* re_match is like re_match_2 except it takes only a single string. */
5473 int
5479 {
5481 # ifdef MBS_SUPPORT
5486 else
5487 # endif
5490 # ifndef REGEX_MALLOC
5491 # ifdef C_ALLOCA
5493 # endif
5494 # endif
5496 }
5497 # ifdef _LIBC
5499 # endif
5504 #ifdef INSIDE_RECURSION
5518 /* re_match_2 matches the compiled pattern in BUFP against the
5519 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5520 and SIZE2, respectively). We start matching at POS, and stop
5521 matching at STOP.
5523 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5524 store offsets for the substring each group matched in REGS. See the
5525 documentation for exactly how many groups we fill.
5527 We return -1 if no match, -2 if an internal error (such as the
5528 failure stack overflowing). Otherwise, we return the length of the
5529 matched substring. */
5531 int
5539 {
5541 # ifdef MBS_SUPPORT
5546 else
5547 # endif
5551 #ifndef REGEX_MALLOC
5552 # ifdef C_ALLOCA
5554 # endif
5555 #endif
5557 }
5558 #ifdef _LIBC
5560 #endif
5564 #ifdef INSIDE_RECURSION
5566 #ifdef WCHAR
5569 /* This check the substring (from 0, to length) of the multibyte string,
5570 to which offset_buffer correspond. And count how many wchar_t_characters
5571 the substring occupy. We use offset_buffer to optimization.
5572 See convert_mbs_to_wcs. */
5574 static int
5578 {
5581 /* Check whether the size is valid. */
5588 /* If there are no multibyte character, offset_buffer[i] == i.
5589 Optmize for this case. */
5593 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5598 {
5606 else
5608 }
5611 }
5614 /* This is a separate function so that we can force an alloca cleanup
5615 afterwards. */
5616 #ifdef WCHAR
5617 static int
5627 /* string1 == string2 == NULL means string1/2, size1/2 and
5628 mbs_offset1/2 need seting up in this function. */
5629 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5631 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5633 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5636 static int
5646 {
5647 /* General temporaries. */
5650 #ifdef WCHAR
5651 /* They hold whether each wchar_t is binary data or not. */
5653 /* If true, we can't free string1/2, mbs_offset1/2. */
5657 /* Just past the end of the corresponding string. */
5660 /* Pointers into string1 and string2, just past the last characters in
5661 each to consider matching. */
5664 /* Where we are in the data, and the end of the current string. */
5667 /* Where we are in the pattern, and the end of the pattern. */
5668 #ifdef WCHAR
5676 /* Mark the opcode just after a start_memory, so we can test for an
5677 empty subpattern when we get to the stop_memory. */
5680 /* We use this to map every character in the string. */
5683 /* Failure point stack. Each place that can handle a failure further
5684 down the line pushes a failure point on this stack. It consists of
5685 restart, regend, and reg_info for all registers corresponding to
5686 the subexpressions we're currently inside, plus the number of such
5687 registers, and, finally, two char *'s. The first char * is where
5688 to resume scanning the pattern; the second one is where to resume
5689 scanning the strings. If the latter is zero, the failure point is
5690 a ``dummy''; if a failure happens and the failure point is a dummy,
5691 it gets discarded and the next next one is tried. */
5694 #endif
5695 #ifdef DEBUG
5698 #endif
5700 #ifdef REL_ALLOC
5701 /* This holds the pointer to the failure stack, when
5702 it is allocated relocatably. */
5704 #endif
5706 /* We fill all the registers internally, independent of what we
5707 return, for use in backreferences. The number here includes
5708 an element for register zero. */
5711 /* The currently active registers. */
5715 /* Information on the contents of registers. These are pointers into
5716 the input strings; they record just what was matched (on this
5717 attempt) by a subexpression part of the pattern, that is, the
5718 regnum-th regstart pointer points to where in the pattern we began
5719 matching and the regnum-th regend points to right after where we
5720 stopped matching the regnum-th subexpression. (The zeroth register
5721 keeps track of what the whole pattern matches.) */
5724 #endif
5726 /* If a group that's operated upon by a repetition operator fails to
5727 match anything, then the register for its start will need to be
5728 restored because it will have been set to wherever in the string we
5729 are when we last see its open-group operator. Similarly for a
5730 register's end. */
5733 #endif
5735 /* The is_active field of reg_info helps us keep track of which (possibly
5736 nested) subexpressions we are currently in. The matched_something
5737 field of reg_info[reg_num] helps us tell whether or not we have
5738 matched any of the pattern so far this time through the reg_num-th
5739 subexpression. These two fields get reset each time through any
5740 loop their register is in. */
5743 #endif
5745 /* The following record the register info as found in the above
5746 variables when we find a match better than any we've seen before.
5747 This happens as we backtrack through the failure points, which in
5748 turn happens only if we have not yet matched the entire string. */
5752 #endif
5754 /* Logically, this is `best_regend[0]'. But we don't want to have to
5755 allocate space for that if we're not allocating space for anything
5756 else (see below). Also, we never need info about register 0 for
5757 any of the other register vectors, and it seems rather a kludge to
5758 treat `best_regend' differently than the rest. So we keep track of
5759 the end of the best match so far in a separate variable. We
5760 initialize this to NULL so that when we backtrack the first time
5761 and need to test it, it's not garbage. */
5764 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5767 /* Used when we pop values we don't care about. */
5771 #endif
5773 #ifdef DEBUG
5774 /* Counts the total number of registers pushed. */
5776 #endif
5782 #ifdef MATCH_MAY_ALLOCATE
5783 /* Do not bother to initialize all the register variables if there are
5784 no groups in the pattern, as it takes a fair amount of time. If
5785 there are groups, we include space for register 0 (the whole
5786 pattern), even though we never use it, since it simplifies the
5787 array indexing. We should fix this. */
5789 {
5802 {
5805 }
5806 }
5807 else
5808 {
5809 /* We must initialize all our variables to NULL, so that
5810 `FREE_VARIABLES' doesn't try to free them. */
5814 }
5817 /* The starting position is bogus. */
5818 #ifdef WCHAR
5822 #endif
5823 {
5826 }
5828 #ifdef WCHAR
5829 /* Allocate wchar_t array for string1 and string2 and
5830 fill them with converted string. */
5832 {
5833 /* We need seting up buffers here. */
5835 /* We must free wcs buffers in this function. */
5839 {
5844 {
5849 }
5850 }
5852 {
5857 {
5864 }
5869 }
5870 }
5872 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5873 pattern to (char*) in regex_compile. */
5879 /* Initialize subexpression text positions to -1 to mark ones that no
5880 start_memory/stop_memory has been seen for. Also initialize the
5881 register information struct. */
5883 {
5891 }
5893 /* We move `string1' into `string2' if the latter's empty -- but not if
5894 `string1' is null. */
5896 {
5901 #ifdef WCHAR
5906 #endif
5907 }
5911 /* Compute where to stop matching, within the two strings. */
5912 #ifdef WCHAR
5914 {
5918 }
5919 else
5920 {
5926 }
5931 }
5932 #else
5934 {
5937 }
5938 else
5939 {
5942 }
5945 /* `p' scans through the pattern as `d' scans through the data.
5946 `dend' is the end of the input string that `d' points within. `d'
5947 is advanced into the following input string whenever necessary, but
5948 this happens before fetching; therefore, at the beginning of the
5949 loop, `d' can be pointing at the end of a string, but it cannot
5950 equal `string2'. */
5951 #ifdef WCHAR
5953 {
5957 }
5958 else
5959 {
5963 }
5969 }
5970 #else
5972 {
5975 }
5976 else
5977 {
5980 }
5989 /* This loops over pattern commands. It exits by returning from the
5990 function if the match is complete, or it drops through if the match
5991 fails at this starting point in the input data. */
5993 {
5994 #ifdef _LIBC
5996 #else
5998 #endif
6004 /* If we haven't matched the entire string, and we want the
6005 longest match, try backtracking. */
6007 {
6008 /* 1 if this match ends in the same string (string1 or string2)
6009 as the best previous match. */
6012 /* 1 if this match is the best seen so far. */
6013 boolean best_match_p;
6015 /* AIX compiler got confused when this was combined
6016 with the previous declaration. */
6019 else
6027 /* If exceeds best match so far, save it. */
6029 {
6036 {
6039 }
6040 }
6042 }
6044 /* If no failure points, don't restore garbage. And if
6045 last match is real best match, don't restore second
6046 best one. */
6048 {
6049 restore_best_regs:
6050 /* Restore best match. It may happen that `dend ==
6051 end_match_1' while the restored d is in string2.
6052 For example, the pattern `x.*y.*z' against the
6053 strings `x-' and `y-z-', if the two strings are
6054 not consecutive in memory. */
6062 {
6065 }
6066 }
6069 succeed_label:
6071 /* If caller wants register contents data back, do it. */
6073 {
6074 /* Have the register data arrays been allocated? */
6077 extra element beyond `num_regs' for the `-1' marker
6078 GNU code uses. */
6083 {
6086 }
6088 }
6091 allocated, reallocate them. If we need fewer, just
6092 leave it alone. */
6094 {
6099 {
6102 }
6103 }
6104 }
6105 else
6106 {
6107 /* These braces fend off a "empty body in an else-statement"
6108 warning under GCC when assert expands to nothing. */
6110 }
6112 /* Convert the pointer data in `regstart' and `regend' to
6113 indices. Register zero has to be set differently,
6114 since we haven't kept track of any info for it. */
6116 {
6118 #ifdef WCHAR
6122 else
6125 #else
6130 }
6132 /* Go through the first `min (num_regs, regs->num_regs)'
6133 registers, since that is all we initialized. */
6135 mcnt++)
6136 {
6139 else
6140 {
6145 }
6146 }
6148 /* If the regs structure we return has more elements than
6149 were in the pattern, set the extra elements to -1. If
6150 we (re)allocated the registers, this is the case,
6151 because we always allocate enough to have at least one
6152 -1 at the end. */
6162 #ifdef WCHAR
6165 else
6167 csize1;
6169 #else
6171 ? string1
6179 }
6181 /* Otherwise match next pattern command. */
6183 {
6184 /* Ignore these. Used to ignore the n of succeed_n's which
6185 currently have n == 0. */
6194 /* Match the next n pattern characters exactly. The following
6195 byte in the pattern defines n, and the n bytes after that
6196 are the characters to match. */
6198 #ifdef MBS_SUPPORT
6200 #endif
6204 /* This is written out as an if-else so we don't waste time
6205 testing `translate' inside the loop. */
6207 {
6208 do
6209 {
6211 #ifdef WCHAR
6213 {
6217 }
6218 else
6219 {
6222 }
6223 #else
6228 }
6230 }
6231 else
6232 {
6233 do
6234 {
6237 }
6239 }
6244 /* Match any character except possibly a newline or a null. */
6256 d++;
6262 {
6264 #ifdef WCHAR
6268 #define WORK_BUFFER_SIZE 128
6270 # ifdef _LIBC
6279 #ifdef WCHAR
6280 # ifdef _LIBC
6289 /* p points charset[6], so the address of the next instruction
6290 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6291 where l=length of char_classes, m=length of collating_symbol,
6292 n=equivalence_class, o=length of char_range,
6293 p'=length of character. */
6295 /* Update p to indicate the next instruction. */
6299 /* match with char_class? */
6301 {
6308 # ifdef _LIBC
6311 # else
6314 # endif
6315 }
6317 /* match with collating_symbol? */
6318 # ifdef _LIBC
6320 {
6325 workp++)
6326 {
6334 {
6335 /* Update d, however d will be incremented at
6336 char_set_matched:, we decrement d here. */
6339 }
6340 }
6341 }
6343 # endif
6344 /* If we can't look up collation data, we use wcscoll
6345 instead. */
6346 {
6348 {
6350 # ifdef _LIBC
6352 # else
6354 # endif
6356 /* If wcscoll(the collating symbol, whole string) > 0,
6357 any substring of the string never match with the
6358 collating symbol. */
6359 # ifdef _LIBC
6361 # else
6363 # endif
6364 {
6367 }
6369 /* First, we compare the collating symbol with
6370 the first character of the string.
6371 If it don't match, we add the next character to
6372 the compare buffer in turn. */
6374 {
6377 {
6382 }
6384 /* add next character to the compare buffer. */
6388 # ifdef _LIBC
6390 # else
6392 # endif
6397 /* (str_buf > workp) indicate (str_buf + X > workp),
6398 because for all X (str_buf + X > str_buf).
6399 So we don't need continue this loop. */
6402 /* Otherwise(str_buf < workp),
6403 (str_buf+next_character) may equals (workp).
6404 So we continue this loop. */
6405 }
6406 /* not matched */
6410 }
6411 }
6412 /* match with equivalence_class? */
6413 # ifdef _LIBC
6415 {
6417 /* Try to match the equivalence class against
6418 those known to the collate implementation. */
6427 /* This #include defines a local function! */
6428 # include <locale/weightwc.h>
6439 /* Write 1 collating element to str_buf, and
6440 get its index. */
6444 {
6447 {
6452 }
6456 }
6458 /* Update d, however d will be incremented at
6459 char_set_matched:, we decrement d here. */
6462 {
6465 else
6466 {
6469 }
6470 }
6475 workp++)
6476 {
6478 /* We already checked idx != 0 in regex_compile. */
6481 {
6489 }
6490 }
6491 /* not matched */
6494 }
6496 # endif
6497 /* If we can't look up collation data, we use wcscoll
6498 instead. */
6499 {
6501 {
6503 # ifdef _LIBC
6505 # else
6507 # endif
6509 /* If wcscoll(the collating symbol, whole string) > 0,
6510 any substring of the string never match with the
6511 collating symbol. */
6512 # ifdef _LIBC
6514 # else
6516 # endif
6517 {
6520 }
6522 /* First, we compare the equivalence class with
6523 the first character of the string.
6524 If it don't match, we add the next character to
6525 the compare buffer in turn. */
6527 {
6530 {
6535 }
6537 /* add next character to the compare buffer. */
6541 # ifdef _LIBC
6543 # else
6545 # endif
6551 /* (str_buf > workp) indicate (str_buf + X > workp),
6552 because for all X (str_buf + X > str_buf).
6553 So we don't need continue this loop. */
6556 /* Otherwise(str_buf < workp),
6557 (str_buf+next_character) may equals (workp).
6558 So we continue this loop. */
6559 }
6560 /* not matched */
6564 }
6565 }
6567 /* match with char_range? */
6568 # ifdef _LIBC
6570 {
6578 {
6581 /* We already compute the collation sequence value
6582 of the characters (or collating symbols). */
6588 }
6589 }
6590 else
6591 # endif
6592 {
6593 /* We set range_start_char at str_buf[0], range_end_char
6594 at str_buf[4], and compared char at str_buf[2]. */
6600 {
6603 /* match if (range_start_char <= c <= range_end_char). */
6605 /* If range_start(or end) < 0, we assume -range_start(end)
6606 is the offset of the collating symbol which is specified
6607 as the character of the range start(end). */
6609 /* range_start */
6612 else
6613 {
6616 }
6618 /* range_end */
6621 else
6622 {
6625 }
6627 # ifdef _LIBC
6630 # else
6633 # endif
6635 }
6636 }
6638 /* match with char? */
6645 char_set_matched:
6647 #else
6648 /* Cast to `unsigned' instead of `unsigned char' in case the
6649 bit list is a full 32 bytes long. */
6657 #undef WORK_BUFFER_SIZE
6660 d++;
6662 }
6665 /* The beginning of a group is represented by start_memory.
6666 The arguments are the register number in the next byte, and the
6667 number of groups inner to this one in the next. The text
6668 matched within the group is recorded (in the internal
6669 registers data structure) under the register number. */
6674 /* Find out if this group can match the empty string. */
6681 /* Save the position in the string where we were the last time
6682 we were at this open-group operator in case the group is
6683 operated upon by a repetition operator, e.g., with `(a*)*b'
6684 against `ab'; then we want to ignore where we are now in
6685 the string in case this attempt to match fails. */
6698 /* Clear this whenever we change the register activity status. */
6701 /* This is the new highest active register. */
6704 /* If nothing was active before, this is the new lowest active
6705 register. */
6709 /* Move past the register number and inner group count. */
6716 /* The stop_memory opcode represents the end of a group. Its
6717 arguments are the same as start_memory's: the register
6718 number, and the number of inner groups. */
6723 /* We need to save the string position the last time we were at
6724 this close-group operator in case the group is operated
6725 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6726 against `aba'; then we want to ignore where we are now in
6727 the string in case this attempt to match fails. */
6737 /* This register isn't active anymore. */
6740 /* Clear this whenever we change the register activity status. */
6743 /* If this was the only register active, nothing is active
6744 anymore. */
6746 {
6749 }
6750 else
6752 it isn't necessarily one less than now: consider
6753 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6754 new highest active register is 1. */
6757 r--;
6759 /* If we end up at register zero, that means that we saved
6760 the registers as the result of an `on_failure_jump', not
6761 a `start_memory', and we jumped to past the innermost
6762 `stop_memory'. For example, in ((.)*) we save
6763 registers 1 and 2 as a result of the *, but when we pop
6764 back to the second ), we are at the stop_memory 1.
6765 Thus, nothing is active. */
6767 {
6770 }
6771 else
6773 }
6775 /* If just failed to match something this time around with a
6776 group that's operated on by a repetition operator, try to
6777 force exit from the ``loop'', and restore the register
6778 information for this group that we had before trying this
6779 last match. */
6783 {
6789 {
6803 }
6806 /* If the next operation is a jump backwards in the pattern
6807 to an on_failure_jump right before the start_memory
6808 corresponding to this stop_memory, exit from the loop
6809 by forcing a failure after pushing on the stack the
6810 on_failure_jump's jump in the pattern, and d. */
6814 {
6815 /* If this group ever matched anything, then restore
6816 what its registers were before trying this last
6817 failed match, e.g., with `(a*)*b' against `ab' for
6818 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6819 against `aba' for regend[3].
6821 Also restore the registers for inner groups for,
6822 e.g., `((a*)(b*))*' against `aba' (register 3 would
6823 otherwise get trashed). */
6826 {
6831 /* Restore this and inner groups' (if any) registers. */
6833 r++)
6834 {
6837 /* xx why this test? */
6840 }
6841 }
6842 p1++;
6847 }
6848 }
6850 /* Move past the register number and the inner group count. */
6855 /* \<digit> has been turned into a `duplicate' command which is
6856 followed by the numeric value of <digit> as the register number. */
6858 {
6863 /* Can't back reference a group which we've never matched. */
6867 /* Where in input to try to start matching. */
6870 /* Where to stop matching; if both the place to start and
6871 the place to stop matching are in the same string, then
6872 set to the place to stop, otherwise, for now have to use
6873 the end of the first string. */
6879 {
6880 /* If necessary, advance to next segment in register
6881 contents. */
6883 {
6887 /* End of string1 => advance to string2. */
6890 }
6891 /* At end of register contents => success */
6894 /* If necessary, advance to next segment in data. */
6897 /* How many characters left in this segment to match. */
6900 /* Want how many consecutive characters we can match in
6901 one shot, so, if necessary, adjust the count. */
6905 /* Compare that many; failure if mismatch, else move
6906 past them. */
6913 /* Do this because we've match some characters. */
6915 }
6916 }
6920 /* begline matches the empty string at the beginning of the string
6921 (unless `not_bol' is set in `bufp'), and, if
6922 `newline_anchor' is set, after newlines. */
6927 {
6929 }
6931 {
6933 }
6934 /* In all other cases, we fail. */
6938 /* endline is the dual of begline. */
6943 {
6945 }
6947 /* We have to ``prefetch'' the next character. */
6950 {
6952 }
6956 /* Match at the very beginning of the data. */
6964 /* Match at the very end of the data. */
6972 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6973 pushes NULL as the value for the string on the stack. Then
6974 `pop_failure_point' will keep the current value for the
6975 string, instead of restoring it. To see why, consider
6976 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6977 then the . fails against the \n. But the next thing we want
6978 to do is match the \n against the \n; if we restored the
6979 string value, we would be back at the foo.
6981 Because this is used only in specific cases, we don't need to
6982 check all the things that `on_failure_jump' does, to make
6983 sure the right things get saved on the stack. Hence we don't
6984 share its code. The only reason to push anything on the
6985 stack at all is that otherwise we would have to change
6986 `anychar's code to do something besides goto fail in this
6987 case; that seems worse than this. */
6992 #ifdef _LIBC
6994 #else
6996 #endif
7002 /* Uses of on_failure_jump:
7004 Each alternative starts with an on_failure_jump that points
7005 to the beginning of the next alternative. Each alternative
7006 except the last ends with a jump that in effect jumps past
7007 the rest of the alternatives. (They really jump to the
7008 ending jump of the following alternative, because tensioning
7009 these jumps is a hassle.)
7011 Repeats start with an on_failure_jump that points past both
7012 the repetition text and either the following jump or
7013 pop_failure_jump back to this on_failure_jump. */
7015 on_failure:
7019 #ifdef _LIBC
7021 #else
7023 #endif
7025 /* If this on_failure_jump comes right before a group (i.e.,
7026 the original * applied to a group), save the information
7027 for that group and all inner ones, so that if we fail back
7028 to this point, the group's information will be correct.
7029 For example, in \(a*\)*\1, we need the preceding group,
7030 and in \(zz\(a*\)b*\)\2, we need the inner group. */
7032 /* We can't use `p' to check ahead because we push
7033 a failure point to `p + mcnt' after we do this. */
7036 /* We need to skip no_op's before we look for the
7037 start_memory in case this on_failure_jump is happening as
7038 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
7039 against aba. */
7041 p1++;
7044 {
7045 /* We have a new highest active register now. This will
7046 get reset at the start_memory we are about to get to,
7047 but we will have saved all the registers relevant to
7048 this repetition op, as described above. */
7052 }
7059 /* A smart repeat ends with `maybe_pop_jump'.
7060 We change it to either `pop_failure_jump' or `jump'. */
7064 {
7067 /* Compare the beginning of the repeat with what in the
7068 pattern follows its end. If we can establish that there
7069 is nothing that they would both match, i.e., that we
7070 would have to backtrack because of (as in, e.g., `a*a')
7071 then we can change to pop_failure_jump, because we'll
7072 never have to backtrack.
7074 This is not true in the case of alternatives: in
7075 `(a|ab)*' we do need to backtrack to the `ab' alternative
7076 (e.g., if the string was `ab'). But instead of trying to
7077 detect that here, the alternative has put on a dummy
7078 failure point which is what we will end up popping. */
7080 /* Skip over open/close-group commands.
7081 If what follows this loop is a ...+ construct,
7082 look at what begins its body, since we will have to
7083 match at least one of that. */
7085 {
7093 else
7095 }
7098 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7099 to the `maybe_finalize_jump' of this case. Examine what
7100 follows. */
7102 /* If we're at the end of the pattern, we can change. */
7104 {
7105 /* Consider what happens when matching ":\(.*\)"
7106 against ":/". I don't really understand this code
7107 yet. */
7109 pop_failure_jump;
7110 DEBUG_PRINT1
7112 }
7115 #ifdef MBS_SUPPORT
7117 #endif
7119 {
7120 register UCHAR_T c
7124 #ifdef MBS_SUPPORT
7126 #endif
7128 {
7130 pop_failure_jump;
7131 #ifdef WCHAR
7135 #else
7139 #endif
7140 }
7142 #ifndef WCHAR
7145 {
7152 /* `not' is equal to 1 if c would match, which means
7153 that we can't change to pop_failure_jump. */
7155 {
7158 }
7159 }
7161 }
7162 #ifndef WCHAR
7164 {
7165 /* We win if the first character of the loop is not part
7166 of the charset. */
7171 {
7174 }
7177 {
7179 /* We win if the charset_not inside the loop
7180 lists every character listed in the charset after. */
7188 {
7191 }
7192 }
7194 {
7196 /* We win if the charset inside the loop
7197 has no overlap with the one after the loop. */
7200 idx++)
7205 {
7208 }
7209 }
7210 }
7212 }
7215 {
7219 }
7220 /* Note fall through. */
7223 /* The end of a simple repeat has a pop_failure_jump back to
7224 its matching on_failure_jump, where the latter will push a
7225 failure point. The pop_failure_jump takes off failure
7226 points put on by this pop_failure_jump's matching
7227 on_failure_jump; we got through the pattern to here from the
7228 matching on_failure_jump, so didn't fail. */
7230 {
7231 /* We need to pass separate storage for the lowest and
7232 highest registers, even though we don't care about the
7233 actual values. Otherwise, we will restore only one
7234 register from the stack, since lowest will == highest in
7235 `pop_failure_point'. */
7244 }
7245 /* Note fall through. */
7247 unconditional_jump:
7248 #ifdef _LIBC
7250 #else
7252 #endif
7253 /* Note fall through. */
7255 /* Unconditionally jump (without popping any failure points). */
7260 #ifdef _LIBC
7262 #else
7264 #endif
7268 /* We need this opcode so we can detect where alternatives end
7269 in `group_match_null_string_p' et al. */
7275 /* Normally, the on_failure_jump pushes a failure point, which
7276 then gets popped at pop_failure_jump. We will end up at
7277 pop_failure_jump, also, and with a pattern of, say, `a+', we
7278 are skipping over the on_failure_jump, so we have to push
7279 something meaningless for pop_failure_jump to pop. */
7282 /* It doesn't matter what we push for the string here. What
7283 the code at `fail' tests is the value for the pattern. */
7288 /* At the end of an alternative, we need to push a dummy failure
7289 point in case we are followed by a `pop_failure_jump', because
7290 we don't want the failure point for the alternative to be
7291 popped. For example, matching `(a|ab)*' against `aab'
7292 requires that we match the `ab' alternative. */
7295 /* See comments just above at `dummy_failure_jump' about the
7296 two zeroes. */
7300 /* Have to succeed matching what follows at least n times.
7301 After that, handle like `on_failure_jump'. */
7307 /* Originally, this is how many times we HAVE to succeed. */
7309 {
7310 mcnt--;
7313 #ifdef _LIBC
7316 #else
7319 #endif
7320 }
7322 {
7323 #ifdef _LIBC
7326 #else
7331 #ifdef WCHAR
7333 #else
7338 }
7345 /* Originally, this is how many times we CAN jump. */
7347 {
7348 mcnt--;
7351 #ifdef _LIBC
7353 mcnt);
7354 #else
7356 mcnt);
7359 }
7360 /* If don't have to jump any more, skip over the rest of command. */
7361 else
7366 {
7372 #ifdef _LIBC
7374 #else
7376 #endif
7379 }
7381 #if 0
7382 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7383 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7384 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7385 macro and introducing temporary variables works around the bug. */
7398 #else
7400 {
7412 }
7415 {
7427 }
7428 #endif
7444 #ifdef emacs
7471 matchsyntax:
7473 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7474 d++;
7488 matchnotsyntax:
7490 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7491 d++;
7504 d++;
7513 d++;
7519 }
7523 /* We goto here if a matching operation fails. */
7524 fail:
7532 /* If this failure point is a dummy, try the next one. */
7536 /* If we failed to the end of the pattern, don't examine *p. */
7539 {
7542 /* If failed to a backwards jump that's part of a repetition
7543 loop, need to pop this failure point and use the next one. */
7545 {
7556 || (!is_a_jump_n
7562 }
7563 }
7567 }
7568 else
7579 \f
7580 /* Subroutine definitions for re_match_2. */
7583 /* We are passed P pointing to a register number after a start_memory.
7585 Return true if the pattern up to the corresponding stop_memory can
7586 match the empty string, and false otherwise.
7588 If we find the matching stop_memory, sets P to point to one past its number.
7589 Otherwise, sets P to an undefined byte less than or equal to END.
7591 We don't handle duplicates properly (yet). */
7593 static boolean
7597 {
7599 /* Point to after the args to the start_memory. */
7603 {
7604 /* Skip over opcodes that can match nothing, and return true or
7605 false, as appropriate, when we get to one that can't, or to the
7606 matching stop_memory. */
7609 {
7610 /* Could be either a loop or a series of alternatives. */
7612 p1++;
7615 /* If the next operation is not a jump backwards in the
7616 pattern. */
7619 {
7620 /* Go through the on_failure_jumps of the alternatives,
7621 seeing if any of the alternatives cannot match nothing.
7622 The last alternative starts with only a jump,
7623 whereas the rest start with on_failure_jump and end
7624 with a jump, e.g., here is the pattern for `a|b|c':
7626 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7627 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7628 /exactn/1/c
7630 So, we have to first go through the first (n-1)
7631 alternatives and then deal with the last one separately. */
7634 /* Deal with the first (n-1) alternatives, which start
7635 with an on_failure_jump (see above) that jumps to right
7636 past a jump_past_alt. */
7639 jump_past_alt)
7640 {
7641 /* `mcnt' holds how many bytes long the alternative
7642 is, including the ending `jump_past_alt' and
7643 its number. */
7647 reg_info))
7650 /* Move to right after this alternative, including the
7651 jump_past_alt. */
7654 /* Break if it's the beginning of an n-th alternative
7655 that doesn't begin with an on_failure_jump. */
7659 /* Still have to check that it's not an n-th
7660 alternative that starts with an on_failure_jump. */
7661 p1++;
7664 jump_past_alt)
7665 {
7666 /* Get to the beginning of the n-th alternative. */
7669 }
7670 }
7672 /* Deal with the last alternative: go back and get number
7673 of the `jump_past_alt' just before it. `mcnt' contains
7674 the length of the alternative. */
7694 }
7701 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7702 It expects P to be the first byte of a single alternative and END one
7703 byte past the last. The alternative can contain groups. */
7705 static boolean
7709 {
7714 {
7715 /* Skip over opcodes that can match nothing, and break when we get
7716 to one that can't. */
7719 {
7720 /* It's a loop. */
7722 p1++;
7730 }
7737 /* Deals with the ops common to group_match_null_string_p and
7738 alt_match_null_string_p.
7740 Sets P to one after the op and its arguments, if any. */
7742 static boolean
7746 {
7748 boolean ret;
7753 {
7763 #ifdef emacs
7767 #endif
7775 /* Have to set this here in case we're checking a group which
7776 contains a group and a back reference to it. */
7785 /* If this is an optimized succeed_n for zero times, make the jump. */
7790 else
7795 /* Get to the number of times to succeed. */
7800 {
7804 }
7805 else
7818 /* All other opcodes mean we cannot match the empty string. */
7820 }
7827 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7828 bytes; nonzero otherwise. */
7830 static int
7834 RE_TRANSLATE_TYPE translate;
7835 {
7839 {
7840 #ifdef WCHAR
7847 len--;
7848 }
7850 }
7851 \f
7855 /* Entry points for GNU code. */
7857 /* re_compile_pattern is the GNU regular expression compiler: it
7858 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7859 Returns 0 if the pattern was valid, otherwise an error string.
7861 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7862 are set in BUFP on entry.
7864 We call regex_compile to do the actual compilation. */
7871 {
7872 reg_errcode_t ret;
7874 /* GNU code is written to assume at least RE_NREGS registers will be set
7875 (and at least one extra will be -1). */
7878 /* And GNU code determines whether or not to get register information
7879 by passing null for the REGS argument to re_match, etc., not by
7880 setting no_sub. */
7883 /* Match anchors at newline. */
7886 # ifdef MBS_SUPPORT
7889 else
7890 # endif
7896 }
7897 #ifdef _LIBC
7899 #endif
7900 \f
7901 /* Entry points compatible with 4.2 BSD regex library. We don't define
7902 them unless specifically requested. */
7904 #if defined _REGEX_RE_COMP || defined _LIBC
7906 /* BSD has one and only one pattern buffer. */
7910 #ifdef _LIBC
7911 /* Make these definitions weak in libc, so POSIX programs can redefine
7912 these names if they don't use our functions, and still use
7913 regcomp/regexec below without link errors. */
7914 weak_function
7915 #endif
7918 {
7919 reg_errcode_t ret;
7922 {
7926 }
7929 {
7938 }
7940 /* Since `re_exec' always passes NULL for the `regs' argument, we
7941 don't need to initialize the pattern buffer fields which affect it. */
7943 /* Match anchors at newlines. */
7946 # ifdef MBS_SUPPORT
7949 else
7950 # endif
7956 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7958 }
7961 int
7962 #ifdef _LIBC
7963 weak_function
7964 #endif
7967 {
7969 return
7971 }
7974 \f
7975 /* POSIX.2 functions. Don't define these for Emacs. */
7977 #ifndef emacs
7979 /* regcomp takes a regular expression as a string and compiles it.
7981 PREG is a regex_t *. We do not expect any fields to be initialized,
7982 since POSIX says we shouldn't. Thus, we set
7984 `buffer' to the compiled pattern;
7985 `used' to the length of the compiled pattern;
7986 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7987 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7988 RE_SYNTAX_POSIX_BASIC;
7989 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7990 `fastmap' to an allocated space for the fastmap;
7991 `fastmap_accurate' to zero;
7992 `re_nsub' to the number of subexpressions in PATTERN.
7994 PATTERN is the address of the pattern string.
7996 CFLAGS is a series of bits which affect compilation.
7998 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
7999 use POSIX basic syntax.
8001 If REG_NEWLINE is set, then . and [^...] don't match newline.
8002 Also, regexec will try a match beginning after every newline.
8004 If REG_ICASE is set, then we considers upper- and lowercase
8005 versions of letters to be equivalent when matching.
8007 If REG_NOSUB is set, then when PREG is passed to regexec, that
8008 routine will report only success or failure, and nothing about the
8009 registers.
8011 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
8012 the return codes and their meanings.) */
8014 int
8019 {
8020 reg_errcode_t ret;
8021 reg_syntax_t syntax
8025 /* regex_compile will allocate the space for the compiled pattern. */
8030 /* Try to allocate space for the fastmap. */
8034 {
8037 preg->translate
8043 /* Map uppercase characters to corresponding lowercase ones. */
8046 }
8047 else
8050 /* If REG_NEWLINE is set, newlines are treated differently. */
8055 /* It also changes the matching behavior. */
8057 }
8058 else
8063 /* POSIX says a null character in the pattern terminates it, so we
8064 can use strlen here in compiling the pattern. */
8065 # ifdef MBS_SUPPORT
8068 else
8069 # endif
8072 /* POSIX doesn't distinguish between an unmatched open-group and an
8073 unmatched close-group: both are REG_EPAREN. */
8077 {
8078 /* Compute the fastmap now, since regexec cannot modify the pattern
8079 buffer. */
8081 {
8082 /* Some error occurred while computing the fastmap, just forget
8083 about it. */
8086 }
8087 }
8090 }
8091 #ifdef _LIBC
8093 #endif
8096 /* regexec searches for a given pattern, specified by PREG, in the
8097 string STRING.
8099 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
8100 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
8101 least NMATCH elements, and we set them to the offsets of the
8102 corresponding matched substrings.
8104 EFLAGS specifies `execution flags' which affect matching: if
8105 REG_NOTBOL is set, then ^ does not match at the beginning of the
8106 string; if REG_NOTEOL is set, then $ does not match at the end.
8108 We return 0 if we find a match and REG_NOMATCH if not. */
8110 int
8115 regmatch_t pmatch[];
8117 {
8120 regex_t private_preg;
8129 /* The user has told us exactly how many registers to return
8130 information about, via `nmatch'. We have to pass that on to the
8131 matching routines. */
8135 {
8141 }
8143 /* Perform the searching operation. */
8148 /* Copy the register information to the POSIX structure. */
8150 {
8152 {
8156 {
8159 }
8160 }
8162 /* If we needed the temporary register info, free the space now. */
8164 }
8166 /* We want zero return to mean success, unlike `re_search'. */
8168 }
8169 #ifdef _LIBC
8171 #endif
8174 /* Returns a message corresponding to an error code, ERRCODE, returned
8175 from either regcomp or regexec. We don't use PREG here. */
8177 size_t
8183 {
8190 /* Only error codes returned by the rest of the code should be passed
8191 to this routine. If we are given anything else, or if other regex
8192 code generates an invalid error code, then the program has a bug.
8193 Dump core so we can fix it. */
8201 {
8203 {
8204 #if defined HAVE_MEMPCPY || defined _LIBC
8206 #else
8209 #endif
8210 }
8211 else
8213 }
8216 }
8217 #ifdef _LIBC
8219 #endif
8222 /* Free dynamically allocated space used by PREG. */
8224 void
8227 {
8243 }
8244 #ifdef _LIBC
8246 #endif
8252 \f
8253 #undef STORE_NUMBER
8254 #undef STORE_NUMBER_AND_INCR
8255 #undef EXTRACT_NUMBER
8256 #undef EXTRACT_NUMBER_AND_INCR
8258 #undef DEBUG_PRINT_COMPILED_PATTERN
8259 #undef DEBUG_PRINT_DOUBLE_STRING
8261 #undef INIT_FAIL_STACK
8262 #undef RESET_FAIL_STACK
8263 #undef DOUBLE_FAIL_STACK
8264 #undef PUSH_PATTERN_OP
8265 #undef PUSH_FAILURE_POINTER
8266 #undef PUSH_FAILURE_INT
8267 #undef PUSH_FAILURE_ELT
8268 #undef POP_FAILURE_POINTER
8269 #undef POP_FAILURE_INT
8270 #undef POP_FAILURE_ELT
8271 #undef DEBUG_PUSH
8272 #undef DEBUG_POP
8273 #undef PUSH_FAILURE_POINT
8274 #undef POP_FAILURE_POINT
8276 #undef REG_UNSET_VALUE
8277 #undef REG_UNSET
8279 #undef PATFETCH
8280 #undef PATFETCH_RAW
8281 #undef PATUNFETCH
8282 #undef TRANSLATE
8284 #undef INIT_BUF_SIZE
8285 #undef GET_BUFFER_SPACE
8286 #undef BUF_PUSH
8287 #undef BUF_PUSH_2
8288 #undef BUF_PUSH_3
8289 #undef STORE_JUMP
8290 #undef STORE_JUMP2
8291 #undef INSERT_JUMP
8292 #undef INSERT_JUMP2
8293 #undef EXTEND_BUFFER
8294 #undef GET_UNSIGNED_NUMBER
8295 #undef FREE_STACK_RETURN
8297 # undef POINTER_TO_OFFSET
8298 # undef MATCHING_IN_FRST_STRING
8299 # undef PREFETCH
8300 # undef AT_STRINGS_BEG
8301 # undef AT_STRINGS_END
8302 # undef WORDCHAR_P
8303 # undef FREE_VAR
8304 # undef FREE_VARIABLES
8305 # undef NO_HIGHEST_ACTIVE_REG
8306 # undef NO_LOWEST_ACTIVE_REG
8308 # undef CHAR_T
8309 # undef UCHAR_T
8310 # undef COMPILED_BUFFER_VAR
8311 # undef OFFSET_ADDRESS_SIZE
8312 # undef CHAR_CLASS_SIZE
8313 # undef PREFIX
8314 # undef ARG_PREFIX
8315 # undef PUT_CHAR
8316 # undef BYTE
8317 # undef WCHAR
8319 # define DEFINED_ONCE