| blob | e3439b2ff634d044cc1d1b1ef95f06035b8de910 |
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 #ifndef PARAMS
40 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
41 # define PARAMS(args) args
42 # else
43 # define PARAMS(args) ()
47 #ifndef INSIDE_RECURSION
49 # if defined STDC_HEADERS && !defined emacs
50 # include <stddef.h>
51 # else
52 /* We need this for `regex.h', and perhaps for the Emacs include files. */
53 # include <sys/types.h>
54 # endif
56 # define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
58 /* For platform which support the ISO C amendement 1 functionality we
59 support user defined character classes. */
60 # if defined _LIBC || WIDE_CHAR_SUPPORT
61 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
62 # include <wchar.h>
63 # include <wctype.h>
64 # endif
66 # ifdef _LIBC
67 /* We have to keep the namespace clean. */
68 # define regfree(preg) __regfree (preg)
69 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
70 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
71 # define regerror(errcode, preg, errbuf, errbuf_size) \
72 __regerror(errcode, preg, errbuf, errbuf_size)
73 # define re_set_registers(bu, re, nu, st, en) \
74 __re_set_registers (bu, re, nu, st, en)
75 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
76 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
77 # define re_match(bufp, string, size, pos, regs) \
78 __re_match (bufp, string, size, pos, regs)
79 # define re_search(bufp, string, size, startpos, range, regs) \
80 __re_search (bufp, string, size, startpos, range, regs)
81 # define re_compile_pattern(pattern, length, bufp) \
82 __re_compile_pattern (pattern, length, bufp)
83 # define re_set_syntax(syntax) __re_set_syntax (syntax)
84 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
85 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
86 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
88 # define btowc __btowc
90 /* We are also using some library internals. */
91 # include <locale/localeinfo.h>
92 # include <locale/elem-hash.h>
93 # include <langinfo.h>
94 # include <locale/coll-lookup.h>
95 # endif
97 /* This is for other GNU distributions with internationalized messages. */
98 # if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
99 # include <libintl.h>
100 # ifdef _LIBC
101 # undef gettext
103 # endif
104 # else
105 # define gettext(msgid) (msgid)
106 # endif
108 # ifndef gettext_noop
109 /* This define is so xgettext can find the internationalizable
110 strings. */
111 # define gettext_noop(String) String
112 # endif
114 /* The `emacs' switch turns on certain matching commands
115 that make sense only in Emacs. */
116 # ifdef emacs
124 /* If we are not linking with Emacs proper,
125 we can't use the relocating allocator
126 even if config.h says that we can. */
127 # undef REL_ALLOC
129 # if defined STDC_HEADERS || defined _LIBC
130 # include <stdlib.h>
131 # else
134 # endif
136 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
137 If nothing else has been done, use the method below. */
138 # ifdef INHIBIT_STRING_HEADER
139 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
140 # if !defined bzero && !defined bcopy
141 # undef INHIBIT_STRING_HEADER
142 # endif
143 # endif
144 # endif
146 /* This is the normal way of making sure we have a bcopy and a bzero.
147 This is used in most programs--a few other programs avoid this
148 by defining INHIBIT_STRING_HEADER. */
149 # ifndef INHIBIT_STRING_HEADER
150 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
151 # include <string.h>
152 # ifndef bzero
153 # ifndef _LIBC
155 # else
156 # define bzero(s, n) __bzero (s, n)
157 # endif
158 # endif
159 # else
160 # include <strings.h>
161 # ifndef memcmp
162 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
163 # endif
164 # ifndef memcpy
165 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
166 # endif
167 # endif
168 # endif
170 /* Define the syntax stuff for \<, \>, etc. */
172 /* This must be nonzero for the wordchar and notwordchar pattern
173 commands in re_match_2. */
174 # ifndef Sword
175 # define Sword 1
176 # endif
178 # ifdef SWITCH_ENUM_BUG
179 # define SWITCH_ENUM_CAST(x) ((int)(x))
180 # else
181 # define SWITCH_ENUM_CAST(x) (x)
182 # endif
186 # if defined _LIBC || HAVE_LIMITS_H
187 # include <limits.h>
188 # endif
190 # ifndef MB_LEN_MAX
191 # define MB_LEN_MAX 1
192 # endif
193 \f
194 /* Get the interface, including the syntax bits. */
197 /* isalpha etc. are used for the character classes. */
198 # include <ctype.h>
200 /* Jim Meyering writes:
202 "... Some ctype macros are valid only for character codes that
203 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
204 using /bin/cc or gcc but without giving an ansi option). So, all
205 ctype uses should be through macros like ISPRINT... If
206 STDC_HEADERS is defined, then autoconf has verified that the ctype
207 macros don't need to be guarded with references to isascii. ...
208 Defining isascii to 1 should let any compiler worth its salt
209 eliminate the && through constant folding."
210 Solaris defines some of these symbols so we must undefine them first. */
212 # undef ISASCII
213 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
214 # define ISASCII(c) 1
215 # else
216 # define ISASCII(c) isascii(c)
217 # endif
219 # ifdef isblank
220 # define ISBLANK(c) (ISASCII (c) && isblank (c))
221 # else
223 # endif
224 # ifdef isgraph
225 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
226 # else
227 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
228 # endif
230 # undef ISPRINT
231 # define ISPRINT(c) (ISASCII (c) && isprint (c))
232 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
233 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
234 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
235 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
236 # define ISLOWER(c) (ISASCII (c) && islower (c))
237 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
238 # define ISSPACE(c) (ISASCII (c) && isspace (c))
239 # define ISUPPER(c) (ISASCII (c) && isupper (c))
240 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
242 # ifdef _tolower
243 # define TOLOWER(c) _tolower(c)
244 # else
245 # define TOLOWER(c) tolower(c)
246 # endif
248 # ifndef NULL
249 # define NULL (void *)0
250 # endif
252 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
253 since ours (we hope) works properly with all combinations of
254 machines, compilers, `char' and `unsigned char' argument types.
255 (Per Bothner suggested the basic approach.) */
256 # undef SIGN_EXTEND_CHAR
257 # if __STDC__
258 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
260 /* As in Harbison and Steele. */
261 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
262 # endif
263 \f
264 # ifndef emacs
265 /* How many characters in the character set. */
266 # define CHAR_SET_SIZE 256
268 # ifdef SYNTAX_TABLE
278 static void
280 {
295 }
299 # define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
302 \f
303 /* Integer type for pointers. */
304 # if !defined _LIBC && !defined HAVE_UINTPTR_T
306 # endif
308 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
309 use `alloca' instead of `malloc'. This is because using malloc in
310 re_search* or re_match* could cause memory leaks when C-g is used in
311 Emacs; also, malloc is slower and causes storage fragmentation. On
312 the other hand, malloc is more portable, and easier to debug.
314 Because we sometimes use alloca, some routines have to be macros,
315 not functions -- `alloca'-allocated space disappears at the end of the
316 function it is called in. */
318 # ifdef REGEX_MALLOC
320 # define REGEX_ALLOCATE malloc
321 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
322 # define REGEX_FREE free
326 /* Emacs already defines alloca, sometimes. */
327 # ifndef alloca
329 /* Make alloca work the best possible way. */
330 # ifdef __GNUC__
331 # define alloca __builtin_alloca
333 # if HAVE_ALLOCA_H
334 # include <alloca.h>
340 # define REGEX_ALLOCATE alloca
342 /* Assumes a `char *destination' variable. */
343 # define REGEX_REALLOCATE(source, osize, nsize) \
344 (destination = (char *) alloca (nsize), \
345 memcpy (destination, source, osize))
347 /* No need to do anything to free, after alloca. */
352 /* Define how to allocate the failure stack. */
354 # if defined REL_ALLOC && defined REGEX_MALLOC
356 # define REGEX_ALLOCATE_STACK(size) \
357 r_alloc (&failure_stack_ptr, (size))
358 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
359 r_re_alloc (&failure_stack_ptr, (nsize))
360 # define REGEX_FREE_STACK(ptr) \
361 r_alloc_free (&failure_stack_ptr)
365 # ifdef REGEX_MALLOC
367 # define REGEX_ALLOCATE_STACK malloc
368 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
369 # define REGEX_FREE_STACK free
373 # define REGEX_ALLOCATE_STACK alloca
375 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
376 REGEX_REALLOCATE (source, osize, nsize)
377 /* No need to explicitly free anything. */
378 # define REGEX_FREE_STACK(arg)
384 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
385 `string1' or just past its end. This works if PTR is NULL, which is
386 a good thing. */
387 # define FIRST_STRING_P(ptr) \
388 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
390 /* (Re)Allocate N items of type T using malloc, or fail. */
391 # define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
392 # define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
393 # define RETALLOC_IF(addr, n, t) \
394 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
395 # define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
399 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
401 # undef MAX
402 # undef MIN
403 # define MAX(a, b) ((a) > (b) ? (a) : (b))
404 # define MIN(a, b) ((a) < (b) ? (a) : (b))
407 # define false 0
408 # define true 1
411 reg_syntax_t syntax,
427 #ifdef MBS_SUPPORT
429 reg_syntax_t syntax,
448 #endif
449 \f
450 /* These are the command codes that appear in compiled regular
451 expressions. Some opcodes are followed by argument bytes. A
452 command code can specify any interpretation whatsoever for its
453 arguments. Zero bytes may appear in the compiled regular expression. */
456 {
459 /* Succeed right away--no more backtracking. */
460 succeed,
462 /* Followed by one byte giving n, then by n literal bytes. */
463 exactn,
465 # ifdef MBS_SUPPORT
466 /* Same as exactn, but contains binary data. */
467 exactn_bin,
468 # endif
470 /* Matches any (more or less) character. */
471 anychar,
473 /* Matches any one char belonging to specified set. First
474 following byte is number of bitmap bytes. Then come bytes
475 for a bitmap saying which chars are in. Bits in each byte
476 are ordered low-bit-first. A character is in the set if its
477 bit is 1. A character too large to have a bit in the map is
478 automatically not in the set. */
479 /* ifdef MBS_SUPPORT, following element is length of character
480 classes, length of collating symbols, length of equivalence
481 classes, length of character ranges, and length of characters.
482 Next, character class element, collating symbols elements,
483 equivalence class elements, range elements, and character
484 elements follow.
485 See regex_compile function. */
486 charset,
488 /* Same parameters as charset, but match any character that is
489 not one of those specified. */
490 charset_not,
492 /* Start remembering the text that is matched, for storing in a
493 register. Followed by one byte with the register number, in
494 the range 0 to one less than the pattern buffer's re_nsub
495 field. Then followed by one byte with the number of groups
496 inner to this one. (This last has to be part of the
497 start_memory only because we need it in the on_failure_jump
498 of re_match_2.) */
499 start_memory,
501 /* Stop remembering the text that is matched and store it in a
502 memory register. Followed by one byte with the register
503 number, in the range 0 to one less than `re_nsub' in the
504 pattern buffer, and one byte with the number of inner groups,
505 just like `start_memory'. (We need the number of inner
506 groups here because we don't have any easy way of finding the
507 corresponding start_memory when we're at a stop_memory.) */
508 stop_memory,
510 /* Match a duplicate of something remembered. Followed by one
511 byte containing the register number. */
512 duplicate,
514 /* Fail unless at beginning of line. */
515 begline,
517 /* Fail unless at end of line. */
518 endline,
520 /* Succeeds if at beginning of buffer (if emacs) or at beginning
521 of string to be matched (if not). */
522 begbuf,
524 /* Analogously, for end of buffer/string. */
525 endbuf,
527 /* Followed by two byte relative address to which to jump. */
528 jump,
530 /* Same as jump, but marks the end of an alternative. */
531 jump_past_alt,
533 /* Followed by two-byte relative address of place to resume at
534 in case of failure. */
535 /* ifdef MBS_SUPPORT, the size of address is 1. */
536 on_failure_jump,
538 /* Like on_failure_jump, but pushes a placeholder instead of the
539 current string position when executed. */
540 on_failure_keep_string_jump,
542 /* Throw away latest failure point and then jump to following
543 two-byte relative address. */
544 /* ifdef MBS_SUPPORT, the size of address is 1. */
545 pop_failure_jump,
547 /* Change to pop_failure_jump if know won't have to backtrack to
548 match; otherwise change to jump. This is used to jump
549 back to the beginning of a repeat. If what follows this jump
550 clearly won't match what the repeat does, such that we can be
551 sure that there is no use backtracking out of repetitions
552 already matched, then we change it to a pop_failure_jump.
553 Followed by two-byte address. */
554 /* ifdef MBS_SUPPORT, the size of address is 1. */
555 maybe_pop_jump,
557 /* Jump to following two-byte address, and push a dummy failure
558 point. This failure point will be thrown away if an attempt
559 is made to use it for a failure. A `+' construct makes this
560 before the first repeat. Also used as an intermediary kind
561 of jump when compiling an alternative. */
562 /* ifdef MBS_SUPPORT, the size of address is 1. */
563 dummy_failure_jump,
565 /* Push a dummy failure point and continue. Used at the end of
566 alternatives. */
567 push_dummy_failure,
569 /* Followed by two-byte relative address and two-byte number n.
570 After matching N times, jump to the address upon failure. */
571 /* ifdef MBS_SUPPORT, the size of address is 1. */
572 succeed_n,
574 /* Followed by two-byte relative address, and two-byte number n.
575 Jump to the address N times, then fail. */
576 /* ifdef MBS_SUPPORT, the size of address is 1. */
577 jump_n,
579 /* Set the following two-byte relative address to the
580 subsequent two-byte number. The address *includes* the two
581 bytes of number. */
582 /* ifdef MBS_SUPPORT, the size of address is 1. */
583 set_number_at,
592 notwordbound /* Succeeds if not at a word boundary. */
594 # ifdef emacs
599 /* Matches any character whose syntax is specified. Followed by
600 a byte which contains a syntax code, e.g., Sword. */
601 syntaxspec,
603 /* Matches any character whose syntax is not that specified. */
604 notsyntaxspec
608 \f
610 #ifdef BYTE
611 # define CHAR_T char
612 # define UCHAR_T unsigned char
613 # define COMPILED_BUFFER_VAR bufp->buffer
614 # define OFFSET_ADDRESS_SIZE 2
615 # if defined (__STDC__) || defined (ALMOST_STDC) || defined (HAVE_STRINGIZE)
616 # define PREFIX(name) byte_##name
617 # else
619 # endif
620 # define ARG_PREFIX(name) name
621 # define PUT_CHAR(c) putchar (c)
622 #else
623 # ifdef WCHAR
624 # define CHAR_T wchar_t
625 # define UCHAR_T wchar_t
626 # define COMPILED_BUFFER_VAR wc_buffer
628 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
629 # if defined (__STDC__) || defined (ALMOST_STDC) || defined (HAVE_STRINGIZE)
630 # define PREFIX(name) wcs_##name
631 # define ARG_PREFIX(name) c##name
632 # else
635 # endif
636 /* Should we use wide stream?? */
638 # define TRUE 1
639 # define FALSE 0
640 # else
641 # ifdef MBS_SUPPORT
642 # define WCHAR
643 # define INSIDE_RECURSION
645 # undef INSIDE_RECURSION
646 # endif
647 # define BYTE
648 # define INSIDE_RECURSION
650 # undef INSIDE_RECURSION
651 # endif
652 #endif
654 #ifdef INSIDE_RECURSION
655 /* Common operations on the compiled pattern. */
657 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
658 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
660 # ifdef WCHAR
661 # define STORE_NUMBER(destination, number) \
662 do { \
663 *(destination) = (UCHAR_T)(number); \
664 } while (0)
666 # define STORE_NUMBER(destination, number) \
667 do { \
668 (destination)[0] = (number) & 0377; \
669 (destination)[1] = (number) >> 8; \
670 } while (0)
673 /* Same as STORE_NUMBER, except increment DESTINATION to
674 the byte after where the number is stored. Therefore, DESTINATION
675 must be an lvalue. */
676 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
678 # define STORE_NUMBER_AND_INCR(destination, number) \
679 do { \
680 STORE_NUMBER (destination, number); \
681 (destination) += OFFSET_ADDRESS_SIZE; \
682 } while (0)
684 /* Put into DESTINATION a number stored in two contiguous bytes starting
685 at SOURCE. */
686 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
688 # ifdef WCHAR
689 # define EXTRACT_NUMBER(destination, source) \
690 do { \
691 (destination) = *(source); \
692 } while (0)
694 # define EXTRACT_NUMBER(destination, source) \
695 do { \
696 (destination) = *(source) & 0377; \
697 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
698 } while (0)
699 # endif
701 # ifdef DEBUG
703 static void
707 {
708 # ifdef WCHAR
714 # endif
715 }
718 # undef EXTRACT_NUMBER
719 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
724 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
725 SOURCE must be an lvalue. */
727 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
728 do { \
729 EXTRACT_NUMBER (destination, source); \
730 (source) += OFFSET_ADDRESS_SIZE; \
731 } while (0)
733 # ifdef DEBUG
736 static void
740 {
743 }
745 # ifndef EXTRACT_MACROS
746 # undef EXTRACT_NUMBER_AND_INCR
747 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
748 PREFIX(extract_number_and_incr) (&dest, &src)
753 \f
755 /* If DEBUG is defined, Regex prints many voluminous messages about what
756 it is doing (if the variable `debug' is nonzero). If linked with the
757 main program in `iregex.c', you can enter patterns and strings
758 interactively. And if linked with the main program in `main.c' and
759 the other test files, you can run the already-written tests. */
761 # ifdef DEBUG
763 # ifndef DEFINED_ONCE
765 /* We use standard I/O for debugging. */
766 # include <stdio.h>
768 /* It is useful to test things that ``must'' be true when debugging. */
769 # include <assert.h>
773 # define DEBUG_STATEMENT(e) e
774 # define DEBUG_PRINT1(x) if (debug) printf (x)
775 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
776 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
777 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
780 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
781 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
782 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
783 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
786 /* Print the fastmap in human-readable form. */
788 # ifndef DEFINED_ONCE
789 void
792 {
797 {
799 {
803 {
805 i++;
806 }
808 {
811 }
812 }
813 }
815 }
819 /* Print a compiled pattern string in human-readable form, starting at
820 the START pointer into it and ending just before the pointer END. */
822 void
826 {
833 {
836 }
838 /* Loop over pattern commands. */
840 {
841 # ifdef _LIBC
843 # else
845 # endif
848 {
856 do
857 {
860 }
864 # ifdef MBS_SUPPORT
868 do
869 {
871 }
896 {
897 # ifdef WCHAR
908 {
914 }
917 {
923 }
926 {
930 }
935 # else
947 {
948 /* Are we starting a range? */
950 {
953 }
954 /* Have we broken a range? */
956 {
959 }
965 }
974 }
987 # ifdef _LIBC
989 # else
991 # endif
996 # ifdef _LIBC
998 # else
1001 # endif
1006 # ifdef _LIBC
1008 # else
1010 # endif
1019 # ifdef _LIBC
1021 # else
1023 # endif
1028 # ifdef _LIBC
1030 # else
1032 # endif
1037 # ifdef _LIBC
1039 # else
1041 # endif
1046 # ifdef _LIBC
1048 # else
1050 # endif
1057 # ifdef _LIBC
1059 # else
1062 # endif
1076 # ifdef _LIBC
1078 # else
1081 # endif
1100 # ifdef emacs
1144 }
1147 }
1149 # ifdef _LIBC
1151 # else
1153 # endif
1154 }
1157 void
1160 {
1169 {
1172 }
1174 # ifdef _LIBC
1176 # else
1178 # endif
1186 /* Perhaps we should print the translate table? */
1187 }
1190 void
1197 {
1202 else
1203 {
1207 {
1212 }
1216 {
1219 {
1222 }
1223 }
1224 }
1225 }
1227 # ifndef DEFINED_ONCE
1228 void
1231 {
1233 }
1234 # endif
1238 # ifndef DEFINED_ONCE
1239 # undef assert
1240 # define assert(e)
1242 # define DEBUG_STATEMENT(e)
1243 # define DEBUG_PRINT1(x)
1244 # define DEBUG_PRINT2(x1, x2)
1245 # define DEBUG_PRINT3(x1, x2, x3)
1246 # define DEBUG_PRINT4(x1, x2, x3, x4)
1248 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1249 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1253 \f
1255 # ifdef WCHAR
1256 /* This convert a multibyte string to a wide character string.
1257 And write their correspondances to offset_buffer(see below)
1258 and write whether each wchar_t is binary data to is_binary.
1259 This assume invalid multibyte sequences as binary data.
1260 We assume offset_buffer and is_binary is already allocated
1261 enough space. */
1266 static size_t
1272 /* It hold correspondances between src(char string) and
1273 dest(wchar_t string) for optimization.
1274 e.g. src = "xxxyzz"
1275 dest = {'X', 'Y', 'Z'}
1276 (each "xxx", "y" and "zz" represent one multibyte character
1277 corresponding to 'X', 'Y' and 'Z'.)
1278 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1279 = {0, 3, 4, 6}
1280 */
1283 {
1293 /* Initialize the conversion state. */
1299 {
1300 #ifdef _LIBC
1302 #else
1304 #endif
1307 /* failed to convert. maybe src contains binary data.
1308 So we consume 1 byte manualy. */
1309 {
1313 }
1314 else
1316 /* In sjis encoding, we use yen sign as escape character in
1317 place of reverse solidus. So we convert 0x5c(yen sign in
1318 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1319 solidus in UCS2). */
1324 }
1326 /* Fill remain of the buffer with sentinel. */
1331 }
1337 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1338 also be assigned to arbitrarily: each pattern buffer stores its own
1339 syntax, so it can be changed between regex compilations. */
1340 /* This has no initializer because initialized variables in Emacs
1341 become read-only after dumping. */
1342 reg_syntax_t re_syntax_options;
1345 /* Specify the precise syntax of regexps for compilation. This provides
1346 for compatibility for various utilities which historically have
1347 different, incompatible syntaxes.
1349 The argument SYNTAX is a bit mask comprised of the various bits
1350 defined in regex.h. We return the old syntax. */
1352 reg_syntax_t
1354 reg_syntax_t syntax;
1355 {
1359 # ifdef DEBUG
1366 }
1367 # ifdef _LIBC
1369 # endif
1370 \f
1371 /* This table gives an error message for each of the error codes listed
1372 in regex.h. Obviously the order here has to be same as there.
1373 POSIX doesn't require that we do anything for REG_NOERROR,
1374 but why not be nice? */
1377 {
1395 };
1396 \f
1399 #ifndef DEFINED_ONCE
1400 /* Avoiding alloca during matching, to placate r_alloc. */
1402 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1403 searching and matching functions should not call alloca. On some
1404 systems, alloca is implemented in terms of malloc, and if we're
1405 using the relocating allocator routines, then malloc could cause a
1406 relocation, which might (if the strings being searched are in the
1407 ralloc heap) shift the data out from underneath the regexp
1408 routines.
1410 Here's another reason to avoid allocation: Emacs
1411 processes input from X in a signal handler; processing X input may
1412 call malloc; if input arrives while a matching routine is calling
1413 malloc, then we're scrod. But Emacs can't just block input while
1414 calling matching routines; then we don't notice interrupts when
1415 they come in. So, Emacs blocks input around all regexp calls
1416 except the matching calls, which it leaves unprotected, in the
1417 faith that they will not malloc. */
1419 /* Normally, this is fine. */
1420 # define MATCH_MAY_ALLOCATE
1422 /* When using GNU C, we are not REALLY using the C alloca, no matter
1423 what config.h may say. So don't take precautions for it. */
1424 # ifdef __GNUC__
1425 # undef C_ALLOCA
1426 # endif
1428 /* The match routines may not allocate if (1) they would do it with malloc
1429 and (2) it's not safe for them to use malloc.
1430 Note that if REL_ALLOC is defined, matching would not use malloc for the
1431 failure stack, but we would still use it for the register vectors;
1432 so REL_ALLOC should not affect this. */
1433 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1434 # undef MATCH_MAY_ALLOCATE
1435 # endif
1437 \f
1438 #ifdef INSIDE_RECURSION
1439 /* Failure stack declarations and macros; both re_compile_fastmap and
1440 re_match_2 use a failure stack. These have to be macros because of
1441 REGEX_ALLOCATE_STACK. */
1444 /* Number of failure points for which to initially allocate space
1445 when matching. If this number is exceeded, we allocate more
1446 space, so it is not a hard limit. */
1447 # ifndef INIT_FAILURE_ALLOC
1448 # define INIT_FAILURE_ALLOC 5
1449 # endif
1451 /* Roughly the maximum number of failure points on the stack. Would be
1452 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1453 This is a variable only so users of regex can assign to it; we never
1454 change it ourselves. */
1456 # ifdef INT_IS_16BIT
1458 # ifndef DEFINED_ONCE
1459 # if defined MATCH_MAY_ALLOCATE
1460 /* 4400 was enough to cause a crash on Alpha OSF/1,
1461 whose default stack limit is 2mb. */
1463 # else
1465 # endif
1466 # endif
1469 {
1472 };
1477 {
1485 # ifndef DEFINED_ONCE
1486 # if defined MATCH_MAY_ALLOCATE
1487 /* 4400 was enough to cause a crash on Alpha OSF/1,
1488 whose default stack limit is 2mb. */
1490 # else
1492 # endif
1493 # endif
1496 {
1499 };
1504 {
1512 # ifndef DEFINED_ONCE
1513 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1514 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1515 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1516 # endif
1519 /* Define macros to initialize and free the failure stack.
1520 Do `return -2' if the alloc fails. */
1522 # ifdef MATCH_MAY_ALLOCATE
1523 # define INIT_FAIL_STACK() \
1524 do { \
1525 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1526 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1527 \
1528 if (fail_stack.stack == NULL) \
1529 return -2; \
1530 \
1531 fail_stack.size = INIT_FAILURE_ALLOC; \
1532 fail_stack.avail = 0; \
1533 } while (0)
1535 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1536 # else
1537 # define INIT_FAIL_STACK() \
1538 do { \
1539 fail_stack.avail = 0; \
1540 } while (0)
1542 # define RESET_FAIL_STACK()
1543 # endif
1546 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1548 Return 1 if succeeds, and 0 if either ran out of memory
1549 allocating space for it or it was already too large.
1551 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1553 # define DOUBLE_FAIL_STACK(fail_stack) \
1554 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1555 ? 0 \
1556 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1557 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1558 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1559 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1560 \
1561 (fail_stack).stack == NULL \
1562 ? 0 \
1563 : ((fail_stack).size <<= 1, \
1564 1)))
1567 /* Push pointer POINTER on FAIL_STACK.
1568 Return 1 if was able to do so and 0 if ran out of memory allocating
1569 space to do so. */
1570 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1571 ((FAIL_STACK_FULL () \
1572 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1573 ? 0 \
1574 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1575 1))
1577 /* Push a pointer value onto the failure stack.
1578 Assumes the variable `fail_stack'. Probably should only
1579 be called from within `PUSH_FAILURE_POINT'. */
1580 # define PUSH_FAILURE_POINTER(item) \
1581 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1583 /* This pushes an integer-valued item onto the failure stack.
1584 Assumes the variable `fail_stack'. Probably should only
1585 be called from within `PUSH_FAILURE_POINT'. */
1586 # define PUSH_FAILURE_INT(item) \
1587 fail_stack.stack[fail_stack.avail++].integer = (item)
1589 /* Push a fail_stack_elt_t value onto the failure stack.
1590 Assumes the variable `fail_stack'. Probably should only
1591 be called from within `PUSH_FAILURE_POINT'. */
1592 # define PUSH_FAILURE_ELT(item) \
1593 fail_stack.stack[fail_stack.avail++] = (item)
1595 /* These three POP... operations complement the three PUSH... operations.
1596 All assume that `fail_stack' is nonempty. */
1597 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1598 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1599 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1601 /* Used to omit pushing failure point id's when we're not debugging. */
1602 # ifdef DEBUG
1603 # define DEBUG_PUSH PUSH_FAILURE_INT
1604 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1605 # else
1606 # define DEBUG_PUSH(item)
1607 # define DEBUG_POP(item_addr)
1608 # endif
1611 /* Push the information about the state we will need
1612 if we ever fail back to it.
1614 Requires variables fail_stack, regstart, regend, reg_info, and
1615 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1616 be declared.
1618 Does `return FAILURE_CODE' if runs out of memory. */
1620 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1621 do { \
1622 char *destination; \
1623 /* Must be int, so when we don't save any registers, the arithmetic \
1625 /* Can't be int, since there is not a shred of a guarantee that int \
1626 is wide enough to hold a value of something to which pointer can \
1628 active_reg_t this_reg; \
1629 \
1630 DEBUG_STATEMENT (failure_id++); \
1631 DEBUG_STATEMENT (nfailure_points_pushed++); \
1635 \
1638 \
1640 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1641 { \
1642 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1643 return failure_code; \
1644 \
1646 (fail_stack).size); \
1648 } \
1649 \
1652 \
1653 if (1) \
1654 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1655 this_reg++) \
1656 { \
1658 DEBUG_STATEMENT (num_regs_pushed++); \
1659 \
1661 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1662 \
1664 PUSH_FAILURE_POINTER (regend[this_reg]); \
1665 \
1667 reg_info[this_reg].word.pointer); \
1669 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1672 MATCHED_SOMETHING (reg_info[this_reg])); \
1674 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1676 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1677 } \
1678 \
1680 PUSH_FAILURE_INT (lowest_active_reg); \
1681 \
1683 PUSH_FAILURE_INT (highest_active_reg); \
1684 \
1686 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1687 PUSH_FAILURE_POINTER (pattern_place); \
1688 \
1690 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1691 size2); \
1693 PUSH_FAILURE_POINTER (string_place); \
1694 \
1696 DEBUG_PUSH (failure_id); \
1697 } while (0)
1699 # ifndef DEFINED_ONCE
1700 /* This is the number of items that are pushed and popped on the stack
1701 for each register. */
1702 # define NUM_REG_ITEMS 3
1704 /* Individual items aside from the registers. */
1705 # ifdef DEBUG
1707 # else
1708 # define NUM_NONREG_ITEMS 4
1709 # endif
1711 /* We push at most this many items on the stack. */
1712 /* We used to use (num_regs - 1), which is the number of registers
1713 this regexp will save; but that was changed to 5
1714 to avoid stack overflow for a regexp with lots of parens. */
1715 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1717 /* We actually push this many items. */
1718 # define NUM_FAILURE_ITEMS \
1719 (((0 \
1720 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1721 * NUM_REG_ITEMS) \
1722 + NUM_NONREG_ITEMS)
1724 /* How many items can still be added to the stack without overflowing it. */
1725 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1729 /* Pops what PUSH_FAIL_STACK pushes.
1731 We restore into the parameters, all of which should be lvalues:
1732 STR -- the saved data position.
1733 PAT -- the saved pattern position.
1734 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1735 REGSTART, REGEND -- arrays of string positions.
1736 REG_INFO -- array of information about each subexpression.
1738 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1739 `pend', `string1', `size1', `string2', and `size2'. */
1740 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1741 { \
1742 DEBUG_STATEMENT (unsigned failure_id;) \
1743 active_reg_t this_reg; \
1744 const UCHAR_T *string_temp; \
1745 \
1746 assert (!FAIL_STACK_EMPTY ()); \
1747 \
1752 \
1753 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1754 \
1755 DEBUG_POP (&failure_id); \
1757 \
1758 /* If the saved string location is NULL, it came from an \
1759 on_failure_keep_string_jump opcode, and we want to throw away the \
1761 string_temp = POP_FAILURE_POINTER (); \
1762 if (string_temp != NULL) \
1763 str = (const CHAR_T *) string_temp; \
1764 \
1766 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1768 \
1769 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1771 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1772 \
1774 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1776 \
1777 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1779 \
1780 if (1) \
1781 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1782 { \
1784 \
1785 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1787 reg_info[this_reg].word.pointer); \
1788 \
1789 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1791 \
1792 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1794 } \
1795 else \
1796 { \
1797 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1798 { \
1799 reg_info[this_reg].word.integer = 0; \
1800 regend[this_reg] = 0; \
1801 regstart[this_reg] = 0; \
1802 } \
1803 highest_active_reg = high_reg; \
1804 } \
1805 \
1806 set_regs_matched_done = 0; \
1807 DEBUG_STATEMENT (nfailure_points_popped++); \
1809 \f
1810 /* Structure for per-register (a.k.a. per-group) information.
1811 Other register information, such as the
1812 starting and ending positions (which are addresses), and the list of
1813 inner groups (which is a bits list) are maintained in separate
1814 variables.
1816 We are making a (strictly speaking) nonportable assumption here: that
1817 the compiler will pack our bit fields into something that fits into
1818 the type of `word', i.e., is something that fits into one item on the
1819 failure stack. */
1822 /* Declarations and macros for re_match_2. */
1825 {
1827 struct
1828 {
1829 /* This field is one if this group can match the empty string,
1830 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1831 # define MATCH_NULL_UNSET_VALUE 3
1839 # ifndef DEFINED_ONCE
1840 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1841 # define IS_ACTIVE(R) ((R).bits.is_active)
1842 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1843 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1846 /* Call this when have matched a real character; it sets `matched' flags
1847 for the subexpressions which we are currently inside. Also records
1848 that those subexprs have matched. */
1849 # define SET_REGS_MATCHED() \
1850 do \
1851 { \
1852 if (!set_regs_matched_done) \
1853 { \
1854 active_reg_t r; \
1855 set_regs_matched_done = 1; \
1856 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1857 { \
1858 MATCHED_SOMETHING (reg_info[r]) \
1859 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1860 = 1; \
1861 } \
1862 } \
1863 } \
1864 while (0)
1867 /* Registers are set to a sentinel when they haven't yet matched. */
1869 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1870 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1872 /* Subroutine declarations and macros for regex_compile. */
1882 reg_syntax_t syntax));
1885 reg_syntax_t syntax));
1886 # ifdef WCHAR
1891 reg_syntax_t syntax,
1900 reg_syntax_t syntax,
1904 /* Fetch the next character in the uncompiled pattern---translating it
1905 if necessary. Also cast from a signed character in the constant
1906 string passed to us by the user to an unsigned char that we can use
1907 as an array index (in, e.g., `translate'). */
1908 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1909 because it is impossible to allocate 4GB array for some encodings
1910 which have 4 byte character_set like UCS4. */
1911 # ifndef PATFETCH
1912 # ifdef WCHAR
1913 # define PATFETCH(c) \
1914 do {if (p == pend) return REG_EEND; \
1915 c = (UCHAR_T) *p++; \
1916 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1917 } while (0)
1919 # define PATFETCH(c) \
1920 do {if (p == pend) return REG_EEND; \
1921 c = (unsigned char) *p++; \
1922 if (translate) c = (unsigned char) translate[c]; \
1923 } while (0)
1925 # endif
1927 /* Fetch the next character in the uncompiled pattern, with no
1928 translation. */
1929 # define PATFETCH_RAW(c) \
1930 do {if (p == pend) return REG_EEND; \
1931 c = (UCHAR_T) *p++; \
1932 } while (0)
1934 /* Go backwards one character in the pattern. */
1935 # define PATUNFETCH p--
1938 /* If `translate' is non-null, return translate[D], else just D. We
1939 cast the subscript to translate because some data is declared as
1940 `char *', to avoid warnings when a string constant is passed. But
1941 when we use a character as a subscript we must make it unsigned. */
1942 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1943 because it is impossible to allocate 4GB array for some encodings
1944 which have 4 byte character_set like UCS4. */
1946 # ifndef TRANSLATE
1947 # ifdef WCHAR
1948 # define TRANSLATE(d) \
1949 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1950 ? (char) translate[(unsigned char) (d)] : (d))
1952 # define TRANSLATE(d) \
1953 (translate ? (char) translate[(unsigned char) (d)] : (d))
1955 # endif
1958 /* Macros for outputting the compiled pattern into `buffer'. */
1960 /* If the buffer isn't allocated when it comes in, use this. */
1961 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1963 /* Make sure we have at least N more bytes of space in buffer. */
1964 # ifdef WCHAR
1965 # define GET_BUFFER_SPACE(n) \
1966 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
1967 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
1968 EXTEND_BUFFER ()
1970 # define GET_BUFFER_SPACE(n) \
1971 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1972 EXTEND_BUFFER ()
1975 /* Make sure we have one more byte of buffer space and then add C to it. */
1976 # define BUF_PUSH(c) \
1977 do { \
1978 GET_BUFFER_SPACE (1); \
1979 *b++ = (UCHAR_T) (c); \
1980 } while (0)
1983 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1984 # define BUF_PUSH_2(c1, c2) \
1985 do { \
1986 GET_BUFFER_SPACE (2); \
1987 *b++ = (UCHAR_T) (c1); \
1988 *b++ = (UCHAR_T) (c2); \
1989 } while (0)
1992 /* As with BUF_PUSH_2, except for three bytes. */
1993 # define BUF_PUSH_3(c1, c2, c3) \
1994 do { \
1995 GET_BUFFER_SPACE (3); \
1996 *b++ = (UCHAR_T) (c1); \
1997 *b++ = (UCHAR_T) (c2); \
1998 *b++ = (UCHAR_T) (c3); \
1999 } while (0)
2001 /* Store a jump with opcode OP at LOC to location TO. We store a
2002 relative address offset by the three bytes the jump itself occupies. */
2003 # define STORE_JUMP(op, loc, to) \
2004 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
2006 /* Likewise, for a two-argument jump. */
2007 # define STORE_JUMP2(op, loc, to, arg) \
2008 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
2010 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
2011 # define INSERT_JUMP(op, loc, to) \
2012 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
2014 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
2015 # define INSERT_JUMP2(op, loc, to, arg) \
2016 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
2017 arg, b)
2019 /* This is not an arbitrary limit: the arguments which represent offsets
2020 into the pattern are two bytes long. So if 2^16 bytes turns out to
2021 be too small, many things would have to change. */
2022 /* Any other compiler which, like MSC, has allocation limit below 2^16
2023 bytes will have to use approach similar to what was done below for
2024 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
2025 reallocating to 0 bytes. Such thing is not going to work too well.
2026 You have been warned!! */
2027 # ifndef DEFINED_ONCE
2028 # if defined _MSC_VER && !defined WIN32
2029 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2030 The REALLOC define eliminates a flurry of conversion warnings,
2031 but is not required. */
2032 # define MAX_BUF_SIZE 65500L
2033 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2034 # else
2035 # define MAX_BUF_SIZE (1L << 16)
2036 # define REALLOC(p,s) realloc ((p), (s))
2037 # endif
2039 /* Extend the buffer by twice its current size via realloc and
2040 reset the pointers that pointed into the old block to point to the
2041 correct places in the new one. If extending the buffer results in it
2042 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2043 # if __BOUNDED_POINTERS__
2044 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2045 # define MOVE_BUFFER_POINTER(P) \
2046 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2047 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2048 else \
2049 { \
2050 SET_HIGH_BOUND (b); \
2051 SET_HIGH_BOUND (begalt); \
2052 if (fixup_alt_jump) \
2053 SET_HIGH_BOUND (fixup_alt_jump); \
2054 if (laststart) \
2055 SET_HIGH_BOUND (laststart); \
2056 if (pending_exact) \
2057 SET_HIGH_BOUND (pending_exact); \
2058 }
2059 # else
2060 # define MOVE_BUFFER_POINTER(P) (P) += incr
2061 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2062 # endif
2065 # ifdef WCHAR
2066 # define EXTEND_BUFFER() \
2067 do { \
2068 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2069 int wchar_count; \
2070 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2071 return REG_ESIZE; \
2072 bufp->allocated <<= 1; \
2073 if (bufp->allocated > MAX_BUF_SIZE) \
2074 bufp->allocated = MAX_BUF_SIZE; \
2076 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2077 if (wchar_count == 0) wchar_count = 1; \
2079 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2080 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2081 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2082 if (COMPILED_BUFFER_VAR == NULL) \
2083 return REG_ESPACE; \
2085 if (old_buffer != COMPILED_BUFFER_VAR) \
2086 { \
2087 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2088 MOVE_BUFFER_POINTER (b); \
2089 MOVE_BUFFER_POINTER (begalt); \
2090 if (fixup_alt_jump) \
2091 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2092 if (laststart) \
2093 MOVE_BUFFER_POINTER (laststart); \
2094 if (pending_exact) \
2095 MOVE_BUFFER_POINTER (pending_exact); \
2096 } \
2097 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2098 } while (0)
2100 # define EXTEND_BUFFER() \
2101 do { \
2102 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2103 if (bufp->allocated == MAX_BUF_SIZE) \
2104 return REG_ESIZE; \
2105 bufp->allocated <<= 1; \
2106 if (bufp->allocated > MAX_BUF_SIZE) \
2107 bufp->allocated = MAX_BUF_SIZE; \
2108 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2109 bufp->allocated); \
2110 if (COMPILED_BUFFER_VAR == NULL) \
2111 return REG_ESPACE; \
2113 if (old_buffer != COMPILED_BUFFER_VAR) \
2114 { \
2115 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2116 MOVE_BUFFER_POINTER (b); \
2117 MOVE_BUFFER_POINTER (begalt); \
2118 if (fixup_alt_jump) \
2119 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2120 if (laststart) \
2121 MOVE_BUFFER_POINTER (laststart); \
2122 if (pending_exact) \
2123 MOVE_BUFFER_POINTER (pending_exact); \
2124 } \
2125 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2126 } while (0)
2129 # ifndef DEFINED_ONCE
2130 /* Since we have one byte reserved for the register number argument to
2131 {start,stop}_memory, the maximum number of groups we can report
2132 things about is what fits in that byte. */
2133 # define MAX_REGNUM 255
2135 /* But patterns can have more than `MAX_REGNUM' registers. We just
2136 ignore the excess. */
2140 /* Macros for the compile stack. */
2142 /* Since offsets can go either forwards or backwards, this type needs to
2143 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2144 /* int may be not enough when sizeof(int) == 2. */
2148 {
2149 pattern_offset_t begalt_offset;
2150 pattern_offset_t fixup_alt_jump;
2151 pattern_offset_t inner_group_offset;
2152 pattern_offset_t laststart_offset;
2153 regnum_t regnum;
2158 {
2165 # define INIT_COMPILE_STACK_SIZE 32
2167 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2168 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2170 /* The next available element. */
2171 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2175 /* Set the bit for character C in a list. */
2176 # ifndef DEFINED_ONCE
2177 # define SET_LIST_BIT(c) \
2178 (b[((unsigned char) (c)) / BYTEWIDTH] \
2179 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2182 /* Get the next unsigned number in the uncompiled pattern. */
2183 # define GET_UNSIGNED_NUMBER(num) \
2184 { \
2185 while (p != pend) \
2186 { \
2187 PATFETCH (c); \
2189 break; \
2190 if (num <= RE_DUP_MAX) \
2191 { \
2192 if (num < 0) \
2193 num = 0; \
2195 } \
2196 } \
2197 }
2199 # ifndef DEFINED_ONCE
2200 # if defined _LIBC || WIDE_CHAR_SUPPORT
2201 /* The GNU C library provides support for user-defined character classes
2202 and the functions from ISO C amendement 1. */
2203 # ifdef CHARCLASS_NAME_MAX
2204 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2205 # else
2206 /* This shouldn't happen but some implementation might still have this
2207 problem. Use a reasonable default value. */
2208 # define CHAR_CLASS_MAX_LENGTH 256
2209 # endif
2211 # ifdef _LIBC
2212 # define IS_CHAR_CLASS(string) __wctype (string)
2213 # else
2214 # define IS_CHAR_CLASS(string) wctype (string)
2215 # endif
2216 # else
2219 # define IS_CHAR_CLASS(string) \
2226 # endif
2228 \f
2229 # ifndef MATCH_MAY_ALLOCATE
2231 /* If we cannot allocate large objects within re_match_2_internal,
2232 we make the fail stack and register vectors global.
2233 The fail stack, we grow to the maximum size when a regexp
2234 is compiled.
2235 The register vectors, we adjust in size each time we
2236 compile a regexp, according to the number of registers it needs. */
2240 /* Size with which the following vectors are currently allocated.
2241 That is so we can make them bigger as needed,
2242 but never make them smaller. */
2243 # ifdef DEFINED_ONCE
2255 /* Make the register vectors big enough for NUM_REGS registers,
2256 but don't make them smaller. */
2258 static void
2261 {
2263 {
2275 }
2276 }
2279 \f
2280 # ifndef DEFINED_ONCE
2282 compile_stack,
2283 regnum_t regnum));
2286 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2287 Returns one of error codes defined in `regex.h', or zero for success.
2289 Assumes the `allocated' (and perhaps `buffer') and `translate'
2290 fields are set in BUFP on entry.
2292 If it succeeds, results are put in BUFP (if it returns an error, the
2293 contents of BUFP are undefined):
2294 `buffer' is the compiled pattern;
2295 `syntax' is set to SYNTAX;
2296 `used' is set to the length of the compiled pattern;
2297 `fastmap_accurate' is zero;
2298 `re_nsub' is the number of subexpressions in PATTERN;
2299 `not_bol' and `not_eol' are zero;
2301 The `fastmap' and `newline_anchor' fields are neither
2302 examined nor set. */
2304 /* Return, freeing storage we allocated. */
2305 # ifdef WCHAR
2306 # define FREE_STACK_RETURN(value) \
2307 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2308 # else
2309 # define FREE_STACK_RETURN(value) \
2310 return (free (compile_stack.stack), value)
2313 static reg_errcode_t
2317 reg_syntax_t syntax;
2319 {
2320 /* We fetch characters from PATTERN here. Even though PATTERN is
2321 `char *' (i.e., signed), we declare these variables as unsigned, so
2322 they can be reliably used as array indices. */
2325 #ifdef WCHAR
2326 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2329 /* offset buffer for optimization. See convert_mbs_to_wc. */
2331 /* It hold whether each wchar_t is binary data or not. */
2333 /* A flag whether exactn is handling binary data or not. */
2337 /* A random temporary spot in PATTERN. */
2340 /* Points to the end of the buffer, where we should append. */
2343 /* Keeps track of unclosed groups. */
2344 compile_stack_type compile_stack;
2346 /* Points to the current (ending) position in the pattern. */
2347 #ifdef WCHAR
2355 /* How to translate the characters in the pattern. */
2358 /* Address of the count-byte of the most recently inserted `exactn'
2359 command. This makes it possible to tell if a new exact-match
2360 character can be added to that command or if the character requires
2361 a new `exactn' command. */
2364 /* Address of start of the most recently finished expression.
2365 This tells, e.g., postfix * where to find the start of its
2366 operand. Reset at the beginning of groups and alternatives. */
2369 /* Address of beginning of regexp, or inside of last group. */
2372 /* Address of the place where a forward jump should go to the end of
2373 the containing expression. Each alternative of an `or' -- except the
2374 last -- ends with a forward jump of this sort. */
2377 /* Counts open-groups as they are encountered. Remembered for the
2378 matching close-group on the compile stack, so the same register
2379 number is put in the stop_memory as the start_memory. */
2382 #ifdef WCHAR
2383 /* Initialize the wchar_t PATTERN and offset_buffer. */
2388 {
2393 }
2398 {
2403 }
2404 #endif
2406 #ifdef DEBUG
2409 {
2415 }
2418 /* Initialize the compile stack. */
2421 {
2422 #ifdef WCHAR
2426 #endif
2428 }
2433 /* Initialize the pattern buffer. */
2438 /* Set `used' to zero, so that if we return an error, the pattern
2439 printer (for debugging) will think there's no pattern. We reset it
2440 at the end. */
2443 /* Always count groups, whether or not bufp->no_sub is set. */
2446 #if !defined emacs && !defined SYNTAX_TABLE
2447 /* Initialize the syntax table. */
2449 #endif
2452 {
2455 enough space. This loses if buffer's address is bogus, but
2456 that is the user's responsibility. */
2457 #ifdef WCHAR
2458 /* Free bufp->buffer and allocate an array for wchar_t pattern
2459 buffer. */
2462 UCHAR_T);
2463 #else
2466 }
2467 else
2470 UCHAR_T);
2471 }
2474 #ifdef WCHAR
2478 }
2479 #ifdef WCHAR
2480 else
2482 #endif
2486 /* Loop through the uncompiled pattern until we're at the end. */
2488 {
2492 {
2494 {
2497 /* If context independent, it's an operator. */
2499 /* Otherwise, depends on what's come before. */
2502 else
2504 }
2509 {
2511 p == pend
2512 /* If context independent, it's an operator. */
2514 /* Otherwise, depends on what's next. */
2517 else
2519 }
2528 handle_plus:
2530 /* If there is no previous pattern... */
2532 {
2537 }
2539 {
2540 /* Are we optimizing this jump? */
2543 /* 1 means zero (many) matches is allowed. */
2546 /* If there is a sequence of repetition chars, collapse it
2547 down to just one (the right one). We can't combine
2548 interval operators with these because of, e.g., `a{2}*',
2549 which should only match an even number of `a's. */
2552 {
2563 ;
2566 {
2571 {
2572 PATUNFETCH;
2573 PATUNFETCH;
2575 }
2578 }
2579 else
2580 {
2581 PATUNFETCH;
2583 }
2585 /* If we get here, we found another repeat character. */
2586 }
2588 /* Star, etc. applied to an empty pattern is equivalent
2589 to an empty pattern. */
2593 /* Now we know whether or not zero matches is allowed
2594 and also whether or not two or more matches is allowed. */
2597 end a backward relative jump from `b' to before the next
2598 jump we're going to put in below (which jumps from
2599 laststart to after this jump).
2601 But if we are at the `*' in the exact sequence `.*\n',
2602 insert an unconditional jump backwards to the .,
2603 instead of the beginning of the loop. This way we only
2604 push a failure point once, instead of every time
2605 through the loop. */
2608 /* Allocate the space for the jump. */
2611 /* We know we are not at the first character of the pattern,
2612 because laststart was nonzero. And we've already
2613 incremented `p', by the way, to be the character after
2614 the `*'. Do we have to do something analogous here
2615 for null bytes, because of RE_DOT_NOT_NULL? */
2617 && zero_times_ok
2623 }
2624 else
2625 /* Anything else. */
2629 /* We've added more stuff to the buffer. */
2631 }
2633 /* On failure, jump from laststart to b + 3, which will be the
2634 end of the buffer after this jump is inserted. */
2635 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2636 'b + 3'. */
2645 {
2646 /* At least one repetition is required, so insert a
2647 `dummy_failure_jump' before the initial
2648 `on_failure_jump' instruction of the loop. This
2649 effects a skip over that instruction the first time
2650 we hit that loop. */
2655 }
2656 }
2667 {
2669 #ifdef WCHAR
2671 #else
2673 #endif
2676 #ifdef WCHAR
2677 /* We assume a charset(_not) structure as a wchar_t array.
2678 charset[0] = (re_opcode_t) charset(_not)
2679 charset[1] = l (= length of char_classes)
2680 charset[2] = m (= length of collating_symbols)
2681 charset[3] = n (= length of equivalence_classes)
2682 charset[4] = o (= length of char_ranges)
2683 charset[5] = p (= length of chars)
2685 charset[6] = char_class (wctype_t)
2686 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2687 ...
2688 charset[l+5] = char_class (wctype_t)
2690 charset[l+6] = collating_symbol (wchar_t)
2691 ...
2692 charset[l+m+5] = collating_symbol (wchar_t)
2693 ifdef _LIBC we use the index if
2694 _NL_COLLATE_SYMB_EXTRAMB instead of
2695 wchar_t string.
2697 charset[l+m+6] = equivalence_classes (wchar_t)
2698 ...
2699 charset[l+m+n+5] = equivalence_classes (wchar_t)
2700 ifdef _LIBC we use the index in
2701 _NL_COLLATE_WEIGHT instead of
2702 wchar_t string.
2704 charset[l+m+n+6] = range_start
2705 charset[l+m+n+7] = range_end
2706 ...
2707 charset[l+m+n+2o+4] = range_start
2708 charset[l+m+n+2o+5] = range_end
2709 ifdef _LIBC we use the value looked up
2710 in _NL_COLLATE_COLLSEQ instead of
2711 wchar_t character.
2713 charset[l+m+n+2o+6] = char
2714 ...
2715 charset[l+m+n+2o+p+5] = char
2717 */
2719 /* We need at least 6 spaces: the opcode, the length of
2720 char_classes, the length of collating_symbols, the length of
2721 equivalence_classes, the length of char_ranges, the length of
2722 chars. */
2725 /* Save b as laststart. And We use laststart as the pointer
2726 to the first element of the charset here.
2727 In other words, laststart[i] indicates charset[i]. */
2730 /* We test `*p == '^' twice, instead of using an if
2731 statement, so we only need one BUF_PUSH. */
2734 p++;
2736 /* Push the length of char_classes, the length of
2737 collating_symbols, the length of equivalence_classes, the
2738 length of char_ranges and the length of chars. */
2742 /* Remember the first position in the bracket expression. */
2745 /* charset_not matches newline according to a syntax bit. */
2748 {
2751 }
2753 /* Read in characters and ranges, setting map bits. */
2755 {
2760 /* \ might escape characters inside [...] and [^...]. */
2762 {
2770 }
2772 /* Could be the end of the bracket expression. If it's
2773 not (i.e., when the bracket expression is `[]' so
2774 far), the ']' character bit gets set way below. */
2778 /* Look ahead to see if it's a range when the last thing
2779 was a character class. */
2783 /* Look ahead to see if it's a range when the last thing
2784 was a character: if this is a hyphen not at the
2785 beginning or the end of a list, then it's the range
2786 operator. */
2791 {
2792 reg_errcode_t ret;
2793 /* Allocate the space for range_start and range_end. */
2795 /* Update the pointer to indicate end of buffer. */
2801 }
2804 reg_errcode_t ret;
2806 /* Move past the `-'. */
2808 /* Allocate the space for range_start and range_end. */
2810 /* Update the pointer to indicate end of buffer. */
2813 laststart);
2816 }
2818 /* See if we're at the beginning of a possible character
2819 class. */
2827 /* If pattern is `[[:'. */
2831 {
2837 else
2838 /* This is in any case an invalid class name. */
2840 }
2843 /* If isn't a word bracketed by `[:' and `:]':
2844 undo the ending character, the letters, and leave
2845 the leading `:' and `[' (but store them as character). */
2847 {
2851 /* Query the character class as wctype_t. */
2856 /* Throw away the ] at the end of the character
2857 class. */
2862 /* Allocate the space for character class. */
2864 /* Update the pointer to indicate end of buffer. */
2866 /* Move data which follow character classes
2867 not to violate the data. */
2874 /* Store the character class. */
2876 /* Update length of char_classes */
2880 }
2881 else
2882 {
2883 c1++;
2885 PATUNFETCH;
2891 }
2892 }
2895 {
2898 # ifdef _LIBC
2901 # endif
2905 /* If pattern is `[[=' or '[[.'. */
2909 {
2915 else
2916 /* This is in any case an invalid class name. */
2918 }
2922 {
2924 /* If we have no collation data we use the default
2925 collation in which each character is in a class
2926 by itself. It also means that ASCII is the
2927 character set and therefore we cannot have character
2928 with more than one byte in the multibyte
2929 representation. */
2931 /* If not defined _LIBC, we push the name and
2932 `\0' for the sake of matching performance. */
2935 # ifdef _LIBC
2938 # endif
2939 {
2942 }
2943 # ifdef _LIBC
2944 else
2945 {
2952 /* This #include defines a local function! */
2953 # include <locale/weightwc.h>
2956 {
2957 /* We push the index for equivalence class. */
2962 _NL_COLLATE_TABLEWC);
2965 _NL_COLLATE_WEIGHTWC);
2968 _NL_COLLATE_EXTRAWC);
2971 _NL_COLLATE_INDIRECTWC);
2975 /* This is no valid character. */
2979 }
2981 {
2982 /* We push collation sequence value
2983 for collating symbol. */
2993 /* We have to convert the name to a single-byte
2994 string. This is possible since the names
2995 consist of ASCII characters and the internal
2996 representation is UCS4. */
3000 table_size =
3002 _NL_COLLATE_SYMB_HASH_SIZEMB);
3005 _NL_COLLATE_SYMB_TABLEMB);
3008 _NL_COLLATE_SYMB_EXTRAMB);
3010 /* Locate the character in the hashing table. */
3017 {
3018 /* First compare the hashing value. */
3024 {
3025 /* Yep, this is the entry. */
3029 }
3031 /* Next entry. */
3033 }
3036 {
3037 /* Compute the index of the byte sequence
3038 in the table. */
3040 /* Adjust for the alignment. */
3044 }
3046 {
3047 /* No valid character. Match it as a
3048 single byte character. */
3051 /* Update the length of characters */
3055 /* Throw away the ] at the end of the
3056 collating symbol. */
3058 /* exit from the switch block. */
3060 }
3061 else
3063 }
3065 }
3066 # endif
3067 /* Throw away the ] at the end of the equivalence
3068 class (or collating symbol). */
3071 /* Allocate the space for the equivalence class
3072 (or collating symbol) (and '\0' if needed). */
3074 /* Update the pointer to indicate end of buffer. */
3079 /* Calculate the offset of char_ranges,
3080 which is next to equivalence_classes. */
3083 /* Insert space. */
3086 /* Write the equivalence_class and \0. */
3090 /* Update the length of equivalence_classes. */
3093 }
3096 /* Calculate the offset of the equivalence_classes,
3097 which is next to collating_symbols. */
3099 /* Insert space and write the collationg_symbol
3100 and \0. */
3105 /* In re_match_2_internal if range_start < -1, we
3106 assume -range_start is the offset of the
3107 collating symbol which is specified as
3108 the character of the range start. So we assign
3109 -(laststart[1] + laststart[2] + 6) to
3110 range_start. */
3112 /* Update the length of collating_symbol. */
3115 }
3116 }
3117 else
3118 {
3119 c1++;
3121 PATUNFETCH;
3127 }
3128 }
3129 else
3130 {
3135 }
3136 }
3139 /* Ensure that we have enough space to push a charset: the
3140 opcode, the length count, and the bitset; 34 bytes in all. */
3145 /* We test `*p == '^' twice, instead of using an if
3146 statement, so we only need one BUF_PUSH. */
3149 p++;
3151 /* Remember the first position in the bracket expression. */
3154 /* Push the number of bytes in the bitmap. */
3157 /* Clear the whole map. */
3160 /* charset_not matches newline according to a syntax bit. */
3165 /* Read in characters and ranges, setting map bits. */
3167 {
3172 /* \ might escape characters inside [...] and [^...]. */
3174 {
3181 }
3183 /* Could be the end of the bracket expression. If it's
3184 not (i.e., when the bracket expression is `[]' so
3185 far), the ']' character bit gets set way below. */
3189 /* Look ahead to see if it's a range when the last thing
3190 was a character class. */
3194 /* Look ahead to see if it's a range when the last thing
3195 was a character: if this is a hyphen not at the
3196 beginning or the end of a list, then it's the range
3197 operator. */
3202 {
3203 reg_errcode_t ret
3208 }
3212 reg_errcode_t ret;
3214 /* Move past the `-'. */
3220 }
3222 /* See if we're at the beginning of a possible character
3223 class. */
3232 /* If pattern is `[[:'. */
3236 {
3242 else
3243 /* This is in any case an invalid class name. */
3245 }
3248 /* If isn't a word bracketed by `[:' and `:]':
3249 undo the ending character, the letters, and leave
3250 the leading `:' and `[' (but set bits for them). */
3252 {
3253 # if defined _LIBC || WIDE_CHAR_SUPPORT
3263 /* Throw away the ] at the end of the character
3264 class. */
3270 {
3271 # ifdef _LIBC
3274 # else
3277 # endif
3282 }
3285 # else
3303 /* Throw away the ] at the end of the character
3304 class. */
3310 {
3311 /* This was split into 3 if's to
3312 avoid an arbitrary limit in some compiler. */
3331 }
3334 }
3335 else
3336 {
3337 c1++;
3339 PATUNFETCH;
3344 }
3345 }
3347 {
3349 # ifdef _LIBC
3352 # endif
3357 /* If pattern is `[[='. */
3361 {
3367 else
3368 /* This is in any case an invalid class name. */
3370 }
3374 {
3375 /* If we have no collation data we use the default
3376 collation in which each character is in a class
3377 by itself. It also means that ASCII is the
3378 character set and therefore we cannot have character
3379 with more than one byte in the multibyte
3380 representation. */
3381 # ifdef _LIBC
3383 # endif
3384 {
3388 /* Throw away the ] at the end of the equivalence
3389 class. */
3392 /* Set the bit for the character. */
3394 }
3395 # ifdef _LIBC
3396 else
3397 {
3398 /* Try to match the byte sequence in `str' against
3399 those known to the collate implementation.
3400 First find out whether the bytes in `str' are
3401 actually from exactly one character. */
3410 /* This #include defines a local function! */
3411 # include <locale/weight.h>
3424 /* This is no valid character. */
3427 /* Throw away the ] at the end of the equivalence
3428 class. */
3431 /* Now we have to go throught the whole table
3432 and find all characters which have the same
3433 first level weight.
3435 XXX Note that this is not entirely correct.
3436 we would have to match multibyte sequences
3437 but this is not possible with the current
3438 implementation. */
3440 /* XXX This test would have to be changed if we
3441 would allow matching multibyte sequences. */
3443 {
3447 /* Test whether the lenghts match. */
3449 {
3450 /* They do. New compare the bytes of
3451 the weight. */
3460 /* They match. Mark the character as
3461 acceptable. */
3463 }
3464 }
3465 }
3466 # endif
3468 }
3469 else
3470 {
3471 c1++;
3473 PATUNFETCH;
3478 }
3479 }
3481 {
3483 # ifdef _LIBC
3486 # endif
3491 /* If pattern is `[[.'. */
3495 {
3501 else
3502 /* This is in any case an invalid class name. */
3504 }
3508 {
3509 /* If we have no collation data we use the default
3510 collation in which each character is the name
3511 for its own class which contains only the one
3512 character. It also means that ASCII is the
3513 character set and therefore we cannot have character
3514 with more than one byte in the multibyte
3515 representation. */
3516 # ifdef _LIBC
3518 # endif
3519 {
3523 /* Throw away the ] at the end of the equivalence
3524 class. */
3527 /* Set the bit for the character. */
3530 }
3531 # ifdef _LIBC
3532 else
3533 {
3534 /* Try to match the byte sequence in `str' against
3535 those known to the collate implementation.
3536 First find out whether the bytes in `str' are
3537 actually from exactly one character. */
3546 table_size =
3548 _NL_COLLATE_SYMB_HASH_SIZEMB);
3551 _NL_COLLATE_SYMB_TABLEMB);
3554 _NL_COLLATE_SYMB_EXTRAMB);
3556 /* Locate the character in the hashing table. */
3563 {
3564 /* First compare the hashing value. */
3571 {
3572 /* Yep, this is the entry. */
3576 }
3578 /* Next entry. */
3580 }
3583 /* This is no valid character. */
3586 /* Throw away the ] at the end of the equivalence
3587 class. */
3590 /* Now add the multibyte character(s) we found
3591 to the accept list.
3593 XXX Note that this is not entirely correct.
3594 we would have to match multibyte sequences
3595 but this is not possible with the current
3596 implementation. Also, we have to match
3597 collating symbols, which expand to more than
3598 one file, as a whole and not allow the
3599 individual bytes. */
3604 {
3607 }
3608 }
3609 # endif
3611 }
3612 else
3613 {
3614 c1++;
3616 PATUNFETCH;
3621 }
3622 }
3623 else
3624 {
3628 }
3629 }
3631 /* Discard any (non)matching list bytes that are all 0 at the
3632 end of the map. Decrease the map-length byte too. */
3637 }
3644 else
3651 else
3658 else
3665 else
3672 else
3679 /* Do not translate the character after the \, so that we can
3680 distinguish, e.g., \B from \b, even if we normally would
3681 translate, e.g., B to b. */
3685 {
3690 handle_open:
3692 regnum++;
3695 {
3697 compile_stack_elt_t);
3701 }
3703 /* These are the values to restore when we hit end of this
3704 group. They are all relative offsets, so that if the
3705 whole pattern moves because of realloc, they will still
3706 be valid. */
3708 COMPILE_STACK_TOP.fixup_alt_jump
3713 /* We will eventually replace the 0 with the number of
3714 groups inner to this one. But do not push a
3715 start_memory for groups beyond the last one we can
3716 represent in the compiled pattern. */
3718 {
3722 }
3729 /* If we've reached MAX_REGNUM groups, then this open
3730 won't actually generate any code, so we'll have to
3731 clear pending_exact explicitly. */
3740 {
3743 else
3745 }
3747 handle_close:
3750 alternative for a possible future
3751 `pop_failure_jump' to pop. See comments at
3752 `push_dummy_failure' in `re_match_2'. */
3755 /* We allocated space for this jump when we assigned
3756 to `fixup_alt_jump', in the `handle_alt' case below. */
3758 }
3760 /* See similar code for backslashed left paren above. */
3762 {
3765 else
3767 }
3769 /* Since we just checked for an empty stack above, this
3770 ``can't happen''. */
3772 {
3773 /* We don't just want to restore into `regnum', because
3774 later groups should continue to be numbered higher,
3775 as in `(ab)c(de)' -- the second group is #2. */
3776 regnum_t this_group_regnum;
3780 fixup_alt_jump
3786 /* If we've reached MAX_REGNUM groups, then this open
3787 won't actually generate any code, so we'll have to
3788 clear pending_exact explicitly. */
3791 /* We're at the end of the group, so now we know how many
3792 groups were inside this one. */
3794 {
3795 UCHAR_T *inner_group_loc
3801 }
3802 }
3809 handle_alt:
3813 /* Insert before the previous alternative a jump which
3814 jumps to this alternative if the former fails. */
3821 /* The alternative before this one has a jump after it
3822 which gets executed if it gets matched. Adjust that
3823 jump so it will jump to this alternative's analogous
3824 jump (put in below, which in turn will jump to the next
3825 (if any) alternative's such jump, etc.). The last such
3826 jump jumps to the correct final destination. A picture:
3827 _____ _____
3828 | | | |
3829 | v | v
3830 a | b | c
3832 If we are at `b', then fixup_alt_jump right now points to a
3833 three-byte space after `a'. We'll put in the jump, set
3834 fixup_alt_jump to right after `b', and leave behind three
3835 bytes which we'll fill in when we get to after `c'. */
3840 /* Mark and leave space for a jump after this alternative,
3841 to be filled in later either by next alternative or
3842 when know we're at the end of a series of alternatives. */
3853 /* If \{ is a literal. */
3855 /* If we're at `\{' and it's not the open-interval
3856 operator. */
3860 handle_interval:
3861 {
3862 /* If got here, then the syntax allows intervals. */
3864 /* At least (most) this many matches must be made. */
3867 /* Place in the uncompiled pattern (i.e., just after
3868 the '{') to go back to if the interval is invalid. */
3877 {
3881 }
3882 else
3883 /* Interval such as `{1}' => match exactly once. */
3890 {
3894 }
3899 /* If it's invalid to have no preceding re. */
3901 {
3907 else
3909 }
3911 /* We just parsed a valid interval. */
3916 /* If the upper bound is zero, don't want to succeed at
3917 all; jump from `laststart' to `b + 3', which will be
3918 the end of the buffer after we insert the jump. */
3919 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3920 instead of 'b + 3'. */
3922 {
3927 }
3929 /* Otherwise, we have a nontrivial interval. When
3930 we're all done, the pattern will look like:
3931 set_number_at <jump count> <upper bound>
3932 set_number_at <succeed_n count> <lower bound>
3933 succeed_n <after jump addr> <succeed_n count>
3934 <body of loop>
3935 jump_n <succeed_n addr> <jump count>
3936 (The upper bound and `jump_n' are omitted if
3937 `upper_bound' is 1, though.) */
3938 else
3940 more at the end of the loop. */
3946 /* Initialize lower bound of the `succeed_n', even
3947 though it will be set during matching by its
3948 attendant `set_number_at' (inserted next),
3949 because `re_compile_fastmap' needs to know.
3950 Jump to the `jump_n' we might insert below. */
3957 /* Code to initialize the lower bound. Insert
3958 before the `succeed_n'. The `5' is the last two
3959 bytes of this `set_number_at', plus 3 bytes of
3960 the following `succeed_n'. */
3961 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3962 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3963 of the following `succeed_n'. */
3970 append a backward jump to the `succeed_n'
3971 that starts this interval.
3973 When we've reached this during matching,
3974 we'll have matched the interval once, so
3975 jump back only `upper_bound - 1' times. */
3981 /* The location we want to set is the second
3982 parameter of the `jump_n'; that is `b-2' as
3983 an absolute address. `laststart' will be
3984 the `set_number_at' we're about to insert;
3985 `laststart+3' the number to set, the source
3986 for the relative address. But we are
3987 inserting into the middle of the pattern --
3988 so everything is getting moved up by 5.
3989 Conclusion: (b - 2) - (laststart + 3) + 5,
3990 i.e., b - laststart.
3992 We insert this at the beginning of the loop
3993 so that if we fail during matching, we'll
3994 reinitialize the bounds. */
3999 }
4000 }
4004 invalid_interval:
4007 unfetch_interval:
4008 /* Match the characters as literals. */
4013 else
4015 }
4017 #ifdef emacs
4018 /* There is no way to specify the before_dot and after_dot
4019 operators. rms says this is ok. --karl */
4100 /* Can't back reference to a subexpression if inside of it. */
4113 else
4117 normal_backslash:
4118 /* You might think it would be useful for \ to mean
4119 not to translate; but if we don't translate it
4120 it will never match anything. */
4123 }
4128 /* Expects the character in `c'. */
4129 normal_char:
4130 /* If no exactn currently being built. */
4132 #ifdef WCHAR
4133 /* If last exactn handle binary(or character) and
4134 new exactn handle character(or binary). */
4138 /* If last exactn not at current position. */
4141 /* We have only one byte following the exactn for the count. */
4144 /* If followed by a repetition operator. */
4153 {
4154 /* Start building a new exactn. */
4158 #ifdef WCHAR
4159 /* Is this exactn binary data or character? */
4163 else
4165 #else
4169 }
4178 /* Through the pattern now. */
4186 /* If we don't want backtracking, force success
4187 the first time we reach the end of the compiled pattern. */
4191 #ifdef WCHAR
4195 #endif
4198 /* We have succeeded; set the length of the buffer. */
4199 #ifdef WCHAR
4201 #else
4203 #endif
4205 #ifdef DEBUG
4207 {
4210 }
4213 #ifndef MATCH_MAY_ALLOCATE
4214 /* Initialize the failure stack to the largest possible stack. This
4215 isn't necessary unless we're trying to avoid calling alloca in
4216 the search and match routines. */
4217 {
4220 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4221 is strictly greater than re_max_failures, the largest possible stack
4222 is 2 * re_max_failures failure points. */
4224 {
4227 # ifdef emacs
4229 fail_stack.stack
4232 else
4233 fail_stack.stack
4239 fail_stack.stack
4242 else
4243 fail_stack.stack
4248 }
4251 }
4257 /* Subroutines for `regex_compile'. */
4259 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4260 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4262 static void
4264 re_opcode_t op;
4267 {
4270 }
4273 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4274 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4276 static void
4278 re_opcode_t op;
4281 {
4285 }
4288 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4289 for OP followed by two-byte integer parameter ARG. */
4290 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4292 static void
4294 re_opcode_t op;
4298 {
4306 }
4309 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4310 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4312 static void
4314 re_opcode_t op;
4318 {
4326 }
4329 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4330 after an alternative or a begin-subexpression. We assume there is at
4331 least one character before the ^. */
4333 static boolean
4336 reg_syntax_t syntax;
4337 {
4341 return
4342 /* After a subexpression? */
4344 /* After an alternative? */
4346 }
4349 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4350 at least one character after the $, i.e., `P < PEND'. */
4352 static boolean
4355 reg_syntax_t syntax;
4356 {
4361 return
4362 /* Before a subexpression? */
4365 /* Before an alternative? */
4368 }
4372 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4373 false if it's not. */
4375 static boolean
4377 compile_stack_type compile_stack;
4378 regnum_t regnum;
4379 {
4384 this_element--)
4389 }
4392 #ifdef INSIDE_RECURSION
4394 #ifdef WCHAR
4395 /* This insert space, which size is "num", into the pattern at "loc".
4396 "end" must point the end of the allocated buffer. */
4397 static void
4402 {
4408 }
4411 #ifdef WCHAR
4412 static reg_errcode_t
4414 char_set)
4415 CHAR_T range_start_char;
4418 RE_TRANSLATE_TYPE translate;
4419 reg_syntax_t syntax;
4420 {
4423 reg_errcode_t ret;
4424 # ifdef _LIBC
4427 # endif
4431 # ifdef _LIBC
4434 {
4436 _NL_COLLATE_COLLSEQWC);
4441 {
4442 /* range_start is a collating symbol. */
4444 /* Retreive the index and get collation sequence value. */
4447 }
4448 else
4453 /* Report an error if the range is empty and the syntax prohibits
4454 this. */
4458 /* Insert space to the end of the char_ranges. */
4463 }
4464 else
4465 # endif
4466 {
4468 range_start_char;
4470 /* Report an error if the range is empty and the syntax prohibits
4471 this. */
4475 /* Insert space to the end of the char_ranges. */
4480 }
4481 /* Have to increment the pointer into the pattern string, so the
4482 caller isn't still at the ending character. */
4486 }
4488 /* Read the ending character of a range (in a bracket expression) from the
4489 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4490 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4491 Then we set the translation of all bits between the starting and
4492 ending characters (inclusive) in the compiled pattern B.
4494 Return an error code.
4496 We use these short variable names so we can use the same macros as
4497 `regex_compile' itself. */
4499 static reg_errcode_t
4503 RE_TRANSLATE_TYPE translate;
4504 reg_syntax_t syntax;
4506 {
4509 reg_errcode_t ret;
4510 # if _LIBC
4514 # else
4516 # endif
4521 /* Have to increment the pointer into the pattern string, so the
4522 caller isn't still at the ending character. */
4525 /* Report an error if the range is empty and the syntax prohibits this. */
4528 # if _LIBC
4530 _NL_COLLATE_COLLSEQMB);
4535 {
4539 {
4542 }
4543 }
4544 # else
4545 /* Here we see why `this_char' has to be larger than an `unsigned
4546 char' -- we would otherwise go into an infinite loop, since all
4547 characters <= 0xff. */
4549 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4550 and some compilers cast it to int implicitly, so following for_loop
4551 may fall to (almost) infinite loop.
4552 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4553 To avoid this, we cast p[0] to unsigned int and truncate it. */
4557 {
4560 }
4561 # endif
4564 }
4566 \f
4567 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4568 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4569 characters can start a string that matches the pattern. This fastmap
4570 is used by re_search to skip quickly over impossible starting points.
4572 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4573 area as BUFP->fastmap.
4575 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4576 the pattern buffer.
4578 Returns 0 if we succeed, -2 if an internal error. */
4580 #ifdef WCHAR
4581 /* local function for re_compile_fastmap.
4582 truncate wchar_t character to char. */
4585 static unsigned char
4587 CHAR_T c;
4588 {
4593 # ifdef _LIBC
4595 # else
4597 # endif
4599 }
4602 static int
4605 {
4607 #ifdef MATCH_MAY_ALLOCATE
4609 #endif
4610 #ifndef REGEX_MALLOC
4612 #endif
4616 #ifdef WCHAR
4617 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4618 pattern to (char*) in regex_compile. */
4627 #ifdef REL_ALLOC
4628 /* This holds the pointer to the failure stack, when
4629 it is allocated relocatably. */
4631 #endif
4633 /* Assume that each path through the pattern can be null until
4634 proven otherwise. We set this false at the bottom of switch
4635 statement, to which we get only if a particular path doesn't
4636 match the empty string. */
4639 /* We aren't doing a `succeed_n' to begin with. */
4650 {
4652 {
4653 /* We have reached the (effective) end of pattern. */
4655 {
4658 /* Reset for next path. */
4664 }
4665 else
4667 }
4669 /* We should never be about to go beyond the end of the pattern. */
4673 {
4675 /* I guess the idea here is to simply not bother with a fastmap
4676 if a backreference is used, since it's too hard to figure out
4677 the fastmap for the corresponding group. Setting
4678 `can_be_null' stops `re_search_2' from using the fastmap, so
4679 that is all we do. */
4685 /* Following are the cases which match a character. These end
4686 with `break'. */
4688 #ifdef WCHAR
4697 #ifdef MBS_SUPPORT
4701 #endif
4703 #ifdef WCHAR
4704 /* It is hard to distinguish fastmap from (multi byte) characters
4705 which depends on current locale. */
4721 /* Chars beyond end of map must be allowed. */
4746 {
4749 /* `.' matches anything ... */
4753 /* ... except perhaps newline. */
4757 /* Return if we have already set `can_be_null'; if we have,
4758 then the fastmap is irrelevant. Something's wrong here. */
4762 /* Otherwise, have to check alternative paths. */
4764 }
4766 #ifdef emacs
4783 /* All cases after this match the empty string. These end with
4784 `continue'. */
4818 /* Jump backward implies we just went through the body of a
4819 loop and matched nothing. Opcode jumped to should be
4820 `on_failure_jump' or `succeed_n'. Just treat it like an
4821 ordinary jump. For a * loop, it has pushed its failure
4822 point already; if so, discard that as redundant. */
4827 p++;
4831 /* If what's on the stack is where we are now, pop it. */
4841 handle_on_failure_jump:
4844 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4845 end of the pattern. We don't want to push such a point,
4846 since when we restore it above, entering the switch will
4847 increment `p' past the end of the pattern. We don't need
4848 to push such a point since we obviously won't find any more
4849 fastmap entries beyond `pend'. Such a pattern can match
4850 the null string, though. */
4852 {
4854 {
4857 }
4858 }
4859 else
4863 {
4866 }
4872 /* Get to the number of times to succeed. */
4875 /* Increment p past the n for when k != 0. */
4878 {
4882 }
4901 /* Getting here means we have found the possible starting
4902 characters for one path of the pattern -- and that the empty
4903 string does not match. We need not follow this path further.
4904 Instead, look at the next alternative (remembered on the
4905 stack), or quit if no more. The test at the top of the loop
4906 does these things. */
4911 /* Set `can_be_null' for the last path (also the first path, if the
4912 pattern is empty). */
4915 done:
4918 }
4922 int
4925 {
4926 # ifdef MBS_SUPPORT
4929 else
4930 # endif
4933 #ifdef _LIBC
4935 #endif
4936 \f
4938 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4939 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4940 this memory for recording register information. STARTS and ENDS
4941 must be allocated using the malloc library routine, and must each
4942 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4944 If NUM_REGS == 0, then subsequent matches should allocate their own
4945 register data.
4947 Unless this function is called, the first search or match using
4948 PATTERN_BUFFER will allocate its own register data, without
4949 freeing the old data. */
4951 void
4957 {
4959 {
4964 }
4965 else
4966 {
4970 }
4971 }
4972 #ifdef _LIBC
4974 #endif
4975 \f
4976 /* Searching routines. */
4978 /* Like re_search_2, below, but only one string is specified, and
4979 doesn't let you say where to stop matching. */
4981 int
4987 {
4990 }
4991 #ifdef _LIBC
4993 #endif
4996 /* Using the compiled pattern in BUFP->buffer, first tries to match the
4997 virtual concatenation of STRING1 and STRING2, starting first at index
4998 STARTPOS, then at STARTPOS + 1, and so on.
5000 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
5002 RANGE is how far to scan while trying to match. RANGE = 0 means try
5003 only at STARTPOS; in general, the last start tried is STARTPOS +
5004 RANGE.
5006 In REGS, return the indices of the virtual concatenation of STRING1
5007 and STRING2 that matched the entire BUFP->buffer and its contained
5008 subexpressions.
5010 Do not consider matching one past the index STOP in the virtual
5011 concatenation of STRING1 and STRING2.
5013 We return either the position in the strings at which the match was
5014 found, -1 if no match, or -2 if error (such as failure
5015 stack overflow). */
5017 int
5026 {
5027 # ifdef MBS_SUPPORT
5031 else
5032 # endif
5036 #ifdef _LIBC
5038 #endif
5042 #ifdef INSIDE_RECURSION
5044 #ifdef MATCH_MAY_ALLOCATE
5045 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
5046 #else
5047 # define FREE_VAR(var) if (var) free (var); var = NULL
5048 #endif
5050 #ifdef WCHAR
5051 # define MAX_ALLOCA_SIZE 2000
5053 # define FREE_WCS_BUFFERS() \
5054 do { \
5055 if (size1 > MAX_ALLOCA_SIZE) \
5056 { \
5057 free (wcs_string1); \
5058 free (mbs_offset1); \
5059 } \
5060 else \
5061 { \
5062 FREE_VAR (wcs_string1); \
5063 FREE_VAR (mbs_offset1); \
5064 } \
5065 if (size2 > MAX_ALLOCA_SIZE) \
5066 { \
5067 free (wcs_string2); \
5068 free (mbs_offset2); \
5069 } \
5070 else \
5071 { \
5072 FREE_VAR (wcs_string2); \
5073 FREE_VAR (mbs_offset2); \
5074 } \
5075 } while (0)
5077 #endif
5080 static int
5090 {
5096 #ifdef WCHAR
5097 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5099 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5101 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5103 /* They hold whether each wchar_t is binary data or not. */
5107 /* Check for out-of-range STARTPOS. */
5111 /* Fix up RANGE if it might eventually take us outside
5112 the virtual concatenation of STRING1 and STRING2.
5113 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5119 /* If the search isn't to be a backwards one, don't waste time in a
5120 search for a pattern that must be anchored. */
5123 /* `begline' is like `begbuf' if it cannot match at newlines. */
5126 {
5129 else
5131 }
5133 #ifdef emacs
5134 /* In a forward search for something that starts with \=.
5135 don't keep searching past point. */
5137 {
5141 }
5144 /* Update the fastmap now if not correct already. */
5149 #ifdef WCHAR
5150 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5151 fill them with converted string. */
5153 {
5155 {
5159 }
5160 else
5161 {
5165 }
5167 {
5169 {
5173 }
5174 else
5175 {
5179 }
5181 }
5187 else
5189 }
5191 {
5193 {
5197 }
5198 else
5199 {
5203 }
5205 {
5209 else
5212 }
5218 else
5220 }
5224 /* Loop through the string, looking for a place to start matching. */
5226 {
5227 /* If a fastmap is supplied, skip quickly over characters that
5228 cannot be the start of a match. If the pattern can match the
5229 null string, however, we don't need to skip characters; we want
5230 the first null string. */
5232 {
5234 {
5244 /* Written out as an if-else to avoid testing `translate'
5245 inside the loop. */
5250 range--;
5251 else
5253 range--;
5256 }
5258 {
5265 }
5266 }
5268 /* If can't match the null string, and that's all we have left, fail. */
5271 {
5272 #ifdef WCHAR
5274 #endif
5276 }
5278 #ifdef WCHAR
5289 #ifndef REGEX_MALLOC
5290 # ifdef C_ALLOCA
5292 # endif
5293 #endif
5296 {
5297 #ifdef WCHAR
5299 #endif
5301 }
5304 {
5305 #ifdef WCHAR
5307 #endif
5309 }
5311 advance:
5315 {
5316 range--;
5317 startpos++;
5318 }
5319 else
5320 {
5321 range++;
5322 startpos--;
5323 }
5324 }
5325 #ifdef WCHAR
5327 #endif
5329 }
5331 #ifdef WCHAR
5332 /* This converts PTR, a pointer into one of the search wchar_t strings
5333 `string1' and `string2' into an multibyte string offset from the
5334 beginning of that string. We use mbs_offset to optimize.
5335 See convert_mbs_to_wcs. */
5336 # define POINTER_TO_OFFSET(ptr) \
5337 (FIRST_STRING_P (ptr) \
5338 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5339 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5340 + csize1)))
5342 /* This converts PTR, a pointer into one of the search strings `string1'
5343 and `string2' into an offset from the beginning of that string. */
5344 # define POINTER_TO_OFFSET(ptr) \
5345 (FIRST_STRING_P (ptr) \
5346 ? ((regoff_t) ((ptr) - string1)) \
5347 : ((regoff_t) ((ptr) - string2 + size1)))
5350 /* Macros for dealing with the split strings in re_match_2. */
5352 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5354 /* Call before fetching a character with *d. This switches over to
5355 string2 if necessary. */
5356 #define PREFETCH() \
5357 while (d == dend) \
5358 { \
5360 if (dend == end_match_2) \
5361 goto fail; \
5363 d = string2; \
5364 dend = end_match_2; \
5365 }
5367 /* Test if at very beginning or at very end of the virtual concatenation
5368 of `string1' and `string2'. If only one string, it's `string2'. */
5369 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5370 #define AT_STRINGS_END(d) ((d) == end2)
5373 /* Test if D points to a character which is word-constituent. We have
5374 two special cases to check for: if past the end of string1, look at
5375 the first character in string2; and if before the beginning of
5376 string2, look at the last character in string1. */
5377 #ifdef WCHAR
5378 /* Use internationalized API instead of SYNTAX. */
5379 # define WORDCHAR_P(d) \
5380 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5381 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5382 || ((d) == end1 ? *string2 \
5385 # define WORDCHAR_P(d) \
5386 (SYNTAX ((d) == end1 ? *string2 \
5387 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5388 == Sword)
5391 /* Disabled due to a compiler bug -- see comment at case wordbound */
5392 #if 0
5393 /* Test if the character before D and the one at D differ with respect
5394 to being word-constituent. */
5395 #define AT_WORD_BOUNDARY(d) \
5396 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5397 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5398 #endif
5400 /* Free everything we malloc. */
5401 #ifdef MATCH_MAY_ALLOCATE
5402 # ifdef WCHAR
5403 # define FREE_VARIABLES() \
5404 do { \
5405 REGEX_FREE_STACK (fail_stack.stack); \
5406 FREE_VAR (regstart); \
5407 FREE_VAR (regend); \
5408 FREE_VAR (old_regstart); \
5409 FREE_VAR (old_regend); \
5410 FREE_VAR (best_regstart); \
5411 FREE_VAR (best_regend); \
5412 FREE_VAR (reg_info); \
5413 FREE_VAR (reg_dummy); \
5414 FREE_VAR (reg_info_dummy); \
5415 if (!cant_free_wcs_buf) \
5416 { \
5417 FREE_VAR (string1); \
5418 FREE_VAR (string2); \
5419 FREE_VAR (mbs_offset1); \
5420 FREE_VAR (mbs_offset2); \
5421 } \
5422 } while (0)
5424 # define FREE_VARIABLES() \
5425 do { \
5426 REGEX_FREE_STACK (fail_stack.stack); \
5427 FREE_VAR (regstart); \
5428 FREE_VAR (regend); \
5429 FREE_VAR (old_regstart); \
5430 FREE_VAR (old_regend); \
5431 FREE_VAR (best_regstart); \
5432 FREE_VAR (best_regend); \
5433 FREE_VAR (reg_info); \
5434 FREE_VAR (reg_dummy); \
5435 FREE_VAR (reg_info_dummy); \
5436 } while (0)
5438 #else
5439 # ifdef WCHAR
5440 # define FREE_VARIABLES() \
5441 do { \
5442 if (!cant_free_wcs_buf) \
5443 { \
5444 FREE_VAR (string1); \
5445 FREE_VAR (string2); \
5446 FREE_VAR (mbs_offset1); \
5447 FREE_VAR (mbs_offset2); \
5448 } \
5449 } while (0)
5455 /* These values must meet several constraints. They must not be valid
5456 register values; since we have a limit of 255 registers (because
5457 we use only one byte in the pattern for the register number), we can
5458 use numbers larger than 255. They must differ by 1, because of
5459 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5460 be larger than the value for the highest register, so we do not try
5461 to actually save any registers when none are active. */
5462 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5463 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5464 \f
5466 /* Matching routines. */
5469 /* re_match is like re_match_2 except it takes only a single string. */
5471 int
5477 {
5479 # ifdef MBS_SUPPORT
5484 else
5485 # endif
5488 # ifndef REGEX_MALLOC
5489 # ifdef C_ALLOCA
5491 # endif
5492 # endif
5494 }
5495 # ifdef _LIBC
5497 # endif
5502 #ifdef INSIDE_RECURSION
5516 /* re_match_2 matches the compiled pattern in BUFP against the
5517 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5518 and SIZE2, respectively). We start matching at POS, and stop
5519 matching at STOP.
5521 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5522 store offsets for the substring each group matched in REGS. See the
5523 documentation for exactly how many groups we fill.
5525 We return -1 if no match, -2 if an internal error (such as the
5526 failure stack overflowing). Otherwise, we return the length of the
5527 matched substring. */
5529 int
5537 {
5539 # ifdef MBS_SUPPORT
5544 else
5545 # endif
5549 #ifndef REGEX_MALLOC
5550 # ifdef C_ALLOCA
5552 # endif
5553 #endif
5555 }
5556 #ifdef _LIBC
5558 #endif
5562 #ifdef INSIDE_RECURSION
5564 #ifdef WCHAR
5567 /* This check the substring (from 0, to length) of the multibyte string,
5568 to which offset_buffer correspond. And count how many wchar_t_characters
5569 the substring occupy. We use offset_buffer to optimization.
5570 See convert_mbs_to_wcs. */
5572 static int
5576 {
5579 /* Check whether the size is valid. */
5586 /* If there are no multibyte character, offset_buffer[i] == i.
5587 Optmize for this case. */
5591 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5596 {
5604 else
5606 }
5609 }
5612 /* This is a separate function so that we can force an alloca cleanup
5613 afterwards. */
5614 #ifdef WCHAR
5615 static int
5625 /* string1 == string2 == NULL means string1/2, size1/2 and
5626 mbs_offset1/2 need seting up in this function. */
5627 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5629 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5631 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5634 static int
5644 {
5645 /* General temporaries. */
5648 #ifdef WCHAR
5649 /* They hold whether each wchar_t is binary data or not. */
5651 /* If true, we can't free string1/2, mbs_offset1/2. */
5655 /* Just past the end of the corresponding string. */
5658 /* Pointers into string1 and string2, just past the last characters in
5659 each to consider matching. */
5662 /* Where we are in the data, and the end of the current string. */
5665 /* Where we are in the pattern, and the end of the pattern. */
5666 #ifdef WCHAR
5674 /* Mark the opcode just after a start_memory, so we can test for an
5675 empty subpattern when we get to the stop_memory. */
5678 /* We use this to map every character in the string. */
5681 /* Failure point stack. Each place that can handle a failure further
5682 down the line pushes a failure point on this stack. It consists of
5683 restart, regend, and reg_info for all registers corresponding to
5684 the subexpressions we're currently inside, plus the number of such
5685 registers, and, finally, two char *'s. The first char * is where
5686 to resume scanning the pattern; the second one is where to resume
5687 scanning the strings. If the latter is zero, the failure point is
5688 a ``dummy''; if a failure happens and the failure point is a dummy,
5689 it gets discarded and the next next one is tried. */
5692 #endif
5693 #ifdef DEBUG
5696 #endif
5698 #ifdef REL_ALLOC
5699 /* This holds the pointer to the failure stack, when
5700 it is allocated relocatably. */
5702 #endif
5704 /* We fill all the registers internally, independent of what we
5705 return, for use in backreferences. The number here includes
5706 an element for register zero. */
5709 /* The currently active registers. */
5713 /* Information on the contents of registers. These are pointers into
5714 the input strings; they record just what was matched (on this
5715 attempt) by a subexpression part of the pattern, that is, the
5716 regnum-th regstart pointer points to where in the pattern we began
5717 matching and the regnum-th regend points to right after where we
5718 stopped matching the regnum-th subexpression. (The zeroth register
5719 keeps track of what the whole pattern matches.) */
5722 #endif
5724 /* If a group that's operated upon by a repetition operator fails to
5725 match anything, then the register for its start will need to be
5726 restored because it will have been set to wherever in the string we
5727 are when we last see its open-group operator. Similarly for a
5728 register's end. */
5731 #endif
5733 /* The is_active field of reg_info helps us keep track of which (possibly
5734 nested) subexpressions we are currently in. The matched_something
5735 field of reg_info[reg_num] helps us tell whether or not we have
5736 matched any of the pattern so far this time through the reg_num-th
5737 subexpression. These two fields get reset each time through any
5738 loop their register is in. */
5741 #endif
5743 /* The following record the register info as found in the above
5744 variables when we find a match better than any we've seen before.
5745 This happens as we backtrack through the failure points, which in
5746 turn happens only if we have not yet matched the entire string. */
5750 #endif
5752 /* Logically, this is `best_regend[0]'. But we don't want to have to
5753 allocate space for that if we're not allocating space for anything
5754 else (see below). Also, we never need info about register 0 for
5755 any of the other register vectors, and it seems rather a kludge to
5756 treat `best_regend' differently than the rest. So we keep track of
5757 the end of the best match so far in a separate variable. We
5758 initialize this to NULL so that when we backtrack the first time
5759 and need to test it, it's not garbage. */
5762 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5765 /* Used when we pop values we don't care about. */
5769 #endif
5771 #ifdef DEBUG
5772 /* Counts the total number of registers pushed. */
5774 #endif
5780 #ifdef MATCH_MAY_ALLOCATE
5781 /* Do not bother to initialize all the register variables if there are
5782 no groups in the pattern, as it takes a fair amount of time. If
5783 there are groups, we include space for register 0 (the whole
5784 pattern), even though we never use it, since it simplifies the
5785 array indexing. We should fix this. */
5787 {
5800 {
5803 }
5804 }
5805 else
5806 {
5807 /* We must initialize all our variables to NULL, so that
5808 `FREE_VARIABLES' doesn't try to free them. */
5812 }
5815 /* The starting position is bogus. */
5816 #ifdef WCHAR
5820 #endif
5821 {
5824 }
5826 #ifdef WCHAR
5827 /* Allocate wchar_t array for string1 and string2 and
5828 fill them with converted string. */
5830 {
5831 /* We need seting up buffers here. */
5833 /* We must free wcs buffers in this function. */
5837 {
5842 {
5847 }
5848 }
5850 {
5855 {
5862 }
5867 }
5868 }
5870 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5871 pattern to (char*) in regex_compile. */
5877 /* Initialize subexpression text positions to -1 to mark ones that no
5878 start_memory/stop_memory has been seen for. Also initialize the
5879 register information struct. */
5881 {
5889 }
5891 /* We move `string1' into `string2' if the latter's empty -- but not if
5892 `string1' is null. */
5894 {
5899 #ifdef WCHAR
5904 #endif
5905 }
5909 /* Compute where to stop matching, within the two strings. */
5910 #ifdef WCHAR
5912 {
5916 }
5917 else
5918 {
5924 }
5929 }
5930 #else
5932 {
5935 }
5936 else
5937 {
5940 }
5943 /* `p' scans through the pattern as `d' scans through the data.
5944 `dend' is the end of the input string that `d' points within. `d'
5945 is advanced into the following input string whenever necessary, but
5946 this happens before fetching; therefore, at the beginning of the
5947 loop, `d' can be pointing at the end of a string, but it cannot
5948 equal `string2'. */
5949 #ifdef WCHAR
5951 {
5955 }
5956 else
5957 {
5961 }
5967 }
5968 #else
5970 {
5973 }
5974 else
5975 {
5978 }
5987 /* This loops over pattern commands. It exits by returning from the
5988 function if the match is complete, or it drops through if the match
5989 fails at this starting point in the input data. */
5991 {
5992 #ifdef _LIBC
5994 #else
5996 #endif
6002 /* If we haven't matched the entire string, and we want the
6003 longest match, try backtracking. */
6005 {
6006 /* 1 if this match ends in the same string (string1 or string2)
6007 as the best previous match. */
6010 /* 1 if this match is the best seen so far. */
6011 boolean best_match_p;
6013 /* AIX compiler got confused when this was combined
6014 with the previous declaration. */
6017 else
6025 /* If exceeds best match so far, save it. */
6027 {
6034 {
6037 }
6038 }
6040 }
6042 /* If no failure points, don't restore garbage. And if
6043 last match is real best match, don't restore second
6044 best one. */
6046 {
6047 restore_best_regs:
6048 /* Restore best match. It may happen that `dend ==
6049 end_match_1' while the restored d is in string2.
6050 For example, the pattern `x.*y.*z' against the
6051 strings `x-' and `y-z-', if the two strings are
6052 not consecutive in memory. */
6060 {
6063 }
6064 }
6067 succeed_label:
6069 /* If caller wants register contents data back, do it. */
6071 {
6072 /* Have the register data arrays been allocated? */
6075 extra element beyond `num_regs' for the `-1' marker
6076 GNU code uses. */
6081 {
6084 }
6086 }
6089 allocated, reallocate them. If we need fewer, just
6090 leave it alone. */
6092 {
6097 {
6100 }
6101 }
6102 }
6103 else
6104 {
6105 /* These braces fend off a "empty body in an else-statement"
6106 warning under GCC when assert expands to nothing. */
6108 }
6110 /* Convert the pointer data in `regstart' and `regend' to
6111 indices. Register zero has to be set differently,
6112 since we haven't kept track of any info for it. */
6114 {
6116 #ifdef WCHAR
6120 else
6123 #else
6128 }
6130 /* Go through the first `min (num_regs, regs->num_regs)'
6131 registers, since that is all we initialized. */
6133 mcnt++)
6134 {
6137 else
6138 {
6143 }
6144 }
6146 /* If the regs structure we return has more elements than
6147 were in the pattern, set the extra elements to -1. If
6148 we (re)allocated the registers, this is the case,
6149 because we always allocate enough to have at least one
6150 -1 at the end. */
6160 #ifdef WCHAR
6163 else
6165 csize1;
6167 #else
6169 ? string1
6177 }
6179 /* Otherwise match next pattern command. */
6181 {
6182 /* Ignore these. Used to ignore the n of succeed_n's which
6183 currently have n == 0. */
6192 /* Match the next n pattern characters exactly. The following
6193 byte in the pattern defines n, and the n bytes after that
6194 are the characters to match. */
6196 #ifdef MBS_SUPPORT
6198 #endif
6202 /* This is written out as an if-else so we don't waste time
6203 testing `translate' inside the loop. */
6205 {
6206 do
6207 {
6209 #ifdef WCHAR
6211 {
6215 }
6216 else
6217 {
6220 }
6221 #else
6226 }
6228 }
6229 else
6230 {
6231 do
6232 {
6235 }
6237 }
6242 /* Match any character except possibly a newline or a null. */
6254 d++;
6260 {
6262 #ifdef WCHAR
6266 #define WORK_BUFFER_SIZE 128
6268 # ifdef _LIBC
6277 #ifdef WCHAR
6278 # ifdef _LIBC
6287 /* p points charset[6], so the address of the next instruction
6288 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6289 where l=length of char_classes, m=length of collating_symbol,
6290 n=equivalence_class, o=length of char_range,
6291 p'=length of character. */
6293 /* Update p to indicate the next instruction. */
6297 /* match with char_class? */
6299 {
6306 # ifdef _LIBC
6309 # else
6312 # endif
6313 }
6315 /* match with collating_symbol? */
6316 # ifdef _LIBC
6318 {
6323 workp++)
6324 {
6332 {
6333 /* Update d, however d will be incremented at
6334 char_set_matched:, we decrement d here. */
6337 }
6338 }
6339 }
6341 # endif
6342 /* If we can't look up collation data, we use wcscoll
6343 instead. */
6344 {
6346 {
6348 # ifdef _LIBC
6350 # else
6352 # endif
6354 /* If wcscoll(the collating symbol, whole string) > 0,
6355 any substring of the string never match with the
6356 collating symbol. */
6357 # ifdef _LIBC
6359 # else
6361 # endif
6362 {
6365 }
6367 /* First, we compare the collating symbol with
6368 the first character of the string.
6369 If it don't match, we add the next character to
6370 the compare buffer in turn. */
6372 {
6375 {
6380 }
6382 /* add next character to the compare buffer. */
6386 # ifdef _LIBC
6388 # else
6390 # endif
6395 /* (str_buf > workp) indicate (str_buf + X > workp),
6396 because for all X (str_buf + X > str_buf).
6397 So we don't need continue this loop. */
6400 /* Otherwise(str_buf < workp),
6401 (str_buf+next_character) may equals (workp).
6402 So we continue this loop. */
6403 }
6404 /* not matched */
6408 }
6409 }
6410 /* match with equivalence_class? */
6411 # ifdef _LIBC
6413 {
6415 /* Try to match the equivalence class against
6416 those known to the collate implementation. */
6425 /* This #include defines a local function! */
6426 # include <locale/weightwc.h>
6437 /* Write 1 collating element to str_buf, and
6438 get its index. */
6442 {
6445 {
6450 }
6454 }
6456 /* Update d, however d will be incremented at
6457 char_set_matched:, we decrement d here. */
6460 {
6463 else
6464 {
6467 }
6468 }
6473 workp++)
6474 {
6476 /* We already checked idx != 0 in regex_compile. */
6479 {
6487 }
6488 }
6489 /* not matched */
6492 }
6494 # endif
6495 /* If we can't look up collation data, we use wcscoll
6496 instead. */
6497 {
6499 {
6501 # ifdef _LIBC
6503 # else
6505 # endif
6507 /* If wcscoll(the collating symbol, whole string) > 0,
6508 any substring of the string never match with the
6509 collating symbol. */
6510 # ifdef _LIBC
6512 # else
6514 # endif
6515 {
6518 }
6520 /* First, we compare the equivalence class with
6521 the first character of the string.
6522 If it don't match, we add the next character to
6523 the compare buffer in turn. */
6525 {
6528 {
6533 }
6535 /* add next character to the compare buffer. */
6539 # ifdef _LIBC
6541 # else
6543 # endif
6549 /* (str_buf > workp) indicate (str_buf + X > workp),
6550 because for all X (str_buf + X > str_buf).
6551 So we don't need continue this loop. */
6554 /* Otherwise(str_buf < workp),
6555 (str_buf+next_character) may equals (workp).
6556 So we continue this loop. */
6557 }
6558 /* not matched */
6562 }
6563 }
6565 /* match with char_range? */
6566 # ifdef _LIBC
6568 {
6576 {
6579 /* We already compute the collation sequence value
6580 of the characters (or collating symbols). */
6586 }
6587 }
6588 else
6589 # endif
6590 {
6591 /* We set range_start_char at str_buf[0], range_end_char
6592 at str_buf[4], and compared char at str_buf[2]. */
6598 {
6601 /* match if (range_start_char <= c <= range_end_char). */
6603 /* If range_start(or end) < 0, we assume -range_start(end)
6604 is the offset of the collating symbol which is specified
6605 as the character of the range start(end). */
6607 /* range_start */
6610 else
6611 {
6614 }
6616 /* range_end */
6619 else
6620 {
6623 }
6625 # ifdef _LIBC
6628 # else
6631 # endif
6633 }
6634 }
6636 /* match with char? */
6643 char_set_matched:
6645 #else
6646 /* Cast to `unsigned' instead of `unsigned char' in case the
6647 bit list is a full 32 bytes long. */
6655 #undef WORK_BUFFER_SIZE
6658 d++;
6660 }
6663 /* The beginning of a group is represented by start_memory.
6664 The arguments are the register number in the next byte, and the
6665 number of groups inner to this one in the next. The text
6666 matched within the group is recorded (in the internal
6667 registers data structure) under the register number. */
6672 /* Find out if this group can match the empty string. */
6679 /* Save the position in the string where we were the last time
6680 we were at this open-group operator in case the group is
6681 operated upon by a repetition operator, e.g., with `(a*)*b'
6682 against `ab'; then we want to ignore where we are now in
6683 the string in case this attempt to match fails. */
6696 /* Clear this whenever we change the register activity status. */
6699 /* This is the new highest active register. */
6702 /* If nothing was active before, this is the new lowest active
6703 register. */
6707 /* Move past the register number and inner group count. */
6714 /* The stop_memory opcode represents the end of a group. Its
6715 arguments are the same as start_memory's: the register
6716 number, and the number of inner groups. */
6721 /* We need to save the string position the last time we were at
6722 this close-group operator in case the group is operated
6723 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6724 against `aba'; then we want to ignore where we are now in
6725 the string in case this attempt to match fails. */
6735 /* This register isn't active anymore. */
6738 /* Clear this whenever we change the register activity status. */
6741 /* If this was the only register active, nothing is active
6742 anymore. */
6744 {
6747 }
6748 else
6750 it isn't necessarily one less than now: consider
6751 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6752 new highest active register is 1. */
6755 r--;
6757 /* If we end up at register zero, that means that we saved
6758 the registers as the result of an `on_failure_jump', not
6759 a `start_memory', and we jumped to past the innermost
6760 `stop_memory'. For example, in ((.)*) we save
6761 registers 1 and 2 as a result of the *, but when we pop
6762 back to the second ), we are at the stop_memory 1.
6763 Thus, nothing is active. */
6765 {
6768 }
6769 else
6771 }
6773 /* If just failed to match something this time around with a
6774 group that's operated on by a repetition operator, try to
6775 force exit from the ``loop'', and restore the register
6776 information for this group that we had before trying this
6777 last match. */
6781 {
6787 {
6801 }
6804 /* If the next operation is a jump backwards in the pattern
6805 to an on_failure_jump right before the start_memory
6806 corresponding to this stop_memory, exit from the loop
6807 by forcing a failure after pushing on the stack the
6808 on_failure_jump's jump in the pattern, and d. */
6812 {
6813 /* If this group ever matched anything, then restore
6814 what its registers were before trying this last
6815 failed match, e.g., with `(a*)*b' against `ab' for
6816 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6817 against `aba' for regend[3].
6819 Also restore the registers for inner groups for,
6820 e.g., `((a*)(b*))*' against `aba' (register 3 would
6821 otherwise get trashed). */
6824 {
6829 /* Restore this and inner groups' (if any) registers. */
6831 r++)
6832 {
6835 /* xx why this test? */
6838 }
6839 }
6840 p1++;
6845 }
6846 }
6848 /* Move past the register number and the inner group count. */
6853 /* \<digit> has been turned into a `duplicate' command which is
6854 followed by the numeric value of <digit> as the register number. */
6856 {
6861 /* Can't back reference a group which we've never matched. */
6865 /* Where in input to try to start matching. */
6868 /* Where to stop matching; if both the place to start and
6869 the place to stop matching are in the same string, then
6870 set to the place to stop, otherwise, for now have to use
6871 the end of the first string. */
6877 {
6878 /* If necessary, advance to next segment in register
6879 contents. */
6881 {
6885 /* End of string1 => advance to string2. */
6888 }
6889 /* At end of register contents => success */
6892 /* If necessary, advance to next segment in data. */
6895 /* How many characters left in this segment to match. */
6898 /* Want how many consecutive characters we can match in
6899 one shot, so, if necessary, adjust the count. */
6903 /* Compare that many; failure if mismatch, else move
6904 past them. */
6911 /* Do this because we've match some characters. */
6913 }
6914 }
6918 /* begline matches the empty string at the beginning of the string
6919 (unless `not_bol' is set in `bufp'), and, if
6920 `newline_anchor' is set, after newlines. */
6925 {
6927 }
6929 {
6931 }
6932 /* In all other cases, we fail. */
6936 /* endline is the dual of begline. */
6941 {
6943 }
6945 /* We have to ``prefetch'' the next character. */
6948 {
6950 }
6954 /* Match at the very beginning of the data. */
6962 /* Match at the very end of the data. */
6970 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6971 pushes NULL as the value for the string on the stack. Then
6972 `pop_failure_point' will keep the current value for the
6973 string, instead of restoring it. To see why, consider
6974 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6975 then the . fails against the \n. But the next thing we want
6976 to do is match the \n against the \n; if we restored the
6977 string value, we would be back at the foo.
6979 Because this is used only in specific cases, we don't need to
6980 check all the things that `on_failure_jump' does, to make
6981 sure the right things get saved on the stack. Hence we don't
6982 share its code. The only reason to push anything on the
6983 stack at all is that otherwise we would have to change
6984 `anychar's code to do something besides goto fail in this
6985 case; that seems worse than this. */
6990 #ifdef _LIBC
6992 #else
6994 #endif
7000 /* Uses of on_failure_jump:
7002 Each alternative starts with an on_failure_jump that points
7003 to the beginning of the next alternative. Each alternative
7004 except the last ends with a jump that in effect jumps past
7005 the rest of the alternatives. (They really jump to the
7006 ending jump of the following alternative, because tensioning
7007 these jumps is a hassle.)
7009 Repeats start with an on_failure_jump that points past both
7010 the repetition text and either the following jump or
7011 pop_failure_jump back to this on_failure_jump. */
7013 on_failure:
7017 #ifdef _LIBC
7019 #else
7021 #endif
7023 /* If this on_failure_jump comes right before a group (i.e.,
7024 the original * applied to a group), save the information
7025 for that group and all inner ones, so that if we fail back
7026 to this point, the group's information will be correct.
7027 For example, in \(a*\)*\1, we need the preceding group,
7028 and in \(zz\(a*\)b*\)\2, we need the inner group. */
7030 /* We can't use `p' to check ahead because we push
7031 a failure point to `p + mcnt' after we do this. */
7034 /* We need to skip no_op's before we look for the
7035 start_memory in case this on_failure_jump is happening as
7036 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
7037 against aba. */
7039 p1++;
7042 {
7043 /* We have a new highest active register now. This will
7044 get reset at the start_memory we are about to get to,
7045 but we will have saved all the registers relevant to
7046 this repetition op, as described above. */
7050 }
7057 /* A smart repeat ends with `maybe_pop_jump'.
7058 We change it to either `pop_failure_jump' or `jump'. */
7062 {
7065 /* Compare the beginning of the repeat with what in the
7066 pattern follows its end. If we can establish that there
7067 is nothing that they would both match, i.e., that we
7068 would have to backtrack because of (as in, e.g., `a*a')
7069 then we can change to pop_failure_jump, because we'll
7070 never have to backtrack.
7072 This is not true in the case of alternatives: in
7073 `(a|ab)*' we do need to backtrack to the `ab' alternative
7074 (e.g., if the string was `ab'). But instead of trying to
7075 detect that here, the alternative has put on a dummy
7076 failure point which is what we will end up popping. */
7078 /* Skip over open/close-group commands.
7079 If what follows this loop is a ...+ construct,
7080 look at what begins its body, since we will have to
7081 match at least one of that. */
7083 {
7091 else
7093 }
7096 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7097 to the `maybe_finalize_jump' of this case. Examine what
7098 follows. */
7100 /* If we're at the end of the pattern, we can change. */
7102 {
7103 /* Consider what happens when matching ":\(.*\)"
7104 against ":/". I don't really understand this code
7105 yet. */
7107 pop_failure_jump;
7108 DEBUG_PRINT1
7110 }
7113 #ifdef MBS_SUPPORT
7115 #endif
7117 {
7118 register UCHAR_T c
7122 #ifdef MBS_SUPPORT
7124 #endif
7126 {
7128 pop_failure_jump;
7129 #ifdef WCHAR
7133 #else
7137 #endif
7138 }
7140 #ifndef WCHAR
7143 {
7150 /* `not' is equal to 1 if c would match, which means
7151 that we can't change to pop_failure_jump. */
7153 {
7156 }
7157 }
7159 }
7160 #ifndef WCHAR
7162 {
7163 /* We win if the first character of the loop is not part
7164 of the charset. */
7169 {
7172 }
7175 {
7177 /* We win if the charset_not inside the loop
7178 lists every character listed in the charset after. */
7186 {
7189 }
7190 }
7192 {
7194 /* We win if the charset inside the loop
7195 has no overlap with the one after the loop. */
7198 idx++)
7203 {
7206 }
7207 }
7208 }
7210 }
7213 {
7217 }
7218 /* Note fall through. */
7221 /* The end of a simple repeat has a pop_failure_jump back to
7222 its matching on_failure_jump, where the latter will push a
7223 failure point. The pop_failure_jump takes off failure
7224 points put on by this pop_failure_jump's matching
7225 on_failure_jump; we got through the pattern to here from the
7226 matching on_failure_jump, so didn't fail. */
7228 {
7229 /* We need to pass separate storage for the lowest and
7230 highest registers, even though we don't care about the
7231 actual values. Otherwise, we will restore only one
7232 register from the stack, since lowest will == highest in
7233 `pop_failure_point'. */
7242 }
7243 /* Note fall through. */
7245 unconditional_jump:
7246 #ifdef _LIBC
7248 #else
7250 #endif
7251 /* Note fall through. */
7253 /* Unconditionally jump (without popping any failure points). */
7258 #ifdef _LIBC
7260 #else
7262 #endif
7266 /* We need this opcode so we can detect where alternatives end
7267 in `group_match_null_string_p' et al. */
7273 /* Normally, the on_failure_jump pushes a failure point, which
7274 then gets popped at pop_failure_jump. We will end up at
7275 pop_failure_jump, also, and with a pattern of, say, `a+', we
7276 are skipping over the on_failure_jump, so we have to push
7277 something meaningless for pop_failure_jump to pop. */
7280 /* It doesn't matter what we push for the string here. What
7281 the code at `fail' tests is the value for the pattern. */
7286 /* At the end of an alternative, we need to push a dummy failure
7287 point in case we are followed by a `pop_failure_jump', because
7288 we don't want the failure point for the alternative to be
7289 popped. For example, matching `(a|ab)*' against `aab'
7290 requires that we match the `ab' alternative. */
7293 /* See comments just above at `dummy_failure_jump' about the
7294 two zeroes. */
7298 /* Have to succeed matching what follows at least n times.
7299 After that, handle like `on_failure_jump'. */
7305 /* Originally, this is how many times we HAVE to succeed. */
7307 {
7308 mcnt--;
7311 #ifdef _LIBC
7314 #else
7317 #endif
7318 }
7320 {
7321 #ifdef _LIBC
7324 #else
7329 #ifdef WCHAR
7331 #else
7336 }
7343 /* Originally, this is how many times we CAN jump. */
7345 {
7346 mcnt--;
7349 #ifdef _LIBC
7351 mcnt);
7352 #else
7354 mcnt);
7357 }
7358 /* If don't have to jump any more, skip over the rest of command. */
7359 else
7364 {
7370 #ifdef _LIBC
7372 #else
7374 #endif
7377 }
7379 #if 0
7380 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7381 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7382 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7383 macro and introducing temporary variables works around the bug. */
7396 #else
7398 {
7410 }
7413 {
7425 }
7426 #endif
7442 #ifdef emacs
7469 matchsyntax:
7471 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7472 d++;
7486 matchnotsyntax:
7488 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7489 d++;
7502 d++;
7511 d++;
7517 }
7521 /* We goto here if a matching operation fails. */
7522 fail:
7530 /* If this failure point is a dummy, try the next one. */
7534 /* If we failed to the end of the pattern, don't examine *p. */
7537 {
7540 /* If failed to a backwards jump that's part of a repetition
7541 loop, need to pop this failure point and use the next one. */
7543 {
7554 || (!is_a_jump_n
7560 }
7561 }
7565 }
7566 else
7577 \f
7578 /* Subroutine definitions for re_match_2. */
7581 /* We are passed P pointing to a register number after a start_memory.
7583 Return true if the pattern up to the corresponding stop_memory can
7584 match the empty string, and false otherwise.
7586 If we find the matching stop_memory, sets P to point to one past its number.
7587 Otherwise, sets P to an undefined byte less than or equal to END.
7589 We don't handle duplicates properly (yet). */
7591 static boolean
7595 {
7597 /* Point to after the args to the start_memory. */
7601 {
7602 /* Skip over opcodes that can match nothing, and return true or
7603 false, as appropriate, when we get to one that can't, or to the
7604 matching stop_memory. */
7607 {
7608 /* Could be either a loop or a series of alternatives. */
7610 p1++;
7613 /* If the next operation is not a jump backwards in the
7614 pattern. */
7617 {
7618 /* Go through the on_failure_jumps of the alternatives,
7619 seeing if any of the alternatives cannot match nothing.
7620 The last alternative starts with only a jump,
7621 whereas the rest start with on_failure_jump and end
7622 with a jump, e.g., here is the pattern for `a|b|c':
7624 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7625 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7626 /exactn/1/c
7628 So, we have to first go through the first (n-1)
7629 alternatives and then deal with the last one separately. */
7632 /* Deal with the first (n-1) alternatives, which start
7633 with an on_failure_jump (see above) that jumps to right
7634 past a jump_past_alt. */
7637 jump_past_alt)
7638 {
7639 /* `mcnt' holds how many bytes long the alternative
7640 is, including the ending `jump_past_alt' and
7641 its number. */
7645 reg_info))
7648 /* Move to right after this alternative, including the
7649 jump_past_alt. */
7652 /* Break if it's the beginning of an n-th alternative
7653 that doesn't begin with an on_failure_jump. */
7657 /* Still have to check that it's not an n-th
7658 alternative that starts with an on_failure_jump. */
7659 p1++;
7662 jump_past_alt)
7663 {
7664 /* Get to the beginning of the n-th alternative. */
7667 }
7668 }
7670 /* Deal with the last alternative: go back and get number
7671 of the `jump_past_alt' just before it. `mcnt' contains
7672 the length of the alternative. */
7692 }
7699 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7700 It expects P to be the first byte of a single alternative and END one
7701 byte past the last. The alternative can contain groups. */
7703 static boolean
7707 {
7712 {
7713 /* Skip over opcodes that can match nothing, and break when we get
7714 to one that can't. */
7717 {
7718 /* It's a loop. */
7720 p1++;
7728 }
7735 /* Deals with the ops common to group_match_null_string_p and
7736 alt_match_null_string_p.
7738 Sets P to one after the op and its arguments, if any. */
7740 static boolean
7744 {
7746 boolean ret;
7751 {
7761 #ifdef emacs
7765 #endif
7773 /* Have to set this here in case we're checking a group which
7774 contains a group and a back reference to it. */
7783 /* If this is an optimized succeed_n for zero times, make the jump. */
7788 else
7793 /* Get to the number of times to succeed. */
7798 {
7802 }
7803 else
7816 /* All other opcodes mean we cannot match the empty string. */
7818 }
7825 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7826 bytes; nonzero otherwise. */
7828 static int
7832 RE_TRANSLATE_TYPE translate;
7833 {
7837 {
7838 #ifdef WCHAR
7845 len--;
7846 }
7848 }
7849 \f
7853 /* Entry points for GNU code. */
7855 /* re_compile_pattern is the GNU regular expression compiler: it
7856 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7857 Returns 0 if the pattern was valid, otherwise an error string.
7859 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7860 are set in BUFP on entry.
7862 We call regex_compile to do the actual compilation. */
7869 {
7870 reg_errcode_t ret;
7872 /* GNU code is written to assume at least RE_NREGS registers will be set
7873 (and at least one extra will be -1). */
7876 /* And GNU code determines whether or not to get register information
7877 by passing null for the REGS argument to re_match, etc., not by
7878 setting no_sub. */
7881 /* Match anchors at newline. */
7884 # ifdef MBS_SUPPORT
7887 else
7888 # endif
7894 }
7895 #ifdef _LIBC
7897 #endif
7898 \f
7899 /* Entry points compatible with 4.2 BSD regex library. We don't define
7900 them unless specifically requested. */
7902 #if defined _REGEX_RE_COMP || defined _LIBC
7904 /* BSD has one and only one pattern buffer. */
7908 #ifdef _LIBC
7909 /* Make these definitions weak in libc, so POSIX programs can redefine
7910 these names if they don't use our functions, and still use
7911 regcomp/regexec below without link errors. */
7912 weak_function
7913 #endif
7916 {
7917 reg_errcode_t ret;
7920 {
7924 }
7927 {
7936 }
7938 /* Since `re_exec' always passes NULL for the `regs' argument, we
7939 don't need to initialize the pattern buffer fields which affect it. */
7941 /* Match anchors at newlines. */
7944 # ifdef MBS_SUPPORT
7947 else
7948 # endif
7954 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7956 }
7959 int
7960 #ifdef _LIBC
7961 weak_function
7962 #endif
7965 {
7967 return
7969 }
7972 \f
7973 /* POSIX.2 functions. Don't define these for Emacs. */
7975 #ifndef emacs
7977 /* regcomp takes a regular expression as a string and compiles it.
7979 PREG is a regex_t *. We do not expect any fields to be initialized,
7980 since POSIX says we shouldn't. Thus, we set
7982 `buffer' to the compiled pattern;
7983 `used' to the length of the compiled pattern;
7984 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7985 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7986 RE_SYNTAX_POSIX_BASIC;
7987 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7988 `fastmap' to an allocated space for the fastmap;
7989 `fastmap_accurate' to zero;
7990 `re_nsub' to the number of subexpressions in PATTERN.
7992 PATTERN is the address of the pattern string.
7994 CFLAGS is a series of bits which affect compilation.
7996 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
7997 use POSIX basic syntax.
7999 If REG_NEWLINE is set, then . and [^...] don't match newline.
8000 Also, regexec will try a match beginning after every newline.
8002 If REG_ICASE is set, then we considers upper- and lowercase
8003 versions of letters to be equivalent when matching.
8005 If REG_NOSUB is set, then when PREG is passed to regexec, that
8006 routine will report only success or failure, and nothing about the
8007 registers.
8009 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
8010 the return codes and their meanings.) */
8012 int
8017 {
8018 reg_errcode_t ret;
8019 reg_syntax_t syntax
8023 /* regex_compile will allocate the space for the compiled pattern. */
8028 /* Try to allocate space for the fastmap. */
8032 {
8035 preg->translate
8041 /* Map uppercase characters to corresponding lowercase ones. */
8044 }
8045 else
8048 /* If REG_NEWLINE is set, newlines are treated differently. */
8053 /* It also changes the matching behavior. */
8055 }
8056 else
8061 /* POSIX says a null character in the pattern terminates it, so we
8062 can use strlen here in compiling the pattern. */
8063 # ifdef MBS_SUPPORT
8066 else
8067 # endif
8070 /* POSIX doesn't distinguish between an unmatched open-group and an
8071 unmatched close-group: both are REG_EPAREN. */
8075 {
8076 /* Compute the fastmap now, since regexec cannot modify the pattern
8077 buffer. */
8079 {
8080 /* Some error occurred while computing the fastmap, just forget
8081 about it. */
8084 }
8085 }
8088 }
8089 #ifdef _LIBC
8091 #endif
8094 /* regexec searches for a given pattern, specified by PREG, in the
8095 string STRING.
8097 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
8098 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
8099 least NMATCH elements, and we set them to the offsets of the
8100 corresponding matched substrings.
8102 EFLAGS specifies `execution flags' which affect matching: if
8103 REG_NOTBOL is set, then ^ does not match at the beginning of the
8104 string; if REG_NOTEOL is set, then $ does not match at the end.
8106 We return 0 if we find a match and REG_NOMATCH if not. */
8108 int
8113 regmatch_t pmatch[];
8115 {
8118 regex_t private_preg;
8127 /* The user has told us exactly how many registers to return
8128 information about, via `nmatch'. We have to pass that on to the
8129 matching routines. */
8133 {
8139 }
8141 /* Perform the searching operation. */
8146 /* Copy the register information to the POSIX structure. */
8148 {
8150 {
8154 {
8157 }
8158 }
8160 /* If we needed the temporary register info, free the space now. */
8162 }
8164 /* We want zero return to mean success, unlike `re_search'. */
8166 }
8167 #ifdef _LIBC
8169 #endif
8172 /* Returns a message corresponding to an error code, ERRCODE, returned
8173 from either regcomp or regexec. We don't use PREG here. */
8175 size_t
8181 {
8188 /* Only error codes returned by the rest of the code should be passed
8189 to this routine. If we are given anything else, or if other regex
8190 code generates an invalid error code, then the program has a bug.
8191 Dump core so we can fix it. */
8199 {
8201 {
8202 #if defined HAVE_MEMPCPY || defined _LIBC
8204 #else
8207 #endif
8208 }
8209 else
8211 }
8214 }
8215 #ifdef _LIBC
8217 #endif
8220 /* Free dynamically allocated space used by PREG. */
8222 void
8225 {
8241 }
8242 #ifdef _LIBC
8244 #endif
8250 \f
8251 #undef STORE_NUMBER
8252 #undef STORE_NUMBER_AND_INCR
8253 #undef EXTRACT_NUMBER
8254 #undef EXTRACT_NUMBER_AND_INCR
8256 #undef DEBUG_PRINT_COMPILED_PATTERN
8257 #undef DEBUG_PRINT_DOUBLE_STRING
8259 #undef INIT_FAIL_STACK
8260 #undef RESET_FAIL_STACK
8261 #undef DOUBLE_FAIL_STACK
8262 #undef PUSH_PATTERN_OP
8263 #undef PUSH_FAILURE_POINTER
8264 #undef PUSH_FAILURE_INT
8265 #undef PUSH_FAILURE_ELT
8266 #undef POP_FAILURE_POINTER
8267 #undef POP_FAILURE_INT
8268 #undef POP_FAILURE_ELT
8269 #undef DEBUG_PUSH
8270 #undef DEBUG_POP
8271 #undef PUSH_FAILURE_POINT
8272 #undef POP_FAILURE_POINT
8274 #undef REG_UNSET_VALUE
8275 #undef REG_UNSET
8277 #undef PATFETCH
8278 #undef PATFETCH_RAW
8279 #undef PATUNFETCH
8280 #undef TRANSLATE
8282 #undef INIT_BUF_SIZE
8283 #undef GET_BUFFER_SPACE
8284 #undef BUF_PUSH
8285 #undef BUF_PUSH_2
8286 #undef BUF_PUSH_3
8287 #undef STORE_JUMP
8288 #undef STORE_JUMP2
8289 #undef INSERT_JUMP
8290 #undef INSERT_JUMP2
8291 #undef EXTEND_BUFFER
8292 #undef GET_UNSIGNED_NUMBER
8293 #undef FREE_STACK_RETURN
8295 # undef POINTER_TO_OFFSET
8296 # undef MATCHING_IN_FRST_STRING
8297 # undef PREFETCH
8298 # undef AT_STRINGS_BEG
8299 # undef AT_STRINGS_END
8300 # undef WORDCHAR_P
8301 # undef FREE_VAR
8302 # undef FREE_VARIABLES
8303 # undef NO_HIGHEST_ACTIVE_REG
8304 # undef NO_LOWEST_ACTIVE_REG
8306 # undef CHAR_T
8307 # undef UCHAR_T
8308 # undef COMPILED_BUFFER_VAR
8309 # undef OFFSET_ADDRESS_SIZE
8310 # undef CHAR_CLASS_SIZE
8311 # undef PREFIX
8312 # undef ARG_PREFIX
8313 # undef PUT_CHAR
8314 # undef BYTE
8315 # undef WCHAR
8317 # define DEFINED_ONCE