| blob | 17091ce9569fdc991efb5f5c939bd56ac089ecf6 |
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.)
6 Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
7 2002, 2005, 2010, 2013 Free Software Foundation, Inc.
8 This file is part of the GNU C Library.
10 The GNU C Library is free software; you can redistribute it and/or
11 modify it under the terms of the GNU Lesser General Public
12 License as published by the Free Software Foundation; either
13 version 2.1 of the License, or (at your option) any later version.
15 The GNU C Library is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 Lesser General Public License for more details.
20 You should have received a copy of the GNU Lesser General Public
21 License along with the GNU C Library; if not, write to the Free
22 Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
23 02110-1301 USA. */
25 /* This file has been modified for usage in libiberty. It includes "xregex.h"
26 instead of <regex.h>. The "xregex.h" header file renames all external
27 routines with an "x" prefix so they do not collide with the native regex
28 routines or with other components regex routines. */
29 /* AIX requires this to be the first thing in the file. */
30 #if defined _AIX && !defined __GNUC__ && !defined REGEX_MALLOC
31 #pragma alloca
32 #endif
34 #undef _GNU_SOURCE
35 #define _GNU_SOURCE
37 #ifndef INSIDE_RECURSION
38 # ifdef HAVE_CONFIG_H
39 # include <config.h>
40 # endif
41 #endif
43 #include <ansidecl.h>
45 #ifndef INSIDE_RECURSION
47 # if defined STDC_HEADERS && !defined emacs
48 # include <stddef.h>
49 # define PTR_INT_TYPE ptrdiff_t
50 # else
51 /* We need this for `regex.h', and perhaps for the Emacs include files. */
52 # include <sys/types.h>
53 # define PTR_INT_TYPE long
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 # define PREFIX(name) byte_##name
616 # define ARG_PREFIX(name) name
617 # define PUT_CHAR(c) putchar (c)
618 #else
619 # ifdef WCHAR
620 # define CHAR_T wchar_t
621 # define UCHAR_T wchar_t
622 # define COMPILED_BUFFER_VAR wc_buffer
624 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
625 # define PREFIX(name) wcs_##name
626 # define ARG_PREFIX(name) c##name
627 /* Should we use wide stream?? */
629 # define TRUE 1
630 # define FALSE 0
631 # else
632 # ifdef MBS_SUPPORT
633 # define WCHAR
634 # define INSIDE_RECURSION
636 # undef INSIDE_RECURSION
637 # endif
638 # define BYTE
639 # define INSIDE_RECURSION
641 # undef INSIDE_RECURSION
642 # endif
643 #endif
645 #ifdef INSIDE_RECURSION
646 /* Common operations on the compiled pattern. */
648 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
649 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
651 # ifdef WCHAR
652 # define STORE_NUMBER(destination, number) \
653 do { \
654 *(destination) = (UCHAR_T)(number); \
655 } while (0)
657 # define STORE_NUMBER(destination, number) \
658 do { \
659 (destination)[0] = (number) & 0377; \
660 (destination)[1] = (number) >> 8; \
661 } while (0)
664 /* Same as STORE_NUMBER, except increment DESTINATION to
665 the byte after where the number is stored. Therefore, DESTINATION
666 must be an lvalue. */
667 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
669 # define STORE_NUMBER_AND_INCR(destination, number) \
670 do { \
671 STORE_NUMBER (destination, number); \
672 (destination) += OFFSET_ADDRESS_SIZE; \
673 } while (0)
675 /* Put into DESTINATION a number stored in two contiguous bytes starting
676 at SOURCE. */
677 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
679 # ifdef WCHAR
680 # define EXTRACT_NUMBER(destination, source) \
681 do { \
682 (destination) = *(source); \
683 } while (0)
685 # define EXTRACT_NUMBER(destination, source) \
686 do { \
687 (destination) = *(source) & 0377; \
688 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
689 } while (0)
690 # endif
692 # ifdef DEBUG
694 static void
696 {
697 # ifdef WCHAR
703 # endif
704 }
707 # undef EXTRACT_NUMBER
708 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
713 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
714 SOURCE must be an lvalue. */
716 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
717 do { \
718 EXTRACT_NUMBER (destination, source); \
719 (source) += OFFSET_ADDRESS_SIZE; \
720 } while (0)
722 # ifdef DEBUG
725 static void
727 {
730 }
732 # ifndef EXTRACT_MACROS
733 # undef EXTRACT_NUMBER_AND_INCR
734 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
735 PREFIX(extract_number_and_incr) (&dest, &src)
740 \f
742 /* If DEBUG is defined, Regex prints many voluminous messages about what
743 it is doing (if the variable `debug' is nonzero). If linked with the
744 main program in `iregex.c', you can enter patterns and strings
745 interactively. And if linked with the main program in `main.c' and
746 the other test files, you can run the already-written tests. */
748 # ifdef DEBUG
750 # ifndef DEFINED_ONCE
752 /* We use standard I/O for debugging. */
753 # include <stdio.h>
755 /* It is useful to test things that ``must'' be true when debugging. */
756 # include <assert.h>
760 # define DEBUG_STATEMENT(e) e
761 # define DEBUG_PRINT1(x) if (debug) printf (x)
762 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
763 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
764 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
767 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
768 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
769 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
770 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
773 /* Print the fastmap in human-readable form. */
775 # ifndef DEFINED_ONCE
776 void
778 {
783 {
785 {
789 {
791 i++;
792 }
794 {
797 }
798 }
799 }
801 }
805 /* Print a compiled pattern string in human-readable form, starting at
806 the START pointer into it and ending just before the pointer END. */
808 void
810 {
817 {
820 }
822 /* Loop over pattern commands. */
824 {
825 # ifdef _LIBC
827 # else
829 # endif
832 {
840 do
841 {
844 }
848 # ifdef MBS_SUPPORT
852 do
853 {
855 }
880 {
881 # ifdef WCHAR
892 {
898 }
901 {
907 }
910 {
914 }
919 # else
931 {
932 /* Are we starting a range? */
934 {
937 }
938 /* Have we broken a range? */
940 {
943 }
949 }
958 }
971 # ifdef _LIBC
973 # else
975 # endif
980 # ifdef _LIBC
982 # else
985 # endif
990 # ifdef _LIBC
992 # else
994 # endif
1003 # ifdef _LIBC
1005 # else
1007 # endif
1012 # ifdef _LIBC
1014 # else
1016 # endif
1021 # ifdef _LIBC
1023 # else
1025 # endif
1030 # ifdef _LIBC
1032 # else
1034 # endif
1041 # ifdef _LIBC
1043 # else
1046 # endif
1060 # ifdef _LIBC
1062 # else
1065 # endif
1084 # ifdef emacs
1128 }
1131 }
1133 # ifdef _LIBC
1135 # else
1137 # endif
1138 }
1141 void
1143 {
1152 {
1155 }
1157 # ifdef _LIBC
1159 # else
1161 # endif
1169 /* Perhaps we should print the translate table? */
1170 }
1173 void
1176 {
1181 else
1182 {
1186 {
1191 }
1195 {
1198 {
1201 }
1202 }
1203 }
1204 }
1206 # ifndef DEFINED_ONCE
1207 void
1209 {
1211 }
1212 # endif
1216 # ifndef DEFINED_ONCE
1217 # undef assert
1218 # define assert(e)
1220 # define DEBUG_STATEMENT(e)
1221 # define DEBUG_PRINT1(x)
1222 # define DEBUG_PRINT2(x1, x2)
1223 # define DEBUG_PRINT3(x1, x2, x3)
1224 # define DEBUG_PRINT4(x1, x2, x3, x4)
1226 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1227 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1231 \f
1233 # ifdef WCHAR
1234 /* This convert a multibyte string to a wide character string.
1235 And write their correspondances to offset_buffer(see below)
1236 and write whether each wchar_t is binary data to is_binary.
1237 This assume invalid multibyte sequences as binary data.
1238 We assume offset_buffer and is_binary is already allocated
1239 enough space. */
1244 static size_t
1247 /* It hold correspondances between src(char string) and
1248 dest(wchar_t string) for optimization.
1249 e.g. src = "xxxyzz"
1250 dest = {'X', 'Y', 'Z'}
1251 (each "xxx", "y" and "zz" represent one multibyte character
1252 corresponding to 'X', 'Y' and 'Z'.)
1253 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1254 = {0, 3, 4, 6}
1255 */
1256 {
1266 /* Initialize the conversion state. */
1272 {
1273 #ifdef _LIBC
1275 #else
1277 #endif
1280 /* failed to convert. maybe src contains binary data.
1281 So we consume 1 byte manualy. */
1282 {
1286 }
1287 else
1289 /* In sjis encoding, we use yen sign as escape character in
1290 place of reverse solidus. So we convert 0x5c(yen sign in
1291 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1292 solidus in UCS2). */
1297 }
1299 /* Fill remain of the buffer with sentinel. */
1304 }
1310 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1311 also be assigned to arbitrarily: each pattern buffer stores its own
1312 syntax, so it can be changed between regex compilations. */
1313 /* This has no initializer because initialized variables in Emacs
1314 become read-only after dumping. */
1315 reg_syntax_t re_syntax_options;
1318 /* Specify the precise syntax of regexps for compilation. This provides
1319 for compatibility for various utilities which historically have
1320 different, incompatible syntaxes.
1322 The argument SYNTAX is a bit mask comprised of the various bits
1323 defined in regex.h. We return the old syntax. */
1325 reg_syntax_t
1327 {
1331 # ifdef DEBUG
1338 }
1339 # ifdef _LIBC
1341 # endif
1342 \f
1343 /* This table gives an error message for each of the error codes listed
1344 in regex.h. Obviously the order here has to be same as there.
1345 POSIX doesn't require that we do anything for REG_NOERROR,
1346 but why not be nice? */
1349 {
1367 };
1368 \f
1371 #ifndef DEFINED_ONCE
1372 /* Avoiding alloca during matching, to placate r_alloc. */
1374 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1375 searching and matching functions should not call alloca. On some
1376 systems, alloca is implemented in terms of malloc, and if we're
1377 using the relocating allocator routines, then malloc could cause a
1378 relocation, which might (if the strings being searched are in the
1379 ralloc heap) shift the data out from underneath the regexp
1380 routines.
1382 Here's another reason to avoid allocation: Emacs
1383 processes input from X in a signal handler; processing X input may
1384 call malloc; if input arrives while a matching routine is calling
1385 malloc, then we're scrod. But Emacs can't just block input while
1386 calling matching routines; then we don't notice interrupts when
1387 they come in. So, Emacs blocks input around all regexp calls
1388 except the matching calls, which it leaves unprotected, in the
1389 faith that they will not malloc. */
1391 /* Normally, this is fine. */
1392 # define MATCH_MAY_ALLOCATE
1394 /* When using GNU C, we are not REALLY using the C alloca, no matter
1395 what config.h may say. So don't take precautions for it. */
1396 # ifdef __GNUC__
1397 # undef C_ALLOCA
1398 # endif
1400 /* The match routines may not allocate if (1) they would do it with malloc
1401 and (2) it's not safe for them to use malloc.
1402 Note that if REL_ALLOC is defined, matching would not use malloc for the
1403 failure stack, but we would still use it for the register vectors;
1404 so REL_ALLOC should not affect this. */
1405 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1406 # undef MATCH_MAY_ALLOCATE
1407 # endif
1409 \f
1410 #ifdef INSIDE_RECURSION
1411 /* Failure stack declarations and macros; both re_compile_fastmap and
1412 re_match_2 use a failure stack. These have to be macros because of
1413 REGEX_ALLOCATE_STACK. */
1416 /* Number of failure points for which to initially allocate space
1417 when matching. If this number is exceeded, we allocate more
1418 space, so it is not a hard limit. */
1419 # ifndef INIT_FAILURE_ALLOC
1420 # define INIT_FAILURE_ALLOC 5
1421 # endif
1423 /* Roughly the maximum number of failure points on the stack. Would be
1424 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1425 This is a variable only so users of regex can assign to it; we never
1426 change it ourselves. */
1428 # ifdef INT_IS_16BIT
1430 # ifndef DEFINED_ONCE
1431 # if defined MATCH_MAY_ALLOCATE
1432 /* 4400 was enough to cause a crash on Alpha OSF/1,
1433 whose default stack limit is 2mb. */
1435 # else
1437 # endif
1438 # endif
1441 {
1444 };
1449 {
1457 # ifndef DEFINED_ONCE
1458 # if defined MATCH_MAY_ALLOCATE
1459 /* 4400 was enough to cause a crash on Alpha OSF/1,
1460 whose default stack limit is 2mb. */
1462 # else
1464 # endif
1465 # endif
1468 {
1471 };
1476 {
1484 # ifndef DEFINED_ONCE
1485 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1486 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1487 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1488 # endif
1491 /* Define macros to initialize and free the failure stack.
1492 Do `return -2' if the alloc fails. */
1494 # ifdef MATCH_MAY_ALLOCATE
1495 # define INIT_FAIL_STACK() \
1496 do { \
1497 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1498 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1499 \
1500 if (fail_stack.stack == NULL) \
1501 return -2; \
1502 \
1503 fail_stack.size = INIT_FAILURE_ALLOC; \
1504 fail_stack.avail = 0; \
1505 } while (0)
1507 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1508 # else
1509 # define INIT_FAIL_STACK() \
1510 do { \
1511 fail_stack.avail = 0; \
1512 } while (0)
1514 # define RESET_FAIL_STACK()
1515 # endif
1518 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1520 Return 1 if succeeds, and 0 if either ran out of memory
1521 allocating space for it or it was already too large.
1523 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1525 # define DOUBLE_FAIL_STACK(fail_stack) \
1526 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1527 ? 0 \
1528 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1529 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1530 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1531 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1532 \
1533 (fail_stack).stack == NULL \
1534 ? 0 \
1535 : ((fail_stack).size <<= 1, \
1536 1)))
1539 /* Push pointer POINTER on FAIL_STACK.
1540 Return 1 if was able to do so and 0 if ran out of memory allocating
1541 space to do so. */
1542 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1543 ((FAIL_STACK_FULL () \
1544 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1545 ? 0 \
1546 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1547 1))
1549 /* Push a pointer value onto the failure stack.
1550 Assumes the variable `fail_stack'. Probably should only
1551 be called from within `PUSH_FAILURE_POINT'. */
1552 # define PUSH_FAILURE_POINTER(item) \
1553 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1555 /* This pushes an integer-valued item onto the failure stack.
1556 Assumes the variable `fail_stack'. Probably should only
1557 be called from within `PUSH_FAILURE_POINT'. */
1558 # define PUSH_FAILURE_INT(item) \
1559 fail_stack.stack[fail_stack.avail++].integer = (item)
1561 /* Push a fail_stack_elt_t value onto the failure stack.
1562 Assumes the variable `fail_stack'. Probably should only
1563 be called from within `PUSH_FAILURE_POINT'. */
1564 # define PUSH_FAILURE_ELT(item) \
1565 fail_stack.stack[fail_stack.avail++] = (item)
1567 /* These three POP... operations complement the three PUSH... operations.
1568 All assume that `fail_stack' is nonempty. */
1569 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1570 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1571 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1573 /* Used to omit pushing failure point id's when we're not debugging. */
1574 # ifdef DEBUG
1575 # define DEBUG_PUSH PUSH_FAILURE_INT
1576 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1577 # else
1578 # define DEBUG_PUSH(item)
1579 # define DEBUG_POP(item_addr)
1580 # endif
1583 /* Push the information about the state we will need
1584 if we ever fail back to it.
1586 Requires variables fail_stack, regstart, regend, reg_info, and
1587 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1588 be declared.
1590 Does `return FAILURE_CODE' if runs out of memory. */
1592 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1593 do { \
1594 char *destination; \
1595 /* Must be int, so when we don't save any registers, the arithmetic \
1597 /* Can't be int, since there is not a shred of a guarantee that int \
1598 is wide enough to hold a value of something to which pointer can \
1600 active_reg_t this_reg; \
1601 \
1602 DEBUG_STATEMENT (failure_id++); \
1603 DEBUG_STATEMENT (nfailure_points_pushed++); \
1607 \
1610 \
1612 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1613 { \
1614 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1615 return failure_code; \
1616 \
1618 (fail_stack).size); \
1620 } \
1621 \
1624 \
1625 if (1) \
1626 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1627 this_reg++) \
1628 { \
1630 DEBUG_STATEMENT (num_regs_pushed++); \
1631 \
1633 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1634 \
1636 PUSH_FAILURE_POINTER (regend[this_reg]); \
1637 \
1639 reg_info[this_reg].word.pointer); \
1641 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1644 MATCHED_SOMETHING (reg_info[this_reg])); \
1646 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1648 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1649 } \
1650 \
1652 PUSH_FAILURE_INT (lowest_active_reg); \
1653 \
1655 PUSH_FAILURE_INT (highest_active_reg); \
1656 \
1658 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1659 PUSH_FAILURE_POINTER (pattern_place); \
1660 \
1662 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1663 size2); \
1665 PUSH_FAILURE_POINTER (string_place); \
1666 \
1668 DEBUG_PUSH (failure_id); \
1669 } while (0)
1671 # ifndef DEFINED_ONCE
1672 /* This is the number of items that are pushed and popped on the stack
1673 for each register. */
1674 # define NUM_REG_ITEMS 3
1676 /* Individual items aside from the registers. */
1677 # ifdef DEBUG
1679 # else
1680 # define NUM_NONREG_ITEMS 4
1681 # endif
1683 /* We push at most this many items on the stack. */
1684 /* We used to use (num_regs - 1), which is the number of registers
1685 this regexp will save; but that was changed to 5
1686 to avoid stack overflow for a regexp with lots of parens. */
1687 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1689 /* We actually push this many items. */
1690 # define NUM_FAILURE_ITEMS \
1691 (((0 \
1692 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1693 * NUM_REG_ITEMS) \
1694 + NUM_NONREG_ITEMS)
1696 /* How many items can still be added to the stack without overflowing it. */
1697 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1701 /* Pops what PUSH_FAIL_STACK pushes.
1703 We restore into the parameters, all of which should be lvalues:
1704 STR -- the saved data position.
1705 PAT -- the saved pattern position.
1706 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1707 REGSTART, REGEND -- arrays of string positions.
1708 REG_INFO -- array of information about each subexpression.
1710 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1711 `pend', `string1', `size1', `string2', and `size2'. */
1712 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1713 { \
1714 DEBUG_STATEMENT (unsigned failure_id;) \
1715 active_reg_t this_reg; \
1716 const UCHAR_T *string_temp; \
1717 \
1718 assert (!FAIL_STACK_EMPTY ()); \
1719 \
1724 \
1725 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1726 \
1727 DEBUG_POP (&failure_id); \
1729 \
1730 /* If the saved string location is NULL, it came from an \
1731 on_failure_keep_string_jump opcode, and we want to throw away the \
1733 string_temp = POP_FAILURE_POINTER (); \
1734 if (string_temp != NULL) \
1735 str = (const CHAR_T *) string_temp; \
1736 \
1738 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1740 \
1741 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1743 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1744 \
1746 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1748 \
1749 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1751 \
1752 if (1) \
1753 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1754 { \
1756 \
1757 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1759 reg_info[this_reg].word.pointer); \
1760 \
1761 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1763 \
1764 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1766 } \
1767 else \
1768 { \
1769 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1770 { \
1771 reg_info[this_reg].word.integer = 0; \
1772 regend[this_reg] = 0; \
1773 regstart[this_reg] = 0; \
1774 } \
1775 highest_active_reg = high_reg; \
1776 } \
1777 \
1778 set_regs_matched_done = 0; \
1779 DEBUG_STATEMENT (nfailure_points_popped++); \
1781 \f
1782 /* Structure for per-register (a.k.a. per-group) information.
1783 Other register information, such as the
1784 starting and ending positions (which are addresses), and the list of
1785 inner groups (which is a bits list) are maintained in separate
1786 variables.
1788 We are making a (strictly speaking) nonportable assumption here: that
1789 the compiler will pack our bit fields into something that fits into
1790 the type of `word', i.e., is something that fits into one item on the
1791 failure stack. */
1794 /* Declarations and macros for re_match_2. */
1797 {
1799 struct
1800 {
1801 /* This field is one if this group can match the empty string,
1802 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1803 # define MATCH_NULL_UNSET_VALUE 3
1811 # ifndef DEFINED_ONCE
1812 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1813 # define IS_ACTIVE(R) ((R).bits.is_active)
1814 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1815 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1818 /* Call this when have matched a real character; it sets `matched' flags
1819 for the subexpressions which we are currently inside. Also records
1820 that those subexprs have matched. */
1821 # define SET_REGS_MATCHED() \
1822 do \
1823 { \
1824 if (!set_regs_matched_done) \
1825 { \
1826 active_reg_t r; \
1827 set_regs_matched_done = 1; \
1828 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1829 { \
1830 MATCHED_SOMETHING (reg_info[r]) \
1831 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1832 = 1; \
1833 } \
1834 } \
1835 } \
1836 while (0)
1839 /* Registers are set to a sentinel when they haven't yet matched. */
1841 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1842 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1844 /* Subroutine declarations and macros for regex_compile. */
1854 reg_syntax_t syntax);
1857 reg_syntax_t syntax);
1858 # ifdef WCHAR
1863 reg_syntax_t syntax,
1872 reg_syntax_t syntax,
1876 /* Fetch the next character in the uncompiled pattern---translating it
1877 if necessary. Also cast from a signed character in the constant
1878 string passed to us by the user to an unsigned char that we can use
1879 as an array index (in, e.g., `translate'). */
1880 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1881 because it is impossible to allocate 4GB array for some encodings
1882 which have 4 byte character_set like UCS4. */
1883 # ifndef PATFETCH
1884 # ifdef WCHAR
1885 # define PATFETCH(c) \
1886 do {if (p == pend) return REG_EEND; \
1887 c = (UCHAR_T) *p++; \
1888 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1889 } while (0)
1891 # define PATFETCH(c) \
1892 do {if (p == pend) return REG_EEND; \
1893 c = (unsigned char) *p++; \
1894 if (translate) c = (unsigned char) translate[c]; \
1895 } while (0)
1897 # endif
1899 /* Fetch the next character in the uncompiled pattern, with no
1900 translation. */
1901 # define PATFETCH_RAW(c) \
1902 do {if (p == pend) return REG_EEND; \
1903 c = (UCHAR_T) *p++; \
1904 } while (0)
1906 /* Go backwards one character in the pattern. */
1907 # define PATUNFETCH p--
1910 /* If `translate' is non-null, return translate[D], else just D. We
1911 cast the subscript to translate because some data is declared as
1912 `char *', to avoid warnings when a string constant is passed. But
1913 when we use a character as a subscript we must make it unsigned. */
1914 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1915 because it is impossible to allocate 4GB array for some encodings
1916 which have 4 byte character_set like UCS4. */
1918 # ifndef TRANSLATE
1919 # ifdef WCHAR
1920 # define TRANSLATE(d) \
1921 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1922 ? (char) translate[(unsigned char) (d)] : (d))
1924 # define TRANSLATE(d) \
1925 (translate ? (char) translate[(unsigned char) (d)] : (char) (d))
1927 # endif
1930 /* Macros for outputting the compiled pattern into `buffer'. */
1932 /* If the buffer isn't allocated when it comes in, use this. */
1933 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1935 /* Make sure we have at least N more bytes of space in buffer. */
1936 # ifdef WCHAR
1937 # define GET_BUFFER_SPACE(n) \
1938 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
1939 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
1940 EXTEND_BUFFER ()
1942 # define GET_BUFFER_SPACE(n) \
1943 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1944 EXTEND_BUFFER ()
1947 /* Make sure we have one more byte of buffer space and then add C to it. */
1948 # define BUF_PUSH(c) \
1949 do { \
1950 GET_BUFFER_SPACE (1); \
1951 *b++ = (UCHAR_T) (c); \
1952 } while (0)
1955 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1956 # define BUF_PUSH_2(c1, c2) \
1957 do { \
1958 GET_BUFFER_SPACE (2); \
1959 *b++ = (UCHAR_T) (c1); \
1960 *b++ = (UCHAR_T) (c2); \
1961 } while (0)
1964 /* As with BUF_PUSH_2, except for three bytes. */
1965 # define BUF_PUSH_3(c1, c2, c3) \
1966 do { \
1967 GET_BUFFER_SPACE (3); \
1968 *b++ = (UCHAR_T) (c1); \
1969 *b++ = (UCHAR_T) (c2); \
1970 *b++ = (UCHAR_T) (c3); \
1971 } while (0)
1973 /* Store a jump with opcode OP at LOC to location TO. We store a
1974 relative address offset by the three bytes the jump itself occupies. */
1975 # define STORE_JUMP(op, loc, to) \
1976 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
1978 /* Likewise, for a two-argument jump. */
1979 # define STORE_JUMP2(op, loc, to, arg) \
1980 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
1982 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1983 # define INSERT_JUMP(op, loc, to) \
1984 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
1986 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1987 # define INSERT_JUMP2(op, loc, to, arg) \
1988 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
1989 arg, b)
1991 /* This is not an arbitrary limit: the arguments which represent offsets
1992 into the pattern are two bytes long. So if 2^16 bytes turns out to
1993 be too small, many things would have to change. */
1994 /* Any other compiler which, like MSC, has allocation limit below 2^16
1995 bytes will have to use approach similar to what was done below for
1996 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1997 reallocating to 0 bytes. Such thing is not going to work too well.
1998 You have been warned!! */
1999 # ifndef DEFINED_ONCE
2000 # if defined _MSC_VER && !defined WIN32
2001 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2002 The REALLOC define eliminates a flurry of conversion warnings,
2003 but is not required. */
2004 # define MAX_BUF_SIZE 65500L
2005 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2006 # else
2007 # define MAX_BUF_SIZE (1L << 16)
2008 # define REALLOC(p,s) realloc ((p), (s))
2009 # endif
2011 /* Extend the buffer by twice its current size via realloc and
2012 reset the pointers that pointed into the old block to point to the
2013 correct places in the new one. If extending the buffer results in it
2014 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2015 # if __BOUNDED_POINTERS__
2016 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2017 # define MOVE_BUFFER_POINTER(P) \
2018 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2019 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2020 else \
2021 { \
2022 SET_HIGH_BOUND (b); \
2023 SET_HIGH_BOUND (begalt); \
2024 if (fixup_alt_jump) \
2025 SET_HIGH_BOUND (fixup_alt_jump); \
2026 if (laststart) \
2027 SET_HIGH_BOUND (laststart); \
2028 if (pending_exact) \
2029 SET_HIGH_BOUND (pending_exact); \
2030 }
2031 # else
2032 # define MOVE_BUFFER_POINTER(P) (P) += incr
2033 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2034 # endif
2037 # ifdef WCHAR
2038 # define EXTEND_BUFFER() \
2039 do { \
2040 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2041 int wchar_count; \
2042 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2043 return REG_ESIZE; \
2044 bufp->allocated <<= 1; \
2045 if (bufp->allocated > MAX_BUF_SIZE) \
2046 bufp->allocated = MAX_BUF_SIZE; \
2048 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2049 if (wchar_count == 0) wchar_count = 1; \
2051 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2052 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2053 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2054 if (COMPILED_BUFFER_VAR == NULL) \
2055 return REG_ESPACE; \
2057 if (old_buffer != COMPILED_BUFFER_VAR) \
2058 { \
2059 PTR_INT_TYPE incr = COMPILED_BUFFER_VAR - old_buffer; \
2060 MOVE_BUFFER_POINTER (b); \
2061 MOVE_BUFFER_POINTER (begalt); \
2062 if (fixup_alt_jump) \
2063 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2064 if (laststart) \
2065 MOVE_BUFFER_POINTER (laststart); \
2066 if (pending_exact) \
2067 MOVE_BUFFER_POINTER (pending_exact); \
2068 } \
2069 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2070 } while (0)
2072 # define EXTEND_BUFFER() \
2073 do { \
2074 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2075 if (bufp->allocated == MAX_BUF_SIZE) \
2076 return REG_ESIZE; \
2077 bufp->allocated <<= 1; \
2078 if (bufp->allocated > MAX_BUF_SIZE) \
2079 bufp->allocated = MAX_BUF_SIZE; \
2080 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2081 bufp->allocated); \
2082 if (COMPILED_BUFFER_VAR == NULL) \
2083 return REG_ESPACE; \
2085 if (old_buffer != COMPILED_BUFFER_VAR) \
2086 { \
2087 PTR_INT_TYPE 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)
2101 # ifndef DEFINED_ONCE
2102 /* Since we have one byte reserved for the register number argument to
2103 {start,stop}_memory, the maximum number of groups we can report
2104 things about is what fits in that byte. */
2105 # define MAX_REGNUM 255
2107 /* But patterns can have more than `MAX_REGNUM' registers. We just
2108 ignore the excess. */
2112 /* Macros for the compile stack. */
2114 /* Since offsets can go either forwards or backwards, this type needs to
2115 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2116 /* int may be not enough when sizeof(int) == 2. */
2120 {
2121 pattern_offset_t begalt_offset;
2122 pattern_offset_t fixup_alt_jump;
2123 pattern_offset_t inner_group_offset;
2124 pattern_offset_t laststart_offset;
2125 regnum_t regnum;
2130 {
2137 # define INIT_COMPILE_STACK_SIZE 32
2139 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2140 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2142 /* The next available element. */
2143 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2147 /* Set the bit for character C in a list. */
2148 # ifndef DEFINED_ONCE
2149 # define SET_LIST_BIT(c) \
2150 (b[((unsigned char) (c)) / BYTEWIDTH] \
2151 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2154 /* Get the next unsigned number in the uncompiled pattern. */
2155 # define GET_UNSIGNED_NUMBER(num) \
2156 { \
2157 while (p != pend) \
2158 { \
2159 PATFETCH (c); \
2161 break; \
2162 if (num <= RE_DUP_MAX) \
2163 { \
2164 if (num < 0) \
2165 num = 0; \
2167 } \
2168 } \
2169 }
2171 # ifndef DEFINED_ONCE
2172 # if defined _LIBC || WIDE_CHAR_SUPPORT
2173 /* The GNU C library provides support for user-defined character classes
2174 and the functions from ISO C amendement 1. */
2175 # ifdef CHARCLASS_NAME_MAX
2176 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2177 # else
2178 /* This shouldn't happen but some implementation might still have this
2179 problem. Use a reasonable default value. */
2180 # define CHAR_CLASS_MAX_LENGTH 256
2181 # endif
2183 # ifdef _LIBC
2184 # define IS_CHAR_CLASS(string) __wctype (string)
2185 # else
2186 # define IS_CHAR_CLASS(string) wctype (string)
2187 # endif
2188 # else
2191 # define IS_CHAR_CLASS(string) \
2198 # endif
2200 \f
2201 # ifndef MATCH_MAY_ALLOCATE
2203 /* If we cannot allocate large objects within re_match_2_internal,
2204 we make the fail stack and register vectors global.
2205 The fail stack, we grow to the maximum size when a regexp
2206 is compiled.
2207 The register vectors, we adjust in size each time we
2208 compile a regexp, according to the number of registers it needs. */
2212 /* Size with which the following vectors are currently allocated.
2213 That is so we can make them bigger as needed,
2214 but never make them smaller. */
2215 # ifdef DEFINED_ONCE
2227 /* Make the register vectors big enough for NUM_REGS registers,
2228 but don't make them smaller. */
2230 static void
2232 {
2234 {
2246 }
2247 }
2250 \f
2251 # ifndef DEFINED_ONCE
2253 regnum_t regnum);
2256 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2257 Returns one of error codes defined in `regex.h', or zero for success.
2259 Assumes the `allocated' (and perhaps `buffer') and `translate'
2260 fields are set in BUFP on entry.
2262 If it succeeds, results are put in BUFP (if it returns an error, the
2263 contents of BUFP are undefined):
2264 `buffer' is the compiled pattern;
2265 `syntax' is set to SYNTAX;
2266 `used' is set to the length of the compiled pattern;
2267 `fastmap_accurate' is zero;
2268 `re_nsub' is the number of subexpressions in PATTERN;
2269 `not_bol' and `not_eol' are zero;
2271 The `fastmap' and `newline_anchor' fields are neither
2272 examined nor set. */
2274 /* Return, freeing storage we allocated. */
2275 # ifdef WCHAR
2276 # define FREE_STACK_RETURN(value) \
2277 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2278 # else
2279 # define FREE_STACK_RETURN(value) \
2280 return (free (compile_stack.stack), value)
2283 static reg_errcode_t
2287 {
2288 /* We fetch characters from PATTERN here. Even though PATTERN is
2289 `char *' (i.e., signed), we declare these variables as unsigned, so
2290 they can be reliably used as array indices. */
2293 #ifdef WCHAR
2294 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2297 /* offset buffer for optimization. See convert_mbs_to_wc. */
2299 /* It hold whether each wchar_t is binary data or not. */
2301 /* A flag whether exactn is handling binary data or not. */
2305 /* A random temporary spot in PATTERN. */
2308 /* Points to the end of the buffer, where we should append. */
2311 /* Keeps track of unclosed groups. */
2312 compile_stack_type compile_stack;
2314 /* Points to the current (ending) position in the pattern. */
2315 #ifdef WCHAR
2323 /* How to translate the characters in the pattern. */
2326 /* Address of the count-byte of the most recently inserted `exactn'
2327 command. This makes it possible to tell if a new exact-match
2328 character can be added to that command or if the character requires
2329 a new `exactn' command. */
2332 /* Address of start of the most recently finished expression.
2333 This tells, e.g., postfix * where to find the start of its
2334 operand. Reset at the beginning of groups and alternatives. */
2337 /* Address of beginning of regexp, or inside of last group. */
2340 /* Address of the place where a forward jump should go to the end of
2341 the containing expression. Each alternative of an `or' -- except the
2342 last -- ends with a forward jump of this sort. */
2345 /* Counts open-groups as they are encountered. Remembered for the
2346 matching close-group on the compile stack, so the same register
2347 number is put in the stop_memory as the start_memory. */
2350 #ifdef WCHAR
2351 /* Initialize the wchar_t PATTERN and offset_buffer. */
2356 {
2361 }
2366 {
2371 }
2372 #endif
2374 #ifdef DEBUG
2377 {
2383 }
2386 /* Initialize the compile stack. */
2389 {
2390 #ifdef WCHAR
2394 #endif
2396 }
2401 /* Initialize the pattern buffer. */
2406 /* Set `used' to zero, so that if we return an error, the pattern
2407 printer (for debugging) will think there's no pattern. We reset it
2408 at the end. */
2411 /* Always count groups, whether or not bufp->no_sub is set. */
2414 #if !defined emacs && !defined SYNTAX_TABLE
2415 /* Initialize the syntax table. */
2417 #endif
2420 {
2423 enough space. This loses if buffer's address is bogus, but
2424 that is the user's responsibility. */
2425 #ifdef WCHAR
2426 /* Free bufp->buffer and allocate an array for wchar_t pattern
2427 buffer. */
2430 UCHAR_T);
2431 #else
2434 }
2435 else
2438 UCHAR_T);
2439 }
2442 #ifdef WCHAR
2446 }
2447 #ifdef WCHAR
2448 else
2450 #endif
2454 /* Loop through the uncompiled pattern until we're at the end. */
2456 {
2460 {
2462 {
2465 /* If context independent, it's an operator. */
2467 /* Otherwise, depends on what's come before. */
2470 else
2472 }
2477 {
2479 p == pend
2480 /* If context independent, it's an operator. */
2482 /* Otherwise, depends on what's next. */
2485 else
2487 }
2496 handle_plus:
2498 /* If there is no previous pattern... */
2500 {
2505 }
2507 {
2508 /* Are we optimizing this jump? */
2511 /* 1 means zero (many) matches is allowed. */
2514 /* If there is a sequence of repetition chars, collapse it
2515 down to just one (the right one). We can't combine
2516 interval operators with these because of, e.g., `a{2}*',
2517 which should only match an even number of `a's. */
2520 {
2531 ;
2534 {
2539 {
2540 PATUNFETCH;
2541 PATUNFETCH;
2543 }
2546 }
2547 else
2548 {
2549 PATUNFETCH;
2551 }
2553 /* If we get here, we found another repeat character. */
2554 }
2556 /* Star, etc. applied to an empty pattern is equivalent
2557 to an empty pattern. */
2561 /* Now we know whether or not zero matches is allowed
2562 and also whether or not two or more matches is allowed. */
2565 end a backward relative jump from `b' to before the next
2566 jump we're going to put in below (which jumps from
2567 laststart to after this jump).
2569 But if we are at the `*' in the exact sequence `.*\n',
2570 insert an unconditional jump backwards to the .,
2571 instead of the beginning of the loop. This way we only
2572 push a failure point once, instead of every time
2573 through the loop. */
2576 /* Allocate the space for the jump. */
2579 /* We know we are not at the first character of the pattern,
2580 because laststart was nonzero. And we've already
2581 incremented `p', by the way, to be the character after
2582 the `*'. Do we have to do something analogous here
2583 for null bytes, because of RE_DOT_NOT_NULL? */
2585 && zero_times_ok
2591 }
2592 else
2593 /* Anything else. */
2597 /* We've added more stuff to the buffer. */
2599 }
2601 /* On failure, jump from laststart to b + 3, which will be the
2602 end of the buffer after this jump is inserted. */
2603 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2604 'b + 3'. */
2613 {
2614 /* At least one repetition is required, so insert a
2615 `dummy_failure_jump' before the initial
2616 `on_failure_jump' instruction of the loop. This
2617 effects a skip over that instruction the first time
2618 we hit that loop. */
2623 }
2624 }
2635 {
2637 #ifdef WCHAR
2639 #else
2641 #endif
2644 #ifdef WCHAR
2645 /* We assume a charset(_not) structure as a wchar_t array.
2646 charset[0] = (re_opcode_t) charset(_not)
2647 charset[1] = l (= length of char_classes)
2648 charset[2] = m (= length of collating_symbols)
2649 charset[3] = n (= length of equivalence_classes)
2650 charset[4] = o (= length of char_ranges)
2651 charset[5] = p (= length of chars)
2653 charset[6] = char_class (wctype_t)
2654 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2655 ...
2656 charset[l+5] = char_class (wctype_t)
2658 charset[l+6] = collating_symbol (wchar_t)
2659 ...
2660 charset[l+m+5] = collating_symbol (wchar_t)
2661 ifdef _LIBC we use the index if
2662 _NL_COLLATE_SYMB_EXTRAMB instead of
2663 wchar_t string.
2665 charset[l+m+6] = equivalence_classes (wchar_t)
2666 ...
2667 charset[l+m+n+5] = equivalence_classes (wchar_t)
2668 ifdef _LIBC we use the index in
2669 _NL_COLLATE_WEIGHT instead of
2670 wchar_t string.
2672 charset[l+m+n+6] = range_start
2673 charset[l+m+n+7] = range_end
2674 ...
2675 charset[l+m+n+2o+4] = range_start
2676 charset[l+m+n+2o+5] = range_end
2677 ifdef _LIBC we use the value looked up
2678 in _NL_COLLATE_COLLSEQ instead of
2679 wchar_t character.
2681 charset[l+m+n+2o+6] = char
2682 ...
2683 charset[l+m+n+2o+p+5] = char
2685 */
2687 /* We need at least 6 spaces: the opcode, the length of
2688 char_classes, the length of collating_symbols, the length of
2689 equivalence_classes, the length of char_ranges, the length of
2690 chars. */
2693 /* Save b as laststart. And We use laststart as the pointer
2694 to the first element of the charset here.
2695 In other words, laststart[i] indicates charset[i]. */
2698 /* We test `*p == '^' twice, instead of using an if
2699 statement, so we only need one BUF_PUSH. */
2702 p++;
2704 /* Push the length of char_classes, the length of
2705 collating_symbols, the length of equivalence_classes, the
2706 length of char_ranges and the length of chars. */
2710 /* Remember the first position in the bracket expression. */
2713 /* charset_not matches newline according to a syntax bit. */
2716 {
2719 }
2721 /* Read in characters and ranges, setting map bits. */
2723 {
2728 /* \ might escape characters inside [...] and [^...]. */
2730 {
2738 }
2740 /* Could be the end of the bracket expression. If it's
2741 not (i.e., when the bracket expression is `[]' so
2742 far), the ']' character bit gets set way below. */
2746 /* Look ahead to see if it's a range when the last thing
2747 was a character class. */
2751 /* Look ahead to see if it's a range when the last thing
2752 was a character: if this is a hyphen not at the
2753 beginning or the end of a list, then it's the range
2754 operator. */
2759 {
2760 reg_errcode_t ret;
2761 /* Allocate the space for range_start and range_end. */
2763 /* Update the pointer to indicate end of buffer. */
2769 }
2772 reg_errcode_t ret;
2774 /* Move past the `-'. */
2776 /* Allocate the space for range_start and range_end. */
2778 /* Update the pointer to indicate end of buffer. */
2781 laststart);
2784 }
2786 /* See if we're at the beginning of a possible character
2787 class. */
2795 /* If pattern is `[[:'. */
2799 {
2805 else
2806 /* This is in any case an invalid class name. */
2808 }
2811 /* If isn't a word bracketed by `[:' and `:]':
2812 undo the ending character, the letters, and leave
2813 the leading `:' and `[' (but store them as character). */
2815 {
2819 /* Query the character class as wctype_t. */
2824 /* Throw away the ] at the end of the character
2825 class. */
2830 /* Allocate the space for character class. */
2832 /* Update the pointer to indicate end of buffer. */
2834 /* Move data which follow character classes
2835 not to violate the data. */
2842 /* Store the character class. */
2844 /* Update length of char_classes */
2848 }
2849 else
2850 {
2851 c1++;
2853 PATUNFETCH;
2859 }
2860 }
2863 {
2866 # ifdef _LIBC
2869 # endif
2873 /* If pattern is `[[=' or '[[.'. */
2877 {
2883 else
2884 /* This is in any case an invalid class name. */
2886 }
2890 {
2892 /* If we have no collation data we use the default
2893 collation in which each character is in a class
2894 by itself. It also means that ASCII is the
2895 character set and therefore we cannot have character
2896 with more than one byte in the multibyte
2897 representation. */
2899 /* If not defined _LIBC, we push the name and
2900 `\0' for the sake of matching performance. */
2903 # ifdef _LIBC
2906 # endif
2907 {
2910 }
2911 # ifdef _LIBC
2912 else
2913 {
2920 /* This #include defines a local function! */
2921 # include <locale/weightwc.h>
2924 {
2925 /* We push the index for equivalence class. */
2930 _NL_COLLATE_TABLEWC);
2933 _NL_COLLATE_WEIGHTWC);
2936 _NL_COLLATE_EXTRAWC);
2939 _NL_COLLATE_INDIRECTWC);
2943 /* This is no valid character. */
2947 }
2949 {
2950 /* We push collation sequence value
2951 for collating symbol. */
2961 /* We have to convert the name to a single-byte
2962 string. This is possible since the names
2963 consist of ASCII characters and the internal
2964 representation is UCS4. */
2968 table_size =
2970 _NL_COLLATE_SYMB_HASH_SIZEMB);
2973 _NL_COLLATE_SYMB_TABLEMB);
2976 _NL_COLLATE_SYMB_EXTRAMB);
2978 /* Locate the character in the hashing table. */
2985 {
2986 /* First compare the hashing value. */
2992 {
2993 /* Yep, this is the entry. */
2997 }
2999 /* Next entry. */
3001 }
3004 {
3005 /* Compute the index of the byte sequence
3006 in the table. */
3008 /* Adjust for the alignment. */
3012 }
3014 {
3015 /* No valid character. Match it as a
3016 single byte character. */
3019 /* Update the length of characters */
3023 /* Throw away the ] at the end of the
3024 collating symbol. */
3026 /* exit from the switch block. */
3028 }
3029 else
3031 }
3033 }
3034 # endif
3035 /* Throw away the ] at the end of the equivalence
3036 class (or collating symbol). */
3039 /* Allocate the space for the equivalence class
3040 (or collating symbol) (and '\0' if needed). */
3042 /* Update the pointer to indicate end of buffer. */
3047 /* Calculate the offset of char_ranges,
3048 which is next to equivalence_classes. */
3051 /* Insert space. */
3054 /* Write the equivalence_class and \0. */
3058 /* Update the length of equivalence_classes. */
3061 }
3064 /* Calculate the offset of the equivalence_classes,
3065 which is next to collating_symbols. */
3067 /* Insert space and write the collationg_symbol
3068 and \0. */
3073 /* In re_match_2_internal if range_start < -1, we
3074 assume -range_start is the offset of the
3075 collating symbol which is specified as
3076 the character of the range start. So we assign
3077 -(laststart[1] + laststart[2] + 6) to
3078 range_start. */
3080 /* Update the length of collating_symbol. */
3083 }
3084 }
3085 else
3086 {
3087 c1++;
3089 PATUNFETCH;
3095 }
3096 }
3097 else
3098 {
3103 }
3104 }
3107 /* Ensure that we have enough space to push a charset: the
3108 opcode, the length count, and the bitset; 34 bytes in all. */
3113 /* We test `*p == '^' twice, instead of using an if
3114 statement, so we only need one BUF_PUSH. */
3117 p++;
3119 /* Remember the first position in the bracket expression. */
3122 /* Push the number of bytes in the bitmap. */
3125 /* Clear the whole map. */
3128 /* charset_not matches newline according to a syntax bit. */
3133 /* Read in characters and ranges, setting map bits. */
3135 {
3140 /* \ might escape characters inside [...] and [^...]. */
3142 {
3149 }
3151 /* Could be the end of the bracket expression. If it's
3152 not (i.e., when the bracket expression is `[]' so
3153 far), the ']' character bit gets set way below. */
3157 /* Look ahead to see if it's a range when the last thing
3158 was a character class. */
3162 /* Look ahead to see if it's a range when the last thing
3163 was a character: if this is a hyphen not at the
3164 beginning or the end of a list, then it's the range
3165 operator. */
3170 {
3171 reg_errcode_t ret
3176 }
3180 reg_errcode_t ret;
3182 /* Move past the `-'. */
3188 }
3190 /* See if we're at the beginning of a possible character
3191 class. */
3200 /* If pattern is `[[:'. */
3204 {
3210 else
3211 /* This is in any case an invalid class name. */
3213 }
3216 /* If isn't a word bracketed by `[:' and `:]':
3217 undo the ending character, the letters, and leave
3218 the leading `:' and `[' (but set bits for them). */
3220 {
3221 # if defined _LIBC || WIDE_CHAR_SUPPORT
3231 /* Throw away the ] at the end of the character
3232 class. */
3238 {
3239 # ifdef _LIBC
3242 # else
3245 # endif
3250 }
3253 # else
3271 /* Throw away the ] at the end of the character
3272 class. */
3278 {
3279 /* This was split into 3 if's to
3280 avoid an arbitrary limit in some compiler. */
3299 }
3302 }
3303 else
3304 {
3305 c1++;
3307 PATUNFETCH;
3312 }
3313 }
3315 {
3317 # ifdef _LIBC
3320 # endif
3325 /* If pattern is `[[='. */
3329 {
3335 else
3336 /* This is in any case an invalid class name. */
3338 }
3342 {
3343 /* If we have no collation data we use the default
3344 collation in which each character is in a class
3345 by itself. It also means that ASCII is the
3346 character set and therefore we cannot have character
3347 with more than one byte in the multibyte
3348 representation. */
3349 # ifdef _LIBC
3351 # endif
3352 {
3356 /* Throw away the ] at the end of the equivalence
3357 class. */
3360 /* Set the bit for the character. */
3362 }
3363 # ifdef _LIBC
3364 else
3365 {
3366 /* Try to match the byte sequence in `str' against
3367 those known to the collate implementation.
3368 First find out whether the bytes in `str' are
3369 actually from exactly one character. */
3378 /* This #include defines a local function! */
3379 # include <locale/weight.h>
3392 /* This is no valid character. */
3395 /* Throw away the ] at the end of the equivalence
3396 class. */
3399 /* Now we have to go throught the whole table
3400 and find all characters which have the same
3401 first level weight.
3403 XXX Note that this is not entirely correct.
3404 we would have to match multibyte sequences
3405 but this is not possible with the current
3406 implementation. */
3408 /* XXX This test would have to be changed if we
3409 would allow matching multibyte sequences. */
3411 {
3415 /* Test whether the lenghts match. */
3417 {
3418 /* They do. New compare the bytes of
3419 the weight. */
3428 /* They match. Mark the character as
3429 acceptable. */
3431 }
3432 }
3433 }
3434 # endif
3436 }
3437 else
3438 {
3439 c1++;
3441 PATUNFETCH;
3446 }
3447 }
3449 {
3451 # ifdef _LIBC
3454 # endif
3459 /* If pattern is `[[.'. */
3463 {
3469 else
3470 /* This is in any case an invalid class name. */
3472 }
3476 {
3477 /* If we have no collation data we use the default
3478 collation in which each character is the name
3479 for its own class which contains only the one
3480 character. It also means that ASCII is the
3481 character set and therefore we cannot have character
3482 with more than one byte in the multibyte
3483 representation. */
3484 # ifdef _LIBC
3486 # endif
3487 {
3491 /* Throw away the ] at the end of the equivalence
3492 class. */
3495 /* Set the bit for the character. */
3498 }
3499 # ifdef _LIBC
3500 else
3501 {
3502 /* Try to match the byte sequence in `str' against
3503 those known to the collate implementation.
3504 First find out whether the bytes in `str' are
3505 actually from exactly one character. */
3514 table_size =
3516 _NL_COLLATE_SYMB_HASH_SIZEMB);
3519 _NL_COLLATE_SYMB_TABLEMB);
3522 _NL_COLLATE_SYMB_EXTRAMB);
3524 /* Locate the character in the hashing table. */
3531 {
3532 /* First compare the hashing value. */
3539 {
3540 /* Yep, this is the entry. */
3544 }
3546 /* Next entry. */
3548 }
3551 /* This is no valid character. */
3554 /* Throw away the ] at the end of the equivalence
3555 class. */
3558 /* Now add the multibyte character(s) we found
3559 to the accept list.
3561 XXX Note that this is not entirely correct.
3562 we would have to match multibyte sequences
3563 but this is not possible with the current
3564 implementation. Also, we have to match
3565 collating symbols, which expand to more than
3566 one file, as a whole and not allow the
3567 individual bytes. */
3572 {
3575 }
3576 }
3577 # endif
3579 }
3580 else
3581 {
3582 c1++;
3584 PATUNFETCH;
3589 }
3590 }
3591 else
3592 {
3596 }
3597 }
3599 /* Discard any (non)matching list bytes that are all 0 at the
3600 end of the map. Decrease the map-length byte too. */
3605 }
3612 else
3619 else
3626 else
3633 else
3640 else
3647 /* Do not translate the character after the \, so that we can
3648 distinguish, e.g., \B from \b, even if we normally would
3649 translate, e.g., B to b. */
3653 {
3658 handle_open:
3660 regnum++;
3663 {
3665 compile_stack_elt_t);
3669 }
3671 /* These are the values to restore when we hit end of this
3672 group. They are all relative offsets, so that if the
3673 whole pattern moves because of realloc, they will still
3674 be valid. */
3676 COMPILE_STACK_TOP.fixup_alt_jump
3681 /* We will eventually replace the 0 with the number of
3682 groups inner to this one. But do not push a
3683 start_memory for groups beyond the last one we can
3684 represent in the compiled pattern. */
3686 {
3690 }
3697 /* If we've reached MAX_REGNUM groups, then this open
3698 won't actually generate any code, so we'll have to
3699 clear pending_exact explicitly. */
3708 {
3711 else
3713 }
3715 handle_close:
3718 alternative for a possible future
3719 `pop_failure_jump' to pop. See comments at
3720 `push_dummy_failure' in `re_match_2'. */
3723 /* We allocated space for this jump when we assigned
3724 to `fixup_alt_jump', in the `handle_alt' case below. */
3726 }
3728 /* See similar code for backslashed left paren above. */
3730 {
3733 else
3735 }
3737 /* Since we just checked for an empty stack above, this
3738 ``can't happen''. */
3740 {
3741 /* We don't just want to restore into `regnum', because
3742 later groups should continue to be numbered higher,
3743 as in `(ab)c(de)' -- the second group is #2. */
3744 regnum_t this_group_regnum;
3748 fixup_alt_jump
3754 /* If we've reached MAX_REGNUM groups, then this open
3755 won't actually generate any code, so we'll have to
3756 clear pending_exact explicitly. */
3759 /* We're at the end of the group, so now we know how many
3760 groups were inside this one. */
3762 {
3763 UCHAR_T *inner_group_loc
3769 }
3770 }
3777 handle_alt:
3781 /* Insert before the previous alternative a jump which
3782 jumps to this alternative if the former fails. */
3789 /* The alternative before this one has a jump after it
3790 which gets executed if it gets matched. Adjust that
3791 jump so it will jump to this alternative's analogous
3792 jump (put in below, which in turn will jump to the next
3793 (if any) alternative's such jump, etc.). The last such
3794 jump jumps to the correct final destination. A picture:
3795 _____ _____
3796 | | | |
3797 | v | v
3798 a | b | c
3800 If we are at `b', then fixup_alt_jump right now points to a
3801 three-byte space after `a'. We'll put in the jump, set
3802 fixup_alt_jump to right after `b', and leave behind three
3803 bytes which we'll fill in when we get to after `c'. */
3808 /* Mark and leave space for a jump after this alternative,
3809 to be filled in later either by next alternative or
3810 when know we're at the end of a series of alternatives. */
3821 /* If \{ is a literal. */
3823 /* If we're at `\{' and it's not the open-interval
3824 operator. */
3828 handle_interval:
3829 {
3830 /* If got here, then the syntax allows intervals. */
3832 /* At least (most) this many matches must be made. */
3835 /* Place in the uncompiled pattern (i.e., just after
3836 the '{') to go back to if the interval is invalid. */
3845 {
3849 }
3850 else
3851 /* Interval such as `{1}' => match exactly once. */
3858 {
3862 }
3867 /* If it's invalid to have no preceding re. */
3869 {
3875 else
3877 }
3879 /* We just parsed a valid interval. */
3884 /* If the upper bound is zero, don't want to succeed at
3885 all; jump from `laststart' to `b + 3', which will be
3886 the end of the buffer after we insert the jump. */
3887 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3888 instead of 'b + 3'. */
3890 {
3895 }
3897 /* Otherwise, we have a nontrivial interval. When
3898 we're all done, the pattern will look like:
3899 set_number_at <jump count> <upper bound>
3900 set_number_at <succeed_n count> <lower bound>
3901 succeed_n <after jump addr> <succeed_n count>
3902 <body of loop>
3903 jump_n <succeed_n addr> <jump count>
3904 (The upper bound and `jump_n' are omitted if
3905 `upper_bound' is 1, though.) */
3906 else
3908 more at the end of the loop. */
3914 /* Initialize lower bound of the `succeed_n', even
3915 though it will be set during matching by its
3916 attendant `set_number_at' (inserted next),
3917 because `re_compile_fastmap' needs to know.
3918 Jump to the `jump_n' we might insert below. */
3925 /* Code to initialize the lower bound. Insert
3926 before the `succeed_n'. The `5' is the last two
3927 bytes of this `set_number_at', plus 3 bytes of
3928 the following `succeed_n'. */
3929 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3930 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3931 of the following `succeed_n'. */
3938 append a backward jump to the `succeed_n'
3939 that starts this interval.
3941 When we've reached this during matching,
3942 we'll have matched the interval once, so
3943 jump back only `upper_bound - 1' times. */
3949 /* The location we want to set is the second
3950 parameter of the `jump_n'; that is `b-2' as
3951 an absolute address. `laststart' will be
3952 the `set_number_at' we're about to insert;
3953 `laststart+3' the number to set, the source
3954 for the relative address. But we are
3955 inserting into the middle of the pattern --
3956 so everything is getting moved up by 5.
3957 Conclusion: (b - 2) - (laststart + 3) + 5,
3958 i.e., b - laststart.
3960 We insert this at the beginning of the loop
3961 so that if we fail during matching, we'll
3962 reinitialize the bounds. */
3967 }
3968 }
3972 invalid_interval:
3975 unfetch_interval:
3976 /* Match the characters as literals. */
3981 else
3983 }
3985 #ifdef emacs
3986 /* There is no way to specify the before_dot and after_dot
3987 operators. rms says this is ok. --karl */
4068 /* Can't back reference to a subexpression if inside of it. */
4081 else
4085 normal_backslash:
4086 /* You might think it would be useful for \ to mean
4087 not to translate; but if we don't translate it
4088 it will never match anything. */
4091 }
4096 /* Expects the character in `c'. */
4097 normal_char:
4098 /* If no exactn currently being built. */
4100 #ifdef WCHAR
4101 /* If last exactn handle binary(or character) and
4102 new exactn handle character(or binary). */
4106 /* If last exactn not at current position. */
4109 /* We have only one byte following the exactn for the count. */
4112 /* If followed by a repetition operator. */
4121 {
4122 /* Start building a new exactn. */
4126 #ifdef WCHAR
4127 /* Is this exactn binary data or character? */
4131 else
4133 #else
4137 }
4146 /* Through the pattern now. */
4154 /* If we don't want backtracking, force success
4155 the first time we reach the end of the compiled pattern. */
4159 #ifdef WCHAR
4163 #endif
4166 /* We have succeeded; set the length of the buffer. */
4167 #ifdef WCHAR
4169 #else
4171 #endif
4173 #ifdef DEBUG
4175 {
4178 }
4181 #ifndef MATCH_MAY_ALLOCATE
4182 /* Initialize the failure stack to the largest possible stack. This
4183 isn't necessary unless we're trying to avoid calling alloca in
4184 the search and match routines. */
4185 {
4188 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4189 is strictly greater than re_max_failures, the largest possible stack
4190 is 2 * re_max_failures failure points. */
4192 {
4195 # ifdef emacs
4197 fail_stack.stack
4200 else
4201 fail_stack.stack
4207 fail_stack.stack
4210 else
4211 fail_stack.stack
4216 }
4219 }
4225 /* Subroutines for `regex_compile'. */
4227 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4228 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4230 static void
4232 {
4235 }
4238 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4239 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4241 static void
4243 {
4247 }
4250 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4251 for OP followed by two-byte integer parameter ARG. */
4252 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4254 static void
4256 {
4264 }
4267 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4268 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4270 static void
4273 {
4281 }
4284 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4285 after an alternative or a begin-subexpression. We assume there is at
4286 least one character before the ^. */
4288 static boolean
4290 reg_syntax_t syntax)
4291 {
4295 return
4296 /* After a subexpression? */
4298 /* After an alternative? */
4300 }
4303 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4304 at least one character after the $, i.e., `P < PEND'. */
4306 static boolean
4308 reg_syntax_t syntax)
4309 {
4314 return
4315 /* Before a subexpression? */
4318 /* Before an alternative? */
4321 }
4325 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4326 false if it's not. */
4328 static boolean
4330 {
4335 this_element--)
4340 }
4343 #ifdef INSIDE_RECURSION
4345 #ifdef WCHAR
4346 /* This insert space, which size is "num", into the pattern at "loc".
4347 "end" must point the end of the allocated buffer. */
4348 static void
4350 {
4356 }
4359 #ifdef WCHAR
4360 static reg_errcode_t
4364 {
4367 reg_errcode_t ret;
4368 # ifdef _LIBC
4371 # endif
4375 # ifdef _LIBC
4378 {
4380 _NL_COLLATE_COLLSEQWC);
4385 {
4386 /* range_start is a collating symbol. */
4388 /* Retreive the index and get collation sequence value. */
4391 }
4392 else
4397 /* Report an error if the range is empty and the syntax prohibits
4398 this. */
4402 /* Insert space to the end of the char_ranges. */
4407 }
4408 else
4409 # endif
4410 {
4412 range_start_char;
4414 /* Report an error if the range is empty and the syntax prohibits
4415 this. */
4419 /* Insert space to the end of the char_ranges. */
4424 }
4425 /* Have to increment the pointer into the pattern string, so the
4426 caller isn't still at the ending character. */
4430 }
4432 /* Read the ending character of a range (in a bracket expression) from the
4433 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4434 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4435 Then we set the translation of all bits between the starting and
4436 ending characters (inclusive) in the compiled pattern B.
4438 Return an error code.
4440 We use these short variable names so we can use the same macros as
4441 `regex_compile' itself. */
4443 static reg_errcode_t
4447 {
4450 reg_errcode_t ret;
4451 # if _LIBC
4455 # else
4457 # endif
4462 /* Have to increment the pointer into the pattern string, so the
4463 caller isn't still at the ending character. */
4466 /* Report an error if the range is empty and the syntax prohibits this. */
4469 # if _LIBC
4471 _NL_COLLATE_COLLSEQMB);
4476 {
4480 {
4483 }
4484 }
4485 # else
4486 /* Here we see why `this_char' has to be larger than an `unsigned
4487 char' -- we would otherwise go into an infinite loop, since all
4488 characters <= 0xff. */
4490 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4491 and some compilers cast it to int implicitly, so following for_loop
4492 may fall to (almost) infinite loop.
4493 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4494 To avoid this, we cast p[0] to unsigned int and truncate it. */
4498 {
4501 }
4502 # endif
4505 }
4507 \f
4508 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4509 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4510 characters can start a string that matches the pattern. This fastmap
4511 is used by re_search to skip quickly over impossible starting points.
4513 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4514 area as BUFP->fastmap.
4516 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4517 the pattern buffer.
4519 Returns 0 if we succeed, -2 if an internal error. */
4521 #ifdef WCHAR
4522 /* local function for re_compile_fastmap.
4523 truncate wchar_t character to char. */
4526 static unsigned char
4528 {
4533 # ifdef _LIBC
4535 # else
4537 # endif
4539 }
4542 static int
4544 {
4546 #ifdef MATCH_MAY_ALLOCATE
4548 #endif
4549 #ifndef REGEX_MALLOC
4551 #endif
4555 #ifdef WCHAR
4556 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4557 pattern to (char*) in regex_compile. */
4566 #ifdef REL_ALLOC
4567 /* This holds the pointer to the failure stack, when
4568 it is allocated relocatably. */
4570 #endif
4572 /* Assume that each path through the pattern can be null until
4573 proven otherwise. We set this false at the bottom of switch
4574 statement, to which we get only if a particular path doesn't
4575 match the empty string. */
4578 /* We aren't doing a `succeed_n' to begin with. */
4589 {
4591 {
4592 /* We have reached the (effective) end of pattern. */
4594 {
4597 /* Reset for next path. */
4603 }
4604 else
4606 }
4608 /* We should never be about to go beyond the end of the pattern. */
4612 {
4614 /* I guess the idea here is to simply not bother with a fastmap
4615 if a backreference is used, since it's too hard to figure out
4616 the fastmap for the corresponding group. Setting
4617 `can_be_null' stops `re_search_2' from using the fastmap, so
4618 that is all we do. */
4624 /* Following are the cases which match a character. These end
4625 with `break'. */
4627 #ifdef WCHAR
4636 #ifdef MBS_SUPPORT
4640 #endif
4642 #ifdef WCHAR
4643 /* It is hard to distinguish fastmap from (multi byte) characters
4644 which depends on current locale. */
4660 /* Chars beyond end of map must be allowed. */
4685 {
4688 /* `.' matches anything ... */
4692 /* ... except perhaps newline. */
4696 /* Return if we have already set `can_be_null'; if we have,
4697 then the fastmap is irrelevant. Something's wrong here. */
4701 /* Otherwise, have to check alternative paths. */
4703 }
4705 #ifdef emacs
4722 /* All cases after this match the empty string. These end with
4723 `continue'. */
4757 /* Jump backward implies we just went through the body of a
4758 loop and matched nothing. Opcode jumped to should be
4759 `on_failure_jump' or `succeed_n'. Just treat it like an
4760 ordinary jump. For a * loop, it has pushed its failure
4761 point already; if so, discard that as redundant. */
4766 p++;
4770 /* If what's on the stack is where we are now, pop it. */
4780 handle_on_failure_jump:
4783 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4784 end of the pattern. We don't want to push such a point,
4785 since when we restore it above, entering the switch will
4786 increment `p' past the end of the pattern. We don't need
4787 to push such a point since we obviously won't find any more
4788 fastmap entries beyond `pend'. Such a pattern can match
4789 the null string, though. */
4791 {
4793 {
4796 }
4797 }
4798 else
4802 {
4805 }
4811 /* Get to the number of times to succeed. */
4814 /* Increment p past the n for when k != 0. */
4817 {
4821 }
4840 /* Getting here means we have found the possible starting
4841 characters for one path of the pattern -- and that the empty
4842 string does not match. We need not follow this path further.
4843 Instead, look at the next alternative (remembered on the
4844 stack), or quit if no more. The test at the top of the loop
4845 does these things. */
4850 /* Set `can_be_null' for the last path (also the first path, if the
4851 pattern is empty). */
4854 done:
4857 }
4861 int
4863 {
4864 # ifdef MBS_SUPPORT
4867 else
4868 # endif
4871 #ifdef _LIBC
4873 #endif
4874 \f
4876 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4877 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4878 this memory for recording register information. STARTS and ENDS
4879 must be allocated using the malloc library routine, and must each
4880 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4882 If NUM_REGS == 0, then subsequent matches should allocate their own
4883 register data.
4885 Unless this function is called, the first search or match using
4886 PATTERN_BUFFER will allocate its own register data, without
4887 freeing the old data. */
4889 void
4893 {
4895 {
4900 }
4901 else
4902 {
4906 }
4907 }
4908 #ifdef _LIBC
4910 #endif
4911 \f
4912 /* Searching routines. */
4914 /* Like re_search_2, below, but only one string is specified, and
4915 doesn't let you say where to stop matching. */
4917 int
4920 {
4923 }
4924 #ifdef _LIBC
4926 #endif
4929 /* Using the compiled pattern in BUFP->buffer, first tries to match the
4930 virtual concatenation of STRING1 and STRING2, starting first at index
4931 STARTPOS, then at STARTPOS + 1, and so on.
4933 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
4935 RANGE is how far to scan while trying to match. RANGE = 0 means try
4936 only at STARTPOS; in general, the last start tried is STARTPOS +
4937 RANGE.
4939 In REGS, return the indices of the virtual concatenation of STRING1
4940 and STRING2 that matched the entire BUFP->buffer and its contained
4941 subexpressions.
4943 Do not consider matching one past the index STOP in the virtual
4944 concatenation of STRING1 and STRING2.
4946 We return either the position in the strings at which the match was
4947 found, -1 if no match, or -2 if error (such as failure
4948 stack overflow). */
4950 int
4954 {
4955 # ifdef MBS_SUPPORT
4959 else
4960 # endif
4964 #ifdef _LIBC
4966 #endif
4970 #ifdef INSIDE_RECURSION
4972 #ifdef MATCH_MAY_ALLOCATE
4973 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
4974 #else
4975 # define FREE_VAR(var) free (var); var = NULL
4976 #endif
4978 #ifdef WCHAR
4979 # define MAX_ALLOCA_SIZE 2000
4981 # define FREE_WCS_BUFFERS() \
4982 do { \
4983 if (size1 > MAX_ALLOCA_SIZE) \
4984 { \
4985 free (wcs_string1); \
4986 free (mbs_offset1); \
4987 } \
4988 else \
4989 { \
4990 FREE_VAR (wcs_string1); \
4991 FREE_VAR (mbs_offset1); \
4992 } \
4993 if (size2 > MAX_ALLOCA_SIZE) \
4994 { \
4995 free (wcs_string2); \
4996 free (mbs_offset2); \
4997 } \
4998 else \
4999 { \
5000 FREE_VAR (wcs_string2); \
5001 FREE_VAR (mbs_offset2); \
5002 } \
5003 } while (0)
5005 #endif
5008 static int
5013 {
5019 #ifdef WCHAR
5020 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5022 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5024 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5026 /* They hold whether each wchar_t is binary data or not. */
5030 /* Check for out-of-range STARTPOS. */
5034 /* Fix up RANGE if it might eventually take us outside
5035 the virtual concatenation of STRING1 and STRING2.
5036 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5042 /* If the search isn't to be a backwards one, don't waste time in a
5043 search for a pattern that must be anchored. */
5046 /* `begline' is like `begbuf' if it cannot match at newlines. */
5049 {
5052 else
5054 }
5056 #ifdef emacs
5057 /* In a forward search for something that starts with \=.
5058 don't keep searching past point. */
5060 {
5064 }
5067 /* Update the fastmap now if not correct already. */
5072 #ifdef WCHAR
5073 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5074 fill them with converted string. */
5076 {
5078 {
5082 }
5083 else
5084 {
5088 }
5090 {
5092 {
5096 }
5097 else
5098 {
5102 }
5104 }
5110 else
5112 }
5114 {
5116 {
5120 }
5121 else
5122 {
5126 }
5128 {
5132 else
5135 }
5141 else
5143 }
5147 /* Loop through the string, looking for a place to start matching. */
5149 {
5150 /* If a fastmap is supplied, skip quickly over characters that
5151 cannot be the start of a match. If the pattern can match the
5152 null string, however, we don't need to skip characters; we want
5153 the first null string. */
5155 {
5157 {
5167 /* Written out as an if-else to avoid testing `translate'
5168 inside the loop. */
5173 range--;
5174 else
5176 range--;
5179 }
5181 {
5188 }
5189 }
5191 /* If can't match the null string, and that's all we have left, fail. */
5194 {
5195 #ifdef WCHAR
5197 #endif
5199 }
5201 #ifdef WCHAR
5212 #ifndef REGEX_MALLOC
5213 # ifdef C_ALLOCA
5215 # endif
5216 #endif
5219 {
5220 #ifdef WCHAR
5222 #endif
5224 }
5227 {
5228 #ifdef WCHAR
5230 #endif
5232 }
5234 advance:
5238 {
5239 range--;
5240 startpos++;
5241 }
5242 else
5243 {
5244 range++;
5245 startpos--;
5246 }
5247 }
5248 #ifdef WCHAR
5250 #endif
5252 }
5254 #ifdef WCHAR
5255 /* This converts PTR, a pointer into one of the search wchar_t strings
5256 `string1' and `string2' into an multibyte string offset from the
5257 beginning of that string. We use mbs_offset to optimize.
5258 See convert_mbs_to_wcs. */
5259 # define POINTER_TO_OFFSET(ptr) \
5260 (FIRST_STRING_P (ptr) \
5261 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5262 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5263 + csize1)))
5265 /* This converts PTR, a pointer into one of the search strings `string1'
5266 and `string2' into an offset from the beginning of that string. */
5267 # define POINTER_TO_OFFSET(ptr) \
5268 (FIRST_STRING_P (ptr) \
5269 ? ((regoff_t) ((ptr) - string1)) \
5270 : ((regoff_t) ((ptr) - string2 + size1)))
5273 /* Macros for dealing with the split strings in re_match_2. */
5275 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5277 /* Call before fetching a character with *d. This switches over to
5278 string2 if necessary. */
5279 #define PREFETCH() \
5280 while (d == dend) \
5281 { \
5283 if (dend == end_match_2) \
5284 goto fail; \
5286 d = string2; \
5287 dend = end_match_2; \
5288 }
5290 /* Test if at very beginning or at very end of the virtual concatenation
5291 of `string1' and `string2'. If only one string, it's `string2'. */
5292 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5293 #define AT_STRINGS_END(d) ((d) == end2)
5296 /* Test if D points to a character which is word-constituent. We have
5297 two special cases to check for: if past the end of string1, look at
5298 the first character in string2; and if before the beginning of
5299 string2, look at the last character in string1. */
5300 #ifdef WCHAR
5301 /* Use internationalized API instead of SYNTAX. */
5302 # define WORDCHAR_P(d) \
5303 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5304 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5305 || ((d) == end1 ? *string2 \
5308 # define WORDCHAR_P(d) \
5309 (SYNTAX ((d) == end1 ? *string2 \
5310 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5311 == Sword)
5314 /* Disabled due to a compiler bug -- see comment at case wordbound */
5315 #if 0
5316 /* Test if the character before D and the one at D differ with respect
5317 to being word-constituent. */
5318 #define AT_WORD_BOUNDARY(d) \
5319 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5320 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5321 #endif
5323 /* Free everything we malloc. */
5324 #ifdef MATCH_MAY_ALLOCATE
5325 # ifdef WCHAR
5326 # define FREE_VARIABLES() \
5327 do { \
5328 REGEX_FREE_STACK (fail_stack.stack); \
5329 FREE_VAR (regstart); \
5330 FREE_VAR (regend); \
5331 FREE_VAR (old_regstart); \
5332 FREE_VAR (old_regend); \
5333 FREE_VAR (best_regstart); \
5334 FREE_VAR (best_regend); \
5335 FREE_VAR (reg_info); \
5336 FREE_VAR (reg_dummy); \
5337 FREE_VAR (reg_info_dummy); \
5338 if (!cant_free_wcs_buf) \
5339 { \
5340 FREE_VAR (string1); \
5341 FREE_VAR (string2); \
5342 FREE_VAR (mbs_offset1); \
5343 FREE_VAR (mbs_offset2); \
5344 } \
5345 } while (0)
5347 # define FREE_VARIABLES() \
5348 do { \
5349 REGEX_FREE_STACK (fail_stack.stack); \
5350 FREE_VAR (regstart); \
5351 FREE_VAR (regend); \
5352 FREE_VAR (old_regstart); \
5353 FREE_VAR (old_regend); \
5354 FREE_VAR (best_regstart); \
5355 FREE_VAR (best_regend); \
5356 FREE_VAR (reg_info); \
5357 FREE_VAR (reg_dummy); \
5358 FREE_VAR (reg_info_dummy); \
5359 } while (0)
5361 #else
5362 # ifdef WCHAR
5363 # define FREE_VARIABLES() \
5364 do { \
5365 if (!cant_free_wcs_buf) \
5366 { \
5367 FREE_VAR (string1); \
5368 FREE_VAR (string2); \
5369 FREE_VAR (mbs_offset1); \
5370 FREE_VAR (mbs_offset2); \
5371 } \
5372 } while (0)
5378 /* These values must meet several constraints. They must not be valid
5379 register values; since we have a limit of 255 registers (because
5380 we use only one byte in the pattern for the register number), we can
5381 use numbers larger than 255. They must differ by 1, because of
5382 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5383 be larger than the value for the highest register, so we do not try
5384 to actually save any registers when none are active. */
5385 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5386 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5387 \f
5389 /* Matching routines. */
5392 /* re_match is like re_match_2 except it takes only a single string. */
5394 int
5397 {
5399 # ifdef MBS_SUPPORT
5404 else
5405 # endif
5408 # ifndef REGEX_MALLOC
5409 # ifdef C_ALLOCA
5411 # endif
5412 # endif
5414 }
5415 # ifdef _LIBC
5417 # endif
5422 #ifdef INSIDE_RECURSION
5436 /* re_match_2 matches the compiled pattern in BUFP against the
5437 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5438 and SIZE2, respectively). We start matching at POS, and stop
5439 matching at STOP.
5441 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5442 store offsets for the substring each group matched in REGS. See the
5443 documentation for exactly how many groups we fill.
5445 We return -1 if no match, -2 if an internal error (such as the
5446 failure stack overflowing). Otherwise, we return the length of the
5447 matched substring. */
5449 int
5453 {
5455 # ifdef MBS_SUPPORT
5460 else
5461 # endif
5465 #ifndef REGEX_MALLOC
5466 # ifdef C_ALLOCA
5468 # endif
5469 #endif
5471 }
5472 #ifdef _LIBC
5474 #endif
5478 #ifdef INSIDE_RECURSION
5480 #ifdef WCHAR
5483 /* This check the substring (from 0, to length) of the multibyte string,
5484 to which offset_buffer correspond. And count how many wchar_t_characters
5485 the substring occupy. We use offset_buffer to optimization.
5486 See convert_mbs_to_wcs. */
5488 static int
5490 {
5493 /* Check whether the size is valid. */
5500 /* If there are no multibyte character, offset_buffer[i] == i.
5501 Optmize for this case. */
5505 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5510 {
5518 else
5520 }
5523 }
5526 /* This is a separate function so that we can force an alloca cleanup
5527 afterwards. */
5528 #ifdef WCHAR
5529 static int
5536 /* string1 == string2 == NULL means string1/2, size1/2 and
5537 mbs_offset1/2 need seting up in this function. */
5538 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5541 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5544 static int
5551 {
5552 /* General temporaries. */
5555 #ifdef WCHAR
5556 /* They hold whether each wchar_t is binary data or not. */
5558 /* If true, we can't free string1/2, mbs_offset1/2. */
5562 /* Just past the end of the corresponding string. */
5565 /* Pointers into string1 and string2, just past the last characters in
5566 each to consider matching. */
5569 /* Where we are in the data, and the end of the current string. */
5572 /* Where we are in the pattern, and the end of the pattern. */
5573 #ifdef WCHAR
5581 /* Mark the opcode just after a start_memory, so we can test for an
5582 empty subpattern when we get to the stop_memory. */
5585 /* We use this to map every character in the string. */
5588 /* Failure point stack. Each place that can handle a failure further
5589 down the line pushes a failure point on this stack. It consists of
5590 restart, regend, and reg_info for all registers corresponding to
5591 the subexpressions we're currently inside, plus the number of such
5592 registers, and, finally, two char *'s. The first char * is where
5593 to resume scanning the pattern; the second one is where to resume
5594 scanning the strings. If the latter is zero, the failure point is
5595 a ``dummy''; if a failure happens and the failure point is a dummy,
5596 it gets discarded and the next next one is tried. */
5599 #endif
5600 #ifdef DEBUG
5603 #endif
5605 #ifdef REL_ALLOC
5606 /* This holds the pointer to the failure stack, when
5607 it is allocated relocatably. */
5609 #endif
5611 /* We fill all the registers internally, independent of what we
5612 return, for use in backreferences. The number here includes
5613 an element for register zero. */
5616 /* The currently active registers. */
5620 /* Information on the contents of registers. These are pointers into
5621 the input strings; they record just what was matched (on this
5622 attempt) by a subexpression part of the pattern, that is, the
5623 regnum-th regstart pointer points to where in the pattern we began
5624 matching and the regnum-th regend points to right after where we
5625 stopped matching the regnum-th subexpression. (The zeroth register
5626 keeps track of what the whole pattern matches.) */
5629 #endif
5631 /* If a group that's operated upon by a repetition operator fails to
5632 match anything, then the register for its start will need to be
5633 restored because it will have been set to wherever in the string we
5634 are when we last see its open-group operator. Similarly for a
5635 register's end. */
5638 #endif
5640 /* The is_active field of reg_info helps us keep track of which (possibly
5641 nested) subexpressions we are currently in. The matched_something
5642 field of reg_info[reg_num] helps us tell whether or not we have
5643 matched any of the pattern so far this time through the reg_num-th
5644 subexpression. These two fields get reset each time through any
5645 loop their register is in. */
5648 #endif
5650 /* The following record the register info as found in the above
5651 variables when we find a match better than any we've seen before.
5652 This happens as we backtrack through the failure points, which in
5653 turn happens only if we have not yet matched the entire string. */
5657 #endif
5659 /* Logically, this is `best_regend[0]'. But we don't want to have to
5660 allocate space for that if we're not allocating space for anything
5661 else (see below). Also, we never need info about register 0 for
5662 any of the other register vectors, and it seems rather a kludge to
5663 treat `best_regend' differently than the rest. So we keep track of
5664 the end of the best match so far in a separate variable. We
5665 initialize this to NULL so that when we backtrack the first time
5666 and need to test it, it's not garbage. */
5669 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5672 /* Used when we pop values we don't care about. */
5676 #endif
5678 #ifdef DEBUG
5679 /* Counts the total number of registers pushed. */
5681 #endif
5687 #ifdef MATCH_MAY_ALLOCATE
5688 /* Do not bother to initialize all the register variables if there are
5689 no groups in the pattern, as it takes a fair amount of time. If
5690 there are groups, we include space for register 0 (the whole
5691 pattern), even though we never use it, since it simplifies the
5692 array indexing. We should fix this. */
5694 {
5707 {
5710 }
5711 }
5712 else
5713 {
5714 /* We must initialize all our variables to NULL, so that
5715 `FREE_VARIABLES' doesn't try to free them. */
5719 }
5722 /* The starting position is bogus. */
5723 #ifdef WCHAR
5727 #endif
5728 {
5731 }
5733 #ifdef WCHAR
5734 /* Allocate wchar_t array for string1 and string2 and
5735 fill them with converted string. */
5737 {
5738 /* We need seting up buffers here. */
5740 /* We must free wcs buffers in this function. */
5744 {
5749 {
5754 }
5755 }
5757 {
5762 {
5769 }
5774 }
5775 }
5777 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5778 pattern to (char*) in regex_compile. */
5784 /* Initialize subexpression text positions to -1 to mark ones that no
5785 start_memory/stop_memory has been seen for. Also initialize the
5786 register information struct. */
5788 {
5796 }
5798 /* We move `string1' into `string2' if the latter's empty -- but not if
5799 `string1' is null. */
5801 {
5806 #ifdef WCHAR
5811 #endif
5812 }
5816 /* Compute where to stop matching, within the two strings. */
5817 #ifdef WCHAR
5819 {
5823 }
5824 else
5825 {
5831 }
5836 }
5837 #else
5839 {
5842 }
5843 else
5844 {
5847 }
5850 /* `p' scans through the pattern as `d' scans through the data.
5851 `dend' is the end of the input string that `d' points within. `d'
5852 is advanced into the following input string whenever necessary, but
5853 this happens before fetching; therefore, at the beginning of the
5854 loop, `d' can be pointing at the end of a string, but it cannot
5855 equal `string2'. */
5856 #ifdef WCHAR
5858 {
5862 }
5863 else
5864 {
5868 }
5874 }
5875 #else
5877 {
5880 }
5881 else
5882 {
5885 }
5894 /* This loops over pattern commands. It exits by returning from the
5895 function if the match is complete, or it drops through if the match
5896 fails at this starting point in the input data. */
5898 {
5899 #ifdef _LIBC
5901 #else
5903 #endif
5909 /* If we haven't matched the entire string, and we want the
5910 longest match, try backtracking. */
5912 {
5913 /* 1 if this match ends in the same string (string1 or string2)
5914 as the best previous match. */
5915 boolean same_str_p;
5917 /* 1 if this match is the best seen so far. */
5918 boolean best_match_p;
5923 /* AIX compiler got confused when this was combined
5924 with the previous declaration. */
5927 else
5935 /* If exceeds best match so far, save it. */
5937 {
5944 {
5947 }
5948 }
5950 }
5952 /* If no failure points, don't restore garbage. And if
5953 last match is real best match, don't restore second
5954 best one. */
5956 {
5957 restore_best_regs:
5958 /* Restore best match. It may happen that `dend ==
5959 end_match_1' while the restored d is in string2.
5960 For example, the pattern `x.*y.*z' against the
5961 strings `x-' and `y-z-', if the two strings are
5962 not consecutive in memory. */
5970 {
5973 }
5974 }
5977 succeed_label:
5979 /* If caller wants register contents data back, do it. */
5981 {
5982 /* Have the register data arrays been allocated? */
5985 extra element beyond `num_regs' for the `-1' marker
5986 GNU code uses. */
5991 {
5994 }
5996 }
5999 allocated, reallocate them. If we need fewer, just
6000 leave it alone. */
6002 {
6007 {
6010 }
6011 }
6012 }
6013 else
6014 {
6015 /* These braces fend off a "empty body in an else-statement"
6016 warning under GCC when assert expands to nothing. */
6018 }
6020 /* Convert the pointer data in `regstart' and `regend' to
6021 indices. Register zero has to be set differently,
6022 since we haven't kept track of any info for it. */
6024 {
6026 #ifdef WCHAR
6030 else
6033 #else
6038 }
6040 /* Go through the first `min (num_regs, regs->num_regs)'
6041 registers, since that is all we initialized. */
6043 mcnt++)
6044 {
6047 else
6048 {
6053 }
6054 }
6056 /* If the regs structure we return has more elements than
6057 were in the pattern, set the extra elements to -1. If
6058 we (re)allocated the registers, this is the case,
6059 because we always allocate enough to have at least one
6060 -1 at the end. */
6070 #ifdef WCHAR
6073 else
6075 csize1;
6077 #else
6079 ? string1
6087 }
6089 /* Otherwise match next pattern command. */
6091 {
6092 /* Ignore these. Used to ignore the n of succeed_n's which
6093 currently have n == 0. */
6102 /* Match the next n pattern characters exactly. The following
6103 byte in the pattern defines n, and the n bytes after that
6104 are the characters to match. */
6106 #ifdef MBS_SUPPORT
6108 #endif
6112 /* This is written out as an if-else so we don't waste time
6113 testing `translate' inside the loop. */
6115 {
6116 do
6117 {
6119 #ifdef WCHAR
6121 {
6125 }
6126 else
6127 {
6130 }
6131 #else
6136 }
6138 }
6139 else
6140 {
6141 do
6142 {
6145 }
6147 }
6152 /* Match any character except possibly a newline or a null. */
6164 d++;
6170 {
6172 #ifdef WCHAR
6176 #define WORK_BUFFER_SIZE 128
6178 # ifdef _LIBC
6187 #ifdef WCHAR
6188 # ifdef _LIBC
6197 /* p points charset[6], so the address of the next instruction
6198 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6199 where l=length of char_classes, m=length of collating_symbol,
6200 n=equivalence_class, o=length of char_range,
6201 p'=length of character. */
6203 /* Update p to indicate the next instruction. */
6207 /* match with char_class? */
6209 {
6216 # ifdef _LIBC
6219 # else
6222 # endif
6223 }
6225 /* match with collating_symbol? */
6226 # ifdef _LIBC
6228 {
6233 workp++)
6234 {
6242 {
6243 /* Update d, however d will be incremented at
6244 char_set_matched:, we decrement d here. */
6247 }
6248 }
6249 }
6251 # endif
6252 /* If we can't look up collation data, we use wcscoll
6253 instead. */
6254 {
6256 {
6258 # ifdef _LIBC
6260 # else
6262 # endif
6264 /* If wcscoll(the collating symbol, whole string) > 0,
6265 any substring of the string never match with the
6266 collating symbol. */
6267 # ifdef _LIBC
6269 # else
6271 # endif
6272 {
6275 }
6277 /* First, we compare the collating symbol with
6278 the first character of the string.
6279 If it don't match, we add the next character to
6280 the compare buffer in turn. */
6282 {
6285 {
6290 }
6292 /* add next character to the compare buffer. */
6296 # ifdef _LIBC
6298 # else
6300 # endif
6305 /* (str_buf > workp) indicate (str_buf + X > workp),
6306 because for all X (str_buf + X > str_buf).
6307 So we don't need continue this loop. */
6310 /* Otherwise(str_buf < workp),
6311 (str_buf+next_character) may equals (workp).
6312 So we continue this loop. */
6313 }
6314 /* not matched */
6318 }
6319 }
6320 /* match with equivalence_class? */
6321 # ifdef _LIBC
6323 {
6325 /* Try to match the equivalence class against
6326 those known to the collate implementation. */
6335 /* This #include defines a local function! */
6336 # include <locale/weightwc.h>
6347 /* Write 1 collating element to str_buf, and
6348 get its index. */
6352 {
6355 {
6360 }
6364 }
6366 /* Update d, however d will be incremented at
6367 char_set_matched:, we decrement d here. */
6370 {
6373 else
6374 {
6377 }
6378 }
6383 workp++)
6384 {
6386 /* We already checked idx != 0 in regex_compile. */
6389 {
6397 }
6398 }
6399 /* not matched */
6402 }
6404 # endif
6405 /* If we can't look up collation data, we use wcscoll
6406 instead. */
6407 {
6409 {
6411 # ifdef _LIBC
6413 # else
6415 # endif
6417 /* If wcscoll(the collating symbol, whole string) > 0,
6418 any substring of the string never match with the
6419 collating symbol. */
6420 # ifdef _LIBC
6422 # else
6424 # endif
6425 {
6428 }
6430 /* First, we compare the equivalence class with
6431 the first character of the string.
6432 If it don't match, we add the next character to
6433 the compare buffer in turn. */
6435 {
6438 {
6443 }
6445 /* add next character to the compare buffer. */
6449 # ifdef _LIBC
6451 # else
6453 # endif
6459 /* (str_buf > workp) indicate (str_buf + X > workp),
6460 because for all X (str_buf + X > str_buf).
6461 So we don't need continue this loop. */
6464 /* Otherwise(str_buf < workp),
6465 (str_buf+next_character) may equals (workp).
6466 So we continue this loop. */
6467 }
6468 /* not matched */
6472 }
6473 }
6475 /* match with char_range? */
6476 # ifdef _LIBC
6478 {
6486 {
6489 /* We already compute the collation sequence value
6490 of the characters (or collating symbols). */
6496 }
6497 }
6498 else
6499 # endif
6500 {
6501 /* We set range_start_char at str_buf[0], range_end_char
6502 at str_buf[4], and compared char at str_buf[2]. */
6508 {
6511 /* match if (range_start_char <= c <= range_end_char). */
6513 /* If range_start(or end) < 0, we assume -range_start(end)
6514 is the offset of the collating symbol which is specified
6515 as the character of the range start(end). */
6517 /* range_start */
6520 else
6521 {
6524 }
6526 /* range_end */
6529 else
6530 {
6533 }
6535 # ifdef _LIBC
6538 # else
6541 # endif
6543 }
6544 }
6546 /* match with char? */
6553 char_set_matched:
6555 #else
6556 /* Cast to `unsigned' instead of `unsigned char' in case the
6557 bit list is a full 32 bytes long. */
6565 #undef WORK_BUFFER_SIZE
6568 d++;
6570 }
6573 /* The beginning of a group is represented by start_memory.
6574 The arguments are the register number in the next byte, and the
6575 number of groups inner to this one in the next. The text
6576 matched within the group is recorded (in the internal
6577 registers data structure) under the register number. */
6582 /* Find out if this group can match the empty string. */
6589 /* Save the position in the string where we were the last time
6590 we were at this open-group operator in case the group is
6591 operated upon by a repetition operator, e.g., with `(a*)*b'
6592 against `ab'; then we want to ignore where we are now in
6593 the string in case this attempt to match fails. */
6606 /* Clear this whenever we change the register activity status. */
6609 /* This is the new highest active register. */
6612 /* If nothing was active before, this is the new lowest active
6613 register. */
6617 /* Move past the register number and inner group count. */
6624 /* The stop_memory opcode represents the end of a group. Its
6625 arguments are the same as start_memory's: the register
6626 number, and the number of inner groups. */
6631 /* We need to save the string position the last time we were at
6632 this close-group operator in case the group is operated
6633 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6634 against `aba'; then we want to ignore where we are now in
6635 the string in case this attempt to match fails. */
6645 /* This register isn't active anymore. */
6648 /* Clear this whenever we change the register activity status. */
6651 /* If this was the only register active, nothing is active
6652 anymore. */
6654 {
6657 }
6658 else
6660 it isn't necessarily one less than now: consider
6661 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6662 new highest active register is 1. */
6665 r--;
6667 /* If we end up at register zero, that means that we saved
6668 the registers as the result of an `on_failure_jump', not
6669 a `start_memory', and we jumped to past the innermost
6670 `stop_memory'. For example, in ((.)*) we save
6671 registers 1 and 2 as a result of the *, but when we pop
6672 back to the second ), we are at the stop_memory 1.
6673 Thus, nothing is active. */
6675 {
6678 }
6679 else
6681 }
6683 /* If just failed to match something this time around with a
6684 group that's operated on by a repetition operator, try to
6685 force exit from the ``loop'', and restore the register
6686 information for this group that we had before trying this
6687 last match. */
6691 {
6697 {
6711 }
6714 /* If the next operation is a jump backwards in the pattern
6715 to an on_failure_jump right before the start_memory
6716 corresponding to this stop_memory, exit from the loop
6717 by forcing a failure after pushing on the stack the
6718 on_failure_jump's jump in the pattern, and d. */
6722 {
6723 /* If this group ever matched anything, then restore
6724 what its registers were before trying this last
6725 failed match, e.g., with `(a*)*b' against `ab' for
6726 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6727 against `aba' for regend[3].
6729 Also restore the registers for inner groups for,
6730 e.g., `((a*)(b*))*' against `aba' (register 3 would
6731 otherwise get trashed). */
6734 {
6739 /* Restore this and inner groups' (if any) registers. */
6741 r++)
6742 {
6745 /* xx why this test? */
6748 }
6749 }
6750 p1++;
6755 }
6756 }
6758 /* Move past the register number and the inner group count. */
6763 /* \<digit> has been turned into a `duplicate' command which is
6764 followed by the numeric value of <digit> as the register number. */
6766 {
6771 /* Can't back reference a group which we've never matched. */
6775 /* Where in input to try to start matching. */
6778 /* Where to stop matching; if both the place to start and
6779 the place to stop matching are in the same string, then
6780 set to the place to stop, otherwise, for now have to use
6781 the end of the first string. */
6787 {
6788 /* If necessary, advance to next segment in register
6789 contents. */
6791 {
6795 /* End of string1 => advance to string2. */
6798 }
6799 /* At end of register contents => success */
6802 /* If necessary, advance to next segment in data. */
6805 /* How many characters left in this segment to match. */
6808 /* Want how many consecutive characters we can match in
6809 one shot, so, if necessary, adjust the count. */
6813 /* Compare that many; failure if mismatch, else move
6814 past them. */
6821 /* Do this because we've match some characters. */
6823 }
6824 }
6828 /* begline matches the empty string at the beginning of the string
6829 (unless `not_bol' is set in `bufp'), and, if
6830 `newline_anchor' is set, after newlines. */
6835 {
6837 }
6839 {
6841 }
6842 /* In all other cases, we fail. */
6846 /* endline is the dual of begline. */
6851 {
6853 }
6855 /* We have to ``prefetch'' the next character. */
6858 {
6860 }
6864 /* Match at the very beginning of the data. */
6872 /* Match at the very end of the data. */
6880 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6881 pushes NULL as the value for the string on the stack. Then
6882 `pop_failure_point' will keep the current value for the
6883 string, instead of restoring it. To see why, consider
6884 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6885 then the . fails against the \n. But the next thing we want
6886 to do is match the \n against the \n; if we restored the
6887 string value, we would be back at the foo.
6889 Because this is used only in specific cases, we don't need to
6890 check all the things that `on_failure_jump' does, to make
6891 sure the right things get saved on the stack. Hence we don't
6892 share its code. The only reason to push anything on the
6893 stack at all is that otherwise we would have to change
6894 `anychar's code to do something besides goto fail in this
6895 case; that seems worse than this. */
6900 #ifdef _LIBC
6902 #else
6904 #endif
6910 /* Uses of on_failure_jump:
6912 Each alternative starts with an on_failure_jump that points
6913 to the beginning of the next alternative. Each alternative
6914 except the last ends with a jump that in effect jumps past
6915 the rest of the alternatives. (They really jump to the
6916 ending jump of the following alternative, because tensioning
6917 these jumps is a hassle.)
6919 Repeats start with an on_failure_jump that points past both
6920 the repetition text and either the following jump or
6921 pop_failure_jump back to this on_failure_jump. */
6923 on_failure:
6927 #ifdef _LIBC
6929 #else
6931 #endif
6933 /* If this on_failure_jump comes right before a group (i.e.,
6934 the original * applied to a group), save the information
6935 for that group and all inner ones, so that if we fail back
6936 to this point, the group's information will be correct.
6937 For example, in \(a*\)*\1, we need the preceding group,
6938 and in \(zz\(a*\)b*\)\2, we need the inner group. */
6940 /* We can't use `p' to check ahead because we push
6941 a failure point to `p + mcnt' after we do this. */
6944 /* We need to skip no_op's before we look for the
6945 start_memory in case this on_failure_jump is happening as
6946 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
6947 against aba. */
6949 p1++;
6952 {
6953 /* We have a new highest active register now. This will
6954 get reset at the start_memory we are about to get to,
6955 but we will have saved all the registers relevant to
6956 this repetition op, as described above. */
6960 }
6967 /* A smart repeat ends with `maybe_pop_jump'.
6968 We change it to either `pop_failure_jump' or `jump'. */
6972 {
6975 /* Compare the beginning of the repeat with what in the
6976 pattern follows its end. If we can establish that there
6977 is nothing that they would both match, i.e., that we
6978 would have to backtrack because of (as in, e.g., `a*a')
6979 then we can change to pop_failure_jump, because we'll
6980 never have to backtrack.
6982 This is not true in the case of alternatives: in
6983 `(a|ab)*' we do need to backtrack to the `ab' alternative
6984 (e.g., if the string was `ab'). But instead of trying to
6985 detect that here, the alternative has put on a dummy
6986 failure point which is what we will end up popping. */
6988 /* Skip over open/close-group commands.
6989 If what follows this loop is a ...+ construct,
6990 look at what begins its body, since we will have to
6991 match at least one of that. */
6993 {
7001 else
7003 }
7006 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7007 to the `maybe_finalize_jump' of this case. Examine what
7008 follows. */
7010 /* If we're at the end of the pattern, we can change. */
7012 {
7013 /* Consider what happens when matching ":\(.*\)"
7014 against ":/". I don't really understand this code
7015 yet. */
7017 pop_failure_jump;
7018 DEBUG_PRINT1
7020 }
7023 #ifdef MBS_SUPPORT
7025 #endif
7027 {
7028 register UCHAR_T c
7032 #ifdef MBS_SUPPORT
7034 #endif
7036 {
7038 pop_failure_jump;
7039 #ifdef WCHAR
7043 #else
7047 #endif
7048 }
7050 #ifndef WCHAR
7053 {
7060 /* `negate' is equal to 1 if c would match, which means
7061 that we can't change to pop_failure_jump. */
7063 {
7066 }
7067 }
7069 }
7070 #ifndef WCHAR
7072 {
7073 /* We win if the first character of the loop is not part
7074 of the charset. */
7079 {
7082 }
7085 {
7087 /* We win if the charset_not inside the loop
7088 lists every character listed in the charset after. */
7096 {
7099 }
7100 }
7102 {
7104 /* We win if the charset inside the loop
7105 has no overlap with the one after the loop. */
7108 idx++)
7113 {
7116 }
7117 }
7118 }
7120 }
7123 {
7127 }
7128 /* Note fall through. */
7131 /* The end of a simple repeat has a pop_failure_jump back to
7132 its matching on_failure_jump, where the latter will push a
7133 failure point. The pop_failure_jump takes off failure
7134 points put on by this pop_failure_jump's matching
7135 on_failure_jump; we got through the pattern to here from the
7136 matching on_failure_jump, so didn't fail. */
7138 {
7139 /* We need to pass separate storage for the lowest and
7140 highest registers, even though we don't care about the
7141 actual values. Otherwise, we will restore only one
7142 register from the stack, since lowest will == highest in
7143 `pop_failure_point'. */
7152 }
7153 /* Note fall through. */
7155 unconditional_jump:
7156 #ifdef _LIBC
7158 #else
7160 #endif
7161 /* Note fall through. */
7163 /* Unconditionally jump (without popping any failure points). */
7168 #ifdef _LIBC
7170 #else
7172 #endif
7176 /* We need this opcode so we can detect where alternatives end
7177 in `group_match_null_string_p' et al. */
7183 /* Normally, the on_failure_jump pushes a failure point, which
7184 then gets popped at pop_failure_jump. We will end up at
7185 pop_failure_jump, also, and with a pattern of, say, `a+', we
7186 are skipping over the on_failure_jump, so we have to push
7187 something meaningless for pop_failure_jump to pop. */
7190 /* It doesn't matter what we push for the string here. What
7191 the code at `fail' tests is the value for the pattern. */
7196 /* At the end of an alternative, we need to push a dummy failure
7197 point in case we are followed by a `pop_failure_jump', because
7198 we don't want the failure point for the alternative to be
7199 popped. For example, matching `(a|ab)*' against `aab'
7200 requires that we match the `ab' alternative. */
7203 /* See comments just above at `dummy_failure_jump' about the
7204 two zeroes. */
7208 /* Have to succeed matching what follows at least n times.
7209 After that, handle like `on_failure_jump'. */
7215 /* Originally, this is how many times we HAVE to succeed. */
7217 {
7218 mcnt--;
7221 #ifdef _LIBC
7224 #else
7227 #endif
7228 }
7230 {
7231 #ifdef _LIBC
7234 #else
7239 #ifdef WCHAR
7241 #else
7246 }
7253 /* Originally, this is how many times we CAN jump. */
7255 {
7256 mcnt--;
7259 #ifdef _LIBC
7261 mcnt);
7262 #else
7264 mcnt);
7267 }
7268 /* If don't have to jump any more, skip over the rest of command. */
7269 else
7274 {
7280 #ifdef _LIBC
7282 #else
7284 #endif
7287 }
7289 #if 0
7290 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7291 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7292 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7293 macro and introducing temporary variables works around the bug. */
7306 #else
7308 {
7320 }
7323 {
7335 }
7336 #endif
7352 #ifdef emacs
7379 matchsyntax:
7381 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7382 d++;
7396 matchnotsyntax:
7398 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7399 d++;
7412 d++;
7421 d++;
7427 }
7431 /* We goto here if a matching operation fails. */
7432 fail:
7440 /* If this failure point is a dummy, try the next one. */
7444 /* If we failed to the end of the pattern, don't examine *p. */
7447 {
7450 /* If failed to a backwards jump that's part of a repetition
7451 loop, need to pop this failure point and use the next one. */
7453 {
7464 || (!is_a_jump_n
7470 }
7471 }
7475 }
7476 else
7487 \f
7488 /* Subroutine definitions for re_match_2. */
7491 /* We are passed P pointing to a register number after a start_memory.
7493 Return true if the pattern up to the corresponding stop_memory can
7494 match the empty string, and false otherwise.
7496 If we find the matching stop_memory, sets P to point to one past its number.
7497 Otherwise, sets P to an undefined byte less than or equal to END.
7499 We don't handle duplicates properly (yet). */
7501 static boolean
7504 {
7506 /* Point to after the args to the start_memory. */
7510 {
7511 /* Skip over opcodes that can match nothing, and return true or
7512 false, as appropriate, when we get to one that can't, or to the
7513 matching stop_memory. */
7516 {
7517 /* Could be either a loop or a series of alternatives. */
7519 p1++;
7522 /* If the next operation is not a jump backwards in the
7523 pattern. */
7526 {
7527 /* Go through the on_failure_jumps of the alternatives,
7528 seeing if any of the alternatives cannot match nothing.
7529 The last alternative starts with only a jump,
7530 whereas the rest start with on_failure_jump and end
7531 with a jump, e.g., here is the pattern for `a|b|c':
7533 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7534 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7535 /exactn/1/c
7537 So, we have to first go through the first (n-1)
7538 alternatives and then deal with the last one separately. */
7541 /* Deal with the first (n-1) alternatives, which start
7542 with an on_failure_jump (see above) that jumps to right
7543 past a jump_past_alt. */
7546 jump_past_alt)
7547 {
7548 /* `mcnt' holds how many bytes long the alternative
7549 is, including the ending `jump_past_alt' and
7550 its number. */
7554 reg_info))
7557 /* Move to right after this alternative, including the
7558 jump_past_alt. */
7561 /* Break if it's the beginning of an n-th alternative
7562 that doesn't begin with an on_failure_jump. */
7566 /* Still have to check that it's not an n-th
7567 alternative that starts with an on_failure_jump. */
7568 p1++;
7571 jump_past_alt)
7572 {
7573 /* Get to the beginning of the n-th alternative. */
7576 }
7577 }
7579 /* Deal with the last alternative: go back and get number
7580 of the `jump_past_alt' just before it. `mcnt' contains
7581 the length of the alternative. */
7601 }
7608 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7609 It expects P to be the first byte of a single alternative and END one
7610 byte past the last. The alternative can contain groups. */
7612 static boolean
7615 {
7620 {
7621 /* Skip over opcodes that can match nothing, and break when we get
7622 to one that can't. */
7625 {
7626 /* It's a loop. */
7628 p1++;
7636 }
7643 /* Deals with the ops common to group_match_null_string_p and
7644 alt_match_null_string_p.
7646 Sets P to one after the op and its arguments, if any. */
7648 static boolean
7651 {
7653 boolean ret;
7658 {
7668 #ifdef emacs
7672 #endif
7680 /* Have to set this here in case we're checking a group which
7681 contains a group and a back reference to it. */
7690 /* If this is an optimized succeed_n for zero times, make the jump. */
7695 else
7700 /* Get to the number of times to succeed. */
7705 {
7709 }
7710 else
7723 /* All other opcodes mean we cannot match the empty string. */
7725 }
7732 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7733 bytes; nonzero otherwise. */
7735 static int
7737 RE_TRANSLATE_TYPE translate)
7738 {
7742 {
7743 #ifdef WCHAR
7750 len--;
7751 }
7753 }
7754 \f
7758 /* Entry points for GNU code. */
7760 /* re_compile_pattern is the GNU regular expression compiler: it
7761 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7762 Returns 0 if the pattern was valid, otherwise an error string.
7764 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7765 are set in BUFP on entry.
7767 We call regex_compile to do the actual compilation. */
7772 {
7773 reg_errcode_t ret;
7775 /* GNU code is written to assume at least RE_NREGS registers will be set
7776 (and at least one extra will be -1). */
7779 /* And GNU code determines whether or not to get register information
7780 by passing null for the REGS argument to re_match, etc., not by
7781 setting no_sub. */
7784 /* Match anchors at newline. */
7787 # ifdef MBS_SUPPORT
7790 else
7791 # endif
7797 }
7798 #ifdef _LIBC
7800 #endif
7801 \f
7802 /* Entry points compatible with 4.2 BSD regex library. We don't define
7803 them unless specifically requested. */
7805 #if defined _REGEX_RE_COMP || defined _LIBC
7807 /* BSD has one and only one pattern buffer. */
7811 #ifdef _LIBC
7812 /* Make these definitions weak in libc, so POSIX programs can redefine
7813 these names if they don't use our functions, and still use
7814 regcomp/regexec below without link errors. */
7815 weak_function
7816 #endif
7818 {
7819 reg_errcode_t ret;
7822 {
7826 }
7829 {
7838 }
7840 /* Since `re_exec' always passes NULL for the `regs' argument, we
7841 don't need to initialize the pattern buffer fields which affect it. */
7843 /* Match anchors at newlines. */
7846 # ifdef MBS_SUPPORT
7849 else
7850 # endif
7856 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7858 }
7861 int
7862 #ifdef _LIBC
7863 weak_function
7864 #endif
7866 {
7868 return
7870 }
7873 \f
7874 /* POSIX.2 functions. Don't define these for Emacs. */
7876 #ifndef emacs
7878 /* regcomp takes a regular expression as a string and compiles it.
7880 PREG is a regex_t *. We do not expect any fields to be initialized,
7881 since POSIX says we shouldn't. Thus, we set
7883 `buffer' to the compiled pattern;
7884 `used' to the length of the compiled pattern;
7885 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7886 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7887 RE_SYNTAX_POSIX_BASIC;
7888 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7889 `fastmap' to an allocated space for the fastmap;
7890 `fastmap_accurate' to zero;
7891 `re_nsub' to the number of subexpressions in PATTERN.
7893 PATTERN is the address of the pattern string.
7895 CFLAGS is a series of bits which affect compilation.
7897 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
7898 use POSIX basic syntax.
7900 If REG_NEWLINE is set, then . and [^...] don't match newline.
7901 Also, regexec will try a match beginning after every newline.
7903 If REG_ICASE is set, then we considers upper- and lowercase
7904 versions of letters to be equivalent when matching.
7906 If REG_NOSUB is set, then when PREG is passed to regexec, that
7907 routine will report only success or failure, and nothing about the
7908 registers.
7910 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
7911 the return codes and their meanings.) */
7913 int
7915 {
7916 reg_errcode_t ret;
7917 reg_syntax_t syntax
7921 /* regex_compile will allocate the space for the compiled pattern. */
7926 /* Try to allocate space for the fastmap. */
7930 {
7933 preg->translate
7939 /* Map uppercase characters to corresponding lowercase ones. */
7942 }
7943 else
7946 /* If REG_NEWLINE is set, newlines are treated differently. */
7951 /* It also changes the matching behavior. */
7953 }
7954 else
7959 /* POSIX says a null character in the pattern terminates it, so we
7960 can use strlen here in compiling the pattern. */
7961 # ifdef MBS_SUPPORT
7964 else
7965 # endif
7968 /* POSIX doesn't distinguish between an unmatched open-group and an
7969 unmatched close-group: both are REG_EPAREN. */
7973 {
7974 /* Compute the fastmap now, since regexec cannot modify the pattern
7975 buffer. */
7977 {
7978 /* Some error occurred while computing the fastmap, just forget
7979 about it. */
7982 }
7983 }
7986 }
7987 #ifdef _LIBC
7989 #endif
7992 /* regexec searches for a given pattern, specified by PREG, in the
7993 string STRING.
7995 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
7996 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
7997 least NMATCH elements, and we set them to the offsets of the
7998 corresponding matched substrings.
8000 EFLAGS specifies `execution flags' which affect matching: if
8001 REG_NOTBOL is set, then ^ does not match at the beginning of the
8002 string; if REG_NOTEOL is set, then $ does not match at the end.
8004 We return 0 if we find a match and REG_NOMATCH if not. */
8006 int
8009 {
8012 regex_t private_preg;
8021 /* The user has told us exactly how many registers to return
8022 information about, via `nmatch'. We have to pass that on to the
8023 matching routines. */
8027 {
8033 }
8035 /* Perform the searching operation. */
8040 /* Copy the register information to the POSIX structure. */
8042 {
8044 {
8048 {
8051 }
8052 }
8054 /* If we needed the temporary register info, free the space now. */
8056 }
8058 /* We want zero return to mean success, unlike `re_search'. */
8060 }
8061 #ifdef _LIBC
8063 #endif
8066 /* Returns a message corresponding to an error code, ERRCODE, returned
8067 from either regcomp or regexec. We don't use PREG here. */
8069 size_t
8072 {
8079 /* Only error codes returned by the rest of the code should be passed
8080 to this routine. If we are given anything else, or if other regex
8081 code generates an invalid error code, then the program has a bug.
8082 Dump core so we can fix it. */
8090 {
8092 {
8093 #if defined HAVE_MEMPCPY || defined _LIBC
8095 #else
8098 #endif
8099 }
8100 else
8102 }
8105 }
8106 #ifdef _LIBC
8108 #endif
8111 /* Free dynamically allocated space used by PREG. */
8113 void
8115 {
8128 }
8129 #ifdef _LIBC
8131 #endif
8137 \f
8138 #undef STORE_NUMBER
8139 #undef STORE_NUMBER_AND_INCR
8140 #undef EXTRACT_NUMBER
8141 #undef EXTRACT_NUMBER_AND_INCR
8143 #undef DEBUG_PRINT_COMPILED_PATTERN
8144 #undef DEBUG_PRINT_DOUBLE_STRING
8146 #undef INIT_FAIL_STACK
8147 #undef RESET_FAIL_STACK
8148 #undef DOUBLE_FAIL_STACK
8149 #undef PUSH_PATTERN_OP
8150 #undef PUSH_FAILURE_POINTER
8151 #undef PUSH_FAILURE_INT
8152 #undef PUSH_FAILURE_ELT
8153 #undef POP_FAILURE_POINTER
8154 #undef POP_FAILURE_INT
8155 #undef POP_FAILURE_ELT
8156 #undef DEBUG_PUSH
8157 #undef DEBUG_POP
8158 #undef PUSH_FAILURE_POINT
8159 #undef POP_FAILURE_POINT
8161 #undef REG_UNSET_VALUE
8162 #undef REG_UNSET
8164 #undef PATFETCH
8165 #undef PATFETCH_RAW
8166 #undef PATUNFETCH
8167 #undef TRANSLATE
8169 #undef INIT_BUF_SIZE
8170 #undef GET_BUFFER_SPACE
8171 #undef BUF_PUSH
8172 #undef BUF_PUSH_2
8173 #undef BUF_PUSH_3
8174 #undef STORE_JUMP
8175 #undef STORE_JUMP2
8176 #undef INSERT_JUMP
8177 #undef INSERT_JUMP2
8178 #undef EXTEND_BUFFER
8179 #undef GET_UNSIGNED_NUMBER
8180 #undef FREE_STACK_RETURN
8182 # undef POINTER_TO_OFFSET
8183 # undef MATCHING_IN_FRST_STRING
8184 # undef PREFETCH
8185 # undef AT_STRINGS_BEG
8186 # undef AT_STRINGS_END
8187 # undef WORDCHAR_P
8188 # undef FREE_VAR
8189 # undef FREE_VARIABLES
8190 # undef NO_HIGHEST_ACTIVE_REG
8191 # undef NO_LOWEST_ACTIVE_REG
8193 # undef CHAR_T
8194 # undef UCHAR_T
8195 # undef COMPILED_BUFFER_VAR
8196 # undef OFFSET_ADDRESS_SIZE
8197 # undef CHAR_CLASS_SIZE
8198 # undef PREFIX
8199 # undef ARG_PREFIX
8200 # undef PUT_CHAR
8201 # undef BYTE
8202 # undef WCHAR
8204 # define DEFINED_ONCE