| blob | 3babf6cbe342874c8c529a708a90d8f4e3203021 |
1 /* Extended regular expression matching and search library,
2 version 0.12.
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
5 Copyright (C) 1993-1999, 2000, 2001, 2002 Free Software Foundation, Inc.
6 This file is part of the GNU C Library.
8 The GNU C Library is free software; you can redistribute it and/or
9 modify it under the terms of the GNU Lesser General Public
10 License as published by the Free Software Foundation; either
11 version 2.1 of the License, or (at your option) any later version.
13 The GNU C Library is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 Lesser General Public License for more details.
18 You should have received a copy of the GNU Lesser General Public
19 License along with the GNU C Library; if not, write to the Free
20 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
21 02111-1307 USA. */
23 /* This file has been modified for usage in libiberty. It includes "xregex.h"
24 instead of <regex.h>. The "xregex.h" header file renames all external
25 routines with an "x" prefix so they do not collide with the native regex
26 routines or with other components regex routines. */
27 /* AIX requires this to be the first thing in the file. */
28 #if defined _AIX && !defined __GNUC__ && !defined REGEX_MALLOC
29 #pragma alloca
30 #endif
32 #undef _GNU_SOURCE
33 #define _GNU_SOURCE
35 #ifndef INSIDE_RECURSION
36 # ifdef HAVE_CONFIG_H
37 # include <config.h>
38 # endif
39 #endif
41 #include <ansidecl.h>
43 #ifndef PARAMS
44 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
45 # define PARAMS(args) args
46 # else
47 # define PARAMS(args) ()
51 #ifndef INSIDE_RECURSION
53 # if defined STDC_HEADERS && !defined emacs
54 # include <stddef.h>
55 # else
56 /* We need this for `regex.h', and perhaps for the Emacs include files. */
57 # include <sys/types.h>
58 # endif
60 # define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
62 /* For platform which support the ISO C amendement 1 functionality we
63 support user defined character classes. */
64 # if defined _LIBC || WIDE_CHAR_SUPPORT
65 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
66 # include <wchar.h>
67 # include <wctype.h>
68 # endif
70 # ifdef _LIBC
71 /* We have to keep the namespace clean. */
72 # define regfree(preg) __regfree (preg)
73 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
74 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
75 # define regerror(errcode, preg, errbuf, errbuf_size) \
76 __regerror(errcode, preg, errbuf, errbuf_size)
77 # define re_set_registers(bu, re, nu, st, en) \
78 __re_set_registers (bu, re, nu, st, en)
79 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
80 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
81 # define re_match(bufp, string, size, pos, regs) \
82 __re_match (bufp, string, size, pos, regs)
83 # define re_search(bufp, string, size, startpos, range, regs) \
84 __re_search (bufp, string, size, startpos, range, regs)
85 # define re_compile_pattern(pattern, length, bufp) \
86 __re_compile_pattern (pattern, length, bufp)
87 # define re_set_syntax(syntax) __re_set_syntax (syntax)
88 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
89 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
90 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
92 # define btowc __btowc
94 /* We are also using some library internals. */
95 # include <locale/localeinfo.h>
96 # include <locale/elem-hash.h>
97 # include <langinfo.h>
98 # include <locale/coll-lookup.h>
99 # endif
101 /* This is for other GNU distributions with internationalized messages. */
102 # if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
103 # include <libintl.h>
104 # ifdef _LIBC
105 # undef gettext
107 # endif
108 # else
109 # define gettext(msgid) (msgid)
110 # endif
112 # ifndef gettext_noop
113 /* This define is so xgettext can find the internationalizable
114 strings. */
115 # define gettext_noop(String) String
116 # endif
118 /* The `emacs' switch turns on certain matching commands
119 that make sense only in Emacs. */
120 # ifdef emacs
128 /* If we are not linking with Emacs proper,
129 we can't use the relocating allocator
130 even if config.h says that we can. */
131 # undef REL_ALLOC
133 # if defined STDC_HEADERS || defined _LIBC
134 # include <stdlib.h>
135 # else
138 # endif
140 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
141 If nothing else has been done, use the method below. */
142 # ifdef INHIBIT_STRING_HEADER
143 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
144 # if !defined bzero && !defined bcopy
145 # undef INHIBIT_STRING_HEADER
146 # endif
147 # endif
148 # endif
150 /* This is the normal way of making sure we have a bcopy and a bzero.
151 This is used in most programs--a few other programs avoid this
152 by defining INHIBIT_STRING_HEADER. */
153 # ifndef INHIBIT_STRING_HEADER
154 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
155 # include <string.h>
156 # ifndef bzero
157 # ifndef _LIBC
159 # else
160 # define bzero(s, n) __bzero (s, n)
161 # endif
162 # endif
163 # else
164 # include <strings.h>
165 # ifndef memcmp
166 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
167 # endif
168 # ifndef memcpy
169 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
170 # endif
171 # endif
172 # endif
174 /* Define the syntax stuff for \<, \>, etc. */
176 /* This must be nonzero for the wordchar and notwordchar pattern
177 commands in re_match_2. */
178 # ifndef Sword
179 # define Sword 1
180 # endif
182 # ifdef SWITCH_ENUM_BUG
183 # define SWITCH_ENUM_CAST(x) ((int)(x))
184 # else
185 # define SWITCH_ENUM_CAST(x) (x)
186 # endif
190 # if defined _LIBC || HAVE_LIMITS_H
191 # include <limits.h>
192 # endif
194 # ifndef MB_LEN_MAX
195 # define MB_LEN_MAX 1
196 # endif
197 \f
198 /* Get the interface, including the syntax bits. */
201 /* isalpha etc. are used for the character classes. */
202 # include <ctype.h>
204 /* Jim Meyering writes:
206 "... Some ctype macros are valid only for character codes that
207 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
208 using /bin/cc or gcc but without giving an ansi option). So, all
209 ctype uses should be through macros like ISPRINT... If
210 STDC_HEADERS is defined, then autoconf has verified that the ctype
211 macros don't need to be guarded with references to isascii. ...
212 Defining isascii to 1 should let any compiler worth its salt
213 eliminate the && through constant folding."
214 Solaris defines some of these symbols so we must undefine them first. */
216 # undef ISASCII
217 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
218 # define ISASCII(c) 1
219 # else
220 # define ISASCII(c) isascii(c)
221 # endif
223 # ifdef isblank
224 # define ISBLANK(c) (ISASCII (c) && isblank (c))
225 # else
227 # endif
228 # ifdef isgraph
229 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
230 # else
231 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
232 # endif
234 # undef ISPRINT
235 # define ISPRINT(c) (ISASCII (c) && isprint (c))
236 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
237 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
238 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
239 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
240 # define ISLOWER(c) (ISASCII (c) && islower (c))
241 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
242 # define ISSPACE(c) (ISASCII (c) && isspace (c))
243 # define ISUPPER(c) (ISASCII (c) && isupper (c))
244 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
246 # ifdef _tolower
247 # define TOLOWER(c) _tolower(c)
248 # else
249 # define TOLOWER(c) tolower(c)
250 # endif
252 # ifndef NULL
253 # define NULL (void *)0
254 # endif
256 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
257 since ours (we hope) works properly with all combinations of
258 machines, compilers, `char' and `unsigned char' argument types.
259 (Per Bothner suggested the basic approach.) */
260 # undef SIGN_EXTEND_CHAR
261 # if __STDC__
262 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
264 /* As in Harbison and Steele. */
265 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
266 # endif
267 \f
268 # ifndef emacs
269 /* How many characters in the character set. */
270 # define CHAR_SET_SIZE 256
272 # ifdef SYNTAX_TABLE
282 static void
284 {
299 }
303 # define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
306 \f
307 /* Integer type for pointers. */
308 # if !defined _LIBC && !defined HAVE_UINTPTR_T
310 # endif
312 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
313 use `alloca' instead of `malloc'. This is because using malloc in
314 re_search* or re_match* could cause memory leaks when C-g is used in
315 Emacs; also, malloc is slower and causes storage fragmentation. On
316 the other hand, malloc is more portable, and easier to debug.
318 Because we sometimes use alloca, some routines have to be macros,
319 not functions -- `alloca'-allocated space disappears at the end of the
320 function it is called in. */
322 # ifdef REGEX_MALLOC
324 # define REGEX_ALLOCATE malloc
325 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
326 # define REGEX_FREE free
330 /* Emacs already defines alloca, sometimes. */
331 # ifndef alloca
333 /* Make alloca work the best possible way. */
334 # ifdef __GNUC__
335 # define alloca __builtin_alloca
337 # if HAVE_ALLOCA_H
338 # include <alloca.h>
344 # define REGEX_ALLOCATE alloca
346 /* Assumes a `char *destination' variable. */
347 # define REGEX_REALLOCATE(source, osize, nsize) \
348 (destination = (char *) alloca (nsize), \
349 memcpy (destination, source, osize))
351 /* No need to do anything to free, after alloca. */
356 /* Define how to allocate the failure stack. */
358 # if defined REL_ALLOC && defined REGEX_MALLOC
360 # define REGEX_ALLOCATE_STACK(size) \
361 r_alloc (&failure_stack_ptr, (size))
362 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
363 r_re_alloc (&failure_stack_ptr, (nsize))
364 # define REGEX_FREE_STACK(ptr) \
365 r_alloc_free (&failure_stack_ptr)
369 # ifdef REGEX_MALLOC
371 # define REGEX_ALLOCATE_STACK malloc
372 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
373 # define REGEX_FREE_STACK free
377 # define REGEX_ALLOCATE_STACK alloca
379 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
380 REGEX_REALLOCATE (source, osize, nsize)
381 /* No need to explicitly free anything. */
382 # define REGEX_FREE_STACK(arg)
388 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
389 `string1' or just past its end. This works if PTR is NULL, which is
390 a good thing. */
391 # define FIRST_STRING_P(ptr) \
392 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
394 /* (Re)Allocate N items of type T using malloc, or fail. */
395 # define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
396 # define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
397 # define RETALLOC_IF(addr, n, t) \
398 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
399 # define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
403 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
405 # undef MAX
406 # undef MIN
407 # define MAX(a, b) ((a) > (b) ? (a) : (b))
408 # define MIN(a, b) ((a) < (b) ? (a) : (b))
411 # define false 0
412 # define true 1
415 reg_syntax_t syntax,
431 #ifdef MBS_SUPPORT
433 reg_syntax_t syntax,
452 #endif
453 \f
454 /* These are the command codes that appear in compiled regular
455 expressions. Some opcodes are followed by argument bytes. A
456 command code can specify any interpretation whatsoever for its
457 arguments. Zero bytes may appear in the compiled regular expression. */
460 {
463 /* Succeed right away--no more backtracking. */
464 succeed,
466 /* Followed by one byte giving n, then by n literal bytes. */
467 exactn,
469 # ifdef MBS_SUPPORT
470 /* Same as exactn, but contains binary data. */
471 exactn_bin,
472 # endif
474 /* Matches any (more or less) character. */
475 anychar,
477 /* Matches any one char belonging to specified set. First
478 following byte is number of bitmap bytes. Then come bytes
479 for a bitmap saying which chars are in. Bits in each byte
480 are ordered low-bit-first. A character is in the set if its
481 bit is 1. A character too large to have a bit in the map is
482 automatically not in the set. */
483 /* ifdef MBS_SUPPORT, following element is length of character
484 classes, length of collating symbols, length of equivalence
485 classes, length of character ranges, and length of characters.
486 Next, character class element, collating symbols elements,
487 equivalence class elements, range elements, and character
488 elements follow.
489 See regex_compile function. */
490 charset,
492 /* Same parameters as charset, but match any character that is
493 not one of those specified. */
494 charset_not,
496 /* Start remembering the text that is matched, for storing in a
497 register. Followed by one byte with the register number, in
498 the range 0 to one less than the pattern buffer's re_nsub
499 field. Then followed by one byte with the number of groups
500 inner to this one. (This last has to be part of the
501 start_memory only because we need it in the on_failure_jump
502 of re_match_2.) */
503 start_memory,
505 /* Stop remembering the text that is matched and store it in a
506 memory register. Followed by one byte with the register
507 number, in the range 0 to one less than `re_nsub' in the
508 pattern buffer, and one byte with the number of inner groups,
509 just like `start_memory'. (We need the number of inner
510 groups here because we don't have any easy way of finding the
511 corresponding start_memory when we're at a stop_memory.) */
512 stop_memory,
514 /* Match a duplicate of something remembered. Followed by one
515 byte containing the register number. */
516 duplicate,
518 /* Fail unless at beginning of line. */
519 begline,
521 /* Fail unless at end of line. */
522 endline,
524 /* Succeeds if at beginning of buffer (if emacs) or at beginning
525 of string to be matched (if not). */
526 begbuf,
528 /* Analogously, for end of buffer/string. */
529 endbuf,
531 /* Followed by two byte relative address to which to jump. */
532 jump,
534 /* Same as jump, but marks the end of an alternative. */
535 jump_past_alt,
537 /* Followed by two-byte relative address of place to resume at
538 in case of failure. */
539 /* ifdef MBS_SUPPORT, the size of address is 1. */
540 on_failure_jump,
542 /* Like on_failure_jump, but pushes a placeholder instead of the
543 current string position when executed. */
544 on_failure_keep_string_jump,
546 /* Throw away latest failure point and then jump to following
547 two-byte relative address. */
548 /* ifdef MBS_SUPPORT, the size of address is 1. */
549 pop_failure_jump,
551 /* Change to pop_failure_jump if know won't have to backtrack to
552 match; otherwise change to jump. This is used to jump
553 back to the beginning of a repeat. If what follows this jump
554 clearly won't match what the repeat does, such that we can be
555 sure that there is no use backtracking out of repetitions
556 already matched, then we change it to a pop_failure_jump.
557 Followed by two-byte address. */
558 /* ifdef MBS_SUPPORT, the size of address is 1. */
559 maybe_pop_jump,
561 /* Jump to following two-byte address, and push a dummy failure
562 point. This failure point will be thrown away if an attempt
563 is made to use it for a failure. A `+' construct makes this
564 before the first repeat. Also used as an intermediary kind
565 of jump when compiling an alternative. */
566 /* ifdef MBS_SUPPORT, the size of address is 1. */
567 dummy_failure_jump,
569 /* Push a dummy failure point and continue. Used at the end of
570 alternatives. */
571 push_dummy_failure,
573 /* Followed by two-byte relative address and two-byte number n.
574 After matching N times, jump to the address upon failure. */
575 /* ifdef MBS_SUPPORT, the size of address is 1. */
576 succeed_n,
578 /* Followed by two-byte relative address, and two-byte number n.
579 Jump to the address N times, then fail. */
580 /* ifdef MBS_SUPPORT, the size of address is 1. */
581 jump_n,
583 /* Set the following two-byte relative address to the
584 subsequent two-byte number. The address *includes* the two
585 bytes of number. */
586 /* ifdef MBS_SUPPORT, the size of address is 1. */
587 set_number_at,
596 notwordbound /* Succeeds if not at a word boundary. */
598 # ifdef emacs
603 /* Matches any character whose syntax is specified. Followed by
604 a byte which contains a syntax code, e.g., Sword. */
605 syntaxspec,
607 /* Matches any character whose syntax is not that specified. */
608 notsyntaxspec
612 \f
614 #ifdef BYTE
615 # define CHAR_T char
616 # define UCHAR_T unsigned char
617 # define COMPILED_BUFFER_VAR bufp->buffer
618 # define OFFSET_ADDRESS_SIZE 2
619 # if defined (__STDC__) || defined (ALMOST_STDC) || defined (HAVE_STRINGIZE)
620 # define PREFIX(name) byte_##name
621 # else
623 # endif
624 # define ARG_PREFIX(name) name
625 # define PUT_CHAR(c) putchar (c)
626 #else
627 # ifdef WCHAR
628 # define CHAR_T wchar_t
629 # define UCHAR_T wchar_t
630 # define COMPILED_BUFFER_VAR wc_buffer
632 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
633 # if defined (__STDC__) || defined (ALMOST_STDC) || defined (HAVE_STRINGIZE)
634 # define PREFIX(name) wcs_##name
635 # define ARG_PREFIX(name) c##name
636 # else
639 # endif
640 /* Should we use wide stream?? */
642 # define TRUE 1
643 # define FALSE 0
644 # else
645 # ifdef MBS_SUPPORT
646 # define WCHAR
647 # define INSIDE_RECURSION
649 # undef INSIDE_RECURSION
650 # endif
651 # define BYTE
652 # define INSIDE_RECURSION
654 # undef INSIDE_RECURSION
655 # endif
656 #endif
658 #ifdef INSIDE_RECURSION
659 /* Common operations on the compiled pattern. */
661 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
662 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
664 # ifdef WCHAR
665 # define STORE_NUMBER(destination, number) \
666 do { \
667 *(destination) = (UCHAR_T)(number); \
668 } while (0)
670 # define STORE_NUMBER(destination, number) \
671 do { \
672 (destination)[0] = (number) & 0377; \
673 (destination)[1] = (number) >> 8; \
674 } while (0)
677 /* Same as STORE_NUMBER, except increment DESTINATION to
678 the byte after where the number is stored. Therefore, DESTINATION
679 must be an lvalue. */
680 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
682 # define STORE_NUMBER_AND_INCR(destination, number) \
683 do { \
684 STORE_NUMBER (destination, number); \
685 (destination) += OFFSET_ADDRESS_SIZE; \
686 } while (0)
688 /* Put into DESTINATION a number stored in two contiguous bytes starting
689 at SOURCE. */
690 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
692 # ifdef WCHAR
693 # define EXTRACT_NUMBER(destination, source) \
694 do { \
695 (destination) = *(source); \
696 } while (0)
698 # define EXTRACT_NUMBER(destination, source) \
699 do { \
700 (destination) = *(source) & 0377; \
701 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
702 } while (0)
703 # endif
705 # ifdef DEBUG
707 static void
711 {
712 # ifdef WCHAR
718 # endif
719 }
722 # undef EXTRACT_NUMBER
723 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
728 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
729 SOURCE must be an lvalue. */
731 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
732 do { \
733 EXTRACT_NUMBER (destination, source); \
734 (source) += OFFSET_ADDRESS_SIZE; \
735 } while (0)
737 # ifdef DEBUG
740 static void
744 {
747 }
749 # ifndef EXTRACT_MACROS
750 # undef EXTRACT_NUMBER_AND_INCR
751 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
752 PREFIX(extract_number_and_incr) (&dest, &src)
757 \f
759 /* If DEBUG is defined, Regex prints many voluminous messages about what
760 it is doing (if the variable `debug' is nonzero). If linked with the
761 main program in `iregex.c', you can enter patterns and strings
762 interactively. And if linked with the main program in `main.c' and
763 the other test files, you can run the already-written tests. */
765 # ifdef DEBUG
767 # ifndef DEFINED_ONCE
769 /* We use standard I/O for debugging. */
770 # include <stdio.h>
772 /* It is useful to test things that ``must'' be true when debugging. */
773 # include <assert.h>
777 # define DEBUG_STATEMENT(e) e
778 # define DEBUG_PRINT1(x) if (debug) printf (x)
779 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
780 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
781 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
784 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
785 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
786 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
787 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
790 /* Print the fastmap in human-readable form. */
792 # ifndef DEFINED_ONCE
793 void
796 {
801 {
803 {
807 {
809 i++;
810 }
812 {
815 }
816 }
817 }
819 }
823 /* Print a compiled pattern string in human-readable form, starting at
824 the START pointer into it and ending just before the pointer END. */
826 void
830 {
837 {
840 }
842 /* Loop over pattern commands. */
844 {
845 # ifdef _LIBC
847 # else
849 # endif
852 {
860 do
861 {
864 }
868 # ifdef MBS_SUPPORT
872 do
873 {
875 }
900 {
901 # ifdef WCHAR
912 {
918 }
921 {
927 }
930 {
934 }
939 # else
951 {
952 /* Are we starting a range? */
954 {
957 }
958 /* Have we broken a range? */
960 {
963 }
969 }
978 }
991 # ifdef _LIBC
993 # else
995 # endif
1000 # ifdef _LIBC
1002 # else
1005 # endif
1010 # ifdef _LIBC
1012 # else
1014 # endif
1023 # ifdef _LIBC
1025 # else
1027 # endif
1032 # ifdef _LIBC
1034 # else
1036 # endif
1041 # ifdef _LIBC
1043 # else
1045 # endif
1050 # ifdef _LIBC
1052 # else
1054 # endif
1061 # ifdef _LIBC
1063 # else
1066 # endif
1080 # ifdef _LIBC
1082 # else
1085 # endif
1104 # ifdef emacs
1148 }
1151 }
1153 # ifdef _LIBC
1155 # else
1157 # endif
1158 }
1161 void
1164 {
1173 {
1176 }
1178 # ifdef _LIBC
1180 # else
1182 # endif
1190 /* Perhaps we should print the translate table? */
1191 }
1194 void
1201 {
1206 else
1207 {
1211 {
1216 }
1220 {
1223 {
1226 }
1227 }
1228 }
1229 }
1231 # ifndef DEFINED_ONCE
1232 void
1235 {
1237 }
1238 # endif
1242 # ifndef DEFINED_ONCE
1243 # undef assert
1244 # define assert(e)
1246 # define DEBUG_STATEMENT(e)
1247 # define DEBUG_PRINT1(x)
1248 # define DEBUG_PRINT2(x1, x2)
1249 # define DEBUG_PRINT3(x1, x2, x3)
1250 # define DEBUG_PRINT4(x1, x2, x3, x4)
1252 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1253 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1257 \f
1259 # ifdef WCHAR
1260 /* This convert a multibyte string to a wide character string.
1261 And write their correspondances to offset_buffer(see below)
1262 and write whether each wchar_t is binary data to is_binary.
1263 This assume invalid multibyte sequences as binary data.
1264 We assume offset_buffer and is_binary is already allocated
1265 enough space. */
1270 static size_t
1276 /* It hold correspondances between src(char string) and
1277 dest(wchar_t string) for optimization.
1278 e.g. src = "xxxyzz"
1279 dest = {'X', 'Y', 'Z'}
1280 (each "xxx", "y" and "zz" represent one multibyte character
1281 corresponding to 'X', 'Y' and 'Z'.)
1282 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1283 = {0, 3, 4, 6}
1284 */
1287 {
1297 /* Initialize the conversion state. */
1303 {
1304 #ifdef _LIBC
1306 #else
1308 #endif
1311 /* failed to convert. maybe src contains binary data.
1312 So we consume 1 byte manualy. */
1313 {
1317 }
1318 else
1320 /* In sjis encoding, we use yen sign as escape character in
1321 place of reverse solidus. So we convert 0x5c(yen sign in
1322 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1323 solidus in UCS2). */
1328 }
1330 /* Fill remain of the buffer with sentinel. */
1335 }
1341 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1342 also be assigned to arbitrarily: each pattern buffer stores its own
1343 syntax, so it can be changed between regex compilations. */
1344 /* This has no initializer because initialized variables in Emacs
1345 become read-only after dumping. */
1346 reg_syntax_t re_syntax_options;
1349 /* Specify the precise syntax of regexps for compilation. This provides
1350 for compatibility for various utilities which historically have
1351 different, incompatible syntaxes.
1353 The argument SYNTAX is a bit mask comprised of the various bits
1354 defined in regex.h. We return the old syntax. */
1356 reg_syntax_t
1358 reg_syntax_t syntax;
1359 {
1363 # ifdef DEBUG
1370 }
1371 # ifdef _LIBC
1373 # endif
1374 \f
1375 /* This table gives an error message for each of the error codes listed
1376 in regex.h. Obviously the order here has to be same as there.
1377 POSIX doesn't require that we do anything for REG_NOERROR,
1378 but why not be nice? */
1381 {
1399 };
1400 \f
1403 #ifndef DEFINED_ONCE
1404 /* Avoiding alloca during matching, to placate r_alloc. */
1406 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1407 searching and matching functions should not call alloca. On some
1408 systems, alloca is implemented in terms of malloc, and if we're
1409 using the relocating allocator routines, then malloc could cause a
1410 relocation, which might (if the strings being searched are in the
1411 ralloc heap) shift the data out from underneath the regexp
1412 routines.
1414 Here's another reason to avoid allocation: Emacs
1415 processes input from X in a signal handler; processing X input may
1416 call malloc; if input arrives while a matching routine is calling
1417 malloc, then we're scrod. But Emacs can't just block input while
1418 calling matching routines; then we don't notice interrupts when
1419 they come in. So, Emacs blocks input around all regexp calls
1420 except the matching calls, which it leaves unprotected, in the
1421 faith that they will not malloc. */
1423 /* Normally, this is fine. */
1424 # define MATCH_MAY_ALLOCATE
1426 /* When using GNU C, we are not REALLY using the C alloca, no matter
1427 what config.h may say. So don't take precautions for it. */
1428 # ifdef __GNUC__
1429 # undef C_ALLOCA
1430 # endif
1432 /* The match routines may not allocate if (1) they would do it with malloc
1433 and (2) it's not safe for them to use malloc.
1434 Note that if REL_ALLOC is defined, matching would not use malloc for the
1435 failure stack, but we would still use it for the register vectors;
1436 so REL_ALLOC should not affect this. */
1437 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1438 # undef MATCH_MAY_ALLOCATE
1439 # endif
1441 \f
1442 #ifdef INSIDE_RECURSION
1443 /* Failure stack declarations and macros; both re_compile_fastmap and
1444 re_match_2 use a failure stack. These have to be macros because of
1445 REGEX_ALLOCATE_STACK. */
1448 /* Number of failure points for which to initially allocate space
1449 when matching. If this number is exceeded, we allocate more
1450 space, so it is not a hard limit. */
1451 # ifndef INIT_FAILURE_ALLOC
1452 # define INIT_FAILURE_ALLOC 5
1453 # endif
1455 /* Roughly the maximum number of failure points on the stack. Would be
1456 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1457 This is a variable only so users of regex can assign to it; we never
1458 change it ourselves. */
1460 # ifdef INT_IS_16BIT
1462 # ifndef DEFINED_ONCE
1463 # if defined MATCH_MAY_ALLOCATE
1464 /* 4400 was enough to cause a crash on Alpha OSF/1,
1465 whose default stack limit is 2mb. */
1467 # else
1469 # endif
1470 # endif
1473 {
1476 };
1481 {
1489 # ifndef DEFINED_ONCE
1490 # if defined MATCH_MAY_ALLOCATE
1491 /* 4400 was enough to cause a crash on Alpha OSF/1,
1492 whose default stack limit is 2mb. */
1494 # else
1496 # endif
1497 # endif
1500 {
1503 };
1508 {
1516 # ifndef DEFINED_ONCE
1517 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1518 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1519 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1520 # endif
1523 /* Define macros to initialize and free the failure stack.
1524 Do `return -2' if the alloc fails. */
1526 # ifdef MATCH_MAY_ALLOCATE
1527 # define INIT_FAIL_STACK() \
1528 do { \
1529 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1530 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1531 \
1532 if (fail_stack.stack == NULL) \
1533 return -2; \
1534 \
1535 fail_stack.size = INIT_FAILURE_ALLOC; \
1536 fail_stack.avail = 0; \
1537 } while (0)
1539 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1540 # else
1541 # define INIT_FAIL_STACK() \
1542 do { \
1543 fail_stack.avail = 0; \
1544 } while (0)
1546 # define RESET_FAIL_STACK()
1547 # endif
1550 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1552 Return 1 if succeeds, and 0 if either ran out of memory
1553 allocating space for it or it was already too large.
1555 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1557 # define DOUBLE_FAIL_STACK(fail_stack) \
1558 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1559 ? 0 \
1560 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1561 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1562 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1563 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1564 \
1565 (fail_stack).stack == NULL \
1566 ? 0 \
1567 : ((fail_stack).size <<= 1, \
1568 1)))
1571 /* Push pointer POINTER on FAIL_STACK.
1572 Return 1 if was able to do so and 0 if ran out of memory allocating
1573 space to do so. */
1574 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1575 ((FAIL_STACK_FULL () \
1576 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1577 ? 0 \
1578 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1579 1))
1581 /* Push a pointer value onto the failure stack.
1582 Assumes the variable `fail_stack'. Probably should only
1583 be called from within `PUSH_FAILURE_POINT'. */
1584 # define PUSH_FAILURE_POINTER(item) \
1585 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1587 /* This pushes an integer-valued item onto the failure stack.
1588 Assumes the variable `fail_stack'. Probably should only
1589 be called from within `PUSH_FAILURE_POINT'. */
1590 # define PUSH_FAILURE_INT(item) \
1591 fail_stack.stack[fail_stack.avail++].integer = (item)
1593 /* Push a fail_stack_elt_t value onto the failure stack.
1594 Assumes the variable `fail_stack'. Probably should only
1595 be called from within `PUSH_FAILURE_POINT'. */
1596 # define PUSH_FAILURE_ELT(item) \
1597 fail_stack.stack[fail_stack.avail++] = (item)
1599 /* These three POP... operations complement the three PUSH... operations.
1600 All assume that `fail_stack' is nonempty. */
1601 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1602 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1603 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1605 /* Used to omit pushing failure point id's when we're not debugging. */
1606 # ifdef DEBUG
1607 # define DEBUG_PUSH PUSH_FAILURE_INT
1608 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1609 # else
1610 # define DEBUG_PUSH(item)
1611 # define DEBUG_POP(item_addr)
1612 # endif
1615 /* Push the information about the state we will need
1616 if we ever fail back to it.
1618 Requires variables fail_stack, regstart, regend, reg_info, and
1619 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1620 be declared.
1622 Does `return FAILURE_CODE' if runs out of memory. */
1624 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1625 do { \
1626 char *destination; \
1627 /* Must be int, so when we don't save any registers, the arithmetic \
1629 /* Can't be int, since there is not a shred of a guarantee that int \
1630 is wide enough to hold a value of something to which pointer can \
1632 active_reg_t this_reg; \
1633 \
1634 DEBUG_STATEMENT (failure_id++); \
1635 DEBUG_STATEMENT (nfailure_points_pushed++); \
1639 \
1642 \
1644 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1645 { \
1646 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1647 return failure_code; \
1648 \
1650 (fail_stack).size); \
1652 } \
1653 \
1656 \
1657 if (1) \
1658 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1659 this_reg++) \
1660 { \
1662 DEBUG_STATEMENT (num_regs_pushed++); \
1663 \
1665 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1666 \
1668 PUSH_FAILURE_POINTER (regend[this_reg]); \
1669 \
1671 reg_info[this_reg].word.pointer); \
1673 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1676 MATCHED_SOMETHING (reg_info[this_reg])); \
1678 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1680 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1681 } \
1682 \
1684 PUSH_FAILURE_INT (lowest_active_reg); \
1685 \
1687 PUSH_FAILURE_INT (highest_active_reg); \
1688 \
1690 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1691 PUSH_FAILURE_POINTER (pattern_place); \
1692 \
1694 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1695 size2); \
1697 PUSH_FAILURE_POINTER (string_place); \
1698 \
1700 DEBUG_PUSH (failure_id); \
1701 } while (0)
1703 # ifndef DEFINED_ONCE
1704 /* This is the number of items that are pushed and popped on the stack
1705 for each register. */
1706 # define NUM_REG_ITEMS 3
1708 /* Individual items aside from the registers. */
1709 # ifdef DEBUG
1711 # else
1712 # define NUM_NONREG_ITEMS 4
1713 # endif
1715 /* We push at most this many items on the stack. */
1716 /* We used to use (num_regs - 1), which is the number of registers
1717 this regexp will save; but that was changed to 5
1718 to avoid stack overflow for a regexp with lots of parens. */
1719 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1721 /* We actually push this many items. */
1722 # define NUM_FAILURE_ITEMS \
1723 (((0 \
1724 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1725 * NUM_REG_ITEMS) \
1726 + NUM_NONREG_ITEMS)
1728 /* How many items can still be added to the stack without overflowing it. */
1729 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1733 /* Pops what PUSH_FAIL_STACK pushes.
1735 We restore into the parameters, all of which should be lvalues:
1736 STR -- the saved data position.
1737 PAT -- the saved pattern position.
1738 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1739 REGSTART, REGEND -- arrays of string positions.
1740 REG_INFO -- array of information about each subexpression.
1742 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1743 `pend', `string1', `size1', `string2', and `size2'. */
1744 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1745 { \
1746 DEBUG_STATEMENT (unsigned failure_id;) \
1747 active_reg_t this_reg; \
1748 const UCHAR_T *string_temp; \
1749 \
1750 assert (!FAIL_STACK_EMPTY ()); \
1751 \
1756 \
1757 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1758 \
1759 DEBUG_POP (&failure_id); \
1761 \
1762 /* If the saved string location is NULL, it came from an \
1763 on_failure_keep_string_jump opcode, and we want to throw away the \
1765 string_temp = POP_FAILURE_POINTER (); \
1766 if (string_temp != NULL) \
1767 str = (const CHAR_T *) string_temp; \
1768 \
1770 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1772 \
1773 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1775 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1776 \
1778 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1780 \
1781 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1783 \
1784 if (1) \
1785 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1786 { \
1788 \
1789 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1791 reg_info[this_reg].word.pointer); \
1792 \
1793 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1795 \
1796 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1798 } \
1799 else \
1800 { \
1801 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1802 { \
1803 reg_info[this_reg].word.integer = 0; \
1804 regend[this_reg] = 0; \
1805 regstart[this_reg] = 0; \
1806 } \
1807 highest_active_reg = high_reg; \
1808 } \
1809 \
1810 set_regs_matched_done = 0; \
1811 DEBUG_STATEMENT (nfailure_points_popped++); \
1813 \f
1814 /* Structure for per-register (a.k.a. per-group) information.
1815 Other register information, such as the
1816 starting and ending positions (which are addresses), and the list of
1817 inner groups (which is a bits list) are maintained in separate
1818 variables.
1820 We are making a (strictly speaking) nonportable assumption here: that
1821 the compiler will pack our bit fields into something that fits into
1822 the type of `word', i.e., is something that fits into one item on the
1823 failure stack. */
1826 /* Declarations and macros for re_match_2. */
1829 {
1831 struct
1832 {
1833 /* This field is one if this group can match the empty string,
1834 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1835 # define MATCH_NULL_UNSET_VALUE 3
1843 # ifndef DEFINED_ONCE
1844 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1845 # define IS_ACTIVE(R) ((R).bits.is_active)
1846 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1847 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1850 /* Call this when have matched a real character; it sets `matched' flags
1851 for the subexpressions which we are currently inside. Also records
1852 that those subexprs have matched. */
1853 # define SET_REGS_MATCHED() \
1854 do \
1855 { \
1856 if (!set_regs_matched_done) \
1857 { \
1858 active_reg_t r; \
1859 set_regs_matched_done = 1; \
1860 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1861 { \
1862 MATCHED_SOMETHING (reg_info[r]) \
1863 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1864 = 1; \
1865 } \
1866 } \
1867 } \
1868 while (0)
1871 /* Registers are set to a sentinel when they haven't yet matched. */
1873 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1874 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1876 /* Subroutine declarations and macros for regex_compile. */
1886 reg_syntax_t syntax));
1889 reg_syntax_t syntax));
1890 # ifdef WCHAR
1895 reg_syntax_t syntax,
1904 reg_syntax_t syntax,
1908 /* Fetch the next character in the uncompiled pattern---translating it
1909 if necessary. Also cast from a signed character in the constant
1910 string passed to us by the user to an unsigned char that we can use
1911 as an array index (in, e.g., `translate'). */
1912 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1913 because it is impossible to allocate 4GB array for some encodings
1914 which have 4 byte character_set like UCS4. */
1915 # ifndef PATFETCH
1916 # ifdef WCHAR
1917 # define PATFETCH(c) \
1918 do {if (p == pend) return REG_EEND; \
1919 c = (UCHAR_T) *p++; \
1920 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1921 } while (0)
1923 # define PATFETCH(c) \
1924 do {if (p == pend) return REG_EEND; \
1925 c = (unsigned char) *p++; \
1926 if (translate) c = (unsigned char) translate[c]; \
1927 } while (0)
1929 # endif
1931 /* Fetch the next character in the uncompiled pattern, with no
1932 translation. */
1933 # define PATFETCH_RAW(c) \
1934 do {if (p == pend) return REG_EEND; \
1935 c = (UCHAR_T) *p++; \
1936 } while (0)
1938 /* Go backwards one character in the pattern. */
1939 # define PATUNFETCH p--
1942 /* If `translate' is non-null, return translate[D], else just D. We
1943 cast the subscript to translate because some data is declared as
1944 `char *', to avoid warnings when a string constant is passed. But
1945 when we use a character as a subscript we must make it unsigned. */
1946 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1947 because it is impossible to allocate 4GB array for some encodings
1948 which have 4 byte character_set like UCS4. */
1950 # ifndef TRANSLATE
1951 # ifdef WCHAR
1952 # define TRANSLATE(d) \
1953 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1954 ? (char) translate[(unsigned char) (d)] : (d))
1956 # define TRANSLATE(d) \
1957 (translate ? (char) translate[(unsigned char) (d)] : (d))
1959 # endif
1962 /* Macros for outputting the compiled pattern into `buffer'. */
1964 /* If the buffer isn't allocated when it comes in, use this. */
1965 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1967 /* Make sure we have at least N more bytes of space in buffer. */
1968 # ifdef WCHAR
1969 # define GET_BUFFER_SPACE(n) \
1970 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
1971 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
1972 EXTEND_BUFFER ()
1974 # define GET_BUFFER_SPACE(n) \
1975 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1976 EXTEND_BUFFER ()
1979 /* Make sure we have one more byte of buffer space and then add C to it. */
1980 # define BUF_PUSH(c) \
1981 do { \
1982 GET_BUFFER_SPACE (1); \
1983 *b++ = (UCHAR_T) (c); \
1984 } while (0)
1987 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1988 # define BUF_PUSH_2(c1, c2) \
1989 do { \
1990 GET_BUFFER_SPACE (2); \
1991 *b++ = (UCHAR_T) (c1); \
1992 *b++ = (UCHAR_T) (c2); \
1993 } while (0)
1996 /* As with BUF_PUSH_2, except for three bytes. */
1997 # define BUF_PUSH_3(c1, c2, c3) \
1998 do { \
1999 GET_BUFFER_SPACE (3); \
2000 *b++ = (UCHAR_T) (c1); \
2001 *b++ = (UCHAR_T) (c2); \
2002 *b++ = (UCHAR_T) (c3); \
2003 } while (0)
2005 /* Store a jump with opcode OP at LOC to location TO. We store a
2006 relative address offset by the three bytes the jump itself occupies. */
2007 # define STORE_JUMP(op, loc, to) \
2008 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
2010 /* Likewise, for a two-argument jump. */
2011 # define STORE_JUMP2(op, loc, to, arg) \
2012 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
2014 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
2015 # define INSERT_JUMP(op, loc, to) \
2016 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
2018 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
2019 # define INSERT_JUMP2(op, loc, to, arg) \
2020 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
2021 arg, b)
2023 /* This is not an arbitrary limit: the arguments which represent offsets
2024 into the pattern are two bytes long. So if 2^16 bytes turns out to
2025 be too small, many things would have to change. */
2026 /* Any other compiler which, like MSC, has allocation limit below 2^16
2027 bytes will have to use approach similar to what was done below for
2028 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
2029 reallocating to 0 bytes. Such thing is not going to work too well.
2030 You have been warned!! */
2031 # ifndef DEFINED_ONCE
2032 # if defined _MSC_VER && !defined WIN32
2033 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2034 The REALLOC define eliminates a flurry of conversion warnings,
2035 but is not required. */
2036 # define MAX_BUF_SIZE 65500L
2037 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2038 # else
2039 # define MAX_BUF_SIZE (1L << 16)
2040 # define REALLOC(p,s) realloc ((p), (s))
2041 # endif
2043 /* Extend the buffer by twice its current size via realloc and
2044 reset the pointers that pointed into the old block to point to the
2045 correct places in the new one. If extending the buffer results in it
2046 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2047 # if __BOUNDED_POINTERS__
2048 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2049 # define MOVE_BUFFER_POINTER(P) \
2050 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2051 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2052 else \
2053 { \
2054 SET_HIGH_BOUND (b); \
2055 SET_HIGH_BOUND (begalt); \
2056 if (fixup_alt_jump) \
2057 SET_HIGH_BOUND (fixup_alt_jump); \
2058 if (laststart) \
2059 SET_HIGH_BOUND (laststart); \
2060 if (pending_exact) \
2061 SET_HIGH_BOUND (pending_exact); \
2062 }
2063 # else
2064 # define MOVE_BUFFER_POINTER(P) (P) += incr
2065 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2066 # endif
2069 # ifdef WCHAR
2070 # define EXTEND_BUFFER() \
2071 do { \
2072 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2073 int wchar_count; \
2074 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2075 return REG_ESIZE; \
2076 bufp->allocated <<= 1; \
2077 if (bufp->allocated > MAX_BUF_SIZE) \
2078 bufp->allocated = MAX_BUF_SIZE; \
2080 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2081 if (wchar_count == 0) wchar_count = 1; \
2083 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2084 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2085 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2086 if (COMPILED_BUFFER_VAR == NULL) \
2087 return REG_ESPACE; \
2089 if (old_buffer != COMPILED_BUFFER_VAR) \
2090 { \
2091 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2092 MOVE_BUFFER_POINTER (b); \
2093 MOVE_BUFFER_POINTER (begalt); \
2094 if (fixup_alt_jump) \
2095 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2096 if (laststart) \
2097 MOVE_BUFFER_POINTER (laststart); \
2098 if (pending_exact) \
2099 MOVE_BUFFER_POINTER (pending_exact); \
2100 } \
2101 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2102 } while (0)
2104 # define EXTEND_BUFFER() \
2105 do { \
2106 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2107 if (bufp->allocated == MAX_BUF_SIZE) \
2108 return REG_ESIZE; \
2109 bufp->allocated <<= 1; \
2110 if (bufp->allocated > MAX_BUF_SIZE) \
2111 bufp->allocated = MAX_BUF_SIZE; \
2112 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2113 bufp->allocated); \
2114 if (COMPILED_BUFFER_VAR == NULL) \
2115 return REG_ESPACE; \
2117 if (old_buffer != COMPILED_BUFFER_VAR) \
2118 { \
2119 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2120 MOVE_BUFFER_POINTER (b); \
2121 MOVE_BUFFER_POINTER (begalt); \
2122 if (fixup_alt_jump) \
2123 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2124 if (laststart) \
2125 MOVE_BUFFER_POINTER (laststart); \
2126 if (pending_exact) \
2127 MOVE_BUFFER_POINTER (pending_exact); \
2128 } \
2129 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2130 } while (0)
2133 # ifndef DEFINED_ONCE
2134 /* Since we have one byte reserved for the register number argument to
2135 {start,stop}_memory, the maximum number of groups we can report
2136 things about is what fits in that byte. */
2137 # define MAX_REGNUM 255
2139 /* But patterns can have more than `MAX_REGNUM' registers. We just
2140 ignore the excess. */
2144 /* Macros for the compile stack. */
2146 /* Since offsets can go either forwards or backwards, this type needs to
2147 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2148 /* int may be not enough when sizeof(int) == 2. */
2152 {
2153 pattern_offset_t begalt_offset;
2154 pattern_offset_t fixup_alt_jump;
2155 pattern_offset_t inner_group_offset;
2156 pattern_offset_t laststart_offset;
2157 regnum_t regnum;
2162 {
2169 # define INIT_COMPILE_STACK_SIZE 32
2171 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2172 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2174 /* The next available element. */
2175 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2179 /* Set the bit for character C in a list. */
2180 # ifndef DEFINED_ONCE
2181 # define SET_LIST_BIT(c) \
2182 (b[((unsigned char) (c)) / BYTEWIDTH] \
2183 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2186 /* Get the next unsigned number in the uncompiled pattern. */
2187 # define GET_UNSIGNED_NUMBER(num) \
2188 { \
2189 while (p != pend) \
2190 { \
2191 PATFETCH (c); \
2193 break; \
2194 if (num <= RE_DUP_MAX) \
2195 { \
2196 if (num < 0) \
2197 num = 0; \
2199 } \
2200 } \
2201 }
2203 # ifndef DEFINED_ONCE
2204 # if defined _LIBC || WIDE_CHAR_SUPPORT
2205 /* The GNU C library provides support for user-defined character classes
2206 and the functions from ISO C amendement 1. */
2207 # ifdef CHARCLASS_NAME_MAX
2208 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2209 # else
2210 /* This shouldn't happen but some implementation might still have this
2211 problem. Use a reasonable default value. */
2212 # define CHAR_CLASS_MAX_LENGTH 256
2213 # endif
2215 # ifdef _LIBC
2216 # define IS_CHAR_CLASS(string) __wctype (string)
2217 # else
2218 # define IS_CHAR_CLASS(string) wctype (string)
2219 # endif
2220 # else
2223 # define IS_CHAR_CLASS(string) \
2230 # endif
2232 \f
2233 # ifndef MATCH_MAY_ALLOCATE
2235 /* If we cannot allocate large objects within re_match_2_internal,
2236 we make the fail stack and register vectors global.
2237 The fail stack, we grow to the maximum size when a regexp
2238 is compiled.
2239 The register vectors, we adjust in size each time we
2240 compile a regexp, according to the number of registers it needs. */
2244 /* Size with which the following vectors are currently allocated.
2245 That is so we can make them bigger as needed,
2246 but never make them smaller. */
2247 # ifdef DEFINED_ONCE
2259 /* Make the register vectors big enough for NUM_REGS registers,
2260 but don't make them smaller. */
2262 static void
2265 {
2267 {
2279 }
2280 }
2283 \f
2284 # ifndef DEFINED_ONCE
2286 compile_stack,
2287 regnum_t regnum));
2290 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2291 Returns one of error codes defined in `regex.h', or zero for success.
2293 Assumes the `allocated' (and perhaps `buffer') and `translate'
2294 fields are set in BUFP on entry.
2296 If it succeeds, results are put in BUFP (if it returns an error, the
2297 contents of BUFP are undefined):
2298 `buffer' is the compiled pattern;
2299 `syntax' is set to SYNTAX;
2300 `used' is set to the length of the compiled pattern;
2301 `fastmap_accurate' is zero;
2302 `re_nsub' is the number of subexpressions in PATTERN;
2303 `not_bol' and `not_eol' are zero;
2305 The `fastmap' and `newline_anchor' fields are neither
2306 examined nor set. */
2308 /* Return, freeing storage we allocated. */
2309 # ifdef WCHAR
2310 # define FREE_STACK_RETURN(value) \
2311 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2312 # else
2313 # define FREE_STACK_RETURN(value) \
2314 return (free (compile_stack.stack), value)
2317 static reg_errcode_t
2321 reg_syntax_t syntax;
2323 {
2324 /* We fetch characters from PATTERN here. Even though PATTERN is
2325 `char *' (i.e., signed), we declare these variables as unsigned, so
2326 they can be reliably used as array indices. */
2329 #ifdef WCHAR
2330 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2333 /* offset buffer for optimization. See convert_mbs_to_wc. */
2335 /* It hold whether each wchar_t is binary data or not. */
2337 /* A flag whether exactn is handling binary data or not. */
2341 /* A random temporary spot in PATTERN. */
2344 /* Points to the end of the buffer, where we should append. */
2347 /* Keeps track of unclosed groups. */
2348 compile_stack_type compile_stack;
2350 /* Points to the current (ending) position in the pattern. */
2351 #ifdef WCHAR
2359 /* How to translate the characters in the pattern. */
2362 /* Address of the count-byte of the most recently inserted `exactn'
2363 command. This makes it possible to tell if a new exact-match
2364 character can be added to that command or if the character requires
2365 a new `exactn' command. */
2368 /* Address of start of the most recently finished expression.
2369 This tells, e.g., postfix * where to find the start of its
2370 operand. Reset at the beginning of groups and alternatives. */
2373 /* Address of beginning of regexp, or inside of last group. */
2376 /* Address of the place where a forward jump should go to the end of
2377 the containing expression. Each alternative of an `or' -- except the
2378 last -- ends with a forward jump of this sort. */
2381 /* Counts open-groups as they are encountered. Remembered for the
2382 matching close-group on the compile stack, so the same register
2383 number is put in the stop_memory as the start_memory. */
2386 #ifdef WCHAR
2387 /* Initialize the wchar_t PATTERN and offset_buffer. */
2392 {
2397 }
2402 {
2407 }
2408 #endif
2410 #ifdef DEBUG
2413 {
2419 }
2422 /* Initialize the compile stack. */
2425 {
2426 #ifdef WCHAR
2430 #endif
2432 }
2437 /* Initialize the pattern buffer. */
2442 /* Set `used' to zero, so that if we return an error, the pattern
2443 printer (for debugging) will think there's no pattern. We reset it
2444 at the end. */
2447 /* Always count groups, whether or not bufp->no_sub is set. */
2450 #if !defined emacs && !defined SYNTAX_TABLE
2451 /* Initialize the syntax table. */
2453 #endif
2456 {
2459 enough space. This loses if buffer's address is bogus, but
2460 that is the user's responsibility. */
2461 #ifdef WCHAR
2462 /* Free bufp->buffer and allocate an array for wchar_t pattern
2463 buffer. */
2466 UCHAR_T);
2467 #else
2470 }
2471 else
2474 UCHAR_T);
2475 }
2478 #ifdef WCHAR
2482 }
2483 #ifdef WCHAR
2484 else
2486 #endif
2490 /* Loop through the uncompiled pattern until we're at the end. */
2492 {
2496 {
2498 {
2501 /* If context independent, it's an operator. */
2503 /* Otherwise, depends on what's come before. */
2506 else
2508 }
2513 {
2515 p == pend
2516 /* If context independent, it's an operator. */
2518 /* Otherwise, depends on what's next. */
2521 else
2523 }
2532 handle_plus:
2534 /* If there is no previous pattern... */
2536 {
2541 }
2543 {
2544 /* Are we optimizing this jump? */
2547 /* 1 means zero (many) matches is allowed. */
2550 /* If there is a sequence of repetition chars, collapse it
2551 down to just one (the right one). We can't combine
2552 interval operators with these because of, e.g., `a{2}*',
2553 which should only match an even number of `a's. */
2556 {
2567 ;
2570 {
2575 {
2576 PATUNFETCH;
2577 PATUNFETCH;
2579 }
2582 }
2583 else
2584 {
2585 PATUNFETCH;
2587 }
2589 /* If we get here, we found another repeat character. */
2590 }
2592 /* Star, etc. applied to an empty pattern is equivalent
2593 to an empty pattern. */
2597 /* Now we know whether or not zero matches is allowed
2598 and also whether or not two or more matches is allowed. */
2601 end a backward relative jump from `b' to before the next
2602 jump we're going to put in below (which jumps from
2603 laststart to after this jump).
2605 But if we are at the `*' in the exact sequence `.*\n',
2606 insert an unconditional jump backwards to the .,
2607 instead of the beginning of the loop. This way we only
2608 push a failure point once, instead of every time
2609 through the loop. */
2612 /* Allocate the space for the jump. */
2615 /* We know we are not at the first character of the pattern,
2616 because laststart was nonzero. And we've already
2617 incremented `p', by the way, to be the character after
2618 the `*'. Do we have to do something analogous here
2619 for null bytes, because of RE_DOT_NOT_NULL? */
2621 && zero_times_ok
2627 }
2628 else
2629 /* Anything else. */
2633 /* We've added more stuff to the buffer. */
2635 }
2637 /* On failure, jump from laststart to b + 3, which will be the
2638 end of the buffer after this jump is inserted. */
2639 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2640 'b + 3'. */
2649 {
2650 /* At least one repetition is required, so insert a
2651 `dummy_failure_jump' before the initial
2652 `on_failure_jump' instruction of the loop. This
2653 effects a skip over that instruction the first time
2654 we hit that loop. */
2659 }
2660 }
2671 {
2673 #ifdef WCHAR
2675 #else
2677 #endif
2680 #ifdef WCHAR
2681 /* We assume a charset(_not) structure as a wchar_t array.
2682 charset[0] = (re_opcode_t) charset(_not)
2683 charset[1] = l (= length of char_classes)
2684 charset[2] = m (= length of collating_symbols)
2685 charset[3] = n (= length of equivalence_classes)
2686 charset[4] = o (= length of char_ranges)
2687 charset[5] = p (= length of chars)
2689 charset[6] = char_class (wctype_t)
2690 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2691 ...
2692 charset[l+5] = char_class (wctype_t)
2694 charset[l+6] = collating_symbol (wchar_t)
2695 ...
2696 charset[l+m+5] = collating_symbol (wchar_t)
2697 ifdef _LIBC we use the index if
2698 _NL_COLLATE_SYMB_EXTRAMB instead of
2699 wchar_t string.
2701 charset[l+m+6] = equivalence_classes (wchar_t)
2702 ...
2703 charset[l+m+n+5] = equivalence_classes (wchar_t)
2704 ifdef _LIBC we use the index in
2705 _NL_COLLATE_WEIGHT instead of
2706 wchar_t string.
2708 charset[l+m+n+6] = range_start
2709 charset[l+m+n+7] = range_end
2710 ...
2711 charset[l+m+n+2o+4] = range_start
2712 charset[l+m+n+2o+5] = range_end
2713 ifdef _LIBC we use the value looked up
2714 in _NL_COLLATE_COLLSEQ instead of
2715 wchar_t character.
2717 charset[l+m+n+2o+6] = char
2718 ...
2719 charset[l+m+n+2o+p+5] = char
2721 */
2723 /* We need at least 6 spaces: the opcode, the length of
2724 char_classes, the length of collating_symbols, the length of
2725 equivalence_classes, the length of char_ranges, the length of
2726 chars. */
2729 /* Save b as laststart. And We use laststart as the pointer
2730 to the first element of the charset here.
2731 In other words, laststart[i] indicates charset[i]. */
2734 /* We test `*p == '^' twice, instead of using an if
2735 statement, so we only need one BUF_PUSH. */
2738 p++;
2740 /* Push the length of char_classes, the length of
2741 collating_symbols, the length of equivalence_classes, the
2742 length of char_ranges and the length of chars. */
2746 /* Remember the first position in the bracket expression. */
2749 /* charset_not matches newline according to a syntax bit. */
2752 {
2755 }
2757 /* Read in characters and ranges, setting map bits. */
2759 {
2764 /* \ might escape characters inside [...] and [^...]. */
2766 {
2774 }
2776 /* Could be the end of the bracket expression. If it's
2777 not (i.e., when the bracket expression is `[]' so
2778 far), the ']' character bit gets set way below. */
2782 /* Look ahead to see if it's a range when the last thing
2783 was a character class. */
2787 /* Look ahead to see if it's a range when the last thing
2788 was a character: if this is a hyphen not at the
2789 beginning or the end of a list, then it's the range
2790 operator. */
2795 {
2796 reg_errcode_t ret;
2797 /* Allocate the space for range_start and range_end. */
2799 /* Update the pointer to indicate end of buffer. */
2805 }
2808 reg_errcode_t ret;
2810 /* Move past the `-'. */
2812 /* Allocate the space for range_start and range_end. */
2814 /* Update the pointer to indicate end of buffer. */
2817 laststart);
2820 }
2822 /* See if we're at the beginning of a possible character
2823 class. */
2831 /* If pattern is `[[:'. */
2835 {
2841 else
2842 /* This is in any case an invalid class name. */
2844 }
2847 /* If isn't a word bracketed by `[:' and `:]':
2848 undo the ending character, the letters, and leave
2849 the leading `:' and `[' (but store them as character). */
2851 {
2855 /* Query the character class as wctype_t. */
2860 /* Throw away the ] at the end of the character
2861 class. */
2866 /* Allocate the space for character class. */
2868 /* Update the pointer to indicate end of buffer. */
2870 /* Move data which follow character classes
2871 not to violate the data. */
2878 /* Store the character class. */
2880 /* Update length of char_classes */
2884 }
2885 else
2886 {
2887 c1++;
2889 PATUNFETCH;
2895 }
2896 }
2899 {
2902 # ifdef _LIBC
2905 # endif
2909 /* If pattern is `[[=' or '[[.'. */
2913 {
2919 else
2920 /* This is in any case an invalid class name. */
2922 }
2926 {
2928 /* If we have no collation data we use the default
2929 collation in which each character is in a class
2930 by itself. It also means that ASCII is the
2931 character set and therefore we cannot have character
2932 with more than one byte in the multibyte
2933 representation. */
2935 /* If not defined _LIBC, we push the name and
2936 `\0' for the sake of matching performance. */
2939 # ifdef _LIBC
2942 # endif
2943 {
2946 }
2947 # ifdef _LIBC
2948 else
2949 {
2956 /* This #include defines a local function! */
2957 # include <locale/weightwc.h>
2960 {
2961 /* We push the index for equivalence class. */
2966 _NL_COLLATE_TABLEWC);
2969 _NL_COLLATE_WEIGHTWC);
2972 _NL_COLLATE_EXTRAWC);
2975 _NL_COLLATE_INDIRECTWC);
2979 /* This is no valid character. */
2983 }
2985 {
2986 /* We push collation sequence value
2987 for collating symbol. */
2997 /* We have to convert the name to a single-byte
2998 string. This is possible since the names
2999 consist of ASCII characters and the internal
3000 representation is UCS4. */
3004 table_size =
3006 _NL_COLLATE_SYMB_HASH_SIZEMB);
3009 _NL_COLLATE_SYMB_TABLEMB);
3012 _NL_COLLATE_SYMB_EXTRAMB);
3014 /* Locate the character in the hashing table. */
3021 {
3022 /* First compare the hashing value. */
3028 {
3029 /* Yep, this is the entry. */
3033 }
3035 /* Next entry. */
3037 }
3040 {
3041 /* Compute the index of the byte sequence
3042 in the table. */
3044 /* Adjust for the alignment. */
3048 }
3050 {
3051 /* No valid character. Match it as a
3052 single byte character. */
3055 /* Update the length of characters */
3059 /* Throw away the ] at the end of the
3060 collating symbol. */
3062 /* exit from the switch block. */
3064 }
3065 else
3067 }
3069 }
3070 # endif
3071 /* Throw away the ] at the end of the equivalence
3072 class (or collating symbol). */
3075 /* Allocate the space for the equivalence class
3076 (or collating symbol) (and '\0' if needed). */
3078 /* Update the pointer to indicate end of buffer. */
3083 /* Calculate the offset of char_ranges,
3084 which is next to equivalence_classes. */
3087 /* Insert space. */
3090 /* Write the equivalence_class and \0. */
3094 /* Update the length of equivalence_classes. */
3097 }
3100 /* Calculate the offset of the equivalence_classes,
3101 which is next to collating_symbols. */
3103 /* Insert space and write the collationg_symbol
3104 and \0. */
3109 /* In re_match_2_internal if range_start < -1, we
3110 assume -range_start is the offset of the
3111 collating symbol which is specified as
3112 the character of the range start. So we assign
3113 -(laststart[1] + laststart[2] + 6) to
3114 range_start. */
3116 /* Update the length of collating_symbol. */
3119 }
3120 }
3121 else
3122 {
3123 c1++;
3125 PATUNFETCH;
3131 }
3132 }
3133 else
3134 {
3139 }
3140 }
3143 /* Ensure that we have enough space to push a charset: the
3144 opcode, the length count, and the bitset; 34 bytes in all. */
3149 /* We test `*p == '^' twice, instead of using an if
3150 statement, so we only need one BUF_PUSH. */
3153 p++;
3155 /* Remember the first position in the bracket expression. */
3158 /* Push the number of bytes in the bitmap. */
3161 /* Clear the whole map. */
3164 /* charset_not matches newline according to a syntax bit. */
3169 /* Read in characters and ranges, setting map bits. */
3171 {
3176 /* \ might escape characters inside [...] and [^...]. */
3178 {
3185 }
3187 /* Could be the end of the bracket expression. If it's
3188 not (i.e., when the bracket expression is `[]' so
3189 far), the ']' character bit gets set way below. */
3193 /* Look ahead to see if it's a range when the last thing
3194 was a character class. */
3198 /* Look ahead to see if it's a range when the last thing
3199 was a character: if this is a hyphen not at the
3200 beginning or the end of a list, then it's the range
3201 operator. */
3206 {
3207 reg_errcode_t ret
3212 }
3216 reg_errcode_t ret;
3218 /* Move past the `-'. */
3224 }
3226 /* See if we're at the beginning of a possible character
3227 class. */
3236 /* If pattern is `[[:'. */
3240 {
3246 else
3247 /* This is in any case an invalid class name. */
3249 }
3252 /* If isn't a word bracketed by `[:' and `:]':
3253 undo the ending character, the letters, and leave
3254 the leading `:' and `[' (but set bits for them). */
3256 {
3257 # if defined _LIBC || WIDE_CHAR_SUPPORT
3267 /* Throw away the ] at the end of the character
3268 class. */
3274 {
3275 # ifdef _LIBC
3278 # else
3281 # endif
3286 }
3289 # else
3307 /* Throw away the ] at the end of the character
3308 class. */
3314 {
3315 /* This was split into 3 if's to
3316 avoid an arbitrary limit in some compiler. */
3335 }
3338 }
3339 else
3340 {
3341 c1++;
3343 PATUNFETCH;
3348 }
3349 }
3351 {
3353 # ifdef _LIBC
3356 # endif
3361 /* If pattern is `[[='. */
3365 {
3371 else
3372 /* This is in any case an invalid class name. */
3374 }
3378 {
3379 /* If we have no collation data we use the default
3380 collation in which each character is in a class
3381 by itself. It also means that ASCII is the
3382 character set and therefore we cannot have character
3383 with more than one byte in the multibyte
3384 representation. */
3385 # ifdef _LIBC
3387 # endif
3388 {
3392 /* Throw away the ] at the end of the equivalence
3393 class. */
3396 /* Set the bit for the character. */
3398 }
3399 # ifdef _LIBC
3400 else
3401 {
3402 /* Try to match the byte sequence in `str' against
3403 those known to the collate implementation.
3404 First find out whether the bytes in `str' are
3405 actually from exactly one character. */
3414 /* This #include defines a local function! */
3415 # include <locale/weight.h>
3428 /* This is no valid character. */
3431 /* Throw away the ] at the end of the equivalence
3432 class. */
3435 /* Now we have to go throught the whole table
3436 and find all characters which have the same
3437 first level weight.
3439 XXX Note that this is not entirely correct.
3440 we would have to match multibyte sequences
3441 but this is not possible with the current
3442 implementation. */
3444 /* XXX This test would have to be changed if we
3445 would allow matching multibyte sequences. */
3447 {
3451 /* Test whether the lenghts match. */
3453 {
3454 /* They do. New compare the bytes of
3455 the weight. */
3464 /* They match. Mark the character as
3465 acceptable. */
3467 }
3468 }
3469 }
3470 # endif
3472 }
3473 else
3474 {
3475 c1++;
3477 PATUNFETCH;
3482 }
3483 }
3485 {
3487 # ifdef _LIBC
3490 # endif
3495 /* If pattern is `[[.'. */
3499 {
3505 else
3506 /* This is in any case an invalid class name. */
3508 }
3512 {
3513 /* If we have no collation data we use the default
3514 collation in which each character is the name
3515 for its own class which contains only the one
3516 character. It also means that ASCII is the
3517 character set and therefore we cannot have character
3518 with more than one byte in the multibyte
3519 representation. */
3520 # ifdef _LIBC
3522 # endif
3523 {
3527 /* Throw away the ] at the end of the equivalence
3528 class. */
3531 /* Set the bit for the character. */
3534 }
3535 # ifdef _LIBC
3536 else
3537 {
3538 /* Try to match the byte sequence in `str' against
3539 those known to the collate implementation.
3540 First find out whether the bytes in `str' are
3541 actually from exactly one character. */
3550 table_size =
3552 _NL_COLLATE_SYMB_HASH_SIZEMB);
3555 _NL_COLLATE_SYMB_TABLEMB);
3558 _NL_COLLATE_SYMB_EXTRAMB);
3560 /* Locate the character in the hashing table. */
3567 {
3568 /* First compare the hashing value. */
3575 {
3576 /* Yep, this is the entry. */
3580 }
3582 /* Next entry. */
3584 }
3587 /* This is no valid character. */
3590 /* Throw away the ] at the end of the equivalence
3591 class. */
3594 /* Now add the multibyte character(s) we found
3595 to the accept list.
3597 XXX Note that this is not entirely correct.
3598 we would have to match multibyte sequences
3599 but this is not possible with the current
3600 implementation. Also, we have to match
3601 collating symbols, which expand to more than
3602 one file, as a whole and not allow the
3603 individual bytes. */
3608 {
3611 }
3612 }
3613 # endif
3615 }
3616 else
3617 {
3618 c1++;
3620 PATUNFETCH;
3625 }
3626 }
3627 else
3628 {
3632 }
3633 }
3635 /* Discard any (non)matching list bytes that are all 0 at the
3636 end of the map. Decrease the map-length byte too. */
3641 }
3648 else
3655 else
3662 else
3669 else
3676 else
3683 /* Do not translate the character after the \, so that we can
3684 distinguish, e.g., \B from \b, even if we normally would
3685 translate, e.g., B to b. */
3689 {
3694 handle_open:
3696 regnum++;
3699 {
3701 compile_stack_elt_t);
3705 }
3707 /* These are the values to restore when we hit end of this
3708 group. They are all relative offsets, so that if the
3709 whole pattern moves because of realloc, they will still
3710 be valid. */
3712 COMPILE_STACK_TOP.fixup_alt_jump
3717 /* We will eventually replace the 0 with the number of
3718 groups inner to this one. But do not push a
3719 start_memory for groups beyond the last one we can
3720 represent in the compiled pattern. */
3722 {
3726 }
3733 /* If we've reached MAX_REGNUM groups, then this open
3734 won't actually generate any code, so we'll have to
3735 clear pending_exact explicitly. */
3744 {
3747 else
3749 }
3751 handle_close:
3754 alternative for a possible future
3755 `pop_failure_jump' to pop. See comments at
3756 `push_dummy_failure' in `re_match_2'. */
3759 /* We allocated space for this jump when we assigned
3760 to `fixup_alt_jump', in the `handle_alt' case below. */
3762 }
3764 /* See similar code for backslashed left paren above. */
3766 {
3769 else
3771 }
3773 /* Since we just checked for an empty stack above, this
3774 ``can't happen''. */
3776 {
3777 /* We don't just want to restore into `regnum', because
3778 later groups should continue to be numbered higher,
3779 as in `(ab)c(de)' -- the second group is #2. */
3780 regnum_t this_group_regnum;
3784 fixup_alt_jump
3790 /* If we've reached MAX_REGNUM groups, then this open
3791 won't actually generate any code, so we'll have to
3792 clear pending_exact explicitly. */
3795 /* We're at the end of the group, so now we know how many
3796 groups were inside this one. */
3798 {
3799 UCHAR_T *inner_group_loc
3805 }
3806 }
3813 handle_alt:
3817 /* Insert before the previous alternative a jump which
3818 jumps to this alternative if the former fails. */
3825 /* The alternative before this one has a jump after it
3826 which gets executed if it gets matched. Adjust that
3827 jump so it will jump to this alternative's analogous
3828 jump (put in below, which in turn will jump to the next
3829 (if any) alternative's such jump, etc.). The last such
3830 jump jumps to the correct final destination. A picture:
3831 _____ _____
3832 | | | |
3833 | v | v
3834 a | b | c
3836 If we are at `b', then fixup_alt_jump right now points to a
3837 three-byte space after `a'. We'll put in the jump, set
3838 fixup_alt_jump to right after `b', and leave behind three
3839 bytes which we'll fill in when we get to after `c'. */
3844 /* Mark and leave space for a jump after this alternative,
3845 to be filled in later either by next alternative or
3846 when know we're at the end of a series of alternatives. */
3857 /* If \{ is a literal. */
3859 /* If we're at `\{' and it's not the open-interval
3860 operator. */
3864 handle_interval:
3865 {
3866 /* If got here, then the syntax allows intervals. */
3868 /* At least (most) this many matches must be made. */
3871 /* Place in the uncompiled pattern (i.e., just after
3872 the '{') to go back to if the interval is invalid. */
3881 {
3885 }
3886 else
3887 /* Interval such as `{1}' => match exactly once. */
3894 {
3898 }
3903 /* If it's invalid to have no preceding re. */
3905 {
3911 else
3913 }
3915 /* We just parsed a valid interval. */
3920 /* If the upper bound is zero, don't want to succeed at
3921 all; jump from `laststart' to `b + 3', which will be
3922 the end of the buffer after we insert the jump. */
3923 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3924 instead of 'b + 3'. */
3926 {
3931 }
3933 /* Otherwise, we have a nontrivial interval. When
3934 we're all done, the pattern will look like:
3935 set_number_at <jump count> <upper bound>
3936 set_number_at <succeed_n count> <lower bound>
3937 succeed_n <after jump addr> <succeed_n count>
3938 <body of loop>
3939 jump_n <succeed_n addr> <jump count>
3940 (The upper bound and `jump_n' are omitted if
3941 `upper_bound' is 1, though.) */
3942 else
3944 more at the end of the loop. */
3950 /* Initialize lower bound of the `succeed_n', even
3951 though it will be set during matching by its
3952 attendant `set_number_at' (inserted next),
3953 because `re_compile_fastmap' needs to know.
3954 Jump to the `jump_n' we might insert below. */
3961 /* Code to initialize the lower bound. Insert
3962 before the `succeed_n'. The `5' is the last two
3963 bytes of this `set_number_at', plus 3 bytes of
3964 the following `succeed_n'. */
3965 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3966 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3967 of the following `succeed_n'. */
3974 append a backward jump to the `succeed_n'
3975 that starts this interval.
3977 When we've reached this during matching,
3978 we'll have matched the interval once, so
3979 jump back only `upper_bound - 1' times. */
3985 /* The location we want to set is the second
3986 parameter of the `jump_n'; that is `b-2' as
3987 an absolute address. `laststart' will be
3988 the `set_number_at' we're about to insert;
3989 `laststart+3' the number to set, the source
3990 for the relative address. But we are
3991 inserting into the middle of the pattern --
3992 so everything is getting moved up by 5.
3993 Conclusion: (b - 2) - (laststart + 3) + 5,
3994 i.e., b - laststart.
3996 We insert this at the beginning of the loop
3997 so that if we fail during matching, we'll
3998 reinitialize the bounds. */
4003 }
4004 }
4008 invalid_interval:
4011 unfetch_interval:
4012 /* Match the characters as literals. */
4017 else
4019 }
4021 #ifdef emacs
4022 /* There is no way to specify the before_dot and after_dot
4023 operators. rms says this is ok. --karl */
4104 /* Can't back reference to a subexpression if inside of it. */
4117 else
4121 normal_backslash:
4122 /* You might think it would be useful for \ to mean
4123 not to translate; but if we don't translate it
4124 it will never match anything. */
4127 }
4132 /* Expects the character in `c'. */
4133 normal_char:
4134 /* If no exactn currently being built. */
4136 #ifdef WCHAR
4137 /* If last exactn handle binary(or character) and
4138 new exactn handle character(or binary). */
4142 /* If last exactn not at current position. */
4145 /* We have only one byte following the exactn for the count. */
4148 /* If followed by a repetition operator. */
4157 {
4158 /* Start building a new exactn. */
4162 #ifdef WCHAR
4163 /* Is this exactn binary data or character? */
4167 else
4169 #else
4173 }
4182 /* Through the pattern now. */
4190 /* If we don't want backtracking, force success
4191 the first time we reach the end of the compiled pattern. */
4195 #ifdef WCHAR
4199 #endif
4202 /* We have succeeded; set the length of the buffer. */
4203 #ifdef WCHAR
4205 #else
4207 #endif
4209 #ifdef DEBUG
4211 {
4214 }
4217 #ifndef MATCH_MAY_ALLOCATE
4218 /* Initialize the failure stack to the largest possible stack. This
4219 isn't necessary unless we're trying to avoid calling alloca in
4220 the search and match routines. */
4221 {
4224 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4225 is strictly greater than re_max_failures, the largest possible stack
4226 is 2 * re_max_failures failure points. */
4228 {
4231 # ifdef emacs
4233 fail_stack.stack
4236 else
4237 fail_stack.stack
4243 fail_stack.stack
4246 else
4247 fail_stack.stack
4252 }
4255 }
4261 /* Subroutines for `regex_compile'. */
4263 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4264 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4266 static void
4268 re_opcode_t op;
4271 {
4274 }
4277 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4278 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4280 static void
4282 re_opcode_t op;
4285 {
4289 }
4292 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4293 for OP followed by two-byte integer parameter ARG. */
4294 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4296 static void
4298 re_opcode_t op;
4302 {
4310 }
4313 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4314 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4316 static void
4318 re_opcode_t op;
4322 {
4330 }
4333 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4334 after an alternative or a begin-subexpression. We assume there is at
4335 least one character before the ^. */
4337 static boolean
4340 reg_syntax_t syntax;
4341 {
4345 return
4346 /* After a subexpression? */
4348 /* After an alternative? */
4350 }
4353 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4354 at least one character after the $, i.e., `P < PEND'. */
4356 static boolean
4359 reg_syntax_t syntax;
4360 {
4365 return
4366 /* Before a subexpression? */
4369 /* Before an alternative? */
4372 }
4376 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4377 false if it's not. */
4379 static boolean
4381 compile_stack_type compile_stack;
4382 regnum_t regnum;
4383 {
4388 this_element--)
4393 }
4396 #ifdef INSIDE_RECURSION
4398 #ifdef WCHAR
4399 /* This insert space, which size is "num", into the pattern at "loc".
4400 "end" must point the end of the allocated buffer. */
4401 static void
4406 {
4412 }
4415 #ifdef WCHAR
4416 static reg_errcode_t
4418 char_set)
4419 CHAR_T range_start_char;
4422 RE_TRANSLATE_TYPE translate;
4423 reg_syntax_t syntax;
4424 {
4427 reg_errcode_t ret;
4428 # ifdef _LIBC
4431 # endif
4435 # ifdef _LIBC
4438 {
4440 _NL_COLLATE_COLLSEQWC);
4445 {
4446 /* range_start is a collating symbol. */
4448 /* Retreive the index and get collation sequence value. */
4451 }
4452 else
4457 /* Report an error if the range is empty and the syntax prohibits
4458 this. */
4462 /* Insert space to the end of the char_ranges. */
4467 }
4468 else
4469 # endif
4470 {
4472 range_start_char;
4474 /* Report an error if the range is empty and the syntax prohibits
4475 this. */
4479 /* Insert space to the end of the char_ranges. */
4484 }
4485 /* Have to increment the pointer into the pattern string, so the
4486 caller isn't still at the ending character. */
4490 }
4492 /* Read the ending character of a range (in a bracket expression) from the
4493 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4494 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4495 Then we set the translation of all bits between the starting and
4496 ending characters (inclusive) in the compiled pattern B.
4498 Return an error code.
4500 We use these short variable names so we can use the same macros as
4501 `regex_compile' itself. */
4503 static reg_errcode_t
4507 RE_TRANSLATE_TYPE translate;
4508 reg_syntax_t syntax;
4510 {
4513 reg_errcode_t ret;
4514 # if _LIBC
4518 # else
4520 # endif
4525 /* Have to increment the pointer into the pattern string, so the
4526 caller isn't still at the ending character. */
4529 /* Report an error if the range is empty and the syntax prohibits this. */
4532 # if _LIBC
4534 _NL_COLLATE_COLLSEQMB);
4539 {
4543 {
4546 }
4547 }
4548 # else
4549 /* Here we see why `this_char' has to be larger than an `unsigned
4550 char' -- we would otherwise go into an infinite loop, since all
4551 characters <= 0xff. */
4553 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4554 and some compilers cast it to int implicitly, so following for_loop
4555 may fall to (almost) infinite loop.
4556 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4557 To avoid this, we cast p[0] to unsigned int and truncate it. */
4561 {
4564 }
4565 # endif
4568 }
4570 \f
4571 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4572 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4573 characters can start a string that matches the pattern. This fastmap
4574 is used by re_search to skip quickly over impossible starting points.
4576 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4577 area as BUFP->fastmap.
4579 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4580 the pattern buffer.
4582 Returns 0 if we succeed, -2 if an internal error. */
4584 #ifdef WCHAR
4585 /* local function for re_compile_fastmap.
4586 truncate wchar_t character to char. */
4589 static unsigned char
4591 CHAR_T c;
4592 {
4597 # ifdef _LIBC
4599 # else
4601 # endif
4603 }
4606 static int
4609 {
4611 #ifdef MATCH_MAY_ALLOCATE
4613 #endif
4614 #ifndef REGEX_MALLOC
4616 #endif
4620 #ifdef WCHAR
4621 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4622 pattern to (char*) in regex_compile. */
4631 #ifdef REL_ALLOC
4632 /* This holds the pointer to the failure stack, when
4633 it is allocated relocatably. */
4635 #endif
4637 /* Assume that each path through the pattern can be null until
4638 proven otherwise. We set this false at the bottom of switch
4639 statement, to which we get only if a particular path doesn't
4640 match the empty string. */
4643 /* We aren't doing a `succeed_n' to begin with. */
4654 {
4656 {
4657 /* We have reached the (effective) end of pattern. */
4659 {
4662 /* Reset for next path. */
4668 }
4669 else
4671 }
4673 /* We should never be about to go beyond the end of the pattern. */
4677 {
4679 /* I guess the idea here is to simply not bother with a fastmap
4680 if a backreference is used, since it's too hard to figure out
4681 the fastmap for the corresponding group. Setting
4682 `can_be_null' stops `re_search_2' from using the fastmap, so
4683 that is all we do. */
4689 /* Following are the cases which match a character. These end
4690 with `break'. */
4692 #ifdef WCHAR
4701 #ifdef MBS_SUPPORT
4705 #endif
4707 #ifdef WCHAR
4708 /* It is hard to distinguish fastmap from (multi byte) characters
4709 which depends on current locale. */
4725 /* Chars beyond end of map must be allowed. */
4750 {
4753 /* `.' matches anything ... */
4757 /* ... except perhaps newline. */
4761 /* Return if we have already set `can_be_null'; if we have,
4762 then the fastmap is irrelevant. Something's wrong here. */
4766 /* Otherwise, have to check alternative paths. */
4768 }
4770 #ifdef emacs
4787 /* All cases after this match the empty string. These end with
4788 `continue'. */
4822 /* Jump backward implies we just went through the body of a
4823 loop and matched nothing. Opcode jumped to should be
4824 `on_failure_jump' or `succeed_n'. Just treat it like an
4825 ordinary jump. For a * loop, it has pushed its failure
4826 point already; if so, discard that as redundant. */
4831 p++;
4835 /* If what's on the stack is where we are now, pop it. */
4845 handle_on_failure_jump:
4848 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4849 end of the pattern. We don't want to push such a point,
4850 since when we restore it above, entering the switch will
4851 increment `p' past the end of the pattern. We don't need
4852 to push such a point since we obviously won't find any more
4853 fastmap entries beyond `pend'. Such a pattern can match
4854 the null string, though. */
4856 {
4858 {
4861 }
4862 }
4863 else
4867 {
4870 }
4876 /* Get to the number of times to succeed. */
4879 /* Increment p past the n for when k != 0. */
4882 {
4886 }
4905 /* Getting here means we have found the possible starting
4906 characters for one path of the pattern -- and that the empty
4907 string does not match. We need not follow this path further.
4908 Instead, look at the next alternative (remembered on the
4909 stack), or quit if no more. The test at the top of the loop
4910 does these things. */
4915 /* Set `can_be_null' for the last path (also the first path, if the
4916 pattern is empty). */
4919 done:
4922 }
4926 int
4929 {
4930 # ifdef MBS_SUPPORT
4933 else
4934 # endif
4937 #ifdef _LIBC
4939 #endif
4940 \f
4942 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4943 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4944 this memory for recording register information. STARTS and ENDS
4945 must be allocated using the malloc library routine, and must each
4946 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4948 If NUM_REGS == 0, then subsequent matches should allocate their own
4949 register data.
4951 Unless this function is called, the first search or match using
4952 PATTERN_BUFFER will allocate its own register data, without
4953 freeing the old data. */
4955 void
4961 {
4963 {
4968 }
4969 else
4970 {
4974 }
4975 }
4976 #ifdef _LIBC
4978 #endif
4979 \f
4980 /* Searching routines. */
4982 /* Like re_search_2, below, but only one string is specified, and
4983 doesn't let you say where to stop matching. */
4985 int
4991 {
4994 }
4995 #ifdef _LIBC
4997 #endif
5000 /* Using the compiled pattern in BUFP->buffer, first tries to match the
5001 virtual concatenation of STRING1 and STRING2, starting first at index
5002 STARTPOS, then at STARTPOS + 1, and so on.
5004 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
5006 RANGE is how far to scan while trying to match. RANGE = 0 means try
5007 only at STARTPOS; in general, the last start tried is STARTPOS +
5008 RANGE.
5010 In REGS, return the indices of the virtual concatenation of STRING1
5011 and STRING2 that matched the entire BUFP->buffer and its contained
5012 subexpressions.
5014 Do not consider matching one past the index STOP in the virtual
5015 concatenation of STRING1 and STRING2.
5017 We return either the position in the strings at which the match was
5018 found, -1 if no match, or -2 if error (such as failure
5019 stack overflow). */
5021 int
5030 {
5031 # ifdef MBS_SUPPORT
5035 else
5036 # endif
5040 #ifdef _LIBC
5042 #endif
5046 #ifdef INSIDE_RECURSION
5048 #ifdef MATCH_MAY_ALLOCATE
5049 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
5050 #else
5051 # define FREE_VAR(var) if (var) free (var); var = NULL
5052 #endif
5054 #ifdef WCHAR
5055 # define MAX_ALLOCA_SIZE 2000
5057 # define FREE_WCS_BUFFERS() \
5058 do { \
5059 if (size1 > MAX_ALLOCA_SIZE) \
5060 { \
5061 free (wcs_string1); \
5062 free (mbs_offset1); \
5063 } \
5064 else \
5065 { \
5066 FREE_VAR (wcs_string1); \
5067 FREE_VAR (mbs_offset1); \
5068 } \
5069 if (size2 > MAX_ALLOCA_SIZE) \
5070 { \
5071 free (wcs_string2); \
5072 free (mbs_offset2); \
5073 } \
5074 else \
5075 { \
5076 FREE_VAR (wcs_string2); \
5077 FREE_VAR (mbs_offset2); \
5078 } \
5079 } while (0)
5081 #endif
5084 static int
5094 {
5100 #ifdef WCHAR
5101 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5103 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5105 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5107 /* They hold whether each wchar_t is binary data or not. */
5111 /* Check for out-of-range STARTPOS. */
5115 /* Fix up RANGE if it might eventually take us outside
5116 the virtual concatenation of STRING1 and STRING2.
5117 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5123 /* If the search isn't to be a backwards one, don't waste time in a
5124 search for a pattern that must be anchored. */
5127 /* `begline' is like `begbuf' if it cannot match at newlines. */
5130 {
5133 else
5135 }
5137 #ifdef emacs
5138 /* In a forward search for something that starts with \=.
5139 don't keep searching past point. */
5141 {
5145 }
5148 /* Update the fastmap now if not correct already. */
5153 #ifdef WCHAR
5154 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5155 fill them with converted string. */
5157 {
5159 {
5163 }
5164 else
5165 {
5169 }
5171 {
5173 {
5177 }
5178 else
5179 {
5183 }
5185 }
5191 else
5193 }
5195 {
5197 {
5201 }
5202 else
5203 {
5207 }
5209 {
5213 else
5216 }
5222 else
5224 }
5228 /* Loop through the string, looking for a place to start matching. */
5230 {
5231 /* If a fastmap is supplied, skip quickly over characters that
5232 cannot be the start of a match. If the pattern can match the
5233 null string, however, we don't need to skip characters; we want
5234 the first null string. */
5236 {
5238 {
5248 /* Written out as an if-else to avoid testing `translate'
5249 inside the loop. */
5254 range--;
5255 else
5257 range--;
5260 }
5262 {
5269 }
5270 }
5272 /* If can't match the null string, and that's all we have left, fail. */
5275 {
5276 #ifdef WCHAR
5278 #endif
5280 }
5282 #ifdef WCHAR
5293 #ifndef REGEX_MALLOC
5294 # ifdef C_ALLOCA
5296 # endif
5297 #endif
5300 {
5301 #ifdef WCHAR
5303 #endif
5305 }
5308 {
5309 #ifdef WCHAR
5311 #endif
5313 }
5315 advance:
5319 {
5320 range--;
5321 startpos++;
5322 }
5323 else
5324 {
5325 range++;
5326 startpos--;
5327 }
5328 }
5329 #ifdef WCHAR
5331 #endif
5333 }
5335 #ifdef WCHAR
5336 /* This converts PTR, a pointer into one of the search wchar_t strings
5337 `string1' and `string2' into an multibyte string offset from the
5338 beginning of that string. We use mbs_offset to optimize.
5339 See convert_mbs_to_wcs. */
5340 # define POINTER_TO_OFFSET(ptr) \
5341 (FIRST_STRING_P (ptr) \
5342 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5343 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5344 + csize1)))
5346 /* This converts PTR, a pointer into one of the search strings `string1'
5347 and `string2' into an offset from the beginning of that string. */
5348 # define POINTER_TO_OFFSET(ptr) \
5349 (FIRST_STRING_P (ptr) \
5350 ? ((regoff_t) ((ptr) - string1)) \
5351 : ((regoff_t) ((ptr) - string2 + size1)))
5354 /* Macros for dealing with the split strings in re_match_2. */
5356 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5358 /* Call before fetching a character with *d. This switches over to
5359 string2 if necessary. */
5360 #define PREFETCH() \
5361 while (d == dend) \
5362 { \
5364 if (dend == end_match_2) \
5365 goto fail; \
5367 d = string2; \
5368 dend = end_match_2; \
5369 }
5371 /* Test if at very beginning or at very end of the virtual concatenation
5372 of `string1' and `string2'. If only one string, it's `string2'. */
5373 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5374 #define AT_STRINGS_END(d) ((d) == end2)
5377 /* Test if D points to a character which is word-constituent. We have
5378 two special cases to check for: if past the end of string1, look at
5379 the first character in string2; and if before the beginning of
5380 string2, look at the last character in string1. */
5381 #ifdef WCHAR
5382 /* Use internationalized API instead of SYNTAX. */
5383 # define WORDCHAR_P(d) \
5384 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5385 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5386 || ((d) == end1 ? *string2 \
5389 # define WORDCHAR_P(d) \
5390 (SYNTAX ((d) == end1 ? *string2 \
5391 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5392 == Sword)
5395 /* Disabled due to a compiler bug -- see comment at case wordbound */
5396 #if 0
5397 /* Test if the character before D and the one at D differ with respect
5398 to being word-constituent. */
5399 #define AT_WORD_BOUNDARY(d) \
5400 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5401 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5402 #endif
5404 /* Free everything we malloc. */
5405 #ifdef MATCH_MAY_ALLOCATE
5406 # ifdef WCHAR
5407 # define FREE_VARIABLES() \
5408 do { \
5409 REGEX_FREE_STACK (fail_stack.stack); \
5410 FREE_VAR (regstart); \
5411 FREE_VAR (regend); \
5412 FREE_VAR (old_regstart); \
5413 FREE_VAR (old_regend); \
5414 FREE_VAR (best_regstart); \
5415 FREE_VAR (best_regend); \
5416 FREE_VAR (reg_info); \
5417 FREE_VAR (reg_dummy); \
5418 FREE_VAR (reg_info_dummy); \
5419 if (!cant_free_wcs_buf) \
5420 { \
5421 FREE_VAR (string1); \
5422 FREE_VAR (string2); \
5423 FREE_VAR (mbs_offset1); \
5424 FREE_VAR (mbs_offset2); \
5425 } \
5426 } while (0)
5428 # define FREE_VARIABLES() \
5429 do { \
5430 REGEX_FREE_STACK (fail_stack.stack); \
5431 FREE_VAR (regstart); \
5432 FREE_VAR (regend); \
5433 FREE_VAR (old_regstart); \
5434 FREE_VAR (old_regend); \
5435 FREE_VAR (best_regstart); \
5436 FREE_VAR (best_regend); \
5437 FREE_VAR (reg_info); \
5438 FREE_VAR (reg_dummy); \
5439 FREE_VAR (reg_info_dummy); \
5440 } while (0)
5442 #else
5443 # ifdef WCHAR
5444 # define FREE_VARIABLES() \
5445 do { \
5446 if (!cant_free_wcs_buf) \
5447 { \
5448 FREE_VAR (string1); \
5449 FREE_VAR (string2); \
5450 FREE_VAR (mbs_offset1); \
5451 FREE_VAR (mbs_offset2); \
5452 } \
5453 } while (0)
5459 /* These values must meet several constraints. They must not be valid
5460 register values; since we have a limit of 255 registers (because
5461 we use only one byte in the pattern for the register number), we can
5462 use numbers larger than 255. They must differ by 1, because of
5463 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5464 be larger than the value for the highest register, so we do not try
5465 to actually save any registers when none are active. */
5466 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5467 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5468 \f
5470 /* Matching routines. */
5473 /* re_match is like re_match_2 except it takes only a single string. */
5475 int
5481 {
5483 # ifdef MBS_SUPPORT
5488 else
5489 # endif
5492 # ifndef REGEX_MALLOC
5493 # ifdef C_ALLOCA
5495 # endif
5496 # endif
5498 }
5499 # ifdef _LIBC
5501 # endif
5506 #ifdef INSIDE_RECURSION
5520 /* re_match_2 matches the compiled pattern in BUFP against the
5521 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5522 and SIZE2, respectively). We start matching at POS, and stop
5523 matching at STOP.
5525 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5526 store offsets for the substring each group matched in REGS. See the
5527 documentation for exactly how many groups we fill.
5529 We return -1 if no match, -2 if an internal error (such as the
5530 failure stack overflowing). Otherwise, we return the length of the
5531 matched substring. */
5533 int
5541 {
5543 # ifdef MBS_SUPPORT
5548 else
5549 # endif
5553 #ifndef REGEX_MALLOC
5554 # ifdef C_ALLOCA
5556 # endif
5557 #endif
5559 }
5560 #ifdef _LIBC
5562 #endif
5566 #ifdef INSIDE_RECURSION
5568 #ifdef WCHAR
5571 /* This check the substring (from 0, to length) of the multibyte string,
5572 to which offset_buffer correspond. And count how many wchar_t_characters
5573 the substring occupy. We use offset_buffer to optimization.
5574 See convert_mbs_to_wcs. */
5576 static int
5580 {
5583 /* Check whether the size is valid. */
5590 /* If there are no multibyte character, offset_buffer[i] == i.
5591 Optmize for this case. */
5595 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5600 {
5608 else
5610 }
5613 }
5616 /* This is a separate function so that we can force an alloca cleanup
5617 afterwards. */
5618 #ifdef WCHAR
5619 static int
5629 /* string1 == string2 == NULL means string1/2, size1/2 and
5630 mbs_offset1/2 need seting up in this function. */
5631 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5633 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5635 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5638 static int
5648 {
5649 /* General temporaries. */
5652 #ifdef WCHAR
5653 /* They hold whether each wchar_t is binary data or not. */
5655 /* If true, we can't free string1/2, mbs_offset1/2. */
5659 /* Just past the end of the corresponding string. */
5662 /* Pointers into string1 and string2, just past the last characters in
5663 each to consider matching. */
5666 /* Where we are in the data, and the end of the current string. */
5669 /* Where we are in the pattern, and the end of the pattern. */
5670 #ifdef WCHAR
5678 /* Mark the opcode just after a start_memory, so we can test for an
5679 empty subpattern when we get to the stop_memory. */
5682 /* We use this to map every character in the string. */
5685 /* Failure point stack. Each place that can handle a failure further
5686 down the line pushes a failure point on this stack. It consists of
5687 restart, regend, and reg_info for all registers corresponding to
5688 the subexpressions we're currently inside, plus the number of such
5689 registers, and, finally, two char *'s. The first char * is where
5690 to resume scanning the pattern; the second one is where to resume
5691 scanning the strings. If the latter is zero, the failure point is
5692 a ``dummy''; if a failure happens and the failure point is a dummy,
5693 it gets discarded and the next next one is tried. */
5696 #endif
5697 #ifdef DEBUG
5700 #endif
5702 #ifdef REL_ALLOC
5703 /* This holds the pointer to the failure stack, when
5704 it is allocated relocatably. */
5706 #endif
5708 /* We fill all the registers internally, independent of what we
5709 return, for use in backreferences. The number here includes
5710 an element for register zero. */
5713 /* The currently active registers. */
5717 /* Information on the contents of registers. These are pointers into
5718 the input strings; they record just what was matched (on this
5719 attempt) by a subexpression part of the pattern, that is, the
5720 regnum-th regstart pointer points to where in the pattern we began
5721 matching and the regnum-th regend points to right after where we
5722 stopped matching the regnum-th subexpression. (The zeroth register
5723 keeps track of what the whole pattern matches.) */
5726 #endif
5728 /* If a group that's operated upon by a repetition operator fails to
5729 match anything, then the register for its start will need to be
5730 restored because it will have been set to wherever in the string we
5731 are when we last see its open-group operator. Similarly for a
5732 register's end. */
5735 #endif
5737 /* The is_active field of reg_info helps us keep track of which (possibly
5738 nested) subexpressions we are currently in. The matched_something
5739 field of reg_info[reg_num] helps us tell whether or not we have
5740 matched any of the pattern so far this time through the reg_num-th
5741 subexpression. These two fields get reset each time through any
5742 loop their register is in. */
5745 #endif
5747 /* The following record the register info as found in the above
5748 variables when we find a match better than any we've seen before.
5749 This happens as we backtrack through the failure points, which in
5750 turn happens only if we have not yet matched the entire string. */
5754 #endif
5756 /* Logically, this is `best_regend[0]'. But we don't want to have to
5757 allocate space for that if we're not allocating space for anything
5758 else (see below). Also, we never need info about register 0 for
5759 any of the other register vectors, and it seems rather a kludge to
5760 treat `best_regend' differently than the rest. So we keep track of
5761 the end of the best match so far in a separate variable. We
5762 initialize this to NULL so that when we backtrack the first time
5763 and need to test it, it's not garbage. */
5766 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5769 /* Used when we pop values we don't care about. */
5773 #endif
5775 #ifdef DEBUG
5776 /* Counts the total number of registers pushed. */
5778 #endif
5784 #ifdef MATCH_MAY_ALLOCATE
5785 /* Do not bother to initialize all the register variables if there are
5786 no groups in the pattern, as it takes a fair amount of time. If
5787 there are groups, we include space for register 0 (the whole
5788 pattern), even though we never use it, since it simplifies the
5789 array indexing. We should fix this. */
5791 {
5804 {
5807 }
5808 }
5809 else
5810 {
5811 /* We must initialize all our variables to NULL, so that
5812 `FREE_VARIABLES' doesn't try to free them. */
5816 }
5819 /* The starting position is bogus. */
5820 #ifdef WCHAR
5824 #endif
5825 {
5828 }
5830 #ifdef WCHAR
5831 /* Allocate wchar_t array for string1 and string2 and
5832 fill them with converted string. */
5834 {
5835 /* We need seting up buffers here. */
5837 /* We must free wcs buffers in this function. */
5841 {
5846 {
5851 }
5852 }
5854 {
5859 {
5866 }
5871 }
5872 }
5874 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5875 pattern to (char*) in regex_compile. */
5881 /* Initialize subexpression text positions to -1 to mark ones that no
5882 start_memory/stop_memory has been seen for. Also initialize the
5883 register information struct. */
5885 {
5893 }
5895 /* We move `string1' into `string2' if the latter's empty -- but not if
5896 `string1' is null. */
5898 {
5903 #ifdef WCHAR
5908 #endif
5909 }
5913 /* Compute where to stop matching, within the two strings. */
5914 #ifdef WCHAR
5916 {
5920 }
5921 else
5922 {
5928 }
5933 }
5934 #else
5936 {
5939 }
5940 else
5941 {
5944 }
5947 /* `p' scans through the pattern as `d' scans through the data.
5948 `dend' is the end of the input string that `d' points within. `d'
5949 is advanced into the following input string whenever necessary, but
5950 this happens before fetching; therefore, at the beginning of the
5951 loop, `d' can be pointing at the end of a string, but it cannot
5952 equal `string2'. */
5953 #ifdef WCHAR
5955 {
5959 }
5960 else
5961 {
5965 }
5971 }
5972 #else
5974 {
5977 }
5978 else
5979 {
5982 }
5991 /* This loops over pattern commands. It exits by returning from the
5992 function if the match is complete, or it drops through if the match
5993 fails at this starting point in the input data. */
5995 {
5996 #ifdef _LIBC
5998 #else
6000 #endif
6006 /* If we haven't matched the entire string, and we want the
6007 longest match, try backtracking. */
6009 {
6010 /* 1 if this match ends in the same string (string1 or string2)
6011 as the best previous match. */
6014 /* 1 if this match is the best seen so far. */
6015 boolean best_match_p;
6017 /* AIX compiler got confused when this was combined
6018 with the previous declaration. */
6021 else
6029 /* If exceeds best match so far, save it. */
6031 {
6038 {
6041 }
6042 }
6044 }
6046 /* If no failure points, don't restore garbage. And if
6047 last match is real best match, don't restore second
6048 best one. */
6050 {
6051 restore_best_regs:
6052 /* Restore best match. It may happen that `dend ==
6053 end_match_1' while the restored d is in string2.
6054 For example, the pattern `x.*y.*z' against the
6055 strings `x-' and `y-z-', if the two strings are
6056 not consecutive in memory. */
6064 {
6067 }
6068 }
6071 succeed_label:
6073 /* If caller wants register contents data back, do it. */
6075 {
6076 /* Have the register data arrays been allocated? */
6079 extra element beyond `num_regs' for the `-1' marker
6080 GNU code uses. */
6085 {
6088 }
6090 }
6093 allocated, reallocate them. If we need fewer, just
6094 leave it alone. */
6096 {
6101 {
6104 }
6105 }
6106 }
6107 else
6108 {
6109 /* These braces fend off a "empty body in an else-statement"
6110 warning under GCC when assert expands to nothing. */
6112 }
6114 /* Convert the pointer data in `regstart' and `regend' to
6115 indices. Register zero has to be set differently,
6116 since we haven't kept track of any info for it. */
6118 {
6120 #ifdef WCHAR
6124 else
6127 #else
6132 }
6134 /* Go through the first `min (num_regs, regs->num_regs)'
6135 registers, since that is all we initialized. */
6137 mcnt++)
6138 {
6141 else
6142 {
6147 }
6148 }
6150 /* If the regs structure we return has more elements than
6151 were in the pattern, set the extra elements to -1. If
6152 we (re)allocated the registers, this is the case,
6153 because we always allocate enough to have at least one
6154 -1 at the end. */
6164 #ifdef WCHAR
6167 else
6169 csize1;
6171 #else
6173 ? string1
6181 }
6183 /* Otherwise match next pattern command. */
6185 {
6186 /* Ignore these. Used to ignore the n of succeed_n's which
6187 currently have n == 0. */
6196 /* Match the next n pattern characters exactly. The following
6197 byte in the pattern defines n, and the n bytes after that
6198 are the characters to match. */
6200 #ifdef MBS_SUPPORT
6202 #endif
6206 /* This is written out as an if-else so we don't waste time
6207 testing `translate' inside the loop. */
6209 {
6210 do
6211 {
6213 #ifdef WCHAR
6215 {
6219 }
6220 else
6221 {
6224 }
6225 #else
6230 }
6232 }
6233 else
6234 {
6235 do
6236 {
6239 }
6241 }
6246 /* Match any character except possibly a newline or a null. */
6258 d++;
6264 {
6266 #ifdef WCHAR
6270 #define WORK_BUFFER_SIZE 128
6272 # ifdef _LIBC
6281 #ifdef WCHAR
6282 # ifdef _LIBC
6291 /* p points charset[6], so the address of the next instruction
6292 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6293 where l=length of char_classes, m=length of collating_symbol,
6294 n=equivalence_class, o=length of char_range,
6295 p'=length of character. */
6297 /* Update p to indicate the next instruction. */
6301 /* match with char_class? */
6303 {
6310 # ifdef _LIBC
6313 # else
6316 # endif
6317 }
6319 /* match with collating_symbol? */
6320 # ifdef _LIBC
6322 {
6327 workp++)
6328 {
6336 {
6337 /* Update d, however d will be incremented at
6338 char_set_matched:, we decrement d here. */
6341 }
6342 }
6343 }
6345 # endif
6346 /* If we can't look up collation data, we use wcscoll
6347 instead. */
6348 {
6350 {
6352 # ifdef _LIBC
6354 # else
6356 # endif
6358 /* If wcscoll(the collating symbol, whole string) > 0,
6359 any substring of the string never match with the
6360 collating symbol. */
6361 # ifdef _LIBC
6363 # else
6365 # endif
6366 {
6369 }
6371 /* First, we compare the collating symbol with
6372 the first character of the string.
6373 If it don't match, we add the next character to
6374 the compare buffer in turn. */
6376 {
6379 {
6384 }
6386 /* add next character to the compare buffer. */
6390 # ifdef _LIBC
6392 # else
6394 # endif
6399 /* (str_buf > workp) indicate (str_buf + X > workp),
6400 because for all X (str_buf + X > str_buf).
6401 So we don't need continue this loop. */
6404 /* Otherwise(str_buf < workp),
6405 (str_buf+next_character) may equals (workp).
6406 So we continue this loop. */
6407 }
6408 /* not matched */
6412 }
6413 }
6414 /* match with equivalence_class? */
6415 # ifdef _LIBC
6417 {
6419 /* Try to match the equivalence class against
6420 those known to the collate implementation. */
6429 /* This #include defines a local function! */
6430 # include <locale/weightwc.h>
6441 /* Write 1 collating element to str_buf, and
6442 get its index. */
6446 {
6449 {
6454 }
6458 }
6460 /* Update d, however d will be incremented at
6461 char_set_matched:, we decrement d here. */
6464 {
6467 else
6468 {
6471 }
6472 }
6477 workp++)
6478 {
6480 /* We already checked idx != 0 in regex_compile. */
6483 {
6491 }
6492 }
6493 /* not matched */
6496 }
6498 # endif
6499 /* If we can't look up collation data, we use wcscoll
6500 instead. */
6501 {
6503 {
6505 # ifdef _LIBC
6507 # else
6509 # endif
6511 /* If wcscoll(the collating symbol, whole string) > 0,
6512 any substring of the string never match with the
6513 collating symbol. */
6514 # ifdef _LIBC
6516 # else
6518 # endif
6519 {
6522 }
6524 /* First, we compare the equivalence class with
6525 the first character of the string.
6526 If it don't match, we add the next character to
6527 the compare buffer in turn. */
6529 {
6532 {
6537 }
6539 /* add next character to the compare buffer. */
6543 # ifdef _LIBC
6545 # else
6547 # endif
6553 /* (str_buf > workp) indicate (str_buf + X > workp),
6554 because for all X (str_buf + X > str_buf).
6555 So we don't need continue this loop. */
6558 /* Otherwise(str_buf < workp),
6559 (str_buf+next_character) may equals (workp).
6560 So we continue this loop. */
6561 }
6562 /* not matched */
6566 }
6567 }
6569 /* match with char_range? */
6570 # ifdef _LIBC
6572 {
6580 {
6583 /* We already compute the collation sequence value
6584 of the characters (or collating symbols). */
6590 }
6591 }
6592 else
6593 # endif
6594 {
6595 /* We set range_start_char at str_buf[0], range_end_char
6596 at str_buf[4], and compared char at str_buf[2]. */
6602 {
6605 /* match if (range_start_char <= c <= range_end_char). */
6607 /* If range_start(or end) < 0, we assume -range_start(end)
6608 is the offset of the collating symbol which is specified
6609 as the character of the range start(end). */
6611 /* range_start */
6614 else
6615 {
6618 }
6620 /* range_end */
6623 else
6624 {
6627 }
6629 # ifdef _LIBC
6632 # else
6635 # endif
6637 }
6638 }
6640 /* match with char? */
6647 char_set_matched:
6649 #else
6650 /* Cast to `unsigned' instead of `unsigned char' in case the
6651 bit list is a full 32 bytes long. */
6659 #undef WORK_BUFFER_SIZE
6662 d++;
6664 }
6667 /* The beginning of a group is represented by start_memory.
6668 The arguments are the register number in the next byte, and the
6669 number of groups inner to this one in the next. The text
6670 matched within the group is recorded (in the internal
6671 registers data structure) under the register number. */
6676 /* Find out if this group can match the empty string. */
6683 /* Save the position in the string where we were the last time
6684 we were at this open-group operator in case the group is
6685 operated upon by a repetition operator, e.g., with `(a*)*b'
6686 against `ab'; then we want to ignore where we are now in
6687 the string in case this attempt to match fails. */
6700 /* Clear this whenever we change the register activity status. */
6703 /* This is the new highest active register. */
6706 /* If nothing was active before, this is the new lowest active
6707 register. */
6711 /* Move past the register number and inner group count. */
6718 /* The stop_memory opcode represents the end of a group. Its
6719 arguments are the same as start_memory's: the register
6720 number, and the number of inner groups. */
6725 /* We need to save the string position the last time we were at
6726 this close-group operator in case the group is operated
6727 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6728 against `aba'; then we want to ignore where we are now in
6729 the string in case this attempt to match fails. */
6739 /* This register isn't active anymore. */
6742 /* Clear this whenever we change the register activity status. */
6745 /* If this was the only register active, nothing is active
6746 anymore. */
6748 {
6751 }
6752 else
6754 it isn't necessarily one less than now: consider
6755 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6756 new highest active register is 1. */
6759 r--;
6761 /* If we end up at register zero, that means that we saved
6762 the registers as the result of an `on_failure_jump', not
6763 a `start_memory', and we jumped to past the innermost
6764 `stop_memory'. For example, in ((.)*) we save
6765 registers 1 and 2 as a result of the *, but when we pop
6766 back to the second ), we are at the stop_memory 1.
6767 Thus, nothing is active. */
6769 {
6772 }
6773 else
6775 }
6777 /* If just failed to match something this time around with a
6778 group that's operated on by a repetition operator, try to
6779 force exit from the ``loop'', and restore the register
6780 information for this group that we had before trying this
6781 last match. */
6785 {
6791 {
6805 }
6808 /* If the next operation is a jump backwards in the pattern
6809 to an on_failure_jump right before the start_memory
6810 corresponding to this stop_memory, exit from the loop
6811 by forcing a failure after pushing on the stack the
6812 on_failure_jump's jump in the pattern, and d. */
6816 {
6817 /* If this group ever matched anything, then restore
6818 what its registers were before trying this last
6819 failed match, e.g., with `(a*)*b' against `ab' for
6820 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6821 against `aba' for regend[3].
6823 Also restore the registers for inner groups for,
6824 e.g., `((a*)(b*))*' against `aba' (register 3 would
6825 otherwise get trashed). */
6828 {
6833 /* Restore this and inner groups' (if any) registers. */
6835 r++)
6836 {
6839 /* xx why this test? */
6842 }
6843 }
6844 p1++;
6849 }
6850 }
6852 /* Move past the register number and the inner group count. */
6857 /* \<digit> has been turned into a `duplicate' command which is
6858 followed by the numeric value of <digit> as the register number. */
6860 {
6865 /* Can't back reference a group which we've never matched. */
6869 /* Where in input to try to start matching. */
6872 /* Where to stop matching; if both the place to start and
6873 the place to stop matching are in the same string, then
6874 set to the place to stop, otherwise, for now have to use
6875 the end of the first string. */
6881 {
6882 /* If necessary, advance to next segment in register
6883 contents. */
6885 {
6889 /* End of string1 => advance to string2. */
6892 }
6893 /* At end of register contents => success */
6896 /* If necessary, advance to next segment in data. */
6899 /* How many characters left in this segment to match. */
6902 /* Want how many consecutive characters we can match in
6903 one shot, so, if necessary, adjust the count. */
6907 /* Compare that many; failure if mismatch, else move
6908 past them. */
6915 /* Do this because we've match some characters. */
6917 }
6918 }
6922 /* begline matches the empty string at the beginning of the string
6923 (unless `not_bol' is set in `bufp'), and, if
6924 `newline_anchor' is set, after newlines. */
6929 {
6931 }
6933 {
6935 }
6936 /* In all other cases, we fail. */
6940 /* endline is the dual of begline. */
6945 {
6947 }
6949 /* We have to ``prefetch'' the next character. */
6952 {
6954 }
6958 /* Match at the very beginning of the data. */
6966 /* Match at the very end of the data. */
6974 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6975 pushes NULL as the value for the string on the stack. Then
6976 `pop_failure_point' will keep the current value for the
6977 string, instead of restoring it. To see why, consider
6978 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6979 then the . fails against the \n. But the next thing we want
6980 to do is match the \n against the \n; if we restored the
6981 string value, we would be back at the foo.
6983 Because this is used only in specific cases, we don't need to
6984 check all the things that `on_failure_jump' does, to make
6985 sure the right things get saved on the stack. Hence we don't
6986 share its code. The only reason to push anything on the
6987 stack at all is that otherwise we would have to change
6988 `anychar's code to do something besides goto fail in this
6989 case; that seems worse than this. */
6994 #ifdef _LIBC
6996 #else
6998 #endif
7004 /* Uses of on_failure_jump:
7006 Each alternative starts with an on_failure_jump that points
7007 to the beginning of the next alternative. Each alternative
7008 except the last ends with a jump that in effect jumps past
7009 the rest of the alternatives. (They really jump to the
7010 ending jump of the following alternative, because tensioning
7011 these jumps is a hassle.)
7013 Repeats start with an on_failure_jump that points past both
7014 the repetition text and either the following jump or
7015 pop_failure_jump back to this on_failure_jump. */
7017 on_failure:
7021 #ifdef _LIBC
7023 #else
7025 #endif
7027 /* If this on_failure_jump comes right before a group (i.e.,
7028 the original * applied to a group), save the information
7029 for that group and all inner ones, so that if we fail back
7030 to this point, the group's information will be correct.
7031 For example, in \(a*\)*\1, we need the preceding group,
7032 and in \(zz\(a*\)b*\)\2, we need the inner group. */
7034 /* We can't use `p' to check ahead because we push
7035 a failure point to `p + mcnt' after we do this. */
7038 /* We need to skip no_op's before we look for the
7039 start_memory in case this on_failure_jump is happening as
7040 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
7041 against aba. */
7043 p1++;
7046 {
7047 /* We have a new highest active register now. This will
7048 get reset at the start_memory we are about to get to,
7049 but we will have saved all the registers relevant to
7050 this repetition op, as described above. */
7054 }
7061 /* A smart repeat ends with `maybe_pop_jump'.
7062 We change it to either `pop_failure_jump' or `jump'. */
7066 {
7069 /* Compare the beginning of the repeat with what in the
7070 pattern follows its end. If we can establish that there
7071 is nothing that they would both match, i.e., that we
7072 would have to backtrack because of (as in, e.g., `a*a')
7073 then we can change to pop_failure_jump, because we'll
7074 never have to backtrack.
7076 This is not true in the case of alternatives: in
7077 `(a|ab)*' we do need to backtrack to the `ab' alternative
7078 (e.g., if the string was `ab'). But instead of trying to
7079 detect that here, the alternative has put on a dummy
7080 failure point which is what we will end up popping. */
7082 /* Skip over open/close-group commands.
7083 If what follows this loop is a ...+ construct,
7084 look at what begins its body, since we will have to
7085 match at least one of that. */
7087 {
7095 else
7097 }
7100 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7101 to the `maybe_finalize_jump' of this case. Examine what
7102 follows. */
7104 /* If we're at the end of the pattern, we can change. */
7106 {
7107 /* Consider what happens when matching ":\(.*\)"
7108 against ":/". I don't really understand this code
7109 yet. */
7111 pop_failure_jump;
7112 DEBUG_PRINT1
7114 }
7117 #ifdef MBS_SUPPORT
7119 #endif
7121 {
7122 register UCHAR_T c
7126 #ifdef MBS_SUPPORT
7128 #endif
7130 {
7132 pop_failure_jump;
7133 #ifdef WCHAR
7137 #else
7141 #endif
7142 }
7144 #ifndef WCHAR
7147 {
7154 /* `not' is equal to 1 if c would match, which means
7155 that we can't change to pop_failure_jump. */
7157 {
7160 }
7161 }
7163 }
7164 #ifndef WCHAR
7166 {
7167 /* We win if the first character of the loop is not part
7168 of the charset. */
7173 {
7176 }
7179 {
7181 /* We win if the charset_not inside the loop
7182 lists every character listed in the charset after. */
7190 {
7193 }
7194 }
7196 {
7198 /* We win if the charset inside the loop
7199 has no overlap with the one after the loop. */
7202 idx++)
7207 {
7210 }
7211 }
7212 }
7214 }
7217 {
7221 }
7222 /* Note fall through. */
7225 /* The end of a simple repeat has a pop_failure_jump back to
7226 its matching on_failure_jump, where the latter will push a
7227 failure point. The pop_failure_jump takes off failure
7228 points put on by this pop_failure_jump's matching
7229 on_failure_jump; we got through the pattern to here from the
7230 matching on_failure_jump, so didn't fail. */
7232 {
7233 /* We need to pass separate storage for the lowest and
7234 highest registers, even though we don't care about the
7235 actual values. Otherwise, we will restore only one
7236 register from the stack, since lowest will == highest in
7237 `pop_failure_point'. */
7246 }
7247 /* Note fall through. */
7249 unconditional_jump:
7250 #ifdef _LIBC
7252 #else
7254 #endif
7255 /* Note fall through. */
7257 /* Unconditionally jump (without popping any failure points). */
7262 #ifdef _LIBC
7264 #else
7266 #endif
7270 /* We need this opcode so we can detect where alternatives end
7271 in `group_match_null_string_p' et al. */
7277 /* Normally, the on_failure_jump pushes a failure point, which
7278 then gets popped at pop_failure_jump. We will end up at
7279 pop_failure_jump, also, and with a pattern of, say, `a+', we
7280 are skipping over the on_failure_jump, so we have to push
7281 something meaningless for pop_failure_jump to pop. */
7284 /* It doesn't matter what we push for the string here. What
7285 the code at `fail' tests is the value for the pattern. */
7290 /* At the end of an alternative, we need to push a dummy failure
7291 point in case we are followed by a `pop_failure_jump', because
7292 we don't want the failure point for the alternative to be
7293 popped. For example, matching `(a|ab)*' against `aab'
7294 requires that we match the `ab' alternative. */
7297 /* See comments just above at `dummy_failure_jump' about the
7298 two zeroes. */
7302 /* Have to succeed matching what follows at least n times.
7303 After that, handle like `on_failure_jump'. */
7309 /* Originally, this is how many times we HAVE to succeed. */
7311 {
7312 mcnt--;
7315 #ifdef _LIBC
7318 #else
7321 #endif
7322 }
7324 {
7325 #ifdef _LIBC
7328 #else
7333 #ifdef WCHAR
7335 #else
7340 }
7347 /* Originally, this is how many times we CAN jump. */
7349 {
7350 mcnt--;
7353 #ifdef _LIBC
7355 mcnt);
7356 #else
7358 mcnt);
7361 }
7362 /* If don't have to jump any more, skip over the rest of command. */
7363 else
7368 {
7374 #ifdef _LIBC
7376 #else
7378 #endif
7381 }
7383 #if 0
7384 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7385 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7386 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7387 macro and introducing temporary variables works around the bug. */
7400 #else
7402 {
7414 }
7417 {
7429 }
7430 #endif
7446 #ifdef emacs
7473 matchsyntax:
7475 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7476 d++;
7490 matchnotsyntax:
7492 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7493 d++;
7506 d++;
7515 d++;
7521 }
7525 /* We goto here if a matching operation fails. */
7526 fail:
7534 /* If this failure point is a dummy, try the next one. */
7538 /* If we failed to the end of the pattern, don't examine *p. */
7541 {
7544 /* If failed to a backwards jump that's part of a repetition
7545 loop, need to pop this failure point and use the next one. */
7547 {
7558 || (!is_a_jump_n
7564 }
7565 }
7569 }
7570 else
7581 \f
7582 /* Subroutine definitions for re_match_2. */
7585 /* We are passed P pointing to a register number after a start_memory.
7587 Return true if the pattern up to the corresponding stop_memory can
7588 match the empty string, and false otherwise.
7590 If we find the matching stop_memory, sets P to point to one past its number.
7591 Otherwise, sets P to an undefined byte less than or equal to END.
7593 We don't handle duplicates properly (yet). */
7595 static boolean
7599 {
7601 /* Point to after the args to the start_memory. */
7605 {
7606 /* Skip over opcodes that can match nothing, and return true or
7607 false, as appropriate, when we get to one that can't, or to the
7608 matching stop_memory. */
7611 {
7612 /* Could be either a loop or a series of alternatives. */
7614 p1++;
7617 /* If the next operation is not a jump backwards in the
7618 pattern. */
7621 {
7622 /* Go through the on_failure_jumps of the alternatives,
7623 seeing if any of the alternatives cannot match nothing.
7624 The last alternative starts with only a jump,
7625 whereas the rest start with on_failure_jump and end
7626 with a jump, e.g., here is the pattern for `a|b|c':
7628 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7629 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7630 /exactn/1/c
7632 So, we have to first go through the first (n-1)
7633 alternatives and then deal with the last one separately. */
7636 /* Deal with the first (n-1) alternatives, which start
7637 with an on_failure_jump (see above) that jumps to right
7638 past a jump_past_alt. */
7641 jump_past_alt)
7642 {
7643 /* `mcnt' holds how many bytes long the alternative
7644 is, including the ending `jump_past_alt' and
7645 its number. */
7649 reg_info))
7652 /* Move to right after this alternative, including the
7653 jump_past_alt. */
7656 /* Break if it's the beginning of an n-th alternative
7657 that doesn't begin with an on_failure_jump. */
7661 /* Still have to check that it's not an n-th
7662 alternative that starts with an on_failure_jump. */
7663 p1++;
7666 jump_past_alt)
7667 {
7668 /* Get to the beginning of the n-th alternative. */
7671 }
7672 }
7674 /* Deal with the last alternative: go back and get number
7675 of the `jump_past_alt' just before it. `mcnt' contains
7676 the length of the alternative. */
7696 }
7703 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7704 It expects P to be the first byte of a single alternative and END one
7705 byte past the last. The alternative can contain groups. */
7707 static boolean
7711 {
7716 {
7717 /* Skip over opcodes that can match nothing, and break when we get
7718 to one that can't. */
7721 {
7722 /* It's a loop. */
7724 p1++;
7732 }
7739 /* Deals with the ops common to group_match_null_string_p and
7740 alt_match_null_string_p.
7742 Sets P to one after the op and its arguments, if any. */
7744 static boolean
7748 {
7750 boolean ret;
7755 {
7765 #ifdef emacs
7769 #endif
7777 /* Have to set this here in case we're checking a group which
7778 contains a group and a back reference to it. */
7787 /* If this is an optimized succeed_n for zero times, make the jump. */
7792 else
7797 /* Get to the number of times to succeed. */
7802 {
7806 }
7807 else
7820 /* All other opcodes mean we cannot match the empty string. */
7822 }
7829 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7830 bytes; nonzero otherwise. */
7832 static int
7836 RE_TRANSLATE_TYPE translate;
7837 {
7841 {
7842 #ifdef WCHAR
7849 len--;
7850 }
7852 }
7853 \f
7857 /* Entry points for GNU code. */
7859 /* re_compile_pattern is the GNU regular expression compiler: it
7860 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7861 Returns 0 if the pattern was valid, otherwise an error string.
7863 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7864 are set in BUFP on entry.
7866 We call regex_compile to do the actual compilation. */
7873 {
7874 reg_errcode_t ret;
7876 /* GNU code is written to assume at least RE_NREGS registers will be set
7877 (and at least one extra will be -1). */
7880 /* And GNU code determines whether or not to get register information
7881 by passing null for the REGS argument to re_match, etc., not by
7882 setting no_sub. */
7885 /* Match anchors at newline. */
7888 # ifdef MBS_SUPPORT
7891 else
7892 # endif
7898 }
7899 #ifdef _LIBC
7901 #endif
7902 \f
7903 /* Entry points compatible with 4.2 BSD regex library. We don't define
7904 them unless specifically requested. */
7906 #if defined _REGEX_RE_COMP || defined _LIBC
7908 /* BSD has one and only one pattern buffer. */
7912 #ifdef _LIBC
7913 /* Make these definitions weak in libc, so POSIX programs can redefine
7914 these names if they don't use our functions, and still use
7915 regcomp/regexec below without link errors. */
7916 weak_function
7917 #endif
7920 {
7921 reg_errcode_t ret;
7924 {
7928 }
7931 {
7940 }
7942 /* Since `re_exec' always passes NULL for the `regs' argument, we
7943 don't need to initialize the pattern buffer fields which affect it. */
7945 /* Match anchors at newlines. */
7948 # ifdef MBS_SUPPORT
7951 else
7952 # endif
7958 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7960 }
7963 int
7964 #ifdef _LIBC
7965 weak_function
7966 #endif
7969 {
7971 return
7973 }
7976 \f
7977 /* POSIX.2 functions. Don't define these for Emacs. */
7979 #ifndef emacs
7981 /* regcomp takes a regular expression as a string and compiles it.
7983 PREG is a regex_t *. We do not expect any fields to be initialized,
7984 since POSIX says we shouldn't. Thus, we set
7986 `buffer' to the compiled pattern;
7987 `used' to the length of the compiled pattern;
7988 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7989 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7990 RE_SYNTAX_POSIX_BASIC;
7991 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7992 `fastmap' to an allocated space for the fastmap;
7993 `fastmap_accurate' to zero;
7994 `re_nsub' to the number of subexpressions in PATTERN.
7996 PATTERN is the address of the pattern string.
7998 CFLAGS is a series of bits which affect compilation.
8000 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
8001 use POSIX basic syntax.
8003 If REG_NEWLINE is set, then . and [^...] don't match newline.
8004 Also, regexec will try a match beginning after every newline.
8006 If REG_ICASE is set, then we considers upper- and lowercase
8007 versions of letters to be equivalent when matching.
8009 If REG_NOSUB is set, then when PREG is passed to regexec, that
8010 routine will report only success or failure, and nothing about the
8011 registers.
8013 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
8014 the return codes and their meanings.) */
8016 int
8021 {
8022 reg_errcode_t ret;
8023 reg_syntax_t syntax
8027 /* regex_compile will allocate the space for the compiled pattern. */
8032 /* Try to allocate space for the fastmap. */
8036 {
8039 preg->translate
8045 /* Map uppercase characters to corresponding lowercase ones. */
8048 }
8049 else
8052 /* If REG_NEWLINE is set, newlines are treated differently. */
8057 /* It also changes the matching behavior. */
8059 }
8060 else
8065 /* POSIX says a null character in the pattern terminates it, so we
8066 can use strlen here in compiling the pattern. */
8067 # ifdef MBS_SUPPORT
8070 else
8071 # endif
8074 /* POSIX doesn't distinguish between an unmatched open-group and an
8075 unmatched close-group: both are REG_EPAREN. */
8079 {
8080 /* Compute the fastmap now, since regexec cannot modify the pattern
8081 buffer. */
8083 {
8084 /* Some error occurred while computing the fastmap, just forget
8085 about it. */
8088 }
8089 }
8092 }
8093 #ifdef _LIBC
8095 #endif
8098 /* regexec searches for a given pattern, specified by PREG, in the
8099 string STRING.
8101 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
8102 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
8103 least NMATCH elements, and we set them to the offsets of the
8104 corresponding matched substrings.
8106 EFLAGS specifies `execution flags' which affect matching: if
8107 REG_NOTBOL is set, then ^ does not match at the beginning of the
8108 string; if REG_NOTEOL is set, then $ does not match at the end.
8110 We return 0 if we find a match and REG_NOMATCH if not. */
8112 int
8117 regmatch_t pmatch[];
8119 {
8122 regex_t private_preg;
8131 /* The user has told us exactly how many registers to return
8132 information about, via `nmatch'. We have to pass that on to the
8133 matching routines. */
8137 {
8143 }
8145 /* Perform the searching operation. */
8150 /* Copy the register information to the POSIX structure. */
8152 {
8154 {
8158 {
8161 }
8162 }
8164 /* If we needed the temporary register info, free the space now. */
8166 }
8168 /* We want zero return to mean success, unlike `re_search'. */
8170 }
8171 #ifdef _LIBC
8173 #endif
8176 /* Returns a message corresponding to an error code, ERRCODE, returned
8177 from either regcomp or regexec. We don't use PREG here. */
8179 size_t
8185 {
8192 /* Only error codes returned by the rest of the code should be passed
8193 to this routine. If we are given anything else, or if other regex
8194 code generates an invalid error code, then the program has a bug.
8195 Dump core so we can fix it. */
8203 {
8205 {
8206 #if defined HAVE_MEMPCPY || defined _LIBC
8208 #else
8211 #endif
8212 }
8213 else
8215 }
8218 }
8219 #ifdef _LIBC
8221 #endif
8224 /* Free dynamically allocated space used by PREG. */
8226 void
8229 {
8245 }
8246 #ifdef _LIBC
8248 #endif
8254 \f
8255 #undef STORE_NUMBER
8256 #undef STORE_NUMBER_AND_INCR
8257 #undef EXTRACT_NUMBER
8258 #undef EXTRACT_NUMBER_AND_INCR
8260 #undef DEBUG_PRINT_COMPILED_PATTERN
8261 #undef DEBUG_PRINT_DOUBLE_STRING
8263 #undef INIT_FAIL_STACK
8264 #undef RESET_FAIL_STACK
8265 #undef DOUBLE_FAIL_STACK
8266 #undef PUSH_PATTERN_OP
8267 #undef PUSH_FAILURE_POINTER
8268 #undef PUSH_FAILURE_INT
8269 #undef PUSH_FAILURE_ELT
8270 #undef POP_FAILURE_POINTER
8271 #undef POP_FAILURE_INT
8272 #undef POP_FAILURE_ELT
8273 #undef DEBUG_PUSH
8274 #undef DEBUG_POP
8275 #undef PUSH_FAILURE_POINT
8276 #undef POP_FAILURE_POINT
8278 #undef REG_UNSET_VALUE
8279 #undef REG_UNSET
8281 #undef PATFETCH
8282 #undef PATFETCH_RAW
8283 #undef PATUNFETCH
8284 #undef TRANSLATE
8286 #undef INIT_BUF_SIZE
8287 #undef GET_BUFFER_SPACE
8288 #undef BUF_PUSH
8289 #undef BUF_PUSH_2
8290 #undef BUF_PUSH_3
8291 #undef STORE_JUMP
8292 #undef STORE_JUMP2
8293 #undef INSERT_JUMP
8294 #undef INSERT_JUMP2
8295 #undef EXTEND_BUFFER
8296 #undef GET_UNSIGNED_NUMBER
8297 #undef FREE_STACK_RETURN
8299 # undef POINTER_TO_OFFSET
8300 # undef MATCHING_IN_FRST_STRING
8301 # undef PREFETCH
8302 # undef AT_STRINGS_BEG
8303 # undef AT_STRINGS_END
8304 # undef WORDCHAR_P
8305 # undef FREE_VAR
8306 # undef FREE_VARIABLES
8307 # undef NO_HIGHEST_ACTIVE_REG
8308 # undef NO_LOWEST_ACTIVE_REG
8310 # undef CHAR_T
8311 # undef UCHAR_T
8312 # undef COMPILED_BUFFER_VAR
8313 # undef OFFSET_ADDRESS_SIZE
8314 # undef CHAR_CLASS_SIZE
8315 # undef PREFIX
8316 # undef ARG_PREFIX
8317 # undef PUT_CHAR
8318 # undef BYTE
8319 # undef WCHAR
8321 # define DEFINED_ONCE