| blob | 0915403450bd70dd958bd801a7885b2a9726065c |
1 /* Extended regular expression matching and search library,
2 version 0.12.
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
6 Copyright (C) 1993-2023 Free Software Foundation, Inc.
7 This file is part of the GNU C Library.
9 The GNU C Library is free software; you can redistribute it and/or
10 modify it under the terms of the GNU Lesser General Public
11 License as published by the Free Software Foundation; either
12 version 2.1 of the License, or (at your option) any later version.
14 The GNU C Library is distributed in the hope that it will be useful,
15 but WITHOUT ANY WARRANTY; without even the implied warranty of
16 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 Lesser General Public License for more details.
19 You should have received a copy of the GNU Lesser General Public
20 License along with the GNU C Library; if not, write to the Free
21 Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
22 02110-1301 USA. */
24 /* This file has been modified for usage in libiberty. It includes "xregex.h"
25 instead of <regex.h>. The "xregex.h" header file renames all external
26 routines with an "x" prefix so they do not collide with the native regex
27 routines or with other components regex routines. */
28 /* AIX requires this to be the first thing in the file. */
29 #if defined _AIX && !defined __GNUC__ && !defined REGEX_MALLOC
30 #pragma alloca
31 #endif
33 #if __GNUC__ >= 12
35 #endif
37 #undef _GNU_SOURCE
38 #define _GNU_SOURCE
40 #ifndef INSIDE_RECURSION
41 # ifdef HAVE_CONFIG_H
42 # include <config.h>
43 # endif
44 #endif
46 #include <ansidecl.h>
48 #ifndef INSIDE_RECURSION
50 # if defined STDC_HEADERS && !defined emacs
51 # include <stddef.h>
52 # define PTR_INT_TYPE ptrdiff_t
53 # else
54 /* We need this for `regex.h', and perhaps for the Emacs include files. */
55 # include <sys/types.h>
56 # define PTR_INT_TYPE long
57 # endif
59 # define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
61 /* For platform which support the ISO C amendement 1 functionality we
62 support user defined character classes. */
63 # if defined _LIBC || WIDE_CHAR_SUPPORT
64 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
65 # include <wchar.h>
66 # include <wctype.h>
67 # endif
69 # ifdef _LIBC
70 /* We have to keep the namespace clean. */
71 # define regfree(preg) __regfree (preg)
72 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
73 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
74 # define regerror(errcode, preg, errbuf, errbuf_size) \
75 __regerror(errcode, preg, errbuf, errbuf_size)
76 # define re_set_registers(bu, re, nu, st, en) \
77 __re_set_registers (bu, re, nu, st, en)
78 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
79 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
80 # define re_match(bufp, string, size, pos, regs) \
81 __re_match (bufp, string, size, pos, regs)
82 # define re_search(bufp, string, size, startpos, range, regs) \
83 __re_search (bufp, string, size, startpos, range, regs)
84 # define re_compile_pattern(pattern, length, bufp) \
85 __re_compile_pattern (pattern, length, bufp)
86 # define re_set_syntax(syntax) __re_set_syntax (syntax)
87 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
88 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
89 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
91 # define btowc __btowc
93 /* We are also using some library internals. */
94 # include <locale/localeinfo.h>
95 # include <locale/elem-hash.h>
96 # include <langinfo.h>
97 # include <locale/coll-lookup.h>
98 # endif
100 /* This is for other GNU distributions with internationalized messages. */
101 # if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
102 # include <libintl.h>
103 # ifdef _LIBC
104 # undef gettext
106 # endif
107 # else
108 # define gettext(msgid) (msgid)
109 # endif
111 # ifndef gettext_noop
112 /* This define is so xgettext can find the internationalizable
113 strings. */
114 # define gettext_noop(String) String
115 # endif
117 /* The `emacs' switch turns on certain matching commands
118 that make sense only in Emacs. */
119 # ifdef emacs
127 /* If we are not linking with Emacs proper,
128 we can't use the relocating allocator
129 even if config.h says that we can. */
130 # undef REL_ALLOC
132 # if defined STDC_HEADERS || defined _LIBC
133 # include <stdlib.h>
134 # else
137 # endif
139 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
140 If nothing else has been done, use the method below. */
141 # ifdef INHIBIT_STRING_HEADER
142 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
143 # if !defined bzero && !defined bcopy
144 # undef INHIBIT_STRING_HEADER
145 # endif
146 # endif
147 # endif
149 /* This is the normal way of making sure we have a bcopy and a bzero.
150 This is used in most programs--a few other programs avoid this
151 by defining INHIBIT_STRING_HEADER. */
152 # ifndef INHIBIT_STRING_HEADER
153 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
154 # include <string.h>
155 # ifndef bzero
156 # ifndef _LIBC
158 # else
159 # define bzero(s, n) __bzero (s, n)
160 # endif
161 # endif
162 # else
163 # include <strings.h>
164 # ifndef memcmp
165 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
166 # endif
167 # ifndef memcpy
168 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
169 # endif
170 # endif
171 # endif
173 /* Define the syntax stuff for \<, \>, etc. */
175 /* This must be nonzero for the wordchar and notwordchar pattern
176 commands in re_match_2. */
177 # ifndef Sword
178 # define Sword 1
179 # endif
181 # ifdef SWITCH_ENUM_BUG
182 # define SWITCH_ENUM_CAST(x) ((int)(x))
183 # else
184 # define SWITCH_ENUM_CAST(x) (x)
185 # endif
189 # if defined _LIBC || HAVE_LIMITS_H
190 # include <limits.h>
191 # endif
193 # ifndef MB_LEN_MAX
194 # define MB_LEN_MAX 1
195 # endif
196 \f
197 /* Get the interface, including the syntax bits. */
200 /* isalpha etc. are used for the character classes. */
201 # include <ctype.h>
203 /* Jim Meyering writes:
205 "... Some ctype macros are valid only for character codes that
206 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
207 using /bin/cc or gcc but without giving an ansi option). So, all
208 ctype uses should be through macros like ISPRINT... If
209 STDC_HEADERS is defined, then autoconf has verified that the ctype
210 macros don't need to be guarded with references to isascii. ...
211 Defining isascii to 1 should let any compiler worth its salt
212 eliminate the && through constant folding."
213 Solaris defines some of these symbols so we must undefine them first. */
215 # undef ISASCII
216 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
217 # define ISASCII(c) 1
218 # else
219 # define ISASCII(c) isascii(c)
220 # endif
222 # ifdef isblank
223 # define ISBLANK(c) (ISASCII (c) && isblank (c))
224 # else
226 # endif
227 # ifdef isgraph
228 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
229 # else
230 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
231 # endif
233 # undef ISPRINT
234 # define ISPRINT(c) (ISASCII (c) && isprint (c))
235 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
236 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
237 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
238 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
239 # define ISLOWER(c) (ISASCII (c) && islower (c))
240 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
241 # define ISSPACE(c) (ISASCII (c) && isspace (c))
242 # define ISUPPER(c) (ISASCII (c) && isupper (c))
243 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
245 # ifdef _tolower
246 # define TOLOWER(c) _tolower(c)
247 # else
248 # define TOLOWER(c) tolower(c)
249 # endif
251 # ifndef NULL
252 # define NULL (void *)0
253 # endif
255 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
256 since ours (we hope) works properly with all combinations of
257 machines, compilers, `char' and `unsigned char' argument types.
258 (Per Bothner suggested the basic approach.) */
259 # undef SIGN_EXTEND_CHAR
260 # if __STDC__
261 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
263 /* As in Harbison and Steele. */
264 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
265 # endif
266 \f
267 # ifndef emacs
268 /* How many characters in the character set. */
269 # define CHAR_SET_SIZE 256
271 # ifdef SYNTAX_TABLE
281 static void
283 {
298 }
302 # define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
305 \f
306 /* Integer type for pointers. */
307 # if !defined _LIBC && !defined HAVE_UINTPTR_T
309 # endif
311 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
312 use `alloca' instead of `malloc'. This is because using malloc in
313 re_search* or re_match* could cause memory leaks when C-g is used in
314 Emacs; also, malloc is slower and causes storage fragmentation. On
315 the other hand, malloc is more portable, and easier to debug.
317 Because we sometimes use alloca, some routines have to be macros,
318 not functions -- `alloca'-allocated space disappears at the end of the
319 function it is called in. */
321 # ifdef REGEX_MALLOC
323 # define REGEX_ALLOCATE malloc
324 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
325 # define REGEX_FREE free
329 /* Emacs already defines alloca, sometimes. */
330 # ifndef alloca
332 /* Make alloca work the best possible way. */
333 # ifdef __GNUC__
334 # define alloca __builtin_alloca
336 # if HAVE_ALLOCA_H
337 # include <alloca.h>
343 # define REGEX_ALLOCATE alloca
345 /* Assumes a `char *destination' variable. */
346 # define REGEX_REALLOCATE(source, osize, nsize) \
347 (destination = (char *) alloca (nsize), \
348 memcpy (destination, source, osize))
350 /* No need to do anything to free, after alloca. */
355 /* Define how to allocate the failure stack. */
357 # if defined REL_ALLOC && defined REGEX_MALLOC
359 # define REGEX_ALLOCATE_STACK(size) \
360 r_alloc (&failure_stack_ptr, (size))
361 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
362 r_re_alloc (&failure_stack_ptr, (nsize))
363 # define REGEX_FREE_STACK(ptr) \
364 r_alloc_free (&failure_stack_ptr)
368 # ifdef REGEX_MALLOC
370 # define REGEX_ALLOCATE_STACK malloc
371 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
372 # define REGEX_FREE_STACK free
376 # define REGEX_ALLOCATE_STACK alloca
378 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
379 REGEX_REALLOCATE (source, osize, nsize)
380 /* No need to explicitly free anything. */
381 # define REGEX_FREE_STACK(arg)
387 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
388 `string1' or just past its end. This works if PTR is NULL, which is
389 a good thing. */
390 # define FIRST_STRING_P(ptr) \
391 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
393 /* (Re)Allocate N items of type T using malloc, or fail. */
394 # define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
395 # define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
396 # define RETALLOC_IF(addr, n, t) \
397 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
398 # define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
402 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
404 # undef MAX
405 # undef MIN
406 # define MAX(a, b) ((a) > (b) ? (a) : (b))
407 # define MIN(a, b) ((a) < (b) ? (a) : (b))
410 # define false 0
411 # define true 1
414 reg_syntax_t syntax,
430 #ifdef MBS_SUPPORT
432 reg_syntax_t syntax,
451 #endif
452 \f
453 /* These are the command codes that appear in compiled regular
454 expressions. Some opcodes are followed by argument bytes. A
455 command code can specify any interpretation whatsoever for its
456 arguments. Zero bytes may appear in the compiled regular expression. */
459 {
462 /* Succeed right away--no more backtracking. */
463 succeed,
465 /* Followed by one byte giving n, then by n literal bytes. */
466 exactn,
468 # ifdef MBS_SUPPORT
469 /* Same as exactn, but contains binary data. */
470 exactn_bin,
471 # endif
473 /* Matches any (more or less) character. */
474 anychar,
476 /* Matches any one char belonging to specified set. First
477 following byte is number of bitmap bytes. Then come bytes
478 for a bitmap saying which chars are in. Bits in each byte
479 are ordered low-bit-first. A character is in the set if its
480 bit is 1. A character too large to have a bit in the map is
481 automatically not in the set. */
482 /* ifdef MBS_SUPPORT, following element is length of character
483 classes, length of collating symbols, length of equivalence
484 classes, length of character ranges, and length of characters.
485 Next, character class element, collating symbols elements,
486 equivalence class elements, range elements, and character
487 elements follow.
488 See regex_compile function. */
489 charset,
491 /* Same parameters as charset, but match any character that is
492 not one of those specified. */
493 charset_not,
495 /* Start remembering the text that is matched, for storing in a
496 register. Followed by one byte with the register number, in
497 the range 0 to one less than the pattern buffer's re_nsub
498 field. Then followed by one byte with the number of groups
499 inner to this one. (This last has to be part of the
500 start_memory only because we need it in the on_failure_jump
501 of re_match_2.) */
502 start_memory,
504 /* Stop remembering the text that is matched and store it in a
505 memory register. Followed by one byte with the register
506 number, in the range 0 to one less than `re_nsub' in the
507 pattern buffer, and one byte with the number of inner groups,
508 just like `start_memory'. (We need the number of inner
509 groups here because we don't have any easy way of finding the
510 corresponding start_memory when we're at a stop_memory.) */
511 stop_memory,
513 /* Match a duplicate of something remembered. Followed by one
514 byte containing the register number. */
515 duplicate,
517 /* Fail unless at beginning of line. */
518 begline,
520 /* Fail unless at end of line. */
521 endline,
523 /* Succeeds if at beginning of buffer (if emacs) or at beginning
524 of string to be matched (if not). */
525 begbuf,
527 /* Analogously, for end of buffer/string. */
528 endbuf,
530 /* Followed by two byte relative address to which to jump. */
531 jump,
533 /* Same as jump, but marks the end of an alternative. */
534 jump_past_alt,
536 /* Followed by two-byte relative address of place to resume at
537 in case of failure. */
538 /* ifdef MBS_SUPPORT, the size of address is 1. */
539 on_failure_jump,
541 /* Like on_failure_jump, but pushes a placeholder instead of the
542 current string position when executed. */
543 on_failure_keep_string_jump,
545 /* Throw away latest failure point and then jump to following
546 two-byte relative address. */
547 /* ifdef MBS_SUPPORT, the size of address is 1. */
548 pop_failure_jump,
550 /* Change to pop_failure_jump if know won't have to backtrack to
551 match; otherwise change to jump. This is used to jump
552 back to the beginning of a repeat. If what follows this jump
553 clearly won't match what the repeat does, such that we can be
554 sure that there is no use backtracking out of repetitions
555 already matched, then we change it to a pop_failure_jump.
556 Followed by two-byte address. */
557 /* ifdef MBS_SUPPORT, the size of address is 1. */
558 maybe_pop_jump,
560 /* Jump to following two-byte address, and push a dummy failure
561 point. This failure point will be thrown away if an attempt
562 is made to use it for a failure. A `+' construct makes this
563 before the first repeat. Also used as an intermediary kind
564 of jump when compiling an alternative. */
565 /* ifdef MBS_SUPPORT, the size of address is 1. */
566 dummy_failure_jump,
568 /* Push a dummy failure point and continue. Used at the end of
569 alternatives. */
570 push_dummy_failure,
572 /* Followed by two-byte relative address and two-byte number n.
573 After matching N times, jump to the address upon failure. */
574 /* ifdef MBS_SUPPORT, the size of address is 1. */
575 succeed_n,
577 /* Followed by two-byte relative address, and two-byte number n.
578 Jump to the address N times, then fail. */
579 /* ifdef MBS_SUPPORT, the size of address is 1. */
580 jump_n,
582 /* Set the following two-byte relative address to the
583 subsequent two-byte number. The address *includes* the two
584 bytes of number. */
585 /* ifdef MBS_SUPPORT, the size of address is 1. */
586 set_number_at,
595 notwordbound /* Succeeds if not at a word boundary. */
597 # ifdef emacs
602 /* Matches any character whose syntax is specified. Followed by
603 a byte which contains a syntax code, e.g., Sword. */
604 syntaxspec,
606 /* Matches any character whose syntax is not that specified. */
607 notsyntaxspec
611 \f
613 #ifdef BYTE
614 # define CHAR_T char
615 # define UCHAR_T unsigned char
616 # define COMPILED_BUFFER_VAR bufp->buffer
617 # define OFFSET_ADDRESS_SIZE 2
618 # define PREFIX(name) byte_##name
619 # define ARG_PREFIX(name) name
620 # define PUT_CHAR(c) putchar (c)
621 #else
622 # ifdef WCHAR
623 # define CHAR_T wchar_t
624 # define UCHAR_T wchar_t
625 # define COMPILED_BUFFER_VAR wc_buffer
627 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
628 # define PREFIX(name) wcs_##name
629 # define ARG_PREFIX(name) c##name
630 /* Should we use wide stream?? */
632 # define TRUE 1
633 # define FALSE 0
634 # else
635 # ifdef MBS_SUPPORT
636 # define WCHAR
637 # define INSIDE_RECURSION
639 # undef INSIDE_RECURSION
640 # endif
641 # define BYTE
642 # define INSIDE_RECURSION
644 # undef INSIDE_RECURSION
645 # endif
646 #endif
648 #ifdef INSIDE_RECURSION
649 /* Common operations on the compiled pattern. */
651 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
652 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
654 # ifdef WCHAR
655 # define STORE_NUMBER(destination, number) \
656 do { \
657 *(destination) = (UCHAR_T)(number); \
658 } while (0)
660 # define STORE_NUMBER(destination, number) \
661 do { \
662 (destination)[0] = (number) & 0377; \
663 (destination)[1] = (number) >> 8; \
664 } while (0)
667 /* Same as STORE_NUMBER, except increment DESTINATION to
668 the byte after where the number is stored. Therefore, DESTINATION
669 must be an lvalue. */
670 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
672 # define STORE_NUMBER_AND_INCR(destination, number) \
673 do { \
674 STORE_NUMBER (destination, number); \
675 (destination) += OFFSET_ADDRESS_SIZE; \
676 } while (0)
678 /* Put into DESTINATION a number stored in two contiguous bytes starting
679 at SOURCE. */
680 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
682 # ifdef WCHAR
683 # define EXTRACT_NUMBER(destination, source) \
684 do { \
685 (destination) = *(source); \
686 } while (0)
688 # define EXTRACT_NUMBER(destination, source) \
689 do { \
690 (destination) = *(source) & 0377; \
691 (destination) += ((unsigned) SIGN_EXTEND_CHAR (*((source) + 1))) << 8; \
692 } while (0)
693 # endif
695 # ifdef DEBUG
697 static void
699 {
700 # ifdef WCHAR
706 # endif
707 }
710 # undef EXTRACT_NUMBER
711 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
716 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
717 SOURCE must be an lvalue. */
719 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
720 do { \
721 EXTRACT_NUMBER (destination, source); \
722 (source) += OFFSET_ADDRESS_SIZE; \
723 } while (0)
725 # ifdef DEBUG
728 static void
730 {
733 }
735 # ifndef EXTRACT_MACROS
736 # undef EXTRACT_NUMBER_AND_INCR
737 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
738 PREFIX(extract_number_and_incr) (&dest, &src)
743 \f
745 /* If DEBUG is defined, Regex prints many voluminous messages about what
746 it is doing (if the variable `debug' is nonzero). If linked with the
747 main program in `iregex.c', you can enter patterns and strings
748 interactively. And if linked with the main program in `main.c' and
749 the other test files, you can run the already-written tests. */
751 # ifdef DEBUG
753 # ifndef DEFINED_ONCE
755 /* We use standard I/O for debugging. */
756 # include <stdio.h>
758 /* It is useful to test things that ``must'' be true when debugging. */
759 # include <assert.h>
763 # define DEBUG_STATEMENT(e) e
764 # define DEBUG_PRINT1(x) if (debug) printf (x)
765 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
766 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
767 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
770 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
771 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
772 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
773 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
776 /* Print the fastmap in human-readable form. */
778 # ifndef DEFINED_ONCE
779 void
781 {
786 {
788 {
792 {
794 i++;
795 }
797 {
800 }
801 }
802 }
804 }
808 /* Print a compiled pattern string in human-readable form, starting at
809 the START pointer into it and ending just before the pointer END. */
811 void
813 {
820 {
823 }
825 /* Loop over pattern commands. */
827 {
828 # ifdef _LIBC
830 # else
832 # endif
835 {
843 do
844 {
847 }
851 # ifdef MBS_SUPPORT
855 do
856 {
858 }
883 {
884 # ifdef WCHAR
895 {
901 }
904 {
910 }
913 {
917 }
922 # else
934 {
935 /* Are we starting a range? */
937 {
940 }
941 /* Have we broken a range? */
943 {
946 }
952 }
961 }
974 # ifdef _LIBC
976 # else
978 # endif
983 # ifdef _LIBC
985 # else
988 # endif
993 # ifdef _LIBC
995 # else
997 # endif
1006 # ifdef _LIBC
1008 # else
1010 # endif
1015 # ifdef _LIBC
1017 # else
1019 # endif
1024 # ifdef _LIBC
1026 # else
1028 # endif
1033 # ifdef _LIBC
1035 # else
1037 # endif
1044 # ifdef _LIBC
1046 # else
1049 # endif
1063 # ifdef _LIBC
1065 # else
1068 # endif
1087 # ifdef emacs
1131 }
1134 }
1136 # ifdef _LIBC
1138 # else
1140 # endif
1141 }
1144 void
1146 {
1155 {
1158 }
1160 # ifdef _LIBC
1162 # else
1164 # endif
1172 /* Perhaps we should print the translate table? */
1173 }
1176 void
1179 {
1184 else
1185 {
1189 {
1194 }
1198 {
1201 {
1204 }
1205 }
1206 }
1207 }
1209 # ifndef DEFINED_ONCE
1210 void
1212 {
1214 }
1215 # endif
1219 # ifndef DEFINED_ONCE
1220 # undef assert
1221 # define assert(e)
1223 # define DEBUG_STATEMENT(e)
1224 # define DEBUG_PRINT1(x)
1225 # define DEBUG_PRINT2(x1, x2)
1226 # define DEBUG_PRINT3(x1, x2, x3)
1227 # define DEBUG_PRINT4(x1, x2, x3, x4)
1229 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1230 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1234 \f
1236 # ifdef WCHAR
1237 /* This convert a multibyte string to a wide character string.
1238 And write their correspondances to offset_buffer(see below)
1239 and write whether each wchar_t is binary data to is_binary.
1240 This assume invalid multibyte sequences as binary data.
1241 We assume offset_buffer and is_binary is already allocated
1242 enough space. */
1247 static size_t
1250 /* It hold correspondances between src(char string) and
1251 dest(wchar_t string) for optimization.
1252 e.g. src = "xxxyzz"
1253 dest = {'X', 'Y', 'Z'}
1254 (each "xxx", "y" and "zz" represent one multibyte character
1255 corresponding to 'X', 'Y' and 'Z'.)
1256 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1257 = {0, 3, 4, 6}
1258 */
1259 {
1269 /* Initialize the conversion state. */
1275 {
1276 #ifdef _LIBC
1278 #else
1280 #endif
1283 /* failed to convert. maybe src contains binary data.
1284 So we consume 1 byte manualy. */
1285 {
1289 }
1290 else
1292 /* In sjis encoding, we use yen sign as escape character in
1293 place of reverse solidus. So we convert 0x5c(yen sign in
1294 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1295 solidus in UCS2). */
1300 }
1302 /* Fill remain of the buffer with sentinel. */
1307 }
1313 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1314 also be assigned to arbitrarily: each pattern buffer stores its own
1315 syntax, so it can be changed between regex compilations. */
1316 /* This has no initializer because initialized variables in Emacs
1317 become read-only after dumping. */
1318 reg_syntax_t re_syntax_options;
1321 /* Specify the precise syntax of regexps for compilation. This provides
1322 for compatibility for various utilities which historically have
1323 different, incompatible syntaxes.
1325 The argument SYNTAX is a bit mask comprised of the various bits
1326 defined in regex.h. We return the old syntax. */
1328 reg_syntax_t
1330 {
1334 # ifdef DEBUG
1341 }
1342 # ifdef _LIBC
1344 # endif
1345 \f
1346 /* This table gives an error message for each of the error codes listed
1347 in regex.h. Obviously the order here has to be same as there.
1348 POSIX doesn't require that we do anything for REG_NOERROR,
1349 but why not be nice? */
1352 {
1370 };
1371 \f
1374 #ifndef DEFINED_ONCE
1375 /* Avoiding alloca during matching, to placate r_alloc. */
1377 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1378 searching and matching functions should not call alloca. On some
1379 systems, alloca is implemented in terms of malloc, and if we're
1380 using the relocating allocator routines, then malloc could cause a
1381 relocation, which might (if the strings being searched are in the
1382 ralloc heap) shift the data out from underneath the regexp
1383 routines.
1385 Here's another reason to avoid allocation: Emacs
1386 processes input from X in a signal handler; processing X input may
1387 call malloc; if input arrives while a matching routine is calling
1388 malloc, then we're scrod. But Emacs can't just block input while
1389 calling matching routines; then we don't notice interrupts when
1390 they come in. So, Emacs blocks input around all regexp calls
1391 except the matching calls, which it leaves unprotected, in the
1392 faith that they will not malloc. */
1394 /* Normally, this is fine. */
1395 # define MATCH_MAY_ALLOCATE
1397 /* When using GNU C, we are not REALLY using the C alloca, no matter
1398 what config.h may say. So don't take precautions for it. */
1399 # ifdef __GNUC__
1400 # undef C_ALLOCA
1401 # endif
1403 /* The match routines may not allocate if (1) they would do it with malloc
1404 and (2) it's not safe for them to use malloc.
1405 Note that if REL_ALLOC is defined, matching would not use malloc for the
1406 failure stack, but we would still use it for the register vectors;
1407 so REL_ALLOC should not affect this. */
1408 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1409 # undef MATCH_MAY_ALLOCATE
1410 # endif
1412 \f
1413 #ifdef INSIDE_RECURSION
1414 /* Failure stack declarations and macros; both re_compile_fastmap and
1415 re_match_2 use a failure stack. These have to be macros because of
1416 REGEX_ALLOCATE_STACK. */
1419 /* Number of failure points for which to initially allocate space
1420 when matching. If this number is exceeded, we allocate more
1421 space, so it is not a hard limit. */
1422 # ifndef INIT_FAILURE_ALLOC
1423 # define INIT_FAILURE_ALLOC 5
1424 # endif
1426 /* Roughly the maximum number of failure points on the stack. Would be
1427 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1428 This is a variable only so users of regex can assign to it; we never
1429 change it ourselves. */
1431 # ifdef INT_IS_16BIT
1433 # ifndef DEFINED_ONCE
1434 # if defined MATCH_MAY_ALLOCATE
1435 /* 4400 was enough to cause a crash on Alpha OSF/1,
1436 whose default stack limit is 2mb. */
1438 # else
1440 # endif
1441 # endif
1444 {
1447 };
1452 {
1460 # ifndef DEFINED_ONCE
1461 # if defined MATCH_MAY_ALLOCATE
1462 /* 4400 was enough to cause a crash on Alpha OSF/1,
1463 whose default stack limit is 2mb. */
1465 # else
1467 # endif
1468 # endif
1471 {
1474 };
1479 {
1487 # ifndef DEFINED_ONCE
1488 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1489 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1490 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1491 # endif
1494 /* Define macros to initialize and free the failure stack.
1495 Do `return -2' if the alloc fails. */
1497 # ifdef MATCH_MAY_ALLOCATE
1498 # define INIT_FAIL_STACK() \
1499 do { \
1500 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1501 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1502 \
1503 if (fail_stack.stack == NULL) \
1504 return -2; \
1505 \
1506 fail_stack.size = INIT_FAILURE_ALLOC; \
1507 fail_stack.avail = 0; \
1508 } while (0)
1510 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1511 # else
1512 # define INIT_FAIL_STACK() \
1513 do { \
1514 fail_stack.avail = 0; \
1515 } while (0)
1517 # define RESET_FAIL_STACK()
1518 # endif
1521 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1523 Return 1 if succeeds, and 0 if either ran out of memory
1524 allocating space for it or it was already too large.
1526 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1528 # define DOUBLE_FAIL_STACK(fail_stack) \
1529 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1530 ? 0 \
1531 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1532 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1533 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1534 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1535 \
1536 (fail_stack).stack == NULL \
1537 ? 0 \
1538 : ((fail_stack).size <<= 1, \
1539 1)))
1542 /* Push pointer POINTER on FAIL_STACK.
1543 Return 1 if was able to do so and 0 if ran out of memory allocating
1544 space to do so. */
1545 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1546 ((FAIL_STACK_FULL () \
1547 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1548 ? 0 \
1549 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1550 1))
1552 /* Push a pointer value onto the failure stack.
1553 Assumes the variable `fail_stack'. Probably should only
1554 be called from within `PUSH_FAILURE_POINT'. */
1555 # define PUSH_FAILURE_POINTER(item) \
1556 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1558 /* This pushes an integer-valued item onto the failure stack.
1559 Assumes the variable `fail_stack'. Probably should only
1560 be called from within `PUSH_FAILURE_POINT'. */
1561 # define PUSH_FAILURE_INT(item) \
1562 fail_stack.stack[fail_stack.avail++].integer = (item)
1564 /* Push a fail_stack_elt_t value onto the failure stack.
1565 Assumes the variable `fail_stack'. Probably should only
1566 be called from within `PUSH_FAILURE_POINT'. */
1567 # define PUSH_FAILURE_ELT(item) \
1568 fail_stack.stack[fail_stack.avail++] = (item)
1570 /* These three POP... operations complement the three PUSH... operations.
1571 All assume that `fail_stack' is nonempty. */
1572 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1573 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1574 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1576 /* Used to omit pushing failure point id's when we're not debugging. */
1577 # ifdef DEBUG
1578 # define DEBUG_PUSH PUSH_FAILURE_INT
1579 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1580 # else
1581 # define DEBUG_PUSH(item)
1582 # define DEBUG_POP(item_addr)
1583 # endif
1586 /* Push the information about the state we will need
1587 if we ever fail back to it.
1589 Requires variables fail_stack, regstart, regend, reg_info, and
1590 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1591 be declared.
1593 Does `return FAILURE_CODE' if runs out of memory. */
1595 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1596 do { \
1597 char *destination; \
1598 /* Must be int, so when we don't save any registers, the arithmetic \
1600 /* Can't be int, since there is not a shred of a guarantee that int \
1601 is wide enough to hold a value of something to which pointer can \
1603 active_reg_t this_reg; \
1604 \
1605 DEBUG_STATEMENT (failure_id++); \
1606 DEBUG_STATEMENT (nfailure_points_pushed++); \
1610 \
1613 \
1615 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1616 { \
1617 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1618 return failure_code; \
1619 \
1621 (fail_stack).size); \
1623 } \
1624 \
1627 \
1628 if (1) \
1629 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1630 this_reg++) \
1631 { \
1633 DEBUG_STATEMENT (num_regs_pushed++); \
1634 \
1636 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1637 \
1639 PUSH_FAILURE_POINTER (regend[this_reg]); \
1640 \
1642 reg_info[this_reg].word.pointer); \
1644 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1647 MATCHED_SOMETHING (reg_info[this_reg])); \
1649 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1651 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1652 } \
1653 \
1655 PUSH_FAILURE_INT (lowest_active_reg); \
1656 \
1658 PUSH_FAILURE_INT (highest_active_reg); \
1659 \
1661 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1662 PUSH_FAILURE_POINTER (pattern_place); \
1663 \
1665 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1666 size2); \
1668 PUSH_FAILURE_POINTER (string_place); \
1669 \
1671 DEBUG_PUSH (failure_id); \
1672 } while (0)
1674 # ifndef DEFINED_ONCE
1675 /* This is the number of items that are pushed and popped on the stack
1676 for each register. */
1677 # define NUM_REG_ITEMS 3
1679 /* Individual items aside from the registers. */
1680 # ifdef DEBUG
1682 # else
1683 # define NUM_NONREG_ITEMS 4
1684 # endif
1686 /* We push at most this many items on the stack. */
1687 /* We used to use (num_regs - 1), which is the number of registers
1688 this regexp will save; but that was changed to 5
1689 to avoid stack overflow for a regexp with lots of parens. */
1690 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1692 /* We actually push this many items. */
1693 # define NUM_FAILURE_ITEMS \
1694 (((0 \
1695 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1696 * NUM_REG_ITEMS) \
1697 + NUM_NONREG_ITEMS)
1699 /* How many items can still be added to the stack without overflowing it. */
1700 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1704 /* Pops what PUSH_FAIL_STACK pushes.
1706 We restore into the parameters, all of which should be lvalues:
1707 STR -- the saved data position.
1708 PAT -- the saved pattern position.
1709 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1710 REGSTART, REGEND -- arrays of string positions.
1711 REG_INFO -- array of information about each subexpression.
1713 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1714 `pend', `string1', `size1', `string2', and `size2'. */
1715 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1716 { \
1717 DEBUG_STATEMENT (unsigned failure_id;) \
1718 active_reg_t this_reg; \
1719 const UCHAR_T *string_temp; \
1720 \
1721 assert (!FAIL_STACK_EMPTY ()); \
1722 \
1727 \
1728 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1729 \
1730 DEBUG_POP (&failure_id); \
1732 \
1733 /* If the saved string location is NULL, it came from an \
1734 on_failure_keep_string_jump opcode, and we want to throw away the \
1736 string_temp = POP_FAILURE_POINTER (); \
1737 if (string_temp != NULL) \
1738 str = (const CHAR_T *) string_temp; \
1739 \
1741 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1743 \
1744 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1746 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1747 \
1749 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1751 \
1752 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1754 \
1755 if (1) \
1756 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1757 { \
1759 \
1760 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1762 reg_info[this_reg].word.pointer); \
1763 \
1764 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1766 \
1767 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1769 } \
1770 else \
1771 { \
1772 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1773 { \
1774 reg_info[this_reg].word.integer = 0; \
1775 regend[this_reg] = 0; \
1776 regstart[this_reg] = 0; \
1777 } \
1778 highest_active_reg = high_reg; \
1779 } \
1780 \
1781 set_regs_matched_done = 0; \
1782 DEBUG_STATEMENT (nfailure_points_popped++); \
1784 \f
1785 /* Structure for per-register (a.k.a. per-group) information.
1786 Other register information, such as the
1787 starting and ending positions (which are addresses), and the list of
1788 inner groups (which is a bits list) are maintained in separate
1789 variables.
1791 We are making a (strictly speaking) nonportable assumption here: that
1792 the compiler will pack our bit fields into something that fits into
1793 the type of `word', i.e., is something that fits into one item on the
1794 failure stack. */
1797 /* Declarations and macros for re_match_2. */
1800 {
1802 struct
1803 {
1804 /* This field is one if this group can match the empty string,
1805 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1806 # define MATCH_NULL_UNSET_VALUE 3
1814 # ifndef DEFINED_ONCE
1815 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1816 # define IS_ACTIVE(R) ((R).bits.is_active)
1817 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1818 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1821 /* Call this when have matched a real character; it sets `matched' flags
1822 for the subexpressions which we are currently inside. Also records
1823 that those subexprs have matched. */
1824 # define SET_REGS_MATCHED() \
1825 do \
1826 { \
1827 if (!set_regs_matched_done) \
1828 { \
1829 active_reg_t r; \
1830 set_regs_matched_done = 1; \
1831 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1832 { \
1833 MATCHED_SOMETHING (reg_info[r]) \
1834 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1835 = 1; \
1836 } \
1837 } \
1838 } \
1839 while (0)
1842 /* Registers are set to a sentinel when they haven't yet matched. */
1844 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1845 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1847 /* Subroutine declarations and macros for regex_compile. */
1857 reg_syntax_t syntax);
1860 reg_syntax_t syntax);
1861 # ifdef WCHAR
1866 reg_syntax_t syntax,
1875 reg_syntax_t syntax,
1879 /* Fetch the next character in the uncompiled pattern---translating it
1880 if necessary. Also cast from a signed character in the constant
1881 string passed to us by the user to an unsigned char that we can use
1882 as an array index (in, e.g., `translate'). */
1883 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1884 because it is impossible to allocate 4GB array for some encodings
1885 which have 4 byte character_set like UCS4. */
1886 # ifndef PATFETCH
1887 # ifdef WCHAR
1888 # define PATFETCH(c) \
1889 do {if (p == pend) return REG_EEND; \
1890 c = (UCHAR_T) *p++; \
1891 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1892 } while (0)
1894 # define PATFETCH(c) \
1895 do {if (p == pend) return REG_EEND; \
1896 c = (unsigned char) *p++; \
1897 if (translate) c = (unsigned char) translate[c]; \
1898 } while (0)
1900 # endif
1902 /* Fetch the next character in the uncompiled pattern, with no
1903 translation. */
1904 # define PATFETCH_RAW(c) \
1905 do {if (p == pend) return REG_EEND; \
1906 c = (UCHAR_T) *p++; \
1907 } while (0)
1909 /* Go backwards one character in the pattern. */
1910 # define PATUNFETCH p--
1913 /* If `translate' is non-null, return translate[D], else just D. We
1914 cast the subscript to translate because some data is declared as
1915 `char *', to avoid warnings when a string constant is passed. But
1916 when we use a character as a subscript we must make it unsigned. */
1917 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1918 because it is impossible to allocate 4GB array for some encodings
1919 which have 4 byte character_set like UCS4. */
1921 # ifndef TRANSLATE
1922 # ifdef WCHAR
1923 # define TRANSLATE(d) \
1924 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1925 ? (char) translate[(unsigned char) (d)] : (d))
1927 # define TRANSLATE(d) \
1928 (translate ? (char) translate[(unsigned char) (d)] : (char) (d))
1930 # endif
1933 /* Macros for outputting the compiled pattern into `buffer'. */
1935 /* If the buffer isn't allocated when it comes in, use this. */
1936 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1938 /* Make sure we have at least N more bytes of space in buffer. */
1939 # ifdef WCHAR
1940 # define GET_BUFFER_SPACE(n) \
1941 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
1942 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
1943 EXTEND_BUFFER ()
1945 # define GET_BUFFER_SPACE(n) \
1946 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1947 EXTEND_BUFFER ()
1950 /* Make sure we have one more byte of buffer space and then add C to it. */
1951 # define BUF_PUSH(c) \
1952 do { \
1953 GET_BUFFER_SPACE (1); \
1954 *b++ = (UCHAR_T) (c); \
1955 } while (0)
1958 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1959 # define BUF_PUSH_2(c1, c2) \
1960 do { \
1961 GET_BUFFER_SPACE (2); \
1962 *b++ = (UCHAR_T) (c1); \
1963 *b++ = (UCHAR_T) (c2); \
1964 } while (0)
1967 /* As with BUF_PUSH_2, except for three bytes. */
1968 # define BUF_PUSH_3(c1, c2, c3) \
1969 do { \
1970 GET_BUFFER_SPACE (3); \
1971 *b++ = (UCHAR_T) (c1); \
1972 *b++ = (UCHAR_T) (c2); \
1973 *b++ = (UCHAR_T) (c3); \
1974 } while (0)
1976 /* Store a jump with opcode OP at LOC to location TO. We store a
1977 relative address offset by the three bytes the jump itself occupies. */
1978 # define STORE_JUMP(op, loc, to) \
1979 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
1981 /* Likewise, for a two-argument jump. */
1982 # define STORE_JUMP2(op, loc, to, arg) \
1983 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
1985 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1986 # define INSERT_JUMP(op, loc, to) \
1987 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
1989 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1990 # define INSERT_JUMP2(op, loc, to, arg) \
1991 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
1992 arg, b)
1994 /* This is not an arbitrary limit: the arguments which represent offsets
1995 into the pattern are two bytes long. So if 2^16 bytes turns out to
1996 be too small, many things would have to change. */
1997 /* Any other compiler which, like MSC, has allocation limit below 2^16
1998 bytes will have to use approach similar to what was done below for
1999 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
2000 reallocating to 0 bytes. Such thing is not going to work too well.
2001 You have been warned!! */
2002 # ifndef DEFINED_ONCE
2003 # if defined _MSC_VER && !defined WIN32
2004 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2005 The REALLOC define eliminates a flurry of conversion warnings,
2006 but is not required. */
2007 # define MAX_BUF_SIZE 65500L
2008 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2009 # else
2010 # define MAX_BUF_SIZE (1L << 16)
2011 # define REALLOC(p,s) realloc ((p), (s))
2012 # endif
2014 /* Extend the buffer by twice its current size via realloc and
2015 reset the pointers that pointed into the old block to point to the
2016 correct places in the new one. If extending the buffer results in it
2017 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2018 # if __BOUNDED_POINTERS__
2019 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2020 # define MOVE_BUFFER_POINTER(P) \
2021 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2022 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2023 else \
2024 { \
2025 SET_HIGH_BOUND (b); \
2026 SET_HIGH_BOUND (begalt); \
2027 if (fixup_alt_jump) \
2028 SET_HIGH_BOUND (fixup_alt_jump); \
2029 if (laststart) \
2030 SET_HIGH_BOUND (laststart); \
2031 if (pending_exact) \
2032 SET_HIGH_BOUND (pending_exact); \
2033 }
2034 # else
2035 # define MOVE_BUFFER_POINTER(P) (P) += incr
2036 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2037 # endif
2040 # ifdef WCHAR
2041 # define EXTEND_BUFFER() \
2042 do { \
2043 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2044 int wchar_count; \
2045 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2046 return REG_ESIZE; \
2047 bufp->allocated <<= 1; \
2048 if (bufp->allocated > MAX_BUF_SIZE) \
2049 bufp->allocated = MAX_BUF_SIZE; \
2051 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2052 if (wchar_count == 0) wchar_count = 1; \
2054 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2055 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2056 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2057 if (COMPILED_BUFFER_VAR == NULL) \
2058 return REG_ESPACE; \
2060 if (old_buffer != COMPILED_BUFFER_VAR) \
2061 { \
2062 PTR_INT_TYPE incr = COMPILED_BUFFER_VAR - old_buffer; \
2063 MOVE_BUFFER_POINTER (b); \
2064 MOVE_BUFFER_POINTER (begalt); \
2065 if (fixup_alt_jump) \
2066 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2067 if (laststart) \
2068 MOVE_BUFFER_POINTER (laststart); \
2069 if (pending_exact) \
2070 MOVE_BUFFER_POINTER (pending_exact); \
2071 } \
2072 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2073 } while (0)
2075 # define EXTEND_BUFFER() \
2076 do { \
2077 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2078 if (bufp->allocated == MAX_BUF_SIZE) \
2079 return REG_ESIZE; \
2080 bufp->allocated <<= 1; \
2081 if (bufp->allocated > MAX_BUF_SIZE) \
2082 bufp->allocated = MAX_BUF_SIZE; \
2083 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2084 bufp->allocated); \
2085 if (COMPILED_BUFFER_VAR == NULL) \
2086 return REG_ESPACE; \
2088 if (old_buffer != COMPILED_BUFFER_VAR) \
2089 { \
2090 PTR_INT_TYPE incr = COMPILED_BUFFER_VAR - old_buffer; \
2091 MOVE_BUFFER_POINTER (b); \
2092 MOVE_BUFFER_POINTER (begalt); \
2093 if (fixup_alt_jump) \
2094 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2095 if (laststart) \
2096 MOVE_BUFFER_POINTER (laststart); \
2097 if (pending_exact) \
2098 MOVE_BUFFER_POINTER (pending_exact); \
2099 } \
2100 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2101 } while (0)
2104 # ifndef DEFINED_ONCE
2105 /* Since we have one byte reserved for the register number argument to
2106 {start,stop}_memory, the maximum number of groups we can report
2107 things about is what fits in that byte. */
2108 # define MAX_REGNUM 255
2110 /* But patterns can have more than `MAX_REGNUM' registers. We just
2111 ignore the excess. */
2115 /* Macros for the compile stack. */
2117 /* Since offsets can go either forwards or backwards, this type needs to
2118 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2119 /* int may be not enough when sizeof(int) == 2. */
2123 {
2124 pattern_offset_t begalt_offset;
2125 pattern_offset_t fixup_alt_jump;
2126 pattern_offset_t inner_group_offset;
2127 pattern_offset_t laststart_offset;
2128 regnum_t regnum;
2133 {
2140 # define INIT_COMPILE_STACK_SIZE 32
2142 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2143 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2145 /* The next available element. */
2146 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2150 /* Set the bit for character C in a list. */
2151 # ifndef DEFINED_ONCE
2152 # define SET_LIST_BIT(c) \
2153 (b[((unsigned char) (c)) / BYTEWIDTH] \
2154 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2157 /* Get the next unsigned number in the uncompiled pattern. */
2158 # define GET_UNSIGNED_NUMBER(num) \
2159 { \
2160 while (p != pend) \
2161 { \
2162 PATFETCH (c); \
2164 break; \
2165 if (num <= RE_DUP_MAX) \
2166 { \
2167 if (num < 0) \
2168 num = 0; \
2170 } \
2171 } \
2172 }
2174 # ifndef DEFINED_ONCE
2175 # if defined _LIBC || WIDE_CHAR_SUPPORT
2176 /* The GNU C library provides support for user-defined character classes
2177 and the functions from ISO C amendement 1. */
2178 # ifdef CHARCLASS_NAME_MAX
2179 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2180 # else
2181 /* This shouldn't happen but some implementation might still have this
2182 problem. Use a reasonable default value. */
2183 # define CHAR_CLASS_MAX_LENGTH 256
2184 # endif
2186 # ifdef _LIBC
2187 # define IS_CHAR_CLASS(string) __wctype (string)
2188 # else
2189 # define IS_CHAR_CLASS(string) wctype (string)
2190 # endif
2191 # else
2194 # define IS_CHAR_CLASS(string) \
2201 # endif
2203 \f
2204 # ifndef MATCH_MAY_ALLOCATE
2206 /* If we cannot allocate large objects within re_match_2_internal,
2207 we make the fail stack and register vectors global.
2208 The fail stack, we grow to the maximum size when a regexp
2209 is compiled.
2210 The register vectors, we adjust in size each time we
2211 compile a regexp, according to the number of registers it needs. */
2215 /* Size with which the following vectors are currently allocated.
2216 That is so we can make them bigger as needed,
2217 but never make them smaller. */
2218 # ifdef DEFINED_ONCE
2230 /* Make the register vectors big enough for NUM_REGS registers,
2231 but don't make them smaller. */
2233 static void
2235 {
2237 {
2249 }
2250 }
2253 \f
2254 # ifndef DEFINED_ONCE
2256 regnum_t regnum);
2259 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2260 Returns one of error codes defined in `regex.h', or zero for success.
2262 Assumes the `allocated' (and perhaps `buffer') and `translate'
2263 fields are set in BUFP on entry.
2265 If it succeeds, results are put in BUFP (if it returns an error, the
2266 contents of BUFP are undefined):
2267 `buffer' is the compiled pattern;
2268 `syntax' is set to SYNTAX;
2269 `used' is set to the length of the compiled pattern;
2270 `fastmap_accurate' is zero;
2271 `re_nsub' is the number of subexpressions in PATTERN;
2272 `not_bol' and `not_eol' are zero;
2274 The `fastmap' and `newline_anchor' fields are neither
2275 examined nor set. */
2277 /* Return, freeing storage we allocated. */
2278 # ifdef WCHAR
2279 # define FREE_STACK_RETURN(value) \
2280 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2281 # else
2282 # define FREE_STACK_RETURN(value) \
2283 return (free (compile_stack.stack), value)
2286 static reg_errcode_t
2290 {
2291 /* We fetch characters from PATTERN here. Even though PATTERN is
2292 `char *' (i.e., signed), we declare these variables as unsigned, so
2293 they can be reliably used as array indices. */
2296 #ifdef WCHAR
2297 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2300 /* offset buffer for optimization. See convert_mbs_to_wc. */
2302 /* It hold whether each wchar_t is binary data or not. */
2304 /* A flag whether exactn is handling binary data or not. */
2308 /* A random temporary spot in PATTERN. */
2311 /* Points to the end of the buffer, where we should append. */
2314 /* Keeps track of unclosed groups. */
2315 compile_stack_type compile_stack;
2317 /* Points to the current (ending) position in the pattern. */
2318 #ifdef WCHAR
2326 /* How to translate the characters in the pattern. */
2329 /* Address of the count-byte of the most recently inserted `exactn'
2330 command. This makes it possible to tell if a new exact-match
2331 character can be added to that command or if the character requires
2332 a new `exactn' command. */
2335 /* Address of start of the most recently finished expression.
2336 This tells, e.g., postfix * where to find the start of its
2337 operand. Reset at the beginning of groups and alternatives. */
2340 /* Address of beginning of regexp, or inside of last group. */
2343 /* Address of the place where a forward jump should go to the end of
2344 the containing expression. Each alternative of an `or' -- except the
2345 last -- ends with a forward jump of this sort. */
2348 /* Counts open-groups as they are encountered. Remembered for the
2349 matching close-group on the compile stack, so the same register
2350 number is put in the stop_memory as the start_memory. */
2353 #ifdef WCHAR
2354 /* Initialize the wchar_t PATTERN and offset_buffer. */
2359 {
2364 }
2369 {
2374 }
2375 #endif
2377 #ifdef DEBUG
2380 {
2386 }
2389 /* Initialize the compile stack. */
2392 {
2393 #ifdef WCHAR
2397 #endif
2399 }
2404 /* Initialize the pattern buffer. */
2409 /* Set `used' to zero, so that if we return an error, the pattern
2410 printer (for debugging) will think there's no pattern. We reset it
2411 at the end. */
2414 /* Always count groups, whether or not bufp->no_sub is set. */
2417 #if !defined emacs && !defined SYNTAX_TABLE
2418 /* Initialize the syntax table. */
2420 #endif
2423 {
2426 enough space. This loses if buffer's address is bogus, but
2427 that is the user's responsibility. */
2428 #ifdef WCHAR
2429 /* Free bufp->buffer and allocate an array for wchar_t pattern
2430 buffer. */
2433 UCHAR_T);
2434 #else
2437 }
2438 else
2441 UCHAR_T);
2442 }
2445 #ifdef WCHAR
2449 }
2450 #ifdef WCHAR
2451 else
2453 #endif
2457 /* Loop through the uncompiled pattern until we're at the end. */
2459 {
2463 {
2465 {
2468 /* If context independent, it's an operator. */
2470 /* Otherwise, depends on what's come before. */
2473 else
2475 }
2480 {
2482 p == pend
2483 /* If context independent, it's an operator. */
2485 /* Otherwise, depends on what's next. */
2488 else
2490 }
2499 /* Fall through. */
2500 handle_plus:
2502 /* If there is no previous pattern... */
2504 {
2509 }
2511 {
2512 /* Are we optimizing this jump? */
2515 /* 1 means zero (many) matches is allowed. */
2518 /* If there is a sequence of repetition chars, collapse it
2519 down to just one (the right one). We can't combine
2520 interval operators with these because of, e.g., `a{2}*',
2521 which should only match an even number of `a's. */
2524 {
2535 ;
2538 {
2543 {
2544 PATUNFETCH;
2545 PATUNFETCH;
2547 }
2550 }
2551 else
2552 {
2553 PATUNFETCH;
2555 }
2557 /* If we get here, we found another repeat character. */
2558 }
2560 /* Star, etc. applied to an empty pattern is equivalent
2561 to an empty pattern. */
2565 /* Now we know whether or not zero matches is allowed
2566 and also whether or not two or more matches is allowed. */
2569 end a backward relative jump from `b' to before the next
2570 jump we're going to put in below (which jumps from
2571 laststart to after this jump).
2573 But if we are at the `*' in the exact sequence `.*\n',
2574 insert an unconditional jump backwards to the .,
2575 instead of the beginning of the loop. This way we only
2576 push a failure point once, instead of every time
2577 through the loop. */
2580 /* Allocate the space for the jump. */
2583 /* We know we are not at the first character of the pattern,
2584 because laststart was nonzero. And we've already
2585 incremented `p', by the way, to be the character after
2586 the `*'. Do we have to do something analogous here
2587 for null bytes, because of RE_DOT_NOT_NULL? */
2589 && zero_times_ok
2595 }
2596 else
2597 /* Anything else. */
2601 /* We've added more stuff to the buffer. */
2603 }
2605 /* On failure, jump from laststart to b + 3, which will be the
2606 end of the buffer after this jump is inserted. */
2607 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2608 'b + 3'. */
2617 {
2618 /* At least one repetition is required, so insert a
2619 `dummy_failure_jump' before the initial
2620 `on_failure_jump' instruction of the loop. This
2621 effects a skip over that instruction the first time
2622 we hit that loop. */
2627 }
2628 }
2639 {
2641 #ifdef WCHAR
2643 #else
2645 #endif
2648 #ifdef WCHAR
2649 /* We assume a charset(_not) structure as a wchar_t array.
2650 charset[0] = (re_opcode_t) charset(_not)
2651 charset[1] = l (= length of char_classes)
2652 charset[2] = m (= length of collating_symbols)
2653 charset[3] = n (= length of equivalence_classes)
2654 charset[4] = o (= length of char_ranges)
2655 charset[5] = p (= length of chars)
2657 charset[6] = char_class (wctype_t)
2658 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2659 ...
2660 charset[l+5] = char_class (wctype_t)
2662 charset[l+6] = collating_symbol (wchar_t)
2663 ...
2664 charset[l+m+5] = collating_symbol (wchar_t)
2665 ifdef _LIBC we use the index if
2666 _NL_COLLATE_SYMB_EXTRAMB instead of
2667 wchar_t string.
2669 charset[l+m+6] = equivalence_classes (wchar_t)
2670 ...
2671 charset[l+m+n+5] = equivalence_classes (wchar_t)
2672 ifdef _LIBC we use the index in
2673 _NL_COLLATE_WEIGHT instead of
2674 wchar_t string.
2676 charset[l+m+n+6] = range_start
2677 charset[l+m+n+7] = range_end
2678 ...
2679 charset[l+m+n+2o+4] = range_start
2680 charset[l+m+n+2o+5] = range_end
2681 ifdef _LIBC we use the value looked up
2682 in _NL_COLLATE_COLLSEQ instead of
2683 wchar_t character.
2685 charset[l+m+n+2o+6] = char
2686 ...
2687 charset[l+m+n+2o+p+5] = char
2689 */
2691 /* We need at least 6 spaces: the opcode, the length of
2692 char_classes, the length of collating_symbols, the length of
2693 equivalence_classes, the length of char_ranges, the length of
2694 chars. */
2697 /* Save b as laststart. And We use laststart as the pointer
2698 to the first element of the charset here.
2699 In other words, laststart[i] indicates charset[i]. */
2702 /* We test `*p == '^' twice, instead of using an if
2703 statement, so we only need one BUF_PUSH. */
2706 p++;
2708 /* Push the length of char_classes, the length of
2709 collating_symbols, the length of equivalence_classes, the
2710 length of char_ranges and the length of chars. */
2714 /* Remember the first position in the bracket expression. */
2717 /* charset_not matches newline according to a syntax bit. */
2720 {
2723 }
2725 /* Read in characters and ranges, setting map bits. */
2727 {
2732 /* \ might escape characters inside [...] and [^...]. */
2734 {
2742 }
2744 /* Could be the end of the bracket expression. If it's
2745 not (i.e., when the bracket expression is `[]' so
2746 far), the ']' character bit gets set way below. */
2750 /* Look ahead to see if it's a range when the last thing
2751 was a character class. */
2755 /* Look ahead to see if it's a range when the last thing
2756 was a character: if this is a hyphen not at the
2757 beginning or the end of a list, then it's the range
2758 operator. */
2763 {
2764 reg_errcode_t ret;
2765 /* Allocate the space for range_start and range_end. */
2767 /* Update the pointer to indicate end of buffer. */
2773 }
2776 reg_errcode_t ret;
2778 /* Move past the `-'. */
2780 /* Allocate the space for range_start and range_end. */
2782 /* Update the pointer to indicate end of buffer. */
2785 laststart);
2788 }
2790 /* See if we're at the beginning of a possible character
2791 class. */
2799 /* If pattern is `[[:'. */
2803 {
2809 else
2810 /* This is in any case an invalid class name. */
2812 }
2815 /* If isn't a word bracketed by `[:' and `:]':
2816 undo the ending character, the letters, and leave
2817 the leading `:' and `[' (but store them as character). */
2819 {
2823 /* Query the character class as wctype_t. */
2828 /* Throw away the ] at the end of the character
2829 class. */
2834 /* Allocate the space for character class. */
2836 /* Update the pointer to indicate end of buffer. */
2838 /* Move data which follow character classes
2839 not to violate the data. */
2846 /* Store the character class. */
2848 /* Update length of char_classes */
2852 }
2853 else
2854 {
2855 c1++;
2857 PATUNFETCH;
2863 }
2864 }
2867 {
2870 # ifdef _LIBC
2873 # endif
2877 /* If pattern is `[[=' or '[[.'. */
2881 {
2887 else
2888 /* This is in any case an invalid class name. */
2890 }
2894 {
2896 /* If we have no collation data we use the default
2897 collation in which each character is in a class
2898 by itself. It also means that ASCII is the
2899 character set and therefore we cannot have character
2900 with more than one byte in the multibyte
2901 representation. */
2903 /* If not defined _LIBC, we push the name and
2904 `\0' for the sake of matching performance. */
2907 # ifdef _LIBC
2910 # endif
2911 {
2914 }
2915 # ifdef _LIBC
2916 else
2917 {
2924 /* This #include defines a local function! */
2925 # include <locale/weightwc.h>
2928 {
2929 /* We push the index for equivalence class. */
2934 _NL_COLLATE_TABLEWC);
2937 _NL_COLLATE_WEIGHTWC);
2940 _NL_COLLATE_EXTRAWC);
2943 _NL_COLLATE_INDIRECTWC);
2947 /* This is no valid character. */
2951 }
2953 {
2954 /* We push collation sequence value
2955 for collating symbol. */
2965 /* We have to convert the name to a single-byte
2966 string. This is possible since the names
2967 consist of ASCII characters and the internal
2968 representation is UCS4. */
2972 table_size =
2974 _NL_COLLATE_SYMB_HASH_SIZEMB);
2977 _NL_COLLATE_SYMB_TABLEMB);
2980 _NL_COLLATE_SYMB_EXTRAMB);
2982 /* Locate the character in the hashing table. */
2989 {
2990 /* First compare the hashing value. */
2996 {
2997 /* Yep, this is the entry. */
3001 }
3003 /* Next entry. */
3005 }
3008 {
3009 /* Compute the index of the byte sequence
3010 in the table. */
3012 /* Adjust for the alignment. */
3016 }
3018 {
3019 /* No valid character. Match it as a
3020 single byte character. */
3023 /* Update the length of characters */
3027 /* Throw away the ] at the end of the
3028 collating symbol. */
3030 /* exit from the switch block. */
3032 }
3033 else
3035 }
3037 }
3038 # endif
3039 /* Throw away the ] at the end of the equivalence
3040 class (or collating symbol). */
3043 /* Allocate the space for the equivalence class
3044 (or collating symbol) (and '\0' if needed). */
3046 /* Update the pointer to indicate end of buffer. */
3051 /* Calculate the offset of char_ranges,
3052 which is next to equivalence_classes. */
3055 /* Insert space. */
3058 /* Write the equivalence_class and \0. */
3062 /* Update the length of equivalence_classes. */
3065 }
3068 /* Calculate the offset of the equivalence_classes,
3069 which is next to collating_symbols. */
3071 /* Insert space and write the collationg_symbol
3072 and \0. */
3077 /* In re_match_2_internal if range_start < -1, we
3078 assume -range_start is the offset of the
3079 collating symbol which is specified as
3080 the character of the range start. So we assign
3081 -(laststart[1] + laststart[2] + 6) to
3082 range_start. */
3084 /* Update the length of collating_symbol. */
3087 }
3088 }
3089 else
3090 {
3091 c1++;
3093 PATUNFETCH;
3099 }
3100 }
3101 else
3102 {
3107 }
3108 }
3111 /* Ensure that we have enough space to push a charset: the
3112 opcode, the length count, and the bitset; 34 bytes in all. */
3117 /* We test `*p == '^' twice, instead of using an if
3118 statement, so we only need one BUF_PUSH. */
3121 p++;
3123 /* Remember the first position in the bracket expression. */
3126 /* Push the number of bytes in the bitmap. */
3129 /* Clear the whole map. */
3132 /* charset_not matches newline according to a syntax bit. */
3137 /* Read in characters and ranges, setting map bits. */
3139 {
3144 /* \ might escape characters inside [...] and [^...]. */
3146 {
3153 }
3155 /* Could be the end of the bracket expression. If it's
3156 not (i.e., when the bracket expression is `[]' so
3157 far), the ']' character bit gets set way below. */
3161 /* Look ahead to see if it's a range when the last thing
3162 was a character class. */
3166 /* Look ahead to see if it's a range when the last thing
3167 was a character: if this is a hyphen not at the
3168 beginning or the end of a list, then it's the range
3169 operator. */
3174 {
3175 reg_errcode_t ret
3180 }
3184 reg_errcode_t ret;
3186 /* Move past the `-'. */
3192 }
3194 /* See if we're at the beginning of a possible character
3195 class. */
3204 /* If pattern is `[[:'. */
3208 {
3214 else
3215 /* This is in any case an invalid class name. */
3217 }
3220 /* If isn't a word bracketed by `[:' and `:]':
3221 undo the ending character, the letters, and leave
3222 the leading `:' and `[' (but set bits for them). */
3224 {
3225 # if defined _LIBC || WIDE_CHAR_SUPPORT
3235 /* Throw away the ] at the end of the character
3236 class. */
3242 {
3243 # ifdef _LIBC
3246 # else
3249 # endif
3254 }
3257 # else
3275 /* Throw away the ] at the end of the character
3276 class. */
3282 {
3283 /* This was split into 3 if's to
3284 avoid an arbitrary limit in some compiler. */
3303 }
3306 }
3307 else
3308 {
3309 c1++;
3311 PATUNFETCH;
3316 }
3317 }
3319 {
3321 # ifdef _LIBC
3324 # endif
3329 /* If pattern is `[[='. */
3333 {
3339 else
3340 /* This is in any case an invalid class name. */
3342 }
3346 {
3347 /* If we have no collation data we use the default
3348 collation in which each character is in a class
3349 by itself. It also means that ASCII is the
3350 character set and therefore we cannot have character
3351 with more than one byte in the multibyte
3352 representation. */
3353 # ifdef _LIBC
3355 # endif
3356 {
3360 /* Throw away the ] at the end of the equivalence
3361 class. */
3364 /* Set the bit for the character. */
3366 }
3367 # ifdef _LIBC
3368 else
3369 {
3370 /* Try to match the byte sequence in `str' against
3371 those known to the collate implementation.
3372 First find out whether the bytes in `str' are
3373 actually from exactly one character. */
3382 /* This #include defines a local function! */
3383 # include <locale/weight.h>
3396 /* This is no valid character. */
3399 /* Throw away the ] at the end of the equivalence
3400 class. */
3403 /* Now we have to go through the whole table
3404 and find all characters which have the same
3405 first level weight.
3407 XXX Note that this is not entirely correct.
3408 we would have to match multibyte sequences
3409 but this is not possible with the current
3410 implementation. */
3412 /* XXX This test would have to be changed if we
3413 would allow matching multibyte sequences. */
3415 {
3419 /* Test whether the lenghts match. */
3421 {
3422 /* They do. New compare the bytes of
3423 the weight. */
3432 /* They match. Mark the character as
3433 acceptable. */
3435 }
3436 }
3437 }
3438 # endif
3440 }
3441 else
3442 {
3443 c1++;
3445 PATUNFETCH;
3450 }
3451 }
3453 {
3455 # ifdef _LIBC
3458 # endif
3463 /* If pattern is `[[.'. */
3467 {
3473 else
3474 /* This is in any case an invalid class name. */
3476 }
3480 {
3481 /* If we have no collation data we use the default
3482 collation in which each character is the name
3483 for its own class which contains only the one
3484 character. It also means that ASCII is the
3485 character set and therefore we cannot have character
3486 with more than one byte in the multibyte
3487 representation. */
3488 # ifdef _LIBC
3490 # endif
3491 {
3495 /* Throw away the ] at the end of the equivalence
3496 class. */
3499 /* Set the bit for the character. */
3502 }
3503 # ifdef _LIBC
3504 else
3505 {
3506 /* Try to match the byte sequence in `str' against
3507 those known to the collate implementation.
3508 First find out whether the bytes in `str' are
3509 actually from exactly one character. */
3518 table_size =
3520 _NL_COLLATE_SYMB_HASH_SIZEMB);
3523 _NL_COLLATE_SYMB_TABLEMB);
3526 _NL_COLLATE_SYMB_EXTRAMB);
3528 /* Locate the character in the hashing table. */
3535 {
3536 /* First compare the hashing value. */
3543 {
3544 /* Yep, this is the entry. */
3548 }
3550 /* Next entry. */
3552 }
3555 /* This is no valid character. */
3558 /* Throw away the ] at the end of the equivalence
3559 class. */
3562 /* Now add the multibyte character(s) we found
3563 to the accept list.
3565 XXX Note that this is not entirely correct.
3566 we would have to match multibyte sequences
3567 but this is not possible with the current
3568 implementation. Also, we have to match
3569 collating symbols, which expand to more than
3570 one file, as a whole and not allow the
3571 individual bytes. */
3576 {
3579 }
3580 }
3581 # endif
3583 }
3584 else
3585 {
3586 c1++;
3588 PATUNFETCH;
3593 }
3594 }
3595 else
3596 {
3600 }
3601 }
3603 /* Discard any (non)matching list bytes that are all 0 at the
3604 end of the map. Decrease the map-length byte too. */
3609 }
3616 else
3623 else
3630 else
3637 else
3644 else
3651 /* Do not translate the character after the \, so that we can
3652 distinguish, e.g., \B from \b, even if we normally would
3653 translate, e.g., B to b. */
3657 {
3662 handle_open:
3664 regnum++;
3667 {
3669 compile_stack_elt_t);
3673 }
3675 /* These are the values to restore when we hit end of this
3676 group. They are all relative offsets, so that if the
3677 whole pattern moves because of realloc, they will still
3678 be valid. */
3680 COMPILE_STACK_TOP.fixup_alt_jump
3685 /* We will eventually replace the 0 with the number of
3686 groups inner to this one. But do not push a
3687 start_memory for groups beyond the last one we can
3688 represent in the compiled pattern. */
3690 {
3694 }
3701 /* If we've reached MAX_REGNUM groups, then this open
3702 won't actually generate any code, so we'll have to
3703 clear pending_exact explicitly. */
3712 {
3715 else
3717 }
3719 handle_close:
3722 alternative for a possible future
3723 `pop_failure_jump' to pop. See comments at
3724 `push_dummy_failure' in `re_match_2'. */
3727 /* We allocated space for this jump when we assigned
3728 to `fixup_alt_jump', in the `handle_alt' case below. */
3730 }
3732 /* See similar code for backslashed left paren above. */
3734 {
3737 else
3739 }
3741 /* Since we just checked for an empty stack above, this
3742 ``can't happen''. */
3744 {
3745 /* We don't just want to restore into `regnum', because
3746 later groups should continue to be numbered higher,
3747 as in `(ab)c(de)' -- the second group is #2. */
3748 regnum_t this_group_regnum;
3752 fixup_alt_jump
3758 /* If we've reached MAX_REGNUM groups, then this open
3759 won't actually generate any code, so we'll have to
3760 clear pending_exact explicitly. */
3763 /* We're at the end of the group, so now we know how many
3764 groups were inside this one. */
3766 {
3767 UCHAR_T *inner_group_loc
3773 }
3774 }
3781 handle_alt:
3785 /* Insert before the previous alternative a jump which
3786 jumps to this alternative if the former fails. */
3793 /* The alternative before this one has a jump after it
3794 which gets executed if it gets matched. Adjust that
3795 jump so it will jump to this alternative's analogous
3796 jump (put in below, which in turn will jump to the next
3797 (if any) alternative's such jump, etc.). The last such
3798 jump jumps to the correct final destination. A picture:
3799 _____ _____
3800 | | | |
3801 | v | v
3802 a | b | c
3804 If we are at `b', then fixup_alt_jump right now points to a
3805 three-byte space after `a'. We'll put in the jump, set
3806 fixup_alt_jump to right after `b', and leave behind three
3807 bytes which we'll fill in when we get to after `c'. */
3812 /* Mark and leave space for a jump after this alternative,
3813 to be filled in later either by next alternative or
3814 when know we're at the end of a series of alternatives. */
3825 /* If \{ is a literal. */
3827 /* If we're at `\{' and it's not the open-interval
3828 operator. */
3832 handle_interval:
3833 {
3834 /* If got here, then the syntax allows intervals. */
3836 /* At least (most) this many matches must be made. */
3839 /* Place in the uncompiled pattern (i.e., just after
3840 the '{') to go back to if the interval is invalid. */
3849 {
3853 }
3854 else
3855 /* Interval such as `{1}' => match exactly once. */
3862 {
3866 }
3871 /* If it's invalid to have no preceding re. */
3873 {
3879 else
3881 }
3883 /* We just parsed a valid interval. */
3888 /* If the upper bound is zero, don't want to succeed at
3889 all; jump from `laststart' to `b + 3', which will be
3890 the end of the buffer after we insert the jump. */
3891 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3892 instead of 'b + 3'. */
3894 {
3899 }
3901 /* Otherwise, we have a nontrivial interval. When
3902 we're all done, the pattern will look like:
3903 set_number_at <jump count> <upper bound>
3904 set_number_at <succeed_n count> <lower bound>
3905 succeed_n <after jump addr> <succeed_n count>
3906 <body of loop>
3907 jump_n <succeed_n addr> <jump count>
3908 (The upper bound and `jump_n' are omitted if
3909 `upper_bound' is 1, though.) */
3910 else
3912 more at the end of the loop. */
3918 /* Initialize lower bound of the `succeed_n', even
3919 though it will be set during matching by its
3920 attendant `set_number_at' (inserted next),
3921 because `re_compile_fastmap' needs to know.
3922 Jump to the `jump_n' we might insert below. */
3929 /* Code to initialize the lower bound. Insert
3930 before the `succeed_n'. The `5' is the last two
3931 bytes of this `set_number_at', plus 3 bytes of
3932 the following `succeed_n'. */
3933 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3934 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3935 of the following `succeed_n'. */
3942 append a backward jump to the `succeed_n'
3943 that starts this interval.
3945 When we've reached this during matching,
3946 we'll have matched the interval once, so
3947 jump back only `upper_bound - 1' times. */
3953 /* The location we want to set is the second
3954 parameter of the `jump_n'; that is `b-2' as
3955 an absolute address. `laststart' will be
3956 the `set_number_at' we're about to insert;
3957 `laststart+3' the number to set, the source
3958 for the relative address. But we are
3959 inserting into the middle of the pattern --
3960 so everything is getting moved up by 5.
3961 Conclusion: (b - 2) - (laststart + 3) + 5,
3962 i.e., b - laststart.
3964 We insert this at the beginning of the loop
3965 so that if we fail during matching, we'll
3966 reinitialize the bounds. */
3971 }
3972 }
3976 invalid_interval:
3979 unfetch_interval:
3980 /* Match the characters as literals. */
3985 else
3987 }
3989 #ifdef emacs
3990 /* There is no way to specify the before_dot and after_dot
3991 operators. rms says this is ok. --karl */
4072 /* Can't back reference to a subexpression if inside of it. */
4085 else
4089 normal_backslash:
4090 /* You might think it would be useful for \ to mean
4091 not to translate; but if we don't translate it
4092 it will never match anything. */
4095 }
4100 /* Expects the character in `c'. */
4101 normal_char:
4102 /* If no exactn currently being built. */
4104 #ifdef WCHAR
4105 /* If last exactn handle binary(or character) and
4106 new exactn handle character(or binary). */
4110 /* If last exactn not at current position. */
4113 /* We have only one byte following the exactn for the count. */
4116 /* If followed by a repetition operator. */
4125 {
4126 /* Start building a new exactn. */
4130 #ifdef WCHAR
4131 /* Is this exactn binary data or character? */
4135 else
4137 #else
4141 }
4150 /* Through the pattern now. */
4158 /* If we don't want backtracking, force success
4159 the first time we reach the end of the compiled pattern. */
4163 #ifdef WCHAR
4167 #endif
4170 /* We have succeeded; set the length of the buffer. */
4171 #ifdef WCHAR
4173 #else
4175 #endif
4177 #ifdef DEBUG
4179 {
4182 }
4185 #ifndef MATCH_MAY_ALLOCATE
4186 /* Initialize the failure stack to the largest possible stack. This
4187 isn't necessary unless we're trying to avoid calling alloca in
4188 the search and match routines. */
4189 {
4192 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4193 is strictly greater than re_max_failures, the largest possible stack
4194 is 2 * re_max_failures failure points. */
4196 {
4199 # ifdef emacs
4201 fail_stack.stack
4204 else
4205 fail_stack.stack
4211 fail_stack.stack
4214 else
4215 fail_stack.stack
4220 }
4223 }
4229 /* Subroutines for `regex_compile'. */
4231 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4232 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4234 static void
4236 {
4239 }
4242 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4243 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4245 static void
4247 {
4251 }
4254 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4255 for OP followed by two-byte integer parameter ARG. */
4256 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4258 static void
4260 {
4268 }
4271 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4272 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4274 static void
4277 {
4285 }
4288 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4289 after an alternative or a begin-subexpression. We assume there is at
4290 least one character before the ^. */
4292 static boolean
4294 reg_syntax_t syntax)
4295 {
4299 return
4300 /* After a subexpression? */
4302 /* After an alternative? */
4304 }
4307 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4308 at least one character after the $, i.e., `P < PEND'. */
4310 static boolean
4312 reg_syntax_t syntax)
4313 {
4318 return
4319 /* Before a subexpression? */
4322 /* Before an alternative? */
4325 }
4329 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4330 false if it's not. */
4332 static boolean
4334 {
4339 this_element--)
4344 }
4347 #ifdef INSIDE_RECURSION
4349 #ifdef WCHAR
4350 /* This insert space, which size is "num", into the pattern at "loc".
4351 "end" must point the end of the allocated buffer. */
4352 static void
4354 {
4360 }
4363 #ifdef WCHAR
4364 static reg_errcode_t
4368 {
4371 reg_errcode_t ret;
4372 # ifdef _LIBC
4375 # endif
4379 # ifdef _LIBC
4382 {
4384 _NL_COLLATE_COLLSEQWC);
4389 {
4390 /* range_start is a collating symbol. */
4392 /* Retreive the index and get collation sequence value. */
4395 }
4396 else
4401 /* Report an error if the range is empty and the syntax prohibits
4402 this. */
4406 /* Insert space to the end of the char_ranges. */
4411 }
4412 else
4413 # endif
4414 {
4416 range_start_char;
4418 /* Report an error if the range is empty and the syntax prohibits
4419 this. */
4423 /* Insert space to the end of the char_ranges. */
4428 }
4429 /* Have to increment the pointer into the pattern string, so the
4430 caller isn't still at the ending character. */
4434 }
4436 /* Read the ending character of a range (in a bracket expression) from the
4437 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4438 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4439 Then we set the translation of all bits between the starting and
4440 ending characters (inclusive) in the compiled pattern B.
4442 Return an error code.
4444 We use these short variable names so we can use the same macros as
4445 `regex_compile' itself. */
4447 static reg_errcode_t
4451 {
4454 reg_errcode_t ret;
4455 # if _LIBC
4459 # else
4461 # endif
4466 /* Have to increment the pointer into the pattern string, so the
4467 caller isn't still at the ending character. */
4470 /* Report an error if the range is empty and the syntax prohibits this. */
4473 # if _LIBC
4475 _NL_COLLATE_COLLSEQMB);
4480 {
4484 {
4487 }
4488 }
4489 # else
4490 /* Here we see why `this_char' has to be larger than an `unsigned
4491 char' -- we would otherwise go into an infinite loop, since all
4492 characters <= 0xff. */
4494 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4495 and some compilers cast it to int implicitly, so following for_loop
4496 may fall to (almost) infinite loop.
4497 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4498 To avoid this, we cast p[0] to unsigned int and truncate it. */
4502 {
4505 }
4506 # endif
4509 }
4511 \f
4512 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4513 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4514 characters can start a string that matches the pattern. This fastmap
4515 is used by re_search to skip quickly over impossible starting points.
4517 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4518 area as BUFP->fastmap.
4520 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4521 the pattern buffer.
4523 Returns 0 if we succeed, -2 if an internal error. */
4525 #ifdef WCHAR
4526 /* local function for re_compile_fastmap.
4527 truncate wchar_t character to char. */
4530 static unsigned char
4532 {
4537 # ifdef _LIBC
4539 # else
4541 # endif
4543 }
4546 static int
4548 {
4550 #ifdef MATCH_MAY_ALLOCATE
4552 #endif
4553 #ifndef REGEX_MALLOC
4555 #endif
4559 #ifdef WCHAR
4560 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4561 pattern to (char*) in regex_compile. */
4570 #ifdef REL_ALLOC
4571 /* This holds the pointer to the failure stack, when
4572 it is allocated relocatably. */
4574 #endif
4576 /* Assume that each path through the pattern can be null until
4577 proven otherwise. We set this false at the bottom of switch
4578 statement, to which we get only if a particular path doesn't
4579 match the empty string. */
4582 /* We aren't doing a `succeed_n' to begin with. */
4593 {
4595 {
4596 /* We have reached the (effective) end of pattern. */
4598 {
4601 /* Reset for next path. */
4607 }
4608 else
4610 }
4612 /* We should never be about to go beyond the end of the pattern. */
4616 {
4618 /* I guess the idea here is to simply not bother with a fastmap
4619 if a backreference is used, since it's too hard to figure out
4620 the fastmap for the corresponding group. Setting
4621 `can_be_null' stops `re_search_2' from using the fastmap, so
4622 that is all we do. */
4628 /* Following are the cases which match a character. These end
4629 with `break'. */
4631 #ifdef WCHAR
4640 #ifdef MBS_SUPPORT
4644 #endif
4646 #ifdef WCHAR
4647 /* It is hard to distinguish fastmap from (multi byte) characters
4648 which depends on current locale. */
4664 /* Chars beyond end of map must be allowed. */
4689 {
4692 /* `.' matches anything ... */
4696 /* ... except perhaps newline. */
4700 /* Return if we have already set `can_be_null'; if we have,
4701 then the fastmap is irrelevant. Something's wrong here. */
4705 /* Otherwise, have to check alternative paths. */
4707 }
4709 #ifdef emacs
4726 /* All cases after this match the empty string. These end with
4727 `continue'. */
4761 /* Jump backward implies we just went through the body of a
4762 loop and matched nothing. Opcode jumped to should be
4763 `on_failure_jump' or `succeed_n'. Just treat it like an
4764 ordinary jump. For a * loop, it has pushed its failure
4765 point already; if so, discard that as redundant. */
4770 p++;
4774 /* If what's on the stack is where we are now, pop it. */
4784 handle_on_failure_jump:
4787 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4788 end of the pattern. We don't want to push such a point,
4789 since when we restore it above, entering the switch will
4790 increment `p' past the end of the pattern. We don't need
4791 to push such a point since we obviously won't find any more
4792 fastmap entries beyond `pend'. Such a pattern can match
4793 the null string, though. */
4795 {
4797 {
4800 }
4801 }
4802 else
4806 {
4809 }
4815 /* Get to the number of times to succeed. */
4818 /* Increment p past the n for when k != 0. */
4821 {
4825 }
4844 /* Getting here means we have found the possible starting
4845 characters for one path of the pattern -- and that the empty
4846 string does not match. We need not follow this path further.
4847 Instead, look at the next alternative (remembered on the
4848 stack), or quit if no more. The test at the top of the loop
4849 does these things. */
4854 /* Set `can_be_null' for the last path (also the first path, if the
4855 pattern is empty). */
4858 done:
4861 }
4865 int
4867 {
4868 # ifdef MBS_SUPPORT
4871 else
4872 # endif
4875 #ifdef _LIBC
4877 #endif
4878 \f
4880 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4881 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4882 this memory for recording register information. STARTS and ENDS
4883 must be allocated using the malloc library routine, and must each
4884 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4886 If NUM_REGS == 0, then subsequent matches should allocate their own
4887 register data.
4889 Unless this function is called, the first search or match using
4890 PATTERN_BUFFER will allocate its own register data, without
4891 freeing the old data. */
4893 void
4897 {
4899 {
4904 }
4905 else
4906 {
4910 }
4911 }
4912 #ifdef _LIBC
4914 #endif
4915 \f
4916 /* Searching routines. */
4918 /* Like re_search_2, below, but only one string is specified, and
4919 doesn't let you say where to stop matching. */
4921 int
4924 {
4927 }
4928 #ifdef _LIBC
4930 #endif
4933 /* Using the compiled pattern in BUFP->buffer, first tries to match the
4934 virtual concatenation of STRING1 and STRING2, starting first at index
4935 STARTPOS, then at STARTPOS + 1, and so on.
4937 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
4939 RANGE is how far to scan while trying to match. RANGE = 0 means try
4940 only at STARTPOS; in general, the last start tried is STARTPOS +
4941 RANGE.
4943 In REGS, return the indices of the virtual concatenation of STRING1
4944 and STRING2 that matched the entire BUFP->buffer and its contained
4945 subexpressions.
4947 Do not consider matching one past the index STOP in the virtual
4948 concatenation of STRING1 and STRING2.
4950 We return either the position in the strings at which the match was
4951 found, -1 if no match, or -2 if error (such as failure
4952 stack overflow). */
4954 int
4958 {
4959 # ifdef MBS_SUPPORT
4963 else
4964 # endif
4968 #ifdef _LIBC
4970 #endif
4974 #ifdef INSIDE_RECURSION
4976 #ifdef MATCH_MAY_ALLOCATE
4977 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
4978 #else
4979 # define FREE_VAR(var) free (var); var = NULL
4980 #endif
4982 #ifdef WCHAR
4983 # define MAX_ALLOCA_SIZE 2000
4985 # define FREE_WCS_BUFFERS() \
4986 do { \
4987 if (size1 > MAX_ALLOCA_SIZE) \
4988 { \
4989 free (wcs_string1); \
4990 free (mbs_offset1); \
4991 } \
4992 else \
4993 { \
4994 FREE_VAR (wcs_string1); \
4995 FREE_VAR (mbs_offset1); \
4996 } \
4997 if (size2 > MAX_ALLOCA_SIZE) \
4998 { \
4999 free (wcs_string2); \
5000 free (mbs_offset2); \
5001 } \
5002 else \
5003 { \
5004 FREE_VAR (wcs_string2); \
5005 FREE_VAR (mbs_offset2); \
5006 } \
5007 } while (0)
5009 #endif
5012 static int
5017 {
5023 #ifdef WCHAR
5024 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5026 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5028 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5030 /* They hold whether each wchar_t is binary data or not. */
5034 /* Check for out-of-range STARTPOS. */
5038 /* Fix up RANGE if it might eventually take us outside
5039 the virtual concatenation of STRING1 and STRING2.
5040 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5046 /* If the search isn't to be a backwards one, don't waste time in a
5047 search for a pattern that must be anchored. */
5050 /* `begline' is like `begbuf' if it cannot match at newlines. */
5053 {
5056 else
5058 }
5060 #ifdef emacs
5061 /* In a forward search for something that starts with \=.
5062 don't keep searching past point. */
5064 {
5068 }
5071 /* Update the fastmap now if not correct already. */
5076 #ifdef WCHAR
5077 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5078 fill them with converted string. */
5080 {
5082 {
5086 }
5087 else
5088 {
5092 }
5094 {
5096 {
5100 }
5101 else
5102 {
5106 }
5108 }
5114 else
5116 }
5118 {
5120 {
5124 }
5125 else
5126 {
5130 }
5132 {
5136 else
5139 }
5145 else
5147 }
5151 /* Loop through the string, looking for a place to start matching. */
5153 {
5154 /* If a fastmap is supplied, skip quickly over characters that
5155 cannot be the start of a match. If the pattern can match the
5156 null string, however, we don't need to skip characters; we want
5157 the first null string. */
5159 {
5161 {
5171 /* Written out as an if-else to avoid testing `translate'
5172 inside the loop. */
5177 range--;
5178 else
5180 range--;
5183 }
5185 {
5192 }
5193 }
5195 /* If can't match the null string, and that's all we have left, fail. */
5198 {
5199 #ifdef WCHAR
5201 #endif
5203 }
5205 #ifdef WCHAR
5216 #ifndef REGEX_MALLOC
5217 # ifdef C_ALLOCA
5219 # endif
5220 #endif
5223 {
5224 #ifdef WCHAR
5226 #endif
5228 }
5231 {
5232 #ifdef WCHAR
5234 #endif
5236 }
5238 advance:
5242 {
5243 range--;
5244 startpos++;
5245 }
5246 else
5247 {
5248 range++;
5249 startpos--;
5250 }
5251 }
5252 #ifdef WCHAR
5254 #endif
5256 }
5258 #ifdef WCHAR
5259 /* This converts PTR, a pointer into one of the search wchar_t strings
5260 `string1' and `string2' into an multibyte string offset from the
5261 beginning of that string. We use mbs_offset to optimize.
5262 See convert_mbs_to_wcs. */
5263 # define POINTER_TO_OFFSET(ptr) \
5264 (FIRST_STRING_P (ptr) \
5265 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5266 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5267 + csize1)))
5269 /* This converts PTR, a pointer into one of the search strings `string1'
5270 and `string2' into an offset from the beginning of that string. */
5271 # define POINTER_TO_OFFSET(ptr) \
5272 (FIRST_STRING_P (ptr) \
5273 ? ((regoff_t) ((ptr) - string1)) \
5274 : ((regoff_t) ((ptr) - string2 + size1)))
5277 /* Macros for dealing with the split strings in re_match_2. */
5279 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5281 /* Call before fetching a character with *d. This switches over to
5282 string2 if necessary. */
5283 #define PREFETCH() \
5284 while (d == dend) \
5285 { \
5287 if (dend == end_match_2) \
5288 goto fail; \
5290 d = string2; \
5291 dend = end_match_2; \
5292 }
5294 /* Test if at very beginning or at very end of the virtual concatenation
5295 of `string1' and `string2'. If only one string, it's `string2'. */
5296 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5297 #define AT_STRINGS_END(d) ((d) == end2)
5300 /* Test if D points to a character which is word-constituent. We have
5301 two special cases to check for: if past the end of string1, look at
5302 the first character in string2; and if before the beginning of
5303 string2, look at the last character in string1. */
5304 #ifdef WCHAR
5305 /* Use internationalized API instead of SYNTAX. */
5306 # define WORDCHAR_P(d) \
5307 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5308 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5309 || ((d) == end1 ? *string2 \
5312 # define WORDCHAR_P(d) \
5313 (SYNTAX ((d) == end1 ? *string2 \
5314 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5315 == Sword)
5318 /* Disabled due to a compiler bug -- see comment at case wordbound */
5319 #if 0
5320 /* Test if the character before D and the one at D differ with respect
5321 to being word-constituent. */
5322 #define AT_WORD_BOUNDARY(d) \
5323 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5324 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5325 #endif
5327 /* Free everything we malloc. */
5328 #ifdef MATCH_MAY_ALLOCATE
5329 # ifdef WCHAR
5330 # define FREE_VARIABLES() \
5331 do { \
5332 REGEX_FREE_STACK (fail_stack.stack); \
5333 FREE_VAR (regstart); \
5334 FREE_VAR (regend); \
5335 FREE_VAR (old_regstart); \
5336 FREE_VAR (old_regend); \
5337 FREE_VAR (best_regstart); \
5338 FREE_VAR (best_regend); \
5339 FREE_VAR (reg_info); \
5340 FREE_VAR (reg_dummy); \
5341 FREE_VAR (reg_info_dummy); \
5342 if (!cant_free_wcs_buf) \
5343 { \
5344 FREE_VAR (string1); \
5345 FREE_VAR (string2); \
5346 FREE_VAR (mbs_offset1); \
5347 FREE_VAR (mbs_offset2); \
5348 } \
5349 } while (0)
5351 # define FREE_VARIABLES() \
5352 do { \
5353 REGEX_FREE_STACK (fail_stack.stack); \
5354 FREE_VAR (regstart); \
5355 FREE_VAR (regend); \
5356 FREE_VAR (old_regstart); \
5357 FREE_VAR (old_regend); \
5358 FREE_VAR (best_regstart); \
5359 FREE_VAR (best_regend); \
5360 FREE_VAR (reg_info); \
5361 FREE_VAR (reg_dummy); \
5362 FREE_VAR (reg_info_dummy); \
5363 } while (0)
5365 #else
5366 # ifdef WCHAR
5367 # define FREE_VARIABLES() \
5368 do { \
5369 if (!cant_free_wcs_buf) \
5370 { \
5371 FREE_VAR (string1); \
5372 FREE_VAR (string2); \
5373 FREE_VAR (mbs_offset1); \
5374 FREE_VAR (mbs_offset2); \
5375 } \
5376 } while (0)
5382 /* These values must meet several constraints. They must not be valid
5383 register values; since we have a limit of 255 registers (because
5384 we use only one byte in the pattern for the register number), we can
5385 use numbers larger than 255. They must differ by 1, because of
5386 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5387 be larger than the value for the highest register, so we do not try
5388 to actually save any registers when none are active. */
5389 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5390 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5391 \f
5393 /* Matching routines. */
5396 /* re_match is like re_match_2 except it takes only a single string. */
5398 int
5401 {
5403 # ifdef MBS_SUPPORT
5408 else
5409 # endif
5412 # ifndef REGEX_MALLOC
5413 # ifdef C_ALLOCA
5415 # endif
5416 # endif
5418 }
5419 # ifdef _LIBC
5421 # endif
5426 #ifdef INSIDE_RECURSION
5440 /* re_match_2 matches the compiled pattern in BUFP against the
5441 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5442 and SIZE2, respectively). We start matching at POS, and stop
5443 matching at STOP.
5445 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5446 store offsets for the substring each group matched in REGS. See the
5447 documentation for exactly how many groups we fill.
5449 We return -1 if no match, -2 if an internal error (such as the
5450 failure stack overflowing). Otherwise, we return the length of the
5451 matched substring. */
5453 int
5457 {
5459 # ifdef MBS_SUPPORT
5464 else
5465 # endif
5469 #ifndef REGEX_MALLOC
5470 # ifdef C_ALLOCA
5472 # endif
5473 #endif
5475 }
5476 #ifdef _LIBC
5478 #endif
5482 #ifdef INSIDE_RECURSION
5484 #ifdef WCHAR
5487 /* This check the substring (from 0, to length) of the multibyte string,
5488 to which offset_buffer correspond. And count how many wchar_t_characters
5489 the substring occupy. We use offset_buffer to optimization.
5490 See convert_mbs_to_wcs. */
5492 static int
5494 {
5497 /* Check whether the size is valid. */
5504 /* If there are no multibyte character, offset_buffer[i] == i.
5505 Optmize for this case. */
5509 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5514 {
5522 else
5524 }
5527 }
5530 /* This is a separate function so that we can force an alloca cleanup
5531 afterwards. */
5532 #ifdef WCHAR
5533 static int
5540 /* string1 == string2 == NULL means string1/2, size1/2 and
5541 mbs_offset1/2 need seting up in this function. */
5542 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5545 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5548 static int
5555 {
5556 /* General temporaries. */
5559 #ifdef WCHAR
5560 /* They hold whether each wchar_t is binary data or not. */
5562 /* If true, we can't free string1/2, mbs_offset1/2. */
5566 /* Just past the end of the corresponding string. */
5569 /* Pointers into string1 and string2, just past the last characters in
5570 each to consider matching. */
5573 /* Where we are in the data, and the end of the current string. */
5576 /* Where we are in the pattern, and the end of the pattern. */
5577 #ifdef WCHAR
5585 /* Mark the opcode just after a start_memory, so we can test for an
5586 empty subpattern when we get to the stop_memory. */
5589 /* We use this to map every character in the string. */
5592 /* Failure point stack. Each place that can handle a failure further
5593 down the line pushes a failure point on this stack. It consists of
5594 restart, regend, and reg_info for all registers corresponding to
5595 the subexpressions we're currently inside, plus the number of such
5596 registers, and, finally, two char *'s. The first char * is where
5597 to resume scanning the pattern; the second one is where to resume
5598 scanning the strings. If the latter is zero, the failure point is
5599 a ``dummy''; if a failure happens and the failure point is a dummy,
5600 it gets discarded and the next next one is tried. */
5603 #endif
5604 #ifdef DEBUG
5607 #endif
5609 #ifdef REL_ALLOC
5610 /* This holds the pointer to the failure stack, when
5611 it is allocated relocatably. */
5613 #endif
5615 /* We fill all the registers internally, independent of what we
5616 return, for use in backreferences. The number here includes
5617 an element for register zero. */
5620 /* The currently active registers. */
5624 /* Information on the contents of registers. These are pointers into
5625 the input strings; they record just what was matched (on this
5626 attempt) by a subexpression part of the pattern, that is, the
5627 regnum-th regstart pointer points to where in the pattern we began
5628 matching and the regnum-th regend points to right after where we
5629 stopped matching the regnum-th subexpression. (The zeroth register
5630 keeps track of what the whole pattern matches.) */
5633 #endif
5635 /* If a group that's operated upon by a repetition operator fails to
5636 match anything, then the register for its start will need to be
5637 restored because it will have been set to wherever in the string we
5638 are when we last see its open-group operator. Similarly for a
5639 register's end. */
5642 #endif
5644 /* The is_active field of reg_info helps us keep track of which (possibly
5645 nested) subexpressions we are currently in. The matched_something
5646 field of reg_info[reg_num] helps us tell whether or not we have
5647 matched any of the pattern so far this time through the reg_num-th
5648 subexpression. These two fields get reset each time through any
5649 loop their register is in. */
5652 #endif
5654 /* The following record the register info as found in the above
5655 variables when we find a match better than any we've seen before.
5656 This happens as we backtrack through the failure points, which in
5657 turn happens only if we have not yet matched the entire string. */
5661 #endif
5663 /* Logically, this is `best_regend[0]'. But we don't want to have to
5664 allocate space for that if we're not allocating space for anything
5665 else (see below). Also, we never need info about register 0 for
5666 any of the other register vectors, and it seems rather a kludge to
5667 treat `best_regend' differently than the rest. So we keep track of
5668 the end of the best match so far in a separate variable. We
5669 initialize this to NULL so that when we backtrack the first time
5670 and need to test it, it's not garbage. */
5673 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5676 /* Used when we pop values we don't care about. */
5680 #endif
5682 #ifdef DEBUG
5683 /* Counts the total number of registers pushed. */
5685 #endif
5691 #ifdef MATCH_MAY_ALLOCATE
5692 /* Do not bother to initialize all the register variables if there are
5693 no groups in the pattern, as it takes a fair amount of time. If
5694 there are groups, we include space for register 0 (the whole
5695 pattern), even though we never use it, since it simplifies the
5696 array indexing. We should fix this. */
5698 {
5711 {
5714 }
5715 }
5716 else
5717 {
5718 /* We must initialize all our variables to NULL, so that
5719 `FREE_VARIABLES' doesn't try to free them. */
5723 }
5726 /* The starting position is bogus. */
5727 #ifdef WCHAR
5731 #endif
5732 {
5735 }
5737 #ifdef WCHAR
5738 /* Allocate wchar_t array for string1 and string2 and
5739 fill them with converted string. */
5741 {
5742 /* We need seting up buffers here. */
5744 /* We must free wcs buffers in this function. */
5748 {
5753 {
5758 }
5759 }
5761 {
5766 {
5773 }
5778 }
5779 }
5781 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5782 pattern to (char*) in regex_compile. */
5788 /* Initialize subexpression text positions to -1 to mark ones that no
5789 start_memory/stop_memory has been seen for. Also initialize the
5790 register information struct. */
5792 {
5800 }
5802 /* We move `string1' into `string2' if the latter's empty -- but not if
5803 `string1' is null. */
5805 {
5810 #ifdef WCHAR
5815 #endif
5816 }
5820 /* Compute where to stop matching, within the two strings. */
5821 #ifdef WCHAR
5823 {
5827 }
5828 else
5829 {
5835 }
5840 }
5841 #else
5843 {
5846 }
5847 else
5848 {
5851 }
5854 /* `p' scans through the pattern as `d' scans through the data.
5855 `dend' is the end of the input string that `d' points within. `d'
5856 is advanced into the following input string whenever necessary, but
5857 this happens before fetching; therefore, at the beginning of the
5858 loop, `d' can be pointing at the end of a string, but it cannot
5859 equal `string2'. */
5860 #ifdef WCHAR
5862 {
5866 }
5867 else
5868 {
5872 }
5878 }
5879 #else
5881 {
5884 }
5885 else
5886 {
5889 }
5898 /* This loops over pattern commands. It exits by returning from the
5899 function if the match is complete, or it drops through if the match
5900 fails at this starting point in the input data. */
5902 {
5903 #ifdef _LIBC
5905 #else
5907 #endif
5913 /* If we haven't matched the entire string, and we want the
5914 longest match, try backtracking. */
5916 {
5917 /* 1 if this match ends in the same string (string1 or string2)
5918 as the best previous match. */
5919 boolean same_str_p;
5921 /* 1 if this match is the best seen so far. */
5922 boolean best_match_p;
5927 /* AIX compiler got confused when this was combined
5928 with the previous declaration. */
5931 else
5939 /* If exceeds best match so far, save it. */
5941 {
5948 {
5951 }
5952 }
5954 }
5956 /* If no failure points, don't restore garbage. And if
5957 last match is real best match, don't restore second
5958 best one. */
5960 {
5961 restore_best_regs:
5962 /* Restore best match. It may happen that `dend ==
5963 end_match_1' while the restored d is in string2.
5964 For example, the pattern `x.*y.*z' against the
5965 strings `x-' and `y-z-', if the two strings are
5966 not consecutive in memory. */
5974 {
5977 }
5978 }
5981 succeed_label:
5983 /* If caller wants register contents data back, do it. */
5985 {
5986 /* Have the register data arrays been allocated? */
5989 extra element beyond `num_regs' for the `-1' marker
5990 GNU code uses. */
5995 {
5998 }
6000 }
6003 allocated, reallocate them. If we need fewer, just
6004 leave it alone. */
6006 {
6011 {
6014 }
6015 }
6016 }
6017 else
6018 {
6019 /* These braces fend off a "empty body in an else-statement"
6020 warning under GCC when assert expands to nothing. */
6022 }
6024 /* Convert the pointer data in `regstart' and `regend' to
6025 indices. Register zero has to be set differently,
6026 since we haven't kept track of any info for it. */
6028 {
6030 #ifdef WCHAR
6034 else
6037 #else
6042 }
6044 /* Go through the first `min (num_regs, regs->num_regs)'
6045 registers, since that is all we initialized. */
6047 mcnt++)
6048 {
6051 else
6052 {
6057 }
6058 }
6060 /* If the regs structure we return has more elements than
6061 were in the pattern, set the extra elements to -1. If
6062 we (re)allocated the registers, this is the case,
6063 because we always allocate enough to have at least one
6064 -1 at the end. */
6074 #ifdef WCHAR
6077 else
6079 csize1;
6081 #else
6083 ? string1
6091 }
6093 /* Otherwise match next pattern command. */
6095 {
6096 /* Ignore these. Used to ignore the n of succeed_n's which
6097 currently have n == 0. */
6106 /* Match the next n pattern characters exactly. The following
6107 byte in the pattern defines n, and the n bytes after that
6108 are the characters to match. */
6110 #ifdef MBS_SUPPORT
6112 #endif
6116 /* This is written out as an if-else so we don't waste time
6117 testing `translate' inside the loop. */
6119 {
6120 do
6121 {
6123 #ifdef WCHAR
6125 {
6129 }
6130 else
6131 {
6134 }
6135 #else
6140 }
6142 }
6143 else
6144 {
6145 do
6146 {
6149 }
6151 }
6156 /* Match any character except possibly a newline or a null. */
6168 d++;
6174 {
6176 #ifdef WCHAR
6180 #define WORK_BUFFER_SIZE 128
6182 # ifdef _LIBC
6191 #ifdef WCHAR
6192 # ifdef _LIBC
6201 /* p points charset[6], so the address of the next instruction
6202 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6203 where l=length of char_classes, m=length of collating_symbol,
6204 n=equivalence_class, o=length of char_range,
6205 p'=length of character. */
6207 /* Update p to indicate the next instruction. */
6211 /* match with char_class? */
6213 {
6220 # ifdef _LIBC
6223 # else
6226 # endif
6227 }
6229 /* match with collating_symbol? */
6230 # ifdef _LIBC
6232 {
6237 workp++)
6238 {
6246 {
6247 /* Update d, however d will be incremented at
6248 char_set_matched:, we decrement d here. */
6251 }
6252 }
6253 }
6255 # endif
6256 /* If we can't look up collation data, we use wcscoll
6257 instead. */
6258 {
6260 {
6262 # ifdef _LIBC
6264 # else
6266 # endif
6268 /* If wcscoll(the collating symbol, whole string) > 0,
6269 any substring of the string never match with the
6270 collating symbol. */
6271 # ifdef _LIBC
6273 # else
6275 # endif
6276 {
6279 }
6281 /* First, we compare the collating symbol with
6282 the first character of the string.
6283 If it don't match, we add the next character to
6284 the compare buffer in turn. */
6286 {
6289 {
6294 }
6296 /* add next character to the compare buffer. */
6300 # ifdef _LIBC
6302 # else
6304 # endif
6309 /* (str_buf > workp) indicate (str_buf + X > workp),
6310 because for all X (str_buf + X > str_buf).
6311 So we don't need continue this loop. */
6314 /* Otherwise(str_buf < workp),
6315 (str_buf+next_character) may equals (workp).
6316 So we continue this loop. */
6317 }
6318 /* not matched */
6322 }
6323 }
6324 /* match with equivalence_class? */
6325 # ifdef _LIBC
6327 {
6329 /* Try to match the equivalence class against
6330 those known to the collate implementation. */
6339 /* This #include defines a local function! */
6340 # include <locale/weightwc.h>
6351 /* Write 1 collating element to str_buf, and
6352 get its index. */
6356 {
6359 {
6364 }
6368 }
6370 /* Update d, however d will be incremented at
6371 char_set_matched:, we decrement d here. */
6374 {
6377 else
6378 {
6381 }
6382 }
6387 workp++)
6388 {
6390 /* We already checked idx != 0 in regex_compile. */
6393 {
6401 }
6402 }
6403 /* not matched */
6406 }
6408 # endif
6409 /* If we can't look up collation data, we use wcscoll
6410 instead. */
6411 {
6413 {
6415 # ifdef _LIBC
6417 # else
6419 # endif
6421 /* If wcscoll(the collating symbol, whole string) > 0,
6422 any substring of the string never match with the
6423 collating symbol. */
6424 # ifdef _LIBC
6426 # else
6428 # endif
6429 {
6432 }
6434 /* First, we compare the equivalence class with
6435 the first character of the string.
6436 If it don't match, we add the next character to
6437 the compare buffer in turn. */
6439 {
6442 {
6447 }
6449 /* add next character to the compare buffer. */
6453 # ifdef _LIBC
6455 # else
6457 # endif
6463 /* (str_buf > workp) indicate (str_buf + X > workp),
6464 because for all X (str_buf + X > str_buf).
6465 So we don't need continue this loop. */
6468 /* Otherwise(str_buf < workp),
6469 (str_buf+next_character) may equals (workp).
6470 So we continue this loop. */
6471 }
6472 /* not matched */
6476 }
6477 }
6479 /* match with char_range? */
6480 # ifdef _LIBC
6482 {
6490 {
6493 /* We already compute the collation sequence value
6494 of the characters (or collating symbols). */
6500 }
6501 }
6502 else
6503 # endif
6504 {
6505 /* We set range_start_char at str_buf[0], range_end_char
6506 at str_buf[4], and compared char at str_buf[2]. */
6512 {
6515 /* match if (range_start_char <= c <= range_end_char). */
6517 /* If range_start(or end) < 0, we assume -range_start(end)
6518 is the offset of the collating symbol which is specified
6519 as the character of the range start(end). */
6521 /* range_start */
6524 else
6525 {
6528 }
6530 /* range_end */
6533 else
6534 {
6537 }
6539 # ifdef _LIBC
6542 # else
6545 # endif
6547 }
6548 }
6550 /* match with char? */
6557 char_set_matched:
6559 #else
6560 /* Cast to `unsigned' instead of `unsigned char' in case the
6561 bit list is a full 32 bytes long. */
6569 #undef WORK_BUFFER_SIZE
6572 d++;
6574 }
6577 /* The beginning of a group is represented by start_memory.
6578 The arguments are the register number in the next byte, and the
6579 number of groups inner to this one in the next. The text
6580 matched within the group is recorded (in the internal
6581 registers data structure) under the register number. */
6586 /* Find out if this group can match the empty string. */
6593 /* Save the position in the string where we were the last time
6594 we were at this open-group operator in case the group is
6595 operated upon by a repetition operator, e.g., with `(a*)*b'
6596 against `ab'; then we want to ignore where we are now in
6597 the string in case this attempt to match fails. */
6610 /* Clear this whenever we change the register activity status. */
6613 /* This is the new highest active register. */
6616 /* If nothing was active before, this is the new lowest active
6617 register. */
6621 /* Move past the register number and inner group count. */
6628 /* The stop_memory opcode represents the end of a group. Its
6629 arguments are the same as start_memory's: the register
6630 number, and the number of inner groups. */
6635 /* We need to save the string position the last time we were at
6636 this close-group operator in case the group is operated
6637 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6638 against `aba'; then we want to ignore where we are now in
6639 the string in case this attempt to match fails. */
6649 /* This register isn't active anymore. */
6652 /* Clear this whenever we change the register activity status. */
6655 /* If this was the only register active, nothing is active
6656 anymore. */
6658 {
6661 }
6662 else
6664 it isn't necessarily one less than now: consider
6665 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6666 new highest active register is 1. */
6669 r--;
6671 /* If we end up at register zero, that means that we saved
6672 the registers as the result of an `on_failure_jump', not
6673 a `start_memory', and we jumped to past the innermost
6674 `stop_memory'. For example, in ((.)*) we save
6675 registers 1 and 2 as a result of the *, but when we pop
6676 back to the second ), we are at the stop_memory 1.
6677 Thus, nothing is active. */
6679 {
6682 }
6683 else
6685 }
6687 /* If just failed to match something this time around with a
6688 group that's operated on by a repetition operator, try to
6689 force exit from the ``loop'', and restore the register
6690 information for this group that we had before trying this
6691 last match. */
6695 {
6701 {
6704 /* Fall through. */
6716 }
6719 /* If the next operation is a jump backwards in the pattern
6720 to an on_failure_jump right before the start_memory
6721 corresponding to this stop_memory, exit from the loop
6722 by forcing a failure after pushing on the stack the
6723 on_failure_jump's jump in the pattern, and d. */
6727 {
6728 /* If this group ever matched anything, then restore
6729 what its registers were before trying this last
6730 failed match, e.g., with `(a*)*b' against `ab' for
6731 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6732 against `aba' for regend[3].
6734 Also restore the registers for inner groups for,
6735 e.g., `((a*)(b*))*' against `aba' (register 3 would
6736 otherwise get trashed). */
6739 {
6744 /* Restore this and inner groups' (if any) registers. */
6746 r++)
6747 {
6750 /* xx why this test? */
6753 }
6754 }
6755 p1++;
6760 }
6761 }
6763 /* Move past the register number and the inner group count. */
6768 /* \<digit> has been turned into a `duplicate' command which is
6769 followed by the numeric value of <digit> as the register number. */
6771 {
6776 /* Can't back reference a group which we've never matched. */
6780 /* Where in input to try to start matching. */
6783 /* Where to stop matching; if both the place to start and
6784 the place to stop matching are in the same string, then
6785 set to the place to stop, otherwise, for now have to use
6786 the end of the first string. */
6792 {
6793 /* If necessary, advance to next segment in register
6794 contents. */
6796 {
6800 /* End of string1 => advance to string2. */
6803 }
6804 /* At end of register contents => success */
6807 /* If necessary, advance to next segment in data. */
6810 /* How many characters left in this segment to match. */
6813 /* Want how many consecutive characters we can match in
6814 one shot, so, if necessary, adjust the count. */
6818 /* Compare that many; failure if mismatch, else move
6819 past them. */
6826 /* Do this because we've match some characters. */
6828 }
6829 }
6833 /* begline matches the empty string at the beginning of the string
6834 (unless `not_bol' is set in `bufp'), and, if
6835 `newline_anchor' is set, after newlines. */
6840 {
6842 }
6844 {
6846 }
6847 /* In all other cases, we fail. */
6851 /* endline is the dual of begline. */
6856 {
6858 }
6860 /* We have to ``prefetch'' the next character. */
6863 {
6865 }
6869 /* Match at the very beginning of the data. */
6877 /* Match at the very end of the data. */
6885 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6886 pushes NULL as the value for the string on the stack. Then
6887 `pop_failure_point' will keep the current value for the
6888 string, instead of restoring it. To see why, consider
6889 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6890 then the . fails against the \n. But the next thing we want
6891 to do is match the \n against the \n; if we restored the
6892 string value, we would be back at the foo.
6894 Because this is used only in specific cases, we don't need to
6895 check all the things that `on_failure_jump' does, to make
6896 sure the right things get saved on the stack. Hence we don't
6897 share its code. The only reason to push anything on the
6898 stack at all is that otherwise we would have to change
6899 `anychar's code to do something besides goto fail in this
6900 case; that seems worse than this. */
6905 #ifdef _LIBC
6907 #else
6909 #endif
6915 /* Uses of on_failure_jump:
6917 Each alternative starts with an on_failure_jump that points
6918 to the beginning of the next alternative. Each alternative
6919 except the last ends with a jump that in effect jumps past
6920 the rest of the alternatives. (They really jump to the
6921 ending jump of the following alternative, because tensioning
6922 these jumps is a hassle.)
6924 Repeats start with an on_failure_jump that points past both
6925 the repetition text and either the following jump or
6926 pop_failure_jump back to this on_failure_jump. */
6928 on_failure:
6932 #ifdef _LIBC
6934 #else
6936 #endif
6938 /* If this on_failure_jump comes right before a group (i.e.,
6939 the original * applied to a group), save the information
6940 for that group and all inner ones, so that if we fail back
6941 to this point, the group's information will be correct.
6942 For example, in \(a*\)*\1, we need the preceding group,
6943 and in \(zz\(a*\)b*\)\2, we need the inner group. */
6945 /* We can't use `p' to check ahead because we push
6946 a failure point to `p + mcnt' after we do this. */
6949 /* We need to skip no_op's before we look for the
6950 start_memory in case this on_failure_jump is happening as
6951 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
6952 against aba. */
6954 p1++;
6957 {
6958 /* We have a new highest active register now. This will
6959 get reset at the start_memory we are about to get to,
6960 but we will have saved all the registers relevant to
6961 this repetition op, as described above. */
6965 }
6972 /* A smart repeat ends with `maybe_pop_jump'.
6973 We change it to either `pop_failure_jump' or `jump'. */
6977 {
6980 /* Compare the beginning of the repeat with what in the
6981 pattern follows its end. If we can establish that there
6982 is nothing that they would both match, i.e., that we
6983 would have to backtrack because of (as in, e.g., `a*a')
6984 then we can change to pop_failure_jump, because we'll
6985 never have to backtrack.
6987 This is not true in the case of alternatives: in
6988 `(a|ab)*' we do need to backtrack to the `ab' alternative
6989 (e.g., if the string was `ab'). But instead of trying to
6990 detect that here, the alternative has put on a dummy
6991 failure point which is what we will end up popping. */
6993 /* Skip over open/close-group commands.
6994 If what follows this loop is a ...+ construct,
6995 look at what begins its body, since we will have to
6996 match at least one of that. */
6998 {
7006 else
7008 }
7011 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7012 to the `maybe_finalize_jump' of this case. Examine what
7013 follows. */
7015 /* If we're at the end of the pattern, we can change. */
7017 {
7018 /* Consider what happens when matching ":\(.*\)"
7019 against ":/". I don't really understand this code
7020 yet. */
7022 pop_failure_jump;
7023 DEBUG_PRINT1
7025 }
7028 #ifdef MBS_SUPPORT
7030 #endif
7032 {
7033 register UCHAR_T c
7037 #ifdef MBS_SUPPORT
7039 #endif
7041 {
7043 pop_failure_jump;
7044 #ifdef WCHAR
7048 #else
7052 #endif
7053 }
7055 #ifndef WCHAR
7058 {
7065 /* `negate' is equal to 1 if c would match, which means
7066 that we can't change to pop_failure_jump. */
7068 {
7071 }
7072 }
7074 }
7075 #ifndef WCHAR
7077 {
7078 /* We win if the first character of the loop is not part
7079 of the charset. */
7084 {
7087 }
7090 {
7092 /* We win if the charset_not inside the loop
7093 lists every character listed in the charset after. */
7101 {
7104 }
7105 }
7107 {
7109 /* We win if the charset inside the loop
7110 has no overlap with the one after the loop. */
7113 idx++)
7118 {
7121 }
7122 }
7123 }
7125 }
7128 {
7132 }
7133 /* Fall through. */
7136 /* The end of a simple repeat has a pop_failure_jump back to
7137 its matching on_failure_jump, where the latter will push a
7138 failure point. The pop_failure_jump takes off failure
7139 points put on by this pop_failure_jump's matching
7140 on_failure_jump; we got through the pattern to here from the
7141 matching on_failure_jump, so didn't fail. */
7143 {
7144 /* We need to pass separate storage for the lowest and
7145 highest registers, even though we don't care about the
7146 actual values. Otherwise, we will restore only one
7147 register from the stack, since lowest will == highest in
7148 `pop_failure_point'. */
7157 }
7158 /* Fall through. */
7160 unconditional_jump:
7161 #ifdef _LIBC
7163 #else
7165 #endif
7166 /* Note fall through. */
7168 /* Unconditionally jump (without popping any failure points). */
7173 #ifdef _LIBC
7175 #else
7177 #endif
7181 /* We need this opcode so we can detect where alternatives end
7182 in `group_match_null_string_p' et al. */
7188 /* Normally, the on_failure_jump pushes a failure point, which
7189 then gets popped at pop_failure_jump. We will end up at
7190 pop_failure_jump, also, and with a pattern of, say, `a+', we
7191 are skipping over the on_failure_jump, so we have to push
7192 something meaningless for pop_failure_jump to pop. */
7195 /* It doesn't matter what we push for the string here. What
7196 the code at `fail' tests is the value for the pattern. */
7201 /* At the end of an alternative, we need to push a dummy failure
7202 point in case we are followed by a `pop_failure_jump', because
7203 we don't want the failure point for the alternative to be
7204 popped. For example, matching `(a|ab)*' against `aab'
7205 requires that we match the `ab' alternative. */
7208 /* See comments just above at `dummy_failure_jump' about the
7209 two zeroes. */
7213 /* Have to succeed matching what follows at least n times.
7214 After that, handle like `on_failure_jump'. */
7220 /* Originally, this is how many times we HAVE to succeed. */
7222 {
7223 mcnt--;
7226 #ifdef _LIBC
7229 #else
7232 #endif
7233 }
7235 {
7236 #ifdef _LIBC
7239 #else
7244 #ifdef WCHAR
7246 #else
7251 }
7258 /* Originally, this is how many times we CAN jump. */
7260 {
7261 mcnt--;
7264 #ifdef _LIBC
7266 mcnt);
7267 #else
7269 mcnt);
7272 }
7273 /* If don't have to jump any more, skip over the rest of command. */
7274 else
7279 {
7285 #ifdef _LIBC
7287 #else
7289 #endif
7292 }
7294 #if 0
7295 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7296 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7297 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7298 macro and introducing temporary variables works around the bug. */
7311 #else
7313 {
7325 }
7328 {
7340 }
7341 #endif
7357 #ifdef emacs
7384 matchsyntax:
7386 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7387 d++;
7401 matchnotsyntax:
7403 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7404 d++;
7417 d++;
7426 d++;
7432 }
7436 /* We goto here if a matching operation fails. */
7437 fail:
7445 /* If this failure point is a dummy, try the next one. */
7449 /* If we failed to the end of the pattern, don't examine *p. */
7452 {
7455 /* If failed to a backwards jump that's part of a repetition
7456 loop, need to pop this failure point and use the next one. */
7458 {
7461 /* Fall through. */
7470 || (!is_a_jump_n
7476 }
7477 }
7481 }
7482 else
7493 \f
7494 /* Subroutine definitions for re_match_2. */
7497 /* We are passed P pointing to a register number after a start_memory.
7499 Return true if the pattern up to the corresponding stop_memory can
7500 match the empty string, and false otherwise.
7502 If we find the matching stop_memory, sets P to point to one past its number.
7503 Otherwise, sets P to an undefined byte less than or equal to END.
7505 We don't handle duplicates properly (yet). */
7507 static boolean
7510 {
7512 /* Point to after the args to the start_memory. */
7516 {
7517 /* Skip over opcodes that can match nothing, and return true or
7518 false, as appropriate, when we get to one that can't, or to the
7519 matching stop_memory. */
7522 {
7523 /* Could be either a loop or a series of alternatives. */
7525 p1++;
7528 /* If the next operation is not a jump backwards in the
7529 pattern. */
7532 {
7533 /* Go through the on_failure_jumps of the alternatives,
7534 seeing if any of the alternatives cannot match nothing.
7535 The last alternative starts with only a jump,
7536 whereas the rest start with on_failure_jump and end
7537 with a jump, e.g., here is the pattern for `a|b|c':
7539 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7540 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7541 /exactn/1/c
7543 So, we have to first go through the first (n-1)
7544 alternatives and then deal with the last one separately. */
7547 /* Deal with the first (n-1) alternatives, which start
7548 with an on_failure_jump (see above) that jumps to right
7549 past a jump_past_alt. */
7552 jump_past_alt)
7553 {
7554 /* `mcnt' holds how many bytes long the alternative
7555 is, including the ending `jump_past_alt' and
7556 its number. */
7560 reg_info))
7563 /* Move to right after this alternative, including the
7564 jump_past_alt. */
7567 /* Break if it's the beginning of an n-th alternative
7568 that doesn't begin with an on_failure_jump. */
7572 /* Still have to check that it's not an n-th
7573 alternative that starts with an on_failure_jump. */
7574 p1++;
7577 jump_past_alt)
7578 {
7579 /* Get to the beginning of the n-th alternative. */
7582 }
7583 }
7585 /* Deal with the last alternative: go back and get number
7586 of the `jump_past_alt' just before it. `mcnt' contains
7587 the length of the alternative. */
7607 }
7614 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7615 It expects P to be the first byte of a single alternative and END one
7616 byte past the last. The alternative can contain groups. */
7618 static boolean
7621 {
7626 {
7627 /* Skip over opcodes that can match nothing, and break when we get
7628 to one that can't. */
7631 {
7632 /* It's a loop. */
7634 p1++;
7642 }
7649 /* Deals with the ops common to group_match_null_string_p and
7650 alt_match_null_string_p.
7652 Sets P to one after the op and its arguments, if any. */
7654 static boolean
7657 {
7659 boolean ret;
7664 {
7674 #ifdef emacs
7678 #endif
7686 /* Have to set this here in case we're checking a group which
7687 contains a group and a back reference to it. */
7696 /* If this is an optimized succeed_n for zero times, make the jump. */
7701 else
7706 /* Get to the number of times to succeed. */
7711 {
7715 }
7716 else
7730 /* All other opcodes mean we cannot match the empty string. */
7732 }
7739 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7740 bytes; nonzero otherwise. */
7742 static int
7744 RE_TRANSLATE_TYPE translate)
7745 {
7749 {
7750 #ifdef WCHAR
7757 len--;
7758 }
7760 }
7761 \f
7765 /* Entry points for GNU code. */
7767 /* re_compile_pattern is the GNU regular expression compiler: it
7768 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7769 Returns 0 if the pattern was valid, otherwise an error string.
7771 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7772 are set in BUFP on entry.
7774 We call regex_compile to do the actual compilation. */
7779 {
7780 reg_errcode_t ret;
7782 /* GNU code is written to assume at least RE_NREGS registers will be set
7783 (and at least one extra will be -1). */
7786 /* And GNU code determines whether or not to get register information
7787 by passing null for the REGS argument to re_match, etc., not by
7788 setting no_sub. */
7791 /* Match anchors at newline. */
7794 # ifdef MBS_SUPPORT
7797 else
7798 # endif
7804 }
7805 #ifdef _LIBC
7807 #endif
7808 \f
7809 /* Entry points compatible with 4.2 BSD regex library. We don't define
7810 them unless specifically requested. */
7812 #if defined _REGEX_RE_COMP || defined _LIBC
7814 /* BSD has one and only one pattern buffer. */
7818 #ifdef _LIBC
7819 /* Make these definitions weak in libc, so POSIX programs can redefine
7820 these names if they don't use our functions, and still use
7821 regcomp/regexec below without link errors. */
7822 weak_function
7823 #endif
7825 {
7826 reg_errcode_t ret;
7829 {
7833 }
7836 {
7845 }
7847 /* Since `re_exec' always passes NULL for the `regs' argument, we
7848 don't need to initialize the pattern buffer fields which affect it. */
7850 /* Match anchors at newlines. */
7853 # ifdef MBS_SUPPORT
7856 else
7857 # endif
7863 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7865 }
7868 int
7869 #ifdef _LIBC
7870 weak_function
7871 #endif
7873 {
7875 return
7877 }
7880 \f
7881 /* POSIX.2 functions. Don't define these for Emacs. */
7883 #ifndef emacs
7885 /* regcomp takes a regular expression as a string and compiles it.
7887 PREG is a regex_t *. We do not expect any fields to be initialized,
7888 since POSIX says we shouldn't. Thus, we set
7890 `buffer' to the compiled pattern;
7891 `used' to the length of the compiled pattern;
7892 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7893 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7894 RE_SYNTAX_POSIX_BASIC;
7895 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7896 `fastmap' to an allocated space for the fastmap;
7897 `fastmap_accurate' to zero;
7898 `re_nsub' to the number of subexpressions in PATTERN.
7900 PATTERN is the address of the pattern string.
7902 CFLAGS is a series of bits which affect compilation.
7904 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
7905 use POSIX basic syntax.
7907 If REG_NEWLINE is set, then . and [^...] don't match newline.
7908 Also, regexec will try a match beginning after every newline.
7910 If REG_ICASE is set, then we considers upper- and lowercase
7911 versions of letters to be equivalent when matching.
7913 If REG_NOSUB is set, then when PREG is passed to regexec, that
7914 routine will report only success or failure, and nothing about the
7915 registers.
7917 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
7918 the return codes and their meanings.) */
7920 int
7922 {
7923 reg_errcode_t ret;
7924 reg_syntax_t syntax
7928 /* regex_compile will allocate the space for the compiled pattern. */
7933 /* Try to allocate space for the fastmap. */
7937 {
7940 preg->translate
7946 /* Map uppercase characters to corresponding lowercase ones. */
7949 }
7950 else
7953 /* If REG_NEWLINE is set, newlines are treated differently. */
7958 /* It also changes the matching behavior. */
7960 }
7961 else
7966 /* POSIX says a null character in the pattern terminates it, so we
7967 can use strlen here in compiling the pattern. */
7968 # ifdef MBS_SUPPORT
7971 else
7972 # endif
7975 /* POSIX doesn't distinguish between an unmatched open-group and an
7976 unmatched close-group: both are REG_EPAREN. */
7980 {
7981 /* Compute the fastmap now, since regexec cannot modify the pattern
7982 buffer. */
7984 {
7985 /* Some error occurred while computing the fastmap, just forget
7986 about it. */
7989 }
7990 }
7993 }
7994 #ifdef _LIBC
7996 #endif
7999 /* regexec searches for a given pattern, specified by PREG, in the
8000 string STRING.
8002 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
8003 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
8004 least NMATCH elements, and we set them to the offsets of the
8005 corresponding matched substrings.
8007 EFLAGS specifies `execution flags' which affect matching: if
8008 REG_NOTBOL is set, then ^ does not match at the beginning of the
8009 string; if REG_NOTEOL is set, then $ does not match at the end.
8011 We return 0 if we find a match and REG_NOMATCH if not. */
8013 int
8016 {
8019 regex_t private_preg;
8028 /* The user has told us exactly how many registers to return
8029 information about, via `nmatch'. We have to pass that on to the
8030 matching routines. */
8034 {
8040 }
8042 /* Perform the searching operation. */
8047 /* Copy the register information to the POSIX structure. */
8049 {
8051 {
8055 {
8058 }
8059 }
8061 /* If we needed the temporary register info, free the space now. */
8063 }
8065 /* We want zero return to mean success, unlike `re_search'. */
8067 }
8068 #ifdef _LIBC
8070 #endif
8073 /* Returns a message corresponding to an error code, ERRCODE, returned
8074 from either regcomp or regexec. We don't use PREG here. */
8076 size_t
8079 {
8086 /* Only error codes returned by the rest of the code should be passed
8087 to this routine. If we are given anything else, or if other regex
8088 code generates an invalid error code, then the program has a bug.
8089 Dump core so we can fix it. */
8097 {
8099 {
8100 #if defined HAVE_MEMPCPY || defined _LIBC
8102 #else
8105 #endif
8106 }
8107 else
8109 }
8112 }
8113 #ifdef _LIBC
8115 #endif
8118 /* Free dynamically allocated space used by PREG. */
8120 void
8122 {
8135 }
8136 #ifdef _LIBC
8138 #endif
8144 \f
8145 #undef STORE_NUMBER
8146 #undef STORE_NUMBER_AND_INCR
8147 #undef EXTRACT_NUMBER
8148 #undef EXTRACT_NUMBER_AND_INCR
8150 #undef DEBUG_PRINT_COMPILED_PATTERN
8151 #undef DEBUG_PRINT_DOUBLE_STRING
8153 #undef INIT_FAIL_STACK
8154 #undef RESET_FAIL_STACK
8155 #undef DOUBLE_FAIL_STACK
8156 #undef PUSH_PATTERN_OP
8157 #undef PUSH_FAILURE_POINTER
8158 #undef PUSH_FAILURE_INT
8159 #undef PUSH_FAILURE_ELT
8160 #undef POP_FAILURE_POINTER
8161 #undef POP_FAILURE_INT
8162 #undef POP_FAILURE_ELT
8163 #undef DEBUG_PUSH
8164 #undef DEBUG_POP
8165 #undef PUSH_FAILURE_POINT
8166 #undef POP_FAILURE_POINT
8168 #undef REG_UNSET_VALUE
8169 #undef REG_UNSET
8171 #undef PATFETCH
8172 #undef PATFETCH_RAW
8173 #undef PATUNFETCH
8174 #undef TRANSLATE
8176 #undef INIT_BUF_SIZE
8177 #undef GET_BUFFER_SPACE
8178 #undef BUF_PUSH
8179 #undef BUF_PUSH_2
8180 #undef BUF_PUSH_3
8181 #undef STORE_JUMP
8182 #undef STORE_JUMP2
8183 #undef INSERT_JUMP
8184 #undef INSERT_JUMP2
8185 #undef EXTEND_BUFFER
8186 #undef GET_UNSIGNED_NUMBER
8187 #undef FREE_STACK_RETURN
8189 # undef POINTER_TO_OFFSET
8190 # undef MATCHING_IN_FRST_STRING
8191 # undef PREFETCH
8192 # undef AT_STRINGS_BEG
8193 # undef AT_STRINGS_END
8194 # undef WORDCHAR_P
8195 # undef FREE_VAR
8196 # undef FREE_VARIABLES
8197 # undef NO_HIGHEST_ACTIVE_REG
8198 # undef NO_LOWEST_ACTIVE_REG
8200 # undef CHAR_T
8201 # undef UCHAR_T
8202 # undef COMPILED_BUFFER_VAR
8203 # undef OFFSET_ADDRESS_SIZE
8204 # undef CHAR_CLASS_SIZE
8205 # undef PREFIX
8206 # undef ARG_PREFIX
8207 # undef PUT_CHAR
8208 # undef BYTE
8209 # undef WCHAR
8211 # define DEFINED_ONCE