| blob | 07b8ac832dcd8470fee336ac51cce4570b8d07b4 |
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) 1993-2019 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 #undef _GNU_SOURCE
34 #define _GNU_SOURCE
36 #ifndef INSIDE_RECURSION
37 # ifdef HAVE_CONFIG_H
38 # include <config.h>
39 # endif
40 #endif
42 #include <ansidecl.h>
44 #ifndef INSIDE_RECURSION
46 # if defined STDC_HEADERS && !defined emacs
47 # include <stddef.h>
48 # define PTR_INT_TYPE ptrdiff_t
49 # else
50 /* We need this for `regex.h', and perhaps for the Emacs include files. */
51 # include <sys/types.h>
52 # define PTR_INT_TYPE long
53 # endif
55 # define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
57 /* For platform which support the ISO C amendement 1 functionality we
58 support user defined character classes. */
59 # if defined _LIBC || WIDE_CHAR_SUPPORT
60 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
61 # include <wchar.h>
62 # include <wctype.h>
63 # endif
65 # ifdef _LIBC
66 /* We have to keep the namespace clean. */
67 # define regfree(preg) __regfree (preg)
68 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
69 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
70 # define regerror(errcode, preg, errbuf, errbuf_size) \
71 __regerror(errcode, preg, errbuf, errbuf_size)
72 # define re_set_registers(bu, re, nu, st, en) \
73 __re_set_registers (bu, re, nu, st, en)
74 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
75 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
76 # define re_match(bufp, string, size, pos, regs) \
77 __re_match (bufp, string, size, pos, regs)
78 # define re_search(bufp, string, size, startpos, range, regs) \
79 __re_search (bufp, string, size, startpos, range, regs)
80 # define re_compile_pattern(pattern, length, bufp) \
81 __re_compile_pattern (pattern, length, bufp)
82 # define re_set_syntax(syntax) __re_set_syntax (syntax)
83 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
84 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
85 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
87 # define btowc __btowc
89 /* We are also using some library internals. */
90 # include <locale/localeinfo.h>
91 # include <locale/elem-hash.h>
92 # include <langinfo.h>
93 # include <locale/coll-lookup.h>
94 # endif
96 /* This is for other GNU distributions with internationalized messages. */
97 # if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
98 # include <libintl.h>
99 # ifdef _LIBC
100 # undef gettext
102 # endif
103 # else
104 # define gettext(msgid) (msgid)
105 # endif
107 # ifndef gettext_noop
108 /* This define is so xgettext can find the internationalizable
109 strings. */
110 # define gettext_noop(String) String
111 # endif
113 /* The `emacs' switch turns on certain matching commands
114 that make sense only in Emacs. */
115 # ifdef emacs
123 /* If we are not linking with Emacs proper,
124 we can't use the relocating allocator
125 even if config.h says that we can. */
126 # undef REL_ALLOC
128 # if defined STDC_HEADERS || defined _LIBC
129 # include <stdlib.h>
130 # else
133 # endif
135 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
136 If nothing else has been done, use the method below. */
137 # ifdef INHIBIT_STRING_HEADER
138 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
139 # if !defined bzero && !defined bcopy
140 # undef INHIBIT_STRING_HEADER
141 # endif
142 # endif
143 # endif
145 /* This is the normal way of making sure we have a bcopy and a bzero.
146 This is used in most programs--a few other programs avoid this
147 by defining INHIBIT_STRING_HEADER. */
148 # ifndef INHIBIT_STRING_HEADER
149 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
150 # include <string.h>
151 # ifndef bzero
152 # ifndef _LIBC
154 # else
155 # define bzero(s, n) __bzero (s, n)
156 # endif
157 # endif
158 # else
159 # include <strings.h>
160 # ifndef memcmp
161 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
162 # endif
163 # ifndef memcpy
164 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
165 # endif
166 # endif
167 # endif
169 /* Define the syntax stuff for \<, \>, etc. */
171 /* This must be nonzero for the wordchar and notwordchar pattern
172 commands in re_match_2. */
173 # ifndef Sword
174 # define Sword 1
175 # endif
177 # ifdef SWITCH_ENUM_BUG
178 # define SWITCH_ENUM_CAST(x) ((int)(x))
179 # else
180 # define SWITCH_ENUM_CAST(x) (x)
181 # endif
185 # if defined _LIBC || HAVE_LIMITS_H
186 # include <limits.h>
187 # endif
189 # ifndef MB_LEN_MAX
190 # define MB_LEN_MAX 1
191 # endif
192 \f
193 /* Get the interface, including the syntax bits. */
196 /* isalpha etc. are used for the character classes. */
197 # include <ctype.h>
199 /* Jim Meyering writes:
201 "... Some ctype macros are valid only for character codes that
202 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
203 using /bin/cc or gcc but without giving an ansi option). So, all
204 ctype uses should be through macros like ISPRINT... If
205 STDC_HEADERS is defined, then autoconf has verified that the ctype
206 macros don't need to be guarded with references to isascii. ...
207 Defining isascii to 1 should let any compiler worth its salt
208 eliminate the && through constant folding."
209 Solaris defines some of these symbols so we must undefine them first. */
211 # undef ISASCII
212 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
213 # define ISASCII(c) 1
214 # else
215 # define ISASCII(c) isascii(c)
216 # endif
218 # ifdef isblank
219 # define ISBLANK(c) (ISASCII (c) && isblank (c))
220 # else
222 # endif
223 # ifdef isgraph
224 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
225 # else
226 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
227 # endif
229 # undef ISPRINT
230 # define ISPRINT(c) (ISASCII (c) && isprint (c))
231 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
232 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
233 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
234 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
235 # define ISLOWER(c) (ISASCII (c) && islower (c))
236 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
237 # define ISSPACE(c) (ISASCII (c) && isspace (c))
238 # define ISUPPER(c) (ISASCII (c) && isupper (c))
239 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
241 # ifdef _tolower
242 # define TOLOWER(c) _tolower(c)
243 # else
244 # define TOLOWER(c) tolower(c)
245 # endif
247 # ifndef NULL
248 # define NULL (void *)0
249 # endif
251 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
252 since ours (we hope) works properly with all combinations of
253 machines, compilers, `char' and `unsigned char' argument types.
254 (Per Bothner suggested the basic approach.) */
255 # undef SIGN_EXTEND_CHAR
256 # if __STDC__
257 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
259 /* As in Harbison and Steele. */
260 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
261 # endif
262 \f
263 # ifndef emacs
264 /* How many characters in the character set. */
265 # define CHAR_SET_SIZE 256
267 # ifdef SYNTAX_TABLE
277 static void
279 {
294 }
298 # define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
301 \f
302 /* Integer type for pointers. */
303 # if !defined _LIBC && !defined HAVE_UINTPTR_T
305 # endif
307 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
308 use `alloca' instead of `malloc'. This is because using malloc in
309 re_search* or re_match* could cause memory leaks when C-g is used in
310 Emacs; also, malloc is slower and causes storage fragmentation. On
311 the other hand, malloc is more portable, and easier to debug.
313 Because we sometimes use alloca, some routines have to be macros,
314 not functions -- `alloca'-allocated space disappears at the end of the
315 function it is called in. */
317 # ifdef REGEX_MALLOC
319 # define REGEX_ALLOCATE malloc
320 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
321 # define REGEX_FREE free
325 /* Emacs already defines alloca, sometimes. */
326 # ifndef alloca
328 /* Make alloca work the best possible way. */
329 # ifdef __GNUC__
330 # define alloca __builtin_alloca
332 # if HAVE_ALLOCA_H
333 # include <alloca.h>
339 # define REGEX_ALLOCATE alloca
341 /* Assumes a `char *destination' variable. */
342 # define REGEX_REALLOCATE(source, osize, nsize) \
343 (destination = (char *) alloca (nsize), \
344 memcpy (destination, source, osize))
346 /* No need to do anything to free, after alloca. */
351 /* Define how to allocate the failure stack. */
353 # if defined REL_ALLOC && defined REGEX_MALLOC
355 # define REGEX_ALLOCATE_STACK(size) \
356 r_alloc (&failure_stack_ptr, (size))
357 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
358 r_re_alloc (&failure_stack_ptr, (nsize))
359 # define REGEX_FREE_STACK(ptr) \
360 r_alloc_free (&failure_stack_ptr)
364 # ifdef REGEX_MALLOC
366 # define REGEX_ALLOCATE_STACK malloc
367 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
368 # define REGEX_FREE_STACK free
372 # define REGEX_ALLOCATE_STACK alloca
374 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
375 REGEX_REALLOCATE (source, osize, nsize)
376 /* No need to explicitly free anything. */
377 # define REGEX_FREE_STACK(arg)
383 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
384 `string1' or just past its end. This works if PTR is NULL, which is
385 a good thing. */
386 # define FIRST_STRING_P(ptr) \
387 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
389 /* (Re)Allocate N items of type T using malloc, or fail. */
390 # define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
391 # define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
392 # define RETALLOC_IF(addr, n, t) \
393 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
394 # define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
398 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
400 # undef MAX
401 # undef MIN
402 # define MAX(a, b) ((a) > (b) ? (a) : (b))
403 # define MIN(a, b) ((a) < (b) ? (a) : (b))
406 # define false 0
407 # define true 1
410 reg_syntax_t syntax,
426 #ifdef MBS_SUPPORT
428 reg_syntax_t syntax,
447 #endif
448 \f
449 /* These are the command codes that appear in compiled regular
450 expressions. Some opcodes are followed by argument bytes. A
451 command code can specify any interpretation whatsoever for its
452 arguments. Zero bytes may appear in the compiled regular expression. */
455 {
458 /* Succeed right away--no more backtracking. */
459 succeed,
461 /* Followed by one byte giving n, then by n literal bytes. */
462 exactn,
464 # ifdef MBS_SUPPORT
465 /* Same as exactn, but contains binary data. */
466 exactn_bin,
467 # endif
469 /* Matches any (more or less) character. */
470 anychar,
472 /* Matches any one char belonging to specified set. First
473 following byte is number of bitmap bytes. Then come bytes
474 for a bitmap saying which chars are in. Bits in each byte
475 are ordered low-bit-first. A character is in the set if its
476 bit is 1. A character too large to have a bit in the map is
477 automatically not in the set. */
478 /* ifdef MBS_SUPPORT, following element is length of character
479 classes, length of collating symbols, length of equivalence
480 classes, length of character ranges, and length of characters.
481 Next, character class element, collating symbols elements,
482 equivalence class elements, range elements, and character
483 elements follow.
484 See regex_compile function. */
485 charset,
487 /* Same parameters as charset, but match any character that is
488 not one of those specified. */
489 charset_not,
491 /* Start remembering the text that is matched, for storing in a
492 register. Followed by one byte with the register number, in
493 the range 0 to one less than the pattern buffer's re_nsub
494 field. Then followed by one byte with the number of groups
495 inner to this one. (This last has to be part of the
496 start_memory only because we need it in the on_failure_jump
497 of re_match_2.) */
498 start_memory,
500 /* Stop remembering the text that is matched and store it in a
501 memory register. Followed by one byte with the register
502 number, in the range 0 to one less than `re_nsub' in the
503 pattern buffer, and one byte with the number of inner groups,
504 just like `start_memory'. (We need the number of inner
505 groups here because we don't have any easy way of finding the
506 corresponding start_memory when we're at a stop_memory.) */
507 stop_memory,
509 /* Match a duplicate of something remembered. Followed by one
510 byte containing the register number. */
511 duplicate,
513 /* Fail unless at beginning of line. */
514 begline,
516 /* Fail unless at end of line. */
517 endline,
519 /* Succeeds if at beginning of buffer (if emacs) or at beginning
520 of string to be matched (if not). */
521 begbuf,
523 /* Analogously, for end of buffer/string. */
524 endbuf,
526 /* Followed by two byte relative address to which to jump. */
527 jump,
529 /* Same as jump, but marks the end of an alternative. */
530 jump_past_alt,
532 /* Followed by two-byte relative address of place to resume at
533 in case of failure. */
534 /* ifdef MBS_SUPPORT, the size of address is 1. */
535 on_failure_jump,
537 /* Like on_failure_jump, but pushes a placeholder instead of the
538 current string position when executed. */
539 on_failure_keep_string_jump,
541 /* Throw away latest failure point and then jump to following
542 two-byte relative address. */
543 /* ifdef MBS_SUPPORT, the size of address is 1. */
544 pop_failure_jump,
546 /* Change to pop_failure_jump if know won't have to backtrack to
547 match; otherwise change to jump. This is used to jump
548 back to the beginning of a repeat. If what follows this jump
549 clearly won't match what the repeat does, such that we can be
550 sure that there is no use backtracking out of repetitions
551 already matched, then we change it to a pop_failure_jump.
552 Followed by two-byte address. */
553 /* ifdef MBS_SUPPORT, the size of address is 1. */
554 maybe_pop_jump,
556 /* Jump to following two-byte address, and push a dummy failure
557 point. This failure point will be thrown away if an attempt
558 is made to use it for a failure. A `+' construct makes this
559 before the first repeat. Also used as an intermediary kind
560 of jump when compiling an alternative. */
561 /* ifdef MBS_SUPPORT, the size of address is 1. */
562 dummy_failure_jump,
564 /* Push a dummy failure point and continue. Used at the end of
565 alternatives. */
566 push_dummy_failure,
568 /* Followed by two-byte relative address and two-byte number n.
569 After matching N times, jump to the address upon failure. */
570 /* ifdef MBS_SUPPORT, the size of address is 1. */
571 succeed_n,
573 /* Followed by two-byte relative address, and two-byte number n.
574 Jump to the address N times, then fail. */
575 /* ifdef MBS_SUPPORT, the size of address is 1. */
576 jump_n,
578 /* Set the following two-byte relative address to the
579 subsequent two-byte number. The address *includes* the two
580 bytes of number. */
581 /* ifdef MBS_SUPPORT, the size of address is 1. */
582 set_number_at,
591 notwordbound /* Succeeds if not at a word boundary. */
593 # ifdef emacs
598 /* Matches any character whose syntax is specified. Followed by
599 a byte which contains a syntax code, e.g., Sword. */
600 syntaxspec,
602 /* Matches any character whose syntax is not that specified. */
603 notsyntaxspec
607 \f
609 #ifdef BYTE
610 # define CHAR_T char
611 # define UCHAR_T unsigned char
612 # define COMPILED_BUFFER_VAR bufp->buffer
613 # define OFFSET_ADDRESS_SIZE 2
614 # define PREFIX(name) byte_##name
615 # define ARG_PREFIX(name) name
616 # define PUT_CHAR(c) putchar (c)
617 #else
618 # ifdef WCHAR
619 # define CHAR_T wchar_t
620 # define UCHAR_T wchar_t
621 # define COMPILED_BUFFER_VAR wc_buffer
623 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
624 # define PREFIX(name) wcs_##name
625 # define ARG_PREFIX(name) c##name
626 /* Should we use wide stream?? */
628 # define TRUE 1
629 # define FALSE 0
630 # else
631 # ifdef MBS_SUPPORT
632 # define WCHAR
633 # define INSIDE_RECURSION
635 # undef INSIDE_RECURSION
636 # endif
637 # define BYTE
638 # define INSIDE_RECURSION
640 # undef INSIDE_RECURSION
641 # endif
642 #endif
644 #ifdef INSIDE_RECURSION
645 /* Common operations on the compiled pattern. */
647 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
648 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
650 # ifdef WCHAR
651 # define STORE_NUMBER(destination, number) \
652 do { \
653 *(destination) = (UCHAR_T)(number); \
654 } while (0)
656 # define STORE_NUMBER(destination, number) \
657 do { \
658 (destination)[0] = (number) & 0377; \
659 (destination)[1] = (number) >> 8; \
660 } while (0)
663 /* Same as STORE_NUMBER, except increment DESTINATION to
664 the byte after where the number is stored. Therefore, DESTINATION
665 must be an lvalue. */
666 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
668 # define STORE_NUMBER_AND_INCR(destination, number) \
669 do { \
670 STORE_NUMBER (destination, number); \
671 (destination) += OFFSET_ADDRESS_SIZE; \
672 } while (0)
674 /* Put into DESTINATION a number stored in two contiguous bytes starting
675 at SOURCE. */
676 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
678 # ifdef WCHAR
679 # define EXTRACT_NUMBER(destination, source) \
680 do { \
681 (destination) = *(source); \
682 } while (0)
684 # define EXTRACT_NUMBER(destination, source) \
685 do { \
686 (destination) = *(source) & 0377; \
687 (destination) += ((unsigned) SIGN_EXTEND_CHAR (*((source) + 1))) << 8; \
688 } while (0)
689 # endif
691 # ifdef DEBUG
693 static void
695 {
696 # ifdef WCHAR
702 # endif
703 }
706 # undef EXTRACT_NUMBER
707 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
712 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
713 SOURCE must be an lvalue. */
715 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
716 do { \
717 EXTRACT_NUMBER (destination, source); \
718 (source) += OFFSET_ADDRESS_SIZE; \
719 } while (0)
721 # ifdef DEBUG
724 static void
726 {
729 }
731 # ifndef EXTRACT_MACROS
732 # undef EXTRACT_NUMBER_AND_INCR
733 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
734 PREFIX(extract_number_and_incr) (&dest, &src)
739 \f
741 /* If DEBUG is defined, Regex prints many voluminous messages about what
742 it is doing (if the variable `debug' is nonzero). If linked with the
743 main program in `iregex.c', you can enter patterns and strings
744 interactively. And if linked with the main program in `main.c' and
745 the other test files, you can run the already-written tests. */
747 # ifdef DEBUG
749 # ifndef DEFINED_ONCE
751 /* We use standard I/O for debugging. */
752 # include <stdio.h>
754 /* It is useful to test things that ``must'' be true when debugging. */
755 # include <assert.h>
759 # define DEBUG_STATEMENT(e) e
760 # define DEBUG_PRINT1(x) if (debug) printf (x)
761 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
762 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
763 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
766 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
767 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
768 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
769 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
772 /* Print the fastmap in human-readable form. */
774 # ifndef DEFINED_ONCE
775 void
777 {
782 {
784 {
788 {
790 i++;
791 }
793 {
796 }
797 }
798 }
800 }
804 /* Print a compiled pattern string in human-readable form, starting at
805 the START pointer into it and ending just before the pointer END. */
807 void
809 {
816 {
819 }
821 /* Loop over pattern commands. */
823 {
824 # ifdef _LIBC
826 # else
828 # endif
831 {
839 do
840 {
843 }
847 # ifdef MBS_SUPPORT
851 do
852 {
854 }
879 {
880 # ifdef WCHAR
891 {
897 }
900 {
906 }
909 {
913 }
918 # else
930 {
931 /* Are we starting a range? */
933 {
936 }
937 /* Have we broken a range? */
939 {
942 }
948 }
957 }
970 # ifdef _LIBC
972 # else
974 # endif
979 # ifdef _LIBC
981 # else
984 # endif
989 # ifdef _LIBC
991 # else
993 # endif
1002 # ifdef _LIBC
1004 # else
1006 # endif
1011 # ifdef _LIBC
1013 # else
1015 # endif
1020 # ifdef _LIBC
1022 # else
1024 # endif
1029 # ifdef _LIBC
1031 # else
1033 # endif
1040 # ifdef _LIBC
1042 # else
1045 # endif
1059 # ifdef _LIBC
1061 # else
1064 # endif
1083 # ifdef emacs
1127 }
1130 }
1132 # ifdef _LIBC
1134 # else
1136 # endif
1137 }
1140 void
1142 {
1151 {
1154 }
1156 # ifdef _LIBC
1158 # else
1160 # endif
1168 /* Perhaps we should print the translate table? */
1169 }
1172 void
1175 {
1180 else
1181 {
1185 {
1190 }
1194 {
1197 {
1200 }
1201 }
1202 }
1203 }
1205 # ifndef DEFINED_ONCE
1206 void
1208 {
1210 }
1211 # endif
1215 # ifndef DEFINED_ONCE
1216 # undef assert
1217 # define assert(e)
1219 # define DEBUG_STATEMENT(e)
1220 # define DEBUG_PRINT1(x)
1221 # define DEBUG_PRINT2(x1, x2)
1222 # define DEBUG_PRINT3(x1, x2, x3)
1223 # define DEBUG_PRINT4(x1, x2, x3, x4)
1225 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1226 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1230 \f
1232 # ifdef WCHAR
1233 /* This convert a multibyte string to a wide character string.
1234 And write their correspondances to offset_buffer(see below)
1235 and write whether each wchar_t is binary data to is_binary.
1236 This assume invalid multibyte sequences as binary data.
1237 We assume offset_buffer and is_binary is already allocated
1238 enough space. */
1243 static size_t
1246 /* It hold correspondances between src(char string) and
1247 dest(wchar_t string) for optimization.
1248 e.g. src = "xxxyzz"
1249 dest = {'X', 'Y', 'Z'}
1250 (each "xxx", "y" and "zz" represent one multibyte character
1251 corresponding to 'X', 'Y' and 'Z'.)
1252 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1253 = {0, 3, 4, 6}
1254 */
1255 {
1265 /* Initialize the conversion state. */
1271 {
1272 #ifdef _LIBC
1274 #else
1276 #endif
1279 /* failed to convert. maybe src contains binary data.
1280 So we consume 1 byte manualy. */
1281 {
1285 }
1286 else
1288 /* In sjis encoding, we use yen sign as escape character in
1289 place of reverse solidus. So we convert 0x5c(yen sign in
1290 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1291 solidus in UCS2). */
1296 }
1298 /* Fill remain of the buffer with sentinel. */
1303 }
1309 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1310 also be assigned to arbitrarily: each pattern buffer stores its own
1311 syntax, so it can be changed between regex compilations. */
1312 /* This has no initializer because initialized variables in Emacs
1313 become read-only after dumping. */
1314 reg_syntax_t re_syntax_options;
1317 /* Specify the precise syntax of regexps for compilation. This provides
1318 for compatibility for various utilities which historically have
1319 different, incompatible syntaxes.
1321 The argument SYNTAX is a bit mask comprised of the various bits
1322 defined in regex.h. We return the old syntax. */
1324 reg_syntax_t
1326 {
1330 # ifdef DEBUG
1337 }
1338 # ifdef _LIBC
1340 # endif
1341 \f
1342 /* This table gives an error message for each of the error codes listed
1343 in regex.h. Obviously the order here has to be same as there.
1344 POSIX doesn't require that we do anything for REG_NOERROR,
1345 but why not be nice? */
1348 {
1366 };
1367 \f
1370 #ifndef DEFINED_ONCE
1371 /* Avoiding alloca during matching, to placate r_alloc. */
1373 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1374 searching and matching functions should not call alloca. On some
1375 systems, alloca is implemented in terms of malloc, and if we're
1376 using the relocating allocator routines, then malloc could cause a
1377 relocation, which might (if the strings being searched are in the
1378 ralloc heap) shift the data out from underneath the regexp
1379 routines.
1381 Here's another reason to avoid allocation: Emacs
1382 processes input from X in a signal handler; processing X input may
1383 call malloc; if input arrives while a matching routine is calling
1384 malloc, then we're scrod. But Emacs can't just block input while
1385 calling matching routines; then we don't notice interrupts when
1386 they come in. So, Emacs blocks input around all regexp calls
1387 except the matching calls, which it leaves unprotected, in the
1388 faith that they will not malloc. */
1390 /* Normally, this is fine. */
1391 # define MATCH_MAY_ALLOCATE
1393 /* When using GNU C, we are not REALLY using the C alloca, no matter
1394 what config.h may say. So don't take precautions for it. */
1395 # ifdef __GNUC__
1396 # undef C_ALLOCA
1397 # endif
1399 /* The match routines may not allocate if (1) they would do it with malloc
1400 and (2) it's not safe for them to use malloc.
1401 Note that if REL_ALLOC is defined, matching would not use malloc for the
1402 failure stack, but we would still use it for the register vectors;
1403 so REL_ALLOC should not affect this. */
1404 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1405 # undef MATCH_MAY_ALLOCATE
1406 # endif
1408 \f
1409 #ifdef INSIDE_RECURSION
1410 /* Failure stack declarations and macros; both re_compile_fastmap and
1411 re_match_2 use a failure stack. These have to be macros because of
1412 REGEX_ALLOCATE_STACK. */
1415 /* Number of failure points for which to initially allocate space
1416 when matching. If this number is exceeded, we allocate more
1417 space, so it is not a hard limit. */
1418 # ifndef INIT_FAILURE_ALLOC
1419 # define INIT_FAILURE_ALLOC 5
1420 # endif
1422 /* Roughly the maximum number of failure points on the stack. Would be
1423 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1424 This is a variable only so users of regex can assign to it; we never
1425 change it ourselves. */
1427 # ifdef INT_IS_16BIT
1429 # ifndef DEFINED_ONCE
1430 # if defined MATCH_MAY_ALLOCATE
1431 /* 4400 was enough to cause a crash on Alpha OSF/1,
1432 whose default stack limit is 2mb. */
1434 # else
1436 # endif
1437 # endif
1440 {
1443 };
1448 {
1456 # ifndef DEFINED_ONCE
1457 # if defined MATCH_MAY_ALLOCATE
1458 /* 4400 was enough to cause a crash on Alpha OSF/1,
1459 whose default stack limit is 2mb. */
1461 # else
1463 # endif
1464 # endif
1467 {
1470 };
1475 {
1483 # ifndef DEFINED_ONCE
1484 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1485 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1486 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1487 # endif
1490 /* Define macros to initialize and free the failure stack.
1491 Do `return -2' if the alloc fails. */
1493 # ifdef MATCH_MAY_ALLOCATE
1494 # define INIT_FAIL_STACK() \
1495 do { \
1496 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1497 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1498 \
1499 if (fail_stack.stack == NULL) \
1500 return -2; \
1501 \
1502 fail_stack.size = INIT_FAILURE_ALLOC; \
1503 fail_stack.avail = 0; \
1504 } while (0)
1506 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1507 # else
1508 # define INIT_FAIL_STACK() \
1509 do { \
1510 fail_stack.avail = 0; \
1511 } while (0)
1513 # define RESET_FAIL_STACK()
1514 # endif
1517 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1519 Return 1 if succeeds, and 0 if either ran out of memory
1520 allocating space for it or it was already too large.
1522 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1524 # define DOUBLE_FAIL_STACK(fail_stack) \
1525 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1526 ? 0 \
1527 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1528 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1529 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1530 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1531 \
1532 (fail_stack).stack == NULL \
1533 ? 0 \
1534 : ((fail_stack).size <<= 1, \
1535 1)))
1538 /* Push pointer POINTER on FAIL_STACK.
1539 Return 1 if was able to do so and 0 if ran out of memory allocating
1540 space to do so. */
1541 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1542 ((FAIL_STACK_FULL () \
1543 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1544 ? 0 \
1545 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1546 1))
1548 /* Push a pointer value onto the failure stack.
1549 Assumes the variable `fail_stack'. Probably should only
1550 be called from within `PUSH_FAILURE_POINT'. */
1551 # define PUSH_FAILURE_POINTER(item) \
1552 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1554 /* This pushes an integer-valued item onto the failure stack.
1555 Assumes the variable `fail_stack'. Probably should only
1556 be called from within `PUSH_FAILURE_POINT'. */
1557 # define PUSH_FAILURE_INT(item) \
1558 fail_stack.stack[fail_stack.avail++].integer = (item)
1560 /* Push a fail_stack_elt_t value onto the failure stack.
1561 Assumes the variable `fail_stack'. Probably should only
1562 be called from within `PUSH_FAILURE_POINT'. */
1563 # define PUSH_FAILURE_ELT(item) \
1564 fail_stack.stack[fail_stack.avail++] = (item)
1566 /* These three POP... operations complement the three PUSH... operations.
1567 All assume that `fail_stack' is nonempty. */
1568 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1569 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1570 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1572 /* Used to omit pushing failure point id's when we're not debugging. */
1573 # ifdef DEBUG
1574 # define DEBUG_PUSH PUSH_FAILURE_INT
1575 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1576 # else
1577 # define DEBUG_PUSH(item)
1578 # define DEBUG_POP(item_addr)
1579 # endif
1582 /* Push the information about the state we will need
1583 if we ever fail back to it.
1585 Requires variables fail_stack, regstart, regend, reg_info, and
1586 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1587 be declared.
1589 Does `return FAILURE_CODE' if runs out of memory. */
1591 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1592 do { \
1593 char *destination; \
1594 /* Must be int, so when we don't save any registers, the arithmetic \
1596 /* Can't be int, since there is not a shred of a guarantee that int \
1597 is wide enough to hold a value of something to which pointer can \
1599 active_reg_t this_reg; \
1600 \
1601 DEBUG_STATEMENT (failure_id++); \
1602 DEBUG_STATEMENT (nfailure_points_pushed++); \
1606 \
1609 \
1611 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1612 { \
1613 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1614 return failure_code; \
1615 \
1617 (fail_stack).size); \
1619 } \
1620 \
1623 \
1624 if (1) \
1625 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1626 this_reg++) \
1627 { \
1629 DEBUG_STATEMENT (num_regs_pushed++); \
1630 \
1632 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1633 \
1635 PUSH_FAILURE_POINTER (regend[this_reg]); \
1636 \
1638 reg_info[this_reg].word.pointer); \
1640 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1643 MATCHED_SOMETHING (reg_info[this_reg])); \
1645 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1647 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1648 } \
1649 \
1651 PUSH_FAILURE_INT (lowest_active_reg); \
1652 \
1654 PUSH_FAILURE_INT (highest_active_reg); \
1655 \
1657 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1658 PUSH_FAILURE_POINTER (pattern_place); \
1659 \
1661 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1662 size2); \
1664 PUSH_FAILURE_POINTER (string_place); \
1665 \
1667 DEBUG_PUSH (failure_id); \
1668 } while (0)
1670 # ifndef DEFINED_ONCE
1671 /* This is the number of items that are pushed and popped on the stack
1672 for each register. */
1673 # define NUM_REG_ITEMS 3
1675 /* Individual items aside from the registers. */
1676 # ifdef DEBUG
1678 # else
1679 # define NUM_NONREG_ITEMS 4
1680 # endif
1682 /* We push at most this many items on the stack. */
1683 /* We used to use (num_regs - 1), which is the number of registers
1684 this regexp will save; but that was changed to 5
1685 to avoid stack overflow for a regexp with lots of parens. */
1686 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1688 /* We actually push this many items. */
1689 # define NUM_FAILURE_ITEMS \
1690 (((0 \
1691 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1692 * NUM_REG_ITEMS) \
1693 + NUM_NONREG_ITEMS)
1695 /* How many items can still be added to the stack without overflowing it. */
1696 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1700 /* Pops what PUSH_FAIL_STACK pushes.
1702 We restore into the parameters, all of which should be lvalues:
1703 STR -- the saved data position.
1704 PAT -- the saved pattern position.
1705 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1706 REGSTART, REGEND -- arrays of string positions.
1707 REG_INFO -- array of information about each subexpression.
1709 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1710 `pend', `string1', `size1', `string2', and `size2'. */
1711 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1712 { \
1713 DEBUG_STATEMENT (unsigned failure_id;) \
1714 active_reg_t this_reg; \
1715 const UCHAR_T *string_temp; \
1716 \
1717 assert (!FAIL_STACK_EMPTY ()); \
1718 \
1723 \
1724 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1725 \
1726 DEBUG_POP (&failure_id); \
1728 \
1729 /* If the saved string location is NULL, it came from an \
1730 on_failure_keep_string_jump opcode, and we want to throw away the \
1732 string_temp = POP_FAILURE_POINTER (); \
1733 if (string_temp != NULL) \
1734 str = (const CHAR_T *) string_temp; \
1735 \
1737 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1739 \
1740 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1742 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1743 \
1745 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1747 \
1748 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1750 \
1751 if (1) \
1752 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1753 { \
1755 \
1756 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1758 reg_info[this_reg].word.pointer); \
1759 \
1760 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1762 \
1763 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1765 } \
1766 else \
1767 { \
1768 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1769 { \
1770 reg_info[this_reg].word.integer = 0; \
1771 regend[this_reg] = 0; \
1772 regstart[this_reg] = 0; \
1773 } \
1774 highest_active_reg = high_reg; \
1775 } \
1776 \
1777 set_regs_matched_done = 0; \
1778 DEBUG_STATEMENT (nfailure_points_popped++); \
1780 \f
1781 /* Structure for per-register (a.k.a. per-group) information.
1782 Other register information, such as the
1783 starting and ending positions (which are addresses), and the list of
1784 inner groups (which is a bits list) are maintained in separate
1785 variables.
1787 We are making a (strictly speaking) nonportable assumption here: that
1788 the compiler will pack our bit fields into something that fits into
1789 the type of `word', i.e., is something that fits into one item on the
1790 failure stack. */
1793 /* Declarations and macros for re_match_2. */
1796 {
1798 struct
1799 {
1800 /* This field is one if this group can match the empty string,
1801 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1802 # define MATCH_NULL_UNSET_VALUE 3
1810 # ifndef DEFINED_ONCE
1811 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1812 # define IS_ACTIVE(R) ((R).bits.is_active)
1813 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1814 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1817 /* Call this when have matched a real character; it sets `matched' flags
1818 for the subexpressions which we are currently inside. Also records
1819 that those subexprs have matched. */
1820 # define SET_REGS_MATCHED() \
1821 do \
1822 { \
1823 if (!set_regs_matched_done) \
1824 { \
1825 active_reg_t r; \
1826 set_regs_matched_done = 1; \
1827 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1828 { \
1829 MATCHED_SOMETHING (reg_info[r]) \
1830 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1831 = 1; \
1832 } \
1833 } \
1834 } \
1835 while (0)
1838 /* Registers are set to a sentinel when they haven't yet matched. */
1840 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1841 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1843 /* Subroutine declarations and macros for regex_compile. */
1853 reg_syntax_t syntax);
1856 reg_syntax_t syntax);
1857 # ifdef WCHAR
1862 reg_syntax_t syntax,
1871 reg_syntax_t syntax,
1875 /* Fetch the next character in the uncompiled pattern---translating it
1876 if necessary. Also cast from a signed character in the constant
1877 string passed to us by the user to an unsigned char that we can use
1878 as an array index (in, e.g., `translate'). */
1879 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1880 because it is impossible to allocate 4GB array for some encodings
1881 which have 4 byte character_set like UCS4. */
1882 # ifndef PATFETCH
1883 # ifdef WCHAR
1884 # define PATFETCH(c) \
1885 do {if (p == pend) return REG_EEND; \
1886 c = (UCHAR_T) *p++; \
1887 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1888 } while (0)
1890 # define PATFETCH(c) \
1891 do {if (p == pend) return REG_EEND; \
1892 c = (unsigned char) *p++; \
1893 if (translate) c = (unsigned char) translate[c]; \
1894 } while (0)
1896 # endif
1898 /* Fetch the next character in the uncompiled pattern, with no
1899 translation. */
1900 # define PATFETCH_RAW(c) \
1901 do {if (p == pend) return REG_EEND; \
1902 c = (UCHAR_T) *p++; \
1903 } while (0)
1905 /* Go backwards one character in the pattern. */
1906 # define PATUNFETCH p--
1909 /* If `translate' is non-null, return translate[D], else just D. We
1910 cast the subscript to translate because some data is declared as
1911 `char *', to avoid warnings when a string constant is passed. But
1912 when we use a character as a subscript we must make it unsigned. */
1913 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1914 because it is impossible to allocate 4GB array for some encodings
1915 which have 4 byte character_set like UCS4. */
1917 # ifndef TRANSLATE
1918 # ifdef WCHAR
1919 # define TRANSLATE(d) \
1920 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1921 ? (char) translate[(unsigned char) (d)] : (d))
1923 # define TRANSLATE(d) \
1924 (translate ? (char) translate[(unsigned char) (d)] : (char) (d))
1926 # endif
1929 /* Macros for outputting the compiled pattern into `buffer'. */
1931 /* If the buffer isn't allocated when it comes in, use this. */
1932 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1934 /* Make sure we have at least N more bytes of space in buffer. */
1935 # ifdef WCHAR
1936 # define GET_BUFFER_SPACE(n) \
1937 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
1938 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
1939 EXTEND_BUFFER ()
1941 # define GET_BUFFER_SPACE(n) \
1942 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1943 EXTEND_BUFFER ()
1946 /* Make sure we have one more byte of buffer space and then add C to it. */
1947 # define BUF_PUSH(c) \
1948 do { \
1949 GET_BUFFER_SPACE (1); \
1950 *b++ = (UCHAR_T) (c); \
1951 } while (0)
1954 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1955 # define BUF_PUSH_2(c1, c2) \
1956 do { \
1957 GET_BUFFER_SPACE (2); \
1958 *b++ = (UCHAR_T) (c1); \
1959 *b++ = (UCHAR_T) (c2); \
1960 } while (0)
1963 /* As with BUF_PUSH_2, except for three bytes. */
1964 # define BUF_PUSH_3(c1, c2, c3) \
1965 do { \
1966 GET_BUFFER_SPACE (3); \
1967 *b++ = (UCHAR_T) (c1); \
1968 *b++ = (UCHAR_T) (c2); \
1969 *b++ = (UCHAR_T) (c3); \
1970 } while (0)
1972 /* Store a jump with opcode OP at LOC to location TO. We store a
1973 relative address offset by the three bytes the jump itself occupies. */
1974 # define STORE_JUMP(op, loc, to) \
1975 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
1977 /* Likewise, for a two-argument jump. */
1978 # define STORE_JUMP2(op, loc, to, arg) \
1979 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
1981 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1982 # define INSERT_JUMP(op, loc, to) \
1983 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
1985 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1986 # define INSERT_JUMP2(op, loc, to, arg) \
1987 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
1988 arg, b)
1990 /* This is not an arbitrary limit: the arguments which represent offsets
1991 into the pattern are two bytes long. So if 2^16 bytes turns out to
1992 be too small, many things would have to change. */
1993 /* Any other compiler which, like MSC, has allocation limit below 2^16
1994 bytes will have to use approach similar to what was done below for
1995 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1996 reallocating to 0 bytes. Such thing is not going to work too well.
1997 You have been warned!! */
1998 # ifndef DEFINED_ONCE
1999 # if defined _MSC_VER && !defined WIN32
2000 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2001 The REALLOC define eliminates a flurry of conversion warnings,
2002 but is not required. */
2003 # define MAX_BUF_SIZE 65500L
2004 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2005 # else
2006 # define MAX_BUF_SIZE (1L << 16)
2007 # define REALLOC(p,s) realloc ((p), (s))
2008 # endif
2010 /* Extend the buffer by twice its current size via realloc and
2011 reset the pointers that pointed into the old block to point to the
2012 correct places in the new one. If extending the buffer results in it
2013 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2014 # if __BOUNDED_POINTERS__
2015 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2016 # define MOVE_BUFFER_POINTER(P) \
2017 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2018 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2019 else \
2020 { \
2021 SET_HIGH_BOUND (b); \
2022 SET_HIGH_BOUND (begalt); \
2023 if (fixup_alt_jump) \
2024 SET_HIGH_BOUND (fixup_alt_jump); \
2025 if (laststart) \
2026 SET_HIGH_BOUND (laststart); \
2027 if (pending_exact) \
2028 SET_HIGH_BOUND (pending_exact); \
2029 }
2030 # else
2031 # define MOVE_BUFFER_POINTER(P) (P) += incr
2032 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2033 # endif
2036 # ifdef WCHAR
2037 # define EXTEND_BUFFER() \
2038 do { \
2039 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2040 int wchar_count; \
2041 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2042 return REG_ESIZE; \
2043 bufp->allocated <<= 1; \
2044 if (bufp->allocated > MAX_BUF_SIZE) \
2045 bufp->allocated = MAX_BUF_SIZE; \
2047 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2048 if (wchar_count == 0) wchar_count = 1; \
2050 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2051 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2052 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2053 if (COMPILED_BUFFER_VAR == NULL) \
2054 return REG_ESPACE; \
2056 if (old_buffer != COMPILED_BUFFER_VAR) \
2057 { \
2058 PTR_INT_TYPE incr = COMPILED_BUFFER_VAR - old_buffer; \
2059 MOVE_BUFFER_POINTER (b); \
2060 MOVE_BUFFER_POINTER (begalt); \
2061 if (fixup_alt_jump) \
2062 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2063 if (laststart) \
2064 MOVE_BUFFER_POINTER (laststart); \
2065 if (pending_exact) \
2066 MOVE_BUFFER_POINTER (pending_exact); \
2067 } \
2068 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2069 } while (0)
2071 # define EXTEND_BUFFER() \
2072 do { \
2073 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2074 if (bufp->allocated == MAX_BUF_SIZE) \
2075 return REG_ESIZE; \
2076 bufp->allocated <<= 1; \
2077 if (bufp->allocated > MAX_BUF_SIZE) \
2078 bufp->allocated = MAX_BUF_SIZE; \
2079 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2080 bufp->allocated); \
2081 if (COMPILED_BUFFER_VAR == NULL) \
2082 return REG_ESPACE; \
2084 if (old_buffer != COMPILED_BUFFER_VAR) \
2085 { \
2086 PTR_INT_TYPE incr = COMPILED_BUFFER_VAR - old_buffer; \
2087 MOVE_BUFFER_POINTER (b); \
2088 MOVE_BUFFER_POINTER (begalt); \
2089 if (fixup_alt_jump) \
2090 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2091 if (laststart) \
2092 MOVE_BUFFER_POINTER (laststart); \
2093 if (pending_exact) \
2094 MOVE_BUFFER_POINTER (pending_exact); \
2095 } \
2096 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2097 } while (0)
2100 # ifndef DEFINED_ONCE
2101 /* Since we have one byte reserved for the register number argument to
2102 {start,stop}_memory, the maximum number of groups we can report
2103 things about is what fits in that byte. */
2104 # define MAX_REGNUM 255
2106 /* But patterns can have more than `MAX_REGNUM' registers. We just
2107 ignore the excess. */
2111 /* Macros for the compile stack. */
2113 /* Since offsets can go either forwards or backwards, this type needs to
2114 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2115 /* int may be not enough when sizeof(int) == 2. */
2119 {
2120 pattern_offset_t begalt_offset;
2121 pattern_offset_t fixup_alt_jump;
2122 pattern_offset_t inner_group_offset;
2123 pattern_offset_t laststart_offset;
2124 regnum_t regnum;
2129 {
2136 # define INIT_COMPILE_STACK_SIZE 32
2138 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2139 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2141 /* The next available element. */
2142 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2146 /* Set the bit for character C in a list. */
2147 # ifndef DEFINED_ONCE
2148 # define SET_LIST_BIT(c) \
2149 (b[((unsigned char) (c)) / BYTEWIDTH] \
2150 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2153 /* Get the next unsigned number in the uncompiled pattern. */
2154 # define GET_UNSIGNED_NUMBER(num) \
2155 { \
2156 while (p != pend) \
2157 { \
2158 PATFETCH (c); \
2160 break; \
2161 if (num <= RE_DUP_MAX) \
2162 { \
2163 if (num < 0) \
2164 num = 0; \
2166 } \
2167 } \
2168 }
2170 # ifndef DEFINED_ONCE
2171 # if defined _LIBC || WIDE_CHAR_SUPPORT
2172 /* The GNU C library provides support for user-defined character classes
2173 and the functions from ISO C amendement 1. */
2174 # ifdef CHARCLASS_NAME_MAX
2175 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2176 # else
2177 /* This shouldn't happen but some implementation might still have this
2178 problem. Use a reasonable default value. */
2179 # define CHAR_CLASS_MAX_LENGTH 256
2180 # endif
2182 # ifdef _LIBC
2183 # define IS_CHAR_CLASS(string) __wctype (string)
2184 # else
2185 # define IS_CHAR_CLASS(string) wctype (string)
2186 # endif
2187 # else
2190 # define IS_CHAR_CLASS(string) \
2197 # endif
2199 \f
2200 # ifndef MATCH_MAY_ALLOCATE
2202 /* If we cannot allocate large objects within re_match_2_internal,
2203 we make the fail stack and register vectors global.
2204 The fail stack, we grow to the maximum size when a regexp
2205 is compiled.
2206 The register vectors, we adjust in size each time we
2207 compile a regexp, according to the number of registers it needs. */
2211 /* Size with which the following vectors are currently allocated.
2212 That is so we can make them bigger as needed,
2213 but never make them smaller. */
2214 # ifdef DEFINED_ONCE
2226 /* Make the register vectors big enough for NUM_REGS registers,
2227 but don't make them smaller. */
2229 static void
2231 {
2233 {
2245 }
2246 }
2249 \f
2250 # ifndef DEFINED_ONCE
2252 regnum_t regnum);
2255 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2256 Returns one of error codes defined in `regex.h', or zero for success.
2258 Assumes the `allocated' (and perhaps `buffer') and `translate'
2259 fields are set in BUFP on entry.
2261 If it succeeds, results are put in BUFP (if it returns an error, the
2262 contents of BUFP are undefined):
2263 `buffer' is the compiled pattern;
2264 `syntax' is set to SYNTAX;
2265 `used' is set to the length of the compiled pattern;
2266 `fastmap_accurate' is zero;
2267 `re_nsub' is the number of subexpressions in PATTERN;
2268 `not_bol' and `not_eol' are zero;
2270 The `fastmap' and `newline_anchor' fields are neither
2271 examined nor set. */
2273 /* Return, freeing storage we allocated. */
2274 # ifdef WCHAR
2275 # define FREE_STACK_RETURN(value) \
2276 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2277 # else
2278 # define FREE_STACK_RETURN(value) \
2279 return (free (compile_stack.stack), value)
2282 static reg_errcode_t
2286 {
2287 /* We fetch characters from PATTERN here. Even though PATTERN is
2288 `char *' (i.e., signed), we declare these variables as unsigned, so
2289 they can be reliably used as array indices. */
2292 #ifdef WCHAR
2293 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2296 /* offset buffer for optimization. See convert_mbs_to_wc. */
2298 /* It hold whether each wchar_t is binary data or not. */
2300 /* A flag whether exactn is handling binary data or not. */
2304 /* A random temporary spot in PATTERN. */
2307 /* Points to the end of the buffer, where we should append. */
2310 /* Keeps track of unclosed groups. */
2311 compile_stack_type compile_stack;
2313 /* Points to the current (ending) position in the pattern. */
2314 #ifdef WCHAR
2322 /* How to translate the characters in the pattern. */
2325 /* Address of the count-byte of the most recently inserted `exactn'
2326 command. This makes it possible to tell if a new exact-match
2327 character can be added to that command or if the character requires
2328 a new `exactn' command. */
2331 /* Address of start of the most recently finished expression.
2332 This tells, e.g., postfix * where to find the start of its
2333 operand. Reset at the beginning of groups and alternatives. */
2336 /* Address of beginning of regexp, or inside of last group. */
2339 /* Address of the place where a forward jump should go to the end of
2340 the containing expression. Each alternative of an `or' -- except the
2341 last -- ends with a forward jump of this sort. */
2344 /* Counts open-groups as they are encountered. Remembered for the
2345 matching close-group on the compile stack, so the same register
2346 number is put in the stop_memory as the start_memory. */
2349 #ifdef WCHAR
2350 /* Initialize the wchar_t PATTERN and offset_buffer. */
2355 {
2360 }
2365 {
2370 }
2371 #endif
2373 #ifdef DEBUG
2376 {
2382 }
2385 /* Initialize the compile stack. */
2388 {
2389 #ifdef WCHAR
2393 #endif
2395 }
2400 /* Initialize the pattern buffer. */
2405 /* Set `used' to zero, so that if we return an error, the pattern
2406 printer (for debugging) will think there's no pattern. We reset it
2407 at the end. */
2410 /* Always count groups, whether or not bufp->no_sub is set. */
2413 #if !defined emacs && !defined SYNTAX_TABLE
2414 /* Initialize the syntax table. */
2416 #endif
2419 {
2422 enough space. This loses if buffer's address is bogus, but
2423 that is the user's responsibility. */
2424 #ifdef WCHAR
2425 /* Free bufp->buffer and allocate an array for wchar_t pattern
2426 buffer. */
2429 UCHAR_T);
2430 #else
2433 }
2434 else
2437 UCHAR_T);
2438 }
2441 #ifdef WCHAR
2445 }
2446 #ifdef WCHAR
2447 else
2449 #endif
2453 /* Loop through the uncompiled pattern until we're at the end. */
2455 {
2459 {
2461 {
2464 /* If context independent, it's an operator. */
2466 /* Otherwise, depends on what's come before. */
2469 else
2471 }
2476 {
2478 p == pend
2479 /* If context independent, it's an operator. */
2481 /* Otherwise, depends on what's next. */
2484 else
2486 }
2495 /* Fall through. */
2496 handle_plus:
2498 /* If there is no previous pattern... */
2500 {
2505 }
2507 {
2508 /* Are we optimizing this jump? */
2511 /* 1 means zero (many) matches is allowed. */
2514 /* If there is a sequence of repetition chars, collapse it
2515 down to just one (the right one). We can't combine
2516 interval operators with these because of, e.g., `a{2}*',
2517 which should only match an even number of `a's. */
2520 {
2531 ;
2534 {
2539 {
2540 PATUNFETCH;
2541 PATUNFETCH;
2543 }
2546 }
2547 else
2548 {
2549 PATUNFETCH;
2551 }
2553 /* If we get here, we found another repeat character. */
2554 }
2556 /* Star, etc. applied to an empty pattern is equivalent
2557 to an empty pattern. */
2561 /* Now we know whether or not zero matches is allowed
2562 and also whether or not two or more matches is allowed. */
2565 end a backward relative jump from `b' to before the next
2566 jump we're going to put in below (which jumps from
2567 laststart to after this jump).
2569 But if we are at the `*' in the exact sequence `.*\n',
2570 insert an unconditional jump backwards to the .,
2571 instead of the beginning of the loop. This way we only
2572 push a failure point once, instead of every time
2573 through the loop. */
2576 /* Allocate the space for the jump. */
2579 /* We know we are not at the first character of the pattern,
2580 because laststart was nonzero. And we've already
2581 incremented `p', by the way, to be the character after
2582 the `*'. Do we have to do something analogous here
2583 for null bytes, because of RE_DOT_NOT_NULL? */
2585 && zero_times_ok
2591 }
2592 else
2593 /* Anything else. */
2597 /* We've added more stuff to the buffer. */
2599 }
2601 /* On failure, jump from laststart to b + 3, which will be the
2602 end of the buffer after this jump is inserted. */
2603 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2604 'b + 3'. */
2613 {
2614 /* At least one repetition is required, so insert a
2615 `dummy_failure_jump' before the initial
2616 `on_failure_jump' instruction of the loop. This
2617 effects a skip over that instruction the first time
2618 we hit that loop. */
2623 }
2624 }
2635 {
2637 #ifdef WCHAR
2639 #else
2641 #endif
2644 #ifdef WCHAR
2645 /* We assume a charset(_not) structure as a wchar_t array.
2646 charset[0] = (re_opcode_t) charset(_not)
2647 charset[1] = l (= length of char_classes)
2648 charset[2] = m (= length of collating_symbols)
2649 charset[3] = n (= length of equivalence_classes)
2650 charset[4] = o (= length of char_ranges)
2651 charset[5] = p (= length of chars)
2653 charset[6] = char_class (wctype_t)
2654 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2655 ...
2656 charset[l+5] = char_class (wctype_t)
2658 charset[l+6] = collating_symbol (wchar_t)
2659 ...
2660 charset[l+m+5] = collating_symbol (wchar_t)
2661 ifdef _LIBC we use the index if
2662 _NL_COLLATE_SYMB_EXTRAMB instead of
2663 wchar_t string.
2665 charset[l+m+6] = equivalence_classes (wchar_t)
2666 ...
2667 charset[l+m+n+5] = equivalence_classes (wchar_t)
2668 ifdef _LIBC we use the index in
2669 _NL_COLLATE_WEIGHT instead of
2670 wchar_t string.
2672 charset[l+m+n+6] = range_start
2673 charset[l+m+n+7] = range_end
2674 ...
2675 charset[l+m+n+2o+4] = range_start
2676 charset[l+m+n+2o+5] = range_end
2677 ifdef _LIBC we use the value looked up
2678 in _NL_COLLATE_COLLSEQ instead of
2679 wchar_t character.
2681 charset[l+m+n+2o+6] = char
2682 ...
2683 charset[l+m+n+2o+p+5] = char
2685 */
2687 /* We need at least 6 spaces: the opcode, the length of
2688 char_classes, the length of collating_symbols, the length of
2689 equivalence_classes, the length of char_ranges, the length of
2690 chars. */
2693 /* Save b as laststart. And We use laststart as the pointer
2694 to the first element of the charset here.
2695 In other words, laststart[i] indicates charset[i]. */
2698 /* We test `*p == '^' twice, instead of using an if
2699 statement, so we only need one BUF_PUSH. */
2702 p++;
2704 /* Push the length of char_classes, the length of
2705 collating_symbols, the length of equivalence_classes, the
2706 length of char_ranges and the length of chars. */
2710 /* Remember the first position in the bracket expression. */
2713 /* charset_not matches newline according to a syntax bit. */
2716 {
2719 }
2721 /* Read in characters and ranges, setting map bits. */
2723 {
2728 /* \ might escape characters inside [...] and [^...]. */
2730 {
2738 }
2740 /* Could be the end of the bracket expression. If it's
2741 not (i.e., when the bracket expression is `[]' so
2742 far), the ']' character bit gets set way below. */
2746 /* Look ahead to see if it's a range when the last thing
2747 was a character class. */
2751 /* Look ahead to see if it's a range when the last thing
2752 was a character: if this is a hyphen not at the
2753 beginning or the end of a list, then it's the range
2754 operator. */
2759 {
2760 reg_errcode_t ret;
2761 /* Allocate the space for range_start and range_end. */
2763 /* Update the pointer to indicate end of buffer. */
2769 }
2772 reg_errcode_t ret;
2774 /* Move past the `-'. */
2776 /* Allocate the space for range_start and range_end. */
2778 /* Update the pointer to indicate end of buffer. */
2781 laststart);
2784 }
2786 /* See if we're at the beginning of a possible character
2787 class. */
2795 /* If pattern is `[[:'. */
2799 {
2805 else
2806 /* This is in any case an invalid class name. */
2808 }
2811 /* If isn't a word bracketed by `[:' and `:]':
2812 undo the ending character, the letters, and leave
2813 the leading `:' and `[' (but store them as character). */
2815 {
2819 /* Query the character class as wctype_t. */
2824 /* Throw away the ] at the end of the character
2825 class. */
2830 /* Allocate the space for character class. */
2832 /* Update the pointer to indicate end of buffer. */
2834 /* Move data which follow character classes
2835 not to violate the data. */
2842 /* Store the character class. */
2844 /* Update length of char_classes */
2848 }
2849 else
2850 {
2851 c1++;
2853 PATUNFETCH;
2859 }
2860 }
2863 {
2866 # ifdef _LIBC
2869 # endif
2873 /* If pattern is `[[=' or '[[.'. */
2877 {
2883 else
2884 /* This is in any case an invalid class name. */
2886 }
2890 {
2892 /* If we have no collation data we use the default
2893 collation in which each character is in a class
2894 by itself. It also means that ASCII is the
2895 character set and therefore we cannot have character
2896 with more than one byte in the multibyte
2897 representation. */
2899 /* If not defined _LIBC, we push the name and
2900 `\0' for the sake of matching performance. */
2903 # ifdef _LIBC
2906 # endif
2907 {
2910 }
2911 # ifdef _LIBC
2912 else
2913 {
2920 /* This #include defines a local function! */
2921 # include <locale/weightwc.h>
2924 {
2925 /* We push the index for equivalence class. */
2930 _NL_COLLATE_TABLEWC);
2933 _NL_COLLATE_WEIGHTWC);
2936 _NL_COLLATE_EXTRAWC);
2939 _NL_COLLATE_INDIRECTWC);
2943 /* This is no valid character. */
2947 }
2949 {
2950 /* We push collation sequence value
2951 for collating symbol. */
2961 /* We have to convert the name to a single-byte
2962 string. This is possible since the names
2963 consist of ASCII characters and the internal
2964 representation is UCS4. */
2968 table_size =
2970 _NL_COLLATE_SYMB_HASH_SIZEMB);
2973 _NL_COLLATE_SYMB_TABLEMB);
2976 _NL_COLLATE_SYMB_EXTRAMB);
2978 /* Locate the character in the hashing table. */
2985 {
2986 /* First compare the hashing value. */
2992 {
2993 /* Yep, this is the entry. */
2997 }
2999 /* Next entry. */
3001 }
3004 {
3005 /* Compute the index of the byte sequence
3006 in the table. */
3008 /* Adjust for the alignment. */
3012 }
3014 {
3015 /* No valid character. Match it as a
3016 single byte character. */
3019 /* Update the length of characters */
3023 /* Throw away the ] at the end of the
3024 collating symbol. */
3026 /* exit from the switch block. */
3028 }
3029 else
3031 }
3033 }
3034 # endif
3035 /* Throw away the ] at the end of the equivalence
3036 class (or collating symbol). */
3039 /* Allocate the space for the equivalence class
3040 (or collating symbol) (and '\0' if needed). */
3042 /* Update the pointer to indicate end of buffer. */
3047 /* Calculate the offset of char_ranges,
3048 which is next to equivalence_classes. */
3051 /* Insert space. */
3054 /* Write the equivalence_class and \0. */
3058 /* Update the length of equivalence_classes. */
3061 }
3064 /* Calculate the offset of the equivalence_classes,
3065 which is next to collating_symbols. */
3067 /* Insert space and write the collationg_symbol
3068 and \0. */
3073 /* In re_match_2_internal if range_start < -1, we
3074 assume -range_start is the offset of the
3075 collating symbol which is specified as
3076 the character of the range start. So we assign
3077 -(laststart[1] + laststart[2] + 6) to
3078 range_start. */
3080 /* Update the length of collating_symbol. */
3083 }
3084 }
3085 else
3086 {
3087 c1++;
3089 PATUNFETCH;
3095 }
3096 }
3097 else
3098 {
3103 }
3104 }
3107 /* Ensure that we have enough space to push a charset: the
3108 opcode, the length count, and the bitset; 34 bytes in all. */
3113 /* We test `*p == '^' twice, instead of using an if
3114 statement, so we only need one BUF_PUSH. */
3117 p++;
3119 /* Remember the first position in the bracket expression. */
3122 /* Push the number of bytes in the bitmap. */
3125 /* Clear the whole map. */
3128 /* charset_not matches newline according to a syntax bit. */
3133 /* Read in characters and ranges, setting map bits. */
3135 {
3140 /* \ might escape characters inside [...] and [^...]. */
3142 {
3149 }
3151 /* Could be the end of the bracket expression. If it's
3152 not (i.e., when the bracket expression is `[]' so
3153 far), the ']' character bit gets set way below. */
3157 /* Look ahead to see if it's a range when the last thing
3158 was a character class. */
3162 /* Look ahead to see if it's a range when the last thing
3163 was a character: if this is a hyphen not at the
3164 beginning or the end of a list, then it's the range
3165 operator. */
3170 {
3171 reg_errcode_t ret
3176 }
3180 reg_errcode_t ret;
3182 /* Move past the `-'. */
3188 }
3190 /* See if we're at the beginning of a possible character
3191 class. */
3200 /* If pattern is `[[:'. */
3204 {
3210 else
3211 /* This is in any case an invalid class name. */
3213 }
3216 /* If isn't a word bracketed by `[:' and `:]':
3217 undo the ending character, the letters, and leave
3218 the leading `:' and `[' (but set bits for them). */
3220 {
3221 # if defined _LIBC || WIDE_CHAR_SUPPORT
3231 /* Throw away the ] at the end of the character
3232 class. */
3238 {
3239 # ifdef _LIBC
3242 # else
3245 # endif
3250 }
3253 # else
3271 /* Throw away the ] at the end of the character
3272 class. */
3278 {
3279 /* This was split into 3 if's to
3280 avoid an arbitrary limit in some compiler. */
3299 }
3302 }
3303 else
3304 {
3305 c1++;
3307 PATUNFETCH;
3312 }
3313 }
3315 {
3317 # ifdef _LIBC
3320 # endif
3325 /* If pattern is `[[='. */
3329 {
3335 else
3336 /* This is in any case an invalid class name. */
3338 }
3342 {
3343 /* If we have no collation data we use the default
3344 collation in which each character is in a class
3345 by itself. It also means that ASCII is the
3346 character set and therefore we cannot have character
3347 with more than one byte in the multibyte
3348 representation. */
3349 # ifdef _LIBC
3351 # endif
3352 {
3356 /* Throw away the ] at the end of the equivalence
3357 class. */
3360 /* Set the bit for the character. */
3362 }
3363 # ifdef _LIBC
3364 else
3365 {
3366 /* Try to match the byte sequence in `str' against
3367 those known to the collate implementation.
3368 First find out whether the bytes in `str' are
3369 actually from exactly one character. */
3378 /* This #include defines a local function! */
3379 # include <locale/weight.h>
3392 /* This is no valid character. */
3395 /* Throw away the ] at the end of the equivalence
3396 class. */
3399 /* Now we have to go through the whole table
3400 and find all characters which have the same
3401 first level weight.
3403 XXX Note that this is not entirely correct.
3404 we would have to match multibyte sequences
3405 but this is not possible with the current
3406 implementation. */
3408 /* XXX This test would have to be changed if we
3409 would allow matching multibyte sequences. */
3411 {
3415 /* Test whether the lenghts match. */
3417 {
3418 /* They do. New compare the bytes of
3419 the weight. */
3428 /* They match. Mark the character as
3429 acceptable. */
3431 }
3432 }
3433 }
3434 # endif
3436 }
3437 else
3438 {
3439 c1++;
3441 PATUNFETCH;
3446 }
3447 }
3449 {
3451 # ifdef _LIBC
3454 # endif
3459 /* If pattern is `[[.'. */
3463 {
3469 else
3470 /* This is in any case an invalid class name. */
3472 }
3476 {
3477 /* If we have no collation data we use the default
3478 collation in which each character is the name
3479 for its own class which contains only the one
3480 character. It also means that ASCII is the
3481 character set and therefore we cannot have character
3482 with more than one byte in the multibyte
3483 representation. */
3484 # ifdef _LIBC
3486 # endif
3487 {
3491 /* Throw away the ] at the end of the equivalence
3492 class. */
3495 /* Set the bit for the character. */
3498 }
3499 # ifdef _LIBC
3500 else
3501 {
3502 /* Try to match the byte sequence in `str' against
3503 those known to the collate implementation.
3504 First find out whether the bytes in `str' are
3505 actually from exactly one character. */
3514 table_size =
3516 _NL_COLLATE_SYMB_HASH_SIZEMB);
3519 _NL_COLLATE_SYMB_TABLEMB);
3522 _NL_COLLATE_SYMB_EXTRAMB);
3524 /* Locate the character in the hashing table. */
3531 {
3532 /* First compare the hashing value. */
3539 {
3540 /* Yep, this is the entry. */
3544 }
3546 /* Next entry. */
3548 }
3551 /* This is no valid character. */
3554 /* Throw away the ] at the end of the equivalence
3555 class. */
3558 /* Now add the multibyte character(s) we found
3559 to the accept list.
3561 XXX Note that this is not entirely correct.
3562 we would have to match multibyte sequences
3563 but this is not possible with the current
3564 implementation. Also, we have to match
3565 collating symbols, which expand to more than
3566 one file, as a whole and not allow the
3567 individual bytes. */
3572 {
3575 }
3576 }
3577 # endif
3579 }
3580 else
3581 {
3582 c1++;
3584 PATUNFETCH;
3589 }
3590 }
3591 else
3592 {
3596 }
3597 }
3599 /* Discard any (non)matching list bytes that are all 0 at the
3600 end of the map. Decrease the map-length byte too. */
3605 }
3612 else
3619 else
3626 else
3633 else
3640 else
3647 /* Do not translate the character after the \, so that we can
3648 distinguish, e.g., \B from \b, even if we normally would
3649 translate, e.g., B to b. */
3653 {
3658 handle_open:
3660 regnum++;
3663 {
3665 compile_stack_elt_t);
3669 }
3671 /* These are the values to restore when we hit end of this
3672 group. They are all relative offsets, so that if the
3673 whole pattern moves because of realloc, they will still
3674 be valid. */
3676 COMPILE_STACK_TOP.fixup_alt_jump
3681 /* We will eventually replace the 0 with the number of
3682 groups inner to this one. But do not push a
3683 start_memory for groups beyond the last one we can
3684 represent in the compiled pattern. */
3686 {
3690 }
3697 /* If we've reached MAX_REGNUM groups, then this open
3698 won't actually generate any code, so we'll have to
3699 clear pending_exact explicitly. */
3708 {
3711 else
3713 }
3715 handle_close:
3718 alternative for a possible future
3719 `pop_failure_jump' to pop. See comments at
3720 `push_dummy_failure' in `re_match_2'. */
3723 /* We allocated space for this jump when we assigned
3724 to `fixup_alt_jump', in the `handle_alt' case below. */
3726 }
3728 /* See similar code for backslashed left paren above. */
3730 {
3733 else
3735 }
3737 /* Since we just checked for an empty stack above, this
3738 ``can't happen''. */
3740 {
3741 /* We don't just want to restore into `regnum', because
3742 later groups should continue to be numbered higher,
3743 as in `(ab)c(de)' -- the second group is #2. */
3744 regnum_t this_group_regnum;
3748 fixup_alt_jump
3754 /* If we've reached MAX_REGNUM groups, then this open
3755 won't actually generate any code, so we'll have to
3756 clear pending_exact explicitly. */
3759 /* We're at the end of the group, so now we know how many
3760 groups were inside this one. */
3762 {
3763 UCHAR_T *inner_group_loc
3769 }
3770 }
3777 handle_alt:
3781 /* Insert before the previous alternative a jump which
3782 jumps to this alternative if the former fails. */
3789 /* The alternative before this one has a jump after it
3790 which gets executed if it gets matched. Adjust that
3791 jump so it will jump to this alternative's analogous
3792 jump (put in below, which in turn will jump to the next
3793 (if any) alternative's such jump, etc.). The last such
3794 jump jumps to the correct final destination. A picture:
3795 _____ _____
3796 | | | |
3797 | v | v
3798 a | b | c
3800 If we are at `b', then fixup_alt_jump right now points to a
3801 three-byte space after `a'. We'll put in the jump, set
3802 fixup_alt_jump to right after `b', and leave behind three
3803 bytes which we'll fill in when we get to after `c'. */
3808 /* Mark and leave space for a jump after this alternative,
3809 to be filled in later either by next alternative or
3810 when know we're at the end of a series of alternatives. */
3821 /* If \{ is a literal. */
3823 /* If we're at `\{' and it's not the open-interval
3824 operator. */
3828 handle_interval:
3829 {
3830 /* If got here, then the syntax allows intervals. */
3832 /* At least (most) this many matches must be made. */
3835 /* Place in the uncompiled pattern (i.e., just after
3836 the '{') to go back to if the interval is invalid. */
3845 {
3849 }
3850 else
3851 /* Interval such as `{1}' => match exactly once. */
3858 {
3862 }
3867 /* If it's invalid to have no preceding re. */
3869 {
3875 else
3877 }
3879 /* We just parsed a valid interval. */
3884 /* If the upper bound is zero, don't want to succeed at
3885 all; jump from `laststart' to `b + 3', which will be
3886 the end of the buffer after we insert the jump. */
3887 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3888 instead of 'b + 3'. */
3890 {
3895 }
3897 /* Otherwise, we have a nontrivial interval. When
3898 we're all done, the pattern will look like:
3899 set_number_at <jump count> <upper bound>
3900 set_number_at <succeed_n count> <lower bound>
3901 succeed_n <after jump addr> <succeed_n count>
3902 <body of loop>
3903 jump_n <succeed_n addr> <jump count>
3904 (The upper bound and `jump_n' are omitted if
3905 `upper_bound' is 1, though.) */
3906 else
3908 more at the end of the loop. */
3914 /* Initialize lower bound of the `succeed_n', even
3915 though it will be set during matching by its
3916 attendant `set_number_at' (inserted next),
3917 because `re_compile_fastmap' needs to know.
3918 Jump to the `jump_n' we might insert below. */
3925 /* Code to initialize the lower bound. Insert
3926 before the `succeed_n'. The `5' is the last two
3927 bytes of this `set_number_at', plus 3 bytes of
3928 the following `succeed_n'. */
3929 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3930 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3931 of the following `succeed_n'. */
3938 append a backward jump to the `succeed_n'
3939 that starts this interval.
3941 When we've reached this during matching,
3942 we'll have matched the interval once, so
3943 jump back only `upper_bound - 1' times. */
3949 /* The location we want to set is the second
3950 parameter of the `jump_n'; that is `b-2' as
3951 an absolute address. `laststart' will be
3952 the `set_number_at' we're about to insert;
3953 `laststart+3' the number to set, the source
3954 for the relative address. But we are
3955 inserting into the middle of the pattern --
3956 so everything is getting moved up by 5.
3957 Conclusion: (b - 2) - (laststart + 3) + 5,
3958 i.e., b - laststart.
3960 We insert this at the beginning of the loop
3961 so that if we fail during matching, we'll
3962 reinitialize the bounds. */
3967 }
3968 }
3972 invalid_interval:
3975 unfetch_interval:
3976 /* Match the characters as literals. */
3981 else
3983 }
3985 #ifdef emacs
3986 /* There is no way to specify the before_dot and after_dot
3987 operators. rms says this is ok. --karl */
4068 /* Can't back reference to a subexpression if inside of it. */
4081 else
4085 normal_backslash:
4086 /* You might think it would be useful for \ to mean
4087 not to translate; but if we don't translate it
4088 it will never match anything. */
4091 }
4096 /* Expects the character in `c'. */
4097 normal_char:
4098 /* If no exactn currently being built. */
4100 #ifdef WCHAR
4101 /* If last exactn handle binary(or character) and
4102 new exactn handle character(or binary). */
4106 /* If last exactn not at current position. */
4109 /* We have only one byte following the exactn for the count. */
4112 /* If followed by a repetition operator. */
4121 {
4122 /* Start building a new exactn. */
4126 #ifdef WCHAR
4127 /* Is this exactn binary data or character? */
4131 else
4133 #else
4137 }
4146 /* Through the pattern now. */
4154 /* If we don't want backtracking, force success
4155 the first time we reach the end of the compiled pattern. */
4159 #ifdef WCHAR
4163 #endif
4166 /* We have succeeded; set the length of the buffer. */
4167 #ifdef WCHAR
4169 #else
4171 #endif
4173 #ifdef DEBUG
4175 {
4178 }
4181 #ifndef MATCH_MAY_ALLOCATE
4182 /* Initialize the failure stack to the largest possible stack. This
4183 isn't necessary unless we're trying to avoid calling alloca in
4184 the search and match routines. */
4185 {
4188 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4189 is strictly greater than re_max_failures, the largest possible stack
4190 is 2 * re_max_failures failure points. */
4192 {
4195 # ifdef emacs
4197 fail_stack.stack
4200 else
4201 fail_stack.stack
4207 fail_stack.stack
4210 else
4211 fail_stack.stack
4216 }
4219 }
4225 /* Subroutines for `regex_compile'. */
4227 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4228 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4230 static void
4232 {
4235 }
4238 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4239 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4241 static void
4243 {
4247 }
4250 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4251 for OP followed by two-byte integer parameter ARG. */
4252 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4254 static void
4256 {
4264 }
4267 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4268 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4270 static void
4273 {
4281 }
4284 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4285 after an alternative or a begin-subexpression. We assume there is at
4286 least one character before the ^. */
4288 static boolean
4290 reg_syntax_t syntax)
4291 {
4295 return
4296 /* After a subexpression? */
4298 /* After an alternative? */
4300 }
4303 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4304 at least one character after the $, i.e., `P < PEND'. */
4306 static boolean
4308 reg_syntax_t syntax)
4309 {
4314 return
4315 /* Before a subexpression? */
4318 /* Before an alternative? */
4321 }
4325 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4326 false if it's not. */
4328 static boolean
4330 {
4335 this_element--)
4340 }
4343 #ifdef INSIDE_RECURSION
4345 #ifdef WCHAR
4346 /* This insert space, which size is "num", into the pattern at "loc".
4347 "end" must point the end of the allocated buffer. */
4348 static void
4350 {
4356 }
4359 #ifdef WCHAR
4360 static reg_errcode_t
4364 {
4367 reg_errcode_t ret;
4368 # ifdef _LIBC
4371 # endif
4375 # ifdef _LIBC
4378 {
4380 _NL_COLLATE_COLLSEQWC);
4385 {
4386 /* range_start is a collating symbol. */
4388 /* Retreive the index and get collation sequence value. */
4391 }
4392 else
4397 /* Report an error if the range is empty and the syntax prohibits
4398 this. */
4402 /* Insert space to the end of the char_ranges. */
4407 }
4408 else
4409 # endif
4410 {
4412 range_start_char;
4414 /* Report an error if the range is empty and the syntax prohibits
4415 this. */
4419 /* Insert space to the end of the char_ranges. */
4424 }
4425 /* Have to increment the pointer into the pattern string, so the
4426 caller isn't still at the ending character. */
4430 }
4432 /* Read the ending character of a range (in a bracket expression) from the
4433 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4434 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4435 Then we set the translation of all bits between the starting and
4436 ending characters (inclusive) in the compiled pattern B.
4438 Return an error code.
4440 We use these short variable names so we can use the same macros as
4441 `regex_compile' itself. */
4443 static reg_errcode_t
4447 {
4450 reg_errcode_t ret;
4451 # if _LIBC
4455 # else
4457 # endif
4462 /* Have to increment the pointer into the pattern string, so the
4463 caller isn't still at the ending character. */
4466 /* Report an error if the range is empty and the syntax prohibits this. */
4469 # if _LIBC
4471 _NL_COLLATE_COLLSEQMB);
4476 {
4480 {
4483 }
4484 }
4485 # else
4486 /* Here we see why `this_char' has to be larger than an `unsigned
4487 char' -- we would otherwise go into an infinite loop, since all
4488 characters <= 0xff. */
4490 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4491 and some compilers cast it to int implicitly, so following for_loop
4492 may fall to (almost) infinite loop.
4493 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4494 To avoid this, we cast p[0] to unsigned int and truncate it. */
4498 {
4501 }
4502 # endif
4505 }
4507 \f
4508 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4509 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4510 characters can start a string that matches the pattern. This fastmap
4511 is used by re_search to skip quickly over impossible starting points.
4513 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4514 area as BUFP->fastmap.
4516 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4517 the pattern buffer.
4519 Returns 0 if we succeed, -2 if an internal error. */
4521 #ifdef WCHAR
4522 /* local function for re_compile_fastmap.
4523 truncate wchar_t character to char. */
4526 static unsigned char
4528 {
4533 # ifdef _LIBC
4535 # else
4537 # endif
4539 }
4542 static int
4544 {
4546 #ifdef MATCH_MAY_ALLOCATE
4548 #endif
4549 #ifndef REGEX_MALLOC
4551 #endif
4555 #ifdef WCHAR
4556 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4557 pattern to (char*) in regex_compile. */
4566 #ifdef REL_ALLOC
4567 /* This holds the pointer to the failure stack, when
4568 it is allocated relocatably. */
4570 #endif
4572 /* Assume that each path through the pattern can be null until
4573 proven otherwise. We set this false at the bottom of switch
4574 statement, to which we get only if a particular path doesn't
4575 match the empty string. */
4578 /* We aren't doing a `succeed_n' to begin with. */
4589 {
4591 {
4592 /* We have reached the (effective) end of pattern. */
4594 {
4597 /* Reset for next path. */
4603 }
4604 else
4606 }
4608 /* We should never be about to go beyond the end of the pattern. */
4612 {
4614 /* I guess the idea here is to simply not bother with a fastmap
4615 if a backreference is used, since it's too hard to figure out
4616 the fastmap for the corresponding group. Setting
4617 `can_be_null' stops `re_search_2' from using the fastmap, so
4618 that is all we do. */
4624 /* Following are the cases which match a character. These end
4625 with `break'. */
4627 #ifdef WCHAR
4636 #ifdef MBS_SUPPORT
4640 #endif
4642 #ifdef WCHAR
4643 /* It is hard to distinguish fastmap from (multi byte) characters
4644 which depends on current locale. */
4660 /* Chars beyond end of map must be allowed. */
4685 {
4688 /* `.' matches anything ... */
4692 /* ... except perhaps newline. */
4696 /* Return if we have already set `can_be_null'; if we have,
4697 then the fastmap is irrelevant. Something's wrong here. */
4701 /* Otherwise, have to check alternative paths. */
4703 }
4705 #ifdef emacs
4722 /* All cases after this match the empty string. These end with
4723 `continue'. */
4757 /* Jump backward implies we just went through the body of a
4758 loop and matched nothing. Opcode jumped to should be
4759 `on_failure_jump' or `succeed_n'. Just treat it like an
4760 ordinary jump. For a * loop, it has pushed its failure
4761 point already; if so, discard that as redundant. */
4766 p++;
4770 /* If what's on the stack is where we are now, pop it. */
4780 handle_on_failure_jump:
4783 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4784 end of the pattern. We don't want to push such a point,
4785 since when we restore it above, entering the switch will
4786 increment `p' past the end of the pattern. We don't need
4787 to push such a point since we obviously won't find any more
4788 fastmap entries beyond `pend'. Such a pattern can match
4789 the null string, though. */
4791 {
4793 {
4796 }
4797 }
4798 else
4802 {
4805 }
4811 /* Get to the number of times to succeed. */
4814 /* Increment p past the n for when k != 0. */
4817 {
4821 }
4840 /* Getting here means we have found the possible starting
4841 characters for one path of the pattern -- and that the empty
4842 string does not match. We need not follow this path further.
4843 Instead, look at the next alternative (remembered on the
4844 stack), or quit if no more. The test at the top of the loop
4845 does these things. */
4850 /* Set `can_be_null' for the last path (also the first path, if the
4851 pattern is empty). */
4854 done:
4857 }
4861 int
4863 {
4864 # ifdef MBS_SUPPORT
4867 else
4868 # endif
4871 #ifdef _LIBC
4873 #endif
4874 \f
4876 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4877 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4878 this memory for recording register information. STARTS and ENDS
4879 must be allocated using the malloc library routine, and must each
4880 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4882 If NUM_REGS == 0, then subsequent matches should allocate their own
4883 register data.
4885 Unless this function is called, the first search or match using
4886 PATTERN_BUFFER will allocate its own register data, without
4887 freeing the old data. */
4889 void
4893 {
4895 {
4900 }
4901 else
4902 {
4906 }
4907 }
4908 #ifdef _LIBC
4910 #endif
4911 \f
4912 /* Searching routines. */
4914 /* Like re_search_2, below, but only one string is specified, and
4915 doesn't let you say where to stop matching. */
4917 int
4920 {
4923 }
4924 #ifdef _LIBC
4926 #endif
4929 /* Using the compiled pattern in BUFP->buffer, first tries to match the
4930 virtual concatenation of STRING1 and STRING2, starting first at index
4931 STARTPOS, then at STARTPOS + 1, and so on.
4933 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
4935 RANGE is how far to scan while trying to match. RANGE = 0 means try
4936 only at STARTPOS; in general, the last start tried is STARTPOS +
4937 RANGE.
4939 In REGS, return the indices of the virtual concatenation of STRING1
4940 and STRING2 that matched the entire BUFP->buffer and its contained
4941 subexpressions.
4943 Do not consider matching one past the index STOP in the virtual
4944 concatenation of STRING1 and STRING2.
4946 We return either the position in the strings at which the match was
4947 found, -1 if no match, or -2 if error (such as failure
4948 stack overflow). */
4950 int
4954 {
4955 # ifdef MBS_SUPPORT
4959 else
4960 # endif
4964 #ifdef _LIBC
4966 #endif
4970 #ifdef INSIDE_RECURSION
4972 #ifdef MATCH_MAY_ALLOCATE
4973 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
4974 #else
4975 # define FREE_VAR(var) free (var); var = NULL
4976 #endif
4978 #ifdef WCHAR
4979 # define MAX_ALLOCA_SIZE 2000
4981 # define FREE_WCS_BUFFERS() \
4982 do { \
4983 if (size1 > MAX_ALLOCA_SIZE) \
4984 { \
4985 free (wcs_string1); \
4986 free (mbs_offset1); \
4987 } \
4988 else \
4989 { \
4990 FREE_VAR (wcs_string1); \
4991 FREE_VAR (mbs_offset1); \
4992 } \
4993 if (size2 > MAX_ALLOCA_SIZE) \
4994 { \
4995 free (wcs_string2); \
4996 free (mbs_offset2); \
4997 } \
4998 else \
4999 { \
5000 FREE_VAR (wcs_string2); \
5001 FREE_VAR (mbs_offset2); \
5002 } \
5003 } while (0)
5005 #endif
5008 static int
5013 {
5019 #ifdef WCHAR
5020 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5022 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5024 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5026 /* They hold whether each wchar_t is binary data or not. */
5030 /* Check for out-of-range STARTPOS. */
5034 /* Fix up RANGE if it might eventually take us outside
5035 the virtual concatenation of STRING1 and STRING2.
5036 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5042 /* If the search isn't to be a backwards one, don't waste time in a
5043 search for a pattern that must be anchored. */
5046 /* `begline' is like `begbuf' if it cannot match at newlines. */
5049 {
5052 else
5054 }
5056 #ifdef emacs
5057 /* In a forward search for something that starts with \=.
5058 don't keep searching past point. */
5060 {
5064 }
5067 /* Update the fastmap now if not correct already. */
5072 #ifdef WCHAR
5073 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5074 fill them with converted string. */
5076 {
5078 {
5082 }
5083 else
5084 {
5088 }
5090 {
5092 {
5096 }
5097 else
5098 {
5102 }
5104 }
5110 else
5112 }
5114 {
5116 {
5120 }
5121 else
5122 {
5126 }
5128 {
5132 else
5135 }
5141 else
5143 }
5147 /* Loop through the string, looking for a place to start matching. */
5149 {
5150 /* If a fastmap is supplied, skip quickly over characters that
5151 cannot be the start of a match. If the pattern can match the
5152 null string, however, we don't need to skip characters; we want
5153 the first null string. */
5155 {
5157 {
5167 /* Written out as an if-else to avoid testing `translate'
5168 inside the loop. */
5173 range--;
5174 else
5176 range--;
5179 }
5181 {
5188 }
5189 }
5191 /* If can't match the null string, and that's all we have left, fail. */
5194 {
5195 #ifdef WCHAR
5197 #endif
5199 }
5201 #ifdef WCHAR
5212 #ifndef REGEX_MALLOC
5213 # ifdef C_ALLOCA
5215 # endif
5216 #endif
5219 {
5220 #ifdef WCHAR
5222 #endif
5224 }
5227 {
5228 #ifdef WCHAR
5230 #endif
5232 }
5234 advance:
5238 {
5239 range--;
5240 startpos++;
5241 }
5242 else
5243 {
5244 range++;
5245 startpos--;
5246 }
5247 }
5248 #ifdef WCHAR
5250 #endif
5252 }
5254 #ifdef WCHAR
5255 /* This converts PTR, a pointer into one of the search wchar_t strings
5256 `string1' and `string2' into an multibyte string offset from the
5257 beginning of that string. We use mbs_offset to optimize.
5258 See convert_mbs_to_wcs. */
5259 # define POINTER_TO_OFFSET(ptr) \
5260 (FIRST_STRING_P (ptr) \
5261 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5262 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5263 + csize1)))
5265 /* This converts PTR, a pointer into one of the search strings `string1'
5266 and `string2' into an offset from the beginning of that string. */
5267 # define POINTER_TO_OFFSET(ptr) \
5268 (FIRST_STRING_P (ptr) \
5269 ? ((regoff_t) ((ptr) - string1)) \
5270 : ((regoff_t) ((ptr) - string2 + size1)))
5273 /* Macros for dealing with the split strings in re_match_2. */
5275 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5277 /* Call before fetching a character with *d. This switches over to
5278 string2 if necessary. */
5279 #define PREFETCH() \
5280 while (d == dend) \
5281 { \
5283 if (dend == end_match_2) \
5284 goto fail; \
5286 d = string2; \
5287 dend = end_match_2; \
5288 }
5290 /* Test if at very beginning or at very end of the virtual concatenation
5291 of `string1' and `string2'. If only one string, it's `string2'. */
5292 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5293 #define AT_STRINGS_END(d) ((d) == end2)
5296 /* Test if D points to a character which is word-constituent. We have
5297 two special cases to check for: if past the end of string1, look at
5298 the first character in string2; and if before the beginning of
5299 string2, look at the last character in string1. */
5300 #ifdef WCHAR
5301 /* Use internationalized API instead of SYNTAX. */
5302 # define WORDCHAR_P(d) \
5303 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5304 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5305 || ((d) == end1 ? *string2 \
5308 # define WORDCHAR_P(d) \
5309 (SYNTAX ((d) == end1 ? *string2 \
5310 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5311 == Sword)
5314 /* Disabled due to a compiler bug -- see comment at case wordbound */
5315 #if 0
5316 /* Test if the character before D and the one at D differ with respect
5317 to being word-constituent. */
5318 #define AT_WORD_BOUNDARY(d) \
5319 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5320 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5321 #endif
5323 /* Free everything we malloc. */
5324 #ifdef MATCH_MAY_ALLOCATE
5325 # ifdef WCHAR
5326 # define FREE_VARIABLES() \
5327 do { \
5328 REGEX_FREE_STACK (fail_stack.stack); \
5329 FREE_VAR (regstart); \
5330 FREE_VAR (regend); \
5331 FREE_VAR (old_regstart); \
5332 FREE_VAR (old_regend); \
5333 FREE_VAR (best_regstart); \
5334 FREE_VAR (best_regend); \
5335 FREE_VAR (reg_info); \
5336 FREE_VAR (reg_dummy); \
5337 FREE_VAR (reg_info_dummy); \
5338 if (!cant_free_wcs_buf) \
5339 { \
5340 FREE_VAR (string1); \
5341 FREE_VAR (string2); \
5342 FREE_VAR (mbs_offset1); \
5343 FREE_VAR (mbs_offset2); \
5344 } \
5345 } while (0)
5347 # define FREE_VARIABLES() \
5348 do { \
5349 REGEX_FREE_STACK (fail_stack.stack); \
5350 FREE_VAR (regstart); \
5351 FREE_VAR (regend); \
5352 FREE_VAR (old_regstart); \
5353 FREE_VAR (old_regend); \
5354 FREE_VAR (best_regstart); \
5355 FREE_VAR (best_regend); \
5356 FREE_VAR (reg_info); \
5357 FREE_VAR (reg_dummy); \
5358 FREE_VAR (reg_info_dummy); \
5359 } while (0)
5361 #else
5362 # ifdef WCHAR
5363 # define FREE_VARIABLES() \
5364 do { \
5365 if (!cant_free_wcs_buf) \
5366 { \
5367 FREE_VAR (string1); \
5368 FREE_VAR (string2); \
5369 FREE_VAR (mbs_offset1); \
5370 FREE_VAR (mbs_offset2); \
5371 } \
5372 } while (0)
5378 /* These values must meet several constraints. They must not be valid
5379 register values; since we have a limit of 255 registers (because
5380 we use only one byte in the pattern for the register number), we can
5381 use numbers larger than 255. They must differ by 1, because of
5382 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5383 be larger than the value for the highest register, so we do not try
5384 to actually save any registers when none are active. */
5385 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5386 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5387 \f
5389 /* Matching routines. */
5392 /* re_match is like re_match_2 except it takes only a single string. */
5394 int
5397 {
5399 # ifdef MBS_SUPPORT
5404 else
5405 # endif
5408 # ifndef REGEX_MALLOC
5409 # ifdef C_ALLOCA
5411 # endif
5412 # endif
5414 }
5415 # ifdef _LIBC
5417 # endif
5422 #ifdef INSIDE_RECURSION
5436 /* re_match_2 matches the compiled pattern in BUFP against the
5437 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5438 and SIZE2, respectively). We start matching at POS, and stop
5439 matching at STOP.
5441 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5442 store offsets for the substring each group matched in REGS. See the
5443 documentation for exactly how many groups we fill.
5445 We return -1 if no match, -2 if an internal error (such as the
5446 failure stack overflowing). Otherwise, we return the length of the
5447 matched substring. */
5449 int
5453 {
5455 # ifdef MBS_SUPPORT
5460 else
5461 # endif
5465 #ifndef REGEX_MALLOC
5466 # ifdef C_ALLOCA
5468 # endif
5469 #endif
5471 }
5472 #ifdef _LIBC
5474 #endif
5478 #ifdef INSIDE_RECURSION
5480 #ifdef WCHAR
5483 /* This check the substring (from 0, to length) of the multibyte string,
5484 to which offset_buffer correspond. And count how many wchar_t_characters
5485 the substring occupy. We use offset_buffer to optimization.
5486 See convert_mbs_to_wcs. */
5488 static int
5490 {
5493 /* Check whether the size is valid. */
5500 /* If there are no multibyte character, offset_buffer[i] == i.
5501 Optmize for this case. */
5505 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5510 {
5518 else
5520 }
5523 }
5526 /* This is a separate function so that we can force an alloca cleanup
5527 afterwards. */
5528 #ifdef WCHAR
5529 static int
5536 /* string1 == string2 == NULL means string1/2, size1/2 and
5537 mbs_offset1/2 need seting up in this function. */
5538 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5541 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5544 static int
5551 {
5552 /* General temporaries. */
5555 #ifdef WCHAR
5556 /* They hold whether each wchar_t is binary data or not. */
5558 /* If true, we can't free string1/2, mbs_offset1/2. */
5562 /* Just past the end of the corresponding string. */
5565 /* Pointers into string1 and string2, just past the last characters in
5566 each to consider matching. */
5569 /* Where we are in the data, and the end of the current string. */
5572 /* Where we are in the pattern, and the end of the pattern. */
5573 #ifdef WCHAR
5581 /* Mark the opcode just after a start_memory, so we can test for an
5582 empty subpattern when we get to the stop_memory. */
5585 /* We use this to map every character in the string. */
5588 /* Failure point stack. Each place that can handle a failure further
5589 down the line pushes a failure point on this stack. It consists of
5590 restart, regend, and reg_info for all registers corresponding to
5591 the subexpressions we're currently inside, plus the number of such
5592 registers, and, finally, two char *'s. The first char * is where
5593 to resume scanning the pattern; the second one is where to resume
5594 scanning the strings. If the latter is zero, the failure point is
5595 a ``dummy''; if a failure happens and the failure point is a dummy,
5596 it gets discarded and the next next one is tried. */
5599 #endif
5600 #ifdef DEBUG
5603 #endif
5605 #ifdef REL_ALLOC
5606 /* This holds the pointer to the failure stack, when
5607 it is allocated relocatably. */
5609 #endif
5611 /* We fill all the registers internally, independent of what we
5612 return, for use in backreferences. The number here includes
5613 an element for register zero. */
5616 /* The currently active registers. */
5620 /* Information on the contents of registers. These are pointers into
5621 the input strings; they record just what was matched (on this
5622 attempt) by a subexpression part of the pattern, that is, the
5623 regnum-th regstart pointer points to where in the pattern we began
5624 matching and the regnum-th regend points to right after where we
5625 stopped matching the regnum-th subexpression. (The zeroth register
5626 keeps track of what the whole pattern matches.) */
5629 #endif
5631 /* If a group that's operated upon by a repetition operator fails to
5632 match anything, then the register for its start will need to be
5633 restored because it will have been set to wherever in the string we
5634 are when we last see its open-group operator. Similarly for a
5635 register's end. */
5638 #endif
5640 /* The is_active field of reg_info helps us keep track of which (possibly
5641 nested) subexpressions we are currently in. The matched_something
5642 field of reg_info[reg_num] helps us tell whether or not we have
5643 matched any of the pattern so far this time through the reg_num-th
5644 subexpression. These two fields get reset each time through any
5645 loop their register is in. */
5648 #endif
5650 /* The following record the register info as found in the above
5651 variables when we find a match better than any we've seen before.
5652 This happens as we backtrack through the failure points, which in
5653 turn happens only if we have not yet matched the entire string. */
5657 #endif
5659 /* Logically, this is `best_regend[0]'. But we don't want to have to
5660 allocate space for that if we're not allocating space for anything
5661 else (see below). Also, we never need info about register 0 for
5662 any of the other register vectors, and it seems rather a kludge to
5663 treat `best_regend' differently than the rest. So we keep track of
5664 the end of the best match so far in a separate variable. We
5665 initialize this to NULL so that when we backtrack the first time
5666 and need to test it, it's not garbage. */
5669 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5672 /* Used when we pop values we don't care about. */
5676 #endif
5678 #ifdef DEBUG
5679 /* Counts the total number of registers pushed. */
5681 #endif
5687 #ifdef MATCH_MAY_ALLOCATE
5688 /* Do not bother to initialize all the register variables if there are
5689 no groups in the pattern, as it takes a fair amount of time. If
5690 there are groups, we include space for register 0 (the whole
5691 pattern), even though we never use it, since it simplifies the
5692 array indexing. We should fix this. */
5694 {
5707 {
5710 }
5711 }
5712 else
5713 {
5714 /* We must initialize all our variables to NULL, so that
5715 `FREE_VARIABLES' doesn't try to free them. */
5719 }
5722 /* The starting position is bogus. */
5723 #ifdef WCHAR
5727 #endif
5728 {
5731 }
5733 #ifdef WCHAR
5734 /* Allocate wchar_t array for string1 and string2 and
5735 fill them with converted string. */
5737 {
5738 /* We need seting up buffers here. */
5740 /* We must free wcs buffers in this function. */
5744 {
5749 {
5754 }
5755 }
5757 {
5762 {
5769 }
5774 }
5775 }
5777 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5778 pattern to (char*) in regex_compile. */
5784 /* Initialize subexpression text positions to -1 to mark ones that no
5785 start_memory/stop_memory has been seen for. Also initialize the
5786 register information struct. */
5788 {
5796 }
5798 /* We move `string1' into `string2' if the latter's empty -- but not if
5799 `string1' is null. */
5801 {
5806 #ifdef WCHAR
5811 #endif
5812 }
5816 /* Compute where to stop matching, within the two strings. */
5817 #ifdef WCHAR
5819 {
5823 }
5824 else
5825 {
5831 }
5836 }
5837 #else
5839 {
5842 }
5843 else
5844 {
5847 }
5850 /* `p' scans through the pattern as `d' scans through the data.
5851 `dend' is the end of the input string that `d' points within. `d'
5852 is advanced into the following input string whenever necessary, but
5853 this happens before fetching; therefore, at the beginning of the
5854 loop, `d' can be pointing at the end of a string, but it cannot
5855 equal `string2'. */
5856 #ifdef WCHAR
5858 {
5862 }
5863 else
5864 {
5868 }
5874 }
5875 #else
5877 {
5880 }
5881 else
5882 {
5885 }
5894 /* This loops over pattern commands. It exits by returning from the
5895 function if the match is complete, or it drops through if the match
5896 fails at this starting point in the input data. */
5898 {
5899 #ifdef _LIBC
5901 #else
5903 #endif
5909 /* If we haven't matched the entire string, and we want the
5910 longest match, try backtracking. */
5912 {
5913 /* 1 if this match ends in the same string (string1 or string2)
5914 as the best previous match. */
5915 boolean same_str_p;
5917 /* 1 if this match is the best seen so far. */
5918 boolean best_match_p;
5923 /* AIX compiler got confused when this was combined
5924 with the previous declaration. */
5927 else
5935 /* If exceeds best match so far, save it. */
5937 {
5944 {
5947 }
5948 }
5950 }
5952 /* If no failure points, don't restore garbage. And if
5953 last match is real best match, don't restore second
5954 best one. */
5956 {
5957 restore_best_regs:
5958 /* Restore best match. It may happen that `dend ==
5959 end_match_1' while the restored d is in string2.
5960 For example, the pattern `x.*y.*z' against the
5961 strings `x-' and `y-z-', if the two strings are
5962 not consecutive in memory. */
5970 {
5973 }
5974 }
5977 succeed_label:
5979 /* If caller wants register contents data back, do it. */
5981 {
5982 /* Have the register data arrays been allocated? */
5985 extra element beyond `num_regs' for the `-1' marker
5986 GNU code uses. */
5991 {
5994 }
5996 }
5999 allocated, reallocate them. If we need fewer, just
6000 leave it alone. */
6002 {
6007 {
6010 }
6011 }
6012 }
6013 else
6014 {
6015 /* These braces fend off a "empty body in an else-statement"
6016 warning under GCC when assert expands to nothing. */
6018 }
6020 /* Convert the pointer data in `regstart' and `regend' to
6021 indices. Register zero has to be set differently,
6022 since we haven't kept track of any info for it. */
6024 {
6026 #ifdef WCHAR
6030 else
6033 #else
6038 }
6040 /* Go through the first `min (num_regs, regs->num_regs)'
6041 registers, since that is all we initialized. */
6043 mcnt++)
6044 {
6047 else
6048 {
6053 }
6054 }
6056 /* If the regs structure we return has more elements than
6057 were in the pattern, set the extra elements to -1. If
6058 we (re)allocated the registers, this is the case,
6059 because we always allocate enough to have at least one
6060 -1 at the end. */
6070 #ifdef WCHAR
6073 else
6075 csize1;
6077 #else
6079 ? string1
6087 }
6089 /* Otherwise match next pattern command. */
6091 {
6092 /* Ignore these. Used to ignore the n of succeed_n's which
6093 currently have n == 0. */
6102 /* Match the next n pattern characters exactly. The following
6103 byte in the pattern defines n, and the n bytes after that
6104 are the characters to match. */
6106 #ifdef MBS_SUPPORT
6108 #endif
6112 /* This is written out as an if-else so we don't waste time
6113 testing `translate' inside the loop. */
6115 {
6116 do
6117 {
6119 #ifdef WCHAR
6121 {
6125 }
6126 else
6127 {
6130 }
6131 #else
6136 }
6138 }
6139 else
6140 {
6141 do
6142 {
6145 }
6147 }
6152 /* Match any character except possibly a newline or a null. */
6164 d++;
6170 {
6172 #ifdef WCHAR
6176 #define WORK_BUFFER_SIZE 128
6178 # ifdef _LIBC
6187 #ifdef WCHAR
6188 # ifdef _LIBC
6197 /* p points charset[6], so the address of the next instruction
6198 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6199 where l=length of char_classes, m=length of collating_symbol,
6200 n=equivalence_class, o=length of char_range,
6201 p'=length of character. */
6203 /* Update p to indicate the next instruction. */
6207 /* match with char_class? */
6209 {
6216 # ifdef _LIBC
6219 # else
6222 # endif
6223 }
6225 /* match with collating_symbol? */
6226 # ifdef _LIBC
6228 {
6233 workp++)
6234 {
6242 {
6243 /* Update d, however d will be incremented at
6244 char_set_matched:, we decrement d here. */
6247 }
6248 }
6249 }
6251 # endif
6252 /* If we can't look up collation data, we use wcscoll
6253 instead. */
6254 {
6256 {
6258 # ifdef _LIBC
6260 # else
6262 # endif
6264 /* If wcscoll(the collating symbol, whole string) > 0,
6265 any substring of the string never match with the
6266 collating symbol. */
6267 # ifdef _LIBC
6269 # else
6271 # endif
6272 {
6275 }
6277 /* First, we compare the collating symbol with
6278 the first character of the string.
6279 If it don't match, we add the next character to
6280 the compare buffer in turn. */
6282 {
6285 {
6290 }
6292 /* add next character to the compare buffer. */
6296 # ifdef _LIBC
6298 # else
6300 # endif
6305 /* (str_buf > workp) indicate (str_buf + X > workp),
6306 because for all X (str_buf + X > str_buf).
6307 So we don't need continue this loop. */
6310 /* Otherwise(str_buf < workp),
6311 (str_buf+next_character) may equals (workp).
6312 So we continue this loop. */
6313 }
6314 /* not matched */
6318 }
6319 }
6320 /* match with equivalence_class? */
6321 # ifdef _LIBC
6323 {
6325 /* Try to match the equivalence class against
6326 those known to the collate implementation. */
6335 /* This #include defines a local function! */
6336 # include <locale/weightwc.h>
6347 /* Write 1 collating element to str_buf, and
6348 get its index. */
6352 {
6355 {
6360 }
6364 }
6366 /* Update d, however d will be incremented at
6367 char_set_matched:, we decrement d here. */
6370 {
6373 else
6374 {
6377 }
6378 }
6383 workp++)
6384 {
6386 /* We already checked idx != 0 in regex_compile. */
6389 {
6397 }
6398 }
6399 /* not matched */
6402 }
6404 # endif
6405 /* If we can't look up collation data, we use wcscoll
6406 instead. */
6407 {
6409 {
6411 # ifdef _LIBC
6413 # else
6415 # endif
6417 /* If wcscoll(the collating symbol, whole string) > 0,
6418 any substring of the string never match with the
6419 collating symbol. */
6420 # ifdef _LIBC
6422 # else
6424 # endif
6425 {
6428 }
6430 /* First, we compare the equivalence class with
6431 the first character of the string.
6432 If it don't match, we add the next character to
6433 the compare buffer in turn. */
6435 {
6438 {
6443 }
6445 /* add next character to the compare buffer. */
6449 # ifdef _LIBC
6451 # else
6453 # endif
6459 /* (str_buf > workp) indicate (str_buf + X > workp),
6460 because for all X (str_buf + X > str_buf).
6461 So we don't need continue this loop. */
6464 /* Otherwise(str_buf < workp),
6465 (str_buf+next_character) may equals (workp).
6466 So we continue this loop. */
6467 }
6468 /* not matched */
6472 }
6473 }
6475 /* match with char_range? */
6476 # ifdef _LIBC
6478 {
6486 {
6489 /* We already compute the collation sequence value
6490 of the characters (or collating symbols). */
6496 }
6497 }
6498 else
6499 # endif
6500 {
6501 /* We set range_start_char at str_buf[0], range_end_char
6502 at str_buf[4], and compared char at str_buf[2]. */
6508 {
6511 /* match if (range_start_char <= c <= range_end_char). */
6513 /* If range_start(or end) < 0, we assume -range_start(end)
6514 is the offset of the collating symbol which is specified
6515 as the character of the range start(end). */
6517 /* range_start */
6520 else
6521 {
6524 }
6526 /* range_end */
6529 else
6530 {
6533 }
6535 # ifdef _LIBC
6538 # else
6541 # endif
6543 }
6544 }
6546 /* match with char? */
6553 char_set_matched:
6555 #else
6556 /* Cast to `unsigned' instead of `unsigned char' in case the
6557 bit list is a full 32 bytes long. */
6565 #undef WORK_BUFFER_SIZE
6568 d++;
6570 }
6573 /* The beginning of a group is represented by start_memory.
6574 The arguments are the register number in the next byte, and the
6575 number of groups inner to this one in the next. The text
6576 matched within the group is recorded (in the internal
6577 registers data structure) under the register number. */
6582 /* Find out if this group can match the empty string. */
6589 /* Save the position in the string where we were the last time
6590 we were at this open-group operator in case the group is
6591 operated upon by a repetition operator, e.g., with `(a*)*b'
6592 against `ab'; then we want to ignore where we are now in
6593 the string in case this attempt to match fails. */
6606 /* Clear this whenever we change the register activity status. */
6609 /* This is the new highest active register. */
6612 /* If nothing was active before, this is the new lowest active
6613 register. */
6617 /* Move past the register number and inner group count. */
6624 /* The stop_memory opcode represents the end of a group. Its
6625 arguments are the same as start_memory's: the register
6626 number, and the number of inner groups. */
6631 /* We need to save the string position the last time we were at
6632 this close-group operator in case the group is operated
6633 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6634 against `aba'; then we want to ignore where we are now in
6635 the string in case this attempt to match fails. */
6645 /* This register isn't active anymore. */
6648 /* Clear this whenever we change the register activity status. */
6651 /* If this was the only register active, nothing is active
6652 anymore. */
6654 {
6657 }
6658 else
6660 it isn't necessarily one less than now: consider
6661 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6662 new highest active register is 1. */
6665 r--;
6667 /* If we end up at register zero, that means that we saved
6668 the registers as the result of an `on_failure_jump', not
6669 a `start_memory', and we jumped to past the innermost
6670 `stop_memory'. For example, in ((.)*) we save
6671 registers 1 and 2 as a result of the *, but when we pop
6672 back to the second ), we are at the stop_memory 1.
6673 Thus, nothing is active. */
6675 {
6678 }
6679 else
6681 }
6683 /* If just failed to match something this time around with a
6684 group that's operated on by a repetition operator, try to
6685 force exit from the ``loop'', and restore the register
6686 information for this group that we had before trying this
6687 last match. */
6691 {
6697 {
6700 /* Fall through. */
6712 }
6715 /* If the next operation is a jump backwards in the pattern
6716 to an on_failure_jump right before the start_memory
6717 corresponding to this stop_memory, exit from the loop
6718 by forcing a failure after pushing on the stack the
6719 on_failure_jump's jump in the pattern, and d. */
6723 {
6724 /* If this group ever matched anything, then restore
6725 what its registers were before trying this last
6726 failed match, e.g., with `(a*)*b' against `ab' for
6727 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6728 against `aba' for regend[3].
6730 Also restore the registers for inner groups for,
6731 e.g., `((a*)(b*))*' against `aba' (register 3 would
6732 otherwise get trashed). */
6735 {
6740 /* Restore this and inner groups' (if any) registers. */
6742 r++)
6743 {
6746 /* xx why this test? */
6749 }
6750 }
6751 p1++;
6756 }
6757 }
6759 /* Move past the register number and the inner group count. */
6764 /* \<digit> has been turned into a `duplicate' command which is
6765 followed by the numeric value of <digit> as the register number. */
6767 {
6772 /* Can't back reference a group which we've never matched. */
6776 /* Where in input to try to start matching. */
6779 /* Where to stop matching; if both the place to start and
6780 the place to stop matching are in the same string, then
6781 set to the place to stop, otherwise, for now have to use
6782 the end of the first string. */
6788 {
6789 /* If necessary, advance to next segment in register
6790 contents. */
6792 {
6796 /* End of string1 => advance to string2. */
6799 }
6800 /* At end of register contents => success */
6803 /* If necessary, advance to next segment in data. */
6806 /* How many characters left in this segment to match. */
6809 /* Want how many consecutive characters we can match in
6810 one shot, so, if necessary, adjust the count. */
6814 /* Compare that many; failure if mismatch, else move
6815 past them. */
6822 /* Do this because we've match some characters. */
6824 }
6825 }
6829 /* begline matches the empty string at the beginning of the string
6830 (unless `not_bol' is set in `bufp'), and, if
6831 `newline_anchor' is set, after newlines. */
6836 {
6838 }
6840 {
6842 }
6843 /* In all other cases, we fail. */
6847 /* endline is the dual of begline. */
6852 {
6854 }
6856 /* We have to ``prefetch'' the next character. */
6859 {
6861 }
6865 /* Match at the very beginning of the data. */
6873 /* Match at the very end of the data. */
6881 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6882 pushes NULL as the value for the string on the stack. Then
6883 `pop_failure_point' will keep the current value for the
6884 string, instead of restoring it. To see why, consider
6885 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6886 then the . fails against the \n. But the next thing we want
6887 to do is match the \n against the \n; if we restored the
6888 string value, we would be back at the foo.
6890 Because this is used only in specific cases, we don't need to
6891 check all the things that `on_failure_jump' does, to make
6892 sure the right things get saved on the stack. Hence we don't
6893 share its code. The only reason to push anything on the
6894 stack at all is that otherwise we would have to change
6895 `anychar's code to do something besides goto fail in this
6896 case; that seems worse than this. */
6901 #ifdef _LIBC
6903 #else
6905 #endif
6911 /* Uses of on_failure_jump:
6913 Each alternative starts with an on_failure_jump that points
6914 to the beginning of the next alternative. Each alternative
6915 except the last ends with a jump that in effect jumps past
6916 the rest of the alternatives. (They really jump to the
6917 ending jump of the following alternative, because tensioning
6918 these jumps is a hassle.)
6920 Repeats start with an on_failure_jump that points past both
6921 the repetition text and either the following jump or
6922 pop_failure_jump back to this on_failure_jump. */
6924 on_failure:
6928 #ifdef _LIBC
6930 #else
6932 #endif
6934 /* If this on_failure_jump comes right before a group (i.e.,
6935 the original * applied to a group), save the information
6936 for that group and all inner ones, so that if we fail back
6937 to this point, the group's information will be correct.
6938 For example, in \(a*\)*\1, we need the preceding group,
6939 and in \(zz\(a*\)b*\)\2, we need the inner group. */
6941 /* We can't use `p' to check ahead because we push
6942 a failure point to `p + mcnt' after we do this. */
6945 /* We need to skip no_op's before we look for the
6946 start_memory in case this on_failure_jump is happening as
6947 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
6948 against aba. */
6950 p1++;
6953 {
6954 /* We have a new highest active register now. This will
6955 get reset at the start_memory we are about to get to,
6956 but we will have saved all the registers relevant to
6957 this repetition op, as described above. */
6961 }
6968 /* A smart repeat ends with `maybe_pop_jump'.
6969 We change it to either `pop_failure_jump' or `jump'. */
6973 {
6976 /* Compare the beginning of the repeat with what in the
6977 pattern follows its end. If we can establish that there
6978 is nothing that they would both match, i.e., that we
6979 would have to backtrack because of (as in, e.g., `a*a')
6980 then we can change to pop_failure_jump, because we'll
6981 never have to backtrack.
6983 This is not true in the case of alternatives: in
6984 `(a|ab)*' we do need to backtrack to the `ab' alternative
6985 (e.g., if the string was `ab'). But instead of trying to
6986 detect that here, the alternative has put on a dummy
6987 failure point which is what we will end up popping. */
6989 /* Skip over open/close-group commands.
6990 If what follows this loop is a ...+ construct,
6991 look at what begins its body, since we will have to
6992 match at least one of that. */
6994 {
7002 else
7004 }
7007 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7008 to the `maybe_finalize_jump' of this case. Examine what
7009 follows. */
7011 /* If we're at the end of the pattern, we can change. */
7013 {
7014 /* Consider what happens when matching ":\(.*\)"
7015 against ":/". I don't really understand this code
7016 yet. */
7018 pop_failure_jump;
7019 DEBUG_PRINT1
7021 }
7024 #ifdef MBS_SUPPORT
7026 #endif
7028 {
7029 register UCHAR_T c
7033 #ifdef MBS_SUPPORT
7035 #endif
7037 {
7039 pop_failure_jump;
7040 #ifdef WCHAR
7044 #else
7048 #endif
7049 }
7051 #ifndef WCHAR
7054 {
7061 /* `negate' is equal to 1 if c would match, which means
7062 that we can't change to pop_failure_jump. */
7064 {
7067 }
7068 }
7070 }
7071 #ifndef WCHAR
7073 {
7074 /* We win if the first character of the loop is not part
7075 of the charset. */
7080 {
7083 }
7086 {
7088 /* We win if the charset_not inside the loop
7089 lists every character listed in the charset after. */
7097 {
7100 }
7101 }
7103 {
7105 /* We win if the charset inside the loop
7106 has no overlap with the one after the loop. */
7109 idx++)
7114 {
7117 }
7118 }
7119 }
7121 }
7124 {
7128 }
7129 /* Fall through. */
7132 /* The end of a simple repeat has a pop_failure_jump back to
7133 its matching on_failure_jump, where the latter will push a
7134 failure point. The pop_failure_jump takes off failure
7135 points put on by this pop_failure_jump's matching
7136 on_failure_jump; we got through the pattern to here from the
7137 matching on_failure_jump, so didn't fail. */
7139 {
7140 /* We need to pass separate storage for the lowest and
7141 highest registers, even though we don't care about the
7142 actual values. Otherwise, we will restore only one
7143 register from the stack, since lowest will == highest in
7144 `pop_failure_point'. */
7153 }
7154 /* Fall through. */
7156 unconditional_jump:
7157 #ifdef _LIBC
7159 #else
7161 #endif
7162 /* Note fall through. */
7164 /* Unconditionally jump (without popping any failure points). */
7169 #ifdef _LIBC
7171 #else
7173 #endif
7177 /* We need this opcode so we can detect where alternatives end
7178 in `group_match_null_string_p' et al. */
7184 /* Normally, the on_failure_jump pushes a failure point, which
7185 then gets popped at pop_failure_jump. We will end up at
7186 pop_failure_jump, also, and with a pattern of, say, `a+', we
7187 are skipping over the on_failure_jump, so we have to push
7188 something meaningless for pop_failure_jump to pop. */
7191 /* It doesn't matter what we push for the string here. What
7192 the code at `fail' tests is the value for the pattern. */
7197 /* At the end of an alternative, we need to push a dummy failure
7198 point in case we are followed by a `pop_failure_jump', because
7199 we don't want the failure point for the alternative to be
7200 popped. For example, matching `(a|ab)*' against `aab'
7201 requires that we match the `ab' alternative. */
7204 /* See comments just above at `dummy_failure_jump' about the
7205 two zeroes. */
7209 /* Have to succeed matching what follows at least n times.
7210 After that, handle like `on_failure_jump'. */
7216 /* Originally, this is how many times we HAVE to succeed. */
7218 {
7219 mcnt--;
7222 #ifdef _LIBC
7225 #else
7228 #endif
7229 }
7231 {
7232 #ifdef _LIBC
7235 #else
7240 #ifdef WCHAR
7242 #else
7247 }
7254 /* Originally, this is how many times we CAN jump. */
7256 {
7257 mcnt--;
7260 #ifdef _LIBC
7262 mcnt);
7263 #else
7265 mcnt);
7268 }
7269 /* If don't have to jump any more, skip over the rest of command. */
7270 else
7275 {
7281 #ifdef _LIBC
7283 #else
7285 #endif
7288 }
7290 #if 0
7291 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7292 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7293 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7294 macro and introducing temporary variables works around the bug. */
7307 #else
7309 {
7321 }
7324 {
7336 }
7337 #endif
7353 #ifdef emacs
7380 matchsyntax:
7382 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7383 d++;
7397 matchnotsyntax:
7399 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7400 d++;
7413 d++;
7422 d++;
7428 }
7432 /* We goto here if a matching operation fails. */
7433 fail:
7441 /* If this failure point is a dummy, try the next one. */
7445 /* If we failed to the end of the pattern, don't examine *p. */
7448 {
7451 /* If failed to a backwards jump that's part of a repetition
7452 loop, need to pop this failure point and use the next one. */
7454 {
7457 /* Fall through. */
7466 || (!is_a_jump_n
7472 }
7473 }
7477 }
7478 else
7489 \f
7490 /* Subroutine definitions for re_match_2. */
7493 /* We are passed P pointing to a register number after a start_memory.
7495 Return true if the pattern up to the corresponding stop_memory can
7496 match the empty string, and false otherwise.
7498 If we find the matching stop_memory, sets P to point to one past its number.
7499 Otherwise, sets P to an undefined byte less than or equal to END.
7501 We don't handle duplicates properly (yet). */
7503 static boolean
7506 {
7508 /* Point to after the args to the start_memory. */
7512 {
7513 /* Skip over opcodes that can match nothing, and return true or
7514 false, as appropriate, when we get to one that can't, or to the
7515 matching stop_memory. */
7518 {
7519 /* Could be either a loop or a series of alternatives. */
7521 p1++;
7524 /* If the next operation is not a jump backwards in the
7525 pattern. */
7528 {
7529 /* Go through the on_failure_jumps of the alternatives,
7530 seeing if any of the alternatives cannot match nothing.
7531 The last alternative starts with only a jump,
7532 whereas the rest start with on_failure_jump and end
7533 with a jump, e.g., here is the pattern for `a|b|c':
7535 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7536 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7537 /exactn/1/c
7539 So, we have to first go through the first (n-1)
7540 alternatives and then deal with the last one separately. */
7543 /* Deal with the first (n-1) alternatives, which start
7544 with an on_failure_jump (see above) that jumps to right
7545 past a jump_past_alt. */
7548 jump_past_alt)
7549 {
7550 /* `mcnt' holds how many bytes long the alternative
7551 is, including the ending `jump_past_alt' and
7552 its number. */
7556 reg_info))
7559 /* Move to right after this alternative, including the
7560 jump_past_alt. */
7563 /* Break if it's the beginning of an n-th alternative
7564 that doesn't begin with an on_failure_jump. */
7568 /* Still have to check that it's not an n-th
7569 alternative that starts with an on_failure_jump. */
7570 p1++;
7573 jump_past_alt)
7574 {
7575 /* Get to the beginning of the n-th alternative. */
7578 }
7579 }
7581 /* Deal with the last alternative: go back and get number
7582 of the `jump_past_alt' just before it. `mcnt' contains
7583 the length of the alternative. */
7603 }
7610 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7611 It expects P to be the first byte of a single alternative and END one
7612 byte past the last. The alternative can contain groups. */
7614 static boolean
7617 {
7622 {
7623 /* Skip over opcodes that can match nothing, and break when we get
7624 to one that can't. */
7627 {
7628 /* It's a loop. */
7630 p1++;
7638 }
7645 /* Deals with the ops common to group_match_null_string_p and
7646 alt_match_null_string_p.
7648 Sets P to one after the op and its arguments, if any. */
7650 static boolean
7653 {
7655 boolean ret;
7660 {
7670 #ifdef emacs
7674 #endif
7682 /* Have to set this here in case we're checking a group which
7683 contains a group and a back reference to it. */
7692 /* If this is an optimized succeed_n for zero times, make the jump. */
7697 else
7702 /* Get to the number of times to succeed. */
7707 {
7711 }
7712 else
7726 /* All other opcodes mean we cannot match the empty string. */
7728 }
7735 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7736 bytes; nonzero otherwise. */
7738 static int
7740 RE_TRANSLATE_TYPE translate)
7741 {
7745 {
7746 #ifdef WCHAR
7753 len--;
7754 }
7756 }
7757 \f
7761 /* Entry points for GNU code. */
7763 /* re_compile_pattern is the GNU regular expression compiler: it
7764 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7765 Returns 0 if the pattern was valid, otherwise an error string.
7767 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7768 are set in BUFP on entry.
7770 We call regex_compile to do the actual compilation. */
7775 {
7776 reg_errcode_t ret;
7778 /* GNU code is written to assume at least RE_NREGS registers will be set
7779 (and at least one extra will be -1). */
7782 /* And GNU code determines whether or not to get register information
7783 by passing null for the REGS argument to re_match, etc., not by
7784 setting no_sub. */
7787 /* Match anchors at newline. */
7790 # ifdef MBS_SUPPORT
7793 else
7794 # endif
7800 }
7801 #ifdef _LIBC
7803 #endif
7804 \f
7805 /* Entry points compatible with 4.2 BSD regex library. We don't define
7806 them unless specifically requested. */
7808 #if defined _REGEX_RE_COMP || defined _LIBC
7810 /* BSD has one and only one pattern buffer. */
7814 #ifdef _LIBC
7815 /* Make these definitions weak in libc, so POSIX programs can redefine
7816 these names if they don't use our functions, and still use
7817 regcomp/regexec below without link errors. */
7818 weak_function
7819 #endif
7821 {
7822 reg_errcode_t ret;
7825 {
7829 }
7832 {
7841 }
7843 /* Since `re_exec' always passes NULL for the `regs' argument, we
7844 don't need to initialize the pattern buffer fields which affect it. */
7846 /* Match anchors at newlines. */
7849 # ifdef MBS_SUPPORT
7852 else
7853 # endif
7859 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7861 }
7864 int
7865 #ifdef _LIBC
7866 weak_function
7867 #endif
7869 {
7871 return
7873 }
7876 \f
7877 /* POSIX.2 functions. Don't define these for Emacs. */
7879 #ifndef emacs
7881 /* regcomp takes a regular expression as a string and compiles it.
7883 PREG is a regex_t *. We do not expect any fields to be initialized,
7884 since POSIX says we shouldn't. Thus, we set
7886 `buffer' to the compiled pattern;
7887 `used' to the length of the compiled pattern;
7888 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7889 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7890 RE_SYNTAX_POSIX_BASIC;
7891 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7892 `fastmap' to an allocated space for the fastmap;
7893 `fastmap_accurate' to zero;
7894 `re_nsub' to the number of subexpressions in PATTERN.
7896 PATTERN is the address of the pattern string.
7898 CFLAGS is a series of bits which affect compilation.
7900 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
7901 use POSIX basic syntax.
7903 If REG_NEWLINE is set, then . and [^...] don't match newline.
7904 Also, regexec will try a match beginning after every newline.
7906 If REG_ICASE is set, then we considers upper- and lowercase
7907 versions of letters to be equivalent when matching.
7909 If REG_NOSUB is set, then when PREG is passed to regexec, that
7910 routine will report only success or failure, and nothing about the
7911 registers.
7913 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
7914 the return codes and their meanings.) */
7916 int
7918 {
7919 reg_errcode_t ret;
7920 reg_syntax_t syntax
7924 /* regex_compile will allocate the space for the compiled pattern. */
7929 /* Try to allocate space for the fastmap. */
7933 {
7936 preg->translate
7942 /* Map uppercase characters to corresponding lowercase ones. */
7945 }
7946 else
7949 /* If REG_NEWLINE is set, newlines are treated differently. */
7954 /* It also changes the matching behavior. */
7956 }
7957 else
7962 /* POSIX says a null character in the pattern terminates it, so we
7963 can use strlen here in compiling the pattern. */
7964 # ifdef MBS_SUPPORT
7967 else
7968 # endif
7971 /* POSIX doesn't distinguish between an unmatched open-group and an
7972 unmatched close-group: both are REG_EPAREN. */
7976 {
7977 /* Compute the fastmap now, since regexec cannot modify the pattern
7978 buffer. */
7980 {
7981 /* Some error occurred while computing the fastmap, just forget
7982 about it. */
7985 }
7986 }
7989 }
7990 #ifdef _LIBC
7992 #endif
7995 /* regexec searches for a given pattern, specified by PREG, in the
7996 string STRING.
7998 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
7999 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
8000 least NMATCH elements, and we set them to the offsets of the
8001 corresponding matched substrings.
8003 EFLAGS specifies `execution flags' which affect matching: if
8004 REG_NOTBOL is set, then ^ does not match at the beginning of the
8005 string; if REG_NOTEOL is set, then $ does not match at the end.
8007 We return 0 if we find a match and REG_NOMATCH if not. */
8009 int
8012 {
8015 regex_t private_preg;
8024 /* The user has told us exactly how many registers to return
8025 information about, via `nmatch'. We have to pass that on to the
8026 matching routines. */
8030 {
8036 }
8038 /* Perform the searching operation. */
8043 /* Copy the register information to the POSIX structure. */
8045 {
8047 {
8051 {
8054 }
8055 }
8057 /* If we needed the temporary register info, free the space now. */
8059 }
8061 /* We want zero return to mean success, unlike `re_search'. */
8063 }
8064 #ifdef _LIBC
8066 #endif
8069 /* Returns a message corresponding to an error code, ERRCODE, returned
8070 from either regcomp or regexec. We don't use PREG here. */
8072 size_t
8075 {
8082 /* Only error codes returned by the rest of the code should be passed
8083 to this routine. If we are given anything else, or if other regex
8084 code generates an invalid error code, then the program has a bug.
8085 Dump core so we can fix it. */
8093 {
8095 {
8096 #if defined HAVE_MEMPCPY || defined _LIBC
8098 #else
8101 #endif
8102 }
8103 else
8105 }
8108 }
8109 #ifdef _LIBC
8111 #endif
8114 /* Free dynamically allocated space used by PREG. */
8116 void
8118 {
8131 }
8132 #ifdef _LIBC
8134 #endif
8140 \f
8141 #undef STORE_NUMBER
8142 #undef STORE_NUMBER_AND_INCR
8143 #undef EXTRACT_NUMBER
8144 #undef EXTRACT_NUMBER_AND_INCR
8146 #undef DEBUG_PRINT_COMPILED_PATTERN
8147 #undef DEBUG_PRINT_DOUBLE_STRING
8149 #undef INIT_FAIL_STACK
8150 #undef RESET_FAIL_STACK
8151 #undef DOUBLE_FAIL_STACK
8152 #undef PUSH_PATTERN_OP
8153 #undef PUSH_FAILURE_POINTER
8154 #undef PUSH_FAILURE_INT
8155 #undef PUSH_FAILURE_ELT
8156 #undef POP_FAILURE_POINTER
8157 #undef POP_FAILURE_INT
8158 #undef POP_FAILURE_ELT
8159 #undef DEBUG_PUSH
8160 #undef DEBUG_POP
8161 #undef PUSH_FAILURE_POINT
8162 #undef POP_FAILURE_POINT
8164 #undef REG_UNSET_VALUE
8165 #undef REG_UNSET
8167 #undef PATFETCH
8168 #undef PATFETCH_RAW
8169 #undef PATUNFETCH
8170 #undef TRANSLATE
8172 #undef INIT_BUF_SIZE
8173 #undef GET_BUFFER_SPACE
8174 #undef BUF_PUSH
8175 #undef BUF_PUSH_2
8176 #undef BUF_PUSH_3
8177 #undef STORE_JUMP
8178 #undef STORE_JUMP2
8179 #undef INSERT_JUMP
8180 #undef INSERT_JUMP2
8181 #undef EXTEND_BUFFER
8182 #undef GET_UNSIGNED_NUMBER
8183 #undef FREE_STACK_RETURN
8185 # undef POINTER_TO_OFFSET
8186 # undef MATCHING_IN_FRST_STRING
8187 # undef PREFETCH
8188 # undef AT_STRINGS_BEG
8189 # undef AT_STRINGS_END
8190 # undef WORDCHAR_P
8191 # undef FREE_VAR
8192 # undef FREE_VARIABLES
8193 # undef NO_HIGHEST_ACTIVE_REG
8194 # undef NO_LOWEST_ACTIVE_REG
8196 # undef CHAR_T
8197 # undef UCHAR_T
8198 # undef COMPILED_BUFFER_VAR
8199 # undef OFFSET_ADDRESS_SIZE
8200 # undef CHAR_CLASS_SIZE
8201 # undef PREFIX
8202 # undef ARG_PREFIX
8203 # undef PUT_CHAR
8204 # undef BYTE
8205 # undef WCHAR
8207 # define DEFINED_ONCE