| blob | 420c7f4a47da26537ce0439a4716859c969fbd01 |
1 /* Extended regular expression matching and search library,
2 version 0.12.
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
6 Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
7 2002, 2005, 2010 Free Software Foundation, Inc.
8 This file is part of the GNU C Library.
10 The GNU C Library is free software; you can redistribute it and/or
11 modify it under the terms of the GNU Lesser General Public
12 License as published by the Free Software Foundation; either
13 version 2.1 of the License, or (at your option) any later version.
15 The GNU C Library is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 Lesser General Public License for more details.
20 You should have received a copy of the GNU Lesser General Public
21 License along with the GNU C Library; if not, write to the Free
22 Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
23 02110-1301 USA. */
25 /* This file has been modified for usage in libiberty. It includes "xregex.h"
26 instead of <regex.h>. The "xregex.h" header file renames all external
27 routines with an "x" prefix so they do not collide with the native regex
28 routines or with other components regex routines. */
29 /* AIX requires this to be the first thing in the file. */
30 #if defined _AIX && !defined __GNUC__ && !defined REGEX_MALLOC
31 #pragma alloca
32 #endif
34 #undef _GNU_SOURCE
35 #define _GNU_SOURCE
37 #ifndef INSIDE_RECURSION
38 # ifdef HAVE_CONFIG_H
39 # include <config.h>
40 # endif
41 #endif
43 #include <ansidecl.h>
45 #ifndef INSIDE_RECURSION
47 # if defined STDC_HEADERS && !defined emacs
48 # include <stddef.h>
49 # else
50 /* We need this for `regex.h', and perhaps for the Emacs include files. */
51 # include <sys/types.h>
52 # endif
54 # define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
56 /* For platform which support the ISO C amendement 1 functionality we
57 support user defined character classes. */
58 # if defined _LIBC || WIDE_CHAR_SUPPORT
59 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
60 # include <wchar.h>
61 # include <wctype.h>
62 # endif
64 # ifdef _LIBC
65 /* We have to keep the namespace clean. */
66 # define regfree(preg) __regfree (preg)
67 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
68 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
69 # define regerror(errcode, preg, errbuf, errbuf_size) \
70 __regerror(errcode, preg, errbuf, errbuf_size)
71 # define re_set_registers(bu, re, nu, st, en) \
72 __re_set_registers (bu, re, nu, st, en)
73 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
74 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
75 # define re_match(bufp, string, size, pos, regs) \
76 __re_match (bufp, string, size, pos, regs)
77 # define re_search(bufp, string, size, startpos, range, regs) \
78 __re_search (bufp, string, size, startpos, range, regs)
79 # define re_compile_pattern(pattern, length, bufp) \
80 __re_compile_pattern (pattern, length, bufp)
81 # define re_set_syntax(syntax) __re_set_syntax (syntax)
82 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
83 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
84 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
86 # define btowc __btowc
88 /* We are also using some library internals. */
89 # include <locale/localeinfo.h>
90 # include <locale/elem-hash.h>
91 # include <langinfo.h>
92 # include <locale/coll-lookup.h>
93 # endif
95 /* This is for other GNU distributions with internationalized messages. */
96 # if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
97 # include <libintl.h>
98 # ifdef _LIBC
99 # undef gettext
101 # endif
102 # else
103 # define gettext(msgid) (msgid)
104 # endif
106 # ifndef gettext_noop
107 /* This define is so xgettext can find the internationalizable
108 strings. */
109 # define gettext_noop(String) String
110 # endif
112 /* The `emacs' switch turns on certain matching commands
113 that make sense only in Emacs. */
114 # ifdef emacs
122 /* If we are not linking with Emacs proper,
123 we can't use the relocating allocator
124 even if config.h says that we can. */
125 # undef REL_ALLOC
127 # if defined STDC_HEADERS || defined _LIBC
128 # include <stdlib.h>
129 # else
132 # endif
134 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
135 If nothing else has been done, use the method below. */
136 # ifdef INHIBIT_STRING_HEADER
137 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
138 # if !defined bzero && !defined bcopy
139 # undef INHIBIT_STRING_HEADER
140 # endif
141 # endif
142 # endif
144 /* This is the normal way of making sure we have a bcopy and a bzero.
145 This is used in most programs--a few other programs avoid this
146 by defining INHIBIT_STRING_HEADER. */
147 # ifndef INHIBIT_STRING_HEADER
148 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
149 # include <string.h>
150 # ifndef bzero
151 # ifndef _LIBC
153 # else
154 # define bzero(s, n) __bzero (s, n)
155 # endif
156 # endif
157 # else
158 # include <strings.h>
159 # ifndef memcmp
160 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
161 # endif
162 # ifndef memcpy
163 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
164 # endif
165 # endif
166 # endif
168 /* Define the syntax stuff for \<, \>, etc. */
170 /* This must be nonzero for the wordchar and notwordchar pattern
171 commands in re_match_2. */
172 # ifndef Sword
173 # define Sword 1
174 # endif
176 # ifdef SWITCH_ENUM_BUG
177 # define SWITCH_ENUM_CAST(x) ((int)(x))
178 # else
179 # define SWITCH_ENUM_CAST(x) (x)
180 # endif
184 # if defined _LIBC || HAVE_LIMITS_H
185 # include <limits.h>
186 # endif
188 # ifndef MB_LEN_MAX
189 # define MB_LEN_MAX 1
190 # endif
191 \f
192 /* Get the interface, including the syntax bits. */
195 /* isalpha etc. are used for the character classes. */
196 # include <ctype.h>
198 /* Jim Meyering writes:
200 "... Some ctype macros are valid only for character codes that
201 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
202 using /bin/cc or gcc but without giving an ansi option). So, all
203 ctype uses should be through macros like ISPRINT... If
204 STDC_HEADERS is defined, then autoconf has verified that the ctype
205 macros don't need to be guarded with references to isascii. ...
206 Defining isascii to 1 should let any compiler worth its salt
207 eliminate the && through constant folding."
208 Solaris defines some of these symbols so we must undefine them first. */
210 # undef ISASCII
211 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
212 # define ISASCII(c) 1
213 # else
214 # define ISASCII(c) isascii(c)
215 # endif
217 # ifdef isblank
218 # define ISBLANK(c) (ISASCII (c) && isblank (c))
219 # else
221 # endif
222 # ifdef isgraph
223 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
224 # else
225 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
226 # endif
228 # undef ISPRINT
229 # define ISPRINT(c) (ISASCII (c) && isprint (c))
230 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
231 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
232 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
233 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
234 # define ISLOWER(c) (ISASCII (c) && islower (c))
235 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
236 # define ISSPACE(c) (ISASCII (c) && isspace (c))
237 # define ISUPPER(c) (ISASCII (c) && isupper (c))
238 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
240 # ifdef _tolower
241 # define TOLOWER(c) _tolower(c)
242 # else
243 # define TOLOWER(c) tolower(c)
244 # endif
246 # ifndef NULL
247 # define NULL (void *)0
248 # endif
250 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
251 since ours (we hope) works properly with all combinations of
252 machines, compilers, `char' and `unsigned char' argument types.
253 (Per Bothner suggested the basic approach.) */
254 # undef SIGN_EXTEND_CHAR
255 # if __STDC__
256 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
258 /* As in Harbison and Steele. */
259 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
260 # endif
261 \f
262 # ifndef emacs
263 /* How many characters in the character set. */
264 # define CHAR_SET_SIZE 256
266 # ifdef SYNTAX_TABLE
276 static void
278 {
293 }
297 # define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
300 \f
301 /* Integer type for pointers. */
302 # if !defined _LIBC && !defined HAVE_UINTPTR_T
304 # endif
306 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
307 use `alloca' instead of `malloc'. This is because using malloc in
308 re_search* or re_match* could cause memory leaks when C-g is used in
309 Emacs; also, malloc is slower and causes storage fragmentation. On
310 the other hand, malloc is more portable, and easier to debug.
312 Because we sometimes use alloca, some routines have to be macros,
313 not functions -- `alloca'-allocated space disappears at the end of the
314 function it is called in. */
316 # ifdef REGEX_MALLOC
318 # define REGEX_ALLOCATE malloc
319 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
320 # define REGEX_FREE free
324 /* Emacs already defines alloca, sometimes. */
325 # ifndef alloca
327 /* Make alloca work the best possible way. */
328 # ifdef __GNUC__
329 # define alloca __builtin_alloca
331 # if HAVE_ALLOCA_H
332 # include <alloca.h>
338 # define REGEX_ALLOCATE alloca
340 /* Assumes a `char *destination' variable. */
341 # define REGEX_REALLOCATE(source, osize, nsize) \
342 (destination = (char *) alloca (nsize), \
343 memcpy (destination, source, osize))
345 /* No need to do anything to free, after alloca. */
350 /* Define how to allocate the failure stack. */
352 # if defined REL_ALLOC && defined REGEX_MALLOC
354 # define REGEX_ALLOCATE_STACK(size) \
355 r_alloc (&failure_stack_ptr, (size))
356 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
357 r_re_alloc (&failure_stack_ptr, (nsize))
358 # define REGEX_FREE_STACK(ptr) \
359 r_alloc_free (&failure_stack_ptr)
363 # ifdef REGEX_MALLOC
365 # define REGEX_ALLOCATE_STACK malloc
366 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
367 # define REGEX_FREE_STACK free
371 # define REGEX_ALLOCATE_STACK alloca
373 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
374 REGEX_REALLOCATE (source, osize, nsize)
375 /* No need to explicitly free anything. */
376 # define REGEX_FREE_STACK(arg)
382 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
383 `string1' or just past its end. This works if PTR is NULL, which is
384 a good thing. */
385 # define FIRST_STRING_P(ptr) \
386 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
388 /* (Re)Allocate N items of type T using malloc, or fail. */
389 # define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
390 # define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
391 # define RETALLOC_IF(addr, n, t) \
392 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
393 # define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
397 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
399 # undef MAX
400 # undef MIN
401 # define MAX(a, b) ((a) > (b) ? (a) : (b))
402 # define MIN(a, b) ((a) < (b) ? (a) : (b))
405 # define false 0
406 # define true 1
409 reg_syntax_t syntax,
425 #ifdef MBS_SUPPORT
427 reg_syntax_t syntax,
446 #endif
447 \f
448 /* These are the command codes that appear in compiled regular
449 expressions. Some opcodes are followed by argument bytes. A
450 command code can specify any interpretation whatsoever for its
451 arguments. Zero bytes may appear in the compiled regular expression. */
454 {
457 /* Succeed right away--no more backtracking. */
458 succeed,
460 /* Followed by one byte giving n, then by n literal bytes. */
461 exactn,
463 # ifdef MBS_SUPPORT
464 /* Same as exactn, but contains binary data. */
465 exactn_bin,
466 # endif
468 /* Matches any (more or less) character. */
469 anychar,
471 /* Matches any one char belonging to specified set. First
472 following byte is number of bitmap bytes. Then come bytes
473 for a bitmap saying which chars are in. Bits in each byte
474 are ordered low-bit-first. A character is in the set if its
475 bit is 1. A character too large to have a bit in the map is
476 automatically not in the set. */
477 /* ifdef MBS_SUPPORT, following element is length of character
478 classes, length of collating symbols, length of equivalence
479 classes, length of character ranges, and length of characters.
480 Next, character class element, collating symbols elements,
481 equivalence class elements, range elements, and character
482 elements follow.
483 See regex_compile function. */
484 charset,
486 /* Same parameters as charset, but match any character that is
487 not one of those specified. */
488 charset_not,
490 /* Start remembering the text that is matched, for storing in a
491 register. Followed by one byte with the register number, in
492 the range 0 to one less than the pattern buffer's re_nsub
493 field. Then followed by one byte with the number of groups
494 inner to this one. (This last has to be part of the
495 start_memory only because we need it in the on_failure_jump
496 of re_match_2.) */
497 start_memory,
499 /* Stop remembering the text that is matched and store it in a
500 memory register. Followed by one byte with the register
501 number, in the range 0 to one less than `re_nsub' in the
502 pattern buffer, and one byte with the number of inner groups,
503 just like `start_memory'. (We need the number of inner
504 groups here because we don't have any easy way of finding the
505 corresponding start_memory when we're at a stop_memory.) */
506 stop_memory,
508 /* Match a duplicate of something remembered. Followed by one
509 byte containing the register number. */
510 duplicate,
512 /* Fail unless at beginning of line. */
513 begline,
515 /* Fail unless at end of line. */
516 endline,
518 /* Succeeds if at beginning of buffer (if emacs) or at beginning
519 of string to be matched (if not). */
520 begbuf,
522 /* Analogously, for end of buffer/string. */
523 endbuf,
525 /* Followed by two byte relative address to which to jump. */
526 jump,
528 /* Same as jump, but marks the end of an alternative. */
529 jump_past_alt,
531 /* Followed by two-byte relative address of place to resume at
532 in case of failure. */
533 /* ifdef MBS_SUPPORT, the size of address is 1. */
534 on_failure_jump,
536 /* Like on_failure_jump, but pushes a placeholder instead of the
537 current string position when executed. */
538 on_failure_keep_string_jump,
540 /* Throw away latest failure point and then jump to following
541 two-byte relative address. */
542 /* ifdef MBS_SUPPORT, the size of address is 1. */
543 pop_failure_jump,
545 /* Change to pop_failure_jump if know won't have to backtrack to
546 match; otherwise change to jump. This is used to jump
547 back to the beginning of a repeat. If what follows this jump
548 clearly won't match what the repeat does, such that we can be
549 sure that there is no use backtracking out of repetitions
550 already matched, then we change it to a pop_failure_jump.
551 Followed by two-byte address. */
552 /* ifdef MBS_SUPPORT, the size of address is 1. */
553 maybe_pop_jump,
555 /* Jump to following two-byte address, and push a dummy failure
556 point. This failure point will be thrown away if an attempt
557 is made to use it for a failure. A `+' construct makes this
558 before the first repeat. Also used as an intermediary kind
559 of jump when compiling an alternative. */
560 /* ifdef MBS_SUPPORT, the size of address is 1. */
561 dummy_failure_jump,
563 /* Push a dummy failure point and continue. Used at the end of
564 alternatives. */
565 push_dummy_failure,
567 /* Followed by two-byte relative address and two-byte number n.
568 After matching N times, jump to the address upon failure. */
569 /* ifdef MBS_SUPPORT, the size of address is 1. */
570 succeed_n,
572 /* Followed by two-byte relative address, and two-byte number n.
573 Jump to the address N times, then fail. */
574 /* ifdef MBS_SUPPORT, the size of address is 1. */
575 jump_n,
577 /* Set the following two-byte relative address to the
578 subsequent two-byte number. The address *includes* the two
579 bytes of number. */
580 /* ifdef MBS_SUPPORT, the size of address is 1. */
581 set_number_at,
590 notwordbound /* Succeeds if not at a word boundary. */
592 # ifdef emacs
597 /* Matches any character whose syntax is specified. Followed by
598 a byte which contains a syntax code, e.g., Sword. */
599 syntaxspec,
601 /* Matches any character whose syntax is not that specified. */
602 notsyntaxspec
606 \f
608 #ifdef BYTE
609 # define CHAR_T char
610 # define UCHAR_T unsigned char
611 # define COMPILED_BUFFER_VAR bufp->buffer
612 # define OFFSET_ADDRESS_SIZE 2
613 # define PREFIX(name) byte_##name
614 # define ARG_PREFIX(name) name
615 # define PUT_CHAR(c) putchar (c)
616 #else
617 # ifdef WCHAR
618 # define CHAR_T wchar_t
619 # define UCHAR_T wchar_t
620 # define COMPILED_BUFFER_VAR wc_buffer
622 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
623 # define PREFIX(name) wcs_##name
624 # define ARG_PREFIX(name) c##name
625 /* Should we use wide stream?? */
627 # define TRUE 1
628 # define FALSE 0
629 # else
630 # ifdef MBS_SUPPORT
631 # define WCHAR
632 # define INSIDE_RECURSION
634 # undef INSIDE_RECURSION
635 # endif
636 # define BYTE
637 # define INSIDE_RECURSION
639 # undef INSIDE_RECURSION
640 # endif
641 #endif
643 #ifdef INSIDE_RECURSION
644 /* Common operations on the compiled pattern. */
646 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
647 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
649 # ifdef WCHAR
650 # define STORE_NUMBER(destination, number) \
651 do { \
652 *(destination) = (UCHAR_T)(number); \
653 } while (0)
655 # define STORE_NUMBER(destination, number) \
656 do { \
657 (destination)[0] = (number) & 0377; \
658 (destination)[1] = (number) >> 8; \
659 } while (0)
662 /* Same as STORE_NUMBER, except increment DESTINATION to
663 the byte after where the number is stored. Therefore, DESTINATION
664 must be an lvalue. */
665 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
667 # define STORE_NUMBER_AND_INCR(destination, number) \
668 do { \
669 STORE_NUMBER (destination, number); \
670 (destination) += OFFSET_ADDRESS_SIZE; \
671 } while (0)
673 /* Put into DESTINATION a number stored in two contiguous bytes starting
674 at SOURCE. */
675 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
677 # ifdef WCHAR
678 # define EXTRACT_NUMBER(destination, source) \
679 do { \
680 (destination) = *(source); \
681 } while (0)
683 # define EXTRACT_NUMBER(destination, source) \
684 do { \
685 (destination) = *(source) & 0377; \
686 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
687 } while (0)
688 # endif
690 # ifdef DEBUG
692 static void
694 {
695 # ifdef WCHAR
701 # endif
702 }
705 # undef EXTRACT_NUMBER
706 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
711 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
712 SOURCE must be an lvalue. */
714 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
715 do { \
716 EXTRACT_NUMBER (destination, source); \
717 (source) += OFFSET_ADDRESS_SIZE; \
718 } while (0)
720 # ifdef DEBUG
723 static void
725 {
728 }
730 # ifndef EXTRACT_MACROS
731 # undef EXTRACT_NUMBER_AND_INCR
732 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
733 PREFIX(extract_number_and_incr) (&dest, &src)
738 \f
740 /* If DEBUG is defined, Regex prints many voluminous messages about what
741 it is doing (if the variable `debug' is nonzero). If linked with the
742 main program in `iregex.c', you can enter patterns and strings
743 interactively. And if linked with the main program in `main.c' and
744 the other test files, you can run the already-written tests. */
746 # ifdef DEBUG
748 # ifndef DEFINED_ONCE
750 /* We use standard I/O for debugging. */
751 # include <stdio.h>
753 /* It is useful to test things that ``must'' be true when debugging. */
754 # include <assert.h>
758 # define DEBUG_STATEMENT(e) e
759 # define DEBUG_PRINT1(x) if (debug) printf (x)
760 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
761 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
762 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
765 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
766 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
767 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
768 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
771 /* Print the fastmap in human-readable form. */
773 # ifndef DEFINED_ONCE
774 void
776 {
781 {
783 {
787 {
789 i++;
790 }
792 {
795 }
796 }
797 }
799 }
803 /* Print a compiled pattern string in human-readable form, starting at
804 the START pointer into it and ending just before the pointer END. */
806 void
808 {
815 {
818 }
820 /* Loop over pattern commands. */
822 {
823 # ifdef _LIBC
825 # else
827 # endif
830 {
838 do
839 {
842 }
846 # ifdef MBS_SUPPORT
850 do
851 {
853 }
878 {
879 # ifdef WCHAR
890 {
896 }
899 {
905 }
908 {
912 }
917 # else
929 {
930 /* Are we starting a range? */
932 {
935 }
936 /* Have we broken a range? */
938 {
941 }
947 }
956 }
969 # ifdef _LIBC
971 # else
973 # endif
978 # ifdef _LIBC
980 # else
983 # endif
988 # ifdef _LIBC
990 # else
992 # endif
1001 # ifdef _LIBC
1003 # else
1005 # endif
1010 # ifdef _LIBC
1012 # else
1014 # endif
1019 # ifdef _LIBC
1021 # else
1023 # endif
1028 # ifdef _LIBC
1030 # else
1032 # endif
1039 # ifdef _LIBC
1041 # else
1044 # endif
1058 # ifdef _LIBC
1060 # else
1063 # endif
1082 # ifdef emacs
1126 }
1129 }
1131 # ifdef _LIBC
1133 # else
1135 # endif
1136 }
1139 void
1141 {
1150 {
1153 }
1155 # ifdef _LIBC
1157 # else
1159 # endif
1167 /* Perhaps we should print the translate table? */
1168 }
1171 void
1174 {
1179 else
1180 {
1184 {
1189 }
1193 {
1196 {
1199 }
1200 }
1201 }
1202 }
1204 # ifndef DEFINED_ONCE
1205 void
1207 {
1209 }
1210 # endif
1214 # ifndef DEFINED_ONCE
1215 # undef assert
1216 # define assert(e)
1218 # define DEBUG_STATEMENT(e)
1219 # define DEBUG_PRINT1(x)
1220 # define DEBUG_PRINT2(x1, x2)
1221 # define DEBUG_PRINT3(x1, x2, x3)
1222 # define DEBUG_PRINT4(x1, x2, x3, x4)
1224 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1225 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1229 \f
1231 # ifdef WCHAR
1232 /* This convert a multibyte string to a wide character string.
1233 And write their correspondances to offset_buffer(see below)
1234 and write whether each wchar_t is binary data to is_binary.
1235 This assume invalid multibyte sequences as binary data.
1236 We assume offset_buffer and is_binary is already allocated
1237 enough space. */
1242 static size_t
1245 /* It hold correspondances between src(char string) and
1246 dest(wchar_t string) for optimization.
1247 e.g. src = "xxxyzz"
1248 dest = {'X', 'Y', 'Z'}
1249 (each "xxx", "y" and "zz" represent one multibyte character
1250 corresponding to 'X', 'Y' and 'Z'.)
1251 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1252 = {0, 3, 4, 6}
1253 */
1254 {
1264 /* Initialize the conversion state. */
1270 {
1271 #ifdef _LIBC
1273 #else
1275 #endif
1278 /* failed to convert. maybe src contains binary data.
1279 So we consume 1 byte manualy. */
1280 {
1284 }
1285 else
1287 /* In sjis encoding, we use yen sign as escape character in
1288 place of reverse solidus. So we convert 0x5c(yen sign in
1289 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1290 solidus in UCS2). */
1295 }
1297 /* Fill remain of the buffer with sentinel. */
1302 }
1308 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1309 also be assigned to arbitrarily: each pattern buffer stores its own
1310 syntax, so it can be changed between regex compilations. */
1311 /* This has no initializer because initialized variables in Emacs
1312 become read-only after dumping. */
1313 reg_syntax_t re_syntax_options;
1316 /* Specify the precise syntax of regexps for compilation. This provides
1317 for compatibility for various utilities which historically have
1318 different, incompatible syntaxes.
1320 The argument SYNTAX is a bit mask comprised of the various bits
1321 defined in regex.h. We return the old syntax. */
1323 reg_syntax_t
1325 {
1329 # ifdef DEBUG
1336 }
1337 # ifdef _LIBC
1339 # endif
1340 \f
1341 /* This table gives an error message for each of the error codes listed
1342 in regex.h. Obviously the order here has to be same as there.
1343 POSIX doesn't require that we do anything for REG_NOERROR,
1344 but why not be nice? */
1347 {
1365 };
1366 \f
1369 #ifndef DEFINED_ONCE
1370 /* Avoiding alloca during matching, to placate r_alloc. */
1372 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1373 searching and matching functions should not call alloca. On some
1374 systems, alloca is implemented in terms of malloc, and if we're
1375 using the relocating allocator routines, then malloc could cause a
1376 relocation, which might (if the strings being searched are in the
1377 ralloc heap) shift the data out from underneath the regexp
1378 routines.
1380 Here's another reason to avoid allocation: Emacs
1381 processes input from X in a signal handler; processing X input may
1382 call malloc; if input arrives while a matching routine is calling
1383 malloc, then we're scrod. But Emacs can't just block input while
1384 calling matching routines; then we don't notice interrupts when
1385 they come in. So, Emacs blocks input around all regexp calls
1386 except the matching calls, which it leaves unprotected, in the
1387 faith that they will not malloc. */
1389 /* Normally, this is fine. */
1390 # define MATCH_MAY_ALLOCATE
1392 /* When using GNU C, we are not REALLY using the C alloca, no matter
1393 what config.h may say. So don't take precautions for it. */
1394 # ifdef __GNUC__
1395 # undef C_ALLOCA
1396 # endif
1398 /* The match routines may not allocate if (1) they would do it with malloc
1399 and (2) it's not safe for them to use malloc.
1400 Note that if REL_ALLOC is defined, matching would not use malloc for the
1401 failure stack, but we would still use it for the register vectors;
1402 so REL_ALLOC should not affect this. */
1403 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1404 # undef MATCH_MAY_ALLOCATE
1405 # endif
1407 \f
1408 #ifdef INSIDE_RECURSION
1409 /* Failure stack declarations and macros; both re_compile_fastmap and
1410 re_match_2 use a failure stack. These have to be macros because of
1411 REGEX_ALLOCATE_STACK. */
1414 /* Number of failure points for which to initially allocate space
1415 when matching. If this number is exceeded, we allocate more
1416 space, so it is not a hard limit. */
1417 # ifndef INIT_FAILURE_ALLOC
1418 # define INIT_FAILURE_ALLOC 5
1419 # endif
1421 /* Roughly the maximum number of failure points on the stack. Would be
1422 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1423 This is a variable only so users of regex can assign to it; we never
1424 change it ourselves. */
1426 # ifdef INT_IS_16BIT
1428 # ifndef DEFINED_ONCE
1429 # if defined MATCH_MAY_ALLOCATE
1430 /* 4400 was enough to cause a crash on Alpha OSF/1,
1431 whose default stack limit is 2mb. */
1433 # else
1435 # endif
1436 # endif
1439 {
1442 };
1447 {
1455 # ifndef DEFINED_ONCE
1456 # if defined MATCH_MAY_ALLOCATE
1457 /* 4400 was enough to cause a crash on Alpha OSF/1,
1458 whose default stack limit is 2mb. */
1460 # else
1462 # endif
1463 # endif
1466 {
1469 };
1474 {
1482 # ifndef DEFINED_ONCE
1483 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1484 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1485 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1486 # endif
1489 /* Define macros to initialize and free the failure stack.
1490 Do `return -2' if the alloc fails. */
1492 # ifdef MATCH_MAY_ALLOCATE
1493 # define INIT_FAIL_STACK() \
1494 do { \
1495 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1496 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1497 \
1498 if (fail_stack.stack == NULL) \
1499 return -2; \
1500 \
1501 fail_stack.size = INIT_FAILURE_ALLOC; \
1502 fail_stack.avail = 0; \
1503 } while (0)
1505 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1506 # else
1507 # define INIT_FAIL_STACK() \
1508 do { \
1509 fail_stack.avail = 0; \
1510 } while (0)
1512 # define RESET_FAIL_STACK()
1513 # endif
1516 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1518 Return 1 if succeeds, and 0 if either ran out of memory
1519 allocating space for it or it was already too large.
1521 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1523 # define DOUBLE_FAIL_STACK(fail_stack) \
1524 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1525 ? 0 \
1526 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1527 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1528 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1529 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1530 \
1531 (fail_stack).stack == NULL \
1532 ? 0 \
1533 : ((fail_stack).size <<= 1, \
1534 1)))
1537 /* Push pointer POINTER on FAIL_STACK.
1538 Return 1 if was able to do so and 0 if ran out of memory allocating
1539 space to do so. */
1540 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1541 ((FAIL_STACK_FULL () \
1542 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1543 ? 0 \
1544 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1545 1))
1547 /* Push a pointer value onto the failure stack.
1548 Assumes the variable `fail_stack'. Probably should only
1549 be called from within `PUSH_FAILURE_POINT'. */
1550 # define PUSH_FAILURE_POINTER(item) \
1551 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1553 /* This pushes an integer-valued item onto the failure stack.
1554 Assumes the variable `fail_stack'. Probably should only
1555 be called from within `PUSH_FAILURE_POINT'. */
1556 # define PUSH_FAILURE_INT(item) \
1557 fail_stack.stack[fail_stack.avail++].integer = (item)
1559 /* Push a fail_stack_elt_t value onto the failure stack.
1560 Assumes the variable `fail_stack'. Probably should only
1561 be called from within `PUSH_FAILURE_POINT'. */
1562 # define PUSH_FAILURE_ELT(item) \
1563 fail_stack.stack[fail_stack.avail++] = (item)
1565 /* These three POP... operations complement the three PUSH... operations.
1566 All assume that `fail_stack' is nonempty. */
1567 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1568 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1569 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1571 /* Used to omit pushing failure point id's when we're not debugging. */
1572 # ifdef DEBUG
1573 # define DEBUG_PUSH PUSH_FAILURE_INT
1574 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1575 # else
1576 # define DEBUG_PUSH(item)
1577 # define DEBUG_POP(item_addr)
1578 # endif
1581 /* Push the information about the state we will need
1582 if we ever fail back to it.
1584 Requires variables fail_stack, regstart, regend, reg_info, and
1585 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1586 be declared.
1588 Does `return FAILURE_CODE' if runs out of memory. */
1590 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1591 do { \
1592 char *destination; \
1593 /* Must be int, so when we don't save any registers, the arithmetic \
1595 /* Can't be int, since there is not a shred of a guarantee that int \
1596 is wide enough to hold a value of something to which pointer can \
1598 active_reg_t this_reg; \
1599 \
1600 DEBUG_STATEMENT (failure_id++); \
1601 DEBUG_STATEMENT (nfailure_points_pushed++); \
1605 \
1608 \
1610 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1611 { \
1612 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1613 return failure_code; \
1614 \
1616 (fail_stack).size); \
1618 } \
1619 \
1622 \
1623 if (1) \
1624 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1625 this_reg++) \
1626 { \
1628 DEBUG_STATEMENT (num_regs_pushed++); \
1629 \
1631 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1632 \
1634 PUSH_FAILURE_POINTER (regend[this_reg]); \
1635 \
1637 reg_info[this_reg].word.pointer); \
1639 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1642 MATCHED_SOMETHING (reg_info[this_reg])); \
1644 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1646 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1647 } \
1648 \
1650 PUSH_FAILURE_INT (lowest_active_reg); \
1651 \
1653 PUSH_FAILURE_INT (highest_active_reg); \
1654 \
1656 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1657 PUSH_FAILURE_POINTER (pattern_place); \
1658 \
1660 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1661 size2); \
1663 PUSH_FAILURE_POINTER (string_place); \
1664 \
1666 DEBUG_PUSH (failure_id); \
1667 } while (0)
1669 # ifndef DEFINED_ONCE
1670 /* This is the number of items that are pushed and popped on the stack
1671 for each register. */
1672 # define NUM_REG_ITEMS 3
1674 /* Individual items aside from the registers. */
1675 # ifdef DEBUG
1677 # else
1678 # define NUM_NONREG_ITEMS 4
1679 # endif
1681 /* We push at most this many items on the stack. */
1682 /* We used to use (num_regs - 1), which is the number of registers
1683 this regexp will save; but that was changed to 5
1684 to avoid stack overflow for a regexp with lots of parens. */
1685 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1687 /* We actually push this many items. */
1688 # define NUM_FAILURE_ITEMS \
1689 (((0 \
1690 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1691 * NUM_REG_ITEMS) \
1692 + NUM_NONREG_ITEMS)
1694 /* How many items can still be added to the stack without overflowing it. */
1695 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1699 /* Pops what PUSH_FAIL_STACK pushes.
1701 We restore into the parameters, all of which should be lvalues:
1702 STR -- the saved data position.
1703 PAT -- the saved pattern position.
1704 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1705 REGSTART, REGEND -- arrays of string positions.
1706 REG_INFO -- array of information about each subexpression.
1708 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1709 `pend', `string1', `size1', `string2', and `size2'. */
1710 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1711 { \
1712 DEBUG_STATEMENT (unsigned failure_id;) \
1713 active_reg_t this_reg; \
1714 const UCHAR_T *string_temp; \
1715 \
1716 assert (!FAIL_STACK_EMPTY ()); \
1717 \
1722 \
1723 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1724 \
1725 DEBUG_POP (&failure_id); \
1727 \
1728 /* If the saved string location is NULL, it came from an \
1729 on_failure_keep_string_jump opcode, and we want to throw away the \
1731 string_temp = POP_FAILURE_POINTER (); \
1732 if (string_temp != NULL) \
1733 str = (const CHAR_T *) string_temp; \
1734 \
1736 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1738 \
1739 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1741 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1742 \
1744 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1746 \
1747 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1749 \
1750 if (1) \
1751 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1752 { \
1754 \
1755 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1757 reg_info[this_reg].word.pointer); \
1758 \
1759 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1761 \
1762 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1764 } \
1765 else \
1766 { \
1767 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1768 { \
1769 reg_info[this_reg].word.integer = 0; \
1770 regend[this_reg] = 0; \
1771 regstart[this_reg] = 0; \
1772 } \
1773 highest_active_reg = high_reg; \
1774 } \
1775 \
1776 set_regs_matched_done = 0; \
1777 DEBUG_STATEMENT (nfailure_points_popped++); \
1779 \f
1780 /* Structure for per-register (a.k.a. per-group) information.
1781 Other register information, such as the
1782 starting and ending positions (which are addresses), and the list of
1783 inner groups (which is a bits list) are maintained in separate
1784 variables.
1786 We are making a (strictly speaking) nonportable assumption here: that
1787 the compiler will pack our bit fields into something that fits into
1788 the type of `word', i.e., is something that fits into one item on the
1789 failure stack. */
1792 /* Declarations and macros for re_match_2. */
1795 {
1797 struct
1798 {
1799 /* This field is one if this group can match the empty string,
1800 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1801 # define MATCH_NULL_UNSET_VALUE 3
1809 # ifndef DEFINED_ONCE
1810 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1811 # define IS_ACTIVE(R) ((R).bits.is_active)
1812 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1813 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1816 /* Call this when have matched a real character; it sets `matched' flags
1817 for the subexpressions which we are currently inside. Also records
1818 that those subexprs have matched. */
1819 # define SET_REGS_MATCHED() \
1820 do \
1821 { \
1822 if (!set_regs_matched_done) \
1823 { \
1824 active_reg_t r; \
1825 set_regs_matched_done = 1; \
1826 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1827 { \
1828 MATCHED_SOMETHING (reg_info[r]) \
1829 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1830 = 1; \
1831 } \
1832 } \
1833 } \
1834 while (0)
1837 /* Registers are set to a sentinel when they haven't yet matched. */
1839 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1840 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1842 /* Subroutine declarations and macros for regex_compile. */
1852 reg_syntax_t syntax);
1855 reg_syntax_t syntax);
1856 # ifdef WCHAR
1861 reg_syntax_t syntax,
1870 reg_syntax_t syntax,
1874 /* Fetch the next character in the uncompiled pattern---translating it
1875 if necessary. Also cast from a signed character in the constant
1876 string passed to us by the user to an unsigned char that we can use
1877 as an array index (in, e.g., `translate'). */
1878 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1879 because it is impossible to allocate 4GB array for some encodings
1880 which have 4 byte character_set like UCS4. */
1881 # ifndef PATFETCH
1882 # ifdef WCHAR
1883 # define PATFETCH(c) \
1884 do {if (p == pend) return REG_EEND; \
1885 c = (UCHAR_T) *p++; \
1886 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1887 } while (0)
1889 # define PATFETCH(c) \
1890 do {if (p == pend) return REG_EEND; \
1891 c = (unsigned char) *p++; \
1892 if (translate) c = (unsigned char) translate[c]; \
1893 } while (0)
1895 # endif
1897 /* Fetch the next character in the uncompiled pattern, with no
1898 translation. */
1899 # define PATFETCH_RAW(c) \
1900 do {if (p == pend) return REG_EEND; \
1901 c = (UCHAR_T) *p++; \
1902 } while (0)
1904 /* Go backwards one character in the pattern. */
1905 # define PATUNFETCH p--
1908 /* If `translate' is non-null, return translate[D], else just D. We
1909 cast the subscript to translate because some data is declared as
1910 `char *', to avoid warnings when a string constant is passed. But
1911 when we use a character as a subscript we must make it unsigned. */
1912 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1913 because it is impossible to allocate 4GB array for some encodings
1914 which have 4 byte character_set like UCS4. */
1916 # ifndef TRANSLATE
1917 # ifdef WCHAR
1918 # define TRANSLATE(d) \
1919 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1920 ? (char) translate[(unsigned char) (d)] : (d))
1922 # define TRANSLATE(d) \
1923 (translate ? (char) translate[(unsigned char) (d)] : (char) (d))
1925 # endif
1928 /* Macros for outputting the compiled pattern into `buffer'. */
1930 /* If the buffer isn't allocated when it comes in, use this. */
1931 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
1933 /* Make sure we have at least N more bytes of space in buffer. */
1934 # ifdef WCHAR
1935 # define GET_BUFFER_SPACE(n) \
1936 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
1937 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
1938 EXTEND_BUFFER ()
1940 # define GET_BUFFER_SPACE(n) \
1941 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1942 EXTEND_BUFFER ()
1945 /* Make sure we have one more byte of buffer space and then add C to it. */
1946 # define BUF_PUSH(c) \
1947 do { \
1948 GET_BUFFER_SPACE (1); \
1949 *b++ = (UCHAR_T) (c); \
1950 } while (0)
1953 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1954 # define BUF_PUSH_2(c1, c2) \
1955 do { \
1956 GET_BUFFER_SPACE (2); \
1957 *b++ = (UCHAR_T) (c1); \
1958 *b++ = (UCHAR_T) (c2); \
1959 } while (0)
1962 /* As with BUF_PUSH_2, except for three bytes. */
1963 # define BUF_PUSH_3(c1, c2, c3) \
1964 do { \
1965 GET_BUFFER_SPACE (3); \
1966 *b++ = (UCHAR_T) (c1); \
1967 *b++ = (UCHAR_T) (c2); \
1968 *b++ = (UCHAR_T) (c3); \
1969 } while (0)
1971 /* Store a jump with opcode OP at LOC to location TO. We store a
1972 relative address offset by the three bytes the jump itself occupies. */
1973 # define STORE_JUMP(op, loc, to) \
1974 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
1976 /* Likewise, for a two-argument jump. */
1977 # define STORE_JUMP2(op, loc, to, arg) \
1978 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
1980 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1981 # define INSERT_JUMP(op, loc, to) \
1982 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
1984 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1985 # define INSERT_JUMP2(op, loc, to, arg) \
1986 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
1987 arg, b)
1989 /* This is not an arbitrary limit: the arguments which represent offsets
1990 into the pattern are two bytes long. So if 2^16 bytes turns out to
1991 be too small, many things would have to change. */
1992 /* Any other compiler which, like MSC, has allocation limit below 2^16
1993 bytes will have to use approach similar to what was done below for
1994 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1995 reallocating to 0 bytes. Such thing is not going to work too well.
1996 You have been warned!! */
1997 # ifndef DEFINED_ONCE
1998 # if defined _MSC_VER && !defined WIN32
1999 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2000 The REALLOC define eliminates a flurry of conversion warnings,
2001 but is not required. */
2002 # define MAX_BUF_SIZE 65500L
2003 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2004 # else
2005 # define MAX_BUF_SIZE (1L << 16)
2006 # define REALLOC(p,s) realloc ((p), (s))
2007 # endif
2009 /* Extend the buffer by twice its current size via realloc and
2010 reset the pointers that pointed into the old block to point to the
2011 correct places in the new one. If extending the buffer results in it
2012 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2013 # if __BOUNDED_POINTERS__
2014 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2015 # define MOVE_BUFFER_POINTER(P) \
2016 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2017 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2018 else \
2019 { \
2020 SET_HIGH_BOUND (b); \
2021 SET_HIGH_BOUND (begalt); \
2022 if (fixup_alt_jump) \
2023 SET_HIGH_BOUND (fixup_alt_jump); \
2024 if (laststart) \
2025 SET_HIGH_BOUND (laststart); \
2026 if (pending_exact) \
2027 SET_HIGH_BOUND (pending_exact); \
2028 }
2029 # else
2030 # define MOVE_BUFFER_POINTER(P) (P) += incr
2031 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2032 # endif
2035 # ifdef WCHAR
2036 # define EXTEND_BUFFER() \
2037 do { \
2038 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2039 int wchar_count; \
2040 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2041 return REG_ESIZE; \
2042 bufp->allocated <<= 1; \
2043 if (bufp->allocated > MAX_BUF_SIZE) \
2044 bufp->allocated = MAX_BUF_SIZE; \
2046 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2047 if (wchar_count == 0) wchar_count = 1; \
2049 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2050 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2051 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2052 if (COMPILED_BUFFER_VAR == NULL) \
2053 return REG_ESPACE; \
2055 if (old_buffer != COMPILED_BUFFER_VAR) \
2056 { \
2057 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2058 MOVE_BUFFER_POINTER (b); \
2059 MOVE_BUFFER_POINTER (begalt); \
2060 if (fixup_alt_jump) \
2061 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2062 if (laststart) \
2063 MOVE_BUFFER_POINTER (laststart); \
2064 if (pending_exact) \
2065 MOVE_BUFFER_POINTER (pending_exact); \
2066 } \
2067 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2068 } while (0)
2070 # define EXTEND_BUFFER() \
2071 do { \
2072 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2073 if (bufp->allocated == MAX_BUF_SIZE) \
2074 return REG_ESIZE; \
2075 bufp->allocated <<= 1; \
2076 if (bufp->allocated > MAX_BUF_SIZE) \
2077 bufp->allocated = MAX_BUF_SIZE; \
2078 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2079 bufp->allocated); \
2080 if (COMPILED_BUFFER_VAR == NULL) \
2081 return REG_ESPACE; \
2083 if (old_buffer != COMPILED_BUFFER_VAR) \
2084 { \
2085 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2086 MOVE_BUFFER_POINTER (b); \
2087 MOVE_BUFFER_POINTER (begalt); \
2088 if (fixup_alt_jump) \
2089 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2090 if (laststart) \
2091 MOVE_BUFFER_POINTER (laststart); \
2092 if (pending_exact) \
2093 MOVE_BUFFER_POINTER (pending_exact); \
2094 } \
2095 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2096 } while (0)
2099 # ifndef DEFINED_ONCE
2100 /* Since we have one byte reserved for the register number argument to
2101 {start,stop}_memory, the maximum number of groups we can report
2102 things about is what fits in that byte. */
2103 # define MAX_REGNUM 255
2105 /* But patterns can have more than `MAX_REGNUM' registers. We just
2106 ignore the excess. */
2110 /* Macros for the compile stack. */
2112 /* Since offsets can go either forwards or backwards, this type needs to
2113 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2114 /* int may be not enough when sizeof(int) == 2. */
2118 {
2119 pattern_offset_t begalt_offset;
2120 pattern_offset_t fixup_alt_jump;
2121 pattern_offset_t inner_group_offset;
2122 pattern_offset_t laststart_offset;
2123 regnum_t regnum;
2128 {
2135 # define INIT_COMPILE_STACK_SIZE 32
2137 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2138 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2140 /* The next available element. */
2141 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2145 /* Set the bit for character C in a list. */
2146 # ifndef DEFINED_ONCE
2147 # define SET_LIST_BIT(c) \
2148 (b[((unsigned char) (c)) / BYTEWIDTH] \
2149 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2152 /* Get the next unsigned number in the uncompiled pattern. */
2153 # define GET_UNSIGNED_NUMBER(num) \
2154 { \
2155 while (p != pend) \
2156 { \
2157 PATFETCH (c); \
2159 break; \
2160 if (num <= RE_DUP_MAX) \
2161 { \
2162 if (num < 0) \
2163 num = 0; \
2165 } \
2166 } \
2167 }
2169 # ifndef DEFINED_ONCE
2170 # if defined _LIBC || WIDE_CHAR_SUPPORT
2171 /* The GNU C library provides support for user-defined character classes
2172 and the functions from ISO C amendement 1. */
2173 # ifdef CHARCLASS_NAME_MAX
2174 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2175 # else
2176 /* This shouldn't happen but some implementation might still have this
2177 problem. Use a reasonable default value. */
2178 # define CHAR_CLASS_MAX_LENGTH 256
2179 # endif
2181 # ifdef _LIBC
2182 # define IS_CHAR_CLASS(string) __wctype (string)
2183 # else
2184 # define IS_CHAR_CLASS(string) wctype (string)
2185 # endif
2186 # else
2189 # define IS_CHAR_CLASS(string) \
2196 # endif
2198 \f
2199 # ifndef MATCH_MAY_ALLOCATE
2201 /* If we cannot allocate large objects within re_match_2_internal,
2202 we make the fail stack and register vectors global.
2203 The fail stack, we grow to the maximum size when a regexp
2204 is compiled.
2205 The register vectors, we adjust in size each time we
2206 compile a regexp, according to the number of registers it needs. */
2210 /* Size with which the following vectors are currently allocated.
2211 That is so we can make them bigger as needed,
2212 but never make them smaller. */
2213 # ifdef DEFINED_ONCE
2225 /* Make the register vectors big enough for NUM_REGS registers,
2226 but don't make them smaller. */
2228 static void
2230 {
2232 {
2244 }
2245 }
2248 \f
2249 # ifndef DEFINED_ONCE
2251 regnum_t regnum);
2254 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2255 Returns one of error codes defined in `regex.h', or zero for success.
2257 Assumes the `allocated' (and perhaps `buffer') and `translate'
2258 fields are set in BUFP on entry.
2260 If it succeeds, results are put in BUFP (if it returns an error, the
2261 contents of BUFP are undefined):
2262 `buffer' is the compiled pattern;
2263 `syntax' is set to SYNTAX;
2264 `used' is set to the length of the compiled pattern;
2265 `fastmap_accurate' is zero;
2266 `re_nsub' is the number of subexpressions in PATTERN;
2267 `not_bol' and `not_eol' are zero;
2269 The `fastmap' and `newline_anchor' fields are neither
2270 examined nor set. */
2272 /* Return, freeing storage we allocated. */
2273 # ifdef WCHAR
2274 # define FREE_STACK_RETURN(value) \
2275 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2276 # else
2277 # define FREE_STACK_RETURN(value) \
2278 return (free (compile_stack.stack), value)
2281 static reg_errcode_t
2285 {
2286 /* We fetch characters from PATTERN here. Even though PATTERN is
2287 `char *' (i.e., signed), we declare these variables as unsigned, so
2288 they can be reliably used as array indices. */
2291 #ifdef WCHAR
2292 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2295 /* offset buffer for optimization. See convert_mbs_to_wc. */
2297 /* It hold whether each wchar_t is binary data or not. */
2299 /* A flag whether exactn is handling binary data or not. */
2303 /* A random temporary spot in PATTERN. */
2306 /* Points to the end of the buffer, where we should append. */
2309 /* Keeps track of unclosed groups. */
2310 compile_stack_type compile_stack;
2312 /* Points to the current (ending) position in the pattern. */
2313 #ifdef WCHAR
2321 /* How to translate the characters in the pattern. */
2324 /* Address of the count-byte of the most recently inserted `exactn'
2325 command. This makes it possible to tell if a new exact-match
2326 character can be added to that command or if the character requires
2327 a new `exactn' command. */
2330 /* Address of start of the most recently finished expression.
2331 This tells, e.g., postfix * where to find the start of its
2332 operand. Reset at the beginning of groups and alternatives. */
2335 /* Address of beginning of regexp, or inside of last group. */
2338 /* Address of the place where a forward jump should go to the end of
2339 the containing expression. Each alternative of an `or' -- except the
2340 last -- ends with a forward jump of this sort. */
2343 /* Counts open-groups as they are encountered. Remembered for the
2344 matching close-group on the compile stack, so the same register
2345 number is put in the stop_memory as the start_memory. */
2348 #ifdef WCHAR
2349 /* Initialize the wchar_t PATTERN and offset_buffer. */
2354 {
2359 }
2364 {
2369 }
2370 #endif
2372 #ifdef DEBUG
2375 {
2381 }
2384 /* Initialize the compile stack. */
2387 {
2388 #ifdef WCHAR
2392 #endif
2394 }
2399 /* Initialize the pattern buffer. */
2404 /* Set `used' to zero, so that if we return an error, the pattern
2405 printer (for debugging) will think there's no pattern. We reset it
2406 at the end. */
2409 /* Always count groups, whether or not bufp->no_sub is set. */
2412 #if !defined emacs && !defined SYNTAX_TABLE
2413 /* Initialize the syntax table. */
2415 #endif
2418 {
2421 enough space. This loses if buffer's address is bogus, but
2422 that is the user's responsibility. */
2423 #ifdef WCHAR
2424 /* Free bufp->buffer and allocate an array for wchar_t pattern
2425 buffer. */
2428 UCHAR_T);
2429 #else
2432 }
2433 else
2436 UCHAR_T);
2437 }
2440 #ifdef WCHAR
2444 }
2445 #ifdef WCHAR
2446 else
2448 #endif
2452 /* Loop through the uncompiled pattern until we're at the end. */
2454 {
2458 {
2460 {
2463 /* If context independent, it's an operator. */
2465 /* Otherwise, depends on what's come before. */
2468 else
2470 }
2475 {
2477 p == pend
2478 /* If context independent, it's an operator. */
2480 /* Otherwise, depends on what's next. */
2483 else
2485 }
2494 handle_plus:
2496 /* If there is no previous pattern... */
2498 {
2503 }
2505 {
2506 /* Are we optimizing this jump? */
2509 /* 1 means zero (many) matches is allowed. */
2512 /* If there is a sequence of repetition chars, collapse it
2513 down to just one (the right one). We can't combine
2514 interval operators with these because of, e.g., `a{2}*',
2515 which should only match an even number of `a's. */
2518 {
2529 ;
2532 {
2537 {
2538 PATUNFETCH;
2539 PATUNFETCH;
2541 }
2544 }
2545 else
2546 {
2547 PATUNFETCH;
2549 }
2551 /* If we get here, we found another repeat character. */
2552 }
2554 /* Star, etc. applied to an empty pattern is equivalent
2555 to an empty pattern. */
2559 /* Now we know whether or not zero matches is allowed
2560 and also whether or not two or more matches is allowed. */
2563 end a backward relative jump from `b' to before the next
2564 jump we're going to put in below (which jumps from
2565 laststart to after this jump).
2567 But if we are at the `*' in the exact sequence `.*\n',
2568 insert an unconditional jump backwards to the .,
2569 instead of the beginning of the loop. This way we only
2570 push a failure point once, instead of every time
2571 through the loop. */
2574 /* Allocate the space for the jump. */
2577 /* We know we are not at the first character of the pattern,
2578 because laststart was nonzero. And we've already
2579 incremented `p', by the way, to be the character after
2580 the `*'. Do we have to do something analogous here
2581 for null bytes, because of RE_DOT_NOT_NULL? */
2583 && zero_times_ok
2589 }
2590 else
2591 /* Anything else. */
2595 /* We've added more stuff to the buffer. */
2597 }
2599 /* On failure, jump from laststart to b + 3, which will be the
2600 end of the buffer after this jump is inserted. */
2601 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2602 'b + 3'. */
2611 {
2612 /* At least one repetition is required, so insert a
2613 `dummy_failure_jump' before the initial
2614 `on_failure_jump' instruction of the loop. This
2615 effects a skip over that instruction the first time
2616 we hit that loop. */
2621 }
2622 }
2633 {
2635 #ifdef WCHAR
2637 #else
2639 #endif
2642 #ifdef WCHAR
2643 /* We assume a charset(_not) structure as a wchar_t array.
2644 charset[0] = (re_opcode_t) charset(_not)
2645 charset[1] = l (= length of char_classes)
2646 charset[2] = m (= length of collating_symbols)
2647 charset[3] = n (= length of equivalence_classes)
2648 charset[4] = o (= length of char_ranges)
2649 charset[5] = p (= length of chars)
2651 charset[6] = char_class (wctype_t)
2652 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2653 ...
2654 charset[l+5] = char_class (wctype_t)
2656 charset[l+6] = collating_symbol (wchar_t)
2657 ...
2658 charset[l+m+5] = collating_symbol (wchar_t)
2659 ifdef _LIBC we use the index if
2660 _NL_COLLATE_SYMB_EXTRAMB instead of
2661 wchar_t string.
2663 charset[l+m+6] = equivalence_classes (wchar_t)
2664 ...
2665 charset[l+m+n+5] = equivalence_classes (wchar_t)
2666 ifdef _LIBC we use the index in
2667 _NL_COLLATE_WEIGHT instead of
2668 wchar_t string.
2670 charset[l+m+n+6] = range_start
2671 charset[l+m+n+7] = range_end
2672 ...
2673 charset[l+m+n+2o+4] = range_start
2674 charset[l+m+n+2o+5] = range_end
2675 ifdef _LIBC we use the value looked up
2676 in _NL_COLLATE_COLLSEQ instead of
2677 wchar_t character.
2679 charset[l+m+n+2o+6] = char
2680 ...
2681 charset[l+m+n+2o+p+5] = char
2683 */
2685 /* We need at least 6 spaces: the opcode, the length of
2686 char_classes, the length of collating_symbols, the length of
2687 equivalence_classes, the length of char_ranges, the length of
2688 chars. */
2691 /* Save b as laststart. And We use laststart as the pointer
2692 to the first element of the charset here.
2693 In other words, laststart[i] indicates charset[i]. */
2696 /* We test `*p == '^' twice, instead of using an if
2697 statement, so we only need one BUF_PUSH. */
2700 p++;
2702 /* Push the length of char_classes, the length of
2703 collating_symbols, the length of equivalence_classes, the
2704 length of char_ranges and the length of chars. */
2708 /* Remember the first position in the bracket expression. */
2711 /* charset_not matches newline according to a syntax bit. */
2714 {
2717 }
2719 /* Read in characters and ranges, setting map bits. */
2721 {
2726 /* \ might escape characters inside [...] and [^...]. */
2728 {
2736 }
2738 /* Could be the end of the bracket expression. If it's
2739 not (i.e., when the bracket expression is `[]' so
2740 far), the ']' character bit gets set way below. */
2744 /* Look ahead to see if it's a range when the last thing
2745 was a character class. */
2749 /* Look ahead to see if it's a range when the last thing
2750 was a character: if this is a hyphen not at the
2751 beginning or the end of a list, then it's the range
2752 operator. */
2757 {
2758 reg_errcode_t ret;
2759 /* Allocate the space for range_start and range_end. */
2761 /* Update the pointer to indicate end of buffer. */
2767 }
2770 reg_errcode_t ret;
2772 /* Move past the `-'. */
2774 /* Allocate the space for range_start and range_end. */
2776 /* Update the pointer to indicate end of buffer. */
2779 laststart);
2782 }
2784 /* See if we're at the beginning of a possible character
2785 class. */
2793 /* If pattern is `[[:'. */
2797 {
2803 else
2804 /* This is in any case an invalid class name. */
2806 }
2809 /* If isn't a word bracketed by `[:' and `:]':
2810 undo the ending character, the letters, and leave
2811 the leading `:' and `[' (but store them as character). */
2813 {
2817 /* Query the character class as wctype_t. */
2822 /* Throw away the ] at the end of the character
2823 class. */
2828 /* Allocate the space for character class. */
2830 /* Update the pointer to indicate end of buffer. */
2832 /* Move data which follow character classes
2833 not to violate the data. */
2840 /* Store the character class. */
2842 /* Update length of char_classes */
2846 }
2847 else
2848 {
2849 c1++;
2851 PATUNFETCH;
2857 }
2858 }
2861 {
2864 # ifdef _LIBC
2867 # endif
2871 /* If pattern is `[[=' or '[[.'. */
2875 {
2881 else
2882 /* This is in any case an invalid class name. */
2884 }
2888 {
2890 /* If we have no collation data we use the default
2891 collation in which each character is in a class
2892 by itself. It also means that ASCII is the
2893 character set and therefore we cannot have character
2894 with more than one byte in the multibyte
2895 representation. */
2897 /* If not defined _LIBC, we push the name and
2898 `\0' for the sake of matching performance. */
2901 # ifdef _LIBC
2904 # endif
2905 {
2908 }
2909 # ifdef _LIBC
2910 else
2911 {
2918 /* This #include defines a local function! */
2919 # include <locale/weightwc.h>
2922 {
2923 /* We push the index for equivalence class. */
2928 _NL_COLLATE_TABLEWC);
2931 _NL_COLLATE_WEIGHTWC);
2934 _NL_COLLATE_EXTRAWC);
2937 _NL_COLLATE_INDIRECTWC);
2941 /* This is no valid character. */
2945 }
2947 {
2948 /* We push collation sequence value
2949 for collating symbol. */
2959 /* We have to convert the name to a single-byte
2960 string. This is possible since the names
2961 consist of ASCII characters and the internal
2962 representation is UCS4. */
2966 table_size =
2968 _NL_COLLATE_SYMB_HASH_SIZEMB);
2971 _NL_COLLATE_SYMB_TABLEMB);
2974 _NL_COLLATE_SYMB_EXTRAMB);
2976 /* Locate the character in the hashing table. */
2983 {
2984 /* First compare the hashing value. */
2990 {
2991 /* Yep, this is the entry. */
2995 }
2997 /* Next entry. */
2999 }
3002 {
3003 /* Compute the index of the byte sequence
3004 in the table. */
3006 /* Adjust for the alignment. */
3010 }
3012 {
3013 /* No valid character. Match it as a
3014 single byte character. */
3017 /* Update the length of characters */
3021 /* Throw away the ] at the end of the
3022 collating symbol. */
3024 /* exit from the switch block. */
3026 }
3027 else
3029 }
3031 }
3032 # endif
3033 /* Throw away the ] at the end of the equivalence
3034 class (or collating symbol). */
3037 /* Allocate the space for the equivalence class
3038 (or collating symbol) (and '\0' if needed). */
3040 /* Update the pointer to indicate end of buffer. */
3045 /* Calculate the offset of char_ranges,
3046 which is next to equivalence_classes. */
3049 /* Insert space. */
3052 /* Write the equivalence_class and \0. */
3056 /* Update the length of equivalence_classes. */
3059 }
3062 /* Calculate the offset of the equivalence_classes,
3063 which is next to collating_symbols. */
3065 /* Insert space and write the collationg_symbol
3066 and \0. */
3071 /* In re_match_2_internal if range_start < -1, we
3072 assume -range_start is the offset of the
3073 collating symbol which is specified as
3074 the character of the range start. So we assign
3075 -(laststart[1] + laststart[2] + 6) to
3076 range_start. */
3078 /* Update the length of collating_symbol. */
3081 }
3082 }
3083 else
3084 {
3085 c1++;
3087 PATUNFETCH;
3093 }
3094 }
3095 else
3096 {
3101 }
3102 }
3105 /* Ensure that we have enough space to push a charset: the
3106 opcode, the length count, and the bitset; 34 bytes in all. */
3111 /* We test `*p == '^' twice, instead of using an if
3112 statement, so we only need one BUF_PUSH. */
3115 p++;
3117 /* Remember the first position in the bracket expression. */
3120 /* Push the number of bytes in the bitmap. */
3123 /* Clear the whole map. */
3126 /* charset_not matches newline according to a syntax bit. */
3131 /* Read in characters and ranges, setting map bits. */
3133 {
3138 /* \ might escape characters inside [...] and [^...]. */
3140 {
3147 }
3149 /* Could be the end of the bracket expression. If it's
3150 not (i.e., when the bracket expression is `[]' so
3151 far), the ']' character bit gets set way below. */
3155 /* Look ahead to see if it's a range when the last thing
3156 was a character class. */
3160 /* Look ahead to see if it's a range when the last thing
3161 was a character: if this is a hyphen not at the
3162 beginning or the end of a list, then it's the range
3163 operator. */
3168 {
3169 reg_errcode_t ret
3174 }
3178 reg_errcode_t ret;
3180 /* Move past the `-'. */
3186 }
3188 /* See if we're at the beginning of a possible character
3189 class. */
3198 /* If pattern is `[[:'. */
3202 {
3208 else
3209 /* This is in any case an invalid class name. */
3211 }
3214 /* If isn't a word bracketed by `[:' and `:]':
3215 undo the ending character, the letters, and leave
3216 the leading `:' and `[' (but set bits for them). */
3218 {
3219 # if defined _LIBC || WIDE_CHAR_SUPPORT
3229 /* Throw away the ] at the end of the character
3230 class. */
3236 {
3237 # ifdef _LIBC
3240 # else
3243 # endif
3248 }
3251 # else
3269 /* Throw away the ] at the end of the character
3270 class. */
3276 {
3277 /* This was split into 3 if's to
3278 avoid an arbitrary limit in some compiler. */
3297 }
3300 }
3301 else
3302 {
3303 c1++;
3305 PATUNFETCH;
3310 }
3311 }
3313 {
3315 # ifdef _LIBC
3318 # endif
3323 /* If pattern is `[[='. */
3327 {
3333 else
3334 /* This is in any case an invalid class name. */
3336 }
3340 {
3341 /* If we have no collation data we use the default
3342 collation in which each character is in a class
3343 by itself. It also means that ASCII is the
3344 character set and therefore we cannot have character
3345 with more than one byte in the multibyte
3346 representation. */
3347 # ifdef _LIBC
3349 # endif
3350 {
3354 /* Throw away the ] at the end of the equivalence
3355 class. */
3358 /* Set the bit for the character. */
3360 }
3361 # ifdef _LIBC
3362 else
3363 {
3364 /* Try to match the byte sequence in `str' against
3365 those known to the collate implementation.
3366 First find out whether the bytes in `str' are
3367 actually from exactly one character. */
3376 /* This #include defines a local function! */
3377 # include <locale/weight.h>
3390 /* This is no valid character. */
3393 /* Throw away the ] at the end of the equivalence
3394 class. */
3397 /* Now we have to go throught the whole table
3398 and find all characters which have the same
3399 first level weight.
3401 XXX Note that this is not entirely correct.
3402 we would have to match multibyte sequences
3403 but this is not possible with the current
3404 implementation. */
3406 /* XXX This test would have to be changed if we
3407 would allow matching multibyte sequences. */
3409 {
3413 /* Test whether the lenghts match. */
3415 {
3416 /* They do. New compare the bytes of
3417 the weight. */
3426 /* They match. Mark the character as
3427 acceptable. */
3429 }
3430 }
3431 }
3432 # endif
3434 }
3435 else
3436 {
3437 c1++;
3439 PATUNFETCH;
3444 }
3445 }
3447 {
3449 # ifdef _LIBC
3452 # endif
3457 /* If pattern is `[[.'. */
3461 {
3467 else
3468 /* This is in any case an invalid class name. */
3470 }
3474 {
3475 /* If we have no collation data we use the default
3476 collation in which each character is the name
3477 for its own class which contains only the one
3478 character. It also means that ASCII is the
3479 character set and therefore we cannot have character
3480 with more than one byte in the multibyte
3481 representation. */
3482 # ifdef _LIBC
3484 # endif
3485 {
3489 /* Throw away the ] at the end of the equivalence
3490 class. */
3493 /* Set the bit for the character. */
3496 }
3497 # ifdef _LIBC
3498 else
3499 {
3500 /* Try to match the byte sequence in `str' against
3501 those known to the collate implementation.
3502 First find out whether the bytes in `str' are
3503 actually from exactly one character. */
3512 table_size =
3514 _NL_COLLATE_SYMB_HASH_SIZEMB);
3517 _NL_COLLATE_SYMB_TABLEMB);
3520 _NL_COLLATE_SYMB_EXTRAMB);
3522 /* Locate the character in the hashing table. */
3529 {
3530 /* First compare the hashing value. */
3537 {
3538 /* Yep, this is the entry. */
3542 }
3544 /* Next entry. */
3546 }
3549 /* This is no valid character. */
3552 /* Throw away the ] at the end of the equivalence
3553 class. */
3556 /* Now add the multibyte character(s) we found
3557 to the accept list.
3559 XXX Note that this is not entirely correct.
3560 we would have to match multibyte sequences
3561 but this is not possible with the current
3562 implementation. Also, we have to match
3563 collating symbols, which expand to more than
3564 one file, as a whole and not allow the
3565 individual bytes. */
3570 {
3573 }
3574 }
3575 # endif
3577 }
3578 else
3579 {
3580 c1++;
3582 PATUNFETCH;
3587 }
3588 }
3589 else
3590 {
3594 }
3595 }
3597 /* Discard any (non)matching list bytes that are all 0 at the
3598 end of the map. Decrease the map-length byte too. */
3603 }
3610 else
3617 else
3624 else
3631 else
3638 else
3645 /* Do not translate the character after the \, so that we can
3646 distinguish, e.g., \B from \b, even if we normally would
3647 translate, e.g., B to b. */
3651 {
3656 handle_open:
3658 regnum++;
3661 {
3663 compile_stack_elt_t);
3667 }
3669 /* These are the values to restore when we hit end of this
3670 group. They are all relative offsets, so that if the
3671 whole pattern moves because of realloc, they will still
3672 be valid. */
3674 COMPILE_STACK_TOP.fixup_alt_jump
3679 /* We will eventually replace the 0 with the number of
3680 groups inner to this one. But do not push a
3681 start_memory for groups beyond the last one we can
3682 represent in the compiled pattern. */
3684 {
3688 }
3695 /* If we've reached MAX_REGNUM groups, then this open
3696 won't actually generate any code, so we'll have to
3697 clear pending_exact explicitly. */
3706 {
3709 else
3711 }
3713 handle_close:
3716 alternative for a possible future
3717 `pop_failure_jump' to pop. See comments at
3718 `push_dummy_failure' in `re_match_2'. */
3721 /* We allocated space for this jump when we assigned
3722 to `fixup_alt_jump', in the `handle_alt' case below. */
3724 }
3726 /* See similar code for backslashed left paren above. */
3728 {
3731 else
3733 }
3735 /* Since we just checked for an empty stack above, this
3736 ``can't happen''. */
3738 {
3739 /* We don't just want to restore into `regnum', because
3740 later groups should continue to be numbered higher,
3741 as in `(ab)c(de)' -- the second group is #2. */
3742 regnum_t this_group_regnum;
3746 fixup_alt_jump
3752 /* If we've reached MAX_REGNUM groups, then this open
3753 won't actually generate any code, so we'll have to
3754 clear pending_exact explicitly. */
3757 /* We're at the end of the group, so now we know how many
3758 groups were inside this one. */
3760 {
3761 UCHAR_T *inner_group_loc
3767 }
3768 }
3775 handle_alt:
3779 /* Insert before the previous alternative a jump which
3780 jumps to this alternative if the former fails. */
3787 /* The alternative before this one has a jump after it
3788 which gets executed if it gets matched. Adjust that
3789 jump so it will jump to this alternative's analogous
3790 jump (put in below, which in turn will jump to the next
3791 (if any) alternative's such jump, etc.). The last such
3792 jump jumps to the correct final destination. A picture:
3793 _____ _____
3794 | | | |
3795 | v | v
3796 a | b | c
3798 If we are at `b', then fixup_alt_jump right now points to a
3799 three-byte space after `a'. We'll put in the jump, set
3800 fixup_alt_jump to right after `b', and leave behind three
3801 bytes which we'll fill in when we get to after `c'. */
3806 /* Mark and leave space for a jump after this alternative,
3807 to be filled in later either by next alternative or
3808 when know we're at the end of a series of alternatives. */
3819 /* If \{ is a literal. */
3821 /* If we're at `\{' and it's not the open-interval
3822 operator. */
3826 handle_interval:
3827 {
3828 /* If got here, then the syntax allows intervals. */
3830 /* At least (most) this many matches must be made. */
3833 /* Place in the uncompiled pattern (i.e., just after
3834 the '{') to go back to if the interval is invalid. */
3843 {
3847 }
3848 else
3849 /* Interval such as `{1}' => match exactly once. */
3856 {
3860 }
3865 /* If it's invalid to have no preceding re. */
3867 {
3873 else
3875 }
3877 /* We just parsed a valid interval. */
3882 /* If the upper bound is zero, don't want to succeed at
3883 all; jump from `laststart' to `b + 3', which will be
3884 the end of the buffer after we insert the jump. */
3885 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3886 instead of 'b + 3'. */
3888 {
3893 }
3895 /* Otherwise, we have a nontrivial interval. When
3896 we're all done, the pattern will look like:
3897 set_number_at <jump count> <upper bound>
3898 set_number_at <succeed_n count> <lower bound>
3899 succeed_n <after jump addr> <succeed_n count>
3900 <body of loop>
3901 jump_n <succeed_n addr> <jump count>
3902 (The upper bound and `jump_n' are omitted if
3903 `upper_bound' is 1, though.) */
3904 else
3906 more at the end of the loop. */
3912 /* Initialize lower bound of the `succeed_n', even
3913 though it will be set during matching by its
3914 attendant `set_number_at' (inserted next),
3915 because `re_compile_fastmap' needs to know.
3916 Jump to the `jump_n' we might insert below. */
3923 /* Code to initialize the lower bound. Insert
3924 before the `succeed_n'. The `5' is the last two
3925 bytes of this `set_number_at', plus 3 bytes of
3926 the following `succeed_n'. */
3927 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
3928 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
3929 of the following `succeed_n'. */
3936 append a backward jump to the `succeed_n'
3937 that starts this interval.
3939 When we've reached this during matching,
3940 we'll have matched the interval once, so
3941 jump back only `upper_bound - 1' times. */
3947 /* The location we want to set is the second
3948 parameter of the `jump_n'; that is `b-2' as
3949 an absolute address. `laststart' will be
3950 the `set_number_at' we're about to insert;
3951 `laststart+3' the number to set, the source
3952 for the relative address. But we are
3953 inserting into the middle of the pattern --
3954 so everything is getting moved up by 5.
3955 Conclusion: (b - 2) - (laststart + 3) + 5,
3956 i.e., b - laststart.
3958 We insert this at the beginning of the loop
3959 so that if we fail during matching, we'll
3960 reinitialize the bounds. */
3965 }
3966 }
3970 invalid_interval:
3973 unfetch_interval:
3974 /* Match the characters as literals. */
3979 else
3981 }
3983 #ifdef emacs
3984 /* There is no way to specify the before_dot and after_dot
3985 operators. rms says this is ok. --karl */
4066 /* Can't back reference to a subexpression if inside of it. */
4079 else
4083 normal_backslash:
4084 /* You might think it would be useful for \ to mean
4085 not to translate; but if we don't translate it
4086 it will never match anything. */
4089 }
4094 /* Expects the character in `c'. */
4095 normal_char:
4096 /* If no exactn currently being built. */
4098 #ifdef WCHAR
4099 /* If last exactn handle binary(or character) and
4100 new exactn handle character(or binary). */
4104 /* If last exactn not at current position. */
4107 /* We have only one byte following the exactn for the count. */
4110 /* If followed by a repetition operator. */
4119 {
4120 /* Start building a new exactn. */
4124 #ifdef WCHAR
4125 /* Is this exactn binary data or character? */
4129 else
4131 #else
4135 }
4144 /* Through the pattern now. */
4152 /* If we don't want backtracking, force success
4153 the first time we reach the end of the compiled pattern. */
4157 #ifdef WCHAR
4161 #endif
4164 /* We have succeeded; set the length of the buffer. */
4165 #ifdef WCHAR
4167 #else
4169 #endif
4171 #ifdef DEBUG
4173 {
4176 }
4179 #ifndef MATCH_MAY_ALLOCATE
4180 /* Initialize the failure stack to the largest possible stack. This
4181 isn't necessary unless we're trying to avoid calling alloca in
4182 the search and match routines. */
4183 {
4186 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4187 is strictly greater than re_max_failures, the largest possible stack
4188 is 2 * re_max_failures failure points. */
4190 {
4193 # ifdef emacs
4195 fail_stack.stack
4198 else
4199 fail_stack.stack
4205 fail_stack.stack
4208 else
4209 fail_stack.stack
4214 }
4217 }
4223 /* Subroutines for `regex_compile'. */
4225 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4226 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4228 static void
4230 {
4233 }
4236 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4237 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4239 static void
4241 {
4245 }
4248 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4249 for OP followed by two-byte integer parameter ARG. */
4250 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4252 static void
4254 {
4262 }
4265 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4266 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4268 static void
4271 {
4279 }
4282 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4283 after an alternative or a begin-subexpression. We assume there is at
4284 least one character before the ^. */
4286 static boolean
4288 reg_syntax_t syntax)
4289 {
4293 return
4294 /* After a subexpression? */
4296 /* After an alternative? */
4298 }
4301 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4302 at least one character after the $, i.e., `P < PEND'. */
4304 static boolean
4306 reg_syntax_t syntax)
4307 {
4312 return
4313 /* Before a subexpression? */
4316 /* Before an alternative? */
4319 }
4323 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4324 false if it's not. */
4326 static boolean
4328 {
4333 this_element--)
4338 }
4341 #ifdef INSIDE_RECURSION
4343 #ifdef WCHAR
4344 /* This insert space, which size is "num", into the pattern at "loc".
4345 "end" must point the end of the allocated buffer. */
4346 static void
4348 {
4354 }
4357 #ifdef WCHAR
4358 static reg_errcode_t
4362 {
4365 reg_errcode_t ret;
4366 # ifdef _LIBC
4369 # endif
4373 # ifdef _LIBC
4376 {
4378 _NL_COLLATE_COLLSEQWC);
4383 {
4384 /* range_start is a collating symbol. */
4386 /* Retreive the index and get collation sequence value. */
4389 }
4390 else
4395 /* Report an error if the range is empty and the syntax prohibits
4396 this. */
4400 /* Insert space to the end of the char_ranges. */
4405 }
4406 else
4407 # endif
4408 {
4410 range_start_char;
4412 /* Report an error if the range is empty and the syntax prohibits
4413 this. */
4417 /* Insert space to the end of the char_ranges. */
4422 }
4423 /* Have to increment the pointer into the pattern string, so the
4424 caller isn't still at the ending character. */
4428 }
4430 /* Read the ending character of a range (in a bracket expression) from the
4431 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4432 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4433 Then we set the translation of all bits between the starting and
4434 ending characters (inclusive) in the compiled pattern B.
4436 Return an error code.
4438 We use these short variable names so we can use the same macros as
4439 `regex_compile' itself. */
4441 static reg_errcode_t
4445 {
4448 reg_errcode_t ret;
4449 # if _LIBC
4453 # else
4455 # endif
4460 /* Have to increment the pointer into the pattern string, so the
4461 caller isn't still at the ending character. */
4464 /* Report an error if the range is empty and the syntax prohibits this. */
4467 # if _LIBC
4469 _NL_COLLATE_COLLSEQMB);
4474 {
4478 {
4481 }
4482 }
4483 # else
4484 /* Here we see why `this_char' has to be larger than an `unsigned
4485 char' -- we would otherwise go into an infinite loop, since all
4486 characters <= 0xff. */
4488 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4489 and some compilers cast it to int implicitly, so following for_loop
4490 may fall to (almost) infinite loop.
4491 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4492 To avoid this, we cast p[0] to unsigned int and truncate it. */
4496 {
4499 }
4500 # endif
4503 }
4505 \f
4506 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4507 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4508 characters can start a string that matches the pattern. This fastmap
4509 is used by re_search to skip quickly over impossible starting points.
4511 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4512 area as BUFP->fastmap.
4514 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4515 the pattern buffer.
4517 Returns 0 if we succeed, -2 if an internal error. */
4519 #ifdef WCHAR
4520 /* local function for re_compile_fastmap.
4521 truncate wchar_t character to char. */
4524 static unsigned char
4526 {
4531 # ifdef _LIBC
4533 # else
4535 # endif
4537 }
4540 static int
4542 {
4544 #ifdef MATCH_MAY_ALLOCATE
4546 #endif
4547 #ifndef REGEX_MALLOC
4549 #endif
4553 #ifdef WCHAR
4554 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4555 pattern to (char*) in regex_compile. */
4564 #ifdef REL_ALLOC
4565 /* This holds the pointer to the failure stack, when
4566 it is allocated relocatably. */
4568 #endif
4570 /* Assume that each path through the pattern can be null until
4571 proven otherwise. We set this false at the bottom of switch
4572 statement, to which we get only if a particular path doesn't
4573 match the empty string. */
4576 /* We aren't doing a `succeed_n' to begin with. */
4587 {
4589 {
4590 /* We have reached the (effective) end of pattern. */
4592 {
4595 /* Reset for next path. */
4601 }
4602 else
4604 }
4606 /* We should never be about to go beyond the end of the pattern. */
4610 {
4612 /* I guess the idea here is to simply not bother with a fastmap
4613 if a backreference is used, since it's too hard to figure out
4614 the fastmap for the corresponding group. Setting
4615 `can_be_null' stops `re_search_2' from using the fastmap, so
4616 that is all we do. */
4622 /* Following are the cases which match a character. These end
4623 with `break'. */
4625 #ifdef WCHAR
4634 #ifdef MBS_SUPPORT
4638 #endif
4640 #ifdef WCHAR
4641 /* It is hard to distinguish fastmap from (multi byte) characters
4642 which depends on current locale. */
4658 /* Chars beyond end of map must be allowed. */
4683 {
4686 /* `.' matches anything ... */
4690 /* ... except perhaps newline. */
4694 /* Return if we have already set `can_be_null'; if we have,
4695 then the fastmap is irrelevant. Something's wrong here. */
4699 /* Otherwise, have to check alternative paths. */
4701 }
4703 #ifdef emacs
4720 /* All cases after this match the empty string. These end with
4721 `continue'. */
4755 /* Jump backward implies we just went through the body of a
4756 loop and matched nothing. Opcode jumped to should be
4757 `on_failure_jump' or `succeed_n'. Just treat it like an
4758 ordinary jump. For a * loop, it has pushed its failure
4759 point already; if so, discard that as redundant. */
4764 p++;
4768 /* If what's on the stack is where we are now, pop it. */
4778 handle_on_failure_jump:
4781 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4782 end of the pattern. We don't want to push such a point,
4783 since when we restore it above, entering the switch will
4784 increment `p' past the end of the pattern. We don't need
4785 to push such a point since we obviously won't find any more
4786 fastmap entries beyond `pend'. Such a pattern can match
4787 the null string, though. */
4789 {
4791 {
4794 }
4795 }
4796 else
4800 {
4803 }
4809 /* Get to the number of times to succeed. */
4812 /* Increment p past the n for when k != 0. */
4815 {
4819 }
4838 /* Getting here means we have found the possible starting
4839 characters for one path of the pattern -- and that the empty
4840 string does not match. We need not follow this path further.
4841 Instead, look at the next alternative (remembered on the
4842 stack), or quit if no more. The test at the top of the loop
4843 does these things. */
4848 /* Set `can_be_null' for the last path (also the first path, if the
4849 pattern is empty). */
4852 done:
4855 }
4859 int
4861 {
4862 # ifdef MBS_SUPPORT
4865 else
4866 # endif
4869 #ifdef _LIBC
4871 #endif
4872 \f
4874 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4875 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4876 this memory for recording register information. STARTS and ENDS
4877 must be allocated using the malloc library routine, and must each
4878 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4880 If NUM_REGS == 0, then subsequent matches should allocate their own
4881 register data.
4883 Unless this function is called, the first search or match using
4884 PATTERN_BUFFER will allocate its own register data, without
4885 freeing the old data. */
4887 void
4891 {
4893 {
4898 }
4899 else
4900 {
4904 }
4905 }
4906 #ifdef _LIBC
4908 #endif
4909 \f
4910 /* Searching routines. */
4912 /* Like re_search_2, below, but only one string is specified, and
4913 doesn't let you say where to stop matching. */
4915 int
4918 {
4921 }
4922 #ifdef _LIBC
4924 #endif
4927 /* Using the compiled pattern in BUFP->buffer, first tries to match the
4928 virtual concatenation of STRING1 and STRING2, starting first at index
4929 STARTPOS, then at STARTPOS + 1, and so on.
4931 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
4933 RANGE is how far to scan while trying to match. RANGE = 0 means try
4934 only at STARTPOS; in general, the last start tried is STARTPOS +
4935 RANGE.
4937 In REGS, return the indices of the virtual concatenation of STRING1
4938 and STRING2 that matched the entire BUFP->buffer and its contained
4939 subexpressions.
4941 Do not consider matching one past the index STOP in the virtual
4942 concatenation of STRING1 and STRING2.
4944 We return either the position in the strings at which the match was
4945 found, -1 if no match, or -2 if error (such as failure
4946 stack overflow). */
4948 int
4952 {
4953 # ifdef MBS_SUPPORT
4957 else
4958 # endif
4962 #ifdef _LIBC
4964 #endif
4968 #ifdef INSIDE_RECURSION
4970 #ifdef MATCH_MAY_ALLOCATE
4971 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
4972 #else
4973 # define FREE_VAR(var) if (var) free (var); var = NULL
4974 #endif
4976 #ifdef WCHAR
4977 # define MAX_ALLOCA_SIZE 2000
4979 # define FREE_WCS_BUFFERS() \
4980 do { \
4981 if (size1 > MAX_ALLOCA_SIZE) \
4982 { \
4983 free (wcs_string1); \
4984 free (mbs_offset1); \
4985 } \
4986 else \
4987 { \
4988 FREE_VAR (wcs_string1); \
4989 FREE_VAR (mbs_offset1); \
4990 } \
4991 if (size2 > MAX_ALLOCA_SIZE) \
4992 { \
4993 free (wcs_string2); \
4994 free (mbs_offset2); \
4995 } \
4996 else \
4997 { \
4998 FREE_VAR (wcs_string2); \
4999 FREE_VAR (mbs_offset2); \
5000 } \
5001 } while (0)
5003 #endif
5006 static int
5011 {
5017 #ifdef WCHAR
5018 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5020 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5022 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5024 /* They hold whether each wchar_t is binary data or not. */
5028 /* Check for out-of-range STARTPOS. */
5032 /* Fix up RANGE if it might eventually take us outside
5033 the virtual concatenation of STRING1 and STRING2.
5034 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5040 /* If the search isn't to be a backwards one, don't waste time in a
5041 search for a pattern that must be anchored. */
5044 /* `begline' is like `begbuf' if it cannot match at newlines. */
5047 {
5050 else
5052 }
5054 #ifdef emacs
5055 /* In a forward search for something that starts with \=.
5056 don't keep searching past point. */
5058 {
5062 }
5065 /* Update the fastmap now if not correct already. */
5070 #ifdef WCHAR
5071 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5072 fill them with converted string. */
5074 {
5076 {
5080 }
5081 else
5082 {
5086 }
5088 {
5090 {
5094 }
5095 else
5096 {
5100 }
5102 }
5108 else
5110 }
5112 {
5114 {
5118 }
5119 else
5120 {
5124 }
5126 {
5130 else
5133 }
5139 else
5141 }
5145 /* Loop through the string, looking for a place to start matching. */
5147 {
5148 /* If a fastmap is supplied, skip quickly over characters that
5149 cannot be the start of a match. If the pattern can match the
5150 null string, however, we don't need to skip characters; we want
5151 the first null string. */
5153 {
5155 {
5165 /* Written out as an if-else to avoid testing `translate'
5166 inside the loop. */
5171 range--;
5172 else
5174 range--;
5177 }
5179 {
5186 }
5187 }
5189 /* If can't match the null string, and that's all we have left, fail. */
5192 {
5193 #ifdef WCHAR
5195 #endif
5197 }
5199 #ifdef WCHAR
5210 #ifndef REGEX_MALLOC
5211 # ifdef C_ALLOCA
5213 # endif
5214 #endif
5217 {
5218 #ifdef WCHAR
5220 #endif
5222 }
5225 {
5226 #ifdef WCHAR
5228 #endif
5230 }
5232 advance:
5236 {
5237 range--;
5238 startpos++;
5239 }
5240 else
5241 {
5242 range++;
5243 startpos--;
5244 }
5245 }
5246 #ifdef WCHAR
5248 #endif
5250 }
5252 #ifdef WCHAR
5253 /* This converts PTR, a pointer into one of the search wchar_t strings
5254 `string1' and `string2' into an multibyte string offset from the
5255 beginning of that string. We use mbs_offset to optimize.
5256 See convert_mbs_to_wcs. */
5257 # define POINTER_TO_OFFSET(ptr) \
5258 (FIRST_STRING_P (ptr) \
5259 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5260 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5261 + csize1)))
5263 /* This converts PTR, a pointer into one of the search strings `string1'
5264 and `string2' into an offset from the beginning of that string. */
5265 # define POINTER_TO_OFFSET(ptr) \
5266 (FIRST_STRING_P (ptr) \
5267 ? ((regoff_t) ((ptr) - string1)) \
5268 : ((regoff_t) ((ptr) - string2 + size1)))
5271 /* Macros for dealing with the split strings in re_match_2. */
5273 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5275 /* Call before fetching a character with *d. This switches over to
5276 string2 if necessary. */
5277 #define PREFETCH() \
5278 while (d == dend) \
5279 { \
5281 if (dend == end_match_2) \
5282 goto fail; \
5284 d = string2; \
5285 dend = end_match_2; \
5286 }
5288 /* Test if at very beginning or at very end of the virtual concatenation
5289 of `string1' and `string2'. If only one string, it's `string2'. */
5290 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5291 #define AT_STRINGS_END(d) ((d) == end2)
5294 /* Test if D points to a character which is word-constituent. We have
5295 two special cases to check for: if past the end of string1, look at
5296 the first character in string2; and if before the beginning of
5297 string2, look at the last character in string1. */
5298 #ifdef WCHAR
5299 /* Use internationalized API instead of SYNTAX. */
5300 # define WORDCHAR_P(d) \
5301 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5302 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5303 || ((d) == end1 ? *string2 \
5306 # define WORDCHAR_P(d) \
5307 (SYNTAX ((d) == end1 ? *string2 \
5308 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5309 == Sword)
5312 /* Disabled due to a compiler bug -- see comment at case wordbound */
5313 #if 0
5314 /* Test if the character before D and the one at D differ with respect
5315 to being word-constituent. */
5316 #define AT_WORD_BOUNDARY(d) \
5317 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5318 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5319 #endif
5321 /* Free everything we malloc. */
5322 #ifdef MATCH_MAY_ALLOCATE
5323 # ifdef WCHAR
5324 # define FREE_VARIABLES() \
5325 do { \
5326 REGEX_FREE_STACK (fail_stack.stack); \
5327 FREE_VAR (regstart); \
5328 FREE_VAR (regend); \
5329 FREE_VAR (old_regstart); \
5330 FREE_VAR (old_regend); \
5331 FREE_VAR (best_regstart); \
5332 FREE_VAR (best_regend); \
5333 FREE_VAR (reg_info); \
5334 FREE_VAR (reg_dummy); \
5335 FREE_VAR (reg_info_dummy); \
5336 if (!cant_free_wcs_buf) \
5337 { \
5338 FREE_VAR (string1); \
5339 FREE_VAR (string2); \
5340 FREE_VAR (mbs_offset1); \
5341 FREE_VAR (mbs_offset2); \
5342 } \
5343 } while (0)
5345 # define FREE_VARIABLES() \
5346 do { \
5347 REGEX_FREE_STACK (fail_stack.stack); \
5348 FREE_VAR (regstart); \
5349 FREE_VAR (regend); \
5350 FREE_VAR (old_regstart); \
5351 FREE_VAR (old_regend); \
5352 FREE_VAR (best_regstart); \
5353 FREE_VAR (best_regend); \
5354 FREE_VAR (reg_info); \
5355 FREE_VAR (reg_dummy); \
5356 FREE_VAR (reg_info_dummy); \
5357 } while (0)
5359 #else
5360 # ifdef WCHAR
5361 # define FREE_VARIABLES() \
5362 do { \
5363 if (!cant_free_wcs_buf) \
5364 { \
5365 FREE_VAR (string1); \
5366 FREE_VAR (string2); \
5367 FREE_VAR (mbs_offset1); \
5368 FREE_VAR (mbs_offset2); \
5369 } \
5370 } while (0)
5376 /* These values must meet several constraints. They must not be valid
5377 register values; since we have a limit of 255 registers (because
5378 we use only one byte in the pattern for the register number), we can
5379 use numbers larger than 255. They must differ by 1, because of
5380 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5381 be larger than the value for the highest register, so we do not try
5382 to actually save any registers when none are active. */
5383 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5384 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5385 \f
5387 /* Matching routines. */
5390 /* re_match is like re_match_2 except it takes only a single string. */
5392 int
5395 {
5397 # ifdef MBS_SUPPORT
5402 else
5403 # endif
5406 # ifndef REGEX_MALLOC
5407 # ifdef C_ALLOCA
5409 # endif
5410 # endif
5412 }
5413 # ifdef _LIBC
5415 # endif
5420 #ifdef INSIDE_RECURSION
5434 /* re_match_2 matches the compiled pattern in BUFP against the
5435 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5436 and SIZE2, respectively). We start matching at POS, and stop
5437 matching at STOP.
5439 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5440 store offsets for the substring each group matched in REGS. See the
5441 documentation for exactly how many groups we fill.
5443 We return -1 if no match, -2 if an internal error (such as the
5444 failure stack overflowing). Otherwise, we return the length of the
5445 matched substring. */
5447 int
5451 {
5453 # ifdef MBS_SUPPORT
5458 else
5459 # endif
5463 #ifndef REGEX_MALLOC
5464 # ifdef C_ALLOCA
5466 # endif
5467 #endif
5469 }
5470 #ifdef _LIBC
5472 #endif
5476 #ifdef INSIDE_RECURSION
5478 #ifdef WCHAR
5481 /* This check the substring (from 0, to length) of the multibyte string,
5482 to which offset_buffer correspond. And count how many wchar_t_characters
5483 the substring occupy. We use offset_buffer to optimization.
5484 See convert_mbs_to_wcs. */
5486 static int
5488 {
5491 /* Check whether the size is valid. */
5498 /* If there are no multibyte character, offset_buffer[i] == i.
5499 Optmize for this case. */
5503 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5508 {
5516 else
5518 }
5521 }
5524 /* This is a separate function so that we can force an alloca cleanup
5525 afterwards. */
5526 #ifdef WCHAR
5527 static int
5534 /* string1 == string2 == NULL means string1/2, size1/2 and
5535 mbs_offset1/2 need seting up in this function. */
5536 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5539 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5542 static int
5549 {
5550 /* General temporaries. */
5553 #ifdef WCHAR
5554 /* They hold whether each wchar_t is binary data or not. */
5556 /* If true, we can't free string1/2, mbs_offset1/2. */
5560 /* Just past the end of the corresponding string. */
5563 /* Pointers into string1 and string2, just past the last characters in
5564 each to consider matching. */
5567 /* Where we are in the data, and the end of the current string. */
5570 /* Where we are in the pattern, and the end of the pattern. */
5571 #ifdef WCHAR
5579 /* Mark the opcode just after a start_memory, so we can test for an
5580 empty subpattern when we get to the stop_memory. */
5583 /* We use this to map every character in the string. */
5586 /* Failure point stack. Each place that can handle a failure further
5587 down the line pushes a failure point on this stack. It consists of
5588 restart, regend, and reg_info for all registers corresponding to
5589 the subexpressions we're currently inside, plus the number of such
5590 registers, and, finally, two char *'s. The first char * is where
5591 to resume scanning the pattern; the second one is where to resume
5592 scanning the strings. If the latter is zero, the failure point is
5593 a ``dummy''; if a failure happens and the failure point is a dummy,
5594 it gets discarded and the next next one is tried. */
5597 #endif
5598 #ifdef DEBUG
5601 #endif
5603 #ifdef REL_ALLOC
5604 /* This holds the pointer to the failure stack, when
5605 it is allocated relocatably. */
5607 #endif
5609 /* We fill all the registers internally, independent of what we
5610 return, for use in backreferences. The number here includes
5611 an element for register zero. */
5614 /* The currently active registers. */
5618 /* Information on the contents of registers. These are pointers into
5619 the input strings; they record just what was matched (on this
5620 attempt) by a subexpression part of the pattern, that is, the
5621 regnum-th regstart pointer points to where in the pattern we began
5622 matching and the regnum-th regend points to right after where we
5623 stopped matching the regnum-th subexpression. (The zeroth register
5624 keeps track of what the whole pattern matches.) */
5627 #endif
5629 /* If a group that's operated upon by a repetition operator fails to
5630 match anything, then the register for its start will need to be
5631 restored because it will have been set to wherever in the string we
5632 are when we last see its open-group operator. Similarly for a
5633 register's end. */
5636 #endif
5638 /* The is_active field of reg_info helps us keep track of which (possibly
5639 nested) subexpressions we are currently in. The matched_something
5640 field of reg_info[reg_num] helps us tell whether or not we have
5641 matched any of the pattern so far this time through the reg_num-th
5642 subexpression. These two fields get reset each time through any
5643 loop their register is in. */
5646 #endif
5648 /* The following record the register info as found in the above
5649 variables when we find a match better than any we've seen before.
5650 This happens as we backtrack through the failure points, which in
5651 turn happens only if we have not yet matched the entire string. */
5655 #endif
5657 /* Logically, this is `best_regend[0]'. But we don't want to have to
5658 allocate space for that if we're not allocating space for anything
5659 else (see below). Also, we never need info about register 0 for
5660 any of the other register vectors, and it seems rather a kludge to
5661 treat `best_regend' differently than the rest. So we keep track of
5662 the end of the best match so far in a separate variable. We
5663 initialize this to NULL so that when we backtrack the first time
5664 and need to test it, it's not garbage. */
5667 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5670 /* Used when we pop values we don't care about. */
5674 #endif
5676 #ifdef DEBUG
5677 /* Counts the total number of registers pushed. */
5679 #endif
5685 #ifdef MATCH_MAY_ALLOCATE
5686 /* Do not bother to initialize all the register variables if there are
5687 no groups in the pattern, as it takes a fair amount of time. If
5688 there are groups, we include space for register 0 (the whole
5689 pattern), even though we never use it, since it simplifies the
5690 array indexing. We should fix this. */
5692 {
5705 {
5708 }
5709 }
5710 else
5711 {
5712 /* We must initialize all our variables to NULL, so that
5713 `FREE_VARIABLES' doesn't try to free them. */
5717 }
5720 /* The starting position is bogus. */
5721 #ifdef WCHAR
5725 #endif
5726 {
5729 }
5731 #ifdef WCHAR
5732 /* Allocate wchar_t array for string1 and string2 and
5733 fill them with converted string. */
5735 {
5736 /* We need seting up buffers here. */
5738 /* We must free wcs buffers in this function. */
5742 {
5747 {
5752 }
5753 }
5755 {
5760 {
5767 }
5772 }
5773 }
5775 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5776 pattern to (char*) in regex_compile. */
5782 /* Initialize subexpression text positions to -1 to mark ones that no
5783 start_memory/stop_memory has been seen for. Also initialize the
5784 register information struct. */
5786 {
5794 }
5796 /* We move `string1' into `string2' if the latter's empty -- but not if
5797 `string1' is null. */
5799 {
5804 #ifdef WCHAR
5809 #endif
5810 }
5814 /* Compute where to stop matching, within the two strings. */
5815 #ifdef WCHAR
5817 {
5821 }
5822 else
5823 {
5829 }
5834 }
5835 #else
5837 {
5840 }
5841 else
5842 {
5845 }
5848 /* `p' scans through the pattern as `d' scans through the data.
5849 `dend' is the end of the input string that `d' points within. `d'
5850 is advanced into the following input string whenever necessary, but
5851 this happens before fetching; therefore, at the beginning of the
5852 loop, `d' can be pointing at the end of a string, but it cannot
5853 equal `string2'. */
5854 #ifdef WCHAR
5856 {
5860 }
5861 else
5862 {
5866 }
5872 }
5873 #else
5875 {
5878 }
5879 else
5880 {
5883 }
5892 /* This loops over pattern commands. It exits by returning from the
5893 function if the match is complete, or it drops through if the match
5894 fails at this starting point in the input data. */
5896 {
5897 #ifdef _LIBC
5899 #else
5901 #endif
5907 /* If we haven't matched the entire string, and we want the
5908 longest match, try backtracking. */
5910 {
5911 /* 1 if this match ends in the same string (string1 or string2)
5912 as the best previous match. */
5913 boolean same_str_p;
5915 /* 1 if this match is the best seen so far. */
5916 boolean best_match_p;
5921 /* AIX compiler got confused when this was combined
5922 with the previous declaration. */
5925 else
5933 /* If exceeds best match so far, save it. */
5935 {
5942 {
5945 }
5946 }
5948 }
5950 /* If no failure points, don't restore garbage. And if
5951 last match is real best match, don't restore second
5952 best one. */
5954 {
5955 restore_best_regs:
5956 /* Restore best match. It may happen that `dend ==
5957 end_match_1' while the restored d is in string2.
5958 For example, the pattern `x.*y.*z' against the
5959 strings `x-' and `y-z-', if the two strings are
5960 not consecutive in memory. */
5968 {
5971 }
5972 }
5975 succeed_label:
5977 /* If caller wants register contents data back, do it. */
5979 {
5980 /* Have the register data arrays been allocated? */
5983 extra element beyond `num_regs' for the `-1' marker
5984 GNU code uses. */
5989 {
5992 }
5994 }
5997 allocated, reallocate them. If we need fewer, just
5998 leave it alone. */
6000 {
6005 {
6008 }
6009 }
6010 }
6011 else
6012 {
6013 /* These braces fend off a "empty body in an else-statement"
6014 warning under GCC when assert expands to nothing. */
6016 }
6018 /* Convert the pointer data in `regstart' and `regend' to
6019 indices. Register zero has to be set differently,
6020 since we haven't kept track of any info for it. */
6022 {
6024 #ifdef WCHAR
6028 else
6031 #else
6036 }
6038 /* Go through the first `min (num_regs, regs->num_regs)'
6039 registers, since that is all we initialized. */
6041 mcnt++)
6042 {
6045 else
6046 {
6051 }
6052 }
6054 /* If the regs structure we return has more elements than
6055 were in the pattern, set the extra elements to -1. If
6056 we (re)allocated the registers, this is the case,
6057 because we always allocate enough to have at least one
6058 -1 at the end. */
6068 #ifdef WCHAR
6071 else
6073 csize1;
6075 #else
6077 ? string1
6085 }
6087 /* Otherwise match next pattern command. */
6089 {
6090 /* Ignore these. Used to ignore the n of succeed_n's which
6091 currently have n == 0. */
6100 /* Match the next n pattern characters exactly. The following
6101 byte in the pattern defines n, and the n bytes after that
6102 are the characters to match. */
6104 #ifdef MBS_SUPPORT
6106 #endif
6110 /* This is written out as an if-else so we don't waste time
6111 testing `translate' inside the loop. */
6113 {
6114 do
6115 {
6117 #ifdef WCHAR
6119 {
6123 }
6124 else
6125 {
6128 }
6129 #else
6134 }
6136 }
6137 else
6138 {
6139 do
6140 {
6143 }
6145 }
6150 /* Match any character except possibly a newline or a null. */
6162 d++;
6168 {
6170 #ifdef WCHAR
6174 #define WORK_BUFFER_SIZE 128
6176 # ifdef _LIBC
6185 #ifdef WCHAR
6186 # ifdef _LIBC
6195 /* p points charset[6], so the address of the next instruction
6196 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6197 where l=length of char_classes, m=length of collating_symbol,
6198 n=equivalence_class, o=length of char_range,
6199 p'=length of character. */
6201 /* Update p to indicate the next instruction. */
6205 /* match with char_class? */
6207 {
6214 # ifdef _LIBC
6217 # else
6220 # endif
6221 }
6223 /* match with collating_symbol? */
6224 # ifdef _LIBC
6226 {
6231 workp++)
6232 {
6240 {
6241 /* Update d, however d will be incremented at
6242 char_set_matched:, we decrement d here. */
6245 }
6246 }
6247 }
6249 # endif
6250 /* If we can't look up collation data, we use wcscoll
6251 instead. */
6252 {
6254 {
6256 # ifdef _LIBC
6258 # else
6260 # endif
6262 /* If wcscoll(the collating symbol, whole string) > 0,
6263 any substring of the string never match with the
6264 collating symbol. */
6265 # ifdef _LIBC
6267 # else
6269 # endif
6270 {
6273 }
6275 /* First, we compare the collating symbol with
6276 the first character of the string.
6277 If it don't match, we add the next character to
6278 the compare buffer in turn. */
6280 {
6283 {
6288 }
6290 /* add next character to the compare buffer. */
6294 # ifdef _LIBC
6296 # else
6298 # endif
6303 /* (str_buf > workp) indicate (str_buf + X > workp),
6304 because for all X (str_buf + X > str_buf).
6305 So we don't need continue this loop. */
6308 /* Otherwise(str_buf < workp),
6309 (str_buf+next_character) may equals (workp).
6310 So we continue this loop. */
6311 }
6312 /* not matched */
6316 }
6317 }
6318 /* match with equivalence_class? */
6319 # ifdef _LIBC
6321 {
6323 /* Try to match the equivalence class against
6324 those known to the collate implementation. */
6333 /* This #include defines a local function! */
6334 # include <locale/weightwc.h>
6345 /* Write 1 collating element to str_buf, and
6346 get its index. */
6350 {
6353 {
6358 }
6362 }
6364 /* Update d, however d will be incremented at
6365 char_set_matched:, we decrement d here. */
6368 {
6371 else
6372 {
6375 }
6376 }
6381 workp++)
6382 {
6384 /* We already checked idx != 0 in regex_compile. */
6387 {
6395 }
6396 }
6397 /* not matched */
6400 }
6402 # endif
6403 /* If we can't look up collation data, we use wcscoll
6404 instead. */
6405 {
6407 {
6409 # ifdef _LIBC
6411 # else
6413 # endif
6415 /* If wcscoll(the collating symbol, whole string) > 0,
6416 any substring of the string never match with the
6417 collating symbol. */
6418 # ifdef _LIBC
6420 # else
6422 # endif
6423 {
6426 }
6428 /* First, we compare the equivalence class with
6429 the first character of the string.
6430 If it don't match, we add the next character to
6431 the compare buffer in turn. */
6433 {
6436 {
6441 }
6443 /* add next character to the compare buffer. */
6447 # ifdef _LIBC
6449 # else
6451 # endif
6457 /* (str_buf > workp) indicate (str_buf + X > workp),
6458 because for all X (str_buf + X > str_buf).
6459 So we don't need continue this loop. */
6462 /* Otherwise(str_buf < workp),
6463 (str_buf+next_character) may equals (workp).
6464 So we continue this loop. */
6465 }
6466 /* not matched */
6470 }
6471 }
6473 /* match with char_range? */
6474 # ifdef _LIBC
6476 {
6484 {
6487 /* We already compute the collation sequence value
6488 of the characters (or collating symbols). */
6494 }
6495 }
6496 else
6497 # endif
6498 {
6499 /* We set range_start_char at str_buf[0], range_end_char
6500 at str_buf[4], and compared char at str_buf[2]. */
6506 {
6509 /* match if (range_start_char <= c <= range_end_char). */
6511 /* If range_start(or end) < 0, we assume -range_start(end)
6512 is the offset of the collating symbol which is specified
6513 as the character of the range start(end). */
6515 /* range_start */
6518 else
6519 {
6522 }
6524 /* range_end */
6527 else
6528 {
6531 }
6533 # ifdef _LIBC
6536 # else
6539 # endif
6541 }
6542 }
6544 /* match with char? */
6551 char_set_matched:
6553 #else
6554 /* Cast to `unsigned' instead of `unsigned char' in case the
6555 bit list is a full 32 bytes long. */
6563 #undef WORK_BUFFER_SIZE
6566 d++;
6568 }
6571 /* The beginning of a group is represented by start_memory.
6572 The arguments are the register number in the next byte, and the
6573 number of groups inner to this one in the next. The text
6574 matched within the group is recorded (in the internal
6575 registers data structure) under the register number. */
6580 /* Find out if this group can match the empty string. */
6587 /* Save the position in the string where we were the last time
6588 we were at this open-group operator in case the group is
6589 operated upon by a repetition operator, e.g., with `(a*)*b'
6590 against `ab'; then we want to ignore where we are now in
6591 the string in case this attempt to match fails. */
6604 /* Clear this whenever we change the register activity status. */
6607 /* This is the new highest active register. */
6610 /* If nothing was active before, this is the new lowest active
6611 register. */
6615 /* Move past the register number and inner group count. */
6622 /* The stop_memory opcode represents the end of a group. Its
6623 arguments are the same as start_memory's: the register
6624 number, and the number of inner groups. */
6629 /* We need to save the string position the last time we were at
6630 this close-group operator in case the group is operated
6631 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6632 against `aba'; then we want to ignore where we are now in
6633 the string in case this attempt to match fails. */
6643 /* This register isn't active anymore. */
6646 /* Clear this whenever we change the register activity status. */
6649 /* If this was the only register active, nothing is active
6650 anymore. */
6652 {
6655 }
6656 else
6658 it isn't necessarily one less than now: consider
6659 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6660 new highest active register is 1. */
6663 r--;
6665 /* If we end up at register zero, that means that we saved
6666 the registers as the result of an `on_failure_jump', not
6667 a `start_memory', and we jumped to past the innermost
6668 `stop_memory'. For example, in ((.)*) we save
6669 registers 1 and 2 as a result of the *, but when we pop
6670 back to the second ), we are at the stop_memory 1.
6671 Thus, nothing is active. */
6673 {
6676 }
6677 else
6679 }
6681 /* If just failed to match something this time around with a
6682 group that's operated on by a repetition operator, try to
6683 force exit from the ``loop'', and restore the register
6684 information for this group that we had before trying this
6685 last match. */
6689 {
6695 {
6709 }
6712 /* If the next operation is a jump backwards in the pattern
6713 to an on_failure_jump right before the start_memory
6714 corresponding to this stop_memory, exit from the loop
6715 by forcing a failure after pushing on the stack the
6716 on_failure_jump's jump in the pattern, and d. */
6720 {
6721 /* If this group ever matched anything, then restore
6722 what its registers were before trying this last
6723 failed match, e.g., with `(a*)*b' against `ab' for
6724 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6725 against `aba' for regend[3].
6727 Also restore the registers for inner groups for,
6728 e.g., `((a*)(b*))*' against `aba' (register 3 would
6729 otherwise get trashed). */
6732 {
6737 /* Restore this and inner groups' (if any) registers. */
6739 r++)
6740 {
6743 /* xx why this test? */
6746 }
6747 }
6748 p1++;
6753 }
6754 }
6756 /* Move past the register number and the inner group count. */
6761 /* \<digit> has been turned into a `duplicate' command which is
6762 followed by the numeric value of <digit> as the register number. */
6764 {
6769 /* Can't back reference a group which we've never matched. */
6773 /* Where in input to try to start matching. */
6776 /* Where to stop matching; if both the place to start and
6777 the place to stop matching are in the same string, then
6778 set to the place to stop, otherwise, for now have to use
6779 the end of the first string. */
6785 {
6786 /* If necessary, advance to next segment in register
6787 contents. */
6789 {
6793 /* End of string1 => advance to string2. */
6796 }
6797 /* At end of register contents => success */
6800 /* If necessary, advance to next segment in data. */
6803 /* How many characters left in this segment to match. */
6806 /* Want how many consecutive characters we can match in
6807 one shot, so, if necessary, adjust the count. */
6811 /* Compare that many; failure if mismatch, else move
6812 past them. */
6819 /* Do this because we've match some characters. */
6821 }
6822 }
6826 /* begline matches the empty string at the beginning of the string
6827 (unless `not_bol' is set in `bufp'), and, if
6828 `newline_anchor' is set, after newlines. */
6833 {
6835 }
6837 {
6839 }
6840 /* In all other cases, we fail. */
6844 /* endline is the dual of begline. */
6849 {
6851 }
6853 /* We have to ``prefetch'' the next character. */
6856 {
6858 }
6862 /* Match at the very beginning of the data. */
6870 /* Match at the very end of the data. */
6878 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
6879 pushes NULL as the value for the string on the stack. Then
6880 `pop_failure_point' will keep the current value for the
6881 string, instead of restoring it. To see why, consider
6882 matching `foo\nbar' against `.*\n'. The .* matches the foo;
6883 then the . fails against the \n. But the next thing we want
6884 to do is match the \n against the \n; if we restored the
6885 string value, we would be back at the foo.
6887 Because this is used only in specific cases, we don't need to
6888 check all the things that `on_failure_jump' does, to make
6889 sure the right things get saved on the stack. Hence we don't
6890 share its code. The only reason to push anything on the
6891 stack at all is that otherwise we would have to change
6892 `anychar's code to do something besides goto fail in this
6893 case; that seems worse than this. */
6898 #ifdef _LIBC
6900 #else
6902 #endif
6908 /* Uses of on_failure_jump:
6910 Each alternative starts with an on_failure_jump that points
6911 to the beginning of the next alternative. Each alternative
6912 except the last ends with a jump that in effect jumps past
6913 the rest of the alternatives. (They really jump to the
6914 ending jump of the following alternative, because tensioning
6915 these jumps is a hassle.)
6917 Repeats start with an on_failure_jump that points past both
6918 the repetition text and either the following jump or
6919 pop_failure_jump back to this on_failure_jump. */
6921 on_failure:
6925 #ifdef _LIBC
6927 #else
6929 #endif
6931 /* If this on_failure_jump comes right before a group (i.e.,
6932 the original * applied to a group), save the information
6933 for that group and all inner ones, so that if we fail back
6934 to this point, the group's information will be correct.
6935 For example, in \(a*\)*\1, we need the preceding group,
6936 and in \(zz\(a*\)b*\)\2, we need the inner group. */
6938 /* We can't use `p' to check ahead because we push
6939 a failure point to `p + mcnt' after we do this. */
6942 /* We need to skip no_op's before we look for the
6943 start_memory in case this on_failure_jump is happening as
6944 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
6945 against aba. */
6947 p1++;
6950 {
6951 /* We have a new highest active register now. This will
6952 get reset at the start_memory we are about to get to,
6953 but we will have saved all the registers relevant to
6954 this repetition op, as described above. */
6958 }
6965 /* A smart repeat ends with `maybe_pop_jump'.
6966 We change it to either `pop_failure_jump' or `jump'. */
6970 {
6973 /* Compare the beginning of the repeat with what in the
6974 pattern follows its end. If we can establish that there
6975 is nothing that they would both match, i.e., that we
6976 would have to backtrack because of (as in, e.g., `a*a')
6977 then we can change to pop_failure_jump, because we'll
6978 never have to backtrack.
6980 This is not true in the case of alternatives: in
6981 `(a|ab)*' we do need to backtrack to the `ab' alternative
6982 (e.g., if the string was `ab'). But instead of trying to
6983 detect that here, the alternative has put on a dummy
6984 failure point which is what we will end up popping. */
6986 /* Skip over open/close-group commands.
6987 If what follows this loop is a ...+ construct,
6988 look at what begins its body, since we will have to
6989 match at least one of that. */
6991 {
6999 else
7001 }
7004 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7005 to the `maybe_finalize_jump' of this case. Examine what
7006 follows. */
7008 /* If we're at the end of the pattern, we can change. */
7010 {
7011 /* Consider what happens when matching ":\(.*\)"
7012 against ":/". I don't really understand this code
7013 yet. */
7015 pop_failure_jump;
7016 DEBUG_PRINT1
7018 }
7021 #ifdef MBS_SUPPORT
7023 #endif
7025 {
7026 register UCHAR_T c
7030 #ifdef MBS_SUPPORT
7032 #endif
7034 {
7036 pop_failure_jump;
7037 #ifdef WCHAR
7041 #else
7045 #endif
7046 }
7048 #ifndef WCHAR
7051 {
7058 /* `negate' is equal to 1 if c would match, which means
7059 that we can't change to pop_failure_jump. */
7061 {
7064 }
7065 }
7067 }
7068 #ifndef WCHAR
7070 {
7071 /* We win if the first character of the loop is not part
7072 of the charset. */
7077 {
7080 }
7083 {
7085 /* We win if the charset_not inside the loop
7086 lists every character listed in the charset after. */
7094 {
7097 }
7098 }
7100 {
7102 /* We win if the charset inside the loop
7103 has no overlap with the one after the loop. */
7106 idx++)
7111 {
7114 }
7115 }
7116 }
7118 }
7121 {
7125 }
7126 /* Note fall through. */
7129 /* The end of a simple repeat has a pop_failure_jump back to
7130 its matching on_failure_jump, where the latter will push a
7131 failure point. The pop_failure_jump takes off failure
7132 points put on by this pop_failure_jump's matching
7133 on_failure_jump; we got through the pattern to here from the
7134 matching on_failure_jump, so didn't fail. */
7136 {
7137 /* We need to pass separate storage for the lowest and
7138 highest registers, even though we don't care about the
7139 actual values. Otherwise, we will restore only one
7140 register from the stack, since lowest will == highest in
7141 `pop_failure_point'. */
7150 }
7151 /* Note fall through. */
7153 unconditional_jump:
7154 #ifdef _LIBC
7156 #else
7158 #endif
7159 /* Note fall through. */
7161 /* Unconditionally jump (without popping any failure points). */
7166 #ifdef _LIBC
7168 #else
7170 #endif
7174 /* We need this opcode so we can detect where alternatives end
7175 in `group_match_null_string_p' et al. */
7181 /* Normally, the on_failure_jump pushes a failure point, which
7182 then gets popped at pop_failure_jump. We will end up at
7183 pop_failure_jump, also, and with a pattern of, say, `a+', we
7184 are skipping over the on_failure_jump, so we have to push
7185 something meaningless for pop_failure_jump to pop. */
7188 /* It doesn't matter what we push for the string here. What
7189 the code at `fail' tests is the value for the pattern. */
7194 /* At the end of an alternative, we need to push a dummy failure
7195 point in case we are followed by a `pop_failure_jump', because
7196 we don't want the failure point for the alternative to be
7197 popped. For example, matching `(a|ab)*' against `aab'
7198 requires that we match the `ab' alternative. */
7201 /* See comments just above at `dummy_failure_jump' about the
7202 two zeroes. */
7206 /* Have to succeed matching what follows at least n times.
7207 After that, handle like `on_failure_jump'. */
7213 /* Originally, this is how many times we HAVE to succeed. */
7215 {
7216 mcnt--;
7219 #ifdef _LIBC
7222 #else
7225 #endif
7226 }
7228 {
7229 #ifdef _LIBC
7232 #else
7237 #ifdef WCHAR
7239 #else
7244 }
7251 /* Originally, this is how many times we CAN jump. */
7253 {
7254 mcnt--;
7257 #ifdef _LIBC
7259 mcnt);
7260 #else
7262 mcnt);
7265 }
7266 /* If don't have to jump any more, skip over the rest of command. */
7267 else
7272 {
7278 #ifdef _LIBC
7280 #else
7282 #endif
7285 }
7287 #if 0
7288 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7289 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7290 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7291 macro and introducing temporary variables works around the bug. */
7304 #else
7306 {
7318 }
7321 {
7333 }
7334 #endif
7350 #ifdef emacs
7377 matchsyntax:
7379 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7380 d++;
7394 matchnotsyntax:
7396 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7397 d++;
7410 d++;
7419 d++;
7425 }
7429 /* We goto here if a matching operation fails. */
7430 fail:
7438 /* If this failure point is a dummy, try the next one. */
7442 /* If we failed to the end of the pattern, don't examine *p. */
7445 {
7448 /* If failed to a backwards jump that's part of a repetition
7449 loop, need to pop this failure point and use the next one. */
7451 {
7462 || (!is_a_jump_n
7468 }
7469 }
7473 }
7474 else
7485 \f
7486 /* Subroutine definitions for re_match_2. */
7489 /* We are passed P pointing to a register number after a start_memory.
7491 Return true if the pattern up to the corresponding stop_memory can
7492 match the empty string, and false otherwise.
7494 If we find the matching stop_memory, sets P to point to one past its number.
7495 Otherwise, sets P to an undefined byte less than or equal to END.
7497 We don't handle duplicates properly (yet). */
7499 static boolean
7502 {
7504 /* Point to after the args to the start_memory. */
7508 {
7509 /* Skip over opcodes that can match nothing, and return true or
7510 false, as appropriate, when we get to one that can't, or to the
7511 matching stop_memory. */
7514 {
7515 /* Could be either a loop or a series of alternatives. */
7517 p1++;
7520 /* If the next operation is not a jump backwards in the
7521 pattern. */
7524 {
7525 /* Go through the on_failure_jumps of the alternatives,
7526 seeing if any of the alternatives cannot match nothing.
7527 The last alternative starts with only a jump,
7528 whereas the rest start with on_failure_jump and end
7529 with a jump, e.g., here is the pattern for `a|b|c':
7531 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7532 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7533 /exactn/1/c
7535 So, we have to first go through the first (n-1)
7536 alternatives and then deal with the last one separately. */
7539 /* Deal with the first (n-1) alternatives, which start
7540 with an on_failure_jump (see above) that jumps to right
7541 past a jump_past_alt. */
7544 jump_past_alt)
7545 {
7546 /* `mcnt' holds how many bytes long the alternative
7547 is, including the ending `jump_past_alt' and
7548 its number. */
7552 reg_info))
7555 /* Move to right after this alternative, including the
7556 jump_past_alt. */
7559 /* Break if it's the beginning of an n-th alternative
7560 that doesn't begin with an on_failure_jump. */
7564 /* Still have to check that it's not an n-th
7565 alternative that starts with an on_failure_jump. */
7566 p1++;
7569 jump_past_alt)
7570 {
7571 /* Get to the beginning of the n-th alternative. */
7574 }
7575 }
7577 /* Deal with the last alternative: go back and get number
7578 of the `jump_past_alt' just before it. `mcnt' contains
7579 the length of the alternative. */
7599 }
7606 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7607 It expects P to be the first byte of a single alternative and END one
7608 byte past the last. The alternative can contain groups. */
7610 static boolean
7613 {
7618 {
7619 /* Skip over opcodes that can match nothing, and break when we get
7620 to one that can't. */
7623 {
7624 /* It's a loop. */
7626 p1++;
7634 }
7641 /* Deals with the ops common to group_match_null_string_p and
7642 alt_match_null_string_p.
7644 Sets P to one after the op and its arguments, if any. */
7646 static boolean
7649 {
7651 boolean ret;
7656 {
7666 #ifdef emacs
7670 #endif
7678 /* Have to set this here in case we're checking a group which
7679 contains a group and a back reference to it. */
7688 /* If this is an optimized succeed_n for zero times, make the jump. */
7693 else
7698 /* Get to the number of times to succeed. */
7703 {
7707 }
7708 else
7721 /* All other opcodes mean we cannot match the empty string. */
7723 }
7730 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7731 bytes; nonzero otherwise. */
7733 static int
7735 RE_TRANSLATE_TYPE translate)
7736 {
7740 {
7741 #ifdef WCHAR
7748 len--;
7749 }
7751 }
7752 \f
7756 /* Entry points for GNU code. */
7758 /* re_compile_pattern is the GNU regular expression compiler: it
7759 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7760 Returns 0 if the pattern was valid, otherwise an error string.
7762 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7763 are set in BUFP on entry.
7765 We call regex_compile to do the actual compilation. */
7770 {
7771 reg_errcode_t ret;
7773 /* GNU code is written to assume at least RE_NREGS registers will be set
7774 (and at least one extra will be -1). */
7777 /* And GNU code determines whether or not to get register information
7778 by passing null for the REGS argument to re_match, etc., not by
7779 setting no_sub. */
7782 /* Match anchors at newline. */
7785 # ifdef MBS_SUPPORT
7788 else
7789 # endif
7795 }
7796 #ifdef _LIBC
7798 #endif
7799 \f
7800 /* Entry points compatible with 4.2 BSD regex library. We don't define
7801 them unless specifically requested. */
7803 #if defined _REGEX_RE_COMP || defined _LIBC
7805 /* BSD has one and only one pattern buffer. */
7809 #ifdef _LIBC
7810 /* Make these definitions weak in libc, so POSIX programs can redefine
7811 these names if they don't use our functions, and still use
7812 regcomp/regexec below without link errors. */
7813 weak_function
7814 #endif
7816 {
7817 reg_errcode_t ret;
7820 {
7824 }
7827 {
7836 }
7838 /* Since `re_exec' always passes NULL for the `regs' argument, we
7839 don't need to initialize the pattern buffer fields which affect it. */
7841 /* Match anchors at newlines. */
7844 # ifdef MBS_SUPPORT
7847 else
7848 # endif
7854 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7856 }
7859 int
7860 #ifdef _LIBC
7861 weak_function
7862 #endif
7864 {
7866 return
7868 }
7871 \f
7872 /* POSIX.2 functions. Don't define these for Emacs. */
7874 #ifndef emacs
7876 /* regcomp takes a regular expression as a string and compiles it.
7878 PREG is a regex_t *. We do not expect any fields to be initialized,
7879 since POSIX says we shouldn't. Thus, we set
7881 `buffer' to the compiled pattern;
7882 `used' to the length of the compiled pattern;
7883 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
7884 REG_EXTENDED bit in CFLAGS is set; otherwise, to
7885 RE_SYNTAX_POSIX_BASIC;
7886 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
7887 `fastmap' to an allocated space for the fastmap;
7888 `fastmap_accurate' to zero;
7889 `re_nsub' to the number of subexpressions in PATTERN.
7891 PATTERN is the address of the pattern string.
7893 CFLAGS is a series of bits which affect compilation.
7895 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
7896 use POSIX basic syntax.
7898 If REG_NEWLINE is set, then . and [^...] don't match newline.
7899 Also, regexec will try a match beginning after every newline.
7901 If REG_ICASE is set, then we considers upper- and lowercase
7902 versions of letters to be equivalent when matching.
7904 If REG_NOSUB is set, then when PREG is passed to regexec, that
7905 routine will report only success or failure, and nothing about the
7906 registers.
7908 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
7909 the return codes and their meanings.) */
7911 int
7913 {
7914 reg_errcode_t ret;
7915 reg_syntax_t syntax
7919 /* regex_compile will allocate the space for the compiled pattern. */
7924 /* Try to allocate space for the fastmap. */
7928 {
7931 preg->translate
7937 /* Map uppercase characters to corresponding lowercase ones. */
7940 }
7941 else
7944 /* If REG_NEWLINE is set, newlines are treated differently. */
7949 /* It also changes the matching behavior. */
7951 }
7952 else
7957 /* POSIX says a null character in the pattern terminates it, so we
7958 can use strlen here in compiling the pattern. */
7959 # ifdef MBS_SUPPORT
7962 else
7963 # endif
7966 /* POSIX doesn't distinguish between an unmatched open-group and an
7967 unmatched close-group: both are REG_EPAREN. */
7971 {
7972 /* Compute the fastmap now, since regexec cannot modify the pattern
7973 buffer. */
7975 {
7976 /* Some error occurred while computing the fastmap, just forget
7977 about it. */
7980 }
7981 }
7984 }
7985 #ifdef _LIBC
7987 #endif
7990 /* regexec searches for a given pattern, specified by PREG, in the
7991 string STRING.
7993 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
7994 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
7995 least NMATCH elements, and we set them to the offsets of the
7996 corresponding matched substrings.
7998 EFLAGS specifies `execution flags' which affect matching: if
7999 REG_NOTBOL is set, then ^ does not match at the beginning of the
8000 string; if REG_NOTEOL is set, then $ does not match at the end.
8002 We return 0 if we find a match and REG_NOMATCH if not. */
8004 int
8007 {
8010 regex_t private_preg;
8019 /* The user has told us exactly how many registers to return
8020 information about, via `nmatch'. We have to pass that on to the
8021 matching routines. */
8025 {
8031 }
8033 /* Perform the searching operation. */
8038 /* Copy the register information to the POSIX structure. */
8040 {
8042 {
8046 {
8049 }
8050 }
8052 /* If we needed the temporary register info, free the space now. */
8054 }
8056 /* We want zero return to mean success, unlike `re_search'. */
8058 }
8059 #ifdef _LIBC
8061 #endif
8064 /* Returns a message corresponding to an error code, ERRCODE, returned
8065 from either regcomp or regexec. We don't use PREG here. */
8067 size_t
8070 {
8077 /* Only error codes returned by the rest of the code should be passed
8078 to this routine. If we are given anything else, or if other regex
8079 code generates an invalid error code, then the program has a bug.
8080 Dump core so we can fix it. */
8088 {
8090 {
8091 #if defined HAVE_MEMPCPY || defined _LIBC
8093 #else
8096 #endif
8097 }
8098 else
8100 }
8103 }
8104 #ifdef _LIBC
8106 #endif
8109 /* Free dynamically allocated space used by PREG. */
8111 void
8113 {
8129 }
8130 #ifdef _LIBC
8132 #endif
8138 \f
8139 #undef STORE_NUMBER
8140 #undef STORE_NUMBER_AND_INCR
8141 #undef EXTRACT_NUMBER
8142 #undef EXTRACT_NUMBER_AND_INCR
8144 #undef DEBUG_PRINT_COMPILED_PATTERN
8145 #undef DEBUG_PRINT_DOUBLE_STRING
8147 #undef INIT_FAIL_STACK
8148 #undef RESET_FAIL_STACK
8149 #undef DOUBLE_FAIL_STACK
8150 #undef PUSH_PATTERN_OP
8151 #undef PUSH_FAILURE_POINTER
8152 #undef PUSH_FAILURE_INT
8153 #undef PUSH_FAILURE_ELT
8154 #undef POP_FAILURE_POINTER
8155 #undef POP_FAILURE_INT
8156 #undef POP_FAILURE_ELT
8157 #undef DEBUG_PUSH
8158 #undef DEBUG_POP
8159 #undef PUSH_FAILURE_POINT
8160 #undef POP_FAILURE_POINT
8162 #undef REG_UNSET_VALUE
8163 #undef REG_UNSET
8165 #undef PATFETCH
8166 #undef PATFETCH_RAW
8167 #undef PATUNFETCH
8168 #undef TRANSLATE
8170 #undef INIT_BUF_SIZE
8171 #undef GET_BUFFER_SPACE
8172 #undef BUF_PUSH
8173 #undef BUF_PUSH_2
8174 #undef BUF_PUSH_3
8175 #undef STORE_JUMP
8176 #undef STORE_JUMP2
8177 #undef INSERT_JUMP
8178 #undef INSERT_JUMP2
8179 #undef EXTEND_BUFFER
8180 #undef GET_UNSIGNED_NUMBER
8181 #undef FREE_STACK_RETURN
8183 # undef POINTER_TO_OFFSET
8184 # undef MATCHING_IN_FRST_STRING
8185 # undef PREFETCH
8186 # undef AT_STRINGS_BEG
8187 # undef AT_STRINGS_END
8188 # undef WORDCHAR_P
8189 # undef FREE_VAR
8190 # undef FREE_VARIABLES
8191 # undef NO_HIGHEST_ACTIVE_REG
8192 # undef NO_LOWEST_ACTIVE_REG
8194 # undef CHAR_T
8195 # undef UCHAR_T
8196 # undef COMPILED_BUFFER_VAR
8197 # undef OFFSET_ADDRESS_SIZE
8198 # undef CHAR_CLASS_SIZE
8199 # undef PREFIX
8200 # undef ARG_PREFIX
8201 # undef PUT_CHAR
8202 # undef BYTE
8203 # undef WCHAR
8205 # define DEFINED_ONCE