| blob | e123284e39eb4fd709ddf9aa878f8dbe26f26fbd |
1 /* Extended regular expression matching and search library, version
2 0.12. (Implements POSIX draft P10003.2/D11.2, except for
3 internationalization features.)
5 Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998 Free Software Foundation, Inc.
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2, or (at your option)
10 any later version.
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with this program; if not, write to the Free Software
19 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
20 USA. */
22 /* AIX requires this to be the first thing in the file. */
23 #if defined (_AIX) && !defined (REGEX_MALLOC)
24 #pragma alloca
25 #endif
27 #undef _GNU_SOURCE
28 #define _GNU_SOURCE
30 #ifdef emacs
31 /* Converts the pointer to the char to BEG-based offset from the start. */
32 #define PTR_TO_OFFSET(d) \
33 POS_AS_IN_BUFFER (MATCHING_IN_FIRST_STRING \
34 ? (d) - string1 : (d) - (string2 - size1))
35 #define POS_AS_IN_BUFFER(p) ((p) + (NILP (re_match_object) || BUFFERP (re_match_object)))
36 #else
37 #define PTR_TO_OFFSET(d) 0
38 #endif
40 #ifdef HAVE_CONFIG_H
41 #include <config.h>
42 #endif
44 /* We need this for `regex.h', and perhaps for the Emacs include files. */
45 #include <sys/types.h>
47 /* This is for other GNU distributions with internationalized messages. */
48 #if HAVE_LIBINTL_H || defined (_LIBC)
49 # include <libintl.h>
50 #else
51 # define gettext(msgid) (msgid)
52 #endif
54 #ifndef gettext_noop
55 /* This define is so xgettext can find the internationalizable
56 strings. */
57 #define gettext_noop(String) String
58 #endif
60 /* The `emacs' switch turns on certain matching commands
61 that make sense only in Emacs. */
62 #ifdef emacs
67 /* Make syntax table lookup grant data in gl_state. */
68 #define SYNTAX_ENTRY_VIA_PROPERTY
74 #define malloc xmalloc
75 #define realloc xrealloc
76 #define free xfree
80 /* If we are not linking with Emacs proper,
81 we can't use the relocating allocator
82 even if config.h says that we can. */
83 #undef REL_ALLOC
85 #if defined (STDC_HEADERS) || defined (_LIBC)
86 #include <stdlib.h>
87 #else
90 #endif
92 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
93 If nothing else has been done, use the method below. */
94 #ifdef INHIBIT_STRING_HEADER
95 #if !(defined (HAVE_BZERO) && defined (HAVE_BCOPY))
96 #if !defined (bzero) && !defined (bcopy)
97 #undef INHIBIT_STRING_HEADER
98 #endif
99 #endif
100 #endif
102 /* This is the normal way of making sure we have a bcopy and a bzero.
103 This is used in most programs--a few other programs avoid this
104 by defining INHIBIT_STRING_HEADER. */
105 #ifndef INHIBIT_STRING_HEADER
106 #if defined (HAVE_STRING_H) || defined (STDC_HEADERS) || defined (_LIBC)
107 #include <string.h>
108 #ifndef bcmp
109 #define bcmp(s1, s2, n) memcmp ((s1), (s2), (n))
110 #endif
111 #ifndef bcopy
112 #define bcopy(s, d, n) memcpy ((d), (s), (n))
113 #endif
114 #ifndef bzero
115 #define bzero(s, n) memset ((s), 0, (n))
116 #endif
117 #else
118 #include <strings.h>
119 #endif
120 #endif
122 /* Define the syntax stuff for \<, \>, etc. */
124 /* This must be nonzero for the wordchar and notwordchar pattern
125 commands in re_match_2. */
126 #ifndef Sword
127 #define Sword 1
128 #endif
130 #ifdef SWITCH_ENUM_BUG
131 #define SWITCH_ENUM_CAST(x) ((int)(x))
132 #else
133 #define SWITCH_ENUM_CAST(x) (x)
134 #endif
136 #ifdef SYNTAX_TABLE
142 /* How many characters in the character set. */
143 #define CHAR_SET_SIZE 256
147 static void
149 {
170 }
174 #define SYNTAX(c) re_syntax_table[c]
176 /* Dummy macros for non-Emacs environments. */
177 #define BASE_LEADING_CODE_P(c) (0)
178 #define WORD_BOUNDARY_P(c1, c2) (0)
179 #define CHAR_HEAD_P(p) (1)
180 #define SINGLE_BYTE_CHAR_P(c) (1)
181 #define SAME_CHARSET_P(c1, c2) (1)
182 #define MULTIBYTE_FORM_LENGTH(p, s) (1)
183 #define STRING_CHAR(p, s) (*(p))
184 #define STRING_CHAR_AND_LENGTH(p, s, actual_len) ((actual_len) = 1, *(p))
185 #define GET_CHAR_AFTER_2(c, p, str1, end1, str2, end2) \
186 (c = ((p) == (end1) ? *(str2) : *(p)))
187 #define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
188 (c = ((p) == (str2) ? *((end1) - 1) : *((p) - 1)))
190 \f
191 /* Get the interface, including the syntax bits. */
194 /* isalpha etc. are used for the character classes. */
195 #include <ctype.h>
197 /* Jim Meyering writes:
199 "... Some ctype macros are valid only for character codes that
200 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
201 using /bin/cc or gcc but without giving an ansi option). So, all
202 ctype uses should be through macros like ISPRINT... If
203 STDC_HEADERS is defined, then autoconf has verified that the ctype
204 macros don't need to be guarded with references to isascii. ...
205 Defining isascii to 1 should let any compiler worth its salt
206 eliminate the && through constant folding." */
208 #if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII))
209 #define ISASCII(c) 1
210 #else
211 #define ISASCII(c) isascii(c)
212 #endif
214 #ifdef isblank
215 #define ISBLANK(c) (ISASCII (c) && isblank (c))
216 #else
218 #endif
219 #ifdef isgraph
220 #define ISGRAPH(c) (ISASCII (c) && isgraph (c))
221 #else
222 #define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
223 #endif
225 #define ISPRINT(c) (ISASCII (c) && isprint (c))
226 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
227 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
228 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
229 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
230 #define ISLOWER(c) (ISASCII (c) && islower (c))
231 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
232 #define ISSPACE(c) (ISASCII (c) && isspace (c))
233 #define ISUPPER(c) (ISASCII (c) && isupper (c))
234 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
236 #ifndef NULL
237 #define NULL (void *)0
238 #endif
240 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
241 since ours (we hope) works properly with all combinations of
242 machines, compilers, `char' and `unsigned char' argument types.
243 (Per Bothner suggested the basic approach.) */
244 #undef SIGN_EXTEND_CHAR
245 #if __STDC__
246 #define SIGN_EXTEND_CHAR(c) ((signed char) (c))
248 /* As in Harbison and Steele. */
249 #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
250 #endif
251 \f
252 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
253 use `alloca' instead of `malloc'. This is because using malloc in
254 re_search* or re_match* could cause memory leaks when C-g is used in
255 Emacs; also, malloc is slower and causes storage fragmentation. On
256 the other hand, malloc is more portable, and easier to debug.
258 Because we sometimes use alloca, some routines have to be macros,
259 not functions -- `alloca'-allocated space disappears at the end of the
260 function it is called in. */
262 #ifdef REGEX_MALLOC
264 #define REGEX_ALLOCATE malloc
265 #define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
266 #define REGEX_FREE free
270 /* Emacs already defines alloca, sometimes. */
271 #ifndef alloca
273 /* Make alloca work the best possible way. */
274 #ifdef __GNUC__
275 #define alloca __builtin_alloca
277 #if HAVE_ALLOCA_H
278 #include <alloca.h>
284 #endif
290 #define REGEX_ALLOCATE alloca
292 /* Assumes a `char *destination' variable. */
293 #define REGEX_REALLOCATE(source, osize, nsize) \
294 (destination = (char *) alloca (nsize), \
295 bcopy (source, destination, osize), \
296 destination)
298 /* No need to do anything to free, after alloca. */
303 /* Define how to allocate the failure stack. */
305 #if defined (REL_ALLOC) && defined (REGEX_MALLOC)
307 #define REGEX_ALLOCATE_STACK(size) \
308 r_alloc (&failure_stack_ptr, (size))
309 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
310 r_re_alloc (&failure_stack_ptr, (nsize))
311 #define REGEX_FREE_STACK(ptr) \
312 r_alloc_free (&failure_stack_ptr)
316 #ifdef REGEX_MALLOC
318 #define REGEX_ALLOCATE_STACK malloc
319 #define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
320 #define REGEX_FREE_STACK free
324 #define REGEX_ALLOCATE_STACK alloca
326 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
327 REGEX_REALLOCATE (source, osize, nsize)
328 /* No need to explicitly free anything. */
329 #define REGEX_FREE_STACK(arg)
335 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
336 `string1' or just past its end. This works if PTR is NULL, which is
337 a good thing. */
338 #define FIRST_STRING_P(ptr) \
339 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
341 /* (Re)Allocate N items of type T using malloc, or fail. */
342 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
343 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
344 #define RETALLOC_IF(addr, n, t) \
345 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
346 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
350 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
352 #undef MAX
353 #undef MIN
354 #define MAX(a, b) ((a) > (b) ? (a) : (b))
355 #define MIN(a, b) ((a) < (b) ? (a) : (b))
358 #define false 0
359 #define true 1
362 \f
363 /* These are the command codes that appear in compiled regular
364 expressions. Some opcodes are followed by argument bytes. A
365 command code can specify any interpretation whatsoever for its
366 arguments. Zero bytes may appear in the compiled regular expression. */
369 {
372 /* Succeed right away--no more backtracking. */
373 succeed,
375 /* Followed by one byte giving n, then by n literal bytes. */
376 exactn,
378 /* Matches any (more or less) character. */
379 anychar,
381 /* Matches any one char belonging to specified set. First
382 following byte is number of bitmap bytes. Then come bytes
383 for a bitmap saying which chars are in. Bits in each byte
384 are ordered low-bit-first. A character is in the set if its
385 bit is 1. A character too large to have a bit in the map is
386 automatically not in the set. */
387 charset,
389 /* Same parameters as charset, but match any character that is
390 not one of those specified. */
391 charset_not,
393 /* Start remembering the text that is matched, for storing in a
394 register. Followed by one byte with the register number, in
395 the range 0 to one less than the pattern buffer's re_nsub
396 field. Then followed by one byte with the number of groups
397 inner to this one. (This last has to be part of the
398 start_memory only because we need it in the on_failure_jump
399 of re_match_2.) */
400 start_memory,
402 /* Stop remembering the text that is matched and store it in a
403 memory register. Followed by one byte with the register
404 number, in the range 0 to one less than `re_nsub' in the
405 pattern buffer, and one byte with the number of inner groups,
406 just like `start_memory'. (We need the number of inner
407 groups here because we don't have any easy way of finding the
408 corresponding start_memory when we're at a stop_memory.) */
409 stop_memory,
411 /* Match a duplicate of something remembered. Followed by one
412 byte containing the register number. */
413 duplicate,
415 /* Fail unless at beginning of line. */
416 begline,
418 /* Fail unless at end of line. */
419 endline,
421 /* Succeeds if at beginning of buffer (if emacs) or at beginning
422 of string to be matched (if not). */
423 begbuf,
425 /* Analogously, for end of buffer/string. */
426 endbuf,
428 /* Followed by two byte relative address to which to jump. */
429 jump,
431 /* Same as jump, but marks the end of an alternative. */
432 jump_past_alt,
434 /* Followed by two-byte relative address of place to resume at
435 in case of failure. */
436 on_failure_jump,
438 /* Like on_failure_jump, but pushes a placeholder instead of the
439 current string position when executed. */
440 on_failure_keep_string_jump,
442 /* Throw away latest failure point and then jump to following
443 two-byte relative address. */
444 pop_failure_jump,
446 /* Change to pop_failure_jump if know won't have to backtrack to
447 match; otherwise change to jump. This is used to jump
448 back to the beginning of a repeat. If what follows this jump
449 clearly won't match what the repeat does, such that we can be
450 sure that there is no use backtracking out of repetitions
451 already matched, then we change it to a pop_failure_jump.
452 Followed by two-byte address. */
453 maybe_pop_jump,
455 /* Jump to following two-byte address, and push a dummy failure
456 point. This failure point will be thrown away if an attempt
457 is made to use it for a failure. A `+' construct makes this
458 before the first repeat. Also used as an intermediary kind
459 of jump when compiling an alternative. */
460 dummy_failure_jump,
462 /* Push a dummy failure point and continue. Used at the end of
463 alternatives. */
464 push_dummy_failure,
466 /* Followed by two-byte relative address and two-byte number n.
467 After matching N times, jump to the address upon failure. */
468 succeed_n,
470 /* Followed by two-byte relative address, and two-byte number n.
471 Jump to the address N times, then fail. */
472 jump_n,
474 /* Set the following two-byte relative address to the
475 subsequent two-byte number. The address *includes* the two
476 bytes of number. */
477 set_number_at,
486 notwordbound /* Succeeds if not at a word boundary. */
488 #ifdef emacs
493 /* Matches any character whose syntax is specified. Followed by
494 a byte which contains a syntax code, e.g., Sword. */
495 syntaxspec,
497 /* Matches any character whose syntax is not that specified. */
498 notsyntaxspec,
500 /* Matches any character whose category-set contains the specified
501 category. The operator is followed by a byte which contains a
502 category code (mnemonic ASCII character). */
503 categoryspec,
505 /* Matches any character whose category-set does not contain the
506 specified category. The operator is followed by a byte which
507 contains the category code (mnemonic ASCII character). */
508 notcategoryspec
511 \f
512 /* Common operations on the compiled pattern. */
514 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
516 #define STORE_NUMBER(destination, number) \
517 do { \
518 (destination)[0] = (number) & 0377; \
519 (destination)[1] = (number) >> 8; \
520 } while (0)
522 /* Same as STORE_NUMBER, except increment DESTINATION to
523 the byte after where the number is stored. Therefore, DESTINATION
524 must be an lvalue. */
526 #define STORE_NUMBER_AND_INCR(destination, number) \
527 do { \
528 STORE_NUMBER (destination, number); \
529 (destination) += 2; \
530 } while (0)
532 /* Put into DESTINATION a number stored in two contiguous bytes starting
533 at SOURCE. */
535 #define EXTRACT_NUMBER(destination, source) \
536 do { \
537 (destination) = *(source) & 0377; \
538 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
539 } while (0)
541 #ifdef DEBUG
542 static void
546 {
550 }
553 #undef EXTRACT_NUMBER
554 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
559 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
560 SOURCE must be an lvalue. */
562 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
563 do { \
564 EXTRACT_NUMBER (destination, source); \
565 (source) += 2; \
566 } while (0)
568 #ifdef DEBUG
569 static void
573 {
576 }
578 #ifndef EXTRACT_MACROS
579 #undef EXTRACT_NUMBER_AND_INCR
580 #define EXTRACT_NUMBER_AND_INCR(dest, src) \
581 extract_number_and_incr (&dest, &src)
585 \f
586 /* Store a multibyte character in three contiguous bytes starting
587 DESTINATION, and increment DESTINATION to the byte after where the
588 character is stored. Therefore, DESTINATION must be an lvalue. */
590 #define STORE_CHARACTER_AND_INCR(destination, character) \
591 do { \
592 (destination)[0] = (character) & 0377; \
593 (destination)[1] = ((character) >> 8) & 0377; \
594 (destination)[2] = (character) >> 16; \
595 (destination) += 3; \
596 } while (0)
598 /* Put into DESTINATION a character stored in three contiguous bytes
599 starting at SOURCE. */
601 #define EXTRACT_CHARACTER(destination, source) \
602 do { \
603 (destination) = ((source)[0] \
604 | ((source)[1] << 8) \
605 | ((source)[2] << 16)); \
606 } while (0)
609 /* Macros for charset. */
611 /* Size of bitmap of charset P in bytes. P is a start of charset,
612 i.e. *P is (re_opcode_t) charset or (re_opcode_t) charset_not. */
613 #define CHARSET_BITMAP_SIZE(p) ((p)[1] & 0x7F)
615 /* Nonzero if charset P has range table. */
616 #define CHARSET_RANGE_TABLE_EXISTS_P(p) ((p)[1] & 0x80)
618 /* Return the address of range table of charset P. But not the start
619 of table itself, but the before where the number of ranges is
620 stored. `2 +' means to skip re_opcode_t and size of bitmap. */
621 #define CHARSET_RANGE_TABLE(p) (&(p)[2 + CHARSET_BITMAP_SIZE (p)])
623 /* Test if C is listed in the bitmap of charset P. */
624 #define CHARSET_LOOKUP_BITMAP(p, c) \
625 ((c) < CHARSET_BITMAP_SIZE (p) * BYTEWIDTH \
626 && (p)[2 + (c) / BYTEWIDTH] & (1 << ((c) % BYTEWIDTH)))
628 /* Return the address of end of RANGE_TABLE. COUNT is number of
629 ranges (which is a pair of (start, end)) in the RANGE_TABLE. `* 2'
630 is start of range and end of range. `* 3' is size of each start
631 and end. */
632 #define CHARSET_RANGE_TABLE_END(range_table, count) \
633 ((range_table) + (count) * 2 * 3)
635 /* Test if C is in RANGE_TABLE. A flag NOT is negated if C is in.
636 COUNT is number of ranges in RANGE_TABLE. */
637 #define CHARSET_LOOKUP_RANGE_TABLE_RAW(not, c, range_table, count) \
638 do \
639 { \
640 int range_start, range_end; \
641 unsigned char *p; \
642 unsigned char *range_table_end \
643 = CHARSET_RANGE_TABLE_END ((range_table), (count)); \
644 \
645 for (p = (range_table); p < range_table_end; p += 2 * 3) \
646 { \
647 EXTRACT_CHARACTER (range_start, p); \
648 EXTRACT_CHARACTER (range_end, p + 3); \
649 \
650 if (range_start <= (c) && (c) <= range_end) \
651 { \
652 (not) = !(not); \
653 break; \
654 } \
655 } \
656 } \
657 while (0)
659 /* Test if C is in range table of CHARSET. The flag NOT is negated if
660 C is listed in it. */
661 #define CHARSET_LOOKUP_RANGE_TABLE(not, c, charset) \
662 do \
663 { \
665 int count; \
666 unsigned char *range_table = CHARSET_RANGE_TABLE (charset); \
667 \
668 EXTRACT_NUMBER_AND_INCR (count, range_table); \
669 CHARSET_LOOKUP_RANGE_TABLE_RAW ((not), (c), range_table, count); \
670 } \
671 while (0)
672 \f
673 /* If DEBUG is defined, Regex prints many voluminous messages about what
674 it is doing (if the variable `debug' is nonzero). If linked with the
675 main program in `iregex.c', you can enter patterns and strings
676 interactively. And if linked with the main program in `main.c' and
677 the other test files, you can run the already-written tests. */
679 #ifdef DEBUG
681 /* We use standard I/O for debugging. */
682 #include <stdio.h>
684 /* It is useful to test things that ``must'' be true when debugging. */
685 #include <assert.h>
689 #define DEBUG_STATEMENT(e) e
690 #define DEBUG_PRINT1(x) if (debug) printf (x)
691 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
692 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
693 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
694 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
695 if (debug) print_partial_compiled_pattern (s, e)
696 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
697 if (debug) print_double_string (w, s1, sz1, s2, sz2)
700 /* Print the fastmap in human-readable form. */
702 void
705 {
710 {
712 {
716 {
718 i++;
719 }
721 {
724 }
725 }
726 }
728 }
731 /* Print a compiled pattern string in human-readable form, starting at
732 the START pointer into it and ending just before the pointer END. */
734 void
738 {
744 {
747 }
749 /* Loop over pattern commands. */
751 {
755 {
763 do
764 {
767 }
791 {
803 {
804 /* Are we starting a range? */
806 {
809 }
810 /* Have we broken a range? */
812 {
815 }
821 }
829 }
912 #ifdef emacs
956 }
959 }
962 }
965 void
968 {
975 {
978 }
988 /* Perhaps we should print the translate table? */
989 }
992 void
999 {
1004 else
1005 {
1007 {
1012 }
1016 }
1017 }
1021 #undef assert
1022 #define assert(e)
1024 #define DEBUG_STATEMENT(e)
1025 #define DEBUG_PRINT1(x)
1026 #define DEBUG_PRINT2(x1, x2)
1027 #define DEBUG_PRINT3(x1, x2, x3)
1028 #define DEBUG_PRINT4(x1, x2, x3, x4)
1029 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1030 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1033 \f
1034 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1035 also be assigned to arbitrarily: each pattern buffer stores its own
1036 syntax, so it can be changed between regex compilations. */
1037 /* This has no initializer because initialized variables in Emacs
1038 become read-only after dumping. */
1039 reg_syntax_t re_syntax_options;
1042 /* Specify the precise syntax of regexps for compilation. This provides
1043 for compatibility for various utilities which historically have
1044 different, incompatible syntaxes.
1046 The argument SYNTAX is a bit mask comprised of the various bits
1047 defined in regex.h. We return the old syntax. */
1049 reg_syntax_t
1051 reg_syntax_t syntax;
1052 {
1057 }
1058 \f
1059 /* This table gives an error message for each of the error codes listed
1060 in regex.h. Obviously the order here has to be same as there.
1061 POSIX doesn't require that we do anything for REG_NOERROR,
1062 but why not be nice? */
1065 {
1083 };
1084 \f
1085 /* Avoiding alloca during matching, to placate r_alloc. */
1087 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1088 searching and matching functions should not call alloca. On some
1089 systems, alloca is implemented in terms of malloc, and if we're
1090 using the relocating allocator routines, then malloc could cause a
1091 relocation, which might (if the strings being searched are in the
1092 ralloc heap) shift the data out from underneath the regexp
1093 routines.
1095 Here's another reason to avoid allocation: Emacs
1096 processes input from X in a signal handler; processing X input may
1097 call malloc; if input arrives while a matching routine is calling
1098 malloc, then we're scrod. But Emacs can't just block input while
1099 calling matching routines; then we don't notice interrupts when
1100 they come in. So, Emacs blocks input around all regexp calls
1101 except the matching calls, which it leaves unprotected, in the
1102 faith that they will not malloc. */
1104 /* Normally, this is fine. */
1105 #define MATCH_MAY_ALLOCATE
1107 /* When using GNU C, we are not REALLY using the C alloca, no matter
1108 what config.h may say. So don't take precautions for it. */
1109 #ifdef __GNUC__
1110 #undef C_ALLOCA
1111 #endif
1113 /* The match routines may not allocate if (1) they would do it with malloc
1114 and (2) it's not safe for them to use malloc.
1115 Note that if REL_ALLOC is defined, matching would not use malloc for the
1116 failure stack, but we would still use it for the register vectors;
1117 so REL_ALLOC should not affect this. */
1118 #if (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (emacs)
1119 #undef MATCH_MAY_ALLOCATE
1120 #endif
1122 \f
1123 /* Failure stack declarations and macros; both re_compile_fastmap and
1124 re_match_2 use a failure stack. These have to be macros because of
1125 REGEX_ALLOCATE_STACK. */
1128 /* Approximate number of failure points for which to initially allocate space
1129 when matching. If this number is exceeded, we allocate more
1130 space, so it is not a hard limit. */
1131 #ifndef INIT_FAILURE_ALLOC
1132 #define INIT_FAILURE_ALLOC 20
1133 #endif
1135 /* Roughly the maximum number of failure points on the stack. Would be
1136 exactly that if always used TYPICAL_FAILURE_SIZE items each time we failed.
1137 This is a variable only so users of regex can assign to it; we never
1138 change it ourselves. */
1139 #if defined (MATCH_MAY_ALLOCATE)
1140 /* Note that 4400 is enough to cause a crash on Alpha OSF/1,
1141 whose default stack limit is 2mb. In order for a larger
1142 value to work reliably, you have to try to make it accord
1143 with the process stack limit. */
1145 #else
1147 #endif
1149 union fail_stack_elt
1150 {
1153 };
1158 {
1164 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1165 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1166 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1169 /* Define macros to initialize and free the failure stack.
1170 Do `return -2' if the alloc fails. */
1172 #ifdef MATCH_MAY_ALLOCATE
1173 #define INIT_FAIL_STACK() \
1174 do { \
1175 fail_stack.stack = (fail_stack_elt_t *) \
1176 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * TYPICAL_FAILURE_SIZE \
1177 * sizeof (fail_stack_elt_t)); \
1178 \
1179 if (fail_stack.stack == NULL) \
1180 return -2; \
1181 \
1182 fail_stack.size = INIT_FAILURE_ALLOC; \
1183 fail_stack.avail = 0; \
1184 } while (0)
1186 #define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1187 #else
1188 #define INIT_FAIL_STACK() \
1189 do { \
1190 fail_stack.avail = 0; \
1191 } while (0)
1193 #define RESET_FAIL_STACK()
1194 #endif
1197 /* Double the size of FAIL_STACK, up to a limit
1198 which allows approximately `re_max_failures' items.
1200 Return 1 if succeeds, and 0 if either ran out of memory
1201 allocating space for it or it was already too large.
1203 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1205 /* Factor to increase the failure stack size by
1206 when we increase it.
1207 This used to be 2, but 2 was too wasteful
1208 because the old discarded stacks added up to as much space
1209 were as ultimate, maximum-size stack. */
1210 #define FAIL_STACK_GROWTH_FACTOR 4
1212 #define GROW_FAIL_STACK(fail_stack) \
1213 (((fail_stack).size * sizeof (fail_stack_elt_t) \
1214 >= re_max_failures * TYPICAL_FAILURE_SIZE) \
1215 ? 0 \
1216 : ((fail_stack).stack \
1217 = (fail_stack_elt_t *) \
1218 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1219 (fail_stack).size * sizeof (fail_stack_elt_t), \
1220 MIN (re_max_failures * TYPICAL_FAILURE_SIZE, \
1221 ((fail_stack).size * sizeof (fail_stack_elt_t) \
1222 * FAIL_STACK_GROWTH_FACTOR))), \
1223 \
1224 (fail_stack).stack == NULL \
1225 ? 0 \
1226 : ((fail_stack).size \
1227 = (MIN (re_max_failures * TYPICAL_FAILURE_SIZE, \
1228 ((fail_stack).size * sizeof (fail_stack_elt_t) \
1229 * FAIL_STACK_GROWTH_FACTOR)) \
1230 / sizeof (fail_stack_elt_t)), \
1231 1)))
1234 /* Push pointer POINTER on FAIL_STACK.
1235 Return 1 if was able to do so and 0 if ran out of memory allocating
1236 space to do so. */
1237 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1238 ((FAIL_STACK_FULL () \
1239 && !GROW_FAIL_STACK (FAIL_STACK)) \
1240 ? 0 \
1241 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1242 1))
1244 /* Push a pointer value onto the failure stack.
1245 Assumes the variable `fail_stack'. Probably should only
1246 be called from within `PUSH_FAILURE_POINT'. */
1247 #define PUSH_FAILURE_POINTER(item) \
1248 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1250 /* This pushes an integer-valued item onto the failure stack.
1251 Assumes the variable `fail_stack'. Probably should only
1252 be called from within `PUSH_FAILURE_POINT'. */
1253 #define PUSH_FAILURE_INT(item) \
1254 fail_stack.stack[fail_stack.avail++].integer = (item)
1256 /* Push a fail_stack_elt_t value onto the failure stack.
1257 Assumes the variable `fail_stack'. Probably should only
1258 be called from within `PUSH_FAILURE_POINT'. */
1259 #define PUSH_FAILURE_ELT(item) \
1260 fail_stack.stack[fail_stack.avail++] = (item)
1262 /* These three POP... operations complement the three PUSH... operations.
1263 All assume that `fail_stack' is nonempty. */
1264 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1265 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1266 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1268 /* Used to omit pushing failure point id's when we're not debugging. */
1269 #ifdef DEBUG
1270 #define DEBUG_PUSH PUSH_FAILURE_INT
1271 #define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1272 #else
1273 #define DEBUG_PUSH(item)
1274 #define DEBUG_POP(item_addr)
1275 #endif
1278 /* Push the information about the state we will need
1279 if we ever fail back to it.
1281 Requires variables fail_stack, regstart, regend, reg_info, and
1282 num_regs be declared. GROW_FAIL_STACK requires `destination' be
1283 declared.
1285 Does `return FAILURE_CODE' if runs out of memory. */
1287 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1288 do { \
1289 char *destination; \
1290 /* Must be int, so when we don't save any registers, the arithmetic \
1292 int this_reg; \
1293 \
1294 DEBUG_STATEMENT (failure_id++); \
1295 DEBUG_STATEMENT (nfailure_points_pushed++); \
1299 \
1302 \
1304 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1305 { \
1306 if (!GROW_FAIL_STACK (fail_stack)) \
1307 return failure_code; \
1308 \
1310 (fail_stack).size); \
1312 } \
1313 \
1316 \
1317 if (1) \
1318 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1319 this_reg++) \
1320 { \
1322 DEBUG_STATEMENT (num_regs_pushed++); \
1323 \
1325 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1326 \
1328 PUSH_FAILURE_POINTER (regend[this_reg]); \
1329 \
1332 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1335 MATCHED_SOMETHING (reg_info[this_reg])); \
1337 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1339 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1340 } \
1341 \
1343 PUSH_FAILURE_INT (lowest_active_reg); \
1344 \
1346 PUSH_FAILURE_INT (highest_active_reg); \
1347 \
1349 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1350 PUSH_FAILURE_POINTER (pattern_place); \
1351 \
1353 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1354 size2); \
1356 PUSH_FAILURE_POINTER (string_place); \
1357 \
1359 DEBUG_PUSH (failure_id); \
1360 } while (0)
1362 /* This is the number of items that are pushed and popped on the stack
1363 for each register. */
1364 #define NUM_REG_ITEMS 3
1366 /* Individual items aside from the registers. */
1367 #ifdef DEBUG
1369 #else
1370 #define NUM_NONREG_ITEMS 4
1371 #endif
1373 /* Estimate the size of data pushed by a typical failure stack entry.
1374 An estimate is all we need, because all we use this for
1375 is to choose a limit for how big to make the failure stack. */
1377 #define TYPICAL_FAILURE_SIZE 20
1379 /* This is how many items we actually use for a failure point.
1380 It depends on the regexp. */
1381 #define NUM_FAILURE_ITEMS \
1382 (((0 \
1383 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1384 * NUM_REG_ITEMS) \
1385 + NUM_NONREG_ITEMS)
1387 /* How many items can still be added to the stack without overflowing it. */
1388 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1391 /* Pops what PUSH_FAIL_STACK pushes.
1393 We restore into the parameters, all of which should be lvalues:
1394 STR -- the saved data position.
1395 PAT -- the saved pattern position.
1396 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1397 REGSTART, REGEND -- arrays of string positions.
1398 REG_INFO -- array of information about each subexpression.
1400 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1401 `pend', `string1', `size1', `string2', and `size2'. */
1403 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1404 { \
1405 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \
1406 int this_reg; \
1407 const unsigned char *string_temp; \
1408 \
1409 assert (!FAIL_STACK_EMPTY ()); \
1410 \
1415 \
1416 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1417 \
1418 DEBUG_POP (&failure_id); \
1420 \
1421 /* If the saved string location is NULL, it came from an \
1422 on_failure_keep_string_jump opcode, and we want to throw away the \
1424 string_temp = POP_FAILURE_POINTER (); \
1425 if (string_temp != NULL) \
1426 str = (const char *) string_temp; \
1427 \
1429 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1431 \
1432 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1434 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1435 \
1437 high_reg = (unsigned) POP_FAILURE_INT (); \
1439 \
1440 low_reg = (unsigned) POP_FAILURE_INT (); \
1442 \
1443 if (1) \
1444 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1445 { \
1447 \
1448 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1450 \
1451 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1453 \
1454 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1456 } \
1457 else \
1458 { \
1459 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1460 { \
1461 reg_info[this_reg].word.integer = 0; \
1462 regend[this_reg] = 0; \
1463 regstart[this_reg] = 0; \
1464 } \
1465 highest_active_reg = high_reg; \
1466 } \
1467 \
1468 set_regs_matched_done = 0; \
1469 DEBUG_STATEMENT (nfailure_points_popped++); \
1473 \f
1474 /* Structure for per-register (a.k.a. per-group) information.
1475 Other register information, such as the
1476 starting and ending positions (which are addresses), and the list of
1477 inner groups (which is a bits list) are maintained in separate
1478 variables.
1480 We are making a (strictly speaking) nonportable assumption here: that
1481 the compiler will pack our bit fields into something that fits into
1482 the type of `word', i.e., is something that fits into one item on the
1483 failure stack. */
1486 {
1487 fail_stack_elt_t word;
1488 struct
1489 {
1490 /* This field is one if this group can match the empty string,
1491 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1492 #define MATCH_NULL_UNSET_VALUE 3
1500 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1501 #define IS_ACTIVE(R) ((R).bits.is_active)
1502 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1503 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1506 /* Call this when have matched a real character; it sets `matched' flags
1507 for the subexpressions which we are currently inside. Also records
1508 that those subexprs have matched. */
1509 #define SET_REGS_MATCHED() \
1510 do \
1511 { \
1512 if (!set_regs_matched_done) \
1513 { \
1514 unsigned r; \
1515 set_regs_matched_done = 1; \
1516 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1517 { \
1518 MATCHED_SOMETHING (reg_info[r]) \
1519 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1520 = 1; \
1521 } \
1522 } \
1523 } \
1524 while (0)
1526 /* Registers are set to a sentinel when they haven't yet matched. */
1528 #define REG_UNSET_VALUE (®_unset_dummy)
1529 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1530 \f
1531 /* Subroutine declarations and macros for regex_compile. */
1538 /* Fetch the next character in the uncompiled pattern---translating it
1539 if necessary. Also cast from a signed character in the constant
1540 string passed to us by the user to an unsigned char that we can use
1541 as an array index (in, e.g., `translate'). */
1542 #ifndef PATFETCH
1543 #define PATFETCH(c) \
1544 do {if (p == pend) return REG_EEND; \
1545 c = (unsigned char) *p++; \
1546 if (RE_TRANSLATE_P (translate)) c = RE_TRANSLATE (translate, c); \
1547 } while (0)
1548 #endif
1550 /* Fetch the next character in the uncompiled pattern, with no
1551 translation. */
1552 #define PATFETCH_RAW(c) \
1553 do {if (p == pend) return REG_EEND; \
1554 c = (unsigned char) *p++; \
1555 } while (0)
1557 /* Go backwards one character in the pattern. */
1558 #define PATUNFETCH p--
1561 /* If `translate' is non-null, return translate[D], else just D. We
1562 cast the subscript to translate because some data is declared as
1563 `char *', to avoid warnings when a string constant is passed. But
1564 when we use a character as a subscript we must make it unsigned. */
1565 #ifndef TRANSLATE
1566 #define TRANSLATE(d) \
1567 (RE_TRANSLATE_P (translate) \
1568 ? (unsigned) RE_TRANSLATE (translate, (unsigned) (d)) : (d))
1569 #endif
1572 /* Macros for outputting the compiled pattern into `buffer'. */
1574 /* If the buffer isn't allocated when it comes in, use this. */
1575 #define INIT_BUF_SIZE 32
1577 /* Make sure we have at least N more bytes of space in buffer. */
1578 #define GET_BUFFER_SPACE(n) \
1579 while (b - bufp->buffer + (n) > bufp->allocated) \
1580 EXTEND_BUFFER ()
1582 /* Make sure we have one more byte of buffer space and then add C to it. */
1583 #define BUF_PUSH(c) \
1584 do { \
1585 GET_BUFFER_SPACE (1); \
1586 *b++ = (unsigned char) (c); \
1587 } while (0)
1590 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1591 #define BUF_PUSH_2(c1, c2) \
1592 do { \
1593 GET_BUFFER_SPACE (2); \
1594 *b++ = (unsigned char) (c1); \
1595 *b++ = (unsigned char) (c2); \
1596 } while (0)
1599 /* As with BUF_PUSH_2, except for three bytes. */
1600 #define BUF_PUSH_3(c1, c2, c3) \
1601 do { \
1602 GET_BUFFER_SPACE (3); \
1603 *b++ = (unsigned char) (c1); \
1604 *b++ = (unsigned char) (c2); \
1605 *b++ = (unsigned char) (c3); \
1606 } while (0)
1609 /* Store a jump with opcode OP at LOC to location TO. We store a
1610 relative address offset by the three bytes the jump itself occupies. */
1611 #define STORE_JUMP(op, loc, to) \
1612 store_op1 (op, loc, (to) - (loc) - 3)
1614 /* Likewise, for a two-argument jump. */
1615 #define STORE_JUMP2(op, loc, to, arg) \
1616 store_op2 (op, loc, (to) - (loc) - 3, arg)
1618 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1619 #define INSERT_JUMP(op, loc, to) \
1620 insert_op1 (op, loc, (to) - (loc) - 3, b)
1622 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1623 #define INSERT_JUMP2(op, loc, to, arg) \
1624 insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
1627 /* This is not an arbitrary limit: the arguments which represent offsets
1628 into the pattern are two bytes long. So if 2^16 bytes turns out to
1629 be too small, many things would have to change. */
1630 #define MAX_BUF_SIZE (1L << 16)
1633 /* Extend the buffer by twice its current size via realloc and
1634 reset the pointers that pointed into the old block to point to the
1635 correct places in the new one. If extending the buffer results in it
1636 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1637 #define EXTEND_BUFFER() \
1638 do { \
1639 unsigned char *old_buffer = bufp->buffer; \
1640 if (bufp->allocated == MAX_BUF_SIZE) \
1641 return REG_ESIZE; \
1642 bufp->allocated <<= 1; \
1643 if (bufp->allocated > MAX_BUF_SIZE) \
1644 bufp->allocated = MAX_BUF_SIZE; \
1645 bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\
1646 if (bufp->buffer == NULL) \
1647 return REG_ESPACE; \
1649 if (old_buffer != bufp->buffer) \
1650 { \
1651 b = (b - old_buffer) + bufp->buffer; \
1652 begalt = (begalt - old_buffer) + bufp->buffer; \
1653 if (fixup_alt_jump) \
1654 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1655 if (laststart) \
1656 laststart = (laststart - old_buffer) + bufp->buffer; \
1657 if (pending_exact) \
1658 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1659 } \
1660 } while (0)
1663 /* Since we have one byte reserved for the register number argument to
1664 {start,stop}_memory, the maximum number of groups we can report
1665 things about is what fits in that byte. */
1666 #define MAX_REGNUM 255
1668 /* But patterns can have more than `MAX_REGNUM' registers. We just
1669 ignore the excess. */
1673 /* Macros for the compile stack. */
1675 /* Since offsets can go either forwards or backwards, this type needs to
1676 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1680 {
1681 pattern_offset_t begalt_offset;
1682 pattern_offset_t fixup_alt_jump;
1683 pattern_offset_t inner_group_offset;
1684 pattern_offset_t laststart_offset;
1685 regnum_t regnum;
1690 {
1697 #define INIT_COMPILE_STACK_SIZE 32
1699 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1700 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1702 /* The next available element. */
1703 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1706 /* Structure to manage work area for range table. */
1707 struct range_table_work_area
1708 {
1712 };
1714 /* Make sure that WORK_AREA can hold more N multibyte characters. */
1715 #define EXTEND_RANGE_TABLE_WORK_AREA(work_area, n) \
1716 do { \
1717 if (((work_area).used + (n)) * sizeof (int) > (work_area).allocated) \
1718 { \
1719 (work_area).allocated += 16 * sizeof (int); \
1720 if ((work_area).table) \
1721 (work_area).table \
1722 = (int *) realloc ((work_area).table, (work_area).allocated); \
1723 else \
1724 (work_area).table \
1725 = (int *) malloc ((work_area).allocated); \
1726 if ((work_area).table == 0) \
1727 FREE_STACK_RETURN (REG_ESPACE); \
1728 } \
1729 } while (0)
1731 /* Set a range (RANGE_START, RANGE_END) to WORK_AREA. */
1732 #define SET_RANGE_TABLE_WORK_AREA(work_area, range_start, range_end) \
1733 do { \
1734 EXTEND_RANGE_TABLE_WORK_AREA ((work_area), 2); \
1735 (work_area).table[(work_area).used++] = (range_start); \
1736 (work_area).table[(work_area).used++] = (range_end); \
1737 } while (0)
1739 /* Free allocated memory for WORK_AREA. */
1740 #define FREE_RANGE_TABLE_WORK_AREA(work_area) \
1741 do { \
1742 if ((work_area).table) \
1743 free ((work_area).table); \
1744 } while (0)
1746 #define CLEAR_RANGE_TABLE_WORK_USED(work_area) ((work_area).used = 0)
1747 #define RANGE_TABLE_WORK_USED(work_area) ((work_area).used)
1748 #define RANGE_TABLE_WORK_ELT(work_area, i) ((work_area).table[i])
1751 /* Set the bit for character C in a list. */
1752 #define SET_LIST_BIT(c) \
1753 (b[((unsigned char) (c)) / BYTEWIDTH] \
1754 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1757 /* Get the next unsigned number in the uncompiled pattern. */
1758 #define GET_UNSIGNED_NUMBER(num) \
1759 { if (p != pend) \
1760 { \
1761 PATFETCH (c); \
1762 while (ISDIGIT (c)) \
1763 { \
1764 if (num < 0) \
1765 num = 0; \
1767 if (p == pend) \
1768 break; \
1769 PATFETCH (c); \
1770 } \
1771 } \
1772 }
1776 #define IS_CHAR_CLASS(string) \
1783 \f
1784 #ifndef MATCH_MAY_ALLOCATE
1786 /* If we cannot allocate large objects within re_match_2_internal,
1787 we make the fail stack and register vectors global.
1788 The fail stack, we grow to the maximum size when a regexp
1789 is compiled.
1790 The register vectors, we adjust in size each time we
1791 compile a regexp, according to the number of registers it needs. */
1795 /* Size with which the following vectors are currently allocated.
1796 That is so we can make them bigger as needed,
1797 but never make them smaller. */
1807 /* Make the register vectors big enough for NUM_REGS registers,
1808 but don't make them smaller. */
1810 static
1813 {
1815 {
1827 }
1828 }
1831 \f
1832 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1833 Returns one of error codes defined in `regex.h', or zero for success.
1835 Assumes the `allocated' (and perhaps `buffer') and `translate'
1836 fields are set in BUFP on entry.
1838 If it succeeds, results are put in BUFP (if it returns an error, the
1839 contents of BUFP are undefined):
1840 `buffer' is the compiled pattern;
1841 `syntax' is set to SYNTAX;
1842 `used' is set to the length of the compiled pattern;
1843 `fastmap_accurate' is zero;
1844 `re_nsub' is the number of subexpressions in PATTERN;
1845 `not_bol' and `not_eol' are zero;
1847 The `fastmap' and `newline_anchor' fields are neither
1848 examined nor set. */
1850 /* Return, freeing storage we allocated. */
1851 #define FREE_STACK_RETURN(value) \
1852 do { \
1853 FREE_RANGE_TABLE_WORK_AREA (range_table_work); \
1854 free (compile_stack.stack); \
1855 return value; \
1856 } while (0)
1858 static reg_errcode_t
1862 reg_syntax_t syntax;
1864 {
1865 /* We fetch characters from PATTERN here. Even though PATTERN is
1866 `char *' (i.e., signed), we declare these variables as unsigned, so
1867 they can be reliably used as array indices. */
1870 /* A random temporary spot in PATTERN. */
1873 /* Points to the end of the buffer, where we should append. */
1876 /* Keeps track of unclosed groups. */
1877 compile_stack_type compile_stack;
1879 /* Points to the current (ending) position in the pattern. */
1880 #ifdef AIX
1881 /* `const' makes AIX compiler fail. */
1883 #else
1885 #endif
1888 /* How to translate the characters in the pattern. */
1891 /* Address of the count-byte of the most recently inserted `exactn'
1892 command. This makes it possible to tell if a new exact-match
1893 character can be added to that command or if the character requires
1894 a new `exactn' command. */
1897 /* Address of start of the most recently finished expression.
1898 This tells, e.g., postfix * where to find the start of its
1899 operand. Reset at the beginning of groups and alternatives. */
1902 /* Address of beginning of regexp, or inside of last group. */
1905 /* Place in the uncompiled pattern (i.e., the {) to
1906 which to go back if the interval is invalid. */
1909 /* Address of the place where a forward jump should go to the end of
1910 the containing expression. Each alternative of an `or' -- except the
1911 last -- ends with a forward jump of this sort. */
1914 /* Counts open-groups as they are encountered. Remembered for the
1915 matching close-group on the compile stack, so the same register
1916 number is put in the stop_memory as the start_memory. */
1919 /* Work area for range table of charset. */
1922 #ifdef DEBUG
1925 {
1931 }
1934 /* Initialize the compile stack. */
1945 /* Initialize the pattern buffer. */
1950 /* Set `used' to zero, so that if we return an error, the pattern
1951 printer (for debugging) will think there's no pattern. We reset it
1952 at the end. */
1955 /* Always count groups, whether or not bufp->no_sub is set. */
1958 #ifdef emacs
1959 /* bufp->multibyte is set before regex_compile is called, so don't alter
1960 it. */
1962 /* Nothing is recognized as a multibyte character. */
1964 #endif
1966 #if !defined (emacs) && !defined (SYNTAX_TABLE)
1967 /* Initialize the syntax table. */
1969 #endif
1972 {
1975 enough space. This loses if buffer's address is bogus, but
1976 that is the user's responsibility. */
1978 }
1979 else
1982 }
1986 }
1990 /* Loop through the uncompiled pattern until we're at the end. */
1992 {
1996 {
1998 {
2001 /* If context independent, it's an operator. */
2003 /* Otherwise, depends on what's come before. */
2006 else
2008 }
2013 {
2015 p == pend
2016 /* If context independent, it's an operator. */
2018 /* Otherwise, depends on what's next. */
2021 else
2023 }
2032 handle_plus:
2034 /* If there is no previous pattern... */
2036 {
2041 }
2043 {
2044 /* Are we optimizing this jump? */
2047 /* 1 means zero (many) matches is allowed. */
2050 /* If there is a sequence of repetition chars, collapse it
2051 down to just one (the right one). We can't combine
2052 interval operators with these because of, e.g., `a{2}*',
2053 which should only match an even number of `a's. */
2056 {
2067 ;
2070 {
2075 {
2076 PATUNFETCH;
2077 PATUNFETCH;
2079 }
2082 }
2083 else
2084 {
2085 PATUNFETCH;
2087 }
2089 /* If we get here, we found another repeat character. */
2090 }
2092 /* Star, etc. applied to an empty pattern is equivalent
2093 to an empty pattern. */
2097 /* Now we know whether or not zero matches is allowed
2098 and also whether or not two or more matches is allowed. */
2101 end a backward relative jump from `b' to before the next
2102 jump we're going to put in below (which jumps from
2103 laststart to after this jump).
2105 But if we are at the `*' in the exact sequence `.*\n',
2106 insert an unconditional jump backwards to the .,
2107 instead of the beginning of the loop. This way we only
2108 push a failure point once, instead of every time
2109 through the loop. */
2112 /* Allocate the space for the jump. */
2115 /* We know we are not at the first character of the pattern,
2116 because laststart was nonzero. And we've already
2117 incremented `p', by the way, to be the character after
2118 the `*'. Do we have to do something analogous here
2119 for null bytes, because of RE_DOT_NOT_NULL? */
2121 && zero_times_ok
2128 }
2129 else
2130 /* Anything else. */
2133 /* We've added more stuff to the buffer. */
2135 }
2137 /* On failure, jump from laststart to b + 3, which will be the
2138 end of the buffer after this jump is inserted. */
2147 {
2148 /* At least one repetition is required, so insert a
2149 `dummy_failure_jump' before the initial
2150 `on_failure_jump' instruction of the loop. This
2151 effects a skip over that instruction the first time
2152 we hit that loop. */
2156 }
2157 }
2168 {
2173 /* Ensure that we have enough space to push a charset: the
2174 opcode, the length count, and the bitset; 34 bytes in all. */
2179 /* We test `*p == '^' twice, instead of using an if
2180 statement, so we only need one BUF_PUSH. */
2183 p++;
2185 /* Remember the first position in the bracket expression. */
2188 /* Push the number of bytes in the bitmap. */
2191 /* Clear the whole map. */
2194 /* charset_not matches newline according to a syntax bit. */
2199 /* Read in characters and ranges, setting map bits. */
2201 {
2209 /* \ might escape characters inside [...] and [^...]. */
2211 {
2216 }
2217 else
2218 {
2219 /* Could be the end of the bracket expression. If it's
2220 not (i.e., when the bracket expression is `[]' so
2221 far), the ']' character bit gets set way below. */
2224 }
2226 /* If C indicates start of multibyte char, get the
2227 actual character code in C, and set the pattern
2228 pointer P to the next character boundary. */
2230 {
2231 PATUNFETCH;
2234 }
2235 /* What should we do for the character which is
2236 greater than 0x7F, but not BASE_LEADING_CODE_P?
2237 XXX */
2239 /* See if we're at the beginning of a possible character
2240 class. */
2244 {
2245 /* Leave room for the null. */
2251 /* If pattern is `[[:'. */
2255 {
2261 }
2264 /* If isn't a word bracketed by `[:' and `:]':
2265 undo the ending character, the letters, and
2266 leave the leading `:' and `[' (but set bits for
2267 them). */
2269 {
2287 /* Throw away the ] at the end of the character
2288 class. */
2294 {
2296 /* This was split into 3 if's to
2297 avoid an arbitrary limit in some compiler. */
2313 }
2315 /* Repeat the loop. */
2317 }
2318 else
2319 {
2320 c1++;
2322 PATUNFETCH;
2325 /* Because the `:' may starts the range, we
2326 can't simply set bit and repeat the loop.
2327 Instead, just set it to C and handle below. */
2329 }
2330 }
2333 {
2335 /* Discard the `-'. */
2338 /* Fetch the character which ends the range. */
2341 {
2342 PATUNFETCH;
2345 }
2349 {
2350 /* Handle a range such as \177-\377 in multibyte mode.
2351 Split that into two ranges,,
2352 the low one ending at 0237, and the high one
2353 starting at ...040. */
2357 }
2360 }
2361 else
2362 /* Range from C to C. */
2365 /* Set the range ... */
2367 /* ... into bitmap. */
2368 {
2372 /* If the start is after the end, the range is empty. */
2374 {
2377 /* Else, repeat the loop. */
2378 }
2379 else
2380 {
2382 this_char++)
2384 }
2385 }
2386 else
2387 /* ... into range table. */
2389 }
2391 /* Discard any (non)matching list bytes that are all 0 at the
2392 end of the map. Decrease the map-length byte too. */
2397 /* Build real range table from work area. */
2399 {
2403 /* Allocate space for COUNT + RANGE_TABLE. Needs two
2404 bytes for COUNT and three bytes for each character. */
2407 /* Indicate the existence of range table. */
2412 STORE_CHARACTER_AND_INCR
2414 }
2415 }
2422 else
2429 else
2436 else
2443 else
2450 else
2457 /* Do not translate the character after the \, so that we can
2458 distinguish, e.g., \B from \b, even if we normally would
2459 translate, e.g., B to b. */
2463 {
2468 handle_open:
2470 regnum++;
2473 {
2475 compile_stack_elt_t);
2479 }
2481 /* These are the values to restore when we hit end of this
2482 group. They are all relative offsets, so that if the
2483 whole pattern moves because of realloc, they will still
2484 be valid. */
2486 COMPILE_STACK_TOP.fixup_alt_jump
2491 /* We will eventually replace the 0 with the number of
2492 groups inner to this one. But do not push a
2493 start_memory for groups beyond the last one we can
2494 represent in the compiled pattern. */
2496 {
2499 }
2506 /* If we've reached MAX_REGNUM groups, then this open
2507 won't actually generate any code, so we'll have to
2508 clear pending_exact explicitly. */
2519 else
2522 handle_close:
2525 alternative for a possible future
2526 `pop_failure_jump' to pop. See comments at
2527 `push_dummy_failure' in `re_match_2'. */
2530 /* We allocated space for this jump when we assigned
2531 to `fixup_alt_jump', in the `handle_alt' case below. */
2533 }
2535 /* See similar code for backslashed left paren above. */
2539 else
2542 /* Since we just checked for an empty stack above, this
2543 ``can't happen''. */
2545 {
2546 /* We don't just want to restore into `regnum', because
2547 later groups should continue to be numbered higher,
2548 as in `(ab)c(de)' -- the second group is #2. */
2549 regnum_t this_group_regnum;
2553 fixup_alt_jump
2559 /* If we've reached MAX_REGNUM groups, then this open
2560 won't actually generate any code, so we'll have to
2561 clear pending_exact explicitly. */
2564 /* We're at the end of the group, so now we know how many
2565 groups were inside this one. */
2567 {
2574 }
2575 }
2582 handle_alt:
2586 /* Insert before the previous alternative a jump which
2587 jumps to this alternative if the former fails. */
2593 /* The alternative before this one has a jump after it
2594 which gets executed if it gets matched. Adjust that
2595 jump so it will jump to this alternative's analogous
2596 jump (put in below, which in turn will jump to the next
2597 (if any) alternative's such jump, etc.). The last such
2598 jump jumps to the correct final destination. A picture:
2599 _____ _____
2600 | | | |
2601 | v | v
2602 a | b | c
2604 If we are at `b', then fixup_alt_jump right now points to a
2605 three-byte space after `a'. We'll put in the jump, set
2606 fixup_alt_jump to right after `b', and leave behind three
2607 bytes which we'll fill in when we get to after `c'. */
2612 /* Mark and leave space for a jump after this alternative,
2613 to be filled in later either by next alternative or
2614 when know we're at the end of a series of alternatives. */
2625 /* If \{ is a literal. */
2627 /* If we're at `\{' and it's not the open-interval
2628 operator. */
2633 handle_interval:
2634 {
2635 /* If got here, then the syntax allows intervals. */
2637 /* At least (most) this many matches must be made. */
2643 {
2646 else
2648 }
2653 {
2656 }
2657 else
2658 /* Interval such as `{1}' => match exactly once. */
2663 {
2666 else
2668 }
2671 {
2675 }
2678 {
2681 else
2683 }
2685 /* We just parsed a valid interval. */
2687 /* If it's invalid to have no preceding re. */
2689 {
2694 else
2696 }
2698 /* If the upper bound is zero, don't want to succeed at
2699 all; jump from `laststart' to `b + 3', which will be
2700 the end of the buffer after we insert the jump. */
2702 {
2706 }
2708 /* Otherwise, we have a nontrivial interval. When
2709 we're all done, the pattern will look like:
2710 set_number_at <jump count> <upper bound>
2711 set_number_at <succeed_n count> <lower bound>
2712 succeed_n <after jump addr> <succeed_n count>
2713 <body of loop>
2714 jump_n <succeed_n addr> <jump count>
2715 (The upper bound and `jump_n' are omitted if
2716 `upper_bound' is 1, though.) */
2717 else
2719 more at the end of the loop. */
2724 /* Initialize lower bound of the `succeed_n', even
2725 though it will be set during matching by its
2726 attendant `set_number_at' (inserted next),
2727 because `re_compile_fastmap' needs to know.
2728 Jump to the `jump_n' we might insert below. */
2731 lower_bound);
2734 /* Code to initialize the lower bound. Insert
2735 before the `succeed_n'. The `5' is the last two
2736 bytes of this `set_number_at', plus 3 bytes of
2737 the following `succeed_n'. */
2743 append a backward jump to the `succeed_n'
2744 that starts this interval.
2746 When we've reached this during matching,
2747 we'll have matched the interval once, so
2748 jump back only `upper_bound - 1' times. */
2753 /* The location we want to set is the second
2754 parameter of the `jump_n'; that is `b-2' as
2755 an absolute address. `laststart' will be
2756 the `set_number_at' we're about to insert;
2757 `laststart+3' the number to set, the source
2758 for the relative address. But we are
2759 inserting into the middle of the pattern --
2760 so everything is getting moved up by 5.
2761 Conclusion: (b - 2) - (laststart + 3) + 5,
2762 i.e., b - laststart.
2764 We insert this at the beginning of the loop
2765 so that if we fail during matching, we'll
2766 reinitialize the bounds. */
2770 }
2771 }
2774 }
2777 unfetch_interval:
2778 /* If an invalid interval, match the characters as literals. */
2783 /* normal_char and normal_backslash need `c'. */
2787 {
2790 }
2793 #ifdef emacs
2794 /* There is no way to specify the before_dot and after_dot
2795 operators. rms says this is ok. --karl */
2872 /* Can't back reference to a subexpression if inside of it. */
2885 else
2889 normal_backslash:
2890 /* You might think it would be useful for \ to mean
2891 not to translate; but if we don't translate it
2892 it will never match anything. */
2895 }
2900 /* Expects the character in `c'. */
2901 normal_char:
2903 #ifdef emacs
2905 {
2908 /* Set P to the next character boundary. */
2910 }
2911 #endif
2912 /* If no exactn currently being built. */
2915 /* If last exactn not at current position. */
2918 /* We have only one byte following the exactn for the count. */
2921 /* If followed by a repetition operator. */
2930 {
2931 /* Start building a new exactn. */
2937 }
2939 #ifdef emacs
2941 {
2946 {
2949 }
2950 }
2951 else
2952 #endif
2953 {
2956 }
2962 /* Through the pattern now. */
2970 /* If we don't want backtracking, force success
2971 the first time we reach the end of the compiled pattern. */
2977 /* We have succeeded; set the length of the buffer. */
2980 #ifdef DEBUG
2982 {
2985 }
2988 #ifndef MATCH_MAY_ALLOCATE
2989 /* Initialize the failure stack to the largest possible stack. This
2990 isn't necessary unless we're trying to avoid calling alloca in
2991 the search and match routines. */
2992 {
2996 {
2999 #ifdef emacs
3001 fail_stack.stack
3004 else
3005 fail_stack.stack
3011 fail_stack.stack
3014 else
3015 fail_stack.stack
3020 }
3023 }
3028 \f
3029 /* Subroutines for `regex_compile'. */
3031 /* Store OP at LOC followed by two-byte integer parameter ARG. */
3033 static void
3035 re_opcode_t op;
3038 {
3041 }
3044 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
3046 static void
3048 re_opcode_t op;
3051 {
3055 }
3058 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
3059 for OP followed by two-byte integer parameter ARG. */
3061 static void
3063 re_opcode_t op;
3067 {
3075 }
3078 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
3080 static void
3082 re_opcode_t op;
3086 {
3094 }
3097 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
3098 after an alternative or a begin-subexpression. We assume there is at
3099 least one character before the ^. */
3101 static boolean
3104 reg_syntax_t syntax;
3105 {
3109 return
3110 /* After a subexpression? */
3112 /* After an alternative? */
3114 }
3117 /* The dual of at_begline_loc_p. This one is for $. We assume there is
3118 at least one character after the $, i.e., `P < PEND'. */
3120 static boolean
3124 {
3129 return
3130 /* Before a subexpression? */
3133 /* Before an alternative? */
3136 }
3139 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
3140 false if it's not. */
3142 static boolean
3144 compile_stack_type compile_stack;
3145 regnum_t regnum;
3146 {
3151 this_element--)
3156 }
3157 \f
3158 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3159 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3160 characters can start a string that matches the pattern. This fastmap
3161 is used by re_search to skip quickly over impossible starting points.
3163 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3164 area as BUFP->fastmap.
3166 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3167 the pattern buffer.
3169 Returns 0 if we succeed, -2 if an internal error. */
3171 int
3174 {
3176 #ifdef MATCH_MAY_ALLOCATE
3177 fail_stack_type fail_stack;
3178 #endif
3179 #ifndef REGEX_MALLOC
3181 #endif
3182 /* We don't push any register information onto the failure stack. */
3191 /* This holds the pointer to the failure stack, when
3192 it is allocated relocatably. */
3195 /* Assume that each path through the pattern can be null until
3196 proven otherwise. We set this false at the bottom of switch
3197 statement, to which we get only if a particular path doesn't
3198 match the empty string. */
3201 /* We aren't doing a `succeed_n' to begin with. */
3204 /* If all elements for base leading-codes in fastmap is set, this
3205 flag is set true. */
3208 /* Maximum code of simple (single byte) character. */
3219 {
3221 {
3222 /* We have reached the (effective) end of pattern. */
3224 {
3227 /* Reset for next path. */
3233 }
3234 else
3236 }
3238 /* We should never be about to go beyond the end of the pattern. */
3242 {
3244 /* I guess the idea here is to simply not bother with a fastmap
3245 if a backreference is used, since it's too hard to figure out
3246 the fastmap for the corresponding group. Setting
3247 `can_be_null' stops `re_search_2' from using the fastmap, so
3248 that is all we do. */
3254 /* Following are the cases which match a character. These end
3255 with `break'. */
3262 #ifndef emacs
3271 /* Chars beyond end of map must be allowed. */
3302 {
3303 /* Set fastmap[I] 1 where I is a base leading code of each
3304 multibyte character in the range table. */
3307 /* Make P points the range table. */
3310 /* Extract the number of ranges in range table into
3311 COUNT. */
3314 {
3315 /* Extract the start of each range. */
3319 }
3320 }
3325 /* Chars beyond end of bitmap are possible matches.
3326 All the single-byte codes can occur in multibyte buffers.
3327 So any that are not listed in the charset
3328 are possible matches, even in multibyte buffers. */
3340 /* Any character set can possibly contain a character
3341 which doesn't match the specified set of characters. */
3342 {
3343 set_fastmap_for_multibyte_characters:
3345 {
3350 }
3351 }
3356 /* All the single-byte codes can occur in multibyte buffers,
3357 and they may have word syntax. So do consider them. */
3364 /* Any character set can possibly contain a character
3365 whose syntax is `Sword'. */
3371 /* All the single-byte codes can occur in multibyte buffers,
3372 and they may not have word syntax. So do consider them. */
3379 /* Any character set can possibly contain a character
3380 whose syntax is not `Sword'. */
3383 #endif
3386 {
3389 /* `.' matches anything, except perhaps newline.
3390 Even in a multibyte buffer, it should match any
3391 conceivable byte value for the fastmap. */
3399 /* ... except perhaps newline. */
3403 /* Return if we have already set `can_be_null'; if we have,
3404 then the fastmap is irrelevant. Something's wrong here. */
3408 /* Otherwise, have to check alternative paths. */
3410 }
3412 #ifdef emacs
3419 /* This match depends on text properties. These end with
3420 aborting optimizations. */
3423 #if 0
3431 /* Any character set can possibly contain a character
3432 whose syntax is K. */
3444 /* Any character set can possibly contain a character
3445 whose syntax is not K. */
3448 #endif
3459 /* Any character set can possibly contain a character
3460 whose category is K. */
3473 /* Any character set can possibly contain a character
3474 whose category is not K. */
3478 /* All cases after this match the empty string. These end with
3479 `continue'. */
3494 #ifndef emacs
3499 #endif
3515 /* Jump backward implies we just went through the body of a
3516 loop and matched nothing. Opcode jumped to should be
3517 `on_failure_jump' or `succeed_n'. Just treat it like an
3518 ordinary jump. For a * loop, it has pushed its failure
3519 point already; if so, discard that as redundant. */
3524 p++;
3528 /* If what's on the stack is where we are now, pop it. */
3538 handle_on_failure_jump:
3541 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3542 end of the pattern. We don't want to push such a point,
3543 since when we restore it above, entering the switch will
3544 increment `p' past the end of the pattern. We don't need
3545 to push such a point since we obviously won't find any more
3546 fastmap entries beyond `pend'. Such a pattern can match
3547 the null string, though. */
3549 {
3551 {
3554 }
3555 }
3556 else
3560 {
3563 }
3569 /* Get to the number of times to succeed. */
3572 /* Increment p past the n for when k != 0. */
3575 {
3579 }
3598 /* Getting here means we have found the possible starting
3599 characters for one path of the pattern -- and that the empty
3600 string does not match. We need not follow this path further.
3601 Instead, look at the next alternative (remembered on the
3602 stack), or quit if no more. The test at the top of the loop
3603 does these things. */
3608 /* Set `can_be_null' for the last path (also the first path, if the
3609 pattern is empty). */
3612 done:
3616 \f
3617 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3618 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3619 this memory for recording register information. STARTS and ENDS
3620 must be allocated using the malloc library routine, and must each
3621 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3623 If NUM_REGS == 0, then subsequent matches should allocate their own
3624 register data.
3626 Unless this function is called, the first search or match using
3627 PATTERN_BUFFER will allocate its own register data, without
3628 freeing the old data. */
3630 void
3636 {
3638 {
3643 }
3644 else
3645 {
3649 }
3650 }
3651 \f
3652 /* Searching routines. */
3654 /* Like re_search_2, below, but only one string is specified, and
3655 doesn't let you say where to stop matching. */
3657 int
3663 {
3666 }
3668 /* End address of virtual concatenation of string. */
3669 #define STOP_ADDR_VSTRING(P) \
3670 (((P) >= size1 ? string2 + size2 : string1 + size1))
3672 /* Address of POS in the concatenation of virtual string. */
3673 #define POS_ADDR_VSTRING(POS) \
3674 (((POS) >= size1 ? string2 - size1 : string1) + (POS))
3676 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3677 virtual concatenation of STRING1 and STRING2, starting first at index
3678 STARTPOS, then at STARTPOS + 1, and so on.
3680 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3682 RANGE is how far to scan while trying to match. RANGE = 0 means try
3683 only at STARTPOS; in general, the last start tried is STARTPOS +
3684 RANGE.
3686 In REGS, return the indices of the virtual concatenation of STRING1
3687 and STRING2 that matched the entire BUFP->buffer and its contained
3688 subexpressions.
3690 Do not consider matching one past the index STOP in the virtual
3691 concatenation of STRING1 and STRING2.
3693 We return either the position in the strings at which the match was
3694 found, -1 if no match, or -2 if error (such as failure
3695 stack overflow). */
3697 int
3706 {
3714 /* Nonzero if we have to concern multibyte character. */
3717 /* Check for out-of-range STARTPOS. */
3721 /* Fix up RANGE if it might eventually take us outside
3722 the virtual concatenation of STRING1 and STRING2.
3723 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3729 /* If the search isn't to be a backwards one, don't waste time in a
3730 search for a pattern anchored at beginning of buffer. */
3732 {
3735 else
3737 }
3739 #ifdef emacs
3740 /* In a forward search for something that starts with \=.
3741 don't keep searching past point. */
3743 {
3747 }
3750 /* Update the fastmap now if not correct already. */
3755 /* See whether the pattern is anchored. */
3759 #ifdef emacs
3761 {
3766 }
3767 #endif
3769 /* Loop through the string, looking for a place to start matching. */
3771 {
3772 /* If the pattern is anchored,
3773 skip quickly past places we cannot match.
3774 We don't bother to treat startpos == 0 specially
3775 because that case doesn't repeat. */
3777 {
3783 }
3785 /* If a fastmap is supplied, skip quickly over characters that
3786 cannot be the start of a match. If the pattern can match the
3787 null string, however, we don't need to skip characters; we want
3788 the first null string. */
3790 {
3797 {
3804 /* Written out as an if-else to avoid testing `translate'
3805 inside the loop. */
3807 {
3810 {
3814 buf_charlen);
3823 }
3824 else
3828 {
3829 d++;
3830 range--;
3831 }
3832 }
3833 else
3835 {
3836 d++;
3837 range--;
3838 }
3841 }
3843 {
3855 }
3856 }
3858 /* If can't match the null string, and that's all we have left, fail. */
3865 #ifndef REGEX_MALLOC
3866 #ifdef C_ALLOCA
3868 #endif
3869 #endif
3877 advance:
3881 {
3882 /* Update STARTPOS to the next character boundary. */
3884 {
3895 }
3896 else
3897 {
3898 range--;
3899 startpos++;
3900 }
3901 }
3902 else
3903 {
3904 range++;
3905 startpos--;
3907 /* Update STARTPOS to the previous character boundary. */
3909 {
3914 /* Find the head of multibyte form. */
3918 /* Adjust it. */
3921 ;
3922 else
3923 #endif
3924 {
3930 }
3931 }
3932 }
3933 }
3936 \f
3937 /* Declarations and macros for re_match_2. */
3944 /* This converts PTR, a pointer into one of the search strings `string1'
3945 and `string2' into an offset from the beginning of that string. */
3946 #define POINTER_TO_OFFSET(ptr) \
3947 (FIRST_STRING_P (ptr) \
3948 ? ((regoff_t) ((ptr) - string1)) \
3949 : ((regoff_t) ((ptr) - string2 + size1)))
3951 /* Macros for dealing with the split strings in re_match_2. */
3953 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3955 /* Call before fetching a character with *d. This switches over to
3956 string2 if necessary. */
3957 #define PREFETCH() \
3958 while (d == dend) \
3959 { \
3961 if (dend == end_match_2) \
3962 goto fail; \
3964 d = string2; \
3965 dend = end_match_2; \
3966 }
3969 /* Test if at very beginning or at very end of the virtual concatenation
3970 of `string1' and `string2'. If only one string, it's `string2'. */
3971 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3972 #define AT_STRINGS_END(d) ((d) == end2)
3975 /* Test if D points to a character which is word-constituent. We have
3976 two special cases to check for: if past the end of string1, look at
3977 the first character in string2; and if before the beginning of
3978 string2, look at the last character in string1. */
3979 #define WORDCHAR_P(d) \
3980 (SYNTAX ((d) == end1 ? *string2 \
3981 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3982 == Sword)
3984 /* Disabled due to a compiler bug -- see comment at case wordbound */
3986 /* The comment at case wordbound is following one, but we don't use
3987 AT_WORD_BOUNDARY anymore to support multibyte form.
3989 The DEC Alpha C compiler 3.x generates incorrect code for the
3990 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
3991 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
3992 macro and introducing temporary variables works around the bug. */
3994 #if 0
3995 /* Test if the character before D and the one at D differ with respect
3996 to being word-constituent. */
3997 #define AT_WORD_BOUNDARY(d) \
3998 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3999 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
4000 #endif
4002 /* Free everything we malloc. */
4003 #ifdef MATCH_MAY_ALLOCATE
4004 #define FREE_VAR(var) if (var) { REGEX_FREE (var); var = NULL; } else
4005 #define FREE_VARIABLES() \
4006 do { \
4007 REGEX_FREE_STACK (fail_stack.stack); \
4008 FREE_VAR (regstart); \
4009 FREE_VAR (regend); \
4010 FREE_VAR (old_regstart); \
4011 FREE_VAR (old_regend); \
4012 FREE_VAR (best_regstart); \
4013 FREE_VAR (best_regend); \
4014 FREE_VAR (reg_info); \
4015 FREE_VAR (reg_dummy); \
4016 FREE_VAR (reg_info_dummy); \
4017 } while (0)
4018 #else
4022 /* These values must meet several constraints. They must not be valid
4023 register values; since we have a limit of 255 registers (because
4024 we use only one byte in the pattern for the register number), we can
4025 use numbers larger than 255. They must differ by 1, because of
4026 NUM_FAILURE_ITEMS above. And the value for the lowest register must
4027 be larger than the value for the highest register, so we do not try
4028 to actually save any registers when none are active. */
4029 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
4030 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
4031 \f
4032 /* Matching routines. */
4035 /* re_match is like re_match_2 except it takes only a single string. */
4037 int
4043 {
4052 }
4055 #ifdef emacs
4056 /* In Emacs, this is the string or buffer in which we
4057 are matching. It is used for looking up syntax properties. */
4058 Lisp_Object re_match_object;
4059 #endif
4061 /* re_match_2 matches the compiled pattern in BUFP against the
4062 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
4063 and SIZE2, respectively). We start matching at POS, and stop
4064 matching at STOP.
4066 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
4067 store offsets for the substring each group matched in REGS. See the
4068 documentation for exactly how many groups we fill.
4070 We return -1 if no match, -2 if an internal error (such as the
4071 failure stack overflowing). Otherwise, we return the length of the
4072 matched substring. */
4074 int
4082 {
4085 #ifdef emacs
4091 #endif
4101 }
4103 /* This is a separate function so that we can force an alloca cleanup
4104 afterwards. */
4105 static int
4113 {
4114 /* General temporaries. */
4118 /* Just past the end of the corresponding string. */
4121 /* Pointers into string1 and string2, just past the last characters in
4122 each to consider matching. */
4125 /* Where we are in the data, and the end of the current string. */
4128 /* Where we are in the pattern, and the end of the pattern. */
4132 /* Mark the opcode just after a start_memory, so we can test for an
4133 empty subpattern when we get to the stop_memory. */
4136 /* We use this to map every character in the string. */
4139 /* Nonzero if we have to concern multibyte character. */
4142 /* Failure point stack. Each place that can handle a failure further
4143 down the line pushes a failure point on this stack. It consists of
4144 restart, regend, and reg_info for all registers corresponding to
4145 the subexpressions we're currently inside, plus the number of such
4146 registers, and, finally, two char *'s. The first char * is where
4147 to resume scanning the pattern; the second one is where to resume
4148 scanning the strings. If the latter is zero, the failure point is
4149 a ``dummy''; if a failure happens and the failure point is a dummy,
4150 it gets discarded and the next next one is tried. */
4152 fail_stack_type fail_stack;
4153 #endif
4154 #ifdef DEBUG
4157 #endif
4159 /* This holds the pointer to the failure stack, when
4160 it is allocated relocatably. */
4163 /* We fill all the registers internally, independent of what we
4164 return, for use in backreferences. The number here includes
4165 an element for register zero. */
4168 /* The currently active registers. */
4172 /* Information on the contents of registers. These are pointers into
4173 the input strings; they record just what was matched (on this
4174 attempt) by a subexpression part of the pattern, that is, the
4175 regnum-th regstart pointer points to where in the pattern we began
4176 matching and the regnum-th regend points to right after where we
4177 stopped matching the regnum-th subexpression. (The zeroth register
4178 keeps track of what the whole pattern matches.) */
4181 #endif
4183 /* If a group that's operated upon by a repetition operator fails to
4184 match anything, then the register for its start will need to be
4185 restored because it will have been set to wherever in the string we
4186 are when we last see its open-group operator. Similarly for a
4187 register's end. */
4190 #endif
4192 /* The is_active field of reg_info helps us keep track of which (possibly
4193 nested) subexpressions we are currently in. The matched_something
4194 field of reg_info[reg_num] helps us tell whether or not we have
4195 matched any of the pattern so far this time through the reg_num-th
4196 subexpression. These two fields get reset each time through any
4197 loop their register is in. */
4200 #endif
4202 /* The following record the register info as found in the above
4203 variables when we find a match better than any we've seen before.
4204 This happens as we backtrack through the failure points, which in
4205 turn happens only if we have not yet matched the entire string. */
4209 #endif
4211 /* Logically, this is `best_regend[0]'. But we don't want to have to
4212 allocate space for that if we're not allocating space for anything
4213 else (see below). Also, we never need info about register 0 for
4214 any of the other register vectors, and it seems rather a kludge to
4215 treat `best_regend' differently than the rest. So we keep track of
4216 the end of the best match so far in a separate variable. We
4217 initialize this to NULL so that when we backtrack the first time
4218 and need to test it, it's not garbage. */
4221 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
4224 /* Used when we pop values we don't care about. */
4228 #endif
4230 #ifdef DEBUG
4231 /* Counts the total number of registers pushed. */
4233 #endif
4239 #ifdef MATCH_MAY_ALLOCATE
4240 /* Do not bother to initialize all the register variables if there are
4241 no groups in the pattern, as it takes a fair amount of time. If
4242 there are groups, we include space for register 0 (the whole
4243 pattern), even though we never use it, since it simplifies the
4244 array indexing. We should fix this. */
4246 {
4259 {
4262 }
4263 }
4264 else
4265 {
4266 /* We must initialize all our variables to NULL, so that
4267 `FREE_VARIABLES' doesn't try to free them. */
4271 }
4274 /* The starting position is bogus. */
4276 {
4279 }
4281 /* Initialize subexpression text positions to -1 to mark ones that no
4282 start_memory/stop_memory has been seen for. Also initialize the
4283 register information struct. */
4285 {
4293 }
4295 /* We move `string1' into `string2' if the latter's empty -- but not if
4296 `string1' is null. */
4298 {
4303 }
4307 /* Compute where to stop matching, within the two strings. */
4309 {
4312 }
4313 else
4314 {
4317 }
4319 /* `p' scans through the pattern as `d' scans through the data.
4320 `dend' is the end of the input string that `d' points within. `d'
4321 is advanced into the following input string whenever necessary, but
4322 this happens before fetching; therefore, at the beginning of the
4323 loop, `d' can be pointing at the end of a string, but it cannot
4324 equal `string2'. */
4326 {
4329 }
4330 else
4331 {
4334 }
4342 /* This loops over pattern commands. It exits by returning from the
4343 function if the match is complete, or it drops through if the match
4344 fails at this starting point in the input data. */
4346 {
4353 /* If we haven't matched the entire string, and we want the
4354 longest match, try backtracking. */
4356 {
4357 /* 1 if this match ends in the same string (string1 or string2)
4358 as the best previous match. */
4361 /* 1 if this match is the best seen so far. */
4362 boolean best_match_p;
4364 /* AIX compiler got confused when this was combined
4365 with the previous declaration. */
4368 else
4376 /* If exceeds best match so far, save it. */
4378 {
4385 {
4388 }
4389 }
4391 }
4393 /* If no failure points, don't restore garbage. And if
4394 last match is real best match, don't restore second
4395 best one. */
4397 {
4398 restore_best_regs:
4399 /* Restore best match. It may happen that `dend ==
4400 end_match_1' while the restored d is in string2.
4401 For example, the pattern `x.*y.*z' against the
4402 strings `x-' and `y-z-', if the two strings are
4403 not consecutive in memory. */
4411 {
4414 }
4415 }
4418 succeed_label:
4421 /* If caller wants register contents data back, do it. */
4423 {
4424 /* Have the register data arrays been allocated? */
4427 extra element beyond `num_regs' for the `-1' marker
4428 GNU code uses. */
4433 {
4436 }
4438 }
4441 allocated, reallocate them. If we need fewer, just
4442 leave it alone. */
4444 {
4449 {
4452 }
4453 }
4454 }
4455 else
4456 {
4457 /* These braces fend off a "empty body in an else-statement"
4458 warning under GCC when assert expands to nothing. */
4460 }
4462 /* Convert the pointer data in `regstart' and `regend' to
4463 indices. Register zero has to be set differently,
4464 since we haven't kept track of any info for it. */
4466 {
4471 }
4473 /* Go through the first `min (num_regs, regs->num_regs)'
4474 registers, since that is all we initialized. */
4476 {
4479 else
4480 {
4485 }
4486 }
4488 /* If the regs structure we return has more elements than
4489 were in the pattern, set the extra elements to -1. If
4490 we (re)allocated the registers, this is the case,
4491 because we always allocate enough to have at least one
4492 -1 at the end. */
4503 ? string1
4510 }
4512 /* Otherwise match next pattern command. */
4514 {
4515 /* Ignore these. Used to ignore the n of succeed_n's which
4516 currently have n == 0. */
4525 /* Match the next n pattern characters exactly. The following
4526 byte in the pattern defines n, and the n bytes after that
4527 are the characters to match. */
4532 /* This is written out as an if-else so we don't waste time
4533 testing `translate' inside the loop. */
4535 {
4536 #ifdef emacs
4538 do
4539 {
4554 }
4556 else
4558 do
4559 {
4564 d++;
4565 }
4567 }
4568 else
4569 {
4570 do
4571 {
4574 }
4576 }
4581 /* Match any character except possibly a newline or a null. */
4583 {
4591 #ifdef emacs
4594 else
4596 {
4599 }
4612 }
4618 {
4623 /* Start of actual range_table, or end of bitmap if there is no
4624 range table. */
4627 /* Nonzero if there is range table. */
4630 /* Number of ranges of range table. Not in bytes. */
4642 else
4653 /* Cast to `unsigned' instead of `unsigned char' in
4654 case the bit list is a full 32 bytes long. */
4658 }
4669 }
4672 /* The beginning of a group is represented by start_memory.
4673 The arguments are the register number in the next byte, and the
4674 number of groups inner to this one in the next. The text
4675 matched within the group is recorded (in the internal
4676 registers data structure) under the register number. */
4680 /* Find out if this group can match the empty string. */
4687 /* Save the position in the string where we were the last time
4688 we were at this open-group operator in case the group is
4689 operated upon by a repetition operator, e.g., with `(a*)*b'
4690 against `ab'; then we want to ignore where we are now in
4691 the string in case this attempt to match fails. */
4704 /* Clear this whenever we change the register activity status. */
4707 /* This is the new highest active register. */
4710 /* If nothing was active before, this is the new lowest active
4711 register. */
4715 /* Move past the register number and inner group count. */
4722 /* The stop_memory opcode represents the end of a group. Its
4723 arguments are the same as start_memory's: the register
4724 number, and the number of inner groups. */
4728 /* We need to save the string position the last time we were at
4729 this close-group operator in case the group is operated
4730 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4731 against `aba'; then we want to ignore where we are now in
4732 the string in case this attempt to match fails. */
4742 /* This register isn't active anymore. */
4745 /* Clear this whenever we change the register activity status. */
4748 /* If this was the only register active, nothing is active
4749 anymore. */
4751 {
4754 }
4755 else
4757 it isn't necessarily one less than now: consider
4758 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4759 new highest active register is 1. */
4762 r--;
4764 /* If we end up at register zero, that means that we saved
4765 the registers as the result of an `on_failure_jump', not
4766 a `start_memory', and we jumped to past the innermost
4767 `stop_memory'. For example, in ((.)*) we save
4768 registers 1 and 2 as a result of the *, but when we pop
4769 back to the second ), we are at the stop_memory 1.
4770 Thus, nothing is active. */
4772 {
4775 }
4776 else
4778 }
4780 /* If just failed to match something this time around with a
4781 group that's operated on by a repetition operator, try to
4782 force exit from the ``loop'', and restore the register
4783 information for this group that we had before trying this
4784 last match. */
4788 {
4794 {
4808 }
4811 /* If the next operation is a jump backwards in the pattern
4812 to an on_failure_jump right before the start_memory
4813 corresponding to this stop_memory, exit from the loop
4814 by forcing a failure after pushing on the stack the
4815 on_failure_jump's jump in the pattern, and d. */
4818 {
4819 /* If this group ever matched anything, then restore
4820 what its registers were before trying this last
4821 failed match, e.g., with `(a*)*b' against `ab' for
4822 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4823 against `aba' for regend[3].
4825 Also restore the registers for inner groups for,
4826 e.g., `((a*)(b*))*' against `aba' (register 3 would
4827 otherwise get trashed). */
4830 {
4835 /* Restore this and inner groups' (if any) registers. */
4837 {
4840 /* xx why this test? */
4843 }
4844 }
4845 p1++;
4850 }
4851 }
4853 /* Move past the register number and the inner group count. */
4858 /* \<digit> has been turned into a `duplicate' command which is
4859 followed by the numeric value of <digit> as the register number. */
4861 {
4866 /* Can't back reference a group which we've never matched. */
4870 /* Where in input to try to start matching. */
4873 /* Where to stop matching; if both the place to start and
4874 the place to stop matching are in the same string, then
4875 set to the place to stop, otherwise, for now have to use
4876 the end of the first string. */
4882 {
4883 /* If necessary, advance to next segment in register
4884 contents. */
4886 {
4890 /* End of string1 => advance to string2. */
4893 }
4894 /* At end of register contents => success */
4897 /* If necessary, advance to next segment in data. */
4900 /* How many characters left in this segment to match. */
4903 /* Want how many consecutive characters we can match in
4904 one shot, so, if necessary, adjust the count. */
4908 /* Compare that many; failure if mismatch, else move
4909 past them. */
4916 /* Do this because we've match some characters. */
4918 }
4919 }
4923 /* begline matches the empty string at the beginning of the string
4924 (unless `not_bol' is set in `bufp'), and, if
4925 `newline_anchor' is set, after newlines. */
4930 {
4932 }
4934 {
4936 }
4937 /* In all other cases, we fail. */
4941 /* endline is the dual of begline. */
4946 {
4948 }
4950 /* We have to ``prefetch'' the next character. */
4953 {
4955 }
4959 /* Match at the very beginning of the data. */
4967 /* Match at the very end of the data. */
4975 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4976 pushes NULL as the value for the string on the stack. Then
4977 `pop_failure_point' will keep the current value for the
4978 string, instead of restoring it. To see why, consider
4979 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4980 then the . fails against the \n. But the next thing we want
4981 to do is match the \n against the \n; if we restored the
4982 string value, we would be back at the foo.
4984 Because this is used only in specific cases, we don't need to
4985 check all the things that `on_failure_jump' does, to make
4986 sure the right things get saved on the stack. Hence we don't
4987 share its code. The only reason to push anything on the
4988 stack at all is that otherwise we would have to change
4989 `anychar's code to do something besides goto fail in this
4990 case; that seems worse than this. */
5001 /* Uses of on_failure_jump:
5003 Each alternative starts with an on_failure_jump that points
5004 to the beginning of the next alternative. Each alternative
5005 except the last ends with a jump that in effect jumps past
5006 the rest of the alternatives. (They really jump to the
5007 ending jump of the following alternative, because tensioning
5008 these jumps is a hassle.)
5010 Repeats start with an on_failure_jump that points past both
5011 the repetition text and either the following jump or
5012 pop_failure_jump back to this on_failure_jump. */
5014 on_failure:
5017 #if defined (WINDOWSNT) && defined (emacs)
5018 QUIT;
5019 #endif
5024 /* If this on_failure_jump comes right before a group (i.e.,
5025 the original * applied to a group), save the information
5026 for that group and all inner ones, so that if we fail back
5027 to this point, the group's information will be correct.
5028 For example, in \(a*\)*\1, we need the preceding group,
5029 and in \(zz\(a*\)b*\)\2, we need the inner group. */
5031 /* We can't use `p' to check ahead because we push
5032 a failure point to `p + mcnt' after we do this. */
5035 /* We need to skip no_op's before we look for the
5036 start_memory in case this on_failure_jump is happening as
5037 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
5038 against aba. */
5040 p1++;
5043 {
5044 /* We have a new highest active register now. This will
5045 get reset at the start_memory we are about to get to,
5046 but we will have saved all the registers relevant to
5047 this repetition op, as described above. */
5051 }
5058 /* A smart repeat ends with `maybe_pop_jump'.
5059 We change it to either `pop_failure_jump' or `jump'. */
5061 #if defined (WINDOWSNT) && defined (emacs)
5062 QUIT;
5063 #endif
5066 {
5069 /* Compare the beginning of the repeat with what in the
5070 pattern follows its end. If we can establish that there
5071 is nothing that they would both match, i.e., that we
5072 would have to backtrack because of (as in, e.g., `a*a')
5073 then we can change to pop_failure_jump, because we'll
5074 never have to backtrack.
5076 This is not true in the case of alternatives: in
5077 `(a|ab)*' we do need to backtrack to the `ab' alternative
5078 (e.g., if the string was `ab'). But instead of trying to
5079 detect that here, the alternative has put on a dummy
5080 failure point which is what we will end up popping. */
5082 /* Skip over open/close-group commands.
5083 If what follows this loop is a ...+ construct,
5084 look at what begins its body, since we will have to
5085 match at least one of that. */
5087 {
5095 else
5097 }
5100 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
5101 to the `maybe_finalize_jump' of this case. Examine what
5102 follows. */
5104 /* If we're at the end of the pattern, we can change. */
5106 {
5107 /* Consider what happens when matching ":\(.*\)"
5108 against ":/". I don't really understand this code
5109 yet. */
5111 DEBUG_PRINT1
5113 }
5117 {
5122 {
5128 {
5132 }
5133 }
5137 {
5144 /* Test if C is listed in charset (or charset_not)
5145 at `&p1[3]'. */
5147 {
5151 }
5155 /* `not' is equal to 1 if c would match, which means
5156 that we can't change to pop_failure_jump. */
5158 {
5161 }
5162 }
5163 }
5165 {
5167 {
5174 /* Test if C is listed in charset at `p2'. */
5176 {
5181 }
5186 {
5189 }
5190 }
5192 /* It is hard to list up all the character in charset
5193 P2 if it includes multibyte character. Give up in
5194 such case. */
5196 {
5197 /* Now, we are sure that P2 has no range table.
5198 So, for the size of bitmap in P2, `p2[1]' is
5199 enough. But P1 may have range table, so the
5200 size of bitmap table of P1 is extracted by
5201 using macro `CHARSET_BITMAP_SIZE'.
5203 Since we know that all the character listed in
5204 P2 is ASCII, it is enough to test only bitmap
5205 table of P1. */
5208 {
5210 /* We win if the charset_not inside the loop lists
5211 every character listed in the charset after. */
5219 {
5222 }
5223 }
5225 {
5227 /* We win if the charset inside the loop
5228 has no overlap with the one after the loop. */
5232 idx++)
5238 {
5241 }
5242 }
5243 }
5244 }
5245 }
5248 {
5252 }
5253 /* Note fall through. */
5256 /* The end of a simple repeat has a pop_failure_jump back to
5257 its matching on_failure_jump, where the latter will push a
5258 failure point. The pop_failure_jump takes off failure
5259 points put on by this pop_failure_jump's matching
5260 on_failure_jump; we got through the pattern to here from the
5261 matching on_failure_jump, so didn't fail. */
5263 {
5264 /* We need to pass separate storage for the lowest and
5265 highest registers, even though we don't care about the
5266 actual values. Otherwise, we will restore only one
5267 register from the stack, since lowest will == highest in
5268 `pop_failure_point'. */
5277 }
5278 /* Note fall through. */
5281 /* Unconditionally jump (without popping any failure points). */
5283 unconditional_jump:
5284 #if defined (WINDOWSNT) && defined (emacs)
5285 QUIT;
5286 #endif
5294 /* We need this opcode so we can detect where alternatives end
5295 in `group_match_null_string_p' et al. */
5301 /* Normally, the on_failure_jump pushes a failure point, which
5302 then gets popped at pop_failure_jump. We will end up at
5303 pop_failure_jump, also, and with a pattern of, say, `a+', we
5304 are skipping over the on_failure_jump, so we have to push
5305 something meaningless for pop_failure_jump to pop. */
5308 /* It doesn't matter what we push for the string here. What
5309 the code at `fail' tests is the value for the pattern. */
5314 /* At the end of an alternative, we need to push a dummy failure
5315 point in case we are followed by a `pop_failure_jump', because
5316 we don't want the failure point for the alternative to be
5317 popped. For example, matching `(a|ab)*' against `aab'
5318 requires that we match the `ab' alternative. */
5321 /* See comments just above at `dummy_failure_jump' about the
5322 two zeroes. */
5326 /* Have to succeed matching what follows at least n times.
5327 After that, handle like `on_failure_jump'. */
5333 /* Originally, this is how many times we HAVE to succeed. */
5335 {
5336 mcnt--;
5340 }
5342 {
5347 }
5354 /* Originally, this is how many times we CAN jump. */
5356 {
5357 mcnt--;
5360 }
5361 /* If don't have to jump any more, skip over the rest of command. */
5362 else
5367 {
5376 }
5381 /* We SUCCEED in one of the following cases: */
5383 /* Case 1: D is at the beginning or the end of string. */
5386 else
5387 {
5388 /* C1 is the character before D, S1 is the syntax of C1, C2
5389 is the character at D, and S2 is the syntax of C2. */
5396 #ifdef emacs
5399 #endif
5401 #ifdef emacs
5403 #endif
5408 /* Case 3: Both of S1 and S2 are Sword, and macro
5409 WORD_BOUNDARY_P (C1, C2) returns nonzero. */
5412 }
5418 /* We FAIL in one of the following cases: */
5420 /* Case 1: D is at the beginning or the end of string. */
5423 else
5424 {
5425 /* C1 is the character before D, S1 is the syntax of C1, C2
5426 is the character at D, and S2 is the syntax of C2. */
5433 #ifdef emacs
5436 #endif
5438 #ifdef emacs
5440 #endif
5445 /* Case 3: Both of S1 and S2 are Sword, and macro
5446 WORD_BOUNDARY_P (C1, C2) returns nonzero. */
5449 }
5455 /* We FAIL in one of the following cases: */
5457 /* Case 1: D is at the end of string. */
5460 else
5461 {
5462 /* C1 is the character before D, S1 is the syntax of C1, C2
5463 is the character at D, and S2 is the syntax of C2. */
5469 #ifdef emacs
5472 #endif
5475 /* Case 2: S2 is not Sword. */
5479 /* Case 3: D is not at the beginning of string ... */
5481 {
5483 #ifdef emacs
5485 #endif
5488 /* ... and S1 is Sword, and WORD_BOUNDARY_P (C1, C2)
5489 returns 0. */
5492 }
5493 }
5499 /* We FAIL in one of the following cases: */
5501 /* Case 1: D is at the beginning of string. */
5504 else
5505 {
5506 /* C1 is the character before D, S1 is the syntax of C1, C2
5507 is the character at D, and S2 is the syntax of C2. */
5513 #ifdef emacs
5516 #endif
5519 /* Case 2: S1 is not Sword. */
5523 /* Case 3: D is not at the end of string ... */
5525 {
5527 #ifdef emacs
5529 #endif
5532 /* ... and S2 is Sword, and WORD_BOUNDARY_P (C1, C2)
5533 returns 0. */
5536 }
5537 }
5540 #ifdef emacs
5567 matchsyntax:
5569 #ifdef emacs
5570 {
5573 }
5574 #endif
5575 {
5579 /* we must concern about multibyte form, ... */
5581 else
5582 /* everything should be handled as ASCII, even though it
5583 looks like multibyte form. */
5589 }
5601 matchnotsyntax:
5603 #ifdef emacs
5604 {
5607 }
5608 #endif
5609 {
5614 else
5620 }
5628 {
5633 else
5639 }
5647 {
5652 else
5658 }
5669 d++;
5678 d++;
5684 }
5688 /* We goto here if a matching operation fails. */
5689 fail:
5690 #if defined (WINDOWSNT) && defined (emacs)
5691 QUIT;
5692 #endif
5700 /* If this failure point is a dummy, try the next one. */
5704 /* If we failed to the end of the pattern, don't examine *p. */
5707 {
5710 /* If failed to a backwards jump that's part of a repetition
5711 loop, need to pop this failure point and use the next one. */
5713 {
5724 || (!is_a_jump_n
5730 }
5731 }
5735 }
5736 else
5747 \f
5748 /* Subroutine definitions for re_match_2. */
5751 /* We are passed P pointing to a register number after a start_memory.
5753 Return true if the pattern up to the corresponding stop_memory can
5754 match the empty string, and false otherwise.
5756 If we find the matching stop_memory, sets P to point to one past its number.
5757 Otherwise, sets P to an undefined byte less than or equal to END.
5759 We don't handle duplicates properly (yet). */
5761 static boolean
5765 {
5767 /* Point to after the args to the start_memory. */
5771 {
5772 /* Skip over opcodes that can match nothing, and return true or
5773 false, as appropriate, when we get to one that can't, or to the
5774 matching stop_memory. */
5777 {
5778 /* Could be either a loop or a series of alternatives. */
5780 p1++;
5783 /* If the next operation is not a jump backwards in the
5784 pattern. */
5787 {
5788 /* Go through the on_failure_jumps of the alternatives,
5789 seeing if any of the alternatives cannot match nothing.
5790 The last alternative starts with only a jump,
5791 whereas the rest start with on_failure_jump and end
5792 with a jump, e.g., here is the pattern for `a|b|c':
5794 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
5795 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
5796 /exactn/1/c
5798 So, we have to first go through the first (n-1)
5799 alternatives and then deal with the last one separately. */
5802 /* Deal with the first (n-1) alternatives, which start
5803 with an on_failure_jump (see above) that jumps to right
5804 past a jump_past_alt. */
5807 {
5808 /* `mcnt' holds how many bytes long the alternative
5809 is, including the ending `jump_past_alt' and
5810 its number. */
5813 reg_info))
5816 /* Move to right after this alternative, including the
5817 jump_past_alt. */
5820 /* Break if it's the beginning of an n-th alternative
5821 that doesn't begin with an on_failure_jump. */
5825 /* Still have to check that it's not an n-th
5826 alternative that starts with an on_failure_jump. */
5827 p1++;
5830 {
5831 /* Get to the beginning of the n-th alternative. */
5834 }
5835 }
5837 /* Deal with the last alternative: go back and get number
5838 of the `jump_past_alt' just before it. `mcnt' contains
5839 the length of the alternative. */
5859 }
5866 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5867 It expects P to be the first byte of a single alternative and END one
5868 byte past the last. The alternative can contain groups. */
5870 static boolean
5874 {
5879 {
5880 /* Skip over opcodes that can match nothing, and break when we get
5881 to one that can't. */
5884 {
5885 /* It's a loop. */
5887 p1++;
5895 }
5902 /* Deals with the ops common to group_match_null_string_p and
5903 alt_match_null_string_p.
5905 Sets P to one after the op and its arguments, if any. */
5907 static boolean
5911 {
5913 boolean ret;
5918 {
5928 #ifdef emacs
5932 #endif
5940 /* Have to set this here in case we're checking a group which
5941 contains a group and a back reference to it. */
5950 /* If this is an optimized succeed_n for zero times, make the jump. */
5955 else
5960 /* Get to the number of times to succeed. */
5965 {
5969 }
5970 else
5983 /* All other opcodes mean we cannot match the empty string. */
5985 }
5992 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5993 bytes; nonzero otherwise. */
5995 static int
5999 RE_TRANSLATE_TYPE translate;
6000 {
6006 {
6018 }
6024 }
6025 \f
6026 /* Entry points for GNU code. */
6028 /* re_compile_pattern is the GNU regular expression compiler: it
6029 compiles PATTERN (of length SIZE) and puts the result in BUFP.
6030 Returns 0 if the pattern was valid, otherwise an error string.
6032 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
6033 are set in BUFP on entry.
6035 We call regex_compile to do the actual compilation. */
6042 {
6043 reg_errcode_t ret;
6045 /* GNU code is written to assume at least RE_NREGS registers will be set
6046 (and at least one extra will be -1). */
6049 /* And GNU code determines whether or not to get register information
6050 by passing null for the REGS argument to re_match, etc., not by
6051 setting no_sub. */
6054 /* Match anchors at newline. */
6062 }
6063 \f
6064 /* Entry points compatible with 4.2 BSD regex library. We don't define
6065 them unless specifically requested. */
6067 #if defined (_REGEX_RE_COMP) || defined (_LIBC)
6069 /* BSD has one and only one pattern buffer. */
6073 #ifdef _LIBC
6074 /* Make these definitions weak in libc, so POSIX programs can redefine
6075 these names if they don't use our functions, and still use
6076 regcomp/regexec below without link errors. */
6077 weak_function
6078 #endif
6081 {
6082 reg_errcode_t ret;
6085 {
6089 }
6092 {
6095 /* CVS: Yes, we're discarding `const' here if !HAVE_LIBINTL. */
6101 /* CVS: Yes, we're discarding `const' here if !HAVE_LIBINTL. */
6103 }
6105 /* Since `re_exec' always passes NULL for the `regs' argument, we
6106 don't need to initialize the pattern buffer fields which affect it. */
6108 /* Match anchors at newlines. */
6116 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
6118 }
6121 int
6122 #ifdef _LIBC
6123 weak_function
6124 #endif
6127 {
6129 return
6131 }
6133 \f
6134 /* POSIX.2 functions. Don't define these for Emacs. */
6136 #ifndef emacs
6138 /* regcomp takes a regular expression as a string and compiles it.
6140 PREG is a regex_t *. We do not expect any fields to be initialized,
6141 since POSIX says we shouldn't. Thus, we set
6143 `buffer' to the compiled pattern;
6144 `used' to the length of the compiled pattern;
6145 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
6146 REG_EXTENDED bit in CFLAGS is set; otherwise, to
6147 RE_SYNTAX_POSIX_BASIC;
6148 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
6149 `fastmap' and `fastmap_accurate' to zero;
6150 `re_nsub' to the number of subexpressions in PATTERN.
6152 PATTERN is the address of the pattern string.
6154 CFLAGS is a series of bits which affect compilation.
6156 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
6157 use POSIX basic syntax.
6159 If REG_NEWLINE is set, then . and [^...] don't match newline.
6160 Also, regexec will try a match beginning after every newline.
6162 If REG_ICASE is set, then we considers upper- and lowercase
6163 versions of letters to be equivalent when matching.
6165 If REG_NOSUB is set, then when PREG is passed to regexec, that
6166 routine will report only success or failure, and nothing about the
6167 registers.
6169 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
6170 the return codes and their meanings.) */
6172 int
6177 {
6178 reg_errcode_t ret;
6179 unsigned syntax
6183 /* regex_compile will allocate the space for the compiled pattern. */
6188 /* Don't bother to use a fastmap when searching. This simplifies the
6189 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
6190 characters after newlines into the fastmap. This way, we just try
6191 every character. */
6195 {
6198 preg->translate
6204 /* Map uppercase characters to corresponding lowercase ones. */
6207 }
6208 else
6211 /* If REG_NEWLINE is set, newlines are treated differently. */
6216 /* It also changes the matching behavior. */
6218 }
6219 else
6224 /* POSIX says a null character in the pattern terminates it, so we
6225 can use strlen here in compiling the pattern. */
6228 /* POSIX doesn't distinguish between an unmatched open-group and an
6229 unmatched close-group: both are REG_EPAREN. */
6233 }
6236 /* regexec searches for a given pattern, specified by PREG, in the
6237 string STRING.
6239 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
6240 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
6241 least NMATCH elements, and we set them to the offsets of the
6242 corresponding matched substrings.
6244 EFLAGS specifies `execution flags' which affect matching: if
6245 REG_NOTBOL is set, then ^ does not match at the beginning of the
6246 string; if REG_NOTEOL is set, then $ does not match at the end.
6248 We return 0 if we find a match and REG_NOMATCH if not. */
6250 int
6255 regmatch_t pmatch[];
6257 {
6260 regex_t private_preg;
6269 /* The user has told us exactly how many registers to return
6270 information about, via `nmatch'. We have to pass that on to the
6271 matching routines. */
6275 {
6281 }
6283 /* Perform the searching operation. */
6288 /* Copy the register information to the POSIX structure. */
6290 {
6292 {
6296 {
6299 }
6300 }
6302 /* If we needed the temporary register info, free the space now. */
6305 }
6307 /* We want zero return to mean success, unlike `re_search'. */
6309 }
6312 /* Returns a message corresponding to an error code, ERRCODE, returned
6313 from either regcomp or regexec. We don't use PREG here. */
6315 size_t
6321 {
6327 /* Only error codes returned by the rest of the code should be passed
6328 to this routine. If we are given anything else, or if other regex
6329 code generates an invalid error code, then the program has a bug.
6330 Dump core so we can fix it. */
6338 {
6340 {
6343 }
6344 else
6346 }
6349 }
6352 /* Free dynamically allocated space used by PREG. */
6354 void
6357 {
6373 }