1 /* String search routines for GNU Emacs.
2 Copyright (C) 1985, 1986, 1987, 1993, 1994, 1997, 1998, 1999, 2001, 2002,
3 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010
4 Free Software Foundation, Inc.
6 This file is part of GNU Emacs.
8 GNU Emacs is free software: you can redistribute it and/or modify
9 it under the terms of the GNU General Public License as published by
10 the Free Software Foundation, either version 3 of the License, or
11 (at your option) any later version.
13 GNU Emacs is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */
28 #include "character.h"
30 #include "region-cache.h"
32 #include "blockinput.h"
33 #include "intervals.h"
35 #include <sys/types.h>
38 #define REGEXP_CACHE_SIZE 20
40 /* If the regexp is non-nil, then the buffer contains the compiled form
41 of that regexp, suitable for searching. */
44 struct regexp_cache
*next
;
45 Lisp_Object regexp
, whitespace_regexp
;
46 /* Syntax table for which the regexp applies. We need this because
47 of character classes. If this is t, then the compiled pattern is valid
48 for any syntax-table. */
49 Lisp_Object syntax_table
;
50 struct re_pattern_buffer buf
;
52 /* Nonzero means regexp was compiled to do full POSIX backtracking. */
56 /* The instances of that struct. */
57 struct regexp_cache searchbufs
[REGEXP_CACHE_SIZE
];
59 /* The head of the linked list; points to the most recently used buffer. */
60 struct regexp_cache
*searchbuf_head
;
63 /* Every call to re_match, etc., must pass &search_regs as the regs
64 argument unless you can show it is unnecessary (i.e., if re_match
65 is certainly going to be called again before region-around-match
68 Since the registers are now dynamically allocated, we need to make
69 sure not to refer to the Nth register before checking that it has
70 been allocated by checking search_regs.num_regs.
72 The regex code keeps track of whether it has allocated the search
73 buffer using bits in the re_pattern_buffer. This means that whenever
74 you compile a new pattern, it completely forgets whether it has
75 allocated any registers, and will allocate new registers the next
76 time you call a searching or matching function. Therefore, we need
77 to call re_set_registers after compiling a new pattern or after
78 setting the match registers, so that the regex functions will be
79 able to free or re-allocate it properly. */
80 static struct re_registers search_regs
;
82 /* The buffer in which the last search was performed, or
83 Qt if the last search was done in a string;
84 Qnil if no searching has been done yet. */
85 static Lisp_Object last_thing_searched
;
87 /* error condition signaled when regexp compile_pattern fails */
89 Lisp_Object Qinvalid_regexp
;
91 /* Error condition used for failing searches */
92 Lisp_Object Qsearch_failed
;
94 Lisp_Object Vsearch_spaces_regexp
;
96 /* If non-nil, the match data will not be changed during call to
97 searching or matching functions. This variable is for internal use
99 Lisp_Object Vinhibit_changing_match_data
;
101 static void set_search_regs (EMACS_INT
, EMACS_INT
);
102 static void save_search_regs (void);
103 static EMACS_INT
simple_search (int, unsigned char *, int, int,
104 Lisp_Object
, EMACS_INT
, EMACS_INT
,
105 EMACS_INT
, EMACS_INT
);
106 static EMACS_INT
boyer_moore (int, unsigned char *, int, int,
107 Lisp_Object
, Lisp_Object
,
108 EMACS_INT
, EMACS_INT
,
109 EMACS_INT
, EMACS_INT
, int);
110 static EMACS_INT
search_buffer (Lisp_Object
, EMACS_INT
, EMACS_INT
,
111 EMACS_INT
, EMACS_INT
, int, int,
112 Lisp_Object
, Lisp_Object
, int);
113 static void matcher_overflow (void) NO_RETURN
;
116 matcher_overflow (void)
118 error ("Stack overflow in regexp matcher");
121 /* Compile a regexp and signal a Lisp error if anything goes wrong.
122 PATTERN is the pattern to compile.
123 CP is the place to put the result.
124 TRANSLATE is a translation table for ignoring case, or nil for none.
125 REGP is the structure that says where to store the "register"
126 values that will result from matching this pattern.
127 If it is 0, we should compile the pattern not to record any
128 subexpression bounds.
129 POSIX is nonzero if we want full backtracking (POSIX style)
130 for this pattern. 0 means backtrack only enough to get a valid match.
132 The behavior also depends on Vsearch_spaces_regexp. */
135 compile_pattern_1 (struct regexp_cache
*cp
, Lisp_Object pattern
, Lisp_Object translate
, struct re_registers
*regp
, int posix
)
141 cp
->buf
.translate
= (! NILP (translate
) ? translate
: make_number (0));
143 cp
->buf
.multibyte
= STRING_MULTIBYTE (pattern
);
144 cp
->buf
.charset_unibyte
= charset_unibyte
;
145 if (STRINGP (Vsearch_spaces_regexp
))
146 cp
->whitespace_regexp
= Vsearch_spaces_regexp
;
148 cp
->whitespace_regexp
= Qnil
;
150 /* rms: I think BLOCK_INPUT is not needed here any more,
151 because regex.c defines malloc to call xmalloc.
152 Using BLOCK_INPUT here means the debugger won't run if an error occurs.
153 So let's turn it off. */
155 old
= re_set_syntax (RE_SYNTAX_EMACS
156 | (posix
? 0 : RE_NO_POSIX_BACKTRACKING
));
158 if (STRINGP (Vsearch_spaces_regexp
))
159 re_set_whitespace_regexp (SDATA (Vsearch_spaces_regexp
));
161 re_set_whitespace_regexp (NULL
);
163 val
= (char *) re_compile_pattern ((char *) SDATA (pattern
),
164 SBYTES (pattern
), &cp
->buf
);
166 /* If the compiled pattern hard codes some of the contents of the
167 syntax-table, it can only be reused with *this* syntax table. */
168 cp
->syntax_table
= cp
->buf
.used_syntax
? current_buffer
->syntax_table
: Qt
;
170 re_set_whitespace_regexp (NULL
);
175 xsignal1 (Qinvalid_regexp
, build_string (val
));
177 cp
->regexp
= Fcopy_sequence (pattern
);
180 /* Shrink each compiled regexp buffer in the cache
181 to the size actually used right now.
182 This is called from garbage collection. */
185 shrink_regexp_cache (void)
187 struct regexp_cache
*cp
;
189 for (cp
= searchbuf_head
; cp
!= 0; cp
= cp
->next
)
191 cp
->buf
.allocated
= cp
->buf
.used
;
193 = (unsigned char *) xrealloc (cp
->buf
.buffer
, cp
->buf
.used
);
197 /* Clear the regexp cache w.r.t. a particular syntax table,
198 because it was changed.
199 There is no danger of memory leak here because re_compile_pattern
200 automagically manages the memory in each re_pattern_buffer struct,
201 based on its `allocated' and `buffer' values. */
203 clear_regexp_cache (void)
207 for (i
= 0; i
< REGEXP_CACHE_SIZE
; ++i
)
208 /* It's tempting to compare with the syntax-table we've actually changed,
209 but it's not sufficient because char-table inheritance means that
210 modifying one syntax-table can change others at the same time. */
211 if (!EQ (searchbufs
[i
].syntax_table
, Qt
))
212 searchbufs
[i
].regexp
= Qnil
;
215 /* Compile a regexp if necessary, but first check to see if there's one in
217 PATTERN is the pattern to compile.
218 TRANSLATE is a translation table for ignoring case, or nil for none.
219 REGP is the structure that says where to store the "register"
220 values that will result from matching this pattern.
221 If it is 0, we should compile the pattern not to record any
222 subexpression bounds.
223 POSIX is nonzero if we want full backtracking (POSIX style)
224 for this pattern. 0 means backtrack only enough to get a valid match. */
226 struct re_pattern_buffer
*
227 compile_pattern (Lisp_Object pattern
, struct re_registers
*regp
, Lisp_Object translate
, int posix
, int multibyte
)
229 struct regexp_cache
*cp
, **cpp
;
231 for (cpp
= &searchbuf_head
; ; cpp
= &cp
->next
)
234 /* Entries are initialized to nil, and may be set to nil by
235 compile_pattern_1 if the pattern isn't valid. Don't apply
236 string accessors in those cases. However, compile_pattern_1
237 is only applied to the cache entry we pick here to reuse. So
238 nil should never appear before a non-nil entry. */
239 if (NILP (cp
->regexp
))
241 if (SCHARS (cp
->regexp
) == SCHARS (pattern
)
242 && STRING_MULTIBYTE (cp
->regexp
) == STRING_MULTIBYTE (pattern
)
243 && !NILP (Fstring_equal (cp
->regexp
, pattern
))
244 && EQ (cp
->buf
.translate
, (! NILP (translate
) ? translate
: make_number (0)))
245 && cp
->posix
== posix
246 && (EQ (cp
->syntax_table
, Qt
)
247 || EQ (cp
->syntax_table
, current_buffer
->syntax_table
))
248 && !NILP (Fequal (cp
->whitespace_regexp
, Vsearch_spaces_regexp
))
249 && cp
->buf
.charset_unibyte
== charset_unibyte
)
252 /* If we're at the end of the cache, compile into the nil cell
253 we found, or the last (least recently used) cell with a
258 compile_pattern_1 (cp
, pattern
, translate
, regp
, posix
);
263 /* When we get here, cp (aka *cpp) contains the compiled pattern,
264 either because we found it in the cache or because we just compiled it.
265 Move it to the front of the queue to mark it as most recently used. */
267 cp
->next
= searchbuf_head
;
270 /* Advise the searching functions about the space we have allocated
271 for register data. */
273 re_set_registers (&cp
->buf
, regp
, regp
->num_regs
, regp
->start
, regp
->end
);
275 /* The compiled pattern can be used both for multibyte and unibyte
276 target. But, we have to tell which the pattern is used for. */
277 cp
->buf
.target_multibyte
= multibyte
;
284 looking_at_1 (Lisp_Object string
, int posix
)
287 unsigned char *p1
, *p2
;
290 struct re_pattern_buffer
*bufp
;
292 if (running_asynch_code
)
295 /* This is so set_image_of_range_1 in regex.c can find the EQV table. */
296 XCHAR_TABLE (current_buffer
->case_canon_table
)->extras
[2]
297 = current_buffer
->case_eqv_table
;
299 CHECK_STRING (string
);
300 bufp
= compile_pattern (string
,
301 (NILP (Vinhibit_changing_match_data
)
302 ? &search_regs
: NULL
),
303 (!NILP (current_buffer
->case_fold_search
)
304 ? current_buffer
->case_canon_table
: Qnil
),
306 !NILP (current_buffer
->enable_multibyte_characters
));
309 QUIT
; /* Do a pending quit right away, to avoid paradoxical behavior */
311 /* Get pointers and sizes of the two strings
312 that make up the visible portion of the buffer. */
315 s1
= GPT_BYTE
- BEGV_BYTE
;
317 s2
= ZV_BYTE
- GPT_BYTE
;
321 s2
= ZV_BYTE
- BEGV_BYTE
;
326 s1
= ZV_BYTE
- BEGV_BYTE
;
330 re_match_object
= Qnil
;
332 i
= re_match_2 (bufp
, (char *) p1
, s1
, (char *) p2
, s2
,
334 (NILP (Vinhibit_changing_match_data
)
335 ? &search_regs
: NULL
),
336 ZV_BYTE
- BEGV_BYTE
);
342 val
= (0 <= i
? Qt
: Qnil
);
343 if (NILP (Vinhibit_changing_match_data
) && i
>= 0)
344 for (i
= 0; i
< search_regs
.num_regs
; i
++)
345 if (search_regs
.start
[i
] >= 0)
348 = BYTE_TO_CHAR (search_regs
.start
[i
] + BEGV_BYTE
);
350 = BYTE_TO_CHAR (search_regs
.end
[i
] + BEGV_BYTE
);
353 /* Set last_thing_searched only when match data is changed. */
354 if (NILP (Vinhibit_changing_match_data
))
355 XSETBUFFER (last_thing_searched
, current_buffer
);
360 DEFUN ("looking-at", Flooking_at
, Slooking_at
, 1, 1, 0,
361 doc
: /* Return t if text after point matches regular expression REGEXP.
362 This function modifies the match data that `match-beginning',
363 `match-end' and `match-data' access; save and restore the match
364 data if you want to preserve them. */)
367 return looking_at_1 (regexp
, 0);
370 DEFUN ("posix-looking-at", Fposix_looking_at
, Sposix_looking_at
, 1, 1, 0,
371 doc
: /* Return t if text after point matches regular expression REGEXP.
372 Find the longest match, in accord with Posix regular expression rules.
373 This function modifies the match data that `match-beginning',
374 `match-end' and `match-data' access; save and restore the match
375 data if you want to preserve them. */)
378 return looking_at_1 (regexp
, 1);
382 string_match_1 (Lisp_Object regexp
, Lisp_Object string
, Lisp_Object start
, int posix
)
385 struct re_pattern_buffer
*bufp
;
386 EMACS_INT pos
, pos_byte
;
389 if (running_asynch_code
)
392 CHECK_STRING (regexp
);
393 CHECK_STRING (string
);
396 pos
= 0, pos_byte
= 0;
399 int len
= SCHARS (string
);
401 CHECK_NUMBER (start
);
403 if (pos
< 0 && -pos
<= len
)
405 else if (0 > pos
|| pos
> len
)
406 args_out_of_range (string
, start
);
407 pos_byte
= string_char_to_byte (string
, pos
);
410 /* This is so set_image_of_range_1 in regex.c can find the EQV table. */
411 XCHAR_TABLE (current_buffer
->case_canon_table
)->extras
[2]
412 = current_buffer
->case_eqv_table
;
414 bufp
= compile_pattern (regexp
,
415 (NILP (Vinhibit_changing_match_data
)
416 ? &search_regs
: NULL
),
417 (!NILP (current_buffer
->case_fold_search
)
418 ? current_buffer
->case_canon_table
: Qnil
),
420 STRING_MULTIBYTE (string
));
422 re_match_object
= string
;
424 val
= re_search (bufp
, (char *) SDATA (string
),
425 SBYTES (string
), pos_byte
,
426 SBYTES (string
) - pos_byte
,
427 (NILP (Vinhibit_changing_match_data
)
428 ? &search_regs
: NULL
));
431 /* Set last_thing_searched only when match data is changed. */
432 if (NILP (Vinhibit_changing_match_data
))
433 last_thing_searched
= Qt
;
437 if (val
< 0) return Qnil
;
439 if (NILP (Vinhibit_changing_match_data
))
440 for (i
= 0; i
< search_regs
.num_regs
; i
++)
441 if (search_regs
.start
[i
] >= 0)
444 = string_byte_to_char (string
, search_regs
.start
[i
]);
446 = string_byte_to_char (string
, search_regs
.end
[i
]);
449 return make_number (string_byte_to_char (string
, val
));
452 DEFUN ("string-match", Fstring_match
, Sstring_match
, 2, 3, 0,
453 doc
: /* Return index of start of first match for REGEXP in STRING, or nil.
454 Matching ignores case if `case-fold-search' is non-nil.
455 If third arg START is non-nil, start search at that index in STRING.
456 For index of first char beyond the match, do (match-end 0).
457 `match-end' and `match-beginning' also give indices of substrings
458 matched by parenthesis constructs in the pattern.
460 You can use the function `match-string' to extract the substrings
461 matched by the parenthesis constructions in REGEXP. */)
462 (Lisp_Object regexp
, Lisp_Object string
, Lisp_Object start
)
464 return string_match_1 (regexp
, string
, start
, 0);
467 DEFUN ("posix-string-match", Fposix_string_match
, Sposix_string_match
, 2, 3, 0,
468 doc
: /* Return index of start of first match for REGEXP in STRING, or nil.
469 Find the longest match, in accord with Posix regular expression rules.
470 Case is ignored if `case-fold-search' is non-nil in the current buffer.
471 If third arg START is non-nil, start search at that index in STRING.
472 For index of first char beyond the match, do (match-end 0).
473 `match-end' and `match-beginning' also give indices of substrings
474 matched by parenthesis constructs in the pattern. */)
475 (Lisp_Object regexp
, Lisp_Object string
, Lisp_Object start
)
477 return string_match_1 (regexp
, string
, start
, 1);
480 /* Match REGEXP against STRING, searching all of STRING,
481 and return the index of the match, or negative on failure.
482 This does not clobber the match data. */
485 fast_string_match (Lisp_Object regexp
, Lisp_Object string
)
488 struct re_pattern_buffer
*bufp
;
490 bufp
= compile_pattern (regexp
, 0, Qnil
,
491 0, STRING_MULTIBYTE (string
));
493 re_match_object
= string
;
495 val
= re_search (bufp
, (char *) SDATA (string
),
502 /* Match REGEXP against STRING, searching all of STRING ignoring case,
503 and return the index of the match, or negative on failure.
504 This does not clobber the match data.
505 We assume that STRING contains single-byte characters. */
507 extern Lisp_Object Vascii_downcase_table
;
510 fast_c_string_match_ignore_case (Lisp_Object regexp
, const char *string
)
513 struct re_pattern_buffer
*bufp
;
514 int len
= strlen (string
);
516 regexp
= string_make_unibyte (regexp
);
517 re_match_object
= Qt
;
518 bufp
= compile_pattern (regexp
, 0,
519 Vascii_canon_table
, 0,
522 val
= re_search (bufp
, string
, len
, 0, len
, 0);
527 /* Like fast_string_match but ignore case. */
530 fast_string_match_ignore_case (Lisp_Object regexp
, Lisp_Object string
)
533 struct re_pattern_buffer
*bufp
;
535 bufp
= compile_pattern (regexp
, 0, Vascii_canon_table
,
536 0, STRING_MULTIBYTE (string
));
538 re_match_object
= string
;
540 val
= re_search (bufp
, (char *) SDATA (string
),
547 /* Match REGEXP against the characters after POS to LIMIT, and return
548 the number of matched characters. If STRING is non-nil, match
549 against the characters in it. In that case, POS and LIMIT are
550 indices into the string. This function doesn't modify the match
554 fast_looking_at (Lisp_Object regexp
, EMACS_INT pos
, EMACS_INT pos_byte
, EMACS_INT limit
, EMACS_INT limit_byte
, Lisp_Object string
)
557 struct re_pattern_buffer
*buf
;
558 unsigned char *p1
, *p2
;
562 if (STRINGP (string
))
565 pos_byte
= string_char_to_byte (string
, pos
);
567 limit_byte
= string_char_to_byte (string
, limit
);
571 s2
= SBYTES (string
);
572 re_match_object
= string
;
573 multibyte
= STRING_MULTIBYTE (string
);
578 pos_byte
= CHAR_TO_BYTE (pos
);
580 limit_byte
= CHAR_TO_BYTE (limit
);
581 pos_byte
-= BEGV_BYTE
;
582 limit_byte
-= BEGV_BYTE
;
584 s1
= GPT_BYTE
- BEGV_BYTE
;
586 s2
= ZV_BYTE
- GPT_BYTE
;
590 s2
= ZV_BYTE
- BEGV_BYTE
;
595 s1
= ZV_BYTE
- BEGV_BYTE
;
598 re_match_object
= Qnil
;
599 multibyte
= ! NILP (current_buffer
->enable_multibyte_characters
);
602 buf
= compile_pattern (regexp
, 0, Qnil
, 0, multibyte
);
604 len
= re_match_2 (buf
, (char *) p1
, s1
, (char *) p2
, s2
,
605 pos_byte
, NULL
, limit_byte
);
612 /* The newline cache: remembering which sections of text have no newlines. */
614 /* If the user has requested newline caching, make sure it's on.
615 Otherwise, make sure it's off.
616 This is our cheezy way of associating an action with the change of
617 state of a buffer-local variable. */
619 newline_cache_on_off (struct buffer
*buf
)
621 if (NILP (buf
->cache_long_line_scans
))
623 /* It should be off. */
624 if (buf
->newline_cache
)
626 free_region_cache (buf
->newline_cache
);
627 buf
->newline_cache
= 0;
632 /* It should be on. */
633 if (buf
->newline_cache
== 0)
634 buf
->newline_cache
= new_region_cache ();
639 /* Search for COUNT instances of the character TARGET between START and END.
641 If COUNT is positive, search forwards; END must be >= START.
642 If COUNT is negative, search backwards for the -COUNTth instance;
643 END must be <= START.
644 If COUNT is zero, do anything you please; run rogue, for all I care.
646 If END is zero, use BEGV or ZV instead, as appropriate for the
647 direction indicated by COUNT.
649 If we find COUNT instances, set *SHORTAGE to zero, and return the
650 position past the COUNTth match. Note that for reverse motion
651 this is not the same as the usual convention for Emacs motion commands.
653 If we don't find COUNT instances before reaching END, set *SHORTAGE
654 to the number of TARGETs left unfound, and return END.
656 If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do
657 except when inside redisplay. */
660 scan_buffer (register int target
, EMACS_INT start
, EMACS_INT end
, int count
, int *shortage
, int allow_quit
)
662 struct region_cache
*newline_cache
;
673 if (! end
) end
= BEGV
;
676 newline_cache_on_off (current_buffer
);
677 newline_cache
= current_buffer
->newline_cache
;
682 immediate_quit
= allow_quit
;
687 /* Our innermost scanning loop is very simple; it doesn't know
688 about gaps, buffer ends, or the newline cache. ceiling is
689 the position of the last character before the next such
690 obstacle --- the last character the dumb search loop should
692 EMACS_INT ceiling_byte
= CHAR_TO_BYTE (end
) - 1;
693 EMACS_INT start_byte
= CHAR_TO_BYTE (start
);
696 /* If we're looking for a newline, consult the newline cache
697 to see where we can avoid some scanning. */
698 if (target
== '\n' && newline_cache
)
702 while (region_cache_forward
703 (current_buffer
, newline_cache
, start_byte
, &next_change
))
704 start_byte
= next_change
;
705 immediate_quit
= allow_quit
;
707 /* START should never be after END. */
708 if (start_byte
> ceiling_byte
)
709 start_byte
= ceiling_byte
;
711 /* Now the text after start is an unknown region, and
712 next_change is the position of the next known region. */
713 ceiling_byte
= min (next_change
- 1, ceiling_byte
);
716 /* The dumb loop can only scan text stored in contiguous
717 bytes. BUFFER_CEILING_OF returns the last character
718 position that is contiguous, so the ceiling is the
719 position after that. */
720 tem
= BUFFER_CEILING_OF (start_byte
);
721 ceiling_byte
= min (tem
, ceiling_byte
);
724 /* The termination address of the dumb loop. */
725 register unsigned char *ceiling_addr
726 = BYTE_POS_ADDR (ceiling_byte
) + 1;
727 register unsigned char *cursor
728 = BYTE_POS_ADDR (start_byte
);
729 unsigned char *base
= cursor
;
731 while (cursor
< ceiling_addr
)
733 unsigned char *scan_start
= cursor
;
736 while (*cursor
!= target
&& ++cursor
< ceiling_addr
)
739 /* If we're looking for newlines, cache the fact that
740 the region from start to cursor is free of them. */
741 if (target
== '\n' && newline_cache
)
742 know_region_cache (current_buffer
, newline_cache
,
743 start_byte
+ scan_start
- base
,
744 start_byte
+ cursor
- base
);
746 /* Did we find the target character? */
747 if (cursor
< ceiling_addr
)
752 return BYTE_TO_CHAR (start_byte
+ cursor
- base
+ 1);
758 start
= BYTE_TO_CHAR (start_byte
+ cursor
- base
);
764 /* The last character to check before the next obstacle. */
765 EMACS_INT ceiling_byte
= CHAR_TO_BYTE (end
);
766 EMACS_INT start_byte
= CHAR_TO_BYTE (start
);
769 /* Consult the newline cache, if appropriate. */
770 if (target
== '\n' && newline_cache
)
774 while (region_cache_backward
775 (current_buffer
, newline_cache
, start_byte
, &next_change
))
776 start_byte
= next_change
;
777 immediate_quit
= allow_quit
;
779 /* Start should never be at or before end. */
780 if (start_byte
<= ceiling_byte
)
781 start_byte
= ceiling_byte
+ 1;
783 /* Now the text before start is an unknown region, and
784 next_change is the position of the next known region. */
785 ceiling_byte
= max (next_change
, ceiling_byte
);
788 /* Stop scanning before the gap. */
789 tem
= BUFFER_FLOOR_OF (start_byte
- 1);
790 ceiling_byte
= max (tem
, ceiling_byte
);
793 /* The termination address of the dumb loop. */
794 register unsigned char *ceiling_addr
= BYTE_POS_ADDR (ceiling_byte
);
795 register unsigned char *cursor
= BYTE_POS_ADDR (start_byte
- 1);
796 unsigned char *base
= cursor
;
798 while (cursor
>= ceiling_addr
)
800 unsigned char *scan_start
= cursor
;
802 while (*cursor
!= target
&& --cursor
>= ceiling_addr
)
805 /* If we're looking for newlines, cache the fact that
806 the region from after the cursor to start is free of them. */
807 if (target
== '\n' && newline_cache
)
808 know_region_cache (current_buffer
, newline_cache
,
809 start_byte
+ cursor
- base
,
810 start_byte
+ scan_start
- base
);
812 /* Did we find the target character? */
813 if (cursor
>= ceiling_addr
)
818 return BYTE_TO_CHAR (start_byte
+ cursor
- base
);
824 start
= BYTE_TO_CHAR (start_byte
+ cursor
- base
);
830 *shortage
= count
* direction
;
834 /* Search for COUNT instances of a line boundary, which means either a
835 newline or (if selective display enabled) a carriage return.
836 Start at START. If COUNT is negative, search backwards.
838 We report the resulting position by calling TEMP_SET_PT_BOTH.
840 If we find COUNT instances. we position after (always after,
841 even if scanning backwards) the COUNTth match, and return 0.
843 If we don't find COUNT instances before reaching the end of the
844 buffer (or the beginning, if scanning backwards), we return
845 the number of line boundaries left unfound, and position at
846 the limit we bumped up against.
848 If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do
849 except in special cases. */
852 scan_newline (EMACS_INT start
, EMACS_INT start_byte
, EMACS_INT limit
, EMACS_INT limit_byte
, register int count
, int allow_quit
)
854 int direction
= ((count
> 0) ? 1 : -1);
856 register unsigned char *cursor
;
860 register unsigned char *ceiling_addr
;
862 int old_immediate_quit
= immediate_quit
;
864 /* The code that follows is like scan_buffer
865 but checks for either newline or carriage return. */
870 start_byte
= CHAR_TO_BYTE (start
);
874 while (start_byte
< limit_byte
)
876 ceiling
= BUFFER_CEILING_OF (start_byte
);
877 ceiling
= min (limit_byte
- 1, ceiling
);
878 ceiling_addr
= BYTE_POS_ADDR (ceiling
) + 1;
879 base
= (cursor
= BYTE_POS_ADDR (start_byte
));
882 while (*cursor
!= '\n' && ++cursor
!= ceiling_addr
)
885 if (cursor
!= ceiling_addr
)
889 immediate_quit
= old_immediate_quit
;
890 start_byte
= start_byte
+ cursor
- base
+ 1;
891 start
= BYTE_TO_CHAR (start_byte
);
892 TEMP_SET_PT_BOTH (start
, start_byte
);
896 if (++cursor
== ceiling_addr
)
902 start_byte
+= cursor
- base
;
907 while (start_byte
> limit_byte
)
909 ceiling
= BUFFER_FLOOR_OF (start_byte
- 1);
910 ceiling
= max (limit_byte
, ceiling
);
911 ceiling_addr
= BYTE_POS_ADDR (ceiling
) - 1;
912 base
= (cursor
= BYTE_POS_ADDR (start_byte
- 1) + 1);
915 while (--cursor
!= ceiling_addr
&& *cursor
!= '\n')
918 if (cursor
!= ceiling_addr
)
922 immediate_quit
= old_immediate_quit
;
923 /* Return the position AFTER the match we found. */
924 start_byte
= start_byte
+ cursor
- base
+ 1;
925 start
= BYTE_TO_CHAR (start_byte
);
926 TEMP_SET_PT_BOTH (start
, start_byte
);
933 /* Here we add 1 to compensate for the last decrement
934 of CURSOR, which took it past the valid range. */
935 start_byte
+= cursor
- base
+ 1;
939 TEMP_SET_PT_BOTH (limit
, limit_byte
);
940 immediate_quit
= old_immediate_quit
;
942 return count
* direction
;
946 find_next_newline_no_quit (EMACS_INT from
, int cnt
)
948 return scan_buffer ('\n', from
, 0, cnt
, (int *) 0, 0);
951 /* Like find_next_newline, but returns position before the newline,
952 not after, and only search up to TO. This isn't just
953 find_next_newline (...)-1, because you might hit TO. */
956 find_before_next_newline (EMACS_INT from
, EMACS_INT to
, int cnt
)
959 int pos
= scan_buffer ('\n', from
, to
, cnt
, &shortage
, 1);
967 /* Subroutines of Lisp buffer search functions. */
970 search_command (Lisp_Object string
, Lisp_Object bound
, Lisp_Object noerror
, Lisp_Object count
, int direction
, int RE
, int posix
)
978 CHECK_NUMBER (count
);
982 CHECK_STRING (string
);
986 lim
= ZV
, lim_byte
= ZV_BYTE
;
988 lim
= BEGV
, lim_byte
= BEGV_BYTE
;
992 CHECK_NUMBER_COERCE_MARKER (bound
);
994 if (n
> 0 ? lim
< PT
: lim
> PT
)
995 error ("Invalid search bound (wrong side of point)");
997 lim
= ZV
, lim_byte
= ZV_BYTE
;
999 lim
= BEGV
, lim_byte
= BEGV_BYTE
;
1001 lim_byte
= CHAR_TO_BYTE (lim
);
1004 /* This is so set_image_of_range_1 in regex.c can find the EQV table. */
1005 XCHAR_TABLE (current_buffer
->case_canon_table
)->extras
[2]
1006 = current_buffer
->case_eqv_table
;
1008 np
= search_buffer (string
, PT
, PT_BYTE
, lim
, lim_byte
, n
, RE
,
1009 (!NILP (current_buffer
->case_fold_search
)
1010 ? current_buffer
->case_canon_table
1012 (!NILP (current_buffer
->case_fold_search
)
1013 ? current_buffer
->case_eqv_table
1019 xsignal1 (Qsearch_failed
, string
);
1021 if (!EQ (noerror
, Qt
))
1023 if (lim
< BEGV
|| lim
> ZV
)
1025 SET_PT_BOTH (lim
, lim_byte
);
1027 #if 0 /* This would be clean, but maybe programs depend on
1028 a value of nil here. */
1036 if (np
< BEGV
|| np
> ZV
)
1041 return make_number (np
);
1044 /* Return 1 if REGEXP it matches just one constant string. */
1047 trivial_regexp_p (Lisp_Object regexp
)
1049 int len
= SBYTES (regexp
);
1050 unsigned char *s
= SDATA (regexp
);
1055 case '.': case '*': case '+': case '?': case '[': case '^': case '$':
1062 case '|': case '(': case ')': case '`': case '\'': case 'b':
1063 case 'B': case '<': case '>': case 'w': case 'W': case 's':
1064 case 'S': case '=': case '{': case '}': case '_':
1065 case 'c': case 'C': /* for categoryspec and notcategoryspec */
1066 case '1': case '2': case '3': case '4': case '5':
1067 case '6': case '7': case '8': case '9':
1075 /* Search for the n'th occurrence of STRING in the current buffer,
1076 starting at position POS and stopping at position LIM,
1077 treating STRING as a literal string if RE is false or as
1078 a regular expression if RE is true.
1080 If N is positive, searching is forward and LIM must be greater than POS.
1081 If N is negative, searching is backward and LIM must be less than POS.
1083 Returns -x if x occurrences remain to be found (x > 0),
1084 or else the position at the beginning of the Nth occurrence
1085 (if searching backward) or the end (if searching forward).
1087 POSIX is nonzero if we want full backtracking (POSIX style)
1088 for this pattern. 0 means backtrack only enough to get a valid match. */
1090 #define TRANSLATE(out, trt, d) \
1096 temp = Faref (trt, make_number (d)); \
1097 if (INTEGERP (temp)) \
1098 out = XINT (temp); \
1107 /* Only used in search_buffer, to record the end position of the match
1108 when searching regexps and SEARCH_REGS should not be changed
1109 (i.e. Vinhibit_changing_match_data is non-nil). */
1110 static struct re_registers search_regs_1
;
1113 search_buffer (string
, pos
, pos_byte
, lim
, lim_byte
, n
,
1114 RE
, trt
, inverse_trt
, posix
)
1123 Lisp_Object inverse_trt
;
1126 int len
= SCHARS (string
);
1127 int len_byte
= SBYTES (string
);
1130 if (running_asynch_code
)
1131 save_search_regs ();
1133 /* Searching 0 times means don't move. */
1134 /* Null string is found at starting position. */
1135 if (len
== 0 || n
== 0)
1137 set_search_regs (pos_byte
, 0);
1141 if (RE
&& !(trivial_regexp_p (string
) && NILP (Vsearch_spaces_regexp
)))
1143 unsigned char *p1
, *p2
;
1145 struct re_pattern_buffer
*bufp
;
1147 bufp
= compile_pattern (string
,
1148 (NILP (Vinhibit_changing_match_data
)
1149 ? &search_regs
: &search_regs_1
),
1151 !NILP (current_buffer
->enable_multibyte_characters
));
1153 immediate_quit
= 1; /* Quit immediately if user types ^G,
1154 because letting this function finish
1155 can take too long. */
1156 QUIT
; /* Do a pending quit right away,
1157 to avoid paradoxical behavior */
1158 /* Get pointers and sizes of the two strings
1159 that make up the visible portion of the buffer. */
1162 s1
= GPT_BYTE
- BEGV_BYTE
;
1164 s2
= ZV_BYTE
- GPT_BYTE
;
1168 s2
= ZV_BYTE
- BEGV_BYTE
;
1173 s1
= ZV_BYTE
- BEGV_BYTE
;
1176 re_match_object
= Qnil
;
1181 val
= re_search_2 (bufp
, (char *) p1
, s1
, (char *) p2
, s2
,
1182 pos_byte
- BEGV_BYTE
, lim_byte
- pos_byte
,
1183 (NILP (Vinhibit_changing_match_data
)
1184 ? &search_regs
: &search_regs_1
),
1185 /* Don't allow match past current point */
1186 pos_byte
- BEGV_BYTE
);
1189 matcher_overflow ();
1193 if (NILP (Vinhibit_changing_match_data
))
1195 pos_byte
= search_regs
.start
[0] + BEGV_BYTE
;
1196 for (i
= 0; i
< search_regs
.num_regs
; i
++)
1197 if (search_regs
.start
[i
] >= 0)
1199 search_regs
.start
[i
]
1200 = BYTE_TO_CHAR (search_regs
.start
[i
] + BEGV_BYTE
);
1202 = BYTE_TO_CHAR (search_regs
.end
[i
] + BEGV_BYTE
);
1204 XSETBUFFER (last_thing_searched
, current_buffer
);
1205 /* Set pos to the new position. */
1206 pos
= search_regs
.start
[0];
1210 pos_byte
= search_regs_1
.start
[0] + BEGV_BYTE
;
1211 /* Set pos to the new position. */
1212 pos
= BYTE_TO_CHAR (search_regs_1
.start
[0] + BEGV_BYTE
);
1225 val
= re_search_2 (bufp
, (char *) p1
, s1
, (char *) p2
, s2
,
1226 pos_byte
- BEGV_BYTE
, lim_byte
- pos_byte
,
1227 (NILP (Vinhibit_changing_match_data
)
1228 ? &search_regs
: &search_regs_1
),
1229 lim_byte
- BEGV_BYTE
);
1232 matcher_overflow ();
1236 if (NILP (Vinhibit_changing_match_data
))
1238 pos_byte
= search_regs
.end
[0] + BEGV_BYTE
;
1239 for (i
= 0; i
< search_regs
.num_regs
; i
++)
1240 if (search_regs
.start
[i
] >= 0)
1242 search_regs
.start
[i
]
1243 = BYTE_TO_CHAR (search_regs
.start
[i
] + BEGV_BYTE
);
1245 = BYTE_TO_CHAR (search_regs
.end
[i
] + BEGV_BYTE
);
1247 XSETBUFFER (last_thing_searched
, current_buffer
);
1248 pos
= search_regs
.end
[0];
1252 pos_byte
= search_regs_1
.end
[0] + BEGV_BYTE
;
1253 pos
= BYTE_TO_CHAR (search_regs_1
.end
[0] + BEGV_BYTE
);
1266 else /* non-RE case */
1268 unsigned char *raw_pattern
, *pat
;
1269 int raw_pattern_size
;
1270 int raw_pattern_size_byte
;
1271 unsigned char *patbuf
;
1272 int multibyte
= !NILP (current_buffer
->enable_multibyte_characters
);
1273 unsigned char *base_pat
;
1274 /* Set to positive if we find a non-ASCII char that need
1275 translation. Otherwise set to zero later. */
1277 int boyer_moore_ok
= 1;
1279 /* MULTIBYTE says whether the text to be searched is multibyte.
1280 We must convert PATTERN to match that, or we will not really
1281 find things right. */
1283 if (multibyte
== STRING_MULTIBYTE (string
))
1285 raw_pattern
= (unsigned char *) SDATA (string
);
1286 raw_pattern_size
= SCHARS (string
);
1287 raw_pattern_size_byte
= SBYTES (string
);
1291 raw_pattern_size
= SCHARS (string
);
1292 raw_pattern_size_byte
1293 = count_size_as_multibyte (SDATA (string
),
1295 raw_pattern
= (unsigned char *) alloca (raw_pattern_size_byte
+ 1);
1296 copy_text (SDATA (string
), raw_pattern
,
1297 SCHARS (string
), 0, 1);
1301 /* Converting multibyte to single-byte.
1303 ??? Perhaps this conversion should be done in a special way
1304 by subtracting nonascii-insert-offset from each non-ASCII char,
1305 so that only the multibyte chars which really correspond to
1306 the chosen single-byte character set can possibly match. */
1307 raw_pattern_size
= SCHARS (string
);
1308 raw_pattern_size_byte
= SCHARS (string
);
1309 raw_pattern
= (unsigned char *) alloca (raw_pattern_size
+ 1);
1310 copy_text (SDATA (string
), raw_pattern
,
1311 SBYTES (string
), 1, 0);
1314 /* Copy and optionally translate the pattern. */
1315 len
= raw_pattern_size
;
1316 len_byte
= raw_pattern_size_byte
;
1317 patbuf
= (unsigned char *) alloca (len
* MAX_MULTIBYTE_LENGTH
);
1319 base_pat
= raw_pattern
;
1322 /* Fill patbuf by translated characters in STRING while
1323 checking if we can use boyer-moore search. If TRT is
1324 non-nil, we can use boyer-moore search only if TRT can be
1325 represented by the byte array of 256 elements. For that,
1326 all non-ASCII case-equivalents of all case-senstive
1327 characters in STRING must belong to the same charset and
1332 unsigned char str_base
[MAX_MULTIBYTE_LENGTH
], *str
;
1333 int c
, translated
, inverse
;
1334 int in_charlen
, charlen
;
1336 /* If we got here and the RE flag is set, it's because we're
1337 dealing with a regexp known to be trivial, so the backslash
1338 just quotes the next character. */
1339 if (RE
&& *base_pat
== '\\')
1347 c
= STRING_CHAR_AND_LENGTH (base_pat
, in_charlen
);
1352 charlen
= in_charlen
;
1356 /* Translate the character. */
1357 TRANSLATE (translated
, trt
, c
);
1358 charlen
= CHAR_STRING (translated
, str_base
);
1361 /* Check if C has any other case-equivalents. */
1362 TRANSLATE (inverse
, inverse_trt
, c
);
1363 /* If so, check if we can use boyer-moore. */
1364 if (c
!= inverse
&& boyer_moore_ok
)
1366 /* Check if all equivalents belong to the same
1367 group of characters. Note that the check of C
1368 itself is done by the last iteration. */
1369 int this_char_base
= -1;
1371 while (boyer_moore_ok
)
1373 if (ASCII_BYTE_P (inverse
))
1375 if (this_char_base
> 0)
1380 else if (CHAR_BYTE8_P (inverse
))
1381 /* Boyer-moore search can't handle a
1382 translation of an eight-bit
1385 else if (this_char_base
< 0)
1387 this_char_base
= inverse
& ~0x3F;
1389 char_base
= this_char_base
;
1390 else if (this_char_base
!= char_base
)
1393 else if ((inverse
& ~0x3F) != this_char_base
)
1397 TRANSLATE (inverse
, inverse_trt
, inverse
);
1402 /* Store this character into the translated pattern. */
1403 memcpy (pat
, str
, charlen
);
1405 base_pat
+= in_charlen
;
1406 len_byte
-= in_charlen
;
1409 /* If char_base is still negative we didn't find any translated
1410 non-ASCII characters. */
1416 /* Unibyte buffer. */
1422 /* If we got here and the RE flag is set, it's because we're
1423 dealing with a regexp known to be trivial, so the backslash
1424 just quotes the next character. */
1425 if (RE
&& *base_pat
== '\\')
1432 TRANSLATE (translated
, trt
, c
);
1433 *pat
++ = translated
;
1437 len_byte
= pat
- patbuf
;
1438 len
= raw_pattern_size
;
1439 pat
= base_pat
= patbuf
;
1442 return boyer_moore (n
, pat
, len
, len_byte
, trt
, inverse_trt
,
1443 pos
, pos_byte
, lim
, lim_byte
,
1446 return simple_search (n
, pat
, len
, len_byte
, trt
,
1447 pos
, pos_byte
, lim
, lim_byte
);
1451 /* Do a simple string search N times for the string PAT,
1452 whose length is LEN/LEN_BYTE,
1453 from buffer position POS/POS_BYTE until LIM/LIM_BYTE.
1454 TRT is the translation table.
1456 Return the character position where the match is found.
1457 Otherwise, if M matches remained to be found, return -M.
1459 This kind of search works regardless of what is in PAT and
1460 regardless of what is in TRT. It is used in cases where
1461 boyer_moore cannot work. */
1464 simple_search (int n
, unsigned char *pat
, int len
, int len_byte
, Lisp_Object trt
, EMACS_INT pos
, EMACS_INT pos_byte
, EMACS_INT lim
, EMACS_INT lim_byte
)
1466 int multibyte
= ! NILP (current_buffer
->enable_multibyte_characters
);
1467 int forward
= n
> 0;
1468 /* Number of buffer bytes matched. Note that this may be different
1469 from len_byte in a multibyte buffer. */
1472 if (lim
> pos
&& multibyte
)
1477 /* Try matching at position POS. */
1478 EMACS_INT this_pos
= pos
;
1479 EMACS_INT this_pos_byte
= pos_byte
;
1481 unsigned char *p
= pat
;
1482 if (pos
+ len
> lim
|| pos_byte
+ len_byte
> lim_byte
)
1485 while (this_len
> 0)
1487 int charlen
, buf_charlen
;
1490 pat_ch
= STRING_CHAR_AND_LENGTH (p
, charlen
);
1491 buf_ch
= STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte
),
1493 TRANSLATE (buf_ch
, trt
, buf_ch
);
1495 if (buf_ch
!= pat_ch
)
1501 this_pos_byte
+= buf_charlen
;
1507 match_byte
= this_pos_byte
- pos_byte
;
1509 pos_byte
+= match_byte
;
1513 INC_BOTH (pos
, pos_byte
);
1523 /* Try matching at position POS. */
1524 EMACS_INT this_pos
= pos
;
1526 unsigned char *p
= pat
;
1528 if (pos
+ len
> lim
)
1531 while (this_len
> 0)
1534 int buf_ch
= FETCH_BYTE (this_pos
);
1535 TRANSLATE (buf_ch
, trt
, buf_ch
);
1537 if (buf_ch
!= pat_ch
)
1556 /* Backwards search. */
1557 else if (lim
< pos
&& multibyte
)
1562 /* Try matching at position POS. */
1563 EMACS_INT this_pos
= pos
;
1564 EMACS_INT this_pos_byte
= pos_byte
;
1566 const unsigned char *p
= pat
+ len_byte
;
1568 if (this_pos
- len
< lim
|| (pos_byte
- len_byte
) < lim_byte
)
1571 while (this_len
> 0)
1576 DEC_BOTH (this_pos
, this_pos_byte
);
1577 PREV_CHAR_BOUNDARY (p
, pat
);
1578 pat_ch
= STRING_CHAR (p
);
1579 buf_ch
= STRING_CHAR (BYTE_POS_ADDR (this_pos_byte
));
1580 TRANSLATE (buf_ch
, trt
, buf_ch
);
1582 if (buf_ch
!= pat_ch
)
1590 match_byte
= pos_byte
- this_pos_byte
;
1592 pos_byte
= this_pos_byte
;
1596 DEC_BOTH (pos
, pos_byte
);
1606 /* Try matching at position POS. */
1607 EMACS_INT this_pos
= pos
- len
;
1609 unsigned char *p
= pat
;
1614 while (this_len
> 0)
1617 int buf_ch
= FETCH_BYTE (this_pos
);
1618 TRANSLATE (buf_ch
, trt
, buf_ch
);
1620 if (buf_ch
!= pat_ch
)
1643 set_search_regs ((multibyte
? pos_byte
: pos
) - match_byte
, match_byte
);
1645 set_search_regs (multibyte
? pos_byte
: pos
, match_byte
);
1655 /* Do Boyer-Moore search N times for the string BASE_PAT,
1656 whose length is LEN/LEN_BYTE,
1657 from buffer position POS/POS_BYTE until LIM/LIM_BYTE.
1658 DIRECTION says which direction we search in.
1659 TRT and INVERSE_TRT are translation tables.
1660 Characters in PAT are already translated by TRT.
1662 This kind of search works if all the characters in BASE_PAT that
1663 have nontrivial translation are the same aside from the last byte.
1664 This makes it possible to translate just the last byte of a
1665 character, and do so after just a simple test of the context.
1666 CHAR_BASE is nonzero if there is such a non-ASCII character.
1668 If that criterion is not satisfied, do not call this function. */
1671 boyer_moore (n
, base_pat
, len
, len_byte
, trt
, inverse_trt
,
1672 pos
, pos_byte
, lim
, lim_byte
, char_base
)
1674 unsigned char *base_pat
;
1677 Lisp_Object inverse_trt
;
1678 EMACS_INT pos
, pos_byte
;
1679 EMACS_INT lim
, lim_byte
;
1682 int direction
= ((n
> 0) ? 1 : -1);
1683 register int dirlen
;
1685 int stride_for_teases
= 0;
1687 register unsigned char *cursor
, *p_limit
;
1689 unsigned char *pat
, *pat_end
;
1690 int multibyte
= ! NILP (current_buffer
->enable_multibyte_characters
);
1692 unsigned char simple_translate
[0400];
1693 /* These are set to the preceding bytes of a byte to be translated
1694 if char_base is nonzero. As the maximum byte length of a
1695 multibyte character is 5, we have to check at most four previous
1697 int translate_prev_byte1
= 0;
1698 int translate_prev_byte2
= 0;
1699 int translate_prev_byte3
= 0;
1700 int translate_prev_byte4
= 0;
1702 /* The general approach is that we are going to maintain that we know
1703 the first (closest to the present position, in whatever direction
1704 we're searching) character that could possibly be the last
1705 (furthest from present position) character of a valid match. We
1706 advance the state of our knowledge by looking at that character
1707 and seeing whether it indeed matches the last character of the
1708 pattern. If it does, we take a closer look. If it does not, we
1709 move our pointer (to putative last characters) as far as is
1710 logically possible. This amount of movement, which I call a
1711 stride, will be the length of the pattern if the actual character
1712 appears nowhere in the pattern, otherwise it will be the distance
1713 from the last occurrence of that character to the end of the
1714 pattern. If the amount is zero we have a possible match. */
1716 /* Here we make a "mickey mouse" BM table. The stride of the search
1717 is determined only by the last character of the putative match.
1718 If that character does not match, we will stride the proper
1719 distance to propose a match that superimposes it on the last
1720 instance of a character that matches it (per trt), or misses
1721 it entirely if there is none. */
1723 dirlen
= len_byte
* direction
;
1725 /* Record position after the end of the pattern. */
1726 pat_end
= base_pat
+ len_byte
;
1727 /* BASE_PAT points to a character that we start scanning from.
1728 It is the first character in a forward search,
1729 the last character in a backward search. */
1731 base_pat
= pat_end
- 1;
1733 /* A character that does not appear in the pattern induces a
1734 stride equal to the pattern length. */
1735 for (i
= 0; i
< 0400; i
++)
1738 /* We use this for translation, instead of TRT itself.
1739 We fill this in to handle the characters that actually
1740 occur in the pattern. Others don't matter anyway! */
1741 for (i
= 0; i
< 0400; i
++)
1742 simple_translate
[i
] = i
;
1746 /* Setup translate_prev_byte1/2/3/4 from CHAR_BASE. Only a
1747 byte following them are the target of translation. */
1748 unsigned char str
[MAX_MULTIBYTE_LENGTH
];
1749 int len
= CHAR_STRING (char_base
, str
);
1751 translate_prev_byte1
= str
[len
- 2];
1754 translate_prev_byte2
= str
[len
- 3];
1757 translate_prev_byte3
= str
[len
- 4];
1759 translate_prev_byte4
= str
[len
- 5];
1767 unsigned char *ptr
= base_pat
+ i
;
1771 /* If the byte currently looking at is the last of a
1772 character to check case-equivalents, set CH to that
1773 character. An ASCII character and a non-ASCII character
1774 matching with CHAR_BASE are to be checked. */
1777 if (ASCII_BYTE_P (*ptr
) || ! multibyte
)
1780 && ((pat_end
- ptr
) == 1 || CHAR_HEAD_P (ptr
[1])))
1782 unsigned char *charstart
= ptr
- 1;
1784 while (! (CHAR_HEAD_P (*charstart
)))
1786 ch
= STRING_CHAR (charstart
);
1787 if (char_base
!= (ch
& ~0x3F))
1792 j
= (ch
& 0x3F) | 0200;
1797 stride_for_teases
= BM_tab
[j
];
1799 BM_tab
[j
] = dirlen
- i
;
1800 /* A translation table is accompanied by its inverse -- see */
1801 /* comment following downcase_table for details */
1804 int starting_ch
= ch
;
1809 TRANSLATE (ch
, inverse_trt
, ch
);
1811 j
= (ch
& 0x3F) | 0200;
1815 /* For all the characters that map into CH,
1816 set up simple_translate to map the last byte
1818 simple_translate
[j
] = starting_j
;
1819 if (ch
== starting_ch
)
1821 BM_tab
[j
] = dirlen
- i
;
1830 stride_for_teases
= BM_tab
[j
];
1831 BM_tab
[j
] = dirlen
- i
;
1833 /* stride_for_teases tells how much to stride if we get a
1834 match on the far character but are subsequently
1835 disappointed, by recording what the stride would have been
1836 for that character if the last character had been
1839 pos_byte
+= dirlen
- ((direction
> 0) ? direction
: 0);
1840 /* loop invariant - POS_BYTE points at where last char (first
1841 char if reverse) of pattern would align in a possible match. */
1845 unsigned char *tail_end_ptr
;
1847 /* It's been reported that some (broken) compiler thinks that
1848 Boolean expressions in an arithmetic context are unsigned.
1849 Using an explicit ?1:0 prevents this. */
1850 if ((lim_byte
- pos_byte
- ((direction
> 0) ? 1 : 0)) * direction
1852 return (n
* (0 - direction
));
1853 /* First we do the part we can by pointers (maybe nothing) */
1856 limit
= pos_byte
- dirlen
+ direction
;
1859 limit
= BUFFER_CEILING_OF (limit
);
1860 /* LIMIT is now the last (not beyond-last!) value POS_BYTE
1861 can take on without hitting edge of buffer or the gap. */
1862 limit
= min (limit
, pos_byte
+ 20000);
1863 limit
= min (limit
, lim_byte
- 1);
1867 limit
= BUFFER_FLOOR_OF (limit
);
1868 /* LIMIT is now the last (not beyond-last!) value POS_BYTE
1869 can take on without hitting edge of buffer or the gap. */
1870 limit
= max (limit
, pos_byte
- 20000);
1871 limit
= max (limit
, lim_byte
);
1873 tail_end
= BUFFER_CEILING_OF (pos_byte
) + 1;
1874 tail_end_ptr
= BYTE_POS_ADDR (tail_end
);
1876 if ((limit
- pos_byte
) * direction
> 20)
1880 p_limit
= BYTE_POS_ADDR (limit
);
1881 p2
= (cursor
= BYTE_POS_ADDR (pos_byte
));
1882 /* In this loop, pos + cursor - p2 is the surrogate for pos. */
1883 while (1) /* use one cursor setting as long as i can */
1885 if (direction
> 0) /* worth duplicating */
1887 while (cursor
<= p_limit
)
1889 if (BM_tab
[*cursor
] == 0)
1891 cursor
+= BM_tab
[*cursor
];
1896 while (cursor
>= p_limit
)
1898 if (BM_tab
[*cursor
] == 0)
1900 cursor
+= BM_tab
[*cursor
];
1903 /* If you are here, cursor is beyond the end of the
1904 searched region. You fail to match within the
1905 permitted region and would otherwise try a character
1906 beyond that region. */
1910 i
= dirlen
- direction
;
1913 while ((i
-= direction
) + direction
!= 0)
1916 cursor
-= direction
;
1917 /* Translate only the last byte of a character. */
1919 || ((cursor
== tail_end_ptr
1920 || CHAR_HEAD_P (cursor
[1]))
1921 && (CHAR_HEAD_P (cursor
[0])
1922 /* Check if this is the last byte of
1923 a translable character. */
1924 || (translate_prev_byte1
== cursor
[-1]
1925 && (CHAR_HEAD_P (translate_prev_byte1
)
1926 || (translate_prev_byte2
== cursor
[-2]
1927 && (CHAR_HEAD_P (translate_prev_byte2
)
1928 || (translate_prev_byte3
== cursor
[-3]))))))))
1929 ch
= simple_translate
[*cursor
];
1938 while ((i
-= direction
) + direction
!= 0)
1940 cursor
-= direction
;
1941 if (pat
[i
] != *cursor
)
1945 cursor
+= dirlen
- i
- direction
; /* fix cursor */
1946 if (i
+ direction
== 0)
1948 EMACS_INT position
, start
, end
;
1950 cursor
-= direction
;
1952 position
= pos_byte
+ cursor
- p2
+ ((direction
> 0)
1953 ? 1 - len_byte
: 0);
1954 set_search_regs (position
, len_byte
);
1956 if (NILP (Vinhibit_changing_match_data
))
1958 start
= search_regs
.start
[0];
1959 end
= search_regs
.end
[0];
1962 /* If Vinhibit_changing_match_data is non-nil,
1963 search_regs will not be changed. So let's
1964 compute start and end here. */
1966 start
= BYTE_TO_CHAR (position
);
1967 end
= BYTE_TO_CHAR (position
+ len_byte
);
1970 if ((n
-= direction
) != 0)
1971 cursor
+= dirlen
; /* to resume search */
1973 return direction
> 0 ? end
: start
;
1976 cursor
+= stride_for_teases
; /* <sigh> we lose - */
1978 pos_byte
+= cursor
- p2
;
1981 /* Now we'll pick up a clump that has to be done the hard
1982 way because it covers a discontinuity. */
1984 limit
= ((direction
> 0)
1985 ? BUFFER_CEILING_OF (pos_byte
- dirlen
+ 1)
1986 : BUFFER_FLOOR_OF (pos_byte
- dirlen
- 1));
1987 limit
= ((direction
> 0)
1988 ? min (limit
+ len_byte
, lim_byte
- 1)
1989 : max (limit
- len_byte
, lim_byte
));
1990 /* LIMIT is now the last value POS_BYTE can have
1991 and still be valid for a possible match. */
1994 /* This loop can be coded for space rather than
1995 speed because it will usually run only once.
1996 (the reach is at most len + 21, and typically
1997 does not exceed len). */
1998 while ((limit
- pos_byte
) * direction
>= 0)
2000 int ch
= FETCH_BYTE (pos_byte
);
2001 if (BM_tab
[ch
] == 0)
2003 pos_byte
+= BM_tab
[ch
];
2005 break; /* ran off the end */
2008 /* Found what might be a match. */
2009 i
= dirlen
- direction
;
2010 while ((i
-= direction
) + direction
!= 0)
2014 pos_byte
-= direction
;
2015 ptr
= BYTE_POS_ADDR (pos_byte
);
2016 /* Translate only the last byte of a character. */
2018 || ((ptr
== tail_end_ptr
2019 || CHAR_HEAD_P (ptr
[1]))
2020 && (CHAR_HEAD_P (ptr
[0])
2021 /* Check if this is the last byte of a
2022 translable character. */
2023 || (translate_prev_byte1
== ptr
[-1]
2024 && (CHAR_HEAD_P (translate_prev_byte1
)
2025 || (translate_prev_byte2
== ptr
[-2]
2026 && (CHAR_HEAD_P (translate_prev_byte2
)
2027 || translate_prev_byte3
== ptr
[-3])))))))
2028 ch
= simple_translate
[*ptr
];
2034 /* Above loop has moved POS_BYTE part or all the way
2035 back to the first pos (last pos if reverse).
2036 Set it once again at the last (first if reverse) char. */
2037 pos_byte
+= dirlen
- i
- direction
;
2038 if (i
+ direction
== 0)
2040 EMACS_INT position
, start
, end
;
2041 pos_byte
-= direction
;
2043 position
= pos_byte
+ ((direction
> 0) ? 1 - len_byte
: 0);
2044 set_search_regs (position
, len_byte
);
2046 if (NILP (Vinhibit_changing_match_data
))
2048 start
= search_regs
.start
[0];
2049 end
= search_regs
.end
[0];
2052 /* If Vinhibit_changing_match_data is non-nil,
2053 search_regs will not be changed. So let's
2054 compute start and end here. */
2056 start
= BYTE_TO_CHAR (position
);
2057 end
= BYTE_TO_CHAR (position
+ len_byte
);
2060 if ((n
-= direction
) != 0)
2061 pos_byte
+= dirlen
; /* to resume search */
2063 return direction
> 0 ? end
: start
;
2066 pos_byte
+= stride_for_teases
;
2069 /* We have done one clump. Can we continue? */
2070 if ((lim_byte
- pos_byte
) * direction
< 0)
2071 return ((0 - n
) * direction
);
2073 return BYTE_TO_CHAR (pos_byte
);
2076 /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES
2077 for the overall match just found in the current buffer.
2078 Also clear out the match data for registers 1 and up. */
2081 set_search_regs (EMACS_INT beg_byte
, EMACS_INT nbytes
)
2085 if (!NILP (Vinhibit_changing_match_data
))
2088 /* Make sure we have registers in which to store
2089 the match position. */
2090 if (search_regs
.num_regs
== 0)
2092 search_regs
.start
= (regoff_t
*) xmalloc (2 * sizeof (regoff_t
));
2093 search_regs
.end
= (regoff_t
*) xmalloc (2 * sizeof (regoff_t
));
2094 search_regs
.num_regs
= 2;
2097 /* Clear out the other registers. */
2098 for (i
= 1; i
< search_regs
.num_regs
; i
++)
2100 search_regs
.start
[i
] = -1;
2101 search_regs
.end
[i
] = -1;
2104 search_regs
.start
[0] = BYTE_TO_CHAR (beg_byte
);
2105 search_regs
.end
[0] = BYTE_TO_CHAR (beg_byte
+ nbytes
);
2106 XSETBUFFER (last_thing_searched
, current_buffer
);
2109 /* Given STRING, a string of words separated by word delimiters,
2110 compute a regexp that matches those exact words separated by
2111 arbitrary punctuation. If LAX is nonzero, the end of the string
2112 need not match a word boundary unless it ends in whitespace. */
2115 wordify (Lisp_Object string
, int lax
)
2117 register unsigned char *p
, *o
;
2118 register int i
, i_byte
, len
, punct_count
= 0, word_count
= 0;
2121 int adjust
, whitespace_at_end
;
2123 CHECK_STRING (string
);
2125 len
= SCHARS (string
);
2127 for (i
= 0, i_byte
= 0; i
< len
; )
2131 FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c
, string
, i
, i_byte
);
2133 if (SYNTAX (c
) != Sword
)
2136 if (i
> 0 && SYNTAX (prev_c
) == Sword
)
2143 if (SYNTAX (prev_c
) == Sword
)
2146 whitespace_at_end
= 0;
2149 whitespace_at_end
= 1;
2152 return empty_unibyte_string
;
2154 adjust
= - punct_count
+ 5 * (word_count
- 1)
2155 + ((lax
&& !whitespace_at_end
) ? 2 : 4);
2156 if (STRING_MULTIBYTE (string
))
2157 val
= make_uninit_multibyte_string (len
+ adjust
,
2161 val
= make_uninit_string (len
+ adjust
);
2168 for (i
= 0, i_byte
= 0; i
< len
; )
2171 int i_byte_orig
= i_byte
;
2173 FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c
, string
, i
, i_byte
);
2175 if (SYNTAX (c
) == Sword
)
2177 memcpy (o
, SDATA (string
) + i_byte_orig
, i_byte
- i_byte_orig
);
2178 o
+= i_byte
- i_byte_orig
;
2180 else if (i
> 0 && SYNTAX (prev_c
) == Sword
&& --word_count
)
2192 if (!lax
|| whitespace_at_end
)
2201 DEFUN ("search-backward", Fsearch_backward
, Ssearch_backward
, 1, 4,
2202 "MSearch backward: ",
2203 doc
: /* Search backward from point for STRING.
2204 Set point to the beginning of the occurrence found, and return point.
2205 An optional second argument bounds the search; it is a buffer position.
2206 The match found must not extend before that position.
2207 Optional third argument, if t, means if fail just return nil (no error).
2208 If not nil and not t, position at limit of search and return nil.
2209 Optional fourth argument is repeat count--search for successive occurrences.
2211 Search case-sensitivity is determined by the value of the variable
2212 `case-fold-search', which see.
2214 See also the functions `match-beginning', `match-end' and `replace-match'. */)
2215 (Lisp_Object string
, Lisp_Object bound
, Lisp_Object noerror
, Lisp_Object count
)
2217 return search_command (string
, bound
, noerror
, count
, -1, 0, 0);
2220 DEFUN ("search-forward", Fsearch_forward
, Ssearch_forward
, 1, 4, "MSearch: ",
2221 doc
: /* Search forward from point for STRING.
2222 Set point to the end of the occurrence found, and return point.
2223 An optional second argument bounds the search; it is a buffer position.
2224 The match found must not extend after that position. A value of nil is
2225 equivalent to (point-max).
2226 Optional third argument, if t, means if fail just return nil (no error).
2227 If not nil and not t, move to limit of search and return nil.
2228 Optional fourth argument is repeat count--search for successive occurrences.
2230 Search case-sensitivity is determined by the value of the variable
2231 `case-fold-search', which see.
2233 See also the functions `match-beginning', `match-end' and `replace-match'. */)
2234 (Lisp_Object string
, Lisp_Object bound
, Lisp_Object noerror
, Lisp_Object count
)
2236 return search_command (string
, bound
, noerror
, count
, 1, 0, 0);
2239 DEFUN ("word-search-backward", Fword_search_backward
, Sword_search_backward
, 1, 4,
2240 "sWord search backward: ",
2241 doc
: /* Search backward from point for STRING, ignoring differences in punctuation.
2242 Set point to the beginning of the occurrence found, and return point.
2243 An optional second argument bounds the search; it is a buffer position.
2244 The match found must not extend before that position.
2245 Optional third argument, if t, means if fail just return nil (no error).
2246 If not nil and not t, move to limit of search and return nil.
2247 Optional fourth argument is repeat count--search for successive occurrences. */)
2248 (Lisp_Object string
, Lisp_Object bound
, Lisp_Object noerror
, Lisp_Object count
)
2250 return search_command (wordify (string
, 0), bound
, noerror
, count
, -1, 1, 0);
2253 DEFUN ("word-search-forward", Fword_search_forward
, Sword_search_forward
, 1, 4,
2255 doc
: /* Search forward from point for STRING, ignoring differences in punctuation.
2256 Set point to the end of the occurrence found, and return point.
2257 An optional second argument bounds the search; it is a buffer position.
2258 The match found must not extend after that position.
2259 Optional third argument, if t, means if fail just return nil (no error).
2260 If not nil and not t, move to limit of search and return nil.
2261 Optional fourth argument is repeat count--search for successive occurrences. */)
2262 (Lisp_Object string
, Lisp_Object bound
, Lisp_Object noerror
, Lisp_Object count
)
2264 return search_command (wordify (string
, 0), bound
, noerror
, count
, 1, 1, 0);
2267 DEFUN ("word-search-backward-lax", Fword_search_backward_lax
, Sword_search_backward_lax
, 1, 4,
2268 "sWord search backward: ",
2269 doc
: /* Search backward from point for STRING, ignoring differences in punctuation.
2270 Set point to the beginning of the occurrence found, and return point.
2272 Unlike `word-search-backward', the end of STRING need not match a word
2273 boundary unless it ends in whitespace.
2275 An optional second argument bounds the search; it is a buffer position.
2276 The match found must not extend before that position.
2277 Optional third argument, if t, means if fail just return nil (no error).
2278 If not nil and not t, move to limit of search and return nil.
2279 Optional fourth argument is repeat count--search for successive occurrences. */)
2280 (Lisp_Object string
, Lisp_Object bound
, Lisp_Object noerror
, Lisp_Object count
)
2282 return search_command (wordify (string
, 1), bound
, noerror
, count
, -1, 1, 0);
2285 DEFUN ("word-search-forward-lax", Fword_search_forward_lax
, Sword_search_forward_lax
, 1, 4,
2287 doc
: /* Search forward from point for STRING, ignoring differences in punctuation.
2288 Set point to the end of the occurrence found, and return point.
2290 Unlike `word-search-forward', the end of STRING need not match a word
2291 boundary unless it ends in whitespace.
2293 An optional second argument bounds the search; it is a buffer position.
2294 The match found must not extend after that position.
2295 Optional third argument, if t, means if fail just return nil (no error).
2296 If not nil and not t, move to limit of search and return nil.
2297 Optional fourth argument is repeat count--search for successive occurrences. */)
2298 (Lisp_Object string
, Lisp_Object bound
, Lisp_Object noerror
, Lisp_Object count
)
2300 return search_command (wordify (string
, 1), bound
, noerror
, count
, 1, 1, 0);
2303 DEFUN ("re-search-backward", Fre_search_backward
, Sre_search_backward
, 1, 4,
2304 "sRE search backward: ",
2305 doc
: /* Search backward from point for match for regular expression REGEXP.
2306 Set point to the beginning of the match, and return point.
2307 The match found is the one starting last in the buffer
2308 and yet ending before the origin of the search.
2309 An optional second argument bounds the search; it is a buffer position.
2310 The match found must start at or after that position.
2311 Optional third argument, if t, means if fail just return nil (no error).
2312 If not nil and not t, move to limit of search and return nil.
2313 Optional fourth argument is repeat count--search for successive occurrences.
2314 See also the functions `match-beginning', `match-end', `match-string',
2315 and `replace-match'. */)
2316 (Lisp_Object regexp
, Lisp_Object bound
, Lisp_Object noerror
, Lisp_Object count
)
2318 return search_command (regexp
, bound
, noerror
, count
, -1, 1, 0);
2321 DEFUN ("re-search-forward", Fre_search_forward
, Sre_search_forward
, 1, 4,
2323 doc
: /* Search forward from point for regular expression REGEXP.
2324 Set point to the end of the occurrence found, and return point.
2325 An optional second argument bounds the search; it is a buffer position.
2326 The match found must not extend after that position.
2327 Optional third argument, if t, means if fail just return nil (no error).
2328 If not nil and not t, move to limit of search and return nil.
2329 Optional fourth argument is repeat count--search for successive occurrences.
2330 See also the functions `match-beginning', `match-end', `match-string',
2331 and `replace-match'. */)
2332 (Lisp_Object regexp
, Lisp_Object bound
, Lisp_Object noerror
, Lisp_Object count
)
2334 return search_command (regexp
, bound
, noerror
, count
, 1, 1, 0);
2337 DEFUN ("posix-search-backward", Fposix_search_backward
, Sposix_search_backward
, 1, 4,
2338 "sPosix search backward: ",
2339 doc
: /* Search backward from point for match for regular expression REGEXP.
2340 Find the longest match in accord with Posix regular expression rules.
2341 Set point to the beginning of the match, and return point.
2342 The match found is the one starting last in the buffer
2343 and yet ending before the origin of the search.
2344 An optional second argument bounds the search; it is a buffer position.
2345 The match found must start at or after that position.
2346 Optional third argument, if t, means if fail just return nil (no error).
2347 If not nil and not t, move to limit of search and return nil.
2348 Optional fourth argument is repeat count--search for successive occurrences.
2349 See also the functions `match-beginning', `match-end', `match-string',
2350 and `replace-match'. */)
2351 (Lisp_Object regexp
, Lisp_Object bound
, Lisp_Object noerror
, Lisp_Object count
)
2353 return search_command (regexp
, bound
, noerror
, count
, -1, 1, 1);
2356 DEFUN ("posix-search-forward", Fposix_search_forward
, Sposix_search_forward
, 1, 4,
2358 doc
: /* Search forward from point for regular expression REGEXP.
2359 Find the longest match in accord with Posix regular expression rules.
2360 Set point to the end of the occurrence found, and return point.
2361 An optional second argument bounds the search; it is a buffer position.
2362 The match found must not extend after that position.
2363 Optional third argument, if t, means if fail just return nil (no error).
2364 If not nil and not t, move to limit of search and return nil.
2365 Optional fourth argument is repeat count--search for successive occurrences.
2366 See also the functions `match-beginning', `match-end', `match-string',
2367 and `replace-match'. */)
2368 (Lisp_Object regexp
, Lisp_Object bound
, Lisp_Object noerror
, Lisp_Object count
)
2370 return search_command (regexp
, bound
, noerror
, count
, 1, 1, 1);
2373 DEFUN ("replace-match", Freplace_match
, Sreplace_match
, 1, 5, 0,
2374 doc
: /* Replace text matched by last search with NEWTEXT.
2375 Leave point at the end of the replacement text.
2377 If second arg FIXEDCASE is non-nil, do not alter case of replacement text.
2378 Otherwise maybe capitalize the whole text, or maybe just word initials,
2379 based on the replaced text.
2380 If the replaced text has only capital letters
2381 and has at least one multiletter word, convert NEWTEXT to all caps.
2382 Otherwise if all words are capitalized in the replaced text,
2383 capitalize each word in NEWTEXT.
2385 If third arg LITERAL is non-nil, insert NEWTEXT literally.
2386 Otherwise treat `\\' as special:
2387 `\\&' in NEWTEXT means substitute original matched text.
2388 `\\N' means substitute what matched the Nth `\\(...\\)'.
2389 If Nth parens didn't match, substitute nothing.
2390 `\\\\' means insert one `\\'.
2391 Case conversion does not apply to these substitutions.
2393 FIXEDCASE and LITERAL are optional arguments.
2395 The optional fourth argument STRING can be a string to modify.
2396 This is meaningful when the previous match was done against STRING,
2397 using `string-match'. When used this way, `replace-match'
2398 creates and returns a new string made by copying STRING and replacing
2399 the part of STRING that was matched.
2401 The optional fifth argument SUBEXP specifies a subexpression;
2402 it says to replace just that subexpression with NEWTEXT,
2403 rather than replacing the entire matched text.
2404 This is, in a vague sense, the inverse of using `\\N' in NEWTEXT;
2405 `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts
2406 NEWTEXT in place of subexp N.
2407 This is useful only after a regular expression search or match,
2408 since only regular expressions have distinguished subexpressions. */)
2409 (Lisp_Object newtext
, Lisp_Object fixedcase
, Lisp_Object literal
, Lisp_Object string
, Lisp_Object subexp
)
2411 enum { nochange
, all_caps
, cap_initial
} case_action
;
2412 register int pos
, pos_byte
;
2413 int some_multiletter_word
;
2416 int some_nonuppercase_initial
;
2417 register int c
, prevc
;
2419 EMACS_INT opoint
, newpoint
;
2421 CHECK_STRING (newtext
);
2423 if (! NILP (string
))
2424 CHECK_STRING (string
);
2426 case_action
= nochange
; /* We tried an initialization */
2427 /* but some C compilers blew it */
2429 if (search_regs
.num_regs
<= 0)
2430 error ("`replace-match' called before any match found");
2436 CHECK_NUMBER (subexp
);
2437 sub
= XINT (subexp
);
2438 if (sub
< 0 || sub
>= search_regs
.num_regs
)
2439 args_out_of_range (subexp
, make_number (search_regs
.num_regs
));
2444 if (search_regs
.start
[sub
] < BEGV
2445 || search_regs
.start
[sub
] > search_regs
.end
[sub
]
2446 || search_regs
.end
[sub
] > ZV
)
2447 args_out_of_range (make_number (search_regs
.start
[sub
]),
2448 make_number (search_regs
.end
[sub
]));
2452 if (search_regs
.start
[sub
] < 0
2453 || search_regs
.start
[sub
] > search_regs
.end
[sub
]
2454 || search_regs
.end
[sub
] > SCHARS (string
))
2455 args_out_of_range (make_number (search_regs
.start
[sub
]),
2456 make_number (search_regs
.end
[sub
]));
2459 if (NILP (fixedcase
))
2461 /* Decide how to casify by examining the matched text. */
2464 pos
= search_regs
.start
[sub
];
2465 last
= search_regs
.end
[sub
];
2468 pos_byte
= CHAR_TO_BYTE (pos
);
2470 pos_byte
= string_char_to_byte (string
, pos
);
2473 case_action
= all_caps
;
2475 /* some_multiletter_word is set nonzero if any original word
2476 is more than one letter long. */
2477 some_multiletter_word
= 0;
2479 some_nonuppercase_initial
= 0;
2486 c
= FETCH_CHAR_AS_MULTIBYTE (pos_byte
);
2487 INC_BOTH (pos
, pos_byte
);
2490 FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c
, string
, pos
, pos_byte
);
2494 /* Cannot be all caps if any original char is lower case */
2497 if (SYNTAX (prevc
) != Sword
)
2498 some_nonuppercase_initial
= 1;
2500 some_multiletter_word
= 1;
2502 else if (UPPERCASEP (c
))
2505 if (SYNTAX (prevc
) != Sword
)
2508 some_multiletter_word
= 1;
2512 /* If the initial is a caseless word constituent,
2513 treat that like a lowercase initial. */
2514 if (SYNTAX (prevc
) != Sword
)
2515 some_nonuppercase_initial
= 1;
2521 /* Convert to all caps if the old text is all caps
2522 and has at least one multiletter word. */
2523 if (! some_lowercase
&& some_multiletter_word
)
2524 case_action
= all_caps
;
2525 /* Capitalize each word, if the old text has all capitalized words. */
2526 else if (!some_nonuppercase_initial
&& some_multiletter_word
)
2527 case_action
= cap_initial
;
2528 else if (!some_nonuppercase_initial
&& some_uppercase
)
2529 /* Should x -> yz, operating on X, give Yz or YZ?
2530 We'll assume the latter. */
2531 case_action
= all_caps
;
2533 case_action
= nochange
;
2536 /* Do replacement in a string. */
2539 Lisp_Object before
, after
;
2541 before
= Fsubstring (string
, make_number (0),
2542 make_number (search_regs
.start
[sub
]));
2543 after
= Fsubstring (string
, make_number (search_regs
.end
[sub
]), Qnil
);
2545 /* Substitute parts of the match into NEWTEXT
2549 EMACS_INT lastpos
= 0;
2550 EMACS_INT lastpos_byte
= 0;
2551 /* We build up the substituted string in ACCUM. */
2554 int length
= SBYTES (newtext
);
2558 for (pos_byte
= 0, pos
= 0; pos_byte
< length
;)
2562 int delbackslash
= 0;
2564 FETCH_STRING_CHAR_ADVANCE (c
, newtext
, pos
, pos_byte
);
2568 FETCH_STRING_CHAR_ADVANCE (c
, newtext
, pos
, pos_byte
);
2572 substart
= search_regs
.start
[sub
];
2573 subend
= search_regs
.end
[sub
];
2575 else if (c
>= '1' && c
<= '9')
2577 if (search_regs
.start
[c
- '0'] >= 0
2578 && c
<= search_regs
.num_regs
+ '0')
2580 substart
= search_regs
.start
[c
- '0'];
2581 subend
= search_regs
.end
[c
- '0'];
2585 /* If that subexp did not match,
2586 replace \\N with nothing. */
2594 error ("Invalid use of `\\' in replacement text");
2598 if (pos
- 2 != lastpos
)
2599 middle
= substring_both (newtext
, lastpos
,
2601 pos
- 2, pos_byte
- 2);
2604 accum
= concat3 (accum
, middle
,
2606 make_number (substart
),
2607 make_number (subend
)));
2609 lastpos_byte
= pos_byte
;
2611 else if (delbackslash
)
2613 middle
= substring_both (newtext
, lastpos
,
2615 pos
- 1, pos_byte
- 1);
2617 accum
= concat2 (accum
, middle
);
2619 lastpos_byte
= pos_byte
;
2624 middle
= substring_both (newtext
, lastpos
,
2630 newtext
= concat2 (accum
, middle
);
2633 /* Do case substitution in NEWTEXT if desired. */
2634 if (case_action
== all_caps
)
2635 newtext
= Fupcase (newtext
);
2636 else if (case_action
== cap_initial
)
2637 newtext
= Fupcase_initials (newtext
);
2639 return concat3 (before
, newtext
, after
);
2642 /* Record point, then move (quietly) to the start of the match. */
2643 if (PT
>= search_regs
.end
[sub
])
2645 else if (PT
> search_regs
.start
[sub
])
2646 opoint
= search_regs
.end
[sub
] - ZV
;
2650 /* If we want non-literal replacement,
2651 perform substitution on the replacement string. */
2654 int length
= SBYTES (newtext
);
2655 unsigned char *substed
;
2656 int substed_alloc_size
, substed_len
;
2657 int buf_multibyte
= !NILP (current_buffer
->enable_multibyte_characters
);
2658 int str_multibyte
= STRING_MULTIBYTE (newtext
);
2659 Lisp_Object rev_tbl
;
2660 int really_changed
= 0;
2664 substed_alloc_size
= length
* 2 + 100;
2665 substed
= (unsigned char *) xmalloc (substed_alloc_size
+ 1);
2668 /* Go thru NEWTEXT, producing the actual text to insert in
2669 SUBSTED while adjusting multibyteness to that of the current
2672 for (pos_byte
= 0, pos
= 0; pos_byte
< length
;)
2674 unsigned char str
[MAX_MULTIBYTE_LENGTH
];
2675 unsigned char *add_stuff
= NULL
;
2681 FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c
, newtext
, pos
, pos_byte
);
2683 c
= multibyte_char_to_unibyte (c
, rev_tbl
);
2687 /* Note that we don't have to increment POS. */
2688 c
= SREF (newtext
, pos_byte
++);
2690 MAKE_CHAR_MULTIBYTE (c
);
2693 /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED,
2694 or set IDX to a match index, which means put that part
2695 of the buffer text into SUBSTED. */
2703 FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c
, newtext
,
2705 if (!buf_multibyte
&& !ASCII_CHAR_P (c
))
2706 c
= multibyte_char_to_unibyte (c
, rev_tbl
);
2710 c
= SREF (newtext
, pos_byte
++);
2712 MAKE_CHAR_MULTIBYTE (c
);
2717 else if (c
>= '1' && c
<= '9' && c
<= search_regs
.num_regs
+ '0')
2719 if (search_regs
.start
[c
- '0'] >= 1)
2723 add_len
= 1, add_stuff
= "\\";
2727 error ("Invalid use of `\\' in replacement text");
2732 add_len
= CHAR_STRING (c
, str
);
2736 /* If we want to copy part of a previous match,
2737 set up ADD_STUFF and ADD_LEN to point to it. */
2740 EMACS_INT begbyte
= CHAR_TO_BYTE (search_regs
.start
[idx
]);
2741 add_len
= CHAR_TO_BYTE (search_regs
.end
[idx
]) - begbyte
;
2742 if (search_regs
.start
[idx
] < GPT
&& GPT
< search_regs
.end
[idx
])
2743 move_gap (search_regs
.start
[idx
]);
2744 add_stuff
= BYTE_POS_ADDR (begbyte
);
2747 /* Now the stuff we want to add to SUBSTED
2748 is invariably ADD_LEN bytes starting at ADD_STUFF. */
2750 /* Make sure SUBSTED is big enough. */
2751 if (substed_len
+ add_len
>= substed_alloc_size
)
2753 substed_alloc_size
= substed_len
+ add_len
+ 500;
2754 substed
= (unsigned char *) xrealloc (substed
,
2755 substed_alloc_size
+ 1);
2758 /* Now add to the end of SUBSTED. */
2761 memcpy (substed
+ substed_len
, add_stuff
, add_len
);
2762 substed_len
+= add_len
;
2770 int nchars
= multibyte_chars_in_text (substed
, substed_len
);
2772 newtext
= make_multibyte_string (substed
, nchars
, substed_len
);
2775 newtext
= make_unibyte_string (substed
, substed_len
);
2780 /* Replace the old text with the new in the cleanest possible way. */
2781 replace_range (search_regs
.start
[sub
], search_regs
.end
[sub
],
2783 newpoint
= search_regs
.start
[sub
] + SCHARS (newtext
);
2785 if (case_action
== all_caps
)
2786 Fupcase_region (make_number (search_regs
.start
[sub
]),
2787 make_number (newpoint
));
2788 else if (case_action
== cap_initial
)
2789 Fupcase_initials_region (make_number (search_regs
.start
[sub
]),
2790 make_number (newpoint
));
2792 /* Adjust search data for this change. */
2794 EMACS_INT oldend
= search_regs
.end
[sub
];
2795 EMACS_INT oldstart
= search_regs
.start
[sub
];
2796 EMACS_INT change
= newpoint
- search_regs
.end
[sub
];
2799 for (i
= 0; i
< search_regs
.num_regs
; i
++)
2801 if (search_regs
.start
[i
] >= oldend
)
2802 search_regs
.start
[i
] += change
;
2803 else if (search_regs
.start
[i
] > oldstart
)
2804 search_regs
.start
[i
] = oldstart
;
2805 if (search_regs
.end
[i
] >= oldend
)
2806 search_regs
.end
[i
] += change
;
2807 else if (search_regs
.end
[i
] > oldstart
)
2808 search_regs
.end
[i
] = oldstart
;
2812 /* Put point back where it was in the text. */
2814 TEMP_SET_PT (opoint
+ ZV
);
2816 TEMP_SET_PT (opoint
);
2818 /* Now move point "officially" to the start of the inserted replacement. */
2819 move_if_not_intangible (newpoint
);
2825 match_limit (Lisp_Object num
, int beginningp
)
2832 args_out_of_range (num
, make_number (0));
2833 if (search_regs
.num_regs
<= 0)
2834 error ("No match data, because no search succeeded");
2835 if (n
>= search_regs
.num_regs
2836 || search_regs
.start
[n
] < 0)
2838 return (make_number ((beginningp
) ? search_regs
.start
[n
]
2839 : search_regs
.end
[n
]));
2842 DEFUN ("match-beginning", Fmatch_beginning
, Smatch_beginning
, 1, 1, 0,
2843 doc
: /* Return position of start of text matched by last search.
2844 SUBEXP, a number, specifies which parenthesized expression in the last
2846 Value is nil if SUBEXPth pair didn't match, or there were less than
2848 Zero means the entire text matched by the whole regexp or whole string. */)
2849 (Lisp_Object subexp
)
2851 return match_limit (subexp
, 1);
2854 DEFUN ("match-end", Fmatch_end
, Smatch_end
, 1, 1, 0,
2855 doc
: /* Return position of end of text matched by last search.
2856 SUBEXP, a number, specifies which parenthesized expression in the last
2858 Value is nil if SUBEXPth pair didn't match, or there were less than
2860 Zero means the entire text matched by the whole regexp or whole string. */)
2861 (Lisp_Object subexp
)
2863 return match_limit (subexp
, 0);
2866 DEFUN ("match-data", Fmatch_data
, Smatch_data
, 0, 3, 0,
2867 doc
: /* Return a list containing all info on what the last search matched.
2868 Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'.
2869 All the elements are markers or nil (nil if the Nth pair didn't match)
2870 if the last match was on a buffer; integers or nil if a string was matched.
2871 Use `set-match-data' to reinstate the data in this list.
2873 If INTEGERS (the optional first argument) is non-nil, always use
2874 integers \(rather than markers) to represent buffer positions. In
2875 this case, and if the last match was in a buffer, the buffer will get
2876 stored as one additional element at the end of the list.
2878 If REUSE is a list, reuse it as part of the value. If REUSE is long
2879 enough to hold all the values, and if INTEGERS is non-nil, no consing
2882 If optional third arg RESEAT is non-nil, any previous markers on the
2883 REUSE list will be modified to point to nowhere.
2885 Return value is undefined if the last search failed. */)
2886 (Lisp_Object integers
, Lisp_Object reuse
, Lisp_Object reseat
)
2888 Lisp_Object tail
, prev
;
2893 for (tail
= reuse
; CONSP (tail
); tail
= XCDR (tail
))
2894 if (MARKERP (XCAR (tail
)))
2896 unchain_marker (XMARKER (XCAR (tail
)));
2897 XSETCAR (tail
, Qnil
);
2900 if (NILP (last_thing_searched
))
2905 data
= (Lisp_Object
*) alloca ((2 * search_regs
.num_regs
+ 1)
2906 * sizeof (Lisp_Object
));
2909 for (i
= 0; i
< search_regs
.num_regs
; i
++)
2911 int start
= search_regs
.start
[i
];
2914 if (EQ (last_thing_searched
, Qt
)
2915 || ! NILP (integers
))
2917 XSETFASTINT (data
[2 * i
], start
);
2918 XSETFASTINT (data
[2 * i
+ 1], search_regs
.end
[i
]);
2920 else if (BUFFERP (last_thing_searched
))
2922 data
[2 * i
] = Fmake_marker ();
2923 Fset_marker (data
[2 * i
],
2924 make_number (start
),
2925 last_thing_searched
);
2926 data
[2 * i
+ 1] = Fmake_marker ();
2927 Fset_marker (data
[2 * i
+ 1],
2928 make_number (search_regs
.end
[i
]),
2929 last_thing_searched
);
2932 /* last_thing_searched must always be Qt, a buffer, or Qnil. */
2938 data
[2 * i
] = data
[2 * i
+ 1] = Qnil
;
2941 if (BUFFERP (last_thing_searched
) && !NILP (integers
))
2943 data
[len
] = last_thing_searched
;
2947 /* If REUSE is not usable, cons up the values and return them. */
2948 if (! CONSP (reuse
))
2949 return Flist (len
, data
);
2951 /* If REUSE is a list, store as many value elements as will fit
2952 into the elements of REUSE. */
2953 for (i
= 0, tail
= reuse
; CONSP (tail
);
2954 i
++, tail
= XCDR (tail
))
2957 XSETCAR (tail
, data
[i
]);
2959 XSETCAR (tail
, Qnil
);
2963 /* If we couldn't fit all value elements into REUSE,
2964 cons up the rest of them and add them to the end of REUSE. */
2966 XSETCDR (prev
, Flist (len
- i
, data
+ i
));
2971 /* We used to have an internal use variant of `reseat' described as:
2973 If RESEAT is `evaporate', put the markers back on the free list
2974 immediately. No other references to the markers must exist in this
2975 case, so it is used only internally on the unwind stack and
2976 save-match-data from Lisp.
2978 But it was ill-conceived: those supposedly-internal markers get exposed via
2979 the undo-list, so freeing them here is unsafe. */
2981 DEFUN ("set-match-data", Fset_match_data
, Sset_match_data
, 1, 2, 0,
2982 doc
: /* Set internal data on last search match from elements of LIST.
2983 LIST should have been created by calling `match-data' previously.
2985 If optional arg RESEAT is non-nil, make markers on LIST point nowhere. */)
2986 (register Lisp_Object list
, Lisp_Object reseat
)
2989 register Lisp_Object marker
;
2991 if (running_asynch_code
)
2992 save_search_regs ();
2996 /* Unless we find a marker with a buffer or an explicit buffer
2997 in LIST, assume that this match data came from a string. */
2998 last_thing_searched
= Qt
;
3000 /* Allocate registers if they don't already exist. */
3002 int length
= XFASTINT (Flength (list
)) / 2;
3004 if (length
> search_regs
.num_regs
)
3006 if (search_regs
.num_regs
== 0)
3009 = (regoff_t
*) xmalloc (length
* sizeof (regoff_t
));
3011 = (regoff_t
*) xmalloc (length
* sizeof (regoff_t
));
3016 = (regoff_t
*) xrealloc (search_regs
.start
,
3017 length
* sizeof (regoff_t
));
3019 = (regoff_t
*) xrealloc (search_regs
.end
,
3020 length
* sizeof (regoff_t
));
3023 for (i
= search_regs
.num_regs
; i
< length
; i
++)
3024 search_regs
.start
[i
] = -1;
3026 search_regs
.num_regs
= length
;
3029 for (i
= 0; CONSP (list
); i
++)
3031 marker
= XCAR (list
);
3032 if (BUFFERP (marker
))
3034 last_thing_searched
= marker
;
3041 search_regs
.start
[i
] = -1;
3050 if (MARKERP (marker
))
3052 if (XMARKER (marker
)->buffer
== 0)
3053 XSETFASTINT (marker
, 0);
3055 XSETBUFFER (last_thing_searched
, XMARKER (marker
)->buffer
);
3058 CHECK_NUMBER_COERCE_MARKER (marker
);
3059 from
= XINT (marker
);
3061 if (!NILP (reseat
) && MARKERP (m
))
3063 unchain_marker (XMARKER (m
));
3064 XSETCAR (list
, Qnil
);
3067 if ((list
= XCDR (list
), !CONSP (list
)))
3070 m
= marker
= XCAR (list
);
3072 if (MARKERP (marker
) && XMARKER (marker
)->buffer
== 0)
3073 XSETFASTINT (marker
, 0);
3075 CHECK_NUMBER_COERCE_MARKER (marker
);
3076 search_regs
.start
[i
] = from
;
3077 search_regs
.end
[i
] = XINT (marker
);
3079 if (!NILP (reseat
) && MARKERP (m
))
3081 unchain_marker (XMARKER (m
));
3082 XSETCAR (list
, Qnil
);
3088 for (; i
< search_regs
.num_regs
; i
++)
3089 search_regs
.start
[i
] = -1;
3095 /* If non-zero the match data have been saved in saved_search_regs
3096 during the execution of a sentinel or filter. */
3097 static int search_regs_saved
;
3098 static struct re_registers saved_search_regs
;
3099 static Lisp_Object saved_last_thing_searched
;
3101 /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data
3102 if asynchronous code (filter or sentinel) is running. */
3104 save_search_regs (void)
3106 if (!search_regs_saved
)
3108 saved_search_regs
.num_regs
= search_regs
.num_regs
;
3109 saved_search_regs
.start
= search_regs
.start
;
3110 saved_search_regs
.end
= search_regs
.end
;
3111 saved_last_thing_searched
= last_thing_searched
;
3112 last_thing_searched
= Qnil
;
3113 search_regs
.num_regs
= 0;
3114 search_regs
.start
= 0;
3115 search_regs
.end
= 0;
3117 search_regs_saved
= 1;
3121 /* Called upon exit from filters and sentinels. */
3123 restore_search_regs (void)
3125 if (search_regs_saved
)
3127 if (search_regs
.num_regs
> 0)
3129 xfree (search_regs
.start
);
3130 xfree (search_regs
.end
);
3132 search_regs
.num_regs
= saved_search_regs
.num_regs
;
3133 search_regs
.start
= saved_search_regs
.start
;
3134 search_regs
.end
= saved_search_regs
.end
;
3135 last_thing_searched
= saved_last_thing_searched
;
3136 saved_last_thing_searched
= Qnil
;
3137 search_regs_saved
= 0;
3142 unwind_set_match_data (Lisp_Object list
)
3144 /* It is NOT ALWAYS safe to free (evaporate) the markers immediately. */
3145 return Fset_match_data (list
, Qt
);
3148 /* Called to unwind protect the match data. */
3150 record_unwind_save_match_data (void)
3152 record_unwind_protect (unwind_set_match_data
,
3153 Fmatch_data (Qnil
, Qnil
, Qnil
));
3156 /* Quote a string to inactivate reg-expr chars */
3158 DEFUN ("regexp-quote", Fregexp_quote
, Sregexp_quote
, 1, 1, 0,
3159 doc
: /* Return a regexp string which matches exactly STRING and nothing else. */)
3160 (Lisp_Object string
)
3162 register unsigned char *in
, *out
, *end
;
3163 register unsigned char *temp
;
3164 int backslashes_added
= 0;
3166 CHECK_STRING (string
);
3168 temp
= (unsigned char *) alloca (SBYTES (string
) * 2);
3170 /* Now copy the data into the new string, inserting escapes. */
3172 in
= SDATA (string
);
3173 end
= in
+ SBYTES (string
);
3176 for (; in
!= end
; in
++)
3179 || *in
== '*' || *in
== '.' || *in
== '\\'
3180 || *in
== '?' || *in
== '+'
3181 || *in
== '^' || *in
== '$')
3182 *out
++ = '\\', backslashes_added
++;
3186 return make_specified_string (temp
,
3187 SCHARS (string
) + backslashes_added
,
3189 STRING_MULTIBYTE (string
));
3193 syms_of_search (void)
3197 for (i
= 0; i
< REGEXP_CACHE_SIZE
; ++i
)
3199 searchbufs
[i
].buf
.allocated
= 100;
3200 searchbufs
[i
].buf
.buffer
= (unsigned char *) xmalloc (100);
3201 searchbufs
[i
].buf
.fastmap
= searchbufs
[i
].fastmap
;
3202 searchbufs
[i
].regexp
= Qnil
;
3203 searchbufs
[i
].whitespace_regexp
= Qnil
;
3204 searchbufs
[i
].syntax_table
= Qnil
;
3205 staticpro (&searchbufs
[i
].regexp
);
3206 staticpro (&searchbufs
[i
].whitespace_regexp
);
3207 staticpro (&searchbufs
[i
].syntax_table
);
3208 searchbufs
[i
].next
= (i
== REGEXP_CACHE_SIZE
-1 ? 0 : &searchbufs
[i
+1]);
3210 searchbuf_head
= &searchbufs
[0];
3212 Qsearch_failed
= intern_c_string ("search-failed");
3213 staticpro (&Qsearch_failed
);
3214 Qinvalid_regexp
= intern_c_string ("invalid-regexp");
3215 staticpro (&Qinvalid_regexp
);
3217 Fput (Qsearch_failed
, Qerror_conditions
,
3218 pure_cons (Qsearch_failed
, pure_cons (Qerror
, Qnil
)));
3219 Fput (Qsearch_failed
, Qerror_message
,
3220 make_pure_c_string ("Search failed"));
3222 Fput (Qinvalid_regexp
, Qerror_conditions
,
3223 pure_cons (Qinvalid_regexp
, pure_cons (Qerror
, Qnil
)));
3224 Fput (Qinvalid_regexp
, Qerror_message
,
3225 make_pure_c_string ("Invalid regexp"));
3227 last_thing_searched
= Qnil
;
3228 staticpro (&last_thing_searched
);
3230 saved_last_thing_searched
= Qnil
;
3231 staticpro (&saved_last_thing_searched
);
3233 DEFVAR_LISP ("search-spaces-regexp", &Vsearch_spaces_regexp
,
3234 doc
: /* Regexp to substitute for bunches of spaces in regexp search.
3235 Some commands use this for user-specified regexps.
3236 Spaces that occur inside character classes or repetition operators
3237 or other such regexp constructs are not replaced with this.
3238 A value of nil (which is the normal value) means treat spaces literally. */);
3239 Vsearch_spaces_regexp
= Qnil
;
3241 DEFVAR_LISP ("inhibit-changing-match-data", &Vinhibit_changing_match_data
,
3242 doc
: /* Internal use only.
3243 If non-nil, the primitive searching and matching functions
3244 such as `looking-at', `string-match', `re-search-forward', etc.,
3245 do not set the match data. The proper way to use this variable
3246 is to bind it with `let' around a small expression. */);
3247 Vinhibit_changing_match_data
= Qnil
;
3249 defsubr (&Slooking_at
);
3250 defsubr (&Sposix_looking_at
);
3251 defsubr (&Sstring_match
);
3252 defsubr (&Sposix_string_match
);
3253 defsubr (&Ssearch_forward
);
3254 defsubr (&Ssearch_backward
);
3255 defsubr (&Sword_search_forward
);
3256 defsubr (&Sword_search_backward
);
3257 defsubr (&Sword_search_forward_lax
);
3258 defsubr (&Sword_search_backward_lax
);
3259 defsubr (&Sre_search_forward
);
3260 defsubr (&Sre_search_backward
);
3261 defsubr (&Sposix_search_forward
);
3262 defsubr (&Sposix_search_backward
);
3263 defsubr (&Sreplace_match
);
3264 defsubr (&Smatch_beginning
);
3265 defsubr (&Smatch_end
);
3266 defsubr (&Smatch_data
);
3267 defsubr (&Sset_match_data
);
3268 defsubr (&Sregexp_quote
);
3271 /* arch-tag: a6059d79-0552-4f14-a2cb-d379a4e3c78f
3272 (do not change this comment) */