Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
bd2cbd56 SM |
2 | Copyright (C) 1985, 86,87,93,94,97,98, 1999, 2004 |
3 | Free Software Foundation, Inc. | |
ca1d1d23 JB |
4 | |
5 | This file is part of GNU Emacs. | |
6 | ||
7 | GNU Emacs is free software; you can redistribute it and/or modify | |
8 | it under the terms of the GNU General Public License as published by | |
7c938215 | 9 | the Free Software Foundation; either version 2, or (at your option) |
ca1d1d23 JB |
10 | any later version. |
11 | ||
12 | GNU Emacs 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. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
18 | along with GNU Emacs; see the file COPYING. If not, write to | |
3b7ad313 EN |
19 | the Free Software Foundation, Inc., 59 Temple Place - Suite 330, |
20 | Boston, MA 02111-1307, USA. */ | |
ca1d1d23 JB |
21 | |
22 | ||
18160b98 | 23 | #include <config.h> |
ca1d1d23 JB |
24 | #include "lisp.h" |
25 | #include "syntax.h" | |
5679531d | 26 | #include "category.h" |
ca1d1d23 | 27 | #include "buffer.h" |
5679531d | 28 | #include "charset.h" |
9169c321 | 29 | #include "region-cache.h" |
ca1d1d23 | 30 | #include "commands.h" |
9ac0d9e0 | 31 | #include "blockinput.h" |
bf1760bb | 32 | #include "intervals.h" |
4746118a | 33 | |
ca1d1d23 JB |
34 | #include <sys/types.h> |
35 | #include "regex.h" | |
36 | ||
1d288aef | 37 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 38 | |
487282dc KH |
39 | /* If the regexp is non-nil, then the buffer contains the compiled form |
40 | of that regexp, suitable for searching. */ | |
1d288aef RS |
41 | struct regexp_cache |
42 | { | |
487282dc | 43 | struct regexp_cache *next; |
ecdb561e | 44 | Lisp_Object regexp, whitespace_regexp; |
487282dc KH |
45 | struct re_pattern_buffer buf; |
46 | char fastmap[0400]; | |
b819a390 RS |
47 | /* Nonzero means regexp was compiled to do full POSIX backtracking. */ |
48 | char posix; | |
487282dc | 49 | }; |
ca1d1d23 | 50 | |
487282dc KH |
51 | /* The instances of that struct. */ |
52 | struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; | |
ca1d1d23 | 53 | |
487282dc KH |
54 | /* The head of the linked list; points to the most recently used buffer. */ |
55 | struct regexp_cache *searchbuf_head; | |
ca1d1d23 | 56 | |
ca1d1d23 | 57 | |
4746118a JB |
58 | /* Every call to re_match, etc., must pass &search_regs as the regs |
59 | argument unless you can show it is unnecessary (i.e., if re_match | |
60 | is certainly going to be called again before region-around-match | |
61 | can be called). | |
62 | ||
63 | Since the registers are now dynamically allocated, we need to make | |
64 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
65 | been allocated by checking search_regs.num_regs. |
66 | ||
67 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
68 | buffer using bits in the re_pattern_buffer. This means that whenever |
69 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
70 | allocated any registers, and will allocate new registers the next |
71 | time you call a searching or matching function. Therefore, we need | |
72 | to call re_set_registers after compiling a new pattern or after | |
73 | setting the match registers, so that the regex functions will be | |
74 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
75 | static struct re_registers search_regs; |
76 | ||
daa37602 JB |
77 | /* The buffer in which the last search was performed, or |
78 | Qt if the last search was done in a string; | |
79 | Qnil if no searching has been done yet. */ | |
80 | static Lisp_Object last_thing_searched; | |
ca1d1d23 | 81 | |
8e6208c5 | 82 | /* error condition signaled when regexp compile_pattern fails */ |
ca1d1d23 JB |
83 | |
84 | Lisp_Object Qinvalid_regexp; | |
85 | ||
41a33295 | 86 | Lisp_Object Vsearch_spaces_regexp; |
f31a9a68 | 87 | |
ca325161 | 88 | static void set_search_regs (); |
044f81f1 | 89 | static void save_search_regs (); |
facdc750 RS |
90 | static int simple_search (); |
91 | static int boyer_moore (); | |
b819a390 RS |
92 | static int search_buffer (); |
93 | ||
ca1d1d23 JB |
94 | static void |
95 | matcher_overflow () | |
96 | { | |
97 | error ("Stack overflow in regexp matcher"); | |
98 | } | |
99 | ||
b819a390 RS |
100 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
101 | PATTERN is the pattern to compile. | |
102 | CP is the place to put the result. | |
facdc750 | 103 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
104 | REGP is the structure that says where to store the "register" |
105 | values that will result from matching this pattern. | |
106 | If it is 0, we should compile the pattern not to record any | |
107 | subexpression bounds. | |
108 | POSIX is nonzero if we want full backtracking (POSIX style) | |
5679531d KH |
109 | for this pattern. 0 means backtrack only enough to get a valid match. |
110 | MULTIBYTE is nonzero if we want to handle multibyte characters in | |
111 | PATTERN. 0 means all multibyte characters are recognized just as | |
ecdb561e RS |
112 | sequences of binary data. |
113 | ||
41a33295 | 114 | The behavior also depends on Vsearch_spaces_regexp. */ |
ca1d1d23 | 115 | |
487282dc | 116 | static void |
5679531d | 117 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte) |
487282dc | 118 | struct regexp_cache *cp; |
ca1d1d23 | 119 | Lisp_Object pattern; |
facdc750 | 120 | Lisp_Object translate; |
487282dc | 121 | struct re_registers *regp; |
b819a390 | 122 | int posix; |
5679531d | 123 | int multibyte; |
ca1d1d23 | 124 | { |
7276d3d8 | 125 | unsigned char *raw_pattern; |
f8bd51c4 | 126 | int raw_pattern_size; |
d451e4db | 127 | char *val; |
b819a390 | 128 | reg_syntax_t old; |
ca1d1d23 | 129 | |
f8bd51c4 KH |
130 | /* MULTIBYTE says whether the text to be searched is multibyte. |
131 | We must convert PATTERN to match that, or we will not really | |
132 | find things right. */ | |
133 | ||
134 | if (multibyte == STRING_MULTIBYTE (pattern)) | |
135 | { | |
d5db4077 KR |
136 | raw_pattern = (unsigned char *) SDATA (pattern); |
137 | raw_pattern_size = SBYTES (pattern); | |
f8bd51c4 KH |
138 | } |
139 | else if (multibyte) | |
140 | { | |
d5db4077 KR |
141 | raw_pattern_size = count_size_as_multibyte (SDATA (pattern), |
142 | SCHARS (pattern)); | |
7276d3d8 | 143 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
144 | copy_text (SDATA (pattern), raw_pattern, |
145 | SCHARS (pattern), 0, 1); | |
f8bd51c4 KH |
146 | } |
147 | else | |
148 | { | |
149 | /* Converting multibyte to single-byte. | |
150 | ||
151 | ??? Perhaps this conversion should be done in a special way | |
152 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
153 | so that only the multibyte chars which really correspond to | |
154 | the chosen single-byte character set can possibly match. */ | |
d5db4077 | 155 | raw_pattern_size = SCHARS (pattern); |
7276d3d8 | 156 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
157 | copy_text (SDATA (pattern), raw_pattern, |
158 | SBYTES (pattern), 1, 0); | |
f8bd51c4 KH |
159 | } |
160 | ||
487282dc | 161 | cp->regexp = Qnil; |
59fab369 | 162 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 163 | cp->posix = posix; |
5679531d | 164 | cp->buf.multibyte = multibyte; |
41a33295 | 165 | cp->whitespace_regexp = Vsearch_spaces_regexp; |
9ac0d9e0 | 166 | BLOCK_INPUT; |
fb4a568d | 167 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 168 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
f31a9a68 | 169 | |
41a33295 RS |
170 | re_set_whitespace_regexp (NILP (Vsearch_spaces_regexp) ? NULL |
171 | : SDATA (Vsearch_spaces_regexp)); | |
f31a9a68 | 172 | |
7276d3d8 RS |
173 | val = (char *) re_compile_pattern ((char *)raw_pattern, |
174 | raw_pattern_size, &cp->buf); | |
f31a9a68 RS |
175 | |
176 | re_set_whitespace_regexp (NULL); | |
177 | ||
b819a390 | 178 | re_set_syntax (old); |
9ac0d9e0 | 179 | UNBLOCK_INPUT; |
ca1d1d23 | 180 | if (val) |
487282dc | 181 | Fsignal (Qinvalid_regexp, Fcons (build_string (val), Qnil)); |
1113d9db | 182 | |
487282dc | 183 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
184 | } |
185 | ||
6efc7887 RS |
186 | /* Shrink each compiled regexp buffer in the cache |
187 | to the size actually used right now. | |
188 | This is called from garbage collection. */ | |
189 | ||
190 | void | |
191 | shrink_regexp_cache () | |
192 | { | |
a968f437 | 193 | struct regexp_cache *cp; |
6efc7887 RS |
194 | |
195 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
196 | { | |
197 | cp->buf.allocated = cp->buf.used; | |
198 | cp->buf.buffer | |
b23c0a83 | 199 | = (unsigned char *) xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
200 | } |
201 | } | |
202 | ||
487282dc | 203 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
204 | the cache. |
205 | PATTERN is the pattern to compile. | |
facdc750 | 206 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
207 | REGP is the structure that says where to store the "register" |
208 | values that will result from matching this pattern. | |
209 | If it is 0, we should compile the pattern not to record any | |
210 | subexpression bounds. | |
211 | POSIX is nonzero if we want full backtracking (POSIX style) | |
212 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
213 | |
214 | struct re_pattern_buffer * | |
0c8533c6 | 215 | compile_pattern (pattern, regp, translate, posix, multibyte) |
487282dc KH |
216 | Lisp_Object pattern; |
217 | struct re_registers *regp; | |
facdc750 | 218 | Lisp_Object translate; |
0c8533c6 | 219 | int posix, multibyte; |
487282dc KH |
220 | { |
221 | struct regexp_cache *cp, **cpp; | |
222 | ||
223 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
224 | { | |
225 | cp = *cpp; | |
f1b9c7c1 KR |
226 | /* Entries are initialized to nil, and may be set to nil by |
227 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
49a5f770 KR |
228 | string accessors in those cases. However, compile_pattern_1 |
229 | is only applied to the cache entry we pick here to reuse. So | |
230 | nil should never appear before a non-nil entry. */ | |
7c752c80 | 231 | if (NILP (cp->regexp)) |
f1b9c7c1 | 232 | goto compile_it; |
d5db4077 | 233 | if (SCHARS (cp->regexp) == SCHARS (pattern) |
cf69b13e | 234 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 235 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 236 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d | 237 | && cp->posix == posix |
ecdb561e | 238 | && cp->buf.multibyte == multibyte |
41a33295 | 239 | && !NILP (Fequal (cp->whitespace_regexp, Vsearch_spaces_regexp))) |
487282dc KH |
240 | break; |
241 | ||
f1b9c7c1 KR |
242 | /* If we're at the end of the cache, compile into the nil cell |
243 | we found, or the last (least recently used) cell with a | |
244 | string value. */ | |
487282dc KH |
245 | if (cp->next == 0) |
246 | { | |
f1b9c7c1 | 247 | compile_it: |
5679531d | 248 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte); |
487282dc KH |
249 | break; |
250 | } | |
251 | } | |
252 | ||
253 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
254 | either because we found it in the cache or because we just compiled it. | |
255 | Move it to the front of the queue to mark it as most recently used. */ | |
256 | *cpp = cp->next; | |
257 | cp->next = searchbuf_head; | |
258 | searchbuf_head = cp; | |
1113d9db | 259 | |
6639708c RS |
260 | /* Advise the searching functions about the space we have allocated |
261 | for register data. */ | |
262 | if (regp) | |
263 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
264 | ||
487282dc | 265 | return &cp->buf; |
ca1d1d23 JB |
266 | } |
267 | ||
268 | /* Error condition used for failing searches */ | |
269 | Lisp_Object Qsearch_failed; | |
270 | ||
271 | Lisp_Object | |
272 | signal_failure (arg) | |
273 | Lisp_Object arg; | |
274 | { | |
275 | Fsignal (Qsearch_failed, Fcons (arg, Qnil)); | |
276 | return Qnil; | |
277 | } | |
278 | \f | |
b819a390 RS |
279 | static Lisp_Object |
280 | looking_at_1 (string, posix) | |
ca1d1d23 | 281 | Lisp_Object string; |
b819a390 | 282 | int posix; |
ca1d1d23 JB |
283 | { |
284 | Lisp_Object val; | |
285 | unsigned char *p1, *p2; | |
286 | int s1, s2; | |
287 | register int i; | |
487282dc | 288 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 289 | |
7074fde6 FP |
290 | if (running_asynch_code) |
291 | save_search_regs (); | |
292 | ||
b7826503 | 293 | CHECK_STRING (string); |
487282dc KH |
294 | bufp = compile_pattern (string, &search_regs, |
295 | (!NILP (current_buffer->case_fold_search) | |
facdc750 | 296 | ? DOWNCASE_TABLE : Qnil), |
0c8533c6 RS |
297 | posix, |
298 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 JB |
299 | |
300 | immediate_quit = 1; | |
301 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
302 | ||
303 | /* Get pointers and sizes of the two strings | |
304 | that make up the visible portion of the buffer. */ | |
305 | ||
306 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 307 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 308 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 309 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
310 | if (s1 < 0) |
311 | { | |
312 | p2 = p1; | |
fa8ed3e0 | 313 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
314 | s1 = 0; |
315 | } | |
316 | if (s2 < 0) | |
317 | { | |
fa8ed3e0 | 318 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
319 | s2 = 0; |
320 | } | |
8bb43c28 RS |
321 | |
322 | re_match_object = Qnil; | |
177c0ea7 | 323 | |
487282dc | 324 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
fa8ed3e0 RS |
325 | PT_BYTE - BEGV_BYTE, &search_regs, |
326 | ZV_BYTE - BEGV_BYTE); | |
de182d70 | 327 | immediate_quit = 0; |
177c0ea7 | 328 | |
ca1d1d23 JB |
329 | if (i == -2) |
330 | matcher_overflow (); | |
331 | ||
332 | val = (0 <= i ? Qt : Qnil); | |
fa8ed3e0 RS |
333 | if (i >= 0) |
334 | for (i = 0; i < search_regs.num_regs; i++) | |
335 | if (search_regs.start[i] >= 0) | |
336 | { | |
337 | search_regs.start[i] | |
338 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
339 | search_regs.end[i] | |
340 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
341 | } | |
a3668d92 | 342 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
343 | return val; |
344 | } | |
345 | ||
b819a390 | 346 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
347 | doc: /* Return t if text after point matches regular expression REGEXP. |
348 | This function modifies the match data that `match-beginning', | |
349 | `match-end' and `match-data' access; save and restore the match | |
350 | data if you want to preserve them. */) | |
351 | (regexp) | |
94f94972 | 352 | Lisp_Object regexp; |
b819a390 | 353 | { |
94f94972 | 354 | return looking_at_1 (regexp, 0); |
b819a390 RS |
355 | } |
356 | ||
357 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
358 | doc: /* Return t if text after point matches regular expression REGEXP. |
359 | Find the longest match, in accord with Posix regular expression rules. | |
360 | This function modifies the match data that `match-beginning', | |
361 | `match-end' and `match-data' access; save and restore the match | |
362 | data if you want to preserve them. */) | |
363 | (regexp) | |
94f94972 | 364 | Lisp_Object regexp; |
b819a390 | 365 | { |
94f94972 | 366 | return looking_at_1 (regexp, 1); |
b819a390 RS |
367 | } |
368 | \f | |
369 | static Lisp_Object | |
370 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 371 | Lisp_Object regexp, string, start; |
b819a390 | 372 | int posix; |
ca1d1d23 JB |
373 | { |
374 | int val; | |
487282dc | 375 | struct re_pattern_buffer *bufp; |
0c8533c6 RS |
376 | int pos, pos_byte; |
377 | int i; | |
ca1d1d23 | 378 | |
7074fde6 FP |
379 | if (running_asynch_code) |
380 | save_search_regs (); | |
381 | ||
b7826503 PJ |
382 | CHECK_STRING (regexp); |
383 | CHECK_STRING (string); | |
ca1d1d23 JB |
384 | |
385 | if (NILP (start)) | |
0c8533c6 | 386 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
387 | else |
388 | { | |
d5db4077 | 389 | int len = SCHARS (string); |
ca1d1d23 | 390 | |
b7826503 | 391 | CHECK_NUMBER (start); |
0c8533c6 RS |
392 | pos = XINT (start); |
393 | if (pos < 0 && -pos <= len) | |
394 | pos = len + pos; | |
395 | else if (0 > pos || pos > len) | |
ca1d1d23 | 396 | args_out_of_range (string, start); |
0c8533c6 | 397 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
398 | } |
399 | ||
487282dc KH |
400 | bufp = compile_pattern (regexp, &search_regs, |
401 | (!NILP (current_buffer->case_fold_search) | |
facdc750 | 402 | ? DOWNCASE_TABLE : Qnil), |
0c8533c6 RS |
403 | posix, |
404 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 405 | immediate_quit = 1; |
8bb43c28 | 406 | re_match_object = string; |
177c0ea7 | 407 | |
d5db4077 KR |
408 | val = re_search (bufp, (char *) SDATA (string), |
409 | SBYTES (string), pos_byte, | |
410 | SBYTES (string) - pos_byte, | |
ca1d1d23 JB |
411 | &search_regs); |
412 | immediate_quit = 0; | |
daa37602 | 413 | last_thing_searched = Qt; |
ca1d1d23 JB |
414 | if (val == -2) |
415 | matcher_overflow (); | |
416 | if (val < 0) return Qnil; | |
0c8533c6 RS |
417 | |
418 | for (i = 0; i < search_regs.num_regs; i++) | |
419 | if (search_regs.start[i] >= 0) | |
420 | { | |
421 | search_regs.start[i] | |
422 | = string_byte_to_char (string, search_regs.start[i]); | |
423 | search_regs.end[i] | |
424 | = string_byte_to_char (string, search_regs.end[i]); | |
425 | } | |
426 | ||
427 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 428 | } |
e59a8453 | 429 | |
b819a390 | 430 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 PJ |
431 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
432 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
433 | If third arg START is non-nil, start search at that index in STRING. | |
434 | For index of first char beyond the match, do (match-end 0). | |
435 | `match-end' and `match-beginning' also give indices of substrings | |
2bd2f32d RS |
436 | matched by parenthesis constructs in the pattern. |
437 | ||
438 | You can use the function `match-string' to extract the substrings | |
439 | matched by the parenthesis constructions in REGEXP. */) | |
8c1a1077 | 440 | (regexp, string, start) |
b819a390 RS |
441 | Lisp_Object regexp, string, start; |
442 | { | |
443 | return string_match_1 (regexp, string, start, 0); | |
444 | } | |
445 | ||
446 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
447 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
448 | Find the longest match, in accord with Posix regular expression rules. | |
449 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
450 | If third arg START is non-nil, start search at that index in STRING. | |
451 | For index of first char beyond the match, do (match-end 0). | |
452 | `match-end' and `match-beginning' also give indices of substrings | |
453 | matched by parenthesis constructs in the pattern. */) | |
454 | (regexp, string, start) | |
b819a390 RS |
455 | Lisp_Object regexp, string, start; |
456 | { | |
457 | return string_match_1 (regexp, string, start, 1); | |
458 | } | |
459 | ||
e59a8453 RS |
460 | /* Match REGEXP against STRING, searching all of STRING, |
461 | and return the index of the match, or negative on failure. | |
462 | This does not clobber the match data. */ | |
463 | ||
464 | int | |
465 | fast_string_match (regexp, string) | |
466 | Lisp_Object regexp, string; | |
467 | { | |
468 | int val; | |
487282dc | 469 | struct re_pattern_buffer *bufp; |
e59a8453 | 470 | |
facdc750 RS |
471 | bufp = compile_pattern (regexp, 0, Qnil, |
472 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 473 | immediate_quit = 1; |
8bb43c28 | 474 | re_match_object = string; |
177c0ea7 | 475 | |
d5db4077 KR |
476 | val = re_search (bufp, (char *) SDATA (string), |
477 | SBYTES (string), 0, | |
478 | SBYTES (string), 0); | |
e59a8453 RS |
479 | immediate_quit = 0; |
480 | return val; | |
481 | } | |
5679531d KH |
482 | |
483 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
484 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
485 | This does not clobber the match data. |
486 | We assume that STRING contains single-byte characters. */ | |
5679531d KH |
487 | |
488 | extern Lisp_Object Vascii_downcase_table; | |
489 | ||
490 | int | |
b4577c63 | 491 | fast_c_string_match_ignore_case (regexp, string) |
5679531d | 492 | Lisp_Object regexp; |
96b80561 | 493 | const char *string; |
5679531d KH |
494 | { |
495 | int val; | |
496 | struct re_pattern_buffer *bufp; | |
497 | int len = strlen (string); | |
498 | ||
0c8533c6 | 499 | regexp = string_make_unibyte (regexp); |
b4577c63 | 500 | re_match_object = Qt; |
5679531d | 501 | bufp = compile_pattern (regexp, 0, |
facdc750 | 502 | Vascii_downcase_table, 0, |
f8bd51c4 | 503 | 0); |
5679531d KH |
504 | immediate_quit = 1; |
505 | val = re_search (bufp, string, len, 0, len, 0); | |
506 | immediate_quit = 0; | |
507 | return val; | |
508 | } | |
be5f4dfb KH |
509 | |
510 | /* Like fast_string_match but ignore case. */ | |
511 | ||
512 | int | |
513 | fast_string_match_ignore_case (regexp, string) | |
514 | Lisp_Object regexp, string; | |
515 | { | |
516 | int val; | |
517 | struct re_pattern_buffer *bufp; | |
518 | ||
519 | bufp = compile_pattern (regexp, 0, Vascii_downcase_table, | |
520 | 0, STRING_MULTIBYTE (string)); | |
521 | immediate_quit = 1; | |
522 | re_match_object = string; | |
523 | ||
524 | val = re_search (bufp, (char *) SDATA (string), | |
525 | SBYTES (string), 0, | |
526 | SBYTES (string), 0); | |
527 | immediate_quit = 0; | |
528 | return val; | |
529 | } | |
ca1d1d23 | 530 | \f |
9169c321 JB |
531 | /* The newline cache: remembering which sections of text have no newlines. */ |
532 | ||
533 | /* If the user has requested newline caching, make sure it's on. | |
534 | Otherwise, make sure it's off. | |
535 | This is our cheezy way of associating an action with the change of | |
536 | state of a buffer-local variable. */ | |
537 | static void | |
538 | newline_cache_on_off (buf) | |
539 | struct buffer *buf; | |
540 | { | |
541 | if (NILP (buf->cache_long_line_scans)) | |
542 | { | |
543 | /* It should be off. */ | |
544 | if (buf->newline_cache) | |
545 | { | |
546 | free_region_cache (buf->newline_cache); | |
547 | buf->newline_cache = 0; | |
548 | } | |
549 | } | |
550 | else | |
551 | { | |
552 | /* It should be on. */ | |
553 | if (buf->newline_cache == 0) | |
554 | buf->newline_cache = new_region_cache (); | |
555 | } | |
556 | } | |
557 | ||
558 | \f | |
559 | /* Search for COUNT instances of the character TARGET between START and END. | |
560 | ||
561 | If COUNT is positive, search forwards; END must be >= START. | |
562 | If COUNT is negative, search backwards for the -COUNTth instance; | |
563 | END must be <= START. | |
564 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
565 | ||
566 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
567 | direction indicated by COUNT. | |
ffd56f97 JB |
568 | |
569 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
a9f2a45f | 570 | position past the COUNTth match. Note that for reverse motion |
5bfe95c9 | 571 | this is not the same as the usual convention for Emacs motion commands. |
ffd56f97 | 572 | |
9169c321 JB |
573 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
574 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 575 | |
087a5f81 RS |
576 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
577 | except when inside redisplay. */ | |
578 | ||
dfcf069d | 579 | int |
9169c321 JB |
580 | scan_buffer (target, start, end, count, shortage, allow_quit) |
581 | register int target; | |
582 | int start, end; | |
583 | int count; | |
584 | int *shortage; | |
087a5f81 | 585 | int allow_quit; |
ca1d1d23 | 586 | { |
9169c321 | 587 | struct region_cache *newline_cache; |
177c0ea7 | 588 | int direction; |
ffd56f97 | 589 | |
9169c321 JB |
590 | if (count > 0) |
591 | { | |
592 | direction = 1; | |
593 | if (! end) end = ZV; | |
594 | } | |
595 | else | |
596 | { | |
597 | direction = -1; | |
598 | if (! end) end = BEGV; | |
599 | } | |
ffd56f97 | 600 | |
9169c321 JB |
601 | newline_cache_on_off (current_buffer); |
602 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
603 | |
604 | if (shortage != 0) | |
605 | *shortage = 0; | |
606 | ||
087a5f81 | 607 | immediate_quit = allow_quit; |
ca1d1d23 | 608 | |
ffd56f97 | 609 | if (count > 0) |
9169c321 | 610 | while (start != end) |
ca1d1d23 | 611 | { |
9169c321 JB |
612 | /* Our innermost scanning loop is very simple; it doesn't know |
613 | about gaps, buffer ends, or the newline cache. ceiling is | |
614 | the position of the last character before the next such | |
615 | obstacle --- the last character the dumb search loop should | |
616 | examine. */ | |
fa8ed3e0 RS |
617 | int ceiling_byte = CHAR_TO_BYTE (end) - 1; |
618 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 619 | int tem; |
9169c321 JB |
620 | |
621 | /* If we're looking for a newline, consult the newline cache | |
622 | to see where we can avoid some scanning. */ | |
623 | if (target == '\n' && newline_cache) | |
624 | { | |
625 | int next_change; | |
626 | immediate_quit = 0; | |
627 | while (region_cache_forward | |
fa8ed3e0 RS |
628 | (current_buffer, newline_cache, start_byte, &next_change)) |
629 | start_byte = next_change; | |
cbe0db0d | 630 | immediate_quit = allow_quit; |
9169c321 | 631 | |
fa8ed3e0 RS |
632 | /* START should never be after END. */ |
633 | if (start_byte > ceiling_byte) | |
634 | start_byte = ceiling_byte; | |
9169c321 JB |
635 | |
636 | /* Now the text after start is an unknown region, and | |
637 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 638 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
639 | } |
640 | ||
641 | /* The dumb loop can only scan text stored in contiguous | |
642 | bytes. BUFFER_CEILING_OF returns the last character | |
643 | position that is contiguous, so the ceiling is the | |
644 | position after that. */ | |
67ce527d KH |
645 | tem = BUFFER_CEILING_OF (start_byte); |
646 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
647 | |
648 | { | |
177c0ea7 | 649 | /* The termination address of the dumb loop. */ |
fa8ed3e0 RS |
650 | register unsigned char *ceiling_addr |
651 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
652 | register unsigned char *cursor | |
653 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
654 | unsigned char *base = cursor; |
655 | ||
656 | while (cursor < ceiling_addr) | |
657 | { | |
658 | unsigned char *scan_start = cursor; | |
659 | ||
660 | /* The dumb loop. */ | |
661 | while (*cursor != target && ++cursor < ceiling_addr) | |
662 | ; | |
663 | ||
664 | /* If we're looking for newlines, cache the fact that | |
665 | the region from start to cursor is free of them. */ | |
666 | if (target == '\n' && newline_cache) | |
667 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
668 | start_byte + scan_start - base, |
669 | start_byte + cursor - base); | |
9169c321 JB |
670 | |
671 | /* Did we find the target character? */ | |
672 | if (cursor < ceiling_addr) | |
673 | { | |
674 | if (--count == 0) | |
675 | { | |
676 | immediate_quit = 0; | |
fa8ed3e0 | 677 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
678 | } |
679 | cursor++; | |
680 | } | |
681 | } | |
682 | ||
fa8ed3e0 | 683 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 684 | } |
ca1d1d23 JB |
685 | } |
686 | else | |
9169c321 JB |
687 | while (start > end) |
688 | { | |
689 | /* The last character to check before the next obstacle. */ | |
fa8ed3e0 RS |
690 | int ceiling_byte = CHAR_TO_BYTE (end); |
691 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 692 | int tem; |
9169c321 JB |
693 | |
694 | /* Consult the newline cache, if appropriate. */ | |
695 | if (target == '\n' && newline_cache) | |
696 | { | |
697 | int next_change; | |
698 | immediate_quit = 0; | |
699 | while (region_cache_backward | |
fa8ed3e0 RS |
700 | (current_buffer, newline_cache, start_byte, &next_change)) |
701 | start_byte = next_change; | |
cbe0db0d | 702 | immediate_quit = allow_quit; |
9169c321 JB |
703 | |
704 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
705 | if (start_byte <= ceiling_byte) |
706 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
707 | |
708 | /* Now the text before start is an unknown region, and | |
709 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 710 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
711 | } |
712 | ||
713 | /* Stop scanning before the gap. */ | |
67ce527d KH |
714 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
715 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
716 | |
717 | { | |
718 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
719 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
720 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
721 | unsigned char *base = cursor; |
722 | ||
723 | while (cursor >= ceiling_addr) | |
724 | { | |
725 | unsigned char *scan_start = cursor; | |
726 | ||
727 | while (*cursor != target && --cursor >= ceiling_addr) | |
728 | ; | |
729 | ||
730 | /* If we're looking for newlines, cache the fact that | |
731 | the region from after the cursor to start is free of them. */ | |
732 | if (target == '\n' && newline_cache) | |
733 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
734 | start_byte + cursor - base, |
735 | start_byte + scan_start - base); | |
9169c321 JB |
736 | |
737 | /* Did we find the target character? */ | |
738 | if (cursor >= ceiling_addr) | |
739 | { | |
740 | if (++count >= 0) | |
741 | { | |
742 | immediate_quit = 0; | |
fa8ed3e0 | 743 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
744 | } |
745 | cursor--; | |
746 | } | |
747 | } | |
748 | ||
fa8ed3e0 | 749 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
750 | } |
751 | } | |
752 | ||
ca1d1d23 JB |
753 | immediate_quit = 0; |
754 | if (shortage != 0) | |
ffd56f97 | 755 | *shortage = count * direction; |
9169c321 | 756 | return start; |
ca1d1d23 | 757 | } |
fa8ed3e0 RS |
758 | \f |
759 | /* Search for COUNT instances of a line boundary, which means either a | |
760 | newline or (if selective display enabled) a carriage return. | |
761 | Start at START. If COUNT is negative, search backwards. | |
762 | ||
763 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
764 | ||
765 | If we find COUNT instances. we position after (always after, | |
766 | even if scanning backwards) the COUNTth match, and return 0. | |
767 | ||
768 | If we don't find COUNT instances before reaching the end of the | |
769 | buffer (or the beginning, if scanning backwards), we return | |
770 | the number of line boundaries left unfound, and position at | |
771 | the limit we bumped up against. | |
772 | ||
773 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 774 | except in special cases. */ |
ca1d1d23 | 775 | |
63fa018d | 776 | int |
fa8ed3e0 RS |
777 | scan_newline (start, start_byte, limit, limit_byte, count, allow_quit) |
778 | int start, start_byte; | |
779 | int limit, limit_byte; | |
780 | register int count; | |
781 | int allow_quit; | |
63fa018d | 782 | { |
fa8ed3e0 RS |
783 | int direction = ((count > 0) ? 1 : -1); |
784 | ||
785 | register unsigned char *cursor; | |
786 | unsigned char *base; | |
787 | ||
788 | register int ceiling; | |
789 | register unsigned char *ceiling_addr; | |
790 | ||
d5d57b92 RS |
791 | int old_immediate_quit = immediate_quit; |
792 | ||
fa8ed3e0 RS |
793 | /* The code that follows is like scan_buffer |
794 | but checks for either newline or carriage return. */ | |
795 | ||
d5d57b92 RS |
796 | if (allow_quit) |
797 | immediate_quit++; | |
fa8ed3e0 RS |
798 | |
799 | start_byte = CHAR_TO_BYTE (start); | |
800 | ||
801 | if (count > 0) | |
802 | { | |
803 | while (start_byte < limit_byte) | |
804 | { | |
805 | ceiling = BUFFER_CEILING_OF (start_byte); | |
806 | ceiling = min (limit_byte - 1, ceiling); | |
807 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
808 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
809 | while (1) | |
810 | { | |
811 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
812 | ; | |
813 | ||
814 | if (cursor != ceiling_addr) | |
815 | { | |
816 | if (--count == 0) | |
817 | { | |
d5d57b92 | 818 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
819 | start_byte = start_byte + cursor - base + 1; |
820 | start = BYTE_TO_CHAR (start_byte); | |
821 | TEMP_SET_PT_BOTH (start, start_byte); | |
822 | return 0; | |
823 | } | |
824 | else | |
825 | if (++cursor == ceiling_addr) | |
826 | break; | |
827 | } | |
828 | else | |
829 | break; | |
830 | } | |
831 | start_byte += cursor - base; | |
832 | } | |
833 | } | |
834 | else | |
835 | { | |
fa8ed3e0 RS |
836 | while (start_byte > limit_byte) |
837 | { | |
838 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
839 | ceiling = max (limit_byte, ceiling); | |
840 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
841 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
842 | while (1) | |
843 | { | |
844 | while (--cursor != ceiling_addr && *cursor != '\n') | |
845 | ; | |
846 | ||
847 | if (cursor != ceiling_addr) | |
848 | { | |
849 | if (++count == 0) | |
850 | { | |
d5d57b92 | 851 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
852 | /* Return the position AFTER the match we found. */ |
853 | start_byte = start_byte + cursor - base + 1; | |
854 | start = BYTE_TO_CHAR (start_byte); | |
855 | TEMP_SET_PT_BOTH (start, start_byte); | |
856 | return 0; | |
857 | } | |
858 | } | |
859 | else | |
860 | break; | |
861 | } | |
862 | /* Here we add 1 to compensate for the last decrement | |
863 | of CURSOR, which took it past the valid range. */ | |
864 | start_byte += cursor - base + 1; | |
865 | } | |
866 | } | |
867 | ||
868 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 869 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
870 | |
871 | return count * direction; | |
63fa018d RS |
872 | } |
873 | ||
ca1d1d23 | 874 | int |
fa8ed3e0 | 875 | find_next_newline_no_quit (from, cnt) |
ca1d1d23 JB |
876 | register int from, cnt; |
877 | { | |
fa8ed3e0 | 878 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
9169c321 JB |
879 | } |
880 | ||
9169c321 JB |
881 | /* Like find_next_newline, but returns position before the newline, |
882 | not after, and only search up to TO. This isn't just | |
883 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 884 | |
9169c321 JB |
885 | int |
886 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 887 | int from, to, cnt; |
9169c321 JB |
888 | { |
889 | int shortage; | |
890 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
891 | ||
892 | if (shortage == 0) | |
893 | pos--; | |
177c0ea7 | 894 | |
9169c321 | 895 | return pos; |
ca1d1d23 JB |
896 | } |
897 | \f | |
ca1d1d23 JB |
898 | /* Subroutines of Lisp buffer search functions. */ |
899 | ||
900 | static Lisp_Object | |
b819a390 | 901 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
902 | Lisp_Object string, bound, noerror, count; |
903 | int direction; | |
904 | int RE; | |
b819a390 | 905 | int posix; |
ca1d1d23 JB |
906 | { |
907 | register int np; | |
9f43ad85 | 908 | int lim, lim_byte; |
ca1d1d23 JB |
909 | int n = direction; |
910 | ||
911 | if (!NILP (count)) | |
912 | { | |
b7826503 | 913 | CHECK_NUMBER (count); |
ca1d1d23 JB |
914 | n *= XINT (count); |
915 | } | |
916 | ||
b7826503 | 917 | CHECK_STRING (string); |
ca1d1d23 | 918 | if (NILP (bound)) |
9f43ad85 RS |
919 | { |
920 | if (n > 0) | |
921 | lim = ZV, lim_byte = ZV_BYTE; | |
922 | else | |
923 | lim = BEGV, lim_byte = BEGV_BYTE; | |
924 | } | |
ca1d1d23 JB |
925 | else |
926 | { | |
b7826503 | 927 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 928 | lim = XINT (bound); |
6ec8bbd2 | 929 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
930 | error ("Invalid search bound (wrong side of point)"); |
931 | if (lim > ZV) | |
9f43ad85 | 932 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 933 | else if (lim < BEGV) |
9f43ad85 | 934 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
935 | else |
936 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
937 | } |
938 | ||
9f43ad85 | 939 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
ca1d1d23 | 940 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 941 | ? current_buffer->case_canon_table |
3135e9fd | 942 | : Qnil), |
ca1d1d23 | 943 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 944 | ? current_buffer->case_eqv_table |
3135e9fd | 945 | : Qnil), |
b819a390 | 946 | posix); |
ca1d1d23 JB |
947 | if (np <= 0) |
948 | { | |
949 | if (NILP (noerror)) | |
950 | return signal_failure (string); | |
951 | if (!EQ (noerror, Qt)) | |
952 | { | |
953 | if (lim < BEGV || lim > ZV) | |
954 | abort (); | |
9f43ad85 | 955 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
956 | return Qnil; |
957 | #if 0 /* This would be clean, but maybe programs depend on | |
958 | a value of nil here. */ | |
481399bf | 959 | np = lim; |
a5f217b8 | 960 | #endif |
ca1d1d23 | 961 | } |
481399bf RS |
962 | else |
963 | return Qnil; | |
ca1d1d23 JB |
964 | } |
965 | ||
966 | if (np < BEGV || np > ZV) | |
967 | abort (); | |
968 | ||
969 | SET_PT (np); | |
970 | ||
971 | return make_number (np); | |
972 | } | |
973 | \f | |
fa8ed3e0 RS |
974 | /* Return 1 if REGEXP it matches just one constant string. */ |
975 | ||
b6d6a51c KH |
976 | static int |
977 | trivial_regexp_p (regexp) | |
978 | Lisp_Object regexp; | |
979 | { | |
d5db4077 KR |
980 | int len = SBYTES (regexp); |
981 | unsigned char *s = SDATA (regexp); | |
b6d6a51c KH |
982 | while (--len >= 0) |
983 | { | |
984 | switch (*s++) | |
985 | { | |
986 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
987 | return 0; | |
988 | case '\\': | |
989 | if (--len < 0) | |
990 | return 0; | |
991 | switch (*s++) | |
992 | { | |
993 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
994 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 995 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 996 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 997 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
998 | case '6': case '7': case '8': case '9': |
999 | return 0; | |
1000 | } | |
1001 | } | |
1002 | } | |
1003 | return 1; | |
1004 | } | |
1005 | ||
ca325161 | 1006 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 1007 | starting at position POS and stopping at position LIM, |
b819a390 | 1008 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
1009 | a regular expression if RE is true. |
1010 | ||
1011 | If N is positive, searching is forward and LIM must be greater than POS. | |
1012 | If N is negative, searching is backward and LIM must be less than POS. | |
1013 | ||
facdc750 | 1014 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 1015 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
1016 | (if searching backward) or the end (if searching forward). |
1017 | ||
1018 | POSIX is nonzero if we want full backtracking (POSIX style) | |
1019 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 1020 | |
aff2ce94 RS |
1021 | #define TRANSLATE(out, trt, d) \ |
1022 | do \ | |
1023 | { \ | |
1024 | if (! NILP (trt)) \ | |
1025 | { \ | |
1026 | Lisp_Object temp; \ | |
1027 | temp = Faref (trt, make_number (d)); \ | |
1028 | if (INTEGERP (temp)) \ | |
1029 | out = XINT (temp); \ | |
1030 | else \ | |
1031 | out = d; \ | |
1032 | } \ | |
1033 | else \ | |
1034 | out = d; \ | |
1035 | } \ | |
1036 | while (0) | |
facdc750 | 1037 | |
b819a390 | 1038 | static int |
9f43ad85 RS |
1039 | search_buffer (string, pos, pos_byte, lim, lim_byte, n, |
1040 | RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
1041 | Lisp_Object string; |
1042 | int pos; | |
9f43ad85 | 1043 | int pos_byte; |
ca1d1d23 | 1044 | int lim; |
9f43ad85 | 1045 | int lim_byte; |
ca1d1d23 JB |
1046 | int n; |
1047 | int RE; | |
facdc750 RS |
1048 | Lisp_Object trt; |
1049 | Lisp_Object inverse_trt; | |
b819a390 | 1050 | int posix; |
ca1d1d23 | 1051 | { |
d5db4077 KR |
1052 | int len = SCHARS (string); |
1053 | int len_byte = SBYTES (string); | |
facdc750 | 1054 | register int i; |
ca1d1d23 | 1055 | |
7074fde6 FP |
1056 | if (running_asynch_code) |
1057 | save_search_regs (); | |
1058 | ||
a7e4cdde | 1059 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1060 | /* Null string is found at starting position. */ |
a7e4cdde | 1061 | if (len == 0 || n == 0) |
ca325161 | 1062 | { |
0353b28f | 1063 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1064 | return pos; |
1065 | } | |
3f57a499 | 1066 | |
41a33295 | 1067 | if (RE && !(trivial_regexp_p (string) && NILP (Vsearch_spaces_regexp))) |
ca1d1d23 | 1068 | { |
facdc750 RS |
1069 | unsigned char *p1, *p2; |
1070 | int s1, s2; | |
487282dc KH |
1071 | struct re_pattern_buffer *bufp; |
1072 | ||
0c8533c6 RS |
1073 | bufp = compile_pattern (string, &search_regs, trt, posix, |
1074 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 | 1075 | |
ca1d1d23 JB |
1076 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1077 | because letting this function finish | |
1078 | can take too long. */ | |
1079 | QUIT; /* Do a pending quit right away, | |
1080 | to avoid paradoxical behavior */ | |
1081 | /* Get pointers and sizes of the two strings | |
1082 | that make up the visible portion of the buffer. */ | |
1083 | ||
1084 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1085 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1086 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1087 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1088 | if (s1 < 0) |
1089 | { | |
1090 | p2 = p1; | |
fa8ed3e0 | 1091 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1092 | s1 = 0; |
1093 | } | |
1094 | if (s2 < 0) | |
1095 | { | |
fa8ed3e0 | 1096 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1097 | s2 = 0; |
1098 | } | |
8bb43c28 | 1099 | re_match_object = Qnil; |
177c0ea7 | 1100 | |
ca1d1d23 JB |
1101 | while (n < 0) |
1102 | { | |
42db823b | 1103 | int val; |
487282dc | 1104 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1105 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1106 | &search_regs, | |
42db823b | 1107 | /* Don't allow match past current point */ |
4996330b | 1108 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1109 | if (val == -2) |
b6d6a51c KH |
1110 | { |
1111 | matcher_overflow (); | |
1112 | } | |
ca1d1d23 JB |
1113 | if (val >= 0) |
1114 | { | |
26aff150 | 1115 | pos_byte = search_regs.start[0] + BEGV_BYTE; |
4746118a | 1116 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1117 | if (search_regs.start[i] >= 0) |
1118 | { | |
fa8ed3e0 RS |
1119 | search_regs.start[i] |
1120 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1121 | search_regs.end[i] | |
1122 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1123 | } |
a3668d92 | 1124 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1125 | /* Set pos to the new position. */ |
1126 | pos = search_regs.start[0]; | |
1127 | } | |
1128 | else | |
1129 | { | |
1130 | immediate_quit = 0; | |
1131 | return (n); | |
1132 | } | |
1133 | n++; | |
1134 | } | |
1135 | while (n > 0) | |
1136 | { | |
42db823b | 1137 | int val; |
487282dc | 1138 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1139 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1140 | &search_regs, | |
1141 | lim_byte - BEGV_BYTE); | |
ca1d1d23 | 1142 | if (val == -2) |
b6d6a51c KH |
1143 | { |
1144 | matcher_overflow (); | |
1145 | } | |
ca1d1d23 JB |
1146 | if (val >= 0) |
1147 | { | |
26aff150 | 1148 | pos_byte = search_regs.end[0] + BEGV_BYTE; |
4746118a | 1149 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1150 | if (search_regs.start[i] >= 0) |
1151 | { | |
fa8ed3e0 RS |
1152 | search_regs.start[i] |
1153 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1154 | search_regs.end[i] | |
1155 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1156 | } |
a3668d92 | 1157 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1158 | pos = search_regs.end[0]; |
1159 | } | |
1160 | else | |
1161 | { | |
1162 | immediate_quit = 0; | |
1163 | return (0 - n); | |
1164 | } | |
1165 | n--; | |
1166 | } | |
1167 | immediate_quit = 0; | |
1168 | return (pos); | |
1169 | } | |
1170 | else /* non-RE case */ | |
1171 | { | |
facdc750 RS |
1172 | unsigned char *raw_pattern, *pat; |
1173 | int raw_pattern_size; | |
1174 | int raw_pattern_size_byte; | |
1175 | unsigned char *patbuf; | |
1176 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); | |
d5db4077 | 1177 | unsigned char *base_pat = SDATA (string); |
facdc750 | 1178 | int charset_base = -1; |
040272ce | 1179 | int boyer_moore_ok = 1; |
facdc750 RS |
1180 | |
1181 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1182 | We must convert PATTERN to match that, or we will not really | |
1183 | find things right. */ | |
1184 | ||
1185 | if (multibyte == STRING_MULTIBYTE (string)) | |
1186 | { | |
d5db4077 KR |
1187 | raw_pattern = (unsigned char *) SDATA (string); |
1188 | raw_pattern_size = SCHARS (string); | |
1189 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1190 | } |
1191 | else if (multibyte) | |
1192 | { | |
d5db4077 | 1193 | raw_pattern_size = SCHARS (string); |
facdc750 | 1194 | raw_pattern_size_byte |
d5db4077 | 1195 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1196 | raw_pattern_size); |
7276d3d8 | 1197 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1198 | copy_text (SDATA (string), raw_pattern, |
1199 | SCHARS (string), 0, 1); | |
facdc750 RS |
1200 | } |
1201 | else | |
1202 | { | |
1203 | /* Converting multibyte to single-byte. | |
1204 | ||
1205 | ??? Perhaps this conversion should be done in a special way | |
1206 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1207 | so that only the multibyte chars which really correspond to | |
1208 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1209 | raw_pattern_size = SCHARS (string); |
1210 | raw_pattern_size_byte = SCHARS (string); | |
7276d3d8 | 1211 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
1212 | copy_text (SDATA (string), raw_pattern, |
1213 | SBYTES (string), 1, 0); | |
facdc750 RS |
1214 | } |
1215 | ||
1216 | /* Copy and optionally translate the pattern. */ | |
1217 | len = raw_pattern_size; | |
1218 | len_byte = raw_pattern_size_byte; | |
1219 | patbuf = (unsigned char *) alloca (len_byte); | |
1220 | pat = patbuf; | |
1221 | base_pat = raw_pattern; | |
1222 | if (multibyte) | |
1223 | { | |
1224 | while (--len >= 0) | |
1225 | { | |
daaa6ed8 | 1226 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
aff2ce94 | 1227 | int c, translated, inverse; |
facdc750 RS |
1228 | int in_charlen, charlen; |
1229 | ||
1230 | /* If we got here and the RE flag is set, it's because we're | |
1231 | dealing with a regexp known to be trivial, so the backslash | |
1232 | just quotes the next character. */ | |
1233 | if (RE && *base_pat == '\\') | |
1234 | { | |
1235 | len--; | |
1236 | len_byte--; | |
1237 | base_pat++; | |
1238 | } | |
1239 | ||
1240 | c = STRING_CHAR_AND_LENGTH (base_pat, len_byte, in_charlen); | |
040272ce | 1241 | |
facdc750 | 1242 | /* Translate the character, if requested. */ |
aff2ce94 | 1243 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1244 | /* If translation changed the byte-length, go back |
1245 | to the original character. */ | |
daaa6ed8 | 1246 | charlen = CHAR_STRING (translated, str); |
facdc750 RS |
1247 | if (in_charlen != charlen) |
1248 | { | |
1249 | translated = c; | |
daaa6ed8 | 1250 | charlen = CHAR_STRING (c, str); |
facdc750 RS |
1251 | } |
1252 | ||
5ffaf437 RS |
1253 | /* If we are searching for something strange, |
1254 | an invalid multibyte code, don't use boyer-moore. */ | |
1255 | if (! ASCII_BYTE_P (translated) | |
1256 | && (charlen == 1 /* 8bit code */ | |
1257 | || charlen != in_charlen /* invalid multibyte code */ | |
1258 | )) | |
1259 | boyer_moore_ok = 0; | |
1260 | ||
aff2ce94 RS |
1261 | TRANSLATE (inverse, inverse_trt, c); |
1262 | ||
facdc750 RS |
1263 | /* Did this char actually get translated? |
1264 | Would any other char get translated into it? */ | |
aff2ce94 | 1265 | if (translated != c || inverse != c) |
facdc750 RS |
1266 | { |
1267 | /* Keep track of which character set row | |
1268 | contains the characters that need translation. */ | |
5ffaf437 | 1269 | int charset_base_code = c & ~CHAR_FIELD3_MASK; |
d2ac725b KH |
1270 | int inverse_charset_base = inverse & ~CHAR_FIELD3_MASK; |
1271 | ||
1272 | if (charset_base_code != inverse_charset_base) | |
1273 | boyer_moore_ok = 0; | |
1274 | else if (charset_base == -1) | |
facdc750 RS |
1275 | charset_base = charset_base_code; |
1276 | else if (charset_base != charset_base_code) | |
1277 | /* If two different rows appear, needing translation, | |
1278 | then we cannot use boyer_moore search. */ | |
040272ce | 1279 | boyer_moore_ok = 0; |
aff2ce94 | 1280 | } |
facdc750 RS |
1281 | |
1282 | /* Store this character into the translated pattern. */ | |
1283 | bcopy (str, pat, charlen); | |
1284 | pat += charlen; | |
1285 | base_pat += in_charlen; | |
1286 | len_byte -= in_charlen; | |
1287 | } | |
1288 | } | |
1289 | else | |
1290 | { | |
040272ce KH |
1291 | /* Unibyte buffer. */ |
1292 | charset_base = 0; | |
facdc750 RS |
1293 | while (--len >= 0) |
1294 | { | |
040272ce | 1295 | int c, translated; |
facdc750 RS |
1296 | |
1297 | /* If we got here and the RE flag is set, it's because we're | |
1298 | dealing with a regexp known to be trivial, so the backslash | |
1299 | just quotes the next character. */ | |
1300 | if (RE && *base_pat == '\\') | |
1301 | { | |
1302 | len--; | |
1303 | base_pat++; | |
1304 | } | |
1305 | c = *base_pat++; | |
aff2ce94 | 1306 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1307 | *pat++ = translated; |
1308 | } | |
1309 | } | |
1310 | ||
1311 | len_byte = pat - patbuf; | |
1312 | len = raw_pattern_size; | |
1313 | pat = base_pat = patbuf; | |
1314 | ||
040272ce | 1315 | if (boyer_moore_ok) |
facdc750 | 1316 | return boyer_moore (n, pat, len, len_byte, trt, inverse_trt, |
aff2ce94 RS |
1317 | pos, pos_byte, lim, lim_byte, |
1318 | charset_base); | |
facdc750 RS |
1319 | else |
1320 | return simple_search (n, pat, len, len_byte, trt, | |
1321 | pos, pos_byte, lim, lim_byte); | |
1322 | } | |
1323 | } | |
1324 | \f | |
1325 | /* Do a simple string search N times for the string PAT, | |
1326 | whose length is LEN/LEN_BYTE, | |
1327 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1328 | TRT is the translation table. | |
f8bd51c4 | 1329 | |
facdc750 RS |
1330 | Return the character position where the match is found. |
1331 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1332 | |
facdc750 RS |
1333 | This kind of search works regardless of what is in PAT and |
1334 | regardless of what is in TRT. It is used in cases where | |
1335 | boyer_moore cannot work. */ | |
1336 | ||
1337 | static int | |
1338 | simple_search (n, pat, len, len_byte, trt, pos, pos_byte, lim, lim_byte) | |
1339 | int n; | |
1340 | unsigned char *pat; | |
1341 | int len, len_byte; | |
1342 | Lisp_Object trt; | |
1343 | int pos, pos_byte; | |
1344 | int lim, lim_byte; | |
1345 | { | |
1346 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); | |
ab228c24 | 1347 | int forward = n > 0; |
facdc750 RS |
1348 | |
1349 | if (lim > pos && multibyte) | |
1350 | while (n > 0) | |
1351 | { | |
1352 | while (1) | |
f8bd51c4 | 1353 | { |
facdc750 RS |
1354 | /* Try matching at position POS. */ |
1355 | int this_pos = pos; | |
1356 | int this_pos_byte = pos_byte; | |
1357 | int this_len = len; | |
1358 | int this_len_byte = len_byte; | |
1359 | unsigned char *p = pat; | |
1360 | if (pos + len > lim) | |
1361 | goto stop; | |
1362 | ||
1363 | while (this_len > 0) | |
1364 | { | |
1365 | int charlen, buf_charlen; | |
ab228c24 | 1366 | int pat_ch, buf_ch; |
facdc750 | 1367 | |
ab228c24 | 1368 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1369 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1370 | ZV_BYTE - this_pos_byte, | |
1371 | buf_charlen); | |
aff2ce94 | 1372 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1373 | |
1374 | if (buf_ch != pat_ch) | |
1375 | break; | |
ab228c24 RS |
1376 | |
1377 | this_len_byte -= charlen; | |
1378 | this_len--; | |
1379 | p += charlen; | |
1380 | ||
1381 | this_pos_byte += buf_charlen; | |
1382 | this_pos++; | |
facdc750 RS |
1383 | } |
1384 | ||
1385 | if (this_len == 0) | |
1386 | { | |
1387 | pos += len; | |
1388 | pos_byte += len_byte; | |
1389 | break; | |
1390 | } | |
1391 | ||
1392 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1393 | } |
facdc750 RS |
1394 | |
1395 | n--; | |
1396 | } | |
1397 | else if (lim > pos) | |
1398 | while (n > 0) | |
1399 | { | |
1400 | while (1) | |
f8bd51c4 | 1401 | { |
facdc750 RS |
1402 | /* Try matching at position POS. */ |
1403 | int this_pos = pos; | |
1404 | int this_len = len; | |
1405 | unsigned char *p = pat; | |
1406 | ||
1407 | if (pos + len > lim) | |
1408 | goto stop; | |
1409 | ||
1410 | while (this_len > 0) | |
1411 | { | |
1412 | int pat_ch = *p++; | |
1413 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1414 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1415 | |
1416 | if (buf_ch != pat_ch) | |
1417 | break; | |
ab228c24 RS |
1418 | |
1419 | this_len--; | |
1420 | this_pos++; | |
facdc750 RS |
1421 | } |
1422 | ||
1423 | if (this_len == 0) | |
1424 | { | |
1425 | pos += len; | |
1426 | break; | |
1427 | } | |
1428 | ||
1429 | pos++; | |
f8bd51c4 | 1430 | } |
facdc750 RS |
1431 | |
1432 | n--; | |
1433 | } | |
1434 | /* Backwards search. */ | |
1435 | else if (lim < pos && multibyte) | |
1436 | while (n < 0) | |
1437 | { | |
1438 | while (1) | |
f8bd51c4 | 1439 | { |
facdc750 RS |
1440 | /* Try matching at position POS. */ |
1441 | int this_pos = pos - len; | |
1442 | int this_pos_byte = pos_byte - len_byte; | |
1443 | int this_len = len; | |
1444 | int this_len_byte = len_byte; | |
1445 | unsigned char *p = pat; | |
1446 | ||
1447 | if (pos - len < lim) | |
1448 | goto stop; | |
1449 | ||
1450 | while (this_len > 0) | |
1451 | { | |
1452 | int charlen, buf_charlen; | |
ab228c24 | 1453 | int pat_ch, buf_ch; |
facdc750 | 1454 | |
ab228c24 | 1455 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1456 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1457 | ZV_BYTE - this_pos_byte, | |
1458 | buf_charlen); | |
aff2ce94 | 1459 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1460 | |
1461 | if (buf_ch != pat_ch) | |
1462 | break; | |
ab228c24 RS |
1463 | |
1464 | this_len_byte -= charlen; | |
1465 | this_len--; | |
1466 | p += charlen; | |
1467 | this_pos_byte += buf_charlen; | |
1468 | this_pos++; | |
facdc750 RS |
1469 | } |
1470 | ||
1471 | if (this_len == 0) | |
1472 | { | |
1473 | pos -= len; | |
1474 | pos_byte -= len_byte; | |
1475 | break; | |
1476 | } | |
1477 | ||
1478 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1479 | } |
1480 | ||
facdc750 RS |
1481 | n++; |
1482 | } | |
1483 | else if (lim < pos) | |
1484 | while (n < 0) | |
1485 | { | |
1486 | while (1) | |
b6d6a51c | 1487 | { |
facdc750 RS |
1488 | /* Try matching at position POS. */ |
1489 | int this_pos = pos - len; | |
1490 | int this_len = len; | |
1491 | unsigned char *p = pat; | |
1492 | ||
1493 | if (pos - len < lim) | |
1494 | goto stop; | |
1495 | ||
1496 | while (this_len > 0) | |
1497 | { | |
1498 | int pat_ch = *p++; | |
1499 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1500 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1501 | |
1502 | if (buf_ch != pat_ch) | |
1503 | break; | |
ab228c24 RS |
1504 | this_len--; |
1505 | this_pos++; | |
facdc750 RS |
1506 | } |
1507 | ||
1508 | if (this_len == 0) | |
b6d6a51c | 1509 | { |
facdc750 RS |
1510 | pos -= len; |
1511 | break; | |
b6d6a51c | 1512 | } |
facdc750 RS |
1513 | |
1514 | pos--; | |
b6d6a51c | 1515 | } |
facdc750 RS |
1516 | |
1517 | n++; | |
b6d6a51c | 1518 | } |
facdc750 RS |
1519 | |
1520 | stop: | |
1521 | if (n == 0) | |
aff2ce94 | 1522 | { |
ab228c24 RS |
1523 | if (forward) |
1524 | set_search_regs ((multibyte ? pos_byte : pos) - len_byte, len_byte); | |
1525 | else | |
1526 | set_search_regs (multibyte ? pos_byte : pos, len_byte); | |
aff2ce94 RS |
1527 | |
1528 | return pos; | |
1529 | } | |
facdc750 RS |
1530 | else if (n > 0) |
1531 | return -n; | |
1532 | else | |
1533 | return n; | |
1534 | } | |
1535 | \f | |
1536 | /* Do Boyer-Moore search N times for the string PAT, | |
1537 | whose length is LEN/LEN_BYTE, | |
1538 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1539 | DIRECTION says which direction we search in. | |
1540 | TRT and INVERSE_TRT are translation tables. | |
1541 | ||
1542 | This kind of search works if all the characters in PAT that have | |
1543 | nontrivial translation are the same aside from the last byte. This | |
1544 | makes it possible to translate just the last byte of a character, | |
1545 | and do so after just a simple test of the context. | |
1546 | ||
1547 | If that criterion is not satisfied, do not call this function. */ | |
1548 | ||
1549 | static int | |
1550 | boyer_moore (n, base_pat, len, len_byte, trt, inverse_trt, | |
aff2ce94 | 1551 | pos, pos_byte, lim, lim_byte, charset_base) |
facdc750 RS |
1552 | int n; |
1553 | unsigned char *base_pat; | |
1554 | int len, len_byte; | |
1555 | Lisp_Object trt; | |
1556 | Lisp_Object inverse_trt; | |
1557 | int pos, pos_byte; | |
1558 | int lim, lim_byte; | |
aff2ce94 | 1559 | int charset_base; |
facdc750 RS |
1560 | { |
1561 | int direction = ((n > 0) ? 1 : -1); | |
1562 | register int dirlen; | |
a968f437 | 1563 | int infinity, limit, stride_for_teases = 0; |
facdc750 RS |
1564 | register int *BM_tab; |
1565 | int *BM_tab_base; | |
177c0ea7 | 1566 | register unsigned char *cursor, *p_limit; |
facdc750 | 1567 | register int i, j; |
cb6792d2 | 1568 | unsigned char *pat, *pat_end; |
facdc750 RS |
1569 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); |
1570 | ||
1571 | unsigned char simple_translate[0400]; | |
6bbd7a29 GM |
1572 | int translate_prev_byte = 0; |
1573 | int translate_anteprev_byte = 0; | |
facdc750 RS |
1574 | |
1575 | #ifdef C_ALLOCA | |
1576 | int BM_tab_space[0400]; | |
1577 | BM_tab = &BM_tab_space[0]; | |
1578 | #else | |
1579 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
1580 | #endif | |
1581 | /* The general approach is that we are going to maintain that we know */ | |
1582 | /* the first (closest to the present position, in whatever direction */ | |
1583 | /* we're searching) character that could possibly be the last */ | |
1584 | /* (furthest from present position) character of a valid match. We */ | |
1585 | /* advance the state of our knowledge by looking at that character */ | |
1586 | /* and seeing whether it indeed matches the last character of the */ | |
1587 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1588 | /* move our pointer (to putative last characters) as far as is */ | |
1589 | /* logically possible. This amount of movement, which I call a */ | |
1590 | /* stride, will be the length of the pattern if the actual character */ | |
1591 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1592 | /* from the last occurrence of that character to the end of the */ | |
1593 | /* pattern. */ | |
1594 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1595 | /* characters that match the last character. This allows use of only */ | |
1596 | /* a single test, a test for having gone past the end of the */ | |
1597 | /* permissible match region, to test for both possible matches (when */ | |
1598 | /* the stride goes past the end immediately) and failure to */ | |
177c0ea7 | 1599 | /* match (where you get nudged past the end one stride at a time). */ |
facdc750 RS |
1600 | |
1601 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1602 | /* is determined only by the last character of the putative match. */ | |
1603 | /* If that character does not match, we will stride the proper */ | |
1604 | /* distance to propose a match that superimposes it on the last */ | |
1605 | /* instance of a character that matches it (per trt), or misses */ | |
177c0ea7 | 1606 | /* it entirely if there is none. */ |
facdc750 RS |
1607 | |
1608 | dirlen = len_byte * direction; | |
1609 | infinity = dirlen - (lim_byte + pos_byte + len_byte + len_byte) * direction; | |
cb6792d2 RS |
1610 | |
1611 | /* Record position after the end of the pattern. */ | |
1612 | pat_end = base_pat + len_byte; | |
1613 | /* BASE_PAT points to a character that we start scanning from. | |
1614 | It is the first character in a forward search, | |
1615 | the last character in a backward search. */ | |
facdc750 | 1616 | if (direction < 0) |
cb6792d2 RS |
1617 | base_pat = pat_end - 1; |
1618 | ||
facdc750 RS |
1619 | BM_tab_base = BM_tab; |
1620 | BM_tab += 0400; | |
1621 | j = dirlen; /* to get it in a register */ | |
1622 | /* A character that does not appear in the pattern induces a */ | |
1623 | /* stride equal to the pattern length. */ | |
1624 | while (BM_tab_base != BM_tab) | |
1625 | { | |
1626 | *--BM_tab = j; | |
1627 | *--BM_tab = j; | |
1628 | *--BM_tab = j; | |
1629 | *--BM_tab = j; | |
1630 | } | |
1631 | ||
1632 | /* We use this for translation, instead of TRT itself. | |
1633 | We fill this in to handle the characters that actually | |
1634 | occur in the pattern. Others don't matter anyway! */ | |
1635 | bzero (simple_translate, sizeof simple_translate); | |
1636 | for (i = 0; i < 0400; i++) | |
1637 | simple_translate[i] = i; | |
1638 | ||
1639 | i = 0; | |
1640 | while (i != infinity) | |
1641 | { | |
cb6792d2 | 1642 | unsigned char *ptr = base_pat + i; |
facdc750 RS |
1643 | i += direction; |
1644 | if (i == dirlen) | |
1645 | i = infinity; | |
1646 | if (! NILP (trt)) | |
ca1d1d23 | 1647 | { |
facdc750 | 1648 | int ch; |
aff2ce94 | 1649 | int untranslated; |
facdc750 RS |
1650 | int this_translated = 1; |
1651 | ||
1652 | if (multibyte | |
cb6792d2 RS |
1653 | /* Is *PTR the last byte of a character? */ |
1654 | && (pat_end - ptr == 1 || CHAR_HEAD_P (ptr[1]))) | |
ca1d1d23 | 1655 | { |
facdc750 RS |
1656 | unsigned char *charstart = ptr; |
1657 | while (! CHAR_HEAD_P (*charstart)) | |
1658 | charstart--; | |
aff2ce94 | 1659 | untranslated = STRING_CHAR (charstart, ptr - charstart + 1); |
6397418a | 1660 | if (charset_base == (untranslated & ~CHAR_FIELD3_MASK)) |
facdc750 | 1661 | { |
ab228c24 | 1662 | TRANSLATE (ch, trt, untranslated); |
aff2ce94 RS |
1663 | if (! CHAR_HEAD_P (*ptr)) |
1664 | { | |
1665 | translate_prev_byte = ptr[-1]; | |
1666 | if (! CHAR_HEAD_P (translate_prev_byte)) | |
1667 | translate_anteprev_byte = ptr[-2]; | |
1668 | } | |
facdc750 | 1669 | } |
aff2ce94 | 1670 | else |
ab228c24 RS |
1671 | { |
1672 | this_translated = 0; | |
1673 | ch = *ptr; | |
1674 | } | |
ca1d1d23 | 1675 | } |
facdc750 | 1676 | else if (!multibyte) |
aff2ce94 | 1677 | TRANSLATE (ch, trt, *ptr); |
ca1d1d23 JB |
1678 | else |
1679 | { | |
facdc750 RS |
1680 | ch = *ptr; |
1681 | this_translated = 0; | |
ca1d1d23 | 1682 | } |
facdc750 | 1683 | |
ab228c24 RS |
1684 | if (ch > 0400) |
1685 | j = ((unsigned char) ch) | 0200; | |
1686 | else | |
1687 | j = (unsigned char) ch; | |
1688 | ||
facdc750 RS |
1689 | if (i == infinity) |
1690 | stride_for_teases = BM_tab[j]; | |
ab228c24 | 1691 | |
facdc750 RS |
1692 | BM_tab[j] = dirlen - i; |
1693 | /* A translation table is accompanied by its inverse -- see */ | |
177c0ea7 | 1694 | /* comment following downcase_table for details */ |
facdc750 | 1695 | if (this_translated) |
ab228c24 RS |
1696 | { |
1697 | int starting_ch = ch; | |
1698 | int starting_j = j; | |
1699 | while (1) | |
1700 | { | |
1701 | TRANSLATE (ch, inverse_trt, ch); | |
1702 | if (ch > 0400) | |
1703 | j = ((unsigned char) ch) | 0200; | |
1704 | else | |
1705 | j = (unsigned char) ch; | |
1706 | ||
1707 | /* For all the characters that map into CH, | |
1708 | set up simple_translate to map the last byte | |
1709 | into STARTING_J. */ | |
1710 | simple_translate[j] = starting_j; | |
1711 | if (ch == starting_ch) | |
1712 | break; | |
1713 | BM_tab[j] = dirlen - i; | |
1714 | } | |
1715 | } | |
facdc750 RS |
1716 | } |
1717 | else | |
1718 | { | |
1719 | j = *ptr; | |
1720 | ||
1721 | if (i == infinity) | |
1722 | stride_for_teases = BM_tab[j]; | |
1723 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1724 | } |
facdc750 RS |
1725 | /* stride_for_teases tells how much to stride if we get a */ |
1726 | /* match on the far character but are subsequently */ | |
1727 | /* disappointed, by recording what the stride would have been */ | |
1728 | /* for that character if the last character had been */ | |
1729 | /* different. */ | |
1730 | } | |
1731 | infinity = dirlen - infinity; | |
1732 | pos_byte += dirlen - ((direction > 0) ? direction : 0); | |
1733 | /* loop invariant - POS_BYTE points at where last char (first | |
1734 | char if reverse) of pattern would align in a possible match. */ | |
1735 | while (n != 0) | |
1736 | { | |
1737 | int tail_end; | |
1738 | unsigned char *tail_end_ptr; | |
1739 | ||
1740 | /* It's been reported that some (broken) compiler thinks that | |
1741 | Boolean expressions in an arithmetic context are unsigned. | |
1742 | Using an explicit ?1:0 prevents this. */ | |
1743 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1744 | < 0) | |
1745 | return (n * (0 - direction)); | |
1746 | /* First we do the part we can by pointers (maybe nothing) */ | |
1747 | QUIT; | |
1748 | pat = base_pat; | |
1749 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1750 | if (direction > 0) |
1751 | { | |
1752 | limit = BUFFER_CEILING_OF (limit); | |
1753 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1754 | can take on without hitting edge of buffer or the gap. */ | |
1755 | limit = min (limit, pos_byte + 20000); | |
1756 | limit = min (limit, lim_byte - 1); | |
1757 | } | |
1758 | else | |
1759 | { | |
1760 | limit = BUFFER_FLOOR_OF (limit); | |
1761 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1762 | can take on without hitting edge of buffer or the gap. */ | |
1763 | limit = max (limit, pos_byte - 20000); | |
1764 | limit = max (limit, lim_byte); | |
1765 | } | |
facdc750 RS |
1766 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1767 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1768 | ||
1769 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1770 | { |
facdc750 RS |
1771 | unsigned char *p2; |
1772 | ||
1773 | p_limit = BYTE_POS_ADDR (limit); | |
1774 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
1775 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ | |
1776 | while (1) /* use one cursor setting as long as i can */ | |
ca1d1d23 | 1777 | { |
facdc750 | 1778 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1779 | { |
facdc750 RS |
1780 | /* Use signed comparison if appropriate |
1781 | to make cursor+infinity sure to be > p_limit. | |
1782 | Assuming that the buffer lies in a range of addresses | |
1783 | that are all "positive" (as ints) or all "negative", | |
1784 | either kind of comparison will work as long | |
1785 | as we don't step by infinity. So pick the kind | |
1786 | that works when we do step by infinity. */ | |
1787 | if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit) | |
1788 | while ((EMACS_INT) cursor <= (EMACS_INT) p_limit) | |
1789 | cursor += BM_tab[*cursor]; | |
ca1d1d23 | 1790 | else |
facdc750 RS |
1791 | while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit) |
1792 | cursor += BM_tab[*cursor]; | |
1793 | } | |
1794 | else | |
1795 | { | |
1796 | if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit) | |
1797 | while ((EMACS_INT) cursor >= (EMACS_INT) p_limit) | |
1798 | cursor += BM_tab[*cursor]; | |
1799 | else | |
1800 | while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit) | |
1801 | cursor += BM_tab[*cursor]; | |
1802 | } | |
ca1d1d23 | 1803 | /* If you are here, cursor is beyond the end of the searched region. */ |
facdc750 RS |
1804 | /* This can happen if you match on the far character of the pattern, */ |
1805 | /* because the "stride" of that character is infinity, a number able */ | |
1806 | /* to throw you well beyond the end of the search. It can also */ | |
1807 | /* happen if you fail to match within the permitted region and would */ | |
1808 | /* otherwise try a character beyond that region */ | |
1809 | if ((cursor - p_limit) * direction <= len_byte) | |
1810 | break; /* a small overrun is genuine */ | |
1811 | cursor -= infinity; /* large overrun = hit */ | |
1812 | i = dirlen - direction; | |
1813 | if (! NILP (trt)) | |
1814 | { | |
1815 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1816 | { |
facdc750 RS |
1817 | int ch; |
1818 | cursor -= direction; | |
1819 | /* Translate only the last byte of a character. */ | |
1820 | if (! multibyte | |
1821 | || ((cursor == tail_end_ptr | |
1822 | || CHAR_HEAD_P (cursor[1])) | |
1823 | && (CHAR_HEAD_P (cursor[0]) | |
1824 | || (translate_prev_byte == cursor[-1] | |
1825 | && (CHAR_HEAD_P (translate_prev_byte) | |
1826 | || translate_anteprev_byte == cursor[-2]))))) | |
1827 | ch = simple_translate[*cursor]; | |
1828 | else | |
1829 | ch = *cursor; | |
1830 | if (pat[i] != ch) | |
1831 | break; | |
ca1d1d23 | 1832 | } |
facdc750 RS |
1833 | } |
1834 | else | |
1835 | { | |
1836 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1837 | { |
facdc750 RS |
1838 | cursor -= direction; |
1839 | if (pat[i] != *cursor) | |
1840 | break; | |
ca1d1d23 | 1841 | } |
facdc750 RS |
1842 | } |
1843 | cursor += dirlen - i - direction; /* fix cursor */ | |
1844 | if (i + direction == 0) | |
1845 | { | |
1846 | int position; | |
0c8533c6 | 1847 | |
facdc750 | 1848 | cursor -= direction; |
1113d9db | 1849 | |
facdc750 RS |
1850 | position = pos_byte + cursor - p2 + ((direction > 0) |
1851 | ? 1 - len_byte : 0); | |
1852 | set_search_regs (position, len_byte); | |
ca325161 | 1853 | |
facdc750 RS |
1854 | if ((n -= direction) != 0) |
1855 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1856 | else |
facdc750 RS |
1857 | return ((direction > 0) |
1858 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1859 | } |
facdc750 RS |
1860 | else |
1861 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1862 | } |
facdc750 RS |
1863 | pos_byte += cursor - p2; |
1864 | } | |
1865 | else | |
1866 | /* Now we'll pick up a clump that has to be done the hard */ | |
1867 | /* way because it covers a discontinuity */ | |
1868 | { | |
1869 | limit = ((direction > 0) | |
1870 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1871 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1872 | limit = ((direction > 0) | |
1873 | ? min (limit + len_byte, lim_byte - 1) | |
1874 | : max (limit - len_byte, lim_byte)); | |
1875 | /* LIMIT is now the last value POS_BYTE can have | |
1876 | and still be valid for a possible match. */ | |
1877 | while (1) | |
ca1d1d23 | 1878 | { |
facdc750 RS |
1879 | /* This loop can be coded for space rather than */ |
1880 | /* speed because it will usually run only once. */ | |
1881 | /* (the reach is at most len + 21, and typically */ | |
177c0ea7 | 1882 | /* does not exceed len) */ |
facdc750 RS |
1883 | while ((limit - pos_byte) * direction >= 0) |
1884 | pos_byte += BM_tab[FETCH_BYTE (pos_byte)]; | |
1885 | /* now run the same tests to distinguish going off the */ | |
1886 | /* end, a match or a phony match. */ | |
1887 | if ((pos_byte - limit) * direction <= len_byte) | |
1888 | break; /* ran off the end */ | |
1889 | /* Found what might be a match. | |
1890 | Set POS_BYTE back to last (first if reverse) pos. */ | |
1891 | pos_byte -= infinity; | |
1892 | i = dirlen - direction; | |
1893 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1894 | { |
facdc750 RS |
1895 | int ch; |
1896 | unsigned char *ptr; | |
1897 | pos_byte -= direction; | |
1898 | ptr = BYTE_POS_ADDR (pos_byte); | |
1899 | /* Translate only the last byte of a character. */ | |
1900 | if (! multibyte | |
1901 | || ((ptr == tail_end_ptr | |
1902 | || CHAR_HEAD_P (ptr[1])) | |
1903 | && (CHAR_HEAD_P (ptr[0]) | |
1904 | || (translate_prev_byte == ptr[-1] | |
1905 | && (CHAR_HEAD_P (translate_prev_byte) | |
1906 | || translate_anteprev_byte == ptr[-2]))))) | |
1907 | ch = simple_translate[*ptr]; | |
1908 | else | |
1909 | ch = *ptr; | |
1910 | if (pat[i] != ch) | |
1911 | break; | |
1912 | } | |
1913 | /* Above loop has moved POS_BYTE part or all the way | |
1914 | back to the first pos (last pos if reverse). | |
1915 | Set it once again at the last (first if reverse) char. */ | |
1916 | pos_byte += dirlen - i- direction; | |
1917 | if (i + direction == 0) | |
1918 | { | |
1919 | int position; | |
1920 | pos_byte -= direction; | |
1113d9db | 1921 | |
facdc750 | 1922 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
0c8533c6 | 1923 | |
facdc750 | 1924 | set_search_regs (position, len_byte); |
ca325161 | 1925 | |
facdc750 RS |
1926 | if ((n -= direction) != 0) |
1927 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 1928 | else |
facdc750 RS |
1929 | return ((direction > 0) |
1930 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1931 | } |
facdc750 RS |
1932 | else |
1933 | pos_byte += stride_for_teases; | |
1934 | } | |
1935 | } | |
1936 | /* We have done one clump. Can we continue? */ | |
1937 | if ((lim_byte - pos_byte) * direction < 0) | |
1938 | return ((0 - n) * direction); | |
ca1d1d23 | 1939 | } |
facdc750 | 1940 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 1941 | } |
ca325161 | 1942 | |
fa8ed3e0 | 1943 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
1944 | for the overall match just found in the current buffer. |
1945 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
1946 | |
1947 | static void | |
fa8ed3e0 RS |
1948 | set_search_regs (beg_byte, nbytes) |
1949 | int beg_byte, nbytes; | |
ca325161 | 1950 | { |
a7e4cdde RS |
1951 | int i; |
1952 | ||
ca325161 RS |
1953 | /* Make sure we have registers in which to store |
1954 | the match position. */ | |
1955 | if (search_regs.num_regs == 0) | |
1956 | { | |
2d4a771a RS |
1957 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
1958 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 1959 | search_regs.num_regs = 2; |
ca325161 RS |
1960 | } |
1961 | ||
a7e4cdde RS |
1962 | /* Clear out the other registers. */ |
1963 | for (i = 1; i < search_regs.num_regs; i++) | |
1964 | { | |
1965 | search_regs.start[i] = -1; | |
1966 | search_regs.end[i] = -1; | |
1967 | } | |
1968 | ||
fa8ed3e0 RS |
1969 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
1970 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 1971 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 1972 | } |
ca1d1d23 JB |
1973 | \f |
1974 | /* Given a string of words separated by word delimiters, | |
1975 | compute a regexp that matches those exact words | |
1976 | separated by arbitrary punctuation. */ | |
1977 | ||
1978 | static Lisp_Object | |
1979 | wordify (string) | |
1980 | Lisp_Object string; | |
1981 | { | |
1982 | register unsigned char *p, *o; | |
0c8533c6 | 1983 | register int i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 1984 | Lisp_Object val; |
0c8533c6 RS |
1985 | int prev_c = 0; |
1986 | int adjust; | |
ca1d1d23 | 1987 | |
b7826503 | 1988 | CHECK_STRING (string); |
d5db4077 KR |
1989 | p = SDATA (string); |
1990 | len = SCHARS (string); | |
ca1d1d23 | 1991 | |
0c8533c6 RS |
1992 | for (i = 0, i_byte = 0; i < len; ) |
1993 | { | |
1994 | int c; | |
177c0ea7 | 1995 | |
eb99a8dd | 1996 | FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte); |
0c8533c6 RS |
1997 | |
1998 | if (SYNTAX (c) != Sword) | |
1999 | { | |
2000 | punct_count++; | |
2001 | if (i > 0 && SYNTAX (prev_c) == Sword) | |
2002 | word_count++; | |
2003 | } | |
ca1d1d23 | 2004 | |
0c8533c6 RS |
2005 | prev_c = c; |
2006 | } | |
2007 | ||
2008 | if (SYNTAX (prev_c) == Sword) | |
2009 | word_count++; | |
2010 | if (!word_count) | |
b07b65aa | 2011 | return empty_string; |
0c8533c6 RS |
2012 | |
2013 | adjust = - punct_count + 5 * (word_count - 1) + 4; | |
8a2df937 RS |
2014 | if (STRING_MULTIBYTE (string)) |
2015 | val = make_uninit_multibyte_string (len + adjust, | |
d5db4077 | 2016 | SBYTES (string) |
8a2df937 RS |
2017 | + adjust); |
2018 | else | |
2019 | val = make_uninit_string (len + adjust); | |
ca1d1d23 | 2020 | |
d5db4077 | 2021 | o = SDATA (val); |
ca1d1d23 JB |
2022 | *o++ = '\\'; |
2023 | *o++ = 'b'; | |
1e9582d4 | 2024 | prev_c = 0; |
ca1d1d23 | 2025 | |
1e9582d4 RS |
2026 | for (i = 0, i_byte = 0; i < len; ) |
2027 | { | |
2028 | int c; | |
2029 | int i_byte_orig = i_byte; | |
177c0ea7 | 2030 | |
eb99a8dd | 2031 | FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte); |
1e9582d4 RS |
2032 | |
2033 | if (SYNTAX (c) == Sword) | |
2034 | { | |
5d69fe10 | 2035 | bcopy (SDATA (string) + i_byte_orig, o, |
1e9582d4 RS |
2036 | i_byte - i_byte_orig); |
2037 | o += i_byte - i_byte_orig; | |
2038 | } | |
2039 | else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count) | |
2040 | { | |
2041 | *o++ = '\\'; | |
2042 | *o++ = 'W'; | |
2043 | *o++ = '\\'; | |
2044 | *o++ = 'W'; | |
2045 | *o++ = '*'; | |
2046 | } | |
2047 | ||
2048 | prev_c = c; | |
2049 | } | |
ca1d1d23 JB |
2050 | |
2051 | *o++ = '\\'; | |
2052 | *o++ = 'b'; | |
2053 | ||
2054 | return val; | |
2055 | } | |
2056 | \f | |
2057 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
8c1a1077 PJ |
2058 | "MSearch backward: ", |
2059 | doc: /* Search backward from point for STRING. | |
2060 | Set point to the beginning of the occurrence found, and return point. | |
2061 | An optional second argument bounds the search; it is a buffer position. | |
2062 | The match found must not extend before that position. | |
2063 | Optional third argument, if t, means if fail just return nil (no error). | |
2064 | If not nil and not t, position at limit of search and return nil. | |
2065 | Optional fourth argument is repeat count--search for successive occurrences. | |
2066 | ||
2067 | Search case-sensitivity is determined by the value of the variable | |
2068 | `case-fold-search', which see. | |
2069 | ||
2070 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2071 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2072 | Lisp_Object string, bound, noerror, count; |
2073 | { | |
b819a390 | 2074 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2075 | } |
2076 | ||
6af43974 | 2077 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2078 | doc: /* Search forward from point for STRING. |
2079 | Set point to the end of the occurrence found, and return point. | |
2080 | An optional second argument bounds the search; it is a buffer position. | |
2081 | The match found must not extend after that position. nil is equivalent | |
2082 | to (point-max). | |
2083 | Optional third argument, if t, means if fail just return nil (no error). | |
2084 | If not nil and not t, move to limit of search and return nil. | |
2085 | Optional fourth argument is repeat count--search for successive occurrences. | |
2086 | ||
2087 | Search case-sensitivity is determined by the value of the variable | |
2088 | `case-fold-search', which see. | |
2089 | ||
2090 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2091 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2092 | Lisp_Object string, bound, noerror, count; |
2093 | { | |
b819a390 | 2094 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2095 | } |
2096 | ||
2097 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
8c1a1077 PJ |
2098 | "sWord search backward: ", |
2099 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2100 | Set point to the beginning of the occurrence found, and return point. | |
2101 | An optional second argument bounds the search; it is a buffer position. | |
2102 | The match found must not extend before that position. | |
2103 | Optional third argument, if t, means if fail just return nil (no error). | |
2104 | If not nil and not t, move to limit of search and return nil. | |
2105 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2106 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2107 | Lisp_Object string, bound, noerror, count; |
2108 | { | |
b819a390 | 2109 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2110 | } |
2111 | ||
2112 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
8c1a1077 PJ |
2113 | "sWord search: ", |
2114 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2115 | Set point to the end of the occurrence found, and return point. | |
2116 | An optional second argument bounds the search; it is a buffer position. | |
2117 | The match found must not extend after that position. | |
2118 | Optional third argument, if t, means if fail just return nil (no error). | |
2119 | If not nil and not t, move to limit of search and return nil. | |
2120 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2121 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2122 | Lisp_Object string, bound, noerror, count; |
2123 | { | |
b819a390 | 2124 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
2125 | } |
2126 | ||
2127 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
8c1a1077 PJ |
2128 | "sRE search backward: ", |
2129 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2130 | Set point to the beginning of the match, and return point. | |
2131 | The match found is the one starting last in the buffer | |
2132 | and yet ending before the origin of the search. | |
2133 | An optional second argument bounds the search; it is a buffer position. | |
2134 | The match found must start at or after that position. | |
2135 | Optional third argument, if t, means if fail just return nil (no error). | |
2136 | If not nil and not t, move to limit of search and return nil. | |
2137 | Optional fourth argument is repeat count--search for successive occurrences. | |
2138 | See also the functions `match-beginning', `match-end', `match-string', | |
2139 | and `replace-match'. */) | |
2140 | (regexp, bound, noerror, count) | |
19c0a730 | 2141 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2142 | { |
b819a390 | 2143 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2144 | } |
2145 | ||
2146 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2147 | "sRE search: ", |
2148 | doc: /* Search forward from point for regular expression REGEXP. | |
2149 | Set point to the end of the occurrence found, and return point. | |
2150 | An optional second argument bounds the search; it is a buffer position. | |
2151 | The match found must not extend after that position. | |
2152 | Optional third argument, if t, means if fail just return nil (no error). | |
2153 | If not nil and not t, move to limit of search and return nil. | |
2154 | Optional fourth argument is repeat count--search for successive occurrences. | |
2155 | See also the functions `match-beginning', `match-end', `match-string', | |
2156 | and `replace-match'. */) | |
2157 | (regexp, bound, noerror, count) | |
19c0a730 | 2158 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2159 | { |
b819a390 RS |
2160 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2161 | } | |
2162 | ||
2163 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2164 | "sPosix search backward: ", |
2165 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2166 | Find the longest match in accord with Posix regular expression rules. | |
2167 | Set point to the beginning of the match, and return point. | |
2168 | The match found is the one starting last in the buffer | |
2169 | and yet ending before the origin of the search. | |
2170 | An optional second argument bounds the search; it is a buffer position. | |
2171 | The match found must start at or after that position. | |
2172 | Optional third argument, if t, means if fail just return nil (no error). | |
2173 | If not nil and not t, move to limit of search and return nil. | |
2174 | Optional fourth argument is repeat count--search for successive occurrences. | |
2175 | See also the functions `match-beginning', `match-end', `match-string', | |
2176 | and `replace-match'. */) | |
2177 | (regexp, bound, noerror, count) | |
b819a390 RS |
2178 | Lisp_Object regexp, bound, noerror, count; |
2179 | { | |
2180 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2181 | } | |
2182 | ||
2183 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2184 | "sPosix search: ", |
2185 | doc: /* Search forward from point for regular expression REGEXP. | |
2186 | Find the longest match in accord with Posix regular expression rules. | |
2187 | Set point to the end of the occurrence found, and return point. | |
2188 | An optional second argument bounds the search; it is a buffer position. | |
2189 | The match found must not extend after that position. | |
2190 | Optional third argument, if t, means if fail just return nil (no error). | |
2191 | If not nil and not t, move to limit of search and return nil. | |
2192 | Optional fourth argument is repeat count--search for successive occurrences. | |
2193 | See also the functions `match-beginning', `match-end', `match-string', | |
2194 | and `replace-match'. */) | |
2195 | (regexp, bound, noerror, count) | |
b819a390 RS |
2196 | Lisp_Object regexp, bound, noerror, count; |
2197 | { | |
2198 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2199 | } |
2200 | \f | |
d7a5ad5f | 2201 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2202 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2203 | Leave point at the end of the replacement text. |
2204 | ||
8c1a1077 PJ |
2205 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. |
2206 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2207 | based on the replaced text. | |
2208 | If the replaced text has only capital letters | |
2209 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
4dd0c271 RS |
2210 | Otherwise if all words are capitalized in the replaced text, |
2211 | capitalize each word in NEWTEXT. | |
8c1a1077 PJ |
2212 | |
2213 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2214 | Otherwise treat `\\' as special: | |
2215 | `\\&' in NEWTEXT means substitute original matched text. | |
2216 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2217 | If Nth parens didn't match, substitute nothing. | |
2218 | `\\\\' means insert one `\\'. | |
4dd0c271 RS |
2219 | Case conversion does not apply to these substitutions. |
2220 | ||
8c1a1077 | 2221 | FIXEDCASE and LITERAL are optional arguments. |
8c1a1077 PJ |
2222 | |
2223 | The optional fourth argument STRING can be a string to modify. | |
2224 | This is meaningful when the previous match was done against STRING, | |
2225 | using `string-match'. When used this way, `replace-match' | |
2226 | creates and returns a new string made by copying STRING and replacing | |
2227 | the part of STRING that was matched. | |
2228 | ||
2229 | The optional fifth argument SUBEXP specifies a subexpression; | |
2230 | it says to replace just that subexpression with NEWTEXT, | |
2231 | rather than replacing the entire matched text. | |
2232 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2233 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2234 | NEWTEXT in place of subexp N. | |
2235 | This is useful only after a regular expression search or match, | |
2236 | since only regular expressions have distinguished subexpressions. */) | |
2237 | (newtext, fixedcase, literal, string, subexp) | |
d7a5ad5f | 2238 | Lisp_Object newtext, fixedcase, literal, string, subexp; |
ca1d1d23 JB |
2239 | { |
2240 | enum { nochange, all_caps, cap_initial } case_action; | |
ac3b28b1 | 2241 | register int pos, pos_byte; |
ca1d1d23 | 2242 | int some_multiletter_word; |
97832bd0 | 2243 | int some_lowercase; |
73dc8771 | 2244 | int some_uppercase; |
208767c3 | 2245 | int some_nonuppercase_initial; |
ca1d1d23 | 2246 | register int c, prevc; |
d7a5ad5f | 2247 | int sub; |
3e18eecf | 2248 | int opoint, newpoint; |
ca1d1d23 | 2249 | |
b7826503 | 2250 | CHECK_STRING (newtext); |
ca1d1d23 | 2251 | |
080c45fd | 2252 | if (! NILP (string)) |
b7826503 | 2253 | CHECK_STRING (string); |
080c45fd | 2254 | |
ca1d1d23 JB |
2255 | case_action = nochange; /* We tried an initialization */ |
2256 | /* but some C compilers blew it */ | |
4746118a JB |
2257 | |
2258 | if (search_regs.num_regs <= 0) | |
2259 | error ("replace-match called before any match found"); | |
2260 | ||
d7a5ad5f RS |
2261 | if (NILP (subexp)) |
2262 | sub = 0; | |
2263 | else | |
2264 | { | |
b7826503 | 2265 | CHECK_NUMBER (subexp); |
d7a5ad5f RS |
2266 | sub = XINT (subexp); |
2267 | if (sub < 0 || sub >= search_regs.num_regs) | |
2268 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2269 | } | |
2270 | ||
080c45fd RS |
2271 | if (NILP (string)) |
2272 | { | |
d7a5ad5f RS |
2273 | if (search_regs.start[sub] < BEGV |
2274 | || search_regs.start[sub] > search_regs.end[sub] | |
2275 | || search_regs.end[sub] > ZV) | |
2276 | args_out_of_range (make_number (search_regs.start[sub]), | |
2277 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2278 | } |
2279 | else | |
2280 | { | |
d7a5ad5f RS |
2281 | if (search_regs.start[sub] < 0 |
2282 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2283 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2284 | args_out_of_range (make_number (search_regs.start[sub]), |
2285 | make_number (search_regs.end[sub])); | |
080c45fd | 2286 | } |
ca1d1d23 JB |
2287 | |
2288 | if (NILP (fixedcase)) | |
2289 | { | |
2290 | /* Decide how to casify by examining the matched text. */ | |
ac3b28b1 | 2291 | int last; |
ca1d1d23 | 2292 | |
ac3b28b1 KH |
2293 | pos = search_regs.start[sub]; |
2294 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2295 | |
2296 | if (NILP (string)) | |
ac3b28b1 | 2297 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2298 | else |
ac3b28b1 | 2299 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2300 | |
ca1d1d23 JB |
2301 | prevc = '\n'; |
2302 | case_action = all_caps; | |
2303 | ||
2304 | /* some_multiletter_word is set nonzero if any original word | |
2305 | is more than one letter long. */ | |
2306 | some_multiletter_word = 0; | |
97832bd0 | 2307 | some_lowercase = 0; |
208767c3 | 2308 | some_nonuppercase_initial = 0; |
73dc8771 | 2309 | some_uppercase = 0; |
ca1d1d23 | 2310 | |
ac3b28b1 | 2311 | while (pos < last) |
ca1d1d23 | 2312 | { |
080c45fd | 2313 | if (NILP (string)) |
ac3b28b1 KH |
2314 | { |
2315 | c = FETCH_CHAR (pos_byte); | |
2316 | INC_BOTH (pos, pos_byte); | |
2317 | } | |
080c45fd | 2318 | else |
ac3b28b1 | 2319 | FETCH_STRING_CHAR_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2320 | |
ca1d1d23 JB |
2321 | if (LOWERCASEP (c)) |
2322 | { | |
2323 | /* Cannot be all caps if any original char is lower case */ | |
2324 | ||
97832bd0 | 2325 | some_lowercase = 1; |
ca1d1d23 | 2326 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2327 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2328 | else |
2329 | some_multiletter_word = 1; | |
2330 | } | |
2331 | else if (!NOCASEP (c)) | |
2332 | { | |
73dc8771 | 2333 | some_uppercase = 1; |
97832bd0 | 2334 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2335 | ; |
97832bd0 | 2336 | else |
ca1d1d23 JB |
2337 | some_multiletter_word = 1; |
2338 | } | |
208767c3 RS |
2339 | else |
2340 | { | |
2341 | /* If the initial is a caseless word constituent, | |
2342 | treat that like a lowercase initial. */ | |
2343 | if (SYNTAX (prevc) != Sword) | |
2344 | some_nonuppercase_initial = 1; | |
2345 | } | |
ca1d1d23 JB |
2346 | |
2347 | prevc = c; | |
2348 | } | |
2349 | ||
97832bd0 RS |
2350 | /* Convert to all caps if the old text is all caps |
2351 | and has at least one multiletter word. */ | |
2352 | if (! some_lowercase && some_multiletter_word) | |
2353 | case_action = all_caps; | |
c4d460ce | 2354 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2355 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2356 | case_action = cap_initial; |
208767c3 | 2357 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2358 | /* Should x -> yz, operating on X, give Yz or YZ? |
2359 | We'll assume the latter. */ | |
2360 | case_action = all_caps; | |
97832bd0 RS |
2361 | else |
2362 | case_action = nochange; | |
ca1d1d23 JB |
2363 | } |
2364 | ||
080c45fd RS |
2365 | /* Do replacement in a string. */ |
2366 | if (!NILP (string)) | |
2367 | { | |
2368 | Lisp_Object before, after; | |
2369 | ||
2370 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2371 | make_number (search_regs.start[sub])); |
2372 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2373 | |
636a5e28 RS |
2374 | /* Substitute parts of the match into NEWTEXT |
2375 | if desired. */ | |
080c45fd RS |
2376 | if (NILP (literal)) |
2377 | { | |
d131e79c RS |
2378 | int lastpos = 0; |
2379 | int lastpos_byte = 0; | |
080c45fd RS |
2380 | /* We build up the substituted string in ACCUM. */ |
2381 | Lisp_Object accum; | |
2382 | Lisp_Object middle; | |
d5db4077 | 2383 | int length = SBYTES (newtext); |
080c45fd RS |
2384 | |
2385 | accum = Qnil; | |
2386 | ||
ac3b28b1 | 2387 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd RS |
2388 | { |
2389 | int substart = -1; | |
6bbd7a29 | 2390 | int subend = 0; |
1e79ec24 | 2391 | int delbackslash = 0; |
080c45fd | 2392 | |
0c8533c6 RS |
2393 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2394 | ||
080c45fd RS |
2395 | if (c == '\\') |
2396 | { | |
0c8533c6 | 2397 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2398 | |
080c45fd RS |
2399 | if (c == '&') |
2400 | { | |
d7a5ad5f RS |
2401 | substart = search_regs.start[sub]; |
2402 | subend = search_regs.end[sub]; | |
080c45fd | 2403 | } |
5fbbc83b | 2404 | else if (c >= '1' && c <= '9') |
080c45fd | 2405 | { |
5fbbc83b RS |
2406 | if (search_regs.start[c - '0'] >= 0 |
2407 | && c <= search_regs.num_regs + '0') | |
080c45fd RS |
2408 | { |
2409 | substart = search_regs.start[c - '0']; | |
2410 | subend = search_regs.end[c - '0']; | |
2411 | } | |
5fbbc83b RS |
2412 | else |
2413 | { | |
2414 | /* If that subexp did not match, | |
2415 | replace \\N with nothing. */ | |
2416 | substart = 0; | |
2417 | subend = 0; | |
2418 | } | |
080c45fd | 2419 | } |
1e79ec24 KH |
2420 | else if (c == '\\') |
2421 | delbackslash = 1; | |
636a5e28 RS |
2422 | else |
2423 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2424 | } |
2425 | if (substart >= 0) | |
2426 | { | |
d131e79c RS |
2427 | if (pos - 2 != lastpos) |
2428 | middle = substring_both (newtext, lastpos, | |
2429 | lastpos_byte, | |
2430 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2431 | else |
2432 | middle = Qnil; | |
2433 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2434 | Fsubstring (string, |
2435 | make_number (substart), | |
080c45fd RS |
2436 | make_number (subend))); |
2437 | lastpos = pos; | |
0c8533c6 | 2438 | lastpos_byte = pos_byte; |
080c45fd | 2439 | } |
1e79ec24 KH |
2440 | else if (delbackslash) |
2441 | { | |
d131e79c RS |
2442 | middle = substring_both (newtext, lastpos, |
2443 | lastpos_byte, | |
2444 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2445 | |
1e79ec24 KH |
2446 | accum = concat2 (accum, middle); |
2447 | lastpos = pos; | |
0c8533c6 | 2448 | lastpos_byte = pos_byte; |
1e79ec24 | 2449 | } |
080c45fd RS |
2450 | } |
2451 | ||
d131e79c RS |
2452 | if (pos != lastpos) |
2453 | middle = substring_both (newtext, lastpos, | |
2454 | lastpos_byte, | |
0c8533c6 | 2455 | pos, pos_byte); |
080c45fd RS |
2456 | else |
2457 | middle = Qnil; | |
2458 | ||
2459 | newtext = concat2 (accum, middle); | |
2460 | } | |
2461 | ||
636a5e28 | 2462 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2463 | if (case_action == all_caps) |
2464 | newtext = Fupcase (newtext); | |
2465 | else if (case_action == cap_initial) | |
2b2eead9 | 2466 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2467 | |
2468 | return concat3 (before, newtext, after); | |
2469 | } | |
2470 | ||
09c4719e | 2471 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2472 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2473 | opoint = PT - ZV; |
9160906f RS |
2474 | else if (PT > search_regs.start[sub]) |
2475 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2476 | else |
2477 | opoint = PT; | |
2478 | ||
886ed6ec RS |
2479 | /* If we want non-literal replacement, |
2480 | perform substitution on the replacement string. */ | |
2481 | if (NILP (literal)) | |
ca1d1d23 | 2482 | { |
d5db4077 | 2483 | int length = SBYTES (newtext); |
68e69fbd RS |
2484 | unsigned char *substed; |
2485 | int substed_alloc_size, substed_len; | |
3bc25e52 KH |
2486 | int buf_multibyte = !NILP (current_buffer->enable_multibyte_characters); |
2487 | int str_multibyte = STRING_MULTIBYTE (newtext); | |
2488 | Lisp_Object rev_tbl; | |
886ed6ec | 2489 | int really_changed = 0; |
3bc25e52 KH |
2490 | |
2491 | rev_tbl= (!buf_multibyte && CHAR_TABLE_P (Vnonascii_translation_table) | |
2492 | ? Fchar_table_extra_slot (Vnonascii_translation_table, | |
2493 | make_number (0)) | |
2494 | : Qnil); | |
ac3b28b1 | 2495 | |
68e69fbd RS |
2496 | substed_alloc_size = length * 2 + 100; |
2497 | substed = (unsigned char *) xmalloc (substed_alloc_size + 1); | |
2498 | substed_len = 0; | |
2499 | ||
3bc25e52 KH |
2500 | /* Go thru NEWTEXT, producing the actual text to insert in |
2501 | SUBSTED while adjusting multibyteness to that of the current | |
2502 | buffer. */ | |
ca1d1d23 | 2503 | |
ac3b28b1 | 2504 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2505 | { |
68e69fbd | 2506 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
f8ce8a0d GM |
2507 | unsigned char *add_stuff = NULL; |
2508 | int add_len = 0; | |
68e69fbd | 2509 | int idx = -1; |
9a76659d | 2510 | |
3bc25e52 KH |
2511 | if (str_multibyte) |
2512 | { | |
eb99a8dd | 2513 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 KH |
2514 | if (!buf_multibyte) |
2515 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2516 | } | |
2517 | else | |
2518 | { | |
2519 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2520 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2521 | if (buf_multibyte) |
2522 | c = unibyte_char_to_multibyte (c); | |
2523 | } | |
ac3b28b1 | 2524 | |
68e69fbd RS |
2525 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2526 | or set IDX to a match index, which means put that part | |
2527 | of the buffer text into SUBSTED. */ | |
2528 | ||
ca1d1d23 JB |
2529 | if (c == '\\') |
2530 | { | |
886ed6ec RS |
2531 | really_changed = 1; |
2532 | ||
3bc25e52 KH |
2533 | if (str_multibyte) |
2534 | { | |
eb99a8dd KH |
2535 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2536 | pos, pos_byte); | |
3bc25e52 KH |
2537 | if (!buf_multibyte && !SINGLE_BYTE_CHAR_P (c)) |
2538 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2539 | } | |
2540 | else | |
2541 | { | |
d5db4077 | 2542 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2543 | if (buf_multibyte) |
2544 | c = unibyte_char_to_multibyte (c); | |
2545 | } | |
2546 | ||
ca1d1d23 | 2547 | if (c == '&') |
68e69fbd | 2548 | idx = sub; |
78445046 | 2549 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2550 | { |
2551 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2552 | idx = c - '0'; |
ca1d1d23 | 2553 | } |
636a5e28 | 2554 | else if (c == '\\') |
68e69fbd | 2555 | add_len = 1, add_stuff = "\\"; |
636a5e28 | 2556 | else |
3bc25e52 KH |
2557 | { |
2558 | xfree (substed); | |
2559 | error ("Invalid use of `\\' in replacement text"); | |
2560 | } | |
ca1d1d23 JB |
2561 | } |
2562 | else | |
68e69fbd RS |
2563 | { |
2564 | add_len = CHAR_STRING (c, str); | |
2565 | add_stuff = str; | |
2566 | } | |
2567 | ||
2568 | /* If we want to copy part of a previous match, | |
2569 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2570 | if (idx >= 0) | |
2571 | { | |
2572 | int begbyte = CHAR_TO_BYTE (search_regs.start[idx]); | |
2573 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; | |
2574 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2575 | move_gap (search_regs.start[idx]); | |
2576 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2577 | } | |
2578 | ||
2579 | /* Now the stuff we want to add to SUBSTED | |
2580 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2581 | ||
2582 | /* Make sure SUBSTED is big enough. */ | |
2583 | if (substed_len + add_len >= substed_alloc_size) | |
2584 | { | |
2585 | substed_alloc_size = substed_len + add_len + 500; | |
2586 | substed = (unsigned char *) xrealloc (substed, | |
2587 | substed_alloc_size + 1); | |
2588 | } | |
2589 | ||
2590 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2591 | if (add_stuff) |
2592 | { | |
2593 | bcopy (add_stuff, substed + substed_len, add_len); | |
2594 | substed_len += add_len; | |
2595 | } | |
ca1d1d23 | 2596 | } |
68e69fbd | 2597 | |
886ed6ec | 2598 | if (really_changed) |
80460525 KH |
2599 | { |
2600 | if (buf_multibyte) | |
2601 | { | |
2602 | int nchars = multibyte_chars_in_text (substed, substed_len); | |
68e69fbd | 2603 | |
80460525 KH |
2604 | newtext = make_multibyte_string (substed, nchars, substed_len); |
2605 | } | |
2606 | else | |
2607 | newtext = make_unibyte_string (substed, substed_len); | |
2608 | } | |
68e69fbd | 2609 | xfree (substed); |
ca1d1d23 JB |
2610 | } |
2611 | ||
886ed6ec RS |
2612 | /* Replace the old text with the new in the cleanest possible way. */ |
2613 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2614 | newtext, 1, 0, 1); | |
d5db4077 | 2615 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2616 | |
2617 | if (case_action == all_caps) | |
886ed6ec RS |
2618 | Fupcase_region (make_number (search_regs.start[sub]), |
2619 | make_number (newpoint)); | |
ca1d1d23 | 2620 | else if (case_action == cap_initial) |
886ed6ec RS |
2621 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2622 | make_number (newpoint)); | |
3e18eecf | 2623 | |
98e942e0 RS |
2624 | /* Adjust search data for this change. */ |
2625 | { | |
5b88a2c5 | 2626 | int oldend = search_regs.end[sub]; |
41c01205 | 2627 | int oldstart = search_regs.start[sub]; |
98e942e0 RS |
2628 | int change = newpoint - search_regs.end[sub]; |
2629 | int i; | |
2630 | ||
2631 | for (i = 0; i < search_regs.num_regs; i++) | |
2632 | { | |
41c01205 | 2633 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2634 | search_regs.start[i] += change; |
41c01205 DK |
2635 | else if (search_regs.start[i] > oldstart) |
2636 | search_regs.start[i] = oldstart; | |
2637 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2638 | search_regs.end[i] += change; |
41c01205 DK |
2639 | else if (search_regs.end[i] > oldstart) |
2640 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2641 | } |
2642 | } | |
2643 | ||
b0eba991 | 2644 | /* Put point back where it was in the text. */ |
8d808a65 | 2645 | if (opoint <= 0) |
fa8ed3e0 | 2646 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2647 | else |
fa8ed3e0 | 2648 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2649 | |
2650 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2651 | move_if_not_intangible (newpoint); |
177c0ea7 | 2652 | |
ca1d1d23 JB |
2653 | return Qnil; |
2654 | } | |
2655 | \f | |
2656 | static Lisp_Object | |
2657 | match_limit (num, beginningp) | |
2658 | Lisp_Object num; | |
2659 | int beginningp; | |
2660 | { | |
2661 | register int n; | |
2662 | ||
b7826503 | 2663 | CHECK_NUMBER (num); |
ca1d1d23 | 2664 | n = XINT (num); |
f90a5bf5 | 2665 | if (n < 0) |
bd2cbd56 | 2666 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2667 | if (search_regs.num_regs <= 0) |
2668 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2669 | if (n >= search_regs.num_regs |
4746118a | 2670 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2671 | return Qnil; |
2672 | return (make_number ((beginningp) ? search_regs.start[n] | |
2673 | : search_regs.end[n])); | |
2674 | } | |
2675 | ||
2676 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
8c1a1077 PJ |
2677 | doc: /* Return position of start of text matched by last search. |
2678 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2679 | regexp. | |
2680 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2681 | SUBEXP pairs. | |
2682 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2683 | (subexp) | |
5806161b | 2684 | Lisp_Object subexp; |
ca1d1d23 | 2685 | { |
5806161b | 2686 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2687 | } |
2688 | ||
2689 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
8c1a1077 PJ |
2690 | doc: /* Return position of end of text matched by last search. |
2691 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2692 | regexp. | |
2693 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2694 | SUBEXP pairs. | |
2695 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2696 | (subexp) | |
5806161b | 2697 | Lisp_Object subexp; |
ca1d1d23 | 2698 | { |
5806161b | 2699 | return match_limit (subexp, 0); |
177c0ea7 | 2700 | } |
ca1d1d23 | 2701 | |
56256c2a | 2702 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 2, 0, |
8c1a1077 PJ |
2703 | doc: /* Return a list containing all info on what the last search matched. |
2704 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2705 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2706 | if the last match was on a buffer; integers or nil if a string was matched. | |
2707 | Use `store-match-data' to reinstate the data in this list. | |
2708 | ||
41c01205 DK |
2709 | If INTEGERS (the optional first argument) is non-nil, always use |
2710 | integers \(rather than markers) to represent buffer positions. In | |
2711 | this case, and if the last match was in a buffer, the buffer will get | |
2712 | stored as one additional element at the end of the list. | |
2713 | ||
8c1a1077 | 2714 | If REUSE is a list, reuse it as part of the value. If REUSE is long enough |
140a6b7e KS |
2715 | to hold all the values, and if INTEGERS is non-nil, no consing is done. |
2716 | ||
2717 | Return value is undefined if the last search failed. */) | |
8c1a1077 | 2718 | (integers, reuse) |
56256c2a | 2719 | Lisp_Object integers, reuse; |
ca1d1d23 | 2720 | { |
56256c2a | 2721 | Lisp_Object tail, prev; |
4746118a | 2722 | Lisp_Object *data; |
ca1d1d23 JB |
2723 | int i, len; |
2724 | ||
daa37602 | 2725 | if (NILP (last_thing_searched)) |
c36bcf1b | 2726 | return Qnil; |
daa37602 | 2727 | |
6bbd7a29 GM |
2728 | prev = Qnil; |
2729 | ||
41c01205 | 2730 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs + 1) |
4746118a JB |
2731 | * sizeof (Lisp_Object)); |
2732 | ||
41c01205 | 2733 | len = 0; |
4746118a | 2734 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2735 | { |
2736 | int start = search_regs.start[i]; | |
2737 | if (start >= 0) | |
2738 | { | |
56256c2a RS |
2739 | if (EQ (last_thing_searched, Qt) |
2740 | || ! NILP (integers)) | |
ca1d1d23 | 2741 | { |
c235cce7 KH |
2742 | XSETFASTINT (data[2 * i], start); |
2743 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2744 | } |
0ed62dc7 | 2745 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2746 | { |
2747 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2748 | Fset_marker (data[2 * i], |
2749 | make_number (start), | |
2750 | last_thing_searched); | |
ca1d1d23 JB |
2751 | data[2 * i + 1] = Fmake_marker (); |
2752 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2753 | make_number (search_regs.end[i]), |
daa37602 | 2754 | last_thing_searched); |
ca1d1d23 | 2755 | } |
daa37602 JB |
2756 | else |
2757 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2758 | abort (); | |
2759 | ||
41c01205 | 2760 | len = 2*(i+1); |
ca1d1d23 JB |
2761 | } |
2762 | else | |
2763 | data[2 * i] = data [2 * i + 1] = Qnil; | |
2764 | } | |
56256c2a | 2765 | |
bd2cbd56 | 2766 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2767 | { |
bd2cbd56 | 2768 | data[len] = last_thing_searched; |
41c01205 DK |
2769 | len++; |
2770 | } | |
2771 | ||
56256c2a RS |
2772 | /* If REUSE is not usable, cons up the values and return them. */ |
2773 | if (! CONSP (reuse)) | |
41c01205 | 2774 | return Flist (len, data); |
56256c2a RS |
2775 | |
2776 | /* If REUSE is a list, store as many value elements as will fit | |
2777 | into the elements of REUSE. */ | |
2778 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2779 | i++, tail = XCDR (tail)) |
56256c2a | 2780 | { |
41c01205 | 2781 | if (i < len) |
f3fbd155 | 2782 | XSETCAR (tail, data[i]); |
56256c2a | 2783 | else |
f3fbd155 | 2784 | XSETCAR (tail, Qnil); |
56256c2a RS |
2785 | prev = tail; |
2786 | } | |
2787 | ||
2788 | /* If we couldn't fit all value elements into REUSE, | |
2789 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2790 | if (i < len) |
2791 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2792 | |
2793 | return reuse; | |
ca1d1d23 JB |
2794 | } |
2795 | ||
2796 | ||
3f1c005b | 2797 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 1, 0, |
8c1a1077 PJ |
2798 | doc: /* Set internal data on last search match from elements of LIST. |
2799 | LIST should have been created by calling `match-data' previously. */) | |
2800 | (list) | |
ca1d1d23 JB |
2801 | register Lisp_Object list; |
2802 | { | |
2803 | register int i; | |
2804 | register Lisp_Object marker; | |
2805 | ||
7074fde6 FP |
2806 | if (running_asynch_code) |
2807 | save_search_regs (); | |
2808 | ||
ca1d1d23 | 2809 | if (!CONSP (list) && !NILP (list)) |
b37902c8 | 2810 | list = wrong_type_argument (Qconsp, list); |
ca1d1d23 | 2811 | |
41c01205 DK |
2812 | /* Unless we find a marker with a buffer or an explicit buffer |
2813 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
2814 | last_thing_searched = Qt; |
2815 | ||
4746118a JB |
2816 | /* Allocate registers if they don't already exist. */ |
2817 | { | |
d084e942 | 2818 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2819 | |
2820 | if (length > search_regs.num_regs) | |
2821 | { | |
1113d9db JB |
2822 | if (search_regs.num_regs == 0) |
2823 | { | |
2824 | search_regs.start | |
2825 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2826 | search_regs.end | |
2827 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2828 | } | |
4746118a | 2829 | else |
1113d9db JB |
2830 | { |
2831 | search_regs.start | |
2832 | = (regoff_t *) xrealloc (search_regs.start, | |
2833 | length * sizeof (regoff_t)); | |
2834 | search_regs.end | |
2835 | = (regoff_t *) xrealloc (search_regs.end, | |
2836 | length * sizeof (regoff_t)); | |
2837 | } | |
4746118a | 2838 | |
e62371e9 KH |
2839 | for (i = search_regs.num_regs; i < length; i++) |
2840 | search_regs.start[i] = -1; | |
2841 | ||
487282dc | 2842 | search_regs.num_regs = length; |
4746118a | 2843 | } |
ca1d1d23 | 2844 | |
c3762cbd | 2845 | for (i = 0;; i++) |
41c01205 DK |
2846 | { |
2847 | marker = Fcar (list); | |
bd2cbd56 | 2848 | if (BUFFERP (marker)) |
c3762cbd | 2849 | { |
bd2cbd56 | 2850 | last_thing_searched = marker; |
c3762cbd DK |
2851 | break; |
2852 | } | |
2853 | if (i >= length) | |
2854 | break; | |
41c01205 DK |
2855 | if (NILP (marker)) |
2856 | { | |
2857 | search_regs.start[i] = -1; | |
2858 | list = Fcdr (list); | |
2859 | } | |
2860 | else | |
2861 | { | |
2862 | int from; | |
e2811828 | 2863 | |
41c01205 DK |
2864 | if (MARKERP (marker)) |
2865 | { | |
2866 | if (XMARKER (marker)->buffer == 0) | |
2867 | XSETFASTINT (marker, 0); | |
2868 | else | |
2869 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
2870 | } | |
e2811828 | 2871 | |
41c01205 DK |
2872 | CHECK_NUMBER_COERCE_MARKER (marker); |
2873 | from = XINT (marker); | |
2874 | list = Fcdr (list); | |
e2811828 | 2875 | |
41c01205 DK |
2876 | marker = Fcar (list); |
2877 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) | |
2878 | XSETFASTINT (marker, 0); | |
e2811828 | 2879 | |
41c01205 DK |
2880 | CHECK_NUMBER_COERCE_MARKER (marker); |
2881 | search_regs.start[i] = from; | |
2882 | search_regs.end[i] = XINT (marker); | |
2883 | } | |
2884 | list = Fcdr (list); | |
2885 | } | |
ca1d1d23 | 2886 | |
41c01205 DK |
2887 | for (; i < search_regs.num_regs; i++) |
2888 | search_regs.start[i] = -1; | |
2889 | } | |
ca1d1d23 | 2890 | |
177c0ea7 | 2891 | return Qnil; |
ca1d1d23 JB |
2892 | } |
2893 | ||
7074fde6 FP |
2894 | /* If non-zero the match data have been saved in saved_search_regs |
2895 | during the execution of a sentinel or filter. */ | |
75ebf74b | 2896 | static int search_regs_saved; |
7074fde6 | 2897 | static struct re_registers saved_search_regs; |
41c01205 | 2898 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
2899 | |
2900 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
2901 | if asynchronous code (filter or sentinel) is running. */ | |
2902 | static void | |
2903 | save_search_regs () | |
2904 | { | |
2905 | if (!search_regs_saved) | |
2906 | { | |
2907 | saved_search_regs.num_regs = search_regs.num_regs; | |
2908 | saved_search_regs.start = search_regs.start; | |
2909 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
2910 | saved_last_thing_searched = last_thing_searched; |
2911 | last_thing_searched = Qnil; | |
7074fde6 | 2912 | search_regs.num_regs = 0; |
2d4a771a RS |
2913 | search_regs.start = 0; |
2914 | search_regs.end = 0; | |
7074fde6 FP |
2915 | |
2916 | search_regs_saved = 1; | |
2917 | } | |
2918 | } | |
2919 | ||
2920 | /* Called upon exit from filters and sentinels. */ | |
2921 | void | |
2922 | restore_match_data () | |
2923 | { | |
2924 | if (search_regs_saved) | |
2925 | { | |
2926 | if (search_regs.num_regs > 0) | |
2927 | { | |
2928 | xfree (search_regs.start); | |
2929 | xfree (search_regs.end); | |
2930 | } | |
2931 | search_regs.num_regs = saved_search_regs.num_regs; | |
2932 | search_regs.start = saved_search_regs.start; | |
2933 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
2934 | last_thing_searched = saved_last_thing_searched; |
2935 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
2936 | search_regs_saved = 0; |
2937 | } | |
2938 | } | |
2939 | ||
ca1d1d23 JB |
2940 | /* Quote a string to inactivate reg-expr chars */ |
2941 | ||
2942 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 PJ |
2943 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
2944 | (string) | |
5806161b | 2945 | Lisp_Object string; |
ca1d1d23 JB |
2946 | { |
2947 | register unsigned char *in, *out, *end; | |
2948 | register unsigned char *temp; | |
0c8533c6 | 2949 | int backslashes_added = 0; |
ca1d1d23 | 2950 | |
b7826503 | 2951 | CHECK_STRING (string); |
ca1d1d23 | 2952 | |
d5db4077 | 2953 | temp = (unsigned char *) alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
2954 | |
2955 | /* Now copy the data into the new string, inserting escapes. */ | |
2956 | ||
d5db4077 KR |
2957 | in = SDATA (string); |
2958 | end = in + SBYTES (string); | |
177c0ea7 | 2959 | out = temp; |
ca1d1d23 JB |
2960 | |
2961 | for (; in != end; in++) | |
2962 | { | |
2963 | if (*in == '[' || *in == ']' | |
2964 | || *in == '*' || *in == '.' || *in == '\\' | |
2965 | || *in == '?' || *in == '+' | |
2966 | || *in == '^' || *in == '$') | |
0c8533c6 | 2967 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
2968 | *out++ = *in; |
2969 | } | |
2970 | ||
3f8100f1 | 2971 | return make_specified_string (temp, |
d5db4077 | 2972 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
2973 | out - temp, |
2974 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 2975 | } |
177c0ea7 | 2976 | \f |
dfcf069d | 2977 | void |
ca1d1d23 JB |
2978 | syms_of_search () |
2979 | { | |
2980 | register int i; | |
2981 | ||
487282dc KH |
2982 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
2983 | { | |
2984 | searchbufs[i].buf.allocated = 100; | |
b23c0a83 | 2985 | searchbufs[i].buf.buffer = (unsigned char *) xmalloc (100); |
487282dc KH |
2986 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
2987 | searchbufs[i].regexp = Qnil; | |
ecdb561e | 2988 | searchbufs[i].whitespace_regexp = Qnil; |
487282dc KH |
2989 | staticpro (&searchbufs[i].regexp); |
2990 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); | |
2991 | } | |
2992 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
2993 | |
2994 | Qsearch_failed = intern ("search-failed"); | |
2995 | staticpro (&Qsearch_failed); | |
2996 | Qinvalid_regexp = intern ("invalid-regexp"); | |
2997 | staticpro (&Qinvalid_regexp); | |
2998 | ||
2999 | Fput (Qsearch_failed, Qerror_conditions, | |
3000 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
3001 | Fput (Qsearch_failed, Qerror_message, | |
3002 | build_string ("Search failed")); | |
3003 | ||
3004 | Fput (Qinvalid_regexp, Qerror_conditions, | |
3005 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
3006 | Fput (Qinvalid_regexp, Qerror_message, | |
3007 | build_string ("Invalid regexp")); | |
3008 | ||
daa37602 JB |
3009 | last_thing_searched = Qnil; |
3010 | staticpro (&last_thing_searched); | |
3011 | ||
0f6af254 DK |
3012 | saved_last_thing_searched = Qnil; |
3013 | staticpro (&saved_last_thing_searched); | |
3014 | ||
41a33295 | 3015 | DEFVAR_LISP ("search-spaces-regexp", &Vsearch_spaces_regexp, |
e2811828 | 3016 | doc: /* Regexp to substitute for bunches of spaces in regexp search. |
f31a9a68 RS |
3017 | Some commands use this for user-specified regexps. |
3018 | Spaces that occur inside character classes or repetition operators | |
3019 | or other such regexp constructs are not replaced with this. | |
3020 | A value of nil (which is the normal value) means treat spaces literally. */); | |
41a33295 | 3021 | Vsearch_spaces_regexp = Qnil; |
f31a9a68 | 3022 | |
ca1d1d23 | 3023 | defsubr (&Slooking_at); |
b819a390 RS |
3024 | defsubr (&Sposix_looking_at); |
3025 | defsubr (&Sstring_match); | |
3026 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
3027 | defsubr (&Ssearch_forward); |
3028 | defsubr (&Ssearch_backward); | |
3029 | defsubr (&Sword_search_forward); | |
3030 | defsubr (&Sword_search_backward); | |
3031 | defsubr (&Sre_search_forward); | |
3032 | defsubr (&Sre_search_backward); | |
b819a390 RS |
3033 | defsubr (&Sposix_search_forward); |
3034 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
3035 | defsubr (&Sreplace_match); |
3036 | defsubr (&Smatch_beginning); | |
3037 | defsubr (&Smatch_end); | |
3038 | defsubr (&Smatch_data); | |
3f1c005b | 3039 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
3040 | defsubr (&Sregexp_quote); |
3041 | } | |
ab5796a9 MB |
3042 | |
3043 | /* arch-tag: a6059d79-0552-4f14-a2cb-d379a4e3c78f | |
3044 | (do not change this comment) */ |