Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
3a22ee35 | 2 | Copyright (C) 1985, 1986, 1987, 1993, 1994 Free Software Foundation, Inc. |
ca1d1d23 JB |
3 | |
4 | This file is part of GNU Emacs. | |
5 | ||
6 | GNU Emacs is free software; you can redistribute it and/or modify | |
7 | it under the terms of the GNU General Public License as published by | |
7c938215 | 8 | the Free Software Foundation; either version 2, or (at your option) |
ca1d1d23 JB |
9 | any later version. |
10 | ||
11 | GNU Emacs is distributed in the hope that it will be useful, | |
12 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
14 | GNU General Public License for more details. | |
15 | ||
16 | You should have received a copy of the GNU General Public License | |
17 | along with GNU Emacs; see the file COPYING. If not, write to | |
3b7ad313 EN |
18 | the Free Software Foundation, Inc., 59 Temple Place - Suite 330, |
19 | Boston, MA 02111-1307, USA. */ | |
ca1d1d23 JB |
20 | |
21 | ||
18160b98 | 22 | #include <config.h> |
ca1d1d23 JB |
23 | #include "lisp.h" |
24 | #include "syntax.h" | |
5679531d | 25 | #include "category.h" |
ca1d1d23 | 26 | #include "buffer.h" |
5679531d | 27 | #include "charset.h" |
9169c321 | 28 | #include "region-cache.h" |
ca1d1d23 | 29 | #include "commands.h" |
9ac0d9e0 | 30 | #include "blockinput.h" |
4746118a | 31 | |
ca1d1d23 JB |
32 | #include <sys/types.h> |
33 | #include "regex.h" | |
34 | ||
1d288aef | 35 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 36 | |
487282dc KH |
37 | /* If the regexp is non-nil, then the buffer contains the compiled form |
38 | of that regexp, suitable for searching. */ | |
1d288aef RS |
39 | struct regexp_cache |
40 | { | |
487282dc KH |
41 | struct regexp_cache *next; |
42 | Lisp_Object regexp; | |
43 | struct re_pattern_buffer buf; | |
44 | char fastmap[0400]; | |
b819a390 RS |
45 | /* Nonzero means regexp was compiled to do full POSIX backtracking. */ |
46 | char posix; | |
487282dc | 47 | }; |
ca1d1d23 | 48 | |
487282dc KH |
49 | /* The instances of that struct. */ |
50 | struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; | |
ca1d1d23 | 51 | |
487282dc KH |
52 | /* The head of the linked list; points to the most recently used buffer. */ |
53 | struct regexp_cache *searchbuf_head; | |
ca1d1d23 | 54 | |
ca1d1d23 | 55 | |
4746118a JB |
56 | /* Every call to re_match, etc., must pass &search_regs as the regs |
57 | argument unless you can show it is unnecessary (i.e., if re_match | |
58 | is certainly going to be called again before region-around-match | |
59 | can be called). | |
60 | ||
61 | Since the registers are now dynamically allocated, we need to make | |
62 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
63 | been allocated by checking search_regs.num_regs. |
64 | ||
65 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
66 | buffer using bits in the re_pattern_buffer. This means that whenever |
67 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
68 | allocated any registers, and will allocate new registers the next |
69 | time you call a searching or matching function. Therefore, we need | |
70 | to call re_set_registers after compiling a new pattern or after | |
71 | setting the match registers, so that the regex functions will be | |
72 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
73 | static struct re_registers search_regs; |
74 | ||
daa37602 JB |
75 | /* The buffer in which the last search was performed, or |
76 | Qt if the last search was done in a string; | |
77 | Qnil if no searching has been done yet. */ | |
78 | static Lisp_Object last_thing_searched; | |
ca1d1d23 | 79 | |
8e6208c5 | 80 | /* error condition signaled when regexp compile_pattern fails */ |
ca1d1d23 JB |
81 | |
82 | Lisp_Object Qinvalid_regexp; | |
83 | ||
ca325161 | 84 | static void set_search_regs (); |
044f81f1 | 85 | static void save_search_regs (); |
ca325161 | 86 | |
b819a390 RS |
87 | static int search_buffer (); |
88 | ||
ca1d1d23 JB |
89 | static void |
90 | matcher_overflow () | |
91 | { | |
92 | error ("Stack overflow in regexp matcher"); | |
93 | } | |
94 | ||
95 | #ifdef __STDC__ | |
96 | #define CONST const | |
97 | #else | |
98 | #define CONST | |
99 | #endif | |
100 | ||
b819a390 RS |
101 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
102 | PATTERN is the pattern to compile. | |
103 | CP is the place to put the result. | |
104 | TRANSLATE is a translation table for ignoring case, or NULL for none. | |
105 | REGP is the structure that says where to store the "register" | |
106 | values that will result from matching this pattern. | |
107 | If it is 0, we should compile the pattern not to record any | |
108 | subexpression bounds. | |
109 | POSIX is nonzero if we want full backtracking (POSIX style) | |
5679531d KH |
110 | for this pattern. 0 means backtrack only enough to get a valid match. |
111 | MULTIBYTE is nonzero if we want to handle multibyte characters in | |
112 | PATTERN. 0 means all multibyte characters are recognized just as | |
113 | sequences of binary data. */ | |
ca1d1d23 | 114 | |
487282dc | 115 | static void |
5679531d | 116 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte) |
487282dc | 117 | struct regexp_cache *cp; |
ca1d1d23 | 118 | Lisp_Object pattern; |
b1428bd8 | 119 | Lisp_Object *translate; |
487282dc | 120 | struct re_registers *regp; |
b819a390 | 121 | int posix; |
5679531d | 122 | int multibyte; |
ca1d1d23 JB |
123 | { |
124 | CONST char *val; | |
b819a390 | 125 | reg_syntax_t old; |
ca1d1d23 | 126 | |
487282dc KH |
127 | cp->regexp = Qnil; |
128 | cp->buf.translate = translate; | |
b819a390 | 129 | cp->posix = posix; |
5679531d | 130 | cp->buf.multibyte = multibyte; |
9ac0d9e0 | 131 | BLOCK_INPUT; |
b819a390 RS |
132 | old = re_set_syntax (RE_SYNTAX_EMACS |
133 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); | |
b90d9e80 | 134 | val = (CONST char *) re_compile_pattern ((char *) XSTRING (pattern)->data, |
487282dc | 135 | XSTRING (pattern)->size, &cp->buf); |
b819a390 | 136 | re_set_syntax (old); |
9ac0d9e0 | 137 | UNBLOCK_INPUT; |
ca1d1d23 | 138 | if (val) |
487282dc | 139 | Fsignal (Qinvalid_regexp, Fcons (build_string (val), Qnil)); |
1113d9db | 140 | |
487282dc | 141 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
142 | } |
143 | ||
144 | /* Compile a regexp if necessary, but first check to see if there's one in | |
b819a390 RS |
145 | the cache. |
146 | PATTERN is the pattern to compile. | |
147 | TRANSLATE is a translation table for ignoring case, or NULL for none. | |
148 | REGP is the structure that says where to store the "register" | |
149 | values that will result from matching this pattern. | |
150 | If it is 0, we should compile the pattern not to record any | |
151 | subexpression bounds. | |
152 | POSIX is nonzero if we want full backtracking (POSIX style) | |
153 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
154 | |
155 | struct re_pattern_buffer * | |
b819a390 | 156 | compile_pattern (pattern, regp, translate, posix) |
487282dc KH |
157 | Lisp_Object pattern; |
158 | struct re_registers *regp; | |
b1428bd8 | 159 | Lisp_Object *translate; |
b819a390 | 160 | int posix; |
487282dc KH |
161 | { |
162 | struct regexp_cache *cp, **cpp; | |
5679531d KH |
163 | /* Should we check it here, or add an argument `multibyte' to this |
164 | function? */ | |
165 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); | |
487282dc KH |
166 | |
167 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
168 | { | |
169 | cp = *cpp; | |
1d288aef RS |
170 | if (XSTRING (cp->regexp)->size == XSTRING (pattern)->size |
171 | && !NILP (Fstring_equal (cp->regexp, pattern)) | |
b819a390 | 172 | && cp->buf.translate == translate |
5679531d KH |
173 | && cp->posix == posix |
174 | && cp->buf.multibyte == multibyte) | |
487282dc KH |
175 | break; |
176 | ||
177 | /* If we're at the end of the cache, compile into the last cell. */ | |
178 | if (cp->next == 0) | |
179 | { | |
5679531d | 180 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte); |
487282dc KH |
181 | break; |
182 | } | |
183 | } | |
184 | ||
185 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
186 | either because we found it in the cache or because we just compiled it. | |
187 | Move it to the front of the queue to mark it as most recently used. */ | |
188 | *cpp = cp->next; | |
189 | cp->next = searchbuf_head; | |
190 | searchbuf_head = cp; | |
1113d9db | 191 | |
6639708c RS |
192 | /* Advise the searching functions about the space we have allocated |
193 | for register data. */ | |
194 | if (regp) | |
195 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
196 | ||
487282dc | 197 | return &cp->buf; |
ca1d1d23 JB |
198 | } |
199 | ||
200 | /* Error condition used for failing searches */ | |
201 | Lisp_Object Qsearch_failed; | |
202 | ||
203 | Lisp_Object | |
204 | signal_failure (arg) | |
205 | Lisp_Object arg; | |
206 | { | |
207 | Fsignal (Qsearch_failed, Fcons (arg, Qnil)); | |
208 | return Qnil; | |
209 | } | |
210 | \f | |
b819a390 RS |
211 | static Lisp_Object |
212 | looking_at_1 (string, posix) | |
ca1d1d23 | 213 | Lisp_Object string; |
b819a390 | 214 | int posix; |
ca1d1d23 JB |
215 | { |
216 | Lisp_Object val; | |
217 | unsigned char *p1, *p2; | |
218 | int s1, s2; | |
219 | register int i; | |
487282dc | 220 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 221 | |
7074fde6 FP |
222 | if (running_asynch_code) |
223 | save_search_regs (); | |
224 | ||
ca1d1d23 | 225 | CHECK_STRING (string, 0); |
487282dc KH |
226 | bufp = compile_pattern (string, &search_regs, |
227 | (!NILP (current_buffer->case_fold_search) | |
b819a390 RS |
228 | ? DOWNCASE_TABLE : 0), |
229 | posix); | |
ca1d1d23 JB |
230 | |
231 | immediate_quit = 1; | |
232 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
233 | ||
234 | /* Get pointers and sizes of the two strings | |
235 | that make up the visible portion of the buffer. */ | |
236 | ||
237 | p1 = BEGV_ADDR; | |
238 | s1 = GPT - BEGV; | |
239 | p2 = GAP_END_ADDR; | |
240 | s2 = ZV - GPT; | |
241 | if (s1 < 0) | |
242 | { | |
243 | p2 = p1; | |
244 | s2 = ZV - BEGV; | |
245 | s1 = 0; | |
246 | } | |
247 | if (s2 < 0) | |
248 | { | |
249 | s1 = ZV - BEGV; | |
250 | s2 = 0; | |
251 | } | |
252 | ||
487282dc | 253 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
6ec8bbd2 | 254 | PT - BEGV, &search_regs, |
ca1d1d23 JB |
255 | ZV - BEGV); |
256 | if (i == -2) | |
257 | matcher_overflow (); | |
258 | ||
259 | val = (0 <= i ? Qt : Qnil); | |
4746118a | 260 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
261 | if (search_regs.start[i] >= 0) |
262 | { | |
263 | search_regs.start[i] += BEGV; | |
264 | search_regs.end[i] += BEGV; | |
265 | } | |
a3668d92 | 266 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
267 | immediate_quit = 0; |
268 | return val; | |
269 | } | |
270 | ||
b819a390 | 271 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
94f94972 | 272 | "Return t if text after point matches regular expression REGEXP.\n\ |
b819a390 RS |
273 | This function modifies the match data that `match-beginning',\n\ |
274 | `match-end' and `match-data' access; save and restore the match\n\ | |
275 | data if you want to preserve them.") | |
94f94972 RS |
276 | (regexp) |
277 | Lisp_Object regexp; | |
b819a390 | 278 | { |
94f94972 | 279 | return looking_at_1 (regexp, 0); |
b819a390 RS |
280 | } |
281 | ||
282 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
94f94972 | 283 | "Return t if text after point matches regular expression REGEXP.\n\ |
b819a390 RS |
284 | Find the longest match, in accord with Posix regular expression rules.\n\ |
285 | This function modifies the match data that `match-beginning',\n\ | |
286 | `match-end' and `match-data' access; save and restore the match\n\ | |
287 | data if you want to preserve them.") | |
94f94972 RS |
288 | (regexp) |
289 | Lisp_Object regexp; | |
b819a390 | 290 | { |
94f94972 | 291 | return looking_at_1 (regexp, 1); |
b819a390 RS |
292 | } |
293 | \f | |
294 | static Lisp_Object | |
295 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 296 | Lisp_Object regexp, string, start; |
b819a390 | 297 | int posix; |
ca1d1d23 JB |
298 | { |
299 | int val; | |
300 | int s; | |
487282dc | 301 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 302 | |
7074fde6 FP |
303 | if (running_asynch_code) |
304 | save_search_regs (); | |
305 | ||
ca1d1d23 JB |
306 | CHECK_STRING (regexp, 0); |
307 | CHECK_STRING (string, 1); | |
308 | ||
309 | if (NILP (start)) | |
310 | s = 0; | |
311 | else | |
312 | { | |
313 | int len = XSTRING (string)->size; | |
314 | ||
315 | CHECK_NUMBER (start, 2); | |
316 | s = XINT (start); | |
317 | if (s < 0 && -s <= len) | |
26faf9f4 | 318 | s = len + s; |
ca1d1d23 JB |
319 | else if (0 > s || s > len) |
320 | args_out_of_range (string, start); | |
321 | } | |
322 | ||
487282dc KH |
323 | bufp = compile_pattern (regexp, &search_regs, |
324 | (!NILP (current_buffer->case_fold_search) | |
b819a390 | 325 | ? DOWNCASE_TABLE : 0), |
24b704fa | 326 | posix); |
ca1d1d23 | 327 | immediate_quit = 1; |
487282dc | 328 | val = re_search (bufp, (char *) XSTRING (string)->data, |
ca1d1d23 JB |
329 | XSTRING (string)->size, s, XSTRING (string)->size - s, |
330 | &search_regs); | |
331 | immediate_quit = 0; | |
daa37602 | 332 | last_thing_searched = Qt; |
ca1d1d23 JB |
333 | if (val == -2) |
334 | matcher_overflow (); | |
335 | if (val < 0) return Qnil; | |
336 | return make_number (val); | |
337 | } | |
e59a8453 | 338 | |
b819a390 RS |
339 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
340 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
341 | If third arg START is non-nil, start search at that index in STRING.\n\ | |
342 | For index of first char beyond the match, do (match-end 0).\n\ | |
343 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
344 | matched by parenthesis constructs in the pattern.") | |
345 | (regexp, string, start) | |
346 | Lisp_Object regexp, string, start; | |
347 | { | |
348 | return string_match_1 (regexp, string, start, 0); | |
349 | } | |
350 | ||
351 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
352 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
353 | Find the longest match, in accord with Posix regular expression rules.\n\ | |
354 | If third arg START is non-nil, start search at that index in STRING.\n\ | |
355 | For index of first char beyond the match, do (match-end 0).\n\ | |
356 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
357 | matched by parenthesis constructs in the pattern.") | |
358 | (regexp, string, start) | |
359 | Lisp_Object regexp, string, start; | |
360 | { | |
361 | return string_match_1 (regexp, string, start, 1); | |
362 | } | |
363 | ||
e59a8453 RS |
364 | /* Match REGEXP against STRING, searching all of STRING, |
365 | and return the index of the match, or negative on failure. | |
366 | This does not clobber the match data. */ | |
367 | ||
368 | int | |
369 | fast_string_match (regexp, string) | |
370 | Lisp_Object regexp, string; | |
371 | { | |
372 | int val; | |
487282dc | 373 | struct re_pattern_buffer *bufp; |
e59a8453 | 374 | |
b819a390 | 375 | bufp = compile_pattern (regexp, 0, 0, 0); |
e59a8453 | 376 | immediate_quit = 1; |
487282dc | 377 | val = re_search (bufp, (char *) XSTRING (string)->data, |
e59a8453 RS |
378 | XSTRING (string)->size, 0, XSTRING (string)->size, |
379 | 0); | |
380 | immediate_quit = 0; | |
381 | return val; | |
382 | } | |
5679531d KH |
383 | |
384 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
385 | and return the index of the match, or negative on failure. | |
386 | This does not clobber the match data. */ | |
387 | ||
388 | extern Lisp_Object Vascii_downcase_table; | |
389 | ||
390 | int | |
391 | fast_string_match_ignore_case (regexp, string) | |
392 | Lisp_Object regexp; | |
393 | char *string; | |
394 | { | |
395 | int val; | |
396 | struct re_pattern_buffer *bufp; | |
397 | int len = strlen (string); | |
398 | ||
399 | bufp = compile_pattern (regexp, 0, | |
400 | XCHAR_TABLE (Vascii_downcase_table)->contents, 0); | |
401 | immediate_quit = 1; | |
402 | val = re_search (bufp, string, len, 0, len, 0); | |
403 | immediate_quit = 0; | |
404 | return val; | |
405 | } | |
ca1d1d23 | 406 | \f |
9169c321 JB |
407 | /* max and min. */ |
408 | ||
409 | static int | |
410 | max (a, b) | |
411 | int a, b; | |
412 | { | |
413 | return ((a > b) ? a : b); | |
414 | } | |
415 | ||
416 | static int | |
417 | min (a, b) | |
418 | int a, b; | |
419 | { | |
420 | return ((a < b) ? a : b); | |
421 | } | |
422 | ||
423 | \f | |
424 | /* The newline cache: remembering which sections of text have no newlines. */ | |
425 | ||
426 | /* If the user has requested newline caching, make sure it's on. | |
427 | Otherwise, make sure it's off. | |
428 | This is our cheezy way of associating an action with the change of | |
429 | state of a buffer-local variable. */ | |
430 | static void | |
431 | newline_cache_on_off (buf) | |
432 | struct buffer *buf; | |
433 | { | |
434 | if (NILP (buf->cache_long_line_scans)) | |
435 | { | |
436 | /* It should be off. */ | |
437 | if (buf->newline_cache) | |
438 | { | |
439 | free_region_cache (buf->newline_cache); | |
440 | buf->newline_cache = 0; | |
441 | } | |
442 | } | |
443 | else | |
444 | { | |
445 | /* It should be on. */ | |
446 | if (buf->newline_cache == 0) | |
447 | buf->newline_cache = new_region_cache (); | |
448 | } | |
449 | } | |
450 | ||
451 | \f | |
452 | /* Search for COUNT instances of the character TARGET between START and END. | |
453 | ||
454 | If COUNT is positive, search forwards; END must be >= START. | |
455 | If COUNT is negative, search backwards for the -COUNTth instance; | |
456 | END must be <= START. | |
457 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
458 | ||
459 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
460 | direction indicated by COUNT. | |
ffd56f97 JB |
461 | |
462 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
5bfe95c9 RS |
463 | position after the COUNTth match. Note that for reverse motion |
464 | this is not the same as the usual convention for Emacs motion commands. | |
ffd56f97 | 465 | |
9169c321 JB |
466 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
467 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 468 | |
087a5f81 RS |
469 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
470 | except when inside redisplay. */ | |
471 | ||
9169c321 JB |
472 | scan_buffer (target, start, end, count, shortage, allow_quit) |
473 | register int target; | |
474 | int start, end; | |
475 | int count; | |
476 | int *shortage; | |
087a5f81 | 477 | int allow_quit; |
ca1d1d23 | 478 | { |
9169c321 JB |
479 | struct region_cache *newline_cache; |
480 | int direction; | |
ffd56f97 | 481 | |
9169c321 JB |
482 | if (count > 0) |
483 | { | |
484 | direction = 1; | |
485 | if (! end) end = ZV; | |
486 | } | |
487 | else | |
488 | { | |
489 | direction = -1; | |
490 | if (! end) end = BEGV; | |
491 | } | |
ffd56f97 | 492 | |
9169c321 JB |
493 | newline_cache_on_off (current_buffer); |
494 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
495 | |
496 | if (shortage != 0) | |
497 | *shortage = 0; | |
498 | ||
087a5f81 | 499 | immediate_quit = allow_quit; |
ca1d1d23 | 500 | |
ffd56f97 | 501 | if (count > 0) |
9169c321 | 502 | while (start != end) |
ca1d1d23 | 503 | { |
9169c321 JB |
504 | /* Our innermost scanning loop is very simple; it doesn't know |
505 | about gaps, buffer ends, or the newline cache. ceiling is | |
506 | the position of the last character before the next such | |
507 | obstacle --- the last character the dumb search loop should | |
508 | examine. */ | |
509 | register int ceiling = end - 1; | |
510 | ||
511 | /* If we're looking for a newline, consult the newline cache | |
512 | to see where we can avoid some scanning. */ | |
513 | if (target == '\n' && newline_cache) | |
514 | { | |
515 | int next_change; | |
516 | immediate_quit = 0; | |
517 | while (region_cache_forward | |
518 | (current_buffer, newline_cache, start, &next_change)) | |
519 | start = next_change; | |
cbe0db0d | 520 | immediate_quit = allow_quit; |
9169c321 JB |
521 | |
522 | /* start should never be after end. */ | |
523 | if (start >= end) | |
524 | start = end - 1; | |
525 | ||
526 | /* Now the text after start is an unknown region, and | |
527 | next_change is the position of the next known region. */ | |
528 | ceiling = min (next_change - 1, ceiling); | |
529 | } | |
530 | ||
531 | /* The dumb loop can only scan text stored in contiguous | |
532 | bytes. BUFFER_CEILING_OF returns the last character | |
533 | position that is contiguous, so the ceiling is the | |
534 | position after that. */ | |
535 | ceiling = min (BUFFER_CEILING_OF (start), ceiling); | |
536 | ||
537 | { | |
538 | /* The termination address of the dumb loop. */ | |
5679531d KH |
539 | register unsigned char *ceiling_addr = POS_ADDR (ceiling) + 1; |
540 | register unsigned char *cursor = POS_ADDR (start); | |
9169c321 JB |
541 | unsigned char *base = cursor; |
542 | ||
543 | while (cursor < ceiling_addr) | |
544 | { | |
545 | unsigned char *scan_start = cursor; | |
546 | ||
547 | /* The dumb loop. */ | |
548 | while (*cursor != target && ++cursor < ceiling_addr) | |
549 | ; | |
550 | ||
551 | /* If we're looking for newlines, cache the fact that | |
552 | the region from start to cursor is free of them. */ | |
553 | if (target == '\n' && newline_cache) | |
554 | know_region_cache (current_buffer, newline_cache, | |
555 | start + scan_start - base, | |
556 | start + cursor - base); | |
557 | ||
558 | /* Did we find the target character? */ | |
559 | if (cursor < ceiling_addr) | |
560 | { | |
561 | if (--count == 0) | |
562 | { | |
563 | immediate_quit = 0; | |
564 | return (start + cursor - base + 1); | |
565 | } | |
566 | cursor++; | |
567 | } | |
568 | } | |
569 | ||
570 | start += cursor - base; | |
571 | } | |
ca1d1d23 JB |
572 | } |
573 | else | |
9169c321 JB |
574 | while (start > end) |
575 | { | |
576 | /* The last character to check before the next obstacle. */ | |
577 | register int ceiling = end; | |
578 | ||
579 | /* Consult the newline cache, if appropriate. */ | |
580 | if (target == '\n' && newline_cache) | |
581 | { | |
582 | int next_change; | |
583 | immediate_quit = 0; | |
584 | while (region_cache_backward | |
585 | (current_buffer, newline_cache, start, &next_change)) | |
586 | start = next_change; | |
cbe0db0d | 587 | immediate_quit = allow_quit; |
9169c321 JB |
588 | |
589 | /* Start should never be at or before end. */ | |
590 | if (start <= end) | |
591 | start = end + 1; | |
592 | ||
593 | /* Now the text before start is an unknown region, and | |
594 | next_change is the position of the next known region. */ | |
595 | ceiling = max (next_change, ceiling); | |
596 | } | |
597 | ||
598 | /* Stop scanning before the gap. */ | |
599 | ceiling = max (BUFFER_FLOOR_OF (start - 1), ceiling); | |
600 | ||
601 | { | |
602 | /* The termination address of the dumb loop. */ | |
5679531d KH |
603 | register unsigned char *ceiling_addr = POS_ADDR (ceiling); |
604 | register unsigned char *cursor = POS_ADDR (start - 1); | |
9169c321 JB |
605 | unsigned char *base = cursor; |
606 | ||
607 | while (cursor >= ceiling_addr) | |
608 | { | |
609 | unsigned char *scan_start = cursor; | |
610 | ||
611 | while (*cursor != target && --cursor >= ceiling_addr) | |
612 | ; | |
613 | ||
614 | /* If we're looking for newlines, cache the fact that | |
615 | the region from after the cursor to start is free of them. */ | |
616 | if (target == '\n' && newline_cache) | |
617 | know_region_cache (current_buffer, newline_cache, | |
618 | start + cursor - base, | |
619 | start + scan_start - base); | |
620 | ||
621 | /* Did we find the target character? */ | |
622 | if (cursor >= ceiling_addr) | |
623 | { | |
624 | if (++count >= 0) | |
625 | { | |
626 | immediate_quit = 0; | |
627 | return (start + cursor - base); | |
628 | } | |
629 | cursor--; | |
630 | } | |
631 | } | |
632 | ||
633 | start += cursor - base; | |
634 | } | |
635 | } | |
636 | ||
ca1d1d23 JB |
637 | immediate_quit = 0; |
638 | if (shortage != 0) | |
ffd56f97 | 639 | *shortage = count * direction; |
9169c321 | 640 | return start; |
ca1d1d23 JB |
641 | } |
642 | ||
63fa018d RS |
643 | int |
644 | find_next_newline_no_quit (from, cnt) | |
645 | register int from, cnt; | |
646 | { | |
9169c321 | 647 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
63fa018d RS |
648 | } |
649 | ||
ca1d1d23 JB |
650 | int |
651 | find_next_newline (from, cnt) | |
652 | register int from, cnt; | |
653 | { | |
9169c321 JB |
654 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 1); |
655 | } | |
656 | ||
657 | ||
658 | /* Like find_next_newline, but returns position before the newline, | |
659 | not after, and only search up to TO. This isn't just | |
660 | find_next_newline (...)-1, because you might hit TO. */ | |
661 | int | |
662 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 663 | int from, to, cnt; |
9169c321 JB |
664 | { |
665 | int shortage; | |
666 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
667 | ||
668 | if (shortage == 0) | |
669 | pos--; | |
670 | ||
671 | return pos; | |
ca1d1d23 JB |
672 | } |
673 | \f | |
c1dc99a1 JB |
674 | Lisp_Object skip_chars (); |
675 | ||
ca1d1d23 | 676 | DEFUN ("skip-chars-forward", Fskip_chars_forward, Sskip_chars_forward, 1, 2, 0, |
3acb9a69 RS |
677 | "Move point forward, stopping before a char not in STRING, or at pos LIM.\n\ |
678 | STRING is like the inside of a `[...]' in a regular expression\n\ | |
ca1d1d23 JB |
679 | except that `]' is never special and `\\' quotes `^', `-' or `\\'.\n\ |
680 | Thus, with arg \"a-zA-Z\", this skips letters stopping before first nonletter.\n\ | |
c1dc99a1 JB |
681 | With arg \"^a-zA-Z\", skips nonletters stopping before first letter.\n\ |
682 | Returns the distance traveled, either zero or positive.") | |
ca1d1d23 JB |
683 | (string, lim) |
684 | Lisp_Object string, lim; | |
685 | { | |
17431c60 | 686 | return skip_chars (1, 0, string, lim); |
ca1d1d23 JB |
687 | } |
688 | ||
689 | DEFUN ("skip-chars-backward", Fskip_chars_backward, Sskip_chars_backward, 1, 2, 0, | |
3acb9a69 | 690 | "Move point backward, stopping after a char not in STRING, or at pos LIM.\n\ |
c1dc99a1 JB |
691 | See `skip-chars-forward' for details.\n\ |
692 | Returns the distance traveled, either zero or negative.") | |
ca1d1d23 JB |
693 | (string, lim) |
694 | Lisp_Object string, lim; | |
695 | { | |
17431c60 RS |
696 | return skip_chars (0, 0, string, lim); |
697 | } | |
698 | ||
699 | DEFUN ("skip-syntax-forward", Fskip_syntax_forward, Sskip_syntax_forward, 1, 2, 0, | |
700 | "Move point forward across chars in specified syntax classes.\n\ | |
701 | SYNTAX is a string of syntax code characters.\n\ | |
702 | Stop before a char whose syntax is not in SYNTAX, or at position LIM.\n\ | |
703 | If SYNTAX starts with ^, skip characters whose syntax is NOT in SYNTAX.\n\ | |
704 | This function returns the distance traveled, either zero or positive.") | |
705 | (syntax, lim) | |
706 | Lisp_Object syntax, lim; | |
707 | { | |
708 | return skip_chars (1, 1, syntax, lim); | |
709 | } | |
710 | ||
711 | DEFUN ("skip-syntax-backward", Fskip_syntax_backward, Sskip_syntax_backward, 1, 2, 0, | |
712 | "Move point backward across chars in specified syntax classes.\n\ | |
713 | SYNTAX is a string of syntax code characters.\n\ | |
714 | Stop on reaching a char whose syntax is not in SYNTAX, or at position LIM.\n\ | |
715 | If SYNTAX starts with ^, skip characters whose syntax is NOT in SYNTAX.\n\ | |
716 | This function returns the distance traveled, either zero or negative.") | |
717 | (syntax, lim) | |
718 | Lisp_Object syntax, lim; | |
719 | { | |
720 | return skip_chars (0, 1, syntax, lim); | |
ca1d1d23 JB |
721 | } |
722 | ||
c1dc99a1 | 723 | Lisp_Object |
17431c60 RS |
724 | skip_chars (forwardp, syntaxp, string, lim) |
725 | int forwardp, syntaxp; | |
ca1d1d23 JB |
726 | Lisp_Object string, lim; |
727 | { | |
728 | register unsigned char *p, *pend; | |
729 | register unsigned char c; | |
5679531d | 730 | register int ch; |
ca1d1d23 | 731 | unsigned char fastmap[0400]; |
5679531d KH |
732 | /* If SYNTAXP is 0, STRING may contain multi-byte form of characters |
733 | of which codes don't fit in FASTMAP. In that case, we set the | |
734 | first byte of multibyte form (i.e. base leading-code) in FASTMAP | |
735 | and set the actual ranges of characters in CHAR_RANGES. In the | |
736 | form "X-Y" of STRING, both X and Y must belong to the same | |
737 | character set because a range striding across character sets is | |
738 | meaningless. */ | |
739 | int *char_ranges | |
740 | = (int *) alloca (XSTRING (string)->size * (sizeof (int)) * 2); | |
741 | int n_char_ranges = 0; | |
ca1d1d23 JB |
742 | int negate = 0; |
743 | register int i; | |
1312cff5 | 744 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); |
ca1d1d23 JB |
745 | |
746 | CHECK_STRING (string, 0); | |
747 | ||
748 | if (NILP (lim)) | |
a3668d92 | 749 | XSETINT (lim, forwardp ? ZV : BEGV); |
ca1d1d23 JB |
750 | else |
751 | CHECK_NUMBER_COERCE_MARKER (lim, 1); | |
752 | ||
ca1d1d23 | 753 | /* In any case, don't allow scan outside bounds of buffer. */ |
c5241910 RS |
754 | /* jla turned this off, for no known reason. |
755 | bfox turned the ZV part on, and rms turned the | |
756 | BEGV part back on. */ | |
757 | if (XINT (lim) > ZV) | |
c235cce7 | 758 | XSETFASTINT (lim, ZV); |
c5241910 | 759 | if (XINT (lim) < BEGV) |
c235cce7 | 760 | XSETFASTINT (lim, BEGV); |
ca1d1d23 JB |
761 | |
762 | p = XSTRING (string)->data; | |
763 | pend = p + XSTRING (string)->size; | |
764 | bzero (fastmap, sizeof fastmap); | |
765 | ||
766 | if (p != pend && *p == '^') | |
767 | { | |
768 | negate = 1; p++; | |
769 | } | |
770 | ||
17431c60 RS |
771 | /* Find the characters specified and set their elements of fastmap. |
772 | If syntaxp, each character counts as itself. | |
5679531d | 773 | Otherwise, handle backslashes and ranges specially. */ |
ca1d1d23 JB |
774 | |
775 | while (p != pend) | |
776 | { | |
5679531d | 777 | c = *p; |
1312cff5 KH |
778 | if (multibyte) |
779 | { | |
780 | ch = STRING_CHAR (p, pend - p); | |
781 | p += BYTES_BY_CHAR_HEAD (*p); | |
782 | } | |
783 | else | |
784 | { | |
785 | ch = c; | |
786 | p++; | |
787 | } | |
17431c60 | 788 | if (syntaxp) |
5caa45d3 | 789 | fastmap[syntax_spec_code[c]] = 1; |
17431c60 | 790 | else |
ca1d1d23 | 791 | { |
17431c60 | 792 | if (c == '\\') |
ca1d1d23 | 793 | { |
17431c60 RS |
794 | if (p == pend) break; |
795 | c = *p++; | |
796 | } | |
797 | if (p != pend && *p == '-') | |
798 | { | |
5679531d KH |
799 | unsigned int ch2; |
800 | ||
17431c60 RS |
801 | p++; |
802 | if (p == pend) break; | |
5679531d KH |
803 | if (SINGLE_BYTE_CHAR_P (ch)) |
804 | while (c <= *p) | |
805 | { | |
806 | fastmap[c] = 1; | |
807 | c++; | |
808 | } | |
809 | else | |
17431c60 | 810 | { |
5679531d KH |
811 | fastmap[c] = 1; /* C is the base leading-code. */ |
812 | ch2 = STRING_CHAR (p, pend - p); | |
813 | if (ch <= ch2) | |
814 | char_ranges[n_char_ranges++] = ch, | |
815 | char_ranges[n_char_ranges++] = ch2; | |
17431c60 | 816 | } |
1312cff5 | 817 | p += multibyte ? BYTES_BY_CHAR_HEAD (*p) : 1; |
ca1d1d23 | 818 | } |
17431c60 | 819 | else |
5679531d KH |
820 | { |
821 | fastmap[c] = 1; | |
822 | if (!SINGLE_BYTE_CHAR_P (ch)) | |
823 | char_ranges[n_char_ranges++] = ch, | |
824 | char_ranges[n_char_ranges++] = ch; | |
825 | } | |
ca1d1d23 | 826 | } |
ca1d1d23 JB |
827 | } |
828 | ||
5679531d KH |
829 | /* If ^ was the first character, complement the fastmap. In |
830 | addition, as all multibyte characters have possibility of | |
831 | matching, set all entries for base leading codes, which is | |
832 | harmless even if SYNTAXP is 1. */ | |
ca1d1d23 JB |
833 | |
834 | if (negate) | |
835 | for (i = 0; i < sizeof fastmap; i++) | |
5679531d | 836 | { |
1312cff5 | 837 | if (!multibyte || !BASE_LEADING_CODE_P (i)) |
5679531d KH |
838 | fastmap[i] ^= 1; |
839 | else | |
840 | fastmap[i] = 1; | |
841 | } | |
ca1d1d23 | 842 | |
c1dc99a1 | 843 | { |
6ec8bbd2 | 844 | int start_point = PT; |
5caa45d3 | 845 | int pos = PT; |
c1dc99a1 JB |
846 | |
847 | immediate_quit = 1; | |
17431c60 | 848 | if (syntaxp) |
c1dc99a1 | 849 | { |
17431c60 RS |
850 | if (forwardp) |
851 | { | |
1312cff5 KH |
852 | if (multibyte) |
853 | while (pos < XINT (lim) | |
854 | && fastmap[(int) SYNTAX (FETCH_CHAR (pos))]) | |
855 | INC_POS (pos); | |
856 | else | |
857 | while (pos < XINT (lim) | |
858 | && fastmap[(int) SYNTAX (FETCH_BYTE (pos))]) | |
859 | pos++; | |
17431c60 RS |
860 | } |
861 | else | |
862 | { | |
1312cff5 KH |
863 | if (multibyte) |
864 | while (pos > XINT (lim)) | |
865 | { | |
866 | int savepos = pos; | |
867 | DEC_POS (pos); | |
868 | if (!fastmap[(int) SYNTAX (FETCH_CHAR (pos))]) | |
869 | { | |
870 | pos = savepos; | |
871 | break; | |
872 | } | |
873 | } | |
874 | else | |
875 | while (pos > XINT (lim) | |
876 | && fastmap[(int) SYNTAX (FETCH_BYTE (pos - 1))]) | |
877 | pos--; | |
17431c60 | 878 | } |
c1dc99a1 JB |
879 | } |
880 | else | |
881 | { | |
17431c60 RS |
882 | if (forwardp) |
883 | { | |
1312cff5 KH |
884 | if (multibyte) |
885 | while (pos < XINT (lim) && fastmap[(c = FETCH_BYTE (pos))]) | |
886 | { | |
887 | if (!BASE_LEADING_CODE_P (c)) | |
888 | pos++; | |
889 | else if (n_char_ranges) | |
890 | { | |
891 | /* We much check CHAR_RANGES for a multibyte | |
892 | character. */ | |
893 | ch = FETCH_MULTIBYTE_CHAR (pos); | |
894 | for (i = 0; i < n_char_ranges; i += 2) | |
895 | if ((ch >= char_ranges[i] && ch <= char_ranges[i + 1])) | |
896 | break; | |
897 | if (!(negate ^ (i < n_char_ranges))) | |
5679531d | 898 | break; |
5679531d | 899 | |
1312cff5 KH |
900 | INC_POS (pos); |
901 | } | |
902 | else | |
903 | { | |
904 | if (!negate) break; | |
905 | INC_POS (pos); | |
906 | } | |
907 | } | |
908 | else | |
909 | while (pos < XINT (lim) && fastmap[FETCH_BYTE (pos)]) | |
910 | pos++; | |
17431c60 RS |
911 | } |
912 | else | |
913 | { | |
1312cff5 KH |
914 | if (multibyte) |
915 | while (pos > XINT (lim)) | |
916 | { | |
917 | int savepos = pos; | |
918 | DEC_POS (pos); | |
919 | if (fastmap[(c = FETCH_BYTE (pos))]) | |
920 | { | |
921 | if (!BASE_LEADING_CODE_P (c)) | |
922 | ; | |
923 | else if (n_char_ranges) | |
924 | { | |
925 | /* We much check CHAR_RANGES for a multibyte | |
926 | character. */ | |
927 | ch = FETCH_MULTIBYTE_CHAR (pos); | |
928 | for (i = 0; i < n_char_ranges; i += 2) | |
929 | if (ch >= char_ranges[i] && ch <= char_ranges[i + 1]) | |
930 | break; | |
931 | if (!(negate ^ (i < n_char_ranges))) | |
932 | { | |
933 | pos = savepos; | |
934 | break; | |
935 | } | |
936 | } | |
937 | else | |
938 | if (!negate) | |
f73d78a2 KH |
939 | { |
940 | pos = savepos; | |
941 | break; | |
942 | } | |
1312cff5 KH |
943 | } |
944 | else | |
945 | { | |
946 | pos = savepos; | |
947 | break; | |
948 | } | |
949 | } | |
950 | else | |
951 | while (pos > XINT (lim) && fastmap[FETCH_BYTE (pos - 1)]) | |
952 | pos--; | |
17431c60 | 953 | } |
c1dc99a1 | 954 | } |
1312cff5 KH |
955 | if (multibyte |
956 | /* INC_POS or DEC_POS might have moved POS over LIM. */ | |
957 | && (forwardp ? (pos > XINT (lim)) : (pos < XINT (lim)))) | |
958 | pos = XINT (lim); | |
959 | ||
5caa45d3 | 960 | SET_PT (pos); |
c1dc99a1 JB |
961 | immediate_quit = 0; |
962 | ||
6ec8bbd2 | 963 | return make_number (PT - start_point); |
c1dc99a1 | 964 | } |
ca1d1d23 JB |
965 | } |
966 | \f | |
967 | /* Subroutines of Lisp buffer search functions. */ | |
968 | ||
969 | static Lisp_Object | |
b819a390 | 970 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
971 | Lisp_Object string, bound, noerror, count; |
972 | int direction; | |
973 | int RE; | |
b819a390 | 974 | int posix; |
ca1d1d23 JB |
975 | { |
976 | register int np; | |
977 | int lim; | |
978 | int n = direction; | |
979 | ||
980 | if (!NILP (count)) | |
981 | { | |
982 | CHECK_NUMBER (count, 3); | |
983 | n *= XINT (count); | |
984 | } | |
985 | ||
986 | CHECK_STRING (string, 0); | |
987 | if (NILP (bound)) | |
988 | lim = n > 0 ? ZV : BEGV; | |
989 | else | |
990 | { | |
991 | CHECK_NUMBER_COERCE_MARKER (bound, 1); | |
992 | lim = XINT (bound); | |
6ec8bbd2 | 993 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
994 | error ("Invalid search bound (wrong side of point)"); |
995 | if (lim > ZV) | |
996 | lim = ZV; | |
997 | if (lim < BEGV) | |
998 | lim = BEGV; | |
999 | } | |
1000 | ||
6ec8bbd2 | 1001 | np = search_buffer (string, PT, lim, n, RE, |
ca1d1d23 | 1002 | (!NILP (current_buffer->case_fold_search) |
b1428bd8 RS |
1003 | ? XCHAR_TABLE (current_buffer->case_canon_table)->contents |
1004 | : 0), | |
ca1d1d23 | 1005 | (!NILP (current_buffer->case_fold_search) |
b1428bd8 RS |
1006 | ? XCHAR_TABLE (current_buffer->case_eqv_table)->contents |
1007 | : 0), | |
b819a390 | 1008 | posix); |
ca1d1d23 JB |
1009 | if (np <= 0) |
1010 | { | |
1011 | if (NILP (noerror)) | |
1012 | return signal_failure (string); | |
1013 | if (!EQ (noerror, Qt)) | |
1014 | { | |
1015 | if (lim < BEGV || lim > ZV) | |
1016 | abort (); | |
a5f217b8 RS |
1017 | SET_PT (lim); |
1018 | return Qnil; | |
1019 | #if 0 /* This would be clean, but maybe programs depend on | |
1020 | a value of nil here. */ | |
481399bf | 1021 | np = lim; |
a5f217b8 | 1022 | #endif |
ca1d1d23 | 1023 | } |
481399bf RS |
1024 | else |
1025 | return Qnil; | |
ca1d1d23 JB |
1026 | } |
1027 | ||
1028 | if (np < BEGV || np > ZV) | |
1029 | abort (); | |
1030 | ||
1031 | SET_PT (np); | |
1032 | ||
1033 | return make_number (np); | |
1034 | } | |
1035 | \f | |
b6d6a51c KH |
1036 | static int |
1037 | trivial_regexp_p (regexp) | |
1038 | Lisp_Object regexp; | |
1039 | { | |
1040 | int len = XSTRING (regexp)->size; | |
1041 | unsigned char *s = XSTRING (regexp)->data; | |
1042 | unsigned char c; | |
1043 | while (--len >= 0) | |
1044 | { | |
1045 | switch (*s++) | |
1046 | { | |
1047 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
1048 | return 0; | |
1049 | case '\\': | |
1050 | if (--len < 0) | |
1051 | return 0; | |
1052 | switch (*s++) | |
1053 | { | |
1054 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
1055 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
866f60fd | 1056 | case 'S': case '=': |
5679531d | 1057 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 1058 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
1059 | case '6': case '7': case '8': case '9': |
1060 | return 0; | |
1061 | } | |
1062 | } | |
1063 | } | |
1064 | return 1; | |
1065 | } | |
1066 | ||
ca325161 | 1067 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 1068 | starting at position POS and stopping at position LIM, |
b819a390 | 1069 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
1070 | a regular expression if RE is true. |
1071 | ||
1072 | If N is positive, searching is forward and LIM must be greater than POS. | |
1073 | If N is negative, searching is backward and LIM must be less than POS. | |
1074 | ||
1075 | Returns -x if only N-x occurrences found (x > 0), | |
1076 | or else the position at the beginning of the Nth occurrence | |
b819a390 RS |
1077 | (if searching backward) or the end (if searching forward). |
1078 | ||
1079 | POSIX is nonzero if we want full backtracking (POSIX style) | |
1080 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 1081 | |
b819a390 RS |
1082 | static int |
1083 | search_buffer (string, pos, lim, n, RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
1084 | Lisp_Object string; |
1085 | int pos; | |
1086 | int lim; | |
1087 | int n; | |
1088 | int RE; | |
b1428bd8 RS |
1089 | Lisp_Object *trt; |
1090 | Lisp_Object *inverse_trt; | |
b819a390 | 1091 | int posix; |
ca1d1d23 JB |
1092 | { |
1093 | int len = XSTRING (string)->size; | |
1094 | unsigned char *base_pat = XSTRING (string)->data; | |
1095 | register int *BM_tab; | |
1096 | int *BM_tab_base; | |
1097 | register int direction = ((n > 0) ? 1 : -1); | |
1098 | register int dirlen; | |
1099 | int infinity, limit, k, stride_for_teases; | |
1100 | register unsigned char *pat, *cursor, *p_limit; | |
1101 | register int i, j; | |
1102 | unsigned char *p1, *p2; | |
1103 | int s1, s2; | |
1104 | ||
7074fde6 FP |
1105 | if (running_asynch_code) |
1106 | save_search_regs (); | |
1107 | ||
ca1d1d23 | 1108 | /* Null string is found at starting position. */ |
3f57a499 | 1109 | if (len == 0) |
ca325161 RS |
1110 | { |
1111 | set_search_regs (pos, 0); | |
1112 | return pos; | |
1113 | } | |
3f57a499 RS |
1114 | |
1115 | /* Searching 0 times means don't move. */ | |
1116 | if (n == 0) | |
ca1d1d23 JB |
1117 | return pos; |
1118 | ||
b6d6a51c | 1119 | if (RE && !trivial_regexp_p (string)) |
ca1d1d23 | 1120 | { |
487282dc KH |
1121 | struct re_pattern_buffer *bufp; |
1122 | ||
b1428bd8 | 1123 | bufp = compile_pattern (string, &search_regs, trt, posix); |
ca1d1d23 | 1124 | |
ca1d1d23 JB |
1125 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1126 | because letting this function finish | |
1127 | can take too long. */ | |
1128 | QUIT; /* Do a pending quit right away, | |
1129 | to avoid paradoxical behavior */ | |
1130 | /* Get pointers and sizes of the two strings | |
1131 | that make up the visible portion of the buffer. */ | |
1132 | ||
1133 | p1 = BEGV_ADDR; | |
1134 | s1 = GPT - BEGV; | |
1135 | p2 = GAP_END_ADDR; | |
1136 | s2 = ZV - GPT; | |
1137 | if (s1 < 0) | |
1138 | { | |
1139 | p2 = p1; | |
1140 | s2 = ZV - BEGV; | |
1141 | s1 = 0; | |
1142 | } | |
1143 | if (s2 < 0) | |
1144 | { | |
1145 | s1 = ZV - BEGV; | |
1146 | s2 = 0; | |
1147 | } | |
1148 | while (n < 0) | |
1149 | { | |
42db823b | 1150 | int val; |
487282dc | 1151 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
42db823b RS |
1152 | pos - BEGV, lim - pos, &search_regs, |
1153 | /* Don't allow match past current point */ | |
1154 | pos - BEGV); | |
ca1d1d23 | 1155 | if (val == -2) |
b6d6a51c KH |
1156 | { |
1157 | matcher_overflow (); | |
1158 | } | |
ca1d1d23 JB |
1159 | if (val >= 0) |
1160 | { | |
1161 | j = BEGV; | |
4746118a | 1162 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1163 | if (search_regs.start[i] >= 0) |
1164 | { | |
1165 | search_regs.start[i] += j; | |
1166 | search_regs.end[i] += j; | |
1167 | } | |
a3668d92 | 1168 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1169 | /* Set pos to the new position. */ |
1170 | pos = search_regs.start[0]; | |
1171 | } | |
1172 | else | |
1173 | { | |
1174 | immediate_quit = 0; | |
1175 | return (n); | |
1176 | } | |
1177 | n++; | |
1178 | } | |
1179 | while (n > 0) | |
1180 | { | |
42db823b | 1181 | int val; |
487282dc | 1182 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
42db823b RS |
1183 | pos - BEGV, lim - pos, &search_regs, |
1184 | lim - BEGV); | |
ca1d1d23 | 1185 | if (val == -2) |
b6d6a51c KH |
1186 | { |
1187 | matcher_overflow (); | |
1188 | } | |
ca1d1d23 JB |
1189 | if (val >= 0) |
1190 | { | |
1191 | j = BEGV; | |
4746118a | 1192 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1193 | if (search_regs.start[i] >= 0) |
1194 | { | |
1195 | search_regs.start[i] += j; | |
1196 | search_regs.end[i] += j; | |
1197 | } | |
a3668d92 | 1198 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1199 | pos = search_regs.end[0]; |
1200 | } | |
1201 | else | |
1202 | { | |
1203 | immediate_quit = 0; | |
1204 | return (0 - n); | |
1205 | } | |
1206 | n--; | |
1207 | } | |
1208 | immediate_quit = 0; | |
1209 | return (pos); | |
1210 | } | |
1211 | else /* non-RE case */ | |
1212 | { | |
1213 | #ifdef C_ALLOCA | |
1214 | int BM_tab_space[0400]; | |
1215 | BM_tab = &BM_tab_space[0]; | |
1216 | #else | |
1217 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
1218 | #endif | |
b6d6a51c KH |
1219 | { |
1220 | unsigned char *patbuf = (unsigned char *) alloca (len); | |
1221 | pat = patbuf; | |
1222 | while (--len >= 0) | |
1223 | { | |
1224 | /* If we got here and the RE flag is set, it's because we're | |
1225 | dealing with a regexp known to be trivial, so the backslash | |
1226 | just quotes the next character. */ | |
1227 | if (RE && *base_pat == '\\') | |
1228 | { | |
1229 | len--; | |
1230 | base_pat++; | |
1231 | } | |
1232 | *pat++ = (trt ? trt[*base_pat++] : *base_pat++); | |
1233 | } | |
1234 | len = pat - patbuf; | |
1235 | pat = base_pat = patbuf; | |
1236 | } | |
ca1d1d23 JB |
1237 | /* The general approach is that we are going to maintain that we know */ |
1238 | /* the first (closest to the present position, in whatever direction */ | |
1239 | /* we're searching) character that could possibly be the last */ | |
1240 | /* (furthest from present position) character of a valid match. We */ | |
1241 | /* advance the state of our knowledge by looking at that character */ | |
1242 | /* and seeing whether it indeed matches the last character of the */ | |
1243 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1244 | /* move our pointer (to putative last characters) as far as is */ | |
1245 | /* logically possible. This amount of movement, which I call a */ | |
1246 | /* stride, will be the length of the pattern if the actual character */ | |
1247 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1248 | /* from the last occurrence of that character to the end of the */ | |
1249 | /* pattern. */ | |
1250 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1251 | /* characters that match the last character. This allows use of only */ | |
1252 | /* a single test, a test for having gone past the end of the */ | |
1253 | /* permissible match region, to test for both possible matches (when */ | |
1254 | /* the stride goes past the end immediately) and failure to */ | |
1255 | /* match (where you get nudged past the end one stride at a time). */ | |
1256 | ||
1257 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1258 | /* is determined only by the last character of the putative match. */ | |
1259 | /* If that character does not match, we will stride the proper */ | |
1260 | /* distance to propose a match that superimposes it on the last */ | |
1261 | /* instance of a character that matches it (per trt), or misses */ | |
1262 | /* it entirely if there is none. */ | |
1263 | ||
1264 | dirlen = len * direction; | |
1265 | infinity = dirlen - (lim + pos + len + len) * direction; | |
1266 | if (direction < 0) | |
1267 | pat = (base_pat += len - 1); | |
1268 | BM_tab_base = BM_tab; | |
1269 | BM_tab += 0400; | |
1270 | j = dirlen; /* to get it in a register */ | |
1271 | /* A character that does not appear in the pattern induces a */ | |
1272 | /* stride equal to the pattern length. */ | |
1273 | while (BM_tab_base != BM_tab) | |
1274 | { | |
1275 | *--BM_tab = j; | |
1276 | *--BM_tab = j; | |
1277 | *--BM_tab = j; | |
1278 | *--BM_tab = j; | |
1279 | } | |
1280 | i = 0; | |
1281 | while (i != infinity) | |
1282 | { | |
1283 | j = pat[i]; i += direction; | |
1284 | if (i == dirlen) i = infinity; | |
8d505039 | 1285 | if (trt != 0) |
ca1d1d23 JB |
1286 | { |
1287 | k = (j = trt[j]); | |
1288 | if (i == infinity) | |
1289 | stride_for_teases = BM_tab[j]; | |
1290 | BM_tab[j] = dirlen - i; | |
1291 | /* A translation table is accompanied by its inverse -- see */ | |
1292 | /* comment following downcase_table for details */ | |
b1428bd8 | 1293 | while ((j = (unsigned char) inverse_trt[j]) != k) |
ca1d1d23 JB |
1294 | BM_tab[j] = dirlen - i; |
1295 | } | |
1296 | else | |
1297 | { | |
1298 | if (i == infinity) | |
1299 | stride_for_teases = BM_tab[j]; | |
1300 | BM_tab[j] = dirlen - i; | |
1301 | } | |
1302 | /* stride_for_teases tells how much to stride if we get a */ | |
1303 | /* match on the far character but are subsequently */ | |
1304 | /* disappointed, by recording what the stride would have been */ | |
1305 | /* for that character if the last character had been */ | |
1306 | /* different. */ | |
1307 | } | |
1308 | infinity = dirlen - infinity; | |
1309 | pos += dirlen - ((direction > 0) ? direction : 0); | |
1310 | /* loop invariant - pos points at where last char (first char if reverse) | |
1311 | of pattern would align in a possible match. */ | |
1312 | while (n != 0) | |
1313 | { | |
b2c71fb4 KH |
1314 | /* It's been reported that some (broken) compiler thinks that |
1315 | Boolean expressions in an arithmetic context are unsigned. | |
1316 | Using an explicit ?1:0 prevents this. */ | |
1317 | if ((lim - pos - ((direction > 0) ? 1 : 0)) * direction < 0) | |
ca1d1d23 JB |
1318 | return (n * (0 - direction)); |
1319 | /* First we do the part we can by pointers (maybe nothing) */ | |
1320 | QUIT; | |
1321 | pat = base_pat; | |
1322 | limit = pos - dirlen + direction; | |
1323 | limit = ((direction > 0) | |
1324 | ? BUFFER_CEILING_OF (limit) | |
1325 | : BUFFER_FLOOR_OF (limit)); | |
1326 | /* LIMIT is now the last (not beyond-last!) value | |
1327 | POS can take on without hitting edge of buffer or the gap. */ | |
1328 | limit = ((direction > 0) | |
1329 | ? min (lim - 1, min (limit, pos + 20000)) | |
1330 | : max (lim, max (limit, pos - 20000))); | |
1331 | if ((limit - pos) * direction > 20) | |
1332 | { | |
5679531d KH |
1333 | p_limit = POS_ADDR (limit); |
1334 | p2 = (cursor = POS_ADDR (pos)); | |
ca1d1d23 JB |
1335 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ |
1336 | while (1) /* use one cursor setting as long as i can */ | |
1337 | { | |
1338 | if (direction > 0) /* worth duplicating */ | |
1339 | { | |
1340 | /* Use signed comparison if appropriate | |
1341 | to make cursor+infinity sure to be > p_limit. | |
1342 | Assuming that the buffer lies in a range of addresses | |
1343 | that are all "positive" (as ints) or all "negative", | |
1344 | either kind of comparison will work as long | |
1345 | as we don't step by infinity. So pick the kind | |
1346 | that works when we do step by infinity. */ | |
8d505039 | 1347 | if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit) |
9fa17f93 | 1348 | while ((EMACS_INT) cursor <= (EMACS_INT) p_limit) |
ca1d1d23 JB |
1349 | cursor += BM_tab[*cursor]; |
1350 | else | |
45b248b4 | 1351 | while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit) |
ca1d1d23 JB |
1352 | cursor += BM_tab[*cursor]; |
1353 | } | |
1354 | else | |
1355 | { | |
8d505039 RS |
1356 | if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit) |
1357 | while ((EMACS_INT) cursor >= (EMACS_INT) p_limit) | |
ca1d1d23 JB |
1358 | cursor += BM_tab[*cursor]; |
1359 | else | |
45b248b4 | 1360 | while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit) |
ca1d1d23 JB |
1361 | cursor += BM_tab[*cursor]; |
1362 | } | |
1363 | /* If you are here, cursor is beyond the end of the searched region. */ | |
1364 | /* This can happen if you match on the far character of the pattern, */ | |
1365 | /* because the "stride" of that character is infinity, a number able */ | |
1366 | /* to throw you well beyond the end of the search. It can also */ | |
1367 | /* happen if you fail to match within the permitted region and would */ | |
1368 | /* otherwise try a character beyond that region */ | |
1369 | if ((cursor - p_limit) * direction <= len) | |
1370 | break; /* a small overrun is genuine */ | |
1371 | cursor -= infinity; /* large overrun = hit */ | |
1372 | i = dirlen - direction; | |
8d505039 | 1373 | if (trt != 0) |
ca1d1d23 JB |
1374 | { |
1375 | while ((i -= direction) + direction != 0) | |
1376 | if (pat[i] != trt[*(cursor -= direction)]) | |
1377 | break; | |
1378 | } | |
1379 | else | |
1380 | { | |
1381 | while ((i -= direction) + direction != 0) | |
1382 | if (pat[i] != *(cursor -= direction)) | |
1383 | break; | |
1384 | } | |
1385 | cursor += dirlen - i - direction; /* fix cursor */ | |
1386 | if (i + direction == 0) | |
1387 | { | |
1388 | cursor -= direction; | |
1113d9db | 1389 | |
ca325161 RS |
1390 | set_search_regs (pos + cursor - p2 + ((direction > 0) |
1391 | ? 1 - len : 0), | |
1392 | len); | |
1393 | ||
ca1d1d23 JB |
1394 | if ((n -= direction) != 0) |
1395 | cursor += dirlen; /* to resume search */ | |
1396 | else | |
1397 | return ((direction > 0) | |
1398 | ? search_regs.end[0] : search_regs.start[0]); | |
1399 | } | |
1400 | else | |
1401 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
1402 | } | |
1403 | pos += cursor - p2; | |
1404 | } | |
1405 | else | |
1406 | /* Now we'll pick up a clump that has to be done the hard */ | |
1407 | /* way because it covers a discontinuity */ | |
1408 | { | |
1409 | limit = ((direction > 0) | |
1410 | ? BUFFER_CEILING_OF (pos - dirlen + 1) | |
1411 | : BUFFER_FLOOR_OF (pos - dirlen - 1)); | |
1412 | limit = ((direction > 0) | |
1413 | ? min (limit + len, lim - 1) | |
1414 | : max (limit - len, lim)); | |
1415 | /* LIMIT is now the last value POS can have | |
1416 | and still be valid for a possible match. */ | |
1417 | while (1) | |
1418 | { | |
1419 | /* This loop can be coded for space rather than */ | |
1420 | /* speed because it will usually run only once. */ | |
1421 | /* (the reach is at most len + 21, and typically */ | |
1422 | /* does not exceed len) */ | |
1423 | while ((limit - pos) * direction >= 0) | |
5679531d | 1424 | pos += BM_tab[FETCH_BYTE (pos)]; |
ca1d1d23 | 1425 | /* now run the same tests to distinguish going off the */ |
eb8c3be9 | 1426 | /* end, a match or a phony match. */ |
ca1d1d23 JB |
1427 | if ((pos - limit) * direction <= len) |
1428 | break; /* ran off the end */ | |
1429 | /* Found what might be a match. | |
1430 | Set POS back to last (first if reverse) char pos. */ | |
1431 | pos -= infinity; | |
1432 | i = dirlen - direction; | |
1433 | while ((i -= direction) + direction != 0) | |
1434 | { | |
1435 | pos -= direction; | |
8d505039 | 1436 | if (pat[i] != (trt != 0 |
5679531d KH |
1437 | ? trt[FETCH_BYTE (pos)] |
1438 | : FETCH_BYTE (pos))) | |
ca1d1d23 JB |
1439 | break; |
1440 | } | |
1441 | /* Above loop has moved POS part or all the way | |
1442 | back to the first char pos (last char pos if reverse). | |
1443 | Set it once again at the last (first if reverse) char. */ | |
1444 | pos += dirlen - i- direction; | |
1445 | if (i + direction == 0) | |
1446 | { | |
1447 | pos -= direction; | |
1113d9db | 1448 | |
ca325161 RS |
1449 | set_search_regs (pos + ((direction > 0) ? 1 - len : 0), |
1450 | len); | |
1451 | ||
ca1d1d23 JB |
1452 | if ((n -= direction) != 0) |
1453 | pos += dirlen; /* to resume search */ | |
1454 | else | |
1455 | return ((direction > 0) | |
1456 | ? search_regs.end[0] : search_regs.start[0]); | |
1457 | } | |
1458 | else | |
1459 | pos += stride_for_teases; | |
1460 | } | |
1461 | } | |
1462 | /* We have done one clump. Can we continue? */ | |
1463 | if ((lim - pos) * direction < 0) | |
1464 | return ((0 - n) * direction); | |
1465 | } | |
1466 | return pos; | |
1467 | } | |
1468 | } | |
ca325161 RS |
1469 | |
1470 | /* Record beginning BEG and end BEG + LEN | |
1471 | for a match just found in the current buffer. */ | |
1472 | ||
1473 | static void | |
1474 | set_search_regs (beg, len) | |
1475 | int beg, len; | |
1476 | { | |
1477 | /* Make sure we have registers in which to store | |
1478 | the match position. */ | |
1479 | if (search_regs.num_regs == 0) | |
1480 | { | |
2d4a771a RS |
1481 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
1482 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 1483 | search_regs.num_regs = 2; |
ca325161 RS |
1484 | } |
1485 | ||
1486 | search_regs.start[0] = beg; | |
1487 | search_regs.end[0] = beg + len; | |
a3668d92 | 1488 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 1489 | } |
ca1d1d23 JB |
1490 | \f |
1491 | /* Given a string of words separated by word delimiters, | |
1492 | compute a regexp that matches those exact words | |
1493 | separated by arbitrary punctuation. */ | |
1494 | ||
1495 | static Lisp_Object | |
1496 | wordify (string) | |
1497 | Lisp_Object string; | |
1498 | { | |
1499 | register unsigned char *p, *o; | |
1500 | register int i, len, punct_count = 0, word_count = 0; | |
1501 | Lisp_Object val; | |
1502 | ||
1503 | CHECK_STRING (string, 0); | |
1504 | p = XSTRING (string)->data; | |
1505 | len = XSTRING (string)->size; | |
1506 | ||
1507 | for (i = 0; i < len; i++) | |
1508 | if (SYNTAX (p[i]) != Sword) | |
1509 | { | |
1510 | punct_count++; | |
1511 | if (i > 0 && SYNTAX (p[i-1]) == Sword) word_count++; | |
1512 | } | |
1513 | if (SYNTAX (p[len-1]) == Sword) word_count++; | |
1514 | if (!word_count) return build_string (""); | |
1515 | ||
1516 | val = make_string (p, len - punct_count + 5 * (word_count - 1) + 4); | |
1517 | ||
1518 | o = XSTRING (val)->data; | |
1519 | *o++ = '\\'; | |
1520 | *o++ = 'b'; | |
1521 | ||
1522 | for (i = 0; i < len; i++) | |
1523 | if (SYNTAX (p[i]) == Sword) | |
1524 | *o++ = p[i]; | |
1525 | else if (i > 0 && SYNTAX (p[i-1]) == Sword && --word_count) | |
1526 | { | |
1527 | *o++ = '\\'; | |
1528 | *o++ = 'W'; | |
1529 | *o++ = '\\'; | |
1530 | *o++ = 'W'; | |
1531 | *o++ = '*'; | |
1532 | } | |
1533 | ||
1534 | *o++ = '\\'; | |
1535 | *o++ = 'b'; | |
1536 | ||
1537 | return val; | |
1538 | } | |
1539 | \f | |
1540 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
1541 | "sSearch backward: ", | |
1542 | "Search backward from point for STRING.\n\ | |
1543 | Set point to the beginning of the occurrence found, and return point.\n\ | |
1544 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1545 | The match found must not extend before that position.\n\ | |
1546 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1547 | If not nil and not t, position at limit of search and return nil.\n\ | |
1548 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1549 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1550 | (string, bound, noerror, count) | |
1551 | Lisp_Object string, bound, noerror, count; | |
1552 | { | |
b819a390 | 1553 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
1554 | } |
1555 | ||
1556 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "sSearch: ", | |
1557 | "Search forward from point for STRING.\n\ | |
1558 | Set point to the end of the occurrence found, and return point.\n\ | |
1559 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1560 | The match found must not extend after that position. nil is equivalent\n\ | |
1561 | to (point-max).\n\ | |
1562 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1563 | If not nil and not t, move to limit of search and return nil.\n\ | |
1564 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1565 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1566 | (string, bound, noerror, count) | |
1567 | Lisp_Object string, bound, noerror, count; | |
1568 | { | |
b819a390 | 1569 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
1570 | } |
1571 | ||
1572 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
1573 | "sWord search backward: ", | |
1574 | "Search backward from point for STRING, ignoring differences in punctuation.\n\ | |
1575 | Set point to the beginning of the occurrence found, and return point.\n\ | |
1576 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1577 | The match found must not extend before that position.\n\ | |
1578 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1579 | If not nil and not t, move to limit of search and return nil.\n\ | |
1580 | Optional fourth argument is repeat count--search for successive occurrences.") | |
1581 | (string, bound, noerror, count) | |
1582 | Lisp_Object string, bound, noerror, count; | |
1583 | { | |
b819a390 | 1584 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
1585 | } |
1586 | ||
1587 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
1588 | "sWord search: ", | |
1589 | "Search forward from point for STRING, ignoring differences in punctuation.\n\ | |
1590 | Set point to the end of the occurrence found, and return point.\n\ | |
1591 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1592 | The match found must not extend after that position.\n\ | |
1593 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1594 | If not nil and not t, move to limit of search and return nil.\n\ | |
1595 | Optional fourth argument is repeat count--search for successive occurrences.") | |
1596 | (string, bound, noerror, count) | |
1597 | Lisp_Object string, bound, noerror, count; | |
1598 | { | |
b819a390 | 1599 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
1600 | } |
1601 | ||
1602 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
1603 | "sRE search backward: ", | |
1604 | "Search backward from point for match for regular expression REGEXP.\n\ | |
1605 | Set point to the beginning of the match, and return point.\n\ | |
1606 | The match found is the one starting last in the buffer\n\ | |
19c0a730 | 1607 | and yet ending before the origin of the search.\n\ |
ca1d1d23 JB |
1608 | An optional second argument bounds the search; it is a buffer position.\n\ |
1609 | The match found must start at or after that position.\n\ | |
1610 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1611 | If not nil and not t, move to limit of search and return nil.\n\ | |
1612 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1613 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
19c0a730 KH |
1614 | (regexp, bound, noerror, count) |
1615 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 1616 | { |
b819a390 | 1617 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
1618 | } |
1619 | ||
1620 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
1621 | "sRE search: ", | |
1622 | "Search forward from point for regular expression REGEXP.\n\ | |
1623 | Set point to the end of the occurrence found, and return point.\n\ | |
1624 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1625 | The match found must not extend after that position.\n\ | |
1626 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1627 | If not nil and not t, move to limit of search and return nil.\n\ | |
1628 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1629 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
19c0a730 KH |
1630 | (regexp, bound, noerror, count) |
1631 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 1632 | { |
b819a390 RS |
1633 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
1634 | } | |
1635 | ||
1636 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
1637 | "sPosix search backward: ", | |
1638 | "Search backward from point for match for regular expression REGEXP.\n\ | |
1639 | Find the longest match in accord with Posix regular expression rules.\n\ | |
1640 | Set point to the beginning of the match, and return point.\n\ | |
1641 | The match found is the one starting last in the buffer\n\ | |
1642 | and yet ending before the origin of the search.\n\ | |
1643 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1644 | The match found must start at or after that position.\n\ | |
1645 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1646 | If not nil and not t, move to limit of search and return nil.\n\ | |
1647 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1648 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1649 | (regexp, bound, noerror, count) | |
1650 | Lisp_Object regexp, bound, noerror, count; | |
1651 | { | |
1652 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
1653 | } | |
1654 | ||
1655 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
1656 | "sPosix search: ", | |
1657 | "Search forward from point for regular expression REGEXP.\n\ | |
1658 | Find the longest match in accord with Posix regular expression rules.\n\ | |
1659 | Set point to the end of the occurrence found, and return point.\n\ | |
1660 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1661 | The match found must not extend after that position.\n\ | |
1662 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1663 | If not nil and not t, move to limit of search and return nil.\n\ | |
1664 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1665 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1666 | (regexp, bound, noerror, count) | |
1667 | Lisp_Object regexp, bound, noerror, count; | |
1668 | { | |
1669 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
1670 | } |
1671 | \f | |
d7a5ad5f | 1672 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
ca1d1d23 JB |
1673 | "Replace text matched by last search with NEWTEXT.\n\ |
1674 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text.\n\ | |
5b9cf4b2 RS |
1675 | Otherwise maybe capitalize the whole text, or maybe just word initials,\n\ |
1676 | based on the replaced text.\n\ | |
1677 | If the replaced text has only capital letters\n\ | |
1678 | and has at least one multiletter word, convert NEWTEXT to all caps.\n\ | |
1679 | If the replaced text has at least one word starting with a capital letter,\n\ | |
1680 | then capitalize each word in NEWTEXT.\n\n\ | |
ca1d1d23 JB |
1681 | If third arg LITERAL is non-nil, insert NEWTEXT literally.\n\ |
1682 | Otherwise treat `\\' as special:\n\ | |
1683 | `\\&' in NEWTEXT means substitute original matched text.\n\ | |
1684 | `\\N' means substitute what matched the Nth `\\(...\\)'.\n\ | |
1685 | If Nth parens didn't match, substitute nothing.\n\ | |
1686 | `\\\\' means insert one `\\'.\n\ | |
1113d9db | 1687 | FIXEDCASE and LITERAL are optional arguments.\n\ |
080c45fd RS |
1688 | Leaves point at end of replacement text.\n\ |
1689 | \n\ | |
1690 | The optional fourth argument STRING can be a string to modify.\n\ | |
1691 | In that case, this function creates and returns a new string\n\ | |
d7a5ad5f RS |
1692 | which is made by replacing the part of STRING that was matched.\n\ |
1693 | \n\ | |
1694 | The optional fifth argument SUBEXP specifies a subexpression of the match.\n\ | |
1695 | It says to replace just that subexpression instead of the whole match.\n\ | |
1696 | This is useful only after a regular expression search or match\n\ | |
1697 | since only regular expressions have distinguished subexpressions.") | |
1698 | (newtext, fixedcase, literal, string, subexp) | |
1699 | Lisp_Object newtext, fixedcase, literal, string, subexp; | |
ca1d1d23 JB |
1700 | { |
1701 | enum { nochange, all_caps, cap_initial } case_action; | |
1702 | register int pos, last; | |
1703 | int some_multiletter_word; | |
97832bd0 | 1704 | int some_lowercase; |
73dc8771 | 1705 | int some_uppercase; |
208767c3 | 1706 | int some_nonuppercase_initial; |
ca1d1d23 JB |
1707 | register int c, prevc; |
1708 | int inslen; | |
d7a5ad5f | 1709 | int sub; |
ca1d1d23 | 1710 | |
16fdc568 | 1711 | CHECK_STRING (newtext, 0); |
ca1d1d23 | 1712 | |
080c45fd RS |
1713 | if (! NILP (string)) |
1714 | CHECK_STRING (string, 4); | |
1715 | ||
ca1d1d23 JB |
1716 | case_action = nochange; /* We tried an initialization */ |
1717 | /* but some C compilers blew it */ | |
4746118a JB |
1718 | |
1719 | if (search_regs.num_regs <= 0) | |
1720 | error ("replace-match called before any match found"); | |
1721 | ||
d7a5ad5f RS |
1722 | if (NILP (subexp)) |
1723 | sub = 0; | |
1724 | else | |
1725 | { | |
1726 | CHECK_NUMBER (subexp, 3); | |
1727 | sub = XINT (subexp); | |
1728 | if (sub < 0 || sub >= search_regs.num_regs) | |
1729 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
1730 | } | |
1731 | ||
080c45fd RS |
1732 | if (NILP (string)) |
1733 | { | |
d7a5ad5f RS |
1734 | if (search_regs.start[sub] < BEGV |
1735 | || search_regs.start[sub] > search_regs.end[sub] | |
1736 | || search_regs.end[sub] > ZV) | |
1737 | args_out_of_range (make_number (search_regs.start[sub]), | |
1738 | make_number (search_regs.end[sub])); | |
080c45fd RS |
1739 | } |
1740 | else | |
1741 | { | |
d7a5ad5f RS |
1742 | if (search_regs.start[sub] < 0 |
1743 | || search_regs.start[sub] > search_regs.end[sub] | |
1744 | || search_regs.end[sub] > XSTRING (string)->size) | |
1745 | args_out_of_range (make_number (search_regs.start[sub]), | |
1746 | make_number (search_regs.end[sub])); | |
080c45fd | 1747 | } |
ca1d1d23 JB |
1748 | |
1749 | if (NILP (fixedcase)) | |
1750 | { | |
1751 | /* Decide how to casify by examining the matched text. */ | |
1752 | ||
d7a5ad5f | 1753 | last = search_regs.end[sub]; |
ca1d1d23 JB |
1754 | prevc = '\n'; |
1755 | case_action = all_caps; | |
1756 | ||
1757 | /* some_multiletter_word is set nonzero if any original word | |
1758 | is more than one letter long. */ | |
1759 | some_multiletter_word = 0; | |
97832bd0 | 1760 | some_lowercase = 0; |
208767c3 | 1761 | some_nonuppercase_initial = 0; |
73dc8771 | 1762 | some_uppercase = 0; |
ca1d1d23 | 1763 | |
d7a5ad5f | 1764 | for (pos = search_regs.start[sub]; pos < last; pos++) |
ca1d1d23 | 1765 | { |
080c45fd | 1766 | if (NILP (string)) |
5679531d | 1767 | c = FETCH_BYTE (pos); |
080c45fd RS |
1768 | else |
1769 | c = XSTRING (string)->data[pos]; | |
1770 | ||
ca1d1d23 JB |
1771 | if (LOWERCASEP (c)) |
1772 | { | |
1773 | /* Cannot be all caps if any original char is lower case */ | |
1774 | ||
97832bd0 | 1775 | some_lowercase = 1; |
ca1d1d23 | 1776 | if (SYNTAX (prevc) != Sword) |
208767c3 | 1777 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
1778 | else |
1779 | some_multiletter_word = 1; | |
1780 | } | |
1781 | else if (!NOCASEP (c)) | |
1782 | { | |
73dc8771 | 1783 | some_uppercase = 1; |
97832bd0 | 1784 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 1785 | ; |
97832bd0 | 1786 | else |
ca1d1d23 JB |
1787 | some_multiletter_word = 1; |
1788 | } | |
208767c3 RS |
1789 | else |
1790 | { | |
1791 | /* If the initial is a caseless word constituent, | |
1792 | treat that like a lowercase initial. */ | |
1793 | if (SYNTAX (prevc) != Sword) | |
1794 | some_nonuppercase_initial = 1; | |
1795 | } | |
ca1d1d23 JB |
1796 | |
1797 | prevc = c; | |
1798 | } | |
1799 | ||
97832bd0 RS |
1800 | /* Convert to all caps if the old text is all caps |
1801 | and has at least one multiletter word. */ | |
1802 | if (! some_lowercase && some_multiletter_word) | |
1803 | case_action = all_caps; | |
c4d460ce | 1804 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 1805 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 1806 | case_action = cap_initial; |
208767c3 | 1807 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
1808 | /* Should x -> yz, operating on X, give Yz or YZ? |
1809 | We'll assume the latter. */ | |
1810 | case_action = all_caps; | |
97832bd0 RS |
1811 | else |
1812 | case_action = nochange; | |
ca1d1d23 JB |
1813 | } |
1814 | ||
080c45fd RS |
1815 | /* Do replacement in a string. */ |
1816 | if (!NILP (string)) | |
1817 | { | |
1818 | Lisp_Object before, after; | |
1819 | ||
1820 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
1821 | make_number (search_regs.start[sub])); |
1822 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd RS |
1823 | |
1824 | /* Do case substitution into NEWTEXT if desired. */ | |
1825 | if (NILP (literal)) | |
1826 | { | |
1827 | int lastpos = -1; | |
1828 | /* We build up the substituted string in ACCUM. */ | |
1829 | Lisp_Object accum; | |
1830 | Lisp_Object middle; | |
1831 | ||
1832 | accum = Qnil; | |
1833 | ||
1834 | for (pos = 0; pos < XSTRING (newtext)->size; pos++) | |
1835 | { | |
1836 | int substart = -1; | |
1837 | int subend; | |
1e79ec24 | 1838 | int delbackslash = 0; |
080c45fd RS |
1839 | |
1840 | c = XSTRING (newtext)->data[pos]; | |
1841 | if (c == '\\') | |
1842 | { | |
1843 | c = XSTRING (newtext)->data[++pos]; | |
1844 | if (c == '&') | |
1845 | { | |
d7a5ad5f RS |
1846 | substart = search_regs.start[sub]; |
1847 | subend = search_regs.end[sub]; | |
080c45fd RS |
1848 | } |
1849 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') | |
1850 | { | |
ad10348f | 1851 | if (search_regs.start[c - '0'] >= 0) |
080c45fd RS |
1852 | { |
1853 | substart = search_regs.start[c - '0']; | |
1854 | subend = search_regs.end[c - '0']; | |
1855 | } | |
1856 | } | |
1e79ec24 KH |
1857 | else if (c == '\\') |
1858 | delbackslash = 1; | |
080c45fd RS |
1859 | } |
1860 | if (substart >= 0) | |
1861 | { | |
1862 | if (pos - 1 != lastpos + 1) | |
1e79ec24 KH |
1863 | middle = Fsubstring (newtext, |
1864 | make_number (lastpos + 1), | |
1865 | make_number (pos - 1)); | |
080c45fd RS |
1866 | else |
1867 | middle = Qnil; | |
1868 | accum = concat3 (accum, middle, | |
1869 | Fsubstring (string, make_number (substart), | |
1870 | make_number (subend))); | |
1871 | lastpos = pos; | |
1872 | } | |
1e79ec24 KH |
1873 | else if (delbackslash) |
1874 | { | |
1875 | middle = Fsubstring (newtext, make_number (lastpos + 1), | |
1876 | make_number (pos)); | |
1877 | accum = concat2 (accum, middle); | |
1878 | lastpos = pos; | |
1879 | } | |
080c45fd RS |
1880 | } |
1881 | ||
1882 | if (pos != lastpos + 1) | |
1e79ec24 KH |
1883 | middle = Fsubstring (newtext, make_number (lastpos + 1), |
1884 | make_number (pos)); | |
080c45fd RS |
1885 | else |
1886 | middle = Qnil; | |
1887 | ||
1888 | newtext = concat2 (accum, middle); | |
1889 | } | |
1890 | ||
1891 | if (case_action == all_caps) | |
1892 | newtext = Fupcase (newtext); | |
1893 | else if (case_action == cap_initial) | |
2b2eead9 | 1894 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
1895 | |
1896 | return concat3 (before, newtext, after); | |
1897 | } | |
1898 | ||
9a76659d JB |
1899 | /* We insert the replacement text before the old text, and then |
1900 | delete the original text. This means that markers at the | |
1901 | beginning or end of the original will float to the corresponding | |
1902 | position in the replacement. */ | |
d7a5ad5f | 1903 | SET_PT (search_regs.start[sub]); |
ca1d1d23 | 1904 | if (!NILP (literal)) |
16fdc568 | 1905 | Finsert_and_inherit (1, &newtext); |
ca1d1d23 JB |
1906 | else |
1907 | { | |
1908 | struct gcpro gcpro1; | |
16fdc568 | 1909 | GCPRO1 (newtext); |
ca1d1d23 | 1910 | |
16fdc568 | 1911 | for (pos = 0; pos < XSTRING (newtext)->size; pos++) |
ca1d1d23 | 1912 | { |
6ec8bbd2 | 1913 | int offset = PT - search_regs.start[sub]; |
9a76659d | 1914 | |
16fdc568 | 1915 | c = XSTRING (newtext)->data[pos]; |
ca1d1d23 JB |
1916 | if (c == '\\') |
1917 | { | |
16fdc568 | 1918 | c = XSTRING (newtext)->data[++pos]; |
ca1d1d23 | 1919 | if (c == '&') |
9a76659d JB |
1920 | Finsert_buffer_substring |
1921 | (Fcurrent_buffer (), | |
d7a5ad5f RS |
1922 | make_number (search_regs.start[sub] + offset), |
1923 | make_number (search_regs.end[sub] + offset)); | |
78445046 | 1924 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
1925 | { |
1926 | if (search_regs.start[c - '0'] >= 1) | |
9a76659d JB |
1927 | Finsert_buffer_substring |
1928 | (Fcurrent_buffer (), | |
1929 | make_number (search_regs.start[c - '0'] + offset), | |
1930 | make_number (search_regs.end[c - '0'] + offset)); | |
ca1d1d23 JB |
1931 | } |
1932 | else | |
1933 | insert_char (c); | |
1934 | } | |
1935 | else | |
1936 | insert_char (c); | |
1937 | } | |
1938 | UNGCPRO; | |
1939 | } | |
1940 | ||
6ec8bbd2 | 1941 | inslen = PT - (search_regs.start[sub]); |
d7a5ad5f | 1942 | del_range (search_regs.start[sub] + inslen, search_regs.end[sub] + inslen); |
ca1d1d23 JB |
1943 | |
1944 | if (case_action == all_caps) | |
6ec8bbd2 | 1945 | Fupcase_region (make_number (PT - inslen), make_number (PT)); |
ca1d1d23 | 1946 | else if (case_action == cap_initial) |
6ec8bbd2 | 1947 | Fupcase_initials_region (make_number (PT - inslen), make_number (PT)); |
ca1d1d23 JB |
1948 | return Qnil; |
1949 | } | |
1950 | \f | |
1951 | static Lisp_Object | |
1952 | match_limit (num, beginningp) | |
1953 | Lisp_Object num; | |
1954 | int beginningp; | |
1955 | { | |
1956 | register int n; | |
1957 | ||
1958 | CHECK_NUMBER (num, 0); | |
1959 | n = XINT (num); | |
4746118a JB |
1960 | if (n < 0 || n >= search_regs.num_regs) |
1961 | args_out_of_range (num, make_number (search_regs.num_regs)); | |
1962 | if (search_regs.num_regs <= 0 | |
1963 | || search_regs.start[n] < 0) | |
ca1d1d23 JB |
1964 | return Qnil; |
1965 | return (make_number ((beginningp) ? search_regs.start[n] | |
1966 | : search_regs.end[n])); | |
1967 | } | |
1968 | ||
1969 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
1970 | "Return position of start of text matched by last search.\n\ | |
5806161b EN |
1971 | SUBEXP, a number, specifies which parenthesized expression in the last\n\ |
1972 | regexp.\n\ | |
1973 | Value is nil if SUBEXPth pair didn't match, or there were less than\n\ | |
1974 | SUBEXP pairs.\n\ | |
ca1d1d23 | 1975 | Zero means the entire text matched by the whole regexp or whole string.") |
5806161b EN |
1976 | (subexp) |
1977 | Lisp_Object subexp; | |
ca1d1d23 | 1978 | { |
5806161b | 1979 | return match_limit (subexp, 1); |
ca1d1d23 JB |
1980 | } |
1981 | ||
1982 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
1983 | "Return position of end of text matched by last search.\n\ | |
5806161b EN |
1984 | SUBEXP, a number, specifies which parenthesized expression in the last\n\ |
1985 | regexp.\n\ | |
1986 | Value is nil if SUBEXPth pair didn't match, or there were less than\n\ | |
1987 | SUBEXP pairs.\n\ | |
ca1d1d23 | 1988 | Zero means the entire text matched by the whole regexp or whole string.") |
5806161b EN |
1989 | (subexp) |
1990 | Lisp_Object subexp; | |
ca1d1d23 | 1991 | { |
5806161b | 1992 | return match_limit (subexp, 0); |
ca1d1d23 JB |
1993 | } |
1994 | ||
56256c2a | 1995 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 2, 0, |
ca1d1d23 JB |
1996 | "Return a list containing all info on what the last search matched.\n\ |
1997 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'.\n\ | |
1998 | All the elements are markers or nil (nil if the Nth pair didn't match)\n\ | |
1999 | if the last match was on a buffer; integers or nil if a string was matched.\n\ | |
56256c2a RS |
2000 | Use `store-match-data' to reinstate the data in this list.\n\ |
2001 | \n\ | |
2002 | If INTEGERS (the optional first argument) is non-nil, always use integers\n\ | |
8ca821e9 | 2003 | \(rather than markers) to represent buffer positions.\n\ |
56256c2a RS |
2004 | If REUSE is a list, reuse it as part of the value. If REUSE is long enough\n\ |
2005 | to hold all the values, and if INTEGERS is non-nil, no consing is done.") | |
2006 | (integers, reuse) | |
2007 | Lisp_Object integers, reuse; | |
ca1d1d23 | 2008 | { |
56256c2a | 2009 | Lisp_Object tail, prev; |
4746118a | 2010 | Lisp_Object *data; |
ca1d1d23 JB |
2011 | int i, len; |
2012 | ||
daa37602 | 2013 | if (NILP (last_thing_searched)) |
c36bcf1b | 2014 | return Qnil; |
daa37602 | 2015 | |
4746118a JB |
2016 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs) |
2017 | * sizeof (Lisp_Object)); | |
2018 | ||
ca1d1d23 | 2019 | len = -1; |
4746118a | 2020 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2021 | { |
2022 | int start = search_regs.start[i]; | |
2023 | if (start >= 0) | |
2024 | { | |
56256c2a RS |
2025 | if (EQ (last_thing_searched, Qt) |
2026 | || ! NILP (integers)) | |
ca1d1d23 | 2027 | { |
c235cce7 KH |
2028 | XSETFASTINT (data[2 * i], start); |
2029 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2030 | } |
0ed62dc7 | 2031 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2032 | { |
2033 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2034 | Fset_marker (data[2 * i], |
2035 | make_number (start), | |
2036 | last_thing_searched); | |
ca1d1d23 JB |
2037 | data[2 * i + 1] = Fmake_marker (); |
2038 | Fset_marker (data[2 * i + 1], | |
daa37602 JB |
2039 | make_number (search_regs.end[i]), |
2040 | last_thing_searched); | |
ca1d1d23 | 2041 | } |
daa37602 JB |
2042 | else |
2043 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2044 | abort (); | |
2045 | ||
ca1d1d23 JB |
2046 | len = i; |
2047 | } | |
2048 | else | |
2049 | data[2 * i] = data [2 * i + 1] = Qnil; | |
2050 | } | |
56256c2a RS |
2051 | |
2052 | /* If REUSE is not usable, cons up the values and return them. */ | |
2053 | if (! CONSP (reuse)) | |
2054 | return Flist (2 * len + 2, data); | |
2055 | ||
2056 | /* If REUSE is a list, store as many value elements as will fit | |
2057 | into the elements of REUSE. */ | |
2058 | for (i = 0, tail = reuse; CONSP (tail); | |
2059 | i++, tail = XCONS (tail)->cdr) | |
2060 | { | |
2061 | if (i < 2 * len + 2) | |
2062 | XCONS (tail)->car = data[i]; | |
2063 | else | |
2064 | XCONS (tail)->car = Qnil; | |
2065 | prev = tail; | |
2066 | } | |
2067 | ||
2068 | /* If we couldn't fit all value elements into REUSE, | |
2069 | cons up the rest of them and add them to the end of REUSE. */ | |
2070 | if (i < 2 * len + 2) | |
2071 | XCONS (prev)->cdr = Flist (2 * len + 2 - i, data + i); | |
2072 | ||
2073 | return reuse; | |
ca1d1d23 JB |
2074 | } |
2075 | ||
2076 | ||
2077 | DEFUN ("store-match-data", Fstore_match_data, Sstore_match_data, 1, 1, 0, | |
2078 | "Set internal data on last search match from elements of LIST.\n\ | |
2079 | LIST should have been created by calling `match-data' previously.") | |
2080 | (list) | |
2081 | register Lisp_Object list; | |
2082 | { | |
2083 | register int i; | |
2084 | register Lisp_Object marker; | |
2085 | ||
7074fde6 FP |
2086 | if (running_asynch_code) |
2087 | save_search_regs (); | |
2088 | ||
ca1d1d23 | 2089 | if (!CONSP (list) && !NILP (list)) |
b37902c8 | 2090 | list = wrong_type_argument (Qconsp, list); |
ca1d1d23 | 2091 | |
daa37602 JB |
2092 | /* Unless we find a marker with a buffer in LIST, assume that this |
2093 | match data came from a string. */ | |
2094 | last_thing_searched = Qt; | |
2095 | ||
4746118a JB |
2096 | /* Allocate registers if they don't already exist. */ |
2097 | { | |
d084e942 | 2098 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2099 | |
2100 | if (length > search_regs.num_regs) | |
2101 | { | |
1113d9db JB |
2102 | if (search_regs.num_regs == 0) |
2103 | { | |
2104 | search_regs.start | |
2105 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2106 | search_regs.end | |
2107 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2108 | } | |
4746118a | 2109 | else |
1113d9db JB |
2110 | { |
2111 | search_regs.start | |
2112 | = (regoff_t *) xrealloc (search_regs.start, | |
2113 | length * sizeof (regoff_t)); | |
2114 | search_regs.end | |
2115 | = (regoff_t *) xrealloc (search_regs.end, | |
2116 | length * sizeof (regoff_t)); | |
2117 | } | |
4746118a | 2118 | |
487282dc | 2119 | search_regs.num_regs = length; |
4746118a JB |
2120 | } |
2121 | } | |
2122 | ||
2123 | for (i = 0; i < search_regs.num_regs; i++) | |
ca1d1d23 JB |
2124 | { |
2125 | marker = Fcar (list); | |
2126 | if (NILP (marker)) | |
2127 | { | |
2128 | search_regs.start[i] = -1; | |
2129 | list = Fcdr (list); | |
2130 | } | |
2131 | else | |
2132 | { | |
0ed62dc7 | 2133 | if (MARKERP (marker)) |
daa37602 JB |
2134 | { |
2135 | if (XMARKER (marker)->buffer == 0) | |
c235cce7 | 2136 | XSETFASTINT (marker, 0); |
daa37602 | 2137 | else |
a3668d92 | 2138 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); |
daa37602 | 2139 | } |
ca1d1d23 JB |
2140 | |
2141 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
2142 | search_regs.start[i] = XINT (marker); | |
2143 | list = Fcdr (list); | |
2144 | ||
2145 | marker = Fcar (list); | |
0ed62dc7 | 2146 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
c235cce7 | 2147 | XSETFASTINT (marker, 0); |
ca1d1d23 JB |
2148 | |
2149 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
2150 | search_regs.end[i] = XINT (marker); | |
2151 | } | |
2152 | list = Fcdr (list); | |
2153 | } | |
2154 | ||
2155 | return Qnil; | |
2156 | } | |
2157 | ||
7074fde6 FP |
2158 | /* If non-zero the match data have been saved in saved_search_regs |
2159 | during the execution of a sentinel or filter. */ | |
75ebf74b | 2160 | static int search_regs_saved; |
7074fde6 FP |
2161 | static struct re_registers saved_search_regs; |
2162 | ||
2163 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
2164 | if asynchronous code (filter or sentinel) is running. */ | |
2165 | static void | |
2166 | save_search_regs () | |
2167 | { | |
2168 | if (!search_regs_saved) | |
2169 | { | |
2170 | saved_search_regs.num_regs = search_regs.num_regs; | |
2171 | saved_search_regs.start = search_regs.start; | |
2172 | saved_search_regs.end = search_regs.end; | |
2173 | search_regs.num_regs = 0; | |
2d4a771a RS |
2174 | search_regs.start = 0; |
2175 | search_regs.end = 0; | |
7074fde6 FP |
2176 | |
2177 | search_regs_saved = 1; | |
2178 | } | |
2179 | } | |
2180 | ||
2181 | /* Called upon exit from filters and sentinels. */ | |
2182 | void | |
2183 | restore_match_data () | |
2184 | { | |
2185 | if (search_regs_saved) | |
2186 | { | |
2187 | if (search_regs.num_regs > 0) | |
2188 | { | |
2189 | xfree (search_regs.start); | |
2190 | xfree (search_regs.end); | |
2191 | } | |
2192 | search_regs.num_regs = saved_search_regs.num_regs; | |
2193 | search_regs.start = saved_search_regs.start; | |
2194 | search_regs.end = saved_search_regs.end; | |
2195 | ||
2196 | search_regs_saved = 0; | |
2197 | } | |
2198 | } | |
2199 | ||
ca1d1d23 JB |
2200 | /* Quote a string to inactivate reg-expr chars */ |
2201 | ||
2202 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
2203 | "Return a regexp string which matches exactly STRING and nothing else.") | |
5806161b EN |
2204 | (string) |
2205 | Lisp_Object string; | |
ca1d1d23 JB |
2206 | { |
2207 | register unsigned char *in, *out, *end; | |
2208 | register unsigned char *temp; | |
2209 | ||
5806161b | 2210 | CHECK_STRING (string, 0); |
ca1d1d23 | 2211 | |
5806161b | 2212 | temp = (unsigned char *) alloca (XSTRING (string)->size * 2); |
ca1d1d23 JB |
2213 | |
2214 | /* Now copy the data into the new string, inserting escapes. */ | |
2215 | ||
5806161b EN |
2216 | in = XSTRING (string)->data; |
2217 | end = in + XSTRING (string)->size; | |
ca1d1d23 JB |
2218 | out = temp; |
2219 | ||
2220 | for (; in != end; in++) | |
2221 | { | |
2222 | if (*in == '[' || *in == ']' | |
2223 | || *in == '*' || *in == '.' || *in == '\\' | |
2224 | || *in == '?' || *in == '+' | |
2225 | || *in == '^' || *in == '$') | |
2226 | *out++ = '\\'; | |
2227 | *out++ = *in; | |
2228 | } | |
2229 | ||
2230 | return make_string (temp, out - temp); | |
2231 | } | |
2232 | \f | |
2233 | syms_of_search () | |
2234 | { | |
2235 | register int i; | |
2236 | ||
487282dc KH |
2237 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
2238 | { | |
2239 | searchbufs[i].buf.allocated = 100; | |
2240 | searchbufs[i].buf.buffer = (unsigned char *) malloc (100); | |
2241 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; | |
2242 | searchbufs[i].regexp = Qnil; | |
2243 | staticpro (&searchbufs[i].regexp); | |
2244 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); | |
2245 | } | |
2246 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
2247 | |
2248 | Qsearch_failed = intern ("search-failed"); | |
2249 | staticpro (&Qsearch_failed); | |
2250 | Qinvalid_regexp = intern ("invalid-regexp"); | |
2251 | staticpro (&Qinvalid_regexp); | |
2252 | ||
2253 | Fput (Qsearch_failed, Qerror_conditions, | |
2254 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
2255 | Fput (Qsearch_failed, Qerror_message, | |
2256 | build_string ("Search failed")); | |
2257 | ||
2258 | Fput (Qinvalid_regexp, Qerror_conditions, | |
2259 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
2260 | Fput (Qinvalid_regexp, Qerror_message, | |
2261 | build_string ("Invalid regexp")); | |
2262 | ||
daa37602 JB |
2263 | last_thing_searched = Qnil; |
2264 | staticpro (&last_thing_searched); | |
2265 | ||
ca1d1d23 | 2266 | defsubr (&Slooking_at); |
b819a390 RS |
2267 | defsubr (&Sposix_looking_at); |
2268 | defsubr (&Sstring_match); | |
2269 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
2270 | defsubr (&Sskip_chars_forward); |
2271 | defsubr (&Sskip_chars_backward); | |
17431c60 RS |
2272 | defsubr (&Sskip_syntax_forward); |
2273 | defsubr (&Sskip_syntax_backward); | |
ca1d1d23 JB |
2274 | defsubr (&Ssearch_forward); |
2275 | defsubr (&Ssearch_backward); | |
2276 | defsubr (&Sword_search_forward); | |
2277 | defsubr (&Sword_search_backward); | |
2278 | defsubr (&Sre_search_forward); | |
2279 | defsubr (&Sre_search_backward); | |
b819a390 RS |
2280 | defsubr (&Sposix_search_forward); |
2281 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
2282 | defsubr (&Sreplace_match); |
2283 | defsubr (&Smatch_beginning); | |
2284 | defsubr (&Smatch_end); | |
2285 | defsubr (&Smatch_data); | |
2286 | defsubr (&Sstore_match_data); | |
2287 | defsubr (&Sregexp_quote); | |
2288 | } |