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 | 123 | { |
d451e4db | 124 | 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)); | |
d451e4db RS |
134 | val = (char *) re_compile_pattern ((char *) XSTRING (pattern)->data, |
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) | |
b0eba991 | 228 | ? XCHAR_TABLE (DOWNCASE_TABLE)->contents : 0), |
b819a390 | 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 | } | |
8bb43c28 RS |
252 | |
253 | re_match_object = Qnil; | |
ca1d1d23 | 254 | |
487282dc | 255 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
6ec8bbd2 | 256 | PT - BEGV, &search_regs, |
ca1d1d23 JB |
257 | ZV - BEGV); |
258 | if (i == -2) | |
259 | matcher_overflow (); | |
260 | ||
261 | val = (0 <= i ? Qt : Qnil); | |
4746118a | 262 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
263 | if (search_regs.start[i] >= 0) |
264 | { | |
265 | search_regs.start[i] += BEGV; | |
266 | search_regs.end[i] += BEGV; | |
267 | } | |
a3668d92 | 268 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
269 | immediate_quit = 0; |
270 | return val; | |
271 | } | |
272 | ||
b819a390 | 273 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
94f94972 | 274 | "Return t if text after point matches regular expression REGEXP.\n\ |
b819a390 RS |
275 | This function modifies the match data that `match-beginning',\n\ |
276 | `match-end' and `match-data' access; save and restore the match\n\ | |
277 | data if you want to preserve them.") | |
94f94972 RS |
278 | (regexp) |
279 | Lisp_Object regexp; | |
b819a390 | 280 | { |
94f94972 | 281 | return looking_at_1 (regexp, 0); |
b819a390 RS |
282 | } |
283 | ||
284 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
94f94972 | 285 | "Return t if text after point matches regular expression REGEXP.\n\ |
b819a390 RS |
286 | Find the longest match, in accord with Posix regular expression rules.\n\ |
287 | This function modifies the match data that `match-beginning',\n\ | |
288 | `match-end' and `match-data' access; save and restore the match\n\ | |
289 | data if you want to preserve them.") | |
94f94972 RS |
290 | (regexp) |
291 | Lisp_Object regexp; | |
b819a390 | 292 | { |
94f94972 | 293 | return looking_at_1 (regexp, 1); |
b819a390 RS |
294 | } |
295 | \f | |
296 | static Lisp_Object | |
297 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 298 | Lisp_Object regexp, string, start; |
b819a390 | 299 | int posix; |
ca1d1d23 JB |
300 | { |
301 | int val; | |
302 | int s; | |
487282dc | 303 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 304 | |
7074fde6 FP |
305 | if (running_asynch_code) |
306 | save_search_regs (); | |
307 | ||
ca1d1d23 JB |
308 | CHECK_STRING (regexp, 0); |
309 | CHECK_STRING (string, 1); | |
310 | ||
311 | if (NILP (start)) | |
312 | s = 0; | |
313 | else | |
314 | { | |
315 | int len = XSTRING (string)->size; | |
316 | ||
317 | CHECK_NUMBER (start, 2); | |
318 | s = XINT (start); | |
319 | if (s < 0 && -s <= len) | |
26faf9f4 | 320 | s = len + s; |
ca1d1d23 JB |
321 | else if (0 > s || s > len) |
322 | args_out_of_range (string, start); | |
323 | } | |
324 | ||
487282dc KH |
325 | bufp = compile_pattern (regexp, &search_regs, |
326 | (!NILP (current_buffer->case_fold_search) | |
b0eba991 | 327 | ? XCHAR_TABLE (DOWNCASE_TABLE)->contents : 0), |
24b704fa | 328 | posix); |
ca1d1d23 | 329 | immediate_quit = 1; |
8bb43c28 RS |
330 | re_match_object = string; |
331 | ||
487282dc | 332 | val = re_search (bufp, (char *) XSTRING (string)->data, |
ca1d1d23 JB |
333 | XSTRING (string)->size, s, XSTRING (string)->size - s, |
334 | &search_regs); | |
335 | immediate_quit = 0; | |
daa37602 | 336 | last_thing_searched = Qt; |
ca1d1d23 JB |
337 | if (val == -2) |
338 | matcher_overflow (); | |
339 | if (val < 0) return Qnil; | |
340 | return make_number (val); | |
341 | } | |
e59a8453 | 342 | |
b819a390 RS |
343 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
344 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
345 | If third arg START is non-nil, start search at that index in STRING.\n\ | |
346 | For index of first char beyond the match, do (match-end 0).\n\ | |
347 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
348 | matched by parenthesis constructs in the pattern.") | |
349 | (regexp, string, start) | |
350 | Lisp_Object regexp, string, start; | |
351 | { | |
352 | return string_match_1 (regexp, string, start, 0); | |
353 | } | |
354 | ||
355 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
356 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
357 | Find the longest match, in accord with Posix regular expression rules.\n\ | |
358 | If third arg START is non-nil, start search at that index in STRING.\n\ | |
359 | For index of first char beyond the match, do (match-end 0).\n\ | |
360 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
361 | matched by parenthesis constructs in the pattern.") | |
362 | (regexp, string, start) | |
363 | Lisp_Object regexp, string, start; | |
364 | { | |
365 | return string_match_1 (regexp, string, start, 1); | |
366 | } | |
367 | ||
e59a8453 RS |
368 | /* Match REGEXP against STRING, searching all of STRING, |
369 | and return the index of the match, or negative on failure. | |
370 | This does not clobber the match data. */ | |
371 | ||
372 | int | |
373 | fast_string_match (regexp, string) | |
374 | Lisp_Object regexp, string; | |
375 | { | |
376 | int val; | |
487282dc | 377 | struct re_pattern_buffer *bufp; |
e59a8453 | 378 | |
b819a390 | 379 | bufp = compile_pattern (regexp, 0, 0, 0); |
e59a8453 | 380 | immediate_quit = 1; |
8bb43c28 RS |
381 | re_match_object = string; |
382 | ||
487282dc | 383 | val = re_search (bufp, (char *) XSTRING (string)->data, |
e59a8453 RS |
384 | XSTRING (string)->size, 0, XSTRING (string)->size, |
385 | 0); | |
386 | immediate_quit = 0; | |
387 | return val; | |
388 | } | |
5679531d KH |
389 | |
390 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
391 | and return the index of the match, or negative on failure. | |
392 | This does not clobber the match data. */ | |
393 | ||
394 | extern Lisp_Object Vascii_downcase_table; | |
395 | ||
396 | int | |
b4577c63 | 397 | fast_c_string_match_ignore_case (regexp, string) |
5679531d KH |
398 | Lisp_Object regexp; |
399 | char *string; | |
400 | { | |
401 | int val; | |
402 | struct re_pattern_buffer *bufp; | |
403 | int len = strlen (string); | |
404 | ||
b4577c63 | 405 | re_match_object = Qt; |
5679531d KH |
406 | bufp = compile_pattern (regexp, 0, |
407 | XCHAR_TABLE (Vascii_downcase_table)->contents, 0); | |
408 | immediate_quit = 1; | |
409 | val = re_search (bufp, string, len, 0, len, 0); | |
410 | immediate_quit = 0; | |
411 | return val; | |
412 | } | |
ca1d1d23 | 413 | \f |
9169c321 JB |
414 | /* max and min. */ |
415 | ||
416 | static int | |
417 | max (a, b) | |
418 | int a, b; | |
419 | { | |
420 | return ((a > b) ? a : b); | |
421 | } | |
422 | ||
423 | static int | |
424 | min (a, b) | |
425 | int a, b; | |
426 | { | |
427 | return ((a < b) ? a : b); | |
428 | } | |
429 | ||
430 | \f | |
431 | /* The newline cache: remembering which sections of text have no newlines. */ | |
432 | ||
433 | /* If the user has requested newline caching, make sure it's on. | |
434 | Otherwise, make sure it's off. | |
435 | This is our cheezy way of associating an action with the change of | |
436 | state of a buffer-local variable. */ | |
437 | static void | |
438 | newline_cache_on_off (buf) | |
439 | struct buffer *buf; | |
440 | { | |
441 | if (NILP (buf->cache_long_line_scans)) | |
442 | { | |
443 | /* It should be off. */ | |
444 | if (buf->newline_cache) | |
445 | { | |
446 | free_region_cache (buf->newline_cache); | |
447 | buf->newline_cache = 0; | |
448 | } | |
449 | } | |
450 | else | |
451 | { | |
452 | /* It should be on. */ | |
453 | if (buf->newline_cache == 0) | |
454 | buf->newline_cache = new_region_cache (); | |
455 | } | |
456 | } | |
457 | ||
458 | \f | |
459 | /* Search for COUNT instances of the character TARGET between START and END. | |
460 | ||
461 | If COUNT is positive, search forwards; END must be >= START. | |
462 | If COUNT is negative, search backwards for the -COUNTth instance; | |
463 | END must be <= START. | |
464 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
465 | ||
466 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
467 | direction indicated by COUNT. | |
ffd56f97 JB |
468 | |
469 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
5bfe95c9 RS |
470 | position after the COUNTth match. Note that for reverse motion |
471 | this is not the same as the usual convention for Emacs motion commands. | |
ffd56f97 | 472 | |
9169c321 JB |
473 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
474 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 475 | |
087a5f81 RS |
476 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
477 | except when inside redisplay. */ | |
478 | ||
9169c321 JB |
479 | scan_buffer (target, start, end, count, shortage, allow_quit) |
480 | register int target; | |
481 | int start, end; | |
482 | int count; | |
483 | int *shortage; | |
087a5f81 | 484 | int allow_quit; |
ca1d1d23 | 485 | { |
9169c321 JB |
486 | struct region_cache *newline_cache; |
487 | int direction; | |
ffd56f97 | 488 | |
9169c321 JB |
489 | if (count > 0) |
490 | { | |
491 | direction = 1; | |
492 | if (! end) end = ZV; | |
493 | } | |
494 | else | |
495 | { | |
496 | direction = -1; | |
497 | if (! end) end = BEGV; | |
498 | } | |
ffd56f97 | 499 | |
9169c321 JB |
500 | newline_cache_on_off (current_buffer); |
501 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
502 | |
503 | if (shortage != 0) | |
504 | *shortage = 0; | |
505 | ||
087a5f81 | 506 | immediate_quit = allow_quit; |
ca1d1d23 | 507 | |
ffd56f97 | 508 | if (count > 0) |
9169c321 | 509 | while (start != end) |
ca1d1d23 | 510 | { |
9169c321 JB |
511 | /* Our innermost scanning loop is very simple; it doesn't know |
512 | about gaps, buffer ends, or the newline cache. ceiling is | |
513 | the position of the last character before the next such | |
514 | obstacle --- the last character the dumb search loop should | |
515 | examine. */ | |
516 | register int ceiling = end - 1; | |
517 | ||
518 | /* If we're looking for a newline, consult the newline cache | |
519 | to see where we can avoid some scanning. */ | |
520 | if (target == '\n' && newline_cache) | |
521 | { | |
522 | int next_change; | |
523 | immediate_quit = 0; | |
524 | while (region_cache_forward | |
525 | (current_buffer, newline_cache, start, &next_change)) | |
526 | start = next_change; | |
cbe0db0d | 527 | immediate_quit = allow_quit; |
9169c321 JB |
528 | |
529 | /* start should never be after end. */ | |
530 | if (start >= end) | |
531 | start = end - 1; | |
532 | ||
533 | /* Now the text after start is an unknown region, and | |
534 | next_change is the position of the next known region. */ | |
535 | ceiling = min (next_change - 1, ceiling); | |
536 | } | |
537 | ||
538 | /* The dumb loop can only scan text stored in contiguous | |
539 | bytes. BUFFER_CEILING_OF returns the last character | |
540 | position that is contiguous, so the ceiling is the | |
541 | position after that. */ | |
542 | ceiling = min (BUFFER_CEILING_OF (start), ceiling); | |
543 | ||
544 | { | |
545 | /* The termination address of the dumb loop. */ | |
5679531d KH |
546 | register unsigned char *ceiling_addr = POS_ADDR (ceiling) + 1; |
547 | register unsigned char *cursor = POS_ADDR (start); | |
9169c321 JB |
548 | unsigned char *base = cursor; |
549 | ||
550 | while (cursor < ceiling_addr) | |
551 | { | |
552 | unsigned char *scan_start = cursor; | |
553 | ||
554 | /* The dumb loop. */ | |
555 | while (*cursor != target && ++cursor < ceiling_addr) | |
556 | ; | |
557 | ||
558 | /* If we're looking for newlines, cache the fact that | |
559 | the region from start to cursor is free of them. */ | |
560 | if (target == '\n' && newline_cache) | |
561 | know_region_cache (current_buffer, newline_cache, | |
562 | start + scan_start - base, | |
563 | start + cursor - base); | |
564 | ||
565 | /* Did we find the target character? */ | |
566 | if (cursor < ceiling_addr) | |
567 | { | |
568 | if (--count == 0) | |
569 | { | |
570 | immediate_quit = 0; | |
571 | return (start + cursor - base + 1); | |
572 | } | |
573 | cursor++; | |
574 | } | |
575 | } | |
576 | ||
577 | start += cursor - base; | |
578 | } | |
ca1d1d23 JB |
579 | } |
580 | else | |
9169c321 JB |
581 | while (start > end) |
582 | { | |
583 | /* The last character to check before the next obstacle. */ | |
584 | register int ceiling = end; | |
585 | ||
586 | /* Consult the newline cache, if appropriate. */ | |
587 | if (target == '\n' && newline_cache) | |
588 | { | |
589 | int next_change; | |
590 | immediate_quit = 0; | |
591 | while (region_cache_backward | |
592 | (current_buffer, newline_cache, start, &next_change)) | |
593 | start = next_change; | |
cbe0db0d | 594 | immediate_quit = allow_quit; |
9169c321 JB |
595 | |
596 | /* Start should never be at or before end. */ | |
597 | if (start <= end) | |
598 | start = end + 1; | |
599 | ||
600 | /* Now the text before start is an unknown region, and | |
601 | next_change is the position of the next known region. */ | |
602 | ceiling = max (next_change, ceiling); | |
603 | } | |
604 | ||
605 | /* Stop scanning before the gap. */ | |
606 | ceiling = max (BUFFER_FLOOR_OF (start - 1), ceiling); | |
607 | ||
608 | { | |
609 | /* The termination address of the dumb loop. */ | |
5679531d KH |
610 | register unsigned char *ceiling_addr = POS_ADDR (ceiling); |
611 | register unsigned char *cursor = POS_ADDR (start - 1); | |
9169c321 JB |
612 | unsigned char *base = cursor; |
613 | ||
614 | while (cursor >= ceiling_addr) | |
615 | { | |
616 | unsigned char *scan_start = cursor; | |
617 | ||
618 | while (*cursor != target && --cursor >= ceiling_addr) | |
619 | ; | |
620 | ||
621 | /* If we're looking for newlines, cache the fact that | |
622 | the region from after the cursor to start is free of them. */ | |
623 | if (target == '\n' && newline_cache) | |
624 | know_region_cache (current_buffer, newline_cache, | |
625 | start + cursor - base, | |
626 | start + scan_start - base); | |
627 | ||
628 | /* Did we find the target character? */ | |
629 | if (cursor >= ceiling_addr) | |
630 | { | |
631 | if (++count >= 0) | |
632 | { | |
633 | immediate_quit = 0; | |
634 | return (start + cursor - base); | |
635 | } | |
636 | cursor--; | |
637 | } | |
638 | } | |
639 | ||
640 | start += cursor - base; | |
641 | } | |
642 | } | |
643 | ||
ca1d1d23 JB |
644 | immediate_quit = 0; |
645 | if (shortage != 0) | |
ffd56f97 | 646 | *shortage = count * direction; |
9169c321 | 647 | return start; |
ca1d1d23 JB |
648 | } |
649 | ||
63fa018d RS |
650 | int |
651 | find_next_newline_no_quit (from, cnt) | |
652 | register int from, cnt; | |
653 | { | |
9169c321 | 654 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
63fa018d RS |
655 | } |
656 | ||
ca1d1d23 JB |
657 | int |
658 | find_next_newline (from, cnt) | |
659 | register int from, cnt; | |
660 | { | |
9169c321 JB |
661 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 1); |
662 | } | |
663 | ||
664 | ||
665 | /* Like find_next_newline, but returns position before the newline, | |
666 | not after, and only search up to TO. This isn't just | |
667 | find_next_newline (...)-1, because you might hit TO. */ | |
668 | int | |
669 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 670 | int from, to, cnt; |
9169c321 JB |
671 | { |
672 | int shortage; | |
673 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
674 | ||
675 | if (shortage == 0) | |
676 | pos--; | |
677 | ||
678 | return pos; | |
ca1d1d23 JB |
679 | } |
680 | \f | |
ca1d1d23 JB |
681 | /* Subroutines of Lisp buffer search functions. */ |
682 | ||
683 | static Lisp_Object | |
b819a390 | 684 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
685 | Lisp_Object string, bound, noerror, count; |
686 | int direction; | |
687 | int RE; | |
b819a390 | 688 | int posix; |
ca1d1d23 JB |
689 | { |
690 | register int np; | |
691 | int lim; | |
692 | int n = direction; | |
693 | ||
694 | if (!NILP (count)) | |
695 | { | |
696 | CHECK_NUMBER (count, 3); | |
697 | n *= XINT (count); | |
698 | } | |
699 | ||
700 | CHECK_STRING (string, 0); | |
701 | if (NILP (bound)) | |
702 | lim = n > 0 ? ZV : BEGV; | |
703 | else | |
704 | { | |
705 | CHECK_NUMBER_COERCE_MARKER (bound, 1); | |
706 | lim = XINT (bound); | |
6ec8bbd2 | 707 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
708 | error ("Invalid search bound (wrong side of point)"); |
709 | if (lim > ZV) | |
710 | lim = ZV; | |
711 | if (lim < BEGV) | |
712 | lim = BEGV; | |
713 | } | |
714 | ||
6ec8bbd2 | 715 | np = search_buffer (string, PT, lim, n, RE, |
ca1d1d23 | 716 | (!NILP (current_buffer->case_fold_search) |
b1428bd8 RS |
717 | ? XCHAR_TABLE (current_buffer->case_canon_table)->contents |
718 | : 0), | |
ca1d1d23 | 719 | (!NILP (current_buffer->case_fold_search) |
b1428bd8 RS |
720 | ? XCHAR_TABLE (current_buffer->case_eqv_table)->contents |
721 | : 0), | |
b819a390 | 722 | posix); |
ca1d1d23 JB |
723 | if (np <= 0) |
724 | { | |
725 | if (NILP (noerror)) | |
726 | return signal_failure (string); | |
727 | if (!EQ (noerror, Qt)) | |
728 | { | |
729 | if (lim < BEGV || lim > ZV) | |
730 | abort (); | |
a5f217b8 RS |
731 | SET_PT (lim); |
732 | return Qnil; | |
733 | #if 0 /* This would be clean, but maybe programs depend on | |
734 | a value of nil here. */ | |
481399bf | 735 | np = lim; |
a5f217b8 | 736 | #endif |
ca1d1d23 | 737 | } |
481399bf RS |
738 | else |
739 | return Qnil; | |
ca1d1d23 JB |
740 | } |
741 | ||
742 | if (np < BEGV || np > ZV) | |
743 | abort (); | |
744 | ||
745 | SET_PT (np); | |
746 | ||
747 | return make_number (np); | |
748 | } | |
749 | \f | |
b6d6a51c KH |
750 | static int |
751 | trivial_regexp_p (regexp) | |
752 | Lisp_Object regexp; | |
753 | { | |
754 | int len = XSTRING (regexp)->size; | |
755 | unsigned char *s = XSTRING (regexp)->data; | |
756 | unsigned char c; | |
757 | while (--len >= 0) | |
758 | { | |
759 | switch (*s++) | |
760 | { | |
761 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
762 | return 0; | |
763 | case '\\': | |
764 | if (--len < 0) | |
765 | return 0; | |
766 | switch (*s++) | |
767 | { | |
768 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
769 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
866f60fd | 770 | case 'S': case '=': |
5679531d | 771 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 772 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
773 | case '6': case '7': case '8': case '9': |
774 | return 0; | |
775 | } | |
776 | } | |
777 | } | |
778 | return 1; | |
779 | } | |
780 | ||
ca325161 | 781 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 782 | starting at position POS and stopping at position LIM, |
b819a390 | 783 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
784 | a regular expression if RE is true. |
785 | ||
786 | If N is positive, searching is forward and LIM must be greater than POS. | |
787 | If N is negative, searching is backward and LIM must be less than POS. | |
788 | ||
789 | Returns -x if only N-x occurrences found (x > 0), | |
790 | or else the position at the beginning of the Nth occurrence | |
b819a390 RS |
791 | (if searching backward) or the end (if searching forward). |
792 | ||
793 | POSIX is nonzero if we want full backtracking (POSIX style) | |
794 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 795 | |
b819a390 RS |
796 | static int |
797 | search_buffer (string, pos, lim, n, RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
798 | Lisp_Object string; |
799 | int pos; | |
800 | int lim; | |
801 | int n; | |
802 | int RE; | |
b1428bd8 RS |
803 | Lisp_Object *trt; |
804 | Lisp_Object *inverse_trt; | |
b819a390 | 805 | int posix; |
ca1d1d23 JB |
806 | { |
807 | int len = XSTRING (string)->size; | |
808 | unsigned char *base_pat = XSTRING (string)->data; | |
809 | register int *BM_tab; | |
810 | int *BM_tab_base; | |
811 | register int direction = ((n > 0) ? 1 : -1); | |
812 | register int dirlen; | |
813 | int infinity, limit, k, stride_for_teases; | |
814 | register unsigned char *pat, *cursor, *p_limit; | |
815 | register int i, j; | |
816 | unsigned char *p1, *p2; | |
817 | int s1, s2; | |
818 | ||
7074fde6 FP |
819 | if (running_asynch_code) |
820 | save_search_regs (); | |
821 | ||
ca1d1d23 | 822 | /* Null string is found at starting position. */ |
3f57a499 | 823 | if (len == 0) |
ca325161 RS |
824 | { |
825 | set_search_regs (pos, 0); | |
826 | return pos; | |
827 | } | |
3f57a499 RS |
828 | |
829 | /* Searching 0 times means don't move. */ | |
830 | if (n == 0) | |
ca1d1d23 JB |
831 | return pos; |
832 | ||
b6d6a51c | 833 | if (RE && !trivial_regexp_p (string)) |
ca1d1d23 | 834 | { |
487282dc KH |
835 | struct re_pattern_buffer *bufp; |
836 | ||
b1428bd8 | 837 | bufp = compile_pattern (string, &search_regs, trt, posix); |
ca1d1d23 | 838 | |
ca1d1d23 JB |
839 | immediate_quit = 1; /* Quit immediately if user types ^G, |
840 | because letting this function finish | |
841 | can take too long. */ | |
842 | QUIT; /* Do a pending quit right away, | |
843 | to avoid paradoxical behavior */ | |
844 | /* Get pointers and sizes of the two strings | |
845 | that make up the visible portion of the buffer. */ | |
846 | ||
847 | p1 = BEGV_ADDR; | |
848 | s1 = GPT - BEGV; | |
849 | p2 = GAP_END_ADDR; | |
850 | s2 = ZV - GPT; | |
851 | if (s1 < 0) | |
852 | { | |
853 | p2 = p1; | |
854 | s2 = ZV - BEGV; | |
855 | s1 = 0; | |
856 | } | |
857 | if (s2 < 0) | |
858 | { | |
859 | s1 = ZV - BEGV; | |
860 | s2 = 0; | |
861 | } | |
8bb43c28 RS |
862 | re_match_object = Qnil; |
863 | ||
ca1d1d23 JB |
864 | while (n < 0) |
865 | { | |
42db823b | 866 | int val; |
487282dc | 867 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
42db823b RS |
868 | pos - BEGV, lim - pos, &search_regs, |
869 | /* Don't allow match past current point */ | |
870 | pos - BEGV); | |
ca1d1d23 | 871 | if (val == -2) |
b6d6a51c KH |
872 | { |
873 | matcher_overflow (); | |
874 | } | |
ca1d1d23 JB |
875 | if (val >= 0) |
876 | { | |
877 | j = BEGV; | |
4746118a | 878 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
879 | if (search_regs.start[i] >= 0) |
880 | { | |
881 | search_regs.start[i] += j; | |
882 | search_regs.end[i] += j; | |
883 | } | |
a3668d92 | 884 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
885 | /* Set pos to the new position. */ |
886 | pos = search_regs.start[0]; | |
887 | } | |
888 | else | |
889 | { | |
890 | immediate_quit = 0; | |
891 | return (n); | |
892 | } | |
893 | n++; | |
894 | } | |
895 | while (n > 0) | |
896 | { | |
42db823b | 897 | int val; |
487282dc | 898 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
42db823b RS |
899 | pos - BEGV, lim - pos, &search_regs, |
900 | lim - BEGV); | |
ca1d1d23 | 901 | if (val == -2) |
b6d6a51c KH |
902 | { |
903 | matcher_overflow (); | |
904 | } | |
ca1d1d23 JB |
905 | if (val >= 0) |
906 | { | |
907 | j = BEGV; | |
4746118a | 908 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
909 | if (search_regs.start[i] >= 0) |
910 | { | |
911 | search_regs.start[i] += j; | |
912 | search_regs.end[i] += j; | |
913 | } | |
a3668d92 | 914 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
915 | pos = search_regs.end[0]; |
916 | } | |
917 | else | |
918 | { | |
919 | immediate_quit = 0; | |
920 | return (0 - n); | |
921 | } | |
922 | n--; | |
923 | } | |
924 | immediate_quit = 0; | |
925 | return (pos); | |
926 | } | |
927 | else /* non-RE case */ | |
928 | { | |
929 | #ifdef C_ALLOCA | |
930 | int BM_tab_space[0400]; | |
931 | BM_tab = &BM_tab_space[0]; | |
932 | #else | |
933 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
934 | #endif | |
b6d6a51c KH |
935 | { |
936 | unsigned char *patbuf = (unsigned char *) alloca (len); | |
937 | pat = patbuf; | |
938 | while (--len >= 0) | |
939 | { | |
940 | /* If we got here and the RE flag is set, it's because we're | |
941 | dealing with a regexp known to be trivial, so the backslash | |
942 | just quotes the next character. */ | |
943 | if (RE && *base_pat == '\\') | |
944 | { | |
945 | len--; | |
946 | base_pat++; | |
947 | } | |
2fd7a4a4 | 948 | *pat++ = (trt ? XINT (trt[*base_pat++]) : *base_pat++); |
b6d6a51c KH |
949 | } |
950 | len = pat - patbuf; | |
951 | pat = base_pat = patbuf; | |
952 | } | |
ca1d1d23 JB |
953 | /* The general approach is that we are going to maintain that we know */ |
954 | /* the first (closest to the present position, in whatever direction */ | |
955 | /* we're searching) character that could possibly be the last */ | |
956 | /* (furthest from present position) character of a valid match. We */ | |
957 | /* advance the state of our knowledge by looking at that character */ | |
958 | /* and seeing whether it indeed matches the last character of the */ | |
959 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
960 | /* move our pointer (to putative last characters) as far as is */ | |
961 | /* logically possible. This amount of movement, which I call a */ | |
962 | /* stride, will be the length of the pattern if the actual character */ | |
963 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
964 | /* from the last occurrence of that character to the end of the */ | |
965 | /* pattern. */ | |
966 | /* As a coding trick, an enormous stride is coded into the table for */ | |
967 | /* characters that match the last character. This allows use of only */ | |
968 | /* a single test, a test for having gone past the end of the */ | |
969 | /* permissible match region, to test for both possible matches (when */ | |
970 | /* the stride goes past the end immediately) and failure to */ | |
971 | /* match (where you get nudged past the end one stride at a time). */ | |
972 | ||
973 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
974 | /* is determined only by the last character of the putative match. */ | |
975 | /* If that character does not match, we will stride the proper */ | |
976 | /* distance to propose a match that superimposes it on the last */ | |
977 | /* instance of a character that matches it (per trt), or misses */ | |
978 | /* it entirely if there is none. */ | |
979 | ||
980 | dirlen = len * direction; | |
981 | infinity = dirlen - (lim + pos + len + len) * direction; | |
982 | if (direction < 0) | |
983 | pat = (base_pat += len - 1); | |
984 | BM_tab_base = BM_tab; | |
985 | BM_tab += 0400; | |
986 | j = dirlen; /* to get it in a register */ | |
987 | /* A character that does not appear in the pattern induces a */ | |
988 | /* stride equal to the pattern length. */ | |
989 | while (BM_tab_base != BM_tab) | |
990 | { | |
991 | *--BM_tab = j; | |
992 | *--BM_tab = j; | |
993 | *--BM_tab = j; | |
994 | *--BM_tab = j; | |
995 | } | |
996 | i = 0; | |
997 | while (i != infinity) | |
998 | { | |
999 | j = pat[i]; i += direction; | |
1000 | if (i == dirlen) i = infinity; | |
8d505039 | 1001 | if (trt != 0) |
ca1d1d23 | 1002 | { |
2fd7a4a4 | 1003 | k = (j = XINT (trt[j])); |
ca1d1d23 JB |
1004 | if (i == infinity) |
1005 | stride_for_teases = BM_tab[j]; | |
1006 | BM_tab[j] = dirlen - i; | |
1007 | /* A translation table is accompanied by its inverse -- see */ | |
1008 | /* comment following downcase_table for details */ | |
2fd7a4a4 | 1009 | while ((j = (unsigned char) XINT (inverse_trt[j])) != k) |
ca1d1d23 JB |
1010 | BM_tab[j] = dirlen - i; |
1011 | } | |
1012 | else | |
1013 | { | |
1014 | if (i == infinity) | |
1015 | stride_for_teases = BM_tab[j]; | |
1016 | BM_tab[j] = dirlen - i; | |
1017 | } | |
1018 | /* stride_for_teases tells how much to stride if we get a */ | |
1019 | /* match on the far character but are subsequently */ | |
1020 | /* disappointed, by recording what the stride would have been */ | |
1021 | /* for that character if the last character had been */ | |
1022 | /* different. */ | |
1023 | } | |
1024 | infinity = dirlen - infinity; | |
1025 | pos += dirlen - ((direction > 0) ? direction : 0); | |
1026 | /* loop invariant - pos points at where last char (first char if reverse) | |
1027 | of pattern would align in a possible match. */ | |
1028 | while (n != 0) | |
1029 | { | |
b2c71fb4 KH |
1030 | /* It's been reported that some (broken) compiler thinks that |
1031 | Boolean expressions in an arithmetic context are unsigned. | |
1032 | Using an explicit ?1:0 prevents this. */ | |
1033 | if ((lim - pos - ((direction > 0) ? 1 : 0)) * direction < 0) | |
ca1d1d23 JB |
1034 | return (n * (0 - direction)); |
1035 | /* First we do the part we can by pointers (maybe nothing) */ | |
1036 | QUIT; | |
1037 | pat = base_pat; | |
1038 | limit = pos - dirlen + direction; | |
1039 | limit = ((direction > 0) | |
1040 | ? BUFFER_CEILING_OF (limit) | |
1041 | : BUFFER_FLOOR_OF (limit)); | |
1042 | /* LIMIT is now the last (not beyond-last!) value | |
1043 | POS can take on without hitting edge of buffer or the gap. */ | |
1044 | limit = ((direction > 0) | |
1045 | ? min (lim - 1, min (limit, pos + 20000)) | |
1046 | : max (lim, max (limit, pos - 20000))); | |
1047 | if ((limit - pos) * direction > 20) | |
1048 | { | |
5679531d KH |
1049 | p_limit = POS_ADDR (limit); |
1050 | p2 = (cursor = POS_ADDR (pos)); | |
ca1d1d23 JB |
1051 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ |
1052 | while (1) /* use one cursor setting as long as i can */ | |
1053 | { | |
1054 | if (direction > 0) /* worth duplicating */ | |
1055 | { | |
1056 | /* Use signed comparison if appropriate | |
1057 | to make cursor+infinity sure to be > p_limit. | |
1058 | Assuming that the buffer lies in a range of addresses | |
1059 | that are all "positive" (as ints) or all "negative", | |
1060 | either kind of comparison will work as long | |
1061 | as we don't step by infinity. So pick the kind | |
1062 | that works when we do step by infinity. */ | |
8d505039 | 1063 | if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit) |
9fa17f93 | 1064 | while ((EMACS_INT) cursor <= (EMACS_INT) p_limit) |
ca1d1d23 JB |
1065 | cursor += BM_tab[*cursor]; |
1066 | else | |
45b248b4 | 1067 | while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit) |
ca1d1d23 JB |
1068 | cursor += BM_tab[*cursor]; |
1069 | } | |
1070 | else | |
1071 | { | |
8d505039 RS |
1072 | if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit) |
1073 | while ((EMACS_INT) cursor >= (EMACS_INT) p_limit) | |
ca1d1d23 JB |
1074 | cursor += BM_tab[*cursor]; |
1075 | else | |
45b248b4 | 1076 | while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit) |
ca1d1d23 JB |
1077 | cursor += BM_tab[*cursor]; |
1078 | } | |
1079 | /* If you are here, cursor is beyond the end of the searched region. */ | |
1080 | /* This can happen if you match on the far character of the pattern, */ | |
1081 | /* because the "stride" of that character is infinity, a number able */ | |
1082 | /* to throw you well beyond the end of the search. It can also */ | |
1083 | /* happen if you fail to match within the permitted region and would */ | |
1084 | /* otherwise try a character beyond that region */ | |
1085 | if ((cursor - p_limit) * direction <= len) | |
1086 | break; /* a small overrun is genuine */ | |
1087 | cursor -= infinity; /* large overrun = hit */ | |
1088 | i = dirlen - direction; | |
8d505039 | 1089 | if (trt != 0) |
ca1d1d23 JB |
1090 | { |
1091 | while ((i -= direction) + direction != 0) | |
2fd7a4a4 | 1092 | if (pat[i] != XINT (trt[*(cursor -= direction)])) |
ca1d1d23 JB |
1093 | break; |
1094 | } | |
1095 | else | |
1096 | { | |
1097 | while ((i -= direction) + direction != 0) | |
1098 | if (pat[i] != *(cursor -= direction)) | |
1099 | break; | |
1100 | } | |
1101 | cursor += dirlen - i - direction; /* fix cursor */ | |
1102 | if (i + direction == 0) | |
1103 | { | |
1104 | cursor -= direction; | |
1113d9db | 1105 | |
ca325161 RS |
1106 | set_search_regs (pos + cursor - p2 + ((direction > 0) |
1107 | ? 1 - len : 0), | |
1108 | len); | |
1109 | ||
ca1d1d23 JB |
1110 | if ((n -= direction) != 0) |
1111 | cursor += dirlen; /* to resume search */ | |
1112 | else | |
1113 | return ((direction > 0) | |
1114 | ? search_regs.end[0] : search_regs.start[0]); | |
1115 | } | |
1116 | else | |
1117 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
1118 | } | |
1119 | pos += cursor - p2; | |
1120 | } | |
1121 | else | |
1122 | /* Now we'll pick up a clump that has to be done the hard */ | |
1123 | /* way because it covers a discontinuity */ | |
1124 | { | |
1125 | limit = ((direction > 0) | |
1126 | ? BUFFER_CEILING_OF (pos - dirlen + 1) | |
1127 | : BUFFER_FLOOR_OF (pos - dirlen - 1)); | |
1128 | limit = ((direction > 0) | |
1129 | ? min (limit + len, lim - 1) | |
1130 | : max (limit - len, lim)); | |
1131 | /* LIMIT is now the last value POS can have | |
1132 | and still be valid for a possible match. */ | |
1133 | while (1) | |
1134 | { | |
1135 | /* This loop can be coded for space rather than */ | |
1136 | /* speed because it will usually run only once. */ | |
1137 | /* (the reach is at most len + 21, and typically */ | |
1138 | /* does not exceed len) */ | |
1139 | while ((limit - pos) * direction >= 0) | |
5679531d | 1140 | pos += BM_tab[FETCH_BYTE (pos)]; |
ca1d1d23 | 1141 | /* now run the same tests to distinguish going off the */ |
eb8c3be9 | 1142 | /* end, a match or a phony match. */ |
ca1d1d23 JB |
1143 | if ((pos - limit) * direction <= len) |
1144 | break; /* ran off the end */ | |
1145 | /* Found what might be a match. | |
1146 | Set POS back to last (first if reverse) char pos. */ | |
1147 | pos -= infinity; | |
1148 | i = dirlen - direction; | |
1149 | while ((i -= direction) + direction != 0) | |
1150 | { | |
1151 | pos -= direction; | |
8d505039 | 1152 | if (pat[i] != (trt != 0 |
2fd7a4a4 | 1153 | ? XINT (trt[FETCH_BYTE (pos)]) |
5679531d | 1154 | : FETCH_BYTE (pos))) |
ca1d1d23 JB |
1155 | break; |
1156 | } | |
1157 | /* Above loop has moved POS part or all the way | |
1158 | back to the first char pos (last char pos if reverse). | |
1159 | Set it once again at the last (first if reverse) char. */ | |
1160 | pos += dirlen - i- direction; | |
1161 | if (i + direction == 0) | |
1162 | { | |
1163 | pos -= direction; | |
1113d9db | 1164 | |
ca325161 RS |
1165 | set_search_regs (pos + ((direction > 0) ? 1 - len : 0), |
1166 | len); | |
1167 | ||
ca1d1d23 JB |
1168 | if ((n -= direction) != 0) |
1169 | pos += dirlen; /* to resume search */ | |
1170 | else | |
1171 | return ((direction > 0) | |
1172 | ? search_regs.end[0] : search_regs.start[0]); | |
1173 | } | |
1174 | else | |
1175 | pos += stride_for_teases; | |
1176 | } | |
1177 | } | |
1178 | /* We have done one clump. Can we continue? */ | |
1179 | if ((lim - pos) * direction < 0) | |
1180 | return ((0 - n) * direction); | |
1181 | } | |
1182 | return pos; | |
1183 | } | |
1184 | } | |
ca325161 RS |
1185 | |
1186 | /* Record beginning BEG and end BEG + LEN | |
1187 | for a match just found in the current buffer. */ | |
1188 | ||
1189 | static void | |
1190 | set_search_regs (beg, len) | |
1191 | int beg, len; | |
1192 | { | |
1193 | /* Make sure we have registers in which to store | |
1194 | the match position. */ | |
1195 | if (search_regs.num_regs == 0) | |
1196 | { | |
2d4a771a RS |
1197 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
1198 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 1199 | search_regs.num_regs = 2; |
ca325161 RS |
1200 | } |
1201 | ||
1202 | search_regs.start[0] = beg; | |
1203 | search_regs.end[0] = beg + len; | |
a3668d92 | 1204 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 1205 | } |
ca1d1d23 JB |
1206 | \f |
1207 | /* Given a string of words separated by word delimiters, | |
1208 | compute a regexp that matches those exact words | |
1209 | separated by arbitrary punctuation. */ | |
1210 | ||
1211 | static Lisp_Object | |
1212 | wordify (string) | |
1213 | Lisp_Object string; | |
1214 | { | |
1215 | register unsigned char *p, *o; | |
1216 | register int i, len, punct_count = 0, word_count = 0; | |
1217 | Lisp_Object val; | |
1218 | ||
1219 | CHECK_STRING (string, 0); | |
1220 | p = XSTRING (string)->data; | |
1221 | len = XSTRING (string)->size; | |
1222 | ||
1223 | for (i = 0; i < len; i++) | |
1224 | if (SYNTAX (p[i]) != Sword) | |
1225 | { | |
1226 | punct_count++; | |
1227 | if (i > 0 && SYNTAX (p[i-1]) == Sword) word_count++; | |
1228 | } | |
1229 | if (SYNTAX (p[len-1]) == Sword) word_count++; | |
1230 | if (!word_count) return build_string (""); | |
1231 | ||
1232 | val = make_string (p, len - punct_count + 5 * (word_count - 1) + 4); | |
1233 | ||
1234 | o = XSTRING (val)->data; | |
1235 | *o++ = '\\'; | |
1236 | *o++ = 'b'; | |
1237 | ||
1238 | for (i = 0; i < len; i++) | |
1239 | if (SYNTAX (p[i]) == Sword) | |
1240 | *o++ = p[i]; | |
1241 | else if (i > 0 && SYNTAX (p[i-1]) == Sword && --word_count) | |
1242 | { | |
1243 | *o++ = '\\'; | |
1244 | *o++ = 'W'; | |
1245 | *o++ = '\\'; | |
1246 | *o++ = 'W'; | |
1247 | *o++ = '*'; | |
1248 | } | |
1249 | ||
1250 | *o++ = '\\'; | |
1251 | *o++ = 'b'; | |
1252 | ||
1253 | return val; | |
1254 | } | |
1255 | \f | |
1256 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
6af43974 | 1257 | "MSearch backward: ", |
ca1d1d23 JB |
1258 | "Search backward from point for STRING.\n\ |
1259 | Set point to the beginning of the occurrence found, and return point.\n\ | |
1260 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1261 | The match found must not extend before that position.\n\ | |
1262 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1263 | If not nil and not t, position at limit of search and return nil.\n\ | |
1264 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1265 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1266 | (string, bound, noerror, count) | |
1267 | Lisp_Object string, bound, noerror, count; | |
1268 | { | |
b819a390 | 1269 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
1270 | } |
1271 | ||
6af43974 | 1272 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
ca1d1d23 JB |
1273 | "Search forward from point for STRING.\n\ |
1274 | Set point to the end of the occurrence found, and return point.\n\ | |
1275 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1276 | The match found must not extend after that position. nil is equivalent\n\ | |
1277 | to (point-max).\n\ | |
1278 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1279 | If not nil and not t, move to limit of search and return nil.\n\ | |
1280 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1281 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1282 | (string, bound, noerror, count) | |
1283 | Lisp_Object string, bound, noerror, count; | |
1284 | { | |
b819a390 | 1285 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
1286 | } |
1287 | ||
1288 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
1289 | "sWord search backward: ", | |
1290 | "Search backward from point for STRING, ignoring differences in punctuation.\n\ | |
1291 | Set point to the beginning of the occurrence found, and return point.\n\ | |
1292 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1293 | The match found must not extend before that position.\n\ | |
1294 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1295 | If not nil and not t, move to limit of search and return nil.\n\ | |
1296 | Optional fourth argument is repeat count--search for successive occurrences.") | |
1297 | (string, bound, noerror, count) | |
1298 | Lisp_Object string, bound, noerror, count; | |
1299 | { | |
b819a390 | 1300 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
1301 | } |
1302 | ||
1303 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
1304 | "sWord search: ", | |
1305 | "Search forward from point for STRING, ignoring differences in punctuation.\n\ | |
1306 | Set point to the end of the occurrence found, and return point.\n\ | |
1307 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1308 | The match found must not extend after that position.\n\ | |
1309 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1310 | If not nil and not t, move to limit of search and return nil.\n\ | |
1311 | Optional fourth argument is repeat count--search for successive occurrences.") | |
1312 | (string, bound, noerror, count) | |
1313 | Lisp_Object string, bound, noerror, count; | |
1314 | { | |
b819a390 | 1315 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
1316 | } |
1317 | ||
1318 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
1319 | "sRE search backward: ", | |
1320 | "Search backward from point for match for regular expression REGEXP.\n\ | |
1321 | Set point to the beginning of the match, and return point.\n\ | |
1322 | The match found is the one starting last in the buffer\n\ | |
19c0a730 | 1323 | and yet ending before the origin of the search.\n\ |
ca1d1d23 JB |
1324 | An optional second argument bounds the search; it is a buffer position.\n\ |
1325 | The match found must start at or after that position.\n\ | |
1326 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1327 | If not nil and not t, move to limit of search and return nil.\n\ | |
1328 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1329 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
19c0a730 KH |
1330 | (regexp, bound, noerror, count) |
1331 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 1332 | { |
b819a390 | 1333 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
1334 | } |
1335 | ||
1336 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
1337 | "sRE search: ", | |
1338 | "Search forward from point for regular expression REGEXP.\n\ | |
1339 | Set point to the end of the occurrence found, and return point.\n\ | |
1340 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1341 | The match found must not extend after that position.\n\ | |
1342 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1343 | If not nil and not t, move to limit of search and return nil.\n\ | |
1344 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1345 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
19c0a730 KH |
1346 | (regexp, bound, noerror, count) |
1347 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 1348 | { |
b819a390 RS |
1349 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
1350 | } | |
1351 | ||
1352 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
1353 | "sPosix search backward: ", | |
1354 | "Search backward from point for match for regular expression REGEXP.\n\ | |
1355 | Find the longest match in accord with Posix regular expression rules.\n\ | |
1356 | Set point to the beginning of the match, and return point.\n\ | |
1357 | The match found is the one starting last in the buffer\n\ | |
1358 | and yet ending before the origin of the search.\n\ | |
1359 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1360 | The match found must start at or after that position.\n\ | |
1361 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1362 | If not nil and not t, move to limit of search and return nil.\n\ | |
1363 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1364 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1365 | (regexp, bound, noerror, count) | |
1366 | Lisp_Object regexp, bound, noerror, count; | |
1367 | { | |
1368 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
1369 | } | |
1370 | ||
1371 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
1372 | "sPosix search: ", | |
1373 | "Search forward from point for regular expression REGEXP.\n\ | |
1374 | Find the longest match in accord with Posix regular expression rules.\n\ | |
1375 | Set point to the end of the occurrence found, and return point.\n\ | |
1376 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1377 | The match found must not extend after that position.\n\ | |
1378 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1379 | If not nil and not t, move to limit of search and return nil.\n\ | |
1380 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1381 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1382 | (regexp, bound, noerror, count) | |
1383 | Lisp_Object regexp, bound, noerror, count; | |
1384 | { | |
1385 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
1386 | } |
1387 | \f | |
d7a5ad5f | 1388 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
ca1d1d23 JB |
1389 | "Replace text matched by last search with NEWTEXT.\n\ |
1390 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text.\n\ | |
5b9cf4b2 RS |
1391 | Otherwise maybe capitalize the whole text, or maybe just word initials,\n\ |
1392 | based on the replaced text.\n\ | |
1393 | If the replaced text has only capital letters\n\ | |
1394 | and has at least one multiletter word, convert NEWTEXT to all caps.\n\ | |
1395 | If the replaced text has at least one word starting with a capital letter,\n\ | |
1396 | then capitalize each word in NEWTEXT.\n\n\ | |
ca1d1d23 JB |
1397 | If third arg LITERAL is non-nil, insert NEWTEXT literally.\n\ |
1398 | Otherwise treat `\\' as special:\n\ | |
1399 | `\\&' in NEWTEXT means substitute original matched text.\n\ | |
1400 | `\\N' means substitute what matched the Nth `\\(...\\)'.\n\ | |
1401 | If Nth parens didn't match, substitute nothing.\n\ | |
1402 | `\\\\' means insert one `\\'.\n\ | |
1113d9db | 1403 | FIXEDCASE and LITERAL are optional arguments.\n\ |
080c45fd RS |
1404 | Leaves point at end of replacement text.\n\ |
1405 | \n\ | |
1406 | The optional fourth argument STRING can be a string to modify.\n\ | |
1407 | In that case, this function creates and returns a new string\n\ | |
d7a5ad5f RS |
1408 | which is made by replacing the part of STRING that was matched.\n\ |
1409 | \n\ | |
1410 | The optional fifth argument SUBEXP specifies a subexpression of the match.\n\ | |
1411 | It says to replace just that subexpression instead of the whole match.\n\ | |
1412 | This is useful only after a regular expression search or match\n\ | |
1413 | since only regular expressions have distinguished subexpressions.") | |
1414 | (newtext, fixedcase, literal, string, subexp) | |
1415 | Lisp_Object newtext, fixedcase, literal, string, subexp; | |
ca1d1d23 JB |
1416 | { |
1417 | enum { nochange, all_caps, cap_initial } case_action; | |
1418 | register int pos, last; | |
1419 | int some_multiletter_word; | |
97832bd0 | 1420 | int some_lowercase; |
73dc8771 | 1421 | int some_uppercase; |
208767c3 | 1422 | int some_nonuppercase_initial; |
ca1d1d23 JB |
1423 | register int c, prevc; |
1424 | int inslen; | |
d7a5ad5f | 1425 | int sub; |
3e18eecf | 1426 | int opoint, newpoint; |
ca1d1d23 | 1427 | |
16fdc568 | 1428 | CHECK_STRING (newtext, 0); |
ca1d1d23 | 1429 | |
080c45fd RS |
1430 | if (! NILP (string)) |
1431 | CHECK_STRING (string, 4); | |
1432 | ||
ca1d1d23 JB |
1433 | case_action = nochange; /* We tried an initialization */ |
1434 | /* but some C compilers blew it */ | |
4746118a JB |
1435 | |
1436 | if (search_regs.num_regs <= 0) | |
1437 | error ("replace-match called before any match found"); | |
1438 | ||
d7a5ad5f RS |
1439 | if (NILP (subexp)) |
1440 | sub = 0; | |
1441 | else | |
1442 | { | |
1443 | CHECK_NUMBER (subexp, 3); | |
1444 | sub = XINT (subexp); | |
1445 | if (sub < 0 || sub >= search_regs.num_regs) | |
1446 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
1447 | } | |
1448 | ||
080c45fd RS |
1449 | if (NILP (string)) |
1450 | { | |
d7a5ad5f RS |
1451 | if (search_regs.start[sub] < BEGV |
1452 | || search_regs.start[sub] > search_regs.end[sub] | |
1453 | || search_regs.end[sub] > ZV) | |
1454 | args_out_of_range (make_number (search_regs.start[sub]), | |
1455 | make_number (search_regs.end[sub])); | |
080c45fd RS |
1456 | } |
1457 | else | |
1458 | { | |
d7a5ad5f RS |
1459 | if (search_regs.start[sub] < 0 |
1460 | || search_regs.start[sub] > search_regs.end[sub] | |
1461 | || search_regs.end[sub] > XSTRING (string)->size) | |
1462 | args_out_of_range (make_number (search_regs.start[sub]), | |
1463 | make_number (search_regs.end[sub])); | |
080c45fd | 1464 | } |
ca1d1d23 JB |
1465 | |
1466 | if (NILP (fixedcase)) | |
1467 | { | |
1468 | /* Decide how to casify by examining the matched text. */ | |
1469 | ||
d7a5ad5f | 1470 | last = search_regs.end[sub]; |
ca1d1d23 JB |
1471 | prevc = '\n'; |
1472 | case_action = all_caps; | |
1473 | ||
1474 | /* some_multiletter_word is set nonzero if any original word | |
1475 | is more than one letter long. */ | |
1476 | some_multiletter_word = 0; | |
97832bd0 | 1477 | some_lowercase = 0; |
208767c3 | 1478 | some_nonuppercase_initial = 0; |
73dc8771 | 1479 | some_uppercase = 0; |
ca1d1d23 | 1480 | |
d7a5ad5f | 1481 | for (pos = search_regs.start[sub]; pos < last; pos++) |
ca1d1d23 | 1482 | { |
080c45fd | 1483 | if (NILP (string)) |
5679531d | 1484 | c = FETCH_BYTE (pos); |
080c45fd RS |
1485 | else |
1486 | c = XSTRING (string)->data[pos]; | |
1487 | ||
ca1d1d23 JB |
1488 | if (LOWERCASEP (c)) |
1489 | { | |
1490 | /* Cannot be all caps if any original char is lower case */ | |
1491 | ||
97832bd0 | 1492 | some_lowercase = 1; |
ca1d1d23 | 1493 | if (SYNTAX (prevc) != Sword) |
208767c3 | 1494 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
1495 | else |
1496 | some_multiletter_word = 1; | |
1497 | } | |
1498 | else if (!NOCASEP (c)) | |
1499 | { | |
73dc8771 | 1500 | some_uppercase = 1; |
97832bd0 | 1501 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 1502 | ; |
97832bd0 | 1503 | else |
ca1d1d23 JB |
1504 | some_multiletter_word = 1; |
1505 | } | |
208767c3 RS |
1506 | else |
1507 | { | |
1508 | /* If the initial is a caseless word constituent, | |
1509 | treat that like a lowercase initial. */ | |
1510 | if (SYNTAX (prevc) != Sword) | |
1511 | some_nonuppercase_initial = 1; | |
1512 | } | |
ca1d1d23 JB |
1513 | |
1514 | prevc = c; | |
1515 | } | |
1516 | ||
97832bd0 RS |
1517 | /* Convert to all caps if the old text is all caps |
1518 | and has at least one multiletter word. */ | |
1519 | if (! some_lowercase && some_multiletter_word) | |
1520 | case_action = all_caps; | |
c4d460ce | 1521 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 1522 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 1523 | case_action = cap_initial; |
208767c3 | 1524 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
1525 | /* Should x -> yz, operating on X, give Yz or YZ? |
1526 | We'll assume the latter. */ | |
1527 | case_action = all_caps; | |
97832bd0 RS |
1528 | else |
1529 | case_action = nochange; | |
ca1d1d23 JB |
1530 | } |
1531 | ||
080c45fd RS |
1532 | /* Do replacement in a string. */ |
1533 | if (!NILP (string)) | |
1534 | { | |
1535 | Lisp_Object before, after; | |
1536 | ||
1537 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
1538 | make_number (search_regs.start[sub])); |
1539 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 1540 | |
636a5e28 RS |
1541 | /* Substitute parts of the match into NEWTEXT |
1542 | if desired. */ | |
080c45fd RS |
1543 | if (NILP (literal)) |
1544 | { | |
1545 | int lastpos = -1; | |
1546 | /* We build up the substituted string in ACCUM. */ | |
1547 | Lisp_Object accum; | |
1548 | Lisp_Object middle; | |
1549 | ||
1550 | accum = Qnil; | |
1551 | ||
1552 | for (pos = 0; pos < XSTRING (newtext)->size; pos++) | |
1553 | { | |
1554 | int substart = -1; | |
1555 | int subend; | |
1e79ec24 | 1556 | int delbackslash = 0; |
080c45fd RS |
1557 | |
1558 | c = XSTRING (newtext)->data[pos]; | |
1559 | if (c == '\\') | |
1560 | { | |
1561 | c = XSTRING (newtext)->data[++pos]; | |
1562 | if (c == '&') | |
1563 | { | |
d7a5ad5f RS |
1564 | substart = search_regs.start[sub]; |
1565 | subend = search_regs.end[sub]; | |
080c45fd RS |
1566 | } |
1567 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') | |
1568 | { | |
ad10348f | 1569 | if (search_regs.start[c - '0'] >= 0) |
080c45fd RS |
1570 | { |
1571 | substart = search_regs.start[c - '0']; | |
1572 | subend = search_regs.end[c - '0']; | |
1573 | } | |
1574 | } | |
1e79ec24 KH |
1575 | else if (c == '\\') |
1576 | delbackslash = 1; | |
636a5e28 RS |
1577 | else |
1578 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
1579 | } |
1580 | if (substart >= 0) | |
1581 | { | |
1582 | if (pos - 1 != lastpos + 1) | |
1e79ec24 KH |
1583 | middle = Fsubstring (newtext, |
1584 | make_number (lastpos + 1), | |
1585 | make_number (pos - 1)); | |
080c45fd RS |
1586 | else |
1587 | middle = Qnil; | |
1588 | accum = concat3 (accum, middle, | |
1589 | Fsubstring (string, make_number (substart), | |
1590 | make_number (subend))); | |
1591 | lastpos = pos; | |
1592 | } | |
1e79ec24 KH |
1593 | else if (delbackslash) |
1594 | { | |
1595 | middle = Fsubstring (newtext, make_number (lastpos + 1), | |
1596 | make_number (pos)); | |
1597 | accum = concat2 (accum, middle); | |
1598 | lastpos = pos; | |
1599 | } | |
080c45fd RS |
1600 | } |
1601 | ||
1602 | if (pos != lastpos + 1) | |
1e79ec24 KH |
1603 | middle = Fsubstring (newtext, make_number (lastpos + 1), |
1604 | make_number (pos)); | |
080c45fd RS |
1605 | else |
1606 | middle = Qnil; | |
1607 | ||
1608 | newtext = concat2 (accum, middle); | |
1609 | } | |
1610 | ||
636a5e28 | 1611 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
1612 | if (case_action == all_caps) |
1613 | newtext = Fupcase (newtext); | |
1614 | else if (case_action == cap_initial) | |
2b2eead9 | 1615 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
1616 | |
1617 | return concat3 (before, newtext, after); | |
1618 | } | |
1619 | ||
b0eba991 RS |
1620 | /* Record point, the move (quietly) to the start of the match. */ |
1621 | if (PT > search_regs.start[sub]) | |
1622 | opoint = PT - ZV; | |
1623 | else | |
1624 | opoint = PT; | |
1625 | ||
1626 | temp_set_point (search_regs.start[sub], current_buffer); | |
1627 | ||
9a76659d JB |
1628 | /* We insert the replacement text before the old text, and then |
1629 | delete the original text. This means that markers at the | |
1630 | beginning or end of the original will float to the corresponding | |
1631 | position in the replacement. */ | |
ca1d1d23 | 1632 | if (!NILP (literal)) |
16fdc568 | 1633 | Finsert_and_inherit (1, &newtext); |
ca1d1d23 JB |
1634 | else |
1635 | { | |
1636 | struct gcpro gcpro1; | |
16fdc568 | 1637 | GCPRO1 (newtext); |
ca1d1d23 | 1638 | |
16fdc568 | 1639 | for (pos = 0; pos < XSTRING (newtext)->size; pos++) |
ca1d1d23 | 1640 | { |
6ec8bbd2 | 1641 | int offset = PT - search_regs.start[sub]; |
9a76659d | 1642 | |
16fdc568 | 1643 | c = XSTRING (newtext)->data[pos]; |
ca1d1d23 JB |
1644 | if (c == '\\') |
1645 | { | |
16fdc568 | 1646 | c = XSTRING (newtext)->data[++pos]; |
ca1d1d23 | 1647 | if (c == '&') |
9a76659d JB |
1648 | Finsert_buffer_substring |
1649 | (Fcurrent_buffer (), | |
d7a5ad5f RS |
1650 | make_number (search_regs.start[sub] + offset), |
1651 | make_number (search_regs.end[sub] + offset)); | |
78445046 | 1652 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
1653 | { |
1654 | if (search_regs.start[c - '0'] >= 1) | |
9a76659d JB |
1655 | Finsert_buffer_substring |
1656 | (Fcurrent_buffer (), | |
1657 | make_number (search_regs.start[c - '0'] + offset), | |
1658 | make_number (search_regs.end[c - '0'] + offset)); | |
ca1d1d23 | 1659 | } |
636a5e28 | 1660 | else if (c == '\\') |
ca1d1d23 | 1661 | insert_char (c); |
636a5e28 RS |
1662 | else |
1663 | error ("Invalid use of `\\' in replacement text"); | |
ca1d1d23 JB |
1664 | } |
1665 | else | |
1666 | insert_char (c); | |
1667 | } | |
1668 | UNGCPRO; | |
1669 | } | |
1670 | ||
6ec8bbd2 | 1671 | inslen = PT - (search_regs.start[sub]); |
d7a5ad5f | 1672 | del_range (search_regs.start[sub] + inslen, search_regs.end[sub] + inslen); |
ca1d1d23 JB |
1673 | |
1674 | if (case_action == all_caps) | |
6ec8bbd2 | 1675 | Fupcase_region (make_number (PT - inslen), make_number (PT)); |
ca1d1d23 | 1676 | else if (case_action == cap_initial) |
6ec8bbd2 | 1677 | Fupcase_initials_region (make_number (PT - inslen), make_number (PT)); |
b0eba991 | 1678 | |
3e18eecf RS |
1679 | newpoint = PT; |
1680 | ||
b0eba991 | 1681 | /* Put point back where it was in the text. */ |
8d808a65 | 1682 | if (opoint <= 0) |
b0eba991 RS |
1683 | temp_set_point (opoint + ZV, current_buffer); |
1684 | else | |
1685 | temp_set_point (opoint, current_buffer); | |
1686 | ||
1687 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 1688 | move_if_not_intangible (newpoint); |
b0eba991 | 1689 | |
ca1d1d23 JB |
1690 | return Qnil; |
1691 | } | |
1692 | \f | |
1693 | static Lisp_Object | |
1694 | match_limit (num, beginningp) | |
1695 | Lisp_Object num; | |
1696 | int beginningp; | |
1697 | { | |
1698 | register int n; | |
1699 | ||
1700 | CHECK_NUMBER (num, 0); | |
1701 | n = XINT (num); | |
4746118a JB |
1702 | if (n < 0 || n >= search_regs.num_regs) |
1703 | args_out_of_range (num, make_number (search_regs.num_regs)); | |
1704 | if (search_regs.num_regs <= 0 | |
1705 | || search_regs.start[n] < 0) | |
ca1d1d23 JB |
1706 | return Qnil; |
1707 | return (make_number ((beginningp) ? search_regs.start[n] | |
1708 | : search_regs.end[n])); | |
1709 | } | |
1710 | ||
1711 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
1712 | "Return position of start of text matched by last search.\n\ | |
5806161b EN |
1713 | SUBEXP, a number, specifies which parenthesized expression in the last\n\ |
1714 | regexp.\n\ | |
1715 | Value is nil if SUBEXPth pair didn't match, or there were less than\n\ | |
1716 | SUBEXP pairs.\n\ | |
ca1d1d23 | 1717 | Zero means the entire text matched by the whole regexp or whole string.") |
5806161b EN |
1718 | (subexp) |
1719 | Lisp_Object subexp; | |
ca1d1d23 | 1720 | { |
5806161b | 1721 | return match_limit (subexp, 1); |
ca1d1d23 JB |
1722 | } |
1723 | ||
1724 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
1725 | "Return position of end of text matched by last search.\n\ | |
5806161b EN |
1726 | SUBEXP, a number, specifies which parenthesized expression in the last\n\ |
1727 | regexp.\n\ | |
1728 | Value is nil if SUBEXPth pair didn't match, or there were less than\n\ | |
1729 | SUBEXP pairs.\n\ | |
ca1d1d23 | 1730 | Zero means the entire text matched by the whole regexp or whole string.") |
5806161b EN |
1731 | (subexp) |
1732 | Lisp_Object subexp; | |
ca1d1d23 | 1733 | { |
5806161b | 1734 | return match_limit (subexp, 0); |
ca1d1d23 JB |
1735 | } |
1736 | ||
56256c2a | 1737 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 2, 0, |
ca1d1d23 JB |
1738 | "Return a list containing all info on what the last search matched.\n\ |
1739 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'.\n\ | |
1740 | All the elements are markers or nil (nil if the Nth pair didn't match)\n\ | |
1741 | if the last match was on a buffer; integers or nil if a string was matched.\n\ | |
56256c2a RS |
1742 | Use `store-match-data' to reinstate the data in this list.\n\ |
1743 | \n\ | |
1744 | If INTEGERS (the optional first argument) is non-nil, always use integers\n\ | |
8ca821e9 | 1745 | \(rather than markers) to represent buffer positions.\n\ |
56256c2a RS |
1746 | If REUSE is a list, reuse it as part of the value. If REUSE is long enough\n\ |
1747 | to hold all the values, and if INTEGERS is non-nil, no consing is done.") | |
1748 | (integers, reuse) | |
1749 | Lisp_Object integers, reuse; | |
ca1d1d23 | 1750 | { |
56256c2a | 1751 | Lisp_Object tail, prev; |
4746118a | 1752 | Lisp_Object *data; |
ca1d1d23 JB |
1753 | int i, len; |
1754 | ||
daa37602 | 1755 | if (NILP (last_thing_searched)) |
c36bcf1b | 1756 | return Qnil; |
daa37602 | 1757 | |
4746118a JB |
1758 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs) |
1759 | * sizeof (Lisp_Object)); | |
1760 | ||
ca1d1d23 | 1761 | len = -1; |
4746118a | 1762 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1763 | { |
1764 | int start = search_regs.start[i]; | |
1765 | if (start >= 0) | |
1766 | { | |
56256c2a RS |
1767 | if (EQ (last_thing_searched, Qt) |
1768 | || ! NILP (integers)) | |
ca1d1d23 | 1769 | { |
c235cce7 KH |
1770 | XSETFASTINT (data[2 * i], start); |
1771 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 1772 | } |
0ed62dc7 | 1773 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
1774 | { |
1775 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
1776 | Fset_marker (data[2 * i], |
1777 | make_number (start), | |
1778 | last_thing_searched); | |
ca1d1d23 JB |
1779 | data[2 * i + 1] = Fmake_marker (); |
1780 | Fset_marker (data[2 * i + 1], | |
daa37602 JB |
1781 | make_number (search_regs.end[i]), |
1782 | last_thing_searched); | |
ca1d1d23 | 1783 | } |
daa37602 JB |
1784 | else |
1785 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
1786 | abort (); | |
1787 | ||
ca1d1d23 JB |
1788 | len = i; |
1789 | } | |
1790 | else | |
1791 | data[2 * i] = data [2 * i + 1] = Qnil; | |
1792 | } | |
56256c2a RS |
1793 | |
1794 | /* If REUSE is not usable, cons up the values and return them. */ | |
1795 | if (! CONSP (reuse)) | |
1796 | return Flist (2 * len + 2, data); | |
1797 | ||
1798 | /* If REUSE is a list, store as many value elements as will fit | |
1799 | into the elements of REUSE. */ | |
1800 | for (i = 0, tail = reuse; CONSP (tail); | |
1801 | i++, tail = XCONS (tail)->cdr) | |
1802 | { | |
1803 | if (i < 2 * len + 2) | |
1804 | XCONS (tail)->car = data[i]; | |
1805 | else | |
1806 | XCONS (tail)->car = Qnil; | |
1807 | prev = tail; | |
1808 | } | |
1809 | ||
1810 | /* If we couldn't fit all value elements into REUSE, | |
1811 | cons up the rest of them and add them to the end of REUSE. */ | |
1812 | if (i < 2 * len + 2) | |
1813 | XCONS (prev)->cdr = Flist (2 * len + 2 - i, data + i); | |
1814 | ||
1815 | return reuse; | |
ca1d1d23 JB |
1816 | } |
1817 | ||
1818 | ||
1819 | DEFUN ("store-match-data", Fstore_match_data, Sstore_match_data, 1, 1, 0, | |
1820 | "Set internal data on last search match from elements of LIST.\n\ | |
1821 | LIST should have been created by calling `match-data' previously.") | |
1822 | (list) | |
1823 | register Lisp_Object list; | |
1824 | { | |
1825 | register int i; | |
1826 | register Lisp_Object marker; | |
1827 | ||
7074fde6 FP |
1828 | if (running_asynch_code) |
1829 | save_search_regs (); | |
1830 | ||
ca1d1d23 | 1831 | if (!CONSP (list) && !NILP (list)) |
b37902c8 | 1832 | list = wrong_type_argument (Qconsp, list); |
ca1d1d23 | 1833 | |
daa37602 JB |
1834 | /* Unless we find a marker with a buffer in LIST, assume that this |
1835 | match data came from a string. */ | |
1836 | last_thing_searched = Qt; | |
1837 | ||
4746118a JB |
1838 | /* Allocate registers if they don't already exist. */ |
1839 | { | |
d084e942 | 1840 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
1841 | |
1842 | if (length > search_regs.num_regs) | |
1843 | { | |
1113d9db JB |
1844 | if (search_regs.num_regs == 0) |
1845 | { | |
1846 | search_regs.start | |
1847 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
1848 | search_regs.end | |
1849 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
1850 | } | |
4746118a | 1851 | else |
1113d9db JB |
1852 | { |
1853 | search_regs.start | |
1854 | = (regoff_t *) xrealloc (search_regs.start, | |
1855 | length * sizeof (regoff_t)); | |
1856 | search_regs.end | |
1857 | = (regoff_t *) xrealloc (search_regs.end, | |
1858 | length * sizeof (regoff_t)); | |
1859 | } | |
4746118a | 1860 | |
487282dc | 1861 | search_regs.num_regs = length; |
4746118a JB |
1862 | } |
1863 | } | |
1864 | ||
1865 | for (i = 0; i < search_regs.num_regs; i++) | |
ca1d1d23 JB |
1866 | { |
1867 | marker = Fcar (list); | |
1868 | if (NILP (marker)) | |
1869 | { | |
1870 | search_regs.start[i] = -1; | |
1871 | list = Fcdr (list); | |
1872 | } | |
1873 | else | |
1874 | { | |
0ed62dc7 | 1875 | if (MARKERP (marker)) |
daa37602 JB |
1876 | { |
1877 | if (XMARKER (marker)->buffer == 0) | |
c235cce7 | 1878 | XSETFASTINT (marker, 0); |
daa37602 | 1879 | else |
a3668d92 | 1880 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); |
daa37602 | 1881 | } |
ca1d1d23 JB |
1882 | |
1883 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
1884 | search_regs.start[i] = XINT (marker); | |
1885 | list = Fcdr (list); | |
1886 | ||
1887 | marker = Fcar (list); | |
0ed62dc7 | 1888 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
c235cce7 | 1889 | XSETFASTINT (marker, 0); |
ca1d1d23 JB |
1890 | |
1891 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
1892 | search_regs.end[i] = XINT (marker); | |
1893 | } | |
1894 | list = Fcdr (list); | |
1895 | } | |
1896 | ||
1897 | return Qnil; | |
1898 | } | |
1899 | ||
7074fde6 FP |
1900 | /* If non-zero the match data have been saved in saved_search_regs |
1901 | during the execution of a sentinel or filter. */ | |
75ebf74b | 1902 | static int search_regs_saved; |
7074fde6 FP |
1903 | static struct re_registers saved_search_regs; |
1904 | ||
1905 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
1906 | if asynchronous code (filter or sentinel) is running. */ | |
1907 | static void | |
1908 | save_search_regs () | |
1909 | { | |
1910 | if (!search_regs_saved) | |
1911 | { | |
1912 | saved_search_regs.num_regs = search_regs.num_regs; | |
1913 | saved_search_regs.start = search_regs.start; | |
1914 | saved_search_regs.end = search_regs.end; | |
1915 | search_regs.num_regs = 0; | |
2d4a771a RS |
1916 | search_regs.start = 0; |
1917 | search_regs.end = 0; | |
7074fde6 FP |
1918 | |
1919 | search_regs_saved = 1; | |
1920 | } | |
1921 | } | |
1922 | ||
1923 | /* Called upon exit from filters and sentinels. */ | |
1924 | void | |
1925 | restore_match_data () | |
1926 | { | |
1927 | if (search_regs_saved) | |
1928 | { | |
1929 | if (search_regs.num_regs > 0) | |
1930 | { | |
1931 | xfree (search_regs.start); | |
1932 | xfree (search_regs.end); | |
1933 | } | |
1934 | search_regs.num_regs = saved_search_regs.num_regs; | |
1935 | search_regs.start = saved_search_regs.start; | |
1936 | search_regs.end = saved_search_regs.end; | |
1937 | ||
1938 | search_regs_saved = 0; | |
1939 | } | |
1940 | } | |
1941 | ||
ca1d1d23 JB |
1942 | /* Quote a string to inactivate reg-expr chars */ |
1943 | ||
1944 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
1945 | "Return a regexp string which matches exactly STRING and nothing else.") | |
5806161b EN |
1946 | (string) |
1947 | Lisp_Object string; | |
ca1d1d23 JB |
1948 | { |
1949 | register unsigned char *in, *out, *end; | |
1950 | register unsigned char *temp; | |
1951 | ||
5806161b | 1952 | CHECK_STRING (string, 0); |
ca1d1d23 | 1953 | |
5806161b | 1954 | temp = (unsigned char *) alloca (XSTRING (string)->size * 2); |
ca1d1d23 JB |
1955 | |
1956 | /* Now copy the data into the new string, inserting escapes. */ | |
1957 | ||
5806161b EN |
1958 | in = XSTRING (string)->data; |
1959 | end = in + XSTRING (string)->size; | |
ca1d1d23 JB |
1960 | out = temp; |
1961 | ||
1962 | for (; in != end; in++) | |
1963 | { | |
1964 | if (*in == '[' || *in == ']' | |
1965 | || *in == '*' || *in == '.' || *in == '\\' | |
1966 | || *in == '?' || *in == '+' | |
1967 | || *in == '^' || *in == '$') | |
1968 | *out++ = '\\'; | |
1969 | *out++ = *in; | |
1970 | } | |
1971 | ||
1972 | return make_string (temp, out - temp); | |
1973 | } | |
1974 | \f | |
1975 | syms_of_search () | |
1976 | { | |
1977 | register int i; | |
1978 | ||
487282dc KH |
1979 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
1980 | { | |
1981 | searchbufs[i].buf.allocated = 100; | |
1982 | searchbufs[i].buf.buffer = (unsigned char *) malloc (100); | |
1983 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; | |
1984 | searchbufs[i].regexp = Qnil; | |
1985 | staticpro (&searchbufs[i].regexp); | |
1986 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); | |
1987 | } | |
1988 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
1989 | |
1990 | Qsearch_failed = intern ("search-failed"); | |
1991 | staticpro (&Qsearch_failed); | |
1992 | Qinvalid_regexp = intern ("invalid-regexp"); | |
1993 | staticpro (&Qinvalid_regexp); | |
1994 | ||
1995 | Fput (Qsearch_failed, Qerror_conditions, | |
1996 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
1997 | Fput (Qsearch_failed, Qerror_message, | |
1998 | build_string ("Search failed")); | |
1999 | ||
2000 | Fput (Qinvalid_regexp, Qerror_conditions, | |
2001 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
2002 | Fput (Qinvalid_regexp, Qerror_message, | |
2003 | build_string ("Invalid regexp")); | |
2004 | ||
daa37602 JB |
2005 | last_thing_searched = Qnil; |
2006 | staticpro (&last_thing_searched); | |
2007 | ||
ca1d1d23 | 2008 | defsubr (&Slooking_at); |
b819a390 RS |
2009 | defsubr (&Sposix_looking_at); |
2010 | defsubr (&Sstring_match); | |
2011 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
2012 | defsubr (&Ssearch_forward); |
2013 | defsubr (&Ssearch_backward); | |
2014 | defsubr (&Sword_search_forward); | |
2015 | defsubr (&Sword_search_backward); | |
2016 | defsubr (&Sre_search_forward); | |
2017 | defsubr (&Sre_search_backward); | |
b819a390 RS |
2018 | defsubr (&Sposix_search_forward); |
2019 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
2020 | defsubr (&Sreplace_match); |
2021 | defsubr (&Smatch_beginning); | |
2022 | defsubr (&Smatch_end); | |
2023 | defsubr (&Smatch_data); | |
2024 | defsubr (&Sstore_match_data); | |
2025 | defsubr (&Sregexp_quote); | |
2026 | } |