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 | |
8 | the Free Software Foundation; either version 1, or (at your option) | |
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 | |
18 | the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */ | |
19 | ||
20 | ||
18160b98 | 21 | #include <config.h> |
ca1d1d23 JB |
22 | #include "lisp.h" |
23 | #include "syntax.h" | |
24 | #include "buffer.h" | |
9169c321 | 25 | #include "region-cache.h" |
ca1d1d23 | 26 | #include "commands.h" |
9ac0d9e0 | 27 | #include "blockinput.h" |
4746118a | 28 | |
ca1d1d23 JB |
29 | #include <sys/types.h> |
30 | #include "regex.h" | |
31 | ||
487282dc | 32 | #define REGEXP_CACHE_SIZE 5 |
ca1d1d23 | 33 | |
487282dc KH |
34 | /* If the regexp is non-nil, then the buffer contains the compiled form |
35 | of that regexp, suitable for searching. */ | |
36 | struct regexp_cache { | |
37 | struct regexp_cache *next; | |
38 | Lisp_Object regexp; | |
39 | struct re_pattern_buffer buf; | |
40 | char fastmap[0400]; | |
b819a390 RS |
41 | /* Nonzero means regexp was compiled to do full POSIX backtracking. */ |
42 | char posix; | |
487282dc | 43 | }; |
ca1d1d23 | 44 | |
487282dc KH |
45 | /* The instances of that struct. */ |
46 | struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; | |
ca1d1d23 | 47 | |
487282dc KH |
48 | /* The head of the linked list; points to the most recently used buffer. */ |
49 | struct regexp_cache *searchbuf_head; | |
ca1d1d23 | 50 | |
ca1d1d23 | 51 | |
4746118a JB |
52 | /* Every call to re_match, etc., must pass &search_regs as the regs |
53 | argument unless you can show it is unnecessary (i.e., if re_match | |
54 | is certainly going to be called again before region-around-match | |
55 | can be called). | |
56 | ||
57 | Since the registers are now dynamically allocated, we need to make | |
58 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
59 | been allocated by checking search_regs.num_regs. |
60 | ||
61 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
62 | buffer using bits in the re_pattern_buffer. This means that whenever |
63 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
64 | allocated any registers, and will allocate new registers the next |
65 | time you call a searching or matching function. Therefore, we need | |
66 | to call re_set_registers after compiling a new pattern or after | |
67 | setting the match registers, so that the regex functions will be | |
68 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
69 | static struct re_registers search_regs; |
70 | ||
daa37602 JB |
71 | /* The buffer in which the last search was performed, or |
72 | Qt if the last search was done in a string; | |
73 | Qnil if no searching has been done yet. */ | |
74 | static Lisp_Object last_thing_searched; | |
ca1d1d23 JB |
75 | |
76 | /* error condition signalled when regexp compile_pattern fails */ | |
77 | ||
78 | Lisp_Object Qinvalid_regexp; | |
79 | ||
ca325161 | 80 | static void set_search_regs (); |
044f81f1 | 81 | static void save_search_regs (); |
ca325161 | 82 | |
b819a390 RS |
83 | static int search_buffer (); |
84 | ||
ca1d1d23 JB |
85 | static void |
86 | matcher_overflow () | |
87 | { | |
88 | error ("Stack overflow in regexp matcher"); | |
89 | } | |
90 | ||
91 | #ifdef __STDC__ | |
92 | #define CONST const | |
93 | #else | |
94 | #define CONST | |
95 | #endif | |
96 | ||
b819a390 RS |
97 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
98 | PATTERN is the pattern to compile. | |
99 | CP is the place to put the result. | |
100 | TRANSLATE is a translation table for ignoring case, or NULL for none. | |
101 | REGP is the structure that says where to store the "register" | |
102 | values that will result from matching this pattern. | |
103 | If it is 0, we should compile the pattern not to record any | |
104 | subexpression bounds. | |
105 | POSIX is nonzero if we want full backtracking (POSIX style) | |
106 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 107 | |
487282dc | 108 | static void |
b819a390 | 109 | compile_pattern_1 (cp, pattern, translate, regp, posix) |
487282dc | 110 | struct regexp_cache *cp; |
ca1d1d23 | 111 | Lisp_Object pattern; |
ca1d1d23 | 112 | char *translate; |
487282dc | 113 | struct re_registers *regp; |
b819a390 | 114 | int posix; |
ca1d1d23 JB |
115 | { |
116 | CONST char *val; | |
b819a390 | 117 | reg_syntax_t old; |
ca1d1d23 | 118 | |
487282dc KH |
119 | cp->regexp = Qnil; |
120 | cp->buf.translate = translate; | |
b819a390 | 121 | cp->posix = posix; |
9ac0d9e0 | 122 | BLOCK_INPUT; |
b819a390 RS |
123 | old = re_set_syntax (RE_SYNTAX_EMACS |
124 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); | |
b90d9e80 | 125 | val = (CONST char *) re_compile_pattern ((char *) XSTRING (pattern)->data, |
487282dc | 126 | XSTRING (pattern)->size, &cp->buf); |
b819a390 | 127 | re_set_syntax (old); |
9ac0d9e0 | 128 | UNBLOCK_INPUT; |
ca1d1d23 | 129 | if (val) |
487282dc | 130 | Fsignal (Qinvalid_regexp, Fcons (build_string (val), Qnil)); |
1113d9db | 131 | |
487282dc | 132 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
133 | } |
134 | ||
135 | /* Compile a regexp if necessary, but first check to see if there's one in | |
b819a390 RS |
136 | the cache. |
137 | PATTERN is the pattern to compile. | |
138 | TRANSLATE is a translation table for ignoring case, or NULL for none. | |
139 | REGP is the structure that says where to store the "register" | |
140 | values that will result from matching this pattern. | |
141 | If it is 0, we should compile the pattern not to record any | |
142 | subexpression bounds. | |
143 | POSIX is nonzero if we want full backtracking (POSIX style) | |
144 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
145 | |
146 | struct re_pattern_buffer * | |
b819a390 | 147 | compile_pattern (pattern, regp, translate, posix) |
487282dc KH |
148 | Lisp_Object pattern; |
149 | struct re_registers *regp; | |
150 | char *translate; | |
b819a390 | 151 | int posix; |
487282dc KH |
152 | { |
153 | struct regexp_cache *cp, **cpp; | |
154 | ||
155 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
156 | { | |
157 | cp = *cpp; | |
158 | if (!NILP (Fstring_equal (cp->regexp, pattern)) | |
b819a390 RS |
159 | && cp->buf.translate == translate |
160 | && cp->posix == posix) | |
487282dc KH |
161 | break; |
162 | ||
163 | /* If we're at the end of the cache, compile into the last cell. */ | |
164 | if (cp->next == 0) | |
165 | { | |
b819a390 | 166 | compile_pattern_1 (cp, pattern, translate, regp, posix); |
487282dc KH |
167 | break; |
168 | } | |
169 | } | |
170 | ||
171 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
172 | either because we found it in the cache or because we just compiled it. | |
173 | Move it to the front of the queue to mark it as most recently used. */ | |
174 | *cpp = cp->next; | |
175 | cp->next = searchbuf_head; | |
176 | searchbuf_head = cp; | |
1113d9db | 177 | |
6639708c RS |
178 | /* Advise the searching functions about the space we have allocated |
179 | for register data. */ | |
180 | if (regp) | |
181 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
182 | ||
487282dc | 183 | return &cp->buf; |
ca1d1d23 JB |
184 | } |
185 | ||
186 | /* Error condition used for failing searches */ | |
187 | Lisp_Object Qsearch_failed; | |
188 | ||
189 | Lisp_Object | |
190 | signal_failure (arg) | |
191 | Lisp_Object arg; | |
192 | { | |
193 | Fsignal (Qsearch_failed, Fcons (arg, Qnil)); | |
194 | return Qnil; | |
195 | } | |
196 | \f | |
b819a390 RS |
197 | static Lisp_Object |
198 | looking_at_1 (string, posix) | |
ca1d1d23 | 199 | Lisp_Object string; |
b819a390 | 200 | int posix; |
ca1d1d23 JB |
201 | { |
202 | Lisp_Object val; | |
203 | unsigned char *p1, *p2; | |
204 | int s1, s2; | |
205 | register int i; | |
487282dc | 206 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 207 | |
7074fde6 FP |
208 | if (running_asynch_code) |
209 | save_search_regs (); | |
210 | ||
ca1d1d23 | 211 | CHECK_STRING (string, 0); |
487282dc KH |
212 | bufp = compile_pattern (string, &search_regs, |
213 | (!NILP (current_buffer->case_fold_search) | |
b819a390 RS |
214 | ? DOWNCASE_TABLE : 0), |
215 | posix); | |
ca1d1d23 JB |
216 | |
217 | immediate_quit = 1; | |
218 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
219 | ||
220 | /* Get pointers and sizes of the two strings | |
221 | that make up the visible portion of the buffer. */ | |
222 | ||
223 | p1 = BEGV_ADDR; | |
224 | s1 = GPT - BEGV; | |
225 | p2 = GAP_END_ADDR; | |
226 | s2 = ZV - GPT; | |
227 | if (s1 < 0) | |
228 | { | |
229 | p2 = p1; | |
230 | s2 = ZV - BEGV; | |
231 | s1 = 0; | |
232 | } | |
233 | if (s2 < 0) | |
234 | { | |
235 | s1 = ZV - BEGV; | |
236 | s2 = 0; | |
237 | } | |
238 | ||
487282dc | 239 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
ca1d1d23 JB |
240 | point - BEGV, &search_regs, |
241 | ZV - BEGV); | |
242 | if (i == -2) | |
243 | matcher_overflow (); | |
244 | ||
245 | val = (0 <= i ? Qt : Qnil); | |
4746118a | 246 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
247 | if (search_regs.start[i] >= 0) |
248 | { | |
249 | search_regs.start[i] += BEGV; | |
250 | search_regs.end[i] += BEGV; | |
251 | } | |
a3668d92 | 252 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
253 | immediate_quit = 0; |
254 | return val; | |
255 | } | |
256 | ||
b819a390 RS |
257 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
258 | "Return t if text after point matches regular expression PAT.\n\ | |
259 | This function modifies the match data that `match-beginning',\n\ | |
260 | `match-end' and `match-data' access; save and restore the match\n\ | |
261 | data if you want to preserve them.") | |
262 | (string) | |
263 | Lisp_Object string; | |
264 | { | |
265 | return looking_at_1 (string, 0); | |
266 | } | |
267 | ||
268 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
269 | "Return t if text after point matches regular expression PAT.\n\ | |
270 | Find the longest match, in accord with Posix regular expression rules.\n\ | |
271 | This function modifies the match data that `match-beginning',\n\ | |
272 | `match-end' and `match-data' access; save and restore the match\n\ | |
273 | data if you want to preserve them.") | |
274 | (string) | |
275 | Lisp_Object string; | |
276 | { | |
277 | return looking_at_1 (string, 1); | |
278 | } | |
279 | \f | |
280 | static Lisp_Object | |
281 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 282 | Lisp_Object regexp, string, start; |
b819a390 | 283 | int posix; |
ca1d1d23 JB |
284 | { |
285 | int val; | |
286 | int s; | |
487282dc | 287 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 288 | |
7074fde6 FP |
289 | if (running_asynch_code) |
290 | save_search_regs (); | |
291 | ||
ca1d1d23 JB |
292 | CHECK_STRING (regexp, 0); |
293 | CHECK_STRING (string, 1); | |
294 | ||
295 | if (NILP (start)) | |
296 | s = 0; | |
297 | else | |
298 | { | |
299 | int len = XSTRING (string)->size; | |
300 | ||
301 | CHECK_NUMBER (start, 2); | |
302 | s = XINT (start); | |
303 | if (s < 0 && -s <= len) | |
26faf9f4 | 304 | s = len + s; |
ca1d1d23 JB |
305 | else if (0 > s || s > len) |
306 | args_out_of_range (string, start); | |
307 | } | |
308 | ||
487282dc KH |
309 | bufp = compile_pattern (regexp, &search_regs, |
310 | (!NILP (current_buffer->case_fold_search) | |
b819a390 RS |
311 | ? DOWNCASE_TABLE : 0), |
312 | 0); | |
ca1d1d23 | 313 | immediate_quit = 1; |
487282dc | 314 | val = re_search (bufp, (char *) XSTRING (string)->data, |
ca1d1d23 JB |
315 | XSTRING (string)->size, s, XSTRING (string)->size - s, |
316 | &search_regs); | |
317 | immediate_quit = 0; | |
daa37602 | 318 | last_thing_searched = Qt; |
ca1d1d23 JB |
319 | if (val == -2) |
320 | matcher_overflow (); | |
321 | if (val < 0) return Qnil; | |
322 | return make_number (val); | |
323 | } | |
e59a8453 | 324 | |
b819a390 RS |
325 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
326 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
327 | If third arg START is non-nil, start search at that index in STRING.\n\ | |
328 | For index of first char beyond the match, do (match-end 0).\n\ | |
329 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
330 | matched by parenthesis constructs in the pattern.") | |
331 | (regexp, string, start) | |
332 | Lisp_Object regexp, string, start; | |
333 | { | |
334 | return string_match_1 (regexp, string, start, 0); | |
335 | } | |
336 | ||
337 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
338 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
339 | Find the longest match, in accord with Posix regular expression rules.\n\ | |
340 | If third arg START is non-nil, start search at that index in STRING.\n\ | |
341 | For index of first char beyond the match, do (match-end 0).\n\ | |
342 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
343 | matched by parenthesis constructs in the pattern.") | |
344 | (regexp, string, start) | |
345 | Lisp_Object regexp, string, start; | |
346 | { | |
347 | return string_match_1 (regexp, string, start, 1); | |
348 | } | |
349 | ||
e59a8453 RS |
350 | /* Match REGEXP against STRING, searching all of STRING, |
351 | and return the index of the match, or negative on failure. | |
352 | This does not clobber the match data. */ | |
353 | ||
354 | int | |
355 | fast_string_match (regexp, string) | |
356 | Lisp_Object regexp, string; | |
357 | { | |
358 | int val; | |
487282dc | 359 | struct re_pattern_buffer *bufp; |
e59a8453 | 360 | |
b819a390 | 361 | bufp = compile_pattern (regexp, 0, 0, 0); |
e59a8453 | 362 | immediate_quit = 1; |
487282dc | 363 | val = re_search (bufp, (char *) XSTRING (string)->data, |
e59a8453 RS |
364 | XSTRING (string)->size, 0, XSTRING (string)->size, |
365 | 0); | |
366 | immediate_quit = 0; | |
367 | return val; | |
368 | } | |
ca1d1d23 | 369 | \f |
9169c321 JB |
370 | /* max and min. */ |
371 | ||
372 | static int | |
373 | max (a, b) | |
374 | int a, b; | |
375 | { | |
376 | return ((a > b) ? a : b); | |
377 | } | |
378 | ||
379 | static int | |
380 | min (a, b) | |
381 | int a, b; | |
382 | { | |
383 | return ((a < b) ? a : b); | |
384 | } | |
385 | ||
386 | \f | |
387 | /* The newline cache: remembering which sections of text have no newlines. */ | |
388 | ||
389 | /* If the user has requested newline caching, make sure it's on. | |
390 | Otherwise, make sure it's off. | |
391 | This is our cheezy way of associating an action with the change of | |
392 | state of a buffer-local variable. */ | |
393 | static void | |
394 | newline_cache_on_off (buf) | |
395 | struct buffer *buf; | |
396 | { | |
397 | if (NILP (buf->cache_long_line_scans)) | |
398 | { | |
399 | /* It should be off. */ | |
400 | if (buf->newline_cache) | |
401 | { | |
402 | free_region_cache (buf->newline_cache); | |
403 | buf->newline_cache = 0; | |
404 | } | |
405 | } | |
406 | else | |
407 | { | |
408 | /* It should be on. */ | |
409 | if (buf->newline_cache == 0) | |
410 | buf->newline_cache = new_region_cache (); | |
411 | } | |
412 | } | |
413 | ||
414 | \f | |
415 | /* Search for COUNT instances of the character TARGET between START and END. | |
416 | ||
417 | If COUNT is positive, search forwards; END must be >= START. | |
418 | If COUNT is negative, search backwards for the -COUNTth instance; | |
419 | END must be <= START. | |
420 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
421 | ||
422 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
423 | direction indicated by COUNT. | |
ffd56f97 JB |
424 | |
425 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
5bfe95c9 RS |
426 | position after the COUNTth match. Note that for reverse motion |
427 | this is not the same as the usual convention for Emacs motion commands. | |
ffd56f97 | 428 | |
9169c321 JB |
429 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
430 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 431 | |
087a5f81 RS |
432 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
433 | except when inside redisplay. */ | |
434 | ||
9169c321 JB |
435 | scan_buffer (target, start, end, count, shortage, allow_quit) |
436 | register int target; | |
437 | int start, end; | |
438 | int count; | |
439 | int *shortage; | |
087a5f81 | 440 | int allow_quit; |
ca1d1d23 | 441 | { |
9169c321 JB |
442 | struct region_cache *newline_cache; |
443 | int direction; | |
ffd56f97 | 444 | |
9169c321 JB |
445 | if (count > 0) |
446 | { | |
447 | direction = 1; | |
448 | if (! end) end = ZV; | |
449 | } | |
450 | else | |
451 | { | |
452 | direction = -1; | |
453 | if (! end) end = BEGV; | |
454 | } | |
ffd56f97 | 455 | |
9169c321 JB |
456 | newline_cache_on_off (current_buffer); |
457 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
458 | |
459 | if (shortage != 0) | |
460 | *shortage = 0; | |
461 | ||
087a5f81 | 462 | immediate_quit = allow_quit; |
ca1d1d23 | 463 | |
ffd56f97 | 464 | if (count > 0) |
9169c321 | 465 | while (start != end) |
ca1d1d23 | 466 | { |
9169c321 JB |
467 | /* Our innermost scanning loop is very simple; it doesn't know |
468 | about gaps, buffer ends, or the newline cache. ceiling is | |
469 | the position of the last character before the next such | |
470 | obstacle --- the last character the dumb search loop should | |
471 | examine. */ | |
472 | register int ceiling = end - 1; | |
473 | ||
474 | /* If we're looking for a newline, consult the newline cache | |
475 | to see where we can avoid some scanning. */ | |
476 | if (target == '\n' && newline_cache) | |
477 | { | |
478 | int next_change; | |
479 | immediate_quit = 0; | |
480 | while (region_cache_forward | |
481 | (current_buffer, newline_cache, start, &next_change)) | |
482 | start = next_change; | |
cbe0db0d | 483 | immediate_quit = allow_quit; |
9169c321 JB |
484 | |
485 | /* start should never be after end. */ | |
486 | if (start >= end) | |
487 | start = end - 1; | |
488 | ||
489 | /* Now the text after start is an unknown region, and | |
490 | next_change is the position of the next known region. */ | |
491 | ceiling = min (next_change - 1, ceiling); | |
492 | } | |
493 | ||
494 | /* The dumb loop can only scan text stored in contiguous | |
495 | bytes. BUFFER_CEILING_OF returns the last character | |
496 | position that is contiguous, so the ceiling is the | |
497 | position after that. */ | |
498 | ceiling = min (BUFFER_CEILING_OF (start), ceiling); | |
499 | ||
500 | { | |
501 | /* The termination address of the dumb loop. */ | |
502 | register unsigned char *ceiling_addr = &FETCH_CHAR (ceiling) + 1; | |
503 | register unsigned char *cursor = &FETCH_CHAR (start); | |
504 | unsigned char *base = cursor; | |
505 | ||
506 | while (cursor < ceiling_addr) | |
507 | { | |
508 | unsigned char *scan_start = cursor; | |
509 | ||
510 | /* The dumb loop. */ | |
511 | while (*cursor != target && ++cursor < ceiling_addr) | |
512 | ; | |
513 | ||
514 | /* If we're looking for newlines, cache the fact that | |
515 | the region from start to cursor is free of them. */ | |
516 | if (target == '\n' && newline_cache) | |
517 | know_region_cache (current_buffer, newline_cache, | |
518 | start + scan_start - base, | |
519 | start + cursor - base); | |
520 | ||
521 | /* Did we find the target character? */ | |
522 | if (cursor < ceiling_addr) | |
523 | { | |
524 | if (--count == 0) | |
525 | { | |
526 | immediate_quit = 0; | |
527 | return (start + cursor - base + 1); | |
528 | } | |
529 | cursor++; | |
530 | } | |
531 | } | |
532 | ||
533 | start += cursor - base; | |
534 | } | |
ca1d1d23 JB |
535 | } |
536 | else | |
9169c321 JB |
537 | while (start > end) |
538 | { | |
539 | /* The last character to check before the next obstacle. */ | |
540 | register int ceiling = end; | |
541 | ||
542 | /* Consult the newline cache, if appropriate. */ | |
543 | if (target == '\n' && newline_cache) | |
544 | { | |
545 | int next_change; | |
546 | immediate_quit = 0; | |
547 | while (region_cache_backward | |
548 | (current_buffer, newline_cache, start, &next_change)) | |
549 | start = next_change; | |
cbe0db0d | 550 | immediate_quit = allow_quit; |
9169c321 JB |
551 | |
552 | /* Start should never be at or before end. */ | |
553 | if (start <= end) | |
554 | start = end + 1; | |
555 | ||
556 | /* Now the text before start is an unknown region, and | |
557 | next_change is the position of the next known region. */ | |
558 | ceiling = max (next_change, ceiling); | |
559 | } | |
560 | ||
561 | /* Stop scanning before the gap. */ | |
562 | ceiling = max (BUFFER_FLOOR_OF (start - 1), ceiling); | |
563 | ||
564 | { | |
565 | /* The termination address of the dumb loop. */ | |
566 | register unsigned char *ceiling_addr = &FETCH_CHAR (ceiling); | |
567 | register unsigned char *cursor = &FETCH_CHAR (start - 1); | |
568 | unsigned char *base = cursor; | |
569 | ||
570 | while (cursor >= ceiling_addr) | |
571 | { | |
572 | unsigned char *scan_start = cursor; | |
573 | ||
574 | while (*cursor != target && --cursor >= ceiling_addr) | |
575 | ; | |
576 | ||
577 | /* If we're looking for newlines, cache the fact that | |
578 | the region from after the cursor to start is free of them. */ | |
579 | if (target == '\n' && newline_cache) | |
580 | know_region_cache (current_buffer, newline_cache, | |
581 | start + cursor - base, | |
582 | start + scan_start - base); | |
583 | ||
584 | /* Did we find the target character? */ | |
585 | if (cursor >= ceiling_addr) | |
586 | { | |
587 | if (++count >= 0) | |
588 | { | |
589 | immediate_quit = 0; | |
590 | return (start + cursor - base); | |
591 | } | |
592 | cursor--; | |
593 | } | |
594 | } | |
595 | ||
596 | start += cursor - base; | |
597 | } | |
598 | } | |
599 | ||
ca1d1d23 JB |
600 | immediate_quit = 0; |
601 | if (shortage != 0) | |
ffd56f97 | 602 | *shortage = count * direction; |
9169c321 | 603 | return start; |
ca1d1d23 JB |
604 | } |
605 | ||
63fa018d RS |
606 | int |
607 | find_next_newline_no_quit (from, cnt) | |
608 | register int from, cnt; | |
609 | { | |
9169c321 | 610 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
63fa018d RS |
611 | } |
612 | ||
ca1d1d23 JB |
613 | int |
614 | find_next_newline (from, cnt) | |
615 | register int from, cnt; | |
616 | { | |
9169c321 JB |
617 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 1); |
618 | } | |
619 | ||
620 | ||
621 | /* Like find_next_newline, but returns position before the newline, | |
622 | not after, and only search up to TO. This isn't just | |
623 | find_next_newline (...)-1, because you might hit TO. */ | |
624 | int | |
625 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 626 | int from, to, cnt; |
9169c321 JB |
627 | { |
628 | int shortage; | |
629 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
630 | ||
631 | if (shortage == 0) | |
632 | pos--; | |
633 | ||
634 | return pos; | |
ca1d1d23 JB |
635 | } |
636 | \f | |
c1dc99a1 JB |
637 | Lisp_Object skip_chars (); |
638 | ||
ca1d1d23 | 639 | DEFUN ("skip-chars-forward", Fskip_chars_forward, Sskip_chars_forward, 1, 2, 0, |
3acb9a69 RS |
640 | "Move point forward, stopping before a char not in STRING, or at pos LIM.\n\ |
641 | STRING is like the inside of a `[...]' in a regular expression\n\ | |
ca1d1d23 JB |
642 | except that `]' is never special and `\\' quotes `^', `-' or `\\'.\n\ |
643 | Thus, with arg \"a-zA-Z\", this skips letters stopping before first nonletter.\n\ | |
c1dc99a1 JB |
644 | With arg \"^a-zA-Z\", skips nonletters stopping before first letter.\n\ |
645 | Returns the distance traveled, either zero or positive.") | |
ca1d1d23 JB |
646 | (string, lim) |
647 | Lisp_Object string, lim; | |
648 | { | |
17431c60 | 649 | return skip_chars (1, 0, string, lim); |
ca1d1d23 JB |
650 | } |
651 | ||
652 | DEFUN ("skip-chars-backward", Fskip_chars_backward, Sskip_chars_backward, 1, 2, 0, | |
3acb9a69 | 653 | "Move point backward, stopping after a char not in STRING, or at pos LIM.\n\ |
c1dc99a1 JB |
654 | See `skip-chars-forward' for details.\n\ |
655 | Returns the distance traveled, either zero or negative.") | |
ca1d1d23 JB |
656 | (string, lim) |
657 | Lisp_Object string, lim; | |
658 | { | |
17431c60 RS |
659 | return skip_chars (0, 0, string, lim); |
660 | } | |
661 | ||
662 | DEFUN ("skip-syntax-forward", Fskip_syntax_forward, Sskip_syntax_forward, 1, 2, 0, | |
663 | "Move point forward across chars in specified syntax classes.\n\ | |
664 | SYNTAX is a string of syntax code characters.\n\ | |
665 | Stop before a char whose syntax is not in SYNTAX, or at position LIM.\n\ | |
666 | If SYNTAX starts with ^, skip characters whose syntax is NOT in SYNTAX.\n\ | |
667 | This function returns the distance traveled, either zero or positive.") | |
668 | (syntax, lim) | |
669 | Lisp_Object syntax, lim; | |
670 | { | |
671 | return skip_chars (1, 1, syntax, lim); | |
672 | } | |
673 | ||
674 | DEFUN ("skip-syntax-backward", Fskip_syntax_backward, Sskip_syntax_backward, 1, 2, 0, | |
675 | "Move point backward across chars in specified syntax classes.\n\ | |
676 | SYNTAX is a string of syntax code characters.\n\ | |
677 | Stop on reaching a char whose syntax is not in SYNTAX, or at position LIM.\n\ | |
678 | If SYNTAX starts with ^, skip characters whose syntax is NOT in SYNTAX.\n\ | |
679 | This function returns the distance traveled, either zero or negative.") | |
680 | (syntax, lim) | |
681 | Lisp_Object syntax, lim; | |
682 | { | |
683 | return skip_chars (0, 1, syntax, lim); | |
ca1d1d23 JB |
684 | } |
685 | ||
c1dc99a1 | 686 | Lisp_Object |
17431c60 RS |
687 | skip_chars (forwardp, syntaxp, string, lim) |
688 | int forwardp, syntaxp; | |
ca1d1d23 JB |
689 | Lisp_Object string, lim; |
690 | { | |
691 | register unsigned char *p, *pend; | |
692 | register unsigned char c; | |
693 | unsigned char fastmap[0400]; | |
694 | int negate = 0; | |
695 | register int i; | |
696 | ||
697 | CHECK_STRING (string, 0); | |
698 | ||
699 | if (NILP (lim)) | |
a3668d92 | 700 | XSETINT (lim, forwardp ? ZV : BEGV); |
ca1d1d23 JB |
701 | else |
702 | CHECK_NUMBER_COERCE_MARKER (lim, 1); | |
703 | ||
ca1d1d23 | 704 | /* In any case, don't allow scan outside bounds of buffer. */ |
c5241910 RS |
705 | /* jla turned this off, for no known reason. |
706 | bfox turned the ZV part on, and rms turned the | |
707 | BEGV part back on. */ | |
708 | if (XINT (lim) > ZV) | |
c235cce7 | 709 | XSETFASTINT (lim, ZV); |
c5241910 | 710 | if (XINT (lim) < BEGV) |
c235cce7 | 711 | XSETFASTINT (lim, BEGV); |
ca1d1d23 JB |
712 | |
713 | p = XSTRING (string)->data; | |
714 | pend = p + XSTRING (string)->size; | |
715 | bzero (fastmap, sizeof fastmap); | |
716 | ||
717 | if (p != pend && *p == '^') | |
718 | { | |
719 | negate = 1; p++; | |
720 | } | |
721 | ||
17431c60 RS |
722 | /* Find the characters specified and set their elements of fastmap. |
723 | If syntaxp, each character counts as itself. | |
724 | Otherwise, handle backslashes and ranges specially */ | |
ca1d1d23 JB |
725 | |
726 | while (p != pend) | |
727 | { | |
728 | c = *p++; | |
17431c60 RS |
729 | if (syntaxp) |
730 | fastmap[c] = 1; | |
731 | else | |
ca1d1d23 | 732 | { |
17431c60 | 733 | if (c == '\\') |
ca1d1d23 | 734 | { |
17431c60 RS |
735 | if (p == pend) break; |
736 | c = *p++; | |
737 | } | |
738 | if (p != pend && *p == '-') | |
739 | { | |
740 | p++; | |
741 | if (p == pend) break; | |
742 | while (c <= *p) | |
743 | { | |
744 | fastmap[c] = 1; | |
745 | c++; | |
746 | } | |
747 | p++; | |
ca1d1d23 | 748 | } |
17431c60 RS |
749 | else |
750 | fastmap[c] = 1; | |
ca1d1d23 | 751 | } |
ca1d1d23 JB |
752 | } |
753 | ||
9239c6c1 RS |
754 | if (syntaxp && fastmap['-'] != 0) |
755 | fastmap[' '] = 1; | |
756 | ||
ca1d1d23 JB |
757 | /* If ^ was the first character, complement the fastmap. */ |
758 | ||
759 | if (negate) | |
760 | for (i = 0; i < sizeof fastmap; i++) | |
761 | fastmap[i] ^= 1; | |
762 | ||
c1dc99a1 JB |
763 | { |
764 | int start_point = point; | |
765 | ||
766 | immediate_quit = 1; | |
17431c60 | 767 | if (syntaxp) |
c1dc99a1 | 768 | { |
17431c60 RS |
769 | |
770 | if (forwardp) | |
771 | { | |
772 | while (point < XINT (lim) | |
773 | && fastmap[(unsigned char) syntax_code_spec[(int) SYNTAX (FETCH_CHAR (point))]]) | |
774 | SET_PT (point + 1); | |
775 | } | |
776 | else | |
777 | { | |
778 | while (point > XINT (lim) | |
779 | && fastmap[(unsigned char) syntax_code_spec[(int) SYNTAX (FETCH_CHAR (point - 1))]]) | |
780 | SET_PT (point - 1); | |
781 | } | |
c1dc99a1 JB |
782 | } |
783 | else | |
784 | { | |
17431c60 RS |
785 | if (forwardp) |
786 | { | |
787 | while (point < XINT (lim) && fastmap[FETCH_CHAR (point)]) | |
788 | SET_PT (point + 1); | |
789 | } | |
790 | else | |
791 | { | |
792 | while (point > XINT (lim) && fastmap[FETCH_CHAR (point - 1)]) | |
793 | SET_PT (point - 1); | |
794 | } | |
c1dc99a1 JB |
795 | } |
796 | immediate_quit = 0; | |
797 | ||
798 | return make_number (point - start_point); | |
799 | } | |
ca1d1d23 JB |
800 | } |
801 | \f | |
802 | /* Subroutines of Lisp buffer search functions. */ | |
803 | ||
804 | static Lisp_Object | |
b819a390 | 805 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
806 | Lisp_Object string, bound, noerror, count; |
807 | int direction; | |
808 | int RE; | |
b819a390 | 809 | int posix; |
ca1d1d23 JB |
810 | { |
811 | register int np; | |
812 | int lim; | |
813 | int n = direction; | |
814 | ||
815 | if (!NILP (count)) | |
816 | { | |
817 | CHECK_NUMBER (count, 3); | |
818 | n *= XINT (count); | |
819 | } | |
820 | ||
821 | CHECK_STRING (string, 0); | |
822 | if (NILP (bound)) | |
823 | lim = n > 0 ? ZV : BEGV; | |
824 | else | |
825 | { | |
826 | CHECK_NUMBER_COERCE_MARKER (bound, 1); | |
827 | lim = XINT (bound); | |
828 | if (n > 0 ? lim < point : lim > point) | |
829 | error ("Invalid search bound (wrong side of point)"); | |
830 | if (lim > ZV) | |
831 | lim = ZV; | |
832 | if (lim < BEGV) | |
833 | lim = BEGV; | |
834 | } | |
835 | ||
836 | np = search_buffer (string, point, lim, n, RE, | |
837 | (!NILP (current_buffer->case_fold_search) | |
838 | ? XSTRING (current_buffer->case_canon_table)->data : 0), | |
839 | (!NILP (current_buffer->case_fold_search) | |
b819a390 RS |
840 | ? XSTRING (current_buffer->case_eqv_table)->data : 0), |
841 | posix); | |
ca1d1d23 JB |
842 | if (np <= 0) |
843 | { | |
844 | if (NILP (noerror)) | |
845 | return signal_failure (string); | |
846 | if (!EQ (noerror, Qt)) | |
847 | { | |
848 | if (lim < BEGV || lim > ZV) | |
849 | abort (); | |
a5f217b8 RS |
850 | SET_PT (lim); |
851 | return Qnil; | |
852 | #if 0 /* This would be clean, but maybe programs depend on | |
853 | a value of nil here. */ | |
481399bf | 854 | np = lim; |
a5f217b8 | 855 | #endif |
ca1d1d23 | 856 | } |
481399bf RS |
857 | else |
858 | return Qnil; | |
ca1d1d23 JB |
859 | } |
860 | ||
861 | if (np < BEGV || np > ZV) | |
862 | abort (); | |
863 | ||
864 | SET_PT (np); | |
865 | ||
866 | return make_number (np); | |
867 | } | |
868 | \f | |
b6d6a51c KH |
869 | static int |
870 | trivial_regexp_p (regexp) | |
871 | Lisp_Object regexp; | |
872 | { | |
873 | int len = XSTRING (regexp)->size; | |
874 | unsigned char *s = XSTRING (regexp)->data; | |
875 | unsigned char c; | |
876 | while (--len >= 0) | |
877 | { | |
878 | switch (*s++) | |
879 | { | |
880 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
881 | return 0; | |
882 | case '\\': | |
883 | if (--len < 0) | |
884 | return 0; | |
885 | switch (*s++) | |
886 | { | |
887 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
888 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
889 | case 'S': case '1': case '2': case '3': case '4': case '5': | |
890 | case '6': case '7': case '8': case '9': | |
891 | return 0; | |
892 | } | |
893 | } | |
894 | } | |
895 | return 1; | |
896 | } | |
897 | ||
ca325161 | 898 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 899 | starting at position POS and stopping at position LIM, |
b819a390 | 900 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
901 | a regular expression if RE is true. |
902 | ||
903 | If N is positive, searching is forward and LIM must be greater than POS. | |
904 | If N is negative, searching is backward and LIM must be less than POS. | |
905 | ||
906 | Returns -x if only N-x occurrences found (x > 0), | |
907 | or else the position at the beginning of the Nth occurrence | |
b819a390 RS |
908 | (if searching backward) or the end (if searching forward). |
909 | ||
910 | POSIX is nonzero if we want full backtracking (POSIX style) | |
911 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 912 | |
b819a390 RS |
913 | static int |
914 | search_buffer (string, pos, lim, n, RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
915 | Lisp_Object string; |
916 | int pos; | |
917 | int lim; | |
918 | int n; | |
919 | int RE; | |
920 | register unsigned char *trt; | |
921 | register unsigned char *inverse_trt; | |
b819a390 | 922 | int posix; |
ca1d1d23 JB |
923 | { |
924 | int len = XSTRING (string)->size; | |
925 | unsigned char *base_pat = XSTRING (string)->data; | |
926 | register int *BM_tab; | |
927 | int *BM_tab_base; | |
928 | register int direction = ((n > 0) ? 1 : -1); | |
929 | register int dirlen; | |
930 | int infinity, limit, k, stride_for_teases; | |
931 | register unsigned char *pat, *cursor, *p_limit; | |
932 | register int i, j; | |
933 | unsigned char *p1, *p2; | |
934 | int s1, s2; | |
935 | ||
7074fde6 FP |
936 | if (running_asynch_code) |
937 | save_search_regs (); | |
938 | ||
ca1d1d23 | 939 | /* Null string is found at starting position. */ |
3f57a499 | 940 | if (len == 0) |
ca325161 RS |
941 | { |
942 | set_search_regs (pos, 0); | |
943 | return pos; | |
944 | } | |
3f57a499 RS |
945 | |
946 | /* Searching 0 times means don't move. */ | |
947 | if (n == 0) | |
ca1d1d23 JB |
948 | return pos; |
949 | ||
b6d6a51c | 950 | if (RE && !trivial_regexp_p (string)) |
ca1d1d23 | 951 | { |
487282dc KH |
952 | struct re_pattern_buffer *bufp; |
953 | ||
b819a390 | 954 | bufp = compile_pattern (string, &search_regs, (char *) trt, posix); |
ca1d1d23 | 955 | |
ca1d1d23 JB |
956 | immediate_quit = 1; /* Quit immediately if user types ^G, |
957 | because letting this function finish | |
958 | can take too long. */ | |
959 | QUIT; /* Do a pending quit right away, | |
960 | to avoid paradoxical behavior */ | |
961 | /* Get pointers and sizes of the two strings | |
962 | that make up the visible portion of the buffer. */ | |
963 | ||
964 | p1 = BEGV_ADDR; | |
965 | s1 = GPT - BEGV; | |
966 | p2 = GAP_END_ADDR; | |
967 | s2 = ZV - GPT; | |
968 | if (s1 < 0) | |
969 | { | |
970 | p2 = p1; | |
971 | s2 = ZV - BEGV; | |
972 | s1 = 0; | |
973 | } | |
974 | if (s2 < 0) | |
975 | { | |
976 | s1 = ZV - BEGV; | |
977 | s2 = 0; | |
978 | } | |
979 | while (n < 0) | |
980 | { | |
42db823b | 981 | int val; |
487282dc | 982 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
42db823b RS |
983 | pos - BEGV, lim - pos, &search_regs, |
984 | /* Don't allow match past current point */ | |
985 | pos - BEGV); | |
ca1d1d23 | 986 | if (val == -2) |
b6d6a51c KH |
987 | { |
988 | matcher_overflow (); | |
989 | } | |
ca1d1d23 JB |
990 | if (val >= 0) |
991 | { | |
992 | j = BEGV; | |
4746118a | 993 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
994 | if (search_regs.start[i] >= 0) |
995 | { | |
996 | search_regs.start[i] += j; | |
997 | search_regs.end[i] += j; | |
998 | } | |
a3668d92 | 999 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1000 | /* Set pos to the new position. */ |
1001 | pos = search_regs.start[0]; | |
1002 | } | |
1003 | else | |
1004 | { | |
1005 | immediate_quit = 0; | |
1006 | return (n); | |
1007 | } | |
1008 | n++; | |
1009 | } | |
1010 | while (n > 0) | |
1011 | { | |
42db823b | 1012 | int val; |
487282dc | 1013 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
42db823b RS |
1014 | pos - BEGV, lim - pos, &search_regs, |
1015 | lim - BEGV); | |
ca1d1d23 | 1016 | if (val == -2) |
b6d6a51c KH |
1017 | { |
1018 | matcher_overflow (); | |
1019 | } | |
ca1d1d23 JB |
1020 | if (val >= 0) |
1021 | { | |
1022 | j = BEGV; | |
4746118a | 1023 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1024 | if (search_regs.start[i] >= 0) |
1025 | { | |
1026 | search_regs.start[i] += j; | |
1027 | search_regs.end[i] += j; | |
1028 | } | |
a3668d92 | 1029 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1030 | pos = search_regs.end[0]; |
1031 | } | |
1032 | else | |
1033 | { | |
1034 | immediate_quit = 0; | |
1035 | return (0 - n); | |
1036 | } | |
1037 | n--; | |
1038 | } | |
1039 | immediate_quit = 0; | |
1040 | return (pos); | |
1041 | } | |
1042 | else /* non-RE case */ | |
1043 | { | |
1044 | #ifdef C_ALLOCA | |
1045 | int BM_tab_space[0400]; | |
1046 | BM_tab = &BM_tab_space[0]; | |
1047 | #else | |
1048 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
1049 | #endif | |
b6d6a51c KH |
1050 | { |
1051 | unsigned char *patbuf = (unsigned char *) alloca (len); | |
1052 | pat = patbuf; | |
1053 | while (--len >= 0) | |
1054 | { | |
1055 | /* If we got here and the RE flag is set, it's because we're | |
1056 | dealing with a regexp known to be trivial, so the backslash | |
1057 | just quotes the next character. */ | |
1058 | if (RE && *base_pat == '\\') | |
1059 | { | |
1060 | len--; | |
1061 | base_pat++; | |
1062 | } | |
1063 | *pat++ = (trt ? trt[*base_pat++] : *base_pat++); | |
1064 | } | |
1065 | len = pat - patbuf; | |
1066 | pat = base_pat = patbuf; | |
1067 | } | |
ca1d1d23 JB |
1068 | /* The general approach is that we are going to maintain that we know */ |
1069 | /* the first (closest to the present position, in whatever direction */ | |
1070 | /* we're searching) character that could possibly be the last */ | |
1071 | /* (furthest from present position) character of a valid match. We */ | |
1072 | /* advance the state of our knowledge by looking at that character */ | |
1073 | /* and seeing whether it indeed matches the last character of the */ | |
1074 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1075 | /* move our pointer (to putative last characters) as far as is */ | |
1076 | /* logically possible. This amount of movement, which I call a */ | |
1077 | /* stride, will be the length of the pattern if the actual character */ | |
1078 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1079 | /* from the last occurrence of that character to the end of the */ | |
1080 | /* pattern. */ | |
1081 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1082 | /* characters that match the last character. This allows use of only */ | |
1083 | /* a single test, a test for having gone past the end of the */ | |
1084 | /* permissible match region, to test for both possible matches (when */ | |
1085 | /* the stride goes past the end immediately) and failure to */ | |
1086 | /* match (where you get nudged past the end one stride at a time). */ | |
1087 | ||
1088 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1089 | /* is determined only by the last character of the putative match. */ | |
1090 | /* If that character does not match, we will stride the proper */ | |
1091 | /* distance to propose a match that superimposes it on the last */ | |
1092 | /* instance of a character that matches it (per trt), or misses */ | |
1093 | /* it entirely if there is none. */ | |
1094 | ||
1095 | dirlen = len * direction; | |
1096 | infinity = dirlen - (lim + pos + len + len) * direction; | |
1097 | if (direction < 0) | |
1098 | pat = (base_pat += len - 1); | |
1099 | BM_tab_base = BM_tab; | |
1100 | BM_tab += 0400; | |
1101 | j = dirlen; /* to get it in a register */ | |
1102 | /* A character that does not appear in the pattern induces a */ | |
1103 | /* stride equal to the pattern length. */ | |
1104 | while (BM_tab_base != BM_tab) | |
1105 | { | |
1106 | *--BM_tab = j; | |
1107 | *--BM_tab = j; | |
1108 | *--BM_tab = j; | |
1109 | *--BM_tab = j; | |
1110 | } | |
1111 | i = 0; | |
1112 | while (i != infinity) | |
1113 | { | |
1114 | j = pat[i]; i += direction; | |
1115 | if (i == dirlen) i = infinity; | |
1116 | if ((int) trt) | |
1117 | { | |
1118 | k = (j = trt[j]); | |
1119 | if (i == infinity) | |
1120 | stride_for_teases = BM_tab[j]; | |
1121 | BM_tab[j] = dirlen - i; | |
1122 | /* A translation table is accompanied by its inverse -- see */ | |
1123 | /* comment following downcase_table for details */ | |
1124 | while ((j = inverse_trt[j]) != k) | |
1125 | BM_tab[j] = dirlen - i; | |
1126 | } | |
1127 | else | |
1128 | { | |
1129 | if (i == infinity) | |
1130 | stride_for_teases = BM_tab[j]; | |
1131 | BM_tab[j] = dirlen - i; | |
1132 | } | |
1133 | /* stride_for_teases tells how much to stride if we get a */ | |
1134 | /* match on the far character but are subsequently */ | |
1135 | /* disappointed, by recording what the stride would have been */ | |
1136 | /* for that character if the last character had been */ | |
1137 | /* different. */ | |
1138 | } | |
1139 | infinity = dirlen - infinity; | |
1140 | pos += dirlen - ((direction > 0) ? direction : 0); | |
1141 | /* loop invariant - pos points at where last char (first char if reverse) | |
1142 | of pattern would align in a possible match. */ | |
1143 | while (n != 0) | |
1144 | { | |
b2c71fb4 KH |
1145 | /* It's been reported that some (broken) compiler thinks that |
1146 | Boolean expressions in an arithmetic context are unsigned. | |
1147 | Using an explicit ?1:0 prevents this. */ | |
1148 | if ((lim - pos - ((direction > 0) ? 1 : 0)) * direction < 0) | |
ca1d1d23 JB |
1149 | return (n * (0 - direction)); |
1150 | /* First we do the part we can by pointers (maybe nothing) */ | |
1151 | QUIT; | |
1152 | pat = base_pat; | |
1153 | limit = pos - dirlen + direction; | |
1154 | limit = ((direction > 0) | |
1155 | ? BUFFER_CEILING_OF (limit) | |
1156 | : BUFFER_FLOOR_OF (limit)); | |
1157 | /* LIMIT is now the last (not beyond-last!) value | |
1158 | POS can take on without hitting edge of buffer or the gap. */ | |
1159 | limit = ((direction > 0) | |
1160 | ? min (lim - 1, min (limit, pos + 20000)) | |
1161 | : max (lim, max (limit, pos - 20000))); | |
1162 | if ((limit - pos) * direction > 20) | |
1163 | { | |
1164 | p_limit = &FETCH_CHAR (limit); | |
1165 | p2 = (cursor = &FETCH_CHAR (pos)); | |
1166 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ | |
1167 | while (1) /* use one cursor setting as long as i can */ | |
1168 | { | |
1169 | if (direction > 0) /* worth duplicating */ | |
1170 | { | |
1171 | /* Use signed comparison if appropriate | |
1172 | to make cursor+infinity sure to be > p_limit. | |
1173 | Assuming that the buffer lies in a range of addresses | |
1174 | that are all "positive" (as ints) or all "negative", | |
1175 | either kind of comparison will work as long | |
1176 | as we don't step by infinity. So pick the kind | |
1177 | that works when we do step by infinity. */ | |
1178 | if ((int) (p_limit + infinity) > (int) p_limit) | |
1179 | while ((int) cursor <= (int) p_limit) | |
1180 | cursor += BM_tab[*cursor]; | |
1181 | else | |
1182 | while ((unsigned int) cursor <= (unsigned int) p_limit) | |
1183 | cursor += BM_tab[*cursor]; | |
1184 | } | |
1185 | else | |
1186 | { | |
1187 | if ((int) (p_limit + infinity) < (int) p_limit) | |
1188 | while ((int) cursor >= (int) p_limit) | |
1189 | cursor += BM_tab[*cursor]; | |
1190 | else | |
1191 | while ((unsigned int) cursor >= (unsigned int) p_limit) | |
1192 | cursor += BM_tab[*cursor]; | |
1193 | } | |
1194 | /* If you are here, cursor is beyond the end of the searched region. */ | |
1195 | /* This can happen if you match on the far character of the pattern, */ | |
1196 | /* because the "stride" of that character is infinity, a number able */ | |
1197 | /* to throw you well beyond the end of the search. It can also */ | |
1198 | /* happen if you fail to match within the permitted region and would */ | |
1199 | /* otherwise try a character beyond that region */ | |
1200 | if ((cursor - p_limit) * direction <= len) | |
1201 | break; /* a small overrun is genuine */ | |
1202 | cursor -= infinity; /* large overrun = hit */ | |
1203 | i = dirlen - direction; | |
1204 | if ((int) trt) | |
1205 | { | |
1206 | while ((i -= direction) + direction != 0) | |
1207 | if (pat[i] != trt[*(cursor -= direction)]) | |
1208 | break; | |
1209 | } | |
1210 | else | |
1211 | { | |
1212 | while ((i -= direction) + direction != 0) | |
1213 | if (pat[i] != *(cursor -= direction)) | |
1214 | break; | |
1215 | } | |
1216 | cursor += dirlen - i - direction; /* fix cursor */ | |
1217 | if (i + direction == 0) | |
1218 | { | |
1219 | cursor -= direction; | |
1113d9db | 1220 | |
ca325161 RS |
1221 | set_search_regs (pos + cursor - p2 + ((direction > 0) |
1222 | ? 1 - len : 0), | |
1223 | len); | |
1224 | ||
ca1d1d23 JB |
1225 | if ((n -= direction) != 0) |
1226 | cursor += dirlen; /* to resume search */ | |
1227 | else | |
1228 | return ((direction > 0) | |
1229 | ? search_regs.end[0] : search_regs.start[0]); | |
1230 | } | |
1231 | else | |
1232 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
1233 | } | |
1234 | pos += cursor - p2; | |
1235 | } | |
1236 | else | |
1237 | /* Now we'll pick up a clump that has to be done the hard */ | |
1238 | /* way because it covers a discontinuity */ | |
1239 | { | |
1240 | limit = ((direction > 0) | |
1241 | ? BUFFER_CEILING_OF (pos - dirlen + 1) | |
1242 | : BUFFER_FLOOR_OF (pos - dirlen - 1)); | |
1243 | limit = ((direction > 0) | |
1244 | ? min (limit + len, lim - 1) | |
1245 | : max (limit - len, lim)); | |
1246 | /* LIMIT is now the last value POS can have | |
1247 | and still be valid for a possible match. */ | |
1248 | while (1) | |
1249 | { | |
1250 | /* This loop can be coded for space rather than */ | |
1251 | /* speed because it will usually run only once. */ | |
1252 | /* (the reach is at most len + 21, and typically */ | |
1253 | /* does not exceed len) */ | |
1254 | while ((limit - pos) * direction >= 0) | |
1255 | pos += BM_tab[FETCH_CHAR(pos)]; | |
1256 | /* now run the same tests to distinguish going off the */ | |
eb8c3be9 | 1257 | /* end, a match or a phony match. */ |
ca1d1d23 JB |
1258 | if ((pos - limit) * direction <= len) |
1259 | break; /* ran off the end */ | |
1260 | /* Found what might be a match. | |
1261 | Set POS back to last (first if reverse) char pos. */ | |
1262 | pos -= infinity; | |
1263 | i = dirlen - direction; | |
1264 | while ((i -= direction) + direction != 0) | |
1265 | { | |
1266 | pos -= direction; | |
1267 | if (pat[i] != (((int) trt) | |
1268 | ? trt[FETCH_CHAR(pos)] | |
1269 | : FETCH_CHAR (pos))) | |
1270 | break; | |
1271 | } | |
1272 | /* Above loop has moved POS part or all the way | |
1273 | back to the first char pos (last char pos if reverse). | |
1274 | Set it once again at the last (first if reverse) char. */ | |
1275 | pos += dirlen - i- direction; | |
1276 | if (i + direction == 0) | |
1277 | { | |
1278 | pos -= direction; | |
1113d9db | 1279 | |
ca325161 RS |
1280 | set_search_regs (pos + ((direction > 0) ? 1 - len : 0), |
1281 | len); | |
1282 | ||
ca1d1d23 JB |
1283 | if ((n -= direction) != 0) |
1284 | pos += dirlen; /* to resume search */ | |
1285 | else | |
1286 | return ((direction > 0) | |
1287 | ? search_regs.end[0] : search_regs.start[0]); | |
1288 | } | |
1289 | else | |
1290 | pos += stride_for_teases; | |
1291 | } | |
1292 | } | |
1293 | /* We have done one clump. Can we continue? */ | |
1294 | if ((lim - pos) * direction < 0) | |
1295 | return ((0 - n) * direction); | |
1296 | } | |
1297 | return pos; | |
1298 | } | |
1299 | } | |
ca325161 RS |
1300 | |
1301 | /* Record beginning BEG and end BEG + LEN | |
1302 | for a match just found in the current buffer. */ | |
1303 | ||
1304 | static void | |
1305 | set_search_regs (beg, len) | |
1306 | int beg, len; | |
1307 | { | |
1308 | /* Make sure we have registers in which to store | |
1309 | the match position. */ | |
1310 | if (search_regs.num_regs == 0) | |
1311 | { | |
1312 | regoff_t *starts, *ends; | |
1313 | ||
1314 | starts = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
1315 | ends = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 1316 | search_regs.num_regs = 2; |
ca325161 RS |
1317 | } |
1318 | ||
1319 | search_regs.start[0] = beg; | |
1320 | search_regs.end[0] = beg + len; | |
a3668d92 | 1321 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 1322 | } |
ca1d1d23 JB |
1323 | \f |
1324 | /* Given a string of words separated by word delimiters, | |
1325 | compute a regexp that matches those exact words | |
1326 | separated by arbitrary punctuation. */ | |
1327 | ||
1328 | static Lisp_Object | |
1329 | wordify (string) | |
1330 | Lisp_Object string; | |
1331 | { | |
1332 | register unsigned char *p, *o; | |
1333 | register int i, len, punct_count = 0, word_count = 0; | |
1334 | Lisp_Object val; | |
1335 | ||
1336 | CHECK_STRING (string, 0); | |
1337 | p = XSTRING (string)->data; | |
1338 | len = XSTRING (string)->size; | |
1339 | ||
1340 | for (i = 0; i < len; i++) | |
1341 | if (SYNTAX (p[i]) != Sword) | |
1342 | { | |
1343 | punct_count++; | |
1344 | if (i > 0 && SYNTAX (p[i-1]) == Sword) word_count++; | |
1345 | } | |
1346 | if (SYNTAX (p[len-1]) == Sword) word_count++; | |
1347 | if (!word_count) return build_string (""); | |
1348 | ||
1349 | val = make_string (p, len - punct_count + 5 * (word_count - 1) + 4); | |
1350 | ||
1351 | o = XSTRING (val)->data; | |
1352 | *o++ = '\\'; | |
1353 | *o++ = 'b'; | |
1354 | ||
1355 | for (i = 0; i < len; i++) | |
1356 | if (SYNTAX (p[i]) == Sword) | |
1357 | *o++ = p[i]; | |
1358 | else if (i > 0 && SYNTAX (p[i-1]) == Sword && --word_count) | |
1359 | { | |
1360 | *o++ = '\\'; | |
1361 | *o++ = 'W'; | |
1362 | *o++ = '\\'; | |
1363 | *o++ = 'W'; | |
1364 | *o++ = '*'; | |
1365 | } | |
1366 | ||
1367 | *o++ = '\\'; | |
1368 | *o++ = 'b'; | |
1369 | ||
1370 | return val; | |
1371 | } | |
1372 | \f | |
1373 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
1374 | "sSearch backward: ", | |
1375 | "Search backward from point for STRING.\n\ | |
1376 | Set point to the beginning of the occurrence found, and return point.\n\ | |
1377 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1378 | The match found must not extend before that position.\n\ | |
1379 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1380 | If not nil and not t, position at limit of search and return nil.\n\ | |
1381 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1382 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1383 | (string, bound, noerror, count) | |
1384 | Lisp_Object string, bound, noerror, count; | |
1385 | { | |
b819a390 | 1386 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
1387 | } |
1388 | ||
1389 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "sSearch: ", | |
1390 | "Search forward from point for STRING.\n\ | |
1391 | Set point to the end of the occurrence found, and return point.\n\ | |
1392 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1393 | The match found must not extend after that position. nil is equivalent\n\ | |
1394 | to (point-max).\n\ | |
1395 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1396 | If not nil and not t, move to limit of search and return nil.\n\ | |
1397 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1398 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1399 | (string, bound, noerror, count) | |
1400 | Lisp_Object string, bound, noerror, count; | |
1401 | { | |
b819a390 | 1402 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
1403 | } |
1404 | ||
1405 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
1406 | "sWord search backward: ", | |
1407 | "Search backward from point for STRING, ignoring differences in punctuation.\n\ | |
1408 | Set point to the beginning of the occurrence found, and return point.\n\ | |
1409 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1410 | The match found must not extend before that position.\n\ | |
1411 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1412 | If not nil and not t, move to limit of search and return nil.\n\ | |
1413 | Optional fourth argument is repeat count--search for successive occurrences.") | |
1414 | (string, bound, noerror, count) | |
1415 | Lisp_Object string, bound, noerror, count; | |
1416 | { | |
b819a390 | 1417 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
1418 | } |
1419 | ||
1420 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
1421 | "sWord search: ", | |
1422 | "Search forward from point for STRING, ignoring differences in punctuation.\n\ | |
1423 | Set point to the end of the occurrence found, and return point.\n\ | |
1424 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1425 | The match found must not extend after that position.\n\ | |
1426 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1427 | If not nil and not t, move to limit of search and return nil.\n\ | |
1428 | Optional fourth argument is repeat count--search for successive occurrences.") | |
1429 | (string, bound, noerror, count) | |
1430 | Lisp_Object string, bound, noerror, count; | |
1431 | { | |
b819a390 | 1432 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
1433 | } |
1434 | ||
1435 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
1436 | "sRE search backward: ", | |
1437 | "Search backward from point for match for regular expression REGEXP.\n\ | |
1438 | Set point to the beginning of the match, and return point.\n\ | |
1439 | The match found is the one starting last in the buffer\n\ | |
19c0a730 | 1440 | and yet ending before the origin of the search.\n\ |
ca1d1d23 JB |
1441 | An optional second argument bounds the search; it is a buffer position.\n\ |
1442 | The match found must start at or after that position.\n\ | |
1443 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1444 | If not nil and not t, move to limit of search and return nil.\n\ | |
1445 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1446 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
19c0a730 KH |
1447 | (regexp, bound, noerror, count) |
1448 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 1449 | { |
b819a390 | 1450 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
1451 | } |
1452 | ||
1453 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
1454 | "sRE search: ", | |
1455 | "Search forward from point for regular expression REGEXP.\n\ | |
1456 | Set point to the end of the occurrence found, and return point.\n\ | |
1457 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1458 | The match found must not extend after that position.\n\ | |
1459 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1460 | If not nil and not t, move to limit of search and return nil.\n\ | |
1461 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1462 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
19c0a730 KH |
1463 | (regexp, bound, noerror, count) |
1464 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 1465 | { |
b819a390 RS |
1466 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
1467 | } | |
1468 | ||
1469 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
1470 | "sPosix search backward: ", | |
1471 | "Search backward from point for match for regular expression REGEXP.\n\ | |
1472 | Find the longest match in accord with Posix regular expression rules.\n\ | |
1473 | Set point to the beginning of the match, and return point.\n\ | |
1474 | The match found is the one starting last in the buffer\n\ | |
1475 | and yet ending before the origin of the search.\n\ | |
1476 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1477 | The match found must start at or after that position.\n\ | |
1478 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1479 | If not nil and not t, move to limit of search and return nil.\n\ | |
1480 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1481 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1482 | (regexp, bound, noerror, count) | |
1483 | Lisp_Object regexp, bound, noerror, count; | |
1484 | { | |
1485 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
1486 | } | |
1487 | ||
1488 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
1489 | "sPosix search: ", | |
1490 | "Search forward from point for regular expression REGEXP.\n\ | |
1491 | Find the longest match in accord with Posix regular expression rules.\n\ | |
1492 | Set point to the end of the occurrence found, and return point.\n\ | |
1493 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1494 | The match found must not extend after that position.\n\ | |
1495 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1496 | If not nil and not t, move to limit of search and return nil.\n\ | |
1497 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1498 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1499 | (regexp, bound, noerror, count) | |
1500 | Lisp_Object regexp, bound, noerror, count; | |
1501 | { | |
1502 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
1503 | } |
1504 | \f | |
080c45fd | 1505 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 4, 0, |
ca1d1d23 JB |
1506 | "Replace text matched by last search with NEWTEXT.\n\ |
1507 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text.\n\ | |
5b9cf4b2 RS |
1508 | Otherwise maybe capitalize the whole text, or maybe just word initials,\n\ |
1509 | based on the replaced text.\n\ | |
1510 | If the replaced text has only capital letters\n\ | |
1511 | and has at least one multiletter word, convert NEWTEXT to all caps.\n\ | |
1512 | If the replaced text has at least one word starting with a capital letter,\n\ | |
1513 | then capitalize each word in NEWTEXT.\n\n\ | |
ca1d1d23 JB |
1514 | If third arg LITERAL is non-nil, insert NEWTEXT literally.\n\ |
1515 | Otherwise treat `\\' as special:\n\ | |
1516 | `\\&' in NEWTEXT means substitute original matched text.\n\ | |
1517 | `\\N' means substitute what matched the Nth `\\(...\\)'.\n\ | |
1518 | If Nth parens didn't match, substitute nothing.\n\ | |
1519 | `\\\\' means insert one `\\'.\n\ | |
1113d9db | 1520 | FIXEDCASE and LITERAL are optional arguments.\n\ |
080c45fd RS |
1521 | Leaves point at end of replacement text.\n\ |
1522 | \n\ | |
1523 | The optional fourth argument STRING can be a string to modify.\n\ | |
1524 | In that case, this function creates and returns a new string\n\ | |
1525 | which is made by replacing the part of STRING that was matched.") | |
1526 | (newtext, fixedcase, literal, string) | |
1527 | Lisp_Object newtext, fixedcase, literal, string; | |
ca1d1d23 JB |
1528 | { |
1529 | enum { nochange, all_caps, cap_initial } case_action; | |
1530 | register int pos, last; | |
1531 | int some_multiletter_word; | |
97832bd0 | 1532 | int some_lowercase; |
73dc8771 | 1533 | int some_uppercase; |
208767c3 | 1534 | int some_nonuppercase_initial; |
ca1d1d23 JB |
1535 | register int c, prevc; |
1536 | int inslen; | |
1537 | ||
16fdc568 | 1538 | CHECK_STRING (newtext, 0); |
ca1d1d23 | 1539 | |
080c45fd RS |
1540 | if (! NILP (string)) |
1541 | CHECK_STRING (string, 4); | |
1542 | ||
ca1d1d23 JB |
1543 | case_action = nochange; /* We tried an initialization */ |
1544 | /* but some C compilers blew it */ | |
4746118a JB |
1545 | |
1546 | if (search_regs.num_regs <= 0) | |
1547 | error ("replace-match called before any match found"); | |
1548 | ||
080c45fd RS |
1549 | if (NILP (string)) |
1550 | { | |
1551 | if (search_regs.start[0] < BEGV | |
1552 | || search_regs.start[0] > search_regs.end[0] | |
1553 | || search_regs.end[0] > ZV) | |
1554 | args_out_of_range (make_number (search_regs.start[0]), | |
1555 | make_number (search_regs.end[0])); | |
1556 | } | |
1557 | else | |
1558 | { | |
1559 | if (search_regs.start[0] < 0 | |
1560 | || search_regs.start[0] > search_regs.end[0] | |
1561 | || search_regs.end[0] > XSTRING (string)->size) | |
1562 | args_out_of_range (make_number (search_regs.start[0]), | |
1563 | make_number (search_regs.end[0])); | |
1564 | } | |
ca1d1d23 JB |
1565 | |
1566 | if (NILP (fixedcase)) | |
1567 | { | |
1568 | /* Decide how to casify by examining the matched text. */ | |
1569 | ||
1570 | last = search_regs.end[0]; | |
1571 | prevc = '\n'; | |
1572 | case_action = all_caps; | |
1573 | ||
1574 | /* some_multiletter_word is set nonzero if any original word | |
1575 | is more than one letter long. */ | |
1576 | some_multiletter_word = 0; | |
97832bd0 | 1577 | some_lowercase = 0; |
208767c3 | 1578 | some_nonuppercase_initial = 0; |
73dc8771 | 1579 | some_uppercase = 0; |
ca1d1d23 JB |
1580 | |
1581 | for (pos = search_regs.start[0]; pos < last; pos++) | |
1582 | { | |
080c45fd RS |
1583 | if (NILP (string)) |
1584 | c = FETCH_CHAR (pos); | |
1585 | else | |
1586 | c = XSTRING (string)->data[pos]; | |
1587 | ||
ca1d1d23 JB |
1588 | if (LOWERCASEP (c)) |
1589 | { | |
1590 | /* Cannot be all caps if any original char is lower case */ | |
1591 | ||
97832bd0 | 1592 | some_lowercase = 1; |
ca1d1d23 | 1593 | if (SYNTAX (prevc) != Sword) |
208767c3 | 1594 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
1595 | else |
1596 | some_multiletter_word = 1; | |
1597 | } | |
1598 | else if (!NOCASEP (c)) | |
1599 | { | |
73dc8771 | 1600 | some_uppercase = 1; |
97832bd0 | 1601 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 1602 | ; |
97832bd0 | 1603 | else |
ca1d1d23 JB |
1604 | some_multiletter_word = 1; |
1605 | } | |
208767c3 RS |
1606 | else |
1607 | { | |
1608 | /* If the initial is a caseless word constituent, | |
1609 | treat that like a lowercase initial. */ | |
1610 | if (SYNTAX (prevc) != Sword) | |
1611 | some_nonuppercase_initial = 1; | |
1612 | } | |
ca1d1d23 JB |
1613 | |
1614 | prevc = c; | |
1615 | } | |
1616 | ||
97832bd0 RS |
1617 | /* Convert to all caps if the old text is all caps |
1618 | and has at least one multiletter word. */ | |
1619 | if (! some_lowercase && some_multiletter_word) | |
1620 | case_action = all_caps; | |
c4d460ce | 1621 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 1622 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 1623 | case_action = cap_initial; |
208767c3 | 1624 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
1625 | /* Should x -> yz, operating on X, give Yz or YZ? |
1626 | We'll assume the latter. */ | |
1627 | case_action = all_caps; | |
97832bd0 RS |
1628 | else |
1629 | case_action = nochange; | |
ca1d1d23 JB |
1630 | } |
1631 | ||
080c45fd RS |
1632 | /* Do replacement in a string. */ |
1633 | if (!NILP (string)) | |
1634 | { | |
1635 | Lisp_Object before, after; | |
1636 | ||
1637 | before = Fsubstring (string, make_number (0), | |
1638 | make_number (search_regs.start[0])); | |
1639 | after = Fsubstring (string, make_number (search_regs.end[0]), Qnil); | |
1640 | ||
1641 | /* Do case substitution into NEWTEXT if desired. */ | |
1642 | if (NILP (literal)) | |
1643 | { | |
1644 | int lastpos = -1; | |
1645 | /* We build up the substituted string in ACCUM. */ | |
1646 | Lisp_Object accum; | |
1647 | Lisp_Object middle; | |
1648 | ||
1649 | accum = Qnil; | |
1650 | ||
1651 | for (pos = 0; pos < XSTRING (newtext)->size; pos++) | |
1652 | { | |
1653 | int substart = -1; | |
1654 | int subend; | |
1655 | ||
1656 | c = XSTRING (newtext)->data[pos]; | |
1657 | if (c == '\\') | |
1658 | { | |
1659 | c = XSTRING (newtext)->data[++pos]; | |
1660 | if (c == '&') | |
1661 | { | |
1662 | substart = search_regs.start[0]; | |
1663 | subend = search_regs.end[0]; | |
1664 | } | |
1665 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') | |
1666 | { | |
1667 | if (search_regs.start[c - '0'] >= 1) | |
1668 | { | |
1669 | substart = search_regs.start[c - '0']; | |
1670 | subend = search_regs.end[c - '0']; | |
1671 | } | |
1672 | } | |
1673 | } | |
1674 | if (substart >= 0) | |
1675 | { | |
1676 | if (pos - 1 != lastpos + 1) | |
1677 | middle = Fsubstring (newtext, lastpos + 1, pos - 1); | |
1678 | else | |
1679 | middle = Qnil; | |
1680 | accum = concat3 (accum, middle, | |
1681 | Fsubstring (string, make_number (substart), | |
1682 | make_number (subend))); | |
1683 | lastpos = pos; | |
1684 | } | |
1685 | } | |
1686 | ||
1687 | if (pos != lastpos + 1) | |
1688 | middle = Fsubstring (newtext, lastpos + 1, pos); | |
1689 | else | |
1690 | middle = Qnil; | |
1691 | ||
1692 | newtext = concat2 (accum, middle); | |
1693 | } | |
1694 | ||
1695 | if (case_action == all_caps) | |
1696 | newtext = Fupcase (newtext); | |
1697 | else if (case_action == cap_initial) | |
1698 | newtext = upcase_initials (newtext); | |
1699 | ||
1700 | return concat3 (before, newtext, after); | |
1701 | } | |
1702 | ||
9a76659d JB |
1703 | /* We insert the replacement text before the old text, and then |
1704 | delete the original text. This means that markers at the | |
1705 | beginning or end of the original will float to the corresponding | |
1706 | position in the replacement. */ | |
1707 | SET_PT (search_regs.start[0]); | |
ca1d1d23 | 1708 | if (!NILP (literal)) |
16fdc568 | 1709 | Finsert_and_inherit (1, &newtext); |
ca1d1d23 JB |
1710 | else |
1711 | { | |
1712 | struct gcpro gcpro1; | |
16fdc568 | 1713 | GCPRO1 (newtext); |
ca1d1d23 | 1714 | |
16fdc568 | 1715 | for (pos = 0; pos < XSTRING (newtext)->size; pos++) |
ca1d1d23 | 1716 | { |
9a76659d JB |
1717 | int offset = point - search_regs.start[0]; |
1718 | ||
16fdc568 | 1719 | c = XSTRING (newtext)->data[pos]; |
ca1d1d23 JB |
1720 | if (c == '\\') |
1721 | { | |
16fdc568 | 1722 | c = XSTRING (newtext)->data[++pos]; |
ca1d1d23 | 1723 | if (c == '&') |
9a76659d JB |
1724 | Finsert_buffer_substring |
1725 | (Fcurrent_buffer (), | |
1726 | make_number (search_regs.start[0] + offset), | |
1727 | make_number (search_regs.end[0] + offset)); | |
78445046 | 1728 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
1729 | { |
1730 | if (search_regs.start[c - '0'] >= 1) | |
9a76659d JB |
1731 | Finsert_buffer_substring |
1732 | (Fcurrent_buffer (), | |
1733 | make_number (search_regs.start[c - '0'] + offset), | |
1734 | make_number (search_regs.end[c - '0'] + offset)); | |
ca1d1d23 JB |
1735 | } |
1736 | else | |
1737 | insert_char (c); | |
1738 | } | |
1739 | else | |
1740 | insert_char (c); | |
1741 | } | |
1742 | UNGCPRO; | |
1743 | } | |
1744 | ||
9a76659d JB |
1745 | inslen = point - (search_regs.start[0]); |
1746 | del_range (search_regs.start[0] + inslen, search_regs.end[0] + inslen); | |
ca1d1d23 JB |
1747 | |
1748 | if (case_action == all_caps) | |
1749 | Fupcase_region (make_number (point - inslen), make_number (point)); | |
1750 | else if (case_action == cap_initial) | |
1751 | upcase_initials_region (make_number (point - inslen), make_number (point)); | |
1752 | return Qnil; | |
1753 | } | |
1754 | \f | |
1755 | static Lisp_Object | |
1756 | match_limit (num, beginningp) | |
1757 | Lisp_Object num; | |
1758 | int beginningp; | |
1759 | { | |
1760 | register int n; | |
1761 | ||
1762 | CHECK_NUMBER (num, 0); | |
1763 | n = XINT (num); | |
4746118a JB |
1764 | if (n < 0 || n >= search_regs.num_regs) |
1765 | args_out_of_range (num, make_number (search_regs.num_regs)); | |
1766 | if (search_regs.num_regs <= 0 | |
1767 | || search_regs.start[n] < 0) | |
ca1d1d23 JB |
1768 | return Qnil; |
1769 | return (make_number ((beginningp) ? search_regs.start[n] | |
1770 | : search_regs.end[n])); | |
1771 | } | |
1772 | ||
1773 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
1774 | "Return position of start of text matched by last search.\n\ | |
16fdc568 BF |
1775 | NUM specifies which parenthesized expression in the last regexp.\n\ |
1776 | Value is nil if NUMth pair didn't match, or there were less than NUM pairs.\n\ | |
ca1d1d23 JB |
1777 | Zero means the entire text matched by the whole regexp or whole string.") |
1778 | (num) | |
1779 | Lisp_Object num; | |
1780 | { | |
1781 | return match_limit (num, 1); | |
1782 | } | |
1783 | ||
1784 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
1785 | "Return position of end of text matched by last search.\n\ | |
1786 | ARG, a number, specifies which parenthesized expression in the last regexp.\n\ | |
1787 | Value is nil if ARGth pair didn't match, or there were less than ARG pairs.\n\ | |
1788 | Zero means the entire text matched by the whole regexp or whole string.") | |
1789 | (num) | |
1790 | Lisp_Object num; | |
1791 | { | |
1792 | return match_limit (num, 0); | |
1793 | } | |
1794 | ||
1795 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 0, 0, | |
1796 | "Return a list containing all info on what the last search matched.\n\ | |
1797 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'.\n\ | |
1798 | All the elements are markers or nil (nil if the Nth pair didn't match)\n\ | |
1799 | if the last match was on a buffer; integers or nil if a string was matched.\n\ | |
1800 | Use `store-match-data' to reinstate the data in this list.") | |
1801 | () | |
1802 | { | |
4746118a | 1803 | Lisp_Object *data; |
ca1d1d23 JB |
1804 | int i, len; |
1805 | ||
daa37602 JB |
1806 | if (NILP (last_thing_searched)) |
1807 | error ("match-data called before any match found"); | |
1808 | ||
4746118a JB |
1809 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs) |
1810 | * sizeof (Lisp_Object)); | |
1811 | ||
ca1d1d23 | 1812 | len = -1; |
4746118a | 1813 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1814 | { |
1815 | int start = search_regs.start[i]; | |
1816 | if (start >= 0) | |
1817 | { | |
daa37602 | 1818 | if (EQ (last_thing_searched, Qt)) |
ca1d1d23 | 1819 | { |
c235cce7 KH |
1820 | XSETFASTINT (data[2 * i], start); |
1821 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 1822 | } |
0ed62dc7 | 1823 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
1824 | { |
1825 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
1826 | Fset_marker (data[2 * i], |
1827 | make_number (start), | |
1828 | last_thing_searched); | |
ca1d1d23 JB |
1829 | data[2 * i + 1] = Fmake_marker (); |
1830 | Fset_marker (data[2 * i + 1], | |
daa37602 JB |
1831 | make_number (search_regs.end[i]), |
1832 | last_thing_searched); | |
ca1d1d23 | 1833 | } |
daa37602 JB |
1834 | else |
1835 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
1836 | abort (); | |
1837 | ||
ca1d1d23 JB |
1838 | len = i; |
1839 | } | |
1840 | else | |
1841 | data[2 * i] = data [2 * i + 1] = Qnil; | |
1842 | } | |
1843 | return Flist (2 * len + 2, data); | |
1844 | } | |
1845 | ||
1846 | ||
1847 | DEFUN ("store-match-data", Fstore_match_data, Sstore_match_data, 1, 1, 0, | |
1848 | "Set internal data on last search match from elements of LIST.\n\ | |
1849 | LIST should have been created by calling `match-data' previously.") | |
1850 | (list) | |
1851 | register Lisp_Object list; | |
1852 | { | |
1853 | register int i; | |
1854 | register Lisp_Object marker; | |
1855 | ||
7074fde6 FP |
1856 | if (running_asynch_code) |
1857 | save_search_regs (); | |
1858 | ||
ca1d1d23 | 1859 | if (!CONSP (list) && !NILP (list)) |
b37902c8 | 1860 | list = wrong_type_argument (Qconsp, list); |
ca1d1d23 | 1861 | |
daa37602 JB |
1862 | /* Unless we find a marker with a buffer in LIST, assume that this |
1863 | match data came from a string. */ | |
1864 | last_thing_searched = Qt; | |
1865 | ||
4746118a JB |
1866 | /* Allocate registers if they don't already exist. */ |
1867 | { | |
d084e942 | 1868 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
1869 | |
1870 | if (length > search_regs.num_regs) | |
1871 | { | |
1113d9db JB |
1872 | if (search_regs.num_regs == 0) |
1873 | { | |
1874 | search_regs.start | |
1875 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
1876 | search_regs.end | |
1877 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
1878 | } | |
4746118a | 1879 | else |
1113d9db JB |
1880 | { |
1881 | search_regs.start | |
1882 | = (regoff_t *) xrealloc (search_regs.start, | |
1883 | length * sizeof (regoff_t)); | |
1884 | search_regs.end | |
1885 | = (regoff_t *) xrealloc (search_regs.end, | |
1886 | length * sizeof (regoff_t)); | |
1887 | } | |
4746118a | 1888 | |
487282dc | 1889 | search_regs.num_regs = length; |
4746118a JB |
1890 | } |
1891 | } | |
1892 | ||
1893 | for (i = 0; i < search_regs.num_regs; i++) | |
ca1d1d23 JB |
1894 | { |
1895 | marker = Fcar (list); | |
1896 | if (NILP (marker)) | |
1897 | { | |
1898 | search_regs.start[i] = -1; | |
1899 | list = Fcdr (list); | |
1900 | } | |
1901 | else | |
1902 | { | |
0ed62dc7 | 1903 | if (MARKERP (marker)) |
daa37602 JB |
1904 | { |
1905 | if (XMARKER (marker)->buffer == 0) | |
c235cce7 | 1906 | XSETFASTINT (marker, 0); |
daa37602 | 1907 | else |
a3668d92 | 1908 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); |
daa37602 | 1909 | } |
ca1d1d23 JB |
1910 | |
1911 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
1912 | search_regs.start[i] = XINT (marker); | |
1913 | list = Fcdr (list); | |
1914 | ||
1915 | marker = Fcar (list); | |
0ed62dc7 | 1916 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
c235cce7 | 1917 | XSETFASTINT (marker, 0); |
ca1d1d23 JB |
1918 | |
1919 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
1920 | search_regs.end[i] = XINT (marker); | |
1921 | } | |
1922 | list = Fcdr (list); | |
1923 | } | |
1924 | ||
1925 | return Qnil; | |
1926 | } | |
1927 | ||
7074fde6 FP |
1928 | /* If non-zero the match data have been saved in saved_search_regs |
1929 | during the execution of a sentinel or filter. */ | |
75ebf74b | 1930 | static int search_regs_saved; |
7074fde6 FP |
1931 | static struct re_registers saved_search_regs; |
1932 | ||
1933 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
1934 | if asynchronous code (filter or sentinel) is running. */ | |
1935 | static void | |
1936 | save_search_regs () | |
1937 | { | |
1938 | if (!search_regs_saved) | |
1939 | { | |
1940 | saved_search_regs.num_regs = search_regs.num_regs; | |
1941 | saved_search_regs.start = search_regs.start; | |
1942 | saved_search_regs.end = search_regs.end; | |
1943 | search_regs.num_regs = 0; | |
1944 | ||
1945 | search_regs_saved = 1; | |
1946 | } | |
1947 | } | |
1948 | ||
1949 | /* Called upon exit from filters and sentinels. */ | |
1950 | void | |
1951 | restore_match_data () | |
1952 | { | |
1953 | if (search_regs_saved) | |
1954 | { | |
1955 | if (search_regs.num_regs > 0) | |
1956 | { | |
1957 | xfree (search_regs.start); | |
1958 | xfree (search_regs.end); | |
1959 | } | |
1960 | search_regs.num_regs = saved_search_regs.num_regs; | |
1961 | search_regs.start = saved_search_regs.start; | |
1962 | search_regs.end = saved_search_regs.end; | |
1963 | ||
1964 | search_regs_saved = 0; | |
1965 | } | |
1966 | } | |
1967 | ||
ca1d1d23 JB |
1968 | /* Quote a string to inactivate reg-expr chars */ |
1969 | ||
1970 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
1971 | "Return a regexp string which matches exactly STRING and nothing else.") | |
1972 | (str) | |
1973 | Lisp_Object str; | |
1974 | { | |
1975 | register unsigned char *in, *out, *end; | |
1976 | register unsigned char *temp; | |
1977 | ||
1978 | CHECK_STRING (str, 0); | |
1979 | ||
1980 | temp = (unsigned char *) alloca (XSTRING (str)->size * 2); | |
1981 | ||
1982 | /* Now copy the data into the new string, inserting escapes. */ | |
1983 | ||
1984 | in = XSTRING (str)->data; | |
1985 | end = in + XSTRING (str)->size; | |
1986 | out = temp; | |
1987 | ||
1988 | for (; in != end; in++) | |
1989 | { | |
1990 | if (*in == '[' || *in == ']' | |
1991 | || *in == '*' || *in == '.' || *in == '\\' | |
1992 | || *in == '?' || *in == '+' | |
1993 | || *in == '^' || *in == '$') | |
1994 | *out++ = '\\'; | |
1995 | *out++ = *in; | |
1996 | } | |
1997 | ||
1998 | return make_string (temp, out - temp); | |
1999 | } | |
2000 | \f | |
2001 | syms_of_search () | |
2002 | { | |
2003 | register int i; | |
2004 | ||
487282dc KH |
2005 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
2006 | { | |
2007 | searchbufs[i].buf.allocated = 100; | |
2008 | searchbufs[i].buf.buffer = (unsigned char *) malloc (100); | |
2009 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; | |
2010 | searchbufs[i].regexp = Qnil; | |
2011 | staticpro (&searchbufs[i].regexp); | |
2012 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); | |
2013 | } | |
2014 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
2015 | |
2016 | Qsearch_failed = intern ("search-failed"); | |
2017 | staticpro (&Qsearch_failed); | |
2018 | Qinvalid_regexp = intern ("invalid-regexp"); | |
2019 | staticpro (&Qinvalid_regexp); | |
2020 | ||
2021 | Fput (Qsearch_failed, Qerror_conditions, | |
2022 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
2023 | Fput (Qsearch_failed, Qerror_message, | |
2024 | build_string ("Search failed")); | |
2025 | ||
2026 | Fput (Qinvalid_regexp, Qerror_conditions, | |
2027 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
2028 | Fput (Qinvalid_regexp, Qerror_message, | |
2029 | build_string ("Invalid regexp")); | |
2030 | ||
daa37602 JB |
2031 | last_thing_searched = Qnil; |
2032 | staticpro (&last_thing_searched); | |
2033 | ||
ca1d1d23 | 2034 | defsubr (&Slooking_at); |
b819a390 RS |
2035 | defsubr (&Sposix_looking_at); |
2036 | defsubr (&Sstring_match); | |
2037 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
2038 | defsubr (&Sskip_chars_forward); |
2039 | defsubr (&Sskip_chars_backward); | |
17431c60 RS |
2040 | defsubr (&Sskip_syntax_forward); |
2041 | defsubr (&Sskip_syntax_backward); | |
ca1d1d23 JB |
2042 | defsubr (&Ssearch_forward); |
2043 | defsubr (&Ssearch_backward); | |
2044 | defsubr (&Sword_search_forward); | |
2045 | defsubr (&Sword_search_backward); | |
2046 | defsubr (&Sre_search_forward); | |
2047 | defsubr (&Sre_search_backward); | |
b819a390 RS |
2048 | defsubr (&Sposix_search_forward); |
2049 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
2050 | defsubr (&Sreplace_match); |
2051 | defsubr (&Smatch_beginning); | |
2052 | defsubr (&Smatch_end); | |
2053 | defsubr (&Smatch_data); | |
2054 | defsubr (&Sstore_match_data); | |
2055 | defsubr (&Sregexp_quote); | |
2056 | } |