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 RS |
80 | static void set_search_regs (); |
81 | ||
b819a390 RS |
82 | static int search_buffer (); |
83 | ||
ca1d1d23 JB |
84 | static void |
85 | matcher_overflow () | |
86 | { | |
87 | error ("Stack overflow in regexp matcher"); | |
88 | } | |
89 | ||
90 | #ifdef __STDC__ | |
91 | #define CONST const | |
92 | #else | |
93 | #define CONST | |
94 | #endif | |
95 | ||
b819a390 RS |
96 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
97 | PATTERN is the pattern to compile. | |
98 | CP is the place to put the result. | |
99 | TRANSLATE is a translation table for ignoring case, or NULL for none. | |
100 | REGP is the structure that says where to store the "register" | |
101 | values that will result from matching this pattern. | |
102 | If it is 0, we should compile the pattern not to record any | |
103 | subexpression bounds. | |
104 | POSIX is nonzero if we want full backtracking (POSIX style) | |
105 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 106 | |
487282dc | 107 | static void |
b819a390 | 108 | compile_pattern_1 (cp, pattern, translate, regp, posix) |
487282dc | 109 | struct regexp_cache *cp; |
ca1d1d23 | 110 | Lisp_Object pattern; |
ca1d1d23 | 111 | char *translate; |
487282dc | 112 | struct re_registers *regp; |
b819a390 | 113 | int posix; |
ca1d1d23 JB |
114 | { |
115 | CONST char *val; | |
b819a390 | 116 | reg_syntax_t old; |
ca1d1d23 | 117 | |
487282dc KH |
118 | cp->regexp = Qnil; |
119 | cp->buf.translate = translate; | |
b819a390 | 120 | cp->posix = posix; |
9ac0d9e0 | 121 | BLOCK_INPUT; |
b819a390 RS |
122 | old = re_set_syntax (RE_SYNTAX_EMACS |
123 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); | |
b90d9e80 | 124 | val = (CONST char *) re_compile_pattern ((char *) XSTRING (pattern)->data, |
487282dc | 125 | XSTRING (pattern)->size, &cp->buf); |
b819a390 | 126 | re_set_syntax (old); |
9ac0d9e0 | 127 | UNBLOCK_INPUT; |
ca1d1d23 | 128 | if (val) |
487282dc | 129 | Fsignal (Qinvalid_regexp, Fcons (build_string (val), Qnil)); |
1113d9db | 130 | |
487282dc | 131 | cp->regexp = Fcopy_sequence (pattern); |
1113d9db JB |
132 | |
133 | /* Advise the searching functions about the space we have allocated | |
134 | for register data. */ | |
9ac0d9e0 | 135 | BLOCK_INPUT; |
ebb9e16f | 136 | if (regp) |
487282dc | 137 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); |
9ac0d9e0 | 138 | UNBLOCK_INPUT; |
487282dc KH |
139 | } |
140 | ||
141 | /* Compile a regexp if necessary, but first check to see if there's one in | |
b819a390 RS |
142 | the cache. |
143 | PATTERN is the pattern to compile. | |
144 | TRANSLATE is a translation table for ignoring case, or NULL for none. | |
145 | REGP is the structure that says where to store the "register" | |
146 | values that will result from matching this pattern. | |
147 | If it is 0, we should compile the pattern not to record any | |
148 | subexpression bounds. | |
149 | POSIX is nonzero if we want full backtracking (POSIX style) | |
150 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
151 | |
152 | struct re_pattern_buffer * | |
b819a390 | 153 | compile_pattern (pattern, regp, translate, posix) |
487282dc KH |
154 | Lisp_Object pattern; |
155 | struct re_registers *regp; | |
156 | char *translate; | |
b819a390 | 157 | int posix; |
487282dc KH |
158 | { |
159 | struct regexp_cache *cp, **cpp; | |
160 | ||
161 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
162 | { | |
163 | cp = *cpp; | |
164 | if (!NILP (Fstring_equal (cp->regexp, pattern)) | |
b819a390 RS |
165 | && cp->buf.translate == translate |
166 | && cp->posix == posix) | |
487282dc KH |
167 | break; |
168 | ||
169 | /* If we're at the end of the cache, compile into the last cell. */ | |
170 | if (cp->next == 0) | |
171 | { | |
b819a390 | 172 | compile_pattern_1 (cp, pattern, translate, regp, posix); |
487282dc KH |
173 | break; |
174 | } | |
175 | } | |
176 | ||
177 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
178 | either because we found it in the cache or because we just compiled it. | |
179 | Move it to the front of the queue to mark it as most recently used. */ | |
180 | *cpp = cp->next; | |
181 | cp->next = searchbuf_head; | |
182 | searchbuf_head = cp; | |
1113d9db | 183 | |
487282dc | 184 | return &cp->buf; |
ca1d1d23 JB |
185 | } |
186 | ||
187 | /* Error condition used for failing searches */ | |
188 | Lisp_Object Qsearch_failed; | |
189 | ||
190 | Lisp_Object | |
191 | signal_failure (arg) | |
192 | Lisp_Object arg; | |
193 | { | |
194 | Fsignal (Qsearch_failed, Fcons (arg, Qnil)); | |
195 | return Qnil; | |
196 | } | |
197 | \f | |
b819a390 RS |
198 | static Lisp_Object |
199 | looking_at_1 (string, posix) | |
ca1d1d23 | 200 | Lisp_Object string; |
b819a390 | 201 | int posix; |
ca1d1d23 JB |
202 | { |
203 | Lisp_Object val; | |
204 | unsigned char *p1, *p2; | |
205 | int s1, s2; | |
206 | register int i; | |
487282dc | 207 | struct re_pattern_buffer *bufp; |
ca1d1d23 JB |
208 | |
209 | CHECK_STRING (string, 0); | |
487282dc KH |
210 | bufp = compile_pattern (string, &search_regs, |
211 | (!NILP (current_buffer->case_fold_search) | |
b819a390 RS |
212 | ? DOWNCASE_TABLE : 0), |
213 | posix); | |
ca1d1d23 JB |
214 | |
215 | immediate_quit = 1; | |
216 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
217 | ||
218 | /* Get pointers and sizes of the two strings | |
219 | that make up the visible portion of the buffer. */ | |
220 | ||
221 | p1 = BEGV_ADDR; | |
222 | s1 = GPT - BEGV; | |
223 | p2 = GAP_END_ADDR; | |
224 | s2 = ZV - GPT; | |
225 | if (s1 < 0) | |
226 | { | |
227 | p2 = p1; | |
228 | s2 = ZV - BEGV; | |
229 | s1 = 0; | |
230 | } | |
231 | if (s2 < 0) | |
232 | { | |
233 | s1 = ZV - BEGV; | |
234 | s2 = 0; | |
235 | } | |
236 | ||
487282dc | 237 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
ca1d1d23 JB |
238 | point - BEGV, &search_regs, |
239 | ZV - BEGV); | |
240 | if (i == -2) | |
241 | matcher_overflow (); | |
242 | ||
243 | val = (0 <= i ? Qt : Qnil); | |
4746118a | 244 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
245 | if (search_regs.start[i] >= 0) |
246 | { | |
247 | search_regs.start[i] += BEGV; | |
248 | search_regs.end[i] += BEGV; | |
249 | } | |
a3668d92 | 250 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
251 | immediate_quit = 0; |
252 | return val; | |
253 | } | |
254 | ||
b819a390 RS |
255 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
256 | "Return t if text after point matches regular expression PAT.\n\ | |
257 | This function modifies the match data that `match-beginning',\n\ | |
258 | `match-end' and `match-data' access; save and restore the match\n\ | |
259 | data if you want to preserve them.") | |
260 | (string) | |
261 | Lisp_Object string; | |
262 | { | |
263 | return looking_at_1 (string, 0); | |
264 | } | |
265 | ||
266 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
267 | "Return t if text after point matches regular expression PAT.\n\ | |
268 | Find the longest match, in accord with Posix regular expression rules.\n\ | |
269 | This function modifies the match data that `match-beginning',\n\ | |
270 | `match-end' and `match-data' access; save and restore the match\n\ | |
271 | data if you want to preserve them.") | |
272 | (string) | |
273 | Lisp_Object string; | |
274 | { | |
275 | return looking_at_1 (string, 1); | |
276 | } | |
277 | \f | |
278 | static Lisp_Object | |
279 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 280 | Lisp_Object regexp, string, start; |
b819a390 | 281 | int posix; |
ca1d1d23 JB |
282 | { |
283 | int val; | |
284 | int s; | |
487282dc | 285 | struct re_pattern_buffer *bufp; |
ca1d1d23 JB |
286 | |
287 | CHECK_STRING (regexp, 0); | |
288 | CHECK_STRING (string, 1); | |
289 | ||
290 | if (NILP (start)) | |
291 | s = 0; | |
292 | else | |
293 | { | |
294 | int len = XSTRING (string)->size; | |
295 | ||
296 | CHECK_NUMBER (start, 2); | |
297 | s = XINT (start); | |
298 | if (s < 0 && -s <= len) | |
26faf9f4 | 299 | s = len + s; |
ca1d1d23 JB |
300 | else if (0 > s || s > len) |
301 | args_out_of_range (string, start); | |
302 | } | |
303 | ||
487282dc KH |
304 | bufp = compile_pattern (regexp, &search_regs, |
305 | (!NILP (current_buffer->case_fold_search) | |
b819a390 RS |
306 | ? DOWNCASE_TABLE : 0), |
307 | 0); | |
ca1d1d23 | 308 | immediate_quit = 1; |
487282dc | 309 | val = re_search (bufp, (char *) XSTRING (string)->data, |
ca1d1d23 JB |
310 | XSTRING (string)->size, s, XSTRING (string)->size - s, |
311 | &search_regs); | |
312 | immediate_quit = 0; | |
daa37602 | 313 | last_thing_searched = Qt; |
ca1d1d23 JB |
314 | if (val == -2) |
315 | matcher_overflow (); | |
316 | if (val < 0) return Qnil; | |
317 | return make_number (val); | |
318 | } | |
e59a8453 | 319 | |
b819a390 RS |
320 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
321 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
322 | If third arg START is non-nil, start search at that index in STRING.\n\ | |
323 | For index of first char beyond the match, do (match-end 0).\n\ | |
324 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
325 | matched by parenthesis constructs in the pattern.") | |
326 | (regexp, string, start) | |
327 | Lisp_Object regexp, string, start; | |
328 | { | |
329 | return string_match_1 (regexp, string, start, 0); | |
330 | } | |
331 | ||
332 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
333 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
334 | Find the longest match, in accord with Posix regular expression rules.\n\ | |
335 | If third arg START is non-nil, start search at that index in STRING.\n\ | |
336 | For index of first char beyond the match, do (match-end 0).\n\ | |
337 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
338 | matched by parenthesis constructs in the pattern.") | |
339 | (regexp, string, start) | |
340 | Lisp_Object regexp, string, start; | |
341 | { | |
342 | return string_match_1 (regexp, string, start, 1); | |
343 | } | |
344 | ||
e59a8453 RS |
345 | /* Match REGEXP against STRING, searching all of STRING, |
346 | and return the index of the match, or negative on failure. | |
347 | This does not clobber the match data. */ | |
348 | ||
349 | int | |
350 | fast_string_match (regexp, string) | |
351 | Lisp_Object regexp, string; | |
352 | { | |
353 | int val; | |
487282dc | 354 | struct re_pattern_buffer *bufp; |
e59a8453 | 355 | |
b819a390 | 356 | bufp = compile_pattern (regexp, 0, 0, 0); |
e59a8453 | 357 | immediate_quit = 1; |
487282dc | 358 | val = re_search (bufp, (char *) XSTRING (string)->data, |
e59a8453 RS |
359 | XSTRING (string)->size, 0, XSTRING (string)->size, |
360 | 0); | |
361 | immediate_quit = 0; | |
362 | return val; | |
363 | } | |
ca1d1d23 | 364 | \f |
9169c321 JB |
365 | /* max and min. */ |
366 | ||
367 | static int | |
368 | max (a, b) | |
369 | int a, b; | |
370 | { | |
371 | return ((a > b) ? a : b); | |
372 | } | |
373 | ||
374 | static int | |
375 | min (a, b) | |
376 | int a, b; | |
377 | { | |
378 | return ((a < b) ? a : b); | |
379 | } | |
380 | ||
381 | \f | |
382 | /* The newline cache: remembering which sections of text have no newlines. */ | |
383 | ||
384 | /* If the user has requested newline caching, make sure it's on. | |
385 | Otherwise, make sure it's off. | |
386 | This is our cheezy way of associating an action with the change of | |
387 | state of a buffer-local variable. */ | |
388 | static void | |
389 | newline_cache_on_off (buf) | |
390 | struct buffer *buf; | |
391 | { | |
392 | if (NILP (buf->cache_long_line_scans)) | |
393 | { | |
394 | /* It should be off. */ | |
395 | if (buf->newline_cache) | |
396 | { | |
397 | free_region_cache (buf->newline_cache); | |
398 | buf->newline_cache = 0; | |
399 | } | |
400 | } | |
401 | else | |
402 | { | |
403 | /* It should be on. */ | |
404 | if (buf->newline_cache == 0) | |
405 | buf->newline_cache = new_region_cache (); | |
406 | } | |
407 | } | |
408 | ||
409 | \f | |
410 | /* Search for COUNT instances of the character TARGET between START and END. | |
411 | ||
412 | If COUNT is positive, search forwards; END must be >= START. | |
413 | If COUNT is negative, search backwards for the -COUNTth instance; | |
414 | END must be <= START. | |
415 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
416 | ||
417 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
418 | direction indicated by COUNT. | |
ffd56f97 JB |
419 | |
420 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
5bfe95c9 RS |
421 | position after the COUNTth match. Note that for reverse motion |
422 | this is not the same as the usual convention for Emacs motion commands. | |
ffd56f97 | 423 | |
9169c321 JB |
424 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
425 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 426 | |
087a5f81 RS |
427 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
428 | except when inside redisplay. */ | |
429 | ||
9169c321 JB |
430 | scan_buffer (target, start, end, count, shortage, allow_quit) |
431 | register int target; | |
432 | int start, end; | |
433 | int count; | |
434 | int *shortage; | |
087a5f81 | 435 | int allow_quit; |
ca1d1d23 | 436 | { |
9169c321 JB |
437 | struct region_cache *newline_cache; |
438 | int direction; | |
ffd56f97 | 439 | |
9169c321 JB |
440 | if (count > 0) |
441 | { | |
442 | direction = 1; | |
443 | if (! end) end = ZV; | |
444 | } | |
445 | else | |
446 | { | |
447 | direction = -1; | |
448 | if (! end) end = BEGV; | |
449 | } | |
ffd56f97 | 450 | |
9169c321 JB |
451 | newline_cache_on_off (current_buffer); |
452 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
453 | |
454 | if (shortage != 0) | |
455 | *shortage = 0; | |
456 | ||
087a5f81 | 457 | immediate_quit = allow_quit; |
ca1d1d23 | 458 | |
ffd56f97 | 459 | if (count > 0) |
9169c321 | 460 | while (start != end) |
ca1d1d23 | 461 | { |
9169c321 JB |
462 | /* Our innermost scanning loop is very simple; it doesn't know |
463 | about gaps, buffer ends, or the newline cache. ceiling is | |
464 | the position of the last character before the next such | |
465 | obstacle --- the last character the dumb search loop should | |
466 | examine. */ | |
467 | register int ceiling = end - 1; | |
468 | ||
469 | /* If we're looking for a newline, consult the newline cache | |
470 | to see where we can avoid some scanning. */ | |
471 | if (target == '\n' && newline_cache) | |
472 | { | |
473 | int next_change; | |
474 | immediate_quit = 0; | |
475 | while (region_cache_forward | |
476 | (current_buffer, newline_cache, start, &next_change)) | |
477 | start = next_change; | |
cbe0db0d | 478 | immediate_quit = allow_quit; |
9169c321 JB |
479 | |
480 | /* start should never be after end. */ | |
481 | if (start >= end) | |
482 | start = end - 1; | |
483 | ||
484 | /* Now the text after start is an unknown region, and | |
485 | next_change is the position of the next known region. */ | |
486 | ceiling = min (next_change - 1, ceiling); | |
487 | } | |
488 | ||
489 | /* The dumb loop can only scan text stored in contiguous | |
490 | bytes. BUFFER_CEILING_OF returns the last character | |
491 | position that is contiguous, so the ceiling is the | |
492 | position after that. */ | |
493 | ceiling = min (BUFFER_CEILING_OF (start), ceiling); | |
494 | ||
495 | { | |
496 | /* The termination address of the dumb loop. */ | |
497 | register unsigned char *ceiling_addr = &FETCH_CHAR (ceiling) + 1; | |
498 | register unsigned char *cursor = &FETCH_CHAR (start); | |
499 | unsigned char *base = cursor; | |
500 | ||
501 | while (cursor < ceiling_addr) | |
502 | { | |
503 | unsigned char *scan_start = cursor; | |
504 | ||
505 | /* The dumb loop. */ | |
506 | while (*cursor != target && ++cursor < ceiling_addr) | |
507 | ; | |
508 | ||
509 | /* If we're looking for newlines, cache the fact that | |
510 | the region from start to cursor is free of them. */ | |
511 | if (target == '\n' && newline_cache) | |
512 | know_region_cache (current_buffer, newline_cache, | |
513 | start + scan_start - base, | |
514 | start + cursor - base); | |
515 | ||
516 | /* Did we find the target character? */ | |
517 | if (cursor < ceiling_addr) | |
518 | { | |
519 | if (--count == 0) | |
520 | { | |
521 | immediate_quit = 0; | |
522 | return (start + cursor - base + 1); | |
523 | } | |
524 | cursor++; | |
525 | } | |
526 | } | |
527 | ||
528 | start += cursor - base; | |
529 | } | |
ca1d1d23 JB |
530 | } |
531 | else | |
9169c321 JB |
532 | while (start > end) |
533 | { | |
534 | /* The last character to check before the next obstacle. */ | |
535 | register int ceiling = end; | |
536 | ||
537 | /* Consult the newline cache, if appropriate. */ | |
538 | if (target == '\n' && newline_cache) | |
539 | { | |
540 | int next_change; | |
541 | immediate_quit = 0; | |
542 | while (region_cache_backward | |
543 | (current_buffer, newline_cache, start, &next_change)) | |
544 | start = next_change; | |
cbe0db0d | 545 | immediate_quit = allow_quit; |
9169c321 JB |
546 | |
547 | /* Start should never be at or before end. */ | |
548 | if (start <= end) | |
549 | start = end + 1; | |
550 | ||
551 | /* Now the text before start is an unknown region, and | |
552 | next_change is the position of the next known region. */ | |
553 | ceiling = max (next_change, ceiling); | |
554 | } | |
555 | ||
556 | /* Stop scanning before the gap. */ | |
557 | ceiling = max (BUFFER_FLOOR_OF (start - 1), ceiling); | |
558 | ||
559 | { | |
560 | /* The termination address of the dumb loop. */ | |
561 | register unsigned char *ceiling_addr = &FETCH_CHAR (ceiling); | |
562 | register unsigned char *cursor = &FETCH_CHAR (start - 1); | |
563 | unsigned char *base = cursor; | |
564 | ||
565 | while (cursor >= ceiling_addr) | |
566 | { | |
567 | unsigned char *scan_start = cursor; | |
568 | ||
569 | while (*cursor != target && --cursor >= ceiling_addr) | |
570 | ; | |
571 | ||
572 | /* If we're looking for newlines, cache the fact that | |
573 | the region from after the cursor to start is free of them. */ | |
574 | if (target == '\n' && newline_cache) | |
575 | know_region_cache (current_buffer, newline_cache, | |
576 | start + cursor - base, | |
577 | start + scan_start - base); | |
578 | ||
579 | /* Did we find the target character? */ | |
580 | if (cursor >= ceiling_addr) | |
581 | { | |
582 | if (++count >= 0) | |
583 | { | |
584 | immediate_quit = 0; | |
585 | return (start + cursor - base); | |
586 | } | |
587 | cursor--; | |
588 | } | |
589 | } | |
590 | ||
591 | start += cursor - base; | |
592 | } | |
593 | } | |
594 | ||
ca1d1d23 JB |
595 | immediate_quit = 0; |
596 | if (shortage != 0) | |
ffd56f97 | 597 | *shortage = count * direction; |
9169c321 | 598 | return start; |
ca1d1d23 JB |
599 | } |
600 | ||
63fa018d RS |
601 | int |
602 | find_next_newline_no_quit (from, cnt) | |
603 | register int from, cnt; | |
604 | { | |
9169c321 | 605 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
63fa018d RS |
606 | } |
607 | ||
ca1d1d23 JB |
608 | int |
609 | find_next_newline (from, cnt) | |
610 | register int from, cnt; | |
611 | { | |
9169c321 JB |
612 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 1); |
613 | } | |
614 | ||
615 | ||
616 | /* Like find_next_newline, but returns position before the newline, | |
617 | not after, and only search up to TO. This isn't just | |
618 | find_next_newline (...)-1, because you might hit TO. */ | |
619 | int | |
620 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 621 | int from, to, cnt; |
9169c321 JB |
622 | { |
623 | int shortage; | |
624 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
625 | ||
626 | if (shortage == 0) | |
627 | pos--; | |
628 | ||
629 | return pos; | |
ca1d1d23 JB |
630 | } |
631 | \f | |
c1dc99a1 JB |
632 | Lisp_Object skip_chars (); |
633 | ||
ca1d1d23 | 634 | DEFUN ("skip-chars-forward", Fskip_chars_forward, Sskip_chars_forward, 1, 2, 0, |
3acb9a69 RS |
635 | "Move point forward, stopping before a char not in STRING, or at pos LIM.\n\ |
636 | STRING is like the inside of a `[...]' in a regular expression\n\ | |
ca1d1d23 JB |
637 | except that `]' is never special and `\\' quotes `^', `-' or `\\'.\n\ |
638 | Thus, with arg \"a-zA-Z\", this skips letters stopping before first nonletter.\n\ | |
c1dc99a1 JB |
639 | With arg \"^a-zA-Z\", skips nonletters stopping before first letter.\n\ |
640 | Returns the distance traveled, either zero or positive.") | |
ca1d1d23 JB |
641 | (string, lim) |
642 | Lisp_Object string, lim; | |
643 | { | |
17431c60 | 644 | return skip_chars (1, 0, string, lim); |
ca1d1d23 JB |
645 | } |
646 | ||
647 | DEFUN ("skip-chars-backward", Fskip_chars_backward, Sskip_chars_backward, 1, 2, 0, | |
3acb9a69 | 648 | "Move point backward, stopping after a char not in STRING, or at pos LIM.\n\ |
c1dc99a1 JB |
649 | See `skip-chars-forward' for details.\n\ |
650 | Returns the distance traveled, either zero or negative.") | |
ca1d1d23 JB |
651 | (string, lim) |
652 | Lisp_Object string, lim; | |
653 | { | |
17431c60 RS |
654 | return skip_chars (0, 0, string, lim); |
655 | } | |
656 | ||
657 | DEFUN ("skip-syntax-forward", Fskip_syntax_forward, Sskip_syntax_forward, 1, 2, 0, | |
658 | "Move point forward across chars in specified syntax classes.\n\ | |
659 | SYNTAX is a string of syntax code characters.\n\ | |
660 | Stop before a char whose syntax is not in SYNTAX, or at position LIM.\n\ | |
661 | If SYNTAX starts with ^, skip characters whose syntax is NOT in SYNTAX.\n\ | |
662 | This function returns the distance traveled, either zero or positive.") | |
663 | (syntax, lim) | |
664 | Lisp_Object syntax, lim; | |
665 | { | |
666 | return skip_chars (1, 1, syntax, lim); | |
667 | } | |
668 | ||
669 | DEFUN ("skip-syntax-backward", Fskip_syntax_backward, Sskip_syntax_backward, 1, 2, 0, | |
670 | "Move point backward across chars in specified syntax classes.\n\ | |
671 | SYNTAX is a string of syntax code characters.\n\ | |
672 | Stop on reaching a char whose syntax is not in SYNTAX, or at position LIM.\n\ | |
673 | If SYNTAX starts with ^, skip characters whose syntax is NOT in SYNTAX.\n\ | |
674 | This function returns the distance traveled, either zero or negative.") | |
675 | (syntax, lim) | |
676 | Lisp_Object syntax, lim; | |
677 | { | |
678 | return skip_chars (0, 1, syntax, lim); | |
ca1d1d23 JB |
679 | } |
680 | ||
c1dc99a1 | 681 | Lisp_Object |
17431c60 RS |
682 | skip_chars (forwardp, syntaxp, string, lim) |
683 | int forwardp, syntaxp; | |
ca1d1d23 JB |
684 | Lisp_Object string, lim; |
685 | { | |
686 | register unsigned char *p, *pend; | |
687 | register unsigned char c; | |
688 | unsigned char fastmap[0400]; | |
689 | int negate = 0; | |
690 | register int i; | |
691 | ||
692 | CHECK_STRING (string, 0); | |
693 | ||
694 | if (NILP (lim)) | |
a3668d92 | 695 | XSETINT (lim, forwardp ? ZV : BEGV); |
ca1d1d23 JB |
696 | else |
697 | CHECK_NUMBER_COERCE_MARKER (lim, 1); | |
698 | ||
ca1d1d23 | 699 | /* In any case, don't allow scan outside bounds of buffer. */ |
c5241910 RS |
700 | /* jla turned this off, for no known reason. |
701 | bfox turned the ZV part on, and rms turned the | |
702 | BEGV part back on. */ | |
703 | if (XINT (lim) > ZV) | |
c235cce7 | 704 | XSETFASTINT (lim, ZV); |
c5241910 | 705 | if (XINT (lim) < BEGV) |
c235cce7 | 706 | XSETFASTINT (lim, BEGV); |
ca1d1d23 JB |
707 | |
708 | p = XSTRING (string)->data; | |
709 | pend = p + XSTRING (string)->size; | |
710 | bzero (fastmap, sizeof fastmap); | |
711 | ||
712 | if (p != pend && *p == '^') | |
713 | { | |
714 | negate = 1; p++; | |
715 | } | |
716 | ||
17431c60 RS |
717 | /* Find the characters specified and set their elements of fastmap. |
718 | If syntaxp, each character counts as itself. | |
719 | Otherwise, handle backslashes and ranges specially */ | |
ca1d1d23 JB |
720 | |
721 | while (p != pend) | |
722 | { | |
723 | c = *p++; | |
17431c60 RS |
724 | if (syntaxp) |
725 | fastmap[c] = 1; | |
726 | else | |
ca1d1d23 | 727 | { |
17431c60 | 728 | if (c == '\\') |
ca1d1d23 | 729 | { |
17431c60 RS |
730 | if (p == pend) break; |
731 | c = *p++; | |
732 | } | |
733 | if (p != pend && *p == '-') | |
734 | { | |
735 | p++; | |
736 | if (p == pend) break; | |
737 | while (c <= *p) | |
738 | { | |
739 | fastmap[c] = 1; | |
740 | c++; | |
741 | } | |
742 | p++; | |
ca1d1d23 | 743 | } |
17431c60 RS |
744 | else |
745 | fastmap[c] = 1; | |
ca1d1d23 | 746 | } |
ca1d1d23 JB |
747 | } |
748 | ||
9239c6c1 RS |
749 | if (syntaxp && fastmap['-'] != 0) |
750 | fastmap[' '] = 1; | |
751 | ||
ca1d1d23 JB |
752 | /* If ^ was the first character, complement the fastmap. */ |
753 | ||
754 | if (negate) | |
755 | for (i = 0; i < sizeof fastmap; i++) | |
756 | fastmap[i] ^= 1; | |
757 | ||
c1dc99a1 JB |
758 | { |
759 | int start_point = point; | |
760 | ||
761 | immediate_quit = 1; | |
17431c60 | 762 | if (syntaxp) |
c1dc99a1 | 763 | { |
17431c60 RS |
764 | |
765 | if (forwardp) | |
766 | { | |
767 | while (point < XINT (lim) | |
768 | && fastmap[(unsigned char) syntax_code_spec[(int) SYNTAX (FETCH_CHAR (point))]]) | |
769 | SET_PT (point + 1); | |
770 | } | |
771 | else | |
772 | { | |
773 | while (point > XINT (lim) | |
774 | && fastmap[(unsigned char) syntax_code_spec[(int) SYNTAX (FETCH_CHAR (point - 1))]]) | |
775 | SET_PT (point - 1); | |
776 | } | |
c1dc99a1 JB |
777 | } |
778 | else | |
779 | { | |
17431c60 RS |
780 | if (forwardp) |
781 | { | |
782 | while (point < XINT (lim) && fastmap[FETCH_CHAR (point)]) | |
783 | SET_PT (point + 1); | |
784 | } | |
785 | else | |
786 | { | |
787 | while (point > XINT (lim) && fastmap[FETCH_CHAR (point - 1)]) | |
788 | SET_PT (point - 1); | |
789 | } | |
c1dc99a1 JB |
790 | } |
791 | immediate_quit = 0; | |
792 | ||
793 | return make_number (point - start_point); | |
794 | } | |
ca1d1d23 JB |
795 | } |
796 | \f | |
797 | /* Subroutines of Lisp buffer search functions. */ | |
798 | ||
799 | static Lisp_Object | |
b819a390 | 800 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
801 | Lisp_Object string, bound, noerror, count; |
802 | int direction; | |
803 | int RE; | |
b819a390 | 804 | int posix; |
ca1d1d23 JB |
805 | { |
806 | register int np; | |
807 | int lim; | |
808 | int n = direction; | |
809 | ||
810 | if (!NILP (count)) | |
811 | { | |
812 | CHECK_NUMBER (count, 3); | |
813 | n *= XINT (count); | |
814 | } | |
815 | ||
816 | CHECK_STRING (string, 0); | |
817 | if (NILP (bound)) | |
818 | lim = n > 0 ? ZV : BEGV; | |
819 | else | |
820 | { | |
821 | CHECK_NUMBER_COERCE_MARKER (bound, 1); | |
822 | lim = XINT (bound); | |
823 | if (n > 0 ? lim < point : lim > point) | |
824 | error ("Invalid search bound (wrong side of point)"); | |
825 | if (lim > ZV) | |
826 | lim = ZV; | |
827 | if (lim < BEGV) | |
828 | lim = BEGV; | |
829 | } | |
830 | ||
831 | np = search_buffer (string, point, lim, n, RE, | |
832 | (!NILP (current_buffer->case_fold_search) | |
833 | ? XSTRING (current_buffer->case_canon_table)->data : 0), | |
834 | (!NILP (current_buffer->case_fold_search) | |
b819a390 RS |
835 | ? XSTRING (current_buffer->case_eqv_table)->data : 0), |
836 | posix); | |
ca1d1d23 JB |
837 | if (np <= 0) |
838 | { | |
839 | if (NILP (noerror)) | |
840 | return signal_failure (string); | |
841 | if (!EQ (noerror, Qt)) | |
842 | { | |
843 | if (lim < BEGV || lim > ZV) | |
844 | abort (); | |
a5f217b8 RS |
845 | SET_PT (lim); |
846 | return Qnil; | |
847 | #if 0 /* This would be clean, but maybe programs depend on | |
848 | a value of nil here. */ | |
481399bf | 849 | np = lim; |
a5f217b8 | 850 | #endif |
ca1d1d23 | 851 | } |
481399bf RS |
852 | else |
853 | return Qnil; | |
ca1d1d23 JB |
854 | } |
855 | ||
856 | if (np < BEGV || np > ZV) | |
857 | abort (); | |
858 | ||
859 | SET_PT (np); | |
860 | ||
861 | return make_number (np); | |
862 | } | |
863 | \f | |
b6d6a51c KH |
864 | static int |
865 | trivial_regexp_p (regexp) | |
866 | Lisp_Object regexp; | |
867 | { | |
868 | int len = XSTRING (regexp)->size; | |
869 | unsigned char *s = XSTRING (regexp)->data; | |
870 | unsigned char c; | |
871 | while (--len >= 0) | |
872 | { | |
873 | switch (*s++) | |
874 | { | |
875 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
876 | return 0; | |
877 | case '\\': | |
878 | if (--len < 0) | |
879 | return 0; | |
880 | switch (*s++) | |
881 | { | |
882 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
883 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
884 | case 'S': case '1': case '2': case '3': case '4': case '5': | |
885 | case '6': case '7': case '8': case '9': | |
886 | return 0; | |
887 | } | |
888 | } | |
889 | } | |
890 | return 1; | |
891 | } | |
892 | ||
ca325161 | 893 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 894 | starting at position POS and stopping at position LIM, |
b819a390 | 895 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
896 | a regular expression if RE is true. |
897 | ||
898 | If N is positive, searching is forward and LIM must be greater than POS. | |
899 | If N is negative, searching is backward and LIM must be less than POS. | |
900 | ||
901 | Returns -x if only N-x occurrences found (x > 0), | |
902 | or else the position at the beginning of the Nth occurrence | |
b819a390 RS |
903 | (if searching backward) or the end (if searching forward). |
904 | ||
905 | POSIX is nonzero if we want full backtracking (POSIX style) | |
906 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 907 | |
b819a390 RS |
908 | static int |
909 | search_buffer (string, pos, lim, n, RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
910 | Lisp_Object string; |
911 | int pos; | |
912 | int lim; | |
913 | int n; | |
914 | int RE; | |
915 | register unsigned char *trt; | |
916 | register unsigned char *inverse_trt; | |
b819a390 | 917 | int posix; |
ca1d1d23 JB |
918 | { |
919 | int len = XSTRING (string)->size; | |
920 | unsigned char *base_pat = XSTRING (string)->data; | |
921 | register int *BM_tab; | |
922 | int *BM_tab_base; | |
923 | register int direction = ((n > 0) ? 1 : -1); | |
924 | register int dirlen; | |
925 | int infinity, limit, k, stride_for_teases; | |
926 | register unsigned char *pat, *cursor, *p_limit; | |
927 | register int i, j; | |
928 | unsigned char *p1, *p2; | |
929 | int s1, s2; | |
930 | ||
931 | /* Null string is found at starting position. */ | |
3f57a499 | 932 | if (len == 0) |
ca325161 RS |
933 | { |
934 | set_search_regs (pos, 0); | |
935 | return pos; | |
936 | } | |
3f57a499 RS |
937 | |
938 | /* Searching 0 times means don't move. */ | |
939 | if (n == 0) | |
ca1d1d23 JB |
940 | return pos; |
941 | ||
b6d6a51c | 942 | if (RE && !trivial_regexp_p (string)) |
ca1d1d23 | 943 | { |
487282dc KH |
944 | struct re_pattern_buffer *bufp; |
945 | ||
b819a390 | 946 | bufp = compile_pattern (string, &search_regs, (char *) trt, posix); |
ca1d1d23 | 947 | |
ca1d1d23 JB |
948 | immediate_quit = 1; /* Quit immediately if user types ^G, |
949 | because letting this function finish | |
950 | can take too long. */ | |
951 | QUIT; /* Do a pending quit right away, | |
952 | to avoid paradoxical behavior */ | |
953 | /* Get pointers and sizes of the two strings | |
954 | that make up the visible portion of the buffer. */ | |
955 | ||
956 | p1 = BEGV_ADDR; | |
957 | s1 = GPT - BEGV; | |
958 | p2 = GAP_END_ADDR; | |
959 | s2 = ZV - GPT; | |
960 | if (s1 < 0) | |
961 | { | |
962 | p2 = p1; | |
963 | s2 = ZV - BEGV; | |
964 | s1 = 0; | |
965 | } | |
966 | if (s2 < 0) | |
967 | { | |
968 | s1 = ZV - BEGV; | |
969 | s2 = 0; | |
970 | } | |
971 | while (n < 0) | |
972 | { | |
42db823b | 973 | int val; |
487282dc | 974 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
42db823b RS |
975 | pos - BEGV, lim - pos, &search_regs, |
976 | /* Don't allow match past current point */ | |
977 | pos - BEGV); | |
ca1d1d23 | 978 | if (val == -2) |
b6d6a51c KH |
979 | { |
980 | matcher_overflow (); | |
981 | } | |
ca1d1d23 JB |
982 | if (val >= 0) |
983 | { | |
984 | j = BEGV; | |
4746118a | 985 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
986 | if (search_regs.start[i] >= 0) |
987 | { | |
988 | search_regs.start[i] += j; | |
989 | search_regs.end[i] += j; | |
990 | } | |
a3668d92 | 991 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
992 | /* Set pos to the new position. */ |
993 | pos = search_regs.start[0]; | |
994 | } | |
995 | else | |
996 | { | |
997 | immediate_quit = 0; | |
998 | return (n); | |
999 | } | |
1000 | n++; | |
1001 | } | |
1002 | while (n > 0) | |
1003 | { | |
42db823b | 1004 | int val; |
487282dc | 1005 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
42db823b RS |
1006 | pos - BEGV, lim - pos, &search_regs, |
1007 | lim - BEGV); | |
ca1d1d23 | 1008 | if (val == -2) |
b6d6a51c KH |
1009 | { |
1010 | matcher_overflow (); | |
1011 | } | |
ca1d1d23 JB |
1012 | if (val >= 0) |
1013 | { | |
1014 | j = BEGV; | |
4746118a | 1015 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1016 | if (search_regs.start[i] >= 0) |
1017 | { | |
1018 | search_regs.start[i] += j; | |
1019 | search_regs.end[i] += j; | |
1020 | } | |
a3668d92 | 1021 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1022 | pos = search_regs.end[0]; |
1023 | } | |
1024 | else | |
1025 | { | |
1026 | immediate_quit = 0; | |
1027 | return (0 - n); | |
1028 | } | |
1029 | n--; | |
1030 | } | |
1031 | immediate_quit = 0; | |
1032 | return (pos); | |
1033 | } | |
1034 | else /* non-RE case */ | |
1035 | { | |
1036 | #ifdef C_ALLOCA | |
1037 | int BM_tab_space[0400]; | |
1038 | BM_tab = &BM_tab_space[0]; | |
1039 | #else | |
1040 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
1041 | #endif | |
b6d6a51c KH |
1042 | { |
1043 | unsigned char *patbuf = (unsigned char *) alloca (len); | |
1044 | pat = patbuf; | |
1045 | while (--len >= 0) | |
1046 | { | |
1047 | /* If we got here and the RE flag is set, it's because we're | |
1048 | dealing with a regexp known to be trivial, so the backslash | |
1049 | just quotes the next character. */ | |
1050 | if (RE && *base_pat == '\\') | |
1051 | { | |
1052 | len--; | |
1053 | base_pat++; | |
1054 | } | |
1055 | *pat++ = (trt ? trt[*base_pat++] : *base_pat++); | |
1056 | } | |
1057 | len = pat - patbuf; | |
1058 | pat = base_pat = patbuf; | |
1059 | } | |
ca1d1d23 JB |
1060 | /* The general approach is that we are going to maintain that we know */ |
1061 | /* the first (closest to the present position, in whatever direction */ | |
1062 | /* we're searching) character that could possibly be the last */ | |
1063 | /* (furthest from present position) character of a valid match. We */ | |
1064 | /* advance the state of our knowledge by looking at that character */ | |
1065 | /* and seeing whether it indeed matches the last character of the */ | |
1066 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1067 | /* move our pointer (to putative last characters) as far as is */ | |
1068 | /* logically possible. This amount of movement, which I call a */ | |
1069 | /* stride, will be the length of the pattern if the actual character */ | |
1070 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1071 | /* from the last occurrence of that character to the end of the */ | |
1072 | /* pattern. */ | |
1073 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1074 | /* characters that match the last character. This allows use of only */ | |
1075 | /* a single test, a test for having gone past the end of the */ | |
1076 | /* permissible match region, to test for both possible matches (when */ | |
1077 | /* the stride goes past the end immediately) and failure to */ | |
1078 | /* match (where you get nudged past the end one stride at a time). */ | |
1079 | ||
1080 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1081 | /* is determined only by the last character of the putative match. */ | |
1082 | /* If that character does not match, we will stride the proper */ | |
1083 | /* distance to propose a match that superimposes it on the last */ | |
1084 | /* instance of a character that matches it (per trt), or misses */ | |
1085 | /* it entirely if there is none. */ | |
1086 | ||
1087 | dirlen = len * direction; | |
1088 | infinity = dirlen - (lim + pos + len + len) * direction; | |
1089 | if (direction < 0) | |
1090 | pat = (base_pat += len - 1); | |
1091 | BM_tab_base = BM_tab; | |
1092 | BM_tab += 0400; | |
1093 | j = dirlen; /* to get it in a register */ | |
1094 | /* A character that does not appear in the pattern induces a */ | |
1095 | /* stride equal to the pattern length. */ | |
1096 | while (BM_tab_base != BM_tab) | |
1097 | { | |
1098 | *--BM_tab = j; | |
1099 | *--BM_tab = j; | |
1100 | *--BM_tab = j; | |
1101 | *--BM_tab = j; | |
1102 | } | |
1103 | i = 0; | |
1104 | while (i != infinity) | |
1105 | { | |
1106 | j = pat[i]; i += direction; | |
1107 | if (i == dirlen) i = infinity; | |
1108 | if ((int) trt) | |
1109 | { | |
1110 | k = (j = trt[j]); | |
1111 | if (i == infinity) | |
1112 | stride_for_teases = BM_tab[j]; | |
1113 | BM_tab[j] = dirlen - i; | |
1114 | /* A translation table is accompanied by its inverse -- see */ | |
1115 | /* comment following downcase_table for details */ | |
1116 | while ((j = inverse_trt[j]) != k) | |
1117 | BM_tab[j] = dirlen - i; | |
1118 | } | |
1119 | else | |
1120 | { | |
1121 | if (i == infinity) | |
1122 | stride_for_teases = BM_tab[j]; | |
1123 | BM_tab[j] = dirlen - i; | |
1124 | } | |
1125 | /* stride_for_teases tells how much to stride if we get a */ | |
1126 | /* match on the far character but are subsequently */ | |
1127 | /* disappointed, by recording what the stride would have been */ | |
1128 | /* for that character if the last character had been */ | |
1129 | /* different. */ | |
1130 | } | |
1131 | infinity = dirlen - infinity; | |
1132 | pos += dirlen - ((direction > 0) ? direction : 0); | |
1133 | /* loop invariant - pos points at where last char (first char if reverse) | |
1134 | of pattern would align in a possible match. */ | |
1135 | while (n != 0) | |
1136 | { | |
b2c71fb4 KH |
1137 | /* It's been reported that some (broken) compiler thinks that |
1138 | Boolean expressions in an arithmetic context are unsigned. | |
1139 | Using an explicit ?1:0 prevents this. */ | |
1140 | if ((lim - pos - ((direction > 0) ? 1 : 0)) * direction < 0) | |
ca1d1d23 JB |
1141 | return (n * (0 - direction)); |
1142 | /* First we do the part we can by pointers (maybe nothing) */ | |
1143 | QUIT; | |
1144 | pat = base_pat; | |
1145 | limit = pos - dirlen + direction; | |
1146 | limit = ((direction > 0) | |
1147 | ? BUFFER_CEILING_OF (limit) | |
1148 | : BUFFER_FLOOR_OF (limit)); | |
1149 | /* LIMIT is now the last (not beyond-last!) value | |
1150 | POS can take on without hitting edge of buffer or the gap. */ | |
1151 | limit = ((direction > 0) | |
1152 | ? min (lim - 1, min (limit, pos + 20000)) | |
1153 | : max (lim, max (limit, pos - 20000))); | |
1154 | if ((limit - pos) * direction > 20) | |
1155 | { | |
1156 | p_limit = &FETCH_CHAR (limit); | |
1157 | p2 = (cursor = &FETCH_CHAR (pos)); | |
1158 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ | |
1159 | while (1) /* use one cursor setting as long as i can */ | |
1160 | { | |
1161 | if (direction > 0) /* worth duplicating */ | |
1162 | { | |
1163 | /* Use signed comparison if appropriate | |
1164 | to make cursor+infinity sure to be > p_limit. | |
1165 | Assuming that the buffer lies in a range of addresses | |
1166 | that are all "positive" (as ints) or all "negative", | |
1167 | either kind of comparison will work as long | |
1168 | as we don't step by infinity. So pick the kind | |
1169 | that works when we do step by infinity. */ | |
1170 | if ((int) (p_limit + infinity) > (int) p_limit) | |
1171 | while ((int) cursor <= (int) p_limit) | |
1172 | cursor += BM_tab[*cursor]; | |
1173 | else | |
1174 | while ((unsigned int) cursor <= (unsigned int) p_limit) | |
1175 | cursor += BM_tab[*cursor]; | |
1176 | } | |
1177 | else | |
1178 | { | |
1179 | if ((int) (p_limit + infinity) < (int) p_limit) | |
1180 | while ((int) cursor >= (int) p_limit) | |
1181 | cursor += BM_tab[*cursor]; | |
1182 | else | |
1183 | while ((unsigned int) cursor >= (unsigned int) p_limit) | |
1184 | cursor += BM_tab[*cursor]; | |
1185 | } | |
1186 | /* If you are here, cursor is beyond the end of the searched region. */ | |
1187 | /* This can happen if you match on the far character of the pattern, */ | |
1188 | /* because the "stride" of that character is infinity, a number able */ | |
1189 | /* to throw you well beyond the end of the search. It can also */ | |
1190 | /* happen if you fail to match within the permitted region and would */ | |
1191 | /* otherwise try a character beyond that region */ | |
1192 | if ((cursor - p_limit) * direction <= len) | |
1193 | break; /* a small overrun is genuine */ | |
1194 | cursor -= infinity; /* large overrun = hit */ | |
1195 | i = dirlen - direction; | |
1196 | if ((int) trt) | |
1197 | { | |
1198 | while ((i -= direction) + direction != 0) | |
1199 | if (pat[i] != trt[*(cursor -= direction)]) | |
1200 | break; | |
1201 | } | |
1202 | else | |
1203 | { | |
1204 | while ((i -= direction) + direction != 0) | |
1205 | if (pat[i] != *(cursor -= direction)) | |
1206 | break; | |
1207 | } | |
1208 | cursor += dirlen - i - direction; /* fix cursor */ | |
1209 | if (i + direction == 0) | |
1210 | { | |
1211 | cursor -= direction; | |
1113d9db | 1212 | |
ca325161 RS |
1213 | set_search_regs (pos + cursor - p2 + ((direction > 0) |
1214 | ? 1 - len : 0), | |
1215 | len); | |
1216 | ||
ca1d1d23 JB |
1217 | if ((n -= direction) != 0) |
1218 | cursor += dirlen; /* to resume search */ | |
1219 | else | |
1220 | return ((direction > 0) | |
1221 | ? search_regs.end[0] : search_regs.start[0]); | |
1222 | } | |
1223 | else | |
1224 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
1225 | } | |
1226 | pos += cursor - p2; | |
1227 | } | |
1228 | else | |
1229 | /* Now we'll pick up a clump that has to be done the hard */ | |
1230 | /* way because it covers a discontinuity */ | |
1231 | { | |
1232 | limit = ((direction > 0) | |
1233 | ? BUFFER_CEILING_OF (pos - dirlen + 1) | |
1234 | : BUFFER_FLOOR_OF (pos - dirlen - 1)); | |
1235 | limit = ((direction > 0) | |
1236 | ? min (limit + len, lim - 1) | |
1237 | : max (limit - len, lim)); | |
1238 | /* LIMIT is now the last value POS can have | |
1239 | and still be valid for a possible match. */ | |
1240 | while (1) | |
1241 | { | |
1242 | /* This loop can be coded for space rather than */ | |
1243 | /* speed because it will usually run only once. */ | |
1244 | /* (the reach is at most len + 21, and typically */ | |
1245 | /* does not exceed len) */ | |
1246 | while ((limit - pos) * direction >= 0) | |
1247 | pos += BM_tab[FETCH_CHAR(pos)]; | |
1248 | /* now run the same tests to distinguish going off the */ | |
eb8c3be9 | 1249 | /* end, a match or a phony match. */ |
ca1d1d23 JB |
1250 | if ((pos - limit) * direction <= len) |
1251 | break; /* ran off the end */ | |
1252 | /* Found what might be a match. | |
1253 | Set POS back to last (first if reverse) char pos. */ | |
1254 | pos -= infinity; | |
1255 | i = dirlen - direction; | |
1256 | while ((i -= direction) + direction != 0) | |
1257 | { | |
1258 | pos -= direction; | |
1259 | if (pat[i] != (((int) trt) | |
1260 | ? trt[FETCH_CHAR(pos)] | |
1261 | : FETCH_CHAR (pos))) | |
1262 | break; | |
1263 | } | |
1264 | /* Above loop has moved POS part or all the way | |
1265 | back to the first char pos (last char pos if reverse). | |
1266 | Set it once again at the last (first if reverse) char. */ | |
1267 | pos += dirlen - i- direction; | |
1268 | if (i + direction == 0) | |
1269 | { | |
1270 | pos -= direction; | |
1113d9db | 1271 | |
ca325161 RS |
1272 | set_search_regs (pos + ((direction > 0) ? 1 - len : 0), |
1273 | len); | |
1274 | ||
ca1d1d23 JB |
1275 | if ((n -= direction) != 0) |
1276 | pos += dirlen; /* to resume search */ | |
1277 | else | |
1278 | return ((direction > 0) | |
1279 | ? search_regs.end[0] : search_regs.start[0]); | |
1280 | } | |
1281 | else | |
1282 | pos += stride_for_teases; | |
1283 | } | |
1284 | } | |
1285 | /* We have done one clump. Can we continue? */ | |
1286 | if ((lim - pos) * direction < 0) | |
1287 | return ((0 - n) * direction); | |
1288 | } | |
1289 | return pos; | |
1290 | } | |
1291 | } | |
ca325161 RS |
1292 | |
1293 | /* Record beginning BEG and end BEG + LEN | |
1294 | for a match just found in the current buffer. */ | |
1295 | ||
1296 | static void | |
1297 | set_search_regs (beg, len) | |
1298 | int beg, len; | |
1299 | { | |
1300 | /* Make sure we have registers in which to store | |
1301 | the match position. */ | |
1302 | if (search_regs.num_regs == 0) | |
1303 | { | |
1304 | regoff_t *starts, *ends; | |
1305 | ||
1306 | starts = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
1307 | ends = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 1308 | search_regs.num_regs = 2; |
ca325161 RS |
1309 | } |
1310 | ||
1311 | search_regs.start[0] = beg; | |
1312 | search_regs.end[0] = beg + len; | |
a3668d92 | 1313 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 1314 | } |
ca1d1d23 JB |
1315 | \f |
1316 | /* Given a string of words separated by word delimiters, | |
1317 | compute a regexp that matches those exact words | |
1318 | separated by arbitrary punctuation. */ | |
1319 | ||
1320 | static Lisp_Object | |
1321 | wordify (string) | |
1322 | Lisp_Object string; | |
1323 | { | |
1324 | register unsigned char *p, *o; | |
1325 | register int i, len, punct_count = 0, word_count = 0; | |
1326 | Lisp_Object val; | |
1327 | ||
1328 | CHECK_STRING (string, 0); | |
1329 | p = XSTRING (string)->data; | |
1330 | len = XSTRING (string)->size; | |
1331 | ||
1332 | for (i = 0; i < len; i++) | |
1333 | if (SYNTAX (p[i]) != Sword) | |
1334 | { | |
1335 | punct_count++; | |
1336 | if (i > 0 && SYNTAX (p[i-1]) == Sword) word_count++; | |
1337 | } | |
1338 | if (SYNTAX (p[len-1]) == Sword) word_count++; | |
1339 | if (!word_count) return build_string (""); | |
1340 | ||
1341 | val = make_string (p, len - punct_count + 5 * (word_count - 1) + 4); | |
1342 | ||
1343 | o = XSTRING (val)->data; | |
1344 | *o++ = '\\'; | |
1345 | *o++ = 'b'; | |
1346 | ||
1347 | for (i = 0; i < len; i++) | |
1348 | if (SYNTAX (p[i]) == Sword) | |
1349 | *o++ = p[i]; | |
1350 | else if (i > 0 && SYNTAX (p[i-1]) == Sword && --word_count) | |
1351 | { | |
1352 | *o++ = '\\'; | |
1353 | *o++ = 'W'; | |
1354 | *o++ = '\\'; | |
1355 | *o++ = 'W'; | |
1356 | *o++ = '*'; | |
1357 | } | |
1358 | ||
1359 | *o++ = '\\'; | |
1360 | *o++ = 'b'; | |
1361 | ||
1362 | return val; | |
1363 | } | |
1364 | \f | |
1365 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
1366 | "sSearch backward: ", | |
1367 | "Search backward from point for STRING.\n\ | |
1368 | Set point to the beginning of the occurrence found, and return point.\n\ | |
1369 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1370 | The match found must not extend before that position.\n\ | |
1371 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1372 | If not nil and not t, position at limit of search and return nil.\n\ | |
1373 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1374 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1375 | (string, bound, noerror, count) | |
1376 | Lisp_Object string, bound, noerror, count; | |
1377 | { | |
b819a390 | 1378 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
1379 | } |
1380 | ||
1381 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "sSearch: ", | |
1382 | "Search forward from point for STRING.\n\ | |
1383 | Set point to the end of the occurrence found, and return point.\n\ | |
1384 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1385 | The match found must not extend after that position. nil is equivalent\n\ | |
1386 | to (point-max).\n\ | |
1387 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1388 | If not nil and not t, move to limit of search and return nil.\n\ | |
1389 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1390 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1391 | (string, bound, noerror, count) | |
1392 | Lisp_Object string, bound, noerror, count; | |
1393 | { | |
b819a390 | 1394 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
1395 | } |
1396 | ||
1397 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
1398 | "sWord search backward: ", | |
1399 | "Search backward from point for STRING, ignoring differences in punctuation.\n\ | |
1400 | Set point to the beginning of the occurrence found, and return point.\n\ | |
1401 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1402 | The match found must not extend before that position.\n\ | |
1403 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1404 | If not nil and not t, move to limit of search and return nil.\n\ | |
1405 | Optional fourth argument is repeat count--search for successive occurrences.") | |
1406 | (string, bound, noerror, count) | |
1407 | Lisp_Object string, bound, noerror, count; | |
1408 | { | |
b819a390 | 1409 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
1410 | } |
1411 | ||
1412 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
1413 | "sWord search: ", | |
1414 | "Search forward from point for STRING, ignoring differences in punctuation.\n\ | |
1415 | Set point to the end of the occurrence found, and return point.\n\ | |
1416 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1417 | The match found must not extend after that position.\n\ | |
1418 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1419 | If not nil and not t, move to limit of search and return nil.\n\ | |
1420 | Optional fourth argument is repeat count--search for successive occurrences.") | |
1421 | (string, bound, noerror, count) | |
1422 | Lisp_Object string, bound, noerror, count; | |
1423 | { | |
b819a390 | 1424 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
1425 | } |
1426 | ||
1427 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
1428 | "sRE search backward: ", | |
1429 | "Search backward from point for match for regular expression REGEXP.\n\ | |
1430 | Set point to the beginning of the match, and return point.\n\ | |
1431 | The match found is the one starting last in the buffer\n\ | |
19c0a730 | 1432 | and yet ending before the origin of the search.\n\ |
ca1d1d23 JB |
1433 | An optional second argument bounds the search; it is a buffer position.\n\ |
1434 | The match found must start at or after that position.\n\ | |
1435 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1436 | If not nil and not t, move to limit of search and return nil.\n\ | |
1437 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1438 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
19c0a730 KH |
1439 | (regexp, bound, noerror, count) |
1440 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 1441 | { |
b819a390 | 1442 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
1443 | } |
1444 | ||
1445 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
1446 | "sRE search: ", | |
1447 | "Search forward from point for regular expression REGEXP.\n\ | |
1448 | Set point to the end of the occurrence found, and return point.\n\ | |
1449 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1450 | The match found must not extend after that position.\n\ | |
1451 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1452 | If not nil and not t, move to limit of search and return nil.\n\ | |
1453 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1454 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
19c0a730 KH |
1455 | (regexp, bound, noerror, count) |
1456 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 1457 | { |
b819a390 RS |
1458 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
1459 | } | |
1460 | ||
1461 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
1462 | "sPosix search backward: ", | |
1463 | "Search backward from point for match for regular expression REGEXP.\n\ | |
1464 | Find the longest match in accord with Posix regular expression rules.\n\ | |
1465 | Set point to the beginning of the match, and return point.\n\ | |
1466 | The match found is the one starting last in the buffer\n\ | |
1467 | and yet ending before the origin of the search.\n\ | |
1468 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1469 | The match found must start at or after that position.\n\ | |
1470 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1471 | If not nil and not t, move to limit of search and return nil.\n\ | |
1472 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1473 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1474 | (regexp, bound, noerror, count) | |
1475 | Lisp_Object regexp, bound, noerror, count; | |
1476 | { | |
1477 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
1478 | } | |
1479 | ||
1480 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
1481 | "sPosix search: ", | |
1482 | "Search forward from point for regular expression REGEXP.\n\ | |
1483 | Find the longest match in accord with Posix regular expression rules.\n\ | |
1484 | Set point to the end of the occurrence found, and return point.\n\ | |
1485 | An optional second argument bounds the search; it is a buffer position.\n\ | |
1486 | The match found must not extend after that position.\n\ | |
1487 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
1488 | If not nil and not t, move to limit of search and return nil.\n\ | |
1489 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
1490 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
1491 | (regexp, bound, noerror, count) | |
1492 | Lisp_Object regexp, bound, noerror, count; | |
1493 | { | |
1494 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
1495 | } |
1496 | \f | |
080c45fd | 1497 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 4, 0, |
ca1d1d23 JB |
1498 | "Replace text matched by last search with NEWTEXT.\n\ |
1499 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text.\n\ | |
5b9cf4b2 RS |
1500 | Otherwise maybe capitalize the whole text, or maybe just word initials,\n\ |
1501 | based on the replaced text.\n\ | |
1502 | If the replaced text has only capital letters\n\ | |
1503 | and has at least one multiletter word, convert NEWTEXT to all caps.\n\ | |
1504 | If the replaced text has at least one word starting with a capital letter,\n\ | |
1505 | then capitalize each word in NEWTEXT.\n\n\ | |
ca1d1d23 JB |
1506 | If third arg LITERAL is non-nil, insert NEWTEXT literally.\n\ |
1507 | Otherwise treat `\\' as special:\n\ | |
1508 | `\\&' in NEWTEXT means substitute original matched text.\n\ | |
1509 | `\\N' means substitute what matched the Nth `\\(...\\)'.\n\ | |
1510 | If Nth parens didn't match, substitute nothing.\n\ | |
1511 | `\\\\' means insert one `\\'.\n\ | |
1113d9db | 1512 | FIXEDCASE and LITERAL are optional arguments.\n\ |
080c45fd RS |
1513 | Leaves point at end of replacement text.\n\ |
1514 | \n\ | |
1515 | The optional fourth argument STRING can be a string to modify.\n\ | |
1516 | In that case, this function creates and returns a new string\n\ | |
1517 | which is made by replacing the part of STRING that was matched.") | |
1518 | (newtext, fixedcase, literal, string) | |
1519 | Lisp_Object newtext, fixedcase, literal, string; | |
ca1d1d23 JB |
1520 | { |
1521 | enum { nochange, all_caps, cap_initial } case_action; | |
1522 | register int pos, last; | |
1523 | int some_multiletter_word; | |
97832bd0 | 1524 | int some_lowercase; |
73dc8771 | 1525 | int some_uppercase; |
208767c3 | 1526 | int some_nonuppercase_initial; |
ca1d1d23 JB |
1527 | register int c, prevc; |
1528 | int inslen; | |
1529 | ||
16fdc568 | 1530 | CHECK_STRING (newtext, 0); |
ca1d1d23 | 1531 | |
080c45fd RS |
1532 | if (! NILP (string)) |
1533 | CHECK_STRING (string, 4); | |
1534 | ||
ca1d1d23 JB |
1535 | case_action = nochange; /* We tried an initialization */ |
1536 | /* but some C compilers blew it */ | |
4746118a JB |
1537 | |
1538 | if (search_regs.num_regs <= 0) | |
1539 | error ("replace-match called before any match found"); | |
1540 | ||
080c45fd RS |
1541 | if (NILP (string)) |
1542 | { | |
1543 | if (search_regs.start[0] < BEGV | |
1544 | || search_regs.start[0] > search_regs.end[0] | |
1545 | || search_regs.end[0] > ZV) | |
1546 | args_out_of_range (make_number (search_regs.start[0]), | |
1547 | make_number (search_regs.end[0])); | |
1548 | } | |
1549 | else | |
1550 | { | |
1551 | if (search_regs.start[0] < 0 | |
1552 | || search_regs.start[0] > search_regs.end[0] | |
1553 | || search_regs.end[0] > XSTRING (string)->size) | |
1554 | args_out_of_range (make_number (search_regs.start[0]), | |
1555 | make_number (search_regs.end[0])); | |
1556 | } | |
ca1d1d23 JB |
1557 | |
1558 | if (NILP (fixedcase)) | |
1559 | { | |
1560 | /* Decide how to casify by examining the matched text. */ | |
1561 | ||
1562 | last = search_regs.end[0]; | |
1563 | prevc = '\n'; | |
1564 | case_action = all_caps; | |
1565 | ||
1566 | /* some_multiletter_word is set nonzero if any original word | |
1567 | is more than one letter long. */ | |
1568 | some_multiletter_word = 0; | |
97832bd0 | 1569 | some_lowercase = 0; |
208767c3 | 1570 | some_nonuppercase_initial = 0; |
73dc8771 | 1571 | some_uppercase = 0; |
ca1d1d23 JB |
1572 | |
1573 | for (pos = search_regs.start[0]; pos < last; pos++) | |
1574 | { | |
080c45fd RS |
1575 | if (NILP (string)) |
1576 | c = FETCH_CHAR (pos); | |
1577 | else | |
1578 | c = XSTRING (string)->data[pos]; | |
1579 | ||
ca1d1d23 JB |
1580 | if (LOWERCASEP (c)) |
1581 | { | |
1582 | /* Cannot be all caps if any original char is lower case */ | |
1583 | ||
97832bd0 | 1584 | some_lowercase = 1; |
ca1d1d23 | 1585 | if (SYNTAX (prevc) != Sword) |
208767c3 | 1586 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
1587 | else |
1588 | some_multiletter_word = 1; | |
1589 | } | |
1590 | else if (!NOCASEP (c)) | |
1591 | { | |
73dc8771 | 1592 | some_uppercase = 1; |
97832bd0 | 1593 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 1594 | ; |
97832bd0 | 1595 | else |
ca1d1d23 JB |
1596 | some_multiletter_word = 1; |
1597 | } | |
208767c3 RS |
1598 | else |
1599 | { | |
1600 | /* If the initial is a caseless word constituent, | |
1601 | treat that like a lowercase initial. */ | |
1602 | if (SYNTAX (prevc) != Sword) | |
1603 | some_nonuppercase_initial = 1; | |
1604 | } | |
ca1d1d23 JB |
1605 | |
1606 | prevc = c; | |
1607 | } | |
1608 | ||
97832bd0 RS |
1609 | /* Convert to all caps if the old text is all caps |
1610 | and has at least one multiletter word. */ | |
1611 | if (! some_lowercase && some_multiletter_word) | |
1612 | case_action = all_caps; | |
c4d460ce | 1613 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 1614 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 1615 | case_action = cap_initial; |
208767c3 | 1616 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
1617 | /* Should x -> yz, operating on X, give Yz or YZ? |
1618 | We'll assume the latter. */ | |
1619 | case_action = all_caps; | |
97832bd0 RS |
1620 | else |
1621 | case_action = nochange; | |
ca1d1d23 JB |
1622 | } |
1623 | ||
080c45fd RS |
1624 | /* Do replacement in a string. */ |
1625 | if (!NILP (string)) | |
1626 | { | |
1627 | Lisp_Object before, after; | |
1628 | ||
1629 | before = Fsubstring (string, make_number (0), | |
1630 | make_number (search_regs.start[0])); | |
1631 | after = Fsubstring (string, make_number (search_regs.end[0]), Qnil); | |
1632 | ||
1633 | /* Do case substitution into NEWTEXT if desired. */ | |
1634 | if (NILP (literal)) | |
1635 | { | |
1636 | int lastpos = -1; | |
1637 | /* We build up the substituted string in ACCUM. */ | |
1638 | Lisp_Object accum; | |
1639 | Lisp_Object middle; | |
1640 | ||
1641 | accum = Qnil; | |
1642 | ||
1643 | for (pos = 0; pos < XSTRING (newtext)->size; pos++) | |
1644 | { | |
1645 | int substart = -1; | |
1646 | int subend; | |
1647 | ||
1648 | c = XSTRING (newtext)->data[pos]; | |
1649 | if (c == '\\') | |
1650 | { | |
1651 | c = XSTRING (newtext)->data[++pos]; | |
1652 | if (c == '&') | |
1653 | { | |
1654 | substart = search_regs.start[0]; | |
1655 | subend = search_regs.end[0]; | |
1656 | } | |
1657 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') | |
1658 | { | |
1659 | if (search_regs.start[c - '0'] >= 1) | |
1660 | { | |
1661 | substart = search_regs.start[c - '0']; | |
1662 | subend = search_regs.end[c - '0']; | |
1663 | } | |
1664 | } | |
1665 | } | |
1666 | if (substart >= 0) | |
1667 | { | |
1668 | if (pos - 1 != lastpos + 1) | |
1669 | middle = Fsubstring (newtext, lastpos + 1, pos - 1); | |
1670 | else | |
1671 | middle = Qnil; | |
1672 | accum = concat3 (accum, middle, | |
1673 | Fsubstring (string, make_number (substart), | |
1674 | make_number (subend))); | |
1675 | lastpos = pos; | |
1676 | } | |
1677 | } | |
1678 | ||
1679 | if (pos != lastpos + 1) | |
1680 | middle = Fsubstring (newtext, lastpos + 1, pos); | |
1681 | else | |
1682 | middle = Qnil; | |
1683 | ||
1684 | newtext = concat2 (accum, middle); | |
1685 | } | |
1686 | ||
1687 | if (case_action == all_caps) | |
1688 | newtext = Fupcase (newtext); | |
1689 | else if (case_action == cap_initial) | |
1690 | newtext = upcase_initials (newtext); | |
1691 | ||
1692 | return concat3 (before, newtext, after); | |
1693 | } | |
1694 | ||
9a76659d JB |
1695 | /* We insert the replacement text before the old text, and then |
1696 | delete the original text. This means that markers at the | |
1697 | beginning or end of the original will float to the corresponding | |
1698 | position in the replacement. */ | |
1699 | SET_PT (search_regs.start[0]); | |
ca1d1d23 | 1700 | if (!NILP (literal)) |
16fdc568 | 1701 | Finsert_and_inherit (1, &newtext); |
ca1d1d23 JB |
1702 | else |
1703 | { | |
1704 | struct gcpro gcpro1; | |
16fdc568 | 1705 | GCPRO1 (newtext); |
ca1d1d23 | 1706 | |
16fdc568 | 1707 | for (pos = 0; pos < XSTRING (newtext)->size; pos++) |
ca1d1d23 | 1708 | { |
9a76659d JB |
1709 | int offset = point - search_regs.start[0]; |
1710 | ||
16fdc568 | 1711 | c = XSTRING (newtext)->data[pos]; |
ca1d1d23 JB |
1712 | if (c == '\\') |
1713 | { | |
16fdc568 | 1714 | c = XSTRING (newtext)->data[++pos]; |
ca1d1d23 | 1715 | if (c == '&') |
9a76659d JB |
1716 | Finsert_buffer_substring |
1717 | (Fcurrent_buffer (), | |
1718 | make_number (search_regs.start[0] + offset), | |
1719 | make_number (search_regs.end[0] + offset)); | |
78445046 | 1720 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
1721 | { |
1722 | if (search_regs.start[c - '0'] >= 1) | |
9a76659d JB |
1723 | Finsert_buffer_substring |
1724 | (Fcurrent_buffer (), | |
1725 | make_number (search_regs.start[c - '0'] + offset), | |
1726 | make_number (search_regs.end[c - '0'] + offset)); | |
ca1d1d23 JB |
1727 | } |
1728 | else | |
1729 | insert_char (c); | |
1730 | } | |
1731 | else | |
1732 | insert_char (c); | |
1733 | } | |
1734 | UNGCPRO; | |
1735 | } | |
1736 | ||
9a76659d JB |
1737 | inslen = point - (search_regs.start[0]); |
1738 | del_range (search_regs.start[0] + inslen, search_regs.end[0] + inslen); | |
ca1d1d23 JB |
1739 | |
1740 | if (case_action == all_caps) | |
1741 | Fupcase_region (make_number (point - inslen), make_number (point)); | |
1742 | else if (case_action == cap_initial) | |
1743 | upcase_initials_region (make_number (point - inslen), make_number (point)); | |
1744 | return Qnil; | |
1745 | } | |
1746 | \f | |
1747 | static Lisp_Object | |
1748 | match_limit (num, beginningp) | |
1749 | Lisp_Object num; | |
1750 | int beginningp; | |
1751 | { | |
1752 | register int n; | |
1753 | ||
1754 | CHECK_NUMBER (num, 0); | |
1755 | n = XINT (num); | |
4746118a JB |
1756 | if (n < 0 || n >= search_regs.num_regs) |
1757 | args_out_of_range (num, make_number (search_regs.num_regs)); | |
1758 | if (search_regs.num_regs <= 0 | |
1759 | || search_regs.start[n] < 0) | |
ca1d1d23 JB |
1760 | return Qnil; |
1761 | return (make_number ((beginningp) ? search_regs.start[n] | |
1762 | : search_regs.end[n])); | |
1763 | } | |
1764 | ||
1765 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
1766 | "Return position of start of text matched by last search.\n\ | |
16fdc568 BF |
1767 | NUM specifies which parenthesized expression in the last regexp.\n\ |
1768 | Value is nil if NUMth pair didn't match, or there were less than NUM pairs.\n\ | |
ca1d1d23 JB |
1769 | Zero means the entire text matched by the whole regexp or whole string.") |
1770 | (num) | |
1771 | Lisp_Object num; | |
1772 | { | |
1773 | return match_limit (num, 1); | |
1774 | } | |
1775 | ||
1776 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
1777 | "Return position of end of text matched by last search.\n\ | |
1778 | ARG, a number, specifies which parenthesized expression in the last regexp.\n\ | |
1779 | Value is nil if ARGth pair didn't match, or there were less than ARG pairs.\n\ | |
1780 | Zero means the entire text matched by the whole regexp or whole string.") | |
1781 | (num) | |
1782 | Lisp_Object num; | |
1783 | { | |
1784 | return match_limit (num, 0); | |
1785 | } | |
1786 | ||
1787 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 0, 0, | |
1788 | "Return a list containing all info on what the last search matched.\n\ | |
1789 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'.\n\ | |
1790 | All the elements are markers or nil (nil if the Nth pair didn't match)\n\ | |
1791 | if the last match was on a buffer; integers or nil if a string was matched.\n\ | |
1792 | Use `store-match-data' to reinstate the data in this list.") | |
1793 | () | |
1794 | { | |
4746118a | 1795 | Lisp_Object *data; |
ca1d1d23 JB |
1796 | int i, len; |
1797 | ||
daa37602 JB |
1798 | if (NILP (last_thing_searched)) |
1799 | error ("match-data called before any match found"); | |
1800 | ||
4746118a JB |
1801 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs) |
1802 | * sizeof (Lisp_Object)); | |
1803 | ||
ca1d1d23 | 1804 | len = -1; |
4746118a | 1805 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1806 | { |
1807 | int start = search_regs.start[i]; | |
1808 | if (start >= 0) | |
1809 | { | |
daa37602 | 1810 | if (EQ (last_thing_searched, Qt)) |
ca1d1d23 | 1811 | { |
c235cce7 KH |
1812 | XSETFASTINT (data[2 * i], start); |
1813 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 1814 | } |
0ed62dc7 | 1815 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
1816 | { |
1817 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
1818 | Fset_marker (data[2 * i], |
1819 | make_number (start), | |
1820 | last_thing_searched); | |
ca1d1d23 JB |
1821 | data[2 * i + 1] = Fmake_marker (); |
1822 | Fset_marker (data[2 * i + 1], | |
daa37602 JB |
1823 | make_number (search_regs.end[i]), |
1824 | last_thing_searched); | |
ca1d1d23 | 1825 | } |
daa37602 JB |
1826 | else |
1827 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
1828 | abort (); | |
1829 | ||
ca1d1d23 JB |
1830 | len = i; |
1831 | } | |
1832 | else | |
1833 | data[2 * i] = data [2 * i + 1] = Qnil; | |
1834 | } | |
1835 | return Flist (2 * len + 2, data); | |
1836 | } | |
1837 | ||
1838 | ||
1839 | DEFUN ("store-match-data", Fstore_match_data, Sstore_match_data, 1, 1, 0, | |
1840 | "Set internal data on last search match from elements of LIST.\n\ | |
1841 | LIST should have been created by calling `match-data' previously.") | |
1842 | (list) | |
1843 | register Lisp_Object list; | |
1844 | { | |
1845 | register int i; | |
1846 | register Lisp_Object marker; | |
1847 | ||
1848 | if (!CONSP (list) && !NILP (list)) | |
b37902c8 | 1849 | list = wrong_type_argument (Qconsp, list); |
ca1d1d23 | 1850 | |
daa37602 JB |
1851 | /* Unless we find a marker with a buffer in LIST, assume that this |
1852 | match data came from a string. */ | |
1853 | last_thing_searched = Qt; | |
1854 | ||
4746118a JB |
1855 | /* Allocate registers if they don't already exist. */ |
1856 | { | |
d084e942 | 1857 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
1858 | |
1859 | if (length > search_regs.num_regs) | |
1860 | { | |
1113d9db JB |
1861 | if (search_regs.num_regs == 0) |
1862 | { | |
1863 | search_regs.start | |
1864 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
1865 | search_regs.end | |
1866 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
1867 | } | |
4746118a | 1868 | else |
1113d9db JB |
1869 | { |
1870 | search_regs.start | |
1871 | = (regoff_t *) xrealloc (search_regs.start, | |
1872 | length * sizeof (regoff_t)); | |
1873 | search_regs.end | |
1874 | = (regoff_t *) xrealloc (search_regs.end, | |
1875 | length * sizeof (regoff_t)); | |
1876 | } | |
4746118a | 1877 | |
487282dc | 1878 | search_regs.num_regs = length; |
4746118a JB |
1879 | } |
1880 | } | |
1881 | ||
1882 | for (i = 0; i < search_regs.num_regs; i++) | |
ca1d1d23 JB |
1883 | { |
1884 | marker = Fcar (list); | |
1885 | if (NILP (marker)) | |
1886 | { | |
1887 | search_regs.start[i] = -1; | |
1888 | list = Fcdr (list); | |
1889 | } | |
1890 | else | |
1891 | { | |
0ed62dc7 | 1892 | if (MARKERP (marker)) |
daa37602 JB |
1893 | { |
1894 | if (XMARKER (marker)->buffer == 0) | |
c235cce7 | 1895 | XSETFASTINT (marker, 0); |
daa37602 | 1896 | else |
a3668d92 | 1897 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); |
daa37602 | 1898 | } |
ca1d1d23 JB |
1899 | |
1900 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
1901 | search_regs.start[i] = XINT (marker); | |
1902 | list = Fcdr (list); | |
1903 | ||
1904 | marker = Fcar (list); | |
0ed62dc7 | 1905 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
c235cce7 | 1906 | XSETFASTINT (marker, 0); |
ca1d1d23 JB |
1907 | |
1908 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
1909 | search_regs.end[i] = XINT (marker); | |
1910 | } | |
1911 | list = Fcdr (list); | |
1912 | } | |
1913 | ||
1914 | return Qnil; | |
1915 | } | |
1916 | ||
1917 | /* Quote a string to inactivate reg-expr chars */ | |
1918 | ||
1919 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
1920 | "Return a regexp string which matches exactly STRING and nothing else.") | |
1921 | (str) | |
1922 | Lisp_Object str; | |
1923 | { | |
1924 | register unsigned char *in, *out, *end; | |
1925 | register unsigned char *temp; | |
1926 | ||
1927 | CHECK_STRING (str, 0); | |
1928 | ||
1929 | temp = (unsigned char *) alloca (XSTRING (str)->size * 2); | |
1930 | ||
1931 | /* Now copy the data into the new string, inserting escapes. */ | |
1932 | ||
1933 | in = XSTRING (str)->data; | |
1934 | end = in + XSTRING (str)->size; | |
1935 | out = temp; | |
1936 | ||
1937 | for (; in != end; in++) | |
1938 | { | |
1939 | if (*in == '[' || *in == ']' | |
1940 | || *in == '*' || *in == '.' || *in == '\\' | |
1941 | || *in == '?' || *in == '+' | |
1942 | || *in == '^' || *in == '$') | |
1943 | *out++ = '\\'; | |
1944 | *out++ = *in; | |
1945 | } | |
1946 | ||
1947 | return make_string (temp, out - temp); | |
1948 | } | |
1949 | \f | |
1950 | syms_of_search () | |
1951 | { | |
1952 | register int i; | |
1953 | ||
487282dc KH |
1954 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
1955 | { | |
1956 | searchbufs[i].buf.allocated = 100; | |
1957 | searchbufs[i].buf.buffer = (unsigned char *) malloc (100); | |
1958 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; | |
1959 | searchbufs[i].regexp = Qnil; | |
1960 | staticpro (&searchbufs[i].regexp); | |
1961 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); | |
1962 | } | |
1963 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
1964 | |
1965 | Qsearch_failed = intern ("search-failed"); | |
1966 | staticpro (&Qsearch_failed); | |
1967 | Qinvalid_regexp = intern ("invalid-regexp"); | |
1968 | staticpro (&Qinvalid_regexp); | |
1969 | ||
1970 | Fput (Qsearch_failed, Qerror_conditions, | |
1971 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
1972 | Fput (Qsearch_failed, Qerror_message, | |
1973 | build_string ("Search failed")); | |
1974 | ||
1975 | Fput (Qinvalid_regexp, Qerror_conditions, | |
1976 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
1977 | Fput (Qinvalid_regexp, Qerror_message, | |
1978 | build_string ("Invalid regexp")); | |
1979 | ||
daa37602 JB |
1980 | last_thing_searched = Qnil; |
1981 | staticpro (&last_thing_searched); | |
1982 | ||
ca1d1d23 | 1983 | defsubr (&Slooking_at); |
b819a390 RS |
1984 | defsubr (&Sposix_looking_at); |
1985 | defsubr (&Sstring_match); | |
1986 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
1987 | defsubr (&Sskip_chars_forward); |
1988 | defsubr (&Sskip_chars_backward); | |
17431c60 RS |
1989 | defsubr (&Sskip_syntax_forward); |
1990 | defsubr (&Sskip_syntax_backward); | |
ca1d1d23 JB |
1991 | defsubr (&Ssearch_forward); |
1992 | defsubr (&Ssearch_backward); | |
1993 | defsubr (&Sword_search_forward); | |
1994 | defsubr (&Sword_search_backward); | |
1995 | defsubr (&Sre_search_forward); | |
1996 | defsubr (&Sre_search_backward); | |
b819a390 RS |
1997 | defsubr (&Sposix_search_forward); |
1998 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
1999 | defsubr (&Sreplace_match); |
2000 | defsubr (&Smatch_beginning); | |
2001 | defsubr (&Smatch_end); | |
2002 | defsubr (&Smatch_data); | |
2003 | defsubr (&Sstore_match_data); | |
2004 | defsubr (&Sregexp_quote); | |
2005 | } |