Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
0b5538bd | 2 | Copyright (C) 1985, 1986, 1987, 1993, 1994, 1997, 1998, 1999, 2002, 2003, |
aaef169d | 3 | 2004, 2005, 2006 Free Software Foundation, Inc. |
ca1d1d23 JB |
4 | |
5 | This file is part of GNU Emacs. | |
6 | ||
7 | GNU Emacs is free software; you can redistribute it and/or modify | |
8 | it under the terms of the GNU General Public License as published by | |
7c938215 | 9 | the Free Software Foundation; either version 2, or (at your option) |
ca1d1d23 JB |
10 | any later version. |
11 | ||
12 | GNU Emacs is distributed in the hope that it will be useful, | |
13 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | GNU General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
18 | along with GNU Emacs; see the file COPYING. If not, write to | |
4fc5845f LK |
19 | the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, |
20 | Boston, MA 02110-1301, USA. */ | |
ca1d1d23 JB |
21 | |
22 | ||
18160b98 | 23 | #include <config.h> |
ca1d1d23 JB |
24 | #include "lisp.h" |
25 | #include "syntax.h" | |
5679531d | 26 | #include "category.h" |
ca1d1d23 | 27 | #include "buffer.h" |
76eb0881 | 28 | #include "character.h" |
9169c321 | 29 | #include "region-cache.h" |
ca1d1d23 | 30 | #include "commands.h" |
9ac0d9e0 | 31 | #include "blockinput.h" |
bf1760bb | 32 | #include "intervals.h" |
4746118a | 33 | |
ca1d1d23 JB |
34 | #include <sys/types.h> |
35 | #include "regex.h" | |
36 | ||
1d288aef | 37 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 38 | |
487282dc KH |
39 | /* If the regexp is non-nil, then the buffer contains the compiled form |
40 | of that regexp, suitable for searching. */ | |
1d288aef RS |
41 | struct regexp_cache |
42 | { | |
487282dc | 43 | struct regexp_cache *next; |
ecdb561e | 44 | Lisp_Object regexp, whitespace_regexp; |
487282dc KH |
45 | struct re_pattern_buffer buf; |
46 | char fastmap[0400]; | |
b819a390 RS |
47 | /* Nonzero means regexp was compiled to do full POSIX backtracking. */ |
48 | char posix; | |
487282dc | 49 | }; |
ca1d1d23 | 50 | |
487282dc KH |
51 | /* The instances of that struct. */ |
52 | struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; | |
ca1d1d23 | 53 | |
487282dc KH |
54 | /* The head of the linked list; points to the most recently used buffer. */ |
55 | struct regexp_cache *searchbuf_head; | |
ca1d1d23 | 56 | |
ca1d1d23 | 57 | |
4746118a JB |
58 | /* Every call to re_match, etc., must pass &search_regs as the regs |
59 | argument unless you can show it is unnecessary (i.e., if re_match | |
60 | is certainly going to be called again before region-around-match | |
61 | can be called). | |
62 | ||
63 | Since the registers are now dynamically allocated, we need to make | |
64 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
65 | been allocated by checking search_regs.num_regs. |
66 | ||
67 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
68 | buffer using bits in the re_pattern_buffer. This means that whenever |
69 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
70 | allocated any registers, and will allocate new registers the next |
71 | time you call a searching or matching function. Therefore, we need | |
72 | to call re_set_registers after compiling a new pattern or after | |
73 | setting the match registers, so that the regex functions will be | |
74 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
75 | static struct re_registers search_regs; |
76 | ||
daa37602 JB |
77 | /* The buffer in which the last search was performed, or |
78 | Qt if the last search was done in a string; | |
79 | Qnil if no searching has been done yet. */ | |
80 | static Lisp_Object last_thing_searched; | |
ca1d1d23 | 81 | |
8e6208c5 | 82 | /* error condition signaled when regexp compile_pattern fails */ |
ca1d1d23 JB |
83 | |
84 | Lisp_Object Qinvalid_regexp; | |
85 | ||
41a33295 | 86 | Lisp_Object Vsearch_spaces_regexp; |
f31a9a68 | 87 | |
ca325161 | 88 | static void set_search_regs (); |
044f81f1 | 89 | static void save_search_regs (); |
facdc750 RS |
90 | static int simple_search (); |
91 | static int boyer_moore (); | |
b819a390 RS |
92 | static int search_buffer (); |
93 | ||
ca1d1d23 JB |
94 | static void |
95 | matcher_overflow () | |
96 | { | |
97 | error ("Stack overflow in regexp matcher"); | |
98 | } | |
99 | ||
b819a390 RS |
100 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
101 | PATTERN is the pattern to compile. | |
102 | CP is the place to put the result. | |
facdc750 | 103 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
104 | REGP is the structure that says where to store the "register" |
105 | values that will result from matching this pattern. | |
106 | If it is 0, we should compile the pattern not to record any | |
107 | subexpression bounds. | |
108 | POSIX is nonzero if we want full backtracking (POSIX style) | |
5679531d | 109 | for this pattern. 0 means backtrack only enough to get a valid match. |
8f924df7 | 110 | MULTIBYTE is nonzero iff a target of match is a multibyte buffer or |
bbc73b48 KH |
111 | string. |
112 | ||
113 | The behavior also depends on Vsearch_spaces_regexp. */ | |
ca1d1d23 | 114 | |
487282dc | 115 | static void |
5679531d | 116 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte) |
487282dc | 117 | struct regexp_cache *cp; |
ca1d1d23 | 118 | Lisp_Object pattern; |
facdc750 | 119 | Lisp_Object translate; |
487282dc | 120 | struct re_registers *regp; |
b819a390 | 121 | int posix; |
5679531d | 122 | int multibyte; |
ca1d1d23 | 123 | { |
d451e4db | 124 | char *val; |
b819a390 | 125 | reg_syntax_t old; |
ca1d1d23 | 126 | |
487282dc | 127 | cp->regexp = Qnil; |
59fab369 | 128 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 129 | cp->posix = posix; |
93daa011 KH |
130 | cp->buf.multibyte = STRING_MULTIBYTE (pattern); |
131 | cp->buf.target_multibyte = multibyte; | |
bbc73b48 | 132 | cp->whitespace_regexp = Vsearch_spaces_regexp; |
9ac0d9e0 | 133 | BLOCK_INPUT; |
fb4a568d | 134 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 135 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
bbc73b48 KH |
136 | re_set_whitespace_regexp (NILP (Vsearch_spaces_regexp) ? NULL |
137 | : SDATA (Vsearch_spaces_regexp)); | |
138 | ||
8f924df7 KH |
139 | val = (char *) re_compile_pattern ((char *) SDATA (pattern), |
140 | SBYTES (pattern), &cp->buf); | |
bbc73b48 KH |
141 | |
142 | re_set_whitespace_regexp (NULL); | |
143 | ||
b819a390 | 144 | re_set_syntax (old); |
9ac0d9e0 | 145 | UNBLOCK_INPUT; |
ca1d1d23 | 146 | if (val) |
487282dc | 147 | Fsignal (Qinvalid_regexp, Fcons (build_string (val), Qnil)); |
1113d9db | 148 | |
487282dc | 149 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
150 | } |
151 | ||
6efc7887 RS |
152 | /* Shrink each compiled regexp buffer in the cache |
153 | to the size actually used right now. | |
154 | This is called from garbage collection. */ | |
155 | ||
156 | void | |
157 | shrink_regexp_cache () | |
158 | { | |
a968f437 | 159 | struct regexp_cache *cp; |
6efc7887 RS |
160 | |
161 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
162 | { | |
163 | cp->buf.allocated = cp->buf.used; | |
164 | cp->buf.buffer | |
b23c0a83 | 165 | = (unsigned char *) xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
166 | } |
167 | } | |
168 | ||
487282dc | 169 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
170 | the cache. |
171 | PATTERN is the pattern to compile. | |
facdc750 | 172 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
173 | REGP is the structure that says where to store the "register" |
174 | values that will result from matching this pattern. | |
175 | If it is 0, we should compile the pattern not to record any | |
176 | subexpression bounds. | |
177 | POSIX is nonzero if we want full backtracking (POSIX style) | |
178 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
179 | |
180 | struct re_pattern_buffer * | |
0c8533c6 | 181 | compile_pattern (pattern, regp, translate, posix, multibyte) |
487282dc KH |
182 | Lisp_Object pattern; |
183 | struct re_registers *regp; | |
facdc750 | 184 | Lisp_Object translate; |
0c8533c6 | 185 | int posix, multibyte; |
487282dc KH |
186 | { |
187 | struct regexp_cache *cp, **cpp; | |
188 | ||
189 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
190 | { | |
191 | cp = *cpp; | |
f1b9c7c1 KR |
192 | /* Entries are initialized to nil, and may be set to nil by |
193 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
49a5f770 KR |
194 | string accessors in those cases. However, compile_pattern_1 |
195 | is only applied to the cache entry we pick here to reuse. So | |
196 | nil should never appear before a non-nil entry. */ | |
7c752c80 | 197 | if (NILP (cp->regexp)) |
f1b9c7c1 | 198 | goto compile_it; |
d5db4077 | 199 | if (SCHARS (cp->regexp) == SCHARS (pattern) |
cf69b13e | 200 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 201 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 202 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d | 203 | && cp->posix == posix |
bbc73b48 KH |
204 | && cp->buf.target_multibyte == multibyte |
205 | && !NILP (Fequal (cp->whitespace_regexp, Vsearch_spaces_regexp))) | |
487282dc KH |
206 | break; |
207 | ||
f1b9c7c1 KR |
208 | /* If we're at the end of the cache, compile into the nil cell |
209 | we found, or the last (least recently used) cell with a | |
210 | string value. */ | |
487282dc KH |
211 | if (cp->next == 0) |
212 | { | |
f1b9c7c1 | 213 | compile_it: |
5679531d | 214 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte); |
487282dc KH |
215 | break; |
216 | } | |
217 | } | |
218 | ||
219 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
220 | either because we found it in the cache or because we just compiled it. | |
221 | Move it to the front of the queue to mark it as most recently used. */ | |
222 | *cpp = cp->next; | |
223 | cp->next = searchbuf_head; | |
224 | searchbuf_head = cp; | |
1113d9db | 225 | |
6639708c RS |
226 | /* Advise the searching functions about the space we have allocated |
227 | for register data. */ | |
228 | if (regp) | |
229 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
230 | ||
487282dc | 231 | return &cp->buf; |
ca1d1d23 JB |
232 | } |
233 | ||
234 | /* Error condition used for failing searches */ | |
235 | Lisp_Object Qsearch_failed; | |
236 | ||
237 | Lisp_Object | |
238 | signal_failure (arg) | |
239 | Lisp_Object arg; | |
240 | { | |
241 | Fsignal (Qsearch_failed, Fcons (arg, Qnil)); | |
242 | return Qnil; | |
243 | } | |
244 | \f | |
b819a390 RS |
245 | static Lisp_Object |
246 | looking_at_1 (string, posix) | |
ca1d1d23 | 247 | Lisp_Object string; |
b819a390 | 248 | int posix; |
ca1d1d23 JB |
249 | { |
250 | Lisp_Object val; | |
251 | unsigned char *p1, *p2; | |
252 | int s1, s2; | |
253 | register int i; | |
487282dc | 254 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 255 | |
7074fde6 FP |
256 | if (running_asynch_code) |
257 | save_search_regs (); | |
258 | ||
b7826503 | 259 | CHECK_STRING (string); |
487282dc KH |
260 | bufp = compile_pattern (string, &search_regs, |
261 | (!NILP (current_buffer->case_fold_search) | |
0190922f | 262 | ? current_buffer->case_canon_table : Qnil), |
0c8533c6 RS |
263 | posix, |
264 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 JB |
265 | |
266 | immediate_quit = 1; | |
267 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
268 | ||
269 | /* Get pointers and sizes of the two strings | |
270 | that make up the visible portion of the buffer. */ | |
271 | ||
272 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 273 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 274 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 275 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
276 | if (s1 < 0) |
277 | { | |
278 | p2 = p1; | |
fa8ed3e0 | 279 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
280 | s1 = 0; |
281 | } | |
282 | if (s2 < 0) | |
283 | { | |
fa8ed3e0 | 284 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
285 | s2 = 0; |
286 | } | |
8bb43c28 RS |
287 | |
288 | re_match_object = Qnil; | |
177c0ea7 | 289 | |
487282dc | 290 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
fa8ed3e0 RS |
291 | PT_BYTE - BEGV_BYTE, &search_regs, |
292 | ZV_BYTE - BEGV_BYTE); | |
de182d70 | 293 | immediate_quit = 0; |
177c0ea7 | 294 | |
ca1d1d23 JB |
295 | if (i == -2) |
296 | matcher_overflow (); | |
297 | ||
298 | val = (0 <= i ? Qt : Qnil); | |
fa8ed3e0 RS |
299 | if (i >= 0) |
300 | for (i = 0; i < search_regs.num_regs; i++) | |
301 | if (search_regs.start[i] >= 0) | |
302 | { | |
303 | search_regs.start[i] | |
304 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
305 | search_regs.end[i] | |
306 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
307 | } | |
a3668d92 | 308 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
309 | return val; |
310 | } | |
311 | ||
b819a390 | 312 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
313 | doc: /* Return t if text after point matches regular expression REGEXP. |
314 | This function modifies the match data that `match-beginning', | |
315 | `match-end' and `match-data' access; save and restore the match | |
316 | data if you want to preserve them. */) | |
317 | (regexp) | |
94f94972 | 318 | Lisp_Object regexp; |
b819a390 | 319 | { |
94f94972 | 320 | return looking_at_1 (regexp, 0); |
b819a390 RS |
321 | } |
322 | ||
323 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
324 | doc: /* Return t if text after point matches regular expression REGEXP. |
325 | Find the longest match, in accord with Posix regular expression rules. | |
326 | This function modifies the match data that `match-beginning', | |
327 | `match-end' and `match-data' access; save and restore the match | |
328 | data if you want to preserve them. */) | |
329 | (regexp) | |
94f94972 | 330 | Lisp_Object regexp; |
b819a390 | 331 | { |
94f94972 | 332 | return looking_at_1 (regexp, 1); |
b819a390 RS |
333 | } |
334 | \f | |
335 | static Lisp_Object | |
336 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 337 | Lisp_Object regexp, string, start; |
b819a390 | 338 | int posix; |
ca1d1d23 JB |
339 | { |
340 | int val; | |
487282dc | 341 | struct re_pattern_buffer *bufp; |
0c8533c6 RS |
342 | int pos, pos_byte; |
343 | int i; | |
ca1d1d23 | 344 | |
7074fde6 FP |
345 | if (running_asynch_code) |
346 | save_search_regs (); | |
347 | ||
b7826503 PJ |
348 | CHECK_STRING (regexp); |
349 | CHECK_STRING (string); | |
ca1d1d23 JB |
350 | |
351 | if (NILP (start)) | |
0c8533c6 | 352 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
353 | else |
354 | { | |
d5db4077 | 355 | int len = SCHARS (string); |
ca1d1d23 | 356 | |
b7826503 | 357 | CHECK_NUMBER (start); |
0c8533c6 RS |
358 | pos = XINT (start); |
359 | if (pos < 0 && -pos <= len) | |
360 | pos = len + pos; | |
361 | else if (0 > pos || pos > len) | |
ca1d1d23 | 362 | args_out_of_range (string, start); |
0c8533c6 | 363 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
364 | } |
365 | ||
487282dc KH |
366 | bufp = compile_pattern (regexp, &search_regs, |
367 | (!NILP (current_buffer->case_fold_search) | |
0190922f | 368 | ? current_buffer->case_canon_table : Qnil), |
0c8533c6 RS |
369 | posix, |
370 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 371 | immediate_quit = 1; |
8bb43c28 | 372 | re_match_object = string; |
177c0ea7 | 373 | |
d5db4077 KR |
374 | val = re_search (bufp, (char *) SDATA (string), |
375 | SBYTES (string), pos_byte, | |
376 | SBYTES (string) - pos_byte, | |
ca1d1d23 JB |
377 | &search_regs); |
378 | immediate_quit = 0; | |
daa37602 | 379 | last_thing_searched = Qt; |
ca1d1d23 JB |
380 | if (val == -2) |
381 | matcher_overflow (); | |
382 | if (val < 0) return Qnil; | |
0c8533c6 RS |
383 | |
384 | for (i = 0; i < search_regs.num_regs; i++) | |
385 | if (search_regs.start[i] >= 0) | |
386 | { | |
387 | search_regs.start[i] | |
388 | = string_byte_to_char (string, search_regs.start[i]); | |
389 | search_regs.end[i] | |
390 | = string_byte_to_char (string, search_regs.end[i]); | |
391 | } | |
392 | ||
393 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 394 | } |
e59a8453 | 395 | |
b819a390 | 396 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 | 397 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
b85acc4b | 398 | Matching ignores case if `case-fold-search' is non-nil. |
8c1a1077 PJ |
399 | If third arg START is non-nil, start search at that index in STRING. |
400 | For index of first char beyond the match, do (match-end 0). | |
401 | `match-end' and `match-beginning' also give indices of substrings | |
2bd2f32d RS |
402 | matched by parenthesis constructs in the pattern. |
403 | ||
404 | You can use the function `match-string' to extract the substrings | |
405 | matched by the parenthesis constructions in REGEXP. */) | |
8c1a1077 | 406 | (regexp, string, start) |
b819a390 RS |
407 | Lisp_Object regexp, string, start; |
408 | { | |
409 | return string_match_1 (regexp, string, start, 0); | |
410 | } | |
411 | ||
412 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
413 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
414 | Find the longest match, in accord with Posix regular expression rules. | |
415 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
416 | If third arg START is non-nil, start search at that index in STRING. | |
417 | For index of first char beyond the match, do (match-end 0). | |
418 | `match-end' and `match-beginning' also give indices of substrings | |
419 | matched by parenthesis constructs in the pattern. */) | |
420 | (regexp, string, start) | |
b819a390 RS |
421 | Lisp_Object regexp, string, start; |
422 | { | |
423 | return string_match_1 (regexp, string, start, 1); | |
424 | } | |
425 | ||
e59a8453 RS |
426 | /* Match REGEXP against STRING, searching all of STRING, |
427 | and return the index of the match, or negative on failure. | |
428 | This does not clobber the match data. */ | |
429 | ||
430 | int | |
431 | fast_string_match (regexp, string) | |
432 | Lisp_Object regexp, string; | |
433 | { | |
434 | int val; | |
487282dc | 435 | struct re_pattern_buffer *bufp; |
e59a8453 | 436 | |
facdc750 RS |
437 | bufp = compile_pattern (regexp, 0, Qnil, |
438 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 439 | immediate_quit = 1; |
8bb43c28 | 440 | re_match_object = string; |
177c0ea7 | 441 | |
d5db4077 KR |
442 | val = re_search (bufp, (char *) SDATA (string), |
443 | SBYTES (string), 0, | |
444 | SBYTES (string), 0); | |
e59a8453 RS |
445 | immediate_quit = 0; |
446 | return val; | |
447 | } | |
5679531d KH |
448 | |
449 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
450 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
451 | This does not clobber the match data. |
452 | We assume that STRING contains single-byte characters. */ | |
5679531d KH |
453 | |
454 | extern Lisp_Object Vascii_downcase_table; | |
455 | ||
456 | int | |
b4577c63 | 457 | fast_c_string_match_ignore_case (regexp, string) |
5679531d | 458 | Lisp_Object regexp; |
96b80561 | 459 | const char *string; |
5679531d KH |
460 | { |
461 | int val; | |
462 | struct re_pattern_buffer *bufp; | |
463 | int len = strlen (string); | |
464 | ||
0c8533c6 | 465 | regexp = string_make_unibyte (regexp); |
b4577c63 | 466 | re_match_object = Qt; |
5679531d | 467 | bufp = compile_pattern (regexp, 0, |
0190922f | 468 | Vascii_canon_table, 0, |
f8bd51c4 | 469 | 0); |
5679531d KH |
470 | immediate_quit = 1; |
471 | val = re_search (bufp, string, len, 0, len, 0); | |
472 | immediate_quit = 0; | |
473 | return val; | |
474 | } | |
be5f4dfb KH |
475 | |
476 | /* Like fast_string_match but ignore case. */ | |
477 | ||
478 | int | |
479 | fast_string_match_ignore_case (regexp, string) | |
480 | Lisp_Object regexp, string; | |
481 | { | |
482 | int val; | |
483 | struct re_pattern_buffer *bufp; | |
484 | ||
0190922f | 485 | bufp = compile_pattern (regexp, 0, Vascii_canon_table, |
be5f4dfb KH |
486 | 0, STRING_MULTIBYTE (string)); |
487 | immediate_quit = 1; | |
488 | re_match_object = string; | |
489 | ||
490 | val = re_search (bufp, (char *) SDATA (string), | |
491 | SBYTES (string), 0, | |
492 | SBYTES (string), 0); | |
493 | immediate_quit = 0; | |
494 | return val; | |
495 | } | |
ca1d1d23 | 496 | \f |
9169c321 JB |
497 | /* The newline cache: remembering which sections of text have no newlines. */ |
498 | ||
499 | /* If the user has requested newline caching, make sure it's on. | |
500 | Otherwise, make sure it's off. | |
501 | This is our cheezy way of associating an action with the change of | |
502 | state of a buffer-local variable. */ | |
503 | static void | |
504 | newline_cache_on_off (buf) | |
505 | struct buffer *buf; | |
506 | { | |
507 | if (NILP (buf->cache_long_line_scans)) | |
508 | { | |
509 | /* It should be off. */ | |
510 | if (buf->newline_cache) | |
511 | { | |
512 | free_region_cache (buf->newline_cache); | |
513 | buf->newline_cache = 0; | |
514 | } | |
515 | } | |
516 | else | |
517 | { | |
518 | /* It should be on. */ | |
519 | if (buf->newline_cache == 0) | |
520 | buf->newline_cache = new_region_cache (); | |
521 | } | |
522 | } | |
523 | ||
524 | \f | |
525 | /* Search for COUNT instances of the character TARGET between START and END. | |
526 | ||
527 | If COUNT is positive, search forwards; END must be >= START. | |
528 | If COUNT is negative, search backwards for the -COUNTth instance; | |
529 | END must be <= START. | |
530 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
531 | ||
532 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
533 | direction indicated by COUNT. | |
ffd56f97 JB |
534 | |
535 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
a9f2a45f | 536 | position past the COUNTth match. Note that for reverse motion |
5bfe95c9 | 537 | this is not the same as the usual convention for Emacs motion commands. |
ffd56f97 | 538 | |
9169c321 JB |
539 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
540 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 541 | |
087a5f81 RS |
542 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
543 | except when inside redisplay. */ | |
544 | ||
dfcf069d | 545 | int |
9169c321 JB |
546 | scan_buffer (target, start, end, count, shortage, allow_quit) |
547 | register int target; | |
548 | int start, end; | |
549 | int count; | |
550 | int *shortage; | |
087a5f81 | 551 | int allow_quit; |
ca1d1d23 | 552 | { |
9169c321 | 553 | struct region_cache *newline_cache; |
177c0ea7 | 554 | int direction; |
ffd56f97 | 555 | |
9169c321 JB |
556 | if (count > 0) |
557 | { | |
558 | direction = 1; | |
559 | if (! end) end = ZV; | |
560 | } | |
561 | else | |
562 | { | |
563 | direction = -1; | |
564 | if (! end) end = BEGV; | |
565 | } | |
ffd56f97 | 566 | |
9169c321 JB |
567 | newline_cache_on_off (current_buffer); |
568 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
569 | |
570 | if (shortage != 0) | |
571 | *shortage = 0; | |
572 | ||
087a5f81 | 573 | immediate_quit = allow_quit; |
ca1d1d23 | 574 | |
ffd56f97 | 575 | if (count > 0) |
9169c321 | 576 | while (start != end) |
ca1d1d23 | 577 | { |
9169c321 JB |
578 | /* Our innermost scanning loop is very simple; it doesn't know |
579 | about gaps, buffer ends, or the newline cache. ceiling is | |
580 | the position of the last character before the next such | |
581 | obstacle --- the last character the dumb search loop should | |
582 | examine. */ | |
fa8ed3e0 RS |
583 | int ceiling_byte = CHAR_TO_BYTE (end) - 1; |
584 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 585 | int tem; |
9169c321 JB |
586 | |
587 | /* If we're looking for a newline, consult the newline cache | |
588 | to see where we can avoid some scanning. */ | |
589 | if (target == '\n' && newline_cache) | |
590 | { | |
591 | int next_change; | |
592 | immediate_quit = 0; | |
593 | while (region_cache_forward | |
fa8ed3e0 RS |
594 | (current_buffer, newline_cache, start_byte, &next_change)) |
595 | start_byte = next_change; | |
cbe0db0d | 596 | immediate_quit = allow_quit; |
9169c321 | 597 | |
fa8ed3e0 RS |
598 | /* START should never be after END. */ |
599 | if (start_byte > ceiling_byte) | |
600 | start_byte = ceiling_byte; | |
9169c321 JB |
601 | |
602 | /* Now the text after start is an unknown region, and | |
603 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 604 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
605 | } |
606 | ||
607 | /* The dumb loop can only scan text stored in contiguous | |
608 | bytes. BUFFER_CEILING_OF returns the last character | |
609 | position that is contiguous, so the ceiling is the | |
610 | position after that. */ | |
67ce527d KH |
611 | tem = BUFFER_CEILING_OF (start_byte); |
612 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
613 | |
614 | { | |
177c0ea7 | 615 | /* The termination address of the dumb loop. */ |
fa8ed3e0 RS |
616 | register unsigned char *ceiling_addr |
617 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
618 | register unsigned char *cursor | |
619 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
620 | unsigned char *base = cursor; |
621 | ||
622 | while (cursor < ceiling_addr) | |
623 | { | |
624 | unsigned char *scan_start = cursor; | |
625 | ||
626 | /* The dumb loop. */ | |
627 | while (*cursor != target && ++cursor < ceiling_addr) | |
628 | ; | |
629 | ||
630 | /* If we're looking for newlines, cache the fact that | |
631 | the region from start to cursor is free of them. */ | |
632 | if (target == '\n' && newline_cache) | |
633 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
634 | start_byte + scan_start - base, |
635 | start_byte + cursor - base); | |
9169c321 JB |
636 | |
637 | /* Did we find the target character? */ | |
638 | if (cursor < ceiling_addr) | |
639 | { | |
640 | if (--count == 0) | |
641 | { | |
642 | immediate_quit = 0; | |
fa8ed3e0 | 643 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
644 | } |
645 | cursor++; | |
646 | } | |
647 | } | |
648 | ||
fa8ed3e0 | 649 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 650 | } |
ca1d1d23 JB |
651 | } |
652 | else | |
9169c321 JB |
653 | while (start > end) |
654 | { | |
655 | /* The last character to check before the next obstacle. */ | |
fa8ed3e0 RS |
656 | int ceiling_byte = CHAR_TO_BYTE (end); |
657 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 658 | int tem; |
9169c321 JB |
659 | |
660 | /* Consult the newline cache, if appropriate. */ | |
661 | if (target == '\n' && newline_cache) | |
662 | { | |
663 | int next_change; | |
664 | immediate_quit = 0; | |
665 | while (region_cache_backward | |
fa8ed3e0 RS |
666 | (current_buffer, newline_cache, start_byte, &next_change)) |
667 | start_byte = next_change; | |
cbe0db0d | 668 | immediate_quit = allow_quit; |
9169c321 JB |
669 | |
670 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
671 | if (start_byte <= ceiling_byte) |
672 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
673 | |
674 | /* Now the text before start is an unknown region, and | |
675 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 676 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
677 | } |
678 | ||
679 | /* Stop scanning before the gap. */ | |
67ce527d KH |
680 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
681 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
682 | |
683 | { | |
684 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
685 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
686 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
687 | unsigned char *base = cursor; |
688 | ||
689 | while (cursor >= ceiling_addr) | |
690 | { | |
691 | unsigned char *scan_start = cursor; | |
692 | ||
693 | while (*cursor != target && --cursor >= ceiling_addr) | |
694 | ; | |
695 | ||
696 | /* If we're looking for newlines, cache the fact that | |
697 | the region from after the cursor to start is free of them. */ | |
698 | if (target == '\n' && newline_cache) | |
699 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
700 | start_byte + cursor - base, |
701 | start_byte + scan_start - base); | |
9169c321 JB |
702 | |
703 | /* Did we find the target character? */ | |
704 | if (cursor >= ceiling_addr) | |
705 | { | |
706 | if (++count >= 0) | |
707 | { | |
708 | immediate_quit = 0; | |
fa8ed3e0 | 709 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
710 | } |
711 | cursor--; | |
712 | } | |
713 | } | |
714 | ||
fa8ed3e0 | 715 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
716 | } |
717 | } | |
718 | ||
ca1d1d23 JB |
719 | immediate_quit = 0; |
720 | if (shortage != 0) | |
ffd56f97 | 721 | *shortage = count * direction; |
9169c321 | 722 | return start; |
ca1d1d23 | 723 | } |
fa8ed3e0 RS |
724 | \f |
725 | /* Search for COUNT instances of a line boundary, which means either a | |
726 | newline or (if selective display enabled) a carriage return. | |
727 | Start at START. If COUNT is negative, search backwards. | |
728 | ||
729 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
730 | ||
731 | If we find COUNT instances. we position after (always after, | |
732 | even if scanning backwards) the COUNTth match, and return 0. | |
733 | ||
734 | If we don't find COUNT instances before reaching the end of the | |
735 | buffer (or the beginning, if scanning backwards), we return | |
736 | the number of line boundaries left unfound, and position at | |
737 | the limit we bumped up against. | |
738 | ||
739 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 740 | except in special cases. */ |
ca1d1d23 | 741 | |
63fa018d | 742 | int |
fa8ed3e0 RS |
743 | scan_newline (start, start_byte, limit, limit_byte, count, allow_quit) |
744 | int start, start_byte; | |
745 | int limit, limit_byte; | |
746 | register int count; | |
747 | int allow_quit; | |
63fa018d | 748 | { |
fa8ed3e0 RS |
749 | int direction = ((count > 0) ? 1 : -1); |
750 | ||
751 | register unsigned char *cursor; | |
752 | unsigned char *base; | |
753 | ||
754 | register int ceiling; | |
755 | register unsigned char *ceiling_addr; | |
756 | ||
d5d57b92 RS |
757 | int old_immediate_quit = immediate_quit; |
758 | ||
fa8ed3e0 RS |
759 | /* The code that follows is like scan_buffer |
760 | but checks for either newline or carriage return. */ | |
761 | ||
d5d57b92 RS |
762 | if (allow_quit) |
763 | immediate_quit++; | |
fa8ed3e0 RS |
764 | |
765 | start_byte = CHAR_TO_BYTE (start); | |
766 | ||
767 | if (count > 0) | |
768 | { | |
769 | while (start_byte < limit_byte) | |
770 | { | |
771 | ceiling = BUFFER_CEILING_OF (start_byte); | |
772 | ceiling = min (limit_byte - 1, ceiling); | |
773 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
774 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
775 | while (1) | |
776 | { | |
777 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
778 | ; | |
779 | ||
780 | if (cursor != ceiling_addr) | |
781 | { | |
782 | if (--count == 0) | |
783 | { | |
d5d57b92 | 784 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
785 | start_byte = start_byte + cursor - base + 1; |
786 | start = BYTE_TO_CHAR (start_byte); | |
787 | TEMP_SET_PT_BOTH (start, start_byte); | |
788 | return 0; | |
789 | } | |
790 | else | |
791 | if (++cursor == ceiling_addr) | |
792 | break; | |
793 | } | |
794 | else | |
795 | break; | |
796 | } | |
797 | start_byte += cursor - base; | |
798 | } | |
799 | } | |
800 | else | |
801 | { | |
fa8ed3e0 RS |
802 | while (start_byte > limit_byte) |
803 | { | |
804 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
805 | ceiling = max (limit_byte, ceiling); | |
806 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
807 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
808 | while (1) | |
809 | { | |
810 | while (--cursor != ceiling_addr && *cursor != '\n') | |
811 | ; | |
812 | ||
813 | if (cursor != ceiling_addr) | |
814 | { | |
815 | if (++count == 0) | |
816 | { | |
d5d57b92 | 817 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
818 | /* Return the position AFTER the match we found. */ |
819 | start_byte = start_byte + cursor - base + 1; | |
820 | start = BYTE_TO_CHAR (start_byte); | |
821 | TEMP_SET_PT_BOTH (start, start_byte); | |
822 | return 0; | |
823 | } | |
824 | } | |
825 | else | |
826 | break; | |
827 | } | |
828 | /* Here we add 1 to compensate for the last decrement | |
829 | of CURSOR, which took it past the valid range. */ | |
830 | start_byte += cursor - base + 1; | |
831 | } | |
832 | } | |
833 | ||
834 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 835 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
836 | |
837 | return count * direction; | |
63fa018d RS |
838 | } |
839 | ||
ca1d1d23 | 840 | int |
fa8ed3e0 | 841 | find_next_newline_no_quit (from, cnt) |
ca1d1d23 JB |
842 | register int from, cnt; |
843 | { | |
fa8ed3e0 | 844 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
9169c321 JB |
845 | } |
846 | ||
9169c321 JB |
847 | /* Like find_next_newline, but returns position before the newline, |
848 | not after, and only search up to TO. This isn't just | |
849 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 850 | |
9169c321 JB |
851 | int |
852 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 853 | int from, to, cnt; |
9169c321 JB |
854 | { |
855 | int shortage; | |
856 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
857 | ||
858 | if (shortage == 0) | |
859 | pos--; | |
177c0ea7 | 860 | |
9169c321 | 861 | return pos; |
ca1d1d23 JB |
862 | } |
863 | \f | |
ca1d1d23 JB |
864 | /* Subroutines of Lisp buffer search functions. */ |
865 | ||
866 | static Lisp_Object | |
b819a390 | 867 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
868 | Lisp_Object string, bound, noerror, count; |
869 | int direction; | |
870 | int RE; | |
b819a390 | 871 | int posix; |
ca1d1d23 JB |
872 | { |
873 | register int np; | |
9f43ad85 | 874 | int lim, lim_byte; |
ca1d1d23 JB |
875 | int n = direction; |
876 | ||
877 | if (!NILP (count)) | |
878 | { | |
b7826503 | 879 | CHECK_NUMBER (count); |
ca1d1d23 JB |
880 | n *= XINT (count); |
881 | } | |
882 | ||
b7826503 | 883 | CHECK_STRING (string); |
ca1d1d23 | 884 | if (NILP (bound)) |
9f43ad85 RS |
885 | { |
886 | if (n > 0) | |
887 | lim = ZV, lim_byte = ZV_BYTE; | |
888 | else | |
889 | lim = BEGV, lim_byte = BEGV_BYTE; | |
890 | } | |
ca1d1d23 JB |
891 | else |
892 | { | |
b7826503 | 893 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 894 | lim = XINT (bound); |
6ec8bbd2 | 895 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
896 | error ("Invalid search bound (wrong side of point)"); |
897 | if (lim > ZV) | |
9f43ad85 | 898 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 899 | else if (lim < BEGV) |
9f43ad85 | 900 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
901 | else |
902 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
903 | } |
904 | ||
9f43ad85 | 905 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
ca1d1d23 | 906 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 907 | ? current_buffer->case_canon_table |
3135e9fd | 908 | : Qnil), |
ca1d1d23 | 909 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 910 | ? current_buffer->case_eqv_table |
3135e9fd | 911 | : Qnil), |
b819a390 | 912 | posix); |
ca1d1d23 JB |
913 | if (np <= 0) |
914 | { | |
915 | if (NILP (noerror)) | |
916 | return signal_failure (string); | |
917 | if (!EQ (noerror, Qt)) | |
918 | { | |
919 | if (lim < BEGV || lim > ZV) | |
920 | abort (); | |
9f43ad85 | 921 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
922 | return Qnil; |
923 | #if 0 /* This would be clean, but maybe programs depend on | |
924 | a value of nil here. */ | |
481399bf | 925 | np = lim; |
a5f217b8 | 926 | #endif |
ca1d1d23 | 927 | } |
481399bf RS |
928 | else |
929 | return Qnil; | |
ca1d1d23 JB |
930 | } |
931 | ||
932 | if (np < BEGV || np > ZV) | |
933 | abort (); | |
934 | ||
935 | SET_PT (np); | |
936 | ||
937 | return make_number (np); | |
938 | } | |
939 | \f | |
fa8ed3e0 RS |
940 | /* Return 1 if REGEXP it matches just one constant string. */ |
941 | ||
b6d6a51c KH |
942 | static int |
943 | trivial_regexp_p (regexp) | |
944 | Lisp_Object regexp; | |
945 | { | |
d5db4077 KR |
946 | int len = SBYTES (regexp); |
947 | unsigned char *s = SDATA (regexp); | |
b6d6a51c KH |
948 | while (--len >= 0) |
949 | { | |
950 | switch (*s++) | |
951 | { | |
952 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
953 | return 0; | |
954 | case '\\': | |
955 | if (--len < 0) | |
956 | return 0; | |
957 | switch (*s++) | |
958 | { | |
959 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
960 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 961 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 962 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 963 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
964 | case '6': case '7': case '8': case '9': |
965 | return 0; | |
966 | } | |
967 | } | |
968 | } | |
969 | return 1; | |
970 | } | |
971 | ||
ca325161 | 972 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 973 | starting at position POS and stopping at position LIM, |
b819a390 | 974 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
975 | a regular expression if RE is true. |
976 | ||
977 | If N is positive, searching is forward and LIM must be greater than POS. | |
978 | If N is negative, searching is backward and LIM must be less than POS. | |
979 | ||
facdc750 | 980 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 981 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
982 | (if searching backward) or the end (if searching forward). |
983 | ||
984 | POSIX is nonzero if we want full backtracking (POSIX style) | |
985 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 986 | |
aff2ce94 RS |
987 | #define TRANSLATE(out, trt, d) \ |
988 | do \ | |
989 | { \ | |
990 | if (! NILP (trt)) \ | |
991 | { \ | |
992 | Lisp_Object temp; \ | |
993 | temp = Faref (trt, make_number (d)); \ | |
994 | if (INTEGERP (temp)) \ | |
995 | out = XINT (temp); \ | |
996 | else \ | |
997 | out = d; \ | |
998 | } \ | |
999 | else \ | |
1000 | out = d; \ | |
1001 | } \ | |
1002 | while (0) | |
facdc750 | 1003 | |
b819a390 | 1004 | static int |
9f43ad85 RS |
1005 | search_buffer (string, pos, pos_byte, lim, lim_byte, n, |
1006 | RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
1007 | Lisp_Object string; |
1008 | int pos; | |
9f43ad85 | 1009 | int pos_byte; |
ca1d1d23 | 1010 | int lim; |
9f43ad85 | 1011 | int lim_byte; |
ca1d1d23 JB |
1012 | int n; |
1013 | int RE; | |
facdc750 RS |
1014 | Lisp_Object trt; |
1015 | Lisp_Object inverse_trt; | |
b819a390 | 1016 | int posix; |
ca1d1d23 | 1017 | { |
d5db4077 KR |
1018 | int len = SCHARS (string); |
1019 | int len_byte = SBYTES (string); | |
facdc750 | 1020 | register int i; |
ca1d1d23 | 1021 | |
7074fde6 FP |
1022 | if (running_asynch_code) |
1023 | save_search_regs (); | |
1024 | ||
a7e4cdde | 1025 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1026 | /* Null string is found at starting position. */ |
a7e4cdde | 1027 | if (len == 0 || n == 0) |
ca325161 | 1028 | { |
0353b28f | 1029 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1030 | return pos; |
1031 | } | |
3f57a499 | 1032 | |
41a33295 | 1033 | if (RE && !(trivial_regexp_p (string) && NILP (Vsearch_spaces_regexp))) |
ca1d1d23 | 1034 | { |
facdc750 RS |
1035 | unsigned char *p1, *p2; |
1036 | int s1, s2; | |
487282dc KH |
1037 | struct re_pattern_buffer *bufp; |
1038 | ||
0c8533c6 RS |
1039 | bufp = compile_pattern (string, &search_regs, trt, posix, |
1040 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 | 1041 | |
ca1d1d23 JB |
1042 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1043 | because letting this function finish | |
1044 | can take too long. */ | |
1045 | QUIT; /* Do a pending quit right away, | |
1046 | to avoid paradoxical behavior */ | |
1047 | /* Get pointers and sizes of the two strings | |
1048 | that make up the visible portion of the buffer. */ | |
1049 | ||
1050 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1051 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1052 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1053 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1054 | if (s1 < 0) |
1055 | { | |
1056 | p2 = p1; | |
fa8ed3e0 | 1057 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1058 | s1 = 0; |
1059 | } | |
1060 | if (s2 < 0) | |
1061 | { | |
fa8ed3e0 | 1062 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1063 | s2 = 0; |
1064 | } | |
8bb43c28 | 1065 | re_match_object = Qnil; |
177c0ea7 | 1066 | |
ca1d1d23 JB |
1067 | while (n < 0) |
1068 | { | |
42db823b | 1069 | int val; |
487282dc | 1070 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1071 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1072 | &search_regs, | |
42db823b | 1073 | /* Don't allow match past current point */ |
4996330b | 1074 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1075 | if (val == -2) |
b6d6a51c KH |
1076 | { |
1077 | matcher_overflow (); | |
1078 | } | |
ca1d1d23 JB |
1079 | if (val >= 0) |
1080 | { | |
26aff150 | 1081 | pos_byte = search_regs.start[0] + BEGV_BYTE; |
4746118a | 1082 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1083 | if (search_regs.start[i] >= 0) |
1084 | { | |
fa8ed3e0 RS |
1085 | search_regs.start[i] |
1086 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1087 | search_regs.end[i] | |
1088 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1089 | } |
a3668d92 | 1090 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1091 | /* Set pos to the new position. */ |
1092 | pos = search_regs.start[0]; | |
1093 | } | |
1094 | else | |
1095 | { | |
1096 | immediate_quit = 0; | |
1097 | return (n); | |
1098 | } | |
1099 | n++; | |
1100 | } | |
1101 | while (n > 0) | |
1102 | { | |
42db823b | 1103 | int val; |
487282dc | 1104 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1105 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1106 | &search_regs, | |
1107 | lim_byte - BEGV_BYTE); | |
ca1d1d23 | 1108 | if (val == -2) |
b6d6a51c KH |
1109 | { |
1110 | matcher_overflow (); | |
1111 | } | |
ca1d1d23 JB |
1112 | if (val >= 0) |
1113 | { | |
26aff150 | 1114 | pos_byte = search_regs.end[0] + BEGV_BYTE; |
4746118a | 1115 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1116 | if (search_regs.start[i] >= 0) |
1117 | { | |
fa8ed3e0 RS |
1118 | search_regs.start[i] |
1119 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1120 | search_regs.end[i] | |
1121 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1122 | } |
a3668d92 | 1123 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1124 | pos = search_regs.end[0]; |
1125 | } | |
1126 | else | |
1127 | { | |
1128 | immediate_quit = 0; | |
1129 | return (0 - n); | |
1130 | } | |
1131 | n--; | |
1132 | } | |
1133 | immediate_quit = 0; | |
1134 | return (pos); | |
1135 | } | |
1136 | else /* non-RE case */ | |
1137 | { | |
facdc750 RS |
1138 | unsigned char *raw_pattern, *pat; |
1139 | int raw_pattern_size; | |
1140 | int raw_pattern_size_byte; | |
1141 | unsigned char *patbuf; | |
1142 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); | |
a967ed62 | 1143 | unsigned char *base_pat; |
49e44e81 KH |
1144 | /* Set to positive if we find a non-ASCII char that need |
1145 | translation. Otherwise set to zero later. */ | |
1146 | int char_base = -1; | |
040272ce | 1147 | int boyer_moore_ok = 1; |
facdc750 RS |
1148 | |
1149 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1150 | We must convert PATTERN to match that, or we will not really | |
1151 | find things right. */ | |
1152 | ||
1153 | if (multibyte == STRING_MULTIBYTE (string)) | |
1154 | { | |
d5db4077 KR |
1155 | raw_pattern = (unsigned char *) SDATA (string); |
1156 | raw_pattern_size = SCHARS (string); | |
1157 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1158 | } |
1159 | else if (multibyte) | |
1160 | { | |
d5db4077 | 1161 | raw_pattern_size = SCHARS (string); |
facdc750 | 1162 | raw_pattern_size_byte |
d5db4077 | 1163 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1164 | raw_pattern_size); |
7276d3d8 | 1165 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1166 | copy_text (SDATA (string), raw_pattern, |
1167 | SCHARS (string), 0, 1); | |
facdc750 RS |
1168 | } |
1169 | else | |
1170 | { | |
1171 | /* Converting multibyte to single-byte. | |
1172 | ||
1173 | ??? Perhaps this conversion should be done in a special way | |
1174 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1175 | so that only the multibyte chars which really correspond to | |
1176 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1177 | raw_pattern_size = SCHARS (string); |
1178 | raw_pattern_size_byte = SCHARS (string); | |
7276d3d8 | 1179 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
1180 | copy_text (SDATA (string), raw_pattern, |
1181 | SBYTES (string), 1, 0); | |
facdc750 RS |
1182 | } |
1183 | ||
1184 | /* Copy and optionally translate the pattern. */ | |
1185 | len = raw_pattern_size; | |
1186 | len_byte = raw_pattern_size_byte; | |
f5f7578b | 1187 | patbuf = (unsigned char *) alloca (len * MAX_MULTIBYTE_LENGTH); |
facdc750 RS |
1188 | pat = patbuf; |
1189 | base_pat = raw_pattern; | |
1190 | if (multibyte) | |
1191 | { | |
a17ae96a KH |
1192 | /* Fill patbuf by translated characters in STRING while |
1193 | checking if we can use boyer-moore search. If TRT is | |
1194 | non-nil, we can use boyer-moore search only if TRT can be | |
1195 | represented by the byte array of 256 elements. For that, | |
1196 | all non-ASCII case-equivalents of all case-senstive | |
1197 | characters in STRING must belong to the same charset and | |
1198 | row. */ | |
1199 | ||
facdc750 RS |
1200 | while (--len >= 0) |
1201 | { | |
a17ae96a | 1202 | unsigned char str_base[MAX_MULTIBYTE_LENGTH], *str; |
aff2ce94 | 1203 | int c, translated, inverse; |
a17ae96a | 1204 | int in_charlen, charlen; |
facdc750 RS |
1205 | |
1206 | /* If we got here and the RE flag is set, it's because we're | |
1207 | dealing with a regexp known to be trivial, so the backslash | |
1208 | just quotes the next character. */ | |
1209 | if (RE && *base_pat == '\\') | |
1210 | { | |
1211 | len--; | |
62221f22 | 1212 | raw_pattern_size--; |
facdc750 RS |
1213 | len_byte--; |
1214 | base_pat++; | |
1215 | } | |
1216 | ||
1217 | c = STRING_CHAR_AND_LENGTH (base_pat, len_byte, in_charlen); | |
040272ce | 1218 | |
a17ae96a | 1219 | if (NILP (trt)) |
facdc750 | 1220 | { |
a17ae96a KH |
1221 | str = base_pat; |
1222 | charlen = in_charlen; | |
1223 | } | |
1224 | else | |
1225 | { | |
1226 | /* Translate the character. */ | |
1227 | TRANSLATE (translated, trt, c); | |
1228 | charlen = CHAR_STRING (translated, str_base); | |
1229 | str = str_base; | |
1230 | ||
1231 | /* Check if C has any other case-equivalents. */ | |
1232 | TRANSLATE (inverse, inverse_trt, c); | |
1233 | /* If so, check if we can use boyer-moore. */ | |
1234 | if (c != inverse && boyer_moore_ok) | |
1235 | { | |
1236 | /* Check if all equivalents belong to the same | |
1237 | group of characters. Note that the check of C | |
49e44e81 KH |
1238 | itself is done by the last iteration. */ |
1239 | int this_char_base = -1; | |
1240 | ||
1241 | while (boyer_moore_ok) | |
a17ae96a | 1242 | { |
49e44e81 | 1243 | if (ASCII_BYTE_P (inverse)) |
a17ae96a | 1244 | { |
49e44e81 KH |
1245 | if (this_char_base > 0) |
1246 | boyer_moore_ok = 0; | |
1247 | else | |
a17ae96a | 1248 | { |
49e44e81 KH |
1249 | this_char_base = 0; |
1250 | if (char_base < 0) | |
1251 | char_base = this_char_base; | |
a17ae96a KH |
1252 | } |
1253 | } | |
49e44e81 KH |
1254 | else if (CHAR_BYTE8_P (inverse)) |
1255 | /* Boyer-moore search can't handle a | |
1256 | translation of an eight-bit | |
1257 | character. */ | |
1258 | boyer_moore_ok = 0; | |
1259 | else if (this_char_base < 0) | |
1260 | { | |
1261 | this_char_base = inverse & ~0x3F; | |
1262 | if (char_base < 0) | |
1263 | char_base = this_char_base; | |
1264 | else if (char_base > 0 | |
1265 | && this_char_base != char_base) | |
1266 | boyer_moore_ok = 0; | |
1267 | } | |
1268 | else if ((inverse & ~0x3F) != this_char_base) | |
1269 | boyer_moore_ok = 0; | |
a17ae96a KH |
1270 | if (c == inverse) |
1271 | break; | |
1272 | TRANSLATE (inverse, inverse_trt, inverse); | |
1273 | } | |
1274 | } | |
aff2ce94 | 1275 | } |
49e44e81 KH |
1276 | if (char_base < 0) |
1277 | char_base = 0; | |
facdc750 RS |
1278 | |
1279 | /* Store this character into the translated pattern. */ | |
a17ae96a KH |
1280 | bcopy (str, pat, charlen); |
1281 | pat += charlen; | |
facdc750 RS |
1282 | base_pat += in_charlen; |
1283 | len_byte -= in_charlen; | |
1284 | } | |
1285 | } | |
1286 | else | |
1287 | { | |
040272ce | 1288 | /* Unibyte buffer. */ |
a17ae96a | 1289 | char_base = 0; |
facdc750 RS |
1290 | while (--len >= 0) |
1291 | { | |
040272ce | 1292 | int c, translated; |
facdc750 RS |
1293 | |
1294 | /* If we got here and the RE flag is set, it's because we're | |
1295 | dealing with a regexp known to be trivial, so the backslash | |
1296 | just quotes the next character. */ | |
1297 | if (RE && *base_pat == '\\') | |
1298 | { | |
1299 | len--; | |
0190922f | 1300 | raw_pattern_size--; |
facdc750 RS |
1301 | base_pat++; |
1302 | } | |
1303 | c = *base_pat++; | |
aff2ce94 | 1304 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1305 | *pat++ = translated; |
1306 | } | |
1307 | } | |
1308 | ||
1309 | len_byte = pat - patbuf; | |
1310 | len = raw_pattern_size; | |
1311 | pat = base_pat = patbuf; | |
1312 | ||
040272ce | 1313 | if (boyer_moore_ok) |
facdc750 | 1314 | return boyer_moore (n, pat, len, len_byte, trt, inverse_trt, |
aff2ce94 | 1315 | pos, pos_byte, lim, lim_byte, |
a17ae96a | 1316 | char_base); |
facdc750 RS |
1317 | else |
1318 | return simple_search (n, pat, len, len_byte, trt, | |
1319 | pos, pos_byte, lim, lim_byte); | |
1320 | } | |
1321 | } | |
1322 | \f | |
1323 | /* Do a simple string search N times for the string PAT, | |
1324 | whose length is LEN/LEN_BYTE, | |
1325 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1326 | TRT is the translation table. | |
f8bd51c4 | 1327 | |
facdc750 RS |
1328 | Return the character position where the match is found. |
1329 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1330 | |
facdc750 RS |
1331 | This kind of search works regardless of what is in PAT and |
1332 | regardless of what is in TRT. It is used in cases where | |
1333 | boyer_moore cannot work. */ | |
1334 | ||
1335 | static int | |
1336 | simple_search (n, pat, len, len_byte, trt, pos, pos_byte, lim, lim_byte) | |
1337 | int n; | |
1338 | unsigned char *pat; | |
1339 | int len, len_byte; | |
1340 | Lisp_Object trt; | |
1341 | int pos, pos_byte; | |
1342 | int lim, lim_byte; | |
1343 | { | |
1344 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); | |
ab228c24 | 1345 | int forward = n > 0; |
49e44e81 KH |
1346 | /* Number of buffer bytes matched. Note that this may be different |
1347 | from len_byte in a multibyte buffer. */ | |
1348 | int match_byte; | |
facdc750 RS |
1349 | |
1350 | if (lim > pos && multibyte) | |
1351 | while (n > 0) | |
1352 | { | |
1353 | while (1) | |
f8bd51c4 | 1354 | { |
facdc750 RS |
1355 | /* Try matching at position POS. */ |
1356 | int this_pos = pos; | |
1357 | int this_pos_byte = pos_byte; | |
1358 | int this_len = len; | |
1359 | int this_len_byte = len_byte; | |
1360 | unsigned char *p = pat; | |
1361 | if (pos + len > lim) | |
1362 | goto stop; | |
1363 | ||
1364 | while (this_len > 0) | |
1365 | { | |
1366 | int charlen, buf_charlen; | |
ab228c24 | 1367 | int pat_ch, buf_ch; |
facdc750 | 1368 | |
ab228c24 | 1369 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1370 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1371 | ZV_BYTE - this_pos_byte, | |
1372 | buf_charlen); | |
aff2ce94 | 1373 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1374 | |
1375 | if (buf_ch != pat_ch) | |
1376 | break; | |
ab228c24 RS |
1377 | |
1378 | this_len_byte -= charlen; | |
1379 | this_len--; | |
1380 | p += charlen; | |
1381 | ||
1382 | this_pos_byte += buf_charlen; | |
1383 | this_pos++; | |
facdc750 RS |
1384 | } |
1385 | ||
1386 | if (this_len == 0) | |
1387 | { | |
49e44e81 | 1388 | match_byte = this_pos_byte - pos_byte; |
facdc750 | 1389 | pos += len; |
49e44e81 | 1390 | pos_byte += match_byte; |
facdc750 RS |
1391 | break; |
1392 | } | |
1393 | ||
1394 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1395 | } |
facdc750 RS |
1396 | |
1397 | n--; | |
1398 | } | |
1399 | else if (lim > pos) | |
1400 | while (n > 0) | |
1401 | { | |
1402 | while (1) | |
f8bd51c4 | 1403 | { |
facdc750 RS |
1404 | /* Try matching at position POS. */ |
1405 | int this_pos = pos; | |
1406 | int this_len = len; | |
1407 | unsigned char *p = pat; | |
1408 | ||
1409 | if (pos + len > lim) | |
1410 | goto stop; | |
1411 | ||
1412 | while (this_len > 0) | |
1413 | { | |
1414 | int pat_ch = *p++; | |
1415 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1416 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1417 | |
1418 | if (buf_ch != pat_ch) | |
1419 | break; | |
ab228c24 RS |
1420 | |
1421 | this_len--; | |
1422 | this_pos++; | |
facdc750 RS |
1423 | } |
1424 | ||
1425 | if (this_len == 0) | |
1426 | { | |
49e44e81 | 1427 | match_byte = len; |
facdc750 RS |
1428 | pos += len; |
1429 | break; | |
1430 | } | |
1431 | ||
1432 | pos++; | |
f8bd51c4 | 1433 | } |
facdc750 RS |
1434 | |
1435 | n--; | |
1436 | } | |
1437 | /* Backwards search. */ | |
1438 | else if (lim < pos && multibyte) | |
1439 | while (n < 0) | |
1440 | { | |
1441 | while (1) | |
f8bd51c4 | 1442 | { |
facdc750 RS |
1443 | /* Try matching at position POS. */ |
1444 | int this_pos = pos - len; | |
130b729c | 1445 | int this_pos_byte; |
facdc750 RS |
1446 | int this_len = len; |
1447 | int this_len_byte = len_byte; | |
1448 | unsigned char *p = pat; | |
1449 | ||
1450 | if (pos - len < lim) | |
1451 | goto stop; | |
130b729c | 1452 | this_pos_byte = CHAR_TO_BYTE (this_pos); |
49e44e81 | 1453 | match_byte = pos_byte - this_pos_byte; |
facdc750 RS |
1454 | |
1455 | while (this_len > 0) | |
1456 | { | |
1457 | int charlen, buf_charlen; | |
ab228c24 | 1458 | int pat_ch, buf_ch; |
facdc750 | 1459 | |
ab228c24 | 1460 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1461 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1462 | ZV_BYTE - this_pos_byte, | |
1463 | buf_charlen); | |
aff2ce94 | 1464 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1465 | |
1466 | if (buf_ch != pat_ch) | |
1467 | break; | |
ab228c24 RS |
1468 | |
1469 | this_len_byte -= charlen; | |
1470 | this_len--; | |
1471 | p += charlen; | |
1472 | this_pos_byte += buf_charlen; | |
1473 | this_pos++; | |
facdc750 RS |
1474 | } |
1475 | ||
1476 | if (this_len == 0) | |
1477 | { | |
1478 | pos -= len; | |
49e44e81 | 1479 | pos_byte -= match_byte; |
facdc750 RS |
1480 | break; |
1481 | } | |
1482 | ||
1483 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1484 | } |
1485 | ||
facdc750 RS |
1486 | n++; |
1487 | } | |
1488 | else if (lim < pos) | |
1489 | while (n < 0) | |
1490 | { | |
1491 | while (1) | |
b6d6a51c | 1492 | { |
facdc750 RS |
1493 | /* Try matching at position POS. */ |
1494 | int this_pos = pos - len; | |
1495 | int this_len = len; | |
1496 | unsigned char *p = pat; | |
1497 | ||
1498 | if (pos - len < lim) | |
1499 | goto stop; | |
1500 | ||
1501 | while (this_len > 0) | |
1502 | { | |
1503 | int pat_ch = *p++; | |
1504 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1505 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1506 | |
1507 | if (buf_ch != pat_ch) | |
1508 | break; | |
ab228c24 RS |
1509 | this_len--; |
1510 | this_pos++; | |
facdc750 RS |
1511 | } |
1512 | ||
1513 | if (this_len == 0) | |
b6d6a51c | 1514 | { |
49e44e81 | 1515 | match_byte = len; |
facdc750 RS |
1516 | pos -= len; |
1517 | break; | |
b6d6a51c | 1518 | } |
facdc750 RS |
1519 | |
1520 | pos--; | |
b6d6a51c | 1521 | } |
facdc750 RS |
1522 | |
1523 | n++; | |
b6d6a51c | 1524 | } |
facdc750 RS |
1525 | |
1526 | stop: | |
1527 | if (n == 0) | |
aff2ce94 | 1528 | { |
ab228c24 | 1529 | if (forward) |
49e44e81 | 1530 | set_search_regs ((multibyte ? pos_byte : pos) - match_byte, match_byte); |
ab228c24 | 1531 | else |
49e44e81 | 1532 | set_search_regs (multibyte ? pos_byte : pos, match_byte); |
aff2ce94 RS |
1533 | |
1534 | return pos; | |
1535 | } | |
facdc750 RS |
1536 | else if (n > 0) |
1537 | return -n; | |
1538 | else | |
1539 | return n; | |
1540 | } | |
1541 | \f | |
0190922f | 1542 | /* Do Boyer-Moore search N times for the string BASE_PAT, |
facdc750 RS |
1543 | whose length is LEN/LEN_BYTE, |
1544 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1545 | DIRECTION says which direction we search in. | |
1546 | TRT and INVERSE_TRT are translation tables. | |
0190922f | 1547 | Characters in PAT are already translated by TRT. |
facdc750 | 1548 | |
0190922f KH |
1549 | This kind of search works if all the characters in BASE_PAT that |
1550 | have nontrivial translation are the same aside from the last byte. | |
1551 | This makes it possible to translate just the last byte of a | |
1552 | character, and do so after just a simple test of the context. | |
a17ae96a | 1553 | CHAR_BASE is nonzero iff there is such a non-ASCII character. |
facdc750 RS |
1554 | |
1555 | If that criterion is not satisfied, do not call this function. */ | |
1556 | ||
1557 | static int | |
1558 | boyer_moore (n, base_pat, len, len_byte, trt, inverse_trt, | |
a17ae96a | 1559 | pos, pos_byte, lim, lim_byte, char_base) |
facdc750 RS |
1560 | int n; |
1561 | unsigned char *base_pat; | |
1562 | int len, len_byte; | |
1563 | Lisp_Object trt; | |
1564 | Lisp_Object inverse_trt; | |
1565 | int pos, pos_byte; | |
1566 | int lim, lim_byte; | |
a17ae96a | 1567 | int char_base; |
facdc750 RS |
1568 | { |
1569 | int direction = ((n > 0) ? 1 : -1); | |
1570 | register int dirlen; | |
a968f437 | 1571 | int infinity, limit, stride_for_teases = 0; |
facdc750 RS |
1572 | register int *BM_tab; |
1573 | int *BM_tab_base; | |
177c0ea7 | 1574 | register unsigned char *cursor, *p_limit; |
facdc750 | 1575 | register int i, j; |
cb6792d2 | 1576 | unsigned char *pat, *pat_end; |
facdc750 RS |
1577 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); |
1578 | ||
1579 | unsigned char simple_translate[0400]; | |
0190922f | 1580 | /* These are set to the preceding bytes of a byte to be translated |
49e44e81 | 1581 | if char_base is nonzero. As the maximum byte length of a |
a17ae96a | 1582 | multibyte character is 5, we have to check at most four previous |
0190922f KH |
1583 | bytes. */ |
1584 | int translate_prev_byte1 = 0; | |
1585 | int translate_prev_byte2 = 0; | |
1586 | int translate_prev_byte3 = 0; | |
a17ae96a | 1587 | int translate_prev_byte4 = 0; |
facdc750 RS |
1588 | |
1589 | #ifdef C_ALLOCA | |
1590 | int BM_tab_space[0400]; | |
1591 | BM_tab = &BM_tab_space[0]; | |
1592 | #else | |
1593 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
1594 | #endif | |
1595 | /* The general approach is that we are going to maintain that we know */ | |
1596 | /* the first (closest to the present position, in whatever direction */ | |
1597 | /* we're searching) character that could possibly be the last */ | |
1598 | /* (furthest from present position) character of a valid match. We */ | |
1599 | /* advance the state of our knowledge by looking at that character */ | |
1600 | /* and seeing whether it indeed matches the last character of the */ | |
1601 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1602 | /* move our pointer (to putative last characters) as far as is */ | |
1603 | /* logically possible. This amount of movement, which I call a */ | |
1604 | /* stride, will be the length of the pattern if the actual character */ | |
1605 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1606 | /* from the last occurrence of that character to the end of the */ | |
1607 | /* pattern. */ | |
1608 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1609 | /* characters that match the last character. This allows use of only */ | |
1610 | /* a single test, a test for having gone past the end of the */ | |
1611 | /* permissible match region, to test for both possible matches (when */ | |
1612 | /* the stride goes past the end immediately) and failure to */ | |
177c0ea7 | 1613 | /* match (where you get nudged past the end one stride at a time). */ |
facdc750 RS |
1614 | |
1615 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1616 | /* is determined only by the last character of the putative match. */ | |
1617 | /* If that character does not match, we will stride the proper */ | |
1618 | /* distance to propose a match that superimposes it on the last */ | |
1619 | /* instance of a character that matches it (per trt), or misses */ | |
177c0ea7 | 1620 | /* it entirely if there is none. */ |
facdc750 RS |
1621 | |
1622 | dirlen = len_byte * direction; | |
1623 | infinity = dirlen - (lim_byte + pos_byte + len_byte + len_byte) * direction; | |
cb6792d2 RS |
1624 | |
1625 | /* Record position after the end of the pattern. */ | |
1626 | pat_end = base_pat + len_byte; | |
1627 | /* BASE_PAT points to a character that we start scanning from. | |
1628 | It is the first character in a forward search, | |
1629 | the last character in a backward search. */ | |
facdc750 | 1630 | if (direction < 0) |
cb6792d2 RS |
1631 | base_pat = pat_end - 1; |
1632 | ||
facdc750 RS |
1633 | BM_tab_base = BM_tab; |
1634 | BM_tab += 0400; | |
1635 | j = dirlen; /* to get it in a register */ | |
1636 | /* A character that does not appear in the pattern induces a */ | |
1637 | /* stride equal to the pattern length. */ | |
1638 | while (BM_tab_base != BM_tab) | |
1639 | { | |
1640 | *--BM_tab = j; | |
1641 | *--BM_tab = j; | |
1642 | *--BM_tab = j; | |
1643 | *--BM_tab = j; | |
1644 | } | |
1645 | ||
1646 | /* We use this for translation, instead of TRT itself. | |
1647 | We fill this in to handle the characters that actually | |
1648 | occur in the pattern. Others don't matter anyway! */ | |
1649 | bzero (simple_translate, sizeof simple_translate); | |
1650 | for (i = 0; i < 0400; i++) | |
1651 | simple_translate[i] = i; | |
1652 | ||
a17ae96a | 1653 | if (char_base) |
0190922f | 1654 | { |
a17ae96a | 1655 | /* Setup translate_prev_byte1/2/3/4 from CHAR_BASE. Only a |
0190922f | 1656 | byte following them are the target of translation. */ |
0190922f | 1657 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
a17ae96a | 1658 | int len = CHAR_STRING (char_base, str); |
0190922f KH |
1659 | |
1660 | translate_prev_byte1 = str[len - 2]; | |
1661 | if (len > 2) | |
1662 | { | |
1663 | translate_prev_byte2 = str[len - 3]; | |
1664 | if (len > 3) | |
a17ae96a KH |
1665 | { |
1666 | translate_prev_byte3 = str[len - 4]; | |
1667 | if (len > 4) | |
1668 | translate_prev_byte4 = str[len - 5]; | |
1669 | } | |
0190922f KH |
1670 | } |
1671 | } | |
1672 | ||
facdc750 RS |
1673 | i = 0; |
1674 | while (i != infinity) | |
1675 | { | |
cb6792d2 | 1676 | unsigned char *ptr = base_pat + i; |
facdc750 RS |
1677 | i += direction; |
1678 | if (i == dirlen) | |
1679 | i = infinity; | |
1680 | if (! NILP (trt)) | |
ca1d1d23 | 1681 | { |
49e44e81 KH |
1682 | /* If the byte currently looking at is the last of a |
1683 | character to check case-equivalents, set CH to that | |
1684 | character. An ASCII character and a non-ASCII character | |
1685 | matching with CHAR_BASE are to be checked. */ | |
a17ae96a KH |
1686 | int ch = -1; |
1687 | ||
1688 | if (ASCII_BYTE_P (*ptr) || ! multibyte) | |
1689 | ch = *ptr; | |
49e44e81 | 1690 | else if (char_base |
86dc6ccb | 1691 | && ((pat_end - ptr) == 1 || CHAR_HEAD_P (ptr[1]))) |
ca1d1d23 | 1692 | { |
49e44e81 KH |
1693 | unsigned char *charstart = ptr - 1; |
1694 | ||
1695 | while (! (CHAR_HEAD_P (*charstart))) | |
1696 | charstart--; | |
1697 | ch = STRING_CHAR (charstart, ptr - charstart + 1); | |
a17ae96a KH |
1698 | if (char_base != (ch & ~0x3F)) |
1699 | ch = -1; | |
ca1d1d23 | 1700 | } |
facdc750 | 1701 | |
6786a107 | 1702 | if (ch >= 0400) |
49e44e81 KH |
1703 | j = (ch & 0x3F) | 0200; |
1704 | else | |
1705 | j = *ptr; | |
1706 | ||
facdc750 RS |
1707 | if (i == infinity) |
1708 | stride_for_teases = BM_tab[j]; | |
ab228c24 | 1709 | |
facdc750 RS |
1710 | BM_tab[j] = dirlen - i; |
1711 | /* A translation table is accompanied by its inverse -- see */ | |
177c0ea7 | 1712 | /* comment following downcase_table for details */ |
a17ae96a | 1713 | if (ch >= 0) |
ab228c24 RS |
1714 | { |
1715 | int starting_ch = ch; | |
49e44e81 | 1716 | int starting_j = j; |
a17ae96a | 1717 | |
ab228c24 RS |
1718 | while (1) |
1719 | { | |
1720 | TRANSLATE (ch, inverse_trt, ch); | |
6786a107 | 1721 | if (ch >= 0400) |
49e44e81 | 1722 | j = (ch & 0x3F) | 0200; |
ab228c24 | 1723 | else |
a17ae96a | 1724 | j = ch; |
ab228c24 RS |
1725 | |
1726 | /* For all the characters that map into CH, | |
1727 | set up simple_translate to map the last byte | |
1728 | into STARTING_J. */ | |
1729 | simple_translate[j] = starting_j; | |
1730 | if (ch == starting_ch) | |
1731 | break; | |
1732 | BM_tab[j] = dirlen - i; | |
1733 | } | |
1734 | } | |
facdc750 RS |
1735 | } |
1736 | else | |
1737 | { | |
1738 | j = *ptr; | |
1739 | ||
1740 | if (i == infinity) | |
1741 | stride_for_teases = BM_tab[j]; | |
1742 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1743 | } |
facdc750 RS |
1744 | /* stride_for_teases tells how much to stride if we get a */ |
1745 | /* match on the far character but are subsequently */ | |
1746 | /* disappointed, by recording what the stride would have been */ | |
1747 | /* for that character if the last character had been */ | |
1748 | /* different. */ | |
1749 | } | |
1750 | infinity = dirlen - infinity; | |
1751 | pos_byte += dirlen - ((direction > 0) ? direction : 0); | |
1752 | /* loop invariant - POS_BYTE points at where last char (first | |
1753 | char if reverse) of pattern would align in a possible match. */ | |
1754 | while (n != 0) | |
1755 | { | |
1756 | int tail_end; | |
1757 | unsigned char *tail_end_ptr; | |
1758 | ||
1759 | /* It's been reported that some (broken) compiler thinks that | |
1760 | Boolean expressions in an arithmetic context are unsigned. | |
1761 | Using an explicit ?1:0 prevents this. */ | |
1762 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1763 | < 0) | |
1764 | return (n * (0 - direction)); | |
1765 | /* First we do the part we can by pointers (maybe nothing) */ | |
1766 | QUIT; | |
1767 | pat = base_pat; | |
1768 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1769 | if (direction > 0) |
1770 | { | |
1771 | limit = BUFFER_CEILING_OF (limit); | |
1772 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1773 | can take on without hitting edge of buffer or the gap. */ | |
1774 | limit = min (limit, pos_byte + 20000); | |
1775 | limit = min (limit, lim_byte - 1); | |
1776 | } | |
1777 | else | |
1778 | { | |
1779 | limit = BUFFER_FLOOR_OF (limit); | |
1780 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1781 | can take on without hitting edge of buffer or the gap. */ | |
1782 | limit = max (limit, pos_byte - 20000); | |
1783 | limit = max (limit, lim_byte); | |
1784 | } | |
facdc750 RS |
1785 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1786 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1787 | ||
1788 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1789 | { |
facdc750 RS |
1790 | unsigned char *p2; |
1791 | ||
1792 | p_limit = BYTE_POS_ADDR (limit); | |
1793 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
1794 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ | |
1795 | while (1) /* use one cursor setting as long as i can */ | |
ca1d1d23 | 1796 | { |
facdc750 | 1797 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1798 | { |
facdc750 RS |
1799 | /* Use signed comparison if appropriate |
1800 | to make cursor+infinity sure to be > p_limit. | |
1801 | Assuming that the buffer lies in a range of addresses | |
1802 | that are all "positive" (as ints) or all "negative", | |
1803 | either kind of comparison will work as long | |
1804 | as we don't step by infinity. So pick the kind | |
1805 | that works when we do step by infinity. */ | |
1806 | if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit) | |
1807 | while ((EMACS_INT) cursor <= (EMACS_INT) p_limit) | |
1808 | cursor += BM_tab[*cursor]; | |
ca1d1d23 | 1809 | else |
facdc750 RS |
1810 | while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit) |
1811 | cursor += BM_tab[*cursor]; | |
1812 | } | |
1813 | else | |
1814 | { | |
1815 | if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit) | |
1816 | while ((EMACS_INT) cursor >= (EMACS_INT) p_limit) | |
1817 | cursor += BM_tab[*cursor]; | |
1818 | else | |
1819 | while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit) | |
1820 | cursor += BM_tab[*cursor]; | |
1821 | } | |
ca1d1d23 | 1822 | /* If you are here, cursor is beyond the end of the searched region. */ |
facdc750 RS |
1823 | /* This can happen if you match on the far character of the pattern, */ |
1824 | /* because the "stride" of that character is infinity, a number able */ | |
1825 | /* to throw you well beyond the end of the search. It can also */ | |
1826 | /* happen if you fail to match within the permitted region and would */ | |
1827 | /* otherwise try a character beyond that region */ | |
1828 | if ((cursor - p_limit) * direction <= len_byte) | |
1829 | break; /* a small overrun is genuine */ | |
1830 | cursor -= infinity; /* large overrun = hit */ | |
1831 | i = dirlen - direction; | |
1832 | if (! NILP (trt)) | |
1833 | { | |
1834 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1835 | { |
facdc750 RS |
1836 | int ch; |
1837 | cursor -= direction; | |
1838 | /* Translate only the last byte of a character. */ | |
1839 | if (! multibyte | |
1840 | || ((cursor == tail_end_ptr | |
1841 | || CHAR_HEAD_P (cursor[1])) | |
1842 | && (CHAR_HEAD_P (cursor[0]) | |
0190922f KH |
1843 | /* Check if this is the last byte of |
1844 | a translable character. */ | |
1845 | || (translate_prev_byte1 == cursor[-1] | |
1846 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1847 | || (translate_prev_byte2 == cursor[-2] | |
1848 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1849 | || (translate_prev_byte3 == cursor[-3])))))))) | |
facdc750 RS |
1850 | ch = simple_translate[*cursor]; |
1851 | else | |
1852 | ch = *cursor; | |
1853 | if (pat[i] != ch) | |
1854 | break; | |
ca1d1d23 | 1855 | } |
facdc750 RS |
1856 | } |
1857 | else | |
1858 | { | |
1859 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1860 | { |
facdc750 RS |
1861 | cursor -= direction; |
1862 | if (pat[i] != *cursor) | |
1863 | break; | |
ca1d1d23 | 1864 | } |
facdc750 RS |
1865 | } |
1866 | cursor += dirlen - i - direction; /* fix cursor */ | |
1867 | if (i + direction == 0) | |
1868 | { | |
1869 | int position; | |
0c8533c6 | 1870 | |
facdc750 | 1871 | cursor -= direction; |
1113d9db | 1872 | |
facdc750 RS |
1873 | position = pos_byte + cursor - p2 + ((direction > 0) |
1874 | ? 1 - len_byte : 0); | |
1875 | set_search_regs (position, len_byte); | |
ca325161 | 1876 | |
facdc750 RS |
1877 | if ((n -= direction) != 0) |
1878 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1879 | else |
facdc750 RS |
1880 | return ((direction > 0) |
1881 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1882 | } |
facdc750 RS |
1883 | else |
1884 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1885 | } |
facdc750 RS |
1886 | pos_byte += cursor - p2; |
1887 | } | |
1888 | else | |
1889 | /* Now we'll pick up a clump that has to be done the hard */ | |
1890 | /* way because it covers a discontinuity */ | |
1891 | { | |
1892 | limit = ((direction > 0) | |
1893 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1894 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1895 | limit = ((direction > 0) | |
1896 | ? min (limit + len_byte, lim_byte - 1) | |
1897 | : max (limit - len_byte, lim_byte)); | |
1898 | /* LIMIT is now the last value POS_BYTE can have | |
1899 | and still be valid for a possible match. */ | |
1900 | while (1) | |
ca1d1d23 | 1901 | { |
facdc750 RS |
1902 | /* This loop can be coded for space rather than */ |
1903 | /* speed because it will usually run only once. */ | |
1904 | /* (the reach is at most len + 21, and typically */ | |
177c0ea7 | 1905 | /* does not exceed len) */ |
facdc750 RS |
1906 | while ((limit - pos_byte) * direction >= 0) |
1907 | pos_byte += BM_tab[FETCH_BYTE (pos_byte)]; | |
1908 | /* now run the same tests to distinguish going off the */ | |
1909 | /* end, a match or a phony match. */ | |
1910 | if ((pos_byte - limit) * direction <= len_byte) | |
1911 | break; /* ran off the end */ | |
1912 | /* Found what might be a match. | |
1913 | Set POS_BYTE back to last (first if reverse) pos. */ | |
1914 | pos_byte -= infinity; | |
1915 | i = dirlen - direction; | |
1916 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1917 | { |
facdc750 RS |
1918 | int ch; |
1919 | unsigned char *ptr; | |
1920 | pos_byte -= direction; | |
1921 | ptr = BYTE_POS_ADDR (pos_byte); | |
1922 | /* Translate only the last byte of a character. */ | |
1923 | if (! multibyte | |
1924 | || ((ptr == tail_end_ptr | |
1925 | || CHAR_HEAD_P (ptr[1])) | |
1926 | && (CHAR_HEAD_P (ptr[0]) | |
0190922f KH |
1927 | /* Check if this is the last byte of a |
1928 | translable character. */ | |
1929 | || (translate_prev_byte1 == ptr[-1] | |
1930 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1931 | || (translate_prev_byte2 == ptr[-2] | |
1932 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1933 | || translate_prev_byte3 == ptr[-3]))))))) | |
facdc750 RS |
1934 | ch = simple_translate[*ptr]; |
1935 | else | |
1936 | ch = *ptr; | |
1937 | if (pat[i] != ch) | |
1938 | break; | |
1939 | } | |
1940 | /* Above loop has moved POS_BYTE part or all the way | |
1941 | back to the first pos (last pos if reverse). | |
1942 | Set it once again at the last (first if reverse) char. */ | |
1943 | pos_byte += dirlen - i- direction; | |
1944 | if (i + direction == 0) | |
1945 | { | |
1946 | int position; | |
1947 | pos_byte -= direction; | |
1113d9db | 1948 | |
facdc750 | 1949 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
0c8533c6 | 1950 | |
facdc750 | 1951 | set_search_regs (position, len_byte); |
ca325161 | 1952 | |
facdc750 RS |
1953 | if ((n -= direction) != 0) |
1954 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 1955 | else |
facdc750 RS |
1956 | return ((direction > 0) |
1957 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1958 | } |
facdc750 RS |
1959 | else |
1960 | pos_byte += stride_for_teases; | |
1961 | } | |
1962 | } | |
1963 | /* We have done one clump. Can we continue? */ | |
1964 | if ((lim_byte - pos_byte) * direction < 0) | |
1965 | return ((0 - n) * direction); | |
ca1d1d23 | 1966 | } |
facdc750 | 1967 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 1968 | } |
ca325161 | 1969 | |
fa8ed3e0 | 1970 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
1971 | for the overall match just found in the current buffer. |
1972 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
1973 | |
1974 | static void | |
fa8ed3e0 RS |
1975 | set_search_regs (beg_byte, nbytes) |
1976 | int beg_byte, nbytes; | |
ca325161 | 1977 | { |
a7e4cdde RS |
1978 | int i; |
1979 | ||
ca325161 RS |
1980 | /* Make sure we have registers in which to store |
1981 | the match position. */ | |
1982 | if (search_regs.num_regs == 0) | |
1983 | { | |
2d4a771a RS |
1984 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
1985 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 1986 | search_regs.num_regs = 2; |
ca325161 RS |
1987 | } |
1988 | ||
a7e4cdde RS |
1989 | /* Clear out the other registers. */ |
1990 | for (i = 1; i < search_regs.num_regs; i++) | |
1991 | { | |
1992 | search_regs.start[i] = -1; | |
1993 | search_regs.end[i] = -1; | |
1994 | } | |
1995 | ||
fa8ed3e0 RS |
1996 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
1997 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 1998 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 1999 | } |
ca1d1d23 JB |
2000 | \f |
2001 | /* Given a string of words separated by word delimiters, | |
2002 | compute a regexp that matches those exact words | |
2003 | separated by arbitrary punctuation. */ | |
2004 | ||
2005 | static Lisp_Object | |
2006 | wordify (string) | |
2007 | Lisp_Object string; | |
2008 | { | |
2009 | register unsigned char *p, *o; | |
0c8533c6 | 2010 | register int i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 2011 | Lisp_Object val; |
0c8533c6 RS |
2012 | int prev_c = 0; |
2013 | int adjust; | |
ca1d1d23 | 2014 | |
b7826503 | 2015 | CHECK_STRING (string); |
d5db4077 KR |
2016 | p = SDATA (string); |
2017 | len = SCHARS (string); | |
ca1d1d23 | 2018 | |
0c8533c6 RS |
2019 | for (i = 0, i_byte = 0; i < len; ) |
2020 | { | |
2021 | int c; | |
177c0ea7 | 2022 | |
93daa011 | 2023 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
0c8533c6 RS |
2024 | |
2025 | if (SYNTAX (c) != Sword) | |
2026 | { | |
2027 | punct_count++; | |
2028 | if (i > 0 && SYNTAX (prev_c) == Sword) | |
2029 | word_count++; | |
2030 | } | |
ca1d1d23 | 2031 | |
0c8533c6 RS |
2032 | prev_c = c; |
2033 | } | |
2034 | ||
2035 | if (SYNTAX (prev_c) == Sword) | |
2036 | word_count++; | |
2037 | if (!word_count) | |
b07b65aa | 2038 | return empty_string; |
0c8533c6 RS |
2039 | |
2040 | adjust = - punct_count + 5 * (word_count - 1) + 4; | |
8a2df937 RS |
2041 | if (STRING_MULTIBYTE (string)) |
2042 | val = make_uninit_multibyte_string (len + adjust, | |
d5db4077 | 2043 | SBYTES (string) |
8a2df937 RS |
2044 | + adjust); |
2045 | else | |
2046 | val = make_uninit_string (len + adjust); | |
ca1d1d23 | 2047 | |
d5db4077 | 2048 | o = SDATA (val); |
ca1d1d23 JB |
2049 | *o++ = '\\'; |
2050 | *o++ = 'b'; | |
1e9582d4 | 2051 | prev_c = 0; |
ca1d1d23 | 2052 | |
1e9582d4 RS |
2053 | for (i = 0, i_byte = 0; i < len; ) |
2054 | { | |
2055 | int c; | |
2056 | int i_byte_orig = i_byte; | |
177c0ea7 | 2057 | |
93daa011 | 2058 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
1e9582d4 RS |
2059 | |
2060 | if (SYNTAX (c) == Sword) | |
2061 | { | |
5d69fe10 | 2062 | bcopy (SDATA (string) + i_byte_orig, o, |
1e9582d4 RS |
2063 | i_byte - i_byte_orig); |
2064 | o += i_byte - i_byte_orig; | |
2065 | } | |
2066 | else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count) | |
2067 | { | |
2068 | *o++ = '\\'; | |
2069 | *o++ = 'W'; | |
2070 | *o++ = '\\'; | |
2071 | *o++ = 'W'; | |
2072 | *o++ = '*'; | |
2073 | } | |
2074 | ||
2075 | prev_c = c; | |
2076 | } | |
ca1d1d23 JB |
2077 | |
2078 | *o++ = '\\'; | |
2079 | *o++ = 'b'; | |
2080 | ||
2081 | return val; | |
2082 | } | |
2083 | \f | |
2084 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
8c1a1077 PJ |
2085 | "MSearch backward: ", |
2086 | doc: /* Search backward from point for STRING. | |
2087 | Set point to the beginning of the occurrence found, and return point. | |
2088 | An optional second argument bounds the search; it is a buffer position. | |
2089 | The match found must not extend before that position. | |
2090 | Optional third argument, if t, means if fail just return nil (no error). | |
2091 | If not nil and not t, position at limit of search and return nil. | |
2092 | Optional fourth argument is repeat count--search for successive occurrences. | |
2093 | ||
2094 | Search case-sensitivity is determined by the value of the variable | |
2095 | `case-fold-search', which see. | |
2096 | ||
2097 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2098 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2099 | Lisp_Object string, bound, noerror, count; |
2100 | { | |
b819a390 | 2101 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2102 | } |
2103 | ||
6af43974 | 2104 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2105 | doc: /* Search forward from point for STRING. |
2106 | Set point to the end of the occurrence found, and return point. | |
2107 | An optional second argument bounds the search; it is a buffer position. | |
2108 | The match found must not extend after that position. nil is equivalent | |
2109 | to (point-max). | |
2110 | Optional third argument, if t, means if fail just return nil (no error). | |
2111 | If not nil and not t, move to limit of search and return nil. | |
2112 | Optional fourth argument is repeat count--search for successive occurrences. | |
2113 | ||
2114 | Search case-sensitivity is determined by the value of the variable | |
2115 | `case-fold-search', which see. | |
2116 | ||
2117 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2118 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2119 | Lisp_Object string, bound, noerror, count; |
2120 | { | |
b819a390 | 2121 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2122 | } |
2123 | ||
2124 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
8c1a1077 PJ |
2125 | "sWord search backward: ", |
2126 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2127 | Set point to the beginning of the occurrence found, and return point. | |
2128 | An optional second argument bounds the search; it is a buffer position. | |
2129 | The match found must not extend before that position. | |
2130 | Optional third argument, if t, means if fail just return nil (no error). | |
2131 | If not nil and not t, move to limit of search and return nil. | |
2132 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2133 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2134 | Lisp_Object string, bound, noerror, count; |
2135 | { | |
b819a390 | 2136 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2137 | } |
2138 | ||
2139 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
8c1a1077 PJ |
2140 | "sWord search: ", |
2141 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2142 | Set point to the end of the occurrence found, and return point. | |
2143 | An optional second argument bounds the search; it is a buffer position. | |
2144 | The match found must not extend after that position. | |
2145 | Optional third argument, if t, means if fail just return nil (no error). | |
2146 | If not nil and not t, move to limit of search and return nil. | |
2147 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2148 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2149 | Lisp_Object string, bound, noerror, count; |
2150 | { | |
b819a390 | 2151 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
2152 | } |
2153 | ||
2154 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
8c1a1077 PJ |
2155 | "sRE search backward: ", |
2156 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2157 | Set point to the beginning of the match, and return point. | |
2158 | The match found is the one starting last in the buffer | |
2159 | and yet ending before the origin of the search. | |
2160 | An optional second argument bounds the search; it is a buffer position. | |
2161 | The match found must start at or after that position. | |
2162 | Optional third argument, if t, means if fail just return nil (no error). | |
2163 | If not nil and not t, move to limit of search and return nil. | |
2164 | Optional fourth argument is repeat count--search for successive occurrences. | |
2165 | See also the functions `match-beginning', `match-end', `match-string', | |
2166 | and `replace-match'. */) | |
2167 | (regexp, bound, noerror, count) | |
19c0a730 | 2168 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2169 | { |
b819a390 | 2170 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2171 | } |
2172 | ||
2173 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2174 | "sRE search: ", |
2175 | doc: /* Search forward from point for regular expression REGEXP. | |
2176 | Set point to the end of the occurrence found, and return point. | |
2177 | An optional second argument bounds the search; it is a buffer position. | |
2178 | The match found must not extend after that position. | |
2179 | Optional third argument, if t, means if fail just return nil (no error). | |
2180 | If not nil and not t, move to limit of search and return nil. | |
2181 | Optional fourth argument is repeat count--search for successive occurrences. | |
2182 | See also the functions `match-beginning', `match-end', `match-string', | |
2183 | and `replace-match'. */) | |
2184 | (regexp, bound, noerror, count) | |
19c0a730 | 2185 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2186 | { |
b819a390 RS |
2187 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2188 | } | |
2189 | ||
2190 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2191 | "sPosix search backward: ", |
2192 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2193 | Find the longest match in accord with Posix regular expression rules. | |
2194 | Set point to the beginning of the match, and return point. | |
2195 | The match found is the one starting last in the buffer | |
2196 | and yet ending before the origin of the search. | |
2197 | An optional second argument bounds the search; it is a buffer position. | |
2198 | The match found must start at or after that position. | |
2199 | Optional third argument, if t, means if fail just return nil (no error). | |
2200 | If not nil and not t, move to limit of search and return nil. | |
2201 | Optional fourth argument is repeat count--search for successive occurrences. | |
2202 | See also the functions `match-beginning', `match-end', `match-string', | |
2203 | and `replace-match'. */) | |
2204 | (regexp, bound, noerror, count) | |
b819a390 RS |
2205 | Lisp_Object regexp, bound, noerror, count; |
2206 | { | |
2207 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2208 | } | |
2209 | ||
2210 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2211 | "sPosix search: ", |
2212 | doc: /* Search forward from point for regular expression REGEXP. | |
2213 | Find the longest match in accord with Posix regular expression rules. | |
2214 | Set point to the end of the occurrence found, and return point. | |
2215 | An optional second argument bounds the search; it is a buffer position. | |
2216 | The match found must not extend after that position. | |
2217 | Optional third argument, if t, means if fail just return nil (no error). | |
2218 | If not nil and not t, move to limit of search and return nil. | |
2219 | Optional fourth argument is repeat count--search for successive occurrences. | |
2220 | See also the functions `match-beginning', `match-end', `match-string', | |
2221 | and `replace-match'. */) | |
2222 | (regexp, bound, noerror, count) | |
b819a390 RS |
2223 | Lisp_Object regexp, bound, noerror, count; |
2224 | { | |
2225 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2226 | } |
2227 | \f | |
d7a5ad5f | 2228 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2229 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2230 | Leave point at the end of the replacement text. |
2231 | ||
8c1a1077 PJ |
2232 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. |
2233 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2234 | based on the replaced text. | |
2235 | If the replaced text has only capital letters | |
2236 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
4dd0c271 RS |
2237 | Otherwise if all words are capitalized in the replaced text, |
2238 | capitalize each word in NEWTEXT. | |
8c1a1077 PJ |
2239 | |
2240 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2241 | Otherwise treat `\\' as special: | |
2242 | `\\&' in NEWTEXT means substitute original matched text. | |
2243 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2244 | If Nth parens didn't match, substitute nothing. | |
2245 | `\\\\' means insert one `\\'. | |
4dd0c271 RS |
2246 | Case conversion does not apply to these substitutions. |
2247 | ||
8c1a1077 | 2248 | FIXEDCASE and LITERAL are optional arguments. |
8c1a1077 PJ |
2249 | |
2250 | The optional fourth argument STRING can be a string to modify. | |
2251 | This is meaningful when the previous match was done against STRING, | |
2252 | using `string-match'. When used this way, `replace-match' | |
2253 | creates and returns a new string made by copying STRING and replacing | |
2254 | the part of STRING that was matched. | |
2255 | ||
2256 | The optional fifth argument SUBEXP specifies a subexpression; | |
2257 | it says to replace just that subexpression with NEWTEXT, | |
2258 | rather than replacing the entire matched text. | |
2259 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2260 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2261 | NEWTEXT in place of subexp N. | |
2262 | This is useful only after a regular expression search or match, | |
2263 | since only regular expressions have distinguished subexpressions. */) | |
2264 | (newtext, fixedcase, literal, string, subexp) | |
d7a5ad5f | 2265 | Lisp_Object newtext, fixedcase, literal, string, subexp; |
ca1d1d23 JB |
2266 | { |
2267 | enum { nochange, all_caps, cap_initial } case_action; | |
ac3b28b1 | 2268 | register int pos, pos_byte; |
ca1d1d23 | 2269 | int some_multiletter_word; |
97832bd0 | 2270 | int some_lowercase; |
73dc8771 | 2271 | int some_uppercase; |
208767c3 | 2272 | int some_nonuppercase_initial; |
ca1d1d23 | 2273 | register int c, prevc; |
d7a5ad5f | 2274 | int sub; |
3e18eecf | 2275 | int opoint, newpoint; |
ca1d1d23 | 2276 | |
b7826503 | 2277 | CHECK_STRING (newtext); |
ca1d1d23 | 2278 | |
080c45fd | 2279 | if (! NILP (string)) |
b7826503 | 2280 | CHECK_STRING (string); |
080c45fd | 2281 | |
ca1d1d23 JB |
2282 | case_action = nochange; /* We tried an initialization */ |
2283 | /* but some C compilers blew it */ | |
4746118a JB |
2284 | |
2285 | if (search_regs.num_regs <= 0) | |
d72cdbfc | 2286 | error ("`replace-match' called before any match found"); |
4746118a | 2287 | |
d7a5ad5f RS |
2288 | if (NILP (subexp)) |
2289 | sub = 0; | |
2290 | else | |
2291 | { | |
b7826503 | 2292 | CHECK_NUMBER (subexp); |
d7a5ad5f RS |
2293 | sub = XINT (subexp); |
2294 | if (sub < 0 || sub >= search_regs.num_regs) | |
2295 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2296 | } | |
2297 | ||
080c45fd RS |
2298 | if (NILP (string)) |
2299 | { | |
d7a5ad5f RS |
2300 | if (search_regs.start[sub] < BEGV |
2301 | || search_regs.start[sub] > search_regs.end[sub] | |
2302 | || search_regs.end[sub] > ZV) | |
2303 | args_out_of_range (make_number (search_regs.start[sub]), | |
2304 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2305 | } |
2306 | else | |
2307 | { | |
d7a5ad5f RS |
2308 | if (search_regs.start[sub] < 0 |
2309 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2310 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2311 | args_out_of_range (make_number (search_regs.start[sub]), |
2312 | make_number (search_regs.end[sub])); | |
080c45fd | 2313 | } |
ca1d1d23 JB |
2314 | |
2315 | if (NILP (fixedcase)) | |
2316 | { | |
2317 | /* Decide how to casify by examining the matched text. */ | |
ac3b28b1 | 2318 | int last; |
ca1d1d23 | 2319 | |
ac3b28b1 KH |
2320 | pos = search_regs.start[sub]; |
2321 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2322 | |
2323 | if (NILP (string)) | |
ac3b28b1 | 2324 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2325 | else |
ac3b28b1 | 2326 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2327 | |
ca1d1d23 JB |
2328 | prevc = '\n'; |
2329 | case_action = all_caps; | |
2330 | ||
2331 | /* some_multiletter_word is set nonzero if any original word | |
2332 | is more than one letter long. */ | |
2333 | some_multiletter_word = 0; | |
97832bd0 | 2334 | some_lowercase = 0; |
208767c3 | 2335 | some_nonuppercase_initial = 0; |
73dc8771 | 2336 | some_uppercase = 0; |
ca1d1d23 | 2337 | |
ac3b28b1 | 2338 | while (pos < last) |
ca1d1d23 | 2339 | { |
080c45fd | 2340 | if (NILP (string)) |
ac3b28b1 | 2341 | { |
93daa011 | 2342 | c = FETCH_CHAR_AS_MULTIBYTE (pos_byte); |
ac3b28b1 KH |
2343 | INC_BOTH (pos, pos_byte); |
2344 | } | |
080c45fd | 2345 | else |
93daa011 | 2346 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2347 | |
ca1d1d23 JB |
2348 | if (LOWERCASEP (c)) |
2349 | { | |
2350 | /* Cannot be all caps if any original char is lower case */ | |
2351 | ||
97832bd0 | 2352 | some_lowercase = 1; |
ca1d1d23 | 2353 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2354 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2355 | else |
2356 | some_multiletter_word = 1; | |
2357 | } | |
d16c2b66 | 2358 | else if (UPPERCASEP (c)) |
ca1d1d23 | 2359 | { |
73dc8771 | 2360 | some_uppercase = 1; |
97832bd0 | 2361 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2362 | ; |
97832bd0 | 2363 | else |
ca1d1d23 JB |
2364 | some_multiletter_word = 1; |
2365 | } | |
208767c3 RS |
2366 | else |
2367 | { | |
2368 | /* If the initial is a caseless word constituent, | |
2369 | treat that like a lowercase initial. */ | |
2370 | if (SYNTAX (prevc) != Sword) | |
2371 | some_nonuppercase_initial = 1; | |
2372 | } | |
ca1d1d23 JB |
2373 | |
2374 | prevc = c; | |
2375 | } | |
2376 | ||
97832bd0 RS |
2377 | /* Convert to all caps if the old text is all caps |
2378 | and has at least one multiletter word. */ | |
2379 | if (! some_lowercase && some_multiletter_word) | |
2380 | case_action = all_caps; | |
c4d460ce | 2381 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2382 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2383 | case_action = cap_initial; |
208767c3 | 2384 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2385 | /* Should x -> yz, operating on X, give Yz or YZ? |
2386 | We'll assume the latter. */ | |
2387 | case_action = all_caps; | |
97832bd0 RS |
2388 | else |
2389 | case_action = nochange; | |
ca1d1d23 JB |
2390 | } |
2391 | ||
080c45fd RS |
2392 | /* Do replacement in a string. */ |
2393 | if (!NILP (string)) | |
2394 | { | |
2395 | Lisp_Object before, after; | |
2396 | ||
2397 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2398 | make_number (search_regs.start[sub])); |
2399 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2400 | |
636a5e28 RS |
2401 | /* Substitute parts of the match into NEWTEXT |
2402 | if desired. */ | |
080c45fd RS |
2403 | if (NILP (literal)) |
2404 | { | |
d131e79c RS |
2405 | int lastpos = 0; |
2406 | int lastpos_byte = 0; | |
080c45fd RS |
2407 | /* We build up the substituted string in ACCUM. */ |
2408 | Lisp_Object accum; | |
2409 | Lisp_Object middle; | |
d5db4077 | 2410 | int length = SBYTES (newtext); |
080c45fd RS |
2411 | |
2412 | accum = Qnil; | |
2413 | ||
ac3b28b1 | 2414 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd RS |
2415 | { |
2416 | int substart = -1; | |
6bbd7a29 | 2417 | int subend = 0; |
1e79ec24 | 2418 | int delbackslash = 0; |
080c45fd | 2419 | |
0c8533c6 RS |
2420 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2421 | ||
080c45fd RS |
2422 | if (c == '\\') |
2423 | { | |
0c8533c6 | 2424 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2425 | |
080c45fd RS |
2426 | if (c == '&') |
2427 | { | |
d7a5ad5f RS |
2428 | substart = search_regs.start[sub]; |
2429 | subend = search_regs.end[sub]; | |
080c45fd | 2430 | } |
76fc1ea2 | 2431 | else if (c >= '1' && c <= '9') |
080c45fd | 2432 | { |
76fc1ea2 KH |
2433 | if (search_regs.start[c - '0'] >= 0 |
2434 | && c <= search_regs.num_regs + '0') | |
080c45fd RS |
2435 | { |
2436 | substart = search_regs.start[c - '0']; | |
2437 | subend = search_regs.end[c - '0']; | |
2438 | } | |
76fc1ea2 KH |
2439 | else |
2440 | { | |
2441 | /* If that subexp did not match, | |
2442 | replace \\N with nothing. */ | |
2443 | substart = 0; | |
2444 | subend = 0; | |
2445 | } | |
080c45fd | 2446 | } |
1e79ec24 KH |
2447 | else if (c == '\\') |
2448 | delbackslash = 1; | |
636a5e28 RS |
2449 | else |
2450 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2451 | } |
2452 | if (substart >= 0) | |
2453 | { | |
d131e79c RS |
2454 | if (pos - 2 != lastpos) |
2455 | middle = substring_both (newtext, lastpos, | |
2456 | lastpos_byte, | |
2457 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2458 | else |
2459 | middle = Qnil; | |
2460 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2461 | Fsubstring (string, |
2462 | make_number (substart), | |
080c45fd RS |
2463 | make_number (subend))); |
2464 | lastpos = pos; | |
0c8533c6 | 2465 | lastpos_byte = pos_byte; |
080c45fd | 2466 | } |
1e79ec24 KH |
2467 | else if (delbackslash) |
2468 | { | |
d131e79c RS |
2469 | middle = substring_both (newtext, lastpos, |
2470 | lastpos_byte, | |
2471 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2472 | |
1e79ec24 KH |
2473 | accum = concat2 (accum, middle); |
2474 | lastpos = pos; | |
0c8533c6 | 2475 | lastpos_byte = pos_byte; |
1e79ec24 | 2476 | } |
080c45fd RS |
2477 | } |
2478 | ||
d131e79c RS |
2479 | if (pos != lastpos) |
2480 | middle = substring_both (newtext, lastpos, | |
2481 | lastpos_byte, | |
0c8533c6 | 2482 | pos, pos_byte); |
080c45fd RS |
2483 | else |
2484 | middle = Qnil; | |
2485 | ||
2486 | newtext = concat2 (accum, middle); | |
2487 | } | |
2488 | ||
636a5e28 | 2489 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2490 | if (case_action == all_caps) |
2491 | newtext = Fupcase (newtext); | |
2492 | else if (case_action == cap_initial) | |
2b2eead9 | 2493 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2494 | |
2495 | return concat3 (before, newtext, after); | |
2496 | } | |
2497 | ||
09c4719e | 2498 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2499 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2500 | opoint = PT - ZV; |
9160906f RS |
2501 | else if (PT > search_regs.start[sub]) |
2502 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2503 | else |
2504 | opoint = PT; | |
2505 | ||
886ed6ec RS |
2506 | /* If we want non-literal replacement, |
2507 | perform substitution on the replacement string. */ | |
2508 | if (NILP (literal)) | |
ca1d1d23 | 2509 | { |
d5db4077 | 2510 | int length = SBYTES (newtext); |
68e69fbd RS |
2511 | unsigned char *substed; |
2512 | int substed_alloc_size, substed_len; | |
3bc25e52 KH |
2513 | int buf_multibyte = !NILP (current_buffer->enable_multibyte_characters); |
2514 | int str_multibyte = STRING_MULTIBYTE (newtext); | |
2515 | Lisp_Object rev_tbl; | |
886ed6ec | 2516 | int really_changed = 0; |
3bc25e52 | 2517 | |
8f924df7 | 2518 | rev_tbl = Qnil; |
ac3b28b1 | 2519 | |
68e69fbd RS |
2520 | substed_alloc_size = length * 2 + 100; |
2521 | substed = (unsigned char *) xmalloc (substed_alloc_size + 1); | |
2522 | substed_len = 0; | |
2523 | ||
3bc25e52 KH |
2524 | /* Go thru NEWTEXT, producing the actual text to insert in |
2525 | SUBSTED while adjusting multibyteness to that of the current | |
2526 | buffer. */ | |
ca1d1d23 | 2527 | |
ac3b28b1 | 2528 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2529 | { |
68e69fbd | 2530 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
f8ce8a0d GM |
2531 | unsigned char *add_stuff = NULL; |
2532 | int add_len = 0; | |
68e69fbd | 2533 | int idx = -1; |
9a76659d | 2534 | |
3bc25e52 KH |
2535 | if (str_multibyte) |
2536 | { | |
eb99a8dd | 2537 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 KH |
2538 | if (!buf_multibyte) |
2539 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2540 | } | |
2541 | else | |
2542 | { | |
2543 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2544 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2545 | if (buf_multibyte) |
2546 | c = unibyte_char_to_multibyte (c); | |
2547 | } | |
ac3b28b1 | 2548 | |
68e69fbd RS |
2549 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2550 | or set IDX to a match index, which means put that part | |
2551 | of the buffer text into SUBSTED. */ | |
2552 | ||
ca1d1d23 JB |
2553 | if (c == '\\') |
2554 | { | |
886ed6ec RS |
2555 | really_changed = 1; |
2556 | ||
3bc25e52 KH |
2557 | if (str_multibyte) |
2558 | { | |
eb99a8dd KH |
2559 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2560 | pos, pos_byte); | |
071ce769 | 2561 | if (!buf_multibyte && !ASCII_CHAR_P (c)) |
3bc25e52 KH |
2562 | c = multibyte_char_to_unibyte (c, rev_tbl); |
2563 | } | |
2564 | else | |
2565 | { | |
d5db4077 | 2566 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2567 | if (buf_multibyte) |
2568 | c = unibyte_char_to_multibyte (c); | |
2569 | } | |
2570 | ||
ca1d1d23 | 2571 | if (c == '&') |
68e69fbd | 2572 | idx = sub; |
78445046 | 2573 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2574 | { |
2575 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2576 | idx = c - '0'; |
ca1d1d23 | 2577 | } |
636a5e28 | 2578 | else if (c == '\\') |
68e69fbd | 2579 | add_len = 1, add_stuff = "\\"; |
636a5e28 | 2580 | else |
3bc25e52 KH |
2581 | { |
2582 | xfree (substed); | |
2583 | error ("Invalid use of `\\' in replacement text"); | |
2584 | } | |
ca1d1d23 JB |
2585 | } |
2586 | else | |
68e69fbd RS |
2587 | { |
2588 | add_len = CHAR_STRING (c, str); | |
2589 | add_stuff = str; | |
2590 | } | |
2591 | ||
2592 | /* If we want to copy part of a previous match, | |
2593 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2594 | if (idx >= 0) | |
2595 | { | |
2596 | int begbyte = CHAR_TO_BYTE (search_regs.start[idx]); | |
2597 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; | |
2598 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2599 | move_gap (search_regs.start[idx]); | |
2600 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2601 | } | |
2602 | ||
2603 | /* Now the stuff we want to add to SUBSTED | |
2604 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2605 | ||
2606 | /* Make sure SUBSTED is big enough. */ | |
2607 | if (substed_len + add_len >= substed_alloc_size) | |
2608 | { | |
2609 | substed_alloc_size = substed_len + add_len + 500; | |
2610 | substed = (unsigned char *) xrealloc (substed, | |
2611 | substed_alloc_size + 1); | |
2612 | } | |
2613 | ||
2614 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2615 | if (add_stuff) |
2616 | { | |
2617 | bcopy (add_stuff, substed + substed_len, add_len); | |
2618 | substed_len += add_len; | |
2619 | } | |
ca1d1d23 | 2620 | } |
68e69fbd | 2621 | |
886ed6ec | 2622 | if (really_changed) |
76fc1ea2 KH |
2623 | { |
2624 | if (buf_multibyte) | |
2625 | { | |
2626 | int nchars = multibyte_chars_in_text (substed, substed_len); | |
68e69fbd | 2627 | |
76fc1ea2 KH |
2628 | newtext = make_multibyte_string (substed, nchars, substed_len); |
2629 | } | |
2630 | else | |
2631 | newtext = make_unibyte_string (substed, substed_len); | |
2632 | } | |
68e69fbd | 2633 | xfree (substed); |
ca1d1d23 JB |
2634 | } |
2635 | ||
886ed6ec RS |
2636 | /* Replace the old text with the new in the cleanest possible way. */ |
2637 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2638 | newtext, 1, 0, 1); | |
d5db4077 | 2639 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2640 | |
2641 | if (case_action == all_caps) | |
886ed6ec RS |
2642 | Fupcase_region (make_number (search_regs.start[sub]), |
2643 | make_number (newpoint)); | |
ca1d1d23 | 2644 | else if (case_action == cap_initial) |
886ed6ec RS |
2645 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2646 | make_number (newpoint)); | |
3e18eecf | 2647 | |
98e942e0 RS |
2648 | /* Adjust search data for this change. */ |
2649 | { | |
5b88a2c5 | 2650 | int oldend = search_regs.end[sub]; |
41c01205 | 2651 | int oldstart = search_regs.start[sub]; |
98e942e0 RS |
2652 | int change = newpoint - search_regs.end[sub]; |
2653 | int i; | |
2654 | ||
2655 | for (i = 0; i < search_regs.num_regs; i++) | |
2656 | { | |
41c01205 | 2657 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2658 | search_regs.start[i] += change; |
41c01205 DK |
2659 | else if (search_regs.start[i] > oldstart) |
2660 | search_regs.start[i] = oldstart; | |
2661 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2662 | search_regs.end[i] += change; |
41c01205 DK |
2663 | else if (search_regs.end[i] > oldstart) |
2664 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2665 | } |
2666 | } | |
2667 | ||
b0eba991 | 2668 | /* Put point back where it was in the text. */ |
8d808a65 | 2669 | if (opoint <= 0) |
fa8ed3e0 | 2670 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2671 | else |
fa8ed3e0 | 2672 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2673 | |
2674 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2675 | move_if_not_intangible (newpoint); |
177c0ea7 | 2676 | |
ca1d1d23 JB |
2677 | return Qnil; |
2678 | } | |
2679 | \f | |
2680 | static Lisp_Object | |
2681 | match_limit (num, beginningp) | |
2682 | Lisp_Object num; | |
2683 | int beginningp; | |
2684 | { | |
2685 | register int n; | |
2686 | ||
b7826503 | 2687 | CHECK_NUMBER (num); |
ca1d1d23 | 2688 | n = XINT (num); |
f90a5bf5 | 2689 | if (n < 0) |
bd2cbd56 | 2690 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2691 | if (search_regs.num_regs <= 0) |
2692 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2693 | if (n >= search_regs.num_regs |
4746118a | 2694 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2695 | return Qnil; |
2696 | return (make_number ((beginningp) ? search_regs.start[n] | |
2697 | : search_regs.end[n])); | |
2698 | } | |
2699 | ||
2700 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
8c1a1077 PJ |
2701 | doc: /* Return position of start of text matched by last search. |
2702 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2703 | regexp. | |
2704 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2705 | SUBEXP pairs. | |
2706 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2707 | (subexp) | |
5806161b | 2708 | Lisp_Object subexp; |
ca1d1d23 | 2709 | { |
5806161b | 2710 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2711 | } |
2712 | ||
2713 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
8c1a1077 PJ |
2714 | doc: /* Return position of end of text matched by last search. |
2715 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2716 | regexp. | |
2717 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2718 | SUBEXP pairs. | |
2719 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2720 | (subexp) | |
5806161b | 2721 | Lisp_Object subexp; |
ca1d1d23 | 2722 | { |
5806161b | 2723 | return match_limit (subexp, 0); |
177c0ea7 | 2724 | } |
ca1d1d23 | 2725 | |
abd0071c | 2726 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 3, 0, |
8c1a1077 PJ |
2727 | doc: /* Return a list containing all info on what the last search matched. |
2728 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2729 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2730 | if the last match was on a buffer; integers or nil if a string was matched. | |
2731 | Use `store-match-data' to reinstate the data in this list. | |
2732 | ||
41c01205 DK |
2733 | If INTEGERS (the optional first argument) is non-nil, always use |
2734 | integers \(rather than markers) to represent buffer positions. In | |
2735 | this case, and if the last match was in a buffer, the buffer will get | |
2736 | stored as one additional element at the end of the list. | |
2737 | ||
abd0071c KS |
2738 | If REUSE is a list, reuse it as part of the value. If REUSE is long |
2739 | enough to hold all the values, and if INTEGERS is non-nil, no consing | |
2740 | is done. | |
2741 | ||
2742 | If optional third arg RESEAT is non-nil, any previous markers on the | |
2743 | REUSE list will be modified to point to nowhere. | |
2744 | ||
140a6b7e | 2745 | Return value is undefined if the last search failed. */) |
abd0071c KS |
2746 | (integers, reuse, reseat) |
2747 | Lisp_Object integers, reuse, reseat; | |
ca1d1d23 | 2748 | { |
56256c2a | 2749 | Lisp_Object tail, prev; |
4746118a | 2750 | Lisp_Object *data; |
ca1d1d23 JB |
2751 | int i, len; |
2752 | ||
abd0071c KS |
2753 | if (!NILP (reseat)) |
2754 | for (tail = reuse; CONSP (tail); tail = XCDR (tail)) | |
2755 | if (MARKERP (XCAR (tail))) | |
2756 | { | |
51f10faa | 2757 | unchain_marker (XMARKER (XCAR (tail))); |
abd0071c KS |
2758 | XSETCAR (tail, Qnil); |
2759 | } | |
2760 | ||
daa37602 | 2761 | if (NILP (last_thing_searched)) |
c36bcf1b | 2762 | return Qnil; |
daa37602 | 2763 | |
6bbd7a29 GM |
2764 | prev = Qnil; |
2765 | ||
41c01205 | 2766 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs + 1) |
4746118a JB |
2767 | * sizeof (Lisp_Object)); |
2768 | ||
41c01205 | 2769 | len = 0; |
4746118a | 2770 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2771 | { |
2772 | int start = search_regs.start[i]; | |
2773 | if (start >= 0) | |
2774 | { | |
56256c2a RS |
2775 | if (EQ (last_thing_searched, Qt) |
2776 | || ! NILP (integers)) | |
ca1d1d23 | 2777 | { |
c235cce7 KH |
2778 | XSETFASTINT (data[2 * i], start); |
2779 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2780 | } |
0ed62dc7 | 2781 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2782 | { |
2783 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2784 | Fset_marker (data[2 * i], |
2785 | make_number (start), | |
2786 | last_thing_searched); | |
ca1d1d23 JB |
2787 | data[2 * i + 1] = Fmake_marker (); |
2788 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2789 | make_number (search_regs.end[i]), |
daa37602 | 2790 | last_thing_searched); |
ca1d1d23 | 2791 | } |
daa37602 JB |
2792 | else |
2793 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2794 | abort (); | |
2795 | ||
abd0071c | 2796 | len = 2 * i + 2; |
ca1d1d23 JB |
2797 | } |
2798 | else | |
abd0071c | 2799 | data[2 * i] = data[2 * i + 1] = Qnil; |
ca1d1d23 | 2800 | } |
56256c2a | 2801 | |
bd2cbd56 | 2802 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2803 | { |
bd2cbd56 | 2804 | data[len] = last_thing_searched; |
41c01205 DK |
2805 | len++; |
2806 | } | |
2807 | ||
56256c2a RS |
2808 | /* If REUSE is not usable, cons up the values and return them. */ |
2809 | if (! CONSP (reuse)) | |
41c01205 | 2810 | return Flist (len, data); |
56256c2a RS |
2811 | |
2812 | /* If REUSE is a list, store as many value elements as will fit | |
2813 | into the elements of REUSE. */ | |
2814 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2815 | i++, tail = XCDR (tail)) |
56256c2a | 2816 | { |
41c01205 | 2817 | if (i < len) |
f3fbd155 | 2818 | XSETCAR (tail, data[i]); |
56256c2a | 2819 | else |
f3fbd155 | 2820 | XSETCAR (tail, Qnil); |
56256c2a RS |
2821 | prev = tail; |
2822 | } | |
2823 | ||
2824 | /* If we couldn't fit all value elements into REUSE, | |
2825 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2826 | if (i < len) |
2827 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2828 | |
2829 | return reuse; | |
ca1d1d23 JB |
2830 | } |
2831 | ||
51f10faa KS |
2832 | /* Internal usage only: |
2833 | If RESEAT is `evaporate', put the markers back on the free list | |
2834 | immediately. No other references to the markers must exist in this case, | |
2835 | so it is used only internally on the unwind stack and save-match-data from | |
2836 | Lisp. */ | |
ca1d1d23 | 2837 | |
abd0071c | 2838 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 2, 0, |
8c1a1077 | 2839 | doc: /* Set internal data on last search match from elements of LIST. |
abd0071c KS |
2840 | LIST should have been created by calling `match-data' previously. |
2841 | ||
51f10faa | 2842 | If optional arg RESEAT is non-nil, make markers on LIST point nowhere. */) |
abd0071c KS |
2843 | (list, reseat) |
2844 | register Lisp_Object list, reseat; | |
ca1d1d23 JB |
2845 | { |
2846 | register int i; | |
2847 | register Lisp_Object marker; | |
2848 | ||
7074fde6 FP |
2849 | if (running_asynch_code) |
2850 | save_search_regs (); | |
2851 | ||
ca1d1d23 | 2852 | if (!CONSP (list) && !NILP (list)) |
b37902c8 | 2853 | list = wrong_type_argument (Qconsp, list); |
ca1d1d23 | 2854 | |
41c01205 DK |
2855 | /* Unless we find a marker with a buffer or an explicit buffer |
2856 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
2857 | last_thing_searched = Qt; |
2858 | ||
4746118a JB |
2859 | /* Allocate registers if they don't already exist. */ |
2860 | { | |
d084e942 | 2861 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2862 | |
2863 | if (length > search_regs.num_regs) | |
2864 | { | |
1113d9db JB |
2865 | if (search_regs.num_regs == 0) |
2866 | { | |
2867 | search_regs.start | |
2868 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2869 | search_regs.end | |
2870 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2871 | } | |
4746118a | 2872 | else |
1113d9db JB |
2873 | { |
2874 | search_regs.start | |
2875 | = (regoff_t *) xrealloc (search_regs.start, | |
2876 | length * sizeof (regoff_t)); | |
2877 | search_regs.end | |
2878 | = (regoff_t *) xrealloc (search_regs.end, | |
2879 | length * sizeof (regoff_t)); | |
2880 | } | |
4746118a | 2881 | |
e62371e9 KH |
2882 | for (i = search_regs.num_regs; i < length; i++) |
2883 | search_regs.start[i] = -1; | |
2884 | ||
487282dc | 2885 | search_regs.num_regs = length; |
4746118a | 2886 | } |
ca1d1d23 | 2887 | |
abd0071c | 2888 | for (i = 0; CONSP (list); i++) |
41c01205 | 2889 | { |
abd0071c | 2890 | marker = XCAR (list); |
bd2cbd56 | 2891 | if (BUFFERP (marker)) |
c3762cbd | 2892 | { |
bd2cbd56 | 2893 | last_thing_searched = marker; |
c3762cbd DK |
2894 | break; |
2895 | } | |
2896 | if (i >= length) | |
2897 | break; | |
41c01205 DK |
2898 | if (NILP (marker)) |
2899 | { | |
2900 | search_regs.start[i] = -1; | |
abd0071c | 2901 | list = XCDR (list); |
41c01205 DK |
2902 | } |
2903 | else | |
2904 | { | |
2905 | int from; | |
abd0071c | 2906 | Lisp_Object m; |
e2811828 | 2907 | |
abd0071c | 2908 | m = marker; |
41c01205 DK |
2909 | if (MARKERP (marker)) |
2910 | { | |
2911 | if (XMARKER (marker)->buffer == 0) | |
2912 | XSETFASTINT (marker, 0); | |
2913 | else | |
2914 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
2915 | } | |
e2811828 | 2916 | |
41c01205 DK |
2917 | CHECK_NUMBER_COERCE_MARKER (marker); |
2918 | from = XINT (marker); | |
e2811828 | 2919 | |
abd0071c KS |
2920 | if (!NILP (reseat) && MARKERP (m)) |
2921 | { | |
2922 | if (EQ (reseat, Qevaporate)) | |
2923 | free_marker (m); | |
2924 | else | |
2925 | unchain_marker (XMARKER (m)); | |
9ad54a7e | 2926 | XSETCAR (list, Qnil); |
abd0071c KS |
2927 | } |
2928 | ||
2929 | if ((list = XCDR (list), !CONSP (list))) | |
2930 | break; | |
2931 | ||
2932 | m = marker = XCAR (list); | |
2933 | ||
41c01205 DK |
2934 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
2935 | XSETFASTINT (marker, 0); | |
e2811828 | 2936 | |
41c01205 DK |
2937 | CHECK_NUMBER_COERCE_MARKER (marker); |
2938 | search_regs.start[i] = from; | |
2939 | search_regs.end[i] = XINT (marker); | |
abd0071c KS |
2940 | |
2941 | if (!NILP (reseat) && MARKERP (m)) | |
2942 | { | |
2943 | if (EQ (reseat, Qevaporate)) | |
2944 | free_marker (m); | |
2945 | else | |
2946 | unchain_marker (XMARKER (m)); | |
9ad54a7e | 2947 | XSETCAR (list, Qnil); |
abd0071c | 2948 | } |
41c01205 | 2949 | } |
abd0071c | 2950 | list = XCDR (list); |
41c01205 | 2951 | } |
ca1d1d23 | 2952 | |
41c01205 DK |
2953 | for (; i < search_regs.num_regs; i++) |
2954 | search_regs.start[i] = -1; | |
2955 | } | |
ca1d1d23 | 2956 | |
177c0ea7 | 2957 | return Qnil; |
ca1d1d23 JB |
2958 | } |
2959 | ||
7074fde6 FP |
2960 | /* If non-zero the match data have been saved in saved_search_regs |
2961 | during the execution of a sentinel or filter. */ | |
75ebf74b | 2962 | static int search_regs_saved; |
7074fde6 | 2963 | static struct re_registers saved_search_regs; |
41c01205 | 2964 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
2965 | |
2966 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
2967 | if asynchronous code (filter or sentinel) is running. */ | |
2968 | static void | |
2969 | save_search_regs () | |
2970 | { | |
2971 | if (!search_regs_saved) | |
2972 | { | |
2973 | saved_search_regs.num_regs = search_regs.num_regs; | |
2974 | saved_search_regs.start = search_regs.start; | |
2975 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
2976 | saved_last_thing_searched = last_thing_searched; |
2977 | last_thing_searched = Qnil; | |
7074fde6 | 2978 | search_regs.num_regs = 0; |
2d4a771a RS |
2979 | search_regs.start = 0; |
2980 | search_regs.end = 0; | |
7074fde6 FP |
2981 | |
2982 | search_regs_saved = 1; | |
2983 | } | |
2984 | } | |
2985 | ||
2986 | /* Called upon exit from filters and sentinels. */ | |
2987 | void | |
abd0071c | 2988 | restore_search_regs () |
7074fde6 FP |
2989 | { |
2990 | if (search_regs_saved) | |
2991 | { | |
2992 | if (search_regs.num_regs > 0) | |
2993 | { | |
2994 | xfree (search_regs.start); | |
2995 | xfree (search_regs.end); | |
2996 | } | |
2997 | search_regs.num_regs = saved_search_regs.num_regs; | |
2998 | search_regs.start = saved_search_regs.start; | |
2999 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
3000 | last_thing_searched = saved_last_thing_searched; |
3001 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
3002 | search_regs_saved = 0; |
3003 | } | |
3004 | } | |
3005 | ||
abd0071c KS |
3006 | static Lisp_Object |
3007 | unwind_set_match_data (list) | |
3008 | Lisp_Object list; | |
3009 | { | |
51f10faa | 3010 | /* It is safe to free (evaporate) the markers immediately. */ |
abd0071c KS |
3011 | return Fset_match_data (list, Qevaporate); |
3012 | } | |
3013 | ||
3014 | /* Called to unwind protect the match data. */ | |
3015 | void | |
3016 | record_unwind_save_match_data () | |
3017 | { | |
3018 | record_unwind_protect (unwind_set_match_data, | |
3019 | Fmatch_data (Qnil, Qnil, Qnil)); | |
3020 | } | |
3021 | ||
ca1d1d23 JB |
3022 | /* Quote a string to inactivate reg-expr chars */ |
3023 | ||
3024 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 PJ |
3025 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
3026 | (string) | |
5806161b | 3027 | Lisp_Object string; |
ca1d1d23 JB |
3028 | { |
3029 | register unsigned char *in, *out, *end; | |
3030 | register unsigned char *temp; | |
0c8533c6 | 3031 | int backslashes_added = 0; |
ca1d1d23 | 3032 | |
b7826503 | 3033 | CHECK_STRING (string); |
ca1d1d23 | 3034 | |
d5db4077 | 3035 | temp = (unsigned char *) alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
3036 | |
3037 | /* Now copy the data into the new string, inserting escapes. */ | |
3038 | ||
d5db4077 KR |
3039 | in = SDATA (string); |
3040 | end = in + SBYTES (string); | |
177c0ea7 | 3041 | out = temp; |
ca1d1d23 JB |
3042 | |
3043 | for (; in != end; in++) | |
3044 | { | |
66bc6082 | 3045 | if (*in == '[' |
ca1d1d23 JB |
3046 | || *in == '*' || *in == '.' || *in == '\\' |
3047 | || *in == '?' || *in == '+' | |
3048 | || *in == '^' || *in == '$') | |
0c8533c6 | 3049 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
3050 | *out++ = *in; |
3051 | } | |
3052 | ||
3f8100f1 | 3053 | return make_specified_string (temp, |
d5db4077 | 3054 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
3055 | out - temp, |
3056 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 3057 | } |
177c0ea7 | 3058 | \f |
dfcf069d | 3059 | void |
ca1d1d23 JB |
3060 | syms_of_search () |
3061 | { | |
3062 | register int i; | |
3063 | ||
487282dc KH |
3064 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
3065 | { | |
3066 | searchbufs[i].buf.allocated = 100; | |
b23c0a83 | 3067 | searchbufs[i].buf.buffer = (unsigned char *) xmalloc (100); |
487282dc KH |
3068 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
3069 | searchbufs[i].regexp = Qnil; | |
ecdb561e | 3070 | searchbufs[i].whitespace_regexp = Qnil; |
487282dc | 3071 | staticpro (&searchbufs[i].regexp); |
aa77b5ce | 3072 | staticpro (&searchbufs[i].whitespace_regexp); |
487282dc KH |
3073 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); |
3074 | } | |
3075 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
3076 | |
3077 | Qsearch_failed = intern ("search-failed"); | |
3078 | staticpro (&Qsearch_failed); | |
3079 | Qinvalid_regexp = intern ("invalid-regexp"); | |
3080 | staticpro (&Qinvalid_regexp); | |
3081 | ||
3082 | Fput (Qsearch_failed, Qerror_conditions, | |
3083 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
3084 | Fput (Qsearch_failed, Qerror_message, | |
3085 | build_string ("Search failed")); | |
3086 | ||
3087 | Fput (Qinvalid_regexp, Qerror_conditions, | |
3088 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
3089 | Fput (Qinvalid_regexp, Qerror_message, | |
3090 | build_string ("Invalid regexp")); | |
3091 | ||
daa37602 JB |
3092 | last_thing_searched = Qnil; |
3093 | staticpro (&last_thing_searched); | |
3094 | ||
0f6af254 DK |
3095 | saved_last_thing_searched = Qnil; |
3096 | staticpro (&saved_last_thing_searched); | |
3097 | ||
41a33295 | 3098 | DEFVAR_LISP ("search-spaces-regexp", &Vsearch_spaces_regexp, |
e2811828 | 3099 | doc: /* Regexp to substitute for bunches of spaces in regexp search. |
f31a9a68 RS |
3100 | Some commands use this for user-specified regexps. |
3101 | Spaces that occur inside character classes or repetition operators | |
3102 | or other such regexp constructs are not replaced with this. | |
3103 | A value of nil (which is the normal value) means treat spaces literally. */); | |
41a33295 | 3104 | Vsearch_spaces_regexp = Qnil; |
f31a9a68 | 3105 | |
ca1d1d23 | 3106 | defsubr (&Slooking_at); |
b819a390 RS |
3107 | defsubr (&Sposix_looking_at); |
3108 | defsubr (&Sstring_match); | |
3109 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
3110 | defsubr (&Ssearch_forward); |
3111 | defsubr (&Ssearch_backward); | |
3112 | defsubr (&Sword_search_forward); | |
3113 | defsubr (&Sword_search_backward); | |
3114 | defsubr (&Sre_search_forward); | |
3115 | defsubr (&Sre_search_backward); | |
b819a390 RS |
3116 | defsubr (&Sposix_search_forward); |
3117 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
3118 | defsubr (&Sreplace_match); |
3119 | defsubr (&Smatch_beginning); | |
3120 | defsubr (&Smatch_end); | |
3121 | defsubr (&Smatch_data); | |
3f1c005b | 3122 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
3123 | defsubr (&Sregexp_quote); |
3124 | } | |
76fc1ea2 KH |
3125 | |
3126 | /* arch-tag: a6059d79-0552-4f14-a2cb-d379a4e3c78f | |
3127 | (do not change this comment) */ |