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" |
5679531d | 28 | #include "charset.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 | 92 | static int search_buffer (); |
01d09305 | 93 | static void matcher_overflow () NO_RETURN; |
b819a390 | 94 | |
ca1d1d23 JB |
95 | static void |
96 | matcher_overflow () | |
97 | { | |
98 | error ("Stack overflow in regexp matcher"); | |
99 | } | |
100 | ||
b819a390 RS |
101 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
102 | PATTERN is the pattern to compile. | |
103 | CP is the place to put the result. | |
facdc750 | 104 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
105 | REGP is the structure that says where to store the "register" |
106 | values that will result from matching this pattern. | |
107 | If it is 0, we should compile the pattern not to record any | |
108 | subexpression bounds. | |
109 | POSIX is nonzero if we want full backtracking (POSIX style) | |
5679531d KH |
110 | for this pattern. 0 means backtrack only enough to get a valid match. |
111 | MULTIBYTE is nonzero if we want to handle multibyte characters in | |
112 | PATTERN. 0 means all multibyte characters are recognized just as | |
ecdb561e RS |
113 | sequences of binary data. |
114 | ||
41a33295 | 115 | The behavior also depends on Vsearch_spaces_regexp. */ |
ca1d1d23 | 116 | |
487282dc | 117 | static void |
5679531d | 118 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte) |
487282dc | 119 | struct regexp_cache *cp; |
ca1d1d23 | 120 | Lisp_Object pattern; |
facdc750 | 121 | Lisp_Object translate; |
487282dc | 122 | struct re_registers *regp; |
b819a390 | 123 | int posix; |
5679531d | 124 | int multibyte; |
ca1d1d23 | 125 | { |
7276d3d8 | 126 | unsigned char *raw_pattern; |
f8bd51c4 | 127 | int raw_pattern_size; |
d451e4db | 128 | char *val; |
b819a390 | 129 | reg_syntax_t old; |
ca1d1d23 | 130 | |
f8bd51c4 KH |
131 | /* MULTIBYTE says whether the text to be searched is multibyte. |
132 | We must convert PATTERN to match that, or we will not really | |
133 | find things right. */ | |
134 | ||
135 | if (multibyte == STRING_MULTIBYTE (pattern)) | |
136 | { | |
d5db4077 KR |
137 | raw_pattern = (unsigned char *) SDATA (pattern); |
138 | raw_pattern_size = SBYTES (pattern); | |
f8bd51c4 KH |
139 | } |
140 | else if (multibyte) | |
141 | { | |
d5db4077 KR |
142 | raw_pattern_size = count_size_as_multibyte (SDATA (pattern), |
143 | SCHARS (pattern)); | |
7276d3d8 | 144 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
145 | copy_text (SDATA (pattern), raw_pattern, |
146 | SCHARS (pattern), 0, 1); | |
f8bd51c4 KH |
147 | } |
148 | else | |
149 | { | |
150 | /* Converting multibyte to single-byte. | |
151 | ||
152 | ??? Perhaps this conversion should be done in a special way | |
153 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
154 | so that only the multibyte chars which really correspond to | |
155 | the chosen single-byte character set can possibly match. */ | |
d5db4077 | 156 | raw_pattern_size = SCHARS (pattern); |
7276d3d8 | 157 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
158 | copy_text (SDATA (pattern), raw_pattern, |
159 | SBYTES (pattern), 1, 0); | |
f8bd51c4 KH |
160 | } |
161 | ||
487282dc | 162 | cp->regexp = Qnil; |
59fab369 | 163 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 164 | cp->posix = posix; |
5679531d | 165 | cp->buf.multibyte = multibyte; |
41a33295 | 166 | cp->whitespace_regexp = Vsearch_spaces_regexp; |
9ac0d9e0 | 167 | BLOCK_INPUT; |
fb4a568d | 168 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 169 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
f31a9a68 | 170 | |
41a33295 RS |
171 | re_set_whitespace_regexp (NILP (Vsearch_spaces_regexp) ? NULL |
172 | : SDATA (Vsearch_spaces_regexp)); | |
f31a9a68 | 173 | |
7276d3d8 RS |
174 | val = (char *) re_compile_pattern ((char *)raw_pattern, |
175 | raw_pattern_size, &cp->buf); | |
f31a9a68 RS |
176 | |
177 | re_set_whitespace_regexp (NULL); | |
178 | ||
b819a390 | 179 | re_set_syntax (old); |
9ac0d9e0 | 180 | UNBLOCK_INPUT; |
ca1d1d23 | 181 | if (val) |
487282dc | 182 | Fsignal (Qinvalid_regexp, Fcons (build_string (val), Qnil)); |
1113d9db | 183 | |
487282dc | 184 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
185 | } |
186 | ||
6efc7887 RS |
187 | /* Shrink each compiled regexp buffer in the cache |
188 | to the size actually used right now. | |
189 | This is called from garbage collection. */ | |
190 | ||
191 | void | |
192 | shrink_regexp_cache () | |
193 | { | |
a968f437 | 194 | struct regexp_cache *cp; |
6efc7887 RS |
195 | |
196 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
197 | { | |
198 | cp->buf.allocated = cp->buf.used; | |
199 | cp->buf.buffer | |
b23c0a83 | 200 | = (unsigned char *) xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
201 | } |
202 | } | |
203 | ||
487282dc | 204 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
205 | the cache. |
206 | PATTERN is the pattern to compile. | |
facdc750 | 207 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
208 | REGP is the structure that says where to store the "register" |
209 | values that will result from matching this pattern. | |
210 | If it is 0, we should compile the pattern not to record any | |
211 | subexpression bounds. | |
212 | POSIX is nonzero if we want full backtracking (POSIX style) | |
213 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
214 | |
215 | struct re_pattern_buffer * | |
0c8533c6 | 216 | compile_pattern (pattern, regp, translate, posix, multibyte) |
487282dc KH |
217 | Lisp_Object pattern; |
218 | struct re_registers *regp; | |
facdc750 | 219 | Lisp_Object translate; |
0c8533c6 | 220 | int posix, multibyte; |
487282dc KH |
221 | { |
222 | struct regexp_cache *cp, **cpp; | |
223 | ||
224 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
225 | { | |
226 | cp = *cpp; | |
f1b9c7c1 KR |
227 | /* Entries are initialized to nil, and may be set to nil by |
228 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
49a5f770 KR |
229 | string accessors in those cases. However, compile_pattern_1 |
230 | is only applied to the cache entry we pick here to reuse. So | |
231 | nil should never appear before a non-nil entry. */ | |
7c752c80 | 232 | if (NILP (cp->regexp)) |
f1b9c7c1 | 233 | goto compile_it; |
d5db4077 | 234 | if (SCHARS (cp->regexp) == SCHARS (pattern) |
cf69b13e | 235 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 236 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 237 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d | 238 | && cp->posix == posix |
ecdb561e | 239 | && cp->buf.multibyte == multibyte |
41a33295 | 240 | && !NILP (Fequal (cp->whitespace_regexp, Vsearch_spaces_regexp))) |
487282dc KH |
241 | break; |
242 | ||
f1b9c7c1 KR |
243 | /* If we're at the end of the cache, compile into the nil cell |
244 | we found, or the last (least recently used) cell with a | |
245 | string value. */ | |
487282dc KH |
246 | if (cp->next == 0) |
247 | { | |
f1b9c7c1 | 248 | compile_it: |
5679531d | 249 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte); |
487282dc KH |
250 | break; |
251 | } | |
252 | } | |
253 | ||
254 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
255 | either because we found it in the cache or because we just compiled it. | |
256 | Move it to the front of the queue to mark it as most recently used. */ | |
257 | *cpp = cp->next; | |
258 | cp->next = searchbuf_head; | |
259 | searchbuf_head = cp; | |
1113d9db | 260 | |
6639708c RS |
261 | /* Advise the searching functions about the space we have allocated |
262 | for register data. */ | |
263 | if (regp) | |
264 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
265 | ||
487282dc | 266 | return &cp->buf; |
ca1d1d23 JB |
267 | } |
268 | ||
269 | /* Error condition used for failing searches */ | |
270 | Lisp_Object Qsearch_failed; | |
271 | ||
272 | Lisp_Object | |
273 | signal_failure (arg) | |
274 | Lisp_Object arg; | |
275 | { | |
276 | Fsignal (Qsearch_failed, Fcons (arg, Qnil)); | |
277 | return Qnil; | |
278 | } | |
279 | \f | |
b819a390 RS |
280 | static Lisp_Object |
281 | looking_at_1 (string, posix) | |
ca1d1d23 | 282 | Lisp_Object string; |
b819a390 | 283 | int posix; |
ca1d1d23 JB |
284 | { |
285 | Lisp_Object val; | |
286 | unsigned char *p1, *p2; | |
287 | int s1, s2; | |
288 | register int i; | |
487282dc | 289 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 290 | |
7074fde6 FP |
291 | if (running_asynch_code) |
292 | save_search_regs (); | |
293 | ||
b7826503 | 294 | CHECK_STRING (string); |
487282dc KH |
295 | bufp = compile_pattern (string, &search_regs, |
296 | (!NILP (current_buffer->case_fold_search) | |
0190922f | 297 | ? current_buffer->case_canon_table : Qnil), |
0c8533c6 RS |
298 | posix, |
299 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 JB |
300 | |
301 | immediate_quit = 1; | |
302 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
303 | ||
304 | /* Get pointers and sizes of the two strings | |
305 | that make up the visible portion of the buffer. */ | |
306 | ||
307 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 308 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 309 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 310 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
311 | if (s1 < 0) |
312 | { | |
313 | p2 = p1; | |
fa8ed3e0 | 314 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
315 | s1 = 0; |
316 | } | |
317 | if (s2 < 0) | |
318 | { | |
fa8ed3e0 | 319 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
320 | s2 = 0; |
321 | } | |
8bb43c28 RS |
322 | |
323 | re_match_object = Qnil; | |
177c0ea7 | 324 | |
487282dc | 325 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
fa8ed3e0 RS |
326 | PT_BYTE - BEGV_BYTE, &search_regs, |
327 | ZV_BYTE - BEGV_BYTE); | |
de182d70 | 328 | immediate_quit = 0; |
177c0ea7 | 329 | |
ca1d1d23 JB |
330 | if (i == -2) |
331 | matcher_overflow (); | |
332 | ||
333 | val = (0 <= i ? Qt : Qnil); | |
fa8ed3e0 RS |
334 | if (i >= 0) |
335 | for (i = 0; i < search_regs.num_regs; i++) | |
336 | if (search_regs.start[i] >= 0) | |
337 | { | |
338 | search_regs.start[i] | |
339 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
340 | search_regs.end[i] | |
341 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
342 | } | |
a3668d92 | 343 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
344 | return val; |
345 | } | |
346 | ||
b819a390 | 347 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
348 | doc: /* Return t if text after point matches regular expression REGEXP. |
349 | This function modifies the match data that `match-beginning', | |
350 | `match-end' and `match-data' access; save and restore the match | |
351 | data if you want to preserve them. */) | |
352 | (regexp) | |
94f94972 | 353 | Lisp_Object regexp; |
b819a390 | 354 | { |
94f94972 | 355 | return looking_at_1 (regexp, 0); |
b819a390 RS |
356 | } |
357 | ||
358 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
359 | doc: /* Return t if text after point matches regular expression REGEXP. |
360 | Find the longest match, in accord with Posix regular expression rules. | |
361 | This function modifies the match data that `match-beginning', | |
362 | `match-end' and `match-data' access; save and restore the match | |
363 | data if you want to preserve them. */) | |
364 | (regexp) | |
94f94972 | 365 | Lisp_Object regexp; |
b819a390 | 366 | { |
94f94972 | 367 | return looking_at_1 (regexp, 1); |
b819a390 RS |
368 | } |
369 | \f | |
370 | static Lisp_Object | |
371 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 372 | Lisp_Object regexp, string, start; |
b819a390 | 373 | int posix; |
ca1d1d23 JB |
374 | { |
375 | int val; | |
487282dc | 376 | struct re_pattern_buffer *bufp; |
0c8533c6 RS |
377 | int pos, pos_byte; |
378 | int i; | |
ca1d1d23 | 379 | |
7074fde6 FP |
380 | if (running_asynch_code) |
381 | save_search_regs (); | |
382 | ||
b7826503 PJ |
383 | CHECK_STRING (regexp); |
384 | CHECK_STRING (string); | |
ca1d1d23 JB |
385 | |
386 | if (NILP (start)) | |
0c8533c6 | 387 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
388 | else |
389 | { | |
d5db4077 | 390 | int len = SCHARS (string); |
ca1d1d23 | 391 | |
b7826503 | 392 | CHECK_NUMBER (start); |
0c8533c6 RS |
393 | pos = XINT (start); |
394 | if (pos < 0 && -pos <= len) | |
395 | pos = len + pos; | |
396 | else if (0 > pos || pos > len) | |
ca1d1d23 | 397 | args_out_of_range (string, start); |
0c8533c6 | 398 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
399 | } |
400 | ||
487282dc KH |
401 | bufp = compile_pattern (regexp, &search_regs, |
402 | (!NILP (current_buffer->case_fold_search) | |
0190922f | 403 | ? current_buffer->case_canon_table : Qnil), |
0c8533c6 RS |
404 | posix, |
405 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 406 | immediate_quit = 1; |
8bb43c28 | 407 | re_match_object = string; |
177c0ea7 | 408 | |
d5db4077 KR |
409 | val = re_search (bufp, (char *) SDATA (string), |
410 | SBYTES (string), pos_byte, | |
411 | SBYTES (string) - pos_byte, | |
ca1d1d23 JB |
412 | &search_regs); |
413 | immediate_quit = 0; | |
daa37602 | 414 | last_thing_searched = Qt; |
ca1d1d23 JB |
415 | if (val == -2) |
416 | matcher_overflow (); | |
417 | if (val < 0) return Qnil; | |
0c8533c6 RS |
418 | |
419 | for (i = 0; i < search_regs.num_regs; i++) | |
420 | if (search_regs.start[i] >= 0) | |
421 | { | |
422 | search_regs.start[i] | |
423 | = string_byte_to_char (string, search_regs.start[i]); | |
424 | search_regs.end[i] | |
425 | = string_byte_to_char (string, search_regs.end[i]); | |
426 | } | |
427 | ||
428 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 429 | } |
e59a8453 | 430 | |
b819a390 | 431 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 | 432 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
b85acc4b | 433 | Matching ignores case if `case-fold-search' is non-nil. |
8c1a1077 PJ |
434 | If third arg START is non-nil, start search at that index in STRING. |
435 | For index of first char beyond the match, do (match-end 0). | |
436 | `match-end' and `match-beginning' also give indices of substrings | |
2bd2f32d RS |
437 | matched by parenthesis constructs in the pattern. |
438 | ||
439 | You can use the function `match-string' to extract the substrings | |
440 | matched by the parenthesis constructions in REGEXP. */) | |
8c1a1077 | 441 | (regexp, string, start) |
b819a390 RS |
442 | Lisp_Object regexp, string, start; |
443 | { | |
444 | return string_match_1 (regexp, string, start, 0); | |
445 | } | |
446 | ||
447 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
448 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
449 | Find the longest match, in accord with Posix regular expression rules. | |
450 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
451 | If third arg START is non-nil, start search at that index in STRING. | |
452 | For index of first char beyond the match, do (match-end 0). | |
453 | `match-end' and `match-beginning' also give indices of substrings | |
454 | matched by parenthesis constructs in the pattern. */) | |
455 | (regexp, string, start) | |
b819a390 RS |
456 | Lisp_Object regexp, string, start; |
457 | { | |
458 | return string_match_1 (regexp, string, start, 1); | |
459 | } | |
460 | ||
e59a8453 RS |
461 | /* Match REGEXP against STRING, searching all of STRING, |
462 | and return the index of the match, or negative on failure. | |
463 | This does not clobber the match data. */ | |
464 | ||
465 | int | |
466 | fast_string_match (regexp, string) | |
467 | Lisp_Object regexp, string; | |
468 | { | |
469 | int val; | |
487282dc | 470 | struct re_pattern_buffer *bufp; |
e59a8453 | 471 | |
facdc750 RS |
472 | bufp = compile_pattern (regexp, 0, Qnil, |
473 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 474 | immediate_quit = 1; |
8bb43c28 | 475 | re_match_object = string; |
177c0ea7 | 476 | |
d5db4077 KR |
477 | val = re_search (bufp, (char *) SDATA (string), |
478 | SBYTES (string), 0, | |
479 | SBYTES (string), 0); | |
e59a8453 RS |
480 | immediate_quit = 0; |
481 | return val; | |
482 | } | |
5679531d KH |
483 | |
484 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
485 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
486 | This does not clobber the match data. |
487 | We assume that STRING contains single-byte characters. */ | |
5679531d KH |
488 | |
489 | extern Lisp_Object Vascii_downcase_table; | |
490 | ||
491 | int | |
b4577c63 | 492 | fast_c_string_match_ignore_case (regexp, string) |
5679531d | 493 | Lisp_Object regexp; |
96b80561 | 494 | const char *string; |
5679531d KH |
495 | { |
496 | int val; | |
497 | struct re_pattern_buffer *bufp; | |
498 | int len = strlen (string); | |
499 | ||
0c8533c6 | 500 | regexp = string_make_unibyte (regexp); |
b4577c63 | 501 | re_match_object = Qt; |
5679531d | 502 | bufp = compile_pattern (regexp, 0, |
0190922f | 503 | Vascii_canon_table, 0, |
f8bd51c4 | 504 | 0); |
5679531d KH |
505 | immediate_quit = 1; |
506 | val = re_search (bufp, string, len, 0, len, 0); | |
507 | immediate_quit = 0; | |
508 | return val; | |
509 | } | |
be5f4dfb KH |
510 | |
511 | /* Like fast_string_match but ignore case. */ | |
512 | ||
513 | int | |
514 | fast_string_match_ignore_case (regexp, string) | |
515 | Lisp_Object regexp, string; | |
516 | { | |
517 | int val; | |
518 | struct re_pattern_buffer *bufp; | |
519 | ||
0190922f | 520 | bufp = compile_pattern (regexp, 0, Vascii_canon_table, |
be5f4dfb KH |
521 | 0, STRING_MULTIBYTE (string)); |
522 | immediate_quit = 1; | |
523 | re_match_object = string; | |
524 | ||
525 | val = re_search (bufp, (char *) SDATA (string), | |
526 | SBYTES (string), 0, | |
527 | SBYTES (string), 0); | |
528 | immediate_quit = 0; | |
529 | return val; | |
530 | } | |
ca1d1d23 | 531 | \f |
9169c321 JB |
532 | /* The newline cache: remembering which sections of text have no newlines. */ |
533 | ||
534 | /* If the user has requested newline caching, make sure it's on. | |
535 | Otherwise, make sure it's off. | |
536 | This is our cheezy way of associating an action with the change of | |
537 | state of a buffer-local variable. */ | |
538 | static void | |
539 | newline_cache_on_off (buf) | |
540 | struct buffer *buf; | |
541 | { | |
542 | if (NILP (buf->cache_long_line_scans)) | |
543 | { | |
544 | /* It should be off. */ | |
545 | if (buf->newline_cache) | |
546 | { | |
547 | free_region_cache (buf->newline_cache); | |
548 | buf->newline_cache = 0; | |
549 | } | |
550 | } | |
551 | else | |
552 | { | |
553 | /* It should be on. */ | |
554 | if (buf->newline_cache == 0) | |
555 | buf->newline_cache = new_region_cache (); | |
556 | } | |
557 | } | |
558 | ||
559 | \f | |
560 | /* Search for COUNT instances of the character TARGET between START and END. | |
561 | ||
562 | If COUNT is positive, search forwards; END must be >= START. | |
563 | If COUNT is negative, search backwards for the -COUNTth instance; | |
564 | END must be <= START. | |
565 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
566 | ||
567 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
568 | direction indicated by COUNT. | |
ffd56f97 JB |
569 | |
570 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
a9f2a45f | 571 | position past the COUNTth match. Note that for reverse motion |
5bfe95c9 | 572 | this is not the same as the usual convention for Emacs motion commands. |
ffd56f97 | 573 | |
9169c321 JB |
574 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
575 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 576 | |
087a5f81 RS |
577 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
578 | except when inside redisplay. */ | |
579 | ||
dfcf069d | 580 | int |
9169c321 JB |
581 | scan_buffer (target, start, end, count, shortage, allow_quit) |
582 | register int target; | |
583 | int start, end; | |
584 | int count; | |
585 | int *shortage; | |
087a5f81 | 586 | int allow_quit; |
ca1d1d23 | 587 | { |
9169c321 | 588 | struct region_cache *newline_cache; |
177c0ea7 | 589 | int direction; |
ffd56f97 | 590 | |
9169c321 JB |
591 | if (count > 0) |
592 | { | |
593 | direction = 1; | |
594 | if (! end) end = ZV; | |
595 | } | |
596 | else | |
597 | { | |
598 | direction = -1; | |
599 | if (! end) end = BEGV; | |
600 | } | |
ffd56f97 | 601 | |
9169c321 JB |
602 | newline_cache_on_off (current_buffer); |
603 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
604 | |
605 | if (shortage != 0) | |
606 | *shortage = 0; | |
607 | ||
087a5f81 | 608 | immediate_quit = allow_quit; |
ca1d1d23 | 609 | |
ffd56f97 | 610 | if (count > 0) |
9169c321 | 611 | while (start != end) |
ca1d1d23 | 612 | { |
9169c321 JB |
613 | /* Our innermost scanning loop is very simple; it doesn't know |
614 | about gaps, buffer ends, or the newline cache. ceiling is | |
615 | the position of the last character before the next such | |
616 | obstacle --- the last character the dumb search loop should | |
617 | examine. */ | |
fa8ed3e0 RS |
618 | int ceiling_byte = CHAR_TO_BYTE (end) - 1; |
619 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 620 | int tem; |
9169c321 JB |
621 | |
622 | /* If we're looking for a newline, consult the newline cache | |
623 | to see where we can avoid some scanning. */ | |
624 | if (target == '\n' && newline_cache) | |
625 | { | |
626 | int next_change; | |
627 | immediate_quit = 0; | |
628 | while (region_cache_forward | |
fa8ed3e0 RS |
629 | (current_buffer, newline_cache, start_byte, &next_change)) |
630 | start_byte = next_change; | |
cbe0db0d | 631 | immediate_quit = allow_quit; |
9169c321 | 632 | |
fa8ed3e0 RS |
633 | /* START should never be after END. */ |
634 | if (start_byte > ceiling_byte) | |
635 | start_byte = ceiling_byte; | |
9169c321 JB |
636 | |
637 | /* Now the text after start is an unknown region, and | |
638 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 639 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
640 | } |
641 | ||
642 | /* The dumb loop can only scan text stored in contiguous | |
643 | bytes. BUFFER_CEILING_OF returns the last character | |
644 | position that is contiguous, so the ceiling is the | |
645 | position after that. */ | |
67ce527d KH |
646 | tem = BUFFER_CEILING_OF (start_byte); |
647 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
648 | |
649 | { | |
177c0ea7 | 650 | /* The termination address of the dumb loop. */ |
fa8ed3e0 RS |
651 | register unsigned char *ceiling_addr |
652 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
653 | register unsigned char *cursor | |
654 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
655 | unsigned char *base = cursor; |
656 | ||
657 | while (cursor < ceiling_addr) | |
658 | { | |
659 | unsigned char *scan_start = cursor; | |
660 | ||
661 | /* The dumb loop. */ | |
662 | while (*cursor != target && ++cursor < ceiling_addr) | |
663 | ; | |
664 | ||
665 | /* If we're looking for newlines, cache the fact that | |
666 | the region from start to cursor is free of them. */ | |
667 | if (target == '\n' && newline_cache) | |
668 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
669 | start_byte + scan_start - base, |
670 | start_byte + cursor - base); | |
9169c321 JB |
671 | |
672 | /* Did we find the target character? */ | |
673 | if (cursor < ceiling_addr) | |
674 | { | |
675 | if (--count == 0) | |
676 | { | |
677 | immediate_quit = 0; | |
fa8ed3e0 | 678 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
679 | } |
680 | cursor++; | |
681 | } | |
682 | } | |
683 | ||
fa8ed3e0 | 684 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 685 | } |
ca1d1d23 JB |
686 | } |
687 | else | |
9169c321 JB |
688 | while (start > end) |
689 | { | |
690 | /* The last character to check before the next obstacle. */ | |
fa8ed3e0 RS |
691 | int ceiling_byte = CHAR_TO_BYTE (end); |
692 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 693 | int tem; |
9169c321 JB |
694 | |
695 | /* Consult the newline cache, if appropriate. */ | |
696 | if (target == '\n' && newline_cache) | |
697 | { | |
698 | int next_change; | |
699 | immediate_quit = 0; | |
700 | while (region_cache_backward | |
fa8ed3e0 RS |
701 | (current_buffer, newline_cache, start_byte, &next_change)) |
702 | start_byte = next_change; | |
cbe0db0d | 703 | immediate_quit = allow_quit; |
9169c321 JB |
704 | |
705 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
706 | if (start_byte <= ceiling_byte) |
707 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
708 | |
709 | /* Now the text before start is an unknown region, and | |
710 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 711 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
712 | } |
713 | ||
714 | /* Stop scanning before the gap. */ | |
67ce527d KH |
715 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
716 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
717 | |
718 | { | |
719 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
720 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
721 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
722 | unsigned char *base = cursor; |
723 | ||
724 | while (cursor >= ceiling_addr) | |
725 | { | |
726 | unsigned char *scan_start = cursor; | |
727 | ||
728 | while (*cursor != target && --cursor >= ceiling_addr) | |
729 | ; | |
730 | ||
731 | /* If we're looking for newlines, cache the fact that | |
732 | the region from after the cursor to start is free of them. */ | |
733 | if (target == '\n' && newline_cache) | |
734 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
735 | start_byte + cursor - base, |
736 | start_byte + scan_start - base); | |
9169c321 JB |
737 | |
738 | /* Did we find the target character? */ | |
739 | if (cursor >= ceiling_addr) | |
740 | { | |
741 | if (++count >= 0) | |
742 | { | |
743 | immediate_quit = 0; | |
fa8ed3e0 | 744 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
745 | } |
746 | cursor--; | |
747 | } | |
748 | } | |
749 | ||
fa8ed3e0 | 750 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
751 | } |
752 | } | |
753 | ||
ca1d1d23 JB |
754 | immediate_quit = 0; |
755 | if (shortage != 0) | |
ffd56f97 | 756 | *shortage = count * direction; |
9169c321 | 757 | return start; |
ca1d1d23 | 758 | } |
fa8ed3e0 RS |
759 | \f |
760 | /* Search for COUNT instances of a line boundary, which means either a | |
761 | newline or (if selective display enabled) a carriage return. | |
762 | Start at START. If COUNT is negative, search backwards. | |
763 | ||
764 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
765 | ||
766 | If we find COUNT instances. we position after (always after, | |
767 | even if scanning backwards) the COUNTth match, and return 0. | |
768 | ||
769 | If we don't find COUNT instances before reaching the end of the | |
770 | buffer (or the beginning, if scanning backwards), we return | |
771 | the number of line boundaries left unfound, and position at | |
772 | the limit we bumped up against. | |
773 | ||
774 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 775 | except in special cases. */ |
ca1d1d23 | 776 | |
63fa018d | 777 | int |
fa8ed3e0 RS |
778 | scan_newline (start, start_byte, limit, limit_byte, count, allow_quit) |
779 | int start, start_byte; | |
780 | int limit, limit_byte; | |
781 | register int count; | |
782 | int allow_quit; | |
63fa018d | 783 | { |
fa8ed3e0 RS |
784 | int direction = ((count > 0) ? 1 : -1); |
785 | ||
786 | register unsigned char *cursor; | |
787 | unsigned char *base; | |
788 | ||
789 | register int ceiling; | |
790 | register unsigned char *ceiling_addr; | |
791 | ||
d5d57b92 RS |
792 | int old_immediate_quit = immediate_quit; |
793 | ||
fa8ed3e0 RS |
794 | /* The code that follows is like scan_buffer |
795 | but checks for either newline or carriage return. */ | |
796 | ||
d5d57b92 RS |
797 | if (allow_quit) |
798 | immediate_quit++; | |
fa8ed3e0 RS |
799 | |
800 | start_byte = CHAR_TO_BYTE (start); | |
801 | ||
802 | if (count > 0) | |
803 | { | |
804 | while (start_byte < limit_byte) | |
805 | { | |
806 | ceiling = BUFFER_CEILING_OF (start_byte); | |
807 | ceiling = min (limit_byte - 1, ceiling); | |
808 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
809 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
810 | while (1) | |
811 | { | |
812 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
813 | ; | |
814 | ||
815 | if (cursor != ceiling_addr) | |
816 | { | |
817 | if (--count == 0) | |
818 | { | |
d5d57b92 | 819 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
820 | start_byte = start_byte + cursor - base + 1; |
821 | start = BYTE_TO_CHAR (start_byte); | |
822 | TEMP_SET_PT_BOTH (start, start_byte); | |
823 | return 0; | |
824 | } | |
825 | else | |
826 | if (++cursor == ceiling_addr) | |
827 | break; | |
828 | } | |
829 | else | |
830 | break; | |
831 | } | |
832 | start_byte += cursor - base; | |
833 | } | |
834 | } | |
835 | else | |
836 | { | |
fa8ed3e0 RS |
837 | while (start_byte > limit_byte) |
838 | { | |
839 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
840 | ceiling = max (limit_byte, ceiling); | |
841 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
842 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
843 | while (1) | |
844 | { | |
845 | while (--cursor != ceiling_addr && *cursor != '\n') | |
846 | ; | |
847 | ||
848 | if (cursor != ceiling_addr) | |
849 | { | |
850 | if (++count == 0) | |
851 | { | |
d5d57b92 | 852 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
853 | /* Return the position AFTER the match we found. */ |
854 | start_byte = start_byte + cursor - base + 1; | |
855 | start = BYTE_TO_CHAR (start_byte); | |
856 | TEMP_SET_PT_BOTH (start, start_byte); | |
857 | return 0; | |
858 | } | |
859 | } | |
860 | else | |
861 | break; | |
862 | } | |
863 | /* Here we add 1 to compensate for the last decrement | |
864 | of CURSOR, which took it past the valid range. */ | |
865 | start_byte += cursor - base + 1; | |
866 | } | |
867 | } | |
868 | ||
869 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 870 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
871 | |
872 | return count * direction; | |
63fa018d RS |
873 | } |
874 | ||
ca1d1d23 | 875 | int |
fa8ed3e0 | 876 | find_next_newline_no_quit (from, cnt) |
ca1d1d23 JB |
877 | register int from, cnt; |
878 | { | |
fa8ed3e0 | 879 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
9169c321 JB |
880 | } |
881 | ||
9169c321 JB |
882 | /* Like find_next_newline, but returns position before the newline, |
883 | not after, and only search up to TO. This isn't just | |
884 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 885 | |
9169c321 JB |
886 | int |
887 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 888 | int from, to, cnt; |
9169c321 JB |
889 | { |
890 | int shortage; | |
891 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
892 | ||
893 | if (shortage == 0) | |
894 | pos--; | |
177c0ea7 | 895 | |
9169c321 | 896 | return pos; |
ca1d1d23 JB |
897 | } |
898 | \f | |
ca1d1d23 JB |
899 | /* Subroutines of Lisp buffer search functions. */ |
900 | ||
901 | static Lisp_Object | |
b819a390 | 902 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
903 | Lisp_Object string, bound, noerror, count; |
904 | int direction; | |
905 | int RE; | |
b819a390 | 906 | int posix; |
ca1d1d23 JB |
907 | { |
908 | register int np; | |
9f43ad85 | 909 | int lim, lim_byte; |
ca1d1d23 JB |
910 | int n = direction; |
911 | ||
912 | if (!NILP (count)) | |
913 | { | |
b7826503 | 914 | CHECK_NUMBER (count); |
ca1d1d23 JB |
915 | n *= XINT (count); |
916 | } | |
917 | ||
b7826503 | 918 | CHECK_STRING (string); |
ca1d1d23 | 919 | if (NILP (bound)) |
9f43ad85 RS |
920 | { |
921 | if (n > 0) | |
922 | lim = ZV, lim_byte = ZV_BYTE; | |
923 | else | |
924 | lim = BEGV, lim_byte = BEGV_BYTE; | |
925 | } | |
ca1d1d23 JB |
926 | else |
927 | { | |
b7826503 | 928 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 929 | lim = XINT (bound); |
6ec8bbd2 | 930 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
931 | error ("Invalid search bound (wrong side of point)"); |
932 | if (lim > ZV) | |
9f43ad85 | 933 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 934 | else if (lim < BEGV) |
9f43ad85 | 935 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
936 | else |
937 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
938 | } |
939 | ||
9f43ad85 | 940 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
ca1d1d23 | 941 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 942 | ? current_buffer->case_canon_table |
3135e9fd | 943 | : Qnil), |
ca1d1d23 | 944 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 945 | ? current_buffer->case_eqv_table |
3135e9fd | 946 | : Qnil), |
b819a390 | 947 | posix); |
ca1d1d23 JB |
948 | if (np <= 0) |
949 | { | |
950 | if (NILP (noerror)) | |
951 | return signal_failure (string); | |
952 | if (!EQ (noerror, Qt)) | |
953 | { | |
954 | if (lim < BEGV || lim > ZV) | |
955 | abort (); | |
9f43ad85 | 956 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
957 | return Qnil; |
958 | #if 0 /* This would be clean, but maybe programs depend on | |
959 | a value of nil here. */ | |
481399bf | 960 | np = lim; |
a5f217b8 | 961 | #endif |
ca1d1d23 | 962 | } |
481399bf RS |
963 | else |
964 | return Qnil; | |
ca1d1d23 JB |
965 | } |
966 | ||
967 | if (np < BEGV || np > ZV) | |
968 | abort (); | |
969 | ||
970 | SET_PT (np); | |
971 | ||
972 | return make_number (np); | |
973 | } | |
974 | \f | |
fa8ed3e0 RS |
975 | /* Return 1 if REGEXP it matches just one constant string. */ |
976 | ||
b6d6a51c KH |
977 | static int |
978 | trivial_regexp_p (regexp) | |
979 | Lisp_Object regexp; | |
980 | { | |
d5db4077 KR |
981 | int len = SBYTES (regexp); |
982 | unsigned char *s = SDATA (regexp); | |
b6d6a51c KH |
983 | while (--len >= 0) |
984 | { | |
985 | switch (*s++) | |
986 | { | |
987 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
988 | return 0; | |
989 | case '\\': | |
990 | if (--len < 0) | |
991 | return 0; | |
992 | switch (*s++) | |
993 | { | |
994 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
995 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 996 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 997 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 998 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
999 | case '6': case '7': case '8': case '9': |
1000 | return 0; | |
1001 | } | |
1002 | } | |
1003 | } | |
1004 | return 1; | |
1005 | } | |
1006 | ||
ca325161 | 1007 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 1008 | starting at position POS and stopping at position LIM, |
b819a390 | 1009 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
1010 | a regular expression if RE is true. |
1011 | ||
1012 | If N is positive, searching is forward and LIM must be greater than POS. | |
1013 | If N is negative, searching is backward and LIM must be less than POS. | |
1014 | ||
facdc750 | 1015 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 1016 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
1017 | (if searching backward) or the end (if searching forward). |
1018 | ||
1019 | POSIX is nonzero if we want full backtracking (POSIX style) | |
1020 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 1021 | |
aff2ce94 RS |
1022 | #define TRANSLATE(out, trt, d) \ |
1023 | do \ | |
1024 | { \ | |
1025 | if (! NILP (trt)) \ | |
1026 | { \ | |
1027 | Lisp_Object temp; \ | |
1028 | temp = Faref (trt, make_number (d)); \ | |
1029 | if (INTEGERP (temp)) \ | |
1030 | out = XINT (temp); \ | |
1031 | else \ | |
1032 | out = d; \ | |
1033 | } \ | |
1034 | else \ | |
1035 | out = d; \ | |
1036 | } \ | |
1037 | while (0) | |
facdc750 | 1038 | |
b819a390 | 1039 | static int |
9f43ad85 RS |
1040 | search_buffer (string, pos, pos_byte, lim, lim_byte, n, |
1041 | RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
1042 | Lisp_Object string; |
1043 | int pos; | |
9f43ad85 | 1044 | int pos_byte; |
ca1d1d23 | 1045 | int lim; |
9f43ad85 | 1046 | int lim_byte; |
ca1d1d23 JB |
1047 | int n; |
1048 | int RE; | |
facdc750 RS |
1049 | Lisp_Object trt; |
1050 | Lisp_Object inverse_trt; | |
b819a390 | 1051 | int posix; |
ca1d1d23 | 1052 | { |
d5db4077 KR |
1053 | int len = SCHARS (string); |
1054 | int len_byte = SBYTES (string); | |
facdc750 | 1055 | register int i; |
ca1d1d23 | 1056 | |
7074fde6 FP |
1057 | if (running_asynch_code) |
1058 | save_search_regs (); | |
1059 | ||
a7e4cdde | 1060 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1061 | /* Null string is found at starting position. */ |
a7e4cdde | 1062 | if (len == 0 || n == 0) |
ca325161 | 1063 | { |
0353b28f | 1064 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1065 | return pos; |
1066 | } | |
3f57a499 | 1067 | |
41a33295 | 1068 | if (RE && !(trivial_regexp_p (string) && NILP (Vsearch_spaces_regexp))) |
ca1d1d23 | 1069 | { |
facdc750 RS |
1070 | unsigned char *p1, *p2; |
1071 | int s1, s2; | |
487282dc KH |
1072 | struct re_pattern_buffer *bufp; |
1073 | ||
0c8533c6 RS |
1074 | bufp = compile_pattern (string, &search_regs, trt, posix, |
1075 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 | 1076 | |
ca1d1d23 JB |
1077 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1078 | because letting this function finish | |
1079 | can take too long. */ | |
1080 | QUIT; /* Do a pending quit right away, | |
1081 | to avoid paradoxical behavior */ | |
1082 | /* Get pointers and sizes of the two strings | |
1083 | that make up the visible portion of the buffer. */ | |
1084 | ||
1085 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1086 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1087 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1088 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1089 | if (s1 < 0) |
1090 | { | |
1091 | p2 = p1; | |
fa8ed3e0 | 1092 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1093 | s1 = 0; |
1094 | } | |
1095 | if (s2 < 0) | |
1096 | { | |
fa8ed3e0 | 1097 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1098 | s2 = 0; |
1099 | } | |
8bb43c28 | 1100 | re_match_object = Qnil; |
177c0ea7 | 1101 | |
ca1d1d23 JB |
1102 | while (n < 0) |
1103 | { | |
42db823b | 1104 | int val; |
487282dc | 1105 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1106 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1107 | &search_regs, | |
42db823b | 1108 | /* Don't allow match past current point */ |
4996330b | 1109 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1110 | if (val == -2) |
b6d6a51c KH |
1111 | { |
1112 | matcher_overflow (); | |
1113 | } | |
ca1d1d23 JB |
1114 | if (val >= 0) |
1115 | { | |
26aff150 | 1116 | pos_byte = search_regs.start[0] + BEGV_BYTE; |
4746118a | 1117 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1118 | if (search_regs.start[i] >= 0) |
1119 | { | |
fa8ed3e0 RS |
1120 | search_regs.start[i] |
1121 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1122 | search_regs.end[i] | |
1123 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1124 | } |
a3668d92 | 1125 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1126 | /* Set pos to the new position. */ |
1127 | pos = search_regs.start[0]; | |
1128 | } | |
1129 | else | |
1130 | { | |
1131 | immediate_quit = 0; | |
1132 | return (n); | |
1133 | } | |
1134 | n++; | |
1135 | } | |
1136 | while (n > 0) | |
1137 | { | |
42db823b | 1138 | int val; |
487282dc | 1139 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1140 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1141 | &search_regs, | |
1142 | lim_byte - BEGV_BYTE); | |
ca1d1d23 | 1143 | if (val == -2) |
b6d6a51c KH |
1144 | { |
1145 | matcher_overflow (); | |
1146 | } | |
ca1d1d23 JB |
1147 | if (val >= 0) |
1148 | { | |
26aff150 | 1149 | pos_byte = search_regs.end[0] + BEGV_BYTE; |
4746118a | 1150 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1151 | if (search_regs.start[i] >= 0) |
1152 | { | |
fa8ed3e0 RS |
1153 | search_regs.start[i] |
1154 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1155 | search_regs.end[i] | |
1156 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1157 | } |
a3668d92 | 1158 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1159 | pos = search_regs.end[0]; |
1160 | } | |
1161 | else | |
1162 | { | |
1163 | immediate_quit = 0; | |
1164 | return (0 - n); | |
1165 | } | |
1166 | n--; | |
1167 | } | |
1168 | immediate_quit = 0; | |
1169 | return (pos); | |
1170 | } | |
1171 | else /* non-RE case */ | |
1172 | { | |
facdc750 RS |
1173 | unsigned char *raw_pattern, *pat; |
1174 | int raw_pattern_size; | |
1175 | int raw_pattern_size_byte; | |
1176 | unsigned char *patbuf; | |
1177 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); | |
a967ed62 | 1178 | unsigned char *base_pat; |
fed91c38 KH |
1179 | /* Set to positive if we find a non-ASCII char that need |
1180 | translation. Otherwise set to zero later. */ | |
1181 | int charset_base = -1; | |
040272ce | 1182 | int boyer_moore_ok = 1; |
facdc750 RS |
1183 | |
1184 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1185 | We must convert PATTERN to match that, or we will not really | |
1186 | find things right. */ | |
1187 | ||
1188 | if (multibyte == STRING_MULTIBYTE (string)) | |
1189 | { | |
d5db4077 KR |
1190 | raw_pattern = (unsigned char *) SDATA (string); |
1191 | raw_pattern_size = SCHARS (string); | |
1192 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1193 | } |
1194 | else if (multibyte) | |
1195 | { | |
d5db4077 | 1196 | raw_pattern_size = SCHARS (string); |
facdc750 | 1197 | raw_pattern_size_byte |
d5db4077 | 1198 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1199 | raw_pattern_size); |
7276d3d8 | 1200 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1201 | copy_text (SDATA (string), raw_pattern, |
1202 | SCHARS (string), 0, 1); | |
facdc750 RS |
1203 | } |
1204 | else | |
1205 | { | |
1206 | /* Converting multibyte to single-byte. | |
1207 | ||
1208 | ??? Perhaps this conversion should be done in a special way | |
1209 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1210 | so that only the multibyte chars which really correspond to | |
1211 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1212 | raw_pattern_size = SCHARS (string); |
1213 | raw_pattern_size_byte = SCHARS (string); | |
7276d3d8 | 1214 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
1215 | copy_text (SDATA (string), raw_pattern, |
1216 | SBYTES (string), 1, 0); | |
facdc750 RS |
1217 | } |
1218 | ||
1219 | /* Copy and optionally translate the pattern. */ | |
1220 | len = raw_pattern_size; | |
1221 | len_byte = raw_pattern_size_byte; | |
1222 | patbuf = (unsigned char *) alloca (len_byte); | |
1223 | pat = patbuf; | |
1224 | base_pat = raw_pattern; | |
1225 | if (multibyte) | |
1226 | { | |
0190922f KH |
1227 | /* Fill patbuf by translated characters in STRING while |
1228 | checking if we can use boyer-moore search. If TRT is | |
1229 | non-nil, we can use boyer-moore search only if TRT can be | |
1230 | represented by the byte array of 256 elements. For that, | |
1231 | all non-ASCII case-equivalents of all case-senstive | |
1232 | characters in STRING must belong to the same charset and | |
1233 | row. */ | |
1234 | ||
facdc750 RS |
1235 | while (--len >= 0) |
1236 | { | |
0190922f | 1237 | unsigned char str_base[MAX_MULTIBYTE_LENGTH], *str; |
aff2ce94 | 1238 | int c, translated, inverse; |
facdc750 RS |
1239 | int in_charlen, charlen; |
1240 | ||
1241 | /* If we got here and the RE flag is set, it's because we're | |
1242 | dealing with a regexp known to be trivial, so the backslash | |
1243 | just quotes the next character. */ | |
1244 | if (RE && *base_pat == '\\') | |
1245 | { | |
1246 | len--; | |
0190922f | 1247 | raw_pattern_size--; |
facdc750 RS |
1248 | len_byte--; |
1249 | base_pat++; | |
1250 | } | |
1251 | ||
1252 | c = STRING_CHAR_AND_LENGTH (base_pat, len_byte, in_charlen); | |
040272ce | 1253 | |
0190922f | 1254 | if (NILP (trt)) |
facdc750 | 1255 | { |
0190922f KH |
1256 | str = base_pat; |
1257 | charlen = in_charlen; | |
facdc750 | 1258 | } |
0190922f | 1259 | else |
facdc750 | 1260 | { |
0190922f KH |
1261 | /* Translate the character. */ |
1262 | TRANSLATE (translated, trt, c); | |
1263 | charlen = CHAR_STRING (translated, str_base); | |
1264 | str = str_base; | |
1265 | ||
1266 | /* Check if C has any other case-equivalents. */ | |
1267 | TRANSLATE (inverse, inverse_trt, c); | |
1268 | /* If so, check if we can use boyer-moore. */ | |
1269 | if (c != inverse && boyer_moore_ok) | |
1270 | { | |
1271 | /* Check if all equivalents belong to the same | |
1272 | charset & row. Note that the check of C | |
1273 | itself is done by the last iteration. Note | |
1274 | also that we don't have to check ASCII | |
1275 | characters because boyer-moore search can | |
1276 | always handle their translation. */ | |
1277 | while (1) | |
1278 | { | |
fed91c38 | 1279 | if (ASCII_BYTE_P (inverse)) |
0190922f | 1280 | { |
fed91c38 | 1281 | if (charset_base > 0) |
0190922f KH |
1282 | { |
1283 | boyer_moore_ok = 0; | |
1284 | break; | |
1285 | } | |
fed91c38 KH |
1286 | charset_base = 0; |
1287 | } | |
1288 | else if (SINGLE_BYTE_CHAR_P (inverse)) | |
1289 | { | |
1290 | /* Boyer-moore search can't handle a | |
1291 | translation of an eight-bit | |
1292 | character. */ | |
1293 | boyer_moore_ok = 0; | |
1294 | break; | |
1295 | } | |
1296 | else if (charset_base < 0) | |
1297 | charset_base = inverse & ~CHAR_FIELD3_MASK; | |
1298 | else if ((inverse & ~CHAR_FIELD3_MASK) | |
1299 | != charset_base) | |
1300 | { | |
1301 | boyer_moore_ok = 0; | |
1302 | break; | |
0190922f KH |
1303 | } |
1304 | if (c == inverse) | |
1305 | break; | |
1306 | TRANSLATE (inverse, inverse_trt, inverse); | |
1307 | } | |
1308 | } | |
aff2ce94 | 1309 | } |
fed91c38 KH |
1310 | if (charset_base < 0) |
1311 | charset_base = 0; | |
facdc750 RS |
1312 | |
1313 | /* Store this character into the translated pattern. */ | |
1314 | bcopy (str, pat, charlen); | |
1315 | pat += charlen; | |
1316 | base_pat += in_charlen; | |
1317 | len_byte -= in_charlen; | |
1318 | } | |
1319 | } | |
1320 | else | |
1321 | { | |
040272ce KH |
1322 | /* Unibyte buffer. */ |
1323 | charset_base = 0; | |
facdc750 RS |
1324 | while (--len >= 0) |
1325 | { | |
040272ce | 1326 | int c, translated; |
facdc750 RS |
1327 | |
1328 | /* If we got here and the RE flag is set, it's because we're | |
1329 | dealing with a regexp known to be trivial, so the backslash | |
1330 | just quotes the next character. */ | |
1331 | if (RE && *base_pat == '\\') | |
1332 | { | |
1333 | len--; | |
0190922f | 1334 | raw_pattern_size--; |
facdc750 RS |
1335 | base_pat++; |
1336 | } | |
1337 | c = *base_pat++; | |
aff2ce94 | 1338 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1339 | *pat++ = translated; |
1340 | } | |
1341 | } | |
1342 | ||
1343 | len_byte = pat - patbuf; | |
1344 | len = raw_pattern_size; | |
1345 | pat = base_pat = patbuf; | |
1346 | ||
040272ce | 1347 | if (boyer_moore_ok) |
facdc750 | 1348 | return boyer_moore (n, pat, len, len_byte, trt, inverse_trt, |
aff2ce94 RS |
1349 | pos, pos_byte, lim, lim_byte, |
1350 | charset_base); | |
facdc750 RS |
1351 | else |
1352 | return simple_search (n, pat, len, len_byte, trt, | |
1353 | pos, pos_byte, lim, lim_byte); | |
1354 | } | |
1355 | } | |
1356 | \f | |
1357 | /* Do a simple string search N times for the string PAT, | |
1358 | whose length is LEN/LEN_BYTE, | |
1359 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1360 | TRT is the translation table. | |
f8bd51c4 | 1361 | |
facdc750 RS |
1362 | Return the character position where the match is found. |
1363 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1364 | |
facdc750 RS |
1365 | This kind of search works regardless of what is in PAT and |
1366 | regardless of what is in TRT. It is used in cases where | |
1367 | boyer_moore cannot work. */ | |
1368 | ||
1369 | static int | |
1370 | simple_search (n, pat, len, len_byte, trt, pos, pos_byte, lim, lim_byte) | |
1371 | int n; | |
1372 | unsigned char *pat; | |
1373 | int len, len_byte; | |
1374 | Lisp_Object trt; | |
1375 | int pos, pos_byte; | |
1376 | int lim, lim_byte; | |
1377 | { | |
1378 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); | |
ab228c24 | 1379 | int forward = n > 0; |
facdc750 RS |
1380 | |
1381 | if (lim > pos && multibyte) | |
1382 | while (n > 0) | |
1383 | { | |
1384 | while (1) | |
f8bd51c4 | 1385 | { |
facdc750 RS |
1386 | /* Try matching at position POS. */ |
1387 | int this_pos = pos; | |
1388 | int this_pos_byte = pos_byte; | |
1389 | int this_len = len; | |
1390 | int this_len_byte = len_byte; | |
1391 | unsigned char *p = pat; | |
1392 | if (pos + len > lim) | |
1393 | goto stop; | |
1394 | ||
1395 | while (this_len > 0) | |
1396 | { | |
1397 | int charlen, buf_charlen; | |
ab228c24 | 1398 | int pat_ch, buf_ch; |
facdc750 | 1399 | |
ab228c24 | 1400 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1401 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1402 | ZV_BYTE - this_pos_byte, | |
1403 | buf_charlen); | |
aff2ce94 | 1404 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1405 | |
1406 | if (buf_ch != pat_ch) | |
1407 | break; | |
ab228c24 RS |
1408 | |
1409 | this_len_byte -= charlen; | |
1410 | this_len--; | |
1411 | p += charlen; | |
1412 | ||
1413 | this_pos_byte += buf_charlen; | |
1414 | this_pos++; | |
facdc750 RS |
1415 | } |
1416 | ||
1417 | if (this_len == 0) | |
1418 | { | |
1419 | pos += len; | |
1420 | pos_byte += len_byte; | |
1421 | break; | |
1422 | } | |
1423 | ||
1424 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1425 | } |
facdc750 RS |
1426 | |
1427 | n--; | |
1428 | } | |
1429 | else if (lim > pos) | |
1430 | while (n > 0) | |
1431 | { | |
1432 | while (1) | |
f8bd51c4 | 1433 | { |
facdc750 RS |
1434 | /* Try matching at position POS. */ |
1435 | int this_pos = pos; | |
1436 | int this_len = len; | |
1437 | unsigned char *p = pat; | |
1438 | ||
1439 | if (pos + len > lim) | |
1440 | goto stop; | |
1441 | ||
1442 | while (this_len > 0) | |
1443 | { | |
1444 | int pat_ch = *p++; | |
1445 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1446 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1447 | |
1448 | if (buf_ch != pat_ch) | |
1449 | break; | |
ab228c24 RS |
1450 | |
1451 | this_len--; | |
1452 | this_pos++; | |
facdc750 RS |
1453 | } |
1454 | ||
1455 | if (this_len == 0) | |
1456 | { | |
1457 | pos += len; | |
1458 | break; | |
1459 | } | |
1460 | ||
1461 | pos++; | |
f8bd51c4 | 1462 | } |
facdc750 RS |
1463 | |
1464 | n--; | |
1465 | } | |
1466 | /* Backwards search. */ | |
1467 | else if (lim < pos && multibyte) | |
1468 | while (n < 0) | |
1469 | { | |
1470 | while (1) | |
f8bd51c4 | 1471 | { |
facdc750 RS |
1472 | /* Try matching at position POS. */ |
1473 | int this_pos = pos - len; | |
1474 | int this_pos_byte = pos_byte - len_byte; | |
1475 | int this_len = len; | |
1476 | int this_len_byte = len_byte; | |
1477 | unsigned char *p = pat; | |
1478 | ||
1479 | if (pos - len < lim) | |
1480 | goto stop; | |
1481 | ||
1482 | while (this_len > 0) | |
1483 | { | |
1484 | int charlen, buf_charlen; | |
ab228c24 | 1485 | int pat_ch, buf_ch; |
facdc750 | 1486 | |
ab228c24 | 1487 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1488 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1489 | ZV_BYTE - this_pos_byte, | |
1490 | buf_charlen); | |
aff2ce94 | 1491 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1492 | |
1493 | if (buf_ch != pat_ch) | |
1494 | break; | |
ab228c24 RS |
1495 | |
1496 | this_len_byte -= charlen; | |
1497 | this_len--; | |
1498 | p += charlen; | |
1499 | this_pos_byte += buf_charlen; | |
1500 | this_pos++; | |
facdc750 RS |
1501 | } |
1502 | ||
1503 | if (this_len == 0) | |
1504 | { | |
1505 | pos -= len; | |
1506 | pos_byte -= len_byte; | |
1507 | break; | |
1508 | } | |
1509 | ||
1510 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1511 | } |
1512 | ||
facdc750 RS |
1513 | n++; |
1514 | } | |
1515 | else if (lim < pos) | |
1516 | while (n < 0) | |
1517 | { | |
1518 | while (1) | |
b6d6a51c | 1519 | { |
facdc750 RS |
1520 | /* Try matching at position POS. */ |
1521 | int this_pos = pos - len; | |
1522 | int this_len = len; | |
1523 | unsigned char *p = pat; | |
1524 | ||
1525 | if (pos - len < lim) | |
1526 | goto stop; | |
1527 | ||
1528 | while (this_len > 0) | |
1529 | { | |
1530 | int pat_ch = *p++; | |
1531 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1532 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1533 | |
1534 | if (buf_ch != pat_ch) | |
1535 | break; | |
ab228c24 RS |
1536 | this_len--; |
1537 | this_pos++; | |
facdc750 RS |
1538 | } |
1539 | ||
1540 | if (this_len == 0) | |
b6d6a51c | 1541 | { |
facdc750 RS |
1542 | pos -= len; |
1543 | break; | |
b6d6a51c | 1544 | } |
facdc750 RS |
1545 | |
1546 | pos--; | |
b6d6a51c | 1547 | } |
facdc750 RS |
1548 | |
1549 | n++; | |
b6d6a51c | 1550 | } |
facdc750 RS |
1551 | |
1552 | stop: | |
1553 | if (n == 0) | |
aff2ce94 | 1554 | { |
ab228c24 RS |
1555 | if (forward) |
1556 | set_search_regs ((multibyte ? pos_byte : pos) - len_byte, len_byte); | |
1557 | else | |
1558 | set_search_regs (multibyte ? pos_byte : pos, len_byte); | |
aff2ce94 RS |
1559 | |
1560 | return pos; | |
1561 | } | |
facdc750 RS |
1562 | else if (n > 0) |
1563 | return -n; | |
1564 | else | |
1565 | return n; | |
1566 | } | |
1567 | \f | |
0190922f | 1568 | /* Do Boyer-Moore search N times for the string BASE_PAT, |
facdc750 RS |
1569 | whose length is LEN/LEN_BYTE, |
1570 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1571 | DIRECTION says which direction we search in. | |
1572 | TRT and INVERSE_TRT are translation tables. | |
0190922f | 1573 | Characters in PAT are already translated by TRT. |
facdc750 | 1574 | |
0190922f KH |
1575 | This kind of search works if all the characters in BASE_PAT that |
1576 | have nontrivial translation are the same aside from the last byte. | |
1577 | This makes it possible to translate just the last byte of a | |
1578 | character, and do so after just a simple test of the context. | |
1579 | CHARSET_BASE is nonzero iff there is such a non-ASCII character. | |
facdc750 RS |
1580 | |
1581 | If that criterion is not satisfied, do not call this function. */ | |
1582 | ||
1583 | static int | |
1584 | boyer_moore (n, base_pat, len, len_byte, trt, inverse_trt, | |
aff2ce94 | 1585 | pos, pos_byte, lim, lim_byte, charset_base) |
facdc750 RS |
1586 | int n; |
1587 | unsigned char *base_pat; | |
1588 | int len, len_byte; | |
1589 | Lisp_Object trt; | |
1590 | Lisp_Object inverse_trt; | |
1591 | int pos, pos_byte; | |
1592 | int lim, lim_byte; | |
aff2ce94 | 1593 | int charset_base; |
facdc750 RS |
1594 | { |
1595 | int direction = ((n > 0) ? 1 : -1); | |
1596 | register int dirlen; | |
a968f437 | 1597 | int infinity, limit, stride_for_teases = 0; |
facdc750 RS |
1598 | register int *BM_tab; |
1599 | int *BM_tab_base; | |
177c0ea7 | 1600 | register unsigned char *cursor, *p_limit; |
facdc750 | 1601 | register int i, j; |
cb6792d2 | 1602 | unsigned char *pat, *pat_end; |
facdc750 RS |
1603 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); |
1604 | ||
1605 | unsigned char simple_translate[0400]; | |
0190922f KH |
1606 | /* These are set to the preceding bytes of a byte to be translated |
1607 | if charset_base is nonzero. As the maximum byte length of a | |
1608 | multibyte character is 4, we have to check at most three previous | |
1609 | bytes. */ | |
1610 | int translate_prev_byte1 = 0; | |
1611 | int translate_prev_byte2 = 0; | |
1612 | int translate_prev_byte3 = 0; | |
facdc750 RS |
1613 | |
1614 | #ifdef C_ALLOCA | |
1615 | int BM_tab_space[0400]; | |
1616 | BM_tab = &BM_tab_space[0]; | |
1617 | #else | |
1618 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
1619 | #endif | |
1620 | /* The general approach is that we are going to maintain that we know */ | |
1621 | /* the first (closest to the present position, in whatever direction */ | |
1622 | /* we're searching) character that could possibly be the last */ | |
1623 | /* (furthest from present position) character of a valid match. We */ | |
1624 | /* advance the state of our knowledge by looking at that character */ | |
1625 | /* and seeing whether it indeed matches the last character of the */ | |
1626 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1627 | /* move our pointer (to putative last characters) as far as is */ | |
1628 | /* logically possible. This amount of movement, which I call a */ | |
1629 | /* stride, will be the length of the pattern if the actual character */ | |
1630 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1631 | /* from the last occurrence of that character to the end of the */ | |
1632 | /* pattern. */ | |
1633 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1634 | /* characters that match the last character. This allows use of only */ | |
1635 | /* a single test, a test for having gone past the end of the */ | |
1636 | /* permissible match region, to test for both possible matches (when */ | |
1637 | /* the stride goes past the end immediately) and failure to */ | |
177c0ea7 | 1638 | /* match (where you get nudged past the end one stride at a time). */ |
facdc750 RS |
1639 | |
1640 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1641 | /* is determined only by the last character of the putative match. */ | |
1642 | /* If that character does not match, we will stride the proper */ | |
1643 | /* distance to propose a match that superimposes it on the last */ | |
1644 | /* instance of a character that matches it (per trt), or misses */ | |
177c0ea7 | 1645 | /* it entirely if there is none. */ |
facdc750 RS |
1646 | |
1647 | dirlen = len_byte * direction; | |
1648 | infinity = dirlen - (lim_byte + pos_byte + len_byte + len_byte) * direction; | |
cb6792d2 RS |
1649 | |
1650 | /* Record position after the end of the pattern. */ | |
1651 | pat_end = base_pat + len_byte; | |
1652 | /* BASE_PAT points to a character that we start scanning from. | |
1653 | It is the first character in a forward search, | |
1654 | the last character in a backward search. */ | |
facdc750 | 1655 | if (direction < 0) |
cb6792d2 RS |
1656 | base_pat = pat_end - 1; |
1657 | ||
facdc750 RS |
1658 | BM_tab_base = BM_tab; |
1659 | BM_tab += 0400; | |
1660 | j = dirlen; /* to get it in a register */ | |
1661 | /* A character that does not appear in the pattern induces a */ | |
1662 | /* stride equal to the pattern length. */ | |
1663 | while (BM_tab_base != BM_tab) | |
1664 | { | |
1665 | *--BM_tab = j; | |
1666 | *--BM_tab = j; | |
1667 | *--BM_tab = j; | |
1668 | *--BM_tab = j; | |
1669 | } | |
1670 | ||
1671 | /* We use this for translation, instead of TRT itself. | |
1672 | We fill this in to handle the characters that actually | |
1673 | occur in the pattern. Others don't matter anyway! */ | |
1674 | bzero (simple_translate, sizeof simple_translate); | |
1675 | for (i = 0; i < 0400; i++) | |
1676 | simple_translate[i] = i; | |
1677 | ||
0190922f KH |
1678 | if (charset_base) |
1679 | { | |
1680 | /* Setup translate_prev_byte1/2/3 from CHARSET_BASE. Only a | |
1681 | byte following them are the target of translation. */ | |
1682 | int sample_char = charset_base | 0x20; | |
1683 | unsigned char str[MAX_MULTIBYTE_LENGTH]; | |
1684 | int len = CHAR_STRING (sample_char, str); | |
1685 | ||
1686 | translate_prev_byte1 = str[len - 2]; | |
1687 | if (len > 2) | |
1688 | { | |
1689 | translate_prev_byte2 = str[len - 3]; | |
1690 | if (len > 3) | |
1691 | translate_prev_byte3 = str[len - 4]; | |
1692 | } | |
1693 | } | |
1694 | ||
facdc750 RS |
1695 | i = 0; |
1696 | while (i != infinity) | |
1697 | { | |
cb6792d2 | 1698 | unsigned char *ptr = base_pat + i; |
facdc750 RS |
1699 | i += direction; |
1700 | if (i == dirlen) | |
1701 | i = infinity; | |
1702 | if (! NILP (trt)) | |
ca1d1d23 | 1703 | { |
620a5178 KH |
1704 | /* If the byte currently looking at is the last of a |
1705 | character to check case-equivalents, set CH to that | |
1706 | character. An ASCII character and a non-ASCII character | |
1707 | matching with CHARSET_BASE are to be checked. */ | |
0190922f KH |
1708 | int ch = -1; |
1709 | ||
1710 | if (ASCII_BYTE_P (*ptr) || ! multibyte) | |
1711 | ch = *ptr; | |
620a5178 | 1712 | else if (charset_base |
86dc6ccb | 1713 | && ((pat_end - ptr) == 1 || CHAR_HEAD_P (ptr[1]))) |
ca1d1d23 | 1714 | { |
620a5178 KH |
1715 | unsigned char *charstart = ptr - 1; |
1716 | ||
1717 | while (! (CHAR_HEAD_P (*charstart))) | |
1718 | charstart--; | |
1719 | ch = STRING_CHAR (charstart, ptr - charstart + 1); | |
0190922f KH |
1720 | if (charset_base != (ch & ~CHAR_FIELD3_MASK)) |
1721 | ch = -1; | |
ca1d1d23 | 1722 | } |
facdc750 | 1723 | |
6786a107 | 1724 | if (ch >= 0400) |
620a5178 KH |
1725 | j = ((unsigned char) ch) | 0200; |
1726 | else | |
1727 | j = *ptr; | |
1728 | ||
facdc750 RS |
1729 | if (i == infinity) |
1730 | stride_for_teases = BM_tab[j]; | |
ab228c24 | 1731 | |
facdc750 RS |
1732 | BM_tab[j] = dirlen - i; |
1733 | /* A translation table is accompanied by its inverse -- see */ | |
177c0ea7 | 1734 | /* comment following downcase_table for details */ |
0190922f | 1735 | if (ch >= 0) |
ab228c24 RS |
1736 | { |
1737 | int starting_ch = ch; | |
620a5178 | 1738 | int starting_j = j; |
0190922f | 1739 | |
ab228c24 RS |
1740 | while (1) |
1741 | { | |
1742 | TRANSLATE (ch, inverse_trt, ch); | |
6786a107 | 1743 | if (ch >= 0400) |
ab228c24 RS |
1744 | j = ((unsigned char) ch) | 0200; |
1745 | else | |
1746 | j = (unsigned char) ch; | |
1747 | ||
1748 | /* For all the characters that map into CH, | |
1749 | set up simple_translate to map the last byte | |
1750 | into STARTING_J. */ | |
1751 | simple_translate[j] = starting_j; | |
1752 | if (ch == starting_ch) | |
1753 | break; | |
1754 | BM_tab[j] = dirlen - i; | |
1755 | } | |
1756 | } | |
facdc750 RS |
1757 | } |
1758 | else | |
1759 | { | |
1760 | j = *ptr; | |
1761 | ||
1762 | if (i == infinity) | |
1763 | stride_for_teases = BM_tab[j]; | |
1764 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1765 | } |
facdc750 RS |
1766 | /* stride_for_teases tells how much to stride if we get a */ |
1767 | /* match on the far character but are subsequently */ | |
1768 | /* disappointed, by recording what the stride would have been */ | |
1769 | /* for that character if the last character had been */ | |
1770 | /* different. */ | |
1771 | } | |
1772 | infinity = dirlen - infinity; | |
1773 | pos_byte += dirlen - ((direction > 0) ? direction : 0); | |
1774 | /* loop invariant - POS_BYTE points at where last char (first | |
1775 | char if reverse) of pattern would align in a possible match. */ | |
1776 | while (n != 0) | |
1777 | { | |
1778 | int tail_end; | |
1779 | unsigned char *tail_end_ptr; | |
1780 | ||
1781 | /* It's been reported that some (broken) compiler thinks that | |
1782 | Boolean expressions in an arithmetic context are unsigned. | |
1783 | Using an explicit ?1:0 prevents this. */ | |
1784 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1785 | < 0) | |
1786 | return (n * (0 - direction)); | |
1787 | /* First we do the part we can by pointers (maybe nothing) */ | |
1788 | QUIT; | |
1789 | pat = base_pat; | |
1790 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1791 | if (direction > 0) |
1792 | { | |
1793 | limit = BUFFER_CEILING_OF (limit); | |
1794 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1795 | can take on without hitting edge of buffer or the gap. */ | |
1796 | limit = min (limit, pos_byte + 20000); | |
1797 | limit = min (limit, lim_byte - 1); | |
1798 | } | |
1799 | else | |
1800 | { | |
1801 | limit = BUFFER_FLOOR_OF (limit); | |
1802 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1803 | can take on without hitting edge of buffer or the gap. */ | |
1804 | limit = max (limit, pos_byte - 20000); | |
1805 | limit = max (limit, lim_byte); | |
1806 | } | |
facdc750 RS |
1807 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1808 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1809 | ||
1810 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1811 | { |
facdc750 RS |
1812 | unsigned char *p2; |
1813 | ||
1814 | p_limit = BYTE_POS_ADDR (limit); | |
1815 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
1816 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ | |
1817 | while (1) /* use one cursor setting as long as i can */ | |
ca1d1d23 | 1818 | { |
facdc750 | 1819 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1820 | { |
facdc750 RS |
1821 | /* Use signed comparison if appropriate |
1822 | to make cursor+infinity sure to be > p_limit. | |
1823 | Assuming that the buffer lies in a range of addresses | |
1824 | that are all "positive" (as ints) or all "negative", | |
1825 | either kind of comparison will work as long | |
1826 | as we don't step by infinity. So pick the kind | |
1827 | that works when we do step by infinity. */ | |
1828 | if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit) | |
1829 | while ((EMACS_INT) cursor <= (EMACS_INT) p_limit) | |
1830 | cursor += BM_tab[*cursor]; | |
ca1d1d23 | 1831 | else |
facdc750 RS |
1832 | while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit) |
1833 | cursor += BM_tab[*cursor]; | |
1834 | } | |
1835 | else | |
1836 | { | |
1837 | if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit) | |
1838 | while ((EMACS_INT) cursor >= (EMACS_INT) p_limit) | |
1839 | cursor += BM_tab[*cursor]; | |
1840 | else | |
1841 | while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit) | |
1842 | cursor += BM_tab[*cursor]; | |
1843 | } | |
ca1d1d23 | 1844 | /* If you are here, cursor is beyond the end of the searched region. */ |
facdc750 RS |
1845 | /* This can happen if you match on the far character of the pattern, */ |
1846 | /* because the "stride" of that character is infinity, a number able */ | |
1847 | /* to throw you well beyond the end of the search. It can also */ | |
1848 | /* happen if you fail to match within the permitted region and would */ | |
1849 | /* otherwise try a character beyond that region */ | |
1850 | if ((cursor - p_limit) * direction <= len_byte) | |
1851 | break; /* a small overrun is genuine */ | |
1852 | cursor -= infinity; /* large overrun = hit */ | |
1853 | i = dirlen - direction; | |
1854 | if (! NILP (trt)) | |
1855 | { | |
1856 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1857 | { |
facdc750 RS |
1858 | int ch; |
1859 | cursor -= direction; | |
1860 | /* Translate only the last byte of a character. */ | |
1861 | if (! multibyte | |
1862 | || ((cursor == tail_end_ptr | |
1863 | || CHAR_HEAD_P (cursor[1])) | |
1864 | && (CHAR_HEAD_P (cursor[0]) | |
0190922f KH |
1865 | /* Check if this is the last byte of |
1866 | a translable character. */ | |
1867 | || (translate_prev_byte1 == cursor[-1] | |
1868 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1869 | || (translate_prev_byte2 == cursor[-2] | |
1870 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1871 | || (translate_prev_byte3 == cursor[-3])))))))) | |
facdc750 RS |
1872 | ch = simple_translate[*cursor]; |
1873 | else | |
1874 | ch = *cursor; | |
1875 | if (pat[i] != ch) | |
1876 | break; | |
ca1d1d23 | 1877 | } |
facdc750 RS |
1878 | } |
1879 | else | |
1880 | { | |
1881 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1882 | { |
facdc750 RS |
1883 | cursor -= direction; |
1884 | if (pat[i] != *cursor) | |
1885 | break; | |
ca1d1d23 | 1886 | } |
facdc750 RS |
1887 | } |
1888 | cursor += dirlen - i - direction; /* fix cursor */ | |
1889 | if (i + direction == 0) | |
1890 | { | |
1891 | int position; | |
0c8533c6 | 1892 | |
facdc750 | 1893 | cursor -= direction; |
1113d9db | 1894 | |
facdc750 RS |
1895 | position = pos_byte + cursor - p2 + ((direction > 0) |
1896 | ? 1 - len_byte : 0); | |
1897 | set_search_regs (position, len_byte); | |
ca325161 | 1898 | |
facdc750 RS |
1899 | if ((n -= direction) != 0) |
1900 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1901 | else |
facdc750 RS |
1902 | return ((direction > 0) |
1903 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1904 | } |
facdc750 RS |
1905 | else |
1906 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1907 | } |
facdc750 RS |
1908 | pos_byte += cursor - p2; |
1909 | } | |
1910 | else | |
1911 | /* Now we'll pick up a clump that has to be done the hard */ | |
1912 | /* way because it covers a discontinuity */ | |
1913 | { | |
1914 | limit = ((direction > 0) | |
1915 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1916 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1917 | limit = ((direction > 0) | |
1918 | ? min (limit + len_byte, lim_byte - 1) | |
1919 | : max (limit - len_byte, lim_byte)); | |
1920 | /* LIMIT is now the last value POS_BYTE can have | |
1921 | and still be valid for a possible match. */ | |
1922 | while (1) | |
ca1d1d23 | 1923 | { |
facdc750 RS |
1924 | /* This loop can be coded for space rather than */ |
1925 | /* speed because it will usually run only once. */ | |
1926 | /* (the reach is at most len + 21, and typically */ | |
177c0ea7 | 1927 | /* does not exceed len) */ |
facdc750 RS |
1928 | while ((limit - pos_byte) * direction >= 0) |
1929 | pos_byte += BM_tab[FETCH_BYTE (pos_byte)]; | |
1930 | /* now run the same tests to distinguish going off the */ | |
1931 | /* end, a match or a phony match. */ | |
1932 | if ((pos_byte - limit) * direction <= len_byte) | |
1933 | break; /* ran off the end */ | |
1934 | /* Found what might be a match. | |
1935 | Set POS_BYTE back to last (first if reverse) pos. */ | |
1936 | pos_byte -= infinity; | |
1937 | i = dirlen - direction; | |
1938 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1939 | { |
facdc750 RS |
1940 | int ch; |
1941 | unsigned char *ptr; | |
1942 | pos_byte -= direction; | |
1943 | ptr = BYTE_POS_ADDR (pos_byte); | |
1944 | /* Translate only the last byte of a character. */ | |
1945 | if (! multibyte | |
1946 | || ((ptr == tail_end_ptr | |
1947 | || CHAR_HEAD_P (ptr[1])) | |
1948 | && (CHAR_HEAD_P (ptr[0]) | |
0190922f KH |
1949 | /* Check if this is the last byte of a |
1950 | translable character. */ | |
1951 | || (translate_prev_byte1 == ptr[-1] | |
1952 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1953 | || (translate_prev_byte2 == ptr[-2] | |
1954 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1955 | || translate_prev_byte3 == ptr[-3]))))))) | |
facdc750 RS |
1956 | ch = simple_translate[*ptr]; |
1957 | else | |
1958 | ch = *ptr; | |
1959 | if (pat[i] != ch) | |
1960 | break; | |
1961 | } | |
1962 | /* Above loop has moved POS_BYTE part or all the way | |
1963 | back to the first pos (last pos if reverse). | |
1964 | Set it once again at the last (first if reverse) char. */ | |
1965 | pos_byte += dirlen - i- direction; | |
1966 | if (i + direction == 0) | |
1967 | { | |
1968 | int position; | |
1969 | pos_byte -= direction; | |
1113d9db | 1970 | |
facdc750 | 1971 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
0c8533c6 | 1972 | |
facdc750 | 1973 | set_search_regs (position, len_byte); |
ca325161 | 1974 | |
facdc750 RS |
1975 | if ((n -= direction) != 0) |
1976 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 1977 | else |
facdc750 RS |
1978 | return ((direction > 0) |
1979 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1980 | } |
facdc750 RS |
1981 | else |
1982 | pos_byte += stride_for_teases; | |
1983 | } | |
1984 | } | |
1985 | /* We have done one clump. Can we continue? */ | |
1986 | if ((lim_byte - pos_byte) * direction < 0) | |
1987 | return ((0 - n) * direction); | |
ca1d1d23 | 1988 | } |
facdc750 | 1989 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 1990 | } |
ca325161 | 1991 | |
fa8ed3e0 | 1992 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
1993 | for the overall match just found in the current buffer. |
1994 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
1995 | |
1996 | static void | |
fa8ed3e0 RS |
1997 | set_search_regs (beg_byte, nbytes) |
1998 | int beg_byte, nbytes; | |
ca325161 | 1999 | { |
a7e4cdde RS |
2000 | int i; |
2001 | ||
ca325161 RS |
2002 | /* Make sure we have registers in which to store |
2003 | the match position. */ | |
2004 | if (search_regs.num_regs == 0) | |
2005 | { | |
2d4a771a RS |
2006 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
2007 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 2008 | search_regs.num_regs = 2; |
ca325161 RS |
2009 | } |
2010 | ||
a7e4cdde RS |
2011 | /* Clear out the other registers. */ |
2012 | for (i = 1; i < search_regs.num_regs; i++) | |
2013 | { | |
2014 | search_regs.start[i] = -1; | |
2015 | search_regs.end[i] = -1; | |
2016 | } | |
2017 | ||
fa8ed3e0 RS |
2018 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
2019 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 2020 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 2021 | } |
ca1d1d23 JB |
2022 | \f |
2023 | /* Given a string of words separated by word delimiters, | |
2024 | compute a regexp that matches those exact words | |
2025 | separated by arbitrary punctuation. */ | |
2026 | ||
2027 | static Lisp_Object | |
2028 | wordify (string) | |
2029 | Lisp_Object string; | |
2030 | { | |
2031 | register unsigned char *p, *o; | |
0c8533c6 | 2032 | register int i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 2033 | Lisp_Object val; |
0c8533c6 RS |
2034 | int prev_c = 0; |
2035 | int adjust; | |
ca1d1d23 | 2036 | |
b7826503 | 2037 | CHECK_STRING (string); |
d5db4077 KR |
2038 | p = SDATA (string); |
2039 | len = SCHARS (string); | |
ca1d1d23 | 2040 | |
0c8533c6 RS |
2041 | for (i = 0, i_byte = 0; i < len; ) |
2042 | { | |
2043 | int c; | |
177c0ea7 | 2044 | |
eb99a8dd | 2045 | FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte); |
0c8533c6 RS |
2046 | |
2047 | if (SYNTAX (c) != Sword) | |
2048 | { | |
2049 | punct_count++; | |
2050 | if (i > 0 && SYNTAX (prev_c) == Sword) | |
2051 | word_count++; | |
2052 | } | |
ca1d1d23 | 2053 | |
0c8533c6 RS |
2054 | prev_c = c; |
2055 | } | |
2056 | ||
2057 | if (SYNTAX (prev_c) == Sword) | |
2058 | word_count++; | |
2059 | if (!word_count) | |
b07b65aa | 2060 | return empty_string; |
0c8533c6 RS |
2061 | |
2062 | adjust = - punct_count + 5 * (word_count - 1) + 4; | |
8a2df937 RS |
2063 | if (STRING_MULTIBYTE (string)) |
2064 | val = make_uninit_multibyte_string (len + adjust, | |
d5db4077 | 2065 | SBYTES (string) |
8a2df937 RS |
2066 | + adjust); |
2067 | else | |
2068 | val = make_uninit_string (len + adjust); | |
ca1d1d23 | 2069 | |
d5db4077 | 2070 | o = SDATA (val); |
ca1d1d23 JB |
2071 | *o++ = '\\'; |
2072 | *o++ = 'b'; | |
1e9582d4 | 2073 | prev_c = 0; |
ca1d1d23 | 2074 | |
1e9582d4 RS |
2075 | for (i = 0, i_byte = 0; i < len; ) |
2076 | { | |
2077 | int c; | |
2078 | int i_byte_orig = i_byte; | |
177c0ea7 | 2079 | |
eb99a8dd | 2080 | FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte); |
1e9582d4 RS |
2081 | |
2082 | if (SYNTAX (c) == Sword) | |
2083 | { | |
5d69fe10 | 2084 | bcopy (SDATA (string) + i_byte_orig, o, |
1e9582d4 RS |
2085 | i_byte - i_byte_orig); |
2086 | o += i_byte - i_byte_orig; | |
2087 | } | |
2088 | else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count) | |
2089 | { | |
2090 | *o++ = '\\'; | |
2091 | *o++ = 'W'; | |
2092 | *o++ = '\\'; | |
2093 | *o++ = 'W'; | |
2094 | *o++ = '*'; | |
2095 | } | |
2096 | ||
2097 | prev_c = c; | |
2098 | } | |
ca1d1d23 JB |
2099 | |
2100 | *o++ = '\\'; | |
2101 | *o++ = 'b'; | |
2102 | ||
2103 | return val; | |
2104 | } | |
2105 | \f | |
2106 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
8c1a1077 PJ |
2107 | "MSearch backward: ", |
2108 | doc: /* Search backward from point for STRING. | |
2109 | Set point to the beginning of the occurrence found, and return point. | |
2110 | An optional second argument bounds the search; it is a buffer position. | |
2111 | The match found must not extend before that position. | |
2112 | Optional third argument, if t, means if fail just return nil (no error). | |
2113 | If not nil and not t, position at limit of search and return nil. | |
2114 | Optional fourth argument is repeat count--search for successive occurrences. | |
2115 | ||
2116 | Search case-sensitivity is determined by the value of the variable | |
2117 | `case-fold-search', which see. | |
2118 | ||
2119 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2120 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2121 | Lisp_Object string, bound, noerror, count; |
2122 | { | |
b819a390 | 2123 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2124 | } |
2125 | ||
6af43974 | 2126 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2127 | doc: /* Search forward from point for STRING. |
2128 | Set point to the end of the occurrence found, and return point. | |
2129 | An optional second argument bounds the search; it is a buffer position. | |
2130 | The match found must not extend after that position. nil is equivalent | |
2131 | to (point-max). | |
2132 | Optional third argument, if t, means if fail just return nil (no error). | |
2133 | If not nil and not t, move to limit of search and return nil. | |
2134 | Optional fourth argument is repeat count--search for successive occurrences. | |
2135 | ||
2136 | Search case-sensitivity is determined by the value of the variable | |
2137 | `case-fold-search', which see. | |
2138 | ||
2139 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2140 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2141 | Lisp_Object string, bound, noerror, count; |
2142 | { | |
b819a390 | 2143 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2144 | } |
2145 | ||
2146 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
8c1a1077 PJ |
2147 | "sWord search backward: ", |
2148 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2149 | Set point to the beginning of the occurrence found, and return point. | |
2150 | An optional second argument bounds the search; it is a buffer position. | |
2151 | The match found must not extend before that position. | |
2152 | Optional third argument, if t, means if fail just return nil (no error). | |
2153 | If not nil and not t, move to limit of search and return nil. | |
2154 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2155 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2156 | Lisp_Object string, bound, noerror, count; |
2157 | { | |
b819a390 | 2158 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2159 | } |
2160 | ||
2161 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
8c1a1077 PJ |
2162 | "sWord search: ", |
2163 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2164 | Set point to the end of the occurrence found, and return point. | |
2165 | An optional second argument bounds the search; it is a buffer position. | |
2166 | The match found must not extend after that position. | |
2167 | Optional third argument, if t, means if fail just return nil (no error). | |
2168 | If not nil and not t, move to limit of search and return nil. | |
2169 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2170 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2171 | Lisp_Object string, bound, noerror, count; |
2172 | { | |
b819a390 | 2173 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
2174 | } |
2175 | ||
2176 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
8c1a1077 PJ |
2177 | "sRE search backward: ", |
2178 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2179 | Set point to the beginning of the match, and return point. | |
2180 | The match found is the one starting last in the buffer | |
2181 | and yet ending before the origin of the search. | |
2182 | An optional second argument bounds the search; it is a buffer position. | |
2183 | The match found must start at or after that position. | |
2184 | Optional third argument, if t, means if fail just return nil (no error). | |
2185 | If not nil and not t, move to limit of search and return nil. | |
2186 | Optional fourth argument is repeat count--search for successive occurrences. | |
2187 | See also the functions `match-beginning', `match-end', `match-string', | |
2188 | and `replace-match'. */) | |
2189 | (regexp, bound, noerror, count) | |
19c0a730 | 2190 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2191 | { |
b819a390 | 2192 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2193 | } |
2194 | ||
2195 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2196 | "sRE search: ", |
2197 | doc: /* Search forward from point for regular expression REGEXP. | |
2198 | Set point to the end of the occurrence found, and return point. | |
2199 | An optional second argument bounds the search; it is a buffer position. | |
2200 | The match found must not extend after that position. | |
2201 | Optional third argument, if t, means if fail just return nil (no error). | |
2202 | If not nil and not t, move to limit of search and return nil. | |
2203 | Optional fourth argument is repeat count--search for successive occurrences. | |
2204 | See also the functions `match-beginning', `match-end', `match-string', | |
2205 | and `replace-match'. */) | |
2206 | (regexp, bound, noerror, count) | |
19c0a730 | 2207 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2208 | { |
b819a390 RS |
2209 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2210 | } | |
2211 | ||
2212 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2213 | "sPosix search backward: ", |
2214 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2215 | Find the longest match in accord with Posix regular expression rules. | |
2216 | Set point to the beginning of the match, and return point. | |
2217 | The match found is the one starting last in the buffer | |
2218 | and yet ending before the origin of the search. | |
2219 | An optional second argument bounds the search; it is a buffer position. | |
2220 | The match found must start at or after that position. | |
2221 | Optional third argument, if t, means if fail just return nil (no error). | |
2222 | If not nil and not t, move to limit of search and return nil. | |
2223 | Optional fourth argument is repeat count--search for successive occurrences. | |
2224 | See also the functions `match-beginning', `match-end', `match-string', | |
2225 | and `replace-match'. */) | |
2226 | (regexp, bound, noerror, count) | |
b819a390 RS |
2227 | Lisp_Object regexp, bound, noerror, count; |
2228 | { | |
2229 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2230 | } | |
2231 | ||
2232 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2233 | "sPosix search: ", |
2234 | doc: /* Search forward from point for regular expression REGEXP. | |
2235 | Find the longest match in accord with Posix regular expression rules. | |
2236 | Set point to the end of the occurrence found, and return point. | |
2237 | An optional second argument bounds the search; it is a buffer position. | |
2238 | The match found must not extend after that position. | |
2239 | Optional third argument, if t, means if fail just return nil (no error). | |
2240 | If not nil and not t, move to limit of search and return nil. | |
2241 | Optional fourth argument is repeat count--search for successive occurrences. | |
2242 | See also the functions `match-beginning', `match-end', `match-string', | |
2243 | and `replace-match'. */) | |
2244 | (regexp, bound, noerror, count) | |
b819a390 RS |
2245 | Lisp_Object regexp, bound, noerror, count; |
2246 | { | |
2247 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2248 | } |
2249 | \f | |
d7a5ad5f | 2250 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2251 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2252 | Leave point at the end of the replacement text. |
2253 | ||
8c1a1077 PJ |
2254 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. |
2255 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2256 | based on the replaced text. | |
2257 | If the replaced text has only capital letters | |
2258 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
4dd0c271 RS |
2259 | Otherwise if all words are capitalized in the replaced text, |
2260 | capitalize each word in NEWTEXT. | |
8c1a1077 PJ |
2261 | |
2262 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2263 | Otherwise treat `\\' as special: | |
2264 | `\\&' in NEWTEXT means substitute original matched text. | |
2265 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2266 | If Nth parens didn't match, substitute nothing. | |
2267 | `\\\\' means insert one `\\'. | |
4dd0c271 RS |
2268 | Case conversion does not apply to these substitutions. |
2269 | ||
8c1a1077 | 2270 | FIXEDCASE and LITERAL are optional arguments. |
8c1a1077 PJ |
2271 | |
2272 | The optional fourth argument STRING can be a string to modify. | |
2273 | This is meaningful when the previous match was done against STRING, | |
2274 | using `string-match'. When used this way, `replace-match' | |
2275 | creates and returns a new string made by copying STRING and replacing | |
2276 | the part of STRING that was matched. | |
2277 | ||
2278 | The optional fifth argument SUBEXP specifies a subexpression; | |
2279 | it says to replace just that subexpression with NEWTEXT, | |
2280 | rather than replacing the entire matched text. | |
2281 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2282 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2283 | NEWTEXT in place of subexp N. | |
2284 | This is useful only after a regular expression search or match, | |
2285 | since only regular expressions have distinguished subexpressions. */) | |
2286 | (newtext, fixedcase, literal, string, subexp) | |
d7a5ad5f | 2287 | Lisp_Object newtext, fixedcase, literal, string, subexp; |
ca1d1d23 JB |
2288 | { |
2289 | enum { nochange, all_caps, cap_initial } case_action; | |
ac3b28b1 | 2290 | register int pos, pos_byte; |
ca1d1d23 | 2291 | int some_multiletter_word; |
97832bd0 | 2292 | int some_lowercase; |
73dc8771 | 2293 | int some_uppercase; |
208767c3 | 2294 | int some_nonuppercase_initial; |
ca1d1d23 | 2295 | register int c, prevc; |
d7a5ad5f | 2296 | int sub; |
3e18eecf | 2297 | int opoint, newpoint; |
ca1d1d23 | 2298 | |
b7826503 | 2299 | CHECK_STRING (newtext); |
ca1d1d23 | 2300 | |
080c45fd | 2301 | if (! NILP (string)) |
b7826503 | 2302 | CHECK_STRING (string); |
080c45fd | 2303 | |
ca1d1d23 JB |
2304 | case_action = nochange; /* We tried an initialization */ |
2305 | /* but some C compilers blew it */ | |
4746118a JB |
2306 | |
2307 | if (search_regs.num_regs <= 0) | |
d72cdbfc | 2308 | error ("`replace-match' called before any match found"); |
4746118a | 2309 | |
d7a5ad5f RS |
2310 | if (NILP (subexp)) |
2311 | sub = 0; | |
2312 | else | |
2313 | { | |
b7826503 | 2314 | CHECK_NUMBER (subexp); |
d7a5ad5f RS |
2315 | sub = XINT (subexp); |
2316 | if (sub < 0 || sub >= search_regs.num_regs) | |
2317 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2318 | } | |
2319 | ||
080c45fd RS |
2320 | if (NILP (string)) |
2321 | { | |
d7a5ad5f RS |
2322 | if (search_regs.start[sub] < BEGV |
2323 | || search_regs.start[sub] > search_regs.end[sub] | |
2324 | || search_regs.end[sub] > ZV) | |
2325 | args_out_of_range (make_number (search_regs.start[sub]), | |
2326 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2327 | } |
2328 | else | |
2329 | { | |
d7a5ad5f RS |
2330 | if (search_regs.start[sub] < 0 |
2331 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2332 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2333 | args_out_of_range (make_number (search_regs.start[sub]), |
2334 | make_number (search_regs.end[sub])); | |
080c45fd | 2335 | } |
ca1d1d23 JB |
2336 | |
2337 | if (NILP (fixedcase)) | |
2338 | { | |
2339 | /* Decide how to casify by examining the matched text. */ | |
ac3b28b1 | 2340 | int last; |
ca1d1d23 | 2341 | |
ac3b28b1 KH |
2342 | pos = search_regs.start[sub]; |
2343 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2344 | |
2345 | if (NILP (string)) | |
ac3b28b1 | 2346 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2347 | else |
ac3b28b1 | 2348 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2349 | |
ca1d1d23 JB |
2350 | prevc = '\n'; |
2351 | case_action = all_caps; | |
2352 | ||
2353 | /* some_multiletter_word is set nonzero if any original word | |
2354 | is more than one letter long. */ | |
2355 | some_multiletter_word = 0; | |
97832bd0 | 2356 | some_lowercase = 0; |
208767c3 | 2357 | some_nonuppercase_initial = 0; |
73dc8771 | 2358 | some_uppercase = 0; |
ca1d1d23 | 2359 | |
ac3b28b1 | 2360 | while (pos < last) |
ca1d1d23 | 2361 | { |
080c45fd | 2362 | if (NILP (string)) |
ac3b28b1 KH |
2363 | { |
2364 | c = FETCH_CHAR (pos_byte); | |
2365 | INC_BOTH (pos, pos_byte); | |
2366 | } | |
080c45fd | 2367 | else |
ac3b28b1 | 2368 | FETCH_STRING_CHAR_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2369 | |
ca1d1d23 JB |
2370 | if (LOWERCASEP (c)) |
2371 | { | |
2372 | /* Cannot be all caps if any original char is lower case */ | |
2373 | ||
97832bd0 | 2374 | some_lowercase = 1; |
ca1d1d23 | 2375 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2376 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2377 | else |
2378 | some_multiletter_word = 1; | |
2379 | } | |
d16c2b66 | 2380 | else if (UPPERCASEP (c)) |
ca1d1d23 | 2381 | { |
73dc8771 | 2382 | some_uppercase = 1; |
97832bd0 | 2383 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2384 | ; |
97832bd0 | 2385 | else |
ca1d1d23 JB |
2386 | some_multiletter_word = 1; |
2387 | } | |
208767c3 RS |
2388 | else |
2389 | { | |
2390 | /* If the initial is a caseless word constituent, | |
2391 | treat that like a lowercase initial. */ | |
2392 | if (SYNTAX (prevc) != Sword) | |
2393 | some_nonuppercase_initial = 1; | |
2394 | } | |
ca1d1d23 JB |
2395 | |
2396 | prevc = c; | |
2397 | } | |
2398 | ||
97832bd0 RS |
2399 | /* Convert to all caps if the old text is all caps |
2400 | and has at least one multiletter word. */ | |
2401 | if (! some_lowercase && some_multiletter_word) | |
2402 | case_action = all_caps; | |
c4d460ce | 2403 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2404 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2405 | case_action = cap_initial; |
208767c3 | 2406 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2407 | /* Should x -> yz, operating on X, give Yz or YZ? |
2408 | We'll assume the latter. */ | |
2409 | case_action = all_caps; | |
97832bd0 RS |
2410 | else |
2411 | case_action = nochange; | |
ca1d1d23 JB |
2412 | } |
2413 | ||
080c45fd RS |
2414 | /* Do replacement in a string. */ |
2415 | if (!NILP (string)) | |
2416 | { | |
2417 | Lisp_Object before, after; | |
2418 | ||
2419 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2420 | make_number (search_regs.start[sub])); |
2421 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2422 | |
636a5e28 RS |
2423 | /* Substitute parts of the match into NEWTEXT |
2424 | if desired. */ | |
080c45fd RS |
2425 | if (NILP (literal)) |
2426 | { | |
d131e79c RS |
2427 | int lastpos = 0; |
2428 | int lastpos_byte = 0; | |
080c45fd RS |
2429 | /* We build up the substituted string in ACCUM. */ |
2430 | Lisp_Object accum; | |
2431 | Lisp_Object middle; | |
d5db4077 | 2432 | int length = SBYTES (newtext); |
080c45fd RS |
2433 | |
2434 | accum = Qnil; | |
2435 | ||
ac3b28b1 | 2436 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd RS |
2437 | { |
2438 | int substart = -1; | |
6bbd7a29 | 2439 | int subend = 0; |
1e79ec24 | 2440 | int delbackslash = 0; |
080c45fd | 2441 | |
0c8533c6 RS |
2442 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2443 | ||
080c45fd RS |
2444 | if (c == '\\') |
2445 | { | |
0c8533c6 | 2446 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2447 | |
080c45fd RS |
2448 | if (c == '&') |
2449 | { | |
d7a5ad5f RS |
2450 | substart = search_regs.start[sub]; |
2451 | subend = search_regs.end[sub]; | |
080c45fd | 2452 | } |
5fbbc83b | 2453 | else if (c >= '1' && c <= '9') |
080c45fd | 2454 | { |
5fbbc83b RS |
2455 | if (search_regs.start[c - '0'] >= 0 |
2456 | && c <= search_regs.num_regs + '0') | |
080c45fd RS |
2457 | { |
2458 | substart = search_regs.start[c - '0']; | |
2459 | subend = search_regs.end[c - '0']; | |
2460 | } | |
5fbbc83b RS |
2461 | else |
2462 | { | |
2463 | /* If that subexp did not match, | |
2464 | replace \\N with nothing. */ | |
2465 | substart = 0; | |
2466 | subend = 0; | |
2467 | } | |
080c45fd | 2468 | } |
1e79ec24 KH |
2469 | else if (c == '\\') |
2470 | delbackslash = 1; | |
636a5e28 RS |
2471 | else |
2472 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2473 | } |
2474 | if (substart >= 0) | |
2475 | { | |
d131e79c RS |
2476 | if (pos - 2 != lastpos) |
2477 | middle = substring_both (newtext, lastpos, | |
2478 | lastpos_byte, | |
2479 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2480 | else |
2481 | middle = Qnil; | |
2482 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2483 | Fsubstring (string, |
2484 | make_number (substart), | |
080c45fd RS |
2485 | make_number (subend))); |
2486 | lastpos = pos; | |
0c8533c6 | 2487 | lastpos_byte = pos_byte; |
080c45fd | 2488 | } |
1e79ec24 KH |
2489 | else if (delbackslash) |
2490 | { | |
d131e79c RS |
2491 | middle = substring_both (newtext, lastpos, |
2492 | lastpos_byte, | |
2493 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2494 | |
1e79ec24 KH |
2495 | accum = concat2 (accum, middle); |
2496 | lastpos = pos; | |
0c8533c6 | 2497 | lastpos_byte = pos_byte; |
1e79ec24 | 2498 | } |
080c45fd RS |
2499 | } |
2500 | ||
d131e79c RS |
2501 | if (pos != lastpos) |
2502 | middle = substring_both (newtext, lastpos, | |
2503 | lastpos_byte, | |
0c8533c6 | 2504 | pos, pos_byte); |
080c45fd RS |
2505 | else |
2506 | middle = Qnil; | |
2507 | ||
2508 | newtext = concat2 (accum, middle); | |
2509 | } | |
2510 | ||
636a5e28 | 2511 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2512 | if (case_action == all_caps) |
2513 | newtext = Fupcase (newtext); | |
2514 | else if (case_action == cap_initial) | |
2b2eead9 | 2515 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2516 | |
2517 | return concat3 (before, newtext, after); | |
2518 | } | |
2519 | ||
09c4719e | 2520 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2521 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2522 | opoint = PT - ZV; |
9160906f RS |
2523 | else if (PT > search_regs.start[sub]) |
2524 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2525 | else |
2526 | opoint = PT; | |
2527 | ||
886ed6ec RS |
2528 | /* If we want non-literal replacement, |
2529 | perform substitution on the replacement string. */ | |
2530 | if (NILP (literal)) | |
ca1d1d23 | 2531 | { |
d5db4077 | 2532 | int length = SBYTES (newtext); |
68e69fbd RS |
2533 | unsigned char *substed; |
2534 | int substed_alloc_size, substed_len; | |
3bc25e52 KH |
2535 | int buf_multibyte = !NILP (current_buffer->enable_multibyte_characters); |
2536 | int str_multibyte = STRING_MULTIBYTE (newtext); | |
2537 | Lisp_Object rev_tbl; | |
886ed6ec | 2538 | int really_changed = 0; |
3bc25e52 KH |
2539 | |
2540 | rev_tbl= (!buf_multibyte && CHAR_TABLE_P (Vnonascii_translation_table) | |
2541 | ? Fchar_table_extra_slot (Vnonascii_translation_table, | |
2542 | make_number (0)) | |
2543 | : Qnil); | |
ac3b28b1 | 2544 | |
68e69fbd RS |
2545 | substed_alloc_size = length * 2 + 100; |
2546 | substed = (unsigned char *) xmalloc (substed_alloc_size + 1); | |
2547 | substed_len = 0; | |
2548 | ||
3bc25e52 KH |
2549 | /* Go thru NEWTEXT, producing the actual text to insert in |
2550 | SUBSTED while adjusting multibyteness to that of the current | |
2551 | buffer. */ | |
ca1d1d23 | 2552 | |
ac3b28b1 | 2553 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2554 | { |
68e69fbd | 2555 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
f8ce8a0d GM |
2556 | unsigned char *add_stuff = NULL; |
2557 | int add_len = 0; | |
68e69fbd | 2558 | int idx = -1; |
9a76659d | 2559 | |
3bc25e52 KH |
2560 | if (str_multibyte) |
2561 | { | |
eb99a8dd | 2562 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 KH |
2563 | if (!buf_multibyte) |
2564 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2565 | } | |
2566 | else | |
2567 | { | |
2568 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2569 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2570 | if (buf_multibyte) |
2571 | c = unibyte_char_to_multibyte (c); | |
2572 | } | |
ac3b28b1 | 2573 | |
68e69fbd RS |
2574 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2575 | or set IDX to a match index, which means put that part | |
2576 | of the buffer text into SUBSTED. */ | |
2577 | ||
ca1d1d23 JB |
2578 | if (c == '\\') |
2579 | { | |
886ed6ec RS |
2580 | really_changed = 1; |
2581 | ||
3bc25e52 KH |
2582 | if (str_multibyte) |
2583 | { | |
eb99a8dd KH |
2584 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2585 | pos, pos_byte); | |
3bc25e52 KH |
2586 | if (!buf_multibyte && !SINGLE_BYTE_CHAR_P (c)) |
2587 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2588 | } | |
2589 | else | |
2590 | { | |
d5db4077 | 2591 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2592 | if (buf_multibyte) |
2593 | c = unibyte_char_to_multibyte (c); | |
2594 | } | |
2595 | ||
ca1d1d23 | 2596 | if (c == '&') |
68e69fbd | 2597 | idx = sub; |
78445046 | 2598 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2599 | { |
2600 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2601 | idx = c - '0'; |
ca1d1d23 | 2602 | } |
636a5e28 | 2603 | else if (c == '\\') |
68e69fbd | 2604 | add_len = 1, add_stuff = "\\"; |
636a5e28 | 2605 | else |
3bc25e52 KH |
2606 | { |
2607 | xfree (substed); | |
2608 | error ("Invalid use of `\\' in replacement text"); | |
2609 | } | |
ca1d1d23 JB |
2610 | } |
2611 | else | |
68e69fbd RS |
2612 | { |
2613 | add_len = CHAR_STRING (c, str); | |
2614 | add_stuff = str; | |
2615 | } | |
2616 | ||
2617 | /* If we want to copy part of a previous match, | |
2618 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2619 | if (idx >= 0) | |
2620 | { | |
2621 | int begbyte = CHAR_TO_BYTE (search_regs.start[idx]); | |
2622 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; | |
2623 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2624 | move_gap (search_regs.start[idx]); | |
2625 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2626 | } | |
2627 | ||
2628 | /* Now the stuff we want to add to SUBSTED | |
2629 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2630 | ||
2631 | /* Make sure SUBSTED is big enough. */ | |
2632 | if (substed_len + add_len >= substed_alloc_size) | |
2633 | { | |
2634 | substed_alloc_size = substed_len + add_len + 500; | |
2635 | substed = (unsigned char *) xrealloc (substed, | |
2636 | substed_alloc_size + 1); | |
2637 | } | |
2638 | ||
2639 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2640 | if (add_stuff) |
2641 | { | |
2642 | bcopy (add_stuff, substed + substed_len, add_len); | |
2643 | substed_len += add_len; | |
2644 | } | |
ca1d1d23 | 2645 | } |
68e69fbd | 2646 | |
886ed6ec | 2647 | if (really_changed) |
80460525 KH |
2648 | { |
2649 | if (buf_multibyte) | |
2650 | { | |
2651 | int nchars = multibyte_chars_in_text (substed, substed_len); | |
68e69fbd | 2652 | |
80460525 KH |
2653 | newtext = make_multibyte_string (substed, nchars, substed_len); |
2654 | } | |
2655 | else | |
2656 | newtext = make_unibyte_string (substed, substed_len); | |
2657 | } | |
68e69fbd | 2658 | xfree (substed); |
ca1d1d23 JB |
2659 | } |
2660 | ||
886ed6ec RS |
2661 | /* Replace the old text with the new in the cleanest possible way. */ |
2662 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2663 | newtext, 1, 0, 1); | |
d5db4077 | 2664 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2665 | |
2666 | if (case_action == all_caps) | |
886ed6ec RS |
2667 | Fupcase_region (make_number (search_regs.start[sub]), |
2668 | make_number (newpoint)); | |
ca1d1d23 | 2669 | else if (case_action == cap_initial) |
886ed6ec RS |
2670 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2671 | make_number (newpoint)); | |
3e18eecf | 2672 | |
98e942e0 RS |
2673 | /* Adjust search data for this change. */ |
2674 | { | |
5b88a2c5 | 2675 | int oldend = search_regs.end[sub]; |
41c01205 | 2676 | int oldstart = search_regs.start[sub]; |
98e942e0 RS |
2677 | int change = newpoint - search_regs.end[sub]; |
2678 | int i; | |
2679 | ||
2680 | for (i = 0; i < search_regs.num_regs; i++) | |
2681 | { | |
41c01205 | 2682 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2683 | search_regs.start[i] += change; |
41c01205 DK |
2684 | else if (search_regs.start[i] > oldstart) |
2685 | search_regs.start[i] = oldstart; | |
2686 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2687 | search_regs.end[i] += change; |
41c01205 DK |
2688 | else if (search_regs.end[i] > oldstart) |
2689 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2690 | } |
2691 | } | |
2692 | ||
b0eba991 | 2693 | /* Put point back where it was in the text. */ |
8d808a65 | 2694 | if (opoint <= 0) |
fa8ed3e0 | 2695 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2696 | else |
fa8ed3e0 | 2697 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2698 | |
2699 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2700 | move_if_not_intangible (newpoint); |
177c0ea7 | 2701 | |
ca1d1d23 JB |
2702 | return Qnil; |
2703 | } | |
2704 | \f | |
2705 | static Lisp_Object | |
2706 | match_limit (num, beginningp) | |
2707 | Lisp_Object num; | |
2708 | int beginningp; | |
2709 | { | |
2710 | register int n; | |
2711 | ||
b7826503 | 2712 | CHECK_NUMBER (num); |
ca1d1d23 | 2713 | n = XINT (num); |
f90a5bf5 | 2714 | if (n < 0) |
bd2cbd56 | 2715 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2716 | if (search_regs.num_regs <= 0) |
2717 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2718 | if (n >= search_regs.num_regs |
4746118a | 2719 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2720 | return Qnil; |
2721 | return (make_number ((beginningp) ? search_regs.start[n] | |
2722 | : search_regs.end[n])); | |
2723 | } | |
2724 | ||
2725 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
8c1a1077 PJ |
2726 | doc: /* Return position of start of text matched by last search. |
2727 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2728 | regexp. | |
2729 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2730 | SUBEXP pairs. | |
2731 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2732 | (subexp) | |
5806161b | 2733 | Lisp_Object subexp; |
ca1d1d23 | 2734 | { |
5806161b | 2735 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2736 | } |
2737 | ||
2738 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
8c1a1077 PJ |
2739 | doc: /* Return position of end of text matched by last search. |
2740 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2741 | regexp. | |
2742 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2743 | SUBEXP pairs. | |
2744 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2745 | (subexp) | |
5806161b | 2746 | Lisp_Object subexp; |
ca1d1d23 | 2747 | { |
5806161b | 2748 | return match_limit (subexp, 0); |
177c0ea7 | 2749 | } |
ca1d1d23 | 2750 | |
abd0071c | 2751 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 3, 0, |
8c1a1077 PJ |
2752 | doc: /* Return a list containing all info on what the last search matched. |
2753 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2754 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2755 | if the last match was on a buffer; integers or nil if a string was matched. | |
2756 | Use `store-match-data' to reinstate the data in this list. | |
2757 | ||
41c01205 DK |
2758 | If INTEGERS (the optional first argument) is non-nil, always use |
2759 | integers \(rather than markers) to represent buffer positions. In | |
2760 | this case, and if the last match was in a buffer, the buffer will get | |
2761 | stored as one additional element at the end of the list. | |
2762 | ||
abd0071c KS |
2763 | If REUSE is a list, reuse it as part of the value. If REUSE is long |
2764 | enough to hold all the values, and if INTEGERS is non-nil, no consing | |
2765 | is done. | |
2766 | ||
2767 | If optional third arg RESEAT is non-nil, any previous markers on the | |
2768 | REUSE list will be modified to point to nowhere. | |
2769 | ||
140a6b7e | 2770 | Return value is undefined if the last search failed. */) |
abd0071c KS |
2771 | (integers, reuse, reseat) |
2772 | Lisp_Object integers, reuse, reseat; | |
ca1d1d23 | 2773 | { |
56256c2a | 2774 | Lisp_Object tail, prev; |
4746118a | 2775 | Lisp_Object *data; |
ca1d1d23 JB |
2776 | int i, len; |
2777 | ||
abd0071c KS |
2778 | if (!NILP (reseat)) |
2779 | for (tail = reuse; CONSP (tail); tail = XCDR (tail)) | |
2780 | if (MARKERP (XCAR (tail))) | |
2781 | { | |
51f10faa | 2782 | unchain_marker (XMARKER (XCAR (tail))); |
abd0071c KS |
2783 | XSETCAR (tail, Qnil); |
2784 | } | |
2785 | ||
daa37602 | 2786 | if (NILP (last_thing_searched)) |
c36bcf1b | 2787 | return Qnil; |
daa37602 | 2788 | |
6bbd7a29 GM |
2789 | prev = Qnil; |
2790 | ||
41c01205 | 2791 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs + 1) |
4746118a JB |
2792 | * sizeof (Lisp_Object)); |
2793 | ||
41c01205 | 2794 | len = 0; |
4746118a | 2795 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2796 | { |
2797 | int start = search_regs.start[i]; | |
2798 | if (start >= 0) | |
2799 | { | |
56256c2a RS |
2800 | if (EQ (last_thing_searched, Qt) |
2801 | || ! NILP (integers)) | |
ca1d1d23 | 2802 | { |
c235cce7 KH |
2803 | XSETFASTINT (data[2 * i], start); |
2804 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2805 | } |
0ed62dc7 | 2806 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2807 | { |
2808 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2809 | Fset_marker (data[2 * i], |
2810 | make_number (start), | |
2811 | last_thing_searched); | |
ca1d1d23 JB |
2812 | data[2 * i + 1] = Fmake_marker (); |
2813 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2814 | make_number (search_regs.end[i]), |
daa37602 | 2815 | last_thing_searched); |
ca1d1d23 | 2816 | } |
daa37602 JB |
2817 | else |
2818 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2819 | abort (); | |
2820 | ||
abd0071c | 2821 | len = 2 * i + 2; |
ca1d1d23 JB |
2822 | } |
2823 | else | |
abd0071c | 2824 | data[2 * i] = data[2 * i + 1] = Qnil; |
ca1d1d23 | 2825 | } |
56256c2a | 2826 | |
bd2cbd56 | 2827 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2828 | { |
bd2cbd56 | 2829 | data[len] = last_thing_searched; |
41c01205 DK |
2830 | len++; |
2831 | } | |
2832 | ||
56256c2a RS |
2833 | /* If REUSE is not usable, cons up the values and return them. */ |
2834 | if (! CONSP (reuse)) | |
41c01205 | 2835 | return Flist (len, data); |
56256c2a RS |
2836 | |
2837 | /* If REUSE is a list, store as many value elements as will fit | |
2838 | into the elements of REUSE. */ | |
2839 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2840 | i++, tail = XCDR (tail)) |
56256c2a | 2841 | { |
41c01205 | 2842 | if (i < len) |
f3fbd155 | 2843 | XSETCAR (tail, data[i]); |
56256c2a | 2844 | else |
f3fbd155 | 2845 | XSETCAR (tail, Qnil); |
56256c2a RS |
2846 | prev = tail; |
2847 | } | |
2848 | ||
2849 | /* If we couldn't fit all value elements into REUSE, | |
2850 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2851 | if (i < len) |
2852 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2853 | |
2854 | return reuse; | |
ca1d1d23 JB |
2855 | } |
2856 | ||
51f10faa KS |
2857 | /* Internal usage only: |
2858 | If RESEAT is `evaporate', put the markers back on the free list | |
2859 | immediately. No other references to the markers must exist in this case, | |
2860 | so it is used only internally on the unwind stack and save-match-data from | |
2861 | Lisp. */ | |
ca1d1d23 | 2862 | |
abd0071c | 2863 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 2, 0, |
8c1a1077 | 2864 | doc: /* Set internal data on last search match from elements of LIST. |
abd0071c KS |
2865 | LIST should have been created by calling `match-data' previously. |
2866 | ||
51f10faa | 2867 | If optional arg RESEAT is non-nil, make markers on LIST point nowhere. */) |
abd0071c KS |
2868 | (list, reseat) |
2869 | register Lisp_Object list, reseat; | |
ca1d1d23 JB |
2870 | { |
2871 | register int i; | |
2872 | register Lisp_Object marker; | |
2873 | ||
7074fde6 FP |
2874 | if (running_asynch_code) |
2875 | save_search_regs (); | |
2876 | ||
ca1d1d23 | 2877 | if (!CONSP (list) && !NILP (list)) |
b37902c8 | 2878 | list = wrong_type_argument (Qconsp, list); |
ca1d1d23 | 2879 | |
41c01205 DK |
2880 | /* Unless we find a marker with a buffer or an explicit buffer |
2881 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
2882 | last_thing_searched = Qt; |
2883 | ||
4746118a JB |
2884 | /* Allocate registers if they don't already exist. */ |
2885 | { | |
d084e942 | 2886 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2887 | |
2888 | if (length > search_regs.num_regs) | |
2889 | { | |
1113d9db JB |
2890 | if (search_regs.num_regs == 0) |
2891 | { | |
2892 | search_regs.start | |
2893 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2894 | search_regs.end | |
2895 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2896 | } | |
4746118a | 2897 | else |
1113d9db JB |
2898 | { |
2899 | search_regs.start | |
2900 | = (regoff_t *) xrealloc (search_regs.start, | |
2901 | length * sizeof (regoff_t)); | |
2902 | search_regs.end | |
2903 | = (regoff_t *) xrealloc (search_regs.end, | |
2904 | length * sizeof (regoff_t)); | |
2905 | } | |
4746118a | 2906 | |
e62371e9 KH |
2907 | for (i = search_regs.num_regs; i < length; i++) |
2908 | search_regs.start[i] = -1; | |
2909 | ||
487282dc | 2910 | search_regs.num_regs = length; |
4746118a | 2911 | } |
ca1d1d23 | 2912 | |
abd0071c | 2913 | for (i = 0; CONSP (list); i++) |
41c01205 | 2914 | { |
abd0071c | 2915 | marker = XCAR (list); |
bd2cbd56 | 2916 | if (BUFFERP (marker)) |
c3762cbd | 2917 | { |
bd2cbd56 | 2918 | last_thing_searched = marker; |
c3762cbd DK |
2919 | break; |
2920 | } | |
2921 | if (i >= length) | |
2922 | break; | |
41c01205 DK |
2923 | if (NILP (marker)) |
2924 | { | |
2925 | search_regs.start[i] = -1; | |
abd0071c | 2926 | list = XCDR (list); |
41c01205 DK |
2927 | } |
2928 | else | |
2929 | { | |
2930 | int from; | |
abd0071c | 2931 | Lisp_Object m; |
e2811828 | 2932 | |
abd0071c | 2933 | m = marker; |
41c01205 DK |
2934 | if (MARKERP (marker)) |
2935 | { | |
2936 | if (XMARKER (marker)->buffer == 0) | |
2937 | XSETFASTINT (marker, 0); | |
2938 | else | |
2939 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
2940 | } | |
e2811828 | 2941 | |
41c01205 DK |
2942 | CHECK_NUMBER_COERCE_MARKER (marker); |
2943 | from = XINT (marker); | |
e2811828 | 2944 | |
abd0071c KS |
2945 | if (!NILP (reseat) && MARKERP (m)) |
2946 | { | |
2947 | if (EQ (reseat, Qevaporate)) | |
2948 | free_marker (m); | |
2949 | else | |
2950 | unchain_marker (XMARKER (m)); | |
9ad54a7e | 2951 | XSETCAR (list, Qnil); |
abd0071c KS |
2952 | } |
2953 | ||
2954 | if ((list = XCDR (list), !CONSP (list))) | |
2955 | break; | |
2956 | ||
2957 | m = marker = XCAR (list); | |
2958 | ||
41c01205 DK |
2959 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
2960 | XSETFASTINT (marker, 0); | |
e2811828 | 2961 | |
41c01205 DK |
2962 | CHECK_NUMBER_COERCE_MARKER (marker); |
2963 | search_regs.start[i] = from; | |
2964 | search_regs.end[i] = XINT (marker); | |
abd0071c KS |
2965 | |
2966 | if (!NILP (reseat) && MARKERP (m)) | |
2967 | { | |
2968 | if (EQ (reseat, Qevaporate)) | |
2969 | free_marker (m); | |
2970 | else | |
2971 | unchain_marker (XMARKER (m)); | |
9ad54a7e | 2972 | XSETCAR (list, Qnil); |
abd0071c | 2973 | } |
41c01205 | 2974 | } |
abd0071c | 2975 | list = XCDR (list); |
41c01205 | 2976 | } |
ca1d1d23 | 2977 | |
41c01205 DK |
2978 | for (; i < search_regs.num_regs; i++) |
2979 | search_regs.start[i] = -1; | |
2980 | } | |
ca1d1d23 | 2981 | |
177c0ea7 | 2982 | return Qnil; |
ca1d1d23 JB |
2983 | } |
2984 | ||
7074fde6 FP |
2985 | /* If non-zero the match data have been saved in saved_search_regs |
2986 | during the execution of a sentinel or filter. */ | |
75ebf74b | 2987 | static int search_regs_saved; |
7074fde6 | 2988 | static struct re_registers saved_search_regs; |
41c01205 | 2989 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
2990 | |
2991 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
2992 | if asynchronous code (filter or sentinel) is running. */ | |
2993 | static void | |
2994 | save_search_regs () | |
2995 | { | |
2996 | if (!search_regs_saved) | |
2997 | { | |
2998 | saved_search_regs.num_regs = search_regs.num_regs; | |
2999 | saved_search_regs.start = search_regs.start; | |
3000 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
3001 | saved_last_thing_searched = last_thing_searched; |
3002 | last_thing_searched = Qnil; | |
7074fde6 | 3003 | search_regs.num_regs = 0; |
2d4a771a RS |
3004 | search_regs.start = 0; |
3005 | search_regs.end = 0; | |
7074fde6 FP |
3006 | |
3007 | search_regs_saved = 1; | |
3008 | } | |
3009 | } | |
3010 | ||
3011 | /* Called upon exit from filters and sentinels. */ | |
3012 | void | |
abd0071c | 3013 | restore_search_regs () |
7074fde6 FP |
3014 | { |
3015 | if (search_regs_saved) | |
3016 | { | |
3017 | if (search_regs.num_regs > 0) | |
3018 | { | |
3019 | xfree (search_regs.start); | |
3020 | xfree (search_regs.end); | |
3021 | } | |
3022 | search_regs.num_regs = saved_search_regs.num_regs; | |
3023 | search_regs.start = saved_search_regs.start; | |
3024 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
3025 | last_thing_searched = saved_last_thing_searched; |
3026 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
3027 | search_regs_saved = 0; |
3028 | } | |
3029 | } | |
3030 | ||
abd0071c KS |
3031 | static Lisp_Object |
3032 | unwind_set_match_data (list) | |
3033 | Lisp_Object list; | |
3034 | { | |
51f10faa | 3035 | /* It is safe to free (evaporate) the markers immediately. */ |
abd0071c KS |
3036 | return Fset_match_data (list, Qevaporate); |
3037 | } | |
3038 | ||
3039 | /* Called to unwind protect the match data. */ | |
3040 | void | |
3041 | record_unwind_save_match_data () | |
3042 | { | |
3043 | record_unwind_protect (unwind_set_match_data, | |
3044 | Fmatch_data (Qnil, Qnil, Qnil)); | |
3045 | } | |
3046 | ||
ca1d1d23 JB |
3047 | /* Quote a string to inactivate reg-expr chars */ |
3048 | ||
3049 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 PJ |
3050 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
3051 | (string) | |
5806161b | 3052 | Lisp_Object string; |
ca1d1d23 JB |
3053 | { |
3054 | register unsigned char *in, *out, *end; | |
3055 | register unsigned char *temp; | |
0c8533c6 | 3056 | int backslashes_added = 0; |
ca1d1d23 | 3057 | |
b7826503 | 3058 | CHECK_STRING (string); |
ca1d1d23 | 3059 | |
d5db4077 | 3060 | temp = (unsigned char *) alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
3061 | |
3062 | /* Now copy the data into the new string, inserting escapes. */ | |
3063 | ||
d5db4077 KR |
3064 | in = SDATA (string); |
3065 | end = in + SBYTES (string); | |
177c0ea7 | 3066 | out = temp; |
ca1d1d23 JB |
3067 | |
3068 | for (; in != end; in++) | |
3069 | { | |
66bc6082 | 3070 | if (*in == '[' |
ca1d1d23 JB |
3071 | || *in == '*' || *in == '.' || *in == '\\' |
3072 | || *in == '?' || *in == '+' | |
3073 | || *in == '^' || *in == '$') | |
0c8533c6 | 3074 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
3075 | *out++ = *in; |
3076 | } | |
3077 | ||
3f8100f1 | 3078 | return make_specified_string (temp, |
d5db4077 | 3079 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
3080 | out - temp, |
3081 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 3082 | } |
177c0ea7 | 3083 | \f |
dfcf069d | 3084 | void |
ca1d1d23 JB |
3085 | syms_of_search () |
3086 | { | |
3087 | register int i; | |
3088 | ||
487282dc KH |
3089 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
3090 | { | |
3091 | searchbufs[i].buf.allocated = 100; | |
b23c0a83 | 3092 | searchbufs[i].buf.buffer = (unsigned char *) xmalloc (100); |
487282dc KH |
3093 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
3094 | searchbufs[i].regexp = Qnil; | |
ecdb561e | 3095 | searchbufs[i].whitespace_regexp = Qnil; |
487282dc | 3096 | staticpro (&searchbufs[i].regexp); |
aa77b5ce | 3097 | staticpro (&searchbufs[i].whitespace_regexp); |
487282dc KH |
3098 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); |
3099 | } | |
3100 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
3101 | |
3102 | Qsearch_failed = intern ("search-failed"); | |
3103 | staticpro (&Qsearch_failed); | |
3104 | Qinvalid_regexp = intern ("invalid-regexp"); | |
3105 | staticpro (&Qinvalid_regexp); | |
3106 | ||
3107 | Fput (Qsearch_failed, Qerror_conditions, | |
3108 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
3109 | Fput (Qsearch_failed, Qerror_message, | |
3110 | build_string ("Search failed")); | |
3111 | ||
3112 | Fput (Qinvalid_regexp, Qerror_conditions, | |
3113 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
3114 | Fput (Qinvalid_regexp, Qerror_message, | |
3115 | build_string ("Invalid regexp")); | |
3116 | ||
daa37602 JB |
3117 | last_thing_searched = Qnil; |
3118 | staticpro (&last_thing_searched); | |
3119 | ||
0f6af254 DK |
3120 | saved_last_thing_searched = Qnil; |
3121 | staticpro (&saved_last_thing_searched); | |
3122 | ||
41a33295 | 3123 | DEFVAR_LISP ("search-spaces-regexp", &Vsearch_spaces_regexp, |
e2811828 | 3124 | doc: /* Regexp to substitute for bunches of spaces in regexp search. |
f31a9a68 RS |
3125 | Some commands use this for user-specified regexps. |
3126 | Spaces that occur inside character classes or repetition operators | |
3127 | or other such regexp constructs are not replaced with this. | |
3128 | A value of nil (which is the normal value) means treat spaces literally. */); | |
41a33295 | 3129 | Vsearch_spaces_regexp = Qnil; |
f31a9a68 | 3130 | |
ca1d1d23 | 3131 | defsubr (&Slooking_at); |
b819a390 RS |
3132 | defsubr (&Sposix_looking_at); |
3133 | defsubr (&Sstring_match); | |
3134 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
3135 | defsubr (&Ssearch_forward); |
3136 | defsubr (&Ssearch_backward); | |
3137 | defsubr (&Sword_search_forward); | |
3138 | defsubr (&Sword_search_backward); | |
3139 | defsubr (&Sre_search_forward); | |
3140 | defsubr (&Sre_search_backward); | |
b819a390 RS |
3141 | defsubr (&Sposix_search_forward); |
3142 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
3143 | defsubr (&Sreplace_match); |
3144 | defsubr (&Smatch_beginning); | |
3145 | defsubr (&Smatch_end); | |
3146 | defsubr (&Smatch_data); | |
3f1c005b | 3147 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
3148 | defsubr (&Sregexp_quote); |
3149 | } | |
ab5796a9 MB |
3150 | |
3151 | /* arch-tag: a6059d79-0552-4f14-a2cb-d379a4e3c78f | |
3152 | (do not change this comment) */ |