Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
68c45bf0 | 2 | Copyright (C) 1985, 86,87,93,94,97,98, 1999 Free Software Foundation, Inc. |
ca1d1d23 JB |
3 | |
4 | This file is part of GNU Emacs. | |
5 | ||
6 | GNU Emacs is free software; you can redistribute it and/or modify | |
7 | it under the terms of the GNU General Public License as published by | |
7c938215 | 8 | the Free Software Foundation; either version 2, or (at your option) |
ca1d1d23 JB |
9 | any later version. |
10 | ||
11 | GNU Emacs is distributed in the hope that it will be useful, | |
12 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
14 | GNU General Public License for more details. | |
15 | ||
16 | You should have received a copy of the GNU General Public License | |
17 | along with GNU Emacs; see the file COPYING. If not, write to | |
3b7ad313 EN |
18 | the Free Software Foundation, Inc., 59 Temple Place - Suite 330, |
19 | Boston, MA 02111-1307, USA. */ | |
ca1d1d23 JB |
20 | |
21 | ||
18160b98 | 22 | #include <config.h> |
ca1d1d23 JB |
23 | #include "lisp.h" |
24 | #include "syntax.h" | |
5679531d | 25 | #include "category.h" |
ca1d1d23 | 26 | #include "buffer.h" |
5679531d | 27 | #include "charset.h" |
9169c321 | 28 | #include "region-cache.h" |
ca1d1d23 | 29 | #include "commands.h" |
9ac0d9e0 | 30 | #include "blockinput.h" |
bf1760bb | 31 | #include "intervals.h" |
4746118a | 32 | |
ca1d1d23 JB |
33 | #include <sys/types.h> |
34 | #include "regex.h" | |
35 | ||
67ce527d KH |
36 | #define min(a, b) ((a) < (b) ? (a) : (b)) |
37 | #define max(a, b) ((a) > (b) ? (a) : (b)) | |
38 | ||
1d288aef | 39 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 40 | |
487282dc KH |
41 | /* If the regexp is non-nil, then the buffer contains the compiled form |
42 | of that regexp, suitable for searching. */ | |
1d288aef RS |
43 | struct regexp_cache |
44 | { | |
487282dc KH |
45 | struct regexp_cache *next; |
46 | Lisp_Object regexp; | |
47 | struct re_pattern_buffer buf; | |
48 | char fastmap[0400]; | |
b819a390 RS |
49 | /* Nonzero means regexp was compiled to do full POSIX backtracking. */ |
50 | char posix; | |
487282dc | 51 | }; |
ca1d1d23 | 52 | |
487282dc KH |
53 | /* The instances of that struct. */ |
54 | struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; | |
ca1d1d23 | 55 | |
487282dc KH |
56 | /* The head of the linked list; points to the most recently used buffer. */ |
57 | struct regexp_cache *searchbuf_head; | |
ca1d1d23 | 58 | |
ca1d1d23 | 59 | |
4746118a JB |
60 | /* Every call to re_match, etc., must pass &search_regs as the regs |
61 | argument unless you can show it is unnecessary (i.e., if re_match | |
62 | is certainly going to be called again before region-around-match | |
63 | can be called). | |
64 | ||
65 | Since the registers are now dynamically allocated, we need to make | |
66 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
67 | been allocated by checking search_regs.num_regs. |
68 | ||
69 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
70 | buffer using bits in the re_pattern_buffer. This means that whenever |
71 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
72 | allocated any registers, and will allocate new registers the next |
73 | time you call a searching or matching function. Therefore, we need | |
74 | to call re_set_registers after compiling a new pattern or after | |
75 | setting the match registers, so that the regex functions will be | |
76 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
77 | static struct re_registers search_regs; |
78 | ||
daa37602 JB |
79 | /* The buffer in which the last search was performed, or |
80 | Qt if the last search was done in a string; | |
81 | Qnil if no searching has been done yet. */ | |
82 | static Lisp_Object last_thing_searched; | |
ca1d1d23 | 83 | |
8e6208c5 | 84 | /* error condition signaled when regexp compile_pattern fails */ |
ca1d1d23 JB |
85 | |
86 | Lisp_Object Qinvalid_regexp; | |
87 | ||
ca325161 | 88 | static void set_search_regs (); |
044f81f1 | 89 | static void save_search_regs (); |
facdc750 RS |
90 | static int simple_search (); |
91 | static int boyer_moore (); | |
b819a390 RS |
92 | static int search_buffer (); |
93 | ||
ca1d1d23 JB |
94 | static void |
95 | matcher_overflow () | |
96 | { | |
97 | error ("Stack overflow in regexp matcher"); | |
98 | } | |
99 | ||
100 | #ifdef __STDC__ | |
101 | #define CONST const | |
102 | #else | |
103 | #define CONST | |
104 | #endif | |
105 | ||
b819a390 RS |
106 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
107 | PATTERN is the pattern to compile. | |
108 | CP is the place to put the result. | |
facdc750 | 109 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
110 | REGP is the structure that says where to store the "register" |
111 | values that will result from matching this pattern. | |
112 | If it is 0, we should compile the pattern not to record any | |
113 | subexpression bounds. | |
114 | POSIX is nonzero if we want full backtracking (POSIX style) | |
5679531d KH |
115 | for this pattern. 0 means backtrack only enough to get a valid match. |
116 | MULTIBYTE is nonzero if we want to handle multibyte characters in | |
117 | PATTERN. 0 means all multibyte characters are recognized just as | |
118 | sequences of binary data. */ | |
ca1d1d23 | 119 | |
487282dc | 120 | static void |
5679531d | 121 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte) |
487282dc | 122 | struct regexp_cache *cp; |
ca1d1d23 | 123 | Lisp_Object pattern; |
facdc750 | 124 | Lisp_Object translate; |
487282dc | 125 | struct re_registers *regp; |
b819a390 | 126 | int posix; |
5679531d | 127 | int multibyte; |
ca1d1d23 | 128 | { |
7276d3d8 | 129 | unsigned char *raw_pattern; |
f8bd51c4 | 130 | int raw_pattern_size; |
d451e4db | 131 | char *val; |
b819a390 | 132 | reg_syntax_t old; |
ca1d1d23 | 133 | |
f8bd51c4 KH |
134 | /* MULTIBYTE says whether the text to be searched is multibyte. |
135 | We must convert PATTERN to match that, or we will not really | |
136 | find things right. */ | |
137 | ||
138 | if (multibyte == STRING_MULTIBYTE (pattern)) | |
139 | { | |
7276d3d8 | 140 | raw_pattern = (unsigned char *) XSTRING (pattern)->data; |
fc932ac6 | 141 | raw_pattern_size = STRING_BYTES (XSTRING (pattern)); |
f8bd51c4 KH |
142 | } |
143 | else if (multibyte) | |
144 | { | |
145 | raw_pattern_size = count_size_as_multibyte (XSTRING (pattern)->data, | |
146 | XSTRING (pattern)->size); | |
7276d3d8 | 147 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
f8bd51c4 KH |
148 | copy_text (XSTRING (pattern)->data, raw_pattern, |
149 | XSTRING (pattern)->size, 0, 1); | |
150 | } | |
151 | else | |
152 | { | |
153 | /* Converting multibyte to single-byte. | |
154 | ||
155 | ??? Perhaps this conversion should be done in a special way | |
156 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
157 | so that only the multibyte chars which really correspond to | |
158 | the chosen single-byte character set can possibly match. */ | |
159 | raw_pattern_size = XSTRING (pattern)->size; | |
7276d3d8 | 160 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
f8bd51c4 | 161 | copy_text (XSTRING (pattern)->data, raw_pattern, |
fc932ac6 | 162 | STRING_BYTES (XSTRING (pattern)), 1, 0); |
f8bd51c4 KH |
163 | } |
164 | ||
487282dc | 165 | cp->regexp = Qnil; |
59fab369 | 166 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 167 | cp->posix = posix; |
5679531d | 168 | cp->buf.multibyte = multibyte; |
9ac0d9e0 | 169 | BLOCK_INPUT; |
d24873d4 | 170 | old = re_set_syntax (RE_SYNTAX_EMACS | RE_CHAR_CLASSES |
b819a390 | 171 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
7276d3d8 RS |
172 | val = (char *) re_compile_pattern ((char *)raw_pattern, |
173 | raw_pattern_size, &cp->buf); | |
b819a390 | 174 | re_set_syntax (old); |
9ac0d9e0 | 175 | UNBLOCK_INPUT; |
ca1d1d23 | 176 | if (val) |
487282dc | 177 | Fsignal (Qinvalid_regexp, Fcons (build_string (val), Qnil)); |
1113d9db | 178 | |
487282dc | 179 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
180 | } |
181 | ||
6efc7887 RS |
182 | /* Shrink each compiled regexp buffer in the cache |
183 | to the size actually used right now. | |
184 | This is called from garbage collection. */ | |
185 | ||
186 | void | |
187 | shrink_regexp_cache () | |
188 | { | |
189 | struct regexp_cache *cp, **cpp; | |
190 | ||
191 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
192 | { | |
193 | cp->buf.allocated = cp->buf.used; | |
194 | cp->buf.buffer | |
195 | = (unsigned char *) realloc (cp->buf.buffer, cp->buf.used); | |
196 | } | |
197 | } | |
198 | ||
487282dc | 199 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
200 | the cache. |
201 | PATTERN is the pattern to compile. | |
facdc750 | 202 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
203 | REGP is the structure that says where to store the "register" |
204 | values that will result from matching this pattern. | |
205 | If it is 0, we should compile the pattern not to record any | |
206 | subexpression bounds. | |
207 | POSIX is nonzero if we want full backtracking (POSIX style) | |
208 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
209 | |
210 | struct re_pattern_buffer * | |
0c8533c6 | 211 | compile_pattern (pattern, regp, translate, posix, multibyte) |
487282dc KH |
212 | Lisp_Object pattern; |
213 | struct re_registers *regp; | |
facdc750 | 214 | Lisp_Object translate; |
0c8533c6 | 215 | int posix, multibyte; |
487282dc KH |
216 | { |
217 | struct regexp_cache *cp, **cpp; | |
218 | ||
219 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
220 | { | |
221 | cp = *cpp; | |
1d288aef RS |
222 | if (XSTRING (cp->regexp)->size == XSTRING (pattern)->size |
223 | && !NILP (Fstring_equal (cp->regexp, pattern)) | |
59fab369 | 224 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d KH |
225 | && cp->posix == posix |
226 | && cp->buf.multibyte == multibyte) | |
487282dc KH |
227 | break; |
228 | ||
229 | /* If we're at the end of the cache, compile into the last cell. */ | |
230 | if (cp->next == 0) | |
231 | { | |
5679531d | 232 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte); |
487282dc KH |
233 | break; |
234 | } | |
235 | } | |
236 | ||
237 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
238 | either because we found it in the cache or because we just compiled it. | |
239 | Move it to the front of the queue to mark it as most recently used. */ | |
240 | *cpp = cp->next; | |
241 | cp->next = searchbuf_head; | |
242 | searchbuf_head = cp; | |
1113d9db | 243 | |
6639708c RS |
244 | /* Advise the searching functions about the space we have allocated |
245 | for register data. */ | |
246 | if (regp) | |
247 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
248 | ||
487282dc | 249 | return &cp->buf; |
ca1d1d23 JB |
250 | } |
251 | ||
252 | /* Error condition used for failing searches */ | |
253 | Lisp_Object Qsearch_failed; | |
254 | ||
255 | Lisp_Object | |
256 | signal_failure (arg) | |
257 | Lisp_Object arg; | |
258 | { | |
259 | Fsignal (Qsearch_failed, Fcons (arg, Qnil)); | |
260 | return Qnil; | |
261 | } | |
262 | \f | |
b819a390 RS |
263 | static Lisp_Object |
264 | looking_at_1 (string, posix) | |
ca1d1d23 | 265 | Lisp_Object string; |
b819a390 | 266 | int posix; |
ca1d1d23 JB |
267 | { |
268 | Lisp_Object val; | |
269 | unsigned char *p1, *p2; | |
270 | int s1, s2; | |
271 | register int i; | |
487282dc | 272 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 273 | |
7074fde6 FP |
274 | if (running_asynch_code) |
275 | save_search_regs (); | |
276 | ||
ca1d1d23 | 277 | CHECK_STRING (string, 0); |
487282dc KH |
278 | bufp = compile_pattern (string, &search_regs, |
279 | (!NILP (current_buffer->case_fold_search) | |
facdc750 | 280 | ? DOWNCASE_TABLE : Qnil), |
0c8533c6 RS |
281 | posix, |
282 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 JB |
283 | |
284 | immediate_quit = 1; | |
285 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
286 | ||
287 | /* Get pointers and sizes of the two strings | |
288 | that make up the visible portion of the buffer. */ | |
289 | ||
290 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 291 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 292 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 293 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
294 | if (s1 < 0) |
295 | { | |
296 | p2 = p1; | |
fa8ed3e0 | 297 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
298 | s1 = 0; |
299 | } | |
300 | if (s2 < 0) | |
301 | { | |
fa8ed3e0 | 302 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
303 | s2 = 0; |
304 | } | |
8bb43c28 RS |
305 | |
306 | re_match_object = Qnil; | |
ca1d1d23 | 307 | |
487282dc | 308 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
fa8ed3e0 RS |
309 | PT_BYTE - BEGV_BYTE, &search_regs, |
310 | ZV_BYTE - BEGV_BYTE); | |
ca1d1d23 JB |
311 | if (i == -2) |
312 | matcher_overflow (); | |
313 | ||
314 | val = (0 <= i ? Qt : Qnil); | |
fa8ed3e0 RS |
315 | if (i >= 0) |
316 | for (i = 0; i < search_regs.num_regs; i++) | |
317 | if (search_regs.start[i] >= 0) | |
318 | { | |
319 | search_regs.start[i] | |
320 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
321 | search_regs.end[i] | |
322 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
323 | } | |
a3668d92 | 324 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
325 | immediate_quit = 0; |
326 | return val; | |
327 | } | |
328 | ||
b819a390 | 329 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
94f94972 | 330 | "Return t if text after point matches regular expression REGEXP.\n\ |
b819a390 RS |
331 | This function modifies the match data that `match-beginning',\n\ |
332 | `match-end' and `match-data' access; save and restore the match\n\ | |
333 | data if you want to preserve them.") | |
94f94972 RS |
334 | (regexp) |
335 | Lisp_Object regexp; | |
b819a390 | 336 | { |
94f94972 | 337 | return looking_at_1 (regexp, 0); |
b819a390 RS |
338 | } |
339 | ||
340 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
94f94972 | 341 | "Return t if text after point matches regular expression REGEXP.\n\ |
b819a390 RS |
342 | Find the longest match, in accord with Posix regular expression rules.\n\ |
343 | This function modifies the match data that `match-beginning',\n\ | |
344 | `match-end' and `match-data' access; save and restore the match\n\ | |
345 | data if you want to preserve them.") | |
94f94972 RS |
346 | (regexp) |
347 | Lisp_Object regexp; | |
b819a390 | 348 | { |
94f94972 | 349 | return looking_at_1 (regexp, 1); |
b819a390 RS |
350 | } |
351 | \f | |
352 | static Lisp_Object | |
353 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 354 | Lisp_Object regexp, string, start; |
b819a390 | 355 | int posix; |
ca1d1d23 JB |
356 | { |
357 | int val; | |
487282dc | 358 | struct re_pattern_buffer *bufp; |
0c8533c6 RS |
359 | int pos, pos_byte; |
360 | int i; | |
ca1d1d23 | 361 | |
7074fde6 FP |
362 | if (running_asynch_code) |
363 | save_search_regs (); | |
364 | ||
ca1d1d23 JB |
365 | CHECK_STRING (regexp, 0); |
366 | CHECK_STRING (string, 1); | |
367 | ||
368 | if (NILP (start)) | |
0c8533c6 | 369 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
370 | else |
371 | { | |
372 | int len = XSTRING (string)->size; | |
373 | ||
374 | CHECK_NUMBER (start, 2); | |
0c8533c6 RS |
375 | pos = XINT (start); |
376 | if (pos < 0 && -pos <= len) | |
377 | pos = len + pos; | |
378 | else if (0 > pos || pos > len) | |
ca1d1d23 | 379 | args_out_of_range (string, start); |
0c8533c6 | 380 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
381 | } |
382 | ||
487282dc KH |
383 | bufp = compile_pattern (regexp, &search_regs, |
384 | (!NILP (current_buffer->case_fold_search) | |
facdc750 | 385 | ? DOWNCASE_TABLE : Qnil), |
0c8533c6 RS |
386 | posix, |
387 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 388 | immediate_quit = 1; |
8bb43c28 RS |
389 | re_match_object = string; |
390 | ||
487282dc | 391 | val = re_search (bufp, (char *) XSTRING (string)->data, |
fc932ac6 RS |
392 | STRING_BYTES (XSTRING (string)), pos_byte, |
393 | STRING_BYTES (XSTRING (string)) - pos_byte, | |
ca1d1d23 JB |
394 | &search_regs); |
395 | immediate_quit = 0; | |
daa37602 | 396 | last_thing_searched = Qt; |
ca1d1d23 JB |
397 | if (val == -2) |
398 | matcher_overflow (); | |
399 | if (val < 0) return Qnil; | |
0c8533c6 RS |
400 | |
401 | for (i = 0; i < search_regs.num_regs; i++) | |
402 | if (search_regs.start[i] >= 0) | |
403 | { | |
404 | search_regs.start[i] | |
405 | = string_byte_to_char (string, search_regs.start[i]); | |
406 | search_regs.end[i] | |
407 | = string_byte_to_char (string, search_regs.end[i]); | |
408 | } | |
409 | ||
410 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 411 | } |
e59a8453 | 412 | |
b819a390 RS |
413 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
414 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
21e01890 | 415 | Case is ignored if `case-fold-search' is non-nil in the current buffer.\n\ |
b819a390 RS |
416 | If third arg START is non-nil, start search at that index in STRING.\n\ |
417 | For index of first char beyond the match, do (match-end 0).\n\ | |
418 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
419 | matched by parenthesis constructs in the pattern.") | |
420 | (regexp, string, start) | |
421 | Lisp_Object regexp, string, start; | |
422 | { | |
423 | return string_match_1 (regexp, string, start, 0); | |
424 | } | |
425 | ||
426 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
427 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
428 | Find the longest match, in accord with Posix regular expression rules.\n\ | |
21e01890 | 429 | Case is ignored if `case-fold-search' is non-nil in the current buffer.\n\ |
b819a390 RS |
430 | If third arg START is non-nil, start search at that index in STRING.\n\ |
431 | For index of first char beyond the match, do (match-end 0).\n\ | |
432 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
433 | matched by parenthesis constructs in the pattern.") | |
434 | (regexp, string, start) | |
435 | Lisp_Object regexp, string, start; | |
436 | { | |
437 | return string_match_1 (regexp, string, start, 1); | |
438 | } | |
439 | ||
e59a8453 RS |
440 | /* Match REGEXP against STRING, searching all of STRING, |
441 | and return the index of the match, or negative on failure. | |
442 | This does not clobber the match data. */ | |
443 | ||
444 | int | |
445 | fast_string_match (regexp, string) | |
446 | Lisp_Object regexp, string; | |
447 | { | |
448 | int val; | |
487282dc | 449 | struct re_pattern_buffer *bufp; |
e59a8453 | 450 | |
facdc750 RS |
451 | bufp = compile_pattern (regexp, 0, Qnil, |
452 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 453 | immediate_quit = 1; |
8bb43c28 RS |
454 | re_match_object = string; |
455 | ||
487282dc | 456 | val = re_search (bufp, (char *) XSTRING (string)->data, |
fc932ac6 RS |
457 | STRING_BYTES (XSTRING (string)), 0, |
458 | STRING_BYTES (XSTRING (string)), 0); | |
e59a8453 RS |
459 | immediate_quit = 0; |
460 | return val; | |
461 | } | |
5679531d KH |
462 | |
463 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
464 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
465 | This does not clobber the match data. |
466 | We assume that STRING contains single-byte characters. */ | |
5679531d KH |
467 | |
468 | extern Lisp_Object Vascii_downcase_table; | |
469 | ||
470 | int | |
b4577c63 | 471 | fast_c_string_match_ignore_case (regexp, string) |
5679531d KH |
472 | Lisp_Object regexp; |
473 | char *string; | |
474 | { | |
475 | int val; | |
476 | struct re_pattern_buffer *bufp; | |
477 | int len = strlen (string); | |
478 | ||
0c8533c6 | 479 | regexp = string_make_unibyte (regexp); |
b4577c63 | 480 | re_match_object = Qt; |
5679531d | 481 | bufp = compile_pattern (regexp, 0, |
facdc750 | 482 | Vascii_downcase_table, 0, |
f8bd51c4 | 483 | 0); |
5679531d KH |
484 | immediate_quit = 1; |
485 | val = re_search (bufp, string, len, 0, len, 0); | |
486 | immediate_quit = 0; | |
487 | return val; | |
488 | } | |
ca1d1d23 | 489 | \f |
9169c321 JB |
490 | /* The newline cache: remembering which sections of text have no newlines. */ |
491 | ||
492 | /* If the user has requested newline caching, make sure it's on. | |
493 | Otherwise, make sure it's off. | |
494 | This is our cheezy way of associating an action with the change of | |
495 | state of a buffer-local variable. */ | |
496 | static void | |
497 | newline_cache_on_off (buf) | |
498 | struct buffer *buf; | |
499 | { | |
500 | if (NILP (buf->cache_long_line_scans)) | |
501 | { | |
502 | /* It should be off. */ | |
503 | if (buf->newline_cache) | |
504 | { | |
505 | free_region_cache (buf->newline_cache); | |
506 | buf->newline_cache = 0; | |
507 | } | |
508 | } | |
509 | else | |
510 | { | |
511 | /* It should be on. */ | |
512 | if (buf->newline_cache == 0) | |
513 | buf->newline_cache = new_region_cache (); | |
514 | } | |
515 | } | |
516 | ||
517 | \f | |
518 | /* Search for COUNT instances of the character TARGET between START and END. | |
519 | ||
520 | If COUNT is positive, search forwards; END must be >= START. | |
521 | If COUNT is negative, search backwards for the -COUNTth instance; | |
522 | END must be <= START. | |
523 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
524 | ||
525 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
526 | direction indicated by COUNT. | |
ffd56f97 JB |
527 | |
528 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
5bfe95c9 RS |
529 | position after the COUNTth match. Note that for reverse motion |
530 | this is not the same as the usual convention for Emacs motion commands. | |
ffd56f97 | 531 | |
9169c321 JB |
532 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
533 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 534 | |
087a5f81 RS |
535 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
536 | except when inside redisplay. */ | |
537 | ||
dfcf069d | 538 | int |
9169c321 JB |
539 | scan_buffer (target, start, end, count, shortage, allow_quit) |
540 | register int target; | |
541 | int start, end; | |
542 | int count; | |
543 | int *shortage; | |
087a5f81 | 544 | int allow_quit; |
ca1d1d23 | 545 | { |
9169c321 JB |
546 | struct region_cache *newline_cache; |
547 | int direction; | |
ffd56f97 | 548 | |
9169c321 JB |
549 | if (count > 0) |
550 | { | |
551 | direction = 1; | |
552 | if (! end) end = ZV; | |
553 | } | |
554 | else | |
555 | { | |
556 | direction = -1; | |
557 | if (! end) end = BEGV; | |
558 | } | |
ffd56f97 | 559 | |
9169c321 JB |
560 | newline_cache_on_off (current_buffer); |
561 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
562 | |
563 | if (shortage != 0) | |
564 | *shortage = 0; | |
565 | ||
087a5f81 | 566 | immediate_quit = allow_quit; |
ca1d1d23 | 567 | |
ffd56f97 | 568 | if (count > 0) |
9169c321 | 569 | while (start != end) |
ca1d1d23 | 570 | { |
9169c321 JB |
571 | /* Our innermost scanning loop is very simple; it doesn't know |
572 | about gaps, buffer ends, or the newline cache. ceiling is | |
573 | the position of the last character before the next such | |
574 | obstacle --- the last character the dumb search loop should | |
575 | examine. */ | |
fa8ed3e0 RS |
576 | int ceiling_byte = CHAR_TO_BYTE (end) - 1; |
577 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 578 | int tem; |
9169c321 JB |
579 | |
580 | /* If we're looking for a newline, consult the newline cache | |
581 | to see where we can avoid some scanning. */ | |
582 | if (target == '\n' && newline_cache) | |
583 | { | |
584 | int next_change; | |
585 | immediate_quit = 0; | |
586 | while (region_cache_forward | |
fa8ed3e0 RS |
587 | (current_buffer, newline_cache, start_byte, &next_change)) |
588 | start_byte = next_change; | |
cbe0db0d | 589 | immediate_quit = allow_quit; |
9169c321 | 590 | |
fa8ed3e0 RS |
591 | /* START should never be after END. */ |
592 | if (start_byte > ceiling_byte) | |
593 | start_byte = ceiling_byte; | |
9169c321 JB |
594 | |
595 | /* Now the text after start is an unknown region, and | |
596 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 597 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
598 | } |
599 | ||
600 | /* The dumb loop can only scan text stored in contiguous | |
601 | bytes. BUFFER_CEILING_OF returns the last character | |
602 | position that is contiguous, so the ceiling is the | |
603 | position after that. */ | |
67ce527d KH |
604 | tem = BUFFER_CEILING_OF (start_byte); |
605 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
606 | |
607 | { | |
608 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
609 | register unsigned char *ceiling_addr |
610 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
611 | register unsigned char *cursor | |
612 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
613 | unsigned char *base = cursor; |
614 | ||
615 | while (cursor < ceiling_addr) | |
616 | { | |
617 | unsigned char *scan_start = cursor; | |
618 | ||
619 | /* The dumb loop. */ | |
620 | while (*cursor != target && ++cursor < ceiling_addr) | |
621 | ; | |
622 | ||
623 | /* If we're looking for newlines, cache the fact that | |
624 | the region from start to cursor is free of them. */ | |
625 | if (target == '\n' && newline_cache) | |
626 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
627 | start_byte + scan_start - base, |
628 | start_byte + cursor - base); | |
9169c321 JB |
629 | |
630 | /* Did we find the target character? */ | |
631 | if (cursor < ceiling_addr) | |
632 | { | |
633 | if (--count == 0) | |
634 | { | |
635 | immediate_quit = 0; | |
fa8ed3e0 | 636 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
637 | } |
638 | cursor++; | |
639 | } | |
640 | } | |
641 | ||
fa8ed3e0 | 642 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 643 | } |
ca1d1d23 JB |
644 | } |
645 | else | |
9169c321 JB |
646 | while (start > end) |
647 | { | |
648 | /* The last character to check before the next obstacle. */ | |
fa8ed3e0 RS |
649 | int ceiling_byte = CHAR_TO_BYTE (end); |
650 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 651 | int tem; |
9169c321 JB |
652 | |
653 | /* Consult the newline cache, if appropriate. */ | |
654 | if (target == '\n' && newline_cache) | |
655 | { | |
656 | int next_change; | |
657 | immediate_quit = 0; | |
658 | while (region_cache_backward | |
fa8ed3e0 RS |
659 | (current_buffer, newline_cache, start_byte, &next_change)) |
660 | start_byte = next_change; | |
cbe0db0d | 661 | immediate_quit = allow_quit; |
9169c321 JB |
662 | |
663 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
664 | if (start_byte <= ceiling_byte) |
665 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
666 | |
667 | /* Now the text before start is an unknown region, and | |
668 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 669 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
670 | } |
671 | ||
672 | /* Stop scanning before the gap. */ | |
67ce527d KH |
673 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
674 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
675 | |
676 | { | |
677 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
678 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
679 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
680 | unsigned char *base = cursor; |
681 | ||
682 | while (cursor >= ceiling_addr) | |
683 | { | |
684 | unsigned char *scan_start = cursor; | |
685 | ||
686 | while (*cursor != target && --cursor >= ceiling_addr) | |
687 | ; | |
688 | ||
689 | /* If we're looking for newlines, cache the fact that | |
690 | the region from after the cursor to start is free of them. */ | |
691 | if (target == '\n' && newline_cache) | |
692 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
693 | start_byte + cursor - base, |
694 | start_byte + scan_start - base); | |
9169c321 JB |
695 | |
696 | /* Did we find the target character? */ | |
697 | if (cursor >= ceiling_addr) | |
698 | { | |
699 | if (++count >= 0) | |
700 | { | |
701 | immediate_quit = 0; | |
fa8ed3e0 | 702 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
703 | } |
704 | cursor--; | |
705 | } | |
706 | } | |
707 | ||
fa8ed3e0 | 708 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
709 | } |
710 | } | |
711 | ||
ca1d1d23 JB |
712 | immediate_quit = 0; |
713 | if (shortage != 0) | |
ffd56f97 | 714 | *shortage = count * direction; |
9169c321 | 715 | return start; |
ca1d1d23 | 716 | } |
fa8ed3e0 RS |
717 | \f |
718 | /* Search for COUNT instances of a line boundary, which means either a | |
719 | newline or (if selective display enabled) a carriage return. | |
720 | Start at START. If COUNT is negative, search backwards. | |
721 | ||
722 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
723 | ||
724 | If we find COUNT instances. we position after (always after, | |
725 | even if scanning backwards) the COUNTth match, and return 0. | |
726 | ||
727 | If we don't find COUNT instances before reaching the end of the | |
728 | buffer (or the beginning, if scanning backwards), we return | |
729 | the number of line boundaries left unfound, and position at | |
730 | the limit we bumped up against. | |
731 | ||
732 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 733 | except in special cases. */ |
ca1d1d23 | 734 | |
63fa018d | 735 | int |
fa8ed3e0 RS |
736 | scan_newline (start, start_byte, limit, limit_byte, count, allow_quit) |
737 | int start, start_byte; | |
738 | int limit, limit_byte; | |
739 | register int count; | |
740 | int allow_quit; | |
63fa018d | 741 | { |
fa8ed3e0 RS |
742 | int direction = ((count > 0) ? 1 : -1); |
743 | ||
744 | register unsigned char *cursor; | |
745 | unsigned char *base; | |
746 | ||
747 | register int ceiling; | |
748 | register unsigned char *ceiling_addr; | |
749 | ||
d5d57b92 RS |
750 | int old_immediate_quit = immediate_quit; |
751 | ||
fa8ed3e0 RS |
752 | /* If we are not in selective display mode, |
753 | check only for newlines. */ | |
754 | int selective_display = (!NILP (current_buffer->selective_display) | |
755 | && !INTEGERP (current_buffer->selective_display)); | |
756 | ||
757 | /* The code that follows is like scan_buffer | |
758 | but checks for either newline or carriage return. */ | |
759 | ||
d5d57b92 RS |
760 | if (allow_quit) |
761 | immediate_quit++; | |
fa8ed3e0 RS |
762 | |
763 | start_byte = CHAR_TO_BYTE (start); | |
764 | ||
765 | if (count > 0) | |
766 | { | |
767 | while (start_byte < limit_byte) | |
768 | { | |
769 | ceiling = BUFFER_CEILING_OF (start_byte); | |
770 | ceiling = min (limit_byte - 1, ceiling); | |
771 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
772 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
773 | while (1) | |
774 | { | |
775 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
776 | ; | |
777 | ||
778 | if (cursor != ceiling_addr) | |
779 | { | |
780 | if (--count == 0) | |
781 | { | |
d5d57b92 | 782 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
783 | start_byte = start_byte + cursor - base + 1; |
784 | start = BYTE_TO_CHAR (start_byte); | |
785 | TEMP_SET_PT_BOTH (start, start_byte); | |
786 | return 0; | |
787 | } | |
788 | else | |
789 | if (++cursor == ceiling_addr) | |
790 | break; | |
791 | } | |
792 | else | |
793 | break; | |
794 | } | |
795 | start_byte += cursor - base; | |
796 | } | |
797 | } | |
798 | else | |
799 | { | |
fa8ed3e0 RS |
800 | while (start_byte > limit_byte) |
801 | { | |
802 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
803 | ceiling = max (limit_byte, ceiling); | |
804 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
805 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
806 | while (1) | |
807 | { | |
808 | while (--cursor != ceiling_addr && *cursor != '\n') | |
809 | ; | |
810 | ||
811 | if (cursor != ceiling_addr) | |
812 | { | |
813 | if (++count == 0) | |
814 | { | |
d5d57b92 | 815 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
816 | /* Return the position AFTER the match we found. */ |
817 | start_byte = start_byte + cursor - base + 1; | |
818 | start = BYTE_TO_CHAR (start_byte); | |
819 | TEMP_SET_PT_BOTH (start, start_byte); | |
820 | return 0; | |
821 | } | |
822 | } | |
823 | else | |
824 | break; | |
825 | } | |
826 | /* Here we add 1 to compensate for the last decrement | |
827 | of CURSOR, which took it past the valid range. */ | |
828 | start_byte += cursor - base + 1; | |
829 | } | |
830 | } | |
831 | ||
832 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 833 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
834 | |
835 | return count * direction; | |
63fa018d RS |
836 | } |
837 | ||
ca1d1d23 | 838 | int |
fa8ed3e0 | 839 | find_next_newline_no_quit (from, cnt) |
ca1d1d23 JB |
840 | register int from, cnt; |
841 | { | |
fa8ed3e0 | 842 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
9169c321 JB |
843 | } |
844 | ||
9169c321 JB |
845 | /* Like find_next_newline, but returns position before the newline, |
846 | not after, and only search up to TO. This isn't just | |
847 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 848 | |
9169c321 JB |
849 | int |
850 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 851 | int from, to, cnt; |
9169c321 JB |
852 | { |
853 | int shortage; | |
854 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
855 | ||
856 | if (shortage == 0) | |
857 | pos--; | |
858 | ||
859 | return pos; | |
ca1d1d23 JB |
860 | } |
861 | \f | |
ca1d1d23 JB |
862 | /* Subroutines of Lisp buffer search functions. */ |
863 | ||
864 | static Lisp_Object | |
b819a390 | 865 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
866 | Lisp_Object string, bound, noerror, count; |
867 | int direction; | |
868 | int RE; | |
b819a390 | 869 | int posix; |
ca1d1d23 JB |
870 | { |
871 | register int np; | |
9f43ad85 | 872 | int lim, lim_byte; |
ca1d1d23 JB |
873 | int n = direction; |
874 | ||
875 | if (!NILP (count)) | |
876 | { | |
877 | CHECK_NUMBER (count, 3); | |
878 | n *= XINT (count); | |
879 | } | |
880 | ||
881 | CHECK_STRING (string, 0); | |
882 | if (NILP (bound)) | |
9f43ad85 RS |
883 | { |
884 | if (n > 0) | |
885 | lim = ZV, lim_byte = ZV_BYTE; | |
886 | else | |
887 | lim = BEGV, lim_byte = BEGV_BYTE; | |
888 | } | |
ca1d1d23 JB |
889 | else |
890 | { | |
891 | CHECK_NUMBER_COERCE_MARKER (bound, 1); | |
892 | lim = XINT (bound); | |
6ec8bbd2 | 893 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
894 | error ("Invalid search bound (wrong side of point)"); |
895 | if (lim > ZV) | |
9f43ad85 | 896 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 897 | else if (lim < BEGV) |
9f43ad85 | 898 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
899 | else |
900 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
901 | } |
902 | ||
9f43ad85 | 903 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
ca1d1d23 | 904 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 905 | ? current_buffer->case_canon_table |
3135e9fd | 906 | : Qnil), |
ca1d1d23 | 907 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 908 | ? current_buffer->case_eqv_table |
3135e9fd | 909 | : Qnil), |
b819a390 | 910 | posix); |
ca1d1d23 JB |
911 | if (np <= 0) |
912 | { | |
913 | if (NILP (noerror)) | |
914 | return signal_failure (string); | |
915 | if (!EQ (noerror, Qt)) | |
916 | { | |
917 | if (lim < BEGV || lim > ZV) | |
918 | abort (); | |
9f43ad85 | 919 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
920 | return Qnil; |
921 | #if 0 /* This would be clean, but maybe programs depend on | |
922 | a value of nil here. */ | |
481399bf | 923 | np = lim; |
a5f217b8 | 924 | #endif |
ca1d1d23 | 925 | } |
481399bf RS |
926 | else |
927 | return Qnil; | |
ca1d1d23 JB |
928 | } |
929 | ||
930 | if (np < BEGV || np > ZV) | |
931 | abort (); | |
932 | ||
933 | SET_PT (np); | |
934 | ||
935 | return make_number (np); | |
936 | } | |
937 | \f | |
fa8ed3e0 RS |
938 | /* Return 1 if REGEXP it matches just one constant string. */ |
939 | ||
b6d6a51c KH |
940 | static int |
941 | trivial_regexp_p (regexp) | |
942 | Lisp_Object regexp; | |
943 | { | |
fc932ac6 | 944 | int len = STRING_BYTES (XSTRING (regexp)); |
b6d6a51c KH |
945 | unsigned char *s = XSTRING (regexp)->data; |
946 | unsigned char c; | |
947 | while (--len >= 0) | |
948 | { | |
949 | switch (*s++) | |
950 | { | |
951 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
952 | return 0; | |
953 | case '\\': | |
954 | if (--len < 0) | |
955 | return 0; | |
956 | switch (*s++) | |
957 | { | |
958 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
959 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
866f60fd | 960 | case 'S': case '=': |
5679531d | 961 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 962 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
963 | case '6': case '7': case '8': case '9': |
964 | return 0; | |
965 | } | |
966 | } | |
967 | } | |
968 | return 1; | |
969 | } | |
970 | ||
ca325161 | 971 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 972 | starting at position POS and stopping at position LIM, |
b819a390 | 973 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
974 | a regular expression if RE is true. |
975 | ||
976 | If N is positive, searching is forward and LIM must be greater than POS. | |
977 | If N is negative, searching is backward and LIM must be less than POS. | |
978 | ||
facdc750 | 979 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 980 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
981 | (if searching backward) or the end (if searching forward). |
982 | ||
983 | POSIX is nonzero if we want full backtracking (POSIX style) | |
984 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 985 | |
aff2ce94 RS |
986 | #define TRANSLATE(out, trt, d) \ |
987 | do \ | |
988 | { \ | |
989 | if (! NILP (trt)) \ | |
990 | { \ | |
991 | Lisp_Object temp; \ | |
992 | temp = Faref (trt, make_number (d)); \ | |
993 | if (INTEGERP (temp)) \ | |
994 | out = XINT (temp); \ | |
995 | else \ | |
996 | out = d; \ | |
997 | } \ | |
998 | else \ | |
999 | out = d; \ | |
1000 | } \ | |
1001 | while (0) | |
facdc750 | 1002 | |
b819a390 | 1003 | static int |
9f43ad85 RS |
1004 | search_buffer (string, pos, pos_byte, lim, lim_byte, n, |
1005 | RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
1006 | Lisp_Object string; |
1007 | int pos; | |
9f43ad85 | 1008 | int pos_byte; |
ca1d1d23 | 1009 | int lim; |
9f43ad85 | 1010 | int lim_byte; |
ca1d1d23 JB |
1011 | int n; |
1012 | int RE; | |
facdc750 RS |
1013 | Lisp_Object trt; |
1014 | Lisp_Object inverse_trt; | |
b819a390 | 1015 | int posix; |
ca1d1d23 JB |
1016 | { |
1017 | int len = XSTRING (string)->size; | |
fc932ac6 | 1018 | int len_byte = STRING_BYTES (XSTRING (string)); |
facdc750 | 1019 | register int i; |
ca1d1d23 | 1020 | |
7074fde6 FP |
1021 | if (running_asynch_code) |
1022 | save_search_regs (); | |
1023 | ||
a7e4cdde | 1024 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1025 | /* Null string is found at starting position. */ |
a7e4cdde | 1026 | if (len == 0 || n == 0) |
ca325161 RS |
1027 | { |
1028 | set_search_regs (pos, 0); | |
1029 | return pos; | |
1030 | } | |
3f57a499 | 1031 | |
b6d6a51c | 1032 | if (RE && !trivial_regexp_p (string)) |
ca1d1d23 | 1033 | { |
facdc750 RS |
1034 | unsigned char *p1, *p2; |
1035 | int s1, s2; | |
487282dc KH |
1036 | struct re_pattern_buffer *bufp; |
1037 | ||
0c8533c6 RS |
1038 | bufp = compile_pattern (string, &search_regs, trt, posix, |
1039 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 | 1040 | |
ca1d1d23 JB |
1041 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1042 | because letting this function finish | |
1043 | can take too long. */ | |
1044 | QUIT; /* Do a pending quit right away, | |
1045 | to avoid paradoxical behavior */ | |
1046 | /* Get pointers and sizes of the two strings | |
1047 | that make up the visible portion of the buffer. */ | |
1048 | ||
1049 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1050 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1051 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1052 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1053 | if (s1 < 0) |
1054 | { | |
1055 | p2 = p1; | |
fa8ed3e0 | 1056 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1057 | s1 = 0; |
1058 | } | |
1059 | if (s2 < 0) | |
1060 | { | |
fa8ed3e0 | 1061 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1062 | s2 = 0; |
1063 | } | |
8bb43c28 RS |
1064 | re_match_object = Qnil; |
1065 | ||
ca1d1d23 JB |
1066 | while (n < 0) |
1067 | { | |
42db823b | 1068 | int val; |
487282dc | 1069 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1070 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1071 | &search_regs, | |
42db823b | 1072 | /* Don't allow match past current point */ |
4996330b | 1073 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1074 | if (val == -2) |
b6d6a51c KH |
1075 | { |
1076 | matcher_overflow (); | |
1077 | } | |
ca1d1d23 JB |
1078 | if (val >= 0) |
1079 | { | |
26aff150 | 1080 | pos_byte = search_regs.start[0] + BEGV_BYTE; |
4746118a | 1081 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1082 | if (search_regs.start[i] >= 0) |
1083 | { | |
fa8ed3e0 RS |
1084 | search_regs.start[i] |
1085 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1086 | search_regs.end[i] | |
1087 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1088 | } |
a3668d92 | 1089 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1090 | /* Set pos to the new position. */ |
1091 | pos = search_regs.start[0]; | |
1092 | } | |
1093 | else | |
1094 | { | |
1095 | immediate_quit = 0; | |
1096 | return (n); | |
1097 | } | |
1098 | n++; | |
1099 | } | |
1100 | while (n > 0) | |
1101 | { | |
42db823b | 1102 | int val; |
487282dc | 1103 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1104 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1105 | &search_regs, | |
1106 | lim_byte - BEGV_BYTE); | |
ca1d1d23 | 1107 | if (val == -2) |
b6d6a51c KH |
1108 | { |
1109 | matcher_overflow (); | |
1110 | } | |
ca1d1d23 JB |
1111 | if (val >= 0) |
1112 | { | |
26aff150 | 1113 | pos_byte = search_regs.end[0] + BEGV_BYTE; |
4746118a | 1114 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1115 | if (search_regs.start[i] >= 0) |
1116 | { | |
fa8ed3e0 RS |
1117 | search_regs.start[i] |
1118 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1119 | search_regs.end[i] | |
1120 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1121 | } |
a3668d92 | 1122 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1123 | pos = search_regs.end[0]; |
1124 | } | |
1125 | else | |
1126 | { | |
1127 | immediate_quit = 0; | |
1128 | return (0 - n); | |
1129 | } | |
1130 | n--; | |
1131 | } | |
1132 | immediate_quit = 0; | |
1133 | return (pos); | |
1134 | } | |
1135 | else /* non-RE case */ | |
1136 | { | |
facdc750 RS |
1137 | unsigned char *raw_pattern, *pat; |
1138 | int raw_pattern_size; | |
1139 | int raw_pattern_size_byte; | |
1140 | unsigned char *patbuf; | |
1141 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); | |
1142 | unsigned char *base_pat = XSTRING (string)->data; | |
1143 | int charset_base = -1; | |
040272ce | 1144 | int boyer_moore_ok = 1; |
facdc750 RS |
1145 | |
1146 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1147 | We must convert PATTERN to match that, or we will not really | |
1148 | find things right. */ | |
1149 | ||
1150 | if (multibyte == STRING_MULTIBYTE (string)) | |
1151 | { | |
7276d3d8 | 1152 | raw_pattern = (unsigned char *) XSTRING (string)->data; |
facdc750 | 1153 | raw_pattern_size = XSTRING (string)->size; |
fc932ac6 | 1154 | raw_pattern_size_byte = STRING_BYTES (XSTRING (string)); |
facdc750 RS |
1155 | } |
1156 | else if (multibyte) | |
1157 | { | |
1158 | raw_pattern_size = XSTRING (string)->size; | |
1159 | raw_pattern_size_byte | |
1160 | = count_size_as_multibyte (XSTRING (string)->data, | |
1161 | raw_pattern_size); | |
7276d3d8 | 1162 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
facdc750 RS |
1163 | copy_text (XSTRING (string)->data, raw_pattern, |
1164 | XSTRING (string)->size, 0, 1); | |
1165 | } | |
1166 | else | |
1167 | { | |
1168 | /* Converting multibyte to single-byte. | |
1169 | ||
1170 | ??? Perhaps this conversion should be done in a special way | |
1171 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1172 | so that only the multibyte chars which really correspond to | |
1173 | the chosen single-byte character set can possibly match. */ | |
1174 | raw_pattern_size = XSTRING (string)->size; | |
1175 | raw_pattern_size_byte = XSTRING (string)->size; | |
7276d3d8 | 1176 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
facdc750 | 1177 | copy_text (XSTRING (string)->data, raw_pattern, |
fc932ac6 | 1178 | STRING_BYTES (XSTRING (string)), 1, 0); |
facdc750 RS |
1179 | } |
1180 | ||
1181 | /* Copy and optionally translate the pattern. */ | |
1182 | len = raw_pattern_size; | |
1183 | len_byte = raw_pattern_size_byte; | |
1184 | patbuf = (unsigned char *) alloca (len_byte); | |
1185 | pat = patbuf; | |
1186 | base_pat = raw_pattern; | |
1187 | if (multibyte) | |
1188 | { | |
1189 | while (--len >= 0) | |
1190 | { | |
daaa6ed8 | 1191 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
aff2ce94 | 1192 | int c, translated, inverse; |
facdc750 RS |
1193 | int in_charlen, charlen; |
1194 | ||
1195 | /* If we got here and the RE flag is set, it's because we're | |
1196 | dealing with a regexp known to be trivial, so the backslash | |
1197 | just quotes the next character. */ | |
1198 | if (RE && *base_pat == '\\') | |
1199 | { | |
1200 | len--; | |
1201 | len_byte--; | |
1202 | base_pat++; | |
1203 | } | |
1204 | ||
1205 | c = STRING_CHAR_AND_LENGTH (base_pat, len_byte, in_charlen); | |
040272ce | 1206 | |
facdc750 | 1207 | /* Translate the character, if requested. */ |
aff2ce94 | 1208 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1209 | /* If translation changed the byte-length, go back |
1210 | to the original character. */ | |
daaa6ed8 | 1211 | charlen = CHAR_STRING (translated, str); |
facdc750 RS |
1212 | if (in_charlen != charlen) |
1213 | { | |
1214 | translated = c; | |
daaa6ed8 | 1215 | charlen = CHAR_STRING (c, str); |
facdc750 RS |
1216 | } |
1217 | ||
5ffaf437 RS |
1218 | /* If we are searching for something strange, |
1219 | an invalid multibyte code, don't use boyer-moore. */ | |
1220 | if (! ASCII_BYTE_P (translated) | |
1221 | && (charlen == 1 /* 8bit code */ | |
1222 | || charlen != in_charlen /* invalid multibyte code */ | |
1223 | )) | |
1224 | boyer_moore_ok = 0; | |
1225 | ||
aff2ce94 RS |
1226 | TRANSLATE (inverse, inverse_trt, c); |
1227 | ||
facdc750 RS |
1228 | /* Did this char actually get translated? |
1229 | Would any other char get translated into it? */ | |
aff2ce94 | 1230 | if (translated != c || inverse != c) |
facdc750 RS |
1231 | { |
1232 | /* Keep track of which character set row | |
1233 | contains the characters that need translation. */ | |
5ffaf437 | 1234 | int charset_base_code = c & ~CHAR_FIELD3_MASK; |
facdc750 RS |
1235 | if (charset_base == -1) |
1236 | charset_base = charset_base_code; | |
1237 | else if (charset_base != charset_base_code) | |
1238 | /* If two different rows appear, needing translation, | |
1239 | then we cannot use boyer_moore search. */ | |
040272ce | 1240 | boyer_moore_ok = 0; |
aff2ce94 | 1241 | } |
facdc750 RS |
1242 | |
1243 | /* Store this character into the translated pattern. */ | |
1244 | bcopy (str, pat, charlen); | |
1245 | pat += charlen; | |
1246 | base_pat += in_charlen; | |
1247 | len_byte -= in_charlen; | |
1248 | } | |
1249 | } | |
1250 | else | |
1251 | { | |
040272ce KH |
1252 | /* Unibyte buffer. */ |
1253 | charset_base = 0; | |
facdc750 RS |
1254 | while (--len >= 0) |
1255 | { | |
040272ce | 1256 | int c, translated; |
facdc750 RS |
1257 | |
1258 | /* If we got here and the RE flag is set, it's because we're | |
1259 | dealing with a regexp known to be trivial, so the backslash | |
1260 | just quotes the next character. */ | |
1261 | if (RE && *base_pat == '\\') | |
1262 | { | |
1263 | len--; | |
1264 | base_pat++; | |
1265 | } | |
1266 | c = *base_pat++; | |
aff2ce94 | 1267 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1268 | *pat++ = translated; |
1269 | } | |
1270 | } | |
1271 | ||
1272 | len_byte = pat - patbuf; | |
1273 | len = raw_pattern_size; | |
1274 | pat = base_pat = patbuf; | |
1275 | ||
040272ce | 1276 | if (boyer_moore_ok) |
facdc750 | 1277 | return boyer_moore (n, pat, len, len_byte, trt, inverse_trt, |
aff2ce94 RS |
1278 | pos, pos_byte, lim, lim_byte, |
1279 | charset_base); | |
facdc750 RS |
1280 | else |
1281 | return simple_search (n, pat, len, len_byte, trt, | |
1282 | pos, pos_byte, lim, lim_byte); | |
1283 | } | |
1284 | } | |
1285 | \f | |
1286 | /* Do a simple string search N times for the string PAT, | |
1287 | whose length is LEN/LEN_BYTE, | |
1288 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1289 | TRT is the translation table. | |
f8bd51c4 | 1290 | |
facdc750 RS |
1291 | Return the character position where the match is found. |
1292 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1293 | |
facdc750 RS |
1294 | This kind of search works regardless of what is in PAT and |
1295 | regardless of what is in TRT. It is used in cases where | |
1296 | boyer_moore cannot work. */ | |
1297 | ||
1298 | static int | |
1299 | simple_search (n, pat, len, len_byte, trt, pos, pos_byte, lim, lim_byte) | |
1300 | int n; | |
1301 | unsigned char *pat; | |
1302 | int len, len_byte; | |
1303 | Lisp_Object trt; | |
1304 | int pos, pos_byte; | |
1305 | int lim, lim_byte; | |
1306 | { | |
1307 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); | |
ab228c24 | 1308 | int forward = n > 0; |
facdc750 RS |
1309 | |
1310 | if (lim > pos && multibyte) | |
1311 | while (n > 0) | |
1312 | { | |
1313 | while (1) | |
f8bd51c4 | 1314 | { |
facdc750 RS |
1315 | /* Try matching at position POS. */ |
1316 | int this_pos = pos; | |
1317 | int this_pos_byte = pos_byte; | |
1318 | int this_len = len; | |
1319 | int this_len_byte = len_byte; | |
1320 | unsigned char *p = pat; | |
1321 | if (pos + len > lim) | |
1322 | goto stop; | |
1323 | ||
1324 | while (this_len > 0) | |
1325 | { | |
1326 | int charlen, buf_charlen; | |
ab228c24 | 1327 | int pat_ch, buf_ch; |
facdc750 | 1328 | |
ab228c24 | 1329 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1330 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1331 | ZV_BYTE - this_pos_byte, | |
1332 | buf_charlen); | |
aff2ce94 | 1333 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1334 | |
1335 | if (buf_ch != pat_ch) | |
1336 | break; | |
ab228c24 RS |
1337 | |
1338 | this_len_byte -= charlen; | |
1339 | this_len--; | |
1340 | p += charlen; | |
1341 | ||
1342 | this_pos_byte += buf_charlen; | |
1343 | this_pos++; | |
facdc750 RS |
1344 | } |
1345 | ||
1346 | if (this_len == 0) | |
1347 | { | |
1348 | pos += len; | |
1349 | pos_byte += len_byte; | |
1350 | break; | |
1351 | } | |
1352 | ||
1353 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1354 | } |
facdc750 RS |
1355 | |
1356 | n--; | |
1357 | } | |
1358 | else if (lim > pos) | |
1359 | while (n > 0) | |
1360 | { | |
1361 | while (1) | |
f8bd51c4 | 1362 | { |
facdc750 RS |
1363 | /* Try matching at position POS. */ |
1364 | int this_pos = pos; | |
1365 | int this_len = len; | |
1366 | unsigned char *p = pat; | |
1367 | ||
1368 | if (pos + len > lim) | |
1369 | goto stop; | |
1370 | ||
1371 | while (this_len > 0) | |
1372 | { | |
1373 | int pat_ch = *p++; | |
1374 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1375 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1376 | |
1377 | if (buf_ch != pat_ch) | |
1378 | break; | |
ab228c24 RS |
1379 | |
1380 | this_len--; | |
1381 | this_pos++; | |
facdc750 RS |
1382 | } |
1383 | ||
1384 | if (this_len == 0) | |
1385 | { | |
1386 | pos += len; | |
1387 | break; | |
1388 | } | |
1389 | ||
1390 | pos++; | |
f8bd51c4 | 1391 | } |
facdc750 RS |
1392 | |
1393 | n--; | |
1394 | } | |
1395 | /* Backwards search. */ | |
1396 | else if (lim < pos && multibyte) | |
1397 | while (n < 0) | |
1398 | { | |
1399 | while (1) | |
f8bd51c4 | 1400 | { |
facdc750 RS |
1401 | /* Try matching at position POS. */ |
1402 | int this_pos = pos - len; | |
1403 | int this_pos_byte = pos_byte - len_byte; | |
1404 | int this_len = len; | |
1405 | int this_len_byte = len_byte; | |
1406 | unsigned char *p = pat; | |
1407 | ||
1408 | if (pos - len < lim) | |
1409 | goto stop; | |
1410 | ||
1411 | while (this_len > 0) | |
1412 | { | |
1413 | int charlen, buf_charlen; | |
ab228c24 | 1414 | int pat_ch, buf_ch; |
facdc750 | 1415 | |
ab228c24 | 1416 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1417 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1418 | ZV_BYTE - this_pos_byte, | |
1419 | buf_charlen); | |
aff2ce94 | 1420 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1421 | |
1422 | if (buf_ch != pat_ch) | |
1423 | break; | |
ab228c24 RS |
1424 | |
1425 | this_len_byte -= charlen; | |
1426 | this_len--; | |
1427 | p += charlen; | |
1428 | this_pos_byte += buf_charlen; | |
1429 | this_pos++; | |
facdc750 RS |
1430 | } |
1431 | ||
1432 | if (this_len == 0) | |
1433 | { | |
1434 | pos -= len; | |
1435 | pos_byte -= len_byte; | |
1436 | break; | |
1437 | } | |
1438 | ||
1439 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1440 | } |
1441 | ||
facdc750 RS |
1442 | n++; |
1443 | } | |
1444 | else if (lim < pos) | |
1445 | while (n < 0) | |
1446 | { | |
1447 | while (1) | |
b6d6a51c | 1448 | { |
facdc750 RS |
1449 | /* Try matching at position POS. */ |
1450 | int this_pos = pos - len; | |
1451 | int this_len = len; | |
1452 | unsigned char *p = pat; | |
1453 | ||
1454 | if (pos - len < lim) | |
1455 | goto stop; | |
1456 | ||
1457 | while (this_len > 0) | |
1458 | { | |
1459 | int pat_ch = *p++; | |
1460 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1461 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1462 | |
1463 | if (buf_ch != pat_ch) | |
1464 | break; | |
ab228c24 RS |
1465 | this_len--; |
1466 | this_pos++; | |
facdc750 RS |
1467 | } |
1468 | ||
1469 | if (this_len == 0) | |
b6d6a51c | 1470 | { |
facdc750 RS |
1471 | pos -= len; |
1472 | break; | |
b6d6a51c | 1473 | } |
facdc750 RS |
1474 | |
1475 | pos--; | |
b6d6a51c | 1476 | } |
facdc750 RS |
1477 | |
1478 | n++; | |
b6d6a51c | 1479 | } |
facdc750 RS |
1480 | |
1481 | stop: | |
1482 | if (n == 0) | |
aff2ce94 | 1483 | { |
ab228c24 RS |
1484 | if (forward) |
1485 | set_search_regs ((multibyte ? pos_byte : pos) - len_byte, len_byte); | |
1486 | else | |
1487 | set_search_regs (multibyte ? pos_byte : pos, len_byte); | |
aff2ce94 RS |
1488 | |
1489 | return pos; | |
1490 | } | |
facdc750 RS |
1491 | else if (n > 0) |
1492 | return -n; | |
1493 | else | |
1494 | return n; | |
1495 | } | |
1496 | \f | |
1497 | /* Do Boyer-Moore search N times for the string PAT, | |
1498 | whose length is LEN/LEN_BYTE, | |
1499 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1500 | DIRECTION says which direction we search in. | |
1501 | TRT and INVERSE_TRT are translation tables. | |
1502 | ||
1503 | This kind of search works if all the characters in PAT that have | |
1504 | nontrivial translation are the same aside from the last byte. This | |
1505 | makes it possible to translate just the last byte of a character, | |
1506 | and do so after just a simple test of the context. | |
1507 | ||
1508 | If that criterion is not satisfied, do not call this function. */ | |
1509 | ||
1510 | static int | |
1511 | boyer_moore (n, base_pat, len, len_byte, trt, inverse_trt, | |
aff2ce94 | 1512 | pos, pos_byte, lim, lim_byte, charset_base) |
facdc750 RS |
1513 | int n; |
1514 | unsigned char *base_pat; | |
1515 | int len, len_byte; | |
1516 | Lisp_Object trt; | |
1517 | Lisp_Object inverse_trt; | |
1518 | int pos, pos_byte; | |
1519 | int lim, lim_byte; | |
aff2ce94 | 1520 | int charset_base; |
facdc750 RS |
1521 | { |
1522 | int direction = ((n > 0) ? 1 : -1); | |
1523 | register int dirlen; | |
1524 | int infinity, limit, k, stride_for_teases; | |
1525 | register int *BM_tab; | |
1526 | int *BM_tab_base; | |
1527 | register unsigned char *cursor, *p_limit; | |
1528 | register int i, j; | |
cb6792d2 | 1529 | unsigned char *pat, *pat_end; |
facdc750 RS |
1530 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); |
1531 | ||
1532 | unsigned char simple_translate[0400]; | |
1533 | int translate_prev_byte; | |
1534 | int translate_anteprev_byte; | |
1535 | ||
1536 | #ifdef C_ALLOCA | |
1537 | int BM_tab_space[0400]; | |
1538 | BM_tab = &BM_tab_space[0]; | |
1539 | #else | |
1540 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
1541 | #endif | |
1542 | /* The general approach is that we are going to maintain that we know */ | |
1543 | /* the first (closest to the present position, in whatever direction */ | |
1544 | /* we're searching) character that could possibly be the last */ | |
1545 | /* (furthest from present position) character of a valid match. We */ | |
1546 | /* advance the state of our knowledge by looking at that character */ | |
1547 | /* and seeing whether it indeed matches the last character of the */ | |
1548 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1549 | /* move our pointer (to putative last characters) as far as is */ | |
1550 | /* logically possible. This amount of movement, which I call a */ | |
1551 | /* stride, will be the length of the pattern if the actual character */ | |
1552 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1553 | /* from the last occurrence of that character to the end of the */ | |
1554 | /* pattern. */ | |
1555 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1556 | /* characters that match the last character. This allows use of only */ | |
1557 | /* a single test, a test for having gone past the end of the */ | |
1558 | /* permissible match region, to test for both possible matches (when */ | |
1559 | /* the stride goes past the end immediately) and failure to */ | |
1560 | /* match (where you get nudged past the end one stride at a time). */ | |
1561 | ||
1562 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1563 | /* is determined only by the last character of the putative match. */ | |
1564 | /* If that character does not match, we will stride the proper */ | |
1565 | /* distance to propose a match that superimposes it on the last */ | |
1566 | /* instance of a character that matches it (per trt), or misses */ | |
1567 | /* it entirely if there is none. */ | |
1568 | ||
1569 | dirlen = len_byte * direction; | |
1570 | infinity = dirlen - (lim_byte + pos_byte + len_byte + len_byte) * direction; | |
cb6792d2 RS |
1571 | |
1572 | /* Record position after the end of the pattern. */ | |
1573 | pat_end = base_pat + len_byte; | |
1574 | /* BASE_PAT points to a character that we start scanning from. | |
1575 | It is the first character in a forward search, | |
1576 | the last character in a backward search. */ | |
facdc750 | 1577 | if (direction < 0) |
cb6792d2 RS |
1578 | base_pat = pat_end - 1; |
1579 | ||
facdc750 RS |
1580 | BM_tab_base = BM_tab; |
1581 | BM_tab += 0400; | |
1582 | j = dirlen; /* to get it in a register */ | |
1583 | /* A character that does not appear in the pattern induces a */ | |
1584 | /* stride equal to the pattern length. */ | |
1585 | while (BM_tab_base != BM_tab) | |
1586 | { | |
1587 | *--BM_tab = j; | |
1588 | *--BM_tab = j; | |
1589 | *--BM_tab = j; | |
1590 | *--BM_tab = j; | |
1591 | } | |
1592 | ||
1593 | /* We use this for translation, instead of TRT itself. | |
1594 | We fill this in to handle the characters that actually | |
1595 | occur in the pattern. Others don't matter anyway! */ | |
1596 | bzero (simple_translate, sizeof simple_translate); | |
1597 | for (i = 0; i < 0400; i++) | |
1598 | simple_translate[i] = i; | |
1599 | ||
1600 | i = 0; | |
1601 | while (i != infinity) | |
1602 | { | |
cb6792d2 | 1603 | unsigned char *ptr = base_pat + i; |
facdc750 RS |
1604 | i += direction; |
1605 | if (i == dirlen) | |
1606 | i = infinity; | |
1607 | if (! NILP (trt)) | |
ca1d1d23 | 1608 | { |
facdc750 | 1609 | int ch; |
aff2ce94 | 1610 | int untranslated; |
facdc750 RS |
1611 | int this_translated = 1; |
1612 | ||
1613 | if (multibyte | |
cb6792d2 RS |
1614 | /* Is *PTR the last byte of a character? */ |
1615 | && (pat_end - ptr == 1 || CHAR_HEAD_P (ptr[1]))) | |
ca1d1d23 | 1616 | { |
facdc750 RS |
1617 | unsigned char *charstart = ptr; |
1618 | while (! CHAR_HEAD_P (*charstart)) | |
1619 | charstart--; | |
aff2ce94 | 1620 | untranslated = STRING_CHAR (charstart, ptr - charstart + 1); |
6397418a | 1621 | if (charset_base == (untranslated & ~CHAR_FIELD3_MASK)) |
facdc750 | 1622 | { |
ab228c24 | 1623 | TRANSLATE (ch, trt, untranslated); |
aff2ce94 RS |
1624 | if (! CHAR_HEAD_P (*ptr)) |
1625 | { | |
1626 | translate_prev_byte = ptr[-1]; | |
1627 | if (! CHAR_HEAD_P (translate_prev_byte)) | |
1628 | translate_anteprev_byte = ptr[-2]; | |
1629 | } | |
facdc750 | 1630 | } |
aff2ce94 | 1631 | else |
ab228c24 RS |
1632 | { |
1633 | this_translated = 0; | |
1634 | ch = *ptr; | |
1635 | } | |
ca1d1d23 | 1636 | } |
facdc750 | 1637 | else if (!multibyte) |
aff2ce94 | 1638 | TRANSLATE (ch, trt, *ptr); |
ca1d1d23 JB |
1639 | else |
1640 | { | |
facdc750 RS |
1641 | ch = *ptr; |
1642 | this_translated = 0; | |
ca1d1d23 | 1643 | } |
facdc750 | 1644 | |
ab228c24 RS |
1645 | if (ch > 0400) |
1646 | j = ((unsigned char) ch) | 0200; | |
1647 | else | |
1648 | j = (unsigned char) ch; | |
1649 | ||
facdc750 RS |
1650 | if (i == infinity) |
1651 | stride_for_teases = BM_tab[j]; | |
ab228c24 | 1652 | |
facdc750 RS |
1653 | BM_tab[j] = dirlen - i; |
1654 | /* A translation table is accompanied by its inverse -- see */ | |
1655 | /* comment following downcase_table for details */ | |
1656 | if (this_translated) | |
ab228c24 RS |
1657 | { |
1658 | int starting_ch = ch; | |
1659 | int starting_j = j; | |
1660 | while (1) | |
1661 | { | |
1662 | TRANSLATE (ch, inverse_trt, ch); | |
1663 | if (ch > 0400) | |
1664 | j = ((unsigned char) ch) | 0200; | |
1665 | else | |
1666 | j = (unsigned char) ch; | |
1667 | ||
1668 | /* For all the characters that map into CH, | |
1669 | set up simple_translate to map the last byte | |
1670 | into STARTING_J. */ | |
1671 | simple_translate[j] = starting_j; | |
1672 | if (ch == starting_ch) | |
1673 | break; | |
1674 | BM_tab[j] = dirlen - i; | |
1675 | } | |
1676 | } | |
facdc750 RS |
1677 | } |
1678 | else | |
1679 | { | |
1680 | j = *ptr; | |
1681 | ||
1682 | if (i == infinity) | |
1683 | stride_for_teases = BM_tab[j]; | |
1684 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1685 | } |
facdc750 RS |
1686 | /* stride_for_teases tells how much to stride if we get a */ |
1687 | /* match on the far character but are subsequently */ | |
1688 | /* disappointed, by recording what the stride would have been */ | |
1689 | /* for that character if the last character had been */ | |
1690 | /* different. */ | |
1691 | } | |
1692 | infinity = dirlen - infinity; | |
1693 | pos_byte += dirlen - ((direction > 0) ? direction : 0); | |
1694 | /* loop invariant - POS_BYTE points at where last char (first | |
1695 | char if reverse) of pattern would align in a possible match. */ | |
1696 | while (n != 0) | |
1697 | { | |
1698 | int tail_end; | |
1699 | unsigned char *tail_end_ptr; | |
1700 | ||
1701 | /* It's been reported that some (broken) compiler thinks that | |
1702 | Boolean expressions in an arithmetic context are unsigned. | |
1703 | Using an explicit ?1:0 prevents this. */ | |
1704 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1705 | < 0) | |
1706 | return (n * (0 - direction)); | |
1707 | /* First we do the part we can by pointers (maybe nothing) */ | |
1708 | QUIT; | |
1709 | pat = base_pat; | |
1710 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1711 | if (direction > 0) |
1712 | { | |
1713 | limit = BUFFER_CEILING_OF (limit); | |
1714 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1715 | can take on without hitting edge of buffer or the gap. */ | |
1716 | limit = min (limit, pos_byte + 20000); | |
1717 | limit = min (limit, lim_byte - 1); | |
1718 | } | |
1719 | else | |
1720 | { | |
1721 | limit = BUFFER_FLOOR_OF (limit); | |
1722 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1723 | can take on without hitting edge of buffer or the gap. */ | |
1724 | limit = max (limit, pos_byte - 20000); | |
1725 | limit = max (limit, lim_byte); | |
1726 | } | |
facdc750 RS |
1727 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1728 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1729 | ||
1730 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1731 | { |
facdc750 RS |
1732 | unsigned char *p2; |
1733 | ||
1734 | p_limit = BYTE_POS_ADDR (limit); | |
1735 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
1736 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ | |
1737 | while (1) /* use one cursor setting as long as i can */ | |
ca1d1d23 | 1738 | { |
facdc750 | 1739 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1740 | { |
facdc750 RS |
1741 | /* Use signed comparison if appropriate |
1742 | to make cursor+infinity sure to be > p_limit. | |
1743 | Assuming that the buffer lies in a range of addresses | |
1744 | that are all "positive" (as ints) or all "negative", | |
1745 | either kind of comparison will work as long | |
1746 | as we don't step by infinity. So pick the kind | |
1747 | that works when we do step by infinity. */ | |
1748 | if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit) | |
1749 | while ((EMACS_INT) cursor <= (EMACS_INT) p_limit) | |
1750 | cursor += BM_tab[*cursor]; | |
ca1d1d23 | 1751 | else |
facdc750 RS |
1752 | while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit) |
1753 | cursor += BM_tab[*cursor]; | |
1754 | } | |
1755 | else | |
1756 | { | |
1757 | if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit) | |
1758 | while ((EMACS_INT) cursor >= (EMACS_INT) p_limit) | |
1759 | cursor += BM_tab[*cursor]; | |
1760 | else | |
1761 | while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit) | |
1762 | cursor += BM_tab[*cursor]; | |
1763 | } | |
ca1d1d23 | 1764 | /* If you are here, cursor is beyond the end of the searched region. */ |
facdc750 RS |
1765 | /* This can happen if you match on the far character of the pattern, */ |
1766 | /* because the "stride" of that character is infinity, a number able */ | |
1767 | /* to throw you well beyond the end of the search. It can also */ | |
1768 | /* happen if you fail to match within the permitted region and would */ | |
1769 | /* otherwise try a character beyond that region */ | |
1770 | if ((cursor - p_limit) * direction <= len_byte) | |
1771 | break; /* a small overrun is genuine */ | |
1772 | cursor -= infinity; /* large overrun = hit */ | |
1773 | i = dirlen - direction; | |
1774 | if (! NILP (trt)) | |
1775 | { | |
1776 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1777 | { |
facdc750 RS |
1778 | int ch; |
1779 | cursor -= direction; | |
1780 | /* Translate only the last byte of a character. */ | |
1781 | if (! multibyte | |
1782 | || ((cursor == tail_end_ptr | |
1783 | || CHAR_HEAD_P (cursor[1])) | |
1784 | && (CHAR_HEAD_P (cursor[0]) | |
1785 | || (translate_prev_byte == cursor[-1] | |
1786 | && (CHAR_HEAD_P (translate_prev_byte) | |
1787 | || translate_anteprev_byte == cursor[-2]))))) | |
1788 | ch = simple_translate[*cursor]; | |
1789 | else | |
1790 | ch = *cursor; | |
1791 | if (pat[i] != ch) | |
1792 | break; | |
ca1d1d23 | 1793 | } |
facdc750 RS |
1794 | } |
1795 | else | |
1796 | { | |
1797 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1798 | { |
facdc750 RS |
1799 | cursor -= direction; |
1800 | if (pat[i] != *cursor) | |
1801 | break; | |
ca1d1d23 | 1802 | } |
facdc750 RS |
1803 | } |
1804 | cursor += dirlen - i - direction; /* fix cursor */ | |
1805 | if (i + direction == 0) | |
1806 | { | |
1807 | int position; | |
0c8533c6 | 1808 | |
facdc750 | 1809 | cursor -= direction; |
1113d9db | 1810 | |
facdc750 RS |
1811 | position = pos_byte + cursor - p2 + ((direction > 0) |
1812 | ? 1 - len_byte : 0); | |
1813 | set_search_regs (position, len_byte); | |
ca325161 | 1814 | |
facdc750 RS |
1815 | if ((n -= direction) != 0) |
1816 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1817 | else |
facdc750 RS |
1818 | return ((direction > 0) |
1819 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1820 | } |
facdc750 RS |
1821 | else |
1822 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1823 | } |
facdc750 RS |
1824 | pos_byte += cursor - p2; |
1825 | } | |
1826 | else | |
1827 | /* Now we'll pick up a clump that has to be done the hard */ | |
1828 | /* way because it covers a discontinuity */ | |
1829 | { | |
1830 | limit = ((direction > 0) | |
1831 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1832 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1833 | limit = ((direction > 0) | |
1834 | ? min (limit + len_byte, lim_byte - 1) | |
1835 | : max (limit - len_byte, lim_byte)); | |
1836 | /* LIMIT is now the last value POS_BYTE can have | |
1837 | and still be valid for a possible match. */ | |
1838 | while (1) | |
ca1d1d23 | 1839 | { |
facdc750 RS |
1840 | /* This loop can be coded for space rather than */ |
1841 | /* speed because it will usually run only once. */ | |
1842 | /* (the reach is at most len + 21, and typically */ | |
1843 | /* does not exceed len) */ | |
1844 | while ((limit - pos_byte) * direction >= 0) | |
1845 | pos_byte += BM_tab[FETCH_BYTE (pos_byte)]; | |
1846 | /* now run the same tests to distinguish going off the */ | |
1847 | /* end, a match or a phony match. */ | |
1848 | if ((pos_byte - limit) * direction <= len_byte) | |
1849 | break; /* ran off the end */ | |
1850 | /* Found what might be a match. | |
1851 | Set POS_BYTE back to last (first if reverse) pos. */ | |
1852 | pos_byte -= infinity; | |
1853 | i = dirlen - direction; | |
1854 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1855 | { |
facdc750 RS |
1856 | int ch; |
1857 | unsigned char *ptr; | |
1858 | pos_byte -= direction; | |
1859 | ptr = BYTE_POS_ADDR (pos_byte); | |
1860 | /* Translate only the last byte of a character. */ | |
1861 | if (! multibyte | |
1862 | || ((ptr == tail_end_ptr | |
1863 | || CHAR_HEAD_P (ptr[1])) | |
1864 | && (CHAR_HEAD_P (ptr[0]) | |
1865 | || (translate_prev_byte == ptr[-1] | |
1866 | && (CHAR_HEAD_P (translate_prev_byte) | |
1867 | || translate_anteprev_byte == ptr[-2]))))) | |
1868 | ch = simple_translate[*ptr]; | |
1869 | else | |
1870 | ch = *ptr; | |
1871 | if (pat[i] != ch) | |
1872 | break; | |
1873 | } | |
1874 | /* Above loop has moved POS_BYTE part or all the way | |
1875 | back to the first pos (last pos if reverse). | |
1876 | Set it once again at the last (first if reverse) char. */ | |
1877 | pos_byte += dirlen - i- direction; | |
1878 | if (i + direction == 0) | |
1879 | { | |
1880 | int position; | |
1881 | pos_byte -= direction; | |
1113d9db | 1882 | |
facdc750 | 1883 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
0c8533c6 | 1884 | |
facdc750 | 1885 | set_search_regs (position, len_byte); |
ca325161 | 1886 | |
facdc750 RS |
1887 | if ((n -= direction) != 0) |
1888 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 1889 | else |
facdc750 RS |
1890 | return ((direction > 0) |
1891 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1892 | } |
facdc750 RS |
1893 | else |
1894 | pos_byte += stride_for_teases; | |
1895 | } | |
1896 | } | |
1897 | /* We have done one clump. Can we continue? */ | |
1898 | if ((lim_byte - pos_byte) * direction < 0) | |
1899 | return ((0 - n) * direction); | |
ca1d1d23 | 1900 | } |
facdc750 | 1901 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 1902 | } |
ca325161 | 1903 | |
fa8ed3e0 | 1904 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
1905 | for the overall match just found in the current buffer. |
1906 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
1907 | |
1908 | static void | |
fa8ed3e0 RS |
1909 | set_search_regs (beg_byte, nbytes) |
1910 | int beg_byte, nbytes; | |
ca325161 | 1911 | { |
a7e4cdde RS |
1912 | int i; |
1913 | ||
ca325161 RS |
1914 | /* Make sure we have registers in which to store |
1915 | the match position. */ | |
1916 | if (search_regs.num_regs == 0) | |
1917 | { | |
2d4a771a RS |
1918 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
1919 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 1920 | search_regs.num_regs = 2; |
ca325161 RS |
1921 | } |
1922 | ||
a7e4cdde RS |
1923 | /* Clear out the other registers. */ |
1924 | for (i = 1; i < search_regs.num_regs; i++) | |
1925 | { | |
1926 | search_regs.start[i] = -1; | |
1927 | search_regs.end[i] = -1; | |
1928 | } | |
1929 | ||
fa8ed3e0 RS |
1930 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
1931 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 1932 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 1933 | } |
ca1d1d23 JB |
1934 | \f |
1935 | /* Given a string of words separated by word delimiters, | |
1936 | compute a regexp that matches those exact words | |
1937 | separated by arbitrary punctuation. */ | |
1938 | ||
1939 | static Lisp_Object | |
1940 | wordify (string) | |
1941 | Lisp_Object string; | |
1942 | { | |
1943 | register unsigned char *p, *o; | |
0c8533c6 | 1944 | register int i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 1945 | Lisp_Object val; |
0c8533c6 RS |
1946 | int prev_c = 0; |
1947 | int adjust; | |
ca1d1d23 JB |
1948 | |
1949 | CHECK_STRING (string, 0); | |
1950 | p = XSTRING (string)->data; | |
1951 | len = XSTRING (string)->size; | |
1952 | ||
0c8533c6 RS |
1953 | for (i = 0, i_byte = 0; i < len; ) |
1954 | { | |
1955 | int c; | |
1956 | ||
1957 | if (STRING_MULTIBYTE (string)) | |
1958 | FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte); | |
1959 | else | |
1960 | c = XSTRING (string)->data[i++]; | |
1961 | ||
1962 | if (SYNTAX (c) != Sword) | |
1963 | { | |
1964 | punct_count++; | |
1965 | if (i > 0 && SYNTAX (prev_c) == Sword) | |
1966 | word_count++; | |
1967 | } | |
ca1d1d23 | 1968 | |
0c8533c6 RS |
1969 | prev_c = c; |
1970 | } | |
1971 | ||
1972 | if (SYNTAX (prev_c) == Sword) | |
1973 | word_count++; | |
1974 | if (!word_count) | |
1975 | return build_string (""); | |
1976 | ||
1977 | adjust = - punct_count + 5 * (word_count - 1) + 4; | |
8a2df937 RS |
1978 | if (STRING_MULTIBYTE (string)) |
1979 | val = make_uninit_multibyte_string (len + adjust, | |
1980 | STRING_BYTES (XSTRING (string)) | |
1981 | + adjust); | |
1982 | else | |
1983 | val = make_uninit_string (len + adjust); | |
ca1d1d23 JB |
1984 | |
1985 | o = XSTRING (val)->data; | |
1986 | *o++ = '\\'; | |
1987 | *o++ = 'b'; | |
1e9582d4 | 1988 | prev_c = 0; |
ca1d1d23 | 1989 | |
1e9582d4 RS |
1990 | for (i = 0, i_byte = 0; i < len; ) |
1991 | { | |
1992 | int c; | |
1993 | int i_byte_orig = i_byte; | |
1994 | ||
1995 | if (STRING_MULTIBYTE (string)) | |
1996 | FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte); | |
1997 | else | |
8a2df937 RS |
1998 | { |
1999 | c = XSTRING (string)->data[i++]; | |
2000 | i_byte++; | |
2001 | } | |
1e9582d4 RS |
2002 | |
2003 | if (SYNTAX (c) == Sword) | |
2004 | { | |
2005 | bcopy (&XSTRING (string)->data[i_byte_orig], o, | |
2006 | i_byte - i_byte_orig); | |
2007 | o += i_byte - i_byte_orig; | |
2008 | } | |
2009 | else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count) | |
2010 | { | |
2011 | *o++ = '\\'; | |
2012 | *o++ = 'W'; | |
2013 | *o++ = '\\'; | |
2014 | *o++ = 'W'; | |
2015 | *o++ = '*'; | |
2016 | } | |
2017 | ||
2018 | prev_c = c; | |
2019 | } | |
ca1d1d23 JB |
2020 | |
2021 | *o++ = '\\'; | |
2022 | *o++ = 'b'; | |
2023 | ||
2024 | return val; | |
2025 | } | |
2026 | \f | |
2027 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
6af43974 | 2028 | "MSearch backward: ", |
ca1d1d23 JB |
2029 | "Search backward from point for STRING.\n\ |
2030 | Set point to the beginning of the occurrence found, and return point.\n\ | |
2031 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2032 | The match found must not extend before that position.\n\ | |
2033 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2034 | If not nil and not t, position at limit of search and return nil.\n\ | |
2035 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
2036 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
2037 | (string, bound, noerror, count) | |
2038 | Lisp_Object string, bound, noerror, count; | |
2039 | { | |
b819a390 | 2040 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2041 | } |
2042 | ||
6af43974 | 2043 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
ca1d1d23 JB |
2044 | "Search forward from point for STRING.\n\ |
2045 | Set point to the end of the occurrence found, and return point.\n\ | |
2046 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2047 | The match found must not extend after that position. nil is equivalent\n\ | |
2048 | to (point-max).\n\ | |
2049 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2050 | If not nil and not t, move to limit of search and return nil.\n\ | |
2051 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
2052 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
2053 | (string, bound, noerror, count) | |
2054 | Lisp_Object string, bound, noerror, count; | |
2055 | { | |
b819a390 | 2056 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2057 | } |
2058 | ||
2059 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
2060 | "sWord search backward: ", | |
2061 | "Search backward from point for STRING, ignoring differences in punctuation.\n\ | |
2062 | Set point to the beginning of the occurrence found, and return point.\n\ | |
2063 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2064 | The match found must not extend before that position.\n\ | |
2065 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2066 | If not nil and not t, move to limit of search and return nil.\n\ | |
2067 | Optional fourth argument is repeat count--search for successive occurrences.") | |
2068 | (string, bound, noerror, count) | |
2069 | Lisp_Object string, bound, noerror, count; | |
2070 | { | |
b819a390 | 2071 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2072 | } |
2073 | ||
2074 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
2075 | "sWord search: ", | |
2076 | "Search forward from point for STRING, ignoring differences in punctuation.\n\ | |
2077 | Set point to the end of the occurrence found, and return point.\n\ | |
2078 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2079 | The match found must not extend after that position.\n\ | |
2080 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2081 | If not nil and not t, move to limit of search and return nil.\n\ | |
2082 | Optional fourth argument is repeat count--search for successive occurrences.") | |
2083 | (string, bound, noerror, count) | |
2084 | Lisp_Object string, bound, noerror, count; | |
2085 | { | |
b819a390 | 2086 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
2087 | } |
2088 | ||
2089 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
2090 | "sRE search backward: ", | |
2091 | "Search backward from point for match for regular expression REGEXP.\n\ | |
2092 | Set point to the beginning of the match, and return point.\n\ | |
2093 | The match found is the one starting last in the buffer\n\ | |
19c0a730 | 2094 | and yet ending before the origin of the search.\n\ |
ca1d1d23 JB |
2095 | An optional second argument bounds the search; it is a buffer position.\n\ |
2096 | The match found must start at or after that position.\n\ | |
2097 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2098 | If not nil and not t, move to limit of search and return nil.\n\ | |
2099 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
2100 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
19c0a730 KH |
2101 | (regexp, bound, noerror, count) |
2102 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 2103 | { |
b819a390 | 2104 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2105 | } |
2106 | ||
2107 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
2108 | "sRE search: ", | |
2109 | "Search forward from point for regular expression REGEXP.\n\ | |
2110 | Set point to the end of the occurrence found, and return point.\n\ | |
2111 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2112 | The match found must not extend after that position.\n\ | |
2113 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2114 | If not nil and not t, move to limit of search and return nil.\n\ | |
2115 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
2116 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
19c0a730 KH |
2117 | (regexp, bound, noerror, count) |
2118 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 2119 | { |
b819a390 RS |
2120 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2121 | } | |
2122 | ||
2123 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
2124 | "sPosix search backward: ", | |
2125 | "Search backward from point for match for regular expression REGEXP.\n\ | |
2126 | Find the longest match in accord with Posix regular expression rules.\n\ | |
2127 | Set point to the beginning of the match, and return point.\n\ | |
2128 | The match found is the one starting last in the buffer\n\ | |
2129 | and yet ending before the origin of the search.\n\ | |
2130 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2131 | The match found must start at or after that position.\n\ | |
2132 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2133 | If not nil and not t, move to limit of search and return nil.\n\ | |
2134 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
2135 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
2136 | (regexp, bound, noerror, count) | |
2137 | Lisp_Object regexp, bound, noerror, count; | |
2138 | { | |
2139 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2140 | } | |
2141 | ||
2142 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
2143 | "sPosix search: ", | |
2144 | "Search forward from point for regular expression REGEXP.\n\ | |
2145 | Find the longest match in accord with Posix regular expression rules.\n\ | |
2146 | Set point to the end of the occurrence found, and return point.\n\ | |
2147 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2148 | The match found must not extend after that position.\n\ | |
2149 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2150 | If not nil and not t, move to limit of search and return nil.\n\ | |
2151 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
2152 | See also the functions `match-beginning', `match-end' and `replace-match'.") | |
2153 | (regexp, bound, noerror, count) | |
2154 | Lisp_Object regexp, bound, noerror, count; | |
2155 | { | |
2156 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2157 | } |
2158 | \f | |
d7a5ad5f | 2159 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
ca1d1d23 JB |
2160 | "Replace text matched by last search with NEWTEXT.\n\ |
2161 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text.\n\ | |
5b9cf4b2 RS |
2162 | Otherwise maybe capitalize the whole text, or maybe just word initials,\n\ |
2163 | based on the replaced text.\n\ | |
2164 | If the replaced text has only capital letters\n\ | |
2165 | and has at least one multiletter word, convert NEWTEXT to all caps.\n\ | |
2166 | If the replaced text has at least one word starting with a capital letter,\n\ | |
2167 | then capitalize each word in NEWTEXT.\n\n\ | |
ca1d1d23 JB |
2168 | If third arg LITERAL is non-nil, insert NEWTEXT literally.\n\ |
2169 | Otherwise treat `\\' as special:\n\ | |
2170 | `\\&' in NEWTEXT means substitute original matched text.\n\ | |
2171 | `\\N' means substitute what matched the Nth `\\(...\\)'.\n\ | |
2172 | If Nth parens didn't match, substitute nothing.\n\ | |
2173 | `\\\\' means insert one `\\'.\n\ | |
1113d9db | 2174 | FIXEDCASE and LITERAL are optional arguments.\n\ |
080c45fd RS |
2175 | Leaves point at end of replacement text.\n\ |
2176 | \n\ | |
2177 | The optional fourth argument STRING can be a string to modify.\n\ | |
2178 | In that case, this function creates and returns a new string\n\ | |
d7a5ad5f RS |
2179 | which is made by replacing the part of STRING that was matched.\n\ |
2180 | \n\ | |
2181 | The optional fifth argument SUBEXP specifies a subexpression of the match.\n\ | |
2182 | It says to replace just that subexpression instead of the whole match.\n\ | |
2183 | This is useful only after a regular expression search or match\n\ | |
2184 | since only regular expressions have distinguished subexpressions.") | |
2185 | (newtext, fixedcase, literal, string, subexp) | |
2186 | Lisp_Object newtext, fixedcase, literal, string, subexp; | |
ca1d1d23 JB |
2187 | { |
2188 | enum { nochange, all_caps, cap_initial } case_action; | |
ac3b28b1 | 2189 | register int pos, pos_byte; |
ca1d1d23 | 2190 | int some_multiletter_word; |
97832bd0 | 2191 | int some_lowercase; |
73dc8771 | 2192 | int some_uppercase; |
208767c3 | 2193 | int some_nonuppercase_initial; |
ca1d1d23 JB |
2194 | register int c, prevc; |
2195 | int inslen; | |
d7a5ad5f | 2196 | int sub; |
3e18eecf | 2197 | int opoint, newpoint; |
ca1d1d23 | 2198 | |
16fdc568 | 2199 | CHECK_STRING (newtext, 0); |
ca1d1d23 | 2200 | |
080c45fd RS |
2201 | if (! NILP (string)) |
2202 | CHECK_STRING (string, 4); | |
2203 | ||
ca1d1d23 JB |
2204 | case_action = nochange; /* We tried an initialization */ |
2205 | /* but some C compilers blew it */ | |
4746118a JB |
2206 | |
2207 | if (search_regs.num_regs <= 0) | |
2208 | error ("replace-match called before any match found"); | |
2209 | ||
d7a5ad5f RS |
2210 | if (NILP (subexp)) |
2211 | sub = 0; | |
2212 | else | |
2213 | { | |
2214 | CHECK_NUMBER (subexp, 3); | |
2215 | sub = XINT (subexp); | |
2216 | if (sub < 0 || sub >= search_regs.num_regs) | |
2217 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2218 | } | |
2219 | ||
080c45fd RS |
2220 | if (NILP (string)) |
2221 | { | |
d7a5ad5f RS |
2222 | if (search_regs.start[sub] < BEGV |
2223 | || search_regs.start[sub] > search_regs.end[sub] | |
2224 | || search_regs.end[sub] > ZV) | |
2225 | args_out_of_range (make_number (search_regs.start[sub]), | |
2226 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2227 | } |
2228 | else | |
2229 | { | |
d7a5ad5f RS |
2230 | if (search_regs.start[sub] < 0 |
2231 | || search_regs.start[sub] > search_regs.end[sub] | |
2232 | || search_regs.end[sub] > XSTRING (string)->size) | |
2233 | args_out_of_range (make_number (search_regs.start[sub]), | |
2234 | make_number (search_regs.end[sub])); | |
080c45fd | 2235 | } |
ca1d1d23 JB |
2236 | |
2237 | if (NILP (fixedcase)) | |
2238 | { | |
2239 | /* Decide how to casify by examining the matched text. */ | |
ac3b28b1 | 2240 | int last; |
ca1d1d23 | 2241 | |
ac3b28b1 KH |
2242 | pos = search_regs.start[sub]; |
2243 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2244 | |
2245 | if (NILP (string)) | |
ac3b28b1 | 2246 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2247 | else |
ac3b28b1 | 2248 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2249 | |
ca1d1d23 JB |
2250 | prevc = '\n'; |
2251 | case_action = all_caps; | |
2252 | ||
2253 | /* some_multiletter_word is set nonzero if any original word | |
2254 | is more than one letter long. */ | |
2255 | some_multiletter_word = 0; | |
97832bd0 | 2256 | some_lowercase = 0; |
208767c3 | 2257 | some_nonuppercase_initial = 0; |
73dc8771 | 2258 | some_uppercase = 0; |
ca1d1d23 | 2259 | |
ac3b28b1 | 2260 | while (pos < last) |
ca1d1d23 | 2261 | { |
080c45fd | 2262 | if (NILP (string)) |
ac3b28b1 KH |
2263 | { |
2264 | c = FETCH_CHAR (pos_byte); | |
2265 | INC_BOTH (pos, pos_byte); | |
2266 | } | |
080c45fd | 2267 | else |
ac3b28b1 | 2268 | FETCH_STRING_CHAR_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2269 | |
ca1d1d23 JB |
2270 | if (LOWERCASEP (c)) |
2271 | { | |
2272 | /* Cannot be all caps if any original char is lower case */ | |
2273 | ||
97832bd0 | 2274 | some_lowercase = 1; |
ca1d1d23 | 2275 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2276 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2277 | else |
2278 | some_multiletter_word = 1; | |
2279 | } | |
2280 | else if (!NOCASEP (c)) | |
2281 | { | |
73dc8771 | 2282 | some_uppercase = 1; |
97832bd0 | 2283 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2284 | ; |
97832bd0 | 2285 | else |
ca1d1d23 JB |
2286 | some_multiletter_word = 1; |
2287 | } | |
208767c3 RS |
2288 | else |
2289 | { | |
2290 | /* If the initial is a caseless word constituent, | |
2291 | treat that like a lowercase initial. */ | |
2292 | if (SYNTAX (prevc) != Sword) | |
2293 | some_nonuppercase_initial = 1; | |
2294 | } | |
ca1d1d23 JB |
2295 | |
2296 | prevc = c; | |
2297 | } | |
2298 | ||
97832bd0 RS |
2299 | /* Convert to all caps if the old text is all caps |
2300 | and has at least one multiletter word. */ | |
2301 | if (! some_lowercase && some_multiletter_word) | |
2302 | case_action = all_caps; | |
c4d460ce | 2303 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2304 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2305 | case_action = cap_initial; |
208767c3 | 2306 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2307 | /* Should x -> yz, operating on X, give Yz or YZ? |
2308 | We'll assume the latter. */ | |
2309 | case_action = all_caps; | |
97832bd0 RS |
2310 | else |
2311 | case_action = nochange; | |
ca1d1d23 JB |
2312 | } |
2313 | ||
080c45fd RS |
2314 | /* Do replacement in a string. */ |
2315 | if (!NILP (string)) | |
2316 | { | |
2317 | Lisp_Object before, after; | |
2318 | ||
2319 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2320 | make_number (search_regs.start[sub])); |
2321 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2322 | |
636a5e28 RS |
2323 | /* Substitute parts of the match into NEWTEXT |
2324 | if desired. */ | |
080c45fd RS |
2325 | if (NILP (literal)) |
2326 | { | |
d131e79c RS |
2327 | int lastpos = 0; |
2328 | int lastpos_byte = 0; | |
080c45fd RS |
2329 | /* We build up the substituted string in ACCUM. */ |
2330 | Lisp_Object accum; | |
2331 | Lisp_Object middle; | |
ac3b28b1 | 2332 | int length = STRING_BYTES (XSTRING (newtext)); |
080c45fd RS |
2333 | |
2334 | accum = Qnil; | |
2335 | ||
ac3b28b1 | 2336 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd RS |
2337 | { |
2338 | int substart = -1; | |
2339 | int subend; | |
1e79ec24 | 2340 | int delbackslash = 0; |
080c45fd | 2341 | |
0c8533c6 RS |
2342 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2343 | ||
080c45fd RS |
2344 | if (c == '\\') |
2345 | { | |
0c8533c6 | 2346 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
080c45fd RS |
2347 | if (c == '&') |
2348 | { | |
d7a5ad5f RS |
2349 | substart = search_regs.start[sub]; |
2350 | subend = search_regs.end[sub]; | |
080c45fd RS |
2351 | } |
2352 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') | |
2353 | { | |
ad10348f | 2354 | if (search_regs.start[c - '0'] >= 0) |
080c45fd RS |
2355 | { |
2356 | substart = search_regs.start[c - '0']; | |
2357 | subend = search_regs.end[c - '0']; | |
2358 | } | |
2359 | } | |
1e79ec24 KH |
2360 | else if (c == '\\') |
2361 | delbackslash = 1; | |
636a5e28 RS |
2362 | else |
2363 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2364 | } |
2365 | if (substart >= 0) | |
2366 | { | |
d131e79c RS |
2367 | if (pos - 2 != lastpos) |
2368 | middle = substring_both (newtext, lastpos, | |
2369 | lastpos_byte, | |
2370 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2371 | else |
2372 | middle = Qnil; | |
2373 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2374 | Fsubstring (string, |
2375 | make_number (substart), | |
080c45fd RS |
2376 | make_number (subend))); |
2377 | lastpos = pos; | |
0c8533c6 | 2378 | lastpos_byte = pos_byte; |
080c45fd | 2379 | } |
1e79ec24 KH |
2380 | else if (delbackslash) |
2381 | { | |
d131e79c RS |
2382 | middle = substring_both (newtext, lastpos, |
2383 | lastpos_byte, | |
2384 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2385 | |
1e79ec24 KH |
2386 | accum = concat2 (accum, middle); |
2387 | lastpos = pos; | |
0c8533c6 | 2388 | lastpos_byte = pos_byte; |
1e79ec24 | 2389 | } |
080c45fd RS |
2390 | } |
2391 | ||
d131e79c RS |
2392 | if (pos != lastpos) |
2393 | middle = substring_both (newtext, lastpos, | |
2394 | lastpos_byte, | |
0c8533c6 | 2395 | pos, pos_byte); |
080c45fd RS |
2396 | else |
2397 | middle = Qnil; | |
2398 | ||
2399 | newtext = concat2 (accum, middle); | |
2400 | } | |
2401 | ||
636a5e28 | 2402 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2403 | if (case_action == all_caps) |
2404 | newtext = Fupcase (newtext); | |
2405 | else if (case_action == cap_initial) | |
2b2eead9 | 2406 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2407 | |
2408 | return concat3 (before, newtext, after); | |
2409 | } | |
2410 | ||
b0eba991 | 2411 | /* Record point, the move (quietly) to the start of the match. */ |
9160906f | 2412 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2413 | opoint = PT - ZV; |
9160906f RS |
2414 | else if (PT > search_regs.start[sub]) |
2415 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2416 | else |
2417 | opoint = PT; | |
2418 | ||
fa8ed3e0 | 2419 | TEMP_SET_PT (search_regs.start[sub]); |
b0eba991 | 2420 | |
9a76659d JB |
2421 | /* We insert the replacement text before the old text, and then |
2422 | delete the original text. This means that markers at the | |
2423 | beginning or end of the original will float to the corresponding | |
2424 | position in the replacement. */ | |
ca1d1d23 | 2425 | if (!NILP (literal)) |
16fdc568 | 2426 | Finsert_and_inherit (1, &newtext); |
ca1d1d23 JB |
2427 | else |
2428 | { | |
2429 | struct gcpro gcpro1; | |
ac3b28b1 KH |
2430 | int length = STRING_BYTES (XSTRING (newtext)); |
2431 | ||
16fdc568 | 2432 | GCPRO1 (newtext); |
ca1d1d23 | 2433 | |
ac3b28b1 | 2434 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2435 | { |
6ec8bbd2 | 2436 | int offset = PT - search_regs.start[sub]; |
9a76659d | 2437 | |
ac3b28b1 KH |
2438 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2439 | ||
ca1d1d23 JB |
2440 | if (c == '\\') |
2441 | { | |
ac3b28b1 | 2442 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
ca1d1d23 | 2443 | if (c == '&') |
9a76659d JB |
2444 | Finsert_buffer_substring |
2445 | (Fcurrent_buffer (), | |
d7a5ad5f RS |
2446 | make_number (search_regs.start[sub] + offset), |
2447 | make_number (search_regs.end[sub] + offset)); | |
78445046 | 2448 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2449 | { |
2450 | if (search_regs.start[c - '0'] >= 1) | |
9a76659d JB |
2451 | Finsert_buffer_substring |
2452 | (Fcurrent_buffer (), | |
2453 | make_number (search_regs.start[c - '0'] + offset), | |
2454 | make_number (search_regs.end[c - '0'] + offset)); | |
ca1d1d23 | 2455 | } |
636a5e28 | 2456 | else if (c == '\\') |
ca1d1d23 | 2457 | insert_char (c); |
636a5e28 RS |
2458 | else |
2459 | error ("Invalid use of `\\' in replacement text"); | |
ca1d1d23 JB |
2460 | } |
2461 | else | |
2462 | insert_char (c); | |
2463 | } | |
2464 | UNGCPRO; | |
2465 | } | |
2466 | ||
6ec8bbd2 | 2467 | inslen = PT - (search_regs.start[sub]); |
d7a5ad5f | 2468 | del_range (search_regs.start[sub] + inslen, search_regs.end[sub] + inslen); |
ca1d1d23 JB |
2469 | |
2470 | if (case_action == all_caps) | |
6ec8bbd2 | 2471 | Fupcase_region (make_number (PT - inslen), make_number (PT)); |
ca1d1d23 | 2472 | else if (case_action == cap_initial) |
6ec8bbd2 | 2473 | Fupcase_initials_region (make_number (PT - inslen), make_number (PT)); |
b0eba991 | 2474 | |
3e18eecf RS |
2475 | newpoint = PT; |
2476 | ||
b0eba991 | 2477 | /* Put point back where it was in the text. */ |
8d808a65 | 2478 | if (opoint <= 0) |
fa8ed3e0 | 2479 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2480 | else |
fa8ed3e0 | 2481 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2482 | |
2483 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2484 | move_if_not_intangible (newpoint); |
b0eba991 | 2485 | |
ca1d1d23 JB |
2486 | return Qnil; |
2487 | } | |
2488 | \f | |
2489 | static Lisp_Object | |
2490 | match_limit (num, beginningp) | |
2491 | Lisp_Object num; | |
2492 | int beginningp; | |
2493 | { | |
2494 | register int n; | |
2495 | ||
2496 | CHECK_NUMBER (num, 0); | |
2497 | n = XINT (num); | |
4746118a JB |
2498 | if (n < 0 || n >= search_regs.num_regs) |
2499 | args_out_of_range (num, make_number (search_regs.num_regs)); | |
2500 | if (search_regs.num_regs <= 0 | |
2501 | || search_regs.start[n] < 0) | |
ca1d1d23 JB |
2502 | return Qnil; |
2503 | return (make_number ((beginningp) ? search_regs.start[n] | |
2504 | : search_regs.end[n])); | |
2505 | } | |
2506 | ||
2507 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
2508 | "Return position of start of text matched by last search.\n\ | |
5806161b EN |
2509 | SUBEXP, a number, specifies which parenthesized expression in the last\n\ |
2510 | regexp.\n\ | |
2511 | Value is nil if SUBEXPth pair didn't match, or there were less than\n\ | |
2512 | SUBEXP pairs.\n\ | |
ca1d1d23 | 2513 | Zero means the entire text matched by the whole regexp or whole string.") |
5806161b EN |
2514 | (subexp) |
2515 | Lisp_Object subexp; | |
ca1d1d23 | 2516 | { |
5806161b | 2517 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2518 | } |
2519 | ||
2520 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
2521 | "Return position of end of text matched by last search.\n\ | |
5806161b EN |
2522 | SUBEXP, a number, specifies which parenthesized expression in the last\n\ |
2523 | regexp.\n\ | |
2524 | Value is nil if SUBEXPth pair didn't match, or there were less than\n\ | |
2525 | SUBEXP pairs.\n\ | |
ca1d1d23 | 2526 | Zero means the entire text matched by the whole regexp or whole string.") |
5806161b EN |
2527 | (subexp) |
2528 | Lisp_Object subexp; | |
ca1d1d23 | 2529 | { |
5806161b | 2530 | return match_limit (subexp, 0); |
ca1d1d23 JB |
2531 | } |
2532 | ||
56256c2a | 2533 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 2, 0, |
ca1d1d23 JB |
2534 | "Return a list containing all info on what the last search matched.\n\ |
2535 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'.\n\ | |
2536 | All the elements are markers or nil (nil if the Nth pair didn't match)\n\ | |
2537 | if the last match was on a buffer; integers or nil if a string was matched.\n\ | |
56256c2a RS |
2538 | Use `store-match-data' to reinstate the data in this list.\n\ |
2539 | \n\ | |
2540 | If INTEGERS (the optional first argument) is non-nil, always use integers\n\ | |
8ca821e9 | 2541 | \(rather than markers) to represent buffer positions.\n\ |
56256c2a RS |
2542 | If REUSE is a list, reuse it as part of the value. If REUSE is long enough\n\ |
2543 | to hold all the values, and if INTEGERS is non-nil, no consing is done.") | |
2544 | (integers, reuse) | |
2545 | Lisp_Object integers, reuse; | |
ca1d1d23 | 2546 | { |
56256c2a | 2547 | Lisp_Object tail, prev; |
4746118a | 2548 | Lisp_Object *data; |
ca1d1d23 JB |
2549 | int i, len; |
2550 | ||
daa37602 | 2551 | if (NILP (last_thing_searched)) |
c36bcf1b | 2552 | return Qnil; |
daa37602 | 2553 | |
4746118a JB |
2554 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs) |
2555 | * sizeof (Lisp_Object)); | |
2556 | ||
ca1d1d23 | 2557 | len = -1; |
4746118a | 2558 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2559 | { |
2560 | int start = search_regs.start[i]; | |
2561 | if (start >= 0) | |
2562 | { | |
56256c2a RS |
2563 | if (EQ (last_thing_searched, Qt) |
2564 | || ! NILP (integers)) | |
ca1d1d23 | 2565 | { |
c235cce7 KH |
2566 | XSETFASTINT (data[2 * i], start); |
2567 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2568 | } |
0ed62dc7 | 2569 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2570 | { |
2571 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2572 | Fset_marker (data[2 * i], |
2573 | make_number (start), | |
2574 | last_thing_searched); | |
ca1d1d23 JB |
2575 | data[2 * i + 1] = Fmake_marker (); |
2576 | Fset_marker (data[2 * i + 1], | |
daa37602 JB |
2577 | make_number (search_regs.end[i]), |
2578 | last_thing_searched); | |
ca1d1d23 | 2579 | } |
daa37602 JB |
2580 | else |
2581 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2582 | abort (); | |
2583 | ||
ca1d1d23 JB |
2584 | len = i; |
2585 | } | |
2586 | else | |
2587 | data[2 * i] = data [2 * i + 1] = Qnil; | |
2588 | } | |
56256c2a RS |
2589 | |
2590 | /* If REUSE is not usable, cons up the values and return them. */ | |
2591 | if (! CONSP (reuse)) | |
2592 | return Flist (2 * len + 2, data); | |
2593 | ||
2594 | /* If REUSE is a list, store as many value elements as will fit | |
2595 | into the elements of REUSE. */ | |
2596 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2597 | i++, tail = XCDR (tail)) |
56256c2a RS |
2598 | { |
2599 | if (i < 2 * len + 2) | |
c1d497be | 2600 | XCAR (tail) = data[i]; |
56256c2a | 2601 | else |
c1d497be | 2602 | XCAR (tail) = Qnil; |
56256c2a RS |
2603 | prev = tail; |
2604 | } | |
2605 | ||
2606 | /* If we couldn't fit all value elements into REUSE, | |
2607 | cons up the rest of them and add them to the end of REUSE. */ | |
2608 | if (i < 2 * len + 2) | |
c1d497be | 2609 | XCDR (prev) = Flist (2 * len + 2 - i, data + i); |
56256c2a RS |
2610 | |
2611 | return reuse; | |
ca1d1d23 JB |
2612 | } |
2613 | ||
2614 | ||
3f1c005b | 2615 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 1, 0, |
ca1d1d23 JB |
2616 | "Set internal data on last search match from elements of LIST.\n\ |
2617 | LIST should have been created by calling `match-data' previously.") | |
2618 | (list) | |
2619 | register Lisp_Object list; | |
2620 | { | |
2621 | register int i; | |
2622 | register Lisp_Object marker; | |
2623 | ||
7074fde6 FP |
2624 | if (running_asynch_code) |
2625 | save_search_regs (); | |
2626 | ||
ca1d1d23 | 2627 | if (!CONSP (list) && !NILP (list)) |
b37902c8 | 2628 | list = wrong_type_argument (Qconsp, list); |
ca1d1d23 | 2629 | |
daa37602 JB |
2630 | /* Unless we find a marker with a buffer in LIST, assume that this |
2631 | match data came from a string. */ | |
2632 | last_thing_searched = Qt; | |
2633 | ||
4746118a JB |
2634 | /* Allocate registers if they don't already exist. */ |
2635 | { | |
d084e942 | 2636 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2637 | |
2638 | if (length > search_regs.num_regs) | |
2639 | { | |
1113d9db JB |
2640 | if (search_regs.num_regs == 0) |
2641 | { | |
2642 | search_regs.start | |
2643 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2644 | search_regs.end | |
2645 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2646 | } | |
4746118a | 2647 | else |
1113d9db JB |
2648 | { |
2649 | search_regs.start | |
2650 | = (regoff_t *) xrealloc (search_regs.start, | |
2651 | length * sizeof (regoff_t)); | |
2652 | search_regs.end | |
2653 | = (regoff_t *) xrealloc (search_regs.end, | |
2654 | length * sizeof (regoff_t)); | |
2655 | } | |
4746118a | 2656 | |
487282dc | 2657 | search_regs.num_regs = length; |
4746118a JB |
2658 | } |
2659 | } | |
2660 | ||
2661 | for (i = 0; i < search_regs.num_regs; i++) | |
ca1d1d23 JB |
2662 | { |
2663 | marker = Fcar (list); | |
2664 | if (NILP (marker)) | |
2665 | { | |
2666 | search_regs.start[i] = -1; | |
2667 | list = Fcdr (list); | |
2668 | } | |
2669 | else | |
2670 | { | |
0ed62dc7 | 2671 | if (MARKERP (marker)) |
daa37602 JB |
2672 | { |
2673 | if (XMARKER (marker)->buffer == 0) | |
c235cce7 | 2674 | XSETFASTINT (marker, 0); |
daa37602 | 2675 | else |
a3668d92 | 2676 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); |
daa37602 | 2677 | } |
ca1d1d23 JB |
2678 | |
2679 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
2680 | search_regs.start[i] = XINT (marker); | |
2681 | list = Fcdr (list); | |
2682 | ||
2683 | marker = Fcar (list); | |
0ed62dc7 | 2684 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
c235cce7 | 2685 | XSETFASTINT (marker, 0); |
ca1d1d23 JB |
2686 | |
2687 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
2688 | search_regs.end[i] = XINT (marker); | |
2689 | } | |
2690 | list = Fcdr (list); | |
2691 | } | |
2692 | ||
2693 | return Qnil; | |
2694 | } | |
2695 | ||
7074fde6 FP |
2696 | /* If non-zero the match data have been saved in saved_search_regs |
2697 | during the execution of a sentinel or filter. */ | |
75ebf74b | 2698 | static int search_regs_saved; |
7074fde6 FP |
2699 | static struct re_registers saved_search_regs; |
2700 | ||
2701 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
2702 | if asynchronous code (filter or sentinel) is running. */ | |
2703 | static void | |
2704 | save_search_regs () | |
2705 | { | |
2706 | if (!search_regs_saved) | |
2707 | { | |
2708 | saved_search_regs.num_regs = search_regs.num_regs; | |
2709 | saved_search_regs.start = search_regs.start; | |
2710 | saved_search_regs.end = search_regs.end; | |
2711 | search_regs.num_regs = 0; | |
2d4a771a RS |
2712 | search_regs.start = 0; |
2713 | search_regs.end = 0; | |
7074fde6 FP |
2714 | |
2715 | search_regs_saved = 1; | |
2716 | } | |
2717 | } | |
2718 | ||
2719 | /* Called upon exit from filters and sentinels. */ | |
2720 | void | |
2721 | restore_match_data () | |
2722 | { | |
2723 | if (search_regs_saved) | |
2724 | { | |
2725 | if (search_regs.num_regs > 0) | |
2726 | { | |
2727 | xfree (search_regs.start); | |
2728 | xfree (search_regs.end); | |
2729 | } | |
2730 | search_regs.num_regs = saved_search_regs.num_regs; | |
2731 | search_regs.start = saved_search_regs.start; | |
2732 | search_regs.end = saved_search_regs.end; | |
2733 | ||
2734 | search_regs_saved = 0; | |
2735 | } | |
2736 | } | |
2737 | ||
ca1d1d23 JB |
2738 | /* Quote a string to inactivate reg-expr chars */ |
2739 | ||
2740 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
2741 | "Return a regexp string which matches exactly STRING and nothing else.") | |
5806161b EN |
2742 | (string) |
2743 | Lisp_Object string; | |
ca1d1d23 JB |
2744 | { |
2745 | register unsigned char *in, *out, *end; | |
2746 | register unsigned char *temp; | |
0c8533c6 | 2747 | int backslashes_added = 0; |
ca1d1d23 | 2748 | |
5806161b | 2749 | CHECK_STRING (string, 0); |
ca1d1d23 | 2750 | |
fc932ac6 | 2751 | temp = (unsigned char *) alloca (STRING_BYTES (XSTRING (string)) * 2); |
ca1d1d23 JB |
2752 | |
2753 | /* Now copy the data into the new string, inserting escapes. */ | |
2754 | ||
5806161b | 2755 | in = XSTRING (string)->data; |
fc932ac6 | 2756 | end = in + STRING_BYTES (XSTRING (string)); |
ca1d1d23 JB |
2757 | out = temp; |
2758 | ||
2759 | for (; in != end; in++) | |
2760 | { | |
2761 | if (*in == '[' || *in == ']' | |
2762 | || *in == '*' || *in == '.' || *in == '\\' | |
2763 | || *in == '?' || *in == '+' | |
2764 | || *in == '^' || *in == '$') | |
0c8533c6 | 2765 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
2766 | *out++ = *in; |
2767 | } | |
2768 | ||
3f8100f1 | 2769 | return make_specified_string (temp, |
0c8533c6 | 2770 | XSTRING (string)->size + backslashes_added, |
3f8100f1 RS |
2771 | out - temp, |
2772 | STRING_MULTIBYTE (string)); | |
ca1d1d23 JB |
2773 | } |
2774 | \f | |
dfcf069d | 2775 | void |
ca1d1d23 JB |
2776 | syms_of_search () |
2777 | { | |
2778 | register int i; | |
2779 | ||
487282dc KH |
2780 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
2781 | { | |
2782 | searchbufs[i].buf.allocated = 100; | |
2783 | searchbufs[i].buf.buffer = (unsigned char *) malloc (100); | |
2784 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; | |
2785 | searchbufs[i].regexp = Qnil; | |
2786 | staticpro (&searchbufs[i].regexp); | |
2787 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); | |
2788 | } | |
2789 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
2790 | |
2791 | Qsearch_failed = intern ("search-failed"); | |
2792 | staticpro (&Qsearch_failed); | |
2793 | Qinvalid_regexp = intern ("invalid-regexp"); | |
2794 | staticpro (&Qinvalid_regexp); | |
2795 | ||
2796 | Fput (Qsearch_failed, Qerror_conditions, | |
2797 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
2798 | Fput (Qsearch_failed, Qerror_message, | |
2799 | build_string ("Search failed")); | |
2800 | ||
2801 | Fput (Qinvalid_regexp, Qerror_conditions, | |
2802 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
2803 | Fput (Qinvalid_regexp, Qerror_message, | |
2804 | build_string ("Invalid regexp")); | |
2805 | ||
daa37602 JB |
2806 | last_thing_searched = Qnil; |
2807 | staticpro (&last_thing_searched); | |
2808 | ||
ca1d1d23 | 2809 | defsubr (&Slooking_at); |
b819a390 RS |
2810 | defsubr (&Sposix_looking_at); |
2811 | defsubr (&Sstring_match); | |
2812 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
2813 | defsubr (&Ssearch_forward); |
2814 | defsubr (&Ssearch_backward); | |
2815 | defsubr (&Sword_search_forward); | |
2816 | defsubr (&Sword_search_backward); | |
2817 | defsubr (&Sre_search_forward); | |
2818 | defsubr (&Sre_search_backward); | |
b819a390 RS |
2819 | defsubr (&Sposix_search_forward); |
2820 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
2821 | defsubr (&Sreplace_match); |
2822 | defsubr (&Smatch_beginning); | |
2823 | defsubr (&Smatch_end); | |
2824 | defsubr (&Smatch_data); | |
3f1c005b | 2825 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
2826 | defsubr (&Sregexp_quote); |
2827 | } |