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 | ||
1d288aef | 36 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 37 | |
487282dc KH |
38 | /* If the regexp is non-nil, then the buffer contains the compiled form |
39 | of that regexp, suitable for searching. */ | |
1d288aef RS |
40 | struct regexp_cache |
41 | { | |
487282dc KH |
42 | struct regexp_cache *next; |
43 | Lisp_Object regexp; | |
44 | struct re_pattern_buffer buf; | |
45 | char fastmap[0400]; | |
b819a390 RS |
46 | /* Nonzero means regexp was compiled to do full POSIX backtracking. */ |
47 | char posix; | |
487282dc | 48 | }; |
ca1d1d23 | 49 | |
487282dc KH |
50 | /* The instances of that struct. */ |
51 | struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; | |
ca1d1d23 | 52 | |
487282dc KH |
53 | /* The head of the linked list; points to the most recently used buffer. */ |
54 | struct regexp_cache *searchbuf_head; | |
ca1d1d23 | 55 | |
ca1d1d23 | 56 | |
4746118a JB |
57 | /* Every call to re_match, etc., must pass &search_regs as the regs |
58 | argument unless you can show it is unnecessary (i.e., if re_match | |
59 | is certainly going to be called again before region-around-match | |
60 | can be called). | |
61 | ||
62 | Since the registers are now dynamically allocated, we need to make | |
63 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
64 | been allocated by checking search_regs.num_regs. |
65 | ||
66 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
67 | buffer using bits in the re_pattern_buffer. This means that whenever |
68 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
69 | allocated any registers, and will allocate new registers the next |
70 | time you call a searching or matching function. Therefore, we need | |
71 | to call re_set_registers after compiling a new pattern or after | |
72 | setting the match registers, so that the regex functions will be | |
73 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
74 | static struct re_registers search_regs; |
75 | ||
daa37602 JB |
76 | /* The buffer in which the last search was performed, or |
77 | Qt if the last search was done in a string; | |
78 | Qnil if no searching has been done yet. */ | |
79 | static Lisp_Object last_thing_searched; | |
ca1d1d23 | 80 | |
8e6208c5 | 81 | /* error condition signaled when regexp compile_pattern fails */ |
ca1d1d23 JB |
82 | |
83 | Lisp_Object Qinvalid_regexp; | |
84 | ||
ca325161 | 85 | static void set_search_regs (); |
044f81f1 | 86 | static void save_search_regs (); |
facdc750 RS |
87 | static int simple_search (); |
88 | static int boyer_moore (); | |
b819a390 RS |
89 | static int search_buffer (); |
90 | ||
ca1d1d23 JB |
91 | static void |
92 | matcher_overflow () | |
93 | { | |
94 | error ("Stack overflow in regexp matcher"); | |
95 | } | |
96 | ||
b819a390 RS |
97 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
98 | PATTERN is the pattern to compile. | |
99 | CP is the place to put the result. | |
facdc750 | 100 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
101 | REGP is the structure that says where to store the "register" |
102 | values that will result from matching this pattern. | |
103 | If it is 0, we should compile the pattern not to record any | |
104 | subexpression bounds. | |
105 | POSIX is nonzero if we want full backtracking (POSIX style) | |
5679531d KH |
106 | for this pattern. 0 means backtrack only enough to get a valid match. |
107 | MULTIBYTE is nonzero if we want to handle multibyte characters in | |
108 | PATTERN. 0 means all multibyte characters are recognized just as | |
109 | sequences of binary data. */ | |
ca1d1d23 | 110 | |
487282dc | 111 | static void |
5679531d | 112 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte) |
487282dc | 113 | struct regexp_cache *cp; |
ca1d1d23 | 114 | Lisp_Object pattern; |
facdc750 | 115 | Lisp_Object translate; |
487282dc | 116 | struct re_registers *regp; |
b819a390 | 117 | int posix; |
5679531d | 118 | int multibyte; |
ca1d1d23 | 119 | { |
7276d3d8 | 120 | unsigned char *raw_pattern; |
f8bd51c4 | 121 | int raw_pattern_size; |
d451e4db | 122 | char *val; |
b819a390 | 123 | reg_syntax_t old; |
ca1d1d23 | 124 | |
f8bd51c4 KH |
125 | /* MULTIBYTE says whether the text to be searched is multibyte. |
126 | We must convert PATTERN to match that, or we will not really | |
127 | find things right. */ | |
128 | ||
129 | if (multibyte == STRING_MULTIBYTE (pattern)) | |
130 | { | |
7276d3d8 | 131 | raw_pattern = (unsigned char *) XSTRING (pattern)->data; |
fc932ac6 | 132 | raw_pattern_size = STRING_BYTES (XSTRING (pattern)); |
f8bd51c4 KH |
133 | } |
134 | else if (multibyte) | |
135 | { | |
136 | raw_pattern_size = count_size_as_multibyte (XSTRING (pattern)->data, | |
137 | XSTRING (pattern)->size); | |
7276d3d8 | 138 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
f8bd51c4 KH |
139 | copy_text (XSTRING (pattern)->data, raw_pattern, |
140 | XSTRING (pattern)->size, 0, 1); | |
141 | } | |
142 | else | |
143 | { | |
144 | /* Converting multibyte to single-byte. | |
145 | ||
146 | ??? Perhaps this conversion should be done in a special way | |
147 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
148 | so that only the multibyte chars which really correspond to | |
149 | the chosen single-byte character set can possibly match. */ | |
150 | raw_pattern_size = XSTRING (pattern)->size; | |
7276d3d8 | 151 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
f8bd51c4 | 152 | copy_text (XSTRING (pattern)->data, raw_pattern, |
fc932ac6 | 153 | STRING_BYTES (XSTRING (pattern)), 1, 0); |
f8bd51c4 KH |
154 | } |
155 | ||
487282dc | 156 | cp->regexp = Qnil; |
59fab369 | 157 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 158 | cp->posix = posix; |
5679531d | 159 | cp->buf.multibyte = multibyte; |
9ac0d9e0 | 160 | BLOCK_INPUT; |
fb4a568d | 161 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 162 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
7276d3d8 RS |
163 | val = (char *) re_compile_pattern ((char *)raw_pattern, |
164 | raw_pattern_size, &cp->buf); | |
b819a390 | 165 | re_set_syntax (old); |
9ac0d9e0 | 166 | UNBLOCK_INPUT; |
ca1d1d23 | 167 | if (val) |
487282dc | 168 | Fsignal (Qinvalid_regexp, Fcons (build_string (val), Qnil)); |
1113d9db | 169 | |
487282dc | 170 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
171 | } |
172 | ||
6efc7887 RS |
173 | /* Shrink each compiled regexp buffer in the cache |
174 | to the size actually used right now. | |
175 | This is called from garbage collection. */ | |
176 | ||
177 | void | |
178 | shrink_regexp_cache () | |
179 | { | |
a968f437 | 180 | struct regexp_cache *cp; |
6efc7887 RS |
181 | |
182 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
183 | { | |
184 | cp->buf.allocated = cp->buf.used; | |
185 | cp->buf.buffer | |
186 | = (unsigned char *) realloc (cp->buf.buffer, cp->buf.used); | |
187 | } | |
188 | } | |
189 | ||
487282dc | 190 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
191 | the cache. |
192 | PATTERN is the pattern to compile. | |
facdc750 | 193 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
194 | REGP is the structure that says where to store the "register" |
195 | values that will result from matching this pattern. | |
196 | If it is 0, we should compile the pattern not to record any | |
197 | subexpression bounds. | |
198 | POSIX is nonzero if we want full backtracking (POSIX style) | |
199 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
200 | |
201 | struct re_pattern_buffer * | |
0c8533c6 | 202 | compile_pattern (pattern, regp, translate, posix, multibyte) |
487282dc KH |
203 | Lisp_Object pattern; |
204 | struct re_registers *regp; | |
facdc750 | 205 | Lisp_Object translate; |
0c8533c6 | 206 | int posix, multibyte; |
487282dc KH |
207 | { |
208 | struct regexp_cache *cp, **cpp; | |
209 | ||
210 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
211 | { | |
212 | cp = *cpp; | |
f1b9c7c1 KR |
213 | /* Entries are initialized to nil, and may be set to nil by |
214 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
215 | XSTRING in those cases. However, compile_pattern_1 is only | |
216 | applied to the cache entry we pick here to reuse. So nil | |
217 | should never appear before a non-nil entry. */ | |
7c752c80 | 218 | if (NILP (cp->regexp)) |
f1b9c7c1 | 219 | goto compile_it; |
1d288aef | 220 | if (XSTRING (cp->regexp)->size == XSTRING (pattern)->size |
cf69b13e | 221 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 222 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 223 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d KH |
224 | && cp->posix == posix |
225 | && cp->buf.multibyte == multibyte) | |
487282dc KH |
226 | break; |
227 | ||
f1b9c7c1 KR |
228 | /* If we're at the end of the cache, compile into the nil cell |
229 | we found, or the last (least recently used) cell with a | |
230 | string value. */ | |
487282dc KH |
231 | if (cp->next == 0) |
232 | { | |
f1b9c7c1 | 233 | compile_it: |
5679531d | 234 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte); |
487282dc KH |
235 | break; |
236 | } | |
237 | } | |
238 | ||
239 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
240 | either because we found it in the cache or because we just compiled it. | |
241 | Move it to the front of the queue to mark it as most recently used. */ | |
242 | *cpp = cp->next; | |
243 | cp->next = searchbuf_head; | |
244 | searchbuf_head = cp; | |
1113d9db | 245 | |
6639708c RS |
246 | /* Advise the searching functions about the space we have allocated |
247 | for register data. */ | |
248 | if (regp) | |
249 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
250 | ||
487282dc | 251 | return &cp->buf; |
ca1d1d23 JB |
252 | } |
253 | ||
254 | /* Error condition used for failing searches */ | |
255 | Lisp_Object Qsearch_failed; | |
256 | ||
257 | Lisp_Object | |
258 | signal_failure (arg) | |
259 | Lisp_Object arg; | |
260 | { | |
261 | Fsignal (Qsearch_failed, Fcons (arg, Qnil)); | |
262 | return Qnil; | |
263 | } | |
264 | \f | |
b819a390 RS |
265 | static Lisp_Object |
266 | looking_at_1 (string, posix) | |
ca1d1d23 | 267 | Lisp_Object string; |
b819a390 | 268 | int posix; |
ca1d1d23 JB |
269 | { |
270 | Lisp_Object val; | |
271 | unsigned char *p1, *p2; | |
272 | int s1, s2; | |
273 | register int i; | |
487282dc | 274 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 275 | |
7074fde6 FP |
276 | if (running_asynch_code) |
277 | save_search_regs (); | |
278 | ||
ca1d1d23 | 279 | CHECK_STRING (string, 0); |
487282dc KH |
280 | bufp = compile_pattern (string, &search_regs, |
281 | (!NILP (current_buffer->case_fold_search) | |
facdc750 | 282 | ? DOWNCASE_TABLE : Qnil), |
0c8533c6 RS |
283 | posix, |
284 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 JB |
285 | |
286 | immediate_quit = 1; | |
287 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
288 | ||
289 | /* Get pointers and sizes of the two strings | |
290 | that make up the visible portion of the buffer. */ | |
291 | ||
292 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 293 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 294 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 295 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
296 | if (s1 < 0) |
297 | { | |
298 | p2 = p1; | |
fa8ed3e0 | 299 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
300 | s1 = 0; |
301 | } | |
302 | if (s2 < 0) | |
303 | { | |
fa8ed3e0 | 304 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
305 | s2 = 0; |
306 | } | |
8bb43c28 RS |
307 | |
308 | re_match_object = Qnil; | |
ca1d1d23 | 309 | |
487282dc | 310 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
fa8ed3e0 RS |
311 | PT_BYTE - BEGV_BYTE, &search_regs, |
312 | ZV_BYTE - BEGV_BYTE); | |
de182d70 GM |
313 | immediate_quit = 0; |
314 | ||
ca1d1d23 JB |
315 | if (i == -2) |
316 | matcher_overflow (); | |
317 | ||
318 | val = (0 <= i ? Qt : Qnil); | |
fa8ed3e0 RS |
319 | if (i >= 0) |
320 | for (i = 0; i < search_regs.num_regs; i++) | |
321 | if (search_regs.start[i] >= 0) | |
322 | { | |
323 | search_regs.start[i] | |
324 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
325 | search_regs.end[i] | |
326 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
327 | } | |
a3668d92 | 328 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
329 | return val; |
330 | } | |
331 | ||
b819a390 | 332 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
333 | doc: /* Return t if text after point matches regular expression REGEXP. |
334 | This function modifies the match data that `match-beginning', | |
335 | `match-end' and `match-data' access; save and restore the match | |
336 | data if you want to preserve them. */) | |
337 | (regexp) | |
94f94972 | 338 | Lisp_Object regexp; |
b819a390 | 339 | { |
94f94972 | 340 | return looking_at_1 (regexp, 0); |
b819a390 RS |
341 | } |
342 | ||
343 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
344 | doc: /* Return t if text after point matches regular expression REGEXP. |
345 | Find the longest match, in accord with Posix regular expression rules. | |
346 | This function modifies the match data that `match-beginning', | |
347 | `match-end' and `match-data' access; save and restore the match | |
348 | data if you want to preserve them. */) | |
349 | (regexp) | |
94f94972 | 350 | Lisp_Object regexp; |
b819a390 | 351 | { |
94f94972 | 352 | return looking_at_1 (regexp, 1); |
b819a390 RS |
353 | } |
354 | \f | |
355 | static Lisp_Object | |
356 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 357 | Lisp_Object regexp, string, start; |
b819a390 | 358 | int posix; |
ca1d1d23 JB |
359 | { |
360 | int val; | |
487282dc | 361 | struct re_pattern_buffer *bufp; |
0c8533c6 RS |
362 | int pos, pos_byte; |
363 | int i; | |
ca1d1d23 | 364 | |
7074fde6 FP |
365 | if (running_asynch_code) |
366 | save_search_regs (); | |
367 | ||
ca1d1d23 JB |
368 | CHECK_STRING (regexp, 0); |
369 | CHECK_STRING (string, 1); | |
370 | ||
371 | if (NILP (start)) | |
0c8533c6 | 372 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
373 | else |
374 | { | |
375 | int len = XSTRING (string)->size; | |
376 | ||
377 | CHECK_NUMBER (start, 2); | |
0c8533c6 RS |
378 | pos = XINT (start); |
379 | if (pos < 0 && -pos <= len) | |
380 | pos = len + pos; | |
381 | else if (0 > pos || pos > len) | |
ca1d1d23 | 382 | args_out_of_range (string, start); |
0c8533c6 | 383 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
384 | } |
385 | ||
487282dc KH |
386 | bufp = compile_pattern (regexp, &search_regs, |
387 | (!NILP (current_buffer->case_fold_search) | |
facdc750 | 388 | ? DOWNCASE_TABLE : Qnil), |
0c8533c6 RS |
389 | posix, |
390 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 391 | immediate_quit = 1; |
8bb43c28 RS |
392 | re_match_object = string; |
393 | ||
487282dc | 394 | val = re_search (bufp, (char *) XSTRING (string)->data, |
fc932ac6 RS |
395 | STRING_BYTES (XSTRING (string)), pos_byte, |
396 | STRING_BYTES (XSTRING (string)) - pos_byte, | |
ca1d1d23 JB |
397 | &search_regs); |
398 | immediate_quit = 0; | |
daa37602 | 399 | last_thing_searched = Qt; |
ca1d1d23 JB |
400 | if (val == -2) |
401 | matcher_overflow (); | |
402 | if (val < 0) return Qnil; | |
0c8533c6 RS |
403 | |
404 | for (i = 0; i < search_regs.num_regs; i++) | |
405 | if (search_regs.start[i] >= 0) | |
406 | { | |
407 | search_regs.start[i] | |
408 | = string_byte_to_char (string, search_regs.start[i]); | |
409 | search_regs.end[i] | |
410 | = string_byte_to_char (string, search_regs.end[i]); | |
411 | } | |
412 | ||
413 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 414 | } |
e59a8453 | 415 | |
b819a390 | 416 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 PJ |
417 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
418 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
419 | If third arg START is non-nil, start search at that index in STRING. | |
420 | For index of first char beyond the match, do (match-end 0). | |
421 | `match-end' and `match-beginning' also give indices of substrings | |
422 | matched by parenthesis constructs in the pattern. */) | |
423 | (regexp, string, start) | |
b819a390 RS |
424 | Lisp_Object regexp, string, start; |
425 | { | |
426 | return string_match_1 (regexp, string, start, 0); | |
427 | } | |
428 | ||
429 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
430 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
431 | Find the longest match, in accord with Posix regular expression rules. | |
432 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
433 | If third arg START is non-nil, start search at that index in STRING. | |
434 | For index of first char beyond the match, do (match-end 0). | |
435 | `match-end' and `match-beginning' also give indices of substrings | |
436 | matched by parenthesis constructs in the pattern. */) | |
437 | (regexp, string, start) | |
b819a390 RS |
438 | Lisp_Object regexp, string, start; |
439 | { | |
440 | return string_match_1 (regexp, string, start, 1); | |
441 | } | |
442 | ||
e59a8453 RS |
443 | /* Match REGEXP against STRING, searching all of STRING, |
444 | and return the index of the match, or negative on failure. | |
445 | This does not clobber the match data. */ | |
446 | ||
447 | int | |
448 | fast_string_match (regexp, string) | |
449 | Lisp_Object regexp, string; | |
450 | { | |
451 | int val; | |
487282dc | 452 | struct re_pattern_buffer *bufp; |
e59a8453 | 453 | |
facdc750 RS |
454 | bufp = compile_pattern (regexp, 0, Qnil, |
455 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 456 | immediate_quit = 1; |
8bb43c28 RS |
457 | re_match_object = string; |
458 | ||
487282dc | 459 | val = re_search (bufp, (char *) XSTRING (string)->data, |
fc932ac6 RS |
460 | STRING_BYTES (XSTRING (string)), 0, |
461 | STRING_BYTES (XSTRING (string)), 0); | |
e59a8453 RS |
462 | immediate_quit = 0; |
463 | return val; | |
464 | } | |
5679531d KH |
465 | |
466 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
467 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
468 | This does not clobber the match data. |
469 | We assume that STRING contains single-byte characters. */ | |
5679531d KH |
470 | |
471 | extern Lisp_Object Vascii_downcase_table; | |
472 | ||
473 | int | |
b4577c63 | 474 | fast_c_string_match_ignore_case (regexp, string) |
5679531d KH |
475 | Lisp_Object regexp; |
476 | char *string; | |
477 | { | |
478 | int val; | |
479 | struct re_pattern_buffer *bufp; | |
480 | int len = strlen (string); | |
481 | ||
0c8533c6 | 482 | regexp = string_make_unibyte (regexp); |
b4577c63 | 483 | re_match_object = Qt; |
5679531d | 484 | bufp = compile_pattern (regexp, 0, |
facdc750 | 485 | Vascii_downcase_table, 0, |
f8bd51c4 | 486 | 0); |
5679531d KH |
487 | immediate_quit = 1; |
488 | val = re_search (bufp, string, len, 0, len, 0); | |
489 | immediate_quit = 0; | |
490 | return val; | |
491 | } | |
ca1d1d23 | 492 | \f |
9169c321 JB |
493 | /* The newline cache: remembering which sections of text have no newlines. */ |
494 | ||
495 | /* If the user has requested newline caching, make sure it's on. | |
496 | Otherwise, make sure it's off. | |
497 | This is our cheezy way of associating an action with the change of | |
498 | state of a buffer-local variable. */ | |
499 | static void | |
500 | newline_cache_on_off (buf) | |
501 | struct buffer *buf; | |
502 | { | |
503 | if (NILP (buf->cache_long_line_scans)) | |
504 | { | |
505 | /* It should be off. */ | |
506 | if (buf->newline_cache) | |
507 | { | |
508 | free_region_cache (buf->newline_cache); | |
509 | buf->newline_cache = 0; | |
510 | } | |
511 | } | |
512 | else | |
513 | { | |
514 | /* It should be on. */ | |
515 | if (buf->newline_cache == 0) | |
516 | buf->newline_cache = new_region_cache (); | |
517 | } | |
518 | } | |
519 | ||
520 | \f | |
521 | /* Search for COUNT instances of the character TARGET between START and END. | |
522 | ||
523 | If COUNT is positive, search forwards; END must be >= START. | |
524 | If COUNT is negative, search backwards for the -COUNTth instance; | |
525 | END must be <= START. | |
526 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
527 | ||
528 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
529 | direction indicated by COUNT. | |
ffd56f97 JB |
530 | |
531 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
5bfe95c9 RS |
532 | position after the COUNTth match. Note that for reverse motion |
533 | this is not the same as the usual convention for Emacs motion commands. | |
ffd56f97 | 534 | |
9169c321 JB |
535 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
536 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 537 | |
087a5f81 RS |
538 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
539 | except when inside redisplay. */ | |
540 | ||
dfcf069d | 541 | int |
9169c321 JB |
542 | scan_buffer (target, start, end, count, shortage, allow_quit) |
543 | register int target; | |
544 | int start, end; | |
545 | int count; | |
546 | int *shortage; | |
087a5f81 | 547 | int allow_quit; |
ca1d1d23 | 548 | { |
9169c321 JB |
549 | struct region_cache *newline_cache; |
550 | int direction; | |
ffd56f97 | 551 | |
9169c321 JB |
552 | if (count > 0) |
553 | { | |
554 | direction = 1; | |
555 | if (! end) end = ZV; | |
556 | } | |
557 | else | |
558 | { | |
559 | direction = -1; | |
560 | if (! end) end = BEGV; | |
561 | } | |
ffd56f97 | 562 | |
9169c321 JB |
563 | newline_cache_on_off (current_buffer); |
564 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
565 | |
566 | if (shortage != 0) | |
567 | *shortage = 0; | |
568 | ||
087a5f81 | 569 | immediate_quit = allow_quit; |
ca1d1d23 | 570 | |
ffd56f97 | 571 | if (count > 0) |
9169c321 | 572 | while (start != end) |
ca1d1d23 | 573 | { |
9169c321 JB |
574 | /* Our innermost scanning loop is very simple; it doesn't know |
575 | about gaps, buffer ends, or the newline cache. ceiling is | |
576 | the position of the last character before the next such | |
577 | obstacle --- the last character the dumb search loop should | |
578 | examine. */ | |
fa8ed3e0 RS |
579 | int ceiling_byte = CHAR_TO_BYTE (end) - 1; |
580 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 581 | int tem; |
9169c321 JB |
582 | |
583 | /* If we're looking for a newline, consult the newline cache | |
584 | to see where we can avoid some scanning. */ | |
585 | if (target == '\n' && newline_cache) | |
586 | { | |
587 | int next_change; | |
588 | immediate_quit = 0; | |
589 | while (region_cache_forward | |
fa8ed3e0 RS |
590 | (current_buffer, newline_cache, start_byte, &next_change)) |
591 | start_byte = next_change; | |
cbe0db0d | 592 | immediate_quit = allow_quit; |
9169c321 | 593 | |
fa8ed3e0 RS |
594 | /* START should never be after END. */ |
595 | if (start_byte > ceiling_byte) | |
596 | start_byte = ceiling_byte; | |
9169c321 JB |
597 | |
598 | /* Now the text after start is an unknown region, and | |
599 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 600 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
601 | } |
602 | ||
603 | /* The dumb loop can only scan text stored in contiguous | |
604 | bytes. BUFFER_CEILING_OF returns the last character | |
605 | position that is contiguous, so the ceiling is the | |
606 | position after that. */ | |
67ce527d KH |
607 | tem = BUFFER_CEILING_OF (start_byte); |
608 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
609 | |
610 | { | |
611 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
612 | register unsigned char *ceiling_addr |
613 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
614 | register unsigned char *cursor | |
615 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
616 | unsigned char *base = cursor; |
617 | ||
618 | while (cursor < ceiling_addr) | |
619 | { | |
620 | unsigned char *scan_start = cursor; | |
621 | ||
622 | /* The dumb loop. */ | |
623 | while (*cursor != target && ++cursor < ceiling_addr) | |
624 | ; | |
625 | ||
626 | /* If we're looking for newlines, cache the fact that | |
627 | the region from start to cursor is free of them. */ | |
628 | if (target == '\n' && newline_cache) | |
629 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
630 | start_byte + scan_start - base, |
631 | start_byte + cursor - base); | |
9169c321 JB |
632 | |
633 | /* Did we find the target character? */ | |
634 | if (cursor < ceiling_addr) | |
635 | { | |
636 | if (--count == 0) | |
637 | { | |
638 | immediate_quit = 0; | |
fa8ed3e0 | 639 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
640 | } |
641 | cursor++; | |
642 | } | |
643 | } | |
644 | ||
fa8ed3e0 | 645 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 646 | } |
ca1d1d23 JB |
647 | } |
648 | else | |
9169c321 JB |
649 | while (start > end) |
650 | { | |
651 | /* The last character to check before the next obstacle. */ | |
fa8ed3e0 RS |
652 | int ceiling_byte = CHAR_TO_BYTE (end); |
653 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 654 | int tem; |
9169c321 JB |
655 | |
656 | /* Consult the newline cache, if appropriate. */ | |
657 | if (target == '\n' && newline_cache) | |
658 | { | |
659 | int next_change; | |
660 | immediate_quit = 0; | |
661 | while (region_cache_backward | |
fa8ed3e0 RS |
662 | (current_buffer, newline_cache, start_byte, &next_change)) |
663 | start_byte = next_change; | |
cbe0db0d | 664 | immediate_quit = allow_quit; |
9169c321 JB |
665 | |
666 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
667 | if (start_byte <= ceiling_byte) |
668 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
669 | |
670 | /* Now the text before start is an unknown region, and | |
671 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 672 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
673 | } |
674 | ||
675 | /* Stop scanning before the gap. */ | |
67ce527d KH |
676 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
677 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
678 | |
679 | { | |
680 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
681 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
682 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
683 | unsigned char *base = cursor; |
684 | ||
685 | while (cursor >= ceiling_addr) | |
686 | { | |
687 | unsigned char *scan_start = cursor; | |
688 | ||
689 | while (*cursor != target && --cursor >= ceiling_addr) | |
690 | ; | |
691 | ||
692 | /* If we're looking for newlines, cache the fact that | |
693 | the region from after the cursor to start is free of them. */ | |
694 | if (target == '\n' && newline_cache) | |
695 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
696 | start_byte + cursor - base, |
697 | start_byte + scan_start - base); | |
9169c321 JB |
698 | |
699 | /* Did we find the target character? */ | |
700 | if (cursor >= ceiling_addr) | |
701 | { | |
702 | if (++count >= 0) | |
703 | { | |
704 | immediate_quit = 0; | |
fa8ed3e0 | 705 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
706 | } |
707 | cursor--; | |
708 | } | |
709 | } | |
710 | ||
fa8ed3e0 | 711 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
712 | } |
713 | } | |
714 | ||
ca1d1d23 JB |
715 | immediate_quit = 0; |
716 | if (shortage != 0) | |
ffd56f97 | 717 | *shortage = count * direction; |
9169c321 | 718 | return start; |
ca1d1d23 | 719 | } |
fa8ed3e0 RS |
720 | \f |
721 | /* Search for COUNT instances of a line boundary, which means either a | |
722 | newline or (if selective display enabled) a carriage return. | |
723 | Start at START. If COUNT is negative, search backwards. | |
724 | ||
725 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
726 | ||
727 | If we find COUNT instances. we position after (always after, | |
728 | even if scanning backwards) the COUNTth match, and return 0. | |
729 | ||
730 | If we don't find COUNT instances before reaching the end of the | |
731 | buffer (or the beginning, if scanning backwards), we return | |
732 | the number of line boundaries left unfound, and position at | |
733 | the limit we bumped up against. | |
734 | ||
735 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 736 | except in special cases. */ |
ca1d1d23 | 737 | |
63fa018d | 738 | int |
fa8ed3e0 RS |
739 | scan_newline (start, start_byte, limit, limit_byte, count, allow_quit) |
740 | int start, start_byte; | |
741 | int limit, limit_byte; | |
742 | register int count; | |
743 | int allow_quit; | |
63fa018d | 744 | { |
fa8ed3e0 RS |
745 | int direction = ((count > 0) ? 1 : -1); |
746 | ||
747 | register unsigned char *cursor; | |
748 | unsigned char *base; | |
749 | ||
750 | register int ceiling; | |
751 | register unsigned char *ceiling_addr; | |
752 | ||
d5d57b92 RS |
753 | int old_immediate_quit = immediate_quit; |
754 | ||
fa8ed3e0 RS |
755 | /* The code that follows is like scan_buffer |
756 | but checks for either newline or carriage return. */ | |
757 | ||
d5d57b92 RS |
758 | if (allow_quit) |
759 | immediate_quit++; | |
fa8ed3e0 RS |
760 | |
761 | start_byte = CHAR_TO_BYTE (start); | |
762 | ||
763 | if (count > 0) | |
764 | { | |
765 | while (start_byte < limit_byte) | |
766 | { | |
767 | ceiling = BUFFER_CEILING_OF (start_byte); | |
768 | ceiling = min (limit_byte - 1, ceiling); | |
769 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
770 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
771 | while (1) | |
772 | { | |
773 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
774 | ; | |
775 | ||
776 | if (cursor != ceiling_addr) | |
777 | { | |
778 | if (--count == 0) | |
779 | { | |
d5d57b92 | 780 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
781 | start_byte = start_byte + cursor - base + 1; |
782 | start = BYTE_TO_CHAR (start_byte); | |
783 | TEMP_SET_PT_BOTH (start, start_byte); | |
784 | return 0; | |
785 | } | |
786 | else | |
787 | if (++cursor == ceiling_addr) | |
788 | break; | |
789 | } | |
790 | else | |
791 | break; | |
792 | } | |
793 | start_byte += cursor - base; | |
794 | } | |
795 | } | |
796 | else | |
797 | { | |
fa8ed3e0 RS |
798 | while (start_byte > limit_byte) |
799 | { | |
800 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
801 | ceiling = max (limit_byte, ceiling); | |
802 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
803 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
804 | while (1) | |
805 | { | |
806 | while (--cursor != ceiling_addr && *cursor != '\n') | |
807 | ; | |
808 | ||
809 | if (cursor != ceiling_addr) | |
810 | { | |
811 | if (++count == 0) | |
812 | { | |
d5d57b92 | 813 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
814 | /* Return the position AFTER the match we found. */ |
815 | start_byte = start_byte + cursor - base + 1; | |
816 | start = BYTE_TO_CHAR (start_byte); | |
817 | TEMP_SET_PT_BOTH (start, start_byte); | |
818 | return 0; | |
819 | } | |
820 | } | |
821 | else | |
822 | break; | |
823 | } | |
824 | /* Here we add 1 to compensate for the last decrement | |
825 | of CURSOR, which took it past the valid range. */ | |
826 | start_byte += cursor - base + 1; | |
827 | } | |
828 | } | |
829 | ||
830 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 831 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
832 | |
833 | return count * direction; | |
63fa018d RS |
834 | } |
835 | ||
ca1d1d23 | 836 | int |
fa8ed3e0 | 837 | find_next_newline_no_quit (from, cnt) |
ca1d1d23 JB |
838 | register int from, cnt; |
839 | { | |
fa8ed3e0 | 840 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
9169c321 JB |
841 | } |
842 | ||
9169c321 JB |
843 | /* Like find_next_newline, but returns position before the newline, |
844 | not after, and only search up to TO. This isn't just | |
845 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 846 | |
9169c321 JB |
847 | int |
848 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 849 | int from, to, cnt; |
9169c321 JB |
850 | { |
851 | int shortage; | |
852 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
853 | ||
854 | if (shortage == 0) | |
855 | pos--; | |
856 | ||
857 | return pos; | |
ca1d1d23 JB |
858 | } |
859 | \f | |
ca1d1d23 JB |
860 | /* Subroutines of Lisp buffer search functions. */ |
861 | ||
862 | static Lisp_Object | |
b819a390 | 863 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
864 | Lisp_Object string, bound, noerror, count; |
865 | int direction; | |
866 | int RE; | |
b819a390 | 867 | int posix; |
ca1d1d23 JB |
868 | { |
869 | register int np; | |
9f43ad85 | 870 | int lim, lim_byte; |
ca1d1d23 JB |
871 | int n = direction; |
872 | ||
873 | if (!NILP (count)) | |
874 | { | |
875 | CHECK_NUMBER (count, 3); | |
876 | n *= XINT (count); | |
877 | } | |
878 | ||
879 | CHECK_STRING (string, 0); | |
880 | if (NILP (bound)) | |
9f43ad85 RS |
881 | { |
882 | if (n > 0) | |
883 | lim = ZV, lim_byte = ZV_BYTE; | |
884 | else | |
885 | lim = BEGV, lim_byte = BEGV_BYTE; | |
886 | } | |
ca1d1d23 JB |
887 | else |
888 | { | |
889 | CHECK_NUMBER_COERCE_MARKER (bound, 1); | |
890 | lim = XINT (bound); | |
6ec8bbd2 | 891 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
892 | error ("Invalid search bound (wrong side of point)"); |
893 | if (lim > ZV) | |
9f43ad85 | 894 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 895 | else if (lim < BEGV) |
9f43ad85 | 896 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
897 | else |
898 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
899 | } |
900 | ||
9f43ad85 | 901 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
ca1d1d23 | 902 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 903 | ? current_buffer->case_canon_table |
3135e9fd | 904 | : Qnil), |
ca1d1d23 | 905 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 906 | ? current_buffer->case_eqv_table |
3135e9fd | 907 | : Qnil), |
b819a390 | 908 | posix); |
ca1d1d23 JB |
909 | if (np <= 0) |
910 | { | |
911 | if (NILP (noerror)) | |
912 | return signal_failure (string); | |
913 | if (!EQ (noerror, Qt)) | |
914 | { | |
915 | if (lim < BEGV || lim > ZV) | |
916 | abort (); | |
9f43ad85 | 917 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
918 | return Qnil; |
919 | #if 0 /* This would be clean, but maybe programs depend on | |
920 | a value of nil here. */ | |
481399bf | 921 | np = lim; |
a5f217b8 | 922 | #endif |
ca1d1d23 | 923 | } |
481399bf RS |
924 | else |
925 | return Qnil; | |
ca1d1d23 JB |
926 | } |
927 | ||
928 | if (np < BEGV || np > ZV) | |
929 | abort (); | |
930 | ||
931 | SET_PT (np); | |
932 | ||
933 | return make_number (np); | |
934 | } | |
935 | \f | |
fa8ed3e0 RS |
936 | /* Return 1 if REGEXP it matches just one constant string. */ |
937 | ||
b6d6a51c KH |
938 | static int |
939 | trivial_regexp_p (regexp) | |
940 | Lisp_Object regexp; | |
941 | { | |
fc932ac6 | 942 | int len = STRING_BYTES (XSTRING (regexp)); |
b6d6a51c | 943 | unsigned char *s = XSTRING (regexp)->data; |
b6d6a51c KH |
944 | while (--len >= 0) |
945 | { | |
946 | switch (*s++) | |
947 | { | |
948 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
949 | return 0; | |
950 | case '\\': | |
951 | if (--len < 0) | |
952 | return 0; | |
953 | switch (*s++) | |
954 | { | |
955 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
956 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
09c4719e | 957 | case 'S': case '=': case '{': case '}': |
5679531d | 958 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 959 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
960 | case '6': case '7': case '8': case '9': |
961 | return 0; | |
962 | } | |
963 | } | |
964 | } | |
965 | return 1; | |
966 | } | |
967 | ||
ca325161 | 968 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 969 | starting at position POS and stopping at position LIM, |
b819a390 | 970 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
971 | a regular expression if RE is true. |
972 | ||
973 | If N is positive, searching is forward and LIM must be greater than POS. | |
974 | If N is negative, searching is backward and LIM must be less than POS. | |
975 | ||
facdc750 | 976 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 977 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
978 | (if searching backward) or the end (if searching forward). |
979 | ||
980 | POSIX is nonzero if we want full backtracking (POSIX style) | |
981 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 982 | |
aff2ce94 RS |
983 | #define TRANSLATE(out, trt, d) \ |
984 | do \ | |
985 | { \ | |
986 | if (! NILP (trt)) \ | |
987 | { \ | |
988 | Lisp_Object temp; \ | |
989 | temp = Faref (trt, make_number (d)); \ | |
990 | if (INTEGERP (temp)) \ | |
991 | out = XINT (temp); \ | |
992 | else \ | |
993 | out = d; \ | |
994 | } \ | |
995 | else \ | |
996 | out = d; \ | |
997 | } \ | |
998 | while (0) | |
facdc750 | 999 | |
b819a390 | 1000 | static int |
9f43ad85 RS |
1001 | search_buffer (string, pos, pos_byte, lim, lim_byte, n, |
1002 | RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
1003 | Lisp_Object string; |
1004 | int pos; | |
9f43ad85 | 1005 | int pos_byte; |
ca1d1d23 | 1006 | int lim; |
9f43ad85 | 1007 | int lim_byte; |
ca1d1d23 JB |
1008 | int n; |
1009 | int RE; | |
facdc750 RS |
1010 | Lisp_Object trt; |
1011 | Lisp_Object inverse_trt; | |
b819a390 | 1012 | int posix; |
ca1d1d23 JB |
1013 | { |
1014 | int len = XSTRING (string)->size; | |
fc932ac6 | 1015 | int len_byte = STRING_BYTES (XSTRING (string)); |
facdc750 | 1016 | register int i; |
ca1d1d23 | 1017 | |
7074fde6 FP |
1018 | if (running_asynch_code) |
1019 | save_search_regs (); | |
1020 | ||
a7e4cdde | 1021 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1022 | /* Null string is found at starting position. */ |
a7e4cdde | 1023 | if (len == 0 || n == 0) |
ca325161 | 1024 | { |
0353b28f | 1025 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1026 | return pos; |
1027 | } | |
3f57a499 | 1028 | |
b6d6a51c | 1029 | if (RE && !trivial_regexp_p (string)) |
ca1d1d23 | 1030 | { |
facdc750 RS |
1031 | unsigned char *p1, *p2; |
1032 | int s1, s2; | |
487282dc KH |
1033 | struct re_pattern_buffer *bufp; |
1034 | ||
0c8533c6 RS |
1035 | bufp = compile_pattern (string, &search_regs, trt, posix, |
1036 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 | 1037 | |
ca1d1d23 JB |
1038 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1039 | because letting this function finish | |
1040 | can take too long. */ | |
1041 | QUIT; /* Do a pending quit right away, | |
1042 | to avoid paradoxical behavior */ | |
1043 | /* Get pointers and sizes of the two strings | |
1044 | that make up the visible portion of the buffer. */ | |
1045 | ||
1046 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1047 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1048 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1049 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1050 | if (s1 < 0) |
1051 | { | |
1052 | p2 = p1; | |
fa8ed3e0 | 1053 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1054 | s1 = 0; |
1055 | } | |
1056 | if (s2 < 0) | |
1057 | { | |
fa8ed3e0 | 1058 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1059 | s2 = 0; |
1060 | } | |
8bb43c28 RS |
1061 | re_match_object = Qnil; |
1062 | ||
ca1d1d23 JB |
1063 | while (n < 0) |
1064 | { | |
42db823b | 1065 | int val; |
487282dc | 1066 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1067 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1068 | &search_regs, | |
42db823b | 1069 | /* Don't allow match past current point */ |
4996330b | 1070 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1071 | if (val == -2) |
b6d6a51c KH |
1072 | { |
1073 | matcher_overflow (); | |
1074 | } | |
ca1d1d23 JB |
1075 | if (val >= 0) |
1076 | { | |
26aff150 | 1077 | pos_byte = search_regs.start[0] + BEGV_BYTE; |
4746118a | 1078 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1079 | if (search_regs.start[i] >= 0) |
1080 | { | |
fa8ed3e0 RS |
1081 | search_regs.start[i] |
1082 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1083 | search_regs.end[i] | |
1084 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1085 | } |
a3668d92 | 1086 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1087 | /* Set pos to the new position. */ |
1088 | pos = search_regs.start[0]; | |
1089 | } | |
1090 | else | |
1091 | { | |
1092 | immediate_quit = 0; | |
1093 | return (n); | |
1094 | } | |
1095 | n++; | |
1096 | } | |
1097 | while (n > 0) | |
1098 | { | |
42db823b | 1099 | int val; |
487282dc | 1100 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1101 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1102 | &search_regs, | |
1103 | lim_byte - BEGV_BYTE); | |
ca1d1d23 | 1104 | if (val == -2) |
b6d6a51c KH |
1105 | { |
1106 | matcher_overflow (); | |
1107 | } | |
ca1d1d23 JB |
1108 | if (val >= 0) |
1109 | { | |
26aff150 | 1110 | pos_byte = search_regs.end[0] + BEGV_BYTE; |
4746118a | 1111 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1112 | if (search_regs.start[i] >= 0) |
1113 | { | |
fa8ed3e0 RS |
1114 | search_regs.start[i] |
1115 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1116 | search_regs.end[i] | |
1117 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1118 | } |
a3668d92 | 1119 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1120 | pos = search_regs.end[0]; |
1121 | } | |
1122 | else | |
1123 | { | |
1124 | immediate_quit = 0; | |
1125 | return (0 - n); | |
1126 | } | |
1127 | n--; | |
1128 | } | |
1129 | immediate_quit = 0; | |
1130 | return (pos); | |
1131 | } | |
1132 | else /* non-RE case */ | |
1133 | { | |
facdc750 RS |
1134 | unsigned char *raw_pattern, *pat; |
1135 | int raw_pattern_size; | |
1136 | int raw_pattern_size_byte; | |
1137 | unsigned char *patbuf; | |
1138 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); | |
1139 | unsigned char *base_pat = XSTRING (string)->data; | |
1140 | int charset_base = -1; | |
040272ce | 1141 | int boyer_moore_ok = 1; |
facdc750 RS |
1142 | |
1143 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1144 | We must convert PATTERN to match that, or we will not really | |
1145 | find things right. */ | |
1146 | ||
1147 | if (multibyte == STRING_MULTIBYTE (string)) | |
1148 | { | |
7276d3d8 | 1149 | raw_pattern = (unsigned char *) XSTRING (string)->data; |
facdc750 | 1150 | raw_pattern_size = XSTRING (string)->size; |
fc932ac6 | 1151 | raw_pattern_size_byte = STRING_BYTES (XSTRING (string)); |
facdc750 RS |
1152 | } |
1153 | else if (multibyte) | |
1154 | { | |
1155 | raw_pattern_size = XSTRING (string)->size; | |
1156 | raw_pattern_size_byte | |
1157 | = count_size_as_multibyte (XSTRING (string)->data, | |
1158 | raw_pattern_size); | |
7276d3d8 | 1159 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
facdc750 RS |
1160 | copy_text (XSTRING (string)->data, raw_pattern, |
1161 | XSTRING (string)->size, 0, 1); | |
1162 | } | |
1163 | else | |
1164 | { | |
1165 | /* Converting multibyte to single-byte. | |
1166 | ||
1167 | ??? Perhaps this conversion should be done in a special way | |
1168 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1169 | so that only the multibyte chars which really correspond to | |
1170 | the chosen single-byte character set can possibly match. */ | |
1171 | raw_pattern_size = XSTRING (string)->size; | |
1172 | raw_pattern_size_byte = XSTRING (string)->size; | |
7276d3d8 | 1173 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
facdc750 | 1174 | copy_text (XSTRING (string)->data, raw_pattern, |
fc932ac6 | 1175 | STRING_BYTES (XSTRING (string)), 1, 0); |
facdc750 RS |
1176 | } |
1177 | ||
1178 | /* Copy and optionally translate the pattern. */ | |
1179 | len = raw_pattern_size; | |
1180 | len_byte = raw_pattern_size_byte; | |
1181 | patbuf = (unsigned char *) alloca (len_byte); | |
1182 | pat = patbuf; | |
1183 | base_pat = raw_pattern; | |
1184 | if (multibyte) | |
1185 | { | |
1186 | while (--len >= 0) | |
1187 | { | |
daaa6ed8 | 1188 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
aff2ce94 | 1189 | int c, translated, inverse; |
facdc750 RS |
1190 | int in_charlen, charlen; |
1191 | ||
1192 | /* If we got here and the RE flag is set, it's because we're | |
1193 | dealing with a regexp known to be trivial, so the backslash | |
1194 | just quotes the next character. */ | |
1195 | if (RE && *base_pat == '\\') | |
1196 | { | |
1197 | len--; | |
1198 | len_byte--; | |
1199 | base_pat++; | |
1200 | } | |
1201 | ||
1202 | c = STRING_CHAR_AND_LENGTH (base_pat, len_byte, in_charlen); | |
040272ce | 1203 | |
facdc750 | 1204 | /* Translate the character, if requested. */ |
aff2ce94 | 1205 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1206 | /* If translation changed the byte-length, go back |
1207 | to the original character. */ | |
daaa6ed8 | 1208 | charlen = CHAR_STRING (translated, str); |
facdc750 RS |
1209 | if (in_charlen != charlen) |
1210 | { | |
1211 | translated = c; | |
daaa6ed8 | 1212 | charlen = CHAR_STRING (c, str); |
facdc750 RS |
1213 | } |
1214 | ||
5ffaf437 RS |
1215 | /* If we are searching for something strange, |
1216 | an invalid multibyte code, don't use boyer-moore. */ | |
1217 | if (! ASCII_BYTE_P (translated) | |
1218 | && (charlen == 1 /* 8bit code */ | |
1219 | || charlen != in_charlen /* invalid multibyte code */ | |
1220 | )) | |
1221 | boyer_moore_ok = 0; | |
1222 | ||
aff2ce94 RS |
1223 | TRANSLATE (inverse, inverse_trt, c); |
1224 | ||
facdc750 RS |
1225 | /* Did this char actually get translated? |
1226 | Would any other char get translated into it? */ | |
aff2ce94 | 1227 | if (translated != c || inverse != c) |
facdc750 RS |
1228 | { |
1229 | /* Keep track of which character set row | |
1230 | contains the characters that need translation. */ | |
5ffaf437 | 1231 | int charset_base_code = c & ~CHAR_FIELD3_MASK; |
facdc750 RS |
1232 | if (charset_base == -1) |
1233 | charset_base = charset_base_code; | |
1234 | else if (charset_base != charset_base_code) | |
1235 | /* If two different rows appear, needing translation, | |
1236 | then we cannot use boyer_moore search. */ | |
040272ce | 1237 | boyer_moore_ok = 0; |
aff2ce94 | 1238 | } |
facdc750 RS |
1239 | |
1240 | /* Store this character into the translated pattern. */ | |
1241 | bcopy (str, pat, charlen); | |
1242 | pat += charlen; | |
1243 | base_pat += in_charlen; | |
1244 | len_byte -= in_charlen; | |
1245 | } | |
1246 | } | |
1247 | else | |
1248 | { | |
040272ce KH |
1249 | /* Unibyte buffer. */ |
1250 | charset_base = 0; | |
facdc750 RS |
1251 | while (--len >= 0) |
1252 | { | |
040272ce | 1253 | int c, translated; |
facdc750 RS |
1254 | |
1255 | /* If we got here and the RE flag is set, it's because we're | |
1256 | dealing with a regexp known to be trivial, so the backslash | |
1257 | just quotes the next character. */ | |
1258 | if (RE && *base_pat == '\\') | |
1259 | { | |
1260 | len--; | |
1261 | base_pat++; | |
1262 | } | |
1263 | c = *base_pat++; | |
aff2ce94 | 1264 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1265 | *pat++ = translated; |
1266 | } | |
1267 | } | |
1268 | ||
1269 | len_byte = pat - patbuf; | |
1270 | len = raw_pattern_size; | |
1271 | pat = base_pat = patbuf; | |
1272 | ||
040272ce | 1273 | if (boyer_moore_ok) |
facdc750 | 1274 | return boyer_moore (n, pat, len, len_byte, trt, inverse_trt, |
aff2ce94 RS |
1275 | pos, pos_byte, lim, lim_byte, |
1276 | charset_base); | |
facdc750 RS |
1277 | else |
1278 | return simple_search (n, pat, len, len_byte, trt, | |
1279 | pos, pos_byte, lim, lim_byte); | |
1280 | } | |
1281 | } | |
1282 | \f | |
1283 | /* Do a simple string search N times for the string PAT, | |
1284 | whose length is LEN/LEN_BYTE, | |
1285 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1286 | TRT is the translation table. | |
f8bd51c4 | 1287 | |
facdc750 RS |
1288 | Return the character position where the match is found. |
1289 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1290 | |
facdc750 RS |
1291 | This kind of search works regardless of what is in PAT and |
1292 | regardless of what is in TRT. It is used in cases where | |
1293 | boyer_moore cannot work. */ | |
1294 | ||
1295 | static int | |
1296 | simple_search (n, pat, len, len_byte, trt, pos, pos_byte, lim, lim_byte) | |
1297 | int n; | |
1298 | unsigned char *pat; | |
1299 | int len, len_byte; | |
1300 | Lisp_Object trt; | |
1301 | int pos, pos_byte; | |
1302 | int lim, lim_byte; | |
1303 | { | |
1304 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); | |
ab228c24 | 1305 | int forward = n > 0; |
facdc750 RS |
1306 | |
1307 | if (lim > pos && multibyte) | |
1308 | while (n > 0) | |
1309 | { | |
1310 | while (1) | |
f8bd51c4 | 1311 | { |
facdc750 RS |
1312 | /* Try matching at position POS. */ |
1313 | int this_pos = pos; | |
1314 | int this_pos_byte = pos_byte; | |
1315 | int this_len = len; | |
1316 | int this_len_byte = len_byte; | |
1317 | unsigned char *p = pat; | |
1318 | if (pos + len > lim) | |
1319 | goto stop; | |
1320 | ||
1321 | while (this_len > 0) | |
1322 | { | |
1323 | int charlen, buf_charlen; | |
ab228c24 | 1324 | int pat_ch, buf_ch; |
facdc750 | 1325 | |
ab228c24 | 1326 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1327 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1328 | ZV_BYTE - this_pos_byte, | |
1329 | buf_charlen); | |
aff2ce94 | 1330 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1331 | |
1332 | if (buf_ch != pat_ch) | |
1333 | break; | |
ab228c24 RS |
1334 | |
1335 | this_len_byte -= charlen; | |
1336 | this_len--; | |
1337 | p += charlen; | |
1338 | ||
1339 | this_pos_byte += buf_charlen; | |
1340 | this_pos++; | |
facdc750 RS |
1341 | } |
1342 | ||
1343 | if (this_len == 0) | |
1344 | { | |
1345 | pos += len; | |
1346 | pos_byte += len_byte; | |
1347 | break; | |
1348 | } | |
1349 | ||
1350 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1351 | } |
facdc750 RS |
1352 | |
1353 | n--; | |
1354 | } | |
1355 | else if (lim > pos) | |
1356 | while (n > 0) | |
1357 | { | |
1358 | while (1) | |
f8bd51c4 | 1359 | { |
facdc750 RS |
1360 | /* Try matching at position POS. */ |
1361 | int this_pos = pos; | |
1362 | int this_len = len; | |
1363 | unsigned char *p = pat; | |
1364 | ||
1365 | if (pos + len > lim) | |
1366 | goto stop; | |
1367 | ||
1368 | while (this_len > 0) | |
1369 | { | |
1370 | int pat_ch = *p++; | |
1371 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1372 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1373 | |
1374 | if (buf_ch != pat_ch) | |
1375 | break; | |
ab228c24 RS |
1376 | |
1377 | this_len--; | |
1378 | this_pos++; | |
facdc750 RS |
1379 | } |
1380 | ||
1381 | if (this_len == 0) | |
1382 | { | |
1383 | pos += len; | |
1384 | break; | |
1385 | } | |
1386 | ||
1387 | pos++; | |
f8bd51c4 | 1388 | } |
facdc750 RS |
1389 | |
1390 | n--; | |
1391 | } | |
1392 | /* Backwards search. */ | |
1393 | else if (lim < pos && multibyte) | |
1394 | while (n < 0) | |
1395 | { | |
1396 | while (1) | |
f8bd51c4 | 1397 | { |
facdc750 RS |
1398 | /* Try matching at position POS. */ |
1399 | int this_pos = pos - len; | |
1400 | int this_pos_byte = pos_byte - len_byte; | |
1401 | int this_len = len; | |
1402 | int this_len_byte = len_byte; | |
1403 | unsigned char *p = pat; | |
1404 | ||
1405 | if (pos - len < lim) | |
1406 | goto stop; | |
1407 | ||
1408 | while (this_len > 0) | |
1409 | { | |
1410 | int charlen, buf_charlen; | |
ab228c24 | 1411 | int pat_ch, buf_ch; |
facdc750 | 1412 | |
ab228c24 | 1413 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1414 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1415 | ZV_BYTE - this_pos_byte, | |
1416 | buf_charlen); | |
aff2ce94 | 1417 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1418 | |
1419 | if (buf_ch != pat_ch) | |
1420 | break; | |
ab228c24 RS |
1421 | |
1422 | this_len_byte -= charlen; | |
1423 | this_len--; | |
1424 | p += charlen; | |
1425 | this_pos_byte += buf_charlen; | |
1426 | this_pos++; | |
facdc750 RS |
1427 | } |
1428 | ||
1429 | if (this_len == 0) | |
1430 | { | |
1431 | pos -= len; | |
1432 | pos_byte -= len_byte; | |
1433 | break; | |
1434 | } | |
1435 | ||
1436 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1437 | } |
1438 | ||
facdc750 RS |
1439 | n++; |
1440 | } | |
1441 | else if (lim < pos) | |
1442 | while (n < 0) | |
1443 | { | |
1444 | while (1) | |
b6d6a51c | 1445 | { |
facdc750 RS |
1446 | /* Try matching at position POS. */ |
1447 | int this_pos = pos - len; | |
1448 | int this_len = len; | |
1449 | unsigned char *p = pat; | |
1450 | ||
1451 | if (pos - len < lim) | |
1452 | goto stop; | |
1453 | ||
1454 | while (this_len > 0) | |
1455 | { | |
1456 | int pat_ch = *p++; | |
1457 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1458 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1459 | |
1460 | if (buf_ch != pat_ch) | |
1461 | break; | |
ab228c24 RS |
1462 | this_len--; |
1463 | this_pos++; | |
facdc750 RS |
1464 | } |
1465 | ||
1466 | if (this_len == 0) | |
b6d6a51c | 1467 | { |
facdc750 RS |
1468 | pos -= len; |
1469 | break; | |
b6d6a51c | 1470 | } |
facdc750 RS |
1471 | |
1472 | pos--; | |
b6d6a51c | 1473 | } |
facdc750 RS |
1474 | |
1475 | n++; | |
b6d6a51c | 1476 | } |
facdc750 RS |
1477 | |
1478 | stop: | |
1479 | if (n == 0) | |
aff2ce94 | 1480 | { |
ab228c24 RS |
1481 | if (forward) |
1482 | set_search_regs ((multibyte ? pos_byte : pos) - len_byte, len_byte); | |
1483 | else | |
1484 | set_search_regs (multibyte ? pos_byte : pos, len_byte); | |
aff2ce94 RS |
1485 | |
1486 | return pos; | |
1487 | } | |
facdc750 RS |
1488 | else if (n > 0) |
1489 | return -n; | |
1490 | else | |
1491 | return n; | |
1492 | } | |
1493 | \f | |
1494 | /* Do Boyer-Moore search N times for the string PAT, | |
1495 | whose length is LEN/LEN_BYTE, | |
1496 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1497 | DIRECTION says which direction we search in. | |
1498 | TRT and INVERSE_TRT are translation tables. | |
1499 | ||
1500 | This kind of search works if all the characters in PAT that have | |
1501 | nontrivial translation are the same aside from the last byte. This | |
1502 | makes it possible to translate just the last byte of a character, | |
1503 | and do so after just a simple test of the context. | |
1504 | ||
1505 | If that criterion is not satisfied, do not call this function. */ | |
1506 | ||
1507 | static int | |
1508 | boyer_moore (n, base_pat, len, len_byte, trt, inverse_trt, | |
aff2ce94 | 1509 | pos, pos_byte, lim, lim_byte, charset_base) |
facdc750 RS |
1510 | int n; |
1511 | unsigned char *base_pat; | |
1512 | int len, len_byte; | |
1513 | Lisp_Object trt; | |
1514 | Lisp_Object inverse_trt; | |
1515 | int pos, pos_byte; | |
1516 | int lim, lim_byte; | |
aff2ce94 | 1517 | int charset_base; |
facdc750 RS |
1518 | { |
1519 | int direction = ((n > 0) ? 1 : -1); | |
1520 | register int dirlen; | |
a968f437 | 1521 | int infinity, limit, stride_for_teases = 0; |
facdc750 RS |
1522 | register int *BM_tab; |
1523 | int *BM_tab_base; | |
1524 | register unsigned char *cursor, *p_limit; | |
1525 | register int i, j; | |
cb6792d2 | 1526 | unsigned char *pat, *pat_end; |
facdc750 RS |
1527 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); |
1528 | ||
1529 | unsigned char simple_translate[0400]; | |
6bbd7a29 GM |
1530 | int translate_prev_byte = 0; |
1531 | int translate_anteprev_byte = 0; | |
facdc750 RS |
1532 | |
1533 | #ifdef C_ALLOCA | |
1534 | int BM_tab_space[0400]; | |
1535 | BM_tab = &BM_tab_space[0]; | |
1536 | #else | |
1537 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
1538 | #endif | |
1539 | /* The general approach is that we are going to maintain that we know */ | |
1540 | /* the first (closest to the present position, in whatever direction */ | |
1541 | /* we're searching) character that could possibly be the last */ | |
1542 | /* (furthest from present position) character of a valid match. We */ | |
1543 | /* advance the state of our knowledge by looking at that character */ | |
1544 | /* and seeing whether it indeed matches the last character of the */ | |
1545 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1546 | /* move our pointer (to putative last characters) as far as is */ | |
1547 | /* logically possible. This amount of movement, which I call a */ | |
1548 | /* stride, will be the length of the pattern if the actual character */ | |
1549 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1550 | /* from the last occurrence of that character to the end of the */ | |
1551 | /* pattern. */ | |
1552 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1553 | /* characters that match the last character. This allows use of only */ | |
1554 | /* a single test, a test for having gone past the end of the */ | |
1555 | /* permissible match region, to test for both possible matches (when */ | |
1556 | /* the stride goes past the end immediately) and failure to */ | |
1557 | /* match (where you get nudged past the end one stride at a time). */ | |
1558 | ||
1559 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1560 | /* is determined only by the last character of the putative match. */ | |
1561 | /* If that character does not match, we will stride the proper */ | |
1562 | /* distance to propose a match that superimposes it on the last */ | |
1563 | /* instance of a character that matches it (per trt), or misses */ | |
1564 | /* it entirely if there is none. */ | |
1565 | ||
1566 | dirlen = len_byte * direction; | |
1567 | infinity = dirlen - (lim_byte + pos_byte + len_byte + len_byte) * direction; | |
cb6792d2 RS |
1568 | |
1569 | /* Record position after the end of the pattern. */ | |
1570 | pat_end = base_pat + len_byte; | |
1571 | /* BASE_PAT points to a character that we start scanning from. | |
1572 | It is the first character in a forward search, | |
1573 | the last character in a backward search. */ | |
facdc750 | 1574 | if (direction < 0) |
cb6792d2 RS |
1575 | base_pat = pat_end - 1; |
1576 | ||
facdc750 RS |
1577 | BM_tab_base = BM_tab; |
1578 | BM_tab += 0400; | |
1579 | j = dirlen; /* to get it in a register */ | |
1580 | /* A character that does not appear in the pattern induces a */ | |
1581 | /* stride equal to the pattern length. */ | |
1582 | while (BM_tab_base != BM_tab) | |
1583 | { | |
1584 | *--BM_tab = j; | |
1585 | *--BM_tab = j; | |
1586 | *--BM_tab = j; | |
1587 | *--BM_tab = j; | |
1588 | } | |
1589 | ||
1590 | /* We use this for translation, instead of TRT itself. | |
1591 | We fill this in to handle the characters that actually | |
1592 | occur in the pattern. Others don't matter anyway! */ | |
1593 | bzero (simple_translate, sizeof simple_translate); | |
1594 | for (i = 0; i < 0400; i++) | |
1595 | simple_translate[i] = i; | |
1596 | ||
1597 | i = 0; | |
1598 | while (i != infinity) | |
1599 | { | |
cb6792d2 | 1600 | unsigned char *ptr = base_pat + i; |
facdc750 RS |
1601 | i += direction; |
1602 | if (i == dirlen) | |
1603 | i = infinity; | |
1604 | if (! NILP (trt)) | |
ca1d1d23 | 1605 | { |
facdc750 | 1606 | int ch; |
aff2ce94 | 1607 | int untranslated; |
facdc750 RS |
1608 | int this_translated = 1; |
1609 | ||
1610 | if (multibyte | |
cb6792d2 RS |
1611 | /* Is *PTR the last byte of a character? */ |
1612 | && (pat_end - ptr == 1 || CHAR_HEAD_P (ptr[1]))) | |
ca1d1d23 | 1613 | { |
facdc750 RS |
1614 | unsigned char *charstart = ptr; |
1615 | while (! CHAR_HEAD_P (*charstart)) | |
1616 | charstart--; | |
aff2ce94 | 1617 | untranslated = STRING_CHAR (charstart, ptr - charstart + 1); |
6397418a | 1618 | if (charset_base == (untranslated & ~CHAR_FIELD3_MASK)) |
facdc750 | 1619 | { |
ab228c24 | 1620 | TRANSLATE (ch, trt, untranslated); |
aff2ce94 RS |
1621 | if (! CHAR_HEAD_P (*ptr)) |
1622 | { | |
1623 | translate_prev_byte = ptr[-1]; | |
1624 | if (! CHAR_HEAD_P (translate_prev_byte)) | |
1625 | translate_anteprev_byte = ptr[-2]; | |
1626 | } | |
facdc750 | 1627 | } |
aff2ce94 | 1628 | else |
ab228c24 RS |
1629 | { |
1630 | this_translated = 0; | |
1631 | ch = *ptr; | |
1632 | } | |
ca1d1d23 | 1633 | } |
facdc750 | 1634 | else if (!multibyte) |
aff2ce94 | 1635 | TRANSLATE (ch, trt, *ptr); |
ca1d1d23 JB |
1636 | else |
1637 | { | |
facdc750 RS |
1638 | ch = *ptr; |
1639 | this_translated = 0; | |
ca1d1d23 | 1640 | } |
facdc750 | 1641 | |
ab228c24 RS |
1642 | if (ch > 0400) |
1643 | j = ((unsigned char) ch) | 0200; | |
1644 | else | |
1645 | j = (unsigned char) ch; | |
1646 | ||
facdc750 RS |
1647 | if (i == infinity) |
1648 | stride_for_teases = BM_tab[j]; | |
ab228c24 | 1649 | |
facdc750 RS |
1650 | BM_tab[j] = dirlen - i; |
1651 | /* A translation table is accompanied by its inverse -- see */ | |
1652 | /* comment following downcase_table for details */ | |
1653 | if (this_translated) | |
ab228c24 RS |
1654 | { |
1655 | int starting_ch = ch; | |
1656 | int starting_j = j; | |
1657 | while (1) | |
1658 | { | |
1659 | TRANSLATE (ch, inverse_trt, ch); | |
1660 | if (ch > 0400) | |
1661 | j = ((unsigned char) ch) | 0200; | |
1662 | else | |
1663 | j = (unsigned char) ch; | |
1664 | ||
1665 | /* For all the characters that map into CH, | |
1666 | set up simple_translate to map the last byte | |
1667 | into STARTING_J. */ | |
1668 | simple_translate[j] = starting_j; | |
1669 | if (ch == starting_ch) | |
1670 | break; | |
1671 | BM_tab[j] = dirlen - i; | |
1672 | } | |
1673 | } | |
facdc750 RS |
1674 | } |
1675 | else | |
1676 | { | |
1677 | j = *ptr; | |
1678 | ||
1679 | if (i == infinity) | |
1680 | stride_for_teases = BM_tab[j]; | |
1681 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1682 | } |
facdc750 RS |
1683 | /* stride_for_teases tells how much to stride if we get a */ |
1684 | /* match on the far character but are subsequently */ | |
1685 | /* disappointed, by recording what the stride would have been */ | |
1686 | /* for that character if the last character had been */ | |
1687 | /* different. */ | |
1688 | } | |
1689 | infinity = dirlen - infinity; | |
1690 | pos_byte += dirlen - ((direction > 0) ? direction : 0); | |
1691 | /* loop invariant - POS_BYTE points at where last char (first | |
1692 | char if reverse) of pattern would align in a possible match. */ | |
1693 | while (n != 0) | |
1694 | { | |
1695 | int tail_end; | |
1696 | unsigned char *tail_end_ptr; | |
1697 | ||
1698 | /* It's been reported that some (broken) compiler thinks that | |
1699 | Boolean expressions in an arithmetic context are unsigned. | |
1700 | Using an explicit ?1:0 prevents this. */ | |
1701 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1702 | < 0) | |
1703 | return (n * (0 - direction)); | |
1704 | /* First we do the part we can by pointers (maybe nothing) */ | |
1705 | QUIT; | |
1706 | pat = base_pat; | |
1707 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1708 | if (direction > 0) |
1709 | { | |
1710 | limit = BUFFER_CEILING_OF (limit); | |
1711 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1712 | can take on without hitting edge of buffer or the gap. */ | |
1713 | limit = min (limit, pos_byte + 20000); | |
1714 | limit = min (limit, lim_byte - 1); | |
1715 | } | |
1716 | else | |
1717 | { | |
1718 | limit = BUFFER_FLOOR_OF (limit); | |
1719 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1720 | can take on without hitting edge of buffer or the gap. */ | |
1721 | limit = max (limit, pos_byte - 20000); | |
1722 | limit = max (limit, lim_byte); | |
1723 | } | |
facdc750 RS |
1724 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1725 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1726 | ||
1727 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1728 | { |
facdc750 RS |
1729 | unsigned char *p2; |
1730 | ||
1731 | p_limit = BYTE_POS_ADDR (limit); | |
1732 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
1733 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ | |
1734 | while (1) /* use one cursor setting as long as i can */ | |
ca1d1d23 | 1735 | { |
facdc750 | 1736 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1737 | { |
facdc750 RS |
1738 | /* Use signed comparison if appropriate |
1739 | to make cursor+infinity sure to be > p_limit. | |
1740 | Assuming that the buffer lies in a range of addresses | |
1741 | that are all "positive" (as ints) or all "negative", | |
1742 | either kind of comparison will work as long | |
1743 | as we don't step by infinity. So pick the kind | |
1744 | that works when we do step by infinity. */ | |
1745 | if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit) | |
1746 | while ((EMACS_INT) cursor <= (EMACS_INT) p_limit) | |
1747 | cursor += BM_tab[*cursor]; | |
ca1d1d23 | 1748 | else |
facdc750 RS |
1749 | while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit) |
1750 | cursor += BM_tab[*cursor]; | |
1751 | } | |
1752 | else | |
1753 | { | |
1754 | if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit) | |
1755 | while ((EMACS_INT) cursor >= (EMACS_INT) p_limit) | |
1756 | cursor += BM_tab[*cursor]; | |
1757 | else | |
1758 | while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit) | |
1759 | cursor += BM_tab[*cursor]; | |
1760 | } | |
ca1d1d23 | 1761 | /* If you are here, cursor is beyond the end of the searched region. */ |
facdc750 RS |
1762 | /* This can happen if you match on the far character of the pattern, */ |
1763 | /* because the "stride" of that character is infinity, a number able */ | |
1764 | /* to throw you well beyond the end of the search. It can also */ | |
1765 | /* happen if you fail to match within the permitted region and would */ | |
1766 | /* otherwise try a character beyond that region */ | |
1767 | if ((cursor - p_limit) * direction <= len_byte) | |
1768 | break; /* a small overrun is genuine */ | |
1769 | cursor -= infinity; /* large overrun = hit */ | |
1770 | i = dirlen - direction; | |
1771 | if (! NILP (trt)) | |
1772 | { | |
1773 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1774 | { |
facdc750 RS |
1775 | int ch; |
1776 | cursor -= direction; | |
1777 | /* Translate only the last byte of a character. */ | |
1778 | if (! multibyte | |
1779 | || ((cursor == tail_end_ptr | |
1780 | || CHAR_HEAD_P (cursor[1])) | |
1781 | && (CHAR_HEAD_P (cursor[0]) | |
1782 | || (translate_prev_byte == cursor[-1] | |
1783 | && (CHAR_HEAD_P (translate_prev_byte) | |
1784 | || translate_anteprev_byte == cursor[-2]))))) | |
1785 | ch = simple_translate[*cursor]; | |
1786 | else | |
1787 | ch = *cursor; | |
1788 | if (pat[i] != ch) | |
1789 | break; | |
ca1d1d23 | 1790 | } |
facdc750 RS |
1791 | } |
1792 | else | |
1793 | { | |
1794 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1795 | { |
facdc750 RS |
1796 | cursor -= direction; |
1797 | if (pat[i] != *cursor) | |
1798 | break; | |
ca1d1d23 | 1799 | } |
facdc750 RS |
1800 | } |
1801 | cursor += dirlen - i - direction; /* fix cursor */ | |
1802 | if (i + direction == 0) | |
1803 | { | |
1804 | int position; | |
0c8533c6 | 1805 | |
facdc750 | 1806 | cursor -= direction; |
1113d9db | 1807 | |
facdc750 RS |
1808 | position = pos_byte + cursor - p2 + ((direction > 0) |
1809 | ? 1 - len_byte : 0); | |
1810 | set_search_regs (position, len_byte); | |
ca325161 | 1811 | |
facdc750 RS |
1812 | if ((n -= direction) != 0) |
1813 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1814 | else |
facdc750 RS |
1815 | return ((direction > 0) |
1816 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1817 | } |
facdc750 RS |
1818 | else |
1819 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1820 | } |
facdc750 RS |
1821 | pos_byte += cursor - p2; |
1822 | } | |
1823 | else | |
1824 | /* Now we'll pick up a clump that has to be done the hard */ | |
1825 | /* way because it covers a discontinuity */ | |
1826 | { | |
1827 | limit = ((direction > 0) | |
1828 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1829 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1830 | limit = ((direction > 0) | |
1831 | ? min (limit + len_byte, lim_byte - 1) | |
1832 | : max (limit - len_byte, lim_byte)); | |
1833 | /* LIMIT is now the last value POS_BYTE can have | |
1834 | and still be valid for a possible match. */ | |
1835 | while (1) | |
ca1d1d23 | 1836 | { |
facdc750 RS |
1837 | /* This loop can be coded for space rather than */ |
1838 | /* speed because it will usually run only once. */ | |
1839 | /* (the reach is at most len + 21, and typically */ | |
1840 | /* does not exceed len) */ | |
1841 | while ((limit - pos_byte) * direction >= 0) | |
1842 | pos_byte += BM_tab[FETCH_BYTE (pos_byte)]; | |
1843 | /* now run the same tests to distinguish going off the */ | |
1844 | /* end, a match or a phony match. */ | |
1845 | if ((pos_byte - limit) * direction <= len_byte) | |
1846 | break; /* ran off the end */ | |
1847 | /* Found what might be a match. | |
1848 | Set POS_BYTE back to last (first if reverse) pos. */ | |
1849 | pos_byte -= infinity; | |
1850 | i = dirlen - direction; | |
1851 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1852 | { |
facdc750 RS |
1853 | int ch; |
1854 | unsigned char *ptr; | |
1855 | pos_byte -= direction; | |
1856 | ptr = BYTE_POS_ADDR (pos_byte); | |
1857 | /* Translate only the last byte of a character. */ | |
1858 | if (! multibyte | |
1859 | || ((ptr == tail_end_ptr | |
1860 | || CHAR_HEAD_P (ptr[1])) | |
1861 | && (CHAR_HEAD_P (ptr[0]) | |
1862 | || (translate_prev_byte == ptr[-1] | |
1863 | && (CHAR_HEAD_P (translate_prev_byte) | |
1864 | || translate_anteprev_byte == ptr[-2]))))) | |
1865 | ch = simple_translate[*ptr]; | |
1866 | else | |
1867 | ch = *ptr; | |
1868 | if (pat[i] != ch) | |
1869 | break; | |
1870 | } | |
1871 | /* Above loop has moved POS_BYTE part or all the way | |
1872 | back to the first pos (last pos if reverse). | |
1873 | Set it once again at the last (first if reverse) char. */ | |
1874 | pos_byte += dirlen - i- direction; | |
1875 | if (i + direction == 0) | |
1876 | { | |
1877 | int position; | |
1878 | pos_byte -= direction; | |
1113d9db | 1879 | |
facdc750 | 1880 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
0c8533c6 | 1881 | |
facdc750 | 1882 | set_search_regs (position, len_byte); |
ca325161 | 1883 | |
facdc750 RS |
1884 | if ((n -= direction) != 0) |
1885 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 1886 | else |
facdc750 RS |
1887 | return ((direction > 0) |
1888 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1889 | } |
facdc750 RS |
1890 | else |
1891 | pos_byte += stride_for_teases; | |
1892 | } | |
1893 | } | |
1894 | /* We have done one clump. Can we continue? */ | |
1895 | if ((lim_byte - pos_byte) * direction < 0) | |
1896 | return ((0 - n) * direction); | |
ca1d1d23 | 1897 | } |
facdc750 | 1898 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 1899 | } |
ca325161 | 1900 | |
fa8ed3e0 | 1901 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
1902 | for the overall match just found in the current buffer. |
1903 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
1904 | |
1905 | static void | |
fa8ed3e0 RS |
1906 | set_search_regs (beg_byte, nbytes) |
1907 | int beg_byte, nbytes; | |
ca325161 | 1908 | { |
a7e4cdde RS |
1909 | int i; |
1910 | ||
ca325161 RS |
1911 | /* Make sure we have registers in which to store |
1912 | the match position. */ | |
1913 | if (search_regs.num_regs == 0) | |
1914 | { | |
2d4a771a RS |
1915 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
1916 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 1917 | search_regs.num_regs = 2; |
ca325161 RS |
1918 | } |
1919 | ||
a7e4cdde RS |
1920 | /* Clear out the other registers. */ |
1921 | for (i = 1; i < search_regs.num_regs; i++) | |
1922 | { | |
1923 | search_regs.start[i] = -1; | |
1924 | search_regs.end[i] = -1; | |
1925 | } | |
1926 | ||
fa8ed3e0 RS |
1927 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
1928 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 1929 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 1930 | } |
ca1d1d23 JB |
1931 | \f |
1932 | /* Given a string of words separated by word delimiters, | |
1933 | compute a regexp that matches those exact words | |
1934 | separated by arbitrary punctuation. */ | |
1935 | ||
1936 | static Lisp_Object | |
1937 | wordify (string) | |
1938 | Lisp_Object string; | |
1939 | { | |
1940 | register unsigned char *p, *o; | |
0c8533c6 | 1941 | register int i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 1942 | Lisp_Object val; |
0c8533c6 RS |
1943 | int prev_c = 0; |
1944 | int adjust; | |
ca1d1d23 JB |
1945 | |
1946 | CHECK_STRING (string, 0); | |
1947 | p = XSTRING (string)->data; | |
1948 | len = XSTRING (string)->size; | |
1949 | ||
0c8533c6 RS |
1950 | for (i = 0, i_byte = 0; i < len; ) |
1951 | { | |
1952 | int c; | |
1953 | ||
eb99a8dd | 1954 | FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte); |
0c8533c6 RS |
1955 | |
1956 | if (SYNTAX (c) != Sword) | |
1957 | { | |
1958 | punct_count++; | |
1959 | if (i > 0 && SYNTAX (prev_c) == Sword) | |
1960 | word_count++; | |
1961 | } | |
ca1d1d23 | 1962 | |
0c8533c6 RS |
1963 | prev_c = c; |
1964 | } | |
1965 | ||
1966 | if (SYNTAX (prev_c) == Sword) | |
1967 | word_count++; | |
1968 | if (!word_count) | |
b07b65aa | 1969 | return empty_string; |
0c8533c6 RS |
1970 | |
1971 | adjust = - punct_count + 5 * (word_count - 1) + 4; | |
8a2df937 RS |
1972 | if (STRING_MULTIBYTE (string)) |
1973 | val = make_uninit_multibyte_string (len + adjust, | |
1974 | STRING_BYTES (XSTRING (string)) | |
1975 | + adjust); | |
1976 | else | |
1977 | val = make_uninit_string (len + adjust); | |
ca1d1d23 JB |
1978 | |
1979 | o = XSTRING (val)->data; | |
1980 | *o++ = '\\'; | |
1981 | *o++ = 'b'; | |
1e9582d4 | 1982 | prev_c = 0; |
ca1d1d23 | 1983 | |
1e9582d4 RS |
1984 | for (i = 0, i_byte = 0; i < len; ) |
1985 | { | |
1986 | int c; | |
1987 | int i_byte_orig = i_byte; | |
1988 | ||
eb99a8dd | 1989 | FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte); |
1e9582d4 RS |
1990 | |
1991 | if (SYNTAX (c) == Sword) | |
1992 | { | |
1993 | bcopy (&XSTRING (string)->data[i_byte_orig], o, | |
1994 | i_byte - i_byte_orig); | |
1995 | o += i_byte - i_byte_orig; | |
1996 | } | |
1997 | else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count) | |
1998 | { | |
1999 | *o++ = '\\'; | |
2000 | *o++ = 'W'; | |
2001 | *o++ = '\\'; | |
2002 | *o++ = 'W'; | |
2003 | *o++ = '*'; | |
2004 | } | |
2005 | ||
2006 | prev_c = c; | |
2007 | } | |
ca1d1d23 JB |
2008 | |
2009 | *o++ = '\\'; | |
2010 | *o++ = 'b'; | |
2011 | ||
2012 | return val; | |
2013 | } | |
2014 | \f | |
2015 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
8c1a1077 PJ |
2016 | "MSearch backward: ", |
2017 | doc: /* Search backward from point for STRING. | |
2018 | Set point to the beginning of the occurrence found, and return point. | |
2019 | An optional second argument bounds the search; it is a buffer position. | |
2020 | The match found must not extend before that position. | |
2021 | Optional third argument, if t, means if fail just return nil (no error). | |
2022 | If not nil and not t, position at limit of search and return nil. | |
2023 | Optional fourth argument is repeat count--search for successive occurrences. | |
2024 | ||
2025 | Search case-sensitivity is determined by the value of the variable | |
2026 | `case-fold-search', which see. | |
2027 | ||
2028 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2029 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2030 | Lisp_Object string, bound, noerror, count; |
2031 | { | |
b819a390 | 2032 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2033 | } |
2034 | ||
6af43974 | 2035 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2036 | doc: /* Search forward from point for STRING. |
2037 | Set point to the end of the occurrence found, and return point. | |
2038 | An optional second argument bounds the search; it is a buffer position. | |
2039 | The match found must not extend after that position. nil is equivalent | |
2040 | to (point-max). | |
2041 | Optional third argument, if t, means if fail just return nil (no error). | |
2042 | If not nil and not t, move to limit of search and return nil. | |
2043 | Optional fourth argument is repeat count--search for successive occurrences. | |
2044 | ||
2045 | Search case-sensitivity is determined by the value of the variable | |
2046 | `case-fold-search', which see. | |
2047 | ||
2048 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2049 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2050 | Lisp_Object string, bound, noerror, count; |
2051 | { | |
b819a390 | 2052 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2053 | } |
2054 | ||
2055 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
8c1a1077 PJ |
2056 | "sWord search backward: ", |
2057 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2058 | Set point to the beginning of the occurrence found, and return point. | |
2059 | An optional second argument bounds the search; it is a buffer position. | |
2060 | The match found must not extend before that position. | |
2061 | Optional third argument, if t, means if fail just return nil (no error). | |
2062 | If not nil and not t, move to limit of search and return nil. | |
2063 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2064 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2065 | Lisp_Object string, bound, noerror, count; |
2066 | { | |
b819a390 | 2067 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2068 | } |
2069 | ||
2070 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
8c1a1077 PJ |
2071 | "sWord search: ", |
2072 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2073 | Set point to the end of the occurrence found, and return point. | |
2074 | An optional second argument bounds the search; it is a buffer position. | |
2075 | The match found must not extend after that position. | |
2076 | Optional third argument, if t, means if fail just return nil (no error). | |
2077 | If not nil and not t, move to limit of search and return nil. | |
2078 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2079 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2080 | Lisp_Object string, bound, noerror, count; |
2081 | { | |
b819a390 | 2082 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
2083 | } |
2084 | ||
2085 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
8c1a1077 PJ |
2086 | "sRE search backward: ", |
2087 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2088 | Set point to the beginning of the match, and return point. | |
2089 | The match found is the one starting last in the buffer | |
2090 | and yet ending before the origin of the search. | |
2091 | An optional second argument bounds the search; it is a buffer position. | |
2092 | The match found must start at or after that position. | |
2093 | Optional third argument, if t, means if fail just return nil (no error). | |
2094 | If not nil and not t, move to limit of search and return nil. | |
2095 | Optional fourth argument is repeat count--search for successive occurrences. | |
2096 | See also the functions `match-beginning', `match-end', `match-string', | |
2097 | and `replace-match'. */) | |
2098 | (regexp, bound, noerror, count) | |
19c0a730 | 2099 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2100 | { |
b819a390 | 2101 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2102 | } |
2103 | ||
2104 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2105 | "sRE search: ", |
2106 | doc: /* Search forward from point for regular expression REGEXP. | |
2107 | Set point to the end of the occurrence found, and return point. | |
2108 | An optional second argument bounds the search; it is a buffer position. | |
2109 | The match found must not extend after that position. | |
2110 | Optional third argument, if t, means if fail just return nil (no error). | |
2111 | If not nil and not t, move to limit of search and return nil. | |
2112 | Optional fourth argument is repeat count--search for successive occurrences. | |
2113 | See also the functions `match-beginning', `match-end', `match-string', | |
2114 | and `replace-match'. */) | |
2115 | (regexp, bound, noerror, count) | |
19c0a730 | 2116 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2117 | { |
b819a390 RS |
2118 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2119 | } | |
2120 | ||
2121 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2122 | "sPosix search backward: ", |
2123 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2124 | Find the longest match in accord with Posix regular expression rules. | |
2125 | Set point to the beginning of the match, and return point. | |
2126 | The match found is the one starting last in the buffer | |
2127 | and yet ending before the origin of the search. | |
2128 | An optional second argument bounds the search; it is a buffer position. | |
2129 | The match found must start at or after that position. | |
2130 | Optional third argument, if t, means if fail just return nil (no error). | |
2131 | If not nil and not t, move to limit of search and return nil. | |
2132 | Optional fourth argument is repeat count--search for successive occurrences. | |
2133 | See also the functions `match-beginning', `match-end', `match-string', | |
2134 | and `replace-match'. */) | |
2135 | (regexp, bound, noerror, count) | |
b819a390 RS |
2136 | Lisp_Object regexp, bound, noerror, count; |
2137 | { | |
2138 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2139 | } | |
2140 | ||
2141 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2142 | "sPosix search: ", |
2143 | doc: /* Search forward from point for regular expression REGEXP. | |
2144 | Find the longest match in accord with Posix regular expression rules. | |
2145 | Set point to the end of the occurrence found, and return point. | |
2146 | An optional second argument bounds the search; it is a buffer position. | |
2147 | The match found must not extend after that position. | |
2148 | Optional third argument, if t, means if fail just return nil (no error). | |
2149 | If not nil and not t, move to limit of search and return nil. | |
2150 | Optional fourth argument is repeat count--search for successive occurrences. | |
2151 | See also the functions `match-beginning', `match-end', `match-string', | |
2152 | and `replace-match'. */) | |
2153 | (regexp, bound, noerror, count) | |
b819a390 RS |
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, |
8c1a1077 PJ |
2160 | doc: /* Replace text matched by last search with NEWTEXT. |
2161 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. | |
2162 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2163 | based on the replaced text. | |
2164 | If the replaced text has only capital letters | |
2165 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
2166 | If the replaced text has at least one word starting with a capital letter, | |
2167 | then capitalize each word in NEWTEXT. | |
2168 | ||
2169 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2170 | Otherwise treat `\\' as special: | |
2171 | `\\&' in NEWTEXT means substitute original matched text. | |
2172 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2173 | If Nth parens didn't match, substitute nothing. | |
2174 | `\\\\' means insert one `\\'. | |
2175 | FIXEDCASE and LITERAL are optional arguments. | |
2176 | Leaves point at end of replacement text. | |
2177 | ||
2178 | The optional fourth argument STRING can be a string to modify. | |
2179 | This is meaningful when the previous match was done against STRING, | |
2180 | using `string-match'. When used this way, `replace-match' | |
2181 | creates and returns a new string made by copying STRING and replacing | |
2182 | the part of STRING that was matched. | |
2183 | ||
2184 | The optional fifth argument SUBEXP specifies a subexpression; | |
2185 | it says to replace just that subexpression with NEWTEXT, | |
2186 | rather than replacing the entire matched text. | |
2187 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2188 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2189 | NEWTEXT in place of subexp N. | |
2190 | This is useful only after a regular expression search or match, | |
2191 | since only regular expressions have distinguished subexpressions. */) | |
2192 | (newtext, fixedcase, literal, string, subexp) | |
d7a5ad5f | 2193 | Lisp_Object newtext, fixedcase, literal, string, subexp; |
ca1d1d23 JB |
2194 | { |
2195 | enum { nochange, all_caps, cap_initial } case_action; | |
ac3b28b1 | 2196 | register int pos, pos_byte; |
ca1d1d23 | 2197 | int some_multiletter_word; |
97832bd0 | 2198 | int some_lowercase; |
73dc8771 | 2199 | int some_uppercase; |
208767c3 | 2200 | int some_nonuppercase_initial; |
ca1d1d23 JB |
2201 | register int c, prevc; |
2202 | int inslen; | |
d7a5ad5f | 2203 | int sub; |
3e18eecf | 2204 | int opoint, newpoint; |
ca1d1d23 | 2205 | |
16fdc568 | 2206 | CHECK_STRING (newtext, 0); |
ca1d1d23 | 2207 | |
080c45fd RS |
2208 | if (! NILP (string)) |
2209 | CHECK_STRING (string, 4); | |
2210 | ||
ca1d1d23 JB |
2211 | case_action = nochange; /* We tried an initialization */ |
2212 | /* but some C compilers blew it */ | |
4746118a JB |
2213 | |
2214 | if (search_regs.num_regs <= 0) | |
2215 | error ("replace-match called before any match found"); | |
2216 | ||
d7a5ad5f RS |
2217 | if (NILP (subexp)) |
2218 | sub = 0; | |
2219 | else | |
2220 | { | |
2221 | CHECK_NUMBER (subexp, 3); | |
2222 | sub = XINT (subexp); | |
2223 | if (sub < 0 || sub >= search_regs.num_regs) | |
2224 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2225 | } | |
2226 | ||
080c45fd RS |
2227 | if (NILP (string)) |
2228 | { | |
d7a5ad5f RS |
2229 | if (search_regs.start[sub] < BEGV |
2230 | || search_regs.start[sub] > search_regs.end[sub] | |
2231 | || search_regs.end[sub] > ZV) | |
2232 | args_out_of_range (make_number (search_regs.start[sub]), | |
2233 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2234 | } |
2235 | else | |
2236 | { | |
d7a5ad5f RS |
2237 | if (search_regs.start[sub] < 0 |
2238 | || search_regs.start[sub] > search_regs.end[sub] | |
2239 | || search_regs.end[sub] > XSTRING (string)->size) | |
2240 | args_out_of_range (make_number (search_regs.start[sub]), | |
2241 | make_number (search_regs.end[sub])); | |
080c45fd | 2242 | } |
ca1d1d23 JB |
2243 | |
2244 | if (NILP (fixedcase)) | |
2245 | { | |
2246 | /* Decide how to casify by examining the matched text. */ | |
ac3b28b1 | 2247 | int last; |
ca1d1d23 | 2248 | |
ac3b28b1 KH |
2249 | pos = search_regs.start[sub]; |
2250 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2251 | |
2252 | if (NILP (string)) | |
ac3b28b1 | 2253 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2254 | else |
ac3b28b1 | 2255 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2256 | |
ca1d1d23 JB |
2257 | prevc = '\n'; |
2258 | case_action = all_caps; | |
2259 | ||
2260 | /* some_multiletter_word is set nonzero if any original word | |
2261 | is more than one letter long. */ | |
2262 | some_multiletter_word = 0; | |
97832bd0 | 2263 | some_lowercase = 0; |
208767c3 | 2264 | some_nonuppercase_initial = 0; |
73dc8771 | 2265 | some_uppercase = 0; |
ca1d1d23 | 2266 | |
ac3b28b1 | 2267 | while (pos < last) |
ca1d1d23 | 2268 | { |
080c45fd | 2269 | if (NILP (string)) |
ac3b28b1 KH |
2270 | { |
2271 | c = FETCH_CHAR (pos_byte); | |
2272 | INC_BOTH (pos, pos_byte); | |
2273 | } | |
080c45fd | 2274 | else |
ac3b28b1 | 2275 | FETCH_STRING_CHAR_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2276 | |
ca1d1d23 JB |
2277 | if (LOWERCASEP (c)) |
2278 | { | |
2279 | /* Cannot be all caps if any original char is lower case */ | |
2280 | ||
97832bd0 | 2281 | some_lowercase = 1; |
ca1d1d23 | 2282 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2283 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2284 | else |
2285 | some_multiletter_word = 1; | |
2286 | } | |
2287 | else if (!NOCASEP (c)) | |
2288 | { | |
73dc8771 | 2289 | some_uppercase = 1; |
97832bd0 | 2290 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2291 | ; |
97832bd0 | 2292 | else |
ca1d1d23 JB |
2293 | some_multiletter_word = 1; |
2294 | } | |
208767c3 RS |
2295 | else |
2296 | { | |
2297 | /* If the initial is a caseless word constituent, | |
2298 | treat that like a lowercase initial. */ | |
2299 | if (SYNTAX (prevc) != Sword) | |
2300 | some_nonuppercase_initial = 1; | |
2301 | } | |
ca1d1d23 JB |
2302 | |
2303 | prevc = c; | |
2304 | } | |
2305 | ||
97832bd0 RS |
2306 | /* Convert to all caps if the old text is all caps |
2307 | and has at least one multiletter word. */ | |
2308 | if (! some_lowercase && some_multiletter_word) | |
2309 | case_action = all_caps; | |
c4d460ce | 2310 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2311 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2312 | case_action = cap_initial; |
208767c3 | 2313 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2314 | /* Should x -> yz, operating on X, give Yz or YZ? |
2315 | We'll assume the latter. */ | |
2316 | case_action = all_caps; | |
97832bd0 RS |
2317 | else |
2318 | case_action = nochange; | |
ca1d1d23 JB |
2319 | } |
2320 | ||
080c45fd RS |
2321 | /* Do replacement in a string. */ |
2322 | if (!NILP (string)) | |
2323 | { | |
2324 | Lisp_Object before, after; | |
2325 | ||
2326 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2327 | make_number (search_regs.start[sub])); |
2328 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2329 | |
636a5e28 RS |
2330 | /* Substitute parts of the match into NEWTEXT |
2331 | if desired. */ | |
080c45fd RS |
2332 | if (NILP (literal)) |
2333 | { | |
d131e79c RS |
2334 | int lastpos = 0; |
2335 | int lastpos_byte = 0; | |
080c45fd RS |
2336 | /* We build up the substituted string in ACCUM. */ |
2337 | Lisp_Object accum; | |
2338 | Lisp_Object middle; | |
ac3b28b1 | 2339 | int length = STRING_BYTES (XSTRING (newtext)); |
080c45fd RS |
2340 | |
2341 | accum = Qnil; | |
2342 | ||
ac3b28b1 | 2343 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd RS |
2344 | { |
2345 | int substart = -1; | |
6bbd7a29 | 2346 | int subend = 0; |
1e79ec24 | 2347 | int delbackslash = 0; |
080c45fd | 2348 | |
0c8533c6 RS |
2349 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2350 | ||
080c45fd RS |
2351 | if (c == '\\') |
2352 | { | |
0c8533c6 | 2353 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
eb99a8dd | 2354 | |
080c45fd RS |
2355 | if (c == '&') |
2356 | { | |
d7a5ad5f RS |
2357 | substart = search_regs.start[sub]; |
2358 | subend = search_regs.end[sub]; | |
080c45fd RS |
2359 | } |
2360 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') | |
2361 | { | |
ad10348f | 2362 | if (search_regs.start[c - '0'] >= 0) |
080c45fd RS |
2363 | { |
2364 | substart = search_regs.start[c - '0']; | |
2365 | subend = search_regs.end[c - '0']; | |
2366 | } | |
2367 | } | |
1e79ec24 KH |
2368 | else if (c == '\\') |
2369 | delbackslash = 1; | |
636a5e28 RS |
2370 | else |
2371 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2372 | } |
2373 | if (substart >= 0) | |
2374 | { | |
d131e79c RS |
2375 | if (pos - 2 != lastpos) |
2376 | middle = substring_both (newtext, lastpos, | |
2377 | lastpos_byte, | |
2378 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2379 | else |
2380 | middle = Qnil; | |
2381 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2382 | Fsubstring (string, |
2383 | make_number (substart), | |
080c45fd RS |
2384 | make_number (subend))); |
2385 | lastpos = pos; | |
0c8533c6 | 2386 | lastpos_byte = pos_byte; |
080c45fd | 2387 | } |
1e79ec24 KH |
2388 | else if (delbackslash) |
2389 | { | |
d131e79c RS |
2390 | middle = substring_both (newtext, lastpos, |
2391 | lastpos_byte, | |
2392 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2393 | |
1e79ec24 KH |
2394 | accum = concat2 (accum, middle); |
2395 | lastpos = pos; | |
0c8533c6 | 2396 | lastpos_byte = pos_byte; |
1e79ec24 | 2397 | } |
080c45fd RS |
2398 | } |
2399 | ||
d131e79c RS |
2400 | if (pos != lastpos) |
2401 | middle = substring_both (newtext, lastpos, | |
2402 | lastpos_byte, | |
0c8533c6 | 2403 | pos, pos_byte); |
080c45fd RS |
2404 | else |
2405 | middle = Qnil; | |
2406 | ||
2407 | newtext = concat2 (accum, middle); | |
2408 | } | |
2409 | ||
636a5e28 | 2410 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2411 | if (case_action == all_caps) |
2412 | newtext = Fupcase (newtext); | |
2413 | else if (case_action == cap_initial) | |
2b2eead9 | 2414 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2415 | |
2416 | return concat3 (before, newtext, after); | |
2417 | } | |
2418 | ||
09c4719e | 2419 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2420 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2421 | opoint = PT - ZV; |
9160906f RS |
2422 | else if (PT > search_regs.start[sub]) |
2423 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2424 | else |
2425 | opoint = PT; | |
2426 | ||
fa8ed3e0 | 2427 | TEMP_SET_PT (search_regs.start[sub]); |
b0eba991 | 2428 | |
9a76659d JB |
2429 | /* We insert the replacement text before the old text, and then |
2430 | delete the original text. This means that markers at the | |
2431 | beginning or end of the original will float to the corresponding | |
2432 | position in the replacement. */ | |
ca1d1d23 | 2433 | if (!NILP (literal)) |
16fdc568 | 2434 | Finsert_and_inherit (1, &newtext); |
ca1d1d23 JB |
2435 | else |
2436 | { | |
ac3b28b1 | 2437 | int length = STRING_BYTES (XSTRING (newtext)); |
68e69fbd RS |
2438 | unsigned char *substed; |
2439 | int substed_alloc_size, substed_len; | |
3bc25e52 KH |
2440 | int buf_multibyte = !NILP (current_buffer->enable_multibyte_characters); |
2441 | int str_multibyte = STRING_MULTIBYTE (newtext); | |
2442 | Lisp_Object rev_tbl; | |
2443 | ||
2444 | rev_tbl= (!buf_multibyte && CHAR_TABLE_P (Vnonascii_translation_table) | |
2445 | ? Fchar_table_extra_slot (Vnonascii_translation_table, | |
2446 | make_number (0)) | |
2447 | : Qnil); | |
ac3b28b1 | 2448 | |
68e69fbd RS |
2449 | substed_alloc_size = length * 2 + 100; |
2450 | substed = (unsigned char *) xmalloc (substed_alloc_size + 1); | |
2451 | substed_len = 0; | |
2452 | ||
3bc25e52 KH |
2453 | /* Go thru NEWTEXT, producing the actual text to insert in |
2454 | SUBSTED while adjusting multibyteness to that of the current | |
2455 | buffer. */ | |
ca1d1d23 | 2456 | |
ac3b28b1 | 2457 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2458 | { |
68e69fbd | 2459 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
f8ce8a0d GM |
2460 | unsigned char *add_stuff = NULL; |
2461 | int add_len = 0; | |
68e69fbd | 2462 | int idx = -1; |
9a76659d | 2463 | |
3bc25e52 KH |
2464 | if (str_multibyte) |
2465 | { | |
eb99a8dd | 2466 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 KH |
2467 | if (!buf_multibyte) |
2468 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2469 | } | |
2470 | else | |
2471 | { | |
2472 | /* Note that we don't have to increment POS. */ | |
2473 | c = XSTRING (newtext)->data[pos_byte++]; | |
2474 | if (buf_multibyte) | |
2475 | c = unibyte_char_to_multibyte (c); | |
2476 | } | |
ac3b28b1 | 2477 | |
68e69fbd RS |
2478 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2479 | or set IDX to a match index, which means put that part | |
2480 | of the buffer text into SUBSTED. */ | |
2481 | ||
ca1d1d23 JB |
2482 | if (c == '\\') |
2483 | { | |
3bc25e52 KH |
2484 | if (str_multibyte) |
2485 | { | |
eb99a8dd KH |
2486 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2487 | pos, pos_byte); | |
3bc25e52 KH |
2488 | if (!buf_multibyte && !SINGLE_BYTE_CHAR_P (c)) |
2489 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2490 | } | |
2491 | else | |
2492 | { | |
2493 | c = XSTRING (newtext)->data[pos_byte++]; | |
2494 | if (buf_multibyte) | |
2495 | c = unibyte_char_to_multibyte (c); | |
2496 | } | |
2497 | ||
ca1d1d23 | 2498 | if (c == '&') |
68e69fbd | 2499 | idx = sub; |
78445046 | 2500 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2501 | { |
2502 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2503 | idx = c - '0'; |
ca1d1d23 | 2504 | } |
636a5e28 | 2505 | else if (c == '\\') |
68e69fbd | 2506 | add_len = 1, add_stuff = "\\"; |
636a5e28 | 2507 | else |
3bc25e52 KH |
2508 | { |
2509 | xfree (substed); | |
2510 | error ("Invalid use of `\\' in replacement text"); | |
2511 | } | |
ca1d1d23 JB |
2512 | } |
2513 | else | |
68e69fbd RS |
2514 | { |
2515 | add_len = CHAR_STRING (c, str); | |
2516 | add_stuff = str; | |
2517 | } | |
2518 | ||
2519 | /* If we want to copy part of a previous match, | |
2520 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2521 | if (idx >= 0) | |
2522 | { | |
2523 | int begbyte = CHAR_TO_BYTE (search_regs.start[idx]); | |
2524 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; | |
2525 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2526 | move_gap (search_regs.start[idx]); | |
2527 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2528 | } | |
2529 | ||
2530 | /* Now the stuff we want to add to SUBSTED | |
2531 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2532 | ||
2533 | /* Make sure SUBSTED is big enough. */ | |
2534 | if (substed_len + add_len >= substed_alloc_size) | |
2535 | { | |
2536 | substed_alloc_size = substed_len + add_len + 500; | |
2537 | substed = (unsigned char *) xrealloc (substed, | |
2538 | substed_alloc_size + 1); | |
2539 | } | |
2540 | ||
2541 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2542 | if (add_stuff) |
2543 | { | |
2544 | bcopy (add_stuff, substed + substed_len, add_len); | |
2545 | substed_len += add_len; | |
2546 | } | |
ca1d1d23 | 2547 | } |
68e69fbd RS |
2548 | |
2549 | /* Now insert what we accumulated. */ | |
2550 | insert_and_inherit (substed, substed_len); | |
2551 | ||
2552 | xfree (substed); | |
ca1d1d23 JB |
2553 | } |
2554 | ||
6ec8bbd2 | 2555 | inslen = PT - (search_regs.start[sub]); |
d7a5ad5f | 2556 | del_range (search_regs.start[sub] + inslen, search_regs.end[sub] + inslen); |
ca1d1d23 JB |
2557 | |
2558 | if (case_action == all_caps) | |
6ec8bbd2 | 2559 | Fupcase_region (make_number (PT - inslen), make_number (PT)); |
ca1d1d23 | 2560 | else if (case_action == cap_initial) |
6ec8bbd2 | 2561 | Fupcase_initials_region (make_number (PT - inslen), make_number (PT)); |
b0eba991 | 2562 | |
3e18eecf RS |
2563 | newpoint = PT; |
2564 | ||
b0eba991 | 2565 | /* Put point back where it was in the text. */ |
8d808a65 | 2566 | if (opoint <= 0) |
fa8ed3e0 | 2567 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2568 | else |
fa8ed3e0 | 2569 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2570 | |
2571 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2572 | move_if_not_intangible (newpoint); |
b0eba991 | 2573 | |
ca1d1d23 JB |
2574 | return Qnil; |
2575 | } | |
2576 | \f | |
2577 | static Lisp_Object | |
2578 | match_limit (num, beginningp) | |
2579 | Lisp_Object num; | |
2580 | int beginningp; | |
2581 | { | |
2582 | register int n; | |
2583 | ||
2584 | CHECK_NUMBER (num, 0); | |
2585 | n = XINT (num); | |
4746118a JB |
2586 | if (n < 0 || n >= search_regs.num_regs) |
2587 | args_out_of_range (num, make_number (search_regs.num_regs)); | |
2588 | if (search_regs.num_regs <= 0 | |
2589 | || search_regs.start[n] < 0) | |
ca1d1d23 JB |
2590 | return Qnil; |
2591 | return (make_number ((beginningp) ? search_regs.start[n] | |
2592 | : search_regs.end[n])); | |
2593 | } | |
2594 | ||
2595 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
8c1a1077 PJ |
2596 | doc: /* Return position of start of text matched by last search. |
2597 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2598 | regexp. | |
2599 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2600 | SUBEXP pairs. | |
2601 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2602 | (subexp) | |
5806161b | 2603 | Lisp_Object subexp; |
ca1d1d23 | 2604 | { |
5806161b | 2605 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2606 | } |
2607 | ||
2608 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
8c1a1077 PJ |
2609 | doc: /* Return position of end of text matched by last search. |
2610 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2611 | regexp. | |
2612 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2613 | SUBEXP pairs. | |
2614 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2615 | (subexp) | |
5806161b | 2616 | Lisp_Object subexp; |
ca1d1d23 | 2617 | { |
5806161b | 2618 | return match_limit (subexp, 0); |
ca1d1d23 JB |
2619 | } |
2620 | ||
56256c2a | 2621 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 2, 0, |
8c1a1077 PJ |
2622 | doc: /* Return a list containing all info on what the last search matched. |
2623 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2624 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2625 | if the last match was on a buffer; integers or nil if a string was matched. | |
2626 | Use `store-match-data' to reinstate the data in this list. | |
2627 | ||
2628 | If INTEGERS (the optional first argument) is non-nil, always use integers | |
2629 | \(rather than markers) to represent buffer positions. | |
2630 | If REUSE is a list, reuse it as part of the value. If REUSE is long enough | |
2631 | to hold all the values, and if INTEGERS is non-nil, no consing is done. */) | |
2632 | (integers, reuse) | |
56256c2a | 2633 | Lisp_Object integers, reuse; |
ca1d1d23 | 2634 | { |
56256c2a | 2635 | Lisp_Object tail, prev; |
4746118a | 2636 | Lisp_Object *data; |
ca1d1d23 JB |
2637 | int i, len; |
2638 | ||
daa37602 | 2639 | if (NILP (last_thing_searched)) |
c36bcf1b | 2640 | return Qnil; |
daa37602 | 2641 | |
6bbd7a29 GM |
2642 | prev = Qnil; |
2643 | ||
4746118a JB |
2644 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs) |
2645 | * sizeof (Lisp_Object)); | |
2646 | ||
ca1d1d23 | 2647 | len = -1; |
4746118a | 2648 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2649 | { |
2650 | int start = search_regs.start[i]; | |
2651 | if (start >= 0) | |
2652 | { | |
56256c2a RS |
2653 | if (EQ (last_thing_searched, Qt) |
2654 | || ! NILP (integers)) | |
ca1d1d23 | 2655 | { |
c235cce7 KH |
2656 | XSETFASTINT (data[2 * i], start); |
2657 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2658 | } |
0ed62dc7 | 2659 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2660 | { |
2661 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2662 | Fset_marker (data[2 * i], |
2663 | make_number (start), | |
2664 | last_thing_searched); | |
ca1d1d23 JB |
2665 | data[2 * i + 1] = Fmake_marker (); |
2666 | Fset_marker (data[2 * i + 1], | |
daa37602 JB |
2667 | make_number (search_regs.end[i]), |
2668 | last_thing_searched); | |
ca1d1d23 | 2669 | } |
daa37602 JB |
2670 | else |
2671 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2672 | abort (); | |
2673 | ||
ca1d1d23 JB |
2674 | len = i; |
2675 | } | |
2676 | else | |
2677 | data[2 * i] = data [2 * i + 1] = Qnil; | |
2678 | } | |
56256c2a RS |
2679 | |
2680 | /* If REUSE is not usable, cons up the values and return them. */ | |
2681 | if (! CONSP (reuse)) | |
2682 | return Flist (2 * len + 2, data); | |
2683 | ||
2684 | /* If REUSE is a list, store as many value elements as will fit | |
2685 | into the elements of REUSE. */ | |
2686 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2687 | i++, tail = XCDR (tail)) |
56256c2a RS |
2688 | { |
2689 | if (i < 2 * len + 2) | |
f3fbd155 | 2690 | XSETCAR (tail, data[i]); |
56256c2a | 2691 | else |
f3fbd155 | 2692 | XSETCAR (tail, Qnil); |
56256c2a RS |
2693 | prev = tail; |
2694 | } | |
2695 | ||
2696 | /* If we couldn't fit all value elements into REUSE, | |
2697 | cons up the rest of them and add them to the end of REUSE. */ | |
2698 | if (i < 2 * len + 2) | |
f3fbd155 | 2699 | XSETCDR (prev, Flist (2 * len + 2 - i, data + i)); |
56256c2a RS |
2700 | |
2701 | return reuse; | |
ca1d1d23 JB |
2702 | } |
2703 | ||
2704 | ||
3f1c005b | 2705 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 1, 0, |
8c1a1077 PJ |
2706 | doc: /* Set internal data on last search match from elements of LIST. |
2707 | LIST should have been created by calling `match-data' previously. */) | |
2708 | (list) | |
ca1d1d23 JB |
2709 | register Lisp_Object list; |
2710 | { | |
2711 | register int i; | |
2712 | register Lisp_Object marker; | |
2713 | ||
7074fde6 FP |
2714 | if (running_asynch_code) |
2715 | save_search_regs (); | |
2716 | ||
ca1d1d23 | 2717 | if (!CONSP (list) && !NILP (list)) |
b37902c8 | 2718 | list = wrong_type_argument (Qconsp, list); |
ca1d1d23 | 2719 | |
daa37602 JB |
2720 | /* Unless we find a marker with a buffer in LIST, assume that this |
2721 | match data came from a string. */ | |
2722 | last_thing_searched = Qt; | |
2723 | ||
4746118a JB |
2724 | /* Allocate registers if they don't already exist. */ |
2725 | { | |
d084e942 | 2726 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2727 | |
2728 | if (length > search_regs.num_regs) | |
2729 | { | |
1113d9db JB |
2730 | if (search_regs.num_regs == 0) |
2731 | { | |
2732 | search_regs.start | |
2733 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2734 | search_regs.end | |
2735 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2736 | } | |
4746118a | 2737 | else |
1113d9db JB |
2738 | { |
2739 | search_regs.start | |
2740 | = (regoff_t *) xrealloc (search_regs.start, | |
2741 | length * sizeof (regoff_t)); | |
2742 | search_regs.end | |
2743 | = (regoff_t *) xrealloc (search_regs.end, | |
2744 | length * sizeof (regoff_t)); | |
2745 | } | |
4746118a | 2746 | |
e62371e9 KH |
2747 | for (i = search_regs.num_regs; i < length; i++) |
2748 | search_regs.start[i] = -1; | |
2749 | ||
487282dc | 2750 | search_regs.num_regs = length; |
4746118a JB |
2751 | } |
2752 | } | |
2753 | ||
2754 | for (i = 0; i < search_regs.num_regs; i++) | |
ca1d1d23 JB |
2755 | { |
2756 | marker = Fcar (list); | |
2757 | if (NILP (marker)) | |
2758 | { | |
2759 | search_regs.start[i] = -1; | |
2760 | list = Fcdr (list); | |
2761 | } | |
2762 | else | |
2763 | { | |
e62371e9 KH |
2764 | int from; |
2765 | ||
0ed62dc7 | 2766 | if (MARKERP (marker)) |
daa37602 JB |
2767 | { |
2768 | if (XMARKER (marker)->buffer == 0) | |
c235cce7 | 2769 | XSETFASTINT (marker, 0); |
daa37602 | 2770 | else |
a3668d92 | 2771 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); |
daa37602 | 2772 | } |
ca1d1d23 JB |
2773 | |
2774 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
e62371e9 | 2775 | from = XINT (marker); |
ca1d1d23 JB |
2776 | list = Fcdr (list); |
2777 | ||
2778 | marker = Fcar (list); | |
0ed62dc7 | 2779 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
c235cce7 | 2780 | XSETFASTINT (marker, 0); |
ca1d1d23 JB |
2781 | |
2782 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
e62371e9 | 2783 | search_regs.start[i] = from; |
ca1d1d23 JB |
2784 | search_regs.end[i] = XINT (marker); |
2785 | } | |
2786 | list = Fcdr (list); | |
2787 | } | |
2788 | ||
2789 | return Qnil; | |
2790 | } | |
2791 | ||
7074fde6 FP |
2792 | /* If non-zero the match data have been saved in saved_search_regs |
2793 | during the execution of a sentinel or filter. */ | |
75ebf74b | 2794 | static int search_regs_saved; |
7074fde6 FP |
2795 | static struct re_registers saved_search_regs; |
2796 | ||
2797 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
2798 | if asynchronous code (filter or sentinel) is running. */ | |
2799 | static void | |
2800 | save_search_regs () | |
2801 | { | |
2802 | if (!search_regs_saved) | |
2803 | { | |
2804 | saved_search_regs.num_regs = search_regs.num_regs; | |
2805 | saved_search_regs.start = search_regs.start; | |
2806 | saved_search_regs.end = search_regs.end; | |
2807 | search_regs.num_regs = 0; | |
2d4a771a RS |
2808 | search_regs.start = 0; |
2809 | search_regs.end = 0; | |
7074fde6 FP |
2810 | |
2811 | search_regs_saved = 1; | |
2812 | } | |
2813 | } | |
2814 | ||
2815 | /* Called upon exit from filters and sentinels. */ | |
2816 | void | |
2817 | restore_match_data () | |
2818 | { | |
2819 | if (search_regs_saved) | |
2820 | { | |
2821 | if (search_regs.num_regs > 0) | |
2822 | { | |
2823 | xfree (search_regs.start); | |
2824 | xfree (search_regs.end); | |
2825 | } | |
2826 | search_regs.num_regs = saved_search_regs.num_regs; | |
2827 | search_regs.start = saved_search_regs.start; | |
2828 | search_regs.end = saved_search_regs.end; | |
2829 | ||
2830 | search_regs_saved = 0; | |
2831 | } | |
2832 | } | |
2833 | ||
ca1d1d23 JB |
2834 | /* Quote a string to inactivate reg-expr chars */ |
2835 | ||
2836 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 PJ |
2837 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
2838 | (string) | |
5806161b | 2839 | Lisp_Object string; |
ca1d1d23 JB |
2840 | { |
2841 | register unsigned char *in, *out, *end; | |
2842 | register unsigned char *temp; | |
0c8533c6 | 2843 | int backslashes_added = 0; |
ca1d1d23 | 2844 | |
5806161b | 2845 | CHECK_STRING (string, 0); |
ca1d1d23 | 2846 | |
fc932ac6 | 2847 | temp = (unsigned char *) alloca (STRING_BYTES (XSTRING (string)) * 2); |
ca1d1d23 JB |
2848 | |
2849 | /* Now copy the data into the new string, inserting escapes. */ | |
2850 | ||
5806161b | 2851 | in = XSTRING (string)->data; |
fc932ac6 | 2852 | end = in + STRING_BYTES (XSTRING (string)); |
ca1d1d23 JB |
2853 | out = temp; |
2854 | ||
2855 | for (; in != end; in++) | |
2856 | { | |
2857 | if (*in == '[' || *in == ']' | |
2858 | || *in == '*' || *in == '.' || *in == '\\' | |
2859 | || *in == '?' || *in == '+' | |
2860 | || *in == '^' || *in == '$') | |
0c8533c6 | 2861 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
2862 | *out++ = *in; |
2863 | } | |
2864 | ||
3f8100f1 | 2865 | return make_specified_string (temp, |
0c8533c6 | 2866 | XSTRING (string)->size + backslashes_added, |
3f8100f1 RS |
2867 | out - temp, |
2868 | STRING_MULTIBYTE (string)); | |
ca1d1d23 JB |
2869 | } |
2870 | \f | |
dfcf069d | 2871 | void |
ca1d1d23 JB |
2872 | syms_of_search () |
2873 | { | |
2874 | register int i; | |
2875 | ||
487282dc KH |
2876 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
2877 | { | |
2878 | searchbufs[i].buf.allocated = 100; | |
2879 | searchbufs[i].buf.buffer = (unsigned char *) malloc (100); | |
2880 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; | |
2881 | searchbufs[i].regexp = Qnil; | |
2882 | staticpro (&searchbufs[i].regexp); | |
2883 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); | |
2884 | } | |
2885 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
2886 | |
2887 | Qsearch_failed = intern ("search-failed"); | |
2888 | staticpro (&Qsearch_failed); | |
2889 | Qinvalid_regexp = intern ("invalid-regexp"); | |
2890 | staticpro (&Qinvalid_regexp); | |
2891 | ||
2892 | Fput (Qsearch_failed, Qerror_conditions, | |
2893 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
2894 | Fput (Qsearch_failed, Qerror_message, | |
2895 | build_string ("Search failed")); | |
2896 | ||
2897 | Fput (Qinvalid_regexp, Qerror_conditions, | |
2898 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
2899 | Fput (Qinvalid_regexp, Qerror_message, | |
2900 | build_string ("Invalid regexp")); | |
2901 | ||
daa37602 JB |
2902 | last_thing_searched = Qnil; |
2903 | staticpro (&last_thing_searched); | |
2904 | ||
ca1d1d23 | 2905 | defsubr (&Slooking_at); |
b819a390 RS |
2906 | defsubr (&Sposix_looking_at); |
2907 | defsubr (&Sstring_match); | |
2908 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
2909 | defsubr (&Ssearch_forward); |
2910 | defsubr (&Ssearch_backward); | |
2911 | defsubr (&Sword_search_forward); | |
2912 | defsubr (&Sword_search_backward); | |
2913 | defsubr (&Sre_search_forward); | |
2914 | defsubr (&Sre_search_backward); | |
b819a390 RS |
2915 | defsubr (&Sposix_search_forward); |
2916 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
2917 | defsubr (&Sreplace_match); |
2918 | defsubr (&Smatch_beginning); | |
2919 | defsubr (&Smatch_end); | |
2920 | defsubr (&Smatch_data); | |
3f1c005b | 2921 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
2922 | defsubr (&Sregexp_quote); |
2923 | } |