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, |
94f94972 | 333 | "Return t if text after point matches regular expression REGEXP.\n\ |
b819a390 RS |
334 | This function modifies the match data that `match-beginning',\n\ |
335 | `match-end' and `match-data' access; save and restore the match\n\ | |
336 | data if you want to preserve them.") | |
94f94972 RS |
337 | (regexp) |
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, | |
94f94972 | 344 | "Return t if text after point matches regular expression REGEXP.\n\ |
b819a390 RS |
345 | Find the longest match, in accord with Posix regular expression rules.\n\ |
346 | This function modifies the match data that `match-beginning',\n\ | |
347 | `match-end' and `match-data' access; save and restore the match\n\ | |
348 | data if you want to preserve them.") | |
94f94972 RS |
349 | (regexp) |
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 RS |
416 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
417 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
21e01890 | 418 | Case is ignored if `case-fold-search' is non-nil in the current buffer.\n\ |
b819a390 RS |
419 | If third arg START is non-nil, start search at that index in STRING.\n\ |
420 | For index of first char beyond the match, do (match-end 0).\n\ | |
421 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
422 | matched by parenthesis constructs in the pattern.") | |
423 | (regexp, string, start) | |
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, | |
430 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ | |
431 | Find the longest match, in accord with Posix regular expression rules.\n\ | |
21e01890 | 432 | Case is ignored if `case-fold-search' is non-nil in the current buffer.\n\ |
b819a390 RS |
433 | If third arg START is non-nil, start search at that index in STRING.\n\ |
434 | For index of first char beyond the match, do (match-end 0).\n\ | |
435 | `match-end' and `match-beginning' also give indices of substrings\n\ | |
436 | matched by parenthesis constructs in the pattern.") | |
437 | (regexp, string, start) | |
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 | /* If we are not in selective display mode, |
756 | check only for newlines. */ | |
757 | int selective_display = (!NILP (current_buffer->selective_display) | |
758 | && !INTEGERP (current_buffer->selective_display)); | |
759 | ||
760 | /* The code that follows is like scan_buffer | |
761 | but checks for either newline or carriage return. */ | |
762 | ||
d5d57b92 RS |
763 | if (allow_quit) |
764 | immediate_quit++; | |
fa8ed3e0 RS |
765 | |
766 | start_byte = CHAR_TO_BYTE (start); | |
767 | ||
768 | if (count > 0) | |
769 | { | |
770 | while (start_byte < limit_byte) | |
771 | { | |
772 | ceiling = BUFFER_CEILING_OF (start_byte); | |
773 | ceiling = min (limit_byte - 1, ceiling); | |
774 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
775 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
776 | while (1) | |
777 | { | |
778 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
779 | ; | |
780 | ||
781 | if (cursor != ceiling_addr) | |
782 | { | |
783 | if (--count == 0) | |
784 | { | |
d5d57b92 | 785 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
786 | start_byte = start_byte + cursor - base + 1; |
787 | start = BYTE_TO_CHAR (start_byte); | |
788 | TEMP_SET_PT_BOTH (start, start_byte); | |
789 | return 0; | |
790 | } | |
791 | else | |
792 | if (++cursor == ceiling_addr) | |
793 | break; | |
794 | } | |
795 | else | |
796 | break; | |
797 | } | |
798 | start_byte += cursor - base; | |
799 | } | |
800 | } | |
801 | else | |
802 | { | |
fa8ed3e0 RS |
803 | while (start_byte > limit_byte) |
804 | { | |
805 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
806 | ceiling = max (limit_byte, ceiling); | |
807 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
808 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
809 | while (1) | |
810 | { | |
811 | while (--cursor != ceiling_addr && *cursor != '\n') | |
812 | ; | |
813 | ||
814 | if (cursor != ceiling_addr) | |
815 | { | |
816 | if (++count == 0) | |
817 | { | |
d5d57b92 | 818 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
819 | /* Return the position AFTER the match we found. */ |
820 | start_byte = start_byte + cursor - base + 1; | |
821 | start = BYTE_TO_CHAR (start_byte); | |
822 | TEMP_SET_PT_BOTH (start, start_byte); | |
823 | return 0; | |
824 | } | |
825 | } | |
826 | else | |
827 | break; | |
828 | } | |
829 | /* Here we add 1 to compensate for the last decrement | |
830 | of CURSOR, which took it past the valid range. */ | |
831 | start_byte += cursor - base + 1; | |
832 | } | |
833 | } | |
834 | ||
835 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 836 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
837 | |
838 | return count * direction; | |
63fa018d RS |
839 | } |
840 | ||
ca1d1d23 | 841 | int |
fa8ed3e0 | 842 | find_next_newline_no_quit (from, cnt) |
ca1d1d23 JB |
843 | register int from, cnt; |
844 | { | |
fa8ed3e0 | 845 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
9169c321 JB |
846 | } |
847 | ||
9169c321 JB |
848 | /* Like find_next_newline, but returns position before the newline, |
849 | not after, and only search up to TO. This isn't just | |
850 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 851 | |
9169c321 JB |
852 | int |
853 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 854 | int from, to, cnt; |
9169c321 JB |
855 | { |
856 | int shortage; | |
857 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
858 | ||
859 | if (shortage == 0) | |
860 | pos--; | |
861 | ||
862 | return pos; | |
ca1d1d23 JB |
863 | } |
864 | \f | |
ca1d1d23 JB |
865 | /* Subroutines of Lisp buffer search functions. */ |
866 | ||
867 | static Lisp_Object | |
b819a390 | 868 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
869 | Lisp_Object string, bound, noerror, count; |
870 | int direction; | |
871 | int RE; | |
b819a390 | 872 | int posix; |
ca1d1d23 JB |
873 | { |
874 | register int np; | |
9f43ad85 | 875 | int lim, lim_byte; |
ca1d1d23 JB |
876 | int n = direction; |
877 | ||
878 | if (!NILP (count)) | |
879 | { | |
880 | CHECK_NUMBER (count, 3); | |
881 | n *= XINT (count); | |
882 | } | |
883 | ||
884 | CHECK_STRING (string, 0); | |
885 | if (NILP (bound)) | |
9f43ad85 RS |
886 | { |
887 | if (n > 0) | |
888 | lim = ZV, lim_byte = ZV_BYTE; | |
889 | else | |
890 | lim = BEGV, lim_byte = BEGV_BYTE; | |
891 | } | |
ca1d1d23 JB |
892 | else |
893 | { | |
894 | CHECK_NUMBER_COERCE_MARKER (bound, 1); | |
895 | lim = XINT (bound); | |
6ec8bbd2 | 896 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
897 | error ("Invalid search bound (wrong side of point)"); |
898 | if (lim > ZV) | |
9f43ad85 | 899 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 900 | else if (lim < BEGV) |
9f43ad85 | 901 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
902 | else |
903 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
904 | } |
905 | ||
9f43ad85 | 906 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
ca1d1d23 | 907 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 908 | ? current_buffer->case_canon_table |
3135e9fd | 909 | : Qnil), |
ca1d1d23 | 910 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 911 | ? current_buffer->case_eqv_table |
3135e9fd | 912 | : Qnil), |
b819a390 | 913 | posix); |
ca1d1d23 JB |
914 | if (np <= 0) |
915 | { | |
916 | if (NILP (noerror)) | |
917 | return signal_failure (string); | |
918 | if (!EQ (noerror, Qt)) | |
919 | { | |
920 | if (lim < BEGV || lim > ZV) | |
921 | abort (); | |
9f43ad85 | 922 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
923 | return Qnil; |
924 | #if 0 /* This would be clean, but maybe programs depend on | |
925 | a value of nil here. */ | |
481399bf | 926 | np = lim; |
a5f217b8 | 927 | #endif |
ca1d1d23 | 928 | } |
481399bf RS |
929 | else |
930 | return Qnil; | |
ca1d1d23 JB |
931 | } |
932 | ||
933 | if (np < BEGV || np > ZV) | |
934 | abort (); | |
935 | ||
936 | SET_PT (np); | |
937 | ||
938 | return make_number (np); | |
939 | } | |
940 | \f | |
fa8ed3e0 RS |
941 | /* Return 1 if REGEXP it matches just one constant string. */ |
942 | ||
b6d6a51c KH |
943 | static int |
944 | trivial_regexp_p (regexp) | |
945 | Lisp_Object regexp; | |
946 | { | |
fc932ac6 | 947 | int len = STRING_BYTES (XSTRING (regexp)); |
b6d6a51c | 948 | unsigned char *s = XSTRING (regexp)->data; |
b6d6a51c KH |
949 | while (--len >= 0) |
950 | { | |
951 | switch (*s++) | |
952 | { | |
953 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
954 | return 0; | |
955 | case '\\': | |
956 | if (--len < 0) | |
957 | return 0; | |
958 | switch (*s++) | |
959 | { | |
960 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
961 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
09c4719e | 962 | case 'S': case '=': case '{': case '}': |
5679531d | 963 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 964 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
965 | case '6': case '7': case '8': case '9': |
966 | return 0; | |
967 | } | |
968 | } | |
969 | } | |
970 | return 1; | |
971 | } | |
972 | ||
ca325161 | 973 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 974 | starting at position POS and stopping at position LIM, |
b819a390 | 975 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
976 | a regular expression if RE is true. |
977 | ||
978 | If N is positive, searching is forward and LIM must be greater than POS. | |
979 | If N is negative, searching is backward and LIM must be less than POS. | |
980 | ||
facdc750 | 981 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 982 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
983 | (if searching backward) or the end (if searching forward). |
984 | ||
985 | POSIX is nonzero if we want full backtracking (POSIX style) | |
986 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 987 | |
aff2ce94 RS |
988 | #define TRANSLATE(out, trt, d) \ |
989 | do \ | |
990 | { \ | |
991 | if (! NILP (trt)) \ | |
992 | { \ | |
993 | Lisp_Object temp; \ | |
994 | temp = Faref (trt, make_number (d)); \ | |
995 | if (INTEGERP (temp)) \ | |
996 | out = XINT (temp); \ | |
997 | else \ | |
998 | out = d; \ | |
999 | } \ | |
1000 | else \ | |
1001 | out = d; \ | |
1002 | } \ | |
1003 | while (0) | |
facdc750 | 1004 | |
b819a390 | 1005 | static int |
9f43ad85 RS |
1006 | search_buffer (string, pos, pos_byte, lim, lim_byte, n, |
1007 | RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
1008 | Lisp_Object string; |
1009 | int pos; | |
9f43ad85 | 1010 | int pos_byte; |
ca1d1d23 | 1011 | int lim; |
9f43ad85 | 1012 | int lim_byte; |
ca1d1d23 JB |
1013 | int n; |
1014 | int RE; | |
facdc750 RS |
1015 | Lisp_Object trt; |
1016 | Lisp_Object inverse_trt; | |
b819a390 | 1017 | int posix; |
ca1d1d23 JB |
1018 | { |
1019 | int len = XSTRING (string)->size; | |
fc932ac6 | 1020 | int len_byte = STRING_BYTES (XSTRING (string)); |
facdc750 | 1021 | register int i; |
ca1d1d23 | 1022 | |
7074fde6 FP |
1023 | if (running_asynch_code) |
1024 | save_search_regs (); | |
1025 | ||
a7e4cdde | 1026 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1027 | /* Null string is found at starting position. */ |
a7e4cdde | 1028 | if (len == 0 || n == 0) |
ca325161 | 1029 | { |
0353b28f | 1030 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1031 | return pos; |
1032 | } | |
3f57a499 | 1033 | |
b6d6a51c | 1034 | if (RE && !trivial_regexp_p (string)) |
ca1d1d23 | 1035 | { |
facdc750 RS |
1036 | unsigned char *p1, *p2; |
1037 | int s1, s2; | |
487282dc KH |
1038 | struct re_pattern_buffer *bufp; |
1039 | ||
0c8533c6 RS |
1040 | bufp = compile_pattern (string, &search_regs, trt, posix, |
1041 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 | 1042 | |
ca1d1d23 JB |
1043 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1044 | because letting this function finish | |
1045 | can take too long. */ | |
1046 | QUIT; /* Do a pending quit right away, | |
1047 | to avoid paradoxical behavior */ | |
1048 | /* Get pointers and sizes of the two strings | |
1049 | that make up the visible portion of the buffer. */ | |
1050 | ||
1051 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1052 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1053 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1054 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1055 | if (s1 < 0) |
1056 | { | |
1057 | p2 = p1; | |
fa8ed3e0 | 1058 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1059 | s1 = 0; |
1060 | } | |
1061 | if (s2 < 0) | |
1062 | { | |
fa8ed3e0 | 1063 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1064 | s2 = 0; |
1065 | } | |
8bb43c28 RS |
1066 | re_match_object = Qnil; |
1067 | ||
ca1d1d23 JB |
1068 | while (n < 0) |
1069 | { | |
42db823b | 1070 | int val; |
487282dc | 1071 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1072 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1073 | &search_regs, | |
42db823b | 1074 | /* Don't allow match past current point */ |
4996330b | 1075 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1076 | if (val == -2) |
b6d6a51c KH |
1077 | { |
1078 | matcher_overflow (); | |
1079 | } | |
ca1d1d23 JB |
1080 | if (val >= 0) |
1081 | { | |
26aff150 | 1082 | pos_byte = search_regs.start[0] + BEGV_BYTE; |
4746118a | 1083 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1084 | if (search_regs.start[i] >= 0) |
1085 | { | |
fa8ed3e0 RS |
1086 | search_regs.start[i] |
1087 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1088 | search_regs.end[i] | |
1089 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1090 | } |
a3668d92 | 1091 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1092 | /* Set pos to the new position. */ |
1093 | pos = search_regs.start[0]; | |
1094 | } | |
1095 | else | |
1096 | { | |
1097 | immediate_quit = 0; | |
1098 | return (n); | |
1099 | } | |
1100 | n++; | |
1101 | } | |
1102 | while (n > 0) | |
1103 | { | |
42db823b | 1104 | int val; |
487282dc | 1105 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1106 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1107 | &search_regs, | |
1108 | lim_byte - BEGV_BYTE); | |
ca1d1d23 | 1109 | if (val == -2) |
b6d6a51c KH |
1110 | { |
1111 | matcher_overflow (); | |
1112 | } | |
ca1d1d23 JB |
1113 | if (val >= 0) |
1114 | { | |
26aff150 | 1115 | pos_byte = search_regs.end[0] + BEGV_BYTE; |
4746118a | 1116 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1117 | if (search_regs.start[i] >= 0) |
1118 | { | |
fa8ed3e0 RS |
1119 | search_regs.start[i] |
1120 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1121 | search_regs.end[i] | |
1122 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1123 | } |
a3668d92 | 1124 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1125 | pos = search_regs.end[0]; |
1126 | } | |
1127 | else | |
1128 | { | |
1129 | immediate_quit = 0; | |
1130 | return (0 - n); | |
1131 | } | |
1132 | n--; | |
1133 | } | |
1134 | immediate_quit = 0; | |
1135 | return (pos); | |
1136 | } | |
1137 | else /* non-RE case */ | |
1138 | { | |
facdc750 RS |
1139 | unsigned char *raw_pattern, *pat; |
1140 | int raw_pattern_size; | |
1141 | int raw_pattern_size_byte; | |
1142 | unsigned char *patbuf; | |
1143 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); | |
1144 | unsigned char *base_pat = XSTRING (string)->data; | |
1145 | int charset_base = -1; | |
040272ce | 1146 | int boyer_moore_ok = 1; |
facdc750 RS |
1147 | |
1148 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1149 | We must convert PATTERN to match that, or we will not really | |
1150 | find things right. */ | |
1151 | ||
1152 | if (multibyte == STRING_MULTIBYTE (string)) | |
1153 | { | |
7276d3d8 | 1154 | raw_pattern = (unsigned char *) XSTRING (string)->data; |
facdc750 | 1155 | raw_pattern_size = XSTRING (string)->size; |
fc932ac6 | 1156 | raw_pattern_size_byte = STRING_BYTES (XSTRING (string)); |
facdc750 RS |
1157 | } |
1158 | else if (multibyte) | |
1159 | { | |
1160 | raw_pattern_size = XSTRING (string)->size; | |
1161 | raw_pattern_size_byte | |
1162 | = count_size_as_multibyte (XSTRING (string)->data, | |
1163 | raw_pattern_size); | |
7276d3d8 | 1164 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
facdc750 RS |
1165 | copy_text (XSTRING (string)->data, raw_pattern, |
1166 | XSTRING (string)->size, 0, 1); | |
1167 | } | |
1168 | else | |
1169 | { | |
1170 | /* Converting multibyte to single-byte. | |
1171 | ||
1172 | ??? Perhaps this conversion should be done in a special way | |
1173 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1174 | so that only the multibyte chars which really correspond to | |
1175 | the chosen single-byte character set can possibly match. */ | |
1176 | raw_pattern_size = XSTRING (string)->size; | |
1177 | raw_pattern_size_byte = XSTRING (string)->size; | |
7276d3d8 | 1178 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
facdc750 | 1179 | copy_text (XSTRING (string)->data, raw_pattern, |
fc932ac6 | 1180 | STRING_BYTES (XSTRING (string)), 1, 0); |
facdc750 RS |
1181 | } |
1182 | ||
1183 | /* Copy and optionally translate the pattern. */ | |
1184 | len = raw_pattern_size; | |
1185 | len_byte = raw_pattern_size_byte; | |
1186 | patbuf = (unsigned char *) alloca (len_byte); | |
1187 | pat = patbuf; | |
1188 | base_pat = raw_pattern; | |
1189 | if (multibyte) | |
1190 | { | |
1191 | while (--len >= 0) | |
1192 | { | |
daaa6ed8 | 1193 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
aff2ce94 | 1194 | int c, translated, inverse; |
facdc750 RS |
1195 | int in_charlen, charlen; |
1196 | ||
1197 | /* If we got here and the RE flag is set, it's because we're | |
1198 | dealing with a regexp known to be trivial, so the backslash | |
1199 | just quotes the next character. */ | |
1200 | if (RE && *base_pat == '\\') | |
1201 | { | |
1202 | len--; | |
1203 | len_byte--; | |
1204 | base_pat++; | |
1205 | } | |
1206 | ||
1207 | c = STRING_CHAR_AND_LENGTH (base_pat, len_byte, in_charlen); | |
040272ce | 1208 | |
facdc750 | 1209 | /* Translate the character, if requested. */ |
aff2ce94 | 1210 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1211 | /* If translation changed the byte-length, go back |
1212 | to the original character. */ | |
daaa6ed8 | 1213 | charlen = CHAR_STRING (translated, str); |
facdc750 RS |
1214 | if (in_charlen != charlen) |
1215 | { | |
1216 | translated = c; | |
daaa6ed8 | 1217 | charlen = CHAR_STRING (c, str); |
facdc750 RS |
1218 | } |
1219 | ||
5ffaf437 RS |
1220 | /* If we are searching for something strange, |
1221 | an invalid multibyte code, don't use boyer-moore. */ | |
1222 | if (! ASCII_BYTE_P (translated) | |
1223 | && (charlen == 1 /* 8bit code */ | |
1224 | || charlen != in_charlen /* invalid multibyte code */ | |
1225 | )) | |
1226 | boyer_moore_ok = 0; | |
1227 | ||
aff2ce94 RS |
1228 | TRANSLATE (inverse, inverse_trt, c); |
1229 | ||
facdc750 RS |
1230 | /* Did this char actually get translated? |
1231 | Would any other char get translated into it? */ | |
aff2ce94 | 1232 | if (translated != c || inverse != c) |
facdc750 RS |
1233 | { |
1234 | /* Keep track of which character set row | |
1235 | contains the characters that need translation. */ | |
5ffaf437 | 1236 | int charset_base_code = c & ~CHAR_FIELD3_MASK; |
facdc750 RS |
1237 | if (charset_base == -1) |
1238 | charset_base = charset_base_code; | |
1239 | else if (charset_base != charset_base_code) | |
1240 | /* If two different rows appear, needing translation, | |
1241 | then we cannot use boyer_moore search. */ | |
040272ce | 1242 | boyer_moore_ok = 0; |
aff2ce94 | 1243 | } |
facdc750 RS |
1244 | |
1245 | /* Store this character into the translated pattern. */ | |
1246 | bcopy (str, pat, charlen); | |
1247 | pat += charlen; | |
1248 | base_pat += in_charlen; | |
1249 | len_byte -= in_charlen; | |
1250 | } | |
1251 | } | |
1252 | else | |
1253 | { | |
040272ce KH |
1254 | /* Unibyte buffer. */ |
1255 | charset_base = 0; | |
facdc750 RS |
1256 | while (--len >= 0) |
1257 | { | |
040272ce | 1258 | int c, translated; |
facdc750 RS |
1259 | |
1260 | /* If we got here and the RE flag is set, it's because we're | |
1261 | dealing with a regexp known to be trivial, so the backslash | |
1262 | just quotes the next character. */ | |
1263 | if (RE && *base_pat == '\\') | |
1264 | { | |
1265 | len--; | |
1266 | base_pat++; | |
1267 | } | |
1268 | c = *base_pat++; | |
aff2ce94 | 1269 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1270 | *pat++ = translated; |
1271 | } | |
1272 | } | |
1273 | ||
1274 | len_byte = pat - patbuf; | |
1275 | len = raw_pattern_size; | |
1276 | pat = base_pat = patbuf; | |
1277 | ||
040272ce | 1278 | if (boyer_moore_ok) |
facdc750 | 1279 | return boyer_moore (n, pat, len, len_byte, trt, inverse_trt, |
aff2ce94 RS |
1280 | pos, pos_byte, lim, lim_byte, |
1281 | charset_base); | |
facdc750 RS |
1282 | else |
1283 | return simple_search (n, pat, len, len_byte, trt, | |
1284 | pos, pos_byte, lim, lim_byte); | |
1285 | } | |
1286 | } | |
1287 | \f | |
1288 | /* Do a simple string search N times for the string PAT, | |
1289 | whose length is LEN/LEN_BYTE, | |
1290 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1291 | TRT is the translation table. | |
f8bd51c4 | 1292 | |
facdc750 RS |
1293 | Return the character position where the match is found. |
1294 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1295 | |
facdc750 RS |
1296 | This kind of search works regardless of what is in PAT and |
1297 | regardless of what is in TRT. It is used in cases where | |
1298 | boyer_moore cannot work. */ | |
1299 | ||
1300 | static int | |
1301 | simple_search (n, pat, len, len_byte, trt, pos, pos_byte, lim, lim_byte) | |
1302 | int n; | |
1303 | unsigned char *pat; | |
1304 | int len, len_byte; | |
1305 | Lisp_Object trt; | |
1306 | int pos, pos_byte; | |
1307 | int lim, lim_byte; | |
1308 | { | |
1309 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); | |
ab228c24 | 1310 | int forward = n > 0; |
facdc750 RS |
1311 | |
1312 | if (lim > pos && multibyte) | |
1313 | while (n > 0) | |
1314 | { | |
1315 | while (1) | |
f8bd51c4 | 1316 | { |
facdc750 RS |
1317 | /* Try matching at position POS. */ |
1318 | int this_pos = pos; | |
1319 | int this_pos_byte = pos_byte; | |
1320 | int this_len = len; | |
1321 | int this_len_byte = len_byte; | |
1322 | unsigned char *p = pat; | |
1323 | if (pos + len > lim) | |
1324 | goto stop; | |
1325 | ||
1326 | while (this_len > 0) | |
1327 | { | |
1328 | int charlen, buf_charlen; | |
ab228c24 | 1329 | int pat_ch, buf_ch; |
facdc750 | 1330 | |
ab228c24 | 1331 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1332 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1333 | ZV_BYTE - this_pos_byte, | |
1334 | buf_charlen); | |
aff2ce94 | 1335 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1336 | |
1337 | if (buf_ch != pat_ch) | |
1338 | break; | |
ab228c24 RS |
1339 | |
1340 | this_len_byte -= charlen; | |
1341 | this_len--; | |
1342 | p += charlen; | |
1343 | ||
1344 | this_pos_byte += buf_charlen; | |
1345 | this_pos++; | |
facdc750 RS |
1346 | } |
1347 | ||
1348 | if (this_len == 0) | |
1349 | { | |
1350 | pos += len; | |
1351 | pos_byte += len_byte; | |
1352 | break; | |
1353 | } | |
1354 | ||
1355 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1356 | } |
facdc750 RS |
1357 | |
1358 | n--; | |
1359 | } | |
1360 | else if (lim > pos) | |
1361 | while (n > 0) | |
1362 | { | |
1363 | while (1) | |
f8bd51c4 | 1364 | { |
facdc750 RS |
1365 | /* Try matching at position POS. */ |
1366 | int this_pos = pos; | |
1367 | int this_len = len; | |
1368 | unsigned char *p = pat; | |
1369 | ||
1370 | if (pos + len > lim) | |
1371 | goto stop; | |
1372 | ||
1373 | while (this_len > 0) | |
1374 | { | |
1375 | int pat_ch = *p++; | |
1376 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1377 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1378 | |
1379 | if (buf_ch != pat_ch) | |
1380 | break; | |
ab228c24 RS |
1381 | |
1382 | this_len--; | |
1383 | this_pos++; | |
facdc750 RS |
1384 | } |
1385 | ||
1386 | if (this_len == 0) | |
1387 | { | |
1388 | pos += len; | |
1389 | break; | |
1390 | } | |
1391 | ||
1392 | pos++; | |
f8bd51c4 | 1393 | } |
facdc750 RS |
1394 | |
1395 | n--; | |
1396 | } | |
1397 | /* Backwards search. */ | |
1398 | else if (lim < pos && multibyte) | |
1399 | while (n < 0) | |
1400 | { | |
1401 | while (1) | |
f8bd51c4 | 1402 | { |
facdc750 RS |
1403 | /* Try matching at position POS. */ |
1404 | int this_pos = pos - len; | |
1405 | int this_pos_byte = pos_byte - len_byte; | |
1406 | int this_len = len; | |
1407 | int this_len_byte = len_byte; | |
1408 | unsigned char *p = pat; | |
1409 | ||
1410 | if (pos - len < lim) | |
1411 | goto stop; | |
1412 | ||
1413 | while (this_len > 0) | |
1414 | { | |
1415 | int charlen, buf_charlen; | |
ab228c24 | 1416 | int pat_ch, buf_ch; |
facdc750 | 1417 | |
ab228c24 | 1418 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1419 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1420 | ZV_BYTE - this_pos_byte, | |
1421 | buf_charlen); | |
aff2ce94 | 1422 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1423 | |
1424 | if (buf_ch != pat_ch) | |
1425 | break; | |
ab228c24 RS |
1426 | |
1427 | this_len_byte -= charlen; | |
1428 | this_len--; | |
1429 | p += charlen; | |
1430 | this_pos_byte += buf_charlen; | |
1431 | this_pos++; | |
facdc750 RS |
1432 | } |
1433 | ||
1434 | if (this_len == 0) | |
1435 | { | |
1436 | pos -= len; | |
1437 | pos_byte -= len_byte; | |
1438 | break; | |
1439 | } | |
1440 | ||
1441 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1442 | } |
1443 | ||
facdc750 RS |
1444 | n++; |
1445 | } | |
1446 | else if (lim < pos) | |
1447 | while (n < 0) | |
1448 | { | |
1449 | while (1) | |
b6d6a51c | 1450 | { |
facdc750 RS |
1451 | /* Try matching at position POS. */ |
1452 | int this_pos = pos - len; | |
1453 | int this_len = len; | |
1454 | unsigned char *p = pat; | |
1455 | ||
1456 | if (pos - len < lim) | |
1457 | goto stop; | |
1458 | ||
1459 | while (this_len > 0) | |
1460 | { | |
1461 | int pat_ch = *p++; | |
1462 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1463 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1464 | |
1465 | if (buf_ch != pat_ch) | |
1466 | break; | |
ab228c24 RS |
1467 | this_len--; |
1468 | this_pos++; | |
facdc750 RS |
1469 | } |
1470 | ||
1471 | if (this_len == 0) | |
b6d6a51c | 1472 | { |
facdc750 RS |
1473 | pos -= len; |
1474 | break; | |
b6d6a51c | 1475 | } |
facdc750 RS |
1476 | |
1477 | pos--; | |
b6d6a51c | 1478 | } |
facdc750 RS |
1479 | |
1480 | n++; | |
b6d6a51c | 1481 | } |
facdc750 RS |
1482 | |
1483 | stop: | |
1484 | if (n == 0) | |
aff2ce94 | 1485 | { |
ab228c24 RS |
1486 | if (forward) |
1487 | set_search_regs ((multibyte ? pos_byte : pos) - len_byte, len_byte); | |
1488 | else | |
1489 | set_search_regs (multibyte ? pos_byte : pos, len_byte); | |
aff2ce94 RS |
1490 | |
1491 | return pos; | |
1492 | } | |
facdc750 RS |
1493 | else if (n > 0) |
1494 | return -n; | |
1495 | else | |
1496 | return n; | |
1497 | } | |
1498 | \f | |
1499 | /* Do Boyer-Moore search N times for the string PAT, | |
1500 | whose length is LEN/LEN_BYTE, | |
1501 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1502 | DIRECTION says which direction we search in. | |
1503 | TRT and INVERSE_TRT are translation tables. | |
1504 | ||
1505 | This kind of search works if all the characters in PAT that have | |
1506 | nontrivial translation are the same aside from the last byte. This | |
1507 | makes it possible to translate just the last byte of a character, | |
1508 | and do so after just a simple test of the context. | |
1509 | ||
1510 | If that criterion is not satisfied, do not call this function. */ | |
1511 | ||
1512 | static int | |
1513 | boyer_moore (n, base_pat, len, len_byte, trt, inverse_trt, | |
aff2ce94 | 1514 | pos, pos_byte, lim, lim_byte, charset_base) |
facdc750 RS |
1515 | int n; |
1516 | unsigned char *base_pat; | |
1517 | int len, len_byte; | |
1518 | Lisp_Object trt; | |
1519 | Lisp_Object inverse_trt; | |
1520 | int pos, pos_byte; | |
1521 | int lim, lim_byte; | |
aff2ce94 | 1522 | int charset_base; |
facdc750 RS |
1523 | { |
1524 | int direction = ((n > 0) ? 1 : -1); | |
1525 | register int dirlen; | |
a968f437 | 1526 | int infinity, limit, stride_for_teases = 0; |
facdc750 RS |
1527 | register int *BM_tab; |
1528 | int *BM_tab_base; | |
1529 | register unsigned char *cursor, *p_limit; | |
1530 | register int i, j; | |
cb6792d2 | 1531 | unsigned char *pat, *pat_end; |
facdc750 RS |
1532 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); |
1533 | ||
1534 | unsigned char simple_translate[0400]; | |
6bbd7a29 GM |
1535 | int translate_prev_byte = 0; |
1536 | int translate_anteprev_byte = 0; | |
facdc750 RS |
1537 | |
1538 | #ifdef C_ALLOCA | |
1539 | int BM_tab_space[0400]; | |
1540 | BM_tab = &BM_tab_space[0]; | |
1541 | #else | |
1542 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
1543 | #endif | |
1544 | /* The general approach is that we are going to maintain that we know */ | |
1545 | /* the first (closest to the present position, in whatever direction */ | |
1546 | /* we're searching) character that could possibly be the last */ | |
1547 | /* (furthest from present position) character of a valid match. We */ | |
1548 | /* advance the state of our knowledge by looking at that character */ | |
1549 | /* and seeing whether it indeed matches the last character of the */ | |
1550 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1551 | /* move our pointer (to putative last characters) as far as is */ | |
1552 | /* logically possible. This amount of movement, which I call a */ | |
1553 | /* stride, will be the length of the pattern if the actual character */ | |
1554 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1555 | /* from the last occurrence of that character to the end of the */ | |
1556 | /* pattern. */ | |
1557 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1558 | /* characters that match the last character. This allows use of only */ | |
1559 | /* a single test, a test for having gone past the end of the */ | |
1560 | /* permissible match region, to test for both possible matches (when */ | |
1561 | /* the stride goes past the end immediately) and failure to */ | |
1562 | /* match (where you get nudged past the end one stride at a time). */ | |
1563 | ||
1564 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1565 | /* is determined only by the last character of the putative match. */ | |
1566 | /* If that character does not match, we will stride the proper */ | |
1567 | /* distance to propose a match that superimposes it on the last */ | |
1568 | /* instance of a character that matches it (per trt), or misses */ | |
1569 | /* it entirely if there is none. */ | |
1570 | ||
1571 | dirlen = len_byte * direction; | |
1572 | infinity = dirlen - (lim_byte + pos_byte + len_byte + len_byte) * direction; | |
cb6792d2 RS |
1573 | |
1574 | /* Record position after the end of the pattern. */ | |
1575 | pat_end = base_pat + len_byte; | |
1576 | /* BASE_PAT points to a character that we start scanning from. | |
1577 | It is the first character in a forward search, | |
1578 | the last character in a backward search. */ | |
facdc750 | 1579 | if (direction < 0) |
cb6792d2 RS |
1580 | base_pat = pat_end - 1; |
1581 | ||
facdc750 RS |
1582 | BM_tab_base = BM_tab; |
1583 | BM_tab += 0400; | |
1584 | j = dirlen; /* to get it in a register */ | |
1585 | /* A character that does not appear in the pattern induces a */ | |
1586 | /* stride equal to the pattern length. */ | |
1587 | while (BM_tab_base != BM_tab) | |
1588 | { | |
1589 | *--BM_tab = j; | |
1590 | *--BM_tab = j; | |
1591 | *--BM_tab = j; | |
1592 | *--BM_tab = j; | |
1593 | } | |
1594 | ||
1595 | /* We use this for translation, instead of TRT itself. | |
1596 | We fill this in to handle the characters that actually | |
1597 | occur in the pattern. Others don't matter anyway! */ | |
1598 | bzero (simple_translate, sizeof simple_translate); | |
1599 | for (i = 0; i < 0400; i++) | |
1600 | simple_translate[i] = i; | |
1601 | ||
1602 | i = 0; | |
1603 | while (i != infinity) | |
1604 | { | |
cb6792d2 | 1605 | unsigned char *ptr = base_pat + i; |
facdc750 RS |
1606 | i += direction; |
1607 | if (i == dirlen) | |
1608 | i = infinity; | |
1609 | if (! NILP (trt)) | |
ca1d1d23 | 1610 | { |
facdc750 | 1611 | int ch; |
aff2ce94 | 1612 | int untranslated; |
facdc750 RS |
1613 | int this_translated = 1; |
1614 | ||
1615 | if (multibyte | |
cb6792d2 RS |
1616 | /* Is *PTR the last byte of a character? */ |
1617 | && (pat_end - ptr == 1 || CHAR_HEAD_P (ptr[1]))) | |
ca1d1d23 | 1618 | { |
facdc750 RS |
1619 | unsigned char *charstart = ptr; |
1620 | while (! CHAR_HEAD_P (*charstart)) | |
1621 | charstart--; | |
aff2ce94 | 1622 | untranslated = STRING_CHAR (charstart, ptr - charstart + 1); |
6397418a | 1623 | if (charset_base == (untranslated & ~CHAR_FIELD3_MASK)) |
facdc750 | 1624 | { |
ab228c24 | 1625 | TRANSLATE (ch, trt, untranslated); |
aff2ce94 RS |
1626 | if (! CHAR_HEAD_P (*ptr)) |
1627 | { | |
1628 | translate_prev_byte = ptr[-1]; | |
1629 | if (! CHAR_HEAD_P (translate_prev_byte)) | |
1630 | translate_anteprev_byte = ptr[-2]; | |
1631 | } | |
facdc750 | 1632 | } |
aff2ce94 | 1633 | else |
ab228c24 RS |
1634 | { |
1635 | this_translated = 0; | |
1636 | ch = *ptr; | |
1637 | } | |
ca1d1d23 | 1638 | } |
facdc750 | 1639 | else if (!multibyte) |
aff2ce94 | 1640 | TRANSLATE (ch, trt, *ptr); |
ca1d1d23 JB |
1641 | else |
1642 | { | |
facdc750 RS |
1643 | ch = *ptr; |
1644 | this_translated = 0; | |
ca1d1d23 | 1645 | } |
facdc750 | 1646 | |
ab228c24 RS |
1647 | if (ch > 0400) |
1648 | j = ((unsigned char) ch) | 0200; | |
1649 | else | |
1650 | j = (unsigned char) ch; | |
1651 | ||
facdc750 RS |
1652 | if (i == infinity) |
1653 | stride_for_teases = BM_tab[j]; | |
ab228c24 | 1654 | |
facdc750 RS |
1655 | BM_tab[j] = dirlen - i; |
1656 | /* A translation table is accompanied by its inverse -- see */ | |
1657 | /* comment following downcase_table for details */ | |
1658 | if (this_translated) | |
ab228c24 RS |
1659 | { |
1660 | int starting_ch = ch; | |
1661 | int starting_j = j; | |
1662 | while (1) | |
1663 | { | |
1664 | TRANSLATE (ch, inverse_trt, ch); | |
1665 | if (ch > 0400) | |
1666 | j = ((unsigned char) ch) | 0200; | |
1667 | else | |
1668 | j = (unsigned char) ch; | |
1669 | ||
1670 | /* For all the characters that map into CH, | |
1671 | set up simple_translate to map the last byte | |
1672 | into STARTING_J. */ | |
1673 | simple_translate[j] = starting_j; | |
1674 | if (ch == starting_ch) | |
1675 | break; | |
1676 | BM_tab[j] = dirlen - i; | |
1677 | } | |
1678 | } | |
facdc750 RS |
1679 | } |
1680 | else | |
1681 | { | |
1682 | j = *ptr; | |
1683 | ||
1684 | if (i == infinity) | |
1685 | stride_for_teases = BM_tab[j]; | |
1686 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1687 | } |
facdc750 RS |
1688 | /* stride_for_teases tells how much to stride if we get a */ |
1689 | /* match on the far character but are subsequently */ | |
1690 | /* disappointed, by recording what the stride would have been */ | |
1691 | /* for that character if the last character had been */ | |
1692 | /* different. */ | |
1693 | } | |
1694 | infinity = dirlen - infinity; | |
1695 | pos_byte += dirlen - ((direction > 0) ? direction : 0); | |
1696 | /* loop invariant - POS_BYTE points at where last char (first | |
1697 | char if reverse) of pattern would align in a possible match. */ | |
1698 | while (n != 0) | |
1699 | { | |
1700 | int tail_end; | |
1701 | unsigned char *tail_end_ptr; | |
1702 | ||
1703 | /* It's been reported that some (broken) compiler thinks that | |
1704 | Boolean expressions in an arithmetic context are unsigned. | |
1705 | Using an explicit ?1:0 prevents this. */ | |
1706 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1707 | < 0) | |
1708 | return (n * (0 - direction)); | |
1709 | /* First we do the part we can by pointers (maybe nothing) */ | |
1710 | QUIT; | |
1711 | pat = base_pat; | |
1712 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1713 | if (direction > 0) |
1714 | { | |
1715 | limit = BUFFER_CEILING_OF (limit); | |
1716 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1717 | can take on without hitting edge of buffer or the gap. */ | |
1718 | limit = min (limit, pos_byte + 20000); | |
1719 | limit = min (limit, lim_byte - 1); | |
1720 | } | |
1721 | else | |
1722 | { | |
1723 | limit = BUFFER_FLOOR_OF (limit); | |
1724 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1725 | can take on without hitting edge of buffer or the gap. */ | |
1726 | limit = max (limit, pos_byte - 20000); | |
1727 | limit = max (limit, lim_byte); | |
1728 | } | |
facdc750 RS |
1729 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1730 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1731 | ||
1732 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1733 | { |
facdc750 RS |
1734 | unsigned char *p2; |
1735 | ||
1736 | p_limit = BYTE_POS_ADDR (limit); | |
1737 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
1738 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ | |
1739 | while (1) /* use one cursor setting as long as i can */ | |
ca1d1d23 | 1740 | { |
facdc750 | 1741 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1742 | { |
facdc750 RS |
1743 | /* Use signed comparison if appropriate |
1744 | to make cursor+infinity sure to be > p_limit. | |
1745 | Assuming that the buffer lies in a range of addresses | |
1746 | that are all "positive" (as ints) or all "negative", | |
1747 | either kind of comparison will work as long | |
1748 | as we don't step by infinity. So pick the kind | |
1749 | that works when we do step by infinity. */ | |
1750 | if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit) | |
1751 | while ((EMACS_INT) cursor <= (EMACS_INT) p_limit) | |
1752 | cursor += BM_tab[*cursor]; | |
ca1d1d23 | 1753 | else |
facdc750 RS |
1754 | while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit) |
1755 | cursor += BM_tab[*cursor]; | |
1756 | } | |
1757 | else | |
1758 | { | |
1759 | if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit) | |
1760 | while ((EMACS_INT) cursor >= (EMACS_INT) p_limit) | |
1761 | cursor += BM_tab[*cursor]; | |
1762 | else | |
1763 | while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit) | |
1764 | cursor += BM_tab[*cursor]; | |
1765 | } | |
ca1d1d23 | 1766 | /* If you are here, cursor is beyond the end of the searched region. */ |
facdc750 RS |
1767 | /* This can happen if you match on the far character of the pattern, */ |
1768 | /* because the "stride" of that character is infinity, a number able */ | |
1769 | /* to throw you well beyond the end of the search. It can also */ | |
1770 | /* happen if you fail to match within the permitted region and would */ | |
1771 | /* otherwise try a character beyond that region */ | |
1772 | if ((cursor - p_limit) * direction <= len_byte) | |
1773 | break; /* a small overrun is genuine */ | |
1774 | cursor -= infinity; /* large overrun = hit */ | |
1775 | i = dirlen - direction; | |
1776 | if (! NILP (trt)) | |
1777 | { | |
1778 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1779 | { |
facdc750 RS |
1780 | int ch; |
1781 | cursor -= direction; | |
1782 | /* Translate only the last byte of a character. */ | |
1783 | if (! multibyte | |
1784 | || ((cursor == tail_end_ptr | |
1785 | || CHAR_HEAD_P (cursor[1])) | |
1786 | && (CHAR_HEAD_P (cursor[0]) | |
1787 | || (translate_prev_byte == cursor[-1] | |
1788 | && (CHAR_HEAD_P (translate_prev_byte) | |
1789 | || translate_anteprev_byte == cursor[-2]))))) | |
1790 | ch = simple_translate[*cursor]; | |
1791 | else | |
1792 | ch = *cursor; | |
1793 | if (pat[i] != ch) | |
1794 | break; | |
ca1d1d23 | 1795 | } |
facdc750 RS |
1796 | } |
1797 | else | |
1798 | { | |
1799 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1800 | { |
facdc750 RS |
1801 | cursor -= direction; |
1802 | if (pat[i] != *cursor) | |
1803 | break; | |
ca1d1d23 | 1804 | } |
facdc750 RS |
1805 | } |
1806 | cursor += dirlen - i - direction; /* fix cursor */ | |
1807 | if (i + direction == 0) | |
1808 | { | |
1809 | int position; | |
0c8533c6 | 1810 | |
facdc750 | 1811 | cursor -= direction; |
1113d9db | 1812 | |
facdc750 RS |
1813 | position = pos_byte + cursor - p2 + ((direction > 0) |
1814 | ? 1 - len_byte : 0); | |
1815 | set_search_regs (position, len_byte); | |
ca325161 | 1816 | |
facdc750 RS |
1817 | if ((n -= direction) != 0) |
1818 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1819 | else |
facdc750 RS |
1820 | return ((direction > 0) |
1821 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1822 | } |
facdc750 RS |
1823 | else |
1824 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1825 | } |
facdc750 RS |
1826 | pos_byte += cursor - p2; |
1827 | } | |
1828 | else | |
1829 | /* Now we'll pick up a clump that has to be done the hard */ | |
1830 | /* way because it covers a discontinuity */ | |
1831 | { | |
1832 | limit = ((direction > 0) | |
1833 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1834 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1835 | limit = ((direction > 0) | |
1836 | ? min (limit + len_byte, lim_byte - 1) | |
1837 | : max (limit - len_byte, lim_byte)); | |
1838 | /* LIMIT is now the last value POS_BYTE can have | |
1839 | and still be valid for a possible match. */ | |
1840 | while (1) | |
ca1d1d23 | 1841 | { |
facdc750 RS |
1842 | /* This loop can be coded for space rather than */ |
1843 | /* speed because it will usually run only once. */ | |
1844 | /* (the reach is at most len + 21, and typically */ | |
1845 | /* does not exceed len) */ | |
1846 | while ((limit - pos_byte) * direction >= 0) | |
1847 | pos_byte += BM_tab[FETCH_BYTE (pos_byte)]; | |
1848 | /* now run the same tests to distinguish going off the */ | |
1849 | /* end, a match or a phony match. */ | |
1850 | if ((pos_byte - limit) * direction <= len_byte) | |
1851 | break; /* ran off the end */ | |
1852 | /* Found what might be a match. | |
1853 | Set POS_BYTE back to last (first if reverse) pos. */ | |
1854 | pos_byte -= infinity; | |
1855 | i = dirlen - direction; | |
1856 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1857 | { |
facdc750 RS |
1858 | int ch; |
1859 | unsigned char *ptr; | |
1860 | pos_byte -= direction; | |
1861 | ptr = BYTE_POS_ADDR (pos_byte); | |
1862 | /* Translate only the last byte of a character. */ | |
1863 | if (! multibyte | |
1864 | || ((ptr == tail_end_ptr | |
1865 | || CHAR_HEAD_P (ptr[1])) | |
1866 | && (CHAR_HEAD_P (ptr[0]) | |
1867 | || (translate_prev_byte == ptr[-1] | |
1868 | && (CHAR_HEAD_P (translate_prev_byte) | |
1869 | || translate_anteprev_byte == ptr[-2]))))) | |
1870 | ch = simple_translate[*ptr]; | |
1871 | else | |
1872 | ch = *ptr; | |
1873 | if (pat[i] != ch) | |
1874 | break; | |
1875 | } | |
1876 | /* Above loop has moved POS_BYTE part or all the way | |
1877 | back to the first pos (last pos if reverse). | |
1878 | Set it once again at the last (first if reverse) char. */ | |
1879 | pos_byte += dirlen - i- direction; | |
1880 | if (i + direction == 0) | |
1881 | { | |
1882 | int position; | |
1883 | pos_byte -= direction; | |
1113d9db | 1884 | |
facdc750 | 1885 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
0c8533c6 | 1886 | |
facdc750 | 1887 | set_search_regs (position, len_byte); |
ca325161 | 1888 | |
facdc750 RS |
1889 | if ((n -= direction) != 0) |
1890 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 1891 | else |
facdc750 RS |
1892 | return ((direction > 0) |
1893 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1894 | } |
facdc750 RS |
1895 | else |
1896 | pos_byte += stride_for_teases; | |
1897 | } | |
1898 | } | |
1899 | /* We have done one clump. Can we continue? */ | |
1900 | if ((lim_byte - pos_byte) * direction < 0) | |
1901 | return ((0 - n) * direction); | |
ca1d1d23 | 1902 | } |
facdc750 | 1903 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 1904 | } |
ca325161 | 1905 | |
fa8ed3e0 | 1906 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
1907 | for the overall match just found in the current buffer. |
1908 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
1909 | |
1910 | static void | |
fa8ed3e0 RS |
1911 | set_search_regs (beg_byte, nbytes) |
1912 | int beg_byte, nbytes; | |
ca325161 | 1913 | { |
a7e4cdde RS |
1914 | int i; |
1915 | ||
ca325161 RS |
1916 | /* Make sure we have registers in which to store |
1917 | the match position. */ | |
1918 | if (search_regs.num_regs == 0) | |
1919 | { | |
2d4a771a RS |
1920 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
1921 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 1922 | search_regs.num_regs = 2; |
ca325161 RS |
1923 | } |
1924 | ||
a7e4cdde RS |
1925 | /* Clear out the other registers. */ |
1926 | for (i = 1; i < search_regs.num_regs; i++) | |
1927 | { | |
1928 | search_regs.start[i] = -1; | |
1929 | search_regs.end[i] = -1; | |
1930 | } | |
1931 | ||
fa8ed3e0 RS |
1932 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
1933 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 1934 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 1935 | } |
ca1d1d23 JB |
1936 | \f |
1937 | /* Given a string of words separated by word delimiters, | |
1938 | compute a regexp that matches those exact words | |
1939 | separated by arbitrary punctuation. */ | |
1940 | ||
1941 | static Lisp_Object | |
1942 | wordify (string) | |
1943 | Lisp_Object string; | |
1944 | { | |
1945 | register unsigned char *p, *o; | |
0c8533c6 | 1946 | register int i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 1947 | Lisp_Object val; |
0c8533c6 RS |
1948 | int prev_c = 0; |
1949 | int adjust; | |
ca1d1d23 JB |
1950 | |
1951 | CHECK_STRING (string, 0); | |
1952 | p = XSTRING (string)->data; | |
1953 | len = XSTRING (string)->size; | |
1954 | ||
0c8533c6 RS |
1955 | for (i = 0, i_byte = 0; i < len; ) |
1956 | { | |
1957 | int c; | |
1958 | ||
eb99a8dd | 1959 | FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte); |
0c8533c6 RS |
1960 | |
1961 | if (SYNTAX (c) != Sword) | |
1962 | { | |
1963 | punct_count++; | |
1964 | if (i > 0 && SYNTAX (prev_c) == Sword) | |
1965 | word_count++; | |
1966 | } | |
ca1d1d23 | 1967 | |
0c8533c6 RS |
1968 | prev_c = c; |
1969 | } | |
1970 | ||
1971 | if (SYNTAX (prev_c) == Sword) | |
1972 | word_count++; | |
1973 | if (!word_count) | |
b07b65aa | 1974 | return empty_string; |
0c8533c6 RS |
1975 | |
1976 | adjust = - punct_count + 5 * (word_count - 1) + 4; | |
8a2df937 RS |
1977 | if (STRING_MULTIBYTE (string)) |
1978 | val = make_uninit_multibyte_string (len + adjust, | |
1979 | STRING_BYTES (XSTRING (string)) | |
1980 | + adjust); | |
1981 | else | |
1982 | val = make_uninit_string (len + adjust); | |
ca1d1d23 JB |
1983 | |
1984 | o = XSTRING (val)->data; | |
1985 | *o++ = '\\'; | |
1986 | *o++ = 'b'; | |
1e9582d4 | 1987 | prev_c = 0; |
ca1d1d23 | 1988 | |
1e9582d4 RS |
1989 | for (i = 0, i_byte = 0; i < len; ) |
1990 | { | |
1991 | int c; | |
1992 | int i_byte_orig = i_byte; | |
1993 | ||
eb99a8dd | 1994 | FETCH_STRING_CHAR_ADVANCE (c, string, i, i_byte); |
1e9582d4 RS |
1995 | |
1996 | if (SYNTAX (c) == Sword) | |
1997 | { | |
1998 | bcopy (&XSTRING (string)->data[i_byte_orig], o, | |
1999 | i_byte - i_byte_orig); | |
2000 | o += i_byte - i_byte_orig; | |
2001 | } | |
2002 | else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count) | |
2003 | { | |
2004 | *o++ = '\\'; | |
2005 | *o++ = 'W'; | |
2006 | *o++ = '\\'; | |
2007 | *o++ = 'W'; | |
2008 | *o++ = '*'; | |
2009 | } | |
2010 | ||
2011 | prev_c = c; | |
2012 | } | |
ca1d1d23 JB |
2013 | |
2014 | *o++ = '\\'; | |
2015 | *o++ = 'b'; | |
2016 | ||
2017 | return val; | |
2018 | } | |
2019 | \f | |
2020 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
6af43974 | 2021 | "MSearch backward: ", |
ca1d1d23 JB |
2022 | "Search backward from point for STRING.\n\ |
2023 | Set point to the beginning of the occurrence found, and return point.\n\ | |
2024 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2025 | The match found must not extend before that position.\n\ | |
2026 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2027 | If not nil and not t, position at limit of search and return nil.\n\ | |
2028 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
d594a73b EZ |
2029 | \n\ |
2030 | Search case-sensitivity is determined by the value of the variable\n\ | |
2031 | `case-fold-search', which see.\n\ | |
2032 | \n\ | |
ca1d1d23 JB |
2033 | See also the functions `match-beginning', `match-end' and `replace-match'.") |
2034 | (string, bound, noerror, count) | |
2035 | Lisp_Object string, bound, noerror, count; | |
2036 | { | |
b819a390 | 2037 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2038 | } |
2039 | ||
6af43974 | 2040 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
ca1d1d23 JB |
2041 | "Search forward from point for STRING.\n\ |
2042 | Set point to the end of the occurrence found, and return point.\n\ | |
2043 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2044 | The match found must not extend after that position. nil is equivalent\n\ | |
2045 | to (point-max).\n\ | |
2046 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2047 | If not nil and not t, move to limit of search and return nil.\n\ | |
2048 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
d594a73b EZ |
2049 | \n\ |
2050 | Search case-sensitivity is determined by the value of the variable\n\ | |
2051 | `case-fold-search', which see.\n\ | |
2052 | \n\ | |
ca1d1d23 JB |
2053 | See also the functions `match-beginning', `match-end' and `replace-match'.") |
2054 | (string, bound, noerror, count) | |
2055 | Lisp_Object string, bound, noerror, count; | |
2056 | { | |
b819a390 | 2057 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2058 | } |
2059 | ||
2060 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
2061 | "sWord search backward: ", | |
2062 | "Search backward from point for STRING, ignoring differences in punctuation.\n\ | |
2063 | Set point to the beginning of the occurrence found, and return point.\n\ | |
2064 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2065 | The match found must not extend before that position.\n\ | |
2066 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2067 | If not nil and not t, move to limit of search and return nil.\n\ | |
2068 | Optional fourth argument is repeat count--search for successive occurrences.") | |
2069 | (string, bound, noerror, count) | |
2070 | Lisp_Object string, bound, noerror, count; | |
2071 | { | |
b819a390 | 2072 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2073 | } |
2074 | ||
2075 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
2076 | "sWord search: ", | |
2077 | "Search forward from point for STRING, ignoring differences in punctuation.\n\ | |
2078 | Set point to the end of the occurrence found, and return point.\n\ | |
2079 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2080 | The match found must not extend after that position.\n\ | |
2081 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2082 | If not nil and not t, move to limit of search and return nil.\n\ | |
2083 | Optional fourth argument is repeat count--search for successive occurrences.") | |
2084 | (string, bound, noerror, count) | |
2085 | Lisp_Object string, bound, noerror, count; | |
2086 | { | |
b819a390 | 2087 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
2088 | } |
2089 | ||
2090 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
2091 | "sRE search backward: ", | |
2092 | "Search backward from point for match for regular expression REGEXP.\n\ | |
2093 | Set point to the beginning of the match, and return point.\n\ | |
2094 | The match found is the one starting last in the buffer\n\ | |
19c0a730 | 2095 | and yet ending before the origin of the search.\n\ |
ca1d1d23 JB |
2096 | An optional second argument bounds the search; it is a buffer position.\n\ |
2097 | The match found must start at or after that position.\n\ | |
2098 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2099 | If not nil and not t, move to limit of search and return nil.\n\ | |
2100 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
9dddb23f | 2101 | See also the functions `match-beginning', `match-end', `match-string',\n\ |
cb6560a1 | 2102 | and `replace-match'.") |
19c0a730 KH |
2103 | (regexp, bound, noerror, count) |
2104 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 2105 | { |
b819a390 | 2106 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2107 | } |
2108 | ||
2109 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
2110 | "sRE search: ", | |
2111 | "Search forward from point for regular expression REGEXP.\n\ | |
2112 | Set point to the end of the occurrence found, and return point.\n\ | |
2113 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2114 | The match found must not extend after that position.\n\ | |
2115 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2116 | If not nil and not t, move to limit of search and return nil.\n\ | |
2117 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
9dddb23f | 2118 | See also the functions `match-beginning', `match-end', `match-string',\n\ |
cb6560a1 | 2119 | and `replace-match'.") |
19c0a730 KH |
2120 | (regexp, bound, noerror, count) |
2121 | Lisp_Object regexp, bound, noerror, count; | |
ca1d1d23 | 2122 | { |
b819a390 RS |
2123 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2124 | } | |
2125 | ||
2126 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
2127 | "sPosix search backward: ", | |
2128 | "Search backward from point for match for regular expression REGEXP.\n\ | |
2129 | Find the longest match in accord with Posix regular expression rules.\n\ | |
2130 | Set point to the beginning of the match, and return point.\n\ | |
2131 | The match found is the one starting last in the buffer\n\ | |
2132 | and yet ending before the origin of the search.\n\ | |
2133 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2134 | The match found must start at or after that position.\n\ | |
2135 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2136 | If not nil and not t, move to limit of search and return nil.\n\ | |
2137 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
9dddb23f | 2138 | See also the functions `match-beginning', `match-end', `match-string',\n\ |
cb6560a1 | 2139 | and `replace-match'.") |
b819a390 RS |
2140 | (regexp, bound, noerror, count) |
2141 | Lisp_Object regexp, bound, noerror, count; | |
2142 | { | |
2143 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2144 | } | |
2145 | ||
2146 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
2147 | "sPosix search: ", | |
2148 | "Search forward from point for regular expression REGEXP.\n\ | |
2149 | Find the longest match in accord with Posix regular expression rules.\n\ | |
2150 | Set point to the end of the occurrence found, and return point.\n\ | |
2151 | An optional second argument bounds the search; it is a buffer position.\n\ | |
2152 | The match found must not extend after that position.\n\ | |
2153 | Optional third argument, if t, means if fail just return nil (no error).\n\ | |
2154 | If not nil and not t, move to limit of search and return nil.\n\ | |
2155 | Optional fourth argument is repeat count--search for successive occurrences.\n\ | |
9dddb23f | 2156 | See also the functions `match-beginning', `match-end', `match-string',\n\ |
cb6560a1 | 2157 | and `replace-match'.") |
b819a390 RS |
2158 | (regexp, bound, noerror, count) |
2159 | Lisp_Object regexp, bound, noerror, count; | |
2160 | { | |
2161 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2162 | } |
2163 | \f | |
d7a5ad5f | 2164 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
ca1d1d23 JB |
2165 | "Replace text matched by last search with NEWTEXT.\n\ |
2166 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text.\n\ | |
5b9cf4b2 RS |
2167 | Otherwise maybe capitalize the whole text, or maybe just word initials,\n\ |
2168 | based on the replaced text.\n\ | |
2169 | If the replaced text has only capital letters\n\ | |
2170 | and has at least one multiletter word, convert NEWTEXT to all caps.\n\ | |
2171 | If the replaced text has at least one word starting with a capital letter,\n\ | |
2172 | then capitalize each word in NEWTEXT.\n\n\ | |
ca1d1d23 JB |
2173 | If third arg LITERAL is non-nil, insert NEWTEXT literally.\n\ |
2174 | Otherwise treat `\\' as special:\n\ | |
2175 | `\\&' in NEWTEXT means substitute original matched text.\n\ | |
2176 | `\\N' means substitute what matched the Nth `\\(...\\)'.\n\ | |
2177 | If Nth parens didn't match, substitute nothing.\n\ | |
2178 | `\\\\' means insert one `\\'.\n\ | |
1113d9db | 2179 | FIXEDCASE and LITERAL are optional arguments.\n\ |
080c45fd RS |
2180 | Leaves point at end of replacement text.\n\ |
2181 | \n\ | |
2182 | The optional fourth argument STRING can be a string to modify.\n\ | |
b615868c RS |
2183 | This is meaningful when the previous match was done against STRING,\n\ |
2184 | using `string-match'. When used this way, `replace-match'\n\ | |
2185 | creates and returns a new string made by copying STRING and replacing\n\ | |
2186 | the part of STRING that was matched.\n\ | |
d7a5ad5f | 2187 | \n\ |
b615868c RS |
2188 | The optional fifth argument SUBEXP specifies a subexpression;\n\ |
2189 | it says to replace just that subexpression with NEWTEXT,\n\ | |
2190 | rather than replacing the entire matched text.\n\ | |
2191 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT;\n\ | |
2192 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts\n\ | |
b0c75ec4 | 2193 | NEWTEXT in place of subexp N.\n\ |
b615868c | 2194 | This is useful only after a regular expression search or match,\n\ |
d7a5ad5f RS |
2195 | since only regular expressions have distinguished subexpressions.") |
2196 | (newtext, fixedcase, literal, string, subexp) | |
2197 | Lisp_Object newtext, fixedcase, literal, string, subexp; | |
ca1d1d23 JB |
2198 | { |
2199 | enum { nochange, all_caps, cap_initial } case_action; | |
ac3b28b1 | 2200 | register int pos, pos_byte; |
ca1d1d23 | 2201 | int some_multiletter_word; |
97832bd0 | 2202 | int some_lowercase; |
73dc8771 | 2203 | int some_uppercase; |
208767c3 | 2204 | int some_nonuppercase_initial; |
ca1d1d23 JB |
2205 | register int c, prevc; |
2206 | int inslen; | |
d7a5ad5f | 2207 | int sub; |
3e18eecf | 2208 | int opoint, newpoint; |
ca1d1d23 | 2209 | |
16fdc568 | 2210 | CHECK_STRING (newtext, 0); |
ca1d1d23 | 2211 | |
080c45fd RS |
2212 | if (! NILP (string)) |
2213 | CHECK_STRING (string, 4); | |
2214 | ||
ca1d1d23 JB |
2215 | case_action = nochange; /* We tried an initialization */ |
2216 | /* but some C compilers blew it */ | |
4746118a JB |
2217 | |
2218 | if (search_regs.num_regs <= 0) | |
2219 | error ("replace-match called before any match found"); | |
2220 | ||
d7a5ad5f RS |
2221 | if (NILP (subexp)) |
2222 | sub = 0; | |
2223 | else | |
2224 | { | |
2225 | CHECK_NUMBER (subexp, 3); | |
2226 | sub = XINT (subexp); | |
2227 | if (sub < 0 || sub >= search_regs.num_regs) | |
2228 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2229 | } | |
2230 | ||
080c45fd RS |
2231 | if (NILP (string)) |
2232 | { | |
d7a5ad5f RS |
2233 | if (search_regs.start[sub] < BEGV |
2234 | || search_regs.start[sub] > search_regs.end[sub] | |
2235 | || search_regs.end[sub] > ZV) | |
2236 | args_out_of_range (make_number (search_regs.start[sub]), | |
2237 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2238 | } |
2239 | else | |
2240 | { | |
d7a5ad5f RS |
2241 | if (search_regs.start[sub] < 0 |
2242 | || search_regs.start[sub] > search_regs.end[sub] | |
2243 | || search_regs.end[sub] > XSTRING (string)->size) | |
2244 | args_out_of_range (make_number (search_regs.start[sub]), | |
2245 | make_number (search_regs.end[sub])); | |
080c45fd | 2246 | } |
ca1d1d23 JB |
2247 | |
2248 | if (NILP (fixedcase)) | |
2249 | { | |
2250 | /* Decide how to casify by examining the matched text. */ | |
ac3b28b1 | 2251 | int last; |
ca1d1d23 | 2252 | |
ac3b28b1 KH |
2253 | pos = search_regs.start[sub]; |
2254 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2255 | |
2256 | if (NILP (string)) | |
ac3b28b1 | 2257 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2258 | else |
ac3b28b1 | 2259 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2260 | |
ca1d1d23 JB |
2261 | prevc = '\n'; |
2262 | case_action = all_caps; | |
2263 | ||
2264 | /* some_multiletter_word is set nonzero if any original word | |
2265 | is more than one letter long. */ | |
2266 | some_multiletter_word = 0; | |
97832bd0 | 2267 | some_lowercase = 0; |
208767c3 | 2268 | some_nonuppercase_initial = 0; |
73dc8771 | 2269 | some_uppercase = 0; |
ca1d1d23 | 2270 | |
ac3b28b1 | 2271 | while (pos < last) |
ca1d1d23 | 2272 | { |
080c45fd | 2273 | if (NILP (string)) |
ac3b28b1 KH |
2274 | { |
2275 | c = FETCH_CHAR (pos_byte); | |
2276 | INC_BOTH (pos, pos_byte); | |
2277 | } | |
080c45fd | 2278 | else |
ac3b28b1 | 2279 | FETCH_STRING_CHAR_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2280 | |
ca1d1d23 JB |
2281 | if (LOWERCASEP (c)) |
2282 | { | |
2283 | /* Cannot be all caps if any original char is lower case */ | |
2284 | ||
97832bd0 | 2285 | some_lowercase = 1; |
ca1d1d23 | 2286 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2287 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2288 | else |
2289 | some_multiletter_word = 1; | |
2290 | } | |
2291 | else if (!NOCASEP (c)) | |
2292 | { | |
73dc8771 | 2293 | some_uppercase = 1; |
97832bd0 | 2294 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2295 | ; |
97832bd0 | 2296 | else |
ca1d1d23 JB |
2297 | some_multiletter_word = 1; |
2298 | } | |
208767c3 RS |
2299 | else |
2300 | { | |
2301 | /* If the initial is a caseless word constituent, | |
2302 | treat that like a lowercase initial. */ | |
2303 | if (SYNTAX (prevc) != Sword) | |
2304 | some_nonuppercase_initial = 1; | |
2305 | } | |
ca1d1d23 JB |
2306 | |
2307 | prevc = c; | |
2308 | } | |
2309 | ||
97832bd0 RS |
2310 | /* Convert to all caps if the old text is all caps |
2311 | and has at least one multiletter word. */ | |
2312 | if (! some_lowercase && some_multiletter_word) | |
2313 | case_action = all_caps; | |
c4d460ce | 2314 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2315 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2316 | case_action = cap_initial; |
208767c3 | 2317 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2318 | /* Should x -> yz, operating on X, give Yz or YZ? |
2319 | We'll assume the latter. */ | |
2320 | case_action = all_caps; | |
97832bd0 RS |
2321 | else |
2322 | case_action = nochange; | |
ca1d1d23 JB |
2323 | } |
2324 | ||
080c45fd RS |
2325 | /* Do replacement in a string. */ |
2326 | if (!NILP (string)) | |
2327 | { | |
2328 | Lisp_Object before, after; | |
2329 | ||
2330 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2331 | make_number (search_regs.start[sub])); |
2332 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2333 | |
636a5e28 RS |
2334 | /* Substitute parts of the match into NEWTEXT |
2335 | if desired. */ | |
080c45fd RS |
2336 | if (NILP (literal)) |
2337 | { | |
d131e79c RS |
2338 | int lastpos = 0; |
2339 | int lastpos_byte = 0; | |
080c45fd RS |
2340 | /* We build up the substituted string in ACCUM. */ |
2341 | Lisp_Object accum; | |
2342 | Lisp_Object middle; | |
ac3b28b1 | 2343 | int length = STRING_BYTES (XSTRING (newtext)); |
080c45fd RS |
2344 | |
2345 | accum = Qnil; | |
2346 | ||
ac3b28b1 | 2347 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd RS |
2348 | { |
2349 | int substart = -1; | |
6bbd7a29 | 2350 | int subend = 0; |
1e79ec24 | 2351 | int delbackslash = 0; |
080c45fd | 2352 | |
0c8533c6 RS |
2353 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2354 | ||
080c45fd RS |
2355 | if (c == '\\') |
2356 | { | |
0c8533c6 | 2357 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
eb99a8dd | 2358 | |
080c45fd RS |
2359 | if (c == '&') |
2360 | { | |
d7a5ad5f RS |
2361 | substart = search_regs.start[sub]; |
2362 | subend = search_regs.end[sub]; | |
080c45fd RS |
2363 | } |
2364 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') | |
2365 | { | |
ad10348f | 2366 | if (search_regs.start[c - '0'] >= 0) |
080c45fd RS |
2367 | { |
2368 | substart = search_regs.start[c - '0']; | |
2369 | subend = search_regs.end[c - '0']; | |
2370 | } | |
2371 | } | |
1e79ec24 KH |
2372 | else if (c == '\\') |
2373 | delbackslash = 1; | |
636a5e28 RS |
2374 | else |
2375 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2376 | } |
2377 | if (substart >= 0) | |
2378 | { | |
d131e79c RS |
2379 | if (pos - 2 != lastpos) |
2380 | middle = substring_both (newtext, lastpos, | |
2381 | lastpos_byte, | |
2382 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2383 | else |
2384 | middle = Qnil; | |
2385 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2386 | Fsubstring (string, |
2387 | make_number (substart), | |
080c45fd RS |
2388 | make_number (subend))); |
2389 | lastpos = pos; | |
0c8533c6 | 2390 | lastpos_byte = pos_byte; |
080c45fd | 2391 | } |
1e79ec24 KH |
2392 | else if (delbackslash) |
2393 | { | |
d131e79c RS |
2394 | middle = substring_both (newtext, lastpos, |
2395 | lastpos_byte, | |
2396 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2397 | |
1e79ec24 KH |
2398 | accum = concat2 (accum, middle); |
2399 | lastpos = pos; | |
0c8533c6 | 2400 | lastpos_byte = pos_byte; |
1e79ec24 | 2401 | } |
080c45fd RS |
2402 | } |
2403 | ||
d131e79c RS |
2404 | if (pos != lastpos) |
2405 | middle = substring_both (newtext, lastpos, | |
2406 | lastpos_byte, | |
0c8533c6 | 2407 | pos, pos_byte); |
080c45fd RS |
2408 | else |
2409 | middle = Qnil; | |
2410 | ||
2411 | newtext = concat2 (accum, middle); | |
2412 | } | |
2413 | ||
636a5e28 | 2414 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2415 | if (case_action == all_caps) |
2416 | newtext = Fupcase (newtext); | |
2417 | else if (case_action == cap_initial) | |
2b2eead9 | 2418 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2419 | |
2420 | return concat3 (before, newtext, after); | |
2421 | } | |
2422 | ||
09c4719e | 2423 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2424 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2425 | opoint = PT - ZV; |
9160906f RS |
2426 | else if (PT > search_regs.start[sub]) |
2427 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2428 | else |
2429 | opoint = PT; | |
2430 | ||
fa8ed3e0 | 2431 | TEMP_SET_PT (search_regs.start[sub]); |
b0eba991 | 2432 | |
9a76659d JB |
2433 | /* We insert the replacement text before the old text, and then |
2434 | delete the original text. This means that markers at the | |
2435 | beginning or end of the original will float to the corresponding | |
2436 | position in the replacement. */ | |
ca1d1d23 | 2437 | if (!NILP (literal)) |
16fdc568 | 2438 | Finsert_and_inherit (1, &newtext); |
ca1d1d23 JB |
2439 | else |
2440 | { | |
ac3b28b1 | 2441 | int length = STRING_BYTES (XSTRING (newtext)); |
68e69fbd RS |
2442 | unsigned char *substed; |
2443 | int substed_alloc_size, substed_len; | |
3bc25e52 KH |
2444 | int buf_multibyte = !NILP (current_buffer->enable_multibyte_characters); |
2445 | int str_multibyte = STRING_MULTIBYTE (newtext); | |
2446 | Lisp_Object rev_tbl; | |
2447 | ||
2448 | rev_tbl= (!buf_multibyte && CHAR_TABLE_P (Vnonascii_translation_table) | |
2449 | ? Fchar_table_extra_slot (Vnonascii_translation_table, | |
2450 | make_number (0)) | |
2451 | : Qnil); | |
ac3b28b1 | 2452 | |
68e69fbd RS |
2453 | substed_alloc_size = length * 2 + 100; |
2454 | substed = (unsigned char *) xmalloc (substed_alloc_size + 1); | |
2455 | substed_len = 0; | |
2456 | ||
3bc25e52 KH |
2457 | /* Go thru NEWTEXT, producing the actual text to insert in |
2458 | SUBSTED while adjusting multibyteness to that of the current | |
2459 | buffer. */ | |
ca1d1d23 | 2460 | |
ac3b28b1 | 2461 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2462 | { |
68e69fbd | 2463 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
f8ce8a0d GM |
2464 | unsigned char *add_stuff = NULL; |
2465 | int add_len = 0; | |
68e69fbd | 2466 | int idx = -1; |
9a76659d | 2467 | |
3bc25e52 KH |
2468 | if (str_multibyte) |
2469 | { | |
eb99a8dd | 2470 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 KH |
2471 | if (!buf_multibyte) |
2472 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2473 | } | |
2474 | else | |
2475 | { | |
2476 | /* Note that we don't have to increment POS. */ | |
2477 | c = XSTRING (newtext)->data[pos_byte++]; | |
2478 | if (buf_multibyte) | |
2479 | c = unibyte_char_to_multibyte (c); | |
2480 | } | |
ac3b28b1 | 2481 | |
68e69fbd RS |
2482 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2483 | or set IDX to a match index, which means put that part | |
2484 | of the buffer text into SUBSTED. */ | |
2485 | ||
ca1d1d23 JB |
2486 | if (c == '\\') |
2487 | { | |
3bc25e52 KH |
2488 | if (str_multibyte) |
2489 | { | |
eb99a8dd KH |
2490 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2491 | pos, pos_byte); | |
3bc25e52 KH |
2492 | if (!buf_multibyte && !SINGLE_BYTE_CHAR_P (c)) |
2493 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2494 | } | |
2495 | else | |
2496 | { | |
2497 | c = XSTRING (newtext)->data[pos_byte++]; | |
2498 | if (buf_multibyte) | |
2499 | c = unibyte_char_to_multibyte (c); | |
2500 | } | |
2501 | ||
ca1d1d23 | 2502 | if (c == '&') |
68e69fbd | 2503 | idx = sub; |
78445046 | 2504 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2505 | { |
2506 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2507 | idx = c - '0'; |
ca1d1d23 | 2508 | } |
636a5e28 | 2509 | else if (c == '\\') |
68e69fbd | 2510 | add_len = 1, add_stuff = "\\"; |
636a5e28 | 2511 | else |
3bc25e52 KH |
2512 | { |
2513 | xfree (substed); | |
2514 | error ("Invalid use of `\\' in replacement text"); | |
2515 | } | |
ca1d1d23 JB |
2516 | } |
2517 | else | |
68e69fbd RS |
2518 | { |
2519 | add_len = CHAR_STRING (c, str); | |
2520 | add_stuff = str; | |
2521 | } | |
2522 | ||
2523 | /* If we want to copy part of a previous match, | |
2524 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2525 | if (idx >= 0) | |
2526 | { | |
2527 | int begbyte = CHAR_TO_BYTE (search_regs.start[idx]); | |
2528 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; | |
2529 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2530 | move_gap (search_regs.start[idx]); | |
2531 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2532 | } | |
2533 | ||
2534 | /* Now the stuff we want to add to SUBSTED | |
2535 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2536 | ||
2537 | /* Make sure SUBSTED is big enough. */ | |
2538 | if (substed_len + add_len >= substed_alloc_size) | |
2539 | { | |
2540 | substed_alloc_size = substed_len + add_len + 500; | |
2541 | substed = (unsigned char *) xrealloc (substed, | |
2542 | substed_alloc_size + 1); | |
2543 | } | |
2544 | ||
2545 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2546 | if (add_stuff) |
2547 | { | |
2548 | bcopy (add_stuff, substed + substed_len, add_len); | |
2549 | substed_len += add_len; | |
2550 | } | |
ca1d1d23 | 2551 | } |
68e69fbd RS |
2552 | |
2553 | /* Now insert what we accumulated. */ | |
2554 | insert_and_inherit (substed, substed_len); | |
2555 | ||
2556 | xfree (substed); | |
ca1d1d23 JB |
2557 | } |
2558 | ||
6ec8bbd2 | 2559 | inslen = PT - (search_regs.start[sub]); |
d7a5ad5f | 2560 | del_range (search_regs.start[sub] + inslen, search_regs.end[sub] + inslen); |
ca1d1d23 JB |
2561 | |
2562 | if (case_action == all_caps) | |
6ec8bbd2 | 2563 | Fupcase_region (make_number (PT - inslen), make_number (PT)); |
ca1d1d23 | 2564 | else if (case_action == cap_initial) |
6ec8bbd2 | 2565 | Fupcase_initials_region (make_number (PT - inslen), make_number (PT)); |
b0eba991 | 2566 | |
3e18eecf RS |
2567 | newpoint = PT; |
2568 | ||
b0eba991 | 2569 | /* Put point back where it was in the text. */ |
8d808a65 | 2570 | if (opoint <= 0) |
fa8ed3e0 | 2571 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2572 | else |
fa8ed3e0 | 2573 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2574 | |
2575 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2576 | move_if_not_intangible (newpoint); |
b0eba991 | 2577 | |
ca1d1d23 JB |
2578 | return Qnil; |
2579 | } | |
2580 | \f | |
2581 | static Lisp_Object | |
2582 | match_limit (num, beginningp) | |
2583 | Lisp_Object num; | |
2584 | int beginningp; | |
2585 | { | |
2586 | register int n; | |
2587 | ||
2588 | CHECK_NUMBER (num, 0); | |
2589 | n = XINT (num); | |
4746118a JB |
2590 | if (n < 0 || n >= search_regs.num_regs) |
2591 | args_out_of_range (num, make_number (search_regs.num_regs)); | |
2592 | if (search_regs.num_regs <= 0 | |
2593 | || search_regs.start[n] < 0) | |
ca1d1d23 JB |
2594 | return Qnil; |
2595 | return (make_number ((beginningp) ? search_regs.start[n] | |
2596 | : search_regs.end[n])); | |
2597 | } | |
2598 | ||
2599 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
2600 | "Return position of start of text matched by last search.\n\ | |
5806161b EN |
2601 | SUBEXP, a number, specifies which parenthesized expression in the last\n\ |
2602 | regexp.\n\ | |
2603 | Value is nil if SUBEXPth pair didn't match, or there were less than\n\ | |
2604 | SUBEXP pairs.\n\ | |
ca1d1d23 | 2605 | Zero means the entire text matched by the whole regexp or whole string.") |
5806161b EN |
2606 | (subexp) |
2607 | Lisp_Object subexp; | |
ca1d1d23 | 2608 | { |
5806161b | 2609 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2610 | } |
2611 | ||
2612 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
2613 | "Return position of end of text matched by last search.\n\ | |
5806161b EN |
2614 | SUBEXP, a number, specifies which parenthesized expression in the last\n\ |
2615 | regexp.\n\ | |
2616 | Value is nil if SUBEXPth pair didn't match, or there were less than\n\ | |
2617 | SUBEXP pairs.\n\ | |
ca1d1d23 | 2618 | Zero means the entire text matched by the whole regexp or whole string.") |
5806161b EN |
2619 | (subexp) |
2620 | Lisp_Object subexp; | |
ca1d1d23 | 2621 | { |
5806161b | 2622 | return match_limit (subexp, 0); |
ca1d1d23 JB |
2623 | } |
2624 | ||
56256c2a | 2625 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 2, 0, |
ca1d1d23 JB |
2626 | "Return a list containing all info on what the last search matched.\n\ |
2627 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'.\n\ | |
2628 | All the elements are markers or nil (nil if the Nth pair didn't match)\n\ | |
2629 | if the last match was on a buffer; integers or nil if a string was matched.\n\ | |
56256c2a RS |
2630 | Use `store-match-data' to reinstate the data in this list.\n\ |
2631 | \n\ | |
2632 | If INTEGERS (the optional first argument) is non-nil, always use integers\n\ | |
8ca821e9 | 2633 | \(rather than markers) to represent buffer positions.\n\ |
56256c2a RS |
2634 | If REUSE is a list, reuse it as part of the value. If REUSE is long enough\n\ |
2635 | to hold all the values, and if INTEGERS is non-nil, no consing is done.") | |
2636 | (integers, reuse) | |
2637 | Lisp_Object integers, reuse; | |
ca1d1d23 | 2638 | { |
56256c2a | 2639 | Lisp_Object tail, prev; |
4746118a | 2640 | Lisp_Object *data; |
ca1d1d23 JB |
2641 | int i, len; |
2642 | ||
daa37602 | 2643 | if (NILP (last_thing_searched)) |
c36bcf1b | 2644 | return Qnil; |
daa37602 | 2645 | |
6bbd7a29 GM |
2646 | prev = Qnil; |
2647 | ||
4746118a JB |
2648 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs) |
2649 | * sizeof (Lisp_Object)); | |
2650 | ||
ca1d1d23 | 2651 | len = -1; |
4746118a | 2652 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2653 | { |
2654 | int start = search_regs.start[i]; | |
2655 | if (start >= 0) | |
2656 | { | |
56256c2a RS |
2657 | if (EQ (last_thing_searched, Qt) |
2658 | || ! NILP (integers)) | |
ca1d1d23 | 2659 | { |
c235cce7 KH |
2660 | XSETFASTINT (data[2 * i], start); |
2661 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2662 | } |
0ed62dc7 | 2663 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2664 | { |
2665 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2666 | Fset_marker (data[2 * i], |
2667 | make_number (start), | |
2668 | last_thing_searched); | |
ca1d1d23 JB |
2669 | data[2 * i + 1] = Fmake_marker (); |
2670 | Fset_marker (data[2 * i + 1], | |
daa37602 JB |
2671 | make_number (search_regs.end[i]), |
2672 | last_thing_searched); | |
ca1d1d23 | 2673 | } |
daa37602 JB |
2674 | else |
2675 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2676 | abort (); | |
2677 | ||
ca1d1d23 JB |
2678 | len = i; |
2679 | } | |
2680 | else | |
2681 | data[2 * i] = data [2 * i + 1] = Qnil; | |
2682 | } | |
56256c2a RS |
2683 | |
2684 | /* If REUSE is not usable, cons up the values and return them. */ | |
2685 | if (! CONSP (reuse)) | |
2686 | return Flist (2 * len + 2, data); | |
2687 | ||
2688 | /* If REUSE is a list, store as many value elements as will fit | |
2689 | into the elements of REUSE. */ | |
2690 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2691 | i++, tail = XCDR (tail)) |
56256c2a RS |
2692 | { |
2693 | if (i < 2 * len + 2) | |
c1d497be | 2694 | XCAR (tail) = data[i]; |
56256c2a | 2695 | else |
c1d497be | 2696 | XCAR (tail) = Qnil; |
56256c2a RS |
2697 | prev = tail; |
2698 | } | |
2699 | ||
2700 | /* If we couldn't fit all value elements into REUSE, | |
2701 | cons up the rest of them and add them to the end of REUSE. */ | |
2702 | if (i < 2 * len + 2) | |
c1d497be | 2703 | XCDR (prev) = Flist (2 * len + 2 - i, data + i); |
56256c2a RS |
2704 | |
2705 | return reuse; | |
ca1d1d23 JB |
2706 | } |
2707 | ||
2708 | ||
3f1c005b | 2709 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 1, 0, |
ca1d1d23 JB |
2710 | "Set internal data on last search match from elements of LIST.\n\ |
2711 | LIST should have been created by calling `match-data' previously.") | |
2712 | (list) | |
2713 | register Lisp_Object list; | |
2714 | { | |
2715 | register int i; | |
2716 | register Lisp_Object marker; | |
2717 | ||
7074fde6 FP |
2718 | if (running_asynch_code) |
2719 | save_search_regs (); | |
2720 | ||
ca1d1d23 | 2721 | if (!CONSP (list) && !NILP (list)) |
b37902c8 | 2722 | list = wrong_type_argument (Qconsp, list); |
ca1d1d23 | 2723 | |
daa37602 JB |
2724 | /* Unless we find a marker with a buffer in LIST, assume that this |
2725 | match data came from a string. */ | |
2726 | last_thing_searched = Qt; | |
2727 | ||
4746118a JB |
2728 | /* Allocate registers if they don't already exist. */ |
2729 | { | |
d084e942 | 2730 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2731 | |
2732 | if (length > search_regs.num_regs) | |
2733 | { | |
1113d9db JB |
2734 | if (search_regs.num_regs == 0) |
2735 | { | |
2736 | search_regs.start | |
2737 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2738 | search_regs.end | |
2739 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2740 | } | |
4746118a | 2741 | else |
1113d9db JB |
2742 | { |
2743 | search_regs.start | |
2744 | = (regoff_t *) xrealloc (search_regs.start, | |
2745 | length * sizeof (regoff_t)); | |
2746 | search_regs.end | |
2747 | = (regoff_t *) xrealloc (search_regs.end, | |
2748 | length * sizeof (regoff_t)); | |
2749 | } | |
4746118a | 2750 | |
e62371e9 KH |
2751 | for (i = search_regs.num_regs; i < length; i++) |
2752 | search_regs.start[i] = -1; | |
2753 | ||
487282dc | 2754 | search_regs.num_regs = length; |
4746118a JB |
2755 | } |
2756 | } | |
2757 | ||
2758 | for (i = 0; i < search_regs.num_regs; i++) | |
ca1d1d23 JB |
2759 | { |
2760 | marker = Fcar (list); | |
2761 | if (NILP (marker)) | |
2762 | { | |
2763 | search_regs.start[i] = -1; | |
2764 | list = Fcdr (list); | |
2765 | } | |
2766 | else | |
2767 | { | |
e62371e9 KH |
2768 | int from; |
2769 | ||
0ed62dc7 | 2770 | if (MARKERP (marker)) |
daa37602 JB |
2771 | { |
2772 | if (XMARKER (marker)->buffer == 0) | |
c235cce7 | 2773 | XSETFASTINT (marker, 0); |
daa37602 | 2774 | else |
a3668d92 | 2775 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); |
daa37602 | 2776 | } |
ca1d1d23 JB |
2777 | |
2778 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
e62371e9 | 2779 | from = XINT (marker); |
ca1d1d23 JB |
2780 | list = Fcdr (list); |
2781 | ||
2782 | marker = Fcar (list); | |
0ed62dc7 | 2783 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
c235cce7 | 2784 | XSETFASTINT (marker, 0); |
ca1d1d23 JB |
2785 | |
2786 | CHECK_NUMBER_COERCE_MARKER (marker, 0); | |
e62371e9 | 2787 | search_regs.start[i] = from; |
ca1d1d23 JB |
2788 | search_regs.end[i] = XINT (marker); |
2789 | } | |
2790 | list = Fcdr (list); | |
2791 | } | |
2792 | ||
2793 | return Qnil; | |
2794 | } | |
2795 | ||
7074fde6 FP |
2796 | /* If non-zero the match data have been saved in saved_search_regs |
2797 | during the execution of a sentinel or filter. */ | |
75ebf74b | 2798 | static int search_regs_saved; |
7074fde6 FP |
2799 | static struct re_registers saved_search_regs; |
2800 | ||
2801 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
2802 | if asynchronous code (filter or sentinel) is running. */ | |
2803 | static void | |
2804 | save_search_regs () | |
2805 | { | |
2806 | if (!search_regs_saved) | |
2807 | { | |
2808 | saved_search_regs.num_regs = search_regs.num_regs; | |
2809 | saved_search_regs.start = search_regs.start; | |
2810 | saved_search_regs.end = search_regs.end; | |
2811 | search_regs.num_regs = 0; | |
2d4a771a RS |
2812 | search_regs.start = 0; |
2813 | search_regs.end = 0; | |
7074fde6 FP |
2814 | |
2815 | search_regs_saved = 1; | |
2816 | } | |
2817 | } | |
2818 | ||
2819 | /* Called upon exit from filters and sentinels. */ | |
2820 | void | |
2821 | restore_match_data () | |
2822 | { | |
2823 | if (search_regs_saved) | |
2824 | { | |
2825 | if (search_regs.num_regs > 0) | |
2826 | { | |
2827 | xfree (search_regs.start); | |
2828 | xfree (search_regs.end); | |
2829 | } | |
2830 | search_regs.num_regs = saved_search_regs.num_regs; | |
2831 | search_regs.start = saved_search_regs.start; | |
2832 | search_regs.end = saved_search_regs.end; | |
2833 | ||
2834 | search_regs_saved = 0; | |
2835 | } | |
2836 | } | |
2837 | ||
ca1d1d23 JB |
2838 | /* Quote a string to inactivate reg-expr chars */ |
2839 | ||
2840 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
2841 | "Return a regexp string which matches exactly STRING and nothing else.") | |
5806161b EN |
2842 | (string) |
2843 | Lisp_Object string; | |
ca1d1d23 JB |
2844 | { |
2845 | register unsigned char *in, *out, *end; | |
2846 | register unsigned char *temp; | |
0c8533c6 | 2847 | int backslashes_added = 0; |
ca1d1d23 | 2848 | |
5806161b | 2849 | CHECK_STRING (string, 0); |
ca1d1d23 | 2850 | |
fc932ac6 | 2851 | temp = (unsigned char *) alloca (STRING_BYTES (XSTRING (string)) * 2); |
ca1d1d23 JB |
2852 | |
2853 | /* Now copy the data into the new string, inserting escapes. */ | |
2854 | ||
5806161b | 2855 | in = XSTRING (string)->data; |
fc932ac6 | 2856 | end = in + STRING_BYTES (XSTRING (string)); |
ca1d1d23 JB |
2857 | out = temp; |
2858 | ||
2859 | for (; in != end; in++) | |
2860 | { | |
2861 | if (*in == '[' || *in == ']' | |
2862 | || *in == '*' || *in == '.' || *in == '\\' | |
2863 | || *in == '?' || *in == '+' | |
2864 | || *in == '^' || *in == '$') | |
0c8533c6 | 2865 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
2866 | *out++ = *in; |
2867 | } | |
2868 | ||
3f8100f1 | 2869 | return make_specified_string (temp, |
0c8533c6 | 2870 | XSTRING (string)->size + backslashes_added, |
3f8100f1 RS |
2871 | out - temp, |
2872 | STRING_MULTIBYTE (string)); | |
ca1d1d23 JB |
2873 | } |
2874 | \f | |
dfcf069d | 2875 | void |
ca1d1d23 JB |
2876 | syms_of_search () |
2877 | { | |
2878 | register int i; | |
2879 | ||
487282dc KH |
2880 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
2881 | { | |
2882 | searchbufs[i].buf.allocated = 100; | |
2883 | searchbufs[i].buf.buffer = (unsigned char *) malloc (100); | |
2884 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; | |
2885 | searchbufs[i].regexp = Qnil; | |
2886 | staticpro (&searchbufs[i].regexp); | |
2887 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); | |
2888 | } | |
2889 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
2890 | |
2891 | Qsearch_failed = intern ("search-failed"); | |
2892 | staticpro (&Qsearch_failed); | |
2893 | Qinvalid_regexp = intern ("invalid-regexp"); | |
2894 | staticpro (&Qinvalid_regexp); | |
2895 | ||
2896 | Fput (Qsearch_failed, Qerror_conditions, | |
2897 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
2898 | Fput (Qsearch_failed, Qerror_message, | |
2899 | build_string ("Search failed")); | |
2900 | ||
2901 | Fput (Qinvalid_regexp, Qerror_conditions, | |
2902 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
2903 | Fput (Qinvalid_regexp, Qerror_message, | |
2904 | build_string ("Invalid regexp")); | |
2905 | ||
daa37602 JB |
2906 | last_thing_searched = Qnil; |
2907 | staticpro (&last_thing_searched); | |
2908 | ||
ca1d1d23 | 2909 | defsubr (&Slooking_at); |
b819a390 RS |
2910 | defsubr (&Sposix_looking_at); |
2911 | defsubr (&Sstring_match); | |
2912 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
2913 | defsubr (&Ssearch_forward); |
2914 | defsubr (&Ssearch_backward); | |
2915 | defsubr (&Sword_search_forward); | |
2916 | defsubr (&Sword_search_backward); | |
2917 | defsubr (&Sre_search_forward); | |
2918 | defsubr (&Sre_search_backward); | |
b819a390 RS |
2919 | defsubr (&Sposix_search_forward); |
2920 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
2921 | defsubr (&Sreplace_match); |
2922 | defsubr (&Smatch_beginning); | |
2923 | defsubr (&Smatch_end); | |
2924 | defsubr (&Smatch_data); | |
3f1c005b | 2925 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
2926 | defsubr (&Sregexp_quote); |
2927 | } |