Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
bd2cbd56 SM |
2 | Copyright (C) 1985, 86,87,93,94,97,98, 1999, 2004 |
3 | Free Software Foundation, Inc. | |
ca1d1d23 JB |
4 | |
5 | This file is part of GNU Emacs. | |
6 | ||
7 | GNU Emacs is free software; you can redistribute it and/or modify | |
8 | it under the terms of the GNU General Public License as published by | |
7c938215 | 9 | the Free Software Foundation; either version 2, or (at your option) |
ca1d1d23 JB |
10 | any later version. |
11 | ||
12 | GNU Emacs is distributed in the hope that it will be useful, | |
13 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | GNU General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
18 | along with GNU Emacs; see the file COPYING. If not, write to | |
3b7ad313 EN |
19 | the Free Software Foundation, Inc., 59 Temple Place - Suite 330, |
20 | Boston, MA 02111-1307, USA. */ | |
ca1d1d23 JB |
21 | |
22 | ||
18160b98 | 23 | #include <config.h> |
ca1d1d23 JB |
24 | #include "lisp.h" |
25 | #include "syntax.h" | |
5679531d | 26 | #include "category.h" |
ca1d1d23 | 27 | #include "buffer.h" |
76eb0881 | 28 | #include "character.h" |
9169c321 | 29 | #include "region-cache.h" |
ca1d1d23 | 30 | #include "commands.h" |
9ac0d9e0 | 31 | #include "blockinput.h" |
bf1760bb | 32 | #include "intervals.h" |
4746118a | 33 | |
ca1d1d23 JB |
34 | #include <sys/types.h> |
35 | #include "regex.h" | |
36 | ||
1d288aef | 37 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 38 | |
487282dc KH |
39 | /* If the regexp is non-nil, then the buffer contains the compiled form |
40 | of that regexp, suitable for searching. */ | |
1d288aef RS |
41 | struct regexp_cache |
42 | { | |
487282dc KH |
43 | struct regexp_cache *next; |
44 | Lisp_Object regexp; | |
45 | struct re_pattern_buffer buf; | |
46 | char fastmap[0400]; | |
b819a390 RS |
47 | /* Nonzero means regexp was compiled to do full POSIX backtracking. */ |
48 | char posix; | |
487282dc | 49 | }; |
ca1d1d23 | 50 | |
487282dc KH |
51 | /* The instances of that struct. */ |
52 | struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; | |
ca1d1d23 | 53 | |
487282dc KH |
54 | /* The head of the linked list; points to the most recently used buffer. */ |
55 | struct regexp_cache *searchbuf_head; | |
ca1d1d23 | 56 | |
ca1d1d23 | 57 | |
4746118a JB |
58 | /* Every call to re_match, etc., must pass &search_regs as the regs |
59 | argument unless you can show it is unnecessary (i.e., if re_match | |
60 | is certainly going to be called again before region-around-match | |
61 | can be called). | |
62 | ||
63 | Since the registers are now dynamically allocated, we need to make | |
64 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
65 | been allocated by checking search_regs.num_regs. |
66 | ||
67 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
68 | buffer using bits in the re_pattern_buffer. This means that whenever |
69 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
70 | allocated any registers, and will allocate new registers the next |
71 | time you call a searching or matching function. Therefore, we need | |
72 | to call re_set_registers after compiling a new pattern or after | |
73 | setting the match registers, so that the regex functions will be | |
74 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
75 | static struct re_registers search_regs; |
76 | ||
daa37602 JB |
77 | /* The buffer in which the last search was performed, or |
78 | Qt if the last search was done in a string; | |
79 | Qnil if no searching has been done yet. */ | |
80 | static Lisp_Object last_thing_searched; | |
ca1d1d23 | 81 | |
8e6208c5 | 82 | /* error condition signaled when regexp compile_pattern fails */ |
ca1d1d23 JB |
83 | |
84 | Lisp_Object Qinvalid_regexp; | |
85 | ||
ca325161 | 86 | static void set_search_regs (); |
044f81f1 | 87 | static void save_search_regs (); |
facdc750 RS |
88 | static int simple_search (); |
89 | static int boyer_moore (); | |
b819a390 RS |
90 | static int search_buffer (); |
91 | ||
ca1d1d23 JB |
92 | static void |
93 | matcher_overflow () | |
94 | { | |
95 | error ("Stack overflow in regexp matcher"); | |
96 | } | |
97 | ||
b819a390 RS |
98 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
99 | PATTERN is the pattern to compile. | |
100 | CP is the place to put the result. | |
facdc750 | 101 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
102 | REGP is the structure that says where to store the "register" |
103 | values that will result from matching this pattern. | |
104 | If it is 0, we should compile the pattern not to record any | |
105 | subexpression bounds. | |
106 | POSIX is nonzero if we want full backtracking (POSIX style) | |
5679531d | 107 | for this pattern. 0 means backtrack only enough to get a valid match. |
8f924df7 KH |
108 | MULTIBYTE is nonzero iff a target of match is a multibyte buffer or |
109 | string. */ | |
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 | { |
d451e4db | 120 | char *val; |
b819a390 | 121 | reg_syntax_t old; |
ca1d1d23 | 122 | |
487282dc | 123 | cp->regexp = Qnil; |
59fab369 | 124 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 125 | cp->posix = posix; |
93daa011 KH |
126 | cp->buf.multibyte = STRING_MULTIBYTE (pattern); |
127 | cp->buf.target_multibyte = multibyte; | |
9ac0d9e0 | 128 | BLOCK_INPUT; |
fb4a568d | 129 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 130 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
8f924df7 KH |
131 | val = (char *) re_compile_pattern ((char *) SDATA (pattern), |
132 | SBYTES (pattern), &cp->buf); | |
b819a390 | 133 | re_set_syntax (old); |
9ac0d9e0 | 134 | UNBLOCK_INPUT; |
ca1d1d23 | 135 | if (val) |
487282dc | 136 | Fsignal (Qinvalid_regexp, Fcons (build_string (val), Qnil)); |
1113d9db | 137 | |
487282dc | 138 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
139 | } |
140 | ||
6efc7887 RS |
141 | /* Shrink each compiled regexp buffer in the cache |
142 | to the size actually used right now. | |
143 | This is called from garbage collection. */ | |
144 | ||
145 | void | |
146 | shrink_regexp_cache () | |
147 | { | |
a968f437 | 148 | struct regexp_cache *cp; |
6efc7887 RS |
149 | |
150 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
151 | { | |
152 | cp->buf.allocated = cp->buf.used; | |
153 | cp->buf.buffer | |
b23c0a83 | 154 | = (unsigned char *) xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
155 | } |
156 | } | |
157 | ||
487282dc | 158 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
159 | the cache. |
160 | PATTERN is the pattern to compile. | |
facdc750 | 161 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
162 | REGP is the structure that says where to store the "register" |
163 | values that will result from matching this pattern. | |
164 | If it is 0, we should compile the pattern not to record any | |
165 | subexpression bounds. | |
166 | POSIX is nonzero if we want full backtracking (POSIX style) | |
167 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
168 | |
169 | struct re_pattern_buffer * | |
0c8533c6 | 170 | compile_pattern (pattern, regp, translate, posix, multibyte) |
487282dc KH |
171 | Lisp_Object pattern; |
172 | struct re_registers *regp; | |
facdc750 | 173 | Lisp_Object translate; |
0c8533c6 | 174 | int posix, multibyte; |
487282dc KH |
175 | { |
176 | struct regexp_cache *cp, **cpp; | |
177 | ||
178 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
179 | { | |
180 | cp = *cpp; | |
f1b9c7c1 KR |
181 | /* Entries are initialized to nil, and may be set to nil by |
182 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
49a5f770 KR |
183 | string accessors in those cases. However, compile_pattern_1 |
184 | is only applied to the cache entry we pick here to reuse. So | |
185 | nil should never appear before a non-nil entry. */ | |
7c752c80 | 186 | if (NILP (cp->regexp)) |
f1b9c7c1 | 187 | goto compile_it; |
d5db4077 | 188 | if (SCHARS (cp->regexp) == SCHARS (pattern) |
cf69b13e | 189 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 190 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 191 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d | 192 | && cp->posix == posix |
09d1b24e | 193 | && cp->buf.target_multibyte == multibyte) |
487282dc KH |
194 | break; |
195 | ||
f1b9c7c1 KR |
196 | /* If we're at the end of the cache, compile into the nil cell |
197 | we found, or the last (least recently used) cell with a | |
198 | string value. */ | |
487282dc KH |
199 | if (cp->next == 0) |
200 | { | |
f1b9c7c1 | 201 | compile_it: |
5679531d | 202 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte); |
487282dc KH |
203 | break; |
204 | } | |
205 | } | |
206 | ||
207 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
208 | either because we found it in the cache or because we just compiled it. | |
209 | Move it to the front of the queue to mark it as most recently used. */ | |
210 | *cpp = cp->next; | |
211 | cp->next = searchbuf_head; | |
212 | searchbuf_head = cp; | |
1113d9db | 213 | |
6639708c RS |
214 | /* Advise the searching functions about the space we have allocated |
215 | for register data. */ | |
216 | if (regp) | |
217 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
218 | ||
487282dc | 219 | return &cp->buf; |
ca1d1d23 JB |
220 | } |
221 | ||
222 | /* Error condition used for failing searches */ | |
223 | Lisp_Object Qsearch_failed; | |
224 | ||
225 | Lisp_Object | |
226 | signal_failure (arg) | |
227 | Lisp_Object arg; | |
228 | { | |
229 | Fsignal (Qsearch_failed, Fcons (arg, Qnil)); | |
230 | return Qnil; | |
231 | } | |
232 | \f | |
b819a390 RS |
233 | static Lisp_Object |
234 | looking_at_1 (string, posix) | |
ca1d1d23 | 235 | Lisp_Object string; |
b819a390 | 236 | int posix; |
ca1d1d23 JB |
237 | { |
238 | Lisp_Object val; | |
239 | unsigned char *p1, *p2; | |
240 | int s1, s2; | |
241 | register int i; | |
487282dc | 242 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 243 | |
7074fde6 FP |
244 | if (running_asynch_code) |
245 | save_search_regs (); | |
246 | ||
b7826503 | 247 | CHECK_STRING (string); |
487282dc KH |
248 | bufp = compile_pattern (string, &search_regs, |
249 | (!NILP (current_buffer->case_fold_search) | |
facdc750 | 250 | ? DOWNCASE_TABLE : Qnil), |
0c8533c6 RS |
251 | posix, |
252 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 JB |
253 | |
254 | immediate_quit = 1; | |
255 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
256 | ||
257 | /* Get pointers and sizes of the two strings | |
258 | that make up the visible portion of the buffer. */ | |
259 | ||
260 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 261 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 262 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 263 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
264 | if (s1 < 0) |
265 | { | |
266 | p2 = p1; | |
fa8ed3e0 | 267 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
268 | s1 = 0; |
269 | } | |
270 | if (s2 < 0) | |
271 | { | |
fa8ed3e0 | 272 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
273 | s2 = 0; |
274 | } | |
8bb43c28 RS |
275 | |
276 | re_match_object = Qnil; | |
177c0ea7 | 277 | |
487282dc | 278 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
fa8ed3e0 RS |
279 | PT_BYTE - BEGV_BYTE, &search_regs, |
280 | ZV_BYTE - BEGV_BYTE); | |
de182d70 | 281 | immediate_quit = 0; |
177c0ea7 | 282 | |
ca1d1d23 JB |
283 | if (i == -2) |
284 | matcher_overflow (); | |
285 | ||
286 | val = (0 <= i ? Qt : Qnil); | |
fa8ed3e0 RS |
287 | if (i >= 0) |
288 | for (i = 0; i < search_regs.num_regs; i++) | |
289 | if (search_regs.start[i] >= 0) | |
290 | { | |
291 | search_regs.start[i] | |
292 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
293 | search_regs.end[i] | |
294 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
295 | } | |
a3668d92 | 296 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
297 | return val; |
298 | } | |
299 | ||
b819a390 | 300 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
301 | doc: /* Return t if text after point matches regular expression REGEXP. |
302 | This function modifies the match data that `match-beginning', | |
303 | `match-end' and `match-data' access; save and restore the match | |
304 | data if you want to preserve them. */) | |
305 | (regexp) | |
94f94972 | 306 | Lisp_Object regexp; |
b819a390 | 307 | { |
94f94972 | 308 | return looking_at_1 (regexp, 0); |
b819a390 RS |
309 | } |
310 | ||
311 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
312 | doc: /* Return t if text after point matches regular expression REGEXP. |
313 | Find the longest match, in accord with Posix regular expression rules. | |
314 | This function modifies the match data that `match-beginning', | |
315 | `match-end' and `match-data' access; save and restore the match | |
316 | data if you want to preserve them. */) | |
317 | (regexp) | |
94f94972 | 318 | Lisp_Object regexp; |
b819a390 | 319 | { |
94f94972 | 320 | return looking_at_1 (regexp, 1); |
b819a390 RS |
321 | } |
322 | \f | |
323 | static Lisp_Object | |
324 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 325 | Lisp_Object regexp, string, start; |
b819a390 | 326 | int posix; |
ca1d1d23 JB |
327 | { |
328 | int val; | |
487282dc | 329 | struct re_pattern_buffer *bufp; |
0c8533c6 RS |
330 | int pos, pos_byte; |
331 | int i; | |
ca1d1d23 | 332 | |
7074fde6 FP |
333 | if (running_asynch_code) |
334 | save_search_regs (); | |
335 | ||
b7826503 PJ |
336 | CHECK_STRING (regexp); |
337 | CHECK_STRING (string); | |
ca1d1d23 JB |
338 | |
339 | if (NILP (start)) | |
0c8533c6 | 340 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
341 | else |
342 | { | |
d5db4077 | 343 | int len = SCHARS (string); |
ca1d1d23 | 344 | |
b7826503 | 345 | CHECK_NUMBER (start); |
0c8533c6 RS |
346 | pos = XINT (start); |
347 | if (pos < 0 && -pos <= len) | |
348 | pos = len + pos; | |
349 | else if (0 > pos || pos > len) | |
ca1d1d23 | 350 | args_out_of_range (string, start); |
0c8533c6 | 351 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
352 | } |
353 | ||
487282dc KH |
354 | bufp = compile_pattern (regexp, &search_regs, |
355 | (!NILP (current_buffer->case_fold_search) | |
facdc750 | 356 | ? DOWNCASE_TABLE : Qnil), |
0c8533c6 RS |
357 | posix, |
358 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 359 | immediate_quit = 1; |
8bb43c28 | 360 | re_match_object = string; |
177c0ea7 | 361 | |
d5db4077 KR |
362 | val = re_search (bufp, (char *) SDATA (string), |
363 | SBYTES (string), pos_byte, | |
364 | SBYTES (string) - pos_byte, | |
ca1d1d23 JB |
365 | &search_regs); |
366 | immediate_quit = 0; | |
daa37602 | 367 | last_thing_searched = Qt; |
ca1d1d23 JB |
368 | if (val == -2) |
369 | matcher_overflow (); | |
370 | if (val < 0) return Qnil; | |
0c8533c6 RS |
371 | |
372 | for (i = 0; i < search_regs.num_regs; i++) | |
373 | if (search_regs.start[i] >= 0) | |
374 | { | |
375 | search_regs.start[i] | |
376 | = string_byte_to_char (string, search_regs.start[i]); | |
377 | search_regs.end[i] | |
378 | = string_byte_to_char (string, search_regs.end[i]); | |
379 | } | |
380 | ||
381 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 382 | } |
e59a8453 | 383 | |
b819a390 | 384 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 PJ |
385 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
386 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
387 | If third arg START is non-nil, start search at that index in STRING. | |
388 | For index of first char beyond the match, do (match-end 0). | |
389 | `match-end' and `match-beginning' also give indices of substrings | |
2bd2f32d RS |
390 | matched by parenthesis constructs in the pattern. |
391 | ||
392 | You can use the function `match-string' to extract the substrings | |
393 | matched by the parenthesis constructions in REGEXP. */) | |
8c1a1077 | 394 | (regexp, string, start) |
b819a390 RS |
395 | Lisp_Object regexp, string, start; |
396 | { | |
397 | return string_match_1 (regexp, string, start, 0); | |
398 | } | |
399 | ||
400 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
401 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
402 | Find the longest match, in accord with Posix regular expression rules. | |
403 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
404 | If third arg START is non-nil, start search at that index in STRING. | |
405 | For index of first char beyond the match, do (match-end 0). | |
406 | `match-end' and `match-beginning' also give indices of substrings | |
407 | matched by parenthesis constructs in the pattern. */) | |
408 | (regexp, string, start) | |
b819a390 RS |
409 | Lisp_Object regexp, string, start; |
410 | { | |
411 | return string_match_1 (regexp, string, start, 1); | |
412 | } | |
413 | ||
e59a8453 RS |
414 | /* Match REGEXP against STRING, searching all of STRING, |
415 | and return the index of the match, or negative on failure. | |
416 | This does not clobber the match data. */ | |
417 | ||
418 | int | |
419 | fast_string_match (regexp, string) | |
420 | Lisp_Object regexp, string; | |
421 | { | |
422 | int val; | |
487282dc | 423 | struct re_pattern_buffer *bufp; |
e59a8453 | 424 | |
facdc750 RS |
425 | bufp = compile_pattern (regexp, 0, Qnil, |
426 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 427 | immediate_quit = 1; |
8bb43c28 | 428 | re_match_object = string; |
177c0ea7 | 429 | |
d5db4077 KR |
430 | val = re_search (bufp, (char *) SDATA (string), |
431 | SBYTES (string), 0, | |
432 | SBYTES (string), 0); | |
e59a8453 RS |
433 | immediate_quit = 0; |
434 | return val; | |
435 | } | |
5679531d KH |
436 | |
437 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
438 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
439 | This does not clobber the match data. |
440 | We assume that STRING contains single-byte characters. */ | |
5679531d KH |
441 | |
442 | extern Lisp_Object Vascii_downcase_table; | |
443 | ||
444 | int | |
b4577c63 | 445 | fast_c_string_match_ignore_case (regexp, string) |
5679531d | 446 | Lisp_Object regexp; |
96b80561 | 447 | const char *string; |
5679531d KH |
448 | { |
449 | int val; | |
450 | struct re_pattern_buffer *bufp; | |
451 | int len = strlen (string); | |
452 | ||
0c8533c6 | 453 | regexp = string_make_unibyte (regexp); |
b4577c63 | 454 | re_match_object = Qt; |
5679531d | 455 | bufp = compile_pattern (regexp, 0, |
facdc750 | 456 | Vascii_downcase_table, 0, |
f8bd51c4 | 457 | 0); |
5679531d KH |
458 | immediate_quit = 1; |
459 | val = re_search (bufp, string, len, 0, len, 0); | |
460 | immediate_quit = 0; | |
461 | return val; | |
462 | } | |
ca1d1d23 | 463 | \f |
9169c321 JB |
464 | /* The newline cache: remembering which sections of text have no newlines. */ |
465 | ||
466 | /* If the user has requested newline caching, make sure it's on. | |
467 | Otherwise, make sure it's off. | |
468 | This is our cheezy way of associating an action with the change of | |
469 | state of a buffer-local variable. */ | |
470 | static void | |
471 | newline_cache_on_off (buf) | |
472 | struct buffer *buf; | |
473 | { | |
474 | if (NILP (buf->cache_long_line_scans)) | |
475 | { | |
476 | /* It should be off. */ | |
477 | if (buf->newline_cache) | |
478 | { | |
479 | free_region_cache (buf->newline_cache); | |
480 | buf->newline_cache = 0; | |
481 | } | |
482 | } | |
483 | else | |
484 | { | |
485 | /* It should be on. */ | |
486 | if (buf->newline_cache == 0) | |
487 | buf->newline_cache = new_region_cache (); | |
488 | } | |
489 | } | |
490 | ||
491 | \f | |
492 | /* Search for COUNT instances of the character TARGET between START and END. | |
493 | ||
494 | If COUNT is positive, search forwards; END must be >= START. | |
495 | If COUNT is negative, search backwards for the -COUNTth instance; | |
496 | END must be <= START. | |
497 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
498 | ||
499 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
500 | direction indicated by COUNT. | |
ffd56f97 JB |
501 | |
502 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
5bfe95c9 RS |
503 | position after the COUNTth match. Note that for reverse motion |
504 | this is not the same as the usual convention for Emacs motion commands. | |
ffd56f97 | 505 | |
9169c321 JB |
506 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
507 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 508 | |
087a5f81 RS |
509 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
510 | except when inside redisplay. */ | |
511 | ||
dfcf069d | 512 | int |
9169c321 JB |
513 | scan_buffer (target, start, end, count, shortage, allow_quit) |
514 | register int target; | |
515 | int start, end; | |
516 | int count; | |
517 | int *shortage; | |
087a5f81 | 518 | int allow_quit; |
ca1d1d23 | 519 | { |
9169c321 | 520 | struct region_cache *newline_cache; |
177c0ea7 | 521 | int direction; |
ffd56f97 | 522 | |
9169c321 JB |
523 | if (count > 0) |
524 | { | |
525 | direction = 1; | |
526 | if (! end) end = ZV; | |
527 | } | |
528 | else | |
529 | { | |
530 | direction = -1; | |
531 | if (! end) end = BEGV; | |
532 | } | |
ffd56f97 | 533 | |
9169c321 JB |
534 | newline_cache_on_off (current_buffer); |
535 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
536 | |
537 | if (shortage != 0) | |
538 | *shortage = 0; | |
539 | ||
087a5f81 | 540 | immediate_quit = allow_quit; |
ca1d1d23 | 541 | |
ffd56f97 | 542 | if (count > 0) |
9169c321 | 543 | while (start != end) |
ca1d1d23 | 544 | { |
9169c321 JB |
545 | /* Our innermost scanning loop is very simple; it doesn't know |
546 | about gaps, buffer ends, or the newline cache. ceiling is | |
547 | the position of the last character before the next such | |
548 | obstacle --- the last character the dumb search loop should | |
549 | examine. */ | |
fa8ed3e0 RS |
550 | int ceiling_byte = CHAR_TO_BYTE (end) - 1; |
551 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 552 | int tem; |
9169c321 JB |
553 | |
554 | /* If we're looking for a newline, consult the newline cache | |
555 | to see where we can avoid some scanning. */ | |
556 | if (target == '\n' && newline_cache) | |
557 | { | |
558 | int next_change; | |
559 | immediate_quit = 0; | |
560 | while (region_cache_forward | |
fa8ed3e0 RS |
561 | (current_buffer, newline_cache, start_byte, &next_change)) |
562 | start_byte = next_change; | |
cbe0db0d | 563 | immediate_quit = allow_quit; |
9169c321 | 564 | |
fa8ed3e0 RS |
565 | /* START should never be after END. */ |
566 | if (start_byte > ceiling_byte) | |
567 | start_byte = ceiling_byte; | |
9169c321 JB |
568 | |
569 | /* Now the text after start is an unknown region, and | |
570 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 571 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
572 | } |
573 | ||
574 | /* The dumb loop can only scan text stored in contiguous | |
575 | bytes. BUFFER_CEILING_OF returns the last character | |
576 | position that is contiguous, so the ceiling is the | |
577 | position after that. */ | |
67ce527d KH |
578 | tem = BUFFER_CEILING_OF (start_byte); |
579 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
580 | |
581 | { | |
177c0ea7 | 582 | /* The termination address of the dumb loop. */ |
fa8ed3e0 RS |
583 | register unsigned char *ceiling_addr |
584 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
585 | register unsigned char *cursor | |
586 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
587 | unsigned char *base = cursor; |
588 | ||
589 | while (cursor < ceiling_addr) | |
590 | { | |
591 | unsigned char *scan_start = cursor; | |
592 | ||
593 | /* The dumb loop. */ | |
594 | while (*cursor != target && ++cursor < ceiling_addr) | |
595 | ; | |
596 | ||
597 | /* If we're looking for newlines, cache the fact that | |
598 | the region from start to cursor is free of them. */ | |
599 | if (target == '\n' && newline_cache) | |
600 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
601 | start_byte + scan_start - base, |
602 | start_byte + cursor - base); | |
9169c321 JB |
603 | |
604 | /* Did we find the target character? */ | |
605 | if (cursor < ceiling_addr) | |
606 | { | |
607 | if (--count == 0) | |
608 | { | |
609 | immediate_quit = 0; | |
fa8ed3e0 | 610 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
611 | } |
612 | cursor++; | |
613 | } | |
614 | } | |
615 | ||
fa8ed3e0 | 616 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 617 | } |
ca1d1d23 JB |
618 | } |
619 | else | |
9169c321 JB |
620 | while (start > end) |
621 | { | |
622 | /* The last character to check before the next obstacle. */ | |
fa8ed3e0 RS |
623 | int ceiling_byte = CHAR_TO_BYTE (end); |
624 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 625 | int tem; |
9169c321 JB |
626 | |
627 | /* Consult the newline cache, if appropriate. */ | |
628 | if (target == '\n' && newline_cache) | |
629 | { | |
630 | int next_change; | |
631 | immediate_quit = 0; | |
632 | while (region_cache_backward | |
fa8ed3e0 RS |
633 | (current_buffer, newline_cache, start_byte, &next_change)) |
634 | start_byte = next_change; | |
cbe0db0d | 635 | immediate_quit = allow_quit; |
9169c321 JB |
636 | |
637 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
638 | if (start_byte <= ceiling_byte) |
639 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
640 | |
641 | /* Now the text before start is an unknown region, and | |
642 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 643 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
644 | } |
645 | ||
646 | /* Stop scanning before the gap. */ | |
67ce527d KH |
647 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
648 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
649 | |
650 | { | |
651 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
652 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
653 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
654 | unsigned char *base = cursor; |
655 | ||
656 | while (cursor >= ceiling_addr) | |
657 | { | |
658 | unsigned char *scan_start = cursor; | |
659 | ||
660 | while (*cursor != target && --cursor >= ceiling_addr) | |
661 | ; | |
662 | ||
663 | /* If we're looking for newlines, cache the fact that | |
664 | the region from after the cursor to start is free of them. */ | |
665 | if (target == '\n' && newline_cache) | |
666 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
667 | start_byte + cursor - base, |
668 | start_byte + scan_start - base); | |
9169c321 JB |
669 | |
670 | /* Did we find the target character? */ | |
671 | if (cursor >= ceiling_addr) | |
672 | { | |
673 | if (++count >= 0) | |
674 | { | |
675 | immediate_quit = 0; | |
fa8ed3e0 | 676 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
677 | } |
678 | cursor--; | |
679 | } | |
680 | } | |
681 | ||
fa8ed3e0 | 682 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
683 | } |
684 | } | |
685 | ||
ca1d1d23 JB |
686 | immediate_quit = 0; |
687 | if (shortage != 0) | |
ffd56f97 | 688 | *shortage = count * direction; |
9169c321 | 689 | return start; |
ca1d1d23 | 690 | } |
fa8ed3e0 RS |
691 | \f |
692 | /* Search for COUNT instances of a line boundary, which means either a | |
693 | newline or (if selective display enabled) a carriage return. | |
694 | Start at START. If COUNT is negative, search backwards. | |
695 | ||
696 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
697 | ||
698 | If we find COUNT instances. we position after (always after, | |
699 | even if scanning backwards) the COUNTth match, and return 0. | |
700 | ||
701 | If we don't find COUNT instances before reaching the end of the | |
702 | buffer (or the beginning, if scanning backwards), we return | |
703 | the number of line boundaries left unfound, and position at | |
704 | the limit we bumped up against. | |
705 | ||
706 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 707 | except in special cases. */ |
ca1d1d23 | 708 | |
63fa018d | 709 | int |
fa8ed3e0 RS |
710 | scan_newline (start, start_byte, limit, limit_byte, count, allow_quit) |
711 | int start, start_byte; | |
712 | int limit, limit_byte; | |
713 | register int count; | |
714 | int allow_quit; | |
63fa018d | 715 | { |
fa8ed3e0 RS |
716 | int direction = ((count > 0) ? 1 : -1); |
717 | ||
718 | register unsigned char *cursor; | |
719 | unsigned char *base; | |
720 | ||
721 | register int ceiling; | |
722 | register unsigned char *ceiling_addr; | |
723 | ||
d5d57b92 RS |
724 | int old_immediate_quit = immediate_quit; |
725 | ||
fa8ed3e0 RS |
726 | /* The code that follows is like scan_buffer |
727 | but checks for either newline or carriage return. */ | |
728 | ||
d5d57b92 RS |
729 | if (allow_quit) |
730 | immediate_quit++; | |
fa8ed3e0 RS |
731 | |
732 | start_byte = CHAR_TO_BYTE (start); | |
733 | ||
734 | if (count > 0) | |
735 | { | |
736 | while (start_byte < limit_byte) | |
737 | { | |
738 | ceiling = BUFFER_CEILING_OF (start_byte); | |
739 | ceiling = min (limit_byte - 1, ceiling); | |
740 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
741 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
742 | while (1) | |
743 | { | |
744 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
745 | ; | |
746 | ||
747 | if (cursor != ceiling_addr) | |
748 | { | |
749 | if (--count == 0) | |
750 | { | |
d5d57b92 | 751 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
752 | start_byte = start_byte + cursor - base + 1; |
753 | start = BYTE_TO_CHAR (start_byte); | |
754 | TEMP_SET_PT_BOTH (start, start_byte); | |
755 | return 0; | |
756 | } | |
757 | else | |
758 | if (++cursor == ceiling_addr) | |
759 | break; | |
760 | } | |
761 | else | |
762 | break; | |
763 | } | |
764 | start_byte += cursor - base; | |
765 | } | |
766 | } | |
767 | else | |
768 | { | |
fa8ed3e0 RS |
769 | while (start_byte > limit_byte) |
770 | { | |
771 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
772 | ceiling = max (limit_byte, ceiling); | |
773 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
774 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
775 | while (1) | |
776 | { | |
777 | while (--cursor != ceiling_addr && *cursor != '\n') | |
778 | ; | |
779 | ||
780 | if (cursor != ceiling_addr) | |
781 | { | |
782 | if (++count == 0) | |
783 | { | |
d5d57b92 | 784 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
785 | /* Return the position AFTER the match we found. */ |
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 | } | |
792 | else | |
793 | break; | |
794 | } | |
795 | /* Here we add 1 to compensate for the last decrement | |
796 | of CURSOR, which took it past the valid range. */ | |
797 | start_byte += cursor - base + 1; | |
798 | } | |
799 | } | |
800 | ||
801 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 802 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
803 | |
804 | return count * direction; | |
63fa018d RS |
805 | } |
806 | ||
ca1d1d23 | 807 | int |
fa8ed3e0 | 808 | find_next_newline_no_quit (from, cnt) |
ca1d1d23 JB |
809 | register int from, cnt; |
810 | { | |
fa8ed3e0 | 811 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
9169c321 JB |
812 | } |
813 | ||
9169c321 JB |
814 | /* Like find_next_newline, but returns position before the newline, |
815 | not after, and only search up to TO. This isn't just | |
816 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 817 | |
9169c321 JB |
818 | int |
819 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 820 | int from, to, cnt; |
9169c321 JB |
821 | { |
822 | int shortage; | |
823 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
824 | ||
825 | if (shortage == 0) | |
826 | pos--; | |
177c0ea7 | 827 | |
9169c321 | 828 | return pos; |
ca1d1d23 JB |
829 | } |
830 | \f | |
ca1d1d23 JB |
831 | /* Subroutines of Lisp buffer search functions. */ |
832 | ||
833 | static Lisp_Object | |
b819a390 | 834 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
835 | Lisp_Object string, bound, noerror, count; |
836 | int direction; | |
837 | int RE; | |
b819a390 | 838 | int posix; |
ca1d1d23 JB |
839 | { |
840 | register int np; | |
9f43ad85 | 841 | int lim, lim_byte; |
ca1d1d23 JB |
842 | int n = direction; |
843 | ||
844 | if (!NILP (count)) | |
845 | { | |
b7826503 | 846 | CHECK_NUMBER (count); |
ca1d1d23 JB |
847 | n *= XINT (count); |
848 | } | |
849 | ||
b7826503 | 850 | CHECK_STRING (string); |
ca1d1d23 | 851 | if (NILP (bound)) |
9f43ad85 RS |
852 | { |
853 | if (n > 0) | |
854 | lim = ZV, lim_byte = ZV_BYTE; | |
855 | else | |
856 | lim = BEGV, lim_byte = BEGV_BYTE; | |
857 | } | |
ca1d1d23 JB |
858 | else |
859 | { | |
b7826503 | 860 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 861 | lim = XINT (bound); |
6ec8bbd2 | 862 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
863 | error ("Invalid search bound (wrong side of point)"); |
864 | if (lim > ZV) | |
9f43ad85 | 865 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 866 | else if (lim < BEGV) |
9f43ad85 | 867 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
868 | else |
869 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
870 | } |
871 | ||
9f43ad85 | 872 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
ca1d1d23 | 873 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 874 | ? current_buffer->case_canon_table |
3135e9fd | 875 | : Qnil), |
ca1d1d23 | 876 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 877 | ? current_buffer->case_eqv_table |
3135e9fd | 878 | : Qnil), |
b819a390 | 879 | posix); |
ca1d1d23 JB |
880 | if (np <= 0) |
881 | { | |
882 | if (NILP (noerror)) | |
883 | return signal_failure (string); | |
884 | if (!EQ (noerror, Qt)) | |
885 | { | |
886 | if (lim < BEGV || lim > ZV) | |
887 | abort (); | |
9f43ad85 | 888 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
889 | return Qnil; |
890 | #if 0 /* This would be clean, but maybe programs depend on | |
891 | a value of nil here. */ | |
481399bf | 892 | np = lim; |
a5f217b8 | 893 | #endif |
ca1d1d23 | 894 | } |
481399bf RS |
895 | else |
896 | return Qnil; | |
ca1d1d23 JB |
897 | } |
898 | ||
899 | if (np < BEGV || np > ZV) | |
900 | abort (); | |
901 | ||
902 | SET_PT (np); | |
903 | ||
904 | return make_number (np); | |
905 | } | |
906 | \f | |
fa8ed3e0 RS |
907 | /* Return 1 if REGEXP it matches just one constant string. */ |
908 | ||
b6d6a51c KH |
909 | static int |
910 | trivial_regexp_p (regexp) | |
911 | Lisp_Object regexp; | |
912 | { | |
d5db4077 KR |
913 | int len = SBYTES (regexp); |
914 | unsigned char *s = SDATA (regexp); | |
b6d6a51c KH |
915 | while (--len >= 0) |
916 | { | |
917 | switch (*s++) | |
918 | { | |
919 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
920 | return 0; | |
921 | case '\\': | |
922 | if (--len < 0) | |
923 | return 0; | |
924 | switch (*s++) | |
925 | { | |
926 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
927 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 928 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 929 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 930 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
931 | case '6': case '7': case '8': case '9': |
932 | return 0; | |
933 | } | |
934 | } | |
935 | } | |
936 | return 1; | |
937 | } | |
938 | ||
ca325161 | 939 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 940 | starting at position POS and stopping at position LIM, |
b819a390 | 941 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
942 | a regular expression if RE is true. |
943 | ||
944 | If N is positive, searching is forward and LIM must be greater than POS. | |
945 | If N is negative, searching is backward and LIM must be less than POS. | |
946 | ||
facdc750 | 947 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 948 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
949 | (if searching backward) or the end (if searching forward). |
950 | ||
951 | POSIX is nonzero if we want full backtracking (POSIX style) | |
952 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 953 | |
aff2ce94 RS |
954 | #define TRANSLATE(out, trt, d) \ |
955 | do \ | |
956 | { \ | |
957 | if (! NILP (trt)) \ | |
958 | { \ | |
959 | Lisp_Object temp; \ | |
960 | temp = Faref (trt, make_number (d)); \ | |
961 | if (INTEGERP (temp)) \ | |
962 | out = XINT (temp); \ | |
963 | else \ | |
964 | out = d; \ | |
965 | } \ | |
966 | else \ | |
967 | out = d; \ | |
968 | } \ | |
969 | while (0) | |
facdc750 | 970 | |
b819a390 | 971 | static int |
9f43ad85 RS |
972 | search_buffer (string, pos, pos_byte, lim, lim_byte, n, |
973 | RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
974 | Lisp_Object string; |
975 | int pos; | |
9f43ad85 | 976 | int pos_byte; |
ca1d1d23 | 977 | int lim; |
9f43ad85 | 978 | int lim_byte; |
ca1d1d23 JB |
979 | int n; |
980 | int RE; | |
facdc750 RS |
981 | Lisp_Object trt; |
982 | Lisp_Object inverse_trt; | |
b819a390 | 983 | int posix; |
ca1d1d23 | 984 | { |
d5db4077 KR |
985 | int len = SCHARS (string); |
986 | int len_byte = SBYTES (string); | |
facdc750 | 987 | register int i; |
ca1d1d23 | 988 | |
7074fde6 FP |
989 | if (running_asynch_code) |
990 | save_search_regs (); | |
991 | ||
a7e4cdde | 992 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 993 | /* Null string is found at starting position. */ |
a7e4cdde | 994 | if (len == 0 || n == 0) |
ca325161 | 995 | { |
0353b28f | 996 | set_search_regs (pos_byte, 0); |
ca325161 RS |
997 | return pos; |
998 | } | |
3f57a499 | 999 | |
b6d6a51c | 1000 | if (RE && !trivial_regexp_p (string)) |
ca1d1d23 | 1001 | { |
facdc750 RS |
1002 | unsigned char *p1, *p2; |
1003 | int s1, s2; | |
487282dc KH |
1004 | struct re_pattern_buffer *bufp; |
1005 | ||
0c8533c6 RS |
1006 | bufp = compile_pattern (string, &search_regs, trt, posix, |
1007 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 | 1008 | |
ca1d1d23 JB |
1009 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1010 | because letting this function finish | |
1011 | can take too long. */ | |
1012 | QUIT; /* Do a pending quit right away, | |
1013 | to avoid paradoxical behavior */ | |
1014 | /* Get pointers and sizes of the two strings | |
1015 | that make up the visible portion of the buffer. */ | |
1016 | ||
1017 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1018 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1019 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1020 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1021 | if (s1 < 0) |
1022 | { | |
1023 | p2 = p1; | |
fa8ed3e0 | 1024 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1025 | s1 = 0; |
1026 | } | |
1027 | if (s2 < 0) | |
1028 | { | |
fa8ed3e0 | 1029 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1030 | s2 = 0; |
1031 | } | |
8bb43c28 | 1032 | re_match_object = Qnil; |
177c0ea7 | 1033 | |
ca1d1d23 JB |
1034 | while (n < 0) |
1035 | { | |
42db823b | 1036 | int val; |
487282dc | 1037 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1038 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1039 | &search_regs, | |
42db823b | 1040 | /* Don't allow match past current point */ |
4996330b | 1041 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1042 | if (val == -2) |
b6d6a51c KH |
1043 | { |
1044 | matcher_overflow (); | |
1045 | } | |
ca1d1d23 JB |
1046 | if (val >= 0) |
1047 | { | |
26aff150 | 1048 | pos_byte = search_regs.start[0] + BEGV_BYTE; |
4746118a | 1049 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1050 | if (search_regs.start[i] >= 0) |
1051 | { | |
fa8ed3e0 RS |
1052 | search_regs.start[i] |
1053 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1054 | search_regs.end[i] | |
1055 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1056 | } |
a3668d92 | 1057 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1058 | /* Set pos to the new position. */ |
1059 | pos = search_regs.start[0]; | |
1060 | } | |
1061 | else | |
1062 | { | |
1063 | immediate_quit = 0; | |
1064 | return (n); | |
1065 | } | |
1066 | n++; | |
1067 | } | |
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, | |
1074 | lim_byte - BEGV_BYTE); | |
ca1d1d23 | 1075 | if (val == -2) |
b6d6a51c KH |
1076 | { |
1077 | matcher_overflow (); | |
1078 | } | |
ca1d1d23 JB |
1079 | if (val >= 0) |
1080 | { | |
26aff150 | 1081 | pos_byte = search_regs.end[0] + BEGV_BYTE; |
4746118a | 1082 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1083 | if (search_regs.start[i] >= 0) |
1084 | { | |
fa8ed3e0 RS |
1085 | search_regs.start[i] |
1086 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1087 | search_regs.end[i] | |
1088 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1089 | } |
a3668d92 | 1090 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1091 | pos = search_regs.end[0]; |
1092 | } | |
1093 | else | |
1094 | { | |
1095 | immediate_quit = 0; | |
1096 | return (0 - n); | |
1097 | } | |
1098 | n--; | |
1099 | } | |
1100 | immediate_quit = 0; | |
1101 | return (pos); | |
1102 | } | |
1103 | else /* non-RE case */ | |
1104 | { | |
facdc750 RS |
1105 | unsigned char *raw_pattern, *pat; |
1106 | int raw_pattern_size; | |
1107 | int raw_pattern_size_byte; | |
1108 | unsigned char *patbuf; | |
1109 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); | |
d5db4077 | 1110 | unsigned char *base_pat = SDATA (string); |
8f924df7 KH |
1111 | /* High bits of char; 0 for ASCII characters, (CHAR & ~0x3F) |
1112 | otherwise. Characters of the same high bits have the same | |
1113 | sequence of bytes but last. To do the BM search, all | |
1114 | characters in STRING must have the same high bits (including | |
1115 | their case translations). */ | |
f5f7578b | 1116 | int char_high_bits = -1; |
040272ce | 1117 | int boyer_moore_ok = 1; |
facdc750 RS |
1118 | |
1119 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1120 | We must convert PATTERN to match that, or we will not really | |
1121 | find things right. */ | |
1122 | ||
1123 | if (multibyte == STRING_MULTIBYTE (string)) | |
1124 | { | |
d5db4077 KR |
1125 | raw_pattern = (unsigned char *) SDATA (string); |
1126 | raw_pattern_size = SCHARS (string); | |
1127 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1128 | } |
1129 | else if (multibyte) | |
1130 | { | |
d5db4077 | 1131 | raw_pattern_size = SCHARS (string); |
facdc750 | 1132 | raw_pattern_size_byte |
d5db4077 | 1133 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1134 | raw_pattern_size); |
7276d3d8 | 1135 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1136 | copy_text (SDATA (string), raw_pattern, |
1137 | SCHARS (string), 0, 1); | |
facdc750 RS |
1138 | } |
1139 | else | |
1140 | { | |
1141 | /* Converting multibyte to single-byte. | |
1142 | ||
1143 | ??? Perhaps this conversion should be done in a special way | |
1144 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1145 | so that only the multibyte chars which really correspond to | |
1146 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1147 | raw_pattern_size = SCHARS (string); |
1148 | raw_pattern_size_byte = SCHARS (string); | |
7276d3d8 | 1149 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
1150 | copy_text (SDATA (string), raw_pattern, |
1151 | SBYTES (string), 1, 0); | |
facdc750 RS |
1152 | } |
1153 | ||
1154 | /* Copy and optionally translate the pattern. */ | |
1155 | len = raw_pattern_size; | |
1156 | len_byte = raw_pattern_size_byte; | |
f5f7578b | 1157 | patbuf = (unsigned char *) alloca (len * MAX_MULTIBYTE_LENGTH); |
facdc750 RS |
1158 | pat = patbuf; |
1159 | base_pat = raw_pattern; | |
1160 | if (multibyte) | |
1161 | { | |
1162 | while (--len >= 0) | |
1163 | { | |
aff2ce94 | 1164 | int c, translated, inverse; |
f5f7578b | 1165 | int in_charlen; |
facdc750 RS |
1166 | |
1167 | /* If we got here and the RE flag is set, it's because we're | |
1168 | dealing with a regexp known to be trivial, so the backslash | |
1169 | just quotes the next character. */ | |
1170 | if (RE && *base_pat == '\\') | |
1171 | { | |
1172 | len--; | |
1173 | len_byte--; | |
1174 | base_pat++; | |
1175 | } | |
1176 | ||
1177 | c = STRING_CHAR_AND_LENGTH (base_pat, len_byte, in_charlen); | |
040272ce | 1178 | |
facdc750 | 1179 | /* Translate the character, if requested. */ |
aff2ce94 | 1180 | TRANSLATE (translated, trt, c); |
aff2ce94 RS |
1181 | TRANSLATE (inverse, inverse_trt, c); |
1182 | ||
facdc750 RS |
1183 | /* Did this char actually get translated? |
1184 | Would any other char get translated into it? */ | |
aff2ce94 | 1185 | if (translated != c || inverse != c) |
facdc750 RS |
1186 | { |
1187 | /* Keep track of which character set row | |
1188 | contains the characters that need translation. */ | |
8f924df7 KH |
1189 | int this_high_bit = ASCII_CHAR_P (c) ? 0 : (c & ~0x3F); |
1190 | int c1 = inverse != c ? inverse : translated; | |
1191 | int trt_high_bit = ASCII_CHAR_P (c1) ? 0 : (c1 & ~0x3F); | |
d2ac725b | 1192 | |
f5f7578b | 1193 | if (this_high_bit != trt_high_bit) |
d2ac725b | 1194 | boyer_moore_ok = 0; |
f5f7578b KH |
1195 | else if (char_high_bits == -1) |
1196 | char_high_bits = this_high_bit; | |
1197 | else if (char_high_bits != this_high_bit) | |
facdc750 RS |
1198 | /* If two different rows appear, needing translation, |
1199 | then we cannot use boyer_moore search. */ | |
040272ce | 1200 | boyer_moore_ok = 0; |
aff2ce94 | 1201 | } |
facdc750 RS |
1202 | |
1203 | /* Store this character into the translated pattern. */ | |
f5f7578b | 1204 | CHAR_STRING_ADVANCE (translated, pat); |
facdc750 RS |
1205 | base_pat += in_charlen; |
1206 | len_byte -= in_charlen; | |
1207 | } | |
1208 | } | |
1209 | else | |
1210 | { | |
040272ce | 1211 | /* Unibyte buffer. */ |
f5f7578b | 1212 | char_high_bits = 0; |
facdc750 RS |
1213 | while (--len >= 0) |
1214 | { | |
040272ce | 1215 | int c, translated; |
facdc750 RS |
1216 | |
1217 | /* If we got here and the RE flag is set, it's because we're | |
1218 | dealing with a regexp known to be trivial, so the backslash | |
1219 | just quotes the next character. */ | |
1220 | if (RE && *base_pat == '\\') | |
1221 | { | |
1222 | len--; | |
1223 | base_pat++; | |
1224 | } | |
1225 | c = *base_pat++; | |
aff2ce94 | 1226 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1227 | *pat++ = translated; |
1228 | } | |
1229 | } | |
1230 | ||
1231 | len_byte = pat - patbuf; | |
1232 | len = raw_pattern_size; | |
1233 | pat = base_pat = patbuf; | |
1234 | ||
040272ce | 1235 | if (boyer_moore_ok) |
facdc750 | 1236 | return boyer_moore (n, pat, len, len_byte, trt, inverse_trt, |
aff2ce94 | 1237 | pos, pos_byte, lim, lim_byte, |
f5f7578b | 1238 | char_high_bits); |
facdc750 RS |
1239 | else |
1240 | return simple_search (n, pat, len, len_byte, trt, | |
1241 | pos, pos_byte, lim, lim_byte); | |
1242 | } | |
1243 | } | |
1244 | \f | |
1245 | /* Do a simple string search N times for the string PAT, | |
1246 | whose length is LEN/LEN_BYTE, | |
1247 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1248 | TRT is the translation table. | |
f8bd51c4 | 1249 | |
facdc750 RS |
1250 | Return the character position where the match is found. |
1251 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1252 | |
facdc750 RS |
1253 | This kind of search works regardless of what is in PAT and |
1254 | regardless of what is in TRT. It is used in cases where | |
1255 | boyer_moore cannot work. */ | |
1256 | ||
1257 | static int | |
1258 | simple_search (n, pat, len, len_byte, trt, pos, pos_byte, lim, lim_byte) | |
1259 | int n; | |
1260 | unsigned char *pat; | |
1261 | int len, len_byte; | |
1262 | Lisp_Object trt; | |
1263 | int pos, pos_byte; | |
1264 | int lim, lim_byte; | |
1265 | { | |
1266 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); | |
ab228c24 | 1267 | int forward = n > 0; |
facdc750 RS |
1268 | |
1269 | if (lim > pos && multibyte) | |
1270 | while (n > 0) | |
1271 | { | |
1272 | while (1) | |
f8bd51c4 | 1273 | { |
facdc750 RS |
1274 | /* Try matching at position POS. */ |
1275 | int this_pos = pos; | |
1276 | int this_pos_byte = pos_byte; | |
1277 | int this_len = len; | |
1278 | int this_len_byte = len_byte; | |
1279 | unsigned char *p = pat; | |
1280 | if (pos + len > lim) | |
1281 | goto stop; | |
1282 | ||
1283 | while (this_len > 0) | |
1284 | { | |
1285 | int charlen, buf_charlen; | |
ab228c24 | 1286 | int pat_ch, buf_ch; |
facdc750 | 1287 | |
ab228c24 | 1288 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1289 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1290 | ZV_BYTE - this_pos_byte, | |
1291 | buf_charlen); | |
aff2ce94 | 1292 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1293 | |
1294 | if (buf_ch != pat_ch) | |
1295 | break; | |
ab228c24 RS |
1296 | |
1297 | this_len_byte -= charlen; | |
1298 | this_len--; | |
1299 | p += charlen; | |
1300 | ||
1301 | this_pos_byte += buf_charlen; | |
1302 | this_pos++; | |
facdc750 RS |
1303 | } |
1304 | ||
1305 | if (this_len == 0) | |
1306 | { | |
1307 | pos += len; | |
1308 | pos_byte += len_byte; | |
1309 | break; | |
1310 | } | |
1311 | ||
1312 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1313 | } |
facdc750 RS |
1314 | |
1315 | n--; | |
1316 | } | |
1317 | else if (lim > pos) | |
1318 | while (n > 0) | |
1319 | { | |
1320 | while (1) | |
f8bd51c4 | 1321 | { |
facdc750 RS |
1322 | /* Try matching at position POS. */ |
1323 | int this_pos = pos; | |
1324 | int this_len = len; | |
1325 | unsigned char *p = pat; | |
1326 | ||
1327 | if (pos + len > lim) | |
1328 | goto stop; | |
1329 | ||
1330 | while (this_len > 0) | |
1331 | { | |
1332 | int pat_ch = *p++; | |
1333 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1334 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1335 | |
1336 | if (buf_ch != pat_ch) | |
1337 | break; | |
ab228c24 RS |
1338 | |
1339 | this_len--; | |
1340 | this_pos++; | |
facdc750 RS |
1341 | } |
1342 | ||
1343 | if (this_len == 0) | |
1344 | { | |
1345 | pos += len; | |
1346 | break; | |
1347 | } | |
1348 | ||
1349 | pos++; | |
f8bd51c4 | 1350 | } |
facdc750 RS |
1351 | |
1352 | n--; | |
1353 | } | |
1354 | /* Backwards search. */ | |
1355 | else if (lim < pos && multibyte) | |
1356 | while (n < 0) | |
1357 | { | |
1358 | while (1) | |
f8bd51c4 | 1359 | { |
facdc750 RS |
1360 | /* Try matching at position POS. */ |
1361 | int this_pos = pos - len; | |
130b729c | 1362 | int this_pos_byte; |
facdc750 RS |
1363 | int this_len = len; |
1364 | int this_len_byte = len_byte; | |
1365 | unsigned char *p = pat; | |
1366 | ||
1367 | if (pos - len < lim) | |
1368 | goto stop; | |
130b729c | 1369 | this_pos_byte = CHAR_TO_BYTE (this_pos); |
facdc750 RS |
1370 | |
1371 | while (this_len > 0) | |
1372 | { | |
1373 | int charlen, buf_charlen; | |
ab228c24 | 1374 | int pat_ch, buf_ch; |
facdc750 | 1375 | |
ab228c24 | 1376 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1377 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1378 | ZV_BYTE - this_pos_byte, | |
1379 | buf_charlen); | |
aff2ce94 | 1380 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1381 | |
1382 | if (buf_ch != pat_ch) | |
1383 | break; | |
ab228c24 RS |
1384 | |
1385 | this_len_byte -= charlen; | |
1386 | this_len--; | |
1387 | p += charlen; | |
1388 | this_pos_byte += buf_charlen; | |
1389 | this_pos++; | |
facdc750 RS |
1390 | } |
1391 | ||
1392 | if (this_len == 0) | |
1393 | { | |
1394 | pos -= len; | |
1395 | pos_byte -= len_byte; | |
1396 | break; | |
1397 | } | |
1398 | ||
1399 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1400 | } |
1401 | ||
facdc750 RS |
1402 | n++; |
1403 | } | |
1404 | else if (lim < pos) | |
1405 | while (n < 0) | |
1406 | { | |
1407 | while (1) | |
b6d6a51c | 1408 | { |
facdc750 RS |
1409 | /* Try matching at position POS. */ |
1410 | int this_pos = pos - len; | |
1411 | int this_len = len; | |
1412 | unsigned char *p = pat; | |
1413 | ||
1414 | if (pos - len < lim) | |
1415 | goto stop; | |
1416 | ||
1417 | while (this_len > 0) | |
1418 | { | |
1419 | int pat_ch = *p++; | |
1420 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1421 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1422 | |
1423 | if (buf_ch != pat_ch) | |
1424 | break; | |
ab228c24 RS |
1425 | this_len--; |
1426 | this_pos++; | |
facdc750 RS |
1427 | } |
1428 | ||
1429 | if (this_len == 0) | |
b6d6a51c | 1430 | { |
facdc750 RS |
1431 | pos -= len; |
1432 | break; | |
b6d6a51c | 1433 | } |
facdc750 RS |
1434 | |
1435 | pos--; | |
b6d6a51c | 1436 | } |
facdc750 RS |
1437 | |
1438 | n++; | |
b6d6a51c | 1439 | } |
facdc750 RS |
1440 | |
1441 | stop: | |
1442 | if (n == 0) | |
aff2ce94 | 1443 | { |
ab228c24 RS |
1444 | if (forward) |
1445 | set_search_regs ((multibyte ? pos_byte : pos) - len_byte, len_byte); | |
1446 | else | |
1447 | set_search_regs (multibyte ? pos_byte : pos, len_byte); | |
aff2ce94 RS |
1448 | |
1449 | return pos; | |
1450 | } | |
facdc750 RS |
1451 | else if (n > 0) |
1452 | return -n; | |
1453 | else | |
1454 | return n; | |
1455 | } | |
1456 | \f | |
1457 | /* Do Boyer-Moore search N times for the string PAT, | |
1458 | whose length is LEN/LEN_BYTE, | |
1459 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1460 | DIRECTION says which direction we search in. | |
1461 | TRT and INVERSE_TRT are translation tables. | |
1462 | ||
1463 | This kind of search works if all the characters in PAT that have | |
1464 | nontrivial translation are the same aside from the last byte. This | |
1465 | makes it possible to translate just the last byte of a character, | |
1466 | and do so after just a simple test of the context. | |
1467 | ||
1468 | If that criterion is not satisfied, do not call this function. */ | |
1469 | ||
1470 | static int | |
1471 | boyer_moore (n, base_pat, len, len_byte, trt, inverse_trt, | |
f5f7578b | 1472 | pos, pos_byte, lim, lim_byte, char_high_bits) |
facdc750 RS |
1473 | int n; |
1474 | unsigned char *base_pat; | |
1475 | int len, len_byte; | |
1476 | Lisp_Object trt; | |
1477 | Lisp_Object inverse_trt; | |
1478 | int pos, pos_byte; | |
1479 | int lim, lim_byte; | |
f5f7578b | 1480 | int char_high_bits; |
facdc750 RS |
1481 | { |
1482 | int direction = ((n > 0) ? 1 : -1); | |
1483 | register int dirlen; | |
a968f437 | 1484 | int infinity, limit, stride_for_teases = 0; |
facdc750 RS |
1485 | register int *BM_tab; |
1486 | int *BM_tab_base; | |
177c0ea7 | 1487 | register unsigned char *cursor, *p_limit; |
facdc750 | 1488 | register int i, j; |
cb6792d2 | 1489 | unsigned char *pat, *pat_end; |
facdc750 RS |
1490 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); |
1491 | ||
1492 | unsigned char simple_translate[0400]; | |
6bbd7a29 GM |
1493 | int translate_prev_byte = 0; |
1494 | int translate_anteprev_byte = 0; | |
facdc750 RS |
1495 | |
1496 | #ifdef C_ALLOCA | |
1497 | int BM_tab_space[0400]; | |
1498 | BM_tab = &BM_tab_space[0]; | |
1499 | #else | |
1500 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
1501 | #endif | |
1502 | /* The general approach is that we are going to maintain that we know */ | |
1503 | /* the first (closest to the present position, in whatever direction */ | |
1504 | /* we're searching) character that could possibly be the last */ | |
1505 | /* (furthest from present position) character of a valid match. We */ | |
1506 | /* advance the state of our knowledge by looking at that character */ | |
1507 | /* and seeing whether it indeed matches the last character of the */ | |
1508 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1509 | /* move our pointer (to putative last characters) as far as is */ | |
1510 | /* logically possible. This amount of movement, which I call a */ | |
1511 | /* stride, will be the length of the pattern if the actual character */ | |
1512 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1513 | /* from the last occurrence of that character to the end of the */ | |
1514 | /* pattern. */ | |
1515 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1516 | /* characters that match the last character. This allows use of only */ | |
1517 | /* a single test, a test for having gone past the end of the */ | |
1518 | /* permissible match region, to test for both possible matches (when */ | |
1519 | /* the stride goes past the end immediately) and failure to */ | |
177c0ea7 | 1520 | /* match (where you get nudged past the end one stride at a time). */ |
facdc750 RS |
1521 | |
1522 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1523 | /* is determined only by the last character of the putative match. */ | |
1524 | /* If that character does not match, we will stride the proper */ | |
1525 | /* distance to propose a match that superimposes it on the last */ | |
1526 | /* instance of a character that matches it (per trt), or misses */ | |
177c0ea7 | 1527 | /* it entirely if there is none. */ |
facdc750 RS |
1528 | |
1529 | dirlen = len_byte * direction; | |
1530 | infinity = dirlen - (lim_byte + pos_byte + len_byte + len_byte) * direction; | |
cb6792d2 RS |
1531 | |
1532 | /* Record position after the end of the pattern. */ | |
1533 | pat_end = base_pat + len_byte; | |
1534 | /* BASE_PAT points to a character that we start scanning from. | |
1535 | It is the first character in a forward search, | |
1536 | the last character in a backward search. */ | |
facdc750 | 1537 | if (direction < 0) |
cb6792d2 RS |
1538 | base_pat = pat_end - 1; |
1539 | ||
facdc750 RS |
1540 | BM_tab_base = BM_tab; |
1541 | BM_tab += 0400; | |
1542 | j = dirlen; /* to get it in a register */ | |
1543 | /* A character that does not appear in the pattern induces a */ | |
1544 | /* stride equal to the pattern length. */ | |
1545 | while (BM_tab_base != BM_tab) | |
1546 | { | |
1547 | *--BM_tab = j; | |
1548 | *--BM_tab = j; | |
1549 | *--BM_tab = j; | |
1550 | *--BM_tab = j; | |
1551 | } | |
1552 | ||
1553 | /* We use this for translation, instead of TRT itself. | |
1554 | We fill this in to handle the characters that actually | |
1555 | occur in the pattern. Others don't matter anyway! */ | |
1556 | bzero (simple_translate, sizeof simple_translate); | |
1557 | for (i = 0; i < 0400; i++) | |
1558 | simple_translate[i] = i; | |
1559 | ||
1560 | i = 0; | |
1561 | while (i != infinity) | |
1562 | { | |
cb6792d2 | 1563 | unsigned char *ptr = base_pat + i; |
facdc750 RS |
1564 | i += direction; |
1565 | if (i == dirlen) | |
1566 | i = infinity; | |
1567 | if (! NILP (trt)) | |
ca1d1d23 | 1568 | { |
facdc750 | 1569 | int ch; |
aff2ce94 | 1570 | int untranslated; |
facdc750 RS |
1571 | int this_translated = 1; |
1572 | ||
1573 | if (multibyte | |
cb6792d2 RS |
1574 | /* Is *PTR the last byte of a character? */ |
1575 | && (pat_end - ptr == 1 || CHAR_HEAD_P (ptr[1]))) | |
ca1d1d23 | 1576 | { |
facdc750 RS |
1577 | unsigned char *charstart = ptr; |
1578 | while (! CHAR_HEAD_P (*charstart)) | |
1579 | charstart--; | |
aff2ce94 | 1580 | untranslated = STRING_CHAR (charstart, ptr - charstart + 1); |
8f924df7 KH |
1581 | if (char_high_bits |
1582 | == (ASCII_CHAR_P (untranslated) ? 0 : untranslated & ~0x3F)) | |
facdc750 | 1583 | { |
ab228c24 | 1584 | TRANSLATE (ch, trt, untranslated); |
aff2ce94 RS |
1585 | if (! CHAR_HEAD_P (*ptr)) |
1586 | { | |
1587 | translate_prev_byte = ptr[-1]; | |
1588 | if (! CHAR_HEAD_P (translate_prev_byte)) | |
1589 | translate_anteprev_byte = ptr[-2]; | |
1590 | } | |
facdc750 | 1591 | } |
aff2ce94 | 1592 | else |
ab228c24 RS |
1593 | { |
1594 | this_translated = 0; | |
1595 | ch = *ptr; | |
1596 | } | |
ca1d1d23 | 1597 | } |
facdc750 | 1598 | else if (!multibyte) |
aff2ce94 | 1599 | TRANSLATE (ch, trt, *ptr); |
ca1d1d23 JB |
1600 | else |
1601 | { | |
facdc750 RS |
1602 | ch = *ptr; |
1603 | this_translated = 0; | |
ca1d1d23 | 1604 | } |
facdc750 | 1605 | |
b9e5a425 KH |
1606 | if (this_translated |
1607 | && ch >= 0200) | |
1608 | j = (ch & 0x3F) | 0200; | |
ab228c24 RS |
1609 | else |
1610 | j = (unsigned char) ch; | |
1611 | ||
facdc750 RS |
1612 | if (i == infinity) |
1613 | stride_for_teases = BM_tab[j]; | |
ab228c24 | 1614 | |
facdc750 RS |
1615 | BM_tab[j] = dirlen - i; |
1616 | /* A translation table is accompanied by its inverse -- see */ | |
177c0ea7 | 1617 | /* comment following downcase_table for details */ |
facdc750 | 1618 | if (this_translated) |
ab228c24 RS |
1619 | { |
1620 | int starting_ch = ch; | |
1621 | int starting_j = j; | |
1622 | while (1) | |
1623 | { | |
1624 | TRANSLATE (ch, inverse_trt, ch); | |
b9e5a425 KH |
1625 | if (ch > 0200) |
1626 | j = (ch & 0x3F) | 0200; | |
ab228c24 RS |
1627 | else |
1628 | j = (unsigned char) ch; | |
1629 | ||
1630 | /* For all the characters that map into CH, | |
1631 | set up simple_translate to map the last byte | |
1632 | into STARTING_J. */ | |
1633 | simple_translate[j] = starting_j; | |
1634 | if (ch == starting_ch) | |
1635 | break; | |
1636 | BM_tab[j] = dirlen - i; | |
1637 | } | |
1638 | } | |
facdc750 RS |
1639 | } |
1640 | else | |
1641 | { | |
1642 | j = *ptr; | |
1643 | ||
1644 | if (i == infinity) | |
1645 | stride_for_teases = BM_tab[j]; | |
1646 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1647 | } |
facdc750 RS |
1648 | /* stride_for_teases tells how much to stride if we get a */ |
1649 | /* match on the far character but are subsequently */ | |
1650 | /* disappointed, by recording what the stride would have been */ | |
1651 | /* for that character if the last character had been */ | |
1652 | /* different. */ | |
1653 | } | |
1654 | infinity = dirlen - infinity; | |
1655 | pos_byte += dirlen - ((direction > 0) ? direction : 0); | |
1656 | /* loop invariant - POS_BYTE points at where last char (first | |
1657 | char if reverse) of pattern would align in a possible match. */ | |
1658 | while (n != 0) | |
1659 | { | |
1660 | int tail_end; | |
1661 | unsigned char *tail_end_ptr; | |
1662 | ||
1663 | /* It's been reported that some (broken) compiler thinks that | |
1664 | Boolean expressions in an arithmetic context are unsigned. | |
1665 | Using an explicit ?1:0 prevents this. */ | |
1666 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1667 | < 0) | |
1668 | return (n * (0 - direction)); | |
1669 | /* First we do the part we can by pointers (maybe nothing) */ | |
1670 | QUIT; | |
1671 | pat = base_pat; | |
1672 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1673 | if (direction > 0) |
1674 | { | |
1675 | limit = BUFFER_CEILING_OF (limit); | |
1676 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1677 | can take on without hitting edge of buffer or the gap. */ | |
1678 | limit = min (limit, pos_byte + 20000); | |
1679 | limit = min (limit, lim_byte - 1); | |
1680 | } | |
1681 | else | |
1682 | { | |
1683 | limit = BUFFER_FLOOR_OF (limit); | |
1684 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1685 | can take on without hitting edge of buffer or the gap. */ | |
1686 | limit = max (limit, pos_byte - 20000); | |
1687 | limit = max (limit, lim_byte); | |
1688 | } | |
facdc750 RS |
1689 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1690 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1691 | ||
1692 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1693 | { |
facdc750 RS |
1694 | unsigned char *p2; |
1695 | ||
1696 | p_limit = BYTE_POS_ADDR (limit); | |
1697 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
1698 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ | |
1699 | while (1) /* use one cursor setting as long as i can */ | |
ca1d1d23 | 1700 | { |
facdc750 | 1701 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1702 | { |
facdc750 RS |
1703 | /* Use signed comparison if appropriate |
1704 | to make cursor+infinity sure to be > p_limit. | |
1705 | Assuming that the buffer lies in a range of addresses | |
1706 | that are all "positive" (as ints) or all "negative", | |
1707 | either kind of comparison will work as long | |
1708 | as we don't step by infinity. So pick the kind | |
1709 | that works when we do step by infinity. */ | |
1710 | if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit) | |
1711 | while ((EMACS_INT) cursor <= (EMACS_INT) p_limit) | |
1712 | cursor += BM_tab[*cursor]; | |
ca1d1d23 | 1713 | else |
facdc750 RS |
1714 | while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit) |
1715 | cursor += BM_tab[*cursor]; | |
1716 | } | |
1717 | else | |
1718 | { | |
1719 | if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit) | |
1720 | while ((EMACS_INT) cursor >= (EMACS_INT) p_limit) | |
1721 | cursor += BM_tab[*cursor]; | |
1722 | else | |
1723 | while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit) | |
1724 | cursor += BM_tab[*cursor]; | |
1725 | } | |
ca1d1d23 | 1726 | /* If you are here, cursor is beyond the end of the searched region. */ |
facdc750 RS |
1727 | /* This can happen if you match on the far character of the pattern, */ |
1728 | /* because the "stride" of that character is infinity, a number able */ | |
1729 | /* to throw you well beyond the end of the search. It can also */ | |
1730 | /* happen if you fail to match within the permitted region and would */ | |
1731 | /* otherwise try a character beyond that region */ | |
1732 | if ((cursor - p_limit) * direction <= len_byte) | |
1733 | break; /* a small overrun is genuine */ | |
1734 | cursor -= infinity; /* large overrun = hit */ | |
1735 | i = dirlen - direction; | |
1736 | if (! NILP (trt)) | |
1737 | { | |
1738 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1739 | { |
facdc750 RS |
1740 | int ch; |
1741 | cursor -= direction; | |
1742 | /* Translate only the last byte of a character. */ | |
1743 | if (! multibyte | |
1744 | || ((cursor == tail_end_ptr | |
1745 | || CHAR_HEAD_P (cursor[1])) | |
1746 | && (CHAR_HEAD_P (cursor[0]) | |
1747 | || (translate_prev_byte == cursor[-1] | |
1748 | && (CHAR_HEAD_P (translate_prev_byte) | |
1749 | || translate_anteprev_byte == cursor[-2]))))) | |
1750 | ch = simple_translate[*cursor]; | |
1751 | else | |
1752 | ch = *cursor; | |
1753 | if (pat[i] != ch) | |
1754 | break; | |
ca1d1d23 | 1755 | } |
facdc750 RS |
1756 | } |
1757 | else | |
1758 | { | |
1759 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1760 | { |
facdc750 RS |
1761 | cursor -= direction; |
1762 | if (pat[i] != *cursor) | |
1763 | break; | |
ca1d1d23 | 1764 | } |
facdc750 RS |
1765 | } |
1766 | cursor += dirlen - i - direction; /* fix cursor */ | |
1767 | if (i + direction == 0) | |
1768 | { | |
1769 | int position; | |
0c8533c6 | 1770 | |
facdc750 | 1771 | cursor -= direction; |
1113d9db | 1772 | |
facdc750 RS |
1773 | position = pos_byte + cursor - p2 + ((direction > 0) |
1774 | ? 1 - len_byte : 0); | |
1775 | set_search_regs (position, len_byte); | |
ca325161 | 1776 | |
facdc750 RS |
1777 | if ((n -= direction) != 0) |
1778 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1779 | else |
facdc750 RS |
1780 | return ((direction > 0) |
1781 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1782 | } |
facdc750 RS |
1783 | else |
1784 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1785 | } |
facdc750 RS |
1786 | pos_byte += cursor - p2; |
1787 | } | |
1788 | else | |
1789 | /* Now we'll pick up a clump that has to be done the hard */ | |
1790 | /* way because it covers a discontinuity */ | |
1791 | { | |
1792 | limit = ((direction > 0) | |
1793 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1794 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1795 | limit = ((direction > 0) | |
1796 | ? min (limit + len_byte, lim_byte - 1) | |
1797 | : max (limit - len_byte, lim_byte)); | |
1798 | /* LIMIT is now the last value POS_BYTE can have | |
1799 | and still be valid for a possible match. */ | |
1800 | while (1) | |
ca1d1d23 | 1801 | { |
facdc750 RS |
1802 | /* This loop can be coded for space rather than */ |
1803 | /* speed because it will usually run only once. */ | |
1804 | /* (the reach is at most len + 21, and typically */ | |
177c0ea7 | 1805 | /* does not exceed len) */ |
facdc750 RS |
1806 | while ((limit - pos_byte) * direction >= 0) |
1807 | pos_byte += BM_tab[FETCH_BYTE (pos_byte)]; | |
1808 | /* now run the same tests to distinguish going off the */ | |
1809 | /* end, a match or a phony match. */ | |
1810 | if ((pos_byte - limit) * direction <= len_byte) | |
1811 | break; /* ran off the end */ | |
1812 | /* Found what might be a match. | |
1813 | Set POS_BYTE back to last (first if reverse) pos. */ | |
1814 | pos_byte -= infinity; | |
1815 | i = dirlen - direction; | |
1816 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1817 | { |
facdc750 RS |
1818 | int ch; |
1819 | unsigned char *ptr; | |
1820 | pos_byte -= direction; | |
1821 | ptr = BYTE_POS_ADDR (pos_byte); | |
1822 | /* Translate only the last byte of a character. */ | |
1823 | if (! multibyte | |
1824 | || ((ptr == tail_end_ptr | |
1825 | || CHAR_HEAD_P (ptr[1])) | |
1826 | && (CHAR_HEAD_P (ptr[0]) | |
1827 | || (translate_prev_byte == ptr[-1] | |
1828 | && (CHAR_HEAD_P (translate_prev_byte) | |
1829 | || translate_anteprev_byte == ptr[-2]))))) | |
1830 | ch = simple_translate[*ptr]; | |
1831 | else | |
1832 | ch = *ptr; | |
1833 | if (pat[i] != ch) | |
1834 | break; | |
1835 | } | |
1836 | /* Above loop has moved POS_BYTE part or all the way | |
1837 | back to the first pos (last pos if reverse). | |
1838 | Set it once again at the last (first if reverse) char. */ | |
1839 | pos_byte += dirlen - i- direction; | |
1840 | if (i + direction == 0) | |
1841 | { | |
1842 | int position; | |
1843 | pos_byte -= direction; | |
1113d9db | 1844 | |
facdc750 | 1845 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
0c8533c6 | 1846 | |
facdc750 | 1847 | set_search_regs (position, len_byte); |
ca325161 | 1848 | |
facdc750 RS |
1849 | if ((n -= direction) != 0) |
1850 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 1851 | else |
facdc750 RS |
1852 | return ((direction > 0) |
1853 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1854 | } |
facdc750 RS |
1855 | else |
1856 | pos_byte += stride_for_teases; | |
1857 | } | |
1858 | } | |
1859 | /* We have done one clump. Can we continue? */ | |
1860 | if ((lim_byte - pos_byte) * direction < 0) | |
1861 | return ((0 - n) * direction); | |
ca1d1d23 | 1862 | } |
facdc750 | 1863 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 1864 | } |
ca325161 | 1865 | |
fa8ed3e0 | 1866 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
1867 | for the overall match just found in the current buffer. |
1868 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
1869 | |
1870 | static void | |
fa8ed3e0 RS |
1871 | set_search_regs (beg_byte, nbytes) |
1872 | int beg_byte, nbytes; | |
ca325161 | 1873 | { |
a7e4cdde RS |
1874 | int i; |
1875 | ||
ca325161 RS |
1876 | /* Make sure we have registers in which to store |
1877 | the match position. */ | |
1878 | if (search_regs.num_regs == 0) | |
1879 | { | |
2d4a771a RS |
1880 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
1881 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 1882 | search_regs.num_regs = 2; |
ca325161 RS |
1883 | } |
1884 | ||
a7e4cdde RS |
1885 | /* Clear out the other registers. */ |
1886 | for (i = 1; i < search_regs.num_regs; i++) | |
1887 | { | |
1888 | search_regs.start[i] = -1; | |
1889 | search_regs.end[i] = -1; | |
1890 | } | |
1891 | ||
fa8ed3e0 RS |
1892 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
1893 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 1894 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 1895 | } |
ca1d1d23 JB |
1896 | \f |
1897 | /* Given a string of words separated by word delimiters, | |
1898 | compute a regexp that matches those exact words | |
1899 | separated by arbitrary punctuation. */ | |
1900 | ||
1901 | static Lisp_Object | |
1902 | wordify (string) | |
1903 | Lisp_Object string; | |
1904 | { | |
1905 | register unsigned char *p, *o; | |
0c8533c6 | 1906 | register int i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 1907 | Lisp_Object val; |
0c8533c6 RS |
1908 | int prev_c = 0; |
1909 | int adjust; | |
ca1d1d23 | 1910 | |
b7826503 | 1911 | CHECK_STRING (string); |
d5db4077 KR |
1912 | p = SDATA (string); |
1913 | len = SCHARS (string); | |
ca1d1d23 | 1914 | |
0c8533c6 RS |
1915 | for (i = 0, i_byte = 0; i < len; ) |
1916 | { | |
1917 | int c; | |
177c0ea7 | 1918 | |
93daa011 | 1919 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
0c8533c6 RS |
1920 | |
1921 | if (SYNTAX (c) != Sword) | |
1922 | { | |
1923 | punct_count++; | |
1924 | if (i > 0 && SYNTAX (prev_c) == Sword) | |
1925 | word_count++; | |
1926 | } | |
ca1d1d23 | 1927 | |
0c8533c6 RS |
1928 | prev_c = c; |
1929 | } | |
1930 | ||
1931 | if (SYNTAX (prev_c) == Sword) | |
1932 | word_count++; | |
1933 | if (!word_count) | |
b07b65aa | 1934 | return empty_string; |
0c8533c6 RS |
1935 | |
1936 | adjust = - punct_count + 5 * (word_count - 1) + 4; | |
8a2df937 RS |
1937 | if (STRING_MULTIBYTE (string)) |
1938 | val = make_uninit_multibyte_string (len + adjust, | |
d5db4077 | 1939 | SBYTES (string) |
8a2df937 RS |
1940 | + adjust); |
1941 | else | |
1942 | val = make_uninit_string (len + adjust); | |
ca1d1d23 | 1943 | |
d5db4077 | 1944 | o = SDATA (val); |
ca1d1d23 JB |
1945 | *o++ = '\\'; |
1946 | *o++ = 'b'; | |
1e9582d4 | 1947 | prev_c = 0; |
ca1d1d23 | 1948 | |
1e9582d4 RS |
1949 | for (i = 0, i_byte = 0; i < len; ) |
1950 | { | |
1951 | int c; | |
1952 | int i_byte_orig = i_byte; | |
177c0ea7 | 1953 | |
93daa011 | 1954 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
1e9582d4 RS |
1955 | |
1956 | if (SYNTAX (c) == Sword) | |
1957 | { | |
5d69fe10 | 1958 | bcopy (SDATA (string) + i_byte_orig, o, |
1e9582d4 RS |
1959 | i_byte - i_byte_orig); |
1960 | o += i_byte - i_byte_orig; | |
1961 | } | |
1962 | else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count) | |
1963 | { | |
1964 | *o++ = '\\'; | |
1965 | *o++ = 'W'; | |
1966 | *o++ = '\\'; | |
1967 | *o++ = 'W'; | |
1968 | *o++ = '*'; | |
1969 | } | |
1970 | ||
1971 | prev_c = c; | |
1972 | } | |
ca1d1d23 JB |
1973 | |
1974 | *o++ = '\\'; | |
1975 | *o++ = 'b'; | |
1976 | ||
1977 | return val; | |
1978 | } | |
1979 | \f | |
1980 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
8c1a1077 PJ |
1981 | "MSearch backward: ", |
1982 | doc: /* Search backward from point for STRING. | |
1983 | Set point to the beginning of the occurrence found, and return point. | |
1984 | An optional second argument bounds the search; it is a buffer position. | |
1985 | The match found must not extend before that position. | |
1986 | Optional third argument, if t, means if fail just return nil (no error). | |
1987 | If not nil and not t, position at limit of search and return nil. | |
1988 | Optional fourth argument is repeat count--search for successive occurrences. | |
1989 | ||
1990 | Search case-sensitivity is determined by the value of the variable | |
1991 | `case-fold-search', which see. | |
1992 | ||
1993 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
1994 | (string, bound, noerror, count) | |
ca1d1d23 JB |
1995 | Lisp_Object string, bound, noerror, count; |
1996 | { | |
b819a390 | 1997 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
1998 | } |
1999 | ||
6af43974 | 2000 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2001 | doc: /* Search forward from point for STRING. |
2002 | Set point to the end of the occurrence found, and return point. | |
2003 | An optional second argument bounds the search; it is a buffer position. | |
2004 | The match found must not extend after that position. nil is equivalent | |
2005 | to (point-max). | |
2006 | Optional third argument, if t, means if fail just return nil (no error). | |
2007 | If not nil and not t, move to limit of search and return nil. | |
2008 | Optional fourth argument is repeat count--search for successive occurrences. | |
2009 | ||
2010 | Search case-sensitivity is determined by the value of the variable | |
2011 | `case-fold-search', which see. | |
2012 | ||
2013 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2014 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2015 | Lisp_Object string, bound, noerror, count; |
2016 | { | |
b819a390 | 2017 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2018 | } |
2019 | ||
2020 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
8c1a1077 PJ |
2021 | "sWord search backward: ", |
2022 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2023 | Set point to the beginning of the occurrence found, and return point. | |
2024 | An optional second argument bounds the search; it is a buffer position. | |
2025 | The match found must not extend before that position. | |
2026 | Optional third argument, if t, means if fail just return nil (no error). | |
2027 | If not nil and not t, move to limit of search and return nil. | |
2028 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2029 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2030 | Lisp_Object string, bound, noerror, count; |
2031 | { | |
b819a390 | 2032 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2033 | } |
2034 | ||
2035 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
8c1a1077 PJ |
2036 | "sWord search: ", |
2037 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2038 | Set point to the end of the occurrence found, and return point. | |
2039 | An optional second argument bounds the search; it is a buffer position. | |
2040 | The match found must not extend after that position. | |
2041 | Optional third argument, if t, means if fail just return nil (no error). | |
2042 | If not nil and not t, move to limit of search and return nil. | |
2043 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2044 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2045 | Lisp_Object string, bound, noerror, count; |
2046 | { | |
b819a390 | 2047 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
2048 | } |
2049 | ||
2050 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
8c1a1077 PJ |
2051 | "sRE search backward: ", |
2052 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2053 | Set point to the beginning of the match, and return point. | |
2054 | The match found is the one starting last in the buffer | |
2055 | and yet ending before the origin of the search. | |
2056 | An optional second argument bounds the search; it is a buffer position. | |
2057 | The match found must start at or after that position. | |
2058 | Optional third argument, if t, means if fail just return nil (no error). | |
2059 | If not nil and not t, move to limit of search and return nil. | |
2060 | Optional fourth argument is repeat count--search for successive occurrences. | |
2061 | See also the functions `match-beginning', `match-end', `match-string', | |
2062 | and `replace-match'. */) | |
2063 | (regexp, bound, noerror, count) | |
19c0a730 | 2064 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2065 | { |
b819a390 | 2066 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2067 | } |
2068 | ||
2069 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2070 | "sRE search: ", |
2071 | doc: /* Search forward from point for regular expression REGEXP. | |
2072 | Set point to the end of the occurrence found, and return point. | |
2073 | An optional second argument bounds the search; it is a buffer position. | |
2074 | The match found must not extend after that position. | |
2075 | Optional third argument, if t, means if fail just return nil (no error). | |
2076 | If not nil and not t, move to limit of search and return nil. | |
2077 | Optional fourth argument is repeat count--search for successive occurrences. | |
2078 | See also the functions `match-beginning', `match-end', `match-string', | |
2079 | and `replace-match'. */) | |
2080 | (regexp, bound, noerror, count) | |
19c0a730 | 2081 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2082 | { |
b819a390 RS |
2083 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2084 | } | |
2085 | ||
2086 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2087 | "sPosix search backward: ", |
2088 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2089 | Find the longest match in accord with Posix regular expression rules. | |
2090 | Set point to the beginning of the match, and return point. | |
2091 | The match found is the one starting last in the buffer | |
2092 | and yet ending before the origin of the search. | |
2093 | An optional second argument bounds the search; it is a buffer position. | |
2094 | The match found must start at or after that position. | |
2095 | Optional third argument, if t, means if fail just return nil (no error). | |
2096 | If not nil and not t, move to limit of search and return nil. | |
2097 | Optional fourth argument is repeat count--search for successive occurrences. | |
2098 | See also the functions `match-beginning', `match-end', `match-string', | |
2099 | and `replace-match'. */) | |
2100 | (regexp, bound, noerror, count) | |
b819a390 RS |
2101 | Lisp_Object regexp, bound, noerror, count; |
2102 | { | |
2103 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2104 | } | |
2105 | ||
2106 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2107 | "sPosix search: ", |
2108 | doc: /* Search forward from point for regular expression REGEXP. | |
2109 | Find the longest match in accord with Posix regular expression rules. | |
2110 | Set point to the end of the occurrence found, and return point. | |
2111 | An optional second argument bounds the search; it is a buffer position. | |
2112 | The match found must not extend after that position. | |
2113 | Optional third argument, if t, means if fail just return nil (no error). | |
2114 | If not nil and not t, move to limit of search and return nil. | |
2115 | Optional fourth argument is repeat count--search for successive occurrences. | |
2116 | See also the functions `match-beginning', `match-end', `match-string', | |
2117 | and `replace-match'. */) | |
2118 | (regexp, bound, noerror, count) | |
b819a390 RS |
2119 | Lisp_Object regexp, bound, noerror, count; |
2120 | { | |
2121 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2122 | } |
2123 | \f | |
d7a5ad5f | 2124 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2125 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2126 | Leave point at the end of the replacement text. |
2127 | ||
8c1a1077 PJ |
2128 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. |
2129 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2130 | based on the replaced text. | |
2131 | If the replaced text has only capital letters | |
2132 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
4dd0c271 RS |
2133 | Otherwise if all words are capitalized in the replaced text, |
2134 | capitalize each word in NEWTEXT. | |
8c1a1077 PJ |
2135 | |
2136 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2137 | Otherwise treat `\\' as special: | |
2138 | `\\&' in NEWTEXT means substitute original matched text. | |
2139 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2140 | If Nth parens didn't match, substitute nothing. | |
2141 | `\\\\' means insert one `\\'. | |
4dd0c271 RS |
2142 | Case conversion does not apply to these substitutions. |
2143 | ||
8c1a1077 | 2144 | FIXEDCASE and LITERAL are optional arguments. |
8c1a1077 PJ |
2145 | |
2146 | The optional fourth argument STRING can be a string to modify. | |
2147 | This is meaningful when the previous match was done against STRING, | |
2148 | using `string-match'. When used this way, `replace-match' | |
2149 | creates and returns a new string made by copying STRING and replacing | |
2150 | the part of STRING that was matched. | |
2151 | ||
2152 | The optional fifth argument SUBEXP specifies a subexpression; | |
2153 | it says to replace just that subexpression with NEWTEXT, | |
2154 | rather than replacing the entire matched text. | |
2155 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2156 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2157 | NEWTEXT in place of subexp N. | |
2158 | This is useful only after a regular expression search or match, | |
2159 | since only regular expressions have distinguished subexpressions. */) | |
2160 | (newtext, fixedcase, literal, string, subexp) | |
d7a5ad5f | 2161 | Lisp_Object newtext, fixedcase, literal, string, subexp; |
ca1d1d23 JB |
2162 | { |
2163 | enum { nochange, all_caps, cap_initial } case_action; | |
ac3b28b1 | 2164 | register int pos, pos_byte; |
ca1d1d23 | 2165 | int some_multiletter_word; |
97832bd0 | 2166 | int some_lowercase; |
73dc8771 | 2167 | int some_uppercase; |
208767c3 | 2168 | int some_nonuppercase_initial; |
ca1d1d23 | 2169 | register int c, prevc; |
d7a5ad5f | 2170 | int sub; |
3e18eecf | 2171 | int opoint, newpoint; |
ca1d1d23 | 2172 | |
b7826503 | 2173 | CHECK_STRING (newtext); |
ca1d1d23 | 2174 | |
080c45fd | 2175 | if (! NILP (string)) |
b7826503 | 2176 | CHECK_STRING (string); |
080c45fd | 2177 | |
ca1d1d23 JB |
2178 | case_action = nochange; /* We tried an initialization */ |
2179 | /* but some C compilers blew it */ | |
4746118a JB |
2180 | |
2181 | if (search_regs.num_regs <= 0) | |
2182 | error ("replace-match called before any match found"); | |
2183 | ||
d7a5ad5f RS |
2184 | if (NILP (subexp)) |
2185 | sub = 0; | |
2186 | else | |
2187 | { | |
b7826503 | 2188 | CHECK_NUMBER (subexp); |
d7a5ad5f RS |
2189 | sub = XINT (subexp); |
2190 | if (sub < 0 || sub >= search_regs.num_regs) | |
2191 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2192 | } | |
2193 | ||
080c45fd RS |
2194 | if (NILP (string)) |
2195 | { | |
d7a5ad5f RS |
2196 | if (search_regs.start[sub] < BEGV |
2197 | || search_regs.start[sub] > search_regs.end[sub] | |
2198 | || search_regs.end[sub] > ZV) | |
2199 | args_out_of_range (make_number (search_regs.start[sub]), | |
2200 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2201 | } |
2202 | else | |
2203 | { | |
d7a5ad5f RS |
2204 | if (search_regs.start[sub] < 0 |
2205 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2206 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2207 | args_out_of_range (make_number (search_regs.start[sub]), |
2208 | make_number (search_regs.end[sub])); | |
080c45fd | 2209 | } |
ca1d1d23 JB |
2210 | |
2211 | if (NILP (fixedcase)) | |
2212 | { | |
2213 | /* Decide how to casify by examining the matched text. */ | |
ac3b28b1 | 2214 | int last; |
ca1d1d23 | 2215 | |
ac3b28b1 KH |
2216 | pos = search_regs.start[sub]; |
2217 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2218 | |
2219 | if (NILP (string)) | |
ac3b28b1 | 2220 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2221 | else |
ac3b28b1 | 2222 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2223 | |
ca1d1d23 JB |
2224 | prevc = '\n'; |
2225 | case_action = all_caps; | |
2226 | ||
2227 | /* some_multiletter_word is set nonzero if any original word | |
2228 | is more than one letter long. */ | |
2229 | some_multiletter_word = 0; | |
97832bd0 | 2230 | some_lowercase = 0; |
208767c3 | 2231 | some_nonuppercase_initial = 0; |
73dc8771 | 2232 | some_uppercase = 0; |
ca1d1d23 | 2233 | |
ac3b28b1 | 2234 | while (pos < last) |
ca1d1d23 | 2235 | { |
080c45fd | 2236 | if (NILP (string)) |
ac3b28b1 | 2237 | { |
93daa011 | 2238 | c = FETCH_CHAR_AS_MULTIBYTE (pos_byte); |
ac3b28b1 KH |
2239 | INC_BOTH (pos, pos_byte); |
2240 | } | |
080c45fd | 2241 | else |
93daa011 | 2242 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2243 | |
ca1d1d23 JB |
2244 | if (LOWERCASEP (c)) |
2245 | { | |
2246 | /* Cannot be all caps if any original char is lower case */ | |
2247 | ||
97832bd0 | 2248 | some_lowercase = 1; |
ca1d1d23 | 2249 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2250 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2251 | else |
2252 | some_multiletter_word = 1; | |
2253 | } | |
2254 | else if (!NOCASEP (c)) | |
2255 | { | |
73dc8771 | 2256 | some_uppercase = 1; |
97832bd0 | 2257 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2258 | ; |
97832bd0 | 2259 | else |
ca1d1d23 JB |
2260 | some_multiletter_word = 1; |
2261 | } | |
208767c3 RS |
2262 | else |
2263 | { | |
2264 | /* If the initial is a caseless word constituent, | |
2265 | treat that like a lowercase initial. */ | |
2266 | if (SYNTAX (prevc) != Sword) | |
2267 | some_nonuppercase_initial = 1; | |
2268 | } | |
ca1d1d23 JB |
2269 | |
2270 | prevc = c; | |
2271 | } | |
2272 | ||
97832bd0 RS |
2273 | /* Convert to all caps if the old text is all caps |
2274 | and has at least one multiletter word. */ | |
2275 | if (! some_lowercase && some_multiletter_word) | |
2276 | case_action = all_caps; | |
c4d460ce | 2277 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2278 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2279 | case_action = cap_initial; |
208767c3 | 2280 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2281 | /* Should x -> yz, operating on X, give Yz or YZ? |
2282 | We'll assume the latter. */ | |
2283 | case_action = all_caps; | |
97832bd0 RS |
2284 | else |
2285 | case_action = nochange; | |
ca1d1d23 JB |
2286 | } |
2287 | ||
080c45fd RS |
2288 | /* Do replacement in a string. */ |
2289 | if (!NILP (string)) | |
2290 | { | |
2291 | Lisp_Object before, after; | |
2292 | ||
2293 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2294 | make_number (search_regs.start[sub])); |
2295 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2296 | |
636a5e28 RS |
2297 | /* Substitute parts of the match into NEWTEXT |
2298 | if desired. */ | |
080c45fd RS |
2299 | if (NILP (literal)) |
2300 | { | |
d131e79c RS |
2301 | int lastpos = 0; |
2302 | int lastpos_byte = 0; | |
080c45fd RS |
2303 | /* We build up the substituted string in ACCUM. */ |
2304 | Lisp_Object accum; | |
2305 | Lisp_Object middle; | |
d5db4077 | 2306 | int length = SBYTES (newtext); |
080c45fd RS |
2307 | |
2308 | accum = Qnil; | |
2309 | ||
ac3b28b1 | 2310 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd RS |
2311 | { |
2312 | int substart = -1; | |
6bbd7a29 | 2313 | int subend = 0; |
1e79ec24 | 2314 | int delbackslash = 0; |
080c45fd | 2315 | |
0c8533c6 RS |
2316 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2317 | ||
080c45fd RS |
2318 | if (c == '\\') |
2319 | { | |
0c8533c6 | 2320 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2321 | |
080c45fd RS |
2322 | if (c == '&') |
2323 | { | |
d7a5ad5f RS |
2324 | substart = search_regs.start[sub]; |
2325 | subend = search_regs.end[sub]; | |
080c45fd | 2326 | } |
76fc1ea2 | 2327 | else if (c >= '1' && c <= '9') |
080c45fd | 2328 | { |
76fc1ea2 KH |
2329 | if (search_regs.start[c - '0'] >= 0 |
2330 | && c <= search_regs.num_regs + '0') | |
080c45fd RS |
2331 | { |
2332 | substart = search_regs.start[c - '0']; | |
2333 | subend = search_regs.end[c - '0']; | |
2334 | } | |
76fc1ea2 KH |
2335 | else |
2336 | { | |
2337 | /* If that subexp did not match, | |
2338 | replace \\N with nothing. */ | |
2339 | substart = 0; | |
2340 | subend = 0; | |
2341 | } | |
080c45fd | 2342 | } |
1e79ec24 KH |
2343 | else if (c == '\\') |
2344 | delbackslash = 1; | |
636a5e28 RS |
2345 | else |
2346 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2347 | } |
2348 | if (substart >= 0) | |
2349 | { | |
d131e79c RS |
2350 | if (pos - 2 != lastpos) |
2351 | middle = substring_both (newtext, lastpos, | |
2352 | lastpos_byte, | |
2353 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2354 | else |
2355 | middle = Qnil; | |
2356 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2357 | Fsubstring (string, |
2358 | make_number (substart), | |
080c45fd RS |
2359 | make_number (subend))); |
2360 | lastpos = pos; | |
0c8533c6 | 2361 | lastpos_byte = pos_byte; |
080c45fd | 2362 | } |
1e79ec24 KH |
2363 | else if (delbackslash) |
2364 | { | |
d131e79c RS |
2365 | middle = substring_both (newtext, lastpos, |
2366 | lastpos_byte, | |
2367 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2368 | |
1e79ec24 KH |
2369 | accum = concat2 (accum, middle); |
2370 | lastpos = pos; | |
0c8533c6 | 2371 | lastpos_byte = pos_byte; |
1e79ec24 | 2372 | } |
080c45fd RS |
2373 | } |
2374 | ||
d131e79c RS |
2375 | if (pos != lastpos) |
2376 | middle = substring_both (newtext, lastpos, | |
2377 | lastpos_byte, | |
0c8533c6 | 2378 | pos, pos_byte); |
080c45fd RS |
2379 | else |
2380 | middle = Qnil; | |
2381 | ||
2382 | newtext = concat2 (accum, middle); | |
2383 | } | |
2384 | ||
636a5e28 | 2385 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2386 | if (case_action == all_caps) |
2387 | newtext = Fupcase (newtext); | |
2388 | else if (case_action == cap_initial) | |
2b2eead9 | 2389 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2390 | |
2391 | return concat3 (before, newtext, after); | |
2392 | } | |
2393 | ||
09c4719e | 2394 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2395 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2396 | opoint = PT - ZV; |
9160906f RS |
2397 | else if (PT > search_regs.start[sub]) |
2398 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2399 | else |
2400 | opoint = PT; | |
2401 | ||
886ed6ec RS |
2402 | /* If we want non-literal replacement, |
2403 | perform substitution on the replacement string. */ | |
2404 | if (NILP (literal)) | |
ca1d1d23 | 2405 | { |
d5db4077 | 2406 | int length = SBYTES (newtext); |
68e69fbd RS |
2407 | unsigned char *substed; |
2408 | int substed_alloc_size, substed_len; | |
3bc25e52 KH |
2409 | int buf_multibyte = !NILP (current_buffer->enable_multibyte_characters); |
2410 | int str_multibyte = STRING_MULTIBYTE (newtext); | |
2411 | Lisp_Object rev_tbl; | |
886ed6ec | 2412 | int really_changed = 0; |
3bc25e52 | 2413 | |
8f924df7 | 2414 | rev_tbl = Qnil; |
ac3b28b1 | 2415 | |
68e69fbd RS |
2416 | substed_alloc_size = length * 2 + 100; |
2417 | substed = (unsigned char *) xmalloc (substed_alloc_size + 1); | |
2418 | substed_len = 0; | |
2419 | ||
3bc25e52 KH |
2420 | /* Go thru NEWTEXT, producing the actual text to insert in |
2421 | SUBSTED while adjusting multibyteness to that of the current | |
2422 | buffer. */ | |
ca1d1d23 | 2423 | |
ac3b28b1 | 2424 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2425 | { |
68e69fbd | 2426 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
f8ce8a0d GM |
2427 | unsigned char *add_stuff = NULL; |
2428 | int add_len = 0; | |
68e69fbd | 2429 | int idx = -1; |
9a76659d | 2430 | |
3bc25e52 KH |
2431 | if (str_multibyte) |
2432 | { | |
eb99a8dd | 2433 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 KH |
2434 | if (!buf_multibyte) |
2435 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2436 | } | |
2437 | else | |
2438 | { | |
2439 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2440 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2441 | if (buf_multibyte) |
2442 | c = unibyte_char_to_multibyte (c); | |
2443 | } | |
ac3b28b1 | 2444 | |
68e69fbd RS |
2445 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2446 | or set IDX to a match index, which means put that part | |
2447 | of the buffer text into SUBSTED. */ | |
2448 | ||
ca1d1d23 JB |
2449 | if (c == '\\') |
2450 | { | |
886ed6ec RS |
2451 | really_changed = 1; |
2452 | ||
3bc25e52 KH |
2453 | if (str_multibyte) |
2454 | { | |
eb99a8dd KH |
2455 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2456 | pos, pos_byte); | |
071ce769 | 2457 | if (!buf_multibyte && !ASCII_CHAR_P (c)) |
3bc25e52 KH |
2458 | c = multibyte_char_to_unibyte (c, rev_tbl); |
2459 | } | |
2460 | else | |
2461 | { | |
d5db4077 | 2462 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2463 | if (buf_multibyte) |
2464 | c = unibyte_char_to_multibyte (c); | |
2465 | } | |
2466 | ||
ca1d1d23 | 2467 | if (c == '&') |
68e69fbd | 2468 | idx = sub; |
78445046 | 2469 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2470 | { |
2471 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2472 | idx = c - '0'; |
ca1d1d23 | 2473 | } |
636a5e28 | 2474 | else if (c == '\\') |
68e69fbd | 2475 | add_len = 1, add_stuff = "\\"; |
636a5e28 | 2476 | else |
3bc25e52 KH |
2477 | { |
2478 | xfree (substed); | |
2479 | error ("Invalid use of `\\' in replacement text"); | |
2480 | } | |
ca1d1d23 JB |
2481 | } |
2482 | else | |
68e69fbd RS |
2483 | { |
2484 | add_len = CHAR_STRING (c, str); | |
2485 | add_stuff = str; | |
2486 | } | |
2487 | ||
2488 | /* If we want to copy part of a previous match, | |
2489 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2490 | if (idx >= 0) | |
2491 | { | |
2492 | int begbyte = CHAR_TO_BYTE (search_regs.start[idx]); | |
2493 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; | |
2494 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2495 | move_gap (search_regs.start[idx]); | |
2496 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2497 | } | |
2498 | ||
2499 | /* Now the stuff we want to add to SUBSTED | |
2500 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2501 | ||
2502 | /* Make sure SUBSTED is big enough. */ | |
2503 | if (substed_len + add_len >= substed_alloc_size) | |
2504 | { | |
2505 | substed_alloc_size = substed_len + add_len + 500; | |
2506 | substed = (unsigned char *) xrealloc (substed, | |
2507 | substed_alloc_size + 1); | |
2508 | } | |
2509 | ||
2510 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2511 | if (add_stuff) |
2512 | { | |
2513 | bcopy (add_stuff, substed + substed_len, add_len); | |
2514 | substed_len += add_len; | |
2515 | } | |
ca1d1d23 | 2516 | } |
68e69fbd | 2517 | |
886ed6ec | 2518 | if (really_changed) |
76fc1ea2 KH |
2519 | { |
2520 | if (buf_multibyte) | |
2521 | { | |
2522 | int nchars = multibyte_chars_in_text (substed, substed_len); | |
68e69fbd | 2523 | |
76fc1ea2 KH |
2524 | newtext = make_multibyte_string (substed, nchars, substed_len); |
2525 | } | |
2526 | else | |
2527 | newtext = make_unibyte_string (substed, substed_len); | |
2528 | } | |
68e69fbd | 2529 | xfree (substed); |
ca1d1d23 JB |
2530 | } |
2531 | ||
886ed6ec RS |
2532 | /* Replace the old text with the new in the cleanest possible way. */ |
2533 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2534 | newtext, 1, 0, 1); | |
d5db4077 | 2535 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2536 | |
2537 | if (case_action == all_caps) | |
886ed6ec RS |
2538 | Fupcase_region (make_number (search_regs.start[sub]), |
2539 | make_number (newpoint)); | |
ca1d1d23 | 2540 | else if (case_action == cap_initial) |
886ed6ec RS |
2541 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2542 | make_number (newpoint)); | |
3e18eecf | 2543 | |
98e942e0 RS |
2544 | /* Adjust search data for this change. */ |
2545 | { | |
5b88a2c5 | 2546 | int oldend = search_regs.end[sub]; |
41c01205 | 2547 | int oldstart = search_regs.start[sub]; |
98e942e0 RS |
2548 | int change = newpoint - search_regs.end[sub]; |
2549 | int i; | |
2550 | ||
2551 | for (i = 0; i < search_regs.num_regs; i++) | |
2552 | { | |
41c01205 | 2553 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2554 | search_regs.start[i] += change; |
41c01205 DK |
2555 | else if (search_regs.start[i] > oldstart) |
2556 | search_regs.start[i] = oldstart; | |
2557 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2558 | search_regs.end[i] += change; |
41c01205 DK |
2559 | else if (search_regs.end[i] > oldstart) |
2560 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2561 | } |
2562 | } | |
2563 | ||
b0eba991 | 2564 | /* Put point back where it was in the text. */ |
8d808a65 | 2565 | if (opoint <= 0) |
fa8ed3e0 | 2566 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2567 | else |
fa8ed3e0 | 2568 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2569 | |
2570 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2571 | move_if_not_intangible (newpoint); |
177c0ea7 | 2572 | |
ca1d1d23 JB |
2573 | return Qnil; |
2574 | } | |
2575 | \f | |
2576 | static Lisp_Object | |
2577 | match_limit (num, beginningp) | |
2578 | Lisp_Object num; | |
2579 | int beginningp; | |
2580 | { | |
2581 | register int n; | |
2582 | ||
b7826503 | 2583 | CHECK_NUMBER (num); |
ca1d1d23 | 2584 | n = XINT (num); |
f90a5bf5 | 2585 | if (n < 0) |
bd2cbd56 | 2586 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2587 | if (search_regs.num_regs <= 0) |
2588 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2589 | if (n >= search_regs.num_regs |
4746118a | 2590 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2591 | return Qnil; |
2592 | return (make_number ((beginningp) ? search_regs.start[n] | |
2593 | : search_regs.end[n])); | |
2594 | } | |
2595 | ||
2596 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
8c1a1077 PJ |
2597 | doc: /* Return position of start of text matched by last search. |
2598 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2599 | regexp. | |
2600 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2601 | SUBEXP pairs. | |
2602 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2603 | (subexp) | |
5806161b | 2604 | Lisp_Object subexp; |
ca1d1d23 | 2605 | { |
5806161b | 2606 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2607 | } |
2608 | ||
2609 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
8c1a1077 PJ |
2610 | doc: /* Return position of end of text matched by last search. |
2611 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2612 | regexp. | |
2613 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2614 | SUBEXP pairs. | |
2615 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2616 | (subexp) | |
5806161b | 2617 | Lisp_Object subexp; |
ca1d1d23 | 2618 | { |
5806161b | 2619 | return match_limit (subexp, 0); |
177c0ea7 | 2620 | } |
ca1d1d23 | 2621 | |
56256c2a | 2622 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 2, 0, |
8c1a1077 PJ |
2623 | doc: /* Return a list containing all info on what the last search matched. |
2624 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2625 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2626 | if the last match was on a buffer; integers or nil if a string was matched. | |
2627 | Use `store-match-data' to reinstate the data in this list. | |
2628 | ||
41c01205 DK |
2629 | If INTEGERS (the optional first argument) is non-nil, always use |
2630 | integers \(rather than markers) to represent buffer positions. In | |
2631 | this case, and if the last match was in a buffer, the buffer will get | |
2632 | stored as one additional element at the end of the list. | |
2633 | ||
8c1a1077 | 2634 | If REUSE is a list, reuse it as part of the value. If REUSE is long enough |
140a6b7e KS |
2635 | to hold all the values, and if INTEGERS is non-nil, no consing is done. |
2636 | ||
2637 | Return value is undefined if the last search failed. */) | |
8c1a1077 | 2638 | (integers, reuse) |
56256c2a | 2639 | Lisp_Object integers, reuse; |
ca1d1d23 | 2640 | { |
56256c2a | 2641 | Lisp_Object tail, prev; |
4746118a | 2642 | Lisp_Object *data; |
ca1d1d23 JB |
2643 | int i, len; |
2644 | ||
daa37602 | 2645 | if (NILP (last_thing_searched)) |
c36bcf1b | 2646 | return Qnil; |
daa37602 | 2647 | |
6bbd7a29 GM |
2648 | prev = Qnil; |
2649 | ||
41c01205 | 2650 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs + 1) |
4746118a JB |
2651 | * sizeof (Lisp_Object)); |
2652 | ||
41c01205 | 2653 | len = 0; |
4746118a | 2654 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2655 | { |
2656 | int start = search_regs.start[i]; | |
2657 | if (start >= 0) | |
2658 | { | |
56256c2a RS |
2659 | if (EQ (last_thing_searched, Qt) |
2660 | || ! NILP (integers)) | |
ca1d1d23 | 2661 | { |
c235cce7 KH |
2662 | XSETFASTINT (data[2 * i], start); |
2663 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2664 | } |
0ed62dc7 | 2665 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2666 | { |
2667 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2668 | Fset_marker (data[2 * i], |
2669 | make_number (start), | |
2670 | last_thing_searched); | |
ca1d1d23 JB |
2671 | data[2 * i + 1] = Fmake_marker (); |
2672 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2673 | make_number (search_regs.end[i]), |
daa37602 | 2674 | last_thing_searched); |
ca1d1d23 | 2675 | } |
daa37602 JB |
2676 | else |
2677 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2678 | abort (); | |
2679 | ||
41c01205 | 2680 | len = 2*(i+1); |
ca1d1d23 JB |
2681 | } |
2682 | else | |
2683 | data[2 * i] = data [2 * i + 1] = Qnil; | |
2684 | } | |
56256c2a | 2685 | |
bd2cbd56 | 2686 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2687 | { |
bd2cbd56 | 2688 | data[len] = last_thing_searched; |
41c01205 DK |
2689 | len++; |
2690 | } | |
2691 | ||
56256c2a RS |
2692 | /* If REUSE is not usable, cons up the values and return them. */ |
2693 | if (! CONSP (reuse)) | |
41c01205 | 2694 | return Flist (len, data); |
56256c2a RS |
2695 | |
2696 | /* If REUSE is a list, store as many value elements as will fit | |
2697 | into the elements of REUSE. */ | |
2698 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2699 | i++, tail = XCDR (tail)) |
56256c2a | 2700 | { |
41c01205 | 2701 | if (i < len) |
f3fbd155 | 2702 | XSETCAR (tail, data[i]); |
56256c2a | 2703 | else |
f3fbd155 | 2704 | XSETCAR (tail, Qnil); |
56256c2a RS |
2705 | prev = tail; |
2706 | } | |
2707 | ||
2708 | /* If we couldn't fit all value elements into REUSE, | |
2709 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2710 | if (i < len) |
2711 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2712 | |
2713 | return reuse; | |
ca1d1d23 JB |
2714 | } |
2715 | ||
2716 | ||
3f1c005b | 2717 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 1, 0, |
8c1a1077 PJ |
2718 | doc: /* Set internal data on last search match from elements of LIST. |
2719 | LIST should have been created by calling `match-data' previously. */) | |
2720 | (list) | |
ca1d1d23 JB |
2721 | register Lisp_Object list; |
2722 | { | |
2723 | register int i; | |
2724 | register Lisp_Object marker; | |
2725 | ||
7074fde6 FP |
2726 | if (running_asynch_code) |
2727 | save_search_regs (); | |
2728 | ||
ca1d1d23 | 2729 | if (!CONSP (list) && !NILP (list)) |
b37902c8 | 2730 | list = wrong_type_argument (Qconsp, list); |
ca1d1d23 | 2731 | |
41c01205 DK |
2732 | /* Unless we find a marker with a buffer or an explicit buffer |
2733 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
2734 | last_thing_searched = Qt; |
2735 | ||
4746118a JB |
2736 | /* Allocate registers if they don't already exist. */ |
2737 | { | |
d084e942 | 2738 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2739 | |
2740 | if (length > search_regs.num_regs) | |
2741 | { | |
1113d9db JB |
2742 | if (search_regs.num_regs == 0) |
2743 | { | |
2744 | search_regs.start | |
2745 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2746 | search_regs.end | |
2747 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2748 | } | |
4746118a | 2749 | else |
1113d9db JB |
2750 | { |
2751 | search_regs.start | |
2752 | = (regoff_t *) xrealloc (search_regs.start, | |
2753 | length * sizeof (regoff_t)); | |
2754 | search_regs.end | |
2755 | = (regoff_t *) xrealloc (search_regs.end, | |
2756 | length * sizeof (regoff_t)); | |
2757 | } | |
4746118a | 2758 | |
e62371e9 KH |
2759 | for (i = search_regs.num_regs; i < length; i++) |
2760 | search_regs.start[i] = -1; | |
2761 | ||
487282dc | 2762 | search_regs.num_regs = length; |
4746118a | 2763 | } |
ca1d1d23 | 2764 | |
c3762cbd | 2765 | for (i = 0;; i++) |
41c01205 DK |
2766 | { |
2767 | marker = Fcar (list); | |
bd2cbd56 | 2768 | if (BUFFERP (marker)) |
c3762cbd | 2769 | { |
bd2cbd56 | 2770 | last_thing_searched = marker; |
c3762cbd DK |
2771 | break; |
2772 | } | |
2773 | if (i >= length) | |
2774 | break; | |
41c01205 DK |
2775 | if (NILP (marker)) |
2776 | { | |
2777 | search_regs.start[i] = -1; | |
2778 | list = Fcdr (list); | |
2779 | } | |
2780 | else | |
2781 | { | |
2782 | int from; | |
2783 | ||
2784 | if (MARKERP (marker)) | |
2785 | { | |
2786 | if (XMARKER (marker)->buffer == 0) | |
2787 | XSETFASTINT (marker, 0); | |
2788 | else | |
2789 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
2790 | } | |
2791 | ||
2792 | CHECK_NUMBER_COERCE_MARKER (marker); | |
2793 | from = XINT (marker); | |
2794 | list = Fcdr (list); | |
2795 | ||
2796 | marker = Fcar (list); | |
2797 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) | |
2798 | XSETFASTINT (marker, 0); | |
2799 | ||
2800 | CHECK_NUMBER_COERCE_MARKER (marker); | |
2801 | search_regs.start[i] = from; | |
2802 | search_regs.end[i] = XINT (marker); | |
2803 | } | |
2804 | list = Fcdr (list); | |
2805 | } | |
ca1d1d23 | 2806 | |
41c01205 DK |
2807 | for (; i < search_regs.num_regs; i++) |
2808 | search_regs.start[i] = -1; | |
2809 | } | |
ca1d1d23 | 2810 | |
177c0ea7 | 2811 | return Qnil; |
ca1d1d23 JB |
2812 | } |
2813 | ||
7074fde6 FP |
2814 | /* If non-zero the match data have been saved in saved_search_regs |
2815 | during the execution of a sentinel or filter. */ | |
75ebf74b | 2816 | static int search_regs_saved; |
7074fde6 | 2817 | static struct re_registers saved_search_regs; |
41c01205 | 2818 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
2819 | |
2820 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
2821 | if asynchronous code (filter or sentinel) is running. */ | |
2822 | static void | |
2823 | save_search_regs () | |
2824 | { | |
2825 | if (!search_regs_saved) | |
2826 | { | |
2827 | saved_search_regs.num_regs = search_regs.num_regs; | |
2828 | saved_search_regs.start = search_regs.start; | |
2829 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
2830 | saved_last_thing_searched = last_thing_searched; |
2831 | last_thing_searched = Qnil; | |
7074fde6 | 2832 | search_regs.num_regs = 0; |
2d4a771a RS |
2833 | search_regs.start = 0; |
2834 | search_regs.end = 0; | |
7074fde6 FP |
2835 | |
2836 | search_regs_saved = 1; | |
2837 | } | |
2838 | } | |
2839 | ||
2840 | /* Called upon exit from filters and sentinels. */ | |
2841 | void | |
2842 | restore_match_data () | |
2843 | { | |
2844 | if (search_regs_saved) | |
2845 | { | |
2846 | if (search_regs.num_regs > 0) | |
2847 | { | |
2848 | xfree (search_regs.start); | |
2849 | xfree (search_regs.end); | |
2850 | } | |
2851 | search_regs.num_regs = saved_search_regs.num_regs; | |
2852 | search_regs.start = saved_search_regs.start; | |
2853 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
2854 | last_thing_searched = saved_last_thing_searched; |
2855 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
2856 | search_regs_saved = 0; |
2857 | } | |
2858 | } | |
2859 | ||
ca1d1d23 JB |
2860 | /* Quote a string to inactivate reg-expr chars */ |
2861 | ||
2862 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 PJ |
2863 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
2864 | (string) | |
5806161b | 2865 | Lisp_Object string; |
ca1d1d23 JB |
2866 | { |
2867 | register unsigned char *in, *out, *end; | |
2868 | register unsigned char *temp; | |
0c8533c6 | 2869 | int backslashes_added = 0; |
ca1d1d23 | 2870 | |
b7826503 | 2871 | CHECK_STRING (string); |
ca1d1d23 | 2872 | |
d5db4077 | 2873 | temp = (unsigned char *) alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
2874 | |
2875 | /* Now copy the data into the new string, inserting escapes. */ | |
2876 | ||
d5db4077 KR |
2877 | in = SDATA (string); |
2878 | end = in + SBYTES (string); | |
177c0ea7 | 2879 | out = temp; |
ca1d1d23 JB |
2880 | |
2881 | for (; in != end; in++) | |
2882 | { | |
2883 | if (*in == '[' || *in == ']' | |
2884 | || *in == '*' || *in == '.' || *in == '\\' | |
2885 | || *in == '?' || *in == '+' | |
2886 | || *in == '^' || *in == '$') | |
0c8533c6 | 2887 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
2888 | *out++ = *in; |
2889 | } | |
2890 | ||
3f8100f1 | 2891 | return make_specified_string (temp, |
d5db4077 | 2892 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
2893 | out - temp, |
2894 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 2895 | } |
177c0ea7 | 2896 | \f |
dfcf069d | 2897 | void |
ca1d1d23 JB |
2898 | syms_of_search () |
2899 | { | |
2900 | register int i; | |
2901 | ||
487282dc KH |
2902 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
2903 | { | |
2904 | searchbufs[i].buf.allocated = 100; | |
b23c0a83 | 2905 | searchbufs[i].buf.buffer = (unsigned char *) xmalloc (100); |
487282dc KH |
2906 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
2907 | searchbufs[i].regexp = Qnil; | |
2908 | staticpro (&searchbufs[i].regexp); | |
2909 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); | |
2910 | } | |
2911 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
2912 | |
2913 | Qsearch_failed = intern ("search-failed"); | |
2914 | staticpro (&Qsearch_failed); | |
2915 | Qinvalid_regexp = intern ("invalid-regexp"); | |
2916 | staticpro (&Qinvalid_regexp); | |
2917 | ||
2918 | Fput (Qsearch_failed, Qerror_conditions, | |
2919 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
2920 | Fput (Qsearch_failed, Qerror_message, | |
2921 | build_string ("Search failed")); | |
2922 | ||
2923 | Fput (Qinvalid_regexp, Qerror_conditions, | |
2924 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
2925 | Fput (Qinvalid_regexp, Qerror_message, | |
2926 | build_string ("Invalid regexp")); | |
2927 | ||
daa37602 JB |
2928 | last_thing_searched = Qnil; |
2929 | staticpro (&last_thing_searched); | |
2930 | ||
0f6af254 DK |
2931 | saved_last_thing_searched = Qnil; |
2932 | staticpro (&saved_last_thing_searched); | |
2933 | ||
ca1d1d23 | 2934 | defsubr (&Slooking_at); |
b819a390 RS |
2935 | defsubr (&Sposix_looking_at); |
2936 | defsubr (&Sstring_match); | |
2937 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
2938 | defsubr (&Ssearch_forward); |
2939 | defsubr (&Ssearch_backward); | |
2940 | defsubr (&Sword_search_forward); | |
2941 | defsubr (&Sword_search_backward); | |
2942 | defsubr (&Sre_search_forward); | |
2943 | defsubr (&Sre_search_backward); | |
b819a390 RS |
2944 | defsubr (&Sposix_search_forward); |
2945 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
2946 | defsubr (&Sreplace_match); |
2947 | defsubr (&Smatch_beginning); | |
2948 | defsubr (&Smatch_end); | |
2949 | defsubr (&Smatch_data); | |
3f1c005b | 2950 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
2951 | defsubr (&Sregexp_quote); |
2952 | } | |
76fc1ea2 KH |
2953 | |
2954 | /* arch-tag: a6059d79-0552-4f14-a2cb-d379a4e3c78f | |
2955 | (do not change this comment) */ |