Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
429ab54e | 2 | Copyright (C) 1985, 1986, 1987, 1993, 1994, 1997, 1998, 1999, 2001, 2002, |
8cabe764 GM |
3 | 2003, 2004, 2005, 2006, 2007, 2008 |
4 | Free Software Foundation, Inc. | |
ca1d1d23 JB |
5 | |
6 | This file is part of GNU Emacs. | |
7 | ||
9ec0b715 | 8 | GNU Emacs is free software: you can redistribute it and/or modify |
ca1d1d23 | 9 | it under the terms of the GNU General Public License as published by |
9ec0b715 GM |
10 | the Free Software Foundation, either version 3 of the License, or |
11 | (at your option) any later version. | |
ca1d1d23 JB |
12 | |
13 | GNU Emacs is distributed in the hope that it will be useful, | |
14 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
15 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
16 | GNU General Public License for more details. | |
17 | ||
18 | You should have received a copy of the GNU General Public License | |
9ec0b715 | 19 | along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */ |
ca1d1d23 JB |
20 | |
21 | ||
18160b98 | 22 | #include <config.h> |
ca1d1d23 JB |
23 | #include "lisp.h" |
24 | #include "syntax.h" | |
5679531d | 25 | #include "category.h" |
ca1d1d23 | 26 | #include "buffer.h" |
76eb0881 | 27 | #include "character.h" |
89ad6492 | 28 | #include "charset.h" |
9169c321 | 29 | #include "region-cache.h" |
ca1d1d23 | 30 | #include "commands.h" |
9ac0d9e0 | 31 | #include "blockinput.h" |
bf1760bb | 32 | #include "intervals.h" |
4746118a | 33 | |
ca1d1d23 JB |
34 | #include <sys/types.h> |
35 | #include "regex.h" | |
36 | ||
1d288aef | 37 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 38 | |
487282dc KH |
39 | /* If the regexp is non-nil, then the buffer contains the compiled form |
40 | of that regexp, suitable for searching. */ | |
1d288aef RS |
41 | struct regexp_cache |
42 | { | |
487282dc | 43 | struct regexp_cache *next; |
ecdb561e | 44 | Lisp_Object regexp, whitespace_regexp; |
b69e3c18 | 45 | /* Syntax table for which the regexp applies. We need this because |
58e95211 SM |
46 | of character classes. If this is t, then the compiled pattern is valid |
47 | for any syntax-table. */ | |
b69e3c18 | 48 | Lisp_Object syntax_table; |
487282dc KH |
49 | struct re_pattern_buffer buf; |
50 | char fastmap[0400]; | |
b819a390 RS |
51 | /* Nonzero means regexp was compiled to do full POSIX backtracking. */ |
52 | char posix; | |
487282dc | 53 | }; |
ca1d1d23 | 54 | |
487282dc KH |
55 | /* The instances of that struct. */ |
56 | struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; | |
ca1d1d23 | 57 | |
487282dc KH |
58 | /* The head of the linked list; points to the most recently used buffer. */ |
59 | struct regexp_cache *searchbuf_head; | |
ca1d1d23 | 60 | |
ca1d1d23 | 61 | |
4746118a JB |
62 | /* Every call to re_match, etc., must pass &search_regs as the regs |
63 | argument unless you can show it is unnecessary (i.e., if re_match | |
64 | is certainly going to be called again before region-around-match | |
65 | can be called). | |
66 | ||
67 | Since the registers are now dynamically allocated, we need to make | |
68 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
69 | been allocated by checking search_regs.num_regs. |
70 | ||
71 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
72 | buffer using bits in the re_pattern_buffer. This means that whenever |
73 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
74 | allocated any registers, and will allocate new registers the next |
75 | time you call a searching or matching function. Therefore, we need | |
76 | to call re_set_registers after compiling a new pattern or after | |
77 | setting the match registers, so that the regex functions will be | |
78 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
79 | static struct re_registers search_regs; |
80 | ||
daa37602 JB |
81 | /* The buffer in which the last search was performed, or |
82 | Qt if the last search was done in a string; | |
83 | Qnil if no searching has been done yet. */ | |
84 | static Lisp_Object last_thing_searched; | |
ca1d1d23 | 85 | |
8e6208c5 | 86 | /* error condition signaled when regexp compile_pattern fails */ |
ca1d1d23 JB |
87 | |
88 | Lisp_Object Qinvalid_regexp; | |
89 | ||
06f77a8a KS |
90 | /* Error condition used for failing searches */ |
91 | Lisp_Object Qsearch_failed; | |
92 | ||
41a33295 | 93 | Lisp_Object Vsearch_spaces_regexp; |
f31a9a68 | 94 | |
ef887810 RS |
95 | /* If non-nil, the match data will not be changed during call to |
96 | searching or matching functions. This variable is for internal use | |
97 | only. */ | |
98 | Lisp_Object Vinhibit_changing_match_data; | |
99 | ||
ca325161 | 100 | static void set_search_regs (); |
044f81f1 | 101 | static void save_search_regs (); |
facdc750 RS |
102 | static int simple_search (); |
103 | static int boyer_moore (); | |
b819a390 | 104 | static int search_buffer (); |
01d09305 | 105 | static void matcher_overflow () NO_RETURN; |
b819a390 | 106 | |
ca1d1d23 JB |
107 | static void |
108 | matcher_overflow () | |
109 | { | |
110 | error ("Stack overflow in regexp matcher"); | |
111 | } | |
112 | ||
b819a390 RS |
113 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
114 | PATTERN is the pattern to compile. | |
115 | CP is the place to put the result. | |
facdc750 | 116 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
117 | REGP is the structure that says where to store the "register" |
118 | values that will result from matching this pattern. | |
119 | If it is 0, we should compile the pattern not to record any | |
120 | subexpression bounds. | |
121 | POSIX is nonzero if we want full backtracking (POSIX style) | |
5679531d | 122 | for this pattern. 0 means backtrack only enough to get a valid match. |
bbc73b48 KH |
123 | |
124 | The behavior also depends on Vsearch_spaces_regexp. */ | |
ca1d1d23 | 125 | |
487282dc | 126 | static void |
89ad6492 | 127 | compile_pattern_1 (cp, pattern, translate, regp, posix) |
487282dc | 128 | struct regexp_cache *cp; |
ca1d1d23 | 129 | Lisp_Object pattern; |
facdc750 | 130 | Lisp_Object translate; |
487282dc | 131 | struct re_registers *regp; |
b819a390 | 132 | int posix; |
ca1d1d23 | 133 | { |
d451e4db | 134 | char *val; |
b819a390 | 135 | reg_syntax_t old; |
ca1d1d23 | 136 | |
487282dc | 137 | cp->regexp = Qnil; |
59fab369 | 138 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 139 | cp->posix = posix; |
93daa011 | 140 | cp->buf.multibyte = STRING_MULTIBYTE (pattern); |
89ad6492 | 141 | cp->buf.charset_unibyte = charset_unibyte; |
d8c85250 CY |
142 | if (STRINGP (Vsearch_spaces_regexp)) |
143 | cp->whitespace_regexp = Vsearch_spaces_regexp; | |
144 | else | |
145 | cp->whitespace_regexp = Qnil; | |
146 | ||
e7b2dd2e RS |
147 | /* rms: I think BLOCK_INPUT is not needed here any more, |
148 | because regex.c defines malloc to call xmalloc. | |
149 | Using BLOCK_INPUT here means the debugger won't run if an error occurs. | |
150 | So let's turn it off. */ | |
151 | /* BLOCK_INPUT; */ | |
fb4a568d | 152 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 153 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
d8c85250 CY |
154 | |
155 | if (STRINGP (Vsearch_spaces_regexp)) | |
156 | re_set_whitespace_regexp (SDATA (Vsearch_spaces_regexp)); | |
157 | else | |
158 | re_set_whitespace_regexp (NULL); | |
bbc73b48 | 159 | |
8f924df7 KH |
160 | val = (char *) re_compile_pattern ((char *) SDATA (pattern), |
161 | SBYTES (pattern), &cp->buf); | |
bbc73b48 | 162 | |
58e95211 SM |
163 | /* If the compiled pattern hard codes some of the contents of the |
164 | syntax-table, it can only be reused with *this* syntax table. */ | |
165 | cp->syntax_table = cp->buf.used_syntax ? current_buffer->syntax_table : Qt; | |
166 | ||
bbc73b48 KH |
167 | re_set_whitespace_regexp (NULL); |
168 | ||
b819a390 | 169 | re_set_syntax (old); |
e7b2dd2e | 170 | /* UNBLOCK_INPUT; */ |
ca1d1d23 | 171 | if (val) |
06f77a8a | 172 | xsignal1 (Qinvalid_regexp, build_string (val)); |
1113d9db | 173 | |
487282dc | 174 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
175 | } |
176 | ||
6efc7887 RS |
177 | /* Shrink each compiled regexp buffer in the cache |
178 | to the size actually used right now. | |
179 | This is called from garbage collection. */ | |
180 | ||
181 | void | |
182 | shrink_regexp_cache () | |
183 | { | |
a968f437 | 184 | struct regexp_cache *cp; |
6efc7887 RS |
185 | |
186 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
187 | { | |
188 | cp->buf.allocated = cp->buf.used; | |
189 | cp->buf.buffer | |
b23c0a83 | 190 | = (unsigned char *) xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
191 | } |
192 | } | |
193 | ||
54dd3310 SM |
194 | /* Clear the regexp cache w.r.t. a particular syntax table, |
195 | because it was changed. | |
e5b94d44 CY |
196 | There is no danger of memory leak here because re_compile_pattern |
197 | automagically manages the memory in each re_pattern_buffer struct, | |
198 | based on its `allocated' and `buffer' values. */ | |
199 | void | |
200 | clear_regexp_cache () | |
201 | { | |
202 | int i; | |
203 | ||
204 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) | |
54dd3310 SM |
205 | /* It's tempting to compare with the syntax-table we've actually changd, |
206 | but it's not sufficient because char-table inheritance mewans that | |
207 | modifying one syntax-table can change others at the same time. */ | |
208 | if (!EQ (searchbufs[i].syntax_table, Qt)) | |
209 | searchbufs[i].regexp = Qnil; | |
e5b94d44 CY |
210 | } |
211 | ||
487282dc | 212 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
213 | the cache. |
214 | PATTERN is the pattern to compile. | |
facdc750 | 215 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
216 | REGP is the structure that says where to store the "register" |
217 | values that will result from matching this pattern. | |
218 | If it is 0, we should compile the pattern not to record any | |
219 | subexpression bounds. | |
220 | POSIX is nonzero if we want full backtracking (POSIX style) | |
221 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
222 | |
223 | struct re_pattern_buffer * | |
0c8533c6 | 224 | compile_pattern (pattern, regp, translate, posix, multibyte) |
487282dc KH |
225 | Lisp_Object pattern; |
226 | struct re_registers *regp; | |
facdc750 | 227 | Lisp_Object translate; |
0c8533c6 | 228 | int posix, multibyte; |
487282dc KH |
229 | { |
230 | struct regexp_cache *cp, **cpp; | |
231 | ||
232 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
233 | { | |
234 | cp = *cpp; | |
f1b9c7c1 KR |
235 | /* Entries are initialized to nil, and may be set to nil by |
236 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
49a5f770 KR |
237 | string accessors in those cases. However, compile_pattern_1 |
238 | is only applied to the cache entry we pick here to reuse. So | |
239 | nil should never appear before a non-nil entry. */ | |
7c752c80 | 240 | if (NILP (cp->regexp)) |
f1b9c7c1 | 241 | goto compile_it; |
d5db4077 | 242 | if (SCHARS (cp->regexp) == SCHARS (pattern) |
cf69b13e | 243 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 244 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 245 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d | 246 | && cp->posix == posix |
58e95211 SM |
247 | && (EQ (cp->syntax_table, Qt) |
248 | || EQ (cp->syntax_table, current_buffer->syntax_table)) | |
89ad6492 KH |
249 | && !NILP (Fequal (cp->whitespace_regexp, Vsearch_spaces_regexp)) |
250 | && cp->buf.charset_unibyte == charset_unibyte) | |
487282dc KH |
251 | break; |
252 | ||
f1b9c7c1 KR |
253 | /* If we're at the end of the cache, compile into the nil cell |
254 | we found, or the last (least recently used) cell with a | |
255 | string value. */ | |
487282dc KH |
256 | if (cp->next == 0) |
257 | { | |
f1b9c7c1 | 258 | compile_it: |
89ad6492 | 259 | compile_pattern_1 (cp, pattern, translate, regp, posix); |
487282dc KH |
260 | break; |
261 | } | |
262 | } | |
263 | ||
264 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
265 | either because we found it in the cache or because we just compiled it. | |
266 | Move it to the front of the queue to mark it as most recently used. */ | |
267 | *cpp = cp->next; | |
268 | cp->next = searchbuf_head; | |
269 | searchbuf_head = cp; | |
1113d9db | 270 | |
6639708c RS |
271 | /* Advise the searching functions about the space we have allocated |
272 | for register data. */ | |
273 | if (regp) | |
274 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
275 | ||
89ad6492 KH |
276 | /* The compiled pattern can be used both for mulitbyte and unibyte |
277 | target. But, we have to tell which the pattern is used for. */ | |
278 | cp->buf.target_multibyte = multibyte; | |
279 | ||
487282dc | 280 | return &cp->buf; |
ca1d1d23 JB |
281 | } |
282 | ||
ca1d1d23 | 283 | \f |
b819a390 RS |
284 | static Lisp_Object |
285 | looking_at_1 (string, posix) | |
ca1d1d23 | 286 | Lisp_Object string; |
b819a390 | 287 | int posix; |
ca1d1d23 JB |
288 | { |
289 | Lisp_Object val; | |
290 | unsigned char *p1, *p2; | |
291 | int s1, s2; | |
292 | register int i; | |
487282dc | 293 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 294 | |
7074fde6 FP |
295 | if (running_asynch_code) |
296 | save_search_regs (); | |
297 | ||
910c747a RS |
298 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
299 | XCHAR_TABLE (current_buffer->case_canon_table)->extras[2] | |
300 | = current_buffer->case_eqv_table; | |
301 | ||
b7826503 | 302 | CHECK_STRING (string); |
ef887810 RS |
303 | bufp = compile_pattern (string, |
304 | (NILP (Vinhibit_changing_match_data) | |
305 | ? &search_regs : NULL), | |
487282dc | 306 | (!NILP (current_buffer->case_fold_search) |
0190922f | 307 | ? current_buffer->case_canon_table : Qnil), |
0c8533c6 RS |
308 | posix, |
309 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 JB |
310 | |
311 | immediate_quit = 1; | |
312 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
313 | ||
314 | /* Get pointers and sizes of the two strings | |
315 | that make up the visible portion of the buffer. */ | |
316 | ||
317 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 318 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 319 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 320 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
321 | if (s1 < 0) |
322 | { | |
323 | p2 = p1; | |
fa8ed3e0 | 324 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
325 | s1 = 0; |
326 | } | |
327 | if (s2 < 0) | |
328 | { | |
fa8ed3e0 | 329 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
330 | s2 = 0; |
331 | } | |
8bb43c28 RS |
332 | |
333 | re_match_object = Qnil; | |
177c0ea7 | 334 | |
487282dc | 335 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
ef887810 RS |
336 | PT_BYTE - BEGV_BYTE, |
337 | (NILP (Vinhibit_changing_match_data) | |
338 | ? &search_regs : NULL), | |
fa8ed3e0 | 339 | ZV_BYTE - BEGV_BYTE); |
de182d70 | 340 | immediate_quit = 0; |
177c0ea7 | 341 | |
ca1d1d23 JB |
342 | if (i == -2) |
343 | matcher_overflow (); | |
344 | ||
345 | val = (0 <= i ? Qt : Qnil); | |
ef887810 | 346 | if (NILP (Vinhibit_changing_match_data) && i >= 0) |
fa8ed3e0 RS |
347 | for (i = 0; i < search_regs.num_regs; i++) |
348 | if (search_regs.start[i] >= 0) | |
349 | { | |
350 | search_regs.start[i] | |
351 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
352 | search_regs.end[i] | |
353 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
354 | } | |
ef887810 RS |
355 | |
356 | /* Set last_thing_searched only when match data is changed. */ | |
357 | if (NILP (Vinhibit_changing_match_data)) | |
358 | XSETBUFFER (last_thing_searched, current_buffer); | |
359 | ||
ca1d1d23 JB |
360 | return val; |
361 | } | |
362 | ||
b819a390 | 363 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
364 | doc: /* Return t if text after point matches regular expression REGEXP. |
365 | This function modifies the match data that `match-beginning', | |
366 | `match-end' and `match-data' access; save and restore the match | |
367 | data if you want to preserve them. */) | |
368 | (regexp) | |
94f94972 | 369 | Lisp_Object regexp; |
b819a390 | 370 | { |
94f94972 | 371 | return looking_at_1 (regexp, 0); |
b819a390 RS |
372 | } |
373 | ||
374 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
375 | doc: /* Return t if text after point matches regular expression REGEXP. |
376 | Find the longest match, in accord with Posix regular expression rules. | |
377 | This function modifies the match data that `match-beginning', | |
378 | `match-end' and `match-data' access; save and restore the match | |
379 | data if you want to preserve them. */) | |
380 | (regexp) | |
94f94972 | 381 | Lisp_Object regexp; |
b819a390 | 382 | { |
94f94972 | 383 | return looking_at_1 (regexp, 1); |
b819a390 RS |
384 | } |
385 | \f | |
386 | static Lisp_Object | |
387 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 388 | Lisp_Object regexp, string, start; |
b819a390 | 389 | int posix; |
ca1d1d23 JB |
390 | { |
391 | int val; | |
487282dc | 392 | struct re_pattern_buffer *bufp; |
0c8533c6 RS |
393 | int pos, pos_byte; |
394 | int i; | |
ca1d1d23 | 395 | |
7074fde6 FP |
396 | if (running_asynch_code) |
397 | save_search_regs (); | |
398 | ||
b7826503 PJ |
399 | CHECK_STRING (regexp); |
400 | CHECK_STRING (string); | |
ca1d1d23 JB |
401 | |
402 | if (NILP (start)) | |
0c8533c6 | 403 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
404 | else |
405 | { | |
d5db4077 | 406 | int len = SCHARS (string); |
ca1d1d23 | 407 | |
b7826503 | 408 | CHECK_NUMBER (start); |
0c8533c6 RS |
409 | pos = XINT (start); |
410 | if (pos < 0 && -pos <= len) | |
411 | pos = len + pos; | |
412 | else if (0 > pos || pos > len) | |
ca1d1d23 | 413 | args_out_of_range (string, start); |
0c8533c6 | 414 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
415 | } |
416 | ||
910c747a RS |
417 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
418 | XCHAR_TABLE (current_buffer->case_canon_table)->extras[2] | |
419 | = current_buffer->case_eqv_table; | |
420 | ||
ef887810 RS |
421 | bufp = compile_pattern (regexp, |
422 | (NILP (Vinhibit_changing_match_data) | |
423 | ? &search_regs : NULL), | |
487282dc | 424 | (!NILP (current_buffer->case_fold_search) |
0190922f | 425 | ? current_buffer->case_canon_table : Qnil), |
0c8533c6 RS |
426 | posix, |
427 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 428 | immediate_quit = 1; |
8bb43c28 | 429 | re_match_object = string; |
177c0ea7 | 430 | |
d5db4077 KR |
431 | val = re_search (bufp, (char *) SDATA (string), |
432 | SBYTES (string), pos_byte, | |
433 | SBYTES (string) - pos_byte, | |
ef887810 RS |
434 | (NILP (Vinhibit_changing_match_data) |
435 | ? &search_regs : NULL)); | |
ca1d1d23 | 436 | immediate_quit = 0; |
ef887810 RS |
437 | |
438 | /* Set last_thing_searched only when match data is changed. */ | |
439 | if (NILP (Vinhibit_changing_match_data)) | |
440 | last_thing_searched = Qt; | |
441 | ||
ca1d1d23 JB |
442 | if (val == -2) |
443 | matcher_overflow (); | |
444 | if (val < 0) return Qnil; | |
0c8533c6 | 445 | |
ef887810 RS |
446 | if (NILP (Vinhibit_changing_match_data)) |
447 | for (i = 0; i < search_regs.num_regs; i++) | |
448 | if (search_regs.start[i] >= 0) | |
449 | { | |
450 | search_regs.start[i] | |
451 | = string_byte_to_char (string, search_regs.start[i]); | |
452 | search_regs.end[i] | |
453 | = string_byte_to_char (string, search_regs.end[i]); | |
454 | } | |
0c8533c6 RS |
455 | |
456 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 457 | } |
e59a8453 | 458 | |
b819a390 | 459 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 | 460 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
b85acc4b | 461 | Matching ignores case if `case-fold-search' is non-nil. |
8c1a1077 PJ |
462 | If third arg START is non-nil, start search at that index in STRING. |
463 | For index of first char beyond the match, do (match-end 0). | |
464 | `match-end' and `match-beginning' also give indices of substrings | |
2bd2f32d RS |
465 | matched by parenthesis constructs in the pattern. |
466 | ||
467 | You can use the function `match-string' to extract the substrings | |
468 | matched by the parenthesis constructions in REGEXP. */) | |
8c1a1077 | 469 | (regexp, string, start) |
b819a390 RS |
470 | Lisp_Object regexp, string, start; |
471 | { | |
472 | return string_match_1 (regexp, string, start, 0); | |
473 | } | |
474 | ||
475 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
476 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
477 | Find the longest match, in accord with Posix regular expression rules. | |
478 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
479 | If third arg START is non-nil, start search at that index in STRING. | |
480 | For index of first char beyond the match, do (match-end 0). | |
481 | `match-end' and `match-beginning' also give indices of substrings | |
482 | matched by parenthesis constructs in the pattern. */) | |
483 | (regexp, string, start) | |
b819a390 RS |
484 | Lisp_Object regexp, string, start; |
485 | { | |
486 | return string_match_1 (regexp, string, start, 1); | |
487 | } | |
488 | ||
e59a8453 RS |
489 | /* Match REGEXP against STRING, searching all of STRING, |
490 | and return the index of the match, or negative on failure. | |
491 | This does not clobber the match data. */ | |
492 | ||
493 | int | |
494 | fast_string_match (regexp, string) | |
495 | Lisp_Object regexp, string; | |
496 | { | |
497 | int val; | |
487282dc | 498 | struct re_pattern_buffer *bufp; |
e59a8453 | 499 | |
facdc750 RS |
500 | bufp = compile_pattern (regexp, 0, Qnil, |
501 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 502 | immediate_quit = 1; |
8bb43c28 | 503 | re_match_object = string; |
177c0ea7 | 504 | |
d5db4077 KR |
505 | val = re_search (bufp, (char *) SDATA (string), |
506 | SBYTES (string), 0, | |
507 | SBYTES (string), 0); | |
e59a8453 RS |
508 | immediate_quit = 0; |
509 | return val; | |
510 | } | |
5679531d KH |
511 | |
512 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
513 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
514 | This does not clobber the match data. |
515 | We assume that STRING contains single-byte characters. */ | |
5679531d KH |
516 | |
517 | extern Lisp_Object Vascii_downcase_table; | |
518 | ||
519 | int | |
b4577c63 | 520 | fast_c_string_match_ignore_case (regexp, string) |
5679531d | 521 | Lisp_Object regexp; |
96b80561 | 522 | const char *string; |
5679531d KH |
523 | { |
524 | int val; | |
525 | struct re_pattern_buffer *bufp; | |
526 | int len = strlen (string); | |
527 | ||
0c8533c6 | 528 | regexp = string_make_unibyte (regexp); |
b4577c63 | 529 | re_match_object = Qt; |
5679531d | 530 | bufp = compile_pattern (regexp, 0, |
0190922f | 531 | Vascii_canon_table, 0, |
f8bd51c4 | 532 | 0); |
5679531d KH |
533 | immediate_quit = 1; |
534 | val = re_search (bufp, string, len, 0, len, 0); | |
535 | immediate_quit = 0; | |
536 | return val; | |
537 | } | |
be5f4dfb KH |
538 | |
539 | /* Like fast_string_match but ignore case. */ | |
540 | ||
541 | int | |
542 | fast_string_match_ignore_case (regexp, string) | |
543 | Lisp_Object regexp, string; | |
544 | { | |
545 | int val; | |
546 | struct re_pattern_buffer *bufp; | |
547 | ||
0190922f | 548 | bufp = compile_pattern (regexp, 0, Vascii_canon_table, |
be5f4dfb KH |
549 | 0, STRING_MULTIBYTE (string)); |
550 | immediate_quit = 1; | |
551 | re_match_object = string; | |
552 | ||
553 | val = re_search (bufp, (char *) SDATA (string), | |
554 | SBYTES (string), 0, | |
555 | SBYTES (string), 0); | |
556 | immediate_quit = 0; | |
557 | return val; | |
558 | } | |
ca1d1d23 | 559 | \f |
9169c321 JB |
560 | /* The newline cache: remembering which sections of text have no newlines. */ |
561 | ||
562 | /* If the user has requested newline caching, make sure it's on. | |
563 | Otherwise, make sure it's off. | |
564 | This is our cheezy way of associating an action with the change of | |
565 | state of a buffer-local variable. */ | |
566 | static void | |
567 | newline_cache_on_off (buf) | |
568 | struct buffer *buf; | |
569 | { | |
570 | if (NILP (buf->cache_long_line_scans)) | |
571 | { | |
572 | /* It should be off. */ | |
573 | if (buf->newline_cache) | |
574 | { | |
575 | free_region_cache (buf->newline_cache); | |
576 | buf->newline_cache = 0; | |
577 | } | |
578 | } | |
579 | else | |
580 | { | |
581 | /* It should be on. */ | |
582 | if (buf->newline_cache == 0) | |
583 | buf->newline_cache = new_region_cache (); | |
584 | } | |
585 | } | |
586 | ||
587 | \f | |
588 | /* Search for COUNT instances of the character TARGET between START and END. | |
589 | ||
590 | If COUNT is positive, search forwards; END must be >= START. | |
591 | If COUNT is negative, search backwards for the -COUNTth instance; | |
592 | END must be <= START. | |
593 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
594 | ||
595 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
596 | direction indicated by COUNT. | |
ffd56f97 JB |
597 | |
598 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
a9f2a45f | 599 | position past the COUNTth match. Note that for reverse motion |
5bfe95c9 | 600 | this is not the same as the usual convention for Emacs motion commands. |
ffd56f97 | 601 | |
9169c321 JB |
602 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
603 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 604 | |
087a5f81 RS |
605 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
606 | except when inside redisplay. */ | |
607 | ||
dfcf069d | 608 | int |
9169c321 JB |
609 | scan_buffer (target, start, end, count, shortage, allow_quit) |
610 | register int target; | |
611 | int start, end; | |
612 | int count; | |
613 | int *shortage; | |
087a5f81 | 614 | int allow_quit; |
ca1d1d23 | 615 | { |
9169c321 | 616 | struct region_cache *newline_cache; |
177c0ea7 | 617 | int direction; |
ffd56f97 | 618 | |
9169c321 JB |
619 | if (count > 0) |
620 | { | |
621 | direction = 1; | |
622 | if (! end) end = ZV; | |
623 | } | |
624 | else | |
625 | { | |
626 | direction = -1; | |
627 | if (! end) end = BEGV; | |
628 | } | |
ffd56f97 | 629 | |
9169c321 JB |
630 | newline_cache_on_off (current_buffer); |
631 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
632 | |
633 | if (shortage != 0) | |
634 | *shortage = 0; | |
635 | ||
087a5f81 | 636 | immediate_quit = allow_quit; |
ca1d1d23 | 637 | |
ffd56f97 | 638 | if (count > 0) |
9169c321 | 639 | while (start != end) |
ca1d1d23 | 640 | { |
9169c321 JB |
641 | /* Our innermost scanning loop is very simple; it doesn't know |
642 | about gaps, buffer ends, or the newline cache. ceiling is | |
643 | the position of the last character before the next such | |
644 | obstacle --- the last character the dumb search loop should | |
645 | examine. */ | |
fa8ed3e0 RS |
646 | int ceiling_byte = CHAR_TO_BYTE (end) - 1; |
647 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 648 | int tem; |
9169c321 JB |
649 | |
650 | /* If we're looking for a newline, consult the newline cache | |
651 | to see where we can avoid some scanning. */ | |
652 | if (target == '\n' && newline_cache) | |
653 | { | |
654 | int next_change; | |
655 | immediate_quit = 0; | |
656 | while (region_cache_forward | |
fa8ed3e0 RS |
657 | (current_buffer, newline_cache, start_byte, &next_change)) |
658 | start_byte = next_change; | |
cbe0db0d | 659 | immediate_quit = allow_quit; |
9169c321 | 660 | |
fa8ed3e0 RS |
661 | /* START should never be after END. */ |
662 | if (start_byte > ceiling_byte) | |
663 | start_byte = ceiling_byte; | |
9169c321 JB |
664 | |
665 | /* Now the text after start is an unknown region, and | |
666 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 667 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
668 | } |
669 | ||
670 | /* The dumb loop can only scan text stored in contiguous | |
671 | bytes. BUFFER_CEILING_OF returns the last character | |
672 | position that is contiguous, so the ceiling is the | |
673 | position after that. */ | |
67ce527d KH |
674 | tem = BUFFER_CEILING_OF (start_byte); |
675 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
676 | |
677 | { | |
177c0ea7 | 678 | /* The termination address of the dumb loop. */ |
fa8ed3e0 RS |
679 | register unsigned char *ceiling_addr |
680 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
681 | register unsigned char *cursor | |
682 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
683 | unsigned char *base = cursor; |
684 | ||
685 | while (cursor < ceiling_addr) | |
686 | { | |
687 | unsigned char *scan_start = cursor; | |
688 | ||
689 | /* The dumb loop. */ | |
690 | while (*cursor != target && ++cursor < ceiling_addr) | |
691 | ; | |
692 | ||
693 | /* If we're looking for newlines, cache the fact that | |
694 | the region from start to cursor is free of them. */ | |
695 | if (target == '\n' && newline_cache) | |
696 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
697 | start_byte + scan_start - base, |
698 | start_byte + cursor - base); | |
9169c321 JB |
699 | |
700 | /* Did we find the target character? */ | |
701 | if (cursor < ceiling_addr) | |
702 | { | |
703 | if (--count == 0) | |
704 | { | |
705 | immediate_quit = 0; | |
fa8ed3e0 | 706 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
707 | } |
708 | cursor++; | |
709 | } | |
710 | } | |
711 | ||
fa8ed3e0 | 712 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 713 | } |
ca1d1d23 JB |
714 | } |
715 | else | |
9169c321 JB |
716 | while (start > end) |
717 | { | |
718 | /* The last character to check before the next obstacle. */ | |
fa8ed3e0 RS |
719 | int ceiling_byte = CHAR_TO_BYTE (end); |
720 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 721 | int tem; |
9169c321 JB |
722 | |
723 | /* Consult the newline cache, if appropriate. */ | |
724 | if (target == '\n' && newline_cache) | |
725 | { | |
726 | int next_change; | |
727 | immediate_quit = 0; | |
728 | while (region_cache_backward | |
fa8ed3e0 RS |
729 | (current_buffer, newline_cache, start_byte, &next_change)) |
730 | start_byte = next_change; | |
cbe0db0d | 731 | immediate_quit = allow_quit; |
9169c321 JB |
732 | |
733 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
734 | if (start_byte <= ceiling_byte) |
735 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
736 | |
737 | /* Now the text before start is an unknown region, and | |
738 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 739 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
740 | } |
741 | ||
742 | /* Stop scanning before the gap. */ | |
67ce527d KH |
743 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
744 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
745 | |
746 | { | |
747 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
748 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
749 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
750 | unsigned char *base = cursor; |
751 | ||
752 | while (cursor >= ceiling_addr) | |
753 | { | |
754 | unsigned char *scan_start = cursor; | |
755 | ||
756 | while (*cursor != target && --cursor >= ceiling_addr) | |
757 | ; | |
758 | ||
759 | /* If we're looking for newlines, cache the fact that | |
760 | the region from after the cursor to start is free of them. */ | |
761 | if (target == '\n' && newline_cache) | |
762 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
763 | start_byte + cursor - base, |
764 | start_byte + scan_start - base); | |
9169c321 JB |
765 | |
766 | /* Did we find the target character? */ | |
767 | if (cursor >= ceiling_addr) | |
768 | { | |
769 | if (++count >= 0) | |
770 | { | |
771 | immediate_quit = 0; | |
fa8ed3e0 | 772 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
773 | } |
774 | cursor--; | |
775 | } | |
776 | } | |
777 | ||
fa8ed3e0 | 778 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
779 | } |
780 | } | |
781 | ||
ca1d1d23 JB |
782 | immediate_quit = 0; |
783 | if (shortage != 0) | |
ffd56f97 | 784 | *shortage = count * direction; |
9169c321 | 785 | return start; |
ca1d1d23 | 786 | } |
fa8ed3e0 RS |
787 | \f |
788 | /* Search for COUNT instances of a line boundary, which means either a | |
789 | newline or (if selective display enabled) a carriage return. | |
790 | Start at START. If COUNT is negative, search backwards. | |
791 | ||
792 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
793 | ||
794 | If we find COUNT instances. we position after (always after, | |
795 | even if scanning backwards) the COUNTth match, and return 0. | |
796 | ||
797 | If we don't find COUNT instances before reaching the end of the | |
798 | buffer (or the beginning, if scanning backwards), we return | |
799 | the number of line boundaries left unfound, and position at | |
800 | the limit we bumped up against. | |
801 | ||
802 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 803 | except in special cases. */ |
ca1d1d23 | 804 | |
63fa018d | 805 | int |
fa8ed3e0 RS |
806 | scan_newline (start, start_byte, limit, limit_byte, count, allow_quit) |
807 | int start, start_byte; | |
808 | int limit, limit_byte; | |
809 | register int count; | |
810 | int allow_quit; | |
63fa018d | 811 | { |
fa8ed3e0 RS |
812 | int direction = ((count > 0) ? 1 : -1); |
813 | ||
814 | register unsigned char *cursor; | |
815 | unsigned char *base; | |
816 | ||
817 | register int ceiling; | |
818 | register unsigned char *ceiling_addr; | |
819 | ||
d5d57b92 RS |
820 | int old_immediate_quit = immediate_quit; |
821 | ||
fa8ed3e0 RS |
822 | /* The code that follows is like scan_buffer |
823 | but checks for either newline or carriage return. */ | |
824 | ||
d5d57b92 RS |
825 | if (allow_quit) |
826 | immediate_quit++; | |
fa8ed3e0 RS |
827 | |
828 | start_byte = CHAR_TO_BYTE (start); | |
829 | ||
830 | if (count > 0) | |
831 | { | |
832 | while (start_byte < limit_byte) | |
833 | { | |
834 | ceiling = BUFFER_CEILING_OF (start_byte); | |
835 | ceiling = min (limit_byte - 1, ceiling); | |
836 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
837 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
838 | while (1) | |
839 | { | |
840 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
841 | ; | |
842 | ||
843 | if (cursor != ceiling_addr) | |
844 | { | |
845 | if (--count == 0) | |
846 | { | |
d5d57b92 | 847 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
848 | start_byte = start_byte + cursor - base + 1; |
849 | start = BYTE_TO_CHAR (start_byte); | |
850 | TEMP_SET_PT_BOTH (start, start_byte); | |
851 | return 0; | |
852 | } | |
853 | else | |
854 | if (++cursor == ceiling_addr) | |
855 | break; | |
856 | } | |
857 | else | |
858 | break; | |
859 | } | |
860 | start_byte += cursor - base; | |
861 | } | |
862 | } | |
863 | else | |
864 | { | |
fa8ed3e0 RS |
865 | while (start_byte > limit_byte) |
866 | { | |
867 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
868 | ceiling = max (limit_byte, ceiling); | |
869 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
870 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
871 | while (1) | |
872 | { | |
873 | while (--cursor != ceiling_addr && *cursor != '\n') | |
874 | ; | |
875 | ||
876 | if (cursor != ceiling_addr) | |
877 | { | |
878 | if (++count == 0) | |
879 | { | |
d5d57b92 | 880 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
881 | /* Return the position AFTER the match we found. */ |
882 | start_byte = start_byte + cursor - base + 1; | |
883 | start = BYTE_TO_CHAR (start_byte); | |
884 | TEMP_SET_PT_BOTH (start, start_byte); | |
885 | return 0; | |
886 | } | |
887 | } | |
888 | else | |
889 | break; | |
890 | } | |
891 | /* Here we add 1 to compensate for the last decrement | |
892 | of CURSOR, which took it past the valid range. */ | |
893 | start_byte += cursor - base + 1; | |
894 | } | |
895 | } | |
896 | ||
897 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 898 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
899 | |
900 | return count * direction; | |
63fa018d RS |
901 | } |
902 | ||
ca1d1d23 | 903 | int |
fa8ed3e0 | 904 | find_next_newline_no_quit (from, cnt) |
ca1d1d23 JB |
905 | register int from, cnt; |
906 | { | |
fa8ed3e0 | 907 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
9169c321 JB |
908 | } |
909 | ||
9169c321 JB |
910 | /* Like find_next_newline, but returns position before the newline, |
911 | not after, and only search up to TO. This isn't just | |
912 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 913 | |
9169c321 JB |
914 | int |
915 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 916 | int from, to, cnt; |
9169c321 JB |
917 | { |
918 | int shortage; | |
919 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
920 | ||
921 | if (shortage == 0) | |
922 | pos--; | |
177c0ea7 | 923 | |
9169c321 | 924 | return pos; |
ca1d1d23 JB |
925 | } |
926 | \f | |
ca1d1d23 JB |
927 | /* Subroutines of Lisp buffer search functions. */ |
928 | ||
929 | static Lisp_Object | |
b819a390 | 930 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
931 | Lisp_Object string, bound, noerror, count; |
932 | int direction; | |
933 | int RE; | |
b819a390 | 934 | int posix; |
ca1d1d23 JB |
935 | { |
936 | register int np; | |
9f43ad85 | 937 | int lim, lim_byte; |
ca1d1d23 JB |
938 | int n = direction; |
939 | ||
940 | if (!NILP (count)) | |
941 | { | |
b7826503 | 942 | CHECK_NUMBER (count); |
ca1d1d23 JB |
943 | n *= XINT (count); |
944 | } | |
945 | ||
b7826503 | 946 | CHECK_STRING (string); |
ca1d1d23 | 947 | if (NILP (bound)) |
9f43ad85 RS |
948 | { |
949 | if (n > 0) | |
950 | lim = ZV, lim_byte = ZV_BYTE; | |
951 | else | |
952 | lim = BEGV, lim_byte = BEGV_BYTE; | |
953 | } | |
ca1d1d23 JB |
954 | else |
955 | { | |
b7826503 | 956 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 957 | lim = XINT (bound); |
6ec8bbd2 | 958 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
959 | error ("Invalid search bound (wrong side of point)"); |
960 | if (lim > ZV) | |
9f43ad85 | 961 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 962 | else if (lim < BEGV) |
9f43ad85 | 963 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
964 | else |
965 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
966 | } |
967 | ||
910c747a RS |
968 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
969 | XCHAR_TABLE (current_buffer->case_canon_table)->extras[2] | |
970 | = current_buffer->case_eqv_table; | |
971 | ||
9f43ad85 | 972 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
ca1d1d23 | 973 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 974 | ? current_buffer->case_canon_table |
3135e9fd | 975 | : Qnil), |
ca1d1d23 | 976 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 977 | ? current_buffer->case_eqv_table |
3135e9fd | 978 | : Qnil), |
b819a390 | 979 | posix); |
ca1d1d23 JB |
980 | if (np <= 0) |
981 | { | |
982 | if (NILP (noerror)) | |
06f77a8a KS |
983 | xsignal1 (Qsearch_failed, string); |
984 | ||
ca1d1d23 JB |
985 | if (!EQ (noerror, Qt)) |
986 | { | |
987 | if (lim < BEGV || lim > ZV) | |
988 | abort (); | |
9f43ad85 | 989 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
990 | return Qnil; |
991 | #if 0 /* This would be clean, but maybe programs depend on | |
992 | a value of nil here. */ | |
481399bf | 993 | np = lim; |
a5f217b8 | 994 | #endif |
ca1d1d23 | 995 | } |
481399bf RS |
996 | else |
997 | return Qnil; | |
ca1d1d23 JB |
998 | } |
999 | ||
1000 | if (np < BEGV || np > ZV) | |
1001 | abort (); | |
1002 | ||
1003 | SET_PT (np); | |
1004 | ||
1005 | return make_number (np); | |
1006 | } | |
1007 | \f | |
fa8ed3e0 RS |
1008 | /* Return 1 if REGEXP it matches just one constant string. */ |
1009 | ||
b6d6a51c KH |
1010 | static int |
1011 | trivial_regexp_p (regexp) | |
1012 | Lisp_Object regexp; | |
1013 | { | |
d5db4077 KR |
1014 | int len = SBYTES (regexp); |
1015 | unsigned char *s = SDATA (regexp); | |
b6d6a51c KH |
1016 | while (--len >= 0) |
1017 | { | |
1018 | switch (*s++) | |
1019 | { | |
1020 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
1021 | return 0; | |
1022 | case '\\': | |
1023 | if (--len < 0) | |
1024 | return 0; | |
1025 | switch (*s++) | |
1026 | { | |
1027 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
1028 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 1029 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 1030 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 1031 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
1032 | case '6': case '7': case '8': case '9': |
1033 | return 0; | |
1034 | } | |
1035 | } | |
1036 | } | |
1037 | return 1; | |
1038 | } | |
1039 | ||
ca325161 | 1040 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 1041 | starting at position POS and stopping at position LIM, |
b819a390 | 1042 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
1043 | a regular expression if RE is true. |
1044 | ||
1045 | If N is positive, searching is forward and LIM must be greater than POS. | |
1046 | If N is negative, searching is backward and LIM must be less than POS. | |
1047 | ||
facdc750 | 1048 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 1049 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
1050 | (if searching backward) or the end (if searching forward). |
1051 | ||
1052 | POSIX is nonzero if we want full backtracking (POSIX style) | |
1053 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 1054 | |
aff2ce94 RS |
1055 | #define TRANSLATE(out, trt, d) \ |
1056 | do \ | |
1057 | { \ | |
1058 | if (! NILP (trt)) \ | |
1059 | { \ | |
1060 | Lisp_Object temp; \ | |
1061 | temp = Faref (trt, make_number (d)); \ | |
1062 | if (INTEGERP (temp)) \ | |
1063 | out = XINT (temp); \ | |
1064 | else \ | |
1065 | out = d; \ | |
1066 | } \ | |
1067 | else \ | |
1068 | out = d; \ | |
1069 | } \ | |
1070 | while (0) | |
facdc750 | 1071 | |
ef887810 RS |
1072 | /* Only used in search_buffer, to record the end position of the match |
1073 | when searching regexps and SEARCH_REGS should not be changed | |
1074 | (i.e. Vinhibit_changing_match_data is non-nil). */ | |
1075 | static struct re_registers search_regs_1; | |
1076 | ||
b819a390 | 1077 | static int |
9f43ad85 RS |
1078 | search_buffer (string, pos, pos_byte, lim, lim_byte, n, |
1079 | RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
1080 | Lisp_Object string; |
1081 | int pos; | |
9f43ad85 | 1082 | int pos_byte; |
ca1d1d23 | 1083 | int lim; |
9f43ad85 | 1084 | int lim_byte; |
ca1d1d23 JB |
1085 | int n; |
1086 | int RE; | |
facdc750 RS |
1087 | Lisp_Object trt; |
1088 | Lisp_Object inverse_trt; | |
b819a390 | 1089 | int posix; |
ca1d1d23 | 1090 | { |
d5db4077 KR |
1091 | int len = SCHARS (string); |
1092 | int len_byte = SBYTES (string); | |
facdc750 | 1093 | register int i; |
ca1d1d23 | 1094 | |
7074fde6 FP |
1095 | if (running_asynch_code) |
1096 | save_search_regs (); | |
1097 | ||
a7e4cdde | 1098 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1099 | /* Null string is found at starting position. */ |
a7e4cdde | 1100 | if (len == 0 || n == 0) |
ca325161 | 1101 | { |
0353b28f | 1102 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1103 | return pos; |
1104 | } | |
3f57a499 | 1105 | |
41a33295 | 1106 | if (RE && !(trivial_regexp_p (string) && NILP (Vsearch_spaces_regexp))) |
ca1d1d23 | 1107 | { |
facdc750 RS |
1108 | unsigned char *p1, *p2; |
1109 | int s1, s2; | |
487282dc KH |
1110 | struct re_pattern_buffer *bufp; |
1111 | ||
ef887810 RS |
1112 | bufp = compile_pattern (string, |
1113 | (NILP (Vinhibit_changing_match_data) | |
1114 | ? &search_regs : &search_regs_1), | |
1115 | trt, posix, | |
0c8533c6 | 1116 | !NILP (current_buffer->enable_multibyte_characters)); |
ca1d1d23 | 1117 | |
ca1d1d23 JB |
1118 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1119 | because letting this function finish | |
1120 | can take too long. */ | |
1121 | QUIT; /* Do a pending quit right away, | |
1122 | to avoid paradoxical behavior */ | |
1123 | /* Get pointers and sizes of the two strings | |
1124 | that make up the visible portion of the buffer. */ | |
1125 | ||
1126 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1127 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1128 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1129 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1130 | if (s1 < 0) |
1131 | { | |
1132 | p2 = p1; | |
fa8ed3e0 | 1133 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1134 | s1 = 0; |
1135 | } | |
1136 | if (s2 < 0) | |
1137 | { | |
fa8ed3e0 | 1138 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1139 | s2 = 0; |
1140 | } | |
8bb43c28 | 1141 | re_match_object = Qnil; |
177c0ea7 | 1142 | |
ca1d1d23 JB |
1143 | while (n < 0) |
1144 | { | |
42db823b | 1145 | int val; |
487282dc | 1146 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1147 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1148 | (NILP (Vinhibit_changing_match_data) |
1149 | ? &search_regs : &search_regs_1), | |
42db823b | 1150 | /* Don't allow match past current point */ |
4996330b | 1151 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1152 | if (val == -2) |
b6d6a51c KH |
1153 | { |
1154 | matcher_overflow (); | |
1155 | } | |
ca1d1d23 JB |
1156 | if (val >= 0) |
1157 | { | |
ef887810 RS |
1158 | if (NILP (Vinhibit_changing_match_data)) |
1159 | { | |
1160 | pos_byte = search_regs.start[0] + BEGV_BYTE; | |
1161 | for (i = 0; i < search_regs.num_regs; i++) | |
1162 | if (search_regs.start[i] >= 0) | |
1163 | { | |
1164 | search_regs.start[i] | |
1165 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1166 | search_regs.end[i] | |
1167 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1168 | } | |
1169 | XSETBUFFER (last_thing_searched, current_buffer); | |
1170 | /* Set pos to the new position. */ | |
1171 | pos = search_regs.start[0]; | |
1172 | } | |
1173 | else | |
1174 | { | |
1175 | pos_byte = search_regs_1.start[0] + BEGV_BYTE; | |
1176 | /* Set pos to the new position. */ | |
1177 | pos = BYTE_TO_CHAR (search_regs_1.start[0] + BEGV_BYTE); | |
1178 | } | |
ca1d1d23 JB |
1179 | } |
1180 | else | |
1181 | { | |
1182 | immediate_quit = 0; | |
1183 | return (n); | |
1184 | } | |
1185 | n++; | |
1186 | } | |
1187 | while (n > 0) | |
1188 | { | |
42db823b | 1189 | int val; |
487282dc | 1190 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1191 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1192 | (NILP (Vinhibit_changing_match_data) |
1193 | ? &search_regs : &search_regs_1), | |
4996330b | 1194 | lim_byte - BEGV_BYTE); |
ca1d1d23 | 1195 | if (val == -2) |
b6d6a51c KH |
1196 | { |
1197 | matcher_overflow (); | |
1198 | } | |
ca1d1d23 JB |
1199 | if (val >= 0) |
1200 | { | |
ef887810 RS |
1201 | if (NILP (Vinhibit_changing_match_data)) |
1202 | { | |
1203 | pos_byte = search_regs.end[0] + BEGV_BYTE; | |
1204 | for (i = 0; i < search_regs.num_regs; i++) | |
1205 | if (search_regs.start[i] >= 0) | |
1206 | { | |
1207 | search_regs.start[i] | |
1208 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1209 | search_regs.end[i] | |
1210 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1211 | } | |
1212 | XSETBUFFER (last_thing_searched, current_buffer); | |
1213 | pos = search_regs.end[0]; | |
1214 | } | |
1215 | else | |
1216 | { | |
1217 | pos_byte = search_regs_1.end[0] + BEGV_BYTE; | |
1218 | pos = BYTE_TO_CHAR (search_regs_1.end[0] + BEGV_BYTE); | |
1219 | } | |
ca1d1d23 JB |
1220 | } |
1221 | else | |
1222 | { | |
1223 | immediate_quit = 0; | |
1224 | return (0 - n); | |
1225 | } | |
1226 | n--; | |
1227 | } | |
1228 | immediate_quit = 0; | |
1229 | return (pos); | |
1230 | } | |
1231 | else /* non-RE case */ | |
1232 | { | |
facdc750 RS |
1233 | unsigned char *raw_pattern, *pat; |
1234 | int raw_pattern_size; | |
1235 | int raw_pattern_size_byte; | |
1236 | unsigned char *patbuf; | |
1237 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); | |
a967ed62 | 1238 | unsigned char *base_pat; |
49e44e81 KH |
1239 | /* Set to positive if we find a non-ASCII char that need |
1240 | translation. Otherwise set to zero later. */ | |
1241 | int char_base = -1; | |
040272ce | 1242 | int boyer_moore_ok = 1; |
facdc750 RS |
1243 | |
1244 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1245 | We must convert PATTERN to match that, or we will not really | |
1246 | find things right. */ | |
1247 | ||
1248 | if (multibyte == STRING_MULTIBYTE (string)) | |
1249 | { | |
d5db4077 KR |
1250 | raw_pattern = (unsigned char *) SDATA (string); |
1251 | raw_pattern_size = SCHARS (string); | |
1252 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1253 | } |
1254 | else if (multibyte) | |
1255 | { | |
d5db4077 | 1256 | raw_pattern_size = SCHARS (string); |
facdc750 | 1257 | raw_pattern_size_byte |
d5db4077 | 1258 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1259 | raw_pattern_size); |
7276d3d8 | 1260 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1261 | copy_text (SDATA (string), raw_pattern, |
1262 | SCHARS (string), 0, 1); | |
facdc750 RS |
1263 | } |
1264 | else | |
1265 | { | |
1266 | /* Converting multibyte to single-byte. | |
1267 | ||
1268 | ??? Perhaps this conversion should be done in a special way | |
1269 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1270 | so that only the multibyte chars which really correspond to | |
1271 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1272 | raw_pattern_size = SCHARS (string); |
1273 | raw_pattern_size_byte = SCHARS (string); | |
7276d3d8 | 1274 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
1275 | copy_text (SDATA (string), raw_pattern, |
1276 | SBYTES (string), 1, 0); | |
facdc750 RS |
1277 | } |
1278 | ||
1279 | /* Copy and optionally translate the pattern. */ | |
1280 | len = raw_pattern_size; | |
1281 | len_byte = raw_pattern_size_byte; | |
f5f7578b | 1282 | patbuf = (unsigned char *) alloca (len * MAX_MULTIBYTE_LENGTH); |
facdc750 RS |
1283 | pat = patbuf; |
1284 | base_pat = raw_pattern; | |
1285 | if (multibyte) | |
1286 | { | |
a17ae96a KH |
1287 | /* Fill patbuf by translated characters in STRING while |
1288 | checking if we can use boyer-moore search. If TRT is | |
1289 | non-nil, we can use boyer-moore search only if TRT can be | |
1290 | represented by the byte array of 256 elements. For that, | |
1291 | all non-ASCII case-equivalents of all case-senstive | |
1292 | characters in STRING must belong to the same charset and | |
1293 | row. */ | |
1294 | ||
facdc750 RS |
1295 | while (--len >= 0) |
1296 | { | |
a17ae96a | 1297 | unsigned char str_base[MAX_MULTIBYTE_LENGTH], *str; |
aff2ce94 | 1298 | int c, translated, inverse; |
a17ae96a | 1299 | int in_charlen, charlen; |
facdc750 RS |
1300 | |
1301 | /* If we got here and the RE flag is set, it's because we're | |
1302 | dealing with a regexp known to be trivial, so the backslash | |
1303 | just quotes the next character. */ | |
1304 | if (RE && *base_pat == '\\') | |
1305 | { | |
1306 | len--; | |
62221f22 | 1307 | raw_pattern_size--; |
facdc750 RS |
1308 | len_byte--; |
1309 | base_pat++; | |
1310 | } | |
1311 | ||
1312 | c = STRING_CHAR_AND_LENGTH (base_pat, len_byte, in_charlen); | |
040272ce | 1313 | |
a17ae96a | 1314 | if (NILP (trt)) |
facdc750 | 1315 | { |
a17ae96a KH |
1316 | str = base_pat; |
1317 | charlen = in_charlen; | |
1318 | } | |
1319 | else | |
1320 | { | |
1321 | /* Translate the character. */ | |
1322 | TRANSLATE (translated, trt, c); | |
1323 | charlen = CHAR_STRING (translated, str_base); | |
1324 | str = str_base; | |
1325 | ||
1326 | /* Check if C has any other case-equivalents. */ | |
1327 | TRANSLATE (inverse, inverse_trt, c); | |
1328 | /* If so, check if we can use boyer-moore. */ | |
1329 | if (c != inverse && boyer_moore_ok) | |
1330 | { | |
1331 | /* Check if all equivalents belong to the same | |
1332 | group of characters. Note that the check of C | |
49e44e81 KH |
1333 | itself is done by the last iteration. */ |
1334 | int this_char_base = -1; | |
1335 | ||
1336 | while (boyer_moore_ok) | |
a17ae96a | 1337 | { |
49e44e81 | 1338 | if (ASCII_BYTE_P (inverse)) |
a17ae96a | 1339 | { |
49e44e81 KH |
1340 | if (this_char_base > 0) |
1341 | boyer_moore_ok = 0; | |
1342 | else | |
dafaabd1 | 1343 | this_char_base = 0; |
a17ae96a | 1344 | } |
49e44e81 KH |
1345 | else if (CHAR_BYTE8_P (inverse)) |
1346 | /* Boyer-moore search can't handle a | |
1347 | translation of an eight-bit | |
1348 | character. */ | |
1349 | boyer_moore_ok = 0; | |
1350 | else if (this_char_base < 0) | |
1351 | { | |
1352 | this_char_base = inverse & ~0x3F; | |
1353 | if (char_base < 0) | |
1354 | char_base = this_char_base; | |
dafaabd1 | 1355 | else if (this_char_base != char_base) |
49e44e81 KH |
1356 | boyer_moore_ok = 0; |
1357 | } | |
1358 | else if ((inverse & ~0x3F) != this_char_base) | |
1359 | boyer_moore_ok = 0; | |
a17ae96a KH |
1360 | if (c == inverse) |
1361 | break; | |
1362 | TRANSLATE (inverse, inverse_trt, inverse); | |
1363 | } | |
1364 | } | |
aff2ce94 | 1365 | } |
facdc750 RS |
1366 | |
1367 | /* Store this character into the translated pattern. */ | |
a17ae96a KH |
1368 | bcopy (str, pat, charlen); |
1369 | pat += charlen; | |
facdc750 RS |
1370 | base_pat += in_charlen; |
1371 | len_byte -= in_charlen; | |
1372 | } | |
dafaabd1 AS |
1373 | |
1374 | /* If char_base is still negative we didn't find any translated | |
1375 | non-ASCII characters. */ | |
1376 | if (char_base < 0) | |
1377 | char_base = 0; | |
facdc750 RS |
1378 | } |
1379 | else | |
1380 | { | |
040272ce | 1381 | /* Unibyte buffer. */ |
a17ae96a | 1382 | char_base = 0; |
facdc750 RS |
1383 | while (--len >= 0) |
1384 | { | |
040272ce | 1385 | int c, translated; |
facdc750 RS |
1386 | |
1387 | /* If we got here and the RE flag is set, it's because we're | |
1388 | dealing with a regexp known to be trivial, so the backslash | |
1389 | just quotes the next character. */ | |
1390 | if (RE && *base_pat == '\\') | |
1391 | { | |
1392 | len--; | |
0190922f | 1393 | raw_pattern_size--; |
facdc750 RS |
1394 | base_pat++; |
1395 | } | |
1396 | c = *base_pat++; | |
aff2ce94 | 1397 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1398 | *pat++ = translated; |
1399 | } | |
1400 | } | |
1401 | ||
1402 | len_byte = pat - patbuf; | |
1403 | len = raw_pattern_size; | |
1404 | pat = base_pat = patbuf; | |
1405 | ||
040272ce | 1406 | if (boyer_moore_ok) |
facdc750 | 1407 | return boyer_moore (n, pat, len, len_byte, trt, inverse_trt, |
aff2ce94 | 1408 | pos, pos_byte, lim, lim_byte, |
a17ae96a | 1409 | char_base); |
facdc750 RS |
1410 | else |
1411 | return simple_search (n, pat, len, len_byte, trt, | |
1412 | pos, pos_byte, lim, lim_byte); | |
1413 | } | |
1414 | } | |
1415 | \f | |
1416 | /* Do a simple string search N times for the string PAT, | |
1417 | whose length is LEN/LEN_BYTE, | |
1418 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1419 | TRT is the translation table. | |
f8bd51c4 | 1420 | |
facdc750 RS |
1421 | Return the character position where the match is found. |
1422 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1423 | |
facdc750 RS |
1424 | This kind of search works regardless of what is in PAT and |
1425 | regardless of what is in TRT. It is used in cases where | |
1426 | boyer_moore cannot work. */ | |
1427 | ||
1428 | static int | |
1429 | simple_search (n, pat, len, len_byte, trt, pos, pos_byte, lim, lim_byte) | |
1430 | int n; | |
1431 | unsigned char *pat; | |
1432 | int len, len_byte; | |
1433 | Lisp_Object trt; | |
1434 | int pos, pos_byte; | |
1435 | int lim, lim_byte; | |
1436 | { | |
1437 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); | |
ab228c24 | 1438 | int forward = n > 0; |
49e44e81 KH |
1439 | /* Number of buffer bytes matched. Note that this may be different |
1440 | from len_byte in a multibyte buffer. */ | |
1441 | int match_byte; | |
facdc750 RS |
1442 | |
1443 | if (lim > pos && multibyte) | |
1444 | while (n > 0) | |
1445 | { | |
1446 | while (1) | |
f8bd51c4 | 1447 | { |
facdc750 RS |
1448 | /* Try matching at position POS. */ |
1449 | int this_pos = pos; | |
1450 | int this_pos_byte = pos_byte; | |
1451 | int this_len = len; | |
1452 | int this_len_byte = len_byte; | |
1453 | unsigned char *p = pat; | |
a9469498 | 1454 | if (pos + len > lim || pos_byte + len_byte > lim_byte) |
facdc750 RS |
1455 | goto stop; |
1456 | ||
1457 | while (this_len > 0) | |
1458 | { | |
1459 | int charlen, buf_charlen; | |
ab228c24 | 1460 | int pat_ch, buf_ch; |
facdc750 | 1461 | |
ab228c24 | 1462 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1463 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1464 | ZV_BYTE - this_pos_byte, | |
1465 | buf_charlen); | |
aff2ce94 | 1466 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1467 | |
1468 | if (buf_ch != pat_ch) | |
1469 | break; | |
ab228c24 RS |
1470 | |
1471 | this_len_byte -= charlen; | |
1472 | this_len--; | |
1473 | p += charlen; | |
1474 | ||
1475 | this_pos_byte += buf_charlen; | |
1476 | this_pos++; | |
facdc750 RS |
1477 | } |
1478 | ||
1479 | if (this_len == 0) | |
1480 | { | |
49e44e81 | 1481 | match_byte = this_pos_byte - pos_byte; |
facdc750 | 1482 | pos += len; |
49e44e81 | 1483 | pos_byte += match_byte; |
facdc750 RS |
1484 | break; |
1485 | } | |
1486 | ||
1487 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1488 | } |
facdc750 RS |
1489 | |
1490 | n--; | |
1491 | } | |
1492 | else if (lim > pos) | |
1493 | while (n > 0) | |
1494 | { | |
1495 | while (1) | |
f8bd51c4 | 1496 | { |
facdc750 RS |
1497 | /* Try matching at position POS. */ |
1498 | int this_pos = pos; | |
1499 | int this_len = len; | |
1500 | unsigned char *p = pat; | |
1501 | ||
1502 | if (pos + len > lim) | |
1503 | goto stop; | |
1504 | ||
1505 | while (this_len > 0) | |
1506 | { | |
1507 | int pat_ch = *p++; | |
1508 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1509 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1510 | |
1511 | if (buf_ch != pat_ch) | |
1512 | break; | |
ab228c24 RS |
1513 | |
1514 | this_len--; | |
1515 | this_pos++; | |
facdc750 RS |
1516 | } |
1517 | ||
1518 | if (this_len == 0) | |
1519 | { | |
49e44e81 | 1520 | match_byte = len; |
facdc750 RS |
1521 | pos += len; |
1522 | break; | |
1523 | } | |
1524 | ||
1525 | pos++; | |
f8bd51c4 | 1526 | } |
facdc750 RS |
1527 | |
1528 | n--; | |
1529 | } | |
1530 | /* Backwards search. */ | |
1531 | else if (lim < pos && multibyte) | |
1532 | while (n < 0) | |
1533 | { | |
1534 | while (1) | |
f8bd51c4 | 1535 | { |
facdc750 RS |
1536 | /* Try matching at position POS. */ |
1537 | int this_pos = pos - len; | |
3dcf0066 | 1538 | int this_pos_byte; |
facdc750 RS |
1539 | int this_len = len; |
1540 | int this_len_byte = len_byte; | |
1541 | unsigned char *p = pat; | |
1542 | ||
3dcf0066 | 1543 | if (this_pos < lim || (pos_byte - len_byte) < lim_byte) |
facdc750 | 1544 | goto stop; |
3dcf0066 | 1545 | this_pos_byte = CHAR_TO_BYTE (this_pos); |
49e44e81 | 1546 | match_byte = pos_byte - this_pos_byte; |
facdc750 RS |
1547 | |
1548 | while (this_len > 0) | |
1549 | { | |
1550 | int charlen, buf_charlen; | |
ab228c24 | 1551 | int pat_ch, buf_ch; |
facdc750 | 1552 | |
ab228c24 | 1553 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1554 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1555 | ZV_BYTE - this_pos_byte, | |
1556 | buf_charlen); | |
aff2ce94 | 1557 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1558 | |
1559 | if (buf_ch != pat_ch) | |
1560 | break; | |
ab228c24 RS |
1561 | |
1562 | this_len_byte -= charlen; | |
1563 | this_len--; | |
1564 | p += charlen; | |
1565 | this_pos_byte += buf_charlen; | |
1566 | this_pos++; | |
facdc750 RS |
1567 | } |
1568 | ||
1569 | if (this_len == 0) | |
1570 | { | |
1571 | pos -= len; | |
49e44e81 | 1572 | pos_byte -= match_byte; |
facdc750 RS |
1573 | break; |
1574 | } | |
1575 | ||
1576 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1577 | } |
1578 | ||
facdc750 RS |
1579 | n++; |
1580 | } | |
1581 | else if (lim < pos) | |
1582 | while (n < 0) | |
1583 | { | |
1584 | while (1) | |
b6d6a51c | 1585 | { |
facdc750 RS |
1586 | /* Try matching at position POS. */ |
1587 | int this_pos = pos - len; | |
1588 | int this_len = len; | |
1589 | unsigned char *p = pat; | |
1590 | ||
a9469498 | 1591 | if (this_pos < lim) |
facdc750 RS |
1592 | goto stop; |
1593 | ||
1594 | while (this_len > 0) | |
1595 | { | |
1596 | int pat_ch = *p++; | |
1597 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1598 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1599 | |
1600 | if (buf_ch != pat_ch) | |
1601 | break; | |
ab228c24 RS |
1602 | this_len--; |
1603 | this_pos++; | |
facdc750 RS |
1604 | } |
1605 | ||
1606 | if (this_len == 0) | |
b6d6a51c | 1607 | { |
49e44e81 | 1608 | match_byte = len; |
facdc750 RS |
1609 | pos -= len; |
1610 | break; | |
b6d6a51c | 1611 | } |
facdc750 RS |
1612 | |
1613 | pos--; | |
b6d6a51c | 1614 | } |
facdc750 RS |
1615 | |
1616 | n++; | |
b6d6a51c | 1617 | } |
facdc750 RS |
1618 | |
1619 | stop: | |
1620 | if (n == 0) | |
aff2ce94 | 1621 | { |
ab228c24 | 1622 | if (forward) |
49e44e81 | 1623 | set_search_regs ((multibyte ? pos_byte : pos) - match_byte, match_byte); |
ab228c24 | 1624 | else |
49e44e81 | 1625 | set_search_regs (multibyte ? pos_byte : pos, match_byte); |
aff2ce94 RS |
1626 | |
1627 | return pos; | |
1628 | } | |
facdc750 RS |
1629 | else if (n > 0) |
1630 | return -n; | |
1631 | else | |
1632 | return n; | |
1633 | } | |
1634 | \f | |
0190922f | 1635 | /* Do Boyer-Moore search N times for the string BASE_PAT, |
facdc750 RS |
1636 | whose length is LEN/LEN_BYTE, |
1637 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1638 | DIRECTION says which direction we search in. | |
1639 | TRT and INVERSE_TRT are translation tables. | |
0190922f | 1640 | Characters in PAT are already translated by TRT. |
facdc750 | 1641 | |
0190922f KH |
1642 | This kind of search works if all the characters in BASE_PAT that |
1643 | have nontrivial translation are the same aside from the last byte. | |
1644 | This makes it possible to translate just the last byte of a | |
1645 | character, and do so after just a simple test of the context. | |
b2e6b10f | 1646 | CHAR_BASE is nonzero if there is such a non-ASCII character. |
facdc750 RS |
1647 | |
1648 | If that criterion is not satisfied, do not call this function. */ | |
1649 | ||
1650 | static int | |
1651 | boyer_moore (n, base_pat, len, len_byte, trt, inverse_trt, | |
a17ae96a | 1652 | pos, pos_byte, lim, lim_byte, char_base) |
facdc750 RS |
1653 | int n; |
1654 | unsigned char *base_pat; | |
1655 | int len, len_byte; | |
1656 | Lisp_Object trt; | |
1657 | Lisp_Object inverse_trt; | |
1658 | int pos, pos_byte; | |
1659 | int lim, lim_byte; | |
a17ae96a | 1660 | int char_base; |
facdc750 RS |
1661 | { |
1662 | int direction = ((n > 0) ? 1 : -1); | |
1663 | register int dirlen; | |
a968f437 | 1664 | int infinity, limit, stride_for_teases = 0; |
facdc750 RS |
1665 | register int *BM_tab; |
1666 | int *BM_tab_base; | |
177c0ea7 | 1667 | register unsigned char *cursor, *p_limit; |
facdc750 | 1668 | register int i, j; |
cb6792d2 | 1669 | unsigned char *pat, *pat_end; |
facdc750 RS |
1670 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); |
1671 | ||
1672 | unsigned char simple_translate[0400]; | |
0190922f | 1673 | /* These are set to the preceding bytes of a byte to be translated |
49e44e81 | 1674 | if char_base is nonzero. As the maximum byte length of a |
a17ae96a | 1675 | multibyte character is 5, we have to check at most four previous |
0190922f KH |
1676 | bytes. */ |
1677 | int translate_prev_byte1 = 0; | |
1678 | int translate_prev_byte2 = 0; | |
1679 | int translate_prev_byte3 = 0; | |
a17ae96a | 1680 | int translate_prev_byte4 = 0; |
facdc750 | 1681 | |
facdc750 | 1682 | BM_tab = (int *) alloca (0400 * sizeof (int)); |
e2a51aa8 | 1683 | |
facdc750 RS |
1684 | /* The general approach is that we are going to maintain that we know */ |
1685 | /* the first (closest to the present position, in whatever direction */ | |
1686 | /* we're searching) character that could possibly be the last */ | |
1687 | /* (furthest from present position) character of a valid match. We */ | |
1688 | /* advance the state of our knowledge by looking at that character */ | |
1689 | /* and seeing whether it indeed matches the last character of the */ | |
1690 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1691 | /* move our pointer (to putative last characters) as far as is */ | |
1692 | /* logically possible. This amount of movement, which I call a */ | |
1693 | /* stride, will be the length of the pattern if the actual character */ | |
1694 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1695 | /* from the last occurrence of that character to the end of the */ | |
1696 | /* pattern. */ | |
1697 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1698 | /* characters that match the last character. This allows use of only */ | |
1699 | /* a single test, a test for having gone past the end of the */ | |
1700 | /* permissible match region, to test for both possible matches (when */ | |
1701 | /* the stride goes past the end immediately) and failure to */ | |
177c0ea7 | 1702 | /* match (where you get nudged past the end one stride at a time). */ |
facdc750 RS |
1703 | |
1704 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1705 | /* is determined only by the last character of the putative match. */ | |
1706 | /* If that character does not match, we will stride the proper */ | |
1707 | /* distance to propose a match that superimposes it on the last */ | |
1708 | /* instance of a character that matches it (per trt), or misses */ | |
177c0ea7 | 1709 | /* it entirely if there is none. */ |
facdc750 RS |
1710 | |
1711 | dirlen = len_byte * direction; | |
1712 | infinity = dirlen - (lim_byte + pos_byte + len_byte + len_byte) * direction; | |
cb6792d2 RS |
1713 | |
1714 | /* Record position after the end of the pattern. */ | |
1715 | pat_end = base_pat + len_byte; | |
1716 | /* BASE_PAT points to a character that we start scanning from. | |
1717 | It is the first character in a forward search, | |
1718 | the last character in a backward search. */ | |
facdc750 | 1719 | if (direction < 0) |
cb6792d2 RS |
1720 | base_pat = pat_end - 1; |
1721 | ||
facdc750 RS |
1722 | BM_tab_base = BM_tab; |
1723 | BM_tab += 0400; | |
1724 | j = dirlen; /* to get it in a register */ | |
1725 | /* A character that does not appear in the pattern induces a */ | |
1726 | /* stride equal to the pattern length. */ | |
1727 | while (BM_tab_base != BM_tab) | |
1728 | { | |
1729 | *--BM_tab = j; | |
1730 | *--BM_tab = j; | |
1731 | *--BM_tab = j; | |
1732 | *--BM_tab = j; | |
1733 | } | |
1734 | ||
1735 | /* We use this for translation, instead of TRT itself. | |
1736 | We fill this in to handle the characters that actually | |
1737 | occur in the pattern. Others don't matter anyway! */ | |
1738 | bzero (simple_translate, sizeof simple_translate); | |
1739 | for (i = 0; i < 0400; i++) | |
1740 | simple_translate[i] = i; | |
1741 | ||
a17ae96a | 1742 | if (char_base) |
0190922f | 1743 | { |
a17ae96a | 1744 | /* Setup translate_prev_byte1/2/3/4 from CHAR_BASE. Only a |
0190922f | 1745 | byte following them are the target of translation. */ |
0190922f | 1746 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
a17ae96a | 1747 | int len = CHAR_STRING (char_base, str); |
0190922f KH |
1748 | |
1749 | translate_prev_byte1 = str[len - 2]; | |
1750 | if (len > 2) | |
1751 | { | |
1752 | translate_prev_byte2 = str[len - 3]; | |
1753 | if (len > 3) | |
a17ae96a KH |
1754 | { |
1755 | translate_prev_byte3 = str[len - 4]; | |
1756 | if (len > 4) | |
1757 | translate_prev_byte4 = str[len - 5]; | |
1758 | } | |
0190922f KH |
1759 | } |
1760 | } | |
1761 | ||
facdc750 RS |
1762 | i = 0; |
1763 | while (i != infinity) | |
1764 | { | |
cb6792d2 | 1765 | unsigned char *ptr = base_pat + i; |
facdc750 RS |
1766 | i += direction; |
1767 | if (i == dirlen) | |
1768 | i = infinity; | |
1769 | if (! NILP (trt)) | |
ca1d1d23 | 1770 | { |
49e44e81 KH |
1771 | /* If the byte currently looking at is the last of a |
1772 | character to check case-equivalents, set CH to that | |
1773 | character. An ASCII character and a non-ASCII character | |
1774 | matching with CHAR_BASE are to be checked. */ | |
a17ae96a KH |
1775 | int ch = -1; |
1776 | ||
1777 | if (ASCII_BYTE_P (*ptr) || ! multibyte) | |
1778 | ch = *ptr; | |
49e44e81 | 1779 | else if (char_base |
86dc6ccb | 1780 | && ((pat_end - ptr) == 1 || CHAR_HEAD_P (ptr[1]))) |
ca1d1d23 | 1781 | { |
49e44e81 KH |
1782 | unsigned char *charstart = ptr - 1; |
1783 | ||
1784 | while (! (CHAR_HEAD_P (*charstart))) | |
1785 | charstart--; | |
1786 | ch = STRING_CHAR (charstart, ptr - charstart + 1); | |
a17ae96a KH |
1787 | if (char_base != (ch & ~0x3F)) |
1788 | ch = -1; | |
ca1d1d23 | 1789 | } |
facdc750 | 1790 | |
f51995fa | 1791 | if (ch >= 0200) |
49e44e81 KH |
1792 | j = (ch & 0x3F) | 0200; |
1793 | else | |
1794 | j = *ptr; | |
1795 | ||
facdc750 RS |
1796 | if (i == infinity) |
1797 | stride_for_teases = BM_tab[j]; | |
ab228c24 | 1798 | |
facdc750 RS |
1799 | BM_tab[j] = dirlen - i; |
1800 | /* A translation table is accompanied by its inverse -- see */ | |
177c0ea7 | 1801 | /* comment following downcase_table for details */ |
a17ae96a | 1802 | if (ch >= 0) |
ab228c24 RS |
1803 | { |
1804 | int starting_ch = ch; | |
49e44e81 | 1805 | int starting_j = j; |
a17ae96a | 1806 | |
ab228c24 RS |
1807 | while (1) |
1808 | { | |
1809 | TRANSLATE (ch, inverse_trt, ch); | |
f51995fa | 1810 | if (ch >= 0200) |
49e44e81 | 1811 | j = (ch & 0x3F) | 0200; |
ab228c24 | 1812 | else |
a17ae96a | 1813 | j = ch; |
ab228c24 RS |
1814 | |
1815 | /* For all the characters that map into CH, | |
1816 | set up simple_translate to map the last byte | |
1817 | into STARTING_J. */ | |
1818 | simple_translate[j] = starting_j; | |
1819 | if (ch == starting_ch) | |
1820 | break; | |
1821 | BM_tab[j] = dirlen - i; | |
1822 | } | |
1823 | } | |
facdc750 RS |
1824 | } |
1825 | else | |
1826 | { | |
1827 | j = *ptr; | |
1828 | ||
1829 | if (i == infinity) | |
1830 | stride_for_teases = BM_tab[j]; | |
1831 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1832 | } |
facdc750 RS |
1833 | /* stride_for_teases tells how much to stride if we get a */ |
1834 | /* match on the far character but are subsequently */ | |
1835 | /* disappointed, by recording what the stride would have been */ | |
1836 | /* for that character if the last character had been */ | |
1837 | /* different. */ | |
1838 | } | |
1839 | infinity = dirlen - infinity; | |
1840 | pos_byte += dirlen - ((direction > 0) ? direction : 0); | |
1841 | /* loop invariant - POS_BYTE points at where last char (first | |
1842 | char if reverse) of pattern would align in a possible match. */ | |
1843 | while (n != 0) | |
1844 | { | |
1845 | int tail_end; | |
1846 | unsigned char *tail_end_ptr; | |
1847 | ||
1848 | /* It's been reported that some (broken) compiler thinks that | |
1849 | Boolean expressions in an arithmetic context are unsigned. | |
1850 | Using an explicit ?1:0 prevents this. */ | |
1851 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1852 | < 0) | |
1853 | return (n * (0 - direction)); | |
1854 | /* First we do the part we can by pointers (maybe nothing) */ | |
1855 | QUIT; | |
1856 | pat = base_pat; | |
1857 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1858 | if (direction > 0) |
1859 | { | |
1860 | limit = BUFFER_CEILING_OF (limit); | |
1861 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1862 | can take on without hitting edge of buffer or the gap. */ | |
1863 | limit = min (limit, pos_byte + 20000); | |
1864 | limit = min (limit, lim_byte - 1); | |
1865 | } | |
1866 | else | |
1867 | { | |
1868 | limit = BUFFER_FLOOR_OF (limit); | |
1869 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1870 | can take on without hitting edge of buffer or the gap. */ | |
1871 | limit = max (limit, pos_byte - 20000); | |
1872 | limit = max (limit, lim_byte); | |
1873 | } | |
facdc750 RS |
1874 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1875 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1876 | ||
1877 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1878 | { |
facdc750 RS |
1879 | unsigned char *p2; |
1880 | ||
1881 | p_limit = BYTE_POS_ADDR (limit); | |
1882 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
1883 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ | |
1884 | while (1) /* use one cursor setting as long as i can */ | |
ca1d1d23 | 1885 | { |
facdc750 | 1886 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1887 | { |
facdc750 RS |
1888 | /* Use signed comparison if appropriate |
1889 | to make cursor+infinity sure to be > p_limit. | |
1890 | Assuming that the buffer lies in a range of addresses | |
1891 | that are all "positive" (as ints) or all "negative", | |
1892 | either kind of comparison will work as long | |
1893 | as we don't step by infinity. So pick the kind | |
1894 | that works when we do step by infinity. */ | |
1895 | if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit) | |
1896 | while ((EMACS_INT) cursor <= (EMACS_INT) p_limit) | |
1897 | cursor += BM_tab[*cursor]; | |
ca1d1d23 | 1898 | else |
facdc750 RS |
1899 | while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit) |
1900 | cursor += BM_tab[*cursor]; | |
1901 | } | |
1902 | else | |
1903 | { | |
1904 | if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit) | |
1905 | while ((EMACS_INT) cursor >= (EMACS_INT) p_limit) | |
1906 | cursor += BM_tab[*cursor]; | |
1907 | else | |
1908 | while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit) | |
1909 | cursor += BM_tab[*cursor]; | |
1910 | } | |
ca1d1d23 | 1911 | /* If you are here, cursor is beyond the end of the searched region. */ |
facdc750 RS |
1912 | /* This can happen if you match on the far character of the pattern, */ |
1913 | /* because the "stride" of that character is infinity, a number able */ | |
1914 | /* to throw you well beyond the end of the search. It can also */ | |
1915 | /* happen if you fail to match within the permitted region and would */ | |
1916 | /* otherwise try a character beyond that region */ | |
1917 | if ((cursor - p_limit) * direction <= len_byte) | |
1918 | break; /* a small overrun is genuine */ | |
1919 | cursor -= infinity; /* large overrun = hit */ | |
1920 | i = dirlen - direction; | |
1921 | if (! NILP (trt)) | |
1922 | { | |
1923 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1924 | { |
facdc750 RS |
1925 | int ch; |
1926 | cursor -= direction; | |
1927 | /* Translate only the last byte of a character. */ | |
1928 | if (! multibyte | |
1929 | || ((cursor == tail_end_ptr | |
1930 | || CHAR_HEAD_P (cursor[1])) | |
1931 | && (CHAR_HEAD_P (cursor[0]) | |
0190922f KH |
1932 | /* Check if this is the last byte of |
1933 | a translable character. */ | |
1934 | || (translate_prev_byte1 == cursor[-1] | |
1935 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1936 | || (translate_prev_byte2 == cursor[-2] | |
1937 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1938 | || (translate_prev_byte3 == cursor[-3])))))))) | |
facdc750 RS |
1939 | ch = simple_translate[*cursor]; |
1940 | else | |
1941 | ch = *cursor; | |
1942 | if (pat[i] != ch) | |
1943 | break; | |
ca1d1d23 | 1944 | } |
facdc750 RS |
1945 | } |
1946 | else | |
1947 | { | |
1948 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1949 | { |
facdc750 RS |
1950 | cursor -= direction; |
1951 | if (pat[i] != *cursor) | |
1952 | break; | |
ca1d1d23 | 1953 | } |
facdc750 RS |
1954 | } |
1955 | cursor += dirlen - i - direction; /* fix cursor */ | |
1956 | if (i + direction == 0) | |
1957 | { | |
ef887810 | 1958 | int position, start, end; |
0c8533c6 | 1959 | |
facdc750 | 1960 | cursor -= direction; |
1113d9db | 1961 | |
facdc750 RS |
1962 | position = pos_byte + cursor - p2 + ((direction > 0) |
1963 | ? 1 - len_byte : 0); | |
1964 | set_search_regs (position, len_byte); | |
ca325161 | 1965 | |
ef887810 RS |
1966 | if (NILP (Vinhibit_changing_match_data)) |
1967 | { | |
1968 | start = search_regs.start[0]; | |
1969 | end = search_regs.end[0]; | |
1970 | } | |
1971 | else | |
1972 | /* If Vinhibit_changing_match_data is non-nil, | |
1973 | search_regs will not be changed. So let's | |
1974 | compute start and end here. */ | |
1975 | { | |
1976 | start = BYTE_TO_CHAR (position); | |
1977 | end = BYTE_TO_CHAR (position + len_byte); | |
1978 | } | |
1979 | ||
facdc750 RS |
1980 | if ((n -= direction) != 0) |
1981 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1982 | else |
ef887810 | 1983 | return direction > 0 ? end : start; |
ca1d1d23 | 1984 | } |
facdc750 RS |
1985 | else |
1986 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1987 | } |
facdc750 RS |
1988 | pos_byte += cursor - p2; |
1989 | } | |
1990 | else | |
1991 | /* Now we'll pick up a clump that has to be done the hard */ | |
1992 | /* way because it covers a discontinuity */ | |
1993 | { | |
1994 | limit = ((direction > 0) | |
1995 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1996 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1997 | limit = ((direction > 0) | |
1998 | ? min (limit + len_byte, lim_byte - 1) | |
1999 | : max (limit - len_byte, lim_byte)); | |
2000 | /* LIMIT is now the last value POS_BYTE can have | |
2001 | and still be valid for a possible match. */ | |
2002 | while (1) | |
ca1d1d23 | 2003 | { |
facdc750 RS |
2004 | /* This loop can be coded for space rather than */ |
2005 | /* speed because it will usually run only once. */ | |
2006 | /* (the reach is at most len + 21, and typically */ | |
177c0ea7 | 2007 | /* does not exceed len) */ |
facdc750 RS |
2008 | while ((limit - pos_byte) * direction >= 0) |
2009 | pos_byte += BM_tab[FETCH_BYTE (pos_byte)]; | |
2010 | /* now run the same tests to distinguish going off the */ | |
2011 | /* end, a match or a phony match. */ | |
2012 | if ((pos_byte - limit) * direction <= len_byte) | |
2013 | break; /* ran off the end */ | |
2014 | /* Found what might be a match. | |
2015 | Set POS_BYTE back to last (first if reverse) pos. */ | |
2016 | pos_byte -= infinity; | |
2017 | i = dirlen - direction; | |
2018 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 2019 | { |
facdc750 RS |
2020 | int ch; |
2021 | unsigned char *ptr; | |
2022 | pos_byte -= direction; | |
2023 | ptr = BYTE_POS_ADDR (pos_byte); | |
2024 | /* Translate only the last byte of a character. */ | |
2025 | if (! multibyte | |
2026 | || ((ptr == tail_end_ptr | |
2027 | || CHAR_HEAD_P (ptr[1])) | |
2028 | && (CHAR_HEAD_P (ptr[0]) | |
0190922f KH |
2029 | /* Check if this is the last byte of a |
2030 | translable character. */ | |
2031 | || (translate_prev_byte1 == ptr[-1] | |
2032 | && (CHAR_HEAD_P (translate_prev_byte1) | |
2033 | || (translate_prev_byte2 == ptr[-2] | |
2034 | && (CHAR_HEAD_P (translate_prev_byte2) | |
2035 | || translate_prev_byte3 == ptr[-3]))))))) | |
facdc750 RS |
2036 | ch = simple_translate[*ptr]; |
2037 | else | |
2038 | ch = *ptr; | |
2039 | if (pat[i] != ch) | |
2040 | break; | |
2041 | } | |
2042 | /* Above loop has moved POS_BYTE part or all the way | |
2043 | back to the first pos (last pos if reverse). | |
2044 | Set it once again at the last (first if reverse) char. */ | |
2045 | pos_byte += dirlen - i- direction; | |
2046 | if (i + direction == 0) | |
2047 | { | |
ef887810 | 2048 | int position, start, end; |
facdc750 | 2049 | pos_byte -= direction; |
1113d9db | 2050 | |
facdc750 | 2051 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
facdc750 | 2052 | set_search_regs (position, len_byte); |
ca325161 | 2053 | |
ef887810 RS |
2054 | if (NILP (Vinhibit_changing_match_data)) |
2055 | { | |
2056 | start = search_regs.start[0]; | |
2057 | end = search_regs.end[0]; | |
2058 | } | |
2059 | else | |
2060 | /* If Vinhibit_changing_match_data is non-nil, | |
2061 | search_regs will not be changed. So let's | |
2062 | compute start and end here. */ | |
2063 | { | |
2064 | start = BYTE_TO_CHAR (position); | |
2065 | end = BYTE_TO_CHAR (position + len_byte); | |
2066 | } | |
2067 | ||
facdc750 RS |
2068 | if ((n -= direction) != 0) |
2069 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 2070 | else |
ef887810 | 2071 | return direction > 0 ? end : start; |
ca1d1d23 | 2072 | } |
facdc750 RS |
2073 | else |
2074 | pos_byte += stride_for_teases; | |
2075 | } | |
2076 | } | |
2077 | /* We have done one clump. Can we continue? */ | |
2078 | if ((lim_byte - pos_byte) * direction < 0) | |
2079 | return ((0 - n) * direction); | |
ca1d1d23 | 2080 | } |
facdc750 | 2081 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 2082 | } |
ca325161 | 2083 | |
fa8ed3e0 | 2084 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
2085 | for the overall match just found in the current buffer. |
2086 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
2087 | |
2088 | static void | |
fa8ed3e0 RS |
2089 | set_search_regs (beg_byte, nbytes) |
2090 | int beg_byte, nbytes; | |
ca325161 | 2091 | { |
a7e4cdde RS |
2092 | int i; |
2093 | ||
ef887810 RS |
2094 | if (!NILP (Vinhibit_changing_match_data)) |
2095 | return; | |
2096 | ||
ca325161 RS |
2097 | /* Make sure we have registers in which to store |
2098 | the match position. */ | |
2099 | if (search_regs.num_regs == 0) | |
2100 | { | |
2d4a771a RS |
2101 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
2102 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 2103 | search_regs.num_regs = 2; |
ca325161 RS |
2104 | } |
2105 | ||
a7e4cdde RS |
2106 | /* Clear out the other registers. */ |
2107 | for (i = 1; i < search_regs.num_regs; i++) | |
2108 | { | |
2109 | search_regs.start[i] = -1; | |
2110 | search_regs.end[i] = -1; | |
2111 | } | |
2112 | ||
fa8ed3e0 RS |
2113 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
2114 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 2115 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 2116 | } |
ca1d1d23 | 2117 | \f |
c6e0e1cc CY |
2118 | /* Given STRING, a string of words separated by word delimiters, |
2119 | compute a regexp that matches those exact words separated by | |
2120 | arbitrary punctuation. If LAX is nonzero, the end of the string | |
2121 | need not match a word boundary unless it ends in whitespace. */ | |
ca1d1d23 JB |
2122 | |
2123 | static Lisp_Object | |
c6e0e1cc | 2124 | wordify (string, lax) |
ca1d1d23 | 2125 | Lisp_Object string; |
c6e0e1cc | 2126 | int lax; |
ca1d1d23 JB |
2127 | { |
2128 | register unsigned char *p, *o; | |
0c8533c6 | 2129 | register int i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 2130 | Lisp_Object val; |
0c8533c6 | 2131 | int prev_c = 0; |
c6e0e1cc | 2132 | int adjust, whitespace_at_end; |
ca1d1d23 | 2133 | |
b7826503 | 2134 | CHECK_STRING (string); |
d5db4077 KR |
2135 | p = SDATA (string); |
2136 | len = SCHARS (string); | |
ca1d1d23 | 2137 | |
0c8533c6 RS |
2138 | for (i = 0, i_byte = 0; i < len; ) |
2139 | { | |
2140 | int c; | |
177c0ea7 | 2141 | |
93daa011 | 2142 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
0c8533c6 RS |
2143 | |
2144 | if (SYNTAX (c) != Sword) | |
2145 | { | |
2146 | punct_count++; | |
2147 | if (i > 0 && SYNTAX (prev_c) == Sword) | |
2148 | word_count++; | |
2149 | } | |
ca1d1d23 | 2150 | |
0c8533c6 RS |
2151 | prev_c = c; |
2152 | } | |
2153 | ||
2154 | if (SYNTAX (prev_c) == Sword) | |
c6e0e1cc CY |
2155 | { |
2156 | word_count++; | |
2157 | whitespace_at_end = 0; | |
2158 | } | |
2159 | else | |
2160 | whitespace_at_end = 1; | |
2161 | ||
0c8533c6 | 2162 | if (!word_count) |
c60416e0 | 2163 | return empty_unibyte_string; |
0c8533c6 | 2164 | |
c6e0e1cc CY |
2165 | adjust = - punct_count + 5 * (word_count - 1) |
2166 | + ((lax && !whitespace_at_end) ? 2 : 4); | |
8a2df937 RS |
2167 | if (STRING_MULTIBYTE (string)) |
2168 | val = make_uninit_multibyte_string (len + adjust, | |
d5db4077 | 2169 | SBYTES (string) |
8a2df937 RS |
2170 | + adjust); |
2171 | else | |
2172 | val = make_uninit_string (len + adjust); | |
ca1d1d23 | 2173 | |
d5db4077 | 2174 | o = SDATA (val); |
ca1d1d23 JB |
2175 | *o++ = '\\'; |
2176 | *o++ = 'b'; | |
1e9582d4 | 2177 | prev_c = 0; |
ca1d1d23 | 2178 | |
1e9582d4 RS |
2179 | for (i = 0, i_byte = 0; i < len; ) |
2180 | { | |
2181 | int c; | |
2182 | int i_byte_orig = i_byte; | |
177c0ea7 | 2183 | |
93daa011 | 2184 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
1e9582d4 RS |
2185 | |
2186 | if (SYNTAX (c) == Sword) | |
2187 | { | |
5d69fe10 | 2188 | bcopy (SDATA (string) + i_byte_orig, o, |
1e9582d4 RS |
2189 | i_byte - i_byte_orig); |
2190 | o += i_byte - i_byte_orig; | |
2191 | } | |
2192 | else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count) | |
2193 | { | |
2194 | *o++ = '\\'; | |
2195 | *o++ = 'W'; | |
2196 | *o++ = '\\'; | |
2197 | *o++ = 'W'; | |
2198 | *o++ = '*'; | |
2199 | } | |
2200 | ||
2201 | prev_c = c; | |
2202 | } | |
ca1d1d23 | 2203 | |
c6e0e1cc CY |
2204 | if (!lax || whitespace_at_end) |
2205 | { | |
2206 | *o++ = '\\'; | |
2207 | *o++ = 'b'; | |
2208 | } | |
ca1d1d23 JB |
2209 | |
2210 | return val; | |
2211 | } | |
2212 | \f | |
2213 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
8c1a1077 PJ |
2214 | "MSearch backward: ", |
2215 | doc: /* Search backward from point for STRING. | |
2216 | Set point to the beginning of the occurrence found, and return point. | |
2217 | An optional second argument bounds the search; it is a buffer position. | |
2218 | The match found must not extend before that position. | |
2219 | Optional third argument, if t, means if fail just return nil (no error). | |
2220 | If not nil and not t, position at limit of search and return nil. | |
2221 | Optional fourth argument is repeat count--search for successive occurrences. | |
2222 | ||
2223 | Search case-sensitivity is determined by the value of the variable | |
2224 | `case-fold-search', which see. | |
2225 | ||
2226 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2227 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2228 | Lisp_Object string, bound, noerror, count; |
2229 | { | |
b819a390 | 2230 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2231 | } |
2232 | ||
6af43974 | 2233 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2234 | doc: /* Search forward from point for STRING. |
2235 | Set point to the end of the occurrence found, and return point. | |
2236 | An optional second argument bounds the search; it is a buffer position. | |
d6aacdcd | 2237 | The match found must not extend after that position. A value of nil is |
48740f01 | 2238 | equivalent to (point-max). |
8c1a1077 PJ |
2239 | Optional third argument, if t, means if fail just return nil (no error). |
2240 | If not nil and not t, move to limit of search and return nil. | |
2241 | Optional fourth argument is repeat count--search for successive occurrences. | |
2242 | ||
2243 | Search case-sensitivity is determined by the value of the variable | |
2244 | `case-fold-search', which see. | |
2245 | ||
2246 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2247 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2248 | Lisp_Object string, bound, noerror, count; |
2249 | { | |
b819a390 | 2250 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2251 | } |
2252 | ||
2253 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
8c1a1077 PJ |
2254 | "sWord search backward: ", |
2255 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2256 | Set point to the beginning of the occurrence found, and return point. | |
2257 | An optional second argument bounds the search; it is a buffer position. | |
2258 | The match found must not extend before that position. | |
2259 | Optional third argument, if t, means if fail just return nil (no error). | |
2260 | If not nil and not t, move to limit of search and return nil. | |
2261 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2262 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2263 | Lisp_Object string, bound, noerror, count; |
2264 | { | |
c6e0e1cc | 2265 | return search_command (wordify (string, 0), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2266 | } |
2267 | ||
2268 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
8c1a1077 PJ |
2269 | "sWord search: ", |
2270 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2271 | Set point to the end of the occurrence found, and return point. | |
2272 | An optional second argument bounds the search; it is a buffer position. | |
2273 | The match found must not extend after that position. | |
2274 | Optional third argument, if t, means if fail just return nil (no error). | |
2275 | If not nil and not t, move to limit of search and return nil. | |
2276 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2277 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2278 | Lisp_Object string, bound, noerror, count; |
2279 | { | |
c6e0e1cc CY |
2280 | return search_command (wordify (string, 0), bound, noerror, count, 1, 1, 0); |
2281 | } | |
2282 | ||
2283 | DEFUN ("word-search-backward-lax", Fword_search_backward_lax, Sword_search_backward_lax, 1, 4, | |
2284 | "sWord search backward: ", | |
2285 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2286 | Set point to the beginning of the occurrence found, and return point. | |
2287 | ||
2288 | Unlike `word-search-backward', the end of STRING need not match a word | |
2289 | boundary unless it ends in whitespace. | |
2290 | ||
2291 | An optional second argument bounds the search; it is a buffer position. | |
2292 | The match found must not extend before that position. | |
2293 | Optional third argument, if t, means if fail just return nil (no error). | |
2294 | If not nil and not t, move to limit of search and return nil. | |
2295 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2296 | (string, bound, noerror, count) | |
2297 | Lisp_Object string, bound, noerror, count; | |
2298 | { | |
2299 | return search_command (wordify (string, 1), bound, noerror, count, -1, 1, 0); | |
2300 | } | |
2301 | ||
2302 | DEFUN ("word-search-forward-lax", Fword_search_forward_lax, Sword_search_forward_lax, 1, 4, | |
2303 | "sWord search: ", | |
2304 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2305 | Set point to the end of the occurrence found, and return point. | |
2306 | ||
2307 | Unlike `word-search-forward', the end of STRING need not match a word | |
2308 | boundary unless it ends in whitespace. | |
2309 | ||
2310 | An optional second argument bounds the search; it is a buffer position. | |
2311 | The match found must not extend after that position. | |
2312 | Optional third argument, if t, means if fail just return nil (no error). | |
2313 | If not nil and not t, move to limit of search and return nil. | |
2314 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2315 | (string, bound, noerror, count) | |
2316 | Lisp_Object string, bound, noerror, count; | |
2317 | { | |
2318 | return search_command (wordify (string, 1), bound, noerror, count, 1, 1, 0); | |
ca1d1d23 JB |
2319 | } |
2320 | ||
2321 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
8c1a1077 PJ |
2322 | "sRE search backward: ", |
2323 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2324 | Set point to the beginning of the match, and return point. | |
2325 | The match found is the one starting last in the buffer | |
2326 | and yet ending before the origin of the search. | |
2327 | An optional second argument bounds the search; it is a buffer position. | |
2328 | The match found must start at or after that position. | |
2329 | Optional third argument, if t, means if fail just return nil (no error). | |
2330 | If not nil and not t, move to limit of search and return nil. | |
2331 | Optional fourth argument is repeat count--search for successive occurrences. | |
2332 | See also the functions `match-beginning', `match-end', `match-string', | |
2333 | and `replace-match'. */) | |
2334 | (regexp, bound, noerror, count) | |
19c0a730 | 2335 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2336 | { |
b819a390 | 2337 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2338 | } |
2339 | ||
2340 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2341 | "sRE search: ", |
2342 | doc: /* Search forward from point for regular expression REGEXP. | |
2343 | Set point to the end of the occurrence found, and return point. | |
2344 | An optional second argument bounds the search; it is a buffer position. | |
2345 | The match found must not extend after that position. | |
2346 | Optional third argument, if t, means if fail just return nil (no error). | |
2347 | If not nil and not t, move to limit of search and return nil. | |
2348 | Optional fourth argument is repeat count--search for successive occurrences. | |
2349 | See also the functions `match-beginning', `match-end', `match-string', | |
2350 | and `replace-match'. */) | |
2351 | (regexp, bound, noerror, count) | |
19c0a730 | 2352 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2353 | { |
b819a390 RS |
2354 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2355 | } | |
2356 | ||
2357 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2358 | "sPosix search backward: ", |
2359 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2360 | Find the longest match in accord with Posix regular expression rules. | |
2361 | Set point to the beginning of the match, and return point. | |
2362 | The match found is the one starting last in the buffer | |
2363 | and yet ending before the origin of the search. | |
2364 | An optional second argument bounds the search; it is a buffer position. | |
2365 | The match found must start at or after that position. | |
2366 | Optional third argument, if t, means if fail just return nil (no error). | |
2367 | If not nil and not t, move to limit of search and return nil. | |
2368 | Optional fourth argument is repeat count--search for successive occurrences. | |
2369 | See also the functions `match-beginning', `match-end', `match-string', | |
2370 | and `replace-match'. */) | |
2371 | (regexp, bound, noerror, count) | |
b819a390 RS |
2372 | Lisp_Object regexp, bound, noerror, count; |
2373 | { | |
2374 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2375 | } | |
2376 | ||
2377 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2378 | "sPosix search: ", |
2379 | doc: /* Search forward from point for regular expression REGEXP. | |
2380 | Find the longest match in accord with Posix regular expression rules. | |
2381 | Set point to the end of the occurrence found, and return point. | |
2382 | An optional second argument bounds the search; it is a buffer position. | |
2383 | The match found must not extend after that position. | |
2384 | Optional third argument, if t, means if fail just return nil (no error). | |
2385 | If not nil and not t, move to limit of search and return nil. | |
2386 | Optional fourth argument is repeat count--search for successive occurrences. | |
2387 | See also the functions `match-beginning', `match-end', `match-string', | |
2388 | and `replace-match'. */) | |
2389 | (regexp, bound, noerror, count) | |
b819a390 RS |
2390 | Lisp_Object regexp, bound, noerror, count; |
2391 | { | |
2392 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2393 | } |
2394 | \f | |
d7a5ad5f | 2395 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2396 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2397 | Leave point at the end of the replacement text. |
2398 | ||
8c1a1077 PJ |
2399 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. |
2400 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2401 | based on the replaced text. | |
2402 | If the replaced text has only capital letters | |
2403 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
4dd0c271 RS |
2404 | Otherwise if all words are capitalized in the replaced text, |
2405 | capitalize each word in NEWTEXT. | |
8c1a1077 PJ |
2406 | |
2407 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2408 | Otherwise treat `\\' as special: | |
2409 | `\\&' in NEWTEXT means substitute original matched text. | |
2410 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2411 | If Nth parens didn't match, substitute nothing. | |
2412 | `\\\\' means insert one `\\'. | |
4dd0c271 RS |
2413 | Case conversion does not apply to these substitutions. |
2414 | ||
8c1a1077 | 2415 | FIXEDCASE and LITERAL are optional arguments. |
8c1a1077 PJ |
2416 | |
2417 | The optional fourth argument STRING can be a string to modify. | |
2418 | This is meaningful when the previous match was done against STRING, | |
2419 | using `string-match'. When used this way, `replace-match' | |
2420 | creates and returns a new string made by copying STRING and replacing | |
2421 | the part of STRING that was matched. | |
2422 | ||
2423 | The optional fifth argument SUBEXP specifies a subexpression; | |
2424 | it says to replace just that subexpression with NEWTEXT, | |
2425 | rather than replacing the entire matched text. | |
2426 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2427 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2428 | NEWTEXT in place of subexp N. | |
2429 | This is useful only after a regular expression search or match, | |
2430 | since only regular expressions have distinguished subexpressions. */) | |
2431 | (newtext, fixedcase, literal, string, subexp) | |
d7a5ad5f | 2432 | Lisp_Object newtext, fixedcase, literal, string, subexp; |
ca1d1d23 JB |
2433 | { |
2434 | enum { nochange, all_caps, cap_initial } case_action; | |
ac3b28b1 | 2435 | register int pos, pos_byte; |
ca1d1d23 | 2436 | int some_multiletter_word; |
97832bd0 | 2437 | int some_lowercase; |
73dc8771 | 2438 | int some_uppercase; |
208767c3 | 2439 | int some_nonuppercase_initial; |
ca1d1d23 | 2440 | register int c, prevc; |
d7a5ad5f | 2441 | int sub; |
3e18eecf | 2442 | int opoint, newpoint; |
ca1d1d23 | 2443 | |
b7826503 | 2444 | CHECK_STRING (newtext); |
ca1d1d23 | 2445 | |
080c45fd | 2446 | if (! NILP (string)) |
b7826503 | 2447 | CHECK_STRING (string); |
080c45fd | 2448 | |
ca1d1d23 JB |
2449 | case_action = nochange; /* We tried an initialization */ |
2450 | /* but some C compilers blew it */ | |
4746118a JB |
2451 | |
2452 | if (search_regs.num_regs <= 0) | |
d72cdbfc | 2453 | error ("`replace-match' called before any match found"); |
4746118a | 2454 | |
d7a5ad5f RS |
2455 | if (NILP (subexp)) |
2456 | sub = 0; | |
2457 | else | |
2458 | { | |
b7826503 | 2459 | CHECK_NUMBER (subexp); |
d7a5ad5f RS |
2460 | sub = XINT (subexp); |
2461 | if (sub < 0 || sub >= search_regs.num_regs) | |
2462 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2463 | } | |
2464 | ||
080c45fd RS |
2465 | if (NILP (string)) |
2466 | { | |
d7a5ad5f RS |
2467 | if (search_regs.start[sub] < BEGV |
2468 | || search_regs.start[sub] > search_regs.end[sub] | |
2469 | || search_regs.end[sub] > ZV) | |
2470 | args_out_of_range (make_number (search_regs.start[sub]), | |
2471 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2472 | } |
2473 | else | |
2474 | { | |
d7a5ad5f RS |
2475 | if (search_regs.start[sub] < 0 |
2476 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2477 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2478 | args_out_of_range (make_number (search_regs.start[sub]), |
2479 | make_number (search_regs.end[sub])); | |
080c45fd | 2480 | } |
ca1d1d23 JB |
2481 | |
2482 | if (NILP (fixedcase)) | |
2483 | { | |
2484 | /* Decide how to casify by examining the matched text. */ | |
ac3b28b1 | 2485 | int last; |
ca1d1d23 | 2486 | |
ac3b28b1 KH |
2487 | pos = search_regs.start[sub]; |
2488 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2489 | |
2490 | if (NILP (string)) | |
ac3b28b1 | 2491 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2492 | else |
ac3b28b1 | 2493 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2494 | |
ca1d1d23 JB |
2495 | prevc = '\n'; |
2496 | case_action = all_caps; | |
2497 | ||
2498 | /* some_multiletter_word is set nonzero if any original word | |
2499 | is more than one letter long. */ | |
2500 | some_multiletter_word = 0; | |
97832bd0 | 2501 | some_lowercase = 0; |
208767c3 | 2502 | some_nonuppercase_initial = 0; |
73dc8771 | 2503 | some_uppercase = 0; |
ca1d1d23 | 2504 | |
ac3b28b1 | 2505 | while (pos < last) |
ca1d1d23 | 2506 | { |
080c45fd | 2507 | if (NILP (string)) |
ac3b28b1 | 2508 | { |
93daa011 | 2509 | c = FETCH_CHAR_AS_MULTIBYTE (pos_byte); |
ac3b28b1 KH |
2510 | INC_BOTH (pos, pos_byte); |
2511 | } | |
080c45fd | 2512 | else |
93daa011 | 2513 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2514 | |
ca1d1d23 JB |
2515 | if (LOWERCASEP (c)) |
2516 | { | |
2517 | /* Cannot be all caps if any original char is lower case */ | |
2518 | ||
97832bd0 | 2519 | some_lowercase = 1; |
ca1d1d23 | 2520 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2521 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2522 | else |
2523 | some_multiletter_word = 1; | |
2524 | } | |
d16c2b66 | 2525 | else if (UPPERCASEP (c)) |
ca1d1d23 | 2526 | { |
73dc8771 | 2527 | some_uppercase = 1; |
97832bd0 | 2528 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2529 | ; |
97832bd0 | 2530 | else |
ca1d1d23 JB |
2531 | some_multiletter_word = 1; |
2532 | } | |
208767c3 RS |
2533 | else |
2534 | { | |
2535 | /* If the initial is a caseless word constituent, | |
2536 | treat that like a lowercase initial. */ | |
2537 | if (SYNTAX (prevc) != Sword) | |
2538 | some_nonuppercase_initial = 1; | |
2539 | } | |
ca1d1d23 JB |
2540 | |
2541 | prevc = c; | |
2542 | } | |
2543 | ||
97832bd0 RS |
2544 | /* Convert to all caps if the old text is all caps |
2545 | and has at least one multiletter word. */ | |
2546 | if (! some_lowercase && some_multiletter_word) | |
2547 | case_action = all_caps; | |
c4d460ce | 2548 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2549 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2550 | case_action = cap_initial; |
208767c3 | 2551 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2552 | /* Should x -> yz, operating on X, give Yz or YZ? |
2553 | We'll assume the latter. */ | |
2554 | case_action = all_caps; | |
97832bd0 RS |
2555 | else |
2556 | case_action = nochange; | |
ca1d1d23 JB |
2557 | } |
2558 | ||
080c45fd RS |
2559 | /* Do replacement in a string. */ |
2560 | if (!NILP (string)) | |
2561 | { | |
2562 | Lisp_Object before, after; | |
2563 | ||
2564 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2565 | make_number (search_regs.start[sub])); |
2566 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2567 | |
636a5e28 RS |
2568 | /* Substitute parts of the match into NEWTEXT |
2569 | if desired. */ | |
080c45fd RS |
2570 | if (NILP (literal)) |
2571 | { | |
d131e79c RS |
2572 | int lastpos = 0; |
2573 | int lastpos_byte = 0; | |
080c45fd RS |
2574 | /* We build up the substituted string in ACCUM. */ |
2575 | Lisp_Object accum; | |
2576 | Lisp_Object middle; | |
d5db4077 | 2577 | int length = SBYTES (newtext); |
080c45fd RS |
2578 | |
2579 | accum = Qnil; | |
2580 | ||
ac3b28b1 | 2581 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd RS |
2582 | { |
2583 | int substart = -1; | |
6bbd7a29 | 2584 | int subend = 0; |
1e79ec24 | 2585 | int delbackslash = 0; |
080c45fd | 2586 | |
0c8533c6 RS |
2587 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2588 | ||
080c45fd RS |
2589 | if (c == '\\') |
2590 | { | |
0c8533c6 | 2591 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2592 | |
080c45fd RS |
2593 | if (c == '&') |
2594 | { | |
d7a5ad5f RS |
2595 | substart = search_regs.start[sub]; |
2596 | subend = search_regs.end[sub]; | |
080c45fd | 2597 | } |
76fc1ea2 | 2598 | else if (c >= '1' && c <= '9') |
080c45fd | 2599 | { |
76fc1ea2 KH |
2600 | if (search_regs.start[c - '0'] >= 0 |
2601 | && c <= search_regs.num_regs + '0') | |
080c45fd RS |
2602 | { |
2603 | substart = search_regs.start[c - '0']; | |
2604 | subend = search_regs.end[c - '0']; | |
2605 | } | |
76fc1ea2 KH |
2606 | else |
2607 | { | |
2608 | /* If that subexp did not match, | |
2609 | replace \\N with nothing. */ | |
2610 | substart = 0; | |
2611 | subend = 0; | |
2612 | } | |
080c45fd | 2613 | } |
1e79ec24 KH |
2614 | else if (c == '\\') |
2615 | delbackslash = 1; | |
636a5e28 RS |
2616 | else |
2617 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2618 | } |
2619 | if (substart >= 0) | |
2620 | { | |
d131e79c RS |
2621 | if (pos - 2 != lastpos) |
2622 | middle = substring_both (newtext, lastpos, | |
2623 | lastpos_byte, | |
2624 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2625 | else |
2626 | middle = Qnil; | |
2627 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2628 | Fsubstring (string, |
2629 | make_number (substart), | |
080c45fd RS |
2630 | make_number (subend))); |
2631 | lastpos = pos; | |
0c8533c6 | 2632 | lastpos_byte = pos_byte; |
080c45fd | 2633 | } |
1e79ec24 KH |
2634 | else if (delbackslash) |
2635 | { | |
d131e79c RS |
2636 | middle = substring_both (newtext, lastpos, |
2637 | lastpos_byte, | |
2638 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2639 | |
1e79ec24 KH |
2640 | accum = concat2 (accum, middle); |
2641 | lastpos = pos; | |
0c8533c6 | 2642 | lastpos_byte = pos_byte; |
1e79ec24 | 2643 | } |
080c45fd RS |
2644 | } |
2645 | ||
d131e79c RS |
2646 | if (pos != lastpos) |
2647 | middle = substring_both (newtext, lastpos, | |
2648 | lastpos_byte, | |
0c8533c6 | 2649 | pos, pos_byte); |
080c45fd RS |
2650 | else |
2651 | middle = Qnil; | |
2652 | ||
2653 | newtext = concat2 (accum, middle); | |
2654 | } | |
2655 | ||
636a5e28 | 2656 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2657 | if (case_action == all_caps) |
2658 | newtext = Fupcase (newtext); | |
2659 | else if (case_action == cap_initial) | |
2b2eead9 | 2660 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2661 | |
2662 | return concat3 (before, newtext, after); | |
2663 | } | |
2664 | ||
09c4719e | 2665 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2666 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2667 | opoint = PT - ZV; |
9160906f RS |
2668 | else if (PT > search_regs.start[sub]) |
2669 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2670 | else |
2671 | opoint = PT; | |
2672 | ||
886ed6ec RS |
2673 | /* If we want non-literal replacement, |
2674 | perform substitution on the replacement string. */ | |
2675 | if (NILP (literal)) | |
ca1d1d23 | 2676 | { |
d5db4077 | 2677 | int length = SBYTES (newtext); |
68e69fbd RS |
2678 | unsigned char *substed; |
2679 | int substed_alloc_size, substed_len; | |
3bc25e52 KH |
2680 | int buf_multibyte = !NILP (current_buffer->enable_multibyte_characters); |
2681 | int str_multibyte = STRING_MULTIBYTE (newtext); | |
2682 | Lisp_Object rev_tbl; | |
886ed6ec | 2683 | int really_changed = 0; |
3bc25e52 | 2684 | |
8f924df7 | 2685 | rev_tbl = Qnil; |
ac3b28b1 | 2686 | |
68e69fbd RS |
2687 | substed_alloc_size = length * 2 + 100; |
2688 | substed = (unsigned char *) xmalloc (substed_alloc_size + 1); | |
2689 | substed_len = 0; | |
2690 | ||
3bc25e52 KH |
2691 | /* Go thru NEWTEXT, producing the actual text to insert in |
2692 | SUBSTED while adjusting multibyteness to that of the current | |
2693 | buffer. */ | |
ca1d1d23 | 2694 | |
ac3b28b1 | 2695 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2696 | { |
68e69fbd | 2697 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
f8ce8a0d GM |
2698 | unsigned char *add_stuff = NULL; |
2699 | int add_len = 0; | |
68e69fbd | 2700 | int idx = -1; |
9a76659d | 2701 | |
3bc25e52 KH |
2702 | if (str_multibyte) |
2703 | { | |
eb99a8dd | 2704 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 KH |
2705 | if (!buf_multibyte) |
2706 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2707 | } | |
2708 | else | |
2709 | { | |
2710 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2711 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2712 | if (buf_multibyte) |
2713 | c = unibyte_char_to_multibyte (c); | |
2714 | } | |
ac3b28b1 | 2715 | |
68e69fbd RS |
2716 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2717 | or set IDX to a match index, which means put that part | |
2718 | of the buffer text into SUBSTED. */ | |
2719 | ||
ca1d1d23 JB |
2720 | if (c == '\\') |
2721 | { | |
886ed6ec RS |
2722 | really_changed = 1; |
2723 | ||
3bc25e52 KH |
2724 | if (str_multibyte) |
2725 | { | |
eb99a8dd KH |
2726 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2727 | pos, pos_byte); | |
071ce769 | 2728 | if (!buf_multibyte && !ASCII_CHAR_P (c)) |
3bc25e52 KH |
2729 | c = multibyte_char_to_unibyte (c, rev_tbl); |
2730 | } | |
2731 | else | |
2732 | { | |
d5db4077 | 2733 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2734 | if (buf_multibyte) |
2735 | c = unibyte_char_to_multibyte (c); | |
2736 | } | |
2737 | ||
ca1d1d23 | 2738 | if (c == '&') |
68e69fbd | 2739 | idx = sub; |
78445046 | 2740 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2741 | { |
2742 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2743 | idx = c - '0'; |
ca1d1d23 | 2744 | } |
636a5e28 | 2745 | else if (c == '\\') |
68e69fbd | 2746 | add_len = 1, add_stuff = "\\"; |
636a5e28 | 2747 | else |
3bc25e52 KH |
2748 | { |
2749 | xfree (substed); | |
2750 | error ("Invalid use of `\\' in replacement text"); | |
2751 | } | |
ca1d1d23 JB |
2752 | } |
2753 | else | |
68e69fbd RS |
2754 | { |
2755 | add_len = CHAR_STRING (c, str); | |
2756 | add_stuff = str; | |
2757 | } | |
2758 | ||
2759 | /* If we want to copy part of a previous match, | |
2760 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2761 | if (idx >= 0) | |
2762 | { | |
2763 | int begbyte = CHAR_TO_BYTE (search_regs.start[idx]); | |
2764 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; | |
2765 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2766 | move_gap (search_regs.start[idx]); | |
2767 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2768 | } | |
2769 | ||
2770 | /* Now the stuff we want to add to SUBSTED | |
2771 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2772 | ||
2773 | /* Make sure SUBSTED is big enough. */ | |
2774 | if (substed_len + add_len >= substed_alloc_size) | |
2775 | { | |
2776 | substed_alloc_size = substed_len + add_len + 500; | |
2777 | substed = (unsigned char *) xrealloc (substed, | |
2778 | substed_alloc_size + 1); | |
2779 | } | |
2780 | ||
2781 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2782 | if (add_stuff) |
2783 | { | |
2784 | bcopy (add_stuff, substed + substed_len, add_len); | |
2785 | substed_len += add_len; | |
2786 | } | |
ca1d1d23 | 2787 | } |
68e69fbd | 2788 | |
886ed6ec | 2789 | if (really_changed) |
76fc1ea2 KH |
2790 | { |
2791 | if (buf_multibyte) | |
2792 | { | |
2793 | int nchars = multibyte_chars_in_text (substed, substed_len); | |
68e69fbd | 2794 | |
76fc1ea2 KH |
2795 | newtext = make_multibyte_string (substed, nchars, substed_len); |
2796 | } | |
2797 | else | |
2798 | newtext = make_unibyte_string (substed, substed_len); | |
2799 | } | |
68e69fbd | 2800 | xfree (substed); |
ca1d1d23 JB |
2801 | } |
2802 | ||
886ed6ec RS |
2803 | /* Replace the old text with the new in the cleanest possible way. */ |
2804 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2805 | newtext, 1, 0, 1); | |
d5db4077 | 2806 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2807 | |
2808 | if (case_action == all_caps) | |
886ed6ec RS |
2809 | Fupcase_region (make_number (search_regs.start[sub]), |
2810 | make_number (newpoint)); | |
ca1d1d23 | 2811 | else if (case_action == cap_initial) |
886ed6ec RS |
2812 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2813 | make_number (newpoint)); | |
3e18eecf | 2814 | |
98e942e0 RS |
2815 | /* Adjust search data for this change. */ |
2816 | { | |
5b88a2c5 | 2817 | int oldend = search_regs.end[sub]; |
41c01205 | 2818 | int oldstart = search_regs.start[sub]; |
98e942e0 RS |
2819 | int change = newpoint - search_regs.end[sub]; |
2820 | int i; | |
2821 | ||
2822 | for (i = 0; i < search_regs.num_regs; i++) | |
2823 | { | |
41c01205 | 2824 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2825 | search_regs.start[i] += change; |
41c01205 DK |
2826 | else if (search_regs.start[i] > oldstart) |
2827 | search_regs.start[i] = oldstart; | |
2828 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2829 | search_regs.end[i] += change; |
41c01205 DK |
2830 | else if (search_regs.end[i] > oldstart) |
2831 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2832 | } |
2833 | } | |
2834 | ||
b0eba991 | 2835 | /* Put point back where it was in the text. */ |
8d808a65 | 2836 | if (opoint <= 0) |
fa8ed3e0 | 2837 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2838 | else |
fa8ed3e0 | 2839 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2840 | |
2841 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2842 | move_if_not_intangible (newpoint); |
177c0ea7 | 2843 | |
ca1d1d23 JB |
2844 | return Qnil; |
2845 | } | |
2846 | \f | |
2847 | static Lisp_Object | |
2848 | match_limit (num, beginningp) | |
2849 | Lisp_Object num; | |
2850 | int beginningp; | |
2851 | { | |
2852 | register int n; | |
2853 | ||
b7826503 | 2854 | CHECK_NUMBER (num); |
ca1d1d23 | 2855 | n = XINT (num); |
f90a5bf5 | 2856 | if (n < 0) |
bd2cbd56 | 2857 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2858 | if (search_regs.num_regs <= 0) |
2859 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2860 | if (n >= search_regs.num_regs |
4746118a | 2861 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2862 | return Qnil; |
2863 | return (make_number ((beginningp) ? search_regs.start[n] | |
2864 | : search_regs.end[n])); | |
2865 | } | |
2866 | ||
2867 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
8c1a1077 PJ |
2868 | doc: /* Return position of start of text matched by last search. |
2869 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2870 | regexp. | |
2871 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2872 | SUBEXP pairs. | |
2873 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2874 | (subexp) | |
5806161b | 2875 | Lisp_Object subexp; |
ca1d1d23 | 2876 | { |
5806161b | 2877 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2878 | } |
2879 | ||
2880 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
8c1a1077 PJ |
2881 | doc: /* Return position of end of text matched by last search. |
2882 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2883 | regexp. | |
2884 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2885 | SUBEXP pairs. | |
2886 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2887 | (subexp) | |
5806161b | 2888 | Lisp_Object subexp; |
ca1d1d23 | 2889 | { |
5806161b | 2890 | return match_limit (subexp, 0); |
177c0ea7 | 2891 | } |
ca1d1d23 | 2892 | |
abd0071c | 2893 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 3, 0, |
8c1a1077 PJ |
2894 | doc: /* Return a list containing all info on what the last search matched. |
2895 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2896 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2897 | if the last match was on a buffer; integers or nil if a string was matched. | |
2898 | Use `store-match-data' to reinstate the data in this list. | |
2899 | ||
41c01205 DK |
2900 | If INTEGERS (the optional first argument) is non-nil, always use |
2901 | integers \(rather than markers) to represent buffer positions. In | |
2902 | this case, and if the last match was in a buffer, the buffer will get | |
2903 | stored as one additional element at the end of the list. | |
2904 | ||
abd0071c KS |
2905 | If REUSE is a list, reuse it as part of the value. If REUSE is long |
2906 | enough to hold all the values, and if INTEGERS is non-nil, no consing | |
2907 | is done. | |
2908 | ||
2909 | If optional third arg RESEAT is non-nil, any previous markers on the | |
2910 | REUSE list will be modified to point to nowhere. | |
2911 | ||
140a6b7e | 2912 | Return value is undefined if the last search failed. */) |
abd0071c KS |
2913 | (integers, reuse, reseat) |
2914 | Lisp_Object integers, reuse, reseat; | |
ca1d1d23 | 2915 | { |
56256c2a | 2916 | Lisp_Object tail, prev; |
4746118a | 2917 | Lisp_Object *data; |
ca1d1d23 JB |
2918 | int i, len; |
2919 | ||
abd0071c KS |
2920 | if (!NILP (reseat)) |
2921 | for (tail = reuse; CONSP (tail); tail = XCDR (tail)) | |
2922 | if (MARKERP (XCAR (tail))) | |
2923 | { | |
51f10faa | 2924 | unchain_marker (XMARKER (XCAR (tail))); |
abd0071c KS |
2925 | XSETCAR (tail, Qnil); |
2926 | } | |
2927 | ||
daa37602 | 2928 | if (NILP (last_thing_searched)) |
c36bcf1b | 2929 | return Qnil; |
daa37602 | 2930 | |
6bbd7a29 GM |
2931 | prev = Qnil; |
2932 | ||
41c01205 | 2933 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs + 1) |
4746118a JB |
2934 | * sizeof (Lisp_Object)); |
2935 | ||
41c01205 | 2936 | len = 0; |
4746118a | 2937 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2938 | { |
2939 | int start = search_regs.start[i]; | |
2940 | if (start >= 0) | |
2941 | { | |
56256c2a RS |
2942 | if (EQ (last_thing_searched, Qt) |
2943 | || ! NILP (integers)) | |
ca1d1d23 | 2944 | { |
c235cce7 KH |
2945 | XSETFASTINT (data[2 * i], start); |
2946 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2947 | } |
0ed62dc7 | 2948 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2949 | { |
2950 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2951 | Fset_marker (data[2 * i], |
2952 | make_number (start), | |
2953 | last_thing_searched); | |
ca1d1d23 JB |
2954 | data[2 * i + 1] = Fmake_marker (); |
2955 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2956 | make_number (search_regs.end[i]), |
daa37602 | 2957 | last_thing_searched); |
ca1d1d23 | 2958 | } |
daa37602 JB |
2959 | else |
2960 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2961 | abort (); | |
2962 | ||
abd0071c | 2963 | len = 2 * i + 2; |
ca1d1d23 JB |
2964 | } |
2965 | else | |
abd0071c | 2966 | data[2 * i] = data[2 * i + 1] = Qnil; |
ca1d1d23 | 2967 | } |
56256c2a | 2968 | |
bd2cbd56 | 2969 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2970 | { |
bd2cbd56 | 2971 | data[len] = last_thing_searched; |
41c01205 DK |
2972 | len++; |
2973 | } | |
2974 | ||
56256c2a RS |
2975 | /* If REUSE is not usable, cons up the values and return them. */ |
2976 | if (! CONSP (reuse)) | |
41c01205 | 2977 | return Flist (len, data); |
56256c2a RS |
2978 | |
2979 | /* If REUSE is a list, store as many value elements as will fit | |
2980 | into the elements of REUSE. */ | |
2981 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2982 | i++, tail = XCDR (tail)) |
56256c2a | 2983 | { |
41c01205 | 2984 | if (i < len) |
f3fbd155 | 2985 | XSETCAR (tail, data[i]); |
56256c2a | 2986 | else |
f3fbd155 | 2987 | XSETCAR (tail, Qnil); |
56256c2a RS |
2988 | prev = tail; |
2989 | } | |
2990 | ||
2991 | /* If we couldn't fit all value elements into REUSE, | |
2992 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2993 | if (i < len) |
2994 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2995 | |
2996 | return reuse; | |
ca1d1d23 JB |
2997 | } |
2998 | ||
b51d6c92 SM |
2999 | /* We used to have an internal use variant of `reseat' described as: |
3000 | ||
3001 | If RESEAT is `evaporate', put the markers back on the free list | |
3002 | immediately. No other references to the markers must exist in this | |
3003 | case, so it is used only internally on the unwind stack and | |
3004 | save-match-data from Lisp. | |
3005 | ||
3006 | But it was ill-conceived: those supposedly-internal markers get exposed via | |
3007 | the undo-list, so freeing them here is unsafe. */ | |
ca1d1d23 | 3008 | |
abd0071c | 3009 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 2, 0, |
8c1a1077 | 3010 | doc: /* Set internal data on last search match from elements of LIST. |
abd0071c KS |
3011 | LIST should have been created by calling `match-data' previously. |
3012 | ||
51f10faa | 3013 | If optional arg RESEAT is non-nil, make markers on LIST point nowhere. */) |
abd0071c KS |
3014 | (list, reseat) |
3015 | register Lisp_Object list, reseat; | |
ca1d1d23 JB |
3016 | { |
3017 | register int i; | |
3018 | register Lisp_Object marker; | |
3019 | ||
7074fde6 FP |
3020 | if (running_asynch_code) |
3021 | save_search_regs (); | |
3022 | ||
29100cea | 3023 | CHECK_LIST (list); |
ca1d1d23 | 3024 | |
41c01205 DK |
3025 | /* Unless we find a marker with a buffer or an explicit buffer |
3026 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
3027 | last_thing_searched = Qt; |
3028 | ||
4746118a JB |
3029 | /* Allocate registers if they don't already exist. */ |
3030 | { | |
d084e942 | 3031 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
3032 | |
3033 | if (length > search_regs.num_regs) | |
3034 | { | |
1113d9db JB |
3035 | if (search_regs.num_regs == 0) |
3036 | { | |
3037 | search_regs.start | |
3038 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
3039 | search_regs.end | |
3040 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
3041 | } | |
4746118a | 3042 | else |
1113d9db JB |
3043 | { |
3044 | search_regs.start | |
3045 | = (regoff_t *) xrealloc (search_regs.start, | |
3046 | length * sizeof (regoff_t)); | |
3047 | search_regs.end | |
3048 | = (regoff_t *) xrealloc (search_regs.end, | |
3049 | length * sizeof (regoff_t)); | |
3050 | } | |
4746118a | 3051 | |
e62371e9 KH |
3052 | for (i = search_regs.num_regs; i < length; i++) |
3053 | search_regs.start[i] = -1; | |
3054 | ||
487282dc | 3055 | search_regs.num_regs = length; |
4746118a | 3056 | } |
ca1d1d23 | 3057 | |
abd0071c | 3058 | for (i = 0; CONSP (list); i++) |
41c01205 | 3059 | { |
abd0071c | 3060 | marker = XCAR (list); |
bd2cbd56 | 3061 | if (BUFFERP (marker)) |
c3762cbd | 3062 | { |
bd2cbd56 | 3063 | last_thing_searched = marker; |
c3762cbd DK |
3064 | break; |
3065 | } | |
3066 | if (i >= length) | |
3067 | break; | |
41c01205 DK |
3068 | if (NILP (marker)) |
3069 | { | |
3070 | search_regs.start[i] = -1; | |
abd0071c | 3071 | list = XCDR (list); |
41c01205 DK |
3072 | } |
3073 | else | |
3074 | { | |
3075 | int from; | |
abd0071c | 3076 | Lisp_Object m; |
e2811828 | 3077 | |
abd0071c | 3078 | m = marker; |
41c01205 DK |
3079 | if (MARKERP (marker)) |
3080 | { | |
3081 | if (XMARKER (marker)->buffer == 0) | |
3082 | XSETFASTINT (marker, 0); | |
3083 | else | |
3084 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
3085 | } | |
e2811828 | 3086 | |
41c01205 DK |
3087 | CHECK_NUMBER_COERCE_MARKER (marker); |
3088 | from = XINT (marker); | |
e2811828 | 3089 | |
abd0071c KS |
3090 | if (!NILP (reseat) && MARKERP (m)) |
3091 | { | |
b51d6c92 | 3092 | unchain_marker (XMARKER (m)); |
9ad54a7e | 3093 | XSETCAR (list, Qnil); |
abd0071c KS |
3094 | } |
3095 | ||
3096 | if ((list = XCDR (list), !CONSP (list))) | |
3097 | break; | |
3098 | ||
3099 | m = marker = XCAR (list); | |
3100 | ||
41c01205 DK |
3101 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
3102 | XSETFASTINT (marker, 0); | |
e2811828 | 3103 | |
41c01205 DK |
3104 | CHECK_NUMBER_COERCE_MARKER (marker); |
3105 | search_regs.start[i] = from; | |
3106 | search_regs.end[i] = XINT (marker); | |
abd0071c KS |
3107 | |
3108 | if (!NILP (reseat) && MARKERP (m)) | |
3109 | { | |
b51d6c92 | 3110 | unchain_marker (XMARKER (m)); |
9ad54a7e | 3111 | XSETCAR (list, Qnil); |
abd0071c | 3112 | } |
41c01205 | 3113 | } |
abd0071c | 3114 | list = XCDR (list); |
41c01205 | 3115 | } |
ca1d1d23 | 3116 | |
41c01205 DK |
3117 | for (; i < search_regs.num_regs; i++) |
3118 | search_regs.start[i] = -1; | |
3119 | } | |
ca1d1d23 | 3120 | |
177c0ea7 | 3121 | return Qnil; |
ca1d1d23 JB |
3122 | } |
3123 | ||
7074fde6 FP |
3124 | /* If non-zero the match data have been saved in saved_search_regs |
3125 | during the execution of a sentinel or filter. */ | |
75ebf74b | 3126 | static int search_regs_saved; |
7074fde6 | 3127 | static struct re_registers saved_search_regs; |
41c01205 | 3128 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
3129 | |
3130 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
3131 | if asynchronous code (filter or sentinel) is running. */ | |
3132 | static void | |
3133 | save_search_regs () | |
3134 | { | |
3135 | if (!search_regs_saved) | |
3136 | { | |
3137 | saved_search_regs.num_regs = search_regs.num_regs; | |
3138 | saved_search_regs.start = search_regs.start; | |
3139 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
3140 | saved_last_thing_searched = last_thing_searched; |
3141 | last_thing_searched = Qnil; | |
7074fde6 | 3142 | search_regs.num_regs = 0; |
2d4a771a RS |
3143 | search_regs.start = 0; |
3144 | search_regs.end = 0; | |
7074fde6 FP |
3145 | |
3146 | search_regs_saved = 1; | |
3147 | } | |
3148 | } | |
3149 | ||
3150 | /* Called upon exit from filters and sentinels. */ | |
3151 | void | |
abd0071c | 3152 | restore_search_regs () |
7074fde6 FP |
3153 | { |
3154 | if (search_regs_saved) | |
3155 | { | |
3156 | if (search_regs.num_regs > 0) | |
3157 | { | |
3158 | xfree (search_regs.start); | |
3159 | xfree (search_regs.end); | |
3160 | } | |
3161 | search_regs.num_regs = saved_search_regs.num_regs; | |
3162 | search_regs.start = saved_search_regs.start; | |
3163 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
3164 | last_thing_searched = saved_last_thing_searched; |
3165 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
3166 | search_regs_saved = 0; |
3167 | } | |
3168 | } | |
3169 | ||
abd0071c KS |
3170 | static Lisp_Object |
3171 | unwind_set_match_data (list) | |
3172 | Lisp_Object list; | |
3173 | { | |
b51d6c92 SM |
3174 | /* It is NOT ALWAYS safe to free (evaporate) the markers immediately. */ |
3175 | return Fset_match_data (list, Qt); | |
abd0071c KS |
3176 | } |
3177 | ||
3178 | /* Called to unwind protect the match data. */ | |
3179 | void | |
3180 | record_unwind_save_match_data () | |
3181 | { | |
3182 | record_unwind_protect (unwind_set_match_data, | |
3183 | Fmatch_data (Qnil, Qnil, Qnil)); | |
3184 | } | |
3185 | ||
ca1d1d23 JB |
3186 | /* Quote a string to inactivate reg-expr chars */ |
3187 | ||
3188 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 PJ |
3189 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
3190 | (string) | |
5806161b | 3191 | Lisp_Object string; |
ca1d1d23 JB |
3192 | { |
3193 | register unsigned char *in, *out, *end; | |
3194 | register unsigned char *temp; | |
0c8533c6 | 3195 | int backslashes_added = 0; |
ca1d1d23 | 3196 | |
b7826503 | 3197 | CHECK_STRING (string); |
ca1d1d23 | 3198 | |
d5db4077 | 3199 | temp = (unsigned char *) alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
3200 | |
3201 | /* Now copy the data into the new string, inserting escapes. */ | |
3202 | ||
d5db4077 KR |
3203 | in = SDATA (string); |
3204 | end = in + SBYTES (string); | |
177c0ea7 | 3205 | out = temp; |
ca1d1d23 JB |
3206 | |
3207 | for (; in != end; in++) | |
3208 | { | |
66bc6082 | 3209 | if (*in == '[' |
ca1d1d23 JB |
3210 | || *in == '*' || *in == '.' || *in == '\\' |
3211 | || *in == '?' || *in == '+' | |
3212 | || *in == '^' || *in == '$') | |
0c8533c6 | 3213 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
3214 | *out++ = *in; |
3215 | } | |
3216 | ||
3f8100f1 | 3217 | return make_specified_string (temp, |
d5db4077 | 3218 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
3219 | out - temp, |
3220 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 3221 | } |
177c0ea7 | 3222 | \f |
dfcf069d | 3223 | void |
ca1d1d23 JB |
3224 | syms_of_search () |
3225 | { | |
3226 | register int i; | |
3227 | ||
487282dc KH |
3228 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
3229 | { | |
3230 | searchbufs[i].buf.allocated = 100; | |
b23c0a83 | 3231 | searchbufs[i].buf.buffer = (unsigned char *) xmalloc (100); |
487282dc KH |
3232 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
3233 | searchbufs[i].regexp = Qnil; | |
ecdb561e | 3234 | searchbufs[i].whitespace_regexp = Qnil; |
b69e3c18 | 3235 | searchbufs[i].syntax_table = Qnil; |
487282dc | 3236 | staticpro (&searchbufs[i].regexp); |
aa77b5ce | 3237 | staticpro (&searchbufs[i].whitespace_regexp); |
b69e3c18 | 3238 | staticpro (&searchbufs[i].syntax_table); |
487282dc KH |
3239 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); |
3240 | } | |
3241 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
3242 | |
3243 | Qsearch_failed = intern ("search-failed"); | |
3244 | staticpro (&Qsearch_failed); | |
3245 | Qinvalid_regexp = intern ("invalid-regexp"); | |
3246 | staticpro (&Qinvalid_regexp); | |
3247 | ||
3248 | Fput (Qsearch_failed, Qerror_conditions, | |
3249 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
3250 | Fput (Qsearch_failed, Qerror_message, | |
3251 | build_string ("Search failed")); | |
3252 | ||
3253 | Fput (Qinvalid_regexp, Qerror_conditions, | |
3254 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
3255 | Fput (Qinvalid_regexp, Qerror_message, | |
3256 | build_string ("Invalid regexp")); | |
3257 | ||
daa37602 JB |
3258 | last_thing_searched = Qnil; |
3259 | staticpro (&last_thing_searched); | |
3260 | ||
0f6af254 DK |
3261 | saved_last_thing_searched = Qnil; |
3262 | staticpro (&saved_last_thing_searched); | |
3263 | ||
41a33295 | 3264 | DEFVAR_LISP ("search-spaces-regexp", &Vsearch_spaces_regexp, |
e2811828 | 3265 | doc: /* Regexp to substitute for bunches of spaces in regexp search. |
f31a9a68 RS |
3266 | Some commands use this for user-specified regexps. |
3267 | Spaces that occur inside character classes or repetition operators | |
3268 | or other such regexp constructs are not replaced with this. | |
3269 | A value of nil (which is the normal value) means treat spaces literally. */); | |
41a33295 | 3270 | Vsearch_spaces_regexp = Qnil; |
f31a9a68 | 3271 | |
ef887810 RS |
3272 | DEFVAR_LISP ("inhibit-changing-match-data", &Vinhibit_changing_match_data, |
3273 | doc: /* Internal use only. | |
39d0bf74 RS |
3274 | If non-nil, the primitive searching and matching functions |
3275 | such as `looking-at', `string-match', `re-search-forward', etc., | |
3276 | do not set the match data. The proper way to use this variable | |
3277 | is to bind it with `let' around a small expression. */); | |
ef887810 RS |
3278 | Vinhibit_changing_match_data = Qnil; |
3279 | ||
ca1d1d23 | 3280 | defsubr (&Slooking_at); |
b819a390 RS |
3281 | defsubr (&Sposix_looking_at); |
3282 | defsubr (&Sstring_match); | |
3283 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
3284 | defsubr (&Ssearch_forward); |
3285 | defsubr (&Ssearch_backward); | |
3286 | defsubr (&Sword_search_forward); | |
3287 | defsubr (&Sword_search_backward); | |
c6e0e1cc CY |
3288 | defsubr (&Sword_search_forward_lax); |
3289 | defsubr (&Sword_search_backward_lax); | |
ca1d1d23 JB |
3290 | defsubr (&Sre_search_forward); |
3291 | defsubr (&Sre_search_backward); | |
b819a390 RS |
3292 | defsubr (&Sposix_search_forward); |
3293 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
3294 | defsubr (&Sreplace_match); |
3295 | defsubr (&Smatch_beginning); | |
3296 | defsubr (&Smatch_end); | |
3297 | defsubr (&Smatch_data); | |
3f1c005b | 3298 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
3299 | defsubr (&Sregexp_quote); |
3300 | } | |
76fc1ea2 KH |
3301 | |
3302 | /* arch-tag: a6059d79-0552-4f14-a2cb-d379a4e3c78f | |
3303 | (do not change this comment) */ |