Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
429ab54e GM |
2 | Copyright (C) 1985, 1986, 1987, 1993, 1994, 1997, 1998, 1999, 2001, 2002, |
3 | 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc. | |
ca1d1d23 JB |
4 | |
5 | This file is part of GNU Emacs. | |
6 | ||
7 | GNU Emacs is free software; you can redistribute it and/or modify | |
8 | it under the terms of the GNU General Public License as published by | |
7c938215 | 9 | the Free Software Foundation; either version 2, or (at your option) |
ca1d1d23 JB |
10 | any later version. |
11 | ||
12 | GNU Emacs is distributed in the hope that it will be useful, | |
13 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | GNU General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
18 | along with GNU Emacs; see the file COPYING. If not, write to | |
4fc5845f LK |
19 | the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, |
20 | Boston, MA 02110-1301, USA. */ | |
ca1d1d23 JB |
21 | |
22 | ||
18160b98 | 23 | #include <config.h> |
ca1d1d23 JB |
24 | #include "lisp.h" |
25 | #include "syntax.h" | |
5679531d | 26 | #include "category.h" |
ca1d1d23 | 27 | #include "buffer.h" |
76eb0881 | 28 | #include "character.h" |
9169c321 | 29 | #include "region-cache.h" |
ca1d1d23 | 30 | #include "commands.h" |
9ac0d9e0 | 31 | #include "blockinput.h" |
bf1760bb | 32 | #include "intervals.h" |
4746118a | 33 | |
ca1d1d23 JB |
34 | #include <sys/types.h> |
35 | #include "regex.h" | |
36 | ||
1d288aef | 37 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 38 | |
487282dc KH |
39 | /* If the regexp is non-nil, then the buffer contains the compiled form |
40 | of that regexp, suitable for searching. */ | |
1d288aef RS |
41 | struct regexp_cache |
42 | { | |
487282dc | 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 | |
ca325161 | 95 | static void set_search_regs (); |
044f81f1 | 96 | static void save_search_regs (); |
facdc750 RS |
97 | static int simple_search (); |
98 | static int boyer_moore (); | |
b819a390 | 99 | static int search_buffer (); |
01d09305 | 100 | static void matcher_overflow () NO_RETURN; |
b819a390 | 101 | |
ca1d1d23 JB |
102 | static void |
103 | matcher_overflow () | |
104 | { | |
105 | error ("Stack overflow in regexp matcher"); | |
106 | } | |
107 | ||
b819a390 RS |
108 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
109 | PATTERN is the pattern to compile. | |
110 | CP is the place to put the result. | |
facdc750 | 111 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
112 | REGP is the structure that says where to store the "register" |
113 | values that will result from matching this pattern. | |
114 | If it is 0, we should compile the pattern not to record any | |
115 | subexpression bounds. | |
116 | POSIX is nonzero if we want full backtracking (POSIX style) | |
5679531d | 117 | for this pattern. 0 means backtrack only enough to get a valid match. |
8f924df7 | 118 | MULTIBYTE is nonzero iff a target of match is a multibyte buffer or |
bbc73b48 KH |
119 | string. |
120 | ||
121 | The behavior also depends on Vsearch_spaces_regexp. */ | |
ca1d1d23 | 122 | |
487282dc | 123 | static void |
5679531d | 124 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte) |
487282dc | 125 | struct regexp_cache *cp; |
ca1d1d23 | 126 | Lisp_Object pattern; |
facdc750 | 127 | Lisp_Object translate; |
487282dc | 128 | struct re_registers *regp; |
b819a390 | 129 | int posix; |
5679531d | 130 | int multibyte; |
ca1d1d23 | 131 | { |
d451e4db | 132 | char *val; |
b819a390 | 133 | reg_syntax_t old; |
ca1d1d23 | 134 | |
487282dc | 135 | cp->regexp = Qnil; |
59fab369 | 136 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 137 | cp->posix = posix; |
93daa011 KH |
138 | cp->buf.multibyte = STRING_MULTIBYTE (pattern); |
139 | cp->buf.target_multibyte = multibyte; | |
bbc73b48 | 140 | cp->whitespace_regexp = Vsearch_spaces_regexp; |
e7b2dd2e RS |
141 | /* rms: I think BLOCK_INPUT is not needed here any more, |
142 | because regex.c defines malloc to call xmalloc. | |
143 | Using BLOCK_INPUT here means the debugger won't run if an error occurs. | |
144 | So let's turn it off. */ | |
145 | /* BLOCK_INPUT; */ | |
fb4a568d | 146 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 147 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
bbc73b48 KH |
148 | re_set_whitespace_regexp (NILP (Vsearch_spaces_regexp) ? NULL |
149 | : SDATA (Vsearch_spaces_regexp)); | |
150 | ||
8f924df7 KH |
151 | val = (char *) re_compile_pattern ((char *) SDATA (pattern), |
152 | SBYTES (pattern), &cp->buf); | |
bbc73b48 | 153 | |
58e95211 SM |
154 | /* If the compiled pattern hard codes some of the contents of the |
155 | syntax-table, it can only be reused with *this* syntax table. */ | |
156 | cp->syntax_table = cp->buf.used_syntax ? current_buffer->syntax_table : Qt; | |
157 | ||
bbc73b48 KH |
158 | re_set_whitespace_regexp (NULL); |
159 | ||
b819a390 | 160 | re_set_syntax (old); |
e7b2dd2e | 161 | /* UNBLOCK_INPUT; */ |
ca1d1d23 | 162 | if (val) |
06f77a8a | 163 | xsignal1 (Qinvalid_regexp, build_string (val)); |
1113d9db | 164 | |
487282dc | 165 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
166 | } |
167 | ||
6efc7887 RS |
168 | /* Shrink each compiled regexp buffer in the cache |
169 | to the size actually used right now. | |
170 | This is called from garbage collection. */ | |
171 | ||
172 | void | |
173 | shrink_regexp_cache () | |
174 | { | |
a968f437 | 175 | struct regexp_cache *cp; |
6efc7887 RS |
176 | |
177 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
178 | { | |
179 | cp->buf.allocated = cp->buf.used; | |
180 | cp->buf.buffer | |
b23c0a83 | 181 | = (unsigned char *) xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
182 | } |
183 | } | |
184 | ||
54dd3310 SM |
185 | /* Clear the regexp cache w.r.t. a particular syntax table, |
186 | because it was changed. | |
e5b94d44 CY |
187 | There is no danger of memory leak here because re_compile_pattern |
188 | automagically manages the memory in each re_pattern_buffer struct, | |
189 | based on its `allocated' and `buffer' values. */ | |
190 | void | |
191 | clear_regexp_cache () | |
192 | { | |
193 | int i; | |
194 | ||
195 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) | |
54dd3310 SM |
196 | /* It's tempting to compare with the syntax-table we've actually changd, |
197 | but it's not sufficient because char-table inheritance mewans that | |
198 | modifying one syntax-table can change others at the same time. */ | |
199 | if (!EQ (searchbufs[i].syntax_table, Qt)) | |
200 | searchbufs[i].regexp = Qnil; | |
e5b94d44 CY |
201 | } |
202 | ||
487282dc | 203 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
204 | the cache. |
205 | PATTERN is the pattern to compile. | |
facdc750 | 206 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
207 | REGP is the structure that says where to store the "register" |
208 | values that will result from matching this pattern. | |
209 | If it is 0, we should compile the pattern not to record any | |
210 | subexpression bounds. | |
211 | POSIX is nonzero if we want full backtracking (POSIX style) | |
212 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
213 | |
214 | struct re_pattern_buffer * | |
0c8533c6 | 215 | compile_pattern (pattern, regp, translate, posix, multibyte) |
487282dc KH |
216 | Lisp_Object pattern; |
217 | struct re_registers *regp; | |
facdc750 | 218 | Lisp_Object translate; |
0c8533c6 | 219 | int posix, multibyte; |
487282dc KH |
220 | { |
221 | struct regexp_cache *cp, **cpp; | |
222 | ||
223 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
224 | { | |
225 | cp = *cpp; | |
f1b9c7c1 KR |
226 | /* Entries are initialized to nil, and may be set to nil by |
227 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
49a5f770 KR |
228 | string accessors in those cases. However, compile_pattern_1 |
229 | is only applied to the cache entry we pick here to reuse. So | |
230 | nil should never appear before a non-nil entry. */ | |
7c752c80 | 231 | if (NILP (cp->regexp)) |
f1b9c7c1 | 232 | goto compile_it; |
d5db4077 | 233 | if (SCHARS (cp->regexp) == SCHARS (pattern) |
cf69b13e | 234 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 235 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 236 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d | 237 | && cp->posix == posix |
bbc73b48 | 238 | && cp->buf.target_multibyte == multibyte |
58e95211 SM |
239 | && (EQ (cp->syntax_table, Qt) |
240 | || EQ (cp->syntax_table, current_buffer->syntax_table)) | |
bbc73b48 | 241 | && !NILP (Fequal (cp->whitespace_regexp, Vsearch_spaces_regexp))) |
487282dc KH |
242 | break; |
243 | ||
f1b9c7c1 KR |
244 | /* If we're at the end of the cache, compile into the nil cell |
245 | we found, or the last (least recently used) cell with a | |
246 | string value. */ | |
487282dc KH |
247 | if (cp->next == 0) |
248 | { | |
f1b9c7c1 | 249 | compile_it: |
5679531d | 250 | compile_pattern_1 (cp, pattern, translate, regp, posix, multibyte); |
487282dc KH |
251 | break; |
252 | } | |
253 | } | |
254 | ||
255 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
256 | either because we found it in the cache or because we just compiled it. | |
257 | Move it to the front of the queue to mark it as most recently used. */ | |
258 | *cpp = cp->next; | |
259 | cp->next = searchbuf_head; | |
260 | searchbuf_head = cp; | |
1113d9db | 261 | |
6639708c RS |
262 | /* Advise the searching functions about the space we have allocated |
263 | for register data. */ | |
264 | if (regp) | |
265 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
266 | ||
487282dc | 267 | return &cp->buf; |
ca1d1d23 JB |
268 | } |
269 | ||
ca1d1d23 | 270 | \f |
b819a390 RS |
271 | static Lisp_Object |
272 | looking_at_1 (string, posix) | |
ca1d1d23 | 273 | Lisp_Object string; |
b819a390 | 274 | int posix; |
ca1d1d23 JB |
275 | { |
276 | Lisp_Object val; | |
277 | unsigned char *p1, *p2; | |
278 | int s1, s2; | |
279 | register int i; | |
487282dc | 280 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 281 | |
7074fde6 FP |
282 | if (running_asynch_code) |
283 | save_search_regs (); | |
284 | ||
910c747a RS |
285 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
286 | XCHAR_TABLE (current_buffer->case_canon_table)->extras[2] | |
287 | = current_buffer->case_eqv_table; | |
288 | ||
b7826503 | 289 | CHECK_STRING (string); |
487282dc KH |
290 | bufp = compile_pattern (string, &search_regs, |
291 | (!NILP (current_buffer->case_fold_search) | |
0190922f | 292 | ? current_buffer->case_canon_table : Qnil), |
0c8533c6 RS |
293 | posix, |
294 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 JB |
295 | |
296 | immediate_quit = 1; | |
297 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
298 | ||
299 | /* Get pointers and sizes of the two strings | |
300 | that make up the visible portion of the buffer. */ | |
301 | ||
302 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 303 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 304 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 305 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
306 | if (s1 < 0) |
307 | { | |
308 | p2 = p1; | |
fa8ed3e0 | 309 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
310 | s1 = 0; |
311 | } | |
312 | if (s2 < 0) | |
313 | { | |
fa8ed3e0 | 314 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
315 | s2 = 0; |
316 | } | |
8bb43c28 RS |
317 | |
318 | re_match_object = Qnil; | |
177c0ea7 | 319 | |
487282dc | 320 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
fa8ed3e0 RS |
321 | PT_BYTE - BEGV_BYTE, &search_regs, |
322 | ZV_BYTE - BEGV_BYTE); | |
de182d70 | 323 | immediate_quit = 0; |
177c0ea7 | 324 | |
ca1d1d23 JB |
325 | if (i == -2) |
326 | matcher_overflow (); | |
327 | ||
328 | val = (0 <= i ? Qt : Qnil); | |
fa8ed3e0 RS |
329 | if (i >= 0) |
330 | for (i = 0; i < search_regs.num_regs; i++) | |
331 | if (search_regs.start[i] >= 0) | |
332 | { | |
333 | search_regs.start[i] | |
334 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
335 | search_regs.end[i] | |
336 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
337 | } | |
a3668d92 | 338 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
339 | return val; |
340 | } | |
341 | ||
b819a390 | 342 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
343 | doc: /* Return t if text after point matches regular expression REGEXP. |
344 | This function modifies the match data that `match-beginning', | |
345 | `match-end' and `match-data' access; save and restore the match | |
346 | data if you want to preserve them. */) | |
347 | (regexp) | |
94f94972 | 348 | Lisp_Object regexp; |
b819a390 | 349 | { |
94f94972 | 350 | return looking_at_1 (regexp, 0); |
b819a390 RS |
351 | } |
352 | ||
353 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
354 | doc: /* Return t if text after point matches regular expression REGEXP. |
355 | Find the longest match, in accord with Posix regular expression rules. | |
356 | This function modifies the match data that `match-beginning', | |
357 | `match-end' and `match-data' access; save and restore the match | |
358 | data if you want to preserve them. */) | |
359 | (regexp) | |
94f94972 | 360 | Lisp_Object regexp; |
b819a390 | 361 | { |
94f94972 | 362 | return looking_at_1 (regexp, 1); |
b819a390 RS |
363 | } |
364 | \f | |
365 | static Lisp_Object | |
366 | string_match_1 (regexp, string, start, posix) | |
ca1d1d23 | 367 | Lisp_Object regexp, string, start; |
b819a390 | 368 | int posix; |
ca1d1d23 JB |
369 | { |
370 | int val; | |
487282dc | 371 | struct re_pattern_buffer *bufp; |
0c8533c6 RS |
372 | int pos, pos_byte; |
373 | int i; | |
ca1d1d23 | 374 | |
7074fde6 FP |
375 | if (running_asynch_code) |
376 | save_search_regs (); | |
377 | ||
b7826503 PJ |
378 | CHECK_STRING (regexp); |
379 | CHECK_STRING (string); | |
ca1d1d23 JB |
380 | |
381 | if (NILP (start)) | |
0c8533c6 | 382 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
383 | else |
384 | { | |
d5db4077 | 385 | int len = SCHARS (string); |
ca1d1d23 | 386 | |
b7826503 | 387 | CHECK_NUMBER (start); |
0c8533c6 RS |
388 | pos = XINT (start); |
389 | if (pos < 0 && -pos <= len) | |
390 | pos = len + pos; | |
391 | else if (0 > pos || pos > len) | |
ca1d1d23 | 392 | args_out_of_range (string, start); |
0c8533c6 | 393 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
394 | } |
395 | ||
910c747a RS |
396 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
397 | XCHAR_TABLE (current_buffer->case_canon_table)->extras[2] | |
398 | = current_buffer->case_eqv_table; | |
399 | ||
487282dc KH |
400 | bufp = compile_pattern (regexp, &search_regs, |
401 | (!NILP (current_buffer->case_fold_search) | |
0190922f | 402 | ? current_buffer->case_canon_table : Qnil), |
0c8533c6 RS |
403 | posix, |
404 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 405 | immediate_quit = 1; |
8bb43c28 | 406 | re_match_object = string; |
177c0ea7 | 407 | |
d5db4077 KR |
408 | val = re_search (bufp, (char *) SDATA (string), |
409 | SBYTES (string), pos_byte, | |
410 | SBYTES (string) - pos_byte, | |
ca1d1d23 JB |
411 | &search_regs); |
412 | immediate_quit = 0; | |
daa37602 | 413 | last_thing_searched = Qt; |
ca1d1d23 JB |
414 | if (val == -2) |
415 | matcher_overflow (); | |
416 | if (val < 0) return Qnil; | |
0c8533c6 RS |
417 | |
418 | for (i = 0; i < search_regs.num_regs; i++) | |
419 | if (search_regs.start[i] >= 0) | |
420 | { | |
421 | search_regs.start[i] | |
422 | = string_byte_to_char (string, search_regs.start[i]); | |
423 | search_regs.end[i] | |
424 | = string_byte_to_char (string, search_regs.end[i]); | |
425 | } | |
426 | ||
427 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 428 | } |
e59a8453 | 429 | |
b819a390 | 430 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 | 431 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
b85acc4b | 432 | Matching ignores case if `case-fold-search' is non-nil. |
8c1a1077 PJ |
433 | If third arg START is non-nil, start search at that index in STRING. |
434 | For index of first char beyond the match, do (match-end 0). | |
435 | `match-end' and `match-beginning' also give indices of substrings | |
2bd2f32d RS |
436 | matched by parenthesis constructs in the pattern. |
437 | ||
438 | You can use the function `match-string' to extract the substrings | |
439 | matched by the parenthesis constructions in REGEXP. */) | |
8c1a1077 | 440 | (regexp, string, start) |
b819a390 RS |
441 | Lisp_Object regexp, string, start; |
442 | { | |
443 | return string_match_1 (regexp, string, start, 0); | |
444 | } | |
445 | ||
446 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
447 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
448 | Find the longest match, in accord with Posix regular expression rules. | |
449 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
450 | If third arg START is non-nil, start search at that index in STRING. | |
451 | For index of first char beyond the match, do (match-end 0). | |
452 | `match-end' and `match-beginning' also give indices of substrings | |
453 | matched by parenthesis constructs in the pattern. */) | |
454 | (regexp, string, start) | |
b819a390 RS |
455 | Lisp_Object regexp, string, start; |
456 | { | |
457 | return string_match_1 (regexp, string, start, 1); | |
458 | } | |
459 | ||
e59a8453 RS |
460 | /* Match REGEXP against STRING, searching all of STRING, |
461 | and return the index of the match, or negative on failure. | |
462 | This does not clobber the match data. */ | |
463 | ||
464 | int | |
465 | fast_string_match (regexp, string) | |
466 | Lisp_Object regexp, string; | |
467 | { | |
468 | int val; | |
487282dc | 469 | struct re_pattern_buffer *bufp; |
e59a8453 | 470 | |
facdc750 RS |
471 | bufp = compile_pattern (regexp, 0, Qnil, |
472 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 473 | immediate_quit = 1; |
8bb43c28 | 474 | re_match_object = string; |
177c0ea7 | 475 | |
d5db4077 KR |
476 | val = re_search (bufp, (char *) SDATA (string), |
477 | SBYTES (string), 0, | |
478 | SBYTES (string), 0); | |
e59a8453 RS |
479 | immediate_quit = 0; |
480 | return val; | |
481 | } | |
5679531d KH |
482 | |
483 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
484 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
485 | This does not clobber the match data. |
486 | We assume that STRING contains single-byte characters. */ | |
5679531d KH |
487 | |
488 | extern Lisp_Object Vascii_downcase_table; | |
489 | ||
490 | int | |
b4577c63 | 491 | fast_c_string_match_ignore_case (regexp, string) |
5679531d | 492 | Lisp_Object regexp; |
96b80561 | 493 | const char *string; |
5679531d KH |
494 | { |
495 | int val; | |
496 | struct re_pattern_buffer *bufp; | |
497 | int len = strlen (string); | |
498 | ||
0c8533c6 | 499 | regexp = string_make_unibyte (regexp); |
b4577c63 | 500 | re_match_object = Qt; |
5679531d | 501 | bufp = compile_pattern (regexp, 0, |
0190922f | 502 | Vascii_canon_table, 0, |
f8bd51c4 | 503 | 0); |
5679531d KH |
504 | immediate_quit = 1; |
505 | val = re_search (bufp, string, len, 0, len, 0); | |
506 | immediate_quit = 0; | |
507 | return val; | |
508 | } | |
be5f4dfb KH |
509 | |
510 | /* Like fast_string_match but ignore case. */ | |
511 | ||
512 | int | |
513 | fast_string_match_ignore_case (regexp, string) | |
514 | Lisp_Object regexp, string; | |
515 | { | |
516 | int val; | |
517 | struct re_pattern_buffer *bufp; | |
518 | ||
0190922f | 519 | bufp = compile_pattern (regexp, 0, Vascii_canon_table, |
be5f4dfb KH |
520 | 0, STRING_MULTIBYTE (string)); |
521 | immediate_quit = 1; | |
522 | re_match_object = string; | |
523 | ||
524 | val = re_search (bufp, (char *) SDATA (string), | |
525 | SBYTES (string), 0, | |
526 | SBYTES (string), 0); | |
527 | immediate_quit = 0; | |
528 | return val; | |
529 | } | |
ca1d1d23 | 530 | \f |
9169c321 JB |
531 | /* The newline cache: remembering which sections of text have no newlines. */ |
532 | ||
533 | /* If the user has requested newline caching, make sure it's on. | |
534 | Otherwise, make sure it's off. | |
535 | This is our cheezy way of associating an action with the change of | |
536 | state of a buffer-local variable. */ | |
537 | static void | |
538 | newline_cache_on_off (buf) | |
539 | struct buffer *buf; | |
540 | { | |
541 | if (NILP (buf->cache_long_line_scans)) | |
542 | { | |
543 | /* It should be off. */ | |
544 | if (buf->newline_cache) | |
545 | { | |
546 | free_region_cache (buf->newline_cache); | |
547 | buf->newline_cache = 0; | |
548 | } | |
549 | } | |
550 | else | |
551 | { | |
552 | /* It should be on. */ | |
553 | if (buf->newline_cache == 0) | |
554 | buf->newline_cache = new_region_cache (); | |
555 | } | |
556 | } | |
557 | ||
558 | \f | |
559 | /* Search for COUNT instances of the character TARGET between START and END. | |
560 | ||
561 | If COUNT is positive, search forwards; END must be >= START. | |
562 | If COUNT is negative, search backwards for the -COUNTth instance; | |
563 | END must be <= START. | |
564 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
565 | ||
566 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
567 | direction indicated by COUNT. | |
ffd56f97 JB |
568 | |
569 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
a9f2a45f | 570 | position past the COUNTth match. Note that for reverse motion |
5bfe95c9 | 571 | this is not the same as the usual convention for Emacs motion commands. |
ffd56f97 | 572 | |
9169c321 JB |
573 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
574 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 575 | |
087a5f81 RS |
576 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
577 | except when inside redisplay. */ | |
578 | ||
dfcf069d | 579 | int |
9169c321 JB |
580 | scan_buffer (target, start, end, count, shortage, allow_quit) |
581 | register int target; | |
582 | int start, end; | |
583 | int count; | |
584 | int *shortage; | |
087a5f81 | 585 | int allow_quit; |
ca1d1d23 | 586 | { |
9169c321 | 587 | struct region_cache *newline_cache; |
177c0ea7 | 588 | int direction; |
ffd56f97 | 589 | |
9169c321 JB |
590 | if (count > 0) |
591 | { | |
592 | direction = 1; | |
593 | if (! end) end = ZV; | |
594 | } | |
595 | else | |
596 | { | |
597 | direction = -1; | |
598 | if (! end) end = BEGV; | |
599 | } | |
ffd56f97 | 600 | |
9169c321 JB |
601 | newline_cache_on_off (current_buffer); |
602 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
603 | |
604 | if (shortage != 0) | |
605 | *shortage = 0; | |
606 | ||
087a5f81 | 607 | immediate_quit = allow_quit; |
ca1d1d23 | 608 | |
ffd56f97 | 609 | if (count > 0) |
9169c321 | 610 | while (start != end) |
ca1d1d23 | 611 | { |
9169c321 JB |
612 | /* Our innermost scanning loop is very simple; it doesn't know |
613 | about gaps, buffer ends, or the newline cache. ceiling is | |
614 | the position of the last character before the next such | |
615 | obstacle --- the last character the dumb search loop should | |
616 | examine. */ | |
fa8ed3e0 RS |
617 | int ceiling_byte = CHAR_TO_BYTE (end) - 1; |
618 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 619 | int tem; |
9169c321 JB |
620 | |
621 | /* If we're looking for a newline, consult the newline cache | |
622 | to see where we can avoid some scanning. */ | |
623 | if (target == '\n' && newline_cache) | |
624 | { | |
625 | int next_change; | |
626 | immediate_quit = 0; | |
627 | while (region_cache_forward | |
fa8ed3e0 RS |
628 | (current_buffer, newline_cache, start_byte, &next_change)) |
629 | start_byte = next_change; | |
cbe0db0d | 630 | immediate_quit = allow_quit; |
9169c321 | 631 | |
fa8ed3e0 RS |
632 | /* START should never be after END. */ |
633 | if (start_byte > ceiling_byte) | |
634 | start_byte = ceiling_byte; | |
9169c321 JB |
635 | |
636 | /* Now the text after start is an unknown region, and | |
637 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 638 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
639 | } |
640 | ||
641 | /* The dumb loop can only scan text stored in contiguous | |
642 | bytes. BUFFER_CEILING_OF returns the last character | |
643 | position that is contiguous, so the ceiling is the | |
644 | position after that. */ | |
67ce527d KH |
645 | tem = BUFFER_CEILING_OF (start_byte); |
646 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
647 | |
648 | { | |
177c0ea7 | 649 | /* The termination address of the dumb loop. */ |
fa8ed3e0 RS |
650 | register unsigned char *ceiling_addr |
651 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
652 | register unsigned char *cursor | |
653 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
654 | unsigned char *base = cursor; |
655 | ||
656 | while (cursor < ceiling_addr) | |
657 | { | |
658 | unsigned char *scan_start = cursor; | |
659 | ||
660 | /* The dumb loop. */ | |
661 | while (*cursor != target && ++cursor < ceiling_addr) | |
662 | ; | |
663 | ||
664 | /* If we're looking for newlines, cache the fact that | |
665 | the region from start to cursor is free of them. */ | |
666 | if (target == '\n' && newline_cache) | |
667 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
668 | start_byte + scan_start - base, |
669 | start_byte + cursor - base); | |
9169c321 JB |
670 | |
671 | /* Did we find the target character? */ | |
672 | if (cursor < ceiling_addr) | |
673 | { | |
674 | if (--count == 0) | |
675 | { | |
676 | immediate_quit = 0; | |
fa8ed3e0 | 677 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
678 | } |
679 | cursor++; | |
680 | } | |
681 | } | |
682 | ||
fa8ed3e0 | 683 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 684 | } |
ca1d1d23 JB |
685 | } |
686 | else | |
9169c321 JB |
687 | while (start > end) |
688 | { | |
689 | /* The last character to check before the next obstacle. */ | |
fa8ed3e0 RS |
690 | int ceiling_byte = CHAR_TO_BYTE (end); |
691 | int start_byte = CHAR_TO_BYTE (start); | |
67ce527d | 692 | int tem; |
9169c321 JB |
693 | |
694 | /* Consult the newline cache, if appropriate. */ | |
695 | if (target == '\n' && newline_cache) | |
696 | { | |
697 | int next_change; | |
698 | immediate_quit = 0; | |
699 | while (region_cache_backward | |
fa8ed3e0 RS |
700 | (current_buffer, newline_cache, start_byte, &next_change)) |
701 | start_byte = next_change; | |
cbe0db0d | 702 | immediate_quit = allow_quit; |
9169c321 JB |
703 | |
704 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
705 | if (start_byte <= ceiling_byte) |
706 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
707 | |
708 | /* Now the text before start is an unknown region, and | |
709 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 710 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
711 | } |
712 | ||
713 | /* Stop scanning before the gap. */ | |
67ce527d KH |
714 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
715 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
716 | |
717 | { | |
718 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
719 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
720 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
721 | unsigned char *base = cursor; |
722 | ||
723 | while (cursor >= ceiling_addr) | |
724 | { | |
725 | unsigned char *scan_start = cursor; | |
726 | ||
727 | while (*cursor != target && --cursor >= ceiling_addr) | |
728 | ; | |
729 | ||
730 | /* If we're looking for newlines, cache the fact that | |
731 | the region from after the cursor to start is free of them. */ | |
732 | if (target == '\n' && newline_cache) | |
733 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
734 | start_byte + cursor - base, |
735 | start_byte + scan_start - base); | |
9169c321 JB |
736 | |
737 | /* Did we find the target character? */ | |
738 | if (cursor >= ceiling_addr) | |
739 | { | |
740 | if (++count >= 0) | |
741 | { | |
742 | immediate_quit = 0; | |
fa8ed3e0 | 743 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
744 | } |
745 | cursor--; | |
746 | } | |
747 | } | |
748 | ||
fa8ed3e0 | 749 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
750 | } |
751 | } | |
752 | ||
ca1d1d23 JB |
753 | immediate_quit = 0; |
754 | if (shortage != 0) | |
ffd56f97 | 755 | *shortage = count * direction; |
9169c321 | 756 | return start; |
ca1d1d23 | 757 | } |
fa8ed3e0 RS |
758 | \f |
759 | /* Search for COUNT instances of a line boundary, which means either a | |
760 | newline or (if selective display enabled) a carriage return. | |
761 | Start at START. If COUNT is negative, search backwards. | |
762 | ||
763 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
764 | ||
765 | If we find COUNT instances. we position after (always after, | |
766 | even if scanning backwards) the COUNTth match, and return 0. | |
767 | ||
768 | If we don't find COUNT instances before reaching the end of the | |
769 | buffer (or the beginning, if scanning backwards), we return | |
770 | the number of line boundaries left unfound, and position at | |
771 | the limit we bumped up against. | |
772 | ||
773 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 774 | except in special cases. */ |
ca1d1d23 | 775 | |
63fa018d | 776 | int |
fa8ed3e0 RS |
777 | scan_newline (start, start_byte, limit, limit_byte, count, allow_quit) |
778 | int start, start_byte; | |
779 | int limit, limit_byte; | |
780 | register int count; | |
781 | int allow_quit; | |
63fa018d | 782 | { |
fa8ed3e0 RS |
783 | int direction = ((count > 0) ? 1 : -1); |
784 | ||
785 | register unsigned char *cursor; | |
786 | unsigned char *base; | |
787 | ||
788 | register int ceiling; | |
789 | register unsigned char *ceiling_addr; | |
790 | ||
d5d57b92 RS |
791 | int old_immediate_quit = immediate_quit; |
792 | ||
fa8ed3e0 RS |
793 | /* The code that follows is like scan_buffer |
794 | but checks for either newline or carriage return. */ | |
795 | ||
d5d57b92 RS |
796 | if (allow_quit) |
797 | immediate_quit++; | |
fa8ed3e0 RS |
798 | |
799 | start_byte = CHAR_TO_BYTE (start); | |
800 | ||
801 | if (count > 0) | |
802 | { | |
803 | while (start_byte < limit_byte) | |
804 | { | |
805 | ceiling = BUFFER_CEILING_OF (start_byte); | |
806 | ceiling = min (limit_byte - 1, ceiling); | |
807 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
808 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
809 | while (1) | |
810 | { | |
811 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
812 | ; | |
813 | ||
814 | if (cursor != ceiling_addr) | |
815 | { | |
816 | if (--count == 0) | |
817 | { | |
d5d57b92 | 818 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
819 | start_byte = start_byte + cursor - base + 1; |
820 | start = BYTE_TO_CHAR (start_byte); | |
821 | TEMP_SET_PT_BOTH (start, start_byte); | |
822 | return 0; | |
823 | } | |
824 | else | |
825 | if (++cursor == ceiling_addr) | |
826 | break; | |
827 | } | |
828 | else | |
829 | break; | |
830 | } | |
831 | start_byte += cursor - base; | |
832 | } | |
833 | } | |
834 | else | |
835 | { | |
fa8ed3e0 RS |
836 | while (start_byte > limit_byte) |
837 | { | |
838 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
839 | ceiling = max (limit_byte, ceiling); | |
840 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
841 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
842 | while (1) | |
843 | { | |
844 | while (--cursor != ceiling_addr && *cursor != '\n') | |
845 | ; | |
846 | ||
847 | if (cursor != ceiling_addr) | |
848 | { | |
849 | if (++count == 0) | |
850 | { | |
d5d57b92 | 851 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
852 | /* Return the position AFTER the match we found. */ |
853 | start_byte = start_byte + cursor - base + 1; | |
854 | start = BYTE_TO_CHAR (start_byte); | |
855 | TEMP_SET_PT_BOTH (start, start_byte); | |
856 | return 0; | |
857 | } | |
858 | } | |
859 | else | |
860 | break; | |
861 | } | |
862 | /* Here we add 1 to compensate for the last decrement | |
863 | of CURSOR, which took it past the valid range. */ | |
864 | start_byte += cursor - base + 1; | |
865 | } | |
866 | } | |
867 | ||
868 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 869 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
870 | |
871 | return count * direction; | |
63fa018d RS |
872 | } |
873 | ||
ca1d1d23 | 874 | int |
fa8ed3e0 | 875 | find_next_newline_no_quit (from, cnt) |
ca1d1d23 JB |
876 | register int from, cnt; |
877 | { | |
fa8ed3e0 | 878 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
9169c321 JB |
879 | } |
880 | ||
9169c321 JB |
881 | /* Like find_next_newline, but returns position before the newline, |
882 | not after, and only search up to TO. This isn't just | |
883 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 884 | |
9169c321 JB |
885 | int |
886 | find_before_next_newline (from, to, cnt) | |
cbe0db0d | 887 | int from, to, cnt; |
9169c321 JB |
888 | { |
889 | int shortage; | |
890 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
891 | ||
892 | if (shortage == 0) | |
893 | pos--; | |
177c0ea7 | 894 | |
9169c321 | 895 | return pos; |
ca1d1d23 JB |
896 | } |
897 | \f | |
ca1d1d23 JB |
898 | /* Subroutines of Lisp buffer search functions. */ |
899 | ||
900 | static Lisp_Object | |
b819a390 | 901 | search_command (string, bound, noerror, count, direction, RE, posix) |
ca1d1d23 JB |
902 | Lisp_Object string, bound, noerror, count; |
903 | int direction; | |
904 | int RE; | |
b819a390 | 905 | int posix; |
ca1d1d23 JB |
906 | { |
907 | register int np; | |
9f43ad85 | 908 | int lim, lim_byte; |
ca1d1d23 JB |
909 | int n = direction; |
910 | ||
911 | if (!NILP (count)) | |
912 | { | |
b7826503 | 913 | CHECK_NUMBER (count); |
ca1d1d23 JB |
914 | n *= XINT (count); |
915 | } | |
916 | ||
b7826503 | 917 | CHECK_STRING (string); |
ca1d1d23 | 918 | if (NILP (bound)) |
9f43ad85 RS |
919 | { |
920 | if (n > 0) | |
921 | lim = ZV, lim_byte = ZV_BYTE; | |
922 | else | |
923 | lim = BEGV, lim_byte = BEGV_BYTE; | |
924 | } | |
ca1d1d23 JB |
925 | else |
926 | { | |
b7826503 | 927 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 928 | lim = XINT (bound); |
6ec8bbd2 | 929 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
930 | error ("Invalid search bound (wrong side of point)"); |
931 | if (lim > ZV) | |
9f43ad85 | 932 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 933 | else if (lim < BEGV) |
9f43ad85 | 934 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
935 | else |
936 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
937 | } |
938 | ||
910c747a RS |
939 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
940 | XCHAR_TABLE (current_buffer->case_canon_table)->extras[2] | |
941 | = current_buffer->case_eqv_table; | |
942 | ||
9f43ad85 | 943 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
ca1d1d23 | 944 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 945 | ? current_buffer->case_canon_table |
3135e9fd | 946 | : Qnil), |
ca1d1d23 | 947 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 948 | ? current_buffer->case_eqv_table |
3135e9fd | 949 | : Qnil), |
b819a390 | 950 | posix); |
ca1d1d23 JB |
951 | if (np <= 0) |
952 | { | |
953 | if (NILP (noerror)) | |
06f77a8a KS |
954 | xsignal1 (Qsearch_failed, string); |
955 | ||
ca1d1d23 JB |
956 | if (!EQ (noerror, Qt)) |
957 | { | |
958 | if (lim < BEGV || lim > ZV) | |
959 | abort (); | |
9f43ad85 | 960 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
961 | return Qnil; |
962 | #if 0 /* This would be clean, but maybe programs depend on | |
963 | a value of nil here. */ | |
481399bf | 964 | np = lim; |
a5f217b8 | 965 | #endif |
ca1d1d23 | 966 | } |
481399bf RS |
967 | else |
968 | return Qnil; | |
ca1d1d23 JB |
969 | } |
970 | ||
971 | if (np < BEGV || np > ZV) | |
972 | abort (); | |
973 | ||
974 | SET_PT (np); | |
975 | ||
976 | return make_number (np); | |
977 | } | |
978 | \f | |
fa8ed3e0 RS |
979 | /* Return 1 if REGEXP it matches just one constant string. */ |
980 | ||
b6d6a51c KH |
981 | static int |
982 | trivial_regexp_p (regexp) | |
983 | Lisp_Object regexp; | |
984 | { | |
d5db4077 KR |
985 | int len = SBYTES (regexp); |
986 | unsigned char *s = SDATA (regexp); | |
b6d6a51c KH |
987 | while (--len >= 0) |
988 | { | |
989 | switch (*s++) | |
990 | { | |
991 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
992 | return 0; | |
993 | case '\\': | |
994 | if (--len < 0) | |
995 | return 0; | |
996 | switch (*s++) | |
997 | { | |
998 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
999 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 1000 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 1001 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 1002 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
1003 | case '6': case '7': case '8': case '9': |
1004 | return 0; | |
1005 | } | |
1006 | } | |
1007 | } | |
1008 | return 1; | |
1009 | } | |
1010 | ||
ca325161 | 1011 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 1012 | starting at position POS and stopping at position LIM, |
b819a390 | 1013 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
1014 | a regular expression if RE is true. |
1015 | ||
1016 | If N is positive, searching is forward and LIM must be greater than POS. | |
1017 | If N is negative, searching is backward and LIM must be less than POS. | |
1018 | ||
facdc750 | 1019 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 1020 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
1021 | (if searching backward) or the end (if searching forward). |
1022 | ||
1023 | POSIX is nonzero if we want full backtracking (POSIX style) | |
1024 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 1025 | |
aff2ce94 RS |
1026 | #define TRANSLATE(out, trt, d) \ |
1027 | do \ | |
1028 | { \ | |
1029 | if (! NILP (trt)) \ | |
1030 | { \ | |
1031 | Lisp_Object temp; \ | |
1032 | temp = Faref (trt, make_number (d)); \ | |
1033 | if (INTEGERP (temp)) \ | |
1034 | out = XINT (temp); \ | |
1035 | else \ | |
1036 | out = d; \ | |
1037 | } \ | |
1038 | else \ | |
1039 | out = d; \ | |
1040 | } \ | |
1041 | while (0) | |
facdc750 | 1042 | |
b819a390 | 1043 | static int |
9f43ad85 RS |
1044 | search_buffer (string, pos, pos_byte, lim, lim_byte, n, |
1045 | RE, trt, inverse_trt, posix) | |
ca1d1d23 JB |
1046 | Lisp_Object string; |
1047 | int pos; | |
9f43ad85 | 1048 | int pos_byte; |
ca1d1d23 | 1049 | int lim; |
9f43ad85 | 1050 | int lim_byte; |
ca1d1d23 JB |
1051 | int n; |
1052 | int RE; | |
facdc750 RS |
1053 | Lisp_Object trt; |
1054 | Lisp_Object inverse_trt; | |
b819a390 | 1055 | int posix; |
ca1d1d23 | 1056 | { |
d5db4077 KR |
1057 | int len = SCHARS (string); |
1058 | int len_byte = SBYTES (string); | |
facdc750 | 1059 | register int i; |
ca1d1d23 | 1060 | |
7074fde6 FP |
1061 | if (running_asynch_code) |
1062 | save_search_regs (); | |
1063 | ||
a7e4cdde | 1064 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1065 | /* Null string is found at starting position. */ |
a7e4cdde | 1066 | if (len == 0 || n == 0) |
ca325161 | 1067 | { |
0353b28f | 1068 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1069 | return pos; |
1070 | } | |
3f57a499 | 1071 | |
41a33295 | 1072 | if (RE && !(trivial_regexp_p (string) && NILP (Vsearch_spaces_regexp))) |
ca1d1d23 | 1073 | { |
facdc750 RS |
1074 | unsigned char *p1, *p2; |
1075 | int s1, s2; | |
487282dc KH |
1076 | struct re_pattern_buffer *bufp; |
1077 | ||
0c8533c6 RS |
1078 | bufp = compile_pattern (string, &search_regs, trt, posix, |
1079 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 | 1080 | |
ca1d1d23 JB |
1081 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1082 | because letting this function finish | |
1083 | can take too long. */ | |
1084 | QUIT; /* Do a pending quit right away, | |
1085 | to avoid paradoxical behavior */ | |
1086 | /* Get pointers and sizes of the two strings | |
1087 | that make up the visible portion of the buffer. */ | |
1088 | ||
1089 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1090 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1091 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1092 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1093 | if (s1 < 0) |
1094 | { | |
1095 | p2 = p1; | |
fa8ed3e0 | 1096 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1097 | s1 = 0; |
1098 | } | |
1099 | if (s2 < 0) | |
1100 | { | |
fa8ed3e0 | 1101 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1102 | s2 = 0; |
1103 | } | |
8bb43c28 | 1104 | re_match_object = Qnil; |
177c0ea7 | 1105 | |
ca1d1d23 JB |
1106 | while (n < 0) |
1107 | { | |
42db823b | 1108 | int val; |
487282dc | 1109 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1110 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1111 | &search_regs, | |
42db823b | 1112 | /* Don't allow match past current point */ |
4996330b | 1113 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1114 | if (val == -2) |
b6d6a51c KH |
1115 | { |
1116 | matcher_overflow (); | |
1117 | } | |
ca1d1d23 JB |
1118 | if (val >= 0) |
1119 | { | |
26aff150 | 1120 | pos_byte = search_regs.start[0] + BEGV_BYTE; |
4746118a | 1121 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1122 | if (search_regs.start[i] >= 0) |
1123 | { | |
fa8ed3e0 RS |
1124 | search_regs.start[i] |
1125 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1126 | search_regs.end[i] | |
1127 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1128 | } |
a3668d92 | 1129 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1130 | /* Set pos to the new position. */ |
1131 | pos = search_regs.start[0]; | |
1132 | } | |
1133 | else | |
1134 | { | |
1135 | immediate_quit = 0; | |
1136 | return (n); | |
1137 | } | |
1138 | n++; | |
1139 | } | |
1140 | while (n > 0) | |
1141 | { | |
42db823b | 1142 | int val; |
487282dc | 1143 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b KH |
1144 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
1145 | &search_regs, | |
1146 | lim_byte - BEGV_BYTE); | |
ca1d1d23 | 1147 | if (val == -2) |
b6d6a51c KH |
1148 | { |
1149 | matcher_overflow (); | |
1150 | } | |
ca1d1d23 JB |
1151 | if (val >= 0) |
1152 | { | |
26aff150 | 1153 | pos_byte = search_regs.end[0] + BEGV_BYTE; |
4746118a | 1154 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
1155 | if (search_regs.start[i] >= 0) |
1156 | { | |
fa8ed3e0 RS |
1157 | search_regs.start[i] |
1158 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1159 | search_regs.end[i] | |
1160 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
ca1d1d23 | 1161 | } |
a3668d92 | 1162 | XSETBUFFER (last_thing_searched, current_buffer); |
ca1d1d23 JB |
1163 | pos = search_regs.end[0]; |
1164 | } | |
1165 | else | |
1166 | { | |
1167 | immediate_quit = 0; | |
1168 | return (0 - n); | |
1169 | } | |
1170 | n--; | |
1171 | } | |
1172 | immediate_quit = 0; | |
1173 | return (pos); | |
1174 | } | |
1175 | else /* non-RE case */ | |
1176 | { | |
facdc750 RS |
1177 | unsigned char *raw_pattern, *pat; |
1178 | int raw_pattern_size; | |
1179 | int raw_pattern_size_byte; | |
1180 | unsigned char *patbuf; | |
1181 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); | |
a967ed62 | 1182 | unsigned char *base_pat; |
49e44e81 KH |
1183 | /* Set to positive if we find a non-ASCII char that need |
1184 | translation. Otherwise set to zero later. */ | |
1185 | int char_base = -1; | |
040272ce | 1186 | int boyer_moore_ok = 1; |
facdc750 RS |
1187 | |
1188 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1189 | We must convert PATTERN to match that, or we will not really | |
1190 | find things right. */ | |
1191 | ||
1192 | if (multibyte == STRING_MULTIBYTE (string)) | |
1193 | { | |
d5db4077 KR |
1194 | raw_pattern = (unsigned char *) SDATA (string); |
1195 | raw_pattern_size = SCHARS (string); | |
1196 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1197 | } |
1198 | else if (multibyte) | |
1199 | { | |
d5db4077 | 1200 | raw_pattern_size = SCHARS (string); |
facdc750 | 1201 | raw_pattern_size_byte |
d5db4077 | 1202 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1203 | raw_pattern_size); |
7276d3d8 | 1204 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1205 | copy_text (SDATA (string), raw_pattern, |
1206 | SCHARS (string), 0, 1); | |
facdc750 RS |
1207 | } |
1208 | else | |
1209 | { | |
1210 | /* Converting multibyte to single-byte. | |
1211 | ||
1212 | ??? Perhaps this conversion should be done in a special way | |
1213 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1214 | so that only the multibyte chars which really correspond to | |
1215 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1216 | raw_pattern_size = SCHARS (string); |
1217 | raw_pattern_size_byte = SCHARS (string); | |
7276d3d8 | 1218 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
1219 | copy_text (SDATA (string), raw_pattern, |
1220 | SBYTES (string), 1, 0); | |
facdc750 RS |
1221 | } |
1222 | ||
1223 | /* Copy and optionally translate the pattern. */ | |
1224 | len = raw_pattern_size; | |
1225 | len_byte = raw_pattern_size_byte; | |
f5f7578b | 1226 | patbuf = (unsigned char *) alloca (len * MAX_MULTIBYTE_LENGTH); |
facdc750 RS |
1227 | pat = patbuf; |
1228 | base_pat = raw_pattern; | |
1229 | if (multibyte) | |
1230 | { | |
a17ae96a KH |
1231 | /* Fill patbuf by translated characters in STRING while |
1232 | checking if we can use boyer-moore search. If TRT is | |
1233 | non-nil, we can use boyer-moore search only if TRT can be | |
1234 | represented by the byte array of 256 elements. For that, | |
1235 | all non-ASCII case-equivalents of all case-senstive | |
1236 | characters in STRING must belong to the same charset and | |
1237 | row. */ | |
1238 | ||
facdc750 RS |
1239 | while (--len >= 0) |
1240 | { | |
a17ae96a | 1241 | unsigned char str_base[MAX_MULTIBYTE_LENGTH], *str; |
aff2ce94 | 1242 | int c, translated, inverse; |
a17ae96a | 1243 | int in_charlen, charlen; |
facdc750 RS |
1244 | |
1245 | /* If we got here and the RE flag is set, it's because we're | |
1246 | dealing with a regexp known to be trivial, so the backslash | |
1247 | just quotes the next character. */ | |
1248 | if (RE && *base_pat == '\\') | |
1249 | { | |
1250 | len--; | |
62221f22 | 1251 | raw_pattern_size--; |
facdc750 RS |
1252 | len_byte--; |
1253 | base_pat++; | |
1254 | } | |
1255 | ||
1256 | c = STRING_CHAR_AND_LENGTH (base_pat, len_byte, in_charlen); | |
040272ce | 1257 | |
a17ae96a | 1258 | if (NILP (trt)) |
facdc750 | 1259 | { |
a17ae96a KH |
1260 | str = base_pat; |
1261 | charlen = in_charlen; | |
1262 | } | |
1263 | else | |
1264 | { | |
1265 | /* Translate the character. */ | |
1266 | TRANSLATE (translated, trt, c); | |
1267 | charlen = CHAR_STRING (translated, str_base); | |
1268 | str = str_base; | |
1269 | ||
1270 | /* Check if C has any other case-equivalents. */ | |
1271 | TRANSLATE (inverse, inverse_trt, c); | |
1272 | /* If so, check if we can use boyer-moore. */ | |
1273 | if (c != inverse && boyer_moore_ok) | |
1274 | { | |
1275 | /* Check if all equivalents belong to the same | |
1276 | group of characters. Note that the check of C | |
49e44e81 KH |
1277 | itself is done by the last iteration. */ |
1278 | int this_char_base = -1; | |
1279 | ||
1280 | while (boyer_moore_ok) | |
a17ae96a | 1281 | { |
49e44e81 | 1282 | if (ASCII_BYTE_P (inverse)) |
a17ae96a | 1283 | { |
49e44e81 KH |
1284 | if (this_char_base > 0) |
1285 | boyer_moore_ok = 0; | |
1286 | else | |
a17ae96a | 1287 | { |
49e44e81 KH |
1288 | this_char_base = 0; |
1289 | if (char_base < 0) | |
1290 | char_base = this_char_base; | |
a17ae96a KH |
1291 | } |
1292 | } | |
49e44e81 KH |
1293 | else if (CHAR_BYTE8_P (inverse)) |
1294 | /* Boyer-moore search can't handle a | |
1295 | translation of an eight-bit | |
1296 | character. */ | |
1297 | boyer_moore_ok = 0; | |
1298 | else if (this_char_base < 0) | |
1299 | { | |
1300 | this_char_base = inverse & ~0x3F; | |
1301 | if (char_base < 0) | |
1302 | char_base = this_char_base; | |
1303 | else if (char_base > 0 | |
1304 | && this_char_base != char_base) | |
1305 | boyer_moore_ok = 0; | |
1306 | } | |
1307 | else if ((inverse & ~0x3F) != this_char_base) | |
1308 | boyer_moore_ok = 0; | |
a17ae96a KH |
1309 | if (c == inverse) |
1310 | break; | |
1311 | TRANSLATE (inverse, inverse_trt, inverse); | |
1312 | } | |
1313 | } | |
aff2ce94 | 1314 | } |
49e44e81 KH |
1315 | if (char_base < 0) |
1316 | char_base = 0; | |
facdc750 RS |
1317 | |
1318 | /* Store this character into the translated pattern. */ | |
a17ae96a KH |
1319 | bcopy (str, pat, charlen); |
1320 | pat += charlen; | |
facdc750 RS |
1321 | base_pat += in_charlen; |
1322 | len_byte -= in_charlen; | |
1323 | } | |
1324 | } | |
1325 | else | |
1326 | { | |
040272ce | 1327 | /* Unibyte buffer. */ |
a17ae96a | 1328 | char_base = 0; |
facdc750 RS |
1329 | while (--len >= 0) |
1330 | { | |
040272ce | 1331 | int c, translated; |
facdc750 RS |
1332 | |
1333 | /* If we got here and the RE flag is set, it's because we're | |
1334 | dealing with a regexp known to be trivial, so the backslash | |
1335 | just quotes the next character. */ | |
1336 | if (RE && *base_pat == '\\') | |
1337 | { | |
1338 | len--; | |
0190922f | 1339 | raw_pattern_size--; |
facdc750 RS |
1340 | base_pat++; |
1341 | } | |
1342 | c = *base_pat++; | |
aff2ce94 | 1343 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1344 | *pat++ = translated; |
1345 | } | |
1346 | } | |
1347 | ||
1348 | len_byte = pat - patbuf; | |
1349 | len = raw_pattern_size; | |
1350 | pat = base_pat = patbuf; | |
1351 | ||
040272ce | 1352 | if (boyer_moore_ok) |
facdc750 | 1353 | return boyer_moore (n, pat, len, len_byte, trt, inverse_trt, |
aff2ce94 | 1354 | pos, pos_byte, lim, lim_byte, |
a17ae96a | 1355 | char_base); |
facdc750 RS |
1356 | else |
1357 | return simple_search (n, pat, len, len_byte, trt, | |
1358 | pos, pos_byte, lim, lim_byte); | |
1359 | } | |
1360 | } | |
1361 | \f | |
1362 | /* Do a simple string search N times for the string PAT, | |
1363 | whose length is LEN/LEN_BYTE, | |
1364 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1365 | TRT is the translation table. | |
f8bd51c4 | 1366 | |
facdc750 RS |
1367 | Return the character position where the match is found. |
1368 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1369 | |
facdc750 RS |
1370 | This kind of search works regardless of what is in PAT and |
1371 | regardless of what is in TRT. It is used in cases where | |
1372 | boyer_moore cannot work. */ | |
1373 | ||
1374 | static int | |
1375 | simple_search (n, pat, len, len_byte, trt, pos, pos_byte, lim, lim_byte) | |
1376 | int n; | |
1377 | unsigned char *pat; | |
1378 | int len, len_byte; | |
1379 | Lisp_Object trt; | |
1380 | int pos, pos_byte; | |
1381 | int lim, lim_byte; | |
1382 | { | |
1383 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); | |
ab228c24 | 1384 | int forward = n > 0; |
49e44e81 KH |
1385 | /* Number of buffer bytes matched. Note that this may be different |
1386 | from len_byte in a multibyte buffer. */ | |
1387 | int match_byte; | |
facdc750 RS |
1388 | |
1389 | if (lim > pos && multibyte) | |
1390 | while (n > 0) | |
1391 | { | |
1392 | while (1) | |
f8bd51c4 | 1393 | { |
facdc750 RS |
1394 | /* Try matching at position POS. */ |
1395 | int this_pos = pos; | |
1396 | int this_pos_byte = pos_byte; | |
1397 | int this_len = len; | |
1398 | int this_len_byte = len_byte; | |
1399 | unsigned char *p = pat; | |
a9469498 | 1400 | if (pos + len > lim || pos_byte + len_byte > lim_byte) |
facdc750 RS |
1401 | goto stop; |
1402 | ||
1403 | while (this_len > 0) | |
1404 | { | |
1405 | int charlen, buf_charlen; | |
ab228c24 | 1406 | int pat_ch, buf_ch; |
facdc750 | 1407 | |
ab228c24 | 1408 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1409 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1410 | ZV_BYTE - this_pos_byte, | |
1411 | buf_charlen); | |
aff2ce94 | 1412 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1413 | |
1414 | if (buf_ch != pat_ch) | |
1415 | break; | |
ab228c24 RS |
1416 | |
1417 | this_len_byte -= charlen; | |
1418 | this_len--; | |
1419 | p += charlen; | |
1420 | ||
1421 | this_pos_byte += buf_charlen; | |
1422 | this_pos++; | |
facdc750 RS |
1423 | } |
1424 | ||
1425 | if (this_len == 0) | |
1426 | { | |
49e44e81 | 1427 | match_byte = this_pos_byte - pos_byte; |
facdc750 | 1428 | pos += len; |
49e44e81 | 1429 | pos_byte += match_byte; |
facdc750 RS |
1430 | break; |
1431 | } | |
1432 | ||
1433 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1434 | } |
facdc750 RS |
1435 | |
1436 | n--; | |
1437 | } | |
1438 | else if (lim > pos) | |
1439 | while (n > 0) | |
1440 | { | |
1441 | while (1) | |
f8bd51c4 | 1442 | { |
facdc750 RS |
1443 | /* Try matching at position POS. */ |
1444 | int this_pos = pos; | |
1445 | int this_len = len; | |
1446 | unsigned char *p = pat; | |
1447 | ||
1448 | if (pos + len > lim) | |
1449 | goto stop; | |
1450 | ||
1451 | while (this_len > 0) | |
1452 | { | |
1453 | int pat_ch = *p++; | |
1454 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1455 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1456 | |
1457 | if (buf_ch != pat_ch) | |
1458 | break; | |
ab228c24 RS |
1459 | |
1460 | this_len--; | |
1461 | this_pos++; | |
facdc750 RS |
1462 | } |
1463 | ||
1464 | if (this_len == 0) | |
1465 | { | |
49e44e81 | 1466 | match_byte = len; |
facdc750 RS |
1467 | pos += len; |
1468 | break; | |
1469 | } | |
1470 | ||
1471 | pos++; | |
f8bd51c4 | 1472 | } |
facdc750 RS |
1473 | |
1474 | n--; | |
1475 | } | |
1476 | /* Backwards search. */ | |
1477 | else if (lim < pos && multibyte) | |
1478 | while (n < 0) | |
1479 | { | |
1480 | while (1) | |
f8bd51c4 | 1481 | { |
facdc750 RS |
1482 | /* Try matching at position POS. */ |
1483 | int this_pos = pos - len; | |
3dcf0066 | 1484 | int this_pos_byte; |
facdc750 RS |
1485 | int this_len = len; |
1486 | int this_len_byte = len_byte; | |
1487 | unsigned char *p = pat; | |
1488 | ||
3dcf0066 | 1489 | if (this_pos < lim || (pos_byte - len_byte) < lim_byte) |
facdc750 | 1490 | goto stop; |
3dcf0066 | 1491 | this_pos_byte = CHAR_TO_BYTE (this_pos); |
49e44e81 | 1492 | match_byte = pos_byte - this_pos_byte; |
facdc750 RS |
1493 | |
1494 | while (this_len > 0) | |
1495 | { | |
1496 | int charlen, buf_charlen; | |
ab228c24 | 1497 | int pat_ch, buf_ch; |
facdc750 | 1498 | |
ab228c24 | 1499 | pat_ch = STRING_CHAR_AND_LENGTH (p, this_len_byte, charlen); |
facdc750 RS |
1500 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
1501 | ZV_BYTE - this_pos_byte, | |
1502 | buf_charlen); | |
aff2ce94 | 1503 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1504 | |
1505 | if (buf_ch != pat_ch) | |
1506 | break; | |
ab228c24 RS |
1507 | |
1508 | this_len_byte -= charlen; | |
1509 | this_len--; | |
1510 | p += charlen; | |
1511 | this_pos_byte += buf_charlen; | |
1512 | this_pos++; | |
facdc750 RS |
1513 | } |
1514 | ||
1515 | if (this_len == 0) | |
1516 | { | |
1517 | pos -= len; | |
49e44e81 | 1518 | pos_byte -= match_byte; |
facdc750 RS |
1519 | break; |
1520 | } | |
1521 | ||
1522 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1523 | } |
1524 | ||
facdc750 RS |
1525 | n++; |
1526 | } | |
1527 | else if (lim < pos) | |
1528 | while (n < 0) | |
1529 | { | |
1530 | while (1) | |
b6d6a51c | 1531 | { |
facdc750 RS |
1532 | /* Try matching at position POS. */ |
1533 | int this_pos = pos - len; | |
1534 | int this_len = len; | |
1535 | unsigned char *p = pat; | |
1536 | ||
a9469498 | 1537 | if (this_pos < lim) |
facdc750 RS |
1538 | goto stop; |
1539 | ||
1540 | while (this_len > 0) | |
1541 | { | |
1542 | int pat_ch = *p++; | |
1543 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1544 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1545 | |
1546 | if (buf_ch != pat_ch) | |
1547 | break; | |
ab228c24 RS |
1548 | this_len--; |
1549 | this_pos++; | |
facdc750 RS |
1550 | } |
1551 | ||
1552 | if (this_len == 0) | |
b6d6a51c | 1553 | { |
49e44e81 | 1554 | match_byte = len; |
facdc750 RS |
1555 | pos -= len; |
1556 | break; | |
b6d6a51c | 1557 | } |
facdc750 RS |
1558 | |
1559 | pos--; | |
b6d6a51c | 1560 | } |
facdc750 RS |
1561 | |
1562 | n++; | |
b6d6a51c | 1563 | } |
facdc750 RS |
1564 | |
1565 | stop: | |
1566 | if (n == 0) | |
aff2ce94 | 1567 | { |
ab228c24 | 1568 | if (forward) |
49e44e81 | 1569 | set_search_regs ((multibyte ? pos_byte : pos) - match_byte, match_byte); |
ab228c24 | 1570 | else |
49e44e81 | 1571 | set_search_regs (multibyte ? pos_byte : pos, match_byte); |
aff2ce94 RS |
1572 | |
1573 | return pos; | |
1574 | } | |
facdc750 RS |
1575 | else if (n > 0) |
1576 | return -n; | |
1577 | else | |
1578 | return n; | |
1579 | } | |
1580 | \f | |
0190922f | 1581 | /* Do Boyer-Moore search N times for the string BASE_PAT, |
facdc750 RS |
1582 | whose length is LEN/LEN_BYTE, |
1583 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1584 | DIRECTION says which direction we search in. | |
1585 | TRT and INVERSE_TRT are translation tables. | |
0190922f | 1586 | Characters in PAT are already translated by TRT. |
facdc750 | 1587 | |
0190922f KH |
1588 | This kind of search works if all the characters in BASE_PAT that |
1589 | have nontrivial translation are the same aside from the last byte. | |
1590 | This makes it possible to translate just the last byte of a | |
1591 | character, and do so after just a simple test of the context. | |
a17ae96a | 1592 | CHAR_BASE is nonzero iff there is such a non-ASCII character. |
facdc750 RS |
1593 | |
1594 | If that criterion is not satisfied, do not call this function. */ | |
1595 | ||
1596 | static int | |
1597 | boyer_moore (n, base_pat, len, len_byte, trt, inverse_trt, | |
a17ae96a | 1598 | pos, pos_byte, lim, lim_byte, char_base) |
facdc750 RS |
1599 | int n; |
1600 | unsigned char *base_pat; | |
1601 | int len, len_byte; | |
1602 | Lisp_Object trt; | |
1603 | Lisp_Object inverse_trt; | |
1604 | int pos, pos_byte; | |
1605 | int lim, lim_byte; | |
a17ae96a | 1606 | int char_base; |
facdc750 RS |
1607 | { |
1608 | int direction = ((n > 0) ? 1 : -1); | |
1609 | register int dirlen; | |
a968f437 | 1610 | int infinity, limit, stride_for_teases = 0; |
facdc750 RS |
1611 | register int *BM_tab; |
1612 | int *BM_tab_base; | |
177c0ea7 | 1613 | register unsigned char *cursor, *p_limit; |
facdc750 | 1614 | register int i, j; |
cb6792d2 | 1615 | unsigned char *pat, *pat_end; |
facdc750 RS |
1616 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); |
1617 | ||
1618 | unsigned char simple_translate[0400]; | |
0190922f | 1619 | /* These are set to the preceding bytes of a byte to be translated |
49e44e81 | 1620 | if char_base is nonzero. As the maximum byte length of a |
a17ae96a | 1621 | multibyte character is 5, we have to check at most four previous |
0190922f KH |
1622 | bytes. */ |
1623 | int translate_prev_byte1 = 0; | |
1624 | int translate_prev_byte2 = 0; | |
1625 | int translate_prev_byte3 = 0; | |
a17ae96a | 1626 | int translate_prev_byte4 = 0; |
facdc750 RS |
1627 | |
1628 | #ifdef C_ALLOCA | |
1629 | int BM_tab_space[0400]; | |
1630 | BM_tab = &BM_tab_space[0]; | |
1631 | #else | |
1632 | BM_tab = (int *) alloca (0400 * sizeof (int)); | |
1633 | #endif | |
1634 | /* The general approach is that we are going to maintain that we know */ | |
1635 | /* the first (closest to the present position, in whatever direction */ | |
1636 | /* we're searching) character that could possibly be the last */ | |
1637 | /* (furthest from present position) character of a valid match. We */ | |
1638 | /* advance the state of our knowledge by looking at that character */ | |
1639 | /* and seeing whether it indeed matches the last character of the */ | |
1640 | /* pattern. If it does, we take a closer look. If it does not, we */ | |
1641 | /* move our pointer (to putative last characters) as far as is */ | |
1642 | /* logically possible. This amount of movement, which I call a */ | |
1643 | /* stride, will be the length of the pattern if the actual character */ | |
1644 | /* appears nowhere in the pattern, otherwise it will be the distance */ | |
1645 | /* from the last occurrence of that character to the end of the */ | |
1646 | /* pattern. */ | |
1647 | /* As a coding trick, an enormous stride is coded into the table for */ | |
1648 | /* characters that match the last character. This allows use of only */ | |
1649 | /* a single test, a test for having gone past the end of the */ | |
1650 | /* permissible match region, to test for both possible matches (when */ | |
1651 | /* the stride goes past the end immediately) and failure to */ | |
177c0ea7 | 1652 | /* match (where you get nudged past the end one stride at a time). */ |
facdc750 RS |
1653 | |
1654 | /* Here we make a "mickey mouse" BM table. The stride of the search */ | |
1655 | /* is determined only by the last character of the putative match. */ | |
1656 | /* If that character does not match, we will stride the proper */ | |
1657 | /* distance to propose a match that superimposes it on the last */ | |
1658 | /* instance of a character that matches it (per trt), or misses */ | |
177c0ea7 | 1659 | /* it entirely if there is none. */ |
facdc750 RS |
1660 | |
1661 | dirlen = len_byte * direction; | |
1662 | infinity = dirlen - (lim_byte + pos_byte + len_byte + len_byte) * direction; | |
cb6792d2 RS |
1663 | |
1664 | /* Record position after the end of the pattern. */ | |
1665 | pat_end = base_pat + len_byte; | |
1666 | /* BASE_PAT points to a character that we start scanning from. | |
1667 | It is the first character in a forward search, | |
1668 | the last character in a backward search. */ | |
facdc750 | 1669 | if (direction < 0) |
cb6792d2 RS |
1670 | base_pat = pat_end - 1; |
1671 | ||
facdc750 RS |
1672 | BM_tab_base = BM_tab; |
1673 | BM_tab += 0400; | |
1674 | j = dirlen; /* to get it in a register */ | |
1675 | /* A character that does not appear in the pattern induces a */ | |
1676 | /* stride equal to the pattern length. */ | |
1677 | while (BM_tab_base != BM_tab) | |
1678 | { | |
1679 | *--BM_tab = j; | |
1680 | *--BM_tab = j; | |
1681 | *--BM_tab = j; | |
1682 | *--BM_tab = j; | |
1683 | } | |
1684 | ||
1685 | /* We use this for translation, instead of TRT itself. | |
1686 | We fill this in to handle the characters that actually | |
1687 | occur in the pattern. Others don't matter anyway! */ | |
1688 | bzero (simple_translate, sizeof simple_translate); | |
1689 | for (i = 0; i < 0400; i++) | |
1690 | simple_translate[i] = i; | |
1691 | ||
a17ae96a | 1692 | if (char_base) |
0190922f | 1693 | { |
a17ae96a | 1694 | /* Setup translate_prev_byte1/2/3/4 from CHAR_BASE. Only a |
0190922f | 1695 | byte following them are the target of translation. */ |
0190922f | 1696 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
a17ae96a | 1697 | int len = CHAR_STRING (char_base, str); |
0190922f KH |
1698 | |
1699 | translate_prev_byte1 = str[len - 2]; | |
1700 | if (len > 2) | |
1701 | { | |
1702 | translate_prev_byte2 = str[len - 3]; | |
1703 | if (len > 3) | |
a17ae96a KH |
1704 | { |
1705 | translate_prev_byte3 = str[len - 4]; | |
1706 | if (len > 4) | |
1707 | translate_prev_byte4 = str[len - 5]; | |
1708 | } | |
0190922f KH |
1709 | } |
1710 | } | |
1711 | ||
facdc750 RS |
1712 | i = 0; |
1713 | while (i != infinity) | |
1714 | { | |
cb6792d2 | 1715 | unsigned char *ptr = base_pat + i; |
facdc750 RS |
1716 | i += direction; |
1717 | if (i == dirlen) | |
1718 | i = infinity; | |
1719 | if (! NILP (trt)) | |
ca1d1d23 | 1720 | { |
49e44e81 KH |
1721 | /* If the byte currently looking at is the last of a |
1722 | character to check case-equivalents, set CH to that | |
1723 | character. An ASCII character and a non-ASCII character | |
1724 | matching with CHAR_BASE are to be checked. */ | |
a17ae96a KH |
1725 | int ch = -1; |
1726 | ||
1727 | if (ASCII_BYTE_P (*ptr) || ! multibyte) | |
1728 | ch = *ptr; | |
49e44e81 | 1729 | else if (char_base |
86dc6ccb | 1730 | && ((pat_end - ptr) == 1 || CHAR_HEAD_P (ptr[1]))) |
ca1d1d23 | 1731 | { |
49e44e81 KH |
1732 | unsigned char *charstart = ptr - 1; |
1733 | ||
1734 | while (! (CHAR_HEAD_P (*charstart))) | |
1735 | charstart--; | |
1736 | ch = STRING_CHAR (charstart, ptr - charstart + 1); | |
a17ae96a KH |
1737 | if (char_base != (ch & ~0x3F)) |
1738 | ch = -1; | |
ca1d1d23 | 1739 | } |
facdc750 | 1740 | |
6786a107 | 1741 | if (ch >= 0400) |
49e44e81 KH |
1742 | j = (ch & 0x3F) | 0200; |
1743 | else | |
1744 | j = *ptr; | |
1745 | ||
facdc750 RS |
1746 | if (i == infinity) |
1747 | stride_for_teases = BM_tab[j]; | |
ab228c24 | 1748 | |
facdc750 RS |
1749 | BM_tab[j] = dirlen - i; |
1750 | /* A translation table is accompanied by its inverse -- see */ | |
177c0ea7 | 1751 | /* comment following downcase_table for details */ |
a17ae96a | 1752 | if (ch >= 0) |
ab228c24 RS |
1753 | { |
1754 | int starting_ch = ch; | |
49e44e81 | 1755 | int starting_j = j; |
a17ae96a | 1756 | |
ab228c24 RS |
1757 | while (1) |
1758 | { | |
1759 | TRANSLATE (ch, inverse_trt, ch); | |
6786a107 | 1760 | if (ch >= 0400) |
49e44e81 | 1761 | j = (ch & 0x3F) | 0200; |
ab228c24 | 1762 | else |
a17ae96a | 1763 | j = ch; |
ab228c24 RS |
1764 | |
1765 | /* For all the characters that map into CH, | |
1766 | set up simple_translate to map the last byte | |
1767 | into STARTING_J. */ | |
1768 | simple_translate[j] = starting_j; | |
1769 | if (ch == starting_ch) | |
1770 | break; | |
1771 | BM_tab[j] = dirlen - i; | |
1772 | } | |
1773 | } | |
facdc750 RS |
1774 | } |
1775 | else | |
1776 | { | |
1777 | j = *ptr; | |
1778 | ||
1779 | if (i == infinity) | |
1780 | stride_for_teases = BM_tab[j]; | |
1781 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1782 | } |
facdc750 RS |
1783 | /* stride_for_teases tells how much to stride if we get a */ |
1784 | /* match on the far character but are subsequently */ | |
1785 | /* disappointed, by recording what the stride would have been */ | |
1786 | /* for that character if the last character had been */ | |
1787 | /* different. */ | |
1788 | } | |
1789 | infinity = dirlen - infinity; | |
1790 | pos_byte += dirlen - ((direction > 0) ? direction : 0); | |
1791 | /* loop invariant - POS_BYTE points at where last char (first | |
1792 | char if reverse) of pattern would align in a possible match. */ | |
1793 | while (n != 0) | |
1794 | { | |
1795 | int tail_end; | |
1796 | unsigned char *tail_end_ptr; | |
1797 | ||
1798 | /* It's been reported that some (broken) compiler thinks that | |
1799 | Boolean expressions in an arithmetic context are unsigned. | |
1800 | Using an explicit ?1:0 prevents this. */ | |
1801 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1802 | < 0) | |
1803 | return (n * (0 - direction)); | |
1804 | /* First we do the part we can by pointers (maybe nothing) */ | |
1805 | QUIT; | |
1806 | pat = base_pat; | |
1807 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1808 | if (direction > 0) |
1809 | { | |
1810 | limit = BUFFER_CEILING_OF (limit); | |
1811 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1812 | can take on without hitting edge of buffer or the gap. */ | |
1813 | limit = min (limit, pos_byte + 20000); | |
1814 | limit = min (limit, lim_byte - 1); | |
1815 | } | |
1816 | else | |
1817 | { | |
1818 | limit = BUFFER_FLOOR_OF (limit); | |
1819 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1820 | can take on without hitting edge of buffer or the gap. */ | |
1821 | limit = max (limit, pos_byte - 20000); | |
1822 | limit = max (limit, lim_byte); | |
1823 | } | |
facdc750 RS |
1824 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1825 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1826 | ||
1827 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1828 | { |
facdc750 RS |
1829 | unsigned char *p2; |
1830 | ||
1831 | p_limit = BYTE_POS_ADDR (limit); | |
1832 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
1833 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ | |
1834 | while (1) /* use one cursor setting as long as i can */ | |
ca1d1d23 | 1835 | { |
facdc750 | 1836 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1837 | { |
facdc750 RS |
1838 | /* Use signed comparison if appropriate |
1839 | to make cursor+infinity sure to be > p_limit. | |
1840 | Assuming that the buffer lies in a range of addresses | |
1841 | that are all "positive" (as ints) or all "negative", | |
1842 | either kind of comparison will work as long | |
1843 | as we don't step by infinity. So pick the kind | |
1844 | that works when we do step by infinity. */ | |
1845 | if ((EMACS_INT) (p_limit + infinity) > (EMACS_INT) p_limit) | |
1846 | while ((EMACS_INT) cursor <= (EMACS_INT) p_limit) | |
1847 | cursor += BM_tab[*cursor]; | |
ca1d1d23 | 1848 | else |
facdc750 RS |
1849 | while ((EMACS_UINT) cursor <= (EMACS_UINT) p_limit) |
1850 | cursor += BM_tab[*cursor]; | |
1851 | } | |
1852 | else | |
1853 | { | |
1854 | if ((EMACS_INT) (p_limit + infinity) < (EMACS_INT) p_limit) | |
1855 | while ((EMACS_INT) cursor >= (EMACS_INT) p_limit) | |
1856 | cursor += BM_tab[*cursor]; | |
1857 | else | |
1858 | while ((EMACS_UINT) cursor >= (EMACS_UINT) p_limit) | |
1859 | cursor += BM_tab[*cursor]; | |
1860 | } | |
ca1d1d23 | 1861 | /* If you are here, cursor is beyond the end of the searched region. */ |
facdc750 RS |
1862 | /* This can happen if you match on the far character of the pattern, */ |
1863 | /* because the "stride" of that character is infinity, a number able */ | |
1864 | /* to throw you well beyond the end of the search. It can also */ | |
1865 | /* happen if you fail to match within the permitted region and would */ | |
1866 | /* otherwise try a character beyond that region */ | |
1867 | if ((cursor - p_limit) * direction <= len_byte) | |
1868 | break; /* a small overrun is genuine */ | |
1869 | cursor -= infinity; /* large overrun = hit */ | |
1870 | i = dirlen - direction; | |
1871 | if (! NILP (trt)) | |
1872 | { | |
1873 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1874 | { |
facdc750 RS |
1875 | int ch; |
1876 | cursor -= direction; | |
1877 | /* Translate only the last byte of a character. */ | |
1878 | if (! multibyte | |
1879 | || ((cursor == tail_end_ptr | |
1880 | || CHAR_HEAD_P (cursor[1])) | |
1881 | && (CHAR_HEAD_P (cursor[0]) | |
0190922f KH |
1882 | /* Check if this is the last byte of |
1883 | a translable character. */ | |
1884 | || (translate_prev_byte1 == cursor[-1] | |
1885 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1886 | || (translate_prev_byte2 == cursor[-2] | |
1887 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1888 | || (translate_prev_byte3 == cursor[-3])))))))) | |
facdc750 RS |
1889 | ch = simple_translate[*cursor]; |
1890 | else | |
1891 | ch = *cursor; | |
1892 | if (pat[i] != ch) | |
1893 | break; | |
ca1d1d23 | 1894 | } |
facdc750 RS |
1895 | } |
1896 | else | |
1897 | { | |
1898 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1899 | { |
facdc750 RS |
1900 | cursor -= direction; |
1901 | if (pat[i] != *cursor) | |
1902 | break; | |
ca1d1d23 | 1903 | } |
facdc750 RS |
1904 | } |
1905 | cursor += dirlen - i - direction; /* fix cursor */ | |
1906 | if (i + direction == 0) | |
1907 | { | |
1908 | int position; | |
0c8533c6 | 1909 | |
facdc750 | 1910 | cursor -= direction; |
1113d9db | 1911 | |
facdc750 RS |
1912 | position = pos_byte + cursor - p2 + ((direction > 0) |
1913 | ? 1 - len_byte : 0); | |
1914 | set_search_regs (position, len_byte); | |
ca325161 | 1915 | |
facdc750 RS |
1916 | if ((n -= direction) != 0) |
1917 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1918 | else |
facdc750 RS |
1919 | return ((direction > 0) |
1920 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1921 | } |
facdc750 RS |
1922 | else |
1923 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1924 | } |
facdc750 RS |
1925 | pos_byte += cursor - p2; |
1926 | } | |
1927 | else | |
1928 | /* Now we'll pick up a clump that has to be done the hard */ | |
1929 | /* way because it covers a discontinuity */ | |
1930 | { | |
1931 | limit = ((direction > 0) | |
1932 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1933 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1934 | limit = ((direction > 0) | |
1935 | ? min (limit + len_byte, lim_byte - 1) | |
1936 | : max (limit - len_byte, lim_byte)); | |
1937 | /* LIMIT is now the last value POS_BYTE can have | |
1938 | and still be valid for a possible match. */ | |
1939 | while (1) | |
ca1d1d23 | 1940 | { |
facdc750 RS |
1941 | /* This loop can be coded for space rather than */ |
1942 | /* speed because it will usually run only once. */ | |
1943 | /* (the reach is at most len + 21, and typically */ | |
177c0ea7 | 1944 | /* does not exceed len) */ |
facdc750 RS |
1945 | while ((limit - pos_byte) * direction >= 0) |
1946 | pos_byte += BM_tab[FETCH_BYTE (pos_byte)]; | |
1947 | /* now run the same tests to distinguish going off the */ | |
1948 | /* end, a match or a phony match. */ | |
1949 | if ((pos_byte - limit) * direction <= len_byte) | |
1950 | break; /* ran off the end */ | |
1951 | /* Found what might be a match. | |
1952 | Set POS_BYTE back to last (first if reverse) pos. */ | |
1953 | pos_byte -= infinity; | |
1954 | i = dirlen - direction; | |
1955 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1956 | { |
facdc750 RS |
1957 | int ch; |
1958 | unsigned char *ptr; | |
1959 | pos_byte -= direction; | |
1960 | ptr = BYTE_POS_ADDR (pos_byte); | |
1961 | /* Translate only the last byte of a character. */ | |
1962 | if (! multibyte | |
1963 | || ((ptr == tail_end_ptr | |
1964 | || CHAR_HEAD_P (ptr[1])) | |
1965 | && (CHAR_HEAD_P (ptr[0]) | |
0190922f KH |
1966 | /* Check if this is the last byte of a |
1967 | translable character. */ | |
1968 | || (translate_prev_byte1 == ptr[-1] | |
1969 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1970 | || (translate_prev_byte2 == ptr[-2] | |
1971 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1972 | || translate_prev_byte3 == ptr[-3]))))))) | |
facdc750 RS |
1973 | ch = simple_translate[*ptr]; |
1974 | else | |
1975 | ch = *ptr; | |
1976 | if (pat[i] != ch) | |
1977 | break; | |
1978 | } | |
1979 | /* Above loop has moved POS_BYTE part or all the way | |
1980 | back to the first pos (last pos if reverse). | |
1981 | Set it once again at the last (first if reverse) char. */ | |
1982 | pos_byte += dirlen - i- direction; | |
1983 | if (i + direction == 0) | |
1984 | { | |
1985 | int position; | |
1986 | pos_byte -= direction; | |
1113d9db | 1987 | |
facdc750 | 1988 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
0c8533c6 | 1989 | |
facdc750 | 1990 | set_search_regs (position, len_byte); |
ca325161 | 1991 | |
facdc750 RS |
1992 | if ((n -= direction) != 0) |
1993 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 1994 | else |
facdc750 RS |
1995 | return ((direction > 0) |
1996 | ? search_regs.end[0] : search_regs.start[0]); | |
ca1d1d23 | 1997 | } |
facdc750 RS |
1998 | else |
1999 | pos_byte += stride_for_teases; | |
2000 | } | |
2001 | } | |
2002 | /* We have done one clump. Can we continue? */ | |
2003 | if ((lim_byte - pos_byte) * direction < 0) | |
2004 | return ((0 - n) * direction); | |
ca1d1d23 | 2005 | } |
facdc750 | 2006 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 2007 | } |
ca325161 | 2008 | |
fa8ed3e0 | 2009 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
2010 | for the overall match just found in the current buffer. |
2011 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
2012 | |
2013 | static void | |
fa8ed3e0 RS |
2014 | set_search_regs (beg_byte, nbytes) |
2015 | int beg_byte, nbytes; | |
ca325161 | 2016 | { |
a7e4cdde RS |
2017 | int i; |
2018 | ||
ca325161 RS |
2019 | /* Make sure we have registers in which to store |
2020 | the match position. */ | |
2021 | if (search_regs.num_regs == 0) | |
2022 | { | |
2d4a771a RS |
2023 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
2024 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 2025 | search_regs.num_regs = 2; |
ca325161 RS |
2026 | } |
2027 | ||
a7e4cdde RS |
2028 | /* Clear out the other registers. */ |
2029 | for (i = 1; i < search_regs.num_regs; i++) | |
2030 | { | |
2031 | search_regs.start[i] = -1; | |
2032 | search_regs.end[i] = -1; | |
2033 | } | |
2034 | ||
fa8ed3e0 RS |
2035 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
2036 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 2037 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 2038 | } |
ca1d1d23 JB |
2039 | \f |
2040 | /* Given a string of words separated by word delimiters, | |
2041 | compute a regexp that matches those exact words | |
2042 | separated by arbitrary punctuation. */ | |
2043 | ||
2044 | static Lisp_Object | |
2045 | wordify (string) | |
2046 | Lisp_Object string; | |
2047 | { | |
2048 | register unsigned char *p, *o; | |
0c8533c6 | 2049 | register int i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 2050 | Lisp_Object val; |
0c8533c6 RS |
2051 | int prev_c = 0; |
2052 | int adjust; | |
ca1d1d23 | 2053 | |
b7826503 | 2054 | CHECK_STRING (string); |
d5db4077 KR |
2055 | p = SDATA (string); |
2056 | len = SCHARS (string); | |
ca1d1d23 | 2057 | |
0c8533c6 RS |
2058 | for (i = 0, i_byte = 0; i < len; ) |
2059 | { | |
2060 | int c; | |
177c0ea7 | 2061 | |
93daa011 | 2062 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
0c8533c6 RS |
2063 | |
2064 | if (SYNTAX (c) != Sword) | |
2065 | { | |
2066 | punct_count++; | |
2067 | if (i > 0 && SYNTAX (prev_c) == Sword) | |
2068 | word_count++; | |
2069 | } | |
ca1d1d23 | 2070 | |
0c8533c6 RS |
2071 | prev_c = c; |
2072 | } | |
2073 | ||
2074 | if (SYNTAX (prev_c) == Sword) | |
2075 | word_count++; | |
2076 | if (!word_count) | |
b07b65aa | 2077 | return empty_string; |
0c8533c6 RS |
2078 | |
2079 | adjust = - punct_count + 5 * (word_count - 1) + 4; | |
8a2df937 RS |
2080 | if (STRING_MULTIBYTE (string)) |
2081 | val = make_uninit_multibyte_string (len + adjust, | |
d5db4077 | 2082 | SBYTES (string) |
8a2df937 RS |
2083 | + adjust); |
2084 | else | |
2085 | val = make_uninit_string (len + adjust); | |
ca1d1d23 | 2086 | |
d5db4077 | 2087 | o = SDATA (val); |
ca1d1d23 JB |
2088 | *o++ = '\\'; |
2089 | *o++ = 'b'; | |
1e9582d4 | 2090 | prev_c = 0; |
ca1d1d23 | 2091 | |
1e9582d4 RS |
2092 | for (i = 0, i_byte = 0; i < len; ) |
2093 | { | |
2094 | int c; | |
2095 | int i_byte_orig = i_byte; | |
177c0ea7 | 2096 | |
93daa011 | 2097 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
1e9582d4 RS |
2098 | |
2099 | if (SYNTAX (c) == Sword) | |
2100 | { | |
5d69fe10 | 2101 | bcopy (SDATA (string) + i_byte_orig, o, |
1e9582d4 RS |
2102 | i_byte - i_byte_orig); |
2103 | o += i_byte - i_byte_orig; | |
2104 | } | |
2105 | else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count) | |
2106 | { | |
2107 | *o++ = '\\'; | |
2108 | *o++ = 'W'; | |
2109 | *o++ = '\\'; | |
2110 | *o++ = 'W'; | |
2111 | *o++ = '*'; | |
2112 | } | |
2113 | ||
2114 | prev_c = c; | |
2115 | } | |
ca1d1d23 JB |
2116 | |
2117 | *o++ = '\\'; | |
2118 | *o++ = 'b'; | |
2119 | ||
2120 | return val; | |
2121 | } | |
2122 | \f | |
2123 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
8c1a1077 PJ |
2124 | "MSearch backward: ", |
2125 | doc: /* Search backward from point for STRING. | |
2126 | Set point to the beginning of the occurrence found, and return point. | |
2127 | An optional second argument bounds the search; it is a buffer position. | |
2128 | The match found must not extend before that position. | |
2129 | Optional third argument, if t, means if fail just return nil (no error). | |
2130 | If not nil and not t, position at limit of search and return nil. | |
2131 | Optional fourth argument is repeat count--search for successive occurrences. | |
2132 | ||
2133 | Search case-sensitivity is determined by the value of the variable | |
2134 | `case-fold-search', which see. | |
2135 | ||
2136 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2137 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2138 | Lisp_Object string, bound, noerror, count; |
2139 | { | |
b819a390 | 2140 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2141 | } |
2142 | ||
6af43974 | 2143 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2144 | doc: /* Search forward from point for STRING. |
2145 | Set point to the end of the occurrence found, and return point. | |
2146 | An optional second argument bounds the search; it is a buffer position. | |
d6aacdcd | 2147 | The match found must not extend after that position. A value of nil is |
48740f01 | 2148 | equivalent to (point-max). |
8c1a1077 PJ |
2149 | Optional third argument, if t, means if fail just return nil (no error). |
2150 | If not nil and not t, move to limit of search and return nil. | |
2151 | Optional fourth argument is repeat count--search for successive occurrences. | |
2152 | ||
2153 | Search case-sensitivity is determined by the value of the variable | |
2154 | `case-fold-search', which see. | |
2155 | ||
2156 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
2157 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2158 | Lisp_Object string, bound, noerror, count; |
2159 | { | |
b819a390 | 2160 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2161 | } |
2162 | ||
2163 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
8c1a1077 PJ |
2164 | "sWord search backward: ", |
2165 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2166 | Set point to the beginning of the occurrence found, and return point. | |
2167 | An optional second argument bounds the search; it is a buffer position. | |
2168 | The match found must not extend before that position. | |
2169 | Optional third argument, if t, means if fail just return nil (no error). | |
2170 | If not nil and not t, move to limit of search and return nil. | |
2171 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2172 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2173 | Lisp_Object string, bound, noerror, count; |
2174 | { | |
b819a390 | 2175 | return search_command (wordify (string), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2176 | } |
2177 | ||
2178 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
8c1a1077 PJ |
2179 | "sWord search: ", |
2180 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2181 | Set point to the end of the occurrence found, and return point. | |
2182 | An optional second argument bounds the search; it is a buffer position. | |
2183 | The match found must not extend after that position. | |
2184 | Optional third argument, if t, means if fail just return nil (no error). | |
2185 | If not nil and not t, move to limit of search and return nil. | |
2186 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
2187 | (string, bound, noerror, count) | |
ca1d1d23 JB |
2188 | Lisp_Object string, bound, noerror, count; |
2189 | { | |
b819a390 | 2190 | return search_command (wordify (string), bound, noerror, count, 1, 1, 0); |
ca1d1d23 JB |
2191 | } |
2192 | ||
2193 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
8c1a1077 PJ |
2194 | "sRE search backward: ", |
2195 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2196 | Set point to the beginning of the match, and return point. | |
2197 | The match found is the one starting last in the buffer | |
2198 | and yet ending before the origin of the search. | |
2199 | An optional second argument bounds the search; it is a buffer position. | |
2200 | The match found must start at or after that position. | |
2201 | Optional third argument, if t, means if fail just return nil (no error). | |
2202 | If not nil and not t, move to limit of search and return nil. | |
2203 | Optional fourth argument is repeat count--search for successive occurrences. | |
2204 | See also the functions `match-beginning', `match-end', `match-string', | |
2205 | and `replace-match'. */) | |
2206 | (regexp, bound, noerror, count) | |
19c0a730 | 2207 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2208 | { |
b819a390 | 2209 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2210 | } |
2211 | ||
2212 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2213 | "sRE search: ", |
2214 | doc: /* Search forward from point for regular expression REGEXP. | |
2215 | Set point to the end of the occurrence found, and return point. | |
2216 | An optional second argument bounds the search; it is a buffer position. | |
2217 | The match found must not extend after that position. | |
2218 | Optional third argument, if t, means if fail just return nil (no error). | |
2219 | If not nil and not t, move to limit of search and return nil. | |
2220 | Optional fourth argument is repeat count--search for successive occurrences. | |
2221 | See also the functions `match-beginning', `match-end', `match-string', | |
2222 | and `replace-match'. */) | |
2223 | (regexp, bound, noerror, count) | |
19c0a730 | 2224 | Lisp_Object regexp, bound, noerror, count; |
ca1d1d23 | 2225 | { |
b819a390 RS |
2226 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2227 | } | |
2228 | ||
2229 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2230 | "sPosix search backward: ", |
2231 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2232 | Find the longest match in accord with Posix regular expression rules. | |
2233 | Set point to the beginning of the match, and return point. | |
2234 | The match found is the one starting last in the buffer | |
2235 | and yet ending before the origin of the search. | |
2236 | An optional second argument bounds the search; it is a buffer position. | |
2237 | The match found must start at or after that position. | |
2238 | Optional third argument, if t, means if fail just return nil (no error). | |
2239 | If not nil and not t, move to limit of search and return nil. | |
2240 | Optional fourth argument is repeat count--search for successive occurrences. | |
2241 | See also the functions `match-beginning', `match-end', `match-string', | |
2242 | and `replace-match'. */) | |
2243 | (regexp, bound, noerror, count) | |
b819a390 RS |
2244 | Lisp_Object regexp, bound, noerror, count; |
2245 | { | |
2246 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2247 | } | |
2248 | ||
2249 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2250 | "sPosix search: ", |
2251 | doc: /* Search forward from point for regular expression REGEXP. | |
2252 | Find the longest match in accord with Posix regular expression rules. | |
2253 | Set point to the end of the occurrence found, and return point. | |
2254 | An optional second argument bounds the search; it is a buffer position. | |
2255 | The match found must not extend after that position. | |
2256 | Optional third argument, if t, means if fail just return nil (no error). | |
2257 | If not nil and not t, move to limit of search and return nil. | |
2258 | Optional fourth argument is repeat count--search for successive occurrences. | |
2259 | See also the functions `match-beginning', `match-end', `match-string', | |
2260 | and `replace-match'. */) | |
2261 | (regexp, bound, noerror, count) | |
b819a390 RS |
2262 | Lisp_Object regexp, bound, noerror, count; |
2263 | { | |
2264 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2265 | } |
2266 | \f | |
d7a5ad5f | 2267 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2268 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2269 | Leave point at the end of the replacement text. |
2270 | ||
8c1a1077 PJ |
2271 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. |
2272 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2273 | based on the replaced text. | |
2274 | If the replaced text has only capital letters | |
2275 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
4dd0c271 RS |
2276 | Otherwise if all words are capitalized in the replaced text, |
2277 | capitalize each word in NEWTEXT. | |
8c1a1077 PJ |
2278 | |
2279 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2280 | Otherwise treat `\\' as special: | |
2281 | `\\&' in NEWTEXT means substitute original matched text. | |
2282 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2283 | If Nth parens didn't match, substitute nothing. | |
2284 | `\\\\' means insert one `\\'. | |
4dd0c271 RS |
2285 | Case conversion does not apply to these substitutions. |
2286 | ||
8c1a1077 | 2287 | FIXEDCASE and LITERAL are optional arguments. |
8c1a1077 PJ |
2288 | |
2289 | The optional fourth argument STRING can be a string to modify. | |
2290 | This is meaningful when the previous match was done against STRING, | |
2291 | using `string-match'. When used this way, `replace-match' | |
2292 | creates and returns a new string made by copying STRING and replacing | |
2293 | the part of STRING that was matched. | |
2294 | ||
2295 | The optional fifth argument SUBEXP specifies a subexpression; | |
2296 | it says to replace just that subexpression with NEWTEXT, | |
2297 | rather than replacing the entire matched text. | |
2298 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2299 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2300 | NEWTEXT in place of subexp N. | |
2301 | This is useful only after a regular expression search or match, | |
2302 | since only regular expressions have distinguished subexpressions. */) | |
2303 | (newtext, fixedcase, literal, string, subexp) | |
d7a5ad5f | 2304 | Lisp_Object newtext, fixedcase, literal, string, subexp; |
ca1d1d23 JB |
2305 | { |
2306 | enum { nochange, all_caps, cap_initial } case_action; | |
ac3b28b1 | 2307 | register int pos, pos_byte; |
ca1d1d23 | 2308 | int some_multiletter_word; |
97832bd0 | 2309 | int some_lowercase; |
73dc8771 | 2310 | int some_uppercase; |
208767c3 | 2311 | int some_nonuppercase_initial; |
ca1d1d23 | 2312 | register int c, prevc; |
d7a5ad5f | 2313 | int sub; |
3e18eecf | 2314 | int opoint, newpoint; |
ca1d1d23 | 2315 | |
b7826503 | 2316 | CHECK_STRING (newtext); |
ca1d1d23 | 2317 | |
080c45fd | 2318 | if (! NILP (string)) |
b7826503 | 2319 | CHECK_STRING (string); |
080c45fd | 2320 | |
ca1d1d23 JB |
2321 | case_action = nochange; /* We tried an initialization */ |
2322 | /* but some C compilers blew it */ | |
4746118a JB |
2323 | |
2324 | if (search_regs.num_regs <= 0) | |
d72cdbfc | 2325 | error ("`replace-match' called before any match found"); |
4746118a | 2326 | |
d7a5ad5f RS |
2327 | if (NILP (subexp)) |
2328 | sub = 0; | |
2329 | else | |
2330 | { | |
b7826503 | 2331 | CHECK_NUMBER (subexp); |
d7a5ad5f RS |
2332 | sub = XINT (subexp); |
2333 | if (sub < 0 || sub >= search_regs.num_regs) | |
2334 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2335 | } | |
2336 | ||
080c45fd RS |
2337 | if (NILP (string)) |
2338 | { | |
d7a5ad5f RS |
2339 | if (search_regs.start[sub] < BEGV |
2340 | || search_regs.start[sub] > search_regs.end[sub] | |
2341 | || search_regs.end[sub] > ZV) | |
2342 | args_out_of_range (make_number (search_regs.start[sub]), | |
2343 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2344 | } |
2345 | else | |
2346 | { | |
d7a5ad5f RS |
2347 | if (search_regs.start[sub] < 0 |
2348 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2349 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2350 | args_out_of_range (make_number (search_regs.start[sub]), |
2351 | make_number (search_regs.end[sub])); | |
080c45fd | 2352 | } |
ca1d1d23 JB |
2353 | |
2354 | if (NILP (fixedcase)) | |
2355 | { | |
2356 | /* Decide how to casify by examining the matched text. */ | |
ac3b28b1 | 2357 | int last; |
ca1d1d23 | 2358 | |
ac3b28b1 KH |
2359 | pos = search_regs.start[sub]; |
2360 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2361 | |
2362 | if (NILP (string)) | |
ac3b28b1 | 2363 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2364 | else |
ac3b28b1 | 2365 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2366 | |
ca1d1d23 JB |
2367 | prevc = '\n'; |
2368 | case_action = all_caps; | |
2369 | ||
2370 | /* some_multiletter_word is set nonzero if any original word | |
2371 | is more than one letter long. */ | |
2372 | some_multiletter_word = 0; | |
97832bd0 | 2373 | some_lowercase = 0; |
208767c3 | 2374 | some_nonuppercase_initial = 0; |
73dc8771 | 2375 | some_uppercase = 0; |
ca1d1d23 | 2376 | |
ac3b28b1 | 2377 | while (pos < last) |
ca1d1d23 | 2378 | { |
080c45fd | 2379 | if (NILP (string)) |
ac3b28b1 | 2380 | { |
93daa011 | 2381 | c = FETCH_CHAR_AS_MULTIBYTE (pos_byte); |
ac3b28b1 KH |
2382 | INC_BOTH (pos, pos_byte); |
2383 | } | |
080c45fd | 2384 | else |
93daa011 | 2385 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2386 | |
ca1d1d23 JB |
2387 | if (LOWERCASEP (c)) |
2388 | { | |
2389 | /* Cannot be all caps if any original char is lower case */ | |
2390 | ||
97832bd0 | 2391 | some_lowercase = 1; |
ca1d1d23 | 2392 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2393 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2394 | else |
2395 | some_multiletter_word = 1; | |
2396 | } | |
d16c2b66 | 2397 | else if (UPPERCASEP (c)) |
ca1d1d23 | 2398 | { |
73dc8771 | 2399 | some_uppercase = 1; |
97832bd0 | 2400 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2401 | ; |
97832bd0 | 2402 | else |
ca1d1d23 JB |
2403 | some_multiletter_word = 1; |
2404 | } | |
208767c3 RS |
2405 | else |
2406 | { | |
2407 | /* If the initial is a caseless word constituent, | |
2408 | treat that like a lowercase initial. */ | |
2409 | if (SYNTAX (prevc) != Sword) | |
2410 | some_nonuppercase_initial = 1; | |
2411 | } | |
ca1d1d23 JB |
2412 | |
2413 | prevc = c; | |
2414 | } | |
2415 | ||
97832bd0 RS |
2416 | /* Convert to all caps if the old text is all caps |
2417 | and has at least one multiletter word. */ | |
2418 | if (! some_lowercase && some_multiletter_word) | |
2419 | case_action = all_caps; | |
c4d460ce | 2420 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2421 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2422 | case_action = cap_initial; |
208767c3 | 2423 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2424 | /* Should x -> yz, operating on X, give Yz or YZ? |
2425 | We'll assume the latter. */ | |
2426 | case_action = all_caps; | |
97832bd0 RS |
2427 | else |
2428 | case_action = nochange; | |
ca1d1d23 JB |
2429 | } |
2430 | ||
080c45fd RS |
2431 | /* Do replacement in a string. */ |
2432 | if (!NILP (string)) | |
2433 | { | |
2434 | Lisp_Object before, after; | |
2435 | ||
2436 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2437 | make_number (search_regs.start[sub])); |
2438 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2439 | |
636a5e28 RS |
2440 | /* Substitute parts of the match into NEWTEXT |
2441 | if desired. */ | |
080c45fd RS |
2442 | if (NILP (literal)) |
2443 | { | |
d131e79c RS |
2444 | int lastpos = 0; |
2445 | int lastpos_byte = 0; | |
080c45fd RS |
2446 | /* We build up the substituted string in ACCUM. */ |
2447 | Lisp_Object accum; | |
2448 | Lisp_Object middle; | |
d5db4077 | 2449 | int length = SBYTES (newtext); |
080c45fd RS |
2450 | |
2451 | accum = Qnil; | |
2452 | ||
ac3b28b1 | 2453 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd RS |
2454 | { |
2455 | int substart = -1; | |
6bbd7a29 | 2456 | int subend = 0; |
1e79ec24 | 2457 | int delbackslash = 0; |
080c45fd | 2458 | |
0c8533c6 RS |
2459 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2460 | ||
080c45fd RS |
2461 | if (c == '\\') |
2462 | { | |
0c8533c6 | 2463 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2464 | |
080c45fd RS |
2465 | if (c == '&') |
2466 | { | |
d7a5ad5f RS |
2467 | substart = search_regs.start[sub]; |
2468 | subend = search_regs.end[sub]; | |
080c45fd | 2469 | } |
76fc1ea2 | 2470 | else if (c >= '1' && c <= '9') |
080c45fd | 2471 | { |
76fc1ea2 KH |
2472 | if (search_regs.start[c - '0'] >= 0 |
2473 | && c <= search_regs.num_regs + '0') | |
080c45fd RS |
2474 | { |
2475 | substart = search_regs.start[c - '0']; | |
2476 | subend = search_regs.end[c - '0']; | |
2477 | } | |
76fc1ea2 KH |
2478 | else |
2479 | { | |
2480 | /* If that subexp did not match, | |
2481 | replace \\N with nothing. */ | |
2482 | substart = 0; | |
2483 | subend = 0; | |
2484 | } | |
080c45fd | 2485 | } |
1e79ec24 KH |
2486 | else if (c == '\\') |
2487 | delbackslash = 1; | |
636a5e28 RS |
2488 | else |
2489 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2490 | } |
2491 | if (substart >= 0) | |
2492 | { | |
d131e79c RS |
2493 | if (pos - 2 != lastpos) |
2494 | middle = substring_both (newtext, lastpos, | |
2495 | lastpos_byte, | |
2496 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2497 | else |
2498 | middle = Qnil; | |
2499 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2500 | Fsubstring (string, |
2501 | make_number (substart), | |
080c45fd RS |
2502 | make_number (subend))); |
2503 | lastpos = pos; | |
0c8533c6 | 2504 | lastpos_byte = pos_byte; |
080c45fd | 2505 | } |
1e79ec24 KH |
2506 | else if (delbackslash) |
2507 | { | |
d131e79c RS |
2508 | middle = substring_both (newtext, lastpos, |
2509 | lastpos_byte, | |
2510 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2511 | |
1e79ec24 KH |
2512 | accum = concat2 (accum, middle); |
2513 | lastpos = pos; | |
0c8533c6 | 2514 | lastpos_byte = pos_byte; |
1e79ec24 | 2515 | } |
080c45fd RS |
2516 | } |
2517 | ||
d131e79c RS |
2518 | if (pos != lastpos) |
2519 | middle = substring_both (newtext, lastpos, | |
2520 | lastpos_byte, | |
0c8533c6 | 2521 | pos, pos_byte); |
080c45fd RS |
2522 | else |
2523 | middle = Qnil; | |
2524 | ||
2525 | newtext = concat2 (accum, middle); | |
2526 | } | |
2527 | ||
636a5e28 | 2528 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2529 | if (case_action == all_caps) |
2530 | newtext = Fupcase (newtext); | |
2531 | else if (case_action == cap_initial) | |
2b2eead9 | 2532 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2533 | |
2534 | return concat3 (before, newtext, after); | |
2535 | } | |
2536 | ||
09c4719e | 2537 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2538 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2539 | opoint = PT - ZV; |
9160906f RS |
2540 | else if (PT > search_regs.start[sub]) |
2541 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2542 | else |
2543 | opoint = PT; | |
2544 | ||
886ed6ec RS |
2545 | /* If we want non-literal replacement, |
2546 | perform substitution on the replacement string. */ | |
2547 | if (NILP (literal)) | |
ca1d1d23 | 2548 | { |
d5db4077 | 2549 | int length = SBYTES (newtext); |
68e69fbd RS |
2550 | unsigned char *substed; |
2551 | int substed_alloc_size, substed_len; | |
3bc25e52 KH |
2552 | int buf_multibyte = !NILP (current_buffer->enable_multibyte_characters); |
2553 | int str_multibyte = STRING_MULTIBYTE (newtext); | |
2554 | Lisp_Object rev_tbl; | |
886ed6ec | 2555 | int really_changed = 0; |
3bc25e52 | 2556 | |
8f924df7 | 2557 | rev_tbl = Qnil; |
ac3b28b1 | 2558 | |
68e69fbd RS |
2559 | substed_alloc_size = length * 2 + 100; |
2560 | substed = (unsigned char *) xmalloc (substed_alloc_size + 1); | |
2561 | substed_len = 0; | |
2562 | ||
3bc25e52 KH |
2563 | /* Go thru NEWTEXT, producing the actual text to insert in |
2564 | SUBSTED while adjusting multibyteness to that of the current | |
2565 | buffer. */ | |
ca1d1d23 | 2566 | |
ac3b28b1 | 2567 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2568 | { |
68e69fbd | 2569 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
f8ce8a0d GM |
2570 | unsigned char *add_stuff = NULL; |
2571 | int add_len = 0; | |
68e69fbd | 2572 | int idx = -1; |
9a76659d | 2573 | |
3bc25e52 KH |
2574 | if (str_multibyte) |
2575 | { | |
eb99a8dd | 2576 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 KH |
2577 | if (!buf_multibyte) |
2578 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2579 | } | |
2580 | else | |
2581 | { | |
2582 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2583 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2584 | if (buf_multibyte) |
2585 | c = unibyte_char_to_multibyte (c); | |
2586 | } | |
ac3b28b1 | 2587 | |
68e69fbd RS |
2588 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2589 | or set IDX to a match index, which means put that part | |
2590 | of the buffer text into SUBSTED. */ | |
2591 | ||
ca1d1d23 JB |
2592 | if (c == '\\') |
2593 | { | |
886ed6ec RS |
2594 | really_changed = 1; |
2595 | ||
3bc25e52 KH |
2596 | if (str_multibyte) |
2597 | { | |
eb99a8dd KH |
2598 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2599 | pos, pos_byte); | |
071ce769 | 2600 | if (!buf_multibyte && !ASCII_CHAR_P (c)) |
3bc25e52 KH |
2601 | c = multibyte_char_to_unibyte (c, rev_tbl); |
2602 | } | |
2603 | else | |
2604 | { | |
d5db4077 | 2605 | c = SREF (newtext, pos_byte++); |
3bc25e52 KH |
2606 | if (buf_multibyte) |
2607 | c = unibyte_char_to_multibyte (c); | |
2608 | } | |
2609 | ||
ca1d1d23 | 2610 | if (c == '&') |
68e69fbd | 2611 | idx = sub; |
78445046 | 2612 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2613 | { |
2614 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2615 | idx = c - '0'; |
ca1d1d23 | 2616 | } |
636a5e28 | 2617 | else if (c == '\\') |
68e69fbd | 2618 | add_len = 1, add_stuff = "\\"; |
636a5e28 | 2619 | else |
3bc25e52 KH |
2620 | { |
2621 | xfree (substed); | |
2622 | error ("Invalid use of `\\' in replacement text"); | |
2623 | } | |
ca1d1d23 JB |
2624 | } |
2625 | else | |
68e69fbd RS |
2626 | { |
2627 | add_len = CHAR_STRING (c, str); | |
2628 | add_stuff = str; | |
2629 | } | |
2630 | ||
2631 | /* If we want to copy part of a previous match, | |
2632 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2633 | if (idx >= 0) | |
2634 | { | |
2635 | int begbyte = CHAR_TO_BYTE (search_regs.start[idx]); | |
2636 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; | |
2637 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2638 | move_gap (search_regs.start[idx]); | |
2639 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2640 | } | |
2641 | ||
2642 | /* Now the stuff we want to add to SUBSTED | |
2643 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2644 | ||
2645 | /* Make sure SUBSTED is big enough. */ | |
2646 | if (substed_len + add_len >= substed_alloc_size) | |
2647 | { | |
2648 | substed_alloc_size = substed_len + add_len + 500; | |
2649 | substed = (unsigned char *) xrealloc (substed, | |
2650 | substed_alloc_size + 1); | |
2651 | } | |
2652 | ||
2653 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2654 | if (add_stuff) |
2655 | { | |
2656 | bcopy (add_stuff, substed + substed_len, add_len); | |
2657 | substed_len += add_len; | |
2658 | } | |
ca1d1d23 | 2659 | } |
68e69fbd | 2660 | |
886ed6ec | 2661 | if (really_changed) |
76fc1ea2 KH |
2662 | { |
2663 | if (buf_multibyte) | |
2664 | { | |
2665 | int nchars = multibyte_chars_in_text (substed, substed_len); | |
68e69fbd | 2666 | |
76fc1ea2 KH |
2667 | newtext = make_multibyte_string (substed, nchars, substed_len); |
2668 | } | |
2669 | else | |
2670 | newtext = make_unibyte_string (substed, substed_len); | |
2671 | } | |
68e69fbd | 2672 | xfree (substed); |
ca1d1d23 JB |
2673 | } |
2674 | ||
886ed6ec RS |
2675 | /* Replace the old text with the new in the cleanest possible way. */ |
2676 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2677 | newtext, 1, 0, 1); | |
d5db4077 | 2678 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2679 | |
2680 | if (case_action == all_caps) | |
886ed6ec RS |
2681 | Fupcase_region (make_number (search_regs.start[sub]), |
2682 | make_number (newpoint)); | |
ca1d1d23 | 2683 | else if (case_action == cap_initial) |
886ed6ec RS |
2684 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2685 | make_number (newpoint)); | |
3e18eecf | 2686 | |
98e942e0 RS |
2687 | /* Adjust search data for this change. */ |
2688 | { | |
5b88a2c5 | 2689 | int oldend = search_regs.end[sub]; |
41c01205 | 2690 | int oldstart = search_regs.start[sub]; |
98e942e0 RS |
2691 | int change = newpoint - search_regs.end[sub]; |
2692 | int i; | |
2693 | ||
2694 | for (i = 0; i < search_regs.num_regs; i++) | |
2695 | { | |
41c01205 | 2696 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2697 | search_regs.start[i] += change; |
41c01205 DK |
2698 | else if (search_regs.start[i] > oldstart) |
2699 | search_regs.start[i] = oldstart; | |
2700 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2701 | search_regs.end[i] += change; |
41c01205 DK |
2702 | else if (search_regs.end[i] > oldstart) |
2703 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2704 | } |
2705 | } | |
2706 | ||
b0eba991 | 2707 | /* Put point back where it was in the text. */ |
8d808a65 | 2708 | if (opoint <= 0) |
fa8ed3e0 | 2709 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2710 | else |
fa8ed3e0 | 2711 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2712 | |
2713 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2714 | move_if_not_intangible (newpoint); |
177c0ea7 | 2715 | |
ca1d1d23 JB |
2716 | return Qnil; |
2717 | } | |
2718 | \f | |
2719 | static Lisp_Object | |
2720 | match_limit (num, beginningp) | |
2721 | Lisp_Object num; | |
2722 | int beginningp; | |
2723 | { | |
2724 | register int n; | |
2725 | ||
b7826503 | 2726 | CHECK_NUMBER (num); |
ca1d1d23 | 2727 | n = XINT (num); |
f90a5bf5 | 2728 | if (n < 0) |
bd2cbd56 | 2729 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2730 | if (search_regs.num_regs <= 0) |
2731 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2732 | if (n >= search_regs.num_regs |
4746118a | 2733 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2734 | return Qnil; |
2735 | return (make_number ((beginningp) ? search_regs.start[n] | |
2736 | : search_regs.end[n])); | |
2737 | } | |
2738 | ||
2739 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
8c1a1077 PJ |
2740 | doc: /* Return position of start of text matched by last search. |
2741 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2742 | regexp. | |
2743 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2744 | SUBEXP pairs. | |
2745 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2746 | (subexp) | |
5806161b | 2747 | Lisp_Object subexp; |
ca1d1d23 | 2748 | { |
5806161b | 2749 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2750 | } |
2751 | ||
2752 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
8c1a1077 PJ |
2753 | doc: /* Return position of end of text matched by last search. |
2754 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2755 | regexp. | |
2756 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2757 | SUBEXP pairs. | |
2758 | Zero means the entire text matched by the whole regexp or whole string. */) | |
2759 | (subexp) | |
5806161b | 2760 | Lisp_Object subexp; |
ca1d1d23 | 2761 | { |
5806161b | 2762 | return match_limit (subexp, 0); |
177c0ea7 | 2763 | } |
ca1d1d23 | 2764 | |
abd0071c | 2765 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 3, 0, |
8c1a1077 PJ |
2766 | doc: /* Return a list containing all info on what the last search matched. |
2767 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2768 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2769 | if the last match was on a buffer; integers or nil if a string was matched. | |
2770 | Use `store-match-data' to reinstate the data in this list. | |
2771 | ||
41c01205 DK |
2772 | If INTEGERS (the optional first argument) is non-nil, always use |
2773 | integers \(rather than markers) to represent buffer positions. In | |
2774 | this case, and if the last match was in a buffer, the buffer will get | |
2775 | stored as one additional element at the end of the list. | |
2776 | ||
abd0071c KS |
2777 | If REUSE is a list, reuse it as part of the value. If REUSE is long |
2778 | enough to hold all the values, and if INTEGERS is non-nil, no consing | |
2779 | is done. | |
2780 | ||
2781 | If optional third arg RESEAT is non-nil, any previous markers on the | |
2782 | REUSE list will be modified to point to nowhere. | |
2783 | ||
140a6b7e | 2784 | Return value is undefined if the last search failed. */) |
abd0071c KS |
2785 | (integers, reuse, reseat) |
2786 | Lisp_Object integers, reuse, reseat; | |
ca1d1d23 | 2787 | { |
56256c2a | 2788 | Lisp_Object tail, prev; |
4746118a | 2789 | Lisp_Object *data; |
ca1d1d23 JB |
2790 | int i, len; |
2791 | ||
abd0071c KS |
2792 | if (!NILP (reseat)) |
2793 | for (tail = reuse; CONSP (tail); tail = XCDR (tail)) | |
2794 | if (MARKERP (XCAR (tail))) | |
2795 | { | |
51f10faa | 2796 | unchain_marker (XMARKER (XCAR (tail))); |
abd0071c KS |
2797 | XSETCAR (tail, Qnil); |
2798 | } | |
2799 | ||
daa37602 | 2800 | if (NILP (last_thing_searched)) |
c36bcf1b | 2801 | return Qnil; |
daa37602 | 2802 | |
6bbd7a29 GM |
2803 | prev = Qnil; |
2804 | ||
41c01205 | 2805 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs + 1) |
4746118a JB |
2806 | * sizeof (Lisp_Object)); |
2807 | ||
41c01205 | 2808 | len = 0; |
4746118a | 2809 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2810 | { |
2811 | int start = search_regs.start[i]; | |
2812 | if (start >= 0) | |
2813 | { | |
56256c2a RS |
2814 | if (EQ (last_thing_searched, Qt) |
2815 | || ! NILP (integers)) | |
ca1d1d23 | 2816 | { |
c235cce7 KH |
2817 | XSETFASTINT (data[2 * i], start); |
2818 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2819 | } |
0ed62dc7 | 2820 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2821 | { |
2822 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2823 | Fset_marker (data[2 * i], |
2824 | make_number (start), | |
2825 | last_thing_searched); | |
ca1d1d23 JB |
2826 | data[2 * i + 1] = Fmake_marker (); |
2827 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2828 | make_number (search_regs.end[i]), |
daa37602 | 2829 | last_thing_searched); |
ca1d1d23 | 2830 | } |
daa37602 JB |
2831 | else |
2832 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2833 | abort (); | |
2834 | ||
abd0071c | 2835 | len = 2 * i + 2; |
ca1d1d23 JB |
2836 | } |
2837 | else | |
abd0071c | 2838 | data[2 * i] = data[2 * i + 1] = Qnil; |
ca1d1d23 | 2839 | } |
56256c2a | 2840 | |
bd2cbd56 | 2841 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2842 | { |
bd2cbd56 | 2843 | data[len] = last_thing_searched; |
41c01205 DK |
2844 | len++; |
2845 | } | |
2846 | ||
56256c2a RS |
2847 | /* If REUSE is not usable, cons up the values and return them. */ |
2848 | if (! CONSP (reuse)) | |
41c01205 | 2849 | return Flist (len, data); |
56256c2a RS |
2850 | |
2851 | /* If REUSE is a list, store as many value elements as will fit | |
2852 | into the elements of REUSE. */ | |
2853 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2854 | i++, tail = XCDR (tail)) |
56256c2a | 2855 | { |
41c01205 | 2856 | if (i < len) |
f3fbd155 | 2857 | XSETCAR (tail, data[i]); |
56256c2a | 2858 | else |
f3fbd155 | 2859 | XSETCAR (tail, Qnil); |
56256c2a RS |
2860 | prev = tail; |
2861 | } | |
2862 | ||
2863 | /* If we couldn't fit all value elements into REUSE, | |
2864 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2865 | if (i < len) |
2866 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2867 | |
2868 | return reuse; | |
ca1d1d23 JB |
2869 | } |
2870 | ||
51f10faa KS |
2871 | /* Internal usage only: |
2872 | If RESEAT is `evaporate', put the markers back on the free list | |
2873 | immediately. No other references to the markers must exist in this case, | |
2874 | so it is used only internally on the unwind stack and save-match-data from | |
2875 | Lisp. */ | |
ca1d1d23 | 2876 | |
abd0071c | 2877 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 2, 0, |
8c1a1077 | 2878 | doc: /* Set internal data on last search match from elements of LIST. |
abd0071c KS |
2879 | LIST should have been created by calling `match-data' previously. |
2880 | ||
51f10faa | 2881 | If optional arg RESEAT is non-nil, make markers on LIST point nowhere. */) |
abd0071c KS |
2882 | (list, reseat) |
2883 | register Lisp_Object list, reseat; | |
ca1d1d23 JB |
2884 | { |
2885 | register int i; | |
2886 | register Lisp_Object marker; | |
2887 | ||
7074fde6 FP |
2888 | if (running_asynch_code) |
2889 | save_search_regs (); | |
2890 | ||
29100cea | 2891 | CHECK_LIST (list); |
ca1d1d23 | 2892 | |
41c01205 DK |
2893 | /* Unless we find a marker with a buffer or an explicit buffer |
2894 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
2895 | last_thing_searched = Qt; |
2896 | ||
4746118a JB |
2897 | /* Allocate registers if they don't already exist. */ |
2898 | { | |
d084e942 | 2899 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2900 | |
2901 | if (length > search_regs.num_regs) | |
2902 | { | |
1113d9db JB |
2903 | if (search_regs.num_regs == 0) |
2904 | { | |
2905 | search_regs.start | |
2906 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2907 | search_regs.end | |
2908 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2909 | } | |
4746118a | 2910 | else |
1113d9db JB |
2911 | { |
2912 | search_regs.start | |
2913 | = (regoff_t *) xrealloc (search_regs.start, | |
2914 | length * sizeof (regoff_t)); | |
2915 | search_regs.end | |
2916 | = (regoff_t *) xrealloc (search_regs.end, | |
2917 | length * sizeof (regoff_t)); | |
2918 | } | |
4746118a | 2919 | |
e62371e9 KH |
2920 | for (i = search_regs.num_regs; i < length; i++) |
2921 | search_regs.start[i] = -1; | |
2922 | ||
487282dc | 2923 | search_regs.num_regs = length; |
4746118a | 2924 | } |
ca1d1d23 | 2925 | |
abd0071c | 2926 | for (i = 0; CONSP (list); i++) |
41c01205 | 2927 | { |
abd0071c | 2928 | marker = XCAR (list); |
bd2cbd56 | 2929 | if (BUFFERP (marker)) |
c3762cbd | 2930 | { |
bd2cbd56 | 2931 | last_thing_searched = marker; |
c3762cbd DK |
2932 | break; |
2933 | } | |
2934 | if (i >= length) | |
2935 | break; | |
41c01205 DK |
2936 | if (NILP (marker)) |
2937 | { | |
2938 | search_regs.start[i] = -1; | |
abd0071c | 2939 | list = XCDR (list); |
41c01205 DK |
2940 | } |
2941 | else | |
2942 | { | |
2943 | int from; | |
abd0071c | 2944 | Lisp_Object m; |
e2811828 | 2945 | |
abd0071c | 2946 | m = marker; |
41c01205 DK |
2947 | if (MARKERP (marker)) |
2948 | { | |
2949 | if (XMARKER (marker)->buffer == 0) | |
2950 | XSETFASTINT (marker, 0); | |
2951 | else | |
2952 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
2953 | } | |
e2811828 | 2954 | |
41c01205 DK |
2955 | CHECK_NUMBER_COERCE_MARKER (marker); |
2956 | from = XINT (marker); | |
e2811828 | 2957 | |
abd0071c KS |
2958 | if (!NILP (reseat) && MARKERP (m)) |
2959 | { | |
2960 | if (EQ (reseat, Qevaporate)) | |
2961 | free_marker (m); | |
2962 | else | |
2963 | unchain_marker (XMARKER (m)); | |
9ad54a7e | 2964 | XSETCAR (list, Qnil); |
abd0071c KS |
2965 | } |
2966 | ||
2967 | if ((list = XCDR (list), !CONSP (list))) | |
2968 | break; | |
2969 | ||
2970 | m = marker = XCAR (list); | |
2971 | ||
41c01205 DK |
2972 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
2973 | XSETFASTINT (marker, 0); | |
e2811828 | 2974 | |
41c01205 DK |
2975 | CHECK_NUMBER_COERCE_MARKER (marker); |
2976 | search_regs.start[i] = from; | |
2977 | search_regs.end[i] = XINT (marker); | |
abd0071c KS |
2978 | |
2979 | if (!NILP (reseat) && MARKERP (m)) | |
2980 | { | |
2981 | if (EQ (reseat, Qevaporate)) | |
2982 | free_marker (m); | |
2983 | else | |
2984 | unchain_marker (XMARKER (m)); | |
9ad54a7e | 2985 | XSETCAR (list, Qnil); |
abd0071c | 2986 | } |
41c01205 | 2987 | } |
abd0071c | 2988 | list = XCDR (list); |
41c01205 | 2989 | } |
ca1d1d23 | 2990 | |
41c01205 DK |
2991 | for (; i < search_regs.num_regs; i++) |
2992 | search_regs.start[i] = -1; | |
2993 | } | |
ca1d1d23 | 2994 | |
177c0ea7 | 2995 | return Qnil; |
ca1d1d23 JB |
2996 | } |
2997 | ||
7074fde6 FP |
2998 | /* If non-zero the match data have been saved in saved_search_regs |
2999 | during the execution of a sentinel or filter. */ | |
75ebf74b | 3000 | static int search_regs_saved; |
7074fde6 | 3001 | static struct re_registers saved_search_regs; |
41c01205 | 3002 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
3003 | |
3004 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
3005 | if asynchronous code (filter or sentinel) is running. */ | |
3006 | static void | |
3007 | save_search_regs () | |
3008 | { | |
3009 | if (!search_regs_saved) | |
3010 | { | |
3011 | saved_search_regs.num_regs = search_regs.num_regs; | |
3012 | saved_search_regs.start = search_regs.start; | |
3013 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
3014 | saved_last_thing_searched = last_thing_searched; |
3015 | last_thing_searched = Qnil; | |
7074fde6 | 3016 | search_regs.num_regs = 0; |
2d4a771a RS |
3017 | search_regs.start = 0; |
3018 | search_regs.end = 0; | |
7074fde6 FP |
3019 | |
3020 | search_regs_saved = 1; | |
3021 | } | |
3022 | } | |
3023 | ||
3024 | /* Called upon exit from filters and sentinels. */ | |
3025 | void | |
abd0071c | 3026 | restore_search_regs () |
7074fde6 FP |
3027 | { |
3028 | if (search_regs_saved) | |
3029 | { | |
3030 | if (search_regs.num_regs > 0) | |
3031 | { | |
3032 | xfree (search_regs.start); | |
3033 | xfree (search_regs.end); | |
3034 | } | |
3035 | search_regs.num_regs = saved_search_regs.num_regs; | |
3036 | search_regs.start = saved_search_regs.start; | |
3037 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
3038 | last_thing_searched = saved_last_thing_searched; |
3039 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
3040 | search_regs_saved = 0; |
3041 | } | |
3042 | } | |
3043 | ||
abd0071c KS |
3044 | static Lisp_Object |
3045 | unwind_set_match_data (list) | |
3046 | Lisp_Object list; | |
3047 | { | |
51f10faa | 3048 | /* It is safe to free (evaporate) the markers immediately. */ |
abd0071c KS |
3049 | return Fset_match_data (list, Qevaporate); |
3050 | } | |
3051 | ||
3052 | /* Called to unwind protect the match data. */ | |
3053 | void | |
3054 | record_unwind_save_match_data () | |
3055 | { | |
3056 | record_unwind_protect (unwind_set_match_data, | |
3057 | Fmatch_data (Qnil, Qnil, Qnil)); | |
3058 | } | |
3059 | ||
ca1d1d23 JB |
3060 | /* Quote a string to inactivate reg-expr chars */ |
3061 | ||
3062 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 PJ |
3063 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
3064 | (string) | |
5806161b | 3065 | Lisp_Object string; |
ca1d1d23 JB |
3066 | { |
3067 | register unsigned char *in, *out, *end; | |
3068 | register unsigned char *temp; | |
0c8533c6 | 3069 | int backslashes_added = 0; |
ca1d1d23 | 3070 | |
b7826503 | 3071 | CHECK_STRING (string); |
ca1d1d23 | 3072 | |
d5db4077 | 3073 | temp = (unsigned char *) alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
3074 | |
3075 | /* Now copy the data into the new string, inserting escapes. */ | |
3076 | ||
d5db4077 KR |
3077 | in = SDATA (string); |
3078 | end = in + SBYTES (string); | |
177c0ea7 | 3079 | out = temp; |
ca1d1d23 JB |
3080 | |
3081 | for (; in != end; in++) | |
3082 | { | |
66bc6082 | 3083 | if (*in == '[' |
ca1d1d23 JB |
3084 | || *in == '*' || *in == '.' || *in == '\\' |
3085 | || *in == '?' || *in == '+' | |
3086 | || *in == '^' || *in == '$') | |
0c8533c6 | 3087 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
3088 | *out++ = *in; |
3089 | } | |
3090 | ||
3f8100f1 | 3091 | return make_specified_string (temp, |
d5db4077 | 3092 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
3093 | out - temp, |
3094 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 3095 | } |
177c0ea7 | 3096 | \f |
dfcf069d | 3097 | void |
ca1d1d23 JB |
3098 | syms_of_search () |
3099 | { | |
3100 | register int i; | |
3101 | ||
487282dc KH |
3102 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
3103 | { | |
3104 | searchbufs[i].buf.allocated = 100; | |
b23c0a83 | 3105 | searchbufs[i].buf.buffer = (unsigned char *) xmalloc (100); |
487282dc KH |
3106 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
3107 | searchbufs[i].regexp = Qnil; | |
ecdb561e | 3108 | searchbufs[i].whitespace_regexp = Qnil; |
b69e3c18 | 3109 | searchbufs[i].syntax_table = Qnil; |
487282dc | 3110 | staticpro (&searchbufs[i].regexp); |
aa77b5ce | 3111 | staticpro (&searchbufs[i].whitespace_regexp); |
b69e3c18 | 3112 | staticpro (&searchbufs[i].syntax_table); |
487282dc KH |
3113 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); |
3114 | } | |
3115 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 JB |
3116 | |
3117 | Qsearch_failed = intern ("search-failed"); | |
3118 | staticpro (&Qsearch_failed); | |
3119 | Qinvalid_regexp = intern ("invalid-regexp"); | |
3120 | staticpro (&Qinvalid_regexp); | |
3121 | ||
3122 | Fput (Qsearch_failed, Qerror_conditions, | |
3123 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); | |
3124 | Fput (Qsearch_failed, Qerror_message, | |
3125 | build_string ("Search failed")); | |
3126 | ||
3127 | Fput (Qinvalid_regexp, Qerror_conditions, | |
3128 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); | |
3129 | Fput (Qinvalid_regexp, Qerror_message, | |
3130 | build_string ("Invalid regexp")); | |
3131 | ||
daa37602 JB |
3132 | last_thing_searched = Qnil; |
3133 | staticpro (&last_thing_searched); | |
3134 | ||
0f6af254 DK |
3135 | saved_last_thing_searched = Qnil; |
3136 | staticpro (&saved_last_thing_searched); | |
3137 | ||
41a33295 | 3138 | DEFVAR_LISP ("search-spaces-regexp", &Vsearch_spaces_regexp, |
e2811828 | 3139 | doc: /* Regexp to substitute for bunches of spaces in regexp search. |
f31a9a68 RS |
3140 | Some commands use this for user-specified regexps. |
3141 | Spaces that occur inside character classes or repetition operators | |
3142 | or other such regexp constructs are not replaced with this. | |
3143 | A value of nil (which is the normal value) means treat spaces literally. */); | |
41a33295 | 3144 | Vsearch_spaces_regexp = Qnil; |
f31a9a68 | 3145 | |
ca1d1d23 | 3146 | defsubr (&Slooking_at); |
b819a390 RS |
3147 | defsubr (&Sposix_looking_at); |
3148 | defsubr (&Sstring_match); | |
3149 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
3150 | defsubr (&Ssearch_forward); |
3151 | defsubr (&Ssearch_backward); | |
3152 | defsubr (&Sword_search_forward); | |
3153 | defsubr (&Sword_search_backward); | |
3154 | defsubr (&Sre_search_forward); | |
3155 | defsubr (&Sre_search_backward); | |
b819a390 RS |
3156 | defsubr (&Sposix_search_forward); |
3157 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
3158 | defsubr (&Sreplace_match); |
3159 | defsubr (&Smatch_beginning); | |
3160 | defsubr (&Smatch_end); | |
3161 | defsubr (&Smatch_data); | |
3f1c005b | 3162 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
3163 | defsubr (&Sregexp_quote); |
3164 | } | |
76fc1ea2 KH |
3165 | |
3166 | /* arch-tag: a6059d79-0552-4f14-a2cb-d379a4e3c78f | |
3167 | (do not change this comment) */ |