Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
429ab54e | 2 | Copyright (C) 1985, 1986, 1987, 1993, 1994, 1997, 1998, 1999, 2001, 2002, |
114f9c96 | 3 | 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010 |
8cabe764 | 4 | Free Software Foundation, Inc. |
ca1d1d23 JB |
5 | |
6 | This file is part of GNU Emacs. | |
7 | ||
9ec0b715 | 8 | GNU Emacs is free software: you can redistribute it and/or modify |
ca1d1d23 | 9 | it under the terms of the GNU General Public License as published by |
9ec0b715 GM |
10 | the Free Software Foundation, either version 3 of the License, or |
11 | (at your option) any later version. | |
ca1d1d23 JB |
12 | |
13 | GNU Emacs is distributed in the hope that it will be useful, | |
14 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
15 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
16 | GNU General Public License for more details. | |
17 | ||
18 | You should have received a copy of the GNU General Public License | |
9ec0b715 | 19 | along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */ |
ca1d1d23 JB |
20 | |
21 | ||
18160b98 | 22 | #include <config.h> |
d7306fe6 | 23 | #include <setjmp.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" |
89ad6492 | 29 | #include "charset.h" |
9169c321 | 30 | #include "region-cache.h" |
ca1d1d23 | 31 | #include "commands.h" |
9ac0d9e0 | 32 | #include "blockinput.h" |
bf1760bb | 33 | #include "intervals.h" |
4746118a | 34 | |
ca1d1d23 JB |
35 | #include <sys/types.h> |
36 | #include "regex.h" | |
37 | ||
1d288aef | 38 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 39 | |
487282dc KH |
40 | /* If the regexp is non-nil, then the buffer contains the compiled form |
41 | of that regexp, suitable for searching. */ | |
1d288aef RS |
42 | struct regexp_cache |
43 | { | |
487282dc | 44 | struct regexp_cache *next; |
ecdb561e | 45 | Lisp_Object regexp, whitespace_regexp; |
b69e3c18 | 46 | /* Syntax table for which the regexp applies. We need this because |
58e95211 SM |
47 | of character classes. If this is t, then the compiled pattern is valid |
48 | for any syntax-table. */ | |
b69e3c18 | 49 | Lisp_Object syntax_table; |
487282dc KH |
50 | struct re_pattern_buffer buf; |
51 | char fastmap[0400]; | |
b819a390 RS |
52 | /* Nonzero means regexp was compiled to do full POSIX backtracking. */ |
53 | char posix; | |
487282dc | 54 | }; |
ca1d1d23 | 55 | |
487282dc KH |
56 | /* The instances of that struct. */ |
57 | struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; | |
ca1d1d23 | 58 | |
487282dc KH |
59 | /* The head of the linked list; points to the most recently used buffer. */ |
60 | struct regexp_cache *searchbuf_head; | |
ca1d1d23 | 61 | |
ca1d1d23 | 62 | |
4746118a JB |
63 | /* Every call to re_match, etc., must pass &search_regs as the regs |
64 | argument unless you can show it is unnecessary (i.e., if re_match | |
65 | is certainly going to be called again before region-around-match | |
66 | can be called). | |
67 | ||
68 | Since the registers are now dynamically allocated, we need to make | |
69 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
70 | been allocated by checking search_regs.num_regs. |
71 | ||
72 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
73 | buffer using bits in the re_pattern_buffer. This means that whenever |
74 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
75 | allocated any registers, and will allocate new registers the next |
76 | time you call a searching or matching function. Therefore, we need | |
77 | to call re_set_registers after compiling a new pattern or after | |
78 | setting the match registers, so that the regex functions will be | |
79 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
80 | static struct re_registers search_regs; |
81 | ||
daa37602 JB |
82 | /* The buffer in which the last search was performed, or |
83 | Qt if the last search was done in a string; | |
84 | Qnil if no searching has been done yet. */ | |
85 | static Lisp_Object last_thing_searched; | |
ca1d1d23 | 86 | |
8e6208c5 | 87 | /* error condition signaled when regexp compile_pattern fails */ |
ca1d1d23 JB |
88 | |
89 | Lisp_Object Qinvalid_regexp; | |
90 | ||
06f77a8a KS |
91 | /* Error condition used for failing searches */ |
92 | Lisp_Object Qsearch_failed; | |
93 | ||
41a33295 | 94 | Lisp_Object Vsearch_spaces_regexp; |
f31a9a68 | 95 | |
ef887810 RS |
96 | /* If non-nil, the match data will not be changed during call to |
97 | searching or matching functions. This variable is for internal use | |
98 | only. */ | |
99 | Lisp_Object Vinhibit_changing_match_data; | |
100 | ||
f57e2426 J |
101 | static void set_search_regs (EMACS_INT, EMACS_INT); |
102 | static void save_search_regs (void); | |
103 | static EMACS_INT simple_search (int, unsigned char *, int, int, | |
104 | Lisp_Object, EMACS_INT, EMACS_INT, | |
105 | EMACS_INT, EMACS_INT); | |
106 | static EMACS_INT boyer_moore (int, unsigned char *, int, int, | |
107 | Lisp_Object, Lisp_Object, | |
108 | EMACS_INT, EMACS_INT, | |
109 | EMACS_INT, EMACS_INT, int); | |
110 | static EMACS_INT search_buffer (Lisp_Object, EMACS_INT, EMACS_INT, | |
111 | EMACS_INT, EMACS_INT, int, int, | |
112 | Lisp_Object, Lisp_Object, int); | |
971de7fb | 113 | static void matcher_overflow (void) NO_RETURN; |
b819a390 | 114 | |
ca1d1d23 | 115 | static void |
971de7fb | 116 | matcher_overflow (void) |
ca1d1d23 JB |
117 | { |
118 | error ("Stack overflow in regexp matcher"); | |
119 | } | |
120 | ||
b819a390 RS |
121 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
122 | PATTERN is the pattern to compile. | |
123 | CP is the place to put the result. | |
facdc750 | 124 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
125 | REGP is the structure that says where to store the "register" |
126 | values that will result from matching this pattern. | |
127 | If it is 0, we should compile the pattern not to record any | |
128 | subexpression bounds. | |
129 | POSIX is nonzero if we want full backtracking (POSIX style) | |
5679531d | 130 | for this pattern. 0 means backtrack only enough to get a valid match. |
bbc73b48 KH |
131 | |
132 | The behavior also depends on Vsearch_spaces_regexp. */ | |
ca1d1d23 | 133 | |
487282dc | 134 | static void |
971de7fb | 135 | compile_pattern_1 (struct regexp_cache *cp, Lisp_Object pattern, Lisp_Object translate, struct re_registers *regp, int posix) |
ca1d1d23 | 136 | { |
d451e4db | 137 | char *val; |
b819a390 | 138 | reg_syntax_t old; |
ca1d1d23 | 139 | |
487282dc | 140 | cp->regexp = Qnil; |
59fab369 | 141 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 142 | cp->posix = posix; |
93daa011 | 143 | cp->buf.multibyte = STRING_MULTIBYTE (pattern); |
89ad6492 | 144 | cp->buf.charset_unibyte = charset_unibyte; |
d8c85250 CY |
145 | if (STRINGP (Vsearch_spaces_regexp)) |
146 | cp->whitespace_regexp = Vsearch_spaces_regexp; | |
147 | else | |
148 | cp->whitespace_regexp = Qnil; | |
149 | ||
e7b2dd2e RS |
150 | /* rms: I think BLOCK_INPUT is not needed here any more, |
151 | because regex.c defines malloc to call xmalloc. | |
152 | Using BLOCK_INPUT here means the debugger won't run if an error occurs. | |
153 | So let's turn it off. */ | |
154 | /* BLOCK_INPUT; */ | |
fb4a568d | 155 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 156 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
d8c85250 CY |
157 | |
158 | if (STRINGP (Vsearch_spaces_regexp)) | |
159 | re_set_whitespace_regexp (SDATA (Vsearch_spaces_regexp)); | |
160 | else | |
161 | re_set_whitespace_regexp (NULL); | |
bbc73b48 | 162 | |
8f924df7 KH |
163 | val = (char *) re_compile_pattern ((char *) SDATA (pattern), |
164 | SBYTES (pattern), &cp->buf); | |
bbc73b48 | 165 | |
58e95211 SM |
166 | /* If the compiled pattern hard codes some of the contents of the |
167 | syntax-table, it can only be reused with *this* syntax table. */ | |
168 | cp->syntax_table = cp->buf.used_syntax ? current_buffer->syntax_table : Qt; | |
169 | ||
bbc73b48 KH |
170 | re_set_whitespace_regexp (NULL); |
171 | ||
b819a390 | 172 | re_set_syntax (old); |
e7b2dd2e | 173 | /* UNBLOCK_INPUT; */ |
ca1d1d23 | 174 | if (val) |
06f77a8a | 175 | xsignal1 (Qinvalid_regexp, build_string (val)); |
1113d9db | 176 | |
487282dc | 177 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
178 | } |
179 | ||
6efc7887 RS |
180 | /* Shrink each compiled regexp buffer in the cache |
181 | to the size actually used right now. | |
182 | This is called from garbage collection. */ | |
183 | ||
184 | void | |
971de7fb | 185 | shrink_regexp_cache (void) |
6efc7887 | 186 | { |
a968f437 | 187 | struct regexp_cache *cp; |
6efc7887 RS |
188 | |
189 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
190 | { | |
191 | cp->buf.allocated = cp->buf.used; | |
192 | cp->buf.buffer | |
b23c0a83 | 193 | = (unsigned char *) xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
194 | } |
195 | } | |
196 | ||
54dd3310 SM |
197 | /* Clear the regexp cache w.r.t. a particular syntax table, |
198 | because it was changed. | |
e5b94d44 CY |
199 | There is no danger of memory leak here because re_compile_pattern |
200 | automagically manages the memory in each re_pattern_buffer struct, | |
201 | based on its `allocated' and `buffer' values. */ | |
202 | void | |
971de7fb | 203 | clear_regexp_cache (void) |
e5b94d44 CY |
204 | { |
205 | int i; | |
206 | ||
207 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) | |
920fd1fc EZ |
208 | /* It's tempting to compare with the syntax-table we've actually changed, |
209 | but it's not sufficient because char-table inheritance means that | |
54dd3310 SM |
210 | modifying one syntax-table can change others at the same time. */ |
211 | if (!EQ (searchbufs[i].syntax_table, Qt)) | |
212 | searchbufs[i].regexp = Qnil; | |
e5b94d44 CY |
213 | } |
214 | ||
487282dc | 215 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
216 | the cache. |
217 | PATTERN is the pattern to compile. | |
facdc750 | 218 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
219 | REGP is the structure that says where to store the "register" |
220 | values that will result from matching this pattern. | |
221 | If it is 0, we should compile the pattern not to record any | |
222 | subexpression bounds. | |
223 | POSIX is nonzero if we want full backtracking (POSIX style) | |
224 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
225 | |
226 | struct re_pattern_buffer * | |
971de7fb | 227 | compile_pattern (Lisp_Object pattern, struct re_registers *regp, Lisp_Object translate, int posix, int multibyte) |
487282dc KH |
228 | { |
229 | struct regexp_cache *cp, **cpp; | |
230 | ||
231 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
232 | { | |
233 | cp = *cpp; | |
f1b9c7c1 KR |
234 | /* Entries are initialized to nil, and may be set to nil by |
235 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
49a5f770 KR |
236 | string accessors in those cases. However, compile_pattern_1 |
237 | is only applied to the cache entry we pick here to reuse. So | |
238 | nil should never appear before a non-nil entry. */ | |
7c752c80 | 239 | if (NILP (cp->regexp)) |
f1b9c7c1 | 240 | goto compile_it; |
d5db4077 | 241 | if (SCHARS (cp->regexp) == SCHARS (pattern) |
cf69b13e | 242 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 243 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 244 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d | 245 | && cp->posix == posix |
58e95211 SM |
246 | && (EQ (cp->syntax_table, Qt) |
247 | || EQ (cp->syntax_table, current_buffer->syntax_table)) | |
89ad6492 KH |
248 | && !NILP (Fequal (cp->whitespace_regexp, Vsearch_spaces_regexp)) |
249 | && cp->buf.charset_unibyte == charset_unibyte) | |
487282dc KH |
250 | break; |
251 | ||
f1b9c7c1 KR |
252 | /* If we're at the end of the cache, compile into the nil cell |
253 | we found, or the last (least recently used) cell with a | |
254 | string value. */ | |
487282dc KH |
255 | if (cp->next == 0) |
256 | { | |
f1b9c7c1 | 257 | compile_it: |
89ad6492 | 258 | compile_pattern_1 (cp, pattern, translate, regp, posix); |
487282dc KH |
259 | break; |
260 | } | |
261 | } | |
262 | ||
263 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
264 | either because we found it in the cache or because we just compiled it. | |
265 | Move it to the front of the queue to mark it as most recently used. */ | |
266 | *cpp = cp->next; | |
267 | cp->next = searchbuf_head; | |
268 | searchbuf_head = cp; | |
1113d9db | 269 | |
6639708c RS |
270 | /* Advise the searching functions about the space we have allocated |
271 | for register data. */ | |
272 | if (regp) | |
273 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
274 | ||
78edd3b7 | 275 | /* The compiled pattern can be used both for multibyte and unibyte |
89ad6492 KH |
276 | target. But, we have to tell which the pattern is used for. */ |
277 | cp->buf.target_multibyte = multibyte; | |
278 | ||
487282dc | 279 | return &cp->buf; |
ca1d1d23 JB |
280 | } |
281 | ||
ca1d1d23 | 282 | \f |
b819a390 | 283 | static Lisp_Object |
971de7fb | 284 | looking_at_1 (Lisp_Object string, int posix) |
ca1d1d23 JB |
285 | { |
286 | Lisp_Object val; | |
287 | unsigned char *p1, *p2; | |
e7deaab0 | 288 | EMACS_INT s1, s2; |
ca1d1d23 | 289 | register int i; |
487282dc | 290 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 291 | |
7074fde6 FP |
292 | if (running_asynch_code) |
293 | save_search_regs (); | |
294 | ||
910c747a RS |
295 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
296 | XCHAR_TABLE (current_buffer->case_canon_table)->extras[2] | |
297 | = current_buffer->case_eqv_table; | |
298 | ||
b7826503 | 299 | CHECK_STRING (string); |
ef887810 RS |
300 | bufp = compile_pattern (string, |
301 | (NILP (Vinhibit_changing_match_data) | |
302 | ? &search_regs : NULL), | |
487282dc | 303 | (!NILP (current_buffer->case_fold_search) |
0190922f | 304 | ? current_buffer->case_canon_table : Qnil), |
0c8533c6 RS |
305 | posix, |
306 | !NILP (current_buffer->enable_multibyte_characters)); | |
ca1d1d23 JB |
307 | |
308 | immediate_quit = 1; | |
309 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
310 | ||
311 | /* Get pointers and sizes of the two strings | |
312 | that make up the visible portion of the buffer. */ | |
313 | ||
314 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 315 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 316 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 317 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
318 | if (s1 < 0) |
319 | { | |
320 | p2 = p1; | |
fa8ed3e0 | 321 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
322 | s1 = 0; |
323 | } | |
324 | if (s2 < 0) | |
325 | { | |
fa8ed3e0 | 326 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
327 | s2 = 0; |
328 | } | |
8bb43c28 RS |
329 | |
330 | re_match_object = Qnil; | |
177c0ea7 | 331 | |
487282dc | 332 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
ef887810 RS |
333 | PT_BYTE - BEGV_BYTE, |
334 | (NILP (Vinhibit_changing_match_data) | |
335 | ? &search_regs : NULL), | |
fa8ed3e0 | 336 | ZV_BYTE - BEGV_BYTE); |
de182d70 | 337 | immediate_quit = 0; |
177c0ea7 | 338 | |
ca1d1d23 JB |
339 | if (i == -2) |
340 | matcher_overflow (); | |
341 | ||
342 | val = (0 <= i ? Qt : Qnil); | |
ef887810 | 343 | if (NILP (Vinhibit_changing_match_data) && i >= 0) |
fa8ed3e0 RS |
344 | for (i = 0; i < search_regs.num_regs; i++) |
345 | if (search_regs.start[i] >= 0) | |
346 | { | |
347 | search_regs.start[i] | |
348 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
349 | search_regs.end[i] | |
350 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
351 | } | |
ef887810 RS |
352 | |
353 | /* Set last_thing_searched only when match data is changed. */ | |
354 | if (NILP (Vinhibit_changing_match_data)) | |
355 | XSETBUFFER (last_thing_searched, current_buffer); | |
356 | ||
ca1d1d23 JB |
357 | return val; |
358 | } | |
359 | ||
b819a390 | 360 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
361 | doc: /* Return t if text after point matches regular expression REGEXP. |
362 | This function modifies the match data that `match-beginning', | |
363 | `match-end' and `match-data' access; save and restore the match | |
364 | data if you want to preserve them. */) | |
5842a27b | 365 | (Lisp_Object regexp) |
b819a390 | 366 | { |
94f94972 | 367 | return looking_at_1 (regexp, 0); |
b819a390 RS |
368 | } |
369 | ||
370 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
371 | doc: /* Return t if text after point matches regular expression REGEXP. |
372 | Find the longest match, in accord with Posix regular expression rules. | |
373 | This function modifies the match data that `match-beginning', | |
374 | `match-end' and `match-data' access; save and restore the match | |
375 | data if you want to preserve them. */) | |
5842a27b | 376 | (Lisp_Object regexp) |
b819a390 | 377 | { |
94f94972 | 378 | return looking_at_1 (regexp, 1); |
b819a390 RS |
379 | } |
380 | \f | |
381 | static Lisp_Object | |
971de7fb | 382 | string_match_1 (Lisp_Object regexp, Lisp_Object string, Lisp_Object start, int posix) |
ca1d1d23 JB |
383 | { |
384 | int val; | |
487282dc | 385 | struct re_pattern_buffer *bufp; |
e7deaab0 | 386 | EMACS_INT pos, pos_byte; |
0c8533c6 | 387 | int i; |
ca1d1d23 | 388 | |
7074fde6 FP |
389 | if (running_asynch_code) |
390 | save_search_regs (); | |
391 | ||
b7826503 PJ |
392 | CHECK_STRING (regexp); |
393 | CHECK_STRING (string); | |
ca1d1d23 JB |
394 | |
395 | if (NILP (start)) | |
0c8533c6 | 396 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
397 | else |
398 | { | |
d5db4077 | 399 | int len = SCHARS (string); |
ca1d1d23 | 400 | |
b7826503 | 401 | CHECK_NUMBER (start); |
0c8533c6 RS |
402 | pos = XINT (start); |
403 | if (pos < 0 && -pos <= len) | |
404 | pos = len + pos; | |
405 | else if (0 > pos || pos > len) | |
ca1d1d23 | 406 | args_out_of_range (string, start); |
0c8533c6 | 407 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
408 | } |
409 | ||
910c747a RS |
410 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
411 | XCHAR_TABLE (current_buffer->case_canon_table)->extras[2] | |
412 | = current_buffer->case_eqv_table; | |
413 | ||
ef887810 RS |
414 | bufp = compile_pattern (regexp, |
415 | (NILP (Vinhibit_changing_match_data) | |
416 | ? &search_regs : NULL), | |
487282dc | 417 | (!NILP (current_buffer->case_fold_search) |
0190922f | 418 | ? current_buffer->case_canon_table : Qnil), |
0c8533c6 RS |
419 | posix, |
420 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 421 | immediate_quit = 1; |
8bb43c28 | 422 | re_match_object = string; |
177c0ea7 | 423 | |
d5db4077 KR |
424 | val = re_search (bufp, (char *) SDATA (string), |
425 | SBYTES (string), pos_byte, | |
426 | SBYTES (string) - pos_byte, | |
ef887810 RS |
427 | (NILP (Vinhibit_changing_match_data) |
428 | ? &search_regs : NULL)); | |
ca1d1d23 | 429 | immediate_quit = 0; |
ef887810 RS |
430 | |
431 | /* Set last_thing_searched only when match data is changed. */ | |
432 | if (NILP (Vinhibit_changing_match_data)) | |
433 | last_thing_searched = Qt; | |
434 | ||
ca1d1d23 JB |
435 | if (val == -2) |
436 | matcher_overflow (); | |
437 | if (val < 0) return Qnil; | |
0c8533c6 | 438 | |
ef887810 RS |
439 | if (NILP (Vinhibit_changing_match_data)) |
440 | for (i = 0; i < search_regs.num_regs; i++) | |
441 | if (search_regs.start[i] >= 0) | |
442 | { | |
443 | search_regs.start[i] | |
444 | = string_byte_to_char (string, search_regs.start[i]); | |
445 | search_regs.end[i] | |
446 | = string_byte_to_char (string, search_regs.end[i]); | |
447 | } | |
0c8533c6 RS |
448 | |
449 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 450 | } |
e59a8453 | 451 | |
b819a390 | 452 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 | 453 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
b85acc4b | 454 | Matching ignores case if `case-fold-search' is non-nil. |
8c1a1077 PJ |
455 | If third arg START is non-nil, start search at that index in STRING. |
456 | For index of first char beyond the match, do (match-end 0). | |
457 | `match-end' and `match-beginning' also give indices of substrings | |
2bd2f32d RS |
458 | matched by parenthesis constructs in the pattern. |
459 | ||
460 | You can use the function `match-string' to extract the substrings | |
461 | matched by the parenthesis constructions in REGEXP. */) | |
5842a27b | 462 | (Lisp_Object regexp, Lisp_Object string, Lisp_Object start) |
b819a390 RS |
463 | { |
464 | return string_match_1 (regexp, string, start, 0); | |
465 | } | |
466 | ||
467 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
468 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
469 | Find the longest match, in accord with Posix regular expression rules. | |
470 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
471 | If third arg START is non-nil, start search at that index in STRING. | |
472 | For index of first char beyond the match, do (match-end 0). | |
473 | `match-end' and `match-beginning' also give indices of substrings | |
474 | matched by parenthesis constructs in the pattern. */) | |
5842a27b | 475 | (Lisp_Object regexp, Lisp_Object string, Lisp_Object start) |
b819a390 RS |
476 | { |
477 | return string_match_1 (regexp, string, start, 1); | |
478 | } | |
479 | ||
e59a8453 RS |
480 | /* Match REGEXP against STRING, searching all of STRING, |
481 | and return the index of the match, or negative on failure. | |
482 | This does not clobber the match data. */ | |
483 | ||
484 | int | |
971de7fb | 485 | fast_string_match (Lisp_Object regexp, Lisp_Object string) |
e59a8453 RS |
486 | { |
487 | int val; | |
487282dc | 488 | struct re_pattern_buffer *bufp; |
e59a8453 | 489 | |
facdc750 RS |
490 | bufp = compile_pattern (regexp, 0, Qnil, |
491 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 492 | immediate_quit = 1; |
8bb43c28 | 493 | re_match_object = string; |
177c0ea7 | 494 | |
d5db4077 KR |
495 | val = re_search (bufp, (char *) SDATA (string), |
496 | SBYTES (string), 0, | |
497 | SBYTES (string), 0); | |
e59a8453 RS |
498 | immediate_quit = 0; |
499 | return val; | |
500 | } | |
5679531d KH |
501 | |
502 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
503 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
504 | This does not clobber the match data. |
505 | We assume that STRING contains single-byte characters. */ | |
5679531d | 506 | |
5679531d | 507 | int |
971de7fb | 508 | fast_c_string_match_ignore_case (Lisp_Object regexp, const char *string) |
5679531d KH |
509 | { |
510 | int val; | |
511 | struct re_pattern_buffer *bufp; | |
512 | int len = strlen (string); | |
513 | ||
0c8533c6 | 514 | regexp = string_make_unibyte (regexp); |
b4577c63 | 515 | re_match_object = Qt; |
5679531d | 516 | bufp = compile_pattern (regexp, 0, |
0190922f | 517 | Vascii_canon_table, 0, |
f8bd51c4 | 518 | 0); |
5679531d KH |
519 | immediate_quit = 1; |
520 | val = re_search (bufp, string, len, 0, len, 0); | |
521 | immediate_quit = 0; | |
522 | return val; | |
523 | } | |
be5f4dfb KH |
524 | |
525 | /* Like fast_string_match but ignore case. */ | |
526 | ||
527 | int | |
971de7fb | 528 | fast_string_match_ignore_case (Lisp_Object regexp, Lisp_Object string) |
be5f4dfb KH |
529 | { |
530 | int val; | |
531 | struct re_pattern_buffer *bufp; | |
532 | ||
0190922f | 533 | bufp = compile_pattern (regexp, 0, Vascii_canon_table, |
be5f4dfb KH |
534 | 0, STRING_MULTIBYTE (string)); |
535 | immediate_quit = 1; | |
536 | re_match_object = string; | |
537 | ||
538 | val = re_search (bufp, (char *) SDATA (string), | |
539 | SBYTES (string), 0, | |
540 | SBYTES (string), 0); | |
541 | immediate_quit = 0; | |
542 | return val; | |
543 | } | |
ca1d1d23 | 544 | \f |
a80ce213 | 545 | /* Match REGEXP against the characters after POS to LIMIT, and return |
8b13507a KH |
546 | the number of matched characters. If STRING is non-nil, match |
547 | against the characters in it. In that case, POS and LIMIT are | |
548 | indices into the string. This function doesn't modify the match | |
549 | data. */ | |
550 | ||
551 | EMACS_INT | |
971de7fb | 552 | fast_looking_at (Lisp_Object regexp, EMACS_INT pos, EMACS_INT pos_byte, EMACS_INT limit, EMACS_INT limit_byte, Lisp_Object string) |
8b13507a KH |
553 | { |
554 | int multibyte; | |
555 | struct re_pattern_buffer *buf; | |
556 | unsigned char *p1, *p2; | |
e7deaab0 | 557 | EMACS_INT s1, s2; |
8b13507a | 558 | EMACS_INT len; |
78edd3b7 | 559 | |
8b13507a KH |
560 | if (STRINGP (string)) |
561 | { | |
562 | if (pos_byte < 0) | |
563 | pos_byte = string_char_to_byte (string, pos); | |
564 | if (limit_byte < 0) | |
565 | limit_byte = string_char_to_byte (string, limit); | |
566 | p1 = NULL; | |
567 | s1 = 0; | |
568 | p2 = SDATA (string); | |
569 | s2 = SBYTES (string); | |
570 | re_match_object = string; | |
571 | multibyte = STRING_MULTIBYTE (string); | |
572 | } | |
573 | else | |
574 | { | |
575 | if (pos_byte < 0) | |
576 | pos_byte = CHAR_TO_BYTE (pos); | |
577 | if (limit_byte < 0) | |
578 | limit_byte = CHAR_TO_BYTE (limit); | |
579 | pos_byte -= BEGV_BYTE; | |
580 | limit_byte -= BEGV_BYTE; | |
581 | p1 = BEGV_ADDR; | |
582 | s1 = GPT_BYTE - BEGV_BYTE; | |
583 | p2 = GAP_END_ADDR; | |
584 | s2 = ZV_BYTE - GPT_BYTE; | |
585 | if (s1 < 0) | |
586 | { | |
587 | p2 = p1; | |
588 | s2 = ZV_BYTE - BEGV_BYTE; | |
589 | s1 = 0; | |
590 | } | |
591 | if (s2 < 0) | |
592 | { | |
593 | s1 = ZV_BYTE - BEGV_BYTE; | |
594 | s2 = 0; | |
595 | } | |
596 | re_match_object = Qnil; | |
597 | multibyte = ! NILP (current_buffer->enable_multibyte_characters); | |
598 | } | |
599 | ||
600 | buf = compile_pattern (regexp, 0, Qnil, 0, multibyte); | |
601 | immediate_quit = 1; | |
602 | len = re_match_2 (buf, (char *) p1, s1, (char *) p2, s2, | |
603 | pos_byte, NULL, limit_byte); | |
604 | immediate_quit = 0; | |
605 | ||
606 | return len; | |
607 | } | |
608 | ||
609 | \f | |
9169c321 JB |
610 | /* The newline cache: remembering which sections of text have no newlines. */ |
611 | ||
612 | /* If the user has requested newline caching, make sure it's on. | |
613 | Otherwise, make sure it's off. | |
614 | This is our cheezy way of associating an action with the change of | |
615 | state of a buffer-local variable. */ | |
616 | static void | |
971de7fb | 617 | newline_cache_on_off (struct buffer *buf) |
9169c321 JB |
618 | { |
619 | if (NILP (buf->cache_long_line_scans)) | |
620 | { | |
621 | /* It should be off. */ | |
622 | if (buf->newline_cache) | |
623 | { | |
624 | free_region_cache (buf->newline_cache); | |
625 | buf->newline_cache = 0; | |
626 | } | |
627 | } | |
628 | else | |
629 | { | |
630 | /* It should be on. */ | |
631 | if (buf->newline_cache == 0) | |
632 | buf->newline_cache = new_region_cache (); | |
633 | } | |
634 | } | |
635 | ||
636 | \f | |
637 | /* Search for COUNT instances of the character TARGET between START and END. | |
638 | ||
639 | If COUNT is positive, search forwards; END must be >= START. | |
640 | If COUNT is negative, search backwards for the -COUNTth instance; | |
641 | END must be <= START. | |
642 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
643 | ||
644 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
645 | direction indicated by COUNT. | |
ffd56f97 JB |
646 | |
647 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
a9f2a45f | 648 | position past the COUNTth match. Note that for reverse motion |
5bfe95c9 | 649 | this is not the same as the usual convention for Emacs motion commands. |
ffd56f97 | 650 | |
9169c321 JB |
651 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
652 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 653 | |
087a5f81 RS |
654 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
655 | except when inside redisplay. */ | |
656 | ||
dfcf069d | 657 | int |
971de7fb | 658 | scan_buffer (register int target, EMACS_INT start, EMACS_INT end, int count, int *shortage, int allow_quit) |
ca1d1d23 | 659 | { |
9169c321 | 660 | struct region_cache *newline_cache; |
177c0ea7 | 661 | int direction; |
ffd56f97 | 662 | |
9169c321 JB |
663 | if (count > 0) |
664 | { | |
665 | direction = 1; | |
666 | if (! end) end = ZV; | |
667 | } | |
668 | else | |
669 | { | |
670 | direction = -1; | |
671 | if (! end) end = BEGV; | |
672 | } | |
ffd56f97 | 673 | |
9169c321 JB |
674 | newline_cache_on_off (current_buffer); |
675 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
676 | |
677 | if (shortage != 0) | |
678 | *shortage = 0; | |
679 | ||
087a5f81 | 680 | immediate_quit = allow_quit; |
ca1d1d23 | 681 | |
ffd56f97 | 682 | if (count > 0) |
9169c321 | 683 | while (start != end) |
ca1d1d23 | 684 | { |
9169c321 JB |
685 | /* Our innermost scanning loop is very simple; it doesn't know |
686 | about gaps, buffer ends, or the newline cache. ceiling is | |
687 | the position of the last character before the next such | |
688 | obstacle --- the last character the dumb search loop should | |
689 | examine. */ | |
e7deaab0 AS |
690 | EMACS_INT ceiling_byte = CHAR_TO_BYTE (end) - 1; |
691 | EMACS_INT start_byte = CHAR_TO_BYTE (start); | |
692 | EMACS_INT tem; | |
9169c321 JB |
693 | |
694 | /* If we're looking for a newline, consult the newline cache | |
695 | to see where we can avoid some scanning. */ | |
696 | if (target == '\n' && newline_cache) | |
697 | { | |
698 | int next_change; | |
699 | immediate_quit = 0; | |
700 | while (region_cache_forward | |
fa8ed3e0 RS |
701 | (current_buffer, newline_cache, start_byte, &next_change)) |
702 | start_byte = next_change; | |
cbe0db0d | 703 | immediate_quit = allow_quit; |
9169c321 | 704 | |
fa8ed3e0 RS |
705 | /* START should never be after END. */ |
706 | if (start_byte > ceiling_byte) | |
707 | start_byte = ceiling_byte; | |
9169c321 JB |
708 | |
709 | /* Now the text after start is an unknown region, and | |
710 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 711 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
712 | } |
713 | ||
714 | /* The dumb loop can only scan text stored in contiguous | |
715 | bytes. BUFFER_CEILING_OF returns the last character | |
716 | position that is contiguous, so the ceiling is the | |
717 | position after that. */ | |
67ce527d KH |
718 | tem = BUFFER_CEILING_OF (start_byte); |
719 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
720 | |
721 | { | |
177c0ea7 | 722 | /* The termination address of the dumb loop. */ |
fa8ed3e0 RS |
723 | register unsigned char *ceiling_addr |
724 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
725 | register unsigned char *cursor | |
726 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
727 | unsigned char *base = cursor; |
728 | ||
729 | while (cursor < ceiling_addr) | |
730 | { | |
731 | unsigned char *scan_start = cursor; | |
732 | ||
733 | /* The dumb loop. */ | |
734 | while (*cursor != target && ++cursor < ceiling_addr) | |
735 | ; | |
736 | ||
737 | /* If we're looking for newlines, cache the fact that | |
738 | the region from start to cursor is free of them. */ | |
739 | if (target == '\n' && newline_cache) | |
740 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
741 | start_byte + scan_start - base, |
742 | start_byte + cursor - base); | |
9169c321 JB |
743 | |
744 | /* Did we find the target character? */ | |
745 | if (cursor < ceiling_addr) | |
746 | { | |
747 | if (--count == 0) | |
748 | { | |
749 | immediate_quit = 0; | |
fa8ed3e0 | 750 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
751 | } |
752 | cursor++; | |
753 | } | |
754 | } | |
755 | ||
fa8ed3e0 | 756 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 757 | } |
ca1d1d23 JB |
758 | } |
759 | else | |
9169c321 JB |
760 | while (start > end) |
761 | { | |
762 | /* The last character to check before the next obstacle. */ | |
e7deaab0 AS |
763 | EMACS_INT ceiling_byte = CHAR_TO_BYTE (end); |
764 | EMACS_INT start_byte = CHAR_TO_BYTE (start); | |
765 | EMACS_INT tem; | |
9169c321 JB |
766 | |
767 | /* Consult the newline cache, if appropriate. */ | |
768 | if (target == '\n' && newline_cache) | |
769 | { | |
770 | int next_change; | |
771 | immediate_quit = 0; | |
772 | while (region_cache_backward | |
fa8ed3e0 RS |
773 | (current_buffer, newline_cache, start_byte, &next_change)) |
774 | start_byte = next_change; | |
cbe0db0d | 775 | immediate_quit = allow_quit; |
9169c321 JB |
776 | |
777 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
778 | if (start_byte <= ceiling_byte) |
779 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
780 | |
781 | /* Now the text before start is an unknown region, and | |
782 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 783 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
784 | } |
785 | ||
786 | /* Stop scanning before the gap. */ | |
67ce527d KH |
787 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
788 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
789 | |
790 | { | |
791 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
792 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
793 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
794 | unsigned char *base = cursor; |
795 | ||
796 | while (cursor >= ceiling_addr) | |
797 | { | |
798 | unsigned char *scan_start = cursor; | |
799 | ||
800 | while (*cursor != target && --cursor >= ceiling_addr) | |
801 | ; | |
802 | ||
803 | /* If we're looking for newlines, cache the fact that | |
804 | the region from after the cursor to start is free of them. */ | |
805 | if (target == '\n' && newline_cache) | |
806 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
807 | start_byte + cursor - base, |
808 | start_byte + scan_start - base); | |
9169c321 JB |
809 | |
810 | /* Did we find the target character? */ | |
811 | if (cursor >= ceiling_addr) | |
812 | { | |
813 | if (++count >= 0) | |
814 | { | |
815 | immediate_quit = 0; | |
fa8ed3e0 | 816 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
817 | } |
818 | cursor--; | |
819 | } | |
820 | } | |
821 | ||
fa8ed3e0 | 822 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
823 | } |
824 | } | |
825 | ||
ca1d1d23 JB |
826 | immediate_quit = 0; |
827 | if (shortage != 0) | |
ffd56f97 | 828 | *shortage = count * direction; |
9169c321 | 829 | return start; |
ca1d1d23 | 830 | } |
fa8ed3e0 RS |
831 | \f |
832 | /* Search for COUNT instances of a line boundary, which means either a | |
833 | newline or (if selective display enabled) a carriage return. | |
834 | Start at START. If COUNT is negative, search backwards. | |
835 | ||
836 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
837 | ||
838 | If we find COUNT instances. we position after (always after, | |
839 | even if scanning backwards) the COUNTth match, and return 0. | |
840 | ||
841 | If we don't find COUNT instances before reaching the end of the | |
842 | buffer (or the beginning, if scanning backwards), we return | |
843 | the number of line boundaries left unfound, and position at | |
844 | the limit we bumped up against. | |
845 | ||
846 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 847 | except in special cases. */ |
ca1d1d23 | 848 | |
63fa018d | 849 | int |
971de7fb | 850 | scan_newline (EMACS_INT start, EMACS_INT start_byte, EMACS_INT limit, EMACS_INT limit_byte, register int count, int allow_quit) |
63fa018d | 851 | { |
fa8ed3e0 RS |
852 | int direction = ((count > 0) ? 1 : -1); |
853 | ||
854 | register unsigned char *cursor; | |
855 | unsigned char *base; | |
856 | ||
e7deaab0 | 857 | EMACS_INT ceiling; |
fa8ed3e0 RS |
858 | register unsigned char *ceiling_addr; |
859 | ||
d5d57b92 RS |
860 | int old_immediate_quit = immediate_quit; |
861 | ||
fa8ed3e0 RS |
862 | /* The code that follows is like scan_buffer |
863 | but checks for either newline or carriage return. */ | |
864 | ||
d5d57b92 RS |
865 | if (allow_quit) |
866 | immediate_quit++; | |
fa8ed3e0 RS |
867 | |
868 | start_byte = CHAR_TO_BYTE (start); | |
869 | ||
870 | if (count > 0) | |
871 | { | |
872 | while (start_byte < limit_byte) | |
873 | { | |
874 | ceiling = BUFFER_CEILING_OF (start_byte); | |
875 | ceiling = min (limit_byte - 1, ceiling); | |
876 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
877 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
878 | while (1) | |
879 | { | |
880 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
881 | ; | |
882 | ||
883 | if (cursor != ceiling_addr) | |
884 | { | |
885 | if (--count == 0) | |
886 | { | |
d5d57b92 | 887 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
888 | start_byte = start_byte + cursor - base + 1; |
889 | start = BYTE_TO_CHAR (start_byte); | |
890 | TEMP_SET_PT_BOTH (start, start_byte); | |
891 | return 0; | |
892 | } | |
893 | else | |
894 | if (++cursor == ceiling_addr) | |
895 | break; | |
896 | } | |
897 | else | |
898 | break; | |
899 | } | |
900 | start_byte += cursor - base; | |
901 | } | |
902 | } | |
903 | else | |
904 | { | |
fa8ed3e0 RS |
905 | while (start_byte > limit_byte) |
906 | { | |
907 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
908 | ceiling = max (limit_byte, ceiling); | |
909 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
910 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
911 | while (1) | |
912 | { | |
913 | while (--cursor != ceiling_addr && *cursor != '\n') | |
914 | ; | |
915 | ||
916 | if (cursor != ceiling_addr) | |
917 | { | |
918 | if (++count == 0) | |
919 | { | |
d5d57b92 | 920 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
921 | /* Return the position AFTER the match we found. */ |
922 | start_byte = start_byte + cursor - base + 1; | |
923 | start = BYTE_TO_CHAR (start_byte); | |
924 | TEMP_SET_PT_BOTH (start, start_byte); | |
925 | return 0; | |
926 | } | |
927 | } | |
928 | else | |
929 | break; | |
930 | } | |
931 | /* Here we add 1 to compensate for the last decrement | |
932 | of CURSOR, which took it past the valid range. */ | |
933 | start_byte += cursor - base + 1; | |
934 | } | |
935 | } | |
936 | ||
937 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 938 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
939 | |
940 | return count * direction; | |
63fa018d RS |
941 | } |
942 | ||
ca1d1d23 | 943 | int |
971de7fb | 944 | find_next_newline_no_quit (EMACS_INT from, int cnt) |
ca1d1d23 | 945 | { |
fa8ed3e0 | 946 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
9169c321 JB |
947 | } |
948 | ||
9169c321 JB |
949 | /* Like find_next_newline, but returns position before the newline, |
950 | not after, and only search up to TO. This isn't just | |
951 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 952 | |
9169c321 | 953 | int |
971de7fb | 954 | find_before_next_newline (EMACS_INT from, EMACS_INT to, int cnt) |
9169c321 JB |
955 | { |
956 | int shortage; | |
957 | int pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
958 | ||
959 | if (shortage == 0) | |
960 | pos--; | |
177c0ea7 | 961 | |
9169c321 | 962 | return pos; |
ca1d1d23 JB |
963 | } |
964 | \f | |
ca1d1d23 JB |
965 | /* Subroutines of Lisp buffer search functions. */ |
966 | ||
967 | static Lisp_Object | |
971de7fb | 968 | search_command (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count, int direction, int RE, int posix) |
ca1d1d23 JB |
969 | { |
970 | register int np; | |
9f43ad85 | 971 | int lim, lim_byte; |
ca1d1d23 JB |
972 | int n = direction; |
973 | ||
974 | if (!NILP (count)) | |
975 | { | |
b7826503 | 976 | CHECK_NUMBER (count); |
ca1d1d23 JB |
977 | n *= XINT (count); |
978 | } | |
979 | ||
b7826503 | 980 | CHECK_STRING (string); |
ca1d1d23 | 981 | if (NILP (bound)) |
9f43ad85 RS |
982 | { |
983 | if (n > 0) | |
984 | lim = ZV, lim_byte = ZV_BYTE; | |
985 | else | |
986 | lim = BEGV, lim_byte = BEGV_BYTE; | |
987 | } | |
ca1d1d23 JB |
988 | else |
989 | { | |
b7826503 | 990 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 991 | lim = XINT (bound); |
6ec8bbd2 | 992 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
993 | error ("Invalid search bound (wrong side of point)"); |
994 | if (lim > ZV) | |
9f43ad85 | 995 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 996 | else if (lim < BEGV) |
9f43ad85 | 997 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
998 | else |
999 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
1000 | } |
1001 | ||
910c747a RS |
1002 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
1003 | XCHAR_TABLE (current_buffer->case_canon_table)->extras[2] | |
1004 | = current_buffer->case_eqv_table; | |
1005 | ||
9f43ad85 | 1006 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
ca1d1d23 | 1007 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 1008 | ? current_buffer->case_canon_table |
3135e9fd | 1009 | : Qnil), |
ca1d1d23 | 1010 | (!NILP (current_buffer->case_fold_search) |
facdc750 | 1011 | ? current_buffer->case_eqv_table |
3135e9fd | 1012 | : Qnil), |
b819a390 | 1013 | posix); |
ca1d1d23 JB |
1014 | if (np <= 0) |
1015 | { | |
1016 | if (NILP (noerror)) | |
06f77a8a KS |
1017 | xsignal1 (Qsearch_failed, string); |
1018 | ||
ca1d1d23 JB |
1019 | if (!EQ (noerror, Qt)) |
1020 | { | |
1021 | if (lim < BEGV || lim > ZV) | |
1022 | abort (); | |
9f43ad85 | 1023 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
1024 | return Qnil; |
1025 | #if 0 /* This would be clean, but maybe programs depend on | |
1026 | a value of nil here. */ | |
481399bf | 1027 | np = lim; |
a5f217b8 | 1028 | #endif |
ca1d1d23 | 1029 | } |
481399bf RS |
1030 | else |
1031 | return Qnil; | |
ca1d1d23 JB |
1032 | } |
1033 | ||
1034 | if (np < BEGV || np > ZV) | |
1035 | abort (); | |
1036 | ||
1037 | SET_PT (np); | |
1038 | ||
1039 | return make_number (np); | |
1040 | } | |
1041 | \f | |
fa8ed3e0 RS |
1042 | /* Return 1 if REGEXP it matches just one constant string. */ |
1043 | ||
b6d6a51c | 1044 | static int |
971de7fb | 1045 | trivial_regexp_p (Lisp_Object regexp) |
b6d6a51c | 1046 | { |
d5db4077 KR |
1047 | int len = SBYTES (regexp); |
1048 | unsigned char *s = SDATA (regexp); | |
b6d6a51c KH |
1049 | while (--len >= 0) |
1050 | { | |
1051 | switch (*s++) | |
1052 | { | |
1053 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
1054 | return 0; | |
1055 | case '\\': | |
1056 | if (--len < 0) | |
1057 | return 0; | |
1058 | switch (*s++) | |
1059 | { | |
1060 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
1061 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 1062 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 1063 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 1064 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
1065 | case '6': case '7': case '8': case '9': |
1066 | return 0; | |
1067 | } | |
1068 | } | |
1069 | } | |
1070 | return 1; | |
1071 | } | |
1072 | ||
ca325161 | 1073 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 1074 | starting at position POS and stopping at position LIM, |
b819a390 | 1075 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
1076 | a regular expression if RE is true. |
1077 | ||
1078 | If N is positive, searching is forward and LIM must be greater than POS. | |
1079 | If N is negative, searching is backward and LIM must be less than POS. | |
1080 | ||
facdc750 | 1081 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 1082 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
1083 | (if searching backward) or the end (if searching forward). |
1084 | ||
1085 | POSIX is nonzero if we want full backtracking (POSIX style) | |
1086 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 1087 | |
aff2ce94 RS |
1088 | #define TRANSLATE(out, trt, d) \ |
1089 | do \ | |
1090 | { \ | |
1091 | if (! NILP (trt)) \ | |
1092 | { \ | |
1093 | Lisp_Object temp; \ | |
1094 | temp = Faref (trt, make_number (d)); \ | |
1095 | if (INTEGERP (temp)) \ | |
1096 | out = XINT (temp); \ | |
1097 | else \ | |
1098 | out = d; \ | |
1099 | } \ | |
1100 | else \ | |
1101 | out = d; \ | |
1102 | } \ | |
1103 | while (0) | |
facdc750 | 1104 | |
ef887810 RS |
1105 | /* Only used in search_buffer, to record the end position of the match |
1106 | when searching regexps and SEARCH_REGS should not be changed | |
1107 | (i.e. Vinhibit_changing_match_data is non-nil). */ | |
1108 | static struct re_registers search_regs_1; | |
1109 | ||
e7deaab0 | 1110 | static EMACS_INT |
d5a3eaaf AS |
1111 | search_buffer (Lisp_Object string, EMACS_INT pos, EMACS_INT pos_byte, |
1112 | EMACS_INT lim, EMACS_INT lim_byte, int n, | |
1113 | int RE, Lisp_Object trt, Lisp_Object inverse_trt, int posix) | |
ca1d1d23 | 1114 | { |
d5db4077 KR |
1115 | int len = SCHARS (string); |
1116 | int len_byte = SBYTES (string); | |
facdc750 | 1117 | register int i; |
ca1d1d23 | 1118 | |
7074fde6 FP |
1119 | if (running_asynch_code) |
1120 | save_search_regs (); | |
1121 | ||
a7e4cdde | 1122 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1123 | /* Null string is found at starting position. */ |
a7e4cdde | 1124 | if (len == 0 || n == 0) |
ca325161 | 1125 | { |
0353b28f | 1126 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1127 | return pos; |
1128 | } | |
3f57a499 | 1129 | |
41a33295 | 1130 | if (RE && !(trivial_regexp_p (string) && NILP (Vsearch_spaces_regexp))) |
ca1d1d23 | 1131 | { |
facdc750 RS |
1132 | unsigned char *p1, *p2; |
1133 | int s1, s2; | |
487282dc KH |
1134 | struct re_pattern_buffer *bufp; |
1135 | ||
ef887810 RS |
1136 | bufp = compile_pattern (string, |
1137 | (NILP (Vinhibit_changing_match_data) | |
1138 | ? &search_regs : &search_regs_1), | |
1139 | trt, posix, | |
0c8533c6 | 1140 | !NILP (current_buffer->enable_multibyte_characters)); |
ca1d1d23 | 1141 | |
ca1d1d23 JB |
1142 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1143 | because letting this function finish | |
1144 | can take too long. */ | |
1145 | QUIT; /* Do a pending quit right away, | |
1146 | to avoid paradoxical behavior */ | |
1147 | /* Get pointers and sizes of the two strings | |
1148 | that make up the visible portion of the buffer. */ | |
1149 | ||
1150 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1151 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1152 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1153 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1154 | if (s1 < 0) |
1155 | { | |
1156 | p2 = p1; | |
fa8ed3e0 | 1157 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1158 | s1 = 0; |
1159 | } | |
1160 | if (s2 < 0) | |
1161 | { | |
fa8ed3e0 | 1162 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1163 | s2 = 0; |
1164 | } | |
8bb43c28 | 1165 | re_match_object = Qnil; |
177c0ea7 | 1166 | |
ca1d1d23 JB |
1167 | while (n < 0) |
1168 | { | |
42db823b | 1169 | int val; |
487282dc | 1170 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1171 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1172 | (NILP (Vinhibit_changing_match_data) |
1173 | ? &search_regs : &search_regs_1), | |
42db823b | 1174 | /* Don't allow match past current point */ |
4996330b | 1175 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1176 | if (val == -2) |
b6d6a51c KH |
1177 | { |
1178 | matcher_overflow (); | |
1179 | } | |
ca1d1d23 JB |
1180 | if (val >= 0) |
1181 | { | |
ef887810 RS |
1182 | if (NILP (Vinhibit_changing_match_data)) |
1183 | { | |
1184 | pos_byte = search_regs.start[0] + BEGV_BYTE; | |
1185 | for (i = 0; i < search_regs.num_regs; i++) | |
1186 | if (search_regs.start[i] >= 0) | |
1187 | { | |
1188 | search_regs.start[i] | |
1189 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1190 | search_regs.end[i] | |
1191 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1192 | } | |
1193 | XSETBUFFER (last_thing_searched, current_buffer); | |
1194 | /* Set pos to the new position. */ | |
1195 | pos = search_regs.start[0]; | |
1196 | } | |
1197 | else | |
1198 | { | |
1199 | pos_byte = search_regs_1.start[0] + BEGV_BYTE; | |
1200 | /* Set pos to the new position. */ | |
1201 | pos = BYTE_TO_CHAR (search_regs_1.start[0] + BEGV_BYTE); | |
1202 | } | |
ca1d1d23 JB |
1203 | } |
1204 | else | |
1205 | { | |
1206 | immediate_quit = 0; | |
1207 | return (n); | |
1208 | } | |
1209 | n++; | |
1210 | } | |
1211 | while (n > 0) | |
1212 | { | |
42db823b | 1213 | int val; |
487282dc | 1214 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1215 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1216 | (NILP (Vinhibit_changing_match_data) |
1217 | ? &search_regs : &search_regs_1), | |
4996330b | 1218 | lim_byte - BEGV_BYTE); |
ca1d1d23 | 1219 | if (val == -2) |
b6d6a51c KH |
1220 | { |
1221 | matcher_overflow (); | |
1222 | } | |
ca1d1d23 JB |
1223 | if (val >= 0) |
1224 | { | |
ef887810 RS |
1225 | if (NILP (Vinhibit_changing_match_data)) |
1226 | { | |
1227 | pos_byte = search_regs.end[0] + BEGV_BYTE; | |
1228 | for (i = 0; i < search_regs.num_regs; i++) | |
1229 | if (search_regs.start[i] >= 0) | |
1230 | { | |
1231 | search_regs.start[i] | |
1232 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1233 | search_regs.end[i] | |
1234 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1235 | } | |
1236 | XSETBUFFER (last_thing_searched, current_buffer); | |
1237 | pos = search_regs.end[0]; | |
1238 | } | |
1239 | else | |
1240 | { | |
1241 | pos_byte = search_regs_1.end[0] + BEGV_BYTE; | |
1242 | pos = BYTE_TO_CHAR (search_regs_1.end[0] + BEGV_BYTE); | |
1243 | } | |
ca1d1d23 JB |
1244 | } |
1245 | else | |
1246 | { | |
1247 | immediate_quit = 0; | |
1248 | return (0 - n); | |
1249 | } | |
1250 | n--; | |
1251 | } | |
1252 | immediate_quit = 0; | |
1253 | return (pos); | |
1254 | } | |
1255 | else /* non-RE case */ | |
1256 | { | |
facdc750 RS |
1257 | unsigned char *raw_pattern, *pat; |
1258 | int raw_pattern_size; | |
1259 | int raw_pattern_size_byte; | |
1260 | unsigned char *patbuf; | |
1261 | int multibyte = !NILP (current_buffer->enable_multibyte_characters); | |
a967ed62 | 1262 | unsigned char *base_pat; |
49e44e81 KH |
1263 | /* Set to positive if we find a non-ASCII char that need |
1264 | translation. Otherwise set to zero later. */ | |
1265 | int char_base = -1; | |
040272ce | 1266 | int boyer_moore_ok = 1; |
facdc750 RS |
1267 | |
1268 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1269 | We must convert PATTERN to match that, or we will not really | |
1270 | find things right. */ | |
1271 | ||
1272 | if (multibyte == STRING_MULTIBYTE (string)) | |
1273 | { | |
d5db4077 KR |
1274 | raw_pattern = (unsigned char *) SDATA (string); |
1275 | raw_pattern_size = SCHARS (string); | |
1276 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1277 | } |
1278 | else if (multibyte) | |
1279 | { | |
d5db4077 | 1280 | raw_pattern_size = SCHARS (string); |
facdc750 | 1281 | raw_pattern_size_byte |
d5db4077 | 1282 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1283 | raw_pattern_size); |
7276d3d8 | 1284 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1285 | copy_text (SDATA (string), raw_pattern, |
1286 | SCHARS (string), 0, 1); | |
facdc750 RS |
1287 | } |
1288 | else | |
1289 | { | |
1290 | /* Converting multibyte to single-byte. | |
1291 | ||
1292 | ??? Perhaps this conversion should be done in a special way | |
1293 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1294 | so that only the multibyte chars which really correspond to | |
1295 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1296 | raw_pattern_size = SCHARS (string); |
1297 | raw_pattern_size_byte = SCHARS (string); | |
7276d3d8 | 1298 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
1299 | copy_text (SDATA (string), raw_pattern, |
1300 | SBYTES (string), 1, 0); | |
facdc750 RS |
1301 | } |
1302 | ||
1303 | /* Copy and optionally translate the pattern. */ | |
1304 | len = raw_pattern_size; | |
1305 | len_byte = raw_pattern_size_byte; | |
f5f7578b | 1306 | patbuf = (unsigned char *) alloca (len * MAX_MULTIBYTE_LENGTH); |
facdc750 RS |
1307 | pat = patbuf; |
1308 | base_pat = raw_pattern; | |
1309 | if (multibyte) | |
1310 | { | |
a17ae96a KH |
1311 | /* Fill patbuf by translated characters in STRING while |
1312 | checking if we can use boyer-moore search. If TRT is | |
1313 | non-nil, we can use boyer-moore search only if TRT can be | |
1314 | represented by the byte array of 256 elements. For that, | |
1315 | all non-ASCII case-equivalents of all case-senstive | |
1316 | characters in STRING must belong to the same charset and | |
1317 | row. */ | |
1318 | ||
facdc750 RS |
1319 | while (--len >= 0) |
1320 | { | |
a17ae96a | 1321 | unsigned char str_base[MAX_MULTIBYTE_LENGTH], *str; |
aff2ce94 | 1322 | int c, translated, inverse; |
a17ae96a | 1323 | int in_charlen, charlen; |
facdc750 RS |
1324 | |
1325 | /* If we got here and the RE flag is set, it's because we're | |
1326 | dealing with a regexp known to be trivial, so the backslash | |
1327 | just quotes the next character. */ | |
1328 | if (RE && *base_pat == '\\') | |
1329 | { | |
1330 | len--; | |
62221f22 | 1331 | raw_pattern_size--; |
facdc750 RS |
1332 | len_byte--; |
1333 | base_pat++; | |
1334 | } | |
1335 | ||
62a6e103 | 1336 | c = STRING_CHAR_AND_LENGTH (base_pat, in_charlen); |
040272ce | 1337 | |
a17ae96a | 1338 | if (NILP (trt)) |
facdc750 | 1339 | { |
a17ae96a KH |
1340 | str = base_pat; |
1341 | charlen = in_charlen; | |
1342 | } | |
1343 | else | |
1344 | { | |
1345 | /* Translate the character. */ | |
1346 | TRANSLATE (translated, trt, c); | |
1347 | charlen = CHAR_STRING (translated, str_base); | |
1348 | str = str_base; | |
1349 | ||
1350 | /* Check if C has any other case-equivalents. */ | |
1351 | TRANSLATE (inverse, inverse_trt, c); | |
1352 | /* If so, check if we can use boyer-moore. */ | |
1353 | if (c != inverse && boyer_moore_ok) | |
1354 | { | |
1355 | /* Check if all equivalents belong to the same | |
1356 | group of characters. Note that the check of C | |
49e44e81 KH |
1357 | itself is done by the last iteration. */ |
1358 | int this_char_base = -1; | |
1359 | ||
1360 | while (boyer_moore_ok) | |
a17ae96a | 1361 | { |
49e44e81 | 1362 | if (ASCII_BYTE_P (inverse)) |
a17ae96a | 1363 | { |
49e44e81 KH |
1364 | if (this_char_base > 0) |
1365 | boyer_moore_ok = 0; | |
1366 | else | |
dafaabd1 | 1367 | this_char_base = 0; |
a17ae96a | 1368 | } |
49e44e81 KH |
1369 | else if (CHAR_BYTE8_P (inverse)) |
1370 | /* Boyer-moore search can't handle a | |
1371 | translation of an eight-bit | |
1372 | character. */ | |
1373 | boyer_moore_ok = 0; | |
1374 | else if (this_char_base < 0) | |
1375 | { | |
1376 | this_char_base = inverse & ~0x3F; | |
1377 | if (char_base < 0) | |
1378 | char_base = this_char_base; | |
dafaabd1 | 1379 | else if (this_char_base != char_base) |
49e44e81 KH |
1380 | boyer_moore_ok = 0; |
1381 | } | |
1382 | else if ((inverse & ~0x3F) != this_char_base) | |
1383 | boyer_moore_ok = 0; | |
a17ae96a KH |
1384 | if (c == inverse) |
1385 | break; | |
1386 | TRANSLATE (inverse, inverse_trt, inverse); | |
1387 | } | |
1388 | } | |
aff2ce94 | 1389 | } |
facdc750 RS |
1390 | |
1391 | /* Store this character into the translated pattern. */ | |
72af86bd | 1392 | memcpy (pat, str, charlen); |
a17ae96a | 1393 | pat += charlen; |
facdc750 RS |
1394 | base_pat += in_charlen; |
1395 | len_byte -= in_charlen; | |
1396 | } | |
dafaabd1 AS |
1397 | |
1398 | /* If char_base is still negative we didn't find any translated | |
1399 | non-ASCII characters. */ | |
1400 | if (char_base < 0) | |
1401 | char_base = 0; | |
facdc750 RS |
1402 | } |
1403 | else | |
1404 | { | |
040272ce | 1405 | /* Unibyte buffer. */ |
a17ae96a | 1406 | char_base = 0; |
facdc750 RS |
1407 | while (--len >= 0) |
1408 | { | |
040272ce | 1409 | int c, translated; |
facdc750 RS |
1410 | |
1411 | /* If we got here and the RE flag is set, it's because we're | |
1412 | dealing with a regexp known to be trivial, so the backslash | |
1413 | just quotes the next character. */ | |
1414 | if (RE && *base_pat == '\\') | |
1415 | { | |
1416 | len--; | |
0190922f | 1417 | raw_pattern_size--; |
facdc750 RS |
1418 | base_pat++; |
1419 | } | |
1420 | c = *base_pat++; | |
aff2ce94 | 1421 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1422 | *pat++ = translated; |
1423 | } | |
1424 | } | |
1425 | ||
1426 | len_byte = pat - patbuf; | |
1427 | len = raw_pattern_size; | |
1428 | pat = base_pat = patbuf; | |
1429 | ||
040272ce | 1430 | if (boyer_moore_ok) |
facdc750 | 1431 | return boyer_moore (n, pat, len, len_byte, trt, inverse_trt, |
aff2ce94 | 1432 | pos, pos_byte, lim, lim_byte, |
a17ae96a | 1433 | char_base); |
facdc750 RS |
1434 | else |
1435 | return simple_search (n, pat, len, len_byte, trt, | |
1436 | pos, pos_byte, lim, lim_byte); | |
1437 | } | |
1438 | } | |
1439 | \f | |
1440 | /* Do a simple string search N times for the string PAT, | |
1441 | whose length is LEN/LEN_BYTE, | |
1442 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1443 | TRT is the translation table. | |
f8bd51c4 | 1444 | |
facdc750 RS |
1445 | Return the character position where the match is found. |
1446 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1447 | |
facdc750 RS |
1448 | This kind of search works regardless of what is in PAT and |
1449 | regardless of what is in TRT. It is used in cases where | |
1450 | boyer_moore cannot work. */ | |
1451 | ||
e7deaab0 | 1452 | static EMACS_INT |
971de7fb | 1453 | simple_search (int n, unsigned char *pat, int len, int len_byte, Lisp_Object trt, EMACS_INT pos, EMACS_INT pos_byte, EMACS_INT lim, EMACS_INT lim_byte) |
facdc750 RS |
1454 | { |
1455 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); | |
ab228c24 | 1456 | int forward = n > 0; |
49e44e81 KH |
1457 | /* Number of buffer bytes matched. Note that this may be different |
1458 | from len_byte in a multibyte buffer. */ | |
1459 | int match_byte; | |
facdc750 RS |
1460 | |
1461 | if (lim > pos && multibyte) | |
1462 | while (n > 0) | |
1463 | { | |
1464 | while (1) | |
f8bd51c4 | 1465 | { |
facdc750 | 1466 | /* Try matching at position POS. */ |
e7deaab0 AS |
1467 | EMACS_INT this_pos = pos; |
1468 | EMACS_INT this_pos_byte = pos_byte; | |
facdc750 | 1469 | int this_len = len; |
facdc750 | 1470 | unsigned char *p = pat; |
a9469498 | 1471 | if (pos + len > lim || pos_byte + len_byte > lim_byte) |
facdc750 RS |
1472 | goto stop; |
1473 | ||
1474 | while (this_len > 0) | |
1475 | { | |
1476 | int charlen, buf_charlen; | |
ab228c24 | 1477 | int pat_ch, buf_ch; |
facdc750 | 1478 | |
62a6e103 | 1479 | pat_ch = STRING_CHAR_AND_LENGTH (p, charlen); |
facdc750 | 1480 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
facdc750 | 1481 | buf_charlen); |
aff2ce94 | 1482 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1483 | |
1484 | if (buf_ch != pat_ch) | |
1485 | break; | |
ab228c24 | 1486 | |
ab228c24 RS |
1487 | this_len--; |
1488 | p += charlen; | |
1489 | ||
1490 | this_pos_byte += buf_charlen; | |
1491 | this_pos++; | |
facdc750 RS |
1492 | } |
1493 | ||
1494 | if (this_len == 0) | |
1495 | { | |
49e44e81 | 1496 | match_byte = this_pos_byte - pos_byte; |
facdc750 | 1497 | pos += len; |
49e44e81 | 1498 | pos_byte += match_byte; |
facdc750 RS |
1499 | break; |
1500 | } | |
1501 | ||
1502 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1503 | } |
facdc750 RS |
1504 | |
1505 | n--; | |
1506 | } | |
1507 | else if (lim > pos) | |
1508 | while (n > 0) | |
1509 | { | |
1510 | while (1) | |
f8bd51c4 | 1511 | { |
facdc750 | 1512 | /* Try matching at position POS. */ |
e7deaab0 | 1513 | EMACS_INT this_pos = pos; |
facdc750 RS |
1514 | int this_len = len; |
1515 | unsigned char *p = pat; | |
1516 | ||
1517 | if (pos + len > lim) | |
1518 | goto stop; | |
1519 | ||
1520 | while (this_len > 0) | |
1521 | { | |
1522 | int pat_ch = *p++; | |
1523 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1524 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1525 | |
1526 | if (buf_ch != pat_ch) | |
1527 | break; | |
ab228c24 RS |
1528 | |
1529 | this_len--; | |
1530 | this_pos++; | |
facdc750 RS |
1531 | } |
1532 | ||
1533 | if (this_len == 0) | |
1534 | { | |
49e44e81 | 1535 | match_byte = len; |
facdc750 RS |
1536 | pos += len; |
1537 | break; | |
1538 | } | |
1539 | ||
1540 | pos++; | |
f8bd51c4 | 1541 | } |
facdc750 RS |
1542 | |
1543 | n--; | |
1544 | } | |
1545 | /* Backwards search. */ | |
1546 | else if (lim < pos && multibyte) | |
1547 | while (n < 0) | |
1548 | { | |
1549 | while (1) | |
f8bd51c4 | 1550 | { |
facdc750 | 1551 | /* Try matching at position POS. */ |
8b264ecb AS |
1552 | EMACS_INT this_pos = pos; |
1553 | EMACS_INT this_pos_byte = pos_byte; | |
facdc750 | 1554 | int this_len = len; |
c525b3f2 | 1555 | const unsigned char *p = pat + len_byte; |
facdc750 | 1556 | |
8b264ecb | 1557 | if (this_pos - len < lim || (pos_byte - len_byte) < lim_byte) |
facdc750 RS |
1558 | goto stop; |
1559 | ||
1560 | while (this_len > 0) | |
1561 | { | |
8b264ecb | 1562 | int charlen; |
ab228c24 | 1563 | int pat_ch, buf_ch; |
facdc750 | 1564 | |
8b264ecb AS |
1565 | DEC_BOTH (this_pos, this_pos_byte); |
1566 | PREV_CHAR_BOUNDARY (p, pat); | |
1567 | pat_ch = STRING_CHAR (p); | |
1568 | buf_ch = STRING_CHAR (BYTE_POS_ADDR (this_pos_byte)); | |
aff2ce94 | 1569 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1570 | |
1571 | if (buf_ch != pat_ch) | |
1572 | break; | |
ab228c24 | 1573 | |
ab228c24 | 1574 | this_len--; |
facdc750 RS |
1575 | } |
1576 | ||
1577 | if (this_len == 0) | |
1578 | { | |
8b264ecb AS |
1579 | match_byte = pos_byte - this_pos_byte; |
1580 | pos = this_pos; | |
1581 | pos_byte = this_pos_byte; | |
facdc750 RS |
1582 | break; |
1583 | } | |
1584 | ||
1585 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1586 | } |
1587 | ||
facdc750 RS |
1588 | n++; |
1589 | } | |
1590 | else if (lim < pos) | |
1591 | while (n < 0) | |
1592 | { | |
1593 | while (1) | |
b6d6a51c | 1594 | { |
facdc750 | 1595 | /* Try matching at position POS. */ |
e7deaab0 | 1596 | EMACS_INT this_pos = pos - len; |
facdc750 RS |
1597 | int this_len = len; |
1598 | unsigned char *p = pat; | |
1599 | ||
a9469498 | 1600 | if (this_pos < lim) |
facdc750 RS |
1601 | goto stop; |
1602 | ||
1603 | while (this_len > 0) | |
1604 | { | |
1605 | int pat_ch = *p++; | |
1606 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1607 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1608 | |
1609 | if (buf_ch != pat_ch) | |
1610 | break; | |
ab228c24 RS |
1611 | this_len--; |
1612 | this_pos++; | |
facdc750 RS |
1613 | } |
1614 | ||
1615 | if (this_len == 0) | |
b6d6a51c | 1616 | { |
49e44e81 | 1617 | match_byte = len; |
facdc750 RS |
1618 | pos -= len; |
1619 | break; | |
b6d6a51c | 1620 | } |
facdc750 RS |
1621 | |
1622 | pos--; | |
b6d6a51c | 1623 | } |
facdc750 RS |
1624 | |
1625 | n++; | |
b6d6a51c | 1626 | } |
facdc750 RS |
1627 | |
1628 | stop: | |
1629 | if (n == 0) | |
aff2ce94 | 1630 | { |
ab228c24 | 1631 | if (forward) |
49e44e81 | 1632 | set_search_regs ((multibyte ? pos_byte : pos) - match_byte, match_byte); |
ab228c24 | 1633 | else |
49e44e81 | 1634 | set_search_regs (multibyte ? pos_byte : pos, match_byte); |
aff2ce94 RS |
1635 | |
1636 | return pos; | |
1637 | } | |
facdc750 RS |
1638 | else if (n > 0) |
1639 | return -n; | |
1640 | else | |
1641 | return n; | |
1642 | } | |
1643 | \f | |
0190922f | 1644 | /* Do Boyer-Moore search N times for the string BASE_PAT, |
facdc750 RS |
1645 | whose length is LEN/LEN_BYTE, |
1646 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1647 | DIRECTION says which direction we search in. | |
1648 | TRT and INVERSE_TRT are translation tables. | |
0190922f | 1649 | Characters in PAT are already translated by TRT. |
facdc750 | 1650 | |
0190922f KH |
1651 | This kind of search works if all the characters in BASE_PAT that |
1652 | have nontrivial translation are the same aside from the last byte. | |
1653 | This makes it possible to translate just the last byte of a | |
1654 | character, and do so after just a simple test of the context. | |
b2e6b10f | 1655 | CHAR_BASE is nonzero if there is such a non-ASCII character. |
facdc750 RS |
1656 | |
1657 | If that criterion is not satisfied, do not call this function. */ | |
1658 | ||
e7deaab0 | 1659 | static EMACS_INT |
d5a3eaaf AS |
1660 | boyer_moore (int n, unsigned char *base_pat, int len, int len_byte, |
1661 | Lisp_Object trt, Lisp_Object inverse_trt, | |
1662 | EMACS_INT pos, EMACS_INT pos_byte, | |
1663 | EMACS_INT lim, EMACS_INT lim_byte, int char_base) | |
facdc750 RS |
1664 | { |
1665 | int direction = ((n > 0) ? 1 : -1); | |
1666 | register int dirlen; | |
e7deaab0 AS |
1667 | EMACS_INT limit; |
1668 | int stride_for_teases = 0; | |
f4646fff | 1669 | int BM_tab[0400]; |
177c0ea7 | 1670 | register unsigned char *cursor, *p_limit; |
facdc750 | 1671 | register int i, j; |
cb6792d2 | 1672 | unsigned char *pat, *pat_end; |
facdc750 RS |
1673 | int multibyte = ! NILP (current_buffer->enable_multibyte_characters); |
1674 | ||
1675 | unsigned char simple_translate[0400]; | |
0190922f | 1676 | /* These are set to the preceding bytes of a byte to be translated |
49e44e81 | 1677 | if char_base is nonzero. As the maximum byte length of a |
a17ae96a | 1678 | multibyte character is 5, we have to check at most four previous |
0190922f KH |
1679 | bytes. */ |
1680 | int translate_prev_byte1 = 0; | |
1681 | int translate_prev_byte2 = 0; | |
1682 | int translate_prev_byte3 = 0; | |
a17ae96a | 1683 | int translate_prev_byte4 = 0; |
facdc750 | 1684 | |
f4646fff AS |
1685 | /* The general approach is that we are going to maintain that we know |
1686 | the first (closest to the present position, in whatever direction | |
1687 | we're searching) character that could possibly be the last | |
1688 | (furthest from present position) character of a valid match. We | |
1689 | advance the state of our knowledge by looking at that character | |
1690 | and seeing whether it indeed matches the last character of the | |
1691 | pattern. If it does, we take a closer look. If it does not, we | |
1692 | move our pointer (to putative last characters) as far as is | |
1693 | logically possible. This amount of movement, which I call a | |
1694 | stride, will be the length of the pattern if the actual character | |
1695 | appears nowhere in the pattern, otherwise it will be the distance | |
1696 | from the last occurrence of that character to the end of the | |
1697 | pattern. If the amount is zero we have a possible match. */ | |
1698 | ||
1699 | /* Here we make a "mickey mouse" BM table. The stride of the search | |
1700 | is determined only by the last character of the putative match. | |
1701 | If that character does not match, we will stride the proper | |
1702 | distance to propose a match that superimposes it on the last | |
1703 | instance of a character that matches it (per trt), or misses | |
1704 | it entirely if there is none. */ | |
facdc750 RS |
1705 | |
1706 | dirlen = len_byte * direction; | |
cb6792d2 RS |
1707 | |
1708 | /* Record position after the end of the pattern. */ | |
1709 | pat_end = base_pat + len_byte; | |
1710 | /* BASE_PAT points to a character that we start scanning from. | |
1711 | It is the first character in a forward search, | |
1712 | the last character in a backward search. */ | |
facdc750 | 1713 | if (direction < 0) |
cb6792d2 RS |
1714 | base_pat = pat_end - 1; |
1715 | ||
f4646fff AS |
1716 | /* A character that does not appear in the pattern induces a |
1717 | stride equal to the pattern length. */ | |
1718 | for (i = 0; i < 0400; i++) | |
1719 | BM_tab[i] = dirlen; | |
facdc750 RS |
1720 | |
1721 | /* We use this for translation, instead of TRT itself. | |
1722 | We fill this in to handle the characters that actually | |
1723 | occur in the pattern. Others don't matter anyway! */ | |
facdc750 RS |
1724 | for (i = 0; i < 0400; i++) |
1725 | simple_translate[i] = i; | |
1726 | ||
a17ae96a | 1727 | if (char_base) |
0190922f | 1728 | { |
a17ae96a | 1729 | /* Setup translate_prev_byte1/2/3/4 from CHAR_BASE. Only a |
0190922f | 1730 | byte following them are the target of translation. */ |
0190922f | 1731 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
a17ae96a | 1732 | int len = CHAR_STRING (char_base, str); |
0190922f KH |
1733 | |
1734 | translate_prev_byte1 = str[len - 2]; | |
1735 | if (len > 2) | |
1736 | { | |
1737 | translate_prev_byte2 = str[len - 3]; | |
1738 | if (len > 3) | |
a17ae96a KH |
1739 | { |
1740 | translate_prev_byte3 = str[len - 4]; | |
1741 | if (len > 4) | |
1742 | translate_prev_byte4 = str[len - 5]; | |
1743 | } | |
0190922f KH |
1744 | } |
1745 | } | |
1746 | ||
facdc750 | 1747 | i = 0; |
f4646fff | 1748 | while (i != dirlen) |
facdc750 | 1749 | { |
cb6792d2 | 1750 | unsigned char *ptr = base_pat + i; |
facdc750 | 1751 | i += direction; |
facdc750 | 1752 | if (! NILP (trt)) |
ca1d1d23 | 1753 | { |
49e44e81 KH |
1754 | /* If the byte currently looking at is the last of a |
1755 | character to check case-equivalents, set CH to that | |
1756 | character. An ASCII character and a non-ASCII character | |
1757 | matching with CHAR_BASE are to be checked. */ | |
a17ae96a KH |
1758 | int ch = -1; |
1759 | ||
1760 | if (ASCII_BYTE_P (*ptr) || ! multibyte) | |
1761 | ch = *ptr; | |
49e44e81 | 1762 | else if (char_base |
86dc6ccb | 1763 | && ((pat_end - ptr) == 1 || CHAR_HEAD_P (ptr[1]))) |
ca1d1d23 | 1764 | { |
49e44e81 KH |
1765 | unsigned char *charstart = ptr - 1; |
1766 | ||
1767 | while (! (CHAR_HEAD_P (*charstart))) | |
1768 | charstart--; | |
62a6e103 | 1769 | ch = STRING_CHAR (charstart); |
a17ae96a KH |
1770 | if (char_base != (ch & ~0x3F)) |
1771 | ch = -1; | |
ca1d1d23 | 1772 | } |
facdc750 | 1773 | |
f51995fa | 1774 | if (ch >= 0200) |
49e44e81 KH |
1775 | j = (ch & 0x3F) | 0200; |
1776 | else | |
1777 | j = *ptr; | |
1778 | ||
f4646fff | 1779 | if (i == dirlen) |
facdc750 | 1780 | stride_for_teases = BM_tab[j]; |
ab228c24 | 1781 | |
facdc750 RS |
1782 | BM_tab[j] = dirlen - i; |
1783 | /* A translation table is accompanied by its inverse -- see */ | |
177c0ea7 | 1784 | /* comment following downcase_table for details */ |
a17ae96a | 1785 | if (ch >= 0) |
ab228c24 RS |
1786 | { |
1787 | int starting_ch = ch; | |
49e44e81 | 1788 | int starting_j = j; |
a17ae96a | 1789 | |
ab228c24 RS |
1790 | while (1) |
1791 | { | |
1792 | TRANSLATE (ch, inverse_trt, ch); | |
f51995fa | 1793 | if (ch >= 0200) |
49e44e81 | 1794 | j = (ch & 0x3F) | 0200; |
ab228c24 | 1795 | else |
a17ae96a | 1796 | j = ch; |
ab228c24 RS |
1797 | |
1798 | /* For all the characters that map into CH, | |
1799 | set up simple_translate to map the last byte | |
1800 | into STARTING_J. */ | |
1801 | simple_translate[j] = starting_j; | |
1802 | if (ch == starting_ch) | |
1803 | break; | |
1804 | BM_tab[j] = dirlen - i; | |
1805 | } | |
1806 | } | |
facdc750 RS |
1807 | } |
1808 | else | |
1809 | { | |
1810 | j = *ptr; | |
1811 | ||
f4646fff | 1812 | if (i == dirlen) |
facdc750 RS |
1813 | stride_for_teases = BM_tab[j]; |
1814 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1815 | } |
f4646fff AS |
1816 | /* stride_for_teases tells how much to stride if we get a |
1817 | match on the far character but are subsequently | |
1818 | disappointed, by recording what the stride would have been | |
1819 | for that character if the last character had been | |
1820 | different. */ | |
facdc750 | 1821 | } |
facdc750 RS |
1822 | pos_byte += dirlen - ((direction > 0) ? direction : 0); |
1823 | /* loop invariant - POS_BYTE points at where last char (first | |
1824 | char if reverse) of pattern would align in a possible match. */ | |
1825 | while (n != 0) | |
1826 | { | |
e7deaab0 | 1827 | EMACS_INT tail_end; |
facdc750 RS |
1828 | unsigned char *tail_end_ptr; |
1829 | ||
1830 | /* It's been reported that some (broken) compiler thinks that | |
1831 | Boolean expressions in an arithmetic context are unsigned. | |
1832 | Using an explicit ?1:0 prevents this. */ | |
1833 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1834 | < 0) | |
1835 | return (n * (0 - direction)); | |
1836 | /* First we do the part we can by pointers (maybe nothing) */ | |
1837 | QUIT; | |
1838 | pat = base_pat; | |
1839 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1840 | if (direction > 0) |
1841 | { | |
1842 | limit = BUFFER_CEILING_OF (limit); | |
1843 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1844 | can take on without hitting edge of buffer or the gap. */ | |
1845 | limit = min (limit, pos_byte + 20000); | |
1846 | limit = min (limit, lim_byte - 1); | |
1847 | } | |
1848 | else | |
1849 | { | |
1850 | limit = BUFFER_FLOOR_OF (limit); | |
1851 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1852 | can take on without hitting edge of buffer or the gap. */ | |
1853 | limit = max (limit, pos_byte - 20000); | |
1854 | limit = max (limit, lim_byte); | |
1855 | } | |
facdc750 RS |
1856 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1857 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1858 | ||
1859 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1860 | { |
facdc750 RS |
1861 | unsigned char *p2; |
1862 | ||
1863 | p_limit = BYTE_POS_ADDR (limit); | |
1864 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
f4646fff | 1865 | /* In this loop, pos + cursor - p2 is the surrogate for pos. */ |
facdc750 | 1866 | while (1) /* use one cursor setting as long as i can */ |
ca1d1d23 | 1867 | { |
facdc750 | 1868 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1869 | { |
f4646fff AS |
1870 | while (cursor <= p_limit) |
1871 | { | |
1872 | if (BM_tab[*cursor] == 0) | |
1873 | goto hit; | |
facdc750 | 1874 | cursor += BM_tab[*cursor]; |
f4646fff | 1875 | } |
facdc750 RS |
1876 | } |
1877 | else | |
1878 | { | |
f4646fff AS |
1879 | while (cursor >= p_limit) |
1880 | { | |
1881 | if (BM_tab[*cursor] == 0) | |
1882 | goto hit; | |
facdc750 | 1883 | cursor += BM_tab[*cursor]; |
f4646fff | 1884 | } |
facdc750 | 1885 | } |
f4646fff AS |
1886 | /* If you are here, cursor is beyond the end of the |
1887 | searched region. You fail to match within the | |
1888 | permitted region and would otherwise try a character | |
1889 | beyond that region. */ | |
1890 | break; | |
1891 | ||
1892 | hit: | |
facdc750 RS |
1893 | i = dirlen - direction; |
1894 | if (! NILP (trt)) | |
1895 | { | |
1896 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1897 | { |
facdc750 RS |
1898 | int ch; |
1899 | cursor -= direction; | |
1900 | /* Translate only the last byte of a character. */ | |
1901 | if (! multibyte | |
1902 | || ((cursor == tail_end_ptr | |
1903 | || CHAR_HEAD_P (cursor[1])) | |
1904 | && (CHAR_HEAD_P (cursor[0]) | |
0190922f KH |
1905 | /* Check if this is the last byte of |
1906 | a translable character. */ | |
1907 | || (translate_prev_byte1 == cursor[-1] | |
1908 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1909 | || (translate_prev_byte2 == cursor[-2] | |
1910 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1911 | || (translate_prev_byte3 == cursor[-3])))))))) | |
facdc750 RS |
1912 | ch = simple_translate[*cursor]; |
1913 | else | |
1914 | ch = *cursor; | |
1915 | if (pat[i] != ch) | |
1916 | break; | |
ca1d1d23 | 1917 | } |
facdc750 RS |
1918 | } |
1919 | else | |
1920 | { | |
1921 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1922 | { |
facdc750 RS |
1923 | cursor -= direction; |
1924 | if (pat[i] != *cursor) | |
1925 | break; | |
ca1d1d23 | 1926 | } |
facdc750 RS |
1927 | } |
1928 | cursor += dirlen - i - direction; /* fix cursor */ | |
1929 | if (i + direction == 0) | |
1930 | { | |
e7deaab0 | 1931 | EMACS_INT position, start, end; |
0c8533c6 | 1932 | |
facdc750 | 1933 | cursor -= direction; |
1113d9db | 1934 | |
facdc750 RS |
1935 | position = pos_byte + cursor - p2 + ((direction > 0) |
1936 | ? 1 - len_byte : 0); | |
1937 | set_search_regs (position, len_byte); | |
ca325161 | 1938 | |
ef887810 RS |
1939 | if (NILP (Vinhibit_changing_match_data)) |
1940 | { | |
1941 | start = search_regs.start[0]; | |
1942 | end = search_regs.end[0]; | |
1943 | } | |
1944 | else | |
1945 | /* If Vinhibit_changing_match_data is non-nil, | |
1946 | search_regs will not be changed. So let's | |
1947 | compute start and end here. */ | |
1948 | { | |
1949 | start = BYTE_TO_CHAR (position); | |
1950 | end = BYTE_TO_CHAR (position + len_byte); | |
1951 | } | |
1952 | ||
facdc750 RS |
1953 | if ((n -= direction) != 0) |
1954 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1955 | else |
ef887810 | 1956 | return direction > 0 ? end : start; |
ca1d1d23 | 1957 | } |
facdc750 RS |
1958 | else |
1959 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1960 | } |
facdc750 RS |
1961 | pos_byte += cursor - p2; |
1962 | } | |
1963 | else | |
f4646fff AS |
1964 | /* Now we'll pick up a clump that has to be done the hard |
1965 | way because it covers a discontinuity. */ | |
facdc750 RS |
1966 | { |
1967 | limit = ((direction > 0) | |
1968 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1969 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1970 | limit = ((direction > 0) | |
1971 | ? min (limit + len_byte, lim_byte - 1) | |
1972 | : max (limit - len_byte, lim_byte)); | |
1973 | /* LIMIT is now the last value POS_BYTE can have | |
1974 | and still be valid for a possible match. */ | |
1975 | while (1) | |
ca1d1d23 | 1976 | { |
f4646fff AS |
1977 | /* This loop can be coded for space rather than |
1978 | speed because it will usually run only once. | |
1979 | (the reach is at most len + 21, and typically | |
1980 | does not exceed len). */ | |
facdc750 | 1981 | while ((limit - pos_byte) * direction >= 0) |
f4646fff AS |
1982 | { |
1983 | int ch = FETCH_BYTE (pos_byte); | |
1984 | if (BM_tab[ch] == 0) | |
1985 | goto hit2; | |
1986 | pos_byte += BM_tab[ch]; | |
1987 | } | |
1988 | break; /* ran off the end */ | |
1989 | ||
1990 | hit2: | |
1991 | /* Found what might be a match. */ | |
facdc750 RS |
1992 | i = dirlen - direction; |
1993 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1994 | { |
facdc750 RS |
1995 | int ch; |
1996 | unsigned char *ptr; | |
1997 | pos_byte -= direction; | |
1998 | ptr = BYTE_POS_ADDR (pos_byte); | |
1999 | /* Translate only the last byte of a character. */ | |
2000 | if (! multibyte | |
2001 | || ((ptr == tail_end_ptr | |
2002 | || CHAR_HEAD_P (ptr[1])) | |
2003 | && (CHAR_HEAD_P (ptr[0]) | |
0190922f KH |
2004 | /* Check if this is the last byte of a |
2005 | translable character. */ | |
2006 | || (translate_prev_byte1 == ptr[-1] | |
2007 | && (CHAR_HEAD_P (translate_prev_byte1) | |
2008 | || (translate_prev_byte2 == ptr[-2] | |
2009 | && (CHAR_HEAD_P (translate_prev_byte2) | |
2010 | || translate_prev_byte3 == ptr[-3]))))))) | |
facdc750 RS |
2011 | ch = simple_translate[*ptr]; |
2012 | else | |
2013 | ch = *ptr; | |
2014 | if (pat[i] != ch) | |
2015 | break; | |
2016 | } | |
2017 | /* Above loop has moved POS_BYTE part or all the way | |
2018 | back to the first pos (last pos if reverse). | |
2019 | Set it once again at the last (first if reverse) char. */ | |
f4646fff | 2020 | pos_byte += dirlen - i - direction; |
facdc750 RS |
2021 | if (i + direction == 0) |
2022 | { | |
e7deaab0 | 2023 | EMACS_INT position, start, end; |
facdc750 | 2024 | pos_byte -= direction; |
1113d9db | 2025 | |
facdc750 | 2026 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
facdc750 | 2027 | set_search_regs (position, len_byte); |
ca325161 | 2028 | |
ef887810 RS |
2029 | if (NILP (Vinhibit_changing_match_data)) |
2030 | { | |
2031 | start = search_regs.start[0]; | |
2032 | end = search_regs.end[0]; | |
2033 | } | |
2034 | else | |
2035 | /* If Vinhibit_changing_match_data is non-nil, | |
2036 | search_regs will not be changed. So let's | |
2037 | compute start and end here. */ | |
2038 | { | |
2039 | start = BYTE_TO_CHAR (position); | |
2040 | end = BYTE_TO_CHAR (position + len_byte); | |
2041 | } | |
2042 | ||
facdc750 RS |
2043 | if ((n -= direction) != 0) |
2044 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 2045 | else |
ef887810 | 2046 | return direction > 0 ? end : start; |
ca1d1d23 | 2047 | } |
facdc750 RS |
2048 | else |
2049 | pos_byte += stride_for_teases; | |
2050 | } | |
2051 | } | |
2052 | /* We have done one clump. Can we continue? */ | |
2053 | if ((lim_byte - pos_byte) * direction < 0) | |
2054 | return ((0 - n) * direction); | |
ca1d1d23 | 2055 | } |
facdc750 | 2056 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 2057 | } |
ca325161 | 2058 | |
fa8ed3e0 | 2059 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
2060 | for the overall match just found in the current buffer. |
2061 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
2062 | |
2063 | static void | |
971de7fb | 2064 | set_search_regs (EMACS_INT beg_byte, EMACS_INT nbytes) |
ca325161 | 2065 | { |
a7e4cdde RS |
2066 | int i; |
2067 | ||
ef887810 RS |
2068 | if (!NILP (Vinhibit_changing_match_data)) |
2069 | return; | |
2070 | ||
ca325161 RS |
2071 | /* Make sure we have registers in which to store |
2072 | the match position. */ | |
2073 | if (search_regs.num_regs == 0) | |
2074 | { | |
2d4a771a RS |
2075 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
2076 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 2077 | search_regs.num_regs = 2; |
ca325161 RS |
2078 | } |
2079 | ||
a7e4cdde RS |
2080 | /* Clear out the other registers. */ |
2081 | for (i = 1; i < search_regs.num_regs; i++) | |
2082 | { | |
2083 | search_regs.start[i] = -1; | |
2084 | search_regs.end[i] = -1; | |
2085 | } | |
2086 | ||
fa8ed3e0 RS |
2087 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
2088 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 2089 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 2090 | } |
ca1d1d23 | 2091 | \f |
c6e0e1cc CY |
2092 | /* Given STRING, a string of words separated by word delimiters, |
2093 | compute a regexp that matches those exact words separated by | |
2094 | arbitrary punctuation. If LAX is nonzero, the end of the string | |
2095 | need not match a word boundary unless it ends in whitespace. */ | |
ca1d1d23 JB |
2096 | |
2097 | static Lisp_Object | |
971de7fb | 2098 | wordify (Lisp_Object string, int lax) |
ca1d1d23 JB |
2099 | { |
2100 | register unsigned char *p, *o; | |
0c8533c6 | 2101 | register int i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 2102 | Lisp_Object val; |
0c8533c6 | 2103 | int prev_c = 0; |
c6e0e1cc | 2104 | int adjust, whitespace_at_end; |
ca1d1d23 | 2105 | |
b7826503 | 2106 | CHECK_STRING (string); |
d5db4077 KR |
2107 | p = SDATA (string); |
2108 | len = SCHARS (string); | |
ca1d1d23 | 2109 | |
0c8533c6 RS |
2110 | for (i = 0, i_byte = 0; i < len; ) |
2111 | { | |
2112 | int c; | |
177c0ea7 | 2113 | |
93daa011 | 2114 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
0c8533c6 RS |
2115 | |
2116 | if (SYNTAX (c) != Sword) | |
2117 | { | |
2118 | punct_count++; | |
2119 | if (i > 0 && SYNTAX (prev_c) == Sword) | |
2120 | word_count++; | |
2121 | } | |
ca1d1d23 | 2122 | |
0c8533c6 RS |
2123 | prev_c = c; |
2124 | } | |
2125 | ||
2126 | if (SYNTAX (prev_c) == Sword) | |
c6e0e1cc CY |
2127 | { |
2128 | word_count++; | |
2129 | whitespace_at_end = 0; | |
2130 | } | |
2131 | else | |
2132 | whitespace_at_end = 1; | |
2133 | ||
0c8533c6 | 2134 | if (!word_count) |
c60416e0 | 2135 | return empty_unibyte_string; |
0c8533c6 | 2136 | |
c6e0e1cc CY |
2137 | adjust = - punct_count + 5 * (word_count - 1) |
2138 | + ((lax && !whitespace_at_end) ? 2 : 4); | |
8a2df937 RS |
2139 | if (STRING_MULTIBYTE (string)) |
2140 | val = make_uninit_multibyte_string (len + adjust, | |
d5db4077 | 2141 | SBYTES (string) |
8a2df937 RS |
2142 | + adjust); |
2143 | else | |
2144 | val = make_uninit_string (len + adjust); | |
ca1d1d23 | 2145 | |
d5db4077 | 2146 | o = SDATA (val); |
ca1d1d23 JB |
2147 | *o++ = '\\'; |
2148 | *o++ = 'b'; | |
1e9582d4 | 2149 | prev_c = 0; |
ca1d1d23 | 2150 | |
1e9582d4 RS |
2151 | for (i = 0, i_byte = 0; i < len; ) |
2152 | { | |
2153 | int c; | |
2154 | int i_byte_orig = i_byte; | |
177c0ea7 | 2155 | |
93daa011 | 2156 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
1e9582d4 RS |
2157 | |
2158 | if (SYNTAX (c) == Sword) | |
2159 | { | |
72af86bd | 2160 | memcpy (o, SDATA (string) + i_byte_orig, i_byte - i_byte_orig); |
1e9582d4 RS |
2161 | o += i_byte - i_byte_orig; |
2162 | } | |
2163 | else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count) | |
2164 | { | |
2165 | *o++ = '\\'; | |
2166 | *o++ = 'W'; | |
2167 | *o++ = '\\'; | |
2168 | *o++ = 'W'; | |
2169 | *o++ = '*'; | |
2170 | } | |
2171 | ||
2172 | prev_c = c; | |
2173 | } | |
ca1d1d23 | 2174 | |
c6e0e1cc CY |
2175 | if (!lax || whitespace_at_end) |
2176 | { | |
2177 | *o++ = '\\'; | |
2178 | *o++ = 'b'; | |
2179 | } | |
ca1d1d23 JB |
2180 | |
2181 | return val; | |
2182 | } | |
2183 | \f | |
2184 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
8c1a1077 PJ |
2185 | "MSearch backward: ", |
2186 | doc: /* Search backward from point for STRING. | |
2187 | Set point to the beginning of the occurrence found, and return point. | |
2188 | An optional second argument bounds the search; it is a buffer position. | |
2189 | The match found must not extend before that position. | |
2190 | Optional third argument, if t, means if fail just return nil (no error). | |
2191 | If not nil and not t, position at limit of search and return nil. | |
2192 | Optional fourth argument is repeat count--search for successive occurrences. | |
2193 | ||
2194 | Search case-sensitivity is determined by the value of the variable | |
2195 | `case-fold-search', which see. | |
2196 | ||
2197 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2198 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2199 | { |
b819a390 | 2200 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2201 | } |
2202 | ||
6af43974 | 2203 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2204 | doc: /* Search forward from point for STRING. |
2205 | Set point to the end of the occurrence found, and return point. | |
2206 | An optional second argument bounds the search; it is a buffer position. | |
d6aacdcd | 2207 | The match found must not extend after that position. A value of nil is |
48740f01 | 2208 | equivalent to (point-max). |
8c1a1077 PJ |
2209 | Optional third argument, if t, means if fail just return nil (no error). |
2210 | If not nil and not t, move to limit of search and return nil. | |
2211 | Optional fourth argument is repeat count--search for successive occurrences. | |
2212 | ||
2213 | Search case-sensitivity is determined by the value of the variable | |
2214 | `case-fold-search', which see. | |
2215 | ||
2216 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2217 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2218 | { |
b819a390 | 2219 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2220 | } |
2221 | ||
2222 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
8c1a1077 PJ |
2223 | "sWord search backward: ", |
2224 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2225 | Set point to the beginning of the occurrence found, and return point. | |
2226 | An optional second argument bounds the search; it is a buffer position. | |
2227 | The match found must not extend before that position. | |
2228 | Optional third argument, if t, means if fail just return nil (no error). | |
2229 | If not nil and not t, move to limit of search and return nil. | |
2230 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2231 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2232 | { |
c6e0e1cc | 2233 | return search_command (wordify (string, 0), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2234 | } |
2235 | ||
2236 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
8c1a1077 PJ |
2237 | "sWord search: ", |
2238 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2239 | Set point to the end of the occurrence found, and return point. | |
2240 | An optional second argument bounds the search; it is a buffer position. | |
2241 | The match found must not extend after that position. | |
2242 | Optional third argument, if t, means if fail just return nil (no error). | |
2243 | If not nil and not t, move to limit of search and return nil. | |
2244 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2245 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2246 | { |
c6e0e1cc CY |
2247 | return search_command (wordify (string, 0), bound, noerror, count, 1, 1, 0); |
2248 | } | |
2249 | ||
2250 | DEFUN ("word-search-backward-lax", Fword_search_backward_lax, Sword_search_backward_lax, 1, 4, | |
2251 | "sWord search backward: ", | |
2252 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2253 | Set point to the beginning of the occurrence found, and return point. | |
2254 | ||
2255 | Unlike `word-search-backward', the end of STRING need not match a word | |
2256 | boundary unless it ends in whitespace. | |
2257 | ||
2258 | An optional second argument bounds the search; it is a buffer position. | |
2259 | The match found must not extend before that position. | |
2260 | Optional third argument, if t, means if fail just return nil (no error). | |
2261 | If not nil and not t, move to limit of search and return nil. | |
2262 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2263 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
c6e0e1cc CY |
2264 | { |
2265 | return search_command (wordify (string, 1), bound, noerror, count, -1, 1, 0); | |
2266 | } | |
2267 | ||
2268 | DEFUN ("word-search-forward-lax", Fword_search_forward_lax, Sword_search_forward_lax, 1, 4, | |
2269 | "sWord search: ", | |
2270 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2271 | Set point to the end of the occurrence found, and return point. | |
2272 | ||
2273 | Unlike `word-search-forward', the end of STRING need not match a word | |
2274 | boundary unless it ends in whitespace. | |
2275 | ||
2276 | An optional second argument bounds the search; it is a buffer position. | |
2277 | The match found must not extend after that position. | |
2278 | Optional third argument, if t, means if fail just return nil (no error). | |
2279 | If not nil and not t, move to limit of search and return nil. | |
2280 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2281 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
c6e0e1cc CY |
2282 | { |
2283 | return search_command (wordify (string, 1), bound, noerror, count, 1, 1, 0); | |
ca1d1d23 JB |
2284 | } |
2285 | ||
2286 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
8c1a1077 PJ |
2287 | "sRE search backward: ", |
2288 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2289 | Set point to the beginning of the match, and return point. | |
2290 | The match found is the one starting last in the buffer | |
2291 | and yet ending before the origin of the search. | |
2292 | An optional second argument bounds the search; it is a buffer position. | |
2293 | The match found must start at or after that position. | |
2294 | Optional third argument, if t, means if fail just return nil (no error). | |
2295 | If not nil and not t, move to limit of search and return nil. | |
2296 | Optional fourth argument is repeat count--search for successive occurrences. | |
2297 | See also the functions `match-beginning', `match-end', `match-string', | |
2298 | and `replace-match'. */) | |
5842a27b | 2299 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2300 | { |
b819a390 | 2301 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2302 | } |
2303 | ||
2304 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2305 | "sRE search: ", |
2306 | doc: /* Search forward from point for regular expression REGEXP. | |
2307 | Set point to the end of the occurrence found, and return point. | |
2308 | An optional second argument bounds the search; it is a buffer position. | |
2309 | The match found must not extend after that position. | |
2310 | Optional third argument, if t, means if fail just return nil (no error). | |
2311 | If not nil and not t, move to limit of search and return nil. | |
2312 | Optional fourth argument is repeat count--search for successive occurrences. | |
2313 | See also the functions `match-beginning', `match-end', `match-string', | |
2314 | and `replace-match'. */) | |
5842a27b | 2315 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2316 | { |
b819a390 RS |
2317 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2318 | } | |
2319 | ||
2320 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2321 | "sPosix search backward: ", |
2322 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2323 | Find the longest match in accord with Posix regular expression rules. | |
2324 | Set point to the beginning of the match, and return point. | |
2325 | The match found is the one starting last in the buffer | |
2326 | and yet ending before the origin of the search. | |
2327 | An optional second argument bounds the search; it is a buffer position. | |
2328 | The match found must start at or after that position. | |
2329 | Optional third argument, if t, means if fail just return nil (no error). | |
2330 | If not nil and not t, move to limit of search and return nil. | |
2331 | Optional fourth argument is repeat count--search for successive occurrences. | |
2332 | See also the functions `match-beginning', `match-end', `match-string', | |
2333 | and `replace-match'. */) | |
5842a27b | 2334 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2335 | { |
2336 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2337 | } | |
2338 | ||
2339 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2340 | "sPosix search: ", |
2341 | doc: /* Search forward from point for regular expression REGEXP. | |
2342 | Find the longest match in accord with Posix regular expression rules. | |
2343 | Set point to the end of the occurrence found, and return point. | |
2344 | An optional second argument bounds the search; it is a buffer position. | |
2345 | The match found must not extend after that position. | |
2346 | Optional third argument, if t, means if fail just return nil (no error). | |
2347 | If not nil and not t, move to limit of search and return nil. | |
2348 | Optional fourth argument is repeat count--search for successive occurrences. | |
2349 | See also the functions `match-beginning', `match-end', `match-string', | |
2350 | and `replace-match'. */) | |
5842a27b | 2351 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2352 | { |
2353 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2354 | } |
2355 | \f | |
d7a5ad5f | 2356 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2357 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2358 | Leave point at the end of the replacement text. |
2359 | ||
8c1a1077 PJ |
2360 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. |
2361 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2362 | based on the replaced text. | |
2363 | If the replaced text has only capital letters | |
2364 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
4dd0c271 RS |
2365 | Otherwise if all words are capitalized in the replaced text, |
2366 | capitalize each word in NEWTEXT. | |
8c1a1077 PJ |
2367 | |
2368 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2369 | Otherwise treat `\\' as special: | |
2370 | `\\&' in NEWTEXT means substitute original matched text. | |
2371 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2372 | If Nth parens didn't match, substitute nothing. | |
2373 | `\\\\' means insert one `\\'. | |
4dd0c271 RS |
2374 | Case conversion does not apply to these substitutions. |
2375 | ||
8c1a1077 | 2376 | FIXEDCASE and LITERAL are optional arguments. |
8c1a1077 PJ |
2377 | |
2378 | The optional fourth argument STRING can be a string to modify. | |
2379 | This is meaningful when the previous match was done against STRING, | |
2380 | using `string-match'. When used this way, `replace-match' | |
2381 | creates and returns a new string made by copying STRING and replacing | |
2382 | the part of STRING that was matched. | |
2383 | ||
2384 | The optional fifth argument SUBEXP specifies a subexpression; | |
2385 | it says to replace just that subexpression with NEWTEXT, | |
2386 | rather than replacing the entire matched text. | |
2387 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2388 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2389 | NEWTEXT in place of subexp N. | |
2390 | This is useful only after a regular expression search or match, | |
2391 | since only regular expressions have distinguished subexpressions. */) | |
5842a27b | 2392 | (Lisp_Object newtext, Lisp_Object fixedcase, Lisp_Object literal, Lisp_Object string, Lisp_Object subexp) |
ca1d1d23 JB |
2393 | { |
2394 | enum { nochange, all_caps, cap_initial } case_action; | |
ac3b28b1 | 2395 | register int pos, pos_byte; |
ca1d1d23 | 2396 | int some_multiletter_word; |
97832bd0 | 2397 | int some_lowercase; |
73dc8771 | 2398 | int some_uppercase; |
208767c3 | 2399 | int some_nonuppercase_initial; |
ca1d1d23 | 2400 | register int c, prevc; |
d7a5ad5f | 2401 | int sub; |
e7deaab0 | 2402 | EMACS_INT opoint, newpoint; |
ca1d1d23 | 2403 | |
b7826503 | 2404 | CHECK_STRING (newtext); |
ca1d1d23 | 2405 | |
080c45fd | 2406 | if (! NILP (string)) |
b7826503 | 2407 | CHECK_STRING (string); |
080c45fd | 2408 | |
ca1d1d23 JB |
2409 | case_action = nochange; /* We tried an initialization */ |
2410 | /* but some C compilers blew it */ | |
4746118a JB |
2411 | |
2412 | if (search_regs.num_regs <= 0) | |
d72cdbfc | 2413 | error ("`replace-match' called before any match found"); |
4746118a | 2414 | |
d7a5ad5f RS |
2415 | if (NILP (subexp)) |
2416 | sub = 0; | |
2417 | else | |
2418 | { | |
b7826503 | 2419 | CHECK_NUMBER (subexp); |
d7a5ad5f RS |
2420 | sub = XINT (subexp); |
2421 | if (sub < 0 || sub >= search_regs.num_regs) | |
2422 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2423 | } | |
2424 | ||
080c45fd RS |
2425 | if (NILP (string)) |
2426 | { | |
d7a5ad5f RS |
2427 | if (search_regs.start[sub] < BEGV |
2428 | || search_regs.start[sub] > search_regs.end[sub] | |
2429 | || search_regs.end[sub] > ZV) | |
2430 | args_out_of_range (make_number (search_regs.start[sub]), | |
2431 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2432 | } |
2433 | else | |
2434 | { | |
d7a5ad5f RS |
2435 | if (search_regs.start[sub] < 0 |
2436 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2437 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2438 | args_out_of_range (make_number (search_regs.start[sub]), |
2439 | make_number (search_regs.end[sub])); | |
080c45fd | 2440 | } |
ca1d1d23 JB |
2441 | |
2442 | if (NILP (fixedcase)) | |
2443 | { | |
2444 | /* Decide how to casify by examining the matched text. */ | |
e7deaab0 | 2445 | EMACS_INT last; |
ca1d1d23 | 2446 | |
ac3b28b1 KH |
2447 | pos = search_regs.start[sub]; |
2448 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2449 | |
2450 | if (NILP (string)) | |
ac3b28b1 | 2451 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2452 | else |
ac3b28b1 | 2453 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2454 | |
ca1d1d23 JB |
2455 | prevc = '\n'; |
2456 | case_action = all_caps; | |
2457 | ||
2458 | /* some_multiletter_word is set nonzero if any original word | |
2459 | is more than one letter long. */ | |
2460 | some_multiletter_word = 0; | |
97832bd0 | 2461 | some_lowercase = 0; |
208767c3 | 2462 | some_nonuppercase_initial = 0; |
73dc8771 | 2463 | some_uppercase = 0; |
ca1d1d23 | 2464 | |
ac3b28b1 | 2465 | while (pos < last) |
ca1d1d23 | 2466 | { |
080c45fd | 2467 | if (NILP (string)) |
ac3b28b1 | 2468 | { |
93daa011 | 2469 | c = FETCH_CHAR_AS_MULTIBYTE (pos_byte); |
ac3b28b1 KH |
2470 | INC_BOTH (pos, pos_byte); |
2471 | } | |
080c45fd | 2472 | else |
93daa011 | 2473 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2474 | |
ca1d1d23 JB |
2475 | if (LOWERCASEP (c)) |
2476 | { | |
2477 | /* Cannot be all caps if any original char is lower case */ | |
2478 | ||
97832bd0 | 2479 | some_lowercase = 1; |
ca1d1d23 | 2480 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2481 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2482 | else |
2483 | some_multiletter_word = 1; | |
2484 | } | |
d16c2b66 | 2485 | else if (UPPERCASEP (c)) |
ca1d1d23 | 2486 | { |
73dc8771 | 2487 | some_uppercase = 1; |
97832bd0 | 2488 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2489 | ; |
97832bd0 | 2490 | else |
ca1d1d23 JB |
2491 | some_multiletter_word = 1; |
2492 | } | |
208767c3 RS |
2493 | else |
2494 | { | |
2495 | /* If the initial is a caseless word constituent, | |
2496 | treat that like a lowercase initial. */ | |
2497 | if (SYNTAX (prevc) != Sword) | |
2498 | some_nonuppercase_initial = 1; | |
2499 | } | |
ca1d1d23 JB |
2500 | |
2501 | prevc = c; | |
2502 | } | |
2503 | ||
97832bd0 RS |
2504 | /* Convert to all caps if the old text is all caps |
2505 | and has at least one multiletter word. */ | |
2506 | if (! some_lowercase && some_multiletter_word) | |
2507 | case_action = all_caps; | |
c4d460ce | 2508 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2509 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2510 | case_action = cap_initial; |
208767c3 | 2511 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2512 | /* Should x -> yz, operating on X, give Yz or YZ? |
2513 | We'll assume the latter. */ | |
2514 | case_action = all_caps; | |
97832bd0 RS |
2515 | else |
2516 | case_action = nochange; | |
ca1d1d23 JB |
2517 | } |
2518 | ||
080c45fd RS |
2519 | /* Do replacement in a string. */ |
2520 | if (!NILP (string)) | |
2521 | { | |
2522 | Lisp_Object before, after; | |
2523 | ||
2524 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2525 | make_number (search_regs.start[sub])); |
2526 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2527 | |
636a5e28 RS |
2528 | /* Substitute parts of the match into NEWTEXT |
2529 | if desired. */ | |
080c45fd RS |
2530 | if (NILP (literal)) |
2531 | { | |
e7deaab0 AS |
2532 | EMACS_INT lastpos = 0; |
2533 | EMACS_INT lastpos_byte = 0; | |
080c45fd RS |
2534 | /* We build up the substituted string in ACCUM. */ |
2535 | Lisp_Object accum; | |
2536 | Lisp_Object middle; | |
d5db4077 | 2537 | int length = SBYTES (newtext); |
080c45fd RS |
2538 | |
2539 | accum = Qnil; | |
2540 | ||
ac3b28b1 | 2541 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd RS |
2542 | { |
2543 | int substart = -1; | |
6bbd7a29 | 2544 | int subend = 0; |
1e79ec24 | 2545 | int delbackslash = 0; |
080c45fd | 2546 | |
0c8533c6 RS |
2547 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2548 | ||
080c45fd RS |
2549 | if (c == '\\') |
2550 | { | |
0c8533c6 | 2551 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2552 | |
080c45fd RS |
2553 | if (c == '&') |
2554 | { | |
d7a5ad5f RS |
2555 | substart = search_regs.start[sub]; |
2556 | subend = search_regs.end[sub]; | |
080c45fd | 2557 | } |
76fc1ea2 | 2558 | else if (c >= '1' && c <= '9') |
080c45fd | 2559 | { |
76fc1ea2 KH |
2560 | if (search_regs.start[c - '0'] >= 0 |
2561 | && c <= search_regs.num_regs + '0') | |
080c45fd RS |
2562 | { |
2563 | substart = search_regs.start[c - '0']; | |
2564 | subend = search_regs.end[c - '0']; | |
2565 | } | |
76fc1ea2 KH |
2566 | else |
2567 | { | |
2568 | /* If that subexp did not match, | |
2569 | replace \\N with nothing. */ | |
2570 | substart = 0; | |
2571 | subend = 0; | |
2572 | } | |
080c45fd | 2573 | } |
1e79ec24 KH |
2574 | else if (c == '\\') |
2575 | delbackslash = 1; | |
636a5e28 RS |
2576 | else |
2577 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2578 | } |
2579 | if (substart >= 0) | |
2580 | { | |
d131e79c RS |
2581 | if (pos - 2 != lastpos) |
2582 | middle = substring_both (newtext, lastpos, | |
2583 | lastpos_byte, | |
2584 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2585 | else |
2586 | middle = Qnil; | |
2587 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2588 | Fsubstring (string, |
2589 | make_number (substart), | |
080c45fd RS |
2590 | make_number (subend))); |
2591 | lastpos = pos; | |
0c8533c6 | 2592 | lastpos_byte = pos_byte; |
080c45fd | 2593 | } |
1e79ec24 KH |
2594 | else if (delbackslash) |
2595 | { | |
d131e79c RS |
2596 | middle = substring_both (newtext, lastpos, |
2597 | lastpos_byte, | |
2598 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2599 | |
1e79ec24 KH |
2600 | accum = concat2 (accum, middle); |
2601 | lastpos = pos; | |
0c8533c6 | 2602 | lastpos_byte = pos_byte; |
1e79ec24 | 2603 | } |
080c45fd RS |
2604 | } |
2605 | ||
d131e79c RS |
2606 | if (pos != lastpos) |
2607 | middle = substring_both (newtext, lastpos, | |
2608 | lastpos_byte, | |
0c8533c6 | 2609 | pos, pos_byte); |
080c45fd RS |
2610 | else |
2611 | middle = Qnil; | |
2612 | ||
2613 | newtext = concat2 (accum, middle); | |
2614 | } | |
2615 | ||
636a5e28 | 2616 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2617 | if (case_action == all_caps) |
2618 | newtext = Fupcase (newtext); | |
2619 | else if (case_action == cap_initial) | |
2b2eead9 | 2620 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2621 | |
2622 | return concat3 (before, newtext, after); | |
2623 | } | |
2624 | ||
09c4719e | 2625 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2626 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2627 | opoint = PT - ZV; |
9160906f RS |
2628 | else if (PT > search_regs.start[sub]) |
2629 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2630 | else |
2631 | opoint = PT; | |
2632 | ||
886ed6ec RS |
2633 | /* If we want non-literal replacement, |
2634 | perform substitution on the replacement string. */ | |
2635 | if (NILP (literal)) | |
ca1d1d23 | 2636 | { |
d5db4077 | 2637 | int length = SBYTES (newtext); |
68e69fbd RS |
2638 | unsigned char *substed; |
2639 | int substed_alloc_size, substed_len; | |
3bc25e52 KH |
2640 | int buf_multibyte = !NILP (current_buffer->enable_multibyte_characters); |
2641 | int str_multibyte = STRING_MULTIBYTE (newtext); | |
2642 | Lisp_Object rev_tbl; | |
886ed6ec | 2643 | int really_changed = 0; |
3bc25e52 | 2644 | |
8f924df7 | 2645 | rev_tbl = Qnil; |
ac3b28b1 | 2646 | |
68e69fbd RS |
2647 | substed_alloc_size = length * 2 + 100; |
2648 | substed = (unsigned char *) xmalloc (substed_alloc_size + 1); | |
2649 | substed_len = 0; | |
2650 | ||
3bc25e52 KH |
2651 | /* Go thru NEWTEXT, producing the actual text to insert in |
2652 | SUBSTED while adjusting multibyteness to that of the current | |
2653 | buffer. */ | |
ca1d1d23 | 2654 | |
ac3b28b1 | 2655 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2656 | { |
68e69fbd | 2657 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
f8ce8a0d GM |
2658 | unsigned char *add_stuff = NULL; |
2659 | int add_len = 0; | |
68e69fbd | 2660 | int idx = -1; |
9a76659d | 2661 | |
3bc25e52 KH |
2662 | if (str_multibyte) |
2663 | { | |
eb99a8dd | 2664 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 KH |
2665 | if (!buf_multibyte) |
2666 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2667 | } | |
2668 | else | |
2669 | { | |
2670 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2671 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2672 | if (buf_multibyte) |
4c0354d7 | 2673 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 | 2674 | } |
ac3b28b1 | 2675 | |
68e69fbd RS |
2676 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2677 | or set IDX to a match index, which means put that part | |
2678 | of the buffer text into SUBSTED. */ | |
2679 | ||
ca1d1d23 JB |
2680 | if (c == '\\') |
2681 | { | |
886ed6ec RS |
2682 | really_changed = 1; |
2683 | ||
3bc25e52 KH |
2684 | if (str_multibyte) |
2685 | { | |
eb99a8dd KH |
2686 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2687 | pos, pos_byte); | |
071ce769 | 2688 | if (!buf_multibyte && !ASCII_CHAR_P (c)) |
3bc25e52 KH |
2689 | c = multibyte_char_to_unibyte (c, rev_tbl); |
2690 | } | |
2691 | else | |
2692 | { | |
d5db4077 | 2693 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2694 | if (buf_multibyte) |
4c0354d7 | 2695 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 KH |
2696 | } |
2697 | ||
ca1d1d23 | 2698 | if (c == '&') |
68e69fbd | 2699 | idx = sub; |
78445046 | 2700 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2701 | { |
2702 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2703 | idx = c - '0'; |
ca1d1d23 | 2704 | } |
636a5e28 | 2705 | else if (c == '\\') |
68e69fbd | 2706 | add_len = 1, add_stuff = "\\"; |
636a5e28 | 2707 | else |
3bc25e52 KH |
2708 | { |
2709 | xfree (substed); | |
2710 | error ("Invalid use of `\\' in replacement text"); | |
2711 | } | |
ca1d1d23 JB |
2712 | } |
2713 | else | |
68e69fbd RS |
2714 | { |
2715 | add_len = CHAR_STRING (c, str); | |
2716 | add_stuff = str; | |
2717 | } | |
2718 | ||
2719 | /* If we want to copy part of a previous match, | |
2720 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2721 | if (idx >= 0) | |
2722 | { | |
e7deaab0 | 2723 | EMACS_INT begbyte = CHAR_TO_BYTE (search_regs.start[idx]); |
68e69fbd RS |
2724 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; |
2725 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2726 | move_gap (search_regs.start[idx]); | |
2727 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2728 | } | |
2729 | ||
2730 | /* Now the stuff we want to add to SUBSTED | |
2731 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2732 | ||
2733 | /* Make sure SUBSTED is big enough. */ | |
2734 | if (substed_len + add_len >= substed_alloc_size) | |
2735 | { | |
2736 | substed_alloc_size = substed_len + add_len + 500; | |
2737 | substed = (unsigned char *) xrealloc (substed, | |
2738 | substed_alloc_size + 1); | |
2739 | } | |
2740 | ||
2741 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2742 | if (add_stuff) |
2743 | { | |
72af86bd | 2744 | memcpy (substed + substed_len, add_stuff, add_len); |
f8ce8a0d GM |
2745 | substed_len += add_len; |
2746 | } | |
ca1d1d23 | 2747 | } |
68e69fbd | 2748 | |
886ed6ec | 2749 | if (really_changed) |
76fc1ea2 KH |
2750 | { |
2751 | if (buf_multibyte) | |
2752 | { | |
2753 | int nchars = multibyte_chars_in_text (substed, substed_len); | |
68e69fbd | 2754 | |
76fc1ea2 KH |
2755 | newtext = make_multibyte_string (substed, nchars, substed_len); |
2756 | } | |
2757 | else | |
2758 | newtext = make_unibyte_string (substed, substed_len); | |
2759 | } | |
68e69fbd | 2760 | xfree (substed); |
ca1d1d23 JB |
2761 | } |
2762 | ||
886ed6ec RS |
2763 | /* Replace the old text with the new in the cleanest possible way. */ |
2764 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2765 | newtext, 1, 0, 1); | |
d5db4077 | 2766 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2767 | |
2768 | if (case_action == all_caps) | |
886ed6ec RS |
2769 | Fupcase_region (make_number (search_regs.start[sub]), |
2770 | make_number (newpoint)); | |
ca1d1d23 | 2771 | else if (case_action == cap_initial) |
886ed6ec RS |
2772 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2773 | make_number (newpoint)); | |
3e18eecf | 2774 | |
98e942e0 RS |
2775 | /* Adjust search data for this change. */ |
2776 | { | |
e7deaab0 AS |
2777 | EMACS_INT oldend = search_regs.end[sub]; |
2778 | EMACS_INT oldstart = search_regs.start[sub]; | |
2779 | EMACS_INT change = newpoint - search_regs.end[sub]; | |
98e942e0 RS |
2780 | int i; |
2781 | ||
2782 | for (i = 0; i < search_regs.num_regs; i++) | |
2783 | { | |
41c01205 | 2784 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2785 | search_regs.start[i] += change; |
41c01205 DK |
2786 | else if (search_regs.start[i] > oldstart) |
2787 | search_regs.start[i] = oldstart; | |
2788 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2789 | search_regs.end[i] += change; |
41c01205 DK |
2790 | else if (search_regs.end[i] > oldstart) |
2791 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2792 | } |
2793 | } | |
2794 | ||
b0eba991 | 2795 | /* Put point back where it was in the text. */ |
8d808a65 | 2796 | if (opoint <= 0) |
fa8ed3e0 | 2797 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2798 | else |
fa8ed3e0 | 2799 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2800 | |
2801 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2802 | move_if_not_intangible (newpoint); |
177c0ea7 | 2803 | |
ca1d1d23 JB |
2804 | return Qnil; |
2805 | } | |
2806 | \f | |
2807 | static Lisp_Object | |
971de7fb | 2808 | match_limit (Lisp_Object num, int beginningp) |
ca1d1d23 JB |
2809 | { |
2810 | register int n; | |
2811 | ||
b7826503 | 2812 | CHECK_NUMBER (num); |
ca1d1d23 | 2813 | n = XINT (num); |
f90a5bf5 | 2814 | if (n < 0) |
bd2cbd56 | 2815 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2816 | if (search_regs.num_regs <= 0) |
2817 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2818 | if (n >= search_regs.num_regs |
4746118a | 2819 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2820 | return Qnil; |
2821 | return (make_number ((beginningp) ? search_regs.start[n] | |
2822 | : search_regs.end[n])); | |
2823 | } | |
2824 | ||
2825 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
8c1a1077 PJ |
2826 | doc: /* Return position of start of text matched by last search. |
2827 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2828 | regexp. | |
2829 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2830 | SUBEXP pairs. | |
2831 | Zero means the entire text matched by the whole regexp or whole string. */) | |
5842a27b | 2832 | (Lisp_Object subexp) |
ca1d1d23 | 2833 | { |
5806161b | 2834 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2835 | } |
2836 | ||
2837 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
8c1a1077 PJ |
2838 | doc: /* Return position of end of text matched by last search. |
2839 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2840 | regexp. | |
2841 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2842 | SUBEXP pairs. | |
2843 | Zero means the entire text matched by the whole regexp or whole string. */) | |
5842a27b | 2844 | (Lisp_Object subexp) |
ca1d1d23 | 2845 | { |
5806161b | 2846 | return match_limit (subexp, 0); |
177c0ea7 | 2847 | } |
ca1d1d23 | 2848 | |
abd0071c | 2849 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 3, 0, |
8c1a1077 PJ |
2850 | doc: /* Return a list containing all info on what the last search matched. |
2851 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2852 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2853 | if the last match was on a buffer; integers or nil if a string was matched. | |
febb8aba | 2854 | Use `set-match-data' to reinstate the data in this list. |
8c1a1077 | 2855 | |
41c01205 DK |
2856 | If INTEGERS (the optional first argument) is non-nil, always use |
2857 | integers \(rather than markers) to represent buffer positions. In | |
2858 | this case, and if the last match was in a buffer, the buffer will get | |
2859 | stored as one additional element at the end of the list. | |
2860 | ||
abd0071c KS |
2861 | If REUSE is a list, reuse it as part of the value. If REUSE is long |
2862 | enough to hold all the values, and if INTEGERS is non-nil, no consing | |
2863 | is done. | |
2864 | ||
2865 | If optional third arg RESEAT is non-nil, any previous markers on the | |
2866 | REUSE list will be modified to point to nowhere. | |
2867 | ||
140a6b7e | 2868 | Return value is undefined if the last search failed. */) |
5842a27b | 2869 | (Lisp_Object integers, Lisp_Object reuse, Lisp_Object reseat) |
ca1d1d23 | 2870 | { |
56256c2a | 2871 | Lisp_Object tail, prev; |
4746118a | 2872 | Lisp_Object *data; |
ca1d1d23 JB |
2873 | int i, len; |
2874 | ||
abd0071c KS |
2875 | if (!NILP (reseat)) |
2876 | for (tail = reuse; CONSP (tail); tail = XCDR (tail)) | |
2877 | if (MARKERP (XCAR (tail))) | |
2878 | { | |
51f10faa | 2879 | unchain_marker (XMARKER (XCAR (tail))); |
abd0071c KS |
2880 | XSETCAR (tail, Qnil); |
2881 | } | |
2882 | ||
daa37602 | 2883 | if (NILP (last_thing_searched)) |
c36bcf1b | 2884 | return Qnil; |
daa37602 | 2885 | |
6bbd7a29 GM |
2886 | prev = Qnil; |
2887 | ||
41c01205 | 2888 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs + 1) |
4746118a JB |
2889 | * sizeof (Lisp_Object)); |
2890 | ||
41c01205 | 2891 | len = 0; |
4746118a | 2892 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2893 | { |
2894 | int start = search_regs.start[i]; | |
2895 | if (start >= 0) | |
2896 | { | |
56256c2a RS |
2897 | if (EQ (last_thing_searched, Qt) |
2898 | || ! NILP (integers)) | |
ca1d1d23 | 2899 | { |
c235cce7 KH |
2900 | XSETFASTINT (data[2 * i], start); |
2901 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2902 | } |
0ed62dc7 | 2903 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2904 | { |
2905 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2906 | Fset_marker (data[2 * i], |
2907 | make_number (start), | |
2908 | last_thing_searched); | |
ca1d1d23 JB |
2909 | data[2 * i + 1] = Fmake_marker (); |
2910 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2911 | make_number (search_regs.end[i]), |
daa37602 | 2912 | last_thing_searched); |
ca1d1d23 | 2913 | } |
daa37602 JB |
2914 | else |
2915 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2916 | abort (); | |
2917 | ||
abd0071c | 2918 | len = 2 * i + 2; |
ca1d1d23 JB |
2919 | } |
2920 | else | |
abd0071c | 2921 | data[2 * i] = data[2 * i + 1] = Qnil; |
ca1d1d23 | 2922 | } |
56256c2a | 2923 | |
bd2cbd56 | 2924 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2925 | { |
bd2cbd56 | 2926 | data[len] = last_thing_searched; |
41c01205 DK |
2927 | len++; |
2928 | } | |
2929 | ||
56256c2a RS |
2930 | /* If REUSE is not usable, cons up the values and return them. */ |
2931 | if (! CONSP (reuse)) | |
41c01205 | 2932 | return Flist (len, data); |
56256c2a RS |
2933 | |
2934 | /* If REUSE is a list, store as many value elements as will fit | |
2935 | into the elements of REUSE. */ | |
2936 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2937 | i++, tail = XCDR (tail)) |
56256c2a | 2938 | { |
41c01205 | 2939 | if (i < len) |
f3fbd155 | 2940 | XSETCAR (tail, data[i]); |
56256c2a | 2941 | else |
f3fbd155 | 2942 | XSETCAR (tail, Qnil); |
56256c2a RS |
2943 | prev = tail; |
2944 | } | |
2945 | ||
2946 | /* If we couldn't fit all value elements into REUSE, | |
2947 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2948 | if (i < len) |
2949 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2950 | |
2951 | return reuse; | |
ca1d1d23 JB |
2952 | } |
2953 | ||
b51d6c92 SM |
2954 | /* We used to have an internal use variant of `reseat' described as: |
2955 | ||
2956 | If RESEAT is `evaporate', put the markers back on the free list | |
2957 | immediately. No other references to the markers must exist in this | |
2958 | case, so it is used only internally on the unwind stack and | |
2959 | save-match-data from Lisp. | |
2960 | ||
2961 | But it was ill-conceived: those supposedly-internal markers get exposed via | |
2962 | the undo-list, so freeing them here is unsafe. */ | |
ca1d1d23 | 2963 | |
abd0071c | 2964 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 2, 0, |
8c1a1077 | 2965 | doc: /* Set internal data on last search match from elements of LIST. |
abd0071c KS |
2966 | LIST should have been created by calling `match-data' previously. |
2967 | ||
51f10faa | 2968 | If optional arg RESEAT is non-nil, make markers on LIST point nowhere. */) |
5842a27b | 2969 | (register Lisp_Object list, Lisp_Object reseat) |
ca1d1d23 JB |
2970 | { |
2971 | register int i; | |
2972 | register Lisp_Object marker; | |
2973 | ||
7074fde6 FP |
2974 | if (running_asynch_code) |
2975 | save_search_regs (); | |
2976 | ||
29100cea | 2977 | CHECK_LIST (list); |
ca1d1d23 | 2978 | |
41c01205 DK |
2979 | /* Unless we find a marker with a buffer or an explicit buffer |
2980 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
2981 | last_thing_searched = Qt; |
2982 | ||
4746118a JB |
2983 | /* Allocate registers if they don't already exist. */ |
2984 | { | |
d084e942 | 2985 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2986 | |
2987 | if (length > search_regs.num_regs) | |
2988 | { | |
1113d9db JB |
2989 | if (search_regs.num_regs == 0) |
2990 | { | |
2991 | search_regs.start | |
2992 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2993 | search_regs.end | |
2994 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2995 | } | |
4746118a | 2996 | else |
1113d9db JB |
2997 | { |
2998 | search_regs.start | |
2999 | = (regoff_t *) xrealloc (search_regs.start, | |
3000 | length * sizeof (regoff_t)); | |
3001 | search_regs.end | |
3002 | = (regoff_t *) xrealloc (search_regs.end, | |
3003 | length * sizeof (regoff_t)); | |
3004 | } | |
4746118a | 3005 | |
e62371e9 KH |
3006 | for (i = search_regs.num_regs; i < length; i++) |
3007 | search_regs.start[i] = -1; | |
3008 | ||
487282dc | 3009 | search_regs.num_regs = length; |
4746118a | 3010 | } |
ca1d1d23 | 3011 | |
abd0071c | 3012 | for (i = 0; CONSP (list); i++) |
41c01205 | 3013 | { |
abd0071c | 3014 | marker = XCAR (list); |
bd2cbd56 | 3015 | if (BUFFERP (marker)) |
c3762cbd | 3016 | { |
bd2cbd56 | 3017 | last_thing_searched = marker; |
c3762cbd DK |
3018 | break; |
3019 | } | |
3020 | if (i >= length) | |
3021 | break; | |
41c01205 DK |
3022 | if (NILP (marker)) |
3023 | { | |
3024 | search_regs.start[i] = -1; | |
abd0071c | 3025 | list = XCDR (list); |
41c01205 DK |
3026 | } |
3027 | else | |
3028 | { | |
e7deaab0 | 3029 | EMACS_INT from; |
abd0071c | 3030 | Lisp_Object m; |
e2811828 | 3031 | |
abd0071c | 3032 | m = marker; |
41c01205 DK |
3033 | if (MARKERP (marker)) |
3034 | { | |
3035 | if (XMARKER (marker)->buffer == 0) | |
3036 | XSETFASTINT (marker, 0); | |
3037 | else | |
3038 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
3039 | } | |
e2811828 | 3040 | |
41c01205 DK |
3041 | CHECK_NUMBER_COERCE_MARKER (marker); |
3042 | from = XINT (marker); | |
e2811828 | 3043 | |
abd0071c KS |
3044 | if (!NILP (reseat) && MARKERP (m)) |
3045 | { | |
b51d6c92 | 3046 | unchain_marker (XMARKER (m)); |
9ad54a7e | 3047 | XSETCAR (list, Qnil); |
abd0071c KS |
3048 | } |
3049 | ||
3050 | if ((list = XCDR (list), !CONSP (list))) | |
3051 | break; | |
3052 | ||
3053 | m = marker = XCAR (list); | |
3054 | ||
41c01205 DK |
3055 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
3056 | XSETFASTINT (marker, 0); | |
e2811828 | 3057 | |
41c01205 DK |
3058 | CHECK_NUMBER_COERCE_MARKER (marker); |
3059 | search_regs.start[i] = from; | |
3060 | search_regs.end[i] = XINT (marker); | |
abd0071c KS |
3061 | |
3062 | if (!NILP (reseat) && MARKERP (m)) | |
3063 | { | |
b51d6c92 | 3064 | unchain_marker (XMARKER (m)); |
9ad54a7e | 3065 | XSETCAR (list, Qnil); |
abd0071c | 3066 | } |
41c01205 | 3067 | } |
abd0071c | 3068 | list = XCDR (list); |
41c01205 | 3069 | } |
ca1d1d23 | 3070 | |
41c01205 DK |
3071 | for (; i < search_regs.num_regs; i++) |
3072 | search_regs.start[i] = -1; | |
3073 | } | |
ca1d1d23 | 3074 | |
177c0ea7 | 3075 | return Qnil; |
ca1d1d23 JB |
3076 | } |
3077 | ||
7074fde6 FP |
3078 | /* If non-zero the match data have been saved in saved_search_regs |
3079 | during the execution of a sentinel or filter. */ | |
75ebf74b | 3080 | static int search_regs_saved; |
7074fde6 | 3081 | static struct re_registers saved_search_regs; |
41c01205 | 3082 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
3083 | |
3084 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
3085 | if asynchronous code (filter or sentinel) is running. */ | |
3086 | static void | |
971de7fb | 3087 | save_search_regs (void) |
7074fde6 FP |
3088 | { |
3089 | if (!search_regs_saved) | |
3090 | { | |
3091 | saved_search_regs.num_regs = search_regs.num_regs; | |
3092 | saved_search_regs.start = search_regs.start; | |
3093 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
3094 | saved_last_thing_searched = last_thing_searched; |
3095 | last_thing_searched = Qnil; | |
7074fde6 | 3096 | search_regs.num_regs = 0; |
2d4a771a RS |
3097 | search_regs.start = 0; |
3098 | search_regs.end = 0; | |
7074fde6 FP |
3099 | |
3100 | search_regs_saved = 1; | |
3101 | } | |
3102 | } | |
3103 | ||
3104 | /* Called upon exit from filters and sentinels. */ | |
3105 | void | |
971de7fb | 3106 | restore_search_regs (void) |
7074fde6 FP |
3107 | { |
3108 | if (search_regs_saved) | |
3109 | { | |
3110 | if (search_regs.num_regs > 0) | |
3111 | { | |
3112 | xfree (search_regs.start); | |
3113 | xfree (search_regs.end); | |
3114 | } | |
3115 | search_regs.num_regs = saved_search_regs.num_regs; | |
3116 | search_regs.start = saved_search_regs.start; | |
3117 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
3118 | last_thing_searched = saved_last_thing_searched; |
3119 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
3120 | search_regs_saved = 0; |
3121 | } | |
3122 | } | |
3123 | ||
abd0071c | 3124 | static Lisp_Object |
971de7fb | 3125 | unwind_set_match_data (Lisp_Object list) |
abd0071c | 3126 | { |
b51d6c92 SM |
3127 | /* It is NOT ALWAYS safe to free (evaporate) the markers immediately. */ |
3128 | return Fset_match_data (list, Qt); | |
abd0071c KS |
3129 | } |
3130 | ||
3131 | /* Called to unwind protect the match data. */ | |
3132 | void | |
971de7fb | 3133 | record_unwind_save_match_data (void) |
abd0071c KS |
3134 | { |
3135 | record_unwind_protect (unwind_set_match_data, | |
3136 | Fmatch_data (Qnil, Qnil, Qnil)); | |
3137 | } | |
3138 | ||
ca1d1d23 JB |
3139 | /* Quote a string to inactivate reg-expr chars */ |
3140 | ||
3141 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 | 3142 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
5842a27b | 3143 | (Lisp_Object string) |
ca1d1d23 JB |
3144 | { |
3145 | register unsigned char *in, *out, *end; | |
3146 | register unsigned char *temp; | |
0c8533c6 | 3147 | int backslashes_added = 0; |
ca1d1d23 | 3148 | |
b7826503 | 3149 | CHECK_STRING (string); |
ca1d1d23 | 3150 | |
d5db4077 | 3151 | temp = (unsigned char *) alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
3152 | |
3153 | /* Now copy the data into the new string, inserting escapes. */ | |
3154 | ||
d5db4077 KR |
3155 | in = SDATA (string); |
3156 | end = in + SBYTES (string); | |
177c0ea7 | 3157 | out = temp; |
ca1d1d23 JB |
3158 | |
3159 | for (; in != end; in++) | |
3160 | { | |
66bc6082 | 3161 | if (*in == '[' |
ca1d1d23 JB |
3162 | || *in == '*' || *in == '.' || *in == '\\' |
3163 | || *in == '?' || *in == '+' | |
3164 | || *in == '^' || *in == '$') | |
0c8533c6 | 3165 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
3166 | *out++ = *in; |
3167 | } | |
3168 | ||
3f8100f1 | 3169 | return make_specified_string (temp, |
d5db4077 | 3170 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
3171 | out - temp, |
3172 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 3173 | } |
177c0ea7 | 3174 | \f |
dfcf069d | 3175 | void |
971de7fb | 3176 | syms_of_search (void) |
ca1d1d23 JB |
3177 | { |
3178 | register int i; | |
3179 | ||
487282dc KH |
3180 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
3181 | { | |
3182 | searchbufs[i].buf.allocated = 100; | |
b23c0a83 | 3183 | searchbufs[i].buf.buffer = (unsigned char *) xmalloc (100); |
487282dc KH |
3184 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
3185 | searchbufs[i].regexp = Qnil; | |
ecdb561e | 3186 | searchbufs[i].whitespace_regexp = Qnil; |
b69e3c18 | 3187 | searchbufs[i].syntax_table = Qnil; |
487282dc | 3188 | staticpro (&searchbufs[i].regexp); |
aa77b5ce | 3189 | staticpro (&searchbufs[i].whitespace_regexp); |
b69e3c18 | 3190 | staticpro (&searchbufs[i].syntax_table); |
487282dc KH |
3191 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); |
3192 | } | |
3193 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 | 3194 | |
d67b4f80 | 3195 | Qsearch_failed = intern_c_string ("search-failed"); |
ca1d1d23 | 3196 | staticpro (&Qsearch_failed); |
d67b4f80 | 3197 | Qinvalid_regexp = intern_c_string ("invalid-regexp"); |
ca1d1d23 JB |
3198 | staticpro (&Qinvalid_regexp); |
3199 | ||
3200 | Fput (Qsearch_failed, Qerror_conditions, | |
d67b4f80 | 3201 | pure_cons (Qsearch_failed, pure_cons (Qerror, Qnil))); |
ca1d1d23 | 3202 | Fput (Qsearch_failed, Qerror_message, |
d67b4f80 | 3203 | make_pure_c_string ("Search failed")); |
ca1d1d23 JB |
3204 | |
3205 | Fput (Qinvalid_regexp, Qerror_conditions, | |
d67b4f80 | 3206 | pure_cons (Qinvalid_regexp, pure_cons (Qerror, Qnil))); |
ca1d1d23 | 3207 | Fput (Qinvalid_regexp, Qerror_message, |
d67b4f80 | 3208 | make_pure_c_string ("Invalid regexp")); |
ca1d1d23 | 3209 | |
daa37602 JB |
3210 | last_thing_searched = Qnil; |
3211 | staticpro (&last_thing_searched); | |
3212 | ||
0f6af254 DK |
3213 | saved_last_thing_searched = Qnil; |
3214 | staticpro (&saved_last_thing_searched); | |
3215 | ||
41a33295 | 3216 | DEFVAR_LISP ("search-spaces-regexp", &Vsearch_spaces_regexp, |
e2811828 | 3217 | doc: /* Regexp to substitute for bunches of spaces in regexp search. |
f31a9a68 RS |
3218 | Some commands use this for user-specified regexps. |
3219 | Spaces that occur inside character classes or repetition operators | |
3220 | or other such regexp constructs are not replaced with this. | |
3221 | A value of nil (which is the normal value) means treat spaces literally. */); | |
41a33295 | 3222 | Vsearch_spaces_regexp = Qnil; |
f31a9a68 | 3223 | |
ef887810 RS |
3224 | DEFVAR_LISP ("inhibit-changing-match-data", &Vinhibit_changing_match_data, |
3225 | doc: /* Internal use only. | |
39d0bf74 RS |
3226 | If non-nil, the primitive searching and matching functions |
3227 | such as `looking-at', `string-match', `re-search-forward', etc., | |
3228 | do not set the match data. The proper way to use this variable | |
3229 | is to bind it with `let' around a small expression. */); | |
ef887810 RS |
3230 | Vinhibit_changing_match_data = Qnil; |
3231 | ||
ca1d1d23 | 3232 | defsubr (&Slooking_at); |
b819a390 RS |
3233 | defsubr (&Sposix_looking_at); |
3234 | defsubr (&Sstring_match); | |
3235 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
3236 | defsubr (&Ssearch_forward); |
3237 | defsubr (&Ssearch_backward); | |
3238 | defsubr (&Sword_search_forward); | |
3239 | defsubr (&Sword_search_backward); | |
c6e0e1cc CY |
3240 | defsubr (&Sword_search_forward_lax); |
3241 | defsubr (&Sword_search_backward_lax); | |
ca1d1d23 JB |
3242 | defsubr (&Sre_search_forward); |
3243 | defsubr (&Sre_search_backward); | |
b819a390 RS |
3244 | defsubr (&Sposix_search_forward); |
3245 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
3246 | defsubr (&Sreplace_match); |
3247 | defsubr (&Smatch_beginning); | |
3248 | defsubr (&Smatch_end); | |
3249 | defsubr (&Smatch_data); | |
3f1c005b | 3250 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
3251 | defsubr (&Sregexp_quote); |
3252 | } | |
76fc1ea2 KH |
3253 | |
3254 | /* arch-tag: a6059d79-0552-4f14-a2cb-d379a4e3c78f | |
3255 | (do not change this comment) */ |