Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
4fb9a543 GM |
2 | |
3 | Copyright (C) 1985-1987, 1993-1994, 1997-1999, 2001-2012 | |
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 | 56 | /* The instances of that struct. */ |
989b29ad | 57 | static struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; |
ca1d1d23 | 58 | |
487282dc | 59 | /* The head of the linked list; points to the most recently used buffer. */ |
989b29ad | 60 | static 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 | |
02b16839 | 87 | /* Error condition signaled when regexp compile_pattern fails. */ |
955cbe7b | 88 | static Lisp_Object Qinvalid_regexp; |
ca1d1d23 | 89 | |
02b16839 | 90 | /* Error condition used for failing searches. */ |
955cbe7b | 91 | static Lisp_Object Qsearch_failed; |
06f77a8a | 92 | |
d311d28c | 93 | static void set_search_regs (ptrdiff_t, ptrdiff_t); |
f57e2426 | 94 | static void save_search_regs (void); |
d311d28c PE |
95 | static EMACS_INT simple_search (EMACS_INT, unsigned char *, ptrdiff_t, |
96 | ptrdiff_t, Lisp_Object, ptrdiff_t, ptrdiff_t, | |
97 | ptrdiff_t, ptrdiff_t); | |
98 | static EMACS_INT boyer_moore (EMACS_INT, unsigned char *, ptrdiff_t, | |
99 | Lisp_Object, Lisp_Object, ptrdiff_t, | |
100 | ptrdiff_t, int); | |
101 | static EMACS_INT search_buffer (Lisp_Object, ptrdiff_t, ptrdiff_t, | |
102 | ptrdiff_t, ptrdiff_t, EMACS_INT, int, | |
f57e2426 | 103 | Lisp_Object, Lisp_Object, int); |
971de7fb | 104 | static void matcher_overflow (void) NO_RETURN; |
b819a390 | 105 | |
ca1d1d23 | 106 | static void |
971de7fb | 107 | matcher_overflow (void) |
ca1d1d23 JB |
108 | { |
109 | error ("Stack overflow in regexp matcher"); | |
110 | } | |
111 | ||
b819a390 RS |
112 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
113 | PATTERN is the pattern to compile. | |
114 | CP is the place to put the result. | |
facdc750 | 115 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 | 116 | POSIX is nonzero if we want full backtracking (POSIX style) |
5679531d | 117 | for this pattern. 0 means backtrack only enough to get a valid match. |
bbc73b48 KH |
118 | |
119 | The behavior also depends on Vsearch_spaces_regexp. */ | |
ca1d1d23 | 120 | |
487282dc | 121 | static void |
cfe0661d | 122 | compile_pattern_1 (struct regexp_cache *cp, Lisp_Object pattern, Lisp_Object translate, int posix) |
ca1d1d23 | 123 | { |
d451e4db | 124 | char *val; |
b819a390 | 125 | reg_syntax_t old; |
ca1d1d23 | 126 | |
487282dc | 127 | cp->regexp = Qnil; |
59fab369 | 128 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 129 | cp->posix = posix; |
93daa011 | 130 | cp->buf.multibyte = STRING_MULTIBYTE (pattern); |
89ad6492 | 131 | cp->buf.charset_unibyte = charset_unibyte; |
d8c85250 CY |
132 | if (STRINGP (Vsearch_spaces_regexp)) |
133 | cp->whitespace_regexp = Vsearch_spaces_regexp; | |
134 | else | |
135 | cp->whitespace_regexp = Qnil; | |
136 | ||
e7b2dd2e RS |
137 | /* rms: I think BLOCK_INPUT is not needed here any more, |
138 | because regex.c defines malloc to call xmalloc. | |
139 | Using BLOCK_INPUT here means the debugger won't run if an error occurs. | |
140 | So let's turn it off. */ | |
141 | /* BLOCK_INPUT; */ | |
fb4a568d | 142 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 143 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
d8c85250 CY |
144 | |
145 | if (STRINGP (Vsearch_spaces_regexp)) | |
42a5b22f | 146 | re_set_whitespace_regexp (SSDATA (Vsearch_spaces_regexp)); |
d8c85250 CY |
147 | else |
148 | re_set_whitespace_regexp (NULL); | |
bbc73b48 | 149 | |
51b59d79 | 150 | val = (char *) re_compile_pattern (SSDATA (pattern), |
8f924df7 | 151 | SBYTES (pattern), &cp->buf); |
bbc73b48 | 152 | |
58e95211 SM |
153 | /* If the compiled pattern hard codes some of the contents of the |
154 | syntax-table, it can only be reused with *this* syntax table. */ | |
4b4deea2 | 155 | cp->syntax_table = cp->buf.used_syntax ? BVAR (current_buffer, syntax_table) : Qt; |
58e95211 | 156 | |
bbc73b48 KH |
157 | re_set_whitespace_regexp (NULL); |
158 | ||
b819a390 | 159 | re_set_syntax (old); |
e7b2dd2e | 160 | /* UNBLOCK_INPUT; */ |
ca1d1d23 | 161 | if (val) |
06f77a8a | 162 | xsignal1 (Qinvalid_regexp, build_string (val)); |
1113d9db | 163 | |
487282dc | 164 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
165 | } |
166 | ||
6efc7887 RS |
167 | /* Shrink each compiled regexp buffer in the cache |
168 | to the size actually used right now. | |
169 | This is called from garbage collection. */ | |
170 | ||
171 | void | |
971de7fb | 172 | shrink_regexp_cache (void) |
6efc7887 | 173 | { |
a968f437 | 174 | struct regexp_cache *cp; |
6efc7887 RS |
175 | |
176 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
177 | { | |
178 | cp->buf.allocated = cp->buf.used; | |
179 | cp->buf.buffer | |
b23c0a83 | 180 | = (unsigned char *) xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
181 | } |
182 | } | |
183 | ||
54dd3310 SM |
184 | /* Clear the regexp cache w.r.t. a particular syntax table, |
185 | because it was changed. | |
e5b94d44 CY |
186 | There is no danger of memory leak here because re_compile_pattern |
187 | automagically manages the memory in each re_pattern_buffer struct, | |
188 | based on its `allocated' and `buffer' values. */ | |
189 | void | |
971de7fb | 190 | clear_regexp_cache (void) |
e5b94d44 CY |
191 | { |
192 | int i; | |
193 | ||
194 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) | |
920fd1fc EZ |
195 | /* It's tempting to compare with the syntax-table we've actually changed, |
196 | but it's not sufficient because char-table inheritance means that | |
54dd3310 SM |
197 | modifying one syntax-table can change others at the same time. */ |
198 | if (!EQ (searchbufs[i].syntax_table, Qt)) | |
199 | searchbufs[i].regexp = Qnil; | |
e5b94d44 CY |
200 | } |
201 | ||
487282dc | 202 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
203 | the cache. |
204 | PATTERN is the pattern to compile. | |
facdc750 | 205 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
206 | REGP is the structure that says where to store the "register" |
207 | values that will result from matching this pattern. | |
208 | If it is 0, we should compile the pattern not to record any | |
209 | subexpression bounds. | |
210 | POSIX is nonzero if we want full backtracking (POSIX style) | |
211 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
212 | |
213 | struct re_pattern_buffer * | |
971de7fb | 214 | compile_pattern (Lisp_Object pattern, struct re_registers *regp, Lisp_Object translate, int posix, int multibyte) |
487282dc KH |
215 | { |
216 | struct regexp_cache *cp, **cpp; | |
217 | ||
218 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
219 | { | |
220 | cp = *cpp; | |
f1b9c7c1 KR |
221 | /* Entries are initialized to nil, and may be set to nil by |
222 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
49a5f770 KR |
223 | string accessors in those cases. However, compile_pattern_1 |
224 | is only applied to the cache entry we pick here to reuse. So | |
225 | nil should never appear before a non-nil entry. */ | |
7c752c80 | 226 | if (NILP (cp->regexp)) |
f1b9c7c1 | 227 | goto compile_it; |
d5db4077 | 228 | if (SCHARS (cp->regexp) == SCHARS (pattern) |
cf69b13e | 229 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 230 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 231 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d | 232 | && cp->posix == posix |
58e95211 | 233 | && (EQ (cp->syntax_table, Qt) |
4b4deea2 | 234 | || EQ (cp->syntax_table, BVAR (current_buffer, syntax_table))) |
89ad6492 KH |
235 | && !NILP (Fequal (cp->whitespace_regexp, Vsearch_spaces_regexp)) |
236 | && cp->buf.charset_unibyte == charset_unibyte) | |
487282dc KH |
237 | break; |
238 | ||
f1b9c7c1 KR |
239 | /* If we're at the end of the cache, compile into the nil cell |
240 | we found, or the last (least recently used) cell with a | |
241 | string value. */ | |
487282dc KH |
242 | if (cp->next == 0) |
243 | { | |
f1b9c7c1 | 244 | compile_it: |
cfe0661d | 245 | compile_pattern_1 (cp, pattern, translate, posix); |
487282dc KH |
246 | break; |
247 | } | |
248 | } | |
249 | ||
250 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
251 | either because we found it in the cache or because we just compiled it. | |
252 | Move it to the front of the queue to mark it as most recently used. */ | |
253 | *cpp = cp->next; | |
254 | cp->next = searchbuf_head; | |
255 | searchbuf_head = cp; | |
1113d9db | 256 | |
6639708c RS |
257 | /* Advise the searching functions about the space we have allocated |
258 | for register data. */ | |
259 | if (regp) | |
260 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
261 | ||
78edd3b7 | 262 | /* The compiled pattern can be used both for multibyte and unibyte |
89ad6492 KH |
263 | target. But, we have to tell which the pattern is used for. */ |
264 | cp->buf.target_multibyte = multibyte; | |
265 | ||
487282dc | 266 | return &cp->buf; |
ca1d1d23 JB |
267 | } |
268 | ||
ca1d1d23 | 269 | \f |
b819a390 | 270 | static Lisp_Object |
971de7fb | 271 | looking_at_1 (Lisp_Object string, int posix) |
ca1d1d23 JB |
272 | { |
273 | Lisp_Object val; | |
274 | unsigned char *p1, *p2; | |
d311d28c PE |
275 | ptrdiff_t s1, s2; |
276 | register ptrdiff_t i; | |
487282dc | 277 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 278 | |
7074fde6 FP |
279 | if (running_asynch_code) |
280 | save_search_regs (); | |
281 | ||
910c747a | 282 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
4b4deea2 TT |
283 | XCHAR_TABLE (BVAR (current_buffer, case_canon_table))->extras[2] |
284 | = BVAR (current_buffer, case_eqv_table); | |
910c747a | 285 | |
b7826503 | 286 | CHECK_STRING (string); |
ef887810 RS |
287 | bufp = compile_pattern (string, |
288 | (NILP (Vinhibit_changing_match_data) | |
289 | ? &search_regs : NULL), | |
4b4deea2 TT |
290 | (!NILP (BVAR (current_buffer, case_fold_search)) |
291 | ? BVAR (current_buffer, case_canon_table) : Qnil), | |
0c8533c6 | 292 | posix, |
4b4deea2 | 293 | !NILP (BVAR (current_buffer, enable_multibyte_characters))); |
ca1d1d23 JB |
294 | |
295 | immediate_quit = 1; | |
296 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
297 | ||
298 | /* Get pointers and sizes of the two strings | |
299 | that make up the visible portion of the buffer. */ | |
300 | ||
301 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 302 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 303 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 304 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
305 | if (s1 < 0) |
306 | { | |
307 | p2 = p1; | |
fa8ed3e0 | 308 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
309 | s1 = 0; |
310 | } | |
311 | if (s2 < 0) | |
312 | { | |
fa8ed3e0 | 313 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
314 | s2 = 0; |
315 | } | |
8bb43c28 RS |
316 | |
317 | re_match_object = Qnil; | |
177c0ea7 | 318 | |
487282dc | 319 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
ef887810 RS |
320 | PT_BYTE - BEGV_BYTE, |
321 | (NILP (Vinhibit_changing_match_data) | |
322 | ? &search_regs : NULL), | |
fa8ed3e0 | 323 | ZV_BYTE - BEGV_BYTE); |
de182d70 | 324 | immediate_quit = 0; |
177c0ea7 | 325 | |
ca1d1d23 JB |
326 | if (i == -2) |
327 | matcher_overflow (); | |
328 | ||
329 | val = (0 <= i ? Qt : Qnil); | |
ef887810 | 330 | if (NILP (Vinhibit_changing_match_data) && i >= 0) |
fa8ed3e0 RS |
331 | for (i = 0; i < search_regs.num_regs; i++) |
332 | if (search_regs.start[i] >= 0) | |
333 | { | |
334 | search_regs.start[i] | |
335 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
336 | search_regs.end[i] | |
337 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
338 | } | |
ef887810 RS |
339 | |
340 | /* Set last_thing_searched only when match data is changed. */ | |
341 | if (NILP (Vinhibit_changing_match_data)) | |
342 | XSETBUFFER (last_thing_searched, current_buffer); | |
343 | ||
ca1d1d23 JB |
344 | return val; |
345 | } | |
346 | ||
b819a390 | 347 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
348 | doc: /* Return t if text after point matches regular expression REGEXP. |
349 | This function modifies the match data that `match-beginning', | |
350 | `match-end' and `match-data' access; save and restore the match | |
351 | data if you want to preserve them. */) | |
5842a27b | 352 | (Lisp_Object regexp) |
b819a390 | 353 | { |
94f94972 | 354 | return looking_at_1 (regexp, 0); |
b819a390 RS |
355 | } |
356 | ||
357 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
358 | doc: /* Return t if text after point matches regular expression REGEXP. |
359 | Find the longest match, in accord with Posix regular expression rules. | |
360 | This function modifies the match data that `match-beginning', | |
361 | `match-end' and `match-data' access; save and restore the match | |
362 | data if you want to preserve them. */) | |
5842a27b | 363 | (Lisp_Object regexp) |
b819a390 | 364 | { |
94f94972 | 365 | return looking_at_1 (regexp, 1); |
b819a390 RS |
366 | } |
367 | \f | |
368 | static Lisp_Object | |
971de7fb | 369 | string_match_1 (Lisp_Object regexp, Lisp_Object string, Lisp_Object start, int posix) |
ca1d1d23 | 370 | { |
d311d28c | 371 | ptrdiff_t val; |
487282dc | 372 | struct re_pattern_buffer *bufp; |
d311d28c PE |
373 | EMACS_INT pos; |
374 | ptrdiff_t pos_byte, i; | |
ca1d1d23 | 375 | |
7074fde6 FP |
376 | if (running_asynch_code) |
377 | save_search_regs (); | |
378 | ||
b7826503 PJ |
379 | CHECK_STRING (regexp); |
380 | CHECK_STRING (string); | |
ca1d1d23 JB |
381 | |
382 | if (NILP (start)) | |
0c8533c6 | 383 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
384 | else |
385 | { | |
d311d28c | 386 | ptrdiff_t len = SCHARS (string); |
ca1d1d23 | 387 | |
b7826503 | 388 | CHECK_NUMBER (start); |
0c8533c6 RS |
389 | pos = XINT (start); |
390 | if (pos < 0 && -pos <= len) | |
391 | pos = len + pos; | |
392 | else if (0 > pos || pos > len) | |
ca1d1d23 | 393 | args_out_of_range (string, start); |
0c8533c6 | 394 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
395 | } |
396 | ||
910c747a | 397 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
4b4deea2 TT |
398 | XCHAR_TABLE (BVAR (current_buffer, case_canon_table))->extras[2] |
399 | = BVAR (current_buffer, case_eqv_table); | |
910c747a | 400 | |
ef887810 RS |
401 | bufp = compile_pattern (regexp, |
402 | (NILP (Vinhibit_changing_match_data) | |
403 | ? &search_regs : NULL), | |
4b4deea2 TT |
404 | (!NILP (BVAR (current_buffer, case_fold_search)) |
405 | ? BVAR (current_buffer, case_canon_table) : Qnil), | |
0c8533c6 RS |
406 | posix, |
407 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 408 | immediate_quit = 1; |
8bb43c28 | 409 | re_match_object = string; |
177c0ea7 | 410 | |
51b59d79 | 411 | val = re_search (bufp, SSDATA (string), |
d5db4077 KR |
412 | SBYTES (string), pos_byte, |
413 | SBYTES (string) - pos_byte, | |
ef887810 RS |
414 | (NILP (Vinhibit_changing_match_data) |
415 | ? &search_regs : NULL)); | |
ca1d1d23 | 416 | immediate_quit = 0; |
ef887810 RS |
417 | |
418 | /* Set last_thing_searched only when match data is changed. */ | |
419 | if (NILP (Vinhibit_changing_match_data)) | |
420 | last_thing_searched = Qt; | |
421 | ||
ca1d1d23 JB |
422 | if (val == -2) |
423 | matcher_overflow (); | |
424 | if (val < 0) return Qnil; | |
0c8533c6 | 425 | |
ef887810 RS |
426 | if (NILP (Vinhibit_changing_match_data)) |
427 | for (i = 0; i < search_regs.num_regs; i++) | |
428 | if (search_regs.start[i] >= 0) | |
429 | { | |
430 | search_regs.start[i] | |
431 | = string_byte_to_char (string, search_regs.start[i]); | |
432 | search_regs.end[i] | |
433 | = string_byte_to_char (string, search_regs.end[i]); | |
434 | } | |
0c8533c6 RS |
435 | |
436 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 437 | } |
e59a8453 | 438 | |
a7ca3326 | 439 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 | 440 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
b85acc4b | 441 | Matching ignores case if `case-fold-search' is non-nil. |
8c1a1077 PJ |
442 | If third arg START is non-nil, start search at that index in STRING. |
443 | For index of first char beyond the match, do (match-end 0). | |
444 | `match-end' and `match-beginning' also give indices of substrings | |
2bd2f32d RS |
445 | matched by parenthesis constructs in the pattern. |
446 | ||
447 | You can use the function `match-string' to extract the substrings | |
448 | matched by the parenthesis constructions in REGEXP. */) | |
5842a27b | 449 | (Lisp_Object regexp, Lisp_Object string, Lisp_Object start) |
b819a390 RS |
450 | { |
451 | return string_match_1 (regexp, string, start, 0); | |
452 | } | |
453 | ||
454 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
455 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
456 | Find the longest match, in accord with Posix regular expression rules. | |
457 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
458 | If third arg START is non-nil, start search at that index in STRING. | |
459 | For index of first char beyond the match, do (match-end 0). | |
460 | `match-end' and `match-beginning' also give indices of substrings | |
461 | matched by parenthesis constructs in the pattern. */) | |
5842a27b | 462 | (Lisp_Object regexp, Lisp_Object string, Lisp_Object start) |
b819a390 RS |
463 | { |
464 | return string_match_1 (regexp, string, start, 1); | |
465 | } | |
466 | ||
e59a8453 RS |
467 | /* Match REGEXP against STRING, searching all of STRING, |
468 | and return the index of the match, or negative on failure. | |
469 | This does not clobber the match data. */ | |
470 | ||
d311d28c | 471 | ptrdiff_t |
971de7fb | 472 | fast_string_match (Lisp_Object regexp, Lisp_Object string) |
e59a8453 | 473 | { |
d311d28c | 474 | ptrdiff_t val; |
487282dc | 475 | struct re_pattern_buffer *bufp; |
e59a8453 | 476 | |
facdc750 RS |
477 | bufp = compile_pattern (regexp, 0, Qnil, |
478 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 479 | immediate_quit = 1; |
8bb43c28 | 480 | re_match_object = string; |
177c0ea7 | 481 | |
51b59d79 | 482 | val = re_search (bufp, SSDATA (string), |
d5db4077 KR |
483 | SBYTES (string), 0, |
484 | SBYTES (string), 0); | |
e59a8453 RS |
485 | immediate_quit = 0; |
486 | return val; | |
487 | } | |
5679531d KH |
488 | |
489 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
490 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
491 | This does not clobber the match data. |
492 | We assume that STRING contains single-byte characters. */ | |
5679531d | 493 | |
d311d28c | 494 | ptrdiff_t |
971de7fb | 495 | fast_c_string_match_ignore_case (Lisp_Object regexp, const char *string) |
5679531d | 496 | { |
d311d28c | 497 | ptrdiff_t val; |
5679531d | 498 | struct re_pattern_buffer *bufp; |
c098fdb8 | 499 | size_t len = strlen (string); |
5679531d | 500 | |
0c8533c6 | 501 | regexp = string_make_unibyte (regexp); |
b4577c63 | 502 | re_match_object = Qt; |
5679531d | 503 | bufp = compile_pattern (regexp, 0, |
0190922f | 504 | Vascii_canon_table, 0, |
f8bd51c4 | 505 | 0); |
5679531d KH |
506 | immediate_quit = 1; |
507 | val = re_search (bufp, string, len, 0, len, 0); | |
508 | immediate_quit = 0; | |
509 | return val; | |
510 | } | |
be5f4dfb KH |
511 | |
512 | /* Like fast_string_match but ignore case. */ | |
513 | ||
d311d28c | 514 | ptrdiff_t |
971de7fb | 515 | fast_string_match_ignore_case (Lisp_Object regexp, Lisp_Object string) |
be5f4dfb | 516 | { |
d311d28c | 517 | ptrdiff_t val; |
be5f4dfb KH |
518 | struct re_pattern_buffer *bufp; |
519 | ||
0190922f | 520 | bufp = compile_pattern (regexp, 0, Vascii_canon_table, |
be5f4dfb KH |
521 | 0, STRING_MULTIBYTE (string)); |
522 | immediate_quit = 1; | |
523 | re_match_object = string; | |
524 | ||
51b59d79 | 525 | val = re_search (bufp, SSDATA (string), |
be5f4dfb KH |
526 | SBYTES (string), 0, |
527 | SBYTES (string), 0); | |
528 | immediate_quit = 0; | |
529 | return val; | |
530 | } | |
ca1d1d23 | 531 | \f |
a80ce213 | 532 | /* Match REGEXP against the characters after POS to LIMIT, and return |
8b13507a KH |
533 | the number of matched characters. If STRING is non-nil, match |
534 | against the characters in it. In that case, POS and LIMIT are | |
535 | indices into the string. This function doesn't modify the match | |
536 | data. */ | |
537 | ||
d311d28c PE |
538 | ptrdiff_t |
539 | fast_looking_at (Lisp_Object regexp, ptrdiff_t pos, ptrdiff_t pos_byte, ptrdiff_t limit, ptrdiff_t limit_byte, Lisp_Object string) | |
8b13507a KH |
540 | { |
541 | int multibyte; | |
542 | struct re_pattern_buffer *buf; | |
543 | unsigned char *p1, *p2; | |
d311d28c PE |
544 | ptrdiff_t s1, s2; |
545 | ptrdiff_t len; | |
78edd3b7 | 546 | |
8b13507a KH |
547 | if (STRINGP (string)) |
548 | { | |
549 | if (pos_byte < 0) | |
550 | pos_byte = string_char_to_byte (string, pos); | |
551 | if (limit_byte < 0) | |
552 | limit_byte = string_char_to_byte (string, limit); | |
553 | p1 = NULL; | |
554 | s1 = 0; | |
555 | p2 = SDATA (string); | |
556 | s2 = SBYTES (string); | |
557 | re_match_object = string; | |
558 | multibyte = STRING_MULTIBYTE (string); | |
559 | } | |
560 | else | |
561 | { | |
562 | if (pos_byte < 0) | |
563 | pos_byte = CHAR_TO_BYTE (pos); | |
564 | if (limit_byte < 0) | |
565 | limit_byte = CHAR_TO_BYTE (limit); | |
566 | pos_byte -= BEGV_BYTE; | |
567 | limit_byte -= BEGV_BYTE; | |
568 | p1 = BEGV_ADDR; | |
569 | s1 = GPT_BYTE - BEGV_BYTE; | |
570 | p2 = GAP_END_ADDR; | |
571 | s2 = ZV_BYTE - GPT_BYTE; | |
572 | if (s1 < 0) | |
573 | { | |
574 | p2 = p1; | |
575 | s2 = ZV_BYTE - BEGV_BYTE; | |
576 | s1 = 0; | |
577 | } | |
578 | if (s2 < 0) | |
579 | { | |
580 | s1 = ZV_BYTE - BEGV_BYTE; | |
581 | s2 = 0; | |
582 | } | |
583 | re_match_object = Qnil; | |
4b4deea2 | 584 | multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
8b13507a KH |
585 | } |
586 | ||
587 | buf = compile_pattern (regexp, 0, Qnil, 0, multibyte); | |
588 | immediate_quit = 1; | |
589 | len = re_match_2 (buf, (char *) p1, s1, (char *) p2, s2, | |
590 | pos_byte, NULL, limit_byte); | |
591 | immediate_quit = 0; | |
592 | ||
593 | return len; | |
594 | } | |
595 | ||
596 | \f | |
9169c321 JB |
597 | /* The newline cache: remembering which sections of text have no newlines. */ |
598 | ||
599 | /* If the user has requested newline caching, make sure it's on. | |
600 | Otherwise, make sure it's off. | |
601 | This is our cheezy way of associating an action with the change of | |
602 | state of a buffer-local variable. */ | |
603 | static void | |
971de7fb | 604 | newline_cache_on_off (struct buffer *buf) |
9169c321 | 605 | { |
4b4deea2 | 606 | if (NILP (BVAR (buf, cache_long_line_scans))) |
9169c321 JB |
607 | { |
608 | /* It should be off. */ | |
609 | if (buf->newline_cache) | |
610 | { | |
611 | free_region_cache (buf->newline_cache); | |
612 | buf->newline_cache = 0; | |
613 | } | |
614 | } | |
615 | else | |
616 | { | |
617 | /* It should be on. */ | |
618 | if (buf->newline_cache == 0) | |
619 | buf->newline_cache = new_region_cache (); | |
620 | } | |
621 | } | |
622 | ||
623 | \f | |
624 | /* Search for COUNT instances of the character TARGET between START and END. | |
625 | ||
626 | If COUNT is positive, search forwards; END must be >= START. | |
627 | If COUNT is negative, search backwards for the -COUNTth instance; | |
628 | END must be <= START. | |
629 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
630 | ||
631 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
632 | direction indicated by COUNT. | |
ffd56f97 JB |
633 | |
634 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
a9f2a45f | 635 | position past the COUNTth match. Note that for reverse motion |
5bfe95c9 | 636 | this is not the same as the usual convention for Emacs motion commands. |
ffd56f97 | 637 | |
9169c321 JB |
638 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
639 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 640 | |
087a5f81 RS |
641 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
642 | except when inside redisplay. */ | |
643 | ||
d311d28c PE |
644 | ptrdiff_t |
645 | scan_buffer (register int target, ptrdiff_t start, ptrdiff_t end, | |
646 | ptrdiff_t count, ptrdiff_t *shortage, int allow_quit) | |
ca1d1d23 | 647 | { |
9169c321 | 648 | struct region_cache *newline_cache; |
177c0ea7 | 649 | int direction; |
ffd56f97 | 650 | |
9169c321 JB |
651 | if (count > 0) |
652 | { | |
653 | direction = 1; | |
654 | if (! end) end = ZV; | |
655 | } | |
656 | else | |
657 | { | |
658 | direction = -1; | |
659 | if (! end) end = BEGV; | |
660 | } | |
ffd56f97 | 661 | |
9169c321 JB |
662 | newline_cache_on_off (current_buffer); |
663 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
664 | |
665 | if (shortage != 0) | |
666 | *shortage = 0; | |
667 | ||
087a5f81 | 668 | immediate_quit = allow_quit; |
ca1d1d23 | 669 | |
ffd56f97 | 670 | if (count > 0) |
9169c321 | 671 | while (start != end) |
ca1d1d23 | 672 | { |
9169c321 JB |
673 | /* Our innermost scanning loop is very simple; it doesn't know |
674 | about gaps, buffer ends, or the newline cache. ceiling is | |
675 | the position of the last character before the next such | |
676 | obstacle --- the last character the dumb search loop should | |
677 | examine. */ | |
d311d28c PE |
678 | ptrdiff_t ceiling_byte = CHAR_TO_BYTE (end) - 1; |
679 | ptrdiff_t start_byte = CHAR_TO_BYTE (start); | |
680 | ptrdiff_t tem; | |
9169c321 JB |
681 | |
682 | /* If we're looking for a newline, consult the newline cache | |
683 | to see where we can avoid some scanning. */ | |
684 | if (target == '\n' && newline_cache) | |
685 | { | |
0065d054 | 686 | ptrdiff_t next_change; |
9169c321 JB |
687 | immediate_quit = 0; |
688 | while (region_cache_forward | |
fa8ed3e0 RS |
689 | (current_buffer, newline_cache, start_byte, &next_change)) |
690 | start_byte = next_change; | |
cbe0db0d | 691 | immediate_quit = allow_quit; |
9169c321 | 692 | |
fa8ed3e0 RS |
693 | /* START should never be after END. */ |
694 | if (start_byte > ceiling_byte) | |
695 | start_byte = ceiling_byte; | |
9169c321 JB |
696 | |
697 | /* Now the text after start is an unknown region, and | |
698 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 699 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
700 | } |
701 | ||
702 | /* The dumb loop can only scan text stored in contiguous | |
703 | bytes. BUFFER_CEILING_OF returns the last character | |
704 | position that is contiguous, so the ceiling is the | |
705 | position after that. */ | |
67ce527d KH |
706 | tem = BUFFER_CEILING_OF (start_byte); |
707 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
708 | |
709 | { | |
177c0ea7 | 710 | /* The termination address of the dumb loop. */ |
fa8ed3e0 RS |
711 | register unsigned char *ceiling_addr |
712 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
713 | register unsigned char *cursor | |
714 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
715 | unsigned char *base = cursor; |
716 | ||
717 | while (cursor < ceiling_addr) | |
718 | { | |
719 | unsigned char *scan_start = cursor; | |
720 | ||
721 | /* The dumb loop. */ | |
722 | while (*cursor != target && ++cursor < ceiling_addr) | |
723 | ; | |
724 | ||
725 | /* If we're looking for newlines, cache the fact that | |
726 | the region from start to cursor is free of them. */ | |
727 | if (target == '\n' && newline_cache) | |
728 | know_region_cache (current_buffer, newline_cache, | |
6c1bd3f3 EZ |
729 | BYTE_TO_CHAR (start_byte + scan_start - base), |
730 | BYTE_TO_CHAR (start_byte + cursor - base)); | |
9169c321 JB |
731 | |
732 | /* Did we find the target character? */ | |
733 | if (cursor < ceiling_addr) | |
734 | { | |
735 | if (--count == 0) | |
736 | { | |
737 | immediate_quit = 0; | |
fa8ed3e0 | 738 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
739 | } |
740 | cursor++; | |
741 | } | |
742 | } | |
743 | ||
fa8ed3e0 | 744 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 745 | } |
ca1d1d23 JB |
746 | } |
747 | else | |
9169c321 JB |
748 | while (start > end) |
749 | { | |
750 | /* The last character to check before the next obstacle. */ | |
d311d28c PE |
751 | ptrdiff_t ceiling_byte = CHAR_TO_BYTE (end); |
752 | ptrdiff_t start_byte = CHAR_TO_BYTE (start); | |
753 | ptrdiff_t tem; | |
9169c321 JB |
754 | |
755 | /* Consult the newline cache, if appropriate. */ | |
756 | if (target == '\n' && newline_cache) | |
757 | { | |
0065d054 | 758 | ptrdiff_t next_change; |
9169c321 JB |
759 | immediate_quit = 0; |
760 | while (region_cache_backward | |
fa8ed3e0 RS |
761 | (current_buffer, newline_cache, start_byte, &next_change)) |
762 | start_byte = next_change; | |
cbe0db0d | 763 | immediate_quit = allow_quit; |
9169c321 JB |
764 | |
765 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
766 | if (start_byte <= ceiling_byte) |
767 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
768 | |
769 | /* Now the text before start is an unknown region, and | |
770 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 771 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
772 | } |
773 | ||
774 | /* Stop scanning before the gap. */ | |
67ce527d KH |
775 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
776 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
777 | |
778 | { | |
779 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
780 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
781 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
782 | unsigned char *base = cursor; |
783 | ||
784 | while (cursor >= ceiling_addr) | |
785 | { | |
786 | unsigned char *scan_start = cursor; | |
787 | ||
788 | while (*cursor != target && --cursor >= ceiling_addr) | |
789 | ; | |
790 | ||
791 | /* If we're looking for newlines, cache the fact that | |
792 | the region from after the cursor to start is free of them. */ | |
793 | if (target == '\n' && newline_cache) | |
794 | know_region_cache (current_buffer, newline_cache, | |
6c1bd3f3 EZ |
795 | BYTE_TO_CHAR (start_byte + cursor - base), |
796 | BYTE_TO_CHAR (start_byte + scan_start - base)); | |
9169c321 JB |
797 | |
798 | /* Did we find the target character? */ | |
799 | if (cursor >= ceiling_addr) | |
800 | { | |
801 | if (++count >= 0) | |
802 | { | |
803 | immediate_quit = 0; | |
fa8ed3e0 | 804 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
805 | } |
806 | cursor--; | |
807 | } | |
808 | } | |
809 | ||
fa8ed3e0 | 810 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
811 | } |
812 | } | |
813 | ||
ca1d1d23 JB |
814 | immediate_quit = 0; |
815 | if (shortage != 0) | |
ffd56f97 | 816 | *shortage = count * direction; |
9169c321 | 817 | return start; |
ca1d1d23 | 818 | } |
fa8ed3e0 RS |
819 | \f |
820 | /* Search for COUNT instances of a line boundary, which means either a | |
821 | newline or (if selective display enabled) a carriage return. | |
822 | Start at START. If COUNT is negative, search backwards. | |
823 | ||
824 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
825 | ||
826 | If we find COUNT instances. we position after (always after, | |
827 | even if scanning backwards) the COUNTth match, and return 0. | |
828 | ||
829 | If we don't find COUNT instances before reaching the end of the | |
830 | buffer (or the beginning, if scanning backwards), we return | |
831 | the number of line boundaries left unfound, and position at | |
832 | the limit we bumped up against. | |
833 | ||
834 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 835 | except in special cases. */ |
ca1d1d23 | 836 | |
c098fdb8 | 837 | EMACS_INT |
d311d28c PE |
838 | scan_newline (ptrdiff_t start, ptrdiff_t start_byte, |
839 | ptrdiff_t limit, ptrdiff_t limit_byte, | |
c098fdb8 | 840 | register EMACS_INT count, int allow_quit) |
63fa018d | 841 | { |
fa8ed3e0 RS |
842 | int direction = ((count > 0) ? 1 : -1); |
843 | ||
844 | register unsigned char *cursor; | |
845 | unsigned char *base; | |
846 | ||
d311d28c | 847 | ptrdiff_t ceiling; |
fa8ed3e0 RS |
848 | register unsigned char *ceiling_addr; |
849 | ||
d5d57b92 RS |
850 | int old_immediate_quit = immediate_quit; |
851 | ||
fa8ed3e0 RS |
852 | /* The code that follows is like scan_buffer |
853 | but checks for either newline or carriage return. */ | |
854 | ||
d5d57b92 RS |
855 | if (allow_quit) |
856 | immediate_quit++; | |
fa8ed3e0 RS |
857 | |
858 | start_byte = CHAR_TO_BYTE (start); | |
859 | ||
860 | if (count > 0) | |
861 | { | |
862 | while (start_byte < limit_byte) | |
863 | { | |
864 | ceiling = BUFFER_CEILING_OF (start_byte); | |
865 | ceiling = min (limit_byte - 1, ceiling); | |
866 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
867 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
868 | while (1) | |
869 | { | |
870 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
871 | ; | |
872 | ||
873 | if (cursor != ceiling_addr) | |
874 | { | |
875 | if (--count == 0) | |
876 | { | |
d5d57b92 | 877 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
878 | start_byte = start_byte + cursor - base + 1; |
879 | start = BYTE_TO_CHAR (start_byte); | |
880 | TEMP_SET_PT_BOTH (start, start_byte); | |
881 | return 0; | |
882 | } | |
883 | else | |
884 | if (++cursor == ceiling_addr) | |
885 | break; | |
886 | } | |
887 | else | |
888 | break; | |
889 | } | |
890 | start_byte += cursor - base; | |
891 | } | |
892 | } | |
893 | else | |
894 | { | |
fa8ed3e0 RS |
895 | while (start_byte > limit_byte) |
896 | { | |
897 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
898 | ceiling = max (limit_byte, ceiling); | |
899 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
900 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
901 | while (1) | |
902 | { | |
903 | while (--cursor != ceiling_addr && *cursor != '\n') | |
904 | ; | |
905 | ||
906 | if (cursor != ceiling_addr) | |
907 | { | |
908 | if (++count == 0) | |
909 | { | |
d5d57b92 | 910 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
911 | /* Return the position AFTER the match we found. */ |
912 | start_byte = start_byte + cursor - base + 1; | |
913 | start = BYTE_TO_CHAR (start_byte); | |
914 | TEMP_SET_PT_BOTH (start, start_byte); | |
915 | return 0; | |
916 | } | |
917 | } | |
918 | else | |
919 | break; | |
920 | } | |
921 | /* Here we add 1 to compensate for the last decrement | |
922 | of CURSOR, which took it past the valid range. */ | |
923 | start_byte += cursor - base + 1; | |
924 | } | |
925 | } | |
926 | ||
927 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 928 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
929 | |
930 | return count * direction; | |
63fa018d RS |
931 | } |
932 | ||
d311d28c PE |
933 | ptrdiff_t |
934 | find_next_newline_no_quit (ptrdiff_t from, ptrdiff_t cnt) | |
ca1d1d23 | 935 | { |
d311d28c | 936 | return scan_buffer ('\n', from, 0, cnt, (ptrdiff_t *) 0, 0); |
9169c321 JB |
937 | } |
938 | ||
9169c321 JB |
939 | /* Like find_next_newline, but returns position before the newline, |
940 | not after, and only search up to TO. This isn't just | |
941 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 942 | |
d311d28c PE |
943 | ptrdiff_t |
944 | find_before_next_newline (ptrdiff_t from, ptrdiff_t to, ptrdiff_t cnt) | |
9169c321 | 945 | { |
d311d28c PE |
946 | ptrdiff_t shortage; |
947 | ptrdiff_t pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
9169c321 JB |
948 | |
949 | if (shortage == 0) | |
950 | pos--; | |
177c0ea7 | 951 | |
9169c321 | 952 | return pos; |
ca1d1d23 JB |
953 | } |
954 | \f | |
ca1d1d23 JB |
955 | /* Subroutines of Lisp buffer search functions. */ |
956 | ||
957 | static Lisp_Object | |
c098fdb8 EZ |
958 | search_command (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, |
959 | Lisp_Object count, int direction, int RE, int posix) | |
ca1d1d23 | 960 | { |
a53e2e89 | 961 | register EMACS_INT np; |
d311d28c PE |
962 | EMACS_INT lim; |
963 | ptrdiff_t lim_byte; | |
a53e2e89 | 964 | EMACS_INT n = direction; |
ca1d1d23 JB |
965 | |
966 | if (!NILP (count)) | |
967 | { | |
b7826503 | 968 | CHECK_NUMBER (count); |
ca1d1d23 JB |
969 | n *= XINT (count); |
970 | } | |
971 | ||
b7826503 | 972 | CHECK_STRING (string); |
ca1d1d23 | 973 | if (NILP (bound)) |
9f43ad85 RS |
974 | { |
975 | if (n > 0) | |
976 | lim = ZV, lim_byte = ZV_BYTE; | |
977 | else | |
978 | lim = BEGV, lim_byte = BEGV_BYTE; | |
979 | } | |
ca1d1d23 JB |
980 | else |
981 | { | |
b7826503 | 982 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 983 | lim = XINT (bound); |
6ec8bbd2 | 984 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
985 | error ("Invalid search bound (wrong side of point)"); |
986 | if (lim > ZV) | |
9f43ad85 | 987 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 988 | else if (lim < BEGV) |
9f43ad85 | 989 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
990 | else |
991 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
992 | } |
993 | ||
910c747a | 994 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
4b4deea2 TT |
995 | XCHAR_TABLE (BVAR (current_buffer, case_canon_table))->extras[2] |
996 | = BVAR (current_buffer, case_eqv_table); | |
910c747a | 997 | |
9f43ad85 | 998 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
4b4deea2 TT |
999 | (!NILP (BVAR (current_buffer, case_fold_search)) |
1000 | ? BVAR (current_buffer, case_canon_table) | |
3135e9fd | 1001 | : Qnil), |
4b4deea2 TT |
1002 | (!NILP (BVAR (current_buffer, case_fold_search)) |
1003 | ? BVAR (current_buffer, case_eqv_table) | |
3135e9fd | 1004 | : Qnil), |
b819a390 | 1005 | posix); |
ca1d1d23 JB |
1006 | if (np <= 0) |
1007 | { | |
1008 | if (NILP (noerror)) | |
06f77a8a KS |
1009 | xsignal1 (Qsearch_failed, string); |
1010 | ||
ca1d1d23 JB |
1011 | if (!EQ (noerror, Qt)) |
1012 | { | |
1013 | if (lim < BEGV || lim > ZV) | |
1014 | abort (); | |
9f43ad85 | 1015 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
1016 | return Qnil; |
1017 | #if 0 /* This would be clean, but maybe programs depend on | |
1018 | a value of nil here. */ | |
481399bf | 1019 | np = lim; |
a5f217b8 | 1020 | #endif |
ca1d1d23 | 1021 | } |
481399bf RS |
1022 | else |
1023 | return Qnil; | |
ca1d1d23 JB |
1024 | } |
1025 | ||
1026 | if (np < BEGV || np > ZV) | |
1027 | abort (); | |
1028 | ||
1029 | SET_PT (np); | |
1030 | ||
1031 | return make_number (np); | |
1032 | } | |
1033 | \f | |
fa8ed3e0 RS |
1034 | /* Return 1 if REGEXP it matches just one constant string. */ |
1035 | ||
b6d6a51c | 1036 | static int |
971de7fb | 1037 | trivial_regexp_p (Lisp_Object regexp) |
b6d6a51c | 1038 | { |
d311d28c | 1039 | ptrdiff_t len = SBYTES (regexp); |
d5db4077 | 1040 | unsigned char *s = SDATA (regexp); |
b6d6a51c KH |
1041 | while (--len >= 0) |
1042 | { | |
1043 | switch (*s++) | |
1044 | { | |
1045 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
1046 | return 0; | |
1047 | case '\\': | |
1048 | if (--len < 0) | |
1049 | return 0; | |
1050 | switch (*s++) | |
1051 | { | |
1052 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
1053 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 1054 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 1055 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 1056 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
1057 | case '6': case '7': case '8': case '9': |
1058 | return 0; | |
1059 | } | |
1060 | } | |
1061 | } | |
1062 | return 1; | |
1063 | } | |
1064 | ||
ca325161 | 1065 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 1066 | starting at position POS and stopping at position LIM, |
b819a390 | 1067 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
1068 | a regular expression if RE is true. |
1069 | ||
1070 | If N is positive, searching is forward and LIM must be greater than POS. | |
1071 | If N is negative, searching is backward and LIM must be less than POS. | |
1072 | ||
facdc750 | 1073 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 1074 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
1075 | (if searching backward) or the end (if searching forward). |
1076 | ||
1077 | POSIX is nonzero if we want full backtracking (POSIX style) | |
1078 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 1079 | |
aff2ce94 RS |
1080 | #define TRANSLATE(out, trt, d) \ |
1081 | do \ | |
1082 | { \ | |
1083 | if (! NILP (trt)) \ | |
1084 | { \ | |
1085 | Lisp_Object temp; \ | |
1086 | temp = Faref (trt, make_number (d)); \ | |
1087 | if (INTEGERP (temp)) \ | |
1088 | out = XINT (temp); \ | |
1089 | else \ | |
1090 | out = d; \ | |
1091 | } \ | |
1092 | else \ | |
1093 | out = d; \ | |
1094 | } \ | |
1095 | while (0) | |
facdc750 | 1096 | |
ef887810 RS |
1097 | /* Only used in search_buffer, to record the end position of the match |
1098 | when searching regexps and SEARCH_REGS should not be changed | |
1099 | (i.e. Vinhibit_changing_match_data is non-nil). */ | |
1100 | static struct re_registers search_regs_1; | |
1101 | ||
e7deaab0 | 1102 | static EMACS_INT |
d311d28c PE |
1103 | search_buffer (Lisp_Object string, ptrdiff_t pos, ptrdiff_t pos_byte, |
1104 | ptrdiff_t lim, ptrdiff_t lim_byte, EMACS_INT n, | |
d5a3eaaf | 1105 | int RE, Lisp_Object trt, Lisp_Object inverse_trt, int posix) |
ca1d1d23 | 1106 | { |
d311d28c PE |
1107 | ptrdiff_t len = SCHARS (string); |
1108 | ptrdiff_t len_byte = SBYTES (string); | |
1109 | register ptrdiff_t i; | |
ca1d1d23 | 1110 | |
7074fde6 FP |
1111 | if (running_asynch_code) |
1112 | save_search_regs (); | |
1113 | ||
a7e4cdde | 1114 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1115 | /* Null string is found at starting position. */ |
a7e4cdde | 1116 | if (len == 0 || n == 0) |
ca325161 | 1117 | { |
0353b28f | 1118 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1119 | return pos; |
1120 | } | |
3f57a499 | 1121 | |
41a33295 | 1122 | if (RE && !(trivial_regexp_p (string) && NILP (Vsearch_spaces_regexp))) |
ca1d1d23 | 1123 | { |
facdc750 | 1124 | unsigned char *p1, *p2; |
d311d28c | 1125 | ptrdiff_t s1, s2; |
487282dc KH |
1126 | struct re_pattern_buffer *bufp; |
1127 | ||
ef887810 RS |
1128 | bufp = compile_pattern (string, |
1129 | (NILP (Vinhibit_changing_match_data) | |
1130 | ? &search_regs : &search_regs_1), | |
1131 | trt, posix, | |
4b4deea2 | 1132 | !NILP (BVAR (current_buffer, enable_multibyte_characters))); |
ca1d1d23 | 1133 | |
ca1d1d23 JB |
1134 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1135 | because letting this function finish | |
1136 | can take too long. */ | |
1137 | QUIT; /* Do a pending quit right away, | |
1138 | to avoid paradoxical behavior */ | |
1139 | /* Get pointers and sizes of the two strings | |
1140 | that make up the visible portion of the buffer. */ | |
1141 | ||
1142 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1143 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1144 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1145 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1146 | if (s1 < 0) |
1147 | { | |
1148 | p2 = p1; | |
fa8ed3e0 | 1149 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1150 | s1 = 0; |
1151 | } | |
1152 | if (s2 < 0) | |
1153 | { | |
fa8ed3e0 | 1154 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1155 | s2 = 0; |
1156 | } | |
8bb43c28 | 1157 | re_match_object = Qnil; |
177c0ea7 | 1158 | |
ca1d1d23 JB |
1159 | while (n < 0) |
1160 | { | |
d311d28c | 1161 | ptrdiff_t val; |
487282dc | 1162 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1163 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1164 | (NILP (Vinhibit_changing_match_data) |
1165 | ? &search_regs : &search_regs_1), | |
42db823b | 1166 | /* Don't allow match past current point */ |
4996330b | 1167 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1168 | if (val == -2) |
b6d6a51c KH |
1169 | { |
1170 | matcher_overflow (); | |
1171 | } | |
ca1d1d23 JB |
1172 | if (val >= 0) |
1173 | { | |
ef887810 RS |
1174 | if (NILP (Vinhibit_changing_match_data)) |
1175 | { | |
1176 | pos_byte = search_regs.start[0] + BEGV_BYTE; | |
1177 | for (i = 0; i < search_regs.num_regs; i++) | |
1178 | if (search_regs.start[i] >= 0) | |
1179 | { | |
1180 | search_regs.start[i] | |
1181 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1182 | search_regs.end[i] | |
1183 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1184 | } | |
1185 | XSETBUFFER (last_thing_searched, current_buffer); | |
1186 | /* Set pos to the new position. */ | |
1187 | pos = search_regs.start[0]; | |
1188 | } | |
1189 | else | |
1190 | { | |
1191 | pos_byte = search_regs_1.start[0] + BEGV_BYTE; | |
1192 | /* Set pos to the new position. */ | |
1193 | pos = BYTE_TO_CHAR (search_regs_1.start[0] + BEGV_BYTE); | |
1194 | } | |
ca1d1d23 JB |
1195 | } |
1196 | else | |
1197 | { | |
1198 | immediate_quit = 0; | |
1199 | return (n); | |
1200 | } | |
1201 | n++; | |
1202 | } | |
1203 | while (n > 0) | |
1204 | { | |
d311d28c | 1205 | ptrdiff_t val; |
487282dc | 1206 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1207 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1208 | (NILP (Vinhibit_changing_match_data) |
1209 | ? &search_regs : &search_regs_1), | |
4996330b | 1210 | lim_byte - BEGV_BYTE); |
ca1d1d23 | 1211 | if (val == -2) |
b6d6a51c KH |
1212 | { |
1213 | matcher_overflow (); | |
1214 | } | |
ca1d1d23 JB |
1215 | if (val >= 0) |
1216 | { | |
ef887810 RS |
1217 | if (NILP (Vinhibit_changing_match_data)) |
1218 | { | |
1219 | pos_byte = search_regs.end[0] + BEGV_BYTE; | |
1220 | for (i = 0; i < search_regs.num_regs; i++) | |
1221 | if (search_regs.start[i] >= 0) | |
1222 | { | |
1223 | search_regs.start[i] | |
1224 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1225 | search_regs.end[i] | |
1226 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1227 | } | |
1228 | XSETBUFFER (last_thing_searched, current_buffer); | |
1229 | pos = search_regs.end[0]; | |
1230 | } | |
1231 | else | |
1232 | { | |
1233 | pos_byte = search_regs_1.end[0] + BEGV_BYTE; | |
1234 | pos = BYTE_TO_CHAR (search_regs_1.end[0] + BEGV_BYTE); | |
1235 | } | |
ca1d1d23 JB |
1236 | } |
1237 | else | |
1238 | { | |
1239 | immediate_quit = 0; | |
1240 | return (0 - n); | |
1241 | } | |
1242 | n--; | |
1243 | } | |
1244 | immediate_quit = 0; | |
1245 | return (pos); | |
1246 | } | |
1247 | else /* non-RE case */ | |
1248 | { | |
facdc750 | 1249 | unsigned char *raw_pattern, *pat; |
d311d28c PE |
1250 | ptrdiff_t raw_pattern_size; |
1251 | ptrdiff_t raw_pattern_size_byte; | |
facdc750 | 1252 | unsigned char *patbuf; |
4b4deea2 | 1253 | int multibyte = !NILP (BVAR (current_buffer, enable_multibyte_characters)); |
a967ed62 | 1254 | unsigned char *base_pat; |
49e44e81 KH |
1255 | /* Set to positive if we find a non-ASCII char that need |
1256 | translation. Otherwise set to zero later. */ | |
1257 | int char_base = -1; | |
040272ce | 1258 | int boyer_moore_ok = 1; |
facdc750 RS |
1259 | |
1260 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1261 | We must convert PATTERN to match that, or we will not really | |
1262 | find things right. */ | |
1263 | ||
1264 | if (multibyte == STRING_MULTIBYTE (string)) | |
1265 | { | |
51b59d79 | 1266 | raw_pattern = SDATA (string); |
d5db4077 KR |
1267 | raw_pattern_size = SCHARS (string); |
1268 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1269 | } |
1270 | else if (multibyte) | |
1271 | { | |
d5db4077 | 1272 | raw_pattern_size = SCHARS (string); |
facdc750 | 1273 | raw_pattern_size_byte |
d5db4077 | 1274 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1275 | raw_pattern_size); |
7276d3d8 | 1276 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1277 | copy_text (SDATA (string), raw_pattern, |
1278 | SCHARS (string), 0, 1); | |
facdc750 RS |
1279 | } |
1280 | else | |
1281 | { | |
1282 | /* Converting multibyte to single-byte. | |
1283 | ||
1284 | ??? Perhaps this conversion should be done in a special way | |
1285 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1286 | so that only the multibyte chars which really correspond to | |
1287 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1288 | raw_pattern_size = SCHARS (string); |
1289 | raw_pattern_size_byte = SCHARS (string); | |
7276d3d8 | 1290 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
1291 | copy_text (SDATA (string), raw_pattern, |
1292 | SBYTES (string), 1, 0); | |
facdc750 RS |
1293 | } |
1294 | ||
1295 | /* Copy and optionally translate the pattern. */ | |
1296 | len = raw_pattern_size; | |
1297 | len_byte = raw_pattern_size_byte; | |
f5f7578b | 1298 | patbuf = (unsigned char *) alloca (len * MAX_MULTIBYTE_LENGTH); |
facdc750 RS |
1299 | pat = patbuf; |
1300 | base_pat = raw_pattern; | |
1301 | if (multibyte) | |
1302 | { | |
a17ae96a KH |
1303 | /* Fill patbuf by translated characters in STRING while |
1304 | checking if we can use boyer-moore search. If TRT is | |
1305 | non-nil, we can use boyer-moore search only if TRT can be | |
1306 | represented by the byte array of 256 elements. For that, | |
9858f6c3 | 1307 | all non-ASCII case-equivalents of all case-sensitive |
a17ae96a KH |
1308 | characters in STRING must belong to the same charset and |
1309 | row. */ | |
1310 | ||
facdc750 RS |
1311 | while (--len >= 0) |
1312 | { | |
a17ae96a | 1313 | unsigned char str_base[MAX_MULTIBYTE_LENGTH], *str; |
aff2ce94 | 1314 | int c, translated, inverse; |
a17ae96a | 1315 | int in_charlen, charlen; |
facdc750 RS |
1316 | |
1317 | /* If we got here and the RE flag is set, it's because we're | |
1318 | dealing with a regexp known to be trivial, so the backslash | |
1319 | just quotes the next character. */ | |
1320 | if (RE && *base_pat == '\\') | |
1321 | { | |
1322 | len--; | |
62221f22 | 1323 | raw_pattern_size--; |
facdc750 RS |
1324 | len_byte--; |
1325 | base_pat++; | |
1326 | } | |
1327 | ||
62a6e103 | 1328 | c = STRING_CHAR_AND_LENGTH (base_pat, in_charlen); |
040272ce | 1329 | |
a17ae96a | 1330 | if (NILP (trt)) |
facdc750 | 1331 | { |
a17ae96a KH |
1332 | str = base_pat; |
1333 | charlen = in_charlen; | |
1334 | } | |
1335 | else | |
1336 | { | |
1337 | /* Translate the character. */ | |
1338 | TRANSLATE (translated, trt, c); | |
1339 | charlen = CHAR_STRING (translated, str_base); | |
1340 | str = str_base; | |
1341 | ||
1342 | /* Check if C has any other case-equivalents. */ | |
1343 | TRANSLATE (inverse, inverse_trt, c); | |
1344 | /* If so, check if we can use boyer-moore. */ | |
1345 | if (c != inverse && boyer_moore_ok) | |
1346 | { | |
1347 | /* Check if all equivalents belong to the same | |
1348 | group of characters. Note that the check of C | |
49e44e81 KH |
1349 | itself is done by the last iteration. */ |
1350 | int this_char_base = -1; | |
1351 | ||
1352 | while (boyer_moore_ok) | |
a17ae96a | 1353 | { |
49e44e81 | 1354 | if (ASCII_BYTE_P (inverse)) |
a17ae96a | 1355 | { |
49e44e81 KH |
1356 | if (this_char_base > 0) |
1357 | boyer_moore_ok = 0; | |
1358 | else | |
dafaabd1 | 1359 | this_char_base = 0; |
a17ae96a | 1360 | } |
49e44e81 KH |
1361 | else if (CHAR_BYTE8_P (inverse)) |
1362 | /* Boyer-moore search can't handle a | |
1363 | translation of an eight-bit | |
1364 | character. */ | |
1365 | boyer_moore_ok = 0; | |
1366 | else if (this_char_base < 0) | |
1367 | { | |
1368 | this_char_base = inverse & ~0x3F; | |
1369 | if (char_base < 0) | |
1370 | char_base = this_char_base; | |
dafaabd1 | 1371 | else if (this_char_base != char_base) |
49e44e81 KH |
1372 | boyer_moore_ok = 0; |
1373 | } | |
1374 | else if ((inverse & ~0x3F) != this_char_base) | |
1375 | boyer_moore_ok = 0; | |
a17ae96a KH |
1376 | if (c == inverse) |
1377 | break; | |
1378 | TRANSLATE (inverse, inverse_trt, inverse); | |
1379 | } | |
1380 | } | |
aff2ce94 | 1381 | } |
facdc750 RS |
1382 | |
1383 | /* Store this character into the translated pattern. */ | |
72af86bd | 1384 | memcpy (pat, str, charlen); |
a17ae96a | 1385 | pat += charlen; |
facdc750 RS |
1386 | base_pat += in_charlen; |
1387 | len_byte -= in_charlen; | |
1388 | } | |
dafaabd1 AS |
1389 | |
1390 | /* If char_base is still negative we didn't find any translated | |
1391 | non-ASCII characters. */ | |
1392 | if (char_base < 0) | |
1393 | char_base = 0; | |
facdc750 RS |
1394 | } |
1395 | else | |
1396 | { | |
040272ce | 1397 | /* Unibyte buffer. */ |
a17ae96a | 1398 | char_base = 0; |
facdc750 RS |
1399 | while (--len >= 0) |
1400 | { | |
040272ce | 1401 | int c, translated; |
facdc750 RS |
1402 | |
1403 | /* If we got here and the RE flag is set, it's because we're | |
1404 | dealing with a regexp known to be trivial, so the backslash | |
1405 | just quotes the next character. */ | |
1406 | if (RE && *base_pat == '\\') | |
1407 | { | |
1408 | len--; | |
0190922f | 1409 | raw_pattern_size--; |
facdc750 RS |
1410 | base_pat++; |
1411 | } | |
1412 | c = *base_pat++; | |
aff2ce94 | 1413 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1414 | *pat++ = translated; |
1415 | } | |
1416 | } | |
1417 | ||
1418 | len_byte = pat - patbuf; | |
facdc750 RS |
1419 | pat = base_pat = patbuf; |
1420 | ||
040272ce | 1421 | if (boyer_moore_ok) |
1db5b1ad JB |
1422 | return boyer_moore (n, pat, len_byte, trt, inverse_trt, |
1423 | pos_byte, lim_byte, | |
a17ae96a | 1424 | char_base); |
facdc750 | 1425 | else |
1db5b1ad | 1426 | return simple_search (n, pat, raw_pattern_size, len_byte, trt, |
facdc750 RS |
1427 | pos, pos_byte, lim, lim_byte); |
1428 | } | |
1429 | } | |
1430 | \f | |
1431 | /* Do a simple string search N times for the string PAT, | |
1432 | whose length is LEN/LEN_BYTE, | |
1433 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1434 | TRT is the translation table. | |
f8bd51c4 | 1435 | |
facdc750 RS |
1436 | Return the character position where the match is found. |
1437 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1438 | |
facdc750 RS |
1439 | This kind of search works regardless of what is in PAT and |
1440 | regardless of what is in TRT. It is used in cases where | |
1441 | boyer_moore cannot work. */ | |
1442 | ||
e7deaab0 | 1443 | static EMACS_INT |
c098fdb8 | 1444 | simple_search (EMACS_INT n, unsigned char *pat, |
d311d28c PE |
1445 | ptrdiff_t len, ptrdiff_t len_byte, Lisp_Object trt, |
1446 | ptrdiff_t pos, ptrdiff_t pos_byte, | |
1447 | ptrdiff_t lim, ptrdiff_t lim_byte) | |
facdc750 | 1448 | { |
4b4deea2 | 1449 | int multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
ab228c24 | 1450 | int forward = n > 0; |
49e44e81 KH |
1451 | /* Number of buffer bytes matched. Note that this may be different |
1452 | from len_byte in a multibyte buffer. */ | |
d311d28c | 1453 | ptrdiff_t match_byte; |
facdc750 RS |
1454 | |
1455 | if (lim > pos && multibyte) | |
1456 | while (n > 0) | |
1457 | { | |
1458 | while (1) | |
f8bd51c4 | 1459 | { |
facdc750 | 1460 | /* Try matching at position POS. */ |
d311d28c PE |
1461 | ptrdiff_t this_pos = pos; |
1462 | ptrdiff_t this_pos_byte = pos_byte; | |
1463 | ptrdiff_t this_len = len; | |
facdc750 | 1464 | unsigned char *p = pat; |
a9469498 | 1465 | if (pos + len > lim || pos_byte + len_byte > lim_byte) |
facdc750 RS |
1466 | goto stop; |
1467 | ||
1468 | while (this_len > 0) | |
1469 | { | |
1470 | int charlen, buf_charlen; | |
ab228c24 | 1471 | int pat_ch, buf_ch; |
facdc750 | 1472 | |
62a6e103 | 1473 | pat_ch = STRING_CHAR_AND_LENGTH (p, charlen); |
facdc750 | 1474 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
facdc750 | 1475 | buf_charlen); |
aff2ce94 | 1476 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1477 | |
1478 | if (buf_ch != pat_ch) | |
1479 | break; | |
ab228c24 | 1480 | |
ab228c24 RS |
1481 | this_len--; |
1482 | p += charlen; | |
1483 | ||
1484 | this_pos_byte += buf_charlen; | |
1485 | this_pos++; | |
facdc750 RS |
1486 | } |
1487 | ||
1488 | if (this_len == 0) | |
1489 | { | |
49e44e81 | 1490 | match_byte = this_pos_byte - pos_byte; |
facdc750 | 1491 | pos += len; |
49e44e81 | 1492 | pos_byte += match_byte; |
facdc750 RS |
1493 | break; |
1494 | } | |
1495 | ||
1496 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1497 | } |
facdc750 RS |
1498 | |
1499 | n--; | |
1500 | } | |
1501 | else if (lim > pos) | |
1502 | while (n > 0) | |
1503 | { | |
1504 | while (1) | |
f8bd51c4 | 1505 | { |
facdc750 | 1506 | /* Try matching at position POS. */ |
d311d28c PE |
1507 | ptrdiff_t this_pos = pos; |
1508 | ptrdiff_t this_len = len; | |
facdc750 RS |
1509 | unsigned char *p = pat; |
1510 | ||
1511 | if (pos + len > lim) | |
1512 | goto stop; | |
1513 | ||
1514 | while (this_len > 0) | |
1515 | { | |
1516 | int pat_ch = *p++; | |
1517 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1518 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1519 | |
1520 | if (buf_ch != pat_ch) | |
1521 | break; | |
ab228c24 RS |
1522 | |
1523 | this_len--; | |
1524 | this_pos++; | |
facdc750 RS |
1525 | } |
1526 | ||
1527 | if (this_len == 0) | |
1528 | { | |
49e44e81 | 1529 | match_byte = len; |
facdc750 RS |
1530 | pos += len; |
1531 | break; | |
1532 | } | |
1533 | ||
1534 | pos++; | |
f8bd51c4 | 1535 | } |
facdc750 RS |
1536 | |
1537 | n--; | |
1538 | } | |
1539 | /* Backwards search. */ | |
1540 | else if (lim < pos && multibyte) | |
1541 | while (n < 0) | |
1542 | { | |
1543 | while (1) | |
f8bd51c4 | 1544 | { |
facdc750 | 1545 | /* Try matching at position POS. */ |
d311d28c PE |
1546 | ptrdiff_t this_pos = pos; |
1547 | ptrdiff_t this_pos_byte = pos_byte; | |
1548 | ptrdiff_t this_len = len; | |
c525b3f2 | 1549 | const unsigned char *p = pat + len_byte; |
facdc750 | 1550 | |
8b264ecb | 1551 | if (this_pos - len < lim || (pos_byte - len_byte) < lim_byte) |
facdc750 RS |
1552 | goto stop; |
1553 | ||
1554 | while (this_len > 0) | |
1555 | { | |
ab228c24 | 1556 | int pat_ch, buf_ch; |
facdc750 | 1557 | |
8b264ecb AS |
1558 | DEC_BOTH (this_pos, this_pos_byte); |
1559 | PREV_CHAR_BOUNDARY (p, pat); | |
1560 | pat_ch = STRING_CHAR (p); | |
1561 | buf_ch = STRING_CHAR (BYTE_POS_ADDR (this_pos_byte)); | |
aff2ce94 | 1562 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1563 | |
1564 | if (buf_ch != pat_ch) | |
1565 | break; | |
ab228c24 | 1566 | |
ab228c24 | 1567 | this_len--; |
facdc750 RS |
1568 | } |
1569 | ||
1570 | if (this_len == 0) | |
1571 | { | |
8b264ecb AS |
1572 | match_byte = pos_byte - this_pos_byte; |
1573 | pos = this_pos; | |
1574 | pos_byte = this_pos_byte; | |
facdc750 RS |
1575 | break; |
1576 | } | |
1577 | ||
1578 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1579 | } |
1580 | ||
facdc750 RS |
1581 | n++; |
1582 | } | |
1583 | else if (lim < pos) | |
1584 | while (n < 0) | |
1585 | { | |
1586 | while (1) | |
b6d6a51c | 1587 | { |
facdc750 | 1588 | /* Try matching at position POS. */ |
d311d28c PE |
1589 | ptrdiff_t this_pos = pos - len; |
1590 | ptrdiff_t this_len = len; | |
facdc750 RS |
1591 | unsigned char *p = pat; |
1592 | ||
a9469498 | 1593 | if (this_pos < lim) |
facdc750 RS |
1594 | goto stop; |
1595 | ||
1596 | while (this_len > 0) | |
1597 | { | |
1598 | int pat_ch = *p++; | |
1599 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1600 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1601 | |
1602 | if (buf_ch != pat_ch) | |
1603 | break; | |
ab228c24 RS |
1604 | this_len--; |
1605 | this_pos++; | |
facdc750 RS |
1606 | } |
1607 | ||
1608 | if (this_len == 0) | |
b6d6a51c | 1609 | { |
49e44e81 | 1610 | match_byte = len; |
facdc750 RS |
1611 | pos -= len; |
1612 | break; | |
b6d6a51c | 1613 | } |
facdc750 RS |
1614 | |
1615 | pos--; | |
b6d6a51c | 1616 | } |
facdc750 RS |
1617 | |
1618 | n++; | |
b6d6a51c | 1619 | } |
facdc750 RS |
1620 | |
1621 | stop: | |
1622 | if (n == 0) | |
aff2ce94 | 1623 | { |
ab228c24 | 1624 | if (forward) |
49e44e81 | 1625 | set_search_regs ((multibyte ? pos_byte : pos) - match_byte, match_byte); |
ab228c24 | 1626 | else |
49e44e81 | 1627 | set_search_regs (multibyte ? pos_byte : pos, match_byte); |
aff2ce94 RS |
1628 | |
1629 | return pos; | |
1630 | } | |
facdc750 RS |
1631 | else if (n > 0) |
1632 | return -n; | |
1633 | else | |
1634 | return n; | |
1635 | } | |
1636 | \f | |
0190922f | 1637 | /* Do Boyer-Moore search N times for the string BASE_PAT, |
1db5b1ad JB |
1638 | whose length is LEN_BYTE, |
1639 | from buffer position POS_BYTE until LIM_BYTE. | |
facdc750 RS |
1640 | DIRECTION says which direction we search in. |
1641 | TRT and INVERSE_TRT are translation tables. | |
0190922f | 1642 | Characters in PAT are already translated by TRT. |
facdc750 | 1643 | |
0190922f KH |
1644 | This kind of search works if all the characters in BASE_PAT that |
1645 | have nontrivial translation are the same aside from the last byte. | |
1646 | This makes it possible to translate just the last byte of a | |
1647 | character, and do so after just a simple test of the context. | |
b2e6b10f | 1648 | CHAR_BASE is nonzero if there is such a non-ASCII character. |
facdc750 RS |
1649 | |
1650 | If that criterion is not satisfied, do not call this function. */ | |
1651 | ||
e7deaab0 | 1652 | static EMACS_INT |
c098fdb8 | 1653 | boyer_moore (EMACS_INT n, unsigned char *base_pat, |
d311d28c | 1654 | ptrdiff_t len_byte, |
d5a3eaaf | 1655 | Lisp_Object trt, Lisp_Object inverse_trt, |
d311d28c | 1656 | ptrdiff_t pos_byte, ptrdiff_t lim_byte, |
1db5b1ad | 1657 | int char_base) |
facdc750 RS |
1658 | { |
1659 | int direction = ((n > 0) ? 1 : -1); | |
d311d28c PE |
1660 | register ptrdiff_t dirlen; |
1661 | ptrdiff_t limit; | |
e7deaab0 | 1662 | int stride_for_teases = 0; |
f4646fff | 1663 | int BM_tab[0400]; |
177c0ea7 | 1664 | register unsigned char *cursor, *p_limit; |
d311d28c | 1665 | register ptrdiff_t i; |
c098fdb8 | 1666 | register int j; |
cb6792d2 | 1667 | unsigned char *pat, *pat_end; |
4b4deea2 | 1668 | int multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
facdc750 RS |
1669 | |
1670 | unsigned char simple_translate[0400]; | |
0190922f | 1671 | /* These are set to the preceding bytes of a byte to be translated |
49e44e81 | 1672 | if char_base is nonzero. As the maximum byte length of a |
a17ae96a | 1673 | multibyte character is 5, we have to check at most four previous |
0190922f KH |
1674 | bytes. */ |
1675 | int translate_prev_byte1 = 0; | |
1676 | int translate_prev_byte2 = 0; | |
1677 | int translate_prev_byte3 = 0; | |
facdc750 | 1678 | |
f4646fff AS |
1679 | /* The general approach is that we are going to maintain that we know |
1680 | the first (closest to the present position, in whatever direction | |
1681 | we're searching) character that could possibly be the last | |
1682 | (furthest from present position) character of a valid match. We | |
1683 | advance the state of our knowledge by looking at that character | |
1684 | and seeing whether it indeed matches the last character of the | |
1685 | pattern. If it does, we take a closer look. If it does not, we | |
1686 | move our pointer (to putative last characters) as far as is | |
1687 | logically possible. This amount of movement, which I call a | |
1688 | stride, will be the length of the pattern if the actual character | |
1689 | appears nowhere in the pattern, otherwise it will be the distance | |
1690 | from the last occurrence of that character to the end of the | |
1691 | pattern. If the amount is zero we have a possible match. */ | |
1692 | ||
1693 | /* Here we make a "mickey mouse" BM table. The stride of the search | |
1694 | is determined only by the last character of the putative match. | |
1695 | If that character does not match, we will stride the proper | |
1696 | distance to propose a match that superimposes it on the last | |
1697 | instance of a character that matches it (per trt), or misses | |
1698 | it entirely if there is none. */ | |
facdc750 RS |
1699 | |
1700 | dirlen = len_byte * direction; | |
cb6792d2 RS |
1701 | |
1702 | /* Record position after the end of the pattern. */ | |
1703 | pat_end = base_pat + len_byte; | |
1704 | /* BASE_PAT points to a character that we start scanning from. | |
1705 | It is the first character in a forward search, | |
1706 | the last character in a backward search. */ | |
facdc750 | 1707 | if (direction < 0) |
cb6792d2 RS |
1708 | base_pat = pat_end - 1; |
1709 | ||
f4646fff AS |
1710 | /* A character that does not appear in the pattern induces a |
1711 | stride equal to the pattern length. */ | |
1712 | for (i = 0; i < 0400; i++) | |
1713 | BM_tab[i] = dirlen; | |
facdc750 RS |
1714 | |
1715 | /* We use this for translation, instead of TRT itself. | |
1716 | We fill this in to handle the characters that actually | |
1717 | occur in the pattern. Others don't matter anyway! */ | |
facdc750 RS |
1718 | for (i = 0; i < 0400; i++) |
1719 | simple_translate[i] = i; | |
1720 | ||
a17ae96a | 1721 | if (char_base) |
0190922f | 1722 | { |
a17ae96a | 1723 | /* Setup translate_prev_byte1/2/3/4 from CHAR_BASE. Only a |
0190922f | 1724 | byte following them are the target of translation. */ |
0190922f | 1725 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
1f3561e4 | 1726 | int cblen = CHAR_STRING (char_base, str); |
0190922f | 1727 | |
1f3561e4 PE |
1728 | translate_prev_byte1 = str[cblen - 2]; |
1729 | if (cblen > 2) | |
0190922f | 1730 | { |
1f3561e4 PE |
1731 | translate_prev_byte2 = str[cblen - 3]; |
1732 | if (cblen > 3) | |
1f1d9321 | 1733 | translate_prev_byte3 = str[cblen - 4]; |
0190922f KH |
1734 | } |
1735 | } | |
1736 | ||
facdc750 | 1737 | i = 0; |
f4646fff | 1738 | while (i != dirlen) |
facdc750 | 1739 | { |
cb6792d2 | 1740 | unsigned char *ptr = base_pat + i; |
facdc750 | 1741 | i += direction; |
facdc750 | 1742 | if (! NILP (trt)) |
ca1d1d23 | 1743 | { |
49e44e81 KH |
1744 | /* If the byte currently looking at is the last of a |
1745 | character to check case-equivalents, set CH to that | |
1746 | character. An ASCII character and a non-ASCII character | |
1747 | matching with CHAR_BASE are to be checked. */ | |
a17ae96a KH |
1748 | int ch = -1; |
1749 | ||
1750 | if (ASCII_BYTE_P (*ptr) || ! multibyte) | |
1751 | ch = *ptr; | |
49e44e81 | 1752 | else if (char_base |
86dc6ccb | 1753 | && ((pat_end - ptr) == 1 || CHAR_HEAD_P (ptr[1]))) |
ca1d1d23 | 1754 | { |
49e44e81 KH |
1755 | unsigned char *charstart = ptr - 1; |
1756 | ||
1757 | while (! (CHAR_HEAD_P (*charstart))) | |
1758 | charstart--; | |
62a6e103 | 1759 | ch = STRING_CHAR (charstart); |
a17ae96a KH |
1760 | if (char_base != (ch & ~0x3F)) |
1761 | ch = -1; | |
ca1d1d23 | 1762 | } |
facdc750 | 1763 | |
edb7b4dc | 1764 | if (ch >= 0200 && multibyte) |
49e44e81 KH |
1765 | j = (ch & 0x3F) | 0200; |
1766 | else | |
1767 | j = *ptr; | |
1768 | ||
f4646fff | 1769 | if (i == dirlen) |
facdc750 | 1770 | stride_for_teases = BM_tab[j]; |
ab228c24 | 1771 | |
facdc750 | 1772 | BM_tab[j] = dirlen - i; |
1db5b1ad JB |
1773 | /* A translation table is accompanied by its inverse -- see |
1774 | comment following downcase_table for details. */ | |
a17ae96a | 1775 | if (ch >= 0) |
ab228c24 RS |
1776 | { |
1777 | int starting_ch = ch; | |
49e44e81 | 1778 | int starting_j = j; |
a17ae96a | 1779 | |
ab228c24 RS |
1780 | while (1) |
1781 | { | |
1782 | TRANSLATE (ch, inverse_trt, ch); | |
edb7b4dc | 1783 | if (ch >= 0200 && multibyte) |
49e44e81 | 1784 | j = (ch & 0x3F) | 0200; |
ab228c24 | 1785 | else |
a17ae96a | 1786 | j = ch; |
ab228c24 RS |
1787 | |
1788 | /* For all the characters that map into CH, | |
1789 | set up simple_translate to map the last byte | |
1790 | into STARTING_J. */ | |
1791 | simple_translate[j] = starting_j; | |
1792 | if (ch == starting_ch) | |
1793 | break; | |
1794 | BM_tab[j] = dirlen - i; | |
1795 | } | |
1796 | } | |
facdc750 RS |
1797 | } |
1798 | else | |
1799 | { | |
1800 | j = *ptr; | |
1801 | ||
f4646fff | 1802 | if (i == dirlen) |
facdc750 RS |
1803 | stride_for_teases = BM_tab[j]; |
1804 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1805 | } |
f4646fff AS |
1806 | /* stride_for_teases tells how much to stride if we get a |
1807 | match on the far character but are subsequently | |
1808 | disappointed, by recording what the stride would have been | |
1809 | for that character if the last character had been | |
1810 | different. */ | |
facdc750 | 1811 | } |
facdc750 RS |
1812 | pos_byte += dirlen - ((direction > 0) ? direction : 0); |
1813 | /* loop invariant - POS_BYTE points at where last char (first | |
1814 | char if reverse) of pattern would align in a possible match. */ | |
1815 | while (n != 0) | |
1816 | { | |
d311d28c | 1817 | ptrdiff_t tail_end; |
facdc750 RS |
1818 | unsigned char *tail_end_ptr; |
1819 | ||
1820 | /* It's been reported that some (broken) compiler thinks that | |
1821 | Boolean expressions in an arithmetic context are unsigned. | |
1822 | Using an explicit ?1:0 prevents this. */ | |
1823 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1824 | < 0) | |
1825 | return (n * (0 - direction)); | |
1826 | /* First we do the part we can by pointers (maybe nothing) */ | |
1827 | QUIT; | |
1828 | pat = base_pat; | |
1829 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1830 | if (direction > 0) |
1831 | { | |
1832 | limit = BUFFER_CEILING_OF (limit); | |
1833 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1834 | can take on without hitting edge of buffer or the gap. */ | |
1835 | limit = min (limit, pos_byte + 20000); | |
1836 | limit = min (limit, lim_byte - 1); | |
1837 | } | |
1838 | else | |
1839 | { | |
1840 | limit = BUFFER_FLOOR_OF (limit); | |
1841 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1842 | can take on without hitting edge of buffer or the gap. */ | |
1843 | limit = max (limit, pos_byte - 20000); | |
1844 | limit = max (limit, lim_byte); | |
1845 | } | |
facdc750 RS |
1846 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1847 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1848 | ||
1849 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1850 | { |
facdc750 RS |
1851 | unsigned char *p2; |
1852 | ||
1853 | p_limit = BYTE_POS_ADDR (limit); | |
1854 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
f4646fff | 1855 | /* In this loop, pos + cursor - p2 is the surrogate for pos. */ |
facdc750 | 1856 | while (1) /* use one cursor setting as long as i can */ |
ca1d1d23 | 1857 | { |
facdc750 | 1858 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1859 | { |
f4646fff AS |
1860 | while (cursor <= p_limit) |
1861 | { | |
1862 | if (BM_tab[*cursor] == 0) | |
1863 | goto hit; | |
facdc750 | 1864 | cursor += BM_tab[*cursor]; |
f4646fff | 1865 | } |
facdc750 RS |
1866 | } |
1867 | else | |
1868 | { | |
f4646fff AS |
1869 | while (cursor >= p_limit) |
1870 | { | |
1871 | if (BM_tab[*cursor] == 0) | |
1872 | goto hit; | |
facdc750 | 1873 | cursor += BM_tab[*cursor]; |
f4646fff | 1874 | } |
facdc750 | 1875 | } |
f4646fff AS |
1876 | /* If you are here, cursor is beyond the end of the |
1877 | searched region. You fail to match within the | |
1878 | permitted region and would otherwise try a character | |
1879 | beyond that region. */ | |
1880 | break; | |
1881 | ||
1882 | hit: | |
facdc750 RS |
1883 | i = dirlen - direction; |
1884 | if (! NILP (trt)) | |
1885 | { | |
1886 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1887 | { |
facdc750 RS |
1888 | int ch; |
1889 | cursor -= direction; | |
1890 | /* Translate only the last byte of a character. */ | |
1891 | if (! multibyte | |
1892 | || ((cursor == tail_end_ptr | |
1893 | || CHAR_HEAD_P (cursor[1])) | |
1894 | && (CHAR_HEAD_P (cursor[0]) | |
0190922f | 1895 | /* Check if this is the last byte of |
cd1181db | 1896 | a translatable character. */ |
0190922f KH |
1897 | || (translate_prev_byte1 == cursor[-1] |
1898 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1899 | || (translate_prev_byte2 == cursor[-2] | |
1900 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1901 | || (translate_prev_byte3 == cursor[-3])))))))) | |
facdc750 RS |
1902 | ch = simple_translate[*cursor]; |
1903 | else | |
1904 | ch = *cursor; | |
1905 | if (pat[i] != ch) | |
1906 | break; | |
ca1d1d23 | 1907 | } |
facdc750 RS |
1908 | } |
1909 | else | |
1910 | { | |
1911 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1912 | { |
facdc750 RS |
1913 | cursor -= direction; |
1914 | if (pat[i] != *cursor) | |
1915 | break; | |
ca1d1d23 | 1916 | } |
facdc750 RS |
1917 | } |
1918 | cursor += dirlen - i - direction; /* fix cursor */ | |
1919 | if (i + direction == 0) | |
1920 | { | |
d311d28c | 1921 | ptrdiff_t position, start, end; |
0c8533c6 | 1922 | |
facdc750 | 1923 | cursor -= direction; |
1113d9db | 1924 | |
facdc750 RS |
1925 | position = pos_byte + cursor - p2 + ((direction > 0) |
1926 | ? 1 - len_byte : 0); | |
1927 | set_search_regs (position, len_byte); | |
ca325161 | 1928 | |
ef887810 RS |
1929 | if (NILP (Vinhibit_changing_match_data)) |
1930 | { | |
1931 | start = search_regs.start[0]; | |
1932 | end = search_regs.end[0]; | |
1933 | } | |
1934 | else | |
1935 | /* If Vinhibit_changing_match_data is non-nil, | |
1936 | search_regs will not be changed. So let's | |
1937 | compute start and end here. */ | |
1938 | { | |
1939 | start = BYTE_TO_CHAR (position); | |
1940 | end = BYTE_TO_CHAR (position + len_byte); | |
1941 | } | |
1942 | ||
facdc750 RS |
1943 | if ((n -= direction) != 0) |
1944 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1945 | else |
ef887810 | 1946 | return direction > 0 ? end : start; |
ca1d1d23 | 1947 | } |
facdc750 RS |
1948 | else |
1949 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1950 | } |
facdc750 RS |
1951 | pos_byte += cursor - p2; |
1952 | } | |
1953 | else | |
f4646fff AS |
1954 | /* Now we'll pick up a clump that has to be done the hard |
1955 | way because it covers a discontinuity. */ | |
facdc750 RS |
1956 | { |
1957 | limit = ((direction > 0) | |
1958 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1959 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1960 | limit = ((direction > 0) | |
1961 | ? min (limit + len_byte, lim_byte - 1) | |
1962 | : max (limit - len_byte, lim_byte)); | |
1963 | /* LIMIT is now the last value POS_BYTE can have | |
1964 | and still be valid for a possible match. */ | |
1965 | while (1) | |
ca1d1d23 | 1966 | { |
f4646fff AS |
1967 | /* This loop can be coded for space rather than |
1968 | speed because it will usually run only once. | |
1969 | (the reach is at most len + 21, and typically | |
1970 | does not exceed len). */ | |
facdc750 | 1971 | while ((limit - pos_byte) * direction >= 0) |
f4646fff AS |
1972 | { |
1973 | int ch = FETCH_BYTE (pos_byte); | |
1974 | if (BM_tab[ch] == 0) | |
1975 | goto hit2; | |
1976 | pos_byte += BM_tab[ch]; | |
1977 | } | |
1978 | break; /* ran off the end */ | |
1979 | ||
1980 | hit2: | |
1981 | /* Found what might be a match. */ | |
facdc750 RS |
1982 | i = dirlen - direction; |
1983 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1984 | { |
facdc750 RS |
1985 | int ch; |
1986 | unsigned char *ptr; | |
1987 | pos_byte -= direction; | |
1988 | ptr = BYTE_POS_ADDR (pos_byte); | |
1989 | /* Translate only the last byte of a character. */ | |
1990 | if (! multibyte | |
1991 | || ((ptr == tail_end_ptr | |
1992 | || CHAR_HEAD_P (ptr[1])) | |
1993 | && (CHAR_HEAD_P (ptr[0]) | |
0190922f | 1994 | /* Check if this is the last byte of a |
cd1181db | 1995 | translatable character. */ |
0190922f KH |
1996 | || (translate_prev_byte1 == ptr[-1] |
1997 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1998 | || (translate_prev_byte2 == ptr[-2] | |
1999 | && (CHAR_HEAD_P (translate_prev_byte2) | |
2000 | || translate_prev_byte3 == ptr[-3]))))))) | |
facdc750 RS |
2001 | ch = simple_translate[*ptr]; |
2002 | else | |
2003 | ch = *ptr; | |
2004 | if (pat[i] != ch) | |
2005 | break; | |
2006 | } | |
2007 | /* Above loop has moved POS_BYTE part or all the way | |
2008 | back to the first pos (last pos if reverse). | |
2009 | Set it once again at the last (first if reverse) char. */ | |
f4646fff | 2010 | pos_byte += dirlen - i - direction; |
facdc750 RS |
2011 | if (i + direction == 0) |
2012 | { | |
d311d28c | 2013 | ptrdiff_t position, start, end; |
facdc750 | 2014 | pos_byte -= direction; |
1113d9db | 2015 | |
facdc750 | 2016 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
facdc750 | 2017 | set_search_regs (position, len_byte); |
ca325161 | 2018 | |
ef887810 RS |
2019 | if (NILP (Vinhibit_changing_match_data)) |
2020 | { | |
2021 | start = search_regs.start[0]; | |
2022 | end = search_regs.end[0]; | |
2023 | } | |
2024 | else | |
2025 | /* If Vinhibit_changing_match_data is non-nil, | |
2026 | search_regs will not be changed. So let's | |
2027 | compute start and end here. */ | |
2028 | { | |
2029 | start = BYTE_TO_CHAR (position); | |
2030 | end = BYTE_TO_CHAR (position + len_byte); | |
2031 | } | |
2032 | ||
facdc750 RS |
2033 | if ((n -= direction) != 0) |
2034 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 2035 | else |
ef887810 | 2036 | return direction > 0 ? end : start; |
ca1d1d23 | 2037 | } |
facdc750 RS |
2038 | else |
2039 | pos_byte += stride_for_teases; | |
2040 | } | |
2041 | } | |
2042 | /* We have done one clump. Can we continue? */ | |
2043 | if ((lim_byte - pos_byte) * direction < 0) | |
2044 | return ((0 - n) * direction); | |
ca1d1d23 | 2045 | } |
facdc750 | 2046 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 2047 | } |
ca325161 | 2048 | |
fa8ed3e0 | 2049 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
2050 | for the overall match just found in the current buffer. |
2051 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
2052 | |
2053 | static void | |
d311d28c | 2054 | set_search_regs (ptrdiff_t beg_byte, ptrdiff_t nbytes) |
ca325161 | 2055 | { |
d311d28c | 2056 | ptrdiff_t i; |
a7e4cdde | 2057 | |
ef887810 RS |
2058 | if (!NILP (Vinhibit_changing_match_data)) |
2059 | return; | |
2060 | ||
ca325161 RS |
2061 | /* Make sure we have registers in which to store |
2062 | the match position. */ | |
2063 | if (search_regs.num_regs == 0) | |
2064 | { | |
2d4a771a RS |
2065 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
2066 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 2067 | search_regs.num_regs = 2; |
ca325161 RS |
2068 | } |
2069 | ||
a7e4cdde RS |
2070 | /* Clear out the other registers. */ |
2071 | for (i = 1; i < search_regs.num_regs; i++) | |
2072 | { | |
2073 | search_regs.start[i] = -1; | |
2074 | search_regs.end[i] = -1; | |
2075 | } | |
2076 | ||
fa8ed3e0 RS |
2077 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
2078 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 2079 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 2080 | } |
ca1d1d23 | 2081 | \f |
ca1d1d23 | 2082 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, |
8c1a1077 PJ |
2083 | "MSearch backward: ", |
2084 | doc: /* Search backward from point for STRING. | |
2085 | Set point to the beginning of the occurrence found, and return point. | |
2086 | An optional second argument bounds the search; it is a buffer position. | |
2087 | The match found must not extend before that position. | |
2088 | Optional third argument, if t, means if fail just return nil (no error). | |
2089 | If not nil and not t, position at limit of search and return nil. | |
acc28cb9 CY |
2090 | Optional fourth argument COUNT, if non-nil, means to search for COUNT |
2091 | successive occurrences. If COUNT is negative, search forward, | |
2092 | instead of backward, for -COUNT occurrences. | |
8c1a1077 PJ |
2093 | |
2094 | Search case-sensitivity is determined by the value of the variable | |
2095 | `case-fold-search', which see. | |
2096 | ||
2097 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2098 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2099 | { |
b819a390 | 2100 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2101 | } |
2102 | ||
6af43974 | 2103 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2104 | doc: /* Search forward from point for STRING. |
2105 | Set point to the end of the occurrence found, and return point. | |
2106 | An optional second argument bounds the search; it is a buffer position. | |
d6aacdcd | 2107 | The match found must not extend after that position. A value of nil is |
48740f01 | 2108 | equivalent to (point-max). |
8c1a1077 PJ |
2109 | Optional third argument, if t, means if fail just return nil (no error). |
2110 | If not nil and not t, move to limit of search and return nil. | |
acc28cb9 CY |
2111 | Optional fourth argument COUNT, if non-nil, means to search for COUNT |
2112 | successive occurrences. If COUNT is negative, search backward, | |
2113 | instead of forward, for -COUNT occurrences. | |
8c1a1077 PJ |
2114 | |
2115 | Search case-sensitivity is determined by the value of the variable | |
2116 | `case-fold-search', which see. | |
2117 | ||
2118 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2119 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2120 | { |
b819a390 | 2121 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2122 | } |
2123 | ||
ca1d1d23 | 2124 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, |
8c1a1077 PJ |
2125 | "sRE search backward: ", |
2126 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2127 | Set point to the beginning of the match, and return point. | |
2128 | The match found is the one starting last in the buffer | |
2129 | and yet ending before the origin of the search. | |
2130 | An optional second argument bounds the search; it is a buffer position. | |
2131 | The match found must start at or after that position. | |
2132 | Optional third argument, if t, means if fail just return nil (no error). | |
2133 | If not nil and not t, move to limit of search and return nil. | |
2134 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2135 | |
2136 | Search case-sensitivity is determined by the value of the variable | |
2137 | `case-fold-search', which see. | |
2138 | ||
8c1a1077 PJ |
2139 | See also the functions `match-beginning', `match-end', `match-string', |
2140 | and `replace-match'. */) | |
5842a27b | 2141 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2142 | { |
b819a390 | 2143 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2144 | } |
2145 | ||
2146 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2147 | "sRE search: ", |
2148 | doc: /* Search forward from point for regular expression REGEXP. | |
2149 | Set point to the end of the occurrence found, and return point. | |
2150 | An optional second argument bounds the search; it is a buffer position. | |
2151 | The match found must not extend after that position. | |
2152 | Optional third argument, if t, means if fail just return nil (no error). | |
2153 | If not nil and not t, move to limit of search and return nil. | |
2154 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2155 | |
2156 | Search case-sensitivity is determined by the value of the variable | |
2157 | `case-fold-search', which see. | |
2158 | ||
8c1a1077 PJ |
2159 | See also the functions `match-beginning', `match-end', `match-string', |
2160 | and `replace-match'. */) | |
5842a27b | 2161 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2162 | { |
b819a390 RS |
2163 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2164 | } | |
2165 | ||
2166 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2167 | "sPosix search backward: ", |
2168 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2169 | Find the longest match in accord with Posix regular expression rules. | |
2170 | Set point to the beginning of the match, and return point. | |
2171 | The match found is the one starting last in the buffer | |
2172 | and yet ending before the origin of the search. | |
2173 | An optional second argument bounds the search; it is a buffer position. | |
2174 | The match found must start at or after that position. | |
2175 | Optional third argument, if t, means if fail just return nil (no error). | |
2176 | If not nil and not t, move to limit of search and return nil. | |
2177 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2178 | |
2179 | Search case-sensitivity is determined by the value of the variable | |
2180 | `case-fold-search', which see. | |
2181 | ||
8c1a1077 PJ |
2182 | See also the functions `match-beginning', `match-end', `match-string', |
2183 | and `replace-match'. */) | |
5842a27b | 2184 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2185 | { |
2186 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2187 | } | |
2188 | ||
2189 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2190 | "sPosix search: ", |
2191 | doc: /* Search forward from point for regular expression REGEXP. | |
2192 | Find the longest match in accord with Posix regular expression rules. | |
2193 | Set point to the end of the occurrence found, and return point. | |
2194 | An optional second argument bounds the search; it is a buffer position. | |
2195 | The match found must not extend after that position. | |
2196 | Optional third argument, if t, means if fail just return nil (no error). | |
2197 | If not nil and not t, move to limit of search and return nil. | |
2198 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2199 | |
2200 | Search case-sensitivity is determined by the value of the variable | |
2201 | `case-fold-search', which see. | |
2202 | ||
8c1a1077 PJ |
2203 | See also the functions `match-beginning', `match-end', `match-string', |
2204 | and `replace-match'. */) | |
5842a27b | 2205 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2206 | { |
2207 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2208 | } |
2209 | \f | |
d7a5ad5f | 2210 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2211 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2212 | Leave point at the end of the replacement text. |
2213 | ||
8c1a1077 PJ |
2214 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. |
2215 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2216 | based on the replaced text. | |
2217 | If the replaced text has only capital letters | |
2218 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
4dd0c271 RS |
2219 | Otherwise if all words are capitalized in the replaced text, |
2220 | capitalize each word in NEWTEXT. | |
8c1a1077 PJ |
2221 | |
2222 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2223 | Otherwise treat `\\' as special: | |
2224 | `\\&' in NEWTEXT means substitute original matched text. | |
2225 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2226 | If Nth parens didn't match, substitute nothing. | |
2227 | `\\\\' means insert one `\\'. | |
4dd0c271 RS |
2228 | Case conversion does not apply to these substitutions. |
2229 | ||
8c1a1077 | 2230 | FIXEDCASE and LITERAL are optional arguments. |
8c1a1077 PJ |
2231 | |
2232 | The optional fourth argument STRING can be a string to modify. | |
2233 | This is meaningful when the previous match was done against STRING, | |
2234 | using `string-match'. When used this way, `replace-match' | |
2235 | creates and returns a new string made by copying STRING and replacing | |
2236 | the part of STRING that was matched. | |
2237 | ||
2238 | The optional fifth argument SUBEXP specifies a subexpression; | |
2239 | it says to replace just that subexpression with NEWTEXT, | |
2240 | rather than replacing the entire matched text. | |
2241 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2242 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2243 | NEWTEXT in place of subexp N. | |
2244 | This is useful only after a regular expression search or match, | |
2245 | since only regular expressions have distinguished subexpressions. */) | |
5842a27b | 2246 | (Lisp_Object newtext, Lisp_Object fixedcase, Lisp_Object literal, Lisp_Object string, Lisp_Object subexp) |
ca1d1d23 JB |
2247 | { |
2248 | enum { nochange, all_caps, cap_initial } case_action; | |
d311d28c | 2249 | register ptrdiff_t pos, pos_byte; |
ca1d1d23 | 2250 | int some_multiletter_word; |
97832bd0 | 2251 | int some_lowercase; |
73dc8771 | 2252 | int some_uppercase; |
208767c3 | 2253 | int some_nonuppercase_initial; |
ca1d1d23 | 2254 | register int c, prevc; |
a0efffc8 | 2255 | ptrdiff_t sub; |
d311d28c | 2256 | ptrdiff_t opoint, newpoint; |
ca1d1d23 | 2257 | |
b7826503 | 2258 | CHECK_STRING (newtext); |
ca1d1d23 | 2259 | |
080c45fd | 2260 | if (! NILP (string)) |
b7826503 | 2261 | CHECK_STRING (string); |
080c45fd | 2262 | |
ca1d1d23 JB |
2263 | case_action = nochange; /* We tried an initialization */ |
2264 | /* but some C compilers blew it */ | |
4746118a JB |
2265 | |
2266 | if (search_regs.num_regs <= 0) | |
d72cdbfc | 2267 | error ("`replace-match' called before any match found"); |
4746118a | 2268 | |
d7a5ad5f RS |
2269 | if (NILP (subexp)) |
2270 | sub = 0; | |
2271 | else | |
2272 | { | |
b7826503 | 2273 | CHECK_NUMBER (subexp); |
a0efffc8 | 2274 | if (! (0 <= XINT (subexp) && XINT (subexp) < search_regs.num_regs)) |
d7a5ad5f | 2275 | args_out_of_range (subexp, make_number (search_regs.num_regs)); |
a0efffc8 | 2276 | sub = XINT (subexp); |
d7a5ad5f RS |
2277 | } |
2278 | ||
080c45fd RS |
2279 | if (NILP (string)) |
2280 | { | |
d7a5ad5f RS |
2281 | if (search_regs.start[sub] < BEGV |
2282 | || search_regs.start[sub] > search_regs.end[sub] | |
2283 | || search_regs.end[sub] > ZV) | |
2284 | args_out_of_range (make_number (search_regs.start[sub]), | |
2285 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2286 | } |
2287 | else | |
2288 | { | |
d7a5ad5f RS |
2289 | if (search_regs.start[sub] < 0 |
2290 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2291 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2292 | args_out_of_range (make_number (search_regs.start[sub]), |
2293 | make_number (search_regs.end[sub])); | |
080c45fd | 2294 | } |
ca1d1d23 JB |
2295 | |
2296 | if (NILP (fixedcase)) | |
2297 | { | |
2298 | /* Decide how to casify by examining the matched text. */ | |
d311d28c | 2299 | ptrdiff_t last; |
ca1d1d23 | 2300 | |
ac3b28b1 KH |
2301 | pos = search_regs.start[sub]; |
2302 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2303 | |
2304 | if (NILP (string)) | |
ac3b28b1 | 2305 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2306 | else |
ac3b28b1 | 2307 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2308 | |
ca1d1d23 JB |
2309 | prevc = '\n'; |
2310 | case_action = all_caps; | |
2311 | ||
2312 | /* some_multiletter_word is set nonzero if any original word | |
2313 | is more than one letter long. */ | |
2314 | some_multiletter_word = 0; | |
97832bd0 | 2315 | some_lowercase = 0; |
208767c3 | 2316 | some_nonuppercase_initial = 0; |
73dc8771 | 2317 | some_uppercase = 0; |
ca1d1d23 | 2318 | |
ac3b28b1 | 2319 | while (pos < last) |
ca1d1d23 | 2320 | { |
080c45fd | 2321 | if (NILP (string)) |
ac3b28b1 | 2322 | { |
93daa011 | 2323 | c = FETCH_CHAR_AS_MULTIBYTE (pos_byte); |
ac3b28b1 KH |
2324 | INC_BOTH (pos, pos_byte); |
2325 | } | |
080c45fd | 2326 | else |
93daa011 | 2327 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2328 | |
5da9919f | 2329 | if (lowercasep (c)) |
ca1d1d23 JB |
2330 | { |
2331 | /* Cannot be all caps if any original char is lower case */ | |
2332 | ||
97832bd0 | 2333 | some_lowercase = 1; |
ca1d1d23 | 2334 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2335 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2336 | else |
2337 | some_multiletter_word = 1; | |
2338 | } | |
5da9919f | 2339 | else if (uppercasep (c)) |
ca1d1d23 | 2340 | { |
73dc8771 | 2341 | some_uppercase = 1; |
97832bd0 | 2342 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2343 | ; |
97832bd0 | 2344 | else |
ca1d1d23 JB |
2345 | some_multiletter_word = 1; |
2346 | } | |
208767c3 RS |
2347 | else |
2348 | { | |
2349 | /* If the initial is a caseless word constituent, | |
2350 | treat that like a lowercase initial. */ | |
2351 | if (SYNTAX (prevc) != Sword) | |
2352 | some_nonuppercase_initial = 1; | |
2353 | } | |
ca1d1d23 JB |
2354 | |
2355 | prevc = c; | |
2356 | } | |
2357 | ||
97832bd0 RS |
2358 | /* Convert to all caps if the old text is all caps |
2359 | and has at least one multiletter word. */ | |
2360 | if (! some_lowercase && some_multiletter_word) | |
2361 | case_action = all_caps; | |
c4d460ce | 2362 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2363 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2364 | case_action = cap_initial; |
208767c3 | 2365 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2366 | /* Should x -> yz, operating on X, give Yz or YZ? |
2367 | We'll assume the latter. */ | |
2368 | case_action = all_caps; | |
97832bd0 RS |
2369 | else |
2370 | case_action = nochange; | |
ca1d1d23 JB |
2371 | } |
2372 | ||
080c45fd RS |
2373 | /* Do replacement in a string. */ |
2374 | if (!NILP (string)) | |
2375 | { | |
2376 | Lisp_Object before, after; | |
2377 | ||
2378 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2379 | make_number (search_regs.start[sub])); |
2380 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2381 | |
636a5e28 RS |
2382 | /* Substitute parts of the match into NEWTEXT |
2383 | if desired. */ | |
080c45fd RS |
2384 | if (NILP (literal)) |
2385 | { | |
d311d28c PE |
2386 | ptrdiff_t lastpos = 0; |
2387 | ptrdiff_t lastpos_byte = 0; | |
080c45fd RS |
2388 | /* We build up the substituted string in ACCUM. */ |
2389 | Lisp_Object accum; | |
2390 | Lisp_Object middle; | |
d311d28c | 2391 | ptrdiff_t length = SBYTES (newtext); |
080c45fd RS |
2392 | |
2393 | accum = Qnil; | |
2394 | ||
ac3b28b1 | 2395 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd | 2396 | { |
d311d28c PE |
2397 | ptrdiff_t substart = -1; |
2398 | ptrdiff_t subend = 0; | |
1e79ec24 | 2399 | int delbackslash = 0; |
080c45fd | 2400 | |
0c8533c6 RS |
2401 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2402 | ||
080c45fd RS |
2403 | if (c == '\\') |
2404 | { | |
0c8533c6 | 2405 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2406 | |
080c45fd RS |
2407 | if (c == '&') |
2408 | { | |
d7a5ad5f RS |
2409 | substart = search_regs.start[sub]; |
2410 | subend = search_regs.end[sub]; | |
080c45fd | 2411 | } |
76fc1ea2 | 2412 | else if (c >= '1' && c <= '9') |
080c45fd | 2413 | { |
d311d28c PE |
2414 | if (c - '0' < search_regs.num_regs |
2415 | && 0 <= search_regs.start[c - '0']) | |
080c45fd RS |
2416 | { |
2417 | substart = search_regs.start[c - '0']; | |
2418 | subend = search_regs.end[c - '0']; | |
2419 | } | |
76fc1ea2 KH |
2420 | else |
2421 | { | |
2422 | /* If that subexp did not match, | |
2423 | replace \\N with nothing. */ | |
2424 | substart = 0; | |
2425 | subend = 0; | |
2426 | } | |
080c45fd | 2427 | } |
1e79ec24 KH |
2428 | else if (c == '\\') |
2429 | delbackslash = 1; | |
636a5e28 RS |
2430 | else |
2431 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2432 | } |
2433 | if (substart >= 0) | |
2434 | { | |
d131e79c RS |
2435 | if (pos - 2 != lastpos) |
2436 | middle = substring_both (newtext, lastpos, | |
2437 | lastpos_byte, | |
2438 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2439 | else |
2440 | middle = Qnil; | |
2441 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2442 | Fsubstring (string, |
2443 | make_number (substart), | |
080c45fd RS |
2444 | make_number (subend))); |
2445 | lastpos = pos; | |
0c8533c6 | 2446 | lastpos_byte = pos_byte; |
080c45fd | 2447 | } |
1e79ec24 KH |
2448 | else if (delbackslash) |
2449 | { | |
d131e79c RS |
2450 | middle = substring_both (newtext, lastpos, |
2451 | lastpos_byte, | |
2452 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2453 | |
1e79ec24 KH |
2454 | accum = concat2 (accum, middle); |
2455 | lastpos = pos; | |
0c8533c6 | 2456 | lastpos_byte = pos_byte; |
1e79ec24 | 2457 | } |
080c45fd RS |
2458 | } |
2459 | ||
d131e79c RS |
2460 | if (pos != lastpos) |
2461 | middle = substring_both (newtext, lastpos, | |
2462 | lastpos_byte, | |
0c8533c6 | 2463 | pos, pos_byte); |
080c45fd RS |
2464 | else |
2465 | middle = Qnil; | |
2466 | ||
2467 | newtext = concat2 (accum, middle); | |
2468 | } | |
2469 | ||
636a5e28 | 2470 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2471 | if (case_action == all_caps) |
2472 | newtext = Fupcase (newtext); | |
2473 | else if (case_action == cap_initial) | |
2b2eead9 | 2474 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2475 | |
2476 | return concat3 (before, newtext, after); | |
2477 | } | |
2478 | ||
09c4719e | 2479 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2480 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2481 | opoint = PT - ZV; |
9160906f RS |
2482 | else if (PT > search_regs.start[sub]) |
2483 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2484 | else |
2485 | opoint = PT; | |
2486 | ||
886ed6ec RS |
2487 | /* If we want non-literal replacement, |
2488 | perform substitution on the replacement string. */ | |
2489 | if (NILP (literal)) | |
ca1d1d23 | 2490 | { |
0065d054 | 2491 | ptrdiff_t length = SBYTES (newtext); |
68e69fbd | 2492 | unsigned char *substed; |
0065d054 | 2493 | ptrdiff_t substed_alloc_size, substed_len; |
4b4deea2 | 2494 | int buf_multibyte = !NILP (BVAR (current_buffer, enable_multibyte_characters)); |
3bc25e52 | 2495 | int str_multibyte = STRING_MULTIBYTE (newtext); |
886ed6ec | 2496 | int really_changed = 0; |
3bc25e52 | 2497 | |
0065d054 PE |
2498 | substed_alloc_size = ((STRING_BYTES_BOUND - 100) / 2 < length |
2499 | ? STRING_BYTES_BOUND | |
2500 | : length * 2 + 100); | |
2501 | substed = (unsigned char *) xmalloc (substed_alloc_size); | |
68e69fbd RS |
2502 | substed_len = 0; |
2503 | ||
3bc25e52 KH |
2504 | /* Go thru NEWTEXT, producing the actual text to insert in |
2505 | SUBSTED while adjusting multibyteness to that of the current | |
2506 | buffer. */ | |
ca1d1d23 | 2507 | |
ac3b28b1 | 2508 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2509 | { |
68e69fbd | 2510 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
8ea90aa3 | 2511 | const unsigned char *add_stuff = NULL; |
0065d054 | 2512 | ptrdiff_t add_len = 0; |
a0efffc8 | 2513 | ptrdiff_t idx = -1; |
9a76659d | 2514 | |
3bc25e52 KH |
2515 | if (str_multibyte) |
2516 | { | |
eb99a8dd | 2517 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 | 2518 | if (!buf_multibyte) |
461c2ab9 | 2519 | c = multibyte_char_to_unibyte (c); |
3bc25e52 KH |
2520 | } |
2521 | else | |
2522 | { | |
2523 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2524 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2525 | if (buf_multibyte) |
4c0354d7 | 2526 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 | 2527 | } |
ac3b28b1 | 2528 | |
68e69fbd RS |
2529 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2530 | or set IDX to a match index, which means put that part | |
2531 | of the buffer text into SUBSTED. */ | |
2532 | ||
ca1d1d23 JB |
2533 | if (c == '\\') |
2534 | { | |
886ed6ec RS |
2535 | really_changed = 1; |
2536 | ||
3bc25e52 KH |
2537 | if (str_multibyte) |
2538 | { | |
eb99a8dd KH |
2539 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2540 | pos, pos_byte); | |
071ce769 | 2541 | if (!buf_multibyte && !ASCII_CHAR_P (c)) |
461c2ab9 | 2542 | c = multibyte_char_to_unibyte (c); |
3bc25e52 KH |
2543 | } |
2544 | else | |
2545 | { | |
d5db4077 | 2546 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2547 | if (buf_multibyte) |
4c0354d7 | 2548 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 KH |
2549 | } |
2550 | ||
ca1d1d23 | 2551 | if (c == '&') |
68e69fbd | 2552 | idx = sub; |
d311d28c | 2553 | else if (c >= '1' && c <= '9' && c - '0' < search_regs.num_regs) |
ca1d1d23 JB |
2554 | { |
2555 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2556 | idx = c - '0'; |
ca1d1d23 | 2557 | } |
636a5e28 | 2558 | else if (c == '\\') |
a7e979a4 | 2559 | add_len = 1, add_stuff = (unsigned char *) "\\"; |
636a5e28 | 2560 | else |
3bc25e52 KH |
2561 | { |
2562 | xfree (substed); | |
2563 | error ("Invalid use of `\\' in replacement text"); | |
2564 | } | |
ca1d1d23 JB |
2565 | } |
2566 | else | |
68e69fbd RS |
2567 | { |
2568 | add_len = CHAR_STRING (c, str); | |
2569 | add_stuff = str; | |
2570 | } | |
2571 | ||
2572 | /* If we want to copy part of a previous match, | |
2573 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2574 | if (idx >= 0) | |
2575 | { | |
0065d054 | 2576 | ptrdiff_t begbyte = CHAR_TO_BYTE (search_regs.start[idx]); |
68e69fbd RS |
2577 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; |
2578 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2579 | move_gap (search_regs.start[idx]); | |
2580 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2581 | } | |
2582 | ||
2583 | /* Now the stuff we want to add to SUBSTED | |
2584 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2585 | ||
2586 | /* Make sure SUBSTED is big enough. */ | |
0065d054 PE |
2587 | if (substed_alloc_size - substed_len < add_len) |
2588 | substed = | |
2589 | xpalloc (substed, &substed_alloc_size, | |
2590 | add_len - (substed_alloc_size - substed_len), | |
2591 | STRING_BYTES_BOUND, 1); | |
68e69fbd RS |
2592 | |
2593 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2594 | if (add_stuff) |
2595 | { | |
72af86bd | 2596 | memcpy (substed + substed_len, add_stuff, add_len); |
f8ce8a0d GM |
2597 | substed_len += add_len; |
2598 | } | |
ca1d1d23 | 2599 | } |
68e69fbd | 2600 | |
886ed6ec | 2601 | if (really_changed) |
76fc1ea2 KH |
2602 | { |
2603 | if (buf_multibyte) | |
2604 | { | |
d311d28c | 2605 | ptrdiff_t nchars = |
c098fdb8 | 2606 | multibyte_chars_in_text (substed, substed_len); |
68e69fbd | 2607 | |
a7e979a4 PE |
2608 | newtext = make_multibyte_string ((char *) substed, nchars, |
2609 | substed_len); | |
76fc1ea2 KH |
2610 | } |
2611 | else | |
a7e979a4 | 2612 | newtext = make_unibyte_string ((char *) substed, substed_len); |
76fc1ea2 | 2613 | } |
68e69fbd | 2614 | xfree (substed); |
ca1d1d23 JB |
2615 | } |
2616 | ||
886ed6ec RS |
2617 | /* Replace the old text with the new in the cleanest possible way. */ |
2618 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2619 | newtext, 1, 0, 1); | |
d5db4077 | 2620 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2621 | |
2622 | if (case_action == all_caps) | |
886ed6ec RS |
2623 | Fupcase_region (make_number (search_regs.start[sub]), |
2624 | make_number (newpoint)); | |
ca1d1d23 | 2625 | else if (case_action == cap_initial) |
886ed6ec RS |
2626 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2627 | make_number (newpoint)); | |
3e18eecf | 2628 | |
98e942e0 RS |
2629 | /* Adjust search data for this change. */ |
2630 | { | |
d311d28c PE |
2631 | ptrdiff_t oldend = search_regs.end[sub]; |
2632 | ptrdiff_t oldstart = search_regs.start[sub]; | |
2633 | ptrdiff_t change = newpoint - search_regs.end[sub]; | |
2634 | ptrdiff_t i; | |
98e942e0 RS |
2635 | |
2636 | for (i = 0; i < search_regs.num_regs; i++) | |
2637 | { | |
41c01205 | 2638 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2639 | search_regs.start[i] += change; |
41c01205 DK |
2640 | else if (search_regs.start[i] > oldstart) |
2641 | search_regs.start[i] = oldstart; | |
2642 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2643 | search_regs.end[i] += change; |
41c01205 DK |
2644 | else if (search_regs.end[i] > oldstart) |
2645 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2646 | } |
2647 | } | |
2648 | ||
b0eba991 | 2649 | /* Put point back where it was in the text. */ |
8d808a65 | 2650 | if (opoint <= 0) |
fa8ed3e0 | 2651 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2652 | else |
fa8ed3e0 | 2653 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2654 | |
2655 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2656 | move_if_not_intangible (newpoint); |
177c0ea7 | 2657 | |
ca1d1d23 JB |
2658 | return Qnil; |
2659 | } | |
2660 | \f | |
2661 | static Lisp_Object | |
971de7fb | 2662 | match_limit (Lisp_Object num, int beginningp) |
ca1d1d23 | 2663 | { |
a0efffc8 | 2664 | EMACS_INT n; |
ca1d1d23 | 2665 | |
b7826503 | 2666 | CHECK_NUMBER (num); |
ca1d1d23 | 2667 | n = XINT (num); |
f90a5bf5 | 2668 | if (n < 0) |
bd2cbd56 | 2669 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2670 | if (search_regs.num_regs <= 0) |
2671 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2672 | if (n >= search_regs.num_regs |
4746118a | 2673 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2674 | return Qnil; |
2675 | return (make_number ((beginningp) ? search_regs.start[n] | |
2676 | : search_regs.end[n])); | |
2677 | } | |
2678 | ||
a7ca3326 | 2679 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, |
8c1a1077 PJ |
2680 | doc: /* Return position of start of text matched by last search. |
2681 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2682 | regexp. | |
2683 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2684 | SUBEXP pairs. | |
2685 | Zero means the entire text matched by the whole regexp or whole string. */) | |
5842a27b | 2686 | (Lisp_Object subexp) |
ca1d1d23 | 2687 | { |
5806161b | 2688 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2689 | } |
2690 | ||
a7ca3326 | 2691 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, |
8c1a1077 PJ |
2692 | doc: /* Return position of end of text matched by last search. |
2693 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2694 | regexp. | |
2695 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2696 | SUBEXP pairs. | |
2697 | Zero means the entire text matched by the whole regexp or whole string. */) | |
5842a27b | 2698 | (Lisp_Object subexp) |
ca1d1d23 | 2699 | { |
5806161b | 2700 | return match_limit (subexp, 0); |
177c0ea7 | 2701 | } |
ca1d1d23 | 2702 | |
a7ca3326 | 2703 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 3, 0, |
8c1a1077 PJ |
2704 | doc: /* Return a list containing all info on what the last search matched. |
2705 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2706 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2707 | if the last match was on a buffer; integers or nil if a string was matched. | |
febb8aba | 2708 | Use `set-match-data' to reinstate the data in this list. |
8c1a1077 | 2709 | |
41c01205 DK |
2710 | If INTEGERS (the optional first argument) is non-nil, always use |
2711 | integers \(rather than markers) to represent buffer positions. In | |
2712 | this case, and if the last match was in a buffer, the buffer will get | |
2713 | stored as one additional element at the end of the list. | |
2714 | ||
abd0071c KS |
2715 | If REUSE is a list, reuse it as part of the value. If REUSE is long |
2716 | enough to hold all the values, and if INTEGERS is non-nil, no consing | |
2717 | is done. | |
2718 | ||
2719 | If optional third arg RESEAT is non-nil, any previous markers on the | |
2720 | REUSE list will be modified to point to nowhere. | |
2721 | ||
140a6b7e | 2722 | Return value is undefined if the last search failed. */) |
5842a27b | 2723 | (Lisp_Object integers, Lisp_Object reuse, Lisp_Object reseat) |
ca1d1d23 | 2724 | { |
56256c2a | 2725 | Lisp_Object tail, prev; |
4746118a | 2726 | Lisp_Object *data; |
d311d28c | 2727 | ptrdiff_t i, len; |
ca1d1d23 | 2728 | |
abd0071c KS |
2729 | if (!NILP (reseat)) |
2730 | for (tail = reuse; CONSP (tail); tail = XCDR (tail)) | |
2731 | if (MARKERP (XCAR (tail))) | |
2732 | { | |
51f10faa | 2733 | unchain_marker (XMARKER (XCAR (tail))); |
abd0071c KS |
2734 | XSETCAR (tail, Qnil); |
2735 | } | |
2736 | ||
daa37602 | 2737 | if (NILP (last_thing_searched)) |
c36bcf1b | 2738 | return Qnil; |
daa37602 | 2739 | |
6bbd7a29 GM |
2740 | prev = Qnil; |
2741 | ||
41c01205 | 2742 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs + 1) |
4746118a JB |
2743 | * sizeof (Lisp_Object)); |
2744 | ||
41c01205 | 2745 | len = 0; |
4746118a | 2746 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 | 2747 | { |
d311d28c | 2748 | ptrdiff_t start = search_regs.start[i]; |
ca1d1d23 JB |
2749 | if (start >= 0) |
2750 | { | |
56256c2a RS |
2751 | if (EQ (last_thing_searched, Qt) |
2752 | || ! NILP (integers)) | |
ca1d1d23 | 2753 | { |
c235cce7 KH |
2754 | XSETFASTINT (data[2 * i], start); |
2755 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2756 | } |
0ed62dc7 | 2757 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2758 | { |
2759 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2760 | Fset_marker (data[2 * i], |
2761 | make_number (start), | |
2762 | last_thing_searched); | |
ca1d1d23 JB |
2763 | data[2 * i + 1] = Fmake_marker (); |
2764 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2765 | make_number (search_regs.end[i]), |
daa37602 | 2766 | last_thing_searched); |
ca1d1d23 | 2767 | } |
daa37602 JB |
2768 | else |
2769 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2770 | abort (); | |
2771 | ||
abd0071c | 2772 | len = 2 * i + 2; |
ca1d1d23 JB |
2773 | } |
2774 | else | |
abd0071c | 2775 | data[2 * i] = data[2 * i + 1] = Qnil; |
ca1d1d23 | 2776 | } |
56256c2a | 2777 | |
bd2cbd56 | 2778 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2779 | { |
bd2cbd56 | 2780 | data[len] = last_thing_searched; |
41c01205 DK |
2781 | len++; |
2782 | } | |
2783 | ||
56256c2a RS |
2784 | /* If REUSE is not usable, cons up the values and return them. */ |
2785 | if (! CONSP (reuse)) | |
41c01205 | 2786 | return Flist (len, data); |
56256c2a RS |
2787 | |
2788 | /* If REUSE is a list, store as many value elements as will fit | |
2789 | into the elements of REUSE. */ | |
2790 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2791 | i++, tail = XCDR (tail)) |
56256c2a | 2792 | { |
41c01205 | 2793 | if (i < len) |
f3fbd155 | 2794 | XSETCAR (tail, data[i]); |
56256c2a | 2795 | else |
f3fbd155 | 2796 | XSETCAR (tail, Qnil); |
56256c2a RS |
2797 | prev = tail; |
2798 | } | |
2799 | ||
2800 | /* If we couldn't fit all value elements into REUSE, | |
2801 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2802 | if (i < len) |
2803 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2804 | |
2805 | return reuse; | |
ca1d1d23 JB |
2806 | } |
2807 | ||
b51d6c92 SM |
2808 | /* We used to have an internal use variant of `reseat' described as: |
2809 | ||
2810 | If RESEAT is `evaporate', put the markers back on the free list | |
2811 | immediately. No other references to the markers must exist in this | |
2812 | case, so it is used only internally on the unwind stack and | |
2813 | save-match-data from Lisp. | |
2814 | ||
2815 | But it was ill-conceived: those supposedly-internal markers get exposed via | |
2816 | the undo-list, so freeing them here is unsafe. */ | |
ca1d1d23 | 2817 | |
a7ca3326 | 2818 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 2, 0, |
8c1a1077 | 2819 | doc: /* Set internal data on last search match from elements of LIST. |
abd0071c KS |
2820 | LIST should have been created by calling `match-data' previously. |
2821 | ||
51f10faa | 2822 | If optional arg RESEAT is non-nil, make markers on LIST point nowhere. */) |
5842a27b | 2823 | (register Lisp_Object list, Lisp_Object reseat) |
ca1d1d23 | 2824 | { |
5f2ab479 | 2825 | ptrdiff_t i; |
ca1d1d23 JB |
2826 | register Lisp_Object marker; |
2827 | ||
7074fde6 FP |
2828 | if (running_asynch_code) |
2829 | save_search_regs (); | |
2830 | ||
29100cea | 2831 | CHECK_LIST (list); |
ca1d1d23 | 2832 | |
41c01205 DK |
2833 | /* Unless we find a marker with a buffer or an explicit buffer |
2834 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
2835 | last_thing_searched = Qt; |
2836 | ||
4746118a JB |
2837 | /* Allocate registers if they don't already exist. */ |
2838 | { | |
d311d28c | 2839 | EMACS_INT length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2840 | |
2841 | if (length > search_regs.num_regs) | |
2842 | { | |
0065d054 | 2843 | ptrdiff_t num_regs = search_regs.num_regs; |
d311d28c PE |
2844 | if (PTRDIFF_MAX < length) |
2845 | memory_full (SIZE_MAX); | |
0065d054 PE |
2846 | search_regs.start = |
2847 | xpalloc (search_regs.start, &num_regs, length - num_regs, | |
2848 | min (PTRDIFF_MAX, UINT_MAX), sizeof (regoff_t)); | |
2849 | search_regs.end = | |
2850 | xrealloc (search_regs.end, num_regs * sizeof (regoff_t)); | |
2851 | ||
2852 | for (i = search_regs.num_regs; i < num_regs; i++) | |
e62371e9 KH |
2853 | search_regs.start[i] = -1; |
2854 | ||
0065d054 | 2855 | search_regs.num_regs = num_regs; |
4746118a | 2856 | } |
ca1d1d23 | 2857 | |
abd0071c | 2858 | for (i = 0; CONSP (list); i++) |
41c01205 | 2859 | { |
abd0071c | 2860 | marker = XCAR (list); |
bd2cbd56 | 2861 | if (BUFFERP (marker)) |
c3762cbd | 2862 | { |
bd2cbd56 | 2863 | last_thing_searched = marker; |
c3762cbd DK |
2864 | break; |
2865 | } | |
2866 | if (i >= length) | |
2867 | break; | |
41c01205 DK |
2868 | if (NILP (marker)) |
2869 | { | |
2870 | search_regs.start[i] = -1; | |
abd0071c | 2871 | list = XCDR (list); |
41c01205 DK |
2872 | } |
2873 | else | |
2874 | { | |
d311d28c | 2875 | Lisp_Object from; |
abd0071c | 2876 | Lisp_Object m; |
e2811828 | 2877 | |
abd0071c | 2878 | m = marker; |
41c01205 DK |
2879 | if (MARKERP (marker)) |
2880 | { | |
2881 | if (XMARKER (marker)->buffer == 0) | |
2882 | XSETFASTINT (marker, 0); | |
2883 | else | |
2884 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
2885 | } | |
e2811828 | 2886 | |
41c01205 | 2887 | CHECK_NUMBER_COERCE_MARKER (marker); |
d311d28c | 2888 | from = marker; |
e2811828 | 2889 | |
abd0071c KS |
2890 | if (!NILP (reseat) && MARKERP (m)) |
2891 | { | |
b51d6c92 | 2892 | unchain_marker (XMARKER (m)); |
9ad54a7e | 2893 | XSETCAR (list, Qnil); |
abd0071c KS |
2894 | } |
2895 | ||
2896 | if ((list = XCDR (list), !CONSP (list))) | |
2897 | break; | |
2898 | ||
2899 | m = marker = XCAR (list); | |
2900 | ||
41c01205 DK |
2901 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
2902 | XSETFASTINT (marker, 0); | |
e2811828 | 2903 | |
41c01205 | 2904 | CHECK_NUMBER_COERCE_MARKER (marker); |
0b1fccc4 PE |
2905 | if ((XINT (from) < 0 |
2906 | ? TYPE_MINIMUM (regoff_t) <= XINT (from) | |
2907 | : XINT (from) <= TYPE_MAXIMUM (regoff_t)) | |
2908 | && (XINT (marker) < 0 | |
2909 | ? TYPE_MINIMUM (regoff_t) <= XINT (marker) | |
2910 | : XINT (marker) <= TYPE_MAXIMUM (regoff_t))) | |
d311d28c PE |
2911 | { |
2912 | search_regs.start[i] = XINT (from); | |
2913 | search_regs.end[i] = XINT (marker); | |
2914 | } | |
2915 | else | |
2916 | { | |
2917 | search_regs.start[i] = -1; | |
2918 | } | |
abd0071c KS |
2919 | |
2920 | if (!NILP (reseat) && MARKERP (m)) | |
2921 | { | |
b51d6c92 | 2922 | unchain_marker (XMARKER (m)); |
9ad54a7e | 2923 | XSETCAR (list, Qnil); |
abd0071c | 2924 | } |
41c01205 | 2925 | } |
abd0071c | 2926 | list = XCDR (list); |
41c01205 | 2927 | } |
ca1d1d23 | 2928 | |
41c01205 DK |
2929 | for (; i < search_regs.num_regs; i++) |
2930 | search_regs.start[i] = -1; | |
2931 | } | |
ca1d1d23 | 2932 | |
177c0ea7 | 2933 | return Qnil; |
ca1d1d23 JB |
2934 | } |
2935 | ||
7074fde6 FP |
2936 | /* If non-zero the match data have been saved in saved_search_regs |
2937 | during the execution of a sentinel or filter. */ | |
75ebf74b | 2938 | static int search_regs_saved; |
7074fde6 | 2939 | static struct re_registers saved_search_regs; |
41c01205 | 2940 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
2941 | |
2942 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
2943 | if asynchronous code (filter or sentinel) is running. */ | |
2944 | static void | |
971de7fb | 2945 | save_search_regs (void) |
7074fde6 FP |
2946 | { |
2947 | if (!search_regs_saved) | |
2948 | { | |
2949 | saved_search_regs.num_regs = search_regs.num_regs; | |
2950 | saved_search_regs.start = search_regs.start; | |
2951 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
2952 | saved_last_thing_searched = last_thing_searched; |
2953 | last_thing_searched = Qnil; | |
7074fde6 | 2954 | search_regs.num_regs = 0; |
2d4a771a RS |
2955 | search_regs.start = 0; |
2956 | search_regs.end = 0; | |
7074fde6 FP |
2957 | |
2958 | search_regs_saved = 1; | |
2959 | } | |
2960 | } | |
2961 | ||
2962 | /* Called upon exit from filters and sentinels. */ | |
2963 | void | |
971de7fb | 2964 | restore_search_regs (void) |
7074fde6 FP |
2965 | { |
2966 | if (search_regs_saved) | |
2967 | { | |
2968 | if (search_regs.num_regs > 0) | |
2969 | { | |
2970 | xfree (search_regs.start); | |
2971 | xfree (search_regs.end); | |
2972 | } | |
2973 | search_regs.num_regs = saved_search_regs.num_regs; | |
2974 | search_regs.start = saved_search_regs.start; | |
2975 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
2976 | last_thing_searched = saved_last_thing_searched; |
2977 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
2978 | search_regs_saved = 0; |
2979 | } | |
2980 | } | |
2981 | ||
abd0071c | 2982 | static Lisp_Object |
971de7fb | 2983 | unwind_set_match_data (Lisp_Object list) |
abd0071c | 2984 | { |
b51d6c92 SM |
2985 | /* It is NOT ALWAYS safe to free (evaporate) the markers immediately. */ |
2986 | return Fset_match_data (list, Qt); | |
abd0071c KS |
2987 | } |
2988 | ||
2989 | /* Called to unwind protect the match data. */ | |
2990 | void | |
971de7fb | 2991 | record_unwind_save_match_data (void) |
abd0071c KS |
2992 | { |
2993 | record_unwind_protect (unwind_set_match_data, | |
2994 | Fmatch_data (Qnil, Qnil, Qnil)); | |
2995 | } | |
2996 | ||
e9a452d9 | 2997 | /* Quote a string to deactivate reg-expr chars */ |
ca1d1d23 JB |
2998 | |
2999 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 | 3000 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
5842a27b | 3001 | (Lisp_Object string) |
ca1d1d23 | 3002 | { |
a7e979a4 PE |
3003 | register char *in, *out, *end; |
3004 | register char *temp; | |
0c8533c6 | 3005 | int backslashes_added = 0; |
ca1d1d23 | 3006 | |
b7826503 | 3007 | CHECK_STRING (string); |
ca1d1d23 | 3008 | |
a7e979a4 | 3009 | temp = (char *) alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
3010 | |
3011 | /* Now copy the data into the new string, inserting escapes. */ | |
3012 | ||
a7e979a4 | 3013 | in = SSDATA (string); |
d5db4077 | 3014 | end = in + SBYTES (string); |
177c0ea7 | 3015 | out = temp; |
ca1d1d23 JB |
3016 | |
3017 | for (; in != end; in++) | |
3018 | { | |
66bc6082 | 3019 | if (*in == '[' |
ca1d1d23 JB |
3020 | || *in == '*' || *in == '.' || *in == '\\' |
3021 | || *in == '?' || *in == '+' | |
3022 | || *in == '^' || *in == '$') | |
0c8533c6 | 3023 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
3024 | *out++ = *in; |
3025 | } | |
3026 | ||
3f8100f1 | 3027 | return make_specified_string (temp, |
d5db4077 | 3028 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
3029 | out - temp, |
3030 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 3031 | } |
177c0ea7 | 3032 | \f |
dfcf069d | 3033 | void |
971de7fb | 3034 | syms_of_search (void) |
ca1d1d23 JB |
3035 | { |
3036 | register int i; | |
3037 | ||
487282dc KH |
3038 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
3039 | { | |
3040 | searchbufs[i].buf.allocated = 100; | |
b23c0a83 | 3041 | searchbufs[i].buf.buffer = (unsigned char *) xmalloc (100); |
487282dc KH |
3042 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
3043 | searchbufs[i].regexp = Qnil; | |
ecdb561e | 3044 | searchbufs[i].whitespace_regexp = Qnil; |
b69e3c18 | 3045 | searchbufs[i].syntax_table = Qnil; |
487282dc | 3046 | staticpro (&searchbufs[i].regexp); |
aa77b5ce | 3047 | staticpro (&searchbufs[i].whitespace_regexp); |
b69e3c18 | 3048 | staticpro (&searchbufs[i].syntax_table); |
487282dc KH |
3049 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); |
3050 | } | |
3051 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 | 3052 | |
cd3520a4 JB |
3053 | DEFSYM (Qsearch_failed, "search-failed"); |
3054 | DEFSYM (Qinvalid_regexp, "invalid-regexp"); | |
ca1d1d23 JB |
3055 | |
3056 | Fput (Qsearch_failed, Qerror_conditions, | |
d67b4f80 | 3057 | pure_cons (Qsearch_failed, pure_cons (Qerror, Qnil))); |
ca1d1d23 | 3058 | Fput (Qsearch_failed, Qerror_message, |
d67b4f80 | 3059 | make_pure_c_string ("Search failed")); |
ca1d1d23 JB |
3060 | |
3061 | Fput (Qinvalid_regexp, Qerror_conditions, | |
d67b4f80 | 3062 | pure_cons (Qinvalid_regexp, pure_cons (Qerror, Qnil))); |
ca1d1d23 | 3063 | Fput (Qinvalid_regexp, Qerror_message, |
d67b4f80 | 3064 | make_pure_c_string ("Invalid regexp")); |
ca1d1d23 | 3065 | |
daa37602 JB |
3066 | last_thing_searched = Qnil; |
3067 | staticpro (&last_thing_searched); | |
3068 | ||
0f6af254 DK |
3069 | saved_last_thing_searched = Qnil; |
3070 | staticpro (&saved_last_thing_searched); | |
3071 | ||
29208e82 | 3072 | DEFVAR_LISP ("search-spaces-regexp", Vsearch_spaces_regexp, |
e2811828 | 3073 | doc: /* Regexp to substitute for bunches of spaces in regexp search. |
f31a9a68 RS |
3074 | Some commands use this for user-specified regexps. |
3075 | Spaces that occur inside character classes or repetition operators | |
3076 | or other such regexp constructs are not replaced with this. | |
3077 | A value of nil (which is the normal value) means treat spaces literally. */); | |
41a33295 | 3078 | Vsearch_spaces_regexp = Qnil; |
f31a9a68 | 3079 | |
29208e82 | 3080 | DEFVAR_LISP ("inhibit-changing-match-data", Vinhibit_changing_match_data, |
ef887810 | 3081 | doc: /* Internal use only. |
39d0bf74 RS |
3082 | If non-nil, the primitive searching and matching functions |
3083 | such as `looking-at', `string-match', `re-search-forward', etc., | |
3084 | do not set the match data. The proper way to use this variable | |
3085 | is to bind it with `let' around a small expression. */); | |
ef887810 RS |
3086 | Vinhibit_changing_match_data = Qnil; |
3087 | ||
ca1d1d23 | 3088 | defsubr (&Slooking_at); |
b819a390 RS |
3089 | defsubr (&Sposix_looking_at); |
3090 | defsubr (&Sstring_match); | |
3091 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
3092 | defsubr (&Ssearch_forward); |
3093 | defsubr (&Ssearch_backward); | |
ca1d1d23 JB |
3094 | defsubr (&Sre_search_forward); |
3095 | defsubr (&Sre_search_backward); | |
b819a390 RS |
3096 | defsubr (&Sposix_search_forward); |
3097 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
3098 | defsubr (&Sreplace_match); |
3099 | defsubr (&Smatch_beginning); | |
3100 | defsubr (&Smatch_end); | |
3101 | defsubr (&Smatch_data); | |
3f1c005b | 3102 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
3103 | defsubr (&Sregexp_quote); |
3104 | } |