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> |
0328b6de | 23 | |
ca1d1d23 JB |
24 | #include "lisp.h" |
25 | #include "syntax.h" | |
5679531d | 26 | #include "category.h" |
76eb0881 | 27 | #include "character.h" |
e5560ff7 | 28 | #include "buffer.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); |
b819a390 | 104 | |
845ca893 | 105 | static _Noreturn void |
971de7fb | 106 | matcher_overflow (void) |
ca1d1d23 JB |
107 | { |
108 | error ("Stack overflow in regexp matcher"); | |
109 | } | |
110 | ||
b819a390 RS |
111 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
112 | PATTERN is the pattern to compile. | |
113 | CP is the place to put the result. | |
facdc750 | 114 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 | 115 | POSIX is nonzero if we want full backtracking (POSIX style) |
5679531d | 116 | for this pattern. 0 means backtrack only enough to get a valid match. |
bbc73b48 KH |
117 | |
118 | The behavior also depends on Vsearch_spaces_regexp. */ | |
ca1d1d23 | 119 | |
487282dc | 120 | static void |
cfe0661d | 121 | compile_pattern_1 (struct regexp_cache *cp, Lisp_Object pattern, Lisp_Object translate, int posix) |
ca1d1d23 | 122 | { |
d451e4db | 123 | char *val; |
b819a390 | 124 | reg_syntax_t old; |
ca1d1d23 | 125 | |
487282dc | 126 | cp->regexp = Qnil; |
59fab369 | 127 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 128 | cp->posix = posix; |
93daa011 | 129 | cp->buf.multibyte = STRING_MULTIBYTE (pattern); |
89ad6492 | 130 | cp->buf.charset_unibyte = charset_unibyte; |
d8c85250 CY |
131 | if (STRINGP (Vsearch_spaces_regexp)) |
132 | cp->whitespace_regexp = Vsearch_spaces_regexp; | |
133 | else | |
134 | cp->whitespace_regexp = Qnil; | |
135 | ||
e7b2dd2e RS |
136 | /* rms: I think BLOCK_INPUT is not needed here any more, |
137 | because regex.c defines malloc to call xmalloc. | |
138 | Using BLOCK_INPUT here means the debugger won't run if an error occurs. | |
139 | So let's turn it off. */ | |
140 | /* BLOCK_INPUT; */ | |
fb4a568d | 141 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 142 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
d8c85250 CY |
143 | |
144 | if (STRINGP (Vsearch_spaces_regexp)) | |
42a5b22f | 145 | re_set_whitespace_regexp (SSDATA (Vsearch_spaces_regexp)); |
d8c85250 CY |
146 | else |
147 | re_set_whitespace_regexp (NULL); | |
bbc73b48 | 148 | |
51b59d79 | 149 | val = (char *) re_compile_pattern (SSDATA (pattern), |
8f924df7 | 150 | SBYTES (pattern), &cp->buf); |
bbc73b48 | 151 | |
58e95211 SM |
152 | /* If the compiled pattern hard codes some of the contents of the |
153 | syntax-table, it can only be reused with *this* syntax table. */ | |
4b4deea2 | 154 | cp->syntax_table = cp->buf.used_syntax ? BVAR (current_buffer, syntax_table) : Qt; |
58e95211 | 155 | |
bbc73b48 KH |
156 | re_set_whitespace_regexp (NULL); |
157 | ||
b819a390 | 158 | re_set_syntax (old); |
e7b2dd2e | 159 | /* UNBLOCK_INPUT; */ |
ca1d1d23 | 160 | if (val) |
06f77a8a | 161 | xsignal1 (Qinvalid_regexp, build_string (val)); |
1113d9db | 162 | |
487282dc | 163 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
164 | } |
165 | ||
6efc7887 RS |
166 | /* Shrink each compiled regexp buffer in the cache |
167 | to the size actually used right now. | |
168 | This is called from garbage collection. */ | |
169 | ||
170 | void | |
971de7fb | 171 | shrink_regexp_cache (void) |
6efc7887 | 172 | { |
a968f437 | 173 | struct regexp_cache *cp; |
6efc7887 RS |
174 | |
175 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
176 | { | |
177 | cp->buf.allocated = cp->buf.used; | |
38182d90 | 178 | cp->buf.buffer = xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
179 | } |
180 | } | |
181 | ||
54dd3310 SM |
182 | /* Clear the regexp cache w.r.t. a particular syntax table, |
183 | because it was changed. | |
e5b94d44 CY |
184 | There is no danger of memory leak here because re_compile_pattern |
185 | automagically manages the memory in each re_pattern_buffer struct, | |
186 | based on its `allocated' and `buffer' values. */ | |
187 | void | |
971de7fb | 188 | clear_regexp_cache (void) |
e5b94d44 CY |
189 | { |
190 | int i; | |
191 | ||
192 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) | |
920fd1fc EZ |
193 | /* It's tempting to compare with the syntax-table we've actually changed, |
194 | but it's not sufficient because char-table inheritance means that | |
54dd3310 SM |
195 | modifying one syntax-table can change others at the same time. */ |
196 | if (!EQ (searchbufs[i].syntax_table, Qt)) | |
197 | searchbufs[i].regexp = Qnil; | |
e5b94d44 CY |
198 | } |
199 | ||
487282dc | 200 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
201 | the cache. |
202 | PATTERN is the pattern to compile. | |
facdc750 | 203 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
204 | REGP is the structure that says where to store the "register" |
205 | values that will result from matching this pattern. | |
206 | If it is 0, we should compile the pattern not to record any | |
207 | subexpression bounds. | |
208 | POSIX is nonzero if we want full backtracking (POSIX style) | |
209 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
210 | |
211 | struct re_pattern_buffer * | |
971de7fb | 212 | compile_pattern (Lisp_Object pattern, struct re_registers *regp, Lisp_Object translate, int posix, int multibyte) |
487282dc KH |
213 | { |
214 | struct regexp_cache *cp, **cpp; | |
215 | ||
216 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
217 | { | |
218 | cp = *cpp; | |
f1b9c7c1 KR |
219 | /* Entries are initialized to nil, and may be set to nil by |
220 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
49a5f770 KR |
221 | string accessors in those cases. However, compile_pattern_1 |
222 | is only applied to the cache entry we pick here to reuse. So | |
223 | nil should never appear before a non-nil entry. */ | |
7c752c80 | 224 | if (NILP (cp->regexp)) |
f1b9c7c1 | 225 | goto compile_it; |
d5db4077 | 226 | if (SCHARS (cp->regexp) == SCHARS (pattern) |
cf69b13e | 227 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 228 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 229 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d | 230 | && cp->posix == posix |
58e95211 | 231 | && (EQ (cp->syntax_table, Qt) |
4b4deea2 | 232 | || EQ (cp->syntax_table, BVAR (current_buffer, syntax_table))) |
89ad6492 KH |
233 | && !NILP (Fequal (cp->whitespace_regexp, Vsearch_spaces_regexp)) |
234 | && cp->buf.charset_unibyte == charset_unibyte) | |
487282dc KH |
235 | break; |
236 | ||
f1b9c7c1 KR |
237 | /* If we're at the end of the cache, compile into the nil cell |
238 | we found, or the last (least recently used) cell with a | |
239 | string value. */ | |
487282dc KH |
240 | if (cp->next == 0) |
241 | { | |
f1b9c7c1 | 242 | compile_it: |
cfe0661d | 243 | compile_pattern_1 (cp, pattern, translate, posix); |
487282dc KH |
244 | break; |
245 | } | |
246 | } | |
247 | ||
248 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
249 | either because we found it in the cache or because we just compiled it. | |
250 | Move it to the front of the queue to mark it as most recently used. */ | |
251 | *cpp = cp->next; | |
252 | cp->next = searchbuf_head; | |
253 | searchbuf_head = cp; | |
1113d9db | 254 | |
6639708c RS |
255 | /* Advise the searching functions about the space we have allocated |
256 | for register data. */ | |
257 | if (regp) | |
258 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
259 | ||
78edd3b7 | 260 | /* The compiled pattern can be used both for multibyte and unibyte |
89ad6492 KH |
261 | target. But, we have to tell which the pattern is used for. */ |
262 | cp->buf.target_multibyte = multibyte; | |
263 | ||
487282dc | 264 | return &cp->buf; |
ca1d1d23 JB |
265 | } |
266 | ||
ca1d1d23 | 267 | \f |
b819a390 | 268 | static Lisp_Object |
971de7fb | 269 | looking_at_1 (Lisp_Object string, int posix) |
ca1d1d23 JB |
270 | { |
271 | Lisp_Object val; | |
272 | unsigned char *p1, *p2; | |
d311d28c PE |
273 | ptrdiff_t s1, s2; |
274 | register ptrdiff_t i; | |
487282dc | 275 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 276 | |
7074fde6 FP |
277 | if (running_asynch_code) |
278 | save_search_regs (); | |
279 | ||
910c747a | 280 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
34dabdb7 | 281 | set_char_table_extras (BVAR (current_buffer, case_canon_table), 2, |
a098c930 | 282 | BVAR (current_buffer, case_eqv_table)); |
910c747a | 283 | |
b7826503 | 284 | CHECK_STRING (string); |
ef887810 RS |
285 | bufp = compile_pattern (string, |
286 | (NILP (Vinhibit_changing_match_data) | |
287 | ? &search_regs : NULL), | |
4b4deea2 TT |
288 | (!NILP (BVAR (current_buffer, case_fold_search)) |
289 | ? BVAR (current_buffer, case_canon_table) : Qnil), | |
0c8533c6 | 290 | posix, |
4b4deea2 | 291 | !NILP (BVAR (current_buffer, enable_multibyte_characters))); |
ca1d1d23 JB |
292 | |
293 | immediate_quit = 1; | |
294 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
295 | ||
296 | /* Get pointers and sizes of the two strings | |
297 | that make up the visible portion of the buffer. */ | |
298 | ||
299 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 300 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 301 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 302 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
303 | if (s1 < 0) |
304 | { | |
305 | p2 = p1; | |
fa8ed3e0 | 306 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
307 | s1 = 0; |
308 | } | |
309 | if (s2 < 0) | |
310 | { | |
fa8ed3e0 | 311 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
312 | s2 = 0; |
313 | } | |
8bb43c28 RS |
314 | |
315 | re_match_object = Qnil; | |
177c0ea7 | 316 | |
487282dc | 317 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
ef887810 RS |
318 | PT_BYTE - BEGV_BYTE, |
319 | (NILP (Vinhibit_changing_match_data) | |
320 | ? &search_regs : NULL), | |
fa8ed3e0 | 321 | ZV_BYTE - BEGV_BYTE); |
de182d70 | 322 | immediate_quit = 0; |
177c0ea7 | 323 | |
ca1d1d23 JB |
324 | if (i == -2) |
325 | matcher_overflow (); | |
326 | ||
327 | val = (0 <= i ? Qt : Qnil); | |
ef887810 | 328 | if (NILP (Vinhibit_changing_match_data) && i >= 0) |
fa8ed3e0 RS |
329 | for (i = 0; i < search_regs.num_regs; i++) |
330 | if (search_regs.start[i] >= 0) | |
331 | { | |
332 | search_regs.start[i] | |
333 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
334 | search_regs.end[i] | |
335 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
336 | } | |
ef887810 RS |
337 | |
338 | /* Set last_thing_searched only when match data is changed. */ | |
339 | if (NILP (Vinhibit_changing_match_data)) | |
340 | XSETBUFFER (last_thing_searched, current_buffer); | |
341 | ||
ca1d1d23 JB |
342 | return val; |
343 | } | |
344 | ||
b819a390 | 345 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
346 | doc: /* Return t if text after point matches regular expression REGEXP. |
347 | This function modifies the match data that `match-beginning', | |
348 | `match-end' and `match-data' access; save and restore the match | |
349 | data if you want to preserve them. */) | |
5842a27b | 350 | (Lisp_Object regexp) |
b819a390 | 351 | { |
94f94972 | 352 | return looking_at_1 (regexp, 0); |
b819a390 RS |
353 | } |
354 | ||
355 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
356 | doc: /* Return t if text after point matches regular expression REGEXP. |
357 | Find the longest match, in accord with Posix regular expression rules. | |
358 | This function modifies the match data that `match-beginning', | |
359 | `match-end' and `match-data' access; save and restore the match | |
360 | data if you want to preserve them. */) | |
5842a27b | 361 | (Lisp_Object regexp) |
b819a390 | 362 | { |
94f94972 | 363 | return looking_at_1 (regexp, 1); |
b819a390 RS |
364 | } |
365 | \f | |
366 | static Lisp_Object | |
971de7fb | 367 | string_match_1 (Lisp_Object regexp, Lisp_Object string, Lisp_Object start, int posix) |
ca1d1d23 | 368 | { |
d311d28c | 369 | ptrdiff_t val; |
487282dc | 370 | struct re_pattern_buffer *bufp; |
d311d28c PE |
371 | EMACS_INT pos; |
372 | ptrdiff_t pos_byte, i; | |
ca1d1d23 | 373 | |
7074fde6 FP |
374 | if (running_asynch_code) |
375 | save_search_regs (); | |
376 | ||
b7826503 PJ |
377 | CHECK_STRING (regexp); |
378 | CHECK_STRING (string); | |
ca1d1d23 JB |
379 | |
380 | if (NILP (start)) | |
0c8533c6 | 381 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
382 | else |
383 | { | |
d311d28c | 384 | ptrdiff_t len = SCHARS (string); |
ca1d1d23 | 385 | |
b7826503 | 386 | CHECK_NUMBER (start); |
0c8533c6 RS |
387 | pos = XINT (start); |
388 | if (pos < 0 && -pos <= len) | |
389 | pos = len + pos; | |
390 | else if (0 > pos || pos > len) | |
ca1d1d23 | 391 | args_out_of_range (string, start); |
0c8533c6 | 392 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
393 | } |
394 | ||
910c747a | 395 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
34dabdb7 | 396 | set_char_table_extras (BVAR (current_buffer, case_canon_table), 2, |
a098c930 | 397 | BVAR (current_buffer, case_eqv_table)); |
910c747a | 398 | |
ef887810 RS |
399 | bufp = compile_pattern (regexp, |
400 | (NILP (Vinhibit_changing_match_data) | |
401 | ? &search_regs : NULL), | |
4b4deea2 TT |
402 | (!NILP (BVAR (current_buffer, case_fold_search)) |
403 | ? BVAR (current_buffer, case_canon_table) : Qnil), | |
0c8533c6 RS |
404 | posix, |
405 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 406 | immediate_quit = 1; |
8bb43c28 | 407 | re_match_object = string; |
177c0ea7 | 408 | |
51b59d79 | 409 | val = re_search (bufp, SSDATA (string), |
d5db4077 KR |
410 | SBYTES (string), pos_byte, |
411 | SBYTES (string) - pos_byte, | |
ef887810 RS |
412 | (NILP (Vinhibit_changing_match_data) |
413 | ? &search_regs : NULL)); | |
ca1d1d23 | 414 | immediate_quit = 0; |
ef887810 RS |
415 | |
416 | /* Set last_thing_searched only when match data is changed. */ | |
417 | if (NILP (Vinhibit_changing_match_data)) | |
418 | last_thing_searched = Qt; | |
419 | ||
ca1d1d23 JB |
420 | if (val == -2) |
421 | matcher_overflow (); | |
422 | if (val < 0) return Qnil; | |
0c8533c6 | 423 | |
ef887810 RS |
424 | if (NILP (Vinhibit_changing_match_data)) |
425 | for (i = 0; i < search_regs.num_regs; i++) | |
426 | if (search_regs.start[i] >= 0) | |
427 | { | |
428 | search_regs.start[i] | |
429 | = string_byte_to_char (string, search_regs.start[i]); | |
430 | search_regs.end[i] | |
431 | = string_byte_to_char (string, search_regs.end[i]); | |
432 | } | |
0c8533c6 RS |
433 | |
434 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 435 | } |
e59a8453 | 436 | |
a7ca3326 | 437 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 | 438 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
b85acc4b | 439 | Matching ignores case if `case-fold-search' is non-nil. |
8c1a1077 PJ |
440 | If third arg START is non-nil, start search at that index in STRING. |
441 | For index of first char beyond the match, do (match-end 0). | |
442 | `match-end' and `match-beginning' also give indices of substrings | |
2bd2f32d RS |
443 | matched by parenthesis constructs in the pattern. |
444 | ||
445 | You can use the function `match-string' to extract the substrings | |
446 | matched by the parenthesis constructions in REGEXP. */) | |
5842a27b | 447 | (Lisp_Object regexp, Lisp_Object string, Lisp_Object start) |
b819a390 RS |
448 | { |
449 | return string_match_1 (regexp, string, start, 0); | |
450 | } | |
451 | ||
452 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
453 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
454 | Find the longest match, in accord with Posix regular expression rules. | |
455 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
456 | If third arg START is non-nil, start search at that index in STRING. | |
457 | For index of first char beyond the match, do (match-end 0). | |
458 | `match-end' and `match-beginning' also give indices of substrings | |
459 | matched by parenthesis constructs in the pattern. */) | |
5842a27b | 460 | (Lisp_Object regexp, Lisp_Object string, Lisp_Object start) |
b819a390 RS |
461 | { |
462 | return string_match_1 (regexp, string, start, 1); | |
463 | } | |
464 | ||
e59a8453 RS |
465 | /* Match REGEXP against STRING, searching all of STRING, |
466 | and return the index of the match, or negative on failure. | |
467 | This does not clobber the match data. */ | |
468 | ||
d311d28c | 469 | ptrdiff_t |
971de7fb | 470 | fast_string_match (Lisp_Object regexp, Lisp_Object string) |
e59a8453 | 471 | { |
d311d28c | 472 | ptrdiff_t val; |
487282dc | 473 | struct re_pattern_buffer *bufp; |
e59a8453 | 474 | |
facdc750 RS |
475 | bufp = compile_pattern (regexp, 0, Qnil, |
476 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 477 | immediate_quit = 1; |
8bb43c28 | 478 | re_match_object = string; |
177c0ea7 | 479 | |
51b59d79 | 480 | val = re_search (bufp, SSDATA (string), |
d5db4077 KR |
481 | SBYTES (string), 0, |
482 | SBYTES (string), 0); | |
e59a8453 RS |
483 | immediate_quit = 0; |
484 | return val; | |
485 | } | |
5679531d KH |
486 | |
487 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
488 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
489 | This does not clobber the match data. |
490 | We assume that STRING contains single-byte characters. */ | |
5679531d | 491 | |
d311d28c | 492 | ptrdiff_t |
d923b542 DA |
493 | fast_c_string_match_ignore_case (Lisp_Object regexp, |
494 | const char *string, ptrdiff_t len) | |
5679531d | 495 | { |
d311d28c | 496 | ptrdiff_t val; |
5679531d | 497 | struct re_pattern_buffer *bufp; |
5679531d | 498 | |
0c8533c6 | 499 | regexp = string_make_unibyte (regexp); |
b4577c63 | 500 | re_match_object = Qt; |
5679531d | 501 | bufp = compile_pattern (regexp, 0, |
0190922f | 502 | Vascii_canon_table, 0, |
f8bd51c4 | 503 | 0); |
5679531d KH |
504 | immediate_quit = 1; |
505 | val = re_search (bufp, string, len, 0, len, 0); | |
506 | immediate_quit = 0; | |
507 | return val; | |
508 | } | |
be5f4dfb KH |
509 | |
510 | /* Like fast_string_match but ignore case. */ | |
511 | ||
d311d28c | 512 | ptrdiff_t |
971de7fb | 513 | fast_string_match_ignore_case (Lisp_Object regexp, Lisp_Object string) |
be5f4dfb | 514 | { |
d311d28c | 515 | ptrdiff_t val; |
be5f4dfb KH |
516 | struct re_pattern_buffer *bufp; |
517 | ||
0190922f | 518 | bufp = compile_pattern (regexp, 0, Vascii_canon_table, |
be5f4dfb KH |
519 | 0, STRING_MULTIBYTE (string)); |
520 | immediate_quit = 1; | |
521 | re_match_object = string; | |
522 | ||
51b59d79 | 523 | val = re_search (bufp, SSDATA (string), |
be5f4dfb KH |
524 | SBYTES (string), 0, |
525 | SBYTES (string), 0); | |
526 | immediate_quit = 0; | |
527 | return val; | |
528 | } | |
ca1d1d23 | 529 | \f |
a80ce213 | 530 | /* Match REGEXP against the characters after POS to LIMIT, and return |
8b13507a KH |
531 | the number of matched characters. If STRING is non-nil, match |
532 | against the characters in it. In that case, POS and LIMIT are | |
533 | indices into the string. This function doesn't modify the match | |
534 | data. */ | |
535 | ||
d311d28c PE |
536 | ptrdiff_t |
537 | 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 |
538 | { |
539 | int multibyte; | |
540 | struct re_pattern_buffer *buf; | |
541 | unsigned char *p1, *p2; | |
d311d28c PE |
542 | ptrdiff_t s1, s2; |
543 | ptrdiff_t len; | |
78edd3b7 | 544 | |
8b13507a KH |
545 | if (STRINGP (string)) |
546 | { | |
547 | if (pos_byte < 0) | |
548 | pos_byte = string_char_to_byte (string, pos); | |
549 | if (limit_byte < 0) | |
550 | limit_byte = string_char_to_byte (string, limit); | |
551 | p1 = NULL; | |
552 | s1 = 0; | |
553 | p2 = SDATA (string); | |
554 | s2 = SBYTES (string); | |
555 | re_match_object = string; | |
556 | multibyte = STRING_MULTIBYTE (string); | |
557 | } | |
558 | else | |
559 | { | |
560 | if (pos_byte < 0) | |
561 | pos_byte = CHAR_TO_BYTE (pos); | |
562 | if (limit_byte < 0) | |
563 | limit_byte = CHAR_TO_BYTE (limit); | |
564 | pos_byte -= BEGV_BYTE; | |
565 | limit_byte -= BEGV_BYTE; | |
566 | p1 = BEGV_ADDR; | |
567 | s1 = GPT_BYTE - BEGV_BYTE; | |
568 | p2 = GAP_END_ADDR; | |
569 | s2 = ZV_BYTE - GPT_BYTE; | |
570 | if (s1 < 0) | |
571 | { | |
572 | p2 = p1; | |
573 | s2 = ZV_BYTE - BEGV_BYTE; | |
574 | s1 = 0; | |
575 | } | |
576 | if (s2 < 0) | |
577 | { | |
578 | s1 = ZV_BYTE - BEGV_BYTE; | |
579 | s2 = 0; | |
580 | } | |
581 | re_match_object = Qnil; | |
4b4deea2 | 582 | multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
8b13507a KH |
583 | } |
584 | ||
585 | buf = compile_pattern (regexp, 0, Qnil, 0, multibyte); | |
586 | immediate_quit = 1; | |
587 | len = re_match_2 (buf, (char *) p1, s1, (char *) p2, s2, | |
588 | pos_byte, NULL, limit_byte); | |
589 | immediate_quit = 0; | |
590 | ||
591 | return len; | |
592 | } | |
593 | ||
594 | \f | |
9169c321 JB |
595 | /* The newline cache: remembering which sections of text have no newlines. */ |
596 | ||
597 | /* If the user has requested newline caching, make sure it's on. | |
598 | Otherwise, make sure it's off. | |
599 | This is our cheezy way of associating an action with the change of | |
600 | state of a buffer-local variable. */ | |
601 | static void | |
971de7fb | 602 | newline_cache_on_off (struct buffer *buf) |
9169c321 | 603 | { |
4b4deea2 | 604 | if (NILP (BVAR (buf, cache_long_line_scans))) |
9169c321 JB |
605 | { |
606 | /* It should be off. */ | |
607 | if (buf->newline_cache) | |
608 | { | |
609 | free_region_cache (buf->newline_cache); | |
610 | buf->newline_cache = 0; | |
611 | } | |
612 | } | |
613 | else | |
614 | { | |
615 | /* It should be on. */ | |
616 | if (buf->newline_cache == 0) | |
617 | buf->newline_cache = new_region_cache (); | |
618 | } | |
619 | } | |
620 | ||
621 | \f | |
622 | /* Search for COUNT instances of the character TARGET between START and END. | |
623 | ||
624 | If COUNT is positive, search forwards; END must be >= START. | |
625 | If COUNT is negative, search backwards for the -COUNTth instance; | |
626 | END must be <= START. | |
627 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
628 | ||
629 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
630 | direction indicated by COUNT. | |
ffd56f97 JB |
631 | |
632 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
a9f2a45f | 633 | position past the COUNTth match. Note that for reverse motion |
5bfe95c9 | 634 | this is not the same as the usual convention for Emacs motion commands. |
ffd56f97 | 635 | |
9169c321 JB |
636 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
637 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 638 | |
087a5f81 RS |
639 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
640 | except when inside redisplay. */ | |
641 | ||
d311d28c PE |
642 | ptrdiff_t |
643 | scan_buffer (register int target, ptrdiff_t start, ptrdiff_t end, | |
644 | ptrdiff_t count, ptrdiff_t *shortage, int allow_quit) | |
ca1d1d23 | 645 | { |
9169c321 | 646 | struct region_cache *newline_cache; |
177c0ea7 | 647 | int direction; |
ffd56f97 | 648 | |
9169c321 JB |
649 | if (count > 0) |
650 | { | |
651 | direction = 1; | |
652 | if (! end) end = ZV; | |
653 | } | |
654 | else | |
655 | { | |
656 | direction = -1; | |
657 | if (! end) end = BEGV; | |
658 | } | |
ffd56f97 | 659 | |
9169c321 JB |
660 | newline_cache_on_off (current_buffer); |
661 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
662 | |
663 | if (shortage != 0) | |
664 | *shortage = 0; | |
665 | ||
087a5f81 | 666 | immediate_quit = allow_quit; |
ca1d1d23 | 667 | |
ffd56f97 | 668 | if (count > 0) |
9169c321 | 669 | while (start != end) |
ca1d1d23 | 670 | { |
9169c321 JB |
671 | /* Our innermost scanning loop is very simple; it doesn't know |
672 | about gaps, buffer ends, or the newline cache. ceiling is | |
673 | the position of the last character before the next such | |
674 | obstacle --- the last character the dumb search loop should | |
675 | examine. */ | |
d311d28c | 676 | ptrdiff_t ceiling_byte = CHAR_TO_BYTE (end) - 1; |
c8b9f1bc | 677 | ptrdiff_t start_byte; |
d311d28c | 678 | ptrdiff_t tem; |
9169c321 JB |
679 | |
680 | /* If we're looking for a newline, consult the newline cache | |
681 | to see where we can avoid some scanning. */ | |
682 | if (target == '\n' && newline_cache) | |
683 | { | |
0065d054 | 684 | ptrdiff_t next_change; |
9169c321 JB |
685 | immediate_quit = 0; |
686 | while (region_cache_forward | |
c8b9f1bc EZ |
687 | (current_buffer, newline_cache, start, &next_change)) |
688 | start = next_change; | |
cbe0db0d | 689 | immediate_quit = allow_quit; |
9169c321 | 690 | |
c8b9f1bc EZ |
691 | start_byte = CHAR_TO_BYTE (start); |
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. */ | |
c8b9f1bc | 699 | ceiling_byte = min (CHAR_TO_BYTE (next_change) - 1, ceiling_byte); |
9169c321 | 700 | } |
c8b9f1bc EZ |
701 | else |
702 | start_byte = CHAR_TO_BYTE (start); | |
9169c321 JB |
703 | |
704 | /* The dumb loop can only scan text stored in contiguous | |
705 | bytes. BUFFER_CEILING_OF returns the last character | |
706 | position that is contiguous, so the ceiling is the | |
707 | position after that. */ | |
67ce527d KH |
708 | tem = BUFFER_CEILING_OF (start_byte); |
709 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
710 | |
711 | { | |
177c0ea7 | 712 | /* The termination address of the dumb loop. */ |
fa8ed3e0 RS |
713 | register unsigned char *ceiling_addr |
714 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
715 | register unsigned char *cursor | |
716 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
717 | unsigned char *base = cursor; |
718 | ||
719 | while (cursor < ceiling_addr) | |
720 | { | |
721 | unsigned char *scan_start = cursor; | |
722 | ||
723 | /* The dumb loop. */ | |
724 | while (*cursor != target && ++cursor < ceiling_addr) | |
725 | ; | |
726 | ||
727 | /* If we're looking for newlines, cache the fact that | |
728 | the region from start to cursor is free of them. */ | |
729 | if (target == '\n' && newline_cache) | |
730 | know_region_cache (current_buffer, newline_cache, | |
6c1bd3f3 EZ |
731 | BYTE_TO_CHAR (start_byte + scan_start - base), |
732 | BYTE_TO_CHAR (start_byte + cursor - base)); | |
9169c321 JB |
733 | |
734 | /* Did we find the target character? */ | |
735 | if (cursor < ceiling_addr) | |
736 | { | |
737 | if (--count == 0) | |
738 | { | |
739 | immediate_quit = 0; | |
fa8ed3e0 | 740 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
741 | } |
742 | cursor++; | |
743 | } | |
744 | } | |
745 | ||
fa8ed3e0 | 746 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 747 | } |
ca1d1d23 JB |
748 | } |
749 | else | |
9169c321 JB |
750 | while (start > end) |
751 | { | |
752 | /* The last character to check before the next obstacle. */ | |
d311d28c | 753 | ptrdiff_t ceiling_byte = CHAR_TO_BYTE (end); |
c8b9f1bc | 754 | ptrdiff_t start_byte; |
d311d28c | 755 | ptrdiff_t tem; |
9169c321 JB |
756 | |
757 | /* Consult the newline cache, if appropriate. */ | |
758 | if (target == '\n' && newline_cache) | |
759 | { | |
0065d054 | 760 | ptrdiff_t next_change; |
9169c321 JB |
761 | immediate_quit = 0; |
762 | while (region_cache_backward | |
c8b9f1bc EZ |
763 | (current_buffer, newline_cache, start, &next_change)) |
764 | start = next_change; | |
cbe0db0d | 765 | immediate_quit = allow_quit; |
9169c321 | 766 | |
c8b9f1bc EZ |
767 | start_byte = CHAR_TO_BYTE (start); |
768 | ||
9169c321 | 769 | /* Start should never be at or before end. */ |
fa8ed3e0 RS |
770 | if (start_byte <= ceiling_byte) |
771 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
772 | |
773 | /* Now the text before start is an unknown region, and | |
774 | next_change is the position of the next known region. */ | |
c8b9f1bc | 775 | ceiling_byte = max (CHAR_TO_BYTE (next_change), ceiling_byte); |
9169c321 | 776 | } |
c8b9f1bc EZ |
777 | else |
778 | start_byte = CHAR_TO_BYTE (start); | |
9169c321 JB |
779 | |
780 | /* Stop scanning before the gap. */ | |
67ce527d KH |
781 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
782 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
783 | |
784 | { | |
785 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
786 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
787 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
788 | unsigned char *base = cursor; |
789 | ||
790 | while (cursor >= ceiling_addr) | |
791 | { | |
792 | unsigned char *scan_start = cursor; | |
793 | ||
794 | while (*cursor != target && --cursor >= ceiling_addr) | |
795 | ; | |
796 | ||
797 | /* If we're looking for newlines, cache the fact that | |
798 | the region from after the cursor to start is free of them. */ | |
799 | if (target == '\n' && newline_cache) | |
800 | know_region_cache (current_buffer, newline_cache, | |
6c1bd3f3 EZ |
801 | BYTE_TO_CHAR (start_byte + cursor - base), |
802 | BYTE_TO_CHAR (start_byte + scan_start - base)); | |
9169c321 JB |
803 | |
804 | /* Did we find the target character? */ | |
805 | if (cursor >= ceiling_addr) | |
806 | { | |
807 | if (++count >= 0) | |
808 | { | |
809 | immediate_quit = 0; | |
fa8ed3e0 | 810 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
811 | } |
812 | cursor--; | |
813 | } | |
814 | } | |
815 | ||
fa8ed3e0 | 816 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
817 | } |
818 | } | |
819 | ||
ca1d1d23 JB |
820 | immediate_quit = 0; |
821 | if (shortage != 0) | |
ffd56f97 | 822 | *shortage = count * direction; |
9169c321 | 823 | return start; |
ca1d1d23 | 824 | } |
fa8ed3e0 RS |
825 | \f |
826 | /* Search for COUNT instances of a line boundary, which means either a | |
827 | newline or (if selective display enabled) a carriage return. | |
828 | Start at START. If COUNT is negative, search backwards. | |
829 | ||
830 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
831 | ||
832 | If we find COUNT instances. we position after (always after, | |
833 | even if scanning backwards) the COUNTth match, and return 0. | |
834 | ||
835 | If we don't find COUNT instances before reaching the end of the | |
836 | buffer (or the beginning, if scanning backwards), we return | |
837 | the number of line boundaries left unfound, and position at | |
838 | the limit we bumped up against. | |
839 | ||
840 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 841 | except in special cases. */ |
ca1d1d23 | 842 | |
c098fdb8 | 843 | EMACS_INT |
d311d28c PE |
844 | scan_newline (ptrdiff_t start, ptrdiff_t start_byte, |
845 | ptrdiff_t limit, ptrdiff_t limit_byte, | |
c098fdb8 | 846 | register EMACS_INT count, int allow_quit) |
63fa018d | 847 | { |
fa8ed3e0 RS |
848 | int direction = ((count > 0) ? 1 : -1); |
849 | ||
850 | register unsigned char *cursor; | |
851 | unsigned char *base; | |
852 | ||
d311d28c | 853 | ptrdiff_t ceiling; |
fa8ed3e0 RS |
854 | register unsigned char *ceiling_addr; |
855 | ||
d5d57b92 RS |
856 | int old_immediate_quit = immediate_quit; |
857 | ||
fa8ed3e0 RS |
858 | /* The code that follows is like scan_buffer |
859 | but checks for either newline or carriage return. */ | |
860 | ||
d5d57b92 RS |
861 | if (allow_quit) |
862 | immediate_quit++; | |
fa8ed3e0 RS |
863 | |
864 | start_byte = CHAR_TO_BYTE (start); | |
865 | ||
866 | if (count > 0) | |
867 | { | |
868 | while (start_byte < limit_byte) | |
869 | { | |
870 | ceiling = BUFFER_CEILING_OF (start_byte); | |
871 | ceiling = min (limit_byte - 1, ceiling); | |
872 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
873 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
874 | while (1) | |
875 | { | |
876 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
877 | ; | |
878 | ||
879 | if (cursor != ceiling_addr) | |
880 | { | |
881 | if (--count == 0) | |
882 | { | |
d5d57b92 | 883 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
884 | start_byte = start_byte + cursor - base + 1; |
885 | start = BYTE_TO_CHAR (start_byte); | |
886 | TEMP_SET_PT_BOTH (start, start_byte); | |
887 | return 0; | |
888 | } | |
889 | else | |
890 | if (++cursor == ceiling_addr) | |
891 | break; | |
892 | } | |
893 | else | |
894 | break; | |
895 | } | |
896 | start_byte += cursor - base; | |
897 | } | |
898 | } | |
899 | else | |
900 | { | |
fa8ed3e0 RS |
901 | while (start_byte > limit_byte) |
902 | { | |
903 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
904 | ceiling = max (limit_byte, ceiling); | |
905 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
906 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
907 | while (1) | |
908 | { | |
909 | while (--cursor != ceiling_addr && *cursor != '\n') | |
910 | ; | |
911 | ||
912 | if (cursor != ceiling_addr) | |
913 | { | |
914 | if (++count == 0) | |
915 | { | |
d5d57b92 | 916 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
917 | /* Return the position AFTER the match we found. */ |
918 | start_byte = start_byte + cursor - base + 1; | |
919 | start = BYTE_TO_CHAR (start_byte); | |
920 | TEMP_SET_PT_BOTH (start, start_byte); | |
921 | return 0; | |
922 | } | |
923 | } | |
924 | else | |
925 | break; | |
926 | } | |
927 | /* Here we add 1 to compensate for the last decrement | |
928 | of CURSOR, which took it past the valid range. */ | |
929 | start_byte += cursor - base + 1; | |
930 | } | |
931 | } | |
932 | ||
933 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 934 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
935 | |
936 | return count * direction; | |
63fa018d RS |
937 | } |
938 | ||
d311d28c PE |
939 | ptrdiff_t |
940 | find_next_newline_no_quit (ptrdiff_t from, ptrdiff_t cnt) | |
ca1d1d23 | 941 | { |
d311d28c | 942 | return scan_buffer ('\n', from, 0, cnt, (ptrdiff_t *) 0, 0); |
9169c321 JB |
943 | } |
944 | ||
9169c321 JB |
945 | /* Like find_next_newline, but returns position before the newline, |
946 | not after, and only search up to TO. This isn't just | |
947 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 948 | |
d311d28c PE |
949 | ptrdiff_t |
950 | find_before_next_newline (ptrdiff_t from, ptrdiff_t to, ptrdiff_t cnt) | |
9169c321 | 951 | { |
d311d28c PE |
952 | ptrdiff_t shortage; |
953 | ptrdiff_t pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); | |
9169c321 JB |
954 | |
955 | if (shortage == 0) | |
956 | pos--; | |
177c0ea7 | 957 | |
9169c321 | 958 | return pos; |
ca1d1d23 JB |
959 | } |
960 | \f | |
ca1d1d23 JB |
961 | /* Subroutines of Lisp buffer search functions. */ |
962 | ||
963 | static Lisp_Object | |
c098fdb8 EZ |
964 | search_command (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, |
965 | Lisp_Object count, int direction, int RE, int posix) | |
ca1d1d23 | 966 | { |
a53e2e89 | 967 | register EMACS_INT np; |
d311d28c PE |
968 | EMACS_INT lim; |
969 | ptrdiff_t lim_byte; | |
a53e2e89 | 970 | EMACS_INT n = direction; |
ca1d1d23 JB |
971 | |
972 | if (!NILP (count)) | |
973 | { | |
b7826503 | 974 | CHECK_NUMBER (count); |
ca1d1d23 JB |
975 | n *= XINT (count); |
976 | } | |
977 | ||
b7826503 | 978 | CHECK_STRING (string); |
ca1d1d23 | 979 | if (NILP (bound)) |
9f43ad85 RS |
980 | { |
981 | if (n > 0) | |
982 | lim = ZV, lim_byte = ZV_BYTE; | |
983 | else | |
984 | lim = BEGV, lim_byte = BEGV_BYTE; | |
985 | } | |
ca1d1d23 JB |
986 | else |
987 | { | |
b7826503 | 988 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 989 | lim = XINT (bound); |
6ec8bbd2 | 990 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
991 | error ("Invalid search bound (wrong side of point)"); |
992 | if (lim > ZV) | |
9f43ad85 | 993 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 994 | else if (lim < BEGV) |
9f43ad85 | 995 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
996 | else |
997 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
998 | } |
999 | ||
910c747a | 1000 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
34dabdb7 | 1001 | set_char_table_extras (BVAR (current_buffer, case_canon_table), 2, |
a098c930 | 1002 | BVAR (current_buffer, case_eqv_table)); |
910c747a | 1003 | |
9f43ad85 | 1004 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
4b4deea2 TT |
1005 | (!NILP (BVAR (current_buffer, case_fold_search)) |
1006 | ? BVAR (current_buffer, case_canon_table) | |
3135e9fd | 1007 | : Qnil), |
4b4deea2 TT |
1008 | (!NILP (BVAR (current_buffer, case_fold_search)) |
1009 | ? BVAR (current_buffer, case_eqv_table) | |
3135e9fd | 1010 | : Qnil), |
b819a390 | 1011 | posix); |
ca1d1d23 JB |
1012 | if (np <= 0) |
1013 | { | |
1014 | if (NILP (noerror)) | |
06f77a8a KS |
1015 | xsignal1 (Qsearch_failed, string); |
1016 | ||
ca1d1d23 JB |
1017 | if (!EQ (noerror, Qt)) |
1018 | { | |
1019 | if (lim < BEGV || lim > ZV) | |
1088b922 | 1020 | emacs_abort (); |
9f43ad85 | 1021 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
1022 | return Qnil; |
1023 | #if 0 /* This would be clean, but maybe programs depend on | |
1024 | a value of nil here. */ | |
481399bf | 1025 | np = lim; |
a5f217b8 | 1026 | #endif |
ca1d1d23 | 1027 | } |
481399bf RS |
1028 | else |
1029 | return Qnil; | |
ca1d1d23 JB |
1030 | } |
1031 | ||
1032 | if (np < BEGV || np > ZV) | |
1088b922 | 1033 | emacs_abort (); |
ca1d1d23 JB |
1034 | |
1035 | SET_PT (np); | |
1036 | ||
1037 | return make_number (np); | |
1038 | } | |
1039 | \f | |
fa8ed3e0 RS |
1040 | /* Return 1 if REGEXP it matches just one constant string. */ |
1041 | ||
b6d6a51c | 1042 | static int |
971de7fb | 1043 | trivial_regexp_p (Lisp_Object regexp) |
b6d6a51c | 1044 | { |
d311d28c | 1045 | ptrdiff_t len = SBYTES (regexp); |
d5db4077 | 1046 | unsigned char *s = SDATA (regexp); |
b6d6a51c KH |
1047 | while (--len >= 0) |
1048 | { | |
1049 | switch (*s++) | |
1050 | { | |
1051 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
1052 | return 0; | |
1053 | case '\\': | |
1054 | if (--len < 0) | |
1055 | return 0; | |
1056 | switch (*s++) | |
1057 | { | |
1058 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
1059 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 1060 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 1061 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 1062 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
1063 | case '6': case '7': case '8': case '9': |
1064 | return 0; | |
1065 | } | |
1066 | } | |
1067 | } | |
1068 | return 1; | |
1069 | } | |
1070 | ||
ca325161 | 1071 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 1072 | starting at position POS and stopping at position LIM, |
b819a390 | 1073 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
1074 | a regular expression if RE is true. |
1075 | ||
1076 | If N is positive, searching is forward and LIM must be greater than POS. | |
1077 | If N is negative, searching is backward and LIM must be less than POS. | |
1078 | ||
facdc750 | 1079 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 1080 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
1081 | (if searching backward) or the end (if searching forward). |
1082 | ||
1083 | POSIX is nonzero if we want full backtracking (POSIX style) | |
1084 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 1085 | |
aff2ce94 RS |
1086 | #define TRANSLATE(out, trt, d) \ |
1087 | do \ | |
1088 | { \ | |
1089 | if (! NILP (trt)) \ | |
1090 | { \ | |
1091 | Lisp_Object temp; \ | |
1092 | temp = Faref (trt, make_number (d)); \ | |
1093 | if (INTEGERP (temp)) \ | |
1094 | out = XINT (temp); \ | |
1095 | else \ | |
1096 | out = d; \ | |
1097 | } \ | |
1098 | else \ | |
1099 | out = d; \ | |
1100 | } \ | |
1101 | while (0) | |
facdc750 | 1102 | |
ef887810 RS |
1103 | /* Only used in search_buffer, to record the end position of the match |
1104 | when searching regexps and SEARCH_REGS should not be changed | |
1105 | (i.e. Vinhibit_changing_match_data is non-nil). */ | |
1106 | static struct re_registers search_regs_1; | |
1107 | ||
e7deaab0 | 1108 | static EMACS_INT |
d311d28c PE |
1109 | search_buffer (Lisp_Object string, ptrdiff_t pos, ptrdiff_t pos_byte, |
1110 | ptrdiff_t lim, ptrdiff_t lim_byte, EMACS_INT n, | |
d5a3eaaf | 1111 | int RE, Lisp_Object trt, Lisp_Object inverse_trt, int posix) |
ca1d1d23 | 1112 | { |
d311d28c PE |
1113 | ptrdiff_t len = SCHARS (string); |
1114 | ptrdiff_t len_byte = SBYTES (string); | |
1115 | register ptrdiff_t i; | |
ca1d1d23 | 1116 | |
7074fde6 FP |
1117 | if (running_asynch_code) |
1118 | save_search_regs (); | |
1119 | ||
a7e4cdde | 1120 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1121 | /* Null string is found at starting position. */ |
a7e4cdde | 1122 | if (len == 0 || n == 0) |
ca325161 | 1123 | { |
0353b28f | 1124 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1125 | return pos; |
1126 | } | |
3f57a499 | 1127 | |
41a33295 | 1128 | if (RE && !(trivial_regexp_p (string) && NILP (Vsearch_spaces_regexp))) |
ca1d1d23 | 1129 | { |
facdc750 | 1130 | unsigned char *p1, *p2; |
d311d28c | 1131 | ptrdiff_t s1, s2; |
487282dc KH |
1132 | struct re_pattern_buffer *bufp; |
1133 | ||
ef887810 RS |
1134 | bufp = compile_pattern (string, |
1135 | (NILP (Vinhibit_changing_match_data) | |
1136 | ? &search_regs : &search_regs_1), | |
1137 | trt, posix, | |
4b4deea2 | 1138 | !NILP (BVAR (current_buffer, enable_multibyte_characters))); |
ca1d1d23 | 1139 | |
ca1d1d23 JB |
1140 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1141 | because letting this function finish | |
1142 | can take too long. */ | |
1143 | QUIT; /* Do a pending quit right away, | |
1144 | to avoid paradoxical behavior */ | |
1145 | /* Get pointers and sizes of the two strings | |
1146 | that make up the visible portion of the buffer. */ | |
1147 | ||
1148 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1149 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1150 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1151 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1152 | if (s1 < 0) |
1153 | { | |
1154 | p2 = p1; | |
fa8ed3e0 | 1155 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1156 | s1 = 0; |
1157 | } | |
1158 | if (s2 < 0) | |
1159 | { | |
fa8ed3e0 | 1160 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1161 | s2 = 0; |
1162 | } | |
8bb43c28 | 1163 | re_match_object = Qnil; |
177c0ea7 | 1164 | |
ca1d1d23 JB |
1165 | while (n < 0) |
1166 | { | |
d311d28c | 1167 | ptrdiff_t val; |
52c55cc7 | 1168 | |
487282dc | 1169 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1170 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1171 | (NILP (Vinhibit_changing_match_data) |
1172 | ? &search_regs : &search_regs_1), | |
42db823b | 1173 | /* Don't allow match past current point */ |
4996330b | 1174 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1175 | if (val == -2) |
b6d6a51c KH |
1176 | { |
1177 | matcher_overflow (); | |
1178 | } | |
ca1d1d23 JB |
1179 | if (val >= 0) |
1180 | { | |
ef887810 RS |
1181 | if (NILP (Vinhibit_changing_match_data)) |
1182 | { | |
1183 | pos_byte = search_regs.start[0] + BEGV_BYTE; | |
1184 | for (i = 0; i < search_regs.num_regs; i++) | |
1185 | if (search_regs.start[i] >= 0) | |
1186 | { | |
1187 | search_regs.start[i] | |
1188 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1189 | search_regs.end[i] | |
1190 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1191 | } | |
1192 | XSETBUFFER (last_thing_searched, current_buffer); | |
1193 | /* Set pos to the new position. */ | |
1194 | pos = search_regs.start[0]; | |
1195 | } | |
1196 | else | |
1197 | { | |
1198 | pos_byte = search_regs_1.start[0] + BEGV_BYTE; | |
1199 | /* Set pos to the new position. */ | |
1200 | pos = BYTE_TO_CHAR (search_regs_1.start[0] + BEGV_BYTE); | |
1201 | } | |
ca1d1d23 JB |
1202 | } |
1203 | else | |
1204 | { | |
1205 | immediate_quit = 0; | |
1206 | return (n); | |
1207 | } | |
1208 | n++; | |
1209 | } | |
1210 | while (n > 0) | |
1211 | { | |
d311d28c | 1212 | ptrdiff_t val; |
52c55cc7 | 1213 | |
487282dc | 1214 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1215 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1216 | (NILP (Vinhibit_changing_match_data) |
1217 | ? &search_regs : &search_regs_1), | |
4996330b | 1218 | lim_byte - BEGV_BYTE); |
ca1d1d23 | 1219 | if (val == -2) |
b6d6a51c KH |
1220 | { |
1221 | matcher_overflow (); | |
1222 | } | |
ca1d1d23 JB |
1223 | if (val >= 0) |
1224 | { | |
ef887810 RS |
1225 | if (NILP (Vinhibit_changing_match_data)) |
1226 | { | |
1227 | pos_byte = search_regs.end[0] + BEGV_BYTE; | |
1228 | for (i = 0; i < search_regs.num_regs; i++) | |
1229 | if (search_regs.start[i] >= 0) | |
1230 | { | |
1231 | search_regs.start[i] | |
1232 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1233 | search_regs.end[i] | |
1234 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1235 | } | |
1236 | XSETBUFFER (last_thing_searched, current_buffer); | |
1237 | pos = search_regs.end[0]; | |
1238 | } | |
1239 | else | |
1240 | { | |
1241 | pos_byte = search_regs_1.end[0] + BEGV_BYTE; | |
1242 | pos = BYTE_TO_CHAR (search_regs_1.end[0] + BEGV_BYTE); | |
1243 | } | |
ca1d1d23 JB |
1244 | } |
1245 | else | |
1246 | { | |
1247 | immediate_quit = 0; | |
1248 | return (0 - n); | |
1249 | } | |
1250 | n--; | |
1251 | } | |
1252 | immediate_quit = 0; | |
1253 | return (pos); | |
1254 | } | |
1255 | else /* non-RE case */ | |
1256 | { | |
facdc750 | 1257 | unsigned char *raw_pattern, *pat; |
d311d28c PE |
1258 | ptrdiff_t raw_pattern_size; |
1259 | ptrdiff_t raw_pattern_size_byte; | |
facdc750 | 1260 | unsigned char *patbuf; |
4b4deea2 | 1261 | int multibyte = !NILP (BVAR (current_buffer, enable_multibyte_characters)); |
a967ed62 | 1262 | unsigned char *base_pat; |
49e44e81 KH |
1263 | /* Set to positive if we find a non-ASCII char that need |
1264 | translation. Otherwise set to zero later. */ | |
1265 | int char_base = -1; | |
040272ce | 1266 | int boyer_moore_ok = 1; |
facdc750 RS |
1267 | |
1268 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1269 | We must convert PATTERN to match that, or we will not really | |
1270 | find things right. */ | |
1271 | ||
1272 | if (multibyte == STRING_MULTIBYTE (string)) | |
1273 | { | |
51b59d79 | 1274 | raw_pattern = SDATA (string); |
d5db4077 KR |
1275 | raw_pattern_size = SCHARS (string); |
1276 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1277 | } |
1278 | else if (multibyte) | |
1279 | { | |
d5db4077 | 1280 | raw_pattern_size = SCHARS (string); |
facdc750 | 1281 | raw_pattern_size_byte |
d5db4077 | 1282 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1283 | raw_pattern_size); |
38182d90 | 1284 | raw_pattern = alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1285 | copy_text (SDATA (string), raw_pattern, |
1286 | SCHARS (string), 0, 1); | |
facdc750 RS |
1287 | } |
1288 | else | |
1289 | { | |
1290 | /* Converting multibyte to single-byte. | |
1291 | ||
1292 | ??? Perhaps this conversion should be done in a special way | |
1293 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1294 | so that only the multibyte chars which really correspond to | |
1295 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1296 | raw_pattern_size = SCHARS (string); |
1297 | raw_pattern_size_byte = SCHARS (string); | |
38182d90 | 1298 | raw_pattern = alloca (raw_pattern_size + 1); |
d5db4077 KR |
1299 | copy_text (SDATA (string), raw_pattern, |
1300 | SBYTES (string), 1, 0); | |
facdc750 RS |
1301 | } |
1302 | ||
1303 | /* Copy and optionally translate the pattern. */ | |
1304 | len = raw_pattern_size; | |
1305 | len_byte = raw_pattern_size_byte; | |
38182d90 | 1306 | patbuf = alloca (len * MAX_MULTIBYTE_LENGTH); |
facdc750 RS |
1307 | pat = patbuf; |
1308 | base_pat = raw_pattern; | |
1309 | if (multibyte) | |
1310 | { | |
a17ae96a KH |
1311 | /* Fill patbuf by translated characters in STRING while |
1312 | checking if we can use boyer-moore search. If TRT is | |
1313 | non-nil, we can use boyer-moore search only if TRT can be | |
1314 | represented by the byte array of 256 elements. For that, | |
9858f6c3 | 1315 | all non-ASCII case-equivalents of all case-sensitive |
a17ae96a KH |
1316 | characters in STRING must belong to the same charset and |
1317 | row. */ | |
1318 | ||
facdc750 RS |
1319 | while (--len >= 0) |
1320 | { | |
a17ae96a | 1321 | unsigned char str_base[MAX_MULTIBYTE_LENGTH], *str; |
aff2ce94 | 1322 | int c, translated, inverse; |
a17ae96a | 1323 | int in_charlen, charlen; |
facdc750 RS |
1324 | |
1325 | /* If we got here and the RE flag is set, it's because we're | |
1326 | dealing with a regexp known to be trivial, so the backslash | |
1327 | just quotes the next character. */ | |
1328 | if (RE && *base_pat == '\\') | |
1329 | { | |
1330 | len--; | |
62221f22 | 1331 | raw_pattern_size--; |
facdc750 RS |
1332 | len_byte--; |
1333 | base_pat++; | |
1334 | } | |
1335 | ||
62a6e103 | 1336 | c = STRING_CHAR_AND_LENGTH (base_pat, in_charlen); |
040272ce | 1337 | |
a17ae96a | 1338 | if (NILP (trt)) |
facdc750 | 1339 | { |
a17ae96a KH |
1340 | str = base_pat; |
1341 | charlen = in_charlen; | |
1342 | } | |
1343 | else | |
1344 | { | |
1345 | /* Translate the character. */ | |
1346 | TRANSLATE (translated, trt, c); | |
1347 | charlen = CHAR_STRING (translated, str_base); | |
1348 | str = str_base; | |
1349 | ||
1350 | /* Check if C has any other case-equivalents. */ | |
1351 | TRANSLATE (inverse, inverse_trt, c); | |
1352 | /* If so, check if we can use boyer-moore. */ | |
1353 | if (c != inverse && boyer_moore_ok) | |
1354 | { | |
1355 | /* Check if all equivalents belong to the same | |
1356 | group of characters. Note that the check of C | |
49e44e81 KH |
1357 | itself is done by the last iteration. */ |
1358 | int this_char_base = -1; | |
1359 | ||
1360 | while (boyer_moore_ok) | |
a17ae96a | 1361 | { |
49e44e81 | 1362 | if (ASCII_BYTE_P (inverse)) |
a17ae96a | 1363 | { |
49e44e81 KH |
1364 | if (this_char_base > 0) |
1365 | boyer_moore_ok = 0; | |
1366 | else | |
dafaabd1 | 1367 | this_char_base = 0; |
a17ae96a | 1368 | } |
49e44e81 KH |
1369 | else if (CHAR_BYTE8_P (inverse)) |
1370 | /* Boyer-moore search can't handle a | |
1371 | translation of an eight-bit | |
1372 | character. */ | |
1373 | boyer_moore_ok = 0; | |
1374 | else if (this_char_base < 0) | |
1375 | { | |
1376 | this_char_base = inverse & ~0x3F; | |
1377 | if (char_base < 0) | |
1378 | char_base = this_char_base; | |
dafaabd1 | 1379 | else if (this_char_base != char_base) |
49e44e81 KH |
1380 | boyer_moore_ok = 0; |
1381 | } | |
1382 | else if ((inverse & ~0x3F) != this_char_base) | |
1383 | boyer_moore_ok = 0; | |
a17ae96a KH |
1384 | if (c == inverse) |
1385 | break; | |
1386 | TRANSLATE (inverse, inverse_trt, inverse); | |
1387 | } | |
1388 | } | |
aff2ce94 | 1389 | } |
facdc750 RS |
1390 | |
1391 | /* Store this character into the translated pattern. */ | |
72af86bd | 1392 | memcpy (pat, str, charlen); |
a17ae96a | 1393 | pat += charlen; |
facdc750 RS |
1394 | base_pat += in_charlen; |
1395 | len_byte -= in_charlen; | |
1396 | } | |
dafaabd1 AS |
1397 | |
1398 | /* If char_base is still negative we didn't find any translated | |
1399 | non-ASCII characters. */ | |
1400 | if (char_base < 0) | |
1401 | char_base = 0; | |
facdc750 RS |
1402 | } |
1403 | else | |
1404 | { | |
040272ce | 1405 | /* Unibyte buffer. */ |
a17ae96a | 1406 | char_base = 0; |
facdc750 RS |
1407 | while (--len >= 0) |
1408 | { | |
040272ce | 1409 | int c, translated; |
facdc750 RS |
1410 | |
1411 | /* If we got here and the RE flag is set, it's because we're | |
1412 | dealing with a regexp known to be trivial, so the backslash | |
1413 | just quotes the next character. */ | |
1414 | if (RE && *base_pat == '\\') | |
1415 | { | |
1416 | len--; | |
0190922f | 1417 | raw_pattern_size--; |
facdc750 RS |
1418 | base_pat++; |
1419 | } | |
1420 | c = *base_pat++; | |
aff2ce94 | 1421 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1422 | *pat++ = translated; |
1423 | } | |
1424 | } | |
1425 | ||
1426 | len_byte = pat - patbuf; | |
facdc750 RS |
1427 | pat = base_pat = patbuf; |
1428 | ||
040272ce | 1429 | if (boyer_moore_ok) |
1db5b1ad JB |
1430 | return boyer_moore (n, pat, len_byte, trt, inverse_trt, |
1431 | pos_byte, lim_byte, | |
a17ae96a | 1432 | char_base); |
facdc750 | 1433 | else |
1db5b1ad | 1434 | return simple_search (n, pat, raw_pattern_size, len_byte, trt, |
facdc750 RS |
1435 | pos, pos_byte, lim, lim_byte); |
1436 | } | |
1437 | } | |
1438 | \f | |
1439 | /* Do a simple string search N times for the string PAT, | |
1440 | whose length is LEN/LEN_BYTE, | |
1441 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1442 | TRT is the translation table. | |
f8bd51c4 | 1443 | |
facdc750 RS |
1444 | Return the character position where the match is found. |
1445 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1446 | |
facdc750 RS |
1447 | This kind of search works regardless of what is in PAT and |
1448 | regardless of what is in TRT. It is used in cases where | |
1449 | boyer_moore cannot work. */ | |
1450 | ||
e7deaab0 | 1451 | static EMACS_INT |
c098fdb8 | 1452 | simple_search (EMACS_INT n, unsigned char *pat, |
d311d28c PE |
1453 | ptrdiff_t len, ptrdiff_t len_byte, Lisp_Object trt, |
1454 | ptrdiff_t pos, ptrdiff_t pos_byte, | |
1455 | ptrdiff_t lim, ptrdiff_t lim_byte) | |
facdc750 | 1456 | { |
4b4deea2 | 1457 | int multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
ab228c24 | 1458 | int forward = n > 0; |
49e44e81 KH |
1459 | /* Number of buffer bytes matched. Note that this may be different |
1460 | from len_byte in a multibyte buffer. */ | |
8ce70ed2 | 1461 | ptrdiff_t match_byte = PTRDIFF_MIN; |
facdc750 RS |
1462 | |
1463 | if (lim > pos && multibyte) | |
1464 | while (n > 0) | |
1465 | { | |
1466 | while (1) | |
f8bd51c4 | 1467 | { |
facdc750 | 1468 | /* Try matching at position POS. */ |
d311d28c PE |
1469 | ptrdiff_t this_pos = pos; |
1470 | ptrdiff_t this_pos_byte = pos_byte; | |
1471 | ptrdiff_t this_len = len; | |
facdc750 | 1472 | unsigned char *p = pat; |
a9469498 | 1473 | if (pos + len > lim || pos_byte + len_byte > lim_byte) |
facdc750 RS |
1474 | goto stop; |
1475 | ||
1476 | while (this_len > 0) | |
1477 | { | |
1478 | int charlen, buf_charlen; | |
ab228c24 | 1479 | int pat_ch, buf_ch; |
facdc750 | 1480 | |
62a6e103 | 1481 | pat_ch = STRING_CHAR_AND_LENGTH (p, charlen); |
facdc750 | 1482 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
facdc750 | 1483 | buf_charlen); |
aff2ce94 | 1484 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1485 | |
1486 | if (buf_ch != pat_ch) | |
1487 | break; | |
ab228c24 | 1488 | |
ab228c24 RS |
1489 | this_len--; |
1490 | p += charlen; | |
1491 | ||
1492 | this_pos_byte += buf_charlen; | |
1493 | this_pos++; | |
facdc750 RS |
1494 | } |
1495 | ||
1496 | if (this_len == 0) | |
1497 | { | |
49e44e81 | 1498 | match_byte = this_pos_byte - pos_byte; |
facdc750 | 1499 | pos += len; |
49e44e81 | 1500 | pos_byte += match_byte; |
facdc750 RS |
1501 | break; |
1502 | } | |
1503 | ||
1504 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1505 | } |
facdc750 RS |
1506 | |
1507 | n--; | |
1508 | } | |
1509 | else if (lim > pos) | |
1510 | while (n > 0) | |
1511 | { | |
1512 | while (1) | |
f8bd51c4 | 1513 | { |
facdc750 | 1514 | /* Try matching at position POS. */ |
d311d28c PE |
1515 | ptrdiff_t this_pos = pos; |
1516 | ptrdiff_t this_len = len; | |
facdc750 RS |
1517 | unsigned char *p = pat; |
1518 | ||
1519 | if (pos + len > lim) | |
1520 | goto stop; | |
1521 | ||
1522 | while (this_len > 0) | |
1523 | { | |
1524 | int pat_ch = *p++; | |
1525 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1526 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1527 | |
1528 | if (buf_ch != pat_ch) | |
1529 | break; | |
ab228c24 RS |
1530 | |
1531 | this_len--; | |
1532 | this_pos++; | |
facdc750 RS |
1533 | } |
1534 | ||
1535 | if (this_len == 0) | |
1536 | { | |
49e44e81 | 1537 | match_byte = len; |
facdc750 RS |
1538 | pos += len; |
1539 | break; | |
1540 | } | |
1541 | ||
1542 | pos++; | |
f8bd51c4 | 1543 | } |
facdc750 RS |
1544 | |
1545 | n--; | |
1546 | } | |
1547 | /* Backwards search. */ | |
1548 | else if (lim < pos && multibyte) | |
1549 | while (n < 0) | |
1550 | { | |
1551 | while (1) | |
f8bd51c4 | 1552 | { |
facdc750 | 1553 | /* Try matching at position POS. */ |
d311d28c PE |
1554 | ptrdiff_t this_pos = pos; |
1555 | ptrdiff_t this_pos_byte = pos_byte; | |
1556 | ptrdiff_t this_len = len; | |
c525b3f2 | 1557 | const unsigned char *p = pat + len_byte; |
facdc750 | 1558 | |
8b264ecb | 1559 | if (this_pos - len < lim || (pos_byte - len_byte) < lim_byte) |
facdc750 RS |
1560 | goto stop; |
1561 | ||
1562 | while (this_len > 0) | |
1563 | { | |
ab228c24 | 1564 | int pat_ch, buf_ch; |
facdc750 | 1565 | |
8b264ecb AS |
1566 | DEC_BOTH (this_pos, this_pos_byte); |
1567 | PREV_CHAR_BOUNDARY (p, pat); | |
1568 | pat_ch = STRING_CHAR (p); | |
1569 | buf_ch = STRING_CHAR (BYTE_POS_ADDR (this_pos_byte)); | |
aff2ce94 | 1570 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1571 | |
1572 | if (buf_ch != pat_ch) | |
1573 | break; | |
ab228c24 | 1574 | |
ab228c24 | 1575 | this_len--; |
facdc750 RS |
1576 | } |
1577 | ||
1578 | if (this_len == 0) | |
1579 | { | |
8b264ecb AS |
1580 | match_byte = pos_byte - this_pos_byte; |
1581 | pos = this_pos; | |
1582 | pos_byte = this_pos_byte; | |
facdc750 RS |
1583 | break; |
1584 | } | |
1585 | ||
1586 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1587 | } |
1588 | ||
facdc750 RS |
1589 | n++; |
1590 | } | |
1591 | else if (lim < pos) | |
1592 | while (n < 0) | |
1593 | { | |
1594 | while (1) | |
b6d6a51c | 1595 | { |
facdc750 | 1596 | /* Try matching at position POS. */ |
d311d28c PE |
1597 | ptrdiff_t this_pos = pos - len; |
1598 | ptrdiff_t this_len = len; | |
facdc750 RS |
1599 | unsigned char *p = pat; |
1600 | ||
a9469498 | 1601 | if (this_pos < lim) |
facdc750 RS |
1602 | goto stop; |
1603 | ||
1604 | while (this_len > 0) | |
1605 | { | |
1606 | int pat_ch = *p++; | |
1607 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1608 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1609 | |
1610 | if (buf_ch != pat_ch) | |
1611 | break; | |
ab228c24 RS |
1612 | this_len--; |
1613 | this_pos++; | |
facdc750 RS |
1614 | } |
1615 | ||
1616 | if (this_len == 0) | |
b6d6a51c | 1617 | { |
49e44e81 | 1618 | match_byte = len; |
facdc750 RS |
1619 | pos -= len; |
1620 | break; | |
b6d6a51c | 1621 | } |
facdc750 RS |
1622 | |
1623 | pos--; | |
b6d6a51c | 1624 | } |
facdc750 RS |
1625 | |
1626 | n++; | |
b6d6a51c | 1627 | } |
facdc750 RS |
1628 | |
1629 | stop: | |
1630 | if (n == 0) | |
aff2ce94 | 1631 | { |
8ce70ed2 | 1632 | eassert (match_byte != PTRDIFF_MIN); |
ab228c24 | 1633 | if (forward) |
49e44e81 | 1634 | set_search_regs ((multibyte ? pos_byte : pos) - match_byte, match_byte); |
ab228c24 | 1635 | else |
49e44e81 | 1636 | set_search_regs (multibyte ? pos_byte : pos, match_byte); |
aff2ce94 RS |
1637 | |
1638 | return pos; | |
1639 | } | |
facdc750 RS |
1640 | else if (n > 0) |
1641 | return -n; | |
1642 | else | |
1643 | return n; | |
1644 | } | |
1645 | \f | |
0190922f | 1646 | /* Do Boyer-Moore search N times for the string BASE_PAT, |
1db5b1ad JB |
1647 | whose length is LEN_BYTE, |
1648 | from buffer position POS_BYTE until LIM_BYTE. | |
facdc750 RS |
1649 | DIRECTION says which direction we search in. |
1650 | TRT and INVERSE_TRT are translation tables. | |
0190922f | 1651 | Characters in PAT are already translated by TRT. |
facdc750 | 1652 | |
0190922f KH |
1653 | This kind of search works if all the characters in BASE_PAT that |
1654 | have nontrivial translation are the same aside from the last byte. | |
1655 | This makes it possible to translate just the last byte of a | |
1656 | character, and do so after just a simple test of the context. | |
b2e6b10f | 1657 | CHAR_BASE is nonzero if there is such a non-ASCII character. |
facdc750 RS |
1658 | |
1659 | If that criterion is not satisfied, do not call this function. */ | |
1660 | ||
e7deaab0 | 1661 | static EMACS_INT |
c098fdb8 | 1662 | boyer_moore (EMACS_INT n, unsigned char *base_pat, |
d311d28c | 1663 | ptrdiff_t len_byte, |
d5a3eaaf | 1664 | Lisp_Object trt, Lisp_Object inverse_trt, |
d311d28c | 1665 | ptrdiff_t pos_byte, ptrdiff_t lim_byte, |
1db5b1ad | 1666 | int char_base) |
facdc750 RS |
1667 | { |
1668 | int direction = ((n > 0) ? 1 : -1); | |
d311d28c PE |
1669 | register ptrdiff_t dirlen; |
1670 | ptrdiff_t limit; | |
e7deaab0 | 1671 | int stride_for_teases = 0; |
f4646fff | 1672 | int BM_tab[0400]; |
177c0ea7 | 1673 | register unsigned char *cursor, *p_limit; |
d311d28c | 1674 | register ptrdiff_t i; |
c098fdb8 | 1675 | register int j; |
cb6792d2 | 1676 | unsigned char *pat, *pat_end; |
4b4deea2 | 1677 | int multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
facdc750 RS |
1678 | |
1679 | unsigned char simple_translate[0400]; | |
0190922f | 1680 | /* These are set to the preceding bytes of a byte to be translated |
49e44e81 | 1681 | if char_base is nonzero. As the maximum byte length of a |
a17ae96a | 1682 | multibyte character is 5, we have to check at most four previous |
0190922f KH |
1683 | bytes. */ |
1684 | int translate_prev_byte1 = 0; | |
1685 | int translate_prev_byte2 = 0; | |
1686 | int translate_prev_byte3 = 0; | |
facdc750 | 1687 | |
f4646fff AS |
1688 | /* The general approach is that we are going to maintain that we know |
1689 | the first (closest to the present position, in whatever direction | |
1690 | we're searching) character that could possibly be the last | |
1691 | (furthest from present position) character of a valid match. We | |
1692 | advance the state of our knowledge by looking at that character | |
1693 | and seeing whether it indeed matches the last character of the | |
1694 | pattern. If it does, we take a closer look. If it does not, we | |
1695 | move our pointer (to putative last characters) as far as is | |
1696 | logically possible. This amount of movement, which I call a | |
1697 | stride, will be the length of the pattern if the actual character | |
1698 | appears nowhere in the pattern, otherwise it will be the distance | |
1699 | from the last occurrence of that character to the end of the | |
1700 | pattern. If the amount is zero we have a possible match. */ | |
1701 | ||
1702 | /* Here we make a "mickey mouse" BM table. The stride of the search | |
1703 | is determined only by the last character of the putative match. | |
1704 | If that character does not match, we will stride the proper | |
1705 | distance to propose a match that superimposes it on the last | |
1706 | instance of a character that matches it (per trt), or misses | |
1707 | it entirely if there is none. */ | |
facdc750 RS |
1708 | |
1709 | dirlen = len_byte * direction; | |
cb6792d2 RS |
1710 | |
1711 | /* Record position after the end of the pattern. */ | |
1712 | pat_end = base_pat + len_byte; | |
1713 | /* BASE_PAT points to a character that we start scanning from. | |
1714 | It is the first character in a forward search, | |
1715 | the last character in a backward search. */ | |
facdc750 | 1716 | if (direction < 0) |
cb6792d2 RS |
1717 | base_pat = pat_end - 1; |
1718 | ||
f4646fff AS |
1719 | /* A character that does not appear in the pattern induces a |
1720 | stride equal to the pattern length. */ | |
1721 | for (i = 0; i < 0400; i++) | |
1722 | BM_tab[i] = dirlen; | |
facdc750 RS |
1723 | |
1724 | /* We use this for translation, instead of TRT itself. | |
1725 | We fill this in to handle the characters that actually | |
1726 | occur in the pattern. Others don't matter anyway! */ | |
facdc750 RS |
1727 | for (i = 0; i < 0400; i++) |
1728 | simple_translate[i] = i; | |
1729 | ||
a17ae96a | 1730 | if (char_base) |
0190922f | 1731 | { |
a17ae96a | 1732 | /* Setup translate_prev_byte1/2/3/4 from CHAR_BASE. Only a |
0190922f | 1733 | byte following them are the target of translation. */ |
0190922f | 1734 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
1f3561e4 | 1735 | int cblen = CHAR_STRING (char_base, str); |
0190922f | 1736 | |
1f3561e4 PE |
1737 | translate_prev_byte1 = str[cblen - 2]; |
1738 | if (cblen > 2) | |
0190922f | 1739 | { |
1f3561e4 PE |
1740 | translate_prev_byte2 = str[cblen - 3]; |
1741 | if (cblen > 3) | |
1f1d9321 | 1742 | translate_prev_byte3 = str[cblen - 4]; |
0190922f KH |
1743 | } |
1744 | } | |
1745 | ||
facdc750 | 1746 | i = 0; |
f4646fff | 1747 | while (i != dirlen) |
facdc750 | 1748 | { |
cb6792d2 | 1749 | unsigned char *ptr = base_pat + i; |
facdc750 | 1750 | i += direction; |
facdc750 | 1751 | if (! NILP (trt)) |
ca1d1d23 | 1752 | { |
49e44e81 KH |
1753 | /* If the byte currently looking at is the last of a |
1754 | character to check case-equivalents, set CH to that | |
1755 | character. An ASCII character and a non-ASCII character | |
1756 | matching with CHAR_BASE are to be checked. */ | |
a17ae96a KH |
1757 | int ch = -1; |
1758 | ||
1759 | if (ASCII_BYTE_P (*ptr) || ! multibyte) | |
1760 | ch = *ptr; | |
49e44e81 | 1761 | else if (char_base |
86dc6ccb | 1762 | && ((pat_end - ptr) == 1 || CHAR_HEAD_P (ptr[1]))) |
ca1d1d23 | 1763 | { |
49e44e81 KH |
1764 | unsigned char *charstart = ptr - 1; |
1765 | ||
1766 | while (! (CHAR_HEAD_P (*charstart))) | |
1767 | charstart--; | |
62a6e103 | 1768 | ch = STRING_CHAR (charstart); |
a17ae96a KH |
1769 | if (char_base != (ch & ~0x3F)) |
1770 | ch = -1; | |
ca1d1d23 | 1771 | } |
facdc750 | 1772 | |
edb7b4dc | 1773 | if (ch >= 0200 && multibyte) |
49e44e81 KH |
1774 | j = (ch & 0x3F) | 0200; |
1775 | else | |
1776 | j = *ptr; | |
1777 | ||
f4646fff | 1778 | if (i == dirlen) |
facdc750 | 1779 | stride_for_teases = BM_tab[j]; |
ab228c24 | 1780 | |
facdc750 | 1781 | BM_tab[j] = dirlen - i; |
1db5b1ad JB |
1782 | /* A translation table is accompanied by its inverse -- see |
1783 | comment following downcase_table for details. */ | |
a17ae96a | 1784 | if (ch >= 0) |
ab228c24 RS |
1785 | { |
1786 | int starting_ch = ch; | |
49e44e81 | 1787 | int starting_j = j; |
a17ae96a | 1788 | |
ab228c24 RS |
1789 | while (1) |
1790 | { | |
1791 | TRANSLATE (ch, inverse_trt, ch); | |
edb7b4dc | 1792 | if (ch >= 0200 && multibyte) |
49e44e81 | 1793 | j = (ch & 0x3F) | 0200; |
ab228c24 | 1794 | else |
a17ae96a | 1795 | j = ch; |
ab228c24 RS |
1796 | |
1797 | /* For all the characters that map into CH, | |
1798 | set up simple_translate to map the last byte | |
1799 | into STARTING_J. */ | |
1800 | simple_translate[j] = starting_j; | |
1801 | if (ch == starting_ch) | |
1802 | break; | |
1803 | BM_tab[j] = dirlen - i; | |
1804 | } | |
1805 | } | |
facdc750 RS |
1806 | } |
1807 | else | |
1808 | { | |
1809 | j = *ptr; | |
1810 | ||
f4646fff | 1811 | if (i == dirlen) |
facdc750 RS |
1812 | stride_for_teases = BM_tab[j]; |
1813 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1814 | } |
f4646fff AS |
1815 | /* stride_for_teases tells how much to stride if we get a |
1816 | match on the far character but are subsequently | |
1817 | disappointed, by recording what the stride would have been | |
1818 | for that character if the last character had been | |
1819 | different. */ | |
facdc750 | 1820 | } |
facdc750 RS |
1821 | pos_byte += dirlen - ((direction > 0) ? direction : 0); |
1822 | /* loop invariant - POS_BYTE points at where last char (first | |
1823 | char if reverse) of pattern would align in a possible match. */ | |
1824 | while (n != 0) | |
1825 | { | |
d311d28c | 1826 | ptrdiff_t tail_end; |
facdc750 RS |
1827 | unsigned char *tail_end_ptr; |
1828 | ||
1829 | /* It's been reported that some (broken) compiler thinks that | |
1830 | Boolean expressions in an arithmetic context are unsigned. | |
1831 | Using an explicit ?1:0 prevents this. */ | |
1832 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1833 | < 0) | |
1834 | return (n * (0 - direction)); | |
1835 | /* First we do the part we can by pointers (maybe nothing) */ | |
1836 | QUIT; | |
1837 | pat = base_pat; | |
1838 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1839 | if (direction > 0) |
1840 | { | |
1841 | limit = BUFFER_CEILING_OF (limit); | |
1842 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1843 | can take on without hitting edge of buffer or the gap. */ | |
1844 | limit = min (limit, pos_byte + 20000); | |
1845 | limit = min (limit, lim_byte - 1); | |
1846 | } | |
1847 | else | |
1848 | { | |
1849 | limit = BUFFER_FLOOR_OF (limit); | |
1850 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1851 | can take on without hitting edge of buffer or the gap. */ | |
1852 | limit = max (limit, pos_byte - 20000); | |
1853 | limit = max (limit, lim_byte); | |
1854 | } | |
facdc750 RS |
1855 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1856 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1857 | ||
1858 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1859 | { |
facdc750 RS |
1860 | unsigned char *p2; |
1861 | ||
1862 | p_limit = BYTE_POS_ADDR (limit); | |
1863 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
f4646fff | 1864 | /* In this loop, pos + cursor - p2 is the surrogate for pos. */ |
facdc750 | 1865 | while (1) /* use one cursor setting as long as i can */ |
ca1d1d23 | 1866 | { |
facdc750 | 1867 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 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 RS |
1875 | } |
1876 | else | |
1877 | { | |
f4646fff AS |
1878 | while (cursor >= p_limit) |
1879 | { | |
1880 | if (BM_tab[*cursor] == 0) | |
1881 | goto hit; | |
facdc750 | 1882 | cursor += BM_tab[*cursor]; |
f4646fff | 1883 | } |
facdc750 | 1884 | } |
f4646fff AS |
1885 | /* If you are here, cursor is beyond the end of the |
1886 | searched region. You fail to match within the | |
1887 | permitted region and would otherwise try a character | |
1888 | beyond that region. */ | |
1889 | break; | |
1890 | ||
1891 | hit: | |
facdc750 RS |
1892 | i = dirlen - direction; |
1893 | if (! NILP (trt)) | |
1894 | { | |
1895 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1896 | { |
facdc750 RS |
1897 | int ch; |
1898 | cursor -= direction; | |
1899 | /* Translate only the last byte of a character. */ | |
1900 | if (! multibyte | |
1901 | || ((cursor == tail_end_ptr | |
1902 | || CHAR_HEAD_P (cursor[1])) | |
1903 | && (CHAR_HEAD_P (cursor[0]) | |
0190922f | 1904 | /* Check if this is the last byte of |
cd1181db | 1905 | a translatable character. */ |
0190922f KH |
1906 | || (translate_prev_byte1 == cursor[-1] |
1907 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1908 | || (translate_prev_byte2 == cursor[-2] | |
1909 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1910 | || (translate_prev_byte3 == cursor[-3])))))))) | |
facdc750 RS |
1911 | ch = simple_translate[*cursor]; |
1912 | else | |
1913 | ch = *cursor; | |
1914 | if (pat[i] != ch) | |
1915 | break; | |
ca1d1d23 | 1916 | } |
facdc750 RS |
1917 | } |
1918 | else | |
1919 | { | |
1920 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1921 | { |
facdc750 RS |
1922 | cursor -= direction; |
1923 | if (pat[i] != *cursor) | |
1924 | break; | |
ca1d1d23 | 1925 | } |
facdc750 RS |
1926 | } |
1927 | cursor += dirlen - i - direction; /* fix cursor */ | |
1928 | if (i + direction == 0) | |
1929 | { | |
d311d28c | 1930 | ptrdiff_t position, start, end; |
0c8533c6 | 1931 | |
facdc750 | 1932 | cursor -= direction; |
1113d9db | 1933 | |
facdc750 RS |
1934 | position = pos_byte + cursor - p2 + ((direction > 0) |
1935 | ? 1 - len_byte : 0); | |
1936 | set_search_regs (position, len_byte); | |
ca325161 | 1937 | |
ef887810 RS |
1938 | if (NILP (Vinhibit_changing_match_data)) |
1939 | { | |
1940 | start = search_regs.start[0]; | |
1941 | end = search_regs.end[0]; | |
1942 | } | |
1943 | else | |
1944 | /* If Vinhibit_changing_match_data is non-nil, | |
1945 | search_regs will not be changed. So let's | |
1946 | compute start and end here. */ | |
1947 | { | |
1948 | start = BYTE_TO_CHAR (position); | |
1949 | end = BYTE_TO_CHAR (position + len_byte); | |
1950 | } | |
1951 | ||
facdc750 RS |
1952 | if ((n -= direction) != 0) |
1953 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1954 | else |
ef887810 | 1955 | return direction > 0 ? end : start; |
ca1d1d23 | 1956 | } |
facdc750 RS |
1957 | else |
1958 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1959 | } |
facdc750 RS |
1960 | pos_byte += cursor - p2; |
1961 | } | |
1962 | else | |
f4646fff AS |
1963 | /* Now we'll pick up a clump that has to be done the hard |
1964 | way because it covers a discontinuity. */ | |
facdc750 RS |
1965 | { |
1966 | limit = ((direction > 0) | |
1967 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1968 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1969 | limit = ((direction > 0) | |
1970 | ? min (limit + len_byte, lim_byte - 1) | |
1971 | : max (limit - len_byte, lim_byte)); | |
1972 | /* LIMIT is now the last value POS_BYTE can have | |
1973 | and still be valid for a possible match. */ | |
1974 | while (1) | |
ca1d1d23 | 1975 | { |
f4646fff AS |
1976 | /* This loop can be coded for space rather than |
1977 | speed because it will usually run only once. | |
1978 | (the reach is at most len + 21, and typically | |
1979 | does not exceed len). */ | |
facdc750 | 1980 | while ((limit - pos_byte) * direction >= 0) |
f4646fff AS |
1981 | { |
1982 | int ch = FETCH_BYTE (pos_byte); | |
1983 | if (BM_tab[ch] == 0) | |
1984 | goto hit2; | |
1985 | pos_byte += BM_tab[ch]; | |
1986 | } | |
1987 | break; /* ran off the end */ | |
1988 | ||
1989 | hit2: | |
1990 | /* Found what might be a match. */ | |
facdc750 RS |
1991 | i = dirlen - direction; |
1992 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1993 | { |
facdc750 RS |
1994 | int ch; |
1995 | unsigned char *ptr; | |
1996 | pos_byte -= direction; | |
1997 | ptr = BYTE_POS_ADDR (pos_byte); | |
1998 | /* Translate only the last byte of a character. */ | |
1999 | if (! multibyte | |
2000 | || ((ptr == tail_end_ptr | |
2001 | || CHAR_HEAD_P (ptr[1])) | |
2002 | && (CHAR_HEAD_P (ptr[0]) | |
0190922f | 2003 | /* Check if this is the last byte of a |
cd1181db | 2004 | translatable character. */ |
0190922f KH |
2005 | || (translate_prev_byte1 == ptr[-1] |
2006 | && (CHAR_HEAD_P (translate_prev_byte1) | |
2007 | || (translate_prev_byte2 == ptr[-2] | |
2008 | && (CHAR_HEAD_P (translate_prev_byte2) | |
2009 | || translate_prev_byte3 == ptr[-3]))))))) | |
facdc750 RS |
2010 | ch = simple_translate[*ptr]; |
2011 | else | |
2012 | ch = *ptr; | |
2013 | if (pat[i] != ch) | |
2014 | break; | |
2015 | } | |
2016 | /* Above loop has moved POS_BYTE part or all the way | |
2017 | back to the first pos (last pos if reverse). | |
2018 | Set it once again at the last (first if reverse) char. */ | |
f4646fff | 2019 | pos_byte += dirlen - i - direction; |
facdc750 RS |
2020 | if (i + direction == 0) |
2021 | { | |
d311d28c | 2022 | ptrdiff_t position, start, end; |
facdc750 | 2023 | pos_byte -= direction; |
1113d9db | 2024 | |
facdc750 | 2025 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
facdc750 | 2026 | set_search_regs (position, len_byte); |
ca325161 | 2027 | |
ef887810 RS |
2028 | if (NILP (Vinhibit_changing_match_data)) |
2029 | { | |
2030 | start = search_regs.start[0]; | |
2031 | end = search_regs.end[0]; | |
2032 | } | |
2033 | else | |
2034 | /* If Vinhibit_changing_match_data is non-nil, | |
2035 | search_regs will not be changed. So let's | |
2036 | compute start and end here. */ | |
2037 | { | |
2038 | start = BYTE_TO_CHAR (position); | |
2039 | end = BYTE_TO_CHAR (position + len_byte); | |
2040 | } | |
2041 | ||
facdc750 RS |
2042 | if ((n -= direction) != 0) |
2043 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 2044 | else |
ef887810 | 2045 | return direction > 0 ? end : start; |
ca1d1d23 | 2046 | } |
facdc750 RS |
2047 | else |
2048 | pos_byte += stride_for_teases; | |
2049 | } | |
2050 | } | |
2051 | /* We have done one clump. Can we continue? */ | |
2052 | if ((lim_byte - pos_byte) * direction < 0) | |
2053 | return ((0 - n) * direction); | |
ca1d1d23 | 2054 | } |
facdc750 | 2055 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 2056 | } |
ca325161 | 2057 | |
fa8ed3e0 | 2058 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
2059 | for the overall match just found in the current buffer. |
2060 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
2061 | |
2062 | static void | |
d311d28c | 2063 | set_search_regs (ptrdiff_t beg_byte, ptrdiff_t nbytes) |
ca325161 | 2064 | { |
d311d28c | 2065 | ptrdiff_t i; |
a7e4cdde | 2066 | |
ef887810 RS |
2067 | if (!NILP (Vinhibit_changing_match_data)) |
2068 | return; | |
2069 | ||
ca325161 RS |
2070 | /* Make sure we have registers in which to store |
2071 | the match position. */ | |
2072 | if (search_regs.num_regs == 0) | |
2073 | { | |
23f86fce DA |
2074 | search_regs.start = xmalloc (2 * sizeof (regoff_t)); |
2075 | search_regs.end = xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 2076 | search_regs.num_regs = 2; |
ca325161 RS |
2077 | } |
2078 | ||
a7e4cdde RS |
2079 | /* Clear out the other registers. */ |
2080 | for (i = 1; i < search_regs.num_regs; i++) | |
2081 | { | |
2082 | search_regs.start[i] = -1; | |
2083 | search_regs.end[i] = -1; | |
2084 | } | |
2085 | ||
fa8ed3e0 RS |
2086 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
2087 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 2088 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 2089 | } |
ca1d1d23 | 2090 | \f |
ca1d1d23 | 2091 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, |
8c1a1077 PJ |
2092 | "MSearch backward: ", |
2093 | doc: /* Search backward from point for STRING. | |
2094 | Set point to the beginning of the occurrence found, and return point. | |
2095 | An optional second argument bounds the search; it is a buffer position. | |
2096 | The match found must not extend before that position. | |
2097 | Optional third argument, if t, means if fail just return nil (no error). | |
2098 | If not nil and not t, position at limit of search and return nil. | |
acc28cb9 CY |
2099 | Optional fourth argument COUNT, if non-nil, means to search for COUNT |
2100 | successive occurrences. If COUNT is negative, search forward, | |
2101 | instead of backward, for -COUNT occurrences. | |
8c1a1077 PJ |
2102 | |
2103 | Search case-sensitivity is determined by the value of the variable | |
2104 | `case-fold-search', which see. | |
2105 | ||
2106 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2107 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2108 | { |
b819a390 | 2109 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2110 | } |
2111 | ||
6af43974 | 2112 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2113 | doc: /* Search forward from point for STRING. |
2114 | Set point to the end of the occurrence found, and return point. | |
2115 | An optional second argument bounds the search; it is a buffer position. | |
d6aacdcd | 2116 | The match found must not extend after that position. A value of nil is |
48740f01 | 2117 | equivalent to (point-max). |
8c1a1077 PJ |
2118 | Optional third argument, if t, means if fail just return nil (no error). |
2119 | If not nil and not t, move to limit of search and return nil. | |
acc28cb9 CY |
2120 | Optional fourth argument COUNT, if non-nil, means to search for COUNT |
2121 | successive occurrences. If COUNT is negative, search backward, | |
2122 | instead of forward, for -COUNT occurrences. | |
8c1a1077 PJ |
2123 | |
2124 | Search case-sensitivity is determined by the value of the variable | |
2125 | `case-fold-search', which see. | |
2126 | ||
2127 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2128 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2129 | { |
b819a390 | 2130 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2131 | } |
2132 | ||
ca1d1d23 | 2133 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, |
8c1a1077 PJ |
2134 | "sRE search backward: ", |
2135 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2136 | Set point to the beginning of the match, and return point. | |
2137 | The match found is the one starting last in the buffer | |
2138 | and yet ending before the origin of the search. | |
2139 | An optional second argument bounds the search; it is a buffer position. | |
2140 | The match found must start at or after that position. | |
2141 | Optional third argument, if t, means if fail just return nil (no error). | |
2142 | If not nil and not t, move to limit of search and return nil. | |
2143 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2144 | |
2145 | Search case-sensitivity is determined by the value of the variable | |
2146 | `case-fold-search', which see. | |
2147 | ||
8c1a1077 PJ |
2148 | See also the functions `match-beginning', `match-end', `match-string', |
2149 | and `replace-match'. */) | |
5842a27b | 2150 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2151 | { |
b819a390 | 2152 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2153 | } |
2154 | ||
2155 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2156 | "sRE search: ", |
2157 | doc: /* Search forward from point for regular expression REGEXP. | |
2158 | Set point to the end of the occurrence found, and return point. | |
2159 | An optional second argument bounds the search; it is a buffer position. | |
2160 | The match found must not extend after that position. | |
2161 | Optional third argument, if t, means if fail just return nil (no error). | |
2162 | If not nil and not t, move to limit of search and return nil. | |
2163 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2164 | |
2165 | Search case-sensitivity is determined by the value of the variable | |
2166 | `case-fold-search', which see. | |
2167 | ||
8c1a1077 PJ |
2168 | See also the functions `match-beginning', `match-end', `match-string', |
2169 | and `replace-match'. */) | |
5842a27b | 2170 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2171 | { |
b819a390 RS |
2172 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2173 | } | |
2174 | ||
2175 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2176 | "sPosix search backward: ", |
2177 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2178 | Find the longest match in accord with Posix regular expression rules. | |
2179 | Set point to the beginning of the match, and return point. | |
2180 | The match found is the one starting last in the buffer | |
2181 | and yet ending before the origin of the search. | |
2182 | An optional second argument bounds the search; it is a buffer position. | |
2183 | The match found must start at or after that position. | |
2184 | Optional third argument, if t, means if fail just return nil (no error). | |
2185 | If not nil and not t, move to limit of search and return nil. | |
2186 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2187 | |
2188 | Search case-sensitivity is determined by the value of the variable | |
2189 | `case-fold-search', which see. | |
2190 | ||
8c1a1077 PJ |
2191 | See also the functions `match-beginning', `match-end', `match-string', |
2192 | and `replace-match'. */) | |
5842a27b | 2193 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2194 | { |
2195 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2196 | } | |
2197 | ||
2198 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2199 | "sPosix search: ", |
2200 | doc: /* Search forward from point for regular expression REGEXP. | |
2201 | Find the longest match in accord with Posix regular expression rules. | |
2202 | Set point to the end of the occurrence found, and return point. | |
2203 | An optional second argument bounds the search; it is a buffer position. | |
2204 | The match found must not extend after that position. | |
2205 | Optional third argument, if t, means if fail just return nil (no error). | |
2206 | If not nil and not t, move to limit of search and return nil. | |
2207 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2208 | |
2209 | Search case-sensitivity is determined by the value of the variable | |
2210 | `case-fold-search', which see. | |
2211 | ||
8c1a1077 PJ |
2212 | See also the functions `match-beginning', `match-end', `match-string', |
2213 | and `replace-match'. */) | |
5842a27b | 2214 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2215 | { |
2216 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2217 | } |
2218 | \f | |
d7a5ad5f | 2219 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2220 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2221 | Leave point at the end of the replacement text. |
2222 | ||
8c1a1077 PJ |
2223 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. |
2224 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2225 | based on the replaced text. | |
2226 | If the replaced text has only capital letters | |
2227 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
4dd0c271 RS |
2228 | Otherwise if all words are capitalized in the replaced text, |
2229 | capitalize each word in NEWTEXT. | |
8c1a1077 PJ |
2230 | |
2231 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2232 | Otherwise treat `\\' as special: | |
2233 | `\\&' in NEWTEXT means substitute original matched text. | |
2234 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2235 | If Nth parens didn't match, substitute nothing. | |
2236 | `\\\\' means insert one `\\'. | |
d32e47af LM |
2237 | `\\?' is treated literally |
2238 | (for compatibility with `query-replace-regexp'). | |
2239 | Any other character following `\\' signals an error. | |
4dd0c271 RS |
2240 | Case conversion does not apply to these substitutions. |
2241 | ||
8c1a1077 | 2242 | FIXEDCASE and LITERAL are optional arguments. |
8c1a1077 PJ |
2243 | |
2244 | The optional fourth argument STRING can be a string to modify. | |
2245 | This is meaningful when the previous match was done against STRING, | |
2246 | using `string-match'. When used this way, `replace-match' | |
2247 | creates and returns a new string made by copying STRING and replacing | |
2248 | the part of STRING that was matched. | |
2249 | ||
2250 | The optional fifth argument SUBEXP specifies a subexpression; | |
2251 | it says to replace just that subexpression with NEWTEXT, | |
2252 | rather than replacing the entire matched text. | |
2253 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2254 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2255 | NEWTEXT in place of subexp N. | |
2256 | This is useful only after a regular expression search or match, | |
2257 | since only regular expressions have distinguished subexpressions. */) | |
5842a27b | 2258 | (Lisp_Object newtext, Lisp_Object fixedcase, Lisp_Object literal, Lisp_Object string, Lisp_Object subexp) |
ca1d1d23 JB |
2259 | { |
2260 | enum { nochange, all_caps, cap_initial } case_action; | |
d311d28c | 2261 | register ptrdiff_t pos, pos_byte; |
ca1d1d23 | 2262 | int some_multiletter_word; |
97832bd0 | 2263 | int some_lowercase; |
73dc8771 | 2264 | int some_uppercase; |
208767c3 | 2265 | int some_nonuppercase_initial; |
ca1d1d23 | 2266 | register int c, prevc; |
a0efffc8 | 2267 | ptrdiff_t sub; |
d311d28c | 2268 | ptrdiff_t opoint, newpoint; |
ca1d1d23 | 2269 | |
b7826503 | 2270 | CHECK_STRING (newtext); |
ca1d1d23 | 2271 | |
080c45fd | 2272 | if (! NILP (string)) |
b7826503 | 2273 | CHECK_STRING (string); |
080c45fd | 2274 | |
ca1d1d23 JB |
2275 | case_action = nochange; /* We tried an initialization */ |
2276 | /* but some C compilers blew it */ | |
4746118a JB |
2277 | |
2278 | if (search_regs.num_regs <= 0) | |
d72cdbfc | 2279 | error ("`replace-match' called before any match found"); |
4746118a | 2280 | |
d7a5ad5f RS |
2281 | if (NILP (subexp)) |
2282 | sub = 0; | |
2283 | else | |
2284 | { | |
b7826503 | 2285 | CHECK_NUMBER (subexp); |
a0efffc8 | 2286 | if (! (0 <= XINT (subexp) && XINT (subexp) < search_regs.num_regs)) |
d7a5ad5f | 2287 | args_out_of_range (subexp, make_number (search_regs.num_regs)); |
a0efffc8 | 2288 | sub = XINT (subexp); |
d7a5ad5f RS |
2289 | } |
2290 | ||
080c45fd RS |
2291 | if (NILP (string)) |
2292 | { | |
d7a5ad5f RS |
2293 | if (search_regs.start[sub] < BEGV |
2294 | || search_regs.start[sub] > search_regs.end[sub] | |
2295 | || search_regs.end[sub] > ZV) | |
2296 | args_out_of_range (make_number (search_regs.start[sub]), | |
2297 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2298 | } |
2299 | else | |
2300 | { | |
d7a5ad5f RS |
2301 | if (search_regs.start[sub] < 0 |
2302 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2303 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2304 | args_out_of_range (make_number (search_regs.start[sub]), |
2305 | make_number (search_regs.end[sub])); | |
080c45fd | 2306 | } |
ca1d1d23 JB |
2307 | |
2308 | if (NILP (fixedcase)) | |
2309 | { | |
2310 | /* Decide how to casify by examining the matched text. */ | |
d311d28c | 2311 | ptrdiff_t last; |
ca1d1d23 | 2312 | |
ac3b28b1 KH |
2313 | pos = search_regs.start[sub]; |
2314 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2315 | |
2316 | if (NILP (string)) | |
ac3b28b1 | 2317 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2318 | else |
ac3b28b1 | 2319 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2320 | |
ca1d1d23 JB |
2321 | prevc = '\n'; |
2322 | case_action = all_caps; | |
2323 | ||
2324 | /* some_multiletter_word is set nonzero if any original word | |
2325 | is more than one letter long. */ | |
2326 | some_multiletter_word = 0; | |
97832bd0 | 2327 | some_lowercase = 0; |
208767c3 | 2328 | some_nonuppercase_initial = 0; |
73dc8771 | 2329 | some_uppercase = 0; |
ca1d1d23 | 2330 | |
ac3b28b1 | 2331 | while (pos < last) |
ca1d1d23 | 2332 | { |
080c45fd | 2333 | if (NILP (string)) |
ac3b28b1 | 2334 | { |
93daa011 | 2335 | c = FETCH_CHAR_AS_MULTIBYTE (pos_byte); |
ac3b28b1 KH |
2336 | INC_BOTH (pos, pos_byte); |
2337 | } | |
080c45fd | 2338 | else |
93daa011 | 2339 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2340 | |
5da9919f | 2341 | if (lowercasep (c)) |
ca1d1d23 JB |
2342 | { |
2343 | /* Cannot be all caps if any original char is lower case */ | |
2344 | ||
97832bd0 | 2345 | some_lowercase = 1; |
ca1d1d23 | 2346 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2347 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2348 | else |
2349 | some_multiletter_word = 1; | |
2350 | } | |
5da9919f | 2351 | else if (uppercasep (c)) |
ca1d1d23 | 2352 | { |
73dc8771 | 2353 | some_uppercase = 1; |
97832bd0 | 2354 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2355 | ; |
97832bd0 | 2356 | else |
ca1d1d23 JB |
2357 | some_multiletter_word = 1; |
2358 | } | |
208767c3 RS |
2359 | else |
2360 | { | |
2361 | /* If the initial is a caseless word constituent, | |
2362 | treat that like a lowercase initial. */ | |
2363 | if (SYNTAX (prevc) != Sword) | |
2364 | some_nonuppercase_initial = 1; | |
2365 | } | |
ca1d1d23 JB |
2366 | |
2367 | prevc = c; | |
2368 | } | |
2369 | ||
97832bd0 RS |
2370 | /* Convert to all caps if the old text is all caps |
2371 | and has at least one multiletter word. */ | |
2372 | if (! some_lowercase && some_multiletter_word) | |
2373 | case_action = all_caps; | |
c4d460ce | 2374 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2375 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2376 | case_action = cap_initial; |
208767c3 | 2377 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2378 | /* Should x -> yz, operating on X, give Yz or YZ? |
2379 | We'll assume the latter. */ | |
2380 | case_action = all_caps; | |
97832bd0 RS |
2381 | else |
2382 | case_action = nochange; | |
ca1d1d23 JB |
2383 | } |
2384 | ||
080c45fd RS |
2385 | /* Do replacement in a string. */ |
2386 | if (!NILP (string)) | |
2387 | { | |
2388 | Lisp_Object before, after; | |
2389 | ||
2390 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2391 | make_number (search_regs.start[sub])); |
2392 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2393 | |
636a5e28 RS |
2394 | /* Substitute parts of the match into NEWTEXT |
2395 | if desired. */ | |
080c45fd RS |
2396 | if (NILP (literal)) |
2397 | { | |
d311d28c PE |
2398 | ptrdiff_t lastpos = 0; |
2399 | ptrdiff_t lastpos_byte = 0; | |
080c45fd RS |
2400 | /* We build up the substituted string in ACCUM. */ |
2401 | Lisp_Object accum; | |
2402 | Lisp_Object middle; | |
d311d28c | 2403 | ptrdiff_t length = SBYTES (newtext); |
080c45fd RS |
2404 | |
2405 | accum = Qnil; | |
2406 | ||
ac3b28b1 | 2407 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd | 2408 | { |
d311d28c PE |
2409 | ptrdiff_t substart = -1; |
2410 | ptrdiff_t subend = 0; | |
1e79ec24 | 2411 | int delbackslash = 0; |
080c45fd | 2412 | |
0c8533c6 RS |
2413 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2414 | ||
080c45fd RS |
2415 | if (c == '\\') |
2416 | { | |
0c8533c6 | 2417 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2418 | |
080c45fd RS |
2419 | if (c == '&') |
2420 | { | |
d7a5ad5f RS |
2421 | substart = search_regs.start[sub]; |
2422 | subend = search_regs.end[sub]; | |
080c45fd | 2423 | } |
76fc1ea2 | 2424 | else if (c >= '1' && c <= '9') |
080c45fd | 2425 | { |
d311d28c PE |
2426 | if (c - '0' < search_regs.num_regs |
2427 | && 0 <= search_regs.start[c - '0']) | |
080c45fd RS |
2428 | { |
2429 | substart = search_regs.start[c - '0']; | |
2430 | subend = search_regs.end[c - '0']; | |
2431 | } | |
76fc1ea2 KH |
2432 | else |
2433 | { | |
2434 | /* If that subexp did not match, | |
2435 | replace \\N with nothing. */ | |
2436 | substart = 0; | |
2437 | subend = 0; | |
2438 | } | |
080c45fd | 2439 | } |
1e79ec24 KH |
2440 | else if (c == '\\') |
2441 | delbackslash = 1; | |
d32e47af | 2442 | else if (c != '?') |
636a5e28 | 2443 | error ("Invalid use of `\\' in replacement text"); |
080c45fd RS |
2444 | } |
2445 | if (substart >= 0) | |
2446 | { | |
d131e79c RS |
2447 | if (pos - 2 != lastpos) |
2448 | middle = substring_both (newtext, lastpos, | |
2449 | lastpos_byte, | |
2450 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2451 | else |
2452 | middle = Qnil; | |
2453 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2454 | Fsubstring (string, |
2455 | make_number (substart), | |
080c45fd RS |
2456 | make_number (subend))); |
2457 | lastpos = pos; | |
0c8533c6 | 2458 | lastpos_byte = pos_byte; |
080c45fd | 2459 | } |
1e79ec24 KH |
2460 | else if (delbackslash) |
2461 | { | |
d131e79c RS |
2462 | middle = substring_both (newtext, lastpos, |
2463 | lastpos_byte, | |
2464 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2465 | |
1e79ec24 KH |
2466 | accum = concat2 (accum, middle); |
2467 | lastpos = pos; | |
0c8533c6 | 2468 | lastpos_byte = pos_byte; |
1e79ec24 | 2469 | } |
080c45fd RS |
2470 | } |
2471 | ||
d131e79c RS |
2472 | if (pos != lastpos) |
2473 | middle = substring_both (newtext, lastpos, | |
2474 | lastpos_byte, | |
0c8533c6 | 2475 | pos, pos_byte); |
080c45fd RS |
2476 | else |
2477 | middle = Qnil; | |
2478 | ||
2479 | newtext = concat2 (accum, middle); | |
2480 | } | |
2481 | ||
636a5e28 | 2482 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2483 | if (case_action == all_caps) |
2484 | newtext = Fupcase (newtext); | |
2485 | else if (case_action == cap_initial) | |
2b2eead9 | 2486 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2487 | |
2488 | return concat3 (before, newtext, after); | |
2489 | } | |
2490 | ||
09c4719e | 2491 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2492 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2493 | opoint = PT - ZV; |
9160906f RS |
2494 | else if (PT > search_regs.start[sub]) |
2495 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2496 | else |
2497 | opoint = PT; | |
2498 | ||
886ed6ec RS |
2499 | /* If we want non-literal replacement, |
2500 | perform substitution on the replacement string. */ | |
2501 | if (NILP (literal)) | |
ca1d1d23 | 2502 | { |
0065d054 | 2503 | ptrdiff_t length = SBYTES (newtext); |
68e69fbd | 2504 | unsigned char *substed; |
0065d054 | 2505 | ptrdiff_t substed_alloc_size, substed_len; |
4b4deea2 | 2506 | int buf_multibyte = !NILP (BVAR (current_buffer, enable_multibyte_characters)); |
3bc25e52 | 2507 | int str_multibyte = STRING_MULTIBYTE (newtext); |
886ed6ec | 2508 | int really_changed = 0; |
3bc25e52 | 2509 | |
0065d054 PE |
2510 | substed_alloc_size = ((STRING_BYTES_BOUND - 100) / 2 < length |
2511 | ? STRING_BYTES_BOUND | |
2512 | : length * 2 + 100); | |
23f86fce | 2513 | substed = xmalloc (substed_alloc_size); |
68e69fbd RS |
2514 | substed_len = 0; |
2515 | ||
3bc25e52 KH |
2516 | /* Go thru NEWTEXT, producing the actual text to insert in |
2517 | SUBSTED while adjusting multibyteness to that of the current | |
2518 | buffer. */ | |
ca1d1d23 | 2519 | |
ac3b28b1 | 2520 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2521 | { |
68e69fbd | 2522 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
8ea90aa3 | 2523 | const unsigned char *add_stuff = NULL; |
0065d054 | 2524 | ptrdiff_t add_len = 0; |
a0efffc8 | 2525 | ptrdiff_t idx = -1; |
9a76659d | 2526 | |
3bc25e52 KH |
2527 | if (str_multibyte) |
2528 | { | |
eb99a8dd | 2529 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 | 2530 | if (!buf_multibyte) |
461c2ab9 | 2531 | c = multibyte_char_to_unibyte (c); |
3bc25e52 KH |
2532 | } |
2533 | else | |
2534 | { | |
2535 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2536 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2537 | if (buf_multibyte) |
4c0354d7 | 2538 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 | 2539 | } |
ac3b28b1 | 2540 | |
68e69fbd RS |
2541 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2542 | or set IDX to a match index, which means put that part | |
2543 | of the buffer text into SUBSTED. */ | |
2544 | ||
ca1d1d23 JB |
2545 | if (c == '\\') |
2546 | { | |
886ed6ec RS |
2547 | really_changed = 1; |
2548 | ||
3bc25e52 KH |
2549 | if (str_multibyte) |
2550 | { | |
eb99a8dd KH |
2551 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2552 | pos, pos_byte); | |
071ce769 | 2553 | if (!buf_multibyte && !ASCII_CHAR_P (c)) |
461c2ab9 | 2554 | c = multibyte_char_to_unibyte (c); |
3bc25e52 KH |
2555 | } |
2556 | else | |
2557 | { | |
d5db4077 | 2558 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2559 | if (buf_multibyte) |
4c0354d7 | 2560 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 KH |
2561 | } |
2562 | ||
ca1d1d23 | 2563 | if (c == '&') |
68e69fbd | 2564 | idx = sub; |
d311d28c | 2565 | else if (c >= '1' && c <= '9' && c - '0' < search_regs.num_regs) |
ca1d1d23 JB |
2566 | { |
2567 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2568 | idx = c - '0'; |
ca1d1d23 | 2569 | } |
636a5e28 | 2570 | else if (c == '\\') |
a7e979a4 | 2571 | add_len = 1, add_stuff = (unsigned char *) "\\"; |
636a5e28 | 2572 | else |
3bc25e52 KH |
2573 | { |
2574 | xfree (substed); | |
2575 | error ("Invalid use of `\\' in replacement text"); | |
2576 | } | |
ca1d1d23 JB |
2577 | } |
2578 | else | |
68e69fbd RS |
2579 | { |
2580 | add_len = CHAR_STRING (c, str); | |
2581 | add_stuff = str; | |
2582 | } | |
2583 | ||
2584 | /* If we want to copy part of a previous match, | |
2585 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2586 | if (idx >= 0) | |
2587 | { | |
0065d054 | 2588 | ptrdiff_t begbyte = CHAR_TO_BYTE (search_regs.start[idx]); |
68e69fbd RS |
2589 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; |
2590 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2591 | move_gap (search_regs.start[idx]); | |
2592 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2593 | } | |
2594 | ||
2595 | /* Now the stuff we want to add to SUBSTED | |
2596 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2597 | ||
2598 | /* Make sure SUBSTED is big enough. */ | |
0065d054 PE |
2599 | if (substed_alloc_size - substed_len < add_len) |
2600 | substed = | |
2601 | xpalloc (substed, &substed_alloc_size, | |
2602 | add_len - (substed_alloc_size - substed_len), | |
2603 | STRING_BYTES_BOUND, 1); | |
68e69fbd RS |
2604 | |
2605 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2606 | if (add_stuff) |
2607 | { | |
72af86bd | 2608 | memcpy (substed + substed_len, add_stuff, add_len); |
f8ce8a0d GM |
2609 | substed_len += add_len; |
2610 | } | |
ca1d1d23 | 2611 | } |
68e69fbd | 2612 | |
886ed6ec | 2613 | if (really_changed) |
76fc1ea2 KH |
2614 | { |
2615 | if (buf_multibyte) | |
2616 | { | |
d311d28c | 2617 | ptrdiff_t nchars = |
c098fdb8 | 2618 | multibyte_chars_in_text (substed, substed_len); |
68e69fbd | 2619 | |
a7e979a4 PE |
2620 | newtext = make_multibyte_string ((char *) substed, nchars, |
2621 | substed_len); | |
76fc1ea2 KH |
2622 | } |
2623 | else | |
a7e979a4 | 2624 | newtext = make_unibyte_string ((char *) substed, substed_len); |
76fc1ea2 | 2625 | } |
68e69fbd | 2626 | xfree (substed); |
ca1d1d23 JB |
2627 | } |
2628 | ||
886ed6ec RS |
2629 | /* Replace the old text with the new in the cleanest possible way. */ |
2630 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2631 | newtext, 1, 0, 1); | |
d5db4077 | 2632 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2633 | |
2634 | if (case_action == all_caps) | |
886ed6ec RS |
2635 | Fupcase_region (make_number (search_regs.start[sub]), |
2636 | make_number (newpoint)); | |
ca1d1d23 | 2637 | else if (case_action == cap_initial) |
886ed6ec RS |
2638 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2639 | make_number (newpoint)); | |
3e18eecf | 2640 | |
98e942e0 RS |
2641 | /* Adjust search data for this change. */ |
2642 | { | |
d311d28c PE |
2643 | ptrdiff_t oldend = search_regs.end[sub]; |
2644 | ptrdiff_t oldstart = search_regs.start[sub]; | |
2645 | ptrdiff_t change = newpoint - search_regs.end[sub]; | |
2646 | ptrdiff_t i; | |
98e942e0 RS |
2647 | |
2648 | for (i = 0; i < search_regs.num_regs; i++) | |
2649 | { | |
41c01205 | 2650 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2651 | search_regs.start[i] += change; |
41c01205 DK |
2652 | else if (search_regs.start[i] > oldstart) |
2653 | search_regs.start[i] = oldstart; | |
2654 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2655 | search_regs.end[i] += change; |
41c01205 DK |
2656 | else if (search_regs.end[i] > oldstart) |
2657 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2658 | } |
2659 | } | |
2660 | ||
b0eba991 | 2661 | /* Put point back where it was in the text. */ |
8d808a65 | 2662 | if (opoint <= 0) |
fa8ed3e0 | 2663 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2664 | else |
fa8ed3e0 | 2665 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2666 | |
2667 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2668 | move_if_not_intangible (newpoint); |
177c0ea7 | 2669 | |
ca1d1d23 JB |
2670 | return Qnil; |
2671 | } | |
2672 | \f | |
2673 | static Lisp_Object | |
971de7fb | 2674 | match_limit (Lisp_Object num, int beginningp) |
ca1d1d23 | 2675 | { |
a0efffc8 | 2676 | EMACS_INT n; |
ca1d1d23 | 2677 | |
b7826503 | 2678 | CHECK_NUMBER (num); |
ca1d1d23 | 2679 | n = XINT (num); |
f90a5bf5 | 2680 | if (n < 0) |
bd2cbd56 | 2681 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2682 | if (search_regs.num_regs <= 0) |
2683 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2684 | if (n >= search_regs.num_regs |
4746118a | 2685 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2686 | return Qnil; |
2687 | return (make_number ((beginningp) ? search_regs.start[n] | |
2688 | : search_regs.end[n])); | |
2689 | } | |
2690 | ||
a7ca3326 | 2691 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, |
8c1a1077 PJ |
2692 | doc: /* Return position of start 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, 1); |
ca1d1d23 JB |
2701 | } |
2702 | ||
a7ca3326 | 2703 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, |
8c1a1077 PJ |
2704 | doc: /* Return position of end of text matched by last search. |
2705 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2706 | regexp. | |
2707 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2708 | SUBEXP pairs. | |
2709 | Zero means the entire text matched by the whole regexp or whole string. */) | |
5842a27b | 2710 | (Lisp_Object subexp) |
ca1d1d23 | 2711 | { |
5806161b | 2712 | return match_limit (subexp, 0); |
177c0ea7 | 2713 | } |
ca1d1d23 | 2714 | |
a7ca3326 | 2715 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 3, 0, |
8c1a1077 PJ |
2716 | doc: /* Return a list containing all info on what the last search matched. |
2717 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2718 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2719 | if the last match was on a buffer; integers or nil if a string was matched. | |
febb8aba | 2720 | Use `set-match-data' to reinstate the data in this list. |
8c1a1077 | 2721 | |
41c01205 DK |
2722 | If INTEGERS (the optional first argument) is non-nil, always use |
2723 | integers \(rather than markers) to represent buffer positions. In | |
2724 | this case, and if the last match was in a buffer, the buffer will get | |
2725 | stored as one additional element at the end of the list. | |
2726 | ||
abd0071c KS |
2727 | If REUSE is a list, reuse it as part of the value. If REUSE is long |
2728 | enough to hold all the values, and if INTEGERS is non-nil, no consing | |
2729 | is done. | |
2730 | ||
2731 | If optional third arg RESEAT is non-nil, any previous markers on the | |
2732 | REUSE list will be modified to point to nowhere. | |
2733 | ||
140a6b7e | 2734 | Return value is undefined if the last search failed. */) |
5842a27b | 2735 | (Lisp_Object integers, Lisp_Object reuse, Lisp_Object reseat) |
ca1d1d23 | 2736 | { |
56256c2a | 2737 | Lisp_Object tail, prev; |
4746118a | 2738 | Lisp_Object *data; |
d311d28c | 2739 | ptrdiff_t i, len; |
ca1d1d23 | 2740 | |
abd0071c KS |
2741 | if (!NILP (reseat)) |
2742 | for (tail = reuse; CONSP (tail); tail = XCDR (tail)) | |
2743 | if (MARKERP (XCAR (tail))) | |
2744 | { | |
51f10faa | 2745 | unchain_marker (XMARKER (XCAR (tail))); |
abd0071c KS |
2746 | XSETCAR (tail, Qnil); |
2747 | } | |
2748 | ||
daa37602 | 2749 | if (NILP (last_thing_searched)) |
c36bcf1b | 2750 | return Qnil; |
daa37602 | 2751 | |
6bbd7a29 GM |
2752 | prev = Qnil; |
2753 | ||
38182d90 | 2754 | data = alloca ((2 * search_regs.num_regs + 1) * sizeof *data); |
4746118a | 2755 | |
41c01205 | 2756 | len = 0; |
4746118a | 2757 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 | 2758 | { |
d311d28c | 2759 | ptrdiff_t start = search_regs.start[i]; |
ca1d1d23 JB |
2760 | if (start >= 0) |
2761 | { | |
56256c2a RS |
2762 | if (EQ (last_thing_searched, Qt) |
2763 | || ! NILP (integers)) | |
ca1d1d23 | 2764 | { |
c235cce7 KH |
2765 | XSETFASTINT (data[2 * i], start); |
2766 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2767 | } |
0ed62dc7 | 2768 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2769 | { |
2770 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2771 | Fset_marker (data[2 * i], |
2772 | make_number (start), | |
2773 | last_thing_searched); | |
ca1d1d23 JB |
2774 | data[2 * i + 1] = Fmake_marker (); |
2775 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2776 | make_number (search_regs.end[i]), |
daa37602 | 2777 | last_thing_searched); |
ca1d1d23 | 2778 | } |
daa37602 JB |
2779 | else |
2780 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
1088b922 | 2781 | emacs_abort (); |
daa37602 | 2782 | |
abd0071c | 2783 | len = 2 * i + 2; |
ca1d1d23 JB |
2784 | } |
2785 | else | |
abd0071c | 2786 | data[2 * i] = data[2 * i + 1] = Qnil; |
ca1d1d23 | 2787 | } |
56256c2a | 2788 | |
bd2cbd56 | 2789 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2790 | { |
bd2cbd56 | 2791 | data[len] = last_thing_searched; |
41c01205 DK |
2792 | len++; |
2793 | } | |
2794 | ||
56256c2a RS |
2795 | /* If REUSE is not usable, cons up the values and return them. */ |
2796 | if (! CONSP (reuse)) | |
41c01205 | 2797 | return Flist (len, data); |
56256c2a RS |
2798 | |
2799 | /* If REUSE is a list, store as many value elements as will fit | |
2800 | into the elements of REUSE. */ | |
2801 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2802 | i++, tail = XCDR (tail)) |
56256c2a | 2803 | { |
41c01205 | 2804 | if (i < len) |
f3fbd155 | 2805 | XSETCAR (tail, data[i]); |
56256c2a | 2806 | else |
f3fbd155 | 2807 | XSETCAR (tail, Qnil); |
56256c2a RS |
2808 | prev = tail; |
2809 | } | |
2810 | ||
2811 | /* If we couldn't fit all value elements into REUSE, | |
2812 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2813 | if (i < len) |
2814 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2815 | |
2816 | return reuse; | |
ca1d1d23 JB |
2817 | } |
2818 | ||
b51d6c92 SM |
2819 | /* We used to have an internal use variant of `reseat' described as: |
2820 | ||
2821 | If RESEAT is `evaporate', put the markers back on the free list | |
2822 | immediately. No other references to the markers must exist in this | |
2823 | case, so it is used only internally on the unwind stack and | |
2824 | save-match-data from Lisp. | |
2825 | ||
2826 | But it was ill-conceived: those supposedly-internal markers get exposed via | |
2827 | the undo-list, so freeing them here is unsafe. */ | |
ca1d1d23 | 2828 | |
a7ca3326 | 2829 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 2, 0, |
8c1a1077 | 2830 | doc: /* Set internal data on last search match from elements of LIST. |
abd0071c KS |
2831 | LIST should have been created by calling `match-data' previously. |
2832 | ||
51f10faa | 2833 | If optional arg RESEAT is non-nil, make markers on LIST point nowhere. */) |
5842a27b | 2834 | (register Lisp_Object list, Lisp_Object reseat) |
ca1d1d23 | 2835 | { |
5f2ab479 | 2836 | ptrdiff_t i; |
ca1d1d23 JB |
2837 | register Lisp_Object marker; |
2838 | ||
7074fde6 FP |
2839 | if (running_asynch_code) |
2840 | save_search_regs (); | |
2841 | ||
29100cea | 2842 | CHECK_LIST (list); |
ca1d1d23 | 2843 | |
41c01205 DK |
2844 | /* Unless we find a marker with a buffer or an explicit buffer |
2845 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
2846 | last_thing_searched = Qt; |
2847 | ||
4746118a JB |
2848 | /* Allocate registers if they don't already exist. */ |
2849 | { | |
d311d28c | 2850 | EMACS_INT length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2851 | |
2852 | if (length > search_regs.num_regs) | |
2853 | { | |
0065d054 | 2854 | ptrdiff_t num_regs = search_regs.num_regs; |
d311d28c PE |
2855 | if (PTRDIFF_MAX < length) |
2856 | memory_full (SIZE_MAX); | |
0065d054 PE |
2857 | search_regs.start = |
2858 | xpalloc (search_regs.start, &num_regs, length - num_regs, | |
2859 | min (PTRDIFF_MAX, UINT_MAX), sizeof (regoff_t)); | |
2860 | search_regs.end = | |
2861 | xrealloc (search_regs.end, num_regs * sizeof (regoff_t)); | |
2862 | ||
2863 | for (i = search_regs.num_regs; i < num_regs; i++) | |
e62371e9 KH |
2864 | search_regs.start[i] = -1; |
2865 | ||
0065d054 | 2866 | search_regs.num_regs = num_regs; |
4746118a | 2867 | } |
ca1d1d23 | 2868 | |
abd0071c | 2869 | for (i = 0; CONSP (list); i++) |
41c01205 | 2870 | { |
abd0071c | 2871 | marker = XCAR (list); |
bd2cbd56 | 2872 | if (BUFFERP (marker)) |
c3762cbd | 2873 | { |
bd2cbd56 | 2874 | last_thing_searched = marker; |
c3762cbd DK |
2875 | break; |
2876 | } | |
2877 | if (i >= length) | |
2878 | break; | |
41c01205 DK |
2879 | if (NILP (marker)) |
2880 | { | |
2881 | search_regs.start[i] = -1; | |
abd0071c | 2882 | list = XCDR (list); |
41c01205 DK |
2883 | } |
2884 | else | |
2885 | { | |
d311d28c | 2886 | Lisp_Object from; |
abd0071c | 2887 | Lisp_Object m; |
e2811828 | 2888 | |
abd0071c | 2889 | m = marker; |
41c01205 DK |
2890 | if (MARKERP (marker)) |
2891 | { | |
2892 | if (XMARKER (marker)->buffer == 0) | |
2893 | XSETFASTINT (marker, 0); | |
2894 | else | |
2895 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
2896 | } | |
e2811828 | 2897 | |
41c01205 | 2898 | CHECK_NUMBER_COERCE_MARKER (marker); |
d311d28c | 2899 | from = marker; |
e2811828 | 2900 | |
abd0071c KS |
2901 | if (!NILP (reseat) && MARKERP (m)) |
2902 | { | |
b51d6c92 | 2903 | unchain_marker (XMARKER (m)); |
9ad54a7e | 2904 | XSETCAR (list, Qnil); |
abd0071c KS |
2905 | } |
2906 | ||
2907 | if ((list = XCDR (list), !CONSP (list))) | |
2908 | break; | |
2909 | ||
2910 | m = marker = XCAR (list); | |
2911 | ||
41c01205 DK |
2912 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
2913 | XSETFASTINT (marker, 0); | |
e2811828 | 2914 | |
41c01205 | 2915 | CHECK_NUMBER_COERCE_MARKER (marker); |
0b1fccc4 PE |
2916 | if ((XINT (from) < 0 |
2917 | ? TYPE_MINIMUM (regoff_t) <= XINT (from) | |
2918 | : XINT (from) <= TYPE_MAXIMUM (regoff_t)) | |
2919 | && (XINT (marker) < 0 | |
2920 | ? TYPE_MINIMUM (regoff_t) <= XINT (marker) | |
2921 | : XINT (marker) <= TYPE_MAXIMUM (regoff_t))) | |
d311d28c PE |
2922 | { |
2923 | search_regs.start[i] = XINT (from); | |
2924 | search_regs.end[i] = XINT (marker); | |
2925 | } | |
2926 | else | |
2927 | { | |
2928 | search_regs.start[i] = -1; | |
2929 | } | |
abd0071c KS |
2930 | |
2931 | if (!NILP (reseat) && MARKERP (m)) | |
2932 | { | |
b51d6c92 | 2933 | unchain_marker (XMARKER (m)); |
9ad54a7e | 2934 | XSETCAR (list, Qnil); |
abd0071c | 2935 | } |
41c01205 | 2936 | } |
abd0071c | 2937 | list = XCDR (list); |
41c01205 | 2938 | } |
ca1d1d23 | 2939 | |
41c01205 DK |
2940 | for (; i < search_regs.num_regs; i++) |
2941 | search_regs.start[i] = -1; | |
2942 | } | |
ca1d1d23 | 2943 | |
177c0ea7 | 2944 | return Qnil; |
ca1d1d23 JB |
2945 | } |
2946 | ||
7074fde6 FP |
2947 | /* If non-zero the match data have been saved in saved_search_regs |
2948 | during the execution of a sentinel or filter. */ | |
75ebf74b | 2949 | static int search_regs_saved; |
7074fde6 | 2950 | static struct re_registers saved_search_regs; |
41c01205 | 2951 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
2952 | |
2953 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
2954 | if asynchronous code (filter or sentinel) is running. */ | |
2955 | static void | |
971de7fb | 2956 | save_search_regs (void) |
7074fde6 FP |
2957 | { |
2958 | if (!search_regs_saved) | |
2959 | { | |
2960 | saved_search_regs.num_regs = search_regs.num_regs; | |
2961 | saved_search_regs.start = search_regs.start; | |
2962 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
2963 | saved_last_thing_searched = last_thing_searched; |
2964 | last_thing_searched = Qnil; | |
7074fde6 | 2965 | search_regs.num_regs = 0; |
2d4a771a RS |
2966 | search_regs.start = 0; |
2967 | search_regs.end = 0; | |
7074fde6 FP |
2968 | |
2969 | search_regs_saved = 1; | |
2970 | } | |
2971 | } | |
2972 | ||
2973 | /* Called upon exit from filters and sentinels. */ | |
2974 | void | |
971de7fb | 2975 | restore_search_regs (void) |
7074fde6 FP |
2976 | { |
2977 | if (search_regs_saved) | |
2978 | { | |
2979 | if (search_regs.num_regs > 0) | |
2980 | { | |
2981 | xfree (search_regs.start); | |
2982 | xfree (search_regs.end); | |
2983 | } | |
2984 | search_regs.num_regs = saved_search_regs.num_regs; | |
2985 | search_regs.start = saved_search_regs.start; | |
2986 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
2987 | last_thing_searched = saved_last_thing_searched; |
2988 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
2989 | search_regs_saved = 0; |
2990 | } | |
2991 | } | |
2992 | ||
abd0071c | 2993 | static Lisp_Object |
971de7fb | 2994 | unwind_set_match_data (Lisp_Object list) |
abd0071c | 2995 | { |
b51d6c92 SM |
2996 | /* It is NOT ALWAYS safe to free (evaporate) the markers immediately. */ |
2997 | return Fset_match_data (list, Qt); | |
abd0071c KS |
2998 | } |
2999 | ||
3000 | /* Called to unwind protect the match data. */ | |
3001 | void | |
971de7fb | 3002 | record_unwind_save_match_data (void) |
abd0071c KS |
3003 | { |
3004 | record_unwind_protect (unwind_set_match_data, | |
3005 | Fmatch_data (Qnil, Qnil, Qnil)); | |
3006 | } | |
3007 | ||
e9a452d9 | 3008 | /* Quote a string to deactivate reg-expr chars */ |
ca1d1d23 JB |
3009 | |
3010 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 | 3011 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
5842a27b | 3012 | (Lisp_Object string) |
ca1d1d23 | 3013 | { |
a7e979a4 PE |
3014 | register char *in, *out, *end; |
3015 | register char *temp; | |
0c8533c6 | 3016 | int backslashes_added = 0; |
ca1d1d23 | 3017 | |
b7826503 | 3018 | CHECK_STRING (string); |
ca1d1d23 | 3019 | |
38182d90 | 3020 | temp = alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
3021 | |
3022 | /* Now copy the data into the new string, inserting escapes. */ | |
3023 | ||
a7e979a4 | 3024 | in = SSDATA (string); |
d5db4077 | 3025 | end = in + SBYTES (string); |
177c0ea7 | 3026 | out = temp; |
ca1d1d23 JB |
3027 | |
3028 | for (; in != end; in++) | |
3029 | { | |
66bc6082 | 3030 | if (*in == '[' |
ca1d1d23 JB |
3031 | || *in == '*' || *in == '.' || *in == '\\' |
3032 | || *in == '?' || *in == '+' | |
3033 | || *in == '^' || *in == '$') | |
0c8533c6 | 3034 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
3035 | *out++ = *in; |
3036 | } | |
3037 | ||
3f8100f1 | 3038 | return make_specified_string (temp, |
d5db4077 | 3039 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
3040 | out - temp, |
3041 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 3042 | } |
177c0ea7 | 3043 | \f |
dfcf069d | 3044 | void |
971de7fb | 3045 | syms_of_search (void) |
ca1d1d23 JB |
3046 | { |
3047 | register int i; | |
3048 | ||
487282dc KH |
3049 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
3050 | { | |
3051 | searchbufs[i].buf.allocated = 100; | |
23f86fce | 3052 | searchbufs[i].buf.buffer = xmalloc (100); |
487282dc KH |
3053 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
3054 | searchbufs[i].regexp = Qnil; | |
ecdb561e | 3055 | searchbufs[i].whitespace_regexp = Qnil; |
b69e3c18 | 3056 | searchbufs[i].syntax_table = Qnil; |
487282dc | 3057 | staticpro (&searchbufs[i].regexp); |
aa77b5ce | 3058 | staticpro (&searchbufs[i].whitespace_regexp); |
b69e3c18 | 3059 | staticpro (&searchbufs[i].syntax_table); |
487282dc KH |
3060 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); |
3061 | } | |
3062 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 | 3063 | |
cd3520a4 JB |
3064 | DEFSYM (Qsearch_failed, "search-failed"); |
3065 | DEFSYM (Qinvalid_regexp, "invalid-regexp"); | |
ca1d1d23 JB |
3066 | |
3067 | Fput (Qsearch_failed, Qerror_conditions, | |
3438fe21 | 3068 | listn (CONSTYPE_PURE, 2, Qsearch_failed, Qerror)); |
ca1d1d23 | 3069 | Fput (Qsearch_failed, Qerror_message, |
2a0213a6 | 3070 | build_pure_c_string ("Search failed")); |
ca1d1d23 JB |
3071 | |
3072 | Fput (Qinvalid_regexp, Qerror_conditions, | |
3438fe21 | 3073 | listn (CONSTYPE_PURE, 2, Qinvalid_regexp, Qerror)); |
ca1d1d23 | 3074 | Fput (Qinvalid_regexp, Qerror_message, |
2a0213a6 | 3075 | build_pure_c_string ("Invalid regexp")); |
ca1d1d23 | 3076 | |
daa37602 JB |
3077 | last_thing_searched = Qnil; |
3078 | staticpro (&last_thing_searched); | |
3079 | ||
0f6af254 DK |
3080 | saved_last_thing_searched = Qnil; |
3081 | staticpro (&saved_last_thing_searched); | |
3082 | ||
29208e82 | 3083 | DEFVAR_LISP ("search-spaces-regexp", Vsearch_spaces_regexp, |
e2811828 | 3084 | doc: /* Regexp to substitute for bunches of spaces in regexp search. |
f31a9a68 RS |
3085 | Some commands use this for user-specified regexps. |
3086 | Spaces that occur inside character classes or repetition operators | |
3087 | or other such regexp constructs are not replaced with this. | |
3088 | A value of nil (which is the normal value) means treat spaces literally. */); | |
41a33295 | 3089 | Vsearch_spaces_regexp = Qnil; |
f31a9a68 | 3090 | |
29208e82 | 3091 | DEFVAR_LISP ("inhibit-changing-match-data", Vinhibit_changing_match_data, |
ef887810 | 3092 | doc: /* Internal use only. |
39d0bf74 RS |
3093 | If non-nil, the primitive searching and matching functions |
3094 | such as `looking-at', `string-match', `re-search-forward', etc., | |
3095 | do not set the match data. The proper way to use this variable | |
3096 | is to bind it with `let' around a small expression. */); | |
ef887810 RS |
3097 | Vinhibit_changing_match_data = Qnil; |
3098 | ||
ca1d1d23 | 3099 | defsubr (&Slooking_at); |
b819a390 RS |
3100 | defsubr (&Sposix_looking_at); |
3101 | defsubr (&Sstring_match); | |
3102 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
3103 | defsubr (&Ssearch_forward); |
3104 | defsubr (&Ssearch_backward); | |
ca1d1d23 JB |
3105 | defsubr (&Sre_search_forward); |
3106 | defsubr (&Sre_search_backward); | |
b819a390 RS |
3107 | defsubr (&Sposix_search_forward); |
3108 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
3109 | defsubr (&Sreplace_match); |
3110 | defsubr (&Smatch_beginning); | |
3111 | defsubr (&Smatch_end); | |
3112 | defsubr (&Smatch_data); | |
3f1c005b | 3113 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
3114 | defsubr (&Sregexp_quote); |
3115 | } |