Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
73b0cd50 | 2 | Copyright (C) 1985-1987, 1993-1994, 1997-1999, 2001-2011 |
8cabe764 | 3 | Free Software Foundation, Inc. |
ca1d1d23 JB |
4 | |
5 | This file is part of GNU Emacs. | |
6 | ||
9ec0b715 | 7 | GNU Emacs is free software: you can redistribute it and/or modify |
ca1d1d23 | 8 | it under the terms of the GNU General Public License as published by |
9ec0b715 GM |
9 | the Free Software Foundation, either version 3 of the License, or |
10 | (at your option) any later version. | |
ca1d1d23 JB |
11 | |
12 | GNU Emacs is distributed in the hope that it will be useful, | |
13 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | GNU General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
9ec0b715 | 18 | along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */ |
ca1d1d23 JB |
19 | |
20 | ||
18160b98 | 21 | #include <config.h> |
d7306fe6 | 22 | #include <setjmp.h> |
ca1d1d23 JB |
23 | #include "lisp.h" |
24 | #include "syntax.h" | |
5679531d | 25 | #include "category.h" |
ca1d1d23 | 26 | #include "buffer.h" |
76eb0881 | 27 | #include "character.h" |
89ad6492 | 28 | #include "charset.h" |
9169c321 | 29 | #include "region-cache.h" |
ca1d1d23 | 30 | #include "commands.h" |
9ac0d9e0 | 31 | #include "blockinput.h" |
bf1760bb | 32 | #include "intervals.h" |
4746118a | 33 | |
ca1d1d23 JB |
34 | #include <sys/types.h> |
35 | #include "regex.h" | |
36 | ||
1d288aef | 37 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 38 | |
487282dc KH |
39 | /* If the regexp is non-nil, then the buffer contains the compiled form |
40 | of that regexp, suitable for searching. */ | |
1d288aef RS |
41 | struct regexp_cache |
42 | { | |
487282dc | 43 | struct regexp_cache *next; |
ecdb561e | 44 | Lisp_Object regexp, whitespace_regexp; |
b69e3c18 | 45 | /* Syntax table for which the regexp applies. We need this because |
58e95211 SM |
46 | of character classes. If this is t, then the compiled pattern is valid |
47 | for any syntax-table. */ | |
b69e3c18 | 48 | Lisp_Object syntax_table; |
487282dc KH |
49 | struct re_pattern_buffer buf; |
50 | char fastmap[0400]; | |
b819a390 RS |
51 | /* Nonzero means regexp was compiled to do full POSIX backtracking. */ |
52 | char posix; | |
487282dc | 53 | }; |
ca1d1d23 | 54 | |
487282dc KH |
55 | /* The instances of that struct. */ |
56 | struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; | |
ca1d1d23 | 57 | |
487282dc KH |
58 | /* The head of the linked list; points to the most recently used buffer. */ |
59 | struct regexp_cache *searchbuf_head; | |
ca1d1d23 | 60 | |
ca1d1d23 | 61 | |
4746118a JB |
62 | /* Every call to re_match, etc., must pass &search_regs as the regs |
63 | argument unless you can show it is unnecessary (i.e., if re_match | |
64 | is certainly going to be called again before region-around-match | |
65 | can be called). | |
66 | ||
67 | Since the registers are now dynamically allocated, we need to make | |
68 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
69 | been allocated by checking search_regs.num_regs. |
70 | ||
71 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
72 | buffer using bits in the re_pattern_buffer. This means that whenever |
73 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
74 | allocated any registers, and will allocate new registers the next |
75 | time you call a searching or matching function. Therefore, we need | |
76 | to call re_set_registers after compiling a new pattern or after | |
77 | setting the match registers, so that the regex functions will be | |
78 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
79 | static struct re_registers search_regs; |
80 | ||
daa37602 JB |
81 | /* The buffer in which the last search was performed, or |
82 | Qt if the last search was done in a string; | |
83 | Qnil if no searching has been done yet. */ | |
84 | static Lisp_Object last_thing_searched; | |
ca1d1d23 | 85 | |
8e6208c5 | 86 | /* error condition signaled when regexp compile_pattern fails */ |
ca1d1d23 | 87 | |
955cbe7b | 88 | static Lisp_Object Qinvalid_regexp; |
ca1d1d23 | 89 | |
06f77a8a | 90 | /* Error condition used for failing searches */ |
955cbe7b | 91 | static Lisp_Object Qsearch_failed; |
06f77a8a | 92 | |
f57e2426 J |
93 | static void set_search_regs (EMACS_INT, EMACS_INT); |
94 | static void save_search_regs (void); | |
c098fdb8 EZ |
95 | static EMACS_INT simple_search (EMACS_INT, unsigned char *, EMACS_INT, |
96 | EMACS_INT, Lisp_Object, EMACS_INT, EMACS_INT, | |
f57e2426 | 97 | EMACS_INT, EMACS_INT); |
1db5b1ad JB |
98 | static EMACS_INT boyer_moore (EMACS_INT, unsigned char *, EMACS_INT, |
99 | Lisp_Object, Lisp_Object, EMACS_INT, | |
100 | EMACS_INT, int); | |
f57e2426 | 101 | static EMACS_INT search_buffer (Lisp_Object, EMACS_INT, EMACS_INT, |
c098fdb8 | 102 | EMACS_INT, EMACS_INT, EMACS_INT, int, |
f57e2426 | 103 | Lisp_Object, Lisp_Object, int); |
971de7fb | 104 | static void matcher_overflow (void) NO_RETURN; |
b819a390 | 105 | |
ca1d1d23 | 106 | static void |
971de7fb | 107 | matcher_overflow (void) |
ca1d1d23 JB |
108 | { |
109 | error ("Stack overflow in regexp matcher"); | |
110 | } | |
111 | ||
b819a390 RS |
112 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
113 | PATTERN is the pattern to compile. | |
114 | CP is the place to put the result. | |
facdc750 | 115 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 | 116 | POSIX is nonzero if we want full backtracking (POSIX style) |
5679531d | 117 | for this pattern. 0 means backtrack only enough to get a valid match. |
bbc73b48 KH |
118 | |
119 | The behavior also depends on Vsearch_spaces_regexp. */ | |
ca1d1d23 | 120 | |
487282dc | 121 | static void |
cfe0661d | 122 | compile_pattern_1 (struct regexp_cache *cp, Lisp_Object pattern, Lisp_Object translate, int posix) |
ca1d1d23 | 123 | { |
d451e4db | 124 | char *val; |
b819a390 | 125 | reg_syntax_t old; |
ca1d1d23 | 126 | |
487282dc | 127 | cp->regexp = Qnil; |
59fab369 | 128 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 129 | cp->posix = posix; |
93daa011 | 130 | cp->buf.multibyte = STRING_MULTIBYTE (pattern); |
89ad6492 | 131 | cp->buf.charset_unibyte = charset_unibyte; |
d8c85250 CY |
132 | if (STRINGP (Vsearch_spaces_regexp)) |
133 | cp->whitespace_regexp = Vsearch_spaces_regexp; | |
134 | else | |
135 | cp->whitespace_regexp = Qnil; | |
136 | ||
e7b2dd2e RS |
137 | /* rms: I think BLOCK_INPUT is not needed here any more, |
138 | because regex.c defines malloc to call xmalloc. | |
139 | Using BLOCK_INPUT here means the debugger won't run if an error occurs. | |
140 | So let's turn it off. */ | |
141 | /* BLOCK_INPUT; */ | |
fb4a568d | 142 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 143 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
d8c85250 CY |
144 | |
145 | if (STRINGP (Vsearch_spaces_regexp)) | |
42a5b22f | 146 | re_set_whitespace_regexp (SSDATA (Vsearch_spaces_regexp)); |
d8c85250 CY |
147 | else |
148 | re_set_whitespace_regexp (NULL); | |
bbc73b48 | 149 | |
51b59d79 | 150 | val = (char *) re_compile_pattern (SSDATA (pattern), |
8f924df7 | 151 | SBYTES (pattern), &cp->buf); |
bbc73b48 | 152 | |
58e95211 SM |
153 | /* If the compiled pattern hard codes some of the contents of the |
154 | syntax-table, it can only be reused with *this* syntax table. */ | |
4b4deea2 | 155 | cp->syntax_table = cp->buf.used_syntax ? BVAR (current_buffer, syntax_table) : Qt; |
58e95211 | 156 | |
bbc73b48 KH |
157 | re_set_whitespace_regexp (NULL); |
158 | ||
b819a390 | 159 | re_set_syntax (old); |
e7b2dd2e | 160 | /* UNBLOCK_INPUT; */ |
ca1d1d23 | 161 | if (val) |
06f77a8a | 162 | xsignal1 (Qinvalid_regexp, build_string (val)); |
1113d9db | 163 | |
487282dc | 164 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
165 | } |
166 | ||
6efc7887 RS |
167 | /* Shrink each compiled regexp buffer in the cache |
168 | to the size actually used right now. | |
169 | This is called from garbage collection. */ | |
170 | ||
171 | void | |
971de7fb | 172 | shrink_regexp_cache (void) |
6efc7887 | 173 | { |
a968f437 | 174 | struct regexp_cache *cp; |
6efc7887 RS |
175 | |
176 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
177 | { | |
178 | cp->buf.allocated = cp->buf.used; | |
179 | cp->buf.buffer | |
b23c0a83 | 180 | = (unsigned char *) xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
181 | } |
182 | } | |
183 | ||
54dd3310 SM |
184 | /* Clear the regexp cache w.r.t. a particular syntax table, |
185 | because it was changed. | |
e5b94d44 CY |
186 | There is no danger of memory leak here because re_compile_pattern |
187 | automagically manages the memory in each re_pattern_buffer struct, | |
188 | based on its `allocated' and `buffer' values. */ | |
189 | void | |
971de7fb | 190 | clear_regexp_cache (void) |
e5b94d44 CY |
191 | { |
192 | int i; | |
193 | ||
194 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) | |
920fd1fc EZ |
195 | /* It's tempting to compare with the syntax-table we've actually changed, |
196 | but it's not sufficient because char-table inheritance means that | |
54dd3310 SM |
197 | modifying one syntax-table can change others at the same time. */ |
198 | if (!EQ (searchbufs[i].syntax_table, Qt)) | |
199 | searchbufs[i].regexp = Qnil; | |
e5b94d44 CY |
200 | } |
201 | ||
487282dc | 202 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
203 | the cache. |
204 | PATTERN is the pattern to compile. | |
facdc750 | 205 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
206 | REGP is the structure that says where to store the "register" |
207 | values that will result from matching this pattern. | |
208 | If it is 0, we should compile the pattern not to record any | |
209 | subexpression bounds. | |
210 | POSIX is nonzero if we want full backtracking (POSIX style) | |
211 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
212 | |
213 | struct re_pattern_buffer * | |
971de7fb | 214 | compile_pattern (Lisp_Object pattern, struct re_registers *regp, Lisp_Object translate, int posix, int multibyte) |
487282dc KH |
215 | { |
216 | struct regexp_cache *cp, **cpp; | |
217 | ||
218 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
219 | { | |
220 | cp = *cpp; | |
f1b9c7c1 KR |
221 | /* Entries are initialized to nil, and may be set to nil by |
222 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
49a5f770 KR |
223 | string accessors in those cases. However, compile_pattern_1 |
224 | is only applied to the cache entry we pick here to reuse. So | |
225 | nil should never appear before a non-nil entry. */ | |
7c752c80 | 226 | if (NILP (cp->regexp)) |
f1b9c7c1 | 227 | goto compile_it; |
d5db4077 | 228 | if (SCHARS (cp->regexp) == SCHARS (pattern) |
cf69b13e | 229 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 230 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 231 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d | 232 | && cp->posix == posix |
58e95211 | 233 | && (EQ (cp->syntax_table, Qt) |
4b4deea2 | 234 | || EQ (cp->syntax_table, BVAR (current_buffer, syntax_table))) |
89ad6492 KH |
235 | && !NILP (Fequal (cp->whitespace_regexp, Vsearch_spaces_regexp)) |
236 | && cp->buf.charset_unibyte == charset_unibyte) | |
487282dc KH |
237 | break; |
238 | ||
f1b9c7c1 KR |
239 | /* If we're at the end of the cache, compile into the nil cell |
240 | we found, or the last (least recently used) cell with a | |
241 | string value. */ | |
487282dc KH |
242 | if (cp->next == 0) |
243 | { | |
f1b9c7c1 | 244 | compile_it: |
cfe0661d | 245 | compile_pattern_1 (cp, pattern, translate, posix); |
487282dc KH |
246 | break; |
247 | } | |
248 | } | |
249 | ||
250 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
251 | either because we found it in the cache or because we just compiled it. | |
252 | Move it to the front of the queue to mark it as most recently used. */ | |
253 | *cpp = cp->next; | |
254 | cp->next = searchbuf_head; | |
255 | searchbuf_head = cp; | |
1113d9db | 256 | |
6639708c RS |
257 | /* Advise the searching functions about the space we have allocated |
258 | for register data. */ | |
259 | if (regp) | |
260 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
261 | ||
78edd3b7 | 262 | /* The compiled pattern can be used both for multibyte and unibyte |
89ad6492 KH |
263 | target. But, we have to tell which the pattern is used for. */ |
264 | cp->buf.target_multibyte = multibyte; | |
265 | ||
487282dc | 266 | return &cp->buf; |
ca1d1d23 JB |
267 | } |
268 | ||
ca1d1d23 | 269 | \f |
b819a390 | 270 | static Lisp_Object |
971de7fb | 271 | looking_at_1 (Lisp_Object string, int posix) |
ca1d1d23 JB |
272 | { |
273 | Lisp_Object val; | |
274 | unsigned char *p1, *p2; | |
e7deaab0 | 275 | EMACS_INT s1, s2; |
c098fdb8 | 276 | register EMACS_INT i; |
487282dc | 277 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 278 | |
7074fde6 FP |
279 | if (running_asynch_code) |
280 | save_search_regs (); | |
281 | ||
910c747a | 282 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
4b4deea2 TT |
283 | XCHAR_TABLE (BVAR (current_buffer, case_canon_table))->extras[2] |
284 | = BVAR (current_buffer, case_eqv_table); | |
910c747a | 285 | |
b7826503 | 286 | CHECK_STRING (string); |
ef887810 RS |
287 | bufp = compile_pattern (string, |
288 | (NILP (Vinhibit_changing_match_data) | |
289 | ? &search_regs : NULL), | |
4b4deea2 TT |
290 | (!NILP (BVAR (current_buffer, case_fold_search)) |
291 | ? BVAR (current_buffer, case_canon_table) : Qnil), | |
0c8533c6 | 292 | posix, |
4b4deea2 | 293 | !NILP (BVAR (current_buffer, enable_multibyte_characters))); |
ca1d1d23 JB |
294 | |
295 | immediate_quit = 1; | |
296 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
297 | ||
298 | /* Get pointers and sizes of the two strings | |
299 | that make up the visible portion of the buffer. */ | |
300 | ||
301 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 302 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 303 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 304 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
305 | if (s1 < 0) |
306 | { | |
307 | p2 = p1; | |
fa8ed3e0 | 308 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
309 | s1 = 0; |
310 | } | |
311 | if (s2 < 0) | |
312 | { | |
fa8ed3e0 | 313 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
314 | s2 = 0; |
315 | } | |
8bb43c28 RS |
316 | |
317 | re_match_object = Qnil; | |
177c0ea7 | 318 | |
487282dc | 319 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
ef887810 RS |
320 | PT_BYTE - BEGV_BYTE, |
321 | (NILP (Vinhibit_changing_match_data) | |
322 | ? &search_regs : NULL), | |
fa8ed3e0 | 323 | ZV_BYTE - BEGV_BYTE); |
de182d70 | 324 | immediate_quit = 0; |
177c0ea7 | 325 | |
ca1d1d23 JB |
326 | if (i == -2) |
327 | matcher_overflow (); | |
328 | ||
329 | val = (0 <= i ? Qt : Qnil); | |
ef887810 | 330 | if (NILP (Vinhibit_changing_match_data) && i >= 0) |
fa8ed3e0 RS |
331 | for (i = 0; i < search_regs.num_regs; i++) |
332 | if (search_regs.start[i] >= 0) | |
333 | { | |
334 | search_regs.start[i] | |
335 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
336 | search_regs.end[i] | |
337 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
338 | } | |
ef887810 RS |
339 | |
340 | /* Set last_thing_searched only when match data is changed. */ | |
341 | if (NILP (Vinhibit_changing_match_data)) | |
342 | XSETBUFFER (last_thing_searched, current_buffer); | |
343 | ||
ca1d1d23 JB |
344 | return val; |
345 | } | |
346 | ||
b819a390 | 347 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
348 | doc: /* Return t if text after point matches regular expression REGEXP. |
349 | This function modifies the match data that `match-beginning', | |
350 | `match-end' and `match-data' access; save and restore the match | |
351 | data if you want to preserve them. */) | |
5842a27b | 352 | (Lisp_Object regexp) |
b819a390 | 353 | { |
94f94972 | 354 | return looking_at_1 (regexp, 0); |
b819a390 RS |
355 | } |
356 | ||
357 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
358 | doc: /* Return t if text after point matches regular expression REGEXP. |
359 | Find the longest match, in accord with Posix regular expression rules. | |
360 | This function modifies the match data that `match-beginning', | |
361 | `match-end' and `match-data' access; save and restore the match | |
362 | data if you want to preserve them. */) | |
5842a27b | 363 | (Lisp_Object regexp) |
b819a390 | 364 | { |
94f94972 | 365 | return looking_at_1 (regexp, 1); |
b819a390 RS |
366 | } |
367 | \f | |
368 | static Lisp_Object | |
971de7fb | 369 | string_match_1 (Lisp_Object regexp, Lisp_Object string, Lisp_Object start, int posix) |
ca1d1d23 | 370 | { |
a53e2e89 | 371 | EMACS_INT val; |
487282dc | 372 | struct re_pattern_buffer *bufp; |
e7deaab0 | 373 | EMACS_INT pos, pos_byte; |
0c8533c6 | 374 | int i; |
ca1d1d23 | 375 | |
7074fde6 FP |
376 | if (running_asynch_code) |
377 | save_search_regs (); | |
378 | ||
b7826503 PJ |
379 | CHECK_STRING (regexp); |
380 | CHECK_STRING (string); | |
ca1d1d23 JB |
381 | |
382 | if (NILP (start)) | |
0c8533c6 | 383 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
384 | else |
385 | { | |
c098fdb8 | 386 | EMACS_INT len = SCHARS (string); |
ca1d1d23 | 387 | |
b7826503 | 388 | CHECK_NUMBER (start); |
0c8533c6 RS |
389 | pos = XINT (start); |
390 | if (pos < 0 && -pos <= len) | |
391 | pos = len + pos; | |
392 | else if (0 > pos || pos > len) | |
ca1d1d23 | 393 | args_out_of_range (string, start); |
0c8533c6 | 394 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
395 | } |
396 | ||
910c747a | 397 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
4b4deea2 TT |
398 | XCHAR_TABLE (BVAR (current_buffer, case_canon_table))->extras[2] |
399 | = BVAR (current_buffer, case_eqv_table); | |
910c747a | 400 | |
ef887810 RS |
401 | bufp = compile_pattern (regexp, |
402 | (NILP (Vinhibit_changing_match_data) | |
403 | ? &search_regs : NULL), | |
4b4deea2 TT |
404 | (!NILP (BVAR (current_buffer, case_fold_search)) |
405 | ? BVAR (current_buffer, case_canon_table) : Qnil), | |
0c8533c6 RS |
406 | posix, |
407 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 408 | immediate_quit = 1; |
8bb43c28 | 409 | re_match_object = string; |
177c0ea7 | 410 | |
51b59d79 | 411 | val = re_search (bufp, SSDATA (string), |
d5db4077 KR |
412 | SBYTES (string), pos_byte, |
413 | SBYTES (string) - pos_byte, | |
ef887810 RS |
414 | (NILP (Vinhibit_changing_match_data) |
415 | ? &search_regs : NULL)); | |
ca1d1d23 | 416 | immediate_quit = 0; |
ef887810 RS |
417 | |
418 | /* Set last_thing_searched only when match data is changed. */ | |
419 | if (NILP (Vinhibit_changing_match_data)) | |
420 | last_thing_searched = Qt; | |
421 | ||
ca1d1d23 JB |
422 | if (val == -2) |
423 | matcher_overflow (); | |
424 | if (val < 0) return Qnil; | |
0c8533c6 | 425 | |
ef887810 RS |
426 | if (NILP (Vinhibit_changing_match_data)) |
427 | for (i = 0; i < search_regs.num_regs; i++) | |
428 | if (search_regs.start[i] >= 0) | |
429 | { | |
430 | search_regs.start[i] | |
431 | = string_byte_to_char (string, search_regs.start[i]); | |
432 | search_regs.end[i] | |
433 | = string_byte_to_char (string, search_regs.end[i]); | |
434 | } | |
0c8533c6 RS |
435 | |
436 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 437 | } |
e59a8453 | 438 | |
16a97296 | 439 | DEFUE ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 | 440 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
b85acc4b | 441 | Matching ignores case if `case-fold-search' is non-nil. |
8c1a1077 PJ |
442 | If third arg START is non-nil, start search at that index in STRING. |
443 | For index of first char beyond the match, do (match-end 0). | |
444 | `match-end' and `match-beginning' also give indices of substrings | |
2bd2f32d RS |
445 | matched by parenthesis constructs in the pattern. |
446 | ||
447 | You can use the function `match-string' to extract the substrings | |
448 | matched by the parenthesis constructions in REGEXP. */) | |
5842a27b | 449 | (Lisp_Object regexp, Lisp_Object string, Lisp_Object start) |
b819a390 RS |
450 | { |
451 | return string_match_1 (regexp, string, start, 0); | |
452 | } | |
453 | ||
454 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
455 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
456 | Find the longest match, in accord with Posix regular expression rules. | |
457 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
458 | If third arg START is non-nil, start search at that index in STRING. | |
459 | For index of first char beyond the match, do (match-end 0). | |
460 | `match-end' and `match-beginning' also give indices of substrings | |
461 | matched by parenthesis constructs in the pattern. */) | |
5842a27b | 462 | (Lisp_Object regexp, Lisp_Object string, Lisp_Object start) |
b819a390 RS |
463 | { |
464 | return string_match_1 (regexp, string, start, 1); | |
465 | } | |
466 | ||
e59a8453 RS |
467 | /* Match REGEXP against STRING, searching all of STRING, |
468 | and return the index of the match, or negative on failure. | |
469 | This does not clobber the match data. */ | |
470 | ||
a53e2e89 | 471 | EMACS_INT |
971de7fb | 472 | fast_string_match (Lisp_Object regexp, Lisp_Object string) |
e59a8453 | 473 | { |
a53e2e89 | 474 | EMACS_INT val; |
487282dc | 475 | struct re_pattern_buffer *bufp; |
e59a8453 | 476 | |
facdc750 RS |
477 | bufp = compile_pattern (regexp, 0, Qnil, |
478 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 479 | immediate_quit = 1; |
8bb43c28 | 480 | re_match_object = string; |
177c0ea7 | 481 | |
51b59d79 | 482 | val = re_search (bufp, SSDATA (string), |
d5db4077 KR |
483 | SBYTES (string), 0, |
484 | SBYTES (string), 0); | |
e59a8453 RS |
485 | immediate_quit = 0; |
486 | return val; | |
487 | } | |
5679531d KH |
488 | |
489 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
490 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
491 | This does not clobber the match data. |
492 | We assume that STRING contains single-byte characters. */ | |
5679531d | 493 | |
a53e2e89 | 494 | EMACS_INT |
971de7fb | 495 | fast_c_string_match_ignore_case (Lisp_Object regexp, const char *string) |
5679531d | 496 | { |
a53e2e89 | 497 | EMACS_INT val; |
5679531d | 498 | struct re_pattern_buffer *bufp; |
c098fdb8 | 499 | size_t len = strlen (string); |
5679531d | 500 | |
0c8533c6 | 501 | regexp = string_make_unibyte (regexp); |
b4577c63 | 502 | re_match_object = Qt; |
5679531d | 503 | bufp = compile_pattern (regexp, 0, |
0190922f | 504 | Vascii_canon_table, 0, |
f8bd51c4 | 505 | 0); |
5679531d KH |
506 | immediate_quit = 1; |
507 | val = re_search (bufp, string, len, 0, len, 0); | |
508 | immediate_quit = 0; | |
509 | return val; | |
510 | } | |
be5f4dfb KH |
511 | |
512 | /* Like fast_string_match but ignore case. */ | |
513 | ||
a53e2e89 | 514 | EMACS_INT |
971de7fb | 515 | fast_string_match_ignore_case (Lisp_Object regexp, Lisp_Object string) |
be5f4dfb | 516 | { |
a53e2e89 | 517 | EMACS_INT val; |
be5f4dfb KH |
518 | struct re_pattern_buffer *bufp; |
519 | ||
0190922f | 520 | bufp = compile_pattern (regexp, 0, Vascii_canon_table, |
be5f4dfb KH |
521 | 0, STRING_MULTIBYTE (string)); |
522 | immediate_quit = 1; | |
523 | re_match_object = string; | |
524 | ||
51b59d79 | 525 | val = re_search (bufp, SSDATA (string), |
be5f4dfb KH |
526 | SBYTES (string), 0, |
527 | SBYTES (string), 0); | |
528 | immediate_quit = 0; | |
529 | return val; | |
530 | } | |
ca1d1d23 | 531 | \f |
a80ce213 | 532 | /* Match REGEXP against the characters after POS to LIMIT, and return |
8b13507a KH |
533 | the number of matched characters. If STRING is non-nil, match |
534 | against the characters in it. In that case, POS and LIMIT are | |
535 | indices into the string. This function doesn't modify the match | |
536 | data. */ | |
537 | ||
538 | EMACS_INT | |
971de7fb | 539 | fast_looking_at (Lisp_Object regexp, EMACS_INT pos, EMACS_INT pos_byte, EMACS_INT limit, EMACS_INT limit_byte, Lisp_Object string) |
8b13507a KH |
540 | { |
541 | int multibyte; | |
542 | struct re_pattern_buffer *buf; | |
543 | unsigned char *p1, *p2; | |
e7deaab0 | 544 | EMACS_INT s1, s2; |
8b13507a | 545 | EMACS_INT len; |
78edd3b7 | 546 | |
8b13507a KH |
547 | if (STRINGP (string)) |
548 | { | |
549 | if (pos_byte < 0) | |
550 | pos_byte = string_char_to_byte (string, pos); | |
551 | if (limit_byte < 0) | |
552 | limit_byte = string_char_to_byte (string, limit); | |
553 | p1 = NULL; | |
554 | s1 = 0; | |
555 | p2 = SDATA (string); | |
556 | s2 = SBYTES (string); | |
557 | re_match_object = string; | |
558 | multibyte = STRING_MULTIBYTE (string); | |
559 | } | |
560 | else | |
561 | { | |
562 | if (pos_byte < 0) | |
563 | pos_byte = CHAR_TO_BYTE (pos); | |
564 | if (limit_byte < 0) | |
565 | limit_byte = CHAR_TO_BYTE (limit); | |
566 | pos_byte -= BEGV_BYTE; | |
567 | limit_byte -= BEGV_BYTE; | |
568 | p1 = BEGV_ADDR; | |
569 | s1 = GPT_BYTE - BEGV_BYTE; | |
570 | p2 = GAP_END_ADDR; | |
571 | s2 = ZV_BYTE - GPT_BYTE; | |
572 | if (s1 < 0) | |
573 | { | |
574 | p2 = p1; | |
575 | s2 = ZV_BYTE - BEGV_BYTE; | |
576 | s1 = 0; | |
577 | } | |
578 | if (s2 < 0) | |
579 | { | |
580 | s1 = ZV_BYTE - BEGV_BYTE; | |
581 | s2 = 0; | |
582 | } | |
583 | re_match_object = Qnil; | |
4b4deea2 | 584 | multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
8b13507a KH |
585 | } |
586 | ||
587 | buf = compile_pattern (regexp, 0, Qnil, 0, multibyte); | |
588 | immediate_quit = 1; | |
589 | len = re_match_2 (buf, (char *) p1, s1, (char *) p2, s2, | |
590 | pos_byte, NULL, limit_byte); | |
591 | immediate_quit = 0; | |
592 | ||
593 | return len; | |
594 | } | |
595 | ||
596 | \f | |
9169c321 JB |
597 | /* The newline cache: remembering which sections of text have no newlines. */ |
598 | ||
599 | /* If the user has requested newline caching, make sure it's on. | |
600 | Otherwise, make sure it's off. | |
601 | This is our cheezy way of associating an action with the change of | |
602 | state of a buffer-local variable. */ | |
603 | static void | |
971de7fb | 604 | newline_cache_on_off (struct buffer *buf) |
9169c321 | 605 | { |
4b4deea2 | 606 | if (NILP (BVAR (buf, cache_long_line_scans))) |
9169c321 JB |
607 | { |
608 | /* It should be off. */ | |
609 | if (buf->newline_cache) | |
610 | { | |
611 | free_region_cache (buf->newline_cache); | |
612 | buf->newline_cache = 0; | |
613 | } | |
614 | } | |
615 | else | |
616 | { | |
617 | /* It should be on. */ | |
618 | if (buf->newline_cache == 0) | |
619 | buf->newline_cache = new_region_cache (); | |
620 | } | |
621 | } | |
622 | ||
623 | \f | |
624 | /* Search for COUNT instances of the character TARGET between START and END. | |
625 | ||
626 | If COUNT is positive, search forwards; END must be >= START. | |
627 | If COUNT is negative, search backwards for the -COUNTth instance; | |
628 | END must be <= START. | |
629 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
630 | ||
631 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
632 | direction indicated by COUNT. | |
ffd56f97 JB |
633 | |
634 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
a9f2a45f | 635 | position past the COUNTth match. Note that for reverse motion |
5bfe95c9 | 636 | this is not the same as the usual convention for Emacs motion commands. |
ffd56f97 | 637 | |
9169c321 JB |
638 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
639 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 640 | |
087a5f81 RS |
641 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
642 | except when inside redisplay. */ | |
643 | ||
c098fdb8 EZ |
644 | EMACS_INT |
645 | scan_buffer (register int target, EMACS_INT start, EMACS_INT end, | |
a53e2e89 | 646 | EMACS_INT count, EMACS_INT *shortage, int allow_quit) |
ca1d1d23 | 647 | { |
9169c321 | 648 | struct region_cache *newline_cache; |
177c0ea7 | 649 | int direction; |
ffd56f97 | 650 | |
9169c321 JB |
651 | if (count > 0) |
652 | { | |
653 | direction = 1; | |
654 | if (! end) end = ZV; | |
655 | } | |
656 | else | |
657 | { | |
658 | direction = -1; | |
659 | if (! end) end = BEGV; | |
660 | } | |
ffd56f97 | 661 | |
9169c321 JB |
662 | newline_cache_on_off (current_buffer); |
663 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
664 | |
665 | if (shortage != 0) | |
666 | *shortage = 0; | |
667 | ||
087a5f81 | 668 | immediate_quit = allow_quit; |
ca1d1d23 | 669 | |
ffd56f97 | 670 | if (count > 0) |
9169c321 | 671 | while (start != end) |
ca1d1d23 | 672 | { |
9169c321 JB |
673 | /* Our innermost scanning loop is very simple; it doesn't know |
674 | about gaps, buffer ends, or the newline cache. ceiling is | |
675 | the position of the last character before the next such | |
676 | obstacle --- the last character the dumb search loop should | |
677 | examine. */ | |
e7deaab0 AS |
678 | EMACS_INT ceiling_byte = CHAR_TO_BYTE (end) - 1; |
679 | EMACS_INT start_byte = CHAR_TO_BYTE (start); | |
680 | EMACS_INT tem; | |
9169c321 JB |
681 | |
682 | /* If we're looking for a newline, consult the newline cache | |
683 | to see where we can avoid some scanning. */ | |
684 | if (target == '\n' && newline_cache) | |
685 | { | |
c098fdb8 | 686 | EMACS_INT next_change; |
9169c321 JB |
687 | immediate_quit = 0; |
688 | while (region_cache_forward | |
fa8ed3e0 RS |
689 | (current_buffer, newline_cache, start_byte, &next_change)) |
690 | start_byte = next_change; | |
cbe0db0d | 691 | immediate_quit = allow_quit; |
9169c321 | 692 | |
fa8ed3e0 RS |
693 | /* START should never be after END. */ |
694 | if (start_byte > ceiling_byte) | |
695 | start_byte = ceiling_byte; | |
9169c321 JB |
696 | |
697 | /* Now the text after start is an unknown region, and | |
698 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 699 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
700 | } |
701 | ||
702 | /* The dumb loop can only scan text stored in contiguous | |
703 | bytes. BUFFER_CEILING_OF returns the last character | |
704 | position that is contiguous, so the ceiling is the | |
705 | position after that. */ | |
67ce527d KH |
706 | tem = BUFFER_CEILING_OF (start_byte); |
707 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
708 | |
709 | { | |
177c0ea7 | 710 | /* The termination address of the dumb loop. */ |
fa8ed3e0 RS |
711 | register unsigned char *ceiling_addr |
712 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
713 | register unsigned char *cursor | |
714 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
715 | unsigned char *base = cursor; |
716 | ||
717 | while (cursor < ceiling_addr) | |
718 | { | |
719 | unsigned char *scan_start = cursor; | |
720 | ||
721 | /* The dumb loop. */ | |
722 | while (*cursor != target && ++cursor < ceiling_addr) | |
723 | ; | |
724 | ||
725 | /* If we're looking for newlines, cache the fact that | |
726 | the region from start to cursor is free of them. */ | |
727 | if (target == '\n' && newline_cache) | |
728 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
729 | start_byte + scan_start - base, |
730 | start_byte + cursor - base); | |
9169c321 JB |
731 | |
732 | /* Did we find the target character? */ | |
733 | if (cursor < ceiling_addr) | |
734 | { | |
735 | if (--count == 0) | |
736 | { | |
737 | immediate_quit = 0; | |
fa8ed3e0 | 738 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
739 | } |
740 | cursor++; | |
741 | } | |
742 | } | |
743 | ||
fa8ed3e0 | 744 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 745 | } |
ca1d1d23 JB |
746 | } |
747 | else | |
9169c321 JB |
748 | while (start > end) |
749 | { | |
750 | /* The last character to check before the next obstacle. */ | |
e7deaab0 AS |
751 | EMACS_INT ceiling_byte = CHAR_TO_BYTE (end); |
752 | EMACS_INT start_byte = CHAR_TO_BYTE (start); | |
753 | EMACS_INT tem; | |
9169c321 JB |
754 | |
755 | /* Consult the newline cache, if appropriate. */ | |
756 | if (target == '\n' && newline_cache) | |
757 | { | |
c098fdb8 | 758 | EMACS_INT next_change; |
9169c321 JB |
759 | immediate_quit = 0; |
760 | while (region_cache_backward | |
fa8ed3e0 RS |
761 | (current_buffer, newline_cache, start_byte, &next_change)) |
762 | start_byte = next_change; | |
cbe0db0d | 763 | immediate_quit = allow_quit; |
9169c321 JB |
764 | |
765 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
766 | if (start_byte <= ceiling_byte) |
767 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
768 | |
769 | /* Now the text before start is an unknown region, and | |
770 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 771 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
772 | } |
773 | ||
774 | /* Stop scanning before the gap. */ | |
67ce527d KH |
775 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
776 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
777 | |
778 | { | |
779 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
780 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
781 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
782 | unsigned char *base = cursor; |
783 | ||
784 | while (cursor >= ceiling_addr) | |
785 | { | |
786 | unsigned char *scan_start = cursor; | |
787 | ||
788 | while (*cursor != target && --cursor >= ceiling_addr) | |
789 | ; | |
790 | ||
791 | /* If we're looking for newlines, cache the fact that | |
792 | the region from after the cursor to start is free of them. */ | |
793 | if (target == '\n' && newline_cache) | |
794 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
795 | start_byte + cursor - base, |
796 | start_byte + scan_start - base); | |
9169c321 JB |
797 | |
798 | /* Did we find the target character? */ | |
799 | if (cursor >= ceiling_addr) | |
800 | { | |
801 | if (++count >= 0) | |
802 | { | |
803 | immediate_quit = 0; | |
fa8ed3e0 | 804 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
805 | } |
806 | cursor--; | |
807 | } | |
808 | } | |
809 | ||
fa8ed3e0 | 810 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
811 | } |
812 | } | |
813 | ||
ca1d1d23 JB |
814 | immediate_quit = 0; |
815 | if (shortage != 0) | |
ffd56f97 | 816 | *shortage = count * direction; |
9169c321 | 817 | return start; |
ca1d1d23 | 818 | } |
fa8ed3e0 RS |
819 | \f |
820 | /* Search for COUNT instances of a line boundary, which means either a | |
821 | newline or (if selective display enabled) a carriage return. | |
822 | Start at START. If COUNT is negative, search backwards. | |
823 | ||
824 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
825 | ||
826 | If we find COUNT instances. we position after (always after, | |
827 | even if scanning backwards) the COUNTth match, and return 0. | |
828 | ||
829 | If we don't find COUNT instances before reaching the end of the | |
830 | buffer (or the beginning, if scanning backwards), we return | |
831 | the number of line boundaries left unfound, and position at | |
832 | the limit we bumped up against. | |
833 | ||
834 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 835 | except in special cases. */ |
ca1d1d23 | 836 | |
c098fdb8 EZ |
837 | EMACS_INT |
838 | scan_newline (EMACS_INT start, EMACS_INT start_byte, | |
839 | EMACS_INT limit, EMACS_INT limit_byte, | |
840 | register EMACS_INT count, int allow_quit) | |
63fa018d | 841 | { |
fa8ed3e0 RS |
842 | int direction = ((count > 0) ? 1 : -1); |
843 | ||
844 | register unsigned char *cursor; | |
845 | unsigned char *base; | |
846 | ||
e7deaab0 | 847 | EMACS_INT ceiling; |
fa8ed3e0 RS |
848 | register unsigned char *ceiling_addr; |
849 | ||
d5d57b92 RS |
850 | int old_immediate_quit = immediate_quit; |
851 | ||
fa8ed3e0 RS |
852 | /* The code that follows is like scan_buffer |
853 | but checks for either newline or carriage return. */ | |
854 | ||
d5d57b92 RS |
855 | if (allow_quit) |
856 | immediate_quit++; | |
fa8ed3e0 RS |
857 | |
858 | start_byte = CHAR_TO_BYTE (start); | |
859 | ||
860 | if (count > 0) | |
861 | { | |
862 | while (start_byte < limit_byte) | |
863 | { | |
864 | ceiling = BUFFER_CEILING_OF (start_byte); | |
865 | ceiling = min (limit_byte - 1, ceiling); | |
866 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
867 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
868 | while (1) | |
869 | { | |
870 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
871 | ; | |
872 | ||
873 | if (cursor != ceiling_addr) | |
874 | { | |
875 | if (--count == 0) | |
876 | { | |
d5d57b92 | 877 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
878 | start_byte = start_byte + cursor - base + 1; |
879 | start = BYTE_TO_CHAR (start_byte); | |
880 | TEMP_SET_PT_BOTH (start, start_byte); | |
881 | return 0; | |
882 | } | |
883 | else | |
884 | if (++cursor == ceiling_addr) | |
885 | break; | |
886 | } | |
887 | else | |
888 | break; | |
889 | } | |
890 | start_byte += cursor - base; | |
891 | } | |
892 | } | |
893 | else | |
894 | { | |
fa8ed3e0 RS |
895 | while (start_byte > limit_byte) |
896 | { | |
897 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
898 | ceiling = max (limit_byte, ceiling); | |
899 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
900 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
901 | while (1) | |
902 | { | |
903 | while (--cursor != ceiling_addr && *cursor != '\n') | |
904 | ; | |
905 | ||
906 | if (cursor != ceiling_addr) | |
907 | { | |
908 | if (++count == 0) | |
909 | { | |
d5d57b92 | 910 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
911 | /* Return the position AFTER the match we found. */ |
912 | start_byte = start_byte + cursor - base + 1; | |
913 | start = BYTE_TO_CHAR (start_byte); | |
914 | TEMP_SET_PT_BOTH (start, start_byte); | |
915 | return 0; | |
916 | } | |
917 | } | |
918 | else | |
919 | break; | |
920 | } | |
921 | /* Here we add 1 to compensate for the last decrement | |
922 | of CURSOR, which took it past the valid range. */ | |
923 | start_byte += cursor - base + 1; | |
924 | } | |
925 | } | |
926 | ||
927 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 928 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
929 | |
930 | return count * direction; | |
63fa018d RS |
931 | } |
932 | ||
c098fdb8 EZ |
933 | EMACS_INT |
934 | find_next_newline_no_quit (EMACS_INT from, EMACS_INT cnt) | |
ca1d1d23 | 935 | { |
a53e2e89 | 936 | return scan_buffer ('\n', from, 0, cnt, (EMACS_INT *) 0, 0); |
9169c321 JB |
937 | } |
938 | ||
9169c321 JB |
939 | /* Like find_next_newline, but returns position before the newline, |
940 | not after, and only search up to TO. This isn't just | |
941 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 942 | |
c098fdb8 EZ |
943 | EMACS_INT |
944 | find_before_next_newline (EMACS_INT from, EMACS_INT to, EMACS_INT cnt) | |
9169c321 | 945 | { |
a53e2e89 | 946 | EMACS_INT shortage; |
c098fdb8 | 947 | EMACS_INT pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); |
9169c321 JB |
948 | |
949 | if (shortage == 0) | |
950 | pos--; | |
177c0ea7 | 951 | |
9169c321 | 952 | return pos; |
ca1d1d23 JB |
953 | } |
954 | \f | |
ca1d1d23 JB |
955 | /* Subroutines of Lisp buffer search functions. */ |
956 | ||
957 | static Lisp_Object | |
c098fdb8 EZ |
958 | search_command (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, |
959 | Lisp_Object count, int direction, int RE, int posix) | |
ca1d1d23 | 960 | { |
a53e2e89 | 961 | register EMACS_INT np; |
c098fdb8 | 962 | EMACS_INT lim, lim_byte; |
a53e2e89 | 963 | EMACS_INT n = direction; |
ca1d1d23 JB |
964 | |
965 | if (!NILP (count)) | |
966 | { | |
b7826503 | 967 | CHECK_NUMBER (count); |
ca1d1d23 JB |
968 | n *= XINT (count); |
969 | } | |
970 | ||
b7826503 | 971 | CHECK_STRING (string); |
ca1d1d23 | 972 | if (NILP (bound)) |
9f43ad85 RS |
973 | { |
974 | if (n > 0) | |
975 | lim = ZV, lim_byte = ZV_BYTE; | |
976 | else | |
977 | lim = BEGV, lim_byte = BEGV_BYTE; | |
978 | } | |
ca1d1d23 JB |
979 | else |
980 | { | |
b7826503 | 981 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 982 | lim = XINT (bound); |
6ec8bbd2 | 983 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
984 | error ("Invalid search bound (wrong side of point)"); |
985 | if (lim > ZV) | |
9f43ad85 | 986 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 987 | else if (lim < BEGV) |
9f43ad85 | 988 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
989 | else |
990 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
991 | } |
992 | ||
910c747a | 993 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
4b4deea2 TT |
994 | XCHAR_TABLE (BVAR (current_buffer, case_canon_table))->extras[2] |
995 | = BVAR (current_buffer, case_eqv_table); | |
910c747a | 996 | |
9f43ad85 | 997 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
4b4deea2 TT |
998 | (!NILP (BVAR (current_buffer, case_fold_search)) |
999 | ? BVAR (current_buffer, case_canon_table) | |
3135e9fd | 1000 | : Qnil), |
4b4deea2 TT |
1001 | (!NILP (BVAR (current_buffer, case_fold_search)) |
1002 | ? BVAR (current_buffer, case_eqv_table) | |
3135e9fd | 1003 | : Qnil), |
b819a390 | 1004 | posix); |
ca1d1d23 JB |
1005 | if (np <= 0) |
1006 | { | |
1007 | if (NILP (noerror)) | |
06f77a8a KS |
1008 | xsignal1 (Qsearch_failed, string); |
1009 | ||
ca1d1d23 JB |
1010 | if (!EQ (noerror, Qt)) |
1011 | { | |
1012 | if (lim < BEGV || lim > ZV) | |
1013 | abort (); | |
9f43ad85 | 1014 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
1015 | return Qnil; |
1016 | #if 0 /* This would be clean, but maybe programs depend on | |
1017 | a value of nil here. */ | |
481399bf | 1018 | np = lim; |
a5f217b8 | 1019 | #endif |
ca1d1d23 | 1020 | } |
481399bf RS |
1021 | else |
1022 | return Qnil; | |
ca1d1d23 JB |
1023 | } |
1024 | ||
1025 | if (np < BEGV || np > ZV) | |
1026 | abort (); | |
1027 | ||
1028 | SET_PT (np); | |
1029 | ||
1030 | return make_number (np); | |
1031 | } | |
1032 | \f | |
fa8ed3e0 RS |
1033 | /* Return 1 if REGEXP it matches just one constant string. */ |
1034 | ||
b6d6a51c | 1035 | static int |
971de7fb | 1036 | trivial_regexp_p (Lisp_Object regexp) |
b6d6a51c | 1037 | { |
c098fdb8 | 1038 | EMACS_INT len = SBYTES (regexp); |
d5db4077 | 1039 | unsigned char *s = SDATA (regexp); |
b6d6a51c KH |
1040 | while (--len >= 0) |
1041 | { | |
1042 | switch (*s++) | |
1043 | { | |
1044 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
1045 | return 0; | |
1046 | case '\\': | |
1047 | if (--len < 0) | |
1048 | return 0; | |
1049 | switch (*s++) | |
1050 | { | |
1051 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
1052 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 1053 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 1054 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 1055 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
1056 | case '6': case '7': case '8': case '9': |
1057 | return 0; | |
1058 | } | |
1059 | } | |
1060 | } | |
1061 | return 1; | |
1062 | } | |
1063 | ||
ca325161 | 1064 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 1065 | starting at position POS and stopping at position LIM, |
b819a390 | 1066 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
1067 | a regular expression if RE is true. |
1068 | ||
1069 | If N is positive, searching is forward and LIM must be greater than POS. | |
1070 | If N is negative, searching is backward and LIM must be less than POS. | |
1071 | ||
facdc750 | 1072 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 1073 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
1074 | (if searching backward) or the end (if searching forward). |
1075 | ||
1076 | POSIX is nonzero if we want full backtracking (POSIX style) | |
1077 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 1078 | |
aff2ce94 RS |
1079 | #define TRANSLATE(out, trt, d) \ |
1080 | do \ | |
1081 | { \ | |
1082 | if (! NILP (trt)) \ | |
1083 | { \ | |
1084 | Lisp_Object temp; \ | |
1085 | temp = Faref (trt, make_number (d)); \ | |
1086 | if (INTEGERP (temp)) \ | |
1087 | out = XINT (temp); \ | |
1088 | else \ | |
1089 | out = d; \ | |
1090 | } \ | |
1091 | else \ | |
1092 | out = d; \ | |
1093 | } \ | |
1094 | while (0) | |
facdc750 | 1095 | |
ef887810 RS |
1096 | /* Only used in search_buffer, to record the end position of the match |
1097 | when searching regexps and SEARCH_REGS should not be changed | |
1098 | (i.e. Vinhibit_changing_match_data is non-nil). */ | |
1099 | static struct re_registers search_regs_1; | |
1100 | ||
e7deaab0 | 1101 | static EMACS_INT |
d5a3eaaf | 1102 | search_buffer (Lisp_Object string, EMACS_INT pos, EMACS_INT pos_byte, |
c098fdb8 | 1103 | EMACS_INT lim, EMACS_INT lim_byte, EMACS_INT n, |
d5a3eaaf | 1104 | int RE, Lisp_Object trt, Lisp_Object inverse_trt, int posix) |
ca1d1d23 | 1105 | { |
c098fdb8 EZ |
1106 | EMACS_INT len = SCHARS (string); |
1107 | EMACS_INT len_byte = SBYTES (string); | |
facdc750 | 1108 | register int i; |
ca1d1d23 | 1109 | |
7074fde6 FP |
1110 | if (running_asynch_code) |
1111 | save_search_regs (); | |
1112 | ||
a7e4cdde | 1113 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1114 | /* Null string is found at starting position. */ |
a7e4cdde | 1115 | if (len == 0 || n == 0) |
ca325161 | 1116 | { |
0353b28f | 1117 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1118 | return pos; |
1119 | } | |
3f57a499 | 1120 | |
41a33295 | 1121 | if (RE && !(trivial_regexp_p (string) && NILP (Vsearch_spaces_regexp))) |
ca1d1d23 | 1122 | { |
facdc750 | 1123 | unsigned char *p1, *p2; |
c098fdb8 | 1124 | EMACS_INT s1, s2; |
487282dc KH |
1125 | struct re_pattern_buffer *bufp; |
1126 | ||
ef887810 RS |
1127 | bufp = compile_pattern (string, |
1128 | (NILP (Vinhibit_changing_match_data) | |
1129 | ? &search_regs : &search_regs_1), | |
1130 | trt, posix, | |
4b4deea2 | 1131 | !NILP (BVAR (current_buffer, enable_multibyte_characters))); |
ca1d1d23 | 1132 | |
ca1d1d23 JB |
1133 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1134 | because letting this function finish | |
1135 | can take too long. */ | |
1136 | QUIT; /* Do a pending quit right away, | |
1137 | to avoid paradoxical behavior */ | |
1138 | /* Get pointers and sizes of the two strings | |
1139 | that make up the visible portion of the buffer. */ | |
1140 | ||
1141 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1142 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1143 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1144 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1145 | if (s1 < 0) |
1146 | { | |
1147 | p2 = p1; | |
fa8ed3e0 | 1148 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1149 | s1 = 0; |
1150 | } | |
1151 | if (s2 < 0) | |
1152 | { | |
fa8ed3e0 | 1153 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1154 | s2 = 0; |
1155 | } | |
8bb43c28 | 1156 | re_match_object = Qnil; |
177c0ea7 | 1157 | |
ca1d1d23 JB |
1158 | while (n < 0) |
1159 | { | |
c098fdb8 | 1160 | EMACS_INT val; |
487282dc | 1161 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1162 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1163 | (NILP (Vinhibit_changing_match_data) |
1164 | ? &search_regs : &search_regs_1), | |
42db823b | 1165 | /* Don't allow match past current point */ |
4996330b | 1166 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1167 | if (val == -2) |
b6d6a51c KH |
1168 | { |
1169 | matcher_overflow (); | |
1170 | } | |
ca1d1d23 JB |
1171 | if (val >= 0) |
1172 | { | |
ef887810 RS |
1173 | if (NILP (Vinhibit_changing_match_data)) |
1174 | { | |
1175 | pos_byte = search_regs.start[0] + BEGV_BYTE; | |
1176 | for (i = 0; i < search_regs.num_regs; i++) | |
1177 | if (search_regs.start[i] >= 0) | |
1178 | { | |
1179 | search_regs.start[i] | |
1180 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1181 | search_regs.end[i] | |
1182 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1183 | } | |
1184 | XSETBUFFER (last_thing_searched, current_buffer); | |
1185 | /* Set pos to the new position. */ | |
1186 | pos = search_regs.start[0]; | |
1187 | } | |
1188 | else | |
1189 | { | |
1190 | pos_byte = search_regs_1.start[0] + BEGV_BYTE; | |
1191 | /* Set pos to the new position. */ | |
1192 | pos = BYTE_TO_CHAR (search_regs_1.start[0] + BEGV_BYTE); | |
1193 | } | |
ca1d1d23 JB |
1194 | } |
1195 | else | |
1196 | { | |
1197 | immediate_quit = 0; | |
1198 | return (n); | |
1199 | } | |
1200 | n++; | |
1201 | } | |
1202 | while (n > 0) | |
1203 | { | |
c098fdb8 | 1204 | EMACS_INT val; |
487282dc | 1205 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1206 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1207 | (NILP (Vinhibit_changing_match_data) |
1208 | ? &search_regs : &search_regs_1), | |
4996330b | 1209 | lim_byte - BEGV_BYTE); |
ca1d1d23 | 1210 | if (val == -2) |
b6d6a51c KH |
1211 | { |
1212 | matcher_overflow (); | |
1213 | } | |
ca1d1d23 JB |
1214 | if (val >= 0) |
1215 | { | |
ef887810 RS |
1216 | if (NILP (Vinhibit_changing_match_data)) |
1217 | { | |
1218 | pos_byte = search_regs.end[0] + BEGV_BYTE; | |
1219 | for (i = 0; i < search_regs.num_regs; i++) | |
1220 | if (search_regs.start[i] >= 0) | |
1221 | { | |
1222 | search_regs.start[i] | |
1223 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1224 | search_regs.end[i] | |
1225 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1226 | } | |
1227 | XSETBUFFER (last_thing_searched, current_buffer); | |
1228 | pos = search_regs.end[0]; | |
1229 | } | |
1230 | else | |
1231 | { | |
1232 | pos_byte = search_regs_1.end[0] + BEGV_BYTE; | |
1233 | pos = BYTE_TO_CHAR (search_regs_1.end[0] + BEGV_BYTE); | |
1234 | } | |
ca1d1d23 JB |
1235 | } |
1236 | else | |
1237 | { | |
1238 | immediate_quit = 0; | |
1239 | return (0 - n); | |
1240 | } | |
1241 | n--; | |
1242 | } | |
1243 | immediate_quit = 0; | |
1244 | return (pos); | |
1245 | } | |
1246 | else /* non-RE case */ | |
1247 | { | |
facdc750 | 1248 | unsigned char *raw_pattern, *pat; |
c098fdb8 EZ |
1249 | EMACS_INT raw_pattern_size; |
1250 | EMACS_INT raw_pattern_size_byte; | |
facdc750 | 1251 | unsigned char *patbuf; |
4b4deea2 | 1252 | int multibyte = !NILP (BVAR (current_buffer, enable_multibyte_characters)); |
a967ed62 | 1253 | unsigned char *base_pat; |
49e44e81 KH |
1254 | /* Set to positive if we find a non-ASCII char that need |
1255 | translation. Otherwise set to zero later. */ | |
1256 | int char_base = -1; | |
040272ce | 1257 | int boyer_moore_ok = 1; |
facdc750 RS |
1258 | |
1259 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1260 | We must convert PATTERN to match that, or we will not really | |
1261 | find things right. */ | |
1262 | ||
1263 | if (multibyte == STRING_MULTIBYTE (string)) | |
1264 | { | |
51b59d79 | 1265 | raw_pattern = SDATA (string); |
d5db4077 KR |
1266 | raw_pattern_size = SCHARS (string); |
1267 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1268 | } |
1269 | else if (multibyte) | |
1270 | { | |
d5db4077 | 1271 | raw_pattern_size = SCHARS (string); |
facdc750 | 1272 | raw_pattern_size_byte |
d5db4077 | 1273 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1274 | raw_pattern_size); |
7276d3d8 | 1275 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1276 | copy_text (SDATA (string), raw_pattern, |
1277 | SCHARS (string), 0, 1); | |
facdc750 RS |
1278 | } |
1279 | else | |
1280 | { | |
1281 | /* Converting multibyte to single-byte. | |
1282 | ||
1283 | ??? Perhaps this conversion should be done in a special way | |
1284 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1285 | so that only the multibyte chars which really correspond to | |
1286 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1287 | raw_pattern_size = SCHARS (string); |
1288 | raw_pattern_size_byte = SCHARS (string); | |
7276d3d8 | 1289 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
1290 | copy_text (SDATA (string), raw_pattern, |
1291 | SBYTES (string), 1, 0); | |
facdc750 RS |
1292 | } |
1293 | ||
1294 | /* Copy and optionally translate the pattern. */ | |
1295 | len = raw_pattern_size; | |
1296 | len_byte = raw_pattern_size_byte; | |
f5f7578b | 1297 | patbuf = (unsigned char *) alloca (len * MAX_MULTIBYTE_LENGTH); |
facdc750 RS |
1298 | pat = patbuf; |
1299 | base_pat = raw_pattern; | |
1300 | if (multibyte) | |
1301 | { | |
a17ae96a KH |
1302 | /* Fill patbuf by translated characters in STRING while |
1303 | checking if we can use boyer-moore search. If TRT is | |
1304 | non-nil, we can use boyer-moore search only if TRT can be | |
1305 | represented by the byte array of 256 elements. For that, | |
1306 | all non-ASCII case-equivalents of all case-senstive | |
1307 | characters in STRING must belong to the same charset and | |
1308 | row. */ | |
1309 | ||
facdc750 RS |
1310 | while (--len >= 0) |
1311 | { | |
a17ae96a | 1312 | unsigned char str_base[MAX_MULTIBYTE_LENGTH], *str; |
aff2ce94 | 1313 | int c, translated, inverse; |
a17ae96a | 1314 | int in_charlen, charlen; |
facdc750 RS |
1315 | |
1316 | /* If we got here and the RE flag is set, it's because we're | |
1317 | dealing with a regexp known to be trivial, so the backslash | |
1318 | just quotes the next character. */ | |
1319 | if (RE && *base_pat == '\\') | |
1320 | { | |
1321 | len--; | |
62221f22 | 1322 | raw_pattern_size--; |
facdc750 RS |
1323 | len_byte--; |
1324 | base_pat++; | |
1325 | } | |
1326 | ||
62a6e103 | 1327 | c = STRING_CHAR_AND_LENGTH (base_pat, in_charlen); |
040272ce | 1328 | |
a17ae96a | 1329 | if (NILP (trt)) |
facdc750 | 1330 | { |
a17ae96a KH |
1331 | str = base_pat; |
1332 | charlen = in_charlen; | |
1333 | } | |
1334 | else | |
1335 | { | |
1336 | /* Translate the character. */ | |
1337 | TRANSLATE (translated, trt, c); | |
1338 | charlen = CHAR_STRING (translated, str_base); | |
1339 | str = str_base; | |
1340 | ||
1341 | /* Check if C has any other case-equivalents. */ | |
1342 | TRANSLATE (inverse, inverse_trt, c); | |
1343 | /* If so, check if we can use boyer-moore. */ | |
1344 | if (c != inverse && boyer_moore_ok) | |
1345 | { | |
1346 | /* Check if all equivalents belong to the same | |
1347 | group of characters. Note that the check of C | |
49e44e81 KH |
1348 | itself is done by the last iteration. */ |
1349 | int this_char_base = -1; | |
1350 | ||
1351 | while (boyer_moore_ok) | |
a17ae96a | 1352 | { |
49e44e81 | 1353 | if (ASCII_BYTE_P (inverse)) |
a17ae96a | 1354 | { |
49e44e81 KH |
1355 | if (this_char_base > 0) |
1356 | boyer_moore_ok = 0; | |
1357 | else | |
dafaabd1 | 1358 | this_char_base = 0; |
a17ae96a | 1359 | } |
49e44e81 KH |
1360 | else if (CHAR_BYTE8_P (inverse)) |
1361 | /* Boyer-moore search can't handle a | |
1362 | translation of an eight-bit | |
1363 | character. */ | |
1364 | boyer_moore_ok = 0; | |
1365 | else if (this_char_base < 0) | |
1366 | { | |
1367 | this_char_base = inverse & ~0x3F; | |
1368 | if (char_base < 0) | |
1369 | char_base = this_char_base; | |
dafaabd1 | 1370 | else if (this_char_base != char_base) |
49e44e81 KH |
1371 | boyer_moore_ok = 0; |
1372 | } | |
1373 | else if ((inverse & ~0x3F) != this_char_base) | |
1374 | boyer_moore_ok = 0; | |
a17ae96a KH |
1375 | if (c == inverse) |
1376 | break; | |
1377 | TRANSLATE (inverse, inverse_trt, inverse); | |
1378 | } | |
1379 | } | |
aff2ce94 | 1380 | } |
facdc750 RS |
1381 | |
1382 | /* Store this character into the translated pattern. */ | |
72af86bd | 1383 | memcpy (pat, str, charlen); |
a17ae96a | 1384 | pat += charlen; |
facdc750 RS |
1385 | base_pat += in_charlen; |
1386 | len_byte -= in_charlen; | |
1387 | } | |
dafaabd1 AS |
1388 | |
1389 | /* If char_base is still negative we didn't find any translated | |
1390 | non-ASCII characters. */ | |
1391 | if (char_base < 0) | |
1392 | char_base = 0; | |
facdc750 RS |
1393 | } |
1394 | else | |
1395 | { | |
040272ce | 1396 | /* Unibyte buffer. */ |
a17ae96a | 1397 | char_base = 0; |
facdc750 RS |
1398 | while (--len >= 0) |
1399 | { | |
040272ce | 1400 | int c, translated; |
facdc750 RS |
1401 | |
1402 | /* If we got here and the RE flag is set, it's because we're | |
1403 | dealing with a regexp known to be trivial, so the backslash | |
1404 | just quotes the next character. */ | |
1405 | if (RE && *base_pat == '\\') | |
1406 | { | |
1407 | len--; | |
0190922f | 1408 | raw_pattern_size--; |
facdc750 RS |
1409 | base_pat++; |
1410 | } | |
1411 | c = *base_pat++; | |
aff2ce94 | 1412 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1413 | *pat++ = translated; |
1414 | } | |
1415 | } | |
1416 | ||
1417 | len_byte = pat - patbuf; | |
facdc750 RS |
1418 | pat = base_pat = patbuf; |
1419 | ||
040272ce | 1420 | if (boyer_moore_ok) |
1db5b1ad JB |
1421 | return boyer_moore (n, pat, len_byte, trt, inverse_trt, |
1422 | pos_byte, lim_byte, | |
a17ae96a | 1423 | char_base); |
facdc750 | 1424 | else |
1db5b1ad | 1425 | return simple_search (n, pat, raw_pattern_size, len_byte, trt, |
facdc750 RS |
1426 | pos, pos_byte, lim, lim_byte); |
1427 | } | |
1428 | } | |
1429 | \f | |
1430 | /* Do a simple string search N times for the string PAT, | |
1431 | whose length is LEN/LEN_BYTE, | |
1432 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1433 | TRT is the translation table. | |
f8bd51c4 | 1434 | |
facdc750 RS |
1435 | Return the character position where the match is found. |
1436 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1437 | |
facdc750 RS |
1438 | This kind of search works regardless of what is in PAT and |
1439 | regardless of what is in TRT. It is used in cases where | |
1440 | boyer_moore cannot work. */ | |
1441 | ||
e7deaab0 | 1442 | static EMACS_INT |
c098fdb8 EZ |
1443 | simple_search (EMACS_INT n, unsigned char *pat, |
1444 | EMACS_INT len, EMACS_INT len_byte, Lisp_Object trt, | |
1445 | EMACS_INT pos, EMACS_INT pos_byte, | |
1446 | EMACS_INT lim, EMACS_INT lim_byte) | |
facdc750 | 1447 | { |
4b4deea2 | 1448 | int multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
ab228c24 | 1449 | int forward = n > 0; |
49e44e81 KH |
1450 | /* Number of buffer bytes matched. Note that this may be different |
1451 | from len_byte in a multibyte buffer. */ | |
c098fdb8 | 1452 | EMACS_INT match_byte; |
facdc750 RS |
1453 | |
1454 | if (lim > pos && multibyte) | |
1455 | while (n > 0) | |
1456 | { | |
1457 | while (1) | |
f8bd51c4 | 1458 | { |
facdc750 | 1459 | /* Try matching at position POS. */ |
e7deaab0 AS |
1460 | EMACS_INT this_pos = pos; |
1461 | EMACS_INT this_pos_byte = pos_byte; | |
c098fdb8 | 1462 | EMACS_INT this_len = len; |
facdc750 | 1463 | unsigned char *p = pat; |
a9469498 | 1464 | if (pos + len > lim || pos_byte + len_byte > lim_byte) |
facdc750 RS |
1465 | goto stop; |
1466 | ||
1467 | while (this_len > 0) | |
1468 | { | |
1469 | int charlen, buf_charlen; | |
ab228c24 | 1470 | int pat_ch, buf_ch; |
facdc750 | 1471 | |
62a6e103 | 1472 | pat_ch = STRING_CHAR_AND_LENGTH (p, charlen); |
facdc750 | 1473 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
facdc750 | 1474 | buf_charlen); |
aff2ce94 | 1475 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1476 | |
1477 | if (buf_ch != pat_ch) | |
1478 | break; | |
ab228c24 | 1479 | |
ab228c24 RS |
1480 | this_len--; |
1481 | p += charlen; | |
1482 | ||
1483 | this_pos_byte += buf_charlen; | |
1484 | this_pos++; | |
facdc750 RS |
1485 | } |
1486 | ||
1487 | if (this_len == 0) | |
1488 | { | |
49e44e81 | 1489 | match_byte = this_pos_byte - pos_byte; |
facdc750 | 1490 | pos += len; |
49e44e81 | 1491 | pos_byte += match_byte; |
facdc750 RS |
1492 | break; |
1493 | } | |
1494 | ||
1495 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1496 | } |
facdc750 RS |
1497 | |
1498 | n--; | |
1499 | } | |
1500 | else if (lim > pos) | |
1501 | while (n > 0) | |
1502 | { | |
1503 | while (1) | |
f8bd51c4 | 1504 | { |
facdc750 | 1505 | /* Try matching at position POS. */ |
e7deaab0 | 1506 | EMACS_INT this_pos = pos; |
c098fdb8 | 1507 | EMACS_INT this_len = len; |
facdc750 RS |
1508 | unsigned char *p = pat; |
1509 | ||
1510 | if (pos + len > lim) | |
1511 | goto stop; | |
1512 | ||
1513 | while (this_len > 0) | |
1514 | { | |
1515 | int pat_ch = *p++; | |
1516 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1517 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1518 | |
1519 | if (buf_ch != pat_ch) | |
1520 | break; | |
ab228c24 RS |
1521 | |
1522 | this_len--; | |
1523 | this_pos++; | |
facdc750 RS |
1524 | } |
1525 | ||
1526 | if (this_len == 0) | |
1527 | { | |
49e44e81 | 1528 | match_byte = len; |
facdc750 RS |
1529 | pos += len; |
1530 | break; | |
1531 | } | |
1532 | ||
1533 | pos++; | |
f8bd51c4 | 1534 | } |
facdc750 RS |
1535 | |
1536 | n--; | |
1537 | } | |
1538 | /* Backwards search. */ | |
1539 | else if (lim < pos && multibyte) | |
1540 | while (n < 0) | |
1541 | { | |
1542 | while (1) | |
f8bd51c4 | 1543 | { |
facdc750 | 1544 | /* Try matching at position POS. */ |
8b264ecb AS |
1545 | EMACS_INT this_pos = pos; |
1546 | EMACS_INT this_pos_byte = pos_byte; | |
c098fdb8 | 1547 | EMACS_INT this_len = len; |
c525b3f2 | 1548 | const unsigned char *p = pat + len_byte; |
facdc750 | 1549 | |
8b264ecb | 1550 | if (this_pos - len < lim || (pos_byte - len_byte) < lim_byte) |
facdc750 RS |
1551 | goto stop; |
1552 | ||
1553 | while (this_len > 0) | |
1554 | { | |
ab228c24 | 1555 | int pat_ch, buf_ch; |
facdc750 | 1556 | |
8b264ecb AS |
1557 | DEC_BOTH (this_pos, this_pos_byte); |
1558 | PREV_CHAR_BOUNDARY (p, pat); | |
1559 | pat_ch = STRING_CHAR (p); | |
1560 | buf_ch = STRING_CHAR (BYTE_POS_ADDR (this_pos_byte)); | |
aff2ce94 | 1561 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1562 | |
1563 | if (buf_ch != pat_ch) | |
1564 | break; | |
ab228c24 | 1565 | |
ab228c24 | 1566 | this_len--; |
facdc750 RS |
1567 | } |
1568 | ||
1569 | if (this_len == 0) | |
1570 | { | |
8b264ecb AS |
1571 | match_byte = pos_byte - this_pos_byte; |
1572 | pos = this_pos; | |
1573 | pos_byte = this_pos_byte; | |
facdc750 RS |
1574 | break; |
1575 | } | |
1576 | ||
1577 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1578 | } |
1579 | ||
facdc750 RS |
1580 | n++; |
1581 | } | |
1582 | else if (lim < pos) | |
1583 | while (n < 0) | |
1584 | { | |
1585 | while (1) | |
b6d6a51c | 1586 | { |
facdc750 | 1587 | /* Try matching at position POS. */ |
e7deaab0 | 1588 | EMACS_INT this_pos = pos - len; |
c098fdb8 | 1589 | EMACS_INT this_len = len; |
facdc750 RS |
1590 | unsigned char *p = pat; |
1591 | ||
a9469498 | 1592 | if (this_pos < lim) |
facdc750 RS |
1593 | goto stop; |
1594 | ||
1595 | while (this_len > 0) | |
1596 | { | |
1597 | int pat_ch = *p++; | |
1598 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1599 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1600 | |
1601 | if (buf_ch != pat_ch) | |
1602 | break; | |
ab228c24 RS |
1603 | this_len--; |
1604 | this_pos++; | |
facdc750 RS |
1605 | } |
1606 | ||
1607 | if (this_len == 0) | |
b6d6a51c | 1608 | { |
49e44e81 | 1609 | match_byte = len; |
facdc750 RS |
1610 | pos -= len; |
1611 | break; | |
b6d6a51c | 1612 | } |
facdc750 RS |
1613 | |
1614 | pos--; | |
b6d6a51c | 1615 | } |
facdc750 RS |
1616 | |
1617 | n++; | |
b6d6a51c | 1618 | } |
facdc750 RS |
1619 | |
1620 | stop: | |
1621 | if (n == 0) | |
aff2ce94 | 1622 | { |
ab228c24 | 1623 | if (forward) |
49e44e81 | 1624 | set_search_regs ((multibyte ? pos_byte : pos) - match_byte, match_byte); |
ab228c24 | 1625 | else |
49e44e81 | 1626 | set_search_regs (multibyte ? pos_byte : pos, match_byte); |
aff2ce94 RS |
1627 | |
1628 | return pos; | |
1629 | } | |
facdc750 RS |
1630 | else if (n > 0) |
1631 | return -n; | |
1632 | else | |
1633 | return n; | |
1634 | } | |
1635 | \f | |
0190922f | 1636 | /* Do Boyer-Moore search N times for the string BASE_PAT, |
1db5b1ad JB |
1637 | whose length is LEN_BYTE, |
1638 | from buffer position POS_BYTE until LIM_BYTE. | |
facdc750 RS |
1639 | DIRECTION says which direction we search in. |
1640 | TRT and INVERSE_TRT are translation tables. | |
0190922f | 1641 | Characters in PAT are already translated by TRT. |
facdc750 | 1642 | |
0190922f KH |
1643 | This kind of search works if all the characters in BASE_PAT that |
1644 | have nontrivial translation are the same aside from the last byte. | |
1645 | This makes it possible to translate just the last byte of a | |
1646 | character, and do so after just a simple test of the context. | |
b2e6b10f | 1647 | CHAR_BASE is nonzero if there is such a non-ASCII character. |
facdc750 RS |
1648 | |
1649 | If that criterion is not satisfied, do not call this function. */ | |
1650 | ||
e7deaab0 | 1651 | static EMACS_INT |
c098fdb8 | 1652 | boyer_moore (EMACS_INT n, unsigned char *base_pat, |
1db5b1ad | 1653 | EMACS_INT len_byte, |
d5a3eaaf | 1654 | Lisp_Object trt, Lisp_Object inverse_trt, |
1db5b1ad JB |
1655 | EMACS_INT pos_byte, EMACS_INT lim_byte, |
1656 | int char_base) | |
facdc750 RS |
1657 | { |
1658 | int direction = ((n > 0) ? 1 : -1); | |
c098fdb8 | 1659 | register EMACS_INT dirlen; |
e7deaab0 AS |
1660 | EMACS_INT limit; |
1661 | int stride_for_teases = 0; | |
f4646fff | 1662 | int BM_tab[0400]; |
177c0ea7 | 1663 | register unsigned char *cursor, *p_limit; |
c098fdb8 EZ |
1664 | register EMACS_INT i; |
1665 | register int j; | |
cb6792d2 | 1666 | unsigned char *pat, *pat_end; |
4b4deea2 | 1667 | int multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
facdc750 RS |
1668 | |
1669 | unsigned char simple_translate[0400]; | |
0190922f | 1670 | /* These are set to the preceding bytes of a byte to be translated |
49e44e81 | 1671 | if char_base is nonzero. As the maximum byte length of a |
a17ae96a | 1672 | multibyte character is 5, we have to check at most four previous |
0190922f KH |
1673 | bytes. */ |
1674 | int translate_prev_byte1 = 0; | |
1675 | int translate_prev_byte2 = 0; | |
1676 | int translate_prev_byte3 = 0; | |
facdc750 | 1677 | |
f4646fff AS |
1678 | /* The general approach is that we are going to maintain that we know |
1679 | the first (closest to the present position, in whatever direction | |
1680 | we're searching) character that could possibly be the last | |
1681 | (furthest from present position) character of a valid match. We | |
1682 | advance the state of our knowledge by looking at that character | |
1683 | and seeing whether it indeed matches the last character of the | |
1684 | pattern. If it does, we take a closer look. If it does not, we | |
1685 | move our pointer (to putative last characters) as far as is | |
1686 | logically possible. This amount of movement, which I call a | |
1687 | stride, will be the length of the pattern if the actual character | |
1688 | appears nowhere in the pattern, otherwise it will be the distance | |
1689 | from the last occurrence of that character to the end of the | |
1690 | pattern. If the amount is zero we have a possible match. */ | |
1691 | ||
1692 | /* Here we make a "mickey mouse" BM table. The stride of the search | |
1693 | is determined only by the last character of the putative match. | |
1694 | If that character does not match, we will stride the proper | |
1695 | distance to propose a match that superimposes it on the last | |
1696 | instance of a character that matches it (per trt), or misses | |
1697 | it entirely if there is none. */ | |
facdc750 RS |
1698 | |
1699 | dirlen = len_byte * direction; | |
cb6792d2 RS |
1700 | |
1701 | /* Record position after the end of the pattern. */ | |
1702 | pat_end = base_pat + len_byte; | |
1703 | /* BASE_PAT points to a character that we start scanning from. | |
1704 | It is the first character in a forward search, | |
1705 | the last character in a backward search. */ | |
facdc750 | 1706 | if (direction < 0) |
cb6792d2 RS |
1707 | base_pat = pat_end - 1; |
1708 | ||
f4646fff AS |
1709 | /* A character that does not appear in the pattern induces a |
1710 | stride equal to the pattern length. */ | |
1711 | for (i = 0; i < 0400; i++) | |
1712 | BM_tab[i] = dirlen; | |
facdc750 RS |
1713 | |
1714 | /* We use this for translation, instead of TRT itself. | |
1715 | We fill this in to handle the characters that actually | |
1716 | occur in the pattern. Others don't matter anyway! */ | |
facdc750 RS |
1717 | for (i = 0; i < 0400; i++) |
1718 | simple_translate[i] = i; | |
1719 | ||
a17ae96a | 1720 | if (char_base) |
0190922f | 1721 | { |
a17ae96a | 1722 | /* Setup translate_prev_byte1/2/3/4 from CHAR_BASE. Only a |
0190922f | 1723 | byte following them are the target of translation. */ |
0190922f | 1724 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
1f3561e4 | 1725 | int cblen = CHAR_STRING (char_base, str); |
0190922f | 1726 | |
1f3561e4 PE |
1727 | translate_prev_byte1 = str[cblen - 2]; |
1728 | if (cblen > 2) | |
0190922f | 1729 | { |
1f3561e4 PE |
1730 | translate_prev_byte2 = str[cblen - 3]; |
1731 | if (cblen > 3) | |
1f1d9321 | 1732 | translate_prev_byte3 = str[cblen - 4]; |
0190922f KH |
1733 | } |
1734 | } | |
1735 | ||
facdc750 | 1736 | i = 0; |
f4646fff | 1737 | while (i != dirlen) |
facdc750 | 1738 | { |
cb6792d2 | 1739 | unsigned char *ptr = base_pat + i; |
facdc750 | 1740 | i += direction; |
facdc750 | 1741 | if (! NILP (trt)) |
ca1d1d23 | 1742 | { |
49e44e81 KH |
1743 | /* If the byte currently looking at is the last of a |
1744 | character to check case-equivalents, set CH to that | |
1745 | character. An ASCII character and a non-ASCII character | |
1746 | matching with CHAR_BASE are to be checked. */ | |
a17ae96a KH |
1747 | int ch = -1; |
1748 | ||
1749 | if (ASCII_BYTE_P (*ptr) || ! multibyte) | |
1750 | ch = *ptr; | |
49e44e81 | 1751 | else if (char_base |
86dc6ccb | 1752 | && ((pat_end - ptr) == 1 || CHAR_HEAD_P (ptr[1]))) |
ca1d1d23 | 1753 | { |
49e44e81 KH |
1754 | unsigned char *charstart = ptr - 1; |
1755 | ||
1756 | while (! (CHAR_HEAD_P (*charstart))) | |
1757 | charstart--; | |
62a6e103 | 1758 | ch = STRING_CHAR (charstart); |
a17ae96a KH |
1759 | if (char_base != (ch & ~0x3F)) |
1760 | ch = -1; | |
ca1d1d23 | 1761 | } |
facdc750 | 1762 | |
f51995fa | 1763 | if (ch >= 0200) |
49e44e81 KH |
1764 | j = (ch & 0x3F) | 0200; |
1765 | else | |
1766 | j = *ptr; | |
1767 | ||
f4646fff | 1768 | if (i == dirlen) |
facdc750 | 1769 | stride_for_teases = BM_tab[j]; |
ab228c24 | 1770 | |
facdc750 | 1771 | BM_tab[j] = dirlen - i; |
1db5b1ad JB |
1772 | /* A translation table is accompanied by its inverse -- see |
1773 | comment following downcase_table for details. */ | |
a17ae96a | 1774 | if (ch >= 0) |
ab228c24 RS |
1775 | { |
1776 | int starting_ch = ch; | |
49e44e81 | 1777 | int starting_j = j; |
a17ae96a | 1778 | |
ab228c24 RS |
1779 | while (1) |
1780 | { | |
1781 | TRANSLATE (ch, inverse_trt, ch); | |
f51995fa | 1782 | if (ch >= 0200) |
49e44e81 | 1783 | j = (ch & 0x3F) | 0200; |
ab228c24 | 1784 | else |
a17ae96a | 1785 | j = ch; |
ab228c24 RS |
1786 | |
1787 | /* For all the characters that map into CH, | |
1788 | set up simple_translate to map the last byte | |
1789 | into STARTING_J. */ | |
1790 | simple_translate[j] = starting_j; | |
1791 | if (ch == starting_ch) | |
1792 | break; | |
1793 | BM_tab[j] = dirlen - i; | |
1794 | } | |
1795 | } | |
facdc750 RS |
1796 | } |
1797 | else | |
1798 | { | |
1799 | j = *ptr; | |
1800 | ||
f4646fff | 1801 | if (i == dirlen) |
facdc750 RS |
1802 | stride_for_teases = BM_tab[j]; |
1803 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1804 | } |
f4646fff AS |
1805 | /* stride_for_teases tells how much to stride if we get a |
1806 | match on the far character but are subsequently | |
1807 | disappointed, by recording what the stride would have been | |
1808 | for that character if the last character had been | |
1809 | different. */ | |
facdc750 | 1810 | } |
facdc750 RS |
1811 | pos_byte += dirlen - ((direction > 0) ? direction : 0); |
1812 | /* loop invariant - POS_BYTE points at where last char (first | |
1813 | char if reverse) of pattern would align in a possible match. */ | |
1814 | while (n != 0) | |
1815 | { | |
e7deaab0 | 1816 | EMACS_INT tail_end; |
facdc750 RS |
1817 | unsigned char *tail_end_ptr; |
1818 | ||
1819 | /* It's been reported that some (broken) compiler thinks that | |
1820 | Boolean expressions in an arithmetic context are unsigned. | |
1821 | Using an explicit ?1:0 prevents this. */ | |
1822 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1823 | < 0) | |
1824 | return (n * (0 - direction)); | |
1825 | /* First we do the part we can by pointers (maybe nothing) */ | |
1826 | QUIT; | |
1827 | pat = base_pat; | |
1828 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1829 | if (direction > 0) |
1830 | { | |
1831 | limit = BUFFER_CEILING_OF (limit); | |
1832 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1833 | can take on without hitting edge of buffer or the gap. */ | |
1834 | limit = min (limit, pos_byte + 20000); | |
1835 | limit = min (limit, lim_byte - 1); | |
1836 | } | |
1837 | else | |
1838 | { | |
1839 | limit = BUFFER_FLOOR_OF (limit); | |
1840 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1841 | can take on without hitting edge of buffer or the gap. */ | |
1842 | limit = max (limit, pos_byte - 20000); | |
1843 | limit = max (limit, lim_byte); | |
1844 | } | |
facdc750 RS |
1845 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1846 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1847 | ||
1848 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1849 | { |
facdc750 RS |
1850 | unsigned char *p2; |
1851 | ||
1852 | p_limit = BYTE_POS_ADDR (limit); | |
1853 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
f4646fff | 1854 | /* In this loop, pos + cursor - p2 is the surrogate for pos. */ |
facdc750 | 1855 | while (1) /* use one cursor setting as long as i can */ |
ca1d1d23 | 1856 | { |
facdc750 | 1857 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1858 | { |
f4646fff AS |
1859 | while (cursor <= p_limit) |
1860 | { | |
1861 | if (BM_tab[*cursor] == 0) | |
1862 | goto hit; | |
facdc750 | 1863 | cursor += BM_tab[*cursor]; |
f4646fff | 1864 | } |
facdc750 RS |
1865 | } |
1866 | else | |
1867 | { | |
f4646fff AS |
1868 | while (cursor >= p_limit) |
1869 | { | |
1870 | if (BM_tab[*cursor] == 0) | |
1871 | goto hit; | |
facdc750 | 1872 | cursor += BM_tab[*cursor]; |
f4646fff | 1873 | } |
facdc750 | 1874 | } |
f4646fff AS |
1875 | /* If you are here, cursor is beyond the end of the |
1876 | searched region. You fail to match within the | |
1877 | permitted region and would otherwise try a character | |
1878 | beyond that region. */ | |
1879 | break; | |
1880 | ||
1881 | hit: | |
facdc750 RS |
1882 | i = dirlen - direction; |
1883 | if (! NILP (trt)) | |
1884 | { | |
1885 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1886 | { |
facdc750 RS |
1887 | int ch; |
1888 | cursor -= direction; | |
1889 | /* Translate only the last byte of a character. */ | |
1890 | if (! multibyte | |
1891 | || ((cursor == tail_end_ptr | |
1892 | || CHAR_HEAD_P (cursor[1])) | |
1893 | && (CHAR_HEAD_P (cursor[0]) | |
0190922f KH |
1894 | /* Check if this is the last byte of |
1895 | a translable character. */ | |
1896 | || (translate_prev_byte1 == cursor[-1] | |
1897 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1898 | || (translate_prev_byte2 == cursor[-2] | |
1899 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1900 | || (translate_prev_byte3 == cursor[-3])))))))) | |
facdc750 RS |
1901 | ch = simple_translate[*cursor]; |
1902 | else | |
1903 | ch = *cursor; | |
1904 | if (pat[i] != ch) | |
1905 | break; | |
ca1d1d23 | 1906 | } |
facdc750 RS |
1907 | } |
1908 | else | |
1909 | { | |
1910 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1911 | { |
facdc750 RS |
1912 | cursor -= direction; |
1913 | if (pat[i] != *cursor) | |
1914 | break; | |
ca1d1d23 | 1915 | } |
facdc750 RS |
1916 | } |
1917 | cursor += dirlen - i - direction; /* fix cursor */ | |
1918 | if (i + direction == 0) | |
1919 | { | |
e7deaab0 | 1920 | EMACS_INT position, start, end; |
0c8533c6 | 1921 | |
facdc750 | 1922 | cursor -= direction; |
1113d9db | 1923 | |
facdc750 RS |
1924 | position = pos_byte + cursor - p2 + ((direction > 0) |
1925 | ? 1 - len_byte : 0); | |
1926 | set_search_regs (position, len_byte); | |
ca325161 | 1927 | |
ef887810 RS |
1928 | if (NILP (Vinhibit_changing_match_data)) |
1929 | { | |
1930 | start = search_regs.start[0]; | |
1931 | end = search_regs.end[0]; | |
1932 | } | |
1933 | else | |
1934 | /* If Vinhibit_changing_match_data is non-nil, | |
1935 | search_regs will not be changed. So let's | |
1936 | compute start and end here. */ | |
1937 | { | |
1938 | start = BYTE_TO_CHAR (position); | |
1939 | end = BYTE_TO_CHAR (position + len_byte); | |
1940 | } | |
1941 | ||
facdc750 RS |
1942 | if ((n -= direction) != 0) |
1943 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1944 | else |
ef887810 | 1945 | return direction > 0 ? end : start; |
ca1d1d23 | 1946 | } |
facdc750 RS |
1947 | else |
1948 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1949 | } |
facdc750 RS |
1950 | pos_byte += cursor - p2; |
1951 | } | |
1952 | else | |
f4646fff AS |
1953 | /* Now we'll pick up a clump that has to be done the hard |
1954 | way because it covers a discontinuity. */ | |
facdc750 RS |
1955 | { |
1956 | limit = ((direction > 0) | |
1957 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1958 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1959 | limit = ((direction > 0) | |
1960 | ? min (limit + len_byte, lim_byte - 1) | |
1961 | : max (limit - len_byte, lim_byte)); | |
1962 | /* LIMIT is now the last value POS_BYTE can have | |
1963 | and still be valid for a possible match. */ | |
1964 | while (1) | |
ca1d1d23 | 1965 | { |
f4646fff AS |
1966 | /* This loop can be coded for space rather than |
1967 | speed because it will usually run only once. | |
1968 | (the reach is at most len + 21, and typically | |
1969 | does not exceed len). */ | |
facdc750 | 1970 | while ((limit - pos_byte) * direction >= 0) |
f4646fff AS |
1971 | { |
1972 | int ch = FETCH_BYTE (pos_byte); | |
1973 | if (BM_tab[ch] == 0) | |
1974 | goto hit2; | |
1975 | pos_byte += BM_tab[ch]; | |
1976 | } | |
1977 | break; /* ran off the end */ | |
1978 | ||
1979 | hit2: | |
1980 | /* Found what might be a match. */ | |
facdc750 RS |
1981 | i = dirlen - direction; |
1982 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1983 | { |
facdc750 RS |
1984 | int ch; |
1985 | unsigned char *ptr; | |
1986 | pos_byte -= direction; | |
1987 | ptr = BYTE_POS_ADDR (pos_byte); | |
1988 | /* Translate only the last byte of a character. */ | |
1989 | if (! multibyte | |
1990 | || ((ptr == tail_end_ptr | |
1991 | || CHAR_HEAD_P (ptr[1])) | |
1992 | && (CHAR_HEAD_P (ptr[0]) | |
0190922f KH |
1993 | /* Check if this is the last byte of a |
1994 | translable character. */ | |
1995 | || (translate_prev_byte1 == ptr[-1] | |
1996 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1997 | || (translate_prev_byte2 == ptr[-2] | |
1998 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1999 | || translate_prev_byte3 == ptr[-3]))))))) | |
facdc750 RS |
2000 | ch = simple_translate[*ptr]; |
2001 | else | |
2002 | ch = *ptr; | |
2003 | if (pat[i] != ch) | |
2004 | break; | |
2005 | } | |
2006 | /* Above loop has moved POS_BYTE part or all the way | |
2007 | back to the first pos (last pos if reverse). | |
2008 | Set it once again at the last (first if reverse) char. */ | |
f4646fff | 2009 | pos_byte += dirlen - i - direction; |
facdc750 RS |
2010 | if (i + direction == 0) |
2011 | { | |
e7deaab0 | 2012 | EMACS_INT position, start, end; |
facdc750 | 2013 | pos_byte -= direction; |
1113d9db | 2014 | |
facdc750 | 2015 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
facdc750 | 2016 | set_search_regs (position, len_byte); |
ca325161 | 2017 | |
ef887810 RS |
2018 | if (NILP (Vinhibit_changing_match_data)) |
2019 | { | |
2020 | start = search_regs.start[0]; | |
2021 | end = search_regs.end[0]; | |
2022 | } | |
2023 | else | |
2024 | /* If Vinhibit_changing_match_data is non-nil, | |
2025 | search_regs will not be changed. So let's | |
2026 | compute start and end here. */ | |
2027 | { | |
2028 | start = BYTE_TO_CHAR (position); | |
2029 | end = BYTE_TO_CHAR (position + len_byte); | |
2030 | } | |
2031 | ||
facdc750 RS |
2032 | if ((n -= direction) != 0) |
2033 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 2034 | else |
ef887810 | 2035 | return direction > 0 ? end : start; |
ca1d1d23 | 2036 | } |
facdc750 RS |
2037 | else |
2038 | pos_byte += stride_for_teases; | |
2039 | } | |
2040 | } | |
2041 | /* We have done one clump. Can we continue? */ | |
2042 | if ((lim_byte - pos_byte) * direction < 0) | |
2043 | return ((0 - n) * direction); | |
ca1d1d23 | 2044 | } |
facdc750 | 2045 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 2046 | } |
ca325161 | 2047 | |
fa8ed3e0 | 2048 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
2049 | for the overall match just found in the current buffer. |
2050 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
2051 | |
2052 | static void | |
971de7fb | 2053 | set_search_regs (EMACS_INT beg_byte, EMACS_INT nbytes) |
ca325161 | 2054 | { |
a7e4cdde RS |
2055 | int i; |
2056 | ||
ef887810 RS |
2057 | if (!NILP (Vinhibit_changing_match_data)) |
2058 | return; | |
2059 | ||
ca325161 RS |
2060 | /* Make sure we have registers in which to store |
2061 | the match position. */ | |
2062 | if (search_regs.num_regs == 0) | |
2063 | { | |
2d4a771a RS |
2064 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
2065 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 2066 | search_regs.num_regs = 2; |
ca325161 RS |
2067 | } |
2068 | ||
a7e4cdde RS |
2069 | /* Clear out the other registers. */ |
2070 | for (i = 1; i < search_regs.num_regs; i++) | |
2071 | { | |
2072 | search_regs.start[i] = -1; | |
2073 | search_regs.end[i] = -1; | |
2074 | } | |
2075 | ||
fa8ed3e0 RS |
2076 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
2077 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 2078 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 2079 | } |
ca1d1d23 | 2080 | \f |
c6e0e1cc CY |
2081 | /* Given STRING, a string of words separated by word delimiters, |
2082 | compute a regexp that matches those exact words separated by | |
2083 | arbitrary punctuation. If LAX is nonzero, the end of the string | |
2084 | need not match a word boundary unless it ends in whitespace. */ | |
ca1d1d23 JB |
2085 | |
2086 | static Lisp_Object | |
971de7fb | 2087 | wordify (Lisp_Object string, int lax) |
ca1d1d23 | 2088 | { |
1f1d9321 | 2089 | register unsigned char *o; |
c098fdb8 | 2090 | register EMACS_INT i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 2091 | Lisp_Object val; |
0c8533c6 | 2092 | int prev_c = 0; |
c098fdb8 EZ |
2093 | EMACS_INT adjust; |
2094 | int whitespace_at_end; | |
ca1d1d23 | 2095 | |
b7826503 | 2096 | CHECK_STRING (string); |
d5db4077 | 2097 | len = SCHARS (string); |
ca1d1d23 | 2098 | |
0c8533c6 RS |
2099 | for (i = 0, i_byte = 0; i < len; ) |
2100 | { | |
2101 | int c; | |
177c0ea7 | 2102 | |
93daa011 | 2103 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
0c8533c6 RS |
2104 | |
2105 | if (SYNTAX (c) != Sword) | |
2106 | { | |
2107 | punct_count++; | |
6f076cc7 | 2108 | if (SYNTAX (prev_c) == Sword) |
0c8533c6 RS |
2109 | word_count++; |
2110 | } | |
ca1d1d23 | 2111 | |
0c8533c6 RS |
2112 | prev_c = c; |
2113 | } | |
2114 | ||
2115 | if (SYNTAX (prev_c) == Sword) | |
c6e0e1cc CY |
2116 | { |
2117 | word_count++; | |
2118 | whitespace_at_end = 0; | |
2119 | } | |
2120 | else | |
6f076cc7 PE |
2121 | { |
2122 | whitespace_at_end = 1; | |
2123 | if (!word_count) | |
2124 | return empty_unibyte_string; | |
2125 | } | |
0c8533c6 | 2126 | |
c6e0e1cc CY |
2127 | adjust = - punct_count + 5 * (word_count - 1) |
2128 | + ((lax && !whitespace_at_end) ? 2 : 4); | |
8a2df937 RS |
2129 | if (STRING_MULTIBYTE (string)) |
2130 | val = make_uninit_multibyte_string (len + adjust, | |
d5db4077 | 2131 | SBYTES (string) |
8a2df937 RS |
2132 | + adjust); |
2133 | else | |
2134 | val = make_uninit_string (len + adjust); | |
ca1d1d23 | 2135 | |
d5db4077 | 2136 | o = SDATA (val); |
ca1d1d23 JB |
2137 | *o++ = '\\'; |
2138 | *o++ = 'b'; | |
1e9582d4 | 2139 | prev_c = 0; |
ca1d1d23 | 2140 | |
1e9582d4 RS |
2141 | for (i = 0, i_byte = 0; i < len; ) |
2142 | { | |
2143 | int c; | |
c098fdb8 | 2144 | EMACS_INT i_byte_orig = i_byte; |
177c0ea7 | 2145 | |
93daa011 | 2146 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
1e9582d4 RS |
2147 | |
2148 | if (SYNTAX (c) == Sword) | |
2149 | { | |
72af86bd | 2150 | memcpy (o, SDATA (string) + i_byte_orig, i_byte - i_byte_orig); |
1e9582d4 RS |
2151 | o += i_byte - i_byte_orig; |
2152 | } | |
6f076cc7 | 2153 | else if (SYNTAX (prev_c) == Sword && --word_count) |
1e9582d4 RS |
2154 | { |
2155 | *o++ = '\\'; | |
2156 | *o++ = 'W'; | |
2157 | *o++ = '\\'; | |
2158 | *o++ = 'W'; | |
2159 | *o++ = '*'; | |
2160 | } | |
2161 | ||
2162 | prev_c = c; | |
2163 | } | |
ca1d1d23 | 2164 | |
c6e0e1cc CY |
2165 | if (!lax || whitespace_at_end) |
2166 | { | |
2167 | *o++ = '\\'; | |
2168 | *o++ = 'b'; | |
2169 | } | |
ca1d1d23 JB |
2170 | |
2171 | return val; | |
2172 | } | |
2173 | \f | |
2174 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
8c1a1077 PJ |
2175 | "MSearch backward: ", |
2176 | doc: /* Search backward from point for STRING. | |
2177 | Set point to the beginning of the occurrence found, and return point. | |
2178 | An optional second argument bounds the search; it is a buffer position. | |
2179 | The match found must not extend before that position. | |
2180 | Optional third argument, if t, means if fail just return nil (no error). | |
2181 | If not nil and not t, position at limit of search and return nil. | |
2182 | Optional fourth argument is repeat count--search for successive occurrences. | |
2183 | ||
2184 | Search case-sensitivity is determined by the value of the variable | |
2185 | `case-fold-search', which see. | |
2186 | ||
2187 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2188 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2189 | { |
b819a390 | 2190 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2191 | } |
2192 | ||
6af43974 | 2193 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2194 | doc: /* Search forward from point for STRING. |
2195 | Set point to the end of the occurrence found, and return point. | |
2196 | An optional second argument bounds the search; it is a buffer position. | |
d6aacdcd | 2197 | The match found must not extend after that position. A value of nil is |
48740f01 | 2198 | equivalent to (point-max). |
8c1a1077 PJ |
2199 | Optional third argument, if t, means if fail just return nil (no error). |
2200 | If not nil and not t, move to limit of search and return nil. | |
2201 | Optional fourth argument is repeat count--search for successive occurrences. | |
2202 | ||
2203 | Search case-sensitivity is determined by the value of the variable | |
2204 | `case-fold-search', which see. | |
2205 | ||
2206 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2207 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2208 | { |
b819a390 | 2209 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2210 | } |
2211 | ||
2212 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
8c1a1077 PJ |
2213 | "sWord search backward: ", |
2214 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2215 | Set point to the beginning of the occurrence found, and return point. | |
2216 | An optional second argument bounds the search; it is a buffer position. | |
2217 | The match found must not extend before that position. | |
2218 | Optional third argument, if t, means if fail just return nil (no error). | |
2219 | If not nil and not t, move to limit of search and return nil. | |
2220 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2221 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2222 | { |
c6e0e1cc | 2223 | return search_command (wordify (string, 0), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2224 | } |
2225 | ||
2226 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
8c1a1077 PJ |
2227 | "sWord search: ", |
2228 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2229 | Set point to the end of the occurrence found, and return point. | |
2230 | An optional second argument bounds the search; it is a buffer position. | |
2231 | The match found must not extend after that position. | |
2232 | Optional third argument, if t, means if fail just return nil (no error). | |
2233 | If not nil and not t, move to limit of search and return nil. | |
2234 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2235 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2236 | { |
c6e0e1cc CY |
2237 | return search_command (wordify (string, 0), bound, noerror, count, 1, 1, 0); |
2238 | } | |
2239 | ||
2240 | DEFUN ("word-search-backward-lax", Fword_search_backward_lax, Sword_search_backward_lax, 1, 4, | |
2241 | "sWord search backward: ", | |
2242 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2243 | Set point to the beginning of the occurrence found, and return point. | |
2244 | ||
2245 | Unlike `word-search-backward', the end of STRING need not match a word | |
2246 | boundary unless it ends in whitespace. | |
2247 | ||
2248 | An optional second argument bounds the search; it is a buffer position. | |
2249 | The match found must not extend before that position. | |
2250 | Optional third argument, if t, means if fail just return nil (no error). | |
2251 | If not nil and not t, move to limit of search and return nil. | |
2252 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2253 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
c6e0e1cc CY |
2254 | { |
2255 | return search_command (wordify (string, 1), bound, noerror, count, -1, 1, 0); | |
2256 | } | |
2257 | ||
2258 | DEFUN ("word-search-forward-lax", Fword_search_forward_lax, Sword_search_forward_lax, 1, 4, | |
2259 | "sWord search: ", | |
2260 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2261 | Set point to the end of the occurrence found, and return point. | |
2262 | ||
2263 | Unlike `word-search-forward', the end of STRING need not match a word | |
2264 | boundary unless it ends in whitespace. | |
2265 | ||
2266 | An optional second argument bounds the search; it is a buffer position. | |
2267 | The match found must not extend after that position. | |
2268 | Optional third argument, if t, means if fail just return nil (no error). | |
2269 | If not nil and not t, move to limit of search and return nil. | |
2270 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2271 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
c6e0e1cc CY |
2272 | { |
2273 | return search_command (wordify (string, 1), bound, noerror, count, 1, 1, 0); | |
ca1d1d23 JB |
2274 | } |
2275 | ||
2276 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
8c1a1077 PJ |
2277 | "sRE search backward: ", |
2278 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2279 | Set point to the beginning of the match, and return point. | |
2280 | The match found is the one starting last in the buffer | |
2281 | and yet ending before the origin of the search. | |
2282 | An optional second argument bounds the search; it is a buffer position. | |
2283 | The match found must start at or after that position. | |
2284 | Optional third argument, if t, means if fail just return nil (no error). | |
2285 | If not nil and not t, move to limit of search and return nil. | |
2286 | Optional fourth argument is repeat count--search for successive occurrences. | |
2287 | See also the functions `match-beginning', `match-end', `match-string', | |
2288 | and `replace-match'. */) | |
5842a27b | 2289 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2290 | { |
b819a390 | 2291 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2292 | } |
2293 | ||
2294 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2295 | "sRE search: ", |
2296 | doc: /* Search forward from point for regular expression REGEXP. | |
2297 | Set point to the end of the occurrence found, and return point. | |
2298 | An optional second argument bounds the search; it is a buffer position. | |
2299 | The match found must not extend after that position. | |
2300 | Optional third argument, if t, means if fail just return nil (no error). | |
2301 | If not nil and not t, move to limit of search and return nil. | |
2302 | Optional fourth argument is repeat count--search for successive occurrences. | |
2303 | See also the functions `match-beginning', `match-end', `match-string', | |
2304 | and `replace-match'. */) | |
5842a27b | 2305 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2306 | { |
b819a390 RS |
2307 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2308 | } | |
2309 | ||
2310 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2311 | "sPosix search backward: ", |
2312 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2313 | Find the longest match in accord with Posix regular expression rules. | |
2314 | Set point to the beginning of the match, and return point. | |
2315 | The match found is the one starting last in the buffer | |
2316 | and yet ending before the origin of the search. | |
2317 | An optional second argument bounds the search; it is a buffer position. | |
2318 | The match found must start at or after that position. | |
2319 | Optional third argument, if t, means if fail just return nil (no error). | |
2320 | If not nil and not t, move to limit of search and return nil. | |
2321 | Optional fourth argument is repeat count--search for successive occurrences. | |
2322 | See also the functions `match-beginning', `match-end', `match-string', | |
2323 | and `replace-match'. */) | |
5842a27b | 2324 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2325 | { |
2326 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2327 | } | |
2328 | ||
2329 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2330 | "sPosix search: ", |
2331 | doc: /* Search forward from point for regular expression REGEXP. | |
2332 | Find the longest match in accord with Posix regular expression rules. | |
2333 | Set point to the end of the occurrence found, and return point. | |
2334 | An optional second argument bounds the search; it is a buffer position. | |
2335 | The match found must not extend after that position. | |
2336 | Optional third argument, if t, means if fail just return nil (no error). | |
2337 | If not nil and not t, move to limit of search and return nil. | |
2338 | Optional fourth argument is repeat count--search for successive occurrences. | |
2339 | See also the functions `match-beginning', `match-end', `match-string', | |
2340 | and `replace-match'. */) | |
5842a27b | 2341 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2342 | { |
2343 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2344 | } |
2345 | \f | |
d7a5ad5f | 2346 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2347 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2348 | Leave point at the end of the replacement text. |
2349 | ||
8c1a1077 PJ |
2350 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. |
2351 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2352 | based on the replaced text. | |
2353 | If the replaced text has only capital letters | |
2354 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
4dd0c271 RS |
2355 | Otherwise if all words are capitalized in the replaced text, |
2356 | capitalize each word in NEWTEXT. | |
8c1a1077 PJ |
2357 | |
2358 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2359 | Otherwise treat `\\' as special: | |
2360 | `\\&' in NEWTEXT means substitute original matched text. | |
2361 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2362 | If Nth parens didn't match, substitute nothing. | |
2363 | `\\\\' means insert one `\\'. | |
4dd0c271 RS |
2364 | Case conversion does not apply to these substitutions. |
2365 | ||
8c1a1077 | 2366 | FIXEDCASE and LITERAL are optional arguments. |
8c1a1077 PJ |
2367 | |
2368 | The optional fourth argument STRING can be a string to modify. | |
2369 | This is meaningful when the previous match was done against STRING, | |
2370 | using `string-match'. When used this way, `replace-match' | |
2371 | creates and returns a new string made by copying STRING and replacing | |
2372 | the part of STRING that was matched. | |
2373 | ||
2374 | The optional fifth argument SUBEXP specifies a subexpression; | |
2375 | it says to replace just that subexpression with NEWTEXT, | |
2376 | rather than replacing the entire matched text. | |
2377 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2378 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2379 | NEWTEXT in place of subexp N. | |
2380 | This is useful only after a regular expression search or match, | |
2381 | since only regular expressions have distinguished subexpressions. */) | |
5842a27b | 2382 | (Lisp_Object newtext, Lisp_Object fixedcase, Lisp_Object literal, Lisp_Object string, Lisp_Object subexp) |
ca1d1d23 JB |
2383 | { |
2384 | enum { nochange, all_caps, cap_initial } case_action; | |
c098fdb8 | 2385 | register EMACS_INT pos, pos_byte; |
ca1d1d23 | 2386 | int some_multiletter_word; |
97832bd0 | 2387 | int some_lowercase; |
73dc8771 | 2388 | int some_uppercase; |
208767c3 | 2389 | int some_nonuppercase_initial; |
ca1d1d23 | 2390 | register int c, prevc; |
d7a5ad5f | 2391 | int sub; |
e7deaab0 | 2392 | EMACS_INT opoint, newpoint; |
ca1d1d23 | 2393 | |
b7826503 | 2394 | CHECK_STRING (newtext); |
ca1d1d23 | 2395 | |
080c45fd | 2396 | if (! NILP (string)) |
b7826503 | 2397 | CHECK_STRING (string); |
080c45fd | 2398 | |
ca1d1d23 JB |
2399 | case_action = nochange; /* We tried an initialization */ |
2400 | /* but some C compilers blew it */ | |
4746118a JB |
2401 | |
2402 | if (search_regs.num_regs <= 0) | |
d72cdbfc | 2403 | error ("`replace-match' called before any match found"); |
4746118a | 2404 | |
d7a5ad5f RS |
2405 | if (NILP (subexp)) |
2406 | sub = 0; | |
2407 | else | |
2408 | { | |
b7826503 | 2409 | CHECK_NUMBER (subexp); |
d7a5ad5f RS |
2410 | sub = XINT (subexp); |
2411 | if (sub < 0 || sub >= search_regs.num_regs) | |
2412 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2413 | } | |
2414 | ||
080c45fd RS |
2415 | if (NILP (string)) |
2416 | { | |
d7a5ad5f RS |
2417 | if (search_regs.start[sub] < BEGV |
2418 | || search_regs.start[sub] > search_regs.end[sub] | |
2419 | || search_regs.end[sub] > ZV) | |
2420 | args_out_of_range (make_number (search_regs.start[sub]), | |
2421 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2422 | } |
2423 | else | |
2424 | { | |
d7a5ad5f RS |
2425 | if (search_regs.start[sub] < 0 |
2426 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2427 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2428 | args_out_of_range (make_number (search_regs.start[sub]), |
2429 | make_number (search_regs.end[sub])); | |
080c45fd | 2430 | } |
ca1d1d23 JB |
2431 | |
2432 | if (NILP (fixedcase)) | |
2433 | { | |
2434 | /* Decide how to casify by examining the matched text. */ | |
e7deaab0 | 2435 | EMACS_INT last; |
ca1d1d23 | 2436 | |
ac3b28b1 KH |
2437 | pos = search_regs.start[sub]; |
2438 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2439 | |
2440 | if (NILP (string)) | |
ac3b28b1 | 2441 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2442 | else |
ac3b28b1 | 2443 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2444 | |
ca1d1d23 JB |
2445 | prevc = '\n'; |
2446 | case_action = all_caps; | |
2447 | ||
2448 | /* some_multiletter_word is set nonzero if any original word | |
2449 | is more than one letter long. */ | |
2450 | some_multiletter_word = 0; | |
97832bd0 | 2451 | some_lowercase = 0; |
208767c3 | 2452 | some_nonuppercase_initial = 0; |
73dc8771 | 2453 | some_uppercase = 0; |
ca1d1d23 | 2454 | |
ac3b28b1 | 2455 | while (pos < last) |
ca1d1d23 | 2456 | { |
080c45fd | 2457 | if (NILP (string)) |
ac3b28b1 | 2458 | { |
93daa011 | 2459 | c = FETCH_CHAR_AS_MULTIBYTE (pos_byte); |
ac3b28b1 KH |
2460 | INC_BOTH (pos, pos_byte); |
2461 | } | |
080c45fd | 2462 | else |
93daa011 | 2463 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2464 | |
5da9919f | 2465 | if (lowercasep (c)) |
ca1d1d23 JB |
2466 | { |
2467 | /* Cannot be all caps if any original char is lower case */ | |
2468 | ||
97832bd0 | 2469 | some_lowercase = 1; |
ca1d1d23 | 2470 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2471 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2472 | else |
2473 | some_multiletter_word = 1; | |
2474 | } | |
5da9919f | 2475 | else if (uppercasep (c)) |
ca1d1d23 | 2476 | { |
73dc8771 | 2477 | some_uppercase = 1; |
97832bd0 | 2478 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2479 | ; |
97832bd0 | 2480 | else |
ca1d1d23 JB |
2481 | some_multiletter_word = 1; |
2482 | } | |
208767c3 RS |
2483 | else |
2484 | { | |
2485 | /* If the initial is a caseless word constituent, | |
2486 | treat that like a lowercase initial. */ | |
2487 | if (SYNTAX (prevc) != Sword) | |
2488 | some_nonuppercase_initial = 1; | |
2489 | } | |
ca1d1d23 JB |
2490 | |
2491 | prevc = c; | |
2492 | } | |
2493 | ||
97832bd0 RS |
2494 | /* Convert to all caps if the old text is all caps |
2495 | and has at least one multiletter word. */ | |
2496 | if (! some_lowercase && some_multiletter_word) | |
2497 | case_action = all_caps; | |
c4d460ce | 2498 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2499 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2500 | case_action = cap_initial; |
208767c3 | 2501 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2502 | /* Should x -> yz, operating on X, give Yz or YZ? |
2503 | We'll assume the latter. */ | |
2504 | case_action = all_caps; | |
97832bd0 RS |
2505 | else |
2506 | case_action = nochange; | |
ca1d1d23 JB |
2507 | } |
2508 | ||
080c45fd RS |
2509 | /* Do replacement in a string. */ |
2510 | if (!NILP (string)) | |
2511 | { | |
2512 | Lisp_Object before, after; | |
2513 | ||
2514 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2515 | make_number (search_regs.start[sub])); |
2516 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2517 | |
636a5e28 RS |
2518 | /* Substitute parts of the match into NEWTEXT |
2519 | if desired. */ | |
080c45fd RS |
2520 | if (NILP (literal)) |
2521 | { | |
e7deaab0 AS |
2522 | EMACS_INT lastpos = 0; |
2523 | EMACS_INT lastpos_byte = 0; | |
080c45fd RS |
2524 | /* We build up the substituted string in ACCUM. */ |
2525 | Lisp_Object accum; | |
2526 | Lisp_Object middle; | |
a53e2e89 | 2527 | EMACS_INT length = SBYTES (newtext); |
080c45fd RS |
2528 | |
2529 | accum = Qnil; | |
2530 | ||
ac3b28b1 | 2531 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd | 2532 | { |
c098fdb8 EZ |
2533 | EMACS_INT substart = -1; |
2534 | EMACS_INT subend = 0; | |
1e79ec24 | 2535 | int delbackslash = 0; |
080c45fd | 2536 | |
0c8533c6 RS |
2537 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2538 | ||
080c45fd RS |
2539 | if (c == '\\') |
2540 | { | |
0c8533c6 | 2541 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2542 | |
080c45fd RS |
2543 | if (c == '&') |
2544 | { | |
d7a5ad5f RS |
2545 | substart = search_regs.start[sub]; |
2546 | subend = search_regs.end[sub]; | |
080c45fd | 2547 | } |
76fc1ea2 | 2548 | else if (c >= '1' && c <= '9') |
080c45fd | 2549 | { |
76fc1ea2 KH |
2550 | if (search_regs.start[c - '0'] >= 0 |
2551 | && c <= search_regs.num_regs + '0') | |
080c45fd RS |
2552 | { |
2553 | substart = search_regs.start[c - '0']; | |
2554 | subend = search_regs.end[c - '0']; | |
2555 | } | |
76fc1ea2 KH |
2556 | else |
2557 | { | |
2558 | /* If that subexp did not match, | |
2559 | replace \\N with nothing. */ | |
2560 | substart = 0; | |
2561 | subend = 0; | |
2562 | } | |
080c45fd | 2563 | } |
1e79ec24 KH |
2564 | else if (c == '\\') |
2565 | delbackslash = 1; | |
636a5e28 RS |
2566 | else |
2567 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2568 | } |
2569 | if (substart >= 0) | |
2570 | { | |
d131e79c RS |
2571 | if (pos - 2 != lastpos) |
2572 | middle = substring_both (newtext, lastpos, | |
2573 | lastpos_byte, | |
2574 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2575 | else |
2576 | middle = Qnil; | |
2577 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2578 | Fsubstring (string, |
2579 | make_number (substart), | |
080c45fd RS |
2580 | make_number (subend))); |
2581 | lastpos = pos; | |
0c8533c6 | 2582 | lastpos_byte = pos_byte; |
080c45fd | 2583 | } |
1e79ec24 KH |
2584 | else if (delbackslash) |
2585 | { | |
d131e79c RS |
2586 | middle = substring_both (newtext, lastpos, |
2587 | lastpos_byte, | |
2588 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2589 | |
1e79ec24 KH |
2590 | accum = concat2 (accum, middle); |
2591 | lastpos = pos; | |
0c8533c6 | 2592 | lastpos_byte = pos_byte; |
1e79ec24 | 2593 | } |
080c45fd RS |
2594 | } |
2595 | ||
d131e79c RS |
2596 | if (pos != lastpos) |
2597 | middle = substring_both (newtext, lastpos, | |
2598 | lastpos_byte, | |
0c8533c6 | 2599 | pos, pos_byte); |
080c45fd RS |
2600 | else |
2601 | middle = Qnil; | |
2602 | ||
2603 | newtext = concat2 (accum, middle); | |
2604 | } | |
2605 | ||
636a5e28 | 2606 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2607 | if (case_action == all_caps) |
2608 | newtext = Fupcase (newtext); | |
2609 | else if (case_action == cap_initial) | |
2b2eead9 | 2610 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2611 | |
2612 | return concat3 (before, newtext, after); | |
2613 | } | |
2614 | ||
09c4719e | 2615 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2616 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2617 | opoint = PT - ZV; |
9160906f RS |
2618 | else if (PT > search_regs.start[sub]) |
2619 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2620 | else |
2621 | opoint = PT; | |
2622 | ||
886ed6ec RS |
2623 | /* If we want non-literal replacement, |
2624 | perform substitution on the replacement string. */ | |
2625 | if (NILP (literal)) | |
ca1d1d23 | 2626 | { |
c098fdb8 | 2627 | EMACS_INT length = SBYTES (newtext); |
68e69fbd | 2628 | unsigned char *substed; |
c098fdb8 | 2629 | EMACS_INT substed_alloc_size, substed_len; |
4b4deea2 | 2630 | int buf_multibyte = !NILP (BVAR (current_buffer, enable_multibyte_characters)); |
3bc25e52 | 2631 | int str_multibyte = STRING_MULTIBYTE (newtext); |
886ed6ec | 2632 | int really_changed = 0; |
3bc25e52 | 2633 | |
68e69fbd RS |
2634 | substed_alloc_size = length * 2 + 100; |
2635 | substed = (unsigned char *) xmalloc (substed_alloc_size + 1); | |
2636 | substed_len = 0; | |
2637 | ||
3bc25e52 KH |
2638 | /* Go thru NEWTEXT, producing the actual text to insert in |
2639 | SUBSTED while adjusting multibyteness to that of the current | |
2640 | buffer. */ | |
ca1d1d23 | 2641 | |
ac3b28b1 | 2642 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2643 | { |
68e69fbd | 2644 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
8ea90aa3 | 2645 | const unsigned char *add_stuff = NULL; |
c098fdb8 | 2646 | EMACS_INT add_len = 0; |
68e69fbd | 2647 | int idx = -1; |
9a76659d | 2648 | |
3bc25e52 KH |
2649 | if (str_multibyte) |
2650 | { | |
eb99a8dd | 2651 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 | 2652 | if (!buf_multibyte) |
461c2ab9 | 2653 | c = multibyte_char_to_unibyte (c); |
3bc25e52 KH |
2654 | } |
2655 | else | |
2656 | { | |
2657 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2658 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2659 | if (buf_multibyte) |
4c0354d7 | 2660 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 | 2661 | } |
ac3b28b1 | 2662 | |
68e69fbd RS |
2663 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2664 | or set IDX to a match index, which means put that part | |
2665 | of the buffer text into SUBSTED. */ | |
2666 | ||
ca1d1d23 JB |
2667 | if (c == '\\') |
2668 | { | |
886ed6ec RS |
2669 | really_changed = 1; |
2670 | ||
3bc25e52 KH |
2671 | if (str_multibyte) |
2672 | { | |
eb99a8dd KH |
2673 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2674 | pos, pos_byte); | |
071ce769 | 2675 | if (!buf_multibyte && !ASCII_CHAR_P (c)) |
461c2ab9 | 2676 | c = multibyte_char_to_unibyte (c); |
3bc25e52 KH |
2677 | } |
2678 | else | |
2679 | { | |
d5db4077 | 2680 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2681 | if (buf_multibyte) |
4c0354d7 | 2682 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 KH |
2683 | } |
2684 | ||
ca1d1d23 | 2685 | if (c == '&') |
68e69fbd | 2686 | idx = sub; |
78445046 | 2687 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2688 | { |
2689 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2690 | idx = c - '0'; |
ca1d1d23 | 2691 | } |
636a5e28 | 2692 | else if (c == '\\') |
a7e979a4 | 2693 | add_len = 1, add_stuff = (unsigned char *) "\\"; |
636a5e28 | 2694 | else |
3bc25e52 KH |
2695 | { |
2696 | xfree (substed); | |
2697 | error ("Invalid use of `\\' in replacement text"); | |
2698 | } | |
ca1d1d23 JB |
2699 | } |
2700 | else | |
68e69fbd RS |
2701 | { |
2702 | add_len = CHAR_STRING (c, str); | |
2703 | add_stuff = str; | |
2704 | } | |
2705 | ||
2706 | /* If we want to copy part of a previous match, | |
2707 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2708 | if (idx >= 0) | |
2709 | { | |
e7deaab0 | 2710 | EMACS_INT begbyte = CHAR_TO_BYTE (search_regs.start[idx]); |
68e69fbd RS |
2711 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; |
2712 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2713 | move_gap (search_regs.start[idx]); | |
2714 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2715 | } | |
2716 | ||
2717 | /* Now the stuff we want to add to SUBSTED | |
2718 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2719 | ||
2720 | /* Make sure SUBSTED is big enough. */ | |
2721 | if (substed_len + add_len >= substed_alloc_size) | |
2722 | { | |
2723 | substed_alloc_size = substed_len + add_len + 500; | |
2724 | substed = (unsigned char *) xrealloc (substed, | |
2725 | substed_alloc_size + 1); | |
2726 | } | |
2727 | ||
2728 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2729 | if (add_stuff) |
2730 | { | |
72af86bd | 2731 | memcpy (substed + substed_len, add_stuff, add_len); |
f8ce8a0d GM |
2732 | substed_len += add_len; |
2733 | } | |
ca1d1d23 | 2734 | } |
68e69fbd | 2735 | |
886ed6ec | 2736 | if (really_changed) |
76fc1ea2 KH |
2737 | { |
2738 | if (buf_multibyte) | |
2739 | { | |
c098fdb8 EZ |
2740 | EMACS_INT nchars = |
2741 | multibyte_chars_in_text (substed, substed_len); | |
68e69fbd | 2742 | |
a7e979a4 PE |
2743 | newtext = make_multibyte_string ((char *) substed, nchars, |
2744 | substed_len); | |
76fc1ea2 KH |
2745 | } |
2746 | else | |
a7e979a4 | 2747 | newtext = make_unibyte_string ((char *) substed, substed_len); |
76fc1ea2 | 2748 | } |
68e69fbd | 2749 | xfree (substed); |
ca1d1d23 JB |
2750 | } |
2751 | ||
886ed6ec RS |
2752 | /* Replace the old text with the new in the cleanest possible way. */ |
2753 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2754 | newtext, 1, 0, 1); | |
d5db4077 | 2755 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2756 | |
2757 | if (case_action == all_caps) | |
886ed6ec RS |
2758 | Fupcase_region (make_number (search_regs.start[sub]), |
2759 | make_number (newpoint)); | |
ca1d1d23 | 2760 | else if (case_action == cap_initial) |
886ed6ec RS |
2761 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2762 | make_number (newpoint)); | |
3e18eecf | 2763 | |
98e942e0 RS |
2764 | /* Adjust search data for this change. */ |
2765 | { | |
e7deaab0 AS |
2766 | EMACS_INT oldend = search_regs.end[sub]; |
2767 | EMACS_INT oldstart = search_regs.start[sub]; | |
2768 | EMACS_INT change = newpoint - search_regs.end[sub]; | |
98e942e0 RS |
2769 | int i; |
2770 | ||
2771 | for (i = 0; i < search_regs.num_regs; i++) | |
2772 | { | |
41c01205 | 2773 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2774 | search_regs.start[i] += change; |
41c01205 DK |
2775 | else if (search_regs.start[i] > oldstart) |
2776 | search_regs.start[i] = oldstart; | |
2777 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2778 | search_regs.end[i] += change; |
41c01205 DK |
2779 | else if (search_regs.end[i] > oldstart) |
2780 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2781 | } |
2782 | } | |
2783 | ||
b0eba991 | 2784 | /* Put point back where it was in the text. */ |
8d808a65 | 2785 | if (opoint <= 0) |
fa8ed3e0 | 2786 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2787 | else |
fa8ed3e0 | 2788 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2789 | |
2790 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2791 | move_if_not_intangible (newpoint); |
177c0ea7 | 2792 | |
ca1d1d23 JB |
2793 | return Qnil; |
2794 | } | |
2795 | \f | |
2796 | static Lisp_Object | |
971de7fb | 2797 | match_limit (Lisp_Object num, int beginningp) |
ca1d1d23 JB |
2798 | { |
2799 | register int n; | |
2800 | ||
b7826503 | 2801 | CHECK_NUMBER (num); |
ca1d1d23 | 2802 | n = XINT (num); |
f90a5bf5 | 2803 | if (n < 0) |
bd2cbd56 | 2804 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2805 | if (search_regs.num_regs <= 0) |
2806 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2807 | if (n >= search_regs.num_regs |
4746118a | 2808 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2809 | return Qnil; |
2810 | return (make_number ((beginningp) ? search_regs.start[n] | |
2811 | : search_regs.end[n])); | |
2812 | } | |
2813 | ||
16a97296 | 2814 | DEFUE ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, |
8c1a1077 PJ |
2815 | doc: /* Return position of start of text matched by last search. |
2816 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2817 | regexp. | |
2818 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2819 | SUBEXP pairs. | |
2820 | Zero means the entire text matched by the whole regexp or whole string. */) | |
5842a27b | 2821 | (Lisp_Object subexp) |
ca1d1d23 | 2822 | { |
5806161b | 2823 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2824 | } |
2825 | ||
16a97296 | 2826 | DEFUE ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, |
8c1a1077 PJ |
2827 | doc: /* Return position of end of text matched by last search. |
2828 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2829 | regexp. | |
2830 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2831 | SUBEXP pairs. | |
2832 | Zero means the entire text matched by the whole regexp or whole string. */) | |
5842a27b | 2833 | (Lisp_Object subexp) |
ca1d1d23 | 2834 | { |
5806161b | 2835 | return match_limit (subexp, 0); |
177c0ea7 | 2836 | } |
ca1d1d23 | 2837 | |
16a97296 | 2838 | DEFUE ("match-data", Fmatch_data, Smatch_data, 0, 3, 0, |
8c1a1077 PJ |
2839 | doc: /* Return a list containing all info on what the last search matched. |
2840 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2841 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2842 | if the last match was on a buffer; integers or nil if a string was matched. | |
febb8aba | 2843 | Use `set-match-data' to reinstate the data in this list. |
8c1a1077 | 2844 | |
41c01205 DK |
2845 | If INTEGERS (the optional first argument) is non-nil, always use |
2846 | integers \(rather than markers) to represent buffer positions. In | |
2847 | this case, and if the last match was in a buffer, the buffer will get | |
2848 | stored as one additional element at the end of the list. | |
2849 | ||
abd0071c KS |
2850 | If REUSE is a list, reuse it as part of the value. If REUSE is long |
2851 | enough to hold all the values, and if INTEGERS is non-nil, no consing | |
2852 | is done. | |
2853 | ||
2854 | If optional third arg RESEAT is non-nil, any previous markers on the | |
2855 | REUSE list will be modified to point to nowhere. | |
2856 | ||
140a6b7e | 2857 | Return value is undefined if the last search failed. */) |
5842a27b | 2858 | (Lisp_Object integers, Lisp_Object reuse, Lisp_Object reseat) |
ca1d1d23 | 2859 | { |
56256c2a | 2860 | Lisp_Object tail, prev; |
4746118a | 2861 | Lisp_Object *data; |
ca1d1d23 JB |
2862 | int i, len; |
2863 | ||
abd0071c KS |
2864 | if (!NILP (reseat)) |
2865 | for (tail = reuse; CONSP (tail); tail = XCDR (tail)) | |
2866 | if (MARKERP (XCAR (tail))) | |
2867 | { | |
51f10faa | 2868 | unchain_marker (XMARKER (XCAR (tail))); |
abd0071c KS |
2869 | XSETCAR (tail, Qnil); |
2870 | } | |
2871 | ||
daa37602 | 2872 | if (NILP (last_thing_searched)) |
c36bcf1b | 2873 | return Qnil; |
daa37602 | 2874 | |
6bbd7a29 GM |
2875 | prev = Qnil; |
2876 | ||
41c01205 | 2877 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs + 1) |
4746118a JB |
2878 | * sizeof (Lisp_Object)); |
2879 | ||
41c01205 | 2880 | len = 0; |
4746118a | 2881 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 | 2882 | { |
a53e2e89 | 2883 | EMACS_INT start = search_regs.start[i]; |
ca1d1d23 JB |
2884 | if (start >= 0) |
2885 | { | |
56256c2a RS |
2886 | if (EQ (last_thing_searched, Qt) |
2887 | || ! NILP (integers)) | |
ca1d1d23 | 2888 | { |
c235cce7 KH |
2889 | XSETFASTINT (data[2 * i], start); |
2890 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2891 | } |
0ed62dc7 | 2892 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2893 | { |
2894 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2895 | Fset_marker (data[2 * i], |
2896 | make_number (start), | |
2897 | last_thing_searched); | |
ca1d1d23 JB |
2898 | data[2 * i + 1] = Fmake_marker (); |
2899 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2900 | make_number (search_regs.end[i]), |
daa37602 | 2901 | last_thing_searched); |
ca1d1d23 | 2902 | } |
daa37602 JB |
2903 | else |
2904 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2905 | abort (); | |
2906 | ||
abd0071c | 2907 | len = 2 * i + 2; |
ca1d1d23 JB |
2908 | } |
2909 | else | |
abd0071c | 2910 | data[2 * i] = data[2 * i + 1] = Qnil; |
ca1d1d23 | 2911 | } |
56256c2a | 2912 | |
bd2cbd56 | 2913 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2914 | { |
bd2cbd56 | 2915 | data[len] = last_thing_searched; |
41c01205 DK |
2916 | len++; |
2917 | } | |
2918 | ||
56256c2a RS |
2919 | /* If REUSE is not usable, cons up the values and return them. */ |
2920 | if (! CONSP (reuse)) | |
41c01205 | 2921 | return Flist (len, data); |
56256c2a RS |
2922 | |
2923 | /* If REUSE is a list, store as many value elements as will fit | |
2924 | into the elements of REUSE. */ | |
2925 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2926 | i++, tail = XCDR (tail)) |
56256c2a | 2927 | { |
41c01205 | 2928 | if (i < len) |
f3fbd155 | 2929 | XSETCAR (tail, data[i]); |
56256c2a | 2930 | else |
f3fbd155 | 2931 | XSETCAR (tail, Qnil); |
56256c2a RS |
2932 | prev = tail; |
2933 | } | |
2934 | ||
2935 | /* If we couldn't fit all value elements into REUSE, | |
2936 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2937 | if (i < len) |
2938 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2939 | |
2940 | return reuse; | |
ca1d1d23 JB |
2941 | } |
2942 | ||
b51d6c92 SM |
2943 | /* We used to have an internal use variant of `reseat' described as: |
2944 | ||
2945 | If RESEAT is `evaporate', put the markers back on the free list | |
2946 | immediately. No other references to the markers must exist in this | |
2947 | case, so it is used only internally on the unwind stack and | |
2948 | save-match-data from Lisp. | |
2949 | ||
2950 | But it was ill-conceived: those supposedly-internal markers get exposed via | |
2951 | the undo-list, so freeing them here is unsafe. */ | |
ca1d1d23 | 2952 | |
16a97296 | 2953 | DEFUE ("set-match-data", Fset_match_data, Sset_match_data, 1, 2, 0, |
8c1a1077 | 2954 | doc: /* Set internal data on last search match from elements of LIST. |
abd0071c KS |
2955 | LIST should have been created by calling `match-data' previously. |
2956 | ||
51f10faa | 2957 | If optional arg RESEAT is non-nil, make markers on LIST point nowhere. */) |
5842a27b | 2958 | (register Lisp_Object list, Lisp_Object reseat) |
ca1d1d23 JB |
2959 | { |
2960 | register int i; | |
2961 | register Lisp_Object marker; | |
2962 | ||
7074fde6 FP |
2963 | if (running_asynch_code) |
2964 | save_search_regs (); | |
2965 | ||
29100cea | 2966 | CHECK_LIST (list); |
ca1d1d23 | 2967 | |
41c01205 DK |
2968 | /* Unless we find a marker with a buffer or an explicit buffer |
2969 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
2970 | last_thing_searched = Qt; |
2971 | ||
4746118a JB |
2972 | /* Allocate registers if they don't already exist. */ |
2973 | { | |
d084e942 | 2974 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2975 | |
2976 | if (length > search_regs.num_regs) | |
2977 | { | |
1113d9db JB |
2978 | if (search_regs.num_regs == 0) |
2979 | { | |
2980 | search_regs.start | |
2981 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2982 | search_regs.end | |
2983 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2984 | } | |
4746118a | 2985 | else |
1113d9db JB |
2986 | { |
2987 | search_regs.start | |
2988 | = (regoff_t *) xrealloc (search_regs.start, | |
2989 | length * sizeof (regoff_t)); | |
2990 | search_regs.end | |
2991 | = (regoff_t *) xrealloc (search_regs.end, | |
2992 | length * sizeof (regoff_t)); | |
2993 | } | |
4746118a | 2994 | |
e62371e9 KH |
2995 | for (i = search_regs.num_regs; i < length; i++) |
2996 | search_regs.start[i] = -1; | |
2997 | ||
487282dc | 2998 | search_regs.num_regs = length; |
4746118a | 2999 | } |
ca1d1d23 | 3000 | |
abd0071c | 3001 | for (i = 0; CONSP (list); i++) |
41c01205 | 3002 | { |
abd0071c | 3003 | marker = XCAR (list); |
bd2cbd56 | 3004 | if (BUFFERP (marker)) |
c3762cbd | 3005 | { |
bd2cbd56 | 3006 | last_thing_searched = marker; |
c3762cbd DK |
3007 | break; |
3008 | } | |
3009 | if (i >= length) | |
3010 | break; | |
41c01205 DK |
3011 | if (NILP (marker)) |
3012 | { | |
3013 | search_regs.start[i] = -1; | |
abd0071c | 3014 | list = XCDR (list); |
41c01205 DK |
3015 | } |
3016 | else | |
3017 | { | |
e7deaab0 | 3018 | EMACS_INT from; |
abd0071c | 3019 | Lisp_Object m; |
e2811828 | 3020 | |
abd0071c | 3021 | m = marker; |
41c01205 DK |
3022 | if (MARKERP (marker)) |
3023 | { | |
3024 | if (XMARKER (marker)->buffer == 0) | |
3025 | XSETFASTINT (marker, 0); | |
3026 | else | |
3027 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
3028 | } | |
e2811828 | 3029 | |
41c01205 DK |
3030 | CHECK_NUMBER_COERCE_MARKER (marker); |
3031 | from = XINT (marker); | |
e2811828 | 3032 | |
abd0071c KS |
3033 | if (!NILP (reseat) && MARKERP (m)) |
3034 | { | |
b51d6c92 | 3035 | unchain_marker (XMARKER (m)); |
9ad54a7e | 3036 | XSETCAR (list, Qnil); |
abd0071c KS |
3037 | } |
3038 | ||
3039 | if ((list = XCDR (list), !CONSP (list))) | |
3040 | break; | |
3041 | ||
3042 | m = marker = XCAR (list); | |
3043 | ||
41c01205 DK |
3044 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
3045 | XSETFASTINT (marker, 0); | |
e2811828 | 3046 | |
41c01205 DK |
3047 | CHECK_NUMBER_COERCE_MARKER (marker); |
3048 | search_regs.start[i] = from; | |
3049 | search_regs.end[i] = XINT (marker); | |
abd0071c KS |
3050 | |
3051 | if (!NILP (reseat) && MARKERP (m)) | |
3052 | { | |
b51d6c92 | 3053 | unchain_marker (XMARKER (m)); |
9ad54a7e | 3054 | XSETCAR (list, Qnil); |
abd0071c | 3055 | } |
41c01205 | 3056 | } |
abd0071c | 3057 | list = XCDR (list); |
41c01205 | 3058 | } |
ca1d1d23 | 3059 | |
41c01205 DK |
3060 | for (; i < search_regs.num_regs; i++) |
3061 | search_regs.start[i] = -1; | |
3062 | } | |
ca1d1d23 | 3063 | |
177c0ea7 | 3064 | return Qnil; |
ca1d1d23 JB |
3065 | } |
3066 | ||
7074fde6 FP |
3067 | /* If non-zero the match data have been saved in saved_search_regs |
3068 | during the execution of a sentinel or filter. */ | |
75ebf74b | 3069 | static int search_regs_saved; |
7074fde6 | 3070 | static struct re_registers saved_search_regs; |
41c01205 | 3071 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
3072 | |
3073 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
3074 | if asynchronous code (filter or sentinel) is running. */ | |
3075 | static void | |
971de7fb | 3076 | save_search_regs (void) |
7074fde6 FP |
3077 | { |
3078 | if (!search_regs_saved) | |
3079 | { | |
3080 | saved_search_regs.num_regs = search_regs.num_regs; | |
3081 | saved_search_regs.start = search_regs.start; | |
3082 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
3083 | saved_last_thing_searched = last_thing_searched; |
3084 | last_thing_searched = Qnil; | |
7074fde6 | 3085 | search_regs.num_regs = 0; |
2d4a771a RS |
3086 | search_regs.start = 0; |
3087 | search_regs.end = 0; | |
7074fde6 FP |
3088 | |
3089 | search_regs_saved = 1; | |
3090 | } | |
3091 | } | |
3092 | ||
3093 | /* Called upon exit from filters and sentinels. */ | |
3094 | void | |
971de7fb | 3095 | restore_search_regs (void) |
7074fde6 FP |
3096 | { |
3097 | if (search_regs_saved) | |
3098 | { | |
3099 | if (search_regs.num_regs > 0) | |
3100 | { | |
3101 | xfree (search_regs.start); | |
3102 | xfree (search_regs.end); | |
3103 | } | |
3104 | search_regs.num_regs = saved_search_regs.num_regs; | |
3105 | search_regs.start = saved_search_regs.start; | |
3106 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
3107 | last_thing_searched = saved_last_thing_searched; |
3108 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
3109 | search_regs_saved = 0; |
3110 | } | |
3111 | } | |
3112 | ||
abd0071c | 3113 | static Lisp_Object |
971de7fb | 3114 | unwind_set_match_data (Lisp_Object list) |
abd0071c | 3115 | { |
b51d6c92 SM |
3116 | /* It is NOT ALWAYS safe to free (evaporate) the markers immediately. */ |
3117 | return Fset_match_data (list, Qt); | |
abd0071c KS |
3118 | } |
3119 | ||
3120 | /* Called to unwind protect the match data. */ | |
3121 | void | |
971de7fb | 3122 | record_unwind_save_match_data (void) |
abd0071c KS |
3123 | { |
3124 | record_unwind_protect (unwind_set_match_data, | |
3125 | Fmatch_data (Qnil, Qnil, Qnil)); | |
3126 | } | |
3127 | ||
ca1d1d23 JB |
3128 | /* Quote a string to inactivate reg-expr chars */ |
3129 | ||
3130 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 | 3131 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
5842a27b | 3132 | (Lisp_Object string) |
ca1d1d23 | 3133 | { |
a7e979a4 PE |
3134 | register char *in, *out, *end; |
3135 | register char *temp; | |
0c8533c6 | 3136 | int backslashes_added = 0; |
ca1d1d23 | 3137 | |
b7826503 | 3138 | CHECK_STRING (string); |
ca1d1d23 | 3139 | |
a7e979a4 | 3140 | temp = (char *) alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
3141 | |
3142 | /* Now copy the data into the new string, inserting escapes. */ | |
3143 | ||
a7e979a4 | 3144 | in = SSDATA (string); |
d5db4077 | 3145 | end = in + SBYTES (string); |
177c0ea7 | 3146 | out = temp; |
ca1d1d23 JB |
3147 | |
3148 | for (; in != end; in++) | |
3149 | { | |
66bc6082 | 3150 | if (*in == '[' |
ca1d1d23 JB |
3151 | || *in == '*' || *in == '.' || *in == '\\' |
3152 | || *in == '?' || *in == '+' | |
3153 | || *in == '^' || *in == '$') | |
0c8533c6 | 3154 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
3155 | *out++ = *in; |
3156 | } | |
3157 | ||
3f8100f1 | 3158 | return make_specified_string (temp, |
d5db4077 | 3159 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
3160 | out - temp, |
3161 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 3162 | } |
177c0ea7 | 3163 | \f |
dfcf069d | 3164 | void |
971de7fb | 3165 | syms_of_search (void) |
ca1d1d23 JB |
3166 | { |
3167 | register int i; | |
3168 | ||
487282dc KH |
3169 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
3170 | { | |
3171 | searchbufs[i].buf.allocated = 100; | |
b23c0a83 | 3172 | searchbufs[i].buf.buffer = (unsigned char *) xmalloc (100); |
487282dc KH |
3173 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
3174 | searchbufs[i].regexp = Qnil; | |
ecdb561e | 3175 | searchbufs[i].whitespace_regexp = Qnil; |
b69e3c18 | 3176 | searchbufs[i].syntax_table = Qnil; |
487282dc | 3177 | staticpro (&searchbufs[i].regexp); |
aa77b5ce | 3178 | staticpro (&searchbufs[i].whitespace_regexp); |
b69e3c18 | 3179 | staticpro (&searchbufs[i].syntax_table); |
487282dc KH |
3180 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); |
3181 | } | |
3182 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 | 3183 | |
d67b4f80 | 3184 | Qsearch_failed = intern_c_string ("search-failed"); |
ca1d1d23 | 3185 | staticpro (&Qsearch_failed); |
d67b4f80 | 3186 | Qinvalid_regexp = intern_c_string ("invalid-regexp"); |
ca1d1d23 JB |
3187 | staticpro (&Qinvalid_regexp); |
3188 | ||
3189 | Fput (Qsearch_failed, Qerror_conditions, | |
d67b4f80 | 3190 | pure_cons (Qsearch_failed, pure_cons (Qerror, Qnil))); |
ca1d1d23 | 3191 | Fput (Qsearch_failed, Qerror_message, |
d67b4f80 | 3192 | make_pure_c_string ("Search failed")); |
ca1d1d23 JB |
3193 | |
3194 | Fput (Qinvalid_regexp, Qerror_conditions, | |
d67b4f80 | 3195 | pure_cons (Qinvalid_regexp, pure_cons (Qerror, Qnil))); |
ca1d1d23 | 3196 | Fput (Qinvalid_regexp, Qerror_message, |
d67b4f80 | 3197 | make_pure_c_string ("Invalid regexp")); |
ca1d1d23 | 3198 | |
daa37602 JB |
3199 | last_thing_searched = Qnil; |
3200 | staticpro (&last_thing_searched); | |
3201 | ||
0f6af254 DK |
3202 | saved_last_thing_searched = Qnil; |
3203 | staticpro (&saved_last_thing_searched); | |
3204 | ||
29208e82 | 3205 | DEFVAR_LISP ("search-spaces-regexp", Vsearch_spaces_regexp, |
e2811828 | 3206 | doc: /* Regexp to substitute for bunches of spaces in regexp search. |
f31a9a68 RS |
3207 | Some commands use this for user-specified regexps. |
3208 | Spaces that occur inside character classes or repetition operators | |
3209 | or other such regexp constructs are not replaced with this. | |
3210 | A value of nil (which is the normal value) means treat spaces literally. */); | |
41a33295 | 3211 | Vsearch_spaces_regexp = Qnil; |
f31a9a68 | 3212 | |
29208e82 | 3213 | DEFVAR_LISP ("inhibit-changing-match-data", Vinhibit_changing_match_data, |
ef887810 | 3214 | doc: /* Internal use only. |
39d0bf74 RS |
3215 | If non-nil, the primitive searching and matching functions |
3216 | such as `looking-at', `string-match', `re-search-forward', etc., | |
3217 | do not set the match data. The proper way to use this variable | |
3218 | is to bind it with `let' around a small expression. */); | |
ef887810 RS |
3219 | Vinhibit_changing_match_data = Qnil; |
3220 | ||
ca1d1d23 | 3221 | defsubr (&Slooking_at); |
b819a390 RS |
3222 | defsubr (&Sposix_looking_at); |
3223 | defsubr (&Sstring_match); | |
3224 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
3225 | defsubr (&Ssearch_forward); |
3226 | defsubr (&Ssearch_backward); | |
3227 | defsubr (&Sword_search_forward); | |
3228 | defsubr (&Sword_search_backward); | |
c6e0e1cc CY |
3229 | defsubr (&Sword_search_forward_lax); |
3230 | defsubr (&Sword_search_backward_lax); | |
ca1d1d23 JB |
3231 | defsubr (&Sre_search_forward); |
3232 | defsubr (&Sre_search_backward); | |
b819a390 RS |
3233 | defsubr (&Sposix_search_forward); |
3234 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
3235 | defsubr (&Sreplace_match); |
3236 | defsubr (&Smatch_beginning); | |
3237 | defsubr (&Smatch_end); | |
3238 | defsubr (&Smatch_data); | |
3f1c005b | 3239 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
3240 | defsubr (&Sregexp_quote); |
3241 | } |