Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
4fb9a543 | 2 | |
ab422c4d PE |
3 | Copyright (C) 1985-1987, 1993-1994, 1997-1999, 2001-2013 Free Software |
4 | Foundation, Inc. | |
ca1d1d23 JB |
5 | |
6 | This file is part of GNU Emacs. | |
7 | ||
9ec0b715 | 8 | GNU Emacs is free software: you can redistribute it and/or modify |
ca1d1d23 | 9 | it under the terms of the GNU General Public License as published by |
9ec0b715 GM |
10 | the Free Software Foundation, either version 3 of the License, or |
11 | (at your option) any later version. | |
ca1d1d23 JB |
12 | |
13 | GNU Emacs is distributed in the hope that it will be useful, | |
14 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
15 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
16 | GNU General Public License for more details. | |
17 | ||
18 | You should have received a copy of the GNU General Public License | |
9ec0b715 | 19 | along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */ |
ca1d1d23 JB |
20 | |
21 | ||
18160b98 | 22 | #include <config.h> |
0328b6de | 23 | |
ca1d1d23 | 24 | #include "lisp.h" |
5679531d | 25 | #include "category.h" |
76eb0881 | 26 | #include "character.h" |
e5560ff7 | 27 | #include "buffer.h" |
45b683a1 | 28 | #include "syntax.h" |
89ad6492 | 29 | #include "charset.h" |
9169c321 | 30 | #include "region-cache.h" |
ca1d1d23 | 31 | #include "commands.h" |
9ac0d9e0 | 32 | #include "blockinput.h" |
bf1760bb | 33 | #include "intervals.h" |
4746118a | 34 | |
ca1d1d23 JB |
35 | #include <sys/types.h> |
36 | #include "regex.h" | |
37 | ||
1d288aef | 38 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 39 | |
487282dc KH |
40 | /* If the regexp is non-nil, then the buffer contains the compiled form |
41 | of that regexp, suitable for searching. */ | |
1d288aef RS |
42 | struct regexp_cache |
43 | { | |
487282dc | 44 | struct regexp_cache *next; |
ecdb561e | 45 | Lisp_Object regexp, whitespace_regexp; |
b69e3c18 | 46 | /* Syntax table for which the regexp applies. We need this because |
58e95211 SM |
47 | of character classes. If this is t, then the compiled pattern is valid |
48 | for any syntax-table. */ | |
b69e3c18 | 49 | Lisp_Object syntax_table; |
487282dc KH |
50 | struct re_pattern_buffer buf; |
51 | char fastmap[0400]; | |
457882c2 PE |
52 | /* True means regexp was compiled to do full POSIX backtracking. */ |
53 | bool posix; | |
487282dc | 54 | }; |
ca1d1d23 | 55 | |
487282dc | 56 | /* The instances of that struct. */ |
989b29ad | 57 | static struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; |
ca1d1d23 | 58 | |
487282dc | 59 | /* The head of the linked list; points to the most recently used buffer. */ |
989b29ad | 60 | static struct regexp_cache *searchbuf_head; |
ca1d1d23 | 61 | |
ca1d1d23 | 62 | |
4746118a JB |
63 | /* Every call to re_match, etc., must pass &search_regs as the regs |
64 | argument unless you can show it is unnecessary (i.e., if re_match | |
65 | is certainly going to be called again before region-around-match | |
66 | can be called). | |
67 | ||
68 | Since the registers are now dynamically allocated, we need to make | |
69 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
70 | been allocated by checking search_regs.num_regs. |
71 | ||
72 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
73 | buffer using bits in the re_pattern_buffer. This means that whenever |
74 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
75 | allocated any registers, and will allocate new registers the next |
76 | time you call a searching or matching function. Therefore, we need | |
77 | to call re_set_registers after compiling a new pattern or after | |
78 | setting the match registers, so that the regex functions will be | |
79 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
80 | static struct re_registers search_regs; |
81 | ||
daa37602 JB |
82 | /* The buffer in which the last search was performed, or |
83 | Qt if the last search was done in a string; | |
84 | Qnil if no searching has been done yet. */ | |
85 | static Lisp_Object last_thing_searched; | |
ca1d1d23 | 86 | |
02b16839 | 87 | /* Error condition signaled when regexp compile_pattern fails. */ |
955cbe7b | 88 | static Lisp_Object Qinvalid_regexp; |
ca1d1d23 | 89 | |
02b16839 | 90 | /* Error condition used for failing searches. */ |
955cbe7b | 91 | static Lisp_Object Qsearch_failed; |
06f77a8a | 92 | |
d311d28c | 93 | static void set_search_regs (ptrdiff_t, ptrdiff_t); |
f57e2426 | 94 | static void save_search_regs (void); |
d311d28c PE |
95 | static EMACS_INT simple_search (EMACS_INT, unsigned char *, ptrdiff_t, |
96 | ptrdiff_t, Lisp_Object, ptrdiff_t, ptrdiff_t, | |
97 | ptrdiff_t, ptrdiff_t); | |
98 | static EMACS_INT boyer_moore (EMACS_INT, unsigned char *, ptrdiff_t, | |
99 | Lisp_Object, Lisp_Object, ptrdiff_t, | |
100 | ptrdiff_t, int); | |
101 | static EMACS_INT search_buffer (Lisp_Object, ptrdiff_t, ptrdiff_t, | |
102 | ptrdiff_t, ptrdiff_t, EMACS_INT, int, | |
457882c2 | 103 | Lisp_Object, Lisp_Object, bool); |
b819a390 | 104 | |
845ca893 | 105 | static _Noreturn void |
971de7fb | 106 | matcher_overflow (void) |
ca1d1d23 JB |
107 | { |
108 | error ("Stack overflow in regexp matcher"); | |
109 | } | |
110 | ||
b819a390 RS |
111 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
112 | PATTERN is the pattern to compile. | |
113 | CP is the place to put the result. | |
facdc750 | 114 | TRANSLATE is a translation table for ignoring case, or nil for none. |
457882c2 PE |
115 | POSIX is true if we want full backtracking (POSIX style) for this pattern. |
116 | False means backtrack only enough to get a valid match. | |
bbc73b48 KH |
117 | |
118 | The behavior also depends on Vsearch_spaces_regexp. */ | |
ca1d1d23 | 119 | |
487282dc | 120 | static void |
457882c2 PE |
121 | compile_pattern_1 (struct regexp_cache *cp, Lisp_Object pattern, |
122 | Lisp_Object translate, bool 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); |
4d7e6e51 | 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; | |
38182d90 | 179 | cp->buf.buffer = xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
180 | } |
181 | } | |
182 | ||
54dd3310 SM |
183 | /* Clear the regexp cache w.r.t. a particular syntax table, |
184 | because it was changed. | |
e5b94d44 CY |
185 | There is no danger of memory leak here because re_compile_pattern |
186 | automagically manages the memory in each re_pattern_buffer struct, | |
187 | based on its `allocated' and `buffer' values. */ | |
188 | void | |
971de7fb | 189 | clear_regexp_cache (void) |
e5b94d44 CY |
190 | { |
191 | int i; | |
192 | ||
193 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) | |
920fd1fc EZ |
194 | /* It's tempting to compare with the syntax-table we've actually changed, |
195 | but it's not sufficient because char-table inheritance means that | |
54dd3310 SM |
196 | modifying one syntax-table can change others at the same time. */ |
197 | if (!EQ (searchbufs[i].syntax_table, Qt)) | |
198 | searchbufs[i].regexp = Qnil; | |
e5b94d44 CY |
199 | } |
200 | ||
487282dc | 201 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
202 | the cache. |
203 | PATTERN is the pattern to compile. | |
facdc750 | 204 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
205 | REGP is the structure that says where to store the "register" |
206 | values that will result from matching this pattern. | |
207 | If it is 0, we should compile the pattern not to record any | |
208 | subexpression bounds. | |
457882c2 PE |
209 | POSIX is true if we want full backtracking (POSIX style) for this pattern. |
210 | False means backtrack only enough to get a valid match. */ | |
487282dc KH |
211 | |
212 | struct re_pattern_buffer * | |
0063fdb1 | 213 | compile_pattern (Lisp_Object pattern, struct re_registers *regp, |
457882c2 | 214 | Lisp_Object translate, bool posix, bool 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 |
457882c2 | 271 | looking_at_1 (Lisp_Object string, bool posix) |
ca1d1d23 JB |
272 | { |
273 | Lisp_Object val; | |
274 | unsigned char *p1, *p2; | |
d311d28c PE |
275 | ptrdiff_t s1, s2; |
276 | register ptrdiff_t i; | |
487282dc | 277 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 278 | |
7074fde6 FP |
279 | if (running_asynch_code) |
280 | save_search_regs (); | |
281 | ||
910c747a | 282 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
34dabdb7 | 283 | set_char_table_extras (BVAR (current_buffer, case_canon_table), 2, |
a098c930 | 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 | ||
908589fd | 329 | val = (i >= 0 ? Qt : Qnil); |
ef887810 | 330 | if (NILP (Vinhibit_changing_match_data) && i >= 0) |
fc30d803 | 331 | { |
fa8ed3e0 RS |
332 | for (i = 0; i < search_regs.num_regs; i++) |
333 | if (search_regs.start[i] >= 0) | |
334 | { | |
335 | search_regs.start[i] | |
336 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
fc30d803 SM |
337 | search_regs.end[i] |
338 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
339 | } | |
340 | /* Set last_thing_searched only when match data is changed. */ | |
ef887810 | 341 | XSETBUFFER (last_thing_searched, current_buffer); |
fc30d803 | 342 | } |
ef887810 | 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 | |
457882c2 PE |
369 | string_match_1 (Lisp_Object regexp, Lisp_Object string, Lisp_Object start, |
370 | bool posix) | |
ca1d1d23 | 371 | { |
d311d28c | 372 | ptrdiff_t val; |
487282dc | 373 | struct re_pattern_buffer *bufp; |
d311d28c PE |
374 | EMACS_INT pos; |
375 | ptrdiff_t pos_byte, i; | |
ca1d1d23 | 376 | |
7074fde6 FP |
377 | if (running_asynch_code) |
378 | save_search_regs (); | |
379 | ||
b7826503 PJ |
380 | CHECK_STRING (regexp); |
381 | CHECK_STRING (string); | |
ca1d1d23 JB |
382 | |
383 | if (NILP (start)) | |
0c8533c6 | 384 | pos = 0, pos_byte = 0; |
ca1d1d23 JB |
385 | else |
386 | { | |
d311d28c | 387 | ptrdiff_t len = SCHARS (string); |
ca1d1d23 | 388 | |
b7826503 | 389 | CHECK_NUMBER (start); |
0c8533c6 RS |
390 | pos = XINT (start); |
391 | if (pos < 0 && -pos <= len) | |
392 | pos = len + pos; | |
393 | else if (0 > pos || pos > len) | |
ca1d1d23 | 394 | args_out_of_range (string, start); |
0c8533c6 | 395 | pos_byte = string_char_to_byte (string, pos); |
ca1d1d23 JB |
396 | } |
397 | ||
910c747a | 398 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
34dabdb7 | 399 | set_char_table_extras (BVAR (current_buffer, case_canon_table), 2, |
a098c930 | 400 | BVAR (current_buffer, case_eqv_table)); |
910c747a | 401 | |
ef887810 RS |
402 | bufp = compile_pattern (regexp, |
403 | (NILP (Vinhibit_changing_match_data) | |
404 | ? &search_regs : NULL), | |
4b4deea2 TT |
405 | (!NILP (BVAR (current_buffer, case_fold_search)) |
406 | ? BVAR (current_buffer, case_canon_table) : Qnil), | |
0c8533c6 RS |
407 | posix, |
408 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 409 | immediate_quit = 1; |
8bb43c28 | 410 | re_match_object = string; |
177c0ea7 | 411 | |
51b59d79 | 412 | val = re_search (bufp, SSDATA (string), |
d5db4077 KR |
413 | SBYTES (string), pos_byte, |
414 | SBYTES (string) - pos_byte, | |
ef887810 RS |
415 | (NILP (Vinhibit_changing_match_data) |
416 | ? &search_regs : NULL)); | |
ca1d1d23 | 417 | immediate_quit = 0; |
ef887810 RS |
418 | |
419 | /* Set last_thing_searched only when match data is changed. */ | |
420 | if (NILP (Vinhibit_changing_match_data)) | |
421 | last_thing_searched = Qt; | |
422 | ||
ca1d1d23 JB |
423 | if (val == -2) |
424 | matcher_overflow (); | |
425 | if (val < 0) return Qnil; | |
0c8533c6 | 426 | |
ef887810 RS |
427 | if (NILP (Vinhibit_changing_match_data)) |
428 | for (i = 0; i < search_regs.num_regs; i++) | |
429 | if (search_regs.start[i] >= 0) | |
430 | { | |
431 | search_regs.start[i] | |
432 | = string_byte_to_char (string, search_regs.start[i]); | |
433 | search_regs.end[i] | |
434 | = string_byte_to_char (string, search_regs.end[i]); | |
435 | } | |
0c8533c6 RS |
436 | |
437 | return make_number (string_byte_to_char (string, val)); | |
ca1d1d23 | 438 | } |
e59a8453 | 439 | |
a7ca3326 | 440 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
8c1a1077 | 441 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
b85acc4b | 442 | Matching ignores case if `case-fold-search' is non-nil. |
8c1a1077 PJ |
443 | If third arg START is non-nil, start search at that index in STRING. |
444 | For index of first char beyond the match, do (match-end 0). | |
445 | `match-end' and `match-beginning' also give indices of substrings | |
2bd2f32d RS |
446 | matched by parenthesis constructs in the pattern. |
447 | ||
448 | You can use the function `match-string' to extract the substrings | |
449 | matched by the parenthesis constructions in REGEXP. */) | |
5842a27b | 450 | (Lisp_Object regexp, Lisp_Object string, Lisp_Object start) |
b819a390 RS |
451 | { |
452 | return string_match_1 (regexp, string, start, 0); | |
453 | } | |
454 | ||
455 | DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0, | |
8c1a1077 PJ |
456 | doc: /* Return index of start of first match for REGEXP in STRING, or nil. |
457 | Find the longest match, in accord with Posix regular expression rules. | |
458 | Case is ignored if `case-fold-search' is non-nil in the current buffer. | |
459 | If third arg START is non-nil, start search at that index in STRING. | |
460 | For index of first char beyond the match, do (match-end 0). | |
461 | `match-end' and `match-beginning' also give indices of substrings | |
462 | matched by parenthesis constructs in the pattern. */) | |
5842a27b | 463 | (Lisp_Object regexp, Lisp_Object string, Lisp_Object start) |
b819a390 RS |
464 | { |
465 | return string_match_1 (regexp, string, start, 1); | |
466 | } | |
467 | ||
e59a8453 RS |
468 | /* Match REGEXP against STRING, searching all of STRING, |
469 | and return the index of the match, or negative on failure. | |
470 | This does not clobber the match data. */ | |
471 | ||
d311d28c | 472 | ptrdiff_t |
971de7fb | 473 | fast_string_match (Lisp_Object regexp, Lisp_Object string) |
e59a8453 | 474 | { |
d311d28c | 475 | ptrdiff_t val; |
487282dc | 476 | struct re_pattern_buffer *bufp; |
e59a8453 | 477 | |
facdc750 RS |
478 | bufp = compile_pattern (regexp, 0, Qnil, |
479 | 0, STRING_MULTIBYTE (string)); | |
e59a8453 | 480 | immediate_quit = 1; |
8bb43c28 | 481 | re_match_object = string; |
177c0ea7 | 482 | |
51b59d79 | 483 | val = re_search (bufp, SSDATA (string), |
d5db4077 KR |
484 | SBYTES (string), 0, |
485 | SBYTES (string), 0); | |
e59a8453 RS |
486 | immediate_quit = 0; |
487 | return val; | |
488 | } | |
5679531d KH |
489 | |
490 | /* Match REGEXP against STRING, searching all of STRING ignoring case, | |
491 | and return the index of the match, or negative on failure. | |
0c8533c6 RS |
492 | This does not clobber the match data. |
493 | We assume that STRING contains single-byte characters. */ | |
5679531d | 494 | |
d311d28c | 495 | ptrdiff_t |
d923b542 DA |
496 | fast_c_string_match_ignore_case (Lisp_Object regexp, |
497 | const char *string, ptrdiff_t len) | |
5679531d | 498 | { |
d311d28c | 499 | ptrdiff_t val; |
5679531d | 500 | struct re_pattern_buffer *bufp; |
5679531d | 501 | |
0c8533c6 | 502 | regexp = string_make_unibyte (regexp); |
b4577c63 | 503 | re_match_object = Qt; |
5679531d | 504 | bufp = compile_pattern (regexp, 0, |
0190922f | 505 | Vascii_canon_table, 0, |
f8bd51c4 | 506 | 0); |
5679531d KH |
507 | immediate_quit = 1; |
508 | val = re_search (bufp, string, len, 0, len, 0); | |
509 | immediate_quit = 0; | |
510 | return val; | |
511 | } | |
be5f4dfb KH |
512 | |
513 | /* Like fast_string_match but ignore case. */ | |
514 | ||
d311d28c | 515 | ptrdiff_t |
971de7fb | 516 | fast_string_match_ignore_case (Lisp_Object regexp, Lisp_Object string) |
be5f4dfb | 517 | { |
d311d28c | 518 | ptrdiff_t val; |
be5f4dfb KH |
519 | struct re_pattern_buffer *bufp; |
520 | ||
0190922f | 521 | bufp = compile_pattern (regexp, 0, Vascii_canon_table, |
be5f4dfb KH |
522 | 0, STRING_MULTIBYTE (string)); |
523 | immediate_quit = 1; | |
524 | re_match_object = string; | |
525 | ||
51b59d79 | 526 | val = re_search (bufp, SSDATA (string), |
be5f4dfb KH |
527 | SBYTES (string), 0, |
528 | SBYTES (string), 0); | |
529 | immediate_quit = 0; | |
530 | return val; | |
531 | } | |
ca1d1d23 | 532 | \f |
a80ce213 | 533 | /* Match REGEXP against the characters after POS to LIMIT, and return |
8b13507a KH |
534 | the number of matched characters. If STRING is non-nil, match |
535 | against the characters in it. In that case, POS and LIMIT are | |
536 | indices into the string. This function doesn't modify the match | |
537 | data. */ | |
538 | ||
d311d28c | 539 | ptrdiff_t |
0063fdb1 DA |
540 | fast_looking_at (Lisp_Object regexp, ptrdiff_t pos, ptrdiff_t pos_byte, |
541 | ptrdiff_t limit, ptrdiff_t limit_byte, Lisp_Object string) | |
8b13507a | 542 | { |
0063fdb1 | 543 | bool multibyte; |
8b13507a KH |
544 | struct re_pattern_buffer *buf; |
545 | unsigned char *p1, *p2; | |
d311d28c PE |
546 | ptrdiff_t s1, s2; |
547 | ptrdiff_t len; | |
78edd3b7 | 548 | |
8b13507a KH |
549 | if (STRINGP (string)) |
550 | { | |
551 | if (pos_byte < 0) | |
552 | pos_byte = string_char_to_byte (string, pos); | |
553 | if (limit_byte < 0) | |
554 | limit_byte = string_char_to_byte (string, limit); | |
555 | p1 = NULL; | |
556 | s1 = 0; | |
557 | p2 = SDATA (string); | |
558 | s2 = SBYTES (string); | |
559 | re_match_object = string; | |
560 | multibyte = STRING_MULTIBYTE (string); | |
561 | } | |
562 | else | |
563 | { | |
564 | if (pos_byte < 0) | |
565 | pos_byte = CHAR_TO_BYTE (pos); | |
566 | if (limit_byte < 0) | |
567 | limit_byte = CHAR_TO_BYTE (limit); | |
568 | pos_byte -= BEGV_BYTE; | |
569 | limit_byte -= BEGV_BYTE; | |
570 | p1 = BEGV_ADDR; | |
571 | s1 = GPT_BYTE - BEGV_BYTE; | |
572 | p2 = GAP_END_ADDR; | |
573 | s2 = ZV_BYTE - GPT_BYTE; | |
574 | if (s1 < 0) | |
575 | { | |
576 | p2 = p1; | |
577 | s2 = ZV_BYTE - BEGV_BYTE; | |
578 | s1 = 0; | |
579 | } | |
580 | if (s2 < 0) | |
581 | { | |
582 | s1 = ZV_BYTE - BEGV_BYTE; | |
583 | s2 = 0; | |
584 | } | |
585 | re_match_object = Qnil; | |
4b4deea2 | 586 | multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
8b13507a KH |
587 | } |
588 | ||
589 | buf = compile_pattern (regexp, 0, Qnil, 0, multibyte); | |
590 | immediate_quit = 1; | |
591 | len = re_match_2 (buf, (char *) p1, s1, (char *) p2, s2, | |
592 | pos_byte, NULL, limit_byte); | |
593 | immediate_quit = 0; | |
594 | ||
595 | return len; | |
596 | } | |
597 | ||
598 | \f | |
9169c321 JB |
599 | /* The newline cache: remembering which sections of text have no newlines. */ |
600 | ||
e30b79c1 | 601 | /* If the user has requested the long scans caching, make sure it's on. |
9169c321 JB |
602 | Otherwise, make sure it's off. |
603 | This is our cheezy way of associating an action with the change of | |
604 | state of a buffer-local variable. */ | |
605 | static void | |
971de7fb | 606 | newline_cache_on_off (struct buffer *buf) |
9169c321 | 607 | { |
e30b79c1 | 608 | if (NILP (BVAR (buf, cache_long_scans))) |
9169c321 JB |
609 | { |
610 | /* It should be off. */ | |
611 | if (buf->newline_cache) | |
612 | { | |
613 | free_region_cache (buf->newline_cache); | |
614 | buf->newline_cache = 0; | |
615 | } | |
616 | } | |
617 | else | |
618 | { | |
619 | /* It should be on. */ | |
620 | if (buf->newline_cache == 0) | |
621 | buf->newline_cache = new_region_cache (); | |
622 | } | |
623 | } | |
624 | ||
625 | \f | |
b5426561 | 626 | /* Search for COUNT newlines between START/START_BYTE and END/END_BYTE. |
9169c321 JB |
627 | |
628 | If COUNT is positive, search forwards; END must be >= START. | |
629 | If COUNT is negative, search backwards for the -COUNTth instance; | |
630 | END must be <= START. | |
631 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
632 | ||
633 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
634 | direction indicated by COUNT. | |
ffd56f97 JB |
635 | |
636 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
a9f2a45f | 637 | position past the COUNTth match. Note that for reverse motion |
5bfe95c9 | 638 | this is not the same as the usual convention for Emacs motion commands. |
ffd56f97 | 639 | |
9169c321 | 640 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
a84b7c53 | 641 | to the number of newlines left unfound, and return END. |
ffd56f97 | 642 | |
2a14a4f1 DA |
643 | If BYTEPOS is not NULL, set *BYTEPOS to the byte position corresponding |
644 | to the returned character position. | |
645 | ||
9fa1de30 | 646 | If ALLOW_QUIT, set immediate_quit. That's good to do |
087a5f81 RS |
647 | except when inside redisplay. */ |
648 | ||
d311d28c | 649 | ptrdiff_t |
b5426561 DA |
650 | find_newline (ptrdiff_t start, ptrdiff_t start_byte, ptrdiff_t end, |
651 | ptrdiff_t end_byte, ptrdiff_t count, ptrdiff_t *shortage, | |
652 | ptrdiff_t *bytepos, bool allow_quit) | |
ca1d1d23 | 653 | { |
9169c321 | 654 | struct region_cache *newline_cache; |
177c0ea7 | 655 | int direction; |
ffd56f97 | 656 | |
9169c321 JB |
657 | if (count > 0) |
658 | { | |
659 | direction = 1; | |
8ca30920 DA |
660 | if (!end) |
661 | end = ZV, end_byte = ZV_BYTE; | |
9169c321 JB |
662 | } |
663 | else | |
664 | { | |
665 | direction = -1; | |
a84b7c53 | 666 | if (!end) |
8ca30920 | 667 | end = BEGV, end_byte = BEGV_BYTE; |
9169c321 | 668 | } |
8ca30920 DA |
669 | if (end_byte == -1) |
670 | end_byte = CHAR_TO_BYTE (end); | |
ffd56f97 | 671 | |
9169c321 JB |
672 | newline_cache_on_off (current_buffer); |
673 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
674 | |
675 | if (shortage != 0) | |
676 | *shortage = 0; | |
677 | ||
087a5f81 | 678 | immediate_quit = allow_quit; |
ca1d1d23 | 679 | |
ffd56f97 | 680 | if (count > 0) |
9169c321 | 681 | while (start != end) |
ca1d1d23 | 682 | { |
9169c321 JB |
683 | /* Our innermost scanning loop is very simple; it doesn't know |
684 | about gaps, buffer ends, or the newline cache. ceiling is | |
685 | the position of the last character before the next such | |
686 | obstacle --- the last character the dumb search loop should | |
687 | examine. */ | |
2a14a4f1 | 688 | ptrdiff_t tem, ceiling_byte = end_byte - 1; |
9169c321 JB |
689 | |
690 | /* If we're looking for a newline, consult the newline cache | |
691 | to see where we can avoid some scanning. */ | |
a84b7c53 | 692 | if (newline_cache) |
9169c321 | 693 | { |
0065d054 | 694 | ptrdiff_t next_change; |
9169c321 JB |
695 | immediate_quit = 0; |
696 | while (region_cache_forward | |
c8b9f1bc EZ |
697 | (current_buffer, newline_cache, start, &next_change)) |
698 | start = next_change; | |
cbe0db0d | 699 | immediate_quit = allow_quit; |
9169c321 | 700 | |
c8b9f1bc EZ |
701 | start_byte = CHAR_TO_BYTE (start); |
702 | ||
fa8ed3e0 RS |
703 | /* START should never be after END. */ |
704 | if (start_byte > ceiling_byte) | |
705 | start_byte = ceiling_byte; | |
9169c321 JB |
706 | |
707 | /* Now the text after start is an unknown region, and | |
708 | next_change is the position of the next known region. */ | |
c8b9f1bc | 709 | ceiling_byte = min (CHAR_TO_BYTE (next_change) - 1, ceiling_byte); |
9169c321 | 710 | } |
b5426561 | 711 | else if (start_byte == -1) |
c8b9f1bc | 712 | start_byte = CHAR_TO_BYTE (start); |
9169c321 JB |
713 | |
714 | /* The dumb loop can only scan text stored in contiguous | |
715 | bytes. BUFFER_CEILING_OF returns the last character | |
716 | position that is contiguous, so the ceiling is the | |
717 | position after that. */ | |
67ce527d KH |
718 | tem = BUFFER_CEILING_OF (start_byte); |
719 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
720 | |
721 | { | |
177c0ea7 | 722 | /* The termination address of the dumb loop. */ |
5389db9e PE |
723 | unsigned char *lim_addr = BYTE_POS_ADDR (ceiling_byte) + 1; |
724 | ptrdiff_t lim_byte = ceiling_byte + 1; | |
9169c321 | 725 | |
5389db9e PE |
726 | /* Nonpositive offsets (relative to LIM_ADDR and LIM_BYTE) |
727 | of the base, the cursor, and the next line. */ | |
728 | ptrdiff_t base = start_byte - lim_byte; | |
729 | ptrdiff_t cursor, next; | |
730 | ||
731 | for (cursor = base; cursor < 0; cursor = next) | |
732 | { | |
9169c321 | 733 | /* The dumb loop. */ |
5389db9e PE |
734 | unsigned char *nl = memchr (lim_addr + cursor, '\n', - cursor); |
735 | next = nl ? nl - lim_addr : 0; | |
9169c321 JB |
736 | |
737 | /* If we're looking for newlines, cache the fact that | |
5389db9e | 738 | this line's region is free of them. */ |
a84b7c53 PE |
739 | if (newline_cache) |
740 | { | |
a84b7c53 | 741 | know_region_cache (current_buffer, newline_cache, |
5389db9e PE |
742 | BYTE_TO_CHAR (lim_byte + cursor), |
743 | BYTE_TO_CHAR (lim_byte + next)); | |
744 | /* know_region_cache can relocate buffer text. */ | |
745 | lim_addr = BYTE_POS_ADDR (ceiling_byte) + 1; | |
a84b7c53 PE |
746 | } |
747 | ||
748 | if (! nl) | |
749 | break; | |
5389db9e | 750 | next++; |
a84b7c53 PE |
751 | |
752 | if (--count == 0) | |
753 | { | |
754 | immediate_quit = 0; | |
2a14a4f1 | 755 | if (bytepos) |
5389db9e PE |
756 | *bytepos = lim_byte + next; |
757 | return BYTE_TO_CHAR (lim_byte + next); | |
85f37d10 | 758 | } |
9169c321 JB |
759 | } |
760 | ||
5389db9e | 761 | start_byte = lim_byte; |
2a14a4f1 | 762 | start = BYTE_TO_CHAR (start_byte); |
9169c321 | 763 | } |
ca1d1d23 JB |
764 | } |
765 | else | |
9169c321 JB |
766 | while (start > end) |
767 | { | |
768 | /* The last character to check before the next obstacle. */ | |
2a14a4f1 | 769 | ptrdiff_t tem, ceiling_byte = end_byte; |
9169c321 JB |
770 | |
771 | /* Consult the newline cache, if appropriate. */ | |
a84b7c53 | 772 | if (newline_cache) |
9169c321 | 773 | { |
0065d054 | 774 | ptrdiff_t next_change; |
9169c321 JB |
775 | immediate_quit = 0; |
776 | while (region_cache_backward | |
c8b9f1bc EZ |
777 | (current_buffer, newline_cache, start, &next_change)) |
778 | start = next_change; | |
cbe0db0d | 779 | immediate_quit = allow_quit; |
9169c321 | 780 | |
c8b9f1bc EZ |
781 | start_byte = CHAR_TO_BYTE (start); |
782 | ||
9169c321 | 783 | /* Start should never be at or before end. */ |
fa8ed3e0 RS |
784 | if (start_byte <= ceiling_byte) |
785 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
786 | |
787 | /* Now the text before start is an unknown region, and | |
788 | next_change is the position of the next known region. */ | |
c8b9f1bc | 789 | ceiling_byte = max (CHAR_TO_BYTE (next_change), ceiling_byte); |
9169c321 | 790 | } |
b5426561 | 791 | else if (start_byte == -1) |
c8b9f1bc | 792 | start_byte = CHAR_TO_BYTE (start); |
9169c321 JB |
793 | |
794 | /* Stop scanning before the gap. */ | |
67ce527d KH |
795 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
796 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
797 | |
798 | { | |
799 | /* The termination address of the dumb loop. */ | |
5389db9e | 800 | unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
9169c321 | 801 | |
5389db9e PE |
802 | /* Offsets (relative to CEILING_ADDR and CEILING_BYTE) of |
803 | the base, the cursor, and the previous line. These | |
804 | offsets are at least -1. */ | |
805 | ptrdiff_t base = start_byte - ceiling_byte; | |
806 | ptrdiff_t cursor, prev; | |
807 | ||
808 | for (cursor = base; 0 < cursor; cursor = prev) | |
9169c321 | 809 | { |
5389db9e PE |
810 | unsigned char *nl = memrchr (ceiling_addr, '\n', cursor); |
811 | prev = nl ? nl - ceiling_addr : -1; | |
9169c321 JB |
812 | |
813 | /* If we're looking for newlines, cache the fact that | |
5389db9e | 814 | this line's region is free of them. */ |
a84b7c53 PE |
815 | if (newline_cache) |
816 | { | |
a84b7c53 | 817 | know_region_cache (current_buffer, newline_cache, |
5389db9e PE |
818 | BYTE_TO_CHAR (ceiling_byte + prev + 1), |
819 | BYTE_TO_CHAR (ceiling_byte + cursor)); | |
820 | /* know_region_cache can relocate buffer text. */ | |
821 | ceiling_addr = BYTE_POS_ADDR (ceiling_byte); | |
a84b7c53 PE |
822 | } |
823 | ||
824 | if (! nl) | |
825 | break; | |
826 | ||
827 | if (++count >= 0) | |
828 | { | |
829 | immediate_quit = 0; | |
2a14a4f1 | 830 | if (bytepos) |
5389db9e PE |
831 | *bytepos = ceiling_byte + prev + 1; |
832 | return BYTE_TO_CHAR (ceiling_byte + prev + 1); | |
85f37d10 | 833 | } |
9169c321 JB |
834 | } |
835 | ||
5389db9e | 836 | start_byte = ceiling_byte; |
2a14a4f1 | 837 | start = BYTE_TO_CHAR (start_byte); |
9169c321 JB |
838 | } |
839 | } | |
840 | ||
ca1d1d23 | 841 | immediate_quit = 0; |
2a14a4f1 | 842 | if (shortage) |
ffd56f97 | 843 | *shortage = count * direction; |
2a14a4f1 DA |
844 | if (bytepos) |
845 | { | |
846 | *bytepos = start_byte == -1 ? CHAR_TO_BYTE (start) : start_byte; | |
847 | eassert (*bytepos == CHAR_TO_BYTE (start)); | |
848 | } | |
9169c321 | 849 | return start; |
ca1d1d23 | 850 | } |
fa8ed3e0 | 851 | \f |
a84b7c53 | 852 | /* Search for COUNT instances of a line boundary. |
fa8ed3e0 RS |
853 | Start at START. If COUNT is negative, search backwards. |
854 | ||
855 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
856 | ||
857 | If we find COUNT instances. we position after (always after, | |
858 | even if scanning backwards) the COUNTth match, and return 0. | |
859 | ||
860 | If we don't find COUNT instances before reaching the end of the | |
861 | buffer (or the beginning, if scanning backwards), we return | |
862 | the number of line boundaries left unfound, and position at | |
863 | the limit we bumped up against. | |
864 | ||
9fa1de30 | 865 | If ALLOW_QUIT, set immediate_quit. That's good to do |
d5d57b92 | 866 | except in special cases. */ |
ca1d1d23 | 867 | |
d2b36813 | 868 | ptrdiff_t |
d311d28c PE |
869 | scan_newline (ptrdiff_t start, ptrdiff_t start_byte, |
870 | ptrdiff_t limit, ptrdiff_t limit_byte, | |
d2b36813 | 871 | ptrdiff_t count, bool allow_quit) |
63fa018d | 872 | { |
d2b36813 | 873 | ptrdiff_t charpos, bytepos, shortage; |
fa8ed3e0 | 874 | |
d2b36813 DA |
875 | charpos = find_newline (start, start_byte, limit, limit_byte, |
876 | count, &shortage, &bytepos, allow_quit); | |
877 | if (shortage) | |
878 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
fa8ed3e0 | 879 | else |
d2b36813 DA |
880 | TEMP_SET_PT_BOTH (charpos, bytepos); |
881 | return shortage; | |
63fa018d RS |
882 | } |
883 | ||
a611149e EZ |
884 | /* Like find_newline, but doesn't allow QUITting and doesn't return |
885 | SHORTAGE. */ | |
d311d28c | 886 | ptrdiff_t |
b5426561 DA |
887 | find_newline_no_quit (ptrdiff_t from, ptrdiff_t frombyte, |
888 | ptrdiff_t cnt, ptrdiff_t *bytepos) | |
ca1d1d23 | 889 | { |
b5426561 | 890 | return find_newline (from, frombyte, 0, -1, cnt, NULL, bytepos, 0); |
9169c321 JB |
891 | } |
892 | ||
a611149e EZ |
893 | /* Like find_newline, but returns position before the newline, not |
894 | after, and only search up to TO. | |
895 | This isn't just find_newline_no_quit (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 896 | |
d311d28c | 897 | ptrdiff_t |
2a14a4f1 DA |
898 | find_before_next_newline (ptrdiff_t from, ptrdiff_t to, |
899 | ptrdiff_t cnt, ptrdiff_t *bytepos) | |
9169c321 | 900 | { |
d311d28c | 901 | ptrdiff_t shortage; |
b5426561 | 902 | ptrdiff_t pos = find_newline (from, -1, to, -1, cnt, &shortage, bytepos, 1); |
9169c321 JB |
903 | |
904 | if (shortage == 0) | |
2a14a4f1 DA |
905 | { |
906 | if (bytepos) | |
907 | DEC_BOTH (pos, *bytepos); | |
908 | else | |
909 | pos--; | |
910 | } | |
9169c321 | 911 | return pos; |
ca1d1d23 JB |
912 | } |
913 | \f | |
ca1d1d23 JB |
914 | /* Subroutines of Lisp buffer search functions. */ |
915 | ||
916 | static Lisp_Object | |
c098fdb8 | 917 | search_command (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, |
457882c2 | 918 | Lisp_Object count, int direction, int RE, bool posix) |
ca1d1d23 | 919 | { |
457882c2 | 920 | EMACS_INT np; |
d311d28c PE |
921 | EMACS_INT lim; |
922 | ptrdiff_t lim_byte; | |
a53e2e89 | 923 | EMACS_INT n = direction; |
ca1d1d23 JB |
924 | |
925 | if (!NILP (count)) | |
926 | { | |
b7826503 | 927 | CHECK_NUMBER (count); |
ca1d1d23 JB |
928 | n *= XINT (count); |
929 | } | |
930 | ||
b7826503 | 931 | CHECK_STRING (string); |
ca1d1d23 | 932 | if (NILP (bound)) |
9f43ad85 RS |
933 | { |
934 | if (n > 0) | |
935 | lim = ZV, lim_byte = ZV_BYTE; | |
936 | else | |
937 | lim = BEGV, lim_byte = BEGV_BYTE; | |
938 | } | |
ca1d1d23 JB |
939 | else |
940 | { | |
b7826503 | 941 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 942 | lim = XINT (bound); |
6ec8bbd2 | 943 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
944 | error ("Invalid search bound (wrong side of point)"); |
945 | if (lim > ZV) | |
9f43ad85 | 946 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 947 | else if (lim < BEGV) |
9f43ad85 | 948 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
949 | else |
950 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
951 | } |
952 | ||
910c747a | 953 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
34dabdb7 | 954 | set_char_table_extras (BVAR (current_buffer, case_canon_table), 2, |
a098c930 | 955 | BVAR (current_buffer, case_eqv_table)); |
910c747a | 956 | |
9f43ad85 | 957 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
4b4deea2 TT |
958 | (!NILP (BVAR (current_buffer, case_fold_search)) |
959 | ? BVAR (current_buffer, case_canon_table) | |
3135e9fd | 960 | : Qnil), |
4b4deea2 TT |
961 | (!NILP (BVAR (current_buffer, case_fold_search)) |
962 | ? BVAR (current_buffer, case_eqv_table) | |
3135e9fd | 963 | : Qnil), |
b819a390 | 964 | posix); |
ca1d1d23 JB |
965 | if (np <= 0) |
966 | { | |
967 | if (NILP (noerror)) | |
06f77a8a KS |
968 | xsignal1 (Qsearch_failed, string); |
969 | ||
ca1d1d23 JB |
970 | if (!EQ (noerror, Qt)) |
971 | { | |
8ca30920 | 972 | eassert (BEGV <= lim && lim <= ZV); |
9f43ad85 | 973 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
974 | return Qnil; |
975 | #if 0 /* This would be clean, but maybe programs depend on | |
976 | a value of nil here. */ | |
481399bf | 977 | np = lim; |
a5f217b8 | 978 | #endif |
ca1d1d23 | 979 | } |
481399bf RS |
980 | else |
981 | return Qnil; | |
ca1d1d23 JB |
982 | } |
983 | ||
8ca30920 | 984 | eassert (BEGV <= np && np <= ZV); |
ca1d1d23 JB |
985 | SET_PT (np); |
986 | ||
987 | return make_number (np); | |
988 | } | |
989 | \f | |
457882c2 | 990 | /* Return true if REGEXP it matches just one constant string. */ |
fa8ed3e0 | 991 | |
457882c2 | 992 | static bool |
971de7fb | 993 | trivial_regexp_p (Lisp_Object regexp) |
b6d6a51c | 994 | { |
d311d28c | 995 | ptrdiff_t len = SBYTES (regexp); |
d5db4077 | 996 | unsigned char *s = SDATA (regexp); |
b6d6a51c KH |
997 | while (--len >= 0) |
998 | { | |
999 | switch (*s++) | |
1000 | { | |
1001 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
1002 | return 0; | |
1003 | case '\\': | |
1004 | if (--len < 0) | |
1005 | return 0; | |
1006 | switch (*s++) | |
1007 | { | |
1008 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
1009 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 1010 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 1011 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 1012 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
1013 | case '6': case '7': case '8': case '9': |
1014 | return 0; | |
1015 | } | |
1016 | } | |
1017 | } | |
1018 | return 1; | |
1019 | } | |
1020 | ||
ca325161 | 1021 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 1022 | starting at position POS and stopping at position LIM, |
b819a390 | 1023 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
1024 | a regular expression if RE is true. |
1025 | ||
1026 | If N is positive, searching is forward and LIM must be greater than POS. | |
1027 | If N is negative, searching is backward and LIM must be less than POS. | |
1028 | ||
facdc750 | 1029 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 1030 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
1031 | (if searching backward) or the end (if searching forward). |
1032 | ||
1033 | POSIX is nonzero if we want full backtracking (POSIX style) | |
1034 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 1035 | |
aff2ce94 RS |
1036 | #define TRANSLATE(out, trt, d) \ |
1037 | do \ | |
1038 | { \ | |
1039 | if (! NILP (trt)) \ | |
1040 | { \ | |
1041 | Lisp_Object temp; \ | |
1042 | temp = Faref (trt, make_number (d)); \ | |
1043 | if (INTEGERP (temp)) \ | |
1044 | out = XINT (temp); \ | |
1045 | else \ | |
1046 | out = d; \ | |
1047 | } \ | |
1048 | else \ | |
1049 | out = d; \ | |
1050 | } \ | |
1051 | while (0) | |
facdc750 | 1052 | |
ef887810 RS |
1053 | /* Only used in search_buffer, to record the end position of the match |
1054 | when searching regexps and SEARCH_REGS should not be changed | |
1055 | (i.e. Vinhibit_changing_match_data is non-nil). */ | |
1056 | static struct re_registers search_regs_1; | |
1057 | ||
e7deaab0 | 1058 | static EMACS_INT |
d311d28c PE |
1059 | search_buffer (Lisp_Object string, ptrdiff_t pos, ptrdiff_t pos_byte, |
1060 | ptrdiff_t lim, ptrdiff_t lim_byte, EMACS_INT n, | |
457882c2 | 1061 | int RE, Lisp_Object trt, Lisp_Object inverse_trt, bool posix) |
ca1d1d23 | 1062 | { |
d311d28c PE |
1063 | ptrdiff_t len = SCHARS (string); |
1064 | ptrdiff_t len_byte = SBYTES (string); | |
1065 | register ptrdiff_t i; | |
ca1d1d23 | 1066 | |
7074fde6 FP |
1067 | if (running_asynch_code) |
1068 | save_search_regs (); | |
1069 | ||
a7e4cdde | 1070 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1071 | /* Null string is found at starting position. */ |
a7e4cdde | 1072 | if (len == 0 || n == 0) |
ca325161 | 1073 | { |
0353b28f | 1074 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1075 | return pos; |
1076 | } | |
3f57a499 | 1077 | |
41a33295 | 1078 | if (RE && !(trivial_regexp_p (string) && NILP (Vsearch_spaces_regexp))) |
ca1d1d23 | 1079 | { |
facdc750 | 1080 | unsigned char *p1, *p2; |
d311d28c | 1081 | ptrdiff_t s1, s2; |
487282dc KH |
1082 | struct re_pattern_buffer *bufp; |
1083 | ||
ef887810 RS |
1084 | bufp = compile_pattern (string, |
1085 | (NILP (Vinhibit_changing_match_data) | |
1086 | ? &search_regs : &search_regs_1), | |
1087 | trt, posix, | |
4b4deea2 | 1088 | !NILP (BVAR (current_buffer, enable_multibyte_characters))); |
ca1d1d23 | 1089 | |
ca1d1d23 JB |
1090 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1091 | because letting this function finish | |
1092 | can take too long. */ | |
1093 | QUIT; /* Do a pending quit right away, | |
1094 | to avoid paradoxical behavior */ | |
1095 | /* Get pointers and sizes of the two strings | |
1096 | that make up the visible portion of the buffer. */ | |
1097 | ||
1098 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1099 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1100 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1101 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1102 | if (s1 < 0) |
1103 | { | |
1104 | p2 = p1; | |
fa8ed3e0 | 1105 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1106 | s1 = 0; |
1107 | } | |
1108 | if (s2 < 0) | |
1109 | { | |
fa8ed3e0 | 1110 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1111 | s2 = 0; |
1112 | } | |
8bb43c28 | 1113 | re_match_object = Qnil; |
177c0ea7 | 1114 | |
ca1d1d23 JB |
1115 | while (n < 0) |
1116 | { | |
d311d28c | 1117 | ptrdiff_t val; |
52c55cc7 | 1118 | |
487282dc | 1119 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1120 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1121 | (NILP (Vinhibit_changing_match_data) |
1122 | ? &search_regs : &search_regs_1), | |
42db823b | 1123 | /* Don't allow match past current point */ |
4996330b | 1124 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1125 | if (val == -2) |
b6d6a51c KH |
1126 | { |
1127 | matcher_overflow (); | |
1128 | } | |
ca1d1d23 JB |
1129 | if (val >= 0) |
1130 | { | |
ef887810 RS |
1131 | if (NILP (Vinhibit_changing_match_data)) |
1132 | { | |
1133 | pos_byte = search_regs.start[0] + BEGV_BYTE; | |
1134 | for (i = 0; i < search_regs.num_regs; i++) | |
1135 | if (search_regs.start[i] >= 0) | |
1136 | { | |
1137 | search_regs.start[i] | |
1138 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1139 | search_regs.end[i] | |
1140 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1141 | } | |
1142 | XSETBUFFER (last_thing_searched, current_buffer); | |
1143 | /* Set pos to the new position. */ | |
1144 | pos = search_regs.start[0]; | |
1145 | } | |
1146 | else | |
1147 | { | |
1148 | pos_byte = search_regs_1.start[0] + BEGV_BYTE; | |
1149 | /* Set pos to the new position. */ | |
1150 | pos = BYTE_TO_CHAR (search_regs_1.start[0] + BEGV_BYTE); | |
1151 | } | |
ca1d1d23 JB |
1152 | } |
1153 | else | |
1154 | { | |
1155 | immediate_quit = 0; | |
1156 | return (n); | |
1157 | } | |
1158 | n++; | |
1159 | } | |
1160 | while (n > 0) | |
1161 | { | |
d311d28c | 1162 | ptrdiff_t val; |
52c55cc7 | 1163 | |
487282dc | 1164 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1165 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1166 | (NILP (Vinhibit_changing_match_data) |
1167 | ? &search_regs : &search_regs_1), | |
4996330b | 1168 | lim_byte - BEGV_BYTE); |
ca1d1d23 | 1169 | if (val == -2) |
b6d6a51c KH |
1170 | { |
1171 | matcher_overflow (); | |
1172 | } | |
ca1d1d23 JB |
1173 | if (val >= 0) |
1174 | { | |
ef887810 RS |
1175 | if (NILP (Vinhibit_changing_match_data)) |
1176 | { | |
1177 | pos_byte = search_regs.end[0] + BEGV_BYTE; | |
1178 | for (i = 0; i < search_regs.num_regs; i++) | |
1179 | if (search_regs.start[i] >= 0) | |
1180 | { | |
1181 | search_regs.start[i] | |
1182 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1183 | search_regs.end[i] | |
1184 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1185 | } | |
1186 | XSETBUFFER (last_thing_searched, current_buffer); | |
1187 | pos = search_regs.end[0]; | |
1188 | } | |
1189 | else | |
1190 | { | |
1191 | pos_byte = search_regs_1.end[0] + BEGV_BYTE; | |
1192 | pos = BYTE_TO_CHAR (search_regs_1.end[0] + BEGV_BYTE); | |
1193 | } | |
ca1d1d23 JB |
1194 | } |
1195 | else | |
1196 | { | |
1197 | immediate_quit = 0; | |
1198 | return (0 - n); | |
1199 | } | |
1200 | n--; | |
1201 | } | |
1202 | immediate_quit = 0; | |
1203 | return (pos); | |
1204 | } | |
1205 | else /* non-RE case */ | |
1206 | { | |
facdc750 | 1207 | unsigned char *raw_pattern, *pat; |
d311d28c PE |
1208 | ptrdiff_t raw_pattern_size; |
1209 | ptrdiff_t raw_pattern_size_byte; | |
facdc750 | 1210 | unsigned char *patbuf; |
0063fdb1 | 1211 | bool multibyte = !NILP (BVAR (current_buffer, enable_multibyte_characters)); |
a967ed62 | 1212 | unsigned char *base_pat; |
49e44e81 KH |
1213 | /* Set to positive if we find a non-ASCII char that need |
1214 | translation. Otherwise set to zero later. */ | |
1215 | int char_base = -1; | |
457882c2 | 1216 | bool boyer_moore_ok = 1; |
facdc750 RS |
1217 | |
1218 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1219 | We must convert PATTERN to match that, or we will not really | |
1220 | find things right. */ | |
1221 | ||
1222 | if (multibyte == STRING_MULTIBYTE (string)) | |
1223 | { | |
51b59d79 | 1224 | raw_pattern = SDATA (string); |
d5db4077 KR |
1225 | raw_pattern_size = SCHARS (string); |
1226 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1227 | } |
1228 | else if (multibyte) | |
1229 | { | |
d5db4077 | 1230 | raw_pattern_size = SCHARS (string); |
facdc750 | 1231 | raw_pattern_size_byte |
d5db4077 | 1232 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1233 | raw_pattern_size); |
38182d90 | 1234 | raw_pattern = alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1235 | copy_text (SDATA (string), raw_pattern, |
1236 | SCHARS (string), 0, 1); | |
facdc750 RS |
1237 | } |
1238 | else | |
1239 | { | |
1240 | /* Converting multibyte to single-byte. | |
1241 | ||
1242 | ??? Perhaps this conversion should be done in a special way | |
1243 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1244 | so that only the multibyte chars which really correspond to | |
1245 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1246 | raw_pattern_size = SCHARS (string); |
1247 | raw_pattern_size_byte = SCHARS (string); | |
38182d90 | 1248 | raw_pattern = alloca (raw_pattern_size + 1); |
d5db4077 KR |
1249 | copy_text (SDATA (string), raw_pattern, |
1250 | SBYTES (string), 1, 0); | |
facdc750 RS |
1251 | } |
1252 | ||
1253 | /* Copy and optionally translate the pattern. */ | |
1254 | len = raw_pattern_size; | |
1255 | len_byte = raw_pattern_size_byte; | |
38182d90 | 1256 | patbuf = alloca (len * MAX_MULTIBYTE_LENGTH); |
facdc750 RS |
1257 | pat = patbuf; |
1258 | base_pat = raw_pattern; | |
1259 | if (multibyte) | |
1260 | { | |
a17ae96a KH |
1261 | /* Fill patbuf by translated characters in STRING while |
1262 | checking if we can use boyer-moore search. If TRT is | |
1263 | non-nil, we can use boyer-moore search only if TRT can be | |
1264 | represented by the byte array of 256 elements. For that, | |
9858f6c3 | 1265 | all non-ASCII case-equivalents of all case-sensitive |
e79d60fe | 1266 | characters in STRING must belong to the same character |
68abf867 KH |
1267 | group (two characters belong to the same group iff their |
1268 | multibyte forms are the same except for the last byte; | |
1269 | i.e. every 64 characters form a group; U+0000..U+003F, | |
e79d60fe | 1270 | U+0040..U+007F, U+0080..U+00BF, ...). */ |
a17ae96a | 1271 | |
facdc750 RS |
1272 | while (--len >= 0) |
1273 | { | |
a17ae96a | 1274 | unsigned char str_base[MAX_MULTIBYTE_LENGTH], *str; |
aff2ce94 | 1275 | int c, translated, inverse; |
a17ae96a | 1276 | int in_charlen, charlen; |
facdc750 RS |
1277 | |
1278 | /* If we got here and the RE flag is set, it's because we're | |
1279 | dealing with a regexp known to be trivial, so the backslash | |
1280 | just quotes the next character. */ | |
1281 | if (RE && *base_pat == '\\') | |
1282 | { | |
1283 | len--; | |
62221f22 | 1284 | raw_pattern_size--; |
facdc750 RS |
1285 | len_byte--; |
1286 | base_pat++; | |
1287 | } | |
1288 | ||
62a6e103 | 1289 | c = STRING_CHAR_AND_LENGTH (base_pat, in_charlen); |
040272ce | 1290 | |
a17ae96a | 1291 | if (NILP (trt)) |
facdc750 | 1292 | { |
a17ae96a KH |
1293 | str = base_pat; |
1294 | charlen = in_charlen; | |
1295 | } | |
1296 | else | |
1297 | { | |
1298 | /* Translate the character. */ | |
1299 | TRANSLATE (translated, trt, c); | |
1300 | charlen = CHAR_STRING (translated, str_base); | |
1301 | str = str_base; | |
1302 | ||
1303 | /* Check if C has any other case-equivalents. */ | |
1304 | TRANSLATE (inverse, inverse_trt, c); | |
1305 | /* If so, check if we can use boyer-moore. */ | |
1306 | if (c != inverse && boyer_moore_ok) | |
1307 | { | |
1308 | /* Check if all equivalents belong to the same | |
1309 | group of characters. Note that the check of C | |
49e44e81 KH |
1310 | itself is done by the last iteration. */ |
1311 | int this_char_base = -1; | |
1312 | ||
1313 | while (boyer_moore_ok) | |
a17ae96a | 1314 | { |
49e44e81 | 1315 | if (ASCII_BYTE_P (inverse)) |
a17ae96a | 1316 | { |
49e44e81 KH |
1317 | if (this_char_base > 0) |
1318 | boyer_moore_ok = 0; | |
1319 | else | |
dafaabd1 | 1320 | this_char_base = 0; |
a17ae96a | 1321 | } |
49e44e81 KH |
1322 | else if (CHAR_BYTE8_P (inverse)) |
1323 | /* Boyer-moore search can't handle a | |
1324 | translation of an eight-bit | |
1325 | character. */ | |
1326 | boyer_moore_ok = 0; | |
1327 | else if (this_char_base < 0) | |
1328 | { | |
1329 | this_char_base = inverse & ~0x3F; | |
1330 | if (char_base < 0) | |
1331 | char_base = this_char_base; | |
dafaabd1 | 1332 | else if (this_char_base != char_base) |
49e44e81 KH |
1333 | boyer_moore_ok = 0; |
1334 | } | |
1335 | else if ((inverse & ~0x3F) != this_char_base) | |
1336 | boyer_moore_ok = 0; | |
a17ae96a KH |
1337 | if (c == inverse) |
1338 | break; | |
1339 | TRANSLATE (inverse, inverse_trt, inverse); | |
1340 | } | |
1341 | } | |
aff2ce94 | 1342 | } |
facdc750 RS |
1343 | |
1344 | /* Store this character into the translated pattern. */ | |
72af86bd | 1345 | memcpy (pat, str, charlen); |
a17ae96a | 1346 | pat += charlen; |
facdc750 RS |
1347 | base_pat += in_charlen; |
1348 | len_byte -= in_charlen; | |
1349 | } | |
dafaabd1 AS |
1350 | |
1351 | /* If char_base is still negative we didn't find any translated | |
1352 | non-ASCII characters. */ | |
1353 | if (char_base < 0) | |
1354 | char_base = 0; | |
facdc750 RS |
1355 | } |
1356 | else | |
1357 | { | |
040272ce | 1358 | /* Unibyte buffer. */ |
a17ae96a | 1359 | char_base = 0; |
facdc750 RS |
1360 | while (--len >= 0) |
1361 | { | |
e8c6e965 | 1362 | int c, translated, inverse; |
facdc750 RS |
1363 | |
1364 | /* If we got here and the RE flag is set, it's because we're | |
1365 | dealing with a regexp known to be trivial, so the backslash | |
1366 | just quotes the next character. */ | |
1367 | if (RE && *base_pat == '\\') | |
1368 | { | |
1369 | len--; | |
0190922f | 1370 | raw_pattern_size--; |
facdc750 RS |
1371 | base_pat++; |
1372 | } | |
1373 | c = *base_pat++; | |
aff2ce94 | 1374 | TRANSLATE (translated, trt, c); |
facdc750 | 1375 | *pat++ = translated; |
e8c6e965 EZ |
1376 | /* Check that none of C's equivalents violates the |
1377 | assumptions of boyer_moore. */ | |
1378 | TRANSLATE (inverse, inverse_trt, c); | |
1379 | while (1) | |
1380 | { | |
1381 | if (inverse >= 0200) | |
1382 | { | |
1383 | boyer_moore_ok = 0; | |
1384 | break; | |
1385 | } | |
1386 | if (c == inverse) | |
1387 | break; | |
1388 | TRANSLATE (inverse, inverse_trt, inverse); | |
1389 | } | |
facdc750 RS |
1390 | } |
1391 | } | |
1392 | ||
1393 | len_byte = pat - patbuf; | |
facdc750 RS |
1394 | pat = base_pat = patbuf; |
1395 | ||
040272ce | 1396 | if (boyer_moore_ok) |
1db5b1ad JB |
1397 | return boyer_moore (n, pat, len_byte, trt, inverse_trt, |
1398 | pos_byte, lim_byte, | |
a17ae96a | 1399 | char_base); |
facdc750 | 1400 | else |
1db5b1ad | 1401 | return simple_search (n, pat, raw_pattern_size, len_byte, trt, |
facdc750 RS |
1402 | pos, pos_byte, lim, lim_byte); |
1403 | } | |
1404 | } | |
1405 | \f | |
1406 | /* Do a simple string search N times for the string PAT, | |
1407 | whose length is LEN/LEN_BYTE, | |
1408 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1409 | TRT is the translation table. | |
f8bd51c4 | 1410 | |
facdc750 RS |
1411 | Return the character position where the match is found. |
1412 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1413 | |
facdc750 RS |
1414 | This kind of search works regardless of what is in PAT and |
1415 | regardless of what is in TRT. It is used in cases where | |
1416 | boyer_moore cannot work. */ | |
1417 | ||
e7deaab0 | 1418 | static EMACS_INT |
c098fdb8 | 1419 | simple_search (EMACS_INT n, unsigned char *pat, |
d311d28c PE |
1420 | ptrdiff_t len, ptrdiff_t len_byte, Lisp_Object trt, |
1421 | ptrdiff_t pos, ptrdiff_t pos_byte, | |
1422 | ptrdiff_t lim, ptrdiff_t lim_byte) | |
facdc750 | 1423 | { |
0063fdb1 DA |
1424 | bool multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
1425 | bool forward = n > 0; | |
49e44e81 KH |
1426 | /* Number of buffer bytes matched. Note that this may be different |
1427 | from len_byte in a multibyte buffer. */ | |
8ce70ed2 | 1428 | ptrdiff_t match_byte = PTRDIFF_MIN; |
facdc750 RS |
1429 | |
1430 | if (lim > pos && multibyte) | |
1431 | while (n > 0) | |
1432 | { | |
1433 | while (1) | |
f8bd51c4 | 1434 | { |
facdc750 | 1435 | /* Try matching at position POS. */ |
d311d28c PE |
1436 | ptrdiff_t this_pos = pos; |
1437 | ptrdiff_t this_pos_byte = pos_byte; | |
1438 | ptrdiff_t this_len = len; | |
facdc750 | 1439 | unsigned char *p = pat; |
a9469498 | 1440 | if (pos + len > lim || pos_byte + len_byte > lim_byte) |
facdc750 RS |
1441 | goto stop; |
1442 | ||
1443 | while (this_len > 0) | |
1444 | { | |
1445 | int charlen, buf_charlen; | |
ab228c24 | 1446 | int pat_ch, buf_ch; |
facdc750 | 1447 | |
62a6e103 | 1448 | pat_ch = STRING_CHAR_AND_LENGTH (p, charlen); |
facdc750 | 1449 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
facdc750 | 1450 | buf_charlen); |
aff2ce94 | 1451 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1452 | |
1453 | if (buf_ch != pat_ch) | |
1454 | break; | |
ab228c24 | 1455 | |
ab228c24 RS |
1456 | this_len--; |
1457 | p += charlen; | |
1458 | ||
1459 | this_pos_byte += buf_charlen; | |
1460 | this_pos++; | |
facdc750 RS |
1461 | } |
1462 | ||
1463 | if (this_len == 0) | |
1464 | { | |
49e44e81 | 1465 | match_byte = this_pos_byte - pos_byte; |
facdc750 | 1466 | pos += len; |
49e44e81 | 1467 | pos_byte += match_byte; |
facdc750 RS |
1468 | break; |
1469 | } | |
1470 | ||
1471 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1472 | } |
facdc750 RS |
1473 | |
1474 | n--; | |
1475 | } | |
1476 | else if (lim > pos) | |
1477 | while (n > 0) | |
1478 | { | |
1479 | while (1) | |
f8bd51c4 | 1480 | { |
facdc750 | 1481 | /* Try matching at position POS. */ |
d311d28c PE |
1482 | ptrdiff_t this_pos = pos; |
1483 | ptrdiff_t this_len = len; | |
facdc750 RS |
1484 | unsigned char *p = pat; |
1485 | ||
1486 | if (pos + len > lim) | |
1487 | goto stop; | |
1488 | ||
1489 | while (this_len > 0) | |
1490 | { | |
1491 | int pat_ch = *p++; | |
1492 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1493 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1494 | |
1495 | if (buf_ch != pat_ch) | |
1496 | break; | |
ab228c24 RS |
1497 | |
1498 | this_len--; | |
1499 | this_pos++; | |
facdc750 RS |
1500 | } |
1501 | ||
1502 | if (this_len == 0) | |
1503 | { | |
49e44e81 | 1504 | match_byte = len; |
facdc750 RS |
1505 | pos += len; |
1506 | break; | |
1507 | } | |
1508 | ||
1509 | pos++; | |
f8bd51c4 | 1510 | } |
facdc750 RS |
1511 | |
1512 | n--; | |
1513 | } | |
1514 | /* Backwards search. */ | |
1515 | else if (lim < pos && multibyte) | |
1516 | while (n < 0) | |
1517 | { | |
1518 | while (1) | |
f8bd51c4 | 1519 | { |
facdc750 | 1520 | /* Try matching at position POS. */ |
d311d28c PE |
1521 | ptrdiff_t this_pos = pos; |
1522 | ptrdiff_t this_pos_byte = pos_byte; | |
1523 | ptrdiff_t this_len = len; | |
c525b3f2 | 1524 | const unsigned char *p = pat + len_byte; |
facdc750 | 1525 | |
8b264ecb | 1526 | if (this_pos - len < lim || (pos_byte - len_byte) < lim_byte) |
facdc750 RS |
1527 | goto stop; |
1528 | ||
1529 | while (this_len > 0) | |
1530 | { | |
ab228c24 | 1531 | int pat_ch, buf_ch; |
facdc750 | 1532 | |
8b264ecb AS |
1533 | DEC_BOTH (this_pos, this_pos_byte); |
1534 | PREV_CHAR_BOUNDARY (p, pat); | |
1535 | pat_ch = STRING_CHAR (p); | |
1536 | buf_ch = STRING_CHAR (BYTE_POS_ADDR (this_pos_byte)); | |
aff2ce94 | 1537 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1538 | |
1539 | if (buf_ch != pat_ch) | |
1540 | break; | |
ab228c24 | 1541 | |
ab228c24 | 1542 | this_len--; |
facdc750 RS |
1543 | } |
1544 | ||
1545 | if (this_len == 0) | |
1546 | { | |
8b264ecb AS |
1547 | match_byte = pos_byte - this_pos_byte; |
1548 | pos = this_pos; | |
1549 | pos_byte = this_pos_byte; | |
facdc750 RS |
1550 | break; |
1551 | } | |
1552 | ||
1553 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1554 | } |
1555 | ||
facdc750 RS |
1556 | n++; |
1557 | } | |
1558 | else if (lim < pos) | |
1559 | while (n < 0) | |
1560 | { | |
1561 | while (1) | |
b6d6a51c | 1562 | { |
facdc750 | 1563 | /* Try matching at position POS. */ |
d311d28c PE |
1564 | ptrdiff_t this_pos = pos - len; |
1565 | ptrdiff_t this_len = len; | |
facdc750 RS |
1566 | unsigned char *p = pat; |
1567 | ||
a9469498 | 1568 | if (this_pos < lim) |
facdc750 RS |
1569 | goto stop; |
1570 | ||
1571 | while (this_len > 0) | |
1572 | { | |
1573 | int pat_ch = *p++; | |
1574 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1575 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1576 | |
1577 | if (buf_ch != pat_ch) | |
1578 | break; | |
ab228c24 RS |
1579 | this_len--; |
1580 | this_pos++; | |
facdc750 RS |
1581 | } |
1582 | ||
1583 | if (this_len == 0) | |
b6d6a51c | 1584 | { |
49e44e81 | 1585 | match_byte = len; |
facdc750 RS |
1586 | pos -= len; |
1587 | break; | |
b6d6a51c | 1588 | } |
facdc750 RS |
1589 | |
1590 | pos--; | |
b6d6a51c | 1591 | } |
facdc750 RS |
1592 | |
1593 | n++; | |
b6d6a51c | 1594 | } |
facdc750 RS |
1595 | |
1596 | stop: | |
1597 | if (n == 0) | |
aff2ce94 | 1598 | { |
8ce70ed2 | 1599 | eassert (match_byte != PTRDIFF_MIN); |
ab228c24 | 1600 | if (forward) |
49e44e81 | 1601 | set_search_regs ((multibyte ? pos_byte : pos) - match_byte, match_byte); |
ab228c24 | 1602 | else |
49e44e81 | 1603 | set_search_regs (multibyte ? pos_byte : pos, match_byte); |
aff2ce94 RS |
1604 | |
1605 | return pos; | |
1606 | } | |
facdc750 RS |
1607 | else if (n > 0) |
1608 | return -n; | |
1609 | else | |
1610 | return n; | |
1611 | } | |
1612 | \f | |
0190922f | 1613 | /* Do Boyer-Moore search N times for the string BASE_PAT, |
1db5b1ad JB |
1614 | whose length is LEN_BYTE, |
1615 | from buffer position POS_BYTE until LIM_BYTE. | |
facdc750 RS |
1616 | DIRECTION says which direction we search in. |
1617 | TRT and INVERSE_TRT are translation tables. | |
0190922f | 1618 | Characters in PAT are already translated by TRT. |
facdc750 | 1619 | |
0190922f KH |
1620 | This kind of search works if all the characters in BASE_PAT that |
1621 | have nontrivial translation are the same aside from the last byte. | |
1622 | This makes it possible to translate just the last byte of a | |
1623 | character, and do so after just a simple test of the context. | |
b2e6b10f | 1624 | CHAR_BASE is nonzero if there is such a non-ASCII character. |
facdc750 RS |
1625 | |
1626 | If that criterion is not satisfied, do not call this function. */ | |
1627 | ||
e7deaab0 | 1628 | static EMACS_INT |
c098fdb8 | 1629 | boyer_moore (EMACS_INT n, unsigned char *base_pat, |
d311d28c | 1630 | ptrdiff_t len_byte, |
d5a3eaaf | 1631 | Lisp_Object trt, Lisp_Object inverse_trt, |
d311d28c | 1632 | ptrdiff_t pos_byte, ptrdiff_t lim_byte, |
1db5b1ad | 1633 | int char_base) |
facdc750 RS |
1634 | { |
1635 | int direction = ((n > 0) ? 1 : -1); | |
d311d28c PE |
1636 | register ptrdiff_t dirlen; |
1637 | ptrdiff_t limit; | |
e7deaab0 | 1638 | int stride_for_teases = 0; |
f4646fff | 1639 | int BM_tab[0400]; |
177c0ea7 | 1640 | register unsigned char *cursor, *p_limit; |
d311d28c | 1641 | register ptrdiff_t i; |
c098fdb8 | 1642 | register int j; |
cb6792d2 | 1643 | unsigned char *pat, *pat_end; |
0063fdb1 | 1644 | bool multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
facdc750 RS |
1645 | |
1646 | unsigned char simple_translate[0400]; | |
0190922f | 1647 | /* These are set to the preceding bytes of a byte to be translated |
49e44e81 | 1648 | if char_base is nonzero. As the maximum byte length of a |
a17ae96a | 1649 | multibyte character is 5, we have to check at most four previous |
0190922f KH |
1650 | bytes. */ |
1651 | int translate_prev_byte1 = 0; | |
1652 | int translate_prev_byte2 = 0; | |
1653 | int translate_prev_byte3 = 0; | |
facdc750 | 1654 | |
f4646fff AS |
1655 | /* The general approach is that we are going to maintain that we know |
1656 | the first (closest to the present position, in whatever direction | |
1657 | we're searching) character that could possibly be the last | |
1658 | (furthest from present position) character of a valid match. We | |
1659 | advance the state of our knowledge by looking at that character | |
1660 | and seeing whether it indeed matches the last character of the | |
1661 | pattern. If it does, we take a closer look. If it does not, we | |
1662 | move our pointer (to putative last characters) as far as is | |
1663 | logically possible. This amount of movement, which I call a | |
1664 | stride, will be the length of the pattern if the actual character | |
1665 | appears nowhere in the pattern, otherwise it will be the distance | |
1666 | from the last occurrence of that character to the end of the | |
1667 | pattern. If the amount is zero we have a possible match. */ | |
1668 | ||
1669 | /* Here we make a "mickey mouse" BM table. The stride of the search | |
1670 | is determined only by the last character of the putative match. | |
1671 | If that character does not match, we will stride the proper | |
1672 | distance to propose a match that superimposes it on the last | |
1673 | instance of a character that matches it (per trt), or misses | |
1674 | it entirely if there is none. */ | |
facdc750 RS |
1675 | |
1676 | dirlen = len_byte * direction; | |
cb6792d2 RS |
1677 | |
1678 | /* Record position after the end of the pattern. */ | |
1679 | pat_end = base_pat + len_byte; | |
1680 | /* BASE_PAT points to a character that we start scanning from. | |
1681 | It is the first character in a forward search, | |
1682 | the last character in a backward search. */ | |
facdc750 | 1683 | if (direction < 0) |
cb6792d2 RS |
1684 | base_pat = pat_end - 1; |
1685 | ||
f4646fff AS |
1686 | /* A character that does not appear in the pattern induces a |
1687 | stride equal to the pattern length. */ | |
1688 | for (i = 0; i < 0400; i++) | |
1689 | BM_tab[i] = dirlen; | |
facdc750 RS |
1690 | |
1691 | /* We use this for translation, instead of TRT itself. | |
1692 | We fill this in to handle the characters that actually | |
1693 | occur in the pattern. Others don't matter anyway! */ | |
facdc750 RS |
1694 | for (i = 0; i < 0400; i++) |
1695 | simple_translate[i] = i; | |
1696 | ||
a17ae96a | 1697 | if (char_base) |
0190922f | 1698 | { |
a17ae96a | 1699 | /* Setup translate_prev_byte1/2/3/4 from CHAR_BASE. Only a |
0190922f | 1700 | byte following them are the target of translation. */ |
0190922f | 1701 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
1f3561e4 | 1702 | int cblen = CHAR_STRING (char_base, str); |
0190922f | 1703 | |
1f3561e4 PE |
1704 | translate_prev_byte1 = str[cblen - 2]; |
1705 | if (cblen > 2) | |
0190922f | 1706 | { |
1f3561e4 PE |
1707 | translate_prev_byte2 = str[cblen - 3]; |
1708 | if (cblen > 3) | |
1f1d9321 | 1709 | translate_prev_byte3 = str[cblen - 4]; |
0190922f KH |
1710 | } |
1711 | } | |
1712 | ||
facdc750 | 1713 | i = 0; |
f4646fff | 1714 | while (i != dirlen) |
facdc750 | 1715 | { |
cb6792d2 | 1716 | unsigned char *ptr = base_pat + i; |
facdc750 | 1717 | i += direction; |
facdc750 | 1718 | if (! NILP (trt)) |
ca1d1d23 | 1719 | { |
49e44e81 KH |
1720 | /* If the byte currently looking at is the last of a |
1721 | character to check case-equivalents, set CH to that | |
1722 | character. An ASCII character and a non-ASCII character | |
1723 | matching with CHAR_BASE are to be checked. */ | |
a17ae96a KH |
1724 | int ch = -1; |
1725 | ||
1726 | if (ASCII_BYTE_P (*ptr) || ! multibyte) | |
1727 | ch = *ptr; | |
49e44e81 | 1728 | else if (char_base |
86dc6ccb | 1729 | && ((pat_end - ptr) == 1 || CHAR_HEAD_P (ptr[1]))) |
ca1d1d23 | 1730 | { |
49e44e81 KH |
1731 | unsigned char *charstart = ptr - 1; |
1732 | ||
1733 | while (! (CHAR_HEAD_P (*charstart))) | |
1734 | charstart--; | |
62a6e103 | 1735 | ch = STRING_CHAR (charstart); |
a17ae96a KH |
1736 | if (char_base != (ch & ~0x3F)) |
1737 | ch = -1; | |
ca1d1d23 | 1738 | } |
facdc750 | 1739 | |
edb7b4dc | 1740 | if (ch >= 0200 && multibyte) |
49e44e81 KH |
1741 | j = (ch & 0x3F) | 0200; |
1742 | else | |
1743 | j = *ptr; | |
1744 | ||
f4646fff | 1745 | if (i == dirlen) |
facdc750 | 1746 | stride_for_teases = BM_tab[j]; |
ab228c24 | 1747 | |
facdc750 | 1748 | BM_tab[j] = dirlen - i; |
1db5b1ad JB |
1749 | /* A translation table is accompanied by its inverse -- see |
1750 | comment following downcase_table for details. */ | |
a17ae96a | 1751 | if (ch >= 0) |
ab228c24 RS |
1752 | { |
1753 | int starting_ch = ch; | |
49e44e81 | 1754 | int starting_j = j; |
a17ae96a | 1755 | |
ab228c24 RS |
1756 | while (1) |
1757 | { | |
1758 | TRANSLATE (ch, inverse_trt, ch); | |
edb7b4dc | 1759 | if (ch >= 0200 && multibyte) |
49e44e81 | 1760 | j = (ch & 0x3F) | 0200; |
ab228c24 | 1761 | else |
a17ae96a | 1762 | j = ch; |
ab228c24 RS |
1763 | |
1764 | /* For all the characters that map into CH, | |
1765 | set up simple_translate to map the last byte | |
1766 | into STARTING_J. */ | |
1767 | simple_translate[j] = starting_j; | |
1768 | if (ch == starting_ch) | |
1769 | break; | |
1770 | BM_tab[j] = dirlen - i; | |
1771 | } | |
1772 | } | |
facdc750 RS |
1773 | } |
1774 | else | |
1775 | { | |
1776 | j = *ptr; | |
1777 | ||
f4646fff | 1778 | if (i == dirlen) |
facdc750 RS |
1779 | stride_for_teases = BM_tab[j]; |
1780 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1781 | } |
f4646fff AS |
1782 | /* stride_for_teases tells how much to stride if we get a |
1783 | match on the far character but are subsequently | |
1784 | disappointed, by recording what the stride would have been | |
1785 | for that character if the last character had been | |
1786 | different. */ | |
facdc750 | 1787 | } |
facdc750 RS |
1788 | pos_byte += dirlen - ((direction > 0) ? direction : 0); |
1789 | /* loop invariant - POS_BYTE points at where last char (first | |
1790 | char if reverse) of pattern would align in a possible match. */ | |
1791 | while (n != 0) | |
1792 | { | |
d311d28c | 1793 | ptrdiff_t tail_end; |
facdc750 RS |
1794 | unsigned char *tail_end_ptr; |
1795 | ||
1796 | /* It's been reported that some (broken) compiler thinks that | |
1797 | Boolean expressions in an arithmetic context are unsigned. | |
1798 | Using an explicit ?1:0 prevents this. */ | |
1799 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1800 | < 0) | |
1801 | return (n * (0 - direction)); | |
1802 | /* First we do the part we can by pointers (maybe nothing) */ | |
1803 | QUIT; | |
1804 | pat = base_pat; | |
1805 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1806 | if (direction > 0) |
1807 | { | |
1808 | limit = BUFFER_CEILING_OF (limit); | |
1809 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1810 | can take on without hitting edge of buffer or the gap. */ | |
1811 | limit = min (limit, pos_byte + 20000); | |
1812 | limit = min (limit, lim_byte - 1); | |
1813 | } | |
1814 | else | |
1815 | { | |
1816 | limit = BUFFER_FLOOR_OF (limit); | |
1817 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1818 | can take on without hitting edge of buffer or the gap. */ | |
1819 | limit = max (limit, pos_byte - 20000); | |
1820 | limit = max (limit, lim_byte); | |
1821 | } | |
facdc750 RS |
1822 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1823 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1824 | ||
1825 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1826 | { |
facdc750 RS |
1827 | unsigned char *p2; |
1828 | ||
1829 | p_limit = BYTE_POS_ADDR (limit); | |
1830 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
f4646fff | 1831 | /* In this loop, pos + cursor - p2 is the surrogate for pos. */ |
facdc750 | 1832 | while (1) /* use one cursor setting as long as i can */ |
ca1d1d23 | 1833 | { |
facdc750 | 1834 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1835 | { |
f4646fff AS |
1836 | while (cursor <= p_limit) |
1837 | { | |
1838 | if (BM_tab[*cursor] == 0) | |
1839 | goto hit; | |
facdc750 | 1840 | cursor += BM_tab[*cursor]; |
f4646fff | 1841 | } |
facdc750 RS |
1842 | } |
1843 | else | |
1844 | { | |
f4646fff AS |
1845 | while (cursor >= p_limit) |
1846 | { | |
1847 | if (BM_tab[*cursor] == 0) | |
1848 | goto hit; | |
facdc750 | 1849 | cursor += BM_tab[*cursor]; |
f4646fff | 1850 | } |
facdc750 | 1851 | } |
f4646fff AS |
1852 | /* If you are here, cursor is beyond the end of the |
1853 | searched region. You fail to match within the | |
1854 | permitted region and would otherwise try a character | |
1855 | beyond that region. */ | |
1856 | break; | |
1857 | ||
1858 | hit: | |
facdc750 RS |
1859 | i = dirlen - direction; |
1860 | if (! NILP (trt)) | |
1861 | { | |
1862 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1863 | { |
facdc750 RS |
1864 | int ch; |
1865 | cursor -= direction; | |
1866 | /* Translate only the last byte of a character. */ | |
1867 | if (! multibyte | |
1868 | || ((cursor == tail_end_ptr | |
1869 | || CHAR_HEAD_P (cursor[1])) | |
1870 | && (CHAR_HEAD_P (cursor[0]) | |
0190922f | 1871 | /* Check if this is the last byte of |
cd1181db | 1872 | a translatable character. */ |
0190922f KH |
1873 | || (translate_prev_byte1 == cursor[-1] |
1874 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1875 | || (translate_prev_byte2 == cursor[-2] | |
1876 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1877 | || (translate_prev_byte3 == cursor[-3])))))))) | |
facdc750 RS |
1878 | ch = simple_translate[*cursor]; |
1879 | else | |
1880 | ch = *cursor; | |
1881 | if (pat[i] != ch) | |
1882 | break; | |
ca1d1d23 | 1883 | } |
facdc750 RS |
1884 | } |
1885 | else | |
1886 | { | |
1887 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1888 | { |
facdc750 RS |
1889 | cursor -= direction; |
1890 | if (pat[i] != *cursor) | |
1891 | break; | |
ca1d1d23 | 1892 | } |
facdc750 RS |
1893 | } |
1894 | cursor += dirlen - i - direction; /* fix cursor */ | |
1895 | if (i + direction == 0) | |
1896 | { | |
d311d28c | 1897 | ptrdiff_t position, start, end; |
0c8533c6 | 1898 | |
facdc750 | 1899 | cursor -= direction; |
1113d9db | 1900 | |
facdc750 RS |
1901 | position = pos_byte + cursor - p2 + ((direction > 0) |
1902 | ? 1 - len_byte : 0); | |
1903 | set_search_regs (position, len_byte); | |
ca325161 | 1904 | |
ef887810 RS |
1905 | if (NILP (Vinhibit_changing_match_data)) |
1906 | { | |
1907 | start = search_regs.start[0]; | |
1908 | end = search_regs.end[0]; | |
1909 | } | |
1910 | else | |
1911 | /* If Vinhibit_changing_match_data is non-nil, | |
1912 | search_regs will not be changed. So let's | |
1913 | compute start and end here. */ | |
1914 | { | |
1915 | start = BYTE_TO_CHAR (position); | |
1916 | end = BYTE_TO_CHAR (position + len_byte); | |
1917 | } | |
1918 | ||
facdc750 RS |
1919 | if ((n -= direction) != 0) |
1920 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1921 | else |
ef887810 | 1922 | return direction > 0 ? end : start; |
ca1d1d23 | 1923 | } |
facdc750 RS |
1924 | else |
1925 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1926 | } |
facdc750 RS |
1927 | pos_byte += cursor - p2; |
1928 | } | |
1929 | else | |
f4646fff AS |
1930 | /* Now we'll pick up a clump that has to be done the hard |
1931 | way because it covers a discontinuity. */ | |
facdc750 RS |
1932 | { |
1933 | limit = ((direction > 0) | |
1934 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1935 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1936 | limit = ((direction > 0) | |
1937 | ? min (limit + len_byte, lim_byte - 1) | |
1938 | : max (limit - len_byte, lim_byte)); | |
1939 | /* LIMIT is now the last value POS_BYTE can have | |
1940 | and still be valid for a possible match. */ | |
1941 | while (1) | |
ca1d1d23 | 1942 | { |
f4646fff AS |
1943 | /* This loop can be coded for space rather than |
1944 | speed because it will usually run only once. | |
1945 | (the reach is at most len + 21, and typically | |
1946 | does not exceed len). */ | |
facdc750 | 1947 | while ((limit - pos_byte) * direction >= 0) |
f4646fff AS |
1948 | { |
1949 | int ch = FETCH_BYTE (pos_byte); | |
1950 | if (BM_tab[ch] == 0) | |
1951 | goto hit2; | |
1952 | pos_byte += BM_tab[ch]; | |
1953 | } | |
1954 | break; /* ran off the end */ | |
1955 | ||
1956 | hit2: | |
1957 | /* Found what might be a match. */ | |
facdc750 RS |
1958 | i = dirlen - direction; |
1959 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1960 | { |
facdc750 RS |
1961 | int ch; |
1962 | unsigned char *ptr; | |
1963 | pos_byte -= direction; | |
1964 | ptr = BYTE_POS_ADDR (pos_byte); | |
1965 | /* Translate only the last byte of a character. */ | |
1966 | if (! multibyte | |
1967 | || ((ptr == tail_end_ptr | |
1968 | || CHAR_HEAD_P (ptr[1])) | |
1969 | && (CHAR_HEAD_P (ptr[0]) | |
0190922f | 1970 | /* Check if this is the last byte of a |
cd1181db | 1971 | translatable character. */ |
0190922f KH |
1972 | || (translate_prev_byte1 == ptr[-1] |
1973 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1974 | || (translate_prev_byte2 == ptr[-2] | |
1975 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1976 | || translate_prev_byte3 == ptr[-3]))))))) | |
facdc750 RS |
1977 | ch = simple_translate[*ptr]; |
1978 | else | |
1979 | ch = *ptr; | |
1980 | if (pat[i] != ch) | |
1981 | break; | |
1982 | } | |
1983 | /* Above loop has moved POS_BYTE part or all the way | |
1984 | back to the first pos (last pos if reverse). | |
1985 | Set it once again at the last (first if reverse) char. */ | |
f4646fff | 1986 | pos_byte += dirlen - i - direction; |
facdc750 RS |
1987 | if (i + direction == 0) |
1988 | { | |
d311d28c | 1989 | ptrdiff_t position, start, end; |
facdc750 | 1990 | pos_byte -= direction; |
1113d9db | 1991 | |
facdc750 | 1992 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
facdc750 | 1993 | set_search_regs (position, len_byte); |
ca325161 | 1994 | |
ef887810 RS |
1995 | if (NILP (Vinhibit_changing_match_data)) |
1996 | { | |
1997 | start = search_regs.start[0]; | |
1998 | end = search_regs.end[0]; | |
1999 | } | |
2000 | else | |
2001 | /* If Vinhibit_changing_match_data is non-nil, | |
2002 | search_regs will not be changed. So let's | |
2003 | compute start and end here. */ | |
2004 | { | |
2005 | start = BYTE_TO_CHAR (position); | |
2006 | end = BYTE_TO_CHAR (position + len_byte); | |
2007 | } | |
2008 | ||
facdc750 RS |
2009 | if ((n -= direction) != 0) |
2010 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 2011 | else |
ef887810 | 2012 | return direction > 0 ? end : start; |
ca1d1d23 | 2013 | } |
facdc750 RS |
2014 | else |
2015 | pos_byte += stride_for_teases; | |
2016 | } | |
2017 | } | |
2018 | /* We have done one clump. Can we continue? */ | |
2019 | if ((lim_byte - pos_byte) * direction < 0) | |
2020 | return ((0 - n) * direction); | |
ca1d1d23 | 2021 | } |
facdc750 | 2022 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 2023 | } |
ca325161 | 2024 | |
fa8ed3e0 | 2025 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
2026 | for the overall match just found in the current buffer. |
2027 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
2028 | |
2029 | static void | |
d311d28c | 2030 | set_search_regs (ptrdiff_t beg_byte, ptrdiff_t nbytes) |
ca325161 | 2031 | { |
d311d28c | 2032 | ptrdiff_t i; |
a7e4cdde | 2033 | |
ef887810 RS |
2034 | if (!NILP (Vinhibit_changing_match_data)) |
2035 | return; | |
2036 | ||
ca325161 RS |
2037 | /* Make sure we have registers in which to store |
2038 | the match position. */ | |
2039 | if (search_regs.num_regs == 0) | |
2040 | { | |
23f86fce DA |
2041 | search_regs.start = xmalloc (2 * sizeof (regoff_t)); |
2042 | search_regs.end = xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 2043 | search_regs.num_regs = 2; |
ca325161 RS |
2044 | } |
2045 | ||
a7e4cdde RS |
2046 | /* Clear out the other registers. */ |
2047 | for (i = 1; i < search_regs.num_regs; i++) | |
2048 | { | |
2049 | search_regs.start[i] = -1; | |
2050 | search_regs.end[i] = -1; | |
2051 | } | |
2052 | ||
fa8ed3e0 RS |
2053 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
2054 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 2055 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 2056 | } |
ca1d1d23 | 2057 | \f |
ca1d1d23 | 2058 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, |
8c1a1077 PJ |
2059 | "MSearch backward: ", |
2060 | doc: /* Search backward from point for STRING. | |
2061 | Set point to the beginning of the occurrence found, and return point. | |
2062 | An optional second argument bounds the search; it is a buffer position. | |
2063 | The match found must not extend before that position. | |
2064 | Optional third argument, if t, means if fail just return nil (no error). | |
2065 | If not nil and not t, position at limit of search and return nil. | |
acc28cb9 CY |
2066 | Optional fourth argument COUNT, if non-nil, means to search for COUNT |
2067 | successive occurrences. If COUNT is negative, search forward, | |
2068 | instead of backward, for -COUNT occurrences. | |
8c1a1077 PJ |
2069 | |
2070 | Search case-sensitivity is determined by the value of the variable | |
2071 | `case-fold-search', which see. | |
2072 | ||
2073 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2074 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2075 | { |
b819a390 | 2076 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2077 | } |
2078 | ||
6af43974 | 2079 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2080 | doc: /* Search forward from point for STRING. |
2081 | Set point to the end of the occurrence found, and return point. | |
2082 | An optional second argument bounds the search; it is a buffer position. | |
d6aacdcd | 2083 | The match found must not extend after that position. A value of nil is |
48740f01 | 2084 | equivalent to (point-max). |
8c1a1077 PJ |
2085 | Optional third argument, if t, means if fail just return nil (no error). |
2086 | If not nil and not t, move to limit of search and return nil. | |
acc28cb9 CY |
2087 | Optional fourth argument COUNT, if non-nil, means to search for COUNT |
2088 | successive occurrences. If COUNT is negative, search backward, | |
2089 | instead of forward, for -COUNT occurrences. | |
8c1a1077 PJ |
2090 | |
2091 | Search case-sensitivity is determined by the value of the variable | |
2092 | `case-fold-search', which see. | |
2093 | ||
2094 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2095 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2096 | { |
b819a390 | 2097 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2098 | } |
2099 | ||
ca1d1d23 | 2100 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, |
8c1a1077 PJ |
2101 | "sRE search backward: ", |
2102 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2103 | Set point to the beginning of the match, and return point. | |
2104 | The match found is the one starting last in the buffer | |
2105 | and yet ending before the origin of the search. | |
2106 | An optional second argument bounds the search; it is a buffer position. | |
2107 | The match found must start at or after that position. | |
2108 | Optional third argument, if t, means if fail just return nil (no error). | |
2109 | If not nil and not t, move to limit of search and return nil. | |
2110 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2111 | |
2112 | Search case-sensitivity is determined by the value of the variable | |
2113 | `case-fold-search', which see. | |
2114 | ||
8c1a1077 PJ |
2115 | See also the functions `match-beginning', `match-end', `match-string', |
2116 | and `replace-match'. */) | |
5842a27b | 2117 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2118 | { |
b819a390 | 2119 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2120 | } |
2121 | ||
2122 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2123 | "sRE search: ", |
2124 | doc: /* Search forward from point for regular expression REGEXP. | |
2125 | Set point to the end of the occurrence found, and return point. | |
2126 | An optional second argument bounds the search; it is a buffer position. | |
2127 | The match found must not extend after that position. | |
2128 | Optional third argument, if t, means if fail just return nil (no error). | |
2129 | If not nil and not t, move to limit of search and return nil. | |
2130 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2131 | |
2132 | Search case-sensitivity is determined by the value of the variable | |
2133 | `case-fold-search', which see. | |
2134 | ||
8c1a1077 PJ |
2135 | See also the functions `match-beginning', `match-end', `match-string', |
2136 | and `replace-match'. */) | |
5842a27b | 2137 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2138 | { |
b819a390 RS |
2139 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2140 | } | |
2141 | ||
2142 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2143 | "sPosix search backward: ", |
2144 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2145 | Find the longest match in accord with Posix regular expression rules. | |
2146 | Set point to the beginning of the match, and return point. | |
2147 | The match found is the one starting last in the buffer | |
2148 | and yet ending before the origin of the search. | |
2149 | An optional second argument bounds the search; it is a buffer position. | |
2150 | The match found must start at or after that position. | |
2151 | Optional third argument, if t, means if fail just return nil (no error). | |
2152 | If not nil and not t, move to limit of search and return nil. | |
2153 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2154 | |
2155 | Search case-sensitivity is determined by the value of the variable | |
2156 | `case-fold-search', which see. | |
2157 | ||
8c1a1077 PJ |
2158 | See also the functions `match-beginning', `match-end', `match-string', |
2159 | and `replace-match'. */) | |
5842a27b | 2160 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2161 | { |
2162 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2163 | } | |
2164 | ||
2165 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2166 | "sPosix search: ", |
2167 | doc: /* Search forward from point for regular expression REGEXP. | |
2168 | Find the longest match in accord with Posix regular expression rules. | |
2169 | Set point to the end of the occurrence found, and return point. | |
2170 | An optional second argument bounds the search; it is a buffer position. | |
2171 | The match found must not extend after that position. | |
2172 | Optional third argument, if t, means if fail just return nil (no error). | |
2173 | If not nil and not t, move to limit of search and return nil. | |
2174 | Optional fourth argument is repeat count--search for successive occurrences. | |
49080b10 LMI |
2175 | |
2176 | Search case-sensitivity is determined by the value of the variable | |
2177 | `case-fold-search', which see. | |
2178 | ||
8c1a1077 PJ |
2179 | See also the functions `match-beginning', `match-end', `match-string', |
2180 | and `replace-match'. */) | |
5842a27b | 2181 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2182 | { |
2183 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2184 | } |
2185 | \f | |
d7a5ad5f | 2186 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2187 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2188 | Leave point at the end of the replacement text. |
2189 | ||
c88b867f CY |
2190 | If optional second arg FIXEDCASE is non-nil, do not alter the case of |
2191 | the replacement text. Otherwise, maybe capitalize the whole text, or | |
2192 | maybe just word initials, based on the replaced text. If the replaced | |
2193 | text has only capital letters and has at least one multiletter word, | |
2194 | convert NEWTEXT to all caps. Otherwise if all words are capitalized | |
2195 | in the replaced text, capitalize each word in NEWTEXT. | |
2196 | ||
2197 | If optional third arg LITERAL is non-nil, insert NEWTEXT literally. | |
8c1a1077 PJ |
2198 | Otherwise treat `\\' as special: |
2199 | `\\&' in NEWTEXT means substitute original matched text. | |
2200 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2201 | If Nth parens didn't match, substitute nothing. | |
2202 | `\\\\' means insert one `\\'. | |
d32e47af LM |
2203 | `\\?' is treated literally |
2204 | (for compatibility with `query-replace-regexp'). | |
2205 | Any other character following `\\' signals an error. | |
4dd0c271 RS |
2206 | Case conversion does not apply to these substitutions. |
2207 | ||
c88b867f CY |
2208 | If optional fourth argument STRING is non-nil, it should be a string |
2209 | to act on; this should be the string on which the previous match was | |
2210 | done via `string-match'. In this case, `replace-match' creates and | |
2211 | returns a new string, made by copying STRING and replacing the part of | |
2212 | STRING that was matched (the original STRING itself is not altered). | |
8c1a1077 PJ |
2213 | |
2214 | The optional fifth argument SUBEXP specifies a subexpression; | |
2215 | it says to replace just that subexpression with NEWTEXT, | |
2216 | rather than replacing the entire matched text. | |
2217 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2218 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2219 | NEWTEXT in place of subexp N. | |
2220 | This is useful only after a regular expression search or match, | |
2221 | since only regular expressions have distinguished subexpressions. */) | |
5842a27b | 2222 | (Lisp_Object newtext, Lisp_Object fixedcase, Lisp_Object literal, Lisp_Object string, Lisp_Object subexp) |
ca1d1d23 JB |
2223 | { |
2224 | enum { nochange, all_caps, cap_initial } case_action; | |
457882c2 PE |
2225 | ptrdiff_t pos, pos_byte; |
2226 | bool some_multiletter_word; | |
2227 | bool some_lowercase; | |
2228 | bool some_uppercase; | |
2229 | bool some_nonuppercase_initial; | |
2230 | int c, prevc; | |
a0efffc8 | 2231 | ptrdiff_t sub; |
d311d28c | 2232 | ptrdiff_t opoint, newpoint; |
ca1d1d23 | 2233 | |
b7826503 | 2234 | CHECK_STRING (newtext); |
ca1d1d23 | 2235 | |
080c45fd | 2236 | if (! NILP (string)) |
b7826503 | 2237 | CHECK_STRING (string); |
080c45fd | 2238 | |
ca1d1d23 JB |
2239 | case_action = nochange; /* We tried an initialization */ |
2240 | /* but some C compilers blew it */ | |
4746118a JB |
2241 | |
2242 | if (search_regs.num_regs <= 0) | |
d72cdbfc | 2243 | error ("`replace-match' called before any match found"); |
4746118a | 2244 | |
d7a5ad5f RS |
2245 | if (NILP (subexp)) |
2246 | sub = 0; | |
2247 | else | |
2248 | { | |
b7826503 | 2249 | CHECK_NUMBER (subexp); |
a0efffc8 | 2250 | if (! (0 <= XINT (subexp) && XINT (subexp) < search_regs.num_regs)) |
d7a5ad5f | 2251 | args_out_of_range (subexp, make_number (search_regs.num_regs)); |
a0efffc8 | 2252 | sub = XINT (subexp); |
d7a5ad5f RS |
2253 | } |
2254 | ||
080c45fd RS |
2255 | if (NILP (string)) |
2256 | { | |
d7a5ad5f RS |
2257 | if (search_regs.start[sub] < BEGV |
2258 | || search_regs.start[sub] > search_regs.end[sub] | |
2259 | || search_regs.end[sub] > ZV) | |
2260 | args_out_of_range (make_number (search_regs.start[sub]), | |
2261 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2262 | } |
2263 | else | |
2264 | { | |
d7a5ad5f RS |
2265 | if (search_regs.start[sub] < 0 |
2266 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2267 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2268 | args_out_of_range (make_number (search_regs.start[sub]), |
2269 | make_number (search_regs.end[sub])); | |
080c45fd | 2270 | } |
ca1d1d23 JB |
2271 | |
2272 | if (NILP (fixedcase)) | |
2273 | { | |
2274 | /* Decide how to casify by examining the matched text. */ | |
d311d28c | 2275 | ptrdiff_t last; |
ca1d1d23 | 2276 | |
ac3b28b1 KH |
2277 | pos = search_regs.start[sub]; |
2278 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2279 | |
2280 | if (NILP (string)) | |
ac3b28b1 | 2281 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2282 | else |
ac3b28b1 | 2283 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2284 | |
ca1d1d23 JB |
2285 | prevc = '\n'; |
2286 | case_action = all_caps; | |
2287 | ||
2288 | /* some_multiletter_word is set nonzero if any original word | |
2289 | is more than one letter long. */ | |
2290 | some_multiletter_word = 0; | |
97832bd0 | 2291 | some_lowercase = 0; |
208767c3 | 2292 | some_nonuppercase_initial = 0; |
73dc8771 | 2293 | some_uppercase = 0; |
ca1d1d23 | 2294 | |
ac3b28b1 | 2295 | while (pos < last) |
ca1d1d23 | 2296 | { |
080c45fd | 2297 | if (NILP (string)) |
ac3b28b1 | 2298 | { |
93daa011 | 2299 | c = FETCH_CHAR_AS_MULTIBYTE (pos_byte); |
ac3b28b1 KH |
2300 | INC_BOTH (pos, pos_byte); |
2301 | } | |
080c45fd | 2302 | else |
93daa011 | 2303 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2304 | |
5da9919f | 2305 | if (lowercasep (c)) |
ca1d1d23 JB |
2306 | { |
2307 | /* Cannot be all caps if any original char is lower case */ | |
2308 | ||
97832bd0 | 2309 | some_lowercase = 1; |
ca1d1d23 | 2310 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2311 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2312 | else |
2313 | some_multiletter_word = 1; | |
2314 | } | |
5da9919f | 2315 | else if (uppercasep (c)) |
ca1d1d23 | 2316 | { |
73dc8771 | 2317 | some_uppercase = 1; |
97832bd0 | 2318 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2319 | ; |
97832bd0 | 2320 | else |
ca1d1d23 JB |
2321 | some_multiletter_word = 1; |
2322 | } | |
208767c3 RS |
2323 | else |
2324 | { | |
2325 | /* If the initial is a caseless word constituent, | |
2326 | treat that like a lowercase initial. */ | |
2327 | if (SYNTAX (prevc) != Sword) | |
2328 | some_nonuppercase_initial = 1; | |
2329 | } | |
ca1d1d23 JB |
2330 | |
2331 | prevc = c; | |
2332 | } | |
2333 | ||
97832bd0 RS |
2334 | /* Convert to all caps if the old text is all caps |
2335 | and has at least one multiletter word. */ | |
2336 | if (! some_lowercase && some_multiletter_word) | |
2337 | case_action = all_caps; | |
c4d460ce | 2338 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2339 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2340 | case_action = cap_initial; |
208767c3 | 2341 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2342 | /* Should x -> yz, operating on X, give Yz or YZ? |
2343 | We'll assume the latter. */ | |
2344 | case_action = all_caps; | |
97832bd0 RS |
2345 | else |
2346 | case_action = nochange; | |
ca1d1d23 JB |
2347 | } |
2348 | ||
080c45fd RS |
2349 | /* Do replacement in a string. */ |
2350 | if (!NILP (string)) | |
2351 | { | |
2352 | Lisp_Object before, after; | |
2353 | ||
2354 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2355 | make_number (search_regs.start[sub])); |
2356 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2357 | |
636a5e28 RS |
2358 | /* Substitute parts of the match into NEWTEXT |
2359 | if desired. */ | |
080c45fd RS |
2360 | if (NILP (literal)) |
2361 | { | |
d311d28c PE |
2362 | ptrdiff_t lastpos = 0; |
2363 | ptrdiff_t lastpos_byte = 0; | |
080c45fd RS |
2364 | /* We build up the substituted string in ACCUM. */ |
2365 | Lisp_Object accum; | |
2366 | Lisp_Object middle; | |
d311d28c | 2367 | ptrdiff_t length = SBYTES (newtext); |
080c45fd RS |
2368 | |
2369 | accum = Qnil; | |
2370 | ||
ac3b28b1 | 2371 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd | 2372 | { |
d311d28c PE |
2373 | ptrdiff_t substart = -1; |
2374 | ptrdiff_t subend = 0; | |
457882c2 | 2375 | bool delbackslash = 0; |
080c45fd | 2376 | |
0c8533c6 RS |
2377 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2378 | ||
080c45fd RS |
2379 | if (c == '\\') |
2380 | { | |
0c8533c6 | 2381 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2382 | |
080c45fd RS |
2383 | if (c == '&') |
2384 | { | |
d7a5ad5f RS |
2385 | substart = search_regs.start[sub]; |
2386 | subend = search_regs.end[sub]; | |
080c45fd | 2387 | } |
76fc1ea2 | 2388 | else if (c >= '1' && c <= '9') |
080c45fd | 2389 | { |
d311d28c | 2390 | if (c - '0' < search_regs.num_regs |
908589fd | 2391 | && search_regs.start[c - '0'] >= 0) |
080c45fd RS |
2392 | { |
2393 | substart = search_regs.start[c - '0']; | |
2394 | subend = search_regs.end[c - '0']; | |
2395 | } | |
76fc1ea2 KH |
2396 | else |
2397 | { | |
2398 | /* If that subexp did not match, | |
2399 | replace \\N with nothing. */ | |
2400 | substart = 0; | |
2401 | subend = 0; | |
2402 | } | |
080c45fd | 2403 | } |
1e79ec24 KH |
2404 | else if (c == '\\') |
2405 | delbackslash = 1; | |
d32e47af | 2406 | else if (c != '?') |
636a5e28 | 2407 | error ("Invalid use of `\\' in replacement text"); |
080c45fd RS |
2408 | } |
2409 | if (substart >= 0) | |
2410 | { | |
d131e79c RS |
2411 | if (pos - 2 != lastpos) |
2412 | middle = substring_both (newtext, lastpos, | |
2413 | lastpos_byte, | |
2414 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2415 | else |
2416 | middle = Qnil; | |
2417 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2418 | Fsubstring (string, |
2419 | make_number (substart), | |
080c45fd RS |
2420 | make_number (subend))); |
2421 | lastpos = pos; | |
0c8533c6 | 2422 | lastpos_byte = pos_byte; |
080c45fd | 2423 | } |
1e79ec24 KH |
2424 | else if (delbackslash) |
2425 | { | |
d131e79c RS |
2426 | middle = substring_both (newtext, lastpos, |
2427 | lastpos_byte, | |
2428 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2429 | |
1e79ec24 KH |
2430 | accum = concat2 (accum, middle); |
2431 | lastpos = pos; | |
0c8533c6 | 2432 | lastpos_byte = pos_byte; |
1e79ec24 | 2433 | } |
080c45fd RS |
2434 | } |
2435 | ||
d131e79c RS |
2436 | if (pos != lastpos) |
2437 | middle = substring_both (newtext, lastpos, | |
2438 | lastpos_byte, | |
0c8533c6 | 2439 | pos, pos_byte); |
080c45fd RS |
2440 | else |
2441 | middle = Qnil; | |
2442 | ||
2443 | newtext = concat2 (accum, middle); | |
2444 | } | |
2445 | ||
636a5e28 | 2446 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2447 | if (case_action == all_caps) |
2448 | newtext = Fupcase (newtext); | |
2449 | else if (case_action == cap_initial) | |
2b2eead9 | 2450 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2451 | |
2452 | return concat3 (before, newtext, after); | |
2453 | } | |
2454 | ||
09c4719e | 2455 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2456 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2457 | opoint = PT - ZV; |
9160906f RS |
2458 | else if (PT > search_regs.start[sub]) |
2459 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2460 | else |
2461 | opoint = PT; | |
2462 | ||
886ed6ec RS |
2463 | /* If we want non-literal replacement, |
2464 | perform substitution on the replacement string. */ | |
2465 | if (NILP (literal)) | |
ca1d1d23 | 2466 | { |
0065d054 | 2467 | ptrdiff_t length = SBYTES (newtext); |
68e69fbd | 2468 | unsigned char *substed; |
0065d054 | 2469 | ptrdiff_t substed_alloc_size, substed_len; |
0063fdb1 DA |
2470 | bool buf_multibyte = !NILP (BVAR (current_buffer, enable_multibyte_characters)); |
2471 | bool str_multibyte = STRING_MULTIBYTE (newtext); | |
457882c2 | 2472 | bool really_changed = 0; |
3bc25e52 | 2473 | |
7216e43b PE |
2474 | substed_alloc_size = (length <= (STRING_BYTES_BOUND - 100) / 2 |
2475 | ? length * 2 + 100 | |
2476 | : STRING_BYTES_BOUND); | |
23f86fce | 2477 | substed = xmalloc (substed_alloc_size); |
68e69fbd RS |
2478 | substed_len = 0; |
2479 | ||
3bc25e52 KH |
2480 | /* Go thru NEWTEXT, producing the actual text to insert in |
2481 | SUBSTED while adjusting multibyteness to that of the current | |
2482 | buffer. */ | |
ca1d1d23 | 2483 | |
ac3b28b1 | 2484 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2485 | { |
68e69fbd | 2486 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
8ea90aa3 | 2487 | const unsigned char *add_stuff = NULL; |
0065d054 | 2488 | ptrdiff_t add_len = 0; |
a0efffc8 | 2489 | ptrdiff_t idx = -1; |
9a76659d | 2490 | |
3bc25e52 KH |
2491 | if (str_multibyte) |
2492 | { | |
eb99a8dd | 2493 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 | 2494 | if (!buf_multibyte) |
461c2ab9 | 2495 | c = multibyte_char_to_unibyte (c); |
3bc25e52 KH |
2496 | } |
2497 | else | |
2498 | { | |
2499 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2500 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2501 | if (buf_multibyte) |
4c0354d7 | 2502 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 | 2503 | } |
ac3b28b1 | 2504 | |
68e69fbd RS |
2505 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2506 | or set IDX to a match index, which means put that part | |
2507 | of the buffer text into SUBSTED. */ | |
2508 | ||
ca1d1d23 JB |
2509 | if (c == '\\') |
2510 | { | |
886ed6ec RS |
2511 | really_changed = 1; |
2512 | ||
3bc25e52 KH |
2513 | if (str_multibyte) |
2514 | { | |
eb99a8dd KH |
2515 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2516 | pos, pos_byte); | |
071ce769 | 2517 | if (!buf_multibyte && !ASCII_CHAR_P (c)) |
461c2ab9 | 2518 | c = multibyte_char_to_unibyte (c); |
3bc25e52 KH |
2519 | } |
2520 | else | |
2521 | { | |
d5db4077 | 2522 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2523 | if (buf_multibyte) |
4c0354d7 | 2524 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 KH |
2525 | } |
2526 | ||
ca1d1d23 | 2527 | if (c == '&') |
68e69fbd | 2528 | idx = sub; |
d311d28c | 2529 | else if (c >= '1' && c <= '9' && c - '0' < search_regs.num_regs) |
ca1d1d23 JB |
2530 | { |
2531 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2532 | idx = c - '0'; |
ca1d1d23 | 2533 | } |
636a5e28 | 2534 | else if (c == '\\') |
a7e979a4 | 2535 | add_len = 1, add_stuff = (unsigned char *) "\\"; |
636a5e28 | 2536 | else |
3bc25e52 KH |
2537 | { |
2538 | xfree (substed); | |
2539 | error ("Invalid use of `\\' in replacement text"); | |
2540 | } | |
ca1d1d23 JB |
2541 | } |
2542 | else | |
68e69fbd RS |
2543 | { |
2544 | add_len = CHAR_STRING (c, str); | |
2545 | add_stuff = str; | |
2546 | } | |
2547 | ||
2548 | /* If we want to copy part of a previous match, | |
2549 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2550 | if (idx >= 0) | |
2551 | { | |
0065d054 | 2552 | ptrdiff_t begbyte = CHAR_TO_BYTE (search_regs.start[idx]); |
68e69fbd RS |
2553 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; |
2554 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
13002885 | 2555 | move_gap_both (search_regs.start[idx], begbyte); |
68e69fbd RS |
2556 | add_stuff = BYTE_POS_ADDR (begbyte); |
2557 | } | |
2558 | ||
2559 | /* Now the stuff we want to add to SUBSTED | |
2560 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2561 | ||
2562 | /* Make sure SUBSTED is big enough. */ | |
0065d054 PE |
2563 | if (substed_alloc_size - substed_len < add_len) |
2564 | substed = | |
2565 | xpalloc (substed, &substed_alloc_size, | |
2566 | add_len - (substed_alloc_size - substed_len), | |
2567 | STRING_BYTES_BOUND, 1); | |
68e69fbd RS |
2568 | |
2569 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2570 | if (add_stuff) |
2571 | { | |
72af86bd | 2572 | memcpy (substed + substed_len, add_stuff, add_len); |
f8ce8a0d GM |
2573 | substed_len += add_len; |
2574 | } | |
ca1d1d23 | 2575 | } |
68e69fbd | 2576 | |
886ed6ec | 2577 | if (really_changed) |
76fc1ea2 KH |
2578 | { |
2579 | if (buf_multibyte) | |
2580 | { | |
d311d28c | 2581 | ptrdiff_t nchars = |
c098fdb8 | 2582 | multibyte_chars_in_text (substed, substed_len); |
68e69fbd | 2583 | |
a7e979a4 PE |
2584 | newtext = make_multibyte_string ((char *) substed, nchars, |
2585 | substed_len); | |
76fc1ea2 KH |
2586 | } |
2587 | else | |
a7e979a4 | 2588 | newtext = make_unibyte_string ((char *) substed, substed_len); |
76fc1ea2 | 2589 | } |
68e69fbd | 2590 | xfree (substed); |
ca1d1d23 JB |
2591 | } |
2592 | ||
886ed6ec RS |
2593 | /* Replace the old text with the new in the cleanest possible way. */ |
2594 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2595 | newtext, 1, 0, 1); | |
d5db4077 | 2596 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2597 | |
2598 | if (case_action == all_caps) | |
886ed6ec RS |
2599 | Fupcase_region (make_number (search_regs.start[sub]), |
2600 | make_number (newpoint)); | |
ca1d1d23 | 2601 | else if (case_action == cap_initial) |
886ed6ec RS |
2602 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2603 | make_number (newpoint)); | |
3e18eecf | 2604 | |
98e942e0 RS |
2605 | /* Adjust search data for this change. */ |
2606 | { | |
d311d28c PE |
2607 | ptrdiff_t oldend = search_regs.end[sub]; |
2608 | ptrdiff_t oldstart = search_regs.start[sub]; | |
2609 | ptrdiff_t change = newpoint - search_regs.end[sub]; | |
2610 | ptrdiff_t i; | |
98e942e0 RS |
2611 | |
2612 | for (i = 0; i < search_regs.num_regs; i++) | |
2613 | { | |
41c01205 | 2614 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2615 | search_regs.start[i] += change; |
41c01205 DK |
2616 | else if (search_regs.start[i] > oldstart) |
2617 | search_regs.start[i] = oldstart; | |
2618 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2619 | search_regs.end[i] += change; |
41c01205 DK |
2620 | else if (search_regs.end[i] > oldstart) |
2621 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2622 | } |
2623 | } | |
2624 | ||
b0eba991 | 2625 | /* Put point back where it was in the text. */ |
8d808a65 | 2626 | if (opoint <= 0) |
fa8ed3e0 | 2627 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2628 | else |
fa8ed3e0 | 2629 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2630 | |
2631 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2632 | move_if_not_intangible (newpoint); |
177c0ea7 | 2633 | |
ca1d1d23 JB |
2634 | return Qnil; |
2635 | } | |
2636 | \f | |
2637 | static Lisp_Object | |
457882c2 | 2638 | match_limit (Lisp_Object num, bool beginningp) |
ca1d1d23 | 2639 | { |
a0efffc8 | 2640 | EMACS_INT n; |
ca1d1d23 | 2641 | |
b7826503 | 2642 | CHECK_NUMBER (num); |
ca1d1d23 | 2643 | n = XINT (num); |
f90a5bf5 | 2644 | if (n < 0) |
bd2cbd56 | 2645 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2646 | if (search_regs.num_regs <= 0) |
2647 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2648 | if (n >= search_regs.num_regs |
4746118a | 2649 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2650 | return Qnil; |
2651 | return (make_number ((beginningp) ? search_regs.start[n] | |
2652 | : search_regs.end[n])); | |
2653 | } | |
2654 | ||
a7ca3326 | 2655 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, |
8c1a1077 PJ |
2656 | doc: /* Return position of start of text matched by last search. |
2657 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2658 | regexp. | |
2659 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2660 | SUBEXP pairs. | |
2661 | Zero means the entire text matched by the whole regexp or whole string. */) | |
5842a27b | 2662 | (Lisp_Object subexp) |
ca1d1d23 | 2663 | { |
5806161b | 2664 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2665 | } |
2666 | ||
a7ca3326 | 2667 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, |
8c1a1077 PJ |
2668 | doc: /* Return position of end of text matched by last search. |
2669 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2670 | regexp. | |
2671 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2672 | SUBEXP pairs. | |
2673 | Zero means the entire text matched by the whole regexp or whole string. */) | |
5842a27b | 2674 | (Lisp_Object subexp) |
ca1d1d23 | 2675 | { |
5806161b | 2676 | return match_limit (subexp, 0); |
177c0ea7 | 2677 | } |
ca1d1d23 | 2678 | |
a7ca3326 | 2679 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 3, 0, |
8c1a1077 PJ |
2680 | doc: /* Return a list containing all info on what the last search matched. |
2681 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2682 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2683 | if the last match was on a buffer; integers or nil if a string was matched. | |
febb8aba | 2684 | Use `set-match-data' to reinstate the data in this list. |
8c1a1077 | 2685 | |
41c01205 DK |
2686 | If INTEGERS (the optional first argument) is non-nil, always use |
2687 | integers \(rather than markers) to represent buffer positions. In | |
2688 | this case, and if the last match was in a buffer, the buffer will get | |
2689 | stored as one additional element at the end of the list. | |
2690 | ||
abd0071c KS |
2691 | If REUSE is a list, reuse it as part of the value. If REUSE is long |
2692 | enough to hold all the values, and if INTEGERS is non-nil, no consing | |
2693 | is done. | |
2694 | ||
2695 | If optional third arg RESEAT is non-nil, any previous markers on the | |
2696 | REUSE list will be modified to point to nowhere. | |
2697 | ||
140a6b7e | 2698 | Return value is undefined if the last search failed. */) |
5842a27b | 2699 | (Lisp_Object integers, Lisp_Object reuse, Lisp_Object reseat) |
ca1d1d23 | 2700 | { |
56256c2a | 2701 | Lisp_Object tail, prev; |
4746118a | 2702 | Lisp_Object *data; |
d311d28c | 2703 | ptrdiff_t i, len; |
ca1d1d23 | 2704 | |
abd0071c KS |
2705 | if (!NILP (reseat)) |
2706 | for (tail = reuse; CONSP (tail); tail = XCDR (tail)) | |
2707 | if (MARKERP (XCAR (tail))) | |
2708 | { | |
51f10faa | 2709 | unchain_marker (XMARKER (XCAR (tail))); |
abd0071c KS |
2710 | XSETCAR (tail, Qnil); |
2711 | } | |
2712 | ||
daa37602 | 2713 | if (NILP (last_thing_searched)) |
c36bcf1b | 2714 | return Qnil; |
daa37602 | 2715 | |
6bbd7a29 GM |
2716 | prev = Qnil; |
2717 | ||
38182d90 | 2718 | data = alloca ((2 * search_regs.num_regs + 1) * sizeof *data); |
4746118a | 2719 | |
41c01205 | 2720 | len = 0; |
4746118a | 2721 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 | 2722 | { |
d311d28c | 2723 | ptrdiff_t start = search_regs.start[i]; |
ca1d1d23 JB |
2724 | if (start >= 0) |
2725 | { | |
56256c2a RS |
2726 | if (EQ (last_thing_searched, Qt) |
2727 | || ! NILP (integers)) | |
ca1d1d23 | 2728 | { |
c235cce7 KH |
2729 | XSETFASTINT (data[2 * i], start); |
2730 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2731 | } |
0ed62dc7 | 2732 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2733 | { |
2734 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2735 | Fset_marker (data[2 * i], |
2736 | make_number (start), | |
2737 | last_thing_searched); | |
ca1d1d23 JB |
2738 | data[2 * i + 1] = Fmake_marker (); |
2739 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2740 | make_number (search_regs.end[i]), |
daa37602 | 2741 | last_thing_searched); |
ca1d1d23 | 2742 | } |
daa37602 JB |
2743 | else |
2744 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
1088b922 | 2745 | emacs_abort (); |
daa37602 | 2746 | |
abd0071c | 2747 | len = 2 * i + 2; |
ca1d1d23 JB |
2748 | } |
2749 | else | |
abd0071c | 2750 | data[2 * i] = data[2 * i + 1] = Qnil; |
ca1d1d23 | 2751 | } |
56256c2a | 2752 | |
bd2cbd56 | 2753 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2754 | { |
bd2cbd56 | 2755 | data[len] = last_thing_searched; |
41c01205 DK |
2756 | len++; |
2757 | } | |
2758 | ||
56256c2a RS |
2759 | /* If REUSE is not usable, cons up the values and return them. */ |
2760 | if (! CONSP (reuse)) | |
41c01205 | 2761 | return Flist (len, data); |
56256c2a RS |
2762 | |
2763 | /* If REUSE is a list, store as many value elements as will fit | |
2764 | into the elements of REUSE. */ | |
2765 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2766 | i++, tail = XCDR (tail)) |
56256c2a | 2767 | { |
41c01205 | 2768 | if (i < len) |
f3fbd155 | 2769 | XSETCAR (tail, data[i]); |
56256c2a | 2770 | else |
f3fbd155 | 2771 | XSETCAR (tail, Qnil); |
56256c2a RS |
2772 | prev = tail; |
2773 | } | |
2774 | ||
2775 | /* If we couldn't fit all value elements into REUSE, | |
2776 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2777 | if (i < len) |
2778 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2779 | |
2780 | return reuse; | |
ca1d1d23 JB |
2781 | } |
2782 | ||
b51d6c92 SM |
2783 | /* We used to have an internal use variant of `reseat' described as: |
2784 | ||
2785 | If RESEAT is `evaporate', put the markers back on the free list | |
2786 | immediately. No other references to the markers must exist in this | |
2787 | case, so it is used only internally on the unwind stack and | |
2788 | save-match-data from Lisp. | |
2789 | ||
2790 | But it was ill-conceived: those supposedly-internal markers get exposed via | |
2791 | the undo-list, so freeing them here is unsafe. */ | |
ca1d1d23 | 2792 | |
a7ca3326 | 2793 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 2, 0, |
8c1a1077 | 2794 | doc: /* Set internal data on last search match from elements of LIST. |
abd0071c KS |
2795 | LIST should have been created by calling `match-data' previously. |
2796 | ||
51f10faa | 2797 | If optional arg RESEAT is non-nil, make markers on LIST point nowhere. */) |
5842a27b | 2798 | (register Lisp_Object list, Lisp_Object reseat) |
ca1d1d23 | 2799 | { |
5f2ab479 | 2800 | ptrdiff_t i; |
ca1d1d23 JB |
2801 | register Lisp_Object marker; |
2802 | ||
7074fde6 FP |
2803 | if (running_asynch_code) |
2804 | save_search_regs (); | |
2805 | ||
29100cea | 2806 | CHECK_LIST (list); |
ca1d1d23 | 2807 | |
41c01205 DK |
2808 | /* Unless we find a marker with a buffer or an explicit buffer |
2809 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
2810 | last_thing_searched = Qt; |
2811 | ||
4746118a JB |
2812 | /* Allocate registers if they don't already exist. */ |
2813 | { | |
d311d28c | 2814 | EMACS_INT length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2815 | |
2816 | if (length > search_regs.num_regs) | |
2817 | { | |
0065d054 | 2818 | ptrdiff_t num_regs = search_regs.num_regs; |
d311d28c PE |
2819 | if (PTRDIFF_MAX < length) |
2820 | memory_full (SIZE_MAX); | |
0065d054 PE |
2821 | search_regs.start = |
2822 | xpalloc (search_regs.start, &num_regs, length - num_regs, | |
2823 | min (PTRDIFF_MAX, UINT_MAX), sizeof (regoff_t)); | |
2824 | search_regs.end = | |
2825 | xrealloc (search_regs.end, num_regs * sizeof (regoff_t)); | |
2826 | ||
2827 | for (i = search_regs.num_regs; i < num_regs; i++) | |
e62371e9 KH |
2828 | search_regs.start[i] = -1; |
2829 | ||
0065d054 | 2830 | search_regs.num_regs = num_regs; |
4746118a | 2831 | } |
ca1d1d23 | 2832 | |
abd0071c | 2833 | for (i = 0; CONSP (list); i++) |
41c01205 | 2834 | { |
abd0071c | 2835 | marker = XCAR (list); |
bd2cbd56 | 2836 | if (BUFFERP (marker)) |
c3762cbd | 2837 | { |
bd2cbd56 | 2838 | last_thing_searched = marker; |
c3762cbd DK |
2839 | break; |
2840 | } | |
2841 | if (i >= length) | |
2842 | break; | |
41c01205 DK |
2843 | if (NILP (marker)) |
2844 | { | |
2845 | search_regs.start[i] = -1; | |
abd0071c | 2846 | list = XCDR (list); |
41c01205 DK |
2847 | } |
2848 | else | |
2849 | { | |
d311d28c | 2850 | Lisp_Object from; |
abd0071c | 2851 | Lisp_Object m; |
e2811828 | 2852 | |
abd0071c | 2853 | m = marker; |
41c01205 DK |
2854 | if (MARKERP (marker)) |
2855 | { | |
2856 | if (XMARKER (marker)->buffer == 0) | |
2857 | XSETFASTINT (marker, 0); | |
2858 | else | |
2859 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
2860 | } | |
e2811828 | 2861 | |
41c01205 | 2862 | CHECK_NUMBER_COERCE_MARKER (marker); |
d311d28c | 2863 | from = marker; |
e2811828 | 2864 | |
abd0071c KS |
2865 | if (!NILP (reseat) && MARKERP (m)) |
2866 | { | |
b51d6c92 | 2867 | unchain_marker (XMARKER (m)); |
9ad54a7e | 2868 | XSETCAR (list, Qnil); |
abd0071c KS |
2869 | } |
2870 | ||
2871 | if ((list = XCDR (list), !CONSP (list))) | |
2872 | break; | |
2873 | ||
2874 | m = marker = XCAR (list); | |
2875 | ||
41c01205 DK |
2876 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
2877 | XSETFASTINT (marker, 0); | |
e2811828 | 2878 | |
41c01205 | 2879 | CHECK_NUMBER_COERCE_MARKER (marker); |
0b1fccc4 PE |
2880 | if ((XINT (from) < 0 |
2881 | ? TYPE_MINIMUM (regoff_t) <= XINT (from) | |
2882 | : XINT (from) <= TYPE_MAXIMUM (regoff_t)) | |
2883 | && (XINT (marker) < 0 | |
2884 | ? TYPE_MINIMUM (regoff_t) <= XINT (marker) | |
2885 | : XINT (marker) <= TYPE_MAXIMUM (regoff_t))) | |
d311d28c PE |
2886 | { |
2887 | search_regs.start[i] = XINT (from); | |
2888 | search_regs.end[i] = XINT (marker); | |
2889 | } | |
2890 | else | |
2891 | { | |
2892 | search_regs.start[i] = -1; | |
2893 | } | |
abd0071c KS |
2894 | |
2895 | if (!NILP (reseat) && MARKERP (m)) | |
2896 | { | |
b51d6c92 | 2897 | unchain_marker (XMARKER (m)); |
9ad54a7e | 2898 | XSETCAR (list, Qnil); |
abd0071c | 2899 | } |
41c01205 | 2900 | } |
abd0071c | 2901 | list = XCDR (list); |
41c01205 | 2902 | } |
ca1d1d23 | 2903 | |
41c01205 DK |
2904 | for (; i < search_regs.num_regs; i++) |
2905 | search_regs.start[i] = -1; | |
2906 | } | |
ca1d1d23 | 2907 | |
177c0ea7 | 2908 | return Qnil; |
ca1d1d23 JB |
2909 | } |
2910 | ||
457882c2 | 2911 | /* If true the match data have been saved in saved_search_regs |
7074fde6 | 2912 | during the execution of a sentinel or filter. */ |
457882c2 | 2913 | static bool search_regs_saved; |
7074fde6 | 2914 | static struct re_registers saved_search_regs; |
41c01205 | 2915 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
2916 | |
2917 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
2918 | if asynchronous code (filter or sentinel) is running. */ | |
2919 | static void | |
971de7fb | 2920 | save_search_regs (void) |
7074fde6 FP |
2921 | { |
2922 | if (!search_regs_saved) | |
2923 | { | |
2924 | saved_search_regs.num_regs = search_regs.num_regs; | |
2925 | saved_search_regs.start = search_regs.start; | |
2926 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
2927 | saved_last_thing_searched = last_thing_searched; |
2928 | last_thing_searched = Qnil; | |
7074fde6 | 2929 | search_regs.num_regs = 0; |
2d4a771a RS |
2930 | search_regs.start = 0; |
2931 | search_regs.end = 0; | |
7074fde6 FP |
2932 | |
2933 | search_regs_saved = 1; | |
2934 | } | |
2935 | } | |
2936 | ||
2937 | /* Called upon exit from filters and sentinels. */ | |
2938 | void | |
971de7fb | 2939 | restore_search_regs (void) |
7074fde6 FP |
2940 | { |
2941 | if (search_regs_saved) | |
2942 | { | |
2943 | if (search_regs.num_regs > 0) | |
2944 | { | |
2945 | xfree (search_regs.start); | |
2946 | xfree (search_regs.end); | |
2947 | } | |
2948 | search_regs.num_regs = saved_search_regs.num_regs; | |
2949 | search_regs.start = saved_search_regs.start; | |
2950 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
2951 | last_thing_searched = saved_last_thing_searched; |
2952 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
2953 | search_regs_saved = 0; |
2954 | } | |
2955 | } | |
2956 | ||
27e498e6 | 2957 | static void |
971de7fb | 2958 | unwind_set_match_data (Lisp_Object list) |
abd0071c | 2959 | { |
b51d6c92 | 2960 | /* It is NOT ALWAYS safe to free (evaporate) the markers immediately. */ |
27e498e6 | 2961 | Fset_match_data (list, Qt); |
abd0071c KS |
2962 | } |
2963 | ||
2964 | /* Called to unwind protect the match data. */ | |
2965 | void | |
971de7fb | 2966 | record_unwind_save_match_data (void) |
abd0071c KS |
2967 | { |
2968 | record_unwind_protect (unwind_set_match_data, | |
2969 | Fmatch_data (Qnil, Qnil, Qnil)); | |
2970 | } | |
2971 | ||
e9a452d9 | 2972 | /* Quote a string to deactivate reg-expr chars */ |
ca1d1d23 JB |
2973 | |
2974 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 | 2975 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
5842a27b | 2976 | (Lisp_Object string) |
ca1d1d23 | 2977 | { |
457882c2 PE |
2978 | char *in, *out, *end; |
2979 | char *temp; | |
2980 | ptrdiff_t backslashes_added = 0; | |
ca1d1d23 | 2981 | |
b7826503 | 2982 | CHECK_STRING (string); |
ca1d1d23 | 2983 | |
38182d90 | 2984 | temp = alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
2985 | |
2986 | /* Now copy the data into the new string, inserting escapes. */ | |
2987 | ||
a7e979a4 | 2988 | in = SSDATA (string); |
d5db4077 | 2989 | end = in + SBYTES (string); |
177c0ea7 | 2990 | out = temp; |
ca1d1d23 JB |
2991 | |
2992 | for (; in != end; in++) | |
2993 | { | |
66bc6082 | 2994 | if (*in == '[' |
ca1d1d23 JB |
2995 | || *in == '*' || *in == '.' || *in == '\\' |
2996 | || *in == '?' || *in == '+' | |
2997 | || *in == '^' || *in == '$') | |
0c8533c6 | 2998 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
2999 | *out++ = *in; |
3000 | } | |
3001 | ||
3f8100f1 | 3002 | return make_specified_string (temp, |
d5db4077 | 3003 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
3004 | out - temp, |
3005 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 3006 | } |
177c0ea7 | 3007 | \f |
dfcf069d | 3008 | void |
971de7fb | 3009 | syms_of_search (void) |
ca1d1d23 JB |
3010 | { |
3011 | register int i; | |
3012 | ||
487282dc KH |
3013 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
3014 | { | |
3015 | searchbufs[i].buf.allocated = 100; | |
23f86fce | 3016 | searchbufs[i].buf.buffer = xmalloc (100); |
487282dc KH |
3017 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
3018 | searchbufs[i].regexp = Qnil; | |
ecdb561e | 3019 | searchbufs[i].whitespace_regexp = Qnil; |
b69e3c18 | 3020 | searchbufs[i].syntax_table = Qnil; |
487282dc | 3021 | staticpro (&searchbufs[i].regexp); |
aa77b5ce | 3022 | staticpro (&searchbufs[i].whitespace_regexp); |
b69e3c18 | 3023 | staticpro (&searchbufs[i].syntax_table); |
487282dc KH |
3024 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); |
3025 | } | |
3026 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 | 3027 | |
cd3520a4 JB |
3028 | DEFSYM (Qsearch_failed, "search-failed"); |
3029 | DEFSYM (Qinvalid_regexp, "invalid-regexp"); | |
ca1d1d23 JB |
3030 | |
3031 | Fput (Qsearch_failed, Qerror_conditions, | |
3438fe21 | 3032 | listn (CONSTYPE_PURE, 2, Qsearch_failed, Qerror)); |
ca1d1d23 | 3033 | Fput (Qsearch_failed, Qerror_message, |
2a0213a6 | 3034 | build_pure_c_string ("Search failed")); |
ca1d1d23 JB |
3035 | |
3036 | Fput (Qinvalid_regexp, Qerror_conditions, | |
3438fe21 | 3037 | listn (CONSTYPE_PURE, 2, Qinvalid_regexp, Qerror)); |
ca1d1d23 | 3038 | Fput (Qinvalid_regexp, Qerror_message, |
2a0213a6 | 3039 | build_pure_c_string ("Invalid regexp")); |
ca1d1d23 | 3040 | |
daa37602 JB |
3041 | last_thing_searched = Qnil; |
3042 | staticpro (&last_thing_searched); | |
3043 | ||
0f6af254 DK |
3044 | saved_last_thing_searched = Qnil; |
3045 | staticpro (&saved_last_thing_searched); | |
3046 | ||
29208e82 | 3047 | DEFVAR_LISP ("search-spaces-regexp", Vsearch_spaces_regexp, |
e2811828 | 3048 | doc: /* Regexp to substitute for bunches of spaces in regexp search. |
f31a9a68 RS |
3049 | Some commands use this for user-specified regexps. |
3050 | Spaces that occur inside character classes or repetition operators | |
3051 | or other such regexp constructs are not replaced with this. | |
3052 | A value of nil (which is the normal value) means treat spaces literally. */); | |
41a33295 | 3053 | Vsearch_spaces_regexp = Qnil; |
f31a9a68 | 3054 | |
29208e82 | 3055 | DEFVAR_LISP ("inhibit-changing-match-data", Vinhibit_changing_match_data, |
ef887810 | 3056 | doc: /* Internal use only. |
39d0bf74 RS |
3057 | If non-nil, the primitive searching and matching functions |
3058 | such as `looking-at', `string-match', `re-search-forward', etc., | |
3059 | do not set the match data. The proper way to use this variable | |
3060 | is to bind it with `let' around a small expression. */); | |
ef887810 RS |
3061 | Vinhibit_changing_match_data = Qnil; |
3062 | ||
ca1d1d23 | 3063 | defsubr (&Slooking_at); |
b819a390 RS |
3064 | defsubr (&Sposix_looking_at); |
3065 | defsubr (&Sstring_match); | |
3066 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
3067 | defsubr (&Ssearch_forward); |
3068 | defsubr (&Ssearch_backward); | |
ca1d1d23 JB |
3069 | defsubr (&Sre_search_forward); |
3070 | defsubr (&Sre_search_backward); | |
b819a390 RS |
3071 | defsubr (&Sposix_search_forward); |
3072 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
3073 | defsubr (&Sreplace_match); |
3074 | defsubr (&Smatch_beginning); | |
3075 | defsubr (&Smatch_end); | |
3076 | defsubr (&Smatch_data); | |
3f1c005b | 3077 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
3078 | defsubr (&Sregexp_quote); |
3079 | } |