Commit | Line | Data |
---|---|---|
ca1d1d23 | 1 | /* String search routines for GNU Emacs. |
73b0cd50 | 2 | Copyright (C) 1985-1987, 1993-1994, 1997-1999, 2001-2011 |
8cabe764 | 3 | Free Software Foundation, Inc. |
ca1d1d23 JB |
4 | |
5 | This file is part of GNU Emacs. | |
6 | ||
9ec0b715 | 7 | GNU Emacs is free software: you can redistribute it and/or modify |
ca1d1d23 | 8 | it under the terms of the GNU General Public License as published by |
9ec0b715 GM |
9 | the Free Software Foundation, either version 3 of the License, or |
10 | (at your option) any later version. | |
ca1d1d23 JB |
11 | |
12 | GNU Emacs is distributed in the hope that it will be useful, | |
13 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | GNU General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
9ec0b715 | 18 | along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */ |
ca1d1d23 JB |
19 | |
20 | ||
18160b98 | 21 | #include <config.h> |
d7306fe6 | 22 | #include <setjmp.h> |
ca1d1d23 JB |
23 | #include "lisp.h" |
24 | #include "syntax.h" | |
5679531d | 25 | #include "category.h" |
ca1d1d23 | 26 | #include "buffer.h" |
76eb0881 | 27 | #include "character.h" |
89ad6492 | 28 | #include "charset.h" |
9169c321 | 29 | #include "region-cache.h" |
ca1d1d23 | 30 | #include "commands.h" |
9ac0d9e0 | 31 | #include "blockinput.h" |
bf1760bb | 32 | #include "intervals.h" |
4746118a | 33 | |
ca1d1d23 JB |
34 | #include <sys/types.h> |
35 | #include "regex.h" | |
36 | ||
1d288aef | 37 | #define REGEXP_CACHE_SIZE 20 |
ca1d1d23 | 38 | |
487282dc KH |
39 | /* If the regexp is non-nil, then the buffer contains the compiled form |
40 | of that regexp, suitable for searching. */ | |
1d288aef RS |
41 | struct regexp_cache |
42 | { | |
487282dc | 43 | struct regexp_cache *next; |
ecdb561e | 44 | Lisp_Object regexp, whitespace_regexp; |
b69e3c18 | 45 | /* Syntax table for which the regexp applies. We need this because |
58e95211 SM |
46 | of character classes. If this is t, then the compiled pattern is valid |
47 | for any syntax-table. */ | |
b69e3c18 | 48 | Lisp_Object syntax_table; |
487282dc KH |
49 | struct re_pattern_buffer buf; |
50 | char fastmap[0400]; | |
b819a390 RS |
51 | /* Nonzero means regexp was compiled to do full POSIX backtracking. */ |
52 | char posix; | |
487282dc | 53 | }; |
ca1d1d23 | 54 | |
487282dc KH |
55 | /* The instances of that struct. */ |
56 | struct regexp_cache searchbufs[REGEXP_CACHE_SIZE]; | |
ca1d1d23 | 57 | |
487282dc KH |
58 | /* The head of the linked list; points to the most recently used buffer. */ |
59 | struct regexp_cache *searchbuf_head; | |
ca1d1d23 | 60 | |
ca1d1d23 | 61 | |
4746118a JB |
62 | /* Every call to re_match, etc., must pass &search_regs as the regs |
63 | argument unless you can show it is unnecessary (i.e., if re_match | |
64 | is certainly going to be called again before region-around-match | |
65 | can be called). | |
66 | ||
67 | Since the registers are now dynamically allocated, we need to make | |
68 | sure not to refer to the Nth register before checking that it has | |
1113d9db JB |
69 | been allocated by checking search_regs.num_regs. |
70 | ||
71 | The regex code keeps track of whether it has allocated the search | |
487282dc KH |
72 | buffer using bits in the re_pattern_buffer. This means that whenever |
73 | you compile a new pattern, it completely forgets whether it has | |
1113d9db JB |
74 | allocated any registers, and will allocate new registers the next |
75 | time you call a searching or matching function. Therefore, we need | |
76 | to call re_set_registers after compiling a new pattern or after | |
77 | setting the match registers, so that the regex functions will be | |
78 | able to free or re-allocate it properly. */ | |
ca1d1d23 JB |
79 | static struct re_registers search_regs; |
80 | ||
daa37602 JB |
81 | /* The buffer in which the last search was performed, or |
82 | Qt if the last search was done in a string; | |
83 | Qnil if no searching has been done yet. */ | |
84 | static Lisp_Object last_thing_searched; | |
ca1d1d23 | 85 | |
8e6208c5 | 86 | /* error condition signaled when regexp compile_pattern fails */ |
ca1d1d23 JB |
87 | |
88 | Lisp_Object Qinvalid_regexp; | |
89 | ||
06f77a8a KS |
90 | /* Error condition used for failing searches */ |
91 | Lisp_Object Qsearch_failed; | |
92 | ||
f57e2426 J |
93 | static void set_search_regs (EMACS_INT, EMACS_INT); |
94 | static void save_search_regs (void); | |
c098fdb8 EZ |
95 | static EMACS_INT simple_search (EMACS_INT, unsigned char *, EMACS_INT, |
96 | EMACS_INT, Lisp_Object, EMACS_INT, EMACS_INT, | |
f57e2426 | 97 | EMACS_INT, EMACS_INT); |
c098fdb8 | 98 | static EMACS_INT boyer_moore (EMACS_INT, unsigned char *, EMACS_INT, EMACS_INT, |
f57e2426 J |
99 | Lisp_Object, Lisp_Object, |
100 | EMACS_INT, EMACS_INT, | |
101 | EMACS_INT, EMACS_INT, int); | |
102 | static EMACS_INT search_buffer (Lisp_Object, EMACS_INT, EMACS_INT, | |
c098fdb8 | 103 | EMACS_INT, EMACS_INT, EMACS_INT, int, |
f57e2426 | 104 | Lisp_Object, Lisp_Object, int); |
971de7fb | 105 | static void matcher_overflow (void) NO_RETURN; |
b819a390 | 106 | |
ca1d1d23 | 107 | static void |
971de7fb | 108 | matcher_overflow (void) |
ca1d1d23 JB |
109 | { |
110 | error ("Stack overflow in regexp matcher"); | |
111 | } | |
112 | ||
b819a390 RS |
113 | /* Compile a regexp and signal a Lisp error if anything goes wrong. |
114 | PATTERN is the pattern to compile. | |
115 | CP is the place to put the result. | |
facdc750 | 116 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 | 117 | POSIX is nonzero if we want full backtracking (POSIX style) |
5679531d | 118 | for this pattern. 0 means backtrack only enough to get a valid match. |
bbc73b48 KH |
119 | |
120 | The behavior also depends on Vsearch_spaces_regexp. */ | |
ca1d1d23 | 121 | |
487282dc | 122 | static void |
cfe0661d | 123 | compile_pattern_1 (struct regexp_cache *cp, Lisp_Object pattern, Lisp_Object translate, int posix) |
ca1d1d23 | 124 | { |
d451e4db | 125 | char *val; |
b819a390 | 126 | reg_syntax_t old; |
ca1d1d23 | 127 | |
487282dc | 128 | cp->regexp = Qnil; |
59fab369 | 129 | cp->buf.translate = (! NILP (translate) ? translate : make_number (0)); |
b819a390 | 130 | cp->posix = posix; |
93daa011 | 131 | cp->buf.multibyte = STRING_MULTIBYTE (pattern); |
89ad6492 | 132 | cp->buf.charset_unibyte = charset_unibyte; |
d8c85250 CY |
133 | if (STRINGP (Vsearch_spaces_regexp)) |
134 | cp->whitespace_regexp = Vsearch_spaces_regexp; | |
135 | else | |
136 | cp->whitespace_regexp = Qnil; | |
137 | ||
e7b2dd2e RS |
138 | /* rms: I think BLOCK_INPUT is not needed here any more, |
139 | because regex.c defines malloc to call xmalloc. | |
140 | Using BLOCK_INPUT here means the debugger won't run if an error occurs. | |
141 | So let's turn it off. */ | |
142 | /* BLOCK_INPUT; */ | |
fb4a568d | 143 | old = re_set_syntax (RE_SYNTAX_EMACS |
b819a390 | 144 | | (posix ? 0 : RE_NO_POSIX_BACKTRACKING)); |
d8c85250 CY |
145 | |
146 | if (STRINGP (Vsearch_spaces_regexp)) | |
42a5b22f | 147 | re_set_whitespace_regexp (SSDATA (Vsearch_spaces_regexp)); |
d8c85250 CY |
148 | else |
149 | re_set_whitespace_regexp (NULL); | |
bbc73b48 | 150 | |
51b59d79 | 151 | val = (char *) re_compile_pattern (SSDATA (pattern), |
8f924df7 | 152 | SBYTES (pattern), &cp->buf); |
bbc73b48 | 153 | |
58e95211 SM |
154 | /* If the compiled pattern hard codes some of the contents of the |
155 | syntax-table, it can only be reused with *this* syntax table. */ | |
4b4deea2 | 156 | cp->syntax_table = cp->buf.used_syntax ? BVAR (current_buffer, syntax_table) : Qt; |
58e95211 | 157 | |
bbc73b48 KH |
158 | re_set_whitespace_regexp (NULL); |
159 | ||
b819a390 | 160 | re_set_syntax (old); |
e7b2dd2e | 161 | /* UNBLOCK_INPUT; */ |
ca1d1d23 | 162 | if (val) |
06f77a8a | 163 | xsignal1 (Qinvalid_regexp, build_string (val)); |
1113d9db | 164 | |
487282dc | 165 | cp->regexp = Fcopy_sequence (pattern); |
487282dc KH |
166 | } |
167 | ||
6efc7887 RS |
168 | /* Shrink each compiled regexp buffer in the cache |
169 | to the size actually used right now. | |
170 | This is called from garbage collection. */ | |
171 | ||
172 | void | |
971de7fb | 173 | shrink_regexp_cache (void) |
6efc7887 | 174 | { |
a968f437 | 175 | struct regexp_cache *cp; |
6efc7887 RS |
176 | |
177 | for (cp = searchbuf_head; cp != 0; cp = cp->next) | |
178 | { | |
179 | cp->buf.allocated = cp->buf.used; | |
180 | cp->buf.buffer | |
b23c0a83 | 181 | = (unsigned char *) xrealloc (cp->buf.buffer, cp->buf.used); |
6efc7887 RS |
182 | } |
183 | } | |
184 | ||
54dd3310 SM |
185 | /* Clear the regexp cache w.r.t. a particular syntax table, |
186 | because it was changed. | |
e5b94d44 CY |
187 | There is no danger of memory leak here because re_compile_pattern |
188 | automagically manages the memory in each re_pattern_buffer struct, | |
189 | based on its `allocated' and `buffer' values. */ | |
190 | void | |
971de7fb | 191 | clear_regexp_cache (void) |
e5b94d44 CY |
192 | { |
193 | int i; | |
194 | ||
195 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) | |
920fd1fc EZ |
196 | /* It's tempting to compare with the syntax-table we've actually changed, |
197 | but it's not sufficient because char-table inheritance means that | |
54dd3310 SM |
198 | modifying one syntax-table can change others at the same time. */ |
199 | if (!EQ (searchbufs[i].syntax_table, Qt)) | |
200 | searchbufs[i].regexp = Qnil; | |
e5b94d44 CY |
201 | } |
202 | ||
487282dc | 203 | /* Compile a regexp if necessary, but first check to see if there's one in |
b819a390 RS |
204 | the cache. |
205 | PATTERN is the pattern to compile. | |
facdc750 | 206 | TRANSLATE is a translation table for ignoring case, or nil for none. |
b819a390 RS |
207 | REGP is the structure that says where to store the "register" |
208 | values that will result from matching this pattern. | |
209 | If it is 0, we should compile the pattern not to record any | |
210 | subexpression bounds. | |
211 | POSIX is nonzero if we want full backtracking (POSIX style) | |
212 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
487282dc KH |
213 | |
214 | struct re_pattern_buffer * | |
971de7fb | 215 | compile_pattern (Lisp_Object pattern, struct re_registers *regp, Lisp_Object translate, int posix, int multibyte) |
487282dc KH |
216 | { |
217 | struct regexp_cache *cp, **cpp; | |
218 | ||
219 | for (cpp = &searchbuf_head; ; cpp = &cp->next) | |
220 | { | |
221 | cp = *cpp; | |
f1b9c7c1 KR |
222 | /* Entries are initialized to nil, and may be set to nil by |
223 | compile_pattern_1 if the pattern isn't valid. Don't apply | |
49a5f770 KR |
224 | string accessors in those cases. However, compile_pattern_1 |
225 | is only applied to the cache entry we pick here to reuse. So | |
226 | nil should never appear before a non-nil entry. */ | |
7c752c80 | 227 | if (NILP (cp->regexp)) |
f1b9c7c1 | 228 | goto compile_it; |
d5db4077 | 229 | if (SCHARS (cp->regexp) == SCHARS (pattern) |
cf69b13e | 230 | && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern) |
1d288aef | 231 | && !NILP (Fstring_equal (cp->regexp, pattern)) |
59fab369 | 232 | && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0))) |
5679531d | 233 | && cp->posix == posix |
58e95211 | 234 | && (EQ (cp->syntax_table, Qt) |
4b4deea2 | 235 | || EQ (cp->syntax_table, BVAR (current_buffer, syntax_table))) |
89ad6492 KH |
236 | && !NILP (Fequal (cp->whitespace_regexp, Vsearch_spaces_regexp)) |
237 | && cp->buf.charset_unibyte == charset_unibyte) | |
487282dc KH |
238 | break; |
239 | ||
f1b9c7c1 KR |
240 | /* If we're at the end of the cache, compile into the nil cell |
241 | we found, or the last (least recently used) cell with a | |
242 | string value. */ | |
487282dc KH |
243 | if (cp->next == 0) |
244 | { | |
f1b9c7c1 | 245 | compile_it: |
cfe0661d | 246 | compile_pattern_1 (cp, pattern, translate, posix); |
487282dc KH |
247 | break; |
248 | } | |
249 | } | |
250 | ||
251 | /* When we get here, cp (aka *cpp) contains the compiled pattern, | |
252 | either because we found it in the cache or because we just compiled it. | |
253 | Move it to the front of the queue to mark it as most recently used. */ | |
254 | *cpp = cp->next; | |
255 | cp->next = searchbuf_head; | |
256 | searchbuf_head = cp; | |
1113d9db | 257 | |
6639708c RS |
258 | /* Advise the searching functions about the space we have allocated |
259 | for register data. */ | |
260 | if (regp) | |
261 | re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end); | |
262 | ||
78edd3b7 | 263 | /* The compiled pattern can be used both for multibyte and unibyte |
89ad6492 KH |
264 | target. But, we have to tell which the pattern is used for. */ |
265 | cp->buf.target_multibyte = multibyte; | |
266 | ||
487282dc | 267 | return &cp->buf; |
ca1d1d23 JB |
268 | } |
269 | ||
ca1d1d23 | 270 | \f |
b819a390 | 271 | static Lisp_Object |
971de7fb | 272 | looking_at_1 (Lisp_Object string, int posix) |
ca1d1d23 JB |
273 | { |
274 | Lisp_Object val; | |
275 | unsigned char *p1, *p2; | |
e7deaab0 | 276 | EMACS_INT s1, s2; |
c098fdb8 | 277 | register EMACS_INT i; |
487282dc | 278 | struct re_pattern_buffer *bufp; |
ca1d1d23 | 279 | |
7074fde6 FP |
280 | if (running_asynch_code) |
281 | save_search_regs (); | |
282 | ||
910c747a | 283 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
4b4deea2 TT |
284 | XCHAR_TABLE (BVAR (current_buffer, case_canon_table))->extras[2] |
285 | = BVAR (current_buffer, case_eqv_table); | |
910c747a | 286 | |
b7826503 | 287 | CHECK_STRING (string); |
ef887810 RS |
288 | bufp = compile_pattern (string, |
289 | (NILP (Vinhibit_changing_match_data) | |
290 | ? &search_regs : NULL), | |
4b4deea2 TT |
291 | (!NILP (BVAR (current_buffer, case_fold_search)) |
292 | ? BVAR (current_buffer, case_canon_table) : Qnil), | |
0c8533c6 | 293 | posix, |
4b4deea2 | 294 | !NILP (BVAR (current_buffer, enable_multibyte_characters))); |
ca1d1d23 JB |
295 | |
296 | immediate_quit = 1; | |
297 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ | |
298 | ||
299 | /* Get pointers and sizes of the two strings | |
300 | that make up the visible portion of the buffer. */ | |
301 | ||
302 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 303 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 304 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 305 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
306 | if (s1 < 0) |
307 | { | |
308 | p2 = p1; | |
fa8ed3e0 | 309 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
310 | s1 = 0; |
311 | } | |
312 | if (s2 < 0) | |
313 | { | |
fa8ed3e0 | 314 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
315 | s2 = 0; |
316 | } | |
8bb43c28 RS |
317 | |
318 | re_match_object = Qnil; | |
177c0ea7 | 319 | |
487282dc | 320 | i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
ef887810 RS |
321 | PT_BYTE - BEGV_BYTE, |
322 | (NILP (Vinhibit_changing_match_data) | |
323 | ? &search_regs : NULL), | |
fa8ed3e0 | 324 | ZV_BYTE - BEGV_BYTE); |
de182d70 | 325 | immediate_quit = 0; |
177c0ea7 | 326 | |
ca1d1d23 JB |
327 | if (i == -2) |
328 | matcher_overflow (); | |
329 | ||
330 | val = (0 <= i ? Qt : Qnil); | |
ef887810 | 331 | if (NILP (Vinhibit_changing_match_data) && i >= 0) |
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); | |
337 | search_regs.end[i] | |
338 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
339 | } | |
ef887810 RS |
340 | |
341 | /* Set last_thing_searched only when match data is changed. */ | |
342 | if (NILP (Vinhibit_changing_match_data)) | |
343 | XSETBUFFER (last_thing_searched, current_buffer); | |
344 | ||
ca1d1d23 JB |
345 | return val; |
346 | } | |
347 | ||
b819a390 | 348 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
8c1a1077 PJ |
349 | doc: /* Return t if text after point matches regular expression REGEXP. |
350 | This function modifies the match data that `match-beginning', | |
351 | `match-end' and `match-data' access; save and restore the match | |
352 | data if you want to preserve them. */) | |
5842a27b | 353 | (Lisp_Object regexp) |
b819a390 | 354 | { |
94f94972 | 355 | return looking_at_1 (regexp, 0); |
b819a390 RS |
356 | } |
357 | ||
358 | DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0, | |
8c1a1077 PJ |
359 | doc: /* Return t if text after point matches regular expression REGEXP. |
360 | Find the longest match, in accord with Posix regular expression rules. | |
361 | This function modifies the match data that `match-beginning', | |
362 | `match-end' and `match-data' access; save and restore the match | |
363 | data if you want to preserve them. */) | |
5842a27b | 364 | (Lisp_Object regexp) |
b819a390 | 365 | { |
94f94972 | 366 | return looking_at_1 (regexp, 1); |
b819a390 RS |
367 | } |
368 | \f | |
369 | static Lisp_Object | |
971de7fb | 370 | string_match_1 (Lisp_Object regexp, Lisp_Object string, Lisp_Object start, int posix) |
ca1d1d23 JB |
371 | { |
372 | int val; | |
487282dc | 373 | struct re_pattern_buffer *bufp; |
e7deaab0 | 374 | EMACS_INT pos, pos_byte; |
0c8533c6 | 375 | int 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 | { | |
c098fdb8 | 387 | EMACS_INT 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. */ |
4b4deea2 TT |
399 | XCHAR_TABLE (BVAR (current_buffer, case_canon_table))->extras[2] |
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 | |
b819a390 | 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 | ||
472 | int | |
971de7fb | 473 | fast_string_match (Lisp_Object regexp, Lisp_Object string) |
e59a8453 RS |
474 | { |
475 | int 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 | |
5679531d | 495 | int |
971de7fb | 496 | fast_c_string_match_ignore_case (Lisp_Object regexp, const char *string) |
5679531d KH |
497 | { |
498 | int val; | |
499 | struct re_pattern_buffer *bufp; | |
c098fdb8 | 500 | size_t len = strlen (string); |
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 | ||
515 | int | |
971de7fb | 516 | fast_string_match_ignore_case (Lisp_Object regexp, Lisp_Object string) |
be5f4dfb KH |
517 | { |
518 | int val; | |
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 | ||
539 | EMACS_INT | |
971de7fb | 540 | fast_looking_at (Lisp_Object regexp, EMACS_INT pos, EMACS_INT pos_byte, EMACS_INT limit, EMACS_INT limit_byte, Lisp_Object string) |
8b13507a KH |
541 | { |
542 | int multibyte; | |
543 | struct re_pattern_buffer *buf; | |
544 | unsigned char *p1, *p2; | |
e7deaab0 | 545 | EMACS_INT s1, s2; |
8b13507a | 546 | EMACS_INT len; |
78edd3b7 | 547 | |
8b13507a KH |
548 | if (STRINGP (string)) |
549 | { | |
550 | if (pos_byte < 0) | |
551 | pos_byte = string_char_to_byte (string, pos); | |
552 | if (limit_byte < 0) | |
553 | limit_byte = string_char_to_byte (string, limit); | |
554 | p1 = NULL; | |
555 | s1 = 0; | |
556 | p2 = SDATA (string); | |
557 | s2 = SBYTES (string); | |
558 | re_match_object = string; | |
559 | multibyte = STRING_MULTIBYTE (string); | |
560 | } | |
561 | else | |
562 | { | |
563 | if (pos_byte < 0) | |
564 | pos_byte = CHAR_TO_BYTE (pos); | |
565 | if (limit_byte < 0) | |
566 | limit_byte = CHAR_TO_BYTE (limit); | |
567 | pos_byte -= BEGV_BYTE; | |
568 | limit_byte -= BEGV_BYTE; | |
569 | p1 = BEGV_ADDR; | |
570 | s1 = GPT_BYTE - BEGV_BYTE; | |
571 | p2 = GAP_END_ADDR; | |
572 | s2 = ZV_BYTE - GPT_BYTE; | |
573 | if (s1 < 0) | |
574 | { | |
575 | p2 = p1; | |
576 | s2 = ZV_BYTE - BEGV_BYTE; | |
577 | s1 = 0; | |
578 | } | |
579 | if (s2 < 0) | |
580 | { | |
581 | s1 = ZV_BYTE - BEGV_BYTE; | |
582 | s2 = 0; | |
583 | } | |
584 | re_match_object = Qnil; | |
4b4deea2 | 585 | multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
8b13507a KH |
586 | } |
587 | ||
588 | buf = compile_pattern (regexp, 0, Qnil, 0, multibyte); | |
589 | immediate_quit = 1; | |
590 | len = re_match_2 (buf, (char *) p1, s1, (char *) p2, s2, | |
591 | pos_byte, NULL, limit_byte); | |
592 | immediate_quit = 0; | |
593 | ||
594 | return len; | |
595 | } | |
596 | ||
597 | \f | |
9169c321 JB |
598 | /* The newline cache: remembering which sections of text have no newlines. */ |
599 | ||
600 | /* If the user has requested newline caching, make sure it's on. | |
601 | Otherwise, make sure it's off. | |
602 | This is our cheezy way of associating an action with the change of | |
603 | state of a buffer-local variable. */ | |
604 | static void | |
971de7fb | 605 | newline_cache_on_off (struct buffer *buf) |
9169c321 | 606 | { |
4b4deea2 | 607 | if (NILP (BVAR (buf, cache_long_line_scans))) |
9169c321 JB |
608 | { |
609 | /* It should be off. */ | |
610 | if (buf->newline_cache) | |
611 | { | |
612 | free_region_cache (buf->newline_cache); | |
613 | buf->newline_cache = 0; | |
614 | } | |
615 | } | |
616 | else | |
617 | { | |
618 | /* It should be on. */ | |
619 | if (buf->newline_cache == 0) | |
620 | buf->newline_cache = new_region_cache (); | |
621 | } | |
622 | } | |
623 | ||
624 | \f | |
625 | /* Search for COUNT instances of the character TARGET between START and END. | |
626 | ||
627 | If COUNT is positive, search forwards; END must be >= START. | |
628 | If COUNT is negative, search backwards for the -COUNTth instance; | |
629 | END must be <= START. | |
630 | If COUNT is zero, do anything you please; run rogue, for all I care. | |
631 | ||
632 | If END is zero, use BEGV or ZV instead, as appropriate for the | |
633 | direction indicated by COUNT. | |
ffd56f97 JB |
634 | |
635 | If we find COUNT instances, set *SHORTAGE to zero, and return the | |
a9f2a45f | 636 | position past the COUNTth match. Note that for reverse motion |
5bfe95c9 | 637 | this is not the same as the usual convention for Emacs motion commands. |
ffd56f97 | 638 | |
9169c321 JB |
639 | If we don't find COUNT instances before reaching END, set *SHORTAGE |
640 | to the number of TARGETs left unfound, and return END. | |
ffd56f97 | 641 | |
087a5f81 RS |
642 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do |
643 | except when inside redisplay. */ | |
644 | ||
c098fdb8 EZ |
645 | EMACS_INT |
646 | scan_buffer (register int target, EMACS_INT start, EMACS_INT end, | |
647 | EMACS_INT count, int *shortage, int allow_quit) | |
ca1d1d23 | 648 | { |
9169c321 | 649 | struct region_cache *newline_cache; |
177c0ea7 | 650 | int direction; |
ffd56f97 | 651 | |
9169c321 JB |
652 | if (count > 0) |
653 | { | |
654 | direction = 1; | |
655 | if (! end) end = ZV; | |
656 | } | |
657 | else | |
658 | { | |
659 | direction = -1; | |
660 | if (! end) end = BEGV; | |
661 | } | |
ffd56f97 | 662 | |
9169c321 JB |
663 | newline_cache_on_off (current_buffer); |
664 | newline_cache = current_buffer->newline_cache; | |
ca1d1d23 JB |
665 | |
666 | if (shortage != 0) | |
667 | *shortage = 0; | |
668 | ||
087a5f81 | 669 | immediate_quit = allow_quit; |
ca1d1d23 | 670 | |
ffd56f97 | 671 | if (count > 0) |
9169c321 | 672 | while (start != end) |
ca1d1d23 | 673 | { |
9169c321 JB |
674 | /* Our innermost scanning loop is very simple; it doesn't know |
675 | about gaps, buffer ends, or the newline cache. ceiling is | |
676 | the position of the last character before the next such | |
677 | obstacle --- the last character the dumb search loop should | |
678 | examine. */ | |
e7deaab0 AS |
679 | EMACS_INT ceiling_byte = CHAR_TO_BYTE (end) - 1; |
680 | EMACS_INT start_byte = CHAR_TO_BYTE (start); | |
681 | EMACS_INT tem; | |
9169c321 JB |
682 | |
683 | /* If we're looking for a newline, consult the newline cache | |
684 | to see where we can avoid some scanning. */ | |
685 | if (target == '\n' && newline_cache) | |
686 | { | |
c098fdb8 | 687 | EMACS_INT next_change; |
9169c321 JB |
688 | immediate_quit = 0; |
689 | while (region_cache_forward | |
fa8ed3e0 RS |
690 | (current_buffer, newline_cache, start_byte, &next_change)) |
691 | start_byte = next_change; | |
cbe0db0d | 692 | immediate_quit = allow_quit; |
9169c321 | 693 | |
fa8ed3e0 RS |
694 | /* START should never be after END. */ |
695 | if (start_byte > ceiling_byte) | |
696 | start_byte = ceiling_byte; | |
9169c321 JB |
697 | |
698 | /* Now the text after start is an unknown region, and | |
699 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 700 | ceiling_byte = min (next_change - 1, ceiling_byte); |
9169c321 JB |
701 | } |
702 | ||
703 | /* The dumb loop can only scan text stored in contiguous | |
704 | bytes. BUFFER_CEILING_OF returns the last character | |
705 | position that is contiguous, so the ceiling is the | |
706 | position after that. */ | |
67ce527d KH |
707 | tem = BUFFER_CEILING_OF (start_byte); |
708 | ceiling_byte = min (tem, ceiling_byte); | |
9169c321 JB |
709 | |
710 | { | |
177c0ea7 | 711 | /* The termination address of the dumb loop. */ |
fa8ed3e0 RS |
712 | register unsigned char *ceiling_addr |
713 | = BYTE_POS_ADDR (ceiling_byte) + 1; | |
714 | register unsigned char *cursor | |
715 | = BYTE_POS_ADDR (start_byte); | |
9169c321 JB |
716 | unsigned char *base = cursor; |
717 | ||
718 | while (cursor < ceiling_addr) | |
719 | { | |
720 | unsigned char *scan_start = cursor; | |
721 | ||
722 | /* The dumb loop. */ | |
723 | while (*cursor != target && ++cursor < ceiling_addr) | |
724 | ; | |
725 | ||
726 | /* If we're looking for newlines, cache the fact that | |
727 | the region from start to cursor is free of them. */ | |
728 | if (target == '\n' && newline_cache) | |
729 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
730 | start_byte + scan_start - base, |
731 | start_byte + cursor - base); | |
9169c321 JB |
732 | |
733 | /* Did we find the target character? */ | |
734 | if (cursor < ceiling_addr) | |
735 | { | |
736 | if (--count == 0) | |
737 | { | |
738 | immediate_quit = 0; | |
fa8ed3e0 | 739 | return BYTE_TO_CHAR (start_byte + cursor - base + 1); |
9169c321 JB |
740 | } |
741 | cursor++; | |
742 | } | |
743 | } | |
744 | ||
fa8ed3e0 | 745 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 | 746 | } |
ca1d1d23 JB |
747 | } |
748 | else | |
9169c321 JB |
749 | while (start > end) |
750 | { | |
751 | /* The last character to check before the next obstacle. */ | |
e7deaab0 AS |
752 | EMACS_INT ceiling_byte = CHAR_TO_BYTE (end); |
753 | EMACS_INT start_byte = CHAR_TO_BYTE (start); | |
754 | EMACS_INT tem; | |
9169c321 JB |
755 | |
756 | /* Consult the newline cache, if appropriate. */ | |
757 | if (target == '\n' && newline_cache) | |
758 | { | |
c098fdb8 | 759 | EMACS_INT next_change; |
9169c321 JB |
760 | immediate_quit = 0; |
761 | while (region_cache_backward | |
fa8ed3e0 RS |
762 | (current_buffer, newline_cache, start_byte, &next_change)) |
763 | start_byte = next_change; | |
cbe0db0d | 764 | immediate_quit = allow_quit; |
9169c321 JB |
765 | |
766 | /* Start should never be at or before end. */ | |
fa8ed3e0 RS |
767 | if (start_byte <= ceiling_byte) |
768 | start_byte = ceiling_byte + 1; | |
9169c321 JB |
769 | |
770 | /* Now the text before start is an unknown region, and | |
771 | next_change is the position of the next known region. */ | |
fa8ed3e0 | 772 | ceiling_byte = max (next_change, ceiling_byte); |
9169c321 JB |
773 | } |
774 | ||
775 | /* Stop scanning before the gap. */ | |
67ce527d KH |
776 | tem = BUFFER_FLOOR_OF (start_byte - 1); |
777 | ceiling_byte = max (tem, ceiling_byte); | |
9169c321 JB |
778 | |
779 | { | |
780 | /* The termination address of the dumb loop. */ | |
fa8ed3e0 RS |
781 | register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte); |
782 | register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1); | |
9169c321 JB |
783 | unsigned char *base = cursor; |
784 | ||
785 | while (cursor >= ceiling_addr) | |
786 | { | |
787 | unsigned char *scan_start = cursor; | |
788 | ||
789 | while (*cursor != target && --cursor >= ceiling_addr) | |
790 | ; | |
791 | ||
792 | /* If we're looking for newlines, cache the fact that | |
793 | the region from after the cursor to start is free of them. */ | |
794 | if (target == '\n' && newline_cache) | |
795 | know_region_cache (current_buffer, newline_cache, | |
fa8ed3e0 RS |
796 | start_byte + cursor - base, |
797 | start_byte + scan_start - base); | |
9169c321 JB |
798 | |
799 | /* Did we find the target character? */ | |
800 | if (cursor >= ceiling_addr) | |
801 | { | |
802 | if (++count >= 0) | |
803 | { | |
804 | immediate_quit = 0; | |
fa8ed3e0 | 805 | return BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
806 | } |
807 | cursor--; | |
808 | } | |
809 | } | |
810 | ||
fa8ed3e0 | 811 | start = BYTE_TO_CHAR (start_byte + cursor - base); |
9169c321 JB |
812 | } |
813 | } | |
814 | ||
ca1d1d23 JB |
815 | immediate_quit = 0; |
816 | if (shortage != 0) | |
ffd56f97 | 817 | *shortage = count * direction; |
9169c321 | 818 | return start; |
ca1d1d23 | 819 | } |
fa8ed3e0 RS |
820 | \f |
821 | /* Search for COUNT instances of a line boundary, which means either a | |
822 | newline or (if selective display enabled) a carriage return. | |
823 | Start at START. If COUNT is negative, search backwards. | |
824 | ||
825 | We report the resulting position by calling TEMP_SET_PT_BOTH. | |
826 | ||
827 | If we find COUNT instances. we position after (always after, | |
828 | even if scanning backwards) the COUNTth match, and return 0. | |
829 | ||
830 | If we don't find COUNT instances before reaching the end of the | |
831 | buffer (or the beginning, if scanning backwards), we return | |
832 | the number of line boundaries left unfound, and position at | |
833 | the limit we bumped up against. | |
834 | ||
835 | If ALLOW_QUIT is non-zero, set immediate_quit. That's good to do | |
d5d57b92 | 836 | except in special cases. */ |
ca1d1d23 | 837 | |
c098fdb8 EZ |
838 | EMACS_INT |
839 | scan_newline (EMACS_INT start, EMACS_INT start_byte, | |
840 | EMACS_INT limit, EMACS_INT limit_byte, | |
841 | register EMACS_INT count, int allow_quit) | |
63fa018d | 842 | { |
fa8ed3e0 RS |
843 | int direction = ((count > 0) ? 1 : -1); |
844 | ||
845 | register unsigned char *cursor; | |
846 | unsigned char *base; | |
847 | ||
e7deaab0 | 848 | EMACS_INT ceiling; |
fa8ed3e0 RS |
849 | register unsigned char *ceiling_addr; |
850 | ||
d5d57b92 RS |
851 | int old_immediate_quit = immediate_quit; |
852 | ||
fa8ed3e0 RS |
853 | /* The code that follows is like scan_buffer |
854 | but checks for either newline or carriage return. */ | |
855 | ||
d5d57b92 RS |
856 | if (allow_quit) |
857 | immediate_quit++; | |
fa8ed3e0 RS |
858 | |
859 | start_byte = CHAR_TO_BYTE (start); | |
860 | ||
861 | if (count > 0) | |
862 | { | |
863 | while (start_byte < limit_byte) | |
864 | { | |
865 | ceiling = BUFFER_CEILING_OF (start_byte); | |
866 | ceiling = min (limit_byte - 1, ceiling); | |
867 | ceiling_addr = BYTE_POS_ADDR (ceiling) + 1; | |
868 | base = (cursor = BYTE_POS_ADDR (start_byte)); | |
869 | while (1) | |
870 | { | |
871 | while (*cursor != '\n' && ++cursor != ceiling_addr) | |
872 | ; | |
873 | ||
874 | if (cursor != ceiling_addr) | |
875 | { | |
876 | if (--count == 0) | |
877 | { | |
d5d57b92 | 878 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
879 | start_byte = start_byte + cursor - base + 1; |
880 | start = BYTE_TO_CHAR (start_byte); | |
881 | TEMP_SET_PT_BOTH (start, start_byte); | |
882 | return 0; | |
883 | } | |
884 | else | |
885 | if (++cursor == ceiling_addr) | |
886 | break; | |
887 | } | |
888 | else | |
889 | break; | |
890 | } | |
891 | start_byte += cursor - base; | |
892 | } | |
893 | } | |
894 | else | |
895 | { | |
fa8ed3e0 RS |
896 | while (start_byte > limit_byte) |
897 | { | |
898 | ceiling = BUFFER_FLOOR_OF (start_byte - 1); | |
899 | ceiling = max (limit_byte, ceiling); | |
900 | ceiling_addr = BYTE_POS_ADDR (ceiling) - 1; | |
901 | base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1); | |
902 | while (1) | |
903 | { | |
904 | while (--cursor != ceiling_addr && *cursor != '\n') | |
905 | ; | |
906 | ||
907 | if (cursor != ceiling_addr) | |
908 | { | |
909 | if (++count == 0) | |
910 | { | |
d5d57b92 | 911 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
912 | /* Return the position AFTER the match we found. */ |
913 | start_byte = start_byte + cursor - base + 1; | |
914 | start = BYTE_TO_CHAR (start_byte); | |
915 | TEMP_SET_PT_BOTH (start, start_byte); | |
916 | return 0; | |
917 | } | |
918 | } | |
919 | else | |
920 | break; | |
921 | } | |
922 | /* Here we add 1 to compensate for the last decrement | |
923 | of CURSOR, which took it past the valid range. */ | |
924 | start_byte += cursor - base + 1; | |
925 | } | |
926 | } | |
927 | ||
928 | TEMP_SET_PT_BOTH (limit, limit_byte); | |
d5d57b92 | 929 | immediate_quit = old_immediate_quit; |
fa8ed3e0 RS |
930 | |
931 | return count * direction; | |
63fa018d RS |
932 | } |
933 | ||
c098fdb8 EZ |
934 | EMACS_INT |
935 | find_next_newline_no_quit (EMACS_INT from, EMACS_INT cnt) | |
ca1d1d23 | 936 | { |
fa8ed3e0 | 937 | return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0); |
9169c321 JB |
938 | } |
939 | ||
9169c321 JB |
940 | /* Like find_next_newline, but returns position before the newline, |
941 | not after, and only search up to TO. This isn't just | |
942 | find_next_newline (...)-1, because you might hit TO. */ | |
fa8ed3e0 | 943 | |
c098fdb8 EZ |
944 | EMACS_INT |
945 | find_before_next_newline (EMACS_INT from, EMACS_INT to, EMACS_INT cnt) | |
9169c321 JB |
946 | { |
947 | int shortage; | |
c098fdb8 | 948 | EMACS_INT pos = scan_buffer ('\n', from, to, cnt, &shortage, 1); |
9169c321 JB |
949 | |
950 | if (shortage == 0) | |
951 | pos--; | |
177c0ea7 | 952 | |
9169c321 | 953 | return pos; |
ca1d1d23 JB |
954 | } |
955 | \f | |
ca1d1d23 JB |
956 | /* Subroutines of Lisp buffer search functions. */ |
957 | ||
958 | static Lisp_Object | |
c098fdb8 EZ |
959 | search_command (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, |
960 | Lisp_Object count, int direction, int RE, int posix) | |
ca1d1d23 JB |
961 | { |
962 | register int np; | |
c098fdb8 | 963 | EMACS_INT lim, lim_byte; |
ca1d1d23 JB |
964 | int n = direction; |
965 | ||
966 | if (!NILP (count)) | |
967 | { | |
b7826503 | 968 | CHECK_NUMBER (count); |
ca1d1d23 JB |
969 | n *= XINT (count); |
970 | } | |
971 | ||
b7826503 | 972 | CHECK_STRING (string); |
ca1d1d23 | 973 | if (NILP (bound)) |
9f43ad85 RS |
974 | { |
975 | if (n > 0) | |
976 | lim = ZV, lim_byte = ZV_BYTE; | |
977 | else | |
978 | lim = BEGV, lim_byte = BEGV_BYTE; | |
979 | } | |
ca1d1d23 JB |
980 | else |
981 | { | |
b7826503 | 982 | CHECK_NUMBER_COERCE_MARKER (bound); |
ca1d1d23 | 983 | lim = XINT (bound); |
6ec8bbd2 | 984 | if (n > 0 ? lim < PT : lim > PT) |
ca1d1d23 JB |
985 | error ("Invalid search bound (wrong side of point)"); |
986 | if (lim > ZV) | |
9f43ad85 | 987 | lim = ZV, lim_byte = ZV_BYTE; |
588d2fd5 | 988 | else if (lim < BEGV) |
9f43ad85 | 989 | lim = BEGV, lim_byte = BEGV_BYTE; |
588d2fd5 KH |
990 | else |
991 | lim_byte = CHAR_TO_BYTE (lim); | |
ca1d1d23 JB |
992 | } |
993 | ||
910c747a | 994 | /* This is so set_image_of_range_1 in regex.c can find the EQV table. */ |
4b4deea2 TT |
995 | XCHAR_TABLE (BVAR (current_buffer, case_canon_table))->extras[2] |
996 | = BVAR (current_buffer, case_eqv_table); | |
910c747a | 997 | |
9f43ad85 | 998 | np = search_buffer (string, PT, PT_BYTE, lim, lim_byte, n, RE, |
4b4deea2 TT |
999 | (!NILP (BVAR (current_buffer, case_fold_search)) |
1000 | ? BVAR (current_buffer, case_canon_table) | |
3135e9fd | 1001 | : Qnil), |
4b4deea2 TT |
1002 | (!NILP (BVAR (current_buffer, case_fold_search)) |
1003 | ? BVAR (current_buffer, case_eqv_table) | |
3135e9fd | 1004 | : Qnil), |
b819a390 | 1005 | posix); |
ca1d1d23 JB |
1006 | if (np <= 0) |
1007 | { | |
1008 | if (NILP (noerror)) | |
06f77a8a KS |
1009 | xsignal1 (Qsearch_failed, string); |
1010 | ||
ca1d1d23 JB |
1011 | if (!EQ (noerror, Qt)) |
1012 | { | |
1013 | if (lim < BEGV || lim > ZV) | |
1014 | abort (); | |
9f43ad85 | 1015 | SET_PT_BOTH (lim, lim_byte); |
a5f217b8 RS |
1016 | return Qnil; |
1017 | #if 0 /* This would be clean, but maybe programs depend on | |
1018 | a value of nil here. */ | |
481399bf | 1019 | np = lim; |
a5f217b8 | 1020 | #endif |
ca1d1d23 | 1021 | } |
481399bf RS |
1022 | else |
1023 | return Qnil; | |
ca1d1d23 JB |
1024 | } |
1025 | ||
1026 | if (np < BEGV || np > ZV) | |
1027 | abort (); | |
1028 | ||
1029 | SET_PT (np); | |
1030 | ||
1031 | return make_number (np); | |
1032 | } | |
1033 | \f | |
fa8ed3e0 RS |
1034 | /* Return 1 if REGEXP it matches just one constant string. */ |
1035 | ||
b6d6a51c | 1036 | static int |
971de7fb | 1037 | trivial_regexp_p (Lisp_Object regexp) |
b6d6a51c | 1038 | { |
c098fdb8 | 1039 | EMACS_INT len = SBYTES (regexp); |
d5db4077 | 1040 | unsigned char *s = SDATA (regexp); |
b6d6a51c KH |
1041 | while (--len >= 0) |
1042 | { | |
1043 | switch (*s++) | |
1044 | { | |
1045 | case '.': case '*': case '+': case '?': case '[': case '^': case '$': | |
1046 | return 0; | |
1047 | case '\\': | |
1048 | if (--len < 0) | |
1049 | return 0; | |
1050 | switch (*s++) | |
1051 | { | |
1052 | case '|': case '(': case ')': case '`': case '\'': case 'b': | |
1053 | case 'B': case '<': case '>': case 'w': case 'W': case 's': | |
29f89fe7 | 1054 | case 'S': case '=': case '{': case '}': case '_': |
5679531d | 1055 | case 'c': case 'C': /* for categoryspec and notcategoryspec */ |
866f60fd | 1056 | case '1': case '2': case '3': case '4': case '5': |
b6d6a51c KH |
1057 | case '6': case '7': case '8': case '9': |
1058 | return 0; | |
1059 | } | |
1060 | } | |
1061 | } | |
1062 | return 1; | |
1063 | } | |
1064 | ||
ca325161 | 1065 | /* Search for the n'th occurrence of STRING in the current buffer, |
ca1d1d23 | 1066 | starting at position POS and stopping at position LIM, |
b819a390 | 1067 | treating STRING as a literal string if RE is false or as |
ca1d1d23 JB |
1068 | a regular expression if RE is true. |
1069 | ||
1070 | If N is positive, searching is forward and LIM must be greater than POS. | |
1071 | If N is negative, searching is backward and LIM must be less than POS. | |
1072 | ||
facdc750 | 1073 | Returns -x if x occurrences remain to be found (x > 0), |
ca1d1d23 | 1074 | or else the position at the beginning of the Nth occurrence |
b819a390 RS |
1075 | (if searching backward) or the end (if searching forward). |
1076 | ||
1077 | POSIX is nonzero if we want full backtracking (POSIX style) | |
1078 | for this pattern. 0 means backtrack only enough to get a valid match. */ | |
ca1d1d23 | 1079 | |
aff2ce94 RS |
1080 | #define TRANSLATE(out, trt, d) \ |
1081 | do \ | |
1082 | { \ | |
1083 | if (! NILP (trt)) \ | |
1084 | { \ | |
1085 | Lisp_Object temp; \ | |
1086 | temp = Faref (trt, make_number (d)); \ | |
1087 | if (INTEGERP (temp)) \ | |
1088 | out = XINT (temp); \ | |
1089 | else \ | |
1090 | out = d; \ | |
1091 | } \ | |
1092 | else \ | |
1093 | out = d; \ | |
1094 | } \ | |
1095 | while (0) | |
facdc750 | 1096 | |
ef887810 RS |
1097 | /* Only used in search_buffer, to record the end position of the match |
1098 | when searching regexps and SEARCH_REGS should not be changed | |
1099 | (i.e. Vinhibit_changing_match_data is non-nil). */ | |
1100 | static struct re_registers search_regs_1; | |
1101 | ||
e7deaab0 | 1102 | static EMACS_INT |
d5a3eaaf | 1103 | search_buffer (Lisp_Object string, EMACS_INT pos, EMACS_INT pos_byte, |
c098fdb8 | 1104 | EMACS_INT lim, EMACS_INT lim_byte, EMACS_INT n, |
d5a3eaaf | 1105 | int RE, Lisp_Object trt, Lisp_Object inverse_trt, int posix) |
ca1d1d23 | 1106 | { |
c098fdb8 EZ |
1107 | EMACS_INT len = SCHARS (string); |
1108 | EMACS_INT len_byte = SBYTES (string); | |
facdc750 | 1109 | register int i; |
ca1d1d23 | 1110 | |
7074fde6 FP |
1111 | if (running_asynch_code) |
1112 | save_search_regs (); | |
1113 | ||
a7e4cdde | 1114 | /* Searching 0 times means don't move. */ |
ca1d1d23 | 1115 | /* Null string is found at starting position. */ |
a7e4cdde | 1116 | if (len == 0 || n == 0) |
ca325161 | 1117 | { |
0353b28f | 1118 | set_search_regs (pos_byte, 0); |
ca325161 RS |
1119 | return pos; |
1120 | } | |
3f57a499 | 1121 | |
41a33295 | 1122 | if (RE && !(trivial_regexp_p (string) && NILP (Vsearch_spaces_regexp))) |
ca1d1d23 | 1123 | { |
facdc750 | 1124 | unsigned char *p1, *p2; |
c098fdb8 | 1125 | EMACS_INT s1, s2; |
487282dc KH |
1126 | struct re_pattern_buffer *bufp; |
1127 | ||
ef887810 RS |
1128 | bufp = compile_pattern (string, |
1129 | (NILP (Vinhibit_changing_match_data) | |
1130 | ? &search_regs : &search_regs_1), | |
1131 | trt, posix, | |
4b4deea2 | 1132 | !NILP (BVAR (current_buffer, enable_multibyte_characters))); |
ca1d1d23 | 1133 | |
ca1d1d23 JB |
1134 | immediate_quit = 1; /* Quit immediately if user types ^G, |
1135 | because letting this function finish | |
1136 | can take too long. */ | |
1137 | QUIT; /* Do a pending quit right away, | |
1138 | to avoid paradoxical behavior */ | |
1139 | /* Get pointers and sizes of the two strings | |
1140 | that make up the visible portion of the buffer. */ | |
1141 | ||
1142 | p1 = BEGV_ADDR; | |
fa8ed3e0 | 1143 | s1 = GPT_BYTE - BEGV_BYTE; |
ca1d1d23 | 1144 | p2 = GAP_END_ADDR; |
fa8ed3e0 | 1145 | s2 = ZV_BYTE - GPT_BYTE; |
ca1d1d23 JB |
1146 | if (s1 < 0) |
1147 | { | |
1148 | p2 = p1; | |
fa8ed3e0 | 1149 | s2 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1150 | s1 = 0; |
1151 | } | |
1152 | if (s2 < 0) | |
1153 | { | |
fa8ed3e0 | 1154 | s1 = ZV_BYTE - BEGV_BYTE; |
ca1d1d23 JB |
1155 | s2 = 0; |
1156 | } | |
8bb43c28 | 1157 | re_match_object = Qnil; |
177c0ea7 | 1158 | |
ca1d1d23 JB |
1159 | while (n < 0) |
1160 | { | |
c098fdb8 | 1161 | EMACS_INT val; |
487282dc | 1162 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1163 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1164 | (NILP (Vinhibit_changing_match_data) |
1165 | ? &search_regs : &search_regs_1), | |
42db823b | 1166 | /* Don't allow match past current point */ |
4996330b | 1167 | pos_byte - BEGV_BYTE); |
ca1d1d23 | 1168 | if (val == -2) |
b6d6a51c KH |
1169 | { |
1170 | matcher_overflow (); | |
1171 | } | |
ca1d1d23 JB |
1172 | if (val >= 0) |
1173 | { | |
ef887810 RS |
1174 | if (NILP (Vinhibit_changing_match_data)) |
1175 | { | |
1176 | pos_byte = search_regs.start[0] + BEGV_BYTE; | |
1177 | for (i = 0; i < search_regs.num_regs; i++) | |
1178 | if (search_regs.start[i] >= 0) | |
1179 | { | |
1180 | search_regs.start[i] | |
1181 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1182 | search_regs.end[i] | |
1183 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1184 | } | |
1185 | XSETBUFFER (last_thing_searched, current_buffer); | |
1186 | /* Set pos to the new position. */ | |
1187 | pos = search_regs.start[0]; | |
1188 | } | |
1189 | else | |
1190 | { | |
1191 | pos_byte = search_regs_1.start[0] + BEGV_BYTE; | |
1192 | /* Set pos to the new position. */ | |
1193 | pos = BYTE_TO_CHAR (search_regs_1.start[0] + BEGV_BYTE); | |
1194 | } | |
ca1d1d23 JB |
1195 | } |
1196 | else | |
1197 | { | |
1198 | immediate_quit = 0; | |
1199 | return (n); | |
1200 | } | |
1201 | n++; | |
1202 | } | |
1203 | while (n > 0) | |
1204 | { | |
c098fdb8 | 1205 | EMACS_INT val; |
487282dc | 1206 | val = re_search_2 (bufp, (char *) p1, s1, (char *) p2, s2, |
4996330b | 1207 | pos_byte - BEGV_BYTE, lim_byte - pos_byte, |
ef887810 RS |
1208 | (NILP (Vinhibit_changing_match_data) |
1209 | ? &search_regs : &search_regs_1), | |
4996330b | 1210 | lim_byte - BEGV_BYTE); |
ca1d1d23 | 1211 | if (val == -2) |
b6d6a51c KH |
1212 | { |
1213 | matcher_overflow (); | |
1214 | } | |
ca1d1d23 JB |
1215 | if (val >= 0) |
1216 | { | |
ef887810 RS |
1217 | if (NILP (Vinhibit_changing_match_data)) |
1218 | { | |
1219 | pos_byte = search_regs.end[0] + BEGV_BYTE; | |
1220 | for (i = 0; i < search_regs.num_regs; i++) | |
1221 | if (search_regs.start[i] >= 0) | |
1222 | { | |
1223 | search_regs.start[i] | |
1224 | = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE); | |
1225 | search_regs.end[i] | |
1226 | = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE); | |
1227 | } | |
1228 | XSETBUFFER (last_thing_searched, current_buffer); | |
1229 | pos = search_regs.end[0]; | |
1230 | } | |
1231 | else | |
1232 | { | |
1233 | pos_byte = search_regs_1.end[0] + BEGV_BYTE; | |
1234 | pos = BYTE_TO_CHAR (search_regs_1.end[0] + BEGV_BYTE); | |
1235 | } | |
ca1d1d23 JB |
1236 | } |
1237 | else | |
1238 | { | |
1239 | immediate_quit = 0; | |
1240 | return (0 - n); | |
1241 | } | |
1242 | n--; | |
1243 | } | |
1244 | immediate_quit = 0; | |
1245 | return (pos); | |
1246 | } | |
1247 | else /* non-RE case */ | |
1248 | { | |
facdc750 | 1249 | unsigned char *raw_pattern, *pat; |
c098fdb8 EZ |
1250 | EMACS_INT raw_pattern_size; |
1251 | EMACS_INT raw_pattern_size_byte; | |
facdc750 | 1252 | unsigned char *patbuf; |
4b4deea2 | 1253 | int multibyte = !NILP (BVAR (current_buffer, enable_multibyte_characters)); |
a967ed62 | 1254 | unsigned char *base_pat; |
49e44e81 KH |
1255 | /* Set to positive if we find a non-ASCII char that need |
1256 | translation. Otherwise set to zero later. */ | |
1257 | int char_base = -1; | |
040272ce | 1258 | int boyer_moore_ok = 1; |
facdc750 RS |
1259 | |
1260 | /* MULTIBYTE says whether the text to be searched is multibyte. | |
1261 | We must convert PATTERN to match that, or we will not really | |
1262 | find things right. */ | |
1263 | ||
1264 | if (multibyte == STRING_MULTIBYTE (string)) | |
1265 | { | |
51b59d79 | 1266 | raw_pattern = SDATA (string); |
d5db4077 KR |
1267 | raw_pattern_size = SCHARS (string); |
1268 | raw_pattern_size_byte = SBYTES (string); | |
facdc750 RS |
1269 | } |
1270 | else if (multibyte) | |
1271 | { | |
d5db4077 | 1272 | raw_pattern_size = SCHARS (string); |
facdc750 | 1273 | raw_pattern_size_byte |
d5db4077 | 1274 | = count_size_as_multibyte (SDATA (string), |
facdc750 | 1275 | raw_pattern_size); |
7276d3d8 | 1276 | raw_pattern = (unsigned char *) alloca (raw_pattern_size_byte + 1); |
d5db4077 KR |
1277 | copy_text (SDATA (string), raw_pattern, |
1278 | SCHARS (string), 0, 1); | |
facdc750 RS |
1279 | } |
1280 | else | |
1281 | { | |
1282 | /* Converting multibyte to single-byte. | |
1283 | ||
1284 | ??? Perhaps this conversion should be done in a special way | |
1285 | by subtracting nonascii-insert-offset from each non-ASCII char, | |
1286 | so that only the multibyte chars which really correspond to | |
1287 | the chosen single-byte character set can possibly match. */ | |
d5db4077 KR |
1288 | raw_pattern_size = SCHARS (string); |
1289 | raw_pattern_size_byte = SCHARS (string); | |
7276d3d8 | 1290 | raw_pattern = (unsigned char *) alloca (raw_pattern_size + 1); |
d5db4077 KR |
1291 | copy_text (SDATA (string), raw_pattern, |
1292 | SBYTES (string), 1, 0); | |
facdc750 RS |
1293 | } |
1294 | ||
1295 | /* Copy and optionally translate the pattern. */ | |
1296 | len = raw_pattern_size; | |
1297 | len_byte = raw_pattern_size_byte; | |
f5f7578b | 1298 | patbuf = (unsigned char *) alloca (len * MAX_MULTIBYTE_LENGTH); |
facdc750 RS |
1299 | pat = patbuf; |
1300 | base_pat = raw_pattern; | |
1301 | if (multibyte) | |
1302 | { | |
a17ae96a KH |
1303 | /* Fill patbuf by translated characters in STRING while |
1304 | checking if we can use boyer-moore search. If TRT is | |
1305 | non-nil, we can use boyer-moore search only if TRT can be | |
1306 | represented by the byte array of 256 elements. For that, | |
1307 | all non-ASCII case-equivalents of all case-senstive | |
1308 | characters in STRING must belong to the same charset and | |
1309 | row. */ | |
1310 | ||
facdc750 RS |
1311 | while (--len >= 0) |
1312 | { | |
a17ae96a | 1313 | unsigned char str_base[MAX_MULTIBYTE_LENGTH], *str; |
aff2ce94 | 1314 | int c, translated, inverse; |
a17ae96a | 1315 | int in_charlen, charlen; |
facdc750 RS |
1316 | |
1317 | /* If we got here and the RE flag is set, it's because we're | |
1318 | dealing with a regexp known to be trivial, so the backslash | |
1319 | just quotes the next character. */ | |
1320 | if (RE && *base_pat == '\\') | |
1321 | { | |
1322 | len--; | |
62221f22 | 1323 | raw_pattern_size--; |
facdc750 RS |
1324 | len_byte--; |
1325 | base_pat++; | |
1326 | } | |
1327 | ||
62a6e103 | 1328 | c = STRING_CHAR_AND_LENGTH (base_pat, in_charlen); |
040272ce | 1329 | |
a17ae96a | 1330 | if (NILP (trt)) |
facdc750 | 1331 | { |
a17ae96a KH |
1332 | str = base_pat; |
1333 | charlen = in_charlen; | |
1334 | } | |
1335 | else | |
1336 | { | |
1337 | /* Translate the character. */ | |
1338 | TRANSLATE (translated, trt, c); | |
1339 | charlen = CHAR_STRING (translated, str_base); | |
1340 | str = str_base; | |
1341 | ||
1342 | /* Check if C has any other case-equivalents. */ | |
1343 | TRANSLATE (inverse, inverse_trt, c); | |
1344 | /* If so, check if we can use boyer-moore. */ | |
1345 | if (c != inverse && boyer_moore_ok) | |
1346 | { | |
1347 | /* Check if all equivalents belong to the same | |
1348 | group of characters. Note that the check of C | |
49e44e81 KH |
1349 | itself is done by the last iteration. */ |
1350 | int this_char_base = -1; | |
1351 | ||
1352 | while (boyer_moore_ok) | |
a17ae96a | 1353 | { |
49e44e81 | 1354 | if (ASCII_BYTE_P (inverse)) |
a17ae96a | 1355 | { |
49e44e81 KH |
1356 | if (this_char_base > 0) |
1357 | boyer_moore_ok = 0; | |
1358 | else | |
dafaabd1 | 1359 | this_char_base = 0; |
a17ae96a | 1360 | } |
49e44e81 KH |
1361 | else if (CHAR_BYTE8_P (inverse)) |
1362 | /* Boyer-moore search can't handle a | |
1363 | translation of an eight-bit | |
1364 | character. */ | |
1365 | boyer_moore_ok = 0; | |
1366 | else if (this_char_base < 0) | |
1367 | { | |
1368 | this_char_base = inverse & ~0x3F; | |
1369 | if (char_base < 0) | |
1370 | char_base = this_char_base; | |
dafaabd1 | 1371 | else if (this_char_base != char_base) |
49e44e81 KH |
1372 | boyer_moore_ok = 0; |
1373 | } | |
1374 | else if ((inverse & ~0x3F) != this_char_base) | |
1375 | boyer_moore_ok = 0; | |
a17ae96a KH |
1376 | if (c == inverse) |
1377 | break; | |
1378 | TRANSLATE (inverse, inverse_trt, inverse); | |
1379 | } | |
1380 | } | |
aff2ce94 | 1381 | } |
facdc750 RS |
1382 | |
1383 | /* Store this character into the translated pattern. */ | |
72af86bd | 1384 | memcpy (pat, str, charlen); |
a17ae96a | 1385 | pat += charlen; |
facdc750 RS |
1386 | base_pat += in_charlen; |
1387 | len_byte -= in_charlen; | |
1388 | } | |
dafaabd1 AS |
1389 | |
1390 | /* If char_base is still negative we didn't find any translated | |
1391 | non-ASCII characters. */ | |
1392 | if (char_base < 0) | |
1393 | char_base = 0; | |
facdc750 RS |
1394 | } |
1395 | else | |
1396 | { | |
040272ce | 1397 | /* Unibyte buffer. */ |
a17ae96a | 1398 | char_base = 0; |
facdc750 RS |
1399 | while (--len >= 0) |
1400 | { | |
040272ce | 1401 | int c, translated; |
facdc750 RS |
1402 | |
1403 | /* If we got here and the RE flag is set, it's because we're | |
1404 | dealing with a regexp known to be trivial, so the backslash | |
1405 | just quotes the next character. */ | |
1406 | if (RE && *base_pat == '\\') | |
1407 | { | |
1408 | len--; | |
0190922f | 1409 | raw_pattern_size--; |
facdc750 RS |
1410 | base_pat++; |
1411 | } | |
1412 | c = *base_pat++; | |
aff2ce94 | 1413 | TRANSLATE (translated, trt, c); |
facdc750 RS |
1414 | *pat++ = translated; |
1415 | } | |
1416 | } | |
1417 | ||
1418 | len_byte = pat - patbuf; | |
1419 | len = raw_pattern_size; | |
1420 | pat = base_pat = patbuf; | |
1421 | ||
040272ce | 1422 | if (boyer_moore_ok) |
facdc750 | 1423 | return boyer_moore (n, pat, len, len_byte, trt, inverse_trt, |
aff2ce94 | 1424 | pos, pos_byte, lim, lim_byte, |
a17ae96a | 1425 | char_base); |
facdc750 RS |
1426 | else |
1427 | return simple_search (n, pat, len, len_byte, trt, | |
1428 | pos, pos_byte, lim, lim_byte); | |
1429 | } | |
1430 | } | |
1431 | \f | |
1432 | /* Do a simple string search N times for the string PAT, | |
1433 | whose length is LEN/LEN_BYTE, | |
1434 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1435 | TRT is the translation table. | |
f8bd51c4 | 1436 | |
facdc750 RS |
1437 | Return the character position where the match is found. |
1438 | Otherwise, if M matches remained to be found, return -M. | |
f8bd51c4 | 1439 | |
facdc750 RS |
1440 | This kind of search works regardless of what is in PAT and |
1441 | regardless of what is in TRT. It is used in cases where | |
1442 | boyer_moore cannot work. */ | |
1443 | ||
e7deaab0 | 1444 | static EMACS_INT |
c098fdb8 EZ |
1445 | simple_search (EMACS_INT n, unsigned char *pat, |
1446 | EMACS_INT len, EMACS_INT len_byte, Lisp_Object trt, | |
1447 | EMACS_INT pos, EMACS_INT pos_byte, | |
1448 | EMACS_INT lim, EMACS_INT lim_byte) | |
facdc750 | 1449 | { |
4b4deea2 | 1450 | int multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
ab228c24 | 1451 | int forward = n > 0; |
49e44e81 KH |
1452 | /* Number of buffer bytes matched. Note that this may be different |
1453 | from len_byte in a multibyte buffer. */ | |
c098fdb8 | 1454 | EMACS_INT match_byte; |
facdc750 RS |
1455 | |
1456 | if (lim > pos && multibyte) | |
1457 | while (n > 0) | |
1458 | { | |
1459 | while (1) | |
f8bd51c4 | 1460 | { |
facdc750 | 1461 | /* Try matching at position POS. */ |
e7deaab0 AS |
1462 | EMACS_INT this_pos = pos; |
1463 | EMACS_INT this_pos_byte = pos_byte; | |
c098fdb8 | 1464 | EMACS_INT this_len = len; |
facdc750 | 1465 | unsigned char *p = pat; |
a9469498 | 1466 | if (pos + len > lim || pos_byte + len_byte > lim_byte) |
facdc750 RS |
1467 | goto stop; |
1468 | ||
1469 | while (this_len > 0) | |
1470 | { | |
1471 | int charlen, buf_charlen; | |
ab228c24 | 1472 | int pat_ch, buf_ch; |
facdc750 | 1473 | |
62a6e103 | 1474 | pat_ch = STRING_CHAR_AND_LENGTH (p, charlen); |
facdc750 | 1475 | buf_ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (this_pos_byte), |
facdc750 | 1476 | buf_charlen); |
aff2ce94 | 1477 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1478 | |
1479 | if (buf_ch != pat_ch) | |
1480 | break; | |
ab228c24 | 1481 | |
ab228c24 RS |
1482 | this_len--; |
1483 | p += charlen; | |
1484 | ||
1485 | this_pos_byte += buf_charlen; | |
1486 | this_pos++; | |
facdc750 RS |
1487 | } |
1488 | ||
1489 | if (this_len == 0) | |
1490 | { | |
49e44e81 | 1491 | match_byte = this_pos_byte - pos_byte; |
facdc750 | 1492 | pos += len; |
49e44e81 | 1493 | pos_byte += match_byte; |
facdc750 RS |
1494 | break; |
1495 | } | |
1496 | ||
1497 | INC_BOTH (pos, pos_byte); | |
f8bd51c4 | 1498 | } |
facdc750 RS |
1499 | |
1500 | n--; | |
1501 | } | |
1502 | else if (lim > pos) | |
1503 | while (n > 0) | |
1504 | { | |
1505 | while (1) | |
f8bd51c4 | 1506 | { |
facdc750 | 1507 | /* Try matching at position POS. */ |
e7deaab0 | 1508 | EMACS_INT this_pos = pos; |
c098fdb8 | 1509 | EMACS_INT this_len = len; |
facdc750 RS |
1510 | unsigned char *p = pat; |
1511 | ||
1512 | if (pos + len > lim) | |
1513 | goto stop; | |
1514 | ||
1515 | while (this_len > 0) | |
1516 | { | |
1517 | int pat_ch = *p++; | |
1518 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1519 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1520 | |
1521 | if (buf_ch != pat_ch) | |
1522 | break; | |
ab228c24 RS |
1523 | |
1524 | this_len--; | |
1525 | this_pos++; | |
facdc750 RS |
1526 | } |
1527 | ||
1528 | if (this_len == 0) | |
1529 | { | |
49e44e81 | 1530 | match_byte = len; |
facdc750 RS |
1531 | pos += len; |
1532 | break; | |
1533 | } | |
1534 | ||
1535 | pos++; | |
f8bd51c4 | 1536 | } |
facdc750 RS |
1537 | |
1538 | n--; | |
1539 | } | |
1540 | /* Backwards search. */ | |
1541 | else if (lim < pos && multibyte) | |
1542 | while (n < 0) | |
1543 | { | |
1544 | while (1) | |
f8bd51c4 | 1545 | { |
facdc750 | 1546 | /* Try matching at position POS. */ |
8b264ecb AS |
1547 | EMACS_INT this_pos = pos; |
1548 | EMACS_INT this_pos_byte = pos_byte; | |
c098fdb8 | 1549 | EMACS_INT this_len = len; |
c525b3f2 | 1550 | const unsigned char *p = pat + len_byte; |
facdc750 | 1551 | |
8b264ecb | 1552 | if (this_pos - len < lim || (pos_byte - len_byte) < lim_byte) |
facdc750 RS |
1553 | goto stop; |
1554 | ||
1555 | while (this_len > 0) | |
1556 | { | |
ab228c24 | 1557 | int pat_ch, buf_ch; |
facdc750 | 1558 | |
8b264ecb AS |
1559 | DEC_BOTH (this_pos, this_pos_byte); |
1560 | PREV_CHAR_BOUNDARY (p, pat); | |
1561 | pat_ch = STRING_CHAR (p); | |
1562 | buf_ch = STRING_CHAR (BYTE_POS_ADDR (this_pos_byte)); | |
aff2ce94 | 1563 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1564 | |
1565 | if (buf_ch != pat_ch) | |
1566 | break; | |
ab228c24 | 1567 | |
ab228c24 | 1568 | this_len--; |
facdc750 RS |
1569 | } |
1570 | ||
1571 | if (this_len == 0) | |
1572 | { | |
8b264ecb AS |
1573 | match_byte = pos_byte - this_pos_byte; |
1574 | pos = this_pos; | |
1575 | pos_byte = this_pos_byte; | |
facdc750 RS |
1576 | break; |
1577 | } | |
1578 | ||
1579 | DEC_BOTH (pos, pos_byte); | |
f8bd51c4 KH |
1580 | } |
1581 | ||
facdc750 RS |
1582 | n++; |
1583 | } | |
1584 | else if (lim < pos) | |
1585 | while (n < 0) | |
1586 | { | |
1587 | while (1) | |
b6d6a51c | 1588 | { |
facdc750 | 1589 | /* Try matching at position POS. */ |
e7deaab0 | 1590 | EMACS_INT this_pos = pos - len; |
c098fdb8 | 1591 | EMACS_INT this_len = len; |
facdc750 RS |
1592 | unsigned char *p = pat; |
1593 | ||
a9469498 | 1594 | if (this_pos < lim) |
facdc750 RS |
1595 | goto stop; |
1596 | ||
1597 | while (this_len > 0) | |
1598 | { | |
1599 | int pat_ch = *p++; | |
1600 | int buf_ch = FETCH_BYTE (this_pos); | |
aff2ce94 | 1601 | TRANSLATE (buf_ch, trt, buf_ch); |
facdc750 RS |
1602 | |
1603 | if (buf_ch != pat_ch) | |
1604 | break; | |
ab228c24 RS |
1605 | this_len--; |
1606 | this_pos++; | |
facdc750 RS |
1607 | } |
1608 | ||
1609 | if (this_len == 0) | |
b6d6a51c | 1610 | { |
49e44e81 | 1611 | match_byte = len; |
facdc750 RS |
1612 | pos -= len; |
1613 | break; | |
b6d6a51c | 1614 | } |
facdc750 RS |
1615 | |
1616 | pos--; | |
b6d6a51c | 1617 | } |
facdc750 RS |
1618 | |
1619 | n++; | |
b6d6a51c | 1620 | } |
facdc750 RS |
1621 | |
1622 | stop: | |
1623 | if (n == 0) | |
aff2ce94 | 1624 | { |
ab228c24 | 1625 | if (forward) |
49e44e81 | 1626 | set_search_regs ((multibyte ? pos_byte : pos) - match_byte, match_byte); |
ab228c24 | 1627 | else |
49e44e81 | 1628 | set_search_regs (multibyte ? pos_byte : pos, match_byte); |
aff2ce94 RS |
1629 | |
1630 | return pos; | |
1631 | } | |
facdc750 RS |
1632 | else if (n > 0) |
1633 | return -n; | |
1634 | else | |
1635 | return n; | |
1636 | } | |
1637 | \f | |
0190922f | 1638 | /* Do Boyer-Moore search N times for the string BASE_PAT, |
facdc750 RS |
1639 | whose length is LEN/LEN_BYTE, |
1640 | from buffer position POS/POS_BYTE until LIM/LIM_BYTE. | |
1641 | DIRECTION says which direction we search in. | |
1642 | TRT and INVERSE_TRT are translation tables. | |
0190922f | 1643 | Characters in PAT are already translated by TRT. |
facdc750 | 1644 | |
0190922f KH |
1645 | This kind of search works if all the characters in BASE_PAT that |
1646 | have nontrivial translation are the same aside from the last byte. | |
1647 | This makes it possible to translate just the last byte of a | |
1648 | character, and do so after just a simple test of the context. | |
b2e6b10f | 1649 | CHAR_BASE is nonzero if there is such a non-ASCII character. |
facdc750 RS |
1650 | |
1651 | If that criterion is not satisfied, do not call this function. */ | |
1652 | ||
e7deaab0 | 1653 | static EMACS_INT |
c098fdb8 EZ |
1654 | boyer_moore (EMACS_INT n, unsigned char *base_pat, |
1655 | EMACS_INT len, EMACS_INT len_byte, | |
d5a3eaaf AS |
1656 | Lisp_Object trt, Lisp_Object inverse_trt, |
1657 | EMACS_INT pos, EMACS_INT pos_byte, | |
1658 | EMACS_INT lim, EMACS_INT lim_byte, int char_base) | |
facdc750 RS |
1659 | { |
1660 | int direction = ((n > 0) ? 1 : -1); | |
c098fdb8 | 1661 | register EMACS_INT dirlen; |
e7deaab0 AS |
1662 | EMACS_INT limit; |
1663 | int stride_for_teases = 0; | |
f4646fff | 1664 | int BM_tab[0400]; |
177c0ea7 | 1665 | register unsigned char *cursor, *p_limit; |
c098fdb8 EZ |
1666 | register EMACS_INT i; |
1667 | register int j; | |
cb6792d2 | 1668 | unsigned char *pat, *pat_end; |
4b4deea2 | 1669 | int multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
facdc750 RS |
1670 | |
1671 | unsigned char simple_translate[0400]; | |
0190922f | 1672 | /* These are set to the preceding bytes of a byte to be translated |
49e44e81 | 1673 | if char_base is nonzero. As the maximum byte length of a |
a17ae96a | 1674 | multibyte character is 5, we have to check at most four previous |
0190922f KH |
1675 | bytes. */ |
1676 | int translate_prev_byte1 = 0; | |
1677 | int translate_prev_byte2 = 0; | |
1678 | int translate_prev_byte3 = 0; | |
a17ae96a | 1679 | int translate_prev_byte4 = 0; |
facdc750 | 1680 | |
f4646fff AS |
1681 | /* The general approach is that we are going to maintain that we know |
1682 | the first (closest to the present position, in whatever direction | |
1683 | we're searching) character that could possibly be the last | |
1684 | (furthest from present position) character of a valid match. We | |
1685 | advance the state of our knowledge by looking at that character | |
1686 | and seeing whether it indeed matches the last character of the | |
1687 | pattern. If it does, we take a closer look. If it does not, we | |
1688 | move our pointer (to putative last characters) as far as is | |
1689 | logically possible. This amount of movement, which I call a | |
1690 | stride, will be the length of the pattern if the actual character | |
1691 | appears nowhere in the pattern, otherwise it will be the distance | |
1692 | from the last occurrence of that character to the end of the | |
1693 | pattern. If the amount is zero we have a possible match. */ | |
1694 | ||
1695 | /* Here we make a "mickey mouse" BM table. The stride of the search | |
1696 | is determined only by the last character of the putative match. | |
1697 | If that character does not match, we will stride the proper | |
1698 | distance to propose a match that superimposes it on the last | |
1699 | instance of a character that matches it (per trt), or misses | |
1700 | it entirely if there is none. */ | |
facdc750 RS |
1701 | |
1702 | dirlen = len_byte * direction; | |
cb6792d2 RS |
1703 | |
1704 | /* Record position after the end of the pattern. */ | |
1705 | pat_end = base_pat + len_byte; | |
1706 | /* BASE_PAT points to a character that we start scanning from. | |
1707 | It is the first character in a forward search, | |
1708 | the last character in a backward search. */ | |
facdc750 | 1709 | if (direction < 0) |
cb6792d2 RS |
1710 | base_pat = pat_end - 1; |
1711 | ||
f4646fff AS |
1712 | /* A character that does not appear in the pattern induces a |
1713 | stride equal to the pattern length. */ | |
1714 | for (i = 0; i < 0400; i++) | |
1715 | BM_tab[i] = dirlen; | |
facdc750 RS |
1716 | |
1717 | /* We use this for translation, instead of TRT itself. | |
1718 | We fill this in to handle the characters that actually | |
1719 | occur in the pattern. Others don't matter anyway! */ | |
facdc750 RS |
1720 | for (i = 0; i < 0400; i++) |
1721 | simple_translate[i] = i; | |
1722 | ||
a17ae96a | 1723 | if (char_base) |
0190922f | 1724 | { |
a17ae96a | 1725 | /* Setup translate_prev_byte1/2/3/4 from CHAR_BASE. Only a |
0190922f | 1726 | byte following them are the target of translation. */ |
0190922f | 1727 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
1f3561e4 | 1728 | int cblen = CHAR_STRING (char_base, str); |
0190922f | 1729 | |
1f3561e4 PE |
1730 | translate_prev_byte1 = str[cblen - 2]; |
1731 | if (cblen > 2) | |
0190922f | 1732 | { |
1f3561e4 PE |
1733 | translate_prev_byte2 = str[cblen - 3]; |
1734 | if (cblen > 3) | |
a17ae96a | 1735 | { |
1f3561e4 PE |
1736 | translate_prev_byte3 = str[cblen - 4]; |
1737 | if (cblen > 4) | |
1738 | translate_prev_byte4 = str[cblen - 5]; | |
a17ae96a | 1739 | } |
0190922f KH |
1740 | } |
1741 | } | |
1742 | ||
facdc750 | 1743 | i = 0; |
f4646fff | 1744 | while (i != dirlen) |
facdc750 | 1745 | { |
cb6792d2 | 1746 | unsigned char *ptr = base_pat + i; |
facdc750 | 1747 | i += direction; |
facdc750 | 1748 | if (! NILP (trt)) |
ca1d1d23 | 1749 | { |
49e44e81 KH |
1750 | /* If the byte currently looking at is the last of a |
1751 | character to check case-equivalents, set CH to that | |
1752 | character. An ASCII character and a non-ASCII character | |
1753 | matching with CHAR_BASE are to be checked. */ | |
a17ae96a KH |
1754 | int ch = -1; |
1755 | ||
1756 | if (ASCII_BYTE_P (*ptr) || ! multibyte) | |
1757 | ch = *ptr; | |
49e44e81 | 1758 | else if (char_base |
86dc6ccb | 1759 | && ((pat_end - ptr) == 1 || CHAR_HEAD_P (ptr[1]))) |
ca1d1d23 | 1760 | { |
49e44e81 KH |
1761 | unsigned char *charstart = ptr - 1; |
1762 | ||
1763 | while (! (CHAR_HEAD_P (*charstart))) | |
1764 | charstart--; | |
62a6e103 | 1765 | ch = STRING_CHAR (charstart); |
a17ae96a KH |
1766 | if (char_base != (ch & ~0x3F)) |
1767 | ch = -1; | |
ca1d1d23 | 1768 | } |
facdc750 | 1769 | |
f51995fa | 1770 | if (ch >= 0200) |
49e44e81 KH |
1771 | j = (ch & 0x3F) | 0200; |
1772 | else | |
1773 | j = *ptr; | |
1774 | ||
f4646fff | 1775 | if (i == dirlen) |
facdc750 | 1776 | stride_for_teases = BM_tab[j]; |
ab228c24 | 1777 | |
facdc750 RS |
1778 | BM_tab[j] = dirlen - i; |
1779 | /* A translation table is accompanied by its inverse -- see */ | |
177c0ea7 | 1780 | /* comment following downcase_table for details */ |
a17ae96a | 1781 | if (ch >= 0) |
ab228c24 RS |
1782 | { |
1783 | int starting_ch = ch; | |
49e44e81 | 1784 | int starting_j = j; |
a17ae96a | 1785 | |
ab228c24 RS |
1786 | while (1) |
1787 | { | |
1788 | TRANSLATE (ch, inverse_trt, ch); | |
f51995fa | 1789 | if (ch >= 0200) |
49e44e81 | 1790 | j = (ch & 0x3F) | 0200; |
ab228c24 | 1791 | else |
a17ae96a | 1792 | j = ch; |
ab228c24 RS |
1793 | |
1794 | /* For all the characters that map into CH, | |
1795 | set up simple_translate to map the last byte | |
1796 | into STARTING_J. */ | |
1797 | simple_translate[j] = starting_j; | |
1798 | if (ch == starting_ch) | |
1799 | break; | |
1800 | BM_tab[j] = dirlen - i; | |
1801 | } | |
1802 | } | |
facdc750 RS |
1803 | } |
1804 | else | |
1805 | { | |
1806 | j = *ptr; | |
1807 | ||
f4646fff | 1808 | if (i == dirlen) |
facdc750 RS |
1809 | stride_for_teases = BM_tab[j]; |
1810 | BM_tab[j] = dirlen - i; | |
ca1d1d23 | 1811 | } |
f4646fff AS |
1812 | /* stride_for_teases tells how much to stride if we get a |
1813 | match on the far character but are subsequently | |
1814 | disappointed, by recording what the stride would have been | |
1815 | for that character if the last character had been | |
1816 | different. */ | |
facdc750 | 1817 | } |
facdc750 RS |
1818 | pos_byte += dirlen - ((direction > 0) ? direction : 0); |
1819 | /* loop invariant - POS_BYTE points at where last char (first | |
1820 | char if reverse) of pattern would align in a possible match. */ | |
1821 | while (n != 0) | |
1822 | { | |
e7deaab0 | 1823 | EMACS_INT tail_end; |
facdc750 RS |
1824 | unsigned char *tail_end_ptr; |
1825 | ||
1826 | /* It's been reported that some (broken) compiler thinks that | |
1827 | Boolean expressions in an arithmetic context are unsigned. | |
1828 | Using an explicit ?1:0 prevents this. */ | |
1829 | if ((lim_byte - pos_byte - ((direction > 0) ? 1 : 0)) * direction | |
1830 | < 0) | |
1831 | return (n * (0 - direction)); | |
1832 | /* First we do the part we can by pointers (maybe nothing) */ | |
1833 | QUIT; | |
1834 | pat = base_pat; | |
1835 | limit = pos_byte - dirlen + direction; | |
67ce527d KH |
1836 | if (direction > 0) |
1837 | { | |
1838 | limit = BUFFER_CEILING_OF (limit); | |
1839 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1840 | can take on without hitting edge of buffer or the gap. */ | |
1841 | limit = min (limit, pos_byte + 20000); | |
1842 | limit = min (limit, lim_byte - 1); | |
1843 | } | |
1844 | else | |
1845 | { | |
1846 | limit = BUFFER_FLOOR_OF (limit); | |
1847 | /* LIMIT is now the last (not beyond-last!) value POS_BYTE | |
1848 | can take on without hitting edge of buffer or the gap. */ | |
1849 | limit = max (limit, pos_byte - 20000); | |
1850 | limit = max (limit, lim_byte); | |
1851 | } | |
facdc750 RS |
1852 | tail_end = BUFFER_CEILING_OF (pos_byte) + 1; |
1853 | tail_end_ptr = BYTE_POS_ADDR (tail_end); | |
1854 | ||
1855 | if ((limit - pos_byte) * direction > 20) | |
ca1d1d23 | 1856 | { |
facdc750 RS |
1857 | unsigned char *p2; |
1858 | ||
1859 | p_limit = BYTE_POS_ADDR (limit); | |
1860 | p2 = (cursor = BYTE_POS_ADDR (pos_byte)); | |
f4646fff | 1861 | /* In this loop, pos + cursor - p2 is the surrogate for pos. */ |
facdc750 | 1862 | while (1) /* use one cursor setting as long as i can */ |
ca1d1d23 | 1863 | { |
facdc750 | 1864 | if (direction > 0) /* worth duplicating */ |
ca1d1d23 | 1865 | { |
f4646fff AS |
1866 | while (cursor <= p_limit) |
1867 | { | |
1868 | if (BM_tab[*cursor] == 0) | |
1869 | goto hit; | |
facdc750 | 1870 | cursor += BM_tab[*cursor]; |
f4646fff | 1871 | } |
facdc750 RS |
1872 | } |
1873 | else | |
1874 | { | |
f4646fff AS |
1875 | while (cursor >= p_limit) |
1876 | { | |
1877 | if (BM_tab[*cursor] == 0) | |
1878 | goto hit; | |
facdc750 | 1879 | cursor += BM_tab[*cursor]; |
f4646fff | 1880 | } |
facdc750 | 1881 | } |
f4646fff AS |
1882 | /* If you are here, cursor is beyond the end of the |
1883 | searched region. You fail to match within the | |
1884 | permitted region and would otherwise try a character | |
1885 | beyond that region. */ | |
1886 | break; | |
1887 | ||
1888 | hit: | |
facdc750 RS |
1889 | i = dirlen - direction; |
1890 | if (! NILP (trt)) | |
1891 | { | |
1892 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1893 | { |
facdc750 RS |
1894 | int ch; |
1895 | cursor -= direction; | |
1896 | /* Translate only the last byte of a character. */ | |
1897 | if (! multibyte | |
1898 | || ((cursor == tail_end_ptr | |
1899 | || CHAR_HEAD_P (cursor[1])) | |
1900 | && (CHAR_HEAD_P (cursor[0]) | |
0190922f KH |
1901 | /* Check if this is the last byte of |
1902 | a translable character. */ | |
1903 | || (translate_prev_byte1 == cursor[-1] | |
1904 | && (CHAR_HEAD_P (translate_prev_byte1) | |
1905 | || (translate_prev_byte2 == cursor[-2] | |
1906 | && (CHAR_HEAD_P (translate_prev_byte2) | |
1907 | || (translate_prev_byte3 == cursor[-3])))))))) | |
facdc750 RS |
1908 | ch = simple_translate[*cursor]; |
1909 | else | |
1910 | ch = *cursor; | |
1911 | if (pat[i] != ch) | |
1912 | break; | |
ca1d1d23 | 1913 | } |
facdc750 RS |
1914 | } |
1915 | else | |
1916 | { | |
1917 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1918 | { |
facdc750 RS |
1919 | cursor -= direction; |
1920 | if (pat[i] != *cursor) | |
1921 | break; | |
ca1d1d23 | 1922 | } |
facdc750 RS |
1923 | } |
1924 | cursor += dirlen - i - direction; /* fix cursor */ | |
1925 | if (i + direction == 0) | |
1926 | { | |
e7deaab0 | 1927 | EMACS_INT position, start, end; |
0c8533c6 | 1928 | |
facdc750 | 1929 | cursor -= direction; |
1113d9db | 1930 | |
facdc750 RS |
1931 | position = pos_byte + cursor - p2 + ((direction > 0) |
1932 | ? 1 - len_byte : 0); | |
1933 | set_search_regs (position, len_byte); | |
ca325161 | 1934 | |
ef887810 RS |
1935 | if (NILP (Vinhibit_changing_match_data)) |
1936 | { | |
1937 | start = search_regs.start[0]; | |
1938 | end = search_regs.end[0]; | |
1939 | } | |
1940 | else | |
1941 | /* If Vinhibit_changing_match_data is non-nil, | |
1942 | search_regs will not be changed. So let's | |
1943 | compute start and end here. */ | |
1944 | { | |
1945 | start = BYTE_TO_CHAR (position); | |
1946 | end = BYTE_TO_CHAR (position + len_byte); | |
1947 | } | |
1948 | ||
facdc750 RS |
1949 | if ((n -= direction) != 0) |
1950 | cursor += dirlen; /* to resume search */ | |
ca1d1d23 | 1951 | else |
ef887810 | 1952 | return direction > 0 ? end : start; |
ca1d1d23 | 1953 | } |
facdc750 RS |
1954 | else |
1955 | cursor += stride_for_teases; /* <sigh> we lose - */ | |
ca1d1d23 | 1956 | } |
facdc750 RS |
1957 | pos_byte += cursor - p2; |
1958 | } | |
1959 | else | |
f4646fff AS |
1960 | /* Now we'll pick up a clump that has to be done the hard |
1961 | way because it covers a discontinuity. */ | |
facdc750 RS |
1962 | { |
1963 | limit = ((direction > 0) | |
1964 | ? BUFFER_CEILING_OF (pos_byte - dirlen + 1) | |
1965 | : BUFFER_FLOOR_OF (pos_byte - dirlen - 1)); | |
1966 | limit = ((direction > 0) | |
1967 | ? min (limit + len_byte, lim_byte - 1) | |
1968 | : max (limit - len_byte, lim_byte)); | |
1969 | /* LIMIT is now the last value POS_BYTE can have | |
1970 | and still be valid for a possible match. */ | |
1971 | while (1) | |
ca1d1d23 | 1972 | { |
f4646fff AS |
1973 | /* This loop can be coded for space rather than |
1974 | speed because it will usually run only once. | |
1975 | (the reach is at most len + 21, and typically | |
1976 | does not exceed len). */ | |
facdc750 | 1977 | while ((limit - pos_byte) * direction >= 0) |
f4646fff AS |
1978 | { |
1979 | int ch = FETCH_BYTE (pos_byte); | |
1980 | if (BM_tab[ch] == 0) | |
1981 | goto hit2; | |
1982 | pos_byte += BM_tab[ch]; | |
1983 | } | |
1984 | break; /* ran off the end */ | |
1985 | ||
1986 | hit2: | |
1987 | /* Found what might be a match. */ | |
facdc750 RS |
1988 | i = dirlen - direction; |
1989 | while ((i -= direction) + direction != 0) | |
ca1d1d23 | 1990 | { |
facdc750 RS |
1991 | int ch; |
1992 | unsigned char *ptr; | |
1993 | pos_byte -= direction; | |
1994 | ptr = BYTE_POS_ADDR (pos_byte); | |
1995 | /* Translate only the last byte of a character. */ | |
1996 | if (! multibyte | |
1997 | || ((ptr == tail_end_ptr | |
1998 | || CHAR_HEAD_P (ptr[1])) | |
1999 | && (CHAR_HEAD_P (ptr[0]) | |
0190922f KH |
2000 | /* Check if this is the last byte of a |
2001 | translable character. */ | |
2002 | || (translate_prev_byte1 == ptr[-1] | |
2003 | && (CHAR_HEAD_P (translate_prev_byte1) | |
2004 | || (translate_prev_byte2 == ptr[-2] | |
2005 | && (CHAR_HEAD_P (translate_prev_byte2) | |
2006 | || translate_prev_byte3 == ptr[-3]))))))) | |
facdc750 RS |
2007 | ch = simple_translate[*ptr]; |
2008 | else | |
2009 | ch = *ptr; | |
2010 | if (pat[i] != ch) | |
2011 | break; | |
2012 | } | |
2013 | /* Above loop has moved POS_BYTE part or all the way | |
2014 | back to the first pos (last pos if reverse). | |
2015 | Set it once again at the last (first if reverse) char. */ | |
f4646fff | 2016 | pos_byte += dirlen - i - direction; |
facdc750 RS |
2017 | if (i + direction == 0) |
2018 | { | |
e7deaab0 | 2019 | EMACS_INT position, start, end; |
facdc750 | 2020 | pos_byte -= direction; |
1113d9db | 2021 | |
facdc750 | 2022 | position = pos_byte + ((direction > 0) ? 1 - len_byte : 0); |
facdc750 | 2023 | set_search_regs (position, len_byte); |
ca325161 | 2024 | |
ef887810 RS |
2025 | if (NILP (Vinhibit_changing_match_data)) |
2026 | { | |
2027 | start = search_regs.start[0]; | |
2028 | end = search_regs.end[0]; | |
2029 | } | |
2030 | else | |
2031 | /* If Vinhibit_changing_match_data is non-nil, | |
2032 | search_regs will not be changed. So let's | |
2033 | compute start and end here. */ | |
2034 | { | |
2035 | start = BYTE_TO_CHAR (position); | |
2036 | end = BYTE_TO_CHAR (position + len_byte); | |
2037 | } | |
2038 | ||
facdc750 RS |
2039 | if ((n -= direction) != 0) |
2040 | pos_byte += dirlen; /* to resume search */ | |
ca1d1d23 | 2041 | else |
ef887810 | 2042 | return direction > 0 ? end : start; |
ca1d1d23 | 2043 | } |
facdc750 RS |
2044 | else |
2045 | pos_byte += stride_for_teases; | |
2046 | } | |
2047 | } | |
2048 | /* We have done one clump. Can we continue? */ | |
2049 | if ((lim_byte - pos_byte) * direction < 0) | |
2050 | return ((0 - n) * direction); | |
ca1d1d23 | 2051 | } |
facdc750 | 2052 | return BYTE_TO_CHAR (pos_byte); |
ca1d1d23 | 2053 | } |
ca325161 | 2054 | |
fa8ed3e0 | 2055 | /* Record beginning BEG_BYTE and end BEG_BYTE + NBYTES |
a7e4cdde RS |
2056 | for the overall match just found in the current buffer. |
2057 | Also clear out the match data for registers 1 and up. */ | |
ca325161 RS |
2058 | |
2059 | static void | |
971de7fb | 2060 | set_search_regs (EMACS_INT beg_byte, EMACS_INT nbytes) |
ca325161 | 2061 | { |
a7e4cdde RS |
2062 | int i; |
2063 | ||
ef887810 RS |
2064 | if (!NILP (Vinhibit_changing_match_data)) |
2065 | return; | |
2066 | ||
ca325161 RS |
2067 | /* Make sure we have registers in which to store |
2068 | the match position. */ | |
2069 | if (search_regs.num_regs == 0) | |
2070 | { | |
2d4a771a RS |
2071 | search_regs.start = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
2072 | search_regs.end = (regoff_t *) xmalloc (2 * sizeof (regoff_t)); | |
487282dc | 2073 | search_regs.num_regs = 2; |
ca325161 RS |
2074 | } |
2075 | ||
a7e4cdde RS |
2076 | /* Clear out the other registers. */ |
2077 | for (i = 1; i < search_regs.num_regs; i++) | |
2078 | { | |
2079 | search_regs.start[i] = -1; | |
2080 | search_regs.end[i] = -1; | |
2081 | } | |
2082 | ||
fa8ed3e0 RS |
2083 | search_regs.start[0] = BYTE_TO_CHAR (beg_byte); |
2084 | search_regs.end[0] = BYTE_TO_CHAR (beg_byte + nbytes); | |
a3668d92 | 2085 | XSETBUFFER (last_thing_searched, current_buffer); |
ca325161 | 2086 | } |
ca1d1d23 | 2087 | \f |
c6e0e1cc CY |
2088 | /* Given STRING, a string of words separated by word delimiters, |
2089 | compute a regexp that matches those exact words separated by | |
2090 | arbitrary punctuation. If LAX is nonzero, the end of the string | |
2091 | need not match a word boundary unless it ends in whitespace. */ | |
ca1d1d23 JB |
2092 | |
2093 | static Lisp_Object | |
971de7fb | 2094 | wordify (Lisp_Object string, int lax) |
ca1d1d23 JB |
2095 | { |
2096 | register unsigned char *p, *o; | |
c098fdb8 | 2097 | register EMACS_INT i, i_byte, len, punct_count = 0, word_count = 0; |
ca1d1d23 | 2098 | Lisp_Object val; |
0c8533c6 | 2099 | int prev_c = 0; |
c098fdb8 EZ |
2100 | EMACS_INT adjust; |
2101 | int whitespace_at_end; | |
ca1d1d23 | 2102 | |
b7826503 | 2103 | CHECK_STRING (string); |
d5db4077 KR |
2104 | p = SDATA (string); |
2105 | len = SCHARS (string); | |
ca1d1d23 | 2106 | |
0c8533c6 RS |
2107 | for (i = 0, i_byte = 0; i < len; ) |
2108 | { | |
2109 | int c; | |
177c0ea7 | 2110 | |
93daa011 | 2111 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
0c8533c6 RS |
2112 | |
2113 | if (SYNTAX (c) != Sword) | |
2114 | { | |
2115 | punct_count++; | |
2116 | if (i > 0 && SYNTAX (prev_c) == Sword) | |
2117 | word_count++; | |
2118 | } | |
ca1d1d23 | 2119 | |
0c8533c6 RS |
2120 | prev_c = c; |
2121 | } | |
2122 | ||
2123 | if (SYNTAX (prev_c) == Sword) | |
c6e0e1cc CY |
2124 | { |
2125 | word_count++; | |
2126 | whitespace_at_end = 0; | |
2127 | } | |
2128 | else | |
2129 | whitespace_at_end = 1; | |
2130 | ||
0c8533c6 | 2131 | if (!word_count) |
c60416e0 | 2132 | return empty_unibyte_string; |
0c8533c6 | 2133 | |
c6e0e1cc CY |
2134 | adjust = - punct_count + 5 * (word_count - 1) |
2135 | + ((lax && !whitespace_at_end) ? 2 : 4); | |
8a2df937 RS |
2136 | if (STRING_MULTIBYTE (string)) |
2137 | val = make_uninit_multibyte_string (len + adjust, | |
d5db4077 | 2138 | SBYTES (string) |
8a2df937 RS |
2139 | + adjust); |
2140 | else | |
2141 | val = make_uninit_string (len + adjust); | |
ca1d1d23 | 2142 | |
d5db4077 | 2143 | o = SDATA (val); |
ca1d1d23 JB |
2144 | *o++ = '\\'; |
2145 | *o++ = 'b'; | |
1e9582d4 | 2146 | prev_c = 0; |
ca1d1d23 | 2147 | |
1e9582d4 RS |
2148 | for (i = 0, i_byte = 0; i < len; ) |
2149 | { | |
2150 | int c; | |
c098fdb8 | 2151 | EMACS_INT i_byte_orig = i_byte; |
177c0ea7 | 2152 | |
93daa011 | 2153 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, i, i_byte); |
1e9582d4 RS |
2154 | |
2155 | if (SYNTAX (c) == Sword) | |
2156 | { | |
72af86bd | 2157 | memcpy (o, SDATA (string) + i_byte_orig, i_byte - i_byte_orig); |
1e9582d4 RS |
2158 | o += i_byte - i_byte_orig; |
2159 | } | |
2160 | else if (i > 0 && SYNTAX (prev_c) == Sword && --word_count) | |
2161 | { | |
2162 | *o++ = '\\'; | |
2163 | *o++ = 'W'; | |
2164 | *o++ = '\\'; | |
2165 | *o++ = 'W'; | |
2166 | *o++ = '*'; | |
2167 | } | |
2168 | ||
2169 | prev_c = c; | |
2170 | } | |
ca1d1d23 | 2171 | |
c6e0e1cc CY |
2172 | if (!lax || whitespace_at_end) |
2173 | { | |
2174 | *o++ = '\\'; | |
2175 | *o++ = 'b'; | |
2176 | } | |
ca1d1d23 JB |
2177 | |
2178 | return val; | |
2179 | } | |
2180 | \f | |
2181 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, | |
8c1a1077 PJ |
2182 | "MSearch backward: ", |
2183 | doc: /* Search backward from point for STRING. | |
2184 | Set point to the beginning of the occurrence found, and return point. | |
2185 | An optional second argument bounds the search; it is a buffer position. | |
2186 | The match found must not extend before that position. | |
2187 | Optional third argument, if t, means if fail just return nil (no error). | |
2188 | If not nil and not t, position at limit of search and return nil. | |
2189 | Optional fourth argument is repeat count--search for successive occurrences. | |
2190 | ||
2191 | Search case-sensitivity is determined by the value of the variable | |
2192 | `case-fold-search', which see. | |
2193 | ||
2194 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2195 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2196 | { |
b819a390 | 2197 | return search_command (string, bound, noerror, count, -1, 0, 0); |
ca1d1d23 JB |
2198 | } |
2199 | ||
6af43974 | 2200 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "MSearch: ", |
8c1a1077 PJ |
2201 | doc: /* Search forward from point for STRING. |
2202 | Set point to the end of the occurrence found, and return point. | |
2203 | An optional second argument bounds the search; it is a buffer position. | |
d6aacdcd | 2204 | The match found must not extend after that position. A value of nil is |
48740f01 | 2205 | equivalent to (point-max). |
8c1a1077 PJ |
2206 | Optional third argument, if t, means if fail just return nil (no error). |
2207 | If not nil and not t, move to limit of search and return nil. | |
2208 | Optional fourth argument is repeat count--search for successive occurrences. | |
2209 | ||
2210 | Search case-sensitivity is determined by the value of the variable | |
2211 | `case-fold-search', which see. | |
2212 | ||
2213 | See also the functions `match-beginning', `match-end' and `replace-match'. */) | |
5842a27b | 2214 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2215 | { |
b819a390 | 2216 | return search_command (string, bound, noerror, count, 1, 0, 0); |
ca1d1d23 JB |
2217 | } |
2218 | ||
2219 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, | |
8c1a1077 PJ |
2220 | "sWord search backward: ", |
2221 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2222 | Set point to the beginning of the occurrence found, and return point. | |
2223 | An optional second argument bounds the search; it is a buffer position. | |
2224 | The match found must not extend before that position. | |
2225 | Optional third argument, if t, means if fail just return nil (no error). | |
2226 | If not nil and not t, move to limit of search and return nil. | |
2227 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2228 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2229 | { |
c6e0e1cc | 2230 | return search_command (wordify (string, 0), bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2231 | } |
2232 | ||
2233 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, | |
8c1a1077 PJ |
2234 | "sWord search: ", |
2235 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2236 | Set point to the end of the occurrence found, and return point. | |
2237 | An optional second argument bounds the search; it is a buffer position. | |
2238 | The match found must not extend after that position. | |
2239 | Optional third argument, if t, means if fail just return nil (no error). | |
2240 | If not nil and not t, move to limit of search and return nil. | |
2241 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2242 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2243 | { |
c6e0e1cc CY |
2244 | return search_command (wordify (string, 0), bound, noerror, count, 1, 1, 0); |
2245 | } | |
2246 | ||
2247 | DEFUN ("word-search-backward-lax", Fword_search_backward_lax, Sword_search_backward_lax, 1, 4, | |
2248 | "sWord search backward: ", | |
2249 | doc: /* Search backward from point for STRING, ignoring differences in punctuation. | |
2250 | Set point to the beginning of the occurrence found, and return point. | |
2251 | ||
2252 | Unlike `word-search-backward', the end of STRING need not match a word | |
2253 | boundary unless it ends in whitespace. | |
2254 | ||
2255 | An optional second argument bounds the search; it is a buffer position. | |
2256 | The match found must not extend before that position. | |
2257 | Optional third argument, if t, means if fail just return nil (no error). | |
2258 | If not nil and not t, move to limit of search and return nil. | |
2259 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2260 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
c6e0e1cc CY |
2261 | { |
2262 | return search_command (wordify (string, 1), bound, noerror, count, -1, 1, 0); | |
2263 | } | |
2264 | ||
2265 | DEFUN ("word-search-forward-lax", Fword_search_forward_lax, Sword_search_forward_lax, 1, 4, | |
2266 | "sWord search: ", | |
2267 | doc: /* Search forward from point for STRING, ignoring differences in punctuation. | |
2268 | Set point to the end of the occurrence found, and return point. | |
2269 | ||
2270 | Unlike `word-search-forward', the end of STRING need not match a word | |
2271 | boundary unless it ends in whitespace. | |
2272 | ||
2273 | An optional second argument bounds the search; it is a buffer position. | |
2274 | The match found must not extend after that position. | |
2275 | Optional third argument, if t, means if fail just return nil (no error). | |
2276 | If not nil and not t, move to limit of search and return nil. | |
2277 | Optional fourth argument is repeat count--search for successive occurrences. */) | |
5842a27b | 2278 | (Lisp_Object string, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
c6e0e1cc CY |
2279 | { |
2280 | return search_command (wordify (string, 1), bound, noerror, count, 1, 1, 0); | |
ca1d1d23 JB |
2281 | } |
2282 | ||
2283 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, | |
8c1a1077 PJ |
2284 | "sRE search backward: ", |
2285 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2286 | Set point to the beginning of the match, and return point. | |
2287 | The match found is the one starting last in the buffer | |
2288 | and yet ending before the origin of the search. | |
2289 | An optional second argument bounds the search; it is a buffer position. | |
2290 | The match found must start at or after that position. | |
2291 | Optional third argument, if t, means if fail just return nil (no error). | |
2292 | If not nil and not t, move to limit of search and return nil. | |
2293 | Optional fourth argument is repeat count--search for successive occurrences. | |
2294 | See also the functions `match-beginning', `match-end', `match-string', | |
2295 | and `replace-match'. */) | |
5842a27b | 2296 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2297 | { |
b819a390 | 2298 | return search_command (regexp, bound, noerror, count, -1, 1, 0); |
ca1d1d23 JB |
2299 | } |
2300 | ||
2301 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, | |
8c1a1077 PJ |
2302 | "sRE search: ", |
2303 | doc: /* Search forward from point for regular expression REGEXP. | |
2304 | Set point to the end of the occurrence found, and return point. | |
2305 | An optional second argument bounds the search; it is a buffer position. | |
2306 | The match found must not extend after that position. | |
2307 | Optional third argument, if t, means if fail just return nil (no error). | |
2308 | If not nil and not t, move to limit of search and return nil. | |
2309 | Optional fourth argument is repeat count--search for successive occurrences. | |
2310 | See also the functions `match-beginning', `match-end', `match-string', | |
2311 | and `replace-match'. */) | |
5842a27b | 2312 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
ca1d1d23 | 2313 | { |
b819a390 RS |
2314 | return search_command (regexp, bound, noerror, count, 1, 1, 0); |
2315 | } | |
2316 | ||
2317 | DEFUN ("posix-search-backward", Fposix_search_backward, Sposix_search_backward, 1, 4, | |
8c1a1077 PJ |
2318 | "sPosix search backward: ", |
2319 | doc: /* Search backward from point for match for regular expression REGEXP. | |
2320 | Find the longest match in accord with Posix regular expression rules. | |
2321 | Set point to the beginning of the match, and return point. | |
2322 | The match found is the one starting last in the buffer | |
2323 | and yet ending before the origin of the search. | |
2324 | An optional second argument bounds the search; it is a buffer position. | |
2325 | The match found must start at or after that position. | |
2326 | Optional third argument, if t, means if fail just return nil (no error). | |
2327 | If not nil and not t, move to limit of search and return nil. | |
2328 | Optional fourth argument is repeat count--search for successive occurrences. | |
2329 | See also the functions `match-beginning', `match-end', `match-string', | |
2330 | and `replace-match'. */) | |
5842a27b | 2331 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2332 | { |
2333 | return search_command (regexp, bound, noerror, count, -1, 1, 1); | |
2334 | } | |
2335 | ||
2336 | DEFUN ("posix-search-forward", Fposix_search_forward, Sposix_search_forward, 1, 4, | |
8c1a1077 PJ |
2337 | "sPosix search: ", |
2338 | doc: /* Search forward from point for regular expression REGEXP. | |
2339 | Find the longest match in accord with Posix regular expression rules. | |
2340 | Set point to the end of the occurrence found, and return point. | |
2341 | An optional second argument bounds the search; it is a buffer position. | |
2342 | The match found must not extend after that position. | |
2343 | Optional third argument, if t, means if fail just return nil (no error). | |
2344 | If not nil and not t, move to limit of search and return nil. | |
2345 | Optional fourth argument is repeat count--search for successive occurrences. | |
2346 | See also the functions `match-beginning', `match-end', `match-string', | |
2347 | and `replace-match'. */) | |
5842a27b | 2348 | (Lisp_Object regexp, Lisp_Object bound, Lisp_Object noerror, Lisp_Object count) |
b819a390 RS |
2349 | { |
2350 | return search_command (regexp, bound, noerror, count, 1, 1, 1); | |
ca1d1d23 JB |
2351 | } |
2352 | \f | |
d7a5ad5f | 2353 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 5, 0, |
8c1a1077 | 2354 | doc: /* Replace text matched by last search with NEWTEXT. |
4dd0c271 RS |
2355 | Leave point at the end of the replacement text. |
2356 | ||
8c1a1077 PJ |
2357 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text. |
2358 | Otherwise maybe capitalize the whole text, or maybe just word initials, | |
2359 | based on the replaced text. | |
2360 | If the replaced text has only capital letters | |
2361 | and has at least one multiletter word, convert NEWTEXT to all caps. | |
4dd0c271 RS |
2362 | Otherwise if all words are capitalized in the replaced text, |
2363 | capitalize each word in NEWTEXT. | |
8c1a1077 PJ |
2364 | |
2365 | If third arg LITERAL is non-nil, insert NEWTEXT literally. | |
2366 | Otherwise treat `\\' as special: | |
2367 | `\\&' in NEWTEXT means substitute original matched text. | |
2368 | `\\N' means substitute what matched the Nth `\\(...\\)'. | |
2369 | If Nth parens didn't match, substitute nothing. | |
2370 | `\\\\' means insert one `\\'. | |
4dd0c271 RS |
2371 | Case conversion does not apply to these substitutions. |
2372 | ||
8c1a1077 | 2373 | FIXEDCASE and LITERAL are optional arguments. |
8c1a1077 PJ |
2374 | |
2375 | The optional fourth argument STRING can be a string to modify. | |
2376 | This is meaningful when the previous match was done against STRING, | |
2377 | using `string-match'. When used this way, `replace-match' | |
2378 | creates and returns a new string made by copying STRING and replacing | |
2379 | the part of STRING that was matched. | |
2380 | ||
2381 | The optional fifth argument SUBEXP specifies a subexpression; | |
2382 | it says to replace just that subexpression with NEWTEXT, | |
2383 | rather than replacing the entire matched text. | |
2384 | This is, in a vague sense, the inverse of using `\\N' in NEWTEXT; | |
2385 | `\\N' copies subexp N into NEWTEXT, but using N as SUBEXP puts | |
2386 | NEWTEXT in place of subexp N. | |
2387 | This is useful only after a regular expression search or match, | |
2388 | since only regular expressions have distinguished subexpressions. */) | |
5842a27b | 2389 | (Lisp_Object newtext, Lisp_Object fixedcase, Lisp_Object literal, Lisp_Object string, Lisp_Object subexp) |
ca1d1d23 JB |
2390 | { |
2391 | enum { nochange, all_caps, cap_initial } case_action; | |
c098fdb8 | 2392 | register EMACS_INT pos, pos_byte; |
ca1d1d23 | 2393 | int some_multiletter_word; |
97832bd0 | 2394 | int some_lowercase; |
73dc8771 | 2395 | int some_uppercase; |
208767c3 | 2396 | int some_nonuppercase_initial; |
ca1d1d23 | 2397 | register int c, prevc; |
d7a5ad5f | 2398 | int sub; |
e7deaab0 | 2399 | EMACS_INT opoint, newpoint; |
ca1d1d23 | 2400 | |
b7826503 | 2401 | CHECK_STRING (newtext); |
ca1d1d23 | 2402 | |
080c45fd | 2403 | if (! NILP (string)) |
b7826503 | 2404 | CHECK_STRING (string); |
080c45fd | 2405 | |
ca1d1d23 JB |
2406 | case_action = nochange; /* We tried an initialization */ |
2407 | /* but some C compilers blew it */ | |
4746118a JB |
2408 | |
2409 | if (search_regs.num_regs <= 0) | |
d72cdbfc | 2410 | error ("`replace-match' called before any match found"); |
4746118a | 2411 | |
d7a5ad5f RS |
2412 | if (NILP (subexp)) |
2413 | sub = 0; | |
2414 | else | |
2415 | { | |
b7826503 | 2416 | CHECK_NUMBER (subexp); |
d7a5ad5f RS |
2417 | sub = XINT (subexp); |
2418 | if (sub < 0 || sub >= search_regs.num_regs) | |
2419 | args_out_of_range (subexp, make_number (search_regs.num_regs)); | |
2420 | } | |
2421 | ||
080c45fd RS |
2422 | if (NILP (string)) |
2423 | { | |
d7a5ad5f RS |
2424 | if (search_regs.start[sub] < BEGV |
2425 | || search_regs.start[sub] > search_regs.end[sub] | |
2426 | || search_regs.end[sub] > ZV) | |
2427 | args_out_of_range (make_number (search_regs.start[sub]), | |
2428 | make_number (search_regs.end[sub])); | |
080c45fd RS |
2429 | } |
2430 | else | |
2431 | { | |
d7a5ad5f RS |
2432 | if (search_regs.start[sub] < 0 |
2433 | || search_regs.start[sub] > search_regs.end[sub] | |
d5db4077 | 2434 | || search_regs.end[sub] > SCHARS (string)) |
d7a5ad5f RS |
2435 | args_out_of_range (make_number (search_regs.start[sub]), |
2436 | make_number (search_regs.end[sub])); | |
080c45fd | 2437 | } |
ca1d1d23 JB |
2438 | |
2439 | if (NILP (fixedcase)) | |
2440 | { | |
2441 | /* Decide how to casify by examining the matched text. */ | |
e7deaab0 | 2442 | EMACS_INT last; |
ca1d1d23 | 2443 | |
ac3b28b1 KH |
2444 | pos = search_regs.start[sub]; |
2445 | last = search_regs.end[sub]; | |
fa8ed3e0 RS |
2446 | |
2447 | if (NILP (string)) | |
ac3b28b1 | 2448 | pos_byte = CHAR_TO_BYTE (pos); |
fa8ed3e0 | 2449 | else |
ac3b28b1 | 2450 | pos_byte = string_char_to_byte (string, pos); |
fa8ed3e0 | 2451 | |
ca1d1d23 JB |
2452 | prevc = '\n'; |
2453 | case_action = all_caps; | |
2454 | ||
2455 | /* some_multiletter_word is set nonzero if any original word | |
2456 | is more than one letter long. */ | |
2457 | some_multiletter_word = 0; | |
97832bd0 | 2458 | some_lowercase = 0; |
208767c3 | 2459 | some_nonuppercase_initial = 0; |
73dc8771 | 2460 | some_uppercase = 0; |
ca1d1d23 | 2461 | |
ac3b28b1 | 2462 | while (pos < last) |
ca1d1d23 | 2463 | { |
080c45fd | 2464 | if (NILP (string)) |
ac3b28b1 | 2465 | { |
93daa011 | 2466 | c = FETCH_CHAR_AS_MULTIBYTE (pos_byte); |
ac3b28b1 KH |
2467 | INC_BOTH (pos, pos_byte); |
2468 | } | |
080c45fd | 2469 | else |
93daa011 | 2470 | FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE (c, string, pos, pos_byte); |
080c45fd | 2471 | |
ca1d1d23 JB |
2472 | if (LOWERCASEP (c)) |
2473 | { | |
2474 | /* Cannot be all caps if any original char is lower case */ | |
2475 | ||
97832bd0 | 2476 | some_lowercase = 1; |
ca1d1d23 | 2477 | if (SYNTAX (prevc) != Sword) |
208767c3 | 2478 | some_nonuppercase_initial = 1; |
ca1d1d23 JB |
2479 | else |
2480 | some_multiletter_word = 1; | |
2481 | } | |
d16c2b66 | 2482 | else if (UPPERCASEP (c)) |
ca1d1d23 | 2483 | { |
73dc8771 | 2484 | some_uppercase = 1; |
97832bd0 | 2485 | if (SYNTAX (prevc) != Sword) |
c4d460ce | 2486 | ; |
97832bd0 | 2487 | else |
ca1d1d23 JB |
2488 | some_multiletter_word = 1; |
2489 | } | |
208767c3 RS |
2490 | else |
2491 | { | |
2492 | /* If the initial is a caseless word constituent, | |
2493 | treat that like a lowercase initial. */ | |
2494 | if (SYNTAX (prevc) != Sword) | |
2495 | some_nonuppercase_initial = 1; | |
2496 | } | |
ca1d1d23 JB |
2497 | |
2498 | prevc = c; | |
2499 | } | |
2500 | ||
97832bd0 RS |
2501 | /* Convert to all caps if the old text is all caps |
2502 | and has at least one multiletter word. */ | |
2503 | if (! some_lowercase && some_multiletter_word) | |
2504 | case_action = all_caps; | |
c4d460ce | 2505 | /* Capitalize each word, if the old text has all capitalized words. */ |
208767c3 | 2506 | else if (!some_nonuppercase_initial && some_multiletter_word) |
ca1d1d23 | 2507 | case_action = cap_initial; |
208767c3 | 2508 | else if (!some_nonuppercase_initial && some_uppercase) |
73dc8771 KH |
2509 | /* Should x -> yz, operating on X, give Yz or YZ? |
2510 | We'll assume the latter. */ | |
2511 | case_action = all_caps; | |
97832bd0 RS |
2512 | else |
2513 | case_action = nochange; | |
ca1d1d23 JB |
2514 | } |
2515 | ||
080c45fd RS |
2516 | /* Do replacement in a string. */ |
2517 | if (!NILP (string)) | |
2518 | { | |
2519 | Lisp_Object before, after; | |
2520 | ||
2521 | before = Fsubstring (string, make_number (0), | |
d7a5ad5f RS |
2522 | make_number (search_regs.start[sub])); |
2523 | after = Fsubstring (string, make_number (search_regs.end[sub]), Qnil); | |
080c45fd | 2524 | |
636a5e28 RS |
2525 | /* Substitute parts of the match into NEWTEXT |
2526 | if desired. */ | |
080c45fd RS |
2527 | if (NILP (literal)) |
2528 | { | |
e7deaab0 AS |
2529 | EMACS_INT lastpos = 0; |
2530 | EMACS_INT lastpos_byte = 0; | |
080c45fd RS |
2531 | /* We build up the substituted string in ACCUM. */ |
2532 | Lisp_Object accum; | |
2533 | Lisp_Object middle; | |
d5db4077 | 2534 | int length = SBYTES (newtext); |
080c45fd RS |
2535 | |
2536 | accum = Qnil; | |
2537 | ||
ac3b28b1 | 2538 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
080c45fd | 2539 | { |
c098fdb8 EZ |
2540 | EMACS_INT substart = -1; |
2541 | EMACS_INT subend = 0; | |
1e79ec24 | 2542 | int delbackslash = 0; |
080c45fd | 2543 | |
0c8533c6 RS |
2544 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
2545 | ||
080c45fd RS |
2546 | if (c == '\\') |
2547 | { | |
0c8533c6 | 2548 | FETCH_STRING_CHAR_ADVANCE (c, newtext, pos, pos_byte); |
177c0ea7 | 2549 | |
080c45fd RS |
2550 | if (c == '&') |
2551 | { | |
d7a5ad5f RS |
2552 | substart = search_regs.start[sub]; |
2553 | subend = search_regs.end[sub]; | |
080c45fd | 2554 | } |
76fc1ea2 | 2555 | else if (c >= '1' && c <= '9') |
080c45fd | 2556 | { |
76fc1ea2 KH |
2557 | if (search_regs.start[c - '0'] >= 0 |
2558 | && c <= search_regs.num_regs + '0') | |
080c45fd RS |
2559 | { |
2560 | substart = search_regs.start[c - '0']; | |
2561 | subend = search_regs.end[c - '0']; | |
2562 | } | |
76fc1ea2 KH |
2563 | else |
2564 | { | |
2565 | /* If that subexp did not match, | |
2566 | replace \\N with nothing. */ | |
2567 | substart = 0; | |
2568 | subend = 0; | |
2569 | } | |
080c45fd | 2570 | } |
1e79ec24 KH |
2571 | else if (c == '\\') |
2572 | delbackslash = 1; | |
636a5e28 RS |
2573 | else |
2574 | error ("Invalid use of `\\' in replacement text"); | |
080c45fd RS |
2575 | } |
2576 | if (substart >= 0) | |
2577 | { | |
d131e79c RS |
2578 | if (pos - 2 != lastpos) |
2579 | middle = substring_both (newtext, lastpos, | |
2580 | lastpos_byte, | |
2581 | pos - 2, pos_byte - 2); | |
080c45fd RS |
2582 | else |
2583 | middle = Qnil; | |
2584 | accum = concat3 (accum, middle, | |
0c8533c6 RS |
2585 | Fsubstring (string, |
2586 | make_number (substart), | |
080c45fd RS |
2587 | make_number (subend))); |
2588 | lastpos = pos; | |
0c8533c6 | 2589 | lastpos_byte = pos_byte; |
080c45fd | 2590 | } |
1e79ec24 KH |
2591 | else if (delbackslash) |
2592 | { | |
d131e79c RS |
2593 | middle = substring_both (newtext, lastpos, |
2594 | lastpos_byte, | |
2595 | pos - 1, pos_byte - 1); | |
0c8533c6 | 2596 | |
1e79ec24 KH |
2597 | accum = concat2 (accum, middle); |
2598 | lastpos = pos; | |
0c8533c6 | 2599 | lastpos_byte = pos_byte; |
1e79ec24 | 2600 | } |
080c45fd RS |
2601 | } |
2602 | ||
d131e79c RS |
2603 | if (pos != lastpos) |
2604 | middle = substring_both (newtext, lastpos, | |
2605 | lastpos_byte, | |
0c8533c6 | 2606 | pos, pos_byte); |
080c45fd RS |
2607 | else |
2608 | middle = Qnil; | |
2609 | ||
2610 | newtext = concat2 (accum, middle); | |
2611 | } | |
2612 | ||
636a5e28 | 2613 | /* Do case substitution in NEWTEXT if desired. */ |
080c45fd RS |
2614 | if (case_action == all_caps) |
2615 | newtext = Fupcase (newtext); | |
2616 | else if (case_action == cap_initial) | |
2b2eead9 | 2617 | newtext = Fupcase_initials (newtext); |
080c45fd RS |
2618 | |
2619 | return concat3 (before, newtext, after); | |
2620 | } | |
2621 | ||
09c4719e | 2622 | /* Record point, then move (quietly) to the start of the match. */ |
9160906f | 2623 | if (PT >= search_regs.end[sub]) |
b0eba991 | 2624 | opoint = PT - ZV; |
9160906f RS |
2625 | else if (PT > search_regs.start[sub]) |
2626 | opoint = search_regs.end[sub] - ZV; | |
b0eba991 RS |
2627 | else |
2628 | opoint = PT; | |
2629 | ||
886ed6ec RS |
2630 | /* If we want non-literal replacement, |
2631 | perform substitution on the replacement string. */ | |
2632 | if (NILP (literal)) | |
ca1d1d23 | 2633 | { |
c098fdb8 | 2634 | EMACS_INT length = SBYTES (newtext); |
68e69fbd | 2635 | unsigned char *substed; |
c098fdb8 | 2636 | EMACS_INT substed_alloc_size, substed_len; |
4b4deea2 | 2637 | int buf_multibyte = !NILP (BVAR (current_buffer, enable_multibyte_characters)); |
3bc25e52 KH |
2638 | int str_multibyte = STRING_MULTIBYTE (newtext); |
2639 | Lisp_Object rev_tbl; | |
886ed6ec | 2640 | int really_changed = 0; |
3bc25e52 | 2641 | |
8f924df7 | 2642 | rev_tbl = Qnil; |
ac3b28b1 | 2643 | |
68e69fbd RS |
2644 | substed_alloc_size = length * 2 + 100; |
2645 | substed = (unsigned char *) xmalloc (substed_alloc_size + 1); | |
2646 | substed_len = 0; | |
2647 | ||
3bc25e52 KH |
2648 | /* Go thru NEWTEXT, producing the actual text to insert in |
2649 | SUBSTED while adjusting multibyteness to that of the current | |
2650 | buffer. */ | |
ca1d1d23 | 2651 | |
ac3b28b1 | 2652 | for (pos_byte = 0, pos = 0; pos_byte < length;) |
ca1d1d23 | 2653 | { |
68e69fbd | 2654 | unsigned char str[MAX_MULTIBYTE_LENGTH]; |
8ea90aa3 | 2655 | const unsigned char *add_stuff = NULL; |
c098fdb8 | 2656 | EMACS_INT add_len = 0; |
68e69fbd | 2657 | int idx = -1; |
9a76659d | 2658 | |
3bc25e52 KH |
2659 | if (str_multibyte) |
2660 | { | |
eb99a8dd | 2661 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, pos, pos_byte); |
3bc25e52 KH |
2662 | if (!buf_multibyte) |
2663 | c = multibyte_char_to_unibyte (c, rev_tbl); | |
2664 | } | |
2665 | else | |
2666 | { | |
2667 | /* Note that we don't have to increment POS. */ | |
5d69fe10 | 2668 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2669 | if (buf_multibyte) |
4c0354d7 | 2670 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 | 2671 | } |
ac3b28b1 | 2672 | |
68e69fbd RS |
2673 | /* Either set ADD_STUFF and ADD_LEN to the text to put in SUBSTED, |
2674 | or set IDX to a match index, which means put that part | |
2675 | of the buffer text into SUBSTED. */ | |
2676 | ||
ca1d1d23 JB |
2677 | if (c == '\\') |
2678 | { | |
886ed6ec RS |
2679 | really_changed = 1; |
2680 | ||
3bc25e52 KH |
2681 | if (str_multibyte) |
2682 | { | |
eb99a8dd KH |
2683 | FETCH_STRING_CHAR_ADVANCE_NO_CHECK (c, newtext, |
2684 | pos, pos_byte); | |
071ce769 | 2685 | if (!buf_multibyte && !ASCII_CHAR_P (c)) |
3bc25e52 KH |
2686 | c = multibyte_char_to_unibyte (c, rev_tbl); |
2687 | } | |
2688 | else | |
2689 | { | |
d5db4077 | 2690 | c = SREF (newtext, pos_byte++); |
3bc25e52 | 2691 | if (buf_multibyte) |
4c0354d7 | 2692 | MAKE_CHAR_MULTIBYTE (c); |
3bc25e52 KH |
2693 | } |
2694 | ||
ca1d1d23 | 2695 | if (c == '&') |
68e69fbd | 2696 | idx = sub; |
78445046 | 2697 | else if (c >= '1' && c <= '9' && c <= search_regs.num_regs + '0') |
ca1d1d23 JB |
2698 | { |
2699 | if (search_regs.start[c - '0'] >= 1) | |
68e69fbd | 2700 | idx = c - '0'; |
ca1d1d23 | 2701 | } |
636a5e28 | 2702 | else if (c == '\\') |
a7e979a4 | 2703 | add_len = 1, add_stuff = (unsigned char *) "\\"; |
636a5e28 | 2704 | else |
3bc25e52 KH |
2705 | { |
2706 | xfree (substed); | |
2707 | error ("Invalid use of `\\' in replacement text"); | |
2708 | } | |
ca1d1d23 JB |
2709 | } |
2710 | else | |
68e69fbd RS |
2711 | { |
2712 | add_len = CHAR_STRING (c, str); | |
2713 | add_stuff = str; | |
2714 | } | |
2715 | ||
2716 | /* If we want to copy part of a previous match, | |
2717 | set up ADD_STUFF and ADD_LEN to point to it. */ | |
2718 | if (idx >= 0) | |
2719 | { | |
e7deaab0 | 2720 | EMACS_INT begbyte = CHAR_TO_BYTE (search_regs.start[idx]); |
68e69fbd RS |
2721 | add_len = CHAR_TO_BYTE (search_regs.end[idx]) - begbyte; |
2722 | if (search_regs.start[idx] < GPT && GPT < search_regs.end[idx]) | |
2723 | move_gap (search_regs.start[idx]); | |
2724 | add_stuff = BYTE_POS_ADDR (begbyte); | |
2725 | } | |
2726 | ||
2727 | /* Now the stuff we want to add to SUBSTED | |
2728 | is invariably ADD_LEN bytes starting at ADD_STUFF. */ | |
2729 | ||
2730 | /* Make sure SUBSTED is big enough. */ | |
2731 | if (substed_len + add_len >= substed_alloc_size) | |
2732 | { | |
2733 | substed_alloc_size = substed_len + add_len + 500; | |
2734 | substed = (unsigned char *) xrealloc (substed, | |
2735 | substed_alloc_size + 1); | |
2736 | } | |
2737 | ||
2738 | /* Now add to the end of SUBSTED. */ | |
f8ce8a0d GM |
2739 | if (add_stuff) |
2740 | { | |
72af86bd | 2741 | memcpy (substed + substed_len, add_stuff, add_len); |
f8ce8a0d GM |
2742 | substed_len += add_len; |
2743 | } | |
ca1d1d23 | 2744 | } |
68e69fbd | 2745 | |
886ed6ec | 2746 | if (really_changed) |
76fc1ea2 KH |
2747 | { |
2748 | if (buf_multibyte) | |
2749 | { | |
c098fdb8 EZ |
2750 | EMACS_INT nchars = |
2751 | multibyte_chars_in_text (substed, substed_len); | |
68e69fbd | 2752 | |
a7e979a4 PE |
2753 | newtext = make_multibyte_string ((char *) substed, nchars, |
2754 | substed_len); | |
76fc1ea2 KH |
2755 | } |
2756 | else | |
a7e979a4 | 2757 | newtext = make_unibyte_string ((char *) substed, substed_len); |
76fc1ea2 | 2758 | } |
68e69fbd | 2759 | xfree (substed); |
ca1d1d23 JB |
2760 | } |
2761 | ||
886ed6ec RS |
2762 | /* Replace the old text with the new in the cleanest possible way. */ |
2763 | replace_range (search_regs.start[sub], search_regs.end[sub], | |
2764 | newtext, 1, 0, 1); | |
d5db4077 | 2765 | newpoint = search_regs.start[sub] + SCHARS (newtext); |
ca1d1d23 JB |
2766 | |
2767 | if (case_action == all_caps) | |
886ed6ec RS |
2768 | Fupcase_region (make_number (search_regs.start[sub]), |
2769 | make_number (newpoint)); | |
ca1d1d23 | 2770 | else if (case_action == cap_initial) |
886ed6ec RS |
2771 | Fupcase_initials_region (make_number (search_regs.start[sub]), |
2772 | make_number (newpoint)); | |
3e18eecf | 2773 | |
98e942e0 RS |
2774 | /* Adjust search data for this change. */ |
2775 | { | |
e7deaab0 AS |
2776 | EMACS_INT oldend = search_regs.end[sub]; |
2777 | EMACS_INT oldstart = search_regs.start[sub]; | |
2778 | EMACS_INT change = newpoint - search_regs.end[sub]; | |
98e942e0 RS |
2779 | int i; |
2780 | ||
2781 | for (i = 0; i < search_regs.num_regs; i++) | |
2782 | { | |
41c01205 | 2783 | if (search_regs.start[i] >= oldend) |
98e942e0 | 2784 | search_regs.start[i] += change; |
41c01205 DK |
2785 | else if (search_regs.start[i] > oldstart) |
2786 | search_regs.start[i] = oldstart; | |
2787 | if (search_regs.end[i] >= oldend) | |
98e942e0 | 2788 | search_regs.end[i] += change; |
41c01205 DK |
2789 | else if (search_regs.end[i] > oldstart) |
2790 | search_regs.end[i] = oldstart; | |
98e942e0 RS |
2791 | } |
2792 | } | |
2793 | ||
b0eba991 | 2794 | /* Put point back where it was in the text. */ |
8d808a65 | 2795 | if (opoint <= 0) |
fa8ed3e0 | 2796 | TEMP_SET_PT (opoint + ZV); |
b0eba991 | 2797 | else |
fa8ed3e0 | 2798 | TEMP_SET_PT (opoint); |
b0eba991 RS |
2799 | |
2800 | /* Now move point "officially" to the start of the inserted replacement. */ | |
3e18eecf | 2801 | move_if_not_intangible (newpoint); |
177c0ea7 | 2802 | |
ca1d1d23 JB |
2803 | return Qnil; |
2804 | } | |
2805 | \f | |
2806 | static Lisp_Object | |
971de7fb | 2807 | match_limit (Lisp_Object num, int beginningp) |
ca1d1d23 JB |
2808 | { |
2809 | register int n; | |
2810 | ||
b7826503 | 2811 | CHECK_NUMBER (num); |
ca1d1d23 | 2812 | n = XINT (num); |
f90a5bf5 | 2813 | if (n < 0) |
bd2cbd56 | 2814 | args_out_of_range (num, make_number (0)); |
f90a5bf5 RS |
2815 | if (search_regs.num_regs <= 0) |
2816 | error ("No match data, because no search succeeded"); | |
9b9ceb61 | 2817 | if (n >= search_regs.num_regs |
4746118a | 2818 | || search_regs.start[n] < 0) |
ca1d1d23 JB |
2819 | return Qnil; |
2820 | return (make_number ((beginningp) ? search_regs.start[n] | |
2821 | : search_regs.end[n])); | |
2822 | } | |
2823 | ||
2824 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, | |
8c1a1077 PJ |
2825 | doc: /* Return position of start of text matched by last search. |
2826 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2827 | regexp. | |
2828 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2829 | SUBEXP pairs. | |
2830 | Zero means the entire text matched by the whole regexp or whole string. */) | |
5842a27b | 2831 | (Lisp_Object subexp) |
ca1d1d23 | 2832 | { |
5806161b | 2833 | return match_limit (subexp, 1); |
ca1d1d23 JB |
2834 | } |
2835 | ||
2836 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, | |
8c1a1077 PJ |
2837 | doc: /* Return position of end of text matched by last search. |
2838 | SUBEXP, a number, specifies which parenthesized expression in the last | |
2839 | regexp. | |
2840 | Value is nil if SUBEXPth pair didn't match, or there were less than | |
2841 | SUBEXP pairs. | |
2842 | Zero means the entire text matched by the whole regexp or whole string. */) | |
5842a27b | 2843 | (Lisp_Object subexp) |
ca1d1d23 | 2844 | { |
5806161b | 2845 | return match_limit (subexp, 0); |
177c0ea7 | 2846 | } |
ca1d1d23 | 2847 | |
abd0071c | 2848 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 3, 0, |
8c1a1077 PJ |
2849 | doc: /* Return a list containing all info on what the last search matched. |
2850 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'. | |
2851 | All the elements are markers or nil (nil if the Nth pair didn't match) | |
2852 | if the last match was on a buffer; integers or nil if a string was matched. | |
febb8aba | 2853 | Use `set-match-data' to reinstate the data in this list. |
8c1a1077 | 2854 | |
41c01205 DK |
2855 | If INTEGERS (the optional first argument) is non-nil, always use |
2856 | integers \(rather than markers) to represent buffer positions. In | |
2857 | this case, and if the last match was in a buffer, the buffer will get | |
2858 | stored as one additional element at the end of the list. | |
2859 | ||
abd0071c KS |
2860 | If REUSE is a list, reuse it as part of the value. If REUSE is long |
2861 | enough to hold all the values, and if INTEGERS is non-nil, no consing | |
2862 | is done. | |
2863 | ||
2864 | If optional third arg RESEAT is non-nil, any previous markers on the | |
2865 | REUSE list will be modified to point to nowhere. | |
2866 | ||
140a6b7e | 2867 | Return value is undefined if the last search failed. */) |
5842a27b | 2868 | (Lisp_Object integers, Lisp_Object reuse, Lisp_Object reseat) |
ca1d1d23 | 2869 | { |
56256c2a | 2870 | Lisp_Object tail, prev; |
4746118a | 2871 | Lisp_Object *data; |
ca1d1d23 JB |
2872 | int i, len; |
2873 | ||
abd0071c KS |
2874 | if (!NILP (reseat)) |
2875 | for (tail = reuse; CONSP (tail); tail = XCDR (tail)) | |
2876 | if (MARKERP (XCAR (tail))) | |
2877 | { | |
51f10faa | 2878 | unchain_marker (XMARKER (XCAR (tail))); |
abd0071c KS |
2879 | XSETCAR (tail, Qnil); |
2880 | } | |
2881 | ||
daa37602 | 2882 | if (NILP (last_thing_searched)) |
c36bcf1b | 2883 | return Qnil; |
daa37602 | 2884 | |
6bbd7a29 GM |
2885 | prev = Qnil; |
2886 | ||
41c01205 | 2887 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs + 1) |
4746118a JB |
2888 | * sizeof (Lisp_Object)); |
2889 | ||
41c01205 | 2890 | len = 0; |
4746118a | 2891 | for (i = 0; i < search_regs.num_regs; i++) |
ca1d1d23 JB |
2892 | { |
2893 | int start = search_regs.start[i]; | |
2894 | if (start >= 0) | |
2895 | { | |
56256c2a RS |
2896 | if (EQ (last_thing_searched, Qt) |
2897 | || ! NILP (integers)) | |
ca1d1d23 | 2898 | { |
c235cce7 KH |
2899 | XSETFASTINT (data[2 * i], start); |
2900 | XSETFASTINT (data[2 * i + 1], search_regs.end[i]); | |
ca1d1d23 | 2901 | } |
0ed62dc7 | 2902 | else if (BUFFERP (last_thing_searched)) |
ca1d1d23 JB |
2903 | { |
2904 | data[2 * i] = Fmake_marker (); | |
daa37602 JB |
2905 | Fset_marker (data[2 * i], |
2906 | make_number (start), | |
2907 | last_thing_searched); | |
ca1d1d23 JB |
2908 | data[2 * i + 1] = Fmake_marker (); |
2909 | Fset_marker (data[2 * i + 1], | |
177c0ea7 | 2910 | make_number (search_regs.end[i]), |
daa37602 | 2911 | last_thing_searched); |
ca1d1d23 | 2912 | } |
daa37602 JB |
2913 | else |
2914 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ | |
2915 | abort (); | |
2916 | ||
abd0071c | 2917 | len = 2 * i + 2; |
ca1d1d23 JB |
2918 | } |
2919 | else | |
abd0071c | 2920 | data[2 * i] = data[2 * i + 1] = Qnil; |
ca1d1d23 | 2921 | } |
56256c2a | 2922 | |
bd2cbd56 | 2923 | if (BUFFERP (last_thing_searched) && !NILP (integers)) |
41c01205 | 2924 | { |
bd2cbd56 | 2925 | data[len] = last_thing_searched; |
41c01205 DK |
2926 | len++; |
2927 | } | |
2928 | ||
56256c2a RS |
2929 | /* If REUSE is not usable, cons up the values and return them. */ |
2930 | if (! CONSP (reuse)) | |
41c01205 | 2931 | return Flist (len, data); |
56256c2a RS |
2932 | |
2933 | /* If REUSE is a list, store as many value elements as will fit | |
2934 | into the elements of REUSE. */ | |
2935 | for (i = 0, tail = reuse; CONSP (tail); | |
c1d497be | 2936 | i++, tail = XCDR (tail)) |
56256c2a | 2937 | { |
41c01205 | 2938 | if (i < len) |
f3fbd155 | 2939 | XSETCAR (tail, data[i]); |
56256c2a | 2940 | else |
f3fbd155 | 2941 | XSETCAR (tail, Qnil); |
56256c2a RS |
2942 | prev = tail; |
2943 | } | |
2944 | ||
2945 | /* If we couldn't fit all value elements into REUSE, | |
2946 | cons up the rest of them and add them to the end of REUSE. */ | |
41c01205 DK |
2947 | if (i < len) |
2948 | XSETCDR (prev, Flist (len - i, data + i)); | |
56256c2a RS |
2949 | |
2950 | return reuse; | |
ca1d1d23 JB |
2951 | } |
2952 | ||
b51d6c92 SM |
2953 | /* We used to have an internal use variant of `reseat' described as: |
2954 | ||
2955 | If RESEAT is `evaporate', put the markers back on the free list | |
2956 | immediately. No other references to the markers must exist in this | |
2957 | case, so it is used only internally on the unwind stack and | |
2958 | save-match-data from Lisp. | |
2959 | ||
2960 | But it was ill-conceived: those supposedly-internal markers get exposed via | |
2961 | the undo-list, so freeing them here is unsafe. */ | |
ca1d1d23 | 2962 | |
abd0071c | 2963 | DEFUN ("set-match-data", Fset_match_data, Sset_match_data, 1, 2, 0, |
8c1a1077 | 2964 | doc: /* Set internal data on last search match from elements of LIST. |
abd0071c KS |
2965 | LIST should have been created by calling `match-data' previously. |
2966 | ||
51f10faa | 2967 | If optional arg RESEAT is non-nil, make markers on LIST point nowhere. */) |
5842a27b | 2968 | (register Lisp_Object list, Lisp_Object reseat) |
ca1d1d23 JB |
2969 | { |
2970 | register int i; | |
2971 | register Lisp_Object marker; | |
2972 | ||
7074fde6 FP |
2973 | if (running_asynch_code) |
2974 | save_search_regs (); | |
2975 | ||
29100cea | 2976 | CHECK_LIST (list); |
ca1d1d23 | 2977 | |
41c01205 DK |
2978 | /* Unless we find a marker with a buffer or an explicit buffer |
2979 | in LIST, assume that this match data came from a string. */ | |
daa37602 JB |
2980 | last_thing_searched = Qt; |
2981 | ||
4746118a JB |
2982 | /* Allocate registers if they don't already exist. */ |
2983 | { | |
d084e942 | 2984 | int length = XFASTINT (Flength (list)) / 2; |
4746118a JB |
2985 | |
2986 | if (length > search_regs.num_regs) | |
2987 | { | |
1113d9db JB |
2988 | if (search_regs.num_regs == 0) |
2989 | { | |
2990 | search_regs.start | |
2991 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2992 | search_regs.end | |
2993 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); | |
2994 | } | |
4746118a | 2995 | else |
1113d9db JB |
2996 | { |
2997 | search_regs.start | |
2998 | = (regoff_t *) xrealloc (search_regs.start, | |
2999 | length * sizeof (regoff_t)); | |
3000 | search_regs.end | |
3001 | = (regoff_t *) xrealloc (search_regs.end, | |
3002 | length * sizeof (regoff_t)); | |
3003 | } | |
4746118a | 3004 | |
e62371e9 KH |
3005 | for (i = search_regs.num_regs; i < length; i++) |
3006 | search_regs.start[i] = -1; | |
3007 | ||
487282dc | 3008 | search_regs.num_regs = length; |
4746118a | 3009 | } |
ca1d1d23 | 3010 | |
abd0071c | 3011 | for (i = 0; CONSP (list); i++) |
41c01205 | 3012 | { |
abd0071c | 3013 | marker = XCAR (list); |
bd2cbd56 | 3014 | if (BUFFERP (marker)) |
c3762cbd | 3015 | { |
bd2cbd56 | 3016 | last_thing_searched = marker; |
c3762cbd DK |
3017 | break; |
3018 | } | |
3019 | if (i >= length) | |
3020 | break; | |
41c01205 DK |
3021 | if (NILP (marker)) |
3022 | { | |
3023 | search_regs.start[i] = -1; | |
abd0071c | 3024 | list = XCDR (list); |
41c01205 DK |
3025 | } |
3026 | else | |
3027 | { | |
e7deaab0 | 3028 | EMACS_INT from; |
abd0071c | 3029 | Lisp_Object m; |
e2811828 | 3030 | |
abd0071c | 3031 | m = marker; |
41c01205 DK |
3032 | if (MARKERP (marker)) |
3033 | { | |
3034 | if (XMARKER (marker)->buffer == 0) | |
3035 | XSETFASTINT (marker, 0); | |
3036 | else | |
3037 | XSETBUFFER (last_thing_searched, XMARKER (marker)->buffer); | |
3038 | } | |
e2811828 | 3039 | |
41c01205 DK |
3040 | CHECK_NUMBER_COERCE_MARKER (marker); |
3041 | from = XINT (marker); | |
e2811828 | 3042 | |
abd0071c KS |
3043 | if (!NILP (reseat) && MARKERP (m)) |
3044 | { | |
b51d6c92 | 3045 | unchain_marker (XMARKER (m)); |
9ad54a7e | 3046 | XSETCAR (list, Qnil); |
abd0071c KS |
3047 | } |
3048 | ||
3049 | if ((list = XCDR (list), !CONSP (list))) | |
3050 | break; | |
3051 | ||
3052 | m = marker = XCAR (list); | |
3053 | ||
41c01205 DK |
3054 | if (MARKERP (marker) && XMARKER (marker)->buffer == 0) |
3055 | XSETFASTINT (marker, 0); | |
e2811828 | 3056 | |
41c01205 DK |
3057 | CHECK_NUMBER_COERCE_MARKER (marker); |
3058 | search_regs.start[i] = from; | |
3059 | search_regs.end[i] = XINT (marker); | |
abd0071c KS |
3060 | |
3061 | if (!NILP (reseat) && MARKERP (m)) | |
3062 | { | |
b51d6c92 | 3063 | unchain_marker (XMARKER (m)); |
9ad54a7e | 3064 | XSETCAR (list, Qnil); |
abd0071c | 3065 | } |
41c01205 | 3066 | } |
abd0071c | 3067 | list = XCDR (list); |
41c01205 | 3068 | } |
ca1d1d23 | 3069 | |
41c01205 DK |
3070 | for (; i < search_regs.num_regs; i++) |
3071 | search_regs.start[i] = -1; | |
3072 | } | |
ca1d1d23 | 3073 | |
177c0ea7 | 3074 | return Qnil; |
ca1d1d23 JB |
3075 | } |
3076 | ||
7074fde6 FP |
3077 | /* If non-zero the match data have been saved in saved_search_regs |
3078 | during the execution of a sentinel or filter. */ | |
75ebf74b | 3079 | static int search_regs_saved; |
7074fde6 | 3080 | static struct re_registers saved_search_regs; |
41c01205 | 3081 | static Lisp_Object saved_last_thing_searched; |
7074fde6 FP |
3082 | |
3083 | /* Called from Flooking_at, Fstring_match, search_buffer, Fstore_match_data | |
3084 | if asynchronous code (filter or sentinel) is running. */ | |
3085 | static void | |
971de7fb | 3086 | save_search_regs (void) |
7074fde6 FP |
3087 | { |
3088 | if (!search_regs_saved) | |
3089 | { | |
3090 | saved_search_regs.num_regs = search_regs.num_regs; | |
3091 | saved_search_regs.start = search_regs.start; | |
3092 | saved_search_regs.end = search_regs.end; | |
41c01205 DK |
3093 | saved_last_thing_searched = last_thing_searched; |
3094 | last_thing_searched = Qnil; | |
7074fde6 | 3095 | search_regs.num_regs = 0; |
2d4a771a RS |
3096 | search_regs.start = 0; |
3097 | search_regs.end = 0; | |
7074fde6 FP |
3098 | |
3099 | search_regs_saved = 1; | |
3100 | } | |
3101 | } | |
3102 | ||
3103 | /* Called upon exit from filters and sentinels. */ | |
3104 | void | |
971de7fb | 3105 | restore_search_regs (void) |
7074fde6 FP |
3106 | { |
3107 | if (search_regs_saved) | |
3108 | { | |
3109 | if (search_regs.num_regs > 0) | |
3110 | { | |
3111 | xfree (search_regs.start); | |
3112 | xfree (search_regs.end); | |
3113 | } | |
3114 | search_regs.num_regs = saved_search_regs.num_regs; | |
3115 | search_regs.start = saved_search_regs.start; | |
3116 | search_regs.end = saved_search_regs.end; | |
41c01205 DK |
3117 | last_thing_searched = saved_last_thing_searched; |
3118 | saved_last_thing_searched = Qnil; | |
7074fde6 FP |
3119 | search_regs_saved = 0; |
3120 | } | |
3121 | } | |
3122 | ||
abd0071c | 3123 | static Lisp_Object |
971de7fb | 3124 | unwind_set_match_data (Lisp_Object list) |
abd0071c | 3125 | { |
b51d6c92 SM |
3126 | /* It is NOT ALWAYS safe to free (evaporate) the markers immediately. */ |
3127 | return Fset_match_data (list, Qt); | |
abd0071c KS |
3128 | } |
3129 | ||
3130 | /* Called to unwind protect the match data. */ | |
3131 | void | |
971de7fb | 3132 | record_unwind_save_match_data (void) |
abd0071c KS |
3133 | { |
3134 | record_unwind_protect (unwind_set_match_data, | |
3135 | Fmatch_data (Qnil, Qnil, Qnil)); | |
3136 | } | |
3137 | ||
ca1d1d23 JB |
3138 | /* Quote a string to inactivate reg-expr chars */ |
3139 | ||
3140 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, | |
8c1a1077 | 3141 | doc: /* Return a regexp string which matches exactly STRING and nothing else. */) |
5842a27b | 3142 | (Lisp_Object string) |
ca1d1d23 | 3143 | { |
a7e979a4 PE |
3144 | register char *in, *out, *end; |
3145 | register char *temp; | |
0c8533c6 | 3146 | int backslashes_added = 0; |
ca1d1d23 | 3147 | |
b7826503 | 3148 | CHECK_STRING (string); |
ca1d1d23 | 3149 | |
a7e979a4 | 3150 | temp = (char *) alloca (SBYTES (string) * 2); |
ca1d1d23 JB |
3151 | |
3152 | /* Now copy the data into the new string, inserting escapes. */ | |
3153 | ||
a7e979a4 | 3154 | in = SSDATA (string); |
d5db4077 | 3155 | end = in + SBYTES (string); |
177c0ea7 | 3156 | out = temp; |
ca1d1d23 JB |
3157 | |
3158 | for (; in != end; in++) | |
3159 | { | |
66bc6082 | 3160 | if (*in == '[' |
ca1d1d23 JB |
3161 | || *in == '*' || *in == '.' || *in == '\\' |
3162 | || *in == '?' || *in == '+' | |
3163 | || *in == '^' || *in == '$') | |
0c8533c6 | 3164 | *out++ = '\\', backslashes_added++; |
ca1d1d23 JB |
3165 | *out++ = *in; |
3166 | } | |
3167 | ||
3f8100f1 | 3168 | return make_specified_string (temp, |
d5db4077 | 3169 | SCHARS (string) + backslashes_added, |
3f8100f1 RS |
3170 | out - temp, |
3171 | STRING_MULTIBYTE (string)); | |
ca1d1d23 | 3172 | } |
177c0ea7 | 3173 | \f |
dfcf069d | 3174 | void |
971de7fb | 3175 | syms_of_search (void) |
ca1d1d23 JB |
3176 | { |
3177 | register int i; | |
3178 | ||
487282dc KH |
3179 | for (i = 0; i < REGEXP_CACHE_SIZE; ++i) |
3180 | { | |
3181 | searchbufs[i].buf.allocated = 100; | |
b23c0a83 | 3182 | searchbufs[i].buf.buffer = (unsigned char *) xmalloc (100); |
487282dc KH |
3183 | searchbufs[i].buf.fastmap = searchbufs[i].fastmap; |
3184 | searchbufs[i].regexp = Qnil; | |
ecdb561e | 3185 | searchbufs[i].whitespace_regexp = Qnil; |
b69e3c18 | 3186 | searchbufs[i].syntax_table = Qnil; |
487282dc | 3187 | staticpro (&searchbufs[i].regexp); |
aa77b5ce | 3188 | staticpro (&searchbufs[i].whitespace_regexp); |
b69e3c18 | 3189 | staticpro (&searchbufs[i].syntax_table); |
487282dc KH |
3190 | searchbufs[i].next = (i == REGEXP_CACHE_SIZE-1 ? 0 : &searchbufs[i+1]); |
3191 | } | |
3192 | searchbuf_head = &searchbufs[0]; | |
ca1d1d23 | 3193 | |
d67b4f80 | 3194 | Qsearch_failed = intern_c_string ("search-failed"); |
ca1d1d23 | 3195 | staticpro (&Qsearch_failed); |
d67b4f80 | 3196 | Qinvalid_regexp = intern_c_string ("invalid-regexp"); |
ca1d1d23 JB |
3197 | staticpro (&Qinvalid_regexp); |
3198 | ||
3199 | Fput (Qsearch_failed, Qerror_conditions, | |
d67b4f80 | 3200 | pure_cons (Qsearch_failed, pure_cons (Qerror, Qnil))); |
ca1d1d23 | 3201 | Fput (Qsearch_failed, Qerror_message, |
d67b4f80 | 3202 | make_pure_c_string ("Search failed")); |
ca1d1d23 JB |
3203 | |
3204 | Fput (Qinvalid_regexp, Qerror_conditions, | |
d67b4f80 | 3205 | pure_cons (Qinvalid_regexp, pure_cons (Qerror, Qnil))); |
ca1d1d23 | 3206 | Fput (Qinvalid_regexp, Qerror_message, |
d67b4f80 | 3207 | make_pure_c_string ("Invalid regexp")); |
ca1d1d23 | 3208 | |
daa37602 JB |
3209 | last_thing_searched = Qnil; |
3210 | staticpro (&last_thing_searched); | |
3211 | ||
0f6af254 DK |
3212 | saved_last_thing_searched = Qnil; |
3213 | staticpro (&saved_last_thing_searched); | |
3214 | ||
29208e82 | 3215 | DEFVAR_LISP ("search-spaces-regexp", Vsearch_spaces_regexp, |
e2811828 | 3216 | doc: /* Regexp to substitute for bunches of spaces in regexp search. |
f31a9a68 RS |
3217 | Some commands use this for user-specified regexps. |
3218 | Spaces that occur inside character classes or repetition operators | |
3219 | or other such regexp constructs are not replaced with this. | |
3220 | A value of nil (which is the normal value) means treat spaces literally. */); | |
41a33295 | 3221 | Vsearch_spaces_regexp = Qnil; |
f31a9a68 | 3222 | |
29208e82 | 3223 | DEFVAR_LISP ("inhibit-changing-match-data", Vinhibit_changing_match_data, |
ef887810 | 3224 | doc: /* Internal use only. |
39d0bf74 RS |
3225 | If non-nil, the primitive searching and matching functions |
3226 | such as `looking-at', `string-match', `re-search-forward', etc., | |
3227 | do not set the match data. The proper way to use this variable | |
3228 | is to bind it with `let' around a small expression. */); | |
ef887810 RS |
3229 | Vinhibit_changing_match_data = Qnil; |
3230 | ||
ca1d1d23 | 3231 | defsubr (&Slooking_at); |
b819a390 RS |
3232 | defsubr (&Sposix_looking_at); |
3233 | defsubr (&Sstring_match); | |
3234 | defsubr (&Sposix_string_match); | |
ca1d1d23 JB |
3235 | defsubr (&Ssearch_forward); |
3236 | defsubr (&Ssearch_backward); | |
3237 | defsubr (&Sword_search_forward); | |
3238 | defsubr (&Sword_search_backward); | |
c6e0e1cc CY |
3239 | defsubr (&Sword_search_forward_lax); |
3240 | defsubr (&Sword_search_backward_lax); | |
ca1d1d23 JB |
3241 | defsubr (&Sre_search_forward); |
3242 | defsubr (&Sre_search_backward); | |
b819a390 RS |
3243 | defsubr (&Sposix_search_forward); |
3244 | defsubr (&Sposix_search_backward); | |
ca1d1d23 JB |
3245 | defsubr (&Sreplace_match); |
3246 | defsubr (&Smatch_beginning); | |
3247 | defsubr (&Smatch_end); | |
3248 | defsubr (&Smatch_data); | |
3f1c005b | 3249 | defsubr (&Sset_match_data); |
ca1d1d23 JB |
3250 | defsubr (&Sregexp_quote); |
3251 | } |