search.c 95.9 KB
Newer Older
Jim Blandy's avatar
Jim Blandy committed
1
/* String search routines for GNU Emacs.
2
   Copyright (C) 1985, 1986, 1987, 1993, 1994, 1997, 1998, 1999, 2001, 2002,
Glenn Morris's avatar
Glenn Morris committed
3
                 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010
Glenn Morris's avatar
Glenn Morris committed
4
                 Free Software Foundation, Inc.
Jim Blandy's avatar
Jim Blandy committed
5 6 7

This file is part of GNU Emacs.

8
GNU Emacs is free software: you can redistribute it and/or modify
Jim Blandy's avatar
Jim Blandy committed
9
it under the terms of the GNU General Public License as published by
10 11
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Jim Blandy's avatar
Jim Blandy committed
12 13 14 15 16 17 18

GNU Emacs is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
19
along with GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.  */
Jim Blandy's avatar
Jim Blandy committed
20 21


22
#include <config.h>
23
#include <setjmp.h>
Jim Blandy's avatar
Jim Blandy committed
24 25
#include "lisp.h"
#include "syntax.h"
Karl Heuer's avatar
Karl Heuer committed
26
#include "category.h"
Jim Blandy's avatar
Jim Blandy committed
27
#include "buffer.h"
28
#include "character.h"
Kenichi Handa's avatar
Kenichi Handa committed
29
#include "charset.h"
30
#include "region-cache.h"
Jim Blandy's avatar
Jim Blandy committed
31
#include "commands.h"
32
#include "blockinput.h"
33
#include "intervals.h"
Jim Blandy's avatar
Jim Blandy committed
34

Jim Blandy's avatar
Jim Blandy committed
35 36 37
#include <sys/types.h>
#include "regex.h"

38
#define REGEXP_CACHE_SIZE 20
Jim Blandy's avatar
Jim Blandy committed
39

40 41
/* If the regexp is non-nil, then the buffer contains the compiled form
   of that regexp, suitable for searching.  */
42 43
struct regexp_cache
{
44
  struct regexp_cache *next;
45
  Lisp_Object regexp, whitespace_regexp;
46
  /* Syntax table for which the regexp applies.  We need this because
47 48
     of character classes.  If this is t, then the compiled pattern is valid
     for any syntax-table.  */
49
  Lisp_Object syntax_table;
50 51
  struct re_pattern_buffer buf;
  char fastmap[0400];
52 53
  /* Nonzero means regexp was compiled to do full POSIX backtracking.  */
  char posix;
54
};
Jim Blandy's avatar
Jim Blandy committed
55

56 57
/* The instances of that struct.  */
struct regexp_cache searchbufs[REGEXP_CACHE_SIZE];
Jim Blandy's avatar
Jim Blandy committed
58

59 60
/* The head of the linked list; points to the most recently used buffer.  */
struct regexp_cache *searchbuf_head;
Jim Blandy's avatar
Jim Blandy committed
61 62


Jim Blandy's avatar
Jim Blandy committed
63 64 65 66 67 68 69
/* Every call to re_match, etc., must pass &search_regs as the regs
   argument unless you can show it is unnecessary (i.e., if re_match
   is certainly going to be called again before region-around-match
   can be called).

   Since the registers are now dynamically allocated, we need to make
   sure not to refer to the Nth register before checking that it has
Jim Blandy's avatar
Jim Blandy committed
70 71 72
   been allocated by checking search_regs.num_regs.

   The regex code keeps track of whether it has allocated the search
73 74
   buffer using bits in the re_pattern_buffer.  This means that whenever
   you compile a new pattern, it completely forgets whether it has
Jim Blandy's avatar
Jim Blandy committed
75 76 77 78 79
   allocated any registers, and will allocate new registers the next
   time you call a searching or matching function.  Therefore, we need
   to call re_set_registers after compiling a new pattern or after
   setting the match registers, so that the regex functions will be
   able to free or re-allocate it properly.  */
Jim Blandy's avatar
Jim Blandy committed
80 81
static struct re_registers search_regs;

Jim Blandy's avatar
Jim Blandy committed
82 83 84 85
/* The buffer in which the last search was performed, or
   Qt if the last search was done in a string;
   Qnil if no searching has been done yet.  */
static Lisp_Object last_thing_searched;
Jim Blandy's avatar
Jim Blandy committed
86

Karl Heuer's avatar
Karl Heuer committed
87
/* error condition signaled when regexp compile_pattern fails */
Jim Blandy's avatar
Jim Blandy committed
88 89 90

Lisp_Object Qinvalid_regexp;

91 92 93
/* Error condition used for failing searches */
Lisp_Object Qsearch_failed;

Richard M. Stallman's avatar
Richard M. Stallman committed
94
Lisp_Object Vsearch_spaces_regexp;
95

96 97 98 99 100
/* If non-nil, the match data will not be changed during call to
   searching or matching functions.  This variable is for internal use
   only.  */
Lisp_Object Vinhibit_changing_match_data;

101 102
static void set_search_regs P_ ((EMACS_INT, EMACS_INT));
static void save_search_regs P_ ((void));
Juanma Barranquero's avatar
Juanma Barranquero committed
103
static EMACS_INT simple_search P_ ((int, unsigned char *, int, int,
104 105
				    Lisp_Object, EMACS_INT, EMACS_INT,
				    EMACS_INT, EMACS_INT));
Juanma Barranquero's avatar
Juanma Barranquero committed
106
static EMACS_INT boyer_moore P_ ((int, unsigned char *, int, int,
107 108 109 110 111 112
				  Lisp_Object, Lisp_Object,
				  EMACS_INT, EMACS_INT,
				  EMACS_INT, EMACS_INT, int));
static EMACS_INT search_buffer P_ ((Lisp_Object, EMACS_INT, EMACS_INT,
				    EMACS_INT, EMACS_INT, int, int,
				    Lisp_Object, Lisp_Object, int));
113
static void matcher_overflow () NO_RETURN;
114

Jim Blandy's avatar
Jim Blandy committed
115 116 117 118 119 120
static void
matcher_overflow ()
{
  error ("Stack overflow in regexp matcher");
}

121 122 123
/* Compile a regexp and signal a Lisp error if anything goes wrong.
   PATTERN is the pattern to compile.
   CP is the place to put the result.
124
   TRANSLATE is a translation table for ignoring case, or nil for none.
125 126 127 128 129
   REGP is the structure that says where to store the "register"
   values that will result from matching this pattern.
   If it is 0, we should compile the pattern not to record any
   subexpression bounds.
   POSIX is nonzero if we want full backtracking (POSIX style)
Karl Heuer's avatar
Karl Heuer committed
130
   for this pattern.  0 means backtrack only enough to get a valid match.
131 132

   The behavior also depends on Vsearch_spaces_regexp.  */
Jim Blandy's avatar
Jim Blandy committed
133

134
static void
Kenichi Handa's avatar
Kenichi Handa committed
135
compile_pattern_1 (cp, pattern, translate, regp, posix)
136
     struct regexp_cache *cp;
Jim Blandy's avatar
Jim Blandy committed
137
     Lisp_Object pattern;
138
     Lisp_Object translate;
139
     struct re_registers *regp;
140
     int posix;
Jim Blandy's avatar
Jim Blandy committed
141
{
142
  char *val;
143
  reg_syntax_t old;
Jim Blandy's avatar
Jim Blandy committed
144

145
  cp->regexp = Qnil;
146
  cp->buf.translate = (! NILP (translate) ? translate : make_number (0));
147
  cp->posix = posix;
148
  cp->buf.multibyte = STRING_MULTIBYTE (pattern);
Kenichi Handa's avatar
Kenichi Handa committed
149
  cp->buf.charset_unibyte = charset_unibyte;
150 151 152 153 154
  if (STRINGP (Vsearch_spaces_regexp))
    cp->whitespace_regexp = Vsearch_spaces_regexp;
  else
    cp->whitespace_regexp = Qnil;

155 156 157 158 159
  /* rms: I think BLOCK_INPUT is not needed here any more,
     because regex.c defines malloc to call xmalloc.
     Using BLOCK_INPUT here means the debugger won't run if an error occurs.
     So let's turn it off.  */
  /*  BLOCK_INPUT;  */
160
  old = re_set_syntax (RE_SYNTAX_EMACS
161
		       | (posix ? 0 : RE_NO_POSIX_BACKTRACKING));
162 163 164 165 166

  if (STRINGP (Vsearch_spaces_regexp))
    re_set_whitespace_regexp (SDATA (Vsearch_spaces_regexp));
  else
    re_set_whitespace_regexp (NULL);
167

Kenichi Handa's avatar
Kenichi Handa committed
168 169
  val = (char *) re_compile_pattern ((char *) SDATA (pattern),
				     SBYTES (pattern), &cp->buf);
170

171 172 173 174
  /* If the compiled pattern hard codes some of the contents of the
     syntax-table, it can only be reused with *this* syntax table.  */
  cp->syntax_table = cp->buf.used_syntax ? current_buffer->syntax_table : Qt;

175 176
  re_set_whitespace_regexp (NULL);

177
  re_set_syntax (old);
178
  /* UNBLOCK_INPUT;  */
Jim Blandy's avatar
Jim Blandy committed
179
  if (val)
180
    xsignal1 (Qinvalid_regexp, build_string (val));
Jim Blandy's avatar
Jim Blandy committed
181

182 183 184
  cp->regexp = Fcopy_sequence (pattern);
}

185 186 187 188 189 190 191
/* Shrink each compiled regexp buffer in the cache
   to the size actually used right now.
   This is called from garbage collection.  */

void
shrink_regexp_cache ()
{
192
  struct regexp_cache *cp;
193 194 195 196 197

  for (cp = searchbuf_head; cp != 0; cp = cp->next)
    {
      cp->buf.allocated = cp->buf.used;
      cp->buf.buffer
198
	= (unsigned char *) xrealloc (cp->buf.buffer, cp->buf.used);
199 200 201
    }
}

202 203
/* Clear the regexp cache w.r.t. a particular syntax table,
   because it was changed.
204 205 206 207 208 209 210 211 212
   There is no danger of memory leak here because re_compile_pattern
   automagically manages the memory in each re_pattern_buffer struct,
   based on its `allocated' and `buffer' values.  */
void
clear_regexp_cache ()
{
  int i;

  for (i = 0; i < REGEXP_CACHE_SIZE; ++i)
213 214
    /* It's tempting to compare with the syntax-table we've actually changed,
       but it's not sufficient because char-table inheritance means that
215 216 217
       modifying one syntax-table can change others at the same time.  */
    if (!EQ (searchbufs[i].syntax_table, Qt))
      searchbufs[i].regexp = Qnil;
218 219
}

220
/* Compile a regexp if necessary, but first check to see if there's one in
221 222
   the cache.
   PATTERN is the pattern to compile.
223
   TRANSLATE is a translation table for ignoring case, or nil for none.
224 225 226 227 228 229
   REGP is the structure that says where to store the "register"
   values that will result from matching this pattern.
   If it is 0, we should compile the pattern not to record any
   subexpression bounds.
   POSIX is nonzero if we want full backtracking (POSIX style)
   for this pattern.  0 means backtrack only enough to get a valid match.  */
230 231

struct re_pattern_buffer *
232
compile_pattern (pattern, regp, translate, posix, multibyte)
233 234
     Lisp_Object pattern;
     struct re_registers *regp;
235
     Lisp_Object translate;
236
     int posix, multibyte;
237 238 239 240 241 242
{
  struct regexp_cache *cp, **cpp;

  for (cpp = &searchbuf_head; ; cpp = &cp->next)
    {
      cp = *cpp;
243 244
      /* Entries are initialized to nil, and may be set to nil by
	 compile_pattern_1 if the pattern isn't valid.  Don't apply
Ken Raeburn's avatar
Ken Raeburn committed
245 246 247
	 string accessors in those cases.  However, compile_pattern_1
	 is only applied to the cache entry we pick here to reuse.  So
	 nil should never appear before a non-nil entry.  */
248
      if (NILP (cp->regexp))
249
	goto compile_it;
250
      if (SCHARS (cp->regexp) == SCHARS (pattern)
251
	  && STRING_MULTIBYTE (cp->regexp) == STRING_MULTIBYTE (pattern)
252
	  && !NILP (Fstring_equal (cp->regexp, pattern))
253
	  && EQ (cp->buf.translate, (! NILP (translate) ? translate : make_number (0)))
Karl Heuer's avatar
Karl Heuer committed
254
	  && cp->posix == posix
255 256
	  && (EQ (cp->syntax_table, Qt)
	      || EQ (cp->syntax_table, current_buffer->syntax_table))
Kenichi Handa's avatar
Kenichi Handa committed
257 258
	  && !NILP (Fequal (cp->whitespace_regexp, Vsearch_spaces_regexp))
	  && cp->buf.charset_unibyte == charset_unibyte)
259 260
	break;

261 262 263
      /* If we're at the end of the cache, compile into the nil cell
	 we found, or the last (least recently used) cell with a
	 string value.  */
264 265
      if (cp->next == 0)
	{
266
	compile_it:
Kenichi Handa's avatar
Kenichi Handa committed
267
	  compile_pattern_1 (cp, pattern, translate, regp, posix);
268 269 270 271 272 273 274 275 276 277
	  break;
	}
    }

  /* When we get here, cp (aka *cpp) contains the compiled pattern,
     either because we found it in the cache or because we just compiled it.
     Move it to the front of the queue to mark it as most recently used.  */
  *cpp = cp->next;
  cp->next = searchbuf_head;
  searchbuf_head = cp;
Jim Blandy's avatar
Jim Blandy committed
278

279 280 281 282 283
  /* Advise the searching functions about the space we have allocated
     for register data.  */
  if (regp)
    re_set_registers (&cp->buf, regp, regp->num_regs, regp->start, regp->end);

Juanma Barranquero's avatar
Juanma Barranquero committed
284
  /* The compiled pattern can be used both for multibyte and unibyte
Kenichi Handa's avatar
Kenichi Handa committed
285 286 287
     target.  But, we have to tell which the pattern is used for. */
  cp->buf.target_multibyte = multibyte;

288
  return &cp->buf;
Jim Blandy's avatar
Jim Blandy committed
289 290 291
}


292 293
static Lisp_Object
looking_at_1 (string, posix)
Jim Blandy's avatar
Jim Blandy committed
294
     Lisp_Object string;
295
     int posix;
Jim Blandy's avatar
Jim Blandy committed
296 297 298
{
  Lisp_Object val;
  unsigned char *p1, *p2;
299
  EMACS_INT s1, s2;
Jim Blandy's avatar
Jim Blandy committed
300
  register int i;
301
  struct re_pattern_buffer *bufp;
Jim Blandy's avatar
Jim Blandy committed
302

303 304 305
  if (running_asynch_code)
    save_search_regs ();

306 307 308 309
  /* This is so set_image_of_range_1 in regex.c can find the EQV table.  */
  XCHAR_TABLE (current_buffer->case_canon_table)->extras[2]
    = current_buffer->case_eqv_table;

310
  CHECK_STRING (string);
311 312 313
  bufp = compile_pattern (string,
			  (NILP (Vinhibit_changing_match_data)
			   ? &search_regs : NULL),
314
			  (!NILP (current_buffer->case_fold_search)
315
			   ? current_buffer->case_canon_table : Qnil),
316 317
			  posix,
			  !NILP (current_buffer->enable_multibyte_characters));
Jim Blandy's avatar
Jim Blandy committed
318 319 320 321 322 323 324 325

  immediate_quit = 1;
  QUIT;			/* Do a pending quit right away, to avoid paradoxical behavior */

  /* Get pointers and sizes of the two strings
     that make up the visible portion of the buffer. */

  p1 = BEGV_ADDR;
326
  s1 = GPT_BYTE - BEGV_BYTE;
Jim Blandy's avatar
Jim Blandy committed
327
  p2 = GAP_END_ADDR;
328
  s2 = ZV_BYTE - GPT_BYTE;
Jim Blandy's avatar
Jim Blandy committed
329 330 331
  if (s1 < 0)
    {
      p2 = p1;
332
      s2 = ZV_BYTE - BEGV_BYTE;
Jim Blandy's avatar
Jim Blandy committed
333 334 335 336
      s1 = 0;
    }
  if (s2 < 0)
    {
337
      s1 = ZV_BYTE - BEGV_BYTE;
Jim Blandy's avatar
Jim Blandy committed
338 339
      s2 = 0;
    }
340 341

  re_match_object = Qnil;
342

343
  i = re_match_2 (bufp, (char *) p1, s1, (char *) p2, s2,
344 345 346
		  PT_BYTE - BEGV_BYTE,
		  (NILP (Vinhibit_changing_match_data)
		   ? &search_regs : NULL),
347
		  ZV_BYTE - BEGV_BYTE);
348
  immediate_quit = 0;
349

Jim Blandy's avatar
Jim Blandy committed
350 351 352 353
  if (i == -2)
    matcher_overflow ();

  val = (0 <= i ? Qt : Qnil);
354
  if (NILP (Vinhibit_changing_match_data) && i >= 0)
355 356 357 358 359 360 361 362
    for (i = 0; i < search_regs.num_regs; i++)
      if (search_regs.start[i] >= 0)
	{
	  search_regs.start[i]
	    = BYTE_TO_CHAR (search_regs.start[i] + BEGV_BYTE);
	  search_regs.end[i]
	    = BYTE_TO_CHAR (search_regs.end[i] + BEGV_BYTE);
	}
363 364 365 366 367

  /* Set last_thing_searched only when match data is changed.  */
  if (NILP (Vinhibit_changing_match_data))
    XSETBUFFER (last_thing_searched, current_buffer);

Jim Blandy's avatar
Jim Blandy committed
368 369 370
  return val;
}

371
DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0,
372 373 374 375 376
       doc: /* Return t if text after point matches regular expression REGEXP.
This function modifies the match data that `match-beginning',
`match-end' and `match-data' access; save and restore the match
data if you want to preserve them.  */)
     (regexp)
377
     Lisp_Object regexp;
378
{
379
  return looking_at_1 (regexp, 0);
380 381 382
}

DEFUN ("posix-looking-at", Fposix_looking_at, Sposix_looking_at, 1, 1, 0,
383 384 385 386 387 388
       doc: /* Return t if text after point matches regular expression REGEXP.
Find the longest match, in accord with Posix regular expression rules.
This function modifies the match data that `match-beginning',
`match-end' and `match-data' access; save and restore the match
data if you want to preserve them.  */)
     (regexp)
389
     Lisp_Object regexp;
390
{
391
  return looking_at_1 (regexp, 1);
392 393 394 395
}

static Lisp_Object
string_match_1 (regexp, string, start, posix)
Jim Blandy's avatar
Jim Blandy committed
396
     Lisp_Object regexp, string, start;
397
     int posix;
Jim Blandy's avatar
Jim Blandy committed
398 399
{
  int val;
400
  struct re_pattern_buffer *bufp;
401
  EMACS_INT pos, pos_byte;
402
  int i;
Jim Blandy's avatar
Jim Blandy committed
403

404 405 406
  if (running_asynch_code)
    save_search_regs ();

407 408
  CHECK_STRING (regexp);
  CHECK_STRING (string);
Jim Blandy's avatar
Jim Blandy committed
409 410

  if (NILP (start))
411
    pos = 0, pos_byte = 0;
Jim Blandy's avatar
Jim Blandy committed
412 413
  else
    {
414
      int len = SCHARS (string);
Jim Blandy's avatar
Jim Blandy committed
415

416
      CHECK_NUMBER (start);
417 418 419 420
      pos = XINT (start);
      if (pos < 0 && -pos <= len)
	pos = len + pos;
      else if (0 > pos || pos > len)
Jim Blandy's avatar
Jim Blandy committed
421
	args_out_of_range (string, start);
422
      pos_byte = string_char_to_byte (string, pos);
Jim Blandy's avatar
Jim Blandy committed
423 424
    }

425 426 427 428
  /* This is so set_image_of_range_1 in regex.c can find the EQV table.  */
  XCHAR_TABLE (current_buffer->case_canon_table)->extras[2]
    = current_buffer->case_eqv_table;

429 430 431
  bufp = compile_pattern (regexp,
			  (NILP (Vinhibit_changing_match_data)
			   ? &search_regs : NULL),
432
			  (!NILP (current_buffer->case_fold_search)
433
			   ? current_buffer->case_canon_table : Qnil),
434 435
			  posix,
			  STRING_MULTIBYTE (string));
Jim Blandy's avatar
Jim Blandy committed
436
  immediate_quit = 1;
437
  re_match_object = string;
438

439 440 441
  val = re_search (bufp, (char *) SDATA (string),
		   SBYTES (string), pos_byte,
		   SBYTES (string) - pos_byte,
442 443
		   (NILP (Vinhibit_changing_match_data)
		    ? &search_regs : NULL));
Jim Blandy's avatar
Jim Blandy committed
444
  immediate_quit = 0;
445 446 447 448 449

  /* Set last_thing_searched only when match data is changed.  */
  if (NILP (Vinhibit_changing_match_data))
    last_thing_searched = Qt;

Jim Blandy's avatar
Jim Blandy committed
450 451 452
  if (val == -2)
    matcher_overflow ();
  if (val < 0) return Qnil;
453

454 455 456 457 458 459 460 461 462
  if (NILP (Vinhibit_changing_match_data))
    for (i = 0; i < search_regs.num_regs; i++)
      if (search_regs.start[i] >= 0)
	{
	  search_regs.start[i]
	    = string_byte_to_char (string, search_regs.start[i]);
	  search_regs.end[i]
	    = string_byte_to_char (string, search_regs.end[i]);
	}
463 464

  return make_number (string_byte_to_char (string, val));
Jim Blandy's avatar
Jim Blandy committed
465
}
Richard M. Stallman's avatar
Richard M. Stallman committed
466

467
DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0,
468
       doc: /* Return index of start of first match for REGEXP in STRING, or nil.
Richard M. Stallman's avatar
Richard M. Stallman committed
469
Matching ignores case if `case-fold-search' is non-nil.
470 471 472
If third arg START is non-nil, start search at that index in STRING.
For index of first char beyond the match, do (match-end 0).
`match-end' and `match-beginning' also give indices of substrings
473 474 475 476
matched by parenthesis constructs in the pattern.

You can use the function `match-string' to extract the substrings
matched by the parenthesis constructions in REGEXP. */)
477
     (regexp, string, start)
478 479 480 481 482 483
     Lisp_Object regexp, string, start;
{
  return string_match_1 (regexp, string, start, 0);
}

DEFUN ("posix-string-match", Fposix_string_match, Sposix_string_match, 2, 3, 0,
484 485 486 487 488 489 490 491
       doc: /* Return index of start of first match for REGEXP in STRING, or nil.
Find the longest match, in accord with Posix regular expression rules.
Case is ignored if `case-fold-search' is non-nil in the current buffer.
If third arg START is non-nil, start search at that index in STRING.
For index of first char beyond the match, do (match-end 0).
`match-end' and `match-beginning' also give indices of substrings
matched by parenthesis constructs in the pattern.  */)
     (regexp, string, start)
492 493 494 495 496
     Lisp_Object regexp, string, start;
{
  return string_match_1 (regexp, string, start, 1);
}

Richard M. Stallman's avatar
Richard M. Stallman committed
497 498 499 500 501 502 503 504 505
/* Match REGEXP against STRING, searching all of STRING,
   and return the index of the match, or negative on failure.
   This does not clobber the match data.  */

int
fast_string_match (regexp, string)
     Lisp_Object regexp, string;
{
  int val;
506
  struct re_pattern_buffer *bufp;
Richard M. Stallman's avatar
Richard M. Stallman committed
507

508 509
  bufp = compile_pattern (regexp, 0, Qnil,
			  0, STRING_MULTIBYTE (string));
Richard M. Stallman's avatar
Richard M. Stallman committed
510
  immediate_quit = 1;
511
  re_match_object = string;
512

513 514 515
  val = re_search (bufp, (char *) SDATA (string),
		   SBYTES (string), 0,
		   SBYTES (string), 0);
Richard M. Stallman's avatar
Richard M. Stallman committed
516 517 518
  immediate_quit = 0;
  return val;
}
Karl Heuer's avatar
Karl Heuer committed
519 520 521

/* Match REGEXP against STRING, searching all of STRING ignoring case,
   and return the index of the match, or negative on failure.
522 523
   This does not clobber the match data.
   We assume that STRING contains single-byte characters.  */
Karl Heuer's avatar
Karl Heuer committed
524 525 526 527

extern Lisp_Object Vascii_downcase_table;

int
528
fast_c_string_match_ignore_case (regexp, string)
Karl Heuer's avatar
Karl Heuer committed
529
     Lisp_Object regexp;
530
     const char *string;
Karl Heuer's avatar
Karl Heuer committed
531 532 533 534 535
{
  int val;
  struct re_pattern_buffer *bufp;
  int len = strlen (string);

536
  regexp = string_make_unibyte (regexp);
537
  re_match_object = Qt;
Karl Heuer's avatar
Karl Heuer committed
538
  bufp = compile_pattern (regexp, 0,
539
			  Vascii_canon_table, 0,
540
			  0);
Karl Heuer's avatar
Karl Heuer committed
541 542 543 544 545
  immediate_quit = 1;
  val = re_search (bufp, string, len, 0, len, 0);
  immediate_quit = 0;
  return val;
}
546 547 548 549 550 551 552 553 554 555

/* Like fast_string_match but ignore case.  */

int
fast_string_match_ignore_case (regexp, string)
     Lisp_Object regexp, string;
{
  int val;
  struct re_pattern_buffer *bufp;

556
  bufp = compile_pattern (regexp, 0, Vascii_canon_table,
557 558 559 560 561 562 563 564 565 566
			  0, STRING_MULTIBYTE (string));
  immediate_quit = 1;
  re_match_object = string;

  val = re_search (bufp, (char *) SDATA (string),
		   SBYTES (string), 0,
		   SBYTES (string), 0);
  immediate_quit = 0;
  return val;
}
567

568
/* Match REGEXP against the characters after POS to LIMIT, and return
569 570 571 572 573 574 575 576 577 578 579 580 581 582
   the number of matched characters.  If STRING is non-nil, match
   against the characters in it.  In that case, POS and LIMIT are
   indices into the string.  This function doesn't modify the match
   data.  */

EMACS_INT
fast_looking_at (regexp, pos, pos_byte, limit, limit_byte, string)
     Lisp_Object regexp;
     EMACS_INT pos, pos_byte, limit, limit_byte;
     Lisp_Object string;
{
  int multibyte;
  struct re_pattern_buffer *buf;
  unsigned char *p1, *p2;
583
  EMACS_INT s1, s2;
584
  EMACS_INT len;
Juanma Barranquero's avatar
Juanma Barranquero committed
585

586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634
  if (STRINGP (string))
    {
      if (pos_byte < 0)
	pos_byte = string_char_to_byte (string, pos);
      if (limit_byte < 0)
	limit_byte = string_char_to_byte (string, limit);
      p1 = NULL;
      s1 = 0;
      p2 = SDATA (string);
      s2 = SBYTES (string);
      re_match_object = string;
      multibyte = STRING_MULTIBYTE (string);
    }
  else
    {
      if (pos_byte < 0)
	pos_byte = CHAR_TO_BYTE (pos);
      if (limit_byte < 0)
	limit_byte = CHAR_TO_BYTE (limit);
      pos_byte -= BEGV_BYTE;
      limit_byte -= BEGV_BYTE;
      p1 = BEGV_ADDR;
      s1 = GPT_BYTE - BEGV_BYTE;
      p2 = GAP_END_ADDR;
      s2 = ZV_BYTE - GPT_BYTE;
      if (s1 < 0)
	{
	  p2 = p1;
	  s2 = ZV_BYTE - BEGV_BYTE;
	  s1 = 0;
	}
      if (s2 < 0)
	{
	  s1 = ZV_BYTE - BEGV_BYTE;
	  s2 = 0;
	}
      re_match_object = Qnil;
      multibyte = ! NILP (current_buffer->enable_multibyte_characters);
    }

  buf = compile_pattern (regexp, 0, Qnil, 0, multibyte);
  immediate_quit = 1;
  len = re_match_2 (buf, (char *) p1, s1, (char *) p2, s2,
		    pos_byte, NULL, limit_byte);
  immediate_quit = 0;

  return len;
}

635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672

/* The newline cache: remembering which sections of text have no newlines.  */

/* If the user has requested newline caching, make sure it's on.
   Otherwise, make sure it's off.
   This is our cheezy way of associating an action with the change of
   state of a buffer-local variable.  */
static void
newline_cache_on_off (buf)
     struct buffer *buf;
{
  if (NILP (buf->cache_long_line_scans))
    {
      /* It should be off.  */
      if (buf->newline_cache)
        {
          free_region_cache (buf->newline_cache);
          buf->newline_cache = 0;
        }
    }
  else
    {
      /* It should be on.  */
      if (buf->newline_cache == 0)
        buf->newline_cache = new_region_cache ();
    }
}


/* Search for COUNT instances of the character TARGET between START and END.

   If COUNT is positive, search forwards; END must be >= START.
   If COUNT is negative, search backwards for the -COUNTth instance;
      END must be <= START.
   If COUNT is zero, do anything you please; run rogue, for all I care.

   If END is zero, use BEGV or ZV instead, as appropriate for the
   direction indicated by COUNT.
Jim Blandy's avatar
Jim Blandy committed
673 674

   If we find COUNT instances, set *SHORTAGE to zero, and return the
Richard M. Stallman's avatar
Richard M. Stallman committed
675
   position past the COUNTth match.  Note that for reverse motion
Richard M. Stallman's avatar
Richard M. Stallman committed
676
   this is not the same as the usual convention for Emacs motion commands.
Jim Blandy's avatar
Jim Blandy committed
677

678 679
   If we don't find COUNT instances before reaching END, set *SHORTAGE
   to the number of TARGETs left unfound, and return END.
Jim Blandy's avatar
Jim Blandy committed
680

681 682 683
   If ALLOW_QUIT is non-zero, set immediate_quit.  That's good to do
   except when inside redisplay.  */

Andreas Schwab's avatar
Andreas Schwab committed
684
int
685 686
scan_buffer (target, start, end, count, shortage, allow_quit)
     register int target;
687
     EMACS_INT start, end;
688 689
     int count;
     int *shortage;
690
     int allow_quit;
Jim Blandy's avatar
Jim Blandy committed
691
{
692
  struct region_cache *newline_cache;
693
  int direction;
Jim Blandy's avatar
Jim Blandy committed
694

695 696 697 698 699 700 701 702 703 704
  if (count > 0)
    {
      direction = 1;
      if (! end) end = ZV;
    }
  else
    {
      direction = -1;
      if (! end) end = BEGV;
    }
Jim Blandy's avatar
Jim Blandy committed
705

706 707
  newline_cache_on_off (current_buffer);
  newline_cache = current_buffer->newline_cache;
Jim Blandy's avatar
Jim Blandy committed
708 709 710 711

  if (shortage != 0)
    *shortage = 0;

712
  immediate_quit = allow_quit;
Jim Blandy's avatar
Jim Blandy committed
713

Jim Blandy's avatar
Jim Blandy committed
714
  if (count > 0)
715
    while (start != end)
Jim Blandy's avatar
Jim Blandy committed
716
      {
717 718 719 720 721
        /* Our innermost scanning loop is very simple; it doesn't know
           about gaps, buffer ends, or the newline cache.  ceiling is
           the position of the last character before the next such
           obstacle --- the last character the dumb search loop should
           examine.  */
722 723 724
	EMACS_INT ceiling_byte = CHAR_TO_BYTE (end) - 1;
	EMACS_INT start_byte = CHAR_TO_BYTE (start);
	EMACS_INT tem;
725 726 727 728 729 730 731 732

        /* If we're looking for a newline, consult the newline cache
           to see where we can avoid some scanning.  */
        if (target == '\n' && newline_cache)
          {
            int next_change;
            immediate_quit = 0;
            while (region_cache_forward
733 734
                   (current_buffer, newline_cache, start_byte, &next_change))
              start_byte = next_change;
735
            immediate_quit = allow_quit;
736

737 738 739
            /* START should never be after END.  */
            if (start_byte > ceiling_byte)
              start_byte = ceiling_byte;
740 741 742

            /* Now the text after start is an unknown region, and
               next_change is the position of the next known region. */
743
            ceiling_byte = min (next_change - 1, ceiling_byte);
744 745 746 747 748 749
          }

        /* The dumb loop can only scan text stored in contiguous
           bytes. BUFFER_CEILING_OF returns the last character
           position that is contiguous, so the ceiling is the
           position after that.  */
750 751
	tem = BUFFER_CEILING_OF (start_byte);
	ceiling_byte = min (tem, ceiling_byte);
752 753

        {
754
          /* The termination address of the dumb loop.  */
755 756 757 758
          register unsigned char *ceiling_addr
	    = BYTE_POS_ADDR (ceiling_byte) + 1;
          register unsigned char *cursor
	    = BYTE_POS_ADDR (start_byte);
759 760 761 762 763 764 765 766 767 768 769 770 771 772
          unsigned char *base = cursor;

          while (cursor < ceiling_addr)
            {
              unsigned char *scan_start = cursor;

              /* The dumb loop.  */
              while (*cursor != target && ++cursor < ceiling_addr)
                ;

              /* If we're looking for newlines, cache the fact that
                 the region from start to cursor is free of them. */
              if (target == '\n' && newline_cache)
                know_region_cache (current_buffer, newline_cache,
773 774
                                   start_byte + scan_start - base,
                                   start_byte + cursor - base);
775 776 777 778 779 780 781

              /* Did we find the target character?  */
              if (cursor < ceiling_addr)
                {
                  if (--count == 0)
                    {
                      immediate_quit = 0;
782
                      return BYTE_TO_CHAR (start_byte + cursor - base + 1);
783 784 785 786 787
                    }
                  cursor++;
                }
            }

788
          start = BYTE_TO_CHAR (start_byte + cursor - base);
789
        }
Jim Blandy's avatar
Jim Blandy committed
790 791
      }
  else
792 793 794
    while (start > end)
      {
        /* The last character to check before the next obstacle.  */
795 796 797
	EMACS_INT ceiling_byte = CHAR_TO_BYTE (end);
	EMACS_INT start_byte = CHAR_TO_BYTE (start);
	EMACS_INT tem;
798 799 800 801 802 803 804

        /* Consult the newline cache, if appropriate.  */
        if (target == '\n' && newline_cache)
          {
            int next_change;
            immediate_quit = 0;
            while (region_cache_backward
805 806
                   (current_buffer, newline_cache, start_byte, &next_change))
              start_byte = next_change;
807
            immediate_quit = allow_quit;
808 809

            /* Start should never be at or before end.  */
810 811
            if (start_byte <= ceiling_byte)
              start_byte = ceiling_byte + 1;
812 813 814

            /* Now the text before start is an unknown region, and
               next_change is the position of the next known region. */
815
            ceiling_byte = max (next_change, ceiling_byte);
816 817 818
          }

        /* Stop scanning before the gap.  */
819 820
	tem = BUFFER_FLOOR_OF (start_byte - 1);
	ceiling_byte = max (tem, ceiling_byte);
821 822 823

        {
          /* The termination address of the dumb loop.  */
824 825
          register unsigned char *ceiling_addr = BYTE_POS_ADDR (ceiling_byte);
          register unsigned char *cursor = BYTE_POS_ADDR (start_byte - 1);
826 827 828 829 830 831 832 833 834 835 836 837 838
          unsigned char *base = cursor;

          while (cursor >= ceiling_addr)
            {
              unsigned char *scan_start = cursor;

              while (*cursor != target && --cursor >= ceiling_addr)
                ;

              /* If we're looking for newlines, cache the fact that
                 the region from after the cursor to start is free of them.  */
              if (target == '\n' && newline_cache)
                know_region_cache (current_buffer, newline_cache,
839 840
                                   start_byte + cursor - base,
                                   start_byte + scan_start - base);
841 842 843 844 845 846 847

              /* Did we find the target character?  */
              if (cursor >= ceiling_addr)
                {
                  if (++count >= 0)
                    {
                      immediate_quit = 0;
848
                      return BYTE_TO_CHAR (start_byte + cursor - base);
849 850 851 852 853
                    }
                  cursor--;
                }
            }

854
	  start = BYTE_TO_CHAR (start_byte + cursor - base);
855 856 857
        }
      }

Jim Blandy's avatar
Jim Blandy committed
858 859
  immediate_quit = 0;
  if (shortage != 0)
Jim Blandy's avatar
Jim Blandy committed
860
    *shortage = count * direction;
861
  return start;
Jim Blandy's avatar
Jim Blandy committed
862
}
863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878

/* Search for COUNT instances of a line boundary, which means either a
   newline or (if selective display enabled) a carriage return.
   Start at START.  If COUNT is negative, search backwards.

   We report the resulting position by calling TEMP_SET_PT_BOTH.

   If we find COUNT instances. we position after (always after,
   even if scanning backwards) the COUNTth match, and return 0.

   If we don't find COUNT instances before reaching the end of the
   buffer (or the beginning, if scanning backwards), we return
   the number of line boundaries left unfound, and position at
   the limit we bumped up against.

   If ALLOW_QUIT is non-zero, set immediate_quit.  That's good to do
879
   except in special cases.  */
Jim Blandy's avatar
Jim Blandy committed
880

881
int
882
scan_newline (start, start_byte, limit, limit_byte, count, allow_quit)
883 884
     EMACS_INT start, start_byte;
     EMACS_INT limit, limit_byte;
885 886
     register int count;
     int allow_quit;
887
{
888 889 890 891 892
  int direction = ((count > 0) ? 1 : -1);

  register unsigned char *cursor;
  unsigned char *base;

893
  EMACS_INT ceiling;
894 895
  register unsigned char *ceiling_addr;

896 897
  int old_immediate_quit = immediate_quit;

898 899 900
  /* The code that follows is like scan_buffer
     but checks for either newline or carriage return.  */

901 902
  if (allow_quit)
    immediate_quit++;
903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922

  start_byte = CHAR_TO_BYTE (start);

  if (count > 0)
    {
      while (start_byte < limit_byte)
	{
	  ceiling =  BUFFER_CEILING_OF (start_byte);
	  ceiling = min (limit_byte - 1, ceiling);
	  ceiling_addr = BYTE_POS_ADDR (ceiling) + 1;
	  base = (cursor = BYTE_POS_ADDR (start_byte));
	  while (1)
	    {
	      while (*cursor != '\n' && ++cursor != ceiling_addr)
		;

	      if (cursor != ceiling_addr)
		{
		  if (--count == 0)
		    {
923
		      immediate_quit = old_immediate_quit;
924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955
		      start_byte = start_byte + cursor - base + 1;
		      start = BYTE_TO_CHAR (start_byte);
		      TEMP_SET_PT_BOTH (start, start_byte);
		      return 0;
		    }
		  else
		    if (++cursor == ceiling_addr)
		      break;
		}
	      else
		break;
	    }
	  start_byte += cursor - base;
	}
    }
  else
    {
      while (start_byte > limit_byte)
	{
	  ceiling = BUFFER_FLOOR_OF (start_byte - 1);
	  ceiling = max (limit_byte, ceiling);
	  ceiling_addr = BYTE_POS_ADDR (ceiling) - 1;
	  base = (cursor = BYTE_POS_ADDR (start_byte - 1) + 1);
	  while (1)
	    {
	      while (--cursor != ceiling_addr && *cursor != '\n')
		;

	      if (cursor != ceiling_addr)
		{
		  if (++count == 0)
		    {
956
		      immediate_quit = old_immediate_quit;
957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973
		      /* Return the position AFTER the match we found.  */
		      start_byte = start_byte + cursor - base + 1;
		      start = BYTE_TO_CHAR (start_byte);
		      TEMP_SET_PT_BOTH (start, start_byte);
		      return 0;
		    }
		}
	      else
		break;
	    }
	  /* Here we add 1 to compensate for the last decrement
	     of CURSOR, which took it past the valid range.  */
	  start_byte += cursor - base + 1;
	}
    }

  TEMP_SET_PT_BOTH (limit, limit_byte);
974
  immediate_quit = old_immediate_quit;
975 976

  return count * direction;
977 978
}

Jim Blandy's avatar
Jim Blandy committed
979
int
980
find_next_newline_no_quit (from, cnt)
981 982
     EMACS_INT from;
     int cnt;
Jim Blandy's avatar
Jim Blandy committed
983
{
984
  return scan_buffer ('\n', from, 0, cnt, (int *) 0, 0);
985 986 987 988 989
}

/* Like find_next_newline, but returns position before the newline,
   not after, and only search up to TO.  This isn't just
   find_next_newline (...)-1, because you might hit TO.  */
990

991 992
int
find_before_next_newline (from, to, cnt)
993 994
     EMACS_INT from, to;
     int cnt;