| 1 | /* String search routines for GNU Emacs. |
| 2 | Copyright (C) 1985, 1986, 1987, 1993 Free Software Foundation, Inc. |
| 3 | |
| 4 | This file is part of GNU Emacs. |
| 5 | |
| 6 | GNU Emacs is free software; you can redistribute it and/or modify |
| 7 | it under the terms of the GNU General Public License as published by |
| 8 | the Free Software Foundation; either version 1, or (at your option) |
| 9 | any later version. |
| 10 | |
| 11 | GNU Emacs is distributed in the hope that it will be useful, |
| 12 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 14 | GNU General Public License for more details. |
| 15 | |
| 16 | You should have received a copy of the GNU General Public License |
| 17 | along with GNU Emacs; see the file COPYING. If not, write to |
| 18 | the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */ |
| 19 | |
| 20 | |
| 21 | #include "config.h" |
| 22 | #include "lisp.h" |
| 23 | #include "syntax.h" |
| 24 | #include "buffer.h" |
| 25 | #include "commands.h" |
| 26 | #include "blockinput.h" |
| 27 | |
| 28 | #include <sys/types.h> |
| 29 | #include "regex.h" |
| 30 | |
| 31 | #define max(a, b) ((a) > (b) ? (a) : (b)) |
| 32 | #define min(a, b) ((a) < (b) ? (a) : (b)) |
| 33 | |
| 34 | /* We compile regexps into this buffer and then use it for searching. */ |
| 35 | |
| 36 | struct re_pattern_buffer searchbuf; |
| 37 | |
| 38 | char search_fastmap[0400]; |
| 39 | |
| 40 | /* Last regexp we compiled */ |
| 41 | |
| 42 | Lisp_Object last_regexp; |
| 43 | |
| 44 | /* Every call to re_match, etc., must pass &search_regs as the regs |
| 45 | argument unless you can show it is unnecessary (i.e., if re_match |
| 46 | is certainly going to be called again before region-around-match |
| 47 | can be called). |
| 48 | |
| 49 | Since the registers are now dynamically allocated, we need to make |
| 50 | sure not to refer to the Nth register before checking that it has |
| 51 | been allocated by checking search_regs.num_regs. |
| 52 | |
| 53 | The regex code keeps track of whether it has allocated the search |
| 54 | buffer using bits in searchbuf. This means that whenever you |
| 55 | compile a new pattern, it completely forgets whether it has |
| 56 | allocated any registers, and will allocate new registers the next |
| 57 | time you call a searching or matching function. Therefore, we need |
| 58 | to call re_set_registers after compiling a new pattern or after |
| 59 | setting the match registers, so that the regex functions will be |
| 60 | able to free or re-allocate it properly. */ |
| 61 | static struct re_registers search_regs; |
| 62 | |
| 63 | /* The buffer in which the last search was performed, or |
| 64 | Qt if the last search was done in a string; |
| 65 | Qnil if no searching has been done yet. */ |
| 66 | static Lisp_Object last_thing_searched; |
| 67 | |
| 68 | /* error condition signalled when regexp compile_pattern fails */ |
| 69 | |
| 70 | Lisp_Object Qinvalid_regexp; |
| 71 | |
| 72 | static void |
| 73 | matcher_overflow () |
| 74 | { |
| 75 | error ("Stack overflow in regexp matcher"); |
| 76 | } |
| 77 | |
| 78 | #ifdef __STDC__ |
| 79 | #define CONST const |
| 80 | #else |
| 81 | #define CONST |
| 82 | #endif |
| 83 | |
| 84 | /* Compile a regexp and signal a Lisp error if anything goes wrong. */ |
| 85 | |
| 86 | compile_pattern (pattern, bufp, regp, translate) |
| 87 | Lisp_Object pattern; |
| 88 | struct re_pattern_buffer *bufp; |
| 89 | struct re_registers *regp; |
| 90 | char *translate; |
| 91 | { |
| 92 | CONST char *val; |
| 93 | Lisp_Object dummy; |
| 94 | |
| 95 | if (EQ (pattern, last_regexp) |
| 96 | && translate == bufp->translate) |
| 97 | return; |
| 98 | |
| 99 | last_regexp = Qnil; |
| 100 | bufp->translate = translate; |
| 101 | BLOCK_INPUT; |
| 102 | val = re_compile_pattern ((char *) XSTRING (pattern)->data, |
| 103 | XSTRING (pattern)->size, |
| 104 | bufp); |
| 105 | UNBLOCK_INPUT; |
| 106 | if (val) |
| 107 | { |
| 108 | dummy = build_string (val); |
| 109 | while (1) |
| 110 | Fsignal (Qinvalid_regexp, Fcons (dummy, Qnil)); |
| 111 | } |
| 112 | |
| 113 | last_regexp = pattern; |
| 114 | |
| 115 | /* Advise the searching functions about the space we have allocated |
| 116 | for register data. */ |
| 117 | BLOCK_INPUT; |
| 118 | if (regp) |
| 119 | re_set_registers (bufp, regp, regp->num_regs, regp->start, regp->end); |
| 120 | UNBLOCK_INPUT; |
| 121 | |
| 122 | return; |
| 123 | } |
| 124 | |
| 125 | /* Error condition used for failing searches */ |
| 126 | Lisp_Object Qsearch_failed; |
| 127 | |
| 128 | Lisp_Object |
| 129 | signal_failure (arg) |
| 130 | Lisp_Object arg; |
| 131 | { |
| 132 | Fsignal (Qsearch_failed, Fcons (arg, Qnil)); |
| 133 | return Qnil; |
| 134 | } |
| 135 | \f |
| 136 | DEFUN ("looking-at", Flooking_at, Slooking_at, 1, 1, 0, |
| 137 | "Return t if text after point matches regular expression PAT.\n\ |
| 138 | This function modifies the match data that `match-beginning',\n\ |
| 139 | `match-end' and `match-data' access; save and restore the match\n\ |
| 140 | data if you want to preserve them.") |
| 141 | (string) |
| 142 | Lisp_Object string; |
| 143 | { |
| 144 | Lisp_Object val; |
| 145 | unsigned char *p1, *p2; |
| 146 | int s1, s2; |
| 147 | register int i; |
| 148 | |
| 149 | CHECK_STRING (string, 0); |
| 150 | compile_pattern (string, &searchbuf, &search_regs, |
| 151 | !NILP (current_buffer->case_fold_search) ? DOWNCASE_TABLE : 0); |
| 152 | |
| 153 | immediate_quit = 1; |
| 154 | QUIT; /* Do a pending quit right away, to avoid paradoxical behavior */ |
| 155 | |
| 156 | /* Get pointers and sizes of the two strings |
| 157 | that make up the visible portion of the buffer. */ |
| 158 | |
| 159 | p1 = BEGV_ADDR; |
| 160 | s1 = GPT - BEGV; |
| 161 | p2 = GAP_END_ADDR; |
| 162 | s2 = ZV - GPT; |
| 163 | if (s1 < 0) |
| 164 | { |
| 165 | p2 = p1; |
| 166 | s2 = ZV - BEGV; |
| 167 | s1 = 0; |
| 168 | } |
| 169 | if (s2 < 0) |
| 170 | { |
| 171 | s1 = ZV - BEGV; |
| 172 | s2 = 0; |
| 173 | } |
| 174 | |
| 175 | i = re_match_2 (&searchbuf, (char *) p1, s1, (char *) p2, s2, |
| 176 | point - BEGV, &search_regs, |
| 177 | ZV - BEGV); |
| 178 | if (i == -2) |
| 179 | matcher_overflow (); |
| 180 | |
| 181 | val = (0 <= i ? Qt : Qnil); |
| 182 | for (i = 0; i < search_regs.num_regs; i++) |
| 183 | if (search_regs.start[i] >= 0) |
| 184 | { |
| 185 | search_regs.start[i] += BEGV; |
| 186 | search_regs.end[i] += BEGV; |
| 187 | } |
| 188 | XSET (last_thing_searched, Lisp_Buffer, current_buffer); |
| 189 | immediate_quit = 0; |
| 190 | return val; |
| 191 | } |
| 192 | |
| 193 | DEFUN ("string-match", Fstring_match, Sstring_match, 2, 3, 0, |
| 194 | "Return index of start of first match for REGEXP in STRING, or nil.\n\ |
| 195 | If third arg START is non-nil, start search at that index in STRING.\n\ |
| 196 | For index of first char beyond the match, do (match-end 0).\n\ |
| 197 | `match-end' and `match-beginning' also give indices of substrings\n\ |
| 198 | matched by parenthesis constructs in the pattern.") |
| 199 | (regexp, string, start) |
| 200 | Lisp_Object regexp, string, start; |
| 201 | { |
| 202 | int val; |
| 203 | int s; |
| 204 | |
| 205 | CHECK_STRING (regexp, 0); |
| 206 | CHECK_STRING (string, 1); |
| 207 | |
| 208 | if (NILP (start)) |
| 209 | s = 0; |
| 210 | else |
| 211 | { |
| 212 | int len = XSTRING (string)->size; |
| 213 | |
| 214 | CHECK_NUMBER (start, 2); |
| 215 | s = XINT (start); |
| 216 | if (s < 0 && -s <= len) |
| 217 | s = len - s; |
| 218 | else if (0 > s || s > len) |
| 219 | args_out_of_range (string, start); |
| 220 | } |
| 221 | |
| 222 | compile_pattern (regexp, &searchbuf, &search_regs, |
| 223 | !NILP (current_buffer->case_fold_search) ? DOWNCASE_TABLE : 0); |
| 224 | immediate_quit = 1; |
| 225 | val = re_search (&searchbuf, (char *) XSTRING (string)->data, |
| 226 | XSTRING (string)->size, s, XSTRING (string)->size - s, |
| 227 | &search_regs); |
| 228 | immediate_quit = 0; |
| 229 | last_thing_searched = Qt; |
| 230 | if (val == -2) |
| 231 | matcher_overflow (); |
| 232 | if (val < 0) return Qnil; |
| 233 | return make_number (val); |
| 234 | } |
| 235 | |
| 236 | /* Match REGEXP against STRING, searching all of STRING, |
| 237 | and return the index of the match, or negative on failure. |
| 238 | This does not clobber the match data. */ |
| 239 | |
| 240 | int |
| 241 | fast_string_match (regexp, string) |
| 242 | Lisp_Object regexp, string; |
| 243 | { |
| 244 | int val; |
| 245 | |
| 246 | compile_pattern (regexp, &searchbuf, 0, 0); |
| 247 | immediate_quit = 1; |
| 248 | val = re_search (&searchbuf, (char *) XSTRING (string)->data, |
| 249 | XSTRING (string)->size, 0, XSTRING (string)->size, |
| 250 | 0); |
| 251 | immediate_quit = 0; |
| 252 | return val; |
| 253 | } |
| 254 | \f |
| 255 | /* Search for COUNT instances of the character TARGET, starting at START. |
| 256 | If COUNT is negative, search backwards. |
| 257 | |
| 258 | If we find COUNT instances, set *SHORTAGE to zero, and return the |
| 259 | position after the COUNTth match. Note that for reverse motion |
| 260 | this is not the same as the usual convention for Emacs motion commands. |
| 261 | |
| 262 | If we don't find COUNT instances before reaching the end of the |
| 263 | buffer (or the beginning, if scanning backwards), set *SHORTAGE to |
| 264 | the number of TARGETs left unfound, and return the end of the |
| 265 | buffer we bumped up against. */ |
| 266 | |
| 267 | scan_buffer (target, start, count, shortage) |
| 268 | int *shortage, start; |
| 269 | register int count, target; |
| 270 | { |
| 271 | int limit = ((count > 0) ? ZV - 1 : BEGV); |
| 272 | int direction = ((count > 0) ? 1 : -1); |
| 273 | |
| 274 | register unsigned char *cursor; |
| 275 | unsigned char *base; |
| 276 | |
| 277 | register int ceiling; |
| 278 | register unsigned char *ceiling_addr; |
| 279 | |
| 280 | if (shortage != 0) |
| 281 | *shortage = 0; |
| 282 | |
| 283 | immediate_quit = 1; |
| 284 | |
| 285 | if (count > 0) |
| 286 | while (start != limit + 1) |
| 287 | { |
| 288 | ceiling = BUFFER_CEILING_OF (start); |
| 289 | ceiling = min (limit, ceiling); |
| 290 | ceiling_addr = &FETCH_CHAR (ceiling) + 1; |
| 291 | base = (cursor = &FETCH_CHAR (start)); |
| 292 | while (1) |
| 293 | { |
| 294 | while (*cursor != target && ++cursor != ceiling_addr) |
| 295 | ; |
| 296 | if (cursor != ceiling_addr) |
| 297 | { |
| 298 | if (--count == 0) |
| 299 | { |
| 300 | immediate_quit = 0; |
| 301 | return (start + cursor - base + 1); |
| 302 | } |
| 303 | else |
| 304 | if (++cursor == ceiling_addr) |
| 305 | break; |
| 306 | } |
| 307 | else |
| 308 | break; |
| 309 | } |
| 310 | start += cursor - base; |
| 311 | } |
| 312 | else |
| 313 | { |
| 314 | start--; /* first character we scan */ |
| 315 | while (start > limit - 1) |
| 316 | { /* we WILL scan under start */ |
| 317 | ceiling = BUFFER_FLOOR_OF (start); |
| 318 | ceiling = max (limit, ceiling); |
| 319 | ceiling_addr = &FETCH_CHAR (ceiling) - 1; |
| 320 | base = (cursor = &FETCH_CHAR (start)); |
| 321 | cursor++; |
| 322 | while (1) |
| 323 | { |
| 324 | while (--cursor != ceiling_addr && *cursor != target) |
| 325 | ; |
| 326 | if (cursor != ceiling_addr) |
| 327 | { |
| 328 | if (++count == 0) |
| 329 | { |
| 330 | immediate_quit = 0; |
| 331 | return (start + cursor - base + 1); |
| 332 | } |
| 333 | } |
| 334 | else |
| 335 | break; |
| 336 | } |
| 337 | start += cursor - base; |
| 338 | } |
| 339 | } |
| 340 | immediate_quit = 0; |
| 341 | if (shortage != 0) |
| 342 | *shortage = count * direction; |
| 343 | return (start + ((direction == 1 ? 0 : 1))); |
| 344 | } |
| 345 | |
| 346 | int |
| 347 | find_next_newline (from, cnt) |
| 348 | register int from, cnt; |
| 349 | { |
| 350 | return (scan_buffer ('\n', from, cnt, (int *) 0)); |
| 351 | } |
| 352 | \f |
| 353 | Lisp_Object skip_chars (); |
| 354 | |
| 355 | DEFUN ("skip-chars-forward", Fskip_chars_forward, Sskip_chars_forward, 1, 2, 0, |
| 356 | "Move point forward, stopping before a char not in CHARS, or at position LIM.\n\ |
| 357 | CHARS is like the inside of a `[...]' in a regular expression\n\ |
| 358 | except that `]' is never special and `\\' quotes `^', `-' or `\\'.\n\ |
| 359 | Thus, with arg \"a-zA-Z\", this skips letters stopping before first nonletter.\n\ |
| 360 | With arg \"^a-zA-Z\", skips nonletters stopping before first letter.\n\ |
| 361 | Returns the distance traveled, either zero or positive.") |
| 362 | (string, lim) |
| 363 | Lisp_Object string, lim; |
| 364 | { |
| 365 | return skip_chars (1, 0, string, lim); |
| 366 | } |
| 367 | |
| 368 | DEFUN ("skip-chars-backward", Fskip_chars_backward, Sskip_chars_backward, 1, 2, 0, |
| 369 | "Move point backward, stopping after a char not in CHARS, or at position LIM.\n\ |
| 370 | See `skip-chars-forward' for details.\n\ |
| 371 | Returns the distance traveled, either zero or negative.") |
| 372 | (string, lim) |
| 373 | Lisp_Object string, lim; |
| 374 | { |
| 375 | return skip_chars (0, 0, string, lim); |
| 376 | } |
| 377 | |
| 378 | DEFUN ("skip-syntax-forward", Fskip_syntax_forward, Sskip_syntax_forward, 1, 2, 0, |
| 379 | "Move point forward across chars in specified syntax classes.\n\ |
| 380 | SYNTAX is a string of syntax code characters.\n\ |
| 381 | Stop before a char whose syntax is not in SYNTAX, or at position LIM.\n\ |
| 382 | If SYNTAX starts with ^, skip characters whose syntax is NOT in SYNTAX.\n\ |
| 383 | This function returns the distance traveled, either zero or positive.") |
| 384 | (syntax, lim) |
| 385 | Lisp_Object syntax, lim; |
| 386 | { |
| 387 | return skip_chars (1, 1, syntax, lim); |
| 388 | } |
| 389 | |
| 390 | DEFUN ("skip-syntax-backward", Fskip_syntax_backward, Sskip_syntax_backward, 1, 2, 0, |
| 391 | "Move point backward across chars in specified syntax classes.\n\ |
| 392 | SYNTAX is a string of syntax code characters.\n\ |
| 393 | Stop on reaching a char whose syntax is not in SYNTAX, or at position LIM.\n\ |
| 394 | If SYNTAX starts with ^, skip characters whose syntax is NOT in SYNTAX.\n\ |
| 395 | This function returns the distance traveled, either zero or negative.") |
| 396 | (syntax, lim) |
| 397 | Lisp_Object syntax, lim; |
| 398 | { |
| 399 | return skip_chars (0, 1, syntax, lim); |
| 400 | } |
| 401 | |
| 402 | Lisp_Object |
| 403 | skip_chars (forwardp, syntaxp, string, lim) |
| 404 | int forwardp, syntaxp; |
| 405 | Lisp_Object string, lim; |
| 406 | { |
| 407 | register unsigned char *p, *pend; |
| 408 | register unsigned char c; |
| 409 | unsigned char fastmap[0400]; |
| 410 | int negate = 0; |
| 411 | register int i; |
| 412 | |
| 413 | CHECK_STRING (string, 0); |
| 414 | |
| 415 | if (NILP (lim)) |
| 416 | XSET (lim, Lisp_Int, forwardp ? ZV : BEGV); |
| 417 | else |
| 418 | CHECK_NUMBER_COERCE_MARKER (lim, 1); |
| 419 | |
| 420 | #if 0 /* This breaks some things... jla. */ |
| 421 | /* In any case, don't allow scan outside bounds of buffer. */ |
| 422 | if (XFASTINT (lim) > ZV) |
| 423 | XFASTINT (lim) = ZV; |
| 424 | if (XFASTINT (lim) < BEGV) |
| 425 | XFASTINT (lim) = BEGV; |
| 426 | #endif |
| 427 | |
| 428 | p = XSTRING (string)->data; |
| 429 | pend = p + XSTRING (string)->size; |
| 430 | bzero (fastmap, sizeof fastmap); |
| 431 | |
| 432 | if (p != pend && *p == '^') |
| 433 | { |
| 434 | negate = 1; p++; |
| 435 | } |
| 436 | |
| 437 | /* Find the characters specified and set their elements of fastmap. |
| 438 | If syntaxp, each character counts as itself. |
| 439 | Otherwise, handle backslashes and ranges specially */ |
| 440 | |
| 441 | while (p != pend) |
| 442 | { |
| 443 | c = *p++; |
| 444 | if (syntaxp) |
| 445 | fastmap[c] = 1; |
| 446 | else |
| 447 | { |
| 448 | if (c == '\\') |
| 449 | { |
| 450 | if (p == pend) break; |
| 451 | c = *p++; |
| 452 | } |
| 453 | if (p != pend && *p == '-') |
| 454 | { |
| 455 | p++; |
| 456 | if (p == pend) break; |
| 457 | while (c <= *p) |
| 458 | { |
| 459 | fastmap[c] = 1; |
| 460 | c++; |
| 461 | } |
| 462 | p++; |
| 463 | } |
| 464 | else |
| 465 | fastmap[c] = 1; |
| 466 | } |
| 467 | } |
| 468 | |
| 469 | /* If ^ was the first character, complement the fastmap. */ |
| 470 | |
| 471 | if (negate) |
| 472 | for (i = 0; i < sizeof fastmap; i++) |
| 473 | fastmap[i] ^= 1; |
| 474 | |
| 475 | { |
| 476 | int start_point = point; |
| 477 | |
| 478 | immediate_quit = 1; |
| 479 | if (syntaxp) |
| 480 | { |
| 481 | |
| 482 | if (forwardp) |
| 483 | { |
| 484 | while (point < XINT (lim) |
| 485 | && fastmap[(unsigned char) syntax_code_spec[(int) SYNTAX (FETCH_CHAR (point))]]) |
| 486 | SET_PT (point + 1); |
| 487 | } |
| 488 | else |
| 489 | { |
| 490 | while (point > XINT (lim) |
| 491 | && fastmap[(unsigned char) syntax_code_spec[(int) SYNTAX (FETCH_CHAR (point - 1))]]) |
| 492 | SET_PT (point - 1); |
| 493 | } |
| 494 | } |
| 495 | else |
| 496 | { |
| 497 | if (forwardp) |
| 498 | { |
| 499 | while (point < XINT (lim) && fastmap[FETCH_CHAR (point)]) |
| 500 | SET_PT (point + 1); |
| 501 | } |
| 502 | else |
| 503 | { |
| 504 | while (point > XINT (lim) && fastmap[FETCH_CHAR (point - 1)]) |
| 505 | SET_PT (point - 1); |
| 506 | } |
| 507 | } |
| 508 | immediate_quit = 0; |
| 509 | |
| 510 | return make_number (point - start_point); |
| 511 | } |
| 512 | } |
| 513 | \f |
| 514 | /* Subroutines of Lisp buffer search functions. */ |
| 515 | |
| 516 | static Lisp_Object |
| 517 | search_command (string, bound, noerror, count, direction, RE) |
| 518 | Lisp_Object string, bound, noerror, count; |
| 519 | int direction; |
| 520 | int RE; |
| 521 | { |
| 522 | register int np; |
| 523 | int lim; |
| 524 | int n = direction; |
| 525 | |
| 526 | if (!NILP (count)) |
| 527 | { |
| 528 | CHECK_NUMBER (count, 3); |
| 529 | n *= XINT (count); |
| 530 | } |
| 531 | |
| 532 | CHECK_STRING (string, 0); |
| 533 | if (NILP (bound)) |
| 534 | lim = n > 0 ? ZV : BEGV; |
| 535 | else |
| 536 | { |
| 537 | CHECK_NUMBER_COERCE_MARKER (bound, 1); |
| 538 | lim = XINT (bound); |
| 539 | if (n > 0 ? lim < point : lim > point) |
| 540 | error ("Invalid search bound (wrong side of point)"); |
| 541 | if (lim > ZV) |
| 542 | lim = ZV; |
| 543 | if (lim < BEGV) |
| 544 | lim = BEGV; |
| 545 | } |
| 546 | |
| 547 | np = search_buffer (string, point, lim, n, RE, |
| 548 | (!NILP (current_buffer->case_fold_search) |
| 549 | ? XSTRING (current_buffer->case_canon_table)->data : 0), |
| 550 | (!NILP (current_buffer->case_fold_search) |
| 551 | ? XSTRING (current_buffer->case_eqv_table)->data : 0)); |
| 552 | if (np <= 0) |
| 553 | { |
| 554 | if (NILP (noerror)) |
| 555 | return signal_failure (string); |
| 556 | if (!EQ (noerror, Qt)) |
| 557 | { |
| 558 | if (lim < BEGV || lim > ZV) |
| 559 | abort (); |
| 560 | SET_PT (lim); |
| 561 | return Qnil; |
| 562 | #if 0 /* This would be clean, but maybe programs depend on |
| 563 | a value of nil here. */ |
| 564 | np = lim; |
| 565 | #endif |
| 566 | } |
| 567 | else |
| 568 | return Qnil; |
| 569 | } |
| 570 | |
| 571 | if (np < BEGV || np > ZV) |
| 572 | abort (); |
| 573 | |
| 574 | SET_PT (np); |
| 575 | |
| 576 | return make_number (np); |
| 577 | } |
| 578 | \f |
| 579 | /* search for the n'th occurrence of STRING in the current buffer, |
| 580 | starting at position POS and stopping at position LIM, |
| 581 | treating PAT as a literal string if RE is false or as |
| 582 | a regular expression if RE is true. |
| 583 | |
| 584 | If N is positive, searching is forward and LIM must be greater than POS. |
| 585 | If N is negative, searching is backward and LIM must be less than POS. |
| 586 | |
| 587 | Returns -x if only N-x occurrences found (x > 0), |
| 588 | or else the position at the beginning of the Nth occurrence |
| 589 | (if searching backward) or the end (if searching forward). */ |
| 590 | |
| 591 | search_buffer (string, pos, lim, n, RE, trt, inverse_trt) |
| 592 | Lisp_Object string; |
| 593 | int pos; |
| 594 | int lim; |
| 595 | int n; |
| 596 | int RE; |
| 597 | register unsigned char *trt; |
| 598 | register unsigned char *inverse_trt; |
| 599 | { |
| 600 | int len = XSTRING (string)->size; |
| 601 | unsigned char *base_pat = XSTRING (string)->data; |
| 602 | register int *BM_tab; |
| 603 | int *BM_tab_base; |
| 604 | register int direction = ((n > 0) ? 1 : -1); |
| 605 | register int dirlen; |
| 606 | int infinity, limit, k, stride_for_teases; |
| 607 | register unsigned char *pat, *cursor, *p_limit; |
| 608 | register int i, j; |
| 609 | unsigned char *p1, *p2; |
| 610 | int s1, s2; |
| 611 | |
| 612 | /* Null string is found at starting position. */ |
| 613 | if (!len) |
| 614 | return pos; |
| 615 | |
| 616 | if (RE) |
| 617 | compile_pattern (string, &searchbuf, &search_regs, (char *) trt); |
| 618 | |
| 619 | if (RE /* Here we detect whether the */ |
| 620 | /* generality of an RE search is */ |
| 621 | /* really needed. */ |
| 622 | /* first item is "exact match" */ |
| 623 | && *(searchbuf.buffer) == (char) RE_EXACTN_VALUE |
| 624 | && searchbuf.buffer[1] + 2 == searchbuf.used) /*first is ONLY item */ |
| 625 | { |
| 626 | RE = 0; /* can do straight (non RE) search */ |
| 627 | pat = (base_pat = (unsigned char *) searchbuf.buffer + 2); |
| 628 | /* trt already applied */ |
| 629 | len = searchbuf.used - 2; |
| 630 | } |
| 631 | else if (!RE) |
| 632 | { |
| 633 | pat = (unsigned char *) alloca (len); |
| 634 | |
| 635 | for (i = len; i--;) /* Copy the pattern; apply trt */ |
| 636 | *pat++ = (((int) trt) ? trt [*base_pat++] : *base_pat++); |
| 637 | pat -= len; base_pat = pat; |
| 638 | } |
| 639 | |
| 640 | if (RE) |
| 641 | { |
| 642 | immediate_quit = 1; /* Quit immediately if user types ^G, |
| 643 | because letting this function finish |
| 644 | can take too long. */ |
| 645 | QUIT; /* Do a pending quit right away, |
| 646 | to avoid paradoxical behavior */ |
| 647 | /* Get pointers and sizes of the two strings |
| 648 | that make up the visible portion of the buffer. */ |
| 649 | |
| 650 | p1 = BEGV_ADDR; |
| 651 | s1 = GPT - BEGV; |
| 652 | p2 = GAP_END_ADDR; |
| 653 | s2 = ZV - GPT; |
| 654 | if (s1 < 0) |
| 655 | { |
| 656 | p2 = p1; |
| 657 | s2 = ZV - BEGV; |
| 658 | s1 = 0; |
| 659 | } |
| 660 | if (s2 < 0) |
| 661 | { |
| 662 | s1 = ZV - BEGV; |
| 663 | s2 = 0; |
| 664 | } |
| 665 | while (n < 0) |
| 666 | { |
| 667 | int val; |
| 668 | val = re_search_2 (&searchbuf, (char *) p1, s1, (char *) p2, s2, |
| 669 | pos - BEGV, lim - pos, &search_regs, |
| 670 | /* Don't allow match past current point */ |
| 671 | pos - BEGV); |
| 672 | if (val == -2) |
| 673 | matcher_overflow (); |
| 674 | if (val >= 0) |
| 675 | { |
| 676 | j = BEGV; |
| 677 | for (i = 0; i < search_regs.num_regs; i++) |
| 678 | if (search_regs.start[i] >= 0) |
| 679 | { |
| 680 | search_regs.start[i] += j; |
| 681 | search_regs.end[i] += j; |
| 682 | } |
| 683 | XSET (last_thing_searched, Lisp_Buffer, current_buffer); |
| 684 | /* Set pos to the new position. */ |
| 685 | pos = search_regs.start[0]; |
| 686 | } |
| 687 | else |
| 688 | { |
| 689 | immediate_quit = 0; |
| 690 | return (n); |
| 691 | } |
| 692 | n++; |
| 693 | } |
| 694 | while (n > 0) |
| 695 | { |
| 696 | int val; |
| 697 | val = re_search_2 (&searchbuf, (char *) p1, s1, (char *) p2, s2, |
| 698 | pos - BEGV, lim - pos, &search_regs, |
| 699 | lim - BEGV); |
| 700 | if (val == -2) |
| 701 | matcher_overflow (); |
| 702 | if (val >= 0) |
| 703 | { |
| 704 | j = BEGV; |
| 705 | for (i = 0; i < search_regs.num_regs; i++) |
| 706 | if (search_regs.start[i] >= 0) |
| 707 | { |
| 708 | search_regs.start[i] += j; |
| 709 | search_regs.end[i] += j; |
| 710 | } |
| 711 | XSET (last_thing_searched, Lisp_Buffer, current_buffer); |
| 712 | pos = search_regs.end[0]; |
| 713 | } |
| 714 | else |
| 715 | { |
| 716 | immediate_quit = 0; |
| 717 | return (0 - n); |
| 718 | } |
| 719 | n--; |
| 720 | } |
| 721 | immediate_quit = 0; |
| 722 | return (pos); |
| 723 | } |
| 724 | else /* non-RE case */ |
| 725 | { |
| 726 | #ifdef C_ALLOCA |
| 727 | int BM_tab_space[0400]; |
| 728 | BM_tab = &BM_tab_space[0]; |
| 729 | #else |
| 730 | BM_tab = (int *) alloca (0400 * sizeof (int)); |
| 731 | #endif |
| 732 | /* The general approach is that we are going to maintain that we know */ |
| 733 | /* the first (closest to the present position, in whatever direction */ |
| 734 | /* we're searching) character that could possibly be the last */ |
| 735 | /* (furthest from present position) character of a valid match. We */ |
| 736 | /* advance the state of our knowledge by looking at that character */ |
| 737 | /* and seeing whether it indeed matches the last character of the */ |
| 738 | /* pattern. If it does, we take a closer look. If it does not, we */ |
| 739 | /* move our pointer (to putative last characters) as far as is */ |
| 740 | /* logically possible. This amount of movement, which I call a */ |
| 741 | /* stride, will be the length of the pattern if the actual character */ |
| 742 | /* appears nowhere in the pattern, otherwise it will be the distance */ |
| 743 | /* from the last occurrence of that character to the end of the */ |
| 744 | /* pattern. */ |
| 745 | /* As a coding trick, an enormous stride is coded into the table for */ |
| 746 | /* characters that match the last character. This allows use of only */ |
| 747 | /* a single test, a test for having gone past the end of the */ |
| 748 | /* permissible match region, to test for both possible matches (when */ |
| 749 | /* the stride goes past the end immediately) and failure to */ |
| 750 | /* match (where you get nudged past the end one stride at a time). */ |
| 751 | |
| 752 | /* Here we make a "mickey mouse" BM table. The stride of the search */ |
| 753 | /* is determined only by the last character of the putative match. */ |
| 754 | /* If that character does not match, we will stride the proper */ |
| 755 | /* distance to propose a match that superimposes it on the last */ |
| 756 | /* instance of a character that matches it (per trt), or misses */ |
| 757 | /* it entirely if there is none. */ |
| 758 | |
| 759 | dirlen = len * direction; |
| 760 | infinity = dirlen - (lim + pos + len + len) * direction; |
| 761 | if (direction < 0) |
| 762 | pat = (base_pat += len - 1); |
| 763 | BM_tab_base = BM_tab; |
| 764 | BM_tab += 0400; |
| 765 | j = dirlen; /* to get it in a register */ |
| 766 | /* A character that does not appear in the pattern induces a */ |
| 767 | /* stride equal to the pattern length. */ |
| 768 | while (BM_tab_base != BM_tab) |
| 769 | { |
| 770 | *--BM_tab = j; |
| 771 | *--BM_tab = j; |
| 772 | *--BM_tab = j; |
| 773 | *--BM_tab = j; |
| 774 | } |
| 775 | i = 0; |
| 776 | while (i != infinity) |
| 777 | { |
| 778 | j = pat[i]; i += direction; |
| 779 | if (i == dirlen) i = infinity; |
| 780 | if ((int) trt) |
| 781 | { |
| 782 | k = (j = trt[j]); |
| 783 | if (i == infinity) |
| 784 | stride_for_teases = BM_tab[j]; |
| 785 | BM_tab[j] = dirlen - i; |
| 786 | /* A translation table is accompanied by its inverse -- see */ |
| 787 | /* comment following downcase_table for details */ |
| 788 | while ((j = inverse_trt[j]) != k) |
| 789 | BM_tab[j] = dirlen - i; |
| 790 | } |
| 791 | else |
| 792 | { |
| 793 | if (i == infinity) |
| 794 | stride_for_teases = BM_tab[j]; |
| 795 | BM_tab[j] = dirlen - i; |
| 796 | } |
| 797 | /* stride_for_teases tells how much to stride if we get a */ |
| 798 | /* match on the far character but are subsequently */ |
| 799 | /* disappointed, by recording what the stride would have been */ |
| 800 | /* for that character if the last character had been */ |
| 801 | /* different. */ |
| 802 | } |
| 803 | infinity = dirlen - infinity; |
| 804 | pos += dirlen - ((direction > 0) ? direction : 0); |
| 805 | /* loop invariant - pos points at where last char (first char if reverse) |
| 806 | of pattern would align in a possible match. */ |
| 807 | while (n != 0) |
| 808 | { |
| 809 | if ((lim - pos - (direction > 0)) * direction < 0) |
| 810 | return (n * (0 - direction)); |
| 811 | /* First we do the part we can by pointers (maybe nothing) */ |
| 812 | QUIT; |
| 813 | pat = base_pat; |
| 814 | limit = pos - dirlen + direction; |
| 815 | limit = ((direction > 0) |
| 816 | ? BUFFER_CEILING_OF (limit) |
| 817 | : BUFFER_FLOOR_OF (limit)); |
| 818 | /* LIMIT is now the last (not beyond-last!) value |
| 819 | POS can take on without hitting edge of buffer or the gap. */ |
| 820 | limit = ((direction > 0) |
| 821 | ? min (lim - 1, min (limit, pos + 20000)) |
| 822 | : max (lim, max (limit, pos - 20000))); |
| 823 | if ((limit - pos) * direction > 20) |
| 824 | { |
| 825 | p_limit = &FETCH_CHAR (limit); |
| 826 | p2 = (cursor = &FETCH_CHAR (pos)); |
| 827 | /* In this loop, pos + cursor - p2 is the surrogate for pos */ |
| 828 | while (1) /* use one cursor setting as long as i can */ |
| 829 | { |
| 830 | if (direction > 0) /* worth duplicating */ |
| 831 | { |
| 832 | /* Use signed comparison if appropriate |
| 833 | to make cursor+infinity sure to be > p_limit. |
| 834 | Assuming that the buffer lies in a range of addresses |
| 835 | that are all "positive" (as ints) or all "negative", |
| 836 | either kind of comparison will work as long |
| 837 | as we don't step by infinity. So pick the kind |
| 838 | that works when we do step by infinity. */ |
| 839 | if ((int) (p_limit + infinity) > (int) p_limit) |
| 840 | while ((int) cursor <= (int) p_limit) |
| 841 | cursor += BM_tab[*cursor]; |
| 842 | else |
| 843 | while ((unsigned int) cursor <= (unsigned int) p_limit) |
| 844 | cursor += BM_tab[*cursor]; |
| 845 | } |
| 846 | else |
| 847 | { |
| 848 | if ((int) (p_limit + infinity) < (int) p_limit) |
| 849 | while ((int) cursor >= (int) p_limit) |
| 850 | cursor += BM_tab[*cursor]; |
| 851 | else |
| 852 | while ((unsigned int) cursor >= (unsigned int) p_limit) |
| 853 | cursor += BM_tab[*cursor]; |
| 854 | } |
| 855 | /* If you are here, cursor is beyond the end of the searched region. */ |
| 856 | /* This can happen if you match on the far character of the pattern, */ |
| 857 | /* because the "stride" of that character is infinity, a number able */ |
| 858 | /* to throw you well beyond the end of the search. It can also */ |
| 859 | /* happen if you fail to match within the permitted region and would */ |
| 860 | /* otherwise try a character beyond that region */ |
| 861 | if ((cursor - p_limit) * direction <= len) |
| 862 | break; /* a small overrun is genuine */ |
| 863 | cursor -= infinity; /* large overrun = hit */ |
| 864 | i = dirlen - direction; |
| 865 | if ((int) trt) |
| 866 | { |
| 867 | while ((i -= direction) + direction != 0) |
| 868 | if (pat[i] != trt[*(cursor -= direction)]) |
| 869 | break; |
| 870 | } |
| 871 | else |
| 872 | { |
| 873 | while ((i -= direction) + direction != 0) |
| 874 | if (pat[i] != *(cursor -= direction)) |
| 875 | break; |
| 876 | } |
| 877 | cursor += dirlen - i - direction; /* fix cursor */ |
| 878 | if (i + direction == 0) |
| 879 | { |
| 880 | cursor -= direction; |
| 881 | |
| 882 | /* Make sure we have registers in which to store |
| 883 | the match position. */ |
| 884 | if (search_regs.num_regs == 0) |
| 885 | { |
| 886 | regoff_t *starts, *ends; |
| 887 | |
| 888 | starts = |
| 889 | (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
| 890 | ends = |
| 891 | (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
| 892 | BLOCK_INPUT; |
| 893 | re_set_registers (&searchbuf, |
| 894 | &search_regs, |
| 895 | 2, starts, ends); |
| 896 | UNBLOCK_INPUT; |
| 897 | } |
| 898 | |
| 899 | search_regs.start[0] |
| 900 | = pos + cursor - p2 + ((direction > 0) |
| 901 | ? 1 - len : 0); |
| 902 | search_regs.end[0] = len + search_regs.start[0]; |
| 903 | XSET (last_thing_searched, Lisp_Buffer, current_buffer); |
| 904 | if ((n -= direction) != 0) |
| 905 | cursor += dirlen; /* to resume search */ |
| 906 | else |
| 907 | return ((direction > 0) |
| 908 | ? search_regs.end[0] : search_regs.start[0]); |
| 909 | } |
| 910 | else |
| 911 | cursor += stride_for_teases; /* <sigh> we lose - */ |
| 912 | } |
| 913 | pos += cursor - p2; |
| 914 | } |
| 915 | else |
| 916 | /* Now we'll pick up a clump that has to be done the hard */ |
| 917 | /* way because it covers a discontinuity */ |
| 918 | { |
| 919 | limit = ((direction > 0) |
| 920 | ? BUFFER_CEILING_OF (pos - dirlen + 1) |
| 921 | : BUFFER_FLOOR_OF (pos - dirlen - 1)); |
| 922 | limit = ((direction > 0) |
| 923 | ? min (limit + len, lim - 1) |
| 924 | : max (limit - len, lim)); |
| 925 | /* LIMIT is now the last value POS can have |
| 926 | and still be valid for a possible match. */ |
| 927 | while (1) |
| 928 | { |
| 929 | /* This loop can be coded for space rather than */ |
| 930 | /* speed because it will usually run only once. */ |
| 931 | /* (the reach is at most len + 21, and typically */ |
| 932 | /* does not exceed len) */ |
| 933 | while ((limit - pos) * direction >= 0) |
| 934 | pos += BM_tab[FETCH_CHAR(pos)]; |
| 935 | /* now run the same tests to distinguish going off the */ |
| 936 | /* end, a match or a phony match. */ |
| 937 | if ((pos - limit) * direction <= len) |
| 938 | break; /* ran off the end */ |
| 939 | /* Found what might be a match. |
| 940 | Set POS back to last (first if reverse) char pos. */ |
| 941 | pos -= infinity; |
| 942 | i = dirlen - direction; |
| 943 | while ((i -= direction) + direction != 0) |
| 944 | { |
| 945 | pos -= direction; |
| 946 | if (pat[i] != (((int) trt) |
| 947 | ? trt[FETCH_CHAR(pos)] |
| 948 | : FETCH_CHAR (pos))) |
| 949 | break; |
| 950 | } |
| 951 | /* Above loop has moved POS part or all the way |
| 952 | back to the first char pos (last char pos if reverse). |
| 953 | Set it once again at the last (first if reverse) char. */ |
| 954 | pos += dirlen - i- direction; |
| 955 | if (i + direction == 0) |
| 956 | { |
| 957 | pos -= direction; |
| 958 | |
| 959 | /* Make sure we have registers in which to store |
| 960 | the match position. */ |
| 961 | if (search_regs.num_regs == 0) |
| 962 | { |
| 963 | regoff_t *starts, *ends; |
| 964 | |
| 965 | starts = |
| 966 | (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
| 967 | ends = |
| 968 | (regoff_t *) xmalloc (2 * sizeof (regoff_t)); |
| 969 | BLOCK_INPUT; |
| 970 | re_set_registers (&searchbuf, |
| 971 | &search_regs, |
| 972 | 2, starts, ends); |
| 973 | UNBLOCK_INPUT; |
| 974 | } |
| 975 | |
| 976 | search_regs.start[0] |
| 977 | = pos + ((direction > 0) ? 1 - len : 0); |
| 978 | search_regs.end[0] = len + search_regs.start[0]; |
| 979 | XSET (last_thing_searched, Lisp_Buffer, current_buffer); |
| 980 | if ((n -= direction) != 0) |
| 981 | pos += dirlen; /* to resume search */ |
| 982 | else |
| 983 | return ((direction > 0) |
| 984 | ? search_regs.end[0] : search_regs.start[0]); |
| 985 | } |
| 986 | else |
| 987 | pos += stride_for_teases; |
| 988 | } |
| 989 | } |
| 990 | /* We have done one clump. Can we continue? */ |
| 991 | if ((lim - pos) * direction < 0) |
| 992 | return ((0 - n) * direction); |
| 993 | } |
| 994 | return pos; |
| 995 | } |
| 996 | } |
| 997 | \f |
| 998 | /* Given a string of words separated by word delimiters, |
| 999 | compute a regexp that matches those exact words |
| 1000 | separated by arbitrary punctuation. */ |
| 1001 | |
| 1002 | static Lisp_Object |
| 1003 | wordify (string) |
| 1004 | Lisp_Object string; |
| 1005 | { |
| 1006 | register unsigned char *p, *o; |
| 1007 | register int i, len, punct_count = 0, word_count = 0; |
| 1008 | Lisp_Object val; |
| 1009 | |
| 1010 | CHECK_STRING (string, 0); |
| 1011 | p = XSTRING (string)->data; |
| 1012 | len = XSTRING (string)->size; |
| 1013 | |
| 1014 | for (i = 0; i < len; i++) |
| 1015 | if (SYNTAX (p[i]) != Sword) |
| 1016 | { |
| 1017 | punct_count++; |
| 1018 | if (i > 0 && SYNTAX (p[i-1]) == Sword) word_count++; |
| 1019 | } |
| 1020 | if (SYNTAX (p[len-1]) == Sword) word_count++; |
| 1021 | if (!word_count) return build_string (""); |
| 1022 | |
| 1023 | val = make_string (p, len - punct_count + 5 * (word_count - 1) + 4); |
| 1024 | |
| 1025 | o = XSTRING (val)->data; |
| 1026 | *o++ = '\\'; |
| 1027 | *o++ = 'b'; |
| 1028 | |
| 1029 | for (i = 0; i < len; i++) |
| 1030 | if (SYNTAX (p[i]) == Sword) |
| 1031 | *o++ = p[i]; |
| 1032 | else if (i > 0 && SYNTAX (p[i-1]) == Sword && --word_count) |
| 1033 | { |
| 1034 | *o++ = '\\'; |
| 1035 | *o++ = 'W'; |
| 1036 | *o++ = '\\'; |
| 1037 | *o++ = 'W'; |
| 1038 | *o++ = '*'; |
| 1039 | } |
| 1040 | |
| 1041 | *o++ = '\\'; |
| 1042 | *o++ = 'b'; |
| 1043 | |
| 1044 | return val; |
| 1045 | } |
| 1046 | \f |
| 1047 | DEFUN ("search-backward", Fsearch_backward, Ssearch_backward, 1, 4, |
| 1048 | "sSearch backward: ", |
| 1049 | "Search backward from point for STRING.\n\ |
| 1050 | Set point to the beginning of the occurrence found, and return point.\n\ |
| 1051 | An optional second argument bounds the search; it is a buffer position.\n\ |
| 1052 | The match found must not extend before that position.\n\ |
| 1053 | Optional third argument, if t, means if fail just return nil (no error).\n\ |
| 1054 | If not nil and not t, position at limit of search and return nil.\n\ |
| 1055 | Optional fourth argument is repeat count--search for successive occurrences.\n\ |
| 1056 | See also the functions `match-beginning', `match-end' and `replace-match'.") |
| 1057 | (string, bound, noerror, count) |
| 1058 | Lisp_Object string, bound, noerror, count; |
| 1059 | { |
| 1060 | return search_command (string, bound, noerror, count, -1, 0); |
| 1061 | } |
| 1062 | |
| 1063 | DEFUN ("search-forward", Fsearch_forward, Ssearch_forward, 1, 4, "sSearch: ", |
| 1064 | "Search forward from point for STRING.\n\ |
| 1065 | Set point to the end of the occurrence found, and return point.\n\ |
| 1066 | An optional second argument bounds the search; it is a buffer position.\n\ |
| 1067 | The match found must not extend after that position. nil is equivalent\n\ |
| 1068 | to (point-max).\n\ |
| 1069 | Optional third argument, if t, means if fail just return nil (no error).\n\ |
| 1070 | If not nil and not t, move to limit of search and return nil.\n\ |
| 1071 | Optional fourth argument is repeat count--search for successive occurrences.\n\ |
| 1072 | See also the functions `match-beginning', `match-end' and `replace-match'.") |
| 1073 | (string, bound, noerror, count) |
| 1074 | Lisp_Object string, bound, noerror, count; |
| 1075 | { |
| 1076 | return search_command (string, bound, noerror, count, 1, 0); |
| 1077 | } |
| 1078 | |
| 1079 | DEFUN ("word-search-backward", Fword_search_backward, Sword_search_backward, 1, 4, |
| 1080 | "sWord search backward: ", |
| 1081 | "Search backward from point for STRING, ignoring differences in punctuation.\n\ |
| 1082 | Set point to the beginning of the occurrence found, and return point.\n\ |
| 1083 | An optional second argument bounds the search; it is a buffer position.\n\ |
| 1084 | The match found must not extend before that position.\n\ |
| 1085 | Optional third argument, if t, means if fail just return nil (no error).\n\ |
| 1086 | If not nil and not t, move to limit of search and return nil.\n\ |
| 1087 | Optional fourth argument is repeat count--search for successive occurrences.") |
| 1088 | (string, bound, noerror, count) |
| 1089 | Lisp_Object string, bound, noerror, count; |
| 1090 | { |
| 1091 | return search_command (wordify (string), bound, noerror, count, -1, 1); |
| 1092 | } |
| 1093 | |
| 1094 | DEFUN ("word-search-forward", Fword_search_forward, Sword_search_forward, 1, 4, |
| 1095 | "sWord search: ", |
| 1096 | "Search forward from point for STRING, ignoring differences in punctuation.\n\ |
| 1097 | Set point to the end of the occurrence found, and return point.\n\ |
| 1098 | An optional second argument bounds the search; it is a buffer position.\n\ |
| 1099 | The match found must not extend after that position.\n\ |
| 1100 | Optional third argument, if t, means if fail just return nil (no error).\n\ |
| 1101 | If not nil and not t, move to limit of search and return nil.\n\ |
| 1102 | Optional fourth argument is repeat count--search for successive occurrences.") |
| 1103 | (string, bound, noerror, count) |
| 1104 | Lisp_Object string, bound, noerror, count; |
| 1105 | { |
| 1106 | return search_command (wordify (string), bound, noerror, count, 1, 1); |
| 1107 | } |
| 1108 | |
| 1109 | DEFUN ("re-search-backward", Fre_search_backward, Sre_search_backward, 1, 4, |
| 1110 | "sRE search backward: ", |
| 1111 | "Search backward from point for match for regular expression REGEXP.\n\ |
| 1112 | Set point to the beginning of the match, and return point.\n\ |
| 1113 | The match found is the one starting last in the buffer\n\ |
| 1114 | and yet ending before the place the origin of the search.\n\ |
| 1115 | An optional second argument bounds the search; it is a buffer position.\n\ |
| 1116 | The match found must start at or after that position.\n\ |
| 1117 | Optional third argument, if t, means if fail just return nil (no error).\n\ |
| 1118 | If not nil and not t, move to limit of search and return nil.\n\ |
| 1119 | Optional fourth argument is repeat count--search for successive occurrences.\n\ |
| 1120 | See also the functions `match-beginning', `match-end' and `replace-match'.") |
| 1121 | (string, bound, noerror, count) |
| 1122 | Lisp_Object string, bound, noerror, count; |
| 1123 | { |
| 1124 | return search_command (string, bound, noerror, count, -1, 1); |
| 1125 | } |
| 1126 | |
| 1127 | DEFUN ("re-search-forward", Fre_search_forward, Sre_search_forward, 1, 4, |
| 1128 | "sRE search: ", |
| 1129 | "Search forward from point for regular expression REGEXP.\n\ |
| 1130 | Set point to the end of the occurrence found, and return point.\n\ |
| 1131 | An optional second argument bounds the search; it is a buffer position.\n\ |
| 1132 | The match found must not extend after that position.\n\ |
| 1133 | Optional third argument, if t, means if fail just return nil (no error).\n\ |
| 1134 | If not nil and not t, move to limit of search and return nil.\n\ |
| 1135 | Optional fourth argument is repeat count--search for successive occurrences.\n\ |
| 1136 | See also the functions `match-beginning', `match-end' and `replace-match'.") |
| 1137 | (string, bound, noerror, count) |
| 1138 | Lisp_Object string, bound, noerror, count; |
| 1139 | { |
| 1140 | return search_command (string, bound, noerror, count, 1, 1); |
| 1141 | } |
| 1142 | \f |
| 1143 | DEFUN ("replace-match", Freplace_match, Sreplace_match, 1, 3, 0, |
| 1144 | "Replace text matched by last search with NEWTEXT.\n\ |
| 1145 | If second arg FIXEDCASE is non-nil, do not alter case of replacement text.\n\ |
| 1146 | Otherwise convert to all caps or cap initials, like replaced text.\n\ |
| 1147 | If third arg LITERAL is non-nil, insert NEWTEXT literally.\n\ |
| 1148 | Otherwise treat `\\' as special:\n\ |
| 1149 | `\\&' in NEWTEXT means substitute original matched text.\n\ |
| 1150 | `\\N' means substitute what matched the Nth `\\(...\\)'.\n\ |
| 1151 | If Nth parens didn't match, substitute nothing.\n\ |
| 1152 | `\\\\' means insert one `\\'.\n\ |
| 1153 | FIXEDCASE and LITERAL are optional arguments.\n\ |
| 1154 | Leaves point at end of replacement text.") |
| 1155 | (string, fixedcase, literal) |
| 1156 | Lisp_Object string, fixedcase, literal; |
| 1157 | { |
| 1158 | enum { nochange, all_caps, cap_initial } case_action; |
| 1159 | register int pos, last; |
| 1160 | int some_multiletter_word; |
| 1161 | int some_lowercase; |
| 1162 | int some_uppercase_initial; |
| 1163 | register int c, prevc; |
| 1164 | int inslen; |
| 1165 | |
| 1166 | CHECK_STRING (string, 0); |
| 1167 | |
| 1168 | case_action = nochange; /* We tried an initialization */ |
| 1169 | /* but some C compilers blew it */ |
| 1170 | |
| 1171 | if (search_regs.num_regs <= 0) |
| 1172 | error ("replace-match called before any match found"); |
| 1173 | |
| 1174 | if (search_regs.start[0] < BEGV |
| 1175 | || search_regs.start[0] > search_regs.end[0] |
| 1176 | || search_regs.end[0] > ZV) |
| 1177 | args_out_of_range (make_number (search_regs.start[0]), |
| 1178 | make_number (search_regs.end[0])); |
| 1179 | |
| 1180 | if (NILP (fixedcase)) |
| 1181 | { |
| 1182 | /* Decide how to casify by examining the matched text. */ |
| 1183 | |
| 1184 | last = search_regs.end[0]; |
| 1185 | prevc = '\n'; |
| 1186 | case_action = all_caps; |
| 1187 | |
| 1188 | /* some_multiletter_word is set nonzero if any original word |
| 1189 | is more than one letter long. */ |
| 1190 | some_multiletter_word = 0; |
| 1191 | some_lowercase = 0; |
| 1192 | some_uppercase_initial = 0; |
| 1193 | |
| 1194 | for (pos = search_regs.start[0]; pos < last; pos++) |
| 1195 | { |
| 1196 | c = FETCH_CHAR (pos); |
| 1197 | if (LOWERCASEP (c)) |
| 1198 | { |
| 1199 | /* Cannot be all caps if any original char is lower case */ |
| 1200 | |
| 1201 | some_lowercase = 1; |
| 1202 | if (SYNTAX (prevc) != Sword) |
| 1203 | ; |
| 1204 | else |
| 1205 | some_multiletter_word = 1; |
| 1206 | } |
| 1207 | else if (!NOCASEP (c)) |
| 1208 | { |
| 1209 | if (SYNTAX (prevc) != Sword) |
| 1210 | some_uppercase_initial = 1; |
| 1211 | else |
| 1212 | some_multiletter_word = 1; |
| 1213 | } |
| 1214 | |
| 1215 | prevc = c; |
| 1216 | } |
| 1217 | |
| 1218 | /* Convert to all caps if the old text is all caps |
| 1219 | and has at least one multiletter word. */ |
| 1220 | if (! some_lowercase && some_multiletter_word) |
| 1221 | case_action = all_caps; |
| 1222 | /* Capitalize each word, if the old text has a capitalized word. */ |
| 1223 | else if (some_uppercase_initial) |
| 1224 | case_action = cap_initial; |
| 1225 | else |
| 1226 | case_action = nochange; |
| 1227 | } |
| 1228 | |
| 1229 | /* We insert the replacement text before the old text, and then |
| 1230 | delete the original text. This means that markers at the |
| 1231 | beginning or end of the original will float to the corresponding |
| 1232 | position in the replacement. */ |
| 1233 | SET_PT (search_regs.start[0]); |
| 1234 | if (!NILP (literal)) |
| 1235 | Finsert (1, &string); |
| 1236 | else |
| 1237 | { |
| 1238 | struct gcpro gcpro1; |
| 1239 | GCPRO1 (string); |
| 1240 | |
| 1241 | for (pos = 0; pos < XSTRING (string)->size; pos++) |
| 1242 | { |
| 1243 | int offset = point - search_regs.start[0]; |
| 1244 | |
| 1245 | c = XSTRING (string)->data[pos]; |
| 1246 | if (c == '\\') |
| 1247 | { |
| 1248 | c = XSTRING (string)->data[++pos]; |
| 1249 | if (c == '&') |
| 1250 | Finsert_buffer_substring |
| 1251 | (Fcurrent_buffer (), |
| 1252 | make_number (search_regs.start[0] + offset), |
| 1253 | make_number (search_regs.end[0] + offset)); |
| 1254 | else if (c >= '1' && c <= search_regs.num_regs + '0') |
| 1255 | { |
| 1256 | if (search_regs.start[c - '0'] >= 1) |
| 1257 | Finsert_buffer_substring |
| 1258 | (Fcurrent_buffer (), |
| 1259 | make_number (search_regs.start[c - '0'] + offset), |
| 1260 | make_number (search_regs.end[c - '0'] + offset)); |
| 1261 | } |
| 1262 | else |
| 1263 | insert_char (c); |
| 1264 | } |
| 1265 | else |
| 1266 | insert_char (c); |
| 1267 | } |
| 1268 | UNGCPRO; |
| 1269 | } |
| 1270 | |
| 1271 | inslen = point - (search_regs.start[0]); |
| 1272 | del_range (search_regs.start[0] + inslen, search_regs.end[0] + inslen); |
| 1273 | |
| 1274 | if (case_action == all_caps) |
| 1275 | Fupcase_region (make_number (point - inslen), make_number (point)); |
| 1276 | else if (case_action == cap_initial) |
| 1277 | upcase_initials_region (make_number (point - inslen), make_number (point)); |
| 1278 | return Qnil; |
| 1279 | } |
| 1280 | \f |
| 1281 | static Lisp_Object |
| 1282 | match_limit (num, beginningp) |
| 1283 | Lisp_Object num; |
| 1284 | int beginningp; |
| 1285 | { |
| 1286 | register int n; |
| 1287 | |
| 1288 | CHECK_NUMBER (num, 0); |
| 1289 | n = XINT (num); |
| 1290 | if (n < 0 || n >= search_regs.num_regs) |
| 1291 | args_out_of_range (num, make_number (search_regs.num_regs)); |
| 1292 | if (search_regs.num_regs <= 0 |
| 1293 | || search_regs.start[n] < 0) |
| 1294 | return Qnil; |
| 1295 | return (make_number ((beginningp) ? search_regs.start[n] |
| 1296 | : search_regs.end[n])); |
| 1297 | } |
| 1298 | |
| 1299 | DEFUN ("match-beginning", Fmatch_beginning, Smatch_beginning, 1, 1, 0, |
| 1300 | "Return position of start of text matched by last search.\n\ |
| 1301 | ARG, a number, specifies which parenthesized expression in the last regexp.\n\ |
| 1302 | Value is nil if ARGth pair didn't match, or there were less than ARG pairs.\n\ |
| 1303 | Zero means the entire text matched by the whole regexp or whole string.") |
| 1304 | (num) |
| 1305 | Lisp_Object num; |
| 1306 | { |
| 1307 | return match_limit (num, 1); |
| 1308 | } |
| 1309 | |
| 1310 | DEFUN ("match-end", Fmatch_end, Smatch_end, 1, 1, 0, |
| 1311 | "Return position of end of text matched by last search.\n\ |
| 1312 | ARG, a number, specifies which parenthesized expression in the last regexp.\n\ |
| 1313 | Value is nil if ARGth pair didn't match, or there were less than ARG pairs.\n\ |
| 1314 | Zero means the entire text matched by the whole regexp or whole string.") |
| 1315 | (num) |
| 1316 | Lisp_Object num; |
| 1317 | { |
| 1318 | return match_limit (num, 0); |
| 1319 | } |
| 1320 | |
| 1321 | DEFUN ("match-data", Fmatch_data, Smatch_data, 0, 0, 0, |
| 1322 | "Return a list containing all info on what the last search matched.\n\ |
| 1323 | Element 2N is `(match-beginning N)'; element 2N + 1 is `(match-end N)'.\n\ |
| 1324 | All the elements are markers or nil (nil if the Nth pair didn't match)\n\ |
| 1325 | if the last match was on a buffer; integers or nil if a string was matched.\n\ |
| 1326 | Use `store-match-data' to reinstate the data in this list.") |
| 1327 | () |
| 1328 | { |
| 1329 | Lisp_Object *data; |
| 1330 | int i, len; |
| 1331 | |
| 1332 | if (NILP (last_thing_searched)) |
| 1333 | error ("match-data called before any match found"); |
| 1334 | |
| 1335 | data = (Lisp_Object *) alloca ((2 * search_regs.num_regs) |
| 1336 | * sizeof (Lisp_Object)); |
| 1337 | |
| 1338 | len = -1; |
| 1339 | for (i = 0; i < search_regs.num_regs; i++) |
| 1340 | { |
| 1341 | int start = search_regs.start[i]; |
| 1342 | if (start >= 0) |
| 1343 | { |
| 1344 | if (EQ (last_thing_searched, Qt)) |
| 1345 | { |
| 1346 | XFASTINT (data[2 * i]) = start; |
| 1347 | XFASTINT (data[2 * i + 1]) = search_regs.end[i]; |
| 1348 | } |
| 1349 | else if (XTYPE (last_thing_searched) == Lisp_Buffer) |
| 1350 | { |
| 1351 | data[2 * i] = Fmake_marker (); |
| 1352 | Fset_marker (data[2 * i], |
| 1353 | make_number (start), |
| 1354 | last_thing_searched); |
| 1355 | data[2 * i + 1] = Fmake_marker (); |
| 1356 | Fset_marker (data[2 * i + 1], |
| 1357 | make_number (search_regs.end[i]), |
| 1358 | last_thing_searched); |
| 1359 | } |
| 1360 | else |
| 1361 | /* last_thing_searched must always be Qt, a buffer, or Qnil. */ |
| 1362 | abort (); |
| 1363 | |
| 1364 | len = i; |
| 1365 | } |
| 1366 | else |
| 1367 | data[2 * i] = data [2 * i + 1] = Qnil; |
| 1368 | } |
| 1369 | return Flist (2 * len + 2, data); |
| 1370 | } |
| 1371 | |
| 1372 | |
| 1373 | DEFUN ("store-match-data", Fstore_match_data, Sstore_match_data, 1, 1, 0, |
| 1374 | "Set internal data on last search match from elements of LIST.\n\ |
| 1375 | LIST should have been created by calling `match-data' previously.") |
| 1376 | (list) |
| 1377 | register Lisp_Object list; |
| 1378 | { |
| 1379 | register int i; |
| 1380 | register Lisp_Object marker; |
| 1381 | |
| 1382 | if (!CONSP (list) && !NILP (list)) |
| 1383 | list = wrong_type_argument (Qconsp, list); |
| 1384 | |
| 1385 | /* Unless we find a marker with a buffer in LIST, assume that this |
| 1386 | match data came from a string. */ |
| 1387 | last_thing_searched = Qt; |
| 1388 | |
| 1389 | /* Allocate registers if they don't already exist. */ |
| 1390 | { |
| 1391 | int length = XFASTINT (Flength (list)) / 2; |
| 1392 | |
| 1393 | if (length > search_regs.num_regs) |
| 1394 | { |
| 1395 | if (search_regs.num_regs == 0) |
| 1396 | { |
| 1397 | search_regs.start |
| 1398 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); |
| 1399 | search_regs.end |
| 1400 | = (regoff_t *) xmalloc (length * sizeof (regoff_t)); |
| 1401 | } |
| 1402 | else |
| 1403 | { |
| 1404 | search_regs.start |
| 1405 | = (regoff_t *) xrealloc (search_regs.start, |
| 1406 | length * sizeof (regoff_t)); |
| 1407 | search_regs.end |
| 1408 | = (regoff_t *) xrealloc (search_regs.end, |
| 1409 | length * sizeof (regoff_t)); |
| 1410 | } |
| 1411 | |
| 1412 | BLOCK_INPUT; |
| 1413 | re_set_registers (&searchbuf, &search_regs, length, |
| 1414 | search_regs.start, search_regs.end); |
| 1415 | UNBLOCK_INPUT; |
| 1416 | } |
| 1417 | } |
| 1418 | |
| 1419 | for (i = 0; i < search_regs.num_regs; i++) |
| 1420 | { |
| 1421 | marker = Fcar (list); |
| 1422 | if (NILP (marker)) |
| 1423 | { |
| 1424 | search_regs.start[i] = -1; |
| 1425 | list = Fcdr (list); |
| 1426 | } |
| 1427 | else |
| 1428 | { |
| 1429 | if (XTYPE (marker) == Lisp_Marker) |
| 1430 | { |
| 1431 | if (XMARKER (marker)->buffer == 0) |
| 1432 | XFASTINT (marker) = 0; |
| 1433 | else |
| 1434 | XSET (last_thing_searched, Lisp_Buffer, |
| 1435 | XMARKER (marker)->buffer); |
| 1436 | } |
| 1437 | |
| 1438 | CHECK_NUMBER_COERCE_MARKER (marker, 0); |
| 1439 | search_regs.start[i] = XINT (marker); |
| 1440 | list = Fcdr (list); |
| 1441 | |
| 1442 | marker = Fcar (list); |
| 1443 | if (XTYPE (marker) == Lisp_Marker |
| 1444 | && XMARKER (marker)->buffer == 0) |
| 1445 | XFASTINT (marker) = 0; |
| 1446 | |
| 1447 | CHECK_NUMBER_COERCE_MARKER (marker, 0); |
| 1448 | search_regs.end[i] = XINT (marker); |
| 1449 | } |
| 1450 | list = Fcdr (list); |
| 1451 | } |
| 1452 | |
| 1453 | return Qnil; |
| 1454 | } |
| 1455 | |
| 1456 | /* Quote a string to inactivate reg-expr chars */ |
| 1457 | |
| 1458 | DEFUN ("regexp-quote", Fregexp_quote, Sregexp_quote, 1, 1, 0, |
| 1459 | "Return a regexp string which matches exactly STRING and nothing else.") |
| 1460 | (str) |
| 1461 | Lisp_Object str; |
| 1462 | { |
| 1463 | register unsigned char *in, *out, *end; |
| 1464 | register unsigned char *temp; |
| 1465 | |
| 1466 | CHECK_STRING (str, 0); |
| 1467 | |
| 1468 | temp = (unsigned char *) alloca (XSTRING (str)->size * 2); |
| 1469 | |
| 1470 | /* Now copy the data into the new string, inserting escapes. */ |
| 1471 | |
| 1472 | in = XSTRING (str)->data; |
| 1473 | end = in + XSTRING (str)->size; |
| 1474 | out = temp; |
| 1475 | |
| 1476 | for (; in != end; in++) |
| 1477 | { |
| 1478 | if (*in == '[' || *in == ']' |
| 1479 | || *in == '*' || *in == '.' || *in == '\\' |
| 1480 | || *in == '?' || *in == '+' |
| 1481 | || *in == '^' || *in == '$') |
| 1482 | *out++ = '\\'; |
| 1483 | *out++ = *in; |
| 1484 | } |
| 1485 | |
| 1486 | return make_string (temp, out - temp); |
| 1487 | } |
| 1488 | \f |
| 1489 | syms_of_search () |
| 1490 | { |
| 1491 | register int i; |
| 1492 | |
| 1493 | searchbuf.allocated = 100; |
| 1494 | searchbuf.buffer = (unsigned char *) malloc (searchbuf.allocated); |
| 1495 | searchbuf.fastmap = search_fastmap; |
| 1496 | |
| 1497 | Qsearch_failed = intern ("search-failed"); |
| 1498 | staticpro (&Qsearch_failed); |
| 1499 | Qinvalid_regexp = intern ("invalid-regexp"); |
| 1500 | staticpro (&Qinvalid_regexp); |
| 1501 | |
| 1502 | Fput (Qsearch_failed, Qerror_conditions, |
| 1503 | Fcons (Qsearch_failed, Fcons (Qerror, Qnil))); |
| 1504 | Fput (Qsearch_failed, Qerror_message, |
| 1505 | build_string ("Search failed")); |
| 1506 | |
| 1507 | Fput (Qinvalid_regexp, Qerror_conditions, |
| 1508 | Fcons (Qinvalid_regexp, Fcons (Qerror, Qnil))); |
| 1509 | Fput (Qinvalid_regexp, Qerror_message, |
| 1510 | build_string ("Invalid regexp")); |
| 1511 | |
| 1512 | last_regexp = Qnil; |
| 1513 | staticpro (&last_regexp); |
| 1514 | |
| 1515 | last_thing_searched = Qnil; |
| 1516 | staticpro (&last_thing_searched); |
| 1517 | |
| 1518 | defsubr (&Sstring_match); |
| 1519 | defsubr (&Slooking_at); |
| 1520 | defsubr (&Sskip_chars_forward); |
| 1521 | defsubr (&Sskip_chars_backward); |
| 1522 | defsubr (&Sskip_syntax_forward); |
| 1523 | defsubr (&Sskip_syntax_backward); |
| 1524 | defsubr (&Ssearch_forward); |
| 1525 | defsubr (&Ssearch_backward); |
| 1526 | defsubr (&Sword_search_forward); |
| 1527 | defsubr (&Sword_search_backward); |
| 1528 | defsubr (&Sre_search_forward); |
| 1529 | defsubr (&Sre_search_backward); |
| 1530 | defsubr (&Sreplace_match); |
| 1531 | defsubr (&Smatch_beginning); |
| 1532 | defsubr (&Smatch_end); |
| 1533 | defsubr (&Smatch_data); |
| 1534 | defsubr (&Sstore_match_data); |
| 1535 | defsubr (&Sregexp_quote); |
| 1536 | } |